United States
Environmental
Protection Agency
Office of Water
(4601)
EPA815-D-03-007
June 2003
Draft
ULTRAVIOLET DISINFECTION GUIDANCE MANUAL
is
202-566-0556
-------
Note on the Ultraviolet Disinfection Guidance Manual, June 2003 Draft
Purpose: The purpose of this guidance manual, when finalized, is solely to provide technical
information on the application of ultraviolet light for the disinfection of drinking water by public
water systems. EPA is developing this manual to support two upcoming drinking water
regulations: the Long Term 2 Enhanced Surface Water Treatment Rule, which would require
certain systems to provide additional treatment for Cryptosporidium, and the Stage 2
Disinfection Byproducts Rule, which would place more stringent limits on certain disinfection
byproducts. Chapter 1 of this manual contains additional information about these regulations.
This guidance is not a substitute for applicable legal requirements, nor is it a regulation itself.
Thus, it does not impose legally-binding requirements on any party, including EPA, states, or the
regulated community. Interested parties are free to raise questions and objections to the
guidance and the appropriateness of using it in a particular situation. Although this manual
covers many aspects of implementing a UV system, it is not comprehensive in terms of all types
of UV systems, design alternatives, and validation protocols that may provide satisfactory
performance. The mention of trade names or commercial products does not constitute
endorsement or recommendation for use.
Authorship: This manual was developed under the direction of EPA's Office of Water, and was
prepared by Malcolm Pirnie, Inc., Carollo Engineers, P.C., The Cadmus Group, Inc., Dr. Karl G.
Linden, and Dr. James P. Malley, Jr. Questions concerning this document should be addressed
to:
Dan Schmelling
U.S. Environmental Protection Agency
Mail Code 4607M
1200 Pennsylvania Avenue NW
Washington, DC 20460-0001
Tel: (202) 564-5281
Fax: (202) 564-3767
Email: schmelling.dan@epa.gov
Request for comments: EPA is releasing this manual in draft form in order to solicit public
review and comment. The Agency would appreciate comments on the content and organization
of technical information presented in this manual. A list of topics for comment pertaining to
specific chapters and appendices is provided later in this manual. Please submit any comments
no later than 90 days after publication of the Long Term 2 Enhanced Surface Water Treatment
Rule proposal in the Federal Register. Detailed procedures for submitting comments are stated
below.
-------
Acknowledgements: The following provided valuable technical information to assist in the
development of this manual:
American Water Works Association
Dave Battigelli (Clancy Environmental Consultants)
William Bellamy (CH2M Hill)
Jim Bolton (Executive Director, International Ultraviolet Association)
Calgon Carbon Corporation
Tom Hargy (Clancy Environmental Consultants)
OlufHoyer(DVGW)
Richard Hubel (American Water)
Chris McMeen (then with Washington Department of Health)
Alexander Mofidi (Metropolitan Water District of Southern California)
Ondeo Degremont
Richard Sakaji (California Department of Health Services)
Severn Trent Services
Regina Sommer (University of Vienna)
Paul Swaim (CH2M Hill)
Trojan Technologies
Wedeco-Ideal Horizons
John Young (American Water)
Procedures for submitting comments: Comments on this draft guidance manual should be
submitted to EPA's Water Docket. You may submit comments electronically, by mail, or
through hand delivery/courier.
. To submit comments using EPA's electronic public docket, go directly to EPA Dockets at
http://www.epa.gov/edocket, and follow the online instructions for submitting comments.
Once in the system, select "search," and then key in Docket ID No. OW-2002-0039.
• To submit comments by e-mail, send comments to OW_Docket@epa.gov, Attention Docket
ID No. OW-2002-0039. If you send an e-mail comment directly to the Docket without going
through EPA's electronic public docket, EPA's e-mail system automatically captures your e-
mail address, which is included as part of the comment that is placed in the official public
docket.
To submit comments on a disk or CD ROM, mail it to the address identified below. These
electronic submissions will be accepted in WordPerfect or ASCII file format. Avoid the use
of special characters and any form of encryption.
» To submit comments by mail, send three copies of your comments and any enclosures to:
Water Docket, Environmental Protection Agency, Mail Code 4101T, 1200 Pennsylvania
Ave., NW, Washington, DC, 20460, Attention Docket ID No. OW-2002-0039.
• To submit comments by hand delivery or courier, deliver your comments to: Water Docket,
EPA Docket Center, Environmental Protection Agency, Room B102, 1301 Constitution
Ave., NW, Washington, DC, Attention Docket ID No. OW-2002-0039.
-------
Please identify the appropriate docket identification number in the subject Hne on the first page
of your comment. If you submit an electronic comment, please include your name, mailing
address, and an e-mail address or other contact information in the body of your comment. Also
include this contact information on the outside of any disk or CD ROM you submit, and in any
cover letter accompanying the disk or CD ROM.
For public commenters, please note that EPA's policy is that public comments, whether
submitted electronically or in paper, will be made available for public viewing in EPA's
electronic public docket as EPA receives them and without change, unless the comment contains
copyrighted material, confidential business information, or other information whose disclosure is
restricted by statute.
-------
Table of Contents
Table of Contents i
List of Figures vi
List of Tables vii
Glossary viii
Acronyms and Abbreviations : i xiii
1.0 Introduction...;..... 1-1
1.1 Guidance Manual Objectives 1-1
1.2 Organization -, 1-2
1.3 Regulations Summary 1-3
.1.3.1 Long Term 2 Enhanced Surface Water Treatment Rule 1-4
1.3.1.1 Filtered Systems 1-5
1.3.1.2 Unfiltered Systems. 1-6
1.3.1.3 UV Requirements For Filtered And Unfiltered Systems 1-7
1.3.2 Stage2DBPR 1-9
1.4 Alternative Approaches for Disinfecting with UV Light 1-9
2.0 Overview of UV Disinfection 2-1
2.1 History of UV Light for Drinking Water Disinfection •. 2-1
2.2 Fundamental Aspects of UV Light 2-2
2.2.1 Nature of UV Light '. 2-2
2.2.2 Propagation of UV Light 2-3
2.3 Microbial Response to UV Light...; ; 2-6
2.3.1 Mechanisms of Microbial Inactivation by UV Light 2-6
2.3.2 Microbial Repair .' '.. 2-7
2.3.3 UV Dose and Dose Distribution ; 2-10
2.3.4 Microbial Response (UV Dose-Response) : 2-11
2.3.5 Microbial Spectral Response 2-11
2.4 UV Reactors 2-12
2.4.1 Reactor Configuration 2-13
2.4.2 UV Lamps 2-14
2.4.3 Lamp Power Supply and Ballasts 2-17
2.4.4 Lamp Sleeves : 2-17
2.4.5 Cleaning Systems : 2-18
2.4.6 UV Intensity Sensors 2-19
2.4.7 UV Transmittance Monitors 2-20
2.4.8 Temperature Sensors 2-21
2.4.9 Monitoring UV Disinfection Performance 2-21
UV Disinfection Guidance Manual i June 2003
Proposal Draft
-------
Table of Contents (Continued)
2.5 Water Quality Impacts and Byproduct Formation 2-22
2.5.1 Water Quality Impacts 2-22
2.5.2 Byproducts from UV Disinfection 2-25
3.0 Planning and Design Aspects for UV Installations 3-1
3.1 UV Installations Planning 3-4
3.1.1 Defining UV Disinfection Goals 3-4
3.1.2 Identifying Potential Locations For UV Installations 3-5
3.1.2.1 Combined Filter Effluent Installation 3-6
3.1.2.2 Individual Filter Effluent Piping Installation 3-6
3.1.2.3 UV Disinfection Downstream of The Clearwell 3-8
3.1.3 Defining Design Parameters 3-9
3.1.3.1 Assessing Water Quality 3-10
3.1.3.2 Determining Design Flow Rate 3-17
3.1.3.3 Assessing Electrical Power 3-17
3.1.4 Evaluating Potential UV Reactors 3-20
3.1.4.1 UV Reactors, 3-20
3.1.4.2 UV Reactor Control Strategies 3-22
3.1.4.3 Equipment Validation Issues 3-22
3.1.5 Evaluating Operational Strategies 3-24
3.1.6 Evaluating Hydraulics and Process Footprint 3-25
3.1.6.1 Hydraulic Considerations 3-25
3.1.6.2 Process Footprint 3-27
3.1.7 Preparing Preliminary Costs and Selecting the UV Installation Option 3-28
3.2 Equipment Procurement Options ; 3-29
3.3 UV Installation Design Elements 3-30
3.3.1 UV Installation Hydraulics 3-30
3.3.1.1 Inlet and Outlet Piping Configuration 3-31
3.3.1.2 Flow Distribution, Control, and Measurement 3-31
3.3.1.3 Level Control 3-36
3.3.1.4 Air Relief and Pressure Control Valves 3-36
3.3.1.5 Flow Control and Isolation Valves 3-37
3.3.1.6 Intermediate Booster Pumps 3-37
3.3.2 Operational Strategy Determination 3-38
3.3.3 Instrumentation and Control 3-38
3.3.3.1 UV Reactor Start-Up 3-39
3.3.3.2 UV Reactor Automation 3-39
3.3.3.3 UV Intensity and Calculated Dose (If Applicable) 3-39
3.3.3.4 UV Transmittance 3-40
3.3.3.5 Flow Measurement 3-40
3.3.3.6 Lamp Age 3-40
3.3.3.7 Lamp and Reactor Status 3-41
3.3.3.8 Alarms and Control Systems Interlocks 3-41
UV Disinfection Guidance Manual ii June 2003
Proposal Draft
-------
Table of Contents (Continued)
3.3.4 Electrical Power Configuration : 3-42
3.3.4.1 Power Requirements 3-43
3.3.4.2 Backup Power Supply 3-43
3.3.4.3 Ground Fault Interrupt and Electrical Lockout 3-44
3.3.5 UV Installation Layouts 3-45
3.3.5.1 Site Layout ! 3-45
3.3.5.2 UV Installation Layout ; 3-45
3.3.6 Elements Of UV Reactor Specifications 3-47
3.3.6.1 Information Provided by Manufacturer in UV
Reactor Bid 3-49
3.3.7 Final UV Installation Design ...„ 3-51
3.3.7.1 Design Drawings 3-51
3.3.7.2 Specifications 3-52
3.4 Reporting to the State 3-52
3.4.1 Planning 3-52
3.4.2 Equipment Procurement 3-53
3.4.3 Drawings and Specifications 3-53
3.4.4 Validation Report/Start-up Confirmation 3-53
4.0 Overview of Validation Testing 4-1
4.1 LT2ESWTR UV Disinfection Requirements ..4-1
4.2 Overview of Validation Process 4-2
4.2.1 Relating the Experimental RED to Log Inactivation Credit 4-4
4.2.1.1 Tier 1 and Tier 2 Approaches for Establishing Inactivation
Credit 4-5
4.2.2 Location and Application of Validation Testing 4-5
4.2.3 Third-Party Oversight : 4-6
4.3 Considerations for Validation Testing '..'. 4-6
4.3.1 Inlet and Outlet Hydraulics 4-7
4.3.2 UV Equipment '. 4-7
4.3.2.1 UV Reactor Documentation 4-7
4.3.2.2 Control Strategies : 4-7
4.3.2.3 UV Intensity Sensor : 4-8
4.3.2.4 Lamp Aging 4-8
4.3.3 Additives Used in Validation Testing 4-8
4.3.3.1 Challenge Microorganism 4-8
4.3.3.2 UV-Absorbing Material 4-9
4.4 Validation Testing 4-9
4.4.1 Microorganism Preparation : 4-9
4.4.2 Collimated Beam Testing 4-9
4.4.3 Biodosimetry of Full-Scale Reactors 4-10
UV Disinfection Guidance Manual iii June 2003
Proposal Draft
-------
Table of Contents (Continued)
4.5 Data Analysis 4-11
4.5.1 Developing Challenge Microorganisms Dose-Response Curve 4-12
4.5.1.1 Calculate Dose-Response Data From Collimated Beam
Testing 4-12
4.5.1.2 Fitting Dose-Response Data to a Curve 4-13
4.5.2 Determining Log Inactivation from Biodosimetry Testing 4-13
4.5.3 Determining the RED 4-14
4.5.3.1 Calculating the RED Values 4-14
4.5.3.2 Selecting the Appropriate RED for Log Inactivation
Credit Determination 4-14
4.5.3.3 Interpolating RED as a Function of Test Conditions 4-15
4.5.4 Determining Inactivation Credit 4-15
4.6 Tier 1 Criteria ; 4-17
4.6.1 UV Intensity Sensors ; 4-17
4.6.2 UV Lamp Output 4-19
4.6.3 Flow Measurements 4-19
4.6.4 Collimated Beam Apparatus 4-19
4.6.5 Challenge Microorganism Dose-Response 4-19
4.6.6 Medium Pressure Lamps 4-20
4.6.7 Biodosimetry Sampling 4-21
5.0 Start-Up and Operation of UV Installations 5-1
5.1 Start-up of UV Installation 5-3
5.1.1 Final Inspection 5-3
5.1.2 Functional Testing 5-3
5.1.2.1 Verification of Mechanical Operation 5-4
5.1.2.2 Verification of Monitoring Equipment 5-4
5.1.2.3 Verification of Instrumentation and Control Systems 5-5
5.1.2.4 Verification of Flow Distribution and Headless 5-6
5.1.3 Performance Testing 5-7
5.1.4 Operations and Maintenance Manual 5-9
5.2 Operation of UV Installations , 5-10
5.2.1 Operational Requirements 5-11
5.2.2 Recommended Operational Tasks 5-11
5.2.3 Start-up and Shutdown of UV Reactors 5-12
5.2.3.1 Routine Start-up 5-12
5.2.3.2 Routine Shutdown 5-13
5.2.3.3 Winterization 5-13
5.3 Maintenance of UV Reactors 5-13
5.3.1 Summary of Recommended Maintenance Tasks 5-14
5.3.2 General Guidelines for UV Reactor Maintenance 5-15
5.3.2.1 UV Lamp Characteristics 5-15
5.3.2.2 UV Intensity Sensors 5-16
5.3.2.3 Lamp Sleeves 5-18
5.3.2.4 Fouling 5-18
UV Disinfection Guidance Manual iv June 2003
Proposal Draft
-------
Table of Contents (Continued)
5.3.2.5 On-line UVT Monitor Calibration 5-20
5.3.2.6 Flowmeter Calibration 5-20
5.3.2.7 UV Reactor Temperature 5-20
5.3.2.8 Electrical Concerns 5-21
5.3.3 Spare Parts 5-22
5.4 Monitoring, Recording, and Reporting of UV Installation Operation 5-24
5.4.1 Monitoring and Recording Frequency for Compliance Parameters 5-24
5.4.2 Monitoring and Recording for Other Operational Parameters 5-25
5.4.3 Reporting to the State 5-26
5.5 Determination of Validated Operational Parameters 5-27
5.6 Operational Challenges '. 5-32
5.6.1 Low UV Intensity or Low Calculated UV Dose 5-32
5.6.2 Low UV Transmittance 5-34
5.6.3 Rapid Flow Increase or High Flow 5-36
5.6.4 Unreliable UV Intensity Sensor Readings 5-36
5.6.5. Power Quality Problems ; 5-37
5.7 Staffing Issues 5-37
5.7.1 Staffing Levels 5-38
5.7.2 Training... 5-38
5.7.3 Safety Issues 5-39
6.0 References 6-1
Appendices
Appendix A Fundamentals of UV Disinfection A-l
Appendix B Derivation of UV Dose-Response Requirements B-l
Appendix C Validation of UV Reactors .' C-
Appendix D Microbiological Methods D-
Appendix E Measuring Challenge Microorganism UV Dose-Response E-
Appendix F Background to the UV Reactor Validation Protocol F-
Appendix G Issues for Unfiltered Systems G-
Appendix H Issues for Ground Water Systems H-
Appendix I Issues for Small Systems I-
Appendix J Pilot-Scale and Demonstration-Scale Testing J-
Appendix K Preliminary Engineering Report K-l
Appendix L Regulatory Timeline L-l
Appendix M Compliance Forms M-l
Appendix N UV Lamp Breakage Issues N-l
Appendix O Case Studies [This appendix will be included in the final
draft when more information being available.] O-l
Appendix P Validation Protocol Calculator Tool P-l
UV Disinfection Guidance Manual
Proposal Draft
June 2003
-------
List of Figures
Figure 2.1 UV Light in the Electromagnetic Spectrum 2-3
Figure 2.2 Refraction of Light 2-4
Figure 2.3 Reflection of Light 2-4
Figure 2.4 Scattering of Light 2-5
Figure 2.5 Structure of DNA and Nucleotide Sequences Within DNA 2-6
Figure 2.6 UV Absorbance of Nucleotides and Nucleic Acid at pH 7 2-7
Figure 2.7 Hypothetical Dose Distributions for Two Reactors with Differing
Hydraulics 2-9
Figure 2.8 Shapes of UV Dose-Response Curves 2-11
Figure 2.9 Comparison of Microbial UV Action and DNA UV Absorbance 2-12
Figure 2.10 UV Disinfection System Schematic 2-13
Figure 2.11 Example of Closed and Open Channel Reactors 2-14
Figure 2.12 UV Output of LP and MP Mercury Vapor Lamps 2-16
Figure 2.13 UV Lamp Output and its Relation to the UV Absorbance of DNA 2-17
Figure 2.14 UV Transmittance of Quartz that is 1 mm Thick at a Zero Degree
Incidence Angle 2-18
Figure 2.15 Mechanical Wiper System and Physical-Chemical Wiper System 2-19
Figure 2.16 UV Intensity Sensor Viewing Lamps within a UV Reactor 2-20
Figure 2.17 UV Transmittance Monitor Design 2-21
Figure 3.1 Flowchart for Planning, Design, and Construction of UV Facilities 3-3
Figure 3.2 Schematic for UV Installation (Upstream of Clearwell) 3-6
Figure 3.3 Schematic of Individual Filter Effluent Piping Installation in Filter Gallery 3-7
Figure 3.4 UV Disinfection Downstream of High Service Pumps 3-8
Figure 3.5 Example CF Diagram for Three Filtered Waters 3-12
Figure 3.6 Example Flow and UV Absorbance (at 254) Data 3-13
Figure 3.7 Example Effect of Pre-ozonation on UV Absorbance if Ozone is
Quenched Prior to UV Disinfection 3-16
Figure 3.8 Open-Channel Flow Distribution Options 3-33
Figure 3.9 Flow Measurement and Control Options 3-36
Figure 4.1 Steps of a Validation Process 4-3
Figure 4.2 Collimated Beam Test Apparatus 4-10
Figure 4.3 Biodosimetry Test Components '. 4-11
Figure 4.4 Examples of UV Intensity Sensor Spectral Response Ranges 4-18
Figure 4.5 Dose-Response with a Shoulder 4-20
Figure 4.6 Criteria for the Minimum UVT of MP UV Systems Under Tier 1 4-21
Figure 5.1 Start-Up and Operation Flowchart 5-2
Figure 5.2 Example 2-Interpoloation of Validation Data to Determine UV Intensity
Setpoints 5-29
Figure 5.3 Interpolation of Validation Data to Determine UV Intensity Setpoints at
Different Flows and Cryptosporidium Inactivation 5-30
Figure 5.4 Low UV Intensity of Low Calculated UV Dose Decision Chart.... 5-33
Figure 5.5 High UV Absorbance Decision Chart 5-35
UV Disinfection Guidance Manual vi June 2003
Proposal Draft
-------
List of Tables
Table 1.1 Summary of Microbia! and Disinfection Byproduct Rules 1-4
Table 1.2 Bin Requirements for Filtered Systems ; 1-6
Table 1.3 Bin Requirements for Unfiltered Systems 1-7
Table 1.4 UV Dose Requirements Used During Validation Testing 1-7
Table 2,1 Mercury Vapor Lamp Characteristics 2-15
Table 2.2 Mercury Vapor Lamp Comparison '. 2-15
Table 2.3 Water Quality Data and Fouling Observed for UV Disinfection Pilot and
Demonstration Studies ; : 2-24
Table 3.1 Potential Method to Determine Design Flow 3-17
Table 3.2 Start and Restart Times for LPHO and MP Lamps 3-18
Table 3.3 UV Reactor Control Strategies ' 3-22
Table 3.4 Summary of Recommended Hydraulic Configurations for
Validation and Installation ..! ; 3-23
Table 3.5 Potential Operational Strategies '. '. .: 3-25
Table 3.6 Potential UV Reactor Procurement Options 3-30
Table 3.7 Comparison of Techniques for UV Installation Flow Measurement 3-35
Table 3.8 Typical Alarm Conditions for UV Systems ; 3-42
Table 3.9 Recommended Content for UV Reactor Specifications 3-48
Table 3.10 Recommended Information to be Provided by UV Manufacturer/Vendor....... 3-50
Table 4.1 Tier 1 RED Targets for UV System with LP of LPHO Lamps 4-16
Table 4.2 Tier 1 RED Targets for UV System with MP Lamps 4-16
Table 5.1 Example Monitoring During a Four Week Performance Test 5-9
Table 5.2 Recommended Operational Tasks for the UV Reactor 5-11
Table 5.3 Recommended Maintenance Tasks 5-14
Table 5.4 Design and Guaranteed Lives of Major UV Components 5-23
Table 5.5 Off-Specification Operations for Each Control Strategy 5-24
Table 5.6 Monitoring Parameters and Recording Frequency 5-25
Table 5.7 Recommended Monitoring Parameters and Recording Frequency 5-26
Table 5.8 UV Reactor Control Strategies 5-27
Table 5.9 Example Validation Data for Variable Setpoint Operation 5-28
Table 5.10 UV Intensity Setpoint for Different Flow Ranges ; 5-28
Table 5.11 Example Validation Data for Variable Setpoint Operation 5-29
Table 5.12 . UV Intensity Setpoint for Different Flow Ranges 5-30
Table 5.13 Dose Setpoints for Various Log Inactivation of Cryptosporidium.. 5-31.
Table 5.14 Factors Impacting Staffing Needs .'. 5-38
UV Disinfection Guidance Manual
Proposal Draft
VII
June 2003
-------
Glossary
The following definitions were derived from existing UV literature, standard physics
textbooks, and/or industry standards and conventions. Some concepts have more than one
acceptable term or definition, but for consistency within the document, only one term is used.
Absorption - the transformation of UV light to other forms of energy as it passes through a
substance.
Action Spectrum - the relative efficiency of UV energy at different wavelengths in inactivating
microorganisms. Each microorganism has a unique action spectrum.
Ballast - provides the proper voltage and current required to initiate and maintain the gas
discharge within the UV lamp.
Bioassay - a procedure used to determine the response of a specific microorganism after
exposure to UV light, usually in UV reactors. Bioassay is a term typically utilized in toxicology,
describing the testing of the bio-toxicity of a contaminant. Bioassay has been used in the UV
disinfection literature in the same context as "biodosimetry" (see biodosimetry).
Biodosimeter - the challenge microorganism used to measure UV inactivation and ultimately
calculate the reduction equivalent dose (RED; see UV dose) in a UV reactor.
Biodosimetry - a procedure used to determine the reduction equivalent dose (RED) of a UV
reactor. Biodosimetry involves measuring the inactivation of a challenge microorganism after
exposure to UV light in a UV reactor and comparing the results to the known UV dose-response
curve of the challenge microorganism (determined using collimated beam testing) to determine
the RED (see UV Dose).
Challenge Organism - a microorganism used in UV reactor biodosimetry testing.
Collimated Beam Test - a carefully controlled bench-scale test that is used to determine the UV
dose-response of a microorganism. Both time and UV light intensity are accurately measured,
resulting in a specific calculation of delivered UV dose for the microorganism being tested.
Collimated beam tests are described in detail in Appendix C.
Dark Repair - an enzyme-mediated microbial process that removes and regenerates a damaged
section of deoxyribonucleic acid (DNA), using an existing complimentary strand of DNA. Dark
repair refers to all microbial repair processes not requiring reactivating light:
Delivered UV Dose - see UV Dose
Dose Control Strategy - the technique used by a UV system to control the delivered dose that
typically involves adjusting the lamp power or turning "on" or "off1 banks of UV lamps to
respond to changes in UV absorbance, lamp intensity, and flow. Typically, the dose control
strategy is different for LP/LPHO and MP systems.
UV Disinfection Guidance Manual viii June 2003
Proposal Draft
-------
Glossary (Continued)
Dose Distribution - see UV Dose, Delivered UV Dose Distribution.
Emission Spectrum - the relative light power emitted by a lamp as a function of wavelength.
Fluence - see UV Dose
Fluence Rate - see UV Intensity
Gas Discharge - a mixture of non-excited atoms, excited atoms, cations, and free electrons
formed when a sufficiently high voltage is applied across a volume of gas. Most commercial UV
lamps use mercury gas discharges to generate UV light.
Germicidal Effectiveness - the relative inactivation efficiency of each UV wavelength in a
polychromatic emission spectrum. This value is usually approximated by the relative absorbance
of DNA at each wavelength, although individual microorganisms may respond differently. By
convention, germicidal effectiveness of the 254 nm emission line by LP UV lamps is considered
to be unity. The germicidal effectiveness is typically used to weight a polychromatic, MP UV
lamp output to reflect the germicidal energy of that specific source.
Germicidal Range - the range of UV wavelengths responsible for microbial inactivation in
water (200 to 300 nm).
Lamp Envelope - the exterior surface of the UV lamp, which is typically made of quartz.
Lamp Sleeve - the quartz tube that surrounds and protects the UV lamp. The exterior is in
direct contact with the water being treated. There is typically an air gap (approximately 1 cm)
between the lamp envelope and the quartz sleeve.
Light Pipe - a quartz cylinder that transmits light from the interior of the UV reactor to the
photodetector of a UV intensity sensor.
Lignin Sulfonate - a commercially available reagent grade chemical that can simulate the UV
absorbance spectrum of natural waters and be used to adjust UV transmittance during validation
testing.
.(*
Low Pressure (LP) Lamp - a mercury vapor lamp that operates at an internal pressure of 0.001
to 0.01 torr (2 x 10"5to 2 x 10~4 psi) and electrical input of 0.5 watts per centimeter. This results
in essentially monochromatic light output at 254 nanometers.
Low Pressure High Output (LPHO) Lamp - a low pressure mercury vapor lamp that operates
under increased electrical input (1.5 to 10 W/cm), resulting in a higher UV intensity than LP
lamps. It also has essentially monochromatic light output at 254 nanometers.
Medium Pressure (MP) Lamp - a mercury vapor lamp that operates at an internal pressure of
100 to 10,000 torr (2 to 200 psi) and electrical input of 50 to 150 W/cm. This results in a
polychromatic (or broad spectrum) output of UV and visible light at multiple wavelengths,
including the germicidal range.
UV Disinfection Guidance Manual ix June 2003
Proposal Draft
-------
Glossary (Continued)
Monochromatic - light output at only one wavelength. For example, because low pressure and
low pressure high output lamps only significantly emit light at 254 nanometers, they are
considered monochromatic UV light sources.
Monitoring Window - a quartz disc that transmits light from the interior of the UV reactor to
the photodetector of a UV intensity sensor.
Offline Chemical Clean (OCC) - a process to clean lamp sleeves where the UV reactor is taken
off-line and a cleaning solution (typically an acid) is manually pumped into the reactor. After
the foulant has dissolved, the reactor is drained, rinsed, and returned to service. Also called
flush-and-rinse systems.
Online Mechanical Clean (OMC) - a process to clean lamp sleeves where an automatic
mechanical wiper (e.g., O-ring, brush) wipes the surface of the lamp sleeve at a prescribed
frequency.
Petri Factor - a ratio used in collimated beam testing that is equal to the average intensity
measured across the surface of a suspension in a petri dish divided by the intensity at the center
of a petri dish. The petri factor is used to help calculate delivered UV dose as described in
Appendix C.
i
Photodetector - a device that produces an electrical current proportional to the UV light
intensity at the detector's surface.
Photoreactivation - a microbial repair process where enzymes activated by light in the near UV
and visible range (310 to 490 nm) split pyrimidine dimers, thereby repairing UV induced
damage. Photoreactivation requires the presence of light.
Polychromatic - light energy output at several wavelengths such as with MP lamps.
Quartz Sleeve - see lamp sleeve
Radiometer - an instrument used to measure UV irradiance
Reduction Equivalent Dose (RED) - see UV Dose, RED.
Reflection - the change in direction of light propagation when deflected by an interface or
surface.
Refraction - the change in direction of light propagation as it passes through one medium to
another.
Scattering - the change in direction of light propagation caused by interaction with a particle.
Spectral UV Absorbance - the determination of UV Absorbance (A) over a range of
wavelengths (e.g. 200 to 400 nm)
UV Disinfection Guidance Manual x June 2003
Proposal Draft
-------
Glossary (Continued)
State - the agency of the state, tribal, or federal government that has jurisdiction over public
water systems.
UV absorbance (A) - a measure of the amount of UV light that is absorbed by a substance (e.g.,
water, microbial DNA, lamp envelope, quartz sleeve) at a specific wavelength (e.g., 254 nm).
This measurement accounts for absorption and scattering in the medium (e.g., water). Typically
the absorbance is measured on a per centimeter (cm) basis in a 1 cm quartz cuvette. Standard
Method 591 OB details this measurement method. However, for UV disinfection applications, the
sample should not be filtered or adjusted for pH as described in Standard Methods.
UV Dose - the energy per unit area incident on a surface, typically in units of ml/cm2 or J/m2
(older literature also used the units mW-s/cm2 where 1 mW-s/cm2 = 1 ml/cm2). The UV dose
received by a waterborne microorganism in a reactor vessel accounts for the effects on UV
intensity of the absorbance of the water, absorbance of the quartz sleeves, reflection and
refraction of light from the water surface and reactor walls, and the germicidal effectiveness of
the UV wavelengths. This guidance also uses the following terms related to UV dose:
• Delivered UV dose distribution - the probability distribution of delivered UV doses
that microorganisms receive in a flow-through UV reactor; typically shown as a
histogram. An example is shown in Figure 2-7.
• Reduction Equivalent Dose (RED) - a calculated dose for a flow through UV
reactor that is based on biodosimetry (i.e., measuring the level of inactivation of a
challenge microorganism with a known UV dose-response). The RED is set equal to
the UV dose in a colliniated beam test that achieves the same level of inactivation of
the challenge microorganism as measured for the flow-through UV reactor during
biodosimetry testing.
UV Dose-Response - the relationship indicating the level of inactivation of a microbe as a
function of UV dose. Inactivation is often plotted as logio(No/N) where NO is the number of
microbes present prior to UV light exposure and N is the number of microbes present after UV
light exposure. Examples are shown in Figure 2-8.
UV Installation - all of the components of the UV disinfection process, including (but not
limited to) UV reactors, control systems, piping, valves, and building or enclosure.
UV Intensity - the power per unit area passing through an area perpendicular to the direction of
propagation. UV intensity is used in this guidance manual to describe the magnitude of UV light
in a UV reactor and in bench-scale UV experiments.
UV Intensity Sensor - a photosensitive detector used to measure the UV intensity at a point
within the UV reactor.
UV Irradiance - the power per unit area incident to the direction of light propagation at all
angles, including normal.
UV Light - electromagnetic radiation with wavelengths from 200 to 400 nm.
UV Disinfection Guidance Manual xi June 2003
Proposal Draft
-------
Glossary (Continued)
UV Reactor - the vessel or chamber where exposure to UV light takes place, consisting of UV
lamps, quartz sleeves, UV intensity sensors, quartz sleeve cleaning systems, and baffles or other
hydraulic controls. The UV reactor also includes additional hardware for controlling UV dose;
typically comprised of (but not limited to): UV intensity sensors, UV transmittance monitors,
ballasts, and control panels.
UV Reactor Validation - a process by which a UV reactor's disinfection performance is
determined relative to operating parameters that can be monitored. Reactors are validated to
indicate that they achieve a certain delivered UV dose for a range of flow, UV intensity, and
water quality conditions (e.g., UV transmittance). Appendix C details the protocol for validating
UV reactors.
UV Transmittance (UVT) - a measure of the fraction of incident light transmitted through the
water column. The UV transmittance is the ratio of the light entering the water to that exiting the
water. The UVT is usually reported for a pathlength of 1 cm. In an alternate pathlength is used,
it should be specified. UVT is often represented as a percentage and is related to the UV
absorbance by the following equation: %UVT = 100 x 10"A. As the UV absorbance increases,
the UV transmittance decreases.
UV Disinfection Guidance Manual xii June 2003
Proposal Draft
-------
List of Acronyms and Abbreviations
A254'.. '
AC
ACGIH
ACS
ANSI
AOC
APHA
ATCC
atm
AWWA
AwwaRF
ultraviolet absorbance at 254 nanometers
alternating current
American Conference of Governmental Industrial Hygienists
Automatic cleaning system
American National Standards Institute
assimilable organic carbon
American Public Health Association
American Type Culture Collection
atmospheres
American Water Works Association
American Water Works Association Research Foundation
BDL
BDOC
°C
CCPP
CF
CFD
CFR
cfii
CIP
cm
CPEL
CSI
CT
CWS
DBP
DBPR
DC
D/DBP
DNA
DOC
DVGW
e
EPA
op
ft
below detectable limits
biodegradable dissolved organic carbon
degrees Centigrade
calcium carbonate precipitation potential
cumulative frequency
computational fluid dynamics
Code of Federal Regulations
colony forming unit •
clean-in-place
centimeter
ceiling level permissible exposure limit
Construction Specifications Institute
residual disinfectant concentration (mg/L) x time (min)
community water system
disinfection byproduct
disinfection byproduct rule
direct current
disinfectants/disinfection by-product
deoxyribonucleic acid
dissolved organic carbon •
Deutsche Vereinigung des Gas- und Wasserfaches (German Association
for Gas and Water)
exponent of the base of the natural logarithm
United States Environmental Protection Agency
degrees Fahrenheit
feet
g •
GAC
gram
granular activated carbon
UV Disinfection Guidance Manual
Proposal Draft
xin
June 2003
-------
List of Acronyms and Abbreviations (Continued)
gal
GFI
gpm
GWR
GWUDI
h
HAA
HOPE
HGL
hp
HPC
HSP
Hz
m
mA
MCL
mg
mgd
min
mJ
mL
mm
'MP
MS2
U€
Urn
gallon
ground fault interrupt
gallons per minute
ground water rule
ground water under the direct influence [of surface water]
hour
haloacetic acid
high-density polyethylene
hydraulic grade line
horsepower
heterotrophic plate count
high service pump
hertz
I UV intensity
IDLH Immediately Dangerous to Life or Health
IDSE initial distribution system evaluation
IESWTR Interim Enhanced Surface Water Treatment Rule
IT UV intensity x time
J joule
kW kilowatt
kW-hr kilowatt-hour
In natural logarithm
LP low pressure
LPHO low pressure high output
LRAA locational running annual average
LSI Langlier Saturation Index
LT1ES WTR Long Term 1 Enhanced Surface Water Treatment Rule
LT2ESWTR Long Term 2 Enhanced Surface Water Treatment Rule
X wavelength
meter
mil Ham p
maximum contaminant level
milligram
million gallons per day
minutes
millijoule
milliliter
millimeter
medium pressure
male specific-2 bacteriophage
microgram
micrometer, micron
UV Disinfection Guidance Manual
Proposal Draft
xiv
June 2003
-------
List of Acronyms and Abbreviations (Continued)
NEL
nm
NIOSH
NIST
NOM
NSF '
NTNCWS
NTU
NWRI
National Electric Code
nanometer
National Institute for Occupational Safety and Health
National Institute of Standards and Technology
natural organic matter
National Science Foundation
non-transient non-community water system
nephelometric turbidity units
National Water Research Institute
O&M
OCC
OMC
ONORM
OSHA
PAC
PEL
%
PER
pfu
pH
PHA
PLC
POE
psi
psig
pve
QA/QC
r
r2
RAA
RCRA
RED
RMS
RNA
rpm
RPZ
s
SARA
SCADA
SDWA
SMCL
SMP
operation and maintenance
offline chemical clean ,
online mechanical clean
Osterreichisches Normungsinstitut (Austrian Standards Institute)
Occupational Safety and Health Administration
powdered activated carbon
permissible exposure limit
percent
preliminary engineering report
plaque forming unit
negative logarithm of the effective hydrogen ion concentration
process hazard analysis
programmable logic controller
point of entry
pounds per square inch
pounds per square inch gauge '
polyvinyl chloride
quality assurance/quality control
radial distance from center
correlation coefficient
running annual average
Resource Conservation and Recovery Act
reduction equivalent dose
root-mean-square
ribonucleic acid
revolutions per minute
reduced pressure zone
second
Superfund Amendments and Reauthorization Act
supervisory control and data acquisition
Safe Drinking Water Act
secondary maximum contaminant level
standard monitoring program
UV Disinfection Guidance Manual
Proposal Draft
xv
June 2003
-------
List of Acronyms and Abbreviations (Continued)
SOP
sss
SUVA
SWTR
T,o
TCLP
TCR
TDH
IDS
THM
TLV
TNTC
TOC
TOX
TSA
TSB
TSS
TTHM
UPS
uv
UVT
VFD
W
WTP
standard operating procedure
system-specific study
specific ultraviolet absorbance
Surface Water Treatment Rule
time at which ten percent of water has passed through the reactor
toxic characteristic leaching procedure
total coliform rule
total dynamic head
total dissolved solids
trihalo methane
threshold limit values
too numerous to count
total organic carbon
total organic halides
tryptic soy agar
tryptic soy broth
total suspended solids
total trihalomethane
uninterruptible power supply
ultraviolet
ultraviolet transmittance
variable frequency drive
watt
water treatment plant
UV Disinfection Guidance Manual
Proposal Draft
xvi
June 2003
-------
Requested Feedback on the
UV Disinfection Guidance Manual
Chapter or Appendix Title
Specific Issues for Comment
Glossary .
1. Are there additional terms that should be defined?
2. Is each definition accurate and clearly presented?
1. Introduction .
1. Does this chapter provide the appropriate amount of information on the relevant
regulations?
2. Overview Of UV Disinfection
1. Is the level of detail appropriate?
2. Is there additional information that should be provided?
3. Planning And Design Aspects For UV Installations
1. Is the overall UV installation design flowchart realistic? Is the chapter organization
reader-friendly?
2. Is the issue of off-specification operation and its implications on the UV installation
design clearly described?
3. Are the recommendations on developing design criteria helpful? Are there other
approaches that should be discussed?
4, Is the power quality information clear? Is more information needed?
5. Are there additional planning or design issues that should be discussed?
4. Overview of Validation
1. Are the elements of validation clearly presented?
2. Is there other information from the detailed validation protocol (Appendix C) that
should be described here?
5. Start-Up And Operation Of UV Installations
1. Are there other elements of the UV installation start-up that should be discussed?
2. Are the organization of the chapter and presentation of information appropriate?
3. Are the operational requirements examples clearly described?
4. Are there other operation and maintenance issues that should be discussed?
• 5. Are the operational challenges described realistic, and are the solutions helpful?
6. References
1. Are there any references that were overlooked that should be added to help clarify
any points made in the UVDGM?
UV Disinfection Guidance Manual xvii June 2003
Proposal Draft
-------
Requested Feedback on the UV Disinfection Guidance Manual
A. Fundamentals of UV Disinfection
1. Is the level of detail appropriate?
2. Is there additional information that should be provided?
B. Derivation of UV Dose-Response Requirements
1. Are there published or unpublished data available that are not included in this
analysis?
C. UV Validation Protocol Testing
1. Is the description of the testing methods clear?
2. Are the distinctions between Tier 1 and 2 clearly described?
3. To provide a better assessment of the RED bias, please provide dose distributions for
UV reactors you have modeled at UVTs of 95, 90, 85, and 80%.
4. Are the Tier 1 criteria acceptable? If not, please provide data and rational to support
alternative criteria.
5. During validation, the uncertainty of some measurements will not be random. In
particular, errors associated with measurements made by the radiometer will likely be
a systematic error (i.e., the radiometer will always read high or read low for the
duration of the validation testing). Other such errors could occur with the intensity
sensors or the reference sensor used to calibrate the duty sensors. Currently, the
UVGM combines these sources of uncertainty with other random sources of
uncertainty to define an expanded uncertainty. Because these sources of error are not
random during a given validation, should the following approach be used:
If the error of a measurement during validation is constant and
systematic, should the uncertainty of the measurement be used to
define a bias error that is applied to the validation results?
Under the current approach, this would apply to the uncertainty of the
radiometer and move it from the expanded uncertainty to its own bias
error. For example, if the uncertainty of the radiometer is 8 %, a
safety factor of 1.08 is added to the RED bias, polychromatic bias, and
expanded uncertainty. This will increase RED targets for Tier 1 and 2.
6. The expanded uncertainty is calculated for an 80 percent confidence interval to ensure
at least nine out often cases of UV system operation meet target dose values. Should
the expanded uncertainty calculation be based on a 90 or 95 percent confidence
interval to ensure a higher percentage of UV systems meet requirements?
D. Validation Microbial Methods
1. The bounds provided for the MS2 and B. subtilis data come from an analysis of data
published in the literature. Should these bounds be used? If not, please provide data to
support using alternative bounds? Should any of the literature data used to develop
these bounds not be included? If yes, please provide a rational for not including that
data.
2. Are the methods for analyzing the collimated beam data and subsequent UV dose-
response curve clearly stated and appropriate? Are there other options that should be
considered?
UV Disinfection Guidance Manual
Proposal Draft
XVlll
June 2003
-------
Requested Feedback on the UV Disinfection Guidance Manual
E. Collimated Beam Apparatus - Measuring Challenge Microbe UV Dose-
Response
1. Is the collimated beam testing description clear?
F. Validation Background
1. The following is an alternate approach for monitoring dose delivery that is not
included in the manual because it has not been applied or referenced.
Calculate the percent UV output from the lamp to the water based on the sensor
readings using the formula:
P, = *100
L S'(UVT)
where
PL = UV output from the lamp to the water (%)
S = Sensor reading
S'(UVT) = Sensor reading expected with a new lamp operating with unfouled
sleeves at a given UVT
UVT = UVT of the water at 254 nm
The calculated lamp power and measured UVT should be above setpoint values
established during validation. The relation S'(UVT) is measured during validation as
opposed to being calculated.
This approach has the following potential benefits:
• No requirement on sensor position
. -' Could measure S'(UVT) with NOM and compare with LSA or coffee as
an experimental approach for reducing the Polychromatic Bias to one.
Should this approach be discussed in the manual?
G. Issues for Unfiltered Systems
1. Is the level of detail appropriate?
2. Is there additional water quality related design or operational concerns for unfiltered
systems that should be addressed?
H. Issues for Ground Water Systems
1. Are there design or operational issues with UV disinfection of groundwater that are
not addressed?
I. Issues for Small Systems
1. Is the level of detail appropriate?
2. Are the design concerns facing small systems adequately addressed?
3. Design information is presented in Chapter 3, and this appendix only includes areas
where small system design differs from the design issues discussed in Chapter 3. Is
this approach effective? -
UV Disinfection Guidance Manual ". xix . June 2003
Proposaf Draft
-------
Requested Feedback on the UV Disinfection Guidance Manual
J. Pilot-Scale and Demonstration-Scale Testing
1. Is the level of detail appropriate? .
2. Were all of the recommended testing methods clearly explained?
3. Are there any other example testing protocols that should be included in this
appendix?
K. Preliminary Engineering Report
1. Are there any elements of this report that would benefit from more detail?
2. Is there any information missing from this report that you would like to see included in a
standard template (i.e., in this example Design Engineering Report)?
L. Regulatory Timeline
1. Is this appendix helpful for UV installation planning?
2. Are the time allocations for the tasks listed in the timeline appropriate?
M. Compliance Forms
1. Are the example compliance forms well organized and easy to complete?
2. Are there other forms that would be helpful to the utility or the State?
N. UV Lamp Breakage Issues
1. Considering available information, are the major issues surrounding lamp breakage
adequately presented in this appendix? Are there additional issues or sources of
information to be discussed?
2. Are there additional methods for the prevention or mitigation of on-line lamp breaks
that should be presented?
O. Case Studies
There are no questions related to this appendix because it is not included in this draft.
P. Validation Protocol Calculator Tool
There are no questions related to this appendix.
UV Disinfection Guidance Manual xx • June 2003
Proposal Draft
-------
1. Introduction
There is growing interest among public water systems in using ultraviolet (UV) light to
disinfect drinking water, based on its ability to inactivate certain microorganisms without
forming harmful disinfection byproducts (DBFs). Some pathogens, such as Cryptosporidium,
are resistant to commonly used disinfectants, whereas UV light has proven effective against these
microorganisms.
The United States Environmental Protection Agency (EPA) is developing the Long Term
2 Enhanced Surface Water Treatment Rule (LT2ESWTR) to further control micfobial •
contamination of drinking water. The rule requires additional treatment for some systems based
on their source water Cryptosporidium concentrations. UV disinfection is one of the options
utilities have to comply with the treatment requirements.
UV light has been widely used to disinfect effluent from wastewater treatment facilities,
particularly those that reuse effluent for irrigation. Until recently, the use of UV treatment for
drinking water applications was primarily limited to small ground water systems, due to the
belief that it was not effective for inactivating protozoa and was not cost-effective for large
systems. In 1998, however, research demonstrated that UV light could effectively inactivate
Cryptosporidium at low dosages (Buhkari et al. 1998), prompting more research to better
understand its potential for widespread application.
UV disinfection design, operation, and maintenance needs differ from those of traditional
chemical disinfectants used in drinking water applications. EPA is therefore developing this
guidance manual to familiarize States' and utilities with these important issues as well as
regulatory requirements. Areas of particular design and operational importance include hydraulic
control, reliability, redundancy, lamp cleaning and replacement, and lamp breakage. Regulatory
requirements are addressed through UV reactor validation, monitoring, and reporting.
1.1 Guidance Manual Objectives
This manual provides guidance to utilities, States, manufacturers, and other interested
parties on the disinfection of drinking water with UV light, including the regulatory requirements
associated with UV disinfection. The LT2ESWTR requirements do not cover all aspects of the
disinfection process. In the areas not directly addressed by the rule, the manual provides
Throughout this document, the terms "State" or "States" are used to refer to all types of primacy agencies,
including U.S. Territories, Indian Tribes, and EPA Regions.
UV Disinfection Guidance Manual . l-l June 2003
Proposal Draft
-------
1. Introduction
recommendations to assist utilities and regulatory agencies in assessing the disinfection
capability and performance of UV installations. The manual's objectives are as follows:
• Provide public water systems and designers with technical information and guidance
on the selection, design, and operation of UV installations and the UV-related
requirements for compliance with the LT2ESWTR.
• Provide States with guidance and the necessary tools to assess UV installations at the
design, start-up, and routine operation phases.
• Provide manufacturers with testing and performance standards for UV components
and systems for treating drinking water.
1.2 Organization
This manual consists of six chapters and appendices:
• Chapter 1 - Introduction. The remainder of this chapter summarizes the LT2ESWTR
and Stage 2 DBPR and discusses regulatory requirements for disinfection of drinking
water with UV light.
• Chapter 2 - Overview of UV Disinfection. This chapter describes the principles of
disinfection with UV light including inactivation mechanisms, dose-response
relationships, water quality impacts, and UV reactors.
• Chapter 3 - Planning and Design Aspects for UV Installations. This chapter
discusses the key design features for UV disinfection facilities and presents some
common approaches to facility design. Key design features include treatment goals,
existing infrastructure, water quality, hydraulics, and operation and control strategies.
• Chapter 4 - Overview of UV Reactor Validation. This chapter describes the •
LT2ESWTR requirements for validating UV reactors and provides an overview of
validation protocol presented in Appendix C.
• Chapter 5 - Start-up and Operation of UV Installations. This chapter discusses
start-up and operation issues of UV disinfection facilities as well as required
monitoring for regulatory compliance.
• Chapter 6 - References. This chapter lists the full references from Chapters 1-5.
UV Disinfection Guidance Manual 1-2 •'• June 2003
Proposal Draft - .
-------
,1. Introduction
The appendices and their titles follow:
Appendix A. Fundamentals of UV Disinfection
Appendix B. Derivation of UV Dose-Response Requirements
Appendix C. Validation of UV Reactors
Appendix D. Microbiological Methods
Appendix E. Collimat^d Beam Apparatus: Measuring Challenge Microorganism
UV Dose-Response
Appendix F. Background to the UV Reactor Validation Protocol
Appendix G. Issues for Unfiltered Systems
Appendix H. Issues for Ground Water Systems
Appendix I. Issues foir Small Systems
Appendix J. Pilot-Scale and Demonstration Scale Testing
Appendix.K. Preliminary Engineering Report
Appendix L, Regulatory Time Line
Appendix M. Compliance Forms
Appendix N. UV Lamp Breakage Issues
Appendix O. Case Studies [This appendix will be included in the final draft at
which time EPA anticipates more information being available.]
Appendix P: Validation Protocol Calculator Tool
1.3 Regulations Summary
c.
This section summarizes the drinking water regulations for microbial and DBF control.
The Stage 2 Disinfectants and Disinfection Byproduct Rule (DBPR) aims to reduce peak DBP
concentrations in the distribution systeW by modifying the Stage 1 DBPR monitoring
requirements and procedures for compliance determination. The LT2ESWTR and Stage 2 DBPR
are to be promulgated together to address the risk-risk trade off between microbial disinfection
and the byproducts formed by commonly used disinfectants. Consequently, when a utility
assesses its disinfection strategy, not only the disinfection of target pathogens is important, but
also the DBP formation from each disinfectant. Table 1.1 summarizes the microbial treatment
requirements and DBP maximum contaminant levels (MCLs) from the Surface Water Treatment
Rule (SWTR), Interim Enhanced Surface Water Treatment Rule (ESWTR), Long Term 1
Enhanced Surface Water Treatment Rule (LT1ESWTR), LT2ESWTR, Stage 1 DBPR, and Stage
2 DBPR.
I;
UV Disinfection Guidance Manual
Proposal Draft
1-3
June 2003
-------
1. Introduction
Table 1.1 Summary of Microbial and Disinfection Byproduct Rules
Surface Water Treatment Rules - Minimum Treatment Requirements
Regulation
SWTR
lESWTR and
LT1ESWTR
LT2ESWTR
Giardia
3 log removal and
inactivation
Vfrus
4 log removal and
inactivation
No change from SWTR
No change from SWTR
Cryptosporidium
Not addressed
2 log removal
0-2. 5 log additional
treatment1
2-3 log treatment2
Disinfection Byproduct Rules - MCLs Based on Run ning Annual Averages (RAAs)
Regulation
Stage 1 DBPR
Stage 2A DBPR3
Stage 2B DBPR4
Trihalomethanes
(TTHM) (ug/L)
80 as RAA
120 as LRAA
80 as LRAA
Haloacetic Acids
(HAAS) (ug/L)
60 as RAA
100 as LRAA
60 as LRAA
Bromate
(ug/L)
10
Chlorite
(ug/L)
1000
No change from Stage 1
No change from Stage 1
'ReqyjrernentforfilteiefLsyjsterns is inadditiqn to removal achieved by conventional treatment comp
lESWTR and LTIEsvvTR. Specific requirements for each plant depend on source wafer momto
Mng with the n
nng^resuits (40
CFR 141.720).
2Unfiltere.d systems must iproyide 2-3 log inactivation; specific requirements for each plant depend on source water
3Stage 2A bases compliance.on a Ipcatjonal.running annual average (LRAA) at the Stage 1 monitoring locations.
Stage 1 KAAs must stillbe met during mis time! Stage 2A oegmsp years after rule promufgatori] for all
systems.
"Stage 2B bases, compliance on an LRMsat revised monitoring locations identified during the Initial Disjribution „
System tvaluation. Stage 2B begins [6 years after rule promulgation] for large systems and |775-B.5 years after
rule promulgation} for small systems dependent on their LT2ESWTR requirements.
1.3.1 Long Term 2 Enhanced Surface Water Treatment Rule
The LT2ESWTR applies to all public water systems that use surface water or ground
water under the direct influence of surface water (GWUDI), except those that purchase all then-
surface and GWUDI water. It builds on the SWTR, lESWTR, and the LTlESWTR by
improving control of microbial pathogens, specifically the contaminant Cryptosporidium. Unlike
the previous rules, the LT2ESWTR bases treatment requirements on a system's source water
Cryptosporidium concentration and type of treatment provided This section describes the rule
requirements for filtered and unfiltered systems.
"UV Disinfection Guidance Manual
Proposal Draft
1-4
June 2003
-------
1. Introduction
1,3.1.1 Filtered Systems
\
The-LT2ESWTR requires systems that use a surface water or GWUDI source (referred to
collectively in this manual as surface water systems) to conduct source water monitoring to
determine average Cryptosporidium concentrations, unless they have historical Cryptosporidium
data equivalent to what is required under the LT2ESWTR (40 CFR 141.701 (a)). Based on its
average source water Cryptosporidium 'concentration, filtered systems will be classified in one of
four possible bins. A system's bin assignment determines the extent of any additional
Cryptosporidium treatment requirements. The rule requires systems to comply with additional
treatment requirements by using one or,more management or treatment techniques from a
toolbox of options (40 CFR 141.720(b». The process is described in more detail below; the full
monitoring requirements are described! in the Source Water Monitoring Guidance Manual for
Public Water Systems for the Long Term 2 Enhanced Surface Water Treatment Rule (USEPA
2003). i:
c
Bin Classification
i
Table 1.2 presents the bin classifications and their corresponding additional treatment
requirements for all filtered systems (40 CFR 141.709 and 40 CFR 141.720). Systems with
average Cryptosporidium concentrations of less than 0.075 oocysts per liter are placed in Bin 1,
for which no additional treatment is required. For concentrations of 0.075 or more, additional
treatment is required on top of that required by existing rules. The additional treatment required
for each bin, specified in terms of log removal, depends on the type of treatment already in place
by the system.
UV Disinfection Guidance Manual 1-5 June 2003
Proposal Draft
-------
1. Introduction
Table 1.2 Bin Requirements for Filtered Systems1
If your
Cryptosporidium
concentration
(oocysts/L) is...
< 0.075
> 0.075 an d< 1.0
>.1.0 and < 3.0
>3.0
Your bin
classification
is...
1
2
3
4
And if you use the following filtration treatment in full
compliance with existing regulations, then your additional
treatment requirements are...
Conventional
Filtration
Treatment
(includes
softening)
No additbnal
treatment
1 log
treatment2
2 log
treatment3
2.5 log
treatment3
Direct
Filtration
No
additional
treatment
1.5 log
treatment2
2.5 tog
treatment3
3 log
treatment3
Slow Sand or
Diatomaceous
Earth
Filtration
No additbnal
treatment
1 log
treatment2
2 log
treatment3
2.5 log
treatment3
Alternative
Filtration
Technologies
No additional
treatment
As determined
by the State2-4
As determined
by the State3'5
As determined
by the State3'6
(40 CFR 141.709 and 40 CFR 141.720)
2 Systems may use any technology or combination of technologies from the microbial toolbox.
3 Systems must achieve at least 1 log of thereauired treatment using ozone, chlorine dioxide, UV disinfection,
memBranes, Dag/cartridge filters; or bankTilrration.
4 Total Cryptosporidium treatment must be at least 4.0 log.
5 Total Cryptosporidium treatment must be at least 5.0 log.
6 Total Cryptosporidium treatment must be at least 5.5 log.
1.3.1.2 Unfiltered Systems
All existing requirements for unfiltered systems under the SWTR (40 CFR 141.71 and
141.72(a)) remain in effect. This includes disinfection to achieve at least 3 log inactivation of
Giardia and 4 log inactivation of viruses and to maintain a disinfectant residual in the
distribution system (e.g., free chlorine or chloramines). The ESWTR and LT1ESWTR did not
change the disinfection requirements for unfiltered systems. The LT2ESWTR requires 2 log or 3
log inactivation of Cryptosporidium, depending on the source water concentration of
Cryptosporidium (40 CFR 141.721(b)).
The arithmetic mean concentration of all Cryptosporidium samples taken is used to
determine the amount of treatment required, as shown in Table 1.3 (40 CFR 141.721 (a)). If the
mean concentration is less than or equal to 0.01 oocysts/L, the system must provide 2 log
inactivation of Cryptosporidium (40 CFR 141.721(b)). If the mean concentration of
Cryptosporidium exceeds 0.01 oocysts/L, the system must provide at least 3 log inactivation of
Cryptosporidium (40 CFR 141.721(b)).
UV Disinfection Guidance Manual
Propo sal Draft
1-6
June 2003
-------
1. Introduction
Table 1.3 Bin Requirements for Unfiltered Systems
Bin
Number
1
2 •
Average Cryptosporidium Concentration
(oocysts/liteir)
<0.01
>0.01 ,
Additional Cryptosporidium inactivatfon
requirements
2 log1
3 log1
Overall disinfection requirements must be met with a minimum of two disinfectants {40 CFR141.721 (d)).
1.3.1.3 UV Disinfection Requirements for Filtered and Unfiltered Systems
';
To receive disinfection credit for a UV reactor, the LT2ESWTR requires utilities to
demonstrate through validation testing:that the reactor, can deliver the required UV dose (40 CFR
141, Subpart W, Appendix D). EPA developed dose requirements for Cryptosporidium,
Giardia, and virus.as presented in Tab]e 1.4 and described in Appendix B of this guidance
manual. These dose requirements account for uncertainty associated with the dose-response of
the microorganisms in controlled experimental conditions. In practical application, other sources
of uncertainty are introduced due to the hydraulic effects of the UV installation, UV reactor
equipment, and monitoring approach (e.g., UV intensity sensors). Therefore, the validation
protocol (described in Chapter 4 and Appendix C of this guidance manual) applies a safety factor
to the Table 1.4 dose requirements to account for these areas of uncertainty and variability.
Table 1.4 UV Dose Requirements Used During Validation Testing1
Cryp tosporidium
Giardia
Virus
" Log Inactivation
0.5
1.6
1.5
39
1.0
2.5
2.1
58
1.5
3.9
3.0
79
2.0
5.8
5.2
100
2.5
8.5
7.7
121
3.0
12 •
11
143
3.5
-
-
163
4.0
-
-
186
1 40CFR141.729(d)
The LT2ESWTR (40 CFR 141.: Subpart W, Appendix D) specifies the following with
respect to reactor validation: j
• Validation testing must determine a range of operating conditions that can be
monitored by the system and under which the reactor delivers the required UV dose.
1;
• Operating conditions must include flowrate, UV intensity, and lamp status, at a
minimum.
UV Disinfection Guidance Manual
Proposal Draft
1-7
June 2003
-------
1. Introduction
• Validated conditions determined by testing must account for UV absorbance of the
water, lamp fouling and aging, measurement uncertainty of on-line UV intensity
sensors, UV dose distributions arising from the velocity profiles through the reactor,
failure of UV lamps or other critical installation components, and inlet and outlet
piping or channel configurations of the UV reactor.
Using the above requirements as a basis, Appendix C provides guidance for several
possible approaches to reactor validation. States may approve modifications to these approaches
or alternative approaches at their discretion.
Monitoring Requirements (40 CFR 141.729fd»
The LT2ESWTR requires utilities to monitor their reactors to demonstrate that they are
operating within the range of conditions that were validated for the required UV dose. At a
minimum, utilities must monitor each reactor for flowrate, lamp outage, UV intensity as
measured by a UV intensity sensor, and any other parameters required by the State. UV
absorbance should also be measured where it used in a dose control strategy. Systems must
check the calibration of UV intensity sensors and must recalibrate sensors in accordance with a
protocol approved by the State. The LT2ESWTR does not specify monitoring frequency (section
5.4 of this guidance describes the monitoring requirements with recommended frequencies).
Reporting Requirements (40 CFR 141.7301
The LT2ESWTR requires utilities to report the following items:
* Initial reporting - Validation test results demonstrating operating conditions that
achieve the UV dose required for the inactivation credit desired for compliance with
the LT2ESWTR.
• Routine reporting - Volume of water entering the distribution system that was not
treated by the UV reactors operating under validated conditions on a monthly basis.
For the purposes of this guidance manual, when a UV reactor is operating outside of its
validated limits, it is considered "off-specification."
Additional Requirement for Unfiltered Systems (40 CFR 141.721(cK2Yt
For unfiltered systems using UV disinfection to meet the LT2ESWTR requirements, the
required Cryptosporidium log inactivation by UV disinfection must be achieved in at least 95
percent of the water delivered to the public during each calendar month.
UV Disinfection Guidance Manual 1-8 June.2003
Propo sal Draft
-------
1. Introduction
1.3.2 Stage2DBPR
i
The requirements of the Stage 2 DBPR will apply to all community water systems
(CWSs) and nontransient noncommumty water systems (NTNCWSs) — both ground and surface
water systems — that add a disinfectant other than UV light, or that deliver water that has been
treated with a disinfectant other than UV light.
Initial Distribution System Evaluations i
The Stage 2 DBPR is designed.to reduce DBF occurrence peaks in the distribution system
by changing compliance monitoring requirements. Compliance monitoring will be preceded by
an initial distribution system evaluation (IDSE) to identify compliance monitoring locations that
represent high TTHM and HAAS levels. The IDSE consists of either a standard monitoring
program (SMP) or a system-specific study (SSS). NTNCWSs serving fewer than 10,000 people
are not required to perform an IDSE, and other systems may receive waivers from the IDSE
requirement.
Compliance Determination and Schedule
The Stage 2 DBPR changes the way sampling results are averaged to determine
compliance. The determination for the Stage 2 DBPR is based on a LRAA (i.e., compliance
must be met at each monitoring location) instead of the system-wide RAA used under the Stage
1 DBPR.
The Stage 2 DBPR will be implemented in two phases, Stage 2A and Stage 2B. Under
Stage 2A, all systems must comply with TTHM/HAA5 MCLs of 120/100 ng/L measured as
LRAAs at each Stage 1 DBPR monitoring site, while continuing to comply with the Stage 1
DBPR MCLs of 80/60 ug/L measured as RAAs. Under Stage 2B, systems must comply with
TTHM/HAA5 MCLs of 80/60 ug/L at locations identified under the IDSE.
Significant Excursion Evaluations
Because Stage 2 DBPR MCL cpmpliance is based on an annual average of DBP
measurements, a system could from time to time have DBP levels significantly higher than the
MCL (referred to as a significant excursion) while still being in compliance. This is because the
high concentration could be averaged with lower concentrations at a given location. If a
significant excursion occurs, a system must conduct a significant excursion evaluation and
discuss the evaluation with the State no; later than the next sanitary survey.
.
1 .4 Alternative Approaches for, Disinfecting with UV Light
This manual provides technical information about using UV disinfection for drinking
water treatment. Although it covers many aspects of implementing aUV installation, from
design and validation to operation, it is not comprehensive in terms of all types of UV
installations, design alternatives, and validation protocols that may provide satisfactory
__ i
UV Disinfection Guidance Manual , 1-9 June 2003
Proposal Draft
-------
1. Introduction
performance. For example, pulsed UV and eximer lamps are two types of UV technologies not
included in this manual, but they may provide effective disinfection. Currently, a significant
level of research is being conducted surrounding UV disinfection and its applications in various
industries. As more information becomes available, other UV equipment or methods of
operation, design, and validation will evolve. States may recognize alternatives in UV installation
design, operation, and validation that are not described in this manual.
UV Disinfection Guidance Manual
Proposal Draft
1-10
June 2003
-------
2. Overview of UV Disinfection
Chapter 2 provides an overview of UV disinfection. The material ranges from an
explanation of the process in terms of basic chemical and physical principles to a description of
the components of a UV installation and performance monitoring. Appendix A, Fundamentals
of UV Disinfection, serves as a companion to this chapter by providing more detailed
information on each of the topics discussed. The corresponding appendix sections are noted
throughout the text. The organization of this chapter is presented below, including the key
question each section addresses.
• What are the fundamental characteristics of UV light, and what
happens to UV light as it propagates through water? Section 2.2
• How does UV light inactivate microorganisms? Section 2.3.1
• Can microorganisms undergo repair and become infectious
. after inactivation by UV light? Section 2.3.2
• How are UV dose and microbial response
determined? .;; .....:. Sections 2.3.3 and 2,3.4
•> *
« How does UV dose vary in1 a UV reactor? .'......Section 2.3.3
• • '•
What affects a microorganism's response to
UV light? ..? Sections 2.3.4 and 2.3.5
• What do UV reactors look like and how do the key
components function? I Section 2.4
. What are the differences between low pressure and medium
pressure lamps? ; Section 2.4.2
. How do the utility and the istate know the UV reactor is
delivering the required UV'dose? Section 2.4.9
. How does water quality affect UV reactor performance? Section 2.5.1
. Do any disinfection byproducts form as a result of UV
disinfection? '. : Section 2.5.2
2.1 History of UV Light for Drinking Water Disinfection
UV disinfection is an established technology supported by decades of fundamental and
applied research and practice in North America and Europe. Downes and Blunt (1887)
discovered the germicidal properties of sunlight. The development of mercury lamps as artificial
UV light sources in 1901 and the use of quartz as a UV transmitting material in 1906 was soon
UV Disinfection Guidance Manual
Proposal Draft .
2-1
June 2003
-------
2. Overview of UV Disinfection
followed by the first drinking water disinfection application in Marseilles, France in 1910. In
1929, Gates identified a link between UV disinfection and absorption of UV light by nucleic
acid. The development of the fluorescent lamp in the 1930s led to the production of germicidal
tubular lamps. Considerable research on the mechanisms of UV disinfection and the inactivation
of microorganisms occurred during the 1950s (Dulbecco 1950; Kelner 1950; Powell 1959;
Brandt and Giese 1956).
While there was substantial research on UV disinfection during the first half of the 20th
century, the low cost of chlorine and operational problems with early UV disinfection systems
limited the growth of UV disinfection as a drinking water treatment technology. The first
reliable applications of UV light for disinfecting municipal drinking water occurred in
Switzerland and Austria in 1955 (Kruithof and van der Leer 1990). By 1985, the number of
installations in these countries had risen to approximately 500 and 600, respectively. With the
discovery of chlorinated disinfection byproducts (DBFs), UV disinfection became popular in
Norway and the Netherlands with the first installations occurring in 1975 and 1980, respectively.
As of 1996, there were over 2000 UV disinfection systems treating drinking water in
Europe (USEPA 1996), primarily treating flows less than 1 million gallons per day (MOD). A
survey conducted in 2000 found that UV disinfection is currently being used to treat larger flows,
including two installations treating a combined flow of 76 MOD in Helsinki, Finland (Toivanen
2000), and that the number of installations is increasing (USEPA 2000). Several large
installations across the United States are currently under design. Because of the susceptibility of
Cryptosporidium to UV disinfection and the emphasis in recent regulations on controlling
Cryptosporidium, the number of utilities using UV disinfection is expected to increase
significantly over the next decade.
2.2 Fundamental Aspects of UV Light
The use of UV light to disinfect drinking water involves (1) the generation of UV light
with the desired germicidal properties and (2) the delivery (or transmission) of that light to
pathogens. This section provides a basic description of how UV light is generated and the
environmental conditions that affect its delivery to pathogens.
2.2.1 Nature of UV tight
UV light is the region of the electromagnetic spectrum that lies between x-rays and
visible light (Figure 2.1). The UV spectrum is divided into four regions as shown in Figure 2.1:
vacuum UV (100 to 200 nm), UV-C (200 to 280 nm), UV-B (280 to 315 nm), and UV-A (315 to
400 nm) (Meulemans 1986). UV disinfection occurs due to the germicidal action of UV-B and
UV-C with microorganisms. The germicidal action of UV-A is small relative to UV-B and
UV-C and therefore needs very long exposure times to be effective as a disinfectant. Light in the
vacuum UV range is very effective in disinfecting microorganisms (Munakata et al. 1991).
However, it is impractical for water disinfection applications because it rapidly attenuates over
very short distances in water. For the purposes of this manual, the practical germicidal
wavelength for UV light ranges between 200 and 300 nm.
UV Disinfection Guidance Manual 2-2 June 2003
Proposal Draft
-------
2. Overview of UV Disinfection
Figure 2.1 UV Light in the Electromagnetic Spectrum
100 nm 400 nm
Gamma
Rays
X-ray
UV
Visible
Infrared
254 nm
Vacuum UV
UV-C
2M
UV-B
nm
31!
UV-A
nm
100 nm
200 nm
300 nm
400 nm
Typically, UV light is generated by applying a voltage across a gas mixture, resulting in a
discharge of photons. The specific wavelengths of light emitted from photon discharge depend
on the elemental composition of the gas and the power level of the lamp (section A. 1.1). Nearly
all UV lamps designed for water treatment use a gas mixture containing mercury vapor.
Mercury is an advantageous gas for UV disinfection applications because it emits light in the
germicidal wavelength range, as discussed in section 2.3.5. The light output depends on the
concentration of mercury atoms, which is directly related to the mercury vapor pressure.
Mercury at low. vapor pressure (near vacuum; 0.001 to 0.01 torr, 2 x 10"5 to 2 x 10"3 psi) and
moderate temperature (40 °C) produces essentially monochromatic UV light at 253.7 nm. At
higher vapor pressures (100 to 10,000 torr, 2 to 200 psi) and higher operating temperatures (600
to 900 °C), the frequency of collisions between mercury atoms increases, producing UV light
over a broad spectrum (polychromatic) with an overall higher intensity. Mercury vapor pressure
between 0.01 and 100 torr does not efficiently produce UV light.
2.2.2 Propagation of UV Light
As UV light propagates from its source, it interacts with the materials it encounters
through absorption, reflection, refraction, arid scattering. In disinfection applications, these
phenomena result from interactions between the emitted UV light and UV reactor components
(i.e., lamp envelopes, lamp sleeves, and reactor walls) and also the water being treated. When
assessing water quality, UV absorbance or UV transmittance is the parameter that incorporates
the impact of absorption and scattering. This section briefly describes both the phenomena that
influence light propagation and measurement techniques to quantify UV light propagation. More
detailed information is provided in sections A.I.2.1 through A.I.2.5.
Absorption is the transformation of light to other forms of energy as it passes through a
substance. UV absorption of a substance will vary with the wavelength of the light. The
components of the reactor and the water passing through the reactor all absorb UV light to
varying degrees, depending on their material composition. When UV light is absorbed, it is no
longer available to disinfect microorganisms. .
UV Disinfection Guidance Manual 2-3 • • . June 2003
Proposal Draft
-------
2. Overview of UV Disinfection
Unlike absorption, the phenomena of refraction, reflection, and scattering change the
direction of UV light, but the UV light is still available to disinfect microorganisms.
Refraction (Figure 2.2) is the change in the direction of light propagation as it passes
from one medium to another. In UV reactors, refraction occurs when light passes from the UV
lamp through an air gap, through the lamp sleeve, and through the water. These changes alter the
angle that UV light strikes target pathogens.
Figure 2.2 Refraction of Light
Quartz
Sleeve
Incident
from UVLonp
Water
Refracted Light
Reflection is the change in direction of light propagation when it is deflected by a surface
(Figure 2:3). Reflection may be classified as specular or diffuse. Specular reflection occurs
from smooth polished surfaces and follows the Law of Reflection (the angle of incidence is equal
to the angle of reflection). Diffuse reflection occurs from rough surfaces and scatters light in all
directions with little dependence on the incident angle. In UV reactors, reflection will take place
at interfaces that do not transmit UV light (e.g., the reactor wall) and also at UV transmitting
interfaces (e.g., the inside of a lamp sleeve). The type of reflection observed and intensity of
light reflected from a surface depends on the material of the surface.
Figure 2.3 Reflection of Light
Incident Light
Reflected Light
Incident Light
Light
Specular Reflect ion
Diffuse Reflection
UV Disinfection Guidance Manual
Proposal Draft
2-4
June 2003
-------
2. Overview of UV Disinfection
Scattering of light is the change in direction of light propagation caused by interaction
with a particle (Figure 2.4). Particles can cause scattering in all directions, including towards the
incident light source (back-scattering). Scattering of light caused by particles smaller than the
wavelength of the light is called Rayleigh scattering (section A.I .2.4). Particles larger than the
wavelength of light scatter more light in the forward direction but also cause some
backscattering. Rayleigh scattering depends inversely on wavelength to the fourth power (1/X4)
and thus is more prominent at shorter wavelengths. Scattering by particles larger that the
wavelength of the light is relatively independent of wavelength.
Figure 2.4 Scattering of Light
Back
Scattered
Light
Incident
Light
90" Scattered Light
Target
Pathogens
Forward
Scattered
Light
UV absorbance (Ais-O is a commonly used water quality parameter that characterizes the
decrease in the amount of incident light as it passes through a water sample over a specified
distance or pathlength. Various procedures call for filtering the sample through a 0.45 urn
membrane before measuring the absorbance. If the measurement is made according to a
modified version of Standard Method 591 OB (APHA et al. 1998), the water sample is not pH
adjusted or filtered. Since most particles in drinking water are strong absorbers of UV light, it is
recommended that absorbance measurements be made without filtering the sample. Therefore,
the modified measurement accounts for scattering and some absorption from particles in the
water sample that may interfere with UV disinfection. Although Standard Methods identifies
this measurement as UV absorption, this manual will refer to it as absorbance since the latter
term is widely used in the water treatment industry.
The term UV transmittance (UVT) has also been used extensively in the literature when
describing the behavior of UV light. UVT is the percentage of light passing through a water
sample over a specified distance and is related to UV absorbance by Equation 2.1:
Equation 2.1
where
UVT =
A254 =
UV transmittance at specified wavelength (e.g., 254 nm) and pathlength (e.g., 1
cm)
UV absorbance at specified wavelength, based on 1 cm pathlength (unitless; UV
absorption as measured by Standard Method 591 OB)
UV Disinfection Guidance Manual
Proposal Draft
2-5
June 2003
-------
2. Overview of UV Disinfection
2.3 Microbia! Response to UV Light
The mechanism of disinfection by UV light differs considerably from chemical
disinfectants such as chlorine and ozone. Chemical disinfectants inactivate microorganisms by
destroying or damaging cellular structures, interfering with metabolism, and hindering
biosynthesis and growth (Snowball and Hornsey 1988). UV light inactivates microorganisms by
damaging their nucleic acid, thereby preventing the microorganism from replicating. A
microorganism that cannot replicate cannot infect a host.
When studying UV disinfection effectiveness, it is important to use microbial assays that
measure infectivity, not viability. Until recently, viability assays such as excystation and vital
dyes were used to determine inactivation. However, these assays do not evaluate changes in the
ability of a microorganism to reproduce and infest a host. The importance of using assays that
measure inactivation is highlighted by the history of UV disinfection for Cryptosporidium. It
was believed that UV disinfection was not effective for Cryptosporidium inactivation because
results of early Cryptosporidium inactivation studies were based on viability assays. The ability
of UV light to inactivate Cryptosporidium at low doses was revealed when infectivity was
assessed by inoculating mice with UV treated water, which showed greater than 4-log
inactivation of Cryptosporidium at doses less than 20 mJ/cm2 (Bukhari et al. 1999).
This section discusses the damage that causes microbial inactivation, the ability of
microorganisms to repair the damage, methods for determining microbial inactivation, and how
wavelength of UV light affects inactivation.
2.3.1 Mechanisms of Microbial Enactivation by UV Light
UV light inactivates microorganisms by damaging deoxyribonucleic acid (DNA) or .
ribonucleic acid (RNA), thereby interfering with replication of the microorganism (section
A.2.2). In normal DNA replication, the double helix strand separates allowing the single strands
to serve as a template for reconstructing the opposite strand of nucleotides: adenine bonds to
thymine and guanine bonds to cytosine (Figure 2.5).
Figure 2.5 Structure of DNA and Nucleotide Sequences Within DNA
Hydrogen Bonded
Nitrogenous
Base Pairs (A, T,G,C)
Sugar-
Phosphate
Backbone
DNA STRUCTURE
-A-T-G-C-G-A-T-C-
III I I I I I
-T-A-C- G-C-T-A-G-
Purines
A= Adenine
G = Guanine
DNA SEQUENCE
Pyrimidines
T = Thymine
C= Cytosine
UV Disinfection Guidance Manual
Proposal Draft
2-6
June 2003
-------
2. Overview of UV Disinfection
Light that is absorbed by a system can induce a chemical reaction. As shown in
Figure 2.6, each of the nucleotides absorbs UV light from 200 to 300 nm (section A.2.2). The
UV absorption of DNA results from the combination of nucleotides and has a peak near 260 nm
and a local minimum near 230 nm. DNA absorbs light in the wavelength range emitted by UV
lamps, enabling photobiological effects that lead to nucleic acid damage.
Figure 2.6 UV Absorbance of Nucleotides (left) and Nucleic Acid (right) at pH 7
(adapted from Jagger 1967)
1.0n
83T0.8
C (B
| (5) 0.6
jjO.4
0.0
Cytosine
8 £
c n
1.0
0.8
0.6
0.4
0.0
DNA
200 220 240 260 280 300
Wavelength (nm)
200 220 240 260 280 300
Wavelength (nm)
Damage to nucleic acid does not prevent the cell from undergoing metabolism and other
cell functions. Although the microbial cell is alive after exposure to UV light, it cannot
reproduce, and therefore it is incapable of infecting a host. To kill the microbial cell, the UV
dose would need to be increased by orders of magnitude as compared to the UV dose needed to
prevent replication.
Variations in DNA content cause microorganisms to absorb UV light differently, thereby
contributing to the differences in microorganism susceptibility to UV disinfection. There can be
significant disparity in the susceptibility of different strains of bacteria and viruses to UV
disinfection (section A.2.7). Among the pathogens of interest in drinking water, viruses are most
resistant to UV disinfection followed by bacteria and Cryptosporidiunt oocysts and Giardia
cysts. Appendix B provides statistical evaluations for dose-response data of Giardia cysts,
Cryptosporidium oocysts, and viruses, and Chapter 1 contains the regulatory requirements for
inactivating these pathogens.
2.3.2 Microbial Repair
Because microorganisms that have been exposed to UV light still retain metabolic
functions, some are able to repair the damage done by UV light to a limited degree as described
in section A.2.3. In some cases, the microorganism regains infectivity. These microorganisms
have evolved enzyme-mediated mechanisms for reversing-UV damage. Repair of UV light-
induced DNA damage includes photoreactivation and dark repair (Knudson 1985). In
photoreactivation (or photorepair), enzymes energized by exposure to light between 310 and 490
UV Disinfection Guidance Manual
Proposal Draft
2-7
June 2003
-------
2. Overview of UV Disinfection
nm (near and in the visible range) repair damaged sections of DNA. Photoreactivation needs the
presence of reactivating light. Dark repair is defined as when a repair process does not need
reactivating light. The term is somewhat misleading because dark repair can occur in the
presence of light, and therefore does not need dark conditions. Excision repair, a form of dark
repair, is an enzyme-mediated process where the damaged section of DNA is removed and
regenerated using the existing complimentary strand of DNA.
Knudson (1985) found that bacteria are able to repair in light and dark conditions,
suggesting that bacteria may have the enzymes necessary for photorepair and dark repair. Viral
DNA lacks the necessary enzymes for repair, but can repair using the enzymes of a host cell
(Rauth 1965). Linden et al. (2002a) did not observe photoreactivation or dark repair ofGiardia
at UV doses typical for UV disinfection applications (16 and 40 mJ/cm2). However, unpublished
data from the same study show Giardia reactivation in light and dark conditions at very low UV
doses (0.5 mJ/cm2; Linden 2002). Shin et al. (2001) reported Cryptosporidium does not regain
infectivity after inactivation by UV light. One study has shown that Cryptosporidium contains
the capability to undergo some DNA repair (Oguma et al. 2001). However, even though the
DNA is repaired, infectivity is not restored.
Knudson (1985) demonstrated that photorepair can be overcome by increasing the
damage to the DNA through higher UV doses. However, it is unknown if higher UV doses can
reduce dark repair because it is more difficult to study experimentally. Research is continuing to
evaluate this phenomenon. At the doses typically used in UV disinfection, microbial repair can
be controlled and accounted for as discussed in section 3.1.1.
2.3.3 UV Dose and Dose Distribution
UV dose is a measurement of the energy per unit area that is incident on a surface. UV
dose is the product of the average intensity acting on a microorganism from all directions and the
exposure time. Units commonly used for UV dose are J/m2, mJ/cm2, and mWs/cm2
(10 J/m2 = 1 mJ/cm2 = 1 mWs/cm2) with mJ/cm2 being the most common units in North America
and J/m2 being the most common in Europe.
In a batch system such as a bench scale collimated beam test (described in Appendix E),
the average intensity is determined mathematically. For collimated beam tests using a low-
pressure lamp, the UV intensity measured by a radiometer, the UV absorbance of the water, the
thickness of the water layer, the distribution of light across the water surface, and the reflection
and refraction of light from the water surface all are considered in calculating the average
intensity. The UV dose can be determined in a batch system by multiplying the calculated
average intensity by the specific exposure time.
When using polychromatic light sources (e.g., medium-pressure lamps), UV dose
calculations in batch, bench scale experiments also incorporate the same parameters as a low-
pressure lamp collimated beam test. In addition, the intensity at each wavelength in the
germicidal range and the germicidal effectiveness at the associated UV wavelengths are also
considered because microorganisms absorb different amounts of UV light at different
wavelengths. The UV dose-response measured with polychromatic lamps will match the UV
UV Disinfection Guidance Manual
Proposal Draft
2-8
June 2003
-------
2. Overview of UV Disinfection
dose-response of monochromatic lamps when the UV dose delivered by the polychromatic
source is properly calculated (Cabaj et al. 2001; section A.2.4.1).
Dose delivery in a continuous-flow UV reactor is subject to hydrodynamic irregularities
and a variable UV intensity distribution and is a function of the UV absorbance of the water, the
flowrate through the reactor, the UV output from the lamps, and the hydraulic characteristics
within the reactor. As such, it is difficult to calculate directly UV dose within a UV reactor. If
the reactor has plug flow with complete mixing perpendicular to that flow, all microorganisms
leaving the reactor receive the same dose, and the reactor would be termed an "ideal" reactor.
However, these ideal conditions do not generally do not exist in continuous-flow UV reactors.
'As such, microorganisms passing through a UV reactor are exposed to different doses. The
difference in UV doses experienced by microorganisms in a flowing reactor is best characterized
by a dose distribution.
A dose distribution is the probability distribution of UV doses that microorganisms
receive in a flow-through UV reactor; typically shown as a histogram (Figure 2.7). Some
microorganisms travel close to the UV lamps and experience a higher dose while others that
travel close to the reactor walls may experience a lower dose. Some microorganisms move
through the reactor quickly while others travel a more circuitous path. A narrow dose
distribution (Figure 2.7a) indicates more ideal hydrodynamic conditions. A wider distribution
(Figure 2.7b) indicates less efficient-reactor performance and results in a greater degree of
"overdosing" to ensure that the minimum desired dose is achieved for the microorganisms at the
lower end of the dose distribution.
Figure 2.7 Hypothetical Dose Distributions for Two
Reactors with Differing Hydraulics
*
JO
8
I
I
o
0.6
0.5-
0.4-
0.3-
0.2-
0.1-
a. Narrow Dose
Distribution
(Better
Hydraulic
Conditions)
15 30 45 75 90
UV Dose (mJ/cm2)
15 30 45 75 90
UV Dose (mJ/crrf)
There are currently no methods to measure directly the dose distribution in a continuous
flow UV reactor, but mathematical models can help to characterize dose distribution. Therefore,
the UV dose in a UV reactor is estimated as the reduction equivalent dose (RED). The RED is a
calculated dose for a flow through UV reactor that is based on biodosimetry (i.e., measuring the
level of inactivation of a challenge microorganism with a known UV dose-response). The RED
is set equal to the UV dose in a collimated beam test that achieves the same level of inactivation
UV Disinfection Guidance Manual 2-9 June 2003
Proposal Draft
-------
2. Overview of UV Disinfection
of the challenge microorganism as measured for the flow-through UV reactor during
biodosimetry testing. Methods for collimated beam testing and biodosimetry are in Appendix E
section 4.2, respectively.
2.3.4 Microbial Response (UV Dose-Response)
The response of microorganisms to UV light is calculated by determining the
concentration of infectious microorganisms before and after exposure to a measured UV dose
and applying Equation 2.2.
N
Log Inactivation = log,0 —- Equation 2.2
N - •
Where
No = Concentration of infectious microorganisms before exposure to UV light
N = Concentration of infectious microorganisms after exposure to UV light
UV dose-response relationships can be expressed as either the proportion of
microorganisms inactivated (log inactivation, results in a dose-response curve with a positive
slope) or the proportion of microorganisms remaining (log survival, results in a dose-response
curve with a negative slope) as a function of UV dose. The proportion of microorganisms
remaining and the log inactivation are typically shown on a logarithmic (base 10) scale, while the
UV dose is typically shown on a linear scale. This manual will present microbial response as log
inactivation since the terminology is widely accepted in the industry. Therefore, all dose-
response curves presented will have a positive slope.
Although several approaches may be used to measure microbial dose-response, the
bench-scale collimated beam test has evolved as the customary method because it has carefully
controlled conditions, allowing for accurate and repeatable determination of UV dose. Accurate
determination of UV dose is beneficial for developing meaningful relationships between UV
dose and microbial response.
Figure 2.8 presents examples of UV dose-response curves. In general, the UV dose-
response of disperse microorganisms follows first order inactivation (Figure 2.8, E. coli curve;
section A.2.5.1). However, some microorganisms are slower to respond, producing a shoulder at
low UV doses followed by near-linear inactivation (Figure 2.8, B. subtilis curve; section
A.2.5.2). UV dose-response is generally independent of how the germicidal UV light is
produced (i.e., low-pressure or medium-pressure UV light), UV absorbance, temperature, and
PH.
UV Disinfection Guidance Manual 2-10 * June 2003
Proposal Draft
-------
2. Overview of UV Disinfection
Figure 2.8 Shapes of UV Dose-Response Curves
(adapted from Chang et al. 1985)
O E co//
D 6. subtilis spores
v Total coliform-wastewater
X Rotavirus
20
40 60 80
UVDose (mJ/cm2)
100
UV dose-response is affected by particle-association and clumping of microorganisms.
Solids present in wastewater samples can cause a tailing or flattening of the dose-response curve
at higher inactivation levels (Figure 2.8, total coliform curve; section A.2.5.3) because clumping
or particle association shields a fraction of the microorganisms from UV light. In these
wastewater experiments, the microorganisms are present in the treated water at very high
concentrations so that any particle association with turbidity reflects the impact of upstream
treatment processes.
Research by Linden et al. (2002b) indicated that the UV dose-response of
microorganisms added to filtered drinking waters is not altered by variation in turbidity that
meets regulatory requirements (40 CFR 141.73). For unfiltered waters, Passantino and Malley
(2001) found that source water turbidity up to 10 NTU does not impact the UV dose-response of
separately added (seeded) microorganisms. In these experiments, however, microorganisms
were added to waters containing various levels of treated or natural turbidity. Therefore, it was
not possible to examine microorganisms associated directly with particles in their natural or
treated states. Consequently, these drinking water studies can only suggest the impact of
turbidity on dose-response as it relates to the impact of UV light scattering by particles, rather
than particle-association or clumping of microorganisms.
2.3.5 Microbial Spectral Response
The action spectrum (also called UV action) of a microorganism is a measure of
inactivation effectiveness as a function of wavelength. Figure 2.9 illustrates the UV action for
three microbial species and also the UV absorbance of DNA as a function of wavelength.
Because of the similarity between UV action and DNA absorbance, and because DNA
absorbance is easier to measure than UV action, the DNA absorbance spectrum of a
microorganism is often used as a surrogate for its UV action spectrum. The scale of the y-axis
UV Disinfection Guidance Manual
Proposal Draft
2-11
June 2003
-------
2. Overview of UV Disinfection
represents the ratio of inactivation effectiveness at a given wavelength to the inactivation
effectiveness at 254 nm. For most microorganisms, the UV action peaks at or near 260 nm, has a
local minimum near 230 nm, and drops to zero near 300 nm. Although the sensitivity of the
organism often increases below 230 nm, the strong absorption of UV light by components in
natural water at these wavelengths offsets the increased organism sensitivity in this region.
Nevertheless, an operating definition of the effective germicidal range for UV light in water
includes wavelengths from 200 to 300 nm.
Figure 2.9 Comparison of Microbial UV Action and DMA UV Absorbance
(adapted from Rauth 1965 and Linden et al. 2001)
2.0 1
DMA
o - Cryptosporidium
O- • MS2
"• ~ Herpes simplex virus
200 220 240 260 280 300
Wavelength (nm)
2.4 UV Reactors
The goal in designing UV reactors for drinking water disinfection is to deliver efficiently
the necessary dose to inactivate pathogenic microorganisms. An example UV reactor is shown
in Figure 2.10. Commercial UV reactors consist of open or closed-channel vessels containing
UV lamps, lamp sleeves, UV intensity sensors, lamp sleeve wipers, and temperature sensors.
UV lamps are housed within the lamp sleeves, which protect and insulate the lamps. Some
reactors include automatic cleaning mechanisms to keep the lamp sleeves free of deposits that
may form due to contact with the water. UV intensity sensors, flow meters, and in some cases,
UVT monitors are used to monitor dose delivery by the reactor. This section briefly describes
UV reactor components. A more detailed discussion of these components is provided in section
A.3.
UV Disinfection Guidance Manual
Proposal Draft
2-12
June 2003
-------
2. Overview of UV Disinfection
Figure 2.10 UV Disinfection System Schematic
(courtesy of Severn Trent Services)
Reactor
Casing
Temperature
Sensor
UV Lamp Housed in
Quartz Slee
Effluent
Pipe
Quartz Sleeve
Wiper
Wiper
Motor
Influent
Pipe'
. Electrical
UV Intensity Connection
Sensor to Lamp
Control
UV Panel
Transmittance
Monitor
2.4.1 Reactor Configuration
UV reactors are typically classified as either open or closed channel. Water flows under
pressure (i.e., no free surface) in closed channel reactors (Figure 2.1 la). Drinking water UV
applications have used only closed reactors to-date. Open channel reactors (Figure 2.1 Ib) are
open basins with channels containing racks of UV lamps. Open channel reactors are most
commonly used in wastewater applications.
UV Disinfection Guidance Manual
Proposal Draft
2-13
June 2003
-------
2. Overview of UV Disinfection
Figure 2.11 Example of Closed (a) and Open (b) Channel Reactors
(courtesy of Trojan Technologies)
a. Closed-Channel Reactor
b. Open-Channel Reactor
Reactors are designed to optimize dose delivery, and the reactor hydrodynamics play an
important role in design. Lamp placement, inlet and outlet conditions, and baffles all affect
mixing within a reactor. Improvements to the hydraulic behavior of a reactor are often obtained
at the expense of headloss. Individual reactor designs employ various methods to optimize dose
delivery (e.g., higher lamp output versus lower lamp output and improved hydrodynamics
through increased headloss).
2.4.2 UV Lamps
UV light can be produced by the following variety of lamps:
• Low-pressure (LP) mercury vapor lamps
• Low-pressure high-output (LPHO) mercury vapor lamps
. Medium-pressure (MP) mercury vapor lamps
. Electrode-less mercury vapor lamps
. Metal halide lamps
. Xenon lamps (pulsed UV)
• Eximer lamps
. UV lasers
Full-scale drinking water applications generally use LP, LPHO, or MP lamps. As such,
the subsequent discussions in this manual are limited to these UV lamp technologies. Table 2.1
UV Disinfection Guidance Manual
Proposal Draft
2-14
June 2003
-------
2. Overview of UV Disinfection
lists characteristics associated with these lamps, and Table 2.2 lists operational advantages and
disadvantages of the lamp types.
Table 2.1 Mercury Vapor Lamp Characteristics
Parameter
Germicidal UV light
Mercury Vapor Pressure (torr)
Operating Temperature (°C)
Electrical Input (W/cm)
Germicidal UV Output (W/cm)
Electrical to Germicidal UV
Conversion Efficiency (%)
Arc length (cm)
Relative Number of Lamps
Needed for a Given Dose
Lifetime (hrs)
Low-pressure
Monochromatic at
254 nm
Optimal at 0.007
Optimal at 40
0.5
0.2 •
.35-38
10-150
High
8,000-10,000
Low-pressure
high-output
Monochromatic
at 254 nm
0.76
130-200
1.5-10
0.5-3.5
30-40
10-150
Intermediate
8,000-12,000
Medium-pressure
Polychromatic, including
germicidal range
(200 to 300 nm)
300-30,000
600 - 900
.50-250
5-30
10-20
5-120
Low
4,000-8,000
Table 2.2 Mercury Vapor Lamp Comparison
Comparative
Advantages
Comparative
Disadvantages
Low-pressure
• Higher germicidal efficiency; nearly all
output at 254 nm
• Smaller power draw per lamp (less
reduction in dose if lamp falls)
« Longer lamp life
• More lamps needed for a given
application
. Larger footprint
Medium-pressure
. Higher power output
• Fewer lamps for a given application
• Smaller reactors
• Smaller footprint
• Higher operating temperature can
accelerate fouling (section 2.5.1)
• Shorter lamp life
• Lower electrical to germicidal UV
conversion efficiency
The light emitted by LP and LPHO lamps is essentially monochromatic at 253.7 nm
(Figure 2.12a) and is near the maximum of the microbial action spectrum. MP lamps emit at a
wide range of wavelengths across the action spectra (Figure 2.12b). Therefore, LPHO lamps
convert power to germicidal light more efficiently. In either lamp type, power not converted to
light is primarily lost as heat.
UV Disinfection Guidance Manual
Proposal Draft
2-15
June 2003
-------
2. Overview of UV Disinfection
Figure 2.12 UV Output of LP (a) and MP (b) Mercury Vapor Lamps
(Sharpless and Linden 2001)
Lamp Output Relative to
Maximum Output in Range
3 o o p o r* r
3 Ko A b> bo o K
a. Low Pressure Lamp
-~ — —
200 250 300
Wavelength
Lamp Output Relative to
Maximum Output in Range
5 O O O O -»
s ro *.
-------
2. Overview of UV Disinfection
Figure 2.13 UV Lamp Output and its Relation to the UV Absorbance of DNA
(courtesy of Bolton Photosciences, Inc.)
SO)
D>
&
13
O E
11
32
200
250
Wavelength (nm)
300
i
UV lamps may be oriented parallel, perpendicular, or diagonal to flow or ground.
Orienting MP lamps horizontally relative to the ground prevents differential heating of the lamps
and reduces the potential for lamp breakage. Lamp breakage is discussed farther in Appendix N.
UV lamps degrade as they age resulting in a reduction in output (section A.3.1.6). MP
lamps may have a shift in spectral output as well. Lamp degradation will impact dose delivery
over time.
2.4.3 Lamp Power Supply And Ballasts
Ballasts supply the UV lamps with the appropriate power to energize and operate the UV
lamps. Ballasts use inductance (coil or transformer), capacitance, and a starting circuit. Power
supplies and ballasts are available in many different configurations and are tailored to a unique
lamp type and application. UV reactors may use electronic ballasts, magnetic ballasts, or
transformers. The various ballast types and their differences are detailed in section A.3.2.
2.4.4 Lamp Sleeves
UV lamps are housed within lamp sleeves to help keep the lamp at optimal operating
temperature and to protect the lamp from breaking. Lamp sleeves are tubes of quartz (or vitreous
silica). The sleeve length is sufficient to include the lamp and associated electrical connections.
The sleeve diameter is typically 2.5 cm for LP lamps and 5 to 10 cm for MP lamps. The distance
between the exterior of the lamp and interior of the lamp sleeve is approximately 1 cm. Sleeve
walls are typically 2 to 3 mm thick and absorb some UV light (Figure 2.14). UV lamps are
usually centered radially within lamp sleeves using spacers.
UV Disinfection Guidance Manual
Proposal Draft
2-17
June 2003
-------
2. Overview of UV Disinfection
Figure 2.14 UV Transmittance of Quartz that is 1 mm Thick at a Zero Degree
Incidence Angle (GE Quartz 2001)
o
o
c
fl
A:
E
S
£
3
100 -I
90-
80-
70-
60-
^^^
^r
/
'
200 220 240 260 280 300 320 340 360 380 400
Wavelength (nm)
Lamp sleeves can fracture and foul,'and their transmittance will decrease as they age.
Fractures can occur from internal stress and external mechanical forces such as wiper jams, -
water hammer, resonant vibration, and impact by objects. Microscopic fractures may also occur
if lamp sleeves are not handled properly when removed for manual cleaning. If the sleeve
fractures while in service, water can enter the sleeve, making the lamp vulnerable to breakage as
a result of temperature differences between the lamp and the water. Lamp breakage is
undesirable due to potential for mercury release. Appendix N discusses the potential effects of
lamp breakage and possible response plans.
Fouling on the internal lamp sleeve surface arises from the deposition of material from
components within the lamp or sleeve due to temperature and exposure to UV light. The UV
reactor manufacturer can control internal lamp sleeve fouling through appropriate material
selection. Fouling on external surfaces is caused by the reaction of compounds in the water with
the lamp sleeve surface. Compounds that contribute to fouling are discussed in section 2.5.1.
External fouling must be removed by cleaning. In addition, exposure of quartz contaminated
with metal cations can cause solarization as lamp sleeves age. Both fouling and solarization can
decrease the UV transmittance of the sleeve.
2.4.5 Cleaning Systems
UV reactor manufacturers have developed different approaches for cleaning lamp
sleeves, depending on the application. These approaches include both off-line chemical cleaning
(OCC) and on-line mechanical cleaning (OMC) methods.
In OCC systems, the reactor is shut down, drained, and flushed with a cleaning solution.
Solutions used to clean lamp sleeves include citric acid, phosphoric acid, or a food grade
UV Disinfection Guidance Manual " 2-18 . June 2003
Proposal Draft
-------
2. Overview of UV Disinfection
proprietary solution provided by the UV reactor manufacturer. The reactor is rinsed and returned
to operation after sufficient time to dissolve the substances fouling the sleeves is allowed. LPHO
systems typically use OCC systems.
OMC systems are built-in UV reactor components that consist of wipers that are driven
by either screws attached to electric motors or pneumatic pistons. There are two types of wipers
used in OMC systems: mechanical wipers and physical-chemical wipers. Mechanical wipers
may consist of stainless steel brush collars or Teflon® rings that move along the lamp sleeve
(Figure 2.15a). Physical-chemical wipers have a collar filled with cleaning solution that moves
along the lamp sleeve (Figure 2.15b). The wiper physically removes fouling on the lamp sleeve
surface while the cleaning solution within the collar dissolves fouling materials. The use of
mechanical and physical-chemical wipers does not necessitate that the UV reactor be drained.
Therefore, the reactor can remain on-line while the lamp sleeves are cleaned. MP systems
typically use OMC systems because the higher lamp temperatures can accelerate fouling under
certain water qualities.
Figure 2.15 (a) Mechanical Wiper System (courtesy of Calgon Carbon
Corporation), (b) Physical-Chemical Wiper System
(courtesy of Trojan Technologies)
2.4.6 UV Intensity Sensors
UV intensity sensors are photosensitive detectors that measure the UV intensity at a point
within the UV reactor (Figure 2.16). Sensors are used to indicate dose delivery by providing
information related to UV intensity at different points in the reactor. The measurement responds
to changes in lamp output due to lamp power setting, lamp aging, lamp sleeve aging, and lamp
sleeve fouling. Depending on sensor position, UV intensity sensors may also respond to changes
in UV absorbance of the water being treated (section A.3.8.2). UV intensity sensors are
composed of optical components, a photodetector, an amplifier, a housing, and an electrical
connector. The optical components may include monitoring windows, light pipes, diffusers,
apertures, and filters. Monitoring windows and light pipes are designed to deliver light to the
UV Disinfection Guidance Manual
Proposal Draft
2-19
June 2003
-------
2. Overview of UV Disinfection
photodetector. DiffUsers and apertures are designed to reduce the amount of UV light reaching
the photodetector, thereby reducing sensor degradation that is caused by UV energy. .Optical
filters are used to modify the spectral response such that the sensor only responds to germicidal
wavelengths (i.e., 200 to 300 nm). At the time of publication, sensors are specific to each
manufacturer and are subject to validation as described in sections 4.3.2.3 and C.4.7.
Figure 2.16 UV Intensity Sensor Viewing Lamps within a UV Reactor
(courtesy of Severn Trent Services)
UV intensity sensors can be classified as wet or dry. Dry sensors monitor UV light
through a monitoring window, whereas wet UV intensity sensors are in direct contact with the
water flowing through the reactor. Monitoring windows and the wetted ends of wet sensors can
foul over time and need cleaning similar to lamp sleeves.
2.4.7 UV Transmittance Monitors
As stated previously, UVT is an important parameter in determining the efficiency of UV
disinfection. Therefore, monitoring UV transmittance (or UV absorbance to calculate UVT) is
critical to ensure the success of a UV disinfection application. UVT can be determined either
through grab samples with a laboratory instrument or on-line. Several commercial UV reactors
use the measurement of UVT to help monitor and control the calculated UV dose in the reactor.
In general, commercial on-line UVT monitors calculate UVT by measuring the UV
intensity at various distances from a lamp. One such monitor is schematically displayed in
Figure 2.17. In this monitor, a stream of water passes through a cavity containing a LP lamp
with three UV intensity sensors located at various distances from the lamp. The difference in
sensor readings is used to calculate UVT.
UV Disinfection Guidance Manual
Proposal Draft
2-20
June 2003
-------
2. Overview of UV Disinfection
Figure 2.17 UV Transmittance Monitor Design
(courtesy of Severn Trent Services)
UV Intensity
Sensor
Inlet
2.4.8 Temperature Sensors
Energy input per unit volume is relatively high for a UV reactor. The water flowing.
through a reactor efficiently absorbs the wasted heat and maintains operating temperatures within
a desirable range. Nevertheless, temperatures can become elevated under the following
circumstances: . ; .
. Water level in the reactor drops and lamps are exposed to air.
« Water stops flowing in the reactor.
UV reactors are equipped with temperature sensors that monitor the water temperature
within the reactor. If the temperature is above the recommended operating temperature range,
the reactor will shut off to minimize the potential for the lamps overheating.
2.4.9 Monitoring UV Disinfection Performance
The performance of an operating UV disinfection system must be monitored to
demonstrate that adequate disinfection is being achieved (40 CFR 141, Subpart W, Appendix D).
Because the concentration of pathogenic organisms cannot be measured continuously in the UV-
treated water and the dose distribution cannot be measured directly in real time, various
strategies have been developed to monitor dose delivery. Any dose monitoring method must be
evaluated during reactor validation (as described in section 4.3.2.2), and the outputs measured
during validation will be part of the monitoring requirements described in section 5.4.1 (40 CFR
141.729(d)).
UV Disinfection Guidance Manual
Proposal Draft
2-21
June 2003
-------
2. Overview of UV Disinfection
Currently, there are three fundamental approaches to monitor UV disinfection
performance in a UV reactor:
1. UV Intensity Setpoint Approach. In this approach, measurements made by the UV
intensity sensor are used to control the UV reactor. The UV intensity sensor is
located in a position that allows it to properly respond to both changes in UV
intensity output of the Jamps and also UVT of the water. The UV intensity sensor
output and the flowrate are used to monitor dose delivery. The setpoiht value for UV
intensity over a range of flowrates is determined during validation.
2. UV Intensity and UVT Setpoint Approach. This approach is similar to the UV
intensity sensor setpoint approach, except that the UV sensor is placed close to the
lamp such that it only responds to changes in UV lamp output. UVT is monitored
separately. For a specific flowrate, the UV intensity and UVT measurements are used
to monitor dose delivery. The setpoints for UV intensity and UVT over a range of
flowrates are determined during validation.
3. Calculated UV Dose Approach. In this approach, the UV intensity sensor is placed
close to the lamp, which is similar to the UV intensity and UVT setpoint approach.
Flowrate, UVT, and UV intensity are all monitored, and the outputs are used to
calculate UV dose via a validated computational algorithm developed by the UV
reactor manufacturer.
The strategy for dose monitoring depends on the manufacturer and may be proprietary.
Dose monitoring recommendations are discussed in section 5.4.2. •
2.5 Water Quality Impacts and Byproduct Formation
Constituents in the water subjected to treatment affect the performance of UV
disinfection. In addition, all disinfectants can form byproducts, and the goal of the overall
disinfection process is to maximize disinfection while minimizing byproduct formation. This
section discusses water quality characteristics impacting UV disinfection performance and
finishes with a discussion of byproducts formed during the UV disinfection process.
2.5.1 Water Quality Impacts -
UVT, particle content, and constituents that foul lamp sleeves and other wetted
components are the most significant water quality factors impacting UV disinfection
effectiveness. In spite of these effects, the impact of water quality on dose delivery can be
adequately addressed in virtually all drinking water applications if carefully considered during
the design of the UV disinfection system, as discussed in section 3.1.3.1.
The most important water quality parameter affecting reactor performance is UVT. As'
UVT decreases, the intensity throughout the reactor decreases for a given lamp configuration.
This results in a reduction in UV dose delivered to the microorganism and the measured UV
UV Disinfection Guidance Manual
Proposal Draft
2-22
June 2003
-------
2. Overview of UV Disinfection
intensity for a given lamp output. Section 3.1.3.1 discusses how to incorporate the impact of
UVT into UV disinfection system design.
Several chemicals used in water treatment processes can decrease the UVT of water (e.g.,
Fe"1"3 and ozone). However, some oxidants (including ozone) can increase the UVT (APHA et al.
1998) by degrading natural organic matter. Water treatment processes upstream of the UV
reactors can be operated to control UVT, thereby optimizing the design and costs of the UV
reactor (section A.4.1.3 and section 3.1.3.1).
Particle content can also impact UV disinfection performance. Particles may scatter light
and reduce the UV intensity delivered to the microorganisms. Particles may also shield
microorganisms from UV light, effectively reducing disinfection performance.
Compounds in the water can cause fouling in a UV reactor on the external surfaces of the
lamp sleeves and other wetted components (e.g., monitoring windows of UV intensity sensors).
Fouling on the lamp sleeves reduces the transmittance of UV light through the sleeve into the
water, thereby reducing power efficiency. Fouling on the monitoring windows impacts UV
intensity and dose monitoring. Hardness, alkalinity, temperature, iron concentration, and pH all
influence the rate of fouling and, subsequently, the frequency of sleeve cleaning. The following
compounds can cause fouling:
• Compounds whose solubility decreases as temperature increases will precipitate (e.g.,
CaCO3, CaSO4, MgCO3, MgSO4, FePO4, FeCO3, A12(SO4)3). These compounds will
foul MP lamps faster than LP lamps due to differences in operating temperature.
. Compounds with low solubility will precipitate (e.g., Fe(OH)3, A1(OH)3).
. Particles will deposit on the lamp sleeve surface due to gravity settling and
turbulence-induced collisions (Lin et al. 1999a).
Fouling rate kinetics have been reported as first order over time following a short
induction period (Lin et al. 1999b). Depending on the water quality and UV lamp type,
significant fouling may occur in hours or take up to several months. Although there is currently
not sufficient information to predict fouling based on water quality, a facility can use the
Langelier Saturation Index (LSI) or the Calcium Carbonate Precipitation Potential (CCPP) as a
tool to determine if precipitation is likely to occur (section A.4.1.4). Data have been generated
from pilot-scale testing on waters of low to moderate hardness and iron content (Mackey et al.
2001 and Mackey et al. 2003). At total and calcium hardness levels less than 140 mg/L and iron
less than 0.1 mg/L, standard cleaning protocols and wiper frequencies (one sweep every 15
minutes to an hour) were sufficient to overcome the impact of sleeve fouling at all'sites tested.
At site's with high hardness or iron in the feed water, it may be advantageous to evaluate fouling
rates as described in section J.5.1 on a site-specific or worst case basis via pilot-scale or
demonstration-scale testing to identify how best to keep the lamp sleeves clean.
Table 2.3 is a summary of water quality data and the fouling observed for various pilot
and full-scale UV reactors. All of the MP systems shown had mechanical cleaning (except at
Boxalls Lane), and the LPHO systems used manual chemical cleaning. The fouling observed at
individual sources is reported as shown in the following list:
UV Disinfection Guidance Manual 2-23 June 2003
Proposal Draft
-------
2. Overview of UV Disinfection
Not Significant - no significant drop in UV intensity (based on UV intensity sensor
readings)
Moderate - slight decrease in UV intensity and slight scale observed on sleeves
Significant - large decrease in UV intensity and significant deposits observed on
sleeves
Table 2.3 Water Quality Data and Fouling Observed for UV Disinfection
Pilot and Demonstration Studies
Name of Plant
Location
Lamp Type
A254 (cm'1)
LSI
Iron (mg/L)
Manganese (mg/L)
Calcium Hardness
(mg/L as CaCO3)
Hardness
(mg/L as CaCO3)
Alkalinity
(mg/L as CaCO3)
PH
Fouling Observed
Boxalls
Lane1
Hampshire,
UK
MP
NA
NA
NA
NA
NA
325-370
260-280
7.1-7.2
not
significant
Atlanta2
Atlanta,
GA
MP/LPHO
0.01-0.04
NA
<0.04
<0.015
NA
21.5
13.7
6.6
not
significant4
moderate5
Ulrich
Water
Treatment
Plant2
Austin, TX
MP/LPHO
0.03-0.08
NA
0.01
<0.001
40
101
60
9.6
not
significant
Central
Utah2
Orem, UT
MP/LPHO
0.01-0.04
0.5
<0.02
<5.03
162
180
159
7.8
moderate6
Neenah
Water
Utility2
Neenah,
Wl
MP/LPHO/
LP
0.03-0.10
0.7
0.02
0.003
54
87
52
9
not
significant
Cudahy
Water
Utility2
Cudahy,
Wl
MP/LPHO/
LP
0.00-0.03
-0.1
0.01
0.012
80
138
125
7.7
not
significant .
Bourgine et al. 1995
2Mackeyetal. 2001
3 Detection Limit
4 Cleaning wipers on (MP system)
5 Cleaning wipers off (MP system)
6 After 8 months of operation (LPHO system)
NA = Not available
None of the systems studied and listed in Table 2.3 exhibited "significant" fouling, and in
all cases, the observed fouling was controllable by regular system maintenance and cleaning.
Lastly, algae may grow upstream or downstream of UV reactors. Visible light emitted
from the lamps is transmitted through water at further distances than germicidal wavelengths.
UV Disinfection Guidance Manual
Proposal Draft
2-24
June 2003
-------
2. Overview of UV Disinfection
Depending on the concentration of nutrients in the water and the amount of visible light
transmitted beyond the reactor, algae growth may need to be controlled through periodic
maintenance.
2.5.2 Byproducts from UV Disinfection
UV DBFs arise either directly through photochemical reactions or indirectly through
reactions with products of photochemical reactions (section A.4.2). Photochemical reactions will
only take place if a chemical species absorbs UV light; and the resulting excited state reacts to
form a new species. The resulting concentration of new species will depend on the concentration
ofthereactantsandtheUVdose. .
In drinking water, research has focused on the impact of UV light on the formation of
halogenated DBFs after subsequent chlorinatidn and the transformation of organic material to
more degradable components. For ground water and filtered drinking water, UV disinfection at
typical doses has been shown not to impact the formation of trihalomethanes or haloacetic acids,
two categories of DBFs currently regulated by the United States Environmental Protection
Agency (EPA) (Malley et al. 1995; Kashinkunti et a!. 2003).
Several studies have shown low-level formation of non-regulated DBFs (e.g., aldehydes)
as a result of applying UV light to wastewater and raw drinking water sources. However, a study
performed with Filtered drinking water indicates no significant change in aldehydes, carboxylic
acids, or total organic halides (TOX) (Kashinkunti et al. 2003). The difference in results can be
attributed to the difference in water quality, most notably the higher concentration of organic
material in raw waters and wastewaters.
Finally, the conversion of nitrate to nitrite is possible with MP lamps that emit at low
wavelengths (von Sonntag and Schuchman 1992). However, due to the low conversion rate
(about 1 percent; Sharpless and Linden 2001), this is of minimal concern in drinking water
applications.
UV Disinfection Guidance Manual 2-25 June 2003
Proposal Draft ' ,
-------
3. Planning and Design Aspects
for UV Installations
This chapter discusses the key planning and design features for UV installations. The
planning section helps identify the parameters and constraints to be considered prior to design of
the UV installation, and the design section presents factors that should be considered during
detailed design. ,
The focus of Chapter 3 is UV disinfection implementation issues, not the determination
of whether UV disinfection is the most appropriate technology. Throughout Chapter 3, it is
assumed, unless otherwise stated, that the water to be disinfected is filtered water meeting
applicable regulatory requirements. Appendices G, H, and I provide additional information on
unfiltered, ground water, and small systems, respectively. The planning and detailed design for
any UV installation is site-specific. Given the wide range of treatment scenarios that are
possible, a document of this nature cannot address or anticipate all possible treatment conditions.
The information presented here should be used within the context of sound engineering judgment
as it can be applied on a case-by-case basis. In addition, this Guidance Manual was written with
the understanding that UV technology will continue to expand and evolve.
The organization of this chapter is presented below by the question that each section
addresses.
• What are the goals of the UV installation? Section 3.1.1
. What are the potential installation locations?.. Section 3.1.2
. What design parameters need to be defined? Section 3.1.3
• How does the UV reactor selection affect design? Section 3.1.4
. What are the options for validation? Section 3.1.4.3
. How should potential installation locations be evaluated? Section 3.1.6
. What are the existing hydraulic conditions and UV installation
hydraulic needs? Section 3.1.6.1
. What should be considered when estimating the process
footprint of the UV installation? Section 3.1.6.2
« How can the installation options be evaluated?.... Section 3.1.7
• What are the options for UV reactor procurement? Section 3.2
• What are the options for addressing hydraulic constraints and
what are the critical hydraulic system components? Section 3.3.1
UV Disinfection Guidance Manual 3-1 ' ' June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
. How does the control strategy influence the design of the
process instrumentation and control for the UV installation? ."...Section 3.3.2
. What are the elements in the process instrumentation and
control system? Section 3.3.3
« What are the necessary electric power arrangements? Section 3.3.4
. What elements need to be considered for the UV installation
layout? .....Section 3.3.5
. What information should the equipment specification include? Section 3.3.6
. What are the necessary drawings and specifications for the UV
installation? Section 3.3.7
» What should be reported to the State and when? Section 3.4
The process of planning and designing a UV installation is presented as a flowchart in
Figure 3.1. In the United States to date, the majority of the utilities undertaking the construction
of UV installations have pre-purchased the UV reactors prior to design. Therefore, the design
flowchart is based on the pre-purchase of the UV reactors and the use of a traditional design-bid-
build approach for the project. Chapter 3 is generally organized to follow the flowchart. UV
installations can be successfully constructed using any of the equipment procurement and
' contractor selection approaches currently used within the industry. It is the utility's and
engineer's responsibility to select the most appropriate project approach. Whatever approach is
utilized, the planning and design components discussed in Chapter 3 should be addressed even
though the actual order of completion may vary.
UV Disinfection Guidance Manual . , 3-2 June 2003
Proposal Draft
-------
1
1
_o
2
to
o
1
O
O
C
a
c
'w
O)
c
_n
0.
,0
(0
o
CO
I
O)
ative UV
Identify aJtern
•2
I
constraints
evaluating:
* . Hydraulic
c
.1
1
D
Si
Define th
i
1 i
1 *
Hi
ii -i
t
luate
rational and
g u
> ex
II] o
'
5
Evaluate equi|
validation
4
•a
1 I
Evaluate pc
UV equipm
•
I
i i
s. .
J
Ib
£-1
•» c .'
it:
A
C
'-C U
^ s ^
« "2 :
•B § J
>
Section 3. 1.6
A
> l/^
5 — ;
3 •
rt **i
3 ^
'
a c
I *
^ J!
'
^
Section 3, 1
1
1 1
e£
* *
1
f*i
2 c/>
•
1
mi£Vl
microorg
ri
0
en
~
Section 3
ons and costs
facility
Compare opti
and select UV
*
Section 3.1. 7
B §
•5 »
II 1
< •
1
-* c « '
S ">
ll 1
o; < 0}
hydraulics
Design systerr
Section 3.3.1
1
|
•o
1 ^
Finalize bye
Section 3.3
1
§•£ «
Ii i
& s &
9
s 1
•8f »
S1 2
Is §
" w 'S
Do. co
, ,
3
O
s-
c
S i-
8 «
"° m
1 1
"iZ t/)
, ^
51
V at
^ W
•e 'S c ^*
III «
If! 1
Q « c o g t5 w
Q S M O fr "?
3
^w
'o. §
£ '1
u «
1
0 J> ^
1 1 "1
1
o
E5
w
Q
-s
^5
"a
§
Jta
2S
B-'?
S-J6
•¥ (9
s s
bo e
U
U
.2 xz
*5 4i
si
l|
!l
b ts
•8?
II
O T3
I!
5.2
e 2
-------
3. Planning and Design Aspects for UV Installations
3.1 UV Installations Planning
The planning process for a UV installation is similar to the process that would be
employed for any retrofit; upgrade, or new construction project at a water treatment plant (WTP).
In the planning phase, it is important to identify alternatives and define criteria needed to select
the appropriate application and to facilitate detailed design. For a UV installation, this includes
the following steps:
• Defining disinfection goals
. Identifying potential locations for UV disinfection
• Defining design parameters-
. Evaluating potential UV reactors
. Evaluating control strategies
. Evaluating hydraulic factors and process footprint
• Preparing preliminary costs and selecting an installation option
This section provides planning guidance for each of these steps with a focus on specific
elements that should be considered for UV disinfection.
3.1.1 Defining UV Disinfection Goals
A comprehensive disinfection strategy provides multiple barriers to reduce microbial risk
while minimizing disinfection byproduct (DBF) formation. UV disinfection is a tool that can
contribute to a comprehensive disinfection strategy by providing a cost-effective method of
inactivating target pathogens that are more resistant to more traditional disinfection methods.
The specific objectives of a given UV installation should be clearly defined during the planning
stages. This can ensure that the design meets the utility's and the State's expectations based on
the regulatory requirements, target microorganism(s), and the overall disinfection strategy.
Chapter 1 presents the regulatory requirements that must be met for the overall water treatment
process and specific requirements for UV disinfection.
The' UV doses necessary for Cryptosporidium and Giardia inactivation are lower than
that those needed to inactivate viruses. Accordingly, the capital costs for inactivating
Cryptosporidium and Giardia should be lower. One study estimated capital costs for
Cryptosporidium and Giardia inactivation by UV disinfection to be approximately 50 percent
lower than the costs associated with the UV inactivation of viruses (Cotton et al. 2002).
Therefore, the target microorganism and inactivation level should be determined early in the
planning process.
Repair of UV light-induced damage is discussed in section 2.3.2. As discussed
previously, repair has not been observed in Cryptosporidium and viruses, and Giardia only
UV Disinfection Guidance Manual 3-4 June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
exhibited repair when exposed to very low UV doses (0.5 mJ/cm2). Therefore, repair of UV-
induced damage of Cryptosporidium, Giardia, and viruses do not need to be considered in the
UV installation design. However, bacteria have been shown to repair of UV damage. The
residual disinfectant concentration (either chlorine or chloramines) in the distribution system will
most likely prevent repair of UV damage in bacteria. Therefore, microbial repair of bacteria also
does not affect UV installation design.
To a degree, UV disinfection can replace chemicals used to disinfect chlorine-resistant
pathogens (e.g., Cryptosporidium and Giardia), thereby reducing DBF formation. However, UV
disinfection is not as efficient at inactivating viruses as more traditional, chlorine-based
disinfection processes. Because of its effectiveness at treating viruses and the need to maintain a
disinfectant residual in the distribution system, some chlorine-based disinfectant (chlorine or .
chloramines) will be needed even if UV disinfection is implemented. Also, chemicals that serve
as disinfectants may be added in the treatment process to oxidize other constituents present in the
water (e.g., iron, manganese, or taste and odor causing compounds). Utilities that currently add a
chemical disinfectant prior to the location of a future UV installation and plan to curtail the use
of such chemicals, following implementation of UV disinfection should assess the effect that a
reduction in pre-oxidant use may have on water quality at the point of UV application.
Therefore, a utility considering a change in disinfection strategy should evaluate all water quality
goals to ensure they are met and must prepare a disinfection benchmark as discussed in
Chapter 1.
3.1.2 Identifying Potential Locations for UV Installations
It is strongly recommended that the UV disinfection process be placed after filtration.
Although UV disinfection can potentially be applied anywhere along the treatment train from the
raw water intake to after high-service pumping, there are significant drawbacks to placing the
UV installation upstream of filtration in conventional WTPs. Prior to filtration, UV absorbance
at 254 nm (Aas4) is higher (UV transmittance (UVT) is lower) due to higher concentrations of
natural organic matter, turbidity, and particles. Coagulation can enmesh microorganisms in floes
and may block the UV light from reaching the microorganisms, which affects the UV dose-
response of the microorganism. In addition, Long Term 2 Enhanced Surface Water Treatment
Rule (LT2ESWTR) UV dose requirements apply only to post-filter and unfiltered supplies that
meet the criteria for filtration avoidance (40 CFR 141.729 (d)). Therefore, this section focuses
on the post-filtration use of UV disinfection.
This section presents the general post-filtration locations that may be considered for the
UV installation. For a location to be feasible, the UV installation hydraulic needs should be met
(section 3.1.6.1) and the equipment must physically fit in the proposed location. Hydraulic
profiles and preliminary drawings should be developed for each location under consideration to
address these controlling criteria. Also, LT2ESWTR requires that all UV reactors be validated
(40 CFR 141.729 (d)), and the validation protocol (Chapter 4) recommends specific piping
configurations for both validation testing and UV installation. These recommendations for inlet
and outlet conditions can affect the feasibility of the potential locations. Detail on the
recommended inlet and outlet hydraulics for both validation and installation is given in
section 3.1.4.3.
UV Disinfection Guidance Manual 3-5 June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
3.1.2.1
Combined Filter Effluent Installation (Upstream of Clearwell)
A combined filter effluent installation is defined here as the application of UV
disinfection to the filter effluent after it has been combined (as opposed to individual filters) and
prior to the clearwell as shown in Figure 3.2. This installation is typically in a separate building.
Of the three options described, the combined filter installation is generally preferred when
conditions permit.
Figure 3.2 Schematic for UV Installation Upstream of Clearwell
Source
Water
Rapid
Mix
•
Flocculation
-»-
ji i~:~fc!»V:
-^
/v
Sedimentation
Basin
Filters
UV
Disinfection
Clearwell
To
Distribution
System
There are several advantages to this type of design and installation:
• The UV reactor operation is largely independent of the operation of individual filters,
which provides flexibility for design and operation.
. If the entire UV installation failed, a WTP could still provide disinfection by adding a
chemical disinfectant to the clearwell. (Note that backup chemical disinfection may
not provide Cryptosporidium inactivation.)
• Surge and pressure issues that are concerns with UV reactors installed immediately
downstream or upstream of high service pumps (HSPs) are usually not an issue for
this installation location.
. Because the UV installation will typically be constructed in a new building for this
installation location, there may be greater flexibility in maintaining the recommended
inlet and outlet hydraulic conditions for the UV reactors (section 3.1.6.1).
The primary disadvantages of this type of installation are that an additional building may
be necessary and that piping and fittings may result in higher headless than alternative
configurations.
3.1.2.2
Individual Filter Effluent Piping Installation
Individual filter effluent piping installations are defined here as installations with UV
reactors dedicated to each individual filter effluent pipe. The installation is typically within the
existing filter gallery. Figure 3.3 schematically represents this type of installation. The main
UV Disinfection Guidance Manual 3-6 June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
advantage of this installation is that a new building would not be necessary, which may lower
construction costs.
Figure 3.3 Schematic of Individual Filter
Effluent Piping Installation in Filter Gallery
Source
Water
Rapid
Mix
FEocculation
Sedimentation
Basin
HSPs
To
Cleanwll Distribution
System
However, there are several disadvantages to this installation location. Many filter
galleries do not have sufficient space within existing effluent piping to accommodate a UV
reactor. The existing piping may also put constraints on how the UV reactor is validated because
of the unique inlet and outlet conditions that may be present (section 3.1.6.1). In addition to
accommodating the UV reactors, there needs to be sufficient space in the filter gallery or a
nearby area for the control panels and electrical equipment. Access to existing equipment may
be impaired by the UV reactor, and access to UV reactor components for maintenance may be
more restricted than for a combined filter effluent installation. Also, the environmental
conditions (e.g., moisture) in the filter gallery may not be appropriate for the installation of the
UV reactors, associated control panels, and electrical equipment without improvements to the
heating, ventilating, and air conditioning system for the area.
The in-line installation may also complicate treatment plant operations and limit
operational flexibility as described below.
• In general, this option results in an increased number of UV reactors compared to a
combined filter installation because the number of filters dictates the number of UV
reactors. This may increase operation and maintenance costs in comparison to the
combined filter effluent installation where the number of UV reactors is determined
by the design flow, water quality constraints, UV reactor capacity, and redundancy
needs.
• The increased headloss of the UV reactors may affect the operation of the filters and
the clearwell.
UV Disinfection Guidance Manual
Proposal Draft
3-7
June 2003
-------
3. Planning and Design Aspects for UV Installations
With one UV reactor for each filter, the operation of each filter will be dependent on
the reliable operation of each UV reactor and vice versa.
The UV reactor operation during a filter backwash can complicate UV reactor
operations. The lamp cooling need to addressed if it.remains energized during a
backwash because the lamps should not be energized in stagnant water or air. If a
UV reactor is off during a backwash, the flow during the UV reactor warm-up
(section 3.1.3.3) is off-specification, which may cause problems with exceeding off-
specification requirements and recommendations (section 3.1.3).
3.12.3
UV Disinfection Downstream of the Clearwell
It may be possible for a WTP to build the UV installation after the clearwell either
upstream or downstream of the high service pumps (Figure13.4). In many WTPs, the HSPs take
water directly from the clearwell, limiting space and the availability of suitable piping for
installation of the UV installation upstream of the HSPs. installation downstream of the HSPs
may provide greater space and flexibility in locating the UV reactors. Either configuration may
be advantageous if there is insufficient space or head to allow installation of the UV reactors
between the filters and the clearwell; however, there are significant disadvantages to these
options.
Figure 3.4 UV Disinfection Downstream of High Service Pumps
Source
Water
Rapid
Mix
Flocculation
Sedimentation
Basin
Filters
Clearwll
UV
Disinfection
To
Distribution
System
UV installations located downstream of the clearwell will experience greater fluctuations
in flowrate since actual flowrates are more closely matched to system demand changes. This
may increase the UV reactor size or more UV reactors to accommodate the flow fluctuations.
In post-HSP installations, the water will be at distribution system pressure. The UV
reactor housing may need to be reinforced because of these high pressures, which would increase
the cost of the UV reactors. In addition, these locations are more prone to water hammer
because of their proximity to the HSPs and subsequent high pressures, which could lead to sleeve
damage. If a lamp sleeve is damaged, the enclosed lamp may break, releasing mercury into the
water. Hydropneumatic tanks or pressure relief valves may be necessary in this installation
location to avoid water hammer. This issue is discussed in more detail in section 3.1.6.1 and
section N.2.1.3.
UV Disinfection Guidance Manual
Proposal Draft
3-8
June 2003
-------
3. Planning and Design Aspects for UV Installations
A UV reactor located after the HSPs will reduce the discharge pressure to the distribution
system, and a UV installation located between the clearwell and HSPs will reduce the suction
head available for the pumps. As a result, discharge pressures and storage utilization could be
impacted at these two locations.
In summary, UV installations downstream of the clearwell are not recommended because
of the increased potential for adverse pressure conditions within the UV reactor and the increased
reliability and size considerations. In general, these installations should only be considered if the
combined filter effluent and in-line filter effluent locations are not feasible.
3.1.3 Defining Design Parameters
Water quality, lamp fouling/aging factor, flowrate, and power quality affect the sizing of
the UV reactors and associated support facilities. These design parameters need to be
determined to ensure compliance with LT2ESWTR requirements.
UV reactors are required to be validated by LT2ESWTR to demonstrate the UV
installation achieves the required UV dose (40 CFR 141.729(d)). Validation testing establishes
the conditions under which the UV reactors must be operated to ensure the required dose
delivery (40 CFR 141.729(d)). Off-specification is defined as a UV reactor that is operating
outside of its validated limits. (For example, the UV reactor is operating with a flowrate that is
higher than the UV reactor was validated.)
To the extent practical, UV reactors should be designed with process monitoring and
control components (e.g., alarms, shut-off valves) to prevent water from entering the distribution
system when a UV reactor is operating outside of validated conditions. Unfiltered systems that
use UV disinfection to meet the Cryptosporidium treatment requirement of the LT2ESWTR must
demonstrate that at least 95 percent of the water delivered to the public during each month is
treated by UV reactors operating within validated limits (40 CFR 141.721(c)(2)). Or in other
words, the UV reactor cannot be off-specification for more than 5 percent of the water delivered
to the public. The LT2ESWTR does not state an off-specification requirement for filtered
systems; however, States may establish requirements for their filtered systems, including a limit
for off-specification operation.
The UV reactors are off-specification when any of the following conditions occur:
• The flow, UV intensity, or lamp status is outside of the validated range.
• The UVT or UV intensity is outside of the validated range (if the UV intensity and
UVT setpoint approach is used (section 3.1.5)).
• The calculated dose is outside of the validated range at a given flow (if the calculated
dose approach is used (section 3.1.5)).
• All UV lamps in all UV reactors are off because of a power interruption or power
quality problem, and water is flowing through the reactors.
UV Disinfection Guidance Manual ' 3-9 ' • . June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
It is important to determine the appropriate design values for water quality, lamp
fouling/aging factor, flowrate, and power quality because of these LT2ESWTR requirements and
recommendations. If the design parameters are not chosen conservatively enough, the UV
reactors may be operating off-specification and be out of compliance. However, overly
conservative design values may result in unnecessarily large UV reactors or more UV reactors
than necessary.
3.1,3.1 Assessing Water Quality
As highlighted in Chapter 2, the following water quality parameters are the primary
parameters that affect UV installation planning and design:
« Parameters that affect UV dose delivery
- UV absorbance/transmittance from 200 - 300 nm (germicidal range)
- Upstream chemical additives
• Parameters that typically determine sleeve and UV intensity sensor fouling
- Calcium
- Alkalinity
- Hardness '
Iron
- PH
Lamp temperature
It should be reiterated that this manual is focused on post-filtration applications;
therefore, it is assumed that turbidity is low (1 nephelometric turbidity units (NTU) or less) and
results in insignificant particle effects on UV dose delivery (Linden et al. 2002b). It is also
assumed that the water meets applicable maximum contaminant levels (MCLs) and secondary
MCLs.
Water quality data should be collected from locations that are representative of the
potential UV installation location. The duration of sampling, number of samples collected, and
data analyses used to evaluate water quality for UV disinfection are similar to the approaches
taken for other water treatment technologies. The data collection frequency should be a based on
flow variability, the consistency of the source and treated water qualities, and the potential for
obtaining cost and energy savings by refining the design criteria. The extent of water quality
data collected should be left to the discretion of the utility and the designer based on experience
and professional judgment. States may desire to provide input on data collection needs.
The four main considerations for assessing water quality are A254, fouling potential, lamp
fouling/ aging factor and upstream chemical impacts. Each of these is discussed in the following
sections.
UV Disinfection Guidance Manual 3-10 , June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
UV Absorbance
As discussed in Chapter 2, the A2541 of the water directly influences UV dose delivery.
The A254 data should be evaluated to select a design A254 value. The design Ajs4 along with the
specified UV dose and flowrate will be used by the UV manufacturer to determine the
appropriate UV reactor. In addition, UV manufacturers may use the A254 range at the WTP to
determine the turndown (i.e., power modulation) needs of the UV reactors.
Overly conservative design A254 values (i.e., low UVT) can result in over-design and
increased capital costs. Conversely, inappropriately low design A254 values can result in UV
reactor operation outside the validated operating range and potential non-compliance. As with
most designs, the larger the data set, the more refined the final design can be. A utility with very
stable A254 might only need one or two months of data, while a utility that experiences seasonal
changes would benefit from more frequent data collection during seasonal events and over a
longer recording period.
The A254 sampling plan should include collection of A2542 measurements in grab samples
or continuously with an on-line A254 monitor. If A254 peaks occur regularly during the Spring
and Fall, increased sampling frequency during these periods will better capture the magnitude
and duration of the peaks. If different sources or combination of sources (i.e., blending) are used
during the year, the A2M of the potential source water blends should be characterized to properly
identify the appropriate water quality conditions. In addition, the maximum A254 may not
correspond to the period of maximum water production. The relationship between seasonal
production rates and Ai54 data should be considered when developing design criteria.
A cumulative frequency (CF) diagram of the A254 data may assist the utility in
determining its design Aas4 value. Figure 3.5 presents a CF diagram for three filtered waters; the
CF percentile (x-axis) shows the percentage of the dataset that is lower than a given value of A254
over the data collection period. For example, if the 90th percentile Ai54 is 0.043 cm"1, then 90
percent were lower and only 10 percent of the measurements were higher than 0.043 cm"1 over
the period of record.
1 A254 in this section implies A254 measurement specifically at 254 nm unless otherwise noted
2 A254 measurements for developing the design basis for UV disinfection systems should be performed on unfiltered
samples, not with the 0.45 um pre-filtered samples typically used to characterize NOM. However, if only
measurements that have been filtered are available, they still provide valuable information. It should be noted that
pre-filtered measurements are typically biased low (in terms of absorbance), but this bias is generally minimal.
UV Disinfection Guidance Manual 3-11 June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
Figure 3.5 Example CF Diagram for Three Filtered Waters
0.120
~ o.ioo
'E
•WT
I 0.080
0.060
•s
o
0.040
0.020
0.000
. I
i
III
j I
I
I
-------
3. Planning and Design Aspects for UV Installations
220
180
Figure 3.6 Example Flow and UV Absorbance (at 254 nm) Data
Filtered Water 3
WTP Capacity: 220 mgd
0.20
— Flow
* UVAbsat254nm
0.00
The design A254 (e.g., a CF percentile) also should be a function of the utility's preferred
level of conservatism and the site-specific Ais4and flow data. The UV reactor sizing and cost are
not directly proportional to A254 but will increase for increased Ai54 design values. However, by
evaluating the CF plot and collaborating with the UV manufacturer to assess the cost
implications of using a lower A254 value, the utility and designer can select the most appropriate
design A254 for the water quality and disinfection objectives of the project.
Typically, the UV manufacturers work in terms of UVT; therefore, the design A254 is
typically converted to a design UVT3. Because UV manufacturers use UVT in their design and
control of the UV reactors, the remainder of this chapter will use UVT as opposed to
The spectral absorbance of the water over a range of wavelengths (200 - 400 nm) should
also be collected, especially if medium pressure (MP) reactors are being considered. MP lamps
emit light at a range of wavelengths across the 200 nm to 300 nm range. The UV absorbance of
water varies with wavelength, typically decreasing with increasing wavelength. As such, the
attenuation of UV light in a UV reactor, the corresponding disinfection performance, and the UV
intensity sensor response depend on the absorbance at each of the emitted wavelengths. Site- '
specific spectral absorbance can be used to model MP reactors and may be incorporated into UV
dose monitoring and control systems by some UV manufacturers. Spectral absorbance may
= 100* 10""^
UV Disinfection Guidance Manual
Proposal Draft
3-13
June 2003
-------
3. Planning and Design Aspects forUV Installations
exhibit seasonal variation; therefore, spectral absorbance should be collected at different times
during the year to assess this variation. Also, the spectral absorbance may be used to determine
the appropriate UV-absorbing chemical for validation of the UV reactors that will be installed,
which is discussed in section 4.3.3.2.
Fouling Potential
The rate of fouling and the corresponding frequency of sleeve cleaning depend on
hardness, alkalinity, lamp temperature, pH, and certain inorganic constituents (e.g., iron and
calcium). Fouling is typically caused by precipitation of compounds with low solubility or
compounds where the solubility decreases as temperature increases (e.g., CaCOj). "If significant
seasonal shifts in any of the parameters are expected, these trends should be captured in the
monitoring period. Again, a CF diagram may assist in the selection of the appropriate design
criteria.
While the specific rate of fouling and optimal cleaning protocol for any given application
cannot currently be predicted, a proper cleaning protocol and sleeve-fouling factor can be
adequately estimated for most water sources without pilot- or demonstration-scale testing and
then adjusted during normal operation. Extensive data have been generated from pilot-scale
testing on waters of low to moderate hardness and iron content (Mackey et al. 2001 and Mackey
and Gushing 2003). At total and calcium hardness levels below 140 mg/L and low iron (less
than 0.1 mg/L), standard cleaning protocols and wiper frequencies (one sweep every 15 minutes
to an hour) were more than adequate to address the impact of sleeve fouling at the sites tested.
At sites with hardness or iron that exceed these levels, it may be advantageous to evaluate
fouling rates on a site-specific or worst case basis via pilot or demonstration testing (described in
Appendix J) of during UV reactor start-up (section 5.1) to identify how best to address fouling.
Although fouling is not expected to be a significant problem for most utilities, the listed
water quality parameters (page 3-10) should be monitored prior to designing the UV installation
unless adequate water quality data are available. It is important to provide these data to the UV
manufacturer to assist them in a qualitative assessment of the fouling potential for their UV
reactors and to assist the designer in determining what cleaning system should be specified. In
addition, the lamp fouling/aging factor will depend on the initial assessment of potential fouling,
which is discussed in the next section.
Lamp Fouling/Aging Factor
Sleeve fouling; lamp aging, and UV intensity sensor window fouling (if applicable) affect
long-term UV reactor performance. Accumulation of foulants on the lamp sleeve surface can
reduce transmittance of the lamp energy to the water. The rate of fouling depends on the factors
discussed in the previous section. In addition, lamp output decreases over time due to its
physical aging. The rate at which lamp output will decrease is a function of the lamp physical
characteristics, lamp hours in operation, number of on/off cycles, and power applied per lamp
length. In MP reactors, UV lamp aging can also result in a change in the spectral output over
time. Lamp aging is discussed in detail in section A.3.1.6.
A reduction in lamp output results in a reduction in UV dose. The effects of these
parameters are typically incorporated into the UV reactor design by specifying a lamp
UV Disinfection Guidance Manual 3-14 June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
fouling/aging factor, which includes the effects of both sleeve fouling and lamp aging. The lamp
fouling/aging factor will be site-specific and based on the assessment of fouling described
previously and lamp aging information. The lamp aging characteristics can be obtained from the
UV manufacturer and should be certified by an independent third party. The lamp fouling/aging
factor is used by the manufacturer to assist in the selection of the appropriate UV reactor. For
example, if a 0.5 lamp fouling/aging factor is specified, the UV manufacturer will choose a UV
reactor the appropriate lamps (or number of lamps) where the specified UV dose can be achieved
at half of the initial UV lamp output (after burn-in) with all the lamps energized at full power.
The lamp fouling/aging factor typically ranges, from 0.5 (NWRI2000) to 0.9.
The lamp fouling/aging factor is typically specified with a corresponding guaranteed UV
lamp life (e.g., 5000 hours). These items are typically specified together to ensure that the UV
lamp replacement frequency does not occur more frequently than specified by the guaranteed
lamp life given the specified lamp fouling/aging factor. The lamp fouling/aging factor can be
estimated based on the designer's experience and UV manufacturer input. In addition, pilot and
demonstration tests can be completed to estimate the lamp fouling/aging factor as described in
Appendix J.
Selection of a lamp fouling/aging factor and a guaranteed lamp life value is a trade-off
between maintenance costs (the frequency of lamp replacement or chemical cleans necessary)
and capital costs (the size of the UV reactors). A lower lamp fouling/aging factor means the
utility will have less frequent lamp replacements because the UV reactors are designed with
higher powered lamps or more.lamps to achieve the necessary UV output at the guaranteed lamp
life. However, designing a UV reactor with higher powered lamps or more lamps will increase
the size of the needed UV reactor. Thus, the use of a fouling/aging factor that is too conservative
could result in the over-design of the UV reactors. Conversely, the use of a lamp fouling/aging
factor that is not conservative enough may result in the underestimated reduction in the output of
the lamp due to fouling/aging and potentially result in off-specification operation or more
frequent lamp replacement. ,
Impacts of Upstream Treatment Processes
Unit processes upstream of UV reactors can have a significant impact on the UV reactor
performance. The three potential ways that upstream processes may affect UV performance are
(1) to increase UVT by increasing organics removal or oxidizing organics, (2) to decrease UVT
because certain chemicals will absorb UV light, and (3) to affect the lamp sleeve fouling rate.
It is possible to increase filtered water UVT by increasing the coagulant dose; however,
the results will be site-specific. In one study, the UVT was increased from 80% to 89% by
increasing the alum dose from 15 to 45 mg/L (Gushing et al 2001). However, the UVT increase
from an increased alum dose should be considered against the increased alum chemical costs and
sludge production. UVT increases would also probably be observed if other iron coagulant and
poly-aluminum chloride coagulant doses were increased.
Properly implemented, ozone disinfection prior to UV disinfection has the potential to
increase the UVT from oxidation of organic matter. Conversely, ozone disinfection can decrease
UVT if a residual ozone concentration is present in the UV reactors. If the ozone residual is
adequately quenched, a net increase in the UVT will be observed (Malley 2002); an example of
UV Disinfection Guidance Manual 3-15 June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
this increase for an unfiltered water is shown in Figure 3.7. If a UVT increase is desired, then
combination of coagulant increase and ozone disinfection will likely give the greatest UVT
increase (Gushing et al 2001).
Figure 3.7 Example Effect of Pre-ozonation on UV Absorbance
if Ozone is Quenched Prior to UV Disinfection
0.5
0.4
8 °-3
I
0.2
0.1
0.0
—— Preozonatad UV Influent
(no detectable ozone residual)
" UV Influent- No ozone
200 220 240 260 280
W avelength (nm )
300
320
Most common water treatment chemicals themselves will not significantly impact UVT.
The following common water treatment chemicals do not significantly affect UVT at typical
concentrations present in filtered water: Alum, aluminum, ammonia, ammonium, zinc,
'phosphate, calcium, hydroxide, and ferrous iron (Fe+3) (Gushing et al 2001).
However, hypochlorite (Clb~), ferric iron (Fe+2), permanganate, and ozone were the only
commonly used chemicals examined that might reduce UVT (Gushing et al 2001) as described
below.
. Residual CIO" has only a slight effect on UVT. For example, a CIO" residual of 3.5
mg/L will cause the UVT to decrease from 91% to 90% (Cushing et al 2001).
However, in most cases, a hypochlorite residual that high will not be flowing through
the UV reactor.
• It is unlikely that ferric iron will be present in filtered waters because ferric iron is
only present when there is low dissolved oxygen.
. Permanganate is a strong absorber of UV light; however, it is typically added in the
raw water to oxidize taste and odor or iron and manganese. Therefore, when applied
to raw water, there should not be a significant permanganate concentration in the
filtered water.
UV Disinfection Guidance Manual
Proposal Draft
3-16
June 2003
-------
3. Planning and Design Aspects for UV Installations
. Ozone residual can be quenched, and then the UVT will not be decreased. Care
should be taken when choosing the quenching agent because one popular choice,
thiosulfate (often used in the form of calcium thiosulfate), is a strong absorber of UV
light (section A.4.1.3, Table A.5) and wilt decrease the UVT. Sodium bisulfite, an
alternative to calcium thtosulfate, will not significantly impact UVT.
! . ^
The possible UVT variation from upstream processes should be assessed by collecting
UVT data during various operating conditions (e.g., a range of alum doses) that are typically
observed. Potential treatment process upsets should also be considered in the water quality
analysis to determine the extent to which they impact the design UVT and cleaning regime.
Some unit processes that use metal-based coagulants may affect the rate of fouling; these
effects will be site-specific. Mackey et at. (2001) found that iron levels less that 0.1 mg/L could
be adequately cleaned by standard protocols as described previously. In addition, lime softening
has been shown to reduce fouling potential (Mackey et al. 2001). Overall, the effect of upstream
coagulant addition and residual metals should be considered in the fouling data monitoring
described previously.
3.1.3.2 Determining Design Flowrate
Flowrate is a fundamental design parameter that, in combination with water quality, UV
dose, and tamp fouling/aging factor determines the necessary size and number of UV reactors.
The design criteria should identify the average, maximum, and minimum flowrates that the UV
reactors will experience. Potential methods for determining the design flow for the three
described retrofit locations are shown in Table 3.1. In addition, potential future changes in plant
capacity should be considered when determining the UV installation design flow.
Table 3.1 Potential Method to Determine Design Flow
Retrofit Location
Combined filter retrofit
Individual filter retrofit
Post-HSP
Design Flow Basis
Combined rated capacity of all duty filters1
Rated design flow for individual filter
Rated capacity of the HSP station
Flow does not include redundant filters
3.1.3.3 Assessing Electrical Power
The sensitivity of UV reactors to power fluctuations make electrical power supply a
critical component of the UV installation planning and design. In addition, the electrical system
design needs to ensure that the UV installation will meet the requirements or recommendation of
operating within validated conditions (i.e., maximum allowed off-specification). Also, it is
impossible to meet inactivation goals if the power quality causes the reactor to go down (i.e., no
disinfection) for longer than the need to obtain the desired treatment level. For example, if a 2-
log Cryptosporidium inactivation is desired, the UV reactors cannot be down while more than 1
percent of the flow passes through them.
UV Disinfection Guidance Manual 3-17 June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
UV lamps can potentially their lose arc if a voltage fluctuation, power quality anomaly,
or a power interruption occurs. For example, voltage sags that vary from 10 to 15 percent from
normal operating conditions for as low as 2 to 5 cycles (0.03 to 0.08 seconds) may cause UV
lamps to lose their arc.
Low pressure (LP) lamps generally can return to full operating status within 15 seconds
after power is restored. However, low-pressure high output (LPHO) and MP reactors that are
more typically used in drinking water applications exhibit significant restart times if power is
interrupted. The start-up and restart behavior for LPHO and MP lamps is summarized in Table
3.2.'
Table 3.2 Start and Restart Times for LPHO and MP Lamps
Lamp Type
LPHO
MP
Cold Start2
2 min warm-up
+
4-5 min to full power
total time: 6-7 minutes
No warm-up or cool down
+
5 min to full power4
total time: 5 minutes
Warm Start3
2 min warm-up
+
2-5 min to full power
total time: 4 -7 minutes
5 min cool down
+
5 min to full power4
total time: 10 minutes
Information shown in table is compiled from Calgon Carbon, Severn Trent, Trojan, and Wedeco.
2 A cold start occurs when UV lamps are started when they have not been operating for a significant
period of time.
3 A warm start occurs when UV lamps are started after they have just lost their arc (e.g., due to voltage
sag).
60 percent intensity is obtained after 3 minutes.
The effects of temperature can increase or decrease the times listed in Table 3.2 and
should be discussed with the UV manufacturer. Individual manufacturers report that colder
water temperatures (below 50 degrees Fahrenheit, 10 degrees Centigrade) can result in slower
startups for LPHO lamps than listed in Table 3.2. Conversely, MP manufacturers report shorter
re-start times with colder temperatures because the cold water accelerates the condensation of
mercury (i.e., cool down), which is necessary for re-striking the arc.
To minimize the potential for off-specification operation, utilities should evaluate the
reliability and quality of their power supply. Local power suppliers can often provide power
quality and reliability data and should be the first source of information on power quality. For
those locations where power quality is unknown, a power quality assessment is recommended.
An assessment may be as simple as reviewing operating records of power quality incidents (if
available) and power interruptions or Supervisory Control and Data Acquisition (SCADA)
information for the existing plant. More advanced assessments may include the installation of
power quality monitors or the retention of an outside consultant to conduct a detailed power
quality assessment. Generally, personnel with a working knowledge of electrical supply and
installation will be able to review power supply data and determine if power quality problems
exist. If a problem is identified, however, tracing it back to its source and determining an
appropriate remedy is often best left to an expert that specializes in this area.
UV Disinfection Guidance Manual
Proposal Draft
3-18
June 2003
-------
3. Planning and Design Aspects for UV Installations
The most common sources of power quality problems are as follows:
• Faulty wiring and grounding
• Off-site accidents (e.g., transformer damaged by a car accident)
• Weather-related damage
* Animal-related damage
. Facility and equipment modifications
• Power transfer to emergency generator or alternate feeders
In specific locations that are subject to frequent power fluctuations or outages, the
following options should be considered to minimize off-specification operation and ensure
regulatory compliance:
1. Installation of a backup generator
2. Connection to a second, independent power source
3. Installation of power conditioning equipment or a battery-supported uninterruptible
power supply (UPS)
These options will have different response and backup periods associated with them. For
example, a backup generator cost-effectively provides backup power if an extended power
interruption occurs; however, it will not ensure a continuous power supply to avoid UV reactor
shutdown due to voltage sags. Connection to a second, independent power source may have the
same issues as the backup generator, depending on the power quality associated with the second
power source.
Power conditioning equipment will provide high quality power even if voltage sags or
other power quality problems occur. However, power-conditioning equipment does not provide
backup power for extended power outages. A battery-supported UPS provides continuous, high
quality power (i.e., prevent voltage sags) and a specific amount of backup power for a longer
outage, UPS systems can provide as much battery backup, as specified; however, typically UPS
systems for this purpose range between 2 and 15 minutes of battery backup.
The most suitable option will depend on the power quality of the utility, requirements
limiting off-specification operation, and preferences of the utility and State. For example, an
unfiltered system with poor power quality that experiences multiple voltage sags everyday and
periodic interruptions lasting over 3 minutes may consider installing a UPS system with 5
minutes of backup batteries to ensure the 95 percent requirement of operating within validated
ranges is met (40 CFR 141.721 (c)(2)). However, a filtered system that experiences two or three
voltage sags a month and no long-term power interruptions may not need to provide any
additional power or power conditioning equipment.
UV Disinfection Guidance Manual
Proposal Draft
3-19
June 2003
-------
3. Planning and Design Aspects for UV Installations
Any equipment needed to address power quality problems affects both the cost and the
feasibility of implementing UV disinfection. For example, the UV reactor cost and installation
footprint has been estimated to increase by approximately 25 percent if a UPS system with
5 minutes of backup capacity is installed (Cotton et al. 2002). Other power conditioning options
without backup power are less expensive and have lower footprint needs.
*.
It is important that a utility have a complete WTP-wide assessment of its power quality
when considering UV disinfection. Any actions involving the electrical system may also affect
the WTP power quality and equipment performance. For example, the impact of the WTP's
maintenance program for backup generators (e.g., routine startup and exercise) should be
considered during the planning and design of the UV reactors to ensure that the program
supports compliance goals and does not cause excessive UV reactor shutdown times. Other
items that may affect power quality include future integration or upgrade of equipment
(particularly equipment with a large power demand or variable frequency operation), testing of
backup power supplies, deterioration of existing facility wiring (resulting in poorly grounded
circuits), overload of electrical circuits, and any other activity that may affect the electrical
supply or distribution within the facility.
3.1.4 Evaluating Potential UV Reactors
It is important to evaluate the available UV reactors in the planning process because each
manufacturer's UV reactors are unique and proprietary. Process footprints and related
installation needs (e.g., UV reactor to control panel distances) are different, depending on the UV
manufacturer. This section provides a brief overview of different UV reactors, their impact on
space requirements, and UV reactor validation issues. More detailed UV reactor information is
presented in section 2.4. In addition, UV manufacturers should be contacted directly to gain a
better understanding of the UV reactors available and what UV reactors are applicable to the
utility's installation locations given the design criteria developed in section 3.1.3.
3.1.4.1 UV Reactors
' There are-different types of UV reactors for disinfecting drinking water with unique
characteristics, such as lamps, lamp configuration in the reactor, cleaning systems, ballasts, and
control systems (section 2.4.). This section briefly highlights the differences in UV reactors that
affect design of the UV installation.
UV reactors can generally be characterized based on lamp type with LPHO and MP •
lamps being the most applicable to WTPs. One of the fundamental differences between LPHO
and MP reactors is the lamp intensity output, which influences the UV reactor configuration and
size, lamp life, number of lamps, electrical needs, and ballasts. Each has its inherent advantages
and disadvantages. While a competitive procurement can be made among these two reactor
types when the construction contract is bid, the overall layout and supporting facilities will be
different for each.
The UV reactor footprint depends on the UV reactor configuration and UV lamp type.
There are several different UV reactor configurations. Typically, LPHO reactors are in-line (i.e.,
UV Disinfection Guidance Manual 3-20 June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
configured like a pipe). However, MP reactors can be in-line, S-shaped, or U-shaped, depending
on the UV manufacturer and the site constraints of the specific installation location. Typically,
LPHO reactors have a larger footprint than MP reactors because more UV lamps are needed to
deliver the same UV dose. MP reactor footprints will also vary, depending on lamp orientation
(e.g., parallel versus perpendicular to flow). When evaluating locations for installation, the
largest UV reactor footprint of those being considered should be used to estimate the UV
installation footprint.
Lamp life also varies between LPHO and MP reactors. Most manufacturers provide
warrantees of 8,000 to 12,000 hours for LPHO lamps. Guaranteed life for MP lamps range from
4,000 to 8,000 hours. Although the lamp life for LPHO is greater than that for MP reactors, due
to the need for a greater number of lamps, the actual number of lamps that are replaced during a
given period may be less for a MP reactor. It is important to consider the labor associated with
lamp replacement, as well as the actual unit cost of the replacement lamps, when estimating the
operating and maintenance costs of the two technologies. In addition, while LPHO reactors
typically have more lamps, the actual power input is less than that for similarly sized MP
reactors because MP lamps are less efficient in converting the power input to germicidal
wavelengths for disinfection. This may result in a higher input power and an increase in the
overall power consumption for MP reactors compared to LPHO reactors.
The lamp sleeve cleaning systems can also be different between LPHO and MP reactors.
LPHO reactors may have off-line chemical cleaning (OCC) systems instead of on-line
mechanical cleaning (OMC) because of the larger number of lamps. With OCC systems, the UV
reactors must be taken offline to be cleaned. OMC and OCC systems are described in
section 2.4.5. This may result in higher maintenance costs for LPHO reactors, depending on the
extent to which cleaning is necessary.
Finally, the type of ballast used will affect the UV installation layout. Ballasts regulate
the power supply at the appropriate level needed for energizing and driving the UV lamps. UV
reactors may use electronic ballasts, electromagnetic ballasts, or transformers. Transformers are
typically more stable than electronic or electromagnetic ballasts and allow a greater separation
distance between the UV reactor and control panel. However, most transformers allow only step
adjustment of lamp intensity. Compared to transformers, ballasts have the capability to provide
almost continuous intensity adjustment but may increase lamp aging and spectral shift and have
lower allowable separation distances between the UV reactor and control panel. It is important
to discuss the implications of these various components with the UV manufacturers to determine
their effect on the UV installation layout and design. Specific items that should be discussed
include ballast cooling needs, allowable separation distances, and intensity adjustment
capabilities.
The differences described above imply that UV reactor evaluation should not be based
solely on capital costs. Operation and maintenance costs, including energy usage and labor, will
be important in an overall life cycle cost comparison. This is discussed in greater detail in
section 3.1.7.
UV Disinfection Guidance Manual ,3-21 June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
3.1.4.2
UV Reactor Control Strategies
There are currently three different control strategies for UV reactors, which affect how
UV reactors are validated and operated. The three general control strategies relate to three
methods for monitoring dose-delivery and are summarized in Table 3.3. The first strategy
utilizes one or more UV intensity sensors located at a distance from the lamps that yields an
intensity signal that is proportional to UV dose'(UV intensity setpoint approach), and the
intensity sensor measurement and flowrate are used to monitor dose delivery. The second and
third methods utilize UV intensity sensors that are positioned close to the lamps (so that there is
minimal absorbance by the water) and separate monitors for UVT. The second approach
incorporates a validated setpoint value for UVT, in addition to setpoints for UV intensity and
flowrate, to ensure a given dose (UV intensity and UVT setpoint approach). In the third
approach, the UV dose is calculated based on these measurements of flowrate, UV intensity, and
UVT via a validated computational algorithm developed by the manufacturer (calculated dose
approach).
Table 3.3 UV Reactor Control Strategies
Control Strategy
UV Intensity Setpoint
UVT and UV Intensity Setpoint
Calculated Dose
Dose Delivery Monitoring and Control Basis
UV intensity sensor measurement
UV intensity sensor and UVT measurement •
The calculated UV dose1
1 The UV reactor calculates a UV dose, using the UV intensity sensor measurement, the
UVT of the water, and the flowrate.
In the planning phase, these control strategies need to be evaluated by the designer and
utility to determine if a particular control strategy is preferable based on the ease of integration
into their existing operation and control system. The impacts of the control strategy on the
instrumentation and controls are discussed in section 3.3.2, and the specific validation
recommendations for each control strategy are presented in section C.4.9.
3.1.4.3
Equipment Validation Issues
The LT2ESWTR requires that UV reactors be validated (40 CFR 141.729(d)). A utility's
approach to UV reactor-validation will affect the UV installation design. The issues to consider
are the hydraulic parameters for validation and whether equipment will be validated on-site or
off-site.
This section describes how these issues affect the design and installation footprint
estimation. Chapter 4 provides an overview of validation, and Appendix C details UV reactor
validation guidelines in detail.
UV Disinfection Guidance Manual
Proposal Draft
3-22
June 2003
-------
3. Planning and Design Aspects for UV Installations
Validation Hydraulics
The inlet and outlet hydraulics of the UV reactor can significantly affect dose delivery;
therefore, the following validation and corresponding installation strategies are recommended in
the validation protocol (section C.3.1.5) and are presented in Table 3.4.
Table 3.4 Summary of Recommended Hydraulic
Configurations for Validation and Installation
Option
1
2
3 ,
Validation
The inlet and outlet configuration is the
same as the installation for 1 0 diameters
upstream and 5 diameters downstream of
the UV reactor.
The UV reactor is validated with a 90-
degree bend directly upstream of the UV
reactor. The UV reactor is defined to
include a specific amount of straight pipe
upstream or downstream of the UV reactor
as specified by the UV manufacturer.
The velocity at the validation facility is
measured at evenly spaced points through
a given cross section of the flow upstream
and downstream of the UV reactor.
UV Installation
Inlet and outlet configuration is the same as
when the UV reactor was validated for 10
diameters upstream and 5 diameters
downstream of the UV reactor.
The UV reactor should be installed with a
minimum of 5 pipe diameters of straight piping
between the UV reactor and any upstream
hydraulic configuration.1
The velocity at the installation is measured at
evenly spaced points through a given cross
section of the flow upstream and downstream
and is within 20 percent of the theoretical
velocity determined during validation.
This approach is not acceptable if the upstream fitting is an expansion or if the upstream valve will be used for flow
control. A valve that will be exclusively used for open/close service (e.g., isolation) is acceptable.
Option 1 is most applicable when unique piping configurations are needed or if the inlet
and outlet conditions validated in Option 1 cannot be achieved because of site constraints. For
example, Option 1 may be the only validation option for an individual filter effluent location,
which probably does not have 5 diameters of straight pipe before the UV reactors (Option 2)
because of existing site constraints.
The validation and installation of a particular UV reactor should meet one of these
options. Option 2 provides more general applicability for validation and installation of UV
reactors. For example, the inlet and outlet piping configuration for installations in a new
building could be designed based on how the procured UV reactor was validated. Option 3 also
provides flexibility but may have the practical limitation of measuring the velocity through a
cross section at the installation.
Off-site Versus On-site Validation
Manufacturers will likely validate UV reactors over a wide range of flowrates and water
quality (e.g., UVT) conditions at off-site testing facilities. The inlet and outlet hydraulic
conditions during validation will probably be selected so the UV reactors can be installed in most
WTPs. Off-site validation has several advantages, including simplicity, cost, and the ability to
design around a UV reactor with known performance characteristics and inlet and outlet
hydraulics. However, the LT2ES WTR requires that the site-specific installation and operating
. UV Disinfection Guidance Manual 3-23 June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
conditions must fall within the range of conditions used when the installed UV reactor was
validated off-site (40 CFR 141.729(d)). If the validation conditions do not encompass the
utility's design criteria or inlet and outlet piping configurations, the utility may request that the
UV manufacturer re-validate the unit off-site under specific testing conditions that closely match
those of the proposed installation. Alternatively, on-site validation can be performed.
The advantage of on-site validation are that the UV reactors can be validated under the
exact piping hydraulic conditions at which it will operate, and the UVT will more accurately
represent the UV installation even if a UV-absorbing chemical is added. In addition, the
equipment necessary for on-site validation will also provide the flexibility for future testing to
optimize the UV reactor performance under specific hydraulic and water quality conditions even
if they are not completed for the initial validation. However, a disadvantage of on-site validation
is that the UV installation is designed and constructed without prior validation of the
performance of the UV reactors. This may lead to the UV installation failing to meet
performance requirements, and it may be difficult to increase UV disinfection efficiency after the
UV reactors are already installed. In addition, on-site validation is limited to the highest UVT
available at the time of testing. Consequently, UV reactor performance characteristics cannot be
determined at higher UVT, and the UV reactors may need to be operated at conditions other than
optimal, resulting in higher power use and faster lamp and ballast replacement frequencies.
Other disadvantages include the logistics and cost of the testing. For example, one unit must be
isolated from the system to allow validation testing to occur, and a permit may be needed to
discharge the non-pathogenic challenge microorganism.
If on-site validation is desired, then the UV installation design should be adapted to
enable testing. The UV reactor design would need to incorporate feed and sample ports, static
mixers, space for tanks near the UV installation (for the addition of the challenge microorganism
and UV absorbing chemical), and adequate facilities for laboratory testing, and discharge of the
treated water.
3.1.5 Evaluating Operational Strategies
The operational strategy is defined in this manual as the method in which the utility
chooses to operate the UV reactors given the UV reactor's control strategy and validation data.
It is important for the utility to understand the control strategies unique to various UV reactors
(section 3.1.4.2) and select equipment consistent with their operating philosophy and energy
efficiency objectives. The control strategy is defined as the method that the UV reactor uses to
monitor and control the UV lamp power based on flow and UVT to deliver the specified UV
dose. For each UV reactor, the operating conditions must be defined based on validation testing
results (40 CFR 141, Subpart W, Appendix D), and the validation data will vary with different
control strategies. The validation data can be utilized in different ways that facilitate a simple or
complex operating strategy; three potential approaches are described in Table 3.5. Detailed
examples of how to determine the operational parameters for these operational strategies are
described in section 5.5.
UV Disinfection Guidance Manual
Proposal Draft
3-24
June 2003
-------
3. Planning and Design Aspects for UV Installations
Table 3.5 Potential Operational Strategies
Operational
Strategy
Single
Operation
Setpoint
Variable
Setpoint
Operation
Setpoint
Interpolation-
Description
One setpoint is used for
all flows and DVT values
that were validated
A setpoint would be used
for a given flowrate and
UVT range using a
lookup table
The setpoints are
calculated as a function
of flowrate, typically
automatically using the
UV reactor controls'
Advantages
Simplest operational •
strategy
Increased energy efficiency
over the single setpoint
approach
The most energy efficient
operation and may reduce
operational hours needed if
operated automatically
Disadvantages
Not as energy efficient
because the UV reactor is
over-dosing at low flows
More complex operation
compared to single setpoint
approach and may
necessitate more advanced
controls for the UV reactor
Potentially more validation
data is needed (which may
increase validation costs)
and necessitates advanced
reactor controls
Only an option for UV intensity setpoint and calculated dose setpoint approach because the UV intensity
and UVT setpoint approach is controlled as function of flowrate and UVT (as opposed to only flowrate)
3.1.6 Evaluating Hydraulics and Process Footprint
The potential locations for UV disinfection identified in section 3.1.2 can be evaluated
based on an understanding of the candidate UV reactors, the hydraulics, and the estimated
process footprint. This section discusses the principle criteria that affect the feasibility of a UV
installation location - (1) hydraulic needs and limitations and (2) space availability and site
constraints.
3.1.6.1
Hydraulic Considerations
When selecting the appropriate location for UV reactors, the hydraulic needs should be
addressed. Headloss through a UV installation is dependent on the specific UV reactor and
flowrate and generally varies from 0.5 to 3 feet. Characteristic headless data should be obtained
from the UV manufacturers) for all candidate UV reactors. In addition to the headloss
associated with the UV reactor itself, the headloss associated with piping, valves, flow meters,
and flow distribution-devices should be considered when assessing the feasibility and location of
the installation. The overall headloss of a UV installation is typically between I and 8 feet.
If the headloss through the UV installation is greater than the available head,
modifications to the plant design and/or operation may be necessary. Some potential
modifications, alone or in combination, that may be considered to address hydraulic limitations
are listed below followed by details about each:
. • Eliminating existing hydraulic inefficiencies within the facility to improve head
conditions (e.g., replace undersized or deteriorated piping and valves)
. Modifying the operation of the clearwell to accommodate the UV installation
. Modifying the operation of the filters to accommodate the UV installation
UV Disinfection Guidance Manual 3-25 - June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
« Installing booster pumps
. Modifying the UV reactor design (through the UV manufacturer) to reduce headless.
If the UV reactor design is modified, it must be validated in its modified condition to
ensure it meets performance requirements
Eliminating Existing Hydraulic Inefficiencies
Replacing undersized piping and valves with larger diameter piping and valves may
increase the available head for the proposed UV installation. Older piping can also produce
excessive headloss if the inner pipe surface is pitted or scaled or if the original pipe material has
a high coefficient of friction. Slip-lining the interior of existing pipe with a lower coefficient of'
friction pipe material (e.g., high density polyethylene)1 is one method of reducing friction losses.
Re-lining the existing pipe interior with a smooth coating will also reduce headloss.
Modifying Clearwell Operation
A utility may increase head available to a UV installation by lowering the surface water
level of the clearwell. However, this strategy decreases the storage volume available to meet
peak demands. In addition, a lower clearwell level will reduce the contact time available in the
clearwell for chemical disinfectants and may affect pump discharge head. It is important to
evaluate any potential reduction in disinfection credit if contact time in the clearwell is used for
calculating CT. The UV installation, though, may reduce the Giardia CT requirements
sufficiently to offset the reduction in contact time .
Modifying Filter Operation
A treatment facility may alter the operation of its filters to increase the head available for
the UV installation. However, this may reduce filter run times, unit filter run volumes, and result
in more frequent backwashing.- If conditions upstream of the filters are such that additional
freeboard and hydraulic head are available, a second option is to increase the water surface
elevation over the filters to help minimize the reduction in head available for filtration.
Installing Booster Pumps
When modifications to the existing facility or operations will not provide adequate head
for the UV reactors, booster pumps can be installed. Booster pumping provides additional
flexibility in the location of the UV reactors. The installation of booster pumps will increase
facility operation and maintenance cost and space requirements. The reliability of the pumps
should also be considered in the evaluation because the pumps become a critical operating
component. Additional detail on booster pump design is provided in section 3.3.1.6.
Modifying UV reactors
Modifying a UV reactor to reduce headloss (e.g., removing baffles) can affect
disinfection performance and should only be considered in careful collaboration with the UV
manufacturer. Any resulting gains in system head must be weighed against diminished
disinfection efficiency, which could result in more UV reactors being heeded to accommodate
UV Disinfection Guidance Manual 3-26 June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
the flow and provide the necessary UV dose. Any modified UV reactors will also need to be
validated in its modified condition.
Other Options to Address Hydraulic Constraints
If none of the above options are feasible, the utility could consider installing the UV
reactors upstream or downstream of the HSPs. If a location adjacent to the HSPs is selected, the
potential for damage from pressure surges is increased and a surge analysis should be completed.
Most lamp sleeves are designed to withstand continuous positive pressures of at Jeast 120 pounds
per square inch gauge (psig) (Roberts 2000; Aquafine 2001; Dinkloh 2001). However, lamp
sleeves are vulnerable to negative gauge pressure transients associated with water hammer. The
tolerance level of the sleeve depends on the quality of the quartz and the thickness and length of
the sleeve. However, pressures of negative 1.5 psig have been shown to negatively affect sleeve
integrity (Dinkloh 2001). Hydropneumatic tanks, surge relief valves, air release valves, or air
vacuum valves on pumps or at different locations along the pipeline can be used to help control
surge conditions.
3.16.2 Process Footprint
The process footprint should be estimated in the planning phase to determine potential
UV installation locations. One critical component needed to estimate the UV installation
footprint is the number, capacity, and configuration of the UV reactors. The number of UV
reactors depends on the redundancy chosen. UV reactor redundancy should be determined early
in the planning process and should use sound engineering approaches similar to those used for
other major equipment (e.g., capacity to provide full treatment with the largest unit out-of-
service). The specific level of redundancy should be determined by the utility based on
operating history and process requirements and should take into account,the site constraints. For
example, one UV reactor dedicated to each filter may have different redundancy needs than UV
installations treating combined filter effluent. Any excess capacity that may be available within
the UV reactors (e.g:, incorporation of additional lamps or change in lamp power) should also be
considered.
The number of UV reactors necessary is also affected by the acceptable power turndown
of the UV reactors and the LT2ESWTR requirement that the UV reactors must operate within
their validated flow range (40 CFR 141.729(d)). Some UV reactors will operate at low power
efficiency at reduced flowrates, and more UV reactors with a lower capacity may increase
energy efficiency, depending on water quality and flowrates. For the potential combinations of
number and capacity of UV reactors, the available turndown should be determined for each
configuration with respect to the anticipated flow range and power modulation capabilities of the
UV reactors.
The overall UV reactor and piping configuration will be affected by site constraints. For
example, a vertical orientation of the UV reactors may be necessary to reduce building footprint
because of little land availability. Ultimately, the selected configuration should balance the
capital cost of the equipment, which may be lower for designs incorporating high capacity
reactors, with the improved operating efficiency and flexibility that may be achieved using a
larger number of lower capacity reactors.
UV Disinfection Guidance Manual 3-27 June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
The following items should be considered when estimating the UV installation footprint
in the planning phase:
, . The number, capacity, and configuration of UV reactors (including redundancy and
connection piping)
« The configuration of the connection piping and the inlet/outlet piping necessary
before and after each UV reactor, based on validated hydraulic conditions and UV
manufacturer recommendations
. Booster pumps (if necessary)
. The space needed for electrical equipment including control panels, transformers,
ballasts, backup generator(s), and possible UPS systems
• The maximum allowable separation distance between the UV reactors and electrical
controls since distance limitations may apply
Access for maintenance and replacement, room for storage of spare parts, and
chemicals (if needed), and lifting capability for heavy equipment
. Provisions for on-site validation (if applicable)
Once the UV installation footprint is estimated, the feasible site locations may be
determined based on the available land and buildings to accommodate the installation footprint.
UV installation layout is discussed in more detail in section 3.3.5.
3.1.7 Preparing Preliminary Costs and Selecting the UV Installation Option
The amount of analysis necessary to determine the appropriate application point for a UV
installation is site-specific. Some options will clearly be infeasible while others may necessitate
a more detailed comparison of the installation options. Once feasible alternatives are identified,
the development of life cycle costs can be useful in selecting among alternatives.
Preliminary life-cycle cost estimates should include both capital cost and operation and
maintenance cost elements. Capital cost elements includes the cost of the UV reactors, pumping
(if necessary), electrical and instrumentation provisions, and site work; contractor overhead and
profit; piloting and validation costs; engineering, legal, and administrative costs. Depending on
the detail of the cost estimates being developed, the existing infrastructure may need to be
evaluated to develop the cost estimate. These issues are discussed in detail in section 3.3.5.
The average conditions for flowrate and UVT are typically the most representative for
determining annual operating costs, as opposed to the maximum design flowrate and minimum
UVT. Nevertheless, the specific operating limitations of the equipment and the electrical cost
rate structure for the installation should be considered. If a utility's electricity charge includes
both a usage and a demand component, the demand charge may need to be estimated based on
the worst-case operating conditions to accurately represent the cost to the utility. Similarly, it
UV Disinfection Guidance Manual 3-28 . June 2003
Proposal Draft
-------
3.. Planning and Design Aspects for UV Installations
may be important to correlate the anticipated energy demand for the UV reactors to the actual
rate structure for the facility if power costs vary based on the time of day and the flowrate and
UVT fluctuate significantly.
Selection of the best option may not be based solely on capital and operation and
maintenance costs. The final selection criteria should also consider the following factors:
• Cost-effectiveness and ability to meet the utility disinfection and design objectives.
• Ease of installation (where applicable).
• Operational flexibility and reliability.
• Specific maintenance needs.
Flexibility for future treatment expansion (if applicable).
3.2 Equipment Procurement Options
The same equipment procurement options that are used to acquire traditional equipment
(e.g., pumps) within the water industry can also be used for UV reactors. Owner pre-purchase;
base bid, under which the design is based on a single UV manufacturer but is open to alternatives
at the discretion of the owner; and contractor selection, in which operating or performance
criteria are established and final equipment selection is left to the discretion of the contractor are
the most common methods of procurement for traditional design-bid-build projects. Because the
use of UV reactors in drinking water treatment plant applications has been limited in the United
States, many of the projects to date have pre-purchased the UV reactors. Pre-purchase allows the
designer to work more closely with the UV manufacturer during design, reducing the potential
for errors that could occur with an evolving technology. However, pre-purchasing may
necessitate that a more detailed assessment be completed during the planning stage of the project
to ensure that the appropriate equipment is selected and that a second set of contract documents
be prepared. Further, this may result in the owner assuming increased responsibility for
equipment delivery and performance when compared to base bid or contractor selection. If
owner pre-purchase is selected, these factors need to be carefully considered and addressed by
the designer during development of the equipment procurement document.
The advantages and disadvantages of the procurement methods with respect to designing
and constructing a UV installation are described in Table 3.6. It should be noted that funding
sources or municipalities might have specific bidding and procurement requirements. These
requirements are site-specific and should be reviewed prior to establishing a project approach to
ensure all requirements are met.
UV Disinfection Guidance Manual 3-29 ' June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
Table 3.6 Potential UV Reactor Procurement Options
Procurement
Method
Advantages
Disadvantages
Owner
Pre-Purchase
> Single design around selected equipment.
> Actual UV reactor pricing is better defined
earlier in project.
• Owner receives equipment warranty
directly from UV manufacturer.
• May result in shorter project schedule if
equipment fabrication time occurs during
design and bidding phases of the UV
installation.
> Option may necessitate the preparation
of two sets of documents; an equipment
procurement document and contract
documents for the construction of the
overall UV installation.
> Option may not be possible under some
procurement codes.
> Except where procurement is assigned,
installation contractor is not single point
of responsibility for equipment.
> Equipment disputes need to be dealt
with by owner.
Base Bid
> Single design around selected UV
reactors.
• Contractor handles all pricing and
coordination with UV manufacturer.
• Low incentive for contractor to bid
alternates to selected UV manufacturer.
' It is difficult to prevent supplier
"packaging" of UV reactors.
> UV reactor disputes are problematic
because contractor was directed to use
equipment.
Contractor
Selection
> Fits most procurement codes.
' UV reactor disputes are the responsibility
of the contractor.
. Contractor is likely to select UV reactors
with lowest capital cost rather than
lowest life-cycle cost.
> Multiple UV installation designs may be
necessary, increasing engineering effort
and cost.
As discussed previously, Chapter 3 is organized in the same manner as the flow chart
shown in Figure 3.1, utilizing equipment pre-purchase and a design-bid-build approach for
project implementation. It should be noted that successful implementation of UV installations
can be accomplished using any of the equipment procurement and contractor selection
approaches currently available.
3.3 UV Installation Design Elements
Additional design concepts are expanded and refined in this section, particularly with
regard to hydraulic issues, control strategy, instrumentation, and electrical power. The section
concludes with considerations for the layout of UV installations.
3.3.1 UV Installation Hydraulics
Following the selection of an installation option during the planning phase, a more
detailed evaluation of system hydraulics should be conducted, including flow control,
distribution, and measurement. It is important that design of the inlet and outlet conditions be
coordinated with the validation process to ensure that the proposed configuration can be cost-
UV Disinfection Guidance Manual
Proposal Draft
3-30
June 2003
-------
3. Planning and Design Aspects for UV Installations
effectively validated and will provide hydraulic conditions that result in dose delivery equal to or
better than that provided during validation testing.
3.3.1.1 Inlet and Outlet Piping Configuration
Optimal hydraulic conditions vary based on the UV reactor design and lamp
configuration, but turbulenj^flow with a reasonably uniform velocity profile is generally
preferred. Turbulent flow conditions can be achieved at very low flowrates when compared to
the actual capacity of a given pipe cross section.
The recommended inlet and outlet conditions for validation and the installation are
summarized in section 3.1.4.3 and described in detail in section C.3.I.5. These
recommendations should be considered when designing the inlet and outlet conditions for the
UV reactors. The designer should contact the UV manufacturer to determine how the procured
UV reactors were validated and what the inlet and outlet piping constraints are for the UV
installation. If on-site validation is planned, the inlet and outlet hydraulics should be designed as
recommended by the UV manufacturer and as the site-specific constraints permit.
3.3.1,2 Flow Distribution, Control, and Measurement
Regulations specify flowrate, UV intensity, and lamp status as the minimum operating
conditions a utility must routinely monitor (40 CFR 141, Subpart W, Appendix D). Accordingly,
proper flow distribution and measurement are essential for compliance monitoring of the UV
reactors. Confirmation of compliance will be dependent on understanding the flow through each
UV reactor, regardless of the dose monitoring or control strategy used by the utility. Moreover,
UV reactors are validated within specific flow ranges and have associated operating .
characteristics that demonstrate dose delivery as a function of flow. Therefore, the flowrate
through the UV reactor must be known to ensure that proper dose delivery is achieved.
This section discusses different methods for ensuring proper flow distribution and
measurement through UV reactors. In some instances, flow can be determined through flow
splitting and proper hydraulic design without an individual flow measurement device for each
UV reactor. Nevertheless, the need for individual flow measurement for each UV reactor is at
the discretion of the State. Utilities implementing UV disinfection are encouraged to discuss
flow measurement requirements with their State during the planning and preliminary design
phases.
Flow Distribution and Control
Two approaches for flow measurement and control have generally been used. The first
involves the installation of a dedicated flow meter and flow control valve for each UV reactor.
The second involves the use of passive flow distribution, with confirmation of equal flow split by
monitoring pressure differential across identical pipe segments (or the UV reactor) or with flow
meters. For identical reactors, the differential pressure across each parallel UV reactor train
should be the same if equal flow distribution is occurring and valves are in the same operating
position. The use of dedicated flow meters and modulating downstream valves to control flow
UV Disinfection Guidance Manual 3-31 June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
through the UV reactors provides the greatest hydraulic control in applications with widely
varying flowrates.
Assuming multiple, parallel UV reactors of the same capacity, the UV reactors should be
sized and configured to provide approximately equal headloss through each treatment train (i.e.,
portion of distribution and recombination channel or manifold, lateral piping, and UV reactor
with associated valves and flow measurement). This is particularly important if passive flow
distribution is used. Because flowrates may deviate from equal distribution, the maximum
design flowrate for each reactor should account for any potential distribution imbalance.
Equation 3.1 can be used to determine the appropriate upper design flowrate for each UV
reactor:
Cfotal
where
Qreactor =
Qtotal =
E
N
N
UV reactor design flow
Plant maximum design flow
Calculated maximum flow distribution error (percent as a decimal)
Number of on-line UV reactors
EquationS.l
The maximum flow distribution error (E) should be determined through hydraulic
calculations or hydraulic modeling of the UV installation. For example, ideally-two identical,
parallel reactors would have a 50/50 flow split. If the actual flow split between the reactors is
calculated or modeled to be 60/40 percent, then a 20 percent (E=0.20) maximum flow
distribution error (E=(60-50)/50=0.2) would be used in the above equation to estimate the proper
design flow for the reactor.
The reactor flow should be estimated over the range of anticipated operating reactors
(i.e., number of operating reactors). In general, with passive distribution, as the number of UV
reactors increases and flowrate decreases, the potential for flow distribution imbalance is
magnified. Effective passive distribution relies on the headloss through each treatment lateral
being significantly greater than the headloss through the common influent manifold or chamber.
Under the conditions of reduced flow and an increased number of operating reactors, the relative
amount of headloss through each lateral becomes less significant when compared to the headloss
through the manifold, resulting in less controlled distribution.
For utilities that use distribution and recombination channels (as opposed to influent and
effluent manifolds), designers typically have two basic choices to achieve passive flow
distribution (Figure 3.8): a series of individual weirs set at the same elevation or a series of
orifices submerged into the individual UV reactor laterals.
UV Disinfection Guidance Manual
Proposal Draft
3-32
June 2003
-------
3. Planning and Design Aspects for UV Installations
Figure 3.8 Open-Channel Flow Distribution Options
Flow
Weir-
D
Plan
Section
A. Flow Splitting Weirs
Plan
Section
B. Submerged Orifices
UV Disinfection Guidance Manual
Proposal Draft
3-33
June 2003
-------
3. Planning and Design Aspects for UV Installations
Flow Measurement
Depending on the design and control strategy of the UV reactor, a number of options are
available for flow measurement. As discussed previously, flow measurement devices installed
specifically for the UV reactors may not be needed in all applications. It should be noted that
some level of inaccuracy or drift is likely to occur with all flow meters. This potential error
should be accounted for during design and validation.
If a single UV reactor is installed, the plant's raw water metering station can be used to
determine a reasonably accurate flow through the reactor. The use of raw water flow metering
data, however, may not account for backwash and residuals flow losses, which would create flow
measurement inaccuracies for UV reactors installed downstream of the filters or clearwell. For
applications where the UV reactor is dedicated to a rate-of-flow control filter, flow information
from the filters may be used to determine the flowrate through the UV reactors.
If equal flow distribution between multiple UV reactors can be achieved passively under
all hydraulic conditions, a single, common flow meter (new or existing) may be used to measure
flow. The total flow can then be divided by the number of operating UV reactors to determine
the flow through each UV reactor. If this approach is selected, some method of confirming the
equal flow split should also be incorporated (e.g., differential pressure measurement).
A single flow meter for the entire UV installation or individual meters (with or without
rate-of-flow control) should be considered to provide increased flow measurement accuracy.
Magnetic flow meters or other meter types, such as doppler, that do not protrude into the flow
path have the least effect on the velocity profile, which minimizes the potential effect on reactor
inlet or outlet hydraulics. The desired means of flow measurement for the UV reactors should be
selected based on the level of flow measurement accuracy needed to accomplish the operating
and control strategy for the installation and satisfy validation criteria, as well as an understanding
of the variability in the plant flowrate. Several options are listed in Table 3.7 and are illustrated
in Figure 3.9.
UV Disinfection Guidance Manual 3-34 June 2003
Proposal Draft
-------
§
I
1
0)
2
Q)
1
U.
o
(A
60
e
I
"o
c
o
CO
a.
O
IV.
«
ra o
CO r- o
S « "P
OJ 3 'C
Q. co o
V
e>
if
<=
l|,C
"O (0 w
£->.£
2
g>
l€
l
|D
Is
Q 8.
-------
3. Planning and Design Aspects for UV Installations
Figure 3.9 Flow Measurement and Control Options
Raw Water
Flowmeter
A. Raw Water Meter (or Flow Measurement
Raw Water
Flowmeter
B. Single UV Disinfection Facility Meter for Flow Measurement
Flowmeter
Raw Water
Flowmeter
~M
C. Individual UV Reactor Flow Measurement and Control
Individual
Flowmeters
3.3.13
Level Control
The UV reactors must be flowing full at all times during operation. Therefore, the
reactors should be placed below the hydraulic grade line elevation. There are two basic options
commonly used to maintain the level in the reactors. One option is with a fixed downstream
weir; in many WTPs, a fixed weir is already located in a clearwell and can be used for this
purpose. If not, another option is to install a weir immediately downstream of the UV reactor or
at another location that ensures full pipe flow through the UV reactors. A final option is to use
flow control valves to monitor and maintain the downstream hydraulic grade line.
3.3.1.4
Air Relief and Pressure Control Valves
UV reactors should be kept free of air to prevent lamp overheating. The formation of
negative pressures or surge effects within the UV reactors should also be prevented to avoid
damage to the lamp sleeve and UV lamps. The use of air release valves, air/vacuum valves, or
combination air valves may be appropriate to prevent air pockets and negative pressure
UV Disinfection Guidance Manual 3-36 June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
conditions. The locations of the valves will be dictated by the specific configuration of the
installation and should be determined during design.
3.3.1.5 Flow Control and Isolation Valves
Each UV reactor should have the capability of being isolated and taken out of service.
This will necessitate a valve, gate, or other isolation device upstream and downstream of the UV
reactor. Valves are generally preferred, since they provide a tighter seal. Utilities that use
passive flow distribution will rely on the valves primarily for isolation and sequencing of UV
reactor operation (as opposed to flow control). Valves downstream of the UV reactor should be
equipped with an actuator to automatically open or close on a critical alarm occurrence and to
enable start-up sequencing.
If the isolation valves are used for flow control, either the upstream or the downstream
valve can be used. However, it is generally recommended that the valve downstream of the UV
reactor be used to minimize disturbance of the flow entering the UV reactor, particularly if the
separation distance between the upstream valve and the UV reactor is relatively small. The flow
characteristic curve of the valve and the operating.speed of the actuator should be matched to the
flow control needs of the UV reactors. During design, the valve configuration should be
discussed with the UV manufacturer to ensure that UV reactor performance will not be adversely
affected by the location or operation of the valves. It is important to coordinate the location of
the valves with the validation conditions for the reactor, as discussed in section C.3.1.1.
Valve seats and other in-pipe seals and fittings within the straight pipe lengths adjacent to
the UV reactors should be constructed of materials that are resistant to UV light to avoid
degradation. If in-place rehabilitation of existing piping is used to improve system hydraulics,
the materials used to slip-line or reline the piping adjacent to the proposed UV reactors should
also be resistant to degradation from exposure to UV light. Organic materials and plastics that
have not incorporated UV-resistant additives are typically most susceptible to UV degradation.
3.3.1.6 Intermediate Booster Pumps
A detailed evaluation and design of a booster pumping system is recommended if it is
determined during the planning phase that head constraints necessitate the installation of a
pumping system. Pumps common in water treatment plants (i.e., vertical turbine, end-suction
centrifugal, and split-case centrifugal pumps) tend to have higher discharge pressures than
intermediate pumping applications and are generally not appropriate for this application. Mixed
flow or axial flow pumps with high-flow and low-head operating characteristics are typically
more appropriate. However, additional headloss may need to be added to the system, based on
the capabilities of the pump. Smaller diameter piping, backpressure valves, or control valves can
be used to increase the system head to more closely match the pump discharge curve.
Pumps may be installed before or after the UV reactors, allowing more flexibility in the
UV installation's design elevations and the location of the UV reactors. Regardless of pump
location, some form of wetwell should be provided upstream of the pump station. Existing
clearwells, recombination channels, or dedicated pump wetwells may be used. Direct connection
UV Disinfection Guidance Manual 3-37 ' June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
clearwells, recombination channels, or dedicated pump wetwells may be used. Direct connection
to filter effluent piping may adversely affect upstream process performance and should be
avoided. Booster pump operation may be controlled by the water level within the upstream
wetwell. The use of variable frequency drives (VFDs) to moderate flow peaks is recommended.
This is especially important if the pump station is upstream of the UV reactors. By minimizing
hydraulic peaks, the UV reactors can be sized to more closely match the flow through the WTP.
If pumps>are located adjacent to the UV reactors, the impact of surge conditions should
be evaluated. Of particular concern is the potential for surge if the pumps are operating and
power is lost. Pump start-up procedures should be carefully selected with possible inclusion of
pump control valves. Control of individual UV reactor isolation valves should be coordinated
with pump starts and stops and with pump control valves where appropriate. Likewise, the
warm-up time associated with the start-up of the UV reactors must be taken into account with the
sequencing of the pump operation.
3.3.2 Operational Strategy Determination
Once the UV reactors are procured, a utility should determine the preferred operational
strategy given the UV reactor's control strategy and available validation data. The different
operational strategies are described in section 3.1.5, and an example of how to interpret the
validation data to develop an operational strategy is described in section 5.5.
The power needs for UV reactors can be moderately high, and an inefficient UV
installation can result in unnecessarily high operating costs. When considering what operational
strategy to use for a particular installation, the operational complexity should be compared to the
potential for energy savings. It should be noted, however, that intensity adjustment does not
correlate directly to the amount of energy that is saved. Lamp output efficiency may decrease as
the lamp intensity is reduced, resulting in a reduced energy savings. Lamp output efficiency as a
factor of intensity should be discussed with the UV manufacturer and considered when
determining the potential cost savings associated with dose pacing. An operational strategy
consistent with the procured UV reactor should be selected to facilitate the instrumentation and
control design.
3.3.3 Instrumentation and Control
After the hydraulic needs of the UV reactors have been addressed and a dose control
strategy has been selected, the instrumentation and controls necessary to satisfy both can be
identified. The level of instrumentation and control that is needed will depend on the flow
control, flow distribution, and flow measurement approach that is selected, as well as the dose
control strategy that is employed. Passive flow distribution with an intensity setpoint dose
control strategy is a relatively simple operation and demands limited instrumentation and control.
Operating flexibility and the ability to optimize UV reactor energy efficiency, however, are
reduced. The use of dedicated flow meters and flow control valves, in combination with on-line
transmittance monitors and dose pacing, demands a higher level of instrumentation and control.
However, this approach provides significant operating flexibility and the ability to optimize
UV Disinfection Guidance Manual 3-38 June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
The specific instrumentation and control elements included with the UV reactors may not
be known until a final UV reactor selection is made. Most of the equipment manufacturers,
however, share common instrumentation and control attributes and alarm conditions in the
designs of their UV reactors. To enable a procurement document to be prepared, a control
strategy should be established. To the extent practicable, the designer should identify the
elements of the control system that are preprogrammed into the UV reactor control panel and
those that will be addressed through the installation of supplemental controls and equipment. At
a minimum, the LT2ESWTR requires that UV lamp intensity, flo'wrate, and lamp status be
monitored (40 CFR 141.729(d)), The final instrumentation and control design can be modified
as needed after equipment is selected.
3.3.3.1 , UV Reactor Start-up
Regardless of the UV reactors that are selected, the start-up cycle will likely be the same.
For a UV reactor that is starting cold (i.e., previously off as opposed to shutdown for a very short
period due to power interruption), a typical control sequence will open the isolation valves, ignite
the lamps, and bring the lamps to full power. During the typical control sequence, the water
being treated will be off-specification until the lamps reach full operating power, which can take
up to 10 minutes. However, the amount of off-specification water can be reduced by providing a
low flow of cooling water that can be discharged to waste. Alternatively, if a LP or LPHO
reactor is procured, the downstream valve may remain closed as the UV lamps are.warming up.
However, the designer should consult the LP or LPHO manufacturer to ensure this strategy is
feasible. It is recommended that the utility discuss these practices with the State to confirm their
acceptance.
3.3.3.2 UV Reactor Automation
Depending on the size and complexity of the UV reactor, its operation can range from
manual to fully automatic. Manual operation includes manual initiation of lamp start-up and
shut down, and appropriate valve actuation. Different levels and types of automation can be
added to the manual sequence. A first level of automation includes the sequencing of lamp start-
up and valve actuation to bring individual UV reactors on-line after manual initiation. Further
levels of automation could include starting up UV reactors, activating rows of lamps, or making
lamp intensity adjustments based on lamp condition, water quality, and/or flowrate.
Automatic UV reactor shutdown under critical alarm conditions (e.g., high temperature,
lamp or sleeve failure, loss.of flow) is important for all operating approaches, including manual
operation. The shutdown cycle will be site-specific. However, to the extent practicable, the
downstream flow control or isolation valve should be closed whenever the UV reactor is shut
down to minimize the distribution of water that has not been disinfected by the UV installation.
3.3.3.3 UV Intensity and Calculated Dose (If Applicable)
Signals from UV intensity sensors should be displayed locally or on the UV reactor
control panel. Because the output from the UV intensity sensor is integral to the determination
UV Disinfection Guidance Manual 3-39 , June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
of adequate dose delivery, the UV intensity sensor output should be monitored continuously. If
the calculated dose control strategy is used, the calculated dose should be displayed locally and
be monitored continuously.
3.3,3,4 UV Transmittance
An on-line UVT monitor or bench-top spectrophotometer may be used to monitor UVT,
depending on the control strategy (section 3.1.4.2). An on-line UVT monitor is typically used
for the UV intensity and UVT setpoint approach and the calculated UV dose setpoint approach.
However, for utilities that have water with a stable UVT, periodic grab samples may be
adequate. Results from a bench:top spectrophotometer can be manually input into a SCADA
system or other control system. Output from an on-line UVT monitor can be input directly into a
control loop for most UV reactors, a SCADA system, or both.
If the UV intensity setpoint approach is used, UVT does not need to be monitored
because the UVT is accounted for in the UV intensity measurement. However, it may be
advantageous to monitor UVT with an on-line UVT monitor or bench-top unit to assist with
troubleshooting UV reactor performance issues.
The specific size and operating characteristics of the UVT monitor will vary dependent
on the UV manufacturer. If an on-line UVT monitor is included in the design, it is important to
provide adequate space and access to an electrical supply for installation of the monitor and to
include appropriate sample taps and drains in the design for the withdrawal and discharge of
sample water. The sample line should be equipped with a valve to isolate the unit. If insufficient
pressure is available in the system, then a sample pump should be installed.
3.3.3.5 Flow Measurement
Flowrate is one of the operating conditions a utility must routinely monitor (40 CFR 141,
Subpart W, Appendix D). To maintain regulatory compliance, the flowrate through a UV reactor
must be known to ensure that flow is within the validated range. Section 3.3.1.2 discusses flow
measurement and control options. If flow meters are installed, the flow signal should be
displayed locally or be input directly into a control loop for the UV reactor, a SCADA system, or
both.
3.3.3.6 Lamp Age
Each lamp or an integral bank of lamps should be monitored for operating time. Lamp
replacement should be based on the dose delivery and the age of the lamp. Initially, the number
of lamp hours used to trigger lamp replacement can be estimated based on UV manufacturer
recommendation and validation data. Later, the actual frequency of replacement should be
correlated to the operating performance of the UV installation. Frequent restarting of the lamps
may reduce their useful life.
UV Disinfection Guidance Manual 3-40 ' June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
3.3.3.7
Lamp and Reactor Status
Lamp status is one of the operating conditions a utility must routinely monitor (40 CFR
141, Subpart W, Appendix D). In addition to the status of individual lamps, whether the reactor
is on-line or off-line should also be monitored and indicated locally and remotely. Power and
valve status are two methods that utilities can consider to perform such monitoring.
3.3.3.8
Alarms and Control Systems Interlocks
Many UV reactor signals and alarms are specific to the UV installation and the level of
automation employed. Alarms may be designated as minor, major, or critical, depending on the
severity of the condition being indicated. The same alarm condition may represent a different
level of severity dependent on the conditions under which the UV reactor was validated, the type
of UV reactor, the control strategy, and the disinfection objectives of the utility. For example, if
a UV reactor was validated with one lamp out of service, a single lamp failure alarm may be a
minor alarm. Had the reactor been validated with all lamps in operation, then a single lamp
failure may be a major alarm. At a minimum, alarm conditions should be displayed locally. The
use of an audible alarm may be beneficial. If UV reactors will frequently be unstaffed,
provisions should also be included in the design to allow remote monitoring.
A minor alarm generally indicates that a UV reactor needs maintenance but that the UV
reactor is not operating out of compliance. For example, a minor alarm would occur when the
end-of-lamp-life is reached, indicating the possible need for lamp replacement. A major alarm
indicates that the UV reactor needs immediate maintenance (e.g., the UV intensity sensor value
has dropped below the validated setpoint) and that the unit may be operating off-specification.
Based on the utility's disinfection objectives, this condition may also be handled as a critical
alarm. A critical alarm typically shuts the unit down until the cause of the alarm condition is
remedied. An example of a more typical critical alarm is the UV reactor temperature exceeding
a pre-determined maximum value, resulting in automatic shutdown to prevent overheating and
equipment damage.
The designer should work with the UV manufacturer to determine what elements of the
control system are integral to the UV reactor and what will be addressed through the installation
of supplemental controls and equipment. For installations with multiple UV reactors, a common,
master control panel may be necessary to enable sequencing of the UV reactors and to allow the
UV reactor operations to be optimized. Table 3.8 summarizes typical UV reactor monitoring and
alarms; additional detail is provided in section 5.4. Many of the alarms shown will be integral to
the UV reactor control panel.
UV Disinfection Guidance Manual
Proposal Draft
3-41
June 2003
-------
3. Planning and Design Aspects for UV Installations
Table 3.8 Typical Alarm Conditions for UV Reactors
Alarm/Sensor
Lamp Age
Calibrate UV Intensity Sensor
Differential Pressure Out of
Range
(When Differential Pressure is
Used for Flow Split
Confirmation)
Low UV Dose
Low UV Intensity
Low UV Transmittance
High/Low Flow
Lamp/Ballast Failure
Low Liquid Level '
High Temperature
Mechanical Wiper Function
Failure
Purpose/Descriptions
• Minor alarm occurs when run-time for lamp Indicates end of
defined operational lamp life.
» Minor alarm occurs when UV intensity sensor needs
calibration based on operating time.
• Necessary only if a single master flow meter is used.
• Minor alarm occurs if pressure drop across parallel, identical
UV reactors indicates unequal flow split.
• Major alarm occurs if differential pressure across a given UV
reactor indicates flow outside of the validated range.
• Major alarm occurs when dose condition falls below required
dose.
• Triggered by signals gathered by control system and
compared to validated UV reactor dose requirements.
• Major alarm occurs when intensity falls below design
conditions.
. Major alarm occurs when UVT falls below design conditions.
• Major alarm occurs when flowrate falls outside of validated
range.
• Based on measurement from dedicated flow meters or
calculated based on total flowrate divided by number of units
operating.
• Major alarm occurs when a single lamp/ballast failure is
identified.
• Critical alarm occurs when multiple lamp/ballast failures are
identified.
• Critical alarm occurs when liquid level within the UV reactor
drops and potential for overheating increases.
. Critical alarm occurs when the temperature within the UV
reactor or ballast exceeds a setpoint.
• Needed only if a mechanical wiper system is used.
• Critical alarm occurs if wiper function fails.
Note: Alarm conditions and relative severity shown above may vary dependent on specific conditions under which
the UV reactor is validated, the type of UV reactor, the control strategy, and the disinfection objectives of the
utility.
3.3.4 Electrical Power Configuration
The electrical power configuration that is used should take into account the findings of
the power quality assessment conducted during the planning phase described in section 3.1.3.3,
the power requirements of the selected equipment, and the disinfection objectives and control
strategy of the utility. Issues that should be addressed include harmonic distortion and off-
specification operation due to power quality problems (fluctuation in line voltage).
UV Disinfection Guidance Manual
Proposal Draft
3-42
June 2003
-------
3. Planning and Design Aspects for UV Installations
3.3.4.1 Power Requirements
The proper supply voltage and total load requirements must be coordinated with the UV
manufacturer, considering the available power supply. In addition, the power needs for each of
the UV reactor components may differ. For example, the UV reactors may require a 3-phase,
480-volt service while the on-line UVT monitor may need a single phase, 110-volt service. The
method of handling the power feed must be carefully coordinated to ensure all electrical
equipment and services are included and to clearly establish the responsible party for each
element of the electrical supply (e.g., primary service, transformer, secondary-service).
Excluding high service pumping, the electrical load from UV reactors will typically be one of the
larger loads at the WTP.
Due to the varying nature of UV reactor loads, current and voltage harmonic distortion
can be induced. Such disturbances can result in electrical system problems, including
overheating of some power supply components and effects on other critical systems such as
VFDs, program logic controllers (PLCs), and computers. Proper selection of the UV reactors,
including a thorough analysis of the potential for the equipment to induce harmonic distortion,
should minimize the potential for harmonic distortion. Another method for controlling
harmonics is to use a transformer with Delta Wye connections to isolate the UV reactors from
the remainder of the WTP power system. The delta-connected primary feed could be designed
and sized to trap and moderate any induced harmonics. The Wye-connected secondary should
be solidly grounded so that the ballasts are powered from a grounded source in accordance with
electrical code requirements. If a separate transformer for the UV reactors is impractical,
harmonic filters could be added to the UV reactor power supply to control distortion. Regardless
of the method used to address harmonic issues, electrical acceptance testing during start-up
should include a harmonic analysis to verify that the UV reactor harmonics are not affecting
other electrical components at the WTP.
3.3.4.2 Backup Power Supply
The continuous operation of the UV reactors is highly dependent on its power supply.
This dependence, when combined with the sensitivity of the UV reactors to power fluctuations,
increases the importance of a high quality, dependable power supply. The utility should work
with the State to establish specific power reliability objectives for the UV installation, as power
reliability may directly affect the utility's ability to meet the State's allowable off-specification
requirements. As discussed in section 3.1.3.3, minor power transients can lead to lamp outages.
If the power reliability objectives, and, consequently, the disinfection objectives, cannot be met
solely by relying on the commercial power supply, then the use of a backup power supply (i.e.,
backup generator, separate commercial service, and/or battery-supported UPS) may be
necessary. If an existing backup power supply is in place, the load capacity of this supply should
be assessed to determine if it is able to accept the additional load associated with the UV
reactors. Additionally, the time needed to transfer from the primary power supply to a backup
power supply and the potential effect of the transfer time on compliance with the State's
allowable off-specification operation should be determined.
An alternate backup power supply may be needed if a backup power supply is not in-
place or the available load capacity is insufficient to handle the new load associated with the UV
UV Disinfection Guidance Manual 3-43 June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
reactor equipment. The type of backup power supply that is needed will depend on the
frequency and duration of the power interruptions and the potential for those interruptions to
result in off-specification operation of the UV reactors. If power quality issues are infrequent
and of short duration (on the order of seconds or minutes), it is possible that a backup power
supply may be unnecessary, or a simple backup power supply may be sufficient. If the
frequency of power outages increases or the duration of the outages increases, the need for a
more extensive backup power supply becomes more significant.
If a backup power supply is necessary, but continuous power is not needed, the use of a
traditional diesel or natural gas-fired backup generator set, a standby UPS, or a rotary UPS may
be adequate. Should a continuous power supply be needed to meet reliability objectives, the use
of a line-interactive UPS will be necessary. The line-interactive UPS provides a continuous
power supply, but is generally less efficient, has a lower starting current, and costs more than a
similarly sized*standby UPS. Typically, a line-interactive UPS would be installed in conjunction
with a backup generator to provide a cost-effective backup power supply for longer duration
power interruptions or for frequent, shorter duration power interruptions. Although unlikely to
be a requirement for compliance monitoring, it may be beneficial to include a data logger that
records instances of UPS operation as part of the UPS system design.
The elements that should be considered when assessing the need for a backup power
supply for a UV reactor are somewhat unique when compared to those associated with more
typical WTP equipment; However, once it is determined that a backup power supply is
necessary, the design for a UV reactor is very similar to that for any other equipment or
treatment process. Factors that should be considered during design include isolation, in-rush
current, purchase and installation cost, maintenance requirements, voltage regulation, electrical
surge protection, attenuation of harmonic current, run-time, transformer continuity, and the
ability to operate with other power supply equipment. In most circumstances, a UPS should not
be used without a backup generator because of the battery reserve necessary to power a UV
installation for longer durations. To minimize capital cost, the battery reserve time should be
sufficient to allow the power supply to switch to the backup generator.
3.3.4.3 Ground Fault Interrupt and Electrical Lockout
Ground fault interrupt (GFI) is an important safety feature for any electrical system in
contact with water, including UV reactors. All UV reactor suppliers should provide GFI circuits
for their lamps, which should be included in the specifications that are developed for equipment
procurement. For a GFI to function properly, the transformer in the UV reactor ballast must not
be isolated from the ground. If the UV reactor ballast isolates the output from the ground,
ground faults will not be properly detected, and safety can be compromised.
Provisions enabling the UV reactors to be isolated and locked out for maintenance, both
hydraulically and electrical|y, should be included in the design. Control of all lockout systems
should remain local; however, when appropriate, the status of local lockouts could be monitored
remotely. In all cases, the design must comply with electrical code and policy requirements for
equipment lockout. ~
UV Disinfection Guidance Manual., 3-44 June 2003
Proposal Draft ' '
-------
3. Planning and Design Aspects for UV Installations
3.3.5 UV Installation Layouts
Once the previous design elements (i.e., section 3.3.1 through section 3.3.4) have been
evaluated, an installation layout can be prepared as part of the equipment procurement document.
The layout should take into account the findings of all previous work. Because the design
process is iterative and many elements of the layout are dependent on the specific UV reactors
that are used and the validation scenario that is proposed, the layout may change after the UV
reactors are selected and any additional space constraints are identified.
3.3.5.1
Site Layout
Site layout for a UV installation is generally similar to the layout of any treatment
process. When locating the UV installation, access for construction, .operation, and maintenance
should be addressed. The availability of adequate existing infrastructure (e.g., power, drains,
lifting devices) is also important. In general, when compared to other treatment processes at a
WTP, the UV installation has a relatively small footprint.
3.3.5.2
UV Installation Layout
In large part, the piping layout will be dictated by the validated hydraulic conditions
because the inlet and outlet conditions for the installed UV reactors should be equal to or better
than the hydraulic conditions used during validation. Additional details on the relationship
between the validated inlet and outlet configuration and the actual installed configuration are
given in section C.3.1.5. Nevertheless, the designer can prepare a reasonable UV installation
layout based on the type of technology (i.e., LPHO versus MP), the number of UV reactors that
is needed, and the manner in which flow is controlled and measured. This layout can then be
used in the selection and procurement of the UV reactors.
Most UV reactors available for drinking water applications are of a closed-chamber type.
Filtered water is conveyed via pipes or covered channels to a series of UV reactors for primary
disinfection. As such, laying out UV installation typically involves designing the method by
which water is divided between UV reactors (channel or piping), and routing the sections of pipe
between inlet and discharge headers in which the UV reactors are inserted via flanged
connections (although other types of connections may be used). The number and configuration
of the UV reactors will vary depending on lamp type/reactor design, reactor size, flow range to
be treated, control strategy, and degree of redundancy.
Although most components of UV reactors are fairly compact, it is important not to
underestimate the necessary space for the building that will house the UV installation. In
addition to those items identified in section 3.1.6.2, the following factors should be considered in
the layout for the UV installation: • .
» The length of straight-run piping before and after each flow meter to achieve the
proper hydraulic conditions for accurate and repeatable flow measurement (if
applicable)
UV Disinfection Guidance Manual
Proposal Draft
3-45
June 2003
-------
3. Planning and Design Aspects for UV Installations
• Field instrumentation
>i
« Isolation valves and flow control devices
• Control and power panels, and code-required clear space
• Potential space for power monitors and UPS systems
. Drain provisions for the process area and to permit UV reactor draining
• Provisions for future expansion of UV disinfection capacity
Components of the UV reactors are typically located inside a building for protection from
the weather and to provide a clean, convenient area for maintenance. The UV reactors
themselves, associated electrical components and controls, and electrical support equipment such
as a UPS should be enclosed. There are installations, however, where UV reactors and control
panels are uncovered. Prior to implementing an uncovered installation, it is recommended that
the State and UV manufacturer be consulted. Any exposed equipment and control panels should
be rated for the anticipated environment and appropriate site security should be in-place to
restrict public access.
The power and control panels associated with UV reactors should be located so that there
is adequate space for panel doors to be opened without interference, and to allow unhindered
access to the UV reactors with panel doors open. In selecting the location of the power and
control panels, UV manufacturer cable length limitations should not be exceeded. The
maximum allowable cable length is UV manufacturer-specific and may be less than 30 feet. If
harmonic feedback is a concern, extra room should be provided for power conditioning
equipment.
When allotting space for maintenance activities, adequate space to remove the lamps and
the lamp wiper assembly should be provided. In some cases, access may be needed on both
sides of the UV reactor. In addition, provisions should be included to collect and convey water
that is discharged during maintenance activities.
• Certain UV reactors need maintenance involving an OCC procedure in which a UV
reactor is taken off-line, isolated, drained, filled with a cleaning solution, cleaned, flushed, and
returned to service. The QCC equipment is typically self-contained and the cleaning chemical is
recirculated. Where applicable, sufficient space around the UV reactors should be maintained to
provide access for the OCC procedure. In addition, the OCC solution often has specific handling
requirements., Appropriate drains, storage, and health and safety equipment (e.g., emergency
eyewash station) should be provided as recommended by the chemical manufacturer.
Sample taps are recommended upstream and downstream of each UV reactor within the
lateral pipe. The sample taps may be used for the collection of water quality samples or may be
used during validation testing if on-site validation is necessary. If on-site validation will be
conducted, the number and location of sample and feed ports should be coordinated with the UV
manufacturer or third party validation service to comply with the recommendations of the
selected validation protocol. Additional detail on the locations of sample taps and other
UV Disinfection Guidance Manual 3-46 June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
validation-related appurtenances, as well as the methods used to validate a reactor are provided
in section C.3.1.4.
Drain valves or plugs should be located on each lateral between the two isolation valves.
In many cases, the UV manufacturer may have already incorporated a drain into the UV reactor
design. Drain valves should also be provided at one or more low points in the UV installation to
enable the UV reactor to be fully drained for future maintenance activities.
3.3.6 Elements of UV Reactor Specifications
Table 3.9 summarizes the elements that should be considered in developing equipment
specifications for the UV reactors. The information included in Table 3.9 is not exhaustive and
should be modified to meet the specific needs of the utility.
UV Disinfection Guidance Manual
Proposal Draft
3-47
June 2003
-------
3. Planning and Design Aspects for UV Installations
Table 3.9 Recommended Content for UV Reactor Specifications
Specification Item
Flowrate
UV Dose
Water Quality and
Environment
UV Intensity Sensors
Redundancy
Hydraulics
Size/Location
Constraints
Validation
Control Strategy and
Operating Sequence
Lamp Sleeves
Purpose/Description
Maximum, minimum, and average flowrates should be clearly identified.
The minimum and maximum flowrates must be within the range of
validation flowrates. The minimum flowrate is important to avoid
overheating with MP reactors.
The required reduction equivalent dose as well as the validation
technique that will be used to measure the dose should be established.
Additional detail is provided in Chapter 4.
The following water quality criteria should be included:
- Influent temperature - Total Hardness
- Turbidity - pH
UV Transmittance at 254 nm - Iron
- Spectral absorbance 200-300
nm (MP reactors only)
For some parameters, a design range may be most appropriate.
It is recommended that at least one UV intensity sensor be specified per
UV reactor. The number of reference sensors that should be determined
based on the time and labor associated with checking and maintaining
the duty sensors.
If a combined filter effluent UV reactors are used, it is recommended that
at least one completely redundant UV reactor be specified as a standby.
For other configurations, the designer should determine the appropriate
redundancy based on the State's requirements and the utility's
disinfection objectives.
The following hydraulic information should be specified:
Maximum system pressure at the UV reactor
- Maximum allowable headloss through the UV reactor
Special surge conditions that may be experienced
Hydraulic constraints based on site-specific conditions and
validated conditions (e.g., upstream and downstream straight pipe
lengths)
Any size constraints or restrictions on the location of the UV reactor or
control panels (e.g., space constraints with in-line installation).
The specifications should establish the validation protocol that will be
followed, provide the conditions under which the validation will be
conducted (e.g., water quality, flow range, hydraulic conditions, UVT),
and require the submittal of a validation report (40 CFR 141 .730).
The specification should provide a narrative description of the operating
sequence and control strategy for the UV reactors.
At a minimum, the following items should be specified:
Lamp sleeves should be annealed to remove internal stress.
- UV manufacturer should perform QA / QC checks of a fraction of
each lot using a polarized light or other approved method.
- , UV manufacturer should submit documentation on the integrity of
their sleeve, monitoring practices, and rationale for using a given
internal QA / QC frequency.
UV manufacturer should submit calculations showing the maximum
allowable pressure for the lamp sleeves and the maximum bending
stress experienced by the lamp sleeves under the maximum
specified flow conditions.
UV Disinfection Guidance Manual
Proposal Draft
3-48
June 2003
-------
3. Planning and Design1 Aspects for UV Installations
Table 3.9 Recommended Content for UV Reactor Specifications (continued)
Specification item
Safeguards
Control Systems
Performance
Guarantee
Warranties
Pu rpose/Description
At a minimum, the following UV reactor alarms should be specified:
Lamp or ballast failure
Low UV intensity or low UV dose (dependent on control strategy
used)
High temperature
Low or high flow
Wiper failure (as applicable)
Other alarms discussed in section 3.3.3.8, as appropriate
At a minimum the following signals and indications should be specified:
UV reactor status
- UV intensity j
- Individual lamp status
Lamp cleaning cycle and history
Accumulated runtime for individual lamps
Influent flowrate
At a minimum the following* UV reactor controls (as applicable) should be
specified: j
- UV dose setpoints, lamp intensity setpoints, or UVT setpoints
(dependent on control strategy used)
- UV reactor on/off control
UV reactor manual/auto control
UV reactor local/remote control
Manual lamp power level control
Manual lamp cleaning cycle control
Automatic lamp cleaning cycle setpoint control
The performance guarantee should specify that the equipment provided
under the UV reactor specification should meet the performance
requirements stated in' the specification for an identified period. The
following specific performance criteria may be included:
- Allowable headloss at each of the design flowrates.
Estimated power consumption under the design operating
conditions. ]
Disinfection capacity of each reactor under the design water quality
conditions. 1
A physical equipment guarantee and UV lamp guarantee should be
specified. The specific requirements of these guarantees will be at the
discretion of the utility and engineer. .
3.3.6.1
Information Provided by Manufacturer in UV Reactor Bid
It is important that UV manufacturers provide adequate information when bidding to
enable the designer to conduct a proper, timely review of the proposed equipment. Suggested
information to be obtained from the UV manufacturer is presented in Table 3.10.
UV Disinfection Guidance Manual
Proposal Draft
3-49
.
June 2003
-------
3. Planning and Design Aspects for UV Installations
Table 3.10 Recommended Information to be Provided by UV
Manufacturer/Vendor
Item
Design
Parameters
Summary of
Design
Reactor
Technical
Specifications
UV
Manufacturer's
Experience
UV Intensity
Sensor
Validation
Data
Upstream and
Downstream
Hydraulic
Requirements
Power
Requirements
Cleaning
Strategy
Control
Strategy
Reactor Data
Safeguards
Warranties
Purpose
Demonstration of an understanding of the design parameters for the UV reactors.
All UV reactor design parameters from the contract documents should be repeated
in the proposed UV reactor submittal information.
A summary of the equipment proposed (number of UV reactors, lamp type) and ,
specify equipment redundancies.
Ability of proposed UV reactors to meet technical specifications and an explanation
of any exceptions taken.
Information on project experience, including previous installations and references.
Information on the UV intensity sensor(s) including acceptance angle, external
dimensions, working range in mW/cm2, spectral response, measurement
uncertainty, environmental requirements, linearity and temperature stability. Data
and calculations should be provided showing how the total measurement
uncertainty of the sensor is derived from the individual sensor properties.
(See sections 4.3.2.3 and C.4.7 )
UV reactor validation data as described in Appendix C of these Guidelines. If on-
site validation is proposed, validation data for the UV reactors from other, similar
installations should be included to provide a baseline comparison to the proposed
operating conditions.
A statement of the length of straight pipe and hydraulic conditions necessary
upstream and downstream from the UV reactor to ensure the desired flow profile is
maintained and the design conditions are met.
The power needs of each UV reactor and which elements, including electrical cable
and wiring, are included as part of their equipment.
The strategy that will be employed for cleaning the UV lamps in the UV reactor.
The proposed UV reactor control strategy, including manual and automatic control
schemes and a listing of inputs, outputs, and the types of signals that are available
for remote monitoring and control.
The materials of construction, dimensions of the UV reactors and ancillary
equipment, a listing of spare parts, and a sample operations and maintenance
manual.
The safeguards built into the UV reactor and accompanying equipment, such as
high temperature protection, wiper failure alarms, and lamp failure alarms.
A statement of the proposed UV reactor guarantees, including the physical
equipment, the UV lamp, and the system performance guarantee. Any exceptions
should be indicated and explained.
Warranties •
The UV reactor specification should include suitable written guarantees regarding ,
physical equipment, UV lamps, and performance.
It is recommended that the UV lamp guarantee specify that each lamp is warranted to
provide the lamp output necessary to meet the required reduction equivalent dose (RED) under
UV Disinfection Guidance Manual 3-50 • •-' June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
the design conditions for a minimum number of operating hours, which will vary depending on
lamp type.. To limit the UV manufacturer's liability and to potentially reduce the contingency
costs included in their equipment bid prices, the guarantee could be prorated after a specified
number of operating hours. It is important that the appropriate lamp fouling/aging factor be
included in the design conditions as discussed in section 3.1.3.1. If these specifications are not
met, significant operation and maintenance costs may occur because lamps may need to be
replaced frequently for the UV reactors to operate within the validated range. The combination
of lamp fouling/aging factor and the guaranteed lamp life will make the UV manufacturer
responsible if the UV lamps do not meet these specifications. The guaranteed lamp life will
depend on the available technology at the time of the UV installation design and will likely
change as lamp technology improves.
3.3.7 Final UV Installation Design
After the equipment procurement document is developed and competitively bid, and all
bids have been carefully reviewed, the UV reactors can be selected. Once the UV reactors are .
selected, the designer can work with the selected UV manufacturer to develop the final
disinfection installation design based on the specific needs and design of the selected equipment.
The hydraulic design, instrumentation and control design, electrical design, and installation
layout should be modified to address the specific needs of the selected equipment and to ensure
that the control strategy can be implemented within the constraints established during the
validation testing.
Particular emphasis should be given to the integration of the overall control strategy with
the alarms, signals, and interlocks that are integral to the UV reactor design. For designs with
multiple UV reactors, a master control panel may be necessary to enable the sequenced operation
of the individual UV reactors and to optimize the efficiency of the UV installation. It is critical
that the final design be coordinated with the validation testing to ensure that validation criteria
are sufficient to implement the proposed control strategy and to ensure that the UV reactors will
meet the utility's disinfection objectives under the anticipated operating conditions.
3.3.7.1 ' Design Drawings
The drawings may include the following content:
» Existing conditions
• Site work
« Structural work
» Architectural work :
. Mechanical work (heating, ventilation, and air conditioning)
UV Disinfection Guidance Manual 3-51 . > June 2003
Proposal Draft f
-------
3. Planning and Design Aspects for UV Installations
» Electrical work
. Instrumentation work |
3.3.7.2 Specifications
The content of the specifications will vary, dependent on the complexity and size of the
UV installation and the selected method of project delivery. However, it is likely that portions of
nearly all of the 16-Division Construction Specifications Institute (CSI) MasterFormat may be
necessary. For those UV installations that pre-purchase the UV reactors, the equipment
procurement document should be included as an appendix to the specifications to facilitate
contractor review and installation of the equipment.
3.4 Reporting To The State
Interaction with the State throughout the planning and design phases, as well as during
development of the reactor validation protocol, is recommended to ensure that the objectives of
both the utility and the State are met.
Given the relatively limited past use of UV disinfection in drinking water treatment and
the unique technical characteristics of this technology, State agencies may not have developed
approval requirements specifically for UV disinfection. This section provides guidance on the
information that may be included in submittals to the State, Utilities are urged to consult with
their State early in their UV disinfection planning process to understand what approvals and
documentation will be required for the use of UV disinfection.
3.4.1 Planning
The State may require that a pre-design report be submitted that summarizes the decision
logic used to identify, evaluate, and select UV disinfection. 'Appendix K is an example pre-
design report, including installation alternatives and analysis. The following items may be
addressed in the pre-design report:
. Disinfection objectives (target organism and inactivation)
. Overall disinfection strategy
« Summary of reasons for incorporating UV disinfection
• Description of the overall process train
. Description of the proposed UV reactors
. Water quality data
UV Disinfection Guidance Manual 3-52 June 2003
Proposal Draft
-------
3. Planning and Design Aspects for UV Installations
• UV reactor reliability targets (i.e., off-specification limits)
• UV reactor validation
3.4.2 Equipment Procurement
If the utility pre-purchases the equipment, a separate procurement document would be
prepared. The equipment procurement document should be consistent with the pre-design report
and should include technical specifications and a preliminary layout of the UV installation.
Details on the recommended content of the specifications are given in section 3.3.7. While the
State may not require submittal of the equipment procurement document prior to equipment
purchase, it is recommended that acceptance of a pre-design report be received from the State
prior to proceeding with equipment purchase. The State should also be notified of any deviations
from the pre-design report.
3.4.3 Drawings and Specifications
The UV installation drawings and specifications should be submitted to the State for
approval. Under the equipment pre-purchase option, the drawings and specifications should
address the installation of the UV reactors and related equipment as well as other necessary
facility modifications. The specific items that would be included in this submittal are discussed
in section 3.3.7. If an alternative approach is used (e.g., design-build or design-build-operate)
the level of detail included in the design documents will differ.
*
3.4.4 Validation Report/Start-up Confirmation
States may request that a validation report or other preliminary testing results be
submitted. As discussed in section 3.1.4.3, validation may occur off-site or on-site. If the UV
reactors are validated at an off-site location, the validation report should be available from the
UV manufacturer and should be a required submittal from the UV manufacturer as part of either
the equipment procurement documents or the UV installation specifications. If on-site validation
is used, a validation protocol should be developed and accepted by the State prior to
implementation. Following completion of the on-site validation, a validation report should be
prepared and submitted to the State. Recommended validation protocols are provided in
Appendix C.
In addition, some States may request that the utility provide as-built documentation (i.e.,
start-up confirmation) certifying construction was completed in accordance with the approved
drawings and specifications. Start-up confirmation may be most important where alternative
project delivery approaches are used and the State does not have the benefit of reviewing 100
percent design drawings and specifications prior to construction.
UV Disinfection Guidance Manual 3-53 June 2003
Proposal Draft
-------
4. Overview of Validation Testing
The Long Term 2 Enhanced Surface Water Treatment Rule (LT2ESWTR) requires the
use of validated UV reactors for receiving Cryptosporidium, Giardia, or virus inactivation credit
(40 CFR i41.729(d)). The purpose of validating a UV reactor is to provide confidence that the
UV reactor can provide the level of inactivation required for a given application. The rule
specifies only basic components of a validation process (presented in section 4.1). Using those
requirements as a framework, this guidance manual describes recommended procedures and data
analysis for one possible approach to validating a UV reactor. Other approaches or
modifications to this approach may be used at the discretion of the State.
The validation protocol provided in this manual has two tiers, specifying two different
methods for addressing uncertainty with a safety factor to determine the log inactivation credit.
These tiers differ in level of complexity. Tier 1 is simplified while Tier 2 is more complex,
potentially allowing for a less conservative safety factor based on detailed knowledge and testing
of equipment performance. Appendix C provides all the necessary procedures and descriptions
to complete a validation for both Tier 1 and Tier 2 methods. This chapter provides a brief
overview of the validation process, describing all the basic steps of the testing procedures and
interpretation of results, with references to Appendix C for more detailed descriptions. For those
conducting a validation test of a given reactor, it is important to understand the background and
detailed procedures described in Appendix C.
4.1 LT2ESWTR UV Disinfection Requirements
This section reviews the LT2ESWTR requirements related to UV reactor validation
specified under 40 CFR 141.729(d) and 40 CFR141, Subpart W, Appendix D.
Validation testing must determine a set of operating conditions that can be monitored by
a utility to ensure that the UV dose required for a given pathogen inactivation credit is delivered;
and the utility must then monitor to demonstrate it is operating within the range of conditions
under which the reactpr was validated.
Validation operating conditions must include, at a minimum, the following: /-
UV intensity (as measured by a UV intensity sensor)
. Flowrate
• Lamp status
».
Many design and equipment factors affect the UV dose delivered by the reactor. The
validated operating conditions must account for the following factors:
• Lamp aging
« Lamp sleeve fouling
UV Disinfection Guidance Manual 4-1 June 2003
Proposal Draft
-------
4. Overview of Validation Testing
. UV transmittance (U VT) of the water
• Inlet and outlet piping or channel configurations of the UV reactor
» Dose distributions arising from the velocity profiles through the reactor
• Failure of UV lamps or other critical system components
• Measurement uncertainty of on-line sensors
Unless the State approves an alternative approach, validation testing must involve the
following:
• Full-scale testing of a UV reactor that conforms uniformly to the reactors used by the
utility.
« Inactivation of a test microorganism whose dose-response characteristics have been
quantified with a low-pressure (LP) mercury vapor lamp.
4.2 Overview of Validation Process
The validation process determines the log inactivation achieved for a specific pathogen
and relates it to the operating conditions at the time of the testing (e.g., UV transmittance at
254 nm, or UVT, flowrate). Figure 4.1 shows the key steps of a validation process, with the
differentiation of Tier 1 and Tier 2 approaches.
The experimental portion of the validation process is referred to as "biodosimetry." It
consists of a UV reactor test that measures log inactivation of a surrogate (challenge)
microorganism under various flowrate, UVT, and lamp power combinations. Log inactivation is
then benchmarked to the corresponding operational conditions and UV intensity sensor values.
Since the true UV doses delivered to the challenge microorganisms cannot be measured directly
by the UV reactor, a separate test must be conducted to relate the inactivation measured in the
field to a UV dose value. Current practice in the UV industry uses a collimated beam to generate
a UV dose-response curve for a given challenge microorganism (log inactivation versus UV
dose). The log inactivation from the biodosimetry test is then related to a UV dose from the UV
dose-response curve. This dose is termed the reduction equivalent dose (RED).
Hydraulic effects, UV reactor equipment, and error in on-line UV intensity sensors all
create uncertainty in translating an RED measured during a validation test to a given level of
pathogen inactivation during routine operation. To account for this uncertainty, a safety factor
should be applied to the required UV dose values for pathogen inactivation credit. The required
UV dose value multiplied by the appropriate safety factor is the RED that should be
demonstrated during a validation test for a given level of pathogen inactivation credit. Tier 1 and
Tier 2 approaches provide methods for incorporating the safety factor to determine the log
inactivation credit.
UV Disinfection Guidance Manual 4-2 June 2003
Proposal Draft
-------
4. Overview of Validation Testing
Figure 4.1 Steps of a Validation Process
Collimated Beam Test
Generate UV-dose Response for
Challenge Microorganism
UV Dose
Challenge
Influent
Sample
Biodosimetry Test
Microorganism
&^
It I
\* J
A
J
Effluent
Sample^
:-:l
'
Log Inactivation and corresponding UV intensity
sensor values
Determine UV Dose for
Log Inactivation Measured
Reduction Equivalent Dose (RED)
!
5
UV Dose (from collimated beam)
Determine Log Inactivation Credit
Tier i (Preset Safety Factors)
- Experimental plan and results should meet
specified criteria
- Uses Tier I RED Target Tables (Table 4.1 and
4.2)
Tier 2 (Derive Safety Factors)
-Calculate uncertainties associated
with lamps, sensors, microbial
measurements, and interpolation of
data.
-Calculate bias associated with RED
measurements of challenge
microorganism vs. pathogen.
-Calculate bias of MP lamp
measurements (if applicable).
Calculate safety factor
from the above three results.
UV Disinfection Guidance Manual
Proposal Draft
4-3
June 2003
-------
4. Overview of Validation Testing
4.2.1 Relating the Experimental RED to Log Inactivation Credit
Chapter 1 presents the UV dose needed to achieve various inactivation credits for
Cryptosporidium, Giardia, and viruses. These dose requirements were derived from batch
(collimated beam) dose-response data and account for the uncertainty and statistical variability in
the dose-response of the pathogen.
There is significant, additional uncertainty associated with applying these batch data to
full-scale, continuous flow testing results. This additional uncertainty associated with UV
reactor validation and on-line dose monitoring should also be considered when determining the
log inactivation credit from UV reactor validation. To account for this uncertainty, the RED
measured during validation should be greater than the dose requirement multiplied by a safety
factor. The safety factor incorporates random uncertainty and corrections for expected variation,
and is defined according to Equation 4.1:
Equation 4.1
where
BRED =
RED bias
Polychromatic bias
Expanded uncertainty expressed as a fraction
The RED bias is a correction that accounts for the difference between the expected dose
delivered to the target pathogen and the actual dose measured using a challenge microorganism
during biodosimetry. That is, the RED measured for two microorganisms is not identical if the
dose-response behavior of the two microorganisms is different. The magnitude of the difference
will depend on the dose distribution of the UV reactor and the unique inactivation kinetics of the
challenge microorganism and target pathogen. If the challenge microorganism is more resistant
to UV light than the target pathogen, the RED measured during validation will be greater than
the expected dose delivered to the pathogen. If the challenge microorganism is as sensitive or
more sensitive to UV light than the target pathogen, the RED bias has a value of one. Appendix
F describes this concept in more detail.
The polychromatic bias is a correction for the spectral differences in the lamp output,
lamp sleeve UV transmittance, water UVT, and action spectrum between validation and
operation of a UV reactor. This bias only applies to polychromatic lamps.
The expanded uncertainty, e, accounts for the uncertainty in the measurements taken
during validation and associated with the equipment (e.g., UV intensity sensors) used to monitor
dose delivery.
Appendix F discusses in greater detail the basis for the uncertainty and bias terms of the
safety factor. Later sections of this chapter and Appendix C describe the application of the
safety factor.
UV Disinfection Guidance Manual
Proposal Draft
4-4
June 2003
-------
4. Overview of Validation Testing
4.2.1.1 Tier 1 and Tier 2 Approaches for Establishing inactivation Credit
As stated previously, the Tier 1 and Tier 2 approaches differ in the complexity of the
method used to determine the log inactivation credit based on the RED measured during
biodosimetry. The Tier 1 approach provides RED target values to be met during validation that
correspond to the log inactivation credit (presented in Tables 4.1 and 4.2). These RED values
incorporate pre-determined safety factors based on characteristics of the UV reactor and
validation testing (section 4.6 provides further details). In the Tier 2 approach, the user
calculates the safety factor using detailed knowledge of the equipment and testing conditions an'd
then applies it to the required dose. This allows the user to optimize their experimental methods
which may reduce the safety factor.
4.2.2 Location and Application of Validation Testing
Validation testing may be conducted either on-site, being the location where the UV
reactor will be installed and operating, or off-site. Off-site validation may be conducted at either
a manufacturer's facility or at a centralized facility dedicated to validating a variety of UV
equipment.
Reactors may be validated for a specific WTP or may validate under a wide array of
conditions for a variety of treatment applications. In addition to a range of operating conditions
(e.g., flowrate, UVT), the reactors may also be validated for a wide range of target doses, thereby
allowing reactor operation to be tailored to achieve different levels of pathogen inactivation
credit at different WTPs. The test conditions and target doses can allow interpolation of the
validation data to conditions of flowrate, UVT, and lamp output specific for application to
various WTP applications. Section C.4.9.3 describes interpolation of validation results as a
function of those variables.
Utilities installing a pre-validated UV reactor should ensure that validation conditions are
appropriate for their plant operations and the quality of testing is acceptable to their State. At a
minimum, the following hydraulic and operating test conditions impact the application of pre-
validated UV reactors:
. UV reactor inlet and outlet configurations
* Flowrate
. UVT
Validating on-site at the WTP is not trivial and should be regarded as a relatively
complex experimental procedure. Utilities conducting on-site validation should consider the
following issues (section C.3.1 provides further details):
• Obtaining water with a sufficiently high UVT to allow validation over the entire UVT
range expected at the WTP
UV Disinfection Guidance Manual 4-5 • June 2003
Proposal Draft
-------
4. Overview of Validation Testing
• Adequate facilities to culture the challenge microorganism to the necessary levels to
demonstrate the desired inactivation
« Adequate facilities and chemicals to adjust UVT to the range expected during ftill-
scale operation
• Providing sufficient mixing of additives prior to entering the UV reactor and mixing
of the challenge microorganisms after the reactor
• Obtaining permits for the disposal of water used for validation
• Verification of the behavior of UV intensity sensors used during validation (sections
C.3.2 and CAT)
• Testing with inlet and outlet conditions representative of those conditions used at the
WTP (issue for off-site validation)
UV reactors previously validated under existing protocols may receive inactivation credit
if the validation used the appropriate challenge microorganism(s) and test conditions met the
needs of the operating conditions at the WTP. Both the Austrian Standard ONORM M 5873-1
and German Guideline DVGW W294 require an RED of 40 ml/cm2, using a microorganism
more representative of Cryptosporidium (B. subtilis) than that used to develop Tier 1 criteria
(MS2 phage). Based on criteria in this document, UV reactors validated with those protocols
should be granted 3 log Cryptosporidium and Giardia inactivation credit. Validation by
NWRI/AwwaRF Guidelines and NSF Standard 55 should be evaluated on a case-by-case basis
as indicated in Appendix C. ,
4.2.3 Third-Party Oversight
Third-party oversight is recommended to ensure that validation testing and data analyses
are conducted in a technically-sound manner and without bias. The validation testing should be
overseen by a registered professional engineer, independent of the UV manufacturer, with
experience in testing and evaluating UV reactors. Furthermore, expert opinion should be sought
from additional parties in areas of UV validation where the engineer has limited experience.
These areas can include, but are not limited to, lamp physics, optics, hydraulics, microbiology,
and electronics.
4.3 Considerations for Validation Testing
This section highlights the key factors that should be considered in the early planning
stages of UV reactor validation.
UV Disinfection Guidance Manual 4-6 June 2003
Proposal Draft
-------
4. Overview of Validation Testing
4.3.1 Inlet and Outlet Hydraulics
The inlet and outlet configurations of the validation location should produce conditions
that result in equal or worse dose delivery than those that will be obtained at the WTP. Sections
3.1.4.3 and C.3.1.5 provide recommended approaches to ensure such hydraulic conditions.
Computation fluid dynamics (CFD)-based dose modeling can also be used in conjunction
with any approach to conservatively address reactor hydraulics during testing. However, due to
uncertainty in the CFD predictions, the predicted dose delivery during validation should be at
least 20 percent greater than the dose delivery predictions at the WTP.
4.3.2 UV Equipment
This section discusses the following UV equipment related issues: documentation,
monitoring control strategies, UV intensity sensors, and lamp aging effects.
4.3.2.1 UV Reactor Documentation
In the weeks prior to testing, the UV manufacturer should provide documentation
identifying and describing the UV reactor to the testing organization (or to third-party oversight
if the manufacturer is conducting the testing with their facilities). This documentation should
include all reactor and component information relating to dose delivery and monitoring, such as
technical descriptions of all internal components, lamp and sleeve specifications, UV intensity
sensor and sensor port information. See section C.2.2 for a complete list and discussion of the
documentation requirements.
4.3.2.2 Control Strategies
The UV reactor's control strategy for monitoring dose delivery affects the selection of
test conditions (i.e., flowrate, UV intensity, and UVT). At present, three strategies are
commonly used for monitoring UV dose delivery. Sections C.4.9.4.1 to C.4.9.4.3 describe these
strategies in detail and recommend validation conditions for each. (Sections 3.1.5 and 5.5 also
describe these strategies with relation to design and operation, respectively).
. UV intensity setooint - relies on UV intensity measurements (i.e., UV intensity
sensors) and flowrate to confirm dose delivery. The system is in compliance
when the measured intensity value is greater than the setpoint at that flowrate.
• UV intensitv/UVT setpoint - relies on the UVT as well as the UV intensity and
flowrate to determine dose delivery. The system is in compliance when both the
UV intensity and UVT are greater than the preset setpoint values.
. Calculated dose - relies on calculated dose delivery from UV intensity, UVT (in
some cases), lamp power and flowrate using an algorithm provided by the UV
reactor manufacturer. Typically, this method is tested over a range of
UV Disinfection Guidance Manual 4-7. . June 2003
Proposal Draft
-------
4. Overview of Validation Testing
combinations of flow, UVT, and lamp power to determine the UV dose and
validate the algorithm.
4.3.2.3 UV Intensity Sensor
Monitoring of the UV dose is achieved through the use of on-line UV intensity sensors.
The properties of both on-line and reference sensors should be measured by an independent
laboratory that is equipped to confirm sensor calibration and measure the sensor's angular and
spectral response, linearity over the working range, and temperature response. The Tier 1
approach specifies criteria for sensor placement in the UV reactor, sensor spectral response, and
measurement uncertainty.
4.3.2.4 Lamp Aging
Prior to the initiation of validation testing, all lamps should undergo 100 hours of burn-in.
This practice improves the stability of lamp output. Additional testing may also be performed, if
requested, in order to assess the effects of lamp age on dose delivery. With time, medium-
pressure (MP) UV lamps can undergo non-uniform aging that causes spectral shifts in output.
These changes can have an impact on the dose delivery registered by the monitoring systems.
Manufacturers should test dose delivery of new and aged lamps to determine if the aged lamps
reduce disinfection performance. If so, validation should be conducted using both new and aged
lamps. (Section C.4.8 describes a procedure for testing new versus aged lamps.)
4.3.3 Additives Used in Validation Testing
4.3.3.1 Challenge Microorganism
UV reactor validations should be performed with a microorganism with the following
characteristics: inactivation kinetics closely resembling those of the target pathogen and the
ability to be cultured in a reproducible manner to high concentrations. Currently, research has
not identified such a microorganism that is ideal for Cryptosporidium. Challenge
microorganisms typically used include MS2 phage and'Bacillus subtilis, both of which are
significantly more resistant to UV than Cryptosporidium.
The RED bias, an important component of the safety factor, is due to the differences in
inactivation kinetics between the challenge microorganism and the target pathogen. Under the
Tier 1, the RED bias is based on MS2 phage as the challenge organism. If a challenge
microorganism is identified in the future that exhibits a dose-response similar to the target
pathogen (e.g., Cryptosporidium), the RED bias could be decreased.
UV Disinfection Guidance Manual 4-8 . June 2003
Proposal Draft
-------
4. Overview of Validation Testing
4.3.3.2 UV-Absorbing Compound
During validation, the UVT can be lowered through the addition of a UV-absorbing
compound to simulate the range of UVT that may be encountered for a given UV application.
For the validation of MP UV systems, the absorbing compound should have a UV absorbance
spectrum similar to the water being treated in the full-scale application. However, obtaining an
exact replica is usually not possible. Coffee and lignin sulfonate are commonly used UV
absorbing compounds; however, sodium thiosulfate and fluorescein have also been used with
some success.
The polychromatic bias, a component of the safety factor for only MP reactors, is
determined as a function of the UV-absorbing compound. The Tier 1 approach specifies criteria
for minimum UVT for MP reactors using UV-absorbing compounds and applies a correction
factor based on validation testing performed to-date with various UV absorbing compounds.
4.4 Validation Testing
Validation provides an assessment of UV reactor dose delivery and monitoring under '
specific conditions of flowrate, UVT, and lamp output. This section briefly discusses the steps
involved in conducting a validation test and provides references to more detailed procedures in
Appendices C and D.
4.4.1 Microorganism Preparation
Challenge microorganisms should be prepared in accordance with peer-reviewed
methods. All information regarding the source of the host, media descriptions, and preparation
steps should be documented. It is expected that the microorganism stock will be prepared by
laboratory personnel familiar with methodologies designed to prevent microbial stock
contamination. The use of these same techniques in the field during validation is critical and any
personnel participating in the validation should be familiar with them to avoid sample
contamination.
Preparation methods for the two most common challenge microorganisms, MS2 phage
and B. subtilis spores, are provided in Appendix D. Note, the same batch of challenge
organisms should be used for .both collimated beam and biodosimetry testing, as described
below.
4,4,2 Collimated Beam Testing
The collimated beam data are used to develop the dose-response curve for the challenge
microorganism. A collimated beam apparatus typically consists of an enclosed low-pressure UV
lamp and a tube with a non-reflective inner surface (see Figure 4.2). A sample of the challenge
microorganism (preferably taken from the influent to the biodosimetry test stand) is placed in a
petri dish and exposed to the UV light for a predetermined amount of time. The UV dose is
UV Disinfection Guidance Manual 4-9 June 2003
Proposal Draft
-------
4. Overview of Validation Testing
calculated using the intensity of the incident UV light, UV absorbance of the water, and exposure
time. Appendix E provides a complete description of collimated beam testing.
At least two water quality conditions should be tested-one with the highest UVT (no
absorbing chemical added) and a second with the lowest UVT used in the biodosimetry test. UV
doses should be selected to target microorganism inactivations of approximately 0.5, 1.0, 2.0,
3.0,4.0, and 5.0 log.
Figure 4.2 Collimated Beam Test Apparatus
Low-Pressure
.Mercury Arc Lamp
Petri Dish Containing
Microbial Suspension
-I 1
Lamp Enclosure
— Collimating Tube
UV Light® 254 nm
Magnetic Stirrer
4.4.3 Biodosimetry of Full-Scale Reactors
The biodosimetry test is used to determine the inactivation of the challenge
microorganism by the UV reactor under continuous-flow test conditions. Figure 4.3 provides a
schematic of the components used in a typical biodosimetry test. Section C.3.1 describes the key
features.
UV Disinfection Guidance Manual
Proposal Draft
4-10
June 2003
-------
4. Overview of Validation Testing
Figure 4.3 Biodosimetry Test Components
UV Challenge
Absorber Microbe
Water S\ T T
Supp,y —XJ"jj
Backflow
" Prevention
Static
Mixer
Pressure
Pressure
Influent Influent
Quenching Sample
Agent Port
Static
Mixer
Effluent
Sample
Port
The following facilities are typically required in biodosimetry testing:
• Injection of the challenge microorganism and UV absorbing compound
» Mixing of the added compounds upstream and downstream of the reactor before
sampling
. Flow measurement
. Pressure measurement upstream and downstream of the reactor
« Sample collection before and after the UV reactor
Proper facilities should be provided, along with appropriate permits, to discharge the
treated water. The testing should be conducted after steady-state conditions are achieved for the
desired matrix of experimental conditions evaluating variations in challenge organism
concentration, flowrate UVT, and lamp power/output. Samples collected from the influent and
effluent sample ports are used to determine the inactivation achieved for the specific reactor
condition being tested. Operational parameters, such as UV intensity, flowrate, UVT, and lamp
power, are measured during the test.
A detailed description of sampling requirements is provided in Appendix C.
4.5 , Data Analysis
Results from the collimated beam testing, biodosimetry testing, and uncertainties
associated with equipment and data are used to determine the log inactivation credit achieved by
the UV reactor. Data analysis consists of four steps:
1. Developing a UV dose-response curve for the challenge microorganism from the
collimated beam test
UV Disinfection Guidance Manual
Proposal Draft
4-11
June 2003
-------
4. Overview of Validation Testing
2. Calculating log inactivation from the biodosimetry test
3. Determining the RED(s) from the results of steps 1 and 2
4. Applying safety factors to determine log inactivation credit (Tier 1 or Tier 2
approach)-
The following sections describe these steps. References to the appropriate sections in
Appendix C are provided for further details and examples.
4.5.1 Developing Challenge Microorganism Dose-Response Curve
Dose-response curves should initially be generated separately for each collimated beam
test condition (a minimum of two conditions—lowest and highest UVT—is recommended). The
curves should predict similar dose-response relationships, as indicated by statistical analyses. If
statistically similar, the data can be combined and one curve generated for the entire dataset. If
the curves are statistically different, the cause of the difference should be determined, and the
test should either be redone or the different dose-response curves should be used for the different
test conditions. Differences in UV dose-response could occur if the dose-response were
determined with different batches of the challenge microorganism or if coagulation or other
water quality interferences impacted the dose-response.
The following sub-sections describe how to calculate the log inactivation from collimated
beam test data and generate a dose-response curve. Section C.4.9.7.2 discusses the statistical
analysis for comparing curves and combining data.
4.5.1.1 Calculate Dose-Response Data From Collimated Beam Testing
The log inactivation for each dose delivered by the collimated beam should be calculated
using Equation 4.2:
-------
4. Overview of Validation Testing
4.5.1.2 Fitting Dose-Response Data to a Curve
The following steps describe how to develop a dose-response curve:
1) Plot log inactivation achieved as a function of UV dose in the collimated
beam test
->
2) Use regression analysis to derive an equation that best fits the data
• For first-order kinetics, a linear equation should fit best:
Dose = A x Log Inactivation + B
• For dose-responses showing tailing effects, a quadratic equation should fit
best:
Dose = Cx Log Inactivation + D x (log Inactivationf
• For dose-response showing shoulder effects, other polynomial equations
should be used.
3) Evaluate fit of equation
\ t
. Equation coefficients should be significant at a 95 percent confidence
level (section C.4.9.7.1 provides an example that uses p-statistics to
evaluate the coefficients).
• Confidence intervals for the fit should be determined at an 80 percent
confidence level. (The Tier 1 approach specifies criteria the confidence
intervals must meet and the Tier 2 approach includes an uncertainty term
for the confidence intervals in the safety factor calculation.)
• The differences between the predicted dose and measured dose at a given
- log inactivation should be randomly distributed around zero and not
dependent on dose. In other words, the data points should be randomly
distributed above and below the curve (section C.4.9.7.1 provides an
example of this evaluation).
4.5.2 Determining Log Inactivation from Biodosimetry Testing
At each test condition—flowrate, UVT, and lamp output—the arithmetic mean and
standard deviation of the log influent and'effluent challenge microorganism concentrations
should be calculated. From the mean concentrations, log inactivation should calculated using the
following equation:
UV Disinfection Guidance Manual 4-13 June 2003
Proposal Draft
-------
4. Overview of Validation Testing
Log Inactivation = \og(Nt ) - log(jV£ ) Equation 4.3
where
log(Nj) = Mean challenge microorganism log concentration of the reactor influent samples
log(NE) = Mean challenge microorganism log concentration of the reactor effluent samples
The standard deviation is used in the safety factor calculation for Tier 2, while Tier 1 specifies a
limit for the standard deviation.
4.5.3 Determining the RED
This section describes how to calculate RED values for all test conditions and select the
appropriate RED for subsequent log inactivation credit determination.
4.5.3.1 Calculating the RED Values
The RED is calculated by inputting the biodosimetry log inactivation values for each test
condition into the equation describing the UV dose-response curve of the challenge
microorganism.
Example. For 0.5 MOD flow, 80 percent UVT, and lamp output of 70 percent, the
inactivation calculated from Equation 4.3 was 4.0 log. The UV dose-response equation
was best fit with the equation:
Dose = 15.5 x Log Inactivation - 6.0
Inputting 4.0 log into the above equation results in an RED of 56 mJ/cm2. This
calculation should be repeated for each test condition (i.e., flowrate, UVT, and lamp
output combination).
4.5.3.2 Selecting the Appropriate RED for Log Inactivation Credit
Determination
Since the biodosimetry test is conducted at various flowrates, UVT, and lamp output
combinations, the validation results will have more than one RED value for each setpoint.
Choosing the appropriate RED to determine log inactivation credit depends first on the
monitoring approach used to indicate dose delivery. The following three approaches are
considered in this text:
• UV intensity setpoint approach - the UV reactor should be rated at the lowest
inactivation observed.for each set point condition tested.
* UV intensity and UVT setpoint approach - the UV reactor should be rated at the
inactivation observed with UV reactor operation under setpoint conditions.
UV Disinfection Guidance Manual 4-14 June 2003
Proposal Draft
-------
4. Overview of Validation Testing
• Calculated dose approach - the UV reactor should be rated at the lowest inactivation
observed for each calculated dose setpoint evaluated.
Section C.4.9.4 recommends validation conditions for each of the above approaches.
Section C.5 provides examples of interpreting validation results for the different approaches.
4.5.3.3 Interpolating RED as a Function of Test Conditions
The RED measured during validation testing can be interpolated as a function of inverse
flowrate, UVT, or UV intensity by fitting an equation to the data being interpolated (e.g., RED as
a function of inverse flowrate). The equation should not be used for extrapolation (i.e.,
projecting RED outside the range of tested conditions). The following provides guidelines for
interpolation:
, . The equation should pass through the origin (0,0) if the RED is interpolated as a
function of measured intensity or inverse flowrate
. The equation coefficients should be significant at a 95 percent confidence level
. The differences between the values measured and predicted by the equation should be
randomly distributed around zero
. An 80 percent confidence interval should be used to determine the uncertainty of the
equation used to interpolate the RED values. For Tier 1, the uncertainty of the
interpolation should be 10 percent or less at an 80 percent confidence level. For Tier
2 it'should be included as an uncertainty term in the safety factor calculation as
described in section C.4.10.2.3.
4.5.4 Determining Inactivation Credit
As discussed in the introduction to this chapter, there are two approaches described for
determining log inactivation.
. Tier 1 - pre-determined safety factor.
•t
• Tier 2 - calculated safety factor from the following dose delivery monitoring and
validation bias and uncertainties:
- RED bias
- Polychromatic bias (for MP reactors)
- Measured RED
- Interpolation of RED as a function of flowrate, UVT, or UV intensity
- Sensors used during validation (UV intensity, UVT)
- On-line and reference sensors used at WTP (UV intensity, UVT)
- Lamp output quantification
UV Disinfection Guidance Manual 4-15 June 2003
Proposal Draft
-------
4. Overview of Validation Testing
The remainder of this chapter describes how to determine the log inactivation credit
achieved using the Tier 1 approach and the criteria that should be met in order to use this
approach. Appendix C contains a detailed description of the basis the Tier 2 approach.
Tier 1 Log Inacfivation Credit
Tables 4.1 and 4.2 present the RED target values for UV reactors using LP/LPHO and
MP lamps, respectively. The values in these tables are derived by multiplying the required dose
values by the Tier 1 safety factors (see Appendix C for details). The values in Table 4.2 (MP)
are higher than in Table 4.1 (LP/LPHO) because they include the polychromatic bias, which is
not a factor in monochromatic (LP/LPHO) reactors.
For a given pathogen and level of log inactivation credit, the RED measured during
validation should be greater than or equal to the corresponding RED target listed in the table.
Note, validation testing with multiple setpoints may result in different log inactivation credits for
the different setpoints.
Example. Using an LP reactor and meeting the Tier 1 validation criteria (see section
4.6), the lowest RED measured for the challenge microorganism during validation was
29 mJ/cm2. Consequently, the log inactivation credits achieved are 2.5 for
Cryptosporidium and 2.5 for Giarida. No inactivation credit is achieved for viruses.
Table 4.1 Tier 1 RED Targets for UV Reactors with LP or LPHO Lamps
Log
Inactivation
Credit
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
RED Target (mJ/cm2)
Cryptosporidium
6.8
11
15
21
28
36
-
-
Giardia
6.6
9.7
13
20
26
34
-
-
Virus
55
81
110
139
169
199
227
259
Table 4.2 Tier 1 RED Targets for UV Reactors with MP Lamps
Log
Inactivation
Credit
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
RED Target (mJ/cm2)
Cryptosporidium
7.7
12
17
24
32
. 42
-
-
Giardia
7.5
11
15
23
30
40
-
Virus
63
94
128
161
195
231
263
300
UV Disinfection Guidance Manual
Proposal Draft
4-16
June 2003
-------
4. Overview of Validation Testing
4.6 Tier 1 Criteria
The safety factors derived for the Tier 1 approach are based on assumed uncertainties and
corrections for given experimental methods. For these assumptions to be practical, and thus the
use of Tier 1 numbers appropriate, the validation conditions should meet the criteria specified in
this section. Note, the equipment criteria should be provided by the UV manufacturer and
reviewed by a third-party for verification.
4.6.1 UV Intensity Sensors
• UV reactors with MP lamps should be equipped with one sensor per lamp. UV
reactors with LP or LPHO lamps should be equipped with at least one sensor per
bank of lamps.
. UV intensity sensors should view a point along the length of the lamp that is between
the electrode (lamp end) and within 25 percent of the arc length away from the
electrode.
• UV intensity sensors should have a spectral response that peaks between 250 and 280
nm. When mounted on the UV reactor and viewing the lamps through water, the
measurement of UV light greater than 300 nm made by the sensor should be less than
10 percent of the total measurement made by the sensor. Conformance to these
criteria can be demonstrated using UV intensity field modeling. Figure 4.4 presents
examples of two sensors where both have the appropriate peaks, but one has too
much UV light in the >300 nm range.
. The UV intensity sensors used during validation and the duty and reference sensors
used during operation of the UV reactor at the WTP should provide National Institute
of Standards and Technology (NIST)-traceable measurements with an uncertainty of
± 15 percent or less at an 80 percent confidence level.
. During operation of the UV reactor at the WTP, measurements made by the duty UV
intensity sensor should be checked using a reference UV intensity sensor. If the duty
sensor reads higher than the reference sensor (i.e., overestimating dose delivery), or
substantially lower, it should be recalibrated. For a recommended control standard,
the duty sensor should not read less than the reference by the following amount:
^-1 xlOO < ((45f + a* P Equation 4.4
V^rf J .
where
IRcf = Intensity measured by the reference sensor
buty = . Intensity measured by the duty sensor
ORef = Measurement uncertainty of the reference sensor (%)
ODuty = Measurement uncertainty of the duty sensor (%)
UV Disinfection Guidance Manual 4-17 June 2003
Proposal Draft
-------
4. Overview of Validation Testing
If the dose-monitoring strategy uses an on-line UVT monitor, the UV absorbance at
254 nm (A254) calculated from the measured UVT should have an uncertainty of ±10
percent or less at an 80 percent confidence level.
Figure 4.4 Examples of UV Intensity Sensor Spectral Response Ranges
250 280
Filtered Sensor
Unfiltered
Sensor
250 300 350
Wavelength (nm)
400
41
g
.2
•o
fc
•5
&
o
Q
25 -,
20 -
15 -
10 -
5
i
I i
1 i
1 ;
lUlvlAl '
_j^fll WUL . -
200 250 300 350 400
Wavelength (nm)
Filtered Sensor. Detected UV light with a 0 cm
sensor-to-lamp water layer. Detected UV > 300 nm
is 0.7% of total UV light detected.
Jetected Irradiance
-» K> W *. Ol
3 O O O O O
•
--KrfUil
. i
200 250 300 350 400
Wavelength (nm)
Unfiltered Sensor. Detected UV light with a 0 cm
sensor-to-lamp water layer. Detected UV > 300 nm
is 41% of total UV light detected.
»
.5
•o
S
•o
S
ft
a
0.0025 -,
0.0020 -
0.0015 -
0.0010 -I
0.0005 -
0.0000 i
- 2(
1 i
1 I
' ili\IL '
Jfr WA. ! 1
)0 250 300 350 400
Wavelength (nm)
Filtered Sensor. Detected UV light with a 20 cm
sensor-to-lamp water layer. Detected UV > 300 nm
is 5% of total UV light detected.
« 0.0400 n
| 0.0300 -
S
± 0.0200 -
•o
| 0.0100 -
a n nnnn
. ,J
»_
200 250 300 350 400
Wavelength (nm)
Unfiltered Sensor. Detected UV light with a 20 cm
sensor-to-lamp water layer. Detected UV > 300 nm is
85% of total UV light detected.
UV Disinfection Guidance Manual
Proposal Draft
4-18
June 2003
-------
4. Overview of Validation Testing
4.6.2 UV Lamp Output
The standard deviation of the UV output of LP or LPHO lamps should be 15 percent
or less of the mean output. The standard deviation should be determined using either
life test or field test data on aged lamps.
4.6.3 Flow Measurements
The flow measurements made during validation and during operation of the UV
reactor at the WTP should have an uncertainty of ± 5 percent or less at an 80 percent
confidence level.
4.6.4 Collimated Beam Apparatus
. The calculated dose delivered by the collimated beam apparatus should have a
measurement uncertainty of ± 15 percent or less at an 80 percent confidence level.
4.6.5 Challenge Microorganism Dose-Response
• Over the range of doses within one log unit of the log inactivation demonstrated
during validation, the UV sensitivity of the challenge microorganism should be less
than or equal to 25 ml/cm2 per log inactivation (the dose-response of a resistant strain
of MS2). For example, if the challenge microorganism log inactivation measured by
the UV reactor ranges between 1.5 and 3.5 log, the dose-response of the challenge
microorganism should be less than or equal to 25 mJ/cm2 per log inactivation
between 0.5 and 4.5 log inactivation.
• If the dose-response of the challenge microorganism has a shoulder, that shoulder
should not occur over a dose range greater than 50 percent of the RED demonstrated
during validation. The shoulder is defined by extrapolating the exponential reduction
region of the dose-response curve to the dose-axis (see Figure 4.5).
. If the dose-response demonstrates tailing, the tailing should not occur until one log .
reduction greater than the log reduction demonstrated during validation.
. A plot of dose versus log inactivation for the collimated beam test should have an 80
percent confidence interval of 10 percent or less at the log inactivation demonstrated
by the UV reactor.
UV Disinfection Guidance Manual 4-19 June 2003
Proposal Draft
-------
4. Overview of Validation Testing
Figure 4.5 Dose-Response With a Shoulder
5 -1
4.5 -
4 -
3.5 -
3 -
This range should be less than 1/2
the RED demonstrated during
validation. In this case, the RED
should be >20 mJ/cm2.
25
50
UV Dose (mJ/cm2)
75
4.6.6 Medium Pressure Lamps
. During validation, the UVT of the water at 254 nm should be greater than the values
specified in Figure 4.6 for a given sensor-to-lamp water layer and UV-absorbing
chemical. The sensor-to-lamp water layer is defined as the distance traveled through
water by UV light passing from the lamp to the sensor. The values in Figure 4.6 were
taken from Figure C.7 of Appendix C for a polychromatic bias of 1.2.
UV Disinfection Guidance Manual
Proposal Draft
4-20
June 2003
-------
4. Overview of Validation Testing
Figure 4.6 Criteria for the Minimum UVT of MP Reactors under Tier 1
1fifi
s?
c yo ~
c
s
OJ on .
*J WJ
§ 85
E
3
E 80-
c
jg
75 -
•
Coffee
— — Lignin Sulphonate
- - -NOM
0
j
/
s
A
>
/
s
r
A
/
f
5 10
J
r
r
f
f
^^^
^>^
X^
^^ +
-------
5. Start-Up and Operation of UV Installations
This chapter describes the start-up activities and routine operational issues associated
with a UV disinfection facility. The start-up discussion focuses on the functional and
performance testing that should be conducted during the start-up process. The remainder of the
chapter describes the requirements and recommendations for operation, maintenance,
monitoring, and reporting for UV installations. The organization of this chapter is presented
below by the key question each section addresses.
• What is included in final UV installation inspection?; Section 5.1.1
• What testing should be completed during start-up? Sections 5.1.2 and 5.1.3
. What items should be included in the operations and
maintenance manual? Section 5.1.4
. What are the operational requirements and recommended
tasks? Sections 5.2.1 and 5.2.2
» What are the routine start-up and shutdown procedures? Section 5.2.3
• What maintenance tasks are recommended? Section 5.3.1
• What spare parts are recommended to be kept on hand? Section 5.3.3
. What monitoring is required for regulatory compliance? Section 5.4.1
• What additional monitoring is recommended? .....Section 5.4.2
. What should be reported to the State? Section 5.4.3
. How do you determine the operational requirements from
validation testing? Section 5.5
. What should be done if there is: Section 5.6
- Low UV intensity?
- High UV absorbance?
- Rapid flow increase/high flow?
Unreliable UV intensity sensor readings?
Power loss?
. What staffing issues are associated with operation,
maintenance, and monitoring of UV installations? Section 5.7
Given the wide range of UV installations and UV reactors available, this document
cannot address or anticipate all scenarios. The guidelines provided in this manual are a
compilation of industry experience and manufacturers' recommendations. Therefore, they may
UV Disinfection Guidance Manual
Proposal Draft
5-1
June 2003
-------
5. Start-up and Operation of UV Installations^
differ from those provided by specific manufacturers for their equipment. In these situations, the
manufacturer's standards should be followed.
The general process to be followed for the start-up and routine operation of a UV
installation is shown in Figure 5.1. A detailed description of each activity is given in the
remainder of this chapter.
Figure 5.1 Start-up and Operation Flowchart
Completion of
Construction and
Inspection
Start-Up Activities
Final Inspection
Section 5.1.1
Functional Testing
Section 5.1.2
Performance Testing
Section 5.1.3
Operations and Maintenance
Development
Section 5.1.4
Routine Operation
Routine Operations
Sections 5.2.1 to 5.2.4
Operational
Challenges
Section 5.6
Report to State
Section 5.4.3
Staffing Issues
Section 5.7
UV Disinfection Guidance Manual
Proposal Draft
5-2
June 2003
-------
5. Start-up and Operation of UV Installations
5.1 Start-up of UV Installation
For the purposes of this manual, the start-up of the UV installation is considered as the
transition from the construction phase to the operation phase. Start-up activities include final
inspection of the UV reactors and ancillary equipment, functional testing, performance testing,
operations and maintenance (O&M) manual development. Functional testing confirms the
mechanical, instrumentation and controls, and hydraulic conditions of the UV installation to
ensure they meet the requirements of the contract documents. It also verifies that the operational
conditions are consistent with the validated conditions. Performance testing verifies that the UV
reactors are operating in accordance with the contract documents. In addition, an O&M manual
should be developed during UV installation start-up.
A start-up plan should be developed in collaboration with the UV installation designer,
plant operations staff, and the UV manufacturer. The designer will be most familiar with the
layout of the reactors, piping, and how to integrate the UV installation with the other treatment
processes. The operations staff will be able to identify potential impacts on routine plant
operations. The manufacturer will be most familiar with operation of the UV reactors. The start-
up plan should include a pre-start checklist, a procedure for checking equipment installation and
calibration (functional testing), a procedure for verifying system operation, and a procedure for
checking alarm settings and system controls (performance testing).
5.1.1 Final Inspection
As the first step in the start-up process, a detailed inspection of the UV installation should
be completed. The inspection should include a visual assessment to ensure that all components
meet the technical specifications and that the UV installation was completed in accordance with
the construction documents. The configuration of the piping and UV reactors should meet the
constraints established during validation testing (see section 4.3.1). If on-site validation will be
performed, the availability of the necessary features (e.g., feed and sample ports, mixing
systems, drains) should be confirmed. In addition, leak testing should be performed, and then all
UV installation components and associated valves and piping should be thoroughly cleaned and
disinfected (State requirements may apply).
5.1.2 Functional Testing
Functional testing consists of a series of short duration tests that assess the ability of each
component of the system to function in accordance with the specifications detailed in the
contract documents. Some of the evaluations are conducted by monitoring performance during
normal operations. However, the majority of functional testing is completed through simulations
of specific operating conditions and monitoring the UV reactor operation and response.
Functional testing entails flooding and energizing the UV reactors to confirm the operation of the
following items:
« UV lamps and UV intensity sensors
• Operating sequence and control logic for the reactor
UV Disinfection Guidance Manual 5-3 June 2003
Proposal Draft
-------
5. Start-up and Operation of UV Installations
. Ancillary equipment, including UV transmittance (UVT) monitors, flowmeters, and
control valves
*
• Electrical system components, including ballasts, uninterruptible or standby power
supplies, and the ballast cooling system
ji
It is strongly recommended that the UV manufacturer inspect the UV installation prior to
energizing the UV reactors and be present when the UV reactors are first energized.
Manufacturers may require the presence of one of their representatives during these activities as
a condition of their equipment warranty.
5.1:2.1 Verification of Mechanical Operation
UV reactors may incorporate mechanical elements such as valves and on-line mechanical
cleaning (OMC). During functional testing, the satisfactory operation of these mechanical
components should be confirmed. The procedures used to confirm valve operation for a UV
installation are not different from those for other applications that use valves for isolation or flow
control and, therefore, are not described here. The OMC system, if provided, should be checked
for proper operation. Specifically, the following items should be verified:
• Smooth movement of the wiper with no jamming or binding of the wiper on the
sleeve
. Extension of wiper stroke to the full length of the sleeve with no impact at the end of
travel that could damage or break the sleeve
» Proper operation of the wiper drive mechanism and motor with no slipping or binding
5.1.2.2 Verification of Monitoring Equipment
The monitoring equipment is important for UV reactor operation, and its proper operation
should be verified during functional testing.
Flowmeter
Accurate'measurement of the flow is essential to ensure that the UV reactors are
operating within the validated conditions. Not all utilities will install dedicated flowmeters. For
those facilities that rely on flow measurement using an existing, common flowmeter (e.g., raw
water flowmeter), the functionality of the flowmeter should be verified in conjunction with its
intended use with the UV installation. Specifically, the accuracy and operating range of the
flowmeter should be verified and the availability of the necessary output signals from the meter
should be confirmed. If pressure gauges are used to monitor the flow split between UV reactors,
the calibration and installation of the pressure gauges should be verified as well.
The uncertainty associated with the existing flowmeter should be determined to ensure
that the appropriate validation constraints were used. It is recommended that the original
UV Disinfection Guidance Manual 5-4 June 2003
Proposal Draft
-------
5. Start-up and Operation of UV Installations
certification of calibration be reviewed in conjunction with the equipment specifications to
establish the measurement uncertainty for the existing flowmeter. There are three methods to
verify the flowmeter operation: flow verificators, a time-discharge test, and a clamp-on
flowmeter. The flow verificators assess the physical condition of the installed equipment relative
to its condition at the time of factory calibration to confirm that the original uncertainty can be
maintained. For example, verification of a magnetic flowmeter would consist of an insulation
test of the entire flowmeter system and cable; testing of the sensor magnetic properties; testing of
signal converter gain, linearity and zero point; testing of digital output; and testing of analog
output. A time-discharge test compares the flowrate measured by the flowmeter against the
value calculated by measuring the volume of water discharged over a predetermined amount of
time (using a bucket, clearwell, or tank of known volume). A temporary or clamp-on flowmeter
can be used to assess the accuracy of the existing flowmeter. It is important to consider the
uncertainty of the reference flowmeter when using this approach.
If a new flowmeter is used to measure the flow through the reactor, the flowmeter
manufacturer should provide a certification of calibration at the time of equipment delivery. It is
also recommended that the manufacturer inspect the UV installation and confirm that it was
completed in accordance with their recommendations to ensure the certified accuracy of the
flowmeter is achieved. The flowmeter measurement uncertainty should be equal to or better than
that used during validation.
On-line UVT Monitor
An on-line UVT monitor may be included as part of the UV reactor, especially if a UV
intensity and UVT setpoint or calculated dose control strategy (section 3.1.4.2) is used. The on-
line UVT monitor should be calibrated and its operation verified. Calibration can be completed
using a buffer solution of known UVT and may be operation may be verified by collecting and
analyzing grab samples, using a bench top spectrophotometer.
5.1.2.3 Verification of Instrumentation and Control Systems
The amount of testing needed for the instrumentation and control systems is proportional
to the complexity of the control strategy that is used. Testing should include verification of
monitoring equipment (including calibration of all instruments), tuning of control loops,
checking operation functions, and verifying all final control actions. As described below, the
UV reactors should be run through a series of simulations that represent the possible operating
scenarios in order to confirm that the appropriate UV reactor response occurs. Typically, the
packaged UV reactor control panel contains all of the components to control and operate the UV
reactor. The panel should provide the operating status, diagnostic information, and operator
interface capability. It should also include (amp status indicators and programmable logic
controllers (PLC) and may include ballasts, and lamp starters. The PLCs are typically used to
control the operation of a UV reactor based on certain input signals. A manufacturer
representative should be present during the simulations to assist in troubleshooting and
addressing any issues that may result from the packaged UV reactor controls.
Simulations should be used to confirm the operation of the UV reactors and the operation
of all ancillary equipment and instrumentation, including valves, flowmeters, and UVT monitors.
UV Disinfection Guidance Manual 5-5 • , June 2003
Proposal Draft
-------
5. Start-up and Operation of UV Installations
As applicable, specific operating conditions that should be simulated include the following
conditions:
* Cold start of the UV reactors
. Cool down and restart of the UV reactors
. Sequencing of the UV reactors in multiple reactor installations
. Adjustment of lamp intensity in response to varying water quality or flowrate
. Shutdown of the UV reactors
. Operation of the UV reactors during line power failure (when backup or
uninterruptible power supplies (UPS) are available)
. Manual override, safety interlocks, and report generation
During these simulations, the utility should record the amount of off-specification time
and discharge volume (i.e., operation outside of validated conditions) associated with each
action. This is necessary to assess the potential effect of the conditions associated with these
actions on the utility's ability to meet its disinfection goals and comply with the State-established
limitations for off-specification operation. In addition to simulating possible operating
conditions, each of the alarm conditions and monitoring functions incorporated in the design
should be verified. Possible monitoring functions and alarm conditions are discussed in section
3.3.3.8 and may include the following conditions:
. Low UV dose and UV intensity
. Low UVT
. . Low and high flowrate
. Lamp age
. Lamp or ballast failure .
. Low liquid level in the UV reactor
. High temperature
. OMC system failure
5.1.2.4 Verification of Flow Distribution and Headloss
If each reactor is not equipped with a dedicated flowmeter,' then it will be necessary to
verify the flow split between reactors over the entire operating flow range. This flow split and
the total plant flow should be used to estimate the flow through each UV reactor and confirm
UV Disinfection Guidance Manual 5-6 June 2003
Proposal Draft
-------
5. Start-up and Operation of UV Installations
operation is within the validated conditions. Clamp-on type flowmeters or differential pressure
readings across each parallel reactor are alternatives for field verification of the flow split.
The allowable difference in flow among reactors (flow split differential) is established
during validation and should be accounted for in the validation protocol safety factor (section
4.2.1 and section F.5). If the actual flow split differential is greater than assumed in validation,
then steps should be taken to improve the flow split. The Tier 1 recommendations for validation
of UV reactors (section 4.6) necessitates a flow split differential of 10 percent or less. If this is
not observed during functional testing, then a Tier 2 analysis for validation safety factor needs to
be completed for the UV reactor. Appendix C provides details-about the Tier 1 and 2 analysis
and Appendix F provides details about the development of the safety factor.
The headloss should be measured for each reactor and compared to the headloss specified
in the contract documents (if applicable). Pressure transducers or pressure gauges can be used to
measure the headloss.
5.1.3 Performance Testing
Performance testing is intended to assess the operating performance of the UV reactor as
a whole, as well as the individual performance of its components. While functional testing is
primarily completed through simulations of specific operating conditions, performance testing is
generally accomplished through extensive monitoring of reactor performance during the early
stages of continuous operation. It is important to note that performance testing is not intended to
validate disinfection performance, which is completed during validation testing (as described in
Chapter 4). However, performance testing can be used to confirm that the actual operating
conditions are within the constraints established during validation testing. Performance testing
focuses on the accuracy, reliability, and repeatability of UV reactor operation, whereas validation
is used to measure the effectiveness of the UV reactor at delivering the UV doses required for
target pathogen inactivation credit.
When UV lamps are first energized, they go through a stabilizing period called "burn-in."
For some UV lamp designs, the initial lamp output may significantly exceed the design value.
During burn-in, the lamp output may rapidly decrease to a value more consistent with the design.
Following burn-in, lamp output becomes relatively stable until the end of lamp life is
approached. Typically, new UV lamps will not have undergone burn-in prior to installation.
Because performance testing should compare actual operating conditions to validated conditions,
it is important that the lamps be in the same condition as they were during validation testing.
Therefore, UV lamps should be burned-in prior to performance testing, which typically takes
100 hours of continuous operation. The actual required burn-in time should be discussed with
the manufacturer and confirmed through documented operating experience at other UV
installations.
The duration of performance testing and the extent of monitoring will be project-specific
and should be established by the utility and designer based on the objectives of the performance
testing. Performance testing may range in duration from as little as 48 hours of uninterrupted
operation to greater than four months of demonstrative operation. Similarly, the scope of the
testing maytange from an increased monitoring frequency to confirm performance to an
UV Disinfection Guidance Manual 5-7 June 2003
Proposal Draft
-------
5. Start-up and Operation of UV Installations
extensive testing protocol to ftilly optimize reactor performance and establish long-term
operating procedures. During performance testing, treated water may be sent to the distribution
system if upstream treatment has not changed and meets existing regulations. However this
should be confirmed with the State.
Performance testing may include the following items:
• Operation of each UV reactor in automatic mode and demonstration that actual
operating conditions are within the constraints established during validation testing
• Demonstration of UV reactor start-up and switchover sequences that result from
water quality and/or flowrate changes
. Observation of operation, including periods of off-specification operation, due to
power quality problems, and other alarm conditions
• Measurement of electrical service voltage, current, and power consumption with
different flow and water quality combinations to optimize energy use within the
constraints established during validation
. Assessment of the effectiveness of the cleaning system by inspecting sleeve clarity
and condition at regular intervals throughout the test period
. Confirmation that the programmed cleaning frequency correlates with the actual
frequency of cleaning
« Verification of UV intensity sensor operation
• Confirmation of duty sensor accuracy using reference sensors (see section 5.3.2.2)
. Observation of ballast temperature and cooling system performance
« Verification of the accuracy and repeatability of the on-line UVT monitor through the
collection of grab samples and analysis using a bench-top spectrophotometer (if
applicable)
. Confirmation of backup generator and/or UPS power transfer to the UV reactor. This
may necessitate simulation of line power failure to trigger the backup power supply.
It is recommended that the backup power supply be tested for a minimum of two
separate one-hour periods.
The performance testing should be tailored to the specific UV installation. An example
monitoring program for a 4-week performance test is shown in Table 5.1.
UV Disinfection Guidance Manual 5-8 June 2003
Proposal Draft
-------
5. Start-up and Operation of UV Installations
Table 5.1 Example Monitoring During a Four Week Performance Test
Frequency
Continuous
Weekly
Twice during
testing period
After 4 weeks,
1000MC
cycles or one
Off-line
chemical clean
(OCC)
Task
Confirm the operating
setpoint(s)
Develop energy
efficient operation
Check the on-line UVT
monitor calibration
Check UV intensity
sensor calibration
Switch to standby
reactor
Switch to standby
power or UPS
Inspect lamp sleeves
for fouling
Notes
Monitor reactor operation to confirm compliance with the
setpoint(s) established during validation.
Monitor the power consumption. Test the automatic
operation and power consumption under the flow and
water quality variations to determine if energy efficiency
improvements can be made within the validation
constraints.
Check the on-line UVT monitor against a bench-top
spectrophotometer to determine if the on-line unit is in
calibration.
Check the duty sensor against a reference sensor, using
the recommended protocol (section 5.3.2.2) to
determine whether the duty sensor is in calibration.
Monitor the time it takes to switch to a standby reactor to
determine if there will be off-specification operation
during switchover.
Monitor the time it takes to switch to the standby power
supply to determine if there will be off-specification
operation because of power transfer.
Remove a sleeve from the reactor and inspect as
recommended in section 5.3.2.3.
Any off-specification time and flow should be recorded during all performance tests, and
these results should be evaluated to ensure that off-specification requirements are met. During
performance testing, any component that is not operating properly should be corrected and
retested to ensure satisfactory operation. This may necessitate manufacturer involvement,
especially if specifications in the contract documents were not met. Following performance
testing, ongoing monitoring and recording of reactor operation should continue at a reduced
frequency as discussed in section 5.4 and as required by the State.
5.1.4 Operations and Maintenance Manual
The O&M manual should be site-specific and based on as-built drawings, manufacturer's
shop drawings, operating procedures, recommended maintenance tasks, and results from the
performance testing. If possible, the O&M manual should be developed prior to routine
operations. At a minimum, O&M manuals should include the following items:
UV Disinfection Guidance Manual
Proposal Draft
5-9
June 2003
-------
5. Start-up and Operation of UV Installations
• Federal and State regulatory requirements and guidelines
• Overall treatment objectives
• Role of the UV installation in the overall disinfection strategy
• Relationship to adjoining unit processes
• UV reactor design criteria
. UV reactor validation criteria
. General description of UV installation
. Controls and monitoring
. Standard operating procedures
« Start-up procedures
. Shutdown procedures (manual and automatic)
. Safety issues
• Emergency procedures and contingency plan
. Alarm response plans
. Preventative maintenance needs and procedures
• Equipment calibration needs and procedures
• Troubleshooting guide
. Equipment component summary
. Spare parts inventory
. Contact information for equipment manufacturers and technical services
5.2 Operation of UV Installations
The operation of UV installations will vary based on the UV manufacturer, the UV
reactor configuration, and the dose control strategy. This section discusses the required and
recommended operational and routine'start-up and shutdown procedures that are common to all
UV reactors. The operational tasks presented in this section are general in nature, and the
specific operational procedures for the installed UV reactors should be developed with assistance
from the manufacturer and UV installation designer. Examples of how to determine the
operational requirements are presented in section 5.5.
Uy Disinfection Guidance Manual 5-10 . June 2003
Proposal Draft
-------
5. Start-up and Operation of UV Installations
5.2.1 Operational Requirements
To receive inactivation credit, the UV reactors are required to operate within the
validated limits (40 CFR 141, Subpart W, Appendix D). When a UV reactor is operating outside
of these limits, the UV reactor isioperating off-specification as described previously. Unfiltered
systems that use UV disinfection to meet the Cryptosporidium treatment requirement of the
Long-Term 2 Enhanced Surface Water Treatment Rule (LT2ESWTR) must demonstrate that at
least 95 percent of the water delivered to the public during each month is treated by UV reactors
operating within validated limits (40 CFR 141.721(c)(2)). In other words, the UV reactors
cannot operate off-specification for more than 5 percent of the water delivered to the public.
The LT2ESWTR does not establish an off-specification requirement for filtered systems;
however, States may adopt a 5 percent off-specification or more stringent requirement.
Although the specific criteria limiting off-specification water are defined by the State, the United
States Environmental Protection Agency (EPA) recommends that the UV reactors be operated to
minimize off-specification water. The UV reactors must operate under the validated conditions
that are determined based on validation testing (section 5.5) (40 CFR 141, Subpart W, Appendix
D). The specific monitoring requirements associated with off-specification are described in
section 5.4.
5.2.2 Recommended Operational Tasks
UV reactors typically use automatic control systems and do not need significant
operational attention. This section outlines the general operational tasks that are recommended
(Table 5.2). Site-specific operational tasks should be determined by the manufacturer, UV
installation designer, and facility operators, and should be described in the O&M manual (section
5.1.4). Recommended maintenance tasks are discussed in section 5.3.1.
Table 5.2 Recommended Operational Tasks for the UV Reactor
Frequency
Daily
Weekly
Monthly
Serrii-
annually
Recommended Tasks
• Perform overall visual inspection of the all UV reactors.
• Ensure system control is on automatic mode (if applicable).
• Check control panel display for status of system components and alarm status and
history.
• Ensure all on-line analyzers, flowmeters, and data recording equipment are operating
normally.
• Review 24-hour monitoring data to ensure that the reactor has been operating within
validated limits during that period.
• Initiate manual operation of wipers (if provided) to ensure proper operation.
• Check lamp run time values. Consider changing lamps if operating hours exceed
design life or UV intensity is low.
• Check ballast cooling fans for unusual noise.
• Check operation of automatic and manual valves.
UV Disinfection Guidance Manual
Proposal Draft
5-11
June 2003
-------
5. Start-up and Operation of UV Installations
5.2.3 Start-up and Shutdown of UV Reactors
UV reactors may be turned on and off regularly in response to varying flowrate and water
quality. This section describes the routine start-up procedures, shutdown procedures, and
winterization of the UV reactors. The routine start-up and shutdown procedures shown are not
all inclusive. Utilities should modify these procedures based on the specific manufacturer's
recommendations and operating requirements for their system.
5.2.3.1 Routine Start-up
The following start-up procedure serves as an example procedure. The UV reactors
should be operating within validated conditions once the start-up sequence is complete.
1. Follow site-specific procedures for removal of lockouts and tag-outs of the power
supply and control panel.
2. Ensure all lamp and ground connections are properly made. Verify that all incoming
power conductors, including ground conductors are properly terminated.
3. Ensure that the lamp ends and all other reactor ports are covered and/or sealed to
eliminate the potential for operator exposure to UV light.
4. Ensure the breakers are turned on, and all electrical cabinets and equipment are clear
and closed.
5. Initiate the UV reactors'start-up sequence.
6. Initiate water flow (if it is not automatically done in UV reactor controls) to the
reactor and gradually increase the flow until the minimum flow required for lamp
codling is reached. The water exiting the reactor is not disinfected and is considered
off-specification.
7. Verify that all air is purged from reactors (i.e., reactor completely full). Check the
top of the reactor for heat buildup, which indicates an air pocket.
' 8'. Check the UV reactor control panel to ensure that all of the lamps are on and all of
the monitoring parameters are being displayed.
9. Check and resolve any system alarms being displayed.
10. Ensure all of the on-line analyzers (UV intensity sensors and UVT monitors, if
applicable) and flowmeters are operating as intended.
11. After lamp warm-up period, increase flow to the minimum validated flow (if flow is
not automatically adjusted with UV reactor control sequence).
UV Disinfection Guidance Manual 5-12 June 2003
Proposal Draft
-------
5. Start-up and Operation of UV Installations
12. Verify correct flow split between parallel UV reactors using flowmeters and/or
differential pressure gauges.
13. Verify that the UV reactor is operating within validated limits.
5.2.3.2 Routine Shutdown
UV reactors will need to be shut down periodically for maintenance or to accommodate
water quality or flow changes. The main steps involved in shutting reactors down are as follows:
h Throttle the effluent valve (if not part of the control sequence) to reduce flow through
the reactor to the minimum required for cooling. If complete closure of the effluent
valve can be accomplished without overheating the lamps, it is recommended.
2. De-energize the reactors.
3. Close effluent valve if not completed in Step 1. The water exiting the de-energizing
reactor is considered off-specification.
4. If maintenance is being performed, the following steps should be followed. If the UV
reactor is to be placed on standby, the following steps are not necessary.
5. Follow lock out and tag-out procedures for the facility.
6. Drain the reactor if necessary for the specific maintenance task.
7. Inspect and repair or replace any necessary equipment.
After an extended shutdown period (greater than 30 days), the operator should perform a
cleaning and then inspect the lamp sleeves for fouling. Additional cleaning may be necessary
prior to start-up.
5.2.3.3 Winterization
In most drinking water applications, the UV reactors will probably be located within a
building. However, in some instances, the reactors may be located in unheated concrete vaults.
When it is necessary to shut down a UV reactor for an extended period of time and freeze
damage is possible, the UV reactors should be winterized in accordance with the manufacturer's
recommendations.
5.3 Maintenance of UV Reactors
There are no specific regulatory requirements for maintenance of a UV reactor.
However, the UV reactors need to be maintained to ensure that disinfection requirements are
met. Poor maintenance may cause the UV reactors to be operating off-specification. As part of
UV Disinfection Guidance Manual 5-13 , June 2003
Proposal Draft
-------
5. Start-up and Operation of UV Installations
the maintenance tasks, UV reactor components will need to be replaced; therefore, an inventory
of spare parts is necessary. These tasks are described in this section.
5.3.1 Summary of Recommended Maintenance Tasks
Table 5.3 summarizes the recommended maintenance tasks and refers to the general
guidelines for those tasks that are discussed in section 5.3.2. Before any maintenance is
.performed, the main electrical supply to the UV reactors should be disconnected, lockout and
tag-out protocol should be followed, and the operator should wait at least 5 minutes (or as
recommended by the manufacturer) for the lamps to cool down and energy to dissipate.
Table 5.3. Recommended Maintenance Tasks
Frequency
Weekly
Monthly
Monthly
WhenUV
intensity
sensor fails
calibration
check
Monthly
(OCC)
Semi-annually
(OMC)
Semi-annually
(OMC)
Annually
Annually
Task
General Guideline
' Section Reference
Check on-line UVT
monitor calibration
section 5.3.2.5
Check reactor
housing, sleeves, and
wiper seals for leaks
UV intensity sensor
calibration check
protocol
section 5.3.2.2
Replace duty sensor
with calibrated backup
sensor
section 5.3.2.2
• Check cleaning
efficiency
section 5.3.2.4
Check cleaning fluid
reservoir (if provided)
section 5.3.2.4
Calibrate reference
sensor
section 5.3.2.2
Test-trip GFI
section 5.3.2.8
Action
Calibrate UVT monitor when manufacturer's guaranteed
measurement uncertainty is exceeded;
Replace housing, sleeve, or wiper seals if damaged or
leaking.
Check the sensor calibration at the lamp power utilized during
routine operating conditions (e.g. , the majority of operation).
A sensor is out of calibration when it fails the criteria shown in
section 5.3.2.2
• Check the reference sensor with second reference sensor
or two other duty sensors to ensure the first reference
sensor is calibrated.
• If reference sensor is properly calibrated, replace the duty
sensor with calibrated sensor, and send the duty sensor
that failed calibration to the manufacturer.
• Check the replaced sensor one hour later.
. Record UV intensity sensor reading.
• Extract one sleeve per reactor (or bank of lamps for low
pressure high output (LPHO) reactors) for inspection.
• Check remaining sleeves if fouling is observed on the first
sleeve.
• Manually clean sleeve(s) if fouling is seen on the sleeves.
> Record UV intensity sensor reading and compare to original
reading after cleaning.
• Replace sleeve if UV intensity is not restored to validated
level.
Replenish solution if the reservoir level is low. Drain and
replace solution if the solution is discolored.
Send the reference sensor to the manufacturer for calibration.
Maintain ground fault interrupt (GFI) breakers in accordance
with the manufacturer's recommendations.
UV Disinfection Guidance Manual
Proposal Draft
5-14
June 2003
-------
5. Start-up and Operation of UV Installations
Table 5.3. Recommended Maintenance Tasks (continued)
Frequency
Manufacturer's
recommended
frequency
Lamp/
manufacturer
specific
When lamps are
replaced
Sleeve/
Manufacturer
specific
Pressure gauge
manufacturer
specific
Manufacturer
specific
Manufacturer
specific
Manufacturer
specific
Task
General Guideline
Section Reference
Check flowmeter
calibration
section 5.3.2.6
Replace lamp
section 5.3.2.1
Properly dispose of
lamps
section 5.3.2.1
Replace sleeve
section 5.3.2.3
Check operation of
the pressure gauges
that are used to
confirm flow split (if
applicable)
section 5.3.2.6
Clean UVT monitor
Inspect OMC drive
mechanism
Inspect ballast
cooling fan
Action
Calibrate flowmeter when manufacturer's guaranteed
measurement uncertainty is exceeded.
Replace lamps when any one of the following conditions
occur:
. Initiation of low UV intensity alarm (UV intensity equal to or
less than set point value) after verifying that this condition is
caused by low lamp output.
• Initiation of lamp failure alarm.
Send spent lamps to a mercury recycling facility or back to the
manufacturer.
Replace sleeve every 3 to 5 years or when damage, cracks,
or excessive fouling significantly decreases UV intensity of an
otherwise acceptable lamp to the minimum validated intensity
level. The replacement frequency should be adjusted based
on operational experience.
Replace the pressure gauge if deemed faulty by
manufacturer's evaluation procedure.
Clean according to manufacturer's recommended procedure.
Inspect and maintain OMC drive routinely as recommended
by the manufacturer.
Check the ballast cooling fans for dust buildup and damage.'
Replace if necessary.
5.3.2 General Guidelines for UV Reactor Maintenance
This section describes general guidelines for UV reactor components that relate to
maintenance tasks. Specific operations, maintenance, and monitoring tasks are described
individually in later sections. These latter sections also refer back to this section as a reminder of
the general recommendations.
5.3.2.1
UVLamp Characteristics
UV lamp output decreases over time, and UV lamps will need to be replaced periodically
to maintain sufficient UV intensity (i.e., the validated UV intensity setpoint). Replacement
lamps should be identical to those used during reactor validation with respect to arc length, lamp
envelope material and dimensions, amount of mercury, and spectral output. If the lamps
supplied are not equal to the lamps used during validation, the UV reactor is not operating as
validated and is considered off-specification.
U V Disinfection Guidance Manual 5-15 June 2003
Proposal Draft
-------
5. Start-up and Operation of UV Installations
If the mercury content or power rating changes, the different lamp should be assessed by
comparing UV intensity sensor readings, after burn-in, to the lamps that were validated to
determine if the new lamps are equal to the validated lamps. If the sensor reading is equal to or
greater than that of the validated lamps after burn-in, the different lamps are acceptable and
comparable to the validated lamps. However, if a utility replaces the lamps with higher power
lamps to receive higher log inactivation credit, validation testing should be performed to confirm
performance. Lamp manufacturers should also provide documentation of lamp output decay
characteristics, guaranteed life, and lamp burn-in period. This information will help the utility
determine the lamp replacement frequency. It should be noted that different lamps might have
different aging characteristics, which may affect operations and maintenance costs.
The frequency of UV lamp replacement can be based on a utility-determined schedule,
lamp operating hours, or the UV intensity reduction as measured by the UV intensity sensor
(after sleeve and sensor window cleaning); lamp replacement recommendations are discussed in
section 5.3.2.1. During replacement, the lamps and sleeves should be handled in accordance
with manufacturer recommendations, using clean cotton, powder-free latex, or vinyl gloves
because fingerprints can cause damage to the lamps or sleeves during operation.
Lamp manufacturers are required to determine whether their products exhibit the toxicity
characteristic for mercury and whether their lamp is regulated as a universal hazardous waste
under Subtitle C of Resource Conservation and Recovery Act (RCRA) [40 CFR Part 260,261,
264 and 273]. Currently, most UV lamps exceed these toxicity characteristics and require
regulated disposal. As such, these lamps should be sent to a mercury recycling facility where the
^mercury is recovered and lamp components are recycled. Some UV reactors and lamp
manufacturers will accept spent or broken lamps for recycling or proper disposal (Dinkloh 2001;
Lienberger 2002; Gump 2002). Utilities should contact their lamp manufacturer to determine if
they accept spent lamps or should contact their State for a list of local mercury recycling
facilities.
5.3.2.2 UV Intensity Sensors
Well performing UV intensity sensors are necessary to assess whether the validated UV
intensity is being achieved. Sensor calibration, rotation, and placement affect operation. This
section describes these effects and provides recommendations to minimize them.
There are two types of sensors used for UV reactor operation: duty and reference sensors.
Duty sensors are on-line sensors and continuously monitor UV intensity, while the reference
sensors are off-line sensors used to assess the duty sensor performance. Therefore, the reference
sensor specifications should exactly match those of the duty sensors, so that a valid comparison
can be completed. Both duty and reference sensors are described in this section.
Duty UV Intensity Sensor Calibration
Prior to installation, manufacturers calibrate the UV intensity sensors. However, over
time the sensor may drift out of calibration. Because these sensors are vital to assessing the UV
disinfection performance, the calibration of each sensor should be checked at least monthly
UV Disinfection Guidance Manual 5-16 June 2003
Proposal Draft
-------
5. Start-up and Operation of UV Installations
against the reference sensor. To assess the calibration, the following sensor calibration check
protocol should be followed:
1. Measure the UV intensity with the duty sensor, and record the measurement result.
2. Replace the duty sensor with the reference sensor in the same location (i.e., port) as
the duty sensor used in Step 1.
3. Measure the UV intensity with the reference sensor and record the measurement
result.
4. Determine if Equation 5.1. holds true for the two UV intensity sensor readings:
fin- "\
I
Ref
where
I Duty
O" Duty
O Reference
Equation 5.1.
Intensity measured with the reference sensor (mW/cm2)
Intensity measured with the on-line sensor (mW/cm2)
Measurement uncertainty of the on-line UV intensity sensor (%) as
provided by the UV manufacturer in the validation report
Measurement uncertainty of the reference UV intensity sensor (%)
as provided by the UV manufacturer in the validation report
5. Replace the duty sensor with another calibrated duty sensor if the relationship
Equation 5.1 does not hold true.
The calibration of the UV intensity sensor is sensitive to the power level of the UV lamps
(Swaim et al. 2002). To most effectively compare the duty sensor to the reference sensor, the
power level should be set at the level typically used during routine operation (e.g., the majority
of operation).
UV Intensity Sensor Rotation
Some UV intensity sensors are sensitive to their rotational alignment within the sensor
port and will have different readings at different rotations. This may be due to the UV intensity
sensor configuration (e.g., acceptance angle). Section A.3.5 discusses UV intensity sensors
configurations in more detail. The sensors should be rotated until the lowest UV intensity
reading is obtained for routine monitoring purposes. Alternatively, UV reactors may be designed
so the UV intensity, sensors are keyed in the same rotational position at all times. This may not
be an issue for all UV intensity sensors.
Measuring Lamp Output Variability
UV lamp output differs for each lamp, depending on lamp age and lot. As discussed in
section 2.4.6, a sensor measures the UV intensity at its location in the UV reactor and cannot
assess lamp output variability unless there is one sensor per lamp. Many low pressure (LP) or
LPHO reactors have one sensor to monitor a bank of lamps, and some MP reactors use one UV
UV Disinfection Guidance Manual 5-17 June 2003
Proposal Draft
-------
5. Start-up and Operation of UV Installations
intensity sensor to monitor more than one lamp in the reactor. The effect of variable lamp.output
is accounted for in the validation protocol safety factor as discussed in section F.3. For routine
operation, the oldest lamp should be placed in the position closest to the UV intensity sensor if
one sensor monitors multiple lamps.
Reference UV Intensity Sensor
The reference sensor should be calibrated at least once per year at a qualified facility
(e.g., manufacturer) to ensure that it is calibrated properly for the regular duty sensor calibration
checks. The reference sensor should not be exposed to UV light for longer than it takes to
perform the reference sensor measurement. When not in use, the reference sensor should be
stored under conditions that will maintain its integrity and accuracy as recommended by the
manufacturer. If the reference sensor is found to be out of calibration, the calibration interval
should be shortened. One indicator that the reference sensor itself may be out of calibration is if
it shows that all on-line sensors are out of calibration. Some utilities may choose to have
multiple reference sensors to help determine if one reference sensor is out of calibration, as a
replacement reference sensor, or to allow multiple duty sensors to be checked simultaneously.
5.3.2.3 Lamp Sleeves '•
Lamp sleeves degrade over time due to solarization (section 2.4.4) and internal sleeve
fouling, resulting in cloudiness and a loss of UV transmittance. Abrasion of the sleeve surface
during handling or mechanical cleaning may also be a contributing factor to the loss of UV
transmittance. Sleeve transmittance loss is reflected in the UV intensity sensor reading and,
therefore, does not need to be monitored. However, a low UV intensity sensor reading may be
from sleeve transmittance loss and should be considered when troubleshooting the cause of this
problem (as discussed in section 5.6.1). Sleeves will need to be replaced in the case of UV
transmittance loss or other damage. '
Sleeves should be replaced every 3 to 5 years or when damage, cracks or excessive
fouling diminishes UV intensity to the minimum validated intensity level, whichever occurs first.
This replacement frequency should be increased or decreased based on operational experience.
Replacement sleeves should be identical to the sleeves used during validation, meet the design
and UV manufacturer's material and construction specifications, and be certified as described in
section F.6.3. The sleeves should be handled in accordance with manufacturer
recommendations, using clean cotton, powder-free latex, or vinyl gloves because fingerprints can
cause damage to the sleeves during operation. When the sleeves are replaced, the manufacturer's
procedure should be closely followed because the lamp sleeve can crack and break from over-
tightening of the compression nuts that hold it in place.
5.3.2.4 Fouling
As discussed in Chapters 2 and 3, the lamp sleeves and UV intensity sensors/windows
may foul over time, depending on the water quality, lamp type, and cleaning regime. This
section describes possible cleaning techniques and provides some specific recommendations for
addressing fouling issues. .
UV Disinfection Guidance Manual 5-18 June 2003
Proposal Draft
-------
5. Start-up and Operation of UV Installations
Sleeve and UV Intensity Sensor Surface/Window Fouling
There are two types of sleeve cleaning techniques as discussed in section 2.4.5. The first
type is an OMC system, which typically utilizes an automatic mechanical wiper (e.g., O-ring,
brush) to wipe the surface of the sleeve at a prescribed frequency. Some OMC systems have O-
rings with cleaning fluid enclosed in them to enhance cleaning. The second type is an OCC,
which is also a called flush and rinse system. OCC systems are off-line, manual systems that
pump cleaning solution (typically an acid) into the reactor and circulate the solution for a period
of time. Helsinki Water uses an OCC system; a description of their cleaning regime is discussed
in Appendix O. Also, OCC systems clean the sensor wetted surface/window; however, OMC
systems may not, depending on the UV reactor.
The frequency of cleaning is site-specific. An appropriate sleeve cleaning frequency
(manual or automatic) can be determined based on the rate of fouling during the start-up period,
which can be assessed by monitoring the UV intensity sensor measurement. For routine
operation, the cleaning frequency should be increased or decreased based on the amount of
fouling left on the sleeves after the cleaning cycle and the loss of UV intensity prior to cleaning.
Sleeves should initially be inspected for fouling every six months if OMC is employed
and every month if OCC is used. This frequency should be adjusted after 2 years of operating
data are available. A decrease in UV intensity may indicate sleeve fouling, and sleeves should
be inspected if fouling is the suspected cause of the UV intensity drop. In addition, the sensor
windows (if applicable) should be inspected for fouling and supplemental cleaning should be
conducted if necessary, according to the manufacturers recommendation.
For sleeve inspection, one sleeve per reactor (or bank of lamps for LP or LPHO reactors)
should be inspected. The sleeves should be handled in the same manner as described for UV
lamps. If damage or fouling is observed, the remaining sleeves should be inspected. External
fouling can be difficult to identify. Sleeve discoloration is more easily seen by laying the sleeve
on a clean, white, lint-free cloth along side of a new sleeve. If streaks are observed, this may
indicate that the OMC wiper material may be worn or damaged or not aligned properly;
therefore, the wiper should also be inspected. If fouling is observed, the cleaning frequency
should be increased, and supplemental manual cleaning should be conducted as necessary.
If manual cleaning (i.e., beyond routine OCC or OMC cleaning) of lamp sleeves is
necessary, this should be done according to manufacturer recommendations and procedures.
Abrasive cleaners or pads that might scratch the lamp sleeve should not be used. In addition, the
inside of the sleeve should be dry prior to re-installation because water or cleaning solutions
could cause a coating to form during operation. One method of drying the sleeve is to use
isopropyl alcohol and a lint-free cloth; however, there should not be any alcohol left inside the
sleeve after this procedure. As noted earlier, when the sleeves are re-installed after inspection,
the manufacturer's procedure should be closely followed to avoid over-tightening of the
compression nuts.
If OMC cleaning is used, the OMC wipers should be checked for deformation or
degradation at the same time the sleeves are checked. If the OMC.cleaning uses a cleaning
solution, the cleaning solution reservoir should be checked every six months to determine
UV Disinfection Guidance Manual 5-19 - June 2003
Proposal Draft
-------
5. Start-up and Operation of UV Installations
whether more solution should be added. In addition the solution should be replaced if it is
discolored or if the OMC system is not effectively cleaning the sleeve.
Fouling During Periods of Standby
When the UV reactors are out-of-service and full of water, the sleeves may become
fouled (Toivanen 2000). The rate of fouling is site-specific and depends on the influent water
quality. UV reactors equipped with OMC should continue to clean the sleeves even though the
UV reactor is off-line. This should prevent fouling of the sleeves. For UV reactors that do not
include OMC, the utility should consider draining the UV reactor if it is off-line for more than
one week. However, this period could be shorter or longer, depending on the water quality.
After an extended shutdown period of greater than 30 days, the operator should perform a
cleaning (OCC or OMC) and then inspect the lamp sleeves for fouling. Additional cleaning may
be necessary prior to start-up after extended periods of standby.
5.3.2.5 On-line UVT Monitor Calibration
On-line UVT measurements should be compared to those obtained using a bench-top
spectrophotometer every week. The grab samples that are used to check calibration should be
collected from a location close to the on-line UVT monitor sampling point. The frequency may
be decreased or increased based on the performance demonstrated over a one-year period. For
example, the frequency could be reduced to once per month if the UVT monitor was consistently
within the calibration specification for over a month during the first year of monitoring.
5.3.2.6 Flowmeter Calibration
The flowmeter calibration should be checked at the frequency recommended by the
manufacturer. Techniques for verifying calibration are discussed in section 5.1.2.2.
Some UV installations will not have dedicated flowmeters and may use a combination of
an upstream flowmeter and differential pressure gauges to verify flow split as described in
section 3.3.1.2. If differential pressure is used to verify the flow split, the calibration of the main
flowmeter should be checked at the manufacturer's recommended frequency and the accuracy of
the pressure gauges should be periodically verified using a reference gauge or redundant gauge
to confirm measurement consistency between the gauges.
5.3.2.7 UV Reactor Temperature
UV lamps operate at high temperatures (as discussed in section 2.4.2) and need water
flow to maintain them at their optimal temperature and to prevent overheating. Another concern
related to overheating is the formation of air pockets in the UV reactor. Air pockets can cause
the UV reactor temperature to increase and may alter the flow pattern in the UV reactor. UV
lamps can break if their threshold temperature is exceeded, which is discussed in more detail in
section N.2.1.2.
UV Disinfection Guidance Manual 5-20 June 2003
Proposal Draft
-------
5. Start-up and Operation of UV Installations
The water temperature should be monitored. If the water temperature exceeds
manufacturer recommendations, the UV reactor should be shut down. Water level monitoring or
reactor temperature monitoring are typically included in the packaged control systems for the
UV reactor. The water level monitoring should detect any air pockets in the UV reactor. During
start-up and whenever necessary, air should be bled from the UV reactors. The UV reactor
surface can become hot during operation if air pockets or stagnant water are present in the UV
reactor. As a result, nothing unrelated to reactor equipment should be in external contact with
the reactor while in service.
5.3.2.8 Electrical Concerns
UV reactors operate at high voltages. Before any maintenance on the UV reactor is
performed, the main electrical supply to the UV reactors should be disconnected and the operator
should wait at least 5 minutes for the lamps to cool down and energy to dissipate. Lockout, tag-
out procedures and all applicable codes should be followed. The UV reactors should not be
operated if any of the control panel doors are open, and water should not be sprayed around the
electrical equipment.
Typically, power to the UV reactors are provided via a distribution transformer, a circuit
breaker, a disconnect switch at the UV reactor, and related wires and conduits. If maintenance is
necessary on the control panel, the main electrical supply should be disconnected. The power to
the lamps is typically delivered through individual GFI circuit breakers and ballasts.
Maintenance of the GFI breakers is important because they are safety devices that protect the
operators when they are working around the powered equipment. The GFI breakers should be
test-tripped at least once per year and should be maintained in accordance with the
manufacturer's recommendations. Ballast output should be monitored through the UV reactor's
control panel. Irregularities or instabilities in ballast output may indicate a problem with the
electrical feed or the ballast itself.
The ballasts, typically connected between the GFI breakers and the lamps, are electrical
components that regulate the line power to match the input requirement of the lamps. Three
types of ballasts are typically used with UV reactors for converting power: electronic ballasts,
electromagnetic ballasts, and transformers. Electromagnetic ballasts and transformers are very
similar in that both contain a specially wound coil of wire that is used to control the current to
the lamp. Typically inductors or capacitors are used to allow step adjustment of the lamp output.
Electronic ballasts, sometimes referred to as solid-state ballasts, contain semiconductors and
other electronic components that allow the ballast to behave like a switching power supply.
Electronic ballast technology allows nearly continuous adjustment of lamp output.
Power regulation, particularly with electromagnetic ballasts and transformers, will result
in significant heat build-up within the ballast enclosure. If the excess heat is not dissipated, it
can damage the ballast electronics and cause failure. A cooling system is normally provided
with LPHO and medium pressure (MP) reactors to maintain the ballast temperature below the
maximum specified limit. LP reactors typically do not need ballast cooling. The ballast cooling
system should be inspected and maintained as recommended by the manufacturer.
UV Disinfection Guidance Manual 5-21 June 2003
Proposal Draft
-------
5. Start-up and Operation of UV Installations
Power use depends on the specific UV reactor and how it adjusts to changes in water
quality and flow. Power use should be monitored as operational adjustments are made for
changes in flow, UV intensity, UVT, lamp aging and output, and other factors. This information
can be used to determine the most energy efficient operating strategies. For example, some UV
reactors can both increase lamp output and energize additional lamps to respond to a low UV
intensity reading. The power use under these two strategies can be compared to determine which
is more energy efficient.
5.3.3 Spare Parts
The actual life of a component is a function of many variables, including operating
conditions, maintenance practices, the quality of the materials of construction, and fabrication
practices. As a consequence, it is impossible to predict the actual life of a component. To
overcome the operational impacts of this uncertainty, an adequate inventory of critical spare
parts should be maintained to ensure reliable and consistent performance of the UV reactors and
minimize the delivery of off-specification water.
All UV components have a design life and a guaranteed life. The design life represents
the expected duration of operation. The guaranteed life incorporates the risk, assumed by the
manufacturer, to account for the uncertainties associated with the quality of materials used,
production, and operating conditions. Generally, guarantees are conditional in nature and are
valid under certain operating conditions. For example, guaranteed lamp life is normally linked to
the lamp power setting or the number of on/off cycles per 24-hour period. If equipment failure
occurs during the warranty period and if all of the warranty conditions are satisfied, the
manufacturer will typically replace the component and charge the owner a prorated fee for the
use of the replaced component.
Table 5.4 provides typical design and guaranteed lives for major UV reactor components.
These represent current industry trends and are likely to change, as more operation and
maintenance information becomes available and technological advances occur. Manufacturers
should be contacted directly for details specific to their equipment.
UV Disinfection Guidance Manual 5-22 June 2003
Proposal Draft
-------
5. Start-up and Operation of UV Installations
Table 5.4. Design and Guaranteed Lives of Major UV Components
(Based on Manufacturers' Input)
Component
Low pressure lamps (LP and LPHO)
MP lamps
Sleeve
UV Intensity Sensor
UVT monitor
Cleaning systems
Ballasts
Design Life1
12,000 hours
10,000 hours
8 to 1 0 years
3 to 1 0 years
3 to 5 years
3 to 5 years
10 to 15 years
Guaranteed Life 2
8,000 -12,000 hours
4,000 - 8,000 hours
1 to 3 years
1 year
1 year
1 to 3 years
1 to 3 years
Expected duration of operation
2 Accounts for variability of material quality, production, and operating conditions.
The following is a suggested minimum inventory of spare parts, expressed as a
percentage of the installed number. A full list of spare parts will vary depending on the specific
equipment installed and should be coordinated with the UV manufacturer. The number of spare
parts needed depends on the guaranteed life of the specific equipment. For example, a higher
percentage of MP lamps may be necessary compared to LP lamps because the guaranteed lamp
life is less for MP lamps, and therefore they need to be replaced more frequently.
• UV lamps-10 percent with a minimum of two lamps
• Sleeves-5 percent with a minimum of one sleeve
• O-ring Seals- 5 percent with a minimum of two seals
• OMC wipers- 5 percent with a minimum of two wipers
• OMC wiper drive mechanisms- 2 percent with a minimum of one drive
. Ballasts- 5 percent with a minimum of one unit
« Ballast cooling fan-1 unit
• Duty UV intensity sensor- minimum of 2 units (adjust number based on operating
experience)
. Reference UV intensity sensor- 2 units
• On-line UVT monitor-1 unit (if used for control strategy)
UV Disinfection Guidance Manual
Proposal Draft
5-23
June 2003
-------
5. Start-up and Operation of UV Installations
5.4 Monitoring, Recording, and Reporting of UV Installation Operation
Operation of the UV reactors should be monitored to ensure the reactors are operating
within validated limits, to diagnose operating problems, to determine when maintenance is
necessary, and to maintain safe operation. This section discusses the required and recommended
monitoring, recording, and reporting activities for UV installations.
5.4.1 Monitoring and Recording Frequency for Compliance Parameters
Utilities must monitor each reactor to determine whether it is operating within validated
conditions. They also must determine the percentage of flow that was treated within validated
limits (40 CFR 141, Subpart W, Appendix D). The flow is off-specification when a reactor is
operating outside of validated limits. The monitoring parameters depend on the control strategy
used and the validation results. Table 5.5 presents the monitoring parameters for each control
strategy, the criteria for when off-specification occurs, and examples of off-specification
operating conditions.
Table 5.5 Off-Specification Operations for Each Control Strategy
Control
Strategy
UV intensity
setpoint
UVTandUV
intensity
setpoints
Calculated
dose
Parameters
Monitored
UV intensity,
flow/rate, lamp
status
UV intensity,
flowrate, UVT;
lamp status
Calculated
dose, flowrate,
UVT, lamp
status
Off-Specification
Anytime these values are
outside of the validated limits
for these parameters
Anytime these values are
outside of the validated limits
for these parameters
Anytime the calculated dose
is below the validated
setpoint (if validation certifies
that the calculated dose can
be used to control the UV
reactor - see section F.2)1
Examples
1 ) UV intensity below setpoint
2} Flowrate outside validated limits
3) UV lamp failure
4) UV intensity sensor failure
1) UV intensity below setpoint
2) Flowrate outside validated limits
3) UV lamp failure
4) UV intensity sensor failure '
5) UVT below setpoint
1) Calculated dose below setpoint
2) Flowrate outside validated limits
3) UV lamp failure
4) UV intensity sensor failure
5) UVT below setpoint
If validation deems that the calculated dose control is not acceptable, the UV reactor should use the
UVT and UV intensity setpoint control strategy.
It is recommended that the required monitoring parameters be continuously monitored for
each UV reactor and that these values be recorded at least once every four hours. These four-
hour records should be used to determine the percentage of flow that is off-specification. Very
small systems (e.g., systems serving less than 500 people) that are unable to record reactor status
every 4 hours (e.g., manual recording is practiced) can consider a reduced recording frequency;
however, the frequency should not be less than once per day and should be approved by the
State. The monitoring guidelines are summarized in Table 5.6.
UV Disinfection Guidance Manual
Proposal Draft
5-24
June 2003
-------
5. Start-up and Operation of UV Installations
Table 5.6 Monitoring Parameters and Recording Frequency
Parameter
General Guideline
Section Reference
(if applicable)
UV intensity
uvr
Calculated dose'
Lamp status
Calibration of UV
intensity sensors
section 5.3.2.2
Recommended
Recording
Frequency
Every 4 hours
Every 4 hours
Every 4 hours
Every 4 hours
Monthly
Notes
The UV intensity must be above the validated
setpoint
The UVTmust be above the validated setpoint. If
not required to be monitored, this information will
assist in determining if low.UV intensity readings
are because of low UVT
The calculated dose must be above the validated
setpoint
The lamps should be energized if water is flowing
through the UV reactor
The UV intensity sensor calibration must be
checked, using sensor calibration check protocol
Only required if necessary for the control strategy (Table 5.11)
5.4.2 Monitoring and Recording for Other Operational Parameters
In order to minimize operational problems, facilitate regulatory compliance, and evaluate
UV reactor performance, it is recommended that additional parameters, beyond those needed for
regulatory compliance, be monitored. Table 5.7 presents these additional parameters
recommended for monitoring and the recommended recording frequency. These recommended
parameters and their monitoring frequency should be adjusted based on site-specific operating
experience. For example, if sleeve fouling is a maintenance issue and supplemental reactor
cleaning is frequent (e.g., monthly), then the fouling parameters should be monitored on a daily
basis as opposed to weekly as shown in the table below.
UV Disinfection Guidance Manual
Proposal Draft
5-25
June 2003
-------
5. Start-up and Operation of UV Installations
Table 5.7 Recommended Monitoring Parameters and Recording Frequency
Parameter
1 General Guideline
Section Reference
(if applicable)
Power draw
section 5.3.2.8
Water Temperature
section 5.3,2. 7
UV lamp on/off cycles
section 5.3.2.1
Turbidity
pH, iron, calcium,
alkalinity, hardness
section 5.3.2.4
UVT monitor calibration
section 5.3.2.5
Age of the following
equipment:
• Lamp
• Ballast
• Sleeve
. UV intensity sensor
Calibration of flowmeter
section 5.3.2.6
Monitoring
Frequency
Continuous
Continuous
Continuous
Daily
Weekly (reduce if
fouling is not
prevalent)
Weekly (reduce if
appropriate based
on operational
experience)
Monthly
Monthly
Recording
Frequency
Every 4 hours
Daily
Weekly
(Total cycles in
a week)
Weekly
Weekly
Weekly
Monthly
Monthly
Notes
This information can be used to
determine the most energy
efficient operation strategies
Monitor to ensure trie high
temperature limit is not exceeded
(usually part of packaged UV
control system)
Monitor to assess status of the
UV lamps since the of on/off
cycles can help assess lamp
aging
Monitor if chemicals (e.g., lime)
are added post-filtration or prior to
UV disinfection (monitoring may
not be necessary for many UV
reactors)
Monitor to help assess fouling
issues if necessary
Information can assist in planning
scheduled maintenance and O&M
budget
Information can assist in planning
scheduled maintenance and O&M
budget
Information can assist in planning
scheduled maintenance and O&M
budget
All data related to UV reactor operation should be gathered, compiled, and stored for
easy retrieval. The recorded data should be stored for at least two years. Appendix M provides
example logs for many of the parameters listed in Table 5.13.
5.4.3 Reporting to the State
Monthly reports must be prepared and submitted to the State. The report must include
the percentage of off-specification flow, which should be based on at least 4-hour records for
each reactor. The State may have additional reporting requirements. In addition, the percentage
of the UV intensity sensors that were checked for calibration must be reported monthly; all
sensors should be checked every month. An example monthly monitoring form is shown in the
Appendix M. The State should be contacted to determine the specific content of the monthly
reports and to coordinate with other reporting requirements.
UV Disinfection Guidance Manual
Proposal Draft
5-26
June 2003
-------
5. Start-up and Operation of UV Installations
5.5 Determination of Validated Operational Parameters
For each UV reactor, the operating conditions associated with a given level of
inactivation credit must be defined based on validation testing results (40 CFR 141, Subpart W,
Appendix D). The validation testing and resultant data that are used to determine these operating
conditions will vary with different control strategies. A detailed discussion of the three common
control strategies is presented in section 3.3.2. A brief description of each of the control
strategies is shown in Table 5.8.
Table 5.8 UV Reactor Control Strategies
Control Strategy
UV Intensity Setpoint
UV Intensity and UVT setpoints
Calculated Dose
Dose Delivery Monitoring and Control Based On
UV intensity sensor measurement
UV intensity sensor and UVT measurement
The calculated UV dose1
1 The UV reactor calculates a UV dose using the UV intensity sensor measurement,
the UVT of the water, and the flowrate.
This section provides example operational requirements based on the validation examples
described in section C.5 of the validation protocol. Each example describes how the operating
requirements are determined based on the control and operation strategy used and the validation
results.
Example 1. UV Intensity Setpoint Control - Single Operational Setpoint for all Conditions
(Section C.5.1)
The simplest operational strategy uses one single UV intensity setpoint for all flows. In
this example, a LPHO reactor that uses the UV intensity setpoint control strategy was validated
at flows between 100 and 500 gallons per minute (gpm) and a UVT range of 84 to 98 percent.
This reactor passed the criteria for 2-log inactivation of Cryptosporidium with an intensity sensor
setpoint of 5 mW/cm2. The validation testing verified that the UV intensity setpoint control
strategy is appropriate for this reactor
Based on this validation, this reactor must operate at a minimum UV intensity sensor
setpoiht of 5.0 mW/cm2 and a flow range between 100 and 500 gpm to claim 2 log
Cryptosporidium credit. The UV intensity setpoint approach accounts for the UVT in the UV
intensity measurement. Therefore, the intensity setpoint of 5.0 mW/cm2 can be used for any
UVT. Although this is a simple and straightforward operating strategy, single setpoint operation
will not be as energy efficient as using a variable setpoint approach, which is described in
Example 2.
Example 2. UV Intensity Setpoint Control - Variable Setpoint Operation for Different
Flow Conditions (Section C.S.2)
The variable UV intensity setpoint approach has a different UV intensity setpoint at
different flowrates. This operation promotes more energy efficient operation compared to the
UV Disinfection Guidance Manual
Proposal Draft
5-27
June 2003
-------
5. Start-up and Operation of UV Installations
single setpoint approach because the UV intensity setpoint can be decreased at lower flows. For
this example, a LPHO reactor that uses the UV intensity setpoint control strategy was validated
under the conditions shown in Table 5,9 and passed the criteria for 3-log inactivation of
Cryptosporidium at each condition.
Table 5.9 Example Validation Data for
Variable Setpoint Operation
Flow (mgd)
0.90
1.2
1.7
2.4
UVT (%)
70
75
83
92
UV Intensity (mW/cm2)
6.1
7.5
10
14
The UV intensity measurements recorded during validation verified that the UV intensity
setpoint approach is appropriate for this reactor. Because of the data collected, this UV reactor
can be operated at a different setpoint for each flow range. These intensity setpoints could be
used in three ways.
1. A single setpoint as described in Example 1. For example, a setpoint of 14 mW/cm2
could be used at between 0.90 and 2.4 mgd with any UVT.
2. Each intensity setpoint could be used over a given flow range as shown in Table 5.10.
The higher UV intensity measurement from each flow range should be used as the
UV intensity setpoint to be conservative.
Table 5.10 UV Intensity Setpoint for Different
Flow Ranges
Minimum Flow
(mgd)
0.90
'• 1.2
1.7
Maximum Flow
(mgd)
1.2
1.7
2.4
UV Intensity
(mW/cm2l_ i
7.5
10
14
3. The intensity setpoints could be interpolated as a function of flowrate. Figure 5.2
presents an equation based on interpolation. For example, for a flowrate of 2 mgd,
interpolation indicates that a setpoint of 12 mW/cm2 is needed to achieve 3-log
inactivation.
UV Disinfection Guidance Manual
Proposal Draft
5-28
June 2003
-------
5. Start-up and Operation of UV Installations
Figure5.2 Example2-Interpolation of
Validation Data to Determine UV Intensity Setpoints
<£ 15 -i
'o
Q. «$—•
"5 E 10
CO O
fl.H
= 5.20x
1 2
Flowrate (mgd)
Example 3. UV Intensity Setpoint Control - Variable Setpoint Operation for Different
Flow Conditions and Inactivation Goals (Section C.5.3) *
s
For this example, a UV manufacturer has completed a matrix of tests at different
flowrates, UVT, and lamp power to develop a relationship between UV intensity readings, log
inactivation credit, and flow. Table 5.11 shows the results of the validation tests.
Table 5.11 Example Validation Data for Variable Setpoint Operation
Flow
(mgd)
5
5
5
10
10
10
. 20
20
20
UV Intensity
(mW/cm2)
5.1
3.3
1.8
9.1
5.6
2.6
15
11
5.6
Cryptosporidium
Log Credit
3.0
2.5
1.0
3.0
2.5
1.0
3.0
2.0
1.0
The UV intensity measurements recorded during validation verified that the UV intensity
setpoint approach is valid for this reactor. In contrast to Example 2, this reactor was validated
for three different levels of Cryptosporidium inactivation credit. For a utility that only is
required to achieve a 2.0-log inactivation, using this reactor would reduce energy costs compared
to a reactor that had only been validated for 3.0-log Cryptosporidium inactivation.
UV Disinfection Guidance Manual
Proposal Draft
5-29
June 2003
-------
5. Start-up and Operation of UV Installations
These intensity setpoints could be used in three ways to achieve 2.0-log Cryptosporidium
inactivation credit with this reactor,
1. A single setpoint as described in Example 1. For example, a setpoint of 11 m W/cm2
could be used at or between 5 and 20 mgd with any UVT.
2. Each intensity setpoint could be used over a given flow range as shown in Table 5.12.
The higher UV intensity measurement from each flow range should be used as the
UV intensity setpoint to be conservative.
Table 5.12 UV Intensity Setpoint for Different
Flow Ranges
Minimum Flow
(mgd)
5
10
Maximum Flow
(mgd)
10
20
UV Intensity
' (mW/cm2)
5.6
11
3. The intensity setpoints could be interpolated as a function of flowrate. Figure 5.3
presents an equation based on interpolation for three different levels of
Cryptosporidium inactivation. For example, for a flowrate of 12 mgd, interpolation
indicates that a setpoint of 3.8 m W/cm2 is needed to achieve 2-log inactivation.
Figure 5.3 Example 3 - Interpolation of Validation Data to Determine UV Intensity
Setpoints at Different Flows and Cryptosporidium Inactivation
8
O
14
12 -
8 -
6 -
4
2 H
0
o 2.0 log Crypto
A 2.5 log Crypto
O 3.0 log Crypto
y = 0.001 x2* 0.517x+ 0.872
= 0.003x2 + 0.346x +0.920
0.968
10 15 20
Flowrate (gpm)
25
UV Disinfection Guidance Manual
Proposal Draft
5-30
June 2003
-------
5. Start-up and Operation of UV Installations
Example 4. UV Intensity and UVT Setpoint Control Strategy - Single Operational Setpoint
for all Conditions (Section C.5.4)
This example uses single operational setpoint as the operating strategy, which is the same
as example 1. However, this example uses both a UV intensity and a UVT setpoint to control
the reactor operation. In this example, a MP reactor that uses the UV intensity and UVT setpoint
control strategy was validated at flows between 0.1 and 0.5 mgd.and a UVT range of 75 to 98
percent. This reactor passed the criteria for 3-log Cryptosporidium inactivation credit with a UV
intensity sensor setpoint of 41 mW/cm2 and a UVT setpoint of 85 percent. .
Therefore, to claim 3-log Cryptosporidium, this reactor must operate under the following
conditions:
. Maintain minimum UV intensity sensor setpoint of 41.0 mW/cm2.
. Operate within a flow range of 0.1 mgd and 0.5 mgd.
. Operate within a UVT range of 85 to 98 percent.
Example 5. Calculated Dose Setpoint Control - Variable Setpoint Operation for Different
Flow Conditions and Inactivation Goals (Section C.5.5)
The calculated dose control strategy uses UVT, UV intensity, and flow measurements to
estimate a UV dose. For this example, a UV manufacturer has completed a matrix of tests at
different flowrates, UVT, and lamp power to develop a relationship between calculated dose, log
inactivation, and flow. A MP reactor that uses the calculated dose control strategy was validated
at flows between 10 to 40 mgd and a UVT range of 75 to 98 percent. Table 5.13 shows the
results of the validation tests.
Table 5.13 Dose Setpoints for Various Log Inactivation of Cryptosporidium
Cryptosporidium
Log Inactivation
1.0
1.5
2.0
2.5
3.0
Calculated Dose
Setpoint
(mJ/cm2)
14
18
23
28
30
UVT Range
(%)
75-98
75-98
75-98
75-98
79-98
The validation tests as described in section C.5.5 verified that the calculated dose
approach is valid for this reactor and that the calculated dose setpoints could be used for the
ranges of flows tested (10-40 mgd). In addition, this reactor can be utilized by utilities that
need different levels of Cryptosporidium inactivation credit. For a utility that only is required to
achieve a 2.0 log inactivation credit, using this reactor would reduce energy costs compared to a
reactor that had only been validated for 3-log Cryptosporidium inactivation credit. Therefore,
this reactor could be operated at any flow between 10 and 40 mgd, the UVT range specified in
UV Disinfection Guidance Manual
Proposal Draft
5-31
June 2003
-------
5. Start-up and Operation ofUV Installations
Table 5.8, and at the specified calculated dose in Table 5.13 to achieve a specific level of
Cryptosporidium inactivation credit. For example, a reactor must operate at a minimum
calculated dose of 28 mj/cm2 and a flow range between 10 and 40 mgd and UVT between 75 and
98 percent to achieve 2,5-log Cryptosporidium inactivation credit.
5.6 Operational Challenges
An excursion from validated limits can be caused by low UV intensity, low UVT, high or
low flowrate, poor UV intensity sensor performance, power quality problems, or a combination
of these conditions. These conditions will need to be resolved quickly to ensure regulatory
compliance because they can result in prolonged off-specification operation. This section
discusses some of the potential operational challenges and suggested corrective measures.
5.6.1 Low UV Intensity or Low Calculated UV Dose
Although the UV intensity and calculated dose control strategies are different, approaches
for addressing either a low UV intensity or low calculated dose are typically the same. This is
because the UV intensity setpoint control strategy uses UV intensity as an indicator for UV dose;
therefore, the causes of a low UV intensity in a UV intensity control strategy and a low
calculated dose in a calculated dose control strategy are similar.
The output of the UV lamps, UV transmittance of the sleeves, status of the UV intensity
sensor, and fouling of both lamp sleeves and sensor windows affect UV intensity sensor
readings. In the UV intensity setpoint control strategy, UV intensity sensors are placed far
enough from the UV lamp to be affected by UVT. In the UV intensity and UVT setpoint or
calculated dose setpoint control strategy, the UV intensity sensors are close to the lamps and
should not be affected by UVT changes.
If one or more UV intensity sensors reads below the required setpoint, the cause could be
low UV lamp output. If the UV lamp life is greater than the design life, the lamp should be
replaced. If the UV intensity is still low, sensor accuracy should be determined by replacing the
duty sensor with the reference sensor. If the duty and reference sensor agree within the required
uncertainty (from validation),, the cause of the low intensity reading may be due to UV intensity
sensor surface or sensor window fouling or sleeve UV transmittance loss. Potential corrective
measures include cleaning of fouled surfaces and replacement of defective sleeves.
Figure 5.4 presents a decision tree for evaluating low UV intensity problems. If the
above strategies cannot be implemented or are not successful in reducing the low UV intensity,
the UV manufacturer or UV installation designer should be contacted to investigate the problem
further. The utility should activate any backup disinfection or consider shutting down the water
treatment plant (WTP) until the UV intensity is within the validated limits. Anytime that the UV
intensity is lower than the validated limit, it should be recorded as off-specification even if this
does not occur at precisely the time (e.g., 4-hour interval) when the 4-hour recording is
completed.
UV Disinfection Guidance Manual 5-32 June 2003
Proposal Draft
-------
5. Start-up and Operation of UV Installations
Figure 5.4 Low UV Intensity or Low Calculated UV Dose Decision Chart
UV Intensity or calculated
dose is below validated
limits
Adjust UV system operation
to compensate for low UV
intensity if not done
automatically by the control
system.
See evaluation of low UVT
(Figure 5.4)
•*Yes
Evaluate and repair
faulty sensor
Is the UV intensity or
calculated dose still low?
Is the UV intensity or
calculated dose still low?
Yes i.
Take Out quartz sleeve
and/or sensor window and
inspect for fouling
Yea».
Clean sleeve and or
sensor surface/ window
Inspect other reactors and
sensors for fouling
Check other lamps and/or
sleeves in other reactors
to see if they need to
be replaced
No*.
Is the quartz sleeves age
beyond the design life?
Is UV intensity or
calculated dosa still low?
Not*.
Yes1-**
Contact UV manufacturer
and/or UV facility designer
to evaluate problem further
and activate backup
disinfection or consider
WTP shutdown
UV Disinfection Guidance Manual
Proposal Draft
5-33
June 2003
-------
5. Start-up and Operation of UV Installations
5.6.2 Low UV Transmittanee
This evaluation of low UVT assumes either that the low intensity evaluation has been
completed and the cause of the low UV intensity was low UVT or that the operational staff has
observed low UVT. Some UV reactors may increase lamp output or number of lamps in service
to accommodate a decrease in UVT. If the system does not sufficiently compensate, or if the UV
reactor cannot adjust lamp output, the UV intensity may go below the validated limits. The steps
for evaluating low UVT are described below.
The first step is to evaluate the UVT monitor function. If UVT is monitored using an on-
line instrument, the utility should verify the low reading with a bench-top spectrophotometer. If
the second measurement differs significantly from the on-line instrument response, appropriate
repair and calibration of the on-line instrument is necessary.
If UVT is determined using grab samples, a duplicate sample should be obtained and
analyzed. If the UVT of the duplicate sample remains low, the spectrophotometer response
should be checked using a phthalate standard (EPA ICR UV254 method or Standard Method
5910). If the spectrophotometer response is determined to be inaccurate, the spectrophotometer
monitor should be calibrated or repaired.
If the low UVT is determined to be real and not due to a faulty instruments, it should be
compared to the validated UVT set point. If UVT is below the validated UVT set point, the
following operational changes should be considered:
. Vary source water blending ratio (if available) to increase UVT.
. Evaluate whether the coagulation process has been optimized for natural organic
matter (NOM) removal and whether the coagulant dose should be increased. Poor
coagulation caused by coagulant under-dosing can lead to increased NOM
concentration and an associated decrease in UVT:
. Increase oxidant dose prior to the UV installation if possible. However, this strategy*
may increase disinfection byproduct (DBF) formation, which must also be evaluated
if this option is used.
. Investigate potential upstream chemical interferences that may be from a process
failure or upset. For example, if the ozone quenching system failed, the UVT would
decrease.
If the above strategies cannot be implemented or are not successful in reducing the low
UVT, the UV manufacturer or UV installation designer should be contacted to investigate the
problem further. The utility may consider shutting down the WTP or activating any backup
disinfection capacity until the UVT is within the validated limits. A decision tree that
summarizes the approach for troubleshooting low UVT is shown on Figure 5.5. Anytime the
UVT is lower than the validated limit, it should be recorded as off-specification even if it does
not occur at precisely the time (e.g., 4-hour interval) that recording is completed.
UV Disinfection Guidance Manual 5-34 June 2003
Proposal Draft
-------
5. Start-up and Operation of UV Installations
Figure 5.5 High UV Absorbance Decision Chart
s on-line UVT monitor used
Is repeat UVT similar
s grab UVT sample
after repeat sampling?
Recalibrate or repair
Is spectrophotometer
response acceptable?
Is the UV system operating
off-specification because of
low UVT?
an UVT be increased through
WTP operation changes?
Contact manufacturer
or UV system designer
Is UVT below the
validation limit?
issue further. Consider
See low UV intensity
Is UV Intensity
or calculate dose below
alidation limi
and calculated dose
No*-
UV Disinfection Guidance Manual
Proposal Draft
5-35
June 2003
-------
5. Start-up and Operation of UV Installations
5.6.3 Rapid Flow Increase or High Flow
It may be possible to compensate for increased flow (depending on validation data) by
completing one or more of the following actions:
« Increasing the output of the UV lamps
• Using additional lamps or banks of lamps
• Using additional UV reactors
The success of these strategies depends on the magnitude of the'flowrate increase and the
type and configuration of the UV reactors. These changes should occur automatically for
reactors that are controlled using PLCs.
If the measured flowrate is higher than the validated limits and cannot be reduced, the
flowmeter and/or differential pressure meter (if used) should be evaluated to determine if it is
functioning properly. Instrument error can be assessed by comparing signals from individual
flowmeters or differential pressure devices to anticipated values based on facility flowrate and
historic operating data. Alternatively, a calibrated clamp-on type flowmeter may be used to
verify flowrates. If the flowmeter is not operating properly, it should be repaired or replaced. If
flow monitoring devices appear to be functioning properly, valve position or blockage may be
the cause of unequal flow distribution and should be evaluated.
If the flow is below the validated limits, one UV reactor should be taken off-line, which
will transfer that flow to the other energized reactors. This change in operation should result in
the UV reactors being within the validated flow range. Anytime the flow is lower or higher than
the validated limit, it should be recorded as off-specification even if it does not occur at precisely
the time (e.g., 4-hour interval) that the recording is completed.
5.6.4 Unreliable UV Intensity Sensor Readings
Consistent UV intensity sensor readings are important to ensure that the UV reactors are
operating within the validated limits. Unreliable UV intensity sensor readings can be described
by one or more of the following behaviors:
• Calibration checks outside of uncertainty specified in the validation testing
. Random fluctuations of greater than 25 percent
» Biased readings (UV intensity sensor reading is offset from the reference sensor
readings by a certain value)
. Unreliable UV intensity sensor readings can be due to UV intensity sensor malfunction,
condensation in the sensor or between the sensor and sensor window, lamp malfunction, poor
grounding, degradation of sensor electronics, or electronic short circuits.
UV Disinfection Guidance Manual 5-36 June 2003
Proposal Draft
-------
5. Start-up and Operation of UV Installations
The UV. intensity sensor and lamp electrical cables should be secured, and a reference or
standby sensor should be compared to the duty sensor reading. If the duty sensor is found to be
defective or. out of calibration, it should be sent to the manufacturer for repair, and the standby
sensor used in its place.
5.6.5 Power Quality Problems
UV lamps can potentially lose their arc if a voltage sag, power quality anomaly, or a
power interruption occurs. Voltage sags as little as 10 to 15 percent from the nominal voltage for
as few as 2 to 5 cycles can cause a UV lamp to lose its arc. LP lamps generally can return to full
operating status within 15 seconds after power is restored. LPHO and MP lamps will need to be
re-struck, which generally requires between 4 and 10 minutes to get to full lamp power, to
restart. LPHO and MP lamps are affected differently from power losses as discussed in more
detail in section 3.1.3.3.
The corrective actions for short-term power failures (e.g., voltage sag) are different for
LPHO and MP reactors. LPHO lamps need to warm-up before the arc can be struck, and MP
lamps need to be cooled before the arc can be struck. Standby MP reactors (i.e., not in operation
when voltage sag occurred) should be energized instead of "warm" reactors because they will
take less time to restore operation to within validated limits because the UV lamps do not have to
cool down before re-striking. However, installations using LPHO reactors should energize their
"warm" reactors (i.e., the reactors on-line when the voltage sag occurred) instead of standby
LPHO reactors because the UV lamp warm-up time will be less compared to a cold LPHO
reactor.
For long-term power failure (e.g., > 5 minutes) without a UPS system, the UV reactors
should be powered by the backup generator until power is restored. When power is restored, the
shift from the backup generator will likely cause the UV lamps to lose their arc again.
Given the restrictions on operation outside of validated limits (section 1.3.1.3), the utility
should stop water flow through the UV reactors when the lamps are not operating. Also, utilities
should consider installing a UPS if power quality problems are frequent because a standby
generator alone may not adequately alleviate frequent, off-specification flows due to power
quality problems. A UPS system delivers consistent, continuous power even when power
problems occur.
5.7 Staffing Issues
In order to provide consistent and reliable operation of UV reactors, the utility needs to
have appropriate staffing, training, and safety measures in place. This section discusses these
issues.
UV Disinfection Guidance Manual
Proposal Draft
5-37
June 2003
-------
5. Start-up and Operation of UV Installations
5.7.1 Staffing Levels
During start-up operation, a UV reactor will need more operator attention to assist with
functional and performance testing and to establish site-specific O&M procedures (described in
section 5.1.4). However, a typical UV installation needs little operator attention during normal
operation, depending on the level of automation. Generally, UV installations use PLCs to
monitor operating parameters, control the UV reactor, and generate alarms. Increased
automation (e.g., remote monitoring capability) may be incorporated to further reduce operator
requirements. Table 5.14 describes how various site-specific factors affect staffing needs for a
UV installation.
Table 5.14 Factors Impacting Staffing Needs
Factor
Type of UV reactor
Instrumentation and control strategy
Water quality
Impact on Staffing
LP and LPMO reactors may need more maintenance compared
to a MP reactor because they have more lamps and usually
employ OCC cleaning. However, MP lamps will probably need
to be replaced more often than LP lamps.
More automated control strategies will result in lower staffing
levels due to enhanced remote operation and monitoring
capability.
Sleeve fouling and cleaning frequency is affected by water
quality and the design of the UV, reactor. These in turn impact
the staffing needs for manual cleaning for OCC systems and for
maintaining the OMC system.
5.7.2 Training
Training is necessary for all personnel who are associated with the UV installation,
including operators, maintenance workers, instrumentation technicians, electricians, laboratory
staff, custodial staff, engineers, and administrators. The training program should incorporate any
State requirements and should emphasize both normal and emergency operating procedures,
safety issues, process control and alarm conditions, validated operation, and response to
deviations.
The UV manufacturer and UV installation designer should provide training on the UV
reactors, UV installation design, and operation and maintenance activities. It is recommended
that training include both classroom instruction and field training. In addition, actively involving
the operating staff during start-up will provide another opportunity to reinforce classroom
instructions. Continued training should be provided when new employees are hired or when a
process or control alteration is made.
UV Disinfection Guidance Manual
Proposal Draft
5-38
June 2003
-------
5. Start-up and Operation of L?V Installations
5.7.3 Safety Issues
The Office of Safety and Health Administration (OSHA) issues regulations and guidance
to support operator safety in the workplace. There may also be specific safety requirements
imposed by the State. In addition to the standards and procedures established for WTP
operations, the following safety issues pertain specifically to UV reactors:
« UV light exposure
» Electrical safety
i
. Burns from hot lamps or equipment
. Abrasions or cuts from broken lamps
. Potential exposure to mercury from broken lamps - Over-exposure to UV light can
cause eye injury and skin damage.
Threshold Limit Values (TLVs) are issued biannualty by the American Conference of
Governmental Industrial Hygienists (ACGIH). The TLVs for UV radiation apply to
occupational exposure to UV incident on the skin or eye. The recommended TLVs depend on
the lamp wavelengths emitted and the irradiance (mW/cm2); the utility can determine the
appropriate TLV for their UV reactors, using the TLVs for Chemical Substances and Physical
Agents and Biological Exposure Indices (ACGIH 2002). These values are not enforceable
standards but should be considered when establishing operational procedures. To limit or
prevent operator exposure to the UV light, UV reactors should have interlocks that deactivate the
lamps when reactors are accessed. Viewing ports, if provided, should be fitted with UV filtering
windows, or operators should wear a UV resistant face shield when looking at lamps or the
reaction chamber. In addition, warning signs should be placed to minimize the danger of
exposure.
To reduce the risk of electrical shock, the main electrical supply to the UV reactors
should be disconnected and the operator should wait at least 5 minutes for the lamps to cool
down and energy to dissipate before any maintenance is performed. All safety and operation
precautions required by the National Electric Code (NEC), OSHA, local electric codes, and the
UV manufacturer should be followed and include the following precautions:
. Proper grounding
. Lockout, tag-out procedures
. Use of proper electrical insulators
. Installation of safety cut-off switches
The ballasts and the reactor chamber can also become .hot during operation. The
temperature of these components should be assessed prior to contact.
Liy Disinfection Guidance Manual 5-39 June 2003
Proposal Draft
-------
5. Start-up and Operation of UV Installations
Broken lamps pose two potential safety hazards. The lamps and sleeves are constructed
of quartz tubing, which can fracture and cause serious cuts or injury. In addition, broken lamps
may release mercury. Operators should be trained in proper mercury cleanup and disposal
procedures to prevent mercury inhalation or absorption through the skin. Appendix N discusses
potential mercury cleanup procedures.
UV Disinfection Guidance Manual
Proposal Draft
5-40
June 2003
-------
6. References
APHA-AWWA-WEF. 1998. Standard methods for the examination of water and wastewater,
20th ed. Washington D.C.
Aquafine Corporation. 2001. CSL series: Installation, maintenance and operation manual. Part
no.115-1.
Bourgine, P.P., J.I. Chapman, H. Kerai, J.I. Duval, J.G. Green and D. Hamilton. 1995. The
degradation of atrazine and other pesticides by photolysis. The chartered institution of
water and environmental management 9:417-422.
Brandt, C.L., and A.C. Giese. 1956. Photoreversal of nuclear and cytoplasmic effects of short
ultraviolet radiation on paramecium caudatum. J. Gen. Physiol. 39, no. 5: 735-751,
Bukhari, Z., T.M Hargy, J.R. Bolton, B. Dussert, and J.L. Clancy. 1999. Medium-pressure UV
foroocyst inactivation. JournalAWWA 91: 86-94.
Cabaj, A., R. Sommer, W. Pribil, and T. Haider. 2001. What means "dose" in UV-disinfection
with medium pressure lamps. Ozone Science & Engineering 23:239-244.
Carollo Engineers. 2001. Weber basin water treatment plant no. 3 expansion. Layton, Utah.
Chang, J.C.H., S.F. Osoff, D.C. Lobe, M.H. Dorfman, C.M. Dumais, R.G. Quails, and J.D.
Johnson. 1985. UV inactivation of pathogenic and indicator microorganisms. Applied
and Environmental Microbiology 49, no.6: 1361-1365.
Gushing, R.S., E.D. Mackey, J.R. Bolton, and M.I. Stefan. 2001. Impact of common water
treatment chemicals on UV disinfection. Proceedings of the AWWA Annual Conference
and Exposition, June 17-21, Washington, D.C.
Dinkloh, L. 2001. Interview by Ben Hauck. Telephone conversation. Tucson, AZ., August 10.
Downes, A. and T.P. Blunt. 1877. Researches on the effect of light upon bacteria and other
organisms. Proceedings Royal Society London. Volume 26.
Dulbecco, R. 1950. Experiments on photoreactivation of bacteriophages inactivated with
ultraviolet radiation. Journal of Bacteriology 59: 329-347.
Gates, F. L. 1929. A Study of the Bactericidal Action of Ultraviolet Light. J. Gen. Physiol. 13:
231-260.
GE Quartz. 2001. Available from Internet location http://www.gequartz.com/en/lamptube.htm.
Internet.
Gump, D. 2002. Severn Trent. Personal communication with Jennifer Hafer, Malcolm Pimie,
Inc., regarding UV reactors. Tucson AZ, November 6.
UV Disinfection Guidance Manual
Proposal Draft
6-1
June 2003
-------
Chapter 6. References
Jagger, J. 1967. Introduction to research in ultravioletphotobiology. Englewood Cliffs, N.J:
Prentice-Hall, Inc.
Kashinkunti, R,, K.G. Linden, G. Shin, D.H. Metz, M.D. Sobsey, M Moran, and A. Samuelson.
2003. Achieving multi-barrier inactivation in Cincinnati: UV, byproducts, and
biostability. JAWWA. Submitted January 9, 2003.
Kelner, A. 1950. Light-induced recovery of microorganisms from ultraviolet radiation injury,
with special reference to Escherichia coli. Bulletin of the New York Academy of Medicine
26: 189-199. '
Knudson, G.B. 1985, Photoreactivation of UV-irradiated Legionella Pneumoplila and other
Legionella species. Applied and Environmental Microbiology 49, no. 4: 975-980.
Kruithof, J.C., and R.C. van der Leer. 1990. Practical experiences with UV-disinfection in the
Netherlands. Proceedings of the AWWA seminar on emerging technologies in practice,
AWWA Annual Conference, June 17-21, Cincinnati, OH.
Lienberger, J. 2002. Trojan Technologies. Personal communication with Jennifer Hafer, Malcolm Pirnie,
Inc, regarding UV reactors. November 5,
Lin, L.S., C.T. Johnston, and E.R. Blatchley III. 1999a. Inorganic fouling at quartz:water
interfaces in ultraviolet photoreactors I: chemical characterization. Water Research 33,
no. 15: 3321-3329.
Lin, L.S., C.T. Johnston, and E.R. Blatchley. 1999. Inorganic fouling at quartz: water interfaces
in ultraviolet photoreactors - II Temporal And Spatial Distribution. Water Research 33,
no. 15: 3330-3338.
Linden, K.G., G.A. Shin, and M.D. Sobsey. 2001. Comparative effectiveness of UV
wavelengths for the inactivation of Cryptosporidium parvum oocysts in water. Water
Science & Technology 43, no. 12: 171-174.
Linden, K.G., G.A. Shin, G. Faubert, W. Cairns, and M.D. Sobsey. 2002a. UV disinfection of
Giardia lamblia cysts in water. Environmental Science and Technology 36, no. 11: 2519-
2522.
Linden, K.G., L. Batch, and C. Schulz. 2002b. UV Disinfection of filtered water supplies: water
quality impacts on MS2 dose-response curves. Proceedings of the AWWA Annual
Conference, June 16-20, New Orleans, LA.
Linden, K.G. 2002. Personal email communication by Laurel Passantino. April 24-May 6.
Mackey, E.D., R.S. Cushing, and H.B. Wright. 2001. Effect of water quality on UV disinfection
of drinking water. Proceedings of the First International Ultraviolet Association
Congress, June 14-16, Washington, D.C.
UV Disinfection Guidance Manual
Proposal Draft
6-2
June 2003
-------
Chapter 6. References
Mackey, E.D., R.S. Gushing, ML. Janex, N. Picard, J.M. Laine, and J.P. Malley. 2003.
Bridging Pilot-Scale Testing to Full-Scale Design of UVDisinfection Systems, Draft
Report. Final report to be published in 2004. Denver, CO: AWWA Research
Foundation.
Malley, J.P, J.P. Shaw and J.R. Ropp. 1995. Evaluation of by-products produced by treatment of
groundwaters with ultraviolet irradiation. Denver, CO.: AWWA Research Foundation.
Malley, J.P. 2002. Historical perspective of UV use. Presented at the AWWA Water Quality
Technology Conference, November 10-14, Seattle, W.A.
Meulemans, C.C.E. 1986. The basic principles of UV-sterilization of water. In: Ozone +
Ultraviolet Water Treatment, Aquatec Amsterdam, Paris: International Ozone
Association, B.I.1-B.1.13.
Munakata, N., M. Daito, and K. Hieda. 1991. Inactivation spectra of Bacillus Subtilis spores in
extended ultraviolet wavelengths (50-300nm) obtained with Synchrotron radiation.
Photochemistry and Photobiology 54, no 5:761-768.
NWRI. 2000. Ultraviolet Disinfection: Guidelines for Drinking Water and Water Reuse.
National Water Research Institute, Fountain Valley, CA.
Oguma K., H. Katayama, H. Mitani, S. Morita, T. Hirata, and S. Ohgaki. 2001. Determination of
pyrimidine dimers in Escherichia coli and Cryptosporidium parvum during UV light
inactivation, photoreactivation and dark repair. Applied and Environmental
Microbiology 67, no 10: 4630 - 4637.
Passantino, L.B. and J.P. Malley. 2001. Impacts of turbidity and algal content of unfiltered
drinking water supplies on the ultraviolet disinfection process. Proceedings of the 2001
AWWA Annual Conference, Washington D.C. ,
Powell, W. F. 1959. Radiosensitivity as an index of Herpes Simplex virus development. Virology
9:1-19.
Rauth, A.M. 1965. The physical state of viral nucleic acid and the sensitivity of viruses to
ultraviolet light. Biophysical Journal 5: 257-273.
Roberts, Andrew. 2000. Letter to Dan Schmelling of the USEPA, April 10.
Sharpless C.M., and K.G. Linden. 2001. UV photolysis of nitrate: effects of natural organic
matter and dissolved inorganic carbon, and implications for UV water disinfection.
Environmental Science and Technology 35, no 14: 2949-2955.
Shin G.A., K.G. Linden, M.J. Arrowood, G. Faubert, and M.D. Sobsey. 2001. DNA repair of .
UV-irradiated Cryptosporidium parvum oocysts and Giardia lamblia cysts. Proceedings
of the First International Ultraviolet Association Congress, June 14-16, Washington, D.C.
UV Disinfection Guidance Manual 6-3 June 2003
Proposal Draft
-------
Chapter 6. References
Snowball, M.R., and I.S. Hornsey. 1988. Purification of water supplies using ultraviolet light.
Developments in food microbiology, edited by R.K. Robinson, 171-191. NY: Elsevier
Applied Science.
Swaim, P.D.3 M. A. Morine, R.G. Brauer, M.A. Neher, J.L Gebhart, W.D. Bellamy. 2002.
Operating data from Henderson's full-scale UV disinfection facility. Presented at the
water quality technology conference, November 2002.
Toivanen, E. 2000. Experiences with UV disinfection at Helsinki water. IUVA News 2, no. 6: 4-8
U.S. Environmental Protection Agency. 2000. Technical memorandum; Worldwide ultraviolet
disinfection installations for drinking water. Office of Groundwater and Drinking Water,
Washington, DC.
U.S. Environmental Protection Agency. 2000. Technological costs for control ofmicrobial
contaminants and disinfection by-products. Office of Groundwater and Drinking Water,
Washington, DC.
U.S. Environmental Protection Agency. 2002. Microbial Toolbox Guidance Manual. Office of
Groundwater and Drinking Water, Washington, DC.
U.S. Environmental Protection Agency. 2003. Public water system guidance manual for source
water monitoring under the LT2ESWTR. Office of Groundwater and Drinking Water,
Washington, DC.
von Sonntag, C., and H.P. Schuchmann. 1992. UV disinfection of drinking water and by-product
formation—some basic considerations. Journal of Water Supply Research and
Technology 41, no 2: 67-74
UV Disinfection Guidance Manual
Proposal Draft
6-4
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
This appendix supplements Chapter 2, Overview of UV Disinfection, with an additional
level of detail. The purpose of this appendix is to provide technical information regarding the
physical mechanisms of UV light generation, biological reactions causing disinfection, and UV
reactor equipment. The organization of this appendix is presented below, including the key
questions addressed by each section.
« HowisUV light generated? Section A.I.1
• What happens to UV light as it propagates through water? Section A.1.2
. How does UV light inactivate microorganisms? '. Section A.2.2
• Can microorganisms undergo repair and become
infectious after inactivation by UV light? '. Section A.2.3
• How is UV dose determined in a bench-scale (batch)
system? Section A.2.4.1
. How does UV dose vary in a UV reactor? Section A.2.4.2
• How do microbial dose-response curves differ? Section A.2.5
« What factors influence microbial dose-response? Section A.2.6
• Do all microorganisms have the same sensitivity to UV
light? Section A.2.7
• What are the components of a UV installation? Section A.3
. How do low pressure, low-pressure high-output, and
medium pressure lamps differ? Section A.3.1.2-A.3.1.4
. What happens to UV lamps as they age? Section A.3.1.6
. How are UV lamps powered? Section A.3.2
• What is the function of the lamp sleeve? Section A.3.3
• How are lamp sleeves cleaned and why is it necessary
to clean them? Section A.3.4
• How is UV light monitored in a reactor? Section A.3.5
• How are the components of a UV reactor arranged? ...Section A.3.8
UV Disinfection Guidance Manual
Proposal Draft
A-l
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
How do the utility and the State know the UV reactor is
delivering the required UV dose? , Section A.3.9
What are the impacts of water quality on UV
disinfection? Section A.4.1
Do any disinfection byproducts form as a result of UV
disinfection? Section A.4.2
A.1 UV Light Generation and Propagation Through Liquid Media
Using UV light to disinfect drinking water involves generating UV light with the desired
germicidal properties and subsequently delivering that light to the target pathogens. This section
describes fundamental concepts related to the generation and transmission of UV light.
A.1.1 UV Light Generation
Atoms and ions emit light when they change from a higher to a lower energy state. An
atom and most ions consist of electrons orbiting a nucleus of protons and neutrons. The
electrons in each orbital occupy a unique energy state, where the electrons closest to the nucleus
have a lower energy and the electrons further away have a higher energy. When an electron
makes a transition from a higher energy state to a lower energy state, a discrete amount of energy
is released as photons of light at a particular wavelength (A,) according to Equation A.I.
E2-El=~ • Equation A.I
where
EI = Lower energy state (J)
£2 = Higher energy state (J)
h . = Planck's Constant (6.626 xlO'34J»s)
c = Speed of light (2.997 xl08nVs)
X = Wavelength (m)
Energy levels of a given atom or ion are unique and depend on the number of electrons,
protons, and neutrons within that atom or ion and their interaction with external force fields. As
such, each element emits a unique spectrum of light. If the difference between energy levels is
appropriate, the light emitted is in the UV range.
A transition from a tower to a higher energy state requires an energy input. This energy
may be derived from the collision of the atom with a photon of light of wavelength X or by
collision with other atoms, ions, or electrons. Energy transferred to the atom may result in an
increase in the atom's kinetic energy, the transfer of an electron to a higher energy level, or the
removal of an electron from the atom. Removal of an electron from the atom is termed
ionization and results in a positively charged cation and a negatively charged free electron. The
energy required to remove an electron from an atom is termed the ionization energy.
UV Disinfection Guidance Manual A-2 June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
Recombination of a free electron and a cation may result in the emission of light. Since
the free electron and cation may have a range of kinetic energies, the wavelength of emitted light
will vary. The wavelength range will be bound by the ionization energy of the atom, and there
will be a peak within the rage that depends on the temperature of the electrons and cations. The
following sections discuss the relationship between atomic energy states and the generation of
UV light through gas and mercury discharges.
A.1.1.1 Gas Discharges
A gas discharge is a mixture of non-excited atoms, excited atoms, cations, and free
electrons formed when a sufficiently high voltage is applied across a volume of gas. The
wavelength of light emitted from the gas discharge depends on the elemental composition of the
gas discharge and the excitation, ionization, and kinetic energy of those elements.
The formation of the gas discharge within a UV lamp involves several stages. When a
voltage is first applied, free electrons and ions present in the gas are accelerated by the electric
field formed between two electrodes. Initially, the concentration of free electrons and ions arises
from natural radioactivity and is very low. With sufficient voltage, the electrons are accelerated
to high kinetic energies. Collisions of the free electrons with atoms result in a transfer of energy
to the atoms. If the energy transferred is sufficient, the atoms are ionized. This ionization
provides a rapid increase in the number of free electrons and cations, a corresponding increase in
lamp current, and a drop in the voltage across the lamp.
Cations colliding with an electrode cause electrons to be emitted. If sufficient electrons
are emitted, a self-sustaining discharge termed a glow discharge occurs. Initially, only a small
fraction of each electrode emits electrons. With an increase in current, this area increases until
the entire electrode is in use. To increase the current beyond that point, the voltage is increased
to provide more kinetic energy to the cations. High energy cations that collide with the electrode
increase the electrode's temperature. At sufficiently high temperatures, the electrode begins to
thermally emit electrons, and a further increase in current reduces the voltage requirement. At
this point, the electrode discharge is termed an arc discharge.
The start voltage, which is the voltage required to start the gas discharge, is typically
higher than the ionization potential of the gas unless a means is used to introduce electrons.
Preheating the electrode or producing a strong local field using a third electrode located close to
one of the electrodes can be used to introduce electrons and aid in starting the gas discharge.
A gas discharge has a negative impedance that is intrinsically unstable unless a ballast is
placed in series to provide a positive impedance to the power supply. With a direct current (DC)
supply powering the gas discharge, the ballast is a resistor. With an alternating current (AC)
supply, the ballast is either an inductor, capacitor, or some combination of those components.
Inductors and capacitors are preferred over resistors because they do not consume power. More
detail on ballasts is presented in section A.3.2.
The frequency of the AC supply impacts the performance of the gas discharge. If the
frequency of the AC supply is low (« 1 kHz), electron-cation recombination extinguishes the
discharge every half cycle of the lamp voltage. Re-ignition during the next half cycle is '
UV Disinfection Guidance Manual A-3 . June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
facilitated by electron emission from the still warm electrodes. If the frequency of the AC
supply is greater than 1 kHz, the free electrons and cations do not have sufficient time to
recombine and the discharge does not extinguish.
A.1.1.2 Mercury Discharges
i
Mercury in a gas discharge is used to generate the UV light produced in most commercial
UV lamps. Mercury is an advantageous element for UV disinfection due to the following
factors:
• Electron transitions within mercury provide electromagnetic energy in the germicidal
wavelength range.
. Mercury at low vapor pressure and near room temperature produces light at
wavelength 253.7 nm from electrical energy with high efficiency. This wavelength is
near optimal for UV disinfection (section A.2.2).
• Mercury at high vapor pressures produces high intensity polychromatic UV light with
, -reasonably high efficiency.
• Mercury.has a low ionization energy; therefore, free electrons and cations required
for the formation of a gas discharge are easily created using a relatively low start
voltage.
« Mercury reacts minimally with the lamp envelope and electrode materials.
The wavelength and magnitude of light output from a mercury discharge depend on the
concentration of mercury atoms, which is directly related to the mercury pressure. At low
pressures of 0.001 to 0.01 torr (2 x 10~5 to 2 x 10"4 psi), the concentration of mercury is low, and
the distance electrons travel between collisions is relatively long. Electrons achieve higher
kinetic energies with the longer travel distance. Collisions between those free electrons and
mercury atoms excite mercury to the first energy state above the lowest or ground state.
Transition of electrons back to ground state results in the emission of electromagnetic energy at
253.7 and 185 nm. UV lamps with this type of mercury discharge are commonly referred to as
low pressure (LP) lamps.
At higher mercury pressures (100 to 10,000 torr; 2 to 20 psi), a much greater collision
frequency occurs between free electrons and mercury. This increases the energy state of the
mercury atoms and cations to near that of the electrons and increases the temperature of the gas
discharge to near 6,000 °C. When the atoms return to lower energy states, electromagnetic
energy at several wavelengths in the UV light and visible light regions is produced.
Recombination of free electrons and mercury cations produces a small continuum of UV light
between 200 and 245 nm. UV lamps with this type of discharge are called medium pressure
(MP) lamps.
UV Disinfection Guidance Manual A-4 June 2003
Proposal Draft
-------
Appendix A.. Fundamentals of UV Disinfection
A.1.2 UV Light Propagation
This section details the effects that the UV reactor and the water being treated have on the
propagation of UV light.- As UV light propagates, it interacts with the materials it encounters
through absorption, reflection, refraction, and scattering.
A.1.2.1 Absorption
Absorption is the transformation of light to other forms of energy as it passes through a
substance. UV absorbance is the water quality parameter that measures the extent to which the
intensity of UV light is reduced as it passes through water. The impact of absorption on the
intensity light as it travels through a substance is calculated as follows:
= io-fliod=io-
,-ar.rf
Equation A.2
where
I]
A254 =
ae =
Light intensity incident on a cell (mW/cm2)
Light intensity passing through a distance, d, in the cell containing a solution with
various absorbing components (mW/cm2)
Distance traveled by light through the cell (cm)
Molar absorption coefficient of component i (L/mol/cm)
Concentration of component i (mol/L)
Decadic (base 10) absorption coefficient, (cm"1)
Decadic (base 10) absorbance (unitless)
Naperian (base e) absorption coefficient (cm'1).
When UV light is absorbed, it is no longer available to disinfect microorganisms.
A.1.2.2
Refraction
Refraction (Figure A.I) is the change in the direction of light propagation as it passes
from one medium to another.
Figure A.1 Refraction of Light
Medium 1 Medium 2
Refracted Light
Incident Lighi
UV Disinfection Guidance Manual
Proposal Draft
A-5
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
Refraction is governed by Snell's Law,-which is shown in Equation A.3:
«, sin <9, = n, sin 0,
Equation A.3
where
0!
92. =
Index of refraction of the first media
Index of refraction of the second media
Incident angle on the interface
Exit angle from the interface
In UV reactors, refraction occurs when light passes from the lamp through an air gap,
through the lamp sleeve, and into the water. Although refracted light is still available for
disinfection, refraction changes the angle that the light strikes target pathogens.
A.1.2.3
Reflection
Reflection is the change in the direction of light propagation when it is deflected by the
interface between two media (Figure A.2). Reflection may be classified as specular or diffuse.
Specular reflection occurs at smooth polished surfaces where the roughness of the surface is
smaller than the wavelength of light. Reflection from specular surfaces follows the Law of
Reflection, which states that the angle of incidence is equal to the angle of reflection. Diffuse
reflection occurs at rough surfaces. Light is scattered in all directions with little dependence on
the incident angle. The intensity of diffuse reflected light is proportional to the cosine of the
reflectance angle. Reflected light is still available for disinfection.
Figure A.2 Specular and Diffuse Reflection of Light
Incident Light
Reflected Light
Incident Light
Light
Specular Reflection
Diffuse Reflection
In a UV reactor, reflection will take place at UV-transmitting interfaces like an air-quartz
interface and at also interfaces that do not transmit UV light like the reactor wall. The intensity
of reflected light from a UV-transmitting interface is governed by FresnePs Law, which is shown
in Equation A.4.
UV Disinfection Guidance Manual
Proposal Draft
A-6
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
nt cos 9\ - n2 cos 02 ,-\ • | «2 cos #( - «t cos 02
^O, COS0, + «2 COS#2 ^Oj COS02 + M2 COS0,
Equation A.4
where
R
m
n2
Qi
62
the ratio of reflected intensity to incident intensity
Index of refraction of the first media
Index of refraction of the second media
Incident angle onto the interface
Reflected angle from the interface
The intensity of reflected light from a non-transmitting interface depends on the material,
incident angle, wavelength of light, and nature of the surface. .Currently, most UV reactors are
constructed of stainless steel, which reflects 24 percent of UV light at 254 nm at a zero degree
incident angle (Jagger 1967). This indicates that 76 percent of the light energy reaching the
reactor wall is lost. In the future, UV reactors may be developed using materials that reflect
more light, which may improve efficiency.
A.1.2.4
Scattering
Scattering of light is the change in direction of light propagation caused by interaction
with a particle (Figure A3). Scattered light is still capable of disinfecting microorganisms.
Figure A.3 Scattering of Light
Back
Scattered
Light
Incident
Light
Forward
Scattered
Light
90° Scattered Light
Rayleigh scattering is the scattering of light by particles that are smaller than the
wavelength of the light. With Rayleigh scattering, light is scattered uniformly in all directions at
an intensity inversely proportional to the wavelength of light to the fourth power (I A,4). As such,
scattering is more evident at shorter wavelengths. For example, the intensity of scattered light at
200 nm is five times greater than at 300 nm because 1/(2004) is over five times greater than
1/(3004). Particles in water that cause Rayleigh scattering of UV light at 254 nm include small
viruses and large molecules (25 to 300 nm). With larger particles, the scattering observed is non-
uniform, and more light is scattered in the forward direction. The larger particles also cause
backscattering, which is nearly independent of the wavelength of light.
UV Disinfection Guidance Manual
Proposal Draft
A-7
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
A.1.2.5 UV Absorbance and UV Transmittance
UV absorbance (Ais*) is a commonly used water quality parameter that characterizes the
decrease in the amount of incident light as it passes through a water sample over a specified
distance or pathlength. If the measurement is made according to a modified version of Standard
Method 59 1 OB (APHA et al. 1 998) where the water sample is not filtered or pH adjusted, the
modified measurement accounts for scattering and some absorption from particles in the water
sample that may interfere with UV disinfection. Although the Standard Method identifies this
measurement as UV absorption, this manual will refer to it as UV absorbance since the latter
term is widely used in the water treatment industry.
The term UV transmittance (UVT) has also been used extensively in the literature when
describing the behavior of UV light. UVT is the percentage of light passing through a water
sample over a specified distance and is related to UV absorbance by Equation A.5:
% UVT = IQQ*\Q-A™ Equation A.5
where
UVT = UV Transmittance at specified wavelength (e.g., 254 nm) and pathlength
(e.g., 1 cm)
A254 = UV Absorbance at specified wavelength, based on 1 cm pathlength (unitless; UV
absorption as measured by Standard Method 591 OB)
Since UV light scattered from particles is capable of disinfecting microorganisms, it
should be considered when assessing UVT. Much of the scattered light is in the forward
direction and is a significant portion of the transmitted UV light. Typically, conventional
spectrophotometers use narrow beams of light and small detectors that will not measure the
forward scattered light and therefore underestimate the effective UVT of the water sample
(Jagger et al.1975; Linden and Darby 1998). However, spectrophotometers can be equipped
with integrating spheres (Linden and Darby 1998) or detectors capable of measuring forward
scattered light (Jagger et al. 1975) in order to provide a proper assessment of the UVT of water
samples with significant scattering.
A.1.2,6 Estimating UV Light Intensity Within a UV Reactor
The distribution of light intensity about a UV lamp is influenced by the shape of the lamp
and the absorption, refraction, scattering, and reflection of light. Complex models factoring all
of these effects can be used to determine the intensity profile about a lamp, and simplified
models can be used to approximate those profiles. These models are useful tools for
understanding the impact of UV absorbance, UV reactor properties, and UV reactor dimensions
on UV dose delivery and measurements of UV intensity.
If the distance from the lamp is greater than the radius of the arc discharge, the lamp can
be treated as a line source to estimate the intensity. For LP lamps, since the arc discharge fills
the entire lamp, the radius is the same as the lamp radius. For MP lamps, the arc discharge is
much smaller than the radius of the lamp. There are two approaches commonly used for
modeling a line source: the radial model and the point source summation model. If the distance
UV Disinfection Guidance Manual A-8 June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
from the lamp is smaller than the radius of the gas discharge, more complex modeling tools must
be used.
The radial model provides a two-dimensional representation of a three-dimensional
intensity profile. The model assumes light is emitted perpendicular from the line source in the
radial direction as per Equation A.6:
2nr
Equation A.6
where
PL - UV power emitted per unit arc length of the line source (mW/cm)
r = Radial distance from the line source (cm)
ae = Naperian (base e) absorption coefficient of the media (cm"1)
I(r) = UV intensity (mW/cm2) at a distance r from the line source
The Point Source Summation model (Jacob and Dranoff 1970) treats the lamp as a
series of point sources radiating uniformly in all directions. The UV intensity at a point within
the reactor is the sum contribution from each of these points as per Equation A.7.
0-<*,fi
Equation A.7
where
PP
i
Pi
ae
• r
= UV power emitted by each point source (mW)
= Number of point sources used to simulate the lamp
= Distance from the ith point source (cm)
= Naperian (base e) absorption coefficient of the media (cm"1)
Radial distance from the lamp (cm)
= Axial distance along the lamp (cm)
UV intensity (mW/cm2) at a coordinate position (r,z)
The radial and axial distance from the lamp are shown in Figure A.4.
UV Disinfection Guidance Manual
Proposal Draft
A-9
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
Figure A.4 Radial and Axial Distance from a UV Lamp
#> z
(axial distance)
(radial distance)
Note: The point where z = 0 is arbitrary. It can be at the lamp
ends or anywhere along the lamp length. •
For a 25 cm long UV lamp housed in a lamp sleeve (radius = 4 cm) and immersed in
water, Figure A.5 presents the intensity profile predicted using Point Source Summation as a
function of radial and axial distance and the water UV absorbance. For a given radial distance,.
the model predicts a greater UV intensity at an axial position corresponding to the center of the
lamp than at an axial distance corresponding to the lamp ends. The model also demonstrates that
UV intensity will decrease with increased distance from the lamp even in water that does not
absorb UV light (i.e., A254 ~ 0) due to the divergence of UV light from the source. Last, the
model predictions show that the water UV absorbance has a profound impact on the UV intensity
profile about a UV source.
Figure A.5 UV Intensity Profile of a 25 cm Medium-Pressure Mercury Arc Lamp
as a Function of (a) Radial and (b) Axial Position for Waters with
Different UV Absorbance
0 cm-1, UVT = 100 %
0.046 cm-', UVT =90%
5 10 15
Radial Distance from Lamp Center (cm)
20
Y-axis =
UV Intensity
10 cm Radial Distance
From Lamp
mW/cm2
150 T
Lamp is 25 cm long,
centered at x » 0
-20 -10 .0 10 20
Axial Distance along Lamp (cm)
30
More advanced models of the intensity profile about a lamp account for the impacts of
refraction and reflection from reactor components as the light propagates from the discharge
UV Disinfection Guidance Manual
Proposal Draft
A-10
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
(Bolton 2000), the three-dimensional nature of the gas discharge (Irazoqui et al. 1973), and the
direction of light emission (Phillips 1983).
A.2 Microbial Response to UV Light
Disinfection by UV light differs from chemical disinfectants such as chlorine and ozone.
Chemical disinfectants inactivate microorganisms by destroying or damaging cellular structures,
thereby interfering with metabolism, biosynthesis, and growth (Snowball and Hornsey 1988). In
UV disinfection, microorganisms are inactivated by inducing damage to their nucleic acid such
that they can no longer reproduce. This section discusses nucleic acid structure, the damage that
causes microbial inactivation, the ability of microorganisms to repair the damage, methods for
determining microbial inactivation, and factors that affect inactivation.
A.2.1 DNA/RNA Structure
Nucleic acid is a fundamental building block of life and is responsible for reproduction
and defining the nature of life. The nucleic acid is either deoxyribonucleic acid (DNA) or
ribonucleic acid (RNA). The nucleic acid within the nucleus of most cells, including bacteria
and protozoa, is composed of double stranded DNA. DNA contains the information necessary
for the synthesis of ribosomal, transfer, and messenger RNA, which are responsible for
synthesizing enzymes that drive metabolic processes within the cell. The genetic material of
viruses may either be DNA or RNA and can be single or double stranded.
DNA and RNA are long polymers comprised of combinations of four nucleotides. In
DNA, the nucleotides are purines (adenine and guanine) and pyrimidines (thymine and cytosine).
In RNA, the nucleotides are the same except that uracil replaces thymine. Each nucleotide can
be broken down into two parts - a sugar-phosphate backbone and a nitrogenous base (Figure
A.6). If the nucleic acid is double-stranded, nucleotides on one strand will compliment those on
the other strand. Adenine pairs with thymine in DNA (or uracil in RNA) while guanine pairs
with cytosine. Hydrogen bonds form between each pair (Figure A.6).
UV Disinfection Guidance Manual A-11 June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
Figure A.6 Structure of DNA and Nucleotide Sequences Within DNA
Hydrogen Bonded
Nitrogenous
Base Pairs (A, T,G,C)
Sugar-
Phosphate
Backbone
DNA STRUCTURE
-A-T-G- C- G-A-T-C-
1 I I I 1 I I I
—T- A-C- G- C-T-A-G-
Pu rines
A = Adenine
G = Guanine
DNA SEQUENCE
Pyrimidines
T = Thymine
C = Cytosine
A.2.2 Mechanism of Inactivation by UV Light
UV light inactivates microorganisms by damaging DNA or RNA, thereby interfering
with replication of the microorganism. Only light that is absorbed by a system can induce a
chemical reaction (First Law of Photochemistry). As shown in Figure A.7, nucleotides absorb
UV light in from 200 to 300 nm, which enables the photobiological effects that lead to nucleic
acid damage. The UV absorption of nucleic acid is a combination of the absorbance of the
nucleotides and has an absorption peak near 260 nm and a local minimum near 230 nm.
Figure A.7 UV Absorbance of Nucleotides and Nucleic Acid at pH 7
(adapted from Jagger 1967)
o ,« u.o
C (0
I #0.6
11 0.41
n JS
0.0
Cytosine *-»
v '. / ••' '•
Adenine
200 220 240 260 280 300
Wavelength (nm)
^
m o
o>
n
0.6-
0.4-
0.2
0.0
DNA
200 220 240 260 280 300
Wavelength (nm)
UV Disinfection Guidance Manual
Proposal Draft
A-12
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
While both purines and pyrimidines strongly absorb UV light, the rate of UV-induced
damage is greater with pyrimidines (Jagger 1967). Absorbed UV light induces six types of
damage within the pyrimidines of nucleic acid (Setlow 1967; Snowball and Hornsey 1988;
Pfeifer 1997), with varying levels of effectiveness dependent on UV dose:
• Single and double strand breaks are only significant with UV doses several orders
of magnitude higher than those practical for UV disinfection.
• DNA-DNA cross-links are covalent bonds between two different strands of DNA,
and they are also only significant with UV doses orders of magnitude higher than
those practical for UV disinfection.
• Protein-DNA cross-links are covalent bonds between a protein and a DNA strand,
and they may be important for the disinfection of certain microorganisms such as
Micrococcus radioduram.
• Pyrimidine hydrates do not contribute to UV disinfection.
• .Pyrimidine (6-4) pyrimidine photoproducts are a major class of UV damage.
• Pyrimidine dimers are covalent bonds between two pyrimidines on the same DNA
strand, and they are the most common damage resulting from UV disinfection.
While it is possible for thymine-thymine, cyctosine-cytosine, and thymine-cytosine
dimers to form within DNA, thymine-thymine dimers are the most common. However, since
thymine is not present in RNA, uracil-uracil and cytosine-cytosine dimers are formed.
Microorganisms with DNA rich in the thymine tend to be more sensitive to UV disinfection
(Adler 1966).
Dimers cause faults in the transcription of information from DNA to RNA, which in turn
results in disruption of cell metabolism. However, damage to nucleic acid does not prevent the
cell from undergoing metabolism and other cell functions. As discussed in the next section,
enzyme mechanisms within the cell are capable of repairing some of the damage to the. nucleic
acid. To directly damage the internal structure of the cell, UV doses much higher than those
required for inactivation are necessary (Brandt and Giese 1956).
A.2.3 Repair Mechanisms
Because microorganisms that have been exposed to UV light still retain metabolic
functions, some are able to repair the damage done by UV light and regain infectivity. Repair of
UV light-induced DNA damage includes photoreactivation and dark repair (Knudson 1985). At
the doses typically used in UV disinfection, microbial repair can be controlled and accounted for
as discussed in section 3.1.1.
UV Disinfection Guidance Manual A-13 June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
A.2.3.1
Photoreactivation
Photreactivation or photorepair is the cleaving of pyrimidine dimers by the enzyme DNA
photolyase (Setlow 1967). The repair mechanism is termed photorepair because exposure of the
enzyme to light between 310 and 490 nm is needed to activate the enzyme and provide it with
the energy necessary to split the paired dimers.
Figure A.8 shows the difference in UV dose necessary to achieve a certain log
inactivation with and without considering photoreactivation for two organisms. Photorepair
varies with different microorganism types, different species, and different strains of a given
species. The extent of photorepair depends on many factors, including type of microorganism,
degree of inactivation, time between exposure to UV light and photoreactivating light, and the
nutrient state of the microorganism.
Figure A.8 Repair of L. Pneumophila and E Coli (adapted from Knudson 1985)
D L pneumophila - Photorepair
O L pneumophila - No Photorepair
X E. coli - Photorepair
X E. coli - No Photorepair
10 20 30
UV Dose (mJ/cm2)
40
Photoreactivation increased the UV dose necessary to achieve 3-log inactivation of seven
Legionella species between 1.1 and 6.3 fold (Knudson 1985). Photoreactivation also increased
the dose necessary for 4-log inactivation of twelve species of bacteria by 1.2 to 3.5 fold (Hoyer
1998). However, Shin et al. (2001) did not observe photorepair with Cryptosporidium parvum.
Although viral DNA does not have the necessary enzymes for repair, the photorepair of
viral DNA can occur using'the enzymes of their host cells. Lytle (1971) reported that the
photorepair of Herpes simplex virus by mammalian cells varies significantly, depending on the
host cell. RNA viruses lack the ability to photorepair in a host cell (Rauth 1965).
Kelner (1950) reported that the ratio of UV dose necessary to achieve a certain log
inactivation with and without considering photorepair is independent of the degree of
inactivation. However, more recent research by Lindenauer and Darby (1994) reported that the
UV Disinfection Guidance Manual '. A-14 June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
effect of photorepair of coliform bacteria in wastewater becomes less pronounced as UV dose
increases. Knudsen (1985) also found a slight reduction in the ability ofLegionella in
wastewater to repair after higher inactivation levels.
The time between UV light exposure and exposure to photoreactivating light has a
significant effect on the ability to photoreactivate. Dulbecco (1950) reported that the ability of
T2 phage to repair using E. coli as a host organism decreases as the time between exposure to
UV light and photoreactivating light increases. Kelner (1950) reported that £. coli at 37 °C in a
nutrient broth lost the ability to photorepair after 140 minutes in the dark after exposure to UV
light (the same time the survivors took to attempt cell division). In the same study, cells kept at
colder temperatures maintained their ability to photorepair for several hours longer.
The rate of photorepair is constant with time until it reaches saturation, where saturation
is defined as the maximum amount of repair possible by the microorganism given its repair
ability and the extent of damage. Kashimada et al. (1996) reported photorepair saturation of E.
coli occurs after 2 hours of exposure under fluorescent lighting. With exposure to sunlight,
however, they reported photorepair saturation after 15 minutes followed by inactivation that was
attributed to the UV component of sunlight. The rate of repair increases with temperature
(Kelner 1950) but is nearly independent of the reactivating light intensity (Setlow 1967),
suggesting photorepair is rate limited by the enzyme concentration within the microorganism.
The nutrient state of the microorganism also impacts the ability to photorepair. Giese et
al. (1954) reported that a starved strain of paramecium, Colpidium colpoda, needed more
reactivating light to reach saturation than organisms with sufficient nutrients.
A.2.3,2 Dark Repair
Dark repair is when repair processes do not need reactivating light. The term is
somewhat misleading because dark repair can occur in the presence of light and therefore does
not need dark conditions. The forms of dark repair include excision repair, recombinational
repair, and inducible error prone repair. Excision repair, the most common form of dark repair,
is an enzyme-mediated process involving four steps:
1. Repair endonnuclease enzyme recognizes the DNA damage and cleaves the DNA
strand.
2. Exonuclease enzyme excises the damaged section,
3. DNA polymerase rebuilds the removed section using the complementary strand as a
template.
4. Polynucleotide ligase rejoins the severed strand.
One study (Knudsen 1985) examined two different strains of E. coli: one that has the
enzymes necessary for repair (B/R strain) and one that lacks the necessary repair enzymes (recA"
uvr" strain). The difference in UV dose needed for 1-log inactivation of the strain capable of
repair was over two orders of magnitude higher than the dose needed for 1 -log inactivation of the
UV Disinfection Guidance Manual A-15 ''• June 2003
Proposal Draft
-------
1
1200^enhnsy'vania Avenue, NW
Washington DC 20460
202-566-0556
' Appendix A. Fundamentals of UV Disinfection
repair deficient strain, indicating that dark repair impacts the UV dose-response of
microorganisms.
Based on the difference in UV sensitivity of repair proficient and deficient bacteria,
Jagger (1967) discovered that roughly 99 percent of repair that occurs is dark repair. However,
the rate at which dark repair occurs is unknown. It is possible that microorganisms have dark
repaired prior to the microbial assay, and dark repair is not detected. Therefore, the effects of
dark repair can be difficult to measure. Unlike bacteria, viruses do not have the enzymes
necessary for dark repair. However, virus can repair in the host cell using the host cells' enzymes
(Rauth 1965).
A.2.4 UV Dose and Dose Distribution
UV dose is a measurement of the amount of the energy per unit area that is incident on a
surface. UV dose is the product of the average intensity acting on a microorganism from all
directions and the exposure time. Units commonly used for UV dose are J/m2, mJ/cm2, and
mWs/cm2 (10 J/m2 = 1 mJ/cm2 = 1 mWs/cm2) with mJ/cm2 being the most common units in
North America and J/m2 being the most common in Europe. This section discusses how UV
dose is calculated in bench-scale, batch systems and also how the UV dose distribution is
determined in continuous flow pilot- or full-scale UV reactors.
A.2.4.1 Calculation of UV Dose in Bench-Scale, Batch Systems
The most carefully controlled method of determining UV dose is in a batch system with a
bench-scale collimated beam apparatus. Appendix E presents procedures for collimated beam
testing. The factors impacting UV dose calculation in collimated beam tests are described in this
section.
The general definition of UV dose is the product of UV intensity multiplied by the
exposure time.
UV Dose = I-t • Equation A.8
where
I = ' UV intensity (mW/cm2)
t = Exposure time (s) •
If intensity is not constant with respect to time, the integral of the intensity output over
the exposure time should be used in place of intensity as in Equation A.9.
i
UV Dose =J/-£// , Equation A.9
0
where
variables are defined as in Equation A.8.
UV Disinfection Guidance Manual '''.' A-16 June 2003
Proposal Draft .';
-------
Appendix A. Fundamentals of UV Disinfection
Due to several conditions present in collimated beam testing, the intensity measured by
the radiometer does not accurately represent the intensity of light that reaches the target
organisms. To get an accurate calculation of the UV dose delivered to the microorganisms, the
following factors are considered as shown in Equation A. 10 (Bolton and Linden 2003):
absorbance/transmittance of the water, thickness of the water layer, distribution of light across
the surface of the suspension, and reflection and refraction of light from the water.
70/>,(1 -/?)(! MO""""')
W Dose,., = Imf-t = ° f: . — - - - • t Equation A. 1 0
*
where
Iavg = Average intensity within the suspension (mW/cm2)
t = Exposure time (s)
Io = Intensity measured at the suspension's surface (mW/cm2)
R = Fraction of light reflected at the suspension's surface (from FresnePs Law)
aio = . Decadic (base 10) absorption coefficient of the suspension, Ais4 (cm"1)
d = Thickness of water layer (cm)
Pf = Petri factor, ratio of measured intensity at the center of the exposure dish to
average intensity across the surface area of the exposure dish (unitless)
Because microorganisms respond differently to different wavelengths of light, if a
polychromatic light source (e.g., MP lamp) is used, it is also critical to incorporate the light
intensity and the inactivation effectiveness of each wavelength in the germicidai range when
determining UV dose. For microorganisms that exhibit inactivation kinetics that are independent
of wavelength, the equivalent dose at 254 nm from a polychromatic source is calculated as
follows (Meulemans 1986):
300
As* = Z7^)0^) ' ' Equation A. 1 1
A=200
where
0254 = UV dose equivalent at 254 nm
A, • = Wavelength of light (nm)
\(k) - . Intensity at wavelength A. over 1 nm increments
G(k) - Relative action spectrum of the microorganism defined as k*/k254
kx - First order inactivation constant at wavelength X
k2S4 = First order inactivation constant at 254 nm wavelength
t = Exposure time (s)
However, if the microorganism does not exhibit the same inactivation kinetics at each
wavelength, the dose-response curve may be characterized by a shoulder (section A.2.5.2), and
the dose equivalent at 254 nm is calculated using Equation A.12 (Cabaj et al. 2001):
AT 300
— = \-(\-e-"™D™y™ = £l-0-*"*lCl)rf' '•> EquationA.12
•^0 4=200
UV Disinfection Guidance Manual A-17 June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
where
Dx-
kx
dx
Dose delivered at wavelength A.
First order inactivation constant at wavelength A,
Intercept of the exponential region with the y-axis at wavelength A,
A.2.4.2
Dose Distribution in Continuous Flow UV Reactors
Determining the UV dose in a continuous flow pilot- or full-scale UV reactor is
complicated by hydrodynamics (particle paths) and variations in UV intensity throughout the
reactor.
In an ideal reactor that has plug-flow conditions and complete mixing perpendicular to
the flow, all microorganisms entering the reactor will receive the same UV dose, which is
calculated according to Equation A. 13.
Equation A. 13
where
lavg
tr
V
Q -
Volume-averaged UV intensity within the reactor (mW/cm2)
Theoretical residence time of the reactor (s)
Volume of water within the reactor (gal)
Flowrate through the reactor (gal/s)
Equation A. 13 calculates the maximum UV dose possible in an ideal reactor. However in
practice, microorganisms take different paths through a reactor and thus do not all receive the
maximum dose. Instead, the UV dose delivered to the organisms is best described using a dose
distribution (Figure A.9). A dose distribution is a curve or histogram that indicates the
probability of a microorganism receiving a certain dose as it travels through the UV reactor.
Figure A.9. Hypothetical Dose Distribution Delivered by a UV Reactor
0 10 20 30 40 50 60 70 80 90 100
UVDosefmJ/cm2)
UV Disinfection Guidance Manual
Proposal Draft
A-18
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
The width of the dose distribution is an indication of the hydraulic conditions in the
reactor. The more narrow the distribution, the better the hydraulic conditions approximate plug-
flow with complete mixing. However, a narrow dose distribution does not always imply
efficient dose delivery. An annular reactor with a thin water layer between the lamp sleeve and
the reactor wall will deliver a narrow dose distribution. However, if the reactor wall absorbs UV
light, energy losses at the wall will be excessive and the reactor will not efficiently utilize the UV
output of the lamp. The most cost effective design of a UV reactor will have a dose distribution
that reflects a compromise between inefficiency due to energy losses at the reactor wall and by
adjacent lamps as well as inefficiency due to hydrodynamics.
The dose distribution of a UV reactor cannot be measured in a practical manner with
current technology. However, by predicting particle trajectories through the intensity field of a
UV reactor using computational fluid dynamics (CFD), dose distributions can be calculated
(Wright and Lawryshyn 2000).
Inactivation achieved by a reactor with a modeled dose distribution can be calculated by
summing the inactivation achieved by each dose in the dose distribution according to
Equation A. 14.
N
Equation A. 14
"0 i
where
pi(Di) = Probability of dose D; occurring
f(D;) = Mathematical function describing microorganism inactivation as a function of
dose
Using the inactivation kinetics of the microorganism, the inactivation is related to a single
dose value termed the reduction equivalent dose (RED) by Equation A. 15.
Equation A. 15
where
N =
NO =
p(Dj) =
f(Dj) =
Di =
RED =
Concentration of organisms after exposure to UV light
Concentration of organisms before exposure to UV light
Probability of Dj occurring
Mathematical function describing inactivation as a function of dose
UVDose
Reduction equivalent dose
UV Disinfection Guidance Manual
Proposal Draft
A-19
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
If microorganism inactivation can be described using first order kinetics (section
V.2,5.1), inactivation is related to RED by Equation A.16.
N V ( \ U> tKFD
— = 2, F\Pi)e ' = e~ . Equation A.16
•"'o »
where
variables are defined as in Equation A. 15 and
k = First order inactivation coefficient :
By re-arranging Equation A.16, the reduction equivalent dose is calculated according to
:ion A. 17.
Equation A. 17
Equation A. 17.
RED=-\n\
k
where
variables are defined as in Equation A. 15 and A.16
Because UV reactors do not exhibit ideal dose delivery, the RED of a reactor delivering a
dose distribution depends on the UV sensitivity of the microorganisms used to calculate RED.
The RED determined when using a challenge microorganism that is more resistant to UV
disinfection will be higher compared to when using a less resistant microorganism. In contrast,
the RED of an ideal reactor has the same value for all microorganisms. Also, the RED of a
reactor delivering a dose distribution will vary in a non-linear fashion with the lamp power and a
flowrate while the ideal reactor model predicts a proportional relationship. Lastly, the
dependence of RED on UV absorbance of the water will be more pronounced with a reactor
delivering a dose distribution than an ideal reactor. The RED will decrease with increased UV
absorbance at a greater rate with the reactor with a dose distribution than is predicted by ideal
models.
The inactivation of a microorganism and the associated RED are measured using
biodosimetry (described in section 4.2).
A.2.5 Dose-Response Relationships
UV dose-response relationships can be expressed as either the proportion of
microorganisms inactivated (log inactivation, results in dose-response curves with positive
slope) or the proportion of microorganisms remaining (log survival; results in dose-response
curves with negative slope) as a function of UV dose. The proportion of microorganisms
remaining and the log inactivation are typically shown on a logarithmic (base 10) scale, while the
UV dose is typically shown on a linear scale. This manual will present microbial response as log
inactivation since the terminology is widely accepted in the industry. Therefore, all dose-
response curves presented will have a positive slope. The log inactivation of the microorganisms
is determined by measuring the concentration of replicating microorganisms after exposure to a
measured UV dose and is calculated according to Equation A. 18.
UV Disinfection Guidance Manual A-20 June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
Log Inactivation = loglo —-
N
Equation A. 18
where
No
N
Concentration of viable microorganisms before exposure to UV light
Concentration of viable microorganisms after exposure to UV light
Many UV dose-response curves for disperse microorganisms follow first order
inactivation, but in some cases, the dose-response curves take other shapes such as shoulders or
tailing. A shoulder is characterized by a period of very little inactivation at lower doses followed
by linear or exponential inactivation. A dose-response curve that exhibits tailing is characterized
by a decrease in the inactivation rate after a certain degree of inactivation has been observed.
Figure A. 10 shows various shapes of dose-response curves. Note that the terms "shoulder" and
"tailing" refer to the shape the curves take when the y-axis of the dose-response curve is
presented as log survival with negative slopes, which is the inverse of log inactivation.
Figure A.10. UV Dose-Response Curves (adapted from Chang et ai. 1985)
OE.
D B. subtilis spores
v Total coliform-wastewater
X Rotavirus
40 60 80
UV Dose (mJ/cm2)
100
A.2.5.1
First Order Response
The E. coli data shown in Figure A. 10 exhibit first order dose-response behavior. The
equation for first order inactivation is shown in Equation A. 19:
UV Disinfection Guidance Manual
Proposal Draft
A-21
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
Equation A. 19
where
NO = Concentration of viable microorganisms before UV exposure
N = Concentration of viable microorganisms after UV exposure
k ~ First order inactivation coefficient of the microorganisms (cm2/mJ)
D = UV dose delivered to the microorganisms (mJ/cm )
DIO = UV dose needed to inactivate microorganisms by one log (i.e., 90 percent
inactivation) (mJ/cm2)
In first-order response, only one photon of light is needed to inactivate a microorganism.
A.2.5.2 Shoulders
The B, subtilis data shown in Figure A. 10 exhibit a shoulder followed by first order dose-
response behavior. The shoulder is attributed to a delayed response of a microorganism when
exposed to UV light. Unlike first order inactivation, more than one photon of light is needed to
inactivate a microorganism. Although the number of photons can not be measured directly, it
can be related to first order response through curve fits of empirical equations. Equation A.20
(Cabaj et al. 2001) is one of the many equations derived from empirical curve fits that can be
used to model inactivation curves with shoulders.
N/N = 1 - (l - e-*0)" Equation A.20
where
NO = Concentration of viable microorganisms before UV exposure
N = • Concentration of viable microorganisms after UV exposure
k = First order inactivation coefficient of the microorganisms (cm2/mJ)
D = UV dose delivered to the microorganisms (mJ/cm2)
d - 'Intercept of the exponential region of the dose-response with the y-axis
Morton and Haynes (1969) reported a decrease in the shoulder with nutrient-depleted E.
coli and proposed that the shoulder was associated with dark repair. Photoreactivation
significantly increased the shoulder observed with E. coli (Hoyer 1998) and Legionella
(Knudson 1985). •
Note that the equation presented is only one of the many equations derived from
empirical curve fits. There are many methods to model UV dose-response data not presented
here that may better describe specific UV dose-response data (Severin et al. 1984).
UV Disinfection Guidance Manual A-22 June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
4.2.5.3 Tailing
If the irradiated microorganisms are a mixture of disperse microorganisms and clumped
or particle-associated microorganisms, the UV dose-response will demonstrate tailing, or a
flattening of the curve at higher UV doses (Parker and Darby 1995). With wastewaters, tailing
begins after 2 to 3 log of disperse microorganism inactivation, with diminishing inactivation
occurring beyond that level despite increasing UV dose (Figure A. 10, Total coliforms). Dose-
response with tailing can be modeled using Equation A.2 1 .
N = JV*" + Npe~k>° Equation A.21
where
N = Concentration of viable microorganisms after UV exposure
No = Concentration of disperse microorganisms before UV exposure
k = First order inactivation coefficient of the microorganisms (cm2/mJ)
D = UV dose delivered to the microorganisms (mJ/cm2)
Np = Concentration of particles containing the microorganisms
kp = Pseudo first order inactivation constant of particle-associated microorganisms
(cm2/mJ)
A.2.6 Factors Impacting Microbial Response
Several factors impact the response of microorganisms to UV light. This section
discusses these factors, including UV intensity, UV absorbance, temperature, pH, particles, and
UV wavelength.
A.2.6.1 UV Intensity
Oliver and Cosgrove (1975) reported that UV dose-response of microorganisms follows
the Law of Reciprocity over an intensity range of 1 to 200 mW/cm2. For example, the
inactivation effectiveness observed with UV intensity of 1 mW/cm2 and an exposure time of 200
seconds is equivalent to the inactivation observed with an exposure time of 1 second and UV
intensity of 200 mW/cm2 as well as all intensity-time combinations between 1 and 200.
Exceptions to this reciprocal relationship between time and intensity occur at very low
and high intensities (Setlow 1967). With low UV intensities and long exposure times, repair
may compete with inactivation. Sommer et al. (1998) found less inactivation of E. coli at a given
dose with low intensities ranging from 0.002 to 0.2 mW/cm2. However, inactivation of B.
subtilis spores, MS2 bacteriophage (MS2), <|>xl74 phage, and B40-8 phage at a given dose was
the same regardless of UV intensity. At UV intensities on the order of 1010 to 1011 mW/cm2
(several orders of magnitude higher than the intensity from lamps used for UV disinfection),
Gurzadyan et al. (1981) reported an increase in single strand breaks and a reduction in
dimerization in the nucleic acid of <|>xl74 phage.
UV Disinfection Guidance Manual A-23 . June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
A.2.6.2
UVAbsorbance
Because the calculation of dose delivered to a microbial suspension (Equation A. 10,
section A.2.4.1) accounts for UV absorbance in bench-scale batch experiments, measured UV
dose-response curves like those presented in Figure A. 10 are independent of the suspension's UV
absorbance. However, as the UV absorbance of the suspension increases, the UV intensity
incident on the sample may need to increase in order to deliver a given dose. In bench-scale,
batch experiments, there are several ways to keep the UV dose constant such as increasing
exposure time or decreasing the depth of the solution, thereby decreasing the pathlength.
4.2.6.3
Temperature
, Temperature affects the configuration of nucleic acid and the activity of repair enzymes;
however, existing research shows temperature effects on UV dose-response are minimal and
depend on the microorganism. Severin et al. (1983) found the UV dose needed for a given log
reduction of E. coli, Candida parapsilosis, and f2 phage increases slightly as temperature
decreased (Table A.I). Maltey (2000) reported the dose-response of MS2 is independent of
temperature from 1 to 23°C (Figure A.I 1).
Table A.1 Impact of Temperature on UV Disinfection
Microorganism
£ coli
C. parapsilosis
f2 phage
UV dose (mJ/cm2) needed to achieve 2 log inactivation at a temperature of
5°C
11.8
30.9
72.5
20 °C
11.2
28.8
65.6
35 °C
10.7
28.0
60.7
Figure A.11 Impact of Temperature on MS2 UV Dose-Response (Malley 2000)
3 •
1 •
10 20 30 40 50 60 70 80 90 100
Delivered UV Dose (mJ/cm2)
UV Disinfection Guidance Manual
Proposal Draft
A-24
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
4.2.6.4
UV dose-response is usually independent of the pH of the water. The UV absorbance of
nucleic acid and repair enzyme activity .vary with pH (Jagger 1967). However, the pH within a
cell is buffered to a relatively constant value, independent of water pH. For example, Malley
(2000) reported the dose-response of MS2 is independent of the suspension pH from pH 6 to 9
(Figure A. 12).
Figure A.12 Impact of pH on MS2 UV Dose-Response (Malley 2000)
i
1
s,.
10 20 30 40 50 60 70 60
Delivered UV Dose (mJ/cm2)
go 100
A.2.6.5
Particle Association
To date, research examining the effects of naturally occurring particles and
microorganisms is limited to wastewater studies. Due to the limited concentration of
microorganisms in drinking water sources, methods of directly examining the impact of particles
do not currently exist. However, the phenomena observed in wastewater studies may also apply
to particle association occurring in drinking water. The effects of individual particles (such as
those that cause turbidity) on UV disinfection are discussed in section A.4.1.2.
Results from research with wastewaters have indicated that clumping or particle
association will shield microorganisms from UV light. The UVT at 260 nm through 10 microns
of cell tissue is roughly 10 percent (Jagger 1967), suggesting that clumps of organisms would
offer protection. The water content of cells and intracellular material and the porous nature of
flocculated particles will influence the penetration of light into waterborne particles. Quails et al.
(1983) reported that filtration of secondary effluent through an 8 micron filter removes the
particles responsible for the tailing in the dose-response of coliforms. With 8 micron filtration,
coliform inactivation at 12 mJ/cm increased from 3 log to over 4.5 log inactivation. Loge et al.
(1999) reported the UV absorbance of wastewater solids varied from 0.33 to 56.9 nm"1 (3,300 to
569,000 cm"1) with the high absorbance associated with activated sludge plant using iron to
remove phosphorus. Petri et al. (2000) reported coagulation of MS2 by iron in ground water
increased the UV dose needed to inactivate MS2 by a factor of 2.5 to 3.5.
UV Disinfection Guidance Manual
Proposal Draft
A-25
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
4.2.6.6 Wavelength
Microbial dose-response varies with the wavelength of UV light. The action spectrum
(also called UV action) of a microorganism is a measure of inactivation as a function of the
wavelength for a given UV dose. The dependence of UV action on wavelength is similar to the
dependence of the UV absorbance of DNA on wavelength (Figure A. 13). UV action peaks near
260 nm, has a local minimum near 230 nm, and drops to zero near 300 nm. While it is generally
believed that microorganisms are most sensitive near 260 nm, there are exceptions. For
example, the UV sensitivities of tobacco mosaic virus (Hollaender and Duggar 1936), reovirus
(Rauth 1965), and Herpes simplex virus (Powell 1959) are greater below 230 nm. Although the
UV action increases below 230 nm for most microorganisms, the UV absorbance of natural
waters at these wavelengths make this region impractical for UV disinfection of microorganisms
in water. Because of the similarity between UV action and DNA absorbance, and because DNA
absorbance is easier to measure than UV action, the DNA absorbance spectrum of a
microorganism is often used as a surrogate for its UV action spectrum.
Figure A.13 Comparison of Microbial Action to DNA UV Absorbance
(adapted from Rauth 1965 and Linden et at. 2001)
DMA
- -o - Crypfosporicfr'um
0--MS2
~* -Herpes simplex virus
.2.01
1.5-
1.0-
0.5-
0.0 H—i 1—i—i' " i i—i 1—r
200 220 240 260 280 300
Wavelength (nm)
A plot of the first order inactivation coefficient as a function of wavelength can be used
to show the action spectrum if the dose-response follows first order inactivation. Plots of two
kinetic terms as a function of wavelength are necessary to plot the action spectrum if the dose-
response has a shoulder (Cabaj et al. 2001) as discussed in section A.2.5.2. Plots of UV action
spectra are often presented relative to some wavelength, typically 254 nm.
UV Disinfection Guidance Manual
Proposal Draft
A-26
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
A.2.7 UV Dose-Response of Differing Microorganisms
The UV dose-responses of microorganisms have been tabulated in a number of review
articles and are summarized in this section.
Data presented in Tables A.2 and A.3 show that the UV sensitivity of microorganisms
varies with different species. Of the pathogens of interest in drinking water, viruses are most
resistant to UV disinfection followed by bacteria, and Cryptosporidium oocysts and Giardia
cysts. The most UV resistant viruses of concern in drinking water are adenovirus Type 40 and
41. Appendix B provides dose-response data for Giardia cysts, Cryptosporidium oocysts, and
viruses, and Chapter 1 (Table 1.4) contains the regulatory UV dose requirements for inactivating
these pathogens.
Table A.2 provides average dose reported without photoreactivation for incremental log
inactivation of various pathogenic bacteria, virus, and protozoa of concern in drinking water.
Table A.3 provides similar information for various non-pathogenic indicator bacteria, spore
forming bacteria, and bacteriophage. All data in Tables A.2 and A.3 are for microorganisms
suspended in water and irradiated using a collimated beam apparatus with UV light at 254 nm.
Spore-forming and gram-positive bacteria are more resistant to UV light than gram
negative bacteria (Jagger 1967). With microorganisms larger than 1 micron, the absorption of
UV light by the cytoplasm can be significant, depending on the wavelength, and therefore can
affect UV sensitivity.
Rauth (19.65) found that the UV sensitivity of virus and bacteriophage varies over two
orders of magnitude from the most sensitive to the most resistant. The same study showed
viruses with single-stranded nucleic acid are ten times more sensitive than viruses with double-
stranded nucleic acid.
UV Disinfection Guidance Manual
Proposal Draft
A-27
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
Table A.2 UV Sensitivity of Pathogenic Microorganisms in Water1
Microorganism
Aeromonas hydrophila
Campylobacterjejuni
Escherichia colt 01 57:H7
Legionella pneumophila
Salmonella anatum
Salmonella enteritidis
Salmonella typhi
Salmonella typhimurium
Shigella dysenteriae
Shigella sonnei
Sfaprty/ococcus aureus
Vibrio cholerae
Yersinia enterocolitica
Adenovirus Type 40 2
Adenovirus Type 41 2
Coxsackievirus B5
Hepatitis A HM1 75
Hepatitis A
Hepatitis A HM1 75
Poliovirus Type 1
Poliovirus Type 1
Poliovirus Type 1
Poliovirus Type 1
Rotavirus SA1 1
Rotavirus SA1 1
Rotavirus SA11
Cryptosporidium parvum z
Cryptosporidium parvum 2
Giardia lamblia 2
Giardia lamblia 2
Type
Bacteria
Bacteria
Bacteria
Bacteria
Bacteria
Bacteria
Bacteria
Bacteria
Bacteria
Bacteria
Bacteria
Bacteria
Bacteria
Virus
Virus
Virus
Virus
Virus
Virus
Virus
Virus
Virus
Virus
Virus
Virus
Virus
Protozoa
Protozoa
Protozoa
Protozoa
UV Dose (mJ/cm2) inactivation
indicated
1-log
1.1
1.6
1.5
3.1
7.5
5
1.8
2
0.5
3.2
3.9
0.8
1.7
30
22
6.9
5.1
5.5
4.1
4.0
6
5.6
5.7
7.6
7.1
9.1
<2
<1
<1
2-log
2.6
3.4
2.8
5
12
7
4.8
3.5
1.2
4.9
5.4
1.4
2.8
59
50
14
14
9.8
8.2
8.7
14
11
11
15
15
19
<3
<3
<3
3-log
3.9
4
4.1
6.9
15.
9
6.4
'5
2
6.5
6.5
2.2
3.7
90
80
21
22
15
12
14
23
16
18
23
25
26
<5
<6
<6
4-log
5
4.6
5.6
9.4
10
8.2
9
3
8.2
10.4
2.9
4.6
120
30
21
16.
21
30
22
13
36
<2
Reference
Wilson etal. 1992
Wilson etal. 1992
Wilson etal. 1992
Wilson et al. 1992
Tosa and Hirata 1998
Tosa and Hirata 1998
Wilson etal. 1992
Tosa and Hirata 1998
Wilson etal. 1992
Chang etal. 1985
Chang etal. 1985
Wilson eta 1. 1992
Wilson etal. 1992
Meng and Gerba 1996
Meng and Gerba 1996
Battigelli etal. 1993
Wilson etal. 1992
Wiedenmann et al. 1993
Battigelli etal. 1993
Meng and Gerba 1996
Harris etal. 1987
Chang etal. 1985
Wilson etal. 1992
Battigelli et at. 1993
Chang etal. 1985
Wilson etal. 1992
Shin et al. 2001
Clancy et al. 2000
Linden et al. 2002a
Mofidi et al. 2002
Adapted from Wright and Sakamoto 1999
2 Additional data for adenovirus, Cryptosporidium, and Giardia are in Appendix B.
UV Disinfection Guidance Manual
Proposal Draft
A-28
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
Table A.3 UV Sensitivity of Non-Pathogenic Bacteria, Bacteriophage, and
Spore-Forming Bacteria in Water1
Microorganism
Escherichia coti
Escherichia coli
Escherichia coli
Escherichia coli
Streptococcus faecatis
Streptococcus faecalis
MS-2
MS-2
MS-2
MS-2
MS-2
MS-2
MS-2
<|>X174
.)>X174
<|»X174
PRD-1
B-40
Bacillus subtilis spores
Bacillus subtilis spores
Type
Bacteria
Bacteria .
Bacteria
Bacteria
Bacteria
Bacteria
Phage
Phage
Phage
Phage
Phage
Phage
Phage
Phage
Phage
Phage
Phage
Phage
Spores
Spores
UV Dose (mJ/cm")
inactivation indicated
1-log
2.5
3
4.0
4.4
6.6
5,5
4
16
12
21
17
14
19
2.2
2.1
4
9.9
12
36
29
2-log
3
4.8
5.3
6.2
8.8
6.5
16
34
30
36
34
29
40
5.3
4.2
8
17
18
49
40
3-log
3.5
6.7
6.4
7.3
9.9
8
38
52
45
61
7.3
6.4
12
24
23
61
51
4-log
5 '
8.4
7.3
.8.1
11
9
68
71
62
.10
8:5
30
28
78
Reference
Harris etal. 1987
Chang etal. 1985
Sommer et al. 1998
Wilson etal. 1992
Chang etal. 1985
Harris etal. 1987
Wiedenmann et al. 1993
Wilson etal. 1992
Tree etal. 1997
Sommer etal. 1998
Rauth 1965
Meng and Gerba 1 996
Oppenheimer et al. 1993
Sommer etal. 1998
Battigelli et al. 1993 '
Oppenheimer et al. 1993
Meng and Gerba 1996
Sommer etal. 1998 .
Chang etal. 1985
Sommer etal. 1998
1 Adapted from Wright and Sakamoto 1999.
A.3 UV Reactors
This section discusses UV reactor components, UV reactor configurations, and how
reactor performance is monitored. The following UV reactor components are discussed:
. Mercury lamps
• ' Ballasts and power supplies
• Lamp sleeves
• Cleaning systems
. UV intensity sensors
• UV transmittance monitors
. Temperature sensors
UV'Disinfection Guidance Manual A-29 June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
A.3.1 Mercury Lamps
This section describes mercury lamps, including how they are constructed, their
components, efficiency, spectral output, and aging. A majority of the material in this section was
derived from Sources and Applications of Ultraviolet Radiation by Roger Phillips (1983).
Section 2.4.2 (Table 2.1) compares the operating characteristics of LP, LPHO, and MP mercury
lamps.
A.3.1.1
Lamp Construction
Mercury vapor discharge lamps consist of a UV-transmitting envelope made from a tube
of quartz sealed at both ends (Figure A.14). An electrode is located at each end of the envelope
connected to the outside through a seal. The envelope is filled with mercury and an inert gas.
Figure A.14 Construction of a UV Lamp
LOW-PRESSURE MERCURY LAMP - HOT CATHODE TYPE
TungstenCoil Electrode Enve|ope
/
*
: r . f c
Mercury & Inert Gas Fill
LOW-PRESSURE HIGH-OUTPUT
Tungsten Coil Electrode
^£^3 *
-------
Appendix A. Fundamentals of UV Disinfection
Lamp Envelope
The envelope of the lamp should transmit germicidal UV light, act as an electrical
insulator, and not react with the lamp's fill gases. A non-crystalline form of quartz, vitreous
silica, is often used for the lamp envelope because of its high UVT and its resistance to high
temperatures. However, some LP lamps use UV-transmitting glass instead of quartz. Envelopes
are approximately 1 to 2 mm thick, and the diameter is selected to optimize the UV output and
lamp life.
As quartz is exposed to high temperatures, it begins to crystallize. Crystallization
reduces the UVT of the quartz and changes its coefficient of expansion, which causes internal
stresses. Envelopes for MP lamps must be able to withstand thermal shocks associated with 600
to 900 °C operating temperatures without the quartz transforming to its crystalline form.' LP
lamps have tower operating temperatures where crystallization is not a concern, which is why
some LP lamps use UV-transmitting glass rather than quartz for the lamp envelope.
Envelopes of MP lamps may be covered with a reflective coating at the ends. This is to
keep the ends warm and prevent the condensation of mercury behind the electrodes;
The UV transmittance of the envelope affects the spectral output of MP mercury lamps,
especially at lower wavelengths. Lamp envelopes can be made from doped quartz, or quartz that
is altered to absorb specific wavelengths, in order to prevent undesirable non-germicidal
photochemical reactions. If the lamp envelope is not made from doped quartz, the lamp sleeves
can also be used to restrict the wavelengths emitted (described in section A.3.3).
Electrodes " "
With a LP mercury lamp, electrode design depends on whether the lamp operates with a
glow or arc discharge. With a glow discharge, free electrons are formed from the bombardment
of the electrode by cations. The electrode used is typically a cylinder of iron or nickel. Lamps
of this type of electrode operate near 150 °C and are termed cold-cathode lamps. With an arc
discharge, free electrons are emitted thermally from a hot electrode operating near 900 °C and
are referred to as hot-cathode lamps. The electrode is made of a coil of tungsten wire embedded
with oxides of calcium, barium, or strontium. The high melting point of tungsten prevents
evaporation of electrode materials that could coat the inside of the lamp and reduce output of UV
light. The oxides embedded within the tungsten coil reduce the temperature needed for the
emission of electrons. LP and LPHO lamps used in UV disinfection are usually hot-cathode
lamps.
In order to reduce the start voltage of a hot-cathode lamp, each electrode may have two
electrical connections to pass current through the electrode. Resistive heating of the electrode
raises the electrode's temperature, thereby facilitating rapid transition from a glow discharge to
an arc discharge at a lower voltage. Rapid transition from a glow to an arc discharge reduces
electrode sputtering and improves lamp life. The process of transitioning from glow to arc
discharges and how it produces UV light is described in section A. 1.1.1.
Electrode design and operation is critical for reliable long term operation. In order for
lamps to operate at an optimal temperature, electrode design should promote heat transfer. The
UV Disinfection Guidance Manual A-31 . June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
electrodes of a MP mercury lamp consist of a tungsten rod wrapped in a coil of tungsten wire.
To improve thermal emission of electrons, thorium or alkaline-earth oxides are embedded within
the coils, and the tungsten rod may contain thorium oxide. The electrode must warm-up within
seconds to allow transition from a glow to an arc discharge and minimize sputtering of tungsten
onto the envelope. The electrode operating temperature must be high enough to promote thermal
emission of electrons and low enough to prevent the evaporation of tungsten.
The electrode of a MP lamp is connected to the external electrical supply via a thin
molybdenum foil sealed in the quartz at the lamp ends. The molybdenum foil is ductile and
therefore does not crack the quartz when the lamp expands and contracts due to changes, in
operating temperature. If the temperature of the seal increases beyond 350 °C, the molybdenum
will oxidize and the seal will fail. Because the seal is located behind the electrode, its
temperature is lower than the temperature of the arc.
Mercury Fill
The mercury fill present in UV lamps can be in the solid, liquid, or vapor phase. At
typical LP and LPHO lamp operating temperatures, mercury remains predominantly in the liquid
or solid amalgam phase with a small proportion in the vapor phase (which is responsible for
producing UV light). An amalgam is an alloy of elemental mercury with another metal
(typically indium or gallium in lamp applications) that can be either solid or liquid at room
temperature, depending on the relative proportions of the two metals. Amalgams are typically
used in LP and LPHO lamps, while MP lamps contain liquid elemental mercury.
Vapor pressure (the pressure of mercury in the vapor phase) depends on the temperature.
LP lamps operate with an envelope temperature near 40 °C, resulting in a mercury vapor
pressure near 0.007 torr (1.4 x 10"4 psi), which is optimal for the production of UV light at 254
nm. MP lamps operate at a much higher envelope temperature (600 to 900 °C), resulting in a
mercury vapor pressure ranging from 100 to 10,000 torr (2 to 200 psi). In MP lamps, the
concentration of mercury in the vapor phase is controlled by the amount of mercury in the lamp,
as opposed to LP and LPHO lamps where an excess of mercury is placed in the lamp and only a
portion of the elemental mercury enters the vapor phase.
With a conventional LP lamp, increasing the operating current will not produce a higher
UV output. Instead the operating temperature will increase causing an increase in vapor pressure
and the UV light output of the lamp will decrease. LPHO lamps hold the mercury vapor pressure
constant at the optimal value, allowing the UV light output to increase as current increases until a
saturation value is reached. Methods of controlling the vapor pressure within the lamp include
using either a mercury amalgam attached to the lamp envelope, a cold spot on the lamp wall, or a
mercury condensation chamber located behind each electrode. With each method, the
temperature of the mercury within the lamp, and hence the vapor pressure, is controlled,
allowing efficient production of UV light at higher currents.
Inert Gas Fill
In addition to mercury, lamps are filled with an inert gas (typically argon) at 1 to 50 torr
(0.02 to 1 psi). The inert gas aids in starting the gas discharge and reduces deterioration of the
electrode. When the lamp is started at room temperature, the concentration of mercury atoms is
UV Disinfection Guidance Manual A-32 June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
low and there are few collisions between free electrons and mercury. However, there are a
significant number of collisions between free electrons and argon atoms. These collisions excite
the argon atoms to a metastable higher energy state that does not return quickly to a ground state.
Collisions between the excited metastable argon and mercury or free electrons ionizes the
mercury and argon, respectively. The free electrons released by ionization reduce the start
voltage and aid in the formation of the gas discharge. However, if lamps are manufactured with
a non-ideal argon pressure (>50 torr; 1 p" si), the collisions between electrons and argon cause
energy losses, and therefore the electrons never achieve sufficient kinetic energy to excite the
mercury atoms.
A.3.1.2 Low-Pressure Lamp Efficiency
LP lamps are designed and manufactured to efficiently convert electrical energy to
germicidal UV light. An optimal LP lamp design typically includes the following components:
• 3.6 cm lamp envelope diameter
• 0.007 torr mercury fi II (1.3 x 10"4 psi)
. 3 torr argon fill (0.06 psi)
• 400 mA operating current
. 40 °C operating temperature
» 0.5 W/cm power input per arc length
Under such conditions, the power input efficiency is as follows:
« 60 percent converted to UV light at 185 and 254 nm
• 3 percent converted to other wavelengths
• 15 percent to electrode losses
. 22 percent to thermal losses from the arc
A.3.1.3 Low-Pressure High Output Lamp Efficiency
Theoretically, LPHO lamps have the same efficiency as LP lamps because they operate at
similar vapor pressures. However in practice, LPHO lamp efficiency can be significantly lower,
depending on lamp construction, ballast operation, power settings, and lamp cooling. The
energy input to a LPHO lamp can be converted to energy in the following forms:
• 30 to 45 percent converted to UV light at and 254 nm
UV Disinfection Guidance Manual A-33 June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
5 to 25 percent converted to light at other wavelengths, primarily 185, 313, 365, and
436 nm
50 to 65 percent to thermal losses from the arc
A.3.1.4 Medium-Pressure Lamp Efficiency
For the purposes of UV disinfection, the efficiency of a MP lamp can be defined as the
ratio of its germicidal output to its electrical input. Equation A.22 defines germicidal efficiency
as a function of power input, lamp output, and the action spectra of a given microorganism.
300nm
£b_ = *;2oo™ - Equation A.22
where
r) = Germicidal efficiency of the lamp
PC = Germicidal lamp output (W)
PE = Electrical power input (W)
A = Wavelength (nm)
P(A,) = Lamp output measured over 1 nm increments at wavelength A. (W)
G(A.) = Action spectra of the microorganism at wavelength A, (unitless)
Figure A. 15 presents the output versus electrical input between 248 and 320 nm for three
MP lamps containing different mercury doses. For the lamps considered, lamp efficiency varied
slightly with input power to the lamp but did not vary with mercury dose. Lamp efficiency on
average was 10 percent. Because lamp data used to generate Figure A. 15 were based only on
lamp output from 248 to 320 nm, the lamp efficiency may be underestimated.
UV Disinfection Guidance Manual A-34 June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
Figure A.15 Germicidal Output from 248 to 320 nm of Three MP
Lamps Calculated for MS2 as a Function of Electrical Power Input
(adapted from Phillips 1983)
1C
Germicidal Output (W/cm)
» u. 3 t
i i
X 4.8 mg/cm Hg dose
D 8.0 mg/cm Hg dose
A 10.1 mg/cm Hg dose
XA
X
. Q
A
x*1
0 50 100 150
Electrical Power Consumed (W/cm)
A.3.1.5 Spectral Output of Lamps
LP lamps emit primarily resonant light at 253.7 nm (Figure A.16a) that is formed from
electron transitions from the first excited states to the ground state of mercury. They also emit
light at 185 nm with intensity varying from 12 to 34 percent of the UV intensity at 253.7 nm
depending on the operating current, wall temperature, and inert gas fill. UV light at 185 nm will
react with oxygen and promote the formation of ozone within the lamp sleeve. Ozone is a
corrosive and toxic compound that absorbs UV light. As such, LP lamps for UV disinfection
applications are manufactured to reduce or eliminate the emission of UV light at 185 nm. Other
wavelengths of light including 313, 365,405,436, and 546 nm also are emitted from LP lamps at
low intensities due to higher energy electron transitions within the mercury.
The spectral output of LPHO lamps is similar to LP lamps. Although all of the
wavelengths emitted are identical, the intensity of light from LPHO lamps is higher.
The spectral output of MP mercury lamps involves peaks overlying a continuum (Figure
A.16b). The combination of free electrons and mercury cations within the arc creates a broad
continuum of UV energy lines between 200 and 245 nm. This continuum does not occur with
LP lamps, where non-radiating recombination occurs at the envelope walls. Electron transitions
within the mercury produce numerous narrow peaks of electromagnetic energy in the visible and
ultraviolet range. These transitions result in a broadening of the emitted light and a shift in its
peak, usually to longer wavelengths. For example, the peak from 260 to 270 nm arises primarily
due to the 254 nm electron transition.
UV Disinfection Guidance Manual
Proposal Draft
A-35
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
Figure A.16. UV Output of LP (a) and MP (b)
Mercury Lamps (Sharpless and Linden 2001)
V
8 w
JJ1.0-
'•** <-
JZ — U.O
(Sf
tj S'O.e
t°
o |°-4
°- P
I*0-2'
°s
2 f\ « ,
a. Low Pressure Lamp
200 250 300 350 400
Wavelength (nm)
5 |10.
§22
!S
— e n o
Lamp Output ReU
i/la xi mum Output i
J O O O C
j KJ » b> c
b. Medium Pressure Lamp
UV
Continuum
^iuJLUI/X.
,]1 II
— -
Jl
;••-
200 250 300 350 400
Wavelength (nm)
A.3.1.6 Lamp Aging
Over time, UV lamps can degrade, resulting in a reduction in output where lower
germicidal wavelengths degrade faster than higher wavelengths. Lamp output will decrease over
time as a function of lamp hours in operation, number of on/off cycles and power applied per
unit (lamp) length.. The rate of decrease in lamp output slows as the lamp ages (Figure A. 17).
UV Disinfection Guidance Manual
Proposal Draft
A-36
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
Figure A.17. Reduction in UV Output of LP and MP Lamps Over Time
(adapted from Schenck 1981)
100-
80-
60-
40
20
0
a. Low-Pressure Mercury Lamps
5000
Time (Mrs)
10000
b. Medium-Pressure Mercury Lamps
500
Time (Hrs)
1000
Lamp aging can be affected by the following factors:
Ballast operations, including power setting, frequency, and harmonic distortion of the
voltage and current driving the lamp
. Water temperature and heat transfer from lamps
. Vibration of the lamp sleeves caused by water flowing through the reactor
• The frequency of on-off cycles
With LP and MP lamps, sputtering of the electrode during the glow phase of start-up can
coat the inside surface of the lamp envelope with tungsten. The tungsten coating is black in
color, non-uniform, concentrated within a few inches of the electrode, and can absorb UV light
(Figure A. 18). Sputtering from the electrode can be reduced by the following conditions:
• Pre-heating the electrode before applying the start voltage
• Driving the lamp with a sinusoidal current waveform
• Using a lamp with a higher argon content
• Minimizing the number of lamp starts during operation
UV Disinfection Guidance Manual
Proposal Draft
A-37
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
Figure A.18. Aged UV Lamp (right) in Comparison to a New UV Lamp (left)
(Mackey et al. 2003)
If a MP lamp is not sufficiently cooled during operation, tungsten and oxides between the
tungsten coils may evaporate and coat the inside of the envelope. LP lamps using
UV-transmitting glass may have mercury combine with sodium in the glass to create a UV
absorbing coating. Any deposits on the inner or outer surfaces of the lamp envelope and by
metallic impurities within the envelope will absorb UV light. The absorption of UV light can
raise the temperature, which may lead to localized overheating of the lamp envelope. If the lamp
envelope is quartz, the increase in temperature can lead to devitrification (crystallization),
contributing to an additional decrease in UVT.
With MP lamps, reaction of the electrode with any water molecules that have entered the
lamp envelope as a result of lamp seal failure will form an oxide and hydrogen and also increase
the start voltage. The molybdenum seal of a MP lamp will oxidize and fail if the seal
temperature exceeds 350 °C. High operating temperatures of a MP lamp can also lead to bubbles
and distortion of the lamp envelope materials and devitrification (crystallization), which leads to
a decrease in UVT. The coefficient of expansion of crystalline quartz is higher than that of non-
crystalline quartz, and rapid changes in temperature will also stress the envelope, which may lead
to lamp breakage.
A.3.2 Lamp Power Supply and Ballasts
UV lamps are typically operated with an AC supply. Unlike an incandescent lamp, a
mercury vapor lamp cannot be connected directly to the electrical service because it has a non-
linear voltage to ampere characteristic (Persson and Kuusisto 1998). In order for the mercury
vapor lamp to function properly, a ballast must be inserted into the circuit to limit the current
flow through the lamp. When placed in series with the lamp, the ballast provides an impedance
to the power supply with a positive voltage-current characteristic. The power supply and ballast
are designed to provide the following features:
. Reliable and rapid starting of the gas discharge
UV Disinfection Guidance Manual
Proposal Draft
A-38
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
• Re-ignition of the gas discharge every half cycle of the power supply
. An appropriate current waveform
. A high power factor
. Stable light output
Resistors, capacitors, inductors, or combinations of these can be used as ballasts;
however, resistors are not used because they consume power and therefore reduce electrical
efficiency. Lamp ballasts are often termed either magnetic (also known as electromagnetic) or
electronic. Magnetic ballasts can be inductive or capacitive and operate at the line frequency.
Electronic ballasts operate at frequencies higher than that of the line voltage and involve solid
state devices or a mixture of solid state devices, inductors, and capacitors.
A.3,2.1 Magnetic Ballasts
There are two types of magnetic ballasts: capacitive (those with capacitors) and inductive
(those with inductors). Each ballast type is designed to control the current to the lamp.
With a capacitive ballast, the current through the lamp is primarily a function of the
capacitance used and does not vary significantly with the applied voltage or the lamp properties.
An advantage of the capacitive ballast is that the power delivered to the lamp and the lamp
output are independent of line voltage. A disadvantage is that electrode sputtering can increase,
which accelerates electrode aging. Capacitive ballasts are more efficient than inductive ballasts,
but less efficient than electronic ballasts. Because of the stored energy in the capacitor and the
coil, capacitive ballasts are less prone to failure as a result of small fluctuations in power quality.
With the inductive ballast, the current through the lamp is a function of the inductance,
the applied voltage, and the lamp properties. As electrical current flows through the inductor, it
generates a magnetic field. The magnetic field opposes the electrical current, and the strength of
the field is proportional to the current passing through the inductor. Therefore, as the current
increases, so does the resistance to the current. This interaction limits the total current flow to
the lamps to a specific amperage. The highest power achieved with the inductive ballast is lower
than with the capacitive ballast. However, electrode sputtering is less than with capacitive
ballasts, leading to extended electrode life. With capacitive ballasts, the UV lamp output varies
with the line voltage. Inductive ballasts provide more stable current output and better resolution
and control than capacitive ballasts, but are generally less efficient, larger, heavier, and more
expensive.
Magnetic ballasts are currently the most common type of ballast used for medium
pressure lamps due to their durability and proven operating stability in the higher power
applications. Medium pressure reactors typically incorporate some form of power adjustment to
optimize energy efficiency and control dose delivery. Because of the manner in which magnetic
ballasts operate, power can only be adjusted by incorporating capacitors or inductors into the
circuit. Adjustment occurs in a series of steps, and the number of steps is limited by the number
of capacitors or inductors that are included in the ballast.
UV Disinfection Guidance Manual A-39 • June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
A.3.2.2
Electronic Ballasts
Electronic ballasts, sometimes referred to as solid state ballasts, contain semiconductors
and other electronic components such as low-pass filters, rectifiers, buffer capacitors, and high
frequency oscillators that allow the ballast to behave like a switching power supply. A chopped
electrical current with up to 50,000 pulses per second of electricity is supplied to the lamp,
whereas a magnetic ballast typically produces only 100 to 120 pulses per second. With an
electronic ballast, the frequency of electrical pulses supplied to the lamp is longer when the lamp
is cold. As the lamp approaches its optimum operating temperature, the electronic ballast
provides shorter and less frequent pulses of current to the lamp.
Electronic ballasts are a newer technology than magnetic ballasts and are therefore less
proven. Although they have limited operational experience, electronic ballasts offer increased
efficiency, smaller size and weight, and the opportunity for nearly continuous power adjustment
over a wide range of settings. Reliability has improved significantly since electronic ballasts
were initially developed. Currently, manufacturers of low pressure reactors and smaller medium
pressure reactors often use electronic ballasts in their design. Because of the reduction in stored
energy, electronic ballasts are more susceptible to failure due to power inconsistencies; however,
by incorporating a buffer capacitor, minor power disturbances can be smoothed out, reducing the
occurrence of lamp failure.
A.3.2.3 Comparison of Ballast Types
Electronic and magnetic ballasts each have specific advantages and disadvantages.
Manufacturers considerthese advantages and disadvantages when determining the technology to
incorporate into their equipment designs. The final selection takes into account the relative
importance of each of the advantages and disadvantages for a given application. A single
manufacturer may have equipment designs based on both ballast types. For example, one UV
manufacturer uses electronic ballasts for its smaller units and magnetic ballasts for its larger
units. A summary of some of the advantages and disadvantages of each ballast technology is
shown in Table A.4.
UV Disinfection Guidance Manual
Proposal Draft
A-40
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
Table A.4. Comparison of Magnetic and Electronic Ballasts.
Magnetic Ballast
Electronic Ballast
Comparative
Advantages
Less potential for power
interference due to stored energy
More resistant to power surges
More resistant to high
temperatures.
Less prone to interference with
electronic devices
Less prone to sputtering (inductive
less than capacitive)
Proven technology (in use for
nearly 70-years)
Less expensive
More efficient
Lighter weight
Smaller size
Less potential for heat generation
Less potential for noise
Continuous power adjustment
Longer lamp operating life
Comparative
Disadvantages
•' Less efficient (capacitive more
efficient than inductive)
. Heavier weight
• Larger size
» More potential for heat generation
» More potential for noise.
. Step-function power adjustment
(number of steps proportional to
number of inductors/capacitors)
. Shorter lamp operating life
More potential for power
interference due to stored energy
(can be minimized by incorporating
a capacitor)
Less resistant to power surges
Less resistant to high temperatures
More prone to interference with
electronic devices
More potential for sputtering
Newer technology (limited operating
experience, especially in larger
sizes)
More expensive
A.3.2.4 Lamp Startup
The voltage applied to the lamps must be sufficiently high to start and operate the lamps.
Step-up transformers are needed to increase the voltage above the mains to start cold-cathode
lamps. Hot-cathode lamps are classified as either instant or switch start. Instant-start lamps have
a single connection with each electrode. Starting instant-start lamps needs the application of a
high voltage. As the electrodes warm-up, the needed voltage drops. Switch-start lamps have
two electrical connections with each electrode, and the electrodes are preheated for 1 to 2
seconds before the start voltage is applied. This reduces the start voltage and lengthens the lamp
life. Because of their relatively high impedance, MP lamps typically need a higher voltage than
LP lamps for starting and stable operation. Operating voltage ranges from 5 to 30 volts/cm,
depending on arc length, mercury dose, lamp diameter, and electrode losses. With the exception
of short lamps, step-up transformers are needed to operate MP lamps and high voltage pulses are
used to start them.
UV Disinfection Guidance Manual
Proposal Draft
A-41
, June 2003
-------
Appendix A. Fundamentals of UV Disinfection
A.3.2.5 Voltage Frequency Converters
With LP and LPHO lamps, frequency converters may be used to increase the voltage
frequency from that of the mains (typically 60 Hz) to 20 to 100 kHz. Typically, the efficiency of
UV light output from electrical power increases by as much as 10 percent as the frequency
increases above 500 Hz, Furthermore, the higher frequency reduces electrode deterioration,
makes the lamps easier to start, and extends the lamp life. These benefits, however, can be offset
by power losses associated with the frequency converter.
A.3.3 Lamp Sleeves
In UV reactors, lamps are housed within lamp sleeves. Sleeve length is sufficient to
include the lamp and associated electrical connections. Sleeve diameter is typically 1 inch (2.5
cm) for LP mercury lamps and 2 to 4 inches (5 to 10 cm) for MP lamps. Sleeve walls are
typically 2 to 3 mm thick and absorb some UV light. Sleeves made of doped quartz are used to
prevent the transmission of low-wavelength UV light, thereby reducing undesirable
photochemical reactions.
Lamp sleeves have several functions other than housing the lamps. They maintain the
lamp temperature at an optimal value and control heat transfer from the lamps. Heat transfer
from the MP lamp prevents failure of the molybdenum seal, distortion of the lamp envelope, and
evaporation of the tungsten electrode. Also, lamp sleeves isolate the lamp and its electrical
connections from the water. Lastly, they protect the lamp from mechanical forces such as water
hammer and protect the lamp from thermal shock arising from differences in water and lamp
envelope temperature.
Typically, LP lamps are centered using Teflon® rings, and MP lamps are centered using
ceramic or metal disks. The positioning of the lamp along the length of the sleeve can influence
dose delivery by the reactor.
Sealing the lamp sleeve assembly prevents water condensation within the sleeve and
contains any ozone formed between the lamp envelope and lamp sleeve. Components within the
sleeve should withstand exposure to UV light, ozone, and high temperatures. If the components
are not made of the appropriate material, exposure can cause component deterioration and off-
gassing of any impurities present in the quartz from manufacturing. Off-gassed materials can
form UV-absorbing deposits on the inner surfaces of the lamp sleeve. Off-gassing and ozone
formation will be a greater issue with MP lamps because they operate at a higher temperature
and emit low-wavelength ozone-forming UV light. Off-gassing can be minimized through
proper manufacturing of the lamp sleeves.
The UVT of a lamp sleeve influences the intensity of UV light transmitted from the lamp
into the water. The UVT is a function of the reflectance and absorbance of UV light by the
sleeve, as per Equation A.23.
UV Disinfection Guidance Manual A-42 June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
UVTS (A) = [1 - RM (A)] [I - Rsw
(-a(i)L)
Equation A.23
where
Rsw(X-)
a(
L
Sleeve UVT at wavelength A.
Reflectance of UV light at the air-sleeve interface at wavelength X •
Reflectance of UV light at the sleeve-water interface at wavelength X
Sleeve absorption (Base e) at wavelength A,
Pathlength of light through the sleeve
Because the refractive indices of the lamp sleeve and water are similar, the reflectance of
UV light at the sleeve-water interface (Rsw) is often considered negligible in this equation. The
absorption coefficient of the sleeve varies strongly with wavelength and the material of the
sleeve. For a zero degree incidence angle, Figure A. 19 presents the UVT over the germicidal
range of two types of quartz: standard and wavelength-selective. Quartz can be manufactured to
select for a variety of wavelengths depending on the desired application. For UV disinfection
applications, wavelength-selective quartz is primarily used to prevent the transmission of low
wavelength (<200 nm) UV light into the water.
Figure A.19. UV Transmittance of Two Types of Quartz Commonly Used to Make
Lamp Sleeves (GE Quartz 2001)
Wavelength-Selective
Quartz
200
220 240 260 280
Wavelength (nm)
300
320
In order to reduce fouling on the sleeve surfaces, some UV reactors using LP lamps have
sleeves made of Teflon® or Teflon®-coated quartz. However, Teflon® sleeves have a lower UV
transmittance, and their transmittance degrades faster than conventional quartz.
Failure mechanisms for sleeves include fractures and fouling. Fractures arise from
internal stresses created during the production of the quartz and external mechanical forces.
Annealing the quartz at high temperatures during production removes internal stresses. Visual
inspection using polarized light can also reveal whether or not sleeves are stress free. Fractures
UV Disinfection Guidance Manual
Proposal Draft
A-43
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
may arise from mechanical forces such as wiper jams, water hammer, resonant vibration, and
impact by foreign objects. Fouling may occur on both internal and external surfaces and is
discussed in more detail in section A.4.1.4. Exposure of quartz contaminated with metal cations
from the manufacturing process can cause solarization and an increase in UV absorption.
A.3.4 Cleaning Systems
Due to fouling on the lamp sleeves, cleaning the external surface of sleeves is important
to maintain dose delivery. UV reactor manufacturers have developed different approaches for
cleaning lamp sleeves, depending on the application. Both manual and automatic cleaning
regimes are used. A reactor must be shut down and drained prior to manual cleaning. The
sleeves are removed once the reactor is drained and wiped with a cloth and cleaning solution.
Manual cleaning is primarily used for smaller systems with relatively few sleeves and lower
fouling potential.
Automatic cleaning approaches are typically used for larger systems. They may be
classified as off-line chemical cleaning (OCC) or on-line mechanical cleaning (OMC). OCC
systems, also referred to as flush and rinse systems, involve a sequence of events controlled by
the UV reactor. In OCC systems, the reactor is shut down, drained, and flushed with a cleaning
solution. Solutions used to clean sleeves include citric acid, phosphoric acid, or a food grade
proprietary solution provided by the UV reactor manufacturer. The reactor is rinsed and returned
to operation after sufficient time to dissolve the substances fouling the sleeves is allowed. OCC
cleaning approaches are typically used by reactors with LPHO lamps.
In OMC systems, the UV reactor remains on-line while the lamp sleeves are cleaned.
OMC systems have mechanical or physical-chemical wipers that are built-in to the UV reactor.
The wipers are either driven by screws attached to electric motors or pneumatic pistons.
Mechanical wipers may consist of steel brush collars or Teflon® rings that move along the lamp
sleeve. Physical-chemical wipers have a collar filled with cleaning solution that move along the
lamp sleeve. The wiper physically removes fouling on the lamp sleeve surface while the
cleaning solution within the collar dissolves fouling materials. UV reactors with MP lamps
typically use wipers because the higher lamp temperatures accelerate fouling under certain water
qualities.
The time between sleeve cleaning will depend on the rate of fouling. Sleeve cleaning can
be initiated manually, at regular intervals, or triggered by a calculated UV dose or measured UV
intensity, depending on the reactor control logic. In physical-chemical wipers, solution
replacement varies with the rate of fouling and is on the order of months. Replacing the cleaning
solution is necessary because reaction with the foulant and dilution with water reduces the ability
of the cleaning solution to dissolve the foulant.
A.3.5 UV Intensity Sensors
UV intensity sensors are photosensitive detectors that are used to indicate dose delivery
by providing information related to UV intensity at different points in the UV reactor. UV
intensity sensors include the following components arranged as shown in Figure A.20.
UV Disinfection Guidance Manual A-44 June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
Monitoring windows and light pipes deliver light to the photodetector. Monitoring
windows are typically quartz discs and light pipes are cylindrical probes made of
quartz (quartz silica probe).
Diffusers and apertures reduce the UV light incident on the photodetector, thereby
reducing UV intensity sensor degradation. Diffusers also modify the UV intensity
sensor's angular response.
Filters limit the light delivered to the diode, often restricting it to germicidal
wavelengths.
Photodetectors are solid-state devices that produce a current proportional to the
irradiance on the detector's active surface. The responsivity of typical photodetector
to UV light is on the order of 0.1 to 0.4 milliamps/mW (mA/mW)-
Amplifiers convert the output of the photodetector from a low-level current to a
standardized output proportional to the irradiance (e.g., converts intensity to a 4 to 20
mA output for use in process control interfaces).
The housing of the UV intensity sensor protects the components from the external
environment. The housing should be electrically grounded to shield the photodetector
and amplifier, thereby reducing electrical noise and bias.
UV Disinfection Guidance Manual
Proposal Draft
A-45
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
Figure A.20. Interior UV Intensity Sensor Schematics
(courtesy of (a) Severn Trent Services and (b) Wedeco-ldeal Horizons)
a.
Signal Amplifier
Printed Circuit Boards
Photodetector
UVC Filtei
Housing
Quartz Silica Prol
, , Signal
Output
UVC Light
^ Housing
Photodetectof
Filter
Quartz
Window
UV Light
Note that the sensor shown in Figure A.20b is cylindrical in shape. All dimensions are
standardized as detailed in the German standards for UV disinfection.
A.3.5.1 UV Intensity Sensor Properties
UV intensity sensor properties that impact the measurement of UV intensity and dose
delivery monitoring include angular response, acceptance angle, spectral response, working
range, detection limit and resolution, linearity, temperature response, long term drift, calibration
factor, and measurement uncertainty. An ideal UV intensity sensor will have a linear response
over the working range, provide a response unaffected by ambient temperature, be stable over
time, have zero measurement noise and bias, respond only to germicidal UV light, and have zero
measurement uncertainty.
Angular response is a plot of the sensor measurement as a function of the incident angle
of UV light at the sensor's window. Angular response is affected by the UV intensity sensor's
aperture size, the size of the photodetector's active surface, the distance between the aperture and
the active surface, and the impact of any diffusers and reflecting surfaces within the UV intensity
sensor. An ideal sensor has a cosine response (Equation A.24) because a cosine response results
in an accurate'measure of the light incident on the surface of the photodetector.
= /,cos(0)
Equation A.24
where
Im
Ii
9
Intensity measured by photodetector
Intensity incident on photodetector's surface
Incident angle at the photodetector surface
UV Disinfection Guidance Manual
Proposal Draft •
A-46
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
In practice, sensors deviate from cosine response; some potential responses are shown in
Figure A.2L
Figure A.21. Angular Response of Two UV Intensity Sensors
Relative to Ideal Cosine Response
8"
k.
« IS
3 "
~~~ Cosine
- - • Sensor 1
Sensor 2
-90 -60 -30
30 60 90
Incident Angle (degrees)
The opening or acceptance angle of the UV intensity sensor is the angle over which the
sensor detects UV light. The opening angle is typically measured by either the threshold
detection of UV light or detection at some percentage of the maximum detection value (e.g., 50
percent). The acceptance angle is a characteristic of the sensor but does not affect sensor
performance.
The spectral response is a measure of the output of the UV intensity sensor as a function
of wavelength. The sensor spectra! response depends on the response of the photodetector and
filters and the UVT of the monitoring windows, light pipes, and filters. Some sensors use filters
to limit the spectral response to the wavelengths within the germicidal range (200 to 300 nm)
because it can be advantageous for sensors to only respond to UV light that causes damage to
microorganisms.
The working range of the UV intensity sensor is the range that the sensor is able to
measure. The low end of the working range is defined by the detection limit of the
measurement. The high end of the measurement range is limited by the saturation of the
photodetector and the amplifier. Saturation is the point at which the sensor can no longer
respond to an increase in intensity.
The detection limit of the UV intensity sensor is the lowest UV intensity that can be
detected and quantified at a known confidence level. The detection limit is calculated as a
confidence of repeated measurements of low intensity UV light, usually at a specific percentage
confidence interval. The measurement resolution is the smallest difference in UV intensity that
can be differentiated at a given confidence limit. The detection limit and the resolution depend
on the measurement noise and on any digitalization of the analog output from the UV intensity
sensor by the system's electronics. The measurement noise is the root-mean-square (RMS) of the
random variation in the sensor measurement over time. The measurement bias is the time-
averaged sensor measurement obtained with no incident light. The measurement bias and noise
UV Disinfection Guidance Manual
Proposal Draft
A-47
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
of a photodetector are increased by electromagnetic fields within the UV reactor if the sensor is
not properly shielded and grounded.
An ideal UV intensity sensor responds proportionally to the intensity incident on the
sensor (Figure A.22). The linearity of the UV intensity sensor is a measure of the deviation of
the sensor response from that proportional relationship. Linearity is affected by bias and
saturation. The linearity is reported as the ratio of the measured response to the known incident
intensity, usually at a specific percentage confidence interval.
Figure A.22 Example of Sensor Linearity
Region of
Linear Response
ideal
Response
Region of
Saturation
1234567
Incident Intensity (mW/ctn1)
UV intensity sensor measurement is also affected by ambient temperature. The changes
in sensor response arise from thermal expansion of the optical components, the photodetector,
and the amplifier. UV intensity sensor electronics can compensate to reduce the effects of
temperature.
The long-term drift of the UV intensity sensor is the change in response as a function of
time. Exposure to UV light damages optical and electronic components within the UV intensity
sensor. The damage caused by UV light is typically greater at higher UV intensities and lower
wavelengths. Degradation of the filter can increase the filter's bandwidth (the wavelength range
passed by the filter), thereby increasing the UV intensity sensor measurement even though the
UV lamp output has not increased. Degradation of the monitoring windows and light pipes may
cause a decrease in UVT due to solarization. Off-gassing from damaged components can coat
optical components, reducing the measured intensity.
The calibration factor of the UV intensity sensor is a value used to convert the standard
electrical output of the UV intensity sensor (mA or volts) to UV, intensity (mW/cm2 or W/m2).,
The calibration factor is the ratio of the known intensity of the UV light to the electrical output of
the sensor. Sensors used in UV reactors equipped with LP or LPHO lamps are calibrated with
UV Disinfection Guidance Manual A-48 June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
UV light at 254 nm. Sensors used in UV reactors equipped with MP lamps can either be
calibrated with light only at 254 nm or can be calibrated with polychromatic UV light from a MP
lamp. . .
The uncertainty of a UV intensity sensor represents the difference in intensity between
that measured by the sensor and an accepted reference sensor. This uncertainty incorporates the
uncertainty that arises due to variability in calibration, linearity, spectral response, angular
response, temperature response, and long-term drift.
A.3.6 UV Transmittance Monitors
As stated previously, UVT is an important parameter in determining the efficiency of UV
disinfection. Therefore, monitoring UVT (or UV absorbance, A254> to calculate UVT) is critical
to the success of a UV disinfection application. UVT can be determined either by grab samples
with a laboratory instrument or by an on-line instrument. Several commercial UV reactors use
the measurement of UVT to help monitor and control the calculated UV dose in the reactor.
In general, commercial on-line UVT monitors calculate transmittance by measuring UV
intensity at various distances from a lamp. One such monitor is schematically displayed in
Figure A.23. In this monitor, a stream of water passes through a cavity containing a short LP
lamp with three UV intensity sensors located at various distances from the lamp. The difference
in sensor readings is used to calculate UVT.
Figure A.23. UV Transmittance Monitor Design
(courtesy of Severn Trent Services)
UV Intensity
Sensor
Inlet
Outlet
UV Disinfection Guidance Manual
Proposal Draft
A-49
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
A.3.7 Temperature Sensors
Energy input per unit volume is relatively high for a UV reactor. The water flowing
through a reactor efficiently absorbs the waste heat and maintains operating temperatures within
a desirable range. Nevertheless, temperatures can become elevated under the following
circumstances:
Water level in the reactor drops and lamps are exposed to air.
i
. Water stops flowing in the reactor.
Most temperature sensors are located at the top of the UV reactor. The temperature
sensor can either measure the water temperature or the reactor shell temperature. In either case,
if the temperature exceeds a setpoint value, it will register a high-temperature alarm. The
temperature alarms can be integrated into a supervisory control and data acquisition (SCADA)
system such that the alarm results in an operations change to reduce the potential for lamp
breakage. For instance, the reactor can be shut down or valves can open or close to change the
flow of water to the reactor.
A.3.8 Reactor Configuration
This section describes the configuration of UV lamps and UV intensity sensors as well as
the hydraulic considerations of the overall reactor design. .
A.3.8.1
Lamp Placement
The lamp configuration in a reactor is designed to optimize dose delivery. UV lamps
may be oriented parallel, perpendicular, or diagonal to flow. Depending on the installation of the
reactor, this can result in lamps oriented horizontally, vertically, or diagonally relative to the
ground. Orienting MP lamps horizontal relative to the ground prevents overheating at the top of
the lamps and reduces the potential for lamp breakage due to temperature differentials.
In a reactor with a square-cross section, typically lamps are placed with lamp arrays
perpendicular to flow. This pattern may be staggered to improve disinfection efficiency. With a
circular cross-section, lamps typically are evenly spaced on one or more concentric circles
parallel to flow- The water layer between lamps and between the lamps and the reactor wall
influences dose delivery. If the water layer is too small, the reactor wall and adjacent lamps will
absorb UV light. If the water layer is too large, water will pass through regions of lower UV
intensity and experience a lower UV dose. The optimal spacing between lamps depends on the
UVT of the water, the output of the lamp, and the degree of hydraulic mixing within the reactor.
A.3.8.2 UV Intensity Sensor Placement
UV intensity sensors may be located to view either one or more lamps. The measurement
of UV intensity from a given lamp depends on the following conditions:
UV Disinfection Guidance Manual A-50 June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
• Output of the lamp
. UVT of the water
. Distance from the lamp to the UV intensity sensor
. Incident angle of the light on the UV intensity sensor
As such, a given measurement by a UV intensity sensor viewing more than one lamp can
have many interpretations, and such measurements should be properly understood to avoid
misinterpretation. Also, UV intensity sensors may be located to view the output from the center
or ends of the lamp. The optimal sensor placement will give a representative or conservative
measure of the lamp output, given that lamp aging and sleeve fouling is non-uniform along the
length of the lamp.
The number of UV intensity sensors used in a reactor can vary from one per lamp to one
per reactor. The appropriate number of sensors will depend on the type of lamp used, the
variance in lamp-to-lamp output (especially after the lamps have aged), and the impact of that
variance on dose delivery and dose monitoring. The implications of the number of sensors used
per reactor are discussed in the background to the validation protocol (section F.3.5)
UV intensity sensors may view the lamps either from a UV intensity sensor port located
on the reactor wall or through a lamp sleeve located within the reactor. UV intensity sensors are
classified as wet or dry. A dry UV intensity sensor views the UV light through a monitoring
window as shown in Figure A.20b. A wet UV intensity sensor is in direct contact with the water
flowing through the reactor and is shown in Figure A.20a. While checking the on-line UV
intensity sensor with a reference UV intensity sensor is easier with a separate monitoring
window, condensation on the window can interfere with the measurement of UV intensity.
A.3.8.3 Hydraulic Considerations
The flow through UV reactors is turbulent with residence times on the order of tenths of a
second for MP lamps or seconds for LP lamps. In theory, optimal dose delivery by a UV reactor
is obtained with plug flow hydrodynamics and complete mixing perpendicular to the flow. In
practice, however, UV reactors do not have these ideal hydrodynamics.
Lamp placement, inlet and outlet conditions, baffles, and mixers all affect hydrodynamics
within a reactor. Turbulence,and eddies form in the wake behind lamp sleeves oriented
perpendicular to flow. Staggered lamp arrays promote mixing within the reactor, thereby
minimizing short-circuiting of flow.
Inlet and outlet conditions can have a significant impact on reactor hydrodynamics.
Ninety-degree inlet and outlets promote short-circuiting, eddies, and dead zones within the
reactor. Straight inlet conditions with gradual changes in cross sectional area can be used to
develop flow for optimal dose delivery.
UV Disinfection Guidance Manual A-51 June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
Some manufacturers insert baffles to improve hydrodynamics in the reactor. Perforated
plates can be used to even the flow throughout the reactor's cross-section. Plates with a single
opening are used to direct flow towards high intensity regions within the reactor. Mixers used
within reactors are designed to promote either turbulent or vortex mixing.
Improvements to the hydrodynamics through the reactor are often obtained at the expense
of headloss. Perforated baffle plates and turbulent mixers can increase dose delivery but will
significantly increase headloss. However, inlet and outlet conditions surrounding the reactor can
be changed to reduce headloss without changing the disinfection effectiveness within the reactor.
Also, using vortex mixers allows the spacing between lamps to increase, thereby reducing
headloss through the reactor. • '••
A.3.9 Monitoring UV Disinfection Performance
Some method of monitoring the performance of an operating UV installation is required
to demonstrate to the utility and primacy agency that adequate disinfection is being achieved (40
CFR 141.729(d)). Because the concentration of pathogenic organisms cannot be measured
continuously in the UV-treated water and the dose cannot be measured directly in real time,
various strategies have been developed to demonstrate adequate dose delivery. Any dose
monitoring method selected must be evaluated during reactor validation (described in Appendix
C) and the outputs measured during validation will be part of the monitoring setpoints.
Currently, there are three fundamental approaches to monitor UV disinfection
performance in a UV reactor:
1. UV Intensity Setpoint Approach. In this approach, measurements made by the UV
intensity sensor are used to control the UV reactor. The UV intensity sensor is
located in a position that allows it to properly respond to both changes in UV
intensity output of the lamps and also UVT of the water. The UV intensity sensor
output and the flowrate are used to monitor dose delivery. The setpoint value for UV
intensity over a range of flowrates is determined during validation (see Chapter 4).
2. UV Intensity and UVT Setpoint Approach. This approach is similar to the UV
intensity sensor setpoint approach, except that the UV sensor is placed close to the
lamp such that it only responds to changes in UV lamp output. UVT is monitored
separately. For a specific flowrate, the UV intensity and UVT measurements are used
to monitor dose delivery. The setpoints for UV intensity and UVT over a range of
flowrates are determined during validation (see Chapter 4).
3. Calculated UV Dose Approach. In this approach, the UV intensity sensor is placed
close to the lamp, which is similar to the UV intensity and UVT setpoint approach.
Flowrate, UVT, and UV intensity are all monitored, and the outputs are used to
calculate UV dose via a validated computational algorithm developed by the UV
reactor manufacturer.
The strategy for dose monitoring depends on the manufacturer and is typically
proprietary. Dose monitoring recommendations are discussed in section 5.4.
UV Disinfection Guidance Manual A-52 June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
A.4 Water Quality Impacts and Byproduct Formation
Constituents in the water affect the performance of UV reactors. In addition, most
disinfectants form byproducts, and the goal of the overall disinfection process is to maximize
disinfection while minimizing byproduct formation.
A.4.1 Water Quality Impacts ,
UV absorbance, particle content, and constituents that foul lamp sleeves and other wetted
components are the most significant water quality factors impacting UV disinfection
effectiveness. In spite of these effects, the impact of water quality on dose delivery can be
adequately addressed in virtually all drinking water applications if carefully considered during
the design of the UV reactors.
A.4.1.1 UV Absorbance
The most important water quality parameter affecting reactor performance is UV
absorbance. As UV absorbance increases, the intensity throughout the reactor decreases for a
given lamp output. This results in a reduction in UV dose delivered to the microorganism and
measured UV intensity. Section 3.1.3.1 discusses how to incorporate the impact of UV
absorbance into UV reactor design.
UV absorbers in typical source waters include humic and fulyic acids, other aromatic
organics (e.g., phenols), metals (e.g., iron), and anions (e.g., nitrates and sulfites) (Yip and
Kpnasewich 1972; DeMers and Renner 1992). Both soluble and particulate forms of these
compounds will absorb UV light. UV absorbance will vary over time due to changing
concentrations of these compounds. Temporal variability in UV absorbance is greater in rivers
and small lakes than in large lakes and reservoirs. UV absorbance will vary seasonally due to
rainfall, lake stratification and destratification (turnover), and changes in biological activity of
organisms within the water source.
Water treatment processes can reduce the UV absorbance of water. Coagulation,
flocculation, and sedimentation'remove soluble and particulate material, and filtration removes
particles. Oxidants such as chlorine and ozone reduce soluble material, precipitate metals, and
reduce UV absorbance. Activated carbon absorption also reduces soluble organics. Because
these treatment processes reduce UV absorbance, the lowest UV absorbance occurs at the end of
the treatment train, and therefore UV disinfection is most effective when applied after filtration.
Chemicals used in the water treatment process can also increase the UV absorbance of the water,
and their impacts are discussed in section A.4.1.3.
A.4.1.2 Particles .
Particle content can also impact UV disinfection performance. Particles may absorb and
scatter light, thereby reducing the UV intensity delivered to the organisms. Particle-associated
microorganisms also may be shielded from UV light, effectively reducing disinfection
UV Disinfection Guidance Manual A-53 June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
performance as discussed in section A.2.6.5 and causing a tailing or flattening of the dose-
response curve when higher inactivation levels are desired. Particles in source waters are diverse
in composition and size and include large molecules, microorganisms, clay particles, algae, and
floes. Sources of particles include wastewater discharges, erosion, runoff, microbial growth, and
animal waste. The particle concentration will vary over time both seasonally and over the short
term. Storm events, lake turn over, and spring runoff are some events that increase the
concentration of particles.
Recent research by Linden et al. (2002b) indicated that the UV dose-response of
microorganisms added to filtered drinking waters is not altered by variation in turbidity of
filtered water that met regulatory requirements (40 CFR 141.73). For unfiltered raw waters,
Passantino and Malley (2001) found that source water turbidity up to 10 NTU does not impact
the UV dose-response of separately added (seeded) organisms. In these experiments, however,
organisms were added to waters containing various levels of treated or natural turbidity.
Therefore, it was not possible to examine microorganisms associated directly with particles in
their natural or. treated states. Consequently, these drinking water studies can only suggest the
impact of turbidity on dose-response as it relates to the impact of UV light scattering by particles
rather than particle-association or clumping of microorganisms.
Water treatment unit processes such as coagulation, flocculation, sedimentation, and
filtration are designed to remove particles from water. Organisms within coagulated and
flocculated particles will be more difficult to inactivate; however, they will typically be removed
during filtration.
A.4.1.3 Water Treatment Chemicals
Water treatment chemicals affect the UVT of the water, the formation of conglomerate
particles, and the fouling potential of the water.
Water treatment processes upstream of the UV reactors can be operated to control and
increase UVT, thereby optimizing the design and costs of the UV reactor. Chemicals such as
chlorine, ozone, and hydrogen peroxide oxidize UV-absorbing compounds but may also absorb
UV light with ozone showing the most pronounced effect on UV absorbance. Oxidant residuals
can be quenched with chemicals such as sodium thiosulfate or sodium bisulfite. However, the
use of these chemicals can also increase the UV absorbance of water.
Table A.5 lists the UV absorption coefficients of common water treatment chemicals and
their "impact threshold concentration", defined as the concentration that will decrease the UVT
from 91 to 90 percent. Of these chemicals, ozone and ferric iron have the greatest potential of
impacting the UV absorbance of water.
UV Disinfection Guidance Manual A-54 June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
Table A.5 UV Absorbance Characteristics of Common Water Treatment
Chemicals (Adapted from Bolton et al. 2001)
Compound ,
Ozone (O3) (aqueous)
Ferric iron (Fe3*)
Permanganate (Mn04~)
Thiosulfate ion {S2O32")
Hypochlorite ion (CIO")
Hydrogen peroxide (H2O2)
Ferrous iron (Fe2*)
Sulfite ion (SO32')
Zinc ion (Zn2+)
Ammonia (NH3)
Ammonium ion (NH/)
Calcium ion (Ca2*)
Hydroxide ion (OH')
Magnesium ion (Mg2*)
Manganese ion (Mn2*)
Phosphate species
Sulfate ion (SO42~)
Molar Absorbtion
Coefficient
(M'1 cm"1)
3,250
4,716
657
201
29.5
18.7
28
16.5
1.7
NSA
NSA
NSA
NSA
NSA
NSA
NSA
NSA
Mass-based
Absorbance
(L/mg cm"1)
0.0677
0.0844
0.0055
0.00178
0.000573
0.00055
0.0005
0.000206
0.000026
NSA
NSA
NSA
NSA
NSA
NSA
NSA
NSA
Impact
Threshold
Concentration 1
(mfl/L)
0.071
0.057
0.91
2.7
8.4
8.7
9.6
23
187
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
NSA No significant absorbance
N/A Not applicable
1 Concentration in mg/L resulting
(A254 increase from 0.041 cm"
in UVT decrease from 91 percent to 90 percent
1 to 0.046 cm"1)
A.4.1.4 Fouling Potential
Wetted components within a UV reactor can become fouled over time. Fouling on the
external surfaces of the lamp sleeve reduces the transmittance of UV light from the lamps into
the water, thereby reducing dose delivery. Fouling on UV intensity sensor windows reduces the
intensity of UV light measured by the sensors, resulting in under prediction of dose delivery.
Fouling on the inside surfaces of the reactor reduces reflection of UV light from those surfaces,
which reduces the amount of UV light available for disinfection.
Fouling on the wetted surfaces of a UV reactor has been attributed to the following
events:
Compounds whose solubility decreases as temperature increases will precipitate (e.g.,
CaCO3, CaSO4> MgCO3, MgSO4, FePO4, FeCO3, Al2(SO4)3). These compounds will
foul MP lamps faster than LP lamps due to differences in operating temperature.
Compounds with low solubility will precipitate (e.g., Fe(OH)3, Al(OH)3).
UV Disinfection Guidance Manual
Proposal Draft
A-55
June 2003
-------
Appendix A, Fundamentals of UV Disinfection
. Particles will deposit on the lamp sleeve surface due to gravity settling and
turbulence-induced collisions (Lin etal. I999a).
Precipitation will depend on the water temperature, pH, alkalinity, ion concentration,
total hardness, and the particle concentration. Residual concentrations of coagulants like ferric
sulfate can also affect fouling. The fouling will vary spatially along and around the lamp sleeve,
and will depend on the operating temperature of the lamp. Precipitation of compounds whose
solubility decreases with increasing temperature is more notable with lamps operating at higher
temperatures (e.g., MP lamps; Sheriff and Gehr 2001). Organic fouling can occur when a reactor
is left off and full of water for an extended period of time (Toivanen 2000).
Fouling rates on lamp sleeves are reported to follow first order kinetics after an initial
induction period (Lin et al. 1999b). Currently, there is not sufficient information to predict
quantitatively the fouling based on water quality. The potential for fouling and the frequency of
sleeve cleaning will be site and equipment specific. The fouling observed during several pilot-
and full-scale UV facilities is shown in section 2.5.1 (Table 2.3).
The Langelier Saturation Index (LSI) or the calcium carbonate precipitation potential
(CCPP) can be used to help indicate fouling potential. The LSI is defined as the difference
between the pH of the water and the pH at which calcium and carbonate are at equilibrium with
solid CaCC>3. The CCPP is the amount of calcium carbonate that will precipitate when
equilibrium conditions in the water have been reached. Both the LSI and CCPP are functions of
temperature, pH, calcium hardness, total dissolved solids (TDS), and alkalinity. For UV
disinfection, the temperature of the lamp sleeve surface should be used to calculate the LSI and
CCPP. The LSI and CCPP will depend on upstream processes, such as pH adjustment and lime
softening, and may vary daily or seasonally.
A.4.1.S Algae Growth
Visible light emitted from UV lamps may promote algae growth in UV reactors and the
surrounding piping'. Depending on the species, algae growth could cause taste and odor
problems in the finished water. Algae growth is a greater issue with MP lamps than LP lamps
because MP lamps produce more light in the visible range. Algae growth also depends on water
temperature, pH, and nutrient concentration (Sterner and Grover 1998).
A.4.2 Disinfection Byproducts
UV disinfection byproducts (DBFs) arise either directly through photochemical reactions
or indirectly through reactions with products of photochemical reactions. Photochemical
reactions will only take place if a chemical species absorbs UV light, and the resulting excited
state reacts to form a new species. The resulting concentration of new species will depend on the
concentration of the reactants and the UV dose.
When UV light is absorbed by an atom, electrons within the atom are excited to higher
energy states. An excited atom may return to its original ground state releasing the absorbed
energy as light, or it may interact with other atoms forming or breaking bonds. The formation or
UV Disinfection Guidance Manual A-56 June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
breaking of bonds between atoms results in the formation of a new chemical species. Chemical
reactions promoted by light are termed photochemical reactions. In some cases, the products of
photochemical reactions are radical species. Radical species may react with other chemicals to
form new chemical species (i.e., UV DBFs).
In drinking water, research has focused on the impact of UV light on the formation of
halogenated DBFs following subsequent chlorination and the transformation of organic material
to more degradabie components. For ground water and filtered drinking water, UV disinfection
at typical doses is not shown to impact the formation of trihalomethanes (THM) or haloacetic
acids (HAA), two categories of DBFs currently regulated by EPA (Malley et al. 1995;
Kashinkunti et al. 2003). Several studies have shown low-level formation of degradabie, non-
regulated DBFs (e.g., aldehydes) as a result of applying UV light to wastewater and raw drinking
water sources. However, a study performed with filtered drinking water indicates no significant
change in aldehydes, carboxylic acids, or total organic halides (TOX) (Kashinkunti et al. 2003).
The difference in results can be attributed to the difference in water quality, most notably the
higher concentration of organic material in raw waters and wastewaters.
Akhlaq et al. (1990) reported that UV doses of 250 mJ/cm2 from an LP lamp do not break
down alginic acid, a model compound for polysaccharides in drinking water. They concluded
that UV disinfection does not increase the assimilable organic carbon (AOC) of drinking water.
With UV doses ranging from 18 to 161 mJ/cm2, Kruithof et al. (1989) reported no increase in
AOC or mutagenicity of a granular activated carbon (GAC) filtrate.
Malley et al. (1995) evaluated the impact of UV doses of 60,130, and 200 mJ/cm2 on
DBF formation in ground waters and treated surface waters. They reported no change in pH,
turbidity, dissolved organic carbon, A254, color, nitrate, nitrite, bromide, iron, or manganese.
Formaldehyde increased from 1.2 to 12.1 p.g/L with one highly colored ground water.
Formaldehyde increased from less than 2 (ig/L up to 14 jo.g/L with untreated surface waters but
only 2 to 3 ng/L with treated surface waters. A small but insignificant increase in AOC was
observed with all waters.
Zheng et al. (1999) observed an 8 to 17 percent decrease in THM and a 9 to 19 percent
increase in HAA when MP UV light was applied at a dose of 2000 mJ/cm2 after chlorination.
However, at a lower dose of 100 mJ/cm2, they observed a 1 to 7 percent decrease in THM and no
change in HAA.
A low conversion of nitrate to nitrite by UV light has been observed (approximately 1
percent; Sharpless and Linden 2001). Von Sonntag and Schuchmann (1992) also reported 0.001
and 0.072 mg/L nitrite formed from 50 mg/L nitrate exposed to 25 mJ/cm2 from LP and MP
lamps, respectively. Conversion is lower with LP lamps than MP lamps because the UV
absorbance of nitrate is higher below 240 nm than it is at 254 nm.
UV Disinfection Guidance Manual
Proposal Draft
A-57
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
A.5 References
Adler, H.I. 1966. The genetic control of radiation sensitivity in microorganisms. Advances in
Radiation 2: 167-191.
Akhlaq, M.S., H.P. Schuchmann, and C. von Sonntag. 1990. Degradation of the polysaccharide
alginic acid: a comparison of the effects of UV light and ozone. Environmental Science
and Technology 24: 379-383.
APHA-AWWA-WEF. 1998. Standard methods for the examination of water and wastewater,
20th ed. Washington D.C.
BattigelH, D.A., M.D. Sobsey, and D.C. Lobe. 1993. The inactivation of Hepatitis A virus and
other model viruses by UV irradiation. Water Science & Technology 27, no. 3-4: 339-
342.
Bolton, J.R. 2000. Calculation of ultraviolet fluence rate distributions in an annular reactor:
significance of refraction and reflection. Water Research 34, no. 13: 3315-3324.
Bolton, J!R. and K.G. Linden. 2003. Standardization of Methods for Fluence (UV Dose)
Determination in Bench-scale UV Experiments. ASCE Journal of Environmental
Engineering 129, no. 3: 209-215.
Bolton, J.R., M.I.'Stefan, R.S. Gushing, and E. Mackey. 2001. The importance of water
absorbance on the efficiency of ultraviolet disinfection reactors. Proceedings of the
IUVA 1st International Congress, June 14-16, Washington, D.C.
Brandt, C.L. and A.C. Giese. 1956. Photoreversal of nuclear and cytoplasmic effects of short
ultraviolet radiation on Paramecium caudatum. Journal of General Physiology 39, no. 5:
735-751.
Cabaj, A., R. Sommer, W. Pribil, and T. Haider. 2001. What means "dose" in UV-disinfection
with medium pressure lamps. Ozone Science & Engineering 23: 239-244.
Chang, J.C.H., S.F. Osoff, D.C. Lobe, M.H. Dorfman, C.M. Dumais, R.G. Quails, and J.D.
Johnson. 1985. UV inactivation of pathogenic and indicator microorganisms. Applied
and Environmental Microbiology 49, no.6: 1361-1365.
Clancy, J.L., Z. Bukhari, T.M. Hargy, J.R. Bolton, B.W. Dussert, and M.M. Marshall. 2000.
Using UV to Inactivate Cryptosporidium. Journal of the American Water Works
Association 92, no. 9:97-104.
DeMers, L.D. and R.C. Renner. 1992. Alternative Disinfection Technologies for Small Drinking
Water Systems. Chapter 5. American Water Works Association Research Foundation,
Denver Colorado.
Dulbecco, R. 1950. Experiments on photoreactivation of bacteriophages inactivated with
ultraviolet radiation. Journal of Bacteriology 59: 329-347.
UV Disinfection Guidance Manual A-58 June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection*
GE Quartz. 2001. Available from Internet location http://www.gequartz.com/en/lamptube.htm.
Internet.
Giese A.C., C.L. Brandt, C. Jacobson, D.C. Shephard, and R.T. Sanders. 1954. The effects of
starvation on photoreactivation in Colpidium colpoda. Physiological Zoology 71:71 -78.
Gurzadyan, G.G., D.N. Nikogosyan, P.G. Kryukov, V.S. Letokhov, T.S. Balmukhanov, A.A.
Belogurov, and G.B. Zavilgelsky. 1981. Mechanism of high power picosecond laser UV
inactivation of viruses and bacterial plasmids. Photochemistry and Photobiology 33: 835-
838.
Harris, G.D., V.D. Adams, D.L. Sorensen, and M.S. Curtis. 1987. Ultraviolet inactivation of,
selected-bacteria and virus with photoreactivation of the bacteria. Water Research 21, no.
6: 687-692.
Hollaender, A. and B.M. Duggar. 1936. Irradiation of plant viruses and of microorganisms with
monochromatic light. III. resistance of the virus of typical tobacco mosaic and
Escherichia Coli to radiation from L 3000 To L 2250 A. Proc. N. A. S .22: 19-24.
Hoyer, 0.1998. Testing performance and monitoring of UV systems for drinking water
disinfection. Water Supply 16, no. 1/2: 424-429.
Irazoqui, H.A., J. Cerda, and A.E. Cassano. 1973. Radiation profiles in an empty annular
photoreactor with a source of finite spatial dimensions. AlChE Journal 19, no. 3: 460-
467.
Jacob, S.M. and J.S. Dranoff. 1970. Light intensity profiles in a perfectly-mixed photoreactor.
AIChE Journal 16, no. 3: 359-363.
Jagger, J. 1967. Introduction to research in ultravioletphotobiology. Englewood Cliffs, NJ:
Prentice Hall, Inc.
Jagger, J., T. Possum, and S. McCaul. 1975. Ultraviolet irradiation of suspensions of micro-
organisms: possible errors involved in the estimation of average fluence per cell.
Photochemistry and Photobiology 21: 3 79-3 82.
Kashimada, K., N. Kamiko, K. Yamamoto, and S. Ohgaki. 1996. Assessment of
. photoreactivation following ultraviolet light disinfection. Water Science & Technology
33, no. 10-11:261-269.
Kashinkunti, R., K.G. Linden, G. Shin, D.H. Metz, M.D. Sobsey, M. Moran, and A. Samuelson.
2003. Achieving multi-barrier inactivation in Cincinnati: UV, byproducts, and
biostability. Journaloj'the American Water Works Association. Submitted January 9,
2003.
UV Disinfection Guidance Manual A-59 June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
Kelner, A. 1950. Light-induced recovery of microorganisms from ultraviolet radiation injury,
with special reference to Escherichia coli. Bulletin of the New York Academy of Medicine
26: 189-199.
Knudson, G.B. 1985. Photoreactivation of UV-irradiated Legionella Pneumoplila and other
Legionella species. Applied and Environmental Microbiology 49, no. 4: 975-980.
Kruithof, J.C., R.C. van der Leer, W.A.M. Hijnen, P.A.N.M. Nuhn, F.A.P. Houtepen, and L.A.C.
Feij. 1989. Ultra violet disinfection of carbon filtered drinking water. Ozone + Win the
treatment of water and other liquids. Edited by W.J.'Masschelein. Ill 3.1 - I1I3.15. Berlin
Lin, L.S., C.T. Johnston, and E.R. Blatchley III. 1999a. Inorganic fouling at quartz:water
interfaces in ultraviolet photoreactors I: chemical characterization. Water Research 33,
no. 15:3321-3329.
Lin, L.S., C.T. Johnston, and E.R. Blatchley III. 1999b. Inorganic fouling at quartz: water
interfaces in ultraviolet photoreactors - II Temporal And Spatial Distribution. Water
Research 33, no. 15: 3330-3338.
Linden, K.G. and J.L. Darby. 1998. UV disinfection of marginal effluents: determining UV
absorbance and subsequent estimation of UV intensity. Water Environment Research 70,
no. 2: 214-223.
Linden, K.G., G.A. Shin, and M.D. Sobsey. 2001. Comparative effectiveness of UV
wavelengths for the inactivation of Cryptosporidium parvum oocysts in water. Water
Science & Technology 43, no. 12: 171-174.
Linden, K.G., G.A. Shin, G. Faubert, W. Cairns, and M.D. Sobsey. 2002a. UV disinfection of
Giardia lamblia cysts in water. Environmental Science & Technology 36, no. 11:2519-
2522.
Linden, K.G., L. Batch, and C. Schulz. 2002b. UV Disinfection of filtered water supplies: water
quality impacts on MS2 dose-response curves. Proceedings of the AWWA Annual
Conference, June 16-20, New Orleans, LA.
Lindenauer, K.G. and J.L. Darby. 1994. Ultraviolet disinfection of wastewater: effect of UV
dose on subsequent photoreactivation. Water Research 28: 805.
Loge, F.J., R.W. Emerick, D.E. Thompson, D.C. Nelson, and J.L. Darby. 1999. Factors
influencing ultraviolet performance. Parti: Light penetration to wastewater particles.
Water Research 71, no. 6: 377-381.
Lytle, C.D. 1971. Host-cell reactivtaion in mammalian cells I. Survival of ultra-violet-irradiated
Herpes virus in different cell-lines. InternationalJoumal of Radiation Biology 19, no. 4:
329-337.
UV Disinfection Guidance Manual
Proposal Draft
A-60
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
Mackey, E.D., R.S. Gushing, M.L. Janex, N. Picard, J.M. Laine, and J.P. Malley. 2003.
Bridging Pilot-Scale Testing to Full-Scale Design of UV Disinfection Systems, Draft
Report. Final report to be published in 2004. Denver, CO: AWWA Research
Foundation.
Malley, J.P. 2000. The state of the art in using UV disinfection for waters and wastewaters in
North America. Proceedings of ENVIRO 2000 - Australian Water Association (AWA),
April 9-13, Sydney, Australia.
Malley, J.P, J.P. Shaw, and J.R. Ropp. 1995. Evaluation of by-products produced by treatment
ofgroundwaters with ultraviolet irradiation. Denver, CO.: AWWA Research
Foundation.
Meng Q.S. and C.P. Gerba. 1996. Comparative inactivation of enteric adenovirus, poliovirus and
coliphages by ultraviolet irradiation. Water Research 30, no. 11:2665-2668.
Meulemans, C.C.E. 1986. The basic principles ofUV-sterilization of water. In: Ozone +
Ultraviolet Water Treatment, Aquatec Amsterdam, Paris: International Ozone
Association, B.I.1-B.l.13.
Mofidi, A.A., E.A. Meyer, P.M. Wallis, C.I. Chou, B.P. Meyer, S. Ramalingam, and B.M.
Coffey. 2002. The effect of UV light on the inactivation ofGiardia lamblia and Giardia
muris cysts as determined by animal infectivity assay (P-2951-01). Water Research 36:
2098-2108.
Morton, R.A. and R.H. Haynes. 1969. Changes in the ultraviolet sensitivity of Escherichia coli
during growth in batch cultures. Journal of Bacteriology 97(3): 1379-1385.
Oliver, B.G. and E.G. Cosgrove. 1975. The disinfection of sewage treatment plant effluents
using ultraviolet light. Canadian Journal of Chemical Engineering 53: 170-174.
Oppenheimer, J.A., J.E. Hoagland, J.-M. Laine, J.G. Jacangelo, and A. Bhamrah. 1993.
Microbial inactivation and characterization of toxicity and by-products occurring in
reclaimed wastewater disinfected with UV radiation. Planning, design & operation of
effluent disinfection systems, 13-24. Whippany, N.J.: the Water Environment Federation.
Parker, J.A. and J.L. Darby. 1995. Particle-associated coliform in secondary effluents: shielding
from ultraviolet light disinfection. Water Environment Research 67, no 7: 1065-1075.
Passantino, L.B. and J.P. Malley. 2001. Impacts of turbidity and algal content of unfiltered
drinking water supplies on the ultraviolet disinfection process. Proceedings of the 2001
AWWA Annual Conference, Washington D.C.
Persson, E, and D. Kuusisto. 1998. A performance comparison of electronic vs. magnetic
ballast for powering gas-discharge UV lamps. Proceedings of the RadTech Conference,
RadTech International, Chicago, IL.
UV Disinfection Guidance Manual
Proposal Draft
A-61
June 2003
-------
Appendix A. Fundamentals of UV Disinfection
Petri, M.P., G. Fang, J.P. Malley, D.C. Moran, and H. Wright. 2000. Groundwater UV
disinfection: challenges and solutions. Proceedings of the Water Quality Technology
Conference, AWWA, November 5-9. Salt Lake City, UT.
Pfeifer, G.P. 1997. Formation and processing of UV photoproducts: effects of DNA sequence
and chromatin environment. Photochemistry & Photobiology 65(2): 270-283.
Phillips, R. 1983. Sources and application of ultraviolet radiation. New York: Academic Press.
Powell, W.F. 1959. Radiosensitivity as an index of Herpes Simplex virus development.
Virology 9: 1-19.
Quails, R.G., M.P. Flynn, and J.D. Johnson. 1983. The role of suspended particles in ultraviolet
disinfection. Journal WPCF 55, no 10: 1280-1285.
Rauth, A.M. 1965. The physical state of viral nucleic acid and the sensitivity of viruses to
ultraviolet light. BiophysicalJournal 5:257-273.
Schenck, G.0.1981. Ultraviolet sterilization. Handbook of water purification. 2nd ed. Edited by
W. Lorch. New York: John Wiley & Sons.
Setlow, J.K. 1967. Comprehensive biochemistry. Chapter 5 in The effects of ultraviolet radiation
andphotoreactivation27: 157-209.
Severin, B.F., M.T. Suidan, and R.S. Engelbrecht. 1983. Effects of temperature on ultraviolet
light disinfection. Environmental Science & Technology 17( 12): 717-721.
Severin, B.F., M.T. Suidan, and R.S. Engelbrecht. 1984. Mixing effects in UV disinfection.
Journal WPC. 56, no 7: 881-888.
Sharpless C.M. and K.G. Linden. 2001. UV photolysis of nitrate: effects of natural organic
matter and dissolved inorganic carbon, and implications for UV water disinfection.
Environmental Science & Technology 35, no 14: 2949-2955.
Sheriff, M, and R. Gehr. 2001. Laboratory investigation of inorganic fouling of low pressure UV
disinfection lamps. Water Quality Research Journal of Canada 36, no 1: 71 -92.
Shin, G-A., K.G. Linden,'M.J. Arrowood, and M.D. Sobsey. 2001: Low-pressure UV
inactivation and DNA repair potential of Cryptosporidium parvum oocysts. Applied &
Environmental Microbiology 67 no. 7: 3029-3032.
Snowball, M.R. and I.S. Hornsey. 1988. Purification of water supplies using ultraviolet light.
Developments in food microbiology, edited by R.K. Robinson, 171-191. NY: Elsevier
Applied Science.
Sommer, R., T. Haider, A. Cabaj, W. Pribil, and M. Lhotsky. 1998. Time dose reciprocity in UV
disinfection of water. Water Science & Technology 38:145-150.
UV Disinfection Guidance Manual A-62 June 2003
Proposal Draft
-------
Appendix A. Fundamentals of UV Disinfection
Sterner, R.W. and J.P. Grover. 1998. Algal growth in warm temperate reservoirs: kinetic
examination of nitrogen, temperature, light, and other nutrients. Water Research 32, no
12: 3539-3548.
Toivanen, E. 2000. Experiences with UV disinfection at Helsinki water. IUVA News 2, no. 6:4-
8.
Tosa, K. and T. Hirata. 1998. HRWM-39: Photoreactivation of Salmonella following UV
disinfection. Vol. 10 Health-Related Water Microbiology. IAWQ 19th Biennial
International Conference, Jun21 -26, 168-174.
Tree, J.A., M.R. Adams, and D.N. Lees. 1997. Virus inactivation during disinfection of
wastewater by chlorination and UV irradiation and the efficacy of F+ bacteriophage as a
viral indicator. Water Science & Technology 35(11-12): 227-232.
von Sonntag, C. and H-P. Schuchmann. 1992. UV disinfection of drinking water and by-product
formation—some basic considerations. Journal of Water Supply Research and
Technology 41, no 2: 67-74 .
Wiedenmann, A., B. Fischer, U. Straub, C.-H. Wang, B. Flehmig, and D. Schoenen. 1993.
Disinfection of Hepatitis A virus and MS-2 coliphage in water by ultraviolet irradiation:
Comparison of UV-susceptibility. Water Science & Technology 27, no 3-4: 335-338.
Wilson, B.R., P-.F. Roessler, E. Van Dellen, M. Abbaszadegan, and C.P. Gerba. 1992. Coliphage
MS-2 as a UV water disinfection efficacy test surrogate for bacterial and viral pathogens,
219-235. Proceedings of the water quality technology conference, Nov. 15-19, Toronto.
Wright, H.B. and G. Sakamoto. 1999. UVdose required to achieve incremental log inactivation
of bacteria, virus, and protozoa. Trojan Technologies, Inc., London, Ontario, Canada.
Wright, H.B. and Y.A. Lawryshyn. 2000. An assessment of the bioassay concept for UV reactor
validation. Disinfection of Wastes in the New Millenium, New Orleans, Louisiana,
March 15-18, 2000; Water Environment Federation, Alexandria, Virginia
Yip, R.W. and D.E. Konasewich. 1972. Ultraviolet sterilization of water - its potential and
\\mitations.WaterandPollutionControl. 14:14-18.
Zheng, M, S.A. Andrews, and J.R. Bolton. 1999. Impacts of medium pressure UV on THM and
HAA formation in pre-UV chlorinated-drinking water. Water Quality Technology
Conference, October 31-November 3, Tampa, F.L.
UV Disinfection Guidance Manual A-63 June 2003
Proposal Draft
-------
Appendix B. Derivation of UV Dose-Response Requirements
In support of the Long Term 2 Enhanced Surface Water Treatment Rule (LT2ESWTR),
the U.S. Environmental Protection Agency (EPA) developed UV dose requirements for
Cryptosporidium, Giardia, and virus inactivation. The requirements represent the UV dose
necessary to achieve a given inactivation level, similar to the concentration * time (CT)
requirements for chemical disinfectants.
The UV dose requirements were developed to account for uncertainty associated with the
dose-response of microorganisms (Cryptosporidium, Giardia, and virus) in controlled
experimental conditions. In practical application, other sources of variability and uncertainty are
introduced due to the hydraulic effects of the UV installation, UV reactor, and UV intensity
sensors, the validation protocol, as described in Chapter 4 and Appendix C, addresses these and
other areas of variability and uncertainty by applying safety factors to the UV dose requirements
derived in this appendix. Therefore, the dose requirements presented in this appendix are not the
actual dose levels at which utilities will be required to validate and operate UV reactors for a
given log inactivation.
This appendix explains the derivation of the UV dose requirements through a three-step
process of data collection, qualitative review to establish working data sets, and mathematical
analyses.
B.1 Data Collection
EPA collected UV dose-response research data for adenovirus, Giardia lamblia, Giardia.
muris, and Cryptosporidium parvum. Adenovirus was evaluated because, of the data available, it
is considered the most resistant to inactivation by UV light of the pathogenic waterborne viruses.
In compiling data, EPA reviewed published and unpublished studies conducted over the past 50
years as provided in published literature, electronic databases, research reports, and conference
proceedings. The experimental conditions varied among batch and continuous flow UV
apparatuses, types of UV lamps, and water quality conditions. Table B.I summarizes these
studies.
UV Disinfection Guidance Manual B-l . June 2003
Proposal Draft
-------
O
u
I
D,
I
e
o
I
Q
CQ-
x
"-3
a.
<
U
.2
"5
O
B
to
I
w
m
(0
i
18
|
1
a>
E
•c
Q.
X
UJ
c
o
•a
n
s.
=
IS
S
u
E
S
o
S
,_
II
'c
m *
UVDOS
Measurerr
11
.jl r*
1
E v
P
•a
Assay Use
I
c
1
CO
1
X
8
w
E
€0
1
(U
o
ffl>
C.
_ ~ Z
t? '5 w £
ct 5
8
§
«
3.
^
in
Cell Cultui
{PLC/PRFy
CM
z
c
3
1
E
>
c
1
o
z
o
m
•e
<3
JD
J3
o
Q.
f
CD
Cell Cultui
(KB cells
S
1
c
1
1
i
'5
o
c
s
£
13
Ifs
So?
X)
C
'6>
•5
CL
g
S
p
Cell Cultui
(Vero)
o>
z
c
(0
X
z
to
s
1
8
2
15
«
ro o>
JD
J3
I is
MP-Calcut:
(DNA),
Radiome
O.
•5
1
gT3
IP
5
1
c
fij
z
1
c
o
Z
1
I
JQ
ffi
_
.2
0.
•5
&
(0
"
•o g>
0> 0)
50
JO
•,_
Radiome
a.
'J
S
ff
o S.
"a "~"
O
in
1
c
at
3
Z
.2
2 •
a.
^
'I
£
11
S4
~s
z
c
1
Z
to
'1
c
i
in
£
"5
c
|l
jE -5
E"6 -
S § «
fi 2 i
-JjU
j_
o
1
ct:
a.
i
a
-------
!
c
O
« •.•&
CO (Q
^
o •
'E
r
i
!
c
I
' 0
s
-Q
W
E
1
^| e_
».ili
°- g ji
Q:-I
Reference
f- "P w.
s = a
•o
•§
o S
1s
0.
5
'S
a
Mouse
Infectivity
(C3H/HeN)
1
s
•£
;§
m
*>
i
«B
1
<5
in
Craik et al.
2000
A
5
-ffi
E
.0
• a;
Q.
_i
g.
S
Mouse
Infectivity
1
z
Hamster
co
S
•2
S
5
2
Hayes et al.
2001
"S
£
•5
c
D
c
•S
•5
Q.
E
§
£
it
z
:
z
i
.">
i
•2
e
(5
o
Oppenheimer
et al. 2002
£»
m
_j
u
E
o
1
a:
a.
_i
|
a
_.$•
c
1
z
c
CO
r
•2
-2
•2
1
S
CM
O (D CN
.Q
1
£
.2
i
0.
_i
1
a
if
z
z
0>
•2
-2
-.2
i
(3
o
75
|g
.£ o
—1 CM
•%$
E §
•81
(I) m
*i
>,
W
ra
.2
CO
^
ti.
|
if
1
z
i
.2
•S
.2
i
5
0
Malley 2000a
J«
_s
g
E
o
=5
2
£L
_l
g
S
• f-f
W .> Q .>
3 ts "S is
^ H !s
z
<
z
c
i
!§
1
•2
1
5
m
Mofidi et al.
2002
-------
I
3
cr
O
u
o
JO
o
u
i
o
ffl
.X
•o
I
Q.
CO
m
ental Information
in
.2
P
|
jjj
2
£
u
I
-
C
(U
C
n (0
UV Dose
Measurement
Q. a)
JS H
"c
1$
r
i
0)
.S
is
to
1
«
S
CO
1
'c
fo
-.5 IP
dj £ *—••
Q£ i— i
Reference
ix >
2 to
I5"
0.
S
10
3 -
Continuo
Flow
Si5
ijo
z
5
g
™*
s
i.
e
1
^
1
Bukhari et al.
1999
jcE
"S J E 51
_l O fl> !n
ro Q- S.
MP-Calculated
(DMA)
Radiometer
Si o.
5 -1
1
m
if I
i
S
S
~
S
§
ffl
g
i
6
g.
(2
0)
Clancy et al.
2000
IS
c
D
Radiometer
MP-Calculated
(DMA)
Q.
5
Si
~*
o
its
* t —
z
5
o
^
0)
.5
i
1
i-
S
Q
o
I Clancy Env
2000
'
a
Radiometer
D.
_i
1 -
«f
CO .>
11
z
D J „ « o
S g> g 'S w
j_ o — S ^
1
£
3
1
S
^
1
II
OOL 2^
•§§£§
-J=*il>
Radiometer
- calculated
(L "•
-1 S
|
|||
i
S
g
w.
(U
|
1
fi
A"
1
Craik et al.
2001
TJ
Untreate
Surface
Water
]i
Q.
S
S
Continuo
Flow
as .*; — -
1
S
g
•~
g
1
6
a
1
3
S.
Q
in
Hargy et al.
2000
3
Radiometer
D.
_l
I.
!~
o^
1
5
o
^
s:
i
E
3
TO
Q. .
S
o
Landis et al.
2000
ill
to
ID
O
m
Q
J£
Q.
Continue
- Flow
Il§
£ "~
1
§
g
*—
0) .
|
1
^
g
(?
o
Mackey et al.
2000
t
0 '
S 1
"[5
9
i
o
'3
0
o .,
UV Disinfecti
Proposal Draf
-------
1
i
1
1
od
S
§
€X
S
Dose-R
•H^
•y
•s
P4
E«
_O
?
•c
4>
Q
cd
_g
•S
1
<
G*
O
ollected (c
U
XS
S
*-
o
£»
re
E '
3
CO
^
CO
O
2
re
£
o
B
c
E
5
^
1
•c
o
Q.
X
UJ
e
i
I
£
£
M
1
O
*•"•
s
General
«t
CO CO
*B
•E
0) 0>
to E
O <]>
3
'^> (0
3 2
S
Q. d)
J3^
1
E g
"S *
Q.I-
X
UJ
•o
0>
w
>•
!
c
'2
w
«
J_
$
'D
CO
E
.52
i
cf
Peer
Reviewed
Literature
Reference (Y/N)
Q- g) 0>
& j£ >
^
"O <1>
i -2 "S
sH «
•s"5
S°
I5
SOL
5
"(5
CO
Cell Culture
(HCT-8) and
RT-PCR
^
z
S
jt
°
0)
,c
m
£
§
§.
a
o
Mofidi et al. No
1999
1
J
ii=
c
D
c
8
7
"S
z
Q.
£
§
CO
m
>>
sl-
ip
^ '
Z
(S
5
2
£
o
H
ro
cc
0.
_j
fi
TO
m
Cell Culture
(MDCK)
1
SS
I*
jO
1
CD '
• £
§
8.
a
*
.1
"55
'S x>
ll
Q- ffi «
>JI5
5J
f£
to
^
0^
0.
_j
5
ro
CO
Cell Culture
(MDCK)
1
„
•£
£
!
CO
e •
s
a
5
*
Kashinkunti No
etal.2002
fcl|
t .E TJ k_ . -tt
1 i Idlli
! i iljtl
^ iE 'S 'C ^ m
o ^ as al^ »
i l^lllli
C J ._ ftt m ^3 d) *«
v ^ re (D ,« J-J ^ TO
_i_ o $ "S o co d^ ^*
T1. ilsill^
Q. c 5 T) o 3 nj i_
e .2 * » -| c p j
— i«* (u ^^ ... o *- *?
Jjf p ^ ^^ J" .V. Q) w
!g ^ £-| CL i 1 1
&. .iS"!^
* -c ^ ^7 ^ 5 e i
N/A - Not applicable; LP - lo'
1 Water Quality Definitions:
Lab - Tap water treated in t
Reclaimed Wastewater - T<
Low Turbidity Wastewater •
Unfiltered -Water that mee
Untreated Surface Water -
WTP Filtered Water -Filtei
Lab Filtered Surface Water
m
O
S
£
|3
§
.S —
-------
Appendix B. Derivation of UV Dose-Response Requirements
B.2 Data Review—Criteria for Inclusion in Statistical Analysis
EPA evaluated the data presented in Table B.I to determine the data sets to be used in
analyzing dose-response for each target microorganism. To be included in the statistical
analysis, the experimental design had to be sufficiently documented with respect to
experimental conditions, methodology, and calculation of results to allow an accurate
assessment of UV dose-response. For instance, studies were not included if the report did
not provide sufficient information to determine the UV dose measurement method or whether
the reported UV dose accounted for appropriate parameters (e.g., UV absorbance). The
statistical dose-response analysis combines data across different experimental designs and
conditions; therefore, it is important to ensure the differences between studies do not affect
the UV dose-response relationship.
B.2.1 Appropriate Experimental Design and Conditions
Research studies with the following criteria were selected for the statistical analyses:.
• Batch experimental design
. Low pressure (LP) lamps as the UV light source
. Filtered water, high quality unfiltered water, laboratory water, or low turbidity
reclaimed wastewater
« UV dose of the target microorganism inactivation directly measured and not derived
from the inactivation response of another microorganism
Data from continuous flow studies were not included in the analyses because flow-
through UV reactors apply a distribution of UV doses as opposed to a single dose. Moreover,
UV dose in a reactor is difficult to calculate precisely due to the variability in hydraulic detention
time and UV intensity distributions in reactors.
Studies were not included if the researchers utilized a UV light source that did not have a
widely accepted dose measurement methodology, such as pulsed UV lamps. Medium pressure
(MP) lamps pose a challenge of dose measurement due to the polychromatic nature of the MP
UV light and the absence of a standard method for calculating dose from MP lamps. The results
of a t-test indicated the LP and MP UV dose-response data, as reported, were statistically
different1; therefore, only LP lamp data were used in the statistical analyses.
Given the potential interference of particles in water and the fact that utilities installing
UV disinfection would need to meet finished water turbidity levels, EPA restricted media to •
water with turbidity values less than or equal to 1 nephelometric turbidity unit (NTU).
1 The t-test was calculated with Cryptosporidium data at low doses. The Giardia data and higher dose
Cryptosporidium data had too many data reported as "greater than a value" (referred to as censored data) and thus,
could not be used in a t-test. The adenovirus data had too few MP data to conduct a t-test.
UV Disinfection Guidance Manual B-6 June 2003
Proposal Draft
-------
Appendix B. Derivation of UV Dose-Response Requirements
Studies utilizing non-standard microbial assay methods (i.e., not generally accepted in
standard microbiological methods references) or studies not providing an evaluation of pathogen
infectivity were not included.
Note that many research studies evaluated multiple experimental conditions, but only the
subset of data meeting the criteria specified for this statistical analysis were used.
B.2.2 .Research Studies and Data Included in Statistical Analysis
r
UV dose-response data sets for adenovirus, Cryptosporidium parvum, and Giardia
lamblia and Giarida muris that met the criteria specified previously are presented in this section.
B.2.2.1 Viruses
For adenovirus, 4 of the 9 studies met the criteria discussed for inclusion in the statistical
analysis. Figure B.I shows the data of the selected studies.
Figure B.1 Observed Adenovirus Data from Selected Research Studies
M.
X
» ,
so
100
150
UV Dose (mjfcm1)
200
250
300
* Garbs 2000 (Type 2)
»Thompson et al,. 2002 (Type 2)
• Thurslon at al.,2002 (Type -40)
» Meng and Ge*a, 1896 (Type 40 and 11) +Shinetal.. 2001a (TypeS)
x Thompson et al.. 2002 (Type 15) •Thompson et al.. 2002 (Type Not Avail.)
UV Disinfection Guidance Manual
Proposal Draft
B-7
June 2003
-------
Appendix B. Derivation of UV Dose-Response Requirements
B.2.2.2 Protozoa
For Cryptosporidium parvum, 9 of the 13 studies met the criteria for inclusion in the
statistical analysis. For Giardia (including both lamblia and muris), 6 of the 8 studies were
included. Figures B.2 and B.3 show the Cryptosporidium parvum and Giardia data of the
selected studies, respectively. The data are both censored and uncensored and noted as such on
each graph. Censored data are those with log inactivation of "greater than" a particular value
rather than an absolute value (termed uncensored).
Figure B.2 Cryptosporidium Data from Selected Research Studies
5.0
4.5
4.0
3.5
c
o
1 2.S
g> 2.0
1.5
1.0
0.5
a
n
o
•-0-0-
o.o o&-
0
10 15 20 25 30
UV Dose (mJ/cm2)
35
40
45
50
* Clancy et al, 2000 "• Clancy et al., 2000 « Clancy et al., 2002 • Clancy et at.. 2002
Ocraiketal., 2001 SCraiketal., 2001 ° Landis et al., 2000 « Landis et al., 2000
A Clancy Envionmental, 2002 * Clancy Envionmental, 2002 G Oppenheimer et al., 2002 * Oppenheimer et al., 2002
O Shin et al., 2001 b * Shin etal., 2001 b OKashinkunti et al., 2002 •Kashinkunti et al., 2002
x Sommer et al., 2001 I n Open symbols report Log InaclivatioO Closed symbols report > Log i.
UV Disinfection Guidance Manual
Proposal Draft
B-8
June 2003
-------
Appendix B. Derivation of UV Dose-Response Requirements
Figure B.3 Giardia Data from Selected Research Studies
o.u
3.5 "
0 an-
2.0
1.0 •
o.o-
c
A «° .
•
8 .
a
D
^ tf •
O
• A° A
©
.*.... n-- — -
*
o
D
0
*
) 5 10 15 20 25 30
UV Dose (mJ/cm2)
0 Mofidi et al., 2002 - G. lamblia • Mofidi et al., 2002 - G. lamblia
A Linden et al., 2002 - G. lamblia A Linden et al., 2002 - G. lamblia
0 Hayes et al., 2001 - G. muris • Hayes et al., 2001 - G. muris
* Mofidi et al., 2002 - G. muris * Mofidi et al., 2002 - G. muris
n Campbell and Wallis, 2002 - G. lamblia ° Oppenheimer et al., 2002 - G. muris
35 40 45 50
o Open symbols report Log Inactivation
• Closed symbols report > Log Inactivation
B.3 Statistical Analysis
To determine the relationships between UV dose and log inactivation of
Cryptosporidium, Giardia, and virus, a mathematical model with hierarchical Bayesian
parameter estimation techniques was used. This model performs a meta-analysis that
summarizes and integrates the findings of multiple research studies. It can be considered as a
compromise of two extreme methods of combining data from different sources. One extreme
method treats the data from different sources as identical replications and computes a regression
as if the data were from a single source. The second extreme method treats each individual study
as totally unrelated to other studies. In this second method, the separately estimated regression
coefficients are pooled only to reflect the possible range. The Bayesian meta-analysis treats the
studies as exchangeable, but not identical or completely unrelated (Hedges 1997). Regression
coefficients for each study are estimated using the same calculations and allowed to differ
between studies. A Bayesian hierarchical modeling approach represents a more general and
reasonable approach for combining information (Gelman et al. 1995; Condon 2001).
UV Disinfection Guidance Manual
Proposal Draft
B-9
June 2003
-------
Appendix B. Derivation of UV Dose-Response Requirements
B.3.1 Model Description
The model used to relate UV dose to Cryptosporidium, Giardia, and virus log
inactivation is described by Equation B.I. Qian et al. (2003) provides a complete description of
the model and further statistical analyses.
Equation B.I
gamma(0.00 1,0.001)
/?- #(0,0.0001)
where
study
I(Cyt) =
Xii =
Log inactivation of the k* observation exposed to the j* UV dose level in the i*
Normal distribution with mean u, and precision T
Censor operator with Cyk as the estimated lower bound of the log inactivation
value for the k* observation exposing to the j* UV dose level in the i* study
j* dose level of study i,
Regression coefficient for study i
Integrated regression coefficient, combining information from all studies
When an observation is known to be greater than a value (right-censored), the reported
value is used as a lower bound value (Q*.). The prior distributions on precision (inverse of
variance), 11,2, are modeled using gamma(0.001, 0.001), which is considered "non-informative"
(the log variance is almost uniform). The prior distribution on ft is N(0, 0.0001), a practically
flat distribution.
One of the benefits of using a Bayesian modeling approach is it allows known
information that can better explain the data relationships to be incorporated into the model. In
this model, two known pieces of information were incorporated: (1) as UV dose increases the
number of microorganisms inactivated increases—incorporated by taking the exponential of (3j in
the second line of Equation B.I, which restricts the slope of the regression between log
inactivation and UV dose to a positive value; and (2) when the UV dose is zero, no
microorganism inactivation due to UV light occurs—incorporated by setting the intercept of the
regression line to zero (the second line in Equation B.I has no intercept term).
A Markov Chain Monte Carlo simulation method is used for estimating the model
parameters. To impute the censored data, an iterative procedure is used. At a given iteration, a
random sample of log inactivation is taken from a normal distribution with the mean and
variance calculated by the current estimates of Pi and tu- If the generated value is less than the
reported value (the lower bound), it is not used and a new value is generated until one that is
larger than the reported value is found. The model is then refitted with new estimates of PJ and
TI^. This process is repeated many times (200,000 in this case). Mathematical theories indicate
that the effect of a set of random initial values for all model coefficients and the censored values
UV Disinfection Guidance Manual B-10 June 2003
Proposal Draft
-------
Appendix B. Derivation of UV Dose-Response Requirements
will gradually disappear, and the samples will converge to their respective posterior marginal
distributions after a number of iterations. In this case, the first 140,000 iterations were discarded
and 1,000 samples for each of the unknown quantities (i.e., coefficients, predictions, and
censored values) were taken from the remaining 60,000 iterations. The computation is
implemented under WinBUGS (Spiegelhalter et al. 1996).
The Bayesian hierarchical model of Equation B.I estimates the integrated model
coefficients using the coefficient estimates from each study. As the model indicated, ft is
assumed to be the mean of the parent distribution of pj. This integration accounts for the
uncertainty of each study and "weights" each study accordingly.
B.3.2 Cryptosporidium and Giardia Modeled Results
The modeled results for Cryptosporidium and Giardia are shown graphically in Figures
B.4 and B.5, respectively. The graphs show the estimated regression for each study. The model
incorporates the coefficients from each study and calculates the predicted median and 80 percent
credible intervals, shown by the black solid line (median) and dark dotted lines (credible
intervals).
UV Disinfection Guidance Manual B-l 1 June 2003
Proposal Draft
-------
i
§
_
i
•
£
o
0)
I
I
"S
Q
03
x
"•a
•o
(0
I
Q
O
I
o
•^t
CQ
o
o
CO
= $
o
CN
LO
O
CO
0)
-------
8
I
I
en
5
>
8
o
cs
j«
S
£
'•5
£
o
I
TJ
£
0.
TJ
C
w
3
O
•o
j>
0)
TJ
O
,w
1
.5
(D
iq
CD
£
i-i
,:il.
1
i
I'
s
5
|2||flg§gw
sir
. • I i I
•ili!
>r>
CO
o
o>
o
-------
Appendix B. Derivation of UV Dose-Response Requirements
B.3.3 Virus Modeled Results
The model for the virus data is slightly different from the Cryptosporidium and Giardia
model described in Equation B.I. First, there were no censored data points; as a result, the term
/(Qt) is not included. Second, based on the data, a log transformation on the UV dose is not
necessary, i.e., the mean is modeled by Pi Xy. Figure B.6 displays the modeled results for
adenovirus.
UV Disinfection Guidance Manual
Proposal Draft
B-14
June 2003
-------
I
f*>
o
8
«J
_
1
0)
!5
TS
£
o
3
Q
"
to
oti
I
O)
iZ
o
in
CM
o
o
CM
O
10
0)
CO
o
o
o
o
o
in
8
6o~|
O
c
Dl
a,
-------
Appendix B. Derivation of UV Dose-Response Requirements
B.3.4 Calculating UV Dose Requirements from Modeled Results
Table B.2 presents the UV dose requirements for Cryptosporidium, Giardia, and viruses.
Each of the graphs presented in Figures B.4 through B.6 show the 80 percent credible interval.
The UV dose requirements for given log inactivatton levels were calculated from the fitted
model's lower bound of the credible interval (as called out in Figures B.4-B.6). Using the lower
bound means that at a given UV dose, the corresponding log inactivation is expected to be
achieved 90 percent of the time.
Table B.2 UV Dose Requirements for Inactivation of Cryptosporidium, Giardia
and Viruses During Validation Testing
Cryptosporidium
Giardia
Virus
Log Inactivation
0.5
1.6
1.5
39
1.0
2.5
2.1
58
1.5
3.9
3.0
79
2.0
5.8
5.2
100
2.5
8.5
7.7
121
3.0
12
11
143
3.5
-
-
163
4.0
-
-
186
B.4 References
Bukhari, Z., T.M. Hargy, J.R. Bolton, B. Dussert, and J.L. Clancy. 1999. Medium-pressure UV
for oocyst inactivation. JAWWA. 91(3):86-94.
Campbell, A.T. and P. Wallis. 2002. The effect of UV irradiation on human derived Giardia
lamblia cysts. Water Research. 36(4):963-969.
Clancy Environmental Consultants. 2000. UV Disinfection of Cryptosporidium andMS2
Coliphage in Catskill-Delaware Water. Report to Hazen and Sawyer.
Clancy, J.L., Z. Bukhari, T.M. Hargy, J.R. Bolton, B.W. Dussert, and M.M. Marshall. 2000.
Using UV to inactivate Cryptosporidium. Journal AWWA. 92(9):97-104.
Clancy, J. L., T. M. Hargy, D.A Battigelli, M. M. Marshall, D. Korich, and W.L. Nicholson.
2002. Susceptibility of Multiple Strains of C. parvum oocysts to UV Light. Final report to
AWWA Research Foundation And American Water Works Association.
Condon, P. 2001. Bayesian Statistical Modeling. Wiley, London.
Craik, S.A., G.R. Finch, J.R. Bolton, and M. Belosevic. 2000. Inactivation of Giardia muris
cysts using medium-pressure ultraviolet radiation in filtered drinking water. Water
Research. 34(18):4325-4332.
Craik, S.A., D. Weldon, G.R. Finch, J.R. Bolton, and M Belosevic. 2001. Inactivation of
Cryptosporidium parvum oocysts using medium- and low-pressure ultraviolet radiation.
Water Research. 35(6):1387-1398.
UV Disinfection Guidance Manual
Proposal Draft
B-16
June 2003
-------
Appendix B. Derivation of UV Dose-Response Requirements
Danielson, R.E., J. Phillips, and F. Soroushian. 2001. Inactivation ofGiardia muris, Bacillus
subtilis spores and MS2 coliphage by ultraviolet light. Proceedings IUVA Congress,
Washington D.C.
Gelman, A., J.B. Carlin, H.S. Stern, and D.B. Rubin. 1995. Bayesian Data Analysis. Chapman &
Hall. London.
Gerba, C.P. et al. 2000. Treatability Screening for CCL Microbial Contaminants - Interim
DRAFT Report to US EPA Office of Water.
Gilead, Z. and H.S. Ginsberg. 1966. Comparison of the rates of ultraviolet inactivation of the
capacity of Type 12 Adenovirus to infect cells and to induce T antigen formation. J.
Bacterial. 92(6): 1853-1854.
Hara, J. et al, 1990, Survival and Disinfection of adenovirus type 19 and enterovirus 70 in
ophthalmic practice. Jpn JopHthamol. 34: 421-427.
Hargy, T.M., J.L. Clancy, Z. Bukhari, and M.M. Marshall. 2000. Shedding UV Light on the
Cryptosporidium Threat. Journal of 'Environmental Health, 63: 19-22.
Hayes, S.L., C.D. Barnes, E.W. Rice, M.W. Ware, and F.W. Schaefer III. 2001. Low Pressure
Ultraviolet Studies for Inactivation for Giardia muris cysts, Proceedings IUVA Congress,
Washington D.C.
Hedges, L. 1997. Bayesian meta-analysis. In Statistical Analysis of Medical Data, eds B. Everitt
and I. Dunn. Arnold. London, pp. 251-275.
Kashinkunti, R.D., D.H. Metz, K.G. Linden, M.D. Sobsey, M. Moran, and A. Samuelson. 2002.
Microbial Inactivation Strategies for the Future: UV, Chlorine, and DBPs. Proceedings
WQTC 2002, Seattle, November 10-14, 2002.
Landis, H.E., Thompson, J.E., Robinson, J.P., Blatchley, E.R. 2000. Inactivation responses of
Cryptosporidium parvum to UV radiation and gamma radiation. Proceedings WQTC
2000, Salt Lake City, WA. November 5-9, 2000.
Linden, K.G., G-A. Shin, G. Faubert, W. Cairns, and M.D. Sobsey. 2002. UV disinfection of
Giardia lamblia in water. Environmental Science and Technology. 36(11):2519-2522.
Mackey, E.D., J.P. Malley, T.M. Hargy, and R.S. Cushing. 2000. MS-2 Bioassays and
Cryptosporidium Challenges: Comparing and Contrasting UV Reactor Validation
Techniques. Proceedings AWWA Water Quality Technical Conference, Salt Lake City,
UT.
Malley, J.P. 2000a. Reduction of Microbial and Chemical Pollution of Coastal Waters Using
Pulsed-UV Disinfection, Interim Report, NOAA.
Malley, J.P. 2000b. AWWARF Project 2593: "Inactivation of Pathogens by Innovative UV
Technologies" Second Quarterly Report.
UV Disinfection Guidance Manual B-17 June 2003
Proposal Draft
-------
Appendix B. Derivation of UV Dose-Response Requirements
Meng, Q.S. and C.P. Gerba. 1996. Comparative inactivation of enteric adenovirus, poliovirus
and coliphages by ultraviolet irradiation. Water Research. 30(ll):2665-2668.
Mofidi, A.A., H. Baribeau and J.F. Green. 1999. Inactivation of Cryptosporidium parvum with
polychromatic systems. Proceedings WQTC, Tampa, Florida, October 31- November 4,
1999.
Mofidi, A.A., E.A. Meyer, P.M. Wallis, C.I. Chou, Meyer, B.P., S. Ramalingam, and B.M.
Coffey. 2002. The effect of UV light on the inactivation ofGiardia lamblia and Giardia
muris cysts as determined by animal infectivity assay. Water Research. 36:2098-2108.
Oppenheitner, J., T. Gillogly, G. Stotarik, and R. Ward. 2002. Comparing the Efficiency of
Low and Medium Pressure UV Light for Inactivating Giardia muris and
Cryptosporidium parvum. Proceedings AWWA Annual Conference, New Orleans, LA.
Qian, S. S., M. Donnelly, D. C. Schmelling, M. Messner, K. G. Linden, arid C. Cotton. 2003.
Ultraviolet light inactivation of Cryptosporidium and Giardia in drinking water: A
Bayesian meta-analysis. Manuscript submitted to Water Research.
Shin, G-A, K. G. Linden, and M.D. Sobsey. 200 la. UV Disinfection of Adenovirus. WERF
Progress Report 6, 98-HHE-2.
Shin, G-A, K. G. Linden, M. Arrowood, and M.D. Sobsey. 200 Ib. Low pressure UV inactivation
and subsequent DNA repair potential of Cryptosporidium parvum oocysts. Appl. Environ.
Micro. 67(7):3029-3032.
Sommer, R., S. Appelt, W. Pribil, and T. Slifko. 2001. UV inactivation of C. parvum evaluated
in cell culture. Unpublished data, personal communication.
i.
Spiegelhalter, D.J., A. Thomas, A., and W.R. Best. 2000. WinBUGS Version 1.3: User Manual.
Medical Research Council, Biostatistics Unit, Institute of Public Health, Cambridge, UK.
Thompson, S.S., J.L. Jackson, M. Suva-Castillo, W.A. Yanko, Z. El Jack, J. Kuo, C.L. Chen,
P.P. Williams, and D.P. Schnurr. 2002. Detection of Infectious Human Adenoviruses in
Tertiary Treated and UV Disinfected Wastewater. Accepted for publication, Water r
Environment Research.
Thurston-Enriquez, J., C. Hass, J. Jacangelo, K. Riley, and C.P. Gerba. 2002. Inactivation of
Feline Calicivirus and Adenovirus Type 40 by Ultraviolet Irradiation. Proceedings, WEF
Disinfection 2002, St. Petersburg FL, February 17-20, 2002.
UV Disinfection Guidance Manual
Proposal Draft
B-18
, June 2003
-------
Appendix C. Validation of UV Reactors
To receive credit for Cryptosporidium, Giardia, or virus inactivation using UV light, the
Long Term 2 Enhanced Surface Water Treatment Rule (LT2ES WTR) requires systems to
demonstrate that the UV reactor can deliver the required dose through validation testing (40 CFR
141.729(d)). Furthermore, validation testing must determine a set of operating conditions that
can be monitored by the control system to ensure that the UV dose required for a given pathogen
inactivation credit is delivered during operation. At a minimum, these operating conditions must
include flowrate, UV intensity measured by a UV intensity sensor, and lamp status. The
validated operating conditions must account for the following factors (40 CFR 141, Subpart W,
Appendix D):
. • Lamp aging
• Lamp sleeve fouling
, . • UV transmittance of the water
. Inlet and outlet piping or channel configurations of the UV reactor
• Dose distributions arising from the velocity profiles through the reactor
• Failure of UV lamps or other critical system components
• . Measurement uncertainty of on-line sensors
Unless the State approves an alternative approach, validation testing must involve the
following components:
« Full-scale testing of a UV reactor, which conforms uniformly to the reactors used by
the system
« Inactivation of a test microorganism whose dose-response characteristics have been
quantified with a low-pressure (LP) mercury vapor lamp,
This appendix presents one approach for validating UV reactors. Other approaches or
modifications to this approach may be used at the discretion of the State. This appendix begins
with an overview of the approach for conducting validation testing and interpreting validation
results. This is followed by a description of the materials, equipment, and personnel used to
conduct validation testing and a description of the steps involved in validating UV reactors. The
appendix ends with descriptive examples showing how validation test results can be related to
inactivation credit.
Appendix F provides more detailed background information on validation testing and
includes several examples. .
U V Disinfection Guidance Manual C-1 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
C.1 Overview -
UV reactor validation should provide confidence that the UV reactor is appropriately
sized for a given disinfection application and should allow a water treatment plant (WTP) to
receive inactivation credit based on on-line measurements of flow, UV intensity, lamp status,
and, in some cases, UV transmittance (UVT) of the water at 254 nm. To ensure a UV reactor is
appropriately-sized for a given WTP, validation testing should provide data on dose delivery and
monitoring under design conditions of flow, UVT, and lamp output. This should be done either
by validating UV reactor performance under those conditions or by validating UV reactor
performance over a range of conditions that can be interpolated to obtain performance under
design conditions. To allow a WTP to obtain inactivation credit with UV disinfection, validation
testing should provide data relating on-line measurements of flow, UV intensity, lamp status, and
UVT to UV dose levels required to achieve target pathogen inactivation credit. This should be
done over the range of those on-line measurements expected with operation of the UV reactor at
the WTP.
UV manufacturers typically produce UV reactors as part of a product line where each
reactor is manufactured to the same specifications. If a representative UV reactor from that
product line undergoes validation testing, the test results can be applied to all other UV reactors
within that product line if those reactors are manufactured to the same specifications as the
validated reactor. If the design specifications of the product line that impact dose delivery and
monitoring change, this new UV reactor design must be re-validated.
C.1.1 Test Protocol
The validation protocol in this guidance document builds on well-established protocols
used in Europe and North America. A UV manufacturer, typically delivers a UV reactor to a test
facility. Test personnel inspect the UV reactor and document features of the design that impact
dose delivery and monitoring (e.g., reactor dimensions and sensor properties). The UV reactor is
installed within a biodosimetry test stand with inlet and outlet piping that should result in equal
or worse dose delivery than with the reactor installed at the WTP. The UV reactor is operated
under various test conditions of flow, UVT, and lamp power. The test condition of UVT is
typically obtained using a UV-absorbing compound injected into the flow upstream of the UV
reactor. A challenge microorganism is injected into the flow upstream of the UV reactor. The
concentration of viable challenge microorganisms is measured in samples collected at the
reactor's inlet and outlet. The results are used to calculate the log inactivation of the challenge
microorganism achieved by the UV reactor. The UV dose-response of the challenge
microorganism present in the inlet sample is measured using a bench-scale device termed a
collimated beam apparatus. The UV dose-response curve is used to relate the log inactivation
observed through the reactor to a UV dose value termed the Reduction Equivalent Dose (RED).
A safety factor is applied to the results to account for any bias and random uncertainty associated
with the validation of the UV reactor and the on-line monitoring approach used to indicate dose
delivery both during validation and during operation at the WTP. Last, a validation report is
prepared that describes the UV reactor tested, the test protocol, the test results, and the
inactivation credits that can be assigned to the UV reactor under given conditions of flow, UVT,
and lamp output. Figure C. 1 presents the organization of this validation protocol and the
sections within this appendix that address each of these issues.
UV Disinfection Guidance Manual C-2 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
Figure C.1 Elements of UV Reactor Validation
Provisions
for Testing
Section C.2.1
UV System
Documentation
Section C.2.2
Biodosimetry
Test Stand
Section C.3.1
Microbiology
Methods
Appendix D
Collimated Beam
Appendix E
UV Intensity
Sensor Test Stand
Section C.3,2 •
Third Party
Oversight
Section C.3.3
Examples
Section C.5
Develop Approved
Test Plan
Sections C.4.1
UV System Inspection
and Installation
Sections C.4.2-C.4.4
Evaluate Influent and
Effluent Mixing
Section C.4.5-C.4.6
Evaluate UV Sensors
Intensity
Section C.4.7
Evaluated Impact of
Lamp Aging
Section C.4.8
Biodosimetry
Section C.4.9
Determining Inactivation
Credit
Section C.4.10
Reporting
Section C.4.11
C.1.2 Relating RED to Target Pathogen Inactivation Credit
t
Chapter 1 (Table 1.4) presents the UV dose needed to achieve various inactivation credits
for Cryptosporidiwn, Giardia, and viruses. The dose values provided in Chapter 1 were
obtained by analyzing UV dose-response data measured using a bench-scale collimated beam
device. To account for variability in the dose-response of the pathogen, an 80 percent predictive
credible interval was used to determine dose values needed to achieve a given log inactivation of
the pathogen. The derivation of the UV dose requirements is presented in Appendix B. This
assessment, however, does not account for the measurement uncertainty associated with UV
reactor validation and on-line dose monitoring. To account for this uncertainty, the RED
UV Disinfection Guidance Manual C-3 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
measured during reactor validation should be equal to or greater than a target RED defined using
the following equation:
Equation C.I
where
REDT =
BRED =
Bpoly =
e =
DP =
Target RED that should be demonstrated during validation
RED bias
. Polychromatic bias
Expanded uncertainty expressed as a fraction
UV dose in Chapter 1 (Table 1.4) required for a given level of target, pathogen
inactivation credit.
The RED bias term accounts for the difference between the dose delivered to the target
pathogen and the dose measured using a challenge microorganism. If the challenge
microorganism is more resistant to UV light than the target pathogen, the RED measured during
validation will be greater than the dose delivered to the pathogen. The magnitude of the
difference will depend on the dose distribution of the UV reactor and the inactivation kinetics of
the challenge microorganism and the target pathogen. If the challenge microorganism is as
sensitive or more sensitive to UV light than the target pathogen, the RED bias has a value of
1.00. A recommended approach for obtaining the value of the RED bias is given in section
C.4.10.2.
The polychromatic bias term accounts for spectral differences in the lamp output, lamp
sleeve UV transmittance, UVT, and action spectrum of the challenge microorganism between
validation and operation of a UV reactor equipped with medium-pressure (MP) lamps. These
differences can cause the dose delivered at the WTP to differ from the dose measured during
validation. Depending on the spectral response and positioning of the UV intensity sensor and
the dose distribution of the UV reactor, the dose delivered at the WTP can be less than dose
measured during validation and indicated by the monitoring system. The polychromatic bias
term accounts for this issue. The polychromatic bias only applies to UV reactors that use
polychromatic UV lamps. With UV reactors using LP or low-pressure high-output (LPHO)
lamps, the polychromatic bias equals 1.00. A recommended approach for obtaining the value of
the polychromatic bias is given in section C.4.10.2.
The expanded uncertainty, e, accounts for the uncertainty in the measurements taken
during validation and used with dose delivery monitoring. In this protocol, the numeric value of
the expanded uncertainty is estimated using an 80 percent confidence level by summing the
individual measurement uncertainties associated with on-line sensors used in the field and during
validation, influent and effluent challenge microorganism concentrations, challenge
microorganism UV dose-response, and quantification of the UV output from the lamps. This
approach is described in section C.4.10.2.
Two approaches, termed Tier 1 and Tier 2, are presented in section C.4.10 for applying
the RED bias, polychromatic bias, and the expanded uncertainty to define the target RED values.
The Tier 1 approach, described in section C.4.10.1, is a standardized approach that uses
prescribed values for the RED bias, the polychromatic bias, and the expanded uncertainty to
UV Disinfection Guidance Manual C-4 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
define RED targets to be demonstrated during validation. To use the Tier 1 approach, the dose
monitoring and validation should meet defined criteria on reactor design, challenge
microorganism UV dose-response, UV absorber used during validation, sensor properties,
monitoring approach, and microbiology.
The Tier 2 approach, described in section C.4.10.2, allows the user to calculate the values
of the RED bias, the polychromatic bias, and the expanded uncertainty, and to use those values
to define the RED target to be demonstrated during validation. The approach does not prescribe
criteria for reactor design, challenge microorganism dose-response, the UV absorber used during
validation, sensor properties, monitoring approach, or microbiology.
C.1.3 Other Validation Protocols
Validation of UV reactors used in drinking water applications has been practiced in North
America and Europe using well-established protocols that include the following, shown in
chronological order of development:
. National Sanitation Foundation/American National Standards Institute (NSF/ANSI)
Standard 55
. Austrian Standards Institute (ONORM ; Osterreichisches Normungsinstitut) M 5873-
1
» German Association for Gas and Water (DVGW; Deutsche Vereinigung des Gas- und
Wasserfaches) W294
• National Water Research Institute/American Water Works Association Research
Foundation (NWRI/AwwaRF) UV Guidelines
UV validation conducted as per DVGW and ONORM demonstrates that a UV reactor
will deliver a RED of 40 mJ/cm2 measured using Bacillus subtilis spores. Validation as per these
protocols should meet criteria for the UV reactor and its validation. UV validation conducted as
per NWRI/AwwaRF Guidelines and NSF Standard 55 both use MS2 bacteriophage (MS2) as a
challenge microorganism. NSF standard 55 specifies a target RED of 40 mJ/cm2 while
NWRI/AwwaRF Guidelines does not specify a target RED. Validation testing as per .
NWRI/AwwaRF Guidelines and NSF Standard 55 should be assessed for consistency with the
guidance for test conditions provided in section C.4.9. Results should be interpreted as per the
guidance provided in sections C.4.9 and C.4.10.
C.1.4 Planning UV Validation
In general, validation testing will be conducted either for a UV manufacturer who wishes
to validate a given UV reactor for the drinking water market or for a utility that wishes to
validate a UV reactor for a specific application. Regardless of the end user, parties conducting
validation testing should develop a test plan that addresses the following questions:
UV Disinfection Guidance Manual C-5 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
. Where will validation take place?
• What test conditions of flow, UVT, arid lamp output should be tested?
. What UV absorbers and challenge microorganisms should be used?
« What inlet and outlet conditions will be used during validation?
• Who will provide the challenge stock solutions and assay water samples?
. How will UV intensity sensor properties during validation be verified?
« Who will conduct collimated beam testing?
• What is the expected safety factor and is it acceptable?
. Who will provide third party oversight?
• What State review and approval is needed for the test protocol?
When planning how validation testing will be done, utilities and manufacturers should
determine if they want to evaluate validation results under Tier 1 or Tier 2. They should assess if
the planned validation will meet the Tier 1 criteria and develop preliminary estimates of the
safety factor that would be applied under Tier 2. They should explore opportunities to optimize
validation testing by identifying approaches that minimize the values of the RED bias,
polychromatic bias, and expanded uncertainty terms used to determine the safety factor. To
provide flexibility in using Tier 1 and 2, one approach would be to ensure validation meets Tier 1
criteria and then to optimize for Tier 2.
C.1.4.1
UV Validation for Manufacturers
UV manufacturers will conduct validation either for a specific WTP or to allow broad
application of their UV reactor to many WTPs. If validation is being done to allow broad
application of the UV reactor, the test conditions of flowrate, UVT, and lamp output will likely
span a larger range than the test conditions that would be used when validating for a specific
WTP. The UV manufacturer may also validate the UV reactor for a range of dose targets that
allow the UV reactor to achieve credit for a range of pathogen log inactivation values. The
number of test conditions and dose targets chosen should be sufficient to allow interpolation of
validation data to conditions of flowrate, UVT, and lamp output specific for a given WTP
application.
For broad application of validation results, inlet and outlet conditions should be chosen to
provide a conservative yet practical representation of inlet/outlet piping used at WTPs. For
example, if the UV reactor is typically applied in a filter gallery, it may make sense to test with a
90 degree bend immediately upstream of the reactor to represent a "worst case" scenario. On the
other hand, if a UV reactor is typically installed with 5 or 10 pipe diameters of straight pipe
upstream of the reactor inlet, it may make sense to test with a 90-degree bend immediately
UV Disinfection Guidance Manual
Proposal Draft
C-6
June 2003
-------
Appendix C. Validation of UV Reactors
upstream of a 5 pipe diameters of straight pipe. UV manufacturers can use computational fluid
dynamics (CFD) as a tool to understand the impact of inlet and outlet conditions on the dose
delivery of their UV reactors in order to best identify the inlet and outlet conditions most
representative of a wide range of applications.
In order to facilitate regulatory approval in the States, validation testing should be
conducted using recognized and accepted protocols. Alternatively, the UV manufacturer should
solicit feedback and approval for the validation test plan from the State(s) before testing.
C.1.4.2 UV Validation for Utilities
Utilities have the option of validating UV reactors either at a UV test facility or on-site at
their WTP. Utilities considering on-site validation should address recommendations on water
quality, disposal, and test train requirements provided in section C.3.I. Potential issues include
obtaining water with a sufficiently high UVT that allows validation over the entire UVT range
expected at the WTP, providing sufficient mixing of additives prior to entering the UV reactor
and mixing of the challenge microorganisms after the reactor, and obtaining permits for the
disposal of the water used for validation. Utilities considering off-site validation at a test facility
should ensure that the inlet and outlet conditions used during validation are representative of
those conditions used at the WTP. Recommendations for inlet and outlet conditions to be used
during UV validation are provided in section C.3.1.5.
\ *.
If on-site validation is considered, the utility should identify who will provide
microbiological support for validation testing. The utility could use either their own
microbiological lab or a third party lab. Regardless of the approach, the microbiology lab should
have demonstrated experience working with the challenge microorganism and be able to provide
timely analysis of water samples collected during validation testing. Appendix D provides detail
on the microbiological lab qualifications and includes growth and assay methods for two
commonly used challenge microorganisms.
With on-site validation, the utility should also identify how it will verify the performance
of UV intensity sensors used during validation. Because utility staff typically do not have
experience in optoelectronic.instrumentation, they should use a third party laboratory to
benchmark sensor performance. Sections C.3.2 and C.4.7 describe the laboratory needs and the
measurements used to benchmark sensor performance.
C.2 UV Reactor
This section describes the hardware and documentation that the UV manufacturer should
provide to the validation facility.
C.2.1 Provisions for Testing
The UV manufacturer should provide for validation a UV reactor with the following
characteristics:
UV Disinfection Guidance Manual C-7 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
A UV reactor that matches the technical description in the documentation provided as
per section C.2.2.
UV lamps that have undergone appropriate burn-in. The recommended burn-in
period is 100 hours.
Lamps aged to give end-of-lamp-life conditions if the reactor is to be tested with aged
lamps. ,.
Provisions to reduce lamp output as per section C.4.9.4.
Provisions to measure the UV output of each lamp and the electrical power delivered
to the lamps as per section C.4.9.2.
On-line and reference UV intensity sensors that meet the technical description
provided in the documentation.
A safety cut-off switch to prevent overheating if LPHO or MP lamps are used.
C.2.2 UV Reactor Documentation
i
Prior to validation testing, the UV manufacturer should provide to the party conducting
the tests documentation identifying and describing the UV reactor. Documentation should
include all reactor and component information that impacts dose delivery and monitoring
including the following:
» Technical descriptions of the reactor and all internal components, including lamps,
sleeves, UV intensity sensors, baffles, and cleaning mechanisms. The technical
description should include dimensions and placement of all wetted components.
. Technical descriptions of the inlet and outlet piping to the reactor undergoing
validation, including the length and cross-sectional dimensions of any pipes,
channels, and bends, and dimensions of any hydraulic structures affecting flow. If
reactors are validated in series, technical descriptions of the piping between reactors
should be provided.
. Lamp specification stating the lamp manufacturer and product number, electrical
power rating, length from electrode to electrode, spectral output of new and aged
lamps, mercury content, and envelope diameter. The spectral output should be
specified for 5 nm intervals or less over a wavelength range that includes the response
range of the UV intensity sensors and the germicidal range.
• Sleeve specifications indicating sleeve dimensions, material, and UV transmittance •
from 200 to 400 nm.
« Technical description of the placement of the lamp within the sleeve.
UV Disinfection Guidance Manual
Proposal Draft
C-8
June 2003
-------
Appendix C. Validation of UV Reactors
Specifications for the reference and on-line UV intensity sensors indicating
manufacturer and product number, external dimensions, and measurement properties.
Measurement properties include spectral and angular response, working range and
linearity, calibration factor, temperature stability, long-term stability, and
measurement uncertainty. Data and calculations should be provided showing how the
total measurement uncertainty of the sensor is derived from the individual sensor
properties. Table C.I gives an example of the calculation of sensor measurement
uncertainty from the uncertainty that arises due to each sensor property.
Table C.1 Example of a UV Intensity Sensor Uncertainty Datasheet
Property
Calibration
Linearity
Temperature response
Angular response
Spectral response
Long term drift
Total Uncertainty1
Uncertainty (%)
8
5 .
3
5
1
10
15
Total uncertainty is calculated as the square root of the
sum of the squared individual uncertainties. In this example,
total uncertainty is (82+52+32+52-H 2+102}1C = 15%.
. Specifications for the UV intensity sensor port indicating all dimensions and
tolerances that impact the positioning of the sensor relative to the lamps.
. If the sensor port contains a monitoring window separate from the sensor,
specifications giving the window material, thickness, and UV transmittance from 200
to 400 nm should be provided.
• Technical description of the algorithm used by the reactor to monitor dose delivery,
including the use of sensors, signal processing, and calculations.
Documentation should also be provided on the proper installation and operation of the
reactor to ensure proper and safe validation testing, including:
» Floivrate, headloss, and pressure rating of the reactor
• Assembly and installation instructions
• Electrical requirements including required line frequency, voltage, amps, and power
• Operation and maintenance manuals that include cleaning procedures, required spare
parts, and safety requirements. Safety requirements should include information on
electrical lockouts, eye and skin protection from UV light, safe handling of lamps,
and mercury cleanup recommendations in the event of a lamp breakage
Lastly, the UV manufacturer should consider providing the following information
relevant to the test procedure:
UV Disinfection Guidance Manual
Proposal Draft
C-9
June 2003
-------
Appendix C. Validation of UV Reactors
Specifications for the challenge microorganism to be used during validation that
includes protocols required for growth and enumeration, expected UV dose-response,
and suitability for use in validation testing as discussed in section F.I .4.
Specifications for the UV absorber to be used during validation.
A description of the test conditions of flowrate, UVT, and lamp output used to
validate the reactor, and the expected measurements of UV intensity and challenge
microorganism RED.
C.3 Test Equipment, Facilities, and Personnel
This section describes the test equipment, laboratory facilities, and personnel that are
typically used during validation testing, including the following components:
. Biodosimetry test stand for measuring challenge microorganism inactivation by the
UV reactor
. UV intensity sensor test stand for measuring sensor properties
« Third party oversight
Appendix D provides information "on the microbiological laboratory with specific
information on the growth and assay of MS2 bacteriophage and B. subtilis spores. Appendix E
provides information on collimated beam apparatus used to measure the UV dose-response of the
challenge microorganism.
C.3.1 Biodosimetry Test Stand
The biodosimetry test stand is used to measure the inactivation of a challenge
microorganism by the UV reactor operating under controlled conditions of flowrate, UVT, and
lamp output.
Figure C.2 presents a block diagram of such a test stand with the following features:
• Water supply with rate-of-flow control and backflow prevention
• Dosing pumps and ports for injecting the challenge microorganism, the UV-absorbing
compound, and, if required, a disinfectant residual-quenching agent
• Influent-mixing device (static mixer or length of pipe) upstream of the reactor to
ensure the challenge microorganism 'and UV-absorbing compound are well-mixed
prior to entering the reactor
• Influent sampling port after the influent-mixing device and before the reactor
UV Disinfection Guidance Manual C-10 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
Inlet and outlet piping to the reactor that results in a dose delivery equal to or less
than the dose delivery expected with the installation of the reactor at a WTP
UV reactor under test
Ports to allow head-loss measurements across the UV reactor
Effluent-mixing device (static mixer or length of pipe) downstream of the reactor to
ensure that the challenge microorganisms that survive inactivation by the reactor are
well-mixed prior to sampling
Effluent sampling port after the effluent-mixing device
Water disposal facilities
Figure C.2 Block Diagram of the Biodosimetry Test Stand
Water
Supply
UV Challenge
Absorber Microbe
Pressure UV prfsure
Flow Gage Reactor Ga9e
meter ^ -• '
Static
Mixer
Valve
Backflow
Prevention
Influent Influent
Quenching Sample
Agent Port
Effluent
Sample
Port
C.3.1.1
Water Supply
Validation testing should prove that the monitoring of dose delivery by the UV reactor is
valid over the full range of UVT values expected with the application of the UV reactor at the
WTP. Typically, the UVT of the water supply used for validation is high and UV absorbing
chemicals are added upstream of the reactor to simulate different, lower UVTs over the test
range. For validation results to be generally applied to all WTPs, the water supply should have a
UVT at 254 nm greater than 97 percent (UV absorption coefficient less than 0.013 cm'1 with a
lOnm path length).
Whether coagulants are naturally present (e.g., reduced iron in ground water) or added as
part of water treatment, they can affect the challenge microorganism concentration, the turbidity,
and the UVT of water samples collected during reactor validation (Petri et al. 2000).
Coagulation of the challenge microorganism can lead to reduced counts and poor sample-to-
sample repeatability. To avoid these effects, the water supply should not contain coagulants that
interfere with the validation results. Alternatively, chelating agents or coffee can be used as an
additive to counter these effects (Petri et al. 2000).
'UV Disinfection Guidance Manual C-l 1 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
The water passing through the reactor should not contain disinfectant residuals that
inactivate the challenge microorganism during testing. If the water does contain a disinfectant
residual, a quenching agent should be injected into the water upstream of the microorganism
injection port. The quenching agent should have a minimal impact on the UVT.
The water supply (volume and flowrate) should be sufficient to allow testing over the
rated flow range of the UV reactor. A flow-control device (e.g., variable speed pump or valve)
can be used to vary the flow over that range. A flowmeter with a known measurement
uncertainty should monitor the flowrate through the UV reactor.
Backflow prevention should be used with a potable water supply. Backflow prevention
can be obtained using reduced pressure zone (RPZ) backflow preventers, air gaps, or check
valves.
C.3.1.2 Dosing of Additives
Challenge microorganisms, UV-absorbing compounds, and possibly disinfectant
quenching agents may be injected into the flow upstream of the UV reactor during validation. If
pumps are used to inject the additives, they should provide a pulseless flowrate or have a cycle
time an order of magnitude less than the residence time of the reactor. The flowrate generated by
the pump should be stable over the time required to take samples as per section C.4.9.5. An
injection port using standardized injector technologies can be used to disperse the additives into
the flow.
C.3.1.3 Mixing of Reactor Influent and Effluent
Additives passed through the reactor should be well-mixed through the cross-section of
the pipe prior to the reactor influent sampling port. The challenge microorganisms surviving UV
disinfection should be well-mixed through the pipe cross-section prior to the reactor effluent
sampling port. Mixing can be achieved either using static mixers or by relying on the turbulent
mixing present in the lengths of pipe upstream of the sampling ports. If the water passed through
the UV reactor is obtained from a large tank, the additives can be premixed in the tank to obtain
a uniform concentration for testing.
C.3.1.4 Sample Taps
The sample taps should be located to provide representative samples of undisinfected
water entering the reactor and the disinfected water leaving the UV reactor. If the influent
sample tap is located too close to the reactor influent, the samples collected may be exposed to
UV light, resulting in underestimation of the influent concentration of the challenge
microorganism. If the effluent sample tap is located too close to the reactor effluent, the effluent
samples will be collected before full exposure to UV light and the effluent concentration of the
challenge microorganism will be overestimated. The UVT of the water can be used to calculate
how far UV light from the reactor penetrates the water upstream and downstream of the reactor.
The sampling points should be located far enough from the UV reactor that the germicidal UV
UV Disinfection Guidance Manual C-12 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
intensity at the point of sampling is less than 0.1 percent of the germicida! intensity within the
UV reactor.
Sample taps may sample from a single point within the flow or from multiple points at
the same time. Samples taken from multiple points within the flow should have the same
concentration of additives and microorganisms within the measurement error.
Sampling taps should remain open over the duration of the test. Sample collection should
meet standards of good practice as defined by Standard Methods Section 9060 (APHA et al.
1995). Samples should be collected in bottles that have been cleaned and sterilized. Samples
collected should be immediately stored on ice within a cooler in the dark until needed for
analysis.
C.3.1.5 UV Reactor Inlet and Outlet Conditions •
As stated previously, the inlet and outlet structures to the UV reactor during validation
should result in equal or worse dose delivery than with the reactor installed at the WTP. EPA
recommends using any one or combination of the following approaches:
• Inlet and outlet conditions used at the WTP match those used during validation for at
least 10 pipe diameters upstream and 5 pipe diameters downstream of the reactor.
. UV reactor is validated either with a 90-degree bend immediately upstream of the
reactor inlet or a with 90-degree bend followed by a length of straight pipe
immediately upstream of the reactor inlet. The reactor is installed at the WTP with a
length of straight pipe immediately upstream of the reactor equal to 5 pipe diameters
plus any length used after the 90-degree bend during validation. To avoid jetting
effects, piping upstream of the straight pipe length should not have expansions for at
least 10 pipe diameters and any valves located in that length of pipe should always be
fully open'during operation of the reactor. With this approach, it is assumed that the
90-degree bend immediately upstream of the reactor inlet provides worse hydraulics
than the installation. This approach assumes that the reactor design has not been
optimized for the 90-degree bend inlet.
• Velocity of the water measured at evenly-spaced points through a given cross section
of the flow upstream and downstream of the reactor is within 20 percent of the
theoretical velocity with both the validation test stand and the installation. The
theoretical velocity is defined as the flowrate divided by the cross-sectional area.
CFD-based dose modeling can be used, in tandem with one of the above-mentioned
approaches, to show that dose delivery with the installation is better than dose delivery during
validation for given conditions of flowrate, UVT, and lamp output. To account for uncertainty in
CFD predictions of dose delivery (Petri and Olson 2001, Wright and Hargreaves 2002), CFD
predictions of dose delivery during validation should be at least 20 percent greater than
predictions of dose delivery at the WTP.
UV Disinfection Guidance Manual C-13 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
C.3.1.6 Quality Assurance and Quality Control
Flowmeters, injection pumps, pressure gauges, and other measuring devices used should
bear evidence of being in calibration. Accuracy of instrumentation should be checked by
comparison with standard measurements. The documentation describing the test facility should
be provided and verified including the following items:
. A description of the validation test stand, including all piping, valves, flowmeters,
mixers, pumps, sampling locations, and measurement instrumentation
• The measurement uncertainty and the last calibration date of all measurement
instrumentation
« Comparisons of on-line instrumentation with standard measurements
C.3.2 UV Intensity Sensor Test Stand
The properties of the on-line and reference UV intensity sensors should be measured by
an independent laboratory that is equipped to confirm sensor calibration and measure the
sensor's angular and spectral response, linearity over the working range, and temperature
response. Measurements should be National Institute of Standards and Technology (NIST)
traceable or equivalent with quantified measurement uncertainties. Personnel who test UV
intensity sensors should be qualified to undertake optical testing, understand the test protocols
for the sensors as provided by the manufacturer, and be aware of all safety requirements
associated with UV-irradiation devices.
C.3.3 Third-Party Oversight
Validation of UV reactors and their components should be conducted at facilities and by
personnel that are acceptable to the State. At a minimum, personnel independent of the
manufacturer of the UV reactor should oversee validation testing. A registered professional
engineer with knowledge and experience in testing and evaluating UV reactors should witness
the validation testing to verify that the documented validation protocol was followed and the
reported data and results are accurate. The engineer should be responsible for supervising the
preparation of the engineering report on validation testing and should review and approve that
report prior to its release. The engineer should not have a personal stake in the outcome of the
validation testing or any conflict of interest with respect to the ultimate use of the UV reactor
being tested. Where necessary, the engineer should use other third parties to provide expert
opinion on various aspects of UV validation testing.
C.4 Testing
This section describes the recommended steps for validating the UV reactor provided by
the UV manufacturer. At the discretion of the State, variations or alternatives to the procedures
or steps may be accepted.
UV Disinfection Guidance Manual " C-14 • June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
C.4.1 Develop Approved Test Plan
The first step in validating a UV reactor should be the development and review of a test
plan. The test plan should be developed with input and approval from the utility, manufacturer,
third party oversight, and the State. The test plan should resolve the questions identified in
section C.I.4,
C.4.2 UV Reactor Inspection
Prior to installing the UV reactor in the biodosimetry test stand, the UV reactor should be
inspected to confirm that it matches the descriptions and dimensions provided in the
manufacturer's documentation as described in section C.2.2.
0.4.3 UV Reactor Installation
The UV reactor and its inlet and outlet piping should be installed at the test facility in
accordance with the manufacturer's installation and assembly instructions. If reactors are
installed in series, the piping between the reactors should conform to specifications provided by
the UV reactor manufacturer. The piping should be inspected to ensure compliance with the
manufacturer's documentation.
C.4.4 Headless and Integrity Evaluation
The physical integrity of the UV reactor and the test train should be checked before
conducting further testing. Personnel who operate the UV reactor during all tests should be
familiar with its operation and maintenance manual and with any safety requirements.
Procedure
1. Pass water through the reactor at the minimum and maximum flowrates.
2. Measure and record the headloss across the reactor at each flowrate.
3. On completion of the test, visually inspect the sleeves, UV intensity sensors, and/or
monitoring windows for mechanical integrity.
4. If the headloss across the reactor exceeds specifications provided by the manufacturer, or
if component integrity has been compromised, investigate the cause and resolve the issue
before further testing.
CAS Evaluation of the Mixing of Additives
The mixing of the UV-absorbing chemical and the challenge microorganism prior to
entering the UV reactor should be confirmed. Mixing can be confirmed by comparing the UV
absorbance of the water at 254 nm (A254) of samples collected at the influent and effluent
UV Disinfection Guidance Manual ' C-15 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
sampling ports using the following procedure. This test should not be necessary if a static mixer
is used between the injection port and the reactor entrance and theflowrate through the static
mixer meets manufacturer specifications.
Procedure
1. Prepare a stock solution of the UV-absorbing compound.
2. Pass water through the reactor at the minimum flowrate.
3. Inject sufficient UV-absorbing compound into the flow of water passing through the
reactor to give a UVT less than the minimum that will be used during challenge
testing.
4. Collect water samples from the influent and effluent sampling ports at 1-minute
intervals and measure the UVT. The sample volume should be less than 5 mL and
collected over a time not exceeding 2 seconds.
5. Calculate the A254 from the measured UVT. Mixing of the injected compounds
should be sufficient if the average Aj54 of the influent samples and the average A254
of the effluent samples agree within 2 percent and the standard deviation of each is
less than 5 percent. If these conditions are not met, the mixing between the injection
port and the influent sampling port should be increased and retested.
C.4.6 Evaluation of the Mixing of Surviving Microorganisms
Mixing of the surviving challenge microorganisms leaving the UV reactor should be
confirmed. Mixing can be confirmed by comparing the challenge microorganism concentration
of samples collected at the effluent sampling port and a sampling port downstream of the effluent
sampling port using the following procedure. This test should not be necessary if a static mixer
is located between the reactor exit and the effluent sampling port and theflowrate through the
static mixer meets manufacturer specifications.
Procedure
1. Prepare a stock solution of the challenge microorganism and a stock solution of the
UV- absorbing compound.
2. Pass water through the reactor at the minimum flowrate that will be used during
challenge testing.
3. Operate the UV reactor with the lamps power set at 100 percent.
4. Inject the challenge microorganism into the water flowing through the reactor.
UV Disinfection Guidance Manual
Proposal Draft
C-16
June 2003
-------
Appendix C. Validation of UV Reactors
5. Collect at least three UV-disinfected samples spaced -1 minute apart from the effluent
sampling point and from a location at least 5 pipe diameters downstream of the
effluent sampling point.
6. Measure the concentration of the challenge microorganism in each sample in
triplicate.
7. If the concentration in the effluent samples is below the detection limit, repeat steps 2
to 6 with the UV absorber injected into the flow to reduce the dose delivery by the
reactor.
8. Repeat steps 3 to 7, passing the water through the reactor at the minimum flowrate
that will be used during the challenge test.
9. The mixing should be sufficient if there is no statistical difference at a 95 percent
confidence level between the geometric means of the samples collected from the two
effluent sample points. If statistical differences are observed, the mixing between the
reactor and the effluent sampling port should be increased and the test repeated.
C.4.7 UV Intensity Sensor Evaluation
The measurement uncertainty of the UV intensity sensors used on the UV reactors should
be confirmed. This may be achieved either by comparing the UV intensity sensor measurements
made on the reactor to a reference measurement, or by measuring the properties of the sensors
using a UV intensity sensor test stand. The following sections discuss each of these approaches.
C.4.7,1 Assessing Uncertainty Using Reference Sensors
If the measurement uncertainty of the reference intensity sensor is known, the following
procedure can be used to check the uncertainty of the UV intensity sensors used during
validation.
Procedure
1. Pass water through the reactor without the addition of UV-absorbing chemicals.
2. Using at least three recently calibrated reference sensors, install each sensor on the UV
reactor at each port and record the measured UV intensity. Repeat using each duty
sensor. If the sensors can be rotated, then measure the minimum and maximum sensor
readings with rotation.
3. Record the water temperature as an indicator of the operating temperature of the sensors.
4. Repeat the test with the UVT decreased to the minimum value expected during testing.
UV Disinfection Guidance Manual C-I7 - June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
5., For a given lamp output and UVT, the difference between the reference sensor
measurements should follow Equation C.2:
Refl
-1
LRef2
:ioo<(af
Ref2
Equation C.2
Intensity measured by a reference sensor designated by the subscript
Measurement uncertainty of reference sensor designated by the subscript
where
I
6. For a given lamp output and UVT, the difference between the reference and duty sensor
measurements should follow Equation C.3:
Duty
-1
Ref
4,
Equation C.3
where
iRef
I Duty
ORef
O~Duty
Intensity measured by the reference sensor
Intensity measured by the duty sensor
Measurement uncertainty of the reference sensor (%)
Measurement uncertainty of the duty sensor (%)
7. UV intensity sensors that do not meet these criteria should be replaced. Alternatively, the
UV manufacturer can re-evaluate their stated measurement uncertainty and use a higher
value.
C.4.7.2 Assessing Uncertainty Using a Sensor Test Stand
The measurement uncertainty of the UV intensity sensors can be assessed by a laboratory
capable of confirming sensor calibration and properties with a known measurement uncertainty.
The laboratory should measure linearity, spectral and angular response, and temperature
response. Results should be used to calculate the measurement uncertainty. Sensors that do not
meet manufacturer specifications should be replaced. Alternatively, the UV manufacturer can
re-evaluate their stated measurement uncertainty and use a higher value.
0.4.8 Evaluation of Lamp and Sleeve Aging on Dose Monitoring
With operation over time, UV lamps and sleeves can experience non-uniform aging along
their length and around their circumference. Lamps can also experience spectral shifts in output
and sleeves can experience spectral shifts in UV transmittance. If these effects have a significant
impact on how the dose delivery indicated by the monitoring system compares to the delivered
dose, validation should be conducted using both new and aged lamps and sleeves. The following
procedure compares dose delivery monitoring with new and aged lamps to identify if validation
U V Disinfection Guidance Manual C-18 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
should be conducted with both new and aged lamps and sleeves. Alternatively, data on the UV
output of new and aged lamps and the UV transmittance of new and aged sleeves can be
compared and used to demonstrate if validation should be conducted with new and aged lamps
and sleeves. In both approaches, an aged lamp or sleeve is one that has reached the end of its
useful service life.
Procedure ' • ' -
1. Prepare a stock solution of the challenge microorganism.
2. Fit the UV reactor with aged lamps and sleeves.
3. Pass water through the reactor at a constant UVT and at the maximum flowrate that will
be used during challenge testing.
4. Operate the UV reactor at peak lamp power.
5. Inject the challenge microorganism into the flow passing through the reactor.
6. Collect at least three microbiological samples spaced one minute apart from the influent
and effluent sampling ports.
7. Record the UV intensity sensor measurements.
8. Fit the UV reactor with new lamps that have undergone 100-hour burn-in and new
sleeves.
9. Lower the lamp power to give a UV intensity sensor reading equivalent to the reading
obtained in step 7.
10. Repeat steps 5 and 6.
II. If the mean log inactivation achieved with new lamps differs from the mean log
inactivation achieved with aged lamps, lamp aging impacts the relationship between dose
delivery and UV intensity sensor reading, and validation with aged lamps and sleeves
should be considered.
C.4.9 Dose Delivery Validation
Dose delivery validation via biodosimetry provides an assessment of dose delivery and
monitoring by the UV reactor under specific conditions of flowrate, UVT, and lamp output.
C.4.9.1 Preparation of Challenge Microorganism Stock Solution
The challenge microorganism is used to measure the dose delivery of the UV reactor
during validation. Because MS2 and B. subtilis spores are typically used, methods for their
UV Disinfection Guidance Manual C-19 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
preparation and assay are provided in this manual in Appendix D. Other peer-reviewed methods
may be used. A rationale for selecting challenge microorganisms other than MS2 and B. subtilis
spores is provided in section F. 1,
The challenge microorganism stock solution should be prepared in accordance with peer-
reviewed methods. The source of the challenge microorganism, the source of the host (if used),
a description of all media used, the steps involved in propagating the challenge microorganism,
and the steps involved in purifying the challenge microorganism to create a mono-disperse stock
solution should be documented. The volume of stock solution needed should be estimated prior
to testing based on the test plan and the expected stock concentration.
C.4.9.2 Reactor Preparation
If the number of sensors is less than the number of lamps, the UV intensity sensors
should be directly monitoring the lamps with the highest output and those lamps should be the
closest lamps to the sensor. The lamps with the highest output can be identified by taking
measurements using either a dedicated test stand or the UV reactor. One approach for using the
UV reactor is described below. This preparation should not be necessary if the UV reactor has
one UV intensity sensor per lamp.
Procedure
1. Install a lamp within a lamp sleeve located near one of the reactor's UV intensity
sensors.
2. Pass water through the reactor at a constant flowrate and UVT.
3. With only the lamp under evaluation on, record the measured UV intensity.
4. Repeat the test for each lamp and rank the results.
5. Install the lamps in the UV reactor so that the lamps with the highest output are
closest to the UV intensity sensors monitoring those lamps.
C.4.9.3
Flowrates
At a minimum, the reactor should be validated at the minimum and maximum flowrates
as defined by the UV manufacturer. Other flowrates within that range can be tested. For
interpolation of validation results as a function of flowrate, a recommended approach for
selecting intermediate flowrates is to approximate a geometric series using Equation C.4:
UV Disinfection Guidance Manual
Proposal Draft
C-20
June 2003
-------
Appendix C. Validation of UV Reactors
Qn ~ QM^P " Equation C.4
where
Qn = nth flowrate to be tested
Qmax - Maximum flowrate to be tested
3 = Rate term with a recommended value between 1.5 and 2
n = Number of flowrates to be tested
The value of p should not exceed 2 and should be sufficient to obtain at least three measured data
points for interpolation.
Example. Interpolation will be used to predict RED as a function of flowrate for a UV
reactor rated over a flow range of 2 to 20 mgd. If a rate term of 2 was used with Equation C.4,
the UV reactor would be validated at flowrates of 20,10, 5, 2.5, and 2 mgd. If a rate term of 1.5
was used with Equation C,4, the UV reactor would be validated at flowrates of 20, 13, 8.9, 5.9,
4.0,2.6, and 2 mgd.
C.4.9.4 Lamp Power and UV Transmittance
At a given flowrate, the UV reactor should be validated under conditions of UVT and
lamp output that demonstrate the UV reactor is sized to deliver a given dose and the UV
reactor's dose monitoring system provides a valid measure of that dose. Typically, the UVT of
the source water used during validation is high and UV absorbing chemicals are added to that
water to achieve the UVT used during validation testing. Different levels of lamp output can be
obtained using one or more of the following approaches:
• Using new and aged lamps
• Using different lamp types with the same spectral output (e.g., using LP and LPHO
lamps)
. Changing the ballasts' power settings
• Using specially modified ballasts capable of operating at different power levels
. Changing the supply voltage to the lamp ballasts
If lamp aging affects the relationship between the inactivation achieved by the UV
reactor and the measurements made by the on-line UV intensity sensor, aged lamps should be
used when validation testing involves reduced lamp output.
The conditions of lamp power and UVT used during validation should depend on the
monitoring approach of the UV reactor. The next three sections describe recommended
approaches for defining these test conditions for UV reactors that use the following monitoring
approaches:
UV Disinfection Guidance Manual C-21 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
» UV intensity setpoint approach
• UV intensity and UVT setpoint approach
. Calculated dose approach
Section F.2 provides background on the development of these approaches.
UV Intensity Setpoint Approach
With the UV intensity setpoint approach, measurements of UV intensity and flowrate are
used directly to indicate dose delivery. Dose delivery at or above a given level is indicated when
the measured intensity reads above an alarm setpoint value defined as a function of flowrate.
With the UV intensity setpoint approach, the UV intensity sensor is positioned within the
UV reactor to respond to the impacts of both lamp output and UVT. As such, dose delivery can
be monitored without the need to measure the UVT.
Strategies for implementing this approach include:
1. Using a single UV intensity setpoint value from minimum to maximum flow to verify
dose delivery at some minimum level.
Example. A UV intensity setpoint of 10 mW/cm2 is used to verify a minimum
MS2 RED of 40 mJ/cm2 from 1 to 5 mgd.
2. Several UV intensity setpoint values are used, each one applying over a specific flow
range.
Example. UV intensity setpoints of 10 and 20 mW/cm2 are used to verify a
minimum MS2 RED of 40 mJ/cm2 from 1 to 2.5 mgd and from 2.5 to 5 mgd,
respectively.
3. UV intensity setpoint values are interpolated as a function of flowrate.
Example. UV intensity setpoints defined by the following equation are used to
indicate an MS2 dose of 39 mJ/cm2 from 1 to 2.4 mgd:
Intensity setpoint (mW/cm2) = 15.6 x flow rate (mgd) + 3.9
4. UV intensity setpoints are defined as a function of flowrate for multiple levels of dose
delivery.
Example. A UV intensity setpoint of 10 mJ/cm2 is used to verify a minimum
RED of 40 mJ/cm2 from 1 to 5 mgd. A UV intensity setpoint value of 7 mW/cm2
is used to verify a minimum RED of 25 mW/cm2 from 1 to 5 mgd.
UV Disinfection Guidance Manual C-22 ' June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
With UV reactors using this monitoring approach, validation testing provides data on the
relationship between dose delivery and measured intensity at a given flowrate. Dose delivery at
a given flowrate and UV intensity is measured under two conditions of lamp power and UVT,
described as follows:
1. Lamps at peak power and the UVT decreased to give a UV intensity sensor reading at
a setppint value."
. High UVT and the lamp power lowered to give a UV intensity sensor reading at a
setpoint value.
The RED assigned to the reactor is the lower value observed between the two test
nn<:
conditions.
If the lamp power cannot be sufficiently lowered to obtain a UV intensity sensor reading
at the setpoint value, an alternative to the second test condition is to test with the lowest possible
lamp power setting and the UVT decreased until an intensity reading at the setpoint is obtained.
This alternative second test condition is acceptable if the following conditions are met:
. The adjusted lamp power results in a lamp output equal to or lower than the lamp
output used for sizing the UV reactor for a WTP. The lamp output used for sizing the
UV reactor is the product of the lamp-aging factor and the fouling factor.
« The RED measured with the second condition is equal to or greater than the RED
measured with the first test condition or the UVT with the second test condition is
less than the UVT expected at the WTP.
There are several approaches for defining the UV intensity setpoint values evaluated
during validation testing:
1. If a UV reactor is being validated for an application with specific design conditions of
flowrate, lamp output, and UVT, the intensity setpoint at design flow is equal to or
greater than the intensity reading obtained with the reactor operating under these
design conditions.
2. A UV reactor manufacturer can usually provide model estimates of dose delivery as a
function of flowrate and UV intensity. The model estimates would be used to define
the intensity setpoint values associated with a target dose delivery. Since model
estimates may not be accurate, trial and error testing may be used to establish the
optimal intensity setpoint necessary for a target level of dose delivery. Alternatively,
testing can be used to define the relation between dose delivery and measured
intensity, and interpolation can be used to define the optimal setpoint associated with
a target dose delivery.
During the validation of a UV reactor using the intensity setpoint monitoring approach,
the UVT used will likely be less than the design UVT and the lamp output will be less than the
design lamp output. While it may appear that these test conditions are more stringent than the
design conditions, it should be recognized that design conditions do not represent the worst-case
UV Disinfection Guidance Manual C-23 June 2003
Proposa/ Draft
-------
Appendix C. Validation of UV Reactors
conditions that can occur at a WTP. For example, lamps can age below their expected end-of-
life output, lamp sleeves can foul internally, wiper mechanisms can fail, and dose-pacing
strategies can reduce lamp output. These factors in combination can result in a UV output well
below the design output. If the design UVT is selected at a 95 percent confidence level, then a
UVT below the design value is expected 5 percent of the time. Because intensity setpoints
should provide a valid measure of dose delivery, regardless of the combination of lamp output
and UVT values, a UV reactor using intensity setpoint monitoring should be validated over the
full range of conditions giving rise to the setpoint, even if they exceed design conditions.
Example. A UV reactor that uses the intensity setpoint approach for monitoring is sized
using a design UVT of 90 percent, a lamp aging/fouling factor of 70 percent, and a flow of 5
mgd. With lamp power and UVT adjusted to 70 and 90 percent, respectively, the UV intensity
sensor reads 14 mW/cm2. The UV reactor is tested at a flow of 5 mgd using the following
conditions of lamp output and UVT that give rise to a UV intensity of 14 mW/cm2:
• 100 percent lamp power, 87 percent UVT
• 27 percent lamp power, 98 percent UVT
By testing the reactor using these conditions, dose delivery associated with a setpoint of
•14 mW/cm2 is validated.
UV Intensity and UVT Setpoint Approach
With the UV intensity/UVT setpoint approach, measurements of UV intensity, UVT, and
flowrate are used to indicate dose delivery. Dose delivery at or above a given level is indicated
when both the measured UV intensity and UVT read above their respective alarm setpoint
values. Strategies for implementing this approach include:
1. Using a single UV intensity setpoint value and UVT setpoint value from minimum to
maximum flowrate to indicate dose delivery at some level.
Example. A minimum MS2 RED of 40 ml/cm2 from 1 to 5 mgd is verified when
the measured UV intensity is equal to or greater than 10 mW/cm2 and the
measured UVT is equal to or greater than 85 percent.
2. Several sets of UV intensity and UVT setpoint values are used, each set applying over
a specific flow range.
Example. For an MS2 RED of 40 mJ/cm2, a UV intensity setpoint value of 10
mW/cm2 and,a UVT setpoint value of 80 percent are used from 1 to 2.5 mgd. A
UV intensity setpoint value of 20 mW/cm and a UVT setpoint value of 85
percent are used from 2.5 to 5 mgd.
3. Sets of UV intensity and UVT setpoint values are interpolated as a function of
flowrate.
UV Disinfection Guidance Manual
Proposal Draft
C-24
June 2003
-------
Appendix C. Validation of UV Reactors
4. Sets of UV intensity and UVT setpoint values are defined as a function of flowrate
for multiple levels of dose delivery.
With UV reactors using this monitoring approach, validation testing provides data on the
dose delivery with the reactor operating at the setpoint values and proof that the sensor is
appropriately positioned for this monitoring approach. As such, each set of UV intensity and
UVT setpoints should be tested using two conditions as follows:
1. UVT decreased to give a reading at the UVT setpoint followed by a decrease in lamp
power to give a UV intensity sensor reading at the UV intensity setpoint.
2. Lamp power at 100 percent and UVT decreased to give a UV intensity sensor reading
at the intensity setpoint.
The first condition provides data on dose delivery with the reactor operating with UV
intensity and UVT at the setpoint values. The second condition provides data on the positioning
of the UV intensity sensor. If the RED measured with the second test condition is greater than
the RED measured with the first, the UV intensity sensor is not appropriately positioned for this
monitoring strategy and this monitoring strategy cannot be used (see section F.2 for a rational for
this criteria).
There are several approaches for defining the UV intensity and UVT setpoints used
during validation testing.
1. At design flow, the UVT setpoint is the design UVT. The intensity setpoint is the UV
intensity measured with the lamp output and UVT adjusted to their design values.
2. At other flowrates, model estimates of dose as a function of UVT and lamp output
can be used to identify the setpoint values that will be assessed during validation
testing. Trial and error testing or interpolation of test results can be used to refine and
optimize those values for a given target dose delivery.
Example. A UV reactor that uses the UV intensity and UVT setpoint approach for
monitoring is sized for a WTP using a design UVT of 90 percent, a design lamp fouling/aging
factor of 70 percent, and a design flowrate of 5 mgd. Operating under those conditions, the
intensity sensor measures 14 mW/cm2. The UV reactor is validated under two test conditions at
a flowrate of 5 mgd:
. 70 percent lamp power and 90 percent UVT resulting in a UV intensity reading of 14
mW/cm2
• . 100 percent lamp power and 75 percent UVT resulting in a UV intensity reading of
14mW/cm2
The first condition provides data on the dose delivery of the reactor operating at the
setpoint. The second condition provides data to assess the positioning of the UV intensity
sensor.
UV Disinfection Guidance Manual C-25 June 2003
Proposal Draft
-------
Appendix C. Validation of UVReactors
Calculated Dose Approach
With the calculated dose approach, dose delivery is calculated from measurements of UV
intensity, UVT, and flowrate using an algorithm developed by the UV reactor manufacturer. For
UV reactors that use this approach, the UV reactor should be tested over a range of combinations
of flowrate, UVT, and lamp power that result in a given calculated dose. At a given flowrate,
that range should include the following combinations:
. Maximum power and decreased UVT
• Maximum UVT and decreased lamp power
« One or two intermediate combinations of UVT and lamp power
If the algorithm for calculating dose accounts for lamps operating at different power
levels or specific lamps operating either on or off, test conditions should include combinations of
these conditions.
Example. A UV reactor that uses a calculated dose for compliance will be used at a
WTP with a design UVT of 90 percent, a design lamp fouling/aging factor of 70 percent, and a
design flowrate of 5 mgd. The target RED is 40 mJ/cm2. At 5 mgd, test conditions that result in
a calculated dose of 40 mJ/cm2 by the monitoring system are as follows:
• 100 percent lamp power, 80 percent UVT
• 5 8 percent lamp power, 90 percent UVT
• 34 percent lamp power, 98 percent UVT
C.4.9.5 Measuring Challenge Microorganism Inactivation by the UV
Reactor
The reactor should be operated at each of the test conditions of flowrate, UVT, and lamp
power in accordance with sections C.4.9.3 and C.4.9.4. Prior to sampling, steady-state
conditions should be confirmed by monitoring the UV intensity sensor measurements and the
UVT. The challenge microorganism should be injected into the flow upstream of the reactor and
well-mixed prior to its entering the UV reactor. At least three influent and effluent samples
should be collected for each test condition. The time interval between sample collections should
be greater than or equal to the residence time between the inlet and outlet sampling ports. Water
samples should be collected by personnel who are familiar with good sampling practices as
specified in Standard Methods (APHA et al. 1995) and the guidance for collecting UV-irradiated
samples. Sample volumes should be sufficient for assessing the challenge microorganism
concentrations in the influent and effluent.
Before and after the samples are collected, the flowrate through the reactor, all UV
intensity sensor measurements, on-line UVT measurements, and any calculated dose values
should be measured and recorded. With the validation of LP or LPHO UV reactors, the UVT
UV Disinfection Guidance Manual C-26 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
should be measured and recorded with each influent sample. With MP reactors, the UVT from
200 to 400 nm should be measured and recorded. The electrical power delivered to the lamps by
each ballast should also be measured and recorded. The challenge test should be repeated if the
flowrate, UV intensity, lamp power, or UVT changes by more than the error of the measurement
over the course of sampling.
The challenge microorganism concentration in the samples should be measured within 24
hours of collection using a peer-reviewed method. Suggested methods for measuring MS2 and
B. subtilis spore concentrations in water samples are provided in Appendix D. Reported
challenge microorganism concentrations should include dilutions, volumes used, and the number
of plaques or colonies counted on each plate.
C.4.9.6 Quality Assurance and Quality Control Samples
f-i
During testing of the UV reactor, samples should be collected to ensure quality assurance
and control (QA/QC) including:
• Trip controls - sample bottles of challenge microorganism stock solution of known
concentration that travel with the stock solution from the microbiological laboratory
to the location of reactor testing and back to the laboratory. The concentration of the
challenge microorganism in the trip controls measured at the beginning and end
should be the same at a 90 percent confidence level.
. Reactor blanks - influent water samples taken without any addition of challenge
microorganism to the flow passing through the reactor. The concentration of the
challenge microorganism measured with the blank should not interfere with the
determination of RED delivered by the reactor.
» Reactor controls - influent and effluent water samples taken with the UV lamps (in
the reactor) turned off. The challenge microorganism concentrations in both samples
should be the same at a 90 percent confidence level.
. Method blanks - sample bottle of sterilized reagent grade water that undergoes the
challenge microorganism assay procedure. The concentration of challenge
microorganism with the method blank should be non-detectable.
C.4.9.7 Challenge Microorganism Dose-Response
The UV dose-response of the challenge microorganism within samples collected from the
reactor influent should be measured with the collimated beam apparatus as described in
Appendix E. At least two dose-response curves should be generated. One sample should have
UVT unadjusted by UV-absorbing additives and one sample should have UVT adjusted to give
the minimum UVT used in section C.4.9.4. A one-liter influent sample should be sufficient for
measuring the challenge microorganism UV dose-response.
UV Disinfection Guidance Manual C-27 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
The collimated beam tests should be conducted within 24 hours of sample collection.
Based on the expected dose-response of the challenge microorganism, UV doses should be
applied to achieve log reductions of approximately 0.5,1.0,2.0, 3.0,4.0, and 5.0. For each log
reduction, at least three aliquots of the influent sample should be irradiated. Three aliquots
should also be collected as zero dose samples. Aliquots should be packed on ice and stored in
the dark until they are assayed. Aliquots should be assayed within 24 hours of irradiation.
The log inactivation for each applied dose delivered by the collimated beam should be
calculated using Equation C.5:
Nf
Log Inactivation = log "
N
Equation C.5
where
No
N
Average concentration of the challenge microorganism in the zero dose aliquots
Challenge microorganism concentration in an aliquot of sample
Fitting Dose-Response Data
The dose-response of the challenge microorganism should be plotted as UV dose versus
log inactivation. An equation that best expresses the UV dose as a function of log (No/N) should
be obtained using regression analysis. A linear equation should best-fit first-order kinetics. A
quadratic equation should provide a better fit with tailing, and other equations should be used if
inactivation kinetics involves shoulders (DVGW 1997, ONORM 2001). Equation coefficients
obtained from the regression analysis should be significant at a 95 percent confidence level. The
differences between the values measured and predicted by the equation should be randomly
distributed around zero and not show a dependence on dose. Confidence intervals for the fit
should be determined at an 80 percent confidence level. The equation should be used for
interpolating dose-response data but should not be used for extrapolation outside of the measured
UV dose range.
Example. The dose-response of MS2, presented in the following table, was measured
using a collimated beam apparatus.
UV Dose
(mJ/cm2)
0
10
30
60
100
Log
Inactivation
0.016
0.805
1.87
3.40
4.71
Log
Inactivation
-0.119
1.06
2.16
3,62
4.83
Regression analysis was used to fit the equations to the MS2 dose-response data:
Dose = A x Log Inactivation + B
and
Dose = C x Log Inactivation + D x (Log Inactivation)2
UV Disinfection Guidance Manual C-28 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
The following table lists the coefficients derived from the regression analysis, the p-
statistics for those coefficients, and the R-squared value for the fit.
Equation
Dose = Ax Log Inactivation + B
R-squared = 0.967
Dose = Cx Log Inactivation + D x (Log Inactivationf
R-squared = 0.995
Coefficient
A
B
C
D
Value
20.5
-6.01
8.90
2.47
p-statistic
3.13 x10'7
0,15
1.22x10^
4.39 x10'6
In evaluating the two equations, a first check was done to determine if the equation
coefficients were significant at a 95 percent confidence level. While the R-squared value for the
first equation was high, the p-statistic for coefficient B was greater than 0.05, indicating that it
was not significant at a 95 percent confidence level. Thus, the first equation was not a good fit to
the dose-response data. On the other hand, the p-statistics for coefficients C and D with the
second equation were both less than 0.05, indicating that they were significant at a 95 percent
confidence level. Thus, Equation 2 was a valid fit to the dose-response data.
A second check of the two equations was to determine if the difference between the
measured and predicted dose was randomly distributed as a function of the log inactivation.
Figure C.3 presents the dose-response data and the fits to the data with confidence levels. As
shown, the first equation under-predicts UV dose at low and high levels of inactivation and over-
predicts dose at mid levels of dose. On the other hand, the second equation does not show a bias
in the prediction of dose as a function of log inactivation. This second check further
demonstrates that the second equation was a valid fit to the dose-response data while the first
equation was not valid.
To illustrate the importance of using an appropriate equation to fit the dose-response data,
the following table compares the dose predicted using the two equations for 2-log inactivation.
Equation
Dose = A x Log Inactivation + B
Dose
= Cx Log Inactivation + D x (Log Inactivationf
UV Dose for 2 log Inactivation
(mJ/cm2)
Mean
35
28
Lower Bound
30
26
As shown, the first equation over-predicts the mean dose .needed for 2-log inactivation by
27 percent, as compared to the second equation. Large errors can occur predicting the UV dose
associated with a given log inactivation if the equation used to fit the data is not appropriate.
UV Disinfection Guidance Manual
Proposal Draft
C-29
June 2003
-------
Appendix C. Validation of UV Reactors
Figure C.3 UV Dose Plotted as a Function of MS2
Log Inactivation and Fitted Using Two Equations
Dose = A* Loglnactivaton + B
-1.
1 2 3.4
Log Inactivation
5
1234
Log Inactivation
Dose - C * log inactivation + D * (log /»ac
0
Combining Dose-Response Data
During validation, the UV dose-response of the challenge microorganism is used to relate
the inactivation measured through the reactor under each test condition to an RED value.
Typically, it is assumed that the dose-response measured with a subset of the test conditions
assessed during validation can be used to calculate the RED for all test conditions. This
assumption is valid if the dose-response of the challenge microorganism does not vary from test
condition to test condition. To prove this assumption, the regression coefficients generated for
each set of dose-response data should be equal at a 95 percent confidence level (Draper and
Smith 1981). If the coefficients are the same, the equation fitting the combined dataset should be
used for determining the RED. If the coefficients are different, the cause of the difference should
UV Disinfection Guidance Manual C-30 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
be determined. Difference in UV dose-response could occur if the dose-response was
determined with different batches of the challenge microorganism or if water quality
interferences are impacting the dose-response (e.g., MS2 coagulation). The following example
presents an approach that can be used to determine if two sets of UV dose-response data can be
combined.
Example. The following table gives the dose-response data for MS2 measured on two
influent samples during validation testing.
UV Dose
(mJ/cm2)
0
10
20
40
60
SO
100
Log Inactivation (No/N)
' Influent Sample 1
0.02
0.33
1.1
1.8
2.7
3.5
3.9
0.09
0.709
1.4
2.4
3.2
4.4
4.4
Influent Sample 2
0.24
0.54
1.0
2.3
L 3.2
3.4
4.3
-0.10
0.40
1.4
2.3
3.3
3.9
4.8
Each dataset can be described using the following equation:
N ^ ( (N '
Dose = AxLog] -^- +Bx Log -f
N I I IN
To determine if the two datasets could be combined, a general equation is defined for
both datasets as:
+ Cxdxl
N
+ D x d x Lo
The term d is set to zero with the first dataset and set to one with the second dataset.
Multiple regression analysis using the full dataset is used to determine the values of coefficients
A, B, C and D with the following results:
Coefficient
A
B
C
D
Value
17.5
1.03
-2.43
0.435
p-statistic
0.000
0.202
0.553
0.689
As shown by the p-statistic, the term A was statistically significant at the 95 percent
confidence level (p < 0.05) and the terms B, C, and D were not (p > 0.05). The regression
analysis was repeated in a step-wise fashion, removing the term with the highest p-value from
the equation. With the second regression, the terms A and B were statistically significant and the
term C was not. With the third regression, the terms A and B were both statistically significant.
Because neither terms C nor D were significant, it can be concluded that the regression
UV Disinfection Guidance Manual
Proposal Draft
C-31
June 2003
-------
Appendix C. Validation of UV Reactors
coefficients generated by the fits to each dose-response are equal at a 95 percent confidence
level. Thus, the two datasets can be combined.
C.4.9.8 Reactor Log Inactivation and RED
For each condition of flowrate, DVT, and lamp output as defined in sections C.4.9.3 and
C.4.9.4, the arithmetic mean and standard deviation of the log of the influent and effluent
challenge microorganism concentrations should be calculated. For each test condition, the log
inactivation should be calculated using equation C.6:
Log Inactivation = log(N, )-log(NE )
Equation C.6
where
log(Ni) = Mean challenge microorganism log concentration of the influent samples
log(Ne) = Mean challenge microorganism log concentration of the effluent samples
The uncertainty of the log inactivation should be calculated using Equation C.7:
Log Inactivation
xlOO%
Equation C.7
where
CTE
Percent uncertainty of the log inactivation through the UV reactor
t-statistic of the influent samples at an 80 percent confidence level
Standard deviation of the challenge microorganism log concentration of the
influent samples
Number of influent samples
t-statistic of the effluent samples at an 80 percent confidence level
Standard deviation of the challenge microorganism log concentration of the
effluent samples
Number of effluent samples
The RED should be calculated from the log inactivation using the equation describing the
UV dose-response curve of the challenge microorganism. The percent measurement uncertainty
of the RED can be calculated using Equation C.8:
UV Disinfection Guidance Manual
Proposal Draft
C-32
June 2003
-------
Appendix C. Validation of UV Reactors
*•' otrr*
'RED
where
URED
UDR
UD =
Equation C.8
Percent uncertainty of the measured RED
Percent uncertainty of the regression equation fitting the challenge
microorganism's UV dose-response data at an 80 percent confidence level (see
section C.4.8.7, Figure C.3)
Percent uncertainty of the collimated beam dose calculation that is not captured in
the variability of the measured dose-response data (see Appendix E). This
typically includes the uncertainty of the radiometer and the Petri factor
Example. A UV reactor was validated using MS2. The UV dose-response measured
using a collimated beam apparatus is given in Figure C.3. The dose-response was fitted using
the following equation:
Dose = 8.90 x Log Inactivation + 2.47 x (Loglnactivation)2
The uncertainty of the radiometer used with the collimated beam apparatus was 8 percent.
The Petri factor was measured with an uncertainty of 2 percent. Thus the uncertainty of the
collimated beam dose calculation, UD is calculated as follows:
The following table presents the microbiology results obtained with the influent and
effluent samples collected with one of the test conditions assessed during validation.
The mean and standard deviation of the influent and effluent log concentrations of the
MS2 are 6.32 ± 0.075 and 4.26 ±0.13, respectively. The log inactivation through the reactor is
calculated as follows:
Log Inactivation = 6.32 - 4.26 = 2.06
Table C.2 Estimated Log Inactivation and Corresponding RED Values Using
Bioassay Results
Before UV
Sample 1
Sample 2
Sample 3
Plate Counts - Dilution = 10*
1
148
173
TNTC
2
180 ,
TNTC
192
3
TNTC
TNTC
150
Plate Counts - Dilution = 10°
1
15
11
37
2
18
32
15
3
20
22
22
Concentration
PFU/mL
1.77x106
2.17x106
2.47x1 06
log
6.24
6.33
6.39
After UV
Sample 1
Sample 2
Sample 3
Plate Counts - Dilution = 10'
1
166
133
165
2
181
TNTC
141
3
TNTC
101
123
Plate Counts - Dilution = 10'
1
17
13
17
2
18
28
14
3
42
10
12
Concentration
PFU/mL
2.57x10"
1.70x10"
1.43x10"
Log
4.40
4.23
4.15
UV Disinfection Guidance Manual
Proposal Draft
C-33
June 2003
-------
Appendix C. Validation of UV Reactors
The t-statistic for 3 samples and an 80 percent confidence level is 1.88. The percent
uncertainty of the log inactivation is calculated as follows:
f(l.88x0.075)2 (1.88x0.13)*^
Uin =± ^—x 100 = 7.91%
2.06
The RED associated with a log inactivation of 2.06 is calculated as follows:
Dose = 8.90 x 2.06 + 2.47 x (2.06)2 = 28.8 mJ/cm2
The percent uncertainty of the regression equation, UDR, at a log inactivation of 2.06 is 6 percent.
The percent uncertainty of the RED is calculated as follows:
URED =(7.92 +62 +8.22)^ =12.9%
C.4.9.9 Interpretation of Results }
Interpretation of the results should depend on the monitoring approach used to guarantee
dose delivery:
« With the UV intensity setpoint approach, the UV reactor should be rated at the lowest
inactivation observed for each setpoint condition evaluated.
* With the UV intensity and UVT setpoint approach, the UV reactor should be rated at
the inactivation observed with UV reactor operation under setpoint conditions.
• With the calculated dose approach, the UV reactor should be rated at the lowest
inactivation observed for each calculated dose setpoint evaluated.
C.4.9.10 Interpolation of Results
The RED measured by validation testing can be interpolated as a function of flowrate,
UVT, and UV intensity by fitting an equation to the data being interpolated. If the RED is
interpolated as a function of the measured intensity or the inverse flowrate, the equation used
should pass through the origin (0,0). The equation coefficients should be significant at a 95
percent confidence level. The differences between the values measured and predicted by the
equation should be randomly distributed around zero. The equation should be used for
interpolating between measured data but should not be used for extrapolation.
The uncertainty of the equation used to interpolate the RED should be assessed by
determining the 80 percent confidence level. If significant, the uncertainty should be included as
an uncertainty term in the determination of the expanded uncertainty, as described in section
C.4.10.2.3.
UV Disinfection Guidance Manual C-34 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
C.4.10 Determining Inactivation Credit
This guidance presents two approaches, termed Tier 1 and Tier 2, which can be used to
relate the RED demonstrated during reactor validation to target pathogen inactivation. Other
approaches or modifications to this approach may be used at the discretion of the State.
With both approaches, the RED demonstrated during validation should be equal to or
greater than a target RED that is related to the dose tables in Chapter 1 using a safety factor.
With Tier 1, fixed safety factors have been defined and applied to the dose tables in Chapter 1 to
define target RED values. The Tier 1 safety factors are based on specific Tier 1 criteria for the
UV reactor and its validation protocol. The Tier 1 approach can be used with a given UV reactor
provided it meets all the Tier 1 criteria. With Tier 2, the safety factors are calculated based on
the validation results for, and certain properties of, the UV reactor that are calculated from the
validation results and certain properties of the UV reactor undergoing validation.
C.4.10.1 . Tier 1 Approach
For a UV reactor using LP or LPHO lamps, Table C.3 presents the Tier I RED values
that should be demonstrated during validation to achieve the specified log-inactivation credits for
Cryptosporidium, Giardia, and virus. Table C.4 presents the Tier I RED values for MP reactors.
The Tier 1 RED values are applicable with all UV reactors that meet the Tier 1 criteria provided
in this section.
Example. To receive 2.5 log Cryptosporidium inactivation credit,.a LP reactor under
Tier 1 should demonstrate an RED of 28 ml/cm2.
Table C.3 Tier 1 RED Targets for UV Reactors with LP or LPHO Lamps
Log
Inactivation
Credit
0.5
• 1.0
1.5
2.0
2.5
3.0
3.5
4.0
RED Target (mJ/cm")
Cryptosporidium
6.8
11
15
21
28
36
-
•
Giardia
6.6
9.7
13
20
26
34
-
-
Virus
55
81
110
139
169
199
227
259
UV Disinfection Guidance Manual
Proposal Draft
C-35
June 2003
-------
Appendix C. Validation of UV Reactors
Table C.4 Tier 1 RED Targets for UV Reactors with MP Lamps
Log
Inactivation
Credit
0.5
1.0
1.5
2,0
2.5
3.0
3.5
4.0
RED Target (mJ/cnV)
Cryptosporidium
7.7
12
17
24
32
42
-
-
Giardia
7.5
11
15
23
30
40
-
-
Virus
63
94
128
161
195
231
263
300
Tier 1 criteria for the UV reactor are as follows:
. UV reactors equipped with MP lamps should be equipped with one sensor per lamp,
UV reactors equipped with LP or LPHO lamps should be equipped with at least one
sensor per bank of lamps.
. The standard deviation of the UV output of LP or LPHO lamps should be 15 percent
or less of the mean output. The standard deviation should be determined using either
life test or field data on aged lamps,
• UV intensity sensors should view a point along the length of the lamp that is within
25 percent of the arc length away from the electrode.
• UV intensity sensors should have a spectral response that peaks between 250 and 280
nm. When mounted on the UV reactor and viewing the lamps through water, the
measurement of UV light greater than 300 nm made by the sensor should be less than
10 percent of the total measurement made by the sensor. Conformance to these
criteria can be demonstrated using UV intensity field modeling. Figure C.4 presents
an example of how two sensors would conform to this criterion.
• The UV intensity sensors used during validation and the duty and reference sensors
used during operation of the UV reactor at the WTP should provide NIST traceable
measurements with a measurement uncertainty of ± 15 percent or less at an 80
percent confidence level.
. During operation of the UV reactor at the WTP, measurements made by the duty UV
intensity sensor should be checked using a reference UV intensity sensor. The
difference between the measurement made by the duty and reference sensors should
meet the following criteria:
Equation C.9
UV Disinfection Guidance Manual
Proposal Draft
C-36
June 2003
-------
Appendix C. Validation of UV Reactors
Figure C.4 Comparison of the Spectral Response of Two UV Intensity Sensors
Estimated Using UV Intensity Field Modeling
250 280
Filtered Sensor
Unfiltered
Sensor
200 250 300 350
Wavelength (nm)
400
300 nm is 0.7% of total UV light detected.
01 50 -,
.1 40 -
Z 30 -
Detected 1
->• 10
o o o
NJ
•
^jj4uidL
_JL
)0 250 300 350 400
Wavelength (nm)
Unfiltered Sensor. Detected UV light with a 0
cm sensor-to-lamp water layer. Detected UV >
300 nm is 41% of total UV light detected.
« 0.0025 -,
J 0.0020 -
f 0.0015 -
| 0.0010 -
£ 0.0005 -
o n nnnn
2(
J
30 250 300 350 400
Wave length (nm)
Filtered Sensor. Detected UV light with a 20
cm sensor-to-lamp water layer. Detected UV >
300 nm is 5% of total UV light detected.
« 0.0400 -,
| 0.0300 -
2
J= 0.0200 -
o 0.0100 -
•S
0 0.0000 -
2(
'
J
I
)0 250 300 350 400
Wavelength (nm)
Unfiltered Sensor. Detected UV light with a 20
cm sensor-to-lamp water layer. Detected UV >
300 nm is 85% of total UV light detected.
UV Disinfection Guidance Manual
Proposal Draft
C-37
June 2003
-------
Appendix C. Validation of UV Reactors
. If the dose monitoring strategy uses an on-line UVT monitor, the Ais4 calculated from
the measured UVT should have a measurement uncertainty of ± 10 percent or less at
an 80 percent confidence level.
Tier 1 criteria for the flow measurements are as follows:
. The flow measurements during validation and during operation of the UV reactor at
the WTP should have a measurement uncertainty of ± 5 percent or less at an 80
percent confidence level.
Tier 1 criteria for the collimated beam apparatus are as follows:
• The calculated dose delivered by the collimated beam apparatus should have a
measurement uncertainty of ± 15 percent or less at an 80 percent confidence level.
Tier 1 criteria for the challenge microorganism dose-response are as follows:
• Over the range of doses within one log of the log reduction demonstrated during
validation, the UV sensitivity of the challenge microorganism should be less than or
equal to 25 ml/cm2 per log inactivation (the dose-response of a resistant strain of
MS2). For example, if you measure log inactivation values between 1.5 and 3.5 log,
the test organism you use should have a dose-response less than or equal to 25.
mJ/cm2 per log inactivation between 0.5 and 4.5 log inactivation.
« If the dose-response of the challenge microorganism has a shoulder, that shoulder
should not occur over a dose range greater than 50 percent of the RED demonstrated
during validation. The shoulder is defined by extrapolating the exponential reduction
region of the dose-response curve to the dose-axis.
» If the dose-response demonstrates tailing, the tailing should not occur until one log
reduction greater than the highest log reduction demonstrated during validation.
Tier 1 criteria for the UVT used for validating UV reactors using medium-pressure lamps
are as follows:
. The UVT at 254 nm of the water during validation should be greater than the values
specified in Figure C.5 for a given sensor-to-lamp water layer and UV-absorbing
chemical (the polychromatic bias should be 1.0). The sensor-to-lamp water layer is
defined as the distance traveled through water by UV light passing from the lamp to
the sensor. The values in Figure C.5 were taken from Figure C.7 for a polychromatic
bias of 1.2.
UV Disinfection Guidance Manual C-38 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
Figure C.5 Criteria for the Minimum UVT of MP UV Reactors under Tier 1
mn
. e
; * DO
I gr
§
I nn ;
.5 OU
C
i .
7C .
f
i Coffee
— ~Lignin Sulphonale
- - -NOM
,
0
/
^r A
j
/
X
- J
f
/
*
'
' -^>*^ff
k
F „
^
-^-*-
t^
* « J
-•
5 10 15 ,20 25 ' 30
, Sensor-to-lamp water layer (cm)
Tier 1 criteria for the challenge microorganism dose-response data are as follows:
. A plot of dose versus log inactivation should have an 80 percent confidence level of
10 percent or less at the log inactivation demonstrated by the UV reactor.
Tier 1 criteria for the challenge microorganism measurements through the reactor are as
follows:
. Five influent and five effluent samples should be collected per test condition
evaluated as per section C.4.9.5.
» The standard deviation of the challenge microorganism concentration measured with
the influent and the effluent samples should be less than or equal to 0.20 log.
Tier 1 criteria for the interpolation of challenge microbe results are as follows:
. The uncertainty of the interpolation should be 10 percent or less at an 80 percent
confidence level.
C.4.10.2 Tier 2 Approach
The safety factor used to relate the RED demonstrated during validation to the dose
required to inactivate the target pathogen should be defined using Equation C.10:
UV Disinfection Guidance Manual
Proposal Draft
C-39
June 2003
-------
Appendix C. Validation of UV Reactors
Equation C. 10
where
SF
BRED =
Safety Factor
RED bias
Polychromatic bias
Expanded uncertainty as a fraction
The following sections describe an approach for defining each of these terms.
Determining the RED Bias
If a single challenge microorganism is used to demonstrate dose delivery during
validation, the RED bias should be determined using Figure C.6 and the following procedure.
(Section F.I provides the background on the development of Figure C.6 and the procedure for
determining the RED bias.)
Procedure
1. Calculate the UV sensitivity of the target pathogen as the dose requirement specified
in Chapter 1 divided by the corresponding log inactivation credit.
2. Calculate the UV sensitivity of the challenge microorganism as the calculated RED
divided by the log inactivation.
3. If the target pathogen is more resistant to UV light than the challenge microorganism,
the RED bias equals 1.0. Otherwise, calculate the RED bias using Equation C.I 1:
RED Bias = REDc
REDD
EquationC.il
where
REDC
REDp
= RED of the challenge microorganism obtained from Figure C.6
= RED of the target pathogen obtained from Figure C.6
Example. An MS2 inactivation of 2 log corresponding to an RED of 36 mJ/cm2 is
measured during validation. A 2-log Cryptosporidium credit is required. The UV dose required
to achieve that level of inactivation from Chapter 1 is 5.8 mJ/cm2. Thus, the UV sensitivity of
MS2 and Cryptosporidium is defined as 36/2.0 = 18 and 5.8/2.0 = 2.9 mJ/cm2 per log
inactivation, respectively. Because MS2 is more resistant.than Cryptosporidium, the RED bias is
greater than one. In Figure C.6, REDs of 19 and 8.2 correspond to UV sensitivities of 18 and 2.9
mJ/cm2 per log, respectively. Thus, using Equation C.I 1, the RED bias is 19/8.2 = 2.3.
Example. An MS2 inactivation of 4 log and a corresponding RED of 80 mJ/cm2 is
measured during validation. A 2.0-log adenovirus credit requiring a dose of 100 mJ/cm2 is
required. Thus, the UV sensitivity of the challenge microorganism and pathogen are 20 and 50
mJ/cm2 per log inactivation, respectively. Because the UV sensitivity of adenovirus is greater
than that of the challenge microorganism, the RED bias equals 1.0.
UV Disinfection Guidance Manual C-40 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
Figure C.6 RED versus Microorganism UV Sensitivity
for Use in Determining the RED Bias
10 20 30 40
UV Sensitivity (mJ/cm2 per log inactivation}
If two challenge microorganisms with different UV sensitivities are used during
validation to demonstrate dose delivery, the RED delivered to this target pathogen can be
determined by interpolation using the following procedure. (Section F.I. 3 provides the
background on the use of two challenge microorganisms to demonstrate RED delivered to a
target pathogen.)
Procedure
1. For a given test condition of flowrate, UVT, and lamp output, calculate the UV
sensitivity of the challenge microorganisms as their respective measured REDs divided
by their corresponding log inactivations.
2. Determine the UV sensitivity of the target pathogen as the dose listed in Chapter 1
divided by the log inactivation.
3. Calculate the RED delivered to the target pathogen using the following equation:
Equation C.I 2
where
REDp = Estimate of the target pathogen's RED
REDci = The RED measured with the first challenge microorganism
REDc2 = The-RED measured with the second challenge microorganism
DlOp - UV sensitivity of the target pathogen (mJ/cm2 per log inactivation)
DlOci = UV sensitivity of the first challenge microorganism (mJ/cm2 per log
inactivation)
D10c2 = UV sensitivity of the second challenge microorganism (mJ/cm2 per log
inactivation)
UV Disinfection Guidance Manual
Proposal Draft
C-41
June 2003
-------
Appendix C. Validation of UV Reactors
4. Calculate the percent uncertainty of the estimated RED of the target pathogen using
Equation C.13:
'-'REDp ~
where
Ut
REDC1URED,C1
)2
Cl .
REDp
Equation C.I3
'REDp
u
RED.C1
Percent uncertainty of the RED estimated for the pathogen
Percent uncertainty of the RED measured with the first challenge
microorganism (see Equation C.8)
As an alternative two-microorganism approach, the log inactivation measured with the
challenge microorganisms can be interpolated as a function of the microorganisms' first-order
inactivation coefficients.
Example. A UV reactor was validated using MS2 and (|>X174 at 1 and 2 mgd. The UV
sensitivities of the MS2 and <|>X174 were 18 and 2 mJ/cm2 per log inactivation, respectively. The
following table gives the RED and percent uncertainties measured with MS2 and (|>X174. At the
lower flowrate of 1 mgd, the <|>X174 was inactivated to below the detection limit and the
measured RED was estimated as greater than 10 mJ/cm2. The table also gives the RED delivered
to Cryptosporidiwn estimated using Equation C.I 2 and the percent uncertainty of that RED
estimated using Equation C.13. These estimations assumed a UV sensitivity of Cryptosporidiwn
of 4.0 mJ/cm2 per log inactivation based on the dose in Chapter 1 for a 3.0-log inactivation
credit.
Flow
(mgd)
1
2
MS2
RED
(mJ/cm2)
40
20
Uncertainty
{%)
6
11
*X174
RED
(mJ/cm2)
>10
9
Uncertainty
(%)
0
4
Cryptosporidium
RED
(mJ/cm2)
14
10
Uncertainty
(%)
3.2
2.9
Determining the Polychromatic Bias
For a UV reactor using a germicidal UV intensity sensor (the spectral response meets
Tier 1 criteria), the polychromatic bias can be assigned a value of one if the UV intensity sensor
is located where dose delivery is proportional to measured UV intensity or closer to the lamps
than that location. This can be shown experimentally by demonstrating under fixed conditions of
flow and measured UV intensity that the RED obtained with peak UVT and lowered lamp power
is greater than or equal to the RED measured with peak lamp power and lowered RED.
• If data are not available showing the UV intensity sensor location meets the above
criteria, the polychromatic bias should be determined by calculating, at a given flowrate, UVT,
UV Disinfection Guidance Manual
Proposal Draft
C-42
June 2003
-------
Appendix C. Validation of UV Reactors
and measured UV intensity, the ratio of RED during validation to the RED at the WTP. This
calculation should be done conservatively by assuming ideal dose delivery where dose is the
product of the average intensity within the reactor and the theoretical mean residence time. The
calculation should include the following factors:
• The spectral UV transmittance of the water during validation and at the WTP.
• The spectral lamp output during validation and expected at the WTP with aged lamps.
. The spectral sleeve UV transmittance during validation and expected at the WTP with
aged and fouled sleeves.
« The spectral response of the sensor used during validation and at the WTP.
. The action spectra of the challenge microorganism used during validation and the
action spectra of the target pathogen taken from the literature.
If the above ratio is less than one, the polychromatic bias should be assigned a value of
one.
Figures C.7 to C.9 present the polychromatic bias for reactors with UV intensity sensor
spectral response curves shown in Figure C.10, Each figure presents, for a given sensor spectral
response, the polychromatic bias as a function of the UVT, the sensor to sleeve water layer, and
the UV absorbing chemical used during validation (coffee, lignin sulphonate, and natural organic
matter (NOM)). The spectral UV absorption coefficient of the UV absorbers and the WTP water
used to define the polychromatic bias values is provided in Figures C.I 1 and C.I2. Figures C.7
to C.9 can be used to determine the polychromatic bias if the spectral response of the UV
intensity sensor used in the figure is representative of the spectral response the UV reactor's
intensity sensor. Alternatively, the polychromatic bias can be calculated using a model that
meets the above-mentioned criteria.
The polychromatic bias shown in Figures C.7 to C.9 was determined for an annular
reactor with a reactor radius of 18.8 cm and a sleeve radius of 3.81 cm. The UV intensity field
was calculated using a radial intensity model. Section F.4.2 presents details on the models used
to develop Figures C.7 to C.9.
The polychromatic bias values in Figures C.7 to C.9 only account for differences between the
spectral UV absorbance during validation and the spectral UV absorbance at the WTP. They do
not account for the impact of spectral shifts in the optical properties of the UV reactor (e.g., lamp
output, sleeve UVT). If spectral shifts in UV reactor properties occur with operation of the UV
reactor at the WTP, the polychromatic bias should be multiplied by terms that account for those
shifts. Section F.4.3 describes spectral shifts and provides estimates of the polychromatic biases
that can occur with those shifts.'
UV Disinfection Guidance Manual
Proposal Draft
C-43
June 2003
-------
Appendix C. Validation of UV Reactors
Figure C.7 Polychromatic Bias as a Function of Water UVT and Sensor-to-Lamp
Water Layer for UV Reactors using Sensors with Germicidal Response (response
A in Figure C.10) Validated using Coffee, Lignin Sulphonate, or MOM
f.fl '•"••• ' : •• ' • '
•H 1 R '.
I
t
In
i i l l
% • • 1 '
-{• -1- -i- 1 -
. T% , t
i t ^ r\
i t i i .
i ^
|'||
i i i
• •"'VJ-^
. -Lll_._.
.(i A 1
1 ' '
, .'..' .'..
*w
• ' ' ^S^
r ** -^
r i i
"Ji V
i "^*P!T'-l"v*
;1- ^-r-
^
"X.. i
"1 * • 1
1 • l\
, ,
'S ' '
"? **J 1
t t
' 1
1 ' ^N
Ai i ii
• i i^i
i ! ' ' ' >>
V 0 "*•"•"
1 T^ l t
1. :i'. ( !• .
[X^ft:j£l-
i * t i
• .1. «.. i
hi - i ":«:
\^t < 1
L BB-.' J
^t t^ ' :
r • .i\^. .
1 H_
i ••. i* i*. i .
.:i-=- ' ••• •
:• 1 1 . 1 -1 =
1 -•••.. -
- | . 1 '• . 1 . " 1 . :
< r r ' . .
..I..-I I-.! '
^il-
S
A
inson Sensor A :
bsoitaer: Coffee
: water layer l
is- • —10 cm
... - 15 cm,
— — 20 cm
.: 25 cm
70 75 80 85 90 95 • 100 :
UVT at 254 nm (%) '
2.0 -
A 0 _
I1'8
m
1 1'6 "
1 1 .4 -
in ,
••crr-i-i*
j j > |
TTT-V
_,.£?•-, -J^
-:- *;--;- -i-
<-f*M*.-Ti
• -^ . . . Y
liK:::::^::::::!::
: :i:i;:-
L L !^
:fe:-::
i iSsJ
:fe:r-
•.•:- •:-;•*-
i i^ i- i
1 1 1 ..:!
"rl^5
.11 i i
-L,«. J_ J-
1 " 1 : . 1^' ( '.:•
1 ' 1 •• 1 1 .
"f-i'TT-
70 75 80 85 90 ""<9
UVT at 254 nm (%)
•T-r^r-i--
...,1 .,1 ,1 ... I!
-l.L.L J-.
• i • r. -. i- i ". .
-T-f »rl^r
:;•.],•-! "!; •
&.t"S;
S
A
:
bsotber: Ligni
Sulphonate
5ensor4o-lamp
water layer
<^~ - -10cm
* * • -15cm
— •— 20 cm
• 25 cm
A
n
s 7:: "'-iiooi^. , •. . ' ' • J
-2.o.q
81-8:
e
1 1-6'
tl 4 ^
1'1:
1 2
1.CH
' •; 7
-i-J.JSJ.
->*!--;- -
• ! V
i i i
-r -wn- -
. W .1- J - _
fci_ t *
111^'
•r-i*,1^" -
.W _l. J? •—
-r-r^r -
1 t :1
mj.
•
T-
I
^5^
' V
.1.1..I-AJ
:|' ' 1 "•(
-*1 cr m~-
- -*^|; ,---
.. -U .U J. .
- ^(*-I*M
1 1 1
-I,-. ,-r-r~
i i i
^-K-
NV>.S
••;-: . s
* $A ji *i ^
ayU[J£^
1 1
•-I-1- • -
1 1
rsd~ : •
Jrstj- •
"*7"«w.
._!_ J^l^^S
•>^i->
:mS^t-Vii
i i i i
-r.-i-Tin-1
.1. 1 ti 1
'I' 1' I- 1
-r*r-r!T"1
1 1 1
^^Wfcir-
.*«!«!
0 75 80 85 '90 '9
UVT at 254 nm (%>
1 1
- ,- -r-r-
i r
i i
- ~ i • i
• - -r-r-
• < i
-T- rrn--
SFffTifli i
5 ' 1C
1 ,
S(
f
•
10
tnsor. Sensor A
tosorber: MOM
sensor-to-lamp
C f*mi
— *• — •™r*O Cm
— - -IQ.cm
— —•• 20:cm;
25cm
UV Disinfection Guidance Manual
Proposal Draft
C-44
June 2003
-------
Appendix C. Validation of UV Reactors
Figure C.8 Polychromatic Bias as a Function of Water UVT and Sensor-To-Lamp
Water Layer for UV Reactors Using Sensors with SiC Response (Response B In
Figure C.10) Validated Using Coffee, Lignin Sulphonate, or NOW
2.0 i
s :
M 1 fi •
ft1'4
£ 1 ?:
1.0 :
7
1 tit
l V "
i J J 1
1 1 t ">
i I I i
^fc"i i V
'- iJSwV'
tiii
•T V_V V.1
•-:--;--!--!-
• i t .1
i - i
i i
i t
1 .; ' .. • 1
till
-"A - t ^. 1. -U -
i-L.i. j. j_,
fcv i i
J V , j
1 1 ^1 1
I j r^Lj >
i i » h.
li>t
. _•_ J_ J_J-
..» J-^.J-
! 1 « -"»
- i*t^iL t i
. .1. J.'J _ J .
I I i i
I I i i
' L 1 '
f I I lT
1 L' ' •
! '!**!"
- L. J. ,
; |^5 ""' ; .
- r - r iS- •
. V . L .'. J. .
r i i • i
. , .., , i
'A 1' \ '
j ' j\L
1 \ 'V.
j * j I
i ;V V
i ^ I\ i "
.j*j.j_V
.j ' ^!.J
M. J.J.Lc
i '^1^1
i i i i
-r-t 1--
i i 1;
: I J
1 1 1
, 1 J
1 1 1
t 1 1 -
\ ; i ;
XV -!-•;-•
' "LIT." '
' H f S ' ~
. r _|^^,_ -j _ ,
.«. 1 .
s
A
1
ensor: Sensor B
bsoiben Coffee
Sensor-to-lamp
.water layer
— - -10 cm
-r ,— 20cm
• — 25 cm
0 75 . 80 85 90 . 95 .; ,100 . ' .
UVTat254nm(%)
,u
S 1-8 •
CO
" 1 fi
1 1'°::
t'4
i£ 1 2
1 n .
:l 1 1 1
lilt
-h-X,--!-
i i t i
• -r -i- T T
_L.t. J.J.
I I t i
* -IT T T T •
"l ."> *1 l" "
. L _^% 4. J m -m
.L.L.L J.-
I I i l
- r - r -r T •
~l "l "l" l"'
1 1 J 1
:i^:i::
-Vi-f-!"
j 'L \ L
-.i-;.-^,
"1 J - L . L -C .
• i - r* iv," r •
" J "i "i '•
_l_l__C_i_.
t •• i- »i
-, - r - r -i- .
^J.X. ->-.
V-!- V
. ^V _ JV_
1 '\
• r 'x:'
.j. r .L_
_"j - . .t!,
»i i
.J- . -V*
• i - - -r •
_J- - if-
'^^'SH.^J^I
_JL J.l'i-1-.
till
H .L J. J- J.
-lilt
;^§s
liSKK
" "i" i " i " t
.-*._!- J. J-
1 1 1 1
" 1 " 1 " 1 " .'
- r -r -i- T •
. L .1. ->- J- •
till
-t_L-l, J* .
1 i i I
1 - r - r -i- 1 • '
* t "*^i ~ 1 "
S
A
ensor: Sensor B
bsorben Lignin
Sulphonate
Sen sor-to- lamp
.water layer
"~"7 — 5 cm-
r- ' -10 cm
• ' • - 15 cm
— — 20 cm
—•25cm
70 T 75' 80, 85 90 V,;95 ; ; ; 100 , - .
20 i
tA <\ Q .
.2
QD
U
? •
I :
•g 1-4:
.& ;
• 1.0 :
7
l*i%i •
, V -'• J-^- <
1 l ' 1 I\
-t-^J.J..
-r--t-^-^--
. .1. J- J. J.
""•to ' ' '
•-:--!--:--:-•
-:--tr7:":r:
\L 1 1 1
-^r-i--)-1
i.^^i i
':bi-?Nt
-L-L-V,'-.
-r-^-r->.
-Kr I"!"!"
.(..(..1.4-
i i i i .
*fcct"!"H"
4-i-r-^
^Ifr-*--!-1
-i- rr-r-
i • i
-JV 'r\-f-
^^ i
-j- \i. .
^N,S
*t- -[••>
-^^.!•*L;
=4*vciii^
--i-i-T-r.*
i i i i
.j.j. j.i.
i t i i •
*4H-I-!--
rT^SijI;-
-T^-vt1*
-j--:-^*-.
..'..<.}. *i
-.|~0_A_i_.
i i i
"i i*~"
-<-.« J. .
•-;-T-V- -
--:--!-i- -
Sfl-i- -
^@^
1 1 1 1
.L.L.I. -«. ,
1 1 1 1
-:--!--!--:-
.(..(.j.j-.
-L .Cj-J-.
i « i .1 .
^SMai- -!- •
St
t
jnsor: Sensor B
Ibsortaer: NOM
Sensor-to-lamp
water layer •'
•T5 cm
^ - -10icm
*. * * '-15; cm
— '— 20 cm
25 cm
D .75 ' 80 85 90 95 100 >
" UVrat254nm(%)
UV Disinfection Guidance Manual
Proposal Draft
C-45
June 2003
-------
Appendix C. Validation of UV Reactors
Figure
W
(Resp
C.9 Polychromatic Bias as a Function of Water UVT and Sensor-To-Lamp
later Layer for UV Reactors Using Sensors with Germicidal Response
onse C In Figure C.10) Validated Using Coffee, Lignin Sulphonate, or NOM
2n
.U
J1 Q .
1.8
CQ
, •— • 1 C .
75 i •"
1 14 :
.C 1 .4
f 12:
1.0 n
L V " J
i .X «
4-i--:«-
..i.j.J.J.
T """!""!"
i i i
* -_^ * •
• . i i ....i^
.1.1. j.' jr*
: riT.^T i- T •
1 . i ; .. i ! : i . I. . |
i L ' I
I'll
- I - L J. J. .
• TT V V
t*-H--:-
. A-^J. .i. .'.
..L.C-L^.,
_ I - L .1. .*-
T r r "n\
i i i i
- T - r TI" ""' '
I . i :, •! •
• T - P Ti* 'I' '
I I i -; 1 ..-.
* iT v •
j f i L
i *"f i i
• ";";"
. i i L . L J A
» • ' i
-v ' .'._
.•jjj'.SZtj
7 '"•';"• "it"
~T"t7T!gr 1
0;.- .: |8
uvr4t25
k • J JVi
.V. I '\,
> 3 i V
TV ft
.J.\.J.i5
i ( \ T
.J.i.iA.
•^]i|r;^
• *i i
..'. •!-•*[. -
--:-i-;^- -'
:S'T>-.V^T^
H^r-r-r^rir-tr.r.-|
ii»i
\L_L J, J.
. -V-1- j.i.
"T\]~1"
t !V '
*K p • *V?
*' v '"L
"fT-^r*
!^^"j.v^j>
^L.tjlj1^
I ' • r.
III).
1 - i ' i
1 II 1
1 1 * 1
.L.t.'.J.,
1 L .L.i.
if i V
ill t
-T"P"<" V
.j.i.-!..;:.
5' 1C
• •;
Sensor Sensor C
Absorber Coffee ;
Sensor-to-lamp
water layer
— - —1Q cm
* - - • 15 cm
5— — 20 cm
: 25cm
)0
j
20 i
J1 Q -
1.8
03
- 1R '
"S '
s..
1 :
•o
Q- 1 7 •
1 n
-J--J- J, J-
-^ - - ^. -t-
^i-- --:--:-
1 " r ™ " * ** i "i
-;- ---;-.-;-
i ' ' '_
^:=T*-
. 1.L1.I..I. .
i , A. .,
-•.L.J.-+.
.;.'..;...:
k. • ' '
-f-W— -
. ' .' »".'?5
M^i.L.L.<-_
^"l'!'^
JL ' 1 L
-•{-•iX^-
--;-.i-{-%
»i--i-7-r-
H->t-i-
.J-^.i-L-
'•i-i'-r-
( 1 1 1 ~
-rri*^
1111
t« • i i "
•\P'J'"J«'
-:-HJH-
.^.;.\J.
T i 1" " "
i i i^ i '
• V"'!""""^
'T.T If "I'
«.! 1 1 1
_L-I__I_ j..
I I I I
*• 1 1: 1
.J..J..-J..
.-..;....;..
-!•-;----;--
^r-| !-•
T*t" "'"
S^s^
.-j..i..rj--i
.1.1. L.k.^
t ' ' • :* ...
1 ' . L.I.
1 1 1 1
-t-J-t-!^
-}-;,;--;--
.|-|.|--L.
1 ill
-T-TTT-
Si
A
snsor: Sensor C
bsorben Lignin
Sulphonate
Sensor-to-lamp '
water layer
"•""""""""5 cm :
— - -10 cm
- - - • 15 cm
— . — 20 cm
-——25 cm
70 75 ' 80 85 .90 95 10d; '!.'•. i
UVTatZ54nm(lK,)
2n
••U
m-— '
11'°:
e
^ 1 4 •
o '•*.
"5
^ 1 T? -
• -i n -
. L .<. .'. JV
•vj- -:--:- i -
.L-V;--:-.
-r—l^s
-r---'--!-1
'!" i^'f\
. .C J. Ji J -
" 1 " •• 1
•W*™ - -i^r—.| w*^. w
H^r -:--;-•
~ 7 " ^^r ~ r "
H-*-rV
• X. 1 -'i: "1
• T^TT -r •
• t -T-Srr-
-i i i^-i_. '. .
"f "I"!""*!"'
1 -I' 1 •: 1
"P?"-r->
-i-T-T-:--
--1-M-I--
..;..;.J.L.
^-i-i-r-
•rfe;
l^H^
r i -j j
^ *lut
i i i i
..;.-;.j.j.
-:--:--:--:-
-i--;-:-:-
1 1 • |: -I
•r ~i***- -i-
-!--!--!-!-
s^ir.r.
*^r ;T*JI*
1 ^ij^
.L-...J.S;
-r -!----:-•
.i.j....;..
. i . L |. .
i i i
-T-r-r-r-
1111
-T-r-r-r-
. L.L.I. J..
,1 i 1 i
*i ^^JPdZ^
i r i
- - -i--:--
- - -*-!•.-
- - -T-rr
i '
• - - - r •
i
. . . . r-
- - - - r -
i
. J. . .L^
S
:
ensor: Sensor C
Absorber: NOM
Sensor-to-lamp
water layer
'— - -10 cm
- * * • 15 cm
20 cm
25 cm
70 .75 80 85 90 95 100 ; ,
UVT at 254 nm (%)
UV Disinfection Guidance Manual
Proposal Draft
C-46
June 2003
-------
Appendix C. Validation of UV Reactors
Figure C.10 Spectral Response of Sensors Used in Defining Figures C.7 to C.9
250 - "300 - 350
Wavelength (nm)
.400
Figure C.11 UV Absorption Coefficient of Coffee, Lignin Sulphonate, and the
Target Water used to Define Figures C.7 to C.9
0.00
200
250 300 350
Wavelength (nm)
us EPA Headquarters Ubrary
202-566-0556
UV Disinfection Guidance Manual
Proposal Draft
C-47
June 2003
-------
Appendix C. Validation of UV Reactors
Figure C.12 Spectral UV Absorbance of MOM and the
Target Water Used to Define Figures C.7 to C.9
0.20-
0.00
200
250 300 350
Wavelength (nm)
400
Example. A UV reactor equipped with "Sensor B" located 10 cm from the lamp sleeve
(10 cm water layer) is validated using coffee as a UV absorbing chemical. The UV reactor is
validated at three intensity setpoints, each tested at lowered UVT values of 95 percent, 90
percent, and 85 percent. The polychromatic bias values taken from Figure C.8 are 1.11,1.29,
and 1.56, for UVT values of 95 percent, 90 percent and 85 percent, respectively.
Example. A UV reactor equipped with "Sensor A" located 15 cm from the sleeve (20
cm water layer) is being considered at a WTP with a design UVT of 80 percent. From Figure
C.7, the polychromatic bias with coffee, (ignin sulphonate, and NOM are 1.7, 1.3, and 1.2,
respectively. Comparing these values, a strong incentive exists to select the UV absorber that
minimizes the polychromatic bias.
Determining the Random Uncertainty
The random uncertainty associated with monitoring and validation should be calculated
at an 80 percent confidence level using the uncertainty of the terms listed in Table C.5. The
expanded uncertainty should be calculated as the square root of the sum of squares of
uncertainties of each term.
If one challenge microorganism is used during validation, the uncertainty of the RED is
calculated using Equation C.8. If two challenge microorganisms are used, the uncertainty of the
RED is calculated using Equation C.13. The uncertainty of the interpolation is obtained from the
confidence bands of the equation used for the interpolation (see section C.4.9.10). The
uncertainty of the UV intensity sensors used during validation and used at the WTP should be
obtained from manufacturer data with supporting documentation as per Table C. 1 in section
UV Disinfection Guidance Manual
Proposal Draft
C-48
June 2003
-------
Appendix C. Validation of UV Reactors
Table C.5 Factors Impacting Expanded Uncertainty of
Dose Delivery Monitoring and Validation
Uncertainty
Measured RED
Any interpolation of RED as a function of flowrate, UVT, or UV intensity
Sensors used during validation (UV intensity, UVT)
On-line and reference sensors used at the WTP (UV intensity, UVT)
Lamp output quantification
C.2.2. If the dose monitoring approach uses a UVT monitor, include the measurement
uncertainty of the UVT monitor obtained from data provided by the manufacturer.
The uncertainty of lamp output quantification is zero if each lamp is monitored by an
individual UV intensity sensor. Otherwise, the uncertainty can be calculated using Equation
C.14:
1 *
Uncertainty = ' . — Equation C.14
where
0 = Standard deviation of lamp-to-lamp output expressed as a percentage of the mean
nj = Number of banks of lamps in series in the reactor
ni = Number of sensors monitoring each bank
The variability of UV output from lamp-to-lamp can be obtained from either life test or
field data on aged lamps.
Example. A UV reactor consists of two banks of four lamps. Each bank is equipped
with two UV intensity sensors. Dose delivery is monitored using the UV intensity setpoint
approach. The manufacturer provides data showing the standard deviation of lamp-to-lamp
output is 12 percent of the mean output at the end of lamp life. Thus, the lamp output
quantification uncertainty is 1.28xl2/(205x205) = 7.7 percent. When operating at a WTP, the on-
line UV intensity sensors have a measurement uncertainty of 20 percent. The on-line sensors
will be checked using a reference sensor with an uncertainty of 5 percent. During validation, the
flowmeter and UV intensity sensors had an uncertainty of 0.5 percent and 5 percent. The
collimated beam dose calculation has an uncertainty of 8 percent. The regression fit to the dose-
response of the phage has an uncertainty of 10 percent. The UV reactor is tested at peak flowrate
with the results shown in Table C.6. The uncertainty of the measured log inactivation is
determined as 4.4 percent. As summarized in Table'C.7, a total uncertainty of 26 percent is
calculated as the square root of the sum of the squares of the individual uncertainties.
UV Disinfection Guidance Manual C-49 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
Table C.6 Sample Calculation of the Log Inactivation Uncertainty
Influent
N
3.60x1 0?
4.90x1 05
4.10x105
Mean
St Dev.
T-statistic
Uncertainty
LogN
5.56
5.69
5.61
5.62
0.067
2.92
0.0729
Effluent
N
154
206
263
Mean
St Dev.
T-statistic
Uncertainty
LogN
2.19
2.31
2.42
2.31
0.116
2.92
0.126
Inactivation
Log Inactivation
Uncertainty
Uncertainty (%)
3.31
0.145
4.40
Table C.7 Sample Calculation of the Expanded Uncertainty
Uncertainty
Log Inactivation by reactor
Collimated beam dose calculation
Regression fit to UV Dose-Response Data
Validation UV intensity sensor
WTP on-line UV intensity sensor
WTP reference UV intensity sensor
Quantification of lamp-to-lamp variability
Expanded Uncertainty
Uncertainty
(%)
4.4.
8
10
5
20
5
7.7
26
Uncertainty
Squared
19
64
100
25
400
25
59
692
Determining the Safety Factor
The safety factor relating the RED measured during validation to the pathogen
inactivation requirements should be calculated as the product of the RED bias, the polychromatic
bias, and the expanded uncertainty as per Equation C. 10.
Example. MS2 inactivation of 2.0 log corresponding to an RED of 40 mJ/cm2 is
measured during validation with a LP reactor. The expanded uncertainty of 35 percent is
calculated. Because LP lamps are used, the polychromatic bias is 1.00. An RED bias of 2.0 is
determined using the observed UV sensitivity of MS2 and the UV sensitivity associated with a
3.0-log Cryptosporidium inactivation credit. A safety factor of (1+0.35) x 2.0 x 1.0 = 2.7 is
calculated. Hence, the Cryptosporidium RED demonstrated by validation is 40 / 2.7 =
15 mJ/cm2. Because the demonstrated Cryptosporidium RED is greater than the 3.0-log
requirement of 12 mJ/cm2, the UV reactor is validated for a 3.0-log Cryptosporidium inactivation
credit.
Example. The UV reactor in the above example is instead equipped with MP lamps
monitored with UV intensity sensors matching the spectral response of Sensor A. The UV
intensity sensors view the lamp through a 15 cm water layer. The UV reactor is validated using
lignin sulphonate at a maximum UVT of 80 percent. Using Figure C.7, the polychromatic bias
UV Disinfection Guidance Manual
Proposal Draft
C-50
June 2003
-------
Appendix C. Validation of UV Reactors
of 1.3 is determined. Thus, the safety factor is (1+0.35) x 2.0 x 1.3 = 3.5 and the
Cryptosporidium RED demonstrated by validation is 40/3.5 = 11.4 mj/cm2. In this case, the
demonstrated RED is less than the required RED of 12 mJ/cm2 for 3.0-log Cryptosporidium
inactivation credit, and the UV reactor should not be considered validated for 3.0-log
inactivation of Cryptosporidium. However, for a 2.5-log Cryptosporidium inactivation requiring
a dose of 8.5 ml/cm2, the RED bias is 2.1, resulting in a safety factor of 3.7 and a demonstrated
Cryptosporidium RED of 40 / 3.7 = 10.8 mJ/cm2. Because the demonstrated RED of 10.8
mJ/cm2 is greater than the target RED of 8.5 mJ/cm2, the UV reactor can be considered validated
for a 2.5 log Cryptosporidium inactivation credit.
C.4.11 Validation Test Report
The engineer responsible for third-party oversight should collect all documentation and
test results and prepare summary and detailed reports.
C.4.11.1 Summary Report
The summary report should describe the UV reactor validated under this protocol in
general terms including the following components:
' . Inlet and outlet conditions
. Number of UV lamps and their location within the reactor
• Lamp characteristics including type, electrical power consumption, and spectral
output
» Monitoring and.controls approach used for dose compliance
• Number of UV intensity sensors and their locations
'. . UVT monitor, if used
• Safety features used to ensure water disinfection
The summary report should provide the challenge microorganism UV dose-response,
including the regression fit and the confidence intervals. The report should tabulate each reactor
test condition evaluated, including the flowrate, UV intensity setpoint, UVT setpoint (if used),
calculated dose (if used), log inactivation achieved, and calculated RED. The number of samples
evaluated, the standard deviation of the influent and effluent samples, and the uncertainty of the
inactivation through the reactor should also be tabulated.
If interpolation of bioassay results is part of dose monitoring, tables or charts should
present the results of the interpolation.
UV Disinfection Guidance Manual C-51 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
If the reactor is evaluated under Tier 1, documentation should be provided supporting that
the validation met Tier 1 criteria. The report should state the pathogen credits that the UV
reactor can achieve based on the Tier 1 designation.
If the UV reactor is evaluated under Tier 2, documentation should be provided describing
the Tier 2 analysis including the determination of the RED bias, polychromatic bias, and
expanded uncertainty. For the expanded uncertainty, each term used in the calculation should be
provided. The report should state the pathogen credits that the UV reactor can achieve based on
the Tier 2 results.
Based on the values used to determine the safety factor applied to the validation data
(Tier 1 or 2), the summary report should specify all criteria for the measurement uncertainty of
the UV intensity sensors, and UVT monitors used at the WTP.
C.4.11.2 Detailed Report
The detailed report should provide a comprehensive description of the test methodology
that includes the following components:
. Identity and qualifications of personnel involved in the validation test
. UV reactor specifications
« UV intensity sensor specifications and calibration documentation
. Physical test set-up
. Summary of QA/QC procedures
« Materials and methods employed during the test
. Complete test results, including raw data and analyses performed
C.5 UV Reactor Validation Examples
This section provides examples of UV reactor validation for the following reactors and
monitoring approach combinations:
LP reactor using a single intensity setpoint (section C.5.1)
« LP reactor using multiple setpoints as a function of flowrate (section C.5.2)
. LP reactor using multiple setpoints as a function of flowrate and dose (section C.5.3)
UV Disinfection Guidance Manual C-52 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
•. MP reactor using a single intensity setpoint and UVT setpoint (section C.5.4)
» MP reactor using the calculated dose method for monitoring (section C.5.5)
C.5.1 LP Reactor Using a Single Intensity Setpoint
A UV reactor consists of two banks in series of nine LPHO lamps oriented perpendicular
to flow. Dose delivery is monitored using the UV intensity setpoint approach. Each bank is
equipped with one UV intensity sensor.
The UV reactor is considered for use at a WTP. The application requires a 2.5 log
inactivation credit of Cryptosporidium. The design flowrate and UVT at the WTP are 500 gpm
and 90 percent, respectively. The UV manufacturer states the lamp fouling/aging factor for the
reactor is 70 percent. During operation at a WTP, the on-line and reference UV intensity sensors
are expected to have a measurement uncertainty of 15 and 5 percent, respectively. The reactor
will operate at the WTP using a single intensity setpoint to indicate dose delivery over a flow
range of 100 to 500 gpm.
The reactor is validated using coffee as the UV absorber and MS2 as the challenge
microorganism. Figure C.I 3 gives the dose-response of the MS2 measured during validation
with a collimated beam. The dose-response is fitted using the following equation:
/N "\
Dose = 20.3 x log ^ +4.3
Figure C.I 3 also provides 80 percent confidence levels for the fit and the percent
uncertainty of the UV dose calculated from those confidence levels.
UV Disinfection Guidance Manual.
Proposal Draft
C-53
June 2003
-------
Appendix C. Validation of UV Reactors
Figure C.13 Dose-Response of the MS2 Challenge,
Microorganism Used in Example C.5.1
140
0
0 12 3 4
Log Inactivation
A reference sensor is used to monitor UV intensity during validation testing. The
intensity setpoint to be validated is determined by operating the reactor under design conditions
of 70 percent lamp output and 90 percent water UVT. Under these conditions, the UV intensity
sensor reads 5.0 mW/cm2.
Table C.8 gives the validation test conditions and results. The reactor is tested at
flowrates of 100 and 500 gpm with the intensity sensor reading 5.0 mW/cm2. At each flowrate,
the reactor is tested under conditions of low UVT - high lamp output and high UVT - low lamp
output. Each test condition is evaluated using five influent and five effluent samples.
Table C.8 Validation Test Conditions and Results for Example C.S.1
Test Conditions
Flow
(gpm)
100
100
500
500
UVT
(%)
98
84
98
84
Lamp
(%)
44
100
44
100
Test Results
UV
Intensity
(mW/crrn
4.98
4.90
4.98
4.92
Influent
(log)
4.97 ± 0.08
5.02 ±0.10
5.03 ± 0.06
5.02 ±0.10
Effluent
dog)
<0
<0
4.02 ± 0.08
3.52 + 0.18
Inactivation
(log)
>4.97
>5.02
1.00
1.49
RED
(mJ/cm2)
>105
>106
24.6
34.6
Note. Influent and effluent data presented as mean ± standard deviation.
Based on the results, the reactor is rated at an MS2 RED of 24.6 mJ/cm2 for a flow range
of 100 to 500 gpm and a sensor setpoint of 5 mW/cm2.
A Tier 2 analysis was used to assess if the reactor achieved 2.5 log Cryptosporidiwn.
UV Disinfection Guidance Manual
Proposal Draft
C-54
June 2003
-------
Appendix C. Validation of UV Reactors
RED Bias. Since a 2.5 log inactivation of Cryptosporidium requires a dose of
8.5 rnJ/cm2, the UV sensitivity of Cryptosporidium is defined as 8.5/2.5 = 3.4 mJ/cm2 per log
inactivation. Since 1.00-log MS2 inactivation occurred with a dose of 24.6 mJ/cm2, the UV
sensitivity of MS2 is defined as 24.6 mJ/cm2 per log inactivation. In Figure C.6, an RED of 9.2
and 21 mJ/cm2 occurs with a UV sensitivity of 3.4 and 24.6 mJ/cm2 per log inactivation,
respectively. Accordingly, the RED bias is 21/9.2 = 2.28.
Polychromatic Bias. The polychromatic bias equals 1 .0 because the UV reactor uses
LPHO lamps.
Expanded uncertainty. A t-statistic of 1 .53 is associated with 5 samples and an 80
percent confidence level. Using the standard deviations for the influent and effluent counts in
Table C.8, the uncertainty of the log inactivation through the reactor is calculated as follows
using Equation C.7:
( (0.06x1. S3)2 (O.Q8xl.53)2V^
I 5 + 5 I
Error = ^ - }— * 1 00% = 6:8%
1.00
The uncertainty of the collimated beam dose calculation was determined to be
8.9 percent. At a UV dose of 24.6 mJ/cm2, the uncertainty in the dose calculation based on the
confidence bands in Figure C.I 3 is 9.6 percent.
The uncertainties of the sensors used during validation and at the WTP are as follows:
, . Validation UV intensity sensor 5 percent
: . . WTP on-line UV intensity sensor 15 percent
. WTP reference UV intensity sensor 5 percent
The total uncertainty of the sensors is calculated according to the following equation:
Error = s2 + 1 52 + 52 2 = 16.6%
The UV vendor states the standard deviation of the UV output from lamp to lamp is 25
percent. Given two banks of lamps and one UV intensity sensor per bank, the uncertainty of the
lamp output is calculated as follows using Equation C.I 4:
Error- = 22.6%
Including each of these random uncertainty terms, the expanded uncertainty is calculated
as follows:
82+8.92 + 9.62, + 16.62+22.62)''2 =31.7%
UV Disinfection Guidance Manual C-55 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
Safety factor. Using Equation C.10, the safety factor is calculated as follows:
SF = (l + 0.317)x 2.28 x 1.00 = 3.00
Based on this safety factor value and the Cryptosporidium dose target for 2.5-log
inactivation credit, the MS2 RED demonstrated during validation should be as follows:
MS2 RED = 3.00 x 8.5 = 25.5 mJ/cm2
Because the demonstrated RED of 24.6 mJ/cm2 is less than this value, the reactor cannot
get 2.5-log Cryptosporidium inactivation credit operating at a sensor setpoint of 15 mW/cm2.
However^ with a 2.0-log Cryptosporidium credit target, the RED bias would be 2.6, resulting in a
safety factor of 3.42 and an MS2 RED target of 19.8 mJ/cm2. Because the demonstrated MS2
RED is greater than this value, the reactor can get 2.0 log Cryptosporidium credit operating at a
setpoint of 15 mW/cm2 over a flow range of 100 to 500 gpm.
The reactor does not meet Tier 1 criteria because the standard deviation of the UV output
from lamp-to-lamp is greater than 15 percent. If the reactor did meet all Tier 1 criteria, the
reactor would receive credit for 2.0-log Cryptosporidium based on a comparison of the
demonstrated MS2 RED of 24.6 mJ/cm2 with the dose criteria in Table C.3.
C.5.2 LP Reactor with a Intensity Setpoint Interpolation as a Function of Flow
A UV reactor consists of four banks of six LPHO lamps oriented perpendicular to the
flow. Dose delivery is monitored using the UV intensity setpoint approach. Each bank is
equipped with two UV intensity sensors.
The UV reactor is rated by the manufacturer for flows ranging from 0.9 to 2.4 mgd. The
manufacturer states that sensor setpoints of 6.0, 7.5, 10, and 14 mW/cm2 should indicate a 3.0-
log Cryptosporidium inactivation credit at flows of 0.9, L2, 1.7 and 2.4 mgd, respectively.
During operation at a WTP, the on-line and reference UV intensity sensors will have a
measurement uncertainty of 1 5 and 5 percent respectively.
The reactor is validated using lignin sulphonate as the UV absorber and MS2 as the
challenge microorganism. Figure C.14 gives the dose-response of the MS2 measured during
validation with a collimated beam. The dose-response is fitted using the following equation:
—2- -0.144
N I
UV Disinfection Guidance Manual
Proposal Draft
C-56
June 2003
-------
Appendix C. Validation of UV Reactors
Figure C.14 Dose-Response of the MS2 Challenge
Microorganism Used in Example C.5.2
.:rf:'h'"";:^i!i!-^DosfcResponse'JUF
••i-,.^$?^^L&vtt--&&l
Confidence intervals are fitted to the data at an 80 percent level.
Table C.9 gives the validation test conditions and results. The reactor is tested at four
flowrates, 0.9,1.2, 1.7 and 2.4 mgd, with the lamp power and UVT adjusted to give a UV
intensity sensor reading at the setpoint values. At each flowrate, the reactor is tested under
conditions of reduced UVT - maximum lamp output and maximum UVT - reduced lamp output.
A reference sensor with an uncertainty of 5 percent is used during validation to measure UV
intensity. Each test condition is evaluated using five influent and five effluent samples.
Table C.9 Validation Test Conditions and Results for Example C.5.2
Test Conditions
Flow
(mgd)
0.90
0.90
1.2
1,2
1.7
1.7
2.4
2.4
UVT
{%)
98
70
98
75
98
83
98
92
Lamp
(%)
37
100
45
100
61
100
83
100
Test Results
UV
Intensity
(mW/cm5)
6.15
6.06
7.48
7.46
10.1
10.1
13.8
13.7
Influent
(Logs)
5.99 ± 0.096
5.94 + 0.127
6.09 ±0.1 00
6.04 + 0.070
6.03 ±0.1 50
5.98 ±0.1 16
6.03 ±0.1 02
6.02 ±0.1 36
Effluent
(Logs)
2.95 + 0.080
3.21 ± 0.087
2.98 ±0.1 08
3.34 ± 0.088
3.27 ±0.1 12
3.45 + 0.120
3.37 ± 0.090
3.37 ± 0.062
Inactivation
Log
3.04
2.73
3.11
2.70
2.76
2.52
2.67
2.66
Uncertainty
(%)
3.8
5.4
4.5
4.0
6.5
6.2
4.9
5.4
RED
(mJ/cm2)
47.5
42.5
48.5
42.1
43.1
39.4
41.6
41.4
Table C.10 presents the MS2 RED and reactor setpoint assigned to each flowrate based
on the validation results. A Tier 2 analysis was used to determine the Cryptosporidium
inactivation credit that can be assigned to the reactor given the validation test results.
UV Disinfection Guidance Manual C-57 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
Table C.10 Summary of Validation Results for Example C.5.2
Flow
(mgd)
0.90
•1.2
1.7
2.4
UV Intensity
Setpoint
(mW/cm2)
6.06
7.46
10.1 .
13.7
MS2 RED
(mJ/cm2)
42.5
42.1
39.4
41.4
RED Bias. Since a 3.0-log inactivation credit for Cryptosporidium requires a dose of 12
mJ/cm2, the UV sensitivity of Cryptosporidium is defined as 12/3.0 = 4.0 mJ/cm2 per log
inactivation. The UV sensitivity of MS2 is 16 mJ/cm2 per log inactivation 42.5/2.73 =
16 mJ/cm2. In Figure C.6, an RED of 10 and 18 mJ/cm2 is associated with a UV sensitivity of
4.0 and 16 mJ/cm per log inactivation. Accordingly, the RED bias is 18/9.8 = 1.84.
Polychromatic Bias. The polychromatic bias equals 1 .0 because the UV reactor uses
LPHO lamps.
Expanded uncertainty. The uncertainty of the log inactivation through the reactor,
calculated using Equation C.7, is tabulated in Table C.9. A mean value of 5.1 percent is used as
the uncertainty of the log inactivation in this analysis. The uncertainty of the collimated beam
dose calculation was determined as 8.9 percent. For an RED near 40 mJ/cm2, the uncertainty in
the RED arising from the scatter in the dose-response in Figure C.I 4 is 4 percent.
The uncertainties of the sensors used during validation and at the WTP are as follows:
. Validation UV intensity sensor 5 percent
. WTP On-line UV intensity sensor 10 percent
. WTP Reference UV intensity sensor 5 percent
The total uncertainty of the sensors is calculated as follows:
Error = (52+102+52}^= 12.2%
The UV vendor states the standard deviation of the UV output from lamp to lamp is 15
percent. Given four banks of lamps and two sensors per bank, the uncertainty associated with
the number of sensors is calculated as follows using Equation C.14:
_, 1.28x15
Error = ,_. ._ = 6.8%
Including each of these random uncertainty terms, the expanded uncertainty is calculated
as follows:
UV Disinfection Guidance Manual
Proposal Draft
C-58
June 2003
-------
Appendix C. Validation of UV Reactors
Error = (5.12 +8.92 +42 +12.22 + 6.82)^ = 17.8%
Safety Factor. Using Equation C.10, the safety factor is calculated as follows:
SF=(l + 0.178)x 1.84x1.00 = 2.17
Cryptosporidium Credit. Using this safety factor, the target RED that should be
demonstrated during validation is 12 x 2.17 = 26 ml/cm2. Because the demonstrated RED of
39.4 ml/cm2 is greater, than this number, the UV reactor operating at the validated intensity
setpoints can get credit for 3.0-log Cryptosporidium inactivation.
The validation results can be used to define three strategies for operating the UV reactor.
at a WTP:
1. The UV reactor can operate using one intensity setpoint over the full range of
flowrates. In this case, a setpoint of 13.7 mW/cm2 can be used to indicate a 3.0-Iog
Cryptosporidium inactivation at all flows of 2.4 mgd or less.
2. The UV reactor can operate using multiple intensity setpoints where each setpoint
functions over a given range of flows. In this case, a setpoint of 13.7 mW/cm2 would
be used at all flows from 1.7 to 2.4 mgd, a setpoint of 10.1 mW/cm2 would be used at
all flows from 1.2 to 1.7 mgd, and a setpoint of 7.46 mW/cm2 would be used at all
flows from 0.90 to 1.2 mgd.
3. The UV reactor can be operated using intensity setpoints interpolated as a function of
flowrate using the validation data. In this case, using the plot of sensor setpoint
versus flowrate in Figure C.I5, a setpoint value of 11.7 mW/cm2 can be used at a
flow of 2 mgd to indicate an MS2 RED of 39.4 mJ/cm2.and hence a 3.0-log
Cryptosporidium inactivation credit.
UV Disinfection Guidance Manual C-59 June 2003
Proposal Draft
-------
Appendix C. Validation of UV Reactors
Figure C.15 Interpolation of Intensity Setpoint Values
Indicating an MS2 Dose of 39.3 mJ/cm2
100
f
o
Q.
C
« 20
80 -
60 -
40 -
6.1 mW/cm2
7.5mW/cm2
- - - 10.1 mW/cm2
— - 13.8mW/cm2
s
60
70 80 90
Water UV Transmittance (%)
100
The UV reactor and validation test conditions met all prerequisites to be considered under
Tier 1. The Tier 1 requirement for 3.0-log inactivation of Cryptosporidium by a LPHO reactor is
36 mJ/cm2. Since the RED demonstrated during validation is greater than this amount, the
reactor can receive 3.0-log Cryptosporidium inactivation credit under Tier 1.
Validation data obtained in this example can be related to design criteria by plotting
combinations of lamp output and water UVT that result in a given measured UV intensity
setpoint value. For example, Figure C.16 plots combinations of lamp output and water UVT that
result in the intensity setpoint values validated in Table C.9. Any combination of UVT and lamp
output along that curve can be used as design criteria for each setpoint value shown. For
example, a setpoint of 10.1 mW/cm2 indicates 3.0-log Cryptosporidium inactivation at a flow of
1.7 mgd. A setpoint of 10.1 mW/cm2 occurs with a combination of 70 percent lamp output and
93 percent UVT. Thus, the reactor could be used in a design application where the design flow,
UVT, and lamp fouling/aging factor are 1.7 mgd, 93 percent, and 70 percent, respectively. The
setpoint of 10.1 mW/cm2 is also obtained with a combination of 80 percent lamp output and 89
percent UVT. Thus, the reactor could also be used in a design application where the design flow,
UVT, and lamp fouling/aging factor is 1.7 mgd, 89 percent, and 80 percent, respectively.
UV Disinfection Guidance Manual
Proposal Draft
C-60
June 2003
-------
Appendix C. Validation of UV Reactors
Figure C.16 Combinations of Lamp Output and Water UV Transmittance that
Result in Given Sensor Setpoint Values
100
E 8° -
*-•
B- 60-
Q. 40'H
c
I 20 ^
6.1 mW/cm2
•7.5mW/cm2
10.1mW/cm2
13.8mW/cm2
.60 - 70 ; 80 , 90
i - Water UV Transmittance (%j
.100
C.5.3 LP Reactor with Intensity Setpoint Interpolation as a Function of Flow and
Target Inactivation
A UV reactor consists of twelve rows of twelve LPHO lamps oriented perpendicular to
flow. Dose delivery is monitored using the UV intensity setpoint approach. Each row is
equipped with one UV intensity sensor.
The UV reactor is rated by the UV vendor for flows ranging from 5 to 20 mgd. During
operation at a WTP, the on-line and reference UV intensity sensors will have a measurement
uncertainty of 15 and 5 percent, respectively.
The UV manufacturer wants to validate the UV reactor using test conditions that allow
interpolation of intensity setpoints as a function of flowrate and measured RED. Table C.I 1
gives the validation test conditions and results. To allow interpolation of sensor setpoints as a
function of flowrate, the reactor is tested at a three flowrates of 5,10, and 20 mgd. To allow
interpolation of sensor setpoints as a function of dose delivery, the reactor is tested at each
flowrate at setpoint values that the manufacturer states will result in MS2 RED values of 10,20,
and 30 mJ/cm . At each setpoint evaluated, the reactor is tested under conditions of reduced
UVT - maximum lamp output and maximum UVT - reduced lamp output. Each test condition is
evaluated using five influent and five effluent samples. A reference sensor with an uncertainty
of 5 percent is used during validation to measure the UV intensity.
UV Disinfection Guidance Manual
Proposal Draft
C-6J
June 2003
-------
Appendix C. Validation of UV Reactors
Table C.11 Validation Test Conditions and Results for Example C.5.3
Test Conditions
Flow
(mgd)
5
5
5
5
5
5
10
10
10
10
10
10
20
20
20
20
20
20
UVT
(%)
98
66
97.5
57
80
47
98
80
98
68
90.5
53
98
95
98
86
98
68
Lamp
(%)
,31
100
20
100
20
100
55
100
33
100
20
100
91
100
67
100
33
100
Test Results
UV
Intensity
(mW/cm5)
5.20
5.10
3.28
3.27
1.81
1.84
9.20
9.10
5.50
5.60
2.62.
2.63
15.1
15.2
11.2
11.2
5.50
5.60
Influent
(logs)
6.00 ± 0.074
5.98 ±0.1 36
6.02 ± 0.088
6.02 ±0.1 29
5.97 ± 0.075
5.96 ±0.1 18
6.09 ±0.141
5.96 ± 0.076
6.00 ± 0.068
6.00 ±0.1 30
5.96 ± 0.066
5.99 ±0.1 35
5.97 ± 0.080
5.97 ±0.1 17
6.03 + 0.117
5.91 ± 0.079
6.00 + 0.167
5.97 ± 0.032
Effluent
(logs)
1.86 ±0.098
2.68 ± 0.090
3.23 + 0.039
3.87 + 0.060
4.55 ±0.1 76
4.84 + 0.110
2.31 ±0.1 14
2.66 ±0.121
3.74 ± 0.070
4.04 ± 0.086
4.91 ± 0.072
4.92 + 0.076
2.93 + 0.104
2.97 ±0.1 25
3.62 + 0.121
3.82 ± 0.038
4.91+0.104
4.89 ±0.1 10
Inactivation
log
4.14
3.30
2.80
2.15
1.42
1,12
3.79
3.30
2.26
1.97
1.05
1.07
3.04
3.00
2.41
2.09
1.09
1.08
Uncertainty
(%)
2.8
4.7
3.3
6.3
12.8
13.7
4.6
4.1
4.1
7.6
8.8
13.7
4.1
5.4
6.7
4.0
17.2
10.2
RED
(mJ/cm2)
42.8
34.4
29.4
- 23.0
15.8
12.8
39.3
34.4
24.1
21.2
12.1
12.3
31.8
31.5
25.6
22.4
12.5
12.4
During validation, Hgnin sulphonate and MS2 are used as the UV absorber and challenge
microorganism, respectively. Figure C.I 7 gives the dose-response of the MS2 measured during
validation with a collimated beam apparatus. The dose-response is fitted using the following
equation:
Dose = 9.9 Ixlofi
+ 1.70
Confidence intervals are fitted to the data at an 80 percent level.
Table C.I2 presents the MS2 RED assigned to each reactor setpoint based on the
validation results. For a given flowrate, Figure C.I8 presents the measured RED interpolated as
a function of measured UV intensity.
UV Disinfection Guidance Manual
Proposal Draft
C-62
June 2003
-------
Appendix C. Validation of UV Reactors
Figure C.17 Dose-Response of the MS2 Challenge
Microorganism Used in Example C.5.3
Log Inactivation
PI*
tM ;
Table C.12 Summary of Validation Results for Example C.5.3
Flow
(mgd)
5
5
5
10
10
10
20
20
20
UV Intensity
tmW/cm2)
5.10
3.27
1.84
9.10
5.60
2.63
15.2
11.2
5.60
MS2 RED
(mJ/cm2)
34.4
23.0
12.8
34.4
21.2
12.3
31.5
22.4
12.4
A Tier 2 analysis was used to determine the Cryptosporidium inactivation credit that can
be assigned to the UV reactor given the validation test results. Because the validation results
will be interpolated as a function of dose delivery, the Tier 2 safety factors are determined as a
function of measured RED. For 1.5, 2.0, 2.5, and 3.0 log Cryptosporidium inactivation, Table
C.I 3 presents the RED bias as a function of the measured RED. RED bias values were
determined using the approach cited in section C.4.10.2. The polychromatic bias equals 1.0
because the UV reactor uses LPHO lamps.
UV Disinfection Guidance Manual
Proposal Draft
C-63
June 2003
-------
I
Appendix C. Validation of UV Reactors
Table C.13 RED Bias as a Function of the Target Pathogen
Target Inactivation and the Demonstrated RED in Example C.5.3
Demonstrated
RED
(mJ/cm2)
12
16
20
24
28
32
36
Challenge
Microorganism
UV Sensitivity
(mJ/cm2 per logL_
11.6
11.3
11.1
11.0
10.8
10.8
10.7
RED Bias for a Cryptosporidium log Inactivation of
1.5 log
1.98
1.96
1.95
194
1.93
1.93
1.92
2.0 log
1.89
1.87
1.86
1.85
1.84
1.84
1.83
2.5 log
1.76
1.74
1.73
1.72
1.71
1.71
1.70
3.0 log
1.65
1.63
1.62
1.61
1.61
1.60
1.60
Figure C.18 Measured RED as a Function of Sensor
Setpoint Values for Given Flowrates
.0
o
111
tt
40.0 -.
35.0 -
30.0 -
25.0 -
20.0 -
15.0 -
10.0 -
5.0 -
0.0
0.0
y = 6.64 + 0.89 y = 3.45 + 2.83
y= 1.97+ 1.09
5.0
10.0
15.0
20.0
Setpoint (mW/cm2)
Table C.I4 presents the random uncertainty terms and the expanded uncertainty as a
function of the demonstrated RED. Using data from Table C.I 1, Figure C.19 presents the
uncertainty of the log inactivation as a function of the demonstrated RED. An empirical fit to
this data was used to obtain the uncertainty of the log inactivation as a function of demonstrated
RED in Table C.I 4. The uncertainty of the RED due to the dose-response data was obtained
from Figure C.17. The uncertainty of the collimated beam dose calculation was 8.9 percent. The
uncertainties of the sensors used during validation and at the WTP are as follows:
. Validation UV intensity sensor 5 percent '
. WTP On-line UV intensity sensor 15 percent
« WTP Reference UV intensity sensor 5 percent
UV Disinfection Guidance Manual
Proposal Draft
C-64
June 2003
-------
Appendix C. Validation of UV Reactors
The total uncertainty of the sensors is calculated as follows:
=16.6%
Error = s2 +15
The UV vendor states the standard deviation of the UV output from lamp to. lamp is 25
percent. Given four rows of lamps and two sensors per row, the uncertainty associated with the
number of sensors is calculated as follows:
Error =
= 9.2%
Table C.14 Random Uncertainty Terms as a Function
of the Demonstrated RED for Example C.5.3
Demonstrated
RED (mJ/cm2)
12
14
16
18
20 .
, 22
24
Uncertainty
Challenge
Microorganism
Log
Inactivation
12.4
10.5
9.1
8.0
7.1
6.4
5.9
Challenge
Microorganism
Dose-response
11.7
9.7
8.2
7.1
6.3
5.6
5.1
Collimated
Beam Dose
Calculation
8.9
8.9
8.9
8.9
8.9
8.9
8.9
%)
Intensity
and
Flow
Sensors
16.6
16.6
16.6
16.6
16.6
16.6
16.6
Number
of .
Sensors
9.2
9.2
9.2
.9.2
9.2
9.2
9.2
Total
Expanded
Uncertainty
27.0
25.3
24.3
23.5
. 23.0
22.6
22.3
Figure C.19 Uncertainty of the Measured Log Inactivation
as a Function of Demonstrated RED
20
I*16
li
O)
o
TO -
5 -
0
,-1.10
0.0
10.0
20.0
30.0
40.0
50.0
Demonstrated RED (mJ/cm )
UV Disinfection Guidance Manual
Proposal Draft
C-65
June 2003
-------
Appendix C. Validation of UV Reactors
Safety factors calculated from the RED bias, polychromatic bias, and expanded
uncertainty are tabulated in Table C. 15. The safety factors were multiplied by the dose
requirements for Cryptosporidium to obtain target RED values which are tabulated in
Table G.I6. For a given demonstrated RED, the UV reactor can achieve a given level of
Cryptosporidium credit if the demonstrated RED is greater than the target RED. Interpolation of
these data can be used to identify the RED required to obtain a given level of Cryptosporidium
inactivation. Using this approach, an RED of 14.0, 19.1, and 25 mJ/cm2 is required to show 2.0,
2.5, and 3.0-log Cryptosporidium inactivation, respectively.
For each flowrate validated, interpolation of the data in Figure C.I 8 will provide the UV
intensity setpoints that will indicate an RED of 14.0, 19.1, and 25 mJ/cm2. For a given RED,
Figure C.20 presents those intensity setpoints as a function of flowrate. Interpolation of the data
in Figure C.20 can be used to identify the intensity setpoint required at a given flowrate. For
example, at a flowrate of 15 mgd, intensity setpoints of 4.7, 6.7, and 9.0 mW/cm2 can be used to
indicate Cryptosporidium log inactivation of 2.0,2.5, and 3.0.
Intensity setpoints obtained from Figure C.20 for a given design flow can be related to
design values of water UVT and lamp output using an approach similar to that used in
section C.5.2 (see Figure C.I6).
Table C.I 5 Safety Factors Applicable to Validation Results
Demonstrated
RED (mJ/cm2)
12
14
16
18
20
22
24
Safety Factors Needed Given a Cryptosporidium inactivation of
1.5 log
2.54
2.48
2.44
2.42
2.40
2.38
2.37
2.0 log
2.42
2.37
2.33
2.30
2.28
2.27
2.26
2.5 log
2.25
2.20
2.17
2.14
2.13
2.11
2.10
3.0 log
2.11
2.06
2.03
2.01
1.99
1.98
1.97
Table C.16 Comparison of Demonstrated RED and RED Required for
Various Log Inactivation of Cryptosporidium for Example C.5.3
Demonstrated
RED (mJ/cm2)
12
14
16
18
20
22
24
RED Needed to Achieve a Crj
1.5 log
9.9
9.7
9.5
9.4
9.4
9.3
9.2
2.0 log
14.0
13.7
13.5
13.3
13.2
13.2
13.1
/ptosporidium Inactivation of
2.5 log
19.1 .
18.7
18.4
18.2
18.1
17.9
17.9
3.0 log
25
25
24
24
24
24
24
UV Disinfection Guidance Manual
Proposal Draft
C-66
June 2003
-------
Appendix C. Validation of UV Reactors
Figure C.20 Intensity Setpoint Values Indicating Various Log
Inactivation of Cryptosporidium for example C.5.2
«~ 14 n
Ji 12 -I
E 10
c
'o
w
CO
8 -
6 -
4 -
2
0
a 2.0 log Crypto
A 2.5 log Crypto
O 3.0 log Crypto
y ='0.001 x2 + 0.517x + 0.872
= 0.003x2 + 0.346x +0.920
0.176x+>b.968
10 15
Flowrate (gpm)
20
25
C.5.4 MP Reactor Using a Single UV Intensity Setpoint and UV Transmittance
Setpoint
A UV reactor consists of two MP lamps oriented parallel to the flow. Each lamp is
monitored by a UV intensity sensor. Lamps are spaced 40 cm apart and 20 cm from the wall.
The lamp sleeve radius is 5 cm. The sensor is located on the wall, 20 cm away from the lamp.
The sensor's spectral response matches that of "Sensor A" in Figure C.10. Dose delivery is
indicated using the UV intensity and UVT setpoint approach.
The reactor is rated for a flow from 0.1 to 0.5 mgd. The reactor will be used with a
design UVT of 85 percent. The manufacturer states that the lamp output at the end-of-lamp life
will be 78 percent of the 100 hr burn-in value. The fouling factor for the reactor is 90 percent.
Accordingly, the lamp output factor for the reactor is 0.78 x 0.90 = 0.70. A UV intensity
setpoint of 2.8 mW/cm2 is obtained by measuring the UV intensity with the water UVT set to 85
percent and the lamp output lowered to 70 percent.
During operation at a WTP, the on-line and reference UV intensity sensors will have a
measurement uncertainty of 10 and 5 percent, respectively.
The UV reactor is validated using lignin sulphonate as a UV absorber and MS2 as a
challenge microorganism; The measured dose-response of the challenge microorganism is
provided in Figure C.14. Table C.17 gives the validation test conditions and results.
UV Disinfection Guidance Manual
Proposal Draft
C-67
June 2003
-------
Appendix C. Validation of UV Reactors
Table C.17 Validation Test Conditions and Results for
Example C.5.4 with the Sensor Located 20 cm from the Lamp
Test Conditions
Flow
(mgd)
0.1
0.5
0.5
0.5
UVT
(%)
85
85
82.7
93.5
Lamp
(%)
70
70
100
20
Test results
UV
Intensity
(mW/cnT)
2.9
2.9
2.8
2.9
Influent
(Logs)
6.00 ± 0.07
6.06 ± 0.08
5.99 ±0.11
6.12 ±0.11
Effluent
(logs)
0.00
3.04 ±0.16
2.13 + 0.14
4.52 + 0.07
Inactivation
Log
>6.0
0
3.03
3.85
1.60
Uncertainty
(%)
-
5.6
4.3
7.6
RED
(mj/cm2)
>82.6
47.1
60.0
24.8
The first two test conditions evaluate dose delivery at minimum and maximum flow with
the reactor operating at the intensity and UVT setpoint values. Based on these results, the UV
reactor is rated at an MS2 RED of 47.1 mJ/cm2 when operating at the setpoint conditions.
The last two test conditions evaluate the sensor position and the validity of using the UV
intensity and UVT setpoint approach for indicating dose delivery. As indicated, the UV reactor
delivers an MS2 RED of 60.0 mJ/cm2 when operating with peak lamp output and the UVT
lowered to 82.7 percent to give a measured intensity at the setpoint. The UV reactor delivers a
dose of 24.8 mJ/cm2 when operating at high UVT and lowered lamp output to give a measured
intensity at the setpoint value. In other words, an intensity setpoint of 2.9 mW/cm2 and a UVT
setpoint of 85 percent does not ensure the reactor delivers an MS2 RED of 47.1 mj/cm2.
The manufacturer has three options for resolving this problem:
• Relocate the sensor closer to the lamp.
• Switch from the dose monitoring method to the calculated dose approach.
• Switch from the dose monitoring method to the intensity setpoint approach either
with the sensor in its current location or with the sensor in a more optimized location.
In this example, the manufacturer chooses to relocate the sensor to 8 cm from the lamp
and revalidates the UV reactor. Table C.I8 gives the test conditions and results. In this case, a
UV intensity of 41.0 mW/cm2 is measured with the UV reactor operating with a UVT of 85
percent and a lamp output of 70 percent. This value is greater than the UV intensity measured
with the sensor on the wall because the sensor is located closer to the lamp. Based on the results,
the UV reactor is rated at an MS2 RED of 48.5 mJ/cm2 when operating at setpoint conditions of
85 percent UVT and a 41.0 mW/cm2 UV intensity value. With the measured intensity at the
intensity setpoint value, the UV reactor delivers an RED greater than 48.5 mJ/cm2 when
operating with a UVT greater than the UVT setpoint value and an RED less than 48.5 mj/cm2
when operating with a UVT less than the setpoint value. Thus, the intensity sensor is properly
located for using the intensity and UVT setpoint approach for indicating dose delivery and the
setpoints will ensure the dose delivery meets an RED of 48.5 mJ/cm2.
UV Disinfection Guidance Manual
Proposal Draft
C-68
June 2003
-------
Appendix C. Validation of UV Reactors
Table C.18 Validation Test Conditions and Results for
Example C.5.4 with the Sensor Located 12 cm from the Lamp
Test Conditions
Flow
(mgd)
0.1
0.5
0.5
0.5
UVT
(%)
85
85
75
98
Lamp
(%)
70
70
100
46
Test Results
UV
Intensity
(mW/cms)
41.0
41.0
41.4
41.1
Influent.
(Logs)
6.00 ±0.07
6.01 ±0.10
6.03 + 0.07
5.96 ±0.12
Effluent
(Logs)
0.00
2.90 ±0.10
3.26 + 0.18
1.59 + 0.12
Inactivatlon
Log
>6.00
3.12
2.77
4.37
Uncertainty
(%)
-
4.3
6.7
3.7
RED
(mJ/cm2)
•> 82.6
48.5
43.0
68.1
A Tier 2 analysis is used to determine the Cryptosporidium inactivation credit that can be
assigned to the UV reactor given the validation test results.
RED Bias. Since a 3.0 log inactivation of Cryptosporidium requires a dose of
12 mJ/cm2, the UV sensitivity of Cryptosporidium is defined as 12/3.0 = 4 mJ/cm2 per log
inactivation. The UV sensitivity of MS2 is 48.5/3.12 =15.5 mJ/cm2. In Figure C.6, a RED of
9.78 and 18.0 mJ/cm2 is associated with UV sensitivities of 4 and 15.5 mJ/cm2 per log
inactivation, respectively. Accordingly, the RED bias is 18.0/9.78 = 1.84.
Polychromatic Bias. The sensor-to-lamp water layer is'3 cm. For a sensor with the
response of "Sensor A" in Figure C.10, a polychromatic bias 1.00 is obtained from Figure C.7
for a UVT of 85 percent.
Expanded uncertainty. The uncertainty of the log inactivation through the UV reactor
calculated using Equation C.7 is 4.3 percent. The uncertainty of the collimated beam dose
calculation was 8.9 percent. The uncertainty in the RED arising from the scatter in the dose-
response obtained from Figure C.14 is 3.9 percent at an RED of 48.5 mJ/cm2. The uncertainties
of the sensors used during validation and at the WTP are as follows:
Validation UV intensity sensor
WTP On-line UV intensity sensor
5 percent
10 percent
• WTP Reference UV intensity sensor 5 percent
The total uncertainty of the sensors is calculated as follows:
Error = s2 + 102 + 52 = 12.2%
' Because each lamp is monitored, the uncertainty knowing the output of the lamps is 0
percent. Including each of these random uncertainty terms, the expanded uncertainty is
calculated as follows:
Error = (4.32 +8.92 +3.92 +12.22 +02)^ =16.2%
UV Disinfection Guidance Manual
Proposal Draft
C-69
June 2003
-------
Appendix C. Validation of UV Reactors
Safety factor. Using Equation C.10, the safety factor is calculated as follows:
SF = (l + 0.162)x 1.84x1.00 = 2.14
Cryptosporidium Credit. Using this safety factor, the target RED that should be
demonstrated during validation is 12 x 2.14 = 26 mJ/cm2. Because the demonstrated RED is
greater than this number, the UV reactor operating at or above the validated intensity and UVT
setpoints can get credit for 3.0-log Cryptosporidium inactivation.
C.5.5 MP Reactor Using Calculated Dose Monitoring
A UV reactor consists of twelve MP lamps oriented perpendicular to the flow. Each
lamp is monitored by a UV intensity sensor whose spectral response matches that of "Sensor A"
in Figure C.10. The UV intensity sensors view the UV lamps through a 15 cm water layer.
During operation at a WTP, the on-line and reference UV intensity sensors will have a
measurement uncertainty of 10 and 5 percent respectively. During validation, a reference sensor
is used to measure UV intensity.
The UV reactor is validated at flows ranging from 10 to 40 mgd using lignin sulphonate
as a UV absorber and MS2 as a challenge microorganism. Figure C.3 gives the dose-response of
the challenge microorganism. The lamp's power supplies vary lamp power from 30 to 100
percent. The UV reactor is validated at flowrate, UVT, and lamp power combinations that give a
calculated UV doses of 30,20 and 10 mJ/cm2. Table C.I9 gives the validation test conditions
and results.
Table C.19 Validation Test Conditions and Results for Example C.5.5
Test Conditions
Flow
(mgd)
40
40
40
20
20
20
10
10
40
40
40
20
20
40
40
40
20
UVT
(%>
98
90
82.8
90
85
75
79
75
96.5
85
73.5
85
75
84.5
75
80
70
Lamp
(%)
40.5
68
100
33.8
44.5
70
30
35.5
30
59.5
100
30
47
30
47
38
30
Test Results
Calculated
RED
(mJ/cm2)
30.1
30.2
30.2
30.0
30.1
29.7
30.3
. 30.1
20.1
20.1
19.9
20.3
19.9
9.9
10.0
10.1
10.4
Intensity
(mW/n/)
22.7
11.8
5.8
5.9
3.6
1.2
1.0
0.6
13.5
4.9
1.3
2.5
0.8
2.3
0.8
1.4
0.2
Inactivation
Log
2.38
2.43
2.38
2.57
2.36
2.31
2.22
2.76
1.42
1.34
1.31
1.75
1.72
0.96
1.24
0.93
0.86
Uncertainty
(%)
3.0
3.2
4.2
3.0
3.6
3.5
4.8
2.9
4.0
6.2
7.3
4.0
2.9
6.5
8.1
7.8
10.4
Measured
RED
(mJ/cm2)
35.2
36.3
35.1
39.1
34.7
33.8
32.0
43.3
17.7
16.3
15.9
23.1
22.7
10.8
14.9
10.4
9.5
UV Disinfection Guidance Manual
Proposal Draft
C-70
June 2003
-------
Appendix C. Validation of UV Reactors
Figure C.2I provides a plot of the measured RED as a function of the calculated RED.
As shown, there is a range of measured RED values associated with a given calculated RED.
The UV reactor is rated at the lower end of that range for a given calculated dose. A power
function (y=AxB) is used to define the relationship between the calculated RED and the lower
bound of the measured RED. •
Figure C.21 Relationship Between Measured and Calculated
Dose for the MP Reactor in Example C.5.5
10 ' 20 . 30:
Calculated RED (mJ/cm2)
A Tier 2 analysis was used to determine the calculated RED values required for 3.0-log
credit for Cryptosporidium. For various log inactivation credit values for Cryptosporidium,
Table C.20 gives the RED bias as a function of the MS2 RED predicted from the calculated dose
using the power function in Figure 21.
Table C.20 RED Bias as a Function of the Target Pathogen
Target Inactivation and the Calculated Dose in Example C.5.5
Calculated
Dose
(mJ/cm2)
10
12
14
16
18
20
22
24
• 26
28
30
MS2 RED
(mJ/cm2)
4.8
6.6
8.6
10.8
13.3
15.9
18.7
21.8
25.0
28.4
32.0,
MS2 Log
Inactivation
0.43
0.58
0.75
0.92
1.11
1.30
1.50
1.69
1.88
2.07
2.25 -
MS2 Sensitivity
(mJ/cm2 per log
Inactivation)
11.2
11.3
11.5
11.7
11.9
12.2
12.5
12.9
13.3
13.7
14.2
RED Bias for Cryptosporidium
log inactivations of
3.0
log
1.61
1.62
1.63
1.64
1.65
1.674
1.68
1.70
1.72
1.74
1.77
2.5
log
1.73
1.74
1.76
1.77
1.78
1.80
1.82
1.84
1.86
1.88
1.91
2.0
log
1.86
1.87
1.89
1.90
1.92
1.93
1.95
1.97
2.00
2.02
2.05
1.5
log
1.95
1.97
1.98
1.99
2.01
2.03
2.05
2.07
2.10
2.12
2.15
1.0
log
1.99
2.00
2.01
2.03
2.04
2.06
2.08
2.11
2.13
2.16
2.19
UV Disinfection Guidance Manual
Proposal Draft
C-71
June 2003
-------
Appendix C. Validation of UV Reactors
Table C.21 gives the random uncertainty terms and the resulting total expanded
uncertainty as a function of the MS2 RED and calculated dose. Using data from Table C.I 9,
Figure C.22 presents the uncertainty of the log inactivation as a function of measured MS2. An
empirical fit to this data was used to predict the uncertainty of the log inactivation as a function
of MS2 RED in Table C.21. The uncertainty of the RED due to the dose-response data was
obtained from Figure C.3. The uncertainty of the collimated beam dose calculation was 8.0
percent. The uncertainties of the sensors used during validation and at the WTP are as follows:
Validation UV intensity sensor
WTP On-line UV intensity sensor
5 percent
10 percent
« WTP. Reference UV intensity sensor 5 percent
The total uncertainty of the sensors is calculated as follows:
Error = s2 + 102 + 52 2 = 12.2%
Because each lamp is monitored by a UV intensity sensor, the uncertainty associated with
quantifying lamp output is zero.
Table C.21 Random Uncertainty Terms as a Function
of the Calculated Dose for Example C.5.5
Calculated
Dose
(mJ/cm2)
10
12
14
16
18
20
22
24
26
28
30
MS2RED
(mJ/cm2)
4.8
6.6
8.6
10.8
13.3
•15.9
18.7
21.8
25.0
•28.4
32.0
Uncertainty (%)
Challenge
Microorganism
Log Inactivation
13.9
11.1
9.2
7.8
6.8
5.9
5.3
4.8
4.3
3.9
3.6
Challenge
Microorganism
Dose-response
1.9
1.3
0.9
0.7
0.5
0.4
0.3
0.3
0.2
0.2
0.2
Collimated
Beam Dose
Calc
8.0
8.0
8.0
8.0
• 8.0
8.0
8.0
8.0
8.0
8.0
8.0
Intensity
Sensors
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
12.2
Total
Expanded
Uncertainty
20.3
18.4
17.3
16.6
16.1
15.8
15.6
15.4
15.3
15.1
15.1
Table C.22 gives the polychromatic bias as a function of the UVT. The polychromatic
bias values were taken from Figure C.7 for sensor "A" located with a 15 cm water, layer and
lignin sulphonate as the UV-absorbing chemical.
UV Disinfection Guidance Manual
Proposal Draft
C-72
June 2003
-------
Appendix C. Validation of UV Reactors
Figure C.22 Uncertainty of the Measured Log Inactivation
as a Function of Demonstrated RED
m c
«£
£•!-
12.0 ,-
10.0 -
8.0 -
. 6.0 -
4.0 -
.2.0 -
0:0
-O
= 42.598xr07118
0.0 10.0 20.0 ,. 30.0 40:0 50.0
Measured RED (mJ/cm2)
Table C.22 Polychromatic Bias for Example C.5.5
UVT
(%)
98
95
90
85
80
75
Polychromatic
Bias
1.03
1.06
1.10
1.20
1.30
1.55
Safety factors calculated from the RED bias, polychromatic bias, and expanded
uncertainty are tabulated in Table C.23. The safety factors were multiplied by the 3.0-log dose
requirement for Cryptosporidium of 12 mJ/cm2 to obtain target RED values which are tabulated
in Table C.24. For a given calculated dose, the UV reactor can achieve receive 3.0-log
Cryptosporidium credit if the measured MS2 RED associated with that calculated dose is greater
than the target RED.
Table C.25 presents the calculated dose needed to achieve a given level of
Cryptosporidium inactivation and the lower limit of UVT over which the calculated dose applies.
The values in Table C.22 for 3.0-log inactivation credit were obtained from Table C.21. The
values for other log inactivation credit levels were obtained by repeating the analysis in Tables
'C.23 and C.24.
UV Disinfection Guidance Manual
Proposal Draft
C-73
June 2003
-------
I
Appendix C. Validation of UV Reactors
Table C.23 Safety Factors for 3-Log Cryptosporidium Inactivation as a Function
of the Calculated Dose for Example C.5.5
Calculated
Dose
(mJ/cm2)
10
12
14
16
18
20
22
24
26
28
30
MS2
RED
(mJ/cm2)
4.8
6.6
8.6
10.8
13.3
15.9
18.7
21.8
25.0
28.4
32.0
Safety Factors for 3.0-log Cryptosporidium Inactivation for UVT of
98%
1.99
1.97
1.97
1.97
1.98
1.99
2.00
2.02
2.04
2.07
2.09
95%
2.05
2.03
2.02
2.03
2.03
2.05
2.06
2.08
2.10
2.13
2.16
90%
2.13
2.10
2.10
2.10
2.11
2.12
2.14
2.16
2.18
2.21
2.24
85%
2.32
2.30
2.29
2.29
2.30
2.32
2.34
2.36
2.38
2.41
2.44
80%
2.51
2.49
2.48
2.48
2.49
2.51
2.53
2.55
2.58
2.61
2.64
75%
3.00
2.97
2.96
2.96
2.97
2.99
3.02
3.04
3.08
3.11
3.15
Table C.24 Comparison of the Target MS2 RED Needed to Demonstrate 3.0-Log
. Cryptosporidium Inactivation Credit to the Calculated Dose and Measured
MS2 RED
Calculated
Dose
(mJ/cm2)
10
12
14
16
18
20
22
24
26
28
30
MS2
RED
(mJ/cm2)
4.8
6.6
8.6
10.8
13.3
15.9
18.7
21.8
25.0
28.4
32.0
Target MS2 RED (mJ/cnV1)
98%
23.9
23.7
23.6
23.6
23.7
23.9
24.1
24.3
24.5
24.8
25.1
95%
24.6
24.3
24,3
24.3
24.4
24.6
24.8
25.0
25.2
25.5
25.9
90%
25.5
25.3
25.2
25.2
25.3
25.5
25.7
25.9
26.2
26.5
26.8
85%
27.8
27.6
27.5
27.5
27.6
27.8
28.0
28.3
28.6
28.9
29.3
80%
30.1
29.9
29.8
29.8
29.9
30.1
30.4
30.6
31.0
31.3
31.7
75%
35.9
35.6
35.5
35.5
35.7
35.9
36.2
36.5
36.9
37.3
37.8
Table C.25 Dose and UVT Alarm Setpoints for
Various Log Inactivation Credit Levels of Cryptosporidium
3.0 log
Dose
(mJ/cm2)
25
26
27
28
29
30
-
UVT
(%)
98.0
95.6
90.4
85.9
82.2
79.2
-
2.5 log
Dose
(mJ/cm2)
22
23
24
25
26
27
28
UVT
(%)
94.7
88.5
83.9
80.5
78.1
76.3
75.0
2.0 log
Dose
(mJ/cm2)
19
20
21
22
23
-
-
UVT
(%)
88.8
83.2
79.4
76.8
75.1
-
-
1.5 log
Dose
(mJ/cm2)
15
. 16
17
18
-
-
-
UVT
(%)
93.0
84.7
79.5
76.3
-
-
-
1.0 log ,
Dose
(mJ/cm2)
12
13
14
15
-
-
-
UVT
(%)
89.6
80.9
76.3
77.5
-
-
-
UV Disinfection Guidance Manual
Proposal Draft
C-74
June 2003
-------
Appendix C. Validation of UV Reactors
• The data in Table C.25 represents calculated dose and UVT alarm setpoints that can be
used to ensure the reactor delivers a given log inactivation of Cryptosporidium, Alternatively, as
shown in Table C.26, a single calculated dose alarm setpoint can be defined over the validated
range of UVT.
Table C.26 Dose Setpoints for Various Log Inactivation Credit Levels of
Cryptosporidium
Cryptosporidium
Log Inactivation
Credit
1.0
1.5
2.0
2.5
3.0
Calculated Dose
Setpoint
(mJ/cmzy
14
18
23
28
30
UVT Range
(%)
75-98
75-98
75-98
75-98
79-98
C.6 References
APHA/AWWA/WEF. 1995. Standard methods for the Examination of Water and Wastewater,
19th Edition. American Public Health Association, American Water Works Association,
and Water Environment Federation, Washington, DC.
Chang, J.C.H., S.F. Osoff, D.C. Lobe, M.H. Dorfman, C.M. Dumais, R.G. Quails and J.D.
Johnson. 1985. UV inactivation of pathogenic and indicator microorganisms. Applied
and Environmental Microbiology 49, no. 6:1361-1365.
Draper, N. and Smith, H. 1981. Applied regression analysis, Second Edition. New York: Wiley.
DVGW. 1997. UV Disinfection Devices for Drinking Water Supply^Requirements and
Testing. German Gas and Water Management Union (DVGW), Bonn, Germany.
NWRI/AwwaRF. 2000. Ultraviolet Disinfection Guidelines for Drinking Water and Water
Reuse. National Water Research Institute and the AwwaRF
ONORM. 2001. ONORM M 5873-1 Plants for the Disinfection of water Using Ultraviolet
Radiation: Requirements and Testing Low Pressure Mercury Lamp Plants.
Osterreichisches Normungsinstitut, Vienna, Austria.
Petri, B.M., G. Fang, J.P. Malley, D.C. Moran, and H. Wright. 2000. Ground water UV
Disinfection: Challenges and Solutions. Proceedings of the American Water Works
Association Water Quality Technology Conference. Salt Lake City, UT, November 5-9,
2000.
UV Disinfection Guidance Manual
Proposal Draft
C-75
June 2003
-------
Appendix C. Validation of UV Reactors
Petri, B. and D. Olson. 2001. Bioassay-validated numerical models for UV reactor design and
scale-up IUVA First International Congress on UV Technologies, Washington, DC, June
14-16,2001.
Wright, N.G. and D.M. Hargreaves. 2001. The use of CFD in the evaluation of UV treatment
systems Journal of Hydroinformatics 3:59-70.
UV Disinfection Guidance Manual
Proposal Draft
C-76
June 2003
-------
Appendix D. Microbiological Methods
D.1 General Recommendations
The challenge microorganism used to validate UV reactors should be cultured and
analyzed by a laboratory staffed by professional microbiologists and equipped to perform
microbiological examinations as per Standard Methods for the Examination of Water and
Wastewater (APHA et al. 1998, sections 1020-1050). Protocols for culturing the challenge
microorganism and measuring its concentration should be based on published and peer-reviewed
methods and should be clearly defined and followed. Measurement of the concentration of the
challenge microorganism before and after exposure to UV light should be initiated within
24 hours of exposure. If the challenge microorganism has the ability to photorepair, exposure of
samples to visible light should be kept to a minimum.
Because MS2 bacteriophage (MS2) and B. subtilis spores are commonly being used as
challenge microorganisms for UV reactor validation, the following sections describe procedures
that can be used for preparing stock solutions of MS2 and B. subtilis spores and assaying the
concentration of those microorganisms in water samples. Procedures for preparing stock
solutions can be scaled to provide the volumes needed for UV reactor validation. Alternative
procedures and challenge microorganisms can be used if they are acceptable to the State.
Section F.I provides a rational for selecting challenge microorganisms.
0.2 MS2 Bacteriophage Stock Preparation
MS2 (ATCC 15597-BI) can be propagated using a variety of host bacteria including
Escherichia coli C3000 (ATCC 15597), E. coli F-amp (ATCC 700891), and others (Meng and
Gerba 1996, Oppenheimer et al. 1993, NWRI/AwwaRF 2000). The following propagation
method was adapted from NWRI/AwwaRF (2000):
1. Inoculate sterile tryptic soy broth (TSB) (DIFCO, Detroit, Michigan) with host
bacterium transferred from a single colony grown on a nutrient agar plate. Incubate
the culture with constant stirring at 35 to 37°C for 18 to 24 hours.
2. Transfer 0.5 mL of the host bacterium culture to 50 mL of fresh TSB and incubate at
35 to 37°C for 4 to 6 hours with continuous shaking at 100 Hz to obtain a culture in
its log growth phase (approx. 3xl08 cfu/mL) (cfu = colony forming units).
3. Dilute stock MS2 using Tris-buffered saline (pH 7.3) to a concentration of
approximately 108 pfu/mL (pfu = plaque forming units).
4. Add ImL of diluted MS2 stock solution to the 50 mL volume of E. coli in TSB and
incubate overnight at 35 to 37°C.
5. Centrifuge the MS2/£. coli culture at 8000 x g (g = 9.82 m/s2) for 10 minutes at 4°C
to remove cellular debris.
UV Disinfection Guidance Manual D-l June 2003
Proposal Draft
-------
Appendix D. Microbiological Methods
6. Filter the supernatant by passing it through a 0.45 urn low protein-binding filter.
7. Assay the concentration of MS2 in the stock solution as per section D.3.
8. Collect and refrigerate the filtrate at 4°C and use within a one-month period.
Propagation should result in a highly concentrated stock solution of essentially mono-
dispersed phage whose UV dose-response follows first order kinetics with minimal tailing.
Figure D.I presents the dose-response of MS2 as reported in the literature. Over the range of
REDs demonstrated during validation testing, the mean dose-response of the MS2 stock solution
should lie within the 90 percent prediction interval of the mean response in Figure D.I. Over a
dose range of 0 to 120 mJ/cm2, the predictions intervals may be defined using the following
equations:
Lower Bound'.log Inactivation =-I Ax. IQ~* xDose2 +7.6xlO"2 xDose
Upper Bound: \oglnactivation = -9.6 x 10~5 x Dose* + 4.5 x 10~2 x Dose
Figure D.1 UV Dose-Response of MS2
Meng and Gerba, 1996
Oppenheimer et al, 1993
Sommeretal, 1998
Tree et al, 1997
Havelaaretal, 1990
Nieuwstad et al, 1994
•Mean
Mean Response 90% Prediction Interval
20
UV Dose (mJ/ctTi)
UV Disinfection Guidance Manual
Proposal Draft
D-2
June 2003
-------
Appendix D. Microbiological Methods
D.3 MS2 Assay
The concentration of MS2 (ATCC 15597-B1) in water samples can be assayed using the
agar overlay technique with E coli (ATCC 15597) as a host bacterium (Adams 1959, Yahya et
al. 1992, Oppenheimer et al. 1993, Meng and Gerba 1996). The following procedure can be
used:
1. Inoculate TSB (Difco, Detroit, MI) with the host bacterium and incubate at 35 to
i 37°C for 18 to 24 hours to obtain an approximate concentration of 108 CFU/mL.
2. Transfer 1 mL of the culture to 50 mL of fresh TSB and incubate at 35 to 37°C for 4
• to 6 hours with continuous shaking at 100 Hz to obtain a culture in its log growth
phase.
3. Obtain serial dilutions of the MS2 sample using a 0.001 M phosphate-saline buffer or
TSB.
4. Combine and gently stir 1 mL of host cell solution, 0.1 mL of diluted MS2 sample,
• and 2 to 3 mL of molten'tryptic soy agar (TSA) (0.7 percent agar, 45 - 48°C) (Difco,
Detroit,- MI).
5. Pour the mixture onto solidified TSA (1.5 percent agar) contained within Petri dishes.
The time between the mixing the MS2 sample with the E. coli host and the plating of
the top agar layer should not exceed 10 minutes. After plating, the agar should
harden in 10 minutes.
6. After the top agar layer hardens, cover, invert the Petri dishes, and incubate 16 to
24 hours at 35 to 37°C.
7. Count the plaques with the aid of a colony counter. Plaques are identified as clear
circular zones 1 to 10 mm in diameter in the lawn of host bacteria.
8, Record the number of plaques per dish, and the MS2 sample volume and dilution. If
it is not possible to distinguish individual plaques because of confluent growth, record
the plate counts as "TNTC" (too numerous to count).
i . -
9. Calculate the MS2 concentration in the water samples:
E«(
Concentration = •=— - Equation D.I
where
nj = The number of counts on each plate
Vj = The volume of undiluted sample used with each plate
Example. A water sample containing MS2 was diluted 10, 100 and 1,000-fold using a
0.1 mL aliquot dilution of the sample for each. Each dilution was assayed in triplicate. Plaque
j
UV Disinfection Guidance Manual D-3 June 2003
Proposal Draft
-------
Appendix D. Microbiological Methods
forming units observed on the plates were 2, 5 and 6 for the 1,000- fold diluted sample and 32,
40, and 47 for the 100-fold diluted sample. With the 10-fold dilution, plate counts were too
numerous to count. The concentration in the original sample is calculated as follows:
Annnn .. .
Concentration = —± - - '— - = 40,000 pfulmL
D.4 Bacillus Subtilis Spore Preparation
Bacillus subtilis spores (ATCC 6633) can be propagated using Schaeffer's media
(Munakata and Rupert 1972, Sommer et al. 1995, DVGW 1997). The following propagation
method was adopted from DVGW (1997):
1. Prepare Columbia agar (Oxoid CM 331) as a 1 L solution of 23.0 g special peptone
(Oxoid L 72), 1.0 g starch, 5.0 g NaCl, and 10.0 g agar (Oxoid L 11) in distilled
water. Adjust pH to 7.0 and autoclave 15 minutes at 121°C.
2. Prepare the sporulation media as a 1 L solution of 280 mg MgSCVI-kO, 1.11 g KC1,
3.1 mg FeSCv7H2O, and 8.9 g nutrient broth (Oxoid CM 67) in distilled water.
Adjust the pH to 7.0 and autoclave it for 15 minutes at 121°C.
3. Inoculate Columbia agar plates (Oxoid CM 331) with three smears of B. subtilis and
incubate 24 hours at 37°C.
4. Inoculate 300 mL of sporulation media with three colonies collected from the agar
plates.
5. Incubate the sporulation media 72 hours at 37°C on a shaker operating at 2 Hz.
6. Sonicate the resulting culture for 10 minutes at 50 kHz and 10°C.
7. Harvest the spores by centrifuging 80 mL aliquots at 5000 g for 10 minutes and 10°C.
8. Wash the spores three times by resuspending in 20 mL of distilled water and
centrifuging at 5000 x g for 10 minutes at 10°C.
9. Resuspend the washed spores in 100 mL of 0.001 M phosphate-saline buffer.
10. Inactivate the vegetative B. subtilis by heat treatment at 80°C for 10 minutes.
11. Sonicate the resulting culture for 10 minutes at 50 kHz and 10°C.
12. Collect the resulting stock solution and assay the B. subtilis spore concentration as per
section D.5.
13. Refrigerate the filtrate at 4°C and use within a one-month period. Sonicate for
10 minutes at 50 kHz and 10°C before use.
UV Disinfection Guidance Manual D-4 June 2003
Proposal Draft
-------
Appendix D. Microbiological Methods
Propagation should result in a highly concentrated stock solution of mono-dispersed
B. subtilis spores with a UV dose-response that follows the dose-response reported in the
literature and presented in Figure D.2. Over the range of reduction equivalent doses (REDs)
demonstrated during validation testing, the mean dose-response of the B, subtilis stock solution
should lie within the 90 percent prediction interval of the mean response provided in Figure D.2,
Over a dose range of 0 to 70 mj/cm2, the predictions intervals may be defined using the
following equations:
- Lower Bound: \oglnactivation = -2.0 x 10~3 x Dose3 + 2.7 x 10"3 x Dose2 -5.3 x 10~2 x Dose
Upper Bound: log Inactivation = 5.7 x 10^ x Dose2 +4.3xlO"2x Dose
Figure D.2 UV Dose-Response of Bacillus Subtilis Spores
DVGW, 1997
Hoyer, 2002
Chang etal, 1985
Sommeretal, 1996
Sommeretal, 1995
Sommeretal, 1998
Sommerand Cabaj, 1993
Sommer and Cabaj, 1993
•Mean
90% Prediction Interval of the Mean Response
50
100
150
UV Dose (mJ/cm )
UV Disinfection Guidance Manual
Proposal Draft
D-5
June 2003
-------
Appendix D. Microbiological Methods
D.5 Bacillus Subtilis Spore Assay
The concentration of B. subtilis spores (ATCC 6633) in water samples can be assayed by
the plate method using plate count agar. The following procedure was adopted from Deutscher
Verein des Gas- und Wasserfaches (DVGW) (1997):
1. Prepare plate count agar (Oxoid CM 325) as a 1 L solution of 5.0 g casein peptone
(Oxoid L 42), 2.5 g yeast extract (Oxoid L 21), 1.0 g glucose, and 9.0 g agar (Oxoid L
11) in distilled water. Adjust pH to 6.8 ± 0.2 and autoclave for 15 minutes at 121° C.
2. Obtain serial dilutions of the B. subtilis spore sample using 0.001 M phosphate-saline
buffer.
3. Vacuum filter 100 mL of diluted sample through a 47 mm x 0.45 um membrane filter
(Gelman Sciences, Ann Arbor, MI).
4. Place filter onto a Petri dish containing hardened agar and cover plates.
5. Incubate plates 24 ± 2 hours at 37 ± 1°C.
6. Count the number of colonies formed with the aid of a colony counter.
7. Record the number of colonies per dish, and the B. subtilis spore sample volume and
dilution. If it is not possible to distinguish individual colonies because of confluent
growth, record the plate counts as "TNTC".
8. Calculate the B. subtilis spore concentration in the original samples as cfu/mL using
Equation D.I.
D.6 References
Adams, M.H. 1959. Bocteriophage.' New York: Interscience publication.
APHA/AWWA/WEF. 1998. Standard methods for the examination of water and wastewater,
20th edition. Washington DC: American Public Health Association, American Water
Works Association, and Water Environment Federation.
DVGW. 1997. UVdisinfection devices for drinking water supply—Requirements and testing.
Bonn, Germany: German Gas and Water Management Union (DVGW).
Havelaar, A.M., C.C.E. Meulemans, W.M. Pot-Hogeboom, and J. Koster. 1990. Inactivation of
bacteriophage MS2 in wastewater effluent with monochromatic and polychromatic
ultraviolet light. Water Research!^, no. 2:1387-1393.
Hoyer, 0.2002. Personal communication.
UV Disinfection Guidance Manual D-6 June 2003
Proposal Draft
-------
Appendix D. Microbiological Methods
Meng, Q.S. and C.P. Gerba. 1996. Comparative inactivation of enteric adenovirus, poliovirus
and coliphages by ultraviolet irradiation. Wat. Res. 30:2665-2668.
Muriakata, N. and C.S. Rupert. 1972. Genetically controlled removal of "spore photoproduct"
: from deoxyribonucleic acid of ultraviolet-irradiated bacillus subtilis spores. J. Bacteriol.
111:192-198. ' ' ,.
Nieuwstad, T.J. and A.H. Havelaar. 1994. The kinetics of batch ultraviolet inactivation of
bacteriophage MS2 and microbiological calibration of an ultraviolet pilot plant. J.
Environ. Sci. Health A29,no.5:1993-2007.
NWRI/AwwaRF. 2000. Ultraviolet disinfection guidelines for drinking water and water reuse.
National Water Research Institute and AwwaRF
Oppenheimer, J.A., I.E. Hbagland, J.-M. Laine, J.G. Jacangelo, and A. Bhamrah. 1993.
Microbial inactivation and characterization of toxicity and by-products occurring in
reclaimed wastewater disinfected with UV radiation. Water Environment Federation
(WEF) Specialty Conference: Planning, design & operation of effluent disinfection
systems. Whippany, New Jersey.
Sommer, R. and A. Cabaj. 1993. Evaluation of the efficiency of a UV plant for drinking water
disinfection. Water Science & Technology 27, no. 7-8:357-362.
Sommer, R., A, Cabaj, D. Schoenen, J. Gebel, A. Kolch, A.H. Havelaar, and P.M. Schets. 1995.
Comparison of three laboratory devices for UV-inactivation of microorganisms. Wat. Sci.
Tech. 31:147-156.
Sommer, R., A. Cabaj, and T. Haider. 1996. Microbiocidal effect of reflected UV radiation in
devices for water disinfection. Water Science & Technology 34, no.5-6:173-177.
Sommer, R., T. Haider, A. Cabaj, W. Pribil, and M. Lhotsky. 1998. Time dose reciprocity in UV
disinfection of water. IAWQ, Vancouver.
Tree, J.A., M.R. Adams, and D.N. Lees. 1997. Virus inactivation during disinfection of
wastewater by chlorination and UV irradiation and the efficacy of f+ bacteriophage as a
Viral indicator'. Water Science & Technology 35:227-232.
Yahya, M.T., T.M. Straub, and C.P.Gerba. 1992. Inactivation of coliphage MS-2 and poliovirus
by copper, silver, and chlorine. Canadian Journal of Microbiology. 38, no.6:430-435.
UV Disinfection Guidance Manual D-7 June 2003
Proposal Draft
-------
Appendix E. Collimated Beam Apparatus: Measuring
Challenge Microorganism UV Dose-Response
The challenge microorganism's UV dose-response should be measured using a bench-
scale device referred to here as a "collimated beam apparatus" (Figure E.I). The apparatus
delivers UV light to a microbial suspension usually contained within a completely mixed batch
reactor. The UV light enters the suspension with a near zero degree angle of incidence and is
relatively homogenous across the surface area. UV dose delivered to the suspension is
calculated using measurements of incident UV intensity, exposure time, suspension depth, and
the absorption coefficient of the suspension. By measuring microbial inactivation in the
suspension as a function of UV dose, the microorganism's dose-response is determined.
Figure E.1 Collimated Beam Apparatus
Low-Pressure
Mercury Arc Lamp
Petri Dish Containing
Microbial Suspension
Lamp Enclosure
""- Collimating Tube
UV Light® 254nm
Magnetic Stirrer
UV Disinfection Guidance Manual
Proposal Draft
E-l
June 2003
-------
Appendix E. Collimated Beam Apparatus: Measuring Challenge Microorganism UV Dose-Response
This appendix describes the following components of collitiiated beam testing:
• Collimated beam apparatus design and operation (section E.I)
. Procedure for irradiating samples using apparatus (section E.2)
. Calculation of UV dose delivered by the apparatus (section E.3)
. Quality Assurance / Quality Control (QA/QC) procedures (section E.4)
« Reporting results (section E.5)
E.1 Apparatus Design and Operation
Because UV dose requirements are based on the pathogen inactivation achieved using
253.7 nm light, the collimated beam apparatus must use a lamp that emits germicidal UV light
only at 254 nm (e.g., a low-pressure lamp). To prevent ozone formation, lamps that emit 185 nm
light should not be used. The output from the lamp measured using a radiometer or equivalent
should not vary more than 5 percent over the exposure time. A stable lamp output can be
obtained by driving the lamp with a constant power source and maintaining the lamp at a
constant operating temperature. A voltage regulator may be used to obtain a stable power supply
to the lamps if the line voltage is not sufficiently stable. A stable temperature can be obtained by
controlling the airflow around the lamp.
The UVlamp should be located far enough above the surface of the microbial suspension
that uniform irradiance is obtained across the sample's surface and UV light enters the
suspension with a near zero degree angle of incidence (Blatchley 1997). A recommended
minimum distance from the lamp to the suspension is six times the longest distance across the
suspension's surface. In order to vary the UV intensity incident on the suspension, the distance
between the suspension and the lamp can be adjusted.
The uniformity of the intensity field across the sample's surface should be assessed by
measuring the "Petri Factor," the ratio of the average irradiance across the suspension surface to
the irradiance measured at the center (Bolton and Linden 2003). The average irradiance is
determined by averaging radiometer measurements taken at each point in a 5 mm spaced grid
across an area defined by the suspension's surface. If the radiometer's sensing window is wider
UV Disinfection Guidance Manual E-2 June 2003
Proposal Draft
-------
Appendix E. Collimated Beam Apparatus: Measuring Challenge Microorganism UV Dose-Response
than 5 mm, it should be reduced using a cover slip, with a small hole. In general, the collimated
beam apparatus should have a Petri Factor greater than 0.9.
The lamp and the light path from the lamp to the suspension should be enclosed to protect
the user from exposure to UV light. A box-like enclosure made of aluminum is often used. A
length of pipe is often used to enclose the light path from the lamp to the microbial suspension.
The inside surface of the pipe should have a low UV reflectance and incorporate apertures to
improve UV light collimation (Blatchley 1997). A shutter mechanism is sometimes used to
control the exposure of the suspension to UV light. The exposure times should be measured with
an uncertainty of 5 percent or less. Exposure times less than 20 seconds are not recommended.
The microbial suspension should be irradiated in an open cylindrical container with a
constant cross-sectional area (e.g., Petri dish). The diameter of the container should be smaller
than the diameter of the light beam incident on the container. Sample depth should be 0.5 to
2 cm. The material of the container should not adsorb the challenge microorganism enough to
impact its measured dose-response.
Sample volumes irradiated in the container should be sufficient for measuring the
challenge microorganism's concentration after irradiation. The microbial suspension should be
mixed using a stir bar and a magnetic stirrer at a rate that does not induce vortices. The volume
and diameter of the stir bar should be small relative to the volume and depth of the sample
volume.
The irradiance at the center of the suspension's surface before and after exposure to UV
light should be measured using a radiometer calibrated at 254 nm. The radiometer calibration
should be National Institute of Standards and Technology (NIST) traceable or equivalent with a
known measurement uncertainty. During measurement, the radiometer's calibration plane should
match the height of the suspension's surface and be perpendicular to the incident UV light. The
calibration plane of the radiometer should be specified in the radiometer's calibration certificate.
The depth of the suspension, including the stir bar volume, should be measured with an
uncertainty of 10 percent or less. The UV absorption coefficient of the microbial suspension at
254 nm should be measured using a spectrophotometer with a measurement uncertainty of
10 percent or less. If scattering of light by the microorganisms and other paniculate matter
within the suspension is significant, the UV absorption coefficient should be measured using a
spectrophotometer equipped with an integrating sphere (Linden and Darby 1998). While 1 cm
cuvettes are typically used for measuring UV absorption coefficients, cuvettes with longer
pathlengths are recommended to reduce the measurement uncertainty with low UV absorbance
samples.
E.2 Procedure
Personnel who perform bench-scale UV irradiation should be experienced with the use
and safety requirements of the equipment. Safety goggles and latex gloves should be worn. Skin
should be shielded from exposure to UV light. The following procedure is recommended for
irradiating a water sample using the collimated beam apparatus:
1. Define the target UV dose.
UV Disinfection Guidance Manual E-3 June 2003
Proposal Draft
-------
Appendix E. Collimated Beam Apparatus: Measuring Challenge Microorganism UV Dose-Response
2. Measure the UV absorption coefficient of the water sample.
3. Place a known" volume from the water sample into a container and add a stir bar.
4. Measure the water depth in the container.
5. Measure the UV irradiance delivered by the collimated beam.
6. Calculate the exposure time to deliver the target dose.
7. Block the light from the collimating tube using a shutter or equivalent.
8. Center the container containing the water sample under the collimating tube.
9, Unblock the light from the collimating tube and start the timer.
10. When the target exposure time has elapsed, block the light from the collimating tube.
11. Remove the container and collect the sample for measurement of the challenge
microorganism concentration. If the sample is not assayed immediately, store in the
dark at 4°C.
12. Re-measure the UV irradiance and calculate the average of the two measurements.
13. Using Equation E.I, calculate the applied dose using the measured irradiance, UV
absorption coefficient, sample depth, and exposure time.
14. Repeat the procedure for various target dose values. The UV dose-response curve is
a plot of the microorganism concentration as a function of the applied dose.
E.3 Dose Calculation
Dose delivered to the sample is calculated using Equation E.I:
D^EsPf(\-R)-^-^^-t EquationE.1
where
D = UV dose in mJ/cm2
Es = UV irradiance at the center of the suspension's surface in mW/cm2
Pf = Petri Factor
R = Reflectance at the air-water interface at 254 nm
L = Distance from lamp centerline to suspension surface in cm
d = Depth of the suspension in cm
a = UV absorption coefficient (Base 10) of the suspension at 254 nm in cm"1
t = Exposure time in seconds
UV Disinfection Guidance Manual E-4 June 2003
Proposal Draft
-------
Appendix E. Collimated Beam Apparatus: Measuring Challenge Microorganism UV Dose-Response
The term L/(d+L) accounts for the divergence of the UV light from the collimated beam
as it passes through the suspension. The reflectance at the air-water interface estimated using
Fresnel's Law is 0.025 given an index of refraction of 1.000 and 1.372 for air and water,
respectively.
Alternatively, given a target dose, the exposure time may be calculated by re-arranging
Equation E.I to form Equation E.2:
ad
Equation E.2
where
variables are defined as in Equation E.I
The measurement uncertainty of the dose delivered by the collimated beam should be
assessed at an 80 percent confidence interval with consideration of each term in Equation E.I.
The measurement uncertainty of each term in Equation E.I can be determined from the
measurement uncertainty stated for the instrumentation used to measure those quantities and the
standard deviation of repeated measurements made with that instrumentation (Taylor 1997). If
the uncertainty of the measurement of the suspension depth and the UV absorption coefficient is
less than 10 percent at a 80 percent confidence level and the product ad is less than 0.1, the
uncertainty of the term (l-10"ad)/ad can be assumed as 4 percent. This assumption allows the use
of the sum of variances approach to calculate the uncertainty of the dose delivered by the
collimated beam.
Example. A pipette with an accuracy of 0.2 mL is used to place a 25 mL microbial
sample in a Petri dish. The incident irradiance of 1.00 mW/cm2 is measured using a radiometer.
The uncertainty of the radiometer measurement indicated by the calibration certificate is
7 percent. The suspension is irradiated for 60 seconds. The irradiation time is monitored using a
stopwatch with an uncertainty of ±1 second. The Petri dish radius, measured using a ruler with
1 mm graduations, is 2.5 ± 0.1 cm. The stir bar volume is estimated as 1' ± 0.1 mL. The UV
absorption coefficient of the microbial sample at 254 nm is 0.050 ± 0.005 cm"1. The Petri factor
of 0.90 ± 0.02 is calculated for the collimated beam apparatus. "The distance from the lamp to the
surface of the suspension is determined using a ruler as 25 ± 1 cm.
The depth in the Petri dish is calculated as the sum of the suspension and stir bar volumes
divided by the area of the Petri dish.
Volume (25±0.2cm3)+(l+0.1cm3)
Area
p(2.5i0.1cm)2
= 1.32 ±0.07 cm
The UV dose is calculated as:
•UV Disinfection Guidance Manual
Proposal Draft
E-5
, June 2003
-------
\
Appendix E. Collimated Beam Apparatus: Measuring Challenge Microorganism UV Dose-Response
(l. 00 mW/cm2)(0.90)(l- 0.025) (l- lO^""-' ')('32'm } f
D = * - -r^ - ~ - -7-i - -T, - r—f-- 60s = 46 ml/cm2
(1.32 cm)
""
Because the uncertainty of the sample depth (± 0.07 cm) and the measured UV
absorption coefficient (± 0.005 cm"1) is less than or equal to 10 percent of the sample depth and
the product of the sample depth and UV absorption coefficient is less than 0.1, the uncertainty of
the term (l-lO'^yad is assumed as 4 percent. The uncertainties of the terms in the dose
calculation are as follows:
• Incidence irradiance 7 percent
. Petri factor 2 percent
. L/(d+L) 0.3 percent
• Time 2 percent
. (l-lO'^/ad 4 percent
The uncertainty of the dose calculation is calculated using the sum of variances approach
as:
Uncertainty= (?2 + 22 + 0.32 + 22 + 42)"2 = 8.5%
E.4 Quality Assurance and Quality Control
QA/QC measures include:
• Designing the collimated beam apparatus with a Petri factor greater than 0.9
• Selecting instrumentation and methods that minimize the measurement uncertainty of
dose delivery by the collimated beam apparatus -
• Calibrating all radiometers at regular intervals as recommended by the manufacturer
« Using a reference radiometer or equivalent method to regularly check the
measurement accuracy of the radiometer used to measure incident irradiance
• Ensuring irradiance measurements before and after exposure to UV light do not differ
by more than 5 percent
i
. Ensuring replicate UV inactivation curves do not differ significantly
UV Disinfection Guidance Manual E-6 June 2003
Proposal Draft
-------
Appendix E. Collimated Beam Apparatus: Measuring Challenge Microorganism UV Dose-Response
« Ensuring the UV dose-response of the challenge microorganism lies within expected
bounds as defined by published dose-response data
E.5 Reporting
The following information should be documented and included with the validation test
report:
Lamp type •
Distance from the light source to the sample surface
Radiometer make and model
Measurement uncertainty of the radiometer and date of last calibration
Comparison of working and reference radiometers
Volume and depth of the microbial suspension
UV absorption coefficient of the microbial suspension measured at 254 nm
Irradiance measurement before and after each irradiation
Petri factor calculations and results
Method of dose determination
UV dose calculations
Uncertainty assessment
E.6 References
Blatchley, E.R. 1997. Numerical modeling of UV intensity: Application to colHmated-beam
reactors and continuous-flow systems. Water Research 31:2205-2218.
Bolton. J. and K. Linden. 2003. Standardization of methods for fluence (UV Dose)
determination in bench-scale UV experiments. J. Environ. Eng. 129, no.3:209-216.
Linden, K.G. and J.L. Darby. 1998. UV Disinfection of Marginal Effluents: determining UV
Absorbance and Subsequent Estimation of UV Intensity. Water Environment Research
70(2).
Taylor, J.R. 1997. An introduction to error analysis: the study of uncertainties in physical
measurements. Sausalito, CA: University Science Books.
UV Disinfection Guidance Manual E-7 June 2003
Proposal Draft
-------
Appendix F. Background to the UV Reactor
Validation Protocol
This appendix provides background material for the validation protocol given in
Appendix C. The background material is organized into the following six sections.
• Dose delivery by UV reactors. Section F, 1 describes how the RED of a challenge
microorganism measured during UV reactor validation is related to the capacity of the
UV reactor to inactivate a target pathogen. This section describes why correction
factors should be applied to the reduction equivalent dose (RED) of the challenge
microorganism to account for systematic errors that arise if the challenge
microorganism is more resistant to UV light as compared to the target pathogen. The
section concludes by describing approaches for selecting one or more challenge
microorganisms to minimize those errors.
• Dose monitoring. Section F.2 describes three approaches whereby measurements of
flowrate, UV intensity, and water UV transmittance (UVT) are used by UV reactors
to indicate dose delivery. This section discusses the importance of UV intensity
sensor placement within a UV reactor and provides a rationale for defining test
conditions to validate UV reactors using a given dose monitoring approach.
• UV intensity sensors. Section F.3 describes the properties of UV intensity sensors,
how those properties impact the sensor's measurement uncertainty, and how that
measurement uncertainty is used to define rejection criteria for using reference
sensors to check the accuracy of duty sensors. The section also discusses how non-
uniform lamp aging and fouling and the variability in lamp output affects the use of
UV intensity sensors.
. Polychromatic considerations. Section F.4 describes systematic errors that can occur
with the validation of UV reactors equipped with medium-pressure UV lamps. This
section provides approaches for assessing those errors and for defining correction
factors mat should be applied to validation data.
. Uncertainty of monitoring and dose factors. Section F.5 provides a rationale for
defining a safety factor that accounts for the random uncertainty associated with UV
reactor validation and monitoring.
• Re-validation. Section F.6 discusses how some changes to a UV reactor design made
by a manufacturer would trigger a need to re-validate the UV reactor.
F.1 Dose Delivery by UV Reactors
Dose delivered to an individual microorganism passing through a UV reactor is defined
as the integral of UV intensity over time:
UV Disinfection Guidance Manual F-l ' June 2003
Proposal Draft
-------
Appendix F. Background to the UV Reactor Validation Protocol
o
where
D
I
t
tr
Equation F.I
Dose delivered to the microorganism by the UV reactor
UV intensity incident on the microorganism as it travels through the UV reactor
time
Residence time of the microorganism within the UV reactor
Because each microorganism passing through the UV reactor follows a unique trajectory,
each microorganism is exposed to a unique dose. For example, microorganisms passing through
the UV reactor close to the lamps are exposed to higher UV intensities as compared to
microorganisms traveling near the reactor walls or between lamps. Microorganisms caught in
eddies or dead zones spend more time within the UV reactor as compared to microorganisms that
pass through the reactor quickly due to hydraulic short-circuiting. Because each microorganism
is exposed to a different UV dose, dose delivery by the UV reactor is best described using a dose
distribution, as opposed to a single dose value. A dose distribution describes the probability that
a microorganism passing through a UV reactor will receive a given dose. Figure F.I presents an
example of a dose distribution for a UV reactor.
Model-based and experimental approaches have been identified to determine the dose
distribution of a UV reactor. Model-based approaches use computational fluid dynamics (CFD)
to predict microorganism trajectories through a UV reactor and hence the dose delivered to each
microorganism. Experimental approaches use microspheres that undergo a chemical reaction
when exposed to UV light. The microspheres are injected upstream of the UV reactor and are
collected downstream. The extent of the UV-induced chemical reaction within each sphere is
measured and used to calculate the dose delivered to that sphere as it traveled through the
reactor. While promising, both model and experimental-based approaches are subjects of current
research. Further effort is necessary to prove these approaches as practical and accurate
predictors of UV reactor performance.
Dose delivery by UV reactors is currently measured using biodosimetry (Quails and
Johnson 1983). With biodosimetry, inactivation of a challenge microorganism passed through
the UV reactor is measured and related to a single dose value based on the known UV
dose-response of that microorganism. This dose value is termed the RED.
UV Disinfection Guidance Manual
Proposal Draft
F-2
June 2003
-------
Appendix F. Background to the UV Reactor Validation Protocol
Figure F.1 Dose Distribution Delivered by a UV Reactor1
UVDose(mJ/cm2)
'(Adapted from Chiu et al. 1999)
F.I .1 Relationship Between RED and the Dose Distribution
The RED of a given microorganism depends on the dose distribution delivered by the
reactor and the UV inactivation kinetics (dose-response) of the challenge microorganism (Cabaj
et al. 1996). A general equation describing this dependence is Equation F.2:
Equation F.2
where
RED =
f
J ~
Pi
RED measured using biodosimetry
Mathematical function describing the inactivation kinetics of the microorganism
Number of dose values in the dose distribution
ith dose in the dose distribution
Probability of occurrence of dose Dj
For example, if the microorganism has first order inactivation kinetics, the function/is
shown in Equation F.3:
N = f(D)=N0exp(-kD.)
Equation F.3
where
N
No
D.
k
Microorganism concentration after exposure to dose D
Microorganism concentration with zero UV dose
Applied UV dose
= Microorganism's first order inactivation coefficient
UV Disinfection Guidance Manual
Proposal Draft
F-3
June 2003
-------
Appendix F. Background to the UV Reactor Validation Protocol
Substituting Equation F.3 into F.2 gives the following equation for the RED of a
microorganism with first-order inactivation kinetics:
RED = - - In T p , exp(- JfeD )
Equation F.4
In equation F.4, the RED depends on the dose distribution of the UV reactor and the first
order inactivation coefficient of the microorganism.
Figure F.2 presents the dependence of the RED on the first order inactivation coefficient
of the challenge microorganism for the dose distribution shown in Figure F.I. The relation was
calculated using Equation F.4. As shown, the RED of a microorganism with a small first-order
inactivation coefficient is greater than the RED of a microorganism with a large first-order
inactivation coefficient. Because the RED depends on the microorganism's UV inactivation
kinetics, the RED of the challenge microorganism is an exact measure of the RED delivered to a
pathogen of interest only when the challenge microorganism has the same inactivation kinetics
as the pathogen (Wright and Lawryshyn 2000).
Example 1. A UV reactor delivers a dose distribution represented by Figure F.I. The
UV reactor is evaluated using biodosimetry. The challenge microorganisms are MS2
bacteriophage (MS2) with a first order coefficient of 0.13 cm2/mJ and 9X174 phage with a first
order coefficient of 1.2 cm2/mJ. As shown in Figure F.2, MS2 would have experienced 1.1 log
inactivation, corresponding to an RED of 19 mJ/cm2. <|>X174 would have experienced.3.6 log
inactivation, corresponding to an RED of 7.3 mJ/cm2. If the pathogen of interest has the same
inactivation kinetics as q>X174, the RED of MS2 would be 2.5 times greater than the RED
delivered to the pathogen, while the RED of
-------
Appendix F. Background to the UV Reactor Validation Protocol
Figure F.2 Microorganism Log Inactivation and RED as a Function of the
Microorganism's First Order Inactivation Coefficient for the UV Reactor
. Represented in Figure F.1
0 T 0.5 1 T. 1.5 . 2 .2.
MS2 4>x174 ...
First Order Inactivation Coefficient (crrfrmJ)
F.1.2 Using RED to Demonstrate Target Pathogen Inactivation
If the UV dose-response of the challenge microorganism differs from that of the target
pathogen and the dose distribution of the UV reactor is not known, the results of biodosimetry
can only be used to estimate the target pathogen inactivation within a range bounded by the
inactivation expected assuming ideal and worst-case hydraulics. Figure F.3 provides a
comparison of the dose distribution of reactors with ideal and worst-case hydraulics to a dose
distribution that might be seen with a real reactor.
Figure F.3 Comparison of the Dose Distributions
of Ideal, Realistic, and Worst-Case UV Reactors
. .0
M
.O
O
Ideal
0 50 100
UV Dose (mJ/cm2) '
UV Dose (mJ/cm2)
Probability
Worst Case
1
0 Infinity
UV Dose (mJ/cm2)
A reactor with ideal hydraulics delivers the same dose to all the microorganisms passing
through the reactor. Its dose distribution is represented by a single dose. Examples of a UV
reactor with ideal hydraulics include the stirred suspension irradiated during the measurement of
UV dose-response by a collimated beam device and a plug flowrate reactor with complete lateral
mixing. In both cases, the UV dose delivered is the product of the average UV intensity within
UV Disinfection Guidance Manual F-5 June 2003
Proposal Draft
-------
Appendix F. Background to the UV Reactor Validation Protocol
the reactor and the residence time. With an ideal reactor, Equation F.5 shows the net rnicrobial
inactivation achieved by the reactor:
Equation F.5
Accordingly, with an ideal reactor, the RED measured with a challenge microorganism is
a measure of the RED delivered to all microorganisms that pass through the reactor. If both the
challenge microorganism and the pathogen have first order inactivation kinetics, the log
inactivation of the pathogen is calculated using Equation F.6:
, N\
log— =-
RED
Equation F.6
where
log (N/N0)P =
kp
RED
Log inactivation of the pathogen
First order inactivation coefficient of the pathogen
RED observed with the pathogen
UV sensitivity of the pathogen expressed as mJ/cm2 per log
The UV sensitivity of the pathogen is related to the first order inactivation coefficient
using Equation F.7:
_ ln(lO) _ 2.30
10 ~ k k
Equation F.7
A UV reactor with worst-case hydraulics delivers a UV dose of zero to all
microorganisms passing through the reactor. However, in the case of a reactor with a
measurable RED, worst-case hydraulics is defined as infinite dose delivered to one fraction of
the flowrate and zero dose delivered to the other fraction. Under these conditions, the net
microbial inactivation achieved by the reactor is calculated according to Equation F.8:
N/ -
/N ~
Equation F.8
As shown, the net inactivation achieved by the worst-case UV reactor with a measurable
RED is constant and independent of the inactivation kinetics of the microorganism. With a
worst-case UV reactor, the measured inactivation is a measure of the inactivation that would
occur with all microorganisms regardless of their UV sensitivity. In other words, the log
inactivation of the pathogen is calculated according to Equation F.9:
UV Disinfection Guidance Manual
Proposal Draft
F-6
June 2003
-------
Appendix F. Background to the UV Reactor Validation Protocol
Equation F.9
where
log (N/No)c = log inactivation of the challenge microorganism
Using the above definitions of an ideal and a worst-case reactor, the log inactivation of a
pathogen estimated from biodosmetry results will have a value between log(N/N0)c and RED/DP.
If the inactivation of the pathogen must be known with absolute confidence, the lower bound of
that range should be used. If the challenge microorganism is more resistant to UV light than the
pathogen, the lower bound is log(N/N0)c. If the challenge microorganism is less resistant to UV
light than the pathogen, the lower bound is RED/DP.
Example 2. A UV reactor is challenged using MS2 with a UV sensitivity of 18 mJ/cm2
per log inactivation. Four log inactivation of the MS2 is observed corresponding to an MS2
RED of 4 x 18 = 72 mJ/cm2. The MS2 results are used to estimate the log inactivation of two
pathogens, one with a UV sensitivity of 10 mJ/cm2 per log inactivation and the other with a UV
sensitivity of 25 mJ/cm2 per log inactivation. The log inactivation of the first pathogen is
estimated between 4.0 and 72/10 = 7.2 log and the log inactivation of the second pathogen is
estimated between 72/25 = 2.9 and 4.0 log. The biodosimetry results can be used to state with
absolute confidence that the inactivation of the first pathogen was at least 4.0 log and the
inactivation of the second pathogen was at least 2.9 log.
Example 3. A UV reactor is designed for 3.0 log Cryptosporidium inactivation. MS2 is
used to measure the performance of the UV reactor. Because MS2 is more resistant to UV light
than Cryptosporidium, 3.0-log MS2 inactivation must be measured to state with absolute
confidence that the reactor achieves 3.0-log Cryptosporidium inactivation.
Example 4. A UV reactor is designed for two log adenovirus inactivation. Two-log
adenovirus inactivation occurs using a UV dose of 100 mJ/cm2. The UV reactor is validated
using MS2. Because adenovirus is more resistant to UV light than MS2, a RED of 100 mJ/cm2
must be measured with MS2 to state with absolute confidence that the UV reactor achieves 2 log
adenovirus inactivation.
Because UV manufacturers strive to optimize the hydraulic design of their UV reactors,
using the worst-case dose distribution represented in Figure F.3 to define the lower bound of
pathogen inactivation is overly conservative. An alternative approach is to use the dose
distribution of a commercial UV reactor that is representative of worst-case reactor hydraulics.
However, defining a worst-case commercial UV reactor is difficult because little data are
available in the peer-reviewed UV disinfection literature on dose distributions. Chiu et al. (1999)
used measured velocity fields and a random walk model to predict the dose distribution delivered
by a wastewater reactor equipped with low-pressure (LP) lamps oriented perpendicular to
flowrate. The dose distribution was bimodal due to a short-circuiting path along the reactor
walls. Wright and Lawryshyn (2000) compared the dose distribution of four reactor designs
using CFD-based dose modeling including the reactor modeled by Chiu et al. Based on this
comparison, the dose distribution developed by Chiu et al. is believed to represent a worst-case
commercial UV reactor.
UV Disinfection Guidance Manual F-7 June 2003
Proposal Draft
-------
Appendix F. Background to the UV Reactor Validation Protocol
Figure F.I presents a dose distribution adapted from Chiu et al.'s data. For that dose
distribution, Figure F.2 presents log inactivation and RED as a function of the microorganism's
UV sensitivity expressed as a first-order inactivation coefficient. Figure F.4 presents the same
relationship, but with UV sensitivity expressed as dose per log inactivation. Using these figures,
the RED delivered to a pathogen by a given UV reactor can be estimated from the measured
RED of the challenge microorganism using Equation F. 10:
RED
RED=REDcx f
p c RED'
Equation F.10
where
REDp =
REDC =
REDP =
REDC =
RED of the pathogen estimated for the UV reactor of interest
RED of the challenge microorganism measured during biodosimetry
RED of the pathogen estimated from Figure F.2 or F.4
RED of the challenge microorganism estimated from Figures F.2 or F.4
The RED determined using Equation F.10 represents the RED that would be delivered if
the reactor under consideration had a dose distribution representative of a worst-case commercial
reactor.
Figure F.4 Microorganism Inactivation and RED as a Function of Microorganism
UV Sensitivity for the UV Reactor Represented in Figure F.1
UV Sensitivity (mJ/cm2 per log inactivation)
Example 5. A UV reactor is challenged using MS2 with a UV sensitivity of 18 mJ/cm2
per log inactivation. Four log inactivation of the MS2 is observed corresponding to an MS2
RED of 4 x 18 = 72 mJ/cm2. The MS2 results are used to estimate the log inactivation of two
pathogens, one with a UV sensitivity of 10 /cm2 per log inactivation and the other with a UV
sensitivity of 25 mJ/cm2 per log inactivation. In Figure F.4, the RED delivered to the
microorganisms with a UV sensitivity of 10, 18, and 25 mJ/cm2 per log inactivation is 15, 19,
UV Disinfection Guidance Manual
Proposal Draft
F-8
June 2003
-------
.Appendix F. Background to the UV Reactor Validation Protocol
and 21 ml/cm2, respectively. Assuming the UV reactor's performance is bounded by a worst
case represented by Figure F.4, the RED delivered to the first pathogen is estimated between 72
ml/cm2 and (72 x 15)/19 = 57 ml/cm2 and the RED delivered to the second pathogen is
estimated between 72 and (72 x 21)719 = 80 mJ/cm2. Inactivation of the first pathogen is
estimated between 5.7 (57/10) and 7.2 (72/10) log and inactivation of the second pathogen is
estimated between 2.9 (72/25) and 3.2 (80/25) log inactivation. This range of inactivation
estimated using the worst-case represented in Figure F.4 is notably less than the range estimated
in Example 3 using the worst-case represented in Figure F.3.
For regulatory purposes, the lower bound of the range of inactivation and RED estimated
for the pathogen should be used when relating challenge microorganism inactivation to target
pathogen inactivation. If the challenge microorganism is more sensitive to UV light than the
pathogen or if both have the same sensitivity, the RED delivered to the pathogen should be
estimated using the RED of the challenge microorganism. If the challenge microorganism is
more resistant to UV light than the pathogen, the RED delivered to the pathogen should be
estimated using Equation F.10.
Example 6. A UV reactor is designed for three log Cryptosporidium inactivation. The
dose needed for 3 log Cryptosporidium taken from Chapter 1 (Table 1.4) is 12 mJ/cm2.
Accordingly, the UV sensitivity of Cryptosporidium is defined as 12/3 = 4 mJ/cm2 per log
inactivation. MS2 with a UV sensitivity of 18 mJ/cm2 per log inactivation is used to measure the
performance of the UV reactor. Because MS2 is more resistant to UV light than
Cryptosporidium, Equation F.10 is used to relate the RED measured using MS2 to the dose
delivered to Cryptosporidium. From Figure F.4, the RED delivered to the microorganisms with
a UV sensitivity of 3.9 and 18 mJ/cm2 per log inactivation is 9.8 and 19, respectively. Thus an .
MS2 RED of 12x19/9.8 = 23 mJ/cm2 should be demonstrated to show 3 log Cryptosporidium
inactivation. ,
Example 7. A UV reactor is designed for one-log adenovirus inactivation. The dose
needed for 1-log adenovirus taken from Chapter 1 (Table 1.4) is 58 mJ/cm2. MS2 with a UV
sensitivity of 18 mJ/cm2 is used to measure the performance of the UV reactor. Because MS2 is
less resistant to UV light than adenovirus, an MS2 RED of 58 mJ/cm2 should be dempnstrated to
show 1-log adenovirus inactivation.
The RED of microorganisms with shoulders and tailing within the dose-response curve
depends on the overlap of the dose distribution with those regions (Cabaj et al.1996, Wright and
Lawryshyn 2000). To use Figure F.2 to define safety factors, the inactivation of the challenge
microorganism should demonstrate an exponential inactivation as a function of dose over the
range of doses in the dose distribution. This creates a dilemma if the dose distribution is not
known. To avoid this issue, the dose-response of an appropriate challenge microorganism
should not demonstrate a shoulder at a dose beyond 50 percent of the demonstrated RED and
should not demonstrate tailing until one log inactivation beyond the demonstrated inactivation.
In the case of a challenge microorganism with a shoulder and tailing in the dose-response, the
UV sensitivity will be defined as the sensitivity over the region of exponential inactivation that
occurs between the shoulder and the onset of tailing. The shoulder of the dose-response is
defined by the intersect of the exponential region with the dose axis (see Figure F.5).
UV Disinfection Guidance Manual F-9 ' June 2003
Proposal Draft
-------
Appendix F. Background to the UV Reactor Validation Protocol
Example 8. Figure F.5 presents the measured UV dose-response of B. subtilis spores.
Because the measured dose-response has a shoulder of 16.5 raJ/cm , the B. subtilis spores should
only be used to demonstrate RED values greater than or equal to 2 x 16.5 = 33 ml/cm2.
Figure F.5. UV Dose-Response of B. subtilis Spores
';;;'",... Vy = 0,0886x -1.4681j
•L Doseaxjsiiihtergept
\. ,n;-s- 16.5:rnJ/cm2;
(m Jcrri?)
(Adapted from Sommer et ai. 1998)
The RED safety factor provides an incentive to select a challenge microorganism whose
UV sensitivity matches that of the target pathogen and a disincentive for overrating UV reactor
performance by using challenge microorganisms whose UV sensitivity is much greater than the
target pathogen.
F.1.3 Biodosimetry Using Two Challenge Microorganisms
In order to provide a better estimate of the target pathogen's log inactivation and RED,
two microorganisms with different UV sensitivities can be used to validate UV reactors. The
target pathogen's log inactivation should be estimated by interpolating the log inactivation of the
two microorganisms as a function of the UV sensitivity defined on a linear scale as a first-order
inactivation coefficient. Alternatively, the target pathogen's RED should be estimated by
interpolating the RED of the two microorganisms as a function of the UV sensitivity defined on a
linear scale as dose per log inactivation. If interpolation does not meet these provisions, the
inactivation of the pathogen will be overestimated.
Example 9. A UV reactor with a dose distribution represented in Figure F.4 is tested
using MS2 and 0X174. The MS2 and X174 have a UV sensitivity of 18 and 2 mJ/cm2 per log
inactivation. Using biodosimetry, 1.1 and 3.6 log inactivation of MS2 and <|>X174 are measured.
These log inactivations correspond to RED values of 20 and 7.2 mJ/cm2, respectively. The RED
measured with MS2 and
-------
Appendix F. Background to the UV Reactor Validation Protocol
RED = 0.731 x UV Sensitivity + 5.83
Hum, defined with a UV
This' equation predicts that the RED delivered to Cryptosporidii
isitivity of 3.9 ml/cm2 per log inactivation, is 8.7 mJ/cm .
sehsitivi
If the inactivation of the more UV-sensitive of the two challenge microorganisms is
greater than the detection limit of the assay, interpolation should be based on the level indicated
by the limitation. Because the inactivation of the UV-sensitive microorganism is
underestimated, the interpolation will be conservative and two-microorganism validation may
not offer an advantage over single microorganism validation.
Example 10. A UV reactor is evaluated using MS2 and <|>X174 phage. MS2 and <|>X174
are injected into the flowrate upstream of the reactor. Influent and effluent samples are collected
and,assayed. The assay has a detection limit of 1 pfu/mL. The concentrations of MS2 and
<|>X174 in the influent is determined as 1,000,000 and 10,000 pfu/mL, respectively. The
concentrations of MS2 and <|>X174 in the effluent samples are 10,000 and 0 pfu/mL, respectively.
The results indicate that the concentration of X174 is below the detection limit of the assay.
Accordingly, the log inactivation of MS2 and <|>X174 is 2 log and > 4 log, respectively. If the
UV sensitivity of MS2 and (j)X174 are determined to be 20 and 2 mJ/cm2 per log, the MS2 RED
is 40 mJ/cm2 and the (j>X174 RED is > 8 mJ/cm2. The following, equation fits the measured RED
as a function of UV sensitivity:
RED = 1.77 x UV Sensitivity + 4.44
This equation predicts that the RED delivered to Cryptosporidium defined with a UV
sensitivity of 3.9 mj/cm2.per log is 11.3 mJ/cm2. This compares to an RED of 20 mJ/cm2 that
would have been predicted by Equation F. 10 using the MS2 data alone. In this case, two-
microorganism biodosimetry estimated lower dose delivery to Cryptosporidium than single
microorganism biodosimetry.
In the past, it has been assumed that the RED measured with a UV-resistant challenge
microorganism can be used to demonstrate compliance with a dose target while the log
inactivation demonstrated with a UV-sensitive challenge microorganism can be used to
demonstrate compliance to a log inactivation target. This approach is not recommended. It is
not possible to demonstrate compliance to a 3-log Cryptosporidium inactivation by using UV-
resistant MS2 to show an RED of 11.7 mJ/cm2 and using UV-sensitive (j)X174 to show
3-log inactivation.
Example 11. In Example 9, even though the RED measured with MS2 was 18 mJ/cm2
and the log inactivation measured with <)>X174 was 3.6 log, Figure F.4 shows that
Cryptosporidium, defined with a UV sensitivity of 3.9 mJ/cm per log, experienced a log
inactivation of 2.5 corresponding to an RED of 9.8 mJ/cm2.
UV Disinfection Guidance Manual F-l 1 ' June 2003
Proposal Draft
-------
Appendix F. Background to the UV Reactor Validation Protocol
F.1.4 Challenge Microorganism Selection
Ideally, UV reactor performance should be validated with a microorganism whose UV
sensitivity matches that of the target pathogen. In this guidance document, the UV sensitivity of
the target microorganisms is given by the dose requirements given in Chapter 1 for
Cryptosporidium, Giardia, and virus. Challenge microorganisms currently used to validate UV
reactors do not have a UV-sensitivity that matches the UV-sensitivity of the target pathogens as
defined in Chapter 1. The UV-resistance of MS2 and B. subtilis spores is notably greater than
that of Cryptosporidium and Giardia, and notably less than that of adenovirus. Furthermore,
demonstrating 3 or 4-log virus inactivation with these challenge microorganisms necessitates
demonstrating REDs greater than 150 ml/cm2. These REDs correspond to greater than 6-log
inactivation of MS2 and B. subtilis spores. Currently, culturing liters of challenge
microorganisms needed to demonstrate greater than 6-log inactivation are not practical.
A challenge microorganism should have reproducible UV inactivation kinetics over the
dose range of interest. The challenge microorganism should be easily prepared in high liters,
easily enumerated by an assay based on microorganism replication, non-pathogenic to humans,
and not harmful to the environment. If the challenge microorganism is a phage, the host bacteria
used to assay the phage concentration should not be pathogenic to humans. MS-2 phage,
non-pathogenic Escherichia coli, B. subtilis spores, and Saccharomyces cerevisae have been
used to bioassay UV reactors designed to treat drinking water. Table F.I summarizes the UV
sensitivity of commonly-used and candidate bioassay microorganisms.
Table F.1 UV Sensitivity of Bioassay Microorganisms and Candidates
Microorganism
MS-2 phage
£ Coli
B. subtilis spores
«|>x174 phage
B40-8 phage
PRD-1 phage .
Dose (mJ/cm') Reported to Achieve
1 log
16
3.0
28
2.2
12
9.9
2 log
34
4.8
39
5.3
18
17
3 log
52
6.7
50
7.3
23
24
4 log
71
8.4
62
11
28
30
Reference
Wilson etal. 1992
Chang etal. 1985
Sommeretal. 1998
Sommeretal. 1998
Sommeretal. 1998
Meng and Gerba 1996
F.2 Dose Monitoring
There are three approaches currently used to monitor dose delivery. In this guidance
document, the terms used are as follows:
• UV intensity setpoint approach ,
• UV intensity and UVT setpoint approach
• Calculated dose approach
UV Disinfection Guidance Manual
Proposal Draft
F-12
June 2003
-------
Appendix F. Background to the UV Reactor Validation Protocol
With the UV intensity setpoint approach, dose delivery is indicated by measured flowrate
and UV intensity. The UV reactor complies with a required dose delivery when the measured
UV intensity is above an alarm setpoint value defined as a function of flowrate through the
reactor. With this approach, the UV intensity sensor should be positioned far enough from the
lamp that it provides measurable responses to changing water UV absorbance (and
corresponding UVT) as well as lamp output. With the UV intensity and UVT setpoint approach,
dose delivery is indicated by measured flowrate, UV intensity, and UVT. The UV reactor
complies with a required dose delivery when the measured UV intensity and UVT are above
alarm setpoint values, both defined as a function of flowrate through the reactor. With this
approach, the UV intensity sensor should be positioned relatively close to the lamp so that it
responds primarily to changing lamp output. With the calculated dose approach, dose delivery is
indicated by a dose value calculated from measured flowrate, UV intensity, and UVT. The UV
reactor complies with'a required dose delivery when the calculated dose is above an alarm
setpoint value. With this approach, there are no requirements for sensor positioning.
To illustrate the UV intensity setpoint approach and the UV intensity and UVT setpoint
approach, Figures F.6, F.7, and F.8 present the relationship between UV dose and measured UV
intensity for an annular reactor containing a single LP lamp. UV intensity was calculated using a
radial UV intensity model and UV dose was calculated assuming ideal hydraulics (Haas and
Sakellaropoulos 1979). UV intensity and dose were calculated for a fixed flowrate of 400 gpm,
water UVT ranging from 60 to 98 percent, and lamp output ranging from 20 to 100 percent. In
each figure, data are presented as plots of dose versus UV intensity sensor reading for values of
UVT specified in the legend. For each of those plots, each point at a given UVT represents, in
order of increasing dose, operation at 20,40,60, 80, and 100 percent lamp power. The
differences between these figures are due to sensor placement.
Figure F.6 Relationship between UV Dose and Intensity for a UV Intensity
Sensor Located to Give Dose Proportional to Measured Irradiance
701
-------
Appendix F. Background to the UV, Reactor Validation Protocol
Figure F.7 Relationship between UV Dose and Intensity
for a UV Intensity Sensor Located Close to the Lamp
W-,
~. 60-
CM
s»-
-3 40
£»-
t)
(
-
400 GPM *
0 A
o A • x
A x o
.A X °
£ - * * J*'
8 * /• s * * • S'
* . , ,
320406080t»120140ie
Sensor (mW/cm2)
UVT
254 nm
• 60% .
X70%
-80% '
Q 85%
X 90%
A 94%
50 098%
Figure F.8 Relationship between UV Dose and Intensity
for a UV Intensity Sensor Located Far from the Lamp
CM
.O
1
^^
W
0
o
"*
70 -I
60 -
50 -
40 -
30
20 -
10 -
0 -
400 GPM .
A
X A 0
0 X A
0 X 0
'--••--••*-•-£-•
JK ~° 1 O '
p x Is :s-
r .
UVT
254 nm
* 60%
x 70%
- 80%
. o 85%
x 90%
A 94%
0 5 10 15 20 o98o/o
Sensor (mW/cm2)
F.2.1 UV Intensity Setpoint Approach
Figure F.6 presents the relationship obtained when the UV intensity sensor is located at a
distance from the lamps where UV dose is proportional to measured UV intensity regardless of
the UVT and lamp output. With an ideal reactor, this sensor location occurs where the measured
intensity equals the average intensity within the reactor. Because of the proportional relationship
between dose delivery and measured intensity, a given intensity can be related to a specific level
of dose delivery.
Example 12. The UV reactor characterized in Figure F.6 is used in a disinfection
application needing a UV dose of 20 ml/cm2. At a flowrate of 400 gpm, a UV intensity value S
UV Disinfection Guidance Manual
Proposal Draft
F-14
June 2003
-------
Appendix F. Background to the UV Reactor Validation Protocol
of 18 mW/cm2 is used as an alarm setpoint to indicate the UV reactor delivers a dose of
20 ml/cm2. This alarm setpoint value will indicate a dose of 20 ml/cm2 regardless of the UVT of
the water and the output of the lamps.
Figure F.7 presents the relationship between dose delivery and measured UV intensity
when the UV intensity sensor is placed closer to the lamp than the sensor in Figure F.6. Because
the sensor views the lamp through a relatively thin water layer, the sensor response to changing
UVT is small compared to that in Figure F.6. Accordingly, the relationship between dose
delivery and measured intensity for different values of UVT cannot be described by a single
proportional relationship. Unlike Figure F.6, a given UV intensity is not related to a specific
level of dose delivery but is related to a range of delivered doses. Accordingly, the measured
UV intensity should only be used to indicate dose delivery at the lower end of that range, which
occurs under conditions of maximum lamp power and reduced UVT.
Example 13. The UV reactor characterized in Figure F.7 is used in an application
needing a UV dose of 20 mJ/cm2. The UV manufacturer states that a UV intensity value 5 of
80 mW/cm2 will indicate a dose of 20 mJ/cm2 under design conditions of 85 percent UVT and 60
percent lamp output. However, as shown in Figure F.7, an intensity of 80 mW/cm2 corresponds
to a dose ranging from 5 to 37 mJ/cm2. The lower end of this range occurs with lamp powers
higher that 60 percent and water UVT lower than 85 percent. For a UV intensity alarm setpoint
to ensure a dose of 20 mJ/cm2 under all possible conditions of the water UVT and lamp output, a
setpoint value 5' of 134 mW/cm2 should be chosen.
Figure F.8 presents the relationship between dose delivery and measured UV intensity
when the UV intensity sensor is located further from the lamps than the sensor in Figure F.6.
Because the sensor views the lamp through a relatively thick water layer, the sensor response to
changing water transmittance is large compared to that in Figure F.6. Like Figure F.7, the
relationship between dose delivery and measured intensity for different values of UVT cannot be
described by a single proportional relationship. As such, a given intensity value is related to a
range of dose values as opposed to a single value. Again, the measured UV intensity should only
be used to indicate dose delivery at the lower end of that range. However, unlike Figure F.7, the
lower end of the range occurs under conditions of reduced lamp power and maximum UVT.
, Example 14. The UV reactor characterized in Figure F.8 is used in an application
needing a UV dose of 20 mJ/cm2. The UV reactor uses the UV intensity setpoint approach to
monitor dose delivery. A UV intensity alarm setpoint value S of 4 mW/cm2 is proposed based on
the UV intensity measured under design conditions of 85 percent UVT and 60 percent lamp
output. However, an intensity of 4 mW/cm2 indicates a dose ranging from 9 to 26 mJ/cm2. To
indicate a dose of 20 ml/cm2 using the UV intensity setpoint approach, a setpoint value S' of 8
mW/cm2 should be chosen.
The location of the UV intensity sensor within a UV reactor is selected by the
manufacturer of the UV reactor. If the UV reactor uses the UV intensity setpoint approach for
dose monitoring, the UV manufacturer should optimize the UV intensity sensor's location to
give a proportional relationship between dose delivery and measured UV intensity similar to the
example given in Figure F.6. If the UV manufacturer does not optimize the UV intensity
sensor's location, a given UV intensity will correspond to a range of UV doses values as opposed
to a single value. While this does not prevent the UV reactor from using the UV intensity
UV Disinfection Guidance Manual F-15 June 2003
Proposal Draft
-------
Appendix F. Background to the UV Reactor Validation Protocol
setpoint approach, the monitoring approach will not be as efficient as with an optimally located
sensor because the UV reactor will be overdosing at many combinations of UVT and lamp
power that given rise to operation at the setpoint.
F.2.2 UV Intensity and UVT Setpoint Approach
If the UV intensity sensor is not at a location optimal for the UV intensity setpoint
approach, measurements of UVT can be used to provide more efficient dose monitoring. UVT
alarm setpoints combined with UV intensity alarm setpoints can be used to indicate dose delivery
providing the UV intensity sensor is placed relatively close to the lamp. With the sensor located
relatively close to the lamp, dose delivery at a given intensity and flowrate decreases with
decreasing UVT (Figure F.7). Accordingly, a UVT alarm setpoint combined with a UV intensity
alarm setpoint provides a meaningful indicator of dose delivery.
Example 15. The UV reactor characterized in Figure F.7 is used in an application
needing a UV dose of 20 ml/cm2. If the UV reactor used the UV intensity setpoint approach to
monitor dose delivery, an alarm setpoint S' of 134 mW/cm2 would be used to indicate a dose
delivery of 20 mJ/cm2. This approach is not efficient because a UV intensity of 134 mW/cm2 is
associated with a UV dose ranging from 20 to 60 mJ/cm2. An alternative approach for dose
monitoring is to use the UV intensity and UVT setpoint approach. Under this approach, a UV
intensity alarm setpoint S of 80 mW/cm2 combined with a UVT'alarm setpoint of'85 percent will
indicate a dose delivery of 20 mJ/cm2. However, the approach is still inefficient because UV
dose may range from 20 to 38 mJ/cm2 with operation of the reactor at the setpoint conditions.
If the UV intensity sensor is located at the optimal position for the UV intensity setpoint
approach (Figure F.6), the UVT reading does not provide any additional information on dose
delivery that is not provided by the measured UV intensity. However, the measured UVT could
be used to indicate whether a UV intensity alarm condition arises from low UVT.
If the UV intensity sensor is located too far from the lamp, dose delivery at a given UV
intensity and flowrate increases with decreasing UVT (Figure F.8). As such, the UVT reading
cannot be used as an alarm setpoint to indicate dose delivery.
Example 16. The UV reactor characterized in Figure F.8 uses the UV intensity and UVT
setpoint approach to show the UV reactor delivers a dose of 20 mJ/cm2. The intensity alarm
setpoint is set to 5 mW/cm2 and the UVT alarm setpoint is set to 90 percent. If the reactor was
operating with a measured UVT and UV intensity of 85 percent and 5 mW/cm2, the delivered
dose would be 28 mJ/cm2. If the reactor was operating with a UVT and UV intensity of 98
percent and 5 mW/cm2, respectively, the delivered dose would be 12 mJ/cm2. Thus the two
alarm setpoint values are not ensuring the UV reactor complies with a dose of 20 ml/cm2. To
remedy this problem, the UV manufacturer should either uses the UV intensity setpoint
approach, move the UV intensity sensor closer to the lamps, or use the calculated dose approach
to monitor dose delivery.
UV Disinfection Guidance Manual F-16 June 2003
Proposal Draft
-------
Appendix F. Background to the UV Reactor Validation Protocol
F.2.3 Calculated Dose Approach
Measurements of flowrate, UV intensity, and UVT can be incorporated into theoretical,
empirical, or semi-empirical calculations of dose delivery. For example, the relationships
represented in Figures F.6 to F.8 could be defined experimentally and used in an empirical
manner to calculate dose. Relationships could also be defined using advanced modeling
approaches and used to relate measured intensity to dose delivery for a given flowrate and UVT.
In theory, the dose calculation does not necessitate that the sensor be placed at any one location
within the reactor. However, if the sensor placed at a location that gives dose delivery
proportional to the sensor reading, the dose calculation does not require UVT as an input
parameter.
. t
F.2.4 Validating Dose Monitoring
The test conditions used to validate a UV reactor should depend on the approach'used to
monitor dose delivery.
If the UV reactor uses the UV intensity setpoint approach, the UV reactor is validated by
measuring the dose delivery with the UV intensity adjusted to the UV intensity alarm setpoint
value. The combination of lamp power and UVT used to achieve operation at the alarm setpoint
should be selected to capture the lower end of the dose range associated with the setpoint. If the
UV intensity sensor is located closer to the lamp than the optimal location, the UV reactor should
be validated at peak lamp power and lowered UVT. If the UV intensity sensor is located further
from the lamp than the optimal location, the UV reactor should be validated at peak UVT and
lowered lamp power. If the positioning of the UV intensity sensor relative to the optimal
location is riot known prior, to validation testing, the UV reactor should be validated using both
test conditions. If the dose values measured with both test conditions are the same, the UV
intensity sensor is at the optimal location.
If the UV reactor uses the UV intensity and UVT setpoint approach, the UV reactor is
validated by measuring dose delivery with the UV intensity and UVT adjusted to the alarm
setpoint values. Validation should also confirm that the UV intensity sensor is located close
enough to the lamp that UVT alarm setpoint values provide a meaningful indicator of dose
delivery. This is accomplished by showing that dose delivery decreases with decreasing UVT
while the UV intensity is held constant at the intensity alarm setpoint value. If dose delivery
increases with decreased UVT, the UV intensity sensor is located too far from the lamp and this
monitoring approach will not work.
If the reactor uses dose calculations, validation testing confirms that dose delivery is
greater than or equal to the calculated dose. Validation testing is conducted at various
combinations of flowrate, lamp output, and UVT that result in performance at a target dose. This
proves the dose calculation is robust over the range of those variables expected with operation of
the reactor at a water treatment plant (WTP).
UV Disinfection Guidance Manual F-17 ' June 2003
Proposal Draft •'"•
-------
Appendix F. Background to the UV Reactor Validation Protocol
F.3 UV Intensity Sensors
UV reactors should be equipped with at least one on-line UV intensity sensor that
measures the UV intensity at some point within the UV reactor. Measurements made by the
on-line UV sensors are used.to indicate dose delivery by the UV reactor. Reference sensors are
used to check that the measurements made by the on-line sensors are valid.
F.3.1 UV Sensor Properties
The UV sensor may or may not measure the UV light through a monitoring window that
is separate from the sensor body. The monitoring windows should have a high UVT over the
spectral response range of the UV sensors.
The UV intensity sensor should detect germicidal UV radiation and produce a
standardized output signal (e.g., 4 to 20 mA) proportional to the UV irradiance incident on the
sensor. UV intensity sensors should be calibrated to an absolute irradiance standard and have a
suitable measurement range, angular response, spectral response, linearity, and stability for
monitoring and controlling UV dose delivery by the UV reactor. An ideal UV intensity sensor
has a linear response to incident UV irradiance that is independent of water temperature and does
not degrade with time. Furthermore, the ideal sensor has a fixed angular response and a
wavelength response that mimics the germicidal response of microorganisms.
UV intensity sensors provided by the manufacturer should be individually calibrated.
UV intensity sensors used to monitor DP lamps are often calibrated using the substitution method
(Larason et al. 1998). With this approach, the intensity of a collimated beam of UV light at 254
nm is measured using the UV sensor and compared to that made using a standard measurement,
such as a National Institute of Standards and Technology (NIST) traceable sensor or chemical
actinometer. The ratio of the standard measurement to the sensor output is the calibration factor.
With sensors designed to measure the output of medium-pressure (MP) lamps, the sensor can be
either calibrated at 254 nm, calibrated as a function of wavelength, or calibrated using
polychromatic light from a MP lamp with a known spectral output. Regardless of the approach
used, the calibration should be traceable to some absolute measurement standard and have a
quantified measurement uncertainty.
Sensor linearity is determined by comparing the sensor output as a function of incident
irradiance to standard measurements of that irradiance. Sensor temperature response is
determined by measuring the dependence of sensor output on the sensor's operating temperature
with the sensor measuring a constant irradiance. Both linearity and temperature response should
be determined over the range of irradiance and temperature expected with the operation of the
UV reactor at the WTP. Angular response of a sensor is determined by measuring the
dependence of the sensor output on the incident angle of collimated UV light of fixed intensity.
The spectral response of a sensor is determined by measuring the dependence of the
sensor output on the wavelength of monochromatic light of known irradiance incident on the
sensor. Spectral response is typically presented as a plot of the ratio of sensor output to incident
irradiance as a function of the wavelength of light. Because it may be affected by infrared
UV Disinfection Guidance Manual
Proposal Draft
F-18
June 2003
-------
Appendix F. Background to the UV Reactor Validation Protocol
transmission of glass filters and fluorescence of diffusers that are part of the sensor (Larason and
Cromer 2001), UV intensity sensor spectral response should be evaluated from 200 to 1000 nm.
The long-term stability of a UV sensor is best-determined using field data but may be
estimated using accelerated life cycle testing. The measurement accuracy of UV sensors can
change over time with operation and environmental exposure. Temperature cycling, exposure to
UV light, mechanical vibration, and other factors will impact the linear, spectral, angular, and
temperature response of a sensor.
The UV sensor manufacturer should conduct regular testing on manufactured UV sensors
to develop a database on sensor properties. While some sensor properties may be measured with
each sensor, other properties, such as long-term stability, can only be measured on a
representative lot size. The sensor manufacturer should have available for inspection the
following information:
s
• Description of the properties measured
• Description of the measurement system used to measure each property
• Description of the measurement standards used
• Documented uncertainty of each measurement
• Description of QA/QC procedures used to ensure the measurements are traceable
• Data collected over time that demonstrates that the properties of the manufactured
sensors meet specifications
F.3.2 UV Intensity Sensor Measurement Uncertainty
The measurement uncertainty of a UV intensity sensor quantifies how the measurement
of UV intensity made by the sensor when mounted on the UV reactor compares with the true
value. For the purposes of this guidance document, UV intensity sensor uncertainty should be
determined at a 90 percent confidence level by summing the uncertainty that arises from the
calibration, linearity, angular and spectral response, temperature response, and long-term
stability.
The uncertainty of sensor calibration depends on the uncertainty of the standards and
instrumentation used to calibrate the sensor, such as voltmeters and amplifiers. Uncertainty
arises from linearity and temperature response because sensor calibration factors, determined at a
given temperature and UV irradiance, are used over a range of temperatures and irradiances with
operation of the sensor with the UV reactor. Uncertainty arises with sensor degradation because
calibration factors are determined on new sensors.
Uncertainty arises with angular response because sensors, calibrated using collimated UV
light, are used in UV reactors to measure UV light impacting from different directions.
Uncertainty arises with spectral response because sensors, calibrated at a fixed wavelength, are
UV Disinfection Guidance Manual F-19 June 2003
Proposal Draft
-------
Appendix F. Background to the UV Reactor Validation Protocol
used in UV reactors equipped with MP lamps. Variability in spectral and angular response from
sensor to sensor will result in a measurement uncertainty not accounted for in calibration. The
impact of spectral and angular response variability on sensor measurement uncertainty can be
determined either by calculation or by measurement. In the first approach, the sensor spectral
and angular response measured on a representative lot size is used as an input to a model that
predicts sensor readings in a UV reactor. The variability in the sensor readings predicted by the
model is used to define an uncertainty term that is included in the calculation of sensor
uncertainty. In the second approach, the variability in measurements made by a representative
number of sensors mounted on the UV reactor is used to define the uncertainty.
Example 17. A UV sensor manufacturer calibrates each manufactured UV intensity
sensor at 20°C with an uncertainty of 5 percent. UV intensity sensor linearity, temperature
response, angular response, and spectral response is evaluated on every tenth sensor
manufactured. Linearity ranges from 1 to 3 percent over the measurement range of the sensor.
Temperature response ranges from 0.1 to 0.2 percent per C°, or an uncertainty of 4 percent from
0 to 40°C. Models predict that the variability in angular and spectral response from sensor to
sensor will cause uncertainties of 8 and 4 percent, respectively. An evaluation of sensors
returned from the field indicates that the long-term drift over a one-year period is 10 percent.
The measurement uncertainty of the sensors is calculated as the square root of the sum of the
squares of the individual uncertainties as per:
Measurement uncertainty = V52 + 32 + 42 + 82 + 42 +102 =15 percent
F.3.3 On-line and Reference UV Intensity Sensors
Degradation in UV intensity sensor performance can lead to significant under- or over-
estimations of dose delivery by the UV reactor's on-line monitoring system. To prevent
underdosing, the measurement uncertainty of the UV intensity sensors should be incorporated as
a safety factor into the sizing and operation of a UV installation and the performance of the on-
line sensor should be regularly checked by use of a reference sensor. Measurements made by the
on-line and reference sensor should meet the following equation:
lRef
1/2
EquationF.il
where
Ikef -
I Duty =
0"Duty
Intensity measured with the reference sensor (W/m2)
Intensity measured with the duty sensor (W/m2)
Measurement Uncertainty of the reference UV intensity sensor (%)
Measurement Uncertainty of the duty UV intensity sensor (%)
If this condition is not met, the cause for the discrepancy should be determined and
resolved. Typically, the discrepancy indicates degradation of the on-line sensor that necessitates
recalibration or replacement.
UV Disinfection Guidance Manual
Proposal Draft
F-20
June 2003
-------
Appendix F. Background to the UV Reactor Validation Protocol
Example 18. A UV reactor uses on-line sensors with an uncertainty of 15 percent. A
reference sensor with an uncertainty of 5 percent is used to check the on-line sensors when the
UV reactor is operating at the WTP. Measurements made by the on-line sensors are considered
out of spec when:
-l xlOO<[l5
'
F.3.4 Positioning of UV Intensity Sensors
While the UV output along the length and around the circumference of a new UV lamp
will be relatively uniform, this may not be true with aged or fouled lamps. Sputtering of
electrode material leads to deposits on the inside of the lamp sleeve within 2 or 3 inches from the
electrode. Discoloration of the lamp sleeve with lamp aging varies along the length of the lamp.
Sleeve fouling varies spatially both along the length and circumference of the lamp sleeve (Lin et
al. 1999).
i If lamps experience non-uniform aging along their length, the UV intensity sensor should
be located to monitor the section along the lamp that experienced the greatest decrease in UV
output with aging. The sensor should not be located to monitor the section that experiences the
least decrease in UV output.
F.3.5 Number of UV Intensity Sensors
Variability in UV output from lamp to lamp impacts both dose delivery and monitoring.
A lamp with a lower output will deliver lower doses to microorganisms passing in its vicinity,
thereby shifting the dose distribution to lower values and reducing the net performance of the
reactor. The shift in the dose distribution will be more pronounced with a reactor with fewer
lamps. Because the dose distribution is affected, the impact on net performance will be greater
with a more UV-sensitive microorganism. If the number of UV intensity sensors is less than the
number of lamps and the sensors monitor those lamps with the highest output, the monitoring
system will overestimate dose delivery by the UV reactor.
The monitoring strategy used to ensure that UV dose delivery meets regulatory targets
should account for the variability of UV output from lamp-to-lamp. If each lamp in the reactor is
monitored by a UV intensity sensor, dose delivery compliance should be based on the lowest
lamp output, unless an accepted and validated dose calculation methodology can account for
lamp-to-lamp variability. If the number of sensors used is less than the number of lamps, either
the lamp with the lowest output should be monitored and used for dose compliance, or the
setpoint used for dose delivery compliance should include a safety factor to account for lamp-to-
lamp variability.
Example 19. A UV reactor installed at a WTP is equipped with four lamps and two UV
intensity sensors. Because of variability in lamp output, the UV intensity 5 cm from each lamp
is 15,10, 8, and 20 mW/cm2, respectively. If one sensor monitors the first lamp and the second
UV Disinfection Guidance Manual F-21 June 2003
Proposal Draft
-------
Appendix F. Background to the UV Reactor Validation Protocol
monitors the forth lamp, the monitoring system will over-estimate the dose delivery by the UV
reactor because microorganisms passing by the second and third lamps will receive lower doses
than the microorganisms passing by the first and fourth lamps.
During UV reactor validation, variability in UV output from lamp to lamp should not
cause the UV reactor to be overrated. If the number of sensors is less than the number of lamps,
the sensors should be monitoring the lamps with the lowest output. If UV intensity sensors
record different values during validation, intensity setpoints and calculations should be based on
the lowest values recorded.
Example 20. A UV reactor undergoing validation is equipped with four lamps and two
sensors. Dose delivery is monitored using the UV intensity setpoint approach. Because of
variability in lamp output, the UV intensity 5 cm from each lamp is 10, 15, 8, and 12 mW/cm2,
respectively. To ensure validation results are meaningful, the sensors should be monitoring the
first and third lamps.
F.4 Polychromatic Considerations
With UV reactors equipped with LP or low pressure high output (LPHO) lamps, dose
delivery and monitoring occurs at a single wavelength of 254 nm. With UV reactors equipped
with MP lamps, dose delivery and monitoring involves a response to multiple wavelengths.
Dose delivery is an integrated response to UV light from 200 to 320 nm. The output from the
UV intensity sensor is an integrated response to UV light over wavelengths spanning the sensor's
spectral response. UV absorbance monitors typically measure UV absorbance at a single
wavelength of 254 nm. If the spectral properties of the UV reactor that influence dose delivery
and monitoring during operation of the UV installation at a WTP are the same as the spectral
properties during validation, then the same dose delivery and monitoring characterized during
validation will occur at the WTP. However, if the spectral properties are different, dose delivery
and monitoring at the WTP will differ from dose delivery and monitoring measured during
validation. The following spectral properties may differ:
• Action spectra of the challenge microorganism used during validation and the target
pathogen
• Spectral UV absorbance of the water during validation and at the WTP
• UV output of the lamps during validation and at the WTP
• UVT of the lamp sleeves during validation'and at the WTP
Safety factors should be applied to the validation data for polychromatic UV reactors if
spectral differences will lead to under dosing at the WTP. This section describes approaches for
assessing the impact of differences in spectral properties and deriving those safety factors.
UV Disinfection Guidance Manual F-22 June 2003
Proposal Draft
-------
Appendix F. Background to the UV Reactor Validation Protocol
F.4.1 Action Spectra
' The dependence of microorganism inactivation kinetics on UV wavelength may be
described using an action spectrum - the UV inactivation sensitivity as a function of wavelength
(Figure F.9). Ideally, the action spectrum of the challenge microorganism used to validate a
polychromatic UV reactor would either match that of the target microorganism or provide a
conservative estimate of inactivation.
Figure F.9 Action Spectra for Various Microorganisms1
2.5 n
Herpes simplex
-e-MS2,R-17,fr,7-S
-K-oX-174
-A-T2
225 235 245 255 265 275 285 295 305
Wavelength (nm)
(Adapted from Rauth 1965)
The impact of various action spectra on UV dose delivery may be estimated by
calculating the germicidal lamp output using Equation F. 12:
320
Equation F.I2
where
Per
X .
P(X)
G(X)
AX
Germicidal output of the MP lamp (W/cm)
Wavelength (nm)
Lamp output (W/nm) measured over 1 nm increments at wavelength X
Relative UV sensitivity of the microorganism at wavelength X
1 nm increment .
Using the action spectra published for fourteen microorganisms (Rauth 1965, Cabaj et al.
2002, Linden 2001), Table F.2 presents the germicidal lamp output calculated for a MP lamp and
the ratio of that output to that of Cryptosporidium. A ratio greater than one indicates that the-
action spectra of the microorganism favors greater inactivation than the action spectra of
Cryptosporidium, If a challenge microorganism with a ratio greater than one is used to validate a
UV Disinfection Guidance Manual
Proposal Draft
F-23
June 2003
-------
Appendix F. Background to the UV Reactor Validation Protocol
MP reactor for Cryptosporidium inactivation, the ratio should be used as a correction factor to
account for the greater inactivation of the challenge that arises from the differences in action
spectra. In the case of MS2 and 5. subtilis, the ratio is close to one and the correction is small.
However, based on the data in Table F.2, a correction factor of 1.16 would be needed with UV
reactors equipped with MP lamps if X174 was used to show Cryptosporidium inactivation.
Table F.2 Germicidal Output Delivered to 14 Microorganisms by a MP Lamp
Microorganism
Cryptosporidium oocysts
MS-2, R-17, fr, 7-S
8. subtilis spores
OX174
Reovirus-3
Polyoma
T2
VSV
Vaccinia
EMC
Herpes simplex
Type / Nucleic acid
(SS = Single Strand,
DS = Double Strand)
DSDNA
Phage / SS RNA
DSDNA
Phage / DS DMA
Animal virus / DS RNA
Animal virus / DS DNA
Phage /DSDNA
Animal virus / RNA
Animal virus / DS DNA
Animal virus / SS RNA
Human virus / DS DNA
Germicidal
Output (W/cm)
5.64
5.78
5.58
6.53
7.46
6.74
6.05
5.53
5.46
5.98
7.00
Germicidal Output
Relative to
Cryptosporidium
1.00
1.04
0.99
1.16
1.32
1.18
1.07
, 0.99
0.98
1.07
1.26
The germicidal output of the MP lamp calculated using the action spectra of B. subtilis
spores and MS2 is equal to or less than that of most of the 14 microorganisms listed in Table F.2.
It is thus reasonable to assume that these microorganisms are acceptable as challenge
microorganisms for many pathogens whose action spectrum is not known, like adenovirus and
Giardia. However, if an alternative challenge microorganism is to be used, its action spectra
should be assessed for suitability.
As an alternate approach to measuring the action spectrum and using Equation F.I2, the
correction factor can also be estimated by comparing the dose-response of the challenge
microorganism to that of MS2 measured with a LP and MP lamp. The correction factor would
be defined as:
Safety Factor = 1.
/Challenge^
Equation F.I3
where
,ku» =
1.04 =
Slope of the dose-response measure with the MP collimated beam (cm2/mj)
Slope of the dose-response measure with the LP collimated beam (cm2/mj)
Germicidal output of MS2 relative to Cryptosporidium, from Table F.2
UV Disinfection Guidance Manual
Proposal Draft
F-24
June 2003
-------
Appendix F. Background to the UV Reactor Validation Protocol
The correction factor that accounts for differences in the action spectra is not the same
correction factor that accounts for differences in the UV sensitivity described in section F.I.2.
The correction factor described in section F.I.2 applies to all UV reactors regardless of lamp
type. The correction factor described in this section is applicable to MP reactors. It should be
used in addition to the correction factor described in section F.I.2.
F.4.2 Water Absorption
During UV reactor validation, a UV-absorbing chemical is added to the water passing
through the reactor in order to simulate high UV absorbance events that could occur at the WTP.
UV-absorbing chemicals that have been used to validate UV reactors include sodium thiosulfate,
fluorescein, coffee, tea, and parahydroxybenzoic acid. Ideally, the spectral absorption of the
water used to validate UV reactors equipped with MP lamps should match the spectral
absorption of the water at the WTP over the wavelength range associated with dose delivery and
monitoring (Figure F.I0).
Figure F. 11 compares the UV absorbance spectra of coffee and lignin sulphonate to that
of two drinking water sources (Water A and Water B). For a given UVT at 254 nm, the UV
absorption at wavelengths above and below 254 nm is greater with coffee, tea, and lignin
sulphonate than with the drinking water sources. If those chemicals are used during validation of
a MP reactor, the RED and UV intensity measured at a given flowrate, lamp output, and water
UVT will be lower during validation than at the WTP.
Figure F.10 Spectral UV Absorption of Water at Various WTPs
0.50
0.40 \
'§ 0.30 H
8 0.20 •
.a
0.00'
200'-, 220 - .240 ' 260 ,S280 300
"''. ' : Wavelength (hmj
UV Disinfection Guidance Manual
Proposal Draft
F-25
June 2003
-------
Appendix F. Background to the UV Reactor Validation Protocol
Figure F.11 Comparison of the UV Absorbance Spectrum of Additives used
during UV Reactor Validation to the UV Absorbance of Two Finished Waters
- - 0.20 -
Lignin sulphonate
. Coffee
;250- 300 350
Wavelength (nm)
400
The impact of the difference in the UV absorbance spectra on the measured intensity will
depend on sensor placement relative to the lamps. If the sensor is located close to the lamps, the
sensor reading during validation will be only slightly lower than the reading at the WTP.
Accordingly, for a given sensor reading, flowrate, and water UVT, the RED delivered at the
WTP will be greater than the RED measured during validation. However, if the sensor is placed
far enough from the lamp, the UV intensity measured during validation will be much lower than
the reading at the WTP. As such, for a given sensor reading, flowrate, and water UVT, the RED
delivered at the WTP will be less than the RED measured during validation. If the UV intensity
sensor's spectral response mimics the microorganism's action spectra and the sensor is located at
a position where the dose delivery is proportional to the sensor reading, the RED delivered at the
WTP will equal the RED measured during validation, even with the differences in the UV
absorbance spectra shown in Figure F. 11 (Wright et al. 2002). However, this relationship will
not hold true if the sensor's spectral response deviates sufficiently from the microorganism's
action spectra.
Modeling approaches can be used to predict and compare the RED and UV intensity
sensor readings obtained during validation to those expected at a WTP. The modeling approach
can be used to define correction factors applicable to validation results to ensure dose monitoring
provides valid measurements at the WTP. UV intensity readings should be predicted using
polychromatic intensity models that factor in the spectral and angular response of the sensor.
While RED predictions could be obtained using CFD-based dose modeling approaches, ideal
dose delivery models should be used to provide conservative correction factors. The ideal dose
delivery model is conservative because the sensor location within a reactor where the dose
delivery is proportional to sensor reading is predicted to occur closer to the lamp with the ideal
model than with a CFD-based dose delivery model (Wright et al. 2002). As such, the transition
to a correction factor greater than one occurs with closer sensor-to-lamp distance with the ideal
dose delivery model than with the CFD-based dose delivery model.
UV Disinfection Guidance Manual
Proposal Draft
F-26
June 2003
-------
Appendix F. Background to the UV Reactor Validation Protocol
Table F.3 provides predictions of dose delivery and sensor measurements for an ideal
annular reactor. The reactor consists of a cylinder with an 1 8.81 -cm radius and a length greater
than the arc length of the lamp. The reactor is equipped with a single MP lamp oriented along
the central axis of the cylinder (i.e., at a radius of 0 cm). The lamp is housed in a lamp sleeve
with a radius of 3.81 cm. The spectral output of the lamp is given in Figure F.12. The spectral
UV absorbances used in the model are provided in Figure F.I 1. UV intensity was modeled using
a polychromatic radial intensity model and the dose was calculated as the product of the average
germicidal intensity and the hydraulic residence time as per the following equation:
EquationF.14
00 -&eV)
where
D = ' Dose delivered by the reactor (ml/cm2)
Larc = Arc length of the lamp (cm)
Tq(X) = Lamp sleeve UVT
O^(X) = Naperian UV absorbance
rwi = Reactor water layer, defined as the radial distance from the sleeve to the reactor
wall (cm)
Q = Flowrate through the reactor (cm3/s)
UV intensity sensor measurements were modeled at different lamp-to-sensor distances
for sensors with the spectral response shown in Figure F.I 3 as per Equation F.I 5:
, ..
/= > - - - Equation F.I 5
where
I = Intensity measured by the sensor
= .Sensor spectral response normalized to unity at 254 nm
= Distance from the sensor to the lamp (cm)
= Lamp sleeve outer radius (cm)
UV Disinfection Guidance Manual F-27 June 2003
Proposal Draft
-------
Appendix F. Background to the UV Reactor Validation Protocol
Table F.3 Dose and UV Intensity Sensor
Measurements Modeled for a MP Annular Reactor
Performance Parameters
MS2 RED (mJ/cm*> .
Sensor
SiC
Filtered SiC
Water Layer (cm)1
2.0
5.0
1 10
15
20
2.0
5.0
10
15
20
Water A
72
Water B
67
Measured UV Intensity {
269
136
59.7
31.9
19.2
112
48.0
15.3
5.02
2.42
256
122
48.7
23.6
13.0
107
44.8
13.7
4.97
1.95
Coffee
60
Lignin
Sulphonate
61
254 nm equivalent roW/cm")
238
101
31.6
11.7
4.77
103
10.9
3.46
1.18
0.410
245
110
40.4
18.2
9.34
104
11.8
4.06
1.51
0.593
1 Water layer is defined as the distance between the lamp sleeve and the UV intensity sensor.
Figure F.12 UV Output of a MP Lamp
3.0 -|
I2"
| 2.0 -
f 1 .0 -
0
> 0.5 -
0.0 :
2(
• ; f ~ - ' ' •
.
! Ji^judluAiA
I
- A
)0 -250 ' 300 350 400
Wavelength (nm)
Table F.3 presents the MS2 RED and sensor measurements predicted for the annular
reactor operating at a flowrate of 200 gpm, a water UVT of 85 percent at 254 nm, and 100
percent lamp power. As expected, the dose delivered with coffee and lignin sulphonate for a
given flowrate, water UVT, and lamp power was less than the dose delivered with both WTP
waters.
For a given sensor reading, flowrate, and UVT, Table F.4 presents the ratio of the dose
measured during validation to the dose delivered at the WTP calculated using the data from
Table F.3. A ratio greater than one indicates that the dose measured during validation will be
greater than the dose delivered at the WTP. As expected, the ratio is less than one with the UV
UV Disinfection Guidance Manual
Proposal Draft
F-28
June 2003
-------
Appendix F. Background to the UV Reactor Validation Protocol
intensity sensor located close to the lamps and greater than one with the UV intensity sensor
located far from the lamps. For a given sensor position, the ratio with lignin sulphonate is closer
to one than the ratio with coffee indicated that lignin sulphonate better matches the UV
absorption spectra of WTP waters. The ratio is also closer to one with a germicidal sensor
spectral response compared to the non-germicidal response. This indicates that validation results
with a germicidal sensor are more representative of performance at a WTP than validation results
with a non-germicidal sensor.
Table F.4 Impact of Water UV Absorbance on the UV Intensity
Sensor Value Associated with a Given UV Dose Delivery
UV Sensor
SIC
Filtered SiC
Water Layer
(cm)
2
5
10
15
20
2
5
10
15
20
Ratio of Dose Delivered During Validation to Dose Delivered at
the WTP for a Given Sensor Reading
Coffee to
Water A
0.93
1.12
1.56
2.25
3.34
0.89
0.98
1.16
1.39
1.70
Coffee to
Water B
0.96
1.09
1.37
1.80
2.44
0.93
0.99
1.12
1.29
1.48
Lignin
Sulphonate to
Water A
0.93
1.04
1.25
1.48
1.74
0.91
0.98
1.09
1.21
1.35
Lignin
Sulphonate to
Water B
0.95
1.01
1.10
1.19
1.27
0.94
0.99
1.06
1.12 '
1.18
For the germicidal sensor, Figure F. 13 presents the ratio of the dose expected with coffee
to the dose expected with finished Water A as a function of sensor position and water UVT.
With the sensor located close to the lamp, the ratio is less than one over a wide range of water
UVT values. However, the ratio increases above one with increased sensor-to-lamp water layer
and, for the most part, increases with decreased UVT.
UV Disinfection Guidance Manual
Proposal Draft
F-29
June 2003
-------
Appendix F. Background to the UV Reactor Validation Protocol
Figure F.13 Comparison of Dose Expected with Coffee as a UV Absorber to Dose
Expected with WTP Water for a MP Reactor Equipped with a Germicidal Sensor
.. » = 3.0 ^
8 - -•
• , a
" o < •'
- «'l
Si :
• .£ £ .
8 "^ *i n
u. 1.0 •
o £
'S "5
" -.2
1- ^0.0'-
•7
*r
1 '*.
' ">''*>
** -—
.«—„-,
V
\
"• y
:'",-
""^-
V
V N
I-, . f
•"-^
-
v
O^x,
•- «. , _h*-
t^SSTfti
Sensor: Filtered 1.
Absorber:;Coffee
2cm
— . ^ cm
— * —10 cm
* * * r 15:cm
20 cm
25 cm
D .75 80 ;/' 85 --90 "95 100,
UVTransmittance 254 nm (%)
For a given UV reactor equipped with MP lamps, the impact of differences in the spectral
UV absorbance between validation and operation at a WTP should be evaluated and used to
establish correction factors. The correction factor is calculated for a given flowrate, sensor
reading, and UVT, as the ratio of the dose expected during validation to the dose expected at the
WTP. If the ratio is less than one, no correction factor is needed.
FAS Spectral Shifts
Spectral shifts in the UV output of MP lamps may occur as MP lamps age. Spectral
shifts in the UVT of light through lamp sleeves may occur as sleeves age and undergo internal
and external fouling. Spectral shifts in the UVT of sensor windows may occur with window
fouling. Spectral shifts associated with the lamp-sleeve assembly will impact both dose delivery
and monitoring, while spectral shifts associated with window fouling will impact monitoring
only.
Figure F.I4 presents reported data on the spectral shift in MP lamp output and lamp
sleeve UVT experienced with aging. Figure F.I 5 presents data comparing the UVT of clean and
fouled lamp sleeves. In both cases, aging and fouling have reduced the output of low-
wavelength UV light from the lamp/sleeve assembly more than the output of higher wavelength
UV light. The impact of lamp and sleeve aging and sleeve fouling can be assessed by validation
testing. Alternatively, the impact can be modeled and used to define a correction factor
applicable to validation results generated using new lamps.
UV Disinfection Guidance Manual
Proposal Draft
F-30
June 2003
-------
Appendix F. Background to the UV Reactor Validation Protocol
Figure F.14 Spectral Shifts in the MP Lamp Output and
Lamp Sleeve UVT Reported with Aging1
S
1.0
0.6 -
o 0.4
I 0.2 H
0.0
Sleeve UVT
Lamp Output
200 250 : 300 350
.'/•";;' Wavelength (nm)
400
1 Adapted from Phillips 1983 and Kawaret al. 1998.
Figure F.15 Comparison of the UVT
of New and Fouled Lamp Sleeves
I 0.8 ^
t>
I
For a measured flowrate, water UVT, and UV intensity, Figures F.I6, F.I7, and F.I8
provide the ratio of the dose delivered with new lamps and sleeves to the dose delivered with
aged lamps, aged sleeves, and fouled sleeves, respectively. In each figure, the dose ratio is
presented as a function of water UVT and sensor-to-lamp water layer for two different sensors.
One sensor had a SiC spectral response while the other had a germicidal response. Dose and UV
UV Disinfection Guidance Manual F-31 . . • June 2003
Proposal Draft
-------
Appendix F. Background to the UV Reactor Validation Protocol
response. Dose and UV intensity values were predicted using Equations F.14 and F.I5 applied
to the annular reactor described in section F.4.2.
Figure F.16 Comparison of Dose Delivered by a MP Reactor
with New and Aged Type 214 Lamp Sleeves
75 80 85 90 * 95
UV Transmittance 254 nm (%)
Sensor: SiC,
Sensor-to-lamp
distance
100
1 ^
2
- £3
a>
eg 12-
>r § 1-2
i t/>
* »
€ o> i 1 .
•|< V1
o> S
' o *
Q i 1 0 -
OS
.2
QC n 9 -
"•*---,
— - _!'
Ihii
^^^m*^—^—
Trfp
"""-~ r^
- . » **
,,j
&-
70 75 80 85 90 95 100
, ' UV Transmittance 254 nm (%)
Sensor:
Filtered SiC
Sensor-to-lamp
distance
— — -r-2 crri
L-ujMMMu^iu-q rm
...,.I»»...MM«II ;j 1, | | |
- - - - 15 cm
20 cm
— ?— 25 cm
\;...
UV Disinfection Guidance Manual
Proposal Draft
F-32
June 2003
-------
Appendix F. Background to the UV Reactor Validation Protocol
Figure F.17 Comparison of Dose Delivered by a
MP Reactor with New and Aged Lamps
£
1.20
Sensor: SiC
Sensor-td-lamp
distance .
75,<;..-.:• ag..-,-: :,B5 ,.•; so . ' 95 100
UV Transmittance 254 nm (%)
;1 j" Sensor;,- -,,
Filtered SiC.
Y Senspr-to-lamp
:' •. distance
UV Disinfection Guidance Manual
Proposal Draft
F-33
June 2003
-------
Appendix F. Background to the UV Reactor Validation Protocol
Figure F.18 Comparison of Dose Delivered by a MP Reactor
with New Type 214 Lamp Sleeves to Fouled Sleeves
1.20
Sensor: SiC
Sensor-to-lamp
distance
75 80 85 90 95
UV Transmittance 254 nm (%)
100
1 in
n qc
o 5
0(1 ^ 0 90
" *• *• «• *,
">••"«••»"•• ..^
-jfrn
•
:?^7=
^^
g^
ssas
70 75 80 85 90 95 100
UV Transmittance 254 nm (%)
Sensor:
Filtered SiC
3ensor-to-lamp
distance
2 cm
-— ~~— 5 cm
— • —10cm
- - - - 1 5 cm
20 cm
25cm
In each figure (Figures F.I6 to F.18), the dose ratio increases with decreased water UVT
and increased sensor-to-lamp distance. The ratio is closer to one with germicidal sensors
compared with sensors with a SiC spectral response.
For a given UV reactor equipped with MP lamps, the impact of spectral shifts in lamp
output and sleeve UVT should be evaluated and used to establish correction factors. The
correction factor is calculated, for a given flowrate, sensor reading, and UVT, as the ratio of the
dose expected with and without the spectral shift expected with operation of the UV reactor at
the WTP. If the ratio is less than one, no correction factor is needed.
UV Disinfection Guidance Manual
Proposal Draft
F-34
June 2003
-------
Appendix P. Background to the UV Reactor Validation Protocol
Spectral shifts associated with lamp and sleeve aging can be avoided by regular
replacement of those components. Spectral shifts arising from fouling on external surfaces of
lamp sleeves and sensor windows can be minimized with good cleaning practices. However,
fouling can also occur on internal surfaces of lamp sleeves and sensor windows.
F.5 Uncertainty of Dose Monitoring and Safety Factors
UV installations should be sized and operated in a manner that accounts for the
measurement uncertainty associated with dose delivery monitoring. The objective of dose
delivery monitoring is to indicate the level of inactivation of the target pathogen. Safety factors
applied to UV installations that account for measurement uncertainty should be chosen to ensure
that UV reactors meet inactivation targets at a 90-percent confidence level. A 90 percent
confidence level is consistent with the confidence level used to define dose values for
Cryptosporidium, Giardia, and virus in Chapter 1 .
F.5.1 Analytical Foundation for Defining Uncertainty
This section derives a measurement equation for UV dose monitoring. This equation is
used in this guidance document as the analytical foundation for defining the uncertainty of dose
monitoring.
Consider a UV installation operating at a WTP. Assuming first order kinetics, the log
inactivation of a target pathogen achieved by the UV reactor at some point in time can be
expressed using Equation F. 16:
RED,
log Np = Equation F. 1 6
Dl0p
where
log Np = Log inactivation of the pathogen
REDp = RED of the pathogen (ml/cm5) .
DIOP = UV sensitivity of the pathogen (mJ/cm2 per log inactivation)
If the UV reactor delivers a dose distribution, the log inactivation
of the pathogen is related to the inactivation of a challenge microorganism using Equation F.17:
RPD ' •' '
log AT, = 5^,-—^- Equation F.17
where
REDC
BRED = Ratio of the RED of the pathogen to.that of the challenge microorganism
REDC = RED of the challenge microorganism (mJ/cm2)
Assuming the challenge microorganism RED is proportional to the measured UV
intensity, log inactivation of the pathogen can be expressed according to Equation F.I 8:
UV Disinfection Guidance Manual F-35 June 2003
Proposal Draft
-------
Appendix F. Background to the UV Reactor Validation Protocol
cd
A
Equation F.I8
I0p
where
I
a
UV intensity measured at the WTP (mW/cm2)
Constant relating challenge microorganism inactivation to measured intensity
(J/W)
The constant k is determined during validation as the ratio of the measured RED of the
challenge microorganism to the measured intensity. Assuming that inactivation is proportional
to flowrate, Equation F.I9 can be used:
D,
10p
L&
i,-Q
Equation F.I9
where
REDCV =
Iv
Qv =
Q
RED of the challenge microorganism measured during validation
UV intensity measured during validation
Flowrate measured during validation (mgd)
Flowrate measured at the WTP (mgd)
If spectral properties such as lamp output, sleeve UVT, and water UV absorbance during
validation differ from those during operation of the UV installation at the WTP, Equation F.I9 is
expressed as Equation F.21:
logtf =B*
I Qv
D 10
uVbf * v V4
Poly
Equation F.20
where
variables are defined as in Equation F. 19
The term Bp0iy is the ratio of challenge microorganism RED expected at the WTP to the
challenge microorganism RED expected during validation for the same conditions of flowrate,
water UVT, and UV intensity.
Assuming the dose-response of the challenge microorganism follows first order kinetics,
the challenge microorganism RED during validation is calculated using the log inactivation of
the challenge microorganism measured through the reactor as per Equation F.21:
Equation F.21
where
DIOC = UV sensitivity of the challenge microorganism (mJ/cm2 per log inactivation)
Njn = Challenge microorganism concentration measured at the reactor influent
Nef = Challenge microorganism concentration measured at the reactor effluent
UV Disinfection Guidance Manual F-36 June 2003
Proposal Draft
-------
Appendix F. Background to the UV Reactor Validation Protocol
The UV sensitivity of the challenge microorganism can be calculated according to
Equation F.22 from the UV dose-response measured using the collirnated beam apparatus:
£> = —2. Equation F.22
log*
where
DCB = Dose delivered by the collimated beam apparatus
log i = Log inactivation of the challenge microorganism observed with dose DCB
The dose delivered by the collimated beam apparatus is defined by Equation E.I (section
E.3). Substituting Equations F.21 and F.22, and E.I into Equation F.20 gives the measurement
equation for dose monitoring using the UV intensity setpoint approach:
R23
Dl0p
F.5.2 Calculating Total Uncertainty
Errors in dose monitoring can be classified as either biases or random uncertainties.
Biases are systematic errors that favor either an over or under estimation of dose delivery.
A bias error will occur with dose monitoring if the monitoring approach does not'account for
differences in the RED measured with the challenge microorganism and the RED delivered to
the target pathogen. A bias error will also occur if the monitoring approach does not account for
differences between the spectral properties of the UV reactor that impact dose delivery and
monitoring during validation and those properties during operation of the UV reactor at the
WTP. A bias error will occur if the radiometer, UV intensity sensor, flowmeter, or UVT monitor
used during validation always reads either high or low. Bias errors should be accounted for
using correction factors. The approaches for defining correction factors to account for bias
errors represented by the terms BRED and Bp0iy in the measurement equation are provided in
Sections F.I and F.4, respectively.
Random uncertainty is associated with every term in the measurement equation
(Equation F.23). If the measurement equation consists of linear relationships of independent
variables whose random uncertainty is normally distributed, standard approaches can be used to
calculate the uncertainty of the measured variable from the uncertainty of each term in the
measurement equation. For example, if the measurement equation is y = xj. -I- \2 or y = Xi - x2,
the uncertainty of y due to the uncertainty of xi and xa is calculated using Equation F.24:
UV Disinfection Guidance Manual F-37 June 2003
Proposal Draft
-------
s =
where
s
Appendix F. Background to the UV Reactor Validation Protocol
Equation F.24
Uncertainty of y in absolute units
Uncertainty of xi in absolute units
Uncertainty of X2 in absolute units
On the other hand, if the measurement equation is y = X] x x2 or y = xi / xa, the
uncertainty of y due to the uncertainty of xi and x2 is calculated using Equation F.25:
s =
where
s
Sl
82
Equation F.25
Uncertainty of y in percent
Uncertainty of xi in percent
Uncertainty of xa in percent
If the measurement equation involves non-linear relations like y= xi expfe), Monte Carlo
approaches should be used to define the uncertainty of y.
Determining the random uncertainty of a measured quantity requires making assumptions
about the statistical distribution of measurements. If the distribution is normal, the uncertainty is
calculated as the product of the sample standard deviation and the t-statistic. If the number of
samples is high, the t-statistic can be approximated by the z-statistic. If the standard deviation of
the population is known, the uncertainty is calculated as the product of the population standard
deviation and the z-statistic. T and z-statistics are often given in the appendices of statistics
texts. The NIST provides recommendations for specifying the uncertainty for quantities that are
not normally distributed.
Table F.5 defines an approach for estimating the uncertainties of each term in the
measurement Equation F.23. The total random uncertainty of dose monitoring can be estimated
by summing the uncertainties associated with each term in Equation F.23 using the above stated
rules. Assuming the terms BRED and Bp0iy are the only bias terms, a safety factor for dose
monitoring can be defined according to Equation F.26:
, x (l + e) Equation F.26
where
e
Total random uncertainty associated with the measurement equation.
UV Disinfection Guidance Manual
Proposal Draft
F-38
June 2003
-------
Appendix F. Background to the UV Reactor Validation Protocol
Table F.5 Terms Used to Define the Uncertainty of Dose Monitoring
Term
BRED and Babs
i
1 and lv
Q and Q¥
DioP
Log(Nin/Na()
DCB
Log(t)
Assumption
No term used if values are selected as safety factors as described in Sections F.1
and F.4. If terms are calculated, use uncertainty of model predictions to define
uncertainty of these terms.
UV intensity measurement uncertainty is often defined by the UV intensity sensor
manufacturer. If a reference sensor is used to check the uncertainty of a duty
sensor, the uncertainty of the duty sensor should be defined as the rejection criteria
used to determine if the on-line sensor is out of tolerance. See Equation F.1 1 .
Use measurement uncertainty defined by flowmeter manufacturer
Accounted for in dose targets provided in Chapter 1
Calculated as a confidence interval using standard deviation and Student's t-statistic
associated with samples collected during validation. See Equation C.7
Calculated as a confidence interval using the measurement uncertainties of the
terms in Equation C.2. See Appendix E and Equation C.8.
Use confidence interval of challenge dose-response. See sections C.4.9.7 and
C.4.9.8
The safety factor defines the relationship between the dose targets provided in Chapter 1
and the RED that should be delivered by the UV reactor at the WTP.
F.6 Re-validation
If the design of a validated UV reactor changes, the UV reactor should be re-validated if
the design change significantly impacts dose delivery or monitoring. Dose delivery and sensor
modeling can be used to assess the impact of the design change and justify the need, or lack of
need, for re-validation. This section discusses UV reactor modifications and provides guidance
on the need for re-validation.
F.6.1 Lamp Assembly
• Design changes to the lamp assembly include changes made by the lamp manufacturer to
the lamp, selection of a new lamp type by the UV manufacturer, and changes made by the UV
manufacturer to the components associated with the lamp assembly. The relationship between
dose delivery and monitoring may be impacted by any design change involving modifications to
the following components:
• Lamp arc length
• Any reflectors, connectors, and spacers used at the lamp ends
•. Lamp envelope diameter
• Lamp envelope UVT from 185 nm to 400 nm
UV Disinfection Guidance Manual
Proposal Draft
F-39
June 2003
-------
Appendix P. Background to the UV Reactor Validation Protocol
• Mercury content of the lamp
• Argon content of the lamp
The lamp's arc length and the use of components at the ends of the lamps (like reflectors,
spacers, and connectors) impact the UV intensity field in the region near the lamp ends. Design
changes to these components could impact dose delivery, especially if the lamps are oriented
perpendicular to flowrate. Design changes could also impact UV intensity sensor measurements
if the lamp ends are within the viewing angle of the sensors. Dose delivery and UV intensity
sensor modeling can be used to assess the impacts on changing the lamp arc length or
components used at the lamp ends. If the impacts are considered significant, the reactor should
be re-validated.
With LP lamps, the UV-emitting plasma occupies the space within the lamp envelope.
With MP lamps, the plasma forms a narrow arc that occupies a portion of the space within the
lamp envelope. In the presence of electromagnetic fields, the plasma within a MP lamp can be
displaced off center within the lamp. The diameter of a plasma centered within the lamp
envelope should have a small impact on the UV intensity field and dose delivery (Bolton 2000).
However, displacement of the plasma off-center within the envelope could impact the intensity
field and dose delivery. The reactor should be re-validated if design changes to the lamp
diameter significantly impact the intensity field.
The UVT of the lamp envelope will impact the UV output of both LP and MP lamps.
. With LP lamps, envelope material can be selected to allow or prevent LP lamps from emitting
UV light at 185 nm. While UV light at 185 nm has a negligible impact on dose delivery and UV
intensity sensor measurements because of the high UV absorbance of water at this wavelength,
185'nm light may promote the formation of ozone within the lamp sleeve. Ozone will absorb
UV light at 254 nm and lower the output from the lamp. Ozone could degrade components
within the lamp assembly leading to internal'sleeve fouling. Typically, LP lamps are selected
with envelopes that prevent output at 185 nm.
With MP lamps, the envelope material has a significant impact on the intensity of UV
light emitted below 260 nm. Lamp envelope material can be selected to eliminate or maximize
UV output at lower wavelengths. Since envelope transmittance decreases with increased
temperature, the UVT of the envelope of a MP lamp should be assessed at the operating
temperature of the lamp. Dose delivery and UV intensity sensor modeling can be used to assess
the impacts of changing lamp material and justify the need for re-validation.
LP lamps typically operate near 40°C with a relatively low mercury vapor pressure that
promotes UV output at 254 nm. Because the amount of mercury added to the lamp is well in
excess of the amount that enters the vapor state during lamp operation, the UV output of a LP
lamp is independent of the mercury dose added to the lamp during lamp manufacture. On the
other hand, MP lamps operate at a high temperature, near 600°C, with all of the added mercury
in the vapor phase. As such, the mercury vapor pressure is dependent on the mercury dose and
the lamp operating temperature. The vapor pressure influences the fraction of mercury that is
ionized or excited to higher energy states, and hence the spectral output of the MP lamp. Table
F.6 presents the calculated impact of mercury dose on the germicidal output and measured
intensity from a MP lamp operating with an electrical input of 70 W/cm. The results suggest that
UV Disinfection Guidance Manual F-40 June 2003
Proposal Draft
-------
Appendix F. Background to the UV Reactor Validation Protocol
a change in mercury dose has no impact on the relationship between dose delivery and
monitoring with germicidal sensors and a small impact on the relationship with SiC sensors.
Table F.6 Impact of the Mercury Dose on the Relationship
Between Germicidal Output and Measured Output of a MP Lamp1
Mercury Dose
(mg/cm)
4.8
8
10.1
UV Output (W/cm) Weighted by
MS2 Action
6.88
6.53
7.10
SiC Sensor
11.4
10.3
10.8
Filtered SiC Sensor
6.82
6.46
7.06
Ratios
SiC:MS2
1.65
1.58
1.52
Filtered SiC:MS2
0.990
0.989
0.993
Adapted from lamp output data from 248 to 400 nm provided by Phillips (1983).
F.6.2 Ballasts
Modifications to lamp ballasts include changing the operating voltage, current,
frequency, and waveform. With LP lamps, modifications will impact the amount of UV
generated by the lamp, but will not impact the relationship between dose delivery and UV
intensity measurements. With MP lamps and some LPHO lamps, changes in lamp operating
temperature and mercury pressure caused by changes in ballast power will impact the spectral
distribution of emitted light. Table F.7 presents the impact of changing the input power from
48 to 92 W/cm on the germicidal output and measured intensity from a MP lamp dosed
with 4.8 mg/cm of mercury. The results suggest a change in lamp operating power has no
impact on the relationship between dose delivery and monitoring with germicidal sensors and a
small impact with SiC sensors.
Table F.7 Impact of Operating Power on the Relationship Between
Germicidal Output and Measured Output of a MP Lamp1
Lamp Input
Power (W/cm)
48
70
92
UV Output (W/cm) Weighted by
MS2 Action
4.13
6.86
9.29
SiC Sensor
7.01
11.3
15.2
Filtered SiC Sensor
4.08
6.78
9.14
Ratios
SIC:MS2
1.70
1.66
1.65
Filtered SiC:MS2
0.99
0.99
0.98
Adapted from lamp output data from 248 to 400 nm provided by Phillips (1983) for a MP lamp dosed with 4.8
mg/cm Hg.
F.6.3 Lamp Sleeves
Design changes to the lamp sleeves include changing the sleeve diameter, thickness, and
material. Changing the sleeve diameter may impact the hydraulics through the reactor, the
measurement of UV intensity, and the optimal placement of UV intensity sensors relative to the
lamp. Changing the thickness and material of the lamp sleeve will impact the spectral UVT,
thereby impacting both dose delivery and UV intensity measurements.
UV Disinfection Guidance Manual F-41 June 2003
Proposal Draft
-------
Appendix F. Background to the UV Reactor Validation Protocol
Dose delivery and UV intensity sensor modeling may be used to assess the impact of
lamp sleeve design changes. Figure F.19 provides the ratio of dose delivered with a standard
sleeve to dose delivered with an "ozone-free" sleeve for a given sensor reading as a function of
water UVT, sensor-to-lamp distance, and sensor spectral response. Dose and UV intensity
values were predicted using Equations F.14 and F.I5 applied to the annular reactor described in
section F.4.2. Sleeve UVT is provided in Figure F.20. The results show that a design change
from a regular sleeve to an ozone-free sleeve described in Figure F.20 would have a small impact
on the relationship between dose delivery and UV intensity sensor readings with a SiC sensor
and a negligible impact with a germicidal sensor. Modeling can also be used to show that the
dose delivery at a given lamp output, water UVT, and flowrate would be approximately 10
percent greater with the standard sleeve than with the ozone-free sleeve. If models indicate the
sleeve design change causes a significant impact on dose delivery and monitoring, the UV
reactor should be re-validated.
Figure F.19 Ratios of Dose Delivered with Standard Sleeve to
Dose Delivered with "Ozone-Free" Sleeves by an Annular Reactor
1 m * -
9 ' •
1 I '1.05-
ss • :
V) tit
fu *i nn
U u-
|fi
-w N n Q^
B p u.s»
£m '•
~ I - :
flQfl -
- -
- — . »
• • -*?»•:
-
**-
i
70 75 80 85 90 95 . 100
UV Trananttance 254 nm (K*
Sensor SiC
Sensor-to-lamp
distance
— 2 cm
„....; 5 cm
— - -10cm
- - ' *15crfi
— — 20cm
-25cm
. "O ' •
" * B> '
r* S 1 n^ -
" fli '
- «BJ
f«
•'B 100;
0) LL.
II) 0)
Q 1 n Q<;
-t? : '
£ 'H QO •
• ' •
L'
^^i^«^^^
=-. te
..
=T,— *
a-.^='g!
F^F-a
-•'-'Hannnn m
T^
70- 75 '80 85 - 90 95 100
, UVTransrnttance254nm(%) "
Sensor:
Filtered SiC
Sensor-to-lamp
d stance
— •-•*— 5 cm
— • —10 cm
- • • • 15 cm
20 cm
.,25 -cm
UV Disinfection Guidance Manual
Proposal Draft
F-42
June 2003
-------
Appendix F. Background to the UV Reactor Validation Protocol
Figure F.20 UVT of Standard and "Ozone-Free" Quartz Assuming Air-Quartz and
Quartz-Water Interfaces
100 1 Standard Quartz
200 , 220. 240 260 280
Wavelength (nm)
300
320
F.6.4 Reactor and Component Dimensions
Modifications to the wetted dimensions and positioning of the components within the
reactor will impact the reactor hydraulics and dose delivery. Modifications could also impact the
intensity field within the reactor and the measurement of UV intensity. Modifications include
changes to the dimensions of the reactor, inlet piping, exit piping, baffles, lamp sleeves, wipers,
and UV intensity sensors. The impact of such modifications on dose delivery and UV intensity
measurements can be insignificant or large. Addition of a baffle plate will likely have a large
impact on dose delivery and a small impact on measured UV intensity, while changing the
position of a UV intensity sensor will likely have a small impact on dose and a large impact on
measured UV intensity. Dose delivery and UV intensity modeling maybe used to assess the
impacts of these modifications. If the impacts are significant, the reactor should be re-validated.
F.6.5 UV Intensity Sensors
Modifications to the UV intensity sensors include changes made by the sensor
manufacturer to the sensor, changes by the UV manufacturer to the sensor housing and
associated optical components, and changes by the UV manufacturer to the number and
positioning of the sensors within the reactor.
Changes to the semi-conductor and optical components within the UV intensity sensor
could impact the sensor's spectral response, linearity, angular response, and temperature
stability. Changes to those properties could impact the sensor's measurement uncertainty. If the
new measurement uncertainty is quantified, it should be used to define a new safety factor for the
UV reactor. If the angular response or spectral response of the sensor changes, measurements
supported by calculations should be used to evaluate the impact of the change on dose delivery
monitoring.
. UV Disinfection Guidance Manual
Proposal Draft
F-43
June 2003
-------
Appendix.F. Background to the UV Reactor Validation Protocol
Changes to the measuring window of the UV intensity sensor include dimensional and
material changes. Changes may impact the UVT of the window and the detection angle.
Measurements supported by calculations should be used to evaluate the impact of the change on
dose delivery monitoring.
Modifications to the positioning of the UV intensity sensor within the reactor could
disturb the flowrate and impact dose delivery. If the impact on dose delivery is negligible,
measurements supported by calculations may be used to compare measured UV intensity at the
two positions and modify the dose monitoring approach without the need for re-validation.
Addition of UV intensity sensors to the reactor could disturb the flowrate through the UV
reactor and impact dose delivery. If sensors are added, they should be positioned relative to the
lamps in a similar manner as the other sensors. For example, if one sensor is positioned to view
two lamps through a 5-cm water layer, then all added sensor should view two lamps through a 5
cm water layer.
F.7 References
Bolton, J.R. 2000. Calculation of ultraviolet fluence rate distributions in an annular reactor:
significance of refraction and reflection. Water Research 34:3315-3324. v
Cabaj, A., R. Sommer, and D. Schoenen. 1996. Biodosimetry: model calculations for U.V. water
disinfection devices with regard to dose distributions. Water Research. 30, no.4:1003-
1009.
Cabaj, A., R. Sommer, W. Pribil, and T. Haider. 2002. The spectral UV sensitivity of
microorgnisms used in biodosimetry. Water Science and Technology: Water Supply,
2(3): 175-181.
Chang, J.C.H., S.F. Osoff, B.C. Lobe, M.H. Dorfman, C.M. Dumais, R.G. Quails, and J.D.
Johnson. 1985. UV inactivation of pathogenic and indicator microorganisms. Applied
and Environmental Microbiology 49(6): 1361-1365.
Chiu, K.-P., D.A. Lyn, P. Savoye, and E.R. Blatchley.1999. Effect of UV system modifications
on disinfection performance. Journal of Environmental Engineering 125(5):459-469.
Haas, C.N. and G.P. Sakellaropoulos.1979. Rational analysis of ultraviolet disinfection.
National Conference on Environmental Engineering, Proc. ASCE Specialty Conf., San
Francisco, CA, July 9-11.
Kawar, K., J. Jenkins, B. Srikanth, and A. Shurtleff. 1998. Ultraviolet light deterioration of the
light transmittance of quartz sleeves due to continuous exposure to UV radiation.
Ultrapure Water October, 67-71.
Larason, T.C., S.S. Bruce, and A.C. Parr. 1998. Spectroradiometric Detector Measurements.
NIST Special Publication 250-41. U.S. Government Printing Office, Washington.
UV Disinfection Guidance Manual
Proposal Draft
F-44
June 2003
-------
Appendix F. Background to the UV Reactor Validation Protocol
Larason T.C. and C. L. Cromer. 2001. Source of error in UV radiation measurements. Journal of
Research of the National Institute of Standards and Technology 106, no.4:649-656.
Lin, L., C.T. Johnston, E.R. Blatchley.1999. Inorganic fouling at quartz: water interfaces in
ultraviolet photoreactors I: chemical characterization. Water Research 33:3321-3329. :
Linden K.G., G. Shin, and M.D. Sobsey. 2001. Comparative effectiveness of UV wavelengths
for the inactivation of Cryptosporidium parvum oocysts in water. Water Science &
Technology 43(12): 171-174.
Meng, Q.S. and C.P. Gerba.1996. Comparative inactivation of enteric adenovirus, poliovirus and
coliphages by ultraviolet irradiation. Wat. Res. 30, no. 11:2665-2668.
Phillips, R.I 983. Sources and Applications of Ultraviolet Radiation. New York: Academic Press,
Quails, R.G. and J.D. Johnson. 1983. Bioassay and dose measurement in UV disinfection.
Applied and Environmental Microbiology 45, no.3:872-877.
Rauth, A.M. 1965. The physical state of viral nucleic acid and the sensitivity of viruses to
ultraviolet light. Biophysical Journal 5:257-273.
Sommer, R., T. Haider, A. Cabaj, W. Pribil, and M. Lhotsky. 1998. Time dose reciprocity in UV
disinfection of water. IAWQ Vancouver 1998 Poster.
Wilson, B.R., P.F. Roessler, E. Van Dellen, M. Abbaszadegan, and C.P. Gerba. 1992 Coliphage
MS-2 as a UV water disinfection efficacy test surrogate for bacterial and viral pathogens.
Proceedings of the Water Quality Technology Conference, Nov. 15-19,1992, Toronto,
219-235.
Wright, H.B. and Y.A. Lawryshyn. 2000. An assessment of the bioassay concept for UV reactor
validation. Disinfection 2000: Disinfection of Wastes in the New Millenium, New
Orleans, Louisiana, March 15-18,2000; Water Environment Federation, Alexandria,
Virginia
Wright, H.B., E. Mackey, and P. White. 2002. UV Disinfection Compliance Monitoring for
Drinking Water Applications. Proceedings of the Water Quality technology Conference
and Exhibition, Seattle, November 10 -14, 2002.
UV Disinfection Guidance Manual F-45 June 2003
Proposal Draft
-------
Appendix G. Issues for Unfiltered Systems
Unfiltered systems are utilities that use surface water sources and meet the filtration
avoidance criteria of the Surface Water Treatment Rule (SWTR) (40 CFR 141.71). The Long
Term 2 Enhanced Surface Water Treatment Rule (LT2ESWTR) requires unfiltered systems to
meet overall disinfection requirements (i.e., Cryptosporidium, Giardia, and virus inactivation)
using a minimum of two disinfectants (40 CFR 141.721(d)). The information presented in this
manual is focused on post-filtration applications of UV disinfection; however, the information is
also relevant to UV disinfection of unfiltered supplies. In addition, the UV dose requirements
presented in section 1.3.1.3 are applicable to both filtered water and water supplies that meet the
regulatory requirements for filtration avoidance (40 CFR 141.729(d)). This appendix identifies
issues that are specific to unfiltered applications of UV disinfection. The following issues are of
particular interest to unfiltered supplies because they make applying UV disinfection different
from post-filter locations:
* Water quality (especially particle content)
» Debris
« Ozone residual (when ozone is applied prior to UV disinfection)
• Off-specification requirements recommendations
G.1 Water Quality
Differences in the quantity and nature of particles in unfiltered surface water supplies are
the most pertinent distinction between post-filtration and unfiltered supply water qualities.
Typically, the turbidity in unfiltered surface waters is less than 1 nephelometric turbidity units
(NTU). However, the SWTR allows turbidity up to 5 NTU immediately prior to the first point of
disinfection application (40 CFR 141,71). Several studies have examined the effects of turbidity
up to 10 NTU on UV disinfection, including changes in UV absorbance measurements made
with a spectrophotometer and inactivation of microorganisms.
Particles in water absorb and scatter UV light to varying degrees based on size and
composition. Particles impact the disinfection process in two distinct manners:
1. Particles can decrease the UV transmittance (UVT) of water and thereby impact UV
dose delivery (section A.4.1.2).
2. Particle association can shield microorganisms from UV light, thereby changing the
characteristics of the UV dose-response curve (section A.2.6.5).
Christensen and Linden (2001) concluded that the light scattering and changes in
absorbance caused by turbidity up to 10 NTU can be accounted for when calculating UV dose in
collimated beam testing provided that the ultraviolet absorbance at 254 nanometers (A254) of the
sample is measured according to a modified version of Standard Method 591 OB (i.e., without
UV Disinfection Guidance Manual G-l June 2003
Proposal Draft
-------
Appendix G. Issues for Unfiltered Systems
0.45 nm filtration). Direct reading spectrophotometers, the most common type of
spectrophotometer, may overestimate the Aa54 of water with turbidity greater than 3 NTU,
resulting in an overly conservative UV dose calculation (Christensen and Linden 2002). To
reduce this overestimation, an'integrating sphere can be installed in a direct-reading
spectrophotometer that will provide accurate ^354 measurements. Regardless of the type of
spectrophotometer used, the effects of increased absorbance due to particles can be accounted for
in the A^ measurement, which can then be used to determine the design UVT. If an appropriate
design UVT is used, the UV reactor will be able to respond to changes in UVT that arise due to
particles.
Particles and microorganisms in a water sample are either dispersed or aggregated
together. Studies have demonstrated that dispersed coliform bacteria in wastewater are easier to
disinfect than aggregated bacteria (Parker and Darby 1995). To date, research examining the
effects of particles in drinking water on UV disinfection has been performed with seeded
organisms and particles. It is unknown at this time how well these studies represent naturally
occurring microorganism and particle interactions. However, since the concentration of
microorganisms in unfiltered sources is typically below detectable limits, methods to examine
this phenomenon directly (without seeding) do not currently exist. Consequently, seeded
drinking water studies can only suggest the impact of turbidity on dose-response as it relates to
the impact of UV light scattering by particles rather than particle-association or clumping of
microorganisms.
Recent research has shown that particles present in supplies meeting regulatory
requirements for unfiltered'drinking water do not impact the UV inactivation of seeded
microorganisms. Passantino and Malley (2001) reported that for unfiltered surface waters,
turbidity up to 7 NTU does not affect the inactivation of seeded male specific-2 bacteriophage
(MS2) in bench-scale, batch, collimated beam testing. In this study, turbidity was increased by
adding natural sediment to waters collected from unfiltered.water supplies. Therefore, naturally
occurring interactions between particles and microorganisms could not be evaluated. In another
study, batch (bench-scale) and continuous-flow (pilot-scale) studies showed that turbidity
ranging from 0.65 to 7 NTU does not affect the UV dose necessary per log inactivation of seeded
MS2, Giardia muris, or Cryptosporidium parvum in unfiltered waters (Oppenheimer et al. 2002),
Womba et al. (2002) evaluated the impact of turbidity on UV inactivation of MS2 at the bench-
and pilot-scale. They found that on the bench-scale, when the impact of turbidity was accounted
for in the UV dose determination, the inactivation of MS2 was not affected by turbidity.
However, in this study on the pilot-scale, because the lamp intensity and flowrate (and therefore
residence time in the reactor) remained constant, the effects of turbidity were not accounted for
in the reactor control strategy. Therefore, the reduction equivalent dose (RED) observed
decreased as turbidity increased.
Unfiltered supplies are also susceptible to algal blooms. Womba et al. (2002) monitored
algae levels in an unfiltered supply reservoir for over one year and found that algal counts were
typically below 30,000 cells/niL; however one algae event had a higher level of nearly 300,000
cells/mL. Although not regulated, the presence of algae may interfere with the UV disinfection
process. Womba et al. (2002) and Passantino and Malley (2001) examined the effects of algae
on UV disinfection of MS2 at the bench-scale in batch, collimated beam testing. Both studies
found that up to algal counts up to 70,000 cells/mL and 42,000 cells/mL, respectively, do not
affect the inactivation of MS2.
UV Disinfection Guidance Manual G-2 June 2003
Proposal Draft
-------
Appendix G. Issues for Unfiltered Systems
G.2 Debris
Relative to post-filter applications of UV disinfection, there may be greater opportunity
for debris to be present in the influent to UV reactors in unfiltered applications. Debris entering
the UV reactor with sufficient momentum could cause lamp sleeve and lamp breakage. The
mass and size of an object that might cause damage is installation-specific and depends on UV
reactor configuration (e.g., horizontal versus vertical reactor orientation) and water velocity
through the reactor. As such, designs should incorporate features that prevent potentially
damaging objects from entering the system; the optimal approach is site-specific. Such features
could include screens, baffles, or low velocity collection areas. Another option is to install the
UV reactors vertically with the inlet closest to the ground, following a low velocity zone. This
arrangement will decrease the momentum of larger debris and reduce the risk of lamp breakage.
The effects of lamp breakage and methods of minimizing it are discussed in Appendix N.
G.3 Ozone Impacts on Absorbance
Some utilities using an unfiltered source may consider applying ozone in addition to UV
disinfection. There are a number of benefits associated with this process combination, including
addressing multi-barrier disinfection requirements. Additionally, if ozone is added prior to UV
disinfection, the A^* of the water can be reduced as shown in Figure G.I, thereby improving the
efficiency of UV disinfection.
Figure G.1 Impact Of Pre-Ozonation On A2M (Malley 2002).
0.5
0.3
•6
8 0.2
0.1
0.0
Preozonated UV Influent
(no detectable ozone residual)
r— UV Influent - No ozone
200 220 240 .260 280 300 320
W avelength (nm)
It should be noted, however, that ozone is a strong UV absorber with a molar absorbance
value of 0.0677 L/mg/cm at 254 nm. Figure G.2 illustrates the impact of ozone concentration on
UVT1 for three baseline transmittance values. If ozone is applied prior to UV reactors and
residual ozone enters the UV reactor, the increase in UV absorbance due to ozone residual can be
UV Disinfection Guidance Manual
Proposal Draft
G-3
June 2003
-------
Appendix G. Issues for Unfiltered Systems
significant and should be considered when determining the design UVT. To address this issue,
utilities can monitor the ozone residual and add an ozone-reducing chemical to maintain the
ozone residual below a specified setpoint value (e.g., 0.1 mg/L). There are several chemicals
that can be used to quench ozone; however, some chemicals (such as sodium thiosulfate) have a
high molar absorbance value (as shown in Table A. 5, section A.4.1.3), and thus have the
potential to decrease the UVT. Such chemicals should not be used prior to UV disinfection.
Sulfite has a lower molar absorbance value and is therefore an acceptable chemical to quench
ozone residual. The impact of water treatment chemicals on UV absorbance can be assessed by
preparing solutions of various concentration and measuring their UV absorbance using a
standard spectrophotometer (Bolton et al. 2001).
Figure G.2. Impact of Ozone Residual on UVT (adapted from Gushing et al. 2001)
100%
D>
f
E
S
90%
80%
70%
60%
0.00
0.25 O.SO 0.75
Aqueous Ozone Concentration (mg/L)
1.00
In at least some cases, the increase in UVT resulting from ozone addition will improve
overall UV disinfection effectiveness provided that any remaining ozone residual is adequately
controlled. Each utility should explore the sequence of disinfectants that best fits their site-
specific objectives and constraints.
G.4 Off-specification Requirements
Off-specification is when the UV reactor is operating outside of its validated limits. UV
installations should be designed with process monitoring and control components (e.g., alarms,
shut-off valves) to prevent water from entering the distribution system when a UV reactor is
operating outside of validated conditions. Unfiltered systems that use UV disinfection to meet
the Cryptosporidiwn treatment requirement of the LT2ESWTR must demonstrate that at least 95
UV Disinfection Guidance Manual G-4 ' June 2003
Proposal Draft
-------
4
Appendix G. Issues for Unfiltered Systems
percent of the water delivered to the public during each month is treated by UV reactors
operating within validated limits (i.e., operating conditions that have been validated to achieve
the necessary log inactivation) (40 CFR 141.721(c)). Failure to demonstrate this will result in a
treatment technique violation.
The UV reactors are off-specification when any of the following conditions occur (40
CFR 141.729(d)):
. The flow, UV intensity, or lamp status is outside of the validated range.
• The UVT or UV intensity is outside of the validated range (if the UV intensity and
UVT setpoint approach is used (section 3.1.5))
• The calculated dose is outside of the validated range at a given flow (if the calculated
dose approach is used (section 3.1.5))
« All UV lamps in all UV reactors are off because of a power interruption or power
quality problem (as discussed in section 3.1.3.3), and water is flowing through the
reactors.
More information on off-specification is in section 3.1.3, and compliance information is
in section 5.4.1.
G.5 References
Bolton, J.R., M.I. Stefan, R.S. Gushing, and E. Mackey. 2001. Importance of water
absorbance/transmittance on the efficiency of ultraviolet disinfection reactors.
Proceedings of the IUVA 1st International Congress, June 14-16, Washington, D.C.
Christensen, J. and K. Linden. 2001. Ultraviolet disinfection of unfiltered drinking water:
particle impacts. Proceedings of the IUVA 1st International Congress, June 14-16,
Washington, D.C.
Christensen, J. and K. Linden. 2002. New findings regarding the impacts of suspended particles
on UV disinfection of drinking water. Proceedings of the AWWA Annual Conference,
June 16-20, New Orleans, L.A.
Gushing, R.S., E.D. Mackey, J.R. Bolton, and M.I. Stefan. 2001. Impact of common water
treatment chemicals on UV disinfection. Proceedings of the AWWA Annual Conference
and Exposition, June 17-21, Washington DC.
Malley, J.P. 2002. Historical.perspective of UV use. Presented at the AWWA Water Quality
Technology Conference, November 10-14, Seattle, W.A.
UV Disinfection Guidance Manual G-5 June 2003
Proposal Draft
-------
Appendix G. Issues for Unfiltered Systems
Oppenheimer J,, T. Gillogly,,G. Stolarik, and R. Ward. 2002. Comparing the efficiency of low
and medium pressure UV light for inactivating Giardia muris and Cryptosporidium
• parvum in waters with low and high levels of turbidity. Proceedings of the AWWA
Annual Conference, June 16-20, New Orleans, L.A.
Parker, J.A. and J.L. Darby. 1995. Particle-associated coliform in secondary effluents: shielding
from ultraviolet light disinfection. Water Environment Research 67:1065-1075.
Passantino, L. and J.P. Malley. 2001. Impacts of turbidity and algal content of unfiltered
drinking water supplies on the ultraviolet disinfection process. Proceedings of the
AWWA Annual Conference and Exposition, June 17-21, Washington D.C.
Womba P., W. Bellamy, J. Malley, and C. Douglas. 2002. UV disinfection and disinfection
byproduct characteristics of an unfiltered \vater supply, Project 2747. Denver, C.O.:
AwwaRF periodic report.
UV Disinfection Guidance Manual
Proposal Draft
G-6
June 2003
-------
Appendix H. Issues for Ground Water Systems
The UV installation design, operation, and maintenance principles presented in Chapters
3 and 5 of this manual are focused on the use of UV reactors to disinfect filtered surface water.
Most of the information presented in those chapters is also applicable to ground water systems.
Additional ground water-specific regulatory requirements and recommendations, site issues,
hydraulic issues, and water quality issues that affect design and operation are discussed in this
appendix.
H.1 Ground Water Systems Background
Regulations should be reviewed to determine the goals and requirements for disinfection.
Existing treatment processes and distribution system parameters should also be analyzed before
selecting a strategy for integrating UV disinfection into the system.
H.1.1 Regulatory Background
Currently, federal regulations do not require ground water systems to provide primary or
secondary disinfection unless the water is a ground water source under the direct influence of
surface water (GWUDI). However, some States require ground water systems to maintain a
residual disinfectant in the distribution system. In addition, ground water systems are required to
meet the requirements of the Total Coliform Rule (TCR) (54 FR 27544) and the Stage 1
Disinfection and Disinfection Byproducts Rule (DBPR) (63 FR 69390) and are expected to be
affected by the upcoming Stage 2 DBPR.
The upcoming Ground Water Rule, as proposed, would require some ground water
systems to provide 4-log removal or inactivation of viruses. Systems with significant
deficiencies, as determined by States during sanitary surveys, and systems that detect fecal
indicators in their source water will be affected. These systems will be required to correct any
deficiencies, provide water from an alternative source, or install treatment that provides 4-log
removal or inactivation of viruses.
Ground water supplies that are designated as GWUDI, as defined in the Surface Water
Treatment Rule (SWTR), 40 CFR Part 141.2, are classified as Subpart H Systems and must meet
the same regulatory requirements as surface water systems. GWUDI systems often use many of
the same treatment strategies as surface water systems, including filtration. The issues involved
with implementing UV disinfection at filtered water utilities (including GWUDI) are discussed
in detail in Chapters 1 through 5 of this manual. GWUDI systems are subject to the Stage 1
DBPR and would be subject to the upcoming Stage 2 DBPR and the Long Term 2 Enhanced
Surface Water Treatment Rule (LT2ESWTR). Both of these regulations are summarized in
section 1.3.
UV Disinfection Guidance Manual
Proposal Draft
H-l
June 2003
-------
Appendix H. Issues for Ground Water Systems
H.1.2 Typical Ground Water System Design
Most ground water systems operate by cycling ground water pumps on and off in
response to demand or storage capacity. Because significant treatment is not usually necessary
beyond secondary disinfection, ground water systems typically do not have a single, centralized
treatment system. Many ground water systems pump to storage (e.g., hydropneumatic tank,
elevated storage tank), but some may discharge directly to the distribution system. A production
well typically consists of a well pump, and may contain a chlorinator (for secondary disinfection)
and corrosion control equipment (for Lead and Copper Rule compliance), air release valves,
vacuum relief valves, and other ancillary equipment necessary for well operation (Figure H.I).
Figure H.1. Typical Ground Water Well Site Layout
1 ; TEE WITH* ••-•
• "i BLIND FLANGE;
. i V '•'••:•".= .'
VACUUM RELIEF VALVE, -i ,
"'''^^CHECK^VA^
.METER:":'.'.
AIR/VACUUM,- , '•'?'.
VALVE'-"-'; ' ;.f. ""
-h- WELL'PUMP;''.,- "• •'•••• ' •' '
- ' '' :•'"•• •'"-" "" -":'•- '
;•>•:•;-' -v. r-\ ::'V- -.;; ;WELL PUMP 'FOUNDATION/
-^- HYDROPNEUMATIC-TANK^:/ . K'. :>..•/ ';; '•.-.;!' ^-PRESSURE",.;'-- :i''....-, : -•.>."•.!
UV reactors may be installed at each well in a production system. If multiple wells are
located in the same area, centralizing the flow through a common header minimizes the number
of UV reactors needed and possibly reduces the project cost. In addition, treatment for other
aesthetic issues (e.g., removal of iron and manganese or stripping of sulfides) may be more
effectively accomplished with centralized treatment. An engineering cost analysis should be
conducted to compare centralized treatment with the installation of individual reactors at each
well.
H.2 Water Quality Issues
Although ground water typically exhibits small variations in water quality, specific
parameters need to be analyzed when planning for a ground water system. As with surface water
systems, UV absorbance at 254 nm (fasi) and the corresponding UV transmittance (UVT) is the
most important parameter when designing a UV installation because it affects the UV reactor
size. In addition, many naturally occurring constituents present in ground water (e.g., calcium,
iron, manganese, aluminum, chloride, carbonate, sulfide, and phosphate) are capable of fouling
the lamp sleeves in UV reactors, and these constituents should be monitored. The potential for
UV'Disinfection Guidance Manual
Proposal Draft
H-2
June 2003
-------
Appendix H. Issues for Ground Water Systems
fouling is greater with medium pressure (MP) reactors than low pressure (LP) and low pressure
high output (LPHO) reactors because MP lamps operate at higher temperatures (section 2.4.2).
Mechanical wipers are often effective at removing fouling on the lamp sleeves. In situations
where the ground water has detectable levels of iron and manganese, chlorination prior to the UV
reactors may cause increased fouling or staining, necessitating chemical cleaning (Malley et al.
2001). A complete discussion of the relevant water quality parameters and the determination of
their design values is presented in section 3.1.3.1.
With ground water systems, it is common for one or more wells to be taken out of service
for an extended period due to fluctuations in water demand, ground water quality, operational
problems, or other planned and unplanned circumstances. Toivanen (2000) reported that the
lamp sleeves and internal surfaces of the UV reactors became fouled when the UV reactors were
out-of-service and full of water. The amount of time it takes to foul the UV reactor while off-line
is site-specific and depends on the water quality. At a minimum, it is recommended that the
reactors be drained if the UV reactor is off-line for more than one week; however, the
appropriate period for this could be shorter or longer depending on the water quality. If the UV
reactor will be off-line for an extended period of time (longer than 30 days), it is recommended
that the reactor be cleaned prior to re-starting the UV reactor. Routine shutdown and start-up
procedures are discussed in section 5.2.3.
H.3 Off-Specification Issues
UV reactors must be validated as discussed in Chapter 4 and operated within the
conditions determined during validation. When a utility is operating outside of the validated
limits, the utility is operating "off-specification."
LT2ESWTR includes requirements limiting off-specification for compliance with the
LT2ESWTR for unfiltered supplies (40 CFR 141.721(c)(2)); however, the rule does not state an
off-specification requirement for filtered systems or ground water systems. States may develop
statewide or site-specific requirements off-specification requirements for ground water systems.
There are two ways that a ground water system could be operating off-specification.
First, off-specification can occur when the flow, UVT, or UV intensity is outside of the validated
conditions. Second, UV lamps can lose arc if a voltage fluctuation, power quality anomaly, or a
power interruption occurs. LP lamps generally can return to full operating status within 15
seconds after power is restored. However, LPHO and'MP UV lamps exhibit restart times
between 4 and 10 minutes if power is interrupted (Cotton et al. 2002). During these restart times,
the water being distributed is inadequately disinfected and is considered off-specification.
H.3.1 Power Quality Assessment
A power quality assessment at each well site should be performed to determine if power
quality might cause off-specification operation. In addition, the reliability of commercial power
at remote sites may be less than that of more populated areas. Backup power or an
uninterruptible power supply (UPS) may be needed, depending on the findings of the power
UV Disinfection Guidance Manual H-3 • June 2003
Proposal Draft
-------
Appendix H. Issues for Ground Water Systems
quality assessment. If backup power is already available for the well pumps, then the backup
power supply should be assessed to determine if sufficient output is available for the UV reactor.
However, UPS may also be needed if there are frequent power quality problems. For systems
that have storage following UV disinfection, it may be possible to isolate the UV reactor and rely
on stored water to meet demand during periods of power failure. Power quality assessments are
discussed in more detail in section 3.1.3.3.
H.3.2 Well Pump Cycling
Well pumps may regularly cycle on and off in response to changes in distribution system
pressure, causing the UV reactor to also be cycled. Frequent lamp cycling reduces lamp life.
Manufacturers recommend that whenever possible lamps remain energized for a minimum of 6
hours (Dinkloh 2001). In addition, the warm-up time when the UV reactor is coming on-line is
considered off-specification until the UV intensity sensor reading reaches the validated value if
water is flowing to the distribution system.
Depending on its current operation and direction from the State, the utility may need to
consider changing the well pump cycling strategy or incorporate UV reactor controls to reduce
off-specification time and to meet the needs of the distribution system. The utility should discuss
its proposed operating strategy with the UV manufacturer to ensure it is appropriate for the
selected UV reactors. While there may be any number of operating strategies that a utility could
use, two operational strategies that could be incorporated to sequence the well pumping with the
operation of the UV reactor are presented below.
The first strategy is to incorporate a delay that prevents the well pump from starting until
the UV reactor reaches its validated UV intensity sensor setpoint (i.e., no flow through the UV
reactor). Under this control strategy there will be a period when the UV reactor will be "on" but
no flow will be passing through it. This control strategy is only effective when LP or LPHO
reactors are used because their lamps can operate for up to 1 hour under no-flow conditions
(Dinkloh 2001) without overheating. However, MP UV reactors may heat the water above the
safe operating temperature of 50 degrees Celsius in 1 to 15 minutes, causing the reactor's
internal safety devices to shut the reactor off (Miller 2001). As such, this control strategy may
be'infeasible for MP reactors unless they incorporate a low flow waste line that allows water to
circulate through the reactor in order for MP lamps to reach the validated UV intensity sensor
setpoint without overheating. The UV manufacturer should be contacted to confirm that this
operational strategy is feasible with or without the waste line.
The second strategy is to provide a system of automated valves that diverts the UV
reactor discharge away from the distribution system until the reactor reaches its validated UV
intensity sensor setpoint. Then the automated valves are repositioned to direct the water from the
UV reactor to the distribution system. This strategy delivers the off-specification water to
"waste" until the validated UV intensity sensor setpoint is reached. This ensures that sufficient
cooling water will flow through MP reactors to prevent overheating and reactor shutdown. For
this strategy, the utility needs to develop an approach for managing the water that is wasted
during reactor warm-up. The water may be wasted to a sanitary sewer, storm sewer, on-site
disposal or drainage system, or temporary storage tank. The utility should coordinate the
discharge location with the State and other involved parties.
UV Disinfection Guidance Manual H-4 June 2003
Proposal Draft
-------
Appendix H. Issues for Ground Water Systems
Both operational strategies introduce a lag between the time when the pump is initiated
and the time when water is introduced into the distribution system. Because of this, existing
controls may need to be adjusted to avoid insufficient system pressure or storage during periods
of UV reactor warm-up. This will be particularly important for those ground water systems that
have frequent on-off cycles or limited storage.
H.4 Well Location Issues
Ground water production wells are sited in a variety of locations, ranging from urban
areas to remote installations. The well location will affect the design and operation of the UV
installation, especially if there is limited space.
H.4.1 Design Considerations
As discussed in section 3.3.5.2, the UV reactor should be installed within a building or
underground vault if possible to facilitate maintenance and protect sensitive equipment. The
need for enclosure of the UV installation will ultimately be based on the manufacturer's
recommendations, local regulatory and code requirements, environmental conditions, and site-
specific constraints. Site security should be appropriate to prevent tampering with the equipment
and water supply and to protect people from injury (e.g., electrocution).
Well sites, particularly in urban areas, may be spatially constrained by adjacent
development. As a result, the amount of exposed pipe and available area for locating equipment
may be limited. In these cases, it may be necessary to modify the pump discharge piping to
accommodate a UV reactor. When constructing a UV installation in an extremely confined
location, the designer must consider the area necessary for operation and maintenance and the
area needed for installation (e.g., staging areas, personnel, and equipment access). In addition,
the inlet and outlet piping should meet the criteria listed in section 3.3.1.1 as compared to the
validated inlet and outlet piping.
UV reactors are susceptible to damage by suspended sand particles or other debris that
may be present in a ground water supply and pass through the well screens. Therefore, it is
important to determine if sand, grit, or fines are present in a well supply and if it is necessary to
install a sand/debris trap or removal equipment prior to UV disinfection. Particles flowing
through the UV reactor may scratch the lamp sleeves, cause the sleeve wiping mechanisms to
jam, or shield pathogens from UV light, thereby decreasing the UV disinfection effectiveness. In
addition, larger particles could break the lamp sleeves and lamps (see Appendix N for lamp
breakage issues).
H.4.2 Operational Issues
Because most well sites are not continuously staffed, UV installations may need
sufficient automation to allow remote monitoring and operation. Controls and alarms should be.
designed to ensure that real-time operational and monitoring data are communicated to the
UV Disinfection Guidance Manual H-5 June 2003
Proposal Draft
-------
Appendix H. Issues for Ground Water Systems
operators. These factors also emphasize the importance of a power quality assessment and the
design of alarms, monitoring capabilities, and backup power facilities.
Disposal of the chemicals used to clean the UV reactors may be an issue if an on-line
chemical cleaning (OCC) system is used (section 2.4.5). If sewer connections or other standard
means of disposal are not available, then chemical waste will need to be transported off-site for
disposal or handled on-site. Utilities should consult with chemical suppliers and the State when
developing disposal strategies.
H.5 Hydraulic Issues
• The hydraulic issues associated with ground water systems include high operating
pressures, piping configuration, air entrainment, and the potential of water hammer and surge
events.
Many well pumps discharge directly to the distribution system or to elevated or
pressurized storage; therefore, the discharge will often be at system pressure. The UV reactor
design may need to be modified to accommodate these higher distribution system pressures.
The actual inlet and outlet hydraulics of the UV reactor should be designed to match the
validated hydraulics as discussed in section 3.3.1. Space is often limited with ground water
installations so valves, flow meters, or other appurtenances may be directly upstream or
downstream of the UV reactor. Consequently, these site constraints may need to be considered
in determining how the UV reactor should be validated. Detailed discussions of UV installation
layout and validation are given in section 3.3.5 and Chapter 4, respectively.
UV reactors should be flooded at all times because air binding can interfere with the UV
disinfection process or cause the lamps to overheat. UV reactors should be located downstream
of any existing or planned air removal equipment (if necessary). Otherwise, the UV installation
design should include a means for automatically releasing air prior to the UV reactor. The UV
reactor may have integral air release valves or valve ports, or air release valves can be installed
in the inlet and outlet piping.
Pressure surge events (water hammer) near the UV reactor may be more likely with
ground water systems than surface water systems because of the UV reactor's proximity to the
well pumps. Surge events can cause positive or negative pressure transients in the well discharge
piping. Negative pressures as small as -1.5 psi may cause the lamp sleeves to break (Dinkloh
2001). A surge analysis is recommended to determine if surge protection is necessary. Many
well sites and distribution systems are already equipped with surge control tanks to dampen
surge effects. These tanks may provide sufficient protection for the UV reactors, depending on
their location relative to the UV reactors.
Other surge control devices, such as air/vacuum release valves, may rely on the
introduction of air into the system to mitigate surge. As discussed previously, the presence of air
can negatively affect the performance of the UV reactors. Air/vacuum valves should only be
used if surge tanks are not an option and the design can eliminate the air prior to the UV reactor
(e.g., install the valve in a section of pipe at a higher elevation than the UV reactor).
UV Disinfection Guidance Manual H-6 June 2003
Proposal Draft
-------
Appendix H. Issues for Ground Water Systems
H.6 References
Cotton, C.A., R.S. Cushing, and D.M. Owen. 2002. The impact of the draft UV disinfection
requirements on UV facility design and operation. Proceedings of the AWWA Annual
Conference, June 16-20, New Orleans, LA.
Dinkloh, L. 2001. Wedeco-Ideal Horizons. Telephone conversations and email correspondence
by Ben Hauck, Malcolm Pirnie Inc., regarding UV reactors. April 25 and August 10.
Malley, J.P., B.A. Petri, G.L. Hunter, D. Moran, M. Nadeau, and J. Leach. 2001. Full-scale
implementation ofUV in groundwater disinfection systems. Denver, CO: AwwaRF Final
Report.
Miller, A. 2001. Trojan Technologies. Telephone conversation by Ben Hauck, Malcolm Pirnie
Inc., regarding UV reactors. August 10.
Toivanen, E. 2000. Experiences with UV disinfection at Helsinki water. IUVA News 2, no.6: 4-8.
UV Disinfection Guidance Manual H-7 June 2003
Proposal Draft
-------
Appendix I. Issues for Small Systems
The objectives of this appendix are to highlight the issues that small systems face when
considering UV disinfection and to reference the more detailed discussion of these issues in this
manual.
To be classified as a public water system, a utility must provide water to a minimum of
15 service connections or serve at least 25 people for at least 60 days per year (40 CFR 141.2).
For the purpose of this appendix, the term small system includes those utilities serving fewer
than 10,000 people or having a daily production rate of less than approximately 1.0 mgd. Most
of the information regarding UV disinfection in Chapters 2,3,4, and 5 is valid for both small
and large systems.
1.1: Is UV Disinfection Applicable to Small Systems?
UV disinfection is applicable to small systems and may be attractive for the, following
reasons:
» It is a relatively low cost technology for the inactivation of Cryptosporidium (Cotton
; etal.2001).
. Chemical use is little to none.
« Operation is relatively simple and maintenance is low.
« Space needs are small. •
1 *" - '
' Two types of UV reactors can potentially be used by small systems, conventional and
point-of-entry (POE) devices. Conventional UV reactors are manufactured for a wide range of
flows (e.g., from 20 gallons per minute (gpm) up to 40 mgd) and are described in section 2.4.
POE units are small UV reactors that are installed at the service connection of the customer.
POE units contain the same components as conventional low-pressure (LP) installations but are
more compact. They are primarily intended for use at individual properties and may be more
suitable for utilities with a limited number of service connections. POE units are required to be
owned, controlled, and maintained by the utility or by a person under contract with the utility to
facilitate proper operation and maintenance and compliance with the treatment requirements
(Safe Drinking Water Act (SDWA) Section 1412(b)(4)(E)). The use of POE units may result in
higher total costs when compared to a centralized, conventional UV installation for all but the
smallest water utilities.
1.2 What Information is Necessary to Assess the Feasibility of UV
Disinfection?
Small systems generally need the same information to assess UV disinfection as larger
systems. Chapter 3 describes the planning and design process for a UV installation in a
UV Disinfection Guidance Manual 1-1 June 2003
Proposal Draft
-------
Appendix 1. Issues for Small Systems
conventional plant and discusses each of the elements that should be considered. In general, the
utility should answer the following questions when assessing the suitability of UV disinfection:
. What are the disinfection goals and can UV disinfection be used to meet these goals?
(section 3.1.1) •
. What are the minimum, average, and maximum flowrates that the UV reactors will
need to treat? (section 3.1.3.2)
. What is the design UV absorbance at 254 nm (A254) and corresponding design UV
transmittance (UVT)? What is the fouling potential of the water supply? What is the
potential for process upset or variability in water quality? Do any of the existing
processes have the potential to interfere with the performance of the UV reactors?
(section 3.1.3.1)
. Can the UV reactors be incorporated into the existing hydraulic profile? If not, can
the existing operations be modified to accommodate the UV reactors, or does
intermediate pumping need.to be installed? (section 3.1.6.1)
. Can the UV installation be incorporated into the existing facility layout? Does a
building need to be constructed to house the UV reactors? (section 3.1.6.2)
. Is the quality and reliability of the electrical power supply adequate? Does a backup
power supply or other supplemental electrical equipment need to be installed?
(sections 3.1.3.3 and 3.3.4)
« How should the UV reactors be controlled? What level of automation and
operational complexity is appropriate? Does the potential for power savings justify
using a more complex operating strategy? Is the existing operations staff sufficient?
(section 3.3.3) . '
. Is the number of UV reactors installed appropriate to efficiently respond to the
anticipated range of flowrates? Does the UV installation have the capability to be
expanded to meet future increases in demand? Is there sufficient redundancy to allow
operating flexibility and to meet the disinfection goals under the operating scenarios?
(section 3.1.3.2)
i ;
• How should the UV reactors be procured? (section 3.2)
. Do the characteristics of the proposed UV application (e.g., flowrate, UVT, UV
intensity) differ from those under which a selected UV reactor was validated? If so,
should the selected equipment be validated on-site or off-site under characteristics
that match those of the intended installation? (Chapter 4 and section 3.1.4.2)
• What is the capital cost of the UV installation? What are the operating costs
associated with a UV installation? (section 3.1.7)
» What is the cost of the UV installation as compared to other disinfection alternatives?
UV Disinfection Guidance Manual 1^2June 2003
Proposal Draft
-------
Appendix I. Issues for Small Systems
• Is there a cost benefit to using POE units as opposed to a centralized UV installation?
If so, how should the utility administer the POE units? Is some form of access
, agreement or water use ordinance necessary to allow administration of the POE units?
1.3 Do the UV Reactors Need to be Housed in a Building?
If possible, the UV reactors should be constructed within a building to facilitate
maintenance and protect the UV reactors. Nonetheless, the need for enclosing the UV reactors
will ultimately be based on manufacturers' recommendations, State requirements, and
environmental conditions. Although some current UV installations do not have a building (e.g.,
Hanovia facility in Australia), local building and electrical codes may necessitate a building or
other protection for the electrical equipment. Regardless of whether the UV reactors are
enclosed, site security is important to prevent tampering with the equipment and water supply
and to protect people from injury (e.g., electrocution). Section 3.3.4 discusses the electrical
equipment issues that should be considered during the planning and design of a UV installation.
1.4 Do the Components of a Small System Differ from Larger UV Reactors?
The main components of a UV reactor (including the necessary instrumentation and
controls) do not differ between large and small systems. Components of the UV reactor may
include the UV lamps, lamp sleeves, UV intensity sensors, ballasts, and cleaning mechanisms,
which are described in section 2.4.
Full-scale drinking water applications generally use LP, low-pressure high-output
(LPHO), or medium-pressure (MP) lamps. Small systems may find reactors that use LP or
LPHO lamps more economical because they convert power into germicidal wavelengths of UV
light more efficiently than MP lamps. Additionally, LP lamps typically have a longer useful life
than either MP or LPHO lamps. For small systems, UV reactors with LP lamps are likely to
represent the most cost-effective disinfection solution. For systems that serve near 10,000
people or treat near 1 mgd, more consideration should be given to LPHO or MP lamps. An
additional discussion of the different lamp types is given in section 2.4.2.
1.5'. What are the Power Needs?
, The power needs depend on the manufacturer. Common manufacturers' power needs are
as follows:
. POE UV units - 120V/60Hz/l phase
. Conventional LP reactors - 120/208V/60Hz/3 phase
. Conventional LPHO reactors - 480V/60Hz/3 phase
. Conventional MP reactors-480V/60Hz/3 phase
UV Disinfection Guidance Manual P5 June 2003
Proposal Draft
-------
Appendix I. Issues for Small Systems
Backup power may be necessary, depending on the type of installation that is selected,
the power quality at the installation site, and the regulatory requirements for the installation.
However, backup power for small systems may not be necessary as some small systems can
accommodate a shutdown for longer periods because there is sufficient storage to meet demand.
Additional detail on the need for backup power and the factors that should be considered when
assessing the power supply are discussed in section 3.1.3.3.
1.6 Do Small UV Reactors Need to be Validated?
All UV reactors, including POE units, are required to be validated (40 CFR 141.729(d)).
Small systems will probably purchase UV reactors that have been validated by the manufacturer.
UV intensity sensor operating setpoints (and potentially UVT setpoints) are established at
specific flowrates during validation testing. These are the setpoints that the systems are required
to operate within to receive inactivation credit. For many small UV reactors and nearly all POE
units, UV reactor control will be limited to "on" and "off" with UV reactor shutdown under
specific critical alarm conditions. Chapter 4 discusses the UV reactor validation requirements,
and Chapter 5 describes operating requirements.
1.7 How are UV Reactors Monitored?
Monitoring UV reactors (conventional and POE) is required to ensure that the UV
reactors are operating within the validated range (40 CFR 141.729(d)). Parameters that must be
monitored include flowrate, UV intensity sensor readings, and UVT (if it is part of the control
strategy) (40 CFR 141.729(d)). POE units should be equipped with mechanical warnings to
ensure that customers are automatically notified of operational problems. Additional detail on
monitoring requirements is provided in section 5.4.1.
1.8 Can the UV Reactors be Operated Remotely?
UV reactors can be operated remotely if the monitoring components provide a 4-20 mA
analog output signal and are integrated into a control strategy. Even though UV reactors can be
operated remotely, routine inspections and on-site maintenance will be necessary to confirm that
the UV reactor is operating properly. Provisions for hydraulic and electrical lockout should be
provided to enable local isolation and lockout for maintenance. Section 3.3.3 provides additional
detail on control strategies for centralized UV installations, and section 5.3 discusses operations
and maintenance needs.
If the utility uses POE units, it may be beneficial to telemeter all alarm conditions to a
central location to facilitate administration and maintenance of the POE units. However,
incorporating this remote capability will likely increase the cost of the UV installation.
UV Disinfection Guidance Manual
Proposal Draft
1-4
June 2003
-------
Appendix 1. Issues for Small Systems
1.9 How Much Maintenance is Needed?
Maintenance is generally limited but will vary depending on the manufacturer and the
specific application. Maintenance may include the following activities:
. Periodic calibration verification of UV intensity sensors, UVT meters, or flowmeters
• Periodic replacement of UV intensity sensors, UVT meters, or flowmeters (if
applicable), depending on calibration or age of the equipment
» Lamp sleeve and reactor cleaning
. Replacement of UV lamps and other components
. Maintenance of other operating components and the electrical systems
Operators should be trained by the UV manufacturer on the proper operation and
maintenance of the UV reactors. The utility should consider contracting trained service
personnel to maintain the UV reactors if this is not possible. Additional detail on operations and
maintenance is given in section 5.3.
1.10 References
Cotton, C.A., D.M. Owen, G.C. Cline, and T.P. Brodeur. 2001. UV disinfection costs for
inactivating Cryptosporidium. Journal of the American Water Works Association 93, no.
2: 67-74.
UV Disinfection Guidance Manual
Proposal Draft
1-5
June 2003
-------
Appendix J. Pilot-Scale and Demonstration-Scale Testing
In some cases, pilot- or demonstration-scale testing may be warranted to aid in selection
of design criteria. For example, long-term UV unit performance will be impacted by lamp aging
and sleeve fouling. With increased use, lamp output decreases due to deposition of inorganic
material on the outside and inside of the sleeve (i.e., "fouling"). Fouling reduces the
transmittance of the lamp energy to the water. Over time, these phenomena will contribute to a
reduction in UV dose. The effect of these parameters should be incorporated into the UV reactor
design. A "lamp aging factor" and a "fouling factor" are usually specified by the design
engineer. Pilot testing can provide useful information for the development of these factors.
This appendix discusses when pilot or demonstration tests may be needed and the types
of tests that may be performed on UV disinfection systems. The purpose(s) of pilot and
demonstration testing is to establish or confirm system design factors, test system reliability, and
evaluate operation and maintenance (O&M) needs. The tests described herein may be performed
individually or in parallel. Validation of reactor microbial inactivation performance is addressed
separately in Appendix C (Validation Protocol).
J.1 When Is Pilot or Demonstration Testing Needed?
Pilot and demonstration tests can be used to meet the following three goals:
1. Assess the impact of unusual water quality conditions (e.g., high calcium or iron
concentrations).
2. Improve estimation of safety factors for large water systems for which such an
investment can yield a high return in reduced life cycle costs.
3. Gain first-hand experience with operating and maintaining a UV installation.
A UV disinfection system should be designed with some knowledge of the likely fouling
potential-of the water and lamp-aging characteristics to ensure the system operates as intended.
If the design and the operation protocol do not properly account for the effects of lamp aging and
sleeve fouling, the system may go into alarm frequently (indicating under dosing);
While pilot or demonstration testing may be warranted in some cases, it is becoming less
necessary as more performance and fouling information is developed. The need for pilot or
demonstration testing should be carefully considered in light of the pre-existing data available on
both system performance and water quality effects on sleeve fouling. Pilot or demonstration
testing may be used to gain operational experience or primacy agency acceptance, as discussed
in the following sections.
Microbiological challenge tests are not recommended during pilot studies because
inactivation efficiency in a pilot system may not be indicative of full-scale performance.
However, UV reactor validation bioassays could be conducted as part of a full-scale
UV Disinfection Guidance Manual
Proposal Draft
J-l
June 2003
-------
Appendix J. Pilot-Scale and Demonstration-Scale Testing
demonstration test if on-site testing is planned. Appendix C presents a detailed discussion of UV
reactor performance validation.
Because UV disinfection is a relatively new drinking water treatment technology in the
United States, State regulatory agency acceptance may depend in part on the confidence in the
technology gained through pilot- and demonstration-scale studies. Identifying previous studies
of similar scope that provide background and precedents may also be helpful in gaining
acceptance of its planned use (see, for example, et al. Mackey 2001).
If a utility chooses to or is required to conduct pilot or demonstration tests, the primacy
agency should understand the objective(s) of the test(s) and the methodologies used. It is
-recommended that the primacy agency be contacted before testing and involved throughout the '
pilot and/or demonstration testing. Identifying and resolving State regulatory agency concerns
when planning testing can help produce a more useful dataset. Additionally, it may be helpful to
include the State in interim briefings on progress and results, and to give them a final report after
completing the testing.
J.1.1 Water Quality Impacts
Extensive data have been generated from pilot-scale testing on waters of low to moderate
hardness and iron content (Mackey et al. 2001, Mackey and Gushing 2003). At total hardness
and calcium levels below 140 mg/L and low iron (less than 0.1 mg/L), standard cleaning
protocols and wiper frequencies (one sweep every 15 minutes to an hour) were more than
adequate to deal with the impact of sleeve fouling at all sites tested. At sites with hardness or
iron in the feed water that exceed these levels, it may be'advantageous to evaluate fouling rates
on a site-specific or worst case basis via pilot or demonstration testing to identify how best to
keep the sleeves clean.
J.1.2 Lamp Fouling Factors for Large Systems
In UV reactor design, a lamp aging factor of 0.7 is commonly used, as discussed in
section 3.1.3.1 of the Manual. For larger systems, it may be economical to pilot or
demonstration test lamp aging to provide data for selecting lamp aging and low-dose-alarm
design factors that will best balance operational costs (how many hours one wants to be able to
operate a lamp before replacing it) with capital costs (the size of the system needed based on
end-of-lamp-life). Lamp aging factors may also be obtained from a certified lamp age testing
program performed by equipment or lamp manufacturers. A lower lamp aging factor means the
utility will have less frequent lamp replacements, but may require a larger system to ensure
compliance at all times.
J.1.3 Gaining Operational Experience
Due to the small number of U.S. drinking water UV installations, very few United States
operators have experience with UV disinfection systems. It may be beneficial for a facility's
UV Disinfection Guidance Manual J-2 June 2003
Proposal Draft '*'.
-------
Appendix J. Pilot-Scale and Demonstration-Scale Testing
staff to obtain operational experience with UV disinfection systems prior to selecting and
implementing UV disinfection. If a utility staff becomes familiar with the operational aspects of
UV disinfection, that staff will be able to provide feedback input on the UV installation design.
In addition, operational experience can help facility managers to determine the staffing/training
needs and help maintenance staff understand and plan for the maintenance needs of the system
(e.g., time to change lamps and calibrate UV intensity sensors and perform manual cleaning).
On-site testing is site-specific depends on the needs and preferences of the utility.
Methods by which facility staff can gain operational experience (besides on-site testing) include:
site visits and partnerships with systems already using UV disinfection; conversations and visits
with manufacturers and attendance of seminars; and on-site training programs (a detailed
discussion of training programs is provided in section 5.7.2).
J.2 Pilot- Versus Demonstration-Scale Testing
Table J.I presents a comparison of the advantages and disadvantages associated with
pilot-scale and demonstration-scale testing. Pilot-scale testing involves operating a smaller
version of a full-sized UV disinfection reactor. It may or may not include all the components of
the full-sized system. Demonstration-scale testing is essentially pilot testing of a full-scale UV
disinfection reactor.
Table J.1 Comparison of Pilot and Demonstration Testing
Testing Method
Advantages
Disadvantages
Pilot-scale
• Smaller footprint needed for UV
reactors
• Less-expensive installation and
operation
• Operators gain O&M experience
• High flexibility in placement of
equipment
• Lesser volumes of water to dispose
Design conditions for UV
disinfection systems may not scale-
up to full-scale systems
In rare cases it may be advisable to
use pilot-scale treatment process
equipment (filters, clarifiers, etc.) to
simulate operational conditions
(e.g., upstream ozone process)
Demonstration-
Scale
Confidence in long-term operation
of the UV unit due to the
representative scale at which
results are obtained
Scale-up factors need not be
developed
Operators gain operations and
maintenance experience on a full-
scale system.
Approval from the primacy agency
may be required to conduct a
demonstration study
Demonstration setups are not as
flexible as pilot studies for
operational experimentation
Installation and operation more
expensive than pilot scale
Greater volumes of water to
dispose.
UV Disinfection Guidance Manual
Proposal Draft
J-3
June 2003
-------
Appendix J. Pilot-Scale and Demonstration-Scale Testing
J.3 Documenting the Test Reactor
For a given test, should the properties of the components that may influence the final
outcome of the test should be identified and recorded. That record may later be used to confirm
that key components of installed UV reactors match those of the systems tested. Table J.2 lists
the components of a UV disinfection system that should be documented and compared between
testing and the final design.
Table J.2 Key Components Associated with UV Reactor Pilot-Scale
and Demonstration Scale Testing
Test
Operational Experience
Fouling Assessment
Head loss Assessment
(demonstration-scale only)
Ballast Performance
Cleaning Mechanism Performance
Lamp Aging/Failure
Sleeve Breakage
Controls/Alarms
Components to Document
Controls, alarms, cleaning mechanisms,
operation, maintenance.
Lamps, sleeves, ballasts, power settings,
UV intensity sensor windows, flow velocity.
Reactor and wetted components, inlet/outlet
conditions.
Lamps, sleeves, ballast, power settings,
operation.
Lamps, sleeves, ballasts, power settings,
ballast operation, UV intensity sensor
windows (if wiper used), cleaning
mechanism, cleaning solutions, wiper
maintenance and operation.
Lamps, sleeves, ballasts, power settings,
ballast operation, cleaning mechanism,
cleaning solutions, wiper maintenance and
operation.
Sleeves, cleaning mechanisms, flow
velocity, water hammer.
Lamps, sleeves, ballasts, UV intensity
sensors, cleaning mechanisms, controls,
operation.
J.4 Testing Objectives
Pilot/demonstration testing may be used to gain information on a specific UV reactor, a
specific water treatment plant (WTP) site, or a combination of the two. Common test objectives
include the following topics:
. The long-term performance and failure modes of the lamps
. The efficacy of cleaning mechanisms for lamp sleeves and UV intensity sensor
windows
• The stability of UV intensity and UVT monitors
UV Disinfection Guidance Manual
Proposal Draft
J-4
June 2003
-------
Appendix J. Pilot-Scale and Demonstration-Scale Testing
• The reliability of controls and alarm systems
• The ease of lamp and UV intensity sensor replacement, the use of reference sensors,
and the maintenance of cleaning devices and solutions
. The rate of fouling on lamp sleeves and UV intensity sensor windows
. The most appropriate cleaning method
• The head loss across the reactor at various flow rates (demonstration-scale only)
• The impact on other unit operations at the WTP
The information obtained during pilot and demonstration testing should be applicable to
the final UV disinfection system installed at the WTP. Accordingly, the equipment tested should
be representative of the UV disinfection system that will be installed. Specific elements of a
pilot/demo-scale system that should be identical include the UV intensity sensors, lamp and
sleeve type, power system, cleaning system, cleaning frequency, and water quality. For
example, lamp-aging data on a 3 kW 25 cm medium pressure (MP) lamp driven by an
electromagnetic ballast cannot be used to predict the aging expected with a 10 kW 50 cm MP
lamp driven by a transformer.
J.5 Testing Protocols
This section describes the major elements and benefits of a range of pilot and
demonstration testing protocols to investigate sleeve fouling and cleaning, lamp aging, head loss
and alarms and controls.
J.5.1 Assessing Fouling
A fouling assessment can be conducted to answer the following questions:
• How fast do the lamps foul?
. How does water quality affect fouling?
« What lamp fouling factor should be specified?
* What lamp cleaning interval is required?
• What sleeve replacement interval is required?
• How do lamp/reactor configurations affect fouling?
• Is fouling of the UV intensity sensor window(s) significant and how should it be
addressed?
UV Disinfection Guidance Manual J-5 June 2003
Proposal Draft
-------
Appendix J. Pilot-Scale and Demonstration-Scale Testing
Fouling may occur, on the inner and outer surfaces of the lamp sleeves, the internal
surfaces of the reactor, and UV intensity sensor windows. Lamp sleeve fouling may have an
impact on dose delivery and cleaning requirements. Sensor window fouling may have an
important impact on assessing dose delivery (e.g., the sensor will not be able to accurately
measure lamp intensity). Fouling on the wetted surfaces of a UV reactor has been attributed to
precipitation of compounds whose solubility decreases as temperature increases, precipitation of
compounds with low solubility, and deposition of particles by gravity settling and turbulence-
induced impaction (Lin et al. 1999a). More detailed discussion on-fouling is provided in sections
3.1.3.1andA.4.1.4.
'. In one method for assessing sleeve fouling similar to that employed by Lin et al. (1999b),
a new lamp can be placed inside the fouled sleeve and ignited. Ultraviolet absorbance at 254
nanometers (A254) is measured by a calibrated radiometer and compared to a similar
measurement made using a new, clean sleeve. The ratio of these two measurements (UV light
passing through the fouled sleeve to that passing through the new sleeve) is the lamp sleeve-
fouling factor.
Manual, chemical cleaning should restore the sleeve A254 to very near that of a new, clean
sleeve. If not, manually clean the inside of the sleeve and measure A254. If it is still low, the
sleeve material has likely degraded. If Aas4 cannot be recovered, further testing may be used to
identify a proper sleeve replacement interval. The lamp sleeve-fouling factor can be plotted as a
function of time. Worst-case results can be analyzed to determine cleaning requirements and
fouling factors for design purposes.
When assessing lamp sleeve fouling, care should be taken to ensure that the results scale-
up to full-scale applications. Some differences in system geometry may lead to erroneous
conclusions based on pilot data alone. For instance, in parallel flow reactors, fouling has been
found to be uneven along the length of the lamps (Lin et al. 1999a). If the lamp and lamp sleeve
geometry (e.g., length or diameter) of the pilot unit is very different from the full-scale system,
the fouling that will occur in the full-scale plant may be markedly different from expectations
based on pilot-scale data. The lamp lengths will be very different and end-effects may be more
pronounced (i.e., the blackened lamp ends of an aged lamp will comprise a greater percentage of
the total length of the lamp).
To assess fouling on the UV intensity sensor windows, clean the sensor monitoring
windows with phosphoric or citric acid at varying time intervals and record the change in sensor
readings. It is expected that the rate of fouling on the lamps will be greater than the rate of
fouling on the sensor windows due to elevated lamp temperature.
J.5.2 Evaluating Cleaning Systems
An evaluation of system cleaning methods can be performed to answer the following
questions:
. Does a particular cleaning protocol work for the UV reactor application?
UV Disinfection Guidance Manual J-6 June 2003
Proposal Draft
-------
Appendix J. Pilot-Scale and Demonstration-Scale Testing
« What is the long term effectiveness of the cleaning method?
. What cleaning frequency is required for each method considered?
Various cleaning methods can be used to periodically remove the foulants that
accumulate on the lamp sleeves and UV intensity sensor windows. Lamp sleeve cleaning
methods include off-line chemical cleaning (OCC) (off-line manual or mechanized cleaning),
on-line mechanical cleaning (OMC) (e.g., brushes or rings), and on-line physicochemical wipers
(acid solution in a wiper collar). Sensor window cleaning methods also include manual cleaning
and mechanical wipers.
J.5.2.1 Assessing Lamp Sleeve Cleaning Protocols
A sleeve cleaning assessment should be performed on at least four lamp sleeves, and the
results be used to identify a sleeve fouling design factor for sizing the UV reactor based on the
lowest individual sleeve-fouling factor observed. This will help ensure proper dose delivery for
the entire life of the sleeve. One method for assessing lamp sleeve cleaning needs is detailed
below:
Pass water through the reactor at the minimum flow rate and operate the lamps at
maximum power. With systems using mechanical or physicochemical wipers, an unwiped
sleeve can be used as a control to verify that fouling is occurring. The manufacturer's
recommendations regarding the maintenance of the cleaning device should be followed.
Record the UV intensity sensor readings before and after the cleaning cycle and use this
data to optimize the cleaning frequency. Sensor windows should be manually cleaned before
measurements to ensure only lamp sleeve fouling is affecting the sensor values. If possible,
check all UV intensity sensor readings with a reference sensor.
At regular time intervals and immediately prior to the scheduled sleeve cleaning cycle,
remove the lamp sleeves and assess the sleeve Azs4 for low pressure (LP) lamps and absorbance
from 200 - 400 nm for MP. The non-destructive method of Lin et al. (I999b) may be used as
discussed in section J.5.1.
After 6 months, or when manual, chemical cleaning is recommended, remove the sleeves
and measure the sleeve Ai54 before and after cleaning the outer surfaces. If the new sleeve
transmittance is not restored by the cleaning, it is likely that the sleeve material has fouled
internally or permanently degraded. Further monitoring and testing may be necessary to identify
the proper sleeve replacement interval.
J,5.2.2 Assessing UV Intensity Sensor Window Cleaning Protocols
To assess fouling of the UV intensity sensor window, one alternative is to operate the
reactor under the conditions suggested in section J.5.2.1 (i.e., minimum water flow rate,
maximum lamp power). After 6 months, or a time interval suggested by the manufacturer, a
chemical cleaning of the monitoring sensor window could be performed. Alternatively, cleaning
UV Disinfection Guidance Manual J-7 June 2003
Proposal Draft
-------
Appendix J/ Pilot-Scale and Demonstration-Scale Testing
could be performed when the sensor reading falls to a minimum value suggested by the
manufacturer. The A254 of the window should be measured before and after cleaning. A sensor
window cleaning frequency can then be estimated as discussed in section 3.1.3.1.
J.5.3 Assessing Head Loss
i . •«•
Head loss assessments should be performed for demonstration-scale (full-scale) systems
to verify that head loss constraints at the final install station will not be exceeded. Head loss data
from pilot scale units should not be used to estimate head loss in a full-scale system.
The head loss, AH, through a UV reactor may be calculated according to Equation J.I:
Kv2
A// = -— Equation J.I
2g . .
where
K = Head loss coefficient (unitless) for the UV reactor
v = Water velocity (m/s) through the reactor . -
g = Gravitational constant (9.8 m/s2)
To assess head loss through a UV reactor, the UV unit can be installed with
instrumentation to measure pressure loss across the reactor (including baffles and specialized
.inlet/outlet piping). Since the head loss coefficient will be higher at lower temperatures due to
decreased water viscosity, it may be desirable, if feasible, to measure head loss at the lowest
water temperature expected at the UV reactor installation to assess the worst-case condition.
If head loss is measured at various flow rates through the reactor, including the minimum
and maximum flow rates, these measured head loss values can be plotted as a function of the
square of the calculated water velocity through the reactor to determine a head loss coefficient.
J.5.4 Lamp Aging and Failure
, A lamp aging evaluation can be conducted to answer the following questions:
, . U.S. EPA Headquarters Library
. What is the actual operating lamp life? . Mail Code 3404T
: . • 1200 Pennsylvania Avenue, NW
. How does lamp output degrade over time? Washington DC 20460
202-566-0556
• What lamp aging factor should be specified?
The service life of a UV lamp extends for thousands of hours. The gerrnicidal output of
the lamp will decline during this period (Phillips 1983). In MP systems, UV lamp aging can also
result in a change in the spectral output over time. With polychromatic (MP) UV lamps, lower
wavelengths will likely decline at a faster rate than will higher wavelengths. The rate and
manner in which a lamp ages is lamp- and operation-specific. A detailed discussion of lamp
UV Disinfection Guidance Manual J-8 June 2003
Proposal Draft
-------
Appendix J. Pilot-Scale and Demonstration-Scale Testing
aging is presented in section A3.1.6. Lamp output will decrease overtime as a function of the
lamp hours in operation,- the number of on/off cycles, and the power applied per unit (lamp)
length. .
Lamp aging tests should be designed to assess the. reduction and variance in lamp
germicidal output over time under defined worst-case operating conditions. Lamp age testing
may use either a pilot/demonstration-scale UV reactor installed at a WTP or a test bed designed
to emulate the reactor (i.e., identical power supply). It is strongly recommended that all tests be
done with the lamps housed in the sleeves and powered by the ballasts that will be used in the
final application. It is best if the lamp sleeves are maintained free of external foulant during
aging tests, in a manner similar to that of the final application.
Factors to consider in designing the test(s) include lamp batch, lamp assembly, electrical
characteristics of the ballasts, heat transfer from the lamps to the water, and operation of the
lamps. Since lamps will be manufactured in batches, it is recommended that lamps from several
different lots be evaluated. During demonstration and pilot scale testing, the lamps should be
operated in a manner and in an environment that reflects conditions expected when the UV
disinfection system is installed at a WTP.
Parameters to monitor over time include electrical power delivered to the ballast,
electrical power delivered to the lamp, and water temperature. If UV intensity sensors alone
monitor lamp output, it is recommended that the A254 of the water also be measured.
During testing, it is recommended that the following analyses be considered:
. Visually inspect the lamps at regular intervals to document any visible degradation of
the lamp assembly, including electrodes and seals, and any darkening of the lamp
envelope;
« Document any fouling on the internal surfaces of the lamp sleeves;
• Measure the germicidal output of the lamp under fixed conditions of ballast operation
(e.g., power setting), heat transfer (e.g., lamp sleeves), and environment (water
temperature and transmission).
- One measurement should be made with the lamps aged 100 hours ("new").
- The germicidal output may be measured using one of the following: a radiometer
equipped with a germicidal filter; a reference UV intensity sensor or radiometer
from 200 to 400 nm; or by bioassay.
- The output from various positions along the lamp may be measured based on
visual inspection (i.e., the pattern of darkening on the lamp).
- If lamp power is variable, lamp output as a function of lamp power setting may be
measured.
• Assess the output from lamps of different lots.
Pilot/demo-scale test data and visual inspections can be used to identify operational
issues and provide operational guidance. The output of the lamps measured under fixed
operating conditions can be plotted over time and fit to provide mean expected performance and
prediction intervals (e.g., 90 ,95th, and 99th percentiles) to estimate the range of performance in
UV Disinfection Guidance Manual
Proposal Draft
J-9
June 2003
-------
Appendix J. Pilot-Scale and Demonstration-Scale Testing
lamp intensity at different lamp ages. In a robust system, all the lamps will age in a similar
manner. If lamps age differently than expected, the results will affect dose delivery and UV
intensity sensor measurements. This data can be used to assess a proper end-of-lamp-life.
J.5.5 Evaluating Controls and Alarms
An evaluation of the UV reactor controls and alarms should be conducted to verify their
performance and to gain familiarity with alarm/control response procedures. A test plan is
typically organized prior to testing, describing the control function to be tested, the test
procedure, and the expected response. Faults may be induced or simulated. Low-dose
conditions may be simulated by reducing lamp power, increasing the A254 of the water, reducing
or increasing flow beyond the validation limits of the reactor, turning off lamps or ballasts, or
disconnecting sensors. Valve failure, high temperature, and ground fault interrupts may be
induced or simulated. Simulating faults may require disconnecting components of the UV
disinfection system or using modified electronics. Accordingly, qualified personnel as identified
by the manufacturer of the UV disinfection system should undertake these simulations. All
operational functions can be verified, including startup and shutdown sequences and cleaning
cycles. Dose pacing if used, may be verified by monitoring lamp power settings and dose
compliance as flow rate, A254, and lamp output are varied.
J.6 References
Lin, L.-S., C.T. Johnston, and E.R. Blatchley. 1999a. Inorganic fouling at Quartz: Water
interfaces in ultraviolet photoreactors -1 Chemical characterization. Wat. Res. 33, no 15:
3321-3329.
Lin, L.-S., C.T. Johnston, and E.R. Blatchley. 1999b. Inorganic fouling at Quartz: Water
interfaces in ultraviolet photoreactors - II Temporal and spatial distribution. Wat. Res. 33,
no 15: 3330-3338.
Mackey, E.D., R.S. Cushing, and G.F. Crozes. 2001. Practical aspects of UV disinfection.
Denver, CO: AWWA Research Foundation and AWWA.
Mackey, E.D. and R.S. Cushing. Publication anticipated in 2003. Bridging pilot-scale testing to
full-scale design of UV disinfection systems. Denver, CO: AWWA Research Foundation
and AWWA.
Phillips, R. 1983. Sources and application of ultraviolet radiation. Academic Press: New York.
UV Disinfection Guidance Manual J-10 June 2003
Proposal Draft
-------
Appendix K. Preliminary Engineering Report
Critical design and implementation issues need to be resolved early in the planning phase
of a UV disinfection facility. The purpose of a preliminary engineering report is 1) to provide
conceptual level layouts and preliminary cost estimates for implementation of UV disinfection at
the water treatment plant (WTP), and 2) to recommend an implementation plan for UV
installation design and construction. Specific components of the preliminary engineering
analysis are listed below:
« Identification of UV reactor design criteria and implementation issues
• Evaluation of UV reactor alternatives and potential locations for the proposed
installations in the plant treatment train
• Determination of the hydraulic characteristics of the UV reactor and incorporate it
into the hydraulic model of the plant
Development of estimates for capital, operational, and life-cycle costs for each
alternative
. Comparison of feasible alternatives and development of implementation
recommendations
This appendix presents an example of a preliminary engineering report (PER) for
retrofitting a UV disinfection facility into an existing WTP. The basic elements involved in the
planning phase of the UV installation are discussed in this report. The format and content of a
site-specific PER should be coordinated with the State. This example report is based largely on
a predesign report prepared for North Shore Water Commission, Wisconsin (Carollo Engineers
2001).
Chapter 3 of the Guidance Manual presents a detailed discussion of UV installation
planning and design principles. A flowchart depicting the planning and design process is
included in Figure 3.1. Table K.I presents a correlation between the flowchart elements
discussed in Chapter 3 and respective sections in this appendix.
Table K.1 Elements of the Planning and Design Process (Ref. Figure 3.1)
Element
Define UV disinfection goals
Identify potential retrofit locations
Determine design parameters
Evaluate potential UV reactors
Evaluate operational and control strategies
Evaluate hydraulic profile and site layouts
Compare retrofit options and costs; select retrofit locations
Chapter 3 Section
3.1.1
3.1.2
3.1.3
3.1.4
3.1.5
3.1.6
3.1.7
Appendix K
Section(s)
K.1, K.2
K.5
K.2.2
K.3
K.4
K.5
K.6, K.7, K.8
UV Disinfection Guidance.Manual
Proposal Draft
K.-1
. June 2003
-------
Appendix K. Preliminary Engineering Report
K.1 Background of Example WTP
A 20 million gallon per day (mgd) surface WTP is used as an example in this appendix.
The water treatment processes employed are coagulation and sedimentation pretreatment,
granular media (anthracite and sand) filtration followed by chlorine disinfection. Since the plant
was put into service in the 1960s, water quality regulations have become more stringent. In
addition, there are growing concerns over chlorine-resistant pathogens (e.g., Cryptosporidium)
and production of chlorinated disinfection byproducts (e.g., trihalomethanes, haloacetic acids).
In order to upgrade the facility to meet current and future regulations and health concerns,
several research studies have been performed involving the use of ozone, membranes, and UV
disinfection. From the results of those studies, it was concluded that the most feasible and cost
effective solution to achieve disinfection of chlorine-resistant pathogens was to add UV
disinfection to the current treatment train.
The following are the general performance goals of the UV installation for the example
WTP:
• Provide 2-log Cryptosporidium inactivation.
. Provide an additional disinfection barrier for other chlorine-resistant pathogens.
K.2 UV Disinfection Criteria
This section includes general information regarding the optimal application point for UV
disinfection at the WTP and design considerations for implementation.
K.2.1 Application Point and UV Transmittance
One of the important parameters controlling UV installation design is the UV
transmittance (UVT) of the water to be treated (section 3.1.3.1). The lower the UVT, the greater
the UV intensity is needed to provide a given UV dose at a given flowrate. UVT typically varies
with source water, seasonally, and through the treatment processes. Consequently, a thorough
UVT analysis was completed during development of design criteria.
K.2.1.1 Point of Application for UV Disinfection
In keeping with the content of this guidance manual, the UV disinfection alternatives
assessed for the WTP were limited to applications after filtration. Based on statistical results of
the filtered water ultraviolet absorbance at 254 nanometers (A25-0 data, the UV reactors are sized
based on a 0.032/cm AIM (93 percent UVT; 10 mm path length; light at 254 nm), which is the
99* percentile minimum of the available A254 data.
UV, Disinfection Guidance Manual K-2 June 2003
Proposal Draft
-------
Appendix K. Preliminary Engineering Report
K.2.1.2 Treatment Chemical Impact on Absorbance
Some chemicals used in water treatment absorb UV light and hence, can influence the
design absorbance value, as discussed in Chapter 3.1.2 of the manual. Ferric iron and
permanganate are two of these, and are used at the example WTP. Ferric iron strongly absorbs
UV light; however, post-filtration iron levels are generally low. Permanganate absorbs UV light,
but at permanganate levels of less than 1 mg/L, which is typically the case post-filtration, the
impact is not significant. Therefore, for the PER, chemical A254 is not considered to influence
the UV design criteria.
K.2.1.3 Power Quality Impact on Absorbance
As stated in section 3.1.3.3, the sensitivity of UV reactors to power fluctuations make
electrical power supply a critical component of the UV installation planning and design.
Preliminary pilot testing of UV reactors over the course of a year at this site did not indicate any
problems with existing water utility power quality for the UV reactor's operational continuity.
Therefore, for this PER, power quality is not considered to negatively impact the UV installation
design.
K.2.2 Inactivation Goals and UV Dose
The goal of UV disinfection at the example WTP is to provide inactivation of chlorine-
resistant pathogens. By using UV disinfection, the general goal of improving public health
protection will be met and compliance with the Long Term 2 Enhanced Surface Water Treatment
Rule (LT2ESWTR) Cryptosporidium inactivation requirements may be achieved, if needed (40
CFR 141.702). (Source water sampling for "Bin" determination has not yet been completed at
this example WTP, so the level of additional credit needed in the future is unknown.) UV
disinfection credit will also be available for Giardia and virus inactivation. This additional UV
disinfection credit will most likely reduce chlorine disinfection requirements, and hence, reduce
disinfection by-product formation.
The desired UV dose (or validated reduction equivalent dose [RED]) depends on the
disinfection strategy of the individual UV installation. The State, utility, and designer must
decide the log inactivation requirements for a target pathogen. Once this information is known,
the UV dose can be established (section 3.1.1). For this PER example, a UV dose of 40 mJ/cm2
is recommended to achieve 2-log Cryptosporidium inactivation and is used for UV disinfection
pre-design purposes.
A 12-month pilot study was conducted to assess the long-term disinfection efficiency and
operation and maintenance (O&M) issues. The study results indicated that lamp fouling and
power quality issues should not be a concern for the facility (Mackey et al. 2001).
UV Disinfection Guidance Manual
Proposal Draft
K-3
June 2003
-------
Appendix K. Preliminary Engineering Report
K.3 UV Installation Equipment
General information on UV reactors and the types of reactor configurations used for
water treatment is provided in this section.
K.3.1 UV Lamp Types
For the flowrates associated with the WTP applications in this example, the number of
lamps needed for a low-pressure (LP) reactor would be excessive, so consideration is limited to
low pressure high output (LPHO) and medium pressure (MP) lamps. The general relative
characteristics of each of these lamp types are listed in Table K.2. The ratio of number of lamps
needed to achieve equivalent RED for LPHO lamps as compared to MP lamps is on the order of
6:1.
Table K.2 Relative Characteristics of LPHO and MP Lamps
Lamp Power Output
Power Efficiency
Number of Lamps Needed
Operating Temperature (°C)
Typical Lamp Life (hours)
LPHO
Low
High
High
130-200
8,000-12,000
MP
High
Low
Low
600 - 900
3,000 - 8,000
K.3.2 UV Reactor Configuration
UV installations can be designed around openrchannel or closed-vessel configurations.
In keeping with the content of this guidance manual and the general trend of the drinking water
industry,.the conceptual designs developed herein are limited to MP and LPHO closed-vessel
reactors.
K.4 UV Reactor Description
WTP.
This section contains information on the UV reactor design criteria for disinfection at the
K.4.1 General UV Reactor Description
Each UV reactor for the WTP should include appropriate control and electrical cabinets
and an off-line chemical cleaning (OCC) system or an on-line mechanical cleaning (OMC)
system for the lamps. The cleaning systems should allow for the removal of organic and
inorganic foulants that have accumulated on the surfaces of the lamp sleeves.
UV Disinfection Guidance Manual
Proposal Draft
K-4
June 2003
-------
Appendix K. Preliminary Engineering Report
K.4.2 Process Control
For disinfection of drinking water, the ability of the UV reactor to deliver the design
RED of 40 mJ/cm2 depends on the flowrate, feed water UVT, and UV intensity. UV intensity is
subject to lamp aging, lamp sleeve cleanliness, and water quality (mainly water UVT). The UV
reactor should be designed to deliver the appropriate dose of UV light to the process flow based
on predetermined maximum flowrate and minimum water quality parameter setpoints with an
appropriate factor of safety (see Chapter 3).
UV intensity sensors in each UV reactor should provide continuous performance
verification of the reactor during operation. In case of lamp failure, the UV reactor
programmable logic controller (PLC) should be programmed to either replace one row of lamps
with another row that was off, or turn the reactor off after replacing it with a stand-by reactor.
(Note that Alternative 1, described in section K.5.1, involves placing a UV reactor on each filter
effluent pipe, therefore stand-by reactors for individual filter installations are not provided). The
failed lamp can then be replaced with minimal interruption of UV reactor operation.
In case of incorrect operation of lamps or low level of UV intensity, the PLC should
display a warning to indicate to the operator that cleaning of the reactor should be performed.
The operator initiates the cleaning of any reactor through the local human machine interface
(HMI). After selection, the UV reactor PLC turns on the stand-by reactor. Then, the PLC closes
the inlet and outlet valves and isolates the reactor to be cleaned.
K.4.3 Expected UV Reactor Maintenance
Although maintenance methods are installation and site-specific, some general
maintenance tasks have been developed and are briefly described in this section. As the UV
reactor represents a critical disinfection process, preventative maintenance should be carried out
on a routine basis to ensure that UV reactors reliably provide the specified dose (40 mJ/cm2).
Inadequate cleaning is a common cause of underdosing in UV reactors. The lamp sleeves should
be cleaned regularly by OMC or periodic OCC, and manually cleaned periodically to supplement
automatic cleaning. The cleaning frequency is dependent on the water quality. Chemical
cleaning is most commonly done with dilute citric or phosphoric acid.
The effective life of the UV lamps depends on the minimum UV dose. The UV lamps
should be replaced either at the end of their expected lifetime or following failure. Generally,
UV lamps are replaced when the intensity has dropped to 70 percent of the original new-lamp
intensity (following cleaning of the chamber). This typically occurs after about 8,000 to 12,000
hours (approximately 300 to 500 days) of operation for LPHO lamps and about 3,000 to 8,000
hours (approximately 100 to 300 days) for MP lamps. The front panel of the enclosures
indicates the cumulative hours each lamp has operated. The lamp run time display will facilitate
monitoring of lamp replacement needs.
UV Disinfection Guidance Manual
Proposal Draft
K-5
June 2003
-------
Appendix K. Preliminary Engineering Report
K.4.4 Power Needs
UV reactor power needs to vary depending on the type of equipment that is installed and
UVT of the water being disinfected. The LPHO reactors used for this pre-design have power
requirements of approximately 20 kW for treating 20 mgd. The MP UV reactors require about
130 kW for treating 20 mgd. As indicated in Table K.3 through K.5, additional power would be
necessary to allow for future expansion of the UV facilities.
K.5 Site Plans and Facility Layouts
The preferred process location for a UV installation at a WTP is downstream of the filters
and upstream of the high-service pumps (section 3.1.2). At the example WTP, there are three
viable alternatives for the UV installation downstream of the filters:
. Alternative 1 - Filter Gallery
• Alternative 2 - Existing Chemical Room
• Alternative 3 - New Building
There are eight granular media filters at the WTP. Alternative 1 involves placing one UV
reactor on the effluent pipe of each filter in the filter gallery between the filters arid clearwell.
Alternative 2 is to construct the UV installation in a chemical room in the WTP between the low
service pumps and the reservoir. Alternative 3 involves constructing a new building between the
low service pumps and the reservoir to house the UV reactors. Figure K.I presents a portion of
the plants hydraulic profile and indicates the vertical locations of the three viable alternatives.
The construction requirements and preliminary drawings for each alternative are
illustrated and described in the following section, along with preliminary design criteria. Costs
for the three alternatives are also compared. The preliminary site-specific design criteria are
provided for example purposes only. Application-specific design criteria should,be provided by
the UV manufacturer for each individual UV disinfection implementation project on a case-by-
case basis.
K.5.1 Alternative 1 - Filter Gallery
In Alternative 1, one 3 mgd UV reactor is installed on the discharge pipe of each of the
eight filters, as shown in Figure K.2. The UV reactors would be installed below the hydraulic
grade-line (HGL) of the existing clearwell to ensure constant submergence.(section 3.1.6.1).
Flow through the UV reactors is by gravity from the filters into the clearwell. During filter
backwashing and filter-to-waste cycles, valves located at the influent of each UV reactor can be
closed to keep the reactor flooded while it is taken off-line.
Compared to the other alternatives, construction of Alternative 1 is the simplest.
Construction would include lowering the level in the clearwell to below the filter discharge
UV Disinfection Guidance Manual - K-6 June 2003
Proposal Draft
-------
Appendix K. Preliminary Engineering Report
pipes, then taking each filter off-line individually to install the new piping, valves, and UV
reactor. This would preclude significant disruption of plant operation during construction.
Figure K.1 Portion of the WTP Hydraulic Profile and Alternative Locations for UV
Implementation (Carollo Engineers 2001)
UV Disinfection Guidance Manual
Proposal Draft
K-7
June 2003
-------
^
maintain water
K-8
June 2003
-------
Appendix K. Preliminary Engineering Report
adversely affect the filter performance. Furthermore, in the event of a UV reactor shutdown, the
filter associated with that UV reactor shutdown would also need to be removed from service.
The preliminary design criteria for Alternative 1 are provided in Table K.3. Due to the
large size of the LPHO reactors, they will not fit into the filter gallery. Therefore, only the MP
reactors are considered for Alternative 1.
Table K.3 Preliminary Design Criteria - Alternative 1 - MP UV Reactors
Description
Treatment plant capacities
Flowrate
Water quality
DVT In a 10mm quartz cell @ 254 nm
Ultraviolet reactors
Type of reactors: medium-pressure
Number of reactors
Number of banks per reactor
Number of lamps per bank
Total number of lamps per reactor
Input power per lamp
Total operating electric load
Total installed electric load
Headless through reactor (at current and future flowrates)
Approximate dimensions of each UV reactor
Length .
Width
Height
Flanges diameter
Unit
mgd
%UVT
No
No.
No.
No.
W
kW
kW
Inches
Inches
Inches
Inches
Inches
Criteria
Current
20
93
8
2
4
8
2000
128
128
12
22
36
26
12
Future .
40
93
16
2
. 4
8
2000
256
256
36
50
36
26
12
As stated previously, the eight UV reactors listed in TabJe K.3 are designed for a
maximum capacity of 3 mgd each. This design is based on the assumption that one UV reactor
would be taken off-line periodically during a filter backwash cycle. The WTP must be able to
treat 20 mgd with one filter out of service, so the remaining seven UV reactors would need to be
able to disinfect the maximum plant flow. This arrangement also provides reactor redundancy.
If one UV reactor were taken out of service, the associated filter would also be taken out of
service.
In this example, future plant expansions needed to be taken into account. For the present
analysis, provisions are made so that future expansion of the UV installation to an ultimate flow
of 40 mgd will be possible. If the filter capacities can be expanded to 40 mgd, the UV
installation expansion will necessitate placing two UV reactors (16 total) in series along each
filter effluent pipe. During the construction of the initial design for 20 mgd, adequate space and
UV Disinfection Guidance Manual K-9 June 2003
Proposal Draft
\
-------
Appendix K. Preliminary Engineering Report
mechanical layouts would be provided for the addition of a second UV reactor. For example, a
section of pipe located at the outlet of the UV reactor for the 20 mgd design could be designed
for easy removal and installation of a second UV reactor.
K.5.2 Alternative 2 - Existing Chemical Room
Alternative 2 consists of installing three 10 mgd reactors (2 operational + 1 stand-by) in
an existing chemical room in the WTP, as shown in Figure K.3. This alternative necessitates
placing the UV reactors above the existing HGL of the plant. The low-lift clearwell pumps
provide the head through the UV reactors. (It is generally more advantageous to place the UV
reactors below the HGL. However, due to the space constraints at the example WTP, and to
provide an example of issues that may arise during design, this option is discussed.)
Figure K.3 Alternative 2 - Existing Chemical Room with UV Reactors
(Caroilo Engineers 2001)
NORTH SHORE WTP - PLAN
In theory, an outlet weir structure would be a viable option to raise the HGL of the plant
to ensure constant submergence of the UV reactors (section 3.1.6.1). In this case, there is not
enough space for such a structure. To ensure constant submergence of the UV reactors, a
vertical pipe at the outlet header would maintain the water level at an elevation above the top of
UV Disinfection Guidance Manual
Proposal Draft
K-10
June 2003
-------
Appendix K. Preliminary Engineering Report
the UV reactors. Air vacuum valves would be installed on the inlet and the effluent vertical pipe
to counteract siphon effects on the UV reactors. The discharge from the effluent header would
then flow by gravity to the reservoirs. The hydraulic design of the inlet and outlet channels
provides a continuous equal flow split between the reactors (section 3.3.1.2).
The construction needs of Alternative 2 are more difficult than Alternative 1. A 36-inch
finished water pipe from the low head pumps would need to be taken out-of-service long enough
to cut the pipe and tie-in a new section with fittings and valves for connection of the new UV
reactors. The existing equipment would need to be moved to alternate locations to accommodate
the new large piping and UV reactors, and the floor would need to be cut to provide clearance .
for the piping to and from the lower floor.
Other issues with Alternative 2 are that space constraints in the chemical room, possible
structural upgrades of the building, and raising the HGL of the plant. To allow for adequate
space for maintenance, piping, and instrumentation, the existing chemical equipment in the room
would need to be removed and reinstalled elsewhere in the plant. Since this is a second level
room, a detailed structural analysis would need to be completed to ensure the floor is able to
withstand the load of the UV installation. If structural upgrades are needed, they could prove to
be expensive and difficult to design and construct. Installing the UV reactors in the second level
room above the HGL would significantly increase the total dynamic head (TDH) placed on the
low-lift clearwell pumps. Therefore, pump upgrades would be necessary to overcome the
additional headloss of the UV installation.
K.5.3 Alternative 3 - New Building
Alternative 3 includes constructing a new building to house three 10 mgd UV reactors (2
operational for 20 mgd + 1 standby) and related equipment.' The building layout and UV design
for Alternative 3 is presented in Figures K.4 and K.5. The 36-inch finished water line from the
low-service clearwell pumps would be modified to provide flow to the UV reactors in the new
building. The new facility would include a two-level structure to house mechanical and
electrical equipment and large diameter piping to convey the filtered water through the UV
reactors. The UV reactors would be installed in the basement of the new building below the
HGL of the plant to ensure constant submergence. The hydraulic design of the inlet and outlet
channels would provide an equal flow split between the reactors, and the discharge would flow
under pressure to the reservoirs. <•
The UV building design and mechanical piping shown in Figures K.4 and K.5 are for
preliminary design and cost estimates only. If this option were selected, the building size and
configuration for this alternative would need to be evaluated in more detail during the design
phase and adjusted as necessary, depending on the final UV reactors selected to be used.
Construction for Alternative 3 would be the most involved of the three options because it
would include excavation and construction of a new building. Besides the building construction,
the project would involve tying into the existing 36-inch finished water pipe in two locations
below grade, and modifying site amenities such as pavement and landscaping. In addition, the
new building would also need new power, control, and security systems as well as plumbing,
HVAC, etc.
UV Disinfection Guidance Manual K-ll . June 2003
Proposal Draft
-------
Appendix K. Preliminary Engineering Report
Figure K.4 Alternative 3 - New Building with UV Reactors - Plan View
(Carollo Engineers 2001)
' 1!'.
ij \
'• >
• 1
•y - t •«•
V v r-'1*- "' s.':'v"
«
s' «"
;.
-
s"
V
•s-.y
a' V - - *
*!• ' • i
UV Disinfection Guidance Manual
Proposal Draft
K-12
June 2003
-------
Appendix K. Preliminary Engineering Report
Figure K.5 Alternative 3 - New Building with UV Reactors - Section View
(Carollo Engineers 2001)
s
1 CTttUUM KvUTC —S *
•*r*s*| l**M '(
H
L^^^ll^n^^ITi1 : -r
SECTION
L~ XX *Hi'JYPiCAt .SECTION
-£;:.: •'«'
k
The preliminary design criteria for Alternatives 2 and 3 using the MP UV reactors are
presented in Table K.4.
UV Disinfection Guidance Manual
Proposal Draft
K-13
June 2003
-------
Appendix K. Preliminary Engineering Report
Table K.4 Preliminary Design Criteria - Alternatives 2 and 3 - MP UV Reactors
Description
Treatment plant capacities
Flowrate
Water quality
UVT in a 10mm quartz cell at 254 nm
Ultraviolet reactors
Type of reactors: medium-pressure
Number of reactors
Number of banks per reactor
Number of lamps per bank
Total number of lamps per reactor
Input power per lamp
Total operating electric load
Total installed electric load
Headless through reactor (at current & future flows)
Approximate dimensions of each UV reactor
Length
Width
Height
Flanges diameter
Unit
mgd
%UVT
No. (Duty •*• Standby)
No.
No.
No.
W
kW
kW
Inches
Inches
Inches
Inches
Inches
Criteria
Current
20.
93
2+1
2
8
16
4000
128
192
10
48
49
41
30
Future
40
93
2 + 1
2
8
16
4500
144
216
48
48
49
41
30
The three MP UV reactors selected have design capacities of 10 mgd each. Three 10
mgd UV reactors for the 20 mgd design provide one stand-by reactor in the event of a
malfunction, cleaning, or maintenance of one UV reactor.
Note that for expansion of the UV installation using the MP reactors given in Table K.4,
the size and number of UV reactors remains constant. In order to provide extra lamp intensity to
meet dose requirements at the ultimate flow, 4000 W lamps would replaced with 4500 W lamps.
(Note that re-validation of the reactors with the 4500 W lamps would be required (40
CFR141.729(d)).
The preliminary design criteria using the LPHO reactors for Alternatives 2 and 3 is
provided in Table K.5.
UV Disinfection Guidance Manual
Proposal Draft
K-14
June 2003
-------
Appendix K. Preliminary Engineering Report
Table K.5 Preliminary Design Criteria - Alternatives 2 and 3 - LPHO UV Reactors
Description
Treatment plant design capacities
Plant fiowrate
Water quality
DVT in a 10mm quartz cell at 254 nm
Ultraviolet reactors
Type of reactors: Low-Pressure High-Output
Number of reactors
Number of rows per reactor111
Number of rows with lamps installed
Number of lamps per row
Total number of lamps per reactor
Input power per lamp
Total operating electric load
Total installed electric load
Headloss through reactor (at current & future flows)
Approximate dimensions of each UV reactor
Length(Z>
Width
Height
Flanges diameter
Unit
mgd
%UVT
No (Duty + Standby)
No.
No.
No.
No.
W
kW
kW
Inches
.
Inches
Inches
Inches
Inches
Criteria
Current
20
•
93
2+1
5
4
12
48
200
19.2
28.8
24
110
51
100
32
Future
40
93
2+1
9
8
12
96
200
38.4
57.6
35
144
51
100
32
1 Preliminary design assumes one spare row in addition to current flow demand requirements for installation of lamps in
the future.
2 Length varies depending on the number of rows installed.
The expansion from 20 mgd to 40 mgd using the LPHO UV reactors would be
accomplished by adding additional rows of lamps to the reactor. The UV manufacturers would
oversize the reactor and additional rows of lamps could be inserted as needed for increasing flow
capacities. However, UV reactor validation would need to be done both with and without the
additional rows for the maximum and ultimate flow conditions.
Alternatively, the UV installation could be sized to allow additional UV reactors to be
installed for expansion. Initially, three 10 mgd reactors would be installed for a capacity of 20
mgd (2 operational and 1 standby). Space would be provided to install two additional reactors in
the future for a capacity of 40 mgd (4 operational and 1 standby).
These examples of UV installation expansion alternatives provide various options to the
designer. To confidently design for future UV installation expansions, it will be critical to have
an accurate flow projection and adequate space for the UV installation expansion. In addition to
the UV reactors needed for an expansion, mechanical piping, controls, instrumentation, and
wiring would need to be considered during the preliminary engineering phase. Furthermore, the
designer should work closely with the UV manufacturer to decide on an expansion plan that has
been proven to work effectively and efficiently for the specific UV installation design.
UV Disinfection Guidance Manual K-15 June 2003
Proposal Draft
-------
Appendix K. Preliminary Engineering Report
K.6 Preliminary Capital and O&M Cost Estimates
The preliminary capital, operational, and maintenance costs for each alternative are
summarized in this section. Estimated costs presented for Alternative 1 are based solely on the
MP design. The LPHO UV reactors used for comparison here would not fit into the filter gallery
and so was not considered for that alternative.
K.6.1 Capital Cost Estimate Summary
The estimated capital improvement costs for each alternative are summarized in Table
K.6. Total project cost includes UV reactors, construction cost, engineering services, and a 20
percent estimating contingency.
>
The cost estimates presented for Alternative 1 in Table K.6 are based on using eight MP
UV reactors. The costs presented for Alternatives 2 and 3 were developed around using three
MP and LPHO UV reactors. The equipment cost for installing the MP UV reactors for all three
Alternatives is higher than installing LPHO UV reactors for Alternatives 2 and 3. However, due
to the relatively simple construction details associated with Alternative 1, the total project cost is
considerably lower than Alternatives 2 and 3, which require significant construction provisions.
A comparison of these alternatives is provided in section K.7.
Table K.6 Preliminary Capital Cost Estimates
Alternative 1- Filter Gallery (MP)
Alternative 2- Chemical Room (MP)
Alternative 2- Chemical Room (LPHO)
Alternative 3- New Building (MP)
Alternative 3- New Building (LPHO)
UV Reactor Cost
$531,000
$556,000
$450,000
$556,000
$450,000
Total Project
Cost
$1,900,000
$2,400,000
$2,300,000
$2,800,000
$2,700.000
Annualized
Capital Cost1
$166,000
$209,000
$201,000
$244,000
$235,000
Annualized costs calculated at 6 percent interest for 20 years.
K.6.2 Preliminary Operating and Maintenance Costs
Table K.7 presents a summary of the estimated O&M costs and total annualized costs for
each alternative (four MP reactors in service for Alternative 1, two MP and LPHO UV reactors
are in service for Alternatives 2 and 3). Detailed O&M costs for an average flow of 10 mgd are
provided for each alternative and UV reactors in Tables K.8 and K.9.
UV Disinfection Guidance Manual
Proposal Draft
K-16
June 2003
-------
Appendix K. Preliminary Engineering Report
Table K.7 Estimated UV Disinfection Costs
Alternative 1
Alternative 2 (MP)
Alternative 2 (LPHO)
Alternative 3 (MP)
Alternative 3 (LPHO)
Annul O&M
Cost1
$72,000
$111,000
$36,000
111,000
$36,000
Annualized
Capital Cost2
$166,000
$209,000
$201,000
$244,000
$235,000
Total Annual
Cost
$238,000
$320,000
$237,000
$355,000
$271,000
Annual
Difference3
$1,000
$83,000
-0-
$118,000
$34,000
1. Costs for UV intensity sensor calibrations, lamp sleeve, ballast and sensor replacement are not included.
2. Annualized costs calculated at 6 percent interest for 20 years.
3. Relative to the least expensive alternative (Alternative 2-LPHO).
Table K.8 MP UV Installation O&M Cost Estimates
for Alternatives 1, 2 and 3
O&M
Alt. 1
Costs
Alts. 2 & 3
Average Plant Flow 1 0 mgd
1 - Power Consumption
Annual power consumption of lamps in kWh
Price of electricity ($/kWh)
Annual Expenses ($)
530,155
0.10
53,015
883.5921
0.10
88,359
2 - Consumables
Lamp replacement # operating
# replaced / yr
Annual Expenses ($)
32 $/Lamp 500 (#1) 17,500
600(#2&3)
35
17,500
21,000
21,000
3 - Labor
Lamps # replaced /yr 35
1 5 min / lamp
Cleaning 1 ctngs / yr / reactor 3
3 hrs / cleaning
Total
Annual Expenses ($)
TOTAL COSTS
1 - Power Consumption
2 - Consumables
3 - Labor
4 - Chemicals
Total Annual Costs
COSTS PER WIG TREATED
Costs per MG Treated
Time (hr) 8.8
Time (hr) 9.0
Time(hr) 17.8 $/hr 65
1,153
53.015
17,500
1,153
100
71,770
$/MG 20.00
65
1,153
88,359
21,000
1,153
100
110,610
30.00
Alternative 1 utilizes eight smaller UV reactors while Alternatives 2 and 3 utilize 2 large reactors. At an average flow of
10 mgd, the large UV reactors operate at the lowest possible setting, which is higher than required at 10mgd.
UV Disinfection Guidance Manual
Proposal Draft
' K-17
June 2003
-------
Appendix K. Preliminary Engineering Report
Table K.9 LPHO UV Installation O&M Cost Estimates
for Alternatives 2 and 3
O&M Costs
Average Plant Flow 1 0 mgd
1 • Power Consumption
Annual power consumption of lamps in kWh
Price of electricity ($/kWh)
Annual Expenses ($)
168,303
0.10
16,830
2 - Consumables
Lamp replacement # operating
# replaced /yr
Annual Expenses ($)
96 $/Lamp 150 12,600
84
12,600
3 - Labor
Lamps # replaced / yr 84
15 min/lamp
Cleaning 1 clngs / yr / reactor 24
3 hrs / cleaning
Total
Annual Expenses ($)
TOTAL COSTS
1 - Power Consumption
2-- Consumables '
3 - Labor
4 - Chemicals
Total Annual Costs
COSTS PER MG TREATED
Time(hr) 21.0
Time (hr) 72.0
Time(hr) 93.0 $/hr 65
6.045
16,830
- , 12,600
6,045
600
36,070
$/MG 10.00
Power consumption represents the majority of the operational cost. The power cost used
for calculation of the annual O&M costs was $0.10/kWh. As expected, the MP reactors have
considerably higher power costs associated with their operation than the LPHO reactors (Tables
K.8 and K.9).
Lamp replacement costs are also significant. Cost of lamp replacement is based on an
estimated lamp life of 10,000 hours and $150/lamp equipment cost for the LPHO reactors. The
MP estimated lamp life is 8,000 hours with a lamp replacement cost of $500/lamp for
Alternative 1 and a lamp cost of $600/lamp for Alternatives 2 and 3.
Estimated labor needs range from 18 hours for the MP UV installation to 93 hours per
year for the LPHO UV installation. Labor estimates are based on lamp replacement at four
lamps per hour and three hours per cleaning.
UV Disinfection Guidance Manual K-18 June 2003
Proposal Draft
-------
Appendix K. Preliminary Engineering Report
K.7 Summary of Alternatives and their Advantages and Disadvantages
A comparison of the three alternatives is presented in this section. The alternative
comparisons are based on cost, feasibility of construction, and ease of maintenance. The
advantages and disadvantages of each alternative are summarized in Table K.10.
Table K.10 Advantages and Disadvantages of Each Alternative
Alternative 1 - Filter Gallery
New building not necessary
Below existing hydraulic grade tine (HGL)
Relatively simple construction
Likely no plant down-time for construction
Lowest capital cost
Alternative 2 - Chemical Building
• New building not necessary
• Main floor access
Alternative 3 - New Building
Ample space for UV reactors and controls
UV reactors placed below the plant HGL
Room for future expansion
Flexibility in UV installation design options
Custom designed space for UV reactors
• Damp during periods of the year
• Needs protective cabinet for each UV reactor
• Tight quarters for construction and maintenance
• Less manufacturer flexibility
• Must take a filter off-line for UV reactor
maintenance
• Does not accommodate expansion easily .
• No redundancy at maximum flow
• UV reactors above the plant HGL
• Difficult construction constraints
• Uncertainty associated with structural upgrades
• Very limited space for relocation of existing
. Chemical equipment to other parts of the plant
• Low lift pump upgrades necessary
• Highest capital and total project cost
• Necessitates a new building
• Longer construction schedule
K.8 Conclusions and Recommendations
Alternative 1, which involves the installation of the UV reactors in the filter gallery, is
the least expensive option. However, there are concerns about the moisture associated with the
location that may adversely affect the performance of the UV reactors and cause maintenance
problems. Servicing and maintaining the UV reactors in the filter gallery and installing the
necessary control panels might be problematic based on space constraints. In addition, to expand
the capacity of the UV installation to accommodate the ultimate flow (40 mgd), the size of the
pipes connecting the UV reactors to the filter effluent pipes would have to be increased and an
additional UV reactor would have to be installed. Furthermore, Alternative 1 does not provide
adequate redundancy at the ultimate flow.
Alternative 2, retrofitting the existing chemical feed room has the advantage of not
requiring a new building. However the disadvantages of this option include the space
limitations, associated pump upgrade needs and the need to find a new location for the chemical
feed equipment currently housed in that room.
UV Disinfection Guidance Manual
Proposal Draft
K-19
June 2003
-------
Appendix K. Preliminary Engineering Report
Alternative 3, constructing a building addition to the WTP, is the most expensive
alternative, but this option has some important advantages over Alternatives 1 and 2. The new
building would offer flexibility in design options. The design would not be limited to using the
MP or LPHO reactors; any UV reactors could be accommodated. There would be room for
future expansion of the UV installation, if necessary, and ample space would be provided for
mechanical and instrument layouts. The UV reactors would be installed below the existing
hydraulic grade line of the plant to ensure submergence of the reactors.
Although Alternative 3 has some distinct advantages over Alternatives 1 and 2, the
capital cost is significantly higher, due to the cost of the new building and appurtenances.
Alternative 1 is the most economical alternative, but does not accommodate expansion easily and
provides no redundancy at maximum flow. The disadvantages of Alternative 2, including lack
of space for the existing chemical equipment, make this alternative the least desirable.
Given the above discussion, and based on both economical and non-economical criteria
for comparison in this example, including anticipated future expansion needs, the ranking of the
alternatives from most desirable to least desirable is as follows:
. Most desirable - Alternative 3, New Building
» Next best option - Alternative 1, Filter Gallery
• Least desirable - Alternative 2, Existing Chemical Room
K.9 References -
Carollo Engineers. 2001. Weber basin water treatment plant No. 3 expansion. Layton, Utah.
Mackey, E.D., R.S. Gushing, and G.F. Crozes. 2001. Practical Aspects of UV Disinfection.
Denver, Colo.: AWWA Research Foundation.
UV Disinfection Guidance Manual K-20 June 2003
Proposal Draft
-------
Appendix L. Regulatory Timeline
The purpose of this appendix is to provide utilities with a timeline (Figure L.I) to assist
in planning and implementation of tasks to achieve compliance with the Long-Term 2 Enhanced
Surface Water Treatment Rule (LT2ESWTR). The timelines present the important tasks that
utilities are likely to complete; however, the tasks and their duration will change based on utility-
specific priorities and constraints.
Tasks have been divided into two general categories: regulatory and engineering.
Compliance dates and resulting planning activities are based on utility size (i.e., systems serving
fewer than 10,000 persons or systems serving 10,000 or more persons).
L.1 Regulatory Tasks
Regulatory tasks and milestones include key dates in the regulatory schedule such as
monitoring requirements and compliance dates.
L.1.1 Cryptosporidium Monitoring
One of the key provisions of the LT2ESWTR is the requirement to conduct monitoring to
determine Cryptosporidium removal/inactivation requirements (40 CFR 141.702). Monitoring
results will be used to determine a "bin classification," which prescribes the Cryptosporidium
inactivation/removal required. More information regarding the monitoring requirements is
available in the Source Water Monitoring Guidance Manual for Public Water Systems for the
LT2ESWTR.
L.1.2 Compliance Deadlines for Cryptosporidium Treatment
For utilities required to provide additional treatment for Cryptosporidium, the compliance
deadline is the date when a utility must have implemented the selected treatment techniques (40
CFR 141.701). Table L.I summarizes the Cryptosporidium treatment compliance deadlines for
the LT2ESWTR.
Table L.1 LT2ESWTR Compliance Schedule Summary1
System Size
Serving 10,000 or more people
Serving fewer than 10,000 people
Compliance Deadline for Systems Making
No Capital Improvements for Compliance
6 years after LT2ESWTR promulgation
8 % years after LT2ESWTR promulgation
1 (40 CFR 141.701)
2 State may grant an additional two years for systems making capital improvements.
UV Disinfection Guidance Manual
Proposal Draft
L-l
June 2003
-------
Appendix L. Regulatory Timeline
L.2 Engineering Tasks and Milestones
Engineering tasks and milestones include tasks that should be completed by a utility to
develop and implement an LT2ESWTR compliance strategy.
L.2.1 Process Evaluation and Planning
Compliance with the LT2ESWTR Cryptosporidium treatment requirements will
necessitate varied levels of process evaluation and planning. After compliance strategy options
have been reviewed (see section 3.1.5) and a decision has been made to implement UV—
disinfection, planning may include one or more of the following activities:
• Engaging the State during planning to ensure the installation of UV disinfection is
approved
' . Conducting disinfection benchmarking and profiling if distribution system total
trihalomethane (TTHM) and five haloacetic acids (HAAS) concentrations are at least
80 percent of the Stage 1 DBPR maximum contaminant levels for TTHM and HAAS
' (40 CFR 141.711-713)
. Developing a capital improvement program that includes the necessary modifications
for LT2ESWTR compliance (i.e., UV disinfection)
« Evaluating and implementing funding alternatives
Utilities are encouraged to seek approval of their LT2ESWTR compliance plan from the
State before implementation of a compliance strategy. This may take several months and can
have a significant impact on the implementation schedule, particularly when the State requires
modifications. Because UV disinfection is a relatively new technology, the State may take
longer to approve UV disinfection or require more significant involvement in the compliance
strategy development.
L.2.2 UV Installation Design
The duration of the facility design phase will be contingent'on a number of •
utility-specific factors, including scope of design (i.e., new facility or retrofit), scale of design
(size of facility), available in-house resources, procurement methods, and validation testing
requirements (discussed in detail in chapters 3 and 4). The design will likely include one or
more of the following tasks:
« Evaluation of equipment and contractor procurement methods
' » UV reactor procurement
• UV installation design
UV Disinfection Guidance Manual L-2 June 2003
Proposal Draft.
-------
Appendix L. Regulatory Timeline
UV reactor validation strategy determination
Many States require final approval of process improvements. As such, utilities should
review the UV installation design and validation strategy with the State. If the State is not
consulted during these phases, additional time may be necessary to receive final approval.
L.2.3 Construction and Startup
The timeline in Figure L.I reflects a construction period of two years for both large and
small utilities. However, the actual duration of construction and startup can vary significantly,
depending on the scope of the project, the significance of the changes to the existing treatment
plant, and other utility specific factors. Utilities should consider these factors during planning
phases and adjust accordingly to ensure regulatory milestones are achieved by the necessary
dates.
L.3 Example Timeline
Figure L, 1 presents example timelines that encompass the regulatory and engineering
tasks discussed in the previous sections. Utilities may have site-specific constraints that may
shorten or extend the duration of the engineering tasks listed; however, regulatory milestones are
not flexible.
UV Disinfection Guidance Manual
Proposal Draft
L-3
June 2003
-------
.1
1
H
b
J
•3
0)
I
.5
"5.
o
O
(0
HI
CM
"a
CO
I
O)
I
11
ft
y*
8
o
IM
I
^
a
§,
S
>
-------
Appendix M. Compliance Forms
This appendix is intended to supplement the monitoring information provided in section
5.4 with examples of monthly compliance report and monitoring log forms that utilities might
use for reporting to the State. (Note, these are only examples; the States may develop their own
compliance forms and require additional monitoring.) The specific monitoring and reporting
requirements for each utility should be confirmed with the State, and the forms should be
modified accordingly. For those utilities with advanced control systems (e.g., Supervisory
Control and Data Acquisition (SCADA)), it may be possible to automatically generate these
reports and compliance forms.
To receive disinfection credit, the Long Term 2 Enhanced Surface Water Treatment Rule
(LT2ESWTR) requires validation testing of UV reactors to demonstrate a set of operating
conditions where the UV reactor will deliver the required dose (40 CFR 141.729 (d)). These
operating conditions must include flowrate, UV intensity, and UV lamp status, and the utility
must monitor these parameters during routine operation to ensure dose delivery. (40 CFR
141.729 (d)). States may specify additional monitoring or reporting requirements. The example
forms presented in this appendix list both required and recommended monitoring parameters
(required parameters are identified with the applicable rule citation).
Table M.1 summarizes the recommended minimum level of monitoring and record
keeping for utilities utilizing UV disinfection. For many of the UV reactor components, the
required or recommended performance level is based on the measurement uncertainty of the
specific equipment that was used when the UV reactor was validated. This uncertainty is used to
determine the validation safety factor and recommended reduction equivalent dose (i.e.,
operational UV dose), as described in section 4.2.
UV Disinfection Guidance Manual
Proposal Draft
M-l
June 2003
-------
Appendix M. Compliance Forms
Table M.1. Summary of Compliance Monitoring and Reporting Activities1'2
Item
Off-
specification/
Validated
Parameters for
UV Dose
Calibration of
UV Intensity
Sensors
Calibration of
UV
Transmittance
(UVT) Monitor
Description
Monitor reactor to
ensure operation
within conditions
validated for
required UV dose
(40CFR141,
Subpart W,
Appendix D).
Calibration checks
compare the duty
sensor to the
reference sensor
and are
recommended at
the power setting
utilized during
normal operation.
It is recommended
that grab samples
be collected to
confirm
performance.
Measured
Parameter
Flowrate, UV
intensity, lamp
status, and other
parameters (e.g.,
UVT) used to
monitor dose.
Percent difference
between duty and
reference sensors
relative to the level
of uncertainty used
in determining the
RED.
(see section C.4.7)
Percent difference
relative to the
manufacturer's
guaranteed
uncertainty.
Recommended
Monitoring
Frequency
Continuously.
Record at least
once every four
.hours (daily for
very small
systems).
Monthly. If a
sensor fails for
three consecutive
months, then the
sensor should be
checked weekly
and the
manufacturer
contacted.
Weekly initially.
Reduced
frequency
following one-
year of
supporting data.
\
Reporting
Frequency
Required monthly
report of off-
specification
operation as a
percent of distributed
flow or operating time
(40CFR141.730).3
Requirements in a
State approved
protocol.
NR.
Section 5.4.2 presents all recommended monitoring activities, including the compliance monitoring shown in this
table.
2 Unless noted in the table with an LT2ESWTR citation, the monitoring is recommended and not required.
3 the reported off-specification value is the percentage of water entering the distribution that was not treated with
UV reactors operating within validated conditions. This is required by the LT2ESWTR (40 CFR 141.730).
NR - No requirement
The LT2ES WTR requires utilities to submit monthly reports to the State (40 CFR
141.730). At a minimum, the reports must detail operating performance during the reporting
period and, specifically, the percent of total distributed volume treated during periods when the
UV reactor(s) was off-specification. An example monthly monitoring form is shown in Table
M.2. Tables M.3 through M.6 present a format that the utility can use to log operating data for
development of the monthly reports. With minor modification, the example forms are applicable
for any of the three control strategies discussed in section 4.3.2.2: UV intensity setpoint, UV
intensity and UV transmittance (UVT) setpoint, and calculated dose.
For those utilities utilizing multiple reactors, the operation of each reactor must be
monitored, recorded, and reported. Requirements for compliance monitoring beyond those
established by the LT2ESWTR and the specific content of the monthly report will be established
by the State and coordinated with all other reporting requirements. Additional information on
UV reactor monitoring and maintenance is provided in Chapter 5.
UV Disinfection Guidance Manual
Proposal Draft
M-2
June 2003
-------
Appendix M. Compliance Forms
Greater detail on each of the example forms is provided below:
• Form M.2 is an example of a summary report that would be completed by the utility
and submitted to the State on a monthly basis.
• Forms M.3A, M.3B, and M.3C are example reference forms for each of the three
control strategies discussed in section 4.3.2.2. These forms would be completed by
the utility based on validation results and then referenced throughout the operation of
the UV installation to confirm compliance.
. Form M.4 is an example operating log that would be completed on a daily basis. The
form would be used to record the operating status of the UV installation and to
estimate the volume of water that was discharged during off-specification operation.
• Form M.5 is an example sensor calibration log. This log would be completed
whenever sensor calibration checks are performed. The tog would be used to record
the results of the calibration testing as well to track any sensor recalibration or repair
work that was completed.
« Form M.6 is an example on-line UVT monitor calibration log. This log would only
be completed by those utilities that have included on-line UVT monitors as part of
their design. The log would be completed whenever UVT monitor calibration checks
are performed. The log would be used to record the results of the calibration testing
as well to track any recalibration or repair work that was completed.
UV Disinfection Guidance Manual
Proposal Draft
M-3
June 2003
-------
S
(0
o
Q.
(0
"Sie
•p £ w
Q) o_ >
°- c§
P»i
&:
£t
1-
#
5
§
•sf
1 >
Z uj
t
S
3
Unit
*r ™
O tfl
5
0)
1
S
S
8
o
Q
>
-------
I
1
<
CO
p
o
o
o
Q.
^
to
Sf
"c
§
1
£
\-
'
o
L. ^5
•2 °0
t- (Q
% c
§ .0
0 o
if
Q V
(A
*i
|i
5 «
11
S o>
.g 5
1
!
§•
1
1
•'"•. S
v/.
. ..f. .
?«
•?';
-F"'
i ^
• i:'
•• ..'"
..it, -
'":••••
•rl'r;.
'V <
'i>''
1l
1
• ,*:
.t>
y
2
:ing Conditions
1
1
;-2
M
e
.2
1
1
O)
I
e
i
>
LL
1
•a
III
S
8
i
T3
'5
O
t3
«a
.«
»>
-------
cs
u
I
'X
O)
3
£
+••
CO
o
o
£
1
IS
0)
Q
;&
10
0)
>
00
e*>
(8
p
o
I
o
1
II
ta
1
O
-------
I
O
u.
ex
E
a
.a
O)
S
o
o
*
£
"5
£
«
O
u
o
o>
2
.w
to
i
£
I
w
(
s
i
^* (H
D> 0
.C C
8 "I
Si
1
1
:"• i'>'i
;f|
'"*%
S
]'&
f/S*
^!
: •""••!
• Vi-j."
' v J.
• T'V:
1
ji 1
«^j
*'•< ,i
:.. <^
Si
S
>;; *-' ^
'•1
/S
!<•>
-!-.-£
fg
;J9
J
"c
o
1
1
1
o
o
•a
5
2
3
75
u
^
'«
c
1
1
1
u.
'
,
,
I
O
e
-------
"E.
I
.*
13
O)
3
1
.2
"5.
E
o
O
W
D)
|
O
•o
CO
t_
o
(0
0>
<0
I
D
1
•• »§z
Q tQ £ 13
•4;'t:
irfc
•II!
fi'ii
Volume
(gal*)
I!
**
•
1* 11 S
li » I |
{! i 1
i| |jj
il li
! !;
! I
•i If
ll 11 I*
f'lll 'il
•llli ill
i! 11 til
!
•I T3
s I
-------
O)
o
0)
o
.2
"5.
o
O
•o
c
(0
o>
Q.
I
1
re
'
O
^
-I
ID
+
~a
jo,
VI
8
% * *
^ - b
Sfi.
' m
Q CQ
iS
i ™
S
«= 1
-S S
1 3,
1 i
s i
I £
*
f l
5 o
£
s
1
8
•
-------
Appendix N. UV Lamp Breakage Issues
. Lamps used in UV reactors typically contain mercury or an amalgam composed of
mercury and another element, such as indium or gallium. Other elements, such as xenon,
cadmium, zinc, and magnesium, are also capable of generating UV light; however, the
temperatures required to volatilize these elements are much higher than to volatilize mercury. In
contrast, mercury has a sufficient vapor pressure at ambient temperatures to provide the optimum
pressure for efficient production of resonance radiation. Moreover, mercury has a low ionization
energy to facilitate starting a lamp (Phillips 1983). In order to provide a cost-efficient lamp
while addressing perceived risks and disposal issues associated with mercury, lamp
manufacturers are continuing to develop ways to reduce the mercury content of lamps without
impacting their efficiency (USEPA 1997b; Walitsky.2001).
The mercury contained within a UV lamp is isolated from exposure to water by a lamp
envelope and surrounding lamp sleeve. In order for mercury to be released into the water, both
the lamp and lamp sleeve must break. For the purposes of this appendix, lamp breakage is
defined as fracture of the lamp sleeve and the lamp envelope. This is further divided into-off-line
and on-line breaks. Off-line breaks occur during handling or maintenance functions when the
lamps are not installed in the reactor. On-line lamp breaks occur while UV reactors are in
operation.
Due to the general public health concern with mercury, this appendix discusses the issues
associated with UV lamps used for drinking water disinfection by addressing potential causes of
lamp breakage, preventive measures, disposal issues, the fate of mercury after release, and
regulatory issues. - •
N.I Off-Line Lamp Breaks
Off-line breaks occur when a lamp breaks during shipping, handling, or storage. These
releases do not pose a hazard to the water consumer but are a concern for operators or employees
in the vicinity of the break.
N.1.1 Potential Causes of Off-Line Lamp Breaks and Corresponding Prevention
Measures
Mercury is sealed in a UV lamp within the lamp envelope; therefore, there is no risk of
mercury exposure from handling an unbroken UV lamp. The UV manufacturer should train
operators in proper handling and maintenance of UV lamps to avoid mishandling and potential
offline breaks. In addition, proper storage procedures will also reduce the potential for lamp
breakage. Lamps should be stored horizontally in individual packaging. Lamps should not be
stacked unpackaged on one another or vertically propped in corners (Dinkloh 200la).
UV Disinfection Guidance Manual N-1 June 2003
Proposal Draft
-------
Appendix N. UV Lamp Breakage Issues
N.1.2 Off-Line Mercury Release Cleanup Procedures
Off-line lamp breaks resulting in a release of mercury can occur; therefore, Standard
Operating Procedures (SOPs) should be developed that describe the procedures for containing
and cleaning the off-line spills. The local poison control center, fire department, or public health
board can assist in the development of SOPs.
Small spills, defined as less than about 0.6 to 2.25 grams (USEPA 1992) or the amount in
a broken thermometer (USEPA 1997a), can be contained and collected with commercially
available mercury spill kits. Mercury and materials used during the cleanup procedure are
regulated as hazardous wastes and should be disposed of properly as described in section N.3.3.
The USEPA Office of Emergency and Remedial Response recommends that "[i]n the event of a
large mercury spill (more than a broken thermometer's worth), immediately evacuate everyone
from the area, seal off the area as well as possible, and call your local authorities for assistance"
(USEPA 1997a). Local authorities can help determine the appropriate response for various spill
sizes to be included in SOPs. Given that the mercury content in a single UV lamp typically
ranges from 0.005 to 0.4 grams (as discussed in section N.4.3), large mercury spill actions would
not be warranted for a single lamp break or multiple lamp breaks that result in release of less
than roughly two grams.
Superfund Amendments and Reauthorization Act (SARA) Title HI regulations address
emergency release, inventory, and release reporting requirements for hazardous materials. The
reportable quantity for mercury spills is one pound (454 grams) as mercury. Based on typical
mercury levels in UV lamps (discussed in section N.4.3), this would necessitate the breakage of
approximately 1,100 medium pressure (MP) lamps and up to 90,000 low pressure (LP) lamps; as
such, spilling more than one pound of mercury is highly unlikely.
N.2 On-Line Lamp Breaks
A recent survey of domestic water and wastewater municipalities, UV lamp
manufacturers, and UV reactor manufacturers identified relatively few instances of on-line lamp
breaks and mercury release (Malley 2001). This section discusses potential causes of lamp
breakage and corresponding prevention measures, followed by a summary of documented
incidents of on-line lamp breaks.
N.2.1 Potential Causes of On-Line Lamp Breaks and Corresponding Prevention
Measures
Lamp breaks can potentially be caused by debris in the water, temperature variations,
exceeding positive or negative pressure limits (water hammer), electrical surges, or improper
maintenance. Lamps may also break as a result of inherent mechanical or physical limitations of
the lamp and improper material selection.
UV Disinfection Guidance Manual N-2 June 2003
Proposal Draft
-------
Appendix N. UV Lamp Breakage Issues
N.2.1.1 Debris
Debris in the water can potentially break the lamp sleeves and lamps. Although the
majority of UV reactors will be installed after the filters in the treatment train, it is possible that
equipment failure upstream may release parts or fragments, such as nuts or bolts. In addition, if
UV disinfection is applied prior to the filters the probability of having debris in the water might
be higher compared to post~filter UV installation. Ground water systems have reported stones or
gravel from wells entering UV reactors and breaking lamps (Malley 2001; Roberts 2000).
" Placement of screens, baffles, or low velocity collection areas upstream of UV reactors or
vertical installation of UV reactors (when applicable) may reduce the risk of debris in the water
from entering the reactor (Caims 2000; Malley 2001, McClean 200Ib). The extent of
containment provided by these safety measures is unknown. Utilities and designers should
determine the applicability of these isolation techniques on a site-specific basis.
N.2.1.2 Loss of Water Flow and Temperature Considerations
UV lamps are designed to operate within a specific temperature range to maximize the
UV light output of the lamp. Without flowing water to cool the lamp, the lamp temperature can
rise to dangerous levels and may break (Dinkloh 2001 a; Malley 2001; Srikanth 2001 a; Srikanth
200Ib). This overheating is more likely to occur with MP than LP lamps (due to lamp operating
temperatures) and occurs much faster in air than stagnant water. Even if upper temperature
levels are not exceeded, after restoration of water flow, the lower temperature water entering the
reactor may cause the lamp sleeve and the lamp to break due to temperature differentials
(Dinkloh 2001a; Malley 2001). In order to prevent lamp breaks, operating procedures should
ensure that the following conditions are met:
j
. Water is flowing through the UV reactor if the UV lamps are energized. •
. The lamps are not energized while the reactor is not flowing full (i.e., no air in the
reactor).
Temperature sensors should be, and typically are, incorporated into the reactor design and
will shut down the reactor before critical temperatures are exceeded (Caims 2000; Dinkloh
2001a; Malley 2001; Srikanth 2001b). Proper hydraulic design is also necessary to ensure that
lamps are submerged at all times during reactor operation. Reactor designs should incorporate
low flow alarms, air relief valves, or other devices to ensure that lamps are operating only when
the reactor is completely flooded and water is flowing. These sensors should be linked to an
alarm and automatic shutoff system (Cairns 2000; Dinkloh 200la; Srikanth 200Ib). Lamp
overheating and temperature differentials could break all the lamps within the affected reactor.
N.2.1.3 Pressure-Related Issues
Hydraulic pressures within the reactor that are not within UV installation operating limits
may also break the lamp sleeve. Although breaking the lamp sleeve does not automatically
break the lamp envelope, the lamp is more vulnerable when its lamp sleeve has been
UV Disinfection Guidance Manual N-3 June 2003
Proposal Draft
-------
Appendix N. UV Lamp Breakage Issues
compromised, potentially allowing the lamp envelope to come into direct contact with the
surrounding water.
Most lamp sleeves are designed to withstand continuous positive pressures of at least 120
pounds per square inch gauge (psig) (Roberts 2000; Aquafine 2001; Dinkloh 200 Ic; Srikanth
200la; Srikanth 200Ib). However, negative gauge pressures below -1.5 psig have been shown to
adversely affect lamp sleeve integrity (Dinkloh 200 Ic). The tolerance level of the lamp sleeve
depends on the quality of the quartz and the thickness and length of the lamp sleeve; therefore,
pressure thresholds vary between lamp sleeves. Positive and negative pressures, such as those
associated with water hammer, that exceed these levels may compromise the integrity of the
lamp sleeve. Manufacturers should provide lamp sleeves with the appropriate material,
thickness, geometry, and seals for the specified pressure and flow ranges of a given UV
installation. Water hammer can affect all UV reactors and break all lamps; therefore, utilities
should perform a surge analysis to determine if water hammer is a potential problem.
N,2,1.4 Procedural Errors
Operation and maintenance training can help prevent lamp breaks during on-line
operations because a lamp damaged by off-line handling or improper maintenance operations
may potentially break under on-line pressure or temperature stresses. For example, a common
procedural error that can occur during lamp replacement is over-tightening compression nuts
when securing the lamp sleeve (Aquafine 2001; Dinkloh 200la; Srikanth 200la; Srikanth 200Ib;
Swaim 2002). Over-tightening can cause a fracture of the lamp sleeve or a leak around the
sleeve or compression nut cavity that may not become apparent until after start-up and operation"
of the UV reactor.
N.2.1.5 UV Reactor Design
The UV reactor manufacturer should design the UV reactor to reduce the possibility of
lamp sleeve and lamp breaks. This subsection describes design problems that may cause lamp
sleeve and/or lamp breakage if not properly addressed.
Electrical Considerations
If the UV installation electrical support system is improperly designed (e.g., inadequate
circuit breakers and ground fault indicator circuits), electrical surges can cause short-circuiting
and lamp socket damage (Srikanth 2001 a; Srikanth 2001b). In addition, system electronics that
can provide voltages that exceed lamp ratings (overdriving lamps) may also result in breaking
the lamp (Malley 2001).
Cleaning Mechanism Considerations
The cleaning mechanism may break the lamp sleeve and lamp envelope if it is not
aligned properly. Although the cleaning mechanism closely surrounds the lamp sleeve for
cleaning, manufacturers should ensure that the mechanism is flexible and able to adjust to minor
misalignment of the lamp sleeves.
UV Disinfection Guidance Manual N-4 * June 2003
Proposal Draft
-------
Appendix N. UV Lamp Breakage Issues
At high lamp temperatures, the cleaning mechanism in some UV reactors may fuse to the
lamp sleeve when not in use. As a result, during the next cleaning event, the lamp sleeve may
crack when the cleaning mechanism is activated or when the cleaning mechanism passes back
over the residual left on the lamp sleeve (Dinkloh 200la). Routine inspection according to
manufacturers' recommendations will help detect problems with the cleaning mechanism before
damage occurs. In some UV reactors, wipers rest away from the lamp sleeve when not in use
and an alarm sounds when the wiper stops along the lamp sleeve.
Thermal Expansion and Contraction
Other potential causes of lamp breaks include improper matching of lamp materials with
respect to thermal expansion characteristics. Compatible materials within the lamp should be
used by the manufacturer to avoid stress and damage that can be caused by thermal expansion
and contraction differences between materials under various operating, shipping, or handling
conditions (Cairns 2000). In addition, improper seal design or lamp envelope swelling may
cause water leaks around the seals that may result in electrical shorts and cracking of lamps
(Cairns 2000).
N.2.1.6 Summary of Potential Causes and Methods of Prevention of On-
U'ne UVLamp Breaks
Table N.I summarizes the potential causes of on-line lamp breaks and provides a brief
description of the preventive measures that UV installation designers and operators can
implement to reduce each risk. There are few documented cases where lamps have been broken
during on-line operations, which are discussed in section N.2.2.
UV Disinfection Guidance Manual N-5 June 2003
Proposal Draft
-------
Appendix N. UV Lamp Breakage Issues
Table N.1 Summary of Potential Causes and
Methods of Prevention of On-Line UV Lamp Breaks
Potential
Cause
Description
Preventive Measure
Debris
Physical impact of debris on
lamp sleeves may cause lamp
breaks.
Installation of screens, baffles, or low
velocity collection areas upstream of UV
reactors or vertical installation of UV
reactors will help prevent debris from
entering the reactor.
Loss of Water
Flow and
Temperature
Considerations
Lamps may overheat and break.
The temperature differential
between stagnant water or air
and flowing water may cause
lamp breaks.
Reactors should always be completely
flooded. Temperature and flow sensors
that are linked to an alarm and automatic
shutoff system can be used to indicate
irregular temperature or flow conditions.
Pressure-
Related
Considerations
Excessive positive or negative
pressures may exceed lamp
sleeve tolerances and break the
lamp sleeve.
A surge analysis should be completed to
determine the occurrence of water
hammer.
Pressure relief valves or other measures
can be used to reduce pressure surges.
Applicable pressure ranges should be
specified for lamp sleeves.
Procedural
Errors
Improper handling or
maintenance may compromise
the integrity of the lamp sleeve
and/or lamp.
Operators and maintenance staff should
be trained by the manufacturer.
UV Reactor
Design
Electrical surges can cause
short-circuiting and lamp socket
damage.
Applying power that exceeds
design rating of lamps can cause
lamps to burst from within.
Adequate circuit breakers/ground fault
indicators should be specified to prevent
damage to the reactor.
Replacement lamps should be electrically
compatible with reactor design.
Misaligned or heat-fused
cleaning mechanism may break
or damage the lamp sleeve and
lamp envelope.
Operators and maintenance staff should
perform routine inspection and
maintenance according to manufacturers'
recommendations.
Thermally incompatible materials
do not allow for expansion and
contraction of lamp components
under required temperature
range.
Designers should specify temperature
ranges likely to be encountered during
shipping, storage, and operation of lamps
to aid the manufacturer in the selection of
thermally compatible materials.
N.2.2 Frequency of On-Line Lamp Breaks
There have been relatively few documented incidents of on-line lamp breaks. As part of
a survey of domestic water and wastewater municipalities, UV lamp manufacturers, and UV
reactor manufacturers, Maliey (2001) identified nine cases of on-line lamp breaks. Both the
lamp sleeve and lamp envelope were damaged in all nine cases, resulting in mercury release
(Table N.2). No cases of on-line failures using LP or low pressure high output (LPHO) lamps
were identified. However, LPHO tamps are relatively new to the UV disinfection market and all
LPHO lamp installations have been operating for 5 years or fewer (Maliey 2001). All nine cases
involved MP lamps. Four of the nine lamp breaks were caused by impacts from stones on lamps
UV Disinfection Guidance Manual N-6 June 2003
Proposal Draft
-------
Appendix N. UV Lamp Breakage Issues
oriented perpendicular to flow. In one of the nine lamp breaks, the applied power exceeded
design rating of lamp (30kW) causing the lamp to burst from within. Differential sleeve heating
resulted in two of the nine documented lamp breaks. The lamps were mounted vertically in the
UV reactor allowing heat to accumulate at the top of the lamp, eventually cracking the sleeve. In
two of the nine instances, operating lamps reached extremely high temperatures (>600 °C) in air
because the reactors lost water flow. When water flow resumed, the cooler water (20 °C) broke
the lamps. Most of these documented cases of lamp failure were the result of design issues that
have been addressed in modern reactor designs. As mentioned previously, temperature and flow
alarms should shut the UV reactor down when the potential for overheating or differential
heating exist.
Another documented instance of MP lamp breakage occurred in a UV-peroxide reactor
designed for well-head treatment of tetrachloroethene-contaminated ground water (Moss 2002a).
The UV reactor was positioned between the ground water extraction pump and distribution
system booster pumps. The 7-foot long MP lamp sleeve sagged and came into contact with the
lamp envelope. The lamp envelope and lamp sleeve broke, releasing mercury to the water in the
reactor. The lamp failure triggered an alarm, shutting down both the ground water extraction and
distribution system booster pumps. Mercury liquid was found settled in the bottom of the
reactor. Water sampling at a nearby fire hydrant detected mercury concentrations below the
maximum contaminant level (MCL) of 2 micrograms per liter (ng/L) (Moss 2002a; Moss
2002b).
European drinking water, utilities have an, extensive history with UV technologies.
Unfortunately, no written documentation of lamp failures was identified; however, two instances
of lamp breakage during UV disinfection of drinking water were noted by European
manufacturers (Roberts 2000; Table N.2). In one instance, a ground water well pump discharged
gravel or stones into the reactor, resulting in a lamp break. A strainer was placed in-line prior to
the reactor to prevent any future instances. The other documented case of a lamp breaking was
due to operator error. A forklift was driven into an operating reactor and physically damaged the
UV reactor. The event activated an alarm and pneumatic valve closure, which contained the
contamination (Roberts 2000). In addition, there was an incident in which equipment debris (a
bolt from the filter underdrain) impacted a lamp sleeve. Although the lamp sleeve was broken,
the lamp envelope remained intact and mercury was not released because of the immediate UV
installation shutdown and prompt operator response (McClean 200la). Table N.2 summarizes
the documented lamp breakages discussed in this section.
UV Disinfection Guidance Manual N-7 June 2003
Proposal Draft
-------
Appendix N. UV Lamp Breakage Issues
Table N.2 Mercury Release Incidents Involving UV Lamp Breaks
Identified
Cause
Number of
Incidents
Description of Incident
Debris
. (4)1 Stones entered the reactors and impacted and broke the lamps.
(1)2 Gravel entered reactor through the booster pump and impacted
and broke the lamp.
Loss of Water
Flow and
Temperature
Considerations
(2V Lamps were left on and allowed to reach high temperatures
(600 °C) in empty non-operating reactors. Restoration of flow
resulted in cooler water (20 °C) breaking the lamps.
Operator Error
(1)3 Forklift collided with on-line reactor resulting in lamp breakage.
Manufacturer
Design
(1)1 Applied power exceeded design rating of lamp (30kW) causing
the lamp to burst from within.
(2)1 Vertical orientation of lamps resulted in differential heating and
eventual cracking of lamp sleeve as surrounding water cooled the
submerged portion of lamp and the exposed portion of the lamp
accumulated heat.
(1)" High operating temperatures resulted in deformation of the lamp
sleeve. The lamp sleeve sagged and on contact with the lamp
envelope, both envelope and lamp sleeve broke.
1 Survey of domestic water, wastewater, and hazardous waste treatment utilities (Malley 2001)
2 European drinking water facilities (Roberts 2000)
3 European brewery (Roberts 2000)
4 UV-peroxide ground water remediation reactor (Moss 2002a)
N.2.3 On-Line Mercury Release Response Plan
On-line lamp breaks are rare occurrences that are preventable with appropriate design and
operation of UV reactors. However, utilities may consider developing a mercury release
response plan for an on-line UV lamp break. The plan may include the following components:
. Site-specific containment measures
• Mercury sampling and compliance monitoring guidelines
• Clean-up procedures
. Reporting requirements
In the event of an on-line lamp failure alarm, the UV reactor should be immediately shut
down and operators should attempt to determine the cause of the alarm. Unfortunately, lamp
failure alarms or sensors cannot typically determine the cause of the alarm, whether it is partial
or complete breakage of the lamp sleeve or lamp envelope (Kolch 2001) or another problem
unrelated to the lamps. Thus, it is recommended that the reactor be taken off-line when
investigating the cause of a lamp failure alarm (Kolch 2001).
UV Disinfection Guidance Manual
Proposal Draft
N-8
June 2003
-------
Appendix N. UV Lamp Breakage Issues
In the event of an on-line lamp break and mercury release, operators should attempt to
isolate the mercury in the reactor or downstream. Utilities may install spring-return actuated
valves with a short closure time on the reactor inlet and outlet piping (McClean 200 Ib) to isolate
the mercury. Given the short residence time of many MP reactors, the outlet-side valve may
need to be located quite a distance downstream so that the valve has time to close and isolate the
mercury upstream. UV installation designers should evaluate valve closure times with respect to
creating water hammer.
Condensed mercury may collect in areas of low water velocity such as the bottom of a
shutdown reactor, sump areas, or a clearwell. In addition, a strainer positioned on the reactor
outlet piping may prevent lamp fragments from entering the water supply system (McClean
200Ib; Srikanth 200la; Srikanth 200Ib). The headloss associated with such measures should be
considered in the hydraulic profile. Designers may also consider installation of drains, vacuum
relief valves, and piping to allow disposal of potentially contaminated water in the reactor to a
waste container or truck.
The extent of containment provided by these safety measures is unknown. Utilities and
designers should determine the applicability of these isolation techniques on a site-specific basis.
Utilities should coordinate with their State primacy agency when developing the
following action items:
« Mercury sampling plan - Sampling procedures may outline sample locations,
sampling frequencies, and analysis methods. Sample locations should be chosen with
consideration of where mercury may settle and to assess the mercury concentrations
potentially reaching the consumer. Sampling frequencies should consider flowrate,
detention time, and travel time to the first potential consumer.
• Site-specific cleanup procedures - Site-specific cleanup procedures should be
incorporated into a utility process hazard analysis (PHA). Issues to consider are
detection and disposal of isolated or condensed mercury, potential disposal or
^ treatment of contaminated water, and cleanup responsibilities (by utility staff or
contracted hazardous materials team).
. Reporting to State - Reporting may include a description of the release, estimated
quantity of release, shutdown or containment procedures, cleanup or disposal
methods, sampling procedures (including sampling locations, frequencies, and
results).
• Public notification requirements, if applicable - Revised public notification
requirements (40 CFR 141.203) outline three tiers of public notification, depending
on the severity of the violation or situation. Exceeding the mercury MCL of 2 ug/L is
classified as a Tier 2 notice, where public notification is required within 30 days,
unless extended to 90 days by the State primacy agency. Public notification
requirements do not specifically address mercury releases due to UV lamp breakage
where the MCL is not exceeded.
UV Disinfection Guidance Manual N-9 June 2003
Proposal Draft
-------
Appendix N. UV Lamp Breakage Issues
N.3 Regulatory Review
This section presents a review of regulations that may apply to the use or breakage of UV
lamps containing mercury in water treatment plants (WTPs).'
N.3.1 Safe Drinking Water Act
The Safe Drinking Water Act (SDWA) established a primary MCL of 2 ng/L for
inorganic mercury (40 CFR 141.62(b)). The required monitoring frequency depends on the
water source and the frequency of detections. Utilities using ground water sources are required
to sample once every 3 years. WTPs using surface water sources are required to sample
annually. If mercury is detected above the MCL in any ground water or surface water utility, the
utility must sample quarterly.
These regulations are independent of the use of UV disinfection at a facility. As
discussed in section N.2.3, utilities should consult with their primacy agencies when developing
a sampling plan for responding to an on-line UV lamp break.
N.3.2 Operator Health and Safety - Exposure Limits
The Mercury Study Report to Congress (USEPA 1997c) provides detailed information on
health effects associated with exposure to elemental mercury and mercury compounds. Mercury
exposure to employees in WTPs falls under the regulatory authority of the Occupational Safety
and Health Administration (OSHA). The exposure limits set by OSHA focus on exposure by
inhalation.
OSHA regulations have established permissible exposure limits (PELs) for mercury
compounds and organo alkyls containing mercury. A PEL is a time weighted average
concentration for an 8-hour workday during a 40-hour work week that is not to be exceeded.
When a PEL is designated as a ceiling level (cPEL), the concentration cannot be exceeded during
any part of the workday. PELs and cPELs are enforceable standards. The National Institute for
Occupational Safety and Health (NIOSH) also publishes Immediately Dangerous to Life or
Health (IDLH) concentrations for a variety of compounds. IDLH concentrations represent the
maximum concentrations that one could escape within 30 minutes without symptoms of
impairment or irreversible health effects. These values are not enforceable, but can be used as
guidance for safety procedures. Table N.3 outlines the PELs, cPELs, and IDLHs for mercury
compounds and organo alkyls containing mercury.
Table N.3 Health and Safety Standards for Mercury Compounds in Air
Compound
Mercury compounds
Organo alkyls containing mercury
PEL (mg-Hg/m3)
NR
0.01
cPEL (mg-Hg/m3)
0.1
0.04
IDLH (mg-Hg/m3)
10
2
NR - not reported.
UV Disinfection Guidance Manual
Proposal Draft
N-10
June 2003
-------
Appendix N. UV Lamp Breakage Issues
In the event of a spill, the volatilization and the resultant concentration of mercury in air
depends on the vapor pressure (0.002 mm Hg; Table N.4), air currents, temperature, surface
area/dispersion of mercury droplets, and time. Calculations using the ideal gas law (PV=nRT)
indicate that these levels may be exceeded if cleanup of the mercury spills does not occur;
however, prompt response and proper cleanup procedures should prevent exposure levels over
these standards.
N.3.3 UV Lamp Disposal Regulations
Lamp manufacturers are required to determine whether their products exhibit the toxicity
characteristic for mercury using a test called the Toxic Characteristic Leaching Procedure
(TCLP, 40 CFR 261). If the TCLP level of a lamp is above the regulatory limit of 0.2 mg/L, the
lamp is regulated as a universal hazardous waste (Universal Waste Rule, 40 CFR 273) under
Subtitle C of the Resource Conservation and Recovery Act (RCRA). As such, these lamps
should be sent to a mercury recycling facility where the mercury is recovered and lamp
components are recycled. Although some mercury lamps do not exceed the TCLP regulatory
level, utilities are encouraged to recycle these lamps to reduce mercury loading to the
environment. Some UV reactor and lamp manufacturers will accept spent or broken lamps for
recycling or proper disposal (Dinkloh 200la; Lienberger 2002; Gump 2002). Alternatively,
utilities should contact their primacy agency for a list of local recycling facilities.
N.4 Additional Factors Affecting Risk
This section provides further information that may be helpful in evaluating risk
associated with on-line lamp breakage. The ultimate fate of mercury after a lamp is broken is
currently unknown but is expected to depend on the following conditions:
. Physical and chemical properties of mercury species in air and water
« Mercury behavior in operating UV lamps
. Quantity of mercury released (type, age, and number of broken lamps)
• Potential mercury reactions in water treatment plants and the distribution system
N.4.1 Physical and Chemical Properties of Mercury
Mercury can exist in three oxidation states: elemental (Hg°), mercurous (Hg+1), and
mercuric (Hg+2). Mercury cycles between oxidation states as a function of the redox conditi
of the surrounding environment and the availability of other reactive compounds.
Elemental mercury is a liquid at ambient temperature and pressure; however, given its
high vapor pressure (Table N.4), elemental mercury is easily vaporized at ambient temperatures.
Other physical and chemical properties of elemental mercury that affect its fate and transport are
outlined in Table N.4.
UV Disinfection Guidance Manual N-11 June 2003
Proposal Draft
-------
Appendix N. UV Lamp Breakage Issues
Table N.4 Physical and Chemical Properties of Elemental Mercury
(Merck & Co., Inc. 1983)
Property
Melting point (°C)
Boiling point (°C)
Density (g/mL at 25 °C)
Solubility (g/L at 25 °C)
Vapor pressure (mm Hg at 25 °C)
Value
-38.87
356.72
13.534
0.061
0.002
Further information regarding mercury solubility in water can
be found in Glewetal. 1971
N.4.2 Mercury Behavior in UV Lamps
It is important to characterize the quantity and form of mercury in an operating lamp
because they represent the starting point for mercury dispersion, speciation, and reaction
chemistry in the water system following a lamp break. However, the quantity and form of
mercury placed in UV lamps typically is considered proprietary information by manufacturers
because these parameters affect the efficiency, operation, and life of the lamp. In general, the
form of mercury contained in a UV lamp is elemental mercury (LP and MP) or a mercury
amalgam (LPHO). An amalgam is an alloy of elemental mercury with another metal (typically
indium in lamp applications) that can be either solid or liquid at room temperature, depending on
the relative proportions of the two metals. In operating lamps, elemental mercury (from pure or
amalgamated mercury) is vaporized in the presence of an inert gas. Vapor phase mercury is
excited and then ionized by the energy transfer from the excited inert gas and the supply of
electrons generated from the applied voltage (Phillips 1983). It is the transition of mercury
electrons from excited state back to ground state that releases energy in the wavelength range of
the UV spectrum.
The concentration of mercury in the vapor phase in LP and LPHO lamps is controlled
predominantly by temperature. Manufacturers of these lamps use different methods to control or
maintain the temperature of the liquid mercury or mercury amalgam to establish the desired
vapor phase mercury concentrations. Methods of controlling the temperature of mercury and,
consequently, the vapor pressure in LP and LPHO include using either a mercury amalgam
attached to the lamp envelope (LPHO only), a cold spot on the lamp wall, or a mercury
condensation chamber located behind each electrode. At typical LP and LPHO lamp operating
temperatures, mercury remains predominantly in the liquid or solid amalgam phase with a small
proportion in the vapor phase.
MP lamps are dosed with elemental mercury liquid. In operating MP lamps, mercury is
primarily present in the vapor phase due to high operating temperatures (600 to 900 °C;
Table 2.1) that cause all liquid elemental mercury to volatilize (Phillips 1983). In order to
control the concentration of vapor phase mercury, manufacturers strictly control the amount of
mercury dosed or added to the lamp during production. This is different from the LP and LPHO
lamps where an excess of mercury is placed in the lamp and only a portion of the elemental
mercury enters the vapor phase. Once elemental mercury enters the vapor phase, mercury
ionization in a MP lamp occurs the same way as in LP or LPHO lamps.
UV Disinfection Guidance Manual N-12 June 2003
Proposal Draft
-------
Appendix N. UV Lamp Breakage Issues
i
The relative proportion of mercury in the liquid/amalgam phase and the vapor phase
becomes important when an operating lamp breaks in water. Vapor phase elemental or ionized
mercury may be released as very fine particles. These particles may more readily dissolve in
water as opposed to condensed liquid or amalgamated mercury that settles in low velocity areas.
In addition to these functional mercury interactions, Altena (2001) reported reactions of
vapor phase mercury with fluorescent lamp components, such as the glass bulb, glass stems,
coatings, and the emission material (electrodes). This process results in the embedding of
mercury in lamp components and the accumulation of mercury-containing deposits, such as
mercury oxide, on the internal lamp envelope surface. Altena (2001) theorized that mercury
reactions with UV lamp components would be comparable to fluorescent lamps. These deposits
represent approximately 2 to 15 percent of the total mercury present in a lamp as calculated from
Altena (2001). After breakage, these deposits are available to dissolve in water; however,
mercury oxide has a low solubility in water (Merck & Co. 1983).
Figure N.I outlines the expected forms of mercury in an operating lamp. Note that all
liquid elemental mercury will volatilize in an operating MP lamp, leaving no mercury in the
liquid phase. Also, amalgams are only used in LPHO lamps.
Figure N.1 Mercury Speciation In Operating UV Lamps
Hg-containing deposits, e.g.
OR
Sonrct rf i7^wr-iio«merciiy
inLPcrMPlSEps
Sauce of mpn-fhia mercury
h LPHO imps
N.4.3 Quantity of Mercury in Lamps
The amount of mercury in a UV reactor is a function of the type of lamp, the number of
lamps in a reactor, and the number of reactors. Mercury content within lamps depends on type
(LP, LPHO, or MP), length, and power rating. Although mercury content data are specific to
manufacturer and lamp, lamps with higher pressures, power ratings, and lengths typically contain
more mercury. Table N.5 summarizes the quantities of elemental mercury dosed into lamps
during manufacturing according to a confidential manufacturer survey and published literature
values.
UV Disinfection Guidance Manual
Proposal Draft
N-13
June 2003
-------
Appendix N. UV Lamp Breakage Issues
Table N.5 Elemental Mercury Content in UV Lamps
Lamp Type
LP
LPHO
MP
Electrical
Power
Rating (W)
15-70
120-260
400
1000
1-25 kW
Mercury Content (mg per lamp)
Phillips (1983)
"a single drop"1
NR
NR
NR
1.4-1 4.5 mg/cm5
Clear etal. (1994)
202
263, 364
75.5
250
NR
•i
Manufacturer Survey
5-50
150
NR
NR .
200-400,
0.3 - 7 mg/cm length,
7.9 mg/cm length
1 Phillips 1983
2 75 W mercury vapor lamp
3 175 W mercury vapor lamp
4 250 W mercury vapor lamp
5 mg per cm of lamp length, reported lamp lengths are 6-300 cm (Primarc Limited 2001)
NR - Not Reported
N.4.4 Quantification of Mercury in Example UV Installations
This section illustrates example calculations of the amount of mercury contained in
hypothetical UV installations. Two UV reactor manufacturers established design parameters for
three treatment flowrates (0.18, 3.5, and 210 million gallons per day (mgd)) with a specified
water quality and design dose (Table N.6). Design parameters included the number of lamps
needed to obtain a dose of 40 mJ/cm2 and the total number of reactors for each of the three
design flows. Calculations assume 50, 150, and 400 mg of mercury per LP, LPHO, and MP
lamp, respectively. Utilities should use site-specific UV installation information to determine
quantities because mercury content varies with lamp type and manufacturer.
UV Disinfection Guidance Manual
Proposal Draft
N-14
June 2003
-------
Appendix N. UV Lamp Breakage Issues
Table N.6 Mercury Quantity in Example UV Installations12
Design Flow
{mgd)
0.18
3,5
210
Average
Flow
(mgd)
0.054
1.4
120
Lamp Type
LP
LPHO
MP
LPHO
MP
LPHO
MP
Average
Number of
Reactors
1
1
1
1
1
6
6
Average Number
of Lamps
"(per reactor)
2
1
1
30
4
72
7
Total Hg in UV
Installation3
(g)
0.1
0.2
0.4
4.5
1.6
64.8
16.8
UV Dose = 40 mJ/cnr
2 Water quality criteria: UVT = 89% (A254 = 0.05 cm'1), Turbidity = 0.1 NTU, Alkalinity = 60 mg/L as CaCO3,
Hardness = 100 mg/L as CaCOa
3 Values given represent the amount of elemental mercury dosed in lamps during manufacturing.
N.4.5 Fate of Mercury After Release
The previous sections define the quantity and form of mercury in an operating lamp and
thus define the starting point for the investigation of the fate of mercury in the water system.
Unfortunately, little documentation exists on the fate of mercury in WTPs or distribution
systems. The few case studies that do exist are 'mainly in the wastewater industry and focus
primarily on removal of influent mercury by the following wastewater treatment processes:
. Primary sedimentation (Lester 1983; Firk 1986; Balogh and Liang 1995; Goldstone et
al. 1990; Oliver and Cosgrove 1974)
. Activated sludge (Gilmour and Bloom 1995; Lester 1983; Chen et al. 1974; and Wu
and Hilger 1985)
• Conventional treatment process (Mugan 1996; Balogh and Liang 1995; Bodaly et al.
1998)
Much of the knowledge about mercury and its potential fate in water systems is derived
from studies performed in natural environments. It is expected that this knowledge of mercury
cycling within the natural environment can be applied to mercury dynamics within a WTP and
distribution system where environmental conditions are largely controlled and remain fairly
stable. However, WTPs employ a number of chemicals that are not typically found in natural
environments. No studies were identified on influence and reaction of mercury with coagulants,
polymers, corrosion inhibitors, ammonia, strong oxidants, and other disinfectants (e.g., chlorine
and ozone). This subsection projects mercury reactions within a WTP and distribution system
based on documented mercury reactions in the natural environment.
UV Disinfection Guidance Manual
Proposal Draft
N-15
June 2003
-------
Appendix N.UV Lamp Breakage Issues
N.4.5.1 Potential Mercury Reactions in Water Treatment Plants and
Distribution System
Liquid phase elemental mercury is considerably denser than water (specific gravity =
13.6; Table N.4) and does not readily dissolve in water. Therefore, liquid elemental mercury and
mercury amalgams may settle in areas of low water velocity, thereby providing an option.for
early containment and removal. For example, in cases where mercury was released from other
water treatment equipment (such as manometers, flow instrumentation, or pump seals), mercury
was found to have condensed and settled in the clearwell. However, the amount of mercury
recovered relative to the amount of mercury released is unknown (Cotton 2002). Kolch (2001)
monitored the mercury concentrations in a 50-liter batch reactor following the destruction of one
LPHO lamp (approximately 150 mg-Hg). Mercury concentrations reached approximately 2.5
ug/L in water. Amalgamated mercury was found settled on the bottom at the reactor (Dinkloh
2001 b). However, it was not reported whether all the 150 mg of mercury present in the
operating lamp was recovered with the amalgam or accounted for in the aqueous phase.
Also, Kolch (2001) did not determine whether the source of aqueous phase mercury was
dissolved mercury from the amalgam or vapor phase mercury present in the lamp prior to when it
was broken. This issue may become important when considering the aqueous behavior of
mercury following a MP lamp break. Mercury in a MP lamp is predominantly in the vapor phase
during operation. It is unknown how the vapor phase mercury will react with the water. Vapor
phase elemental and ionized mercury may become very fine particles when contacting the water
as opposed to liquid or amalgamated mercury that settles in low velocity areas.
Depending on the age of the lamp, mercury-containing deposits, such as mercury oxide,
may accumulate on the inner surface of the lamp envelope in all lamp types. Dissolution of the
deposits would result in additional ionized mercury entering the water. Prompt response to a
lamp break would include removal of lamp fragments; therefore, the compounds on the lamp
fragments should not have the opportunity to enter the WTP.
Once in the water, aqueous (dissolved) elemental and ionized mercury are expected to
cycle between phases and oxidation states as determined by temperature, pH, organic carbon
concentration, minerals and inorganic species, and dissolved oxygen level. The mercurous ion
(Hg2+2) is formed by the oxidation of elemental mercury or the reduction of the mercuric ion
(Hg+2). The mercurous ion is capable of bonding with inorganic constituents; however, it does
not bind with organic compounds. Hg2+2 is rarely stable under typical environmental conditions
and is readily reduced to Hg° or oxidized to Hg .
Inorganic reactions involving the mercuric ion (Hg+2) include binding with inorganic
ligands such as chloride, hydroxide, and sulfide. Considering sulfide is present in anoxic
environments, mercury sulfide (HgS) is not expected to form in a water treatment environment.
Reactions of Hg+2 with chloride and hydroxide resulting in mercuric chloride and mercuric
hydroxide compounds depend on the pH and chloride concentrations (Beckvar et al. 1996).
Additional discrepancies arise in the comparison of mercury reactions and fate in a
drinking water environment versus the natural environment when organic carbon concentrations
and existing microbial populations are considered. The mercuric ion (Hg+2) is the only oxidation
state of mercury capable of association with organic compounds such as phenyl and methyl
UV Disinfection Guidance Manual N-16 June 2003
Proposal Draft
-------
Appendix N. UV Lamp Breakage Issues
groups. The resultant organic compounds, commonly known as methylated mercury, have
different chemical, physical, and toxicological properties than inorganic mercury and offer more
cause for concern due to toxicity and bioaccumulatioh properties (Beckvar et al. 1996).
Methylation of mercury to form methyl and dimethyl mercury is primarily a biological process
involving sulfate-reducing bacteria although it can also occur abiotically. The extent to which
methylatton occurs depends on the availability of Hg+2 and the presence of an appropriate
microbial population. Methylation rates are higher under anoxic conditions, low pH, elevated
temperatures, and high organic matter concentrations (USEPA 1997c). Therefore, even though
Hg may be present in the water column, all of the above factors oppose the occurrence of
methylation in a drinking water environment.
Another divergence of a water treatment environment from the natural environment is the
presence of treatment chemicals such as coagulants, strong oxidants, polymers, corrosion
inhibitors, and ammonia. Seigneur (1994) researched the reaction chemistry of mercury with
inorganic species and strong oxidants such as chlorine and ozone in the aqueous and gas phases
that are present in the atmosphere. Ionized mercury can form inorganic compounds with
chloride and hydroxide ions. Depending on reactant concentrations, these compounds may be
present in the aqueous phase and as solid precipitate. Also, based on reduction oxidation
potentials, it is possible that the oxidation of elemental mercury would occur in the presence of
chlorine and ozone, forming mercury ions and thereby increasing the solubility of elemental
mercury.
1 Further research and investigation is necessary to determine the mechanisms of any
potential mercury reactions. Figure N.2 outlines this preliminary assessment of mercury
speciation and reaction in a drinking water environment.
UV Disinfection Guidance Manual N-17 June 2003
Proposal Draft , "
-------
Appendix N. U V Lamp Breakage Issues
Figure N.2 Potential Reactions of Mercury in a Drinking Water Environment
(Compounds released from a broken lamp are in boldface type.)
AIR'')'" '$
li
-• HtiXfaytoafx- OH^Ci,-;S
'WATER1
w
PIPINGItOR CONTAINMENT STRUCTURES
Mercury (methylated and ionic) sorption to dissolved and paniculate organic matter is
commonly found in natural environments (Beckvar et al. 1996; USEPA 1997c). This
observation was also made in water and wastewater treatment plant studies, where mercury
became incorporated into coagulant floes and activated sludge waste, respectively (Logsdon
1973; Gilmour and Bloom 1995; Lester 1983; Chen et al. 1974; Wu and Hilger 1985).
If UV disinfection of raw water is used and a UV lamp breaks, the mercury could
potentially be removed within the WTP. Logsdon (1973) investigated the efficiency of mercury
removal in conventional WTPs. Bench-scale laboratory tests indicated that inorganic mercury
was removed via coagulation, softening, adsorption on turbidity, powdered activated carbon
(PAC) adsorption, and granular activated carbon (GAC) column adsorption.
N.5 Summary and Conclusions
The risk associated with a mercury release to the water system depends on many factors.
More research is needed to close the knowledge gap that exists regarding the fate of mercury in a
drinking water environment following a UV lamp break. The influence of treatment chemicals
such as oxidants, disinfectants, and coagulants is largely unstudied. Likewise, dispersion and
transport of mercury through a WTP and distribution system has yet to be evaluated. Although
UV Disinfection Guidance Manual
Proposal Draft
N-18
June 2003
-------
Appendix N. UV Lamp Breakage Issues
these issues are, at present time, largely unknown, there are procedures and actions that can be
taken to reduce or mitigate mercury release caused by UV lamp breakage.
For the purposes of this appendix, lamp breakage was defined as fracture of the lamp
sleeve and the lamp envelope. This was further divided into off-line and on-line breaks. Off-line
lamp breaks typically occur during storage or handling and cause small spills (< 2g). Small
spills should be contained, cleaned up, and disposed of properly.
On-line lamp breaks occur while the UV reactor is in operation. There have been
reported incidents of on-line UV lamp breaks associated with impact from debris, loss of water
flow, temperature differentials, faulty electrical equipment and design, and procedural errors.
However, on-line lamp breaks are a rare occurrence and are largely preventable with appropriate
design, operation, maintenance, and operator care. The following engineering and administrative
methods have been proposed that may help prevent UV lamp breakage:
. Screens, baffles, or low velocity collection areas prior to the reactor influent to
prevent entrance of debris
• Temperature and flow sensors and alarms to detect critical conditions and shut the
UV reactors down
• Surge analysis to "determine if water hammer may be a potential problem
• Comprehensive training and maintenance program
i '
In the event of a mercury release, the following engineering controls are additional
precautions that may aid in the containment and collection of mercury:
• Strainers and low velocity collection areas downstream of the reactor
» Isolation valves activated by an alarm to attempt to isolate potentially contaminated
water
The extent of containment and prevention provided by these measures is unknown.
Utilities and designers should consider the applicability of these isolation techniques on a site-
specific basis. Utilities should consult with their State primacy agency in the development of
standard operating procedures, clean-up procedures, and reporting requirements in preparation
for a potential UV lamp break and mercury release. It is recommended that a utility prepare a
mercury release response plan to address these issues. This plan should address compliance with
the SDWA, OSHA health and safety standards, and RCRA universal waste rules. Utilities are
encouraged to recycle or return all mercury-containing lamps to mercury re-generating facilities
or the lamp manufacturer.
U V Disinfection Guidance Manual N-19 June 2003
Proposal Draft
-------
Appendix N. UV Lamp Breakage Issues
N.6 References
Aquafine Corporation. 2001. CSL Series: Installation, Maintenance, and Operation Manual
Part No. 115-1.
Altena, F.W., J.B.J. van Overveld, and H. Oilier. 2001. Technological advances in disinfection
lamps leading to more compact UV sources. Presented at the First International Congress
on Ultraviolet Technologies, Washington, DC, June 14-16.
Balogh, S. and L. Liang. 1995. Mercury pathways in municipal wastewater treatment plants.
Water, Air, and Soil Pollution 80: 1181-1190.
Beckvar, N., J. Field, S..Salazar, and R. Hoff. 1996. Contaminants in aquatic habitats at
hazardous waste site: mercury. Hazardous Materials Response and Assessment Division
of the National Oceanic and Atmospheric Administration Technical Memorandum NOS
ORCA100. Seattle, WA.
Bodaly, R.A., J.W.M. Rudd, and R.J. Flett. 1998. Effect of urban sewage treatment on total and
methyl mercury concentrations in effluents. Biogeochemistry 40: 279-291.
Cairns, B. 2000. Comments on USEPA questions about the risk of using UV disinfection
technologies in the potable water industry. Trojan Technologies Inc., London, Ontario,
Canada, May 24.
Chen, K.Y., C.S. Young and N. Rohatgi. 1974. Trace metals in wastewater effluents. J. Water
. Poll. Control Fed. 46: 2663.
Clear, R., and S. Berman. 1994. Environmental and health aspects of lighting: mercury.
Journal of the Illuminating Engineering Society 138-156.
Cotton, C. 2002. Confidential communication with major water municipality. June.
Dinkloh, L. 200la. Wedeco Ideal Horizons. Personal communication with Jennifer Hafer,
Malcolm Pimie, regarding UV reactors. Tucson AZ, May 10.
Dinkloh, L. 200Ib. Wedeco-Ideal Horizons. Personal communication with Jennifer Hafer,
Malcolm Pirnie, regarding UV reactors. Tucson AZ, August 10.
Dinkloh, L. 200Ic. Wedeco-Ideal Horizons. Email to Christine Cotton regarding UV negative
pressure. Tucson AZ, August 15.
Firk, W. 1986. Heavy metals in sewage and occurring sludges - the balance of three sewage
treatment plants, edited by J.N. Lester, R. Perry and R.M Sterritt. Lisbon: International
Conference Chemicals in the Environment.
UV Disinfection Guidance Manual .N-20 June 2003
Proposal Draft
-------
Appendix N. UV Lamp Breakage Issues
Gilmour, C.C. and N.S. Bloom. 1995. A case study of mercury and methylmercury dynamics in
a Hg-contaminated municipal wastewater treatment plant. Water Air and Soil Pollution
80: 799-803.
Glew, D.N. and D.A. Hames. 1971. Aqueous nonelectrolyte solution: mercury solubility in
. water. Canadian Journal of Chemistry 49: 3114-3118.
Goldstone, M.E., C. Atkinson, P.W.W. Kirk, and J.N. Lester. 1990, The behavior of heavy
metals during wastewater treatment III: mercury and arsenic. The Science of the Total
• Environment 95:271-294.
Gump, D. 2002. Severn Trent. Personal communication with Jennifer Hafer, Malcolm Pirnie,
Inc., regarding UV reactors. Tucson AZ, November 6.
Kolch, A. 2001. Assessing the risk of mercury release from ultraviolet discharge lamps in
drinking water. Wedeco AG Water Technology.
Lester, J.N. 1983. Significance and behaviour of heavy metals in waste water treatment
processes. I. sewage treatment and effluent discharge. Science of the Total Environment
1 30: 1-44.
Lienberger, J. 2002. Trojan Technologies. Personal communication with Jennifer Hafer,
Malcolm Pirnie, Inc, regarding UV reactors. November 5.
Logsdon, G. and J. Symons. 1973. Mercury removal by conventional water-treatment
techniques. Journal American Water Works Association 554-562.
Malley, J. 2001. Addressing concerns over mercury releases from UV lamps. IUVA News 3, no
3:28-29.
McClean, J. 200la. Hanovia Corporation. Personal communication with Doug Owen and
Christine Cotton, Malcolm Pirnie, Inc., regarding UV reactors. Slough, England, April 9.
McClean, J. 200 Ib. Hanovia Corporation. Response to memo by Christine Cotton, Malcolm
Pirnie, Inc., regarding UV reactors. Tucson AZ, May 4. • •
Merck & Co., Inc. 1983. The Merck Index ltfh Edition, edited by M.' Windholz. Rahway,NJ.
Moss, L. 2002a. Personal communication with Jennifer Hafer, Malcolm Pirnie, Inc., regarding
UV reactors. Tucson AZ, April 29.
Moss, L. 2002b. Personal communication with Jennifer Hafer, Malcolm Pirnie, Inc., regarding
' UV reactors. Tucson AZ, November 18.
Mugan, TJ. 1996. Quantification of total mercury discharges from publicly owned treatment
works to Wisconsin surface waters. Water Environment Research 68: 2,229-234.
UV Disinfection Guidance Manual N-21 June 2003
Proposal Draft
-------
Appendix N. UV Lamp Breakage Issues
Oliver, B.G., and E.G. Cosgrove. 1974. Efficiency of heavy-metal removal by a conventional
activated-sludge treatment plant. Water Research 8, no 11: 869-874.
Phillips, R. 1983. Sources and Applications of Ultraviolet Radiation. Academic Press Inc.,
New York, New York.
Primarc Limited. 2001. September 5,. www.primarc.com.
Roberts, A. 2000. Hanovia Corporation. Letter to Dan Schmelling, EPA. April 10.
Seigneur, C., J. Wrobel, and E. Constantinou. 1994. A chemical kinetic mechanism for
atmospheric inorganic mercury. Environmental Science and Technology 28: 1589-1597.
Srikanth, B. 2001a. Aquafine Corporation. Letter to James Malley, UNH, regarding accidental
UV lamp breakage and potential scenarios and prevention strategies. June 13.
Srikanth, B. 2001b. Aquafine Corporation. Personal communication with Jennifer Hafer,
Malcolm Pirnie, Inc., regarding UV reactors. Tucson, AZ, June 21.
Swaim, P.O., M.A. Morine, R.G. Brauer, M.A. Neher, J.L. Gebhart, and W.D. Bellamy. 2002.
Operating data from Henderson's full-scale UV disinfection facility. Proceedings from
the Water Quality Technology Conference, November 10-14, Seattle, WA.
U.S. Environmental Protection Agency. 1992. Characterization of Products Containing
Mercury in Municipal Solid Waste in the United States, 1970-2000. Office of Solid '
Waste, Washington, D.C.
U.S. Environmental Protection Agency. 1997a. Mercury - Emergency Spill & Release Facts.
EPA 540-K-97-004, OSWER 9378.0-WFS, PB97-963405. Office of Emergency and
Remedial Response, Washington, D.C.
U.S. Environmental Protection Agency. 1997b. Mercury emissions from the disposal of
fluorescent lamps. Office of Solid Waste, Washington, D.C. June 30.
U.S. Environmental Protection Agency. 1997c. Mercury Study Report to Congress. EPA-
425/R-97-004: 2-14. Office of Air Quality Planning and Standards and Office of
Research and Development, Washington, D.C.
Walitsky, P. Philips Lighting. 2001. Personal communication with Jennifer Hafer, Malcolm
Pirnie, Inc., regarding UV reactors. Tucson AZ, May 14.
Wu, J.S. and H. Hilger. 1985. Chemodynamic behavior of mercury in activated sludge process.
Am. Inst. Chem. Eng. 81: 109.
UV Disinfection Guidance Manual
Proposal Draft
N-22
June 2003
-------
Appendix O. Case Studies
This appendix will be included in the final draft when more information is available.
UV Disinfection Guidance Manual 0-1 June 2003
Proposal Draft
-------
Appendix P. Validation Protocol Calculator Tool
The validation protocol described in Chapter 4 and Appendix C of this guidance manual
involves several calculations to determine the log inactivation credit achieved during validation
of a UV reactor. For this protocol, a safety factor calculated from uncertainties and known
variability associated with UV reactors, monitoring, and validation methods is used to relate the
reduction equivalent dose (RED) demonstrated during validation to the UV dose required to
achieve a specified log inactivation credit (Table 1.4). EPA developed a spreadsheet that enables
a user to input information associated with the uncertainty of validation and monitoring and
calculate the safety factor and resulting target RED. The calculator was used to develop Tier 1
RED targets and can be used to apply the Tier 2 approach. (See section C.4.10.2 for a
description of the Tier 2 approach, including the safety factor calculation.)
The Microsoft Excel® spreadsheet contains the following five worksheets:
• Instructions - provides step-by-step instructions for entering data into the "RED
Bias", "Polychromatic Bias", and "Safety Factor" worksheets and executing macros
to calculate safety factor and resulting target RED.
. RED Bias - calculates RED bias from input of Chapter 1 UV dose requirements and
inactivation and RED measured during validation.
. Polychromatic Bias - calculates the polychromatic bias for medium-pressure UV
systems using spectral data on the lamp output, sleeve UV transmittance, water UV
transmittance, and sensor response.
. Safety Factor - calculates safety factor from RED bias, polychromatic bias, and
expanded uncertainty associated with reactor validation and monitoring.
• Default Data - contains assumed data for calculating the polychromatic bias;
alternatively, the user can provide validation testing data as specified in the
instruction worksheet.
UV Disinfection Guidance Manual
Proposal Draft
P-l
June 2003
-------
&ER&
United States
Environmental Protection
Agency
Source Water Monitoring Guidance
Manual for Public Water Systems for the
Long Term 2 Enhanced Surface Water
Treatment Rule (LT2 Rule)
June 2003
Draft
.
-------
Office of Water (4607)
EPA 815-D-03-005
http://www.epa.gov/safewater/lt2/tndex.htnil
June 2003
Printed on Recycled Paper
-------
Disclaimer
The Standards and Risk Management Division, of the Office of Ground Water and Drinking Water, has
reviewed and approved this guidance for publication. Neither the United States Government nor any of its
employees, contractors, or their employees make any warranty, expressed or implied, or assumes any
legal liability or responsibility for any third party's use of or the results of such use of any information,
apparatus, product, or process discussed in this report, or represents that its use by such party would not
infringe on privately owned rights. Mention of trade names or commercial products does not constitute
endorsement or recommendation for use.
Questions concerning this document or its application should be addressed to:
Mary Ann Feige
U.S. EPA Office of Ground Water and Drinking Water
Technical Support Center
Room 127
26 West Martin Luther King Drive
Cincinnati, OH 45268-1320
(513)569-7944
(513) 569-7191 (facsimile)
feige.marvann(a!epa.eov
-------
TABLE OF CONTENTS
Section 1: Introduction 1
1.1 Background 2
1.2 Large System Requirements ; 2
1.3 Small System Requirements 3
\ .4 Use of Cryptosporidium Data 5
1.4.1 Cryptosporidium Monitoring Sample Data 5
1.4.2 Cryptosporidium Matrix Spike Data 6
1.5 Use of E. coli Data 7
Section 2: Grandfathering Cryptosporidium Data 8
2.1 General Guidelines for Generating Cryptosporidium Data for Grandfathering 8
2.1.1 Sample Collection Location 8
2.1.2 Sample Collection Schedule 9
2.1.3 Cryptosporidium Analytical Methods for Grandfathered Data 9
2.1.4 Cryptosporidium Laboratories for Grandfathered Data 11
2.1.5 E. coli and Turbidity Measurements 11
2.2 Reporting Grandfathered Data 11
2.2.1 Data Package Contents 11
2.2.2 Schedule for Submission of Grandfathered Data 13
2.2.3 Procedures for Submission of Grandfathered Data 13
2.3 Checklists for Grandfathering Cryptosporidium Data 13
Section 3: Understanding Cryptosporidium Analyses 14
3.1 Summary of EPA Methods 1622 and 1623 14
3.2 Cryptosporidium Laboratory Quality Control 15
3.2.1 Initial Precision and Recovery Test 15
3.2.2 Method Blank Test 16
3.2.3 Ongoing Precision and Recovery Test 16
3.2.4 Holding Time Requirements 16
3.2.5 Staining Controls 17
3.2.6 Proficiency Testing Samples 17
3.2.7 Matrix Spike Samples 17
3.3 Archiving Examination Results 17
Section 4: Understanding E. coli Analyses 18
4.1 Summary of LT2 Rule E. coli Methods 18
4.1.1 Most Probable Number (MPN) Methods 18
4.1.2 Membrane Filtration (MF) Methods 19
4.2 E. coli Laboratory Quality Control 21
4.2.1 Dilution/Rinse Water Sterility Check 21
4.2.2 Media Sterility Check 21
4.2:3 Positive/Negative Controls 21
4.2.4 Media Storage ' 22
4.2.5 Filtration Unit Sterilization 22
Draft
June 2003
-------
4.2.6 Preparation Blanks ; 22
4.2.7 Verification 22
Section 5: Contracting for Cryptosporidium Laboratory Services 23
5.1 Defining Your Needs and Developing a Contract 23
5.1.1 Client Information 24
5.1.2 Sample Information 24
5.1.3 Sampling Schedules 26
5.1.4 Analytical Methodology 26
5.1.5 Data Deliverables and Other Contract Issues 28
5.2 Developing a Bid Sheet 31
' 5.3 Soliciting the Contract • 32
5.3.1 Approved Laboratories 32
5.3.2 Primary and Backup Laboratory Contracts 33
5.4 Evaluating Bids 33
5.4.1 Identifying Responsive Bidders 33
5.4.2 References 34
5.5 Communicating with the Laboratory 34
Section 6: Collecting and Shipping Source Water Samples 35
6.1 Sample Volumes 36
6.2 Sample Collection Location 37
6.2.1 Plants That Do Not Have a Sampling Tap Located Prior to Any Treatment .. 37
6.2.2 Plants That Use Different Water Sources at the Same Time 37
6.2.3 Plants That Use Presedimentation 37
6.2.4 Plants That Use Raw Water Off-Stream Storage 38
6.2.5 Plants That Use Bank Filtration 38
6.3 Source Water Monitoring Schedule 38
6.4 Sample Collection Guidance 39
6.4.1 Sample Collection Documentation 40
6.4.2 Cryptosporidium Sample Collection 41
6.4.3 E. coli Sample Collection 44
6.4.4 Measuring Turbidity 44
6.4.5 Monitoring Sample Temperature 46
Section 7: Reviewing Cryptosporidium Data 48
7.1 Cryptosporidium Data Recording at the Laboratory 48
7.1.1 LT2 Sample Collection Form 48
7.1.2 Method 1622/1623 Bench Sheet 48
7.1.3 Method 1622/1623 Cryptosporidium Slide Examination Form 49
7.2 Submitting Cryptosporidium Data through the LT2 Data Collection System 49
7.2.1 Data Entry/Upload 50
7.2.2 PWS Data Review 51
7.2.3 EPA/State Review 51
7.3 What Do the Sample Examination Results Mean? 51
7.3.1 Immunofluorescent Assay (IFA) 51
7.3.2 4',6-diamadino-2-phenylindole (DAPI) Examination 52
7.3.3 Differential Interference Contrast (DIG) Examination 52
7.5 Reviewing and Validating Raw Cryptosporidium Data (Optional) 52
7.5.1 Data Completeness Check 53
7.5.2 Evaluation of Data Against Method Quality Control Requirements 53
Draft \\ June 2003
-------
7.5.3 Calculation Verification 54
7.6 Data Archiving Requirements 56
Section 8: Reviewing E. coll Data 57
8.1 E. coli Laboratory Data Recording at the Laboratory 57
8.1.1 Sample Identification Information 57
8.1.2 Primary Data 57
8.1.3 Sample Processing and Quality Control Information 57
8.1.4 Sample Results 58
8.2 Submission of E. coli Data through the LT2 Data Collection System 58
8.2.1 Data Entry/Upload 58
8.2.2 PWS Data Review 59
8.2.3 EPA/State Review 59
8.3 Reviewing and Validating E. coli Data (Optional) 59
8.3.1 Data Completeness Check 59
8.3.2 Evaluation of Data Against Method Quality Control Requirements 60
8.3.3 Calculation Verification '. 61
8.4 Data Archiving Requirements 67
Section 9: References 68
Section 10: Acronyms 69
Draft
in
June 2003
-------
TABLES
Table 1-1. Timeline for Large Systems Regulated under the LT2 Rule 3
Table 1-2. Timeline for Small Systems Regulated under the LT2 Rule 4
Table 1-3. Bin Classifications 5
Table 1-4. Effect of the Number of Oocysts on Bin Classification Based on Mean of 12 Samples . 6
Table 1-5. Effect of the Number of Oocysts on Bin Classification Based on Mean of 48 Samples . 6
Table 6-1. Summary of LT2 Rule Monitoring Requirements 35
Table 6-2. Sample Collection Activities Required for Each Plant Type 40
Table 6-3. Minimum Data Elements to Record During Sample Collection 40
Table 6-4. Contacts for Filters Approved for Use in EPA Method 1622/1623 43
Table 7-1. LT2 Data Collection System Data Entry, Review, and Transfer Process 50
Table 8-1. Incubation Times and Temperatures for Approved E. Coli Methods 6!
Table 8-2. Examples of Different Combinations of Positive Tubes 66
Draft
IV
June 2003
-------
APPENDICES
Appendix A Checklist for Beginning Grandfathered Cryptosporidium Monitoring
Appendix B Checklist for Submitting Grandfathered Cryptosporidium Data
Appendix C Example LT2 Sample Collection Form
Appendix D Example Bulk Sample Collection Protocol for Cryptosporidium
Appendix E Example Envirochek™ Field Filtration Protocol for Cryptosporidium
Appendix F Example Filta-Max™ Field Filtration Protocol for Cryptosporidium
Appendix G Example E. coli Sample Collection Protocol
Appendix H Method 2130B for Turbidity Measurement
Appendix I Great Lakes Instrument Method 2 for Turbidity Measurement
Appendix J Revised EPA Method 180.1 for Turbidity Measurement
Draft
June 2003
-------
SECTION 1: INTRODUCTION
The Long Term 2 Enhanced Surface Water Treatment Rule (LT2ESWTR or LT2 rule) requires public
water systems (PWSs) that use surface water or ground water under the direct influence of surface water
to monitor their source water (influent water prior to treatment) for Cryptosporidiwn, E. coli, and
turbidity for a limited period [40 CFR part 141.701 (a)-(h)]. In support of the monitoring requirements
specified by the rule, three documents have been developed to provide guidance to the affected PWSs and
the laboratories that support them:
• Source Water Monitoring Guidance Manual for Public Water Systems for the Long Term 2 Enhanced
Surface Water Treatment Rule (LT2 Rule) (this document). This guidance manual for PWSs affected
by the rule provides information on laboratory contracting, sample collection procedures, and data
.evaluation and interpretation advice.
• Microbial Laboratory Guidance Manual for the Long Term 2 Enhanced Surface Water Treatment
Rule (LT2 Rule). The goal of this manual is to provide Cryptosporidium and E. coli laboratories
analyzing samples in support of the LT2 rule with guidance and detailed procedures for all aspects of
microbial analyses under the rule to maximize data quality and consistency.
• Users' Manual for the Long Term 2 Enhanced Surface Water Treatment Rule (LT2 Rule) Data
Collection System. This manual provides PWSs and laboratories with instructions on the entry,
review, and approval of electronic data using the LT2 Data Collection System, and for generating
reports using the system.
All of these manuals are available at http://www.epa.gov/safewater/lt2/index.htm!. Responses to
frequently asked questions (FAQs) on sampling, analysis, and data reporting questions for the LT2 rule
also are available on this website.
This guidance document is provided to help implement the LT2 rule. This guidance document does not,
however, substitute for the LT2 rule or the analytical methods approved for use under the rule. The
material presented is intended solely for guidance and does not alter any regulatory or analytical method
requirements not altered by the LT2 rule itself.
This manual provides guidance on the following aspects of the LT2 rule:
• Section 1: Overview of the rule's monitoring requirements and how the Cryptosporidium and E. coli
data collected under the rule will be used
• Section 2: Guidance on submitting historical data ("grandfathering")
• Section 3: Understanding Cryptosporidium analyses
• Section 4: Understanding E. coli analyses
• Section 5: Establishing a Cryptosporidium laboratory contract
• Section 6: Guidance on collecting and shipping LT2 monitoring samples
• Section 7: Reviewing Cryptosporidium data
• Section 8: Reviewing E. coli data
Draft \ June2003
-------
Section 1: Introduction
1.1 Background
The LT2 rule is a National Primary Drinking Water Regulation that requires monitoring, reporting, and
public notification requirements for all PWSs that use surface water sources. The rule requires additional
treatment techniques for some systems, based on Cryptosporidium monitoring results (40 CFR part
141.720 - 141.721). The LT2 rule was developed to improve control of microbial pathogens, including
specifically the protozoan Cryptosporidium, in drinking water and to address risk trade-offs with
disinfection byproducts.
The LT2 rule provides for increased protection against microbial pathogens in public water systems that
use surface water sources. The rule focuses on Cryptosporidium, a protozoan pathogen that is widespread
in surface waters. EPA is particularly concerned about Cryptosporidium because it is highly resistant to
inactivation by standard disinfection practices. Ihgestion of Cryptosporidium oocysts can cause acute
gastrointestinal illness, and symptoms in sensitive subpopulations may be severe, including
risk of mortality. Cryptosporidium has been identified as the pathogenic agent in a number of waterborne
disease outbreaks.
EPA convened a Federal Advisory Committee to develop recommendations for both the Stage 2
Disinfectants and Disinfection Byproducts Rule and the LT2 rule. As recommended by the Federal
Advisory Committee,.the LT2 rule requires public water systems that use surface water or ground water
under the direct influence of surface water to monitor their source water (influent water prior to treatment
plant) for Cryptosporidium, E. coli, and turbidity [40 CFR part 141.701 (a)-(h)]. These data would be
used to determine whether additional treatment is needed at PWSs and to assess whether a relationship
could be established between the Cryptosporidium and E. coli levels in source water.
1.2 Large System Requirements
Large systems affected by the LT2 rule include both filtered and unfiltered systems.
• A large, filtered system in the LT2 rule is a system that:
• Uses surface water or ground water under the direct influence of surface water
• Serves at least 10,000 people
• Provides filtration or is unfiltered, but required to install filtration because the system no longer
meets all filtration avoidance criteria
Large, filtered systems are required to conduct initial source water monitoring that includes
Cryptosporidium, E. coli, and turbidity sampling [40 CFR part 141.701 (b)].
• A large, unfiltered system in the LT2 rule is a system that:
• Uses surface water or ground water under the direct influence of surface water
• Serves at least 10,000 people
• Does not currently provide filtration and meets all filtration avoidance criteria
Large unfiltered systems are required to conduct initial source water monitoring that includes
Cryptosporidium sampling only [40 CFR part 141.701 (d)].
All of the Cryptosporidium sampling requirements and guidance discussed in this document apply equally
to both filtered and unfiltered systems and both filtered and unfiltered systems that serve at least 10,000
Draft
June 2003
-------
Section 1: Introduction
people are referred to as large systems in this document. However, the E. colt and turbidity guidance in
this document does not apply to large unfiltered systems.
The steps required for LT2 rule compliance for large systems, and the schedule for these steps, are
summarized in Table 1-1
Table 1-1. Timeline for Large Systems Regulated under the LT2 Rule
Event
Establish contract with a
Cryptosporidium laboratory
pending approval under EPA's Lab
QA Program (Section 2.4.1, below)
Verify that your utility laboratory is
certified under the drinking water
laboratory certification program to
perform the technique you plan to
use for performing £. coli analyses
under LT2"
Submit grandfathered
Cryptosporidium data package
Work with your Cryptosporidium
laboratory to establish a mutually
acceptable sampling schedule
Submit sampling schedule through
the LT2 Data Collection System
Collect monitoring samples"
Submit monitoring results"
Schedule
As soon as possible
As soon as possible
Within 2 months* of rule promulgation1"
Within 8 months of rule promulgation'
As soon as possible after establishing contract
Within 3 months of rule promulgation
Beginning 6 months after rule promulgation
No later than 10 days after the end of the first month
following the month that the sample was collected
(approximately 40 to 70 days after sample collection,
depending on when during the month the sample is
collected)
Duration
N/A - single event
N/A - single event
N/A - single event
N/A - single event
N/A • single event
At least once per
month for 2 years*
At least once per
month for 2 years"
Not applicable to large, unfi/tered systems because these systems are not required to monitor for £ coli or tumidity
b PWSs with at least 2 years of grandfathered data at the time of LT2 rule promulgation and that intend to use these
data in lieu of monitoring under the LT2 rule
c PWSs with fewer than 2 years of grandfathered data at the time of LT2 rule promulgation, or that have at least 2
years of grandfathered data but intend to conduct monitoring under the LT2 rule
" PWSs may be eligible to use historical (grandfathered) data in lieu of these requirements if certain quality
assurance and quality control criteria are met (see Section 2)
" PWSs monitoring for Cryptosporidium may collect more than one sample per month if sampling is evenly spaced
over the monitoring period
N/A = Not applicable
1.3 Small System Requirements
A small system in the LT2 rule is a system that:
• Uses surface water or ground water under the direct influence of surface water
• Serves fewer than 10,000 people
• Provides filtration or is unfiltered but required to install filtration because the system no longer meets
all filtration avoidance criteria
Draft
June 2003
-------
Section 1: Introduction
• Does not currently provide filtration and meets all filtration avoidance criteria
These systems are required to conduct initial source water monitoring for E. coli as an indicator of
Cryptosporidium and, for those systems exceeding E. coli trigger levels, Cryptosporidium monitoring [40
CFRpartl41.701(c)].
The steps required for LT2 rule compliance for small systems, and the schedule for these steps, are
summarized in Table 1-2.
Table 1-2. Timeline for Small Systems Regulated under the LT2 Rule
Event
Verify that your utility laboratory is
certified under the drinking water
laboratory certification program to
perform the technique you plan to
use for perform £ coli analyses
under LT2
Submit sampling schedule through
the LT2 Data Collection System
Collect £ coli samples
Submit E. coli monitoring results
Schedule
Prior to rule promulgation
Within 27 months of rule promulgation
Beginning 30 months (2.5 years) after
rule promulgation
No later than 10 days after the end of
the first month following the month that
the sample was collected
(approximately 40 to 70 days after
sample collection, depending on when
during the month the sample is
collected)
Duration
N/A - single event
N/A - single event
1 year (2 samples per month)
At least once per month for 1 year
Possible additional monitoring requirement for Cryptosporidium if small systems exceed £ coli trigger levels*
Establish contract with a
Cryptosporidium laboratory
pending approval under EPA's Lab
QA Program (Section 2.4.1, below)
Submit sampling schedule through
the LT2 Data Collection System •
Work with your Cryptosporidium
laboratory to establish a mutually
acceptable sampling schedule
Collect Cryptosporidium samples
Submit Cryptosporidium monitoring
results
As soon as possible after you are
notified that your plant has exceeded
the£. co// trigger levels
Within 45 months of rule promulgation
Within 2 months of rule promulgation
48 months (4 years) after
promulgation
No later than 1 0 days after the end of
the first month following the month that
the sample was collected
(approximately 40 to 70 days after
sample collection, depending on when
during the month the sample is
collected)
N/A - single event
N/A - single event
N/A - single event
1 year (2 samples per month)1"
At least once per month for 1 year
Small systems may be required to monitor for Cryptosporidium for 1 year, beginning 6 months after completion of
£ coli monitoring; Cryptosporidium monitoring required if the £ coli annual mean concentrations exceed 10 £
CO///100 mL for systems using lakes/reservoirs or exceed 50 £ coW100 ml for systems using flowing streams
b PWSs monitoring for Cryptosporidium may collect more than two samples per month if sampling is evenly spaced
over the monitoring period
N/A = Not applicable
Draft
June 2003
-------
Section 1: Introduction
Details on the use of the Cryptosporidium and E. coli data collected under the LT2 rule are provided in
Sections 1.4 and 1.5.
1.4 Use of Cryptosporidium Data
Two types of Cryptosporidium data are collected under the LT2 rule: Cryptosporidium occurrence data
from the analysis of monitoring samples, and method performance data from the analysts of matrix spike
(MS) samples. The use of occurrence data from monitoring samples is discussed in Section 1.4.1; the use
of method performance data from MS samples is discussed in Section 1.4.2.
1.4.1 Cryptosporidium Monitoring Sample Data
The concentration of Cryptosporidium oocysts in source water samples analyzed during the LT2 rule will
be used to calculate a mean Cryptosporidium concentration for a PWS and classify the PWSs into a
treatment requirements "bin" (40 CFR part 141.709). These bin classifications are provided in Table 1-3.
The treatment bin classification established for each PWSs will be used to determine whether additional
treatment is needed. PWSs in Bin 1 are not required to implement additional treatment. PWSs in Bins 2 -
4 will be required to implement increasing levels of treatment and source water protection'to address their
greater risk for high Cryptosporidium source water concentrations.
Table 1-3. Bin Classifications
- Cryptosporidium Bin Concentration
Cryptosporidium < 0.075 oocysts/L
0.075 oocysts/L <: Cryptosporidium < 1 .0 oocyst/L
1 .0 oocyst/L & Cryptosporidium < 3.0 oocysts/L
Cryptosporidium z 3.0 oocysts/L
Bin Classification
Bih1
. Bin 2
Bin 3
Bin 4
1.4.1.1 Calculating Bin Classifications
The method used to average individual sample concentrations to determine a PWS's bin classification
depends on the number of samples collected and the length of the sampling period.
For a PWS serving at least 10,000 people, bin classification would be based on the following:
• , For PWSs that collect at least 48 samples during the required monitoring period, the Cryptosporidium
bin calculation is equal to the arithmetic mean of all sample concentrations
• For PWSs that collect at least 24 samples, but not more than 47 samples, during the required
monitoring period, the Cryptosporidium bin calculation is equal to the highest arithmetic mean of all
1 sample concentrations in any 12 consecutive months in the monitoring period
For PWS serving fewer than 10,000 people, and that monitor for Cryptosporidium for 1 year, bin
classification would be based on the simple arithmetic mean of all sample concentrations.
In all cases, the bin concentration is calculated using individual sample concentrations. These
concentrations are calculated as "number of oocysts detected / volume (in L) analyzed." Individual
sample concentrations are not calculated as "oocysts detected / 10 L," nor are bin concentrations
calculated as the "sum of the oocysts detected / the sum of the volumes analyzed." As a result, each
sample has an equal weight on the final bin concentration. In cases where no oocysts are detected, the
number of oocysts used to calculate the sample concentration is "0."
Draft 5 June 2003
-------
Section 1: Introduction
1.4.1.2 Number of Oocysts Detected Versus Bin Classification
To better understand the relationship between the absolute number of oocysts detected in your samples
and the resulting bin classification, several crosswalks are provided below. Table 1-4 applies to large
plants conducting monthly monitoring over 2 years. This table provides a crosswalk between the sum of
the oocysts detected in 10- and 50-L samples during the highest 12-month period and the corresponding
bin classification.
Table 1-4. Effect of the Number of Oocysts on Bin Classification Based on Mean of 12 Samples
Sum of oocysts
found in 12, 10-L
samples*
0 - 8 oocysts
9- 125 oocysts
126 -365 oocysts
366 or more oocysts
Sum of oocysts
found in 12, 50-L
samples'1
0-44 oocysts
45 - 629 oocysts
630- 1829 oocysts
1830 or more oocysts
Corresponding range of mean
Cryptosporidium concentrations
From
To
< 0.075 oocystS/L
0.075 oocysts/L
1 .0 oocyst/L
< 1 .0 oocyst/L
< 3.0 oocysts/L
i 3.0 oocysts/L
Corresponding bin
classification
1
2
3
4
* Representing the highest 12-month mean; assumes that 10-L samples are analyzed for each event
b Representing the highest 12-month mean; assumes that 50-L samples are analyzed for each event
Table 1-5 applies to large plants conducting semimonthly monitoring over 2 years. This table provides a
crosswalk between the sum of the number of oocysts detected in samples during the entire 2-year
monitoring period and the corresponding bin classification. Again, because this crosswalk is based on
analysis of exactly 10 L or 50 L for all samples, it may not apply to all plants that monitor" for
Cryptosporidium on a semimonthly basis, but it helps put into perspective the impact that one high
sample result may have on bin classification.
Table 1-5.. Effect of the Number of Oocysts on Bin Classification Based on Mean of 48 Samples
Sum of oocysts
found in 48, 10-L
samples*
0-35
36 - 503
504 - 1463
1464 or more
Sum of oocysts
found in 48, 50-L
samples"
0-179 oocysts
180-2519oocyst
2520 -731 9 oocysts
7320 or more oocysts
Corresponding range of mean
Cryptosporidium concentrations
From
To
< 0.075 oocysts/L
0.075 oocysts/L
1.0 oocyst/L
< 1 .0 oocyst/L
< 3.0 oocysts/L
* 3.0 oocysts/L
Corresponding bin
classification
1
2
3
4
* Assumes that 10-L samples are analyzed for each event
b Assumes that 50-L samples are analyzed for each event
Systems may analyze larger volumes than 10 L, and larger volumes analyzed should increase analytical
sensitivity (detection limit), provided method performance is acceptable. Because these tables are based
on analysis of exactly 10 L or exactly 50 L for all samples, it may not apply to all plants that monitor
monthly for Cryptosporidium, but it helps put into perspective the impact that one high sample result may
have on bin classification. In addition, filtering higher volumes may not result in the same high volume
analyzed if the source is turbid and the PWS analyzes only a portion of the concentrated sample. The
calculations used to determine the volume analyzed if less than the entire sample volume is analyzed are
discussed in Section 7.5.3.
1.4.2 Cryptosporidium Matrix Spike Data
During LT2 rule Cryptosporidium monitoring, PWSs are required to collect one matrix spike (MS)
sample for every 20 monitoring samples from their source water, per the requirements in EPA Methods
Draft
June 2003
-------
Section I: Introduction
1622/1623 (Section 9.1.8). A description of MS samples is provided in Section 3.2.7 of this document.
For large systems that perform monthly monitoring for 2 years and collect 24 monitoring samples and for
small systems that are triggered into monitoring for 1 year and collect 24 monitoring samples, two MS
samples will be analyzed. For large systems that perform semimonthly or more frequent monitoring for 2
years and collect 48 or more samples, a minimum of three MS samples will be analyzed.
Although MS sample results will not be used to adjust Cryptosporidium recoveries at any individual
source water, the results will be used collectively to assess overall recovery and variability for EPA
Method 1622/1623 in source water. The descriptive statistics of the MS sample results will be compared
to the performance of the methods during the Information Collection Rule Supplemental Surveys to
verify the assumptions on method performance upon which the LT2 rule is based.
When considering the method performance that could be achieved for analysis of Cryptosporidium under
the LT2 rule, the Federal Advisory Committee (FACA) evaluated the results of EPA Methods 1622/1623
in the ICRSS, which involved 87 PWSs sampling twice per month over 1 year for Cryptosporidium and
other parameters. During the ICRSS, the mean Cryptosporidium recovery and mean relative standard
deviation of the MS samples were 43% and 49%, respectively (Reference 9.1).
1.5 Use of E. co// Data
E. coli data are being collected by large systems during LT2 rule monitoring to assess whether a
relationship can be established between the Cryptosporidium and E. coli levels in source water and a
microbial index developed to establish trigger levels for E. coli that would indicate high Cryptosporidium
concentrations in a source water. If a relationship can be established, small systems initially will be
permitted to monitor for E. coli, rather than.conducting more expensive Cryptosporidium analyses. Only
those systems with E. coli levels above the trigger level established in the microbial index would then be
required to monitor for Cryptosporidium to determine bin placement (40 CFR part 141.702).
A preliminary index was developed during development of the FACA agreement using data from the
Information Collection Rule (ICR) and ICRSS (Reference 9.2). These data were evaluated for parameters
that could indicate the likelihood that a source water mean Cryptosporidium level would be above the Bin
2 threshold concentration of 0.075 oocysts/L. Fecal coliforms, total coliforms, E. coli, viruses (ICR only),
and turbidity were assessed for development of the microbial index. Data analyses placed greater
emphasis on E. coli and fecal coliforms because of the direct relationship between these parameters and
fecal contamination.
E. coli was determined to provide the best performance as a Cryptosporidium indicator with the available
data. Based on the data from the ICR and ICRSS, the preliminary E. coli trigger levels were set at a mean
of 10 E. coli/lQQ mL for reservoir/lake-type source waters and 50 E, co/»'/100 mL for flowing stream-type
source waters.
These levels may potentially be revised based on the much larger, more reliable Cryptosporidium and E.
coli data set collected through LT2 rule monitoring.
Draft 1 • June 2003
-------
SECTION 2: GRANDFATHERING CRYPTOSPORIDIUM DATA
"Grandfathered" Cryptosporidium data are results generated before monitoring under the Long Term 2
Enhanced Surface Water Treatment Rule (LT2ESWTR or LT2 rule) starts and that a public water system
(PWS) intends to use in determining its bin classification (Section 1.4.1) under the rule. Grandfathered
data may be used in lieu of, or in addition to, results generated during LT2 rule implementation (40 CFR
part 141.708). This section of the manual is designed to assist PWSs in producing grandfathered data that
should be equivalent to the data collected during LT2 rule implementation and, therefore, eligible for use
in bin classification. The final LT2 rule will establish requirements for reporting and acceptance of
grandfathered monitoring results.
2.1 General Guidelines for Generating Cryptosporidium Data for
Grandfathering
/
A PWS's grandfathered Cryptosporidium data package should meet the following general conditions (40
CFR part 141.708):
• Samples were collected from the appropriate location(s)
• Samples were representative of a plant's source water(s) and the source water(s) have not changed
• Samples were collected no less frequently than each calendar month on a regular schedule, beginning
no earlier than January 1999 (when EPA Method 1622 was first released as an interlaboratory-
validated method)
• Samples were collected in equal intervals of time over the entire collection period (e.g., weekly,
twice-per-month, monthly)
« The data set includes all source water Cryptosporidium monitoring results generated during the
grandfathered data monitoring period (see details below—data from monitoring not directed towards
LT2 rule binning will not be a component of the binning data set)
• Sample volumes of at least 10 L were analyzed or, in cases where 10 L were not analyzed, at least 2
mL of packed pellet volume were analyzed (additional details below)
• The data were generated using the validated versions of EPA Methods 1622 or 1623
• The data are fully compliant with the QA/QC criteria specified in the version of Method 1622 or
Method 1623 used to generate the data, including analysis of matrix spike (MS) samples at a
frequency of at least 5% (1 MS sample for every 20 monitoring samples)
The following sections discuss these recommendations in more detail.
2.1.1 Sample Collection Location
The sample collection location requirements are the same for LT2 rule monitoring and for grandfathered
data and are discussed in Section 6.2. If the PWS does not collect samples as recommended in Section
6.2, the data may not be acceptable for grandfathering.
Draft
June 2003
-------
Section 2: Grandfathering Cryptosporidium Data
2.1.2 Sample Collection Schedule
During LT2 rule monitoring, PWSs will be required to collect samples at least monthly and in accordance
with a schedule established by the PWS prior to initiation of monitoring (40 CFR part 141.703). PWSs
may collect samples more frequently (e.g., twice-per-month, weekly), provided the same frequency is
maintained throughout the monitoring period [40 CFR part 141.701 (e)]. Sampling for grandfathered data
should follow these same criteria.
EPA recommends that, prior to initiation of grandfathered monitoring, PWSs develop a schedule listing
the calendar date on which each Cryptosporidium sample will be collected and include this schedule
when submitting the grandfathered data package to EPA. PWSs that have begun grandfathered
monitoring without establishing a sampling schedule should develop a schedule for the collection of
remaining samples. PWSs should collect samples within 2 days before or after the dates indicated in their
sampling schedules. Exceptions to the sampling schedule are noted as follows:
• If extreme conditions or situations exist that may pose danger to the sampler, or which are unforeseen
or cannot be avoided and which cause the system to be unable to sample in the required time frame,
the PWS should sample as close to the scheduled date as feasible and submit an explanation for the
alternative sampling date with the analytical results.
• PWSs that are unable to report a valid Cryptosporidium analytical result for a scheduled sampling
date due to failure to comply with the analytical method quality control standards (e.g., sample is lost
; or contaminated; laboratory exceeds an analytical method holding time) should collect a replacement
sample within 14 days of being notified by the laboratory that a result cannot be reported for that
date. PWSs should submit an explanation for the replacement sample with the analytical results.
Alternative sample collection dates should be timed so as not to coincide with another scheduled
Cryptosporidium sample collection date. Documentation of alternate sample collection, including the
reason, should be provided with the grandfathered data package.
Water treatment plants that use surface water or ground water under the direct influence (GWUDI), but
are operated only seasonally (e.g., during times of high-water demand) should monitor at least monthly
during the period when the plant is in operation.
The Federal Advisory Committee Agreement in Principle (Agreement) for the LT2 rule recommends that
if PWSs collect a total of at least 48 samples (regardless of whether all of the samples were collected
before LT2 rule promulgation or some were collected before and some after rule promulgation), the
Cryptosporidium bin concentration will be equal to the arithmetic mean of all sample concentrations [40
CFR part 141.709 (b)(l)]. For PWSs that collect a total of at least 24 samples, but not more than 47
samples, the Cryptosporidium bin concentration will be equal to the highest arithmetic mean of all sample
concentrations in any 12 consecutive months during which Cryptosporidium samples were collected [40
CFR part 141.709 (b)(2)]..
2.1.3 Cryptosporidium Analytical Methods for Grandfathered Data
Methods 1622 or 1623 should be used for Cryptosporidium analyses for the LT2 rule [40 CFR part
141.708 (b)(l)]. The following are EPA-validated versions of Methods 1622 and 1623 acceptable for
monitoring for Cryptosporidium before LT2 rule implementation:
• Method 1623: Cryptosporidium and Giardia in Water by Filtration/IMS/FA. U.S. Environmental
Protection Agency, Office of Water. 2001. EPA-821-R-01-025
• Method 1622: Cryptosporidium in Water by Filtration/IMS/FA. U.S. Environmental Protection
Agency, Office of Water. 2001. EPA-821-R-01-026
Draft , ' 9 June 2003
-------
Section 2: Grandfathering Cryptosporidium Data
• Method 1623: Cryptosporidium and Giardia in Water by Filtration/IMS/FA. U.S. Environmental
Protection Agency, Office of Water. 1999. EPA-821-R-99-006 (Note: The 2001 version of the
method should be used to generate data after January I, 2002.)
• Method 1622: Cryptosporidium in Water by Filtration/IMS/FA. U.S. Environmental Protection
Agency, Office of Water. 1999. EPA-821-R-99-001 (Note: The 2001 version of the method should be
used to generate data after January 1, 2002.)
The procedures in EPA Method 1622/1623 are performance-based, and allow for modifications. The 2001
versions of EPA Method 1622/1623 also approve for nationwide use modified versions of the methods
using the following components:
• Whatman Nuclepore CrypTest® filter
• IDEXX Filta-Max™ filter
• Waterborne Aqua-Glo™ G/C Direct FL antibody stain
* Waterborne Crypt-a-Glo™ and Giardi-a-Glo™ antibody stains
Since release of the 2001 versions of Methods 1622/1623, EPA also has approved a modified version of
the methods using the Pall Gelman Envirochek™ HV filter and has approved the use of irradiated, flow
cytometer-sorted spiking suspensions for routine QC sample spiking.
Laboratories that analyze Cryptosporidium samples using other modified procedures, as allowed under
the performance criteria of Methods 1622/1623, should be approved to use the modified procedures under
the Lab QA Program discussed in Section 2.1.4, below, and in detail in the Microbial Laboratory
Guidance Manual for the Long Term 2 Enhanced Surface Water Treatment Rule (LT2 Rule).
Other notable differences between the 1999 and 2001 versions of EPA Method 1622/1623 include the
following:
* Clarified sample acceptance criteria
* Modified capsule filter elution procedure
• Modified concentrate aspiration procedure
• Modified IMS acid dissociation procedure
Updated QC acceptance criteria for initial precision and recovery (IPR) and ongoing precision and
recovery (OPR) tests
Addition of a troubleshooting section for QC failures
• Modified holding times
• Inclusion of flow cytometry-sorted spiking suspensions (required for spiked samples analyzed during
LT2 monitoring)
2.1.3.1 Minimum Sample Volume and Subsampling Analysis
The requirements for sample volume analyses are the same for LT2 rule monitoring and for grandfathered
data [40 CFR part 141.708 (b)(5)]. These requirements are discussed in Section 6.1 of this manual.
2.1.3.2 Analysts of Matrix Spike Samples
The requirements for analysis of matrix spike (MS) samples are the same for LT2 rule monitoring and for
grandfathered data [40 CFR part 141.708 (e)]. These requirements, and guidance on MS sample
collection, are discussed in Section 6.4.2 of this manual.
Draft
10
June 2003
-------
Section 2: Grandfathering Cryptosporidium Data
2.1 A Cryptosporidium Laboratories for Grandfathered Data
EPA^has established the Laboratory Quality Assurance Evaluation Program for the Analysis of
Cryptosporidium in Water (Lab QA Program) to approve laboratories to perform Cryptosporidium
analyses under the LT2 rule (see http://www.epa.gov/safewater/lt2/index.html1. EPA recognizes that
some PWSs could begin generating grandfathered Cryptosporidium data prior to when the Lab QA
Program is fully implemented (e.g., before EPA is able to evaluate all laboratories that will participate in
the program). Consequently, PWSs should ensure that their grandfathered Cryptosporidium samples are
analyzed by laboratories that will be evaluated under the Lab QA Program before the data are submitted
to EPA. Note that PWSs will not submit grandfathered data packages until after the LT2 rule is final,
currently scheduled for mid- or late 2004. Samples analyzed by laboratories that do not meet the criteria
for approval under the LT2 rule may not be accepted for grandfathering.
Laboratories should also participate in the EPA Protozoa PT Program. EPA does not expect there to be
restrictions on the number of laboratories involved in the generation of a PWS's grandfathered data.
2.1.5 £. co// and Turbidity Measurements
The Agreement would not exclude the use of previously collected Cryptosporidium data if E. coli and
turbidity samples are not collected. However, the Agreement recommends that PWSs serving at least
10,000 people should collect E. coli and turbidity samples along with Cryptosporidium samples when
monitoring under the LT2 rule. EPA recommends that PWSs conducting early (i.e., grandfathered)
monitoring collect and analyze E. coli samples with each Cryptosporidium sample and measure turbidity
during each sampling event.
2.2 Reporting Grandfathered Data
The final LT2 rule will establish reporting requirements for grandfathered data. The following
recommendations are intended to give PWSs an indication of potential reporting requirements for
consideration when establishing their grandfathered data monitoring programs.
For consideration of grandfathered data, PWSs should submit to EPA a complete data package as
described below.
2.2.1 Data Package Contents
The grandfathered data package should include the following:
1. A signed cover letter from the PWS certifying that the data represent the plant's current source water
and that all source water Cryptosporidium monitoring results collected during the LT2 rule
monitoring period (defined below) are included in the package
2. Sample collection schedule established before beginning monitoring
3. Where applicable, documentation addressing the dates and reason(s) for re-sampling, as well as the
use of presedimentation, off-stream storage, or bank filtration during monitoring
4. A list of the field and MS samples submitted in the data package (see Section 2.2.1.1, below, for
details), identified by sample ID and collection date •
5. Sample results for all field and MS samples (see Section 2.2.1.2, below, for details) and
6. Documentation that all method-required quality control requirements were acceptable for every field
and MS sample submitted with the package (see Section 2.2.1.3, below, for details).
Draft 11 June 2003
-------
Section 2: Grandfathering Cryptosporidium Data
2.2.1.1 Sample Results to be Reported
PWSs that conduct monitoring for grandfathering should submit results for all source water
Cryptosporidium samples analyzed during the LT2 rule monitoring period, as defined below (40 CFR part
141.707). This will include all samples that were:
• Collected from the sampling location used for LT2 rule monitoring,
• Not spiked, and
• Analyzed using the laboratory's routine process for Method 1622/1623 analyses, including analytical
technique and QA/QC.
EPA plans that the LT2 rule monitoring period for a specific PWS will begin with the collection of the
first sample submitted for LT2 rule binning and end with the collection of the final sample submitted for
LT2 rule binning (as long as a minimum of 2 years of acceptable data have been submitted). With the use
of grandfathered data, the final sample may be collected before the end of the LT2 rule implementation
schedule. Sample results generated after the last sample result in the PWS's data package would be
considered outside the PWS's LT2 rule monitoring period and would not need to be submitted to EPA for
LT2 rule binning purposes. However, these results may be subject to reporting requirements under other
federal or State regulations.
2.2.1.2 Data Elements to be Reported for Each Sample Result
The following data elements, at a minimum, must be submitted for each Cryptosporidium monitoring
sample and MS sample [40 CFR part 141.708 (d)]:
• PWS ID
• Facility ID
• Sample collection point
• Sample collection date
• Sample type (field or MS)
• Sample volume filtered (L), to nearest % L '
• Number of oocysts counted
• For samples in which less than 10 L is filtered or less than 100% of the sample volume is examined,
PWSs should also report the number of filters used and the packed pellet volume.
• For samples in which less than 100% of sample volume is examined, PWSs should also report the
volume of resuspended concentrate and volume of this resuspension processed through
immunomagnetic separation.
• For matrix spike samples, PWSs should also report the sample volume spiked and estimated number
of oocysts spiked. These data are not applicable to monitoring samples.
EPA recommends that these data elements be reported by submitting a completed sample collection form,
laboratory bench sheet, and Cryptosporidium report form for each sample. Example bench sheets and
report forms are provided as attachments in the Microbial Laboratory Guidance Manual for the Long
Term 2 Enhanced Surface Water Treatment Rule (LT2 Rule), available for download from
http://www.epa.gov/safewater/lt2/index.html. Sample documentation forms that are different from these
examples, but that contain the minimum required data elements listed above, may be acceptable..
Draft
12
June 2003
-------
Section 2: GrandfatheringCTyptosporidmm Data
2.2.1.3 Supporting Quality Control Information
The data package should include a signed letter from the laboratory certifying that all method-required
quality control elements (including sample temperature upon receipt, ongoing precision and recovery and
method blank results, holding times; and positive and negative staining controls) were performed at the
required frequency, and were acceptable for every monitoring and MS sample submitted with the package
(however, the actual MS sample results are not required to meet the methods' MS QC acceptance
criteria). The letter should include a list of the applicable monitoring and MS samples, and the
corresponding OPR and method blank sample ID for each.
Alternately, the PWS may include the bench sheet and Cryptosporidium report form (or comparable
detailed data reporting forms) for each OPR and method blank sample associated with the field and MS
samples in the grandfathered data package. If this option is selected, the letter from the laboratory still
should certify that sample temperature upon receipt, holding times, and positive and negative staining
controls were acceptable for all samples. (The letter is not necessary if detailed data reporting forms
containing this information are submitted for the field and MS sample results.)
2.2.2 Schedule for Submission of Grandfathered Data
EPA's current intent is that PWSs with at least 2 years of grandfathered data at the time of LT2 rule
promulgation and that intend to use these data in lieu of monitoring under the LT2 rule (i.e., do NOT
intend to conduct additional monitoring) should submit these data to EPA within 2 months following LT2
rule promulgation (currently planned for mid- or late 2004). EPA plans to notify these PWSs within 4
months following LT2 rule promulgation as to whether their data are.sufficient for bin classification [40
CFR part 141.708 (f)].
PWSs with fewer than 2 years of grandfathered data at the time of LT2 rule promulgation, or that have at
least 2 years of grandfathered data but intend to conduct monitoring under the LT2 rule, should submit
these data to EPA within 8 months of LT2 rule promulgation (which provides the systems with 2 months
to review data from the last potential historical sampling event). Data collected when LT2 rule monitoring
begins (6 months after promulgation) will be submitted through the LT2 Data Collection System [40 CFR
part 141.708 (g)].
Under the Agreement, PWSs should conduct monitoring under the LT2 rule unless notified in writing by
EPA that they have 2 years of acceptable data.
2.2.3 Procedures for Submission of Grandfathered Data
EPA does not intend to formally accept grandfathered Cryptosporidium data until the LT2 rule is
finalized. The final rule will include procedures for submission of grandfathered data.
2.3 Checklists for Grandfathering Cryptosporidium Data
To help PWSs interested in monitoring for Cryptosporidium before LT2ESWTR apply the information
provided in this guidance, two checklists have been developed. The "Checklist for Beginning
Grandfathered Cryptosporidium Monitoring"(Appendix A) is designed to be used by PWSs to check
their monitoring plans against this guidance document before proceeding with monitoring. The "Checklist
for Submitting Grandfathered Cryptosporidium Data" (Appendix B) is designed to be used by PWSs to
check their data package against the information in this guidance document before submitting the data
package to EPA for review.
Draft 13 June 2003
-------
SECTION 3: UNDERSTANDING CRYPTOSPORIDIUM ANALYSES
The LT2 rule requires the use of EPA Method 1622 or EPA Method 1623 for Cryptosporidium
monitoring [40 CFR part 141.705 (a)]. This section provides utility personnel unfamiliar with
Cryptosporidium sample analyses with information on how the analyses are performed and on the quality
control (QC) measures the laboratory uses to verify data quality.
3.1 Summary of EPA Methods 1622 and 1623
EPA Methods 1622 and 1623 resulted from an EPA effort initiated in 1996 to identify new and
innovative technologies for analysis of source water samples for Cryptosporidium and Giardia. The
methods are identical in most respects, generally differing only in the addition of Giardia antibodies in
EPA Method 1623's purification and staining procedures. Both EPA Methods 1622 and 1623 were
subjected to Intel-laboratory validation studies using various source waters, and used in a national survey
of 87 surface water plants (the Information Collection Rule Supplemental Surveys) to provide EPA with a
realistic indication of how the methods would perform when they were used in the monitoring study
(Reference 9.1).
Both EPA Methods 1622 and 1623 also were developed as "performance-based" methods. The methods
include quantitative criteria requirements (minimum recovery and maximum variability) for initial and
ongoing QC samples. These criteria are used to verify acceptable laboratory performance using the
version of the method originally validated or to determine whether a modified version of the method
performs acceptably.
In EPA Methods 1622 and 1623, the following steps are performed:
• Filtration. The sample is filtered in the field or in the laboratory using one of the filters approved for
use with EPA Methods 1622 and 1623:
• Pall Gelman Envirochek™ capsule filter
• Pall Gelman Envirochek™ HV capsule filter
• IDEXX Filta-Max™ foam filter
The oocysts, cysts, and extraneous materials are retained on the filter.
• Elution. Materials on the filter are removed by elution with an aqueous buffered salt detergent
solution. This elution process is performed differently for each filter:
* For the Pall Gelman Envirochek™ and Envirochek™ HV filters, elution is performed by filling
the capsule with elution buffer, attaching the filter to a "wrist shaker" type lab shaker, and
allowing the filter to shake for 5 minutes at a time in three different orientations.
• For the IDEXX Filta-Max™ filter, the elution technique differs by laboratory. Some laboratories
may add the foam filter and elution buffer to a manual plunger chamber to expand the foam filter
and flush any oocysts out of the pores in the foam. Other laboratories may add the foam filter
rings and elution buffer to a stomacher bag and use a stomacher to elute the filter.
Draft
14
June 2003
-------
Section 3: Understanding Cryptosporidium Analyses
• For the Whatman CrypTest® filter, elution is performed by adding elution buffer to the filter
housing and using sonication and pressurized backwashing to separate oocysts from the filter
! fabric.
Concentration. After the filter is eluted, the eluate is centrifuged to concentrate the eluted particles
into a "packed pellet" at the bottom of the centrifuge tube. This packed pellet is measured by the
laboratory analyst. If the pellet volume is £ 2 mL (and 10 L was filtered) the entire sample must be
analyzed. If the pellet volume is > 2 mL, only 2 mL is required to be analyzed under the LT2 rule
(although the utility may request that more be analyzed).
Aspiration and resuspension. The analyst aspirates the supernatant from the top of the packed pellet
to minimize the total sample volume, and resuspends the pellet material by vortexing the sample. The
analyst measures the total resuspended concentrate volume. If the packed pellet volume was > 2 mL,
and the entire sample volume will not be analyzed, only a portion of the concentrate volume will be
processed through the remainder of the method. By dividing the concentrate volume processed
through the remainder of the method by the total concentrate volume, the laboratory can determine
what percent of the sample volume filtered was actually analyzed. By multiplying this percentage by
' the sample volume filtered, the laboratory can determine the volume analyzed.
, Purification. Magnetic beads conjugated to snti-Cryptosporidium antibodies are added to the sample
concentrate and allowed to mix with the sample, where they attach themselves to any oocysts present.
The magnetized oocysts are separated from the extraneous materials using a magnet, and the
extraneous materials are discarded. The magnetic bead complex is then detached from the oocysts.
' Application of the purified sample to a slide. After immunomagnetic separation, the purified
sample is applied to a microscope slide.
Drying the sample. The sample is dried to the slide for several hours to several days to allow the
sample to be stained and rinsed Without loss of organisms.
Staining the sample. Two stains are added to the sample before it is examined to help the analyst
identify any Cryptosporidium that may be present. The oocysts and cysts are stained on the slide with
fluorescently labeled monoclonal antibodies and 4',6-diamidino-2-phenylindole (DAPI).
Examining the sample. During microscopic examination of the slide, three evaluation techniques are
required by EPA Methods 1622 and 1623 to determine whether an object is a Cryptosporidium
oocyst. (Guidance on interpreting examination results is provided in Section 7.3.)
3.2 Cryptosporidium Laboratory Quality Control
As required by both EPA Method 1622/1623 and the Laboratory QA Program, laboratories approved to
perform Cryptosporidium analyses for the LT2 rule must perform specific quality control (QC) steps
during sample analyses to demonstrate that data are reliable [40 CFR part 141.705 (a)(3)]. These QC steps
are described below, in Sections 3.2.1 - 3.2.7.
3.2.1 Initial Precision and Recovery Test
Before performing field sample analyses using EPA Methods 1622 or 1623, the laboratory must
demonstrate acceptable performance. This is demonstrated by the initial precision and recovery (IPR) test,
which consists of four reagent water samples spiked with 100 to 500 oocysts. The results of the four
analyses are used to calculate the average percent recovery and the relative standard deviation (RSD) of
Draft 15 June 2003
-------
Section 3: Understanding Cryptosporidium Analyses
the recoveries for Cryptosporidium. For EPA Methods 1622/1623, the mean Cryptosporidium recovery
must be in the range of 24% to 100% and the RSD.of the four recoveries must be less than 55%. If more
than one process will be used for filtration and/or separation of samples, a separate set of IPR samples
must be analyzed for each process.
3.2.2 Method Blank Test
The method blank test in EPA Method 1622/1623 consists of analysis of an unspiked reagent water
sample to demonstrate freedom from contamination. One method blank sample must be analyzed each
week or every 20 samples, whichever is more frequent. If more than one process will be used for filtration
and/or separation of samples, a separate method blank must be analyzed for each process. If one or more
Cryptosporidium oocysts are found in a blank, analysis of additional samples is halted until the source of
contamination is eliminated and a blank shows no evidence of contamination.
3.2.3 Ongoing Precision and Recovery Test
The ongoing precision and recovery (OPR) in EPA Method 1622/1623 entails analysis of a reagent water
sample spiked with 100 to 500 oocysts to demonstrate ongoing acceptable performance. One OPR sample
must be analyzed each week or every 20 samples, whichever is more frequent. If more than one process
will be used for filtration and/or separation of samples, a separate OPR sample must be analyzed for each
process. OPR samples must be analyzed before any monitoring samples are processed for each batch to
verify acceptable performance. OPR Cryptosporidium recovery must be in the range of 11% to 100% to
be considered acceptable.
3.2.4 Holding Time Requirements
During Cryptosporidium analyses for the LT2 rule, sample processing should be completed as soon as
possible by the laboratory. The laboratory should complete sample filtration, elution, concentration,
purification, and staining the day the sample is received wherever possible. However, the laboratory is
permitted to split up the sample processing steps if processing a sample completely in one day is not
possible. If this is necessary, sample processing can be halted after filtration, application of the purified
sample onto the slide, or staining.
The following holding times must be met for samples analyzed by EPA Methods 1622/1623 during the
LT2 rule:
• Sample collection and filtration. Sample elution must be initiated within 96 hours of sample
collection (if shipped to the laboratory as a bulk sample) or filtration (if filtered in the field).
• Sample elution, concentration, and purification. The laboratory must complete the elution,
concentration, and purification in one work day. It is critical that these steps be completed in one
work day to minimize the time that any target organisms present in the sample sit in eluate or
concentrated matrix. This process ends with the application of the purified sample on the slide for
drying.
• Staining. The sample must be stained within 72 hours of application of the purified sample to the
slide.
• Examination. Although fluorescence assay (FA) and 4',6-diamidino-2-phenylindole (DAPI) and
differential interference contrast (DIG) microscopy examination and confirmation should be
performed immediately after staining is complete, laboratories have up to 7 days from completion of
sample staining to complete the examination and confirmation of samples. However, if
fading/diffusion of fluorescien isothiocyanate (FITC) or DAPI staining is noticed, the laboratory must
reduce this holding time. In addition, the laboratory may adjust the concentration of the DAPI
staining solution so that fading/diffusion does not occur.
Draft
16
June 2003
-------
Section 3: Understanding Cryptosporidium Analyses
3.2.5 Staining Controls
Positive staining controls entail staining and examination of a slide with positive antigen or 200 to 400
intact oocysts to verify that the stain is fluorescing appropriately. These controls are prepared with each
batch of slides that are stained. Negative staining controls entail staining and examining a slide with
phosphate buffered saline solution to verify that no oocysts or interfering particulates are present.
3.2.6 Proficiency Testing Samples
As part of the Lab QA Program, laboratories must successfully analyze initial proficiency testing (IPT)
samples initially, and an ongoing proficiency testing (OPT) samples three times per year. These samples
and the Lab QA Program are discussed in more detail in the Microbial Laboratory Guidance Manual for
the Long Term 2 Enhanced Surface Water Treatment Rule (LT2 Rule).
3.2.7 Matrix Spike Samples
The matrix spike (MS) test in EPA Method 1622/1623 entails analysis of a separate sample aliquot spiked
with 100 to 500 oocysts to determine the effect of the matrix on the method's oocyst recovery.
One MS sample must be analyzed for every 20 samples from your PWS. The first MS sample should be
collected and analyzed during the first sampling event under the monitoring program and at least 12
months must elapse between the first and last MS sample. You should evaluate the MS recoveries, as well
as other attributes of sample processing and examination, and work with the laboratory to determine
whether sample filtration and processing procedures are working acceptably, or need to be re-evaluated.
If it is not possible to analyze an MS sample for the first sampling event due to laboratory sample
processing burden or other reasons, the first MS sample should be analyzed as soon as possible to identify
potential method performance issues with the matrix. The requirement that at least 12 months must elapse
between the first and last MS sample still applies. For example, if a PWS that is monitoring monthly for
24 months is unable to process an MS sample until the 8* sampling event, due to laboratory sample
processing load, the second MS sample can be processed no earlier than the 20th sampling event.
EPA Method 1622/1623 specifies the following additional requirements for MS sample analyses:
• The MS sample volume analyzed must be within 10% of the volume analyzed for the associated field
sample.
• The MS sample must be analyzed in the same QC batch as the field sample, using the same method.
• The MS sample must be collected as a split sample or immediately before or after the associated field
sample.
Under the LT2 rule, If the volume of the MS sample is greater than 10 L,' the system is permitted to filter
all but 10 L of the MS sample in the field, and ship the filtered sample and the remaining 10 L of source
water to the laboratory to have the laboratory spike the remaining 10 L of water and filter it through the
filter used to collect the balance of the sample in the field [40 CFR part 141.705 (a)(2)(ii)].
3.3 Archiving Examination Results
Although not required, laboratories also can archive slides and/or take photographs of slides to maintain
for clients. Slides should be stored in a humid chamber in the dark' at 0°C to 10°C. An alternative
mounting medium also may be used, which may potentially preserve slides longer. Details are provided in
the Microbial Laboratory Guidance Manual for the Long Term 2 Enhanced Surface Water Treatment
Rule (LT2 Rule). -
Draft . 17 June 2003
-------
SECTION 4: UNDERSTANDING £. coo ANALYSES
As noted in Section 1, E. coli and turbidity data generated under the LT2 rule are used differently for
large systems than small systems. E. coli and turbidity are reported with Cryptosporidium data by large
systems to enable EPA to determine whether an E. coli trigger level can be established through the
microbial index. If a defensible trigger level can be established between K coli concentrations and
Cryptosporidium levels, small systems will be able to perform less-expensive E. coli analyses initially to
determine whether more expensive Cryptosporidium monitoring is even necessary.
Although E. coli data will not be used to determine whether additional treatment is needed for large
systems, as Cryptosporidium data will, it is nonetheless critical that the large systems generate reliable E.
coli data to establish relevant trigger levels for use by the small systems. The E. coli data generated by
small systems will be used to determine whether Cryptosporidium monitoring is required, so it is critical
that these data-be reliable, as well.
This section provides utility personnel unfamiliar with E. coli sample analyses with an overview of the
methods used under the LT2 rule and the quality control (QC) measures the laboratory uses to verify data
quality.
4.1 Summary of LT2 Rule E. coli Methods
E. coli sample analyses performed under the LT2 rule must be quantitative; presence/absence E. coli
results are unacceptable under LT2. The methods described below are approved for the analysis of E. coli
samples under the LT2 rule [40 CFR part 141.705 (b)].
4.1.1 Most Probable Number (MPN) Methods
4.1.1.1 Standard Methods 9223B: Colilert® and Colilert-18®
Colilert® and Colilert-18® tests are chromogenic/fluorogenic enzyme substrate tests for the simultaneous
determination of total coliforms and E. coli in water. These tests use commercially available media
containing the chromogenic substrate ortho-nitrophenyl-p-D-galactopyranoside (ONPG), to detect total
coliforms and the fluorogenic substrate 4-methylumbelliferyl-p-D-glucuronide (MUG), to detect E. coli.
Media formulations are available in disposable tubes for the multiple-tube procedure or packets for the
multiple-well procedure. Appropriate preweighed portions of media for mixing and dispensing into
multiple-tubes and wells are also available. The use of commercially prepared media is required for
quality assurance and uniformity. All tests must be conducted in a format that provides quantitative
results [40 CFR part 141.705 (b)].
• Multiple-Tube. For the multiple-tube procedure, a well-mixed sample and/or sample dilution/volume
is added to tubes containing predispensed media. Tubes are then capped and mixed vigorously to
dissolve the media. Alternatively, this procedure can be performed by adding appropriate amounts of
substrate media to a bulk diluted sample (with appropriate dilutions for enumeration), then mixing
and dispensing into multiple-tubes. A 15-tube MPN should be used to obtain quantitative results. The
Draft
18
June 2003
-------
Section 4: Understanding E. coli Analyses
number of dilutions/volumes are determined based on the type, quality, and character of the water
sample.
• Multiple-Well. A multiple-well procedure may be performed with sterilized disposable packets. The
commercially available Quanti-Tray® or Quanti-Tray®/2000 multiple-well tests use Colilert® or
Colilert-18® media to determine E. coli (IDEXX, 1999b,c). In these tests, the packet containing
media is added to a 100-mL sample (or appropriate dilutions for enumeration). The sample is then
mixed and poured into the tray. A tray sealer separates the sample into 51 wells (Quanti-Tray) or 97
wells (Quanti-Tray/2000) and seals the package. . .
After the appropriate sample dilutions/volumes are added, the tubes or trays are incubated at 35°C ±
0.5°C for 18 h when using Colilert-18® or 24 h when using Colilert®. If the response is questionable
after the specified incubation period, the sample is incubated for up to an additional 4 h at 35°C ±
0.5°C for both Colilert® tests. Each tube or well is then compared to the reference color "comparator"
provided with the media. A yellow color greater or equal to the comparator indicates the presence of
total coliforms in the sample, and the tube or well is then checked for fluorescence under long-
wavelength UV light (365-nm). The presence of fluorescence greater than or equal to the comparator
is a positive test for E, coli. If water samples contain humic acid or colored substances, inoculated
tubes or wells should also be compared to a sample water blank without Colilert® reagent added. The
concentration in MPN/100 mL is then calculated from the number of positive tubes or wells using
MPN tables provided by the manufacturer.
4.1.1.2 Standard Methods 9221B/9221F: LTB -EC-MUG
The multiple-tube fermentation method for enumerating E. coli in water uses multiple-tubes and
dilutions/volumes in a two-step procedure to determine E. coli concentrations. In the first step, or
"presumptive phase," a series of tubes containing lauryl tryptose broth (LTB) are inoculated with
undiluted samples and/or dilutions/volumes of the samples and mixed. Inoculated tubes are incubated for
24 ± 2 h at 35°C ± 0.5°C. Each tube then is swirled gently and examined for growth (i.e., turbidity) and
production of gas in the inner Durham tube. If there is no growth, acid, or gas, tubes are re-incubated for
24 ± 2 h at 35°C ± 0.5°C and re-examined. Production of growth and gas within 48 ± 3 h constitutes a
positive presumptive test for coliforms, which include E. coli.
After enrichment in the presumptive medium, positive tubes are subjected to a second step for
enumeration of E. coli. Presumptive tubes are agitated, and growth is transferred using a sterile loop or
applicator stick to tubes containing EC broth supplemented with 4-methylumbelliferyl-p-D-glucuronide
(MUG). Inoculated tubes are incubated at 44.5°C ± 0.2°C for 24 ± 2 h in a water bath. All tubes
exhibiting growth and gas production are examined for bright blue fluorescence under long-wavelength
UV light (366-nm) indicating a positive test for E. coli. The density of £. coli in MPN/100 mL is then
calculated from the number of positive EC-MUG tubes, using MPN tables or formulas. A 15-tube MPN is
required under the LT2 Rule.
4.1.2 Membrane Filtration (MF) Methods
4.1.2.1 Standard Methods 9222B/9222G: mEndo/LES-Endo-NA-MUG and Standard Methods
9222D/9222G: mFC-NA-MUG
These membrane filter methods for enumerating E. coli are two-step incubation procedures. First, a
sample is filtered through a 0.45 ^m filter, then the filter is placed on a pad saturated with mEndo broth or
a plate containing mEndo or LES-Endo agar and incubated for 24 ± 2 h at 35°C ± 0.5°C. Pink to red
colonies with a metallic (golden-green) sheen on the filter are considered to be total coliforms. If initial
determination of fecal coliforms is desired, mFC media can be substituted for mEndo/LES-Endo.
Following initial isolation of total coliforms (or fecal coliforms), the filter is transferred to nutrient agar
Draft
19
June 2003
-------
.Section 4: Understanding E. coli Analyses
containing 4-tnethylumbelliferyl-p-D-glucuronide (NA-MUG) and incubated for 4 h at 35°C ± 0.5°C.
Sheen colonies on mEndo or blue colonies on mFC that fluoresce under a long-wavelength UV light
(366-nm) are positive for E. coli. If high levels of non-£. coli total coliforms interfere with the ability to
accurately enumerate E. coli despite additional dilutions, transfer from mFC or an alternate method (e.g.,
SM 9213D, EPA Method 1603) should be used. ,
4.1.2.2 Standard Methods 9213D: mTEC
The mTEC agar method is a two-step procedure that provides a direct count of £. coli in water, based on
the development of colonies on the surface of a membrane filter when placed on a selective nutrient and
substrate media. This method originally was developed by EPA to monitor the quality of recreational
water. This method was also used in health studies to develop the bacteriological ambient water quality
criteria for E, coli. In this method, a water sample is filtered through a 0.45|am membrane filter, the filter
is placed on mTEC agar (a selective primary isolation medium), and the plate is incubated first at 35°C ±
0,5°C for 2 h to resuscitate injured or stressed bacteria and then at 44.5°C ± 0.2°C for 22-24 h in a water
bath. Following incubation, the filter is transferred to a filter pad saturated with urea substrate medium.
After 15 minutes, all yellow or yellow-brown colonies (occasionally yellow-green) are counted as
positive for E. coli using a fluorescent lamp and either a magnifying lens or a stereoscopic microscope.
4.1.2.3 EPA Method 1603: Modified mTEC
The modified mTEC agar method is a single-step MF procedure that provides a direct count of £. coli in
water based on the development of colonies on the surface of a filter when placed on selective modified
mTEC media. This is a modification of the standard mTEC media that eliminates bromcresol purple and
bromphenol red from the medium, adds the chromogen 5-bromo-6-chloro-3-indolyl-p-D-glucuronide
(Magenta Glue), and eliminates the transfer of the filter to a second substrate medium. In this method, a
water sample is filtered through a 0.45fim membrane filter, the filter is placed on modified mTEC agar,
incubated at 35°C ± 0.5°C for 2 h to resuscitate injured or stressed bacteria, and then incubated for 22-24
h in a 44.5°C ± 0.2°C water bath. Following incubation, all red or magenta colonies are counted as E.
coli.
4.1.2.4 EPA Method 1604: MI Medium
The MI medium method is a single-step membrane filtration procedure used to simultaneously enumerate
total coHforms and E. coli. In this EPA-developed method, a water sample is filtered through a 0.45-|jm
membrane filter, the filter is placed on an MI agar or broth plate, and the medium is incubated at 35°C ±
0.5°C for 24 h. If high levels of non-£. coli total coliforms interfere with the ability to accurately
enumerate E. coli despite additional dilutions, an alternate method (e.g., SM 9213D, EPA Method 1603)
should be used.
E. coli colonies exhibit a blue color and also may fluoresce under a long-wavelength UV light (366-nm).
If desired, the plates can also be observed under long-wavelength UV light (366-nm) for the presence of
total coliform species that fluoresce. Because the blue color from the breakdown of indoxyl-p-D-
glucuronide (IBDG) can mask fluorescence, non-fluorescent blue colonies are included in the total
coliform count. Water samples with high turbidity can clog the membrane filter, interfering with filtration
and potentially interfering with the identification of target colonies.
Draft
20
June 2003
-------
Section 4: Understanding E. coli Analyses
4.1.2.5 m-ColiBlue24® Broth
This broth method is a single-step MF test for enumerating total coli forms and E. coli. As with NA-MUG,
modified mTEC, and MI media, the selective identification of E, coli is based on the detection of the |J-
glucuronidase enzyme. The test medium includes the chromogen 5-bromo-4-chloro-3-indoxyl-p-D-
glucuronide (BCIG or X-Gluc). The chromogen BCIG is hydrolyzed by p-glucuronidase, releasing an
insoluble indoxyl salt that causes the colonies to exhibit a blue color. M-ColiBlue24® broth is a
commercially available format of this method and contains a nutritive lactose-based medium containing
inhibitors to eliminate the growth of non-coliforms. With m-ColiBlue24® broth, a water sample is filtered
through a 0.45nm membrane filter, and the filter is transferred to a plate containing an absorbent pad
saturated with m-ColiBlue24® broth. The filter is incubated at 35°C ± 0.5°C for 24 h and examined for
colony growth. The presence of E. coli is indicated by blue colonies. The presence of total coliforms
(noh-£. coli) is indicated by red colonies. If enumeration of total coliforms is desired, blue and red
colonies should be included in the total coliform count. If high levels of non-£. coli total coliforms
interfere with the ability to accurately enumerate E. coli despite additional dilutions, an alternate method
(e.g., SM 9213D, EPA Method 1603) should be used.
4.2 E. coli Laboratory Quality Control
E. coli sample results reported under the LT2 rule should meet the quality control (QC) specifications set
forth in the approved versions of the methods described above. Sections 4.2.1 - 4.2.7 describe quality
control specifications for E. coli analyses performed under the LT2 rule. This guidance is provided to
help summarize the QC specifications in the methods and does not substitute for or alter the method
specifications. Sample results that do not meet these specifications are not considered valid, and cannot be
reported under the LT2 rule. Additional information on the QC specifications is available in Section 4.2
of the Microbial Laboratory Guidance Manual for the Long-Term 2 Enhanced Surface Water Treatment
Rule (LT2 Rule).
4.2.1 Dilution/Rinse Water Sterility Check
Each batch (or lot, if commercially prepared) of dilution/rinse water should be checked for sterility by
adding 50 mL of water to 50 mL of a double-strength non-selective broth (e.g., tryptic soy, trypticase soy,
or tryptose broth). Incubate at 35°C ± 0.5°C, check for growth after 24 hours and 48 hours (or for the
longest incubation time specified in the method), and record results. The dilution/rinse water batch should
be discarded if growth is detected.
4.2.2 Media Sterility Check
To test sterility of newly prepared media prior to the analysis of field samples, incubate one plate per each
media batch at the appropriate temperature for 24 and 48 hours (or for the longest incubation time
specified in the method) and observe for growth. If any contamination is observed, determine the cause,
correct, and reject any data from samples tested with the media.
4.2.3 Positive/Negative Controls
For each new lot or batch of medium, check the analytical procedures and integrity of the medium before
use by testing with known1 positive and negative control cultures. Laboratories using commercially-
prepared media with manufacturer shelf-lives of greater than 90 days should run positive and negative
controls each quarter in addition to running the batch/lot-specific controls and sterility checks.
Laboratories are encouraged to perform positive and negative control tests each day that field samples are
analyzed. Positive and negative controls should be chosen based on the method-specific requirements. For
example if a 44.5°C water bath is not required by the method, it is not necessary to include Enterobacter
aerogenes as a negative control.
Draft . 21 June 2003
-------
Section 4: Understanding E. coli Analyses
4.2.4 Media Storage
The following media storage specifications should be met for E. coli analyses:
• Agar plates may be held for up to 2 weeks at 1°C to 5°C in plastic bags or containers. Protect media
containing dyes from exposure to light.
Broth in loose fitting caps (e.g., snap caps) should be stored at 1°C to <30°C for no more than 2
weeks
• Broth in tight fitting caps (e.g., screw caps) should be stored at 1°C to <30°C for no longer than 3
months
* All media should be at room temperature prior to use ' -
• Media exhibiting growth or gas should be discarded '
4.2.5 Filtration Unit Sterilization
Membrane filter equipment should be autoclaved before the beginning of a filtration series. A filtration
series ends when 30 minutes or longer elapses after a sample is filtered. Ultraviolet (UV) light (254 nm)
may be used to sanitize equipment (after initial autoclaving for sterilization), if all supplies are pre-
sterilized. UV light can also be used to reduce bacterial carry-over between samples during a filtration
series. The UV lamp should be tested quarterly with a UV light meter or an agar plate. Appropriate
corrective actions should be taken, if necessary.
4.2.6 Preparation Blanks
Preparation blanks should be analyzed to detect potential contamination of dilution/rinse water during the
course of analyses.
4.2.6.1 Membrane Filter Preparation Blank
If membrane filtration is used, an MF preparation blank is performed at the beginning and the end of each
filtration series by filtering 20-30 mL of dilution water through the membrane filter and testing for
growth. If the control indicates contamination with the target organism, all data from affected samples
should be rejected. A filtration series ends when 30 minutes or more elapse between sample filtrations.
4.2.6.2 Most Probable Number Preparation Blank
EPA recommends that a volume of sterilized, buffered water be analyzed exactly like a field sample each
day samples are analyzed. The preparation blank should be incubated with the sample batch and observed
for growth of the target organism. If the control indicates contamination with the target organism, all data
from affected samples should be rejected.
4.2.7 Verification
Verification specifications are detailed in the Certification Manual (Reference 9.3), Standard Methods
(Reference 9.4), and Appendices J through L of the Microbial Laboratory Guidance Manual for the
Long-Term 2 Enhanced Surface Water Treatment Rule (LT2 Rule).
Draft
22
June 2003
-------
SECTION 5: CONTRACTING FOR CRYPTOSPORIDIUM
LABORATORY SERVICES
Although many public water systems (PWSs) have established procedures and policies governing the
purchase of services and supplies, these procedures seldom lend themselves to the purchase of analytical
services. This section provides a basic framework for defining the technical and contractual requirements
associated with purchasing laboratory services for Cryptosporidium analyses for the LT2 rule, awarding
contracts, and working with a contract laboratory.
Successfully contracting for Cryptosporidium laboratory services for LT2 rule monitoring relies on the
following steps:
Step 1: Define the scope of your analytical requirements to develop a detailed contract and
standardized bid sheet
Step 3: Solicit qualified laboratories
Step 4: Award contracts to a primary laboratory and a backup laboratory
Step 5: Work closely with your laboratory before monitoring begins and maintain communications
throughout monitoring
Each of these general steps, and details on the activities associated with each, are discussed in Sections
5.1 through 5.5.
5.1 Defining Your Needs and Developing a Contract
The first step in developing an analytical services contract for Cryptosporidium analyses for LT2 rule
monitoring is identifying the "who, " "what, " "when, " and "how " of the project for your system (the
"why" is the LT2 rule itself). A well-written contract will address each of these issues, as well as the
administrative issues, such as laboratory payments and adjustments.
The best way to ensure that you get the data you need for LT2 rule Cryptosporidium monitoring within
the required time period is to specify your requirements in detail in the contract. A well-written contract
can minimize or eliminate many common problems in procuring analytical services, and enable you to
collect reliable and timely results.
Recommendations on the factors to consider in defining the scope of the services you need, and the
information you should be sure to include in your contract are provided below.
Draft 23 June 2003
-------
Section 5; Contracting for Cryptosporidium Laboratory Services
5.1.1 Client Information
"Who" defines your PWS to the laboratories that you would like to submit bids for the project. Will you
be contracting for laboratory services for a single plant or will this contract require Cryptosporidium
analyses to fulfill monitoring requirements for multiple plants in a system?
Clearly identify in your contract the name and identification number of your PWS, as well as
the name(s) and identification number of the facility(ies) for which samples need to be
analyzed. This information ultimately will be used to identify your samples in the L T2 bata
Collection System, and the laboratory you use for Cryptosporidium sample analyses will need
to know this information. (Alternately, you can provide this information after award to the
awarded laboratory only.)
5.1.2 Sample Information
"What" describes the samples to be analyzed. As noted in Sections 5.1.2.1 through 5.1.2.5, this
encompasses a variety of factors, each of which needs to be evaluated and defined before you develop
your contract.
5.1.2.1 Number of Samples
What is the total number of samples the laboratory will need to analyze? This total includes not only
routine monitoring samples (field samples), but also the matrix spike (MS) samples (Section 3.2.7) that
are required at a frequency of 1 per 20 field samples. Field samples and MS samples are considered
"billable" samples (sample analyses for which the laboratory will be paid their per-sample cost). Internal
laboratory quality control (QC) samples, such as method blanks and ongoing precision and recovery
(OPR) samples should be considered "unbillable" samples—sample analyses that are required, but apply
to multiple PWS clients. Rather than charging clients for these samples directly, laboratories typically
will amortize the costs of these samples across billable samples.
If a sample is collected and sent to the laboratory, but cannot be submitted under the LT2 rule because of
a problem unrelated to laboratory performance (such as shipping delays that violate the sample holding
time), your PWS will be required to collect a "make-up" sample (see Section 6.3 for details). You should
add, as an option to be exercised at your direction in such an event, two additional sample analyses to the
total.
Clearly indicate in your contract the total number of: (1) field samples and (2) MS samples
that the laboratory will be required to analyze. Add two additional, optional, sample analyses
to be exercised if "make-up " samples are required due to problems unrelated to laboratory
performance.
5.1.2.2 Type of Samples
Will your PWS collect and ship bulk water samples to the laboratory for filtration and processing or will
your PWS filter samples on-site and ship the filter to the laboratory? Shipping and analytical costs are
likely to be lower if you filter your samples on-site, but you will need to purchase or rent sample filtration
equipment (see Section 6.4 for details) and have staff trained to use the required procedures or pay for the
laboratory or another firm to perform these tasks.
Draft
24
June 2003
-------
Section 5: Contracting for Cryptosporidium Laboratory Services
Clearly specify in the SOW whether the laboratory will receive bulk water samples or
filtered samples. If filtered samples will be sent, indicate which filter you will use (see
Section 5.1.4.2).
If you will be filtering on-site, and will be using your own equipment to filter the samples, you should
consider purchasing filters directly from the vendor, rather than from the laboratory, to reduce costs.
(Additional information on filtering samples on-site and purchasing filters is provided in Section 6.4.2).
v If your PWS will be purchasing filters directly, specify this in the contract, so the
laboratory knows not to include this in their per~sample price.
5.1.2.3 Anticipated Sample Volume
The LT2 rule will require that at least 10 L be analyzed for each sample (with some exceptions - see
Section 6.1) [40 CFR part 141.705 (a)(l)]. Will your PWS collect 10-L samples or collect higher-volume
samples, such as 50-L samples? If your PWS will be shipping bulk water samples to the laboratory,
greater sample volumes will result in higher shipping costs and will likely result in higher analytical costs.
If your PWS will be filtering samples on-site, and shipping filters to the laboratory, the sample volume
should not affect shipping or analytical costs, but the greater sample volumes filtered may result in higher
packed pellet volume and multiple subsamples (Section 5.1.2.4).
ef Clearly indicate in your contract the volume you anticipate collecting for each sample.
5.1.2.4 Subsamples and Filter Clogs
As noted in Section 3.1, additional steps are required at the laboratory for samples that generate a larger
packed pellet volume than can be processed as one sample through the method's purification step.
Specifically, the laboratory will need to process the packed pellet from the sample as two or more
"subsamples" through the remainder of the method (purification, staining, and examination) to meet LT2
rule sample volume analysis requirements. If a sample clogs before 10 L have been filtered, at least two
filters must be used to meet LT2 rule sample volume analysis requirements [40 CFR part 141.705 (a)(l)].
If the.source water(s) to be monitored by your PWS are characterized by high turbidity, some of your
samples may need to be processed as multiple subsamples or may require two filters to enable you to
meet LT2 rule monitoring requirements. Even if your source water(s) typically is characterized by low
turbidity, you should allow for the possibility that some samples may result in larger packed pellet
volumes on occasion. By including this in the original contract, you will avoid changes to the contract on
short notice if subsamples are required during monitoring.
Clearly indicate in your contract that different sample prices are needed for: (1) full sample
analyses, (2) subsample analyses, and (3) extra filters and the cost of analysis of the extra
filters.
Draft 25 June 2003
-------
Section 5: Contracting for Cryptosporidium Laboratory Services
5.1.2.5 Extra Services
Will any additional services be required of the laboratory outside of actual sample analyses? Possible
services include:
• Sampling kit rental for on-site filtration
• Sample shipping containers
• Sample archiving (laboratories can archive slides and some can take photographs of slides to maintain
for clients)
Some of these services may be included in the sample analysis cost by some laboratories. Defining the
specific services your PWS will need, and specifying these services clearly in the contract will enable the
laboratories to better assess whether the requested services are included in their routine costs or are extra,
and respond accordingly.
«s* Clearly specify in your contract any services required in addition to routine sample analysis.
5.1.3 Sampling Schedules
"When " refers to your anticipated schedule for shipping samples to the laboratory. Will your PWS begin
monitoring before implementation of the LT2 rule with the intent to grandfather some or all of the data or
will your PWS monitor according to the rule schedule?
The minimum monitoring frequency for the LT2 rule is once per month [40 CFR part 141.701 (e)].
During LT2 monitoring, will your PWS collect and ship samples once per month, or will you monitor
more often?
If at all possible, do not establish a firm sampling schedule with specific dates at this point. Most of the
laboratories available to perform Cryptosporidium analyses have multiple PWS clients and need to evenly
distribute their sample load within each week and across weeks in a month to meet holding time
requirements. Rather than dictating a sample collection schedule to the laboratory-and potentially
discouraging laboratories from bidding on the work or risk violating holding times during
monitoring—work with the awarded laboratory to establish a schedule that is will comply with LT2 rule
requirements and is mutually acceptable to your PWS and the laboratory:
Indicate in your contract the month that you plan to begin monitoring and whether you will be
monitoring on a monthly or more frequent basis. If possible, do not specify actual sample
collection dates and days during the week; work with the awarded laboratory to establish a
schedule that meets your needs and does not cause problems for the laboratory.
5.1.4 Analytical Methodology
"How " describes the analytical method that the laboratory will use. This involves two sets of options:
which method to use (EPA Method 1622 or EPA Method 1623) and which filter to use, regardless of
method. It also refers to the QC requirements that must be met during sample processing and analysis.
Draft
26
June 2003
-------
Section 5: Contracting for Cryptosporidium Laboratory Services
5.1.4.1 EPA Method 1622 Versus EPA Method 1623
Will your PWS monitor for Cryptosporidium only or Cryptosporidium and Giardial Most laboratories
analyze samples for both Cryptosporidium and Giardia using EPA Method 1623. If EPA Method 1623 is
used by the laboratory to analyze your LT2 rule samples, only Cryptosporidium data need to be
submitted. If Giardia data are collected, they do not need to be submitted to EPA.
Your contract should specify that EPA Method 1622 be used only if you are interested in monitoring for
Cryptosporidium only (this method only detects Cryptosporidium). Although reagent costs for this
method are slightly less than for EPA Method 1623, actual sample analysis costs may not be lower
because laboratories may not be able to allocate the QC sample costs for this method across as many
clients.
5.1.4.2 Filter Options
Although EPA validated EPA Method 1622 and EPA Method 1623 using one filter type, modified
versions of the methods using alternate filter options have been approved by EPA since validation. The
following available filters are considered acceptable by EPA for use with EPA Methods 1622 and 1623:
•. Original Pall Gelman Envirochek™ capsule filter
• IDEXX Filta-Max™ foam filter
• Pall Gelman Envirochek™ HV capsule filter
Unless your PWS has experience with Cryptosporidium sampling, and a basis for requesting a specific
filter type, you should indicate in the contract that all are acceptable.
If your PWS has experience monitoring for Cryptosporidium and has a filter preference, you will need to
indicate this to the laboratories interested in bidding on the project, as not all laboratories are approved by
EPA through the Lab QA Program to perform all versions of the methods.
*** If your PWS has experience with Cryptosporidium sampling and would like analyses
performed using a specific filter, clearly indicate this in the contract. Otherwise, do not
specify a filter type.
5.1.4.3 Quality Control Requirements
t
Although EPA Methods 1622 and 1623 specify the QC requirements that must be met during
performance of the method, your contract should reiterate that the following QC tests must be performed
at the required frequency during processing and analysis of your samples:
• Method blank test (Section 3.2.2)
• Ongoing precision and recovery (OPR) test (Section 3.2.3)
• Holding time requirements (Section 3.2.4)
• Staining controls (Section 3.2.5)
None of these QC measures should be billable, however. As noted above, in Section 5.1.2.1, the costs for
the'method blank, OPR, and staining control tests should be amortized by the laboratory across the cost of
monitoring samples for all of their clients.
Draft 21 June 2003
-------
Section 5: Contracting for Cryptosporidium Laboratory Services
Reiterate in the contract that method blanks, OPRs, and staining controls must be performed
at the frequency required in the method, and that all holding times must be met. I
5.1.5 Data Deliverables and Other Contract Issues
In addition to the "who," "what, " "when, " and "how" questions that need to be addressed by the
contract, you also will need to provide details on data delivery, adjustments for lateness, and sample
reanalysis cost issues. These issues are discussed in Sections 5.1.5.1 through 5.1.5.5.
5.1.5.1 Data Submission
EPA has developed the web-based LT2 Data Collection System to allow laboratories to report data to
PWSs electronically and allow PWSs to verify the data electronically before submitting the monitoring
results to EPA. This reporting process is summarized in Section 7.2 for Cryptosporidium data, and
discussed in detail in the Users' Manual for the Long Term 2 Enhanced Surface Water Treatment Rule
(LT2 Rule) Data Collection System. The laboratory, at a minimum, will need to submit the results for each
Cryptosporidium monitoring sample to you electronically. (Although your PWS also could enter these
data, based on hardcopy results from the laboratory, this is strongly discouraged, as the potential for error
increases when personnel unfamiliar with the generation of the data for a sample enter these data into the
LT2 Data Collection System.)
*s* Clearly indicate in your contract that the laboratory is required to enter Cryptosporidium
monitoring results for your samples into the LT2 Oata Collection System.
5.1.5.2 Hardcopy Data Deliverables
Note: If you do not intend to review all of the raw data generated by the laboratory, this section is not
relevant, andean be ignored-If your PWS does intend to review all of the raw data associated with your
LT2 samples (discussed in Section 7), you should request copies of the forms used by the laboratory to
record sample measurements, sample processing times, and sample examination results, as well as
information on the QC samples associated with your monitoring sample. (If your PWS will store and
maintain all sample results, rather than the laboratory, then the original forms should be requested.)
»
Suggestions for the materials that should be requested include the following:
• Sample result summary sheet, which should include the following:
• Monitoring sample identification information
• Monitoring sample result, in oocysts/L
• Laboratory quality control batch associated with the sample
• ID number and result for the ongoing precision and recovery (OPR) sample analyzed for this QC
batch
• ID number and result for the method blank sample analyzed for this QC batch
• LT2 sample collection form initiated by your utility and completed with sample receipt information
by the laboratory
Draft 28 • . June 2003
-------
Section 5: Contracting for Cryptosporidium Laboratory Services
Method 1622/1623 Bench Sheet with raw data associated with the monitoring sample (and MS
sample, if applicable)
Method 1622/1623 Cryptosporidium Slide Examination Form with raw data for the monitoring
sample (and MS sample, if applicable)
Laboratory comments. If the laboratory provided comments on the sample analyses or results that
require follow-up, contact the laboratory to discuss, if necessary. Comments may include any
applicable data qualifiers. The following is a list of potential data qualifiers:
• The recovery for the associated ongoing precision and recovery (OPR) sample did not meet
method requirements
• Oocysts were detected in the method blank
• Positive and negative staining controls were not acceptable or not examined
.
• Method holding times were hot met
• Sample arrived at the laboratory in unacceptable condition
If you need the laboratory to submit hardcopy results (this is not necessary, unless you
intend to review all of the raw data), clearly indicate in your contract the materials that are
required.
5.1.5.3 Data Turnaround Requirements
Under the LT2 rule, PWSs are required to submit data no later than 10 days after the end of the first
month following the month when the sample is collected (approximately 40 to 70 days after sample
collection, depending on when during the month the sample is collected) [40 CFR part 141.707 (d)J. For
example, if a sample is collected on March 17, data must be submitted by May 10.
The required data turnaround must be stated clearly in the contract. This turnaround time should be
expressed in calendar days (not working days), and should start from the sample collection date. The data
turnaround time calculations should consider the day that the sample is collected "day zero," and the
following day as "day one." (Data turnaround times in analytical contracts typically start from the receipt
of the sample at the laboratory, but calculating it from the sample collection date is more logical in this
case because the LT2 rule's data submission requirements are based on sample collection date.)
If the data turnaround time starts from sample collection, rather than sample receipt by the laboratory, this
turnaround should accommodate the potential for shipping delays that will be outside of the laboratory's
control. As a general rule, the data turnaround time should not be shorter than the sum of the maximum
holding times in the method—15 days. This includes up to 4 days between sample collection and
initiation of the elution step, which effectively is the maximum time for any shipping delay, as samples
received more than 4 days after collection will not be valid, and cannot be submitted through the LT2
Data Collection System.
Using the 15 days allowed for sample analysis by the methods (plus additional time to compile the data
package and mail the results) as the shortest realistic turnaround time, determine when you will actually
need the results. The same turnaround time can be specified for both submission of electronic data and
receipt of hardcopy materials.
Draft 29 June 2003
-------
Section 5: Contracting for Cryptosporidium Laboratory Services
Do not specify a data turnaround time shorter than you really need, as it may increase the per-sample
price quoted by the laboratories. This turnaround time should be short enough to provide time to carefully
evaluate the results before they must be submitted to EPA, but long enough that it does not unreasonably
burden the laboratory and potentially increase the per-sample quotes you receive when you solicit the
project.
Specify in the contract the data turnaround requirement for electronic andhardcopy
submission of data. This turnaround time should be calculated as the time between sample
collection and receipt of the Hardcopy data by your PWS.
5.1.5.4 Liquidated Damages and Penalties
You should consider including penalty or damage clauses in your contracts as incentives to preclude
laboratories from submitting data late or performing analyses improperly. Due to the nature of the
services provided, it is often difficult to assess actual damages caused by improperly performed analyses.
Liquidated damages often are used in analytical services contracts in lieu of actual damages. Liquidated
damages typically specify that, if the laboratory fails to deliver the data specified in the deliverables
section of the contract, or fails to perform the services within the specified data turnaround time, the
laboratory will pay a fixed, agreed, price to compensate the organization to whom the services should
have been delivered. For example, some EPA contracts for analytical services specify that the laboratory .
will pay, as fixed, agreed, and liquidated damages, 2% of the analysis price per calendar day of delay, to a
maximum reduction of 50% of the analysis price.
If liquidated damages or penalties are involved, they should (1) be based on actual damage caused (in
terms of cost) by each day of lateness, (2) be strong enough to discourage late delivery, and (3) be
reasonable enough that they will not discourage laboratories from bidding. If liquidated damages or
penalties will be applied to meet the required data turnaround time, this information should be included.
The contract should specify that the laboratory will not be charged with liquidated damages when the
delay in delivery or performance arises out of causes beyond the control and without the fault or
negligence of the laboratory. It also may be necessary to limit damages to a certain dollar value or scope.
Other types of damages that should be considered, and may be included in the contract, include costs for
resampling and administrative costs associated with the evaluation and processing of unacceptable data
(data that do not meet the requirements specified in the contract or the QC requirements specified in the
analytical method).
Clearly indicate in your contract whether liquidated damages will be applied to late data or
other problems, how these liquidated damages are calculated,, and the limits and conditions
associated with the damages.
5.1.5.5 Re-Analysis Costs
Every laboratory periodically produces data that are associated with unacceptable QC data or are invalid
for other reasons. The contract should stipulate that the laboratory will reanalyze samples at no cost to
your PWS if the problems are due to laboratory error. If the problems are due to an error outside of the
laboratory's control (such as the laboratory's rejection of a sample received at > 10°C that results in
resampling by your PWS), the laboratory should not be responsible for the additional costs, that may
result.
Draft*
30
June 2003
-------
Section 5: Contracting for Cryptosporidium Laboratory Services
**" Clearly indicate in your contract when the laboratory would be required to bear the costs of
sample re-analysis costs and when these costs will be borne by your PWS.
The contract also should state that you have the right to inspect the results, and if they do not meet the
requirements in the contract, you have the right to reject the data, returning them to the laboratory without
payment. Rejection of data should be based on sound technical review of the results. It also obligates you
to make no use of those results-without making some payment to the laboratory. f
Clearly indicate in your contract that your PWS has the right to inspect results and reject
the results if they do not meet contract requirements.
5.2 Developing a Bid Sheet
After all project requirements have been established, you should develop a bid sheet to accompany the
analytical requirements summary during the solicitation. The bid sheet allows laboratories to submit bids
in the same format, making bid evaluations easier, and also helps to clarify the project. Development and
use of a bid sheet is recommended regardless of whether your PWS solicits the project competitively to
multiple laboratories, or is simply requesting a quote from a laboratory you already know you will be
using, as it provides a very clear vehicle for submitting and evaluating costs.
Bid sheets for analytical services typically are formatted as a chart, with analytical requirements along
one axis and number of samples and prices along the other.
The bid sheet should include the following information:
Project identifier (e.g. "LT2 Cryptosporidium Monitoring Sample Analyses for [PWS name and/or
facility name]")
• Space for laboratory identification information
• Day, date, and time (including time zone) of the bid deadline
• Bid submission information (contact and mailing address, fax number, and/or email address)
• Estimated award date
• Laboratory period of performance (period of time during which the laboratory is obliged to resolve
: issues associated with analysis of the samples—generally 6 months after shipment of last sample)
• Required delivery date (data turnaround time and the basis of its calculation, such as from collection
of each sample)
• Bid validity period (period of time during which bid prices are considered valid—generally 45 days
after the bid deadline; if the project is awarded after the period you specify, you must contact bidding
laboratories to determine whether their bid is still valid, or needs to be revised)
Draft 31 June 2003
-------
Section 5: Contracting for Cryptosporidium Laboratory Services
A summary of the analytical requirements:
* Method (e.g., Cryptosporidium and Giardia by EPA Method 1623)
• Filter preference, if any (this should not be specified, unless your PWS has experience with
Cryptosporidium, and a basis for requesting the use of a specific filter; if you know that you wil!
be field filtering using a specific filter, and shipping this to the laboratory, it is important that you
specify this)
• Whether samples will be shipped as filtered samples or bulk water samples
• Sample volume (e.g., 10 L, 50 L)
Total number of field samples to be analyzed, plus two extra, in case of "make-up" samples
Total number of MS samples to be analyzed
Total number of potential subsamples to be analyzed (expressed as "Up to [no.] subsamples")
• The number generally should not exceed three per sample
• If you have high-turbidity water, you may need to specify up to three subsamples for all of your
.field and MS samples
• If you have a low-turbidity water, you should specify a minimal number, just in case the need
arises
(These costs would not be incurred unless subsamples actually need to be analyzed)
Total number of potential extra filters (in case one or more samples clog during LT2 rule monitoring:
• If you will be shipping bulk samples to the laboratory, express this as "Up to [no.] extra
filters/elutions"
• If you will be filtering samples in the field, but receiving filters from the laboratory, express this
as "Up to [no.] extra filters"
(These costs would not be incurred unless more than one filter actually needs to be used)
Columns for laboratories to enter per-analysis and total costs
5.3 Soliciting the Contract
Procedures for soliciting and awarding contracts to perform analytical services can vary, depending upon
the scope of the project and purchasing requirements within the organization that is issuing the contract.
At one end of the spectrum are contracts that are awarded after placing a single phone call and obtaining a
quote from a single laboratory. The opposite end of the spectrum are contracts awarded after a
competitive solicitation and bidding process involving the distribution of a detailed project description
and a formal bid sheet via fax or mail.
5.3.1 Approved Laboratories
Regardless of whether you will be soliciting the project to multiple laboratories or working with a single
laboratory (although a backup laboratory is strongly recommended—see below), you will need to limit
your laboratories to only those approved by EPA through the Laboratory Quality Assurance Evaluation
Program for Analysis of Cryptosporidium Under the Safe Drinking Water Act (Laboratory QA Program)
(67 FR 9731, March 4,2002). Information on the Laboratory QA Program is posted on
Draft ,
32
June 2003
-------
Section 5: Contracting for Cryptosporidium Laboratory Services
http://www.epa.gov/safewiater/lt2/index.html and this program is described in detail in the Microbial
Laboratory Guidance Manual for the Long Term 2 Enhanced Surface Water Treatment Rule (LT2 Rule).
Briefly, the objectives of the program are to evaluate laboratories' capacity and competency to reliably
measure for the occurrence of Cryptosporidium in surface water using EPA Method 1622/1623. Each
laboratory participating in the program is required to complete the following steps to be qualified through
this program:
• Acceptably perform initial proficiency testing (IPT) on blind samples
Participate in an on-site evaluation of their technical, data management, and quality assurance
procedures
• Acceptably perform ongoing proficiency testing (OPT) on blind samples every four months
To improve Cryptosporidium data quality and consistency during LT2 rule monitoring, EPA requires that
only those laboratories approved for Cryptosporidium analysis under the Lab QA Program be used for
LT2 rule monitoring analyses [40 CFR part 141.706 (a)]. A list of laboratories approved through the Lab
QA Program is available from http://www.epa.gov/safewater/lt2/index.html.
5,3.2 Primary and Backup Laboratory Contracts
Because a laboratory's approval status may change during the LT2 rule monitoring period, you should
plan to award a primary contract and a backup contract. If no performance problems or other problems are
encountered during the LT2 rule monitoring period by the laboratory awarded the primary contract, then
this laboratory would provide uninterrupted sample analysis support for the entire monitoring period.
However, if the laboratory encountered performance problems and was disapproved, or was otherwise
unable to meet contract requirements, your PWS could switch sample analyses to the backup laboratory
under the contract you established with this laboratory before monitoring began.
The award of primary and backup contracts should be discussed in the contract solicitation. All other
things considered equal, the award for the primary contract could be made to the lowest responsive,
responsible bidder and the award for the backup contract could be made to the second lowest responsive,
responsible bidder.
5.4 Evaluating Bids
After the laboratories have received the solicitation and submitted their bids, you must evaluate the bids
to identify the laboratory that will be awarded the analytical services contract. Specific procedures for
evaluating bids may vary, depending upon the requirements of your organization, but the bid evaluation
process generally entails evaluation and comparison of each laboratory's proposed cost and capability to
meet the analysis requirements.
5.4.1 Identifying Responsive Bidders
You should consult your legal department or purchasing department to identify any applicable
requirements for evaluating competitive bids within their organization. At a minimum, however, you
should review all bids and recalculate subtotals and totals to ensure that the bidding laboratories did not
make any mathematical errors. In addition, you should verify that there are no unacceptable contingencies
associated with any of the bids, such as the use of a filter other than the filter that was specified in the
contract solicitation. Either eliminate from consideration bids from laboratories that bid with
contingencies or contact the laboratory(ies) to discuss the bid and verify that the laboratory cannot
perform the specified services.
Draft 33 June 2003
-------
Section 5: Contracting for Cryptosporidium Laboratory Services
Of the remaining (responsive) bids, identify the lowest bidder to award the primary contract and the
second lowest bidder to award the backup contract. If additional assessments of a laboratory's
performance or responsibility are needed, you may want to contact references.
5.4.2 References
If you have not worked with a particular laboratory before and would like to verify that the laboratory
will meet your needs throughout the monitoring period, you can ask the laboratory to provide contacts
and phone numbers of utility or government clients for which the laboratory has performed
Cryptosporidium sample analyses or other comparable services.
Questions to ask the references include:
• Did the laboratory provide data by the required due date?
• Were the data provided by the laboratory of acceptable quality and compliant with contract
requirements?
• Were laboratory personnel easy to work with when problems arose during ail phases of the project,
including sample scheduling, sample analysis, and data review? If problems were noted during data
review, was the laboratory prompt and responsive in addressing your concerns?
• Do you have any reservations in recommending this laboratory?
5.5 Communicating with the Laboratory
After the analytical services contract is awarded, you should request from the laboratory contact
information for the following roles, and provide the laboratory with PWS contacts for the same roles:
• A technical contact for analytical questions or problems
A sample control contact for shipping delays on the PWS end and sample receipt problems on the
laboratory end
• An administrative contact for invoicing and payment
Maintaining communications with the laboratory is critical to identifying and resolving problems quickly
and minimizing the need for resampling and reshipments. At a minimum, you should always notify the
laboratory of sample shipments and confirm that the laboratory received the sample on time and in
acceptable condition.
Although most communications are typically conducted over the phone, these communications also can
be conducted via email, which has the added benefit of providing your PWS and the laboratory with a ,
written record of sample receipt confirmations, problem notifications, and problem resolutions.
Draft
34
June 2003
-------
SECTION 6: COLLECTING AND
SHIPPING SOURCE WATER SAMPLES
Large systems (PWSs serving a population of at least 10,000 people) monitoring under the LT2 rule are
required to collect and analyze source water samples for Cryptosporidium, £. coll, and turbidity for a
minimum of 2 years. Small systems (PWSs that serve fewer than 10,000 people) are required to monitor
their source water for E. coli for a minimum of I year. A subset of small systems would then be required
to conduct Cryptosporidium analyses over a 1-year period if they exceed E. coli trigger levels (40 CFR
part 141.701).
Monitoring requirements for each system size and the schedule for each stage of monitoring is described
in Table 6-1.
Table 6-1. Summary of LT2 Rule Monitoring Requirements
Public water
system size
Large systems
(serving 10,000 or
more people)
Small systems
(serving fewer than
10,000 people)
Monitoring begins
6 months after
promulgation of LT2
rule
30 months (2 %
years) after
promulgation of LT2
rule
Monitoring
duration
2 years'
1 year1*
Monitoring parameters and sample
frequency requirements
Cryptosporidium
minimum
1 sample/month'
see below s
E. coli
minimum
1 sample/month8
1 every 2 weeks
5 Possible additional monitoring requirement for Cryptosporidium
If small systems exceed £ coli trigger levels, then. ..
Small systems
(serving fewer than
10,000 people)
48 months (4 years)
after promulgation
ofLT2rule
1 year
2 sample/month
N/A
PWSs may be eligible to use historical (grandfathered) data in lieu of these requirements if certain quality
assurance and quality control criteria are met (see Section 2)
" Small systems may be required to monitor for Cryptosporidium for 1 year, beginning 6 months after completion of
E. coli monitoring; Cryptosporidium monitoring would be required if the E. coli annual mean concentrations exceed
10 E. coW100 ml for systems using lakes/reservoirs or exceed 50 £. cofi/100 ml_ for systems using flowing streams
c PWSs monitoring for Cryptosporidium may collect more than 1 sample per month if sampling is evenly spaced over
the monitoring period
" Large unfiltered systems are required to conduct source water monitoring that includes Cryptosporidium sampling
only
N/A = Not applicable. No monitoring required.
Draft
35
June 2003
-------
Section 6: Collecting and Shipping Source Water Samples
6.1 Sample Volumes
Sample volume guidance is provided in Section 6.1.1 for Cryptosporidium samples and Section 6.1.2 for
E. coli samples.
6.1.1 Cryptosporidium Samples
Under LT2 rule Cryptosporidium sample volume requirements [40 CFR part 141.705 (a) (1)], PWSs are
required to analyze, at a minimum, either:
• 10 L of sample, or
• 2 mL of packed pellet volume, or
• As much volume as two filters can accommodate before clogging (this condition applies only to
filters that have been approved by EPA for nationwide use with EPA Method 1622/1623—the Pall
Gelman Envirochek™ and Envirochek™ HV filters, or the IDEXX FiltaMax™ foam filter).
The LT2 rule sample volume analysis requirement of 10 L (rather than 10.0 or 10.00 L) accommodates
the potential for imprecisely filled sample containers or filters. Sample volumes s ##.5 L would be
rounded up and sample volumes s ##.4 L would be rounded down. For example, 9.8 L would be rounded
to 10 L, and would meet rule requirements.
Systems may analyze larger volumes than 10 L, and larger volumes analyzed should increase analytical
sensitivity (detection limit), provided method performance is acceptable. EPA encourages systems to
analyze similar sample volumes throughout the monitoring period. However, data sets including different
samples volumes will be accepted, provided the system analyzes the minimum sample volume
requirements noted above.
PWSs with highly turbid water may be able to collect the required minimum packed pellet volume by
avoiding filtration altogether, and shipping a bulk water sample to the laboratory for centrifugation. The
laboratory can mix the sample thoroughly and centrifuge 250-mL or greater aliquot volumes sequentially
according to Section 13.2 of Method 1622/1623, until 2 mL of packed pellet volume is generated.
If the PWS encounters variable water quality that clogs the filter unpredictably, the PWSs should
routinely bring two filters plus a cubitainer to the sampling point for each sampling event:
• If the water quality allows a full 10 L to be filtered without clogging, the PWS can simply ship the
filter to the laboratory and save the remaining materials for subsequent events.
• If the first filter clogs after 5 L or more have been filtered, and the volume is not anticipated to yield 2
mL of packed pellet volume, the PWS should be able to filter the remaining volume through the
second filter and ship both filters to the laboratory for processing.
6.1.2 E. coli Samples
PWSs should analyze up to 100-mL of sample for LT2 monitoring. EPA recommends that the PWS
collect and ship more than 100-mL of sample to ensure sufficient volume for sample analysis is available
in the event of spillage at the laboratory. If spillage or leakage occurs during shipment, there is an
opportunity for sample contamination to occur and the sample should not be analyzed (see Section 8.3.1).
Additional details on sample collection procedures are provided in Section 6.4.3. The capacity of sample
containers should be 120-mL (6 oz.) or 250-mL (8 oz.) to allow for sufficient sample volume and at least
a 1-inch head space to facilitate mixing of the sample by shaking prior to analysis.
Draft 36 June 2003
-------
Section 6: Collecting and Shipping Source Water Samples
6.2 Sample Collection Location
LT2 rule monitoring is intended to assess the mean Cryptosporidium level in the influent to drinking
water plants that treat surface water or ground water under the direct influence (GWUDI) of surface
water. Generally, monitoring is required for each plant that treats a surface water or GWUDI source.
However, where multiple plants receive all of their water from the same influent (e.g., multiple plants
draw water from the same pipe), the same set of monitoring results may be applied to each plant. E. coli
samples should be collected at the same location as Cryptosporidium samples.
PWSs are required to collect source water samples for the LT2 rule from the plant intake prior to any
treatment [40 CFR part 141.704 (a)]. Guidance on sampling at plants where this may not be feasible, or
where other factors, such as the use of multiple sources, need to be addressed, is provided below, in
Sections 6.2.1 through 6.2.5.
6.2.1 Plants That Do Not Have a Sampling Tap Located Prior to Any Treatment
Plants in this situation should pursue one of the following options:
• Manually collect source water samples as close to the intake as is feasible, at a similar depth and
distance from shore.
• Establish a sampling location prior to treatment
*
6.2.2 Plants That Use Different Water Sources at the Same Time
This includes multiple surface water sources and blended surface water and ground water sources. Plants
in this situation should pursue one of the following options:
* If there is a sampling tap where the sources are combined prior to treatment, the sample should be
collected from the tap.
• Samples can be manually collected at each source near the intake on the same day and composited
. into one sample. The volume of sample from each source should be weighted according to the
proportion of that source used by the plant. For example, if a plant has two sources and 75% of the
drinking water is from Source A and 25% is from Source B, then for a 10-L sample, 7.5 L would be
collected from Source A and combined with 2.5 L collected from Source B. Compositing of samples
should reflect plant operation at the time the sample is collected and may change during the
monitoring period.
• Separate samples can be manually collected at each source near the intake on the same day and
analyzed independently. The results would then be used to calculate a weighted average of the
analysis results. The weighted average would be calculated by multiplying the analysis result for each
source by the fraction of the source contribution to total plant flow at the time the samples were
collected, and then summing these values. For example, if a plant has two sources and 75% of the
drinking water is from Source A and 25% is from Source B, then one sample would be collected from
each source and analyzed independently. If the concentration of oocysts for the sample from Source
A was 5 oocysts/L and the concentration of the sample from Source B was 2 oocysts/L, the final
result for the plant for this sampling event would be 4 oocysts/L ([5 oocysts/L x 0.75] + [1 oocyst/L x
0.25]).
6.2.3 Plants That Use Presedimentation
For these plants, source water samples must be collected after the presedimentation basin but before any
other treatment [40 CFR 141.704 (b)]. Use of presedimentation basins during monitoring should be
consistent with routine operational practice and should be documented. For systems taken samples after
Draft 37 June 2003
-------
Section 6: Collecting and Shipping Source Water Samples
presedimentation basin, no "Microbial Toolbox" credits will be allowed for presedimentation, if the plant
is classified into a bin that requires additional treatment [40 CFR 141.726 (a)].
6.2.4 Plants That Use Raw Water Off-Stream Storage
For these plants, source water samples must be collected after the off-stream storage reservoir [CFR
141.704 (c)]. Use of off-stream storage during monitoring should be consistent with routine operational
practice and should be documented.
6.2.5 Plants That Use Bank Filtration
The correct sampling location for PWSs with plants using bank filtration differs depending on whether
the bank filtered water is treated fay subsequent filtration for compliance with the Surface Water
Treatment Rule (SWTR) [40 CFR 141.704 (c)].
• PWSs using bank filtered water that is treated by subsequent filtration for compliance with the SWTR
must collect source water samples from the well (i.e., after bank filtration) but before any other
treatment. Use of bank filtration during monitoring should be consistent with routine operational
practice and should be documented. Systems collecting samples after a bank filtration process may
not receive microbial toolbox credit for the bank filtration [40 CFR 141.726 (c)].
• PWSs using bank filtered water without additional filtration must take source water samples in the
surface water source (e.g., the river). Use of bank filtration during monitoring should be consistent
with routine operational practice and should be documented.
Before monitoring begins, all plants must establish a source water monitoring schedule, as discussed in
Section 6.3.
6.3 Source Water Monitoring Schedule
PWSs are required to collect samples at least monthly and in accordance with a schedule established by
the PWS prior to initiation of monitoring. PWSs may collect samples more frequently (e.g., twice-per-
month, weekly), provided the same frequency is maintained throughout the monitoring period [40 CFR
part 141.701 (e)].
Water treatment plants that use surface water or ground water under the direct influence (GWUDI), but
are operated only seasonally (e.g., during times of high-water demand) should monitor at least monthly
during the period when the plant is in operation.-
Systems regulated under the LT2 rule are required to submit source water monitoring schedule to EPA
within 3 months of rule promulgation [40 CFR part 141.703 (a)]. The schedule is entered using the
scheduler function within the LT2 Data Collection System. Details on the use of the scheduler are
provided in the Users' Manual for the Long Term 2 Enhanced Surface Water Treatment Rule (LT2 Rule)
Data Collection System. Systems are required to collect samples within 2 days before or after the dates
indicated in their sampling schedules [40 CFR part 141.703 (b)].
The scheduler function will be available for PWSs to establish their LT2 monitoring schedule for a 3-
month period, beginning on the date of final rule publication. The use of a predetermined monthly or
semimonthly sampling schedule at each PWS during LT2 is designed to capture storm events and other
factors that affect water quality on a periodic basis. Because a PWS can potentially bias the results of the
monitoring by avoiding sample collection during periods of low water quality, the submission of pre-
scheduled sampling dates will be used to assess compliance.
Draft
38
June 2003
-------
Section 6: Collecting and Shipping Source Water Samples
6.4 Sample Scheduling Compliance Issues
Permissible exceptions to the sampling schedule are noted as follows:
« If extreme conditions or situations exist that may pose danger to the sampler, or which are unforeseen
or cannot be avoided and which cause the system to be unable to sample in the required time frame,
the system should sample as close to the scheduled date as feasible and submit an explanation for the
alternative sampling date to EPA concurrent with shipment of the sample to the laboratory.
EPA will evaluate the explanation and update the schedule in the LT2 Data Collection System, if
acceptable, to permit the analytical result to be submitted through the system (results with sample
collection dates that do not comply with the schedule entered by the PWS before monitoring began
will be rejected from the system).
• .Systems that are unable to report a valid Cryptosporidium analytical result for a scheduled sampling
• date due to failure to comply with the analytical method quality control requirements (e.g., sample is
lost or contaminated; laboratory exceeds analytical method holding time) must collect a replacement
sample within 14 days of being notified by the laboratory that a result cannot be reported for that
date. Systems must submit an explanation for the replacement sample with the analytical results.
Systems should collect an E. coli sample at the same time as the Cryptosporidium replacement
' sample.
Alternative sample collection dates should be timed so as not to coincide with another scheduled
Cryptosporidium sample collection date. Documentation of alternate sample collection, including the
reason, should be provided with the grandfathered data package.
6.4 Sample Collection Guidance
\
Large plants must begin collecting source water samples 6 months after rule promulgation and small
plants must begin 30 months after rule promulgation. Because the LT2 monitoring program is designed to
assess source water Cryptosporidium and E. coli concentrations, not the concentrations of these
organisms at points after any treatment, samples must be collected prior to any treatment and where the
water is no longer subject to surface runoff during LT2 monitoring (40 CFR part 141.704).
During each of the scheduled sampling events, several actions must be performed in addition to collecting
the sample. These actions, and an indication of which plant types each applies to, are summarized in
Table 6-2.
Draft 39 June 2003
-------
Section 6: Collecting and Shipping Source Water Samples
Table 6-2. Sample Collection Activities Required for Each Plant Type
Action
Document sample collection
information
Collect Cryptosporidium
sample
Collect E. coli sample
Measure turbidity
Monitor sample temperature
during sample transport
Large filtered
plants
/
/
/
/
/'
Large unfiltered
plants
/
/
/*
All small plants
/
/
/"
Small plants that
exceed the E.
coli trigger level
/
/
/
* Those utilities with on-site Cryptosporidium analytical capabilities will not need to transport samples unless the
laboratory is not located in close proximity to the sample collection location
" Those small plants with on-site E. coli analytical capabilities will not need to transport samples unless the laboratory
is not located in close proximity to the sample collection location
Guidance and procedures for each of these sample collection activities is provided in Sections 6.4.1 -
6.4.5, below.
6.4.1 Sample Collection Documentation
The information in Table 6-3 should be recorded during sample collection to link the monitoring result to
the plant, and to provide information required for development of the microbial index.
Table 6-3. Minimum Data Elements to Record During Sample Collection
Sampling Information
PWS name
Public Water System Identification
(PWSID) number*
Facility name
Facility ID"
Sample collection point name
Sample collection point ID*
Sample collection date"
Source water type"
Requested analysis
Sample collection time (start time
for field-filtered samples)
Meter readings (for field-filtered
samples only)
Required
/
/
/
/
/
Recommended
/
/
/
/
/
/
Draft
40
June 2003
-------
Section 6: Collecting and Shipping Source Water Samples
Sample collection stop time (for
field-filtered samples only)
Turbidity"
V
/
The combination of these elements constitute the unique sample identifier for LT2 monitoring samples
b This information should be recorded with the £ coli sample collection information, as it will be entered into the LT2
data collection system with the E coli sample results, for use in reassessing the microbial index. It does not need to
be reported with the Cryptosporidium sample collection information
For samples that are shipped off-site, this information should be documented on an LT2 sample collection
form (Appendix C), or similar form provided by your contract laboratory. For samples analyzed on-site
by your utility's laboratory, this information can be documented in a sampling log book or other standard
form used by your utility; the LT2 sample collection form can also be used.
The source water type for the sample will be used to reassess the relationship between Cryptosporidium
and E. coli concentrations (the microbial index discussed in Section ! .5). Sample collection personnel
must select from four source water types on the LT2 sample collection form:
• i Flowing stream (defined under the LT2 rule as "a course of running water flowing in a definite
channel")
• Reservoir/lake (defined under the LT2 rule as "a natural or man made basin or hollow on the Earth's
surface in which water collects or is stored that may or may not have a current or single direction of
flow")
• Ground water under the direct influence (GWUDI) of flowing stream surface water
• G WUDI of reservoir/lake surface water
The source water type should be selected based on the type of source water that accounts for the majority
of the surface water used as source water at the time of sample collection. For example, if the plant uses a
mix of approximately 55% reservoir/lake water and 45% flowing/stream water, the "reservoir/lake"
option should be circled on the LT2 sample collection form.
The majority of source water for plants that use GWUDI is ground water. However, as noted above, the
selection of source water type under the LT2 rule is based on the majority of surface water used as source
water. As a result, the selection of source water type is based on the type of surface water that accounts
for the majority of the influence of the ground water source.
The turbidity of the source water also needs to be measured. Cryptosporidium sample collection
procedures are discussed in Section 6.4.2; E. coli sample collection procedures and turbidity measurement
procedures are discussed in Section 6.4.3 and 6.4.4, respectively.
6.4.2 Cryptosporidium Sample Collection
Several options are available to the PWS in collecting untreated surface water samples for
Cryptosporidium analysis, including the following:
• Collection of bulk water samples for shipment to the laboratory for filtration and analysis. A detailed
protocol for collecting, packing, and shipping bulk samples is provided as Appendix D.
• ,On-site filtration of water samples using the Pall Gelman Envirochek™ or Envirochek™ HV capsule
.filter. A detailed protocol for filtering samples on-site from pressurized or unpressurized sources is
provided as Appendix E.
• On-site filtration of water samples using the IDEXX™ Filta-Max foam filter. A detailed protocol for
filtering samples on-site from pressurized or unpressurized sources is provided as Appendix F.
Draft
41
June 2003
-------
Section 6: Collecting and Shipping Source Water Samples
Regardless of the procedure used to collect Cryptosporidium samples, the sample must be eluted from the
filter within 96 hours of sample collection, per EPA Method 1622/1623 (Section 8.2). If this holding time
is violated, the laboratory will reject the sample, and your PWS will be required to recollect and reship
the sample.
LTZ rule requirement:
Each sample must meet the QC criteria for the methods [40 CFK
part 141705 (a) (3)1 Per EPA Me thod 1622/1623, samples must be
processed or examined within each of the holding times specified by
the method (Section 8.2).
6.4.2.1 Matrix Spike Samples
Method 1622/1623 requires matrix spike (MS) samples to be analyzed at a frequency of 1 MS sample for
every 20 monitoring samples from each plant. This frequency translates to the following, for each plant
category:
• For large PWSs that perform monthly monitoring for 2 years (resulting in 24 monitoring samples), 2
MS samples must be collected and analyzed
• For large PWSs that perform semi-monthly or more frequent monitoring for 2 years (resulting in 48
or more samples), a minimum of 3 MS samples will be collected and analyzed
• For small PWSs that are triggered into Cryptosporidium monitoring and collect semi-monthly
samples for 1 year (resulting in 24 samples), 2 MS samples must be collected and analyzed
The MS sample and the associated unspiked sample must be analyzed by the same procedure and the MS
sample must be the same volume as the associated monitoring sample. If the volume of the MS sample is
greater than 10 L, the system is permitted to filter all but 10 L of the MS sample in the field, and ship the
filtered sample and the remaining 10 L of source water to the laboratory to have the laboratory spike the
remaining 10 L of water and filter it through the filter used to collect the balance of the sample in the
field.
Utilities collecting and shipping bulk water samples for filtration and analysis at the laboratory should
split their sample stream and collect the monitoring sample volume and MS sample volume
simultaneously.
The sample stream should be split using flow controllers on both sides of the split to regulate the
pressure difference between the side being subjected to filtration (resulting in higher pressure) and the
side flowing into a bulk sample container. A mixing chamber (filter housing without filter) can be
added immediately upstream from the Y to aid in equalizing the distribution of sample particulates to
either side.
• If splitting the sample stream is not practical, the utility should collect the MS sample immediately
before or after the monitoring sample.
MS sample results would not be used to adjust Cryptosporidium recoveries at any individual source
water; but MS results would be used collectively to assess overall recovery and variability for EPA
Method 1622/1623 in source water. No resampling would be necessary for MS samples that do not meet
Method 1622/1623 recovery guidelines.
Draft
42
June 2003
-------
Section 6: Collecting and Shipping Source Water Samples
LT2 rule requirements:
(1) The MS and field sample must be collected from the same
sampling location by splitting the sample stream or collecting the
samples sequentially. (2) The volume of the MS sample analyzed
must be within 10% of the volume of the field sample analyzed. (3)
The MS and field sample must be analyzed by the same procedure
[40 CFR part 141705 (a) (2) (i)].
6.4.2.2 Purchasing Filters
If one of the field filtration options is used, you may want to consider purchasing filters in bulk from the
manufacturer (or the manufacturer's local distributor), as it may be cheaper than purchasing the filters
from your Cryptosporidium contract laboratory as part of the sampling kit. This approach also provides
your PWS with a ready supply of extra filters on-site, if a filter clogs during a sampling event. Plants
wishing to explore this option should call one of the contacts in Table 6-4.
Table 6-4. Contacts for Filters Approved for Use in EPA Method 1622/1623 ""
Pall Life Sciences
(Envirochek™ and Envirochek™ HV capsule filters)
IDEXX
(Filta-Max™ foam filters)
www.pall.com/gelman
600 South Wagner Road
Ann Arbor, Ml 48103
Sales:
Phone: (800) 521-1520 ext.2
Fax:(734)913-6495
Technical Support:
Phone: (800) 521-1520 ext.3
Fax:(734)913-6495
www.idexx.com
Sales:
Phone: (800) 321-0207 ext.1
Fax: (207) 856-0630
Technical Support:
Phone: (800) 321-0207 ext.2
Fax:(207)856-0630 .
E-mail: water@idexx.com
The-PWS also can purchase and assemble the entire sampling kit and maintain this kit on site, rather than
shipping it back and forth between the Cryptosporidium laboratory and the plant. If the filters you use
have associated shelf lives and storage conditions, ensure that the filters are stored according to the
manufacturers', directions and are not used past the specified shelf life.
The components and part numbers for the sampling kit are specified in the individual protocols for each
filter. If the sampling kit is maintained on-site by the utility, the utility should use disposable materials
wherever possible to mitigate the risk of cross-contamination between samples or sampling events, and
must disinfect the non-disposable sampling equipment between uses (if the laboratory provides the
sampling kit, this disinfection step is performed at the laboratory.)
Sampling kit cleaning should consist of the following:
• Cleaning equipment by scrubbing with warm detergent solution and exposing to hypochlorite
solution (minimum of a 5% solution of bleach and water) for at least 30 minutes at room temperature
Rinsing the equipment with reagent water and placing the equipment in an area free of potential
Cryptosporidium contamination until dry
6.4.2.3 Filter Clogs and Highly Turbid Water Samples .
PWSs with highly turbid source waters are likely to generate larger packed pellet volumes after
centrifugation and to clog filters than PWSs with low-turbidity waters. As noted in Section 6.1, at least 2
mL of packed pellet volume must be analyzed (for samples in which 10 L is filtered), or as much volume
Draft,
43
June 2003
-------
Section 6': Collecting and Shipping Source Water Samples
as two filters can accommodate before clogging. (If more than 10 L is filtered, then less of the packed
pellet volume needs to be analyzed.) - •
PWSs with highly turbid water may be able to collect the required minimum packed pellet volume by
avoiding filtration altogether, and shipping a bulk water sample to the laboratory for centrifuging. The
laboratory can centrifuge 250-mL or greater aliquot volumes sequentially, until a packed pellet volume of
2 mL is generated.
6.4.3 E. co// Sample Collection
For most large systems, E. coli analyses will be conducted on-site, so samples will not be shipped in most
cases, unlike Cryptosporidium samples. However, many small systems will collect E. coli samples and
ship them off-site for analysis. Regardless of whether the samples are analyzed by the utility's own
laboratory or by a commercial laboratory, laboratories analyzing E. coli samples for the LT2 rule must
use an E. coli method approved for use under the rule and must be certified under the drinking water
certification program for the general coliforrn analysis technique corresponding to the method the
laboratory plans to use for LT2 rule monitoring [40 CFR part 141.705 (b) and 141.706 (b)]. Approved E.
coli methods and their corresponding drinking water certification program coliforrn techniques are
discussed in the Microbial Laboratory Guidance Manual for the Long Term 2 Enhanced Surface Water
Treatment Rule (LT2 Rule). Summary information on these methods is also provided in Section 4 of this
document. :
Collect E. coli samples in sterile, non-toxic, plastic, or glass containers with a leak-proof lid. The capacity
of sample containers should be 120-mL (6 oz.) or 250-mL (8 oz.) to allow for sufficient sample volume
and at least a 1-inch head space to facilitate mixing of the sample by shaking prior to analysis A detailed
protocol for collecting source water samples for E. coli analysis, as well as packing and shipping
guidance for utilities that transport samples off-site for analysis, is provided as Appendix G.
EPA strongly encourages laboratories to analyze samples as soon as possible after collection. E. coli
samples must be analyzed within 24 hours of sample collection [40 CFR part 141.705 (b)(l)]. Note: This
is a longer time period than currently permitted in Standard Methods and the Manual for the Certification
of Laboratories Analyzing Drinking Water, and is based on data demonstrating that surface water samples
could be held, chilled, for up to 24 hours and still yield valid results (Reference 9.5).
Samples should be maintained above freezing and below 10°C in a refrigerator or in a cooler with wet ice,
blue ice, or gel packs, etc. Additional guidance on monitoring sample temperature is available in Section
6.4.5 of this manual.
6.4.4 Measuring Turbidity
PWSs must measure the turbidity of the source at the time of Cryptosporidium and E. coli sample
collection during LT2 rule monitoring. Turbidity must be measured by a party approved by the State [40
CFR part 141.706 (c)] using methods for turbidity measurement approved at 40 CFR part 141.74 [40CFR
part 141.705 (c)]. These methods include:
• Method 2130B, published in Standard Methods for the Examination of Water and Wastewater (19th
or 20th Edition). The full text of the 19* Edition is provided as Appendix H.
• Great Lakes Instrument (GLI) Method 2. The full text of this method is provided as Appendix I.
• Revised EPA Method 180.1, approved in August 1993 in Methods for the Determination of Inorganic
Substances in Environmental Samples (EPA-600/R-93-100). The full text of this method is provided
as Appendix J.
Draft
44
June 2003
-------
Section 6: Collecting and Shipping Source Water Samples
Systems must use turbidimeters that conform to one of the approved methods for measuring turbidity,
such as Hach Turbidimeter 1720D with EPA Method 180.1, GLI Turbidimeter Accu 4 with GLI Method
2, or equivalents (Note: These examples do not constitute an endorsement of specific instrumentation.
Approved methods provide specifications that turbidimeters must meet, and conformance of instruments
with these particular specifications must be determined prior to analysis.)- For regulatory reporting
purposes, either an on-line or a benchtop turbidimeter may be used, and systems must comply with all
quality control requirements specified in methods and regulations. If a system chooses to utilize on-line
units for monitoring, the system must validate the continuous measurements for accuracy on a regular
basis using a protocol approved by the State [40 CFR part 141.74 (c) (1)].
I
6.4.4.1 Measuring Sample Turbidity During LT2 Monitoring
When measuring turbidity, cuvettes must be clear, colorless glass or plastic. The tube must be kept clean,
both inside and out, to provide accurate readings. If a sample tube is scratched, it must be discarded.
* Measuring Sample Turbidity Using SM 2130B. Measure turbidity immediately after sample
collection to prevent temperature changes, particle flocculation, and sedimentation from changing
sample characteristics. Shake sample well before pouring into cuvette. Gently agitate to remove air
bubbles from the inside of the sample before pouring the sample into cell. Wait until all the air
bubbles disappear and remove all moisture from the outside of the sample cell before placing it into
the instrument. If fogging occurs, warm the sample by warm water bath for a short time, then re-
. agitate the sample before placing it in the turbidimeter. Read turbidity directly from instrument
1 display. Note: Measurements should be within the calibration range.
• Measuring Sample Turbidity Using GLI Method 2 or Revised EPA Method 180.1. Different
procedures should be followed, depending on the turbidity of the sample:
• For turbidities estimated to be less than 40 MTU. Shake the sample thoroughly to disperse the
solids. After waiting for the air bubbles to disappear, pour the sample into the turbidimeter tube
and read directly from the instrument scale.
• For turbidities estimated to be greater than 40 NTU. Dilute the sample with turbidity-free water
and compute the turbidity with the dilution factor included.
i • .
6.4.4.2 General Quality Control for Turbidity Measurements
Utilities performing environmental sample measurements must be approved by the State (or EPA Region,
fonstates that do not have primacy) under the drinking water laboratory certification program [40 CFR
part 141.706 (c)]. Each utility laboratory is required to operate a formal quality control (QC) program and
to maintain performance records that define the quality of the data generated. Two types of calibration are
required for turbidity measurements:
• A primary suspension standard. The primary suspension standard should be used to calibrate the
turbidimeter initially and at least every four months in order to prevent instrument drift. The
calibration should be documented. The standards should be replaced when they exceed the expiration
date.
Acceptable primary suspensions include Formazin (a recipe for preparation can be found at EPA
Method 180.1 and Standard Method 2130B), AMCO-AEPA-1 (available from Advanced Polymer
Systems), and Hach StablCal Stabilized Formazin Standards (available from Hach Company). Please
note that Formazin standards are relatively unstable, particularly at low concentrations. Therefore,
dilutions used for calibration need to be prepared on the day they will be used. Stock solutions may
be stable for a month (at 400 NTU) to 1 year (at 4000 NTU). Consult an approved method for more
information.
Draft • • 45 June 2003
-------
Section 6: Collecting and Shipping Source Water Samples
• A secondary suspension standard. The secondary suspension standard is used daily to check the
calibration of the instrument. The calibration should be documented, and should not vary by more
than 10% from the initial calibration values (if they do vary by more than 10%, the system should be
corrected so that performance is acceptable). The standards should be replaced when they exceed the
expiration date.
Acceptable secondary standards include all primary standards, or other materials that are suggested
by instrument manufacturers - such as sealed sample cells filled with a labeled suspension or metal
oxide particulates in a polymer gel, or a turbid glass cube. The purpose of the secondary standard is to
provide a quick check of calibration, The secondary standards should have a fixed turbidity that does
not vary from use to use.
6.4.5 Monitoring Sample Temperature
Source water samples are dynamic environments and, depending on sample constituents and '
environmental conditions, Cryptosporidium oocysts present in a sample can degrade and E. coli present in
a sample can grow or die off, biasing analytical results. Cryptosporidium and E, coli samples for LT2 rule
monitoring that are not analyzed the same day they are collected must be maintained below 10°C to
reduce biological activity. This is specified in Section 8.0 of the June 2003 versions of EPA Method
1622/1623 for Cryptosporidium samples and at 40 CFR part 705 (b) (1) and Chapter V, Section 6.3, of
the Laboratory Certification Manual (Reference 5.2) for£, coli samples.
Samples for all analyses should remain above freezing at all times. This is a requirement in Section 8.0 of
the June 2003 versions of EPA Method 1622/1623. Although not a significant concern for 10-L water
samples, this is a real concern for Cryptosporidium filters and 120- or 250-mL E. coli samples that are
shipped off-site with coolant materials, such as wet ice, blue ice, or gel packs. E. coli holding time studies
performed in support of the LT2 rule (Reference 9.5) demonstrated that E. coli samples can freeze under
these conditions if samples are not packed properly.
The sample collection protocols procedures in Appendices D, E, F, and G provide sample packing
procedures for E. coli and Cryptosporidium samples. Utility personnel should follow these procedures to
ensure that samples remain at acceptable temperatures during shipment.
Because Cryptosporidium samples collected for the LT2 rule must meet the QC criteria in the methods
[40 CFR part 705 (a) (3)], and because these QC criteria include receipt of samples at <10°C and not
frozen, laboratories must reject LT2 Cryptosporidium samples received at >10°C or frozen (this is
discussed further in Section 3.3.12 in this manual). In these cases, the PWS must re-collect and re-ship the
sample.
LT2 rule requirement:
Each sample must meet the QC criteria for the methods [4O CFR
part 141.705 (a) (3)]. Per EPA Method 1622/1623, samples not '
processed on the day of collection must be received at the
laboratory at < WC and not frozen (Section 8.1)
The sample collection protocols discussed in Section 6.4.2 for Cryptosporidium samples and Section
6.4.3 for E. coli samples provide guidance on packing samples to maintain appropriate temperatures.
Utility personnel should follow these procedures to ensure that samples remain at acceptable temperatures
during shipment.
Draft
46
, June 2003
-------
Section 6: Collecting and Shipping Source Water Samples
Several options are available to measure sample temperature upon receipt at the laboratory and, in some
cases, during shipment:
• Temperature sample. One option, for Cryptosporidium filtered samples (not for 10-L bulk samples)
and E. coli 120- and 250-mL samples, is for the PWS to fill a small, inexpensive sample bottle with
water and pack this "temperature sample" next to the field sample. The temperature of this extra
sample volume is measured upon receipt to estimate the temperature of the field sample. Temperature
sample bottles are not appropriate for use with bulk samples because of the potential effect that the
difference in sample volume may have in temperature equilibration in the sample cooler. Example
product: Cole Farmer cat. no. U-06252-20.
• Thermometer vial. A similar option is to use a thermometer that is securely housed in a liquid-filled
vial. Unlike temperature samples, the laboratory does not need to perform an additional step to
monitor the temperature of the vial upon receipt, but instead just reads the thermometer. Example
product: Eagle-Picher Sentry Temperature Vial 3TR-40CS-F or 3TR-40CS.
• {Button. Another option for measuring the sample temperature during shipment and upon receipt is a
Thermocron® iButton. An iButton is a small, waterproof device that contains a computer chip to
record temperature at different time intervals. The information is then downloaded from the iButton
onto a computer. The iButton should be placed in a temperature sample in the cooler, rather than
placed directly in the cooler, where it may be affected by close contact with the coolant. Information
on Thermocron® iButtons is available from http://www. ibutton.com/. Distributors include
http://www.pointsix.com/. http://www.rdsdistributing.com. and http://www.scigiene.com/.
• Stick-on temperature strips. Another option is for the laboratory to apply a stick-on temperature
strip to the outside of the sample container upon receipt at the laboratory. This option does not
measure temperature as precisely as the other options, but still mitigates the risk of sample
contamination while providing an indication of sample temperature to verify that the sample
temperature is acceptable. Example product: Cole Farmer cat. no. U-90316-00.
All temperature measurement devices should be calibrated routinely to ensure accurate measurements.
See the U.S. EPA Manual for the Certification of Laboratories Analyzing Drinking Water (Reference 9.3)
for more information.
Draft
47
June 2003
-------
SECTION 7: REVIEWING CRYPTOSPORIDIUM DATA
When Cryptosporidium samples are processed and analyzed by the laboratory, data on sample
measurements, sample processing times, and slide examination results are recorded at the laboratory and
reported to the PWS through the LT2 Data Collection System and via hardcopy forms. This section
provides an overview of the data recording and reporting processes and discusses the significance of the
examination results reported by the laboratory. This section also provides guidance to those PWSs
interested in reviewing laboratory data.
7.1 Cryptosporidium Data Recording at the Laboratory
The Cryptosporidium laboratory records LT2 rule monitoring data using the following standardized
forms:
7.1.1 LT2 Sample Collection Form
This form (an example of which is provided as Appendix C) is initiated at the plant upon sample
collection and is completed at the laboratory. The following information is recorded on this form by the
Cryptosporidium laboratory:
• Date and time of sample receipt
• Laboratory personnel receiving the sample
• Sample temperature upon receipt
• Sample condition upon receipt
Although none of this information is entered into the LT2 data collection system, it provides
documentation for the utility, the laboratory, and EPA or State officials on sample receipt information
relevant to LT2 rule requirements regarding sample temperatures and sample holding times.
7.1.2 Method 1622/1623 Bench Sheet
The laboratory uses the bench sheet to record all information associated with sample processing, up to,
but not including, sample examination. Information on filtration (if performed in the laboratory), elution,
concentration, immunomagnetic separation, and sample staining are documented on this form. These data
include:
• Sample ID
• Dates and times for all steps associated with method-required holding times
All primary measurements used to calculate sample volume analyzed, if less than 100% of the volume
filtered was analyzed. This information includes the following:
• The volume of the sample after the concentrate (packed pellet) has been resuspended
• The volume of this resuspended concentrate that was actually analyzed
(These two values are used to calculate the percent of the sample volume analyzed, if less than 100%
of the volume filtered was analyzed.)
Draft
48
June 2003
-------
Section 7: Reviewing Cryptosporidium Data
• , Filter clog and packed pellet information, which may need to be provided to demonstrate compliance
with LT2 rule sample analysis requirements if less than 10 L was analyzed
• Cryptosporidium spiking information for OPR and MS samples
• Analyst names or initials for each step
• Reagent and filter lot information
7.1.3 Method 1622/1623 Cryptosporidium Slide Examination Form
The laboratory uses the slide examination form to document detailed information on slide examination.
This information includes the following:
• Sample ID
• • Date and time the examination was completed
• Positive and negative staining control results
• Detailed information on the characteristics of each object on the slide that the analyst determined was
a Cryptosporidium oocyst, including the following:
• Size of the oocyst
• Shape of the oocyst
,• Whether the DAPI stain applied to the sample revealed the presence of nuclei, and, if so, how
many were observed by the analyst
• Whether the analyst observed internal structures during DIG examination
7.2 Submitting Cryptosporidium Data through the LT2 Data
Collection System
During the LT2 rule, laboratories will report Cryptosporidium data to their PWS clients electronically
through EPA's LT2 Data Collection System. The LT2 Data Collection System is a web-based application
that allows laboratory users to enter or upload data, then electronically "release" the data to the PWS for
review, approval, and submission to EPA and the State. Although ownership of the data resides with the
PWS throughout this process, the LT2 Data Collection System increases the ease and efficiency of the
data entry and transfer process from one party to another by transferring the ability to access the data
from the laboratory to the PWS to EPA and the State, and ensuring that data cannot be viewed or changed
by unauthorized parties. A summary of the data entry, review, and transfer process through the LT2 Data
Collection System is provided in Table 7-1, below.
Draft 49 June 2003
-------
Section 7: Reviewing Cryptosporidium Data
Table 7-1. LT2 Data Collection System Data Entry, Review, and Transfer Process
Laboratory actions
Laboratory posts analytical results to the LT2 Data Collection System
LT2 Data Collection System reduces data and checks data for completeness and compliance with LT2
rule requirements
Laboratory Principal Analyst confirms that data meet quality control requirements
• Laboratory "releases" results electronically to the PWS for review
Laboratory user cannot edit data after it is released to the PWS
!
LU
PWS actions
PWS cannot edit data - only review data and either return to laboratory to resolve errors or submit to
EPA
• PWS reviews electronic data through LT2 Data Collection System
PWS "releases" data back to the laboratory if questions
If no questions, PWS submits data to EPA as "approved" or "contested" (indicating that samples have
been correctly analyzed, but that the PWS contends are not valid for use in LT2 binning)
EPA and State actions
• EPA and State users cannot edit data - only review data
• EPA and State review data through LT2 Data Collection System
The data reporting process is discussed in more detail below, in Sections 7.2.1 through 7.2.3, and
discussed in detail in the Users' Manual for the Long Term 2 Enhanced Surface Water Treatment Rule
(LT2 Rule) Data Collection System. The LT2 data system users' guide also provides detailed information
on the PWS user registration process. Information on the LT2 Data Collection System, as well as a
downloadable users' manual, is available at http://www.epa.gov/safewater/lt2/index.html.
7.2.1 Data Entry/Upload
The analyst or another laboratory staff member enters a subset of the data recorded at the bench (Section
7.1) into the LT2 Data Collection System, either by entering the data using web forms or by uploading
data in XML format. This information includes the following: ••
• PWS ID
• Facility ID
• Sample collection point
• Sample collection date
• Sample type (field or MS)
• Sample volume filtered (L), to nearest % L
• Was 100% of filtered volume examined?
• Number of oocysts counted
• For samples in which less than 10 L is filtered or less than 100% of the sample volume is examined,
the laboratory also must enter or upload the number of filters used and the packed pellet volume.
• For samples in which less than 100% of sample volume is examined, the laboratory also must report
the volume of resuspended concentrate and volume of this resuspension processed through
immunomagnetic separation.
Draft
50
June 2003
-------
Section 7: Reviewing Cryptosporidium Data
• . For matrix spike samples, the laboratory also must report the sample volume spiked and estimated
number of oocysts. These data are not required for field samples,
The laboratory must verify that all holding times and other QC requirements were met.
After the information has been entered or uploaded into the system, the system will reduce the data to
yield final sample results, in oocysts/L', verify that LT2 rule Cryptosporidium sample volume analysis
requirements were met for samples in which less than 10 L were analyzed (see Section 6.1), and calculate
MS recoveries. '•
The laboratory's Primary Analyst under the Lab QA Program is responsible for verifying the quality and
accuracy of all sample results in the laboratory, and is required to review and approve the results before
they are submitted to the PWS for review. If inaccuracies or other problems are identified, the primary
analyst discusses the sample information with the analyst or data entry staff and resolves the issues before
the. data are approved for PWS review.
If no inaccuracies or other issues are identified, the Primary Analyst approves the reported data for
"release" to the PWS for review (EPA does not receive the data at this point). When the data are
approved, the rights to the data are transferred electronically by the system to the PWS, and the data can
no longer be changed by the laboratory.
7.2.2 PWS Data Review
After the laboratory has released Cryptosporidium data electronically to the PWS using the LT2 Data
Collection System, the PWS will review the results. The PWS user cannot edit the data, but if the PWS
has an issue with the sample result, such as if the PWS believes that the sample collection point ID or
collection date is incorrect, the PWS can release the results back to the laboratory for issue resolution. In
addition to noting the reason in the LT2 Data Collection System for the return of the data to the
laboratory, you also should contact the laboratory verbally to discuss the issue.
If the PWS determines that the data are accurate, the PWS releases the results to EPA (and the State, if
applicable) as "approved" results. If the PWS determines that the data are accurate, but believes that the
data are not valid for LT2 binning purposes, the PWS can release the results to EPA and the State as
"contested." Contested samples are those that have been correctly analyzed, but that you contend are not
valid for use in LT2 binning, and have submitted to EPA for evaluation.
7.2.3 EPA/State Review
After the PWS has released the results as approved or contested, they are available to EPA and State users
to review through the LT2 Data Collection System. EPA and State users cannot edit the data.
7.3 What Do the Sample Examination Results Mean?
As noted in Section 3.1, the laboratory applies two stains to a sample slide, and then examines the sample
using three different techniques to determine whether objects that cannot be ruled out as Cryptosporidium
oocysts are on a sample slide. A description of these stains and techniques-and how each is used to
evaluate objects examined by the analyst, is provided below.
7.3.1 Immunofluorescent Assay (IFA)
One of the two stains added to the sample before examination is a fluorescent antibody stain that reacts
with Cryptosporidium. The antibodies in this stain, which exhibit an intense apple-green fluorescence
Draft ' 51 June 2003
-------
Section 7: Reviewing Cryptosporidium Data
when the slide is examined using ultraviolet light, wiil attach to Cryptosporidium oocysts that may be
present in the sample. During IFA, the analyst scans the entire well at relatively low magnification (200X)
for apple-green fluorescing objects the size and shape of oocysts. If such an object is located, the analyst
proceeds to the next step in the examination process. The analyst cannot conclude at this stage that an
apple-green fluorescing organism the size and shape of a Cryptosporidium oocyst is indeed an oocyst
because the object may be another organism that has cross-reacted with the antibody stain. Additional
examination procedures are used to determine whether this is the case.
7.3.2 4',6-diamadino-2-phenylindole (DAPi) Examination
The second stain added to the sample before examination is DAPI, a dye that interacts with nucleic acids
and stains nuclei that may be present within the oocyst. The DAPI stain fluoresces when the slide is
examined using ultraviolet light. During the DAPI examination, the analyst observes the object at medium
magnification (400X) to determine whether it contains stained nuclei. Cryptosporidium oocysts contain
four nuclei.
Although looking for four nuclei during DAPI examination, if the object has less than four nuclei, the
analyst cannot rule out the possibility that the organism is a Cryptosporidium oocyst. For example, if less
than four stained nuclei are observed, the object may actually have four nuclei, but some may not be
visible in the plane of focus. Similarly, objects in which no stained nuclei are observed may be organisms
other than Cryptosporidium, may be dead Cryptosporidium oocysts, or may even be live oocysts that are
resistant to DAPI staining.
The DAPI examination is one of several tools for the analyst to use to determine whether an object is an
oocyst. The analyst cannot conclude whether the object is an oocyst based on this examination alone, nor
can the analyst conclude, based on negative results, that the organism is non-infectious. As a result, the
analyst must proceed to the next step in the examination process, even if less than four nuclei are
observed.
7.3.3 Differential Interference Contrast (DIG) Examination
The third evaluation performed by the analyst is an examination of the object at high magnification
(1000X). Using DIG, the analyst looks at the object's external or internal morphological characteristics
(this does not require the use of a stain). The analyst looks for characteristics atypical of Cryptosporidium
oocysts (e.g., spikes, stalks, appendages, pores, one or two large nuclei filling the cell, crystals, spores,
etc.). If atypical structures are not observed, and the object cannot be ruled out as an oocyst based on the
results of the IFA and the DAPI examination, the analyst reports this object as a Cryptosporidium oocyst.
Based on the DIC examination, the size of the object is determined and compared to the acceptable range
for the target organism. If the size and shape of the object is within the acceptable range, the analyst
records the size and shape and characterizes the Cryptosporidium oocyst in one of three ways: (1) an
oocyst with internal structures, i.e., those having recognizable structures consistent with Cryptosporidium,
(2) an oocyst with amorphous structures, or (3) an empty oocyst. Assignment of these characterizations is
dependent on analyst judgement and none of these characterizations is a direct indicator of whether
oocysts are viable and infectious.
7.5 Reviewing and Validating Raw Cryptosporidium Data (Optional)
If your PWS plans to review the raw data generated by the laboratory, you should request from the
laboratory the hardcopy data needed to verify the electronic results (see Section 5.1.5). However, this step
is not required. However, for those PWSs interested in taking this extra step, Sections 7.5.1 through 7,5.3
provide guidance on how to review and validate hardcopy data and verify accuracy.
Draft 52 June 2003
-------
Section 7: Reviewing Cryptosporidium Data
7.5.1 Data Completeness Check
Upon receipt of the hardcopy sample results for a monitoring sample, verify that the laboratory has
submitted the following materials:
• • Sample result summary sheet, which should include the following:
• Monitoring sample identification information
• Monitoring sample result, in oocysts/L
• Laboratory quality control batch associated with the sample
' • Result for the ongoing precision and recovery (OPR) sample analyzed for this QC batch
• Result for the method blank sample analyzed for this QC batch
• LT2 sample collection form initiated by your utility and completed with sample receipt information
by the laboratory
• Method 1622/1623 Bench Sheet with raw data associated with the monitoring sample (and MS
sample, if applicable)
• Method 1622/1623 Cryptosporidium Slide Examination Form with raw data for the monitoring
sample (and MS sample, if applicable)
• Laboratory comments. If the laboratory provided comments on the sample analyses or results that
require follow-up, contact the laboratory to discuss, if necessary. Comments may include any
applicable data qualifiers. The following is a list of potential data qualifiers:
• The recovery for the associated ongoing precision and recovery (OPR) sample did not met
method requirements
• Oocysts were detected in the method blank
• Positive and negative staining controls were not acceptable or not examined
• Method holding times were not met
* Sample arrived at the laboratory in unacceptable condition
Any of the above data qualifiers would result in the sample being considered invalid for LT2 use and the
laboratory should not report the results for the sample to EPA. The PWS may be required to resample.
If forms are missing, incomplete, or incorrect, contact the laboratory immediately to discuss and request
resubmission of the missing forms and/or spreadsheets.
j
7.5.2 Evaluation of Data Against Method Quality Control Requirements
To verify that the laboratory analyzed your monitoring sample within the analytical controls specified by
the method, check the following information:
• Sample condition upon receipt. Verify on the completed LT2 sample collection form that your
sample was received in acceptable condition (not leaking, etc.), and at a temperature between 0°C and
10°C, and not frozen.
• Method blank. Verify that the laboratory analyzed a method blank with the monitoring sample's QC
batch and confirm that the method blank did not contain any oocysts.
Draft 53 June 2003
-------
Section 7: Reviewing Cryptosporidium Data
• Ongoing precision and recovery sample. Verify that the laboratory analyzed an OPR sample with
the monitoring sample's QC batch and that the OPR sample recovery was between 11% and 100%, as
required by EPA Methods 1622 and 1623.
• Holding times. Using the sample collection date and time on the LT2 data collection form and the
dates and times of the method steps recorded by the laboratory on the Method 1622/1623 bench sheet
and report form for the monitoring sample, verify the following:
• The laboratory began elution no more than 96 hours from sample collection
• The laboratory performed the elution, concentration, purification, and slide preparation
• (application of the purified sample to the slide) within 1 working day (the date of the elution step
should be the same as the date of the slide preparation step)
• The laboratory stained the slide within 72 hours of application of the purified sample to the slide
(generally, the date next to the slide staining step should be no more than 3 days later than the
date next to the slide preparation step)
• The laboratory examined the slide within 7 days of staining (the examination date should be no
more than 7 days from the slide staining date)
• Positive and negative staining controls. Based on the information at the top of the Method
1622/1623 Cryptosporidium reporting form, verify that the laboratory performed positive and
negative staining controls, and that the results of these controls were acceptable.
7.5.3 Calculation Verification
The laboratory does not directly report the final concentration of oocysts/L in the sample to EPA. Instead,
they report a series of primary measurements that are used by the LT2 data system to automatically
calculate the final concentration. The volume filtered, the total volume of resuspended concentrate, and
the volume transferred to IMS are used to determine the volume analyzed. The laboratory also records the
total count of oocysts detected, which is divided by the volume analyzed to determine the final
concentration of oocysts/L. Although the final results are automatically calculated by the LT2 data
collection system using the primary measurements supplied by the laboratory, the plant still may wish to
verify them. Guidance on recalculating and verifying final results using primary measurements is
provided below.
7.5.3.1 Field Sample Calculations
To calculate the concentration of Cryptosporidium in your field sample, reported as oocysts/L, the
following information is needed:
• Number of oocysts detected in the sample (recorded as a primary measurement from the examination
results form)
• Volume analyzed
Using these two data elements, the final concentration is calculated as:
final concentration =
oocysts detected in the sample
volume analyzed (L)
If 100% of the sample volume filtered is examined, then the volume analyzed equals the volume filtered.
This applies whether one filter or more than one filter was used; if more than one filter was used, and all
Draft
54
June 2003
-------
Section 7: Reviewing Cryptosporidium Data
of the volume filtered through the multiple filters is processed through the remainder of the method, then
the volume examined is simply the sum of the volumes filtered through each of the filters used.
If < 100% of the volume filtered was processed through the remainder of the method, then additional
calculations are needed to determine the volume analyzed. This is discussed below.
Determining Volume Analyzed when Less than 100% of Sample Was Examined
When <100% of the sample filtered is processed through the remainder of the method and examined
(such as when the volume filtered yields > 2 mL of packed pellet volume after centrifugation), then the
volume analyzed must be determined using the following equations to determine the percentage of the
sample that was examined.
total volume ,of resuspended concentrate transferred to IMS (see Section 7.1.2)
percent examined =
total volume of resuspended concentrate produced
volume analyzed (L) = percent examined « sample volume filtered
Determining the Volume of Resuspended Concentrate to Use for Packed Pellets > 5 mL
Packed pellets with a volume >0.5 mL must be divided into subsamples. Use the formula below to
determine the total volume of resuspension required in the centrifuge tube before separating the
concentrate into two or more subsamples and transferring to IMS.
pellet volume (mL) after centrifugation
total volume of resuspended concentrate (ml) required = x 5 mL
0.5 mL
Example. A 10-L field sample was filtered and processed, producing a packed pellet volume of 2.7 mL.
The laboratory transferred 20 mL of the total resuspended concentrate to IMS and
examination. The laboratory detected 20 oocysts during examination. The following
calculations were performed to determine the volume analyzed and final concentration.
2.7 mL
total volume of resuspended concentrate (mL) required = x 5 mL = 27 mL
0.5 mL
20 mL
percent examined = = 0.74 (74%)
27 mL
volume analyzed (L) = 0.74 x 10 L = 7.4 L
20 oocysts
final concentration (oocysts/L) = = 2.7 oocysts/L
7.4 L
7.5.3.2 Matrix Spike Sample Calculations
For matrix spike (MS) samples, the laboratory records all of the same information that is recorded for
field samples, in addition to information specific to matrix spike samples. The sample volume spiked and
estimated number of oocysts spiked into the sample are used to calculate the concentration of spiked
Draft ' - 55 June 2003
-------
Section 7: Reviewing Cryptosporidium Data
organisms in the sample. To correct for background concentration, the number of organisms detected in
the unspiked field sample is subtracted from the number of oocysts detected in the MS sample.
To determine the percent recovery for a matrix spike (MS) sample, the following information is needed:
• The number of oocysts detected in the MS sample
• The true value of the oocysts spiked into the MS sample
• The number of oocysts detected in the unspiked field sample (to correct for background
concentration)
percent recovery =
oocysts counted in MS sample - oocysts counted in unspiked field sample
oocysts spiked into MS sample
100%
7.6 Data Archiving Requirements
LT2 rule monitoring data must keep monitoring results until 36 months after source water monitoring has
been completed. Although it is the PWS's responsibility to meet LT2 rule data storage requirements for
compliance monitoring samples, the PWS may designate this responsibility to the laboratory.
Although not required, laboratories also can archive slides and/or take photographs of slides to maintain
for clients. As noted in Section 5.1.2.5, this may be considered an extra service and result in extra costs,
as these steps may not be routinely performed by the laboratory. Slides should be stored in a humid .
chamber in the dark at 0°C to 10°C. An alternative mounting medium also may be used, which may
potentially preserve slides longer. Details are provided in the Microbial Laboratory Guidance Manual for
the Long Term 2 Enhanced Surface Water Treatment Rule (LT2 Rule),
Draft
56
June 2003
-------
SECTION 8: REVIEWING £. COL/DATA
When E. coli samples are processed and analyzed by the laboratory, data on sample measurements,
sample processing times, and slide examination results are recorded at the laboratory and reported to the
PWS through the LT2 Data Collection System. This section provides an overview of the data recording
and reporting processes and provides guidance on how to review the data you receive from the laboratory.
8.1 E. coli Laboratory Data Recording at the Laboratory
The laboratories performing E. coli analyses during the LT2 rule record the following general types of
information:
• , Sample identification information
• AH primary measurements used to calculate the final E. coli concentration for each sample
• The incubation start and read times for each method to verify that method requirements were met
• The name of the analyst performing the sample analysis
• Quality control (QC) analysis results (e.g., positive/negative controls, blanks, etc.)
8.1.1 Sample Identification Information
Sample identification information is used to track the sample through sample collection, analysis, and
data reporting. At a minimum, the laboratory records the sample ID, the target parameter (E. coli), and the
method being used (e.g., Membrane Filtration: SM 9222D/SM 9222G).
8.1.2 Primary Data
The laboratory records all primary measurements needed to calculate the final concentration of E. coli per
100 mL. Primary measurements for membrane filtration methods will include the volumes filtered and the
plate counts for each volume filtered. The multiple-well and multiple-tube formats will include the
volumes or dilutions of samples analyzed and the number of positive wells or tubes per each volume
analyzed.
8.1.3 Sample Processing and Quality Control Information
The laboratory records information on the bench sheet on how they processed and analyzed the sample,
including incubation start/end date and times and temperature, and the analyst performing each step of the
method. The lot numbers of reagents, media, and materials used to process the sample and the results of
QC analyses should be recorded in a media log book or QC checklist. In addition to being used to resolve
questions regarding validity of results, this information may be used by the laboratory to determine the
source of any problems the laboratory is having with method performance.
Draft 57 June 2003
-------
Section 8: Reviewing E. coli Data
8.1.4 Sample Results
The final E. coli concentration for field samples will be reported as CFU/100 mL or MPN/100 mL
depending on the method used for analysis. If no E. coli are detected in the sample, a low censored value
based on the volume of sample analyzed must be reported (e.g. <1CFU /100 mL or <1.8 MPN/100 mL).
E. coli concentration will never be reported as a zero.
8.2 Submission of E. coli Data through the LT2 Data Collection
System
During the LT2 rule, laboratories will report E. coli data electronically through EPA's LT2 Data
Collection System to the PWS staff responsible for approving and submitting monitoring results to EPA.
The LT2 Data Collection System is a web-based application that allows laboratory users to enter or
upload data, then electronically "release" the data to the appropriate PWS staff for review, approval, and
submission to EPA and the State. Although ownership of the data resides with the PWS throughout this
process, the LT2 Data Collection System increases the ease and efficiency of the data entry and transfer
process from one party to another by transferring the ability to access the data from the laboratory to the
PWS to EPA and the State, and ensuring that data cannot be viewed or changed by unauthorized parties.
A summary of the data entry, review, and transfer process through the LT2 Data Collection System for
both Cryptospohdium and E. coli samples is provided in Table 7-1, in Section 7.2, above.
The data reporting process is summarized below, in Sections 8.2.1 through 8.2.3, and discussed in detail
in the Users' Manual for the Long Term 2 Enhanced Surface Water Treatment Rule (LT2 Rule) Data
Collection System. The LT2 data system users' guide also provides detailed information on the laboratory
registration process. Information on the LT2 Data Collection System, as well as a downloadable users'
manual, is available at http://www.epa.gov/safewater/lt2/index.html.
8.2.1 Data Entry/Upload
The analyst or another laboratory staff member enters a subset of the data recorded at the bench (Section
8.1) into the LT2 Data Collection System either by entering the data using web forms or by uploading
data in XML format. This information includes the following;
• PWS ID
• Facility ID
• Sample collection point
• , Sample collection date
• Analytical method number
• Method type
• Source water type (provided by PWS on sample collection form)
* Turbidity result (provided by PWS on sample collection form)
• E. coli/WQ mL (see note below)
Note: The laboratory may then enter the final result for the sample that was calculated at the laboratory or
may enter the primary measurements recorded at the bench, and have the LT2 Data Collection System ,
automatically calculate the final sample concentration. Because this information is specific to method
type (membrane filtration, multiple tube fermentation, 51-well, and 97-well), the system provides
Draft 58 June 2003
-------
Section 8: Reviewing E. coli Data
different entry screens for each method type; The laboratory staff entering the data should verify that all
holding times and other QC specifications were met.
The laboratory's official contact is responsible for verifying the quality and accuracy of all sample results
in the laboratory, and is required to review and approve the results before they are submitted to the PWS
for review. If inaccuracies or other problems are identified, the official contact discusses the sample
information with the analyst or data entry staff and resolves the issues before the data are approved for
PWS review.
If no inaccuracies or other issues are identified, the laboratory's official contact approves the data for
"release" to the PWS for review (EPA does not receive the data at this point). When the data are
approved, the rights to the data are transferred electronically by the system to the PWS, and the data can
no longer be changed by the laboratory.
8.2.2 PWS Data Review
After the laboratory has released E. coli data electronically to the PWS using the LT2 Data Collection
System, the PWS will review the results. The PWS user cannot edit the data, but if the PWS has an issue
with the sample result, such as if the PWS believes that the sample collection point ID or collection date
is incorrect, the PWS can release the results back to the laboratory for issue resolution. In addition to
noting the reason in the LT2 Data Collection System for the return of the data to the laboratory, you also
should contact the laboratory verbally to discuss the issue.
If the PWS determines that the data are accurate, the PWS releases the results to EPA (and the State, if
applicable) as "approved" results. If the PWS determines that the data are accurate, but believes that the
data are not valid for other reasons, the PWS can release the results as "contested."
8.2.3 EPA/State Review
After the PWS has released the results as approved or contested, they are available to EPA and State users
to review through the LT2 Data Collection System. EPA and State users cannot edit the date.
8.3 Reviewing and Validating E. coli Data (Optional)
If the PWS staff responsible for submitting data to EPA plans to review the raw data generated by the
laboratory, the original, hardcopy records (whether generated by an in-house laboratory or a contract
laboratory) should be compared to the electronic results. However, this step is not required. Sections 8.3.1
through 8.3.3 provides guidance on how to review and validate the hardcopy data and verify accuracy.
8.3.1 Data Completeness Check
Upon receipt of hardcopy sample results for a monitoring sample, verify that the following information is
included:
• Sample result summary sheet, which should include the following:
• Monitoring sample identification information
• Monitoring sample result, in E. coli/lQQ mL
• Laboratory quality control checklist (or other verification from the laboratory that all QC
specifications were met)
Draft 59 . June 2003
-------
Section 8: Reviewing E. coli Data
• LT2 sample collection form initiated at the time of sample collection and completed with sample
receipt information by the laboratory
• £". coli Method Bench Sheet completed by the laboratory with primary sample processing and
analysis data associated with the monitoring sample
* Laboratory comments. If the laboratory provided comments on the sample analyses or results that
require follow-up, contact the laboratory to discuss, if necessary. Comments may include any
applicable data qualifiers. The following is a list of potential data qualifiers:
• ' Sample arrived at the laboratory in unacceptable condition (i.e., leaking)
• Sample holding time exceeded
• Sample holding temperature not within acceptable range
Unacceptable blank sample result
• Unacceptable positive or negative control result
• Media sterility checks were not acceptable
• Method incubation times or temperatures were not within acceptable range
• Membrane filtration: Too much sediment on the filter
• Membrane filtration: Confluent growth of non-target organism
• Membrane filtration: Colonies too numerous to count (TNTC)
• Membrane filtration: Pre-.or post- filtration series sterility check not acceptable (e.g.,
contamination with E. coli organism)
• Quanti-Tray® was damaged or leaked
• Sample was not distributed to all wells in Quanti-Tray®
• All rows of tubes were not inoculated
• Positive presumptive tubes were not transferred into the appropriate confirmatory medium
Any of the above data qualifiers would result in the sample being considered invalid for LT2 use and the
results for the sample should not be entered into the LT2 data collection system. If the laboratory enters
the results into the LT2 data collection system, the PWS should not submit the results to EPA.
If forms are missing, incomplete, or incorrect, contact the laboratory immediately to discuss and request
resubmission of the missing forms and/or spreadsheets.
8.3.2 Evaluation of Data Against Method Quality Control Requirements
To verify that the laboratory analyzed your monitoring sample within the analytical controls specified by
the method, check the following information:
• Sample condition upon receipt. If the sample was shipped to the laboratory, verify on the completed
LT2 sample collection form that your sample was received in acceptable condition (e:g., not leaking,
etc.), and at a temperature below 10°C, but not frozen. •
• QC samples associated field samples. The frequency of analysis of quality control samples
including method blanks, positive and negative controls, etc. varies according to method requirements
Draft
60
June 2003
-------
Section 8: Reviewing E. coH Data
and specifications in the Certification Manual. Verify that the required QC samples were run with the
•field sample. A summary of these QC specifications is provided in Section 4.2 of this document.
• Holding time. Using the sample collection date and time on the LT2 data collection form and the
date and time of the first method step recorded by the laboratory on the E. coli method bench sheet,
verify that the laboratory began sample analysis within 24 hours of sample collection.
• Incubation times and temperatures. Using the dates and times of the method steps recorded by the
laboratory on the E. coli method bench sheet, verify that the method-specified incubation times and
temperatures, specified in Table 8-1 were met.
Table 8-1. Incubation Times and Temperatures for Approved E. Coli Methods
Method
Standard Methods 9223B
Standard Methods 9221 B/F .
Standard Methods 9222B/9222G
Standard Methods 9222D/9222G
Standard Methods 921 3D
EPA 1603
. EPA 1604
Other Membrane Filter Method
Media
Colilert®
Colilert-18®
LTB
EC-MUG
mENDO -» NA-MUG
LES-ENDO -» NA-MUG
mFC -> NA-MUG
mTEC agar
Modified mTEC
Ml medium
m-ColiBlue24®® Broth
Incubation Time/Temperature
24 to 28 hours at 35°C ± 0.5'C
18 to 22 hours at 35°C ± 0.5'C
24 ±2 and 48 ± 3 hours at 35'C ± 0.5'C
24 ± 2 hours at 44.5°C ± 0.2°C
24 ±2 hours at 35°C ± 0.5°C -* 4 hours at
35°C ± 0.5°C
24 ±2 hours at 35"C ± 0.5°C -» 4 hours at
35°C ± 0.5°C
24 ±2 hours at 44.5°C ± 0.2*C -» 4 hours at
35°C ± 0.5°C
2 hours at 35°C 1 0.5°C -* 22 to 24 hours at
44.5"C±0.2*C
2 hours at 35'C ± 0.5°C -> 22 to 24 hours at
"44.5°C ± 0.2°C
24 hours at 35°C ± 0.5"C
24 hours at 35'C ± 0.5°C
8.3.3 Calculation Verification
Method-specific data to record for each of the individual method types as well as standardized
calculations for each method type are discussed in Sections 8.3.3.1 through 8.3.3.4.
8.3.3.1 Calculations for Determining the E. coli Concentration Using the Colilert® Quanti-Tray
2000® (97-well)
A. Select appropriate dilution to yield countable results. If multiple dilutions are used, the tray
exhibiting positive wells in the 40% to 80% range (39 to 78 total positive large and small wells)
should be used to determine MPN value.
B. Determine MPN. Using the number of large positive wells and small positive wells from the
appropriate dilution, identify the corresponding MPN/100 mL in the table provided by the vendor.
Large well values are located in the left column; small well values are located in the top row. For
example, if a 100-mL sample was analyzed, and there were 29 large positive wells and 5 small
positive wells, the corresponding MPN would be 49.6 MPN/100 mL.
Draft
61
June 2003
-------
Section 8: Reviewing E. col! Data
C. Adjust for dilution factor. Because the MPN/100 mL values in the table are based on 100-mL
samples, the MPN value should be adjusted if less than 100-mL of sample volume was analyzed. Use
the following calculation to adjust the MPN to account for the dilution:
MPN value
Analytical result =
mL of sample analyzed
Example:
Volume analyzed = 10 mL of sample (in 90 mL of dilution water)
Large wells positive = 39
Small wells positive = 5
The MPN value calculated based on the intersection of 10 and 2 in the table.
MPN = 81.3
Analytical result = 81.3 x ^— = 813 E. coli MPN/100 mL
8.3.3.2 Calculations for Determining the E. coli Concentration Using the Colilert® Quanti-Tray
51® (51-well)
A. Select appropriate dilution. If multiple dilutions are used, the tray exhibiting positive wells around
the 80% range (41 positive wells) should be used to determine MPN value.
B. Determine MPN. Using the number of positive wells from the appropriate dilution, identify the
corresponding MPN/100 mL in the table provided by the vendor. For example, if a 100-mL sample
was analyzed, and there were 26 positive wells, the corresponding MPN would be 36.4 MPN/100 mL
C. Adjust for dilution factor. Because the MPN/100 mL values in the table are based on 100-mL
samples, the MPN value should be adjusted if less than 100-mL of sample volume was analyzed. Use
the following calculation to adjust the MPN to account for the dilution:
100
MPN value x = £ coli MPN/100 mL
mL sample analyzed
Example:
Volume analyzed (mL) = 10 mL (in 90 mL of dilution water)
Number of positive wells = 41
MPN = 83.1
The analytical result is calculated as follows:
100
83.1 x = 831 £ coli MPN/100 mL
10
•Draft 62 June 2003
-------
Section 8: Reviewing E. coli Data
8.3.3.3 Calculations for determining the El coli concentration using membrane filter data (adapted
from Reference 9.4)
A. E. coli counts should be determined from the volume(s) filtered that yielded 20 to 80 E. coli colonies
(20-60 for mFC-NA-MUG), and not more than 200 total colonies per plate. (Guidance for samples
; that do not yield countable plates is provided in Sections E and F)
Note: The analytical result can be automatically calculated using the LT2 Data Collection System,
' See Section 8.2.1 for additional information.
B. If there are greater than 200 colonies per membrane, even for the lowest dilution, the result is
: recorded as "too numerous to count" (TNTC). These results cannot be reported for LT2 monitoring,
as they cannot be used for the required data analyses. During the next sampling event, analyze an
additional, lower dilution volume (the highest dilution volume may be omitted) unless conditions
were unusual (e.g., heavy rains, flooding, etc.) during the sampling event yielding TNTC for all
dilutions.
C.' If colonies are not sufficiently distinct for accurate counting, the result is recorded as "confluent
growth" (CNFG). To prevent CNFG from occurring, smaller sample aHquots should be filtered. For
example, if sample volumes of 100, 10, 1 and 0.1 mL are analyzed and even the 0.1-mL plates results
in CNFG, then potentially 0.01 mL should be analyzed during the next sampling event. The 100-mL
volume can be eliminated. Note: If growth is due to high levels of total coliforms but low E. coli then
. another method should be chosen for analyses that does not rely on total coliform determination
prior to or simultaneously with E. coli determination.
r
Note: Results that are TNTC or CNFG are not appropriate for L T2 microbial data analysis,
and cannot be entered into the L T2 Data Collection System.
D. Using the E. coli counts from the appropriate dilution, E. coli CFU/100 mL is calculated based on the
following equation:
100
Eco//CPU x = Eco//CFU/100 ml
mL sample filtered
'Example 1:
Filter 1 volume = 100 mL CPU = TNTC
Filter 2 volume = 10 mL CFU = 40
Filter 3 volume = 1.0 mL CFU = 9
Filter 4 volume = 0.1 mL CFU = 0
Using the guidance on countable colonies in Step A, the counts from the 10:mL plate will be used
to calculate the E. coli concentration for the sample:
100
40 Eco//CFU x = 400 Eco//CFU/100 mL
10 ml
E. If no E. coli colonies are present, the detection limit is calculated as '
< largest volume filtered per 100 mL.
Draft 63 June 2003
-------
Section 8: Reviewing E. coli Data
Example 2:
Filter 1 volume (mL) = 100 mL CPU = 0
Filter 2 volume (mL) = 10 mL CPU = 0
Filter 3 volume (mL) = 1.0 mL CPU = 0
Detection limit =
100mL
Largest volume filtered
= Eco//CFU/100mL
100mL
100 mL
£co///100mL
Example 3:
Filter 1 volume (mL) = 100 mL
Filter 2 volume (mL) = 10 mL
Filter 3 volume (mL) = 1.0 mL
Calculation of E. co///100 mL:
CPU = Lab accident, no data available
CFU = 0
CFU=0
100 mL
10 mL
. co/;CFU/100mL
F. If there are no filters with E, coli counts in the 20-80 colony range (20-60 for mFC-NA-MUG), sum
the E. coli counts on all filters, divide by the volume filtered and report as number per 100 mL.
Example 4:
Filter 1 volume (mL) = 50 mL
Filter 2 volume (mL) = 25 mL
Filter 3 volume (mL) = 10 mL
The analytical result is calculated as:
CPU =15
CPU = 6
CFU = 0
+ 6 + 0)
100
(50+25+10)
= 25E. co//CFU/100mL
Example 5:
Filter 1 volume (mL) = 50 mL CPU = 105
Filter 2 volume (mL) = 25 mL CPU = 92
Filter 3 volume (mL) = 10 mL CPU = 85
The analytical result is calculated as:
(105 + 92 + 85) *
100
(50+ 25-+10)
= 332Eco//CFU/100mL
ff^jgj^-.
Draft
64
June 2003
-------
Section 8: Reviewing E. coli Data
Example 6:
Filter 1 volume (tnL) = 100 mL
Filter 2 volume (mL) = 10 mL
Filter 3 volume (mL) = 1.0 mL
The analytical result is calculated as:
CPU = 82
CPU = 18
CFU = 0
(82 + 18 + 0)
100
(100 +10+1)
= 90Eco//CFU/100mL
Example 7:
Filter I volume (mL) = 50 mL
Filter 2 volume (mL) = 25 mL
Filter 3 volume (mL) = 10 mL
The analytical result is calculated as:
CPU = TNTC
CPU = TNTC
CPU == 83
83
100
10
= 830 E coli CFU/100 mL
8.3.3.4 Calculation off. coli Concentrations Using Multiple-Tube Methods (adapted from
Reference 9.6):
The guidance and examples for determining K coli concentrations using multiple-tube methods are based
on the revision of Standard Methods 9221C included in the 2001 Supplement to the 20th Edition of
Standard Methods, approved by the Standard Methods Committee in 1999.
Note: The analytical result can be automatically calculated using the LT2 Data Collection System.
See Section 8.2.1 for additional information.
A.; For each sample volume (e.g., 10, 1, 0.1, and 0.01 mL or additional sample volumes as necessary),
determine the number of positive tubes out of five.
B. A dilution refers to the volume of original sample that was inoculated into each series of tubes. Only
three of the dilution series will be used to estimate the MPN. The three selected dilutions are called
significant dilutions and are selected according to the following criteria. Examples of significant
dilution selections are provided in Table 8-2, below.
, • Choose the highest dilution (the most dilute, with the least amount of sample) giving positive
results in all five tubes inoculated and the two succeeding higher (more dilute) dilutions. (Table
8-2, Example A).
• If the lowest dilution (least dilute) tested has less than five tubes with positive results, select it and
the two next succeeding higher dilutions (Table 8-2, Examples B and C).
• When a positive result occurs in a dilution higher (more dilute) than the three significant dilutions
selected according to the rules above, change the selection to the lowest dilution (least dilute) that
has less than five positive results and the next two higher dilutions (more dilute) (Table 8-2,
Example D).
Draft
65
June 2003
-------
Section 8: Reviewing E. coli Data
• When the selection rules above have left unselected any higher dilutions (more dilute) with
positive results, add those higher-dilution positive results to the results for.the highest selected
dilution (Table 8-2, Example E).
• If there were not enough higher dilutions tested to select three dilutions, then select the next lower
dilution (Table 8-2, Example F).
C. MPN values need to be adjusted based on the significant dilutions series selected above. Because the
MPN/100 mL values in the table are based on the 10 mL, 1 mL, and 0.1 mL dilution series, per
method requirements, the MPN value must be adjusted if these are not the significant dilution series
selected. Use the following calculation to adjust the MPN when the 10 mL, 1 mL, and 0.1 mL
dilution series are not the significant dilution series selected:
Analytical result =
MPN value
# of mL in middle dilution
= £co//MPN/100 ml
Table 8-2. Examples of Different Combinations of Positive Tubes (Significant Dilution Results Are
in Bold and Underlined)
Example
A
B
C
D
E
F
Least dilute i Most dilute
(Lowest) p (Highest)
10
mL
5
4
o
5
5
5
1
mL
5
5
o
4
4
5
0.1
mL
1
1
1
4
4
5
0.01
mL
0
0
0
1
0
5
0.001
mL
0
0
0
0
1
1
Combination
of positives
5-1-0
4-5-1
0-0-1
4-4-1
4-4-1
5-5-2
MPN Index from
Standard Methods
33
48
1.8
40
40
540
E. CO///100 mL
(after adjustment)
330
48
1.8
400
400
54,000
Example A: The significant dilution series for the 5-1-0 combination of positives includes the 1
mL, 0.1 mL, and 0.01 mL dilutions. Since the 10 mL, 1 mL, and 0.1 mL dilutions
were not selected, an adjustment is necessary to account for the dilutions selected:
Analytical result =
33
0.1
<330E. co///100mL
Example B: Since the 10 mL, 1 mL, and 0.1 mL dilutions are the significant dilutions, no
adjustment is necessary and the result is 48 E. co///100 mL.
Example C: Since the 10 mL, 1 mL, and 0.1 mL dilutions are the significant dilutions, no
adjustment is necessary and the result is 1.8 E. co/z/100 mL.
Examples D and E:
The significant dilution series for the 4-4-1 combination of positives includes the
1 mL, 0.1'mL, and 0.01 mL dilutions. Since the 10 mL, 1 mL, and 0.1 mL
dilutions were not selected, an adjustment is necessary to account for the
dilutions selected:
Draft
66
June 2003
-------
Section 8: Reviewing E. coll Data
40
Analytical result = =400£. colil 100 mL
0.1
Example F: The significant dilution series for the 5-5-2 combination of positives includes the 0.1 mL,
0.01 mL and 0.001 mL dilutions. Since the 10 mL, 1 mL, and 0.1 mL dilutions were not
selected, an adjustment is necessary to account for the dilutions selected:
540
Analytical result = = 54,000 £ co///100mL
0.01
8.4 Data Archiving Requirements
Under the LT2 rule, monitoring data must keep until 36 months after source water monitoring has been
completed [40 CFR part 141.731 (a)]. Although it is the PWS's responsibility to meet LT2 rule data
storage requirements for compliance monitoring samples, the PWS may designate this responsibility to
the laboratory.
Draft 67 June 2003
-------
SECTION 9: REFERENCES
9.1 Connell, Kevin, et al. 2000. ICRSS - Building a Better Protozoa Data Set, J. AWWA. 91(10): 30
-43.
9.2 Pope, Misty, et al. 2003. "Using E. coli To Indicate Source Water Susceptibility to High
Concentrations of Cryptosporidium," in Information Collection Rule Data Analysis. AWWARF,
Denver, CO.
9.3 USEPA. 1997. Manual for the Certification of Laboratories Analyzing Drinking Water; Criteria
and Procedures; Quality Assurance: Fourth Edition. EPA 815-B-97-001.
9.4 APHA. 1998. Standard Methods for the Examination of Water and Wastewater; 20th Edition.
American Public Health Association, American Water Works Association, Washington, D.C.
9.5 Pope, M., et al. 2002. Assessment of the effects of holding time and temperature on E. coli
concentrations in surface water samples. Appl. Environ. Micro, (submitted).
9.6 2001 Supplement to the 20* Edition of Standard Methods 9221 C: Explanation of Bacterial
Density. This supplement is available for download at http://www.techstreet.com/cgi-
bin/detail?product i
Draft
68
June 2003
-------
SECTION 10: ACRONYMS
CPU
CNFG
DAPI
DIC
EPA
FA
FITC
GWUD1
ICR
IFA
IMS
IPR
IPT
L
LT2 rule
LT2ESWTR
MPN
MS
MS/MSD
um
NA-MUG
nm
NPDWR
NTU
OPR
OPT
PBMS
PT
PWS
QA
Colony-forming unit
Confluent growth
4, 6-diamidino-2-phenylindole
Differential interference contrast
U.S. Environmental Protection Agency
Immunofluorescense assay
Fluorescien isothiocyanate
Ground water under the direct influence of surface water
Information Collection Rule
Immunofluorescence assay
r
Immunomagnetic separation
Initial precision and recovery
Initial proficiency testing
Liter
Long Term 2 Enhanced Surface Water Treatment Rule
Long Term 2 Enhanced Surface Water Treatment Rule
Most probable number
Matrix spike
Matrix spike/matrix spike duplicate
Micrometer
Nutrient agar (NA) with 4-methylumbelliferyl-beta-D-glucuronide (MUG)
Nanometer
National Primary Drinking Water Regulations
Nephelometric turbidity unit
Ongoing precision and recovery
Ongoing proficiency testing
Performance-based measurement system
Proficiency testing
Public water system
Quality assurance
Draft
69
June 2003
-------
Section 10: Acronyms
QAP
QC
RSD
SDWA
TNTC
UV
Quality assurance plan
Quality control
Relative standard deviation
Safe Drinking Water Act
Too numerous to count
Ultraviolet
U.S. ERA Headquarters Library
Mail Code 3404T
1200 Pennsylvania Avenue, NW
Washington DC 20460
202-566-0556
Draft
70
June2003
-------
|