United States
Environmental Protection
»mAgency
Distribution System
Water Quality Monitoring:
Sensor Technology Evaluation
Methodology and Results
An Updated Guide for Sensor
Manufacturers and Water Utilities

-------

-------
February 2021
Distribution System Water Quality Monitoring:
Sensor Technology Evaluation Methodology and Results
An Updated Guide for
Sensor Manufacturers and Water Utilities
John S. Hall
Jeffrey G. Szabo
Homeland Security & Materials Management Division
Center for Environmental Solutions & Emergency Response
Sue Witt
Don Schupp
Aptim Federal Services, LLC
Cincinnati, Ohio
U.S. Environmental Protection Agency
Office of Research and Development
Homeland Security & Materials Management Division
Center for Environmental Solutions & Emergency Response
Cincinnati, Ohio
Recycled/Recyclable
Printed with vegetable-based ink on
paper that contains a minimum of
50% post-consumer fiber content
processed chlorine free

-------
Disclaimer
The U.S. Environmental Protection Agency (EPA) through its Office of Research and Development funded
and managed the research described herein under Contract Nos. EP-C-04-034, EP-C-09-041, EP-C-
14-012, and 68HERC19D0009 with Aptim Federal Services, LLC (APTIM). This document has been
reviewed by the Agency but does not necessarily reflect the Agency's views. No official endorsement
should be inferred.The U.S. Environmental Protection Agency does not endorse the purchase or sale of
any commercial products or services.

-------
Foreword
The U.S. Environmental Protection Agency (EPA) is charged by Congress with protecting the Nation's
land, air, and water resources. Under a mandate of national environmental laws, the Agency strives to
formulate and implement actions leading to a compatible balance between human activities and the
ability of natural systems to support and nurture life. To meet this mandate, EPA's research program is
providing data and technical support for solving environmental problems today and building a science
knowledge base necessary to manage our ecological resources wisely, understand how pollutants affect
our health, and prevent or reduce environmental risks in the future.
The Center for Environmental Solutions and Emergency Response (CESER) within the Office of
Research and Development (ORD) conducts applied, stakeholder-driven research and provides
responsive technical support to help solve the Nation's environmental challenges. The Center's research
focuses on innovative approaches to address environmental challenges associated with the built
environment. We develop technologies and decision-support tools to help safeguard public water
systems and groundwater, guide sustainable materials management, remediate sites from traditional
contamination sources and emerging environmental stressors, and address potential threats from
terrorism and natural disasters. CESER collaborates with both public and private sector partners to foster
technologies that improve the effectiveness and reduce the cost of compliance, while anticipating
emerging problems. We provide technical support to EPA regions and programs, states, tribal nations,
and federal partners, and serve as the interagency liaison for EPA in homeland security research and
technology. The Center is a leader in providing scientific solutions to protect human health and the
environment.
Gregory Sayles, Director
Center for Environmental Solutions and Emergency Response

-------
Table of Contents
Disclaimer	ii
Foreword	iii
List of Tables and Figures	viii
Acronyms and Abbreviations	ix
List of Revisions (Updated Guide, October 2019)	xi
Acknowledgements	xii
Notice of Trademarks and Product Names	xiii
Executive Summary	xiv
1.0 Introduction	1-1
1.1	Background	1-1
1.2	Definitions, Representations and Units	1-1
1.3	Concept of Operations for Contamination Warning Systems	1-2
1.4	Research Overview	1-4
1.5	Report Outline	1-4
2.0 Online Detection Equipment and Testing	2-1
2.1	Description of Testing Apparatus	2-1
2.1.1	Recirculating DSS Loop No. 6	2-1
2.1.2	Single Pass DSS	2-2
2.2	Test Contaminants and Water Matrices	2-3
2.2.1	Tested Contaminants	2-3
2.2.2	Test Water Matrix	2-4
2.3	Water Quality Measurement	2-4
2.3.1	Measured Water Quality Parameters	2-6
2.3.2	DSS Loop No. 6 Online Instrumentation	2-6
2.3.3	Single Pass DSS Online Instrumentation	2-7
2.3.4	Single Pass DSS Online Optical Instruments	2-7
2.4	Data Collection and Analysis	2-7
2.4.1	Data Collection	2-8
2.4.2	Data Analysis	2-8
2.5	Teaming with EPA's Water Security Initiative	2-8
2.6	EPA's Future Water Quality Sensor Research	2-8
3.0 Instrument Setup and Data Acquisition	3-1
3.1	Site-Specific Requirements	3-1
3.1.1	Environmentally Protected Housing	3-1
3.1.2	Access for Servicing the Instrumentation	3-2
3.1.3	Pressure-controlled Water Supply	3-2
3.1.4	Drainage Access	3-3
3.1.5	Power Supply and Electrical Protection	3-3
3.1.6	Transmission Media Access	3-4
3.1.7	Source Water Quality Adjustment	3-4
3.1.8	Instrument-Specific Accessories	3-4
3.2	Calibration Materials/Reagents and Onsite Accessories	3-5
3.3	Data Acquisition System	3-5
3.3.1	4 to 20 Milliamperes Current Output	3-5
3.3.2	Serial Protocols	3-6
3.3.3	Data Communication Protocols	3-6
3.3.4	SCADA Setup and Poll Rate	3-6
3.3.5	Data Marking	3-6
3.3.6	Data Transmission and Storage	3-6

-------
3.4 Best Practices for Instrument Setup and Data Acquisition	3-7
4.0 Testing Procedures and Safety Precautions	4-1
4.1	Blank/Control Injection	4-1
4.2	Contaminant Injection Procedures	4-1
4.2.1	Concentration of the Injected Contaminant	4-1
4.2.2	Duration of Injection	4-1
4.2.3	Water Main Flow Rate and Injection Rate	4-1
4.2.4	Neat Compounds Versus Commercial Off-the-Shelf Products	4-2
4.2.5	Wastewater and Ground Water Injections	4-2
4.3	Testing and Analytical Confirmation	4-2
4.3.1	Testing Confirmation	4-2
4.3.2	Analytical Confirmation	4-2
4.4	Flushing and Baseline Establishment	4-2
4.5	Health and Safety Precautions	4-3
4.6	Disposal of Contaminated Water From Test Runs	4-4
4.7	Best Practices for Testing and Safety Precautions	4-4
5.0 Data Analysis	5-1
5.1	Non-Algorithmic Sensor Response Evaluation	5-2
5.1.1	Single Pass DSS Data Analysis	5-2
5.1.2	Recirculating DSS Loop No. 6 Data Analysis	5-3
5.1.3	Edgewood Chemical and Biological Center Test Loop Data Analysis	5-6
5.2	Automated Algorithmic Evaluation of Sensor Response	5-7
5.3	Data Analysis Best Practices	5-9
6.0 Operation, Maintenance and Calibration of Online Instrumentation	6-1
6.1	Operation and Maintenance Labor Costs	6-1
6.2	Equipment-Specific Maintenance and Consumable Costs	6-1
6.3	Total Organic Carbon Instrumentation	6-2
6.3.1 Hach astroTOC™ UV Process Total Organic Carbon Analyzer	6-2
6.32 Sievers® 900 On-Line Total Organic Carbon Analyzer	6-2
6.3.3	Spectra: :lyzer™/Carbo::lyzer™	6-2
6.4	Chlorine Instrumentation	6-3
6.4.1 Hach CL-17 Free and Total Chlorine Analyzer	6-3
6.42 Wallace & Tiernan® Depolox® 3 plus	6-3
64.3 YSI 6920DW	6-3
6.4.4	Analytical Technology, Inc., Model A15/62 Free Chlorine Monitor	6-3
6.4.5	Rosemount Analytical Model FCL	6-4
6.5	Conductivity Instrumentation	6-4
6.6	pH/ Oxygen Reduction Potential Instrumentation	6-4
6.7	Turbidity	6-4
6.8	Dissolved Oxygen	6-5
6.9	Other Conventional Water Quality Parameter/Instrumentation	6-5
6.10	Online Optical Instrumentation	6-5
6.10.1	FlowCAM®	6-5
6.10.2	Hach FilterTrak™ 660 scLaser Nephelometer and Hach 2200 PCX Particle Counter	6-6
6.10.3	BioSentry®	6-6
6.10.4	Spectro::lyzer™/Carbo::lyzer™	6-6
6.10.5	ZAPS MP-1	6-6
6.10.6	ZAPS LiqulD	6-7
6.10.7	Turner TD1000C	6-7
6.10.8	HachUVAS	6-7
V

-------
6.10.9	RealTech UVT	6-7
6.10.10	ChemScan® UV-2150 Process Analyzer	6-7
6.10.11	Optiqua Refractive Index	6-8
6.11	Custom Sensors	6-8
6.11.1	PeCOD®P100 COD Analyzer	6-8
6.11.2	EZ-ATP® Analyzer	6-9
6.11.3	Sporian Inline Biosensor System (IBS)	6-9
6.11.4	QBiSci Online Water Sensor	6-9
6.12	Best Practices and Lessons Learned	6-10
7.0 Bibliography	7-1

-------
List of Tables
Table 2.1 Test Contaminant Matrix	2-4
Table 2.2 Measured Water Quality Parameters	2-5
Table 5.1 Parameter-Specific Significant Change Thresholds	5-1
Table 5.2 Percent Change in Sensor Parameter Response to Injected Chemical Contaminants in
Chlorinated Water - Single Pass DSS	5-2
Table 5.3 Normalized, Signal-to-Noise Corrected Sensor Parameter Response to Injected Chemical
Contaminants in Chlorinated Water - Single Pass DSS	5-4
Table 5.4 Percent Change in Sensor Parameter Response to Injected Biological Contaminants and
Growth Media in Chlorinated Water - Single Pass DSS	5-5
Table 5.5 Normalized, Signal-to-Noise Corrected Sensor Parameter Response to Injected Biological
Contaminants and Growth Media in Chlorinated Water - Single Pass DSS	5-5
Table 5.6 Percent Change in Sensor Parameter Response to Bacillus globigii Injection in Chlorinated
Water - Single Pass DSS	5-5
Table 5.7 Normalized, Signal-to-Noise Corrected Sensor Parameter Response to Bacillus globigii
Injection in Chlorinated Water - Single Pass DSS	5-5
Table 5.8 Quantitative Sensor Parameter Response Matrix to Contaminants in Chloraminated
Cincinnati Tap Water	5-6
Table 5.9 Percent Change in Sensor Parameter Response to Injected Warfare Agents in Chlorinated
Water - Edgewood Chemical and Biological CenterTest Loop	5-7
Table 5.10 Normalized, Signal-to-Noise Corrected Sensor Parameter Response to Injected Warfare
Agents in Chlorinated Water - Edgewood Chemical and Biological CenterTest Loop	5-7
List of Figures
Figure 1.1 Architecture ofthe EPA Contamination Warning System (EPA, 2007a)	1-3
Figure 2.1 Schematic of DSS Loop No. 6	2-1
Figure 2.2 DSS Loop No. 6 Sensor Manifold and Instrumentation Rack	2-2
Figure 2.3 Schematic of Single Pass DSS	2-3
Figure 2.4 Single Pass DSS - Longitudinal View	2-3
Figure 2.5 Single Pass DSS - Connecting Pipe Elbows	2-3
Figure 2.6 Single Pass DSS - Sampling Ports	2-3
Figure 2.7 DSS Loop No. 6 - Online Instrumentation	2-7
Figure 2.8 Single Pass DSS Instrument Panels	2-7
Figure 2.9 Various Single Pass DSS Optical Instruments	2-7
Figure 2.10 Technical Associates Radiation Monitoring Device	2-7
Figure 2.11 NexSens iSIC Data Acquisition System	2-8
Figure 2.12 First Pilot Utility - Water Security Initiative Instrument PanelType A	2-8
Figure 2.13 First Pilot Utility - Water Security Initiative Instrument PanelType B	2-9
Figure 3.1 Single Pass DSS Instrument Panel at 80-foot Sampling Location	3-2
Figure 3.2 Single Pass DSS Instrument Panel at 1,180-foot Sampling Location	3-2
Figure 3.3 Example Constant Head Mechanism for Hach 2200 PCX Particle Counter	3-3
Figure 3.4 Field Communications Enclosure	3-3
Figure 3.5 Hach astroTOC™ UV Process Total Organic Carbon Analyzer	3-4
Figure 3.6 Sievers® 900 On-Line Total Organic Carbon Analyzer	3-5
Figure 3.7 SCADA Data Flow Schematic	3-5
Figure 3.8 T&E Facility NexSens iSIC Datalogger	3-7
Figure 4.1 Injection Apparatus for the Single Pass DSS	4-1
Figure 4.2 Glyphosate Triplicate Injection Run Results	4-3
Figure 4.3 Glyphosate Injections at Varying Concentrations	4-3
Figure 5.1 CANARY Operation Schematic	5-8
v

-------
Acronyms and Abbreviations
jjm	Micrometer
jjS/cm	Microsiemens/centimeter
°C	Degrees Centigrade or Celsius
°F	Degrees Fahrenheit
AC	Absolute Change
ANL	Argonne National Laboratory
APTIM	Aptim Federal Services, LLC
ATP	Adenosine Triphosphate
AVWVA	American Water WorksAssociation
AwwaRF	American Water WorksAssociation
Research Foundation
BED	Binomial Event Discriminator
B. globigii	Bacillus globigii
BOD	Biological Oxygen Demand
CESER	Center for Environmental
Solutions and Emergency
Response
CFU	Colony Forming Units
CI"	Chloride ion
CL2	Chlorine
C02	Carbon dioxide
COD	Chemical Oxygen Demand
CWS	Contamination Warning System
DCE	Data Circuit-Terminating Equipment
Dl	De-ionized
DOC	Dissolved Oxygen
DOC	Dissolved Organic Carbon
DSS	Distribution System Simulator
DTE	Data Terminal Equipment
ECBC	Edgewood Chemical and Biological
Center
E. coli	Escherichia coli
EDS	Event Detection Software
EIA	Electronic Industries Alliance
EPA	U.S. Environmental Protection Agency
eq.	Equivalent
ETV	Environmental Technology Verification
ft/sec	Foot per second
FTI	Frontier Technology, Inc.
GAC	Granular activated carbon
GCWW Greater Cincinnati Water Works
GFP	Green Fluorescent Protein
gpm	Gallons per minute
H+	Hydrogen ion (a proton)
HASP	Health and Safety Plan
HMI	Human-Machine Interface
HOCL"	Hypochlorous acid
HPLC	High-Performance (or High-Pressure) Liquid
Chromatography
hr	Hour
HSPD Homeland Security Presidential
Directive
I/O	Input and Output
IBS	Inline Biosensor System
ICR Inorganic Carbon Remover
IDLH Immediately Dangerous to Life or
Health
IEC	International Electrotechnical
Commission
IP	Internet Protocol
ISE	Ion Selective Electrode
iSIC Intelligent Sensor Interface and
Control
KCN Potassium cyanide
KHP Potassium hydrogen phthalate
mA	Milliamperes
MALS Multi-Angle Light Scattering
MCHM	4-Methylcyclohexane-
methanol
MCL	Maximum contaminant level
MDE	Molecular Detection Element
mg/L	Milligrams perliter
min	Minutes
mL	Milliliter
mm	Millimeter
mNTU	Milli-Nephelometric Turbidity Unit
MSD Metropolitan Sewer District of
Greater Cincinnati
MS2	Male-specific 2
mV	Millivolts
MZI	Mach Zehnder Interferometer
N/A	Not available or not applicable

-------
NAREL
NEMA
NH2CI
nm
NTU
National Air and Radiation
Environmental Laboratory
National Electrical
Manufacturers Association
Chloramine
(monochloramine)
Nanometers
Nephelometric Turbidity
Unit
o2
Oxygen

O&M
Operation and Maintenance
UC
OCL"
Hypochlorite ion
UPS
ODBC
Open Database Connectivity
U.S.

OGWDW Office of Ground

Water and Drinking Water
USB
ORP
Oxidation Reduction Potential
UV
PC
Personal Computer
pg/mL
picograms per milliliter
UV-Vis
PH
Potential of hydrogen in standard
units
UV254
v/v
PLC
Programmable Logic Controller

ppb
parts per billion
VX
ppm
parts per million

QAPP
Quality Assurance Project Plan
WATEF
QBiSci
Quantitative Biosciences, Inc.
Rl
Refractive Index

ROC
Receiver Operating Characteristic
WDMP
RP-570
RTU Protocol based on IEC 57 Part
WSD

5-1 (present IEC 870) Version 0 or 1
WSi
RS-232
Recommended Standard 232
RS-485
Recommended Standard 485
XML
RTUs
Remote Terminal Units
ZAPS
q
wpeak
Peak sensor value between first in

contact of the contaminant with the
sensor until 15 minutes after the initial
contact

q
^baseline
Baseline (mean) sensor value for
one hour immediately preceding an
injection test

S/N
Signal-to-Noise

SCADA
Supervisory Control and Data
Acquisition

SDWA
Safe Drinking Water Act

SNL
Sandia National Laboratories

T&E
Test & Evaluation

TA	Technical Associates
TCP	Transmission Control Protocol
TEVA	Threat Ensemble Vulnerability
Assessment
TEVA-SPOT Threat Ensemble Vulnerability
Assessment - Sensor Placement
Optimization Tool
TOC	Total Organic Carbon
TTEP	Technology Testing and Evaluation
Program
University of Cincinnati
Uninterrupted Power Supply
United States
Universal Serial Bus
Ultraviolet
Ultraviolet-Visible
Ultraviolet 254 nanometer wavelength
Volume/Volume Percent
V-series Nerve Agent (S-[2-
(diisopropylamino)ethyl]-0-ethyl
methylphosphonothioate)
Water Awareness Technology Evaluation
Research and Security
Water Distribution Monitoring Panel
Water Security Division
Water Security Initiative
extensible Markup Language
Zero Angle Photon Spectrometer
IX

-------
List of Revisions (Updated Guide, October 2019)
Item
Revision
Title
An Updated Guide for Sensor Manufacturers and Water Utilities
Date
October 2019
Authors
Sue Witt and Don Schupp
Company
Aptim Federal Services, LLC
Contracts
Included Contract Nos. EP-C-14-012 and 68HERC19D0009
Table of Contents
Updated
Acronyms and Abbreviations
Updated
Acknowledgements
Included new authors and acknowledged previous authors
Notice of Trademarks and
Included 10 new sensors and identified them by asterisk
Product Names

Executive Summary
Stated that the Updated Guide includes the latest 10 years of
sensor testing
Section 1
Stated that the Updated Guide includes the latest 10 years of
sensor testing and did not update previous data or costs
Section 2
Updated Tables 2-1 and 2-2 and Sections 2.3.3 and 2.3.4 to
include the new contaminants and sensors
Section 3
No changes
Section 4
No changes
Section 5
Updated Tables 5.2 through 5.5
Section 6
Included 10 new sensors
Section 7
Included Original Sensor Handbook, Fluorescence Report, and
Online Biological Sensors Report as references

-------
Acknowledgements
The principal authors of this document, titled "Distribution System Water Quality Monitoring: Sensor
Technology Evaluation Methodology and Results-An Updated Guide forSensor Manufacturers and Water
Utilities," are Mr. John S. Hall, Dr. Jeffrey G. Szabo, P.E., Ms. Suzan M. Wtt, and Mr. Donald A. Schupp,
P.E.
The authors wish to acknowledge the contributions of the following individuals and organizations towards
the development and review of this document:
Technical reviews of the document were performed by:
Mr. Steve Allgeier, U.S. Environmental Protection Agency's Office of Ground Water and Drinking
Water
Mr. Stanley States, Pittsburg Water and Sewer Authority
Mr. Alan Roberson, Director of Security and Regulatory Affairs, American Water Works Association
U.S. Environmental Protection Agency quality assurance review was performed by:
Ms. Eletha Brady-Roberts, Quality Assurance Manager
Initial report prepared by:
Mr. Srinivas Panguluri, P.E., Aptim Federal Services, LLC
Mr. Greg Meiners, Aptim Federal Services, LLC
Illustrations and publication design were performed by:
Mr. James I. Scott, Aptim Federal Services, LLC
Cover Photo Credits are as follows:
Unknown Rural Town, West Virginia - photograph from archives of Aptim Federal Services, LLC
Water Tower, Florence, Kentucky - photograph by Ms. Jennifer Panguluri
Child drinking water from a tap - photograph of Mr. Ravi Panguluri taken by Ms. Jennifer Panguluri
X!

-------
Notice of Trademarks and Product Names
Trademark or Product Name
Manufacturer's Name, City, State
Web Site
1720D Turbidimeter
Hach Company, Loveland, Colorado
http://www.hach.com
Analytical Technology, Inc. Model A15/62
Free Chlorine Monitor
Analytical Technology, Inc., Collegeville, Pennsylvania
http://www.analyticaltechnology.com
BioSentry®
JMAR Technologies, Inc., San Diego, California
http://www.jmar.com
ChemScan® UV-2150 Process Analyzer*
ASA Analytics, Waukesha, Wsconsin
http://www.chemscan.com
EZ-ATP®*
AppliTek, (a Hach Company), Nazareth, Belgium
http://www.applitek.com
FlowCAM®
Fluid Imaging Technologies, Yarmouth, Maine
http://www.fluidimaging.com
GuardianBlue™ Early Warning System
Hach Company, Loveland, Colorado
http://www.hach.com
H20 Sentinel™
Frontier Technology Inc., Goleta, California
http://www.fti-net.com
Hach 2200 PCX Particle Counter
Hach Company, Loveland, Colorado
http://www.hach.com
Hach astroTOC™ UV Process Total Organic
Carbon Analyzer
Hach Company, Loveland, Colorado
http://www.hach.com
Hach CL17 Free Chlorine Analyzer
Hach Company, Loveland, Colorado
http://www.hach.com
Hach CL17 Total Chlorine Analyzer
Hach Company, Loveland, Colorado
http://www.hach.com
Hach Event Monitor™ Trigger System
Hach Company, Loveland, Colorado
http://www.hach.com
Hach FilterTrak™ 660 sc Laser Nephelom-
eter
Hach Company, Loveland, Colorado
http://www.hach.com
Hach/GLI Model C53 Conductivity Analyzer
Hach Company, Loveland, Colorado
http://www.hach.com
Hach/GLI Model P53 pH/ORP Analyzer
Hach Company, Loveland, Colorado
http://www.hach.com
Hach UVAS*
Hach Company, Loveland, Colorado
http://www.hach.com
Hach Water Distribution Monitoring Panel
(WDMP) sc
Hach Company, Loveland, Colorado
http://www.hach.com
Hydrolab® DS5
Hach Company, Loveland, Colorado
http://www.hach.com
Optiqua Refractive Index*
Optiqua Technologies, The Netherlands
http://optiqua.com
PeCOD® P100 COD Analyzer*
Mantech Inc., Guelph, Ontario, Canada
http://mantech-inc.com
QBiSci Online Water Sensor *
Quantitative Biosciences, Inc., San Diego, California
htto://www.abisci.com
Real Kill®
Realex Corporation, Spectrum Brands, St. Louis, Mis-
souri
http://www.spectrumbrands.com
RealTech UVT*
Real Tech Inc., Whitby, Ontario, Canada
http://realtechwater.com
Rosemount Analytical Model FCL
Emerson Process Management, Irvine, California
http://www.raihome.com
Roundup®
The Scotts Company, LLC or its affiliates, Marysville,
Ohio
http://www.scotts.com
Sievers® 900 On-Line Total Organic Carbon
Analyzer
GE Analytical Instruments, Boulder, Colorado
http://www.geinstruments.com
Sievers® RL
GE Analytical Instruments, Boulder, Colorado
http://www.geinstruments.com
Six-CENSE™
CENSAR Technologies, Sarasota, Florida
http://www.censar.com
Spectro::lyser™ or Carbo::lyser™
scan Messtechnik GmbH, Vienna, Austria
http://www.s-can.at
Sporian Inline Biosensor System (IBS)*
Sporian Microsystems, Lafayette, Colorado
http://www.sporian.com
SSS-33-5FT
Technical Associates, Canoga Park, California
http://www.tech-associates.com
TROLL® 9000
In-Situ® Inc., Ft. Collins, Colorado
http://www.in-situ.com
Turner TD1000C*
Turner Designs Hydrocarbon Instruments, Inc., Fresno,
California
http://www.turnerdesigns.com
Wallace & Tiernan® Depolox® 3 plus
Siemens Water Technologies, Kent, United Kingdom
http://www.wallace-tiernan.com
YSI 6600
YSI Inc.,Yellow Springs, Ohio
http://www.ysi.com
YSI 6920DW
YSI Inc.,Yellow Springs, Ohio
http://www.ysi.com
ZAPS LiqulD*
ZAPS Technologies Inc., Corvallis, Oregon
http://www.zapstechnologies.com
ZAPS MP-1
ZAPS Technologies Inc., Corvallis, Oregon
http://www.zapstechnologies.com
* Designates sensors new to the Updated Guide (October 2019)
xii

-------
Executive Summary
This report, titled "Distribution System Water Quality Monitoring: Sensor Technology Evaluation
Methodology and Results - An Updated Guide for Sensor Manufacturers and Water Utilities," provides an
overview of the U.S. Environmental Protection Agency's (EPA's) research results from investigating
water quality monitoring sensor technologies that might be used to serve as a real-time contamination
warning system (CWS) when a contaminant is introduced into a drinking water distribution system. EPA's
concept of CWS for protecting water distribution systems is discussed in Chapter 1.0. A principal
component of such a system is online water quality monitoring. The original report was prepared in
October 2009. This updated report includes sensors tested between 2009 and 2019.
Based on a review of available online waterquality monitoring sensortechnologies, an early determination
was made that it was not technically feasible to accurately identify and quantify the many different types
of contaminants that could potentially be introduced into the drinking water supply/distribution system.
Furthermore, because online sensortechnologies need to be economically suitable for mass deployment
within a distribution system, EPA focused its research on identifying sensortechnologies that could be
used to detect anomalous changes in water quality due to contamination event(s). Once a waterquality
anomaly is detected, the water utility operator is alerted, and further actions (e.g., sampling and analysis)
could be undertaken by the operator to identify and quantify the contaminant if necessary. This report
focuses on EPA's research on pilot-scale evaluations of available online waterquality monitoring sensor
instrumentation.
This report first describes the testing apparatus (the recirculating Distribution System Simulator (DSS)
- Loop No. 6, Single Pass DSS, and the online instrumentation) used for the pilot-scale evaluations at
the EPA Test and Evaluation (T&E) Facility in Cincinnati, Ohio (Chapter 2.0). The instrument setup and
data acquisition specifics are described in Chapter 3.0. The detailed testing procedures and safety
precautions are described in Chapter 4.0. The data analysis procedures are presented in Chapter 5.0.
Operation and maintenance specifics for selected instruments are provided in Chapter 6.0. In addition,
each chapter includes a best practices summary at the end with key points that are designed to deliver
the "lessons learned" through this research. A bibliography of selected references is included as Chapter
7.0.

-------

-------
1.0 Introduction
The safety of drinking water supplied to the consumers
by water treatment plant operators is dependent upon
many factors: quality of raw water (surface water and/
or ground water), application of appropriate treatment
technology/disinfection (as needed), and monitoring of
treated/finished water within the water distribution
system network. Appropriate treatment/disinfection
technologies for both surface and ground water sources
are identified by the U.S. Environmental Protection
Agency (EPA) in various regulations that were
promulgated pursuant to the Safe Drinking Water Act
(SDWA) of 1974 and its amendments. Although the
treated water leaving a treatment plant typically meets
EPA's water quality requirements, the water could un-
dergo transformation within the various distribution
system components (e.g., storage tanks and pipes),
which alters the quality, potentially making it unsuit-
able for human consumption. To address these issues,
EPA has developed specific regulations that mandate
periodic monitoring of water quality within distribu-
tion systems.
Research related to water quality monitoring within the
distribution system has increased significantly since the
events of September 11, 2001, when improving the
security of our nation's water infrastructure became a
major priority. Homeland Security Presidential
Directive 7 (HSPD-7), issued on December 17, 2003,
established a national policy for federal departments
and agencies to identify and prioritize United States
critical infrastructure and to protect the infrastructure
from terrorist attacks. Thereafter, HSPD-9, issued on
January 30, 2004, directed EPA to "develop robust,
comprehensive, and fully coordinated surveillance and
monitoring systems, that provide early detection and
awareness of disease, pest, or poisonous agents." EPA
plays a critical role in this effort as the lead federal
agency for water security. Subsequent to these
directives, in March 2004, EPA released the peer-
reviewed Water Security Research and Technical
Support Action Plan (Action Plan - EPA, 2004a), which
identified important water security related issues and
outlined research and technical support needs to address
these issues. In addition, the EPA Action Plan identified
a list of projects to be undertaken in response to the
identified needs. Furthermore, the Action Plan
identified several products proposed to be developed to
enhance the security of drinking water and wastewater
systems. This report is one of the products designed to
meet the Action Plan requirements specified under
Section 3.3.d.2 - Standard Operating Procedures and
Quality Assurance and Control Practices to Guide the
Evaluation of Monitor-
ing Technologies and Section 3.3.d.5 - Standard Op-
erating Procedures for Evaluating Monitoring Tech-
nologies.
ckground
The analytical methods and water quality sensors used
to address EPA regulations pursuant to SDWA were not
designed to address water security threats. Conse-
quently, data necessary to identify a serious threat to the
water supply caused by either an accidental release or
by an intentional act might not be captured during
routine periodic monitoring at drinking water treatment
plants and various distribution system locations. Over
the past five years, as part of the overall Water
Awareness Technology Evaluation Research and Secu-
rity (WATERS) program at the EPA Test & Evaluation
(T&E) Facility in Cincinnati, Ohio, EPA investigated
online water quality monitoring technologies that might
be used to achieve the goal of serving as early warning
indicators to detect contaminant introduction into the
drinking water supply. The WATERS program testing
efforts were sponsored by EPA's Homeland Security
Research Program. During this study, a variety of
commercially available online sensors/in-struments
were evaluated.
Based on a review of available online water quality
monitoring sensor technologies, an early determination
was made that it was not technically feasible to accu-
rately identify and quantify the many different types of
contaminants that could potentially be introduced into a
drinking water supply/distribution system. Further-
more, these online technologies needed to be economi-
cally suitable for mass deployment within a distribution
system. Therefore, EPA focused its research to identify
online sensor technologies that could be used to detect
anomalous changes in the baseline water quality with-
out specific regard to precision, accuracy or identifica-
tion of the contaminant. Once an anomaly is detected
and the water utility operator is alerted, further actions
(e.g., grab sampling and analysis) could be undertaken
by the operator to identify and quantify the contami-
nant whenever possible. This report focuses on EPA's
research on pilot-scale evaluations of available online
water quality monitoring sensor instrumentation.
1	esentations
For the purposes of this document, a "sensor" is defined
as an electro-mechanical device (e.g., membrane, elec-
trode, or microchip) that measures a physical or chemi-
cal characteristic of water and converts it into a "signal"
or measured value, which is typically processed further
by an instrument.
1-1

-------
An instrument is defined as an electro-mechanical device
(or a collection of electro-mechanical devices) that can
manipulate (e.g., amplify) a measured output from an as-
sociated "sensor" and transmit the measured value (e.g.,
analog or digital output value) to a data acquisition system.
Some instruments (e.g., optical instruments) perform
measurements without an associated sensing element (as
defined in this section) and contain additional devices
that transmit data. Therefore, in general, an instrument
is meant to refer collectively to a sensor that provides the
overall measurement functionality. Furthermore, a "sen-
sor" or "instrument" response is intended to define the
change in measured or recorded output value of the rel-
evant water quality parameter. The term "equipment" is
used to refer collectively to electro-mechanical devices
that might include one or more sensors, instruments, and
additional appurtenances such as plumbing, data collec-
tion and/or recording devices that are necessary to make
the overall manufactured device functional. Depending
upon the focus of the discussion, and to improve docu-
ment readability, the terms "sensor," "instrument," and
"equipment" have been used somewhat interchangeably
throughout the document.
The research did not address instrument-specific preci-
sion, accuracy, or ability to identify contaminants.
The term "sensor manufacturer(s)" is intended to in-
clude instrument manufacturer(s) and vendor(s) who
might simply resell or repackage a manufactured prod-
uct. A listing of tested sensor/instrument technologies
and their associated registered or unregistered trade-
marks is included under the Notice of Trademarks and
Product Names (page xiii) in this report. The tested
equipment referred to in the original Guide (EPA, 2009)
was procured over time and used for testing during the
period of 2003 to 2008. This Updated Guide includes
additional equipment that was tested between 2009 -
2019. Subsequent to the testing, there could have been
design changes and/or improvements to the equipment
by the manufacturers. These devices might perform
differently under the same tested conditions but could
bear the same registered or unregistered trademark.
Neither the authors nor EPA make any representations
on the usefulness or general performance of these de-
vices outside the context of the testing described in this
report. The use of these manufacturer-specific names
and model numbers throughout the document is to pro-
mote clarity so that the reader can identify the tested
equipment. Any rights associated with these registered
or unregistered trademarks are the sole property of the
trademark holders. It is recommended that water utilities
and other researchers apply their own judgment priorto
choosing any equipment for water quality monitoring.
English standard units that are commonly used by the
U.S. water utility personnel have been used throughout
this document. For example, volume is reported inU.S.
gallons and velocity in feet per second (ft/sec).
However, in keeping with industry usage, contaminant
concentrations are reported in metric units, in
milligrams per liter (mg/L). Unless otherwise stated for
computational purposes, the values from the
instruments are presented as reported in the output from
the individual instrument(s) without any conversions
provided.
Dincept of Operations for
Contamination Warning
Systems
Real-time water quality monitoring to control treatment
process operations has been successfully performed at
water treatment plants for many years. EPA's concept
for contamination warning systems (CWS) is designed
to extend this monitoring approach to multiple locations
within a water distribution system (Kessler et al., 1998;
ISLI, 1999; AwwaRF, 2002; Kirmeyer et al., 2002;
EPA, 2005 a-d; Roberson and Morley, 2005; Allgeier et
al., 2006; Dawsey et al., 2006). Consequently, baseline
water quality conditions can be monitored continuously
in real-time such that a sudden change in water quality
parameter(s) can trigger a contamination warning.
Monitoring baseline water quality parameters within
the distribution system will also provide multiple
benefits of improved water quality closer to the point-
of-use and additional security for detecting intentional
or unintentional contamination events within the
system. The capital, operational, and maintenance costs
for CWS will be difficult to sustain unless multiple
benefits are identified. For water utilities, it is important
to first maximize the security benefits by strategically
placing the selected online monitors in the network and
utilizing suitable techniques to evaluate the online
sensor responses. Therefore, in addition to evaluating
on-line water quality monitoring and sensor
technologies, EPA has collaborated with various
research entities to develop two key software tools that
provide these functionalities, described below.
EPA, in collaboration with research organizations
including the Sandia National Laboratories (SNL),
Argonne National Laboratory (ANL), University of
Cincinnati (UC) and the American Water Works As-
sociation (AWWA), has developed a software program
referred to as the Threat Ensemble Vulnerability
Assessment - Sensor Placement Optimization Tool (TE-
VA-SPOT). TEVA-SPOT can be used to determine the
optimum number and locations for monitoring stations
within a water distribution system. The software allows
the user to specify a wide range of performance
objectives including: 1) Population-based health
measures, 2) Time to detection, 3) Extent of
1-2

-------
contamination, 4) Volume of contaminated water
consumed, and 5) Number of contamination events
detected. TEVA-SPOT facilitates interactive design of a
water quality monitoring system by allowing the user to
specify constraints to ensure that the performance
objective is satisfied. For example, a TEVA-SPOT user
can integrate expert knowledge during the design
process by identifying either existing or unfeasible
sensor locations. Installation and maintenance costs for
sensor placement can also be factored into the analysis.
More information on the TEVA Research Program and
SPOT can be obtained online at:
https://www.epa.gov/emergency-response-
research/water-infrastructure-modeling-tools.
EPA, in collaboration with SNL, also developed the
CANARY algorithm to evaluate water quality sensor re-
sponses and identify changes in water quality that could
indicate a contamination event. The name CANARY is
not an acronym but suggests a parallel with the historic
"canary in the coal mine" event detection approach in
which the coal miners used canaries to detect poison gas
events. Similarly, the CANARY software evaluates real-
time water quality data obtained from various instru-
ments and uses mathematical and statistical techniques
to identify the onset of anomalous water quality events.
The CANARY software allows for the following: 1) the
use of a standard data format for input and output of
water quality and operations data, 2) the ability to se-
lect different detection algorithms (the program contains
three different mathematical approaches for analyzing
the data), 3) the ability to select various water utility and
location-specific configuration options, 4) an online op-
erations mode and an off-line evaluation/training mode,
and 5) the ability to generate data needed to establish
performance metrics (e.g., false alarm rates). Thisalgo-
rithmic approach enhances the detection sensitivity of
the field equipment and simultaneously reduces the false
positive alarm events. CANARY is freely available for
download through the EPA website. More information
on CANARY can be obtained online at:
https://www.epa.gov/emergency-response-
research/water-infrastructure-modeling-tools
Regardless of the approach used by the utility to evalu-
ate the data collected from online sensors, establishing
a protocol to verify and respond to alarms triggered by
the online water quality monitoring instruments is
important. Note that online water quality monitoring
represents only one component of a holistic CWS. Ad-
ditional data inputs from the utility and public health
agencies should be collected and evaluated to comple-
ment the benefits of online water quality monitoring
(See Figure 1.1).
EPA's Office of Ground Water and Drinking Water
(OGWDW), Water Security Division (WSD), has field-
deployed a pilot project called the Water Security Initia-
tive (WSi), that is based upon the concepts identifiedin
Figure 1.1. The WSi program is being implemented in
the following three phases:
•	Phase I: develop the conceptual design of a
system for timely detection and appropriate
response to drinking water contamination
incidents to mitigate public health and economic
impacts;
•	Phase II: test and demonstrate CWS
through pilots at drinking water utilities and
municipalities and make refinements to the
design based upon pilot results; and
1-3

-------
Routine Operation
Operational Strategy
Fi
Monitoring & Surveillance
Online water quality
Sampling & analysis
Utility
Enhanced security [-~ data
storage
Consumer complaints
Public Health
911 call data
Emergency Medical
Services data
Poison control data
Syndromic
surveillance data
Over-the-counter
medication sales
Public
health
storage
Event Detection
Possible Determination
Drinking water utility
event detection and
initial trigger
validation
Public health event
detection and trigger
validation
Consequence Management
Consequence Management Plan
Credible Determination
>
Response actions


Operational response


Public health response


Site Characterization


Laboratory confirmation


Risk communication

J
Is contamination
Yes
credible?

No
Confirmed Determination
Response actions
Operational response
Public health response
Expanded sampling
Laboratory confirmation
Risk communication
Remediation & Recovery

Planning


System
characterization


Remedy
selection


Remedial
action


Post-re mediation
activity


Risk
communication



Return to routine monitoring & surveillance
Figure 1.1 Architecture of the EPA Contamination Warning System (EPA, 2007a)
4

-------
~ Water Utilities and
Sensor Manufacturers
•	Online water quality monitoring alone will not
provide a holistic CWS.
•	Integration of data streams such as consumer
complaint surveillance, enhanced security
monitoring, public health surveillance, and
triggered sampling and analysis with the online
water quality monitoring is necessary for realizing
the full benefits from a CWS.
•	EPA has developed several guidance and guideline
documents, a modular response protocol toolbox,
and other software tools for utilities planning to
establish a comprehensive CWS. The relevant
software tools include TEVA-SPOT for locating
online sensors and CANARY event detection
software. The bibliography section includes a
listing of the related EPA documents.
•	Manufacturers should design flexibility into the
sensor equipment to output real-time data streams
in a variety of formats, which allows for analysis
by both external and/or internal event detection
algorithms.
•	In addition to helping achieve regulatory
compliance (e.g., monitoring residual disinfectant
levels), sustainable online CWS equipment
can provide other benefits that can lead to
improvements in: distribution system water quality;
treatment process control; distribution system
control; customer service; and overall security.
• Phase III: develop practical guidance and
outreach to promote voluntary national
adoption of effective and sustainable drinking
water CWS.
Based on information collected from the ongoing Phase
I and Phase II activities, WSD has developed a variety of
guidance and interim guidance documents on related
topics including: WaterSentinel system architecture
(EPA, 2005c), planning for CWS deployment (EPA
2007b), developing an operational strategy for CWS
(EPA, 2008a), developing consequence management
plans (EPA, 2008b), and the Cincinnati pilot post-imple-
mentation system status (EPA 2008c). In addition, WSD
had previously developed a modular response protocol
toolbox to assist water utilities for planning and respond-
ing to contamination threats (EPA, 2004 [c through j]).
More information on the EPA WSi can be obtained on-
line at: https://www.epa.gov/waterresilience
1.4 Research Overview
The vast majority of the research described in this re-
port was conducted at the EPA T&E Facility in Cincin-
nati, Ohio. Since the early 1990's, at this facility, EPA
has conducted research using simulated drinking water
distribution systems. A number of pilot-scale distribu-
tion system simulators (DSSs) are in use at the T&E Fa-
cility. EPA operates, maintains, and modifies the DSSs
as needed to accommodate evolving study designs. For
the research results reported in this document, EPA em-
ployed two types of DSSs at the T&E Facility to inves-
tigate water quality monitoring sensor technologies that
might be used to serve as a real-time early warning sys-
tem when a contaminant is introduced into the drink-
ing water supply. Only online sensors were evaluated,
because the response time is critical for achieving the
project objective of contamination warning.
To evaluate the selected sensors, a series of test runs was
conducted by injecting known quantities of potential
contaminants into the selected DSSs. After injection,
sensor data were collected continuously and electroni-
cally archived. After injection, grab samples were col-
lected periodically to confirm the sensor results. These
studies were focused on providing independent third-
party data to decision makers in the following areas:
1.	What water quality parameters will be most
useful in CWS?
2.	Can online water quality sensors be used
to reliably trigger alarms in response to
contamination events within a water distribution
system?
3.	What are the operational and maintenance costs
associated with online water quality monitoring
systems?
1.5 Report Outline
The following chapters of this report summarize the
findings related to this research. Chapter 2.0 presents a
summary of the various online detection sensors/
instrumentation evaluated and the evaluation-specific
research activities performed at the EPA T&E Facility
in Cincinnati, Ohio and other field locations. Chapter
3.0 describes general instrument setup and data acqui-
sition. Chapter 4.0 contains a description of the testing
procedures and safety precautions. Chapter 5.0 outlines
the data analysis procedures. Chapter 6.0 describes the
operation and maintenance (O&M) and calibration re-
quirements of the tested instrumentation. At the end of
each chapter (starting in Chapter 3.0), a summary of
applicable best practices is presented for the targeted
audience, which includes sensor manufacturers and wa-
ter utilities.
The original report was prepared in October 2009
(EPA, 2009). This updated report includes sensors
tested during the following ten years. The new sensor

-------
descriptions and results have been included in the
appropriate chapters of this Updated Guide. Previous
data and associated costs have not been updated in
this version of the Guide.

-------
1 ~I

-------

-------
2.0	Online Detection
Equipment and Testing
The focus of this research was to identify water qual-
ity parameters and online sensor technologies that
could be used to detect anomalous changes in water
quality due to contamination event(s) within a water
distribution system. The sections of this chapterbriefly
describe the following: testing apparatus, con-
taminants and injected concentrations, disinfectants,
water quality parameters and online instrumentation,
data collection and analysis, event detection, and field
applications.
2.1	Description of Testing
Apparatus
The first round of testing for online water quality sen-
sor instrumentation was conducted using recirculat-
ing DSS Loop No. 6 located at the T&E Facility in
Cincinnati, Ohio. DSS Loop No. 6 was essentially
operated as a closed system during the sensor testing
period. At the conclusion of the first round of tests,
some of the research stakeholders expressed concern
that the recirculation mode operation of DSS Loop No.
6 enhanced the detection ability of the sensors. In this
mode, the contaminant is recirculated within the
distribution system, thereby allowing the sensor to
detect the same slug of contaminant multiple times.
Subsequently, later rounds of testing involved the use
of the Single Pass DSS, also located at the T&E Facil-
ity in Cincinnati, Ohio.
Concurrent to the DSS Loop No. 6 and Single Pass DSS
testing, EPA conducted a series of bench-scale
minimum dosing tests. In these tests, the selected con-
taminants in a water matrix (at various concentrations)
were exposed to the online sensors to establish the
minimum dosage/concentration of the contaminant
where a "response" to various water quality parameters
was produced by the sensor instrumentation.
21.1 Recirculating DSS Loop No. 6
Recirculating DSS Loop No. 6 consists of a 15-year old,
6-inch-diameter unlined ductile iron pipe and is one of
six pipe loops within the DSS (Loop Nos. 1 through 6).
DSS Loop No. 6 is approximately 75 feet long and has
a total capacity of approximately 150 gallons. DSS
Loop No. 6 is equipped with a 3-horse-power pump
capable of circulating water through the loop at a rate of
up to 110 gallons per minute (gpm). The loop is
normally operated at a flow rate of 88 gpm, which
produces a velocity of 1 foot per second (ft/sec) in the
main pipe. The process flow schematic of tlieDSS Loop
No. 6 used for these tests (including modifications for
this research) is presented in Figure 2.1.
For the purposes of this testing, DSS Loop No. 6 was
operated in recirculation mode using municipal tap
water supplied by the Greater Cincinnati Water Works
(GCWW). In this mode, the feed tanks and the 100-gal-
Online
Instrument
Panel
Ji2~|
Chlorine
injection
pump
Total Organic Carbon ¦
Chlorine ¦
Con du di vity/T em p erature
pH/Oxidation Reduction Potential
Turbidity
Biofilm sampling coupon
— Water flow-
Heat
exchanger
Oxidation Reduction Potential
Dissolved Oxygen
Potable
water
30-gallon
feed
water
tank

or
2T
Feed
water
pump
Recirculation
pump
(0-0.25 gpm)
h8i-
Incoming makeup water
Loop recirculation/test water
Drain
Chemical addition
Sensor loop instrumentation
Biofilm
- sampling

100-gallon
recirculation/mix
tank
Contaminant
feed tank
Figure 2.1 Schematic of DSS Loop No. 6
2-1

-------
Ion recirculation tank are kept inline with the system.
Operation in this mode effectively increases the volume
of water in the system by 85 gallons, to a total of ap-
proximately 235 gallons. When operating in recircula-
tion mode, potable water is added to the system from
the 30-gallon feed-water tank at a rate of 0.16 gpm. At
this rate, the entire volume of the system is exchanged
in 24 hours. However, due to mixing in the recirculation
tank, the time required to completely exchange the con-
tents of the system via dilution is considerably longer.
Injected contaminants reached the sensors in approxi-
mately 75 seconds and quickly become homogeneously
mixed with the 250 gallons of water in the system. Dye
tests were performed to confirm the travel time and
mixing. The response profiles to injected contaminants
reflect this design. An initial response after the contami-
nant first reaches the sensors is recorded for those sen-
sors capable of detecting the contaminant. The response
persist as the contaminant becomes dispersed in the
DSS Loop No. 6, and in the sensor manifold, followed
by a period of recovery due to dilution or consumption
of the injected material via hydrolysis or reaction with
free chlorine present in the tap water or through pipe
wall reaction.
DSS Loop No. 6 is equipped with one 10-gallon chemi-
cal feed tank and a pump used to add treatment chemi-
cals to the system. The feed tank was used to add chlo-
rine when establishing baseline conditions prior to the
addition of contaminants. Chlorine additions continued
during test runs in order to keep the disinfectant levels
stable during injections. The DSS Loop No. 6 setup also
allowed for testing using cliloramine as the disinfectant.
Two hardware modifications to the flow system of DSS
Loop No. 6 were made to support the sensor evaluation
studies. A 50-gallon feed tank with a delivery line to the
intake side of the recirculation pump was added for the
purpose of introducing contaminants into DSS Loop
No. 6. Also, a sensor loop manifold (see Figure 2.1
and Figure 2.2) was fabricated for the purpose of
diverting water flow from the DSS Loop No. 6 to the
online monitors under evaluation, and to collect grab
samples for field and laboratory analyses.
DSS Loop No. 6 was equipped with a sensor manifold
incorporating the needed online sensors so that the
studies could begin quickly. Since DSS Loop No. 6 was
operated in essentially a closed mode, the observed
sensor responses were typical of a batch reactor
operation. Essentially, the sensor response seen for the
duration of a test run was similar to the case where a
contaminant slug would travel through the system for
the entire test duration (assuming minimal dispersion,
mixing and general disruption of slug due to flow
Figure 2.2 DSS Loop No. 6 Sensor Manifold and
Instrumentation Rack
variations). The recirculation mode within the tank also
dilutes the concentration of the contaminant in 24 hours
and does not represent a true plug flow system. Because
there are some technically valid differences as
compared to a "real world" distribution system, the
recirculation mode allowed for safer contained tests,
eliminated wastage of water, and allowed for easy
identification of viable sensors, prior to embarking on
studies using the Single Pass DSS as outlined in the next
section.
212 Single Pass DSS
The Single Pass DSS was constructed of 3-inch-diam-
eter glass-lined ductile iron pipe and spans the entire
length (150 feet) of the T&E facility high-bay area and
wraps back and forth across this expanse eight times.
The combined length of t his pipe is approximately 1,200
feet and the Single Pass DSS has a total capacity of
approximately 440 gallons. The pipe is gravity fed with
tap water via a 750-gallon stainless steel tank mounted
near the ceiling of the facility. This tank is supplied
from a floor-mounted 1,000-gallon stainless steel lank.
In-situ chemical feed tanks and mixers can be used for
chlorine dosing, chemical addition, or other similar
purpose. The contaminant injection port was in-stalled
immediately downstream of the 750-gallon feed tank. In
addition, two sampling ports were installed at 80-foot
and 1,180-foot distances from the contaminant injection
port. The two sampling ports supply sample water to
multiple instrumentation racks. Figure 2.3 shows a
schematic of the Single Pass DSS within the T&E
Facility.
Figures 2.4 and 2.5 show the Single Pass DSS running
the length of the T&E Facility high bay and wrapping
its length 4 times on the east side of the pipe rack. Fig-
ure 2.6 shows the sampling ports for the inlet located at
the top near the 80-foot mark and the outlet located
directly below this port at the 1,180-foot distance.

-------
water
pump
Figure 2.3 Schematic of Single Pass DSS
into the U.S. water supply. Municipal tap water sup-
plied by the GCWW was used as the water matrix for
this testing.
2.2.1 Tested Contaminants
Table 2.1 presents a summary of the broad classes of con-
taminants and specific contaminants tested by EPA along
with the associated test water matrix. The online instru-
mentation used to measure the individual water quality
parameter responses during the testing varied due to
various logistical reasons and the evolution of the testing
activity during the course of the research. For example,
most of the advanced optical instruments such as the Bio-
Sentn u. I-low C AM n. Spectro::lyser™, and Hach Fil-
terTrak™660 sc Laser Nephelometer were not procured
prior to beginning testing that utilized the recirculating
DSS Loop No. 6. These optical devices were purchased
later to evaluate their efficacy in detecting biological
contaminants. The BioSentrv® and FlowCAM® instru-
ments are designed to count and identify the injected
biological cells. Therefore, for the puiposes of evaluating
these instruments, the following biological contaminants,
surrogates, and growth (or carrier) media such as nutri-
ent broths were injected into the Single Pass DSS: three
micron beads, Escherichia coli jE coli), Jt coli (in de-
chlorinated water), bacteriophage male-specific (MS2),
Bacillus globigii (B. globigii), B. globigii (in dechlorinat-
ed water), secondaiy effluent from wastewater treatment,
sporulation media, sucrose. Terrific Broth, nutrientbroth
and Tiypticase soy™ broth. The biological contamina-
tion tests were performed in three distinct ways: 1) test
cells (centrifuged to isolate the contaminant only) inject-
ed with tap water, 2) test cells in nutrient or broth solu-
tions, 3) test cells in nutrient and broth solutions preceded
by treatment with dechlorinating agents such as sodium
thiosulfate pentahydride and sodium thiosulfate anhy-
drous. The last test was performed because real-world
contamination events might be conducted in conjunction
with dechlorination in an attempt to make the cells more
2.2 Test Contaminants and Water
Matrices
Target contaminants for the study were selected to be
representative of broad classes of biological and chemi-
cal contaminants that could be potentially introduced
Figure 2.4 Single Pass DSS - Longitudinal View
Figure 2.5 Single Pass DSS - Connecting Pipe Elbows
Figure 2.6 Single Pass DSS - Sampling Ports
2-3

-------
Table 2.1 Test Contaminant Matrix
Contaminant
Specific
Contaminant
Recirculating
Loop
Single
Pass
Class
Cl2a
NH2Clb
Cl2
Biologicals
Bacillus globigii


X

Bacteriophage MS2


X

Escherichia coli
X
X
X

Surrogate beads


X

Bacillus subtilis


X
Insecticides
Aldicarb


X

Nicotine

X
X

Real Kill®/Malathion
X
X
X

Dichlorvos


X

Phorate

X
X
Herbicides
Roundup® /Glyphosate
X
X
X

Dicamba


X

Basagran®


X
Culture Broths
Nutrient broth


X

Sporulation media


X

Terrific broth


X

Tryptic soy broth


X
Inorganics
Arsenic trioxide


X

Cadmium nitrate


X

Cesium chloride


X

Cobalt chloride


X

Lead nitrate


X

Mercuric chloride


X

Potassium cyanide


X

Potassium ferricyanide
X

X

Sodium arsenite

X
X

Sodium thiosulfate


X

Sodium fluoride


X
Warfare
Ricin
xc


Agents
G-type nerve agent
xc



V-series nerve agent
xc



Potassium cyanide
xc

X
Others
Blank (GAC water)

X
X

Secondary effluent
X
X
X

Colchicine


X

Dimethyl sulfoxide


X

Dye
X

X

Sucrose


X

Sodium fluoroacetate


X

Methanol


X

Airplane deicer


X

Antifreeze


X

Diesel fuel


X

DISPERSIT®


X
2-4
Contaminant
Specific
Contaminant
Recirculating
Loop
Single
Pass
Class
Cl2a
NH2Clb
Cl2
Others
Sodium Chloride


X
4-Methylcyclohexanemethanol


X
aChlorine bChloramines
^Testing conducted at the U.S. Army's Aberdeen Proving Ground's Edgewood
Chemical Biological Center (ECBC) Facility.
viable. The presence of free chlorine at the typical residu-
al levels [~1 milligrams per liter (mg/L)] is deleterious to
many biological organisms and reduces the efficacy of a
biological attack. The bacteriophage MS2 tests were per-
formed to simulate a viral threat. To evaluate the impact
of nutrient broth and the dechlorinating agents, the fol-
lowing "control" injections were also performed: sodium
thiosulfate pentahydride, sodium thiosulfate anhydrous,
sucrose, terrific broth and nutrient broth.
2.2.2 Test Water Matrix
The GCWW water supply to the T&E Facility comes
from the Miller Plant, which treats water from the Ohio
River. GCWW uses chlorine as the residualdisinfectant
for water distribution. The background range of values
for the routinely measured water quality parameters at
the T&E Facility are as follows: free chlorine - 0.8 to
1.1 mg/L, specific conductance - 300 to 600 microsie-
mens per centimeter (|iS/cm). oxidation reduction po-
tential (ORP) - 500 to 700 millivolts (mV), potentialof
hydrogen in standard units (pH) - 8.5 to 8.8, turbidity
<0.1 nephelometric turbidity units (NTU), and total or-
ganic carbon (TOC) - 0.3 to 1.3 mg/L. Only the free
chlorine levels were adjusted as needed (prior to test-
ing) such that the levels were approximately 1 mg/L.
The chloraminated water was prepared in batches us-
ing a 2,400-gallon tank. GCWW-supplied tap water was
collected in a 2,400-gallon tank at the EPA T&E Facility
and tested for total chlorine residual. Calculations were
made to determine the correct amount of sodium
hypochlorite necessary to raise the total chlorine
concentration to the desired level, usually 2 mg/L.
When this concentration was achieved and verified by
analysis, ammonium hydroxide was added insufficient
quantity (chlorine to ammonia ratio of 4:1) to convert
the free chlorine into combined chlorine. The resulting
chloraminated water was mixed for 15 to 20 minutes
and retested for both free and total chlorine.
2.3 Water Quality Measurement
Prior to introduction of contaminants, water-quality
sensors located within the selected test apparatus (i.e.,
DSS Loop No. 6 or Single Pass DSS) were typically
monitored for an hour to establish normal (baseline)
conditions. After contaminant injection, data from the
various sensors were monitored and recorded. The sen-

-------
Table 2.2 Measured Water Quality Parameters
Parameter
Measurement Type
Online Instrumenta-
tion Tested
Parameter Applicability
Ammonia - nitrogen
Continuous and grab
YSI 6600, YSI 6920DW
Naturally occurring form of nitrogen in the nitrogen cycle. Dis-
solved ammonia gas is toxic to aquatic life at concentrations as
low as 0.2 milligram per liter (mg/L). Will be converted to
chloramine in chlorinated drinking water.
Apparent color
Grab
Various laboratory in-
struments, Six-Cense™
Visible color resulting from turbidity and dissolved materials
(humic material, dissolved metals, dyes, algae). Potable water is
normally colorless after treatment.
Biological Agents
Continuous
EZ-ATP® Analyzer,
Sporian, inline
biosensor system (IBS),
ZAPS LiqulD
Biological agents in drinking water can cause numerous adverse
health effects.
Chemical Oxygen
Demand (COD)
Continuous
PeCOD® P100 COD
Analyzer, RealTech
UVT
May be correlated to total organic carbon.
Chloride
Continuous and grab
YSI 6600, YSI 6920DW
Indicator of salinity. Associated with a secondary maximum
contaminant level (MCL) of 250 mg/L in drinking water.
Conductivity
measured as
specific
conductance3
Continuous and grab
YSI 6600, YSI 6920DW
Hydrolab® DS5, Troll®
9000, Six-Cense™,
Hach/GLI Model C53
Conductivity Analyzer
Ability of water to carry an electrical current. Strong indicator of
dissolved salts. Serves as a surrogate for total dissolved solids.
Dissolved oxygen
(DO)
Continuous and grab
YSI 6600, Hydrolab®
DS5, Troll® 9000, Six-
Cense™
Concentration of oxygen dissolved in water can serve as an
indicator of chemical and biochemical activity in water.
Fluorescence
(total, humic and
bacterial)
Continuous
Spectrophotometric
ZAPS MP-1, ZAPS
LiqulD, Turner TD1000C
Instrumental measure of fluorescence at various wavelengths.
Free chlorine
Continuous and grab
YSI 6920DW, Hydro-
lab® DS5, Troll® 9000,
Six-Cense™, Hach CL17
Free Chlorine Analyzer
Chlorine is added to the DSSb in the form of sodium hypochlorite.
Chlorine levels in drinking water are controlled at ~1 mg/L.
Metals
Continuous
QBiSci Online Water
Sensor
Metals in drinking water may be toxic to living organisms.
Multi-angle light
scattering (MALS)
Continuous
BioSentry®
Utilizes laser-produced MALS technology to generate unique bio-
optical signatures for classification using JMAR's pathogen
detection library.
Multi-spectrum (UV-
Vis) absorption
Continuous
Spectrophotometric
Spectro::lyser™ or
Carb::olyser™,
ZAPS LiqulD,
PeCOD® P100
COD Analyzer,
ChemScan® UV-
2150
UV-Vis excitation that provides a means of estimating absorption
at various wavelengths. Nitrate and/or nitrite concentration, DOCc,
TOC, CODd and BODe (depending on the used algorithm), and
turbidity. Information at nearly any wavelength between 200 and
750 nm.
Nitrate - nitrogen
Continuous, grab, and
spectrophotometric
YSI 6600, YSI 6920DW
Essential nutrient for plants and animals. Nitrate is the most
soluble form of nitrogen. Causes health problems in humans.
Drinking water standard is 10 mg/L.
Oxidation-reduction
potential (ORP)
Continuous and grab
YSI 6600, YSI 6920DW,
Hydrolab® DS5, Troll®
9000, Six-Cense™,
Hach/GLI Model P53
pH/ORP Analyzer
Indicator of dissolved oxidizing and reducing agents (metal
salts, chlorine, sulfite ion). ORP values above 700 millivolts
(mV) kill unwanted organisms in drinking water. A ground water
incursion may lower ORP by increasing chlorine demand.
Chlorination of drinking water produces an ORP background of
~700 millivolts in GCWW water.
Particle Count
Continuous
Hach 2200 PCX Particle
Counter
Counts all particles that are between 2 and 750 |jm in size.
The counted particles can be subdivided into 32 size ranges
to identify particles of interest. For example, the particle size
ranges could be selected to correspond to biological organisms
such as Giardia (6-10 |jm) and Cryptosporidium spp. (2-5 |jm).
2-5

-------
Particle count
and image-based
identification
Continuous
FlowCAM®
Measures particle size, count and shape. Images particles
between 2 |jm and 3 mm in size. Helps to identify and classify
particles based on library of images.

-------
Table 2.2 (continued) Measured Water Quality Parameters
Parameter
Measurement Type
Online Instrumenta-
tion Tested
Usefulness of Parameter for Water Quality
PH
Continuous and grab
YSI 6600, YSI 6920DW,
Hydrolab® DS5, Troll®
9000, Six-Cense™,
Hach/GLI Model P53
pH/ORP Analyzer
Indicator of hydrogen ion activity (acidity or alkalinity) of water.
Most chemical and biochemical processes are pH dependent.
Carbon dioxide/bicarbonate/ carbonate and ammonia/ammonium
equilibria are pH dependent. pH of drinking water is well
established and controlled. A change of more than 0.5 pH unit
indicates a problem.
Temperature
Continuous and grab
YSI 6600, YSI 6920DW,
Hydrolab® DS5, Troll®
9000, Six-Cense™,
Hach/GLI Model C53
Conductivity Analyzer,
Hach/GLI Model P53
pH/ORP Analyzer
A measurement indicator of how hot or cold the water is. DO and
specific conductance change with temperature. Biological and
chemical activities are heavily influenced by water temperature.
Total cyanide,
malathion, and
glyphosate
Grab
Various laboratory
instruments
Compound-specific laboratory analysis for the purpose of deter-
mining the fate of these three contaminants in the DSS.
Total organic
carbon (TOC)
Continuous and grab
Hach astroTOC™ UV
Process Total Organic
Carbon Analyzer, Siev-
ers® 900 On-Line Total
Organic Carbon Ana-
lyzer, Spectro::lyser™
or Carb::olyser™,
RealTech UVT
Dissolved plus particulate organic compounds. Can range from 0.5
to 25 mg/L in drinking water in the U.S. May be correlated to
chemical and biological oxygen demand.
Transmission
Continuous
Spectrophotometric
ZAPS MP-1,
Spectro::lyser™ or
Carb::olyser™, ZAPS
LiqulD
Measure of color based on Beer's Law as measured by photon
transmission through water [800 nanometers (nm) for this study].
Turbidity
Continuous and grab
YSI 6600, YSI 6920DW,
Hydrolab® DS5, Troll®
9000, Six-Cense™,
1720D Turbidimeter,
Hach FilterTrak™ 660 sc
Laser Nephelometer
Indicator of suspended matter and microscopic organisms. Patho-
gens are more likely to be present in highly turbid waters.
Ultraviolet 254
nanometer
wavelength (UV254)
absorption
Continuous
Spectrophotometric
ZAPS MP-1,
Spectro::lyser™ or
Carb::olyser™, RealTech
UVT, Hach UVAS
Measure of organic compounds that absorb photons at 254 nm.
Indicative of organic compounds with aromatic chemical structure
and conjugation.
aSpecific conductance is defined as the raw solution conductivity, compensated to 77°F (25°C).
bDSS = Distribution System Simulator.
cDOC = Dissolved Organic Carbon.
dCOD = Chemical Oxygen Demand.
eBOD = Biological Oxygen Demand.
sor data were supported by the analysis of grab samples
taken from the test apparatus at discrete intervals. For
experimental control, uncontaminated test water
matrix was injected into the test apparatus. During the
testing, it was verified that the act of injection did not
affect baseline conditions as characterized by sensor
response.
2.3.1 Measured Water Quality Parameters
A variety of water quality parameters was measured
during the testing period. The specific instrumentation
used in individual test runs for both DSS Loop No. 6
and the Single Pass DSS was dependent on the
availability of instrumentation during the testing pe-
riod. Table 2.2 presents an overall summary of the
2-7
measured water quality parameters and a summary of
the usefulness of each measurement in terms of water
quality.
2.3.2 DSS Loop No. 6 Online Instrumentation
The following are online water quality monitoring
sensor instruments that were evaluated during the var-
ious DSS Loop No. 6 test runs: YSI 6600, Hydrolab®
DS5, Troll® 9000, Six-CENSE™, Hach Water Dis-
tribution Monitoring Panel (WDMP), and Zero Angle
Photon Spectrometer (ZAPS) MP-1. Figure 2.2 (previ-
ously shown) and Figure 2.7 depict most of the online
instrumentation evaluated during the DSS Loop No. 6
testing.

-------
UV-2150, and Optiqua Refractive Index. Figure 2.9
depicts the instrument panel that contains the controller
for the Carbo::lyser™, controller for the Hach
Filter/Trak™ 660 sc Laser Nephelometer, and the
Mow C AM R device.
In addition to these instruments, EPA is also evaluating
the radiation monitor (Technical Associates. Canoga
Park, California) at the National Air and Radiation En-
vironmental Laboratory (NAREL) in Montgomery, Ala-
bama. The results from these tests were not available at
the time of production of this document. Figure 2.10 de-
picts the radiation monitor.
Figure 2.10 Technical Associates Radiation Monitoring
Device
Figure 2.7 DSS Loop No. 6 - Online
Instrumentation
2.3.3 Single Pass DSS Online Instrumentation
The following are online water quality monitoring
sensor instruments that were evaluated during the
various Single Pass DSS test rans: Hach CL17 free
chlorine analyzer; Analytical Technology, Inc.
Model A15/62 free chlorine monitor; YSI 6920DW;
Wallace & Tiernan® Depolox® 3 plus; Hach astro-
TOC™ UV process TOC analyzer; Hach WD MP:
Sievers® RL; and Sievers® 900 On-Line TOC
Analyzer. New custom instruments tested for this
Updated Guide include the following: PeCOD® P100
COD Analyzer, EZ-ATP® Analyzer, Sporian IBS,
and QBiSci Online Water Sensor. Figure 2.8 shows
two Single Pass DSS instrument panels.
Figure 2.8 Single Pass DSS Instrument Panels
2.3.4 Single Pass DSS Online Optical
Instruments
The following are online optical instruments that were
evaluated during the various Single Pass DSS test runs:
Carbo::lyser™ and Spectro::lyser™, BioSentry®,
Mow CAM M. Hach FilterTrak™ 660 sc Laser Neph-
elometer, and Hach 2200 PCX Particle Counter. New
online optical instalments tested for this Updated Guide
include the following: ZAPS LiquID. Turner
TD1000C, HachUVAS, RealTechUVT, ChemScan®
Figure 2.9 Various Single Pass DSS Optical
Instruments
2.4 Data Collection andAnalysis
Data collected for each parameter from the online wa-
ter quality sensor instruments were complemented by
laboratory analyses of grab samples. To facilitate com-
parisons between the online monitoring results and lab-
oratory analyses, sensor responses to contaminants for
each parameter were plotted along with associated grab

-------
Figure 2.11 NexSens iSIC Data Acquisition System
sample results. These plots allowed a graphic inter-
pretation of the data to 1) evaluate changes in baseline
conditions due to contaminant introduction, 2) compare
sensors (using different technologies to measure the
same parameter), and 3) recognize false negative/false
positive responses by visual comparison to the gxab
sample data.
2.4.1 Data Collection
Wherever possible, each of the online sensors was
connected to a data acquisition system. The intelli-
gent Sensor Interface and Control (iSIC) system was
connected to the data collection personal computer
(PC) via hardwire or radio (as appropriate). The data
collection PC ran the iChart software program, which
polled the connected iSIC(s) and monitoring devices
every 2 minutes and recorded the data reported by the
instrumentation. The 2-minute data collection cycle
was considered to be optimum because of the number
of instruments concurrently tested that needed to be
polled for data and the measurement cycle limita-
tions of some tested devices. The iSIC/iChart system
was selected as the data collection platform because
it incorporated many pre-built device drivers that
could communicate with the widest variety of online
instrumentation tested at the T&E Facility. A more
detailed discussion of the data collection system is
presented in Chapter 4.0. Figure 2.11 shows the Nex-
Sens iSIC data acquisition system.
2.4.2 Data Analysis
The data plots generated from the tests conducted at the
T&E Facility were analyzed visually to construct a
qualitative response matrix for the contaminants test-
ed. The criteria for determining a "significant change"
was subjective at the early stages of the research The
sensor responses were plotted over the course of the test
rims and analyzed for visually significant changes.
Thereafter, a more robust analysis was performed
where the absolute change, percent change, and sig-
nal-to-noise (S/N) ratio for each measured parameter
was computed. See Chapter 5.0 for further details.
At the onset of this testing effort, EPA determined
that an automated algorithmic analysis of the online
data was essential. Therefore, concurrent to the test-
ing, EPA initiated collaboration with SNL for the
development of the CANARY Algorithm (previously
described in Chapter 1.0). In addition, EPA continues
to evaluate other commercial data analysis
algorithms/products as they became available
(Umberg etal., 2009).
2.5 Teaming with EPA's Water Security
Initiative
The work conducted at the T&E Facility assisted
EPA's Water Security initiative (WSi - formerly,
WaterSentinel). As described in Section 1.2, the
EPA's OGWDW-WSD worked collaboratively with
NHSRC to deploy a pilot network of water quality
monitoring instrumentation at GCWW as a part of
the WSi pilot in Cincinnati (EPA, 2008c). Figures
2.12 and 2.13 show two types of instrument panels
deployed at the first pilot utility. The panels contain
online instrumentation to measure free chlorine,
TOC, pH, ORP, conductivity, temperature and tur-
bidity. The Type A panels utilize all Hach
instrumentation, whereas the Type B panels utilize
instrumentation from manufacturers other than
Hach. As will be discussed in Chapter 5.0, free
chlorine and TOC were found to be most useful trigger
parameters in chlorinated water systems.
Figure 2.12 First Pilot Utility - Water Security
Initiative Instrument Panel Type A
2.6 EPA's Future Water Quality
Sensor Research
EPA, through their Technology Testing and Evaluation
Program (TTEP) and testing activities at the T&E
2-9

-------
~ Water Utilities arid
Sensor Manufacturers
•	Online sensors were tested in simulated distribution
systems using chlorinated and chloraininated
waters. The simulated systems were injected with
a variety of target contaminants to evaluate the
individual sensor/parameter response.
•	Grab sampling is critical to verify online sensor
responses.
•	A data polling frequency of two minutes was found
to be optimal for the wide range of sensors tested,
but utilities might want to evaluate other polling
frequencies.
•	A robust Supervisory Control and Data Acquisition
(SCADA) system is needed to fully utilize and
process in real-time the large volumes of data that
are generated.
•	The EPA T&E Facility distribution system
simulators attempt to replicate field conditions, but
the effects of varying water demands were not
simulated during the tests. In addition, the
background water quality parameter levels are very
stable at the T&E Facility. Therefore, for utilities
with varying background water quality parameters
compounded with vary ing demands, the simulated
tests might result in different sensor/parameter
response.
•	Instrument manufacturers need to design for
allowing an automated grab sample to be collected
to validate the instrument response as needed.
Figure 2.13 First Pilot Utility - Water Security Initiative
Instrument Panel Type B
Facility, will continue to identify and evaluate promising
sensor technologies for potential use in CWS, as funding
allows. Radiological and low density biological
detection equipment testing are the key current sensor-
related data gaps. New technologies are needed to reduce
the current capital, operational, and maintenance costs in
order for CWS programs to be sustainable. Information
on sensor evaluation programs can be obtained at:
https://www.epa.gov/waterresilience.
WSi is a program designed to address the risk of in-
tentional contamination of drinking water distribution
systems. Initiated by OGWDW in response to HSPD-9,
the overall goal of WSi is to design and deploy CWS for
drinking water utilities. EPA is implementing the WSi in
three phases: (1) development of a conceptual design that
achieves timely detection and appropriate response to
drinking water contamination incidents; (2)
demonstration and evaluation of the conceptual design in
full-scale pilots at drinking water utilities; and (3)
issuance of guidance and conduct of outreach activities
to promote voluntary national adoption of effective and
sustainable drinking water CWS. The initial full-scale
pilot was implemented in Cincinnati, Ohio. EPA-
OGWDW plans to implement more pilot studies utilizing
the CWS concept presented in Section 1.3. These pilot
studies will be conducted at several utilities to
demonstrate that a functional CWS can be deployed
under a variety of real-world conditions. Additional
information related to EPA's water security research can
be obtained at: https://www.epa.gov/emergency-
response-research/water-security/.
2-10

-------
2-11

-------

-------
3.0 Instrument Setup
and Data Acquisition
As previously mentioned in Section 2.5 (and will be dis-
cussed later in Chapter 5.0), the tests conducted at the
EPA T&E Facility show that free chlorine and TOC are
the most useful water quality parameters for detecting
changes indicative of contamination in chlorinated wa-
ter systems. The prototype monitoring panels installed
by the WSi pilot utility included online instrumentation
to measure free chlorine, TOC, pH, ORP, conductivity,
temperature and turbidity. Each utility should evaluate its
needs and resources (both capital and labor), and re-
view the test results associated with its water distribution
system before selecting a suite of online parameters and
associated instrumentation. The use of chloramines as
disinfectant should also be taken into account when se-
lecting the parameters and instrumentation. Once the pa-
rameters and instruments have been selected, they should
be set up in accordance with the instructions provided by
the manufacturer for flow, pressure, and sample condi-
tioning requirements. This chapter discusses in detail the
various requirements for setting up online water quality
sensor instrumentation at a specific site.
e-Specif • ' rements
EPA-developed software such as TEVA-SPOT should
be used to identify the optimal locations of a fixed
number of sensors. After a potential monitoring site has
been identified using TEVA-SPOT, a site visit should be
performed to ensure that the selected site has:
•	sufficient environmentally protected secure
space for housing the selected instrumentation
•	access and a clear path for transporting and
servicing the instrumentation to conduct
installation and maintenance activities
•	adequate source of pressurized and pressure-
controlled water supply for the proposed
instrumentation
•	drainage access to discharge the water analyzed
by the online instrumentation
•	necessary power supply and backup
(uninterrupted power supply) to power the
online instrumentation, data collection, and data
transmission systems
•	appropriate media (wired or wireless) for
transmitting the online data in real-time to a
specified data collection center
•	water quality characteristics that are suitable (or
can be appropriately conditioned) for analysis
by selected online instrumentation
Depending upon the threat and vulnerability analysis,
some of the selected sites might not meet all of the
requirements. For such sites, alternate means of
meeting a site-specific requirement should be inves-
tigated. For example, all sites might be not suited for
deploying a single communication technology. In such
cases, a combination of wired and wireless
communication technology should be investigated.
Another example could involve a situation where the
initial water quality is not suitable for selected
instrumentation. In this case, either alternate instru-
mentation should be investigated or site/instrument-
specific sample water conditioning could be per-
formed such as pH buffering, degassing, or removing
iron and salts.
3.1.1 Environmentally Protected Housing
The selected site should be environmentally protect-
ed and secure. Many of the online sensors are typi-
cally contained in a National Electrical Manufactur-
ers Association (NEMA) class 4- or 4X-compliant
corrosion-proof enclosures and protected from wind-
blown dust, rain, sleet and external icing. Specifi-
cally, a NEMA4-compliant enclosure has to pass the
"Hose Test," which is described as: a 1-inch nozzle,
delivering 65 gpm of water, from a distance of 10 feet,
from all directions, for a 5-minute time period, with
no water leak to the interior. Class 4X enclosures have
additional protection against corrosion. Preferred
materials for mounting (or housing) the online
instrumentation are polyester/glass, stainless steel,
and epoxy coatings. Although the selected enclosure
might be suited for general outdoor application, there
is an additional need for temperature and humidity
control because the advanced devices are equipped
with onboard computers and electronics that might not
withstand the temperature, humidity, and altitude
extremes. The environmental tolerances are
instrument-specific, and the manufacturer instructions
should be followed to ensure the suit- ability of the
selected housing. In general, it is not recommended
that the instruments be housed in an environment
where the temperature exceeds 90°F (32.2°C) or falls
below 40°F (4.4°C). Appropriate cooling and/or
heating devices should be installed at the site as
needed.
Furthermore, the selected instrumentation might have
humidity specifications, for example, a range of 5 to
95%. High humidity might result in corrosion of elec-
tronic components and/or could lead to short circuits
and malfunction. Humidity can increase the conductivi-
ty of the embedded electronics, leading to short circuits
and malfunction. Condensation is another problem that
can cause electronic devices to malfunction. For exam-
ple, when an instrument is moved from a colder place

-------
to a wanner and more humid place, condensationcould
coat circuit boards and other insulators, leading to short
circuiting inside the equipment. Such short circuits
might cause substantial permanent damage if the equip-
ment is powered on before the condensation has evapo-
rated. Electronic equipment should be acclimatized for
several hours (as specified by the manufacturer) before
powering on.
3.1.2 Access for Servicing the Instrumentation
The online instrumentation generally requires peri-
odic servicing, calibration and reagent replacement.
In a multi-instrument setup, individual instruments
are mounted on panels that are fabricated to fit the
space requirement and also provide easy access for
servicing each instrument. The instruments are set up
such that the servicing need for a single instrument
does not disrupt the function of other instru-
mentation. Also, the water intake and drain lines are
configured in a manner such that they are generally
below the instrumentation so that any water line
failure does not damage the instrumentation. In
addition, the power conditioning, data logging, and
communication equipment are separated from the
online instrumentation. Power conditioning devices
are designed to regulate the voltage and improve the
power quality (e.g., electrical noise suppression and
transient impulse protection). Figures 3.1 and 3.2
show instrument panels that have been designed spe-
cifically to facilitate online monitoring at the EPA
T&E Facility. For example, the sensor shown in Fig-
ures 3.1 and 3.2 has all the water lines at the bottom.
The sample inlet line to each instrument is isolated.
The drain lines are connected to the available floor
drain. The data logging equipment is on the back of
the panel and the power lines are on top.
The operator should be able to see and service the
instrumentation/data collection components with
Figure 3.1 Single Pass DSS Instrument Panel at 80-foot
Sampling Location
ease. If the instrument panel is improperly designed, the
operator might take shortcuts while servicing, which
could lead to lower data quality or equipment
malfunctions due to improper servicing.
3.1.3 Pressure-controlled Water Supply
The majority of the water quality instrumentation is
sensitive to fluctuations in water supply pressure.
Pressure changes can create bubbles (degassing) in the
sampled water, resulting in erroneous data. Pressure
regulator valves are used to allow water from a high-
pressure supply line (or tank) to be reduced to a safer
preset level specified by the instrument
manufacturer(s). Pressure regulators are also suscep-
tible to changes in the water supply pressure. Some-
times, it might be necessary to have multiple layers of
pressure regulation to dampen any effects of pressure
fluctuations on the instrument readings. Instruments
like particle counters require separate mounted
constant-head overflow weir mechanisms so that the
sample outlet can be raised or lowered to the height that
will produce the desired flow. Figure 3.3 shows the
constant head mechanism for a particle counting device.
By pushing water up a fixed-height column and col-
lecting the sample stream from that column, a constant
foot Sampling Location
3-2

-------
cuits to each instrument should be isolated via a circuit
breaker, and both the panel and instruments should be
appropriately grounded. The electrical isolation allows
for servicing of individual instrumentation without dis-
rupting the other equipment installed at the location.
Circuit breakers protect the instrumentation from elec-
trical surges and short circuits. Connection to ground is
a safety issue designed to protect the personnel ser-
vicing the instrumentation and the online instrumenta-
tion. The ground connection also helps limit build-up of
static electricity on the instrumentation. In areas where
the line voltage is known to fluctuate, a surge protector
is also recommended. The surge protector regulates the
voltage supplied to the instrument by either blocking or
by shorting to ground connection when voltages above
safe instrumentation thresholds are sensed in the circuit.
A UPS or an inline battery backup lasting four to eight
hours is recommended, because it continuously powers
and protects the instrumentation from the previously
described power problems. A UPS is also known as a
power or line conditioner because of this ability. AUPS
generally contains a lead-acid battery for storing power.
During electrical outages, the energy reserves stored in
the UPS are used to power the instrumentation. Figure
3.4 shows a field data communications NEMA 4
enclosure with backup UPS Power.
Figure 3.3 Example Constant Head Mechanism for
Hach 2200 PCX Particle Counter.
pressure is delivered to the instrument. This method of
regulated sample delivery, although simple, is veiy
effective in controlling pressure and flow fluctuations.
Hach uses this method of sample delivery for Hach CL-
17 free (or total) chlorine analyzers and Hach 2200 PCX
Particle Counters. The Bio Sentry® unit also employs a
similar method for delivering constant sample flow.
3.1.4	Drainage Access
The sampled water drawn from the online instrumen-
tation panel needs to be discharged appropriately to
meet local discharge requirements. Generally, accessto
a sanitary sewer line is sufficient. In certain locations,
such access might not be easy. Care should be taken so
that water does not pool near the instrumentation,
causing a slipping hazard. A drain manifold is recom-
mended for locations with multiple online instruments.
The drain line should be sized adequately, taking into
account any instrument and inlet line failures.
3.1.5	Power Supply and Electrical Protection
Adequate power supply (preferably 3-phase) with a
backup device for uninterrupted power supply (UPS)
intended to provide sufficient power for the online in-
strumentation, data collection, and data transmission
systems is recommended. In addition, the electrical cir-
Figure 3.4 Field Communications Enclosure.
3-3

-------
Some instruments provide the option for portable or line
power. For example, the YSI 6920DW instrument can
be powered by batten or line power. Line power is
preferred in installations where the line is backed up
with an appropriately sized UPS.
3.1.6	Transmission Media Access
In order to fully realize the benefits of online instru-
mentation, appropriate media (wired or wireless)
should be used in real-time for transmitting the data or
information to a pre-specified data collection location.
Wired media generally provide higher bandwidth but
could be cost-proliibitive in certain locations. In these
cases, the use of wireless media (e.g., licensed or un-
licensed radio, cellular, satellite-based transmission
media options) should be investigated. In some cases,
depending upon the location and media used, the data
transmission might be susceptible to various types of
interferences. In such cases, additional programmatic
error control techniques should be applied to mitigate
the errors during transmission. For example, in a poll-
based data collection platform, it might be necessary to
either increase the number of retries on a failed poll
event or adjust the data packet reception window based
on the bandwidth and latency limitations of the selected
media.
3.1.7	Source Water Quality Adjustment
Generally, the selected instruments need to be suitable
to analyze the source water quality. In some cases, the
source water quality can be adjusted to meet the instru-
ment specifications. For example, certain free chlorine
measuring devices require the pH of the water to be
below 8.5 standard units. If the pH at the selected lo-
cation is above 8.5, appropriate buffering agents (e.g.,
carbon dioxide) should be used to condition the pH of
the sampled water or an alternate online monitoring
instrument should be selected for that parameter. For
example, the Sievers® RL unit is not appropriate for
liighpH water (>8.5).
3.1.8	Instrument-Specific Accessories
As discussed in the previous section, for some instru-
ments, there might be a need for either peripheral
support equipment (or accessories) that precondition
the sample water or for carrier gases to complete the
analysis. The Hach TOC monitor is another example
of an instrument that requires specific accessory
equipment as identified below.
In general. TOC monitors are one of the more complex
instruments to operate. Hach uses the ultraviolet (UV)
persulfate method; tliis method requires reagents
(sodium persulfate and phosphoric acid) to drive the
oxidation reaction. These reagents are supplied in 5-
gallon carboys and are bulky to handle.
3-4
This instrument also requires a clean, carbon dioxide
(C02)-free air source to carry sample flow to the C02
detector. The C02-free air source is supplied either by
a cylinder of liquid nitrogen, or a zero-air generator. If
a zero-air generator is used, an air compressor is needed
to supply a constant stream of air. A considerable
amount of space is required to house this monitor and
its supporting equipment . These units have proven to be
fairly labor-intensive to operate and require a highly
skilled technician to perform maintenance and
calibration procedures. Figure 3.5 shows a Hach
astroTOC™ UV process TOC analyzer instrument and
associated zero air system.
The Sievers® 900 On-Line TOC Analyzer also uses
the UV persulfate method. Similar to the Hach unit,
the Sievers® 900 On-Line TOC Analyzer can be
fairly labor-intensive to operate and requires a highly
skilled technician to perform maintenance and cali-
bration procedures. However, this instrument and its
reagent packs are more compact than the Hach unit.
Also, the Sievers® 900 On-Line TOC Analyzer does
not require an external zero air system/compressor or
a liquid nitrogen Dewar. This unit does require an
inorganic carbon remover (ICR) for waters that are
heavily laden with inorganic carbon. Figure 3.6
shows a Sievers® 900 On-Line TOC Analyzer (the
ICR is contained inside the instrument enclosure).
Unless the utility has extensive in-house experience
with these instruments, it might be prudent to procure
service contracts for each of the aforementioned TOC
units. Surrogate TOC monitoring equipment using
UV-visible (UV-Vis) spectral absorbance has been
found to be less labor intensive, but trade-offs in its
limited ability to detect a variety of potential organic
contaminants should be taken into consideration.
Figure 3.5 Hach astroTOC™ UV Process Total Organic
Carbon Analyzer.

-------
water utility perspective, a SCADA system generally
consists of the following four components:
1.	a Human-Machine Interface (HMI), which is a
combination of computer software and hardware
that presents information to an operator; the
operator is able to monitor and control the
process/instrumentation through this interface.
2.	a supervisory or a central node that gathers data
from a programmable logic controller (PLC)
and/or a Remote Terminal Unit (RTU) for
presentation to the operator through the HMI
and sends commands to the PLC/RTU based on
the operator inputs from the HMI.
3.	PLCs/RTUs connected to the online
instrumentation that convert sensor signals to
digital data (inputs) and send commands to
connected automated devices (such as sampling
devices and pumps) to perform a pre-defined
task based on operator commands from the
HMI.
4.	a data communication infrastructure connecting
the supervisory system to the PLC/RTU.
Figure 3.6 Sievers® 900 On-Line Total Organic Carbon
Analyzer.
3.2 Calibration Materials/Reagents
and Onsite Accessories
During the setup, instrument calibration material,
reagents and accessories ought to be available (as
needed) to ensure that the instruments are operating as
recommended by the manufacturer. Many reagents and
calibration solutions have expiration dates; therefore,
these reagents should be ordered according to the
instrument-specific maintenance schedule. While per-
forming calibration and maintenance activities, each
manufacturer's procedure needs to be followed to en-
sure that the instrument is performing properly and is
measuring the water quality in the designated range.
Also, proper calibration ensures that the quality of the
data is reliable.
3.3 Data Acquisition System
Most water utilities implementing a network of online
instrumentation generally have some type of SCADA
system. SCADA systems are also known as industrial
control systems and are capable of monitoring and con-
trolling a process. Generally, water treatment plants are
automated with some type of SCADA system. From a
Figure 3.7 shows a data flow schematic from field-de-
ployed online instrumentation to the operator in a con-
trol center.
Historically, SCADA system hardware and software
tend to be proprietary. Water utilities that have invested
in a particular manufacturer's solution might find them-
selves restricted to limited choices for equipment when
considering system expansions or upgrades. However,
most SCADA systems can communicate with sensors
or instrumentation that can provide their data output in
4 to 20 milliamperes (inA), or through serial protocols
such as Recommended Standard 232(RS-232)/Recom-
mended Standard 485 (RS-485). The RS-232/RS-485
electrical specifications are defined by the Electronic
Industries Alliance (EIA) for a serial communications
channel.
I/O

Remote

Comms

Master
Meters 4

~ PLC 4

^ Protocols 4

^ SCADA Server
Sensors

RTU Controller

Ethernet

HMI
Field Devices



Serial






Wireless


Field	Control
Devices	Center
Figure 3.7 SCADA Data Flow Schematic
3.3.1 4 to 20 Milliamperes Current Output
Developed in the 1950s, the 4 to 20 inA instrument out-
puts are still widely used by SCADA and instrument
3-5

-------
manufacturers. This output format is ideally suited for
low-cost instruments that provide one or two analog
output values. Generally, the output signal can travel
distances of around 50 meters. In other words, the PLC/
RTU capturing this data output from the instrument
must be located within 50 meters of this instrument.
This output is easy to understand and troubleshoot: a
signal of 4 mA represents zero percent of the output
span, and 20 mA represents one hundred percent signal
output span. For example, a chlorine monitor calibrated
to measure a span between 0 and 5 parts per million
(ppm) will provide a corresponding analog output be-
tween 4 and 20 mA when reporting these values. Trou-
bleshooting the output is simple, requiring only a digi-
tal voltmeter to read the values inline.
3.3.2	Serial Protocols
Developed in the 1960s, the RS-232 is a serial protocol
for sending and receiving signals between a Data Ter-
minal Equipment (DTE) and Data Circuit-terminating
Equipment (DCE). Prior to the popularization of the
Universal Serial Bus (USB), the RS-232 serial port was
commonly available with all types of personal com-
puters. The RS-232 connection (at a minimum) needs 3
wires to communicate where one wire is dedicated to
transmitting data, one to receiving data, and one is
ground. The RS-232 can even use a two-wire connec-
tion (data and ground) if the data flow occurs one way.
The RS-232 standard defines the voltage levels that
correspond to logical one and logical zero levels. Valid
signals are plus or minus 3 to 15 volts.
RS-485 (also known as EIA-485) is a multipoint serial
communications channel that can span distances of up
to 4,000 feet. The multipoint communication is often in
a master-slave arrangement when one device dubbed
"the master" initiates all communication activity with
other devices in the network. RS-485 is used as the
underlying protocol in many standard and proprietary
SCADA protocols, including the most common ver-
sions of Modbus.
3.3.3	Data Communication Protocols
Typical legacy SCADA communications protocols in-
clude Modbus (developed by Modicon), RTU Protocol
(RP-570), and PROFIBUS. These communication pro-
tocols are all proprietary and SCADA-manufacturer
specific but are widely adopted and used. During the
late 1990s, many of the SCADA manufacturers shifted
toward more open communication protocols and ad-
opted the "tie facto" open message structure offered by
Modbus over serial communications protocols such as
RS-232/RS-485. Since 2000, most SCADA manufac-
turers are offering greater open interfacing operability
by adopting standards such as Modbus Transmission
Control Protocol (TCP) over Ethernet and Internet Pro-
tocol (IP). Other standard communications protocols
include: International Electrotechnical Commission
(IEC) 60870-5-101 or 104, IEC 61850 and Distributed
Network Protocol 3. These protocols are standardized
and recognized by all of the major SCADA manufac-
turers. Similar to Modbus, many of these protocols now
contain extensions to operate over TCP/IP.
3.3.4	SCADA Setup and Poll Rate
Generally speaking, most water utilities embarking on
online monitoring have some type of SCADA system in
place. It is generally cost-effective to expand on existing
SCADA systems to accommodate online water quality
monitoring and integrate it with distribution system and
treatment plant operations as needed. For a large utility
(serving >100,000 persons), it is common to have tens
of thousands of SCADA tags [or SCADA input and out-
put (I/O) values] that are polled by the SCADA "mas-
ter device" periodically. Depending upon the poll cycle
and available data bandwidth, polling most of these
SCADA I/O values every one to five minutes is com-
mon. The online water quality instruments themselves
have a sampling cycle, and a vast majority of these have
sampling and reporting cycles of less than one minute.
However, in some cases, the sample cycles for achiev-
ing peak measured values might be between four and
eight minutes (e.g., Hach astroTOC™ UVprocess TOC
analyzer and Sievers® 900 On-Line TOC Analyzer).
The data acquisition system used at the EPA T&E Facil-
ity was set to poll every two minutes. Based on a review
of the data generated during this testing, the researchers
at the T&E Facility conclude that a device poll rate of
every two minutes is sufficient to produce data quality
that can reliably be processed by algorithms to evaluate
significant changes in water quality that is protective of
human health for most locations. However, the utilities
might want to evaluate other polling frequencies.
3.3.5	Data Marking
The SCADA system should be set up so that calibration
events, bad data, and instrument warnings (low re-
agent) are captured and "marked" within the SCADA
water quality database. This will permit the algorithms
analyzing the data in real-time to exclude these marked
data from further analysis, as any anomalies resulting
from these data are unlikely to be actionable.
3,3,8 Dal .mission arid Storage
Data transmission at the T&E Facility and nearby as-
sociated locations use a variety of communication me-
dia, including wired and wireless (radio and cellular)
technologies.
For large SCADA implementations, the majority of the
newer SCADA software manufacturers recommend the
use of a centralized (or distributed) database as the
back-end data repository. Generally, older data that are
not needed for any real-time analysis or computations
3-6

-------
are archived/stored in a database traditionally referred
to as the "historian." The real-time and the near real-
time values are usually run through an event-driven
calculation engine that either automatically performs a
task when predefined conditions are met or raises an
alarm for the operator to intervene or acknowledge the
SCADA value exception. Typically, the commercial
SCADA systems only allow the operator to define set
points (both at high and low levels) for each monitored
parameter to trigger an alarm.
The database storage and retrieval mechanism allows
for add-on algorithm-type programs, which can evalu-
ate the water quality data in real-time, compare it to the
baseline data in the database, and raise alerts and alarms
based on computed values and data trends.
The T&E Facility water quality data are stored at mul-
tiple locations. The iSIC datalogger and the iCliart
application store the data in proprietary formats. The
iChart stores the data in encrypted extensible Markup
Language (XML) and also pushes the data to a sepa-
rate Open Database Connectivity (ODBC)-compliant
MySQL database for storage and retrieval over the net-
work using commonly available tools. Figure 3.8 shows
the T&E Facility NexSens iSIC datalogger.
Figure 3.8 T&E Facility NexSens iSIC Datalogger
3.4 Best Practices for Instrument
Setup and Data Acquisition
Each utility , equipment developer/manufacturer should
review the site-specific requirements identified in Sec-
tion 3.1 of this document. For the utilities, if techni-
cally feasible, a single standard type of panel mountfor
housing all of the instrumentation/ SCADA is recom-
mended. In cases where a one-size-fits-all solution is not
possible (due to space constraints), no more than two or
three types of standard panel designs are rec-
ommended for field implementation. Each of the addi-
tional panel types could be designed to eliminate site-
specific length, width or depth constraint(s). The panel
standardization makes the fabrication and maintenance
easier. Within each type of panel, the following design
factors are of prime importance:
•	Each instrument should be both electrically and
hydraulically isolated (i.e., each instrument has
its own circuit breaker, separate water inlet with
a ball valve).
•	Flow monitoring devices should be non-
fouling (i.e., a rotameter without flow control
or float guide-wire, which tends to accumulate
biological growth and particle debris) and
instrument-specific in the correct flow range.
•	To conserve water, some of the instruments can
be designed to accept the discharge of another
non-reagent-based instrument.
•	In cases where the inlet water tends to be colder
than the environmental housing, degassing
(bubbles) can negatively impact the performance
of some instruments. The discharge side can be
pressurized for some instruments to minimize
the degassing effect. In other cases, a bubble
trap or a constant head mechanism could prove
effective.
•	The panels should be accessible and well-lit
(with an external light source)
•	The sites should have sufficient space to be
ergonomically efficient. This will prevent the
operator from taking shortcuts while performing
maintenance activities.
•	There should be a workbench, restrooms, a
place to store supplies and chemicals onsite to
maximize operator efficiency.
Equipment manufacturers should try to minimize the
footprint of their device and ensure that the housing is
NEMA-compliant. Wherever possible, the fluid lines
should include moisture sensors and be below the elec-
trical and data acquisition components to minimize
damage in case of a leak. In case of malfunctions, the
instruments should be robust and have an alarm func-
tion and self-restarting capability.
Data acquisition using field RTUs should be standard-
ized by the water utility so that the programming can
be simplified and replicated across sites. If data trans-
mission at a particular location is prone to interfer-
ences, programmatic error control techniques should
be applied to mitigate the errors. The data acquisition
and communication units should be UPS-backed and
equipped with lightning/surge protection. The

-------
manufacturers should ensure that the sensor is able to
communicate with the field RTUs, using the common
SCADA communication protocols.
Wireless data transmission should use secure protocols
where possible. Other SCADA related cyber-security
recommendations should be implemented whenever
possible. The data should be stored in databases with
ODBC connectivity and routinely backed up. The
ODBC connectivity enables the use of third-party data
analysis tools/applications or event detection algo-
rithms such as CANARY to interface with the data in
real-time. In addition, data should be marked for instru-
ment alarms, errors, and calibration events so that they
can be filtered out by the algorithms while analyzing the
data for anomalies.
~ Water Utilities and
Sensor Manufacturers
•	Trade-offs should be considered when locating
online sensors at the optimal TEVA-SPOT
identified location or at an alternate nearby
utility-owned location that meets site-specific
requirements identified in Section 3.1.
•	Pressure fluctuations, flow control, bubble
formation, and higher pH values might impact data
quality of many online sensors. Manufacturers
should provide robust non-fouling flow controls
with the sensor and eliminate the potential for
bubble formation in their equipment. Both utilities
and manufacturers should consider the addition of
pressure regulators and constant head devices prior
to sensor elements.
•	Manufacturers should add alarm output channels
to identify instrument-related problems such as
low reagents, instrument calibration drifts, etc.
Also, manufacturers should provide a variety of
interface options for SCADA communications
protocols. In addition, the instruments should be
designed to have a small footprint with built-in
self-restarting capability in case of malfunctions.
•	Utilities considering setting up a panel of
instruments should review the important panel
design factors identified in Section 3.4. Utilities
should also standardize the data acquisition
approach and follow the best practices identified in
Section 3.4.
•	Online TOC monitoring equipment employing
UV-persulfate methods are expensive and difficult
to maintain. Factory service contracts are
recommended. One of the TOC instruments, as
tested at the T&E Facility, requires a carbon-free
air source (i.e., a compressor/generator or nitrogen
tanks). Manufacturers should design and fabricate
simplified TOC monitoring devices.
3-8

-------
4.0	Testing Procedures
and Safety Precautions
Prior to evaluating various online sensors at the T&E
Facility, a Quality Assurance Project Plan (QAPP) and
a Health and Safety Plan (HASP) were developed. The
QAPP outlined the experimental and analytical objec-
tives. The HASP outlined the safety precautions nec-
essary for handling the selected contaminants. Some of
the critical elements obtained from the QAPP and
HASP that are applicable to the testing program (which
might benefit water utilities and manufacturers of sen-
sors considering such internal testing) are described in
the following sections.
4.1	Blank/Control Injection
Prior to injecting any contaminant into the DSS (Single
Pass or Loop No. 6), a control run was made to ensure
that there were no significant contributions to the base-
line water quality sensor response from the pumping ac-
tion of the injection apparatus, the DSS itself, or any-
associated instrumentation. The blank/control injection
matrix was designed to match the water matrix foreach
specific contaminant and was either Cincinnati tap water
or granular activated carbon-filtered Cincinnati tap wa-
ter. Tliis procedure ensures that significant changes are
not caused by either the absence of the selected disinfec-
tant (chlorine, chloramines) or naturally occurring mate-
rial in the injected water. Figure 4.1 shows the injection
apparatus used for testing on the Single PassDSS.
4.2	Contaminant Injection
Procedures
As discussed previously in Section 2.2, various contam-
inants, surrogates, carrier/growth media were selected
to represent a range of chemical, biological and radio-
logical agents that might be accidentally or intention-
ally introduced into a water distribution system. Tliis
section provides details on the injection specifics.
4.2.1 Concentration of the Injected
Contaminant
Several factors were considered while establishing the
injected concentration/dosage of the selected contami-
nants. These included the following: solubility of the
selected contaminant in water, results from the bench-
scale minimum dose sensor response study, and target
concentrations lower than the Immediately Dangerous
to Life or Health (IDLH) level for the selected contami-
nant. Mixing times and solubility observations were
made from beaker tests before performing the injection
event. The bench-scale minimum dose sensor response
study was performed to determine the "detection limit"
associated with a particular water quality monitoring
sensor for the selected contaminant. The purpose of the
bench-scale and the DSS Loop No. 6 and Single Pass
DSS studies was to determine if it was possible to inject
a contaminant at a concentration that was high enough
to cause health effects but could not be detected by the
array of sensors. Contaminant concentrations of 1 mg/L
were typical for both the DSS Loop No. 6 and the Sin-
gle Pass DSS injections. Tliis concentration was usu-
ally detectable by at least one water quality sensor; yet,
for a vast majority of the contaminants, it represented a
concentration well below the IDLH level. In compari-
son, other EPA-sponsored Environmental Technology
Verification (ETV) studies have been conducted using
contaminant injection concentrations of 10 mg/L, which
are generally well within the detection range of most
instruments and suitable for tracking the precision and
accuracy of the test instruments.
4.2.2	Duration of Injection
For the purposes of determining the minimum duration
necessary to detect a water quality baseline change, 2-
minute injections were performed. These short-
duration injections were successfully detected by the
sensors, even though some of the instrument sampling
durations exceed the 2-minute injection period. To eval-
uate total dosage necessary to cause potential harm to
humans, a longer 20-minute injection duration was se-
lected. Tliis duration also allowed for stable tests with a
maximum response time long enough to see a change in
baseline that could be detected by automated algo-
rithms. After injection, data from the various sensors
were monitored and recorded for at least 4 hours for the
DSS Loop No. 6 tests, and for at least 1 hour for the
Single Pass DSS tests. The algorithms and data analysis
techniques are discussed in Chapter 5.
4.2.3	Water Main Flow Rate and Injection Rate
The flow rate through DSS Loop No. 6 was typically
kept at 88 gpm, which also translates to a velocity of
Figure 4.11njection Apparatus for the Single Pass DSS
4-1

-------
1 foot per second (ft/sec) through the 6-inch pipe. This
velocity is commonly encountered in a distribution
system. Similarly, most of the testing that was con-
ducted on the Single Pass DSS was performed at a flow
rate of 22 gpm, which also yielded a velocity of ap-
proximately 1 ft/sec in the 3-inch pipe. All of the test-
ing was conducted under turbulent flow conditions. In
some of the tests, the flow rate of the Single Pass DSS
was varied to obtain the desired contaminant dilution
effect based on available stock concentrations. The fol-
lowing flow rates were used for testing in the Single
Pass DSS: 5, 10.7, 22, and 40 gpm. The typical injec-
tion rate was 0.5 liter per minute (0.13 gpm). When-
ever technically possible (based on solubility and any
other constraints discussed in Section 4.2.1), a batch of
10 liters of contaminant in water was injected over a20-
minute period.
4.2.4	Neat Compounds Versus Commercial
Off-the-Sht	ucts
For the purposes of evaluating how effective the online
sensor instrumentation is in detecting herbicides and
pesticides, manufacturers of some of the commercially
available off-the-shelf products were contacted to ob-
tain the neat (or pure) form of the active ingredient in
the product. During the bench-scale studies, it was
discovered that the inactive ingredient in commercial
off-the-shelf herbicides/pesticides might change water
quality in a detectable manner. For the DSS Loop No. 6
testing, Real Kill® (pesticide) and Roundup® (her-
bicide) were used to represent large groups of similar
commercially available compounds that are readily
available and accessible.
4.2.5	Wastewater and Ground Water Injections
Wastewater and ground water injections were con-
ducted to simulate natural or accidental contamination
events such as cross-connections and broken mains.
There have been cross-connection and back flow events
reported where contaminated wastewater has entered
the distribution system. Also, it is possible for mains
under the water table to seep or infiltrate ground water.
stir • i Analytical
nfirmation
In addition to the blank and control injections described
in the previous section, the DSS tests were repeated
both for DSS Loop No. 6 (in triplicate) and Single Pass
DSS (in duplicate) to ensure that the sensor responses
were valid and repeatable. The EPA ETV studies con-
ducted at the T&E Facility also evaluated inter-unit re-
producibility by deploying multiple units concurrently
for testing purposes. In addition, NHSRC's TTEP is
designed to provide reliable information regarding the
performance of homeland security related technolo-
gies.
ting Confirmation
The initial rounds of triplicate testing in DSS Loop No.
6 yielded consistent results based on the direction of
parameter-specific change. For the later rounds of
testing, only duplicate runs were performed. In case a
test run yielded inconsistent results due to equipment
malfunction, an additional test run was performed as
needed. Figure 4.2 shows a sample graph of triplicate
test results for glyphosate conducted on DSS Loop No.
6. Figure 4.3 shows a sample instrument response with
increasing injected contaminant (glyphosate) concen-
trations, conducted in the Single Pass DSS.
4,3,2 Analytical Confirmation
Bench-top analytical tests were performed to confirm
the water quality parameter readings of the online in-
strumentation. As shown in Figure 4.2, the grab samples
matched the results of the online instrumentation, with
the exception of ORP. The ORP readings are altered
when the sample is exposed to atmosphere during the
grab sampling event. In addition, for some of the con-
taminants (malathion and glyphosate), to ensure that the
injected contaminant was not absorbed/adsorbed into
the biofilm or pipe material, grab sampling from the
sample taps of online sensor instrumentation was
performed. These grab samples were submitted to an
outside laboratory to perform analytical confirmation.
Although the analytical results confirmed the presence
of these contaminants, the concentration levels were
found to vary. The varied results were attributed to the
following: 1) relatively poor analytical methods, which
were chosen by the outside laboratory; 2) injected com-
pounds interacted with free chlorine in the test water,
which might have resulted in the generation of other by-
products that were not measured; and 3) possible
adsorption/absorption to the biofilm. However, as the
changes in the measured water quality parameters were
consistent with the injected level of contaminants (as
shown in Figure 4.3), the analytical confirmation was
abandoned to keep up with the rapid pace of testing and
to reduce project costs.
¦ • • "¦ seli
tablishment
Between the test runs, DSS Loop No. 6 was continuous-
ly operated to flush the system. In addition, prior to the
test runs, DSS Loop No. 6 was sufficiently flushed so
that the water quality parameters (especially turbidity
and temperature) equilibrated and remained stable dur-
ing the test. This parameter stability was confirmed by
4-2

-------
1.30
1.10
0.90 -
-	- Free Chlorine (Grab) DPD - colorimetric Run 1
Free Chlorine (OnLine) DPD - colorimetric Run 1
-	-¦ - Free Chlorine (Grab) DPD - colorimetric Run 2
Free Chlorine (OnLine) DPD - colorimetric Run 2
-	-a - Free Chlorine (Grab) DPD - colorimetric Run 3
Free Chlorine (OnLine) DPD - colorimetric Run 3
T-PraGrab
INJECTION at INJECTION + 4
10:35
0.30
0.10 -


-0.10 -
4- ~	f 			
3 min. T-60min. T-2Hr.	T-3Hr.
I i i i i I i i i i I i i i i I i i i i I i i i i I i i i
-2.00 -1.50 -1.00 -0.50 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 5.50 6.00
TIME (hours)
Figure 4.2 Glyphosate Triplicate Injection Run Results
evaluating the online instrument baseline data to ensure
that the instrument readings were within the "normal"
range of operation. If the instrument readings deviated
from normal conditions (based on operatorexperience),
the instrument was recalibrated to ensure accuracy and
repeatability.
4.5 Health and Safety Precautions
Standard laboratory personal protective equipment such
the biological activity of real pathogens (Edberg et al.,
2000; Lytle and Rice, 2002; Rice et al.. 2005;
Sivaganesan et al., 2006). However, as part of good
laboratory practice, standard Biosafety Level 1
measures were implemented. Personnel had to change
gloves after coming in contact with items that might
carry biological contaminants. Gloves could not be
placed near the face after exposure to biological
contaminants. Any positive reference materials were
as laboratory coats, gloves, 1.0
safety glasses, and safety shoes
were required during the
experiments. For chemical
contaminants, additional test/ _
contaminant-specific protec- ^
tive gear might be required in —0.6
accordance with the Material
Safety Data Sheet or
contaminant-specific HASP.
0.9
0.8
0.7
0.5
0.4
For biological contaminants,
depending upon the contami-
nant, the biohazards and the
risk of infection should be
minimized. All of the bio-
logical contaminants used for
the testing at the T&E Facility
were non-pathogenic. The
surrogates closely represent
' 0.3
0.2
0.1
0.0
¦Test 1 - 0.4 mg/L
—Test 1 Dup - 0.4 mg/L
•Test 2 -1.5 mg/L
—Test 2 Dup -1.5 mg/L
¦Test 3 - 3.0 mg/L
—Test 3 Dup - 3.0 mg/L
-0.40 -0.20
0.00
0.20
0.60
0.40
Time (hours)
Figure 4.3 Glyphosate Injections at Varying Concentrations
0.80
1.00
1.20
4-3

-------
handled with gloves in an appropriate laboratory hood.
Equipment and supplies that came into contact with
suspected biohazard materials had to be sterilized prior
to disposal or reuse. The contaminated equipment/sup-
plies were sterilized by either standard autoclaving or
wiping with 0.02% bleach solution, depending on the
extent of contamination and the type of material consti-
tuting the equipment/supplies. Waste samples had to be
autoclaved prior to disposal. Special precautions, such
as donning heat-resistant gloves, were required for au-
toclaving activities.
)isposal of Contaminated
Water Frc	' ••
The EPA T&E Facility operates under a discharge per-
mit from the Metropolitan Sewer District of Greater
Cincinnati (MSD). This permit authorizes the direct
discharge of specified lev- els of contaminants to the
local, publicly owned waste- water treatment facility.
The aforementioned sensor technology testing at the
T&E Facility was conducted so that all of the test water
could be directly discharged to the sewer system.
Utilities and sensor manufacturers considering such
testing at their facilities should evalu- ate their
contaminant-specific discharge limits prior to initiating
a testing program. If the discharge limits for the selected
contaminants are too low, it might be nec- essary to
make alternate arrangements for disposal of the test
water (e.g., local treatment before discharge, offsite
shipment) or modification of the permit.
I Practices for Testii• • i
ecautions
In order to avoid positive bias from any of the injection
equipment/sampling or monitoring equipment, all test-
ing components should be disinfected and calibrated so
that an accurate baseline is established prior to testing.
Test contaminants (or surrogates) should be selected so
that they represent a broad class of potential threat
agents. The target concentrations should be at or below
the levels where human health can be adversely
affected, including considerations for sensitive/suscep-
tible subpopulations. The selected duration of injection
should be optimized to minimize the use of contami-
nants for both cost control and waste discharge consid-
erations. Whenever possible, for instruments measuring
physical and chemical parameters, bench-scale testing
is recommended prior to pilot-scale or full-scale test-
ing to determine the levels at which the selected instru-
ments can detect the selected contaminants.
A QAPP can help establish a detailed experimental plan
that identifies specific types and quantities of the
4-4
contaminant(s) involved during the testing. The QAPP

-------
~ Water Utilities and
Sensor Manufacturers
•	Control and blank injections should be performed
to ensure that the water quality sensors are not
impacted by the injection apparatus.
•	The testing at the T&E Facility revealed that the
commercially available online water quality
sensor equipment can generate reproducible data
responses with duplicate contaminant injections
and also at varying concentration levels.
•	Stable or predictable baseline water quality levels
are needed to obtain useful data from the online
water quality sensors. The variation in background
values should be considered when locating online
sensors. Also, baseline data should be collected for
a sufficient time period to capture normal water
quality variability for each location.
•	Without varying water demands, contaminants were
found to travel as a slug or in plug flow within the
Single Pass system. Recirculating DSS Loop No. 6
experienced fully mixed conditions within several
minutes.
•	Utilities and manufacturers considering inhouse
testing should select contaminants that represent
a broad class of potential threat agents, develop
a detailed experimental plan/QAPP/HASP, and
evaluate potential disposal options prior to
conducting any test runs.
medical surveillance based on the contaminant and
concentration used.
can also help to define the overall
experimental objec-tives, standardize the
experimental procedures, estab-lish protocols
for instrument calibration (prior to test- ing),
and establish data quality that can be
technically defensible when reviewed.
Prior to any testing, a HASP should be
developed, re-viewed, and approved by
appropriately trained person-nel so that the
tests can be performed safely. The HASP will
identify minimum job hazards and controls,
sample handling techniques, personal
protective equipment, work practices and
engineering controls, and spill/ emergency
procedures. This documentation also helps in
determining if the test water can be directly
and safe-ly discharged (without treatment)
based on the facility's existing discharge
permit. Otherwise, it will be neces-sary to
make arrangements for appropriate waste-han-
dling procedures. In addition (if needed),
HASPs can identify appropriate safety training
programs, personal monitoring needs, and

-------

-------
5.0 Data Analysis
Water quality sensor response data generated at the T&E
Facility enabled EPA to construct a qualitative sensor
parameter response matrix for the contaminants tested.
Analysis of this data required that the concept of a
significant change from baseline water quality condi-
tions be defined. A significant change is a large enough
deviation from normal water quality parameters that
could be used to trigger an alarm to be transmitted to the
data user (i.e., a drinking water utility). The criteria for
determining a significant change in the sensor pa-
rameter was subjective at the early stages of the investi-
gations. Initially, significant change determination was
based on visual qualitative inspection (e.g., drop infree
chlorine and increase in measured TOC value) of the
plotted sensor responses over the course of the contami-
nant injection test runs. Later on the significant change
determination was based on a quantitative approach: the
maximum change observed within a short time period
of contact (defined as 15 minutes) of the sensor with the
contaminant was divided by the baseline value of the
parameter to compute a percent change or deviation
(Hall et al., 2007). Although this method was simple and
straightforward, it omitted several critical factors such
as slow sensor response times and noisy background
data. Therefore, the same response data were also
evaluated using basic statistical methods that are
described in Section 5.1 of this report. As demonstrated
later in this chapter, the significant change threshold is
dependent upon the variability of the baseline water
quality data at a particular monitoring site. Based on the
testing conducted at the T&E Facility, EPA developed
and utilized the significant change thresholds presented
in Table 5.1 for evaluating contaminant injection sen-
sor response data. These threshold values are not en-
tirely based on measured or statistically derivedvalues;
Table 5.1 Parameter-Specific Significant Change
Thresholds
Water Quality Parameter
Deviation from
Baseline Classified as
"Significant Change"
Temperature
± 0.15°C (~0.27°F)
Specific Conductance
> 5% increase
Dissolved Oxygen
± 0.2 milligrams per liter
Oxygen Reduction Potential
± 20 millivolts
Nitrate
± 10%
Chloride
± 15%
Ammonia
± 20%
Turbidity
> 200% increase
Free Chlorine
> 5% decrease
Total Organic Carbon
> 0.1 mg/l increase
the operator's understanding of the variability of water
quality at each location should also be taken into ac-
count while developing these parameter-specific signif-
icant change thresholds. For example, the baseline wa-
ter quality parameters observed at the T&E Facility are
stable with little variance. Therefore, using these values
at a location where the water quality baseline is highly
variable can lead to triggering of an excessive number
of false positive alarms. Keeping these observations in
mind, it is recommended that the end-users employing
this methodology should develop their own site-specific
significant change thresholds for evaluating real-time
water quality data.
During the course of this research EPA was aware that
the significant change data analysis approach, which
used visual inspection of time series data and percent
change from baseline, could lead to variable results
caused by site-specific water quality differences and
analyst bias. Therefore, a more sophisticated quantita-
tive approach was undertaken: individual sensor re-
sponses were analyzed by computing absolute change,
percent change, and S/N ratio. Absolute change and
percent change analysis employ the same mathematical
techniques described previously in the development of
significant change thresholds. The S/N ratio analysis is
designed to filter the level of "background noise" caused
by frequent fluctuations in the baseline data. The S/N
ratio is defined as the ratio of a measured value to the
background noise. A low S/N ratio indicates that the
change in measured value of the parameter might be
caused by the background noise (representing routine
fluctuations in measured baseline) rather than resulting
from a real change in water quality due to the presence
of contaminants (Szabo et al., 2006, 2008a, and 2008b).
Furthermore, EPA's field installation experience has in-
dicated that the baseline water quality at certain loca-
tions (that are immediately influenced by utility opera-
tions) can change significantly over a short time period.
For example, monitored parameters might fluctuate dra-
matically with changes in the operation of tanks, pumps,
and valves. The monitored parameters are also affected
by daily and seasonal changes in the source and finished
water quality, as well as fluctuations in demand. EPA
collaborated with SNL to build an automated algorith-
mic data analysis tool that combines and enhances some
of the previously mentioned qualitative and quantitative
approaches to distinguish between normal variations in
water quality and changes in water quality triggered by
the presence of contaminants (McKenna, et al., 2006).
These types of tools are often referred to as event de-
tection algorithms, which can read SCADA data (water
quality signals, operations data, etc.), perform analysis
in near real-time, and return a 0/1 result (indicating pres-
ence or absence of an alarm). All of these approaches are
discussed further in this chapter.
5-1

-------
5.1 Non-Algorithmic Sensor
Response Evaluation
The qualitative and quantitative approaches employed
to evaluate the data generated at the T&E Facility and
Edgewood Chemical and Biological Center (ECBC) are
discussed in the following subsections.
5.1.1 Single Pass DSS Data Analysis
The data tabulated and reported in this section was
generated by injecting selected contaminants (as listed
in Tables 5.2 through 5.7) into the Single Pass DSS
using chlorinated tap water available at the T&E
Facility (as supplied by GCWW). As mentioned pre-
viously, the background values are generally stable at
this location. The range of background values for the
routinely measured water quality parameters are as
follows: free chlorine - 0.8 to 1.1 mg/L, specific
conductance - 300 to 600 |iS/cm. ORP - 500 to 700
mV, pH - 8.5 to 8.8 standard units, turbidity < 0.1NTU,
and TOC - 0.3 to 1.3 mg/L. Table 5.2 shows the
percent change of several water quality parameters for
various injected contaminants in the Single Pass DSS
(described in Section 2.1.2). The qualitative response
information in Table 5.2 is color-coded to show
changes that exceed 10% from the baseline value.
Percent change is calculated by first calculating the
difference between the baseline mean over one hour
before injection, termed absolute change (AC):
AC Speak - S
'baseline
Where, Speak = the peak sensor value between when the
contaminant is in contact with the sensor until 15
minutes after and Sbaseline = the mean baseline value for
one hour immediately preceding the contaminant
injection test.
Percent change is then calculated as follows:
AC
% Change = 	
^baseline
Table 5.2 allows for easy identification of online water
quality monitoring parameters that would potentially
respond to the specified injected contaminants. An
examination of Table 5.2 reveals that free/total chlorine
and turbidity provide a significant change signal for a
Table 5.2 Percent Change in Sensor Parameter Response to Injected Chemical Contaminants in
Chlorinated Water - Single Pass DSS
Chemical
Contaminants
Initial In-Pipe
Concentration
(mg/L)
Free Chlorine
Total Chlorine
Chloride (CI )
Specific
Conductance
Dissolved
Oxygen (DO)
Oxidation
Reduction
Potential (ORP)
IE
Q.
Turbidity
Aldicarb
0.2
-9.0%
-8.8%
2.3%
0.2%
0.0%
-0.3%
0.3%
188.5%

1.1
-43.6%
-43.5%
-2.8%
0.0%
-1.2%
-0.6%
0.0%
487.0%

2.2
-87.6%
-82.9%
2.7%
0.0%
0.1%
-1.2%
-0.1%
-100.0%
Glyphosate
0.4
-34.4%
-17.1%
3.0%
-0.0%
1.2%
0.1%
-0.1%
1003.5%
1.5
-77.9%
-39.3%
2.0%
0.0%
1.3%
-0.9%
-1.4%
329.2%

3.0
-95.2%
-52.4%
2.6%
-0.0%
0.6%
-2.6%
-4.6%
-100.0%
Colchicine
0.4
-2.7%
0.3%
1.6%
0.1%
3.2%
0.2%
0.3%
593.3%

1.8
-4.3%
-3.9%
0.1%
0.4%
1.3%
-0.4%
0.2%
157.0%

3.6
-5.6%
-4.2%
-0.6%
0.0%
-5.0%
-0.7%
0.2%
111.1%
Dicamba
0.8
-2.8%
0.0%
-1.4%
-0.0%
-0.6%
0.6%
-0.3%
0.0%

1.3
-3.1%
-1.7%
-1.3%
-0.3%
-0.5%
0.6%
-0.5%
-6.9%


-1.9%
0.5%
-0.6%
0.2%
-0.6%
0.5%
-0.7%
-7.1%
Dimethyl Sulfoxide
0.6
-11.8%
-9.5%
-1.5%
0.1%
-0.4%
0.2%
0.2%
4.4%

-29.2%
-26.0%
-0.1%
0.1%
-0.9%
-0.3%
0.3%
1.0%

4.0
-46.9%
-42.6%
0.1%
0.1%
-1.3%
-0.3%
0.1%
2.0%
Lead Nitrate
0.6
-3.9%
1.3%
-5.4%
0.0%
2.3%
0.2%
-0.2%
400.0%

0.7
-2.9%
-2.0%
-0.8%
-0.1%
-3.4%
-0.1%
-0.1%
137.2%

1.4
-0.3%
-2.0%
-0.7%
-0.1%
-0.9%
-0.9%
-0.1%
538.4%
Mercuric Chloride
0.4
-2.5%
-1.2%
-1.6%
0.0%
0.0%
0.2%
-0.1%
26.3%

1.1
-1.0%
-1.5%
-1.7%
0.3%
-7.3%
0.5%
-0.7%
46.9%


-1.1%
0.1%
-2.3%
0.4%
-8.0%
0.8%
-1.2%
-1.4%
Nicotine
0.4
-14.3%
-7.9%
2.4%
0.2%
1.4%
0.2%
0.3%
1200.0%

1.9
-49.3%
-28.6%
0.5%
0.0%
1.5%
-2.4%
0.7%
82.0%

3.8
-84.7%
-47.8%
2.4%
0.1%
-1.5%
-4.8%
1.0%
143.4%
Potassium
0.6
-3.7%
0.8%
11.9%
-0.1%
-0.4%
0.5%
0.1%
287.7%
Ferricyanide
1.6
-5.7%
8.0%
14.5%
0.5%
-0.6%
-0.3%
-0.2%
503.2%
3.2
-4.1%
21.2%
26.3%
0.8%
-0.1%
-0.1%
-0.1%
50.0%
Sodium Thiosulfate
0.2
-15.1%
-13.3%
2.8%
0.4%
-0.1%
0.1%
0.3%
883.3%
(Anhydrous)
1.3
-75.5%
-72.1%
2.2%
0.6%
-1.2%
-1.1%
-0.1%
385.4%

-98.8%
-95.1%
4.9%
0.8%
-2.6%
-5.3%
0.1%
645.2%
Sucrose
0.6
-2.9%
-0.7%
-1.0%
0.1%
1.4%
3.7%
0.0%
167.8%

1.8
-3.2%
-0.4%
-0.9%
0.0%
-5.7%
1.1%
-0.1%
449.1%

3.6
-2.6%
-0.1%
-0.7%
0.0%
-3.8%
0.4%
-0.1%
433.3%
Control Blank
0
2.4%
1.6%
5.5%
0.5%
0.5%
1.8%
0.2%
52.5%

0
1.0%
0.2%
0.6%
1.1%
1.0%
2.0%
0.1%
71.8%

AVGa
1.7%
0.9%
3.1%
0.8%
0.8%
1.9%
0.2%
62.1%
5-2

-------
Table 5.2 Percent Change in Sensor Parameter Response to Injected Chemical Contaminants in
Chlorinated Water - Single Pass DSS (continued)
Chemical
Contaminants
Initial In-Pipe
Concentration
(mg/L)
Free Chlorine
Total Chlorine
Chloride (CI )
Specific
Conductance
Dissolved
Oxygen (DO)
Oxidation
Reduction
Potential (0RP)
IE
Q.
Turbidity
Antifreeze
1
0.3%
N/A
N/A
0.2%
N/A
0.0%
0.1%
5.8%

10
4.7%
N/A
N/A
0.8%
N/A
0.1%
0.2%
9.0%
Bleach
5
334.8%
N/A
N/A
9.7%
N/A
2.8%
1.9%
6.6%
De-icer
1
1.5%
N/A
N/A
0.5%
N/A
0.1%
0.3%
3.9%

10
1.5%
N/A
N/A
0.3%
N/A
0.1%
0.1%
90.8%
Diesel Fuel
1
1.0%
N/A
N/A
0.2%
N/A
0.0%
0.2%
66.7%

5
0.1%
N/A
N/A
0.0%
N/A
0.1%
0.2%
184.1%
Dispersant
1
11.2%
N/A
N/A
0.1%
N/A
0.5%
0.1%
237.4%
10
87.5%
N/A
N/A
0.0%
N/A
1.1%
0.5%
2122.2%
Ethylene Glycol
1
0.3%
N/A
N/A
0.1%
N/A
0.1%
0.1%
3.5%
10
1.4%
N/A
N/A
0.2%
N/A
0.0%
0.1%
8.8%
Pepsin
1
12.9%
N/A
N/A
0.3%
N/A
0.2%
0.4%
36.0%
10
98.4%
N/A
N/A
0.6%
N/A
1.4%
0.7%
198.4%
Basagran
1
1.3%
N/A
N/A
0.0%
N/A
0.1%
0.0%
4.5%
10
2.6%
N/A
N/A
0.7%
N/A
0.1%
0.0%
5.3%
~ indicates that the percent change was >10% of the baseline within 15 minutes.
a Average of two control blank runs. Data presented in this table was not corrected to accommodate for the control blank response.
N/A - Results not available.
majority of the contaminants tested.
Table 5.3 shows further manipulation of the same da-
taset (presented previously in Table 5.2). The data pre-
sented in Table 5.3 have been normalized and adjusted
to correct for the S/N ratio for each monitored
parameter. The S/N ratio was calculated as follows:
AC
S/N= 	
^baseline
where CTbaseline = the standard deviation in the baseline
for one hour prior to injection.
In Table 5.3, further manipulation of the original data-
set (see Table 5.2), reveals that the turbidity parameter
is no longer a good indicator of significant change, as
shown in Table 5.2, because the baseline signal is too
noisy. The data in Table 5.3 also reveal that there are
other water quality parameters that might be good at
detecting contamination, but do not appear so at first
glance in Table 5.2. For example, pH andORP changed
in response to five and nine of the 19 injected
contaminants, respectively. Both parameters have
stable baselines (low standard deviation) and produced
small changes when contaminants were injected.
However, because the baselines were so stable, the
normalized and S/N-adjusted change was relatively
large. Tables 5.2 and 5.3 provide examples of how
non-algorithmic analyses of water quality data can be
useful. The tables also identify some pitfalls to be
avoided when interpreting online data. Additional data
on the response of water quality sensors to biological
suspensions and culture broth are presented in Tables
5.4 through 5.7.
5.1.2 Recirculating DSS Loop No. 6 Data
Analysis
The data tabulated and reported in this section were
generated by injecting selected contaminants (as listed in
the individual tables) into recirculating DSS Loop No. 6
(described in Section 2.1.1) with chloraminated water
prepared at the T&E Facility. The preparation
methodology for the chloraminated water was described
previously in Section 2.2.1. For performing these tests, the
background chloramine level was set at 2 mg/L (measured
as total chlorine). The background values of other
measured water quality parameters are the same as
mentioned previously in Section 5.1.1. Table 5.8 shows the
quantitative sensor responses to contaminants injected in
DSS Loop No. 6 as absolute change, percent change, and
S/N ratio. Also, Table 5.8 shows that the TOC parameter
responded to each contaminant except sodium arsenite,
which was expected since sodium arsenite is not an organic
(carbon-containing) compound. The TOC sensor responses
are comparable to those observed during the same
injections in the chlorinated GCWW water. Responses
were also similar for other sensors, except total chlorine.
Total chlorine, as a measure of chloramines, showed little
change for the contaminants tested, except for decreases of
0.22 mg/L (9.5 percent decrease, S/N: 60) and 0.31 mg/L
(14.6 per- cent decrease, S/N: 48.6) forphorate and sodium
arsenite, respectively.
However, since chloramines (mostly monochloramine)
react more slowly with the injected organic contaminants
than with free chlorine, the changes reported in Table 5.8
occurred over a period of four hours. In comparison, the
free chlorine values changed almost instantaneously when
the same contaminants were introduced into DSS
5-3

-------
Table 5.3 Normalized, Signal-to-Noise Corrected Sensor Parameter Response to Injected Chemical Contaminants in
Chlorinated Water - Single Pass DSS

Initial In-Pipe
tu
_e
-£=
tu
_e
-£=
o
O
tu

-------
water quality, while at the same time limiting the
number of false alarms that occur. CANARY is trained
on "normal" baseline water quality data by the user
during the setup, and the configuration parameters are
selected to accommodate the normal site-specific
variability of water quality parameters. Therefore,
these configuration parameters could vary from one
utility to the next and might even vary across
monitoring locations within a single utility. CANARY
can be set up to receive data from a SCADA database,
and return alarms to the SCADA system. In addition, it
can be run "offline" on historical data to help set the
configuration parameters (or train the algorithm) to
provide the desired balance between event detection
sensitivity and false alarm rates.
CANARY'S open source code is designed to be custom-
izable, allowing outside researchers to develop new al-
gorithms that can be added to CANARY. In the current
version (Version 4.2), CANARY has three change
detection algorithms: time series increments, a linear
filter, and a multivariate nearest-neighbor algorithm.
These algorithms identify a background "water quality
signature" for each water quality sensor and compare
each new water quality measurement to the
background to determine if the new measurement is an
outlier (anomalous) or not. The definition of the water
quality back- ground is updated continuously as new
data become available. A binomial event discriminator
(BED) examines multiple outliers within a prescribed
time window to determine the onset of either an
anomalous event or a change in the water quality
baseline. Figure 5.1 shows the schematic operation of
CANARY software.
In addition to the CANARY software, EPA has also
tested the commercially available Hach Event
Monitor™ Trigger System. Hach's patented
technology utilizes the Hach Event Monitor™ Trigger
System to analyze five commonly measured water
quality parameters monitored from the Hach WDMP
(chlorine, pH, turbidity, conductivity) and the Hach
astroTOC™ UV process TOC analyzer to estimate a
water distribution system's operating baseline (i.e.,
water quality under normal operating conditions).
Thereafter, every minute, the Hach Event Monitor™
Trigger System analyzes the sensor data and computes
a trigger signal, which indicates the level of deviation
from the water quality baseline. If significant
deviations occur, the trigger signal sends alarms to the
operators in real-time. Once a deviation is detected,
the Hach Event Monitor™ Trigger System signals the
(optional) automatic water sampler to capture a water
sample at the designated monitoring location. The
system subsequently compares the computed
algorithmic values to the "Agent Library" and "Plant
~ Water Utilities and
Sensor Manufacturers
•	Online monitoring data should be evaluated both
qualitatively and quantitatively to identify which
parameters provide significant change signals and
have relatively stable baselines with low S/N ratio.
These criteria should be factored into the selected
algorithmic approach of data analysis.
•	Free chlorine and TOC were found to be the most
responsive trigger parameters in chlorinated
systems. Total chlorine was not an effective trigger
parameter in chloraminated systems. TOC or TOC
surrogate monitoring should be considered for both
chlorinated and chloraminated systems.
•	To capture and evaluate sensor responses in real-
time, SCADA equipment and algorithmic analysis
are highly recommended. The SCADA database
should be designed so that it can be easily
interfaced with one or more automated algorithms
for real-time analysis of data.
•	Algorithms should be designed to learn or predict
baseline values of parameters when monitoring
location(s) with relatively unstable baseline
conditions. Known routine system events (such as
valve closures or tank fill and discharge cycles)
need to be incorporated into the algorithmic
evaluation(s) to reduce false positive events.
•	For chlorinated systems, the algorithms and
sensors should be designed to co-relate free
chlorine and TOC data. Manufacturers should
consider providing alarming or algorithm software
with the online sensor equipment that is capable of
identifying bad data and other instrument
operational problems. This will prevent bad data
from being analyzed by the algorithms, resulting in
fewer false positive events.
Read Configuration and Setup
+
Read
New Water Quality Data
Wait For	Process
Next Time Step	Event Detection Algorithms
+
Report
Probable Events
When Finished
Print Results Files and Exit
Figure 5.1 CANARY Operation Schematic

-------
Library" to classify the deviation. The subscription-
based "Agent Library" (Hach GuardianBlue™ Early
Warning System) contains "fingerprints" for a wide
variety of threat contaminants, ranging fromV-series
nerve agent (VX) and ricin to arsenic and herbicides.
The site-specific operator-developed "Plant Library"
contains "fingerprints" of operational and naturally
occurring events specific to each water distribution
system. The plant library can be used to detect and
classify real-world events such as water main breaks,
switching water sources, and caustic overfeeds.
EPA is also evaluating other commercial event detec-
tion technologies such as the Frontier Technology, Inc.
(FTI) Event Detection Software (EDS) tool calledH20
Sentinel™ for contaminant detection. FTI has devel-
oped a proprietary software to monitor a set of standard
water quality parameters measured by sensor stations
placed within a utility's water distribution system and
detect anomalous events that might be indicative of
possible contamination incidents.
EPA plans to continue evaluating other commercially
available event detection algorithms as they become
available (Einfeld et al., 2007; Umberg et al., 2009).
Because true contamination events are rare, the perfor-
mance of event detection systems is difficult to evalu-
ate. It is tempting to set the sensitivity of the algorithm
at a low level so that few alarms are generated, since
true contamination events are costly to investigate.
However, high sensitivity algorithms can result in the
generation of many alarms, which can result in "alarm
fatigue." EPA continues to evaluate the alarm predic-
tion accuracy of these algorithms by simulating con-
tamination events. EPA is considering the use of modi-
fied receiver operating characteristic (ROC) curves, a
data classification methodology that can plot the frac-
tion of true positive alarms versus the fraction of false
positive alarms generated by the individual algorithm.
Modified ROC curves can help determine the efficiency
of these algorithmic approaches.
: ' ' ces
When the contaminant (or surrogate) injection tests
were performed at the T&E Facility, the algorithmic
approaches of data evaluation were not available, with
the exception of the previously-mentioned ETV study.
The qualitative approach, although robust, is not a vi-
able technique for real-time event detection. There are
normal/natural changes in water quality that could
mimic some of the qualitative changes shown in the
testing. For example, at a monitoring station near a stor-
age tank or reservoir, the chlorine levels might change
dramatically depending upon the source of water and
tank operation. Similarly, quantitative changes have a
drawback: in real life, the concentration of the injected
contaminants is unknown, and the resulting amount of
change does not necessarily correlate with prescribed
quantitative values. Therefore, an algorithmic approach
is the preferred approach for event detection. However,
a utility with existing SCADA systems (which general-
ly allow for high-low alarm set points) in the process of
deploying water quality monitoring stations can utilize
significant threshold and other non-algorithmic meth-
odologies described in this chapter to alert the local op-
erator for further investigation. These types of prelimi-
nary data evaluations will assist in establishing and fine-
tuning parameter-specific change thresholds and time-
windows for the algorithmic approaches. The
evaluations of existing algorithms have so far
demonstrated only limited success. Also, in an
algorithmic approach, there is a need to optimize the
sensitivity of the algorithm so that the false positives are
minimized while retaining the algorithm's ability to
detect contamination events. EPA's research for fine
tuning individual algorithms and the search for new
approaches are ongoing.
R.Q
%J \J

-------

-------
6.0 Operation,
Maintenance and
Calibration of Online
Instrumentation
The capital costs of the equipment tested at the EPA
T&E Facility are readily available from the manufac-
turers and are subject to change. However, the O&M
costs are not as readily obtainable or well-defined. The
experience gained at the T&E Facility indicates that, for
most online sensors, O&M costs will quickly exceed
capital costs. To keep the costs under control for the
longer term, manufacturer recommendations should be
followed when performing O&M activities. Also,
maintenance requirements vary significantly, depending
upon the parameter and the device. The on-line
instrumentation evaluated at the T&E Facility was
calibrated and maintained as needed for testing efforts
described in the previous chapters. For year-round op-
eration, maintaining a tight maintenance schedule is
necessary to obtain optimal instrument performance.
The scheduled maintenance activity, as well as cost of
consumable(s), depends on the individual sensor.
This chapter describes general O&M activities and as-
sociated costs for the instruments evaluated at the T&E
Facility. For this purpose, EPA tracked the labor and
consumable costs required to operate and maintain the
tested sensor equipment. The O&M costs presented
here do not include travel time to service the individual
monitoring locations. The travel costs will vary sig-
nificantly, depending upon where service personnel are
operating relative to the geographic distribution of the
monitoring sites. Furthermore, the total labor cost will
depend on operator skill and training, sensor complex-
ity, service contract terms, and the number of sensor
stations deployed. Based on discussions with utilities
participating in the WSi, the optimal service require-
ment for a sensor station deployed to the field was de-
termined to be one O&M site visit per month, with each
visit taking less than four hours.
/laintenance
Labor Costs
The labor hours expended for O&M vary significantly,
depending upon the type of instrument used (or parame-
ter that is monitored). As mentioned previously, the goal
of the WSi's pilot implementation in Cincinnati was to
achieve a service level of four labor hours per monitor-
ing station per month. It was well understood that, dur-
ing the initial phase of installation and shake-down, the
labor costs were going to be much higher. Also, depend-
ing upon the size and complexity of an implementation,
the shake-down period can range from a month to ayear.
The labor hours per instrument also vary significantly.
Of the conventional instruments (TOC, chlorine,
conductivity, pH/ORP and Turbidity), TOC instruments
were by far the most labor intensive from an O&M per-
spective. They also required the highest technician skill
level to operate and maintain. However, the TOC instru-
ments are being continuously redesigned and the labor
level required is expected to be lower in the coming
years. Overall, the data collected from the WSi initiative
(for a 10-month period between January and September
2008) showed that approximately 1.5 person(s) working
on a full-time basis were needed to operate and maintain
17 monitoring stations. Between 60% and 80% of this
labor cost was associated with the O&M for the TOC in-
struments. Also, as indicated previously, the labor hour
estimate does not include travel time to the monitoring
stations, which could vary significantly depending upon
the geographic configuration of the monitoring network.
Experience indicates that labor estimates per site can
vary widely because there might be some sites with ad-
verse water quality (or other site-specific anomaly) that
can cause numerous O&M problems. The best current
estimate for labor hours per site is 1.5 days per month on
average, which is three times the goal of the WSi.
Additional labor costs will be incurred for data analysis
and event detection efforts. Because it is too labor
intensive to have the operator monitor the data on a
continuous basis, an automated event detection or alarm
system is necessary. It is expected that the existing
SCADA operator at a water utility can be assigned the
additional duty of checking the operational status of the
data collection system during every shift to ensure that
data is being collected and analyzed by automated tools.
To ensure everything is operating normally, the utility
should plan to assign a person to skim through the
historical data and monitor the alarm software on adaily
basis. This task is expected to add approximately 30 to
60 minutes per day or per shift depending upon how the
task is assigned. The reviewer should be a staff member
who already understands the operations and the
monitored water quality data and can make decisions on
O&M and alarm response needs as warranted by the
quick review of the data and alarm history.
enl-Specific
Maintenan	sumable
Costs
As mentioned previously, the maintenance requirements
vary significantly, depending upon the parameter and
the device. Based on the experience gained at the T&E
Facility, EPA identified TOC, free chlorine, conductiv-
ity, pH/ORP, turbidity and temperature as the key online
6-1

-------
6-2
monitoring parameters. These parameters are listed in
the order of importance from an event detection per-
spective. The following sections (Section 6.3 through
6.7) describe these parameters (except temperature) and
the instrument-specific O&M activity for the equipment
evaluated at the T&E Facility. The temperature probes
are inexpensive and very robust with almost no main-
tenance requirements and are, therefore, not discussed
in this report. The prices for consumables for the instal-
ments mentioned in this chapter are based on pricing
information obtained during the years 2007 and 2008.
Pricing information for sensors tested between 2009-
2019 correspond to the time period when they were
tested.
ital ' ' ' Carbon
Instrumentation
TOC instrumentation responded to a wide range of
contaminants tested at the T&E Facility. Especially in
chloraminated waters, TOC instrumentation was more
responsive to contaminants than the instrumentation
used for measuring total chlorine/monochloramines. A
skilled technician was required to reliably operate and
maintain the TOC instrumentation on a continuous ba-
sis, making it the most expensive and time-consuming
equipment to operate. It is highly recommended that the
person responsible for TOC instrumentation O&M is-
sues obtain manufacturer-provided training, or the wa-
ter utility should consider purchasing a manufacturer
maintenance contract. EPA also evaluated an optical in-
strument (Carbo::lyser™) that estimates TOC levels by
measuring UV-Vis spectra between 200 and 750 nano-
meter (nm) wavelengths.
6.3.1 Hach a	cess Total
Organic Carbon Analyzer
This instrument requires monthly replenishment of re-
agents, quarterly scheduled maintenance, and approxi-
mately $4,000 per year for consumables. The following
routine maintenance activities need to be performed:
•	Replace reagent/acid (monthly), reagent/oxidizer
(quarterly)
•	Change the pump tubing (quarterly)
•	Replace the semi-permeable sparger membrane
(quarterly)
•	Replace the hydrophobic filter (quarterly)
•	Calibrate the infra-red detector with carbon
dioxide (quarterly)
•	Calibrate the wet-side of the unit with potassium
hydrogen phthalate (KHP) standard (monthly/
quarterly)
•	Replace the UV lamps (annually)
•	Replace the consumables in the carrier gas
generator (annually)
•	Clean infra-red detector window (annually)
A significant amount of time was required for trouble-
shooting and repairing the Hach astroTOC™ UV Pro-
cess Total Organic Carbon Analyzer, primarily due to
instrument malfunctions. The types of malfunctions
observed were mostly related to plugged or interrupted
flow (liquid/gaseous):
•	There was a build-up of silica crystals in the
resample block, which was rectified by Hach by
redesigning the resample block
•	The sparger orifice would get plugged (semi-
permeable membrane) and result in shut-down
of the instrument
Monthly calibration is essential to electronically adjust
for the instrument drift prior to the quarterly mainte-
nance/calibration event.
8.3.2	Sievers®!	inic
Carbon Analyzer
This instrument requires quarterly scheduled mainte-
nance. Depending upon the initial water quality, main-
tenance costs can vary between $2,000 and $4,000 per
year for consumables. The following routine mainte-
nance activities need to be performed:
•	Replace reagent/acid/oxidizer (quarterly/semi-
annual)
•	Change the pump tubing (semi-annual)
•	Replace the resin bed (semi-annual)
•	Replace UV lamp (semi-annual)
•	Replace ICR degasser, chemical trap, and pump
rebuilt kit (annual)
•	Replace in-line particulate filter (annual)
•	Replace oxidizer syringe (annual)
•	Replace restrictor tubing (annual)
Comparatively, a significant amount of time was re-
quired for troubleshooting and repairing the Sievers®
900 On-Line TOC Analyzer, primarily due to
instrument malfunctions. Malfunctions observed were
mostly related to plugged or interrupted liquid flow
(e.g., restrictor tube blockage). In areas with high-
carbonate water, reagents need to be replenished
quarterly.
8.3.3	Spectro;;lyzer™/Carbo;;lyzer™
As indicated earlier, this sensor is an optical instrument
that estimates TOC levels by measuring UV-Vis spectra

-------
between 200 and 750 nm wavelengths. Only TOC that
has some absorption property in the specified
wavelengths is detected. The instrument has minimal
O&M requirements. At first, EPA procured a unit made
of high-grade aluminum. However, the body of this in-
strument corroded in chlorinated Cincinnati tap water.
A replacement stainless steel unit was provided by the
manufacturer free of charge. EPA recommends that only
a stainless steel unit be purchased for locations with
water quality that might be aggressive to aluminum.
The unit in global calibration mode does not require any
calibration. If local TOC values are known, the unit can
be calibrated using locally available high and low
values. For obtaining a zero value in the local
calibration mode, a highly-purified High-Performance
(or High-Pressure) Liquid Chromatography (HPLC)
grade reagent or a high-grade distilled water should be
used. Deionized and nanopure waters are not suitable
for this purpose.
umentation
Free chlorine instruments responded to the majority of
contaminants tested at the T&E Facility on chlorinated
Cincinnati tap water. Both free and total chlorine instru-
ments use either a reagent-based (colorimetric) method
or a reagent-free method (amperometric/polarographic/
galvanic - electrode/membrane-based). The reagent-
free method has varied pH dependence, depending upon
the use of buffering agents. The reagent-based method
consumables include buffering and indicator solutions.
The reagent-free method consumables include electro-
lyte solutions (and membranes, if applicable).
.sCliC	arid Total Chlorine
Analyzer
Hach CL-17 Free and Total Chlorine Analyzers both
utilize the colorimetric method. Buffer and reagent
solutions, which adjust the pH and react with chlorine to
produce a color, are added to the sample. The color
depth is proportional to the amount of chlorine. Buffer
and reagent solutions last about a month. The estimated
cost of the buffer and reagent solution consumables is
between $750 and $1,000 ayear. In addition, the tubing,
stir-bar, and plastic-tube connectors need to be replaced
every six months. The colorimeter needs to be cleaned
every six months. This unit requires no calibration.
6.4.2 Wallac	man® Depolox® 3 plus
The Wallace & Tiernan® Depolox® 3 plus employs a
reagent free (potentiostatic 3-electrode-amperometric)
method for measurement. There is an option for either a
bare electrode or membrane-type measurement. The op-
tion selection is based on water hardness, conductivity,
and variation in pH. The membrane-type instrument is
recommended for a higher pH range (6 to 10 pH usable
range) and low conductivity (10 - 2,500 |iS/cm) waters.
In high pH waters, buffering (e.g., using C02) is needed.
Free chlorine consists of chlorine molecules (Cl2),
hypochlorous acid (HOC1) and hypochlorite ions
(OC1"). The presence of each component is mostly de-
pendent upon pH with some influence of temperature.
The HOC1 component is the most effective component
of free chlorine for disinfection. The electrodes
typically measure the HOC1 component and report it as
free chlorine. In membrane-type instruments, the
membrane is designed to allow only the HOC1 acid
through the membrane, which is measured and re-
ported as free chlorine. At a pH of > 8.5, the majority
of the HOC1 is converted to OC1", thus, interfering with
the accurate measurement of free chlorine. From a
maintenance perspective, the following consider-
ations are noted:
•	Replace the electrolyte in the electrode
reservoir (semi-annually)
•	In the bare electrode model, replace the
grit (that self-cleans the electrode) (semi-
annually)
•	In the membrane type model, clean the
electrode tip with abrasive paper and replace the
membrane (every three years)
The cost of consumables ranges between $350 and
$1,000 ayear. The calibration of the bare electrode is
performed by turning the sample flow off and setting
to zero (after waiting several minutes). The water is
then turned back on, and after waiting for the reading
to stabilize, a grab sample is collected, a N, N-
die thy l-/?-phcnylcnedia mine measurement is per-
formed and the span set to match the test result. For
the membrane-type unit, there is zero calibration.
6,4,3 YS1 6920DW
The YSI 6920DW is a multi-parameter (free chlorine,
turbidity, temperature, conductivity, pH, and ORP) in-
strument that uses a reagent-free (amperometric mem-
brane) method for free chlorine measurement (similar
to the instrument described in Section 6.4.2). From a
maintenance perspective:
•	Replace the electrolyte and membrane
(quarterly)
The annual cost of consumables is approximately
$2,100 ayear.
lalytica! Technology, Inc., Model
A15/62 Free Chlorine Monitor
The Analytical Technology, Inc., Model A15/62 free
chlorine monitor instrument uses a reagent-free (po-
larographic membrane) method for free chlorine mea-
surement (similar to the instrument described in Sec-
tion 6.4.2).
6-3

-------
•	Replace the electrolyte and membrane
(quarterly)
The annual cost of consumables is approximately $100
a year. The consumable costs fortius instrument are less
expensive when compared to the Wallace & Tiernan®
Depolox® 3 plus instrument, because there is no pH
electrode to be replaced.
6.4.6 Rosemount Analytical Model FCL
The Rosemount Analytical Model FCL uses a reagent-
free (membrane type amperometric) method for free
chlorine measurement (similar to the instrument de-
scribed in Section 6.4.2). In addition, the Rosemount
Analytical Model FCL-01 (with manual pH adjust-
ment) chlorine sensor can measure both HOC1 and OC1"
forms of chlorine. The sensor responds differently to
each form and compensates for both sample pH and
temperature. From a maintenance perspective:
•	Replace the electrolyte and membrane
(quarterly)
•	In addition, the membrane needs to be cleaned
(monthly)
During the contaminant injections, this membrane ap-
peared to be more prone to fouling than the other mem-
brane-type instruments. The annual cost of consum-
ables is approximately $400 a year.
, . )nductivi i ntation
The presence of dissolved mineral substances such as
chloride, nitrate, sulfate, and phosphate anions (ions
that carry a negative charge) or sodium, magnesium,
calcium, iron, and aluminum cations (ions that carry a
positive charge) dissolved in water can be measured as
conductivity (or temperature-compensated specific
conductance). Conductivity is basically a measure-
ment of the sample water's ability to carry an electrical
current. There are several factors that affect the
conductivity of water including: concentration of ions;
mobility of ions; oxidation state (valence); and tem-
perature. The testing at the T&E Facility indicated that
a high volume of contaminant injection was needed for
the conductivity value to change significantly.
Conductivity is typically measured directly by either
measuring the voltage drop or the current flow through
a sample.
The performance of the conductivity probes was similar
for the probes tested at the T&E Facility. The probes
evaluated at the facility included the following: Hach/
GLI Model C53 Conductivity Analyzer, YSI 6600, YSI
6920DW, Hydrolab® DS5 and Troll® 9000. From a
maintenance perspective:
•	These probes need to be cleaned and calibrated
(quarterly)
•	The probe requires replacement as needed
(generally lasts at least a year)
The consumables include calibration solutions that cost
approximately $200 per year.
tion
Potential Instrumentation
The majority of the manufacturers combine the pH/
ORP measurement with a reference electrode. The pH
value is a measure of the activity of hydrogen ions (H+)
in the water sample. Therefore, it a measure of the de-
gree of acidity or alkalinity of the water sample. ORP is
a measure of the tendency of the water sample to oxidize
or reduce another chemical substance. Typically, ORP is
measured using an inert metal electrode (platinum),
which will donate electrons to an oxidizing agent or
accept electrons from a reducing agent. The ORP
electrode continues to accept or donate electrons until it
develops a potential that is equal to the ORP of the
solution. ORP is sometimes utilized for estimating the
concentration of chlorine in water. However, ORP
measurement is affected by many factors and might not
be a good surrogate for chlorine. The testing at the T&E
Facility indicated that the ORP probes took longer than
pH probes to return to baseline and grab samples for
ORP are not reliable. The performance of the pH/ ORP
probes was similar for all of the probes tested at the T&E
Facility. The probes evaluated at the facility included the
following: Hach/GLI Model P53 pH/ORP Analyzer,
YSI 6600, YSI 6920DW, Hydrolab® DS5 and Troll®
9000. From a maintenance perspective:
•	These probes need to be cleaned and calibrated
(quarterly)
•	These probes require replacement as needed
(generally lasts at least a year)
•	The Hach/GLI Model P53 pH/ORP Analyzer
requires annual replacement of the salt-bridge
and electrolyte solution, which extends the
instrument life to about 3 years.
The consumables include pH calibration solutions and
probes that cost approximately $1,000 per year.
irbi
Turbidity is a measure of cloudiness or haziness of the
water sample caused by the suspended particles. Tur-
bidity is typically determined by shining a light beam of
wavelengths between 830 and 890 nm into the sample
solution and then measuring the light (at 90-degrees)
scattered by the suspended particles. Online nephelo-
6-4

-------
metric turbidity measuring devices were evaluated. The
turbidity sensors evaluated include the Hydrolab®DS5
and Hach FilterTrak™ 660 sc Laser Nephelometer.
From a maintenance perspective:
•	The Hach FilterTrak™ 660 sc Laser
Nephelometer requires quarterly cleaning/
calibration and annual replacement of the light
source
•	Hydrolab® DS5 turbidimeters require semi-
annual replacement of the wiper
The Hydrolab® DS5 optical turbidimeter probes last
about three years and cost approximately $1,500 per
unit. The annualized cost for consumables, including
the optical probe and calibration solutions, is approxi-
mately $1,000.
The annualized consumables cost for the Hach Filter-
Trak™ 660 sc Laser Nephelometer, including the bulb
replacement and calibration solution, is approximately
$300.
' ssolve "¦ ' • ygen
Dissolved oxygen probes are used to measure the
amount of gaseous oxygen (02) dissolved in the water
sample. Dissolved oxygen levels did not significantly
change after the initial contaminant injections and
therefore was removed from further evaluation.
' her Conventio
Water	meter/
Instrumentation
During the initial phases of testing at the T&E Facility,
various other ion-selective electrodes (ISEs) were also
evaluated for their efficacy and usefulness in detecting
changes in water quality. The other ISEs evaluated for
their ability to measure the following parameters: am-
monia, nitrate, and chloride. Although some of these
parameters changed in the presence of the contami-
nants, free chlorine interfered with ISE calibration and
prevented the repeatability of the measurement. In addi-
tion, some of the ISEs burned out too quickly and were
expensive to replace. Therefore, these parameters were
excluded from further testing.
8,	:al
Instrumentation
The following optical instruments (listed alphabeti-
cally) were evaluated mostly for their ability to detect
biological agents and growth media at the T&E Fa-
cility: FlowCAM®, Hach 2200 PCX Particle Counter,
Hach FilterTrak™ 660 sc Laser Nephelometer,
BioSentry®, Spectro::lyzer™ and Carbo::lyzer™ and
ZAPS MP-1. In addition, because of their ability to
detect contaminants in addition to biological organ-
isms, the Spectro::lyzer™ and Carbo::lyzer™ and the
ZAPS MP-1 were tested using various chemical injec-
tions. Generally, the biological organism tests were per-
formed by injecting 100, 600, 1,000, and 25,000 cells/
mL (both in chlorinated and dechlorinated waters).
Based on previous testing, EPA had discovered that the
online chlorine and TOC monitors generated reliable
responses at injections of 100,000 cells/mL and some
responses at 25,000 cells/mL, but the response faded
below this level.
Generally, these optical instruments were evaluated as
they became available for the research and, therefore,
evaluations utilizing all contaminants with each instru-
ment were not performed. The following is a brief sum-
mary of their capability and equipment performance
observation.
For the Updated Guide, the following optical
instruments were also tested: ZAPS LiquID, Turner
TD1000C, Hach UVAS, RealTech UVT, ChemScan®
UV-2150 Process Analyzer, and Optiqua Refractive
Index.
6,10,1 FlowCAM®
FlowCAM® is an online particle imaging and flow-
cytometry system that takes high-resolution digital
images of particles and cells in the water sample. The
images are analyzed by a proprietary software program
based on Microsoft® Office Excel® that captures and
analyzes the parameters of the particles such as count,
size, length, shape, and equivalent spherical diameter. In
addition, the instrument captures the intensity, trans-
parency, color, bio-volume, compactness, roughness,
and elongations of the particles.
At the T&E Facility, Ankistrodesmus (20 to 100 mi-
crometers | Selenastrum (10 |im). and Saccha-
romyces y east (1.5 |im) were inj ected. The unit was able
to recognize all of the particles. In order to capture
individual images, the flow cytometry setup sig-
nificantly reduces the flow volume to the unit. This
restrictive flow path causes significant delays in the
measurement from the injection time. At concentra-
tions below 1,000 cells/mL, the instrument is unable to
differentiate baseline noise from injected contami-
nants. Sub-micron particles such as E. coli (0.8 to 0.9
|im). bacteriophage MS2 (0.02 to 0.03 |im) and B. glo-
bigii (0.5 to 0.9 |im - spores are smaller than the cells)
were not identifiable with the existing camera optical
resolution and flow cell. However, at sufficiently high
concentrations, the instrument is able to show an in-
creased count.
6-5

-------
6.10.2	Hach FilterTrak™ 880	ir
Nephelometer and Hach 2200 PCX
Particle Counter
The Hach FilterTrak™ 660 laser nephelometer uses a
collimated light source with high beam density and a
distinct wavelength to detect baseline turbidity change
as low as 0.5 milli-nephelometric turbidity units
(mNTU) (0.0005 NTU). The instrument is optimized to
detect particles in the 0.1 to 0.5 |im range. The Hach
2200 PCX Particle Counter is a laser-diode-based
particle counter designed for drinking water
applications. The instrument is optimized to detect
particles in the range of 2-750 |im.
The instruments were challenged with the following
biological injections performed at the T&E Facility:
E. coli, MS2 and B. globigii. The Hach Filter/Trak™
660 sc Laser Nephelometer was able to detect only E.
coli and B. globigii at injection levels of 25,000 cells/
mL. The Hach 2200 PCX Particle Counter was unable
to detect any of the biological injections at these levels.
The particle counter is designed for detecting larger
biological particles such as Cryptosporidium spp. and
Giardia lamblia.
8.10.3	BioSentry®
The BioSentry® system is a laser-based, continuous,
online, real-time monitoring device for detecting
microorganisms in water. The unit utilizes laser-pro-
duced, multi-angle light scattering (MALS) tech-
nology to generate unique bio-optical signatures for
classification using BioSentry®'s pathogen detection
library. BioSentry® can be set up to detect microor-
ganisms and identify suspected pathogens.
The biological injections performed at the T&E Fa-
cility included the following: 3-micron beads (sur-
rogates for Cryptosporidium spp.), B. globigii and E
coli. In the current design (as tested), the unit has to be
programmed to identify a specific contaminant,
whereas all others (even when detected) are identified
as unknowns. Prior to any injection(s), the unit should
be programmed to recognize the injected particle, i.e.,
the unique bio-optical MALS signature should be de-
veloped and put to use (using the local water and pure
form of the injected particle). The unit tested at the
T&E Facility was able to reliably recognize injection
events and identify injected particles between 1,000
and 10,000 cells/mL level. At lower cell concentra-
tions, the detection was not consistent across the test
runs.
8.10.4	Spectro::lyzer™/Carbo::lyzer™
These units are based on UV-Vis spectroscopic absorp-
tion measurements. Contaminants that respond to the
UV-Vis absorption spectra can be detected by this in-
strument. The Spectro::lyser™ was programmed to
measure the optical equivalents of turbidity, nitrate,
TOC, and dissolved organic carbon (DOC). In addition,
the Spectro ::lyzer™ was connected to the con::stat™
process control terminal, which ran the proprietary soft-
ware to compute four pre-set alarm parameters based
on computations of spectral channels that were consid-
ered to be important by the manufacturer.
At the T&E Facility, both chemical and biological in-
jections were performed using the Single Pass DSS.
The chemical injections included: humic acid, sodium
fluoroacetate, aldicarb, dicamba, and gasoline. The in-
strument was able to detect the water quality changes
for all of the contaminants, except sodium fluoroace-
tate. The biological injections included the following:
sucrose, E. coli and B. globigii. The instrument was able
to detect changes at injection levels of approximately
25,000 cells/mL.
Results at the T&E Facility indicated that this device
was capable of serving as a good surrogate for traditional
TOC measurement. The operation and maintenance re-
quirements for this device were minimal when compared
to the traditional UV-persulfate method-based TOC
measuring devices. Also, the size of this device is much
smaller than the traditional TOC measuring devices. As
mentioned previously in Chapters 2, 3, and 5, TOC is a
critical water quality trigger parameter. EPA recom-
mends the use of this type of device especially at loca-
tions where the traditional TOC devices are difficult to
deploy, either due to size or due to ongoing operational
and maintenance costs. Subsequently, the Carbo: :lyzer™
was successfully deployed at two locations under the
WSi pilot study in Cincinnati alongside the traditional
TOC measuring devices. Furthermore, this instrument
and associated software are capable of analyzing the full
spectrum of UV-Vis absorbance, which if fully exploited,
can yield additional information about the water quality
changes that are not captured by the other devices evalu-
ated by EPA at the T&EFacility.
8,10,1	P-1
The ZAPS MP-1 is an online water quality monitoring
device that can be programmed to measure up to 100
slices of optical wavelengths (using optical filters)
between 200 and 800 nanometers. The optical data that
can be captured includes absorption, fluorescence, and
total reflection bands. The ZAPS MP-1 configuration at
the T&E Facility was set up to measure the following
parameters: dark counts, pinhole, nitrate, ultraviolet 254
nanometer wavelength (UV254) absorbance, bacterial
fluorescence, humic fluorescence, total fluorescence,
rhodamine, and transmission. The individual excitation
and response wavelengths were pre-set by the
manufacturer to these parameters.
6-6

-------
At the T&E Facility, nitrate, formazin, Saccharomyces
cerevisiae (yeast) and E. coli were injected to evaluate
responses on this instrument. The instrument respond-
ed well to nitrate injections at 0.14 mg/L. The instru-
ment responded well to the injection of formazin (tur-
bidity standard) at test levels of 13.3 NTU and 26.6
NTU; a linear response was observed in UV254, total
fluorescence, bacterial fluorescence, and nitrate chan-
nels. However, for Saccharomyces cerevisiae (yeast)
and E. coli injections, the instrument showed no re-
sponse below 100,000 cells/mL. Therefore, no further
testing was performed with the biological agents.
6.10.6	quID
The ZAPS LiquID has been used to detect airplane
deicer in runoff from airports. The unit is based on UV-
Vis absorption, fluorescence, and reflectance
measurements (EPA, 2012). It was tested as a complex
multiple wavelength sensor. The unit requires factory
calibration.
The ZAPS LiquID was able to detect antifreeze at 10
mg/L, bleach at 5 mg/L, diesel fuel at 5 mg/L,
DISPERSIT® at 10 mg/L, E. coli at 1,150 colony
forming units (CFU)/mL, sodium thiosulfate at 9 mg/L,
ethylene glycol at 1 mg/L, and pepsin at 10 mg/L.
It was not able to detect antifreeze at 1 mg/L, deicer at
10 mg/L, diesel fuel at 1 mg/L, DISPERSIT® at 1 mg/L,
or pepsin at 1 mg/L.
It was leased for one month during testing for $3,000.
The capital cost was quoted as $60,000. No
maintenance was performed on this unit during the
testing period.
6.10.7	Turner TD1000C
The Turner TD1000C is an online fluorometer with a
single excitation and emission wavelength (EPA, 2012).
It works with fluorescent hydrocarbons. It was utilized
as an example of a relatively simple off-the-shelf
fluorometric sensor.
The Turner unit was able to detect DISPERSIT® at 10
mg/L, E. coli at 1,150 CFU/mL, sodium thiosulfate at 9
mg/L, and Basagran® at 1 mg/L.
It was not able to detect antifreeze at 10 mg/L, bleach at
5 mg/L, deicer at 10 mg/L, diesel fuel at 5 mg/L,
DISPERSIT® at 1 mg/L, ethylene glycol at 10 mg/L, or
pepsin at 10 mg/L.
The Turner TD1000C was purchased for $12,000, and
maintenance costs are estimated to be $200/year based
on labor costs and replacement of disposable items such
as tubing.
6.10.8	Hach UVAS
The Hach UVAS is based on UV absorption
measurement (2-beam technique) at 254 nm (EPA,
2012).
The Hach UVAS was able to detect antifreeze at 10
mg/L, bleach at 5 mg/L, DISPERSIT® at 1 mg/L, E. coli
at 1,150 CFU/mL, sodium thiosulfate at 9 mg/L, pepsin
at 1 mg/L, and Basagran® at 1 mg/L.
It was not able to detect antifreeze at 1 mg/L, deicer at
10 mg/L, diesel fuel at 5 mg/L, or ethylene glycol at 10
mg/L.
The Hach UVAS sensor was purchased for $15,000.
Minimal maintenance is needed, and maintenance costs
are estimated at $200/year, including labor and
disposable items such as new tubing and lens cleaning
materials.
6.10.9	RealTech UVT
The RealTech UVT is based on UV absorption
measurement with anti-drift compensation at 254 nm
(EPA, 2012).
The RealTech UVT was able to detect antifreeze at 10
mg/L, bleach at 5 mg/L, DISPERSIT® at 1 mg/L, E. coli
at 1,150 CFU/mL, sodium thiosulfate at 9 mg/L, pepsin
at 1 mg/L, and Basagran® at 1 mg/L.
It was not able to detect antifreeze at 1 mg/L, deicer at
10 mg/L, diesel fuel at 5 mg/L, or ethylene glycol at 10
mg/L.
The RealTech UVT sensor "as tested" capital cost was
$7,000. Minimal maintenance is needed, and
maintenance costs are estimated at $200/year, including
labor and disposable items such as new tubing and
lamps.
6.10.10	ChemScan® UV-2150 Process Analyzer
The ChemScan® UV-2150 Process Analyzer uses
reagent-assisted, multiple-wavelength ultra-violet (UV)-
absorbance methodology, to measure a customized set of
water quality parameters. This instrument can be
customized to measure one or more of the following four
reagent-based water quality parameters (from up to two
sample lines): free ammonia, total ammonia, mono-
chloramine, and total chlorine. In addition to these
reagent-based parameters, the instrument can report the
raw UV-absorbance values.
Chemicals that possess natural absorbance
characteristics can be detected directly using primary
absorbance techniques. However, for chemicals that do
not possess natural absorbance, ASA Analytics (the
manufacturer of ChemScan®) has created proprietary
6-7

-------
reagents and secondary analysis techniques of
measurement. The instrument evaluated in this report
primarily uses a reagent-based technique to detect total
chlorine. In addition to the reagent-based detection of
total chlorine, the raw UV-absorbance values for the
following selected wavelengths were also recorded during
the evaluation: 230 nanometers (nm), 240 nm, 254 nm,
270 nm, 290 nm, 350 nm and 448 nm.
The ChemScan® instrument was tested with sodium
thiosulfate pentahydrate, malathion, and glyphosate. The
ChemScan® unit showed a decrease in chlorine
concentration when testing sodium thiosulfate. The
ChemScan® instrument also detected 5 mg/L of
malathion and 10 mg/L of glyphosate.
The list price of the ChemScan® instrument is $25,000
for a one sample line, four-output system. The
consumables for the ChemScan® UV-2150 unit include:
reagents (potassium iodide and tripolyphosphate mix),
buffer (sulfuric acid), zero standard (DI water) and
cleaning solution (dilute muriatic/hydrochloric acid). The
total estimated cost of consumables is $l,000/year.
The inj ector pump failed during the first year of operation.
The nominal replacement cost for the pump is $780.
However, the pump was replaced at no cost under
warranty.
tiqua Refractive Index
EventLab is based on Optiqua's patented lab-on-chip
sensor technology. A single optical sensor technology (a
Mach Zehnder Interferometer [MZI]) continuously
measures refractive index (RI) changes in the water
matrix induced by contaminant(s). The sensor
measurements are processed by dedicated electronics, and
software and the data are communicated to a central
server (hosted by Optiqua Technologies) via a cellular
modem through the internet. The data are processed by
Optiqua's proprietary algorithms to detect changes in
water quality and to alert the user of anomalous changes
in water quality.
The Optiqua instrument was tested with sodium
thiosulfate pentahydrate, malathion, glyphosate, sugar,
salt, and 4-methylcyclohexanemethanol (MCHM). The
Optiqua unit showed a decrease in chlorine concentration
when testing thiosulfate and detected 10 mg/L glyphosate
and 30 mg/L sugar. It did not detect 10 mg/L of malathion,
10 mg/L sodium chloride, or 10 mg/L MCHM.
The Optiqua instrument was purchased for $12,000. This
included the probe, control box, and the web-based central
data-processing service for the duration of the
experiments.
Based on single unit use, the annual cost of the multiple
cleaning agents required to maintain the device is
estimated to be roughly $130. The purchase/replacement
cost for the 1-micron in-line filter housing is $56, and
each filter costs roughly $5. With a quarterly filter
replacement schedule, the annual maintenance cost
(excluding labor) is estimated to be $76.
;tom Sensors
After the original Guide was published (EPA, 2009),
EPA began testing additional sensors. A number of these
sensors did not fit into the existing sensor categories of
the original Guide (e.g., TOC, chlorine, conductivity,
pH/ORP, turbidity, DO, other conventional water quality
parameter/ instrumentation, and online optical
instrumentation). These additional sensors include
chemical oxygen demand (COD), biological, and metal
detecting sensors.
8,11,1 PeCOD® PI00 COD Analyzer
The PeCOD® P100 COD Analyzer employs a unique
patented technology that can directly measure the
photocurrent originating from the oxidation of organic
species contained in a sample. The core of the
technology is the PeCOD® sensor, which consists of a
UV-activated nano-particulate Ti02 (titanium dioxide)
photocatalyst coupled to an external circuit. The sample
is introduced into a microcell containing the sensor. The
Ti02 is irradiated by UV light, and a potential bias is
applied. The UV light creates a photohole in the Ti02
sensor with a very high oxidizing power (3.1 eV), and
organic contents in the cell are oxidized (the chemical
potential of the dichromate method is only 1.8 eV). The
PeCOD® COD analyzer exhaustively oxidizes organics
and counts the electrons that are liberated to provide a
direct measure of equivalent COD. It gives a direct
measurement of the oxidation of organic compounds,
thus providing a real measurement of organic pollutants
and not inferred ones. The reaction produces mainly
carbon dioxide and water from organic C, H, and O, as
well as from organic N, mineral acids, organic halides
(CI, Br, etc.), S, or P.
The PeCOD® P100 COD Analyzer was challenged with
60 mg/L sugar, 60 mg/L sodium chloride, 10 mg/L
glyphosate, and 10 mg/L malathion injections as well as
with non-chlorinated secondary effluent from MSD's
wastewater treatment plant (located adjacent to the T&E
Facility). The PeCOD® unit was able to detect the
compounds as they passed through the unit; however, it
was determined that the data were not consistent due to
instrument background noise, false positives, and
maintenance issues.
Based on analyzing one sample per hour, 24 hours per
day, 5 days per week, calibrating the unit after every 10
samples, the annual operating costs for the PeCOD®
PI00 COD Analyzer were estimated to be approximately
$5,700 for electrolyte, $800 for calibrant, and $6,000 for
sensors for a total estimated annual operating cost of
$12,500.

-------
?> Analyzer
According to the manufacturer, the EZ-ATP® Analyzer
can be used as an early warning system in various
applications, where high adenosine triphosphate (ATP)
values indicate a potential risk in surpassing a threshold
value of microorganisms as established in safe drinking
water regulations. In addition to drinking water,
application fields for this device include: bottling, raw
water intake, industrial cooling systems, data center
HVAC, RO demineralization, desalination, and oil and
gas.
The EZ-ATP® Analyzer quantifies different ATP
portions, providing operators data on:
•	Extracellular ATP representing the portion of
ATP released by dead cells (Free ATP)
•	Total ATP
•	Intracellular ATP, representing the portion of
ATP from the metabolism of living
microorganisms (Live Cell ATP)
During pressurized line tests in the Single Pass DSS, the
EZ-ATP® Analyzer was challenged with test
concentrations of E. coli between 1.0E+03 cells/mL and
1.0E+06 cells/mL. The lowest E. coli concentration
with a positive ATP response was 1.36E+04 cells/mL;
therefore, the detection limit for the instrument falls
between 1.32E+03 cells/mL and 1.36E+04 cells/mL. E.
coli event detections are based on the ATP response of
0.5 pg/mL above the baseline reading for either total
ATP, free ATP or live bacteria.
The lowest E. coli concentration tested and detected in
the grab sample/manual mode using GAC matrix water
was 1.50E+03 cells/mL. A similar concentration of E.
coli (1.32E+03 cells/mL) was not detected in the
pressurized line tests. It is possible that the presence of
the dechlorinating agent (sodium thiosulfate) in the
pressurized DSS line negatively impacted the EZ-
ATP® Analyzer response at lower E. coli
concentrations during these tests.
Additional testing, using non-chlorinated secondary
effluent from MSD's wastewater treatment plant was
used to simulate the contamination of a water source
with sewage or some other source of bacterial
contamination, such as concentrated animal feeding
operations (CAFOs) runoff. The secondary effluent was
titrated into GAC water by several orders of magnitude
to establish an ATP detection limit.
Multiple dilutions from 1:100 to 1:10,000 were made
and sampled by the EZ-ATP® device. A positive result
(ATP > 0.5 mg/L) was obtained at every dilution less
than 1:2,000. This indicates that the EZ-ATP® device
can be used for determining the presence of biological
contaminant(s) in a water source.
8,11,3 Sporian Inline Biosensor System (IBS)
The Sporian IBS is a flow-through device that is
equipped with a proprietary sensor cartridge designed to
detect biological contamination. The IBS sensor
cartridge contains a proprietary molecular detection
element (MDE), which consists of a fluorescing (emits
light) media. When biological contamination is present,
the fluorescing media is released from the MDE and
bound to the contaminant. The rate of change of MDE
media is roughly proportional to target concentration
within the environment (how much target interacts with
the MDE), and as such, high target concentrations may
result in a very fast response (in the order of minutes)
and depletion of the cartridge. A positive signal response
is indicated by a decrease in fluorescing signal over
elapsed time (in microseconds) as the MDE is exposed
to the target biological contaminant within the
environment. Due to processing variations, the initial
value will differ from cartridge to cartridge. Over a
period of time, the MDE becomes completely exhausted
and needs to be replaced (EPA 2012).
The IBS was tested with E. coli and Bacillus subtilis
spores. The unit was capable of detecting both dead and
live biological contaminants, with a higher response
from E. coli than B. subtilis. Future evaluation would
need to be conducted to improve the ability of the MDE
to provide confirmation of specific biological
contamination and reduce aspects of non-specific
binding that can occur with the MDE cartridge. This
could then lead to improved target specific MDE
cartridges for contaminant detection and then allow for
specific concentrations of the target to be calculated
(Smith et al. 2020).
Sci Online Water Sensor
According to the manufacturer, the Quantitative
Biosciences, Inc. (QBiSci) Online Water Sensor uses
genetically-modified E. coli to continuously detect the
presence of contaminants in a water supply. Specifically,
eight different engineered cell strains differentially
produce a green fluorescent protein (GFP) reporter in the
presence of any of the following species: As(III), As(V),
Cd(II), Hg(II), Pb(II), Sb(III), tributyl phosphate, and
U(VI). The strains are arrayed in quadruplicate
configuration within a microfluidic device. The device
periodically images the array of cells to visualize the
pattern of fluorescence. A classifier algorithm interprets
the combination of strain fluorescence intensities in each
image in real time to determine the presence or absence
of each contaminant at the provided detection limit. A
presence/absence alarm for each contaminant, as well as
the confidence in detection of each contaminant (on a 0-
1 scale), are outputs via the Modbus TCP/IP or RS-232
RTU protocol. With online connectivity, the results can
be delivered to remote users.
6-9

-------
The QBiSci unit was tested with both sodium arsenite and
cadmium nitrate. The QBiSci Online Water Sensor clearly
detected 200 ppb cadmium and 200 ppb arsenic; however,
lower concentrations of 5 ppb and 10 ppb of the same
contaminants produced responses near the background
levels of unspiked source water. However, if an absolute
change (change from baseline), a percent change, and
signal-to-noise ratio (the absolute change divided by the
standard deviation of the baseline period) are considered
together, it appears that cadmium had positive responses at
5,10 and 200 ppb, and arsenic only had a positive response
at 200 ppb.
6.12 Best Practices and Lessons
Learned
As presented in Sections 6.1 through 6.11, many
conventional instruments are relatively easy to operate
and maintain. However, the TOC instruments are more
complex and require a relatively higher level of technical
skill. A formal training class provided by the
manufacturer for each type of instrument is highly
recommended. Just by following the instruction manuals,
experienced instrumentation technicians and field
engineers who have several years of experience are
usually capable of installing, setting up, calibrating, and
operating new or unfamiliar instrumentation. Less
experienced staff will require assistance from the
instrument manufacturer to avoid some of the pitfalls that
can cause serious damage to an instrument and result in
improper/inefficient operation. For example, the
Sievers® 900 on-line TOC analyzer requires the de-
ionized loop reservoir to be filled prior to power up;
otherwise, air is trapped in the measurement modules,
which will require factory service. Furthermore, factory
default settings for an instrument might not be suitable for
the location; the operator should be aware of the exact
settings for each location prior to service. As another
example, when installing a probe to an analyzer such as
Wallace & Tiernan® Depolox® 3 plus analyzer, it is
necessary to complete a formal setup in the instrument
analyzer to recognize the probe correctly. It is always
advisable to attend training workshops offered by most
instrument manufacturers. Adequate staff training for
setup and maintenance activities is essential for optimal
operations. In addition, for more complex instruments
(such as the optical instrumentation), it is essential for the
maintenance technician to receive formal manufacturer
training.
The operation of a single or a few instruments is a
relatively straightforward process; however, the operation
of a multi-station network or multiple networks can be a
logistical nightmare. Loss of data, false alarms, and other
malfunctions can lead to improper analysis of data by
algorithms and inappropriate actions. The operator should
consult with the instrument manufacturer(s) and develop
~ Water Utilities and
Sensor Manufacturers
•	Over the life of the equipment operation, in
general, the O&M cost will exceed capital cost
for most online sensing equipment. At a
minimum, the water utilities should set aside a
budget for annual labor and consumable costs,
based on the detailed information presented in
this chapter.
•	The evaluations at the EPA T&E Facility and
during the WSi pilot study revealed that almost
60% to 80% of the O&M labor cost associated with
the water quality monitoring sensors was related to
TOC instruments.
•	Instrument technicians should be appropriately
factory-trained for optimal operations.
•	O&M activities should be appropriately scheduled.
Consumables should be purchased in a timely
manner so that they do not expire before they can
be used up.
•	Ideally, sensor manufacturers need to develop
reagent-free sensors that result in lower labor and
consumable costs.
•	In the absence of better TOC instrumentation, TOC
surrogate monitoring (such as the Spectro::lyser™
- see section 6.10.4) should be seriously
considered for both chlorinated and chloraminated
systems as the maintenance requirement for this
type of device is minimal.
a thorough understanding of the instrument outputs
during power outages and other malfunctions such as the
loss of reagents. It is important to develop a monitoring
plan for scheduled maintenance, order expendable
supplies in a timely manner, maintain calibration
standards, and schedule sufficient manpower to cover
network operations. Following a good monitoring plan
will ensure the collection of high-quality data that meets
the monitoring requirements.
In general, most instrument problems are related to flow
(sample and/or reagent issues). The sample flow problem
could be related to restrictions in the flow manifold or
restriction of flow through the instrument. Reagent flow
blockage can result in diminished or un-stable readings.
If the reagents run out completely, the readings typically
drop to zero and are easy to spot. Some instruments are
factory-set to hold the last good reading; if the numbers
do not change over a significant period, it is an indication
of instrument failure. For membrane-type probes, the
reading usually drifts downwards as the membrane is
clogged or nearing its useful life. For electrodes, failure
is generally indicated by problems during calibration
6-10

-------
where either the slope or the gain or the cell constant is
outside of the manufacturer-recommended tolerance
range. Degassing or bubbles can cause improper readings.
Degassing is generally an issue where the incoming
sample water is colder than the environmental housing of
the instrument.
When purchasing consumables, one needs to be sure to
use them prior to their expiration dates. For some ISEs,
the shelf life begins from the date of manufacture and not
the date of installation. When planning to purchase spare
parts and consumables, the shelf life and projected use by
date should be taken into consideration. Water utilities are
encouraged to work with manufacturers to negotiate
purchase price of equipment based on volume of purchase
and any negotiated long-term service contracts.
6-11

-------
7.0 Bibliography
The references included in this bibliography contain
additional detailed information for readers who wish to
pursue, in greater detail, the specific topics discussed in
this guide. Many of these references (especially the
EPA references) are freely available on the internet. The
references are listed alphabetically, based on the last
name of the first author(s). In cases where there are two
or more works by the same author (e.g., EPA), the
entries are listed chronologically.
Allgeier, S.C., J. Pulz, and R. Murray, 2006. Con-
ceptual design of a contamination warning system.
Proceedings of the AWWA Water Security Congress,
Washington, DC.
AwwaRF, 2002. Online monitoring for drinking water
utilities. Editor, Erika Hargesheimer, Awwa Research
Foundation and CRS PRO AQUA, American Water
Works Association, Denver, CO.
Dawsey, W.J., B.S. Minsker, and V.L. VanBlaricum,
2006. Bay esian belief networks to integrate monitoring
evidence of water distribution system contamination.
Journal of Water Resources and Planning, ASCE,
132:4:234.
Department of Homeland Security, 2006. National
Infrastructure Protection Plan.
Edberg, S. C., E.W. Rice, R. J. Karlin, and M. J. Allen,
2000. Escherichia coli: The Best Biological Drinking
Water Indicator for Public Health Protection. Journal
of Applied Microbiology, 88 (S6).
Einfeld, W., S. McKenna, and M.P. Wilson, 2008.
A Simulation Tool to Assess Contaminant Warning
System Sensor Performance Characteristics.
AwwaRF, Denver, CO.
EPA (U.S. Environmental Protection Agency), 2003.
Contamination threat management guide - Module 2.
EPA/817/D-03/002, U.S. Environmental Protection
Agency, Washington, DC.
EPA, 2004a. The Water Security Research and Techni-
cal Support Action Plan. EPA/600/R-04/063, U.S.
Environmental Protection Agency, Washington, DC.
EPA, 2004b. Public Drinking Water Systems Pro-
grams : http: //www. epa. gov/dwreginfo/.
EPA, 2004c. Response Protocol Toolbox Overview.
EPA/817/D-03/007, U.S. Environmental Protection
Agency, Washington, DC.
EPA, 2004d. Response Protocol Toolbox: Water
Utility Planning Guide - Module 1. EPA/817/D-
03/001, U.S. Environmental Protection Agency,
Washington, DC.
EPA, 2004e. Response Protocol Toolbox: Con-
tamination Threat Management Guide - Module 2.
EPA/817/D-03/002, U.S. Environmental
Protection Agency, Washington, DC.
EPA, 2004f. Response Protocol Toolbox: Site
Characterization and Sampling Guide - Module 3.
EPA/817/D-03/003, U.S. Environmental
Protection Agency, Washington, DC.
EPA, 2004g. Response Protocol Toolbox: Analytical
Guide - Module 4. EPA/817/D-03/004, U.S.
Environmental Protection Agency, Washington, DC.
EPA, 2004h. Response Protocol Toolbox:
Public Health Response Guide - Module 5.
EPA/817/D-03/005, U.S. Environmental
Protection Agency, Washington, DC.
EPA, 2004i. Response Protocol Toolbox: Remediation
and Recovery Guide - Module 6. EPA/817/D-03/006,
U.S. Environmental Protection Agency, Washington,
DC.
EPA, 2004j. Response Protocol Toolbox: Planning
for and Responding to Drinking Water Contamina-
tion Threats and Incidents Response Guidelines.
EPA/817/D-04/001, U.S. Environmental Protection Agency,
Washington, DC.
EPA, 2004k. Homeland Security Strategy. U.S.
Environmental Protection Agency, Washington, DC.
EPA, 2005a. Evaluation of Water Quality Sensors as
Devices to Warn of Intentional Contamination in Water
Distribution Systems. EPA/600/R-05/105, Washington,
DC. Available only through WaterlSAC.
EPA, 2005b. Water Sentinel Online Water Quality
Monitoring as an Indicator of Drinking Water Con-
tamination. EPA/817/D-05/002, U.S. Environmental
Protection Agency, Washington, DC.
EPA, 2005c. Water Sentinel System Architecture.
EPA/817/D-05/003, U.S. Environmental
Protection Agency, Washington, DC.
EPA, 2005d. Technologies and Techniques for Early
Warning Systems to Monitor and Evaluate Drink-
ing Water Quality: A State-of-the-Art Review.
EPA/600/R-05/156, U.S. Environmental Protection
Agency, Washington, DC.

-------
EPA, 2007a. Water Quality Sensor Response to Con-
tamination in a Single Pass Water Distribution System
Simulator. EPA/600/R-07/001, Washington, DC.
Available only through WaterlSAC.
EPA, 2007b. Water Security Initiative: Interim Guid-
ance on Planning for Contamination Warning System
Deployment. EPA/817/R-07/002, U.S. Environmental
Protection Agency, Washington, DC.
EPA, 2008a. Water Security Initiative: Interim Guid-
ance on Developing an Operational Strategy for
Contamination Warning Systems. EPA/817/R-08/002,
U.S. Environmental Protection Agency, Washington,
DC.
EPA, 2008b. Water Security Initiative: Interim Guid-
ance on Developing Consequence Management Plans
for Drinking Water Utilities. EPA/817/R-O8/001, U.S.
Environmental Protection Agency, Washington, DC.
EPA, 2008c. Water Security Initiative: Cincinnati Pilot
Post-Implementation System Status - Covering the
Pilot Period: December 2005 through December 2007.
EPA/817/R-08/004, U.S. Environmental Protection
Agency, Washington, DC.
EPA, 2009. Distribution System Water Quality
Monitoring: Sensor Technology Evaluation
Methodology and Results - A Guide for Sensor
Manufacturers and Water Utilities. EPA/600/R-
09/076, U.S. Environmental Protection Agency,
Washington, DC.
Homeland Security Presidential Directive- 9/HSPD-9, 2004.
Defense of United States Agriculture and Food.
https://www.hsdl.org/?abstract&did=444013
ISLI, 1999. Early warning monitoring to detect haz-ardous
events in water supplies. ISLI Press, Washington, DC.
Kessler, A., A. Ostfeld, and G. Sinai, 1998. Detecting accidental
contaminations in municipal water networks. Journal of Water
Resources Planning andManagement-ASCE, 124:4:192.
Kirmeyer, G., M. Friedman, andK. Martel, 2002. Guid-ance
manual for monitoring distribution system water quality. Awwa
Research Foundation, Denver, CO.
Kroll, D., and K. King, 2007. To switch or not to switch.
Potential security considerations in the free chlorine versus
monochloramine debate for drinking water disinfection.
Proceedings of the ASM Bio-defense and Emerging Diseases
Research Meeting, Washington, DC.
Lytle, J., and E. W. Rice, 2002. A Systematic Comparison of the
Electrokinetic Properties of Environmentally Important
Microorganisms in Water. Colloids and Surfaces B:
Biointerfaces, 24, 91-101.
McKenna, S.A., K.A Klise, and M.P. Wilson, 2006. Testing
water quality change detection algorithms. Proceedings of the
8th Annual Water Distribution Systems Analysis (WDSA)
Symposium, Cincinnati, OH.
EPA, 2010. Detection of Biological Suspensions in a
Drinking Water Distribution System Simulator.
EPA/600/R-10/005, U.S. Environmental Protection
Agency, Washington, DC.
EPA, 2012. Sporian In Line Biosensor (IBS)
Evaluation Summary. EPA/600/R-12/580, U.S.
Environmental Protection Agency, Washington, D.C.
EPA, 2012. Detection of Contamination in Drinking
Water Using Fluorescence and Light Absorption
Based Online Sensors. EPA/600/R-12/672, U.S.
Environmental Protection Agency, Washington, DC.
Hall, J.S., A.D. Zaffiro, R.B. Marx, P.C. Kefauver,
E.R. Krishnan, R.C. Haught, and J.G. Herrmann,
2007. On-line water quality parameters as indicators of
distribution system contamination. Journal of Ameri-
can Water Works Association, 99:1:66.
O'Halloran, R., R. Jarrett, G. Robinson, and P. Toscas, 2009.
Data Processing and Analysis for Online Distribution System
Monitoring, AwwaRF, Denver, CO.
Rice, E. W., N. J. Adcock, M. Sivaganesan, and L. J. Rose,
2005. Inactivation of spores of Bacillus anthracis Sterne,
Bacillus cereus, and Bacillus thuringiensis subsp. israelensis by
chlorination. Applied and Environmental Microbiology. 9: 5587-
5589.
Roberson, J.A. and K.M. Morley, 2005, Guidance document,
contamination warning systems for water: An approach for
providing actionable information to decision-makers, AWWA.
Sivaganesan, M., N.J. Adcock, and E.W. Rice, 2006.
Inactivation of Bacillus globigii by chlorination: A Bayesian
model. Journal of Water Supply Research Technology - Aqua,
55.1: 33-43.
Homeland Security Presidential Directive-7/HSPD-7,
2003. Critical Infrastructure Identification, Prioritiza-
tion, and Protection: https://www.cisa.gov/homeland-
security-presidential-directive-7.

-------
Smith, C.B., A.F. Fisher, A.T. Ly, M.J.
Anderson, 2020. Autonomous Querying and
PATHogen Threat Agent Sensor System
(AQUA PATH) Monitoring Source Waters
with Geospatially Wirelessly Networked
Distributed Sensing Systems. U.S. Army
ERDC Technical Report: ERDC/GRL TR-
20-8,
http://dx.doi.org/10.21079/11681/37533
Szabo, J.G., J.S. Hall, and G.C. Meiners,
2006. Water quality sensor responses to
injected contaminants in a chloraminated
pipe loop. Proceedings of the AWWA Water
Security Congress, Washington, DC.
Szabo, J.G., J.S. Hall, and G.C. Meiners,
2008a. Sensor responses to contamination in
chloraminated drinking water. Journal of
American Water Works Association, April
2008, Denver, CO.
Szabo, J.G., J.S. Hall, and G.C. Meiners,
2008b. Detection of biological suspensions
using online detectors in a model drinking
water distribution system simulator. AWWA
Water Security Congress, Cincinnati, OH.
Umberg, K., N. Mix, B. Pickard, and C. Dangel, 2009.
Event Detection System Evaluation, Selection, and
Optimization. AWWA Water Security Congress,
Washington, DC.

-------
&EPA
PRESORTED STANDARD
POSTAGE & FEES PAID
EPA
PERMIT No. G-35
United States
Environmental Protection
Agency
Office of Research and Development
National Homeland Security
Research Center
Cincinnati, OH 45268
Official Business
Penalty for Private Use
$300
EPA/600/R-09/076
October 2009
www.epa.gov
Recycled/Recyclable
Printed with vegetable-based ink on
paper that contains a minimum of
50% post-consumer fiber content
processed chlorine free

-------