September 2001
NSF 01/08/EPADW395
Environmental Technology
Verification Report
Physical Removal of Giardia- and
Cryptosporidium-s\zed Particles in
Drinking Water
Rosedale Products, Inc.
Bag and Rigid Cartridge Filter System
Model GFS-302P2-150S-ESBB
Prepared by
®
NSF International
Under a Cooperative Agreement with
U.S. Environmental Protection Agency
-------
THE ENVIRONMENTAL TECHNOLOGY VERIFICATION
PROGRAM
&EPA
r ixwvjIvvm a
ETV
U.S. Emironmental Protection Agency NSF International
ETV Joint Verification Statement
TECHNOLOGY TYPE:
BAG AND CARTRIDGE FILTER USED IN DRINKING
WATER TREATMENT SYSTEMS
APPLICATION:
PHYSICAL REMOVAL OF GIARDIA- AND
CR YPTOSPORIDIUM-SIZED PARTICLES IN DRINKING
WATER
TECHNOLOGY NAME:
MODEL GFS-302P2 -150S-ESBB
BAG AND RIGID CARTRIDGE FILTER SYSTEM
COMPANY:
ROSEDALE PRODUCTS, INC.
ADDRESS:
3730 WEST LIBERTY STREET PHONE: (734)665-8201
ANN ARBOR, MICHIGAN 48106 FAX: (734) 665-2214
WEB SITE:
www.rosedaleproducts.com
EMAIL:
jima@rosedaleproducts.com
The U.S. Environmental Protection Agency (EPA) has created the Environmental Technology
Verification (ETV) Program to facilitate the deployment of innovative or improved environmental
technologies through performance verification and dissemination of information. The goal of the ETV
program is to further environmental protection by substantially accelerating the acceptance and use of
improved and more cost-effective technologies. ETV seeks to achieve this goal by providing high
quality, peer reviewed data on technology performance to those involved in the design, distribution,
permitting, purchase, and use of environmental technologies.
ETV works in partnership with recognized standards and testing organizations; stakeholders groups which
consist of buyers, vendor organizations, and permitters; and with the full participation of individual
technology developers. The program evaluates the performance of innovative technologies by developing
test plans that are responsive to the needs of stakeholders, conducting field or laboratory tests (as
appropriate), collecting and analyzing data, and preparing peer reviewed reports. All evaluations are
conducted in accordance with rigorous quality assurance protocols to ensure that data of known and
adequate quality are generated and that the results are defensible.
NSF International (NSF) in cooperation with the EPA operates the Drinking Water Treatment Systems
(DWTS) Pilot, one of 12 technology areas under ETV. The DWTS Pilot recently evaluated the
performance of a bag and cartridge system used in drinking water treatment system applications. This
verification statement provides a summary of the test results for the Rosedale Products, Inc. (RPI) Model
GFS-302P2-150S-ESBB Bag and Rigid Cartridge Filter System. Cartwright, Olsen and Associates, an
NSF-qualified field testing organization (FTO), performed the verification testing.
01/08/EPADW395 The accompanying notice is an integral part of this verification statement. September 2001
VS-i
-------
ABSTRACT
The verification testing of the RPI Model GFS-302P2-150S-ESBB Bag and Rigid Cartridge Filter System
occurred at the Minneapolis Municipal Water Works (MWW) facility during a 32-day verification test
period conducted between March and April 2000. The system employed a Model GD-PO-523-2 bag
filter element and a Model PL-520-PPP-141 rigid cartridge filter element. The source water was a blend
of untreated river water and finished water. The system was operated for 23 hours per day with a one-
hour stoppage. There were a total of 22 filter runs with an average low rate of 9.7 gpm. The
manufacturer specified 15 pounds per square inch (psi) as terminal headloss. Following a brief ripening
period during each filter run, on-line turbidity on average over twenty-two filter runs was 1.08 NTU
influent and 0.21 NTU effluent. Three fluorescent microsphere challenges were performed during three
filter runs, for a total of nine challenges. The challenges occurred at the beginning of the run, at roughly
the mid-point as determined by headloss, and then again at a point between 90% headloss and terminal
headloss. The number of microspheres added to the feed water during the nine challenges was
approximately 11,746 particles/mL. Fifty percent of the microspheres used were from a 3.4 |am
microsphere stock solution (further evaluation of the 3.4 |am stock solution indicated that the stock
solution actually contained microspheres with a mean size of approximately 3 |a,m) and the remaining
50% were 5 |am and 6 |a,m in size. Particle counters were used to measure the number of particles in the
feed and finished water, and samples were collected of the feed and finished water and analyzed by
microscopic enumeration. The RPI bag and cartridge system demonstrated 1.1 to 2.1 logio removal of
seeded microspheres (2.5-7.0 |am) based on the microscopic enumeration results, and 1.9 to 2.7 logio
removal of microspheres and indigenous particles sized 2.0 to 7.0 |am based on the on-line particle
counter data that was adjusted for the number of fluorescent microspheres added (as described later).
TECHNOLOGY DESCRIPTION
The system consists of two connected stainless steel filter housings. The first housing contained a Model
GD-PO-523-2 bag filter element. The second housing contained a Model PL-520-PPP-141 rigid cartridge
filter element (which replaced the Model GLR-PO-825-2 filter element used during Phase I initial
operations). Valves and other components are also made of stainless steel or of materials that will not
degrade in water. The flow through both the bag and cartridge filter is from inside to outside. The filter
housings are designed to accommodate a flow rate of 20 gpm, but were operated at 10 gpm during the
verification testing to limit possible filter loadings by high turbidity levels. The system is designed to
operate with surface waters that have turbidity levels of 1 NTU or less and with pressures of less than 60
psi. This testing used 15 psi as a terminal pressure loss value. Liquid chlorine bleach (sodium
hypochlorite) was added during the verification testing to limit any microbial growth within the filters.
The bleach-metering pump was stopped during microsphere challenge events.
The system is designed to act as a final barrier and to capture/contain particles in the size range of C.
parvum (approximately 3-7 |am). Since G. lamblia cysts are larger than C.parvum oocysts, it is assumed
that if the smaller oocysts are contained, the larger cysts will be contained at least the same level1.
Accordingly, while this system is applicable to G. lamblia removal as well as C. parvum removal, focus
was placed on C. parvum sized particles.
The filter system is suited to small public water systems where water treatment plant operators typically
have minimal technical training. The system itself requires no additional chemicals beyond normal
disinfection and relatively limited on-site supervision, for tasks such as reading pressure gauges and
changing filters. No special licensing is required for the use of the filters. Training in bag/element
1 Niemiinski, Eva C. Removal of Cryptosporidium and Giardia through Conventional Water Treatment and Direct
Filtration. EPA/600/SR-97/025, 1997.
01/08/EPADW395 The accompanying notice is an integral part of this verification statement. September 2001
VS-ii
-------
replacement is minimal and is explahed in the Operations and Maintenance (O&M) Manual, as supplied
by the manufacturer (see Verification Report).
VERIFICATION TESTING DESCRIPTION
Test Site
The host site for this demonstration was the Minneapolis Municipal Water Works (MWW) located in
Fridley, Minnesota, a suburb adjacent to and directly north of the City of Minneapolis. The testing
equipment was located in Pump House #5. Pump House #5 is the intake point from the Mississippi river.
Source Water
The source water for the verification testing was a blend of raw water from the Mississippi River and
finished water from the MWW treatment plant. Water at the MWW is softened with lime and treated
with alum for removal of color and turbidity. Powdered activated carbon and occasionally potassium
permanganate are also added to remove taste and odor. The water is then treated with carbon dioxide to
lower the pH and stabilize the remaining hardness prior to being pumped to one of two filtration plants.
At the filtration plant, chlorine and ammonia are added for initial disinfection, fluoride is added for tooth
decay prevention and ferric chloride is added as a coagulant to remove remaining color and turbidity. The
water then enters a series of coagulation/sedimentation basins after which the water is filtered with single,
dual or mixed media filters. Blended poly/ortho phosphate is later added as a corrosion control/inhibitor.
The water is post-chlorinated for final adjustment of the disinfectant residual before being fed into the
reservoirs and pumped into the distribution system. Finished water was blended with raw river water to
obtain a turbidity level between 1-3 NTU.
Methods and Procedures
The verification test was divided into tasks that evaluated the system's treatment performance,
specifically its ability to physically remove polystyrene microspheres in the size range of 3 to 6 |a,m from
the feed water, and documented the system's operational parameters.
Prior to the 32-day verification test, cartridge filter elements underwent filter variability testing to
evaluate the variations between and within filter production lots. Phase I was designed to determine
variations within a production lot number of Model GLR-PO-825-2 cartridge filter elements. Based on
the results of the first phase of variability testing, Rosedale chose to change the cartridge filter to the
Model PL-520-PPP-141 cartridge filter for the remainder of the testing. Phase II variability testing was
designed to show variations between production lots. Each phase included 10 days of system operation
with 23 hours of operation and one hour off line (no flow).
The 32-day verification test was performed to evaluate the total number of gallons treated per filter
system (bag and cartridge) and the finished water characteristics. The bags and cartridges were replaced
if terminal headloss (15 psi) or turbidity breakthrough, as established by the manufacturer, was reached.
Water quality parameters monitored during the verification test included: pH, temperature, turbidity,
particle counts, free chlorine residual, alkalinity, total hardness, total organic carbon (TOC), ultraviolet
absorbance (UVA) at 254 nanometers (mil), true color, aluminum, iron, manganese, algae, and total
coliforms. Laboratory analyses were performed in accordance with the procedures and protocols
established in Standard Methods for the Examination of Water and Wastewater, 19th Edition (SM) or
EPA-approved methods as listed in the report
During the testing, microspheres in the size range of 3 to 6 |am were injected into the pilot installation
feed water via a metering pump to demonstrate 3+ logio removal. Fifty percent of the microspheres used
01/08/EPADW395 The accompanying notice is an integral part of this verification statement. September 2001
VS-iii
-------
were from a 3.4 |am microsphere stock solution (further evaluation of the 3.4 |am stock solution indicated
that the stock solution actually contained microspheres with a mean size of approximately 3 |a,m) and the
remaining 50% were 5 |am and 6 |am in size. Three microsphere challenges were performed during three
filter runs, for a total of nine challenges. The challenges occurred at the beginning of the run, at roughly
the mid-point as determined by headloss, and then again at a point between 90% headloss and terminal
headloss. The feed and finished water were evaluated for the presence of microspheres by using on-line
particle counters and enumeration of samples collected with hemacytometer techniques and/or membrane
filtration.
Operating conditions were documented during each day of verification testing, including: filter flow rate,
filter headloss, hours of operation, filtered water production, and frequency of filter replacement.
VERIFICATION OF PERFORMANCE
Filter Element Variability
Phase I filter element variability testing began on June 24, 1999, with three Model GLR-PO-825-2
cartridge filters from the same production lot (No. 88-4546). The bag filters, used as pre-filters within the
filter train, all were from the same manufacturing lot. The flowrate was 20 gpm per filter and the target
turbidity level was achieved by blending raw river water with finished water to approximately 3.0 NTU.
By the second day of Phase I, the bags and cartridge filters had been replaced once and the filters were
again approaching terminal headloss. Accordingly, the system was shut down on June 25 to reevaluate
the operating parameters. After discussions with the manufacturer, it was decided to reduce influent
turbidity to 1 NTU and decrease the flow rate to 10 gpm to reduce rate of filter loading. It was also
decided that only finished drinking water would serve as the feed water when the equipment was not
attended by an operator to avoid reaching terminal headloss during unmanned periods. Due to concerns
expressed by the manufacturer regarding the cartridges from production lot No. 88-4546, the
manufacturer provided replacement cartridges from a different production lot, No. 6-2-99. Phase I testing
recommenced on June 29 and ended July 7, 1999. Bag and cartridge filters were replaced twice during the
remaining portion of Phase I. Based on the results of Phase I, the manufacturer elected to address
concerns pertaining to the manufacturing process of the Model GLR-PO-825-2 cartridge filter element.
Subsequently, for Phase II of filter element variability testing, the manufacturer provided cartridge filter
elements with a different model number (PL-520-PPP-141) and internal seals within the filter housing.
Phase II of the filter element variability testing occurred between January 10 through 20, 2000 with
Model PL-520-PPP141 cartridge filters from 3 different production lots (Numbers 990541-5, 990541-4,
990541-3). Again, the bag filters used as pre-filters within the filter train were from the same
manufacturing lot. Bag and cartridge filters were replaced twice during Phase II. Headlosses at time of
filter replacement on January 13 were 12 psi, 8 psi, and 15 psi respectively for filter trains #1, #2, and #3.
Corresponding logio reductions of indigenous particles sized 2 to 15 |am as measured by particle counters
were 1.4, 1.2, and 1.6. Head losses at time of filter replacement on January 17 were 12 psi, 8 psi, and 9
psi respectively for filter trains #1, #2, and #3 and corresponding 2-15 |a,m particle count log10 reductions
were 1.5, 1.5, and 1.6. Head losses at time of shut-down on January 20 were 6 psi, 6 psi, and 5.5 psi
respectively for filter trains #1, #2, and #3. Corresponding 2-15 |a,m particle count logio reductions were
1.4, 0.81, and 1.4. Filter train #2 demonstrated comparatively poor particle reduction performances
during Phase II. This was attributed to a faulty pressure differential gauge that bypassed feed water into
the filtered water stream. Due to the limited number of filters evaluated within each production lot,
conclusions regarding variation in filter performance between production lots cannot be offered with any
degree of certainty.
01/08/EPADW395 The accompanying notice is an integral part of this verification statement. September 2001
VS-iv
-------
Operation and Maintenance
The verification testing for the system began on March 7, 2000, and ended its 32-day period on April 20,
2000. The system was operated for 23 hours per day with a one-hour stoppage. There were a total of 22
filter runs (bag and cartridges replaced at the start of each filter run unless otherwise noted). The average
flow rate over the 22 filter runs was 9.7 gpm. The average terminal headloss, volume of water produced,
and duration of the 22 filter runs are summarized in the following table:
Operating Data - 22 Filter Runs (March 7 - April 20, 2000)
Filter Run
Terminal
Headloss
(psi)
Water Produced
Filter Run
Number
(Gallons)
Duration (hours)
Average
16.3
22,789
38.04
Minimum
11.0
10,980
19.25
Maximum
25.5
74,173
135.25
Std Dev.
3.6
15,434
27.76
95% Confidence Interval
14.8,17.9
16,340,29,239
25.88, 50.18
The manufacturer supplied O&M Manual illustrates the equipment and shows the proper configuration of
the housings. The system start up and element replacement procedures are instructive and thorough. A
parts list is included.
Microsphere Removal
The fluorescent microsphere challenge was performed between April 16 and 20, 2000. Particle counters
were used to measure the number of particles in the feed and finished water, and samples were collected
of the feed and finished water and analyzed by microscopic enumeration and a laboratory optical particle
counter. The system demonstrated 1.1 to 2.1 logio removal of the seeded microspheres based on the
microscopic evaluations by Huffman Environmental Consulting; however, it was noted by the laboratory
that, upon visual inspection, a considerable number of microspheres were smaller than 3 |am. The 3.4 |am
microsphere stock solution obtained from Bangs Laboratories was reanalyzed by Bangs and the results
indicated that the true particle median size was not 3.4 |am as specified, but was actually 2.98 |am with a
standard deviation of 0.66 |am or 21.2%. Further evaluation of the particle count data indicated that 1.9 to
2.7 logio removals of particles sized 2 to 7 |am were achieved during the fluorescent microsphere
challenge testing based on normalized on-line particle counter data which involved adding the number of
seeding microspheres (approximately 11,746 particles/mL) to the source water's indigenous material
particle counter value and comparing with the effluent particle counter value (details regarding the
normalized particle count data are described in the Verification Report). The duplicate set of samples
collected during the microsphere challenge were sent to Micro Measurement Laboratories, Inc. for
analysis by a laboratory optical particle counter called an Accusizer. Logio reductions calculated with the
use data as analyzed with the Accusizer were not performed because an analysis of the control sample
container demonstrated a suspected level of contamination (approximately 315 particles/mL). However,
influent particle count data as provided from these analyses were helpful in validating influent
particle/microsphere concentrations used to calculate logio reductions of particles/microspheres sized
between 2|am and 7|am. Results are summarized in the following table:
01/08/EPADW395 The accompanying notice is an integral part of this verification statement. September 2001
VS-v
-------
Hi o
Microscopic
Normalized On-Line
Seeding
Enumeration (2-7 |_im
Particle Counters (2-7 |_im indigenous
microspheres)
particles and microspheres)
First Challenae Run
No headloss
1.1
1.9
Midpoint
2.1
2.3
90% headloss
1.8
2.0
Second Challenae Run.
No headloss
1.5
1.9
Midpoint
2.1
2.7
90% headloss
1.9
2.6
Third Challenae Run.
No headloss
1.5
2.0
Midpoint
1.8
2.7
90% headloss
1.6
2.7
Following the 50% headloss seeding challenges, the flow through the system was interrupted for a brief
interval and then restarted to determine the level of particle sloughing following resumption of flow.
Particles were sloughed for less than three recording cycles of the particle counter, or less than three
minutes. The results are discussed more fully in the Verification Report but point to the necessity for a
brief filter to waste cycle following an interruption in flow.
Water Quality
The following table summarizes the results of the influent and effluent samples collected during the
verification testing period.
Feed/Filtered Water Quality (March 7-April 20,2000)
Parameter
#of
Samples
Average
Minimum
Maximum
Standard
Deviation
95% Confidence
Interval
38/0
7.3/-
3.9/-
11.0/-
2.2/-
6.7, 8.0/-
37/0
8.5/-
8.0/-
8.9/-
0.2/-
8.4, 8.5/-
7/7
70/66
55/54
110/100
18/16
57, 84/55,78
7/7
24/2
<1/<1
110/6
40/3
<1,54/<1,4
7/7
94/95
82/82
130/130
16/16
82,107/
83, 107
7/7
7.8/7.5
6.8/6.4
11/8.8
1.4/0.8
6.7, 8.9/
6.9, 8.1
7/7
14/10
10/5
25/15
6/4
10,18/7, 13
7/7
0.140/0.130
0.180/0.109
0.229/0.156
0.042/0.017
0.109, 0.171/
0.117, 0.143
continuous
1.08/0.21
0.68/0.17
1.46/0.26
0.20/0.02
0.98, 1.16/
0.20, 0.22
continuous
7,329/91
3,784/39
10,056/300
1,737/59
6567, 8090/
65, 117
7/7
0.1/0.1
<0.1/<0.1
0.4/0.6
0.1/0.2
<0.1,0.2/
<0.1,0.3
7/7
0.02/0.1
<0.010.01
0.04/0.04
0.01/0.01
0.01,0.03/
<0.01,0.02
27/0
1.4/-
0.7/-
3.5/-
0.82/-
1.1,1.7/-
27/0
0.6/-
0.1/-
2.5/-
0.6/-
0.4, 0.8/-
Temperature (°C)
pH
Total Alkalinity (mg/L)
Total Coliform (cfu/lOOmL)
Total Hardness (mg/L)
TOC (mg/L)
True Color (TCU)
UVA254 (cm4)
On-line Turbidity (NTU)*
On-line Total Particle Counts
(#/mL)*
Iron (mg/L)
Manganese (mg/L)
Total Chlorine (mg/L)
Free Chlorine (mg/L)
Note: All calculations involving
*Measurements are the average
results with below detection limit values used half the detection limit in the calculation,
of the filter run averages.
01/08/EPADW395 The accompanying notice is an integral part of this verification statement. September 2001
VS-vi
-------
Turbidity removals were consistent and generally good throughout the verification period. Following a
brief ripening period, the average on-line turbidity over the 22 filter runs was 1.08 NTU for the feed and
0.21 NTU in the filtered water. No algae were detected in the filtered water samples.
Original Signed by
E. Timothy Oppelt 9/20/01
E. Timothy Oppelt Date
Director
National Risk Management Research Laboratory
Office of Research and Development
United States Environmental Protection Agency
Original Signed by
Gordon Bellen 9/22/01
Gordon Bellen Date
Vice President
Federal Programs
NSF International
NOTICE: Verifications are based on an evaluation of technology performance under specific,
predetermined criteria and the appropriate quality assurance procedures. EPA and NSF make no
expressed or implied warranties as to the performance of the technology and do not certify that a
technology will always operate as verified. The end user is solely responsible for complying with
any and all applicable federal, state, and local requirements. Mention of corporate names, trade
names, or commercial products does not constitute endorsement or recommendation for use of
specific products. This report is not a NSF Certification of the specific product mentioned herein.
Availability of Supporting Documents
Copies of the ETV Protocol for Equipment Verification Testing for Physical Removal of
Microbiological and Particulate Contaminants dated May 14, 1999, the Verification
Statement, and the Verification Report (NSF Report # 01/08/EPADW395) are available
from the following sources:
(NOTE: Appendices are not included in the Verification Report. Appendices are
available from NSF upon request.)
1. Drinking Water Treatment Systems ETV Pilot Manager (order hard copy)
NSF International
P.O. Box 130140
Ann Arbor, Michigan 48113-0140
2. NSF web site: http://www.nsf.org/etv (electronic copy)
3. EPA web site: http://www.epa.gov/etv (electronic copy)
01/08/EPADW395 The accompanying notice is an integral part of this verification statement. September 2001
VS-vii
-------
September 2001
Environmental Technology Verification Report
Physical Removal of Giardia- and Cryptosporidium-sized Particles in
Drinking Water
Rosedale Products, Inc.
Bag and Rigid Cartridge Filter System
Model GFS-302P2-150S-ESBB
Prepared for:
NSF International
Ann Arbor, Michigan 48105
Prepared by
Philip C. Olsen
Cartwright, Olsen and Associates, LLC
Cedar, Minnesota 55011
Under a cooperative agreement with the U.S. Environmental Protection Agency
Jeffrey Q. Adams, Project Officer
National Risk Management Research Laboratory
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
-------
Notice
The U.S. Environmental Protection Agency (EPA) through its Office of Research and
Development has financially supported and collaborated with NSF International (NSF) under
Cooperative Agreement No. CR 824815. This verification effort was supported by the Drinking
Water Treatment Systems Pilot operating under the Environmental Technology Verification
(ETV) Program. This document has been peer reviewed and reviewed by NSF and EPA and
recommended for public release.
11
-------
Foreword
The following is the final report on an Environmental Technology Verification (ETV) test
performed for NSF International (NSF) and the United States Environmental Protection Agency
(EPA) by Cartwright, Olsen and Associates, LLC, in cooperation with Rosedale Products Inc.
The test was conducted during March and April of 2000 at the Minneapolis Municipal Water
Works Pump House #5 in Fridley, Minnesota USA.
Throughout its history, the EPA has evaluated the effectiveness of innovative technologies to
protect human health and the environment. A new EPA program, the Environmental
Technology Verification Program (ETV) has been instituted to verify the performance of
innovative technical solutions to environmental pollution or human health threats. ETV was
created to substantially accelerate the entrance of new environmental technologies into the
domestic and international marketplace. Verifiable, high quality data on the performance of new
technologies is made available to regulators, developers, consulting engineers, and those in the
public health and environmental protection industries. This encourages more rapid availability
of approaches to better protect the environment.
The EPA has partnered with NSF, an independent, not-for-profit testing and certification
organization dedicated to public health, safety and protection of the environment, to verify
performance of small package drinking water systems that serve small communities under the
Drinking Water Treatment Systems (DWTS) ETV Pilot. A goal of verification testing is to
enhance and facilitate the acceptance of small package drinking water treatment equipment by
state drinking water regulatory officials and consulting engineers while reducing the need for
testing of equipment at each location where the equipment's use is contemplated. NSF will meet
this goal by working with manufacturers and NSF-qualified Field Testing Organizations (FTO)
to conduct verification testing under the approved protocols.
The ETV DWTS is being conducted by NSF with participation of manufacturers, under the
sponsorship of the EPA Office of Research and Development, National Risk Management
Research Laboratory, Water Supply and Water Resources Division, Cincinnati, Ohio. It is
important to note that verification of the equipment does not mean that the equipment is
"certified" by NSF or "accepted" by EPA. Rather, it recognizes that the performance of the
equipment has been determined and verified by these organizations for those conditions tested by
the FTO.
111
-------
Table of Contents
Section Page
Verification Statement VS-i
Title Page i
Notice ii
Foreword iii
Table of Contents iv
Abbreviations and Acronyms x
Definitions xii
Acknowledgments xiii
Chapter 1: Introduction 1
1.1 ETV Purpose and Program Operation 1
1.2 Testing Participants and Responsibilities 1
1.2.1 NSF International 2
1.2.2 Field Testing Organization 2
1.2.3 Manufacturer 3
1.2.4 Analytical Laboratories 3
1.2.5 Minneapolis Municipal Water Works 4
1.2.6 U.S. Environmental Protection Agency 4
1.3 Verification Testing Site 5
1.3.1 Source Water 5
1.3.2 Pilot Effluent Discharge 7
Chapter 2: Equipment Description and Operating Processes 8
2.1 Historical Background 8
2.2 Equipment Description 9
2.2.1 Equipment Installation 12
2.3 Operating Process 13
Chapter 3: Methods and Procedures 14
3.1 Experimental Design 14
3.1.1 Objectives 14
3.1.1.1 Evaluation of Stated Equipment Capabilities 14
3.1.1.2 Evaluation of Equipment Performance Relative To Water Quality
Regulations 15
3.1.1.3 Evaluation of Operational and Maintenance (O&M) Requirements 15
3.1.1.4 Evaluation of Equipment Characteristics 15
3.2 Initial Operations 16
3.2.1 Characterizati on of F eed Water 16
3.2.2 Initial Test Runs 17
3.2.3 Filter Element Variability Testing 17
3.3 Verification Task Procedures 18
3.3.1 Task 1 - Verification Testing Runs and Routine Equipment Operation 19
3.3.2 Task 2 - Feed and Finished Water Quality Characterization 19
iv
-------
Table of Contents (continued)
Section Page
3.3.3 Task 3 - Documentation of Operating Conditions and Treatment Equipment
Performance 20
3.3.4 Task 4 - Microbiological Contaminant Removal 21
3.3.4.1 Preparation of Microbial Surrogate Doses 22
3.3.4.2 Description of Fluorescent Microsphere Seedings 23
3.3.4.3 Data Evaluation 24
3.4 Data Recording, Communications, Logistics and Data Handling Protocol 25
3.4.1 Procedures 25
3.4.1.1 Field Notebooks 25
3.4.1.2 Photographs 26
3.4.1.3 Chain of Custody 26
3.4.1.4 On-line Measurements 26
3.4.1.5 Spreadsheets 26
3.5 Calculation of Data Quality Indicators 27
3.5.1 Representativeness 27
3.5.2 Statistical Uncertainty 27
3.5.3 Accuracy 27
3.5.4 Precision 28
3.6 Verification Testing Schedule 29
3.7 Field Operations Procedures 29
3.7.1 Operations 29
3.7.2 Analytical Equipment 29
3.8 QA/QC Procedures 30
3.8.1 QA/QC Verifications 30
3.8.2 On-Site Analytical Methods 31
3.8.2.1 pH 31
3.8.2.2 Temperature 32
3.8.2.3 Turbidity 32
3.8.2.4 Particle Counting 32
3.8.2.5 Particle Free Water 33
3.8.3 Off-Site Analysis For Chemical and Biological Samples 34
3.8.3.1 Organic Parameters, Total Organic Carbon and UV Absorbance 34
3.8.3.2 Microbial Samples: Coliform and Algae 34
3.8.3.3 Inorganic Samples 34
3.8.3.4 Microspheres 35
3.8.3.5 True Color 35
Chapter 4: Results and Discussion 36
4.1 Introducti on 36
4.2 Initial Operations Period Results 36
4.2.1 Characterizati on of F eed Water 36
4.2.2 Initial Test Runs 37
4.2.3 Filter Element Variability Testing 37
4.3 Verification Testing Results and Discussions 41
v
-------
Table of Contents (continued)
Section Page
4.3.1. Task 1 - Verification Testing Runs And Routine Equipment Operation 41
4.3.2 Task 2 - Influent and Effluent Water Quality Characterization 41
4.2.3 Task 3 - Documentation Of Operating Conditions And Treatment Equipment
Performance 44
4.3.4 Task 4 - Microbiological Contaminant Removal 49
4.3.4.1 Non-Fluorescing Microsphere Challenge Results 49
4.3.4.2 Fluorescing Microsphere Challenge Results 51
4.3.4.3 On-line Particle Counter Analysis During Fluorescing Microsphere
Challenge 52
4.3.4.4 Microscopic Analysis 56
4.3.4.5 Laboratory Optical Particle Analysis 64
4.3.4.6 Discussion of Results of Fluorescent Microsphere Challenges 64
4.3.4.7 Stop/Start Event Evaluation 69
4.4 Equipment Characteristics Results 73
4.4.1 Qualitative Factors 73
4.4.1.1 Filter Element Replacement 73
4.4.1.2 Head Loss 74
4.4.1.3 Other Operational Factors 74
4.4.1.4 Evaluation of O&M Manual 75
4.4.2 Quantati ve F actors 75
4.4.2.1 Filter Elements Replacement 75
4.4.2.2 Anomalous Conditions That Require Operator Response 75
4.4.2.4 Length of Operating Cycle 75
4.5 QA/QC Results 76
4.5.1 Data Correctness 76
4.5.1.1 Representativeness 76
4.5.1.2 Statistical Uncertainty 76
4.5.1.3 Accuracy 77
4.5.1.4 Precision 77
4.5.2 Daily QA/QC Results 77
4.5.3 Bi-Weekly QA/QC Verification Results 78
4.5.4 Results Of QA/QC Verifications At The Start Of Each Testing Period 78
4.5.5 Analytical Laboratory QA/QC 83
4.5.6 Microbiological Laboratory QA/QC 83
4.5.7 QA/QC Procedures for Accusizer Measurements 83
4.6 Limitations 84
Chapter 5: References 86
Tables Page
Table 1-1. GFS Filter System Feed Water Quality (March 7 to April 20, 2000) 7
Table 3-1. Analytical Data Collection Schedule 20
Table 3-2. Operating Data 21
vi
-------
Table of Contents (continued)
Table Page
Table 4-1. Phase I Initial Operations Filter Run Summary 38
Table 4-2. Phase II Variability Testing RPI GFS Filter System Filter Run #1 Particle Count &
Turbidity Results 39
Table 4-3. Phase II Variability Testing RPI GFS Filter System Filter Run #2 Particle Count &
Turbidity Results 39
Table 4-4. Phase II Variability Testing RPI GFS Filter System Filter Run #3 Particle Count &
Turbidity Results 40
Table 4-5 Phase II RPI GFS Filter System Filter Operating Data 41
Table 4-6. Influent Water Quality (March 7 - April 20, 2000) 42
Table 4-7. Influent Water Samples Algae (CFU/lOOmL) 42
Table 4-8. Effluent Water Quality (March 7 - April 20, 2000) 43
Table 4-9. Operating Parameters (Summary of 22 Filter System Runs) 44
Table 4-10. Filter Run Averages For Verification Period (March 7 - April 20) 45
Table 4-11. Stop/Start Stabilization Time 49
Table 4-12. Monosphere Manufacturer Specification 50
Table 4-13. Average 3-7 |j,mParticle Counts Non-Fluorescing Microsphere Challenge Filter
Event #1 50
Table 4-14. Average 3-7 |j,m Particle Counts Non-Fluorescing Microsphere Challenge Filter
Event #3 51
Table 4-15. 3-7 |j,m Influent Particles During Fluorescent Microsphere Challenge (Filter Runs
#20, 21 & 22) 53
Table 4-16. 3-7 |j,m Effluent Particles During Fluorescent Microsphere Challenge (Filter Runs
#20, 21 & 22) 54
Table 4-17. Influent Microscopic Analysis Results Of Fluorescent Challenge Events 57
Table 4-18. Effluent Microscopic Analysis Results Of Fluorescent Challenge Events 57
Table 4-19. Calculation of Percent Removal, Fluorescent Challenge Events 58
Table 4-20. Summary of Number Weight Cumulative Distribution by Histogram 59
Table 4-21. On-line Particle Counts and Added Microsphere Logio Reductions (2.0-7.0|j,m)....61
Table 4-22. Microshcopic Microshpere Counts and Logio Reductions (2.5-7.0|j,m) 65
Table 4-23. On-line Particle Counts Plus Microshperes Logio Reductions (2.0-7.0|j,m) 66
Table 4-24. Calculation of Filter Efficiency With 1st and 2nd Seeding Events and One
Stop/Start Event 69
Table 4-25. Calculation of Filter Efficiency With 2nd Seeding Event and One Stop/Start
Event 69
Figures Page
Figure 2-1. GFS Filter System Schematic 12
Figure 4-1. Influent Turbidity & Gallons Per Filter Run 46
Figure 4-2. RPI GFS Average Filter Influent & Effluent Turbidity 47
Figure 4-3. RPI GFS Logio Removal For Indigenous Particle Sized 2-15|j,m 47
Figure 4-4. RPI GFS Filter Average Filter Run Pressure Drop 48
vii
-------
Table of Contents (continued)
Figures Page
Figure 4-5. Representative Particle Count During Fluorescent Microsphere Seeding, Run #20,
1st Seed 53
Figure 4-6. Fluorescent Microsphere Challenge Event #1 - On-Line Particle Count vs. Filter
Run Time 55
Figure 4-7. Fluorescent Microsphere Challenge Event #2 - On-Line Particle Count vs. Filter
Run Time 55
Figure 4-8. Fluorescent Microsphere Challenge Event #3 - On-Line Particle Count vs. Filter
Run Time 56
Figure 4-9. Historgram Particle Sizing 60
Figure 4-10. Fluorescent Microsphere Challenge Event #1 - Logio Reductions Performance
Throughout the Filter Run with Indigenous Particles 62
Figure 4-11. Fluorescent Microsphere Challenge Event #2 - Logio Reductions Performance
Throughout the Filter Run with Indigenous Particles 63
Figure 4-12. Fluorescent Microsphere Challenge Event #3 - Logio Reductions Performance
Throughout the Filter Run with Indigenous Particles 63
Figure 4-13. Fluorescent Microsphere Challenge Event #1 - Logio Reductions Comparison
between Challenges Analysis and Reductions Throughout the Filter Run with
Indigenous Particles 67
Figure 4-14. Fluorescent Microsphere Challenge Event #2 - Logio Reductions Comparison
between Challenges Analysis and Reductions Throughout the Filter Run with
Indigenous Particles 67
Figure 4-15. Fluorescent Microsphere Challenge Event #3 - Logio Reductions Comparison
between Challenges Analysis and Reductions Throughout the Filter Run with
Indigenous Particles 68
Figure 4-16. Rosedale Filter Run #20 Effluent 3-7 |j,m Particle Count Stop/Start 70
Figure 4-17. Rosedale Filter Run #21 Effluent 3-7 |j,m Particle Count Stop/Start 70
Figure 4-18. Rosedale Filter Run #22 Effluent 3-7 |j,m Particle Count Stop/Start 71
Figure 4-19. Rosedale Filter Run #20 Effluent 3-7 |j,m Turbidity Stop/Start 71
Figure 4-20. Rosedale Filter Run #21 Effluent 3-7 |j,m Turbidity Stop/Start 72
Figure 4-21. Rosedale Filter Run #22 Effluent 3-7 |j,m Turbidity Stop/Start 72
Figure 4-22. Verification of 3|j,m Influent Particles 79
Figure 4-23. Verification of Mix of 3, 10 & 15-|j,m Influent Particles 79
Figure 4-24. Verification of 3-|j,m Effluent Particles 80
Figure 4-25. Verification of 10-|j,m Effluent Particles 80
Figure 4-26. Verification of 15-|j,m Effluent Particles 81
Figure 4-27. Verification of 3, 10 & 15-|j,m Effluent Particles 81
Photographs Page
Photograph 1 - RPI GFS Filter System 10
Photograph 2 - RPI GFS Filter System on site at the Minneapolis Water Works 11
viii
-------
Table of Contents (continued)
Appendices
A.
Manufacturer's Operation and Maintenance Manual
B.
Data Logbook
C.
Data Sheets
D.
Laboratory Chain of Custody Forms
E.
Laboratories Reports
F.
Filter Runs Data
G.
Equipment QA/QC Documentation
H.
Log Reduction Calculations
I.
QA/QC Documentation Tables for On-Site Measurements
IX
-------
Abbreviations and Acronyms
APHA
American Public Health Association
ASTM
American Society for Testing and Materials
AWWA
American Water Works Association
AWWARF
American Water Works Association Research Foundation
cfh
Cubic feet per hour
°C
Degrees Celsius
cfm
Cubic feet per minute
CFU
Colony-forming units
COA
Cartwright, Olsen and Associates, LLC
C. parvum
Cryptosporidium parvum
CV
Coefficient of Variation
DI
Deionized (demineralized) water
E. coli
Escherichia coli
EPA
U.S. Environmental Protection Agency
ETV
Environmental Technology Verification
°F
Degrees Fahrenheit
Flowrates
Flowrates are expressed as US gallons per minute (gpm)
FOD
Field Operations Document
FTO
Field Testing Organization
gallons
Gallons are expressed as US gallons, 1 gal = 3.785 liters =1.2
gallons imperial 1 gallon imperial = 4.54 liters = .833 gallons US
GFS Filter System
Rosedale Products GFS-302P2-150S-ESBB Rigid Cartridge Filter
System
G. lamblia
Giardia lamblia
gpm
Gallons per minute
hp
Horsepower
ICR
Information Collection Rule
IESWTR
Interim Enhanced Surface Water Treatment Rule
Log
Logarithm to the base 10
Ln
Logarithm to the base e
|j,m
Micron = 10"6 meter
mgd
Million gallons per day
mg/L
Milligram per liter
mL
Milliliter
MML
Micro Measurement Laboratories
MPA
Microscopic Particulate Analysis
MWW
Minneapolis Water Works
NIST
National Institute of Standards and Technology
NSF
NSF International, formerly known as National Sanitation
Foundation
NTU
Nephelometric turbidity unit
O&M
Operation & Maintenance
(oo)cyst
Will be used to refer to both cysts and oocysts when used together
DWTS
Drinking Water Treatment Systems
X
-------
PFW
pH
PLC
PQL
psi
Psig
QA/QC
RPD
RPI
RPZ
SM
SWTR
TCU
TOC
Ten State's Standards
USP
USGS
WEF
Particle Free Water
A measure of the degree of the acidity of the alkalinity of the
solution as measured on a scale of 0 to 14.
Programmable Logic Computer
Practical Quantification Limit
Pounds per square inch
Pounds per square inch gauge
Quality Assurance/Quality Control
Relative percent difference
Rosedale Products Inc.
Reduced Pressure Zone
Standard Methods for the Examination of Water and Wastewater
Surface Water Treatment Rule
Total Color Units
Total Organic Carbon
Great Lakes-Upper Mississippi River Board of State Public Health
and Environmental Managers, Recommended Standards for Water
Works
United States Pharmacopeia
U.S. Geological Survey
Water Environment Federation
XI
-------
Definitions
Bag Filter
Cartridge Filters
Filtration
Predisinfection
Prefiltration
A disposable, quickly replaceable fabric filter, normally non-rigid
and contained either singly or in multiples within a pressure vessel.
The flow of water is normally from inside to outside. The bags can
be designed for a wide variety of filter applications and are
commonly used without coagulating or pre-coat chemicals. Those
designed for protozoan (oo)cyst capture have pore sizes that are
uniform and while small enough to contain the (oo)cysts will pass
bacteria, viruses and fine colloids.
Rigid, or semi-rigid, disposable fabric or polymer elements that
like the bag filter can be single or grouped into a filter pressure
vessel. Unlike bag filters the common flow for cartridges is from
the outside to the inner core of the filter. Pore sizes can be
manufactured in many nominal or absolute sizes, with the pressure
losses increasing as the pores decrease. As with bag filters,
unnecessarily small pore sizes contribute to more rapid loading,
pressure losses and thus more frequent element exchanges.
Removal of particulate contaminants by flow through a porous
media. Media can be granulated particles such as sand or coal, or
fabric, fiber or membrane. Bag and cartridge filters are commonly
fabric made from synthetic fibers.
Chemical disinfection of the water prior to passage through the
filter. This is often done to limit biofilm formation on the filter
element that might limit effectiveness or foreshorten filter runs,
and to assure a sanitary supply.
Coarse, often backwashable granular media filtration and
occasionally cartridge filtration or both prior to the bag or cartridge
to eliminate larger material in the water stream, thus limiting the
bag or cartridge filter element to the removal of finer particles in
the size range of the (oo)cyst. Prefiltration reduces the number of
more costly bag exchanges.
Xll
-------
Acknowledgments
The Field Testing Organization, Cartwright, Olsen & Associates (COA), was responsible for all
elements in the testing sequence, including collection of samples, calibration and verification of
instruments, data collection and analysis, data management, data interpretation and the
preparation of this report.
Cartwright, Olsen & Associates, LLC
19406 East Bethel Blvd.
Cedar, Minnesota 55011
(763)434-1300
Fax (763)434-8450
Contact: Philip C. Olsen
Email: p.olsen@ix.netcom.com
The laboratory that conducted the analytical work of this study was:
Spectrum Labs Inc.
301 West County Road E2
St. Paul, MN 55112
Phone: (651) 633-0101
Fax (651)633-1402
Contact: Gerard Herro, Laboratory Manager
E-mail: gherro@spectrum-labs.com
The hemacytometer and microscopic analyses of fluorescent microspheres was performed by:
Debra Huffman Environmental Consulting
6762 Millstone Drive
New Port Richey, FL 34655
Phone: (727) 553-3946
Fax: (727) 893-1189
Contact: Debra Huffman, Ph.D.
E-mail: dhuffman@marine.usf.edu
The Manufacturer of the Equipment was:
Rosedale Products, Inc.
3730 West Liberty Rd.
Ann Arbor Michigan 48106
Phone: (734) 665-8201
Fax: (734) 665-2214
Contact Person: Jim Arnold, Operations Manager
Email: j arnold@rosedaleproducts. com
Xlll
-------
COA wishes to thank NSF International, especially Mr. Bruce Bartley, Project Manger, and
Carol Becker and Kristie Wilhelm, Environmental Engineers, for providing guidance and
program management.
Jim Arnold, Operations Manager, Daniel Morosky, Vice President, Marketing, and John D.
Busch, Product Development and Research Division, Rosedale Products, Inc. are to be
commended for providing the treatment system and the excellent technical and product expertise.
Our gratitude to the Minneapolis Municipal Water Works staff for their generous cooperation
and hospitality during the pilot operation. We especially wish to thank the personnel at Pump
House #5 where the testing was performed.
COA also wishes to thank the Minnesota Department of Health, Drinking Water Protection for
their invaluable analytical and operational assistance, especially Gerald Smith, P.E., Public
Health Engineer, and Anita C. Anderson, Public Health Engineer.
xiv
-------
Chapter 1
Introduction
1.1 ETV Purpose and Program Operation
The U.S. Environmental Protection Agency (EPA) has created the Environmental Technology
Verification (ETV) Program to facilitate the deployment of innovative or improved
environmental technologies through performance verification and dissemination of information.
The goal of the ETV program is to further environmental protection by substantially accelerating
the acceptance and use of improved and more cost-effective technologies. ETV seeks to achieve
this goal by providing high quality, peer reviewed data on technology performance to those
involved in the design, distribution, permitting, purchase, and use of environmental technologies.
ETV works in partnership with recognized standards and testing organizations; stakeholders
groups which consist of buyers, vendor organizations, and permitters; and with the full
participation of individual technology developers. The program evaluates the performance of
innovative technologies by developing test plans that are responsive to the needs of stakeholders,
conducting field or laboratory (as appropriate) testing, collecting and analyzing data, and
preparing peer reviewed reports. All evaluations are conducted in accordance with rigorous
quality assurance protocols to ensure that data of known and adequate quality are generated and
that the results are defensible.
NSF International (NSF) in cooperation with the EPA operates the Drinking Water Treatment
Systems (DWTS) Pilot, one of 12 technology areas under ETV. The DWTS Pilot evaluated the
performance Rosedale Products Inc. (RPI) GFS-302P2-150S-ESBB Rigid Cartridge Filter
System (RPI GFS Filter System), which is a cartridge/bag filter system used in package drinking
water treatment system applications. The system was evaluated during field testing to assess the
system's logio removal capabilities for particles of 3 micron (|j,m) or larger at flow rates of 10
gallons per minute (gpm). The verification testing included a seeding of microspheres sized 3
|im or larger as a non-pathogenic surrogate for Cryptosporidium parvum (C. parvum). This
document provides the verification test results for the RPI GFS Filter System.
1.2 Testing Participants and Responsibilities
The ETV testing of the RPI GFS Filter System was a cooperative effort between the following
participants:
NSF International
Cartwright, 01 sen & Associates, LLC
Rosedale Products, Inc.
Analytical Laboratories
Minneapolis Municipal Water Works
U.S. Environmental Protection Agency
The following is a brief description of each ETV participant and their roles and responsibilities.
1
-------
1.2.1 NSF International
NSF is a not-for-profit standards and certification organization dedicated to public health safety
and the protection of the environment. Founded in 1946 and located in Ann Arbor, Michigan,
NSF has been instrumental in Ihe development of consensus standards for the protection of
public health and the environment. NSF also provides testing and certification services to ensure
that products bearing the NSF Name, Logo and/or Mark meet those standards. The EPA
partnered with the NSF to verify the performance of drinking water treatment systems through
the EPA's ETV Program.
NSF provided technical oversight of the verification testing. An audit of the field analytical and
data gathering and recording procedures was conducted by NSF. NSF also reviewed the Field
Operations Document (FOD) to assure its conformance with pertinent ETV generic protocol and
test plan. NSF also conducted a review of this report and coordinated the EPA and technical
reviews of this report.
Contact Information:
NSF International
789 N. Dixboro Rd.
Ann Arbor, MI 48105
Phone: (734) 769-8010
Fax: (734) 769-0109
Contact: Bruce Bartley, Project Manager
Email: bartley@nsf.org
1.2.2 Field Testing Organization
Cartwright, Olsen & Associates (COA), a Limited Liability Company, conducted the verification
testing of RPI GFS Filter System. COA is a NSF-qualified Field Testing Organization (FTO) for
the DWTS ETV pilot project.
The FTO was responsible for conducting the verification testing. The FTO provided all needed
logistical support, established a communications network, and scheduled and coordinated
activities of all participants. The FTO was responsible for ensuring that the testing location and
feed water conditions were such that the verification testing could meet its stated objectives. The
FTO prepared the FOD, oversaw the pilot testing, managed, evaluated and interpreted the data
generated by the testing, as well as evaluated the performance of the technology. The FTO also
prepared this verification report.
FTO associates and personnel provided by the Minnesota Department of Health conducted the
onsite analyses and data recording during the testing. Oversight of the daily test activity was
provided by the FTO's Project Manager.
Contact Information:
Cartwright, Olsen & Associates, LLC
19406 East Bethel Blvd.
2
-------
Cedar, MN 55011
Phone: (763)434-1300
Fax: (763) 434-8450
Contact: Philip C. Olsen, Project Manager
Email: p.olsen@ix.netcom.com
1.2.3 Manufacturer
The treatment system is manufactured by Rosedale Products, Inc. (RPI). RPI is a 20 year old,
privately held company. RPI is one of the largest manufacturers of bag filter hardware in the
world. The products range from simplex and duplex strainers, to automatic backwashing filters
and Giardia/Cryptosporidium removal systems.
RPI was responsible for supplying a field-ready GFS Filter System equipped with all necessary
components including treatment equipment, instrumentation and controls and an Operations and
Maintenance (O&M) manual. RPI was also responsible for providing logistical and technical
support as needed as well as providing technical assistance to the FTO during operation and
monitoring of the equipment undergoing field verification testing.
Contact Information:
Rosedale Products, Inc.
3730 West Liberty Rd.
Ann Arbor Michigan 48106
Phone: (800) 821-5373
Fax: (734) 665-2214
Contact: Jim Arnold, Operations Manager
Email: j arnonld@rosedaleproducts. com
1.2.4 Analytical Laboratories
Analytical work performed in the laboratory was performed by Spectrum Labs, Inc. Spectrum's
laboratory provided analytical services for Total Alkalinity, Total Hardness, Total Organic
Carbon (TOC), UV254 Absorbance, True Color, Total Coliform, Algae (number and species),
Iron and Manganese.
Contact Information:
Spectrum Labs Inc.
301 West County Road E2
St. Paul, MN 55112
Phone: (651) 633-0101
Fax (651)633-1402
Contact: Gerard Herro, Laboratory Manager
Email: gherro@spectrum-labs.com
3
-------
The microscopic particle analysis including fluorescent microspheres was performed by:
Contact Information:
Debra Huffman Environmental Consulting
6762 Millstone Drive
New Port Richey, FL 34655
Phone: (727) 553-3946
Fax: (727) 893-1189
Contact: Debra Huffman, Ph.D.
E-mail: dhuffman@marine.usf.edu
Additional particle counting analysis was provided by:
Contact Information:
Micro Measurement Laboratories, Inc.
1300 South Wolf Road
Wheeling, IL 60090
Phone: (847) 459-6540
Fax: (847) 459-3088
Contact: Dan Berdovich, Manager of Quality Control and Regulatory Affairs
1.2.5 Minneapolis Municipal Water Works
The Minneapolis Municipal Water Works (MWW) was established in 1867 for fire fighting
protection, and in 1872 for drinking water distribution. The MWW service area has a combined
population of nearly 500,000, with over 100,000 service connections, 14,000 valves and 8,000
hydrants. About 40% of the total city usage (excluding suburbs) is for residential purposes, 45%
is for institutional, commercial, industrial, and 15% is municipal and other uses. The MWW
campus used for this verification testing is located in Fridley, Minnesota, a suburb adjacent to
and directly north of the City of Minneapolis. The testing equipment was located in Pump
House #5.
Contact Information:
City of Minneapolis
Minneapolis Municipal Water Works
Pump House #5,
4100 Marshall Street NE
Fridley, Minnesota 55421-2600
Phone: (612) 661-4946
Fax: (612) 661-4914
Contact: Charles Kocourek
1.2.6 U.S. Environmental Protection Agency
The EPA through its Office of Research and Development has financially supported and
collaborated with NSF under Cooperative Agreement No. CR 824815. This verification effort
4
-------
was supported by the Drinking Water Treatment Systems Pilot operating under the ETV
Program. This document was peer reviewed and reviewed for technical and quality content by
NSF and the EPA and recommended for public release.
1.3 Verification Testing Site
The verification testing of the RPI GFS Filter System took place at Pump House #5 on the
campus of the Minneapolis Municipal Water Works. Pump House #5 is the intake point from
the Mississippi river and as its name suggests consists of two levels of pumps. The lower level
has raw water, high volume low pressure pumps; the upper level contains high volume, high
pressure distribution pumps. The location had the advantage of being a busy, active facility
subject to the many variations afforded by a major metropolitan water treatment facility. While
it had the benefit of real world dynamics, it offered the FTO a challenge unlike that of a highly
controlled, laboratory facility. The treatment plant, and the test station, were exposed to changes
in flows and pressures, some predictable (such as the daily variations in demand in the AM and
PM) and others unexpected. These variations were reflected in many instrument readings.
The location also limited certain aspects of the study since the study technicians were necessarily
mindful of the presence and needs of the MWW staff. Motors and pumps were switched on and
off in accordance with city demands and not test station convenience. Pressures fluctuated as
demands changed and as valves were opened or closed, which added to the unpredictability of
the flows and turbidities in the test station. The test station was nestled between 500, 1,000,
1,800 and 2,000 horsepower (hp) pumps and motors. Earplugs were required which made
communication between technicians difficult, especially during seeding events. That considered,
however, the test site offered excellent, comprehensive real world conditions.
1.3.1 Source Water
The source water for the verification testing was a blend of raw water from the Mississippi River
and finished water from the MWW treatment plant. In Minnesota, the Mississippi River is in the
Upper Mississippi River Basin area. Geology, geomorphology, climate, hydrology and land
covering this area control the occurrence and flow of water, and the distribution of water-quality
constituents. Landforms within this Upper Mississippi River Basin are primarily results of
Pleistocene glaciation. Soils developed on glacial deposits range from heavy, poorly-drained
clayey soils developed on ground moraine to light, well-drained sands on outwash plains.
Agriculture is the dominant land use in the southern and western parts of the basin area: forests
cover much of the northern and eastern parts of the basin area, and the Twin Cities Metro
(location of the MWW) dominates the east-central part of the basin area (USGS, 1999).
The Upper Mississippi River's Basin is underlain by glacial sediments and by a thick sequence
of limestone, shale, shaley sandstone and sandstone of Precambrian and Paleozoic age (USGS,
1999).
The climate of the Fridley, Minnesota area is sub-humid continental. The average monthly
temperature ranges from -12 Celsius (°C) or 11 degrees Fahrenheit (°F) in January to 23°C (74
5
-------
°F) in July. Average precipitation at the MWW is 30 inches. About three-quarters of the annual
precipitation falls from April to September (USGS, 1999).
Mississippi River water is treated at the MWW. The treatment plant is the largest water utility in
the upper Midwest. The MWW produces an average of 70 million gallons per day (mgd). Peak
rate during the summer may be as high as 180 mgd.
At the MWW, water is withdrawn from the river at Pump House #5. From the pumping station,
the water is delivered to a softening plant where lime is used for softening, and alum is used for
removal of color and turbidity. Dilute lime and alum slurry precipitates and settles out during
the softening process. Powdered activated carbon and occasionally potassium permanganate are
also added to remove taste and order. The water is then treated with carbon dioxide to lower the
pH and stabilize the remaining hardness prior to being pumped to one of two filtration plants.
At the filtration plant, chlorine and ammonia are added for initial disinfection, fluoride is added
for tooth decay prevention and ferric chloride is added as a coagulant to remove remaining color
and turbidity. The water then enters a series of coagulation/sedimentation basins after which the
water is filtered with single, dual or mixed media filters. Blended poly/ortho phosphate is later
added as a corrosion control/inhibitor. The water is post chlorinated for final adjustment of the
disinfectant residual before being fed into the reservoirs and pumped into the distribution system.
The quality of the water is assured and controlled through the various stages of treatment by
plant and laboratory tests. An average of 500 chemical, physical and bacteriological
examinations are done each and every day (182,500 tests per year).
During the 32 days of the ETV test period, the blend of river water and treated water exhibited
the following characteristics: turbidity concentrations average of 1.1 Nephelometric turbidity
unit (NTU); temperature range of 3.9°C to 11°C; pH range 8.0-8.9; total alkalinity of 71
milligrams per liter (mg/L); total hardness of 96 mg/L; TOC concentration average of 7.8 mg/L;
UVA254 range of 0.108 to 0.229 cm"1, true color of 14 total color units (TCU), total coliform of
23 colony forming units per 100 milliliters (CFU/90 mL), iron equal to or less than 0.4 mg/L,
manganese less than 0.04, free chlorine average of 0.6, and total chlorine average of 1.4. A
summary of the feed water quality information is presented in Table 1-1 below.
6
-------
Table 1-1. GFS Filter System Feed Water Quality (March 7 to April 20, 2000)
„ , ^ °f a . Standard
Parameter samples Average Minimum Maximum . .
^ Deviation
95% Confidence
Interval
Practical
Quantification
Limit (PQL)
Total Alkalinity (mg/L)
7
70
55
110
18
57, 84
10 mg/L
Total Hardness (mg/L)
7
94
82
130
16
82, 107
10 mg/L
True Color (TCU)
7
14
10
25
6
10, 18
1 TCU
Total Coliform (CFU/100/mL)
7
24*
<1
110
40
<1, 54
1 CFU
TOC (mg/L)
7
7.8
6.8
11.0
1.4
6.7, 8.9
0.4 mg/L
UVA254 (cm4)
7
0.140
0.108
0.229
0.042
0.109,0.171
-
On-line Turbidity (NTU)**
-
1.1
0.7
1.5
0.2
1.0, 1.2
-
Total Chlorine (mg/L)
27
1.4
0.7
3.5
0.82
1.1, 1.7
-
Free Chlorine (mg/L)
27
0.6
0.1
2.5
0.6
0.4. 0.8
-
Iron (mg/L)
7
0.1*
<0.1
0.4
0.1
<0.1, 0.2
0.1 mg/L
Magnesium (mg/L)
7
0.02*
<0.01
0.04
0.01
0.01, 0.03
0.01 mg/L
Temperature (°C)
38
7.3
3.9
11.0
2.2
6.7, 8.0
-
PH
37
8.5
8.0
8.9
0.2
8.4, 8.5
-
* All calculations involving results with below PQL values used half the PQL in the calculation.
** Turbidity values are the on-line values and the average results of each filter run.
1.3.2 Pilot Effluent Discharge
The effluent of the pilot treatment unit was discharged to Minneapolis Metropolitan sanitary
sewer. The Metropolitan Environmental Authority, which encompasses the Minneapolis Metro
Area, maintains a primary sewage treatment plant that discharges to the Mississippi River
downstream of the MWW. No discharge permits were required.
7
-------
Chapter 2
Equipment Description and Operating Processes
2.1 Historical Background
Conventional methods of water treatment, including gravity filtration and chlorination, have not
been as effective against protozoan (oo)cysts, especially Cryptosporidium parvum (C. parvum)
in part because of its size and resistance to chemicals. Treatment plants that are otherwise in
compliance with public health treatment standards are thus vulnerable to outbreaks of disease
(Kaminski, LeChevallier, Korich).
In recent years, protozoan cysts have been determined to be the cause of widespread illness.
These cysts are more resistant to traditional disinfection practices and because of their small size
and pliability, protozoan cysts and oocysts have been known to pass through fiber media filters.
Two such microorganisms are the protozoan (oo)cysts Giardia lamblia (G. lamblia) and C.
parvum. These pathogenic microbes can cause significant gastrointestinal distress, and even
fatalities in the cases of immunocompromised individuals and are thus of considerable interest to
the public health and water treatment communities. Assurances will be required before small
public water systems throughout the country dependent on surface water sources that are likely
contaminated with the pathogen can be confident in employing bag/cartridge filters as a part of
their treatment regimen.
Filtration, in which particles are removed from a water stream by passing water through a
medium that captures and contains them, has an ancient history. Earthen filters using granulated
media such as sand or coal are used worldwide to clarify water. The exact mechanism of
containment is not fully understood, however, it is generally agreed that one mechanism consists
of "straining", where the particle is too hrge to pass through the pores between the media. In
addition, electro-static forces inherent on the media cause the particles to attach. Still other
mechanisms have been proposed that explain the process. In the case of porous fibers and
cartridge filters straining or bridging is presumed to be the primary mechanism of capture
(Maschio).
Filtration has progressed beyond that first employed by civil engineers. Newer, high strength
materials engineered to withstand greater pressures and with a high degree of uniformity in pore
size allows for application of bags and cartridges to more exotic filtration requirements. Bag and
cartridge filters are routinely employed in process fluid applications, even in the cases of highly
viscous fluids. RPI has designed a bag and cartridge filtration system for capture of protozoan
cysts and oocysts, specifically G. Lamblia and C. parvum.
The advantages of Bag and Cartridge Filters include (NRC, 1997):
Designed for simple operation; they do not use coagulant chemicals.
The limitations inherent with technology include (NRC, 1997):
8
-------
Limited to removal of particles from water; will not remove chemical contaminants
present in solution.
Limited to treatment of high quality water sources and not appropriate for waters with
elevated turbidity without pretreatment.
2.2 Equipment Description
The RPI GFS Filter System (Model # GFS-302P2-1505 ESBB) is a bag and rigid cartridge
system that consists of two connected filter housings, the first with a Model # GD-PO-523-2 bag
filter element, and the second with a Model # PI^520-PPP-141 rigid cartridge filter element,
which replaced the Model # GLR-PO-825-2 filter element used during initial operations. The
ETV testing of the RPI GFS Filter System was concluded in April 2000. In 2001, Rosedale
made a product modification on the Model # PL-520-PPP141 rigid cartridge filter element by
changing the seals from an o-ring type seal to a u-cup type seal. The cartridge filter with the new
u-cup seals is marketed under the Model # GLR-520-P2. The Model # GLR-520-P2 is NSF
listed. The cartridge element with the new u-cup seals (Model # GLR-520-P2) will be the
subject of a separate ETV evaluation.
Although the cartridge is rigid, the flow through the cartridge filter is like that of a bag, from
inside to outside. The housings are designed to operate at 20 gpm but were operated at 10 gpm
during the verification testing. It was determined during the initial operations period that the
housing at this site should be operated at 10 gpm t) reduce the filter load. During initial
operations the filters loaded up overnight when operated at 20 gpm and at higher turbidities.
The equipment tested was a filter system designed to capture and contain particles in the size
range of C. parvum. Since G. lamblia cysts are larger it is assumed that if the smaller oocysts are
contained, the larger cysts will be contained at least the same level (Nieminski). Accordingly,
while this filtration system is applicable to G. lamblia removal as well as C. parvum removal,
focus will be on C. parvum sized particles. The system is designed as a final barrier to operate in
waters of 1 NTU turbidity or less, and with pressures of less than 60 pounds per square inch
(psi). This equipment is expected to be applied to small systems where the containment of G.
lamblia and C. parvum is of concern. It is suitable for surface water of 1 NTU turbidity or less,
or for a final barrier following roughing pre-filtration.
The filter housings are made of stainless steel. Valves and other components are also stainless or
of materials that will not degrade in water.
The only chemical that was consumed in the operation of the equipment during the ETV
verification test was liquid chlorine. Liquid chlorine bleach (sodium hypochlorite) was added
during the verification period to limit any microbial growth within the filters. Adding bleach to
feed water is commonly done for surface water systems in front of filters to limit microbial
growth. The bleach-metering pump was stopped during verification challenges. Since the blend
of raw and finished water already contained low levels of chlorine and chloramines, the bleach
was added to compensate for the chlorine demand added by the raw river water. Bleach was also
added to the container of spent elements to control odor prior to their inspection. The addition of
chlorine did not represent an O&M issue as far as the Rosedale equipment was concerned.
9
-------
No special licensing is required for the use of the filters. Training in bag/element replacement is
minimal and is fully explained in Appendix A, Operations and Maintenance Manual, as supplied
by the manufacturer.
The filter system is suited to small public water systems where water treatment plant operators
typically have minimal technical training. The system itself requires no additional chemicals
beyond normal disinfection and relatively limited on-site supervision, for tasks such as reading
pressure gauges and changing filters. Additional controls, meters, and other instrumentation can
be added to facilitate ease in monitoring performance. The filter system itself requires no power.
However, a source water pump may be required.
Photograph 1 illustrates the RPI GFS Filter Systems on location at the MWW. The bag and
cartridge elements are shown in Photograph 2.
Photograph 1 - RPI GFS Filter System
10
-------
Photograph 2- Bag and Cartridge Filter Elements
11
-------
Figure 2-1 is a schematic showing the position of the filters.
Figure 2-1. GFS Filter System Schematic
2.2.1 Equipment Installation
The connection to the train of filters for the initial operations, and later for the single filter during
verification, was through a system for blending raw Mississippi river water with finished water
supplied by the City of Minneapolis.
Pump House #5 which is at the Mississippi river intake location supplied raw water. The water
was screened for large debris, and then pumped through a four-foot diameter pipe to the Lime
Softening Plant. The connection to the pilot blending control valves was a two-inch flexible pipe
attached to the top of this four-foot pipe. The raw water pressure at the pipe was approximately
20 psi. This was not sufficient pressure to supply the pilot, thus a 1.5 Horsepower (hp) booster
pump was added.
12
-------
After installation, and during initial operations, this pump lost its prime on occasion due to
excess air in the line, so a detention tank with an air release valve was added. It was later
determined that the air was introduced when one specific city raw water pump was placed on line
without sufficient priming, hence the occasional, unpredictable and intermittent air events.
The finished city water came from a two-inch supply line, which while direct from the city
service pumps, was a line used to supply the domestic needs of the pump house. A reduced
pressure zone (RPZ) backflow preventor was placed between this supply line and the plumbing
providing water to the ETV pilot installation.
The finished water was supplied at 103 psi, and was reduced through a metering valve to closely
match the raw water pressure. Separate flow control valves allowed the operator to adjust
proportions of raw and finished water. Following the blending station, an on-line static mixer
was used for thorough mixing of blended water, chlorine, and microspheres. The blended water
flowed through a balanced header and then into the three housings. Each housing had an influent
and effluent butterfly valve and a third butterfly valve on the bottom to accommodate a drain
line. Following initial operations, one of the filter trains was removed and the pipe and valve
was used as a by-pass line.
The effluent from each housing was directed first through a water volume meter then through a
flow rotometer, a metering valve, a pressure gauge, and into a discharge line. A sample port
directed a proportion of this flow through a manifold system with valves that allowed the
operator to select which of the three filter effluent lines would be directed to the particle counter
and turbidimeter. Influent samples were withdrawn from a point following the static mixer.
Effluent sample ports were located on the exit side of the rigid filter element vessel, at the
pressure gauge port.
2.3 Operating Process
Operation of a bag and cartridge filtration system is straightforward. Water containing
particulate matter is directed at a steady flow rate through the housings containing the filter
elements and matter is trapped and contained within. Elements are changed when either the
pressure loss through the housing is so great as to reduce the flow or to threaten to burst the
element, or when matter is known to leak through the elements, which ever is first. With no
power requirements (other than those added to monitor performance), simple replacement
procedures and limited operator attention, these filters have been attractive to small, individual
surface water treatment systems.
13
-------
Chapter 3
Methods and Procedures
3.1 Experimental Design
The experimental design of this verification study was developed to provide accurate information
regarding the performance of the treatment system. The impact of field operations as they relate
to data validity was minimized, as much as possible, through the use of standard sampling and
analytical methodology. Due to the unpredictability of environmental conditions and mechanical
equipment performance, this document should not be viewed in the same light as scientific
research conducted in a controlled laboratory setting.
One task of the verification testing involved challenging the RPI GFS Filter System with
polystyrene microsphere surrogates in the size range of C. parvum oocysts were seeded into a
blend of Mississippi River water and Minneapolis finished drinking water.
3.1.1 Objectives
The verification testing was undertaken to evaluate the performance of the RPI GFS Filter
System. Specifically evaluated were RPI's stated equipment capabilities and equipment
performance relative to water quality regulations. Also evaluated were the operational
requirements and maintenance requirements of the system. The details of each of these
evaluations are discussed below.
3.1.1.1 Evaluation of Stated Equipment Capabilities
In March and April of 2000, the ability of the RPI GFS Filter System to remove particles in the
range of C. parvum was tested at the City of Minneapolis Water Works. The testing employed
polystyrene (latex) microspheres as a non-pathogenic surrogate. The accepted size of the C.
parvum oocyst is subject to some discrepancy. Authoritative sources cite differing sizes for the
oocyst ranging from 2-5 |j,m (US EPA April 1999) to 4.6 - 5.5 |j,m (Harter) to 3.9-5.9 |j,m
(Medema). It is possible that different isolates may have slightly different average sizes as well.
While C. parvum oocysts are most often considered to be 4-6 |j,m in size (Bukhari, Davis 1998),
they are also known to be pliable and slightly disc shaped, thereby allowing for occasional
passage through pores smaller than their average diameter. EPA methods 1622 and 1623 employ
l|j,m filters, and it is generally accepted that pores of this size will capture essentially all of the
oocysts. The use of 3 |j,m particles was intended to compensate for this variation, although
oocysts may be as small as 2 |j,m. Particle counter bins were set to conform to the ICR sizing, of
2-3 |j,m, 3-5 |j,m, 5-7 |j,m, 7-10 |j,m, 10-15 |j,m and 15+ |j,m, thus while the primary ranges of
interest for this evaluation were 3-5 |j,m and 5-7 |j,m, particle counts in the bin size 2-3 |j,m may
also be of value. Additional water quality data against which the equipment was tested are
included so that state regulators may draw conclusions about possible performance in other field
applications.
14
-------
3.1.1.2 Evaluation of Equipment Performance Relative To Water Quality Regulations
With increased awareness of pathogens resistant to traditional disinfection and removal
techniques and the fact that the EPA's rules for surface filtration are becoming increasing
stringent, it is expected that the search for alternative disinfection and removal technologies will
grow significantly. This verification study specifically addresses removal of particles in the size
range of 3-7 |j,m.
Small public water treatment systems are particularly subject to changes in process flow and
assurance that particles will not detach or be driven through the barrier during these episodes is
of considerable concern by small system operators and purveyors. Tests to verify the
performance under these conditions are included as part of the test plan, by stopping the flow at
significant points following seeding and then restarting.
The system was tested for removal of particles of 3 |im or larger at flow rates of 10 gpm per
system. While the upper limit of pressure differential for the pressure filtration system is 15
pounds per square inch gauge (psig), it was anticipated that turbidity breakthrough might occur
at a lower pressure. While the equipment can withstand pressure differentials exceeding 15 psi,
this test used 15 psi as a terminal pressure loss value.
3.1.1.3 Evaluation of Operational and Maintenance (O&M) Requirements
An overall evaluation of the operational requirements for the treatment system was undertaken as
part of this verification. This evaluation was qualitative in nature. The manufacturer's O&M
manual and experiences during the daily operation were used to develop a subjective judgment
of the operational requirements of this system. The O&M manual is attached to this report as
Appendix A.
Verification testing also evaluated the maintenance requirements of the treatment system. Not all
of the system's maintenance requirements were necessary due to the short duration of the testing
cycle. The O&M manual details various maintenance activities and their frequencies. This
information, as well as experience with common pieces of equipment (i.e., valves, etc.) was used
to evaluate the maintenance requirements of the treatment system.
3.1.1.4 Evaluation of Equipment Characteristics
The qualitative, quantitative and cost factors of the tested equipment were identified, in so far as
possible, during the verification testing. The relatively short duration of the testing cycle creates
difficulty in reliably identifying some of the qualitative, quantitative and cost factors. The
qualitative factors examined during the verification were operational aspects of the RPI GFS
Filter System, for example, the ease to which filter elements can be exchanged, the measurement
of head loss, and the other operational factors that might impact on performance. Among
quantitative factors examined during the verification testing are costs associated with filter
element replacement, any occasional, anomalous conditions that might require operator response
such as high levels of algae growth, excessive turbidity spikes or frequent filter clogging, and
15
-------
length of operating cycle. This treatment system operated at 10 gpm with feed water turbidity of
1 + 0.2 NTU. Costs will change with changes in flow or feed water quality.
3.2 Initial Operations
An initial operations period was performed to allow the equipment manufacturer to refine the
unit's operating procedures and to make operational adjustments as needed to optimize treatment
performance. Initial operations procedures included a characterization of feed water task, a
system start-up task, and a filter element variability task. The MWW has extensive historical
water quality data, which was reviewed prior to and concurrent with the first initial operations
period.
Equipment information gathered during system start-up and optimizations were used to refine the
test system. Adjustments made to the FOD included a reduction in process flow from 20 to 10
gpm, a redesign and replacement of the rigid filter element and a redesign and replacement of the
housing seals. The redesign of the cartridge and seals delayed the start of Phase II of initial
operations until after January 1, 2000.
3.2.1 Characterization of Feed Water
The primary purpose of this initial operations task was to determine the appropriateness of the
feed water for this study. To that end, the characterization of the water included researching the
watershed, including the nature of the water, the source, and the uses of the water upstream.
This task was done in part prior to selection of the site as suitable for testing, and additionally
during the initial operations phase, before the verification testing period.
The suitability of the feed water to the application of this technology was reviewed before testing
during initial operations. Data from 1997 was obtained from the City of Minneapolis, Municipal
Water Works department for the same time frame as the verification testing period (March and
April). This data was compiled and analyzed with respect to the biological, physical and
chemical characteristics of the water. Parameters studied at the verification testing site include
the following: turbidity, temperature, pH, total alkalinity, total hardness, and true color. Review
of this historical data as detailed in Chapter 4, Results and Discussions, indicated that the
technology should be appropriate for the site.
Uses of the watershed, whether industrial, agricultural, or other human activity such as waste
deposit, mining or boat traffic, which may have an impact on the character of the water, were
examined. Incidental conditions, such as storms, ice-out, or unusual boat traffic, may have a
consequence on the performance and were documented in the logs as they occurred.
As a part of this initial operations task the analysis of parameters of the water that affected the
character of this test. Included are those parameters that were required as a part of the regular
scheduled testing. They included: temperature, turbidity, UV254 absorbency, free chlorine, total
organic carbon, true color, pH, total alkalinity, hardness, iron, manganese and total suspended
solids. These were the same water quality parameters that were analyzed during the verification
runs. Microbiological tests included coliform bacteria and algae.
16
-------
3.2.2 Initial Test Runs
The purpose of initial test runs or start-up testing was to conduct and evaluate trial runs of the
filtration system under study. COA and RPI supervised the installation of the equipment and
start-up, and established initial operating conditions. Trial runs of the system were performed.
During this period COA additionally calibrated and standardized the testing apparatus, measured
and controlled feed water blending to assure smooth test performance during the verification
period. These runs were performed to evaluate operating conditions for the Verification Test and
accordingly had no strict format.
3.2.3 Filter Element Variability Testing
The pilot installation for filter element variability testing consisted of three identical RPI GFS
Filter Systems plumbed in parallel for simultaneous testing. This permitted a controlled study of
filter bags and cartridges (rigid filter elements) with variations between and within manufactured
lots.
The filter element variability testing period was divided into two phases. Phase I was designed
to determine variations within a manufactured lot number of cartridge filter elements. Phase II
was designed to show variations between manufactured lot numbers of cartridge filter elements.
Prior to these two phases, the feedwater was characterized for suitability to this technology.
Each phase included of 10 days of system operation and data recording. During both phases the
filter system was to be on-line for 23 hours and off line (no flow) for 1 hour. An operator was
present 8 hours each day for data recording and operation of the filter system and test station.
The operating and data recording schedule for filter element variability testing were as follows:
Data were recorded during 8 hours per day in split shifts, one four hour AM shift and one four
hour PM shift. This was done to better monitor the pressure changes and to observe variability
through the day. In addition, until the FTO was comfortable with the operation, it was important
to have frequent records, especially of pressure differentials. During each four-hour shift,
particle count, turbidity, flow rate, and pressure differentials were noted for each filter system
once each hour. The system was shut down daily at about 16:00 for one hour then flow resumed
at 17:00. Following daily shutdown it took between 30-60 minutes to once again stabilize
influent turbidity and flowrates. Turbidity was maintained near 1 NTU during the shift periods,
when an operator was on site; during the periods where no operator was on site, the turbidity was
reduced to less than 1 NTU by allowing only finished water to pass through the system.
Terminal headloss for the filter element variability testing period was established at a 15-psi
differential between influent and effluent pressure. Pressure differential was determined by
gauges measuring pressure on the inlet and outlet of the system. Flowrates were maintained at
20 gpm per filter train. During Phase I and II, when terminal headloss was reached, or
breakthrough occurred, the filter elements were replaced with another from the same lot and the
run continued until the ten day period had elapsed.
17
-------
Performance data from filter element variability testing included: turbidity, particle count, flow
rate and head loss across each filter. Particle counts and turbidity measurements for each filter
system were collected sequentially each hour during a minimum eight-hour daily workday.
Thus, filter system #1 was measured at 20 minutes past the hour, filter system #2 at 40 minutes
past the hour and filter system #3 on the hour. Each filter had 80 particle removal data points
during the ten-day period, along with flow rate and head loss data. There were a total of 240
data entries for each of the two ten day periods.
During Phase I, the particle counter was set to count at the intervals required in the original test
plan: 1-3 |j,m, 3-10 |j,m, 10-15 |j,m, < 2 |j,m and > 15 |j,m. Due to the small orifice size of the
sensors used with the HIAC-Royco 8000A particle counters, they easily became plugged and
were not conducive to field cleaning. Accordingly, the HIAC-Royco particle counters were
replaced with MetOne PCX counters. The MetOne counters do not have a < 2 |j,m bin so the
counters were set to the same bin sizes except the l-3|j,m was replaced with 2-3|j,m.
COA used variability test data along with the following confidence formula, to determine the
suitability of the manufactured lots for verification testing. The confidence formula employed
was:
confidence interval =X + tn-\ 1-— (S /¦fn)
' 2
S = standard deviation
n = number of measurements in data set
t = distribution value with n-1 degrees of freedom
a= the significance level defined for 95% confidence as: 1- 0.95 = 0.05.
95% confidence interval = X + tn-l,0.975 (S / -Jn)
3.3 Verification Task Procedures
The procedures for each task of the verification testing were developed in accordance with the
requirements of the EPA/NSF ETV Protocol (EPA/NSF, 1998). The Verification Tasks were as
follows:
Task 1 - Verification Testing Runs and Routine Equipment Operation
Task 2 - Feed And Finished Water Quality Characterization
Task 3 - Documentation of Operating Conditions and Treatment Equipment Performance
• Task 4 - Microbiological Contaminant Removal
Detailed descriptions of each task are provided in the following sections.
18
-------
3.3.1 Task 1 - Verification Testing Runs and Routine Equipment Operation
The objectives of this task were to operate the equipment for the prescribed period of thirty days,
or longer if required to reach terminal headloss or turbidity breakthrough, and to evaluate
equipment control features.
During Task 1, treatment conditions of the blended feedwater were characterized with the
differing water conditions caused by different blending ratios. Changes in the nature of the
feedwater or in the individual nature of the river water character and or finished water
pretreatment were annotated.
The total number of gallons, as measured per square foot (cartridge), as well as per filter system
(cartridge and bag), were computed for each RPI GFS Filter System and logged to allow
comparisons of water quality and volume.
Operating parameters of the equipment were logged during the 32-day test period. Frequency of
filter element replacement, changes in pressure loss or turbidity breakthrough were recorded.
Also incident to this task was the documentation of any repairs and maintenance required. The
performance verification period included a single season; varying water quality parameters and
other conditions impacted performance and were noted accordingly.
Factors that effected the treatment performance that were recorded and measured included:
High turbidity and low turbidity periods and their cause, for example, changes due to ice
out or snow melt, rainfall, or excessive river traffic.
Algal blooms, incurred in summer and then again in late spring.
Changes as the result of increased pumping requirements, often on a daily basis.
Elevated natural organic matter from runoff.
Changes in feed (blended) water quality.
Changes in line pressures due to city demands.
3.3.2 Task 2 - Feed and Finished Water Quality Characterization
The purpose of this task was to provide water quality data relating to the test so that State,
Municipal and other Public Health authorities can determine the applicability of a specific water
source to this type of treatment. This task evaluated the water quality matrices of the influent
water and effluent water and the relationships to the terminal headloss and/or turbidity
breakthrough point.
Factors that could influence water chemistry, such as weather, recreational or commercial boat
traffic, in and out-flows, and river bottom composition were recorded during testing when
appropriate. Also included is a discussion of the human impact upon the source; for example,
whether the source was utilized for other activities, or whether it accepted wastewater of any
description.
The parameters, which were analyzed as part of this testing and the sampling frequency, are
presented below in Table 3-1. Samples of both feedwater and filtered water were analyzed.
19
-------
Table 3-1. Analytical Data Collection Schedule
Parameter
Frequency
Feed
Treated
On-Site Analyses
Temperature
pH
Turbidity
Particle Counts
Free Chlorine
Continuous
Continuous
Daily
Daily
Daily
X
X
X
X
X
X
X
Laboratory Analyses
Total Alkalinity
Total Organic Carbon
Total Hardness
UV Absorbance (254)
True color
Total Coliform
Algae
Iron
Manganese
Weekly
Weekly
Weekly
Weekly
Weekly
Weekly
Weekly
Weekly
Weekly
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
All testing was performed in accordance with the procedures and protocols established in
Standard Methods (SM) and/or EPA methods All on-site testing instrumentation or procedures
were calibrated and/or standardized at scheduled intervals by FTO staff.
Turbidity and particle counters were both continuous and on-line. The on-line turbidity meter
was checked daily against a bench turbidimeter that was checked against turbidity standards.
Particle counts were evaluated by recording the change between influent and effluent particle
counts in the size ranges of 2-3 |j,m, 3-5 |j,m, 5-7 |j,m, 7-10 |j,m, 10-15 |j,m, and 15+ |j,m. Logio
removals were calculated for the ranges of concern, 3-5 |j,m and 5-7 |j,m particles continuously
via computer during the verification period.
3.3.3 Task 3 - Documentation of Operating Conditions and Treatment Equipment
Performance
The operation of the equipment was documented to demonstrate performance and applicability to
small systems. Small systems are characterized by lower volume demands, and by lower flow
rates, but more important to this task, they are also characterized by reduced maintenance and
operating staff. Accordingly, important to the small system application is the ability to employ
"hands off operation, and the introduction of back up and alarm systems.
Among the items recorded daily as a part of this task were the readings or measurements of the
equipment's performance, including the rates of flow through the system, total volume of water
filtered, and condition of the filter elements, replacement frequency and production run readings
The operational parameters and frequency of the readings are listed below in Table 3-2.
20
-------
Table 3-2. Operating Data
Parameter
Frequency
Feed and Filter Flow
Checked and recorded twice daily; flow was adjusted if it varied
more than 10%.
Influent and effluent pressures recorded at the start of each filter
run, and thereafter two times daily. Prefilter headloss prior to
replacement or backwash was also noted, along with pressure
readings at the start of each filter run, and thereafter two times
daily.
Recorded volume of water for each filter element for each run, and
the daily total.
Recorded the date and time for each replacement, the volume of
water treated before replacement and the reason for replacement
(headloss or turbidity breakthrough).
Recorded the visual condition of any replaced element for integrity,
excessive inorganic fouling etc.
Recorded daily in logbook at beginning of first shift.
No action was required because the GFS-302PS-150S ESBB Rigid
Cartridge Filter System has no power connection requirements.
Filter Headloss
Filtered Water Production
Element Replacement
Element Condition
Hours of Operation
Electric Power
Also documented were changes in the pretreatment chemistry or filtration rates. The condition
of the pretreated water was measured (in Task 2 above) and identification of any changes to the
pretreatment regimen was recorded.
Filter elements were replaced when the total pressure differential across the RPI GFS Filter
System reached a 15-psi, when turbidity breakthrough was detected, or when it was expected that
differential was reached during the next on line period. With the exception of a brief period
during the verification test (Runs 17 and 18, for element conservation until stock could be re-
supplied), filter elements were always replaced in pairs. The time, pressure differential and
condition of each filter element was noted in the logbook. This information was tabulated,
assembled and used in conjunction with performance data to include operations and maintenance
factors.
3.3.4 Task 4 - Microbiological Contaminant Removal
The ability of the RPI GFS Filter System to remove particles to the size of C. parvum from water
was the primary focus of this task.
During the verification period challenge, microspheres were injected into the pilot installation
feed water via a metering pump at concentrations capable of demonstrating 3+ logio removal
through the RPI GFS Filter System. There were three challenges employing polystyrene
monospheres added to the source water to demonstrate removal in each of three filter runs, for a
total of nine challenges. The challenges occurred at the beginning of the run, at roughly the mid-
point as determined by headloss, and then again at a point between 90% headloss and terminal.
Each seeding consisted of 10,000 particles per milliliter added to a half-liter of dilution water and
was fed over a five-minute period through a metering pump. Downstream of the microsphere
injection point an on-line static mixer was used to assure proper mixing of microspheres in the
feedwater, previous to entry into the RPI GFS Filter System.
21
-------
Continuous particle counting via electronic particle counter, and sample collection for
enumeration were employed as a means of measuring the removal capabilities of the filter
element. Particle counts were measured in both the influent water and the effluent water.
3.3.4.1 Preparation of Microbial Surrogate Doses
Microspheres in the size range of 3-6 |j,m were used to evaluate the removal capability of the RPI
GFS Filter System. The polystyrene microspheres with a nominal diameter of 3-6 |j,m, of which
at least 50% were in the 3-4 |j,m ranges, were employed for challenge testing. The three sizes
employed were: 3.2 |im, 4.5 |im and 6 |im. The challenge mixtures were composed of 50% 3.2
|im and 25% each 4.5 |im and 6 |im.
The procedure for the preparation of microsphere suspensions was as follows:
A clean, 500-milliliter (mL) National Institute of Standards and Technology (NIST) traceable
(C30913) volumetric flask was filled to the measure line with particle free water (PFW) as
described in Section 3.8.2.5. The flask for the microsphere concentration was washed with hot
water and a lab glass cleaner, and rinsed with PFW following each microsphere injection
procedure and again prior to use. With a clean pipette, approximately 10 mL of dilution water
was withdrawn and set aside. Tween 20 (to .01%) was added to the flask and swirled gently.
The concentrated microsphere suspensions in their shipping bottles were vortexed for 10
seconds, inverted and rortexed again for ten seconds. The appropriate volumes of each size
microsphere concentrate were added to the flask using a wide mouth, disposable serological
pipette.
If the microsphere shipping bottles contained the correct number of microspheres the entire
content was added to the flask. The pipettes (or shipping bottles) were rinsed with PFW and the
rinse added to the flask.
Following the addition of the microspheres, the withdrawn PFW was returned to the flask to the
volume line. After which, a magnetic stir bar was rinsed with PFW and added to the flask.
The volume of suspension was 500 mL and was added at the rate of 100 mL per minute over five
minutes.
The number of particles in the concentrated suspension is inversely related to the cube of the
diameter of the microsphere, and is calculated by the following formula:
6W x 1(?2
n/mL = j
p x n x 0
Where n is the number of particles, W = the grams of polymer per milliliter of latex (which
varies for each size and manufacturer, but which is noted on each container), p = density of
polymer (1.05 for polystyrene) and
-------
The microsphere suspension was injected by a variable pulse/stroke diaphragm-metering pump
through an injection quill into the front of a static mixer. The metering pump was set at its
maximum pulse frequency and the stroke was adjusted to achieve the designed microsphere
injection rate. This rate was measured prior to the seeding by using dilution water. The rate was
further measured by timing the duration of the seeding. The static mixer showed a headloss of
0.3 to 0.5 feet during seeding. During seeding, the suspension was continuously mixed with a
magnetic stir bar.
Removal capabilities of the RPI GFS Filter System were demonstrated by measuring the particle
distribution of influent and effluent streams, following the addition of 10,000 particles per mL,
with on-line electronic optical particle counters. The resulting data were ambiguous.
The addition of a measured concentration of 10,000 particles per milliliter during seeding
challenges did not result in the same increase over the indigenous particles in the influent water.
In all trials with the addition of 10,000 particles per milliliter resulted in a net increase of only
4,000 to 5,000 particles per milliliter as measured by the on-line particle counter. An additional
challenge was performed with a microsphere concentration of 20,000 particles per milliliter in
which the on-line particle counter measured as a net increase of only 7,000 particles per
milliliter.
These results were discussed with the manufacturer of the particle counter, the manufacturer of
the microspheres and the writers of the referenced paper describing the methodology in question.
It was concluded that on-line particle count data measuring high concentrations of microspheres
in the filter influent water could not be employed with confidence to demonstrate filter
performance. Thus, it was decided to augment the on-line particle counting and turbidity data
with the technique described by Li, using hemacytometer counts of fluorescent microspheres
(Li). Challenges were then repeated with fluorescent microspheres as detailed below.
3.3.4.2 Description of Fluorescent Microsphere Seedings
Fluorescent microspheres are not available in the sizes indicated in the test plan. Three sizes of
fluorescing microspheres at 3.4 |j,m, 5 |j,m and 6 |j,m were available from three separate
microsphere manufacturers:
Size Manufacturer Lot#
3.4|j,m Bangs Laboratories 2200
5.0 |j,m Duke Scientific Corp. 21755
6.0 |j,m Polysciences, Inc. 500045
The microspheres were prepared as in the case of the regular polystyrene spheres using effluent
water as dilution water. The seedings were performed as described on April 17, 18 19, and 20,
2000, however, it was necessary to collect samples of the influent and effluent water for
hemacytometer and microscopic evaluation. Those samples were collected from the discharge of
the particle counter.
23
-------
The particle counts were observed on the on-line particle counter and when the peak
concentration was reached and stabilized, samples were collected from both the influent and
effluent sample streams. The effluent sample lagged the influent by about 2 minutes. Each
sample was distributed into two aliquots, one for shipment to Dr. Debra Huffman for
examination and the other refrigerated as a back up.
Samples of the first challenge run were collected sequentially, with the first immediately
followed by the second. The first was shipped to Dr. Huffman, and the second placed in
refrigerated storage as a backup. Samples for the second and third challenge runs were collected
in two single grabs from both the influent and effluent counter for three minutes, (300 mL).
Those samples were then divided into two aliquots, one to be shipped overnight express to Debra
Huffman Environmental Consulting, and the other refrigerated as a backup.
Challenges occurred as follows, with the details recorded as noted:
• Challenge Seeding #1, at 0 headloss: (date, computer time, volume reading, pressure loss
etc.) Start time, stop time for injection of particles. During the injection, the particle
counter readouts were observed to note appropriate distribution.
• Challenge Seeding #2, at mid-point of filter run (50% headloss): readings same as
#1,
• Stop/Start: Filter flow was stopped for a brief period (approximately 5 minutes) and then
resumed.
• Challenge Seeding #3, at 90% of terminal headloss: same as #1.
This schedule was repeated for each of the three filter runs.
3.3.4.3 Data Evaluation
Continuous dectronic particle count data were evaluated by calculating the change in total
particle count from feed water to filtered water, expressing the change in logio reduction. The
aggregate of particle counting data obtained during each verification testing period was analyzed
to determine the average logio removal and 95* percentile logio removal during the 32-day
verification testing period.
One-minute time intervals were used for analysis of particle counting data for logio reduction of
particles in both unfiltered and filtered water. In addition, because particle count data was
continuous, it was possible to present a trend of particle counts with passage of time.
Samples were collected and sent to the laboratory for microscopic enumeration of the fluorescent
spheres using hemacytometer techniques and/or membrane filtration, as appropriate. The
hemacytometer was used when the samples contained high numbers of particles (influent
samples); when the counts were low (effluent samples) the particles were counted
microscopically after filtration through a membrane.
24
-------
3.4 Data Recording, Communications, Logistics and Data Handling Protocol
The objective of the data handling protocol was to tabulate the collection of data for
completeness and accuracy, and to permit ready retrieval for analysis and reporting. In addition,
the use of computer spread sheets allowed manipulation of the data for arrangement into forms,
such as tables or charts, useful for evaluation. A second objective was the statistical analysis of
the data as described in the "NSF/EPA Protocol for Equipment Verification Testing for Physical
Removal of Microbiological and Particulate Contaminants" (EPA/NSF 1998).
Documentation of study events was facilitated through the use of logbooks, photographs, data
sheets, and chain of custody forms. The data management system used in the verification testing
program also involved the use of computer spreadsheet software and manual recording methods
for recording operational parameters on a daily basis.
The chemical parameters and operating data were maintained in a bound logbook (Appendix B)
and on specially-prepared data log sheets (Appendix C). In addition to the items noted in the
data sheets, variations in the treatment plant regimen were noted. Among the changes possible
were changes intended to respond to varying biological contamination and turbidity due to
unusual source water episodes, such as weather related incidents (ice outs, storms) or unusual
traffic or contaminant spills.
3.4.1 Procedures
Procedures existed for the use of the logbooks used for recording the operational data, the
documentation of photographs taken during the study, the use of chain of custody forms, the
gathering of on-line measurements, entry of data into the customized spreadsheets, and the
method for performing statistical analyses. The following is a description of these procedures.
3.4.1.1 Field Notebooks
COA as the FTO for the project was responsible for the maintenance of the field notebooks.
Data were collected in a bound field notebook (Appendix B) and on specially-prepared data log
sheets (Appendix C) from the instrumentation panels and individual testing instruments. The
master field notebook contained flowrate, volume, and pressure variations across several
portions of the system, headloss across the filter housings, bag replacement frequencies and other
variables, as well as notes on the challenge seedings. On-line particle counters and turbidimeters
were linked to a computer with appropriate software for automatic data logging. The test official
time clock was that of the computer; other timepieces such as stopwatches and sweep hands were
used to measure flows or processes.
Each page of the field notebook was sequentially numbered and identified as Rosedale ETV
Test. After October 15, 1999, when on-site staff was notified, each log page was initialed by the
on-site staff member. Prior to that date, since the log was for initial operations and not the
verification period, staff members had understood that the daily visitor log, which they signed,
would suffice for identification. Errors were crossed with a single line and initialed. Deviations
from the FOD whether by error or by a change in the conditions of either the test equipment or
25
-------
the water conditions were noted in the field notebook. The field notebook included a carbon
copy of each page. The original field notebook was stored on-site, the carbon copy sheets
forwarded to the project engineer of COA at least once per week. This not only eased
referencing the original data, but offered protection of the original record of results.
The COA office was the central data collection point and all raw data and notes are on file.
3.4.1.2 Photographs
Photographs were logged into the field logbook. These entries include time, date, and identity of
the photographer.
3.4.1.3 Chain of Custody
Original chain of custody forms traveled with the samples from the test site to the laboratory
(copies of which are attached as Appendix D).
3.4.1.4 On-line Measurements
On-line measurements included particle counters (MetOne PCX) and turbidimeters (HACH
1720C). These instruments were linked to a computer with software designed to record data at
selected intervals. Data were recorded every 2 minutes, except during the challenge testing when
the frequency of recording was changed to one-minute intervals. These data were displayed in
real time and digitally stored within a computer. Digitally stored information was backed up on
a ZIP® disk daily and delivered to COA's office. Manual logbooks were used to record data not
connected to automatic recorders such as flow rates, on-site chemical analysis and pressure. All
data was maintained by the FTO and the data was entered into a spreadsheet database.
3.4.1.5 Spreadsheets
Table 3-1 (Section 3.2.2.2) lists the daily, weekly and monthly water quality samples that were
collected. The results of the daily on-site analyses were recorded in the field notebooks. All
details affecting the operation of the equipment, whether by COA staff, or by State of Minnesota,
Department of Health staff, were also logged in the field notebooks, consolidated and entered
into computer spreadsheets. The data spreadsheets are attached to this report as Appendix C.
A COA associate entered data into a computer spreadsheet program (Microsoft© Excel) on a
daily basis from the field notebooks and any analytical reports. A back-up copy of the computer
data was maintained off site. The database for the project was set up in the form of custom-
designed spreadsheets. All data from the field notebooks were entered into the appropriate
spreadsheet. All recorded calculations were checked at this time. Following data entry, the
spreadsheet was printed out and the printout was checked against the handwritten field
notebooks. Corrections were noted on the hard copies and corrected on the screen, and then a
corrected version of the spreadsheet was printed out. Each step of the verification process was
initialized by the COA operator or engineer performing the entry or verification step. This log is
consistent with standard laboratory practices.
26
-------
Data entered on-site was transferred to the COA offices on diskettes.
3.5 Calculation of Data Quality Indicators
3.5.1 Representativeness
Water quality parameter samples for the RPI GFS Filter System were taken as indicated in Table
3-1. Off-site samples were delivered to the laboratory for analysis. The holding times are those
indicated in EPA 40 CFR, Ch. 1, § 136.3 and SM1060. On-site samples were taken utilizing SM
1060 sampling techniques.
Operating data, such as flow rate, volume measurements and pressure gauges were recorded and
the time noted. Operational parameters were recorded and graphed.
3.5.2 Statistical Uncertainty
Statistical 95% confidence calculations were performed for the water quality parameters listed in
Table 3-1. Each of the water quality parameters was analyzed, and confidence intervals
determined by taking a minimum of three discrete samples for each of the parameters at one
operating set during the testing period.
The formula used for confidence interval calculations was:
confidence interval =X + tn-\ 1-— (S /¦fn)
' 2
S = standard deviation
n = number of measurements in data set
t = distribution value with n-1 degrees of freedom
a= the significance level defined for 95% confidence as: 1- 0.95 = 0.05.
95% confidence interval =X ± fn-\ o 975 )
3.5.3 Accuracy
For water quality parameters, the accuracy referred to the difference between the sample result
and the true or reference value. Care in sampling, calibration and standardization of
instrumentation and consistency in analytical technique ensured accuracy.
For operating parameters such as flow rates and pressures, high levels of accuracy were ensured
by redundant testing by confirming flow meters with bucket and stopwatch measurements.
Pressure gauges were verified by reference to NIST-traceable standard gauges.
Performance evaluation was established by calibration of instruments used on-site and by
conformance to SM and EPA protocol.
27
-------
Accuracy was measured by spiking a known value to a solute, or by using a standard sample.
The spiked (or standard) sample was analyzed and the following equations were used:
For a spiked sample:
For a standard:
where:
%R =
A
B
S
3.5.4 Precision
%R = 100
A - B
%R = 100 x
Observed
True
Percent recovery
Result of spiked sample
Result of un-spiked sample
Spike value
Precision was the measure of the degree of consistency from test to test, and was assured by
replication. In the case of on-site testing for water quality, precision was ensured by triplicate
tests and averaging; for single reading parameters, such as pressure and flow rate, precision was
ensured by redundant readings from operator to operator.
Travel blanks were not required for this testing.
Matrix and method blanks were used for turbidity measurements, pH standardization, and for
calibration of the particle counter both with respect to enumeration and size distribution.
The equation employed for precision for duplicate samples was:
Pi ~ D2
RPD =
(Di + D2)/2
x 100
RPD =
Dl
D2
Relative percent difference
First sample value
Second sample value
The equation employed for precision for triplicate samples was:
S(100)
% Relative Standard Deviation =
x
where:
Standard deviation
28
-------
x = Mean of recovery values
3.6 Verification Testing Schedule
The verification testing started on March 7, 2000, and consisted of a 32-day period conducted
over a single season. Daily testing concluded with the final microsphere challenge on April 20,
2000. Data were logged for a total of 781 hours of treatment system operation. The system was
shut down three times due to various problems. The first time the system was shut down on
March 17, 2000, for one hour due to a failure of the raw water feed pump. Testing resumed
when the back-up pump was placed on line. Data were lost due to a failure of the effluent on-
line turbidimeter on March 18, 2000. The system was shut down for 25 hours between March 18
and March 19 while the turbidimeter was replaced with a back up instrument. The system was
also shut down from April 11 through April 15, for a total of 5 days due to the lead-time needed
to secure the fluorescent microspheres for reseeding, and obtaining additional bags and
cartridges from Rosedale Products, Inc. The system was brought back on line on April 16 and
data recording and challenge testing resumed as soon as the microspheres were received.
Fluorescent microsphere challenge testing was performed on April 16 through April 20, 2000.
During the verification period, aspects of the operation were evaluated to determine insofar as is
possible over a brief period, the degree of maintenance and "hands on" attention required. For
this observation the equipment was run continuously except for the one hour interruption or filter
element replacement times and monitored 8 hours a day until the completion of a period of 32
days.
3.7 Field Operations Procedures
In order to assure data validity, the EPA/NSF Verification Test Plan procedures were followed.
This ensured the accurate documentation of both water quality and equipment performance.
Strict adherence to these procedures resulted in verifiable performance of equipment.
3.7.1 Operations
The operating procedures for the RPI GFS Filter System are described in the O&M Manual. The
O&M Manual for the treatment system was maintained on-site and is attached to this document
as Appendix A. Additionally, operating procedures and equipment descriptions were described
in detail in Chapter 2 of this report. Analytical procedures are described in Spectrum
Laboratory's Quality Assurance Plan, as detailed in the FOD.
3.7.2 Analytical Equipment
The following analytical equipment were used on-site during the verification testing:
A Hach 21 OOP portable turbidimeter was used for benchtop turbidity analysis.
Pressure gauges were Ametek 556L (0 to 100 psi.) with calibration field verified with a
NIST-traceable pressure gauge. There were two gauges on the system, one measuring inlet
29
-------
pressure to the RPI GFS filter system and one measuring outlet pressure. Filter elements
were replaced when the total pressure differential reached 15 psi, when turbidity
breakthrough was detected, or when it was expected that differential would be reached
during a period when the pilot was not staffed.
• A NIST-traceable Miller Weber Thermometer, Model T-775/63CGC Serial Number
3C0611 was used for temperature. The temperature was measured in °C, in 0.1°
increments.
A rotometer (Blue and White model F451004LHN (0 to 40 gpm) was used to measure flow
rate. Rotometer accuracy was verified using the bucket and stopwatch technique.
On-line turbidity measurements were taken with HACH 1720C turbidimeter.
On-line influent and effluent particle count measurements were taken with MetOne PCX
particle counters.
3.8 QA/QC Procedures
The objectives of the Quality Assurance/Quality Control (QA/QC) procedures were to assure
that the data collected during the verification test is representative of the equipment and that the
data is not corrupted by either procedural or recording anomalies. To that end, the FTO was
responsible for the administration of the test and for the flow of data, and the individual
laboratories and agents are accountable for their areas of responsibility.
Adherence to analytical methods as published in Standard Methods or EPA approved
methodology was assured. Moreover, instrumentation and standard reagents were referenced to
NIST. Instruments used to gather data were standardized and calibrated in accordance with the
schedules noted below.
3.8.1 QA/QC Verifications
Measurements of flowrate, volume, pressure variations across the several portions of the system,
headloss across the filter housings, bag replacement frequencies and other variables were noted
in the logbook. To the degree possible, all measurements were taken at the same interval during
the 32-day verification period. All changes from either the expected or the prior measurement
were noted.
Any failures in equipment, however incidental, were noted and the time and position in the
testing cycle logged. Major equipment failures requiring cessation of the flow or major repairs
were logged both as a means of establishing the value of the data recorded during the period of
failure and as a means of determining the cause of failure.
Laboratory results of water quality parameters were reported in standardized formats.
Microbiological surrogate testing were reported both as raw numerical data and in standard
statistical formats. Particle count and distribution data and turbidity data were measured with the
use of on-line sensors and logged digitally on a continuous basis.
All grab samples, filter cartridges, and travel blanks shipped to outside laboratories were
collected, packaged and shipped as required by SM and/or EPA standards. Sample bottles were
30
-------
provided by the laboratory and were shipped in coolers with ice packs. Chain of custody forms
accompanied the samples.
Flowmeters were calibrated and verified using the bucket and stopwatch technique.
A totalizing water meter was included to accurately measure volume. These meters were
calibrated by bucket and stopwatch as well. Flowmeters and totalizing meters were also
compared to each other.
Daily QA/QC Verifications included:
On-line turbidimeter flow rates were verified (bucket and stopwatch). Flows were
measured with sweepwatch or stop watch and a 1,000 mL graduated cylinder.
Although this was a task specified daily by the ETV test plan, the FTO found it
prudent to verify turbidimeter flow much more often than required, to include any
time the turbidimeter flow was stopped and resumed.
On-line turbidity readings standardized against a calibrated bench turbidimeter.
Batch and on-line particle counter flow rates were verified (bucket and
stopwatch). Flows were verified with a 100 mL graduated cylinder and either a
sweep watch or stopwatch.
Bi-Weekly QA/QC Verification included:
Flow rate rotometers were verified with the use of calibrated 50 gallon tank and
stopwatch.
QA/QC Verification at the beginning of each testing period included:
Cleaning and recalibration of on-line turbidimeters.
Verify particle counter calibration with gradated microspheres.
Check differential pressure transmitter signal and pressure gauge readings with
pressure meter. There was no differential pressure transmitter attached to this
equipment. Gauges were verified by comparing the pressure showing on the
gauge with the same pressure showing on a NIST-traceable pressure gauge. The
NIST-traceable pressure gauge was connected to the same port via an in line "T".
Visual inspection particle counter and turbidimeter tubing for unimpeded flow
and integrity.
Further descriptions of these verifications are provided in the results and discussion sections
below.
3.8.2 On-Site A nalytical Methods
Specific Instrumentation methods for on site QA/QC accuracy were conducted during the
verification testing. Water quality parameters were measured by analytical or instrument
methods outlined in SM. On-site instruments were calibrated daily. Sample ports and sampling
techniques remained consistent.
3.8.2.1 pH
pH was recorded in accordance with SM 4500-H . The pH meter calibration was verified daily
with a two-point calibration against NIST-traceable pH standards at pH 7.0 and pH 10.0.
31
-------
3.8.2.2 Temperature
Temperatures were recorded daily with a NIST-traceable thermometer accurate to 0.1 °C, as per
SM2550. The temperature was taken by immersing the thermometer to an index line scribed on
the body into running water and allowing the mercury to stabilize. The thermometer was held
upright during the readings.
3.8.2.3 Turbidity
SM2130 was used for both the bench-top and on-line turbidimeter. An on-line turbidimeter was
correlated to a portable bench-top turbidimeter. The bench-top turbidimeter was calibrated at the
beginning of the verification period, and daily thereafter against secondary standards generated
by the calibration procedure, and then also against secondary standards of 0.1, 0.5, and 3.0 NTU.
Since the measurement systems are different, it was not necessary to have identical readings
between the bench-top and on-line turbidimeters however; measurements should be and were
consistent and comparable.
Samples were collected from a sample tap at a slow steady stream and along the side of a triple
rinsed dedicated beaker to avoid air entrapment. The sample was poured from the beaker into a
double rinsed clean sample vial and inserted into the chamber. This was repeated for influent
and effluent samples, and the reading of the on-line turbidimeter was noted when the sample was
drawn
All glassware for turbidity measurements was kept clean and handled with lint-free laboratory
tissue. Sample cells were additionally wiped with a silicone oiled velvet cloth.
3.8.2.4 Particle Counting
Particle counting is a rapid and efficient means of determining with some accuracy the size
distribution and enumeration of particles in a sample. While it conveys more information than
turbidity, it cannot alone identify the source or nature of any particle matter. The manufacturer
generally calibrates particle counters against NIST microspheres. Particle counters used on site
had a factory calibration certificate dated March 3, 2000, serial numbers 971000353 and
971000354. Calibration was again verified on site with NIST mono-sized polymer
microspheres.
Dilution water was prepared by filtering commercially prepared deionized water through 0.2
micron filters. To one liter of dilution water an amount of particle suspension was added to
measure approximately 2,000 particles per milliliter. The particle sizes were NIST-traceable for
size and included 3|j,m, 10|j,m and 15|j,m particles. Batch and true sizes are noted in the logbook
as follows:
Duke Scientific Corp 3.0 ± 0.027 |j,m
10.0 ± 0.061 |j,m
15.0 ± 0.08 |j,m
32
-------
Particle counter verification was performed for size distribution only, although counts were
corroborated. Particle counters cannot be field verified for count accuracy.
The procedure for monosphere verification noted in the test plan pertains to bench-top particle
counters, however, the procedure can be and was amended for application to on-line particle
counters as follows: Black teflon hoses as supplied by the particle counter manufacturer were
attached to the influent and effluent ports of the counter's sensor. The influent hose was inserted
into a flask containing either dilution water or the particle suspension, and the effluent hose
attached to a metering pump.
A suspension containing 2,000 measured particles per milliliter of a single size was prepared.
Dilution water was suctioned through the particle counter and the pump rate adjusted to 100
mL/min. The influent line to the particle counter was fed the low particle dilution water for
several minutes, until the lines were flushed and a background count was obtained. When the
counts and flows were stable, the influent hose was switched to the particle suspension, which
was mixed gently with a magnetic mixer. Those particle counts were logged and the distribution
noted to assure separation into the proper particle count bin, and the time noted for correlation to
the computer data recorder. After several sensor readings (determined by the volume of
suspension and the counter sample frequency), the hose was switched back to the dilution water
to clear the sensor and to stabilize the counter.
This procedure was performed eight times, four each for the influent and effluent counters.
Although the test plan specified 2 |j,m, 10 |j,m and 15 |j,m sizes, CO A requested of NSF that the 2
|j,m size be replaced with 3 |j,m particles. Particle counting is done by segregating the particles
into bins and since the lower limit of the counter was 2 |j,m, the count of particles at that level
would be uncertain. The verifications were then performed with 3 |j,m, 10 |j,m 15 |j,m mono-
sizes, and once with a mixture of all three sizes at the 1,000 particles per milliliter, or 3,000
counts/mL total.
The results of this verification procedure are discussed and displayed in Chapter 4 of this report.
During the procedure, the flow was carefully controlled at 100 mL/min, and exceptions noted
since reductions or increases in the flow rate alter the counts significantly.
Maintenance of the particle counter is important. Manufacturer recommended maintenance was
followed and noted in the logbook.
Procedures for particle counting were those as noted in SM 2560 (and subsections appropriate to
the equipment in use).
3.8.2.5 Particle Free Water
Particle free water (PFW) was a necessary component of the testing procedure and was prepared
fresh and as often as storage limitations will allow. Fresh PFW was necessary to limit biological
growth that could affect the particle counts. The PFW for this study was initially commercially
available deionized water that had been additionally filtered through a 0.2 |j,m cartridge filter.
33
-------
Field conditions made the production of PFW in accordance with SM difficult, however,
although commercially prepared DI water, filtered on site thorough a 0.2 |j,m filter was
considered suitable for particle counting and other reagent preparation in this application. This
water was used for the NIST-traceable suspensions used to verify the particle counter accuracy.
In the case of the seeding suspensions however, particle free water, even DI water filtered
through a 0.2 |j,m filter, was subject to contamination by airborne particles. Following
consultation with the particle counter manufacturer, the FTO used the test equipment effluent
water as dilution water for the seed suspension. This was deemed preferable to DI water since it
had the same chemical composition as the feed water and the test equipment effluent contained
near 100 particles per mL measuring between 2 to 7 microns in diameter. Particle count of the
suspension was near 4.4 million/mL. Thus, there were 4.4 million particles/mL of seed to 100
particles/mL possible background particles in the dilution water. As with turbidity, glassware
associated with the particle counters was dedicated and cleaned with laboratory glassware
detergent, and then triple rinsed with PFW.
3.8.3 Off-Site Analysis For Chemical and Biological Samples
Tables la and lb of the Code of Federal Regulations 40, Parts 136.3 cross-reference SM, EPA
methods, American Society for Testing and Materials (ASTM) methods and U.S. Geological
Survey (USGS) methods. Spectrum Labs follows EPA, SM or other accepted methodology for
all of their analytical procedures. For example, to analyze alkalinity, EPA method §310.1 is
used; this correlates to SM 2320B, which is the same as ASTM 1067-92 and USGS i-1030-85.
All four of the testing methods are the same.
3.8.3.1 Organic Parameters, Total Organic Carbon and UV Absorbance
Samples for examination were collected in glass bottles furnished by the laboratory, prepared as
in SM 5010B and shipped at 4°C to Spectrum Labs within 8 hours of collection. Samples were
analyzed at the laboratory for TOC by EPA method 415.1. UV254 was analyzed using SM
5910B.
3.8.3.2 Microbial Samples: Coliform and Algae
Since the feedwater is surface water, microbiological samples were collected for analysis of
coliform bacteria and algae. Samples were collected in glass bottles supplied by Spectrum Labs
and kept at 4°C in the proper shipping cooler. Because the travel time was so brief and the
samples were cooled, Spectrum Labs decided it was not necessary to use Lugol's solution as a
preservative. Total Coliform Bacteria were analyzed at the laboratory using SM 9222B, algae
analyzed using SM 10200F (when algae were found, SM 10900 was used for speciation), and E.
coli Bacteria were analyzed using SM 9221F.
3.8.3.3 Inorganic Samples
Inorganic Samples were collected, preserved and shipped in accordance with SM3010B and C
and 1060 and EPA §136.3, 40 CFR Ch.l. Proper bottles and preservatives where required (Iron
34
-------
and Manganese for example) was used. Sample bottles for metals analysis were supplied by the
laboratory. Although the travel time was brief, samples were shipped cooled. Samples were
analyzed at the laboratory in accordance with the following methods: total alkalinity - EPA
method §310.2, color - EPA method §110.2, total hardness - EPA method §130.1, iron and
manganese used EPA method §200.7
3.8.3.4 Microspheres
The samples for microscopic analysis were shipped to Debra Huffman Environmental
Consulting. At Dr. Huffman's laboratory they were examined microscopically and the
fluorescent spheres were counted using hemacytometer techniques and/or membrane filtration as
appropriate. The hemacytometer was used when the samples contained high numbers of
particles (influent samples); when the counts were low (effluent samples) the particles were
counted microscopically after filtration through a membrane. Hemacytometer and membrane
filtration counting was performed as outlined in EPA Method 1622, Section 11.3, EPA Method
1623, Section 11.0, and SM 10200F. Hemacytometer and membrane filtration counting was
performed microscopically at the laboratory. EPA Methods 1622 and 1623 refer to the use of
live cysts and oocysts, and SM 10200F refers to live organisms as well. Because this study
employed synthetic microspheres, the requirement of preserving, dying and handling specific to
live organisms was unnecessary. Accordingly, the techniques employed were those covered by
standard microscopic evaluation procedures as outlined in SM 10200, but without the need for
techniques specific to live organisms.
3.8.3.5 True Color
True color was measured in accordance with SM2120 with a spectrophotometer at 455 nm. The
samples were collected in glass vials and maintained at a temperature of 4°C during shipment to
Spectrum Labs. The samples were warmed to room temperature before analysis. Samples were
analyzed in accordance with EPA method §110.2.
35
-------
Chapter 4
Results and Discussion
4.1 Introduction
Initial operations took place in two phases. The first phase began June 24, 1999 and ended July
9, 1999. The second phase began January 10, 2000 and ended on January 20, 2000. The
verification testing for the RPI GFS Filter System commenced on March 7, 2000, and concluded
a 32-day testing period on April 20, 2000. Microsphere challenge testing was conducted
between April 6 and April 20, 2000 in two sessions, the first using regular polystyrene (latex)
microspheres, and the second, from April 16 through April 20, 2000, using fluorescent
microspheres.
This section of the verification report presents the results of the testing and offers a discussion of
the results. Results and discussions of the following are included: initial operations, equipment
characteristics, finished water quality, polystyrene microsphere surrogate removal, and QA/QC.
4.2 Initial Operations Period Results
The initial operations period allowed the manufacturer to characterize feed water quality and to
optimize treatment efficiency of the equipment. This period was also used for filter element
variability testing to identify variations in system performance attributable to the manufacturing
process of the rigid cartridge filters identified by the manufacturer as the physical barrier of
parasitic cysts within the equipment package. The bag filter was identified as a pre-filter to the
rigid cartridge filter within the equipment package. Accordingly, bag filters from the same
manufacturing lot were used throughout variability testing within, and between, manufacturing
lots of rigid cartridge filters.
The filter element variability testing period was divided into two phases. Phase I was designed
to determine variations within a manufactured lot number of cartridge filter elements. Phase II
was designed to show variations between manufactured lot numbers of cartridge filter elements.
Prior to these two phases, the feedwater was characterized for suitability to this technology.
There is a certain imprecision in the manufacture of filter elements causing pore sizes to vary
slightly. To account for this variability, the RPI GFS Filter System containing filter elements
from same manufacture lot number were subjected to the same influent water conditions during
this phase of the testing. Particle counts were monitored to check filter lot performance.
Turbidity was measured not as a surrogate, but because it represents the cumulative effect of
substances that had the ability to load filters and shorten run times, thus contributing to the
overall performance.
4.2.1 Characterization of Feed Water
Historical untreated surface water quality data was obtained from the City of Minneapolis,
Municipal Water Works department, for the same time frame as the verification testing period
(March and April of 1997). The untreated surface water exhibited the following characteristics:
36
-------
the temperature varied from 0.3°C to 13.2°C; pH was in the range of 7.6 to 8.2; total alkalinity
averaged from 103 mg/L to 169 mg/L; total hardness averaged between 122 mg/L and 188 mg/L;
true color averaged between 31 and 69 TCU and turbidity averaged between 5.2 and 18.6 NTU.
During Initial Operations, between June 29, 1999, and July 9, 1999, untreated river water was
blended with finished Minneapolis drinking water to achieve a level of turbidity near 1 NTU.
Chlorination of blended water was maintained near 0.5 mg/L by injection of sodium
hypochlorite. Additional characteristics of this blended water are as follows: Temperature
varied from 22.9°C to 26°C; pH averaged 8.5; total alkalinity averaged 45 mg/L; total hardness
of averaged 78 mg/L; TOC concentration less than or equal to 6.1mg/L; UV254 Absorption of
0.1; true color of 15 TCU; total coliform was not detected or was detected below the PQL of 1
CFU/lOOmL, iron was not detected or was below the PQL of 0.1 mg/L; manganese was less than
0.02mg/L; and the conductivity was 360 |j,mhos/cm. Algae were not detected or were below the
PQL of 1 Algae/mL in the influent or effluent sample waters in the initial operations testing
period.
4.2.2 Initial Test Runs
COA and RPI supervised the installation of the equipment and start-up, and established operation
of the test station. Trial runs of the system were performed.
During this period COA additionally calibrated and standardized the test station and evaluated
general functionality the station and specifically that of the untreated/treated water blending
system. These exercises were performed previous to the controlled ETV test period and
accordingly had no strict format or reporting requirements.
4.2.3 Filter Element Variability Testing
Phase I of filter element variability testing began at 12:33 on June 24, 1999. Cartridge filters
with the same manufacturing lot number (88-4546) were inserted into the three filter trains. The
flowrate was 20 gpm per filter, and turbidity was controlled by blending raw river water with
finished city water to achieve approximately 3.0 NTU as measured by on-line turbidimeters.
Terminal headloss had been established by the manufacturer at 15 psi across each filter train,
consisting of a bag filter as a pre-filter, and the rigid cartridge filter, described above. Although
each filter housing had an individual pressure differential gauge, it was the sum of the pressures
across both housings that was registered by the pressure gauges located on the instrument panel
that were used to determine system losses. At 17:58 on June 24, the operator's shift was
complete and the equipment was left in operation until an operator returned the following
morning.
By 9:15 June 25, the three filters showed pressure losses of (#1) 7.5 psi, (#2) 30 psi and (#3) 12.5
psi. Flowrates in the three systems had decreased to (#1) 19.5 gpm, (#2) 15.5 gpm and (#3) 12.5
gpm.
By 9:49 (within 34 minutes) the filters had the following headlosses (#1) 24 psi, (#2) 37 psi and
(#3) 31 psi and the three filter trains were shut down for cleaning and filter element replacement.
The cumulative volumes of water filtered at time of shutdown were:
37
-------
Filter System 1 25,899 gallons
Filter System 2 21,997 gallons
Filter System 3 25,083 gallons
New bags and cartridges of the same manufacturing lot number were installed and the filters
placed on line at 12:45. By 15:00 (within 3 hours, 15 minutes) the filter loading rates were
excessive and it was clear that terminal head loss would be achieved within several hours after
the operator's shift concluded. Accordingly, the system was shut down at that time (June 25, at
15:00 hours) to reevaluate the operating parameters.
After discussions with the manufacturer, it was decided that influent turbidity should be reduced
to an average of 1 NTU and the flow rate decreased to 10 gpm. It was also decided that non-
blended Minneapolis finished drinking water serve as the feed water overnight when the
equipment was unattended by an operator. Further, due to concerns expressed by the
manufacturer with the previous lot of cartridges, the manufacturer also provided replacement
cartridges with a different manufacturing lot number (6-2-99).
Following cleaning and purging, the system was restarted with the new operating parameters and
cartridges on June 29, 1999. Filter runs conducted after influent turbidity was reduced to 1 NTU
and flow rate to 10 gpm are summarized in Table 4-1.
Table 4-1. Phase I Initial Operations Filter Run Summary
Filter System #
Dates
First Elements Gallons
Element Replacement
Second Element Set
(total)
(Stop time)
Gallons
1
6/29-7/7
229,472
7/7 16:00
50,997
2
6/29-7/7
250,726
7/8 10:00
29,960
3
6/29-7/7
214,063
7/7 9:00
58,100
Headlosses for the second set of three filters at the end of the 10-day period were: (#1) 4 psi, (#2)
5 psi and (#3) 12 psi. These filter runs were shortened due to the conclusion of Phase I of the
initial operations.
Phase I concluded with 261 hours, 32 minutes (10.90 days) of equipment operation. Based on
the results of Phase I, the manufacturer elected to address concerns pertaining to the
manufacturing process of the rigid cartridge filter element (model number GLR-PO-825-2).
Subsequently, for Phase II of filter element variability testing, the manufacturer provided rigid
cartridge filter elements with a different model number (PL-520-PPP141) and internal seals
within the filter housing. Because the rigid cartridge filter and housing seals serve as primary
components within the equipment package their replacement fundamentally changes the
description of equipment package being marketed by the manufacturer. Accordingly inclusion
of specific performance data collected during phase I is omitted from this report.
Phase II of the filter element variability testing began on January 10, 2000 at 14:20 and
continued through January 20, 2000 to 18:30 for a total period of operation of 244:10 hours
(10.21 days). The purpose of Phase II was to observe variability between manufacturing lots of
rigid cartridge filters. Rigid cartridge filters from 3 different manufacturing lots were used
(990541-5, 990541-4, 990541-3). The bag filters, used as pre-filters within the filter train, all
38
-------
were from the same manufacturing lot (P0525A2). In addition to the amendments made to the
equipment package, influent turbidity was reduced to an average of less than 1 NTU and was
maintained at that level 24 hours/day. The filter flow rate was also reduced from 20 to 10 gpm.
Because of the improbability that terminal head loss would occur across each filter train within
the 8 hours per day an operator was present, filters within all filter trains were replaced
simultaneously when terminal head loss occurred within one filter train. It should be noted that
while this practice inherently created performance data that understates true treatment capacity
for each treatment train, differences in head losses (psi) recorded previous to each filter
replacement suggest variability in treatment capacity between filter trains (refer to Table 4-4).
During Phase II the same schedule of data recording was followed as for Phase I with two shifts
of four hours each, with each shift separated by several hours. Performance results for the
individual filter runs for each of the three filter trains for Phase II are summarized in Tables 4-2,
4-3 and 4-4. Table 4-5 summarizes the operating data for individual filter runs for each of the
three filter trains.
Table 4-2. Phase II Variability Testing RPI GFS Filter System Run #1 Particle Count & Turbidity Results
Filter Train 1
Filter Train 2
Filter Train 3
Filter Run
Mfg. Lot 990541-5
Mfg. Lot 990541-4
Mfg. Lot 990541-3
Ave.
Std.
95%
Ave.
Std.
95%
Ave.
Std.
95%
Dev.
Conf. Int.
Dev.
Conf. Int.
Dev.
Conf. Int.
Influent 3 -7 |im
1,466.37
290.70
1,354.63,
1,474.61
247.52
1,379.46,
1,446.40 278.77
1,339.25,
Particle Counts
1,578.11
1,569.75
1,553.56
(counts/mL)
Effluent 3-7 |im
63.10
24.65
53.62,
106.26
45.00
88.96,
36.65
15.17
30.82,
Particle Counts
72.58
123.56
42.48
(counts/mL)
Particle Count
1.4
0.2
1.3, 1.5
1.2
0.2
1.1, 1.2
1.6
0.2
1.6, 1.7
Logio Removal
Influent On-Line
0.85
0.09
0.81, 0.88
0.86
0.08
0.83, 0.89
0.85
0.09
0.81,
Turbidity (NTU)
0.88
Effluent On-Line
0.31
0.04
0.30, 0.33
0.32
0.03
0.31, 0.34
0.30
0.051
0.28,
Turbidity (NTU)
0.32
Table 4-3. Phase II Variability Testing RPI GFS Filter System Run #2 Particle Count & Turbidity Results
Filter Train 1
Filter Train 2
Filter Train 3
Filter Run
Mfg. Lot 990541-5
Mfg. Lot 990541-4
Mfg. Lot 990541-3
Ave.
Std.
95%
Ave.
Std.
95%
Ave.
Std.
95%
Dev.
Conf. Int.
Dev.
Conf. Int.
Dev.
Conf. Int.
Influent 3 -7 |im
1,322.42
603.39
1,098.93,
1,312.02
604.08
1,088.27,
1,337.53
567.25
1,127.42,
Particle Counts
1,545.92
1,535.77
1,547.63
(counts/mL)
Effluent 3-7 |im
39.53
22.62
31.15,
42.29
24.12
33.93,
31.83
17.93
25.19,
Particle Counts
47.91
51.22
38.47
(counts/mL)
Particle Count
1.5
0.4
1.3, 1.6
1.5
0.4
1.3, 1.6
1.6
0.4
1.5, 1.8
Logio Removal
Influent On-Line
0.95
0.19
0.88. 1.03
0.93
0.09
0.90, 0.96
0.93
0.08
0.90, 0.96
Turbidity (NTU)
Effluent On-Line
0.31
0.05
0.29, 0.33
0.32
0.03
0.31, 0.33
0.31
0.02
0.30, 0.32
Turbidity (NTU)
39
-------
Table 4-4. Phase II Variability Testing RPI GFS Filter System Run #3 Particle Count & Turbidity Results
Filter Train 1
Filter Train 2
Filter Train 3
Filter Run
Mfg. Lot 990541-5
Mfg. Lot 990541-4
Mfg. Lot 990541-3
Ave.
Std.
95% Conf.
Ave.
Std.
95% Conf.
Ave.
Std.
95% Conf.
Dev.
Int.
Dev.
Int.
Dev.
Int.
Influent 3 -7 |im
658.47
346.41
522.68,
628.99
350.85
491.46,
616.75
347.75
480.44,
Particle Counts
794.26
766.52
753.07
(counts/mL)
Effluent 3-7 |im
22.99
12.59
18.05,
78.38
24.07
68.94,
19.94
6.67
17.32,
Particle Counts
27.92
89.81
22.55
(counts/mL)
Particle Count
1.4
0.4
1.3, 1.6
0.81
0.40
0.66, 0.97
1.4
0.4
1.3, 1.5
Log10 Removal
Influent On-Line
0.77
0.13
0.72, 0.82
0.78
0.13
0.73, 0.83
0.78
0.13
0.73, .83
Turbidity (NTU
Effluent On-L ins
0.33
0.10
0.29, 0.37
0.35
0.07
0.33, 0.38
0.32
0.08
0.29, 0.36
Turbidity (NTU)
While these results suggest some variability in particle reduction performance between filter
trains, the degree of variability can be attributed to variations in filter ripening caused by the
method selected to initiate filter replacement, described above. Particle reduction performance
generally improves as terminal head loss is approached. Accordingly, when the filters were
replaced previous to terminal head loss, lower logio reductions were noted. This is observed in
run #1 (Table 4-2). Head losses at time of filter replacement on January 13 were 12 psi, 8 psi,
and 15 psi respectively for filter trains 1, 2, and 3. Corresponding particle count logio reductions
were 1.4, 1.2, and 1.6.
During Phase II run # 2, filter train 1 approached terminal head loss near the end of the daily data
collection period. Leaving the existing filters in operation over night would have caused head
losses to significantly exceed terminal head loss previous to the next data collection shift the
following morning. Accordingly, all filters were replaced. Head losses at time of filter
replacement on January 17 were 12 psi, 8 psi, and 9 psi respectively for filter trains 1, 2, and 3.
Corresponding particle count logio reductions were 1.5, 1.5, and 1.6.
The conclusion of the Phase II filter variability testing period on January 20 caused the
termination of run # 3. Head losses at time of shut-down were 6 psi, 6 psi, and 5.5 psi
respectively for filter trains 1, 2, and 3. Corresponding particle count logio reductions were 1.4,
0.81, and 1.4. During this filter run influent particle counts were significantly lower than what
was observed during runs # 1 and # 2. Also, filter train 2 demonstrated comparatively poor
particle reduction performances. This was attributed to a faulty pressure differential gauge.
Table 4-5 is a summary of the volumes of water treated and the terminal headloss in the Phase II
runs.
40
-------
Table 4-5 Phase II RPI GFS Filter System Filter Operating Data
Filter Run# Filter Train 1 Filter Train 2 Filter Train 3
Mfg. Lot 990541-5 Mfg. Lot 990541-4 Mfg. Lot 990541-3
Run #1
Total Water Processed (gallons)
79,265.5
82,604.0
79,664.6
Max. Change In Pressure Drop (psi)
12.0
8.0
15.0
Run #2
Total Water Processed (gallons)
57,826.0
62,020.0
61,767.0
Max. Change In Pressure Drop (psi)
12.0
8.0
9.0
*Run #3
Total Water Volume (gallons)
44,338.2
43,925.6
44,743.4
Max. Change In Pressure Drop (psi)
6.0
6.0
5.5
* Filter Run #3 discontinued before to terminal headloss.
When the operating data described in Table 4-5 is compared to particle reduction data described
in tables 4-1 and 4-2 it becomes somewhat evident that rigid cartridge filters used in filter train #
2 (lot # 990541-4) offered greater loading capacity. These data also suggest inconsistencies in
performance from rigid cartridge filters used in filter train # 3 (lot 990541-3). Due to the limited
number of filters evaluated within each manufacturing lot, conclusions regarding variation in
filter performance between manufacturing lots cannot be offered with any degree of certainty.
It was also noted that filter train # 2 offered lower particle reduction values. Upon later
examination, it was determined that of the Orange Research pressure differential gauges installed
within each filter train, the one installed in filter train # 2 was faulty and had been bypassing
influent water into the filtered water stream.
4.3 Verification Testing Results and Discussions
The results and discussions of testing runs, routine equipment operations, feed and finished water
quality, operating conditions and equipment performance, and microbiological removal tasks of
the verification testing are presented below.
4.3.1. Task 1 - Verification Testing Runs And Routine Equipment Operation
The objective of this task was to operate the equipment provided by the manufacturer for a
minimum 30-day testing period and assess its ability to meet water quality goals and other
performance characteristics specified by RPI.
The verification testing for the RPI GFS Filter System began on March 7, 2000, and ended its
32-day period on April 20, 2000. During the testing period, one RPI GFS Filter system was
operated for 23-hours each day with flow stopped for one hour each day. The duration of each
filter run from start to terminal headloss and the number of gallons of water produced during
each run are summarized in the results discussion for Task 3, Section 4.2.3.
4.3.2 Task 2 - Influent and Effluent Water Quality Characterization
Results of testing for turbidity in the influent and effluent water were examined to verify the
manufacturer's stated turbidity treatment ability. Examination of TOC and UVA254 testing
41
-------
results, as well as testing results for the inorganic parameters total alkalinity, total hardness, true
color, total coliform, iron and manganese are shown in Table 4-6. Samples for E. coli were
collected on March 13, 2000 and March 30, 2000. E. coli was not detected or was below the
PQL of 1 CFU/lOOmL. Samples for Aluminum analysis were collected on March 13, 2000;
Aluminum was found in the influent samples at 0.83 mg/L.
Table 4-6. Influent Water Quality (March 7 - April 20, 2000)
Parameter
#of
samples
Average
Minimum
Maximum
Standard
Deviation
95%
Confidence
Interval
PQL
Total Alkalinity (mg/L)
7
70
55
110
18
55, 78
10 mg/L
Total Hardness (mg/L)
7
94
82
130
16
82, 107
10 mg/L
True Color (TCU)
7
14
10
25
6
10, 18
1 TCU
Total Coliform (CFU/100/mL)
7
24*
<1
110
40
<1, 54
1 CFU
TOC (mg/L)
7
7.8
6.8
11.0
1.4
6.7, 8.9
0.4 mg/L
UVA254 (cm4)
7
0.140
0.108
0.229
0.042
0.109,0.171
-
On-Line Turbidity (NTU)**
-
1.1
0.7
1.5
0.2
1.0, 1.2
-
Total Chlorine (mg/L)
27
1.4
0.7
3.5
0.82
1.1, 1.7
-
Free Chlorine (mg/L)
27
0.6
0.1
2.5
0.6
0.4, 0.8
-
Iron (mg/L)
7
0.1*
<0.1
0.4
0.1
<0.1, 0.2
0.1 mg/L
Magnesium (mg/L)
7
0.02*
<0.01
0.04
0.01
0.01, 0.03
0.01 mg/L
Temperature (°C)
38
7.3
3.9
11.0
2.2
6.7, 8.0
-
PH
37
8.5
8.0
8.9
0.2
8.4, 8.5
-
*A11 calculations involving results with below PQL values used half the PQL in the calculation.
** Turbidity values are the on-line values and the average results of each filter run.
One influent water sample for Total Coliform Bacteria did not contain a 100 mL water sample;
therefore a 90 mL analysis was performed. Drinking water compliance samples (SDWA) must
be 100 mL volumes to report <1 coliform/lOOmL. This sample analysis must therefore be
reported as
-------
coliform, iron and manganese are shown in Table 4-8. No algae were detected in the finished
water samples. Laboratory reports are provided in Appendix E.
Table 4-8. Effluent Water Quality (March 7 - April 20, 2000)
, # Of .
Parameter , Average
samples b
Minimum
Maximum
Standard
Deviation
95% Confidence
Interval
PQL
Total Alkalinity (mg/L)
7 66
54
100
16
55, 78
10 mg/L
Total Hardness (mg/L)
7 95
82
130
16
83, 107
10 mg/L
True Color (TCU)
7 10
5
15
4
7, 13
1 TCU
Total Coliform (cfu/100 mL)
7 2*
<1
6
3
<1, 4
1 cfu/100 mL
TOC (mg/L)
7 7.5
6.4
8.8
0.8
6.9, 8.1
0.4 mg/L
UVA254 (cm-1)
7 0.130
0.109
0.156
0.017
0.117,0.143
-
Iron (mg/L)
7 0.1*
<0.1
0.6
0.2
<0.1, 0.3
0.1 mg/L
Manganese (mg/L)
7 0.1*
<0.01
0.04
0.01
<0.01,0.02
0.01 mg/L
*A11 calculations involving results with below PQL values used half the PQL in the calculation.
Several times throughout the testing period turbidity spikes were observed that could be directly
related to work being performed on the water distribution system by MWW staff. These
turbidity spikes occurred when the city pressure dropped below 100 psi and additional city
pumps were brought on line. These spikes did not make any significant changes in the total run
averages though. On March 31, it was observed that MWW workers were in the process of
cleaning the intake screens. The cleaning of the screens may have had an effect on the turbidity
readings for the period between March 23 and April 2. During the period of March 23 to April
2, the average turbidity was 1.16 NTU, the highest average for the testing period. The turbidity
before this cleaning averaged 0.73 NTU, and the turbidity between April 3 and April 20
averaged 1.14. Turbidity spikes were unpredictable and gave no warning, thus operators could
not quickly respond to adjust the blended water.
The influent turbidity between April 16 and April 20 may have been elevated due to increases in
algal blooms. During the last run (#22), the average influent turbidity was 1.46 NTU, again
likely due to algal blooms in the river water. The increase in turbidity and particle count
continued past the verification period.
Water temperature of the blended Mississippi River water and the MWW plant water varied
considerably during the verification period, due to the river water temperature warming up with
the season. A high of 11 °C, and a low of 3.9°C were measured in the influent water (average of
7.3° C). At one point (in January during Phase II) the raw Mississippi river water was only a few
tenths above the freezing point. Feed water temperatures lagged river water temperatures by
several days due to interim storage following treatment. Water temperature did not steadily
increase during the period, but advanced and declined as the air temperature changed.
The pH of the feed water was stable during the testing period. pH ranged from a low of 8.0 to a
high of 8.9 with an average pH of 8.4.
43
-------
4.2.3 Task 3 - Documentation Of Operating Conditions And Treatment Equipment
Performance
The purpose of this task was to accurately and fully document the operating conditions during
treatment and the performance of the RPI GFS Filter System during verification testing. Table
4-9 lists the operating parameters that were documented during the verification testing period.
Table 4-9. Operating Parameters (Summary of 22 Filter System Runs)
Parameter Average Minimum Maximum Std Deviation
95% Confidence
Interval
Flow Rate (gpm)
Gallons per filter run
9.7
22,789.1
5.0
10,980.0
11.0 1.0
74,173.4 15,434.2
9.5, 9.9
16,339.7, 29,238.5
The influent water flow rate for the verification period averaged 9.7 gpm. The flow naturally
declined as the pressure drop across the filters increased; the flow was adjusted as per the test
plan whenever it fell by more than 10%, occasional lower readings resulted from lower flows
noted prior to adjustment.
Wastes consisted of spent filter elements, which were examined and then disposed of in solid
waste containers on site. Filter elements are not considered hazardous and could be included
with other site trash. Effluent water was directed to the Metropolitan Sewer System.
The RPI GFS Filter System had no power requirements. Therefore, the daily power
consumption of the treatment system was not recorded.
Table 4-10 summarizes each filter run of the RPI GFS Filter System during the verification
period. Note that filter runs 17 and 18 data are not included in the averaging because the FTO
had run out of cartridge elements thus only the bag was replaced (water usage data is included
for comparison only, and is not used in averages). This interruption was due to a shipping delay
of additional bags and cartridges from the manufacturer.
44
-------
Table 4-10. Filter Run Averages For Verification Period (March 7 - April 20)
Average Total
Influent
Average Total
Effluent
Average
Influent
Turbidity
(NTU)
Average
Effluent
Turbidity
(NTU)
Max Pressure
Change (psi)
Water
Filter Run
Run Number
Particles (2-15
Urn)
(counts/mL)
Particles (2-15
Urn)
(counts/mL)
Produced
(Gallons)
Duration
(hours)
Filter Run 1
4,561.81
43.66
0.81
0.20
16
74,173.4
135.25
Filter Run 2
3,783.90
78.01
0.77
0.22
11
34,525.6
54.25
Filter Run 3
4,174.34
76.70
0.68
0.20
16
50,564.4
**82.5
Filter Run 4
6,235.43
178.05
1.05
0.26
14.75
16,488.4
27.00
Filter Run 5
6,124.01
39.06
0.97
0.21
14
19,567.1
31.75
Filter Run 6
6,430.37
299.54
1.03
0.22
12
23,824.8
39.50
Filter Run 7
6,947.62
60.14
1.06
0.23
16
27,846.1
45.50
Filter Run 8
8,268.76
63.93
1.17
0.18
14
19,840.7
32.50
Filter Run 9
8,719.92
57.79
1.34
0.20
25.5
14,393.6
23.50
Filter Run 10
8,695.56
124.43
1.28
0.21
21
21,245.7
35.25
Filter Run 11
6,750.24
65.28
0.93
0.19
16
29,920.0
52.00
Filter Run 12
9,290.52
112.39
1.33
0.21
15
12,690.0
19.75
Filter Run 13
8,452.42
60.80
1.16
0.21
17
14,950.0
25.50
Filter Run 14
7,541.34
69.35
1.06
0.19
20
14,020.0
23.25
Filter Run 15
7,268.12
90.02
0.97
0.17
16
17,042.4
27.75
Filter Run 16
7,935.23
85.53
0.99
0.20
12
12,454.8
19.25
Filter Run 17
-
-
-
-
-
15,264.7*
-
Filter Run 18
-
-
-
-
-
6,895.3*
-
Filter Run 19
8,486.83
89.03
1.12
0.22
22.5
14,665.6
23.75
Filter Run 20
9,385.81
108.55
1.26
0.23
18
10,980.0
19.25
Filter Run 21
7,463.88
67.38
1.10
0.21
15
14,150.0
23.25
Filter Run 22
10,055.87
47.60
1.46
0.23
15
12,440.0
19.50
Average
7,328.60
90.86
1.08
0.21
16.3
22,789.1
38.04
Minimum
3,783.90
39.06
0.68
0.17
11.0
10,980.0
19.25
Maximum
10,055.87
299.54
1.46
0.26
25.5
74,173.4
135.25
Std Dev.
1,737.35
58.74
0.20
0.02
3.6
15,434.2
27.76
95%
6567, 8090
65, 117
0.98,1.16
0.20, 0.22
14.8, 17.9
16,339.7,
25.88, 50.18
Confidence
29,238.5
* - Runs 17 & 18 water usage data not included in totals, data shown for comparison information only.
** Run 3 was interrupted one hour to replace the raw water pump, and 25 hours because of turbidimeter failure. The
raw water pump failure, which was not noticed immediately, had the effect of reducing the total flow through the
system, and of eliminating the raw water flow so only finished city water was supplied. This accounts for the long
run time to terminal headloss.
Note: Particle counter calibration and verification procedures were performed in Runs 5, 11, 15, and 17.
Filter runs 20, 21, and 22 were fluorescent microsphere challenge runs that are described in more
detail in Task 4.
In the following text the verification runs are summarized showing turbidities, number of gallons
processed, pressure drops and other relative comparisons. Included are explanations of
interruptions and exceptions that should be considered when reviewing the run data.
Turbidity removals were consistent and generally good throughout the verification period.
Following a brief ripening period, on-line turbidity on average over the twenty-two filter runs
was: 1.08 NTU influent and 0.21 NTU effluent for an average 0.64 logio reduction.
45
-------
Figure 4-1 shows the relationship of the influent turbidity to the number of gallons processed per
filter run. Note, filter run 17 and 18 data are not graphed in the Figure 4-1.
1.60
1.40
1.20
1.00
~ 0.80
¦g
= 0.60
0.40
0.20
¦
~
~
~
~
~
¦
~
n
~
u
u
~
~
~
u
~
u
~
~
U
¦
~
¦
¦
¦
¦
1
¦
1
¦
¦
¦
¦
¦
¦
¦
¦
¦
¦
1—I—I—I—I—I—I—I—I—I 1—I—I—I—I—l-B-l-H—l—l—l
80,000
70,000
60,000 "l
50,000 &
40,000 «
30,000 =
_o
20,000 O
10,000
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Filter Run
~ Average Influent Turbidity ¦ Water Usage
Figure 4-1. Influent Turbidity & Gallons Per Filter Run
Figures 4-2, 4-3 and 4-4 are representative of the data recorded during the individual filter runs.
Figure 4-2 illustrates the average influent and effluent turbidity, Figure 4-3 graphs the average
logio reductions based on average influent/effluent particle counts per filter run, and Figure 4-4
presents the change in pressure drop per individual filter run. Note, filter run 17 and 18 data are
not graphed in the following three figures.
46
-------
1.60
1.20
£ 0.80
!a
0.40
• •
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Filter Run
~ Average Effluent Turbidity • Average Influent Turbidity
Figure 4-2. RPI GFS Average Filter Run Influent & Effluent Turbidity
2.50
£0 0
£ 1
<1) -Q
0£ "o
-------
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Filter Run
¦ Change in Pressure
Figure 4-1. RPI GFS Average Filter Run Pressure Drop
Additional graphs showing the influent and effluent turbidity, particle count, flow rate and
change in pressure per individual filter run are attached as Appendix F.
Run #3 was interrupted by the failure of a turbidimeter. The flow was stopped until a
replacement turbidimeter could be installed, or about 25 hours. The flow was then resumed with
no change of the filter elements and the run continued until terminal headloss. Run's #17 and
#18 received only new bag elements when headloss was exceeded. The cartridge element was
left in place.
For each filter run which lasted longer than 24 hours, a stop/start sequence was initiated. Table 4-
11 lists the time required by the RPI GFS Filter System to stabilize following a stop/start. The
stabilization period was defined as the length of time required after the resumption of flow to the
point when the effluent particle counter displayed values that were similar to those prior to the
cessation of flow. Because the particle counter displayed a new reading every minute,
stabilization periods (minutes) include an error factor of up to 1 minute.
48
-------
Table 4-11. Stop/Start Stabilization Time (based upon effluent
on-line particle count data)
Filter Run Number Minutes To Reach Stability
1.1
6
1.2
11
1.3
14
1.4
12
1.5
14
2
8
3
12
4
12
5
8
6.1
8
6.2
16
7
Q
8
Q
O
10
O
10
11.1
10
11.2
12
13
10
15
6
20
7
21
6
22
6
Average
9.4
Minimum
6
Maximum
16
Standard Deviation
3
95% Confidence Int.
8, 11
Particle count graphs for each filter run stop/start sequence are attached in Appendix F.
4.3.4 Task 4 - Microbiological Contaminant Removal
Microbiological removal capabilities were assessed by challenging the filter system with
polystyrene (latex) monospheres. Microsphere challenge testing was conducted between April 6
and April 11, 2000 in two sections, the first using regular, non-fluorescing latex microspheres,
and the second, from April 16 through 20, 2000, using fluorescing latex microspheres.
4.3.4.1 Non-Fluorescing Microsphere Challenge Results
The test plan specified that challenges be made over three filter runs, with monosphere injections
at the beginning of a run, at the approximate midpoint (as determined by headloss) and again at a
point between 90% and terminal headloss. The middle run was also to be followed by a
"stop/start" sequence, and while no additional monospheres were to be added then, samples
would be collected or particle counts enumerated to observe any particle breakthrough as the
result of the interruption and resumption of flow.
Monospheres in a quantity sufficient to demonstrate 3+logio removal were to be injected, and the
influent and effluent particle count and distribution measured and recorded by particle counter.
49
-------
Accordingly, monospheres in three sizes, 3.2 |j,m, 4.5 |j,m and 6 |j,m were mixed in the ratio of
50% 3.2 |j,m and 25% each of 4.5 |j,m and 6 |j,m to a level of approximately 10,000 total particles
per milliliter. This mixture was injected over a period of ten minutes at the rate of 100 mL/min
into the feed stream and the particle counter observed. The monospheres obtained for this
challenge are listed in Table 4-12.
Table 4-12. Monosphere Manufacturer Specification
Size Manufacturer
Lot
CV
3.2|im Duke Scientific
21693
43%
4.5|im Duke Scientific
21328
20.0%
5.832|im Polysciences, Inc.
493498
SD: 0.273|_im
The particle counter was unable to measure the additional particles. Repeated trials resulted in
the same condition: the addition of 10,000 microspheres to water containing indigenous particles
were counted by the particle counter as an addition of only 3,000 to 5,000 microspheres. It was
also noticed that there was no strict proportionality to the additions. In the initial seeding
sequence, the increase in particles was fairly evenly distributed proportionate to the added
particles, and the effluent particles were much lower, as expected. Anticipated logio removals
could not be calculated, however, due to the particle counter and computer recording only those
particles seen by the particle counters, thus undercounting the influent stream.
In continuing discussions with the particle counter manufacturer, particle manufacturers and with
the writers of the cited paper (Li and Goodrich from Li et al.) it was concluded that coincidence
errors were the probable cause of unreliable influent counts. It appears that the particle counter
cannot easily identify spherical or nearly spherical particles when included with high
concentrations of natural, irregularly shaped particles, and errors are also introduced when the
counter is inundated with particles smaller than the counter's detection limit. Other researchers
have encountered the same difficulties (Van Gelder).
A summary of the 3-7 |j,m particle counts in tabular form, showing the particle counts following
the addition of 10,000 counts/mL as measured by the on-line particle counters for challenge run
number 1 follows as Table 4-13.
Table 4-13. Average 3-7 |lm Particle Counts Non-Fluorescing Microsphere Challenge Filter Event #1
Particle Count Immediately Particle Count Peak Particle Count Immediately
Challenge Before Seeding Seeding After Seeding
(Counts/mL) (Counts/mL) (Counts/mL)
Influent Particle Counts
No headloss
3,692.6
6,401.3
3,605.6
Midpoint
4,111.2
6,881.8
4,057.2
90%
3,600.6
6,804.0
3,531.9
Effluent Particle Counts
No headloss
55.1
136.9
55.5
Midpoint
26.7
50.2
29.5
90%
18.1
35.1
19.3
Note: Particle count results are average of three on-line particle counter measurements approximately one minute
apart.
50
-------
It was observed in challenge Event #1 with non-fluorescent microspheres—and later—that the
addition of 10,000 particles, prepared in a suspension as detailed above, increased total influent
counts by only a small fraction, with about 25-30% of the added particles. The lower, effluent
particle counts can be viewed with more confidence however. Calculating logio removal of only
the 10,000 added microspheres against the measured effluent minus the averaged pre-challenge
and post-challenge background effluent counts suggests that removals of 2.1 logs, 2.7 logs and
2.8 logs were reached in this trial at the no headloss, midpoint and 90% headloss seeding events,
respectively.
The second challenge produced the same ambiguous influent particle count data and following
the zero headloss seeding, this second challenge was abandoned.
On April 11, 2000, an additional seeding challenge with non-fluorescent microspheres
(Challenge #3) was performed at the prescribed intervals, and the results of this challenge were
similar to the first, as shown in the following Table 4-14.
Table 4-14. Average 3-7 |lm Particle Counts Non-Fluorescing Microsphere Challenge Filter Event #3
Particle Count Immediately Particle Count Peak Particle Count Immediately
Challenge Before Seeding Seeding After Seeding
(Counts/mL) (Counts/mL) (Counts/mL)
Influent Particle Counts
No headloss
4,077.1
6,852.4
4,091.8
Midpoint
4,458.8
6,966.2
4,509.1
90%
4,362.5
7,019.9
4,187.8
Effluent Particle Counts
No headloss
55.7
191.2
58.3
Midpoint
19.0
28.2
22.2
90%
19.2
27.9
19.9
* Particle count results are average of three in-line particle counter measurements approximately one minute apart.
Following discussions with NSF, arrangements were then made to repeat the three series of
challenges with fluorescent microspheres, to allow enumeration via microscopic count.
4.3.4.2 Fluorescent Microsphere Challenge Results
Fluorescent spheres of the requisite sizes were not available from a single manufacturer, thus
fluorescent microspheres were obtained from three separate manufacturers, and in the sizes 3.4
|j,m, 5 |j,m and 6 |j,m. The microspheres were blended as above, with 50% at 3.4 |j,m and the
remaining 50% divided equally between the other two sizes.
The number of particles in the concentrated suspension is inversely related to the cube of the
diameter of the microspheres, and is calculated by the following formula:
6W x 1(?2
n/mL = j
p x n x 0
Where n is the number of particles, W = the grams of polymer per milliliter of latex (which
varies for each size and manufacturer, but which is noted on each container), p = density of
51
-------
polymer (1.05 for polystyrene) and 0 = the diameter of particle in microns. For the three sizes
of fluorescent particles these values were:
Diameter
Particles/mL
3.4 |j,m
4.63 x 108
5.0 |j,m
1.46 x 108
6.00 |j,m
2.23 x 108
The concentration of 3.4 |j,m particles were confirmed with the manufacturer's Certificate of
Analysis (Appendix G).
The microsphere suspensions were blended in accordance with the test plan to contain twice as
many 3.4 particles as the other two sizes via the following calculation:
At 10 gpm, over five minutes there are 1.89 x 105 milliliters, and at 10,000 particles per
milliliter, 1.89 x 109 total are required in suspension. For the 5 minutes, 500 mL injection at
10,000 particles per mL, the proportions were:
3.4 |j,m particles 2 mL = 9.26 x 108
5.0 |j,m particles 3mL = 4.35xl08
6.0 |j,m particles 2 mL = 4.46 x 108
= 18.07 x 108 total particles = 1.8 x 109
As before, the suspension was introduced in front of a static mixer over a period of five minutes
at the rate of 100 mL/min. The particle counts were observed and when the peak concentration
was reached and stabilized, forming a plateau, samples were collected from both the influent and
effluent sample streams. The effluent sample lagged the influent by about 2 minutes. Each
sample was distributed into two aliquots, one for shipment to Huffman Environmental
Consulting for examination and the other refrigerated as a back up.
4.3.4.3 On-line Particle Counter Analysis During Fluorescing Microsphere Challenge
Again, the on-line particle counters detected only a small fraction of the additional particles
introduced into the influent stream, which is similar to the situation observed during the non-
fluorescent microsphere challenge. The probable cause again was attributed to particle counter
coincidence error. An example of the seedings for the total 3-7 |j,m influent particles as read by
the particle counter during the challenge runs, is shown as Table 4-15. Figure 4-5 illustrates the
curve as shown by the particle counter.
52
-------
Representative particle count data with in-line particle counter during
fluorescent seeding challenge (Run #20, 1st seed).
Seeding challenge = 10,000 microspheres
(V
a. 5,500 .
"E
TO
Q_
V - fc> A ^ ^
-------
Table 4-16 shows an abstraction of the 3-7 |j,m particle counts in the effluent as read by effluent
particle counter during the challenge periods.
89.66
12.20
29.40
74.98
16.55
14.70
42.48
12.88
13.90
Table 4-16. 3-7 |lm Effluent Particles During Fluorescent Microsphere Challenge (Filter Runs #20, 21 & 22)
Immediately Before „ , „ , ¦ Immediately After
J Peak Seeding J
Sample ID Seeding rrmmto/mT ^ Seeding
(Counts/mL) 1 j (Counts/mL)
First Challenge Event
No headloss
Midpoint
90% headloss
Second Challenge Event.
No headloss
Midpoint
90% headloss
Third Challenge Event
No headloss
Midpoint
90% headloss
122.43
63.83
226.45
97.38
20.10
21.08
75.88
16.38
19.58
83.10
12.40
30.36
73.00
15.45
15.45
42.73
13.08
14.15
On-line particle counters, although uncertain on the influent water stream during seedings, were
valuable in confirming overall filter performance with effluent particle counts.
Figures 4-6, 4-7 and 4-8 present information on particle counts for each of the three fluorescent
particle seeding runs, showing influent and effluent at 3-7 |j,m and at 2-7 |j,m. The data points
were subjected to curve fitting via polynomial analysis to show the best fit. Note: arrows point
to seeding events.
54
-------
Fluorescent Microsphere Challenge Event #1
On-Line Particle Counts vs. Filter Run Time
50% & 90% Headloss Challenge
0% Headloss Challenge
0 H
r 0
T-OOOO h-COLO^rcOCNT-OOOO h-COLO^i-COCNT-OOCOh-COm ^ CO CM
LO
-------
Fluorescent Microsphere Challenge Event #3
On-Line Particle Counts vs. Filter Run Time
COt-cOt-COt-COt-COt-COt-COt-COt-COt-COt-COt-COt-CO
^0)COCOCNNr(£)OlOO)t(OCON(N(DT— lfiO^O)COCO(N
T-rCN0JC0C0^^tL0l0(DCDNNC0(00)0)0)OOT-
Filter Run Time (minutes)
Figure 4-8. Fluorescent Microsphere Challenge Event #3 - On-Line Particle Count vs. Filter Run Time
4.3.4.4 Microscopic Analysis
The samples shipped to Huffman Environmental Consulting were examined microscopically and
the fluorescent spheres were counted using hemacytometer techniques and/or membrane
filtration as appropriate. The results of those analyses are tabulated below. The hemacytometer
was used when the samples contained high numbers of particles (influent samples); when the
counts were low (effluent samples) the particles were counted microscopically after filtration
through a membrane. Hemacytometer and membrane filtration counting was performed as
outlined in EPA Method 1622, Section 11.3, EPA Method 1623, Section 11.0, and SM 10200F,
however by amending the procedure as appropriate to recognize the employment of synthetic,
fluorescing microspheres and not dyed/fluorescing oocysts. Tables 4-17 and 4-18 summarize the
results of the microscopic analyses for the influent and effluent samples, respectively. Based on
the microscopic analyses results presented in Tables 4-17 and 4-18, removal calculations are
provided in Table 4-19.
56
-------
Table 4-17. Influent Microscopic Analysis Results Of Fluorescent
. Enumeration Raw Count Volume
sample ID Method (Mean±S.D) Filtered
Challenge Events
Raw Count/hem.
(Mean±S.D)
Count/mL
First Challenge Event
No headloss
Midpoint
Stop/Start *
90% headloss
Second Challenge Event
No headloss
Midpoint
Stop/Start
Hemacytometer
Hemacytometer
Membrane filta"
Hemacytometer
Hemacytometer
Hemacytometer
Membrane filter
90% headloss Hemacytometer
Third Challenge Event
No headloss Hemacytometer
Midpoint Hemacytometer
Stop/Start Membrane filter
90% headloss
Hemacytometer
70
66,60
(63±4.0)
100,74
(87±18)
120 mL
20 mL
10 mL
1, 1,2, 1 (1.2±0.5)
11,4,6,8 (7.2±3)
4, 3, 4, 4 (3.8±0.5)
2,7,3,3 (4.0±2.0)
3,3,3,2 (3±0.5)
4,2,5,3 (3.5±1.0)
4,4,4,2 (3.5±1.0)
2,2,5,5 (3.5±2.0)
2,1,1,3 (2.0±1.0)
0.3 x 104
1.8 x 104
0.6
1.0 x 104
1.0 x 104
0.8 x 104
3.2
0.9 x 104
0.8 x 104
0.8 x 104
8.7
0.5 x 104
Entire grab sample was filtered
Table 4-18. Effluent Microscopic Analysis Results of Fluorescent Challenge Events
Enumeration Membrane Raw Count
Method Volume filtered (Mean±S.D)
Sample ID
Count/mL
First Challenge
2.3 x 102/mL
No headloss
Membrane
1.0 mL
238, 220 (229±13)
Midpoint
Membrane
1.0 mL
157,152 (154±3.5)
1.5 x 102/mL
Stop/Start
Membrane
1.0 mL
30, 28 (29±1.4)
2.9 x lOVmL
90% headloss
Membrane
1.0 mL
168, 122 (145±32)
1.5 x 102/mL
Second Challenge.
3.0 x 102/mL
No headloss
Membrane
1.0 mL
300,290 (295±7)
Midpoint
Membrane
1.0 mL
56,76 (66±14)
6.6 x lOVmL
Stop/Start
Membrane
50.0 mL
149,150 (149±1)
3.0/mL
90% headloss
Membrane
1.0 mL
115, 127 (121±8)
1.2 x 102/mL
Third Challenge
2.3 x 102/mL
No headloss
Membrane
1.0 mL
265, 200 (232±46)
Midpoint
Membrane
1.0 mL
158, 110 (134±34)
1.3 x 102/mL
Stop/Start
Membrane
50.0 mL
65, 58 (61±5.0)
6.1 x lOVmL
90% headloss
Membrane
1.0 mL
115, 136 (125±15)
1.3 x 102/mL
57
-------
Table 4-19. Microscopic Analysis Calculation of Percent Removal, Fluorescent Challenge Events
Sample ID Percent Removal DL°8»,
count/mL counts/mL Removal
First Challenge
2.3 x 102
No headloss
0.3 x 104
92.3 %
1.1
Midpoint
1.8 x 104
1.5 x 102
99.2 %
2.1
90% headloss
1.0 x 104
1.5 x 102
98.5 %
1.8
Second Challenge.
3.0 x 102/mL
No headloss
1.0 x 104
97.0 %
1.5
Midpoint
0.8 x 104
6.6 x lOVrnL
91.8 %
2.1
90% headloss
0.9 x 104
1.2 x 102/mL
98.7 %
1.9
Third Challenge
2.3 x 102/mL
No headloss
0.8 x 104
97.1 %
1.5
Midpoint
0.8 x 104
1.3 x 102/mL
98.4 %
1.8
90% headloss
0.5 x 104
1.3 x 102/mL
97.4 %
1.6
NA = Not Applicable as effluent values were greater than influent values.
The microscopic analyses of the samples clearly indicate the seeded microspheres did in fact
enter the water and flow into the filters where the majority was captured. This examination
included analysis of florescent microspheres exclusively and did not count particles indigenous
to the source water, although, the samples were characterized by the examiner as being very
"muddy".
The microscopic examination also revealed a significant number of fluorescing microspheres
smaller than those suggested by the sizes seeded. It was suspected that some of these were
calcite particles formed by the lime softening process, and much smaller than the 3 |j,m threshold.
This may indeed be the case for some of the particles, however, following the microscopic
examination under UV light, it was concluded that only fluorescing microspheres were measured
in Huffman Environmental Consultings' examination and included in the results above although
many were below 3 |j,m in diameter.
Discussions with the particle manufacturer, Bangs Laboratories, and further analysis of the batch
by Bangs with a Particle Sizing Systems, Inc., Model 770 AccuSizer, revealed that the true
particle median size was not 3.4 |j,m as specified, but was actually 2.98 |j,m with a standard
deviation of 0.66 |j,m or 21.2%. The histographic spectrum showed 50% were below 2.98 |j,m
and 90% of the particles were below the stated diameter of 3.45 |j,m.
Table 4-20 is the size analysis as provided by Bangs Laboratories on the Lot in question (#2200)
showing the distribution of the particles and the true standard deviation. The particles tail off
below 2.5 |j,m, peak to the median at 2.98 |j,m, and again tail off above 5.5 |j,m. Eighty percent of
the particles in this batch lie between 2.66 |j,m and 3.45 |j,m. The FTO also obtained standard
deviation data from the two other microsphere suppliers. Those particles had much narrower
size distribution curves. The 5 |j,m particles from Duke Scientific (G0500) had a measured mean
diameter of 5.1 |j,m and a coefficient of variation (CV) of < 5%, and the 6 (j,m particles from
Polysciences, a measured number average of 5.8950 (im with a standard deviation of 0.3750 (im
and a CV of 6.4%. The contribution of these batches outside the area of interest was negligible.
58
-------
For the 3.4 |j,m Bangs lot 2200 however, the contribution below 3 |j,m is significant and is shown
in Table 4-20 and in the accompanying histogram.
Table 4-20. Summary of Number Weight Cumulative Distribution by Histogram
% Of Particles
Micron Size
5% of total particle number <
2.56 microns
10% of total particle number <
2.66 microns
15% of total particle number <
2.73 microns
20% of total particle number <
2.77microns
25% of total particle number <
2.82 microns
30% of total particle number <
2.85 microns
35% of total particle number <
2.89 microns
40% of total particle number <
2.92 microns
45% of total particle number <
2.95 microns
50% of total particle number <
2.98 microns
55% of total particle number <
3.01 microns
60% of total particle number <
3.04 microns
65% of total particle number <
3.08 microns
70% of total particle number <
3.11 microns
75% of total particle number <
3.15 microns
80% of total particle number <
3.20 microns
85% of total particle number <
3.28 microns
90% of total particle number <
3.45 microns
95% of total particle number <
4.64 microns
99% of total particle number <
5.55 microns
59
-------
Figure 4-9 shows the histogram provided by Bangs Laboratories, Inc. It has been scaled to show
the particles found in the 2 to 5.47 |j,m size range. This historgram displays the distribution of
the 3.4 um particles supplied by Bangs Laboratories. The particle counter employed was an
Accusizer 770. The sample size was 60 ml with a dilution factor of 1.33.
Lot 2200 Bangs 3.4 micron
30,000
Particle Size (|im)
Figure 4 -9. Histogram Particle Sizing
Note: COA discussed the discrepancy with Bangs Laboratories and NSF. From these discussions, it was decided to reevaluate
the fluorescent challenges using the backup set of samples by laboratory optical particle analysis.
The 3.4 |j,m fluorescent microspheres were not distributed around 3.4 |j,m as the laboratory
reported. For the 3.4 |j,m Bangs lot 2200, the contribution below 3 |j,m skews the curve
downward and produces a median diameter of 2.98 |j,m. Using the diameter of 2.9 |j,m in place
of 3.4 |j,m, the particles per mL in the concentrated suspension were recalculated as follows:
Diameter
Particles/mL
2.98 |j,m
6.87 x 108
5.0 |j,m
1.37 x 108
6.00 |j,m
2.35 x 108
The size distributions of the two other microsphere lots were much narrower. The 5 |j,m particles
from Duke Scientific (G0500) had a measured mean diameter of 5.1 |j,m and a coefficient of
variation (CV) of < 5%, and the 6|j,m particles from Polysciences, a measured number average of
5.8950 |j,m with a standard deviation of 0.3750 |j,m and a CV of 6.4%. The contribution of these
batches outside the area of interest was negligible, and while the contribution of ihe 6 |j,m
particles was larger than previously calculated it was partially offset by the reduced contribution
of the 5 |j,m particles.
60
-------
If the particle distribution is recalculated, using the 2 mL volume of concentrated suspension, but
at the higher contribution from the Lot 2200 particles, the suspension had a different profile, as
follows:
2 mL 2.98 particles = 1.34 x 109 = 13.4 x 108
3 mL 5.1 particles =4.11xl08
2mL 6.0 particles =4.70xl08
= 22.2 x 108 = 2.22 x 109
Thus, the actual suspension seeded contained approximately 11,746 particles/mL as measured by
pipette, with 4,661 of those as a contribution from the 5 |j,m and 6 |j,m particles.
The three fluorescent microsphere seeding challenges were recalculated in the following manner.
The influent was established by adding 11,746 to the average of the pre-and post-seeding
background particle counts. These calculated influent counts were confirmed with
corresponding Accusizer influent particle counts (as discussed in Section 4.3.4.5). The effluent
particle count values were the particle counts as determined by the on-line particle counts
measured during the seeding plateau. Counts were compared within the range of 2 |j,m to 7 |j,m
rather than 3 |j,m to 7 |j,m to account for microspheres >2.5 |j,m and < 3.0 |j,m. Table 4-21
summarizes the on-line particle count values before adding the 11,746 particles/mL that were
seeded as fluorescent microspheres and the corresponding logio reductions (see Appendix H).
Table 4-21 On-line Particle Counts and Added Microsphere Log10 Reductions (2.0 -7.0 |lm)
On-line Particle Influent Effluent
Counts Pre/Post On-line Particle Counts On-line Particle Reduction
Seeding1 Plus Microspheres2 Counts3 (logo)
(counts/ml) (counts/mL) (counts/mL)
Seeding
First Verification Challenge Runl
No headloss
6,125
17,871
228
1.9
Midpoint
2,279
14,025
76
2.3
90% headloss
5,418
17,164
169
2.0
Second Verification Challenge Run
No headloss
5,974
17,720
204
1.9
Midpoint
5,471
17,217
38
2.7
90% headloss
5,648
17,394
40
2.6
Third Verification Challenge Run.
No headloss
6,188
17,934
163
2.0
Midpoint
6,662
18,408
35
2.7
90% headloss
6,536
18,282
40
2.7
"''influent counts from on-line particle counter represent the average of approximately 20 data points recorded every
60 seconds. Data points included in this average generally represent 10 data points prior to and after the seeding
event.
2Sum of on-line particle counts pre/post seeding and added fluorescent microspheres (11,746 particles/ml).
3Effluent on-line particle counts represent the average of 3 data points recorded on plateau of elevated counts during
microsphere seeding.
The analysis of adding the number of seeded microspheres to the particle count results suggests
the RPI bag and cartridge filter system demonstrated 1.9 to 2.7 logio reductions of microspheres
and indigenous particles sized 2.0 |j,m to 7.0 |j,m during the fluorescent microsphere challenge
events.
61
-------
Also of significance is the comprehensive display of each of these runs, showing numerical
reductions on both sides of the seedings. Figures 4-10 through Figure 4-12 are illustrations for
each of the last three runs, which included fluorescent particle seeding, showing influent, and
effluent at 2-7 |j,m and at 3-7 |j,m. Note: the data points were curve fitted using a polynomial
curve fitting program.
Fluorescent Microsphere Challenge Event #1
Logio Reductions
Performance Throughout the Filter Run with Indigenous Particles
Run Time (minutes)
-- On-Line 2.0 -7.0 Micron On-Line 3.0 - 7.0 Micron
Figure 4-10. Fluorescent Microsphere Challenge Event #1 - Logio Reductions Performance Throughout the
Filter Run with Indigenous Particles
62
-------
Fluorescent Microsphere Challenge Event #2
Logio Reductions
Performance Throughout the Filter Run with Indigenous Particles
Run Time (minutes)
On-Line 2.0 - 7.0 Micron — On-Line 3.0 - 7.0 Micron
Figure 4-11 Fluorescent Microsphere Challenge Event #2 - Logio Reductions Performance Throughout the
Filter Run with Indigenous Particles
Fluorescent Microsphere Challenge Event #3
Logio Reductions
Performance Throughout the Filter Run with Indigenous Particles
Filter Run Time (minutes)
On-Line 2.0 - 7.0 Micron On-Llne 3.0 - 7.0 Micron
Figure 4-12 Fluorescent Microsphere Challenge Event #3 - Logio Reductions Performance Throughout the
Filter Run with Indigenous Particles
63
-------
4.3.4.5 Laboratory Optical Particle Analysis
The instrument employed for this evaluation was a laboratory optical particle counter, an
Accusizer, often used in the evaluation of materials for the pharmaceutical industry. The
Accusizer has a broad range of both count and sizing accuracy as noted by its application to the
determination of particle sizing for particulate matter in the preparation of injectibles such as
vaccines and serums, United States Pharmacopeia (USP) Reference Standard 788. It is capable
of sizing particles from 0.5 |j,m to 400 |j,m in as many as 512 individual bins. QA/QC procedures
for this instrument and employed in this analysis were in accordance with USP Reference
Standard 788 and ASTM 658.
The second set of samples, collected as a back-up for the microscopic analyses described above,
were forwarded to Micro Measurement Laboratories (MML) where they were analyzed by the
Accusizer in accordance with USP Reference Standard 788 and ASTM 658. Because the
analyses originally planned for these samples were for fluorescing microspheres exclusively,
sample contamination from non-fluorescing, natural particles introduced from the environment
or the sample container itself was considered of no consequence. No efforts had been
undertaken to establish a clean room environment on site nor was there a concern for particle
free sample bottles. It should be noted that the Accusizer counted indigenous particles in the
sample as well as fluorescent microspheres.
As a control, two empty sample collection containers were forwarded to MML to measure the
level of possible sample contamination. Analysis of the control sample container demonstrated a
suspected level of contamination (approximately 315 particles/mL). While this represents a very
small fraction of the total particles measured within the influent samples, they do represent a
very large fraction of total particles measured within the effluent samples. Control samples for
the Accusizer evaluations are discussed below in Section 4.5.7.
Accordingly, logio reductions calculated with the use of the effluent data, as analyzed with the
Accusizer are not included within this report. However, influent particle count data as provided
from these analyses were helpful in validating influent particle/microsphere concentrations used
to calculate logio reductions of particles/microspheres sized between 2\xm and 7|j,m (see
Appendix H).
4.3.4.6 Discussion of Results of Fluorescent Microsphere Challenges
The data from on-line particle counters became uncertain at a concentration above
approximately 10-12,000 particles per milliliter, regardless of the particle size. It appears that
when inundated with smaller particles, even if beneath the threshold of the instrument, the
counters reached coincidence error. Since the influent count was uncertain—and well below
the calculated level—the logio reduction observed through the filters was lower than expected.
The microscopic analysis was accurate, however, because the fluorescent microspheres were
significantly smaller than expected, and because the microscopic enumeration counted only
particles that fluoresced, those counts included a high number of particles beneath the level of
interest. The influent counts were determined by hemacytometer because of the high
64
-------
concentration of microspheres, while the effluent counts were determined by filtration through
membrane and microscopic count. These counts were statistically extrapolated to establish
concentration per milliliter.
The Accusizer, as a laboratory instrument subject to rigorous calibration and quality control,
was better able than the on-line particle counters to size the particles in the samples, however
the data represent a single momentary sample in each case. Again, unlike the microscopic
analysis, which was performed visually for only fluorescent particles, the Accusizer counted
all of the particles in the sample, not just those fluorescent microspheres that were added
during the challenge seedings. Hence, the counts from the Accusizer included both natural
particles and the fluorescent microspheres with no ability to distinguish between them.
Moreover, because backup samples originally secured for microscopic analysis of florescent
microspheres were used, sample contamination from non-fluorescing, natural particles
introduced from the environment or the sample container itself, originally offering no
consequence, may have influenced the results of the AccuSizer analyses. While the suspected
level of sample contamination (315 particles/mL) represents a very small fraction of the total
particles measured within the influent samples, they do represent a very large fraction of total
particles measured within the effluent samples. Because logio reductions calculated with the
use of these data would be misleading, they have not been included within this report.
However, Accusizer data representative of influent particle/microsphere concentrations were
used to validate calculated influent particle/microsphere concentrations used to calculate logio
reductions of particles/microspheres sized between 2\x.rs\ and l\x.rs\ (see Appendix H).
Direct comparisons of the data must be exercised with caution to avoid misleading
interpretations. At the same time however, the availability of two sets of data representing the
same test offers an opportunity unique to the testing of bag and cartridge filters (refer to Tables
4-22 and 4-23). Note the on-line particle counter counted particles, both natural and
fluorescent, in the range of 2.0-7.0 |j,m. Further, the microscopic analysis only accounts for the
fluorescing particles, thus the total counts cannot be expected to be comparable. However, the
reductions, when expressed as a logarithm are comparable. Table 2-22 provides a summary of
the microscopic enumeration log reductions and a summary of the on-line particle counter plus
the addition of the seeded microspheres log reductions are presented in Table 4-23.
Table 4-22. Microscopic Microsphere Counts and Logio Reductions (2.5-7.0 |lm)
Influent
Effluent
Reduction
Seeding
Microscopic
Microscopic
(counts/mL)
(counts/mL)
(logio)
First Verification Challenge Runl
No headloss
3,000
230
1.1
Midpoint
18,000
150
2.1
90% headloss
10,000
150
1.8
Second Verification Challenge Run.
No headloss
10,000
300
1.5
Midpoint
8,000
66
2.1
90% headloss
9,000
120
1.9
Third Verification Challenge Run.
No headloss
8,000
230
1.5
Midpoint
8,000
130
1.8
90% headloss
5,000
130
1.6
65
-------
Table 4-23 On-line Particle Counts Plus Microspheres Log
io Reductions (2.0 -7.0 (lm)
Influent
Effluent
Reduction
Seeding
On-line +
On-line
Microspheres
(counts/mL)
(counts/mL)
(logio)
First Verification Challenge Runl
No headloss
17,871
228
1.9
Midpoint
14,025
76
2.3
90% headloss
17,164
169
2.0
Second Verification Challenge Run.
No headloss
17,720
204
1.9
Midp oint
17,217
38
2.7
90% headloss
17,394
40
2.6
Third Verification Challenge Run.
No headloss
17,934
163
2.0
Midpoint
18,408
35
2.7
90% headloss
18,282
40
2.7
* Based on calculations described in section 4.3.4.4
Logio reductions calculated fom data secured during fluorescent microsphere challenges were
expected to be higher than what is demonstrated between challenges with much lower influent
counts of particles indigenous to the source water. During each challenge microsphere counts of
the size of interest were sufficient to allow effluent counts high enough to not be significantly
affected by limitations of counting instrumentation and analyses while supporting the logio
reductions expected. Figures 4-13, 4-14, and 4-15 provide a comparison of logio reductions from
the various analyses and the logio reductions, between the ranges of 2-7 |j,m and 3-7 |j,m,
achieved throughout each filter run which challenges were conducted. In review of these figures
it can be noted that logio results generated from microscopic data from seeding challenges
(Figure 4-23) typically demonstrate lower reductions than reductions calculated from on-line
particle counter data secured between challenges with lower influent counts. Conversely,
analyses of on-line particle count data (Figure 4-24) demonstrated significantly higher logio
reduction values.
In summary, the RPI bag and cartridge system demonstrated 1.1 to 2.1 logio removal of seeded
microspheres (2.5-7.0 |j,m) based on the microscopic enumeration results, and 1.9 to 2.7 logio
removal of microspheres and indigenous particles sized 2.0 to 7.0 |j,m based on normalized on-
line particle counter data.
66
-------
Fluorescent Microsphere Challenge Event #1
Logio Reductions
Comparison between Challenges Analyses
and
Reductions Throughout the Filter Run with Indigenous Particles
Run Time (minutes)
• 2.5-7.0 Micro Analysis I 2.0-7.0 Normalized On-line
^^"Continuous On-Line 2.0 -7.0 Micron Continuous On-Line 3.0 - 7.0 Micron
Figure 4-13. Fluorescent Microsphere Challenge Event #1 - Logio Reductions Comparison between
Challenges Analysis and Reductions Throughout the Filter Run with Indigenous Particles
Fluorescent Microsphere Challenge Event #2
Log10 Reductions
Comparison between Challenges Analyses
and
Reductions Throughout the Filter Run with Indigenous Particles
Run Time (minutes)
• 2.5-7.0 Micro Analysis I 2.0-7.0 Normalized On-line
Continuous On-Line 2.0 - 7.0 Micron Continuous On-Line 3.0 - 7.0 Micron
Figure 4-14. Fluorescent Microsphere Challenge Event #2 - Log10 Reductions Comparison between
Challenges Analysis and Reductions Throughout the Filter Run with Indigenous Particles
67
-------
Fluorescent Microsphere Challenge Event #3
Logm Reductions
Comparison between Challenges Analyses
and
Reductions Throughout the Filter Run with Indigenous Particles
Filter Run Time (minutes)
• 2.5-7.0 Micro Analysis ~ 2.0 - 7.0 Normalized On-line
Continuous On-Line 2.0 - 7.0 Micron Continuous On-Llne 3.0 - 7.0 Micron
Figure 4-15. Fluorescent Microsphere Challenge Event #3 - Logio Reductions Comparison between
Challenges Analysis and Reductions Throughout the Filter Run with Indigenous Particles
68
-------
4.3.4.7 Stop/Start Event Evaluation
Stop/Start data generated during filter runs was evaluated by analyzing the microscopic data
provided by Huffman Environmental Consulting in addition to the discussion on stabilization
time as discussed in Section 4.2.3, Task 3.
Assuming that microspheres did not break through or slough off the filter media during steady
state operation between challenge events, filter removal efficiencies with one stop/start event
were evaluated given results of the microscopic analyses presented in Tables 4-17 and 4-18. The
microsphere concentrations from influent and effluent samples were compared to evaluate filter
efficiency. Results are provided below in Tables 4-24 and 4-25.
Table 4-24.Calculation of Filter Efficiency With 1st and 2nd Seeding Events and One Stop/Start Event
Influent Effluent
Percent
Logic
Sample ID No headloss + Midpoint No headloss + Midpoint seedings Removai (»/) Removal
First Challenge
21,000
309
98.5
1.83
Stop/Start
Second Challenge
18,000
369
98.0
1.69
Stop/Start
Third Challenge
16,087
421
97.4
1.58
Stop/Start
Table 4-25. Calculation of Filter Efficiency With 2nd Seeding Event and one Stop/Start Event
Influent Effluent
No headloss + Midpoint No headloss + Midpoint
Sample ID
Percent
Removal (%)
Logo
Removal
First Challenge
18,000
179
99.0
2.00
Stop/Start
Second Challenge
8,000
69
99.1
2.06
Stop/Start
Third Challenge
8,000
191
97.6
1.62
Stop/Start
The purpose of a cessation and resumption of flow is also designed to indicate the duration of
time in which previously captured particles are released if flow is interrupted then resumed. The
analysis of a single sample is then of less interest than the determination of a peak following a
stop, and the duration of the peak. From that view, the actual number of particles is of less
interest than an approximation of the size distribution, and the length of time until the filter is
back to normal performance. For these evaluations, the in-line particle counter data, along with
sizing data from the Accusizer, and both confirmed by the microscopic data is of some interest.
The duration of the peak following interruption is short, two to three minutes; and the loss in
performance is, as would be expected, greater with the smaller particles. However, reductions
shown as logio removal are relatively small in the range of interest, at approximately 0.5 logio.
69
-------
The following graphs (Figures 4-16 through 4-18) illustrate the particle breakthrough against
time in the range of 3 |j,m to 7 |j,m and the duration of breakthrough. The duration of the spike is
approximately 3 to 4 minutes. The length depends some on the point in the sensor cycle that the
particle count resumes, but it is clearly a brief jump whereupon the filter stabilizes and returns to
the prior levels.
£
o ^
CO 0)
" &
£ t/>
c 0)
= o
O =
O i
_0) Q.
O ' '
'¦E
ra
Q.
500
400
300
200
100
start
stop
7 f
0
I i I—I—h 1 1 1 1 1 r~
1 i , i
i 1 i
i i i i
nT r, V , Y* _ Y* K V- ,r , Y" _ Y* „ Y" , Y"
& /O' .•£ .»•?> rfr rfr rfi rft & &
K. K. K. K. K • K • K." K • K. K.
O O O O O O O O O O
Time
Figure 4-16. Rosedale Filter Run #20 Effluent 3-7 (lm Particle Count Stop/Start
c
o
s-
o
E
_i
E
h«-
co
o
s—¦"
Q.
c
3
0)
o
o
o
s-
©
o
Q-
s-
a.
600
500
400
300
200
100
0
start
stop
. ...I. i
f
1 ~ 1 I ~ I ~ I I I
"I 1 1 1-
Time
Figure 4-17. Rosedale Filter Run #21 Effluent 3-7 (lm Particle Count Stop/Start
70
-------
400
300
- sr200
0)
o
'€
® Q- 100
start
I I I I ~ I I
stop
\ I ~
I I
^ ^ r^.N r)/ r)/ q^M r)/ (p> r^?
Time
fly
Figure 4-18. Rosedale Filter Run #22 Effluent 3-7 (lm Particle Count Stop/Start
The following graphs (Figures 4-19 through Figure 4-21) illustrate the turbidity peak following a
stop/start sequence. As in the case of the particle counts, the duration is brief, however, since
turbidity is composed of a large number of particles beneath the level of interest (that is, below
3(j,m) the duration is slightly longer. Moreover, the response time of the turbidimeter is
somewhat longer than that of the particle counter.
Figure 4-19. Rosedale Filter Run #20 Effluent 3-7 (lm Turbidity Stop/Start
71
-------
Figure 4-20. Rosedale Filter Run #21 Effluent 3-7 (lm Turbidity Stop/Start
Figure 4-21. Rosedale Filter Run #22 Effluent 3-7 (lm Turbidity Stop/Start
72
-------
4.4 Equipment Characteristics Results
The qualitative, quantitative and cost factors of the tested equipment were identified during the
verification period, in so far as possible. The results of these three factors are limited due to the
relatively short duration of the testing period.
4.4.1 Qualitative Factors
Qualitative factors that were examined during the verification testing were the ease of which
filter elements were exchanged, measurement of head loss, and other operational factors that
might impact on performance of the equipment.
4.4.1.1 Filter Element Replacement
Filter elements on the RPI GFS Filter System were replaced when the headloss across the system
approached 15 psi. Filter elements were replaced in pairs, except for one period (filter runs 17
and 18) when only the first element, the bag, was replaced to conserve cartridge elements.
During the filter replacement seals in the top of the basket were also routinely replaced. In
normal operation replacement of the seals is not necessary. Only on one occasion was the seal
found to be defective. Additionally, an 'O' ring on top of the housing was replaced on one
occasion when it was determined to have been pinched during replacement of the elements.
Replacement of both elements took about 15 minutes, including the time required to drain down
the housing and purge the air following replacement. If rushed it could take still less time,
although the operators of the system usually took longer in order to assure that the housings and
elements were properly placed. Instructions for replacement included in the operating
instructions supplied by the manufacturer were helpful.
Installation of the bag requires that it be fully pushed into the basket, and smoothed, and that the
top rim be within the basket and held in place by a spring. The rigid element is also placed in the
basket, and should be pushed firmly and fully snug so the seal at the top is in place. Wetting the
seal and rim of the basket with filtered water allowed for an easier fit.
Removal of the elements should be accomplished by removing the baskets from the housings and
then the elements from the basket. Because of the force of the seal plate and spring at the top,
these elements resisted removal in most cases, and the operators found it was better done with
two people pulling in opposition.
The bag and cartridge filter element fabric contained a chemical that produced milky foam when
wetted. It is important to bleed off this residue before placing the filters on line, to prevent
introduction into the filtrate stream and cause erroneous readings in the turbidimeters and
particle counters. Although care was exercised during replacement, occasional spikes in
turbidity and counts especially immediately following replacement may be the product of this
material.
73
-------
4.4.1.2 Head Loss
Headloss across the filters was determined by pressure gauges showing influent and effluent
pressures. Manufacturer supplied Orange Research differential pressure gauges mounted on the
housings were ambiguous and were removed previous to the beginning of the verification period.
After examination of the pressure differential gauge located on the bag filter housing in filter
train #2 it was discovered the diaphragm separating the influent from the effluent chambers was
ruptured and decayed, thus allowing bypass between the chambers. This may have contributed
to rigid cartridge filters loading during the initial operations period. Examination of other
Orange Research gauges indicated others too had failed during initial operations. Of the six
gauges supplied, only two were functional. The broken differential gauge was replaced with two
pressure gauges, but these readings were disregarded except as an indication of overall housing
pressure during bag replacement. Throughout the verification study, pressure gauges mounted
on the instrument panel that measured inlet and outlet pressures across both housings were used
as the pressure measurement of record. These gauges were verified against a NIST-traceable
pressure gauge.
As expected, headlosses accelerated toward the end of a filter run. With the variation in river
water turbidity, even with stable flows predicting filter run lengths was difficult and uncertain
The filter headloss was established by the manufacturer at 15 psi and was not challenged to
breakthrough by the FTO. Casual observation noted that performance improved toward the end
of a filter run, as headlosses increased, likely due to particle bridging.
4.4.1.3 Other Operational Factors
Evaluation of equipment safety was conducted as part of the verification testing. Evaluation of
the safety of the treatment system was done by examination of the components of the system and
identification of hazards associated with these components. A judgment as to the safety of the
treatment system was made from these evaluations.
Prior to opening the filter housings, pressure must be relieved. Pressure was simply bled off
through the air release valve at the top of the housing and offered no difficulty. Replacement of
the filters without some water spillage however proved to be nearly impossible, thus it was
important to mop up the area following replacement to prevent slips and falls.
No injuries or accidents occurred during verification testing.
A handle on one of the butterfly valves, provided by Bray Manufacturing, broke during normal
operation and was replaced. Examination of the handle showed that it had been cast poorly, with
a flaw that had not been noticed. There was evidence of a crack almost through the entire
casting but covered with paint. There was no interruption of either testing or the flow through
the system because of this failure.
74
-------
4.4.1.4 Evaluation of O&M Manual
The manufacturer supplies an O&M Manual that illustrates the equipment and shows the proper
configuration of the housings. The filter start up and filter media element replacement
procedures are instructive and thorough. A spare parts list used for the RPI GFS Filter System
was included. The manufacturer also describes warrantees pertaining to the RPI GFS Filter
System.
The O&M manual was reviewed for completeness and used during equipment installation, start-
up, and system operation. The manual is brief and concise, however it appears to be a general
O&M booklet intended to be applicable to a number of similar systems. Since the system
supplied is a simple filter system with only pressure gauges as instruments, the discussion of
operational or performance procedures is necessarily limited in scope. However, since this
system is being marketed in a public health arena, some attention to operational considerations
relating to health performance could be included. While it was found the manual provides
adequate instruction for the ETV tasks, additional operational guides might be helpful.
The use of a differential pressure gauge might be insuitable in some applications and separate
gauges for the influent and effluent pressure ports of each housing could be added, along with
instructions on their use. Sample taps can also be added at those points. Dimensions, especially
those relating to the sizes and threads of fittings and to housing mounting distances could aid in
installation. In addition, some discussion on what to look for in assessing seal and gasket
integrity might be useful to inexperienced operators.
4.4.2 Quantitative Factors
4.4.2.1 Filter Elements Replacement
A total of 20 bags and 20 cartridge filter elements were used during the study. In field practice,
bags and filter elements would not necessarily be replaced at the same time. The less costly bag
is likely to be replaced more frequently than the more expensive rigid cartridge. In many cases
where seasonal algae or sediment loads are heavier, small system users may benefit from pre-
filtration, to limit the bag and rigid element removal to smaller suspended matter.
4.4.2.2 Anomalous Conditions That Require Operator Response
Operator response is required primarily to observe and monitor pressure losses as an indication
of the filter system performance.
4.4.2.4 Length of Operating Cycle
As would be expected, the length of operation of a filter element is directly proportional to the
loading rate as measured by the turbidimeter and particle counter. The loading rate could be
controlled by changing the blend of raw and finished water. To what degree changing the blend
would directly effect the loading rate was not measured. Operators allowed a slight variation in
turbidity to occur naturally, responding to changes only when they exceeded predetermined
75
-------
limits. It is likely that small systems operators will employ additional pretreatment, so the RPI
GFS system will be used as a final barrier.
In addition, the test plan required that the flow be adjusted whenever it strayed from the
determined rate. In many small system applications, the filters are allowed to follow a declining
rate, and in those cases improved performance or extended filter life may be possible. Typically
filter rates slowed as the headloss increased, and at the same time removal was slightly
improved. It should also be noted that within this performance evaluation, the filter train is
treated as a whole. As such both the pre-filter and final filter were replaced when terminal head
loss across the filter train had been met. It is expected in field applications filters will be
replaced based upon individual, as compared to combined, head loss. Thus, increasing the
probability of extending the life of the final filter.
4.5 QA/QC Results
The objective of this task is to assure the high quality and integrity of all measurements of
operational and water quality parameters during the ETV project. QA/QC verifications were
recorded in the laboratory logbooks or spread sheets. QA/QC documentation and calibration
certifications are attached to this report as Appendix G.
4.5.1 Data Correctness
Data correctness refers to data quality, for which there are four indicators:
• Representativeness
• Statistical Uncertainty
• Accuracy
• Precision
Calculation of all of the above data quality indicators were outlined in the Chapter 3, Methods &
Procedures. All water quality samples were collected according to the sampling procedures
specified by the EPA/NSF ETV protocols, which ensured the representativeness of the samples.
4.5.1.1 Representativeness
Operational parameters graphs and discussions are included under Task 3 - Documentation of
Operations Conditions and Treatment Equipment Performance. Individual operational
parameters, such as flow rate, particle count data, turbidity data, and testing equipment
verification are presented below in discussions on Daily, Bi-Weekly and Start of Testing Period
QA/QC Results.
4.5.1.2 Statistical Uncertainty
Ninety-five percent confidence intervals were calculated for the water quality parameters of the
RPI GFS Filter System. These include influent and effluent turbidity, particle count, flow rates,
and various other filter runs performance data as discussed in Task 3 - Documentation of
76
-------
Operations Conditions and Treatment Equipment Performance. Ninety-five percent confidence
intervals were also presented in the water samples summary tables in the discussion of Task 2 -
Influent and Effluent Water Quality Characterization.
4.5.1.3 Accuracy
For this ETV study, the accuracy refers to the difference between the sample result, and the true
or reference value. Calculations of data accuracy were made to ensure the accuracy of the
testing equipment in this study. Accuracy of parameters as flow rate, particle count data,
turbidity data, and testing equipment verification are presented below in discussions on Daily,
Bi-Weekly and Start of Testing Period QA/QC Results.
4.5.1.4 Precision
Precision refers to the degree of mutual agreement among individual measurements and provides
an estimate of random error. Precision was ensured by calculating the percent relative standard
deviation or the relative percent difference, and having it be equal to or less than 30%. For
single reading parameters, such as pressure and flow rates, precision was ensured by redundant
readings from operator to operator. Samples were analyzed in triplicate for those on-site
parameters consequential to the testing: bench-top turbidity, pH and bench-top particle counts
associated with the calibration of the equipment. These calibration procedures and results are
presented in discussions on Daily, Bi-Weekly and Start of Testing Period QA/QC Results.
4.5.2 Daily QA/QC Results
Daily readings for water quality were listed in the logbook and then transcribed to computer
format. Logbooks contained carbon paper second sheets that were separated and maintained off
site at the COA offices. Computer diskettes were used to download data and then transferred
physically to the COA offices.
The influent on-line turbidimeter flow rate averaged 458 mL/minute. The effluent on-line
turbidimeter flow rate averaged 446 mL/minute. These averages were calculated only to show
that the limits were observed. To determine the flow rate of the on-line turbidimeters the flow
was measured with stopwatch or sweep-watch and a 1,000 mL graduated cylinder. The
maximum rate during the testing period for the influent turbidimeter was 660 mL/minute, the
minimum was 310 mL/minute for the effluent turbidimeter the maximum rate during the testing
period was 740 mL/minute, the minimum was 272 mL/minute The acceptable range of flows as
specified by the manufacturer is 250 mL/minute to 750 mL/minute. The turbidimeter readings
are accurate within those ranges; however, the time from beginning of flow to stable turbidity
indication is lengthened with the slower flows. The manufacturer notes that the first Sep
response time is 2.5 seconds and 90% stability is reached in 5 minutes when the flow is 750
mL/min.
The influent readout from the Hach 1720C on-line turbidity averaged 1.09 NTU during the
period; the average from the Hach 2100P benchtop turbidimeter was 1.08 NTU. The effluent
readout from the Hach on-line turbidity averaged 0.28 NTU during the period; the average from
77
-------
the Hach 21 OOP benchtop turbidity was 0.24. This narrow difference is accidental as the on-line
and bench turbidimeters are not expected to read the same, only to track in a relative manner.
The on-line and bench-top readings were compared daily. Ten (of 22 readings) for effluent
turbidity were outside the 30% RPD due to the proximity of turbidity to the instrument's
measuring limit. One (of 22 readings) for influent turbidity was outside the 30% RPD. The
bench-top readings were within acceptable limits of 30% of RPD. Additional documentation and
tables can be found in Appendix I.
The influent feed water particle counter flow rate averaged 102 mL/minute. To determine the
flow rate of the on-line feed water turbidimeter the flow rate was measured using a graduated
cylinder and stopwatch. The maximum flow rate measured was 110 mL/minute, the minimum
was 97 mL/minute. The target flow rate specified by the manufacturer is 100 mL/minute.
Efforts were made to keep the flow rate between 95 mL/minute to 105 mL/minute and the flow
was adjusted whenever those boundaries were crossed. The finished water particle counter flow
rate averaged 98 mL/minute. The flow was measured using a graduated cylinder and stopwatch.
The temperature was recorded daily in the evening with a NIST-traceable Miller Weber
Thermometer, Model T-775/63CGC.
The pH meter was calibrated daily against NIST-traceable pH buffers at 7.00 and 10.00 daily.
The pH meter was a Cole Palmer Oakton® WD-35615 Series. The pH calibration buffers were
Oakton pH Singles 7.00 (model #35653-02), and pH Singles 10.00 (model #35653-03). The pH
calibration was performed prior to the recorded influent pH measurement.
The tubing and all water lines used on the treatment system were inspected before testing began
and daily after March 15, 2000. The tubing and lines remained in good condition and
replacements were not necessary.
4.5.3 Bi-Weekly QA/QC Verification Results
Every two weeks checks were made on the on-line flow meters, the meters were cleaned out if
necessary, and the flow readouts were verified. The flow meters were supplied with clean,
filtered water and did not foul. The 30-day test period only required one scheduled verification
of the on-line flow meters. The on-line flow meters were verified (bucket and stopwatch), using
a measured container on March 18, 2000. The flow was measured at 10 gpm x four times. The
deviation was found to be 1 second in 4 minutes at 10 gpm over 40 gallons.
4.5.4 Results Of QA/QC Verifications At The Start Of Each Testing Period
The particle counters were calibrated by Pacific Scientific Instruments using polystyrene latex
spheres traceable to NIST standards. Particle counters used on site were MetOne PCX models.
The MetOne particle counters had a factory calibration certificate dated March 3, 2000, serial
numbers 971000353 and 971000354. Calibration was again verified on site with NIST mono-
sized polymer microspheres as described in Section 3.8.2.4 above. Particle counter verification
was performed for size distribution only, although counts were corroborated.
78
-------
The following figures show the distribution as counted by the MetOne particle counter during the
NIST-traceable verification.
Figure 4-22 shows the particle counts during the influent 3 |j,m verification. The Figure
the addition of the added particles as would be expected.
shows
O
r
(U
0_
lnfluent:>15
lnfluent:10-15
lnfluent:2-3
Influent: 3-5
Influent: 5-7
¦°—lnfluent:7-10
,\N .v3 /v5 .»•?> r£? r£- rp
15
—a— Influent: 10-15
^ lnfluent:2-3
X lnfluent:3-5
A lnfluent:5-7
—o lnfluent:7-10
14:51 14:52 14:53 14:54 14:55 14:56 14:57 14:58
Time (minutes)
Figure 4-23. Verification of Mix of 3,10 & 15 (lm Influent Particles
79
-------
Figure 4-24 shows the particle counts during the effluent 3 |j,m verification. The Figure shows
the addition of the added particles in the 3 |j,m size range as expected.
¦ Effluent:>15
-Effluent: 10-15
¦ Effluent:2-3
" Effluent: 3-5
-Effluent: 5-7
"Effluent:7-10
14:27 14:28 14:29 14:30 14:31 14:32 14:33 14:34 14:35 14:36
Time (minutes)
Figure 4-24. Verification of 3 |lm Effluent Particles
Figure 4-25 illustrates the particle counts during the 10 |j,m effluent verification. This Figure
shows the addition of the added particles in the 10 |j,m size range as expected.
1,800
1,600
1,400
1,200
1,000
800
600
400
200
" Effluent
" Effluent
- Effluent
" Effluent
- Effluent
>15
10-15
2-3
3-5
5-7
14:11 14:13 14:14 14:15 14:16 14:17
Time (minutes)
14:18 14:19
Figure 4-25. Verification of 10 Jim Effluent Particles
80
-------
Figure 4-26 illustrates the particle counts during the 15 |j,m effluent verification. The Figure
shows the addition of the added particles in the 15 |j,m size range as expected.
0
O
r
re
Q_
13:37 13:38 13:39 13:40 13:41 13:42 13:43 13:44 13:45
Time (minutes)
Figure 4-26. Verification of 15 \lm Effluent Particles
Figure 4-27 illustrates the particle counts during the mix of 3, 10, and 15 |j,m effluent
verification. The Figure shows the addition of the added particles in the |j,m size range as
expected.
"Effluent
"Effluent
-Effluent
"Effluent
-Effluent
>15
10-15
2-3
3-5
5-7
14:43 14:44
Time (minutes)
14:46
Figure 4-27. Verification of 3,10 & 15 |lm Effluent Particles
81
-------
Particles that were added were:
Duke Scientific Corp 3.0 ± 0.027|j,m
10.0 ± 0.061|j,m
15.0 ± 0.08|j,m
Visual inspections of the particle counter and turbidimeter tubing at the beginning of testing and
daily thereafter showed unimpeded flow and integrity. The tubing was in good condition and
replacements were not necessary.
There was no differential pressure transmitter attached to this equipment. Gauges were verified
on March 15, 2000 and again on April 2, 2000 by comparing the pressure shown on the gauge
with the same pressure shown on a NIST-traceable pressure gauge. The NIST-traceable pressure
gauge verified the board pressure gauges at 30 psig.
The effluent turbidimeter failed on March 18, and was replaced with a back up on March 19,
2000. The replacement meter was calibrated with Formazin suspension to 20 NTU in
accordance with manufacturer's instructions following restart.
Before the challenge testing of the RPI GFS Filter System began, the Minnesota Department of
Health performed calibration procedures on the bench top, Hach 21 OOP turbidimeter. The
instrument was calibrated to the manufacturer's recommended standards of 20, 100 and 800 NTU
with fresh Formazin suspensions. The manufacturer explains that since the response signal is
linear from 0-20 NTU efforts to standardize to lower levels are fruitless and may instead throw
the readings off. Calibration standards are further required to be at least 65 NTU apart. In
addition, weighting the curve to the range of interest (in this case at levels less than 5 NTU) also
provide the opportunity for increasing error. The manufacturer's recommended settings were
also observed in subsequent calibrations.
The turbidimeter was calibrated against freshly prepared Formazin dilutions from a standard
suspension (4000 NTU). The standards were prepared using NIST-traceable glassware,
including pipettes and volumetric flasks.
Gel ex secondary standards were also calibrated following manufacturer's instructions during the
instrument calibration, and additional secondary standards were prepared or purchased from
Hach. These standards were referenced daily in the ranges of concern. While the standards at
0.5, 1 and 3 NTU were relatively stable, the reference of 0.1 NTU was somewhat ambiguous as
it is at or near the limit of detection for this instrument.
Turbidity samples were collected from a sample tap at a slow steady stream and along the side of
a triple rinsed dedicated beaker to avoid air entrapment. The sample was poured from the beaker
into a double rinsed clean sample vial.
All glassware for turbidity measurements was kept clean and handled with lint free laboratory
tissue. The sample cells were further wiped with velvet, silicon oilcloth.
82
-------
4.5.5 Analytical Laboratory QA/QC
Samples for analyses conducted on feed and finished water are listed in Table 4-2 and Table 4-3.
QA/QC procedures are based on SM, 199h Ed., (APHA, 1995) and Methods for Chemical
Analysis of Water and Wastes, (EPA, 1995).
Calibration results of the analytical instrumentation used to conduct the analyses listed in Table
4-3 on finished water is recorded and kept on file at Spectrum Labs, Inc. QA/QC procedures and
documentation pertinent to this verification test are on file at Spectrum Laboratories and
Cartwright, Olsen & Associates, LLC.
It was noted that the Spectrum QC data documentation lacked the reviewer's initials and the date
of review. The written response from Spectrum regarding this issue indicated that they believed
that the review occurred, however, the documents lack the notation of the review. A review of
the QC data and results of analytical instrumentation indicate that adequate controls were in
place to render the data obtained acceptable.
4.5.6 Microbiological Laboratory QA/QC
Influent analyses used hemacytometer procedures, and effluent counts, which were much lower,
used membrane filtration and microscopic counting. Hemacytometer and membrane filtration
counting was performed as outlined in EPA Method 1622, Section 11.3, EPA Method 1623,
Section 11.0, and SM 10200F but altered for the counting of synthetic, fluorescing microspheres.
A review of all QA/QC procedures and results of analytical instrumentation indicate that
adequate controls were in place to render the data obtained acceptable.
4.5.7 QA/QC Procedures for Accusizer Measurements
QA/QC procedures for the Accusizer measuring system were based on protocols established by
Standard Methods, ASTM 658 and the Pharmaceutical Industry USP 788.
Prior to testing the Accusizer was calibrated to known particle sizes with NIST-traceable
particles at 2 |j,m and 5 |j,m, and additionally at 2.92 |j,m and 3.7 |j,m. A review of size accuracy
suggests that the accuracy of optical particle devices is ± 5% at near and below 2 |j,m and at near
and above 5 |j,m, but closer to ± 10% at the interval of 2.6 |j,m to 4.0 |j,m. This anomaly is due to
electronic characteristics of optical particle counting instruments and should be taken into
account in interpreting any data. Unfortunately, this distortion is at the range of interest in this
study.
The instrument sensor was also standardized to size against voltage, and the curve for that
procedure is attached in the Appendix G.
The samples were kept cool from the time of sampling and shipped chilled. When received they
were refrigerated until removed for testing. Each sample was labeled to indicate challenge run
83
-------
and seed and whether influent or effluent. The samples were contained in Dow Chemical snap
cap containers, and were approximately 100-120 mL each in volume.
When tested, each sample was sonicated for 30 seconds, and then inverted 25 times. This was
done to limit any air bubbles, break up agglomerated particles and to mix homogeneously so the
particles would be distributed throughout the sample. The required volume (15 mL) was then
drawn from the sample and injected into the counter.
Two empty, unused containers were sent to MML as a control to determine the number of
particles introduced into the test samples from sample containers used. Since the intent of these
samples were originally to back-up samples for fluorescing microscopic analysis, procedures that
might have limited to the contribution of non-fluorescing particulates, including sample
container selection itself, were not taken into consideration at the time of sample collection.
PFW was added to the empty sample containers and prepared as other test samples, with
sonification and inversion to eliminate air and break up agglomerated particles. The distribution
of particles from these control containers suggests contribution to the test samples of 315
particles/mL between 3 |j,m to 7 |j,m. Because negative effluent counts would result if measured
data alone were used in Table 4-21, counts were adjusted to include sample container
contribution of 315 particles/mL.
A review of all QA/QC procedures and results of analytical instrumentation indicate that
adequate controls were in place to render the data obtained acceptable.
4.6 Limitations
Measurements used to characterize filter performance included on-line turbidimeters and particle
counters, both of which had severe limitations. The turbidity limitation has been addressed by
others; all that can safely be stated is that there is a vague relationship between turbidity
reduction and filtration. Its use as a defining factor however, is suspect. Turbidity can be the
product of a few large particles, or many smaller ones, and their nature is not revealed.
The use of on-line particle counting also has limitations, especially when it is used to calculate
logio removals. The lower level of particle counts, for example in either fully finished water or
in water that has been filtered, has confidence. The upper levels, especially when particle counts
exceed 10,000 per milliliter in the sizes seen by the sensor, or when there are high numbers of
smaller particles, are uncertain. As a means of establishing policy, the strict application of
particle count data may be severely limited; as a means of evaluating filter performance
however, particle counting can be a rapid and insightful method of determining effectiveness.
Also limiting in this study were the hemacytometer counts of the influent particles. Other
researchers have noted: "recovery values calculated using hemacytometer counts were
consistently lower and significantly different from the recovery values calculated using
membrane counts" (Klonicki). The effect of the hemacytometer count reductions may have
influenced the reduced logio removal values suggested by microscopic analysis.
84
-------
Bench particle counting, using laboratory instrumentation such as a Coulter counter or the
Accusizer employed in this study, can be performed with greater accuracy, especially when
using dilutions and automatic pipettes, however, these tests do not lend themselves to field
applications where sample contamination is probable and rapid feedback is required (Van
Gelder).
The same is true with the use of fluorescent microspheres where, again, laboratory analyses
allow for a single sample, and determination of particle counts require statistical methodology
and a high degree of measuring precision, not easily performed in the field
(Li, et. al.).
Recommendations that can be drawn without contention include the need for proper bag and
element installation and the requirement that there be a minimum of two bed volumes of filtered
water discharged to waste cycle following an interruption in flow. In addition, other operational
details, such as flow rate limitations and allowable pressure differentials may also be inferred
from these data that may aid regulators and small system designers in the application of this
technology to small community water supplies.
85
-------
Chapter 5
References
The following references were used in the preparation of this report:
American Public Health Association, American Water Works Association, Water Environmental
Federation, Standard Methods for the Examination of Water and Wastewater, 19th Edition,
APHA, AWWA, WEF, Washington D C., 1995.
Atherholt, T. B., M. W. LeChevallier, W. D. Norton and J. S. Rosen. Effect of rainfall on
Giardia and Crypto. Journal AWWA 90 (9): 66-80 (1998).
American Water Works Association. "Particle Counting and Sizing in Operational Control of
Coagulation and Filtration Processes." Manual of Water Supply Practices M37, Denver CO 53-
71 (1992).
Bukhari, Z., R.C. McCuin, C.R. Flicker, and J.L. Clancy, Immunomagnetic Separation of
Cryptosporidium parvum from Source Water Samples of Various Turbidities. Applied and
Environmental Microbiology. 64 (11): 4495-4499 (1998).
Davis, Lawrence J. and Rosemary Soave, Cryptosporidium, Isopora, Cyclospora, Microsporidia,
and Dientamoeba, in Infectious Diseases, 2nd Ed. Gorbach, Bartlett and Blacklow, eds.
Philadelphia: W.B. Saunders Company, 1998.
Frey, M.N. et.al., "A Synthesis Report of Cryptosporidium." Paper presented at the American
Water Works Association Annual Conference, Atlanta, Georgia, June 18, 1997.
Hancock, C.M, Ward, J.V., Hancock, K.W., Klonicki P.T. and Sturbaum C.D. Assessing Plant
Performance Using MP A. Journal AWWA 12(88): 24-34(1996).
Harter, Thomas, Sonja Wagner and Edward R. Atwill. Colloid Transport and Filtration of
Cryptosporidium parvum in Sandy Soils and Aquifer Sediments. Environmental Science and
Technology 34(1): 62-70 (2000).
Kiminski, J.C. Correspondence in New England Journal of Medicine 331 (22): 1529-1530
(1994).
Klonicki, Patricia T., Carrie M. Hancock, Timothy M. Straub, Stephanie I. Harris, Keith W.
Hancock, Ali N. Alyaseri, Charles J. Meyer and Gregory D. Sturnbaum, Cypto Research: are
fundamental data missing? Journal AWWA 89 (9): 97-103 (1997).
LeChevallier, M.W. et al. Giardia and Cryptosporidium spp. in Filtered Drinking Water
Supplies. Applied and Environmental Microbiology 57 (9): 2617-2621 (1991).
Li, Sylvana Y. et.al. Reliability of surrogates for determining Cryptosporidium removal, Journal
AWWA 89 (5): 90-99 (1997).
86
-------
MacKenzie, W.R. et al. A Massive Outbreak in Milwaukee of Cryptosporidium Infection
Transmitted through the Public Water Supply. New England Journal of Medicine 331 (3): 161 -
167 (1994).
Maschio, Celio, and A. C.F. Arruda. Application of X-Ray Computerized Tomography to
Characterize Particle Retention Within Depth Filters. Particle and Particle Systems
Characterization 17 (1): 28-32 (2000).
Medema, G.J., F.M. Schets, P.F.M. Teunis, and A.H. Havelaar. Sedimentation of Free and
Attached Cryptosporidium Oocysts and Giardia Cysts in Water. Applied and Environmental
Microbiology 64 (11): 4460-4466 (1998).
National Research Council. Safe Water from Every Tap: Improving Water Services To Small
Communities, by Committee on Small Water Supply Systems, the Water Science and
Technology Board and the Commission on Geosciences, Environment and Resources,
Washington, D.C.: National Academy Press, 1977, 112-114.
Nieminski, Eva C. Removal of Cryptosporidium and Giardia through Conventional Water
Treatment and Direct Filtration. EPA/600/SR-97/025, 1997.
U.S. Environmental Protection Agency, Enhanced Surface Water Treatment Rule (ESWTR) - 40
CFR Parts 9, 141 and 142, EPA, February 19, 1999.
U.S. Environmental Protection Agency, Office of Drinking Water, Guidance Manual for
Compliance with the Filtration and Disinfection Requirements for Public Water System Using
Surface Water Sources, EPA No. 68-1-6989, U.S. EPA, Washington, D.C., March 1991.
U.S. Environmental Protection Agency, Office of Drinking Water, Guidance Manual for
Compliance with the Interim Enhanced Surface Water Treatment Rule: Turbidity Provisions,
EPA No. 815-R-99-010, U.S. EPA, Washington, D.C., April 1999.
U.S. Environmental Protection Agency, Methods for Chemical Analysis of Water and Wastes.
EPA 600/479-020, 1995.
U.S. Environmental Protection Agency, Microscopic Particulate Analysis (MPA) for Filtration
Plant Optimization. EPA 910-R-96-001, 1996.
U.S. Environmental Protection Agency, National Primary Drinking Water Regulations: Interim
Enhanced Surface Water Treatment Rule Notice of Data Availability, 62 Fed. Reg. 59486-59577
(1997) (to be codified at 40 C.F.R. pt 141, 142)(proposed).
U.S. Environmental Protection Agency/NSF International. ETV Protocol: Protocol for
Equipment Verification Testing for Physical Removal of Microbiological and Particulate
Contaminants, EPA/NSF, April 1998, Revised May 1999.
87
-------
U.S. Environmental Protection Agency/NSF International. EPA Test Plan: NSF Equipment
Verification Testing Plan: Bag Filters and Cartridge Filters for the Removal of Microbiological
and Particulate Contaminates, Final Draft of Test Plan, EPA/NSF, April 1998, Revised May
1999.
"U.S. Geological Survey National Water-Quality Assessment Program" Upper Mississippi River
Basin Study Unit, July 12, 1997, http:/www.cr.usgs.gov/umis/descrip.html, (November 10,
1999).
Van Gelder, A. M, Zaid K. Chowdhury and D. Lawler. Conscientious particle counting. Journal
AWWA 91 (12) 64-76 (1999).
Watanabe, M.E. New Cryptosporidium Testing Methods. Environmental Science and
Technology 30 (12): 532A-535A (1996).
88
------- |