United States
Environmental Protection
Agency
Office of Research and
Development
Washington, D.C. 20460
EPA/600/R-98/160
May 1999
\>EPA
Environmental Technology
Verification Report
NSF International • Ann Arbor, Michigan
Inactivation of Cryptosporidium
parvum oocysts in Drinking Water
Calgon Carbon Corporation's
Sentinel™ Ultraviolet Reactor
-------
EPA/600/R-98/160
May 1999
Inactivation of Cryptosporidium parvum
oocysts in Drinking Water
Calgon Carbon Corporation's
Sentinel™ Ultraviolet Reactor
Edited by
C. Bruce Bartley
Carol A. Becker
NSF International
Ann Arbor, Michigan 48105
Prepared by
Cartwright, Olsen and Associates, LLC
Cedar, Minnesota 55011
Project Officer
Jeffrey Q. Adams
National Risk Management Research Laboratory
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC
-------
Notice
The information in this document has been funded wholly or in part by the U.S. Environmental
Protection Agency (USEPA) under Cooperative Agreement No. C R 824815 with NSF
International (NSF). This verification effort was supported by Package Drinking Water
Treatment Systems Pilot operating under the Environmental Technology Verification (ETV)
Program. It has been subjected to the Agency's peer and administrative review, and it has been
approved for publication as an EPA document. Mention of corporate names, trade names, or
commercial products does not constitute endorsement or recommendation for use of specific
products. This report is not an NSF Certification of the specific product mentioned herein.
This document is copyrighted in its entirety by NSF International.
11
-------
Foreword
The following is the final report on an Environmental Technology Verification (ETV) test
performed for the NSF/USEPA by Cartwright, Olsen and Associates, LLC, in cooperation with
Calgon Carbon Corporation. The test was conducted during March and April of 1998 at the
Mannheim Water Treatment Plant in Kitchener, Ontario, Canada.
Throughout its history, the USEPA has evaluated technologies to determine their effectiveness in
preventing, controlling, and cleaning up pollution. EPA is now expanding these efforts by
instituting a new program, the Environmental Technology Verification Program—or ETV—to
verify the performance of a larger universe of innovative technical solutions to problems that
threaten human health or the environment. ETV was created to substantially accelerate the
entrance of new environmental technologies into the domestic and international marketplace. It
supplies technology buyers and developers, consulting engineers, states, and the U.S. EPA
regions with high quality data on the performance of new technologies. This encourages more
rapid availability of approaches to better protect the environment.
The USEPA has partnered with NSF, an independent, not-for-profit organization dedicated to
public health, safety and protection of the environment, to verify performance of small package
drinking water systems that serve small communities. 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.
This verification testing program is being conducted by NSF International 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
(WSWRD) - Cincinnati, Ohio. It is important to note that verification of the equipment does not
mean that the equipment is "certified" by the NSF or EPA. Rather, it recognizes that the
performance of the equipment has been determined and verified by these organizations.
in
-------
Preface
The following is a report on the Environmental Technology Verification (ETV) test of the
Calgon Carbon Corporation Ultraviolet Reactor being marketed under the name Sentinel™. The
study was conducted at the Mannheim Water Treatment Plant in Kitchener, Ontario, Canada in
the spring of 1998.
It is important to note that the purpose of the ETV program is to verify field performance of
commercially available, innovative drinking water technologies. As such, it is clearly not a
scientific study of discovery and is not intended to establish scientific principles or to explore
new findings; nor is the intent to establish or evaluate testing and analytical protocols. Scientific
methods are followed to the extent that they are methodical and rigorous, and that the data are
gathered without contamination, but these are tests conducted in the field, and subject to the field
conditions of unpredictability.
The purpose of the testing is to present to state regulators and water treatment specialists
reasonable expectations of performance to allow them to evaluate the technology for application
to specific needs, and is specifically targeted to small systems where extensive pilot testing may
prove prohibitive. It can not be expected to answer all their questions, however, it may serve as
an introduction to those technologies.
To that end, the basics of the technology are presented, and a fair and comprehensive report of
the procedures and methods of the test is illustrated. It is not intended as a tutorial of the
technology and certainly not of the underlying scientific principles either of the technology or of
the testing and analytical procedures. It is expected that if additional information at a deeper
level is required, the reviewers will be capable of independent research.
The study was conducted by Cartwright, Olsen and Associates, LLC (COA), as a qualified Field
Testing Organization (FTO) on behalf of the NSF and EPA. The design of the test was based on
Protocols and Test Plans developed through consensus of stakeholders and approved by the EPA
and NSF, and was specified in a Field Operations Document (FOD). The FOD was approved by
NSF and EPA prior to beginning the verification study. The FOD is an on site, working
document which is specific to the technology under study, and is based on the NSF/EPA test
plan. It is an effort to anticipate on site problems and performance, and to guide the operators in
conducting the test.
The challenge employed live Cryptosporidium parvum oocysts and Giardia muris cysts. The test
plan specified C. parvum, the manufacturer elected to challenge G. muris as well. G. muris is
accepted as a test organism by the NSF/EPA test plan and has the advantage of being non-
infectious to humans, however, G. muris is considered by some to be a more fragile cyst than G.
lamblia. In hindsight, the challenge of G. muris was probably irrelevant and unnecessary, but the
design included that challenge in an effort to establish a dose/kill relationship. Given the
limitations of time and budget however, laboratory facilities allowed for a near simultaneous
challenge of the two organisms selected. Accordingly, this report will focus on the effect of the
IV
-------
reactor on C. parvum. No inference can be made concerning G. muris with respect to
inactivation nor to infectivity.
C. parvum was studied by vital dye, excystation and animal infectivity. The vital dye or
excystation were required by the test plan; however, in collimated beam bench testing, Calgon
Carbon Corporation (CCC) had determined that animal infectivity was a better indication of UV
performance. The reasons are not known, and could be the subject of future research, but
because of the bench testing CCC elected to add the optional animal infectivity study to the
mandated vital dye and excystation procedures.
Different researchers have utilized several different methods and designs to determine UV
dosages so care must be taken when comparing different studies. The information presented in
this report is scientifically sound and promising but warrants additional research and independent
confirmation by other investigators to insure its validity.
It was hoped that a dose response would be established, although that was not the primary intent
of the study. The UV reactor proved to be more effective than expected at lower irradiance
levels in inactivating the organism, hence too few data points were available to establish a UV
dose response curve. None-the-less, the reactor proved to inactivate the organism as verified by
animal infectivity
Several essential but controversial issues arose: 1) there is no consensus in the microbiological
community with respect to methods to determine viability/infectivity of C. parvum; 2) the
methods through which UV irradiance is measured and/or calculated within a reactor is also
subject to disagreement. There are also differences in individual preferences for establishing UV
dose, proper selection of organism for test, methods of quality control and nomenclature. COA
does not pretend to be a referee in these discussions, nor to have an opinion beyond that specific
to this testing and verification study.
The report, along with the study, is a collaborative effort, and hence represents the styles of the
various contributors. Appendices have been left intact and all commentary by COA has been
confined to the document proper. The value of this report is to its intended audience, the
purveyors of small system water plants, and any ancillary benefits to either the microbiological or
UV scientific communities are beyond the concerns of this document.
The ETV program is an evolving program and thus subject to review and change. By its nature it
will ever be addressing new and innovative technologies, where analytical and scientific
techniques may be non-standard or uncertain, and often subject to dispute. As the program
develops standard procedures may become formulated, and to that degree we are pleased to have
made a contribution.
Please note that in the tables used for data collection as well as other support documentation the
use of the term Ray ox® UV Tower. The Sentinel™ UV Reactor (not Ray ox® UV Tower) is the
correct name for the equipment package studied in this performance evaluation.
-------
Contents
Notice ii
Foreword iii
Preface iv
Contents vi
Acronyms, Abbreviations, Formulas and Symbols x
Acknowledgments xiii
Chapter 1 Verification Statement 1
Chapter 2 Introduction 7
2.1 Historical Background 7
Chapter 3 Procedures & Methods Used In Testing 9
3.1 Equipment Capabilities and Description 9
3.7.7 Equipment Description 9
3.7.2 Instrumentation and Controls 10
3.2 Testing Overview 11
3.3 Field Testing Methods 12
3.3.7 Water Quality 12
3.3.1.1 Test Site Description and Characterization 12
3.3.1.2 Water Analysis 13
3.3.2 Equipment Installation 15
3.3.3 Flow Measurements 16
3.3.4 Operating Parameters 16
3.3.5 Irradiance Measurement 17
3.3.6 Microbiological Challenge Methods 18
3.3.6.1 Description of Cryptosporidium parvum 18
3.3.6.2 Enumeration of oocyst Suspensions 19
3.3.6.3 Challenge Seeding Schedule 19
3.3.6.4 Viability/Infectivity Analysis 20
3.3.6.5 In vitro Viability Assays 20
3.3.6.6 Neonatal Mouse Infectivity Assays 21
3.3.6.6.1 Preparation of Infectious Doses 21
3.3.6.6.2 Infectivity Assays 21
3.3.6.7 Oocyst Trip Controls, Holding Times and
Temperature 21
3.3.7 Logw Inactivation Calculation 22
vi
-------
3.3.8 Health and Safety Measures 22
3.4 Calculation of UV Dose 23
3.4.1 Estimation of Exposure Time 23
3.4.2 The UV Dose Concept 23
3.4.3 Recognized Methods for Estimating UV Dose 24
3.4.4 Multiple Point Source Summation Model for
Estimating UV Dose 25
3.4.4.1 Model Assumptions 25
3.4.4.2 Model Adjustments for Factors Affecting
UVDose 25
3.5 Data Management 26
Chapter 4 Results & Discussions 27
4.1 Feed Water Quality 27
4.1.1 On-Site Test Results 27
4.1.2 Laboratory Test Results 27
4.1.3 UV254Scan 27
4.2 UV Dose Calculation 28
4.2.1 Flow Rate Studies 28
4.2.2 Exposure Time 28
4.2.3 Multiple Point Summation Model Results 29
4.2.4 UVIrradiance by Direct Measurement 30
4.3 Microbiological Challenge Results 32
4.3.1 In vitro Viability Assays 32
4.3.2 Neonatal Mouse Infectivity Assays 32
4.3.3 Dose Response Infectivity Calculations 34
4.3.4 Comparison of InactivationMethods 35
4.3.5 Affect ofoocyst Trip Controls, Holding Times and
Temperature on Results 37
4.4 Operations and Maintenance 38
4.4.1 Ease of Operation 38
4.4.2 UV Sensor 39
4.4.3 Lamp Fouling/Cleaning 39
4.4.4 Lamp Hours and Power Consumption 41
4.4.5 O&MManual 43
Chapter 5 Limitations 45
5.1 Method for Determining Viability and Inactivation 45
5.2 UV Dose Estimates and Measurements 45
vn
-------
5.3 Low Flow Rate Testing 46
5.4 Feed Water Conditions 46
5.5 Exclusion of Giardia Data 46
Chapter 6 Conclusions 47
Chapter 7 Recommendations 48
Chapter 8 References 49
LIST OF TABLES
Table 3-1 - Testing Schedule For Water Quality Parameters 14
Table 3-2 - Schedule of Collection of Operating Data 17
Table 3-3 - Cryptosporidium parvum Challenge Seeding Schedule 19
Table 4-1 - On-Site Testing for the Verification Period 27
Table 4-2 - Parameters Measured By Spectrum Laboratories 28
Table 4-3 - UV Dose in mW-s/cm2as Calculated by the MPSS Model 30
Table 4-4 - Effects of UV Exposure on Cryptosporidium parvum viability 32
Table 4-5 - Normalized (percent) oocyst/cyst Inactivation Using the Sentinel™ System
and in vitro Viability Assays 33
Table 4-6 - Effects of UV exposure on Cryptosporidium parvum oocyst infectivity 33
Table 4-7 - Log inactivation of Cryptosporidium parvum following exposure to UV light:
Comparison of in vitro methods with mouse infectivity assays 36
Table 4-8 - Wiper Cycle Frequency 40
Table 4-9 - Power Consumed during Challenge Test Periods C. parvum 42
Vlll
-------
LIST OF FIGURES
Figure 3-1 - Illustration Of The Sentinel™ UV Reactor 10
Figure 4-1 - Average of UV Irradiance Over Time 31
Figure 4-2 - Turbidity vs. UV Irradiance 31
Figure 4-3 - UV Dose Response Curve for C. parvum oocyst infectivity 35
Figure 4-4 - Cryptosporidiumparvum Log Inactivation Ratio vs. UV Dose 37
Figure 4-5 - Power Used During Testing 41
Figure 4-6 - Voltage Used During Testing Period 42
Figure 4-7 - Amperage Used During Testing Period 43
LIST OF APPENDICES
A. Mannheim (Waterloo) Water Treatment Plant Authority Feed Water Analysis
B. Operating Conditions Treatment Equipment Performance
C. Test Site Photos, Illustrations & Flow Charts of Equipment
D. Calculation of the Irradiance from the UV Lamp in the Sentinel™ UV Reactor
E. Laboratory and Equipment Certifications
F. Work Sheets for Calculation of UV Dose
G. Package Treatment Plant Operating Data Spreadsheet
H. Daily Field Notes
I. Laboratory Analysis of Inorganic Chemical Parameters, Total Organic Carbon and UV
Absorption (performed at Spectrum Labs. Inc.)
J. Representative Scan & Graph Of UV Absorbance
K. Work Sheets and Data for Animal Infectivity Studies (performed at University
of Arizona)
L. Work Sheets and Data for Excystation and Vital Dye Studies (performed at Clancy
Environmental Consulting)
IX
-------
ACRONYMS, ABBREVIATIONS, FORMULA AND SYMBOLS
AOC - Assimilable Organic Carbon
AWWA - American Water Works Association
AWWARF - American Water Works Association Research Foundation
CCC - Calgon Carbon Corporation
CEC - Clancy Environmental Consultants, Inc.
COA - Cartwright, Olsen and Associates, LLC
DOC - dissolved organic carbon
ETV - Environmental Technology Verification.
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 = 0.833 imperial gallons
1 imperial gallon = 4.54 liters = 1.2 US gallons.
Lamp Fouling - Lamp fouling is the reduction in UV Irradiance caused by the presence of
certain organic and inorganic ions in the water that can result in the accumulation of mineral
deposits or biofilm on the quartz sleeves covering the lamps. Chemical or mechanical cleaning is
needed to restore the UV Irradiance to design conditions.
LED - light emitting diode
Low Pressure Lamps - Low pressure lamps operate at a temperature between 38° and 49°C
(100 and 120°F) to produce near monochromatic radiation at 253.7 nm. These lamps typically
have a linear power density of about 0.3 W/cm.
Medium Pressure Lamps - Medium pressure lamps produce a broad spectrum of UV light
(extending over the 200-300 nm range of microbiological sensitivity with a maximum output at
about 255 nm) with a higher irradiance and operating at a much higher operating temperature
(surface temperatures >500°C) than do low pressure Hg lamps. The linear power density is also
much higher (typically 100-300 W/cm).
MPN - Most Probable Number
-------
MPSS - Multiple Point Source Summation
NIST - National Institute of Standards and Technology
NSF - NSF International
O&M - Operations and Maintenance
OSHA - Occupational Safety and Health Administration
PLC - Programmable logic controller
RMP - Residual Management Plant
SCADA - Supervisory Control and Data Acquisition
SD - Standard Deviation
SiC - Silicon carbide
SWTR - Surface Water Treatment Rule
TOC - Total Organic Carbon
USEPA - United States Environmental Protection Agency
UV Absorbance - Absorbance through a fixed pathlength is related as:
A = - log T.
The relationship is often expressed as Trammittance where:
%T= 100 x 10 -A.
UV Absorption is the transfer of energy from an electromagnetic field to a molecular entity. It is
expressed by an absorption coefficient which is the absorbance divided by the optical pathlength,
/. Thus,
a = A// = -(I//) log T Since A is dimensionless, a is often given the unit m"1 or cm"
UV Absorption was measured as in Standard Method 5910.
UV Dose - The UV energy is quantified to a dose by multiplying the average UV irradiance (the
UV irradiance averaged over the exposed volume of the reactor) by the actual exposure time:
XI
-------
Dose (|J,W s/cm2) = Average UV Irradiance (|j,W/cm2) x Exposure Time (s)
To avoid the use of large numbers and the potential for error, UV dose is often indicated as mW
s/cm^, where 1 mW s = 1000 |j,W s.
UV Intensity - Irradiance is often referred to as intensity. Intensity is the traditional term for
photon flux, fluence rate, irradiance or radiant power (radiant flux). It is recommended the term
be used only for qualitative descriptions.
UV Irradiance - The UV power on a specific area (usually 1 cm2) coming from all incident
directions and expressed as microwatts per square centimeter (|iW/cm2) or milliwatts per square
centimeter (mW/cm2). This is occasionally (and erroneously) referred to as UV Intensity.
UV Output - The amount of power (in the wavelength range of 200-300 nm) delivered from the
lamp into the water and described in terms of watts (W) per lamp. The absolute free-standing
UV power of the lamp is decreased by end loss and by transmission losses through the quartz
sleeve. The UV output can be reduced because of lamp aging, water temperature, and lamp
fouling.
UV Power - The amount of power at all wavelengths in a specific range delivered to the water.
UV Transmittance - The ability of the water to transmit UV light. Transmittance of a water
sample is generally measured as the percentage (%T) of the ratio of transmitted light irradiance
(E) to incident light irradiance (E0) through an operationally defined pathlength (L). Many
commercially available spectrophotometers actually report the Absorbance (A) for a fixed
pathlength (L) of the sample. Percent Transmittance and Absorbance can be related through an
operationally defined pathlength (L) as %T = 100 x 10"A. Many naturally occurring organic and
inorganic constituents (e.g., natural organic matter, iron, nitrate) will absorb energy in the UV
wavelengths, thus reducing the transmittance of the water. This reduced transmittance often
interferes with the disinfection efficiency of a UV disinfection system.
Vital Dye - DAPI/PI (4', 6-diamidino-2-phenylindole and propidium iodide)
WSWRD - Water Supply and Water Resources Division
WTP - Water Treatment Plant
xn
-------
ACKNOWLEDGMENTS
The Field Testing Organization, Cartwright, Olsen and Associates, LLC (COA) was responsible
for all elements in the testing sequence, including collection of samples, calibration and
verification of instruments, data collection and analysis, test data management, data
interpretation and the preparation of this report.
Cartwright, Olsen & Associates, LLC
19406 East Bethel Blvd.
Cedar, Minnesota 55011
(612)434-1300 Fax (612) 434-8450
The laboratory selected for microbiological analysis of this validation was:
Clancy Environmental Consultants, Inc.
272 North Main Street
St. Albans, VT 05478
(802) 527-2460 Fax (802) 524-3809
Animal Infectivity Studies:
Marilyn M. Marshall
Sterling Parasitology Laboratories
University of Arizona
Tucson, Arizona
(520)621-4439 Fax (520) 621-3588
Additional, non-microbiological, analytical work was performed by:
Spectrum Labs Inc.
301 West County Road E2
St. Paul, MN 55112
(612)633-0101 Fax (612) 633-1402
The Manufacturer of the Equipment was:
Calgon Carbon Corporation
130 Royal Crest Court
Ontario, Canada, L3R 041
(905) 477-9242 Fax (905) 477-4511
We wish to thank the participants in this test, especially Brian Pett and staff of the Mannheim
Water Treatment Plant for their generous cooperation and hospitality, and the members of the
Grand River Conservation District for their courtesy and informative resources.
Xlll
-------
Chapter 1
Verification Statement
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, D.C. 20460
ENVIRONMENTAL TECHNOLOGY VERIFICATION PROGRAM
VERIFICATION STATEMENT
TECHNOLOGY TYPE: ULTRAVIOLET RADIATION USED IN PACKAGED
DRINKING WATER TREATMENT SYSTEMS
APPLICATION: MICROBIOLOGICAL CONTAMINANT INACTIVATION
TECHNOLOGY NAME: SENTINEL™ ULTRAVIOLET REACTOR (R-ll, Model 6-1)
COMPANY: CALGON CARBON CORPORATION
OXIDATION TECHNOLOGIES
ADDRESS: 130 ROYAL CREST COURT
MARKHAM, ONTARIO, CANADA L3ROA1
TELEPHONE: (905) 477-9242
The U.S. Environmental Protection Agency (EPA) has created a program to facilitate the deployment of
innovative technologies through performance verification and information dissemination. The goal of
the Environmental Technology Verification (ETV) Program is to further environmental protection by
substantially accelerating the acceptance and use of improved and more cost effective technologies. The
ETV Program is intended to assist and inform those involved in the design, distribution, permitting, and
purchase of environmental technologies. This verification statement provides a summary of the
performance results for the Calgon Carbon Corporation (CCC) Sentinel™ Ultraviolet Reactor, R-ll,
Model 6-1 (Sentinel™). The Sentinel™ is a package drinking water treatment system that uses medium-
pressure ultraviolet (UV) lamps operating at a higher temperature than low-pressure lamps to produce a
broad spectrum of UV light with a higher irradiance to inactivate microbiological contaminants.
ABSTRACT
The EPA and NSF International (NSF) verified the performance of the Sentinel™ under the EPA's ETV
program. The Sentinel™ obtained an estimated 3.9 logic inactivation of Cryptosporidium parvum (C.
parvum) as determined by animal infectivity methods, at an estimated UV dose of 20 mW-s/cm , when
fed finished (treated but not chlorinated) water that was seeded with C. parvum at a flow rate of
approximately 215 gallons per minute (gpm). When using other methods (vital dyes and in vitro
excystation), a maximum of 1.2 logic inactivation of C. parvum was observed. During the
microbiological seeding challenge, the finished water fed to the system had these characteristics:
• turbidity less than or equal to 0.11 NTU
• nitrate less than or equal to 4.0 milligrams per liter (mg/L)
• pH range of 7.40-7.76
-------
• temperature range of 8.8 - 12.3 °C
• UV254 absorption coefficient ranges 0.02-0.06 cm"1.
At a flow rate of 25 gpm, the Sentinel™'s power requirements were verified as 1.046 + 0.046 kW per
lamp at full power. Three of the UV irradiance sensors failed and some of the automatic quartz sleeve
wipers had operational difficulties including a broken weld. CCC informed NSF of their intent to
improve these portions of the Sentinel™.
Details of the verification testing, including the testing data and discussion of results, may be found in
the report entitled "Environmental Technology Verification Report: Inactivation of Cryptosporidium
parvum oocysts in Drinking Water: Calgon Carbon Corporation's Sentinel™ Ultraviolet Reactor"
(EPA/600/R-98/160).
PROGRAM OPERATION
The EPA, in partnership with recognized verification organizations, objectively and systematically
evaluates the performance of innovative technologies. Together, with the full participation of the
technology developer, they develop plans, conduct tests, collect and analyze data, and report findings.
The evaluations are conducted according to a rigorous demonstration plan and established protocols.
NSF, a not-for-profit organization dedicated to public health safety and protection of the environment,
assured data quality objectives were met during testing through their oversight and management of
verification activities. The verification testing of the Sentinel™ was performed by Cartwright, Olsen and
Associates, LLC (COA), an NSF-qualified Field Testing Organization for the Package Drinking Water
Treatment Systems (PDWTS) ETV Pilot.
TECHNOLOGY DESCRIPTION
The Sentinel™ is a medium pressure UV water treatment system designed to inactivate microbiological
contaminants. There are two major UV light technologies: low-pressure and medium pressure lamps.
The low-pressure lamp UV light technology emits most of its energy at the 253.7 nm wavelength. The
medium pressure lamps produce a broad spectrum of UV light (extending over the 200-300 nm range
with a maximum output at about 255 nm) with a higher irradiance and operating at a much higher
operating temperature (surface temperatures >500°C) than low pressure lamps. The linear power density
is also much higher (typically 100-300 W/cm).
The system is a skid-mounted, stand-alone system equipped with a control panel, power supply,
transformer, and fail-safe and monitoring controls. The system has two UV reaction chambers contained
within a stainless steel column (dimensions: 10" diameter, 80" tall). Each reaction chamber has three 1
kW medium pressure ultraviolet lamps. Each lamp can be operated at full or reduced power. Each lamp
is contained within a quartz sleeve aligned perpendicular to and across the flow of the water. The UV
dose for the system is calculated using a multiple point source summation (MPSS) model that is
undergoing a peer review. The hydraulic design for the system is for continuous flow rates up to 500
gpm (0.7 mgd). Throughout the verification testing period, the system was operated at a flow rate of 25
gpm during regular flow conditions and at approximately 215 gpm (814 L/min) during inoculated
feedwater conditions.
VERIFICATION TESTING DESCRIPTION
In March and April of 1998, the ability of the Sentinel™ Ultraviolet Reactor to inactivate the protozoa C.
parvum oocysts was tested at the Mannheim Water Treatment Plant in Kitchener, Ontario, Canada.
The Grand River is the source water for the Mannheim Water Treatment Plant. Pretreated surface water
(treated by coagulation, flocculation, sedimentation, ozonation, and filtration) was inoculated with C.
parvum oocysts and fed to the Sentinel™. The pretreated surface water exhibited the following
-------
characteristics during the microbiological seeding portion of the verification testing: turbidity
concentrations less than or equal to 0.11 NTU; pH range 7.40-7.76; temperature range 8.8-12.3 °C;
nitrate concentration less than or equal to 4.0 milligrams per liter (mg/L); total organic carbon (TOC)
concentration less than or equal to 4.5 mg/L; UV254 absorption coefficient range 0.02-0.06 cm"1.
Methods
During each day of the verification test, samples of the feed and finished water were collected, labeled
and analyzed. All analyses were performed in accordance with the procedures in Standard Methods.
The purpose of the microbiological challenge test was to demonstrate the effectiveness of the application
of medium pressure UV lamps as configured in the Sentinel™ equipment in inactivating the protozoan
oocysts in the field. The challenge testing was performed on finished water representing a uniform water
quality matrix.
The Sentinel™ was challenged with live oocysts and consisted of the following steps:
1) the introduction of live oocysts into the water stream and their passage through the
Sentinel™,
2) the recovery of the oocysts from the water stream,
3) the determination of their viability and/or infectivity,
4) the calculation of logic inactivation.
The organisms were introduced upstream of a static mixer ahead of the reactor and collected on 1 \\m
filters after the reactor. The overall flow rate during the tests was approximately 215 gpm (814 L/min).
The filters were shipped to Clancy Environmental Consultants, Inc. (CEC) in Vermont where the
organisms were isolated, concentrated and subjected to analysis by in vitro methods to determine
viability. Additionally, for C. parvum oocysts, animal infectivity experiments were also conducted to
ascertain the levels of inactivation demonstrated by in vitro assays and to provide further evidence for the
correlation between in vitro methods and neonatal mouse infectivity, following oocyst exposure to UV
light. The details of the seeding, recovery, and viability assays are found in Clancy et al. (1998).
VERIFICATION OF PERFORMANCE
The following is a summary of the findings of the verification testing of the Sentinel™:
Water Quality Results
The following two tables present the mean, minimum, and maximum water quality parameter
concentration results of the influent and effluent samples collected during the verification testing:
On-Site Water Quality Sampling Results (March 30 through April 13)
Temp. Bench In-Line
("Q1 pH2 Turbidity3 Turbidity4
Mean
Minimum
Maximum
10.4/11.4
8.8/8.8
11.0/12.4
7.6/7.6
7.4/7.4
7.8/7.7
0.095/0.094
0.056/0.053
0.147/0.134
0.072/0.077
0.041/0.041
0.112/0.147
1 - Temperature from influent/effluent of reactor.
2 - pH from influent/effluent of reactor.
3 - Turbidity in NTU from bench influent/effluent of reactor.
4 - Turbidity in NTU from on-line turbidimeter, filter 3/filter 4.
-------
Laboratory Water Quality Sampling Results (microbial challenge test days)
Alk
(mg/1)1
Al
(mg/1)1
Color
(TCU)1
Iron
(mg/1)1
Mang
(mg/1)1
NO3
(mg/1)1
UV254
(cm"1)1
Mean
Minimum
Maximum
164/164
150/150
180/180
0.26/0.3
0.06/0.08
0.86/0.48
5/5
5/5
5/5
0.15/ND
ND/ND
0.5/ND
0.01/.01
ND/ND
0.03/0.02
3.58/3.34
3/3
4/3.7
0.0464/0.0366
0.0365/0.0214
0.0551/0.0427
1 - Concentration from influent/effluent of reactor.
ND = Not Detected
Microbiological Results
Results of the C. parvum inactivation by the Sentinel™ as determined by animal infectivity, vital dyes,
and in vitro excystation studies are presented in the following table:
Summary of the Results of C. parvum inactivation by the Sentinel
TM
Challenge
Date
3/31/98
4/6/98
4/7/98
4/1/98
4/8/98
Sentinel™
UV Dose at 215 gpm
(mW-s/cm2)
167 (High) - 2 lamps full
152 (High) - 2 lamps full
137 (High) - 2 lamps full
69 (Medium) - 2 lamps
reduced
20 (Low) - 1 lamp reduced
%Trans-
mittance
90.0
89.4
87.9
90.1
91.1
Logio Inactivation
via animal
infectivity
>4
Not Done
Not Done
>4
3.9
Vital dyes
assay
(D API/PI)
1.2
0.9
0.5
0
0
In vitro
excystation
0.4
0.4
0.2
0
0
Mouse infectivity assays with high and medium UV doses demonstrated no infection in neonatal mice
despite oral inoculation of up to IxlO5 oocysts. The oocysts which had been exposed to a low UV dose
resulted in 4.5% infection (1 of 22 mice) with an inoculum of IxlO5 UV exposed oocysts per mouse;
however, no mice were infected when inocula of either IxlO4 or IxlO3 UV exposed oocysts were
administered into a total of 36 mice.
Operations and Maintenance Results
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 and monitored 24 hours a day until the completion of a period of 27
days. Results observed included:
• Three of the contained irradiance sensors failed due to unexpected electronics problems. CCC is
taking action to redesign the sensor circuit board.
• The automatic quartz sleeve wipers ceased operating for many reasons including a broken weld.
The wiper mechanism is being redesigned and will be the subject of a separate ETV evaluation.
CCC has determined that the cause of wiper failure was the impact force of the brush with the
wiper stop at the end of the extended travel position.
• During the maintenance period the power consumption was approximately 1.046+0.046 kW per
lamp. Assuming daily operation of six lamps at full power, the power demand is estimated at
150.6 kW per day.
-------
• The O&M manual supplied by the manufacturer was specific to this equipment and included all
the components of the pilot plant. Drawings and illustrations showing the positions of the meters
and controls are included along with explanations of control functions and step-by-step
instructions for common maintenance functions, such as: replacement of lamps, quartz tube
cleaning and reactor cleaning. Complete instructions for equipment start-up and shut-down
procedures were listed in this guide. The control panel is thoroughly explained so that all
programmable functions, including wiper cycles, lamp set-points for alarms and other PLC
parameters are easily learned by even inexperienced personnel. Safety measures included
detailed instructions concerning high voltage, protection against UV irradiance, and the
procedures for mercury spills in the event of lamp breakage. A trouble shooting guide was
furnished.
Conclusions
Through this testing it was established that at a process flow rate of approximately 215 gpm the
Sentinel™ could obtain an estimated 3.9 logic inactivation of C. parvum oocysts as determined by animal
infectivity results with one lamp illuminated (out of six) at reduced power (0.5 kW). Greater (> 4 logio)
inactivation was achieved at 215 gpm with higher UV doses, respectively, with two lamps at reduced
power (0.5 kW each), and with two lamps at full power (1.0 kW each), again as determined by animal
infectivity results.
Furthermore, the use of in vitro methods (vital dyes and in vitro excystation) significantly under-
estimated oocyst inactivation when compared to neonatal mouse infectivity.
During the verification period, water quality parameters that influence UV absorbance were measured to
assist in evaluating other waters for application of this UV system. During the challenge periods, UV254
absorption coefficient was between 0.02 and 0.06; turbidity was <0.11 NTU. No iron or manganese was
detected in the sample water; nitrates were no greater than 3.7 mg/L and total organic carbon was no
greater than 4.3 mg/L.
Also of importance to this study was the operation of the equipment in the field. Several deficiencies
were noted with wiper failures, irradiance sensor, and attenuation tubes. CCC has informed COA and
NSF that they are taking action to improve these portions of the system.
Limitations
The PDWTS ETV Pilot verifies the performance of innovative water treatment systems using consensus
methods and procedures. This verification identified limitations associated with the use of non-standard
methods. For example, the verification identified concerns about the methods for assessing oocyst
viability and estimating UV dose. The lack of consensus on evaluation methods and procedures or the
application of a technology is a reflection of the uncertainties associated with emerging technologies,
developing analytical techniques and engineering applications. The resolution of these uncertainties is
within the purview of rigorous scientific research and not the ETV program. A detailed description of
the methodology limitations associated with this performance testing is provided in the Verification
Report (EPA/600/R-98/160).
-------
Original Signed by
E. Timothy Oppelt
E. Timothy Oppelt
Director
National Risk Management Laboratory
Office of Research and Development
United States Environmental Protection Agency
Original Signed by
Tom Bruursema
5/13/99
Tom Bruursema
General Manager
Environmental and Research Services
NSF International
NOTICE: Verifications are based on an evaluation of technology performance under specific,
predetermined criteria and the appropriate quality assurance procedures. 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 Inactivation of
Microbiological Contaminants dated March 8, 1998, the Verification Statement
(EPA/600/R-98/160VS), and the Verification Report (EPA/600/R-98/160) are available
from the following sources:
(NOTE: Appendices are not included in the Verification Report.
available from NSF upon request.)
1. Drinking Water 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/etv (electronic copy)
3. EPA web site https://www.epa.gov/etv (electronic copy)
Appendices are
-------
Chapter 2
Introduction
2.1 Historical Background
Ultraviolet light for disinfection of water and wastewater has grown in popularity throughout
North America in recent years, although it has been employed in Europe since the 1950s. There
are some 500 UV installations in North America, and it is estimated there are over 2,000 in
Europe (USEPA 1996). Improvements in the lamps, both mechanically and in their irradiance
and power, have made UV a more attractive solution to disinfection problems. It is often
preferred over chemical disinfection methods because:
• it has a proven effectiveness in inactivating many pathogens, especially bacteria and viruses
(USEPA 1996);
• it requires relatively short contact times;
• it has reduced O&M costs and space requirements;
• there is lesser risk of by-product formation than chlorine, ozone or chlorine dioxide;
• there are lower capital costs than many alternative technologies;
• there is no likelihood of overdosing, or residual formation.
The disadvantages inherent in the technology include:
• the relative lack of information about inactivation of protozoa;
• the lack of residual (although it is acknowledged that by employing UV, lower levels of
traditional chemical disinfectants can be used);
• some uncertainties in measuring UV dose (an issue discussed in Section 3.4);
• some possible limitations on total capacity (Wolfe, 1990);
• UV irradiance in water may have restraints determined by water chemistry, for example,
turbidity, pH, temperature, UV absorption, nitrates, TOC and True Color.
Ultraviolet light has often been used in water and wastewater treatment to destroy bacteria and
viruses, an application for which there is considerable documentation. In recent years, especially
with the outbreaks of infection from protozoan cysts and oocysts, such as Giardia lamblia (G.
lamblia) and Cryptosporidium parvum (C. parvum) respectively, and with the increased public
awareness and concern (MacKenzie et al., 1994), efforts to capture or inactivate these microbes
have accelerated. US drinking water utilities rank Cryptosporidium research as their highest
priority resulting in increased research funding by US utilities, AWWARF and the USEPA (Frey
etal., 1997).
Conventional methods of water treatment including gravity filtration and chlorination have not
been effective against protozoan oocysts, especially Cryptosporidium, in part because of their
size and resistance to chemicals. Treatment plants that are otherwise in compliance with public
health treatment standards are thus vulnerable to outbreaks of disease (Kiminski 1994,
LeChevallier et al., 1991, Korich et al., 1990).
-------
With the increased awareness of pathogens resistant to traditional disinfection techniques,
specifically the parasitic oocysts of G. lamblia and C. parvum, and with the implementation of
the Enhanced Surface Water Treatment Rule (SWTR) and the Groundwater Rule in the near
future, it is expected that the search for alternative disinfection technologies will grow
significantly. This verification study specifically addresses C. parvum.
As of this date, it is generally accepted that low pressure, low power UV lamps have not been
effective against protozoan (oo)cysts (USEPA 1996). While this study does not attempt to
address the parameters of either wavelength or irradiance specific to the inactivation of the
pathogens, it was intended to demonstrate the effectiveness of the Sentinel™ UV Reactor in their
inactivation. The Sentinel™ UV Reactor differs from more common UV schemes in that it
utilizes medium pressure, high powered UV lamps.
Among the concerns inherent in the use of UV disinfection is that of UV dose and its
measurement. UV dose is related to the UV energy delivered to the microorganism by the UV
lamp through the water and is a function of residence time (exposure) and average UV irradiance
in the water. Factors that affect the irradiance delivered are the wattage and the spectral radiant
power of the lamp (the spectrum of which is dependent on the pressure). Factors inherent in the
water include the absorbance of the water over the wavelength range of interest, which is itself a
function of the true color, turbidity and the composition of the water, especially the dissolved
organic matter. The geometry of the reactor was addressed in the test design for the Sentinel™
UV Reactor. Thus, careful attention has been paid to the feedwater characteristics, the physics of
the reactor and to the calculation of the irradiance of the lamps. The ability to perform in a test
situation, however, is not always sufficient for a proper evaluation. Thus, the ETV Test Plan
requires additional parameters to offer engineers and public health regulators a forthright
appraisal of the equipment and its applicability to small drinking water systems.
-------
Chapter 3
Procedures & Methods Used In Testing
3.1 Equipment Capabilities and Description
The equipment tested was a medium pressure ultraviolet light disinfection system designed to
inactivate microorganisms including C. parvum protozoan oocysts.
3.1.1 Equipment Description
The equipment tested was referred to during testing as a CCC 6 kW Rayox® Ultraviolet Tower.
Subsequent to the testing and verification period, CCC changed the name of the UV reactor to
the Sentinel™ Ultraviolet Reactor System for commercial marketing purposes. The equipment
specifications are the same.
Two UV reaction chambers are contained within a stainless steel column with six 1 kW medium
pressure ultraviolet lamps, three in each chamber. Each lamp can be powered at full or reduced
power. The lamps are each contained within a quartz sleeve aligned perpendicular to and across
the flow of the water. Figure 3-1 shows an illustration of the Sentinel™ UV Reactor.
There are two major UV light technologies: low-pressure and medium pressure lamps. The low-
pressure lamp UV light technology emits most of its energy at the 253.7 nm wavelength. The
medium pressure lamps produce a broad spectrum of UV light (extending over the 200-300 nm
range with a maximum output at about 255 nm) with a higher irradiance and operate at a much
higher temperature (surface temperatures >500°C) than low pressure lamps. The linear power
density is also much higher (typically 100-300 W/cm).
The maximum design temperature is 60°C and maximum design pressure 50 psi. The equipment
that was furnished on site was R-l 1, Model 6-1. It has a footprint of 44-1/2 inches by 150-7/8
inches (1.13 meters by 3.83 meters) including the static mixing cylinder. The main body is 44-
1/2 inches by 81 inches (1.13 meters by 2.06 meters). It stands 94 inches tall (2.39 meters). The
UV reactor proper is a cylindrical vessel 10 inches in diameter and 80 inches tall. The system is
a stand-alone with control panel, power supply and transformer, along with fail-safe and
monitoring controls on a skid mounted platform.
The lamps are contained within a quartz sleeve; the sleeves are cleaned automatically by a
pneumatically powered stainless steel brush. The automatic cleaning device was proprietary to
CCC and has been patented. The cleaning cycles are controlled by a programmable logic
controller (PLC) and can be programmed according to the fouling conditions of the subject
water. A pneumatic valve was triggered by the PLC and the brushes move along the quartz tube.
Following a brief (and adjustable) period, the brushes return and are reset for the next sequence.
UV sensors are located within the chamber, 7.5 centimeters from—and aimed at—the center of
each lamp, through an external Teflon probe. They measure lamp irradiance as transmitted to the
sensor through the Teflon, the quartz sleeve and the surrounding water. The UV irradiance
-------
detectors are mounted on a circuit board located within the chambers to measure the UV
irradiance and are calibrated to read out in mW/cm2 to a meter. Factors included in calculating
the response of the irradiance sensors include the quartz sleeve, the spectral emission of the UV
lamp, the UV absorbance of the water, and the attenuation of the Teflon probe ends. The
measured irradiance was converted through factors to account for the Teflon tube, the fiber optic
cable and the sensor.
/-SAMPLE PORT
RAYOX REACTOR--.
TEMPERATURE SWITCH, TSH-1118 ^••-**<--
UV SENSOR, 6 PLACES -"--,. I f
SAMPLE PORT ----.„
UV PORT, 6 PLACES
REACTOR ENCLOSURE ---.
l»f ,'~TRANSMITTANCE CONTROLLER
'""'" -•'' 8 PLACES
v FLOWMETER. FT-1162
'-SAMPLE PORT
•^ DRAIN LINE
1" HOSE CONNECTION (HIDDEN)
Figure 3-1. Illustration Of The Sentinel™ UV Reactor
3.1.2 Instrumentation and Controls
The package plant as shipped contains a flow meter, irradiance sensors, volt and amp meters for
each lamp. In addition, there were a number of instruments added to the package plant specific
for this test.
The fail-safe system includes a power drop sensor that indicates if a lamp has failed and can be
programmed to send an alarm, shut down the system (ceasing the flow) or both. There is a dual
point sensor such that an alarm can be triggered at one level or the unit aborted if required.
The flowmeter used to measure the flowrate through the treatment reactor is an Endress and
Hauser Promag 3OF, factory calibrated and shop verified.
The on-board SiC detectors are calibrated with a portable IL 1700 radiometer. Each in-line
sensor reads to a PLC driven digital display that shows lamp hours, starts, irradiance levels and
other system parameters including flow rates and lamp cleaning cycles.
10
-------
3.2 Testing Overview
The testing was conducted during portions of March and April 1998 and consisted of an
operations study where the performance of the package system was evaluated, and a verification
study, where the package plant was challenged with C. parvum. According to the Mannheim
WTP's records (Appendix A), the dissolved organic carbon (DOC) was higher in summer and
autumn of 1997 than in the spring of that year, along with true color. The testing period was
selected in part because it represents an "ice out" and higher spring run-off river flow period that
may tax the water treatment plant. Also, this season allowed for expediency of the testing.
Because the manufacturer recommends use of this product on pretreated surface water feed, only
one performance verification period was required. This selection assumes that the pre-treatment
within the water treatment plant will provide relatively consistent quality throughout the year, a
condition that was confirmed by prior treatment plant records (Appendix A).
While the river is changing from season to season and, for that matter, from day to day, the water
treatment plant is producing water of consistent quality. The readings and measurements such as
absorbance and chemical and biological parameters are taken to assure consistency (even the best
water plants are subject to some variability) and to account for changes in treatment methods.
During each day of the test, samples of both the feed and finished water were taken, labeled and
tested. These samples were taken prior to any microbiological inoculation.
The equipment was operated continuously for the verification testing period of 320 hours and the
27 day total operating period; however, brief interruptions for repair of the wipers and quartz
shields were required. These interruptions are listed in Appendix B.
While the hydraulic design is for continuous flow at rates up to 500 gpm, or in excess of 0.7
mgd, this testing period challenged the device at flow rates closer to 215 gpm. The limiting
factor in the challenge testing flow rate was the oocyst capture filters, which are limited to 50
gpm per cartridge. The housing contains four cartridges and has a total flow rate capacity of 215
gpm.
Previous work performed with a collimated beam unit using a medium pressure 1-kW UV lamp
showed >3.9 log inactivation of C. parvum at dosages as low as 41 mW-s/cm2 using the animal
infectivity assay (Bukhari et al., 1999). The purpose of the microbiological challenge test was to
demonstrate the effectiveness of the application of medium pressure UV lamps as configured in
the Sentinel™ equipment in inactivating the protozoan oocysts in the field. The challenge testing
was performed on finished water representing a uniform water quality matrix. Additional studies
of water quality were performed to assure conditions of the test and to enable the manufacturer to
predict effectiveness in other applications.
Each test introduced either 2 or 4 x 108 live oocysts concentrated as a suspension in 200
milliliters of water. If vital dye and excystation alone was employed, 2 x 108 live oocysts were
injected; if infectivity by inoculation into mice was employed in addition, 4 x 108 live oocysts
11
-------
were injected. The injections took place evenly over a period of about two minutes, during which
approximately 400 gallons of water were inoculated.
The static mixer was a Ross 4" x six -element motionless mixer in a stainless steel chamber.
Vanes within the mixer assure complete mixing and distribution of the oocysts throughout the
influent water. The expected pressure drop across the mixer at 215 gpm was 3.36 psig; the
pressure drop was measured at 3.5 psig.
A 6 inch Watts RPZ Backflow prevention device was installed upstream of the challenge
organism injection point to assure microbial contaminants were not able to migrate into the
treatment process stream of the WTP.
The microbiological samples were analyzed using vital dye and excystation. Prior studies have
indicated limited and conflicting results about UV inactivation as determined by excystation and
vital dye (Campbell et al., 1995, Clancy et al., 1998). Due to these concerns, CCC elected to
confirm C. parvum oocyst inactivation with animal infectivity studies.
More detailed procedures for vital dye, excystation and animal infectivity are summarized in the
following sections.
3.3 Field Testing Methods
3.3.1 Water Quality
3.3.1.1 Test Site Description and Characterization
The site selected for this verification testing is a surface water source located within the Regional
Municipality of Waterloo, Ontario, Canada. The cities of Kitchener and Waterloo (along with
one other city and four townships in the Region) obtain their water from both ground and surface
sources. Groundwater from 126 wells supplies 85% of the water to the communities served and
the remaining 15% is taken from the Grand River. Many of the wells within the groundwater
supplies in the area are heavily laden with iron and manganese, and others contain high levels of
sulfur. The Grand River is a reliable source of water for the community with a flow from May
through October averaging 10 cubic meters per second.
Grand River water is treated at the Mannheim Water Treatment Plant (WTP). The treatment
plant is a state of the art facility with construction completed in 1993 and capable of treating 13.3
million gallons per day. It presently processes 5 million gallons per day.
Water is withdrawn from the river, piped and stored in a 26.6 million gallon reservoir for several
days for turbidity reduction. It then is treated by coagulation, flocculation, sedimentation,
ozonation, filtration and chlorination. This finished water is sent to a blending station for
blending with groundwater prior to distribution to the municipality.
The Grand River is a major watershed of Southern Ontario and extends over 298 km (185 miles)
in length, terminating at Lake Erie. The watershed itself is 6800 km2 (2600 mi2). Four primary
12
-------
tributaries feed the Grand—the Speed, Eramosa, Nith and Conestogo rivers along with a number
of creeks.
While the region is primarily agricultural, there is a population of 732,000 within the watershed
of which some 525,000 are urban. Up river from the Mannheim WTP are dairy farms, urban
centers with wastewater effluent and chemical plants. One such plant in City of Elmira was the
site of former toxic waste location. Although the river is not a major source of power, there are
and have been power plants on the river in the past. There are three dams on the river used for
flood control.
In 1994 the river was designated a Heritage River and is thus subject to careful scrutiny and
control. A Grand River Conservation District oversees all river activities and is responsible for
monitoring the quality of the watershed.
A preliminary analysis of the feedwater, as reported by the Waterloo WTP Authority, follows as
Appendix A. This report contains analyses of the Grand River water at the effluent side of the
filter and represents the general quality of water that was directed to the test station.
The water discharged during challenge was directed to the Residue Management Plant (RMP)
which is an on site facility designed to contain the backwash water from the filters along with the
sludge from the settling basins. The effluent from the test station was directed there during
challenge periods.
At the RMP, a polymer is added, the resultant sludge is filter pressed and the cake is sent to a
landfill. Prior to the 1993 outbreak of Cryptosporidiosis in the Waterloo region the supernatant
water was recycled from the RMP to the head of the treatment plant. Following operational
changes made after the outbreak, that practice was discontinued. The filter backwash water along
with the settling sludge is assumed to have high levels of microorganisms, including G. lamblia
and C. parvum. The facility meets the requirements of the Canadian/Ontario Ministry of the
Environment and Energy as well as the Region's authorities. Supernatant now flows to a
designated collection/management area.
3.3.1.2 Water Analysis
In addition to earlier written reports of water quality from regional authorities, the parameters of
the water that impact on the verification study were identified and measured. Included are those
parameters that are required as a part of the regular scheduled testing. They include:
temperature, turbidity, UV absorption, ozone, total organic carbon, true color, pH, total
alkalinity, calcium hardness, nitrate, iron, manganese and aluminum. Although the test plan
suggests coliform testing, since the source is filtered and ozonated, and because the WTP
routinely tests for coliform bacteria and does not allow release of water with a coliform count,
this test was eliminated. Moreover, the presence or absence of coliform bacteria would not affect
the verification of oocyst inactivation.
During each day of the test, samples of both the feed and finished water were taken, labeled and
tested. The testing schedule for water quality parameters is shown in Table 3-1.
13
-------
14
-------
All testing was performed in accordance with the procedures and protocols established in
Standard Me thods.
Table 3-1. Testing Schedule for Water Quality Parameters
Parameter
Temperature
pH
Turbidity
Ozone
Total Alkalinity
Total Hardness
Total Organic Carbon
True color
Nitrate
Iron
Manganese
Aluminum
UV Absorption
Frequency
daily
daily
daily
daily
semi -weekly
semi -weekly
semi -weekly
semi -weekly
semi -weekly
semi -weekly
semi -weekly
semi -weekly
semi -weekly
Where tested
on- site
on- site
on- site
on- site
lab
lab
lab
lab
lab
lab
lab
lab
lab
Standard
Method
2550B
4500H+
2130B
4500 03B
2320B
2340C
5310C
Hach SM2120
4500NO3:E
3113B
3120
3120
5910
All on-site testing instrumentation and procedures were calibrated and/or standardized daily by
COA staffer agent. The on-site analyses were performed by Mannheim WTP Laboratory.
Because the test location preceded the chlorination point, tests for free and total chlorine were
eliminated; however, the possibility for residual ozone remained, so testing was performed daily
and immediately preceding challenge testing to verify the absence of ozone in the process stream
used for inoculation of challenge organisms.
Ozone, pH, temperature and turbidity were tested daily by the Mannheim WTP lab technician.
On-site instruments were calibrated daily.
The on-site bench turbidimeter is a Hach Model 2100N and was used to reference the in-line
meters. The in-line meters are Hach 1720C low range turbidimeters. Each filter has an in-line
turbidimeter which are labeled 1, 2, 3, and 4. There is also a blended water turbidimeter. During
this verification study, filters 3 and 4 supplied water to the pilot plant and thus only those in line
turbidity meters were of concern. Turbidity was measured in accordance with SM 2130, bench
top and in-line. On-line turbidimeters were compared daily to the bench turbidimeter.
There are two pH meters on site: an Orion Model 720A and a Hach EC-110. The pH probe is a
Ross combination pH electrode and was calibrated daily at pH 4 and pH 7 with National Institute
of Standards and Technology (NIST) traceable standard solutions in accordance with SM
4500-H+.
15
-------
The ozone testing was performed with a Hach Spectrophotometer, Model DR-2000 and employs
the indigo tri-sulfonate procedure. Reagents for this procedure were prepared weekly on-site.
Tests were performed in triplicate and averaged.
The ozone residual calculation used the following formula:
03 = IYABS blank x 100 mis) - (ABS sample x TV sample)
0.42 x SV
Where:
ABS sample is the absorbance of the sample at 600 nm
TV sample is the total volume of the sample = SV +10 mis.
SV = sample volume (ml) = (final weight - initial weight) x 1.0 mL/g
0.42 = slope of calibration curve at 600 nm = constant
Samples for total organic carbon and UV absorption measurements were collected in furnished
glass bottles, prepared as in SM 501 OB and shipped at 4°C by overnight express to Spectrum
Laboratories. Inorganic samples were also collected in accordance with SM 3010C and shipped
to Spectrum Laboratories overnight at a temperature of 2-8°C.
Since medium pressure lamps are broad range across the spectrum between 200 and 300 nm,
Spectrum Labs was asked to determine the UV absorption coefficient at 4 nm intervals between
200 and 300 nm.
All grab samples for water quality analyses, filter cartridges, travel blanks, and other material
sent to outside laboratories for analytical work were taken, packaged and shipped with chain of
custody forms within the same packaging.
3.3.2 Equipment Installation
The connection from the water treatment plant to the Sentinel™ UV Reactor included a Watts
RPZ backflow prevention device along with a shut off ball valve and a manually adjustable
diaphragm flow control valve. The adjustable diaphragm valve was used in conjunction with the
flow meter to regulate the flow through the system. A schematic of the Sentinel™ UV Reactor
and its installation in the plant is attached as Appendix C.
The Sentinel™ UV Reactor was installed in the lowest level of the Mannheim WTP, beneath the
filter galleries. Thirty-six inch stainless steel pipe connected the bottom drains of the filters. The
four filters are connected in pairs to this pipe, and separated by a full size automatic gate valve.
The two filters supplying water to the CCC Sentinel™ UV Reactor were filters number 3 and 4.
The thirty six inch automatic gate valve connecting the two filter banks remained closed during
the entire testing period; no water from filters 1 and 2 was introduced into the pilot plant.
16
-------
Feed water to the pilot plant was supplied through a 4" pipe connected to the 36" linking pipe. A
15 hp, single stage centrifugal booster pump was used to maintain 200 gpm across the pilot plant
and capture filters. This pump was installed prior to the RPZ and included a recirculation loop
with a diaphragm metering valve to reduce forward flow to the pilot plant between challenge test
periods.
Although the equipment was installed prior to chlorination, there was a possibility of residual
ozone. A chemical feed pump to inject sodium thiosulfate was installed and inserted into the
feed line upstream of the RPZ.
Following installation of the reactor, it was determined it was not possible to return the effluent
to the front of the plant as originally planned. The location of the pilot plant within the treatment
facility prevented such return. That meant all the water would have to be directed to the RMP.
Although the plant is intended to accept backwash and overflow water, it was not designed to
accept an additional 200 gpm for 13.3 days, the period of the verification testing (3.8 million
gallons). Flow through the system was thus reduced to 25 gpm during non-challenge periods.
Flow rates were increased to approximately 215 gpm during the challenge periods.
3.3.3 Flow Measurements
The flow meter included as a part of the UV reactor is a Promag 30 series that uses the principle
of voltage induced into a conductive fluid moving through a magnetic field. The induced voltage
is proportional to the flow velocity, which can be calibrated. The instrument has a circuit that
permits a stable zero point and is thus independent of the medium. The instrument was verified
on-site by "bucket and stopwatch."
3.3.4 Operating Parameters
Among the items to be recorded daily were measurements of the equipment's physical
performance, the pretreatment chemistry and filtration.
The conditions of the UV Reactor lamp output and irradiance were measured at two hour
intervals, 24 hours per day. Rates of flow through the UV Reactor were also recorded. In
addition, wattage measurements, cleaning cycles and the condition of the lamps was noted at two
hour intervals.
Table 3-2 presents the schedule of collection of Operating Data.
The wiping mechanism consists of a circular stainless steel brush attached to a push rod that is
activated by compressed air. On each cycle the brush travels the length of the quartz tube, pauses
for a brief and adjustable time, usually about 5-10 seconds, then returns to the start position. The
mechanism can be manually triggered by a special release valve using a pointed object, such as a
pencil. The observer can look through a UV safe viewing glass and see the brush in the rest
position. When the brush travels the length of the tube, the observer may see only a shadow
pass, and then a second shadow as the wiper returns to the rest position.
17
-------
Table 3-2. Schedule of Collection of Operating Data
Parameter Instrument Record Frequency
Flow rate Flow Meter Every two hours (must not
vary more than 10%)
UV Irradiance Instrument Every two hours.
UV Sensor In line monitor Note changes following
cleaning
Lamp Fouling/Cleaning Note frequency and time
Lamp hours Daily
Power Consumption Meter Daily
Lamp cycles Note on/off cycles
COA decided that due to the short duration of the test, a better demonstration of durability would
be accomplished by stressing the wipers. The frequency of the wiping mechanism cycle was set
for 300 seconds. Early on the first day of testing (3/31), the frequency was further shortened to
150 seconds, and then on 4/1 to 60 seconds.
3.3.5 Irradiance Measurement
The UV irradiance measurements were to be conducted via UV sensors contained within the
reactor, with one sensor at the center of each lamp. The sensors are a silicon carbide (SiC)
semiconductor that is sensitive to the range of 230-310 nm. The sensors begin to saturate at
irradiance levels in excess of 0.5 mW/cm2 so the UV must be attenuated before it reaches the
sensor. Teflon tubes, inserted into the reactor, were used to attenuate the irradiance. The UV
irradiance is thus measured as seen in the water through the quartz sleeve and through the Teflon
tube, along with the intervening water. Factors included in calculating the irradiance from the
sensors include the quartz sleeve, the spectral emission of the UV lamp, the UV absorbance
spectrum of the water, and the fiber optic cable.
By using a fiber optic probe to read each attenuation tube (and thus the individual sensors), each
of the UV sensors was compared and an attenuation factor established (see Appendix D). With
that factor, the absolute irradiance of each tube was calculated. Finally, a PLC was programmed
to convert the mA output from each UV sensor to a panel readout in mW/cm2. A full calibration
of each lamp was conducted before and after each challenge test period. Any changes in
calibration were noted on a linear time line against irradiance readings.
During the initial operations period, three of the contained system irradiance detectors failed due
to electronic problems. As a result, irradiance was then hand measured for all six lamps over the
course of the verification period. For this, two-International Light radiometers were used. The
first radiometer was model number IL1400A (serial number 2557) with UV detector probe model
number SED 240 (serial number 2813). Calibration certificates are located in Appendix E.
During April 8, 1998 of the verification period, it was observed that after the output reading
18
-------
of this radiometer stabilized, it began to drift. Accordingly, a second radiometer (model/serial
number IL1700/681) was calibrated to the first radiometer and used for the duration of the study.
Data was recorded for re-evaluation apart from this study with the three remaining system
irradiance detectors that remained functional. These sensors are not a part of the fail safe system,
which measures the power to the lamps and can be programmed to send an alarm signal, or shut
down the plant (or both) upon failure. The consequences of the loss of the system sensors are
beyond the bounds of this study; however, they do impact the future discussions regarding the
establishment of a standardized protocol for irradiance measurement.
3.3.6 Microbiological Challenge Methods
The UV reactor was challenged with live oocysts and consisted of the following steps:
1) the introduction of live oocysts into the water stream and their passage through the
CCC Sentinel™ UV Reactor,
2) the recovery of the oocysts from the water stream,
3) the determination of their viability and/or infectivity,
4) the calculation of logic inactivation.
The organisms were introduced upstream of a static mixer ahead of the reactor and collected on 1
jam filters after the reactor. The overall flow rate during the tests was about 215 gpm (814
L/min). The filters were shipped to Clancy Environmental Consultants, Inc. (CEC) in Vermont
where the organisms were isolated, concentrated and subjected to analysis by in vitro methods to
determine viability. Additionally, for C. parvum oocysts, animal infectivity experiments were
also conducted to ascertain the levels of inactivation demonstrated by in vivo assays and to
provide further evidence for the correlation between in vitro methods and neonatal mouse
infectivity, following oocyst exposure to UV light. The details of the seeding, recovery, and
viability assays are found in Clancy et al. (1998).
3.3.6.1 Description of Cryptosporidiumparvum
The C. parvum isolate used in this study was purchased from the University of Arizona and is
also referred to as the Harley Moon or Iowa strain. This strain was originally isolated from a calf
and has been maintained by passage through neonatal calves. A lot number was assigned to each
calf on the day the calf was infected and a batch number was given for the day the oocysts were
shed. These lot and batch numbers are recorded to validate oocysts' age. The oocysts excreted
in the feces of experimentally infected calves were isolated from the feces by discontinuous
sucrose gradients followed by microcentrifuge-scale cesium chloride gradients (Arrowood and
Sterling, 1987; Arrowood and Donaldson, 1996). The purified oocysts were stored at 4°C in
0.01% Tween 20 solution containing 100 U of penicillin, 100 jig of streptomycin, and 100 jig of
gentamicin per ml to retard bacterial growth.
19
-------
3.3.6.2 Enumeration of oocyst Suspensions
A known number of oocysts were purchased and their numbers were confirmed by using a
hemocytometer, according to the procedures detailed in USEPA Method 1622 (1998). The
demonstration phase consisted of trip controls, a process control, three replicates at a high UV
dose, one replicate at a medium UV dose and one replicate at a low UV dose. For the process
o
controls and UV disinfection trials 2-4 xlO oocysts were used.
3.3.6.3 Challenge Seeding Schedule
The organisms were introduced upstream of a static mixer ahead of the reactor according
to the schedule presented below. Sodium thiosulfate was injected prior to all seedings to
neutralize any ozone residual.
There was a bypass around the capture filter housing so that the Sentinel™ UV Reactor could
continue to operate while the filters were removed from the housing for shipment to the
laboratory. During the challenge testing this bypass section was physically removed to insure
that all microorganisms passed through the capture filter. The filters used in these studies
typically capture greater than 7 log oocysts; the impact of the test over the natural number of
oocysts in the RMP was minimal.
Table 3-3. Cryptosporidium parvum Challenge Seeding Schedule
Date
3/30
3/31
4/1
4/6
4/7
4/7
4/8
Run Type
Process
Control #lf
"High"
"Medium"
"High"
Process
Control #2
"High"
"Low"
Sample
Number
98090-6
98091-1
98092-1
98093-8
98094-13
98094-14
98099-7
Flow Rate
211 gpm
212 gpm
212 gpm
209 gpm}
213 gpm
213 gpm
214 gpm
Lamp
Power
HW/cm2
Off
#2—11.5
#5—9.4
#2—6.15
#5—6.5
#2—7.8
#5—7.5
Off
#2—7.25
#5—7.0
#5—3.5
UVDose
mW-s/cm2
Calculated*
0
167
69
152
0
137
20
* Calculations are shown in Appendix F.
t The sizes of the process control doses were chosen to detect up to a two log decrease in oocyst
viability caused by the process alone without UV treatment.
{ The flow declined from 209 gpm to 191 gpm during the seeding.
The spiking protocol followed that indicated in the EPA/NSF Protocol (EPA/NSF, 1998).
20
-------
The oocyst seeding protocol consisted of the following steps:
• The flow rate through the UV Reactor was adjusted to 190-215 gpm and the wipers
were turned off.
• The metering pump was energized to inject 0.025 molar sodium thiosulfate solution
into the feed water line at the rate of 100-150 mL/min at 78-80 psig.
• A sample of feedwater, downstream of sodium thiosulfate injection, was taken for
absorption analysis.
• A feedwater sample, again downstream of sodium thiosulfate injection, was taken and
immediately analyzed for ozone concentration to assure no ozone was present prior to
oocyst seeding.
• The entire effluent stream from UV reactor was diverted through a stainless steel
housing containing four each 3" diameter by 20" long 1.0 micron absolute track-etch
polycarbonate membrane filter cartridges (Nucleopore, Inc.). The surface area of each
filter was 2.8 m2 (30.14 ft2) for a total filter area of 120.5 ft2. At 200 gpm the
approach flowrate was 1.65 gpm/ft2.
• The protozoan oocyst injection utilized a 250 mL graduated cylinder into which a
suspension of 2 or 4 x 108 oocysts was placed. A Blue and White Model C-1500N
metering pump equipped with PTFE tubing injected the organisms into the feed
stream at a rate of 50 mL/min. The microorganisms were injected through a 1/4 inch
compression fitting at the inlet end of the static mixer and out through a probe
inserted to the approximate center of the mixing chamber.
• When the cylinder was « 95% empty, it was refilled with incoming feed water and the
flow rate was increased to 80 mL/min to ensure that all organisms were fed into the
Sentinel™ unit and to flush the injection system.
Upon completion of the seeding operation, flow was diverted around the filter housing, the
cartridges were removed, double bagged and shipped to CEC laboratories for processing.
3.3.6.4 Viability/Infectivity Analysis
The oocysts were isolated from the 1 jim filters, concentrated by centrifugation and divided for
analysis by two in vitro methods (fluorogenic vital dyes; DAPI and PI and in vitro excystation)
and neonatal mouse infectivity assays.
3.3.6.5 In vitro Viability Assays
The fluorogenic vital dyes assay, utilizing DAPI and PI, was performed according to the
procedures described by Campbell et al. (1992) and in vitro excystation was performed by the
procedures described by Robertson et al. (1993).
21
-------
3.3.6.6 Neonatal Mouse Infectivity Assays
3.3.6.6.1 Preparation of Infectious Doses
The doses were prepared by removing an aliquot of the enumerated suspension and diluting with
deionized water to a volume containing the target number of oocysts. These doses were again
enumerated with a hemocytometer and analyzed with the Fisher Chi-Squared index. If the
variance exceeded chance, the suspension was re-sampled and re-counted until an acceptable
variance for a minimum of five replicate counts was obtained. The hemocytometer counts were
then used to calculate the mean number of oocysts per 10 |jL dose.
3.3.6.6.2 Infectivity Assays
One process control, one high UV dose, one medium UV dose and one low UV dose were
evaluated by the neonatal mouse infectivity assays for Cryptosporidium. The UV doses (high,
medium and low) were determined from the results of the previous bench-scale study on medium
pressure lamps (Bukhari et al., 1999). No replicates were conducted due to the costliness of the
animal infectivity procedures. The procedures for mouse infectivity assays were modifications of
the original procedures described by Finch et al. (1993) and are described in Bukhari et al. (1999)
and Korich et al. (1990).
3.3.6.7 Oocyst Trip Controls, Holding Times and Temperature
The trip controls were held at 4°C throughout the study, including all travel to and from CEC and
the field site. The experimental and process control oocysts remained at 4°C until they were
vortexed for 30 minutes (still chilled), then mixed in system water (approximately 10°C, see
Table 4-1). The oocysts were then seeded into the test system and held on capture filters for the
duration of the test, for a total time of 25 minutes at the temperature of the water. After
collection on the filters, they were immediately chilled to 4°C, placed on ice packs and shipped
to CEC for elution. The ice packs were still frozen upon arrival, maintaining the temperature at
4°C. The total time the oocysts were at temperatures higher than 4°C in the field was about 2
hours. Once they arrived at CEC, each filter was individually eluted, which took approximately
about two hours per filter, so they were exposed to room temperature for this time, which
involves elution, centrifugation, washing and centrifugation. Once they were eluted and
concentrated, they were divided into two portions: one for shipment to the University of Arizona
and the other to remain at CEC. CEC then held its portion at 4°C and conducted the vital dye
and excystation the following day, which was the same day the University of Arizona was
infecting the mice. CEC held the in vitro assays until the University of Arizona began infecting
the mice so that all assays were done at the same time, eliminating another variable.
This means that CEC varied its procedures in that sample hold times and temperature to an
extent were not met. However, this was inherent in the study design. The sample hold times
were designed to be minimal, but this still meant a day for the exposure, a day to get the filters
back to CEC and eluted, and a day to get the concentrates to the University of Arizona for
infectivity, or a minimum of three days. The original excystation rates were done 1-2 days prior
22
-------
to this at CEC. These were compared with the excystation rates generated at the University of
Arizona when the oocysts were shipped to CEC. Losses of viability prior to the start of the study
were then considered. There was excellent correlation between the two labs. This means that as
many as 5-6 days elapsed from measuring the original excystation rate at CEC prior to the study
and measuring the final excystation rate at the time the mice were infected. This is the nature of
this work and cannot be avoided.
3.3.7 Logw Inactivation Calculation
Logistic analysis, as proposed by Finch, et al. (1993) was used for analyzing oocyst dose
response data. This method applies a logarithmic transformation that converts the normal dose
response data into a form that can be readily analyzed by linear regression. Linear regression
analysis yields an equation for the straight line of the type y = b + mx where b and m are the
intercept and slope of the line, respectively.
The transformation was accomplished by first defining the term response LOGIT for a given
oocyst dose as the natural logarithm (In) of the proportion of mice infected divided by one minus
the proportion of mice.
That is: response logit = ln[P/(l-P)], where P is the proportion of mice infected with a given
dose of oocysts (number of mice infected/number of mice inoculated).
The response logit values obtained experimentally were treated as the dependent (Y) variable for
regression analysis with the logic of the number of oocysts in each dose as the independent (X)
variable. A regression analysis was used to perform the least squares regression, provide the
regression equation parameters (b,m), and to test the validity of the resulting regression model
equation.
The logit dose response model proposed by Finch and analyzed here produces a linear regression
of the dose response function where the response lies between zero and 100%. Logarithmic
transformations of zero and 100% responses cannot be done and are, therefore, not used in the
logit model.
3.3.8 Health and Safety Measures
There were two major safety concerns for on-site staff with respect to this testing procedure.
1) The equipment to be tested is powered by 600 volt AC electricity and
2) The microbes to be tested are highly infectious.
Accordingly, built into the equipment were a number of safety features. Since this equipment
has been designed for installation in water treatment plants, interlock connections, breakers and
other protective devices have been included in its manufacture.
23
-------
For protection against accidental infection by oocysts, strict environmental laboratory procedures
were followed. Protective clothing such as gloves, glasses and lab coats was on hand and used
for shipment in protective containers. Handling of all live oocysts and oocyst containing
materials was done by laboratory personnel trained in biological safety.
environment of the test station was in the underdrain area of the filter plant and test staff were
exposed to the loud noise of rushing water and air scouring during backwash periods. Earplugs
The water treatment plant is located in Canada and thus not under the regulation of the
Occupational Safety and Health Administration (OSHA); however, Canadian authorities have
requirements.
3.4
The UV dose is a function of the average UV irradiance (which is determined by the lamp power)
and exposure time (which is determined by the flow rate through the reactor, and by the reactor
3.4.1 Estimation of Exposure Time
reactor, the velocity of the water through the reactor and thus the brief residence time (8.8
seconds at 200 gpm) within the reactor. The exposed volume of the reactor and the flow rate was
3.4.2
To determine the UV dose applied to achieve inactivation of a target microorganism requires a
measurement of the UV irradiance and the time the organism is exposed to the UV radiation.
used for other water disinfectants. In the case of chemical disinfectants, such as chlorine, there is
a standard procedure. The concentration of the chemical dose multiplied by the residence time
Ct value which can then be used to compute disinfection
Ct value are disinfectant demand (that of the
chemical, the blending of waters, the motion through a reactor and the mixing rates. Tracer tests
can be performed for large size systems to follow the flow paths of the chemically treated and
In the case of UV disinfection, there is no residual whatsoever; UV light is effective only when a
particle is exposed to it, thus further complicating the measurement. As a particle enters a
-------
reactor containing more than one lamp, it will receive varying illumination levels from each lamp
(in the case of the Sentinel™ equipment, several lamps). Depending on the particle's distance
from each lamp, the exposure time will depend on the specific path of the particle through the
reactor. The exposure of any specific particle to UV is difficult to determine, but certain
assumptions can safely be made, one of which assumed that every particle that passes through the
reactor with multiple sources, e.g., lamps, will be exposed to UV light.
The amount of UV light can be detected by silicon carbide detectors, and the irradiance levels
can be calibrated to a standard measurement. The illumination detected however, is only that
which is measured at a specific point, and only those photons that enter the detector on a plane
can be detected. The sensors included in the Sentinel™ Reactor were designed to measure the
irradiance at a point midway along the length of the UV lamp, and at a distance of 7.5 cm from
the lamp (at the wall of the reactor). The sensors used were saturated by the UV intensity, and
hence required an "attenuation" tube to limit the power of the UV light reaching them.
Conditions that contribute to the difficulty in measuring and thus limit the estimation of the UV
dose include (USEPA, 1996):
• The UV absorbance of the feed water may vary temporally.
• Typically, UV irradiance is measured at only one point within the reactor and only that
irradiance which enters the tube on a plane surface.
• The diode silicon carbide detectors may measure UV irradiance outside of the range of the
path length of interest.
• There is difficulty with determining with confidence the path of any single particle as it
passes through the reactor and hence the exposures to UV light.
3.4.3 Recognized Methods for Estimating UV Dose
The EPA recognizes three other means of establishing UV dose (USEPA, 1996) such as through
"bioassay", actinometry, and via calculation.
In bioassay method, microorganisms are exposed to collimated beams of UV light. Collimated
beams are those that reach the target in parallel beams and which can be produced in a
laboratory. The irradiance is measured by a radiometer and the kill rate of the microorganism,
along with a dose response, is established. The dose response can then be used to back calculate
the UV dose. This procedure is explained in detail in the ANSI/NSF Standard 55, Annex B.
Among the microbes that have been targeted are Bacillus subtilis and MS-2 phage. A standard
method using C. parvum is not final, however, the results using this method have been compared
to this verification test in a paper presented recently at the annual American Water Works
Association (AWWA) conference (Bolton et al., 1998).
Actinometry measures UV light through a photochemical reaction. A chemical sensitive to UV
light at the wavelength of interest is exposed and the resulting photochemical changes are
studied. A cell containing the chemical is inserted into the reactor and exposed to the UV light.
The chemical change produces a chemical product at a known rate and concentration relative to
25
-------
the irradiance that can be used in part to calibrate photometers. This method is of value in
calibrating radiometer readings. The measurement of UV irradiance by this method was
described in a Section 3.3.5. As described in a later section of this report, there were difficulties
in the direct measurement of UV irradiance using this method.
The third method cited by EPA is via calculation of the average irradiance across the reactor. It
is this third method that was employed in this verification test, through a Multiple Point Source
Summation (MPSS) model derived by Dr. Bolton. The model was then used to establish the UV
doses during the challenge periods of the verification testing.
3.4.4 Multiple Point Source Summation Model for Estimating VV Dose
Following is a discussion of the model employed in this verification test which is similar to the
MPSS method. This discussion is directed toward the layperson, is non-mathematical, and hence
limited in scope, and is intended only as an introduction to some of the important concepts. For a
more comprehensive explanation of the model, the reader is urged to consult Dr. Bolton's
original paper, which is in press and will be available in early 1999.
Most UV light sources are long narrow lamps. The light output from such a lamp may be
approximated by a large number (n) of "point sources" equally spaced along the lamp axis. The
light from each point source is assumed to radiate equally in all directions and the irradiance
across a small volume element in the reactor is then obtained by summing the irradiance at that
volume element from all n point sources.
3.4.4.1 Model Assumptions
This MPSS model makes the assumption that the total illumination of a tube type lamp consists
of a series of illuminating points along a line. It is further assumed that the light emanates
radially, and in all directions from each point and that the total energy delivered is the sum of all
the smaller, "point" sources. For a reactor that has several lamps then, additional summations are
required for the illumination along the axis of the reactor as well, of course thereby complicating
the formula. Add to that the notion that the light source is circular in cross section, and that a
second and third lamp are illuminated in the same reactor, and the resulting determination of total
irradiance at any point, or even average point, in the reactor is complex.
3.4.4.2 Model Adjustments for Factors Affecting UV Dose
The establishment of UV dose then must take into account the power of the lamp, the irradiance
(which is the radiant power over an area) along with the factors that affect the passage through
from the lamp to the point of interest or to the particle. These factors include the lamp glass
itself, the air between the lamp and a quartz sleeve that protects the lamp, the quartz sleeve and
its interface with both the air on one side and the water on the other, and the water itself. At each
of the interfaces are reflections and refractions which are well described in optical physics. Add
to this the characteristics of absorbance of the water and the calculation is very sophisticated.
26
-------
The model employed here also describes a "volume of influence" for each lamp, and then
calculates an average irradiance for several of the transmittance percentages (%T) within that
volume. The irradiance values outside of that volume are estimated from the increasing distance
from the lamp, and by assuming an average irradiance from the lamp. When the percentage
transmittance (%T) is less than 65%, the contribution outside the volume of influence can safely
be disregarded.
For this verification test only two lamps were illuminated during the challenge periods. The
volume of influence in each reactor was 12.32 liters, thus each reactor had a "dark" volume of
43.8 liters. This was considered in the calculations.
When all of these considerations are taken together, a complex formula can be derived that
allows a calculation, however tedious. These calculations are made for each of the challenge
periods, for each water condition and lamp power, and are shown in Appendix F. A more
detailed summary of the MPSS equations and assumptions of the model are presented in
Appendix D.
3.5 Data Management
Data were collected in a bound logbook and on charts from the instrumentation panels and
individual testing instruments. The visitors log, on-site testing data and turbidity meter readings
were contained in pre-printed, bound pages. Preprinting the form served as a prompt for readings
and allowed for distribution of tasks.
There is a single master logbook containing all on-site operating data which remained on site and
contained instrument readings, on-site analyses and any comments concerning the test run with
respect to either the nature of the feedwater or the operation of the equipment. This master
logbook contained flow, irradiance, notes on the challenge seedings and equipment. Data were
entered into a computer spreadsheet program on a daily basis from the logbook and from all
analytical reports. A back-up copy of the log book and computer data was maintained off site.
This log is consistent with standard laboratory practice.
All details affecting the operation of the equipment, whether by COA staff, laboratory staffer by
CCC staff were logged in the experimental logbook, consolidated and entered into computer
spreadsheets. The computer spreadsheets follow as Appendix G.
Each page of the logbook was sequentially numbered and signed by the on duty COA staff.
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 the water conditions, were noted
in the logbook. Copies of the logbook are attached as Appendix H; copies of chain of custody
forms are included as Appendix I. Original chain of custody forms traveled with the samples.
Although data were collected at four locations: the test site, CEC laboratories, the University of
Arizona and Spectrum Labs, the COA offices were the central data collection point and all raw
data and notes are on file.
27
-------
Chapter 4
Results & Discussion
4.1 Feed Water Quality
4.1.1 On-Site Test Results
Table 4-1 presents the listing of on-site testing for the verification period.
Table 4-1. On-Site Testing for the Verification Period
Date
Temp.1
pH
Ozone
Turbidity
Turbidity
3/30*
3/31*
4/1*
4/2
4/3
4/4
4/5
4/6*
4/7*
4/8*
4/9
4/10
4/11
4/12
4/13
9.9/10.6
10.6/11.7
11/12.3
10.8/11.9
11.0/12.4
10.9/11.9
11/11.9
11/11.9
10.3/11.3
8.8/8.8
10.1/11.0
10.0/11.0
10.2/11.2
10.2/11.3
10.6/11.6
7.76/7.71
7.71/7.68
7.68/7.66
7.63/7.61
7.58/7.55
7.60/7.60
7.54/7.55
7.59/7.58
7.60/7.61
7.50/7.50
7.47/7.46
7.47/7.47
7.40/7.40
7.45/7.47
7.62/7.65
0.03/0.05
0.03/0.04
0.03/0.05
0.03/0.05
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0.09/0.09
0.093/0.109
0.096/0.106
0.065/0.064
0.088/0.101
0.080/0.081
0.056/0.053
0.098/0.092
0.068/0.053
0.094/0.093
0.087/0.077
0.136/0.132
0.132/0.123
0.098/0.098
0.147/0.134
0.064/0.058
0.101/0.110
0.077/0.076
0.054/0.046
0.068/0.050
0.049/0.074
0.043/0.043
0.088/0.047
0.041/0.041
0.096/0.093
0.064/0.095
0.067/0.147
0.100/0.088
0.052/0.074
0.112/0.114
Notes:
* Challenge test days.
1 Temperature in °C from influent/effluent of reactor.
2 Ozone in mg/L from influent/effluent of reactor.
3 Turbidity in NTU from bench influent/effluent of reactor.
4 Turbidity in NTU from on-line turbidimeter, filter 3/filter 4.
4.1.2 Laboratory Test Results
Table 4-2 presents parameters measured at Spectrum Laboratories.
4.1.3 UV254Scan
A representative scan and graph of the UV absorption coefficient at 4 nm intervals between 200
and 300 nm is included as Appendix J. This data was then used by the MPSS model in the
calculation of UV Dose.
28
-------
Table 4-2. Parameters Measured by Spectrum Laboratories
Date
3-311
3-3 IE
4-021
4-02E
4-081
4-08E
4-101
4-10E
4-131
4-13E
Alk
mg/L
180
180
150
160
160
150
160
160
170
170
Al
mg/L
0.86
0.26
0.20
0.48
0.08
0.08
0.10
0.23
0.06
0.45
Color
TCU
5
5
5
5
5
5
5
5
5
5
T.Hard
mg/L
260
250
210
210
220
220
230
230
250
250
Iron
mg/L
0.1
ND
0.5
ND
ND
ND
ND
ND
ND
ND
Mang
mg/L
0.03
ND
ND
0.01
ND
ND
ND
0.01
ND
0.02
N03
mg/L
4.0
3.7
3.7
3.1
3.0
3.0
3.4
3.3
3.8
3.6
TOC
mg/L
4.5
4.3
4.0
3.8
3.5
3.5
3.4
3.2
3.7
3.4
UV254
cm"1
0.0367
0.0383
0.0519
0.0214
0.0551
0.0427
0.0517
0.0423
0.0365
0.0385
I = Influent; E = Effluent
4.2 UV Dose Calculation
4.2.1 Flow Rate Studies
The flow meter on-site was verified by "bucket and stopwatch". Following is the calculation
performed. The flowmeter registered 256.8 gpm at the beginning and 257.2 gpm following for
an average of 257 gpm. The entire volume was directed into a 36 inch diameter tank and the
time to fill a 36 inch rise was measured.
The measured tank volume was 158.62 gallons, and it filled to that level in 36.24 seconds.
158.62/36.24 x 60 = 262.62 gpm.
The error from flowmeter to measured flow is +2.1%
4.2.2 Exposure Time
The following sections describe procedures for estimating exposure time and the model used to
calculate the UV Dose. The reactor geometry has been described by CCC as follows:
2 identical cylindrical sections at top and bottom Va= 14.673 L
2 "flattened" cylindrical sections where the UV
lamps are mounted Vb= 27.378 L
1 cylindrical section in the middle Vc= 29.346 L
less the volume of the 6 quartz sleeves Vq= 0.342L
2Va + 2Vb +VC -6Vq = Vt = 111.396 L/ 3.785 =29.43 gallons
29
-------
An estimate of exposure time was calculated from the flow rate and the volume of the reactor
based on its geometry:
Exposure time (min) = reactor volume (gal)/flow rate (gpm)
Since the lamps illuminate the entire reactor (both chambers), the exposure time is equal to the
residence time within the reactor itself.
At 200 gpm, the residence time is 8.829 seconds based on the flow rate and the geometry of the
reactor. Each half chamber contains 14.71 gallons and the exposure time during those test
periods when only one lamp was illuminated was 4.42 seconds.
The fluid dynamics of the system are as follows: at 200 gpm the velocity through the 4 inch
static mixer was « 5.1 feet per second. The water enters the first chamber at the bottom in a pipe
tee 1-7/8 inches off the bottom of the chamber. This can be represented as the entrance of a pipe
into a larger pipe or a tank at a right angle. The chamber is 10 inches in diameter, thus flow is
turned upward through hydraulic resistance across and from the bottom of the chamber. The turn
introduces turbulence. In the passage through the first chamber, the perpendicularly placed UV
lamp sleeves introduce additional turbulence; at the midpoint and between the two chambers a
separation plate narrows the flow to 4 inches in diameter through an orifice plate. Since all the
water must pass through this orifice, additional turbulence is introduced as the water velocity
increases, along with a pressure drop at the exit of the plate. The velocity of the water through
the 10 inch diameter reactor at 200 gpm was .82 feet/sec.
The turbulent flow introduces a pressure drop across the system of approximately 16 psi at 200
gpm. To some degree a higher pressure loss across the reactor represents more turbulent flow.
Measuring the pressure drops at several different flow rates allows for a prediction of turbulence.
4.2.3 Multiple Point Summation Model Results
The challenge testing was originally intended to produce UV doses of 400, 200, 100 and 0
mW-s/cm2 (the last as a process control). Following initial lab testing (Bukhari et al., 1999) for
UV254 and model calculation, those doses were adjusted downward to 200, 100, 50, and 0
mW-s/cm2. To provide variable UV dose at the same flow, the theoretical challenge illumination
for different numbers of lamps and at different power levels was calculated by the model. The
theoretical doses are shown in Table 4-3.
The target doses for the demonstration study were selected as high (300 mW-s/cm2), medium
(100 mW-s/cm ) and low (50 mW-s/cm ) UV doses, delivered with medium-pressure lamps.
However, as the absorbance of the water affects the actual UV doses that are delivered to the
target organisms, re-calculation of the actual doses by using the MPSS model (see Sections 3.4
and 4.2) indicated that the actual UV doses that oocysts were exposed to were considerably lower
than anticipated. Final doses as calculated by the model were 167, 152, 137, 69, and 20 mW-
s/cm2 (see Appendix F for the details of the MPSS model calculation).
30
-------
The MPSS modeling which relates the above factors via a point source summation compares the
irradiance measured at the sensor to the average irradiance of the entire reactor. The calculations
are included as Appendix F. Correlation of radiometer readings to the MPSS model is discussed
in Appendix D.
Although the radiometer readings were correlated to the MPSS model results, the readings were
limited by the use of filters that attenuated the readings. Furthermore, the model UV dose
estimates require validation through another method such as the bioassay method (see Section
3.4.3 for more discussion). Comparison of the UV dose estimated in this study by the MPSS
method to the UV dose measured or estimated by another method is not recommended.
Table 4-3. UV Dose in mW-s/cm2 as Calculated by the MPSS Model*
Number of Lamps on Full Power Reduced Power
6
5
4
3
2
1
814
678
543
407
271
136
280
233
187
140
93.2
46.8
*Note: At the time these calculations were performed, the absorbance of the water was
relatively low (before spring run-off). When the actual tests were run, the absorbance of the
water was higher and thus the delivered UV doses were lower.
4.2.4 UV Irradiance by Direct Measurement
During the testing and verification period (on 4/7/98) the treatment plant changed the
pretreatment coagulant from poly aluminum chloride to aluminum sulfate as part of a routine
springtime procedure. This change resulted in difficulties in maintaining plant turbidity levels
below the goal of 0.1 NTU. The change, while it had no noticeable impact on the challenge
periods did result in a diminution of the irradiance measurements during the latter stage of the
verification period.
At the time it was thought that the diminution might be due to either increased turbidity or to a
reduction in lamp irradiance caused by aging or some other problem. When the treatment plant
stabilized and the turbidity levels dropped, the irradiance levels rose slightly, suggesting that the
water quality was in part a cause of the reduction. The quartz tubes were not noticeably fouled,
and absorption coefficients were not appreciably different. The attenuation tubes had bleached
out (from UV light) and that, along with a slow degradation in the irradiance due to lamp aging
and a change in water quality and absorbance, may have contributed to the loss.
The average irradiance, as shown in Figure 4-1, shows a decline over time.
31
-------
OO CO CO CO CO CO CO CO CO CO CO CO CO CO CO CO
en en en en en en en en en en en en en en en en
READINGS EVERY 2 HOURS
Figure 4-1. Average of UV Irradiance Over Time
0.25
0.00
oooooooooooooooooooooooooooo
CO CO
READINGS EVERY 2 HOURS
-TURBIDITY UV IRRADIANCE
Figure 4-2. Turbidity vs. UV Irradiance
32
-------
Turbidity alone does not correlate well with irradiance. Turbidity spikes appear to be related to
filter backwash periods. The overall turbidity of the plant increased following the change in
pretreatment chemistry on April 7 and following that change the filters were backwashed several
times a day during the restoration period.
4.3 Microbiological Challenge Results
4.3.1 In vitro Viability Assays
Viability data obtained with vital dyes and in vitro excystation have been summarized in Table
4-4. In each assay, 100 oocysts were visualized in duplicate. For comparative purposes the
process control viability values were normalized to 100% and the respective viability values for
all remaining test samples were adjusted according to the normalization factor. These data have
been presented in Table 4-5.
Table 4-4. Effects of UV Exposure on Cryptosporidium parvum Viability
Cryptosporidium parvum viability (n=3)
Lab ID
number
98090-7
98090-6
98091-2
98091-1
98092-2
98092-1
98093-9
98093-8
98094-13
98094-14
98099-8
98099-7
Pilot testing
parameters
Trip control 1
Process control 1
Trip control 2
UV dose (High)
Trip control 3
UV dose (Medium)
Trip control 4
UV dose (High)
Process control 2
UV dose (High)
Trip control 5
UV dose (Low)
Vital dyes
82.0 ±4.0
71.7 ± 15.4
61.7±4.9*
90.4 ± 1.2
4.9 ±4.5
88.7 ± 1.0
73.6 ±4.1
76.6 ±4.4
8.5 ± 1.4
45. 3 ± 16.8
25 ±8. 9
79.7 ±5.1
75.3 ±6.8
Excystation
81.4±8.1
29.7± 1.1
45 ±7*
79.8 ±4.3
13.7±3.1
56 ±9.8
89 ±2.7*
72.7 ±2.3
80.4 ± 1.7
12.1 ± 1.5
49.3 ±4.0
34.3 ±4.0
77.3 ±4.2
47 ±7.8
72.7 ± 1.5*
* Repeat analysis on an aliquot taken from same concentrate.
4.3.2 Neonatal Mouse Infectivity Assays
Mouse infectivity assays with high and medium UV doses demonstrated no infection in neonatal
mice despite oral inoculation of up to IxlO5 oocysts (Table 4-6). The oocysts which had been
exposed to a low UV dose resulted in 4.5% infection (1 of 22 mice) with an inoculum of IxlO5
UV exposed oocysts per mouse; however, no mice were infected when inocula of either IxlO4 or
IxlO3 UV exposed oocysts were administered into a total of 36 mice (Table 4-6).
33
-------
Table 4-5. Normalized (Percent) oocyst/cyst Inactivation Using the
,TM
Sentinel System and in vitro Viability Assays
Cryptosporidium parvum viability
UV Dose (mW-s/cnO
Process control
20
69
137
152
167
Vital Dyes
100
104.7 ±9.5
102.3± 5.7
34.8± 12.4
11. 8± 1.9
6.8 ±6.3
Excystation
100
160 ±3.3
160±5.1
75. 5 ±8. 8
26.6 ±3.3
30.1 ±6.8
Note: values over 100% should be considered to be 100%.
Table 4-6. Effects of UV Exposure on Cryptosporidium parvum oocyst Infectivity
Lab ID
number
98090-7
98090-6
98091-1
98092-1
98099-7
Pilot scale testing
parameters
Trip control 1
Oocysts per mouse
# of total mice
# of mice infected
% infectivity
Process control 1
Oocysts per mouse
# of total mice
# of mice infected
% infectivity
UV dose (High)
Oocysts per mouse
# of total mice
# of mice infected
% infectivity
UV dose (Medium)
Oocysts per mouse
# of total mice
# of mice infected
% infectivity
UV dose (Low)
Oocysts per mouse
# of total mice
# of mice infected
% infectivity
Cryptosporidium parvum infectivity
Inoculum 1 Inoculum 2 Inoculum 3
25
38
2
5.3%
50
25
11
44%
1000
24
0
0%
1000
22
0
0%
1000
18
0
0%
75
40
14
35%
500
20
20
100%
10,000
12|
0
0%
10,000
26
0
0%
10,000
18
0
0%
150
27
15
55.6%
5000
23
23
100%
100,000
24
0
0%
100,000
25
0
0%
100,000
22
1
4.5%
fData from one of two cages of 12 mice was invalid due to laboratory accident.
34
-------
4.3.3 Dose Response Infectivity Calculations
The assay procedure and "Logit Dose Response Calculation" are presented in detail in Section
3.3.7. The initial analysis of the dose response data yielded the following equation which
describes the dose response of fresh oocysts in neonatal CD-I mice:
Response Logit = -6.752 + 3.611LogioDose
This equation enables the estimation of the number of infective oocysts in a dose administered to
the mice based on the proportion of mice infected and eliminates the need to construct dose-
response curves for each level of disinfection to be studied.
The linear regression equation of the model (e.g. response logit = -6.752 + 3.61 llogiodose) is a
mathematical representation of a straight line of the type y = b + mx. Here, response logit (y) is
calculated from the expression ln(P/l-P) where P is the proportion of animals infected. The
intercept (b,-6.752) and the slope (m,3.611) are provided by linear regression analysis whereas
the dose of fresh oocysts (x) is a known quantity. When the proportion of infected animals is
known for a given dose of oocysts, solving for the number of infectious organisms in a given
dose of oocysts becomes quite easy. Determining the number of infective oocysts in a given dose
involves calculating the response logit [ln(P/l-P)] from the proportion of infected animals and
solving the model equation for logio dose. Taking the antilog provides the number of infectious
oocysts in the administered dose. For example, of 20 mice administered 300 oocysts, 14 became
infected and 6 did not. To determine the number of infective oocysts administered to the mice
and the log reduction in oocyst infectivity use the model: [Response logit = -6.752 + 3.611 Logio
Dose] as illustrated below.
1. Determine the proportion of mice infected: 14/20 = 0.7
2. Calculate the response logit: In [P/(1-P)] = In [0.77(1-0.7)] = In 2.333 = 0.8473
3. Substitute into the model: 0.8473 = -6.752 + 3.611 Logio Dose
4. Solve for Log dose: Log Dose = [(0.8473 + 6.752)73.611] = 2.1045
5. Find the antilog: antilog 2.1045 = 127 This is the number of infective oocysts present in the
dose of 300 oocysts administered to the 20 mice.
The final infectivity logio inactivation calculations are based on the LOGIT dose response model:
Response LOGIT = -6.752 + 3.611 LogioDose
Where:
Response LOGIT = LN [proportion of mice infected + (1 - proportion of mice infected)]
The logio inactivation of C. parvum upon exposure to UV is given by the following calculations
for treated oocysts (low dose).
Proportion of mice infected = 0.045; 97,500 oocysts administered.
Response LOGIT = LN[0.045 7 (1-0.045)] = -3.055
35
-------
Substituting into the model equation yields the number of infective oocysts in the administered
dose:
-3.055 = -6.752 + 3.611 LogioDose
Logi0Dose = 3.697 73.611 = 1.024
Calculated Infective Dose = 10.6 oocysts
Log Change of Infectivity = Log(infective dose / administered dose)
Log Decrease of Infectivity = Log(10.6 / 97,500) = -4.0 Log Change (decrease).
Thus, the low dose has a 4 log decrease in oocyst infectivity.
Response Logit = -6.752 + 3.611 Log10 Dose
4
3 -
2 -
1
0
-1
-2
-3 -
-4
o Infectivity Data
2 -r
Regression Line (r =0.70)
90% Prediction Limits
90% Confidence Limits
-------
Nonetheless measuring oocyst inactivation at the high, medium and low UV doses indicated that
the in vitro methods demonstrated marginal reductions in oocyst viability when compared to the
process control. In a comparison between the vital dyes and in vitro excystation, data indicated
that the vital dyes assay provided a greater decline in oocyst viability than in vitro excystation.
For example, a UV dose of 152 mW-s/cm resulted in a normalized viability value of 11.8%,
which decreased to 6.8% as the UV dose was increased to 167 mW-s/cm2. However, at these
respective UV doses, in vitro excystation indicated a normalized viability value of 26.6%, which
increased to 30.1%.
Table 4-7. Log Inactivation of Cryptosporidium parvum Following Exposure
to UV Light: Comparison of in vitro Methods with Mouse Infectivity Assays
Cryptosporidium parvum
UV Dose (mW-
s/cm2)
Process control
(NoUV)
20 (Low)
69 (Medium)
137 (High)
152 (High)
167 (High)
Lab ID
Number
98090-6
98099-7
98092-1
98094-14
98093-8
98091-1
Vital dyes assay
(D API/PI)
0
0
0
0.5
0.9
1.2
In vitro
excystation
0
0
0
0.2
0.4
0.4
Mouse
infectivity
0
3.9
>4
Not Done
Not Done
>4
The demonstration study samples contained large amounts of contaminating debris/particulates
that were derived from filtration of large volumes of finished water. It is well documented that in
vitro excystation has limited application in environmental samples, due to the ease with which
oocysts can be masked by contaminating debris. Masking of oocysts clearly creates a scenario
where increased errors are likely to occur in ascertaining oocyst population viability. This may
be a possible reason for greater variability with in vitro excystation than with the vital dyes.
Despite this, these data indicated that both in vitro assays failed to demonstrate in excess of 1.2
log inactivation at the three UV doses that were examined. However, when aliquots of oocysts
derived from the same concentrates were subjected to mouse infectivity assays, > 4 log
inactivation was noted (Figure 4-4) with oocysts that were exposed to the high and medium UV
doses, whereas 3.9 log inactivation occurred with the oocysts which were exposed to the low UV
dose (Table 4-7).
These results demonstrate that in vitro methods for C. parvum oocysts (vital dyes or in vitro
excystation) significantly under-estimate oocyst inactivation of UV exposed oocysts when
compared to the mouse infectivity assays. Previously, it has been suggested that the sensitivity of
the mouse infectivity assays was attributable to their capacity for enabling analysis of large
numbers of oocysts than was feasible with the in vitro assays. However, data from recent studies
suggest that in vitro methods may be sensitive to the mechanisms of oocyst inactivation and in
turn less reliable than mouse infectivity assays for determining oocyst inactivation (Clancy et al.,
37
-------
1998). Belosevic et al. (1997) noted previously that excystation did not correlate with animal
infectivity using this same CD-I mouse model.
5.00 -
4.50 -
4.00 -
c 3.50 -
o
is 3.00 -
>
I 2.50
I 2'°°
-1 1.50 -
1.00
0.50 -
(
J t
/
/
/ o - — -^
ANIMAL INFECTIVITY
— •— EXCYSTATION
—A— VITAL DYE
5 25 50 75 100 125 150 175
UV Dose mW-s/cm2
| Note: UV dose calculated at 20, 69, 137, 152, and 167 |
Figure 4-4. Cryptosporidiumparvum Log Inactivation Ratio vs. UV Dose
A recent study comparing four viability assays (DAPI/PI, excystation, cell culture, and animal
infectivity) showed a 4.6 log inactivation of C. parvum using the animal infectivity model
(Slifko, 1998). The results of the cell culture most probable number (MPN) analysis showed a
4.07 log inactivation of C. parvum.
4.3.5 Affect ofoocyst Trip Controls, Holding Time and Temperature on Results
The critical excystation rate, expected to be 80% or higher (as noted in the test procedures), is
that of the oocysts when assayed prior to delivery into the field. The excystation rate of the trip
and process controls after they returned from the field varied, and may be (and were) lower than
80% due to the rigors of the travel (trip control) and experimentation (process control).
However, this does not mean that the testing is invalid because the excystation rates were lower
than 80%. For example, if the excystation rates for trip or process controls were extremely low
(<50% viable when they began at 80+%), even normalizing the data would present problems. A
50% excystation rate means that the oocyst preparation were half dead (consequently half were
alive). The excystation rates in this study were all within the 80% range (when considering
standard deviation (SD)) except for the Sample 98094-13 Process Control 2 which was 49.3%
(Table 4-4).
38
-------
The effect of the process control on the log decrease is as follows:
Proportion of mice infected = 0.44; 51.9 oocysts administered.
Response LOGIT = LN[0.44/(1-0.44)] = -0.241
-0.241 = -6.752 + 3.611 LogioDose
LogioDose = 6.511 73.611 = 1.803
Infective Dose = 63.5 oocysts
Log Change of Infectivity = Log(63.5 / 51.9) = 0.09 Log Change (increase).
The effect of the trip control on the log decrease in oocyst infectivity is as follows:
Proportion of mice infected = 0.55; 158 oocysts administered.
Response LOGIT = LN[0.55/(1-0.55)] = 0.201
0.201 = -6.752 + 3.611 LogioDose
LogioDose = 6.953 73.611 = 1.925
Infective Dose = 84.2 oocysts
Log Change of Infectivity = Log(84.2 / 158) = -0.27 Log Change (decrease).
The overall effect of process and trip controls is negligible on the log decrease in oocyst
infectivity. Hence, the log inactivation was assumed at the low dose to be 3.9 (-4.0 + 0.09 =
-3.9). The trip control was not considered in the calculation.
4.4 Operations and Maintenance
To illustrate the application of the system for small drinking water systems, studies of operating
conditions and power consumption were measured and evaluated. These records and a
recapitulation of all measured operating parameters in the log are included on the spread sheets
as Appendix G.
4.4.1 Ease of Operation
The duration of the test was too brief to make a full determination of the ease of operations and
maintenance. Operator involvement would be related to the size of the community served and
the reporting requirements, and to the degree of automation afforded the treatment plant. Prior
studies have suggested that one of the advantages of UV disinfection is its relative ease of
operation (USEPA, 1996).
The Sentinel™ UV System can contain a high level of sophisticated controls, suitable for
attachment to Supervisory Control and Data Acquisition (SCADA) or other remote monitoring
systems. It also has on-board fail safe controls that can be programmed to sound an alarm, shut
down the system or both upon lamp failure. Additional controls to illuminate a back-up lamp
can be added if desired.
39
-------
Even without automated controls however, operator time is minimal: it will take approximately
ten minutes per day (or per shift if required) to check the operation of the lamps, wipers, observe
and note irradiance measurements, power consumption, lamp turn-on frequency, gallons
processed and flow rate, all of which are automatically recorded and LED displayed by the
system. Operator time may be somewhat longer if chemical testing is required (for example,
parameters such as turbidity, pH and alkalinity). Continuous chart monitoring can readily be
attached if necessary. The UV chamber has viewing ports to observe lamps and wipers.
Major repairs to the system, including the power supply and instrumentation, would necessarily
be performed by qualified service technicians.
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 and monitored 24 hours a day until the
completion of a period of 27 days. During this time few problems were noted apart from those
already enumerated above.
During the continuing operations stage following the challenges, the reactor operated with little
intervention. It was noted at one time that the quartz cleaning brush was stopping in mid-travel,
but that was an aberration which corrected itself. As the operators grew more confident in their
understanding of the system, small anomalies became understood and accepted. The inspections
of the lamps, after seeing the same patterns for several days in a row, were less of a concern;
shadows and opaque areas were identified as normal and not as scaling intrusions or failures.
4.4.2 UV Sensor
Three of the contained irradiance sensors failed due to unexpected electronics problems.
Consequently, the irradiance was hand measured with the portable radiometer at all six lamps
during the entire verification test, and compared to the three in line irradiance instruments. The
in-line data were not used in the UV Dose calculations but were registered to accommodate
corrections to the on-board sensors.
4.4.3 Lamp Fouling/Cleaning
During the operations performance period, and more critically during the verification period, the
automatic quartz sleeve wipers ceased operating for many reasons including a broken weld.
Wiping was initially set at 15 minute cycles, and then increased to five minute cycles and then to
one minute cycles (see Table 4-8). This frequency may have contributed to breakage. Following
the wiper repair the frequency of cleaning was reduced from one minute, then to 150 seconds,
and then to 500 seconds for the duration of the verification period.
On 4/2 a brush disconnected from the wiper rod. On 4/3, during the repair of the brush, a part of
the mechanism fell on and cracked the quartz tube of lamp #1. (The lamps are numbered 1
through 6 from the bottom up, to correspond to the flow path of water.) Also on 4/3, the wiper
40
-------
for lamp 5 stuck in the extended position. On 4/4 the #5 wiper was repaired, and the wipers on
#6 and #3 failed. On 4/5 the #3 and #6 wipers were removed and repaired.
Table 4-8. Wiper Cycle Frequency
Date
3-30
3-31
3-31
4-1
4-2
4-3
4-4
4-5
4-6
4-7
4-8
4-9
4-10
4-11
4-12
4-13
Frequency
900 seconds
900 seconds
300 seconds
150 seconds
60 seconds
60 seconds
wiper failure
60 seconds
60 seconds
150 seconds
150 seconds
500 seconds
500 seconds
500 seconds
500 seconds
500 seconds
On 4/6 the wiper frequency changed to 150 seconds. On 4/7, the #5 wiper again failed and was
replaced on 4/8. On 4/8 the frequency was changed to 500 seconds for the duration of the study.
Following the last repair, no additional failure occurred. Later, wipers were noted to stop along
the lamp, but they self corrected upon activation of the next cycle.
The wiper mechanism is being redesigned and will be the subject of a separate ETV evaluation.
CCC determined that the cause of wiper failure was the impact force of the brush with the wiper
stop at the end of the extended travel position.
Because the water flow was reduced during non-challenge periods to a rate far below the design
flowrate (to 25 gpm instead of the design flow of 500 gpm), there was a greater likelihood of
deposits on the quartz tubes (the lamps will have run hotter). During a reevaluation study it
would be instructive to reduce the rate of wiping frequency until deposits appear on the quartz
tubes, and then increase frequency until deposits do not form as a means of establishing wiper
frequency under those conditions.
41
-------
4.4.4 Lamp Hours and Power Consumption
Power was measured by a utility kilowatt-meter. The meter was installed on the power line that
supplied the test unit on 3/31. The Valhalla Digital power meter originally planned for the test
was not available.
The life of the lamps is guaranteed by CCC at 3,000 hours. Lamp life will vary with local
conditions. Lamp life can degrade by fouling or by power losses. There is some depreciation as
the lamps age as well.
Figure 4-5 below shows power consumption during the testing period. Because during the
challenge period only at most two lamps were on, power consumption is not constant over time.
During the maintenance period the power consumption was approximately 1.046+0.046 kW per
lamp. Assuming daily operation of six lamps at full power, the power demand is estimated at
150.6 kW per day.
200
180
160
>- 140
Q 120
LU
a.
(0
K
I
100
80
60
40
20
0
\
\
No Meter
oo
O)
O
£2
CO
oo
O)
DATE
Figure 4-5. Power Used During Testing
Table 4-9 presents kW-h/1000 gal during challenge periods relating to UV dose, and the power
consumption (as measured) during the balance of the period (regardless of flowrate).
Following the microorganism challenge periods, from midnight on 4/9/98 and through 2:00 am
4/14/98 (122 hours), the equipment was run continuously at 25 gpm with all six lamps on full
power. During this period 840 kWh were used or 6.88 kW/hour.
42
-------
Table 4-9. Power Consumed during Challenge Test Periods C. parvum
Date
3/31
3/31
4/6
4/7
4/8
Lamps
2&5
High
2&5
Reduced
2&5
High
2&5
High
5
Reduced
KWh*
2.0845
0.984
2.095
2.109
0.552
gpm
212
214
200.4
213
214
Calc.
mW-
s/cm2
167
69
152
159
20
kW per
1000
gallons
0.164
0.077
0.174
0.165
0.043
* - KWh was determined by multiplying voltage and amperage by 1000 for one hour (KVA)
Voltage and amperage readings, except during lamp off times, were stable throughout the period
(Figures 4-6 and 4-7). Increased power demand following a reduced power period suggests that
initial illumination requires a higher level of power.
300
250
200
O
o
CHALLENGE TESTING
(Lamps 1,3,4 & 0 Turned Off,
150
100
oo
g>
o
CO
oo
o>
CO
CO
oo
O>
oo
O>
oo
O>
oo
O>
oo
o>
CO
?r
oo
o>
5
•^r
oo
o>
5
•^r
oo
CT>
£?
^
oo
O)
oo
O5
^
?r
oo
>
oo
>
oo
>
oo
>
oo
o>
oo
o>
oo
o>
oo
en
DATE
Figure 4-6. Voltage Used During Testing Period
43
-------
(Lamps l ,3,4 & b I urned ortl
CHALLENGE TESTING
(Lamps 1,3,4 & 6 Turned Off
DATE
Figure 4-7. Amperage Used During Testing Period
4.4.5 O&M Manual
The Operations and Maintenance (O&M) manual supplied by the manufacturer was specific to
this equipment and included all the components of the pilot plant. Drawings and illustrations
showing the positions of the meters and controls are included along with explanations of control
functions and step-by-step instructions for common maintenance functions, such as: replacement
of lamps, quartz tube cleaning and reactor cleaning. Complete instructions for equipment start-
up and shut-down procedures were listed in this guide. The control panel is thoroughly
explained so that all programmable functions, including wiper cycles, lamp set-points for alarms
and other PLC parameters are easily learned by even inexperienced personnel.
Safety measures included detailed instructions concerning high voltage, protection against UV
irradiance, and the procedures for mercury spills in the event of lamp breakage. A trouble
shooting guide was furnished that included the following potential problems.
• No power
• Emergency stop button engaged
• Transmittance controller not working
• Faulty solenoid
44
-------
Low air pressure
Jammed cylinder
Water leaks
UV lamp not operating
High temperature
Low or no amp or voltmeter reading
45
-------
Chapter 5
Limitations
With every drinking water treatment product evaluation, there are often limitations in the
assessment methodology, the conditions of testing and the technology itself. The lack of
consensus on evaluation methods and procedures or the application of a technology is a reflection
of the uncertainties associated with emerging technologies, developing analytical techniques and
engineering applications. The resolution of these uncertainties is within the purview of rigorous
scientific research and not the ETV program. Rather the ETV Package Drinking Water
Treatment Systems pilot verifies the performance of innovative water treatment systems using
consensus methods and procedures. The following section describes the limitations of the
methods and procedures that were followed in the verification of the performance of the CCC
TA/f
Sentinel Ultraviolet radiation technology.
5.1 Method for Determining Viability and Inactivation
There is a need in the water treatment and microbiological communities to establish a consensus
for determining C. parvum viability/infectivity and inactivation. While methodologies for
determining viability of bacteria and other microbes have general acceptance, there is no
consensus in the scientific community on C. parvum viability/infectivity methods. Research has
not yet proven that infectivity in mice correlates to infectivity in humans; human studies to
determine the infective dose of C. parvum are rare, but have been performed (DuPont et al.,
1995). Water quality scientists do not agree that animal (mouse) infectivity is the irrefutable
method for estimating. Most water quality scientists agree that the animal infectivity method is
the best method presently available for estimating infectivity of the oocyst. The statistical
guidance does not exist that allows interpretation of the test results in a manner that is irrefutable.
Water quality investigators, scientists and engineers are reluctant to use animal infectivity studies
due to their expense. Additional tests of the performance of the Sentinel™ would be very
informative, and provide a more definitive dose response curve. For example, the experimental
design had not anticipated that the lower UV dose was effective in inactivating C. parvum.
Consequently, there was no UV dose response curve formed from the data. For a dose response
curve to be meaningful, many additional data points would be required, and this is beyond the
purpose of this study.
5.2 UV Dose Estimates and Measurements
There is a general lack of agreement among drinking water scientists and engineers as well as
understanding as to the best ways to determine and report UV dosage. In drinking water, three
approaches are used to determine the UV dose:
1. Determination of UV irradiance from sensor readings in mW/cm and multiplication of
irradiance by the residence time in seconds that is determined either theoretically or by tracer
studies.
2. Determination of UV dose using a biological indicator such as Bacillus subtilis spores or
MS-2 bactedophage. The dose response curve is determined from bench-scale collimated
46
-------
beam tests and then the full-scale unit is challenged with the biological indicator, inactivation
determined and actual UV dose is calculated from the dose response curves established from
the collimated beam tests.
3. Determination of dosage using a theoretical MPSS model first termed in the wastewater UV
literature as the point-source summation model.
The work presented here uses a very sophisticated mathematical model similar to the point-
source summation model, to calculate dosage. This model is undergoing peer review as part of
its publication in a scientific journal. However, the model results are limited by a lack of
verification data from more reliable electronic radiometer and actual velocity measurements or
from biological indicator results. Any comparison of the UV dose estimated in this study by the
MPSS method to the UV dose measured or estimated by another method is neither intended nor
recommended.
5.3 Low Flow Rate Testing
The system was designed to handle up to 500 gpm. It was tested at approximately 215 gallons
per minute (gpm) during the microbiological challenge testing and run at 25 gpm, during periods
when the information on O&M was collected due the limitations of the test site. Unfortunately,
the effects on the ability of the system to inactivate (oo)cysts as well as lamp operation, sensor
life, cleaning frequency and other O&M parameters at the higher flow rate of 500 gpm were not
collected in this study.
5.4 Feed Water Conditions
The verification testing was performed on a finished water source whose turbidity was <0.11
NTU, True Color was 5 TCU and whose UV254 absorption coefficient was between 0.02 and 0.06
cm"1 during microbiological challenge testing. System performance is highly dependent on feed
water quality characteristics. UV irradiance in water may be limited to waters of low turbidity,
and may have additional restraints determined by water chemistry, for example, pH, temperature,
UV absorption, nitrates, TOC and True Color.
5.5 Exclusion of Giardia Data
CCC elected to evaluate the performance of the Sentinel™ in inactivating G. muris by using the
excystation method. Although the C. parvum results demonstrate inactivation via animal
infectivity results, the same conclusion cannot be drawn for G. muris. G. muris may not
necessarily fully represent the human pathogen, Giardia lamblia. However, if interested, the
reader may review G. muris data in Appendix L.
47
-------
Chapter 6
Conclusions
Through this testing it was established that at a process flow rate of 215 gpm the Calgon Carbon
Corporation Sentinel™ UV Reactor could obtain an estimated 3.9 log inactivation of C. parvum
oocysts as determined by animal infectivity results with one lamp illuminated (out of six) at
reduced power (0.5 kW). Greater (> 4 log) inactivation was achieved at 215 gpm with higher
UV doses, respectively, with two lamps at reduced power (0.5 kW each), and with two lamps at
full power (1.0 kW each), again as determined by animal infectivity results.
Furthermore, the use of in vitro methods like the vital dyes and in vitro excystation significantly
under-estimated oocyst inactivation when compared to neonatal mouse infectivity (Figure 4-3).
During the verification period, water quality parameters that influence UV absorbance were
measured to assist in evaluating other waters for application of this UV system. During the
challenge periods, UV254 absorption coefficient was between 0.02 and 0.06; turbidity was <0.11
NTU. No iron or manganese was detected in the sample water; nitrates were no greater than 3.7
mg/L and total organic carbon was no greater than 4.3 mg/L. Other water quality conditions
would require short duration, site specific testing to establish performance.
Also of importance to this study was the operation of the equipment in the field. Several
deficiencies were noted with wiper failures, irradiance sensor, and attenuation tubes.
CCC has informed CO A and NSF that they are taking action to improve these portions of the
system.
Deficiency Action Taken
Wiper Failures More intensive plant quality control over
their manufacture.
Irradiance Sensor A re-design of the sensor circuit board.
Attenuation Tubes Changing the materials from Teflon to an
aluminum mirror configuration.
These elements will be required to undergo additional and separate verification testing.
48
-------
Chapter 7
Recommendations
Although the results of this study are very promising, they warrant additional research and
independent confirmation by other investigators to insure their validity. For example, the
threshold of UV dose at which there is a consistent logio reduction of oocysts needs to be
established to develop the appropriate safely factors for the field application of UV equipment. It
is important to regulators and users that the effective threshold UV dose be established. The
threshold UV dose can be developed from an IDso curve (a curve that defines the UV dose at
which fifty percent of the test animals become infected when fed treated oocysts). The
development of the IDso curve need only be performed under controlled laboratory conditions.
Consensus is needed on the use of the neonatal mouse infectivity method to represent the
inactivation of oocysts. The relationship between various viability methods, the study of which
is already in progress, is necessary for deciding upon a final method for assessing the inactivation
of cysts and oocysts. Once the scientific community can agree upon a method, then NSF and the
EPA can modify the ETV Test Plan accordingly and develop a standardized methodology. The
cost of performing the method must be affordable as well accurate, precise and reproducible.
The variance associated with the predictions of the existing animal infectivity logistical dose
response model (LOGIT) needs to be addressed with additional laboratory studies. Of particular
concern is when the LOGIT model extrapolates the decrease in oocyst infectivity outside the
initial calibration range of the model.
The estimation of UV dose in the field needs additional research and refinement. The MPSS
model needs verification using bioassay methods or radiometers that are not attenuated. NSF
TA/f
recommends that the UV dose produced by the Sentinel with the number of lamps ranging
from one to six be, at a minimum, estimated using a bioassay technique.
Since flow rate affects UV dose, all UV equipment should be verified at its maximum design
flow rate. In addition, the flow rate can affect the build-up of scale and other UV inhibiting
substances on the quartz sleeve of the UV lamps. For the verification of operation and
maintenance parameters, UV equipment should be tested under the minimum, median and
maximum flow rates. NSF will recommend these modifications to the ETV test plan for
Microbiological Inactivation by UV Technologies.
This ETV test focused on the inactivation of C. parvum and did not address other microbial
pathogens such as bacteria or viruses that might be present. Site specific testing would also be
suggested in determining applicability to other microbes.
49
-------
Chapter 8
References
References consulted in the preparation of the initial FOD and this report include the following:
ANSI/NSF 55-191 Ultraviolet Microbiological Water Treatment Systems, 1991.
APHA, Standard Methods for the Examination of Water and Wastewater, 19th ed. 1995.
Arrowood, M. J., and C. R. Sterling. Isolation of Cryptosporidium oocysts and sporozoites using
discontinuous sucrose and isopycnic percoll gradients. Journal of Par asitology 73: 314-319
(1987).
Arrowood, M. J., and K. Donaldson. Improved purification methods for calf-derived
Cryptosporidiumparvum oocysts using discontinuous sucrose and cesium gradients. Journal of
Eukaryotic Microbiology 43: 89S (1996).
AWWA, Guidance Manual for Compliance with the Filtration and Disinfection Requirements
for Public Water Systems Using Surface Water Sources, AWWA, Denver, Colorado, 1991.
Belosevic, M., R.A. Guy, R. Taghi-Kilani, N.F. Neumann, L.L. Gyurek, L.R.J. Liyange, PJ.
Millard, and G.R. Finch. Nucleic acid stains as indicators of Cryptosporidium parvum oocysts
viability. International Journal of Par asitology 21 (7): 787-798 (1997).
Bolton, James R., Bertrand Dussert, Zia Bukhari, Thomas Hargy and Jennifer L. Clancy,
"Inactivation of Cryptosporidium parvum by Ultraviolet Light: Comparison of Laboratory- and
Pilot-scale Results on Finished Water". Paper presented at Annual AWWA Conference, Dallas,
Texas, June 21-25, 1998.
Bolton, James R., Detailed Procedure for the calculation of the average irradiance in an
ultraviolet disinfection reactor using broadband medium-pressure mercury UV lamps, manuscript
in preparation, 1998.
Bukhari, Zia, Thomas M. Hargy, James R. Bolton, Bertrand Dussert, and Jennifer L. Clancy,
"Medium-Pressure UV for oocyst inactivation". Journal AWWA 91 (3): 86-94 (1999).
Campbell, A.T., L. J. Robertson, M.R. Snowball and H.V. Smith. Viability of Cryptosporidium
parvum Oocysts: Correlation of In Vitro Excystation with Inclusion or Exclusion of Fluorogenic
Vital Dyes. Applied and Environmental Microbiology 58 (11): 3488-3493 (1992).
Campbell, A.T., L. J. Robertson, M.R. Snowball and H.V. Smith. Inactivation of oocysts of
Cryptosporidium parvum by ultraviolet radiation. Water Research 29 (11): 2583-2586 (1995).
Clancy, Jennifer L. "Inactivation of Cryptosporidium parvum by Ultraviolet Light: Dose
Response Curves for Low and Medium Pressure Mercury lamps and Pilot Scale Demonstration
50
-------
Studies". Paper presented at Annual AWWA Water Quality Technology Conference, San Diego,
California, November 1-4, 1998.
Clancy, Jennifer L., Thomas M. Hargy, Marilyn M. Marshall, and John E. Dyksen. Innovative
Electrotechnologiesfor Cryptosporidium Inactivation, Electric Power Research Institute,
Denver, Colorado, 1998.
DuPont, Herbert L, M.D., Cynthis L. Chappel, Ph.D., Charles R Sterling, Ph.D., Pablo C.
Okhuysen, M.D., Joan B. Rose, Ph.D., and Walter Jakubowski. The Infectivity of
Cryptosporidium parvum in Healthy Volunteers, New England Journal of Medicine 332 (13):
855-859(1995).
EPA/NSF, Protocol for Equipment Verification Testing of Microbiological Contaminant
Inactivation by Packaged and/or Modular Drinking Water Treatment Systems for Small Public
or Private Water Supplies, Draft of February 20, 1998.
Finch, G.R., C.W. Daniels, E.K. Black F.W. Schaefer III and M. Belosevic. Dose Response of
Cryptosporidium parvum in Outbred Neonatal CD-I Mice. Applied and Environmental
Microbiology 59 (11): 3661-3665 (1993).
Frey, M.N., Carrie Hancock, Gary S. Logsdon. "A Synthesis Report of Cryptosporidium", Poster
presented at the AWWA Annual Conference, Atlanta, Georgia, June 18, 1997.
Huffman, Debra E. "Efficacy of Pulsed White Light to Inactivate Microorganisms," Paper
presented at Annual AWWA Water Quality Technology Conference, San Diego, California,
November 1-4, 1998.
Kiminski, J.C. Correspondence in New England Journal of Medicine 331 (22): 1529-1530
(1994).
Korich, D.G., J.R. Mead, M.S. Madore, N.A. Sinclair and C.R. Sterling. Effects of Ozone,
Chlorine Dioxide, Chlorine and Monochlorimine on Cryptosporidium parvum Oocyst Viability,
Applied and Environmental Microbiology 56(5): 1423-1428(1990).
LeChevallier, M.W., William D. Norton and Ramon G. Lee. Giardia and Cryptosporidium spp.
in Surface Water Supplies, Applied and Environmental Microbiology 57 (9): 2610-2616 (1991).
LeChevallier, M.W., William D. Norton and Ramon G. Lee. Giardia and Cryptosporidium spp.
in Filtered Drinking Water Supplies, Applied and Environmental Microbiology 57 (9): 2617-
2621 (1991).
MacKenzie, W.R., Neil J, Hoxie, Mary E. Proctor, M. Stephen Gradus, Kathleen A. Blair, Dan
E. Peterson, James J. Kazmierczak, David G. Addiss, Kim R. Fox, Joan B. Rose and Jeffrey P.
Davis. A Massive Outbreak in Milwaukee of Cryptosporidium Infection Transmitted through the
Public Water Supply, New England Journal of Medicine 331 (3): 161-167 (1994).
51
-------
Meulemans, C.C.E. The Basic Principles of UV-Disinfection of Water, Ozone Science and
Engineering (9): 299-314 (1987).
Parrotta, M. J. and Faysal Bekdash. UV Disinfection of small groundwater supplies, Journal
AWWA90(2):71-81 (1998).
Pett, Brian, Franklyn Smith, Douglas Stendahl, and Roland Welker. "Cryptosporidiosis Outbreak
from an Operations Point of View, Kitchener-Waterloo, Ontario, Spring 1993", Paper presented
atAWWA Water Quality Technology Conference, Miami, Florida, November 1993.
Regional Municipality of Waterloo (Water Services Division), Water Quality Data Report, July
1996.
Rice, Eugene W., and John C. Hoff. Inactivation ofGiardia lamblia cysts by Ultraviolet
Irradiation, AppliedandEnvironmentalMicrobiology 42 (3): 546-547 (1981).
Robertson, L.J., A.T. Campbell and H.V. Smith. In vitro excystation of Cryptosporidium
parvum. Journal of Parasitology 106: 13-19 (1993).
Severin, Elaine F. and Peter F. Roessler. Resolving UV Photometer Outputs with Modeled
Intensity Profiles, Water Research 32 (5): 1718-1724 (1998).
Slifko, Theresa R., Debra Friedman, Joan B. Rose and Walter Jakubowski. An In Vitro Method
for Detecting Infectious Cryptosporidium Oocysts with Cell Culture, Applied and Environmental
Microbiology 63 (9): 3669-3675 (1997).
Slifko, Theresa R., Debra Friedman, and Joan B. Rose. "Comparison of 4 Cryptosporidium
parvum Variability Assays: DAPI/PI, Excystation, Cell Culture and Animal Infectivity", Paper
presented atAWWA Water Quality Technology Conference, San Diego, CA, November 1998.
USEP A, (1996). Ultraviolet Light Disinfection Technology in Drinking Water Application—an
Overview, Office of Water. Washington, D.C. EPA 811-R-96-002.
USEP A, (1998). Method 1622: Cryptosporidium in Water by Filtration/IMS/FA. Office of Water.
Washington, DC. EPA 821-R-97-023.
Verkoeven, J.W. Glossary of Terms Used in Photochemistry, Pure and Applied Chemistry 68
(12): 2223-2286 (1996).
Watanabe, Myrna E. New Cryptosporidium Testing Methods, Environmental Science and
Technology 30 (12): 532A-535A (1996).
Wolfe, R.L. Ultraviolet disinfection of potable water, Environmental Science Technology 24 (6):
768-773 (1990).
52
------- |