United States
Environmental Protection
Agency
Office of Research and
Development
Washington DC 20460
EPA/600/R-01/110
December 2001
Controlling Disinfection
By-Products and
Microbial Contaminants in
Drinking Water

-------
                                                 EPA/600/R-01/110
                                                   December 2001
      Controlling  Disinfection
    By-Products  and Microbial
Contaminants in  Drinking Water
                      Edited by

                    Robert M. Clark
           Water Supply and Water Resources Division
          National Risk Management Research Laboratory
              Office of Research and Development
              U.S. Environmental Protection Agency

                        and

                    Brenda K. Boutin
           National Center for Environmental Assessment
              Office of Research and Development
              U.S. Environmental Protection Agency
          National Risk Management Research Laboratory
              Office of Research and Development
              U.S. Environmental Protection Agency
                  Cincinnati, Ohio 45268

-------
                                          Notice
    The information in this report has been subjected to Agency peer and administrative review and has
been approved for publication as an EPA document. Mention of trade names or commercial products does
not constitute endorsement or recommendation for use.
                                         Abstract

    Historically drinking water utilities in the United States (U.S.) have played a major role in protecting
public health through the reduction of waterborne disease. These reductions in waterborne disease out-
breaks were brought about by the use of sand filtration, disinfection and the application of drinking water
standards. Coincident with the passage of the SDWA of 1974, it was discovered that chloroform was a
disinfection by-product (DBP) resulting from the interaction of chlorine with natural organic matter in water.
Chloroform is one of a class of compounds called trihalomethanes. This finding posed a serious dilemma
because it raised the possibility that chemical disinfection, which clearly reduced the risk of infectious
disease, might also result in the formation of potentially harmful chemical by-products. Although disinfec-
tion of public drinking water had dramatically  reduced outbreaks of diseases attributable to waterborne
pathogens, the identification of chloroform in drinking water raised questions about possible health risks
associated with these exposures. In the United States, since 1974, additional DBFs have been identified
and concerns  have  intensified about health risks resulting from exposures to them. Although a causal
relationship between DBP exposures and these health risks has not been  conclusively established, risk
managers have responded, in the interest of protecting public health, by developing alternative treatment
systems and issuing rules and regulations designed to  maintain protective levels of disinfection while
reducing potentially harmful levels of DBPs. In 1981, the USERA issued a report intended to summarize the
"state-of-the-art" regarding the control of disinfection by-products in drinking water. However, EPA's current
drinking water research program is more sophisticated than it was twenty years ago. For example, when
the treatment technology manual was published in  1981, it reported primarily on treatment oriented re-
search. Twenty years later, the technology research program includes source water protection, treatment
technology and distribution system studies. The research  also reflects a concern over balancing the risks
of potential carcinogenic exposure against the risks from microbial infection. This document is intended to
summarize the research that has been conducted in technology research by EPA since the publication of
the 1981 treatment technology document.

-------
                                       Foreword
    The U.S. Environmental Protection Agency is charged by Congress with protecting the Nation's land,
air, and water resources. Under a mandate of national environmental laws, the Agency strives to formu-
late and implement actions leading to a compatible balance between human activities and the ability of
natural systems to support and nurture life. To meet this mandate, ERA'S research program is providing
data and technical support for solving environmental problems today and building a science knowledge
base necessary to manage our ecological resources wisely, understand how pollutants affect our health,
and prevent or reduce environmental risks in the future.


    The National Risk Management Research Laboratory is  the Agency's center for investigation of
technological and management approaches for preventing and reducing risks from pollution that threat-
ens human health and the environment. The focus of the Laboratory's research program is on methods
and their cost-effectiveness  for prevention and control of pollution to air, land, water, and subsurface
resources; protection of water quality in public water systems;  remediation of contaminated sites, sedi-
ments and ground water; prevention and control of indoor air pollution; and restoration of ecosystems.
NRMRL collaborates with  both public and  private sector partners to foster technologies that reduce the
cost of compliance and to anticipate emerging problems.  NRMRL's research provides solutions to envi-
ronmental problems by: developing and promoting technologies that protect and improve the environ-
ment; advancing scientific and  engineering information to support regulatory and policy decisions; and
providing the technical support and information transferto ensure implementation of environmental regu-
lations and strategies at the national, state, and community levels.


    This publication has been produced as part of the Laboratory's strategic long-term research plan. It
is published and made available by ERA'S  Office of Research and Development to assist the user com-
munity and to link researchers with their clients.
                                            E. Timothy Oppelt, Director
                                            National Risk Management Research Laboratory

-------
                                         Preface

    As part of his response to a letter from Robert Hooke, Sir Isaac Newton wrote, "If I have seen further
... it is by standing on the shoulders of giants." This statement, made in the  17th century, could easily
describe the progress we have made over the last 25 years in identifying and  controlling disinfection by-
products (DBFs) and microbial contaminants in drinking water. In 1972, I was assigned to the staff of Dr.
Andrew W. Breidenbach, director of EPA's first National Environmental Research Center, and worked with Mr.
Gordon Robeck to help bring the Water Supply Research Laboratory into existence. At the time, I had no idea
of the impact that the program would have on drinking water in the United States and the world, nor did I have
the slightest idea that I would eventually serve as the program's director for 14 years.

    I joined the staff of the Water Supply Research Laboratory shortly before the  passage of the Safe Drink-
ing Water Act in 1974. At that time, we were just beginning to recognize that chloroform was a by-product
resulting from the interaction of chlorine with natural organic matter in drinking water. During my early years
in the program, I had a chance to learn from not only Gordon Robeck, but from other such eminent scientists
as Dr. James Symons, Mr. Leeland McCabe, and Mr. Edwin Geldreich. Although Robeck, Symons, McCabe,
and Geldreich were "giants," there were many other talented and highly productive individuals who contrib-
uted to the success of the program as well.  In  1981, we published a document titled "Treatment Techniques
for Controlling Trihalomethanes in Drinking Water," which was an attempt to summarize contemporary knowl-
edge in that important area. It was highly successful and, at the time, was considered a benchmark in the
field. It was so successful that, after the USEPA's supply of the report was exhausted, it was republished by
the American Waterworks Association and by the Japan Waterworks Association. This document is in-
tended to complement the 1981 volume by describing research completed by the National Risk Management
Research Laboratory in the 20 years between 1981 and 2001.

    In 1972, all of the drinking water research activities conducted by the USERA were concentrated in the
Water Supply Research Laboratory. When the Safe Drinking Water Act was passed in 1974, the Laboratory
was blessed with a generous allotment of funds and staff. However, through reorganizations and redirections
overthe past 20 years, various aspects of the program have been transferred to other organizational units in
EPA. In the early 1990s, there was serious  consideration of its elimination because it  represented "mature
technology." Support from the American Waterworks Association (AWWA), the American Waterworks As-
sociation Research Foundation  (AWWARF), and the Association of State Drinking Water Administrators
(ASDWA) helped it survive that difficult period.

    Drinking water research has now become an integral part of the USEPA's base research program. Each
of the laboratories and centers in EPA's Office of Research and Development (ORD)  has a core research
program devoted to various  aspects of drinking water research.  It is effectively coordinated by Dr. Fred
Hauchman, who serves as National Drinking Water Research Program Manager, and by Mr. E.  Timothy
Oppelt who, as Director of the National Risk Management Research Laboratory  (NRMRL), is the Executive
Lead. The EPA works collaboratively with other organizations and research programs including the AWWA,
AWWARF, ASDWA, the National Association of Water Companies, and the Association of Metropolitan Water
Authorities.

    Just as the state of drinking water research in EPA has changed, the nature of the science and engineer-
ing support for the program has also changed. During the past 20 years, research in the areas of disinfection
by-products and microbial contaminant control has become complex and scientifically challenging. NRMRL's
Water Supply and Water Resources Division (WSWRD), the direct successor of the Water Supply Research
Laboratory that initially focused on treatment technology, has evolved into a program that is fundamental and
science-based. It researches small systems technologies, distribution systems, and source water protection,
and has sponsored projects throughout the  world.

    The ORD drinking water  program has a long and  productive history in EPA and has evolved into a
broadly based, complex, and scientifically challenging program. I believe that the material contained in this
document reflects the progress that we have made in research  related to the control of DBPs and microbial
control in drinking water during the past 20 years. The underlying principle that continues to govern all of our
research in WSWRD is protecting the public health of the American drinking water consumer.

 Robert M. Clark
 Senior Research Engineering Advisor
 Water Supply and Water Resources Division
 National Risk Management Research Laboratory, USERA

                                          iv

-------
                                  Table of Contents
Abstract	ii
Preface	iv
Abbreviations	viii
Acknowledgements	xiii

Chapter 1 Control of Microbes and DBFs in Drinking Water: An Overview	1-1
Introduction	1-1
Drinking Water Regulations in the United States	1-3
Trends in Compliance with the SDWA	1-4
Current DBP and Microbial Regulations	1-6
Chemistry of DBP Formation	1-6
Source Water Protection	1-7
Microbial Pathogen Disinfection	1-8
Controlling Alternative DBPs	1-8
Control of DBPs Using Biological Filtration	1-9
Controlling Microbial Contaminants Using Filtration	1-9
Controlling DBPs and Microbes Using GAG and Membranes	 1-10
Removing DBP Precursors Using Enhanced Coagulation	 1-11
Controlling Microbes and DBPs in Small Systems	 1-11
Modeling Chlorine Residuals and DBP Formation	 1-12
Distribution System Water Quality	 1-12
Cost of Controlling DBPs and Microbial Contaminants	 1-13
ERA'S Technology Research Program: Some Final Thoughts	1-14

Chapter 2 A Review of Federal Drinking Water Regulations in the U.S	2-1
Introduction	2-1
The Safe Drinking Water Act (SDWA)

Chapter 3 Disinfection By-Product (DBP) Chemistry: Formation and Determination	3-1
Introduction	3-1
General Issues in Disinfection: Disinfectants and Source Material for DBPs	3-6
An Overview of Disinfection By-Product Formation Source Material	3-11
EPA Research into DBP Formation and Chemistry	3-13
Directions in  DBP Analytical Chemistry Research	3-23

Chapter 4 Source Water Protection: Its Role in Controlling Disinfection	4-1
  By-Products (DBPs) and Microbial Contaminants
Introduction	4-1
The Safe Drinking Water Act and Source Water Protection	4-2
Threats to Source Water Quality	4-4
Sources of Oocysts	4-10
Measuring and Monitoring Pathogens in Source Waters	4-13
Protecting Source Waters	4-15
Best Management Practices	4-19
Source Water Protection and Watershed Management	4-22
Modeling and Source Water Protection	4-23
Summary and Conclusions	4-25

Chapters Disinfection	5-1
Introduction	5-1
Chlorine	5-1
Chloramine	5-7

-------
                                  Contents, Cont'd.

Chlorine Dioxide	5-9
Ozone	5-10
Ultraviolet (UV) Irradiation	5-11
Summary	5-12

Chapter 6 Alternative Disinfectants	6-1
Introduction	6-1
Models for Assessing Halogenated DBP Precursors	6-1
Chloramines	6-2
Chlorine Dioxide	6-6
Ozone	6-9
Summary	6-15

Chapter/ Disinfection By-Product Control Through Biological Filtration	7-1
Introduction	7-1
EPA-Funded Studies	7-1
Discussion	7-14

Chapters Microbiological Removal by Filtration Processes	8-1
Introduction	8-1
Slow Sand Filtration	8-3
Diatomaceous Earth Filtration	8-5
Granular Media Filtration	8-6
Summary	8-16

Chapter 9 Activated Carbon and Membrane Processes for	9-1
  Disinfection By-Product (DBP) and Microbial Control
Introduction	9-1
Activated Carbon 	9-2
Membranes	9-13
Conclusions	9-23

Chapter 10 Coagulation	10-1
Introduction	10-1
Background	10-1
Conventional vs. Enhanced vs. Optimized Coagulation	10-3
Enhanced Coagulation's Role in Water Quality	10-4
Coagulation With and Without Acid Addition	10-6
Comparing Alum and Iron Coagulation	10-7
Fractionation	10-8
Speciation	10-9
Scale Up	10-10
Secondary Effects	10-11
Summary	10-13

Chapter 11  Controlling Disinfection By-Products (DBFs) and	11-1
  Microbial Contaminants in Small Public Water Systems (PWSs)
Introduction	11-1
SDWA Coverage	11-2
Research Approach	11-5
Performance Results	11-9
Controlled Turbidity Challenge Experimental Results	11-11

                                          vi

-------
                                 Contents, Cont'd.


Disinfection	11-19
Field-Scale Demonstration Projects	11-22
Small System Remote Monitoring and Control Technology	11-30
Summary	11-34

Chapter 12 Modeling Chlorine Decay and the Formation of	12-1
  Disinfection By-Products (DBFs) in Drinking Water
Introduction	12-1
Chemistry of Disinfectants in Water	12-2
Modeling the Decay of Chlorine Residuals	12-3
EPA Research Activities	12-5
Modeling the Formation of DBPs	12-10
EPA Research Activities	12-14
Exposure to DBPs from Distribution Systems	12-18
Modeling Chlorine Decay and TTHMs in Distribution Systems	12-18
Evolution of System Modeling	12-24
Water Quality and Tanks	12-26
Policy Issues	12-28
Summary and Conclusions	12-28

Chapter 13 Biofilms in Drinking Water Distribution Systems	13-1
Introduction	13-1
Previous Research	13-2
Pilot Systems Research	13-5
On-Going and Future Biofilm Research	13-7
Summary and Conclusions	13-8

Chapter 14 Control of Microbial Contaminants and Disinfection	14-1
  By-Products (DBPs): Cost and Performance
Introduction	14-1
The Role of Disinfection in the U.S	14-1
Microbiological Control	14-2
Formation of DBPs	14-3
Treatment Strategies for Controlling DBPs	14-4
Comparative Analysis	14-10
Summary and Conclusions	14-10

Glossary	G-1
                                         VII

-------
                           Abbreviations
AOC	assimilable organic carbon
AOP	Advanced Oxidation Process
APO	antibiotic-preserved oocysts
ARC/INFO®       ESRI's® (Environmental Systems Research Institute, Inc.) commercial
 	CIS (Geographic Information System) software
ARM-II	Agricultural Runoff Management Model
ASTM	American Society forTesting and Materials
AWBERC	Andrew W. Breidenbach Environmental Research Center
AWWA	American Water Works Association
AWWARF	American Water Works Association Research Foundation
BAA	bromoacetic acid
BAG	biological activated carbon
BAT	best available technology
BCAA	bromochloroaceticacid
BCAN	bromochloroacetonitrile
BDOC	biodegradable dissolved organic carbon
BMP	best management practice
BOM	biodegradable organic matter
BWC	backwash chlorinated
CAA	chloroaceticacid
CCC	Chlorine Chemistry Council
CCL	contaminant candidate list
CFU	colony forming unit
CGR	coliform growth response
CH	chloral hydrate
Ci	curie
CLR	chlorine residual
CLSA	closed loop stripping apparatus
CP	chloropicrin
CSO	combined storm water-sewage overflows
CSTR	continuous-flow stirred-tank reactor
CSU	Colorado State University
CT	contact time
CT-ORW	conventionally treated Ohio River water
CTA	cellulose triacetate
CWA	Clean Water Act
DBAA	dibromoaceticacid
DBAN	dibromoacetonitrile
DBMS	data base management system
DBP	disinfection by-product
DBPFP	disinfection by-product formation potential
DBPR	Disinfection By-products Rule
DCAA	dichloroaceticacid
DCAN	dichloroacetonitrile
D/DBP	disinfectants/disinfection by-products
DE	diatomaceous earth (filtration)

                                 viii

-------
                      Abbreviations, Cont'd.
DIG	differential interference contrast
DO	dissolved oxygen
DOC	dissolved organic carbon
DOM	dissolved organic matter
DPO	dichromate-preserved oocyst
DSS	Distribution System Simulator
DVM	Discrete-Volume Method
DWF	dry-weather flow
DWQM	Dynamic Water Quality Model
DWSRF	Drinking Water State Revolving Fund
EBCT	empty bed contact time
ECD	electron-capture detection
EDM	Event-Driven Method
EFL	East Fork Lake
EPA	Environmental Protection Agency (US)
EPANET	An EPA-developed computer program that performs simulation of hydraulic
 	and water quality behavior within drinking water distribution systems
EPS	extracellular polysaccharide
ESCA	Electron Spectroscopy for Chemical Analysis
ESWTR	Enhanced Surface Water Treatment Rule
ETV	Environmental Technology Verification (Program)
FAC	filtration avoidance criteria (Chapter 4)
FAC	free available chlorine (Chapter 11)
FBDOC	fast biodegradable dissolved organic carbon
FC	fecal coliform
FDM	Finite-Difference Method
FID	flame-ionization detection
FISH	fluorescent in situ hybridization
FLOWSED	A one-dimensional mathematical model that computes FLOW conditions
 	and SEDiment movement
FP	formation potential
FRT	filter run time
FS	fecal streptococci
FY	fiscal year
GAG	granular activated carbon
GC	gas chromatography
GC/MS	gas chromatography/mass spectroscopy
CIS	Geographic Information System
GM	geometric mean
GMR	Great Miami River
GS	Green Swamp
GT	coagulant mixing intensities
GWR	Ground Water Rule
GWUDI	ground water underthe direct influence of surface water
HAA	haloacetic acid
HAAFP	haloacetic acid formation potential
HAN	haloacetonitrile

                                  ix

-------
                      Abbreviations, Cont'd.
HFTF	high-flow thin film
HL	head loss
HPC	heterotrophic plate count
http	hyper text transfer protocol
HX	hydrohalic acid
1C	ion chromatography
IC-ICP-MS	ion chromatography inductively coupled plasma mass spectrometry
ICP-MS	inductively coupled plasma mass spectrometry
ICR	Information Collection Rule
IESWTR	Interim Enhanced Surface Water Treatment Rule
IFA	indirect fluorescent monoclonal antibody
IMS	integrated membrane system
IR	infrared (spectroscopy)
JWWA	Japan Water Works Association
L	liter
LCR	Lead and Copper Rule
LTD	long-term demand
LTESWTR	Long Term Enhanced Surface Water Treatment Rule
MCLG	maximum contaminant level goal
MCL	maximum contaminant level
M/DBP	microbial pathogens/disinfection by-products
MF	microfiltration
mg/L	milligrams per liter
MIOX	mixed oxidants
MR	Mississippi River
M/R	monitoring and reporting
MRDL	maximum residual disinfectant level
MRDLG	maximum residual disinfectant level goal
MS	mass spectroscopy
MS	molecular size
MSL	mean sea level
MWCO	molecular-weight cutoff
MWL	Miami Whitewater Lake
MX	MutagenX
MAS	National Academy of Sciences
NBDOC	non-biodegradable dissolved organic carbon
NC	non-chlorinated
NCER	National Center for Environmental Research
ND	not detected
NF	nanofiltration
NIEHS	National Institute of Environmental and Health Sciences
NIPDWR	National Interim Primary Drinking Water Regulations
NMR	nuclear magnetic resonance
NMWD	North Marin Water District
NOM	natural organic matter

-------
                       Abbreviations, Cont'd

NPDES	National Pollutant Discharge Elimination System
NPDWR	National Primary Drinking Water Regulations
NPOX	nonpurgeable organic halide
NSDWR	National Secondary Drinking Water Regulations
NPWA	North Penn Water Authority
NRCS	Natural Resources Conservation Service
NRMRL	National Risk Management Research Laboratory
NTM	non-tuberculosis mycobacteria
NTU	nephelometric turbidity unit
O&M	operation and maintenance
OBP	ozone by-products
OR	Ohio River
ORD	Office of Research and Development
ORP	oxidation reduction potential
ORSANCO	Ohio River Valley Water Sanitation Commission
ORW	Ohio River water
PAC	powdered activated carbon
PBS	phosphate buffered saline
PC	prechlorinated
PCA	plate count agar
PCB	polychlorinated biphenyl
PCR	polymerase chain reaction
PFU/L	plaque forming unit/liter
PHS	Public Health Service
PM	precursor material
POE	point-of-entry
POD	point-of-use
POX	purgeable organic halide
POXFP	purgeable organic halide formation potential
PVC	polyvinyl chloride
PWS	public water system
QUALNET	A temporal and spatial prediction model of chlorine distribution in a pipe
 	network under unsteady-flow conditions
R2A	A low-nutrient-content growth medium for performing heterotrophic plate
 	counts
Reg-Neg	Regulatory-Negotiation Committee
RO	reverse osmosis
RF1	A topological and geographic (CIS) coverage of the primary rivers and
 	streams in the coterminous United States
RSSCT	Rapid Small Scale Column Test
RTS	Remote Telemetry System
RTU	remote telemetry unit
SAC	spectral absorption coefficient
SAR	structure-activity relationship
SBDOC	slowly biodegradable dissolved organic carbon
SCADA	Supervisory Control and Data Acquisition
SCCRWA	South Central Connecticut Regional Water Authority
                                 XI

-------
                      Abbreviations, Cont'd.

SDS	simulated distribution system
SDWA	Safe Drinking Water Act
SDWAA	Safe Drinking Water Act Amendments
SL	Stonelick Lake
SMCL	secondary maximum contaminant level
SOC	synthetic organic chemical
SSF	slow sand filtration
SSO	sanitary sewer overflow
STP	sewage treatment plant
SUVA	specific ultraviolet absorbance
SWP	source water protection
SWTR	Surface Water Treatment Rule
TC	total coliform
TCAA	trichloroacetic acid
TCAN	trichloroacetonitrile
TCR	Total Coliform Rule
TDM	Time-Driven Method
TDS	total dissolved solids
T&E	test and evaluation
THM	trihalomethane
THMFP	trihalomethane formation potential
TOC	total organic carbon
TOX	total organic halide
TOXFP	total organic halide formation potential
TPC	total particle count
TSA-SB	tryptic soy agar-sheep's blood
TSS	Ten-State Standards
TT	treatment technique
TTHM	total trihalomethane
TTHMFP	total trihalomethane formation potential
UF	ultrafiltration
UFC	uniform formation condition
UK	United  Kingdom
USDHEW	United  States Department of Health, Education and Welfare
USERA	United  States Environmental Protection Agency
USGS	United  States Geological Survey
UV	ultraviolet
UVA	ultraviolet absorbance
WASP4	Water Quality Analysis Simulation Program (release number four)
WSWRD	Water Supply and Water Resources Division
www	World Wide Web
XAD®            Afunctionalized poly (styrene-di-vinylbenzene) resin
                                 XII

-------
                               Acknowledgements
    Many individuals have worked diligently to make this document possible. Both Mr. Oppelt and Mr.
Clyde Dempsey, NRMRL's Assistant Laboratory Director for Water, have provided critical support that
has made the research described in this document possible. Mr. Lee Mulkey, Associate Laboratory
Director for Ecology in NRMRL, has consistently encouraged us to broaden our research to include
watershed management and source water protection. Chapter 4 is a reflection of this more expansive
approach. NRMRL's Associate Laboratory Director for Health, Dr. Hugh Mckinnon, has been a con-
stant source of support and encouragement and has continually reminded us of the need to maintain
our concern for public health at the forefront of our research.

    NRMRL's Technology Transfer and Support Division has  partnered with WSWRD by providing a
substantial portion of the funds and staff time that made this document possible. We truly appreciate
the support provided by Mr. Dan Murray, Director of the Technology Transfer and Support Division
(TTSD), and Messers. Randy Revetta, Pat Burke, Steve Wilson, and Ms. Carol Grove, all ofTTSD. Mrs.
Sue Schock, TTSD, played an indispensable role as Work Assignment Manager for a support contract
with SAIC underwhich the document was produced and edited. Dr. Faysal Bekdash of SAIC developed
the glossary and list of abbreviations. Custom Editorial Productions subcontracted with SAIC to per-
form the editorial and desktop publishing tasks under the management of Ms. Jan Clavey.
Robert M. Clark

Water Supply and Water Resources Division
National Risk Management Research Laboratory USEPA
                                         XIII

-------
                                      CHAPTER 1

         Control of Microbes and DBFs in Drinking Water: An Overview1

Introduction

Historically, drinking water utilities in the United States (U. S.) have played the maj or role in protecting
public health through the reduction of waterborne disease. For example, in the 1880s, the typhoid death
rate was 158 deaths per 100,000 in Pittsburgh, PA. But by 1935, the typhoid death rate had declined to
5 per 100,000 (Clark et al. 1991). These reductions in waterborne disease outbreaks were brought about
by the use of sand filtration, disinfection, and the application of drinking water standards. Despite this
excellent record, the occurrence of occasional drinking water quality problems are a reminder of the
need for constant vigilance. In 1993, Milwaukee, WI, suffered a Cryptosporidiosis outbreak; it was
estimated  that over 400,000 people were made ill  and an  estimated 75-100 immune-compromised
people died (Blair 1994). In July  of  1993, Manhattan, NY,  was placed on a boil water order, as was
Washington, D.C., in December of 1993. Both systems experienced microbial maximum contaminant
level (MCL) violations under the Federal Safe Drinking Water Act (SDWA) (Clark et al. 1999).

Growing national concern  over the need to protect drinking water quality in the United States was
reinforced by the U.S. Congress  on  December 16, 1974. On  that date it passed the  SDWA, which
established the first set of federally enforceable drinking water regulations in the history of the U.S.
Section 1401  (1)(D) of the SDWA, Public Law 93-523, states that "the term 'primary  drinking water
regulation' means a regulation which contains criteria and procedures to assure a supply of drinking
water which dependably complies with such maximum contaminant levels. .." and Section 1412 (a)(2)
states that "National interim primary drinking water regulations promulgated under paragraph (a)(l)
shall protect health to the extent feasible, using technology, treatment techniques, and other means,
which the Administrator determines are generally available  (taking costs into consideration). . ." This
provision  of the  SDWA established the  requirement that a "Treatment Techniques" document must
accompany the establishment of an MCL for any regulated contaminant (SDWA 1974).

Coincident with the passage of the SDWA, Rook (1974) and Bellar and Lichtenberg (1974) reported,
nearly simultaneously, that chloroform was a disinfection by-product (DBF) resulting from the interac-
tion of chlorine with natural organic matter in water. Chloroform is one of a class of compounds called
trihalomethanes.  This finding posed a serious dilemma because it raised the possibility that chemical
disinfection, which clearly reduced the risk of infectious disease, might also result in the formation of
potentially harmful chemical by-products. In the United States,  since 1974, additional DBFs have been
identified, and concerns have intensified about health risks resulting from exposures to them (Bull
1993). Although a causal relationship between DBF  exposures and these health risks has not been
conclusively established, risk managers have responded, in the interest of protecting public health, by
developing alternative treatment systems and issuing rules and regulations designed to maintain pro-
tective levels of disinfection while reducing potentially harmful levels of DBFs  (USEPA 1998a, b).
Robert M. Clark and Robert C. Thurnau: ORD/NRMRL/WSWRD, AWBERC Mailstop 689, 26
West Martin Luther King Dr., Cincinnati, OH 45268. Corresponding Author: Robert M. Clark, 513-
569-7201, clark.robertm@epa.gov.


                                            1-1

-------
After discovering that chloroform was, in fact, a by-product of disinfection, it was also determined,
using the best information available, that chloroform was a possible carcinogen (Symons et al. 1981).
This finding sparked a flurry of research both in and outside the U.S. Environmental Protection Agency
(EPA). Results from these studies formed the basis for an amendment to the National Interim Primary
Drinking Water Regulations issued on November 29, 1979. The amendment established an MCL of
0.10 mg/L total trihalomethanes (TTHM) in drinking water (Federal Register 1979).

These events caused the EPA to focus its drinking water research program on establishing the scientific
basis for control of DBFs. It attempted to answer the following questions (Symons et al. 1981):

    1.  How does the consumption of trihalomethanes affect the health of consumers?
    2.  How should trihalomethanes be measured?
    3.  How do water quality conditions influence trihalomethane formation?
    4.  What treatment technique(s) can a drinking water utility use to reduce trihalomethane con-
       centrations in distributed water?
    5.  What effect will altering treatment procedures for controlling trihalomethanes have on the
       microbiological quality of distributed water?
    6.  What are the costs of the various treatment alternatives for controlling trihalomethanes?

In 1981, the Drinking Water Research Division of EPA's Municipal Environmental Research Labora-
tory issued a report entitled "Treatment Techniques for Controlling Trihalomethanes in Drinking Wa-
ter" (Symons etal. 1981). This document was intended to satisfy the need for a "Treatment Technique"
document for the TTHM regulation. It reviewed, as much as was known at that time, the health impli-
cations of trihalomethanes in drinking water. It summarized the results of research intended to answer
the questions listed above and attempted to help the reader choose the most cost-effective treatment
techniques for TTHM control. The EPA published 2000 copies of the document, and it was then repub-
lished by the American Water Works Association (AWWA). It was later translated into Japanese and
was distributed by the Japan Water Works Association (JWWA).

In the intervening twenty years, EPA has conducted a great deal of research in an attempt to provide
answers to the six questions listed  above.  In 1997, the EPA issued a Research Plan for Microbial
Pathogens and Disinfection By-products in Drinking Water (M/DBP) (USEPA 1997a). The M/DBP
plan is a comprehensive summary of the multi-disciplinary research program being conducted by EPA
to deal with identification and control of disinfection by-products and microbial contaminants in drink-
ing water.

After the passage of the 1986 SDWA Amendments, there was growing concern that a single focus on
controlling DBFs could be detrimental to protecting against microbial contamination in drinking water
distribution systems. This concern resulted in the Surface Water Treatment Rule (SWTR) and the Total
Coilform Rule (TCR), both of which were promulgated to insure that microbes in drinking water sys-
tems would be controlled as well as DBFs (USEPA 1989a,b). This concern is also reflected  in the
research program outlined in the M/DBP plan.

This report herein is intended to provide an overview of the treatment technology research conducted
by EPA since the publication of "Treatment Techniques for Controlling Trihalomethanes  in Drinking
Water" (Symons et al. 1981). It  describes the nature and scope of the regulations which have been
promulgated since the passage of the SDWA. It discusses recently identified disinfection by-products
and related regulations and reviews some of the trends that have developed with respect to compliance
with the rules and regulations under the SDWA. Finally, it summarizes and reviews treatment technol-
ogy research efforts conducted by EPA since 1981.
                                            1-2

-------
Drinking Water Regulations in the United States

The U.S. has nearly 60,000 community water supplies serving over 226 million people. Most of the
community water systems supply water to less than 500 people. Over 63% of these systems supply
water to less than 2.4% of the population, while 5.4% supply water to 78.5% of the population. Clearly,
a few large systems supply drinking water to most of the U.S. population.  In addition to the 60,000
community water supplies, there are 140,000 non-community water systems that serve schools, recre-
ational areas, trailer parks, etc. Nearly 98% of the non-community systems use ground water as a
source, and approximately 80% of the community systems use ground water as a source. Many of the
utilities that use ground water practice disinfection only, and a large number do not practice any treat-
ment at all (USEPA 1999a).

In its original form, the SDWA established a set of regulations known as the National Interim Primary
Drinking Water Regulations (NIPDWR) (SDWA 1974). The regulations included MCLs for ten inor-
ganic contaminants, six organic contaminants, turbidity, coliform, radium-226, radium-228, gross al-
pha activity, and man-made radionuclides. The NIPDWR also  established monitoring and analytical
requirements for determining compliance. Public water systems were defined as those which provided
piped water to the public for human consumption and had at least 15 service connections or regularly
served an average of 25 persons at least 60 days out of the year.

The MCLs for coliforms, nitrate,  and turbidity (surface water only) applied to both community and
non-community systems, while the other  MCLs applied only to community systems. The SDWA of
1974 required EPA to review and revise the NIPDWR systematically as appropriate. As mentioned
previously, a maj or revision of the NIPDWR occurred in 1979, when an MCL for total trihalomethanes
was promulgated (USEPA 1979).

On June 19, 1986, the SDWA was amended by Public Law 99-339, known as the Safe Drinking Water
Act Amendments of 1986. The 1986 Amendments promulgated a requirement and a schedule for EPA
to implement regulations for 83 contaminants which  had been published  previously in "Advanced
Notices of Proposed Rulemaking" as contaminants being considered for regulation. EPA was allowed
to substitute up to seven contaminants onto the list of 83. In addition, EPA was required to publish a list
every three years of contaminants "known or anticipated to occur in public water systems and which
may require regulation under this act" (those which may have any adverse effect on the health of
persons).

The 1986 Amendments stipulated that regulations should contain a maximum contaminant level goal
(MCLG), a health-based concentration "at which no known or anticipated adverse effects on the health
of persons occur and which allows an adequate margin of safety." MCLs were to be set at a level as
close to the MCLG as feasible. Feasible was defined in the law as "feasible with the use of the best
technology, treatment techniques and other means which the Administrator finds, after examination for
efficacy under field conditions and not solely under laboratory conditions are available (taking cost into
consideration)." Granular activated carbon (GAC) was designated in the law as feasible for the control
of synthetic organic chemicals, and any technology, treatment technique, or other means found to be the
best available for control of synthetic organic chemicals had to be at least as effective as GAC. Each
regulation which established an MCL was required to list the technology, treatment technique, or other
means  which the Administrator found to  be feasible for purposes of meeting the MCL. These have
become known as best available technology (BAT).

In the event that EPA found it was not "economically or technologically feasible to ascertain the level
of the contaminant," the law authorized the Administrator to promulgate a regulation that required the
                                            1-3

-------
use of a treatment technique in lieu of an MCL. A far-reaching provision of the law required EPA to
promulgate a regulation specifying criteria under which filtration was required as a treatment technique
for public water systems supplied by surface water sources. Similarly, EPA was directed to promulgate
regulations requiring disinfection as a treatment technique for all public water systems.

Passage of the 1996 Amendments to the SDWA focused the attention of water  utility managers and
public health and regulatory officials on source water protection and its role in protecting public water
supplies. Concern over source water protection was not limited to surface water supplies because many
ground water supplies proved to be vulnerable as well. Based on the 1996 Amendments, the States will
implement programs to decide if a system's source of water is threatened, as well as determine the
means to prevent pollution. Communities will be allowed to ask for state assistance, and a certain
percentage of the State Revolving Loan Fund has been earmarked to assist with source water protection
(Howell 1997).

The Standard Setting Process

The process of developing drinking water standards in the U.S. is complex. This  complexity is prima-
rily due to the need to integrate scientific knowledge with legal requirements and current societal val-
ues. The process flows from determining health risks of various contaminants, or risk assessment, to
developing regulatory control options, or risk management (Cox 1997).

The risk assessment process begins by reviewing all possible adverse effects of  a particular contami-
nant and determining which  effects are significant via drinking water. An analysis is then made of
carcinogenic and noncarcinogenic  effects. For carcinogens, a classification scheme based on strength
of evidence is used as well as quantitative risk extrapolation models. As a policy choice, the health goal
level for possible and probable human carcinogens is zero. For noncarcinogens and contaminants with
equivocal evidence of carcinogenicity, a safe nonzero level can be set as the health  goal. Once the
health goal is established, the risk  management process is used to determine the regulatory approach
and the feasible, enforceable level  for each contaminant.

The risk management process begins with an assessment of monitoring feasibility. If a particular con-
taminant can be monitored, then a MCL is set. If monitoring is not feasible, then the  law specifies a
treatment technique requirement. In either case, EPA must determine the feasibility of treating con-
taminated water to levels equaling  or approaching the health goal.

Costs must also be considered in terms of both individual systems applying appropriate technologies
and the total national costs of various regulatory options. In the final analysis, the protection of public
health, including an adequate margin of safety, is the predominant factor.

Trends in Compliance with the SDWA

The SDWA has been in existence long enough to monitor trends for compliance. In order to give a brief
picture of trends over the last ten years, compliance with the TTHM rule, nitrate regulations, the total
coliform rule, and the number of community water systems with any type of MCL violation  are dis-
cussed.

Compliance with the TTHM and Nitrate Rules

TTHMs form when disinfectants react with natural organic matter in water, and they may have poten-
tial chronic health effects. They tend to occur mostly  in surface waters. Nitrates have potential  acute
health effects  and occur mostly  in ground  water systems (which tend to be smaller). Because
                                            1-4

-------
trihalomethanes and nitrates are two of EPA's earliest regulated contaminants, tracking compliance
with these standards provides a general sense of public drinking water quality over time (USEPA 1999b).

The 1979 standard for TTHMs applies to approximately 3,500 community water systems (those serv-
ing at least 10,000 people). The number of community water systems with at least one violation of the
TTHM MCL in one year has been decreasing fairly steadily since the mid-1980s, going from a peak of
about 70 system (2 percent of the total number of systems that must comply) violations in 1985 to fewer
than 10 system violations in 1998. The number of community water systems with monitoring  and
reporting violations for TTHMs has also been decreasing fairly steadily, going from about 180 systems
in 1985 to about 70 in 1998.

The nitrate standard applies to all types and sizes of public water systems. The number of community
water systems with MCL violations for nitrate has been decreasing slightly since the mid-1980s, going
from a peak of about 340 system violations in 1985 to approximately 190 system violations in 1998. As
with TTHMs, the peak number of systems with reported violations represents  a small fraction—less
than one percent—of the total systems which must comply with the nitrate MCL.

Total Coliform Rule (TCR)

The Total Coliform Rule became effective in December 1990, although a less stringent standard for total
coliform existed (in combination with a turbidity level standard) as one of the interim regulations under
the original SDWA.  Community water systems with total coliform violations have accounted for the vast
majority of community water systems with MCL violations each year. Monitoring is required more  fre-
quently for total coliform, thus creating more opportunities for detecting MCL violations (USEPA 1999b).

The number of systems with total coliform MCL violations has decreased fairly  steadily since 1980, at
a rate of about 200 systems per year. Since 1980, over 80 percent of all community water systems with
any MCL violation had a violation for total coliform. However, even the peak  number of systems
violating the total coliform MCL (approximately  7,000 systems in 1980) represents only about 13
percent of the total number of community water systems that must comply with the standard.

The number of systems with MCL violations for total coliform did not increase after the 1990 rule went
into effect. However, the population affected by community water systems with TCR MCL violations
more than doubled between 1990 and 1993, going from roughly 12.5 million people affected in 1990 to
28 million in 1993.  The population affected has declined  steadily by about 4 million people per year
since 1993 to about 8 million in 1998.

Compliance with the Surface Water Treatment Rule

The SWTR took effect in December  1990. The number  of community water systems identified as
violating the rule's treatment technique requirements increased from about 10 in 1991 to approxi-
mately 1,500 in 1994, then  dropped to just under 1,000  by  1998. When noncompliance was at its
highest, the number of systems violating the SWTR represented about 14 percent of the total number of
community water systems that must comply with the rule (USEPA  1999b).

The population affected by these violations increased from about 140,000 people in 1991 to about 26
million in 1994. This affected population was higher than for any other contaminant or rule, with the
exception of the TCR in 1993. The population affected gradually decreased to about 18 million in 1998.

The number of systems with monitoring and reporting violations of the SWTR rose from about  120
systems in 1992 to a peak of approximately 600 systems in  1994 and has generally decreased since.  The
number of people served by systems with violations of monitoring and reporting requirements peaked
                                            1-5

-------
at 5 million people in 1994 and declined to about 2 million in 1997. The population affected then rose
to about 3.7 million in 1998.

One reason for the high number of systems with treatment technique violations as compared to moni-
toring and reporting violations is that many systems received treatment technique violations for failure
to filter. Because installing filtration is expensive, many large systems have needed more time than the
regulations allow to place filtration systems in service.

Effect of System Size and Compliance

Generally, larger systems have more resources available to comply with regulations, so fewer viola-
tions are incurred, despite the fact that larger systems must comply with more regulations than smaller
systems (USEPA 1999b). In recent years, it appears that the gap between the percentage of small,
medium, large and very large systems with violations has been closing. However, very small systems
are  still almost 50 percent more likely to incur violations than all other system sizes.

Current DBF and Microbial Regulations

Chapter 2, "A Review of Federal Drinking Water Regulations in the U.S.," provides an overview of the
history of the SDWA in the U.S. He reviews the SDWA and summarizes current MCLs and treatment
requirements. He also reviews the regulations that EPA is promulgating or considering for promulgation.

The 1986, SDWA amendments listed disinfectants and disinfection by-products (D/DBPs) among the
contaminants that EPA must regulate. Because of the difficult issues associated with this requirement,
EPA implemented the Negotiated Rulemaking Act of 1990. A Regulatory-Negotiation (Reg-Neg) com-
mittee with representatives from state and local agencies, public water systems (PWSs), elected offi-
cials, consumer groups, and environmental organizations met periodically from November 1992 through
June 1993 (Cox 1997). Based on the recommendation of the Reg-Neg committee and in response to a
wide range of technical comments from stakeholders and members of the public, EPA developed three
sets of rules to control microbial pathogens and DBFs.  These three rules are as follows: the Information
Collection Rule (ICR), a two-stage DBF rule, and a similarly staged Enhanced Surface Water Treat-
ment Rule (ESWTR) (USEPA 1998a, 1998b, 1998c, 1999a).

The 1996 Amendments to the SDWA required EPA to establish a Contaminant Candidate List (CCL)
for  future regulatory action. EPAs ORD will identify emerging pathogens and chemicals of public
health concern and assess the nature and magnitude of health effects associated with these waterborne
agents (USEPA 1998c).

There are many other requirements that follow from the specific details of the rules and regulations that
have been promulgated or may be promulgated under the SDWA for both DBF and microbial contami-
nant control. EPAs ORD has developed an agenda that is targeted to finding solutions to these prob-
lems as discussed in  the following sections.

Chemistry of DBF Formation

In Chapter 3, "Disinfection By-Product (DBF) Chemistry: Formation and Determination," the authors
address some of the  major issues associated with DBF formation chemistry. They focus primarily on
EPA-sponsored or in-house research. The primary disinfectants used in the United States are chlorine,
chlorine dioxide, chloramines, ozone, and potassium permanganate. Some disinfectants are generally
more effective than  others and  some are more effective against specific organisms than others. For
example, chlorine, which is the most widely used disinfectant in the U.S., is very effective against
bacteria and viruses, but is relatively ineffective against parasites such as Cryptosporidium. Ozone is

                                            1-6

-------
the most generally effective disinfectant, but is so reactive that it would be difficult to maintain a
residual in a distribution system.

As has been discussed, the disadvantage of applying a disinfectant is that it may be powerful enough to
non-selectively react with substances in the water, other than microorganisms, to form DBFs. There are
thousands of DBFs, and they may be categorized into three major classes: inorganic by-products, or-
ganic oxidation by-products, and halogenated organic by-products. The health effects of some of these
by-products are of little concern, but some are suspected to be carcinogenic or have other health effects
and are therefore subject to regulation as discussed previously.

By-products form when the disinfectant reacts with organic or inorganic "precursor" material; how-
ever, natural organic matter (NOM) has probably received the most attention. Much of the engineering
focus on controlling DBFs is on either controlling the by-products themselves or removing the precur-
sor material to keep the by-products from forming.

Source Water Protection

Many of the drinking water utilities in the U.S. invest  a great deal of time, energy,  and capital in
developing mechanisms for protecting against the impact of sudden changes in influent water quality.
Some of these mechanisms include investment in excess capacity and development of emergency pro-
cedures (Miller 1989).

Chapter 4,  "Source Water Protection: Its Role in Controlling Disinfection By-Products (DBFs) and
Microbial Contaminants," explores Source Water Protection (SWP) as it relates to the control of DBFs
and microbial contamination. They discuss the nature of threats to source water quality; techniques and
methods for monitoring and assessment of pathogens; technologies for control of water quality; the use
of models to assess water utility vulnerability; and the relationship of source water protection to water-
shed management.

Passage of  the 1996 amendments to the Safe Drinking Water Act  (SDWAA) focused the attention of
water utility managers, public health, and regulatory officials on SWP and its potential for protecting
public water supplies. Events such as the 1993 Cryptosporidiosis outbreak in Milwaukee, WI, rein-
forced the idea that water suppliers which meet all of the SWTR requirements of the SDWA are still
vulnerable to microbial breakthrough (Okun et  al. 1997; Fox and Lytle 1996). The Milwaukee experi-
ence demonstrated that water treatment and/or disinfection alone may not be enough to ensure the
provision of potable and safe water to the consumer.

Based on the 1996 amendments, the states will have to implement programs  to decide if a system's
source of supply is threatened as well  as determine the means to prevent pollution. Communities  will
be allowed  to ask for state assistance,  and a certain percentage of the State Revolving Loan Fund has
been earmarked to assist with source water protection activities (Howell  1987).

Although the SDWA was passed in 1974 and amended in 1986 and  1996, concerns about SWP actually
began with  the SDWAA of 1986. The 1986 amendments included provisions for "Protection of Ground
Water Sources of Water." The two programs set up under this requirement were the "Sole Source
Aquifer Demonstration Program," to establish demonstration programs to protect critical aquifer areas
from degradation; and the "Wellhead Protection Program," which requires states to develop programs
for protecting areas around public water supply wells to prevent contamination from residential, indus-
trial, and farming-use activities.

Managing microbial risk requires identification and quantification of organisms. The potential sources
of pathogens in source water are many and varied, including nonpoint runoff, discharges from treated
and untreated sewage, and combined sewer overflows.  From a waterborne outbreak and public health

                                            1-7

-------
viewpoint, both Giardia and Cryptosporidium are of primary concern. Monitoring regulations often
specify indicator organisms for determining water quality because the analytical methods are easier,
faster, and more cost effective than methods for specific organisms. The limitations of relying on indi-
cator organisms for determining the presence of pathogens include the occurrence of false positives
and the fact that indicator organisms measure bacteria that live not only in human enteric tracts, but
also in the enteric tracts of other animals (Toranzos and McFeters 1997).

Modeling can assist in identifying the vulnerability of a drinking water utility to threats from source
water contamination. These models can be used in assessing the impact of upstream point source dis-
charges on downstream users as well as the potential for contamination from nonpoint sources (Clark
et al.  1998). Another aspect of contamination modeling is the overland transport of pathogens. Al-
though efforts to model overland transport of Cryptosporidium oocysts have been limited, such models
are needed to predict oocyst loads and estimate the effectiveness of management practices.

Microbial Pathogen Disinfection

Chapter 5, "Disinfection," discusses over twenty years of EPAs research  and studies on microbial
pathogen inactivation. These pathogens, including bacteria, viral, and protozoan species,  comprise a
diverse group of organisms which serve as the etiological agents of waterborne disease. Although unit
processes, such as  coagulation, clarification, and filtration, may  dramatically reduce the number of
microbial pathogens, disinfection frequently serves as the final and, in some cases, the only barrier to
the entry of these organisms into finished water.

The disinfection process may be affected by physical and chemical factors such as temperature and pH,
as they are known to play an important role in the inactivation process for most commonly used disin-
fectants (Hoff 1986). Turbidity and particle protection influence disinfection efficiency as well as clump-
ing of individual microorganisms (Berman et al. 1988). Resistance to chemical disinfection may vary
greatly among the various microorganisms of interest and also between different life-stages of indi-
vidual species, such as is seen with bacterial endospores or encysted forms of protozoa.

Making comparisons among various studies of microbial inactivation are often difficult due to differ-
ences in methodology. Factors such as mixing, the type of bioassays employed to determine viability,
the volume of sample analyzed, and the reporting of residual versus initial dosing concentrations of the
disinfectant may vary greatly from one  study to another. Often these parameters  are not described in
sufficient detail in  scientific manuscripts to make a proper evaluation. Data collected under field or
pilot-scale conditions may show marked differences from the results of laboratory experiments con-
ducted under oxidant demand-free conditions. These discrepancies, along with the need to determine
the efficacy of disinfection for new and emerging waterborne pathogens, have been a major focus of the
EPAs research program on  microbial inactivation.  Rice discusses potable water disinfection,  as cat-
egorized  by individual oxidants, and summarizes the microbial inactivation research which has been
conducted or sponsored by EPA during the time period from 1980 to 1999.

Controlling Alternative DBFs

Chapter 6, "Alternative Disinfectants," discusses EPAs research devoted to characterizing and control-
ling DBFs. Recent  studies conducted by or funded by the EPAs ORD in Cincinnati that examine the
use of three alternative oxidants are presented: chloramine, chlorine dioxide, and ozone. As discussed
previously, chlorination of drinking water results in the formation of numerous DBFs, several of which
are regulated. Water systems seeking to meet MCLs  for regulated DBFs might consider various ap-
proaches to limiting DBFs such as: removing precursor compounds before the disinfectant is applied,
                                            1-8

-------
using less chlorine, using alternative disinfectants to chlorine, or removing DBFs after their formation.
Combinations of these approaches might also be considered. As mentioned previously, removing DBFs
after their formation is a method that is generally not considered but, no matter which approach is
selected, the effectiveness of the disinfection process must not be jeopardized.

Based on the research cited by Miltner, formation of DBFs by chloramines is significantly lower than
by free chlorine, with the exception of the formation of cyanogen chloride. Formation of non-haloge-
nated DBFs such as aldehydes and assimilable organic carbon  (AOC) is found to be minimal with
chloramination.

The formation of DBFs by chlorine dioxide (C1O2) is also significantly lower than with free chlorine.
Chlorine dioxide oxidizes DBF precursors in the treated water to the extent that lower concentrations
of DBFs are formed with subsequent chlorination. Using C1O2 results in the formation of non-haloge-
nated DBFs such as aldehydes, ketones, and AOC.

The use of C1O2 can result in the formation of chlorite and chlorate. Chlorite can be controlled by GAC
and through the use of reducing agents. Sulfite and metabisulfite can reduce chlorite, but may form
chlorate. Thiosulfate can reduce chlorite without forming chlorate. Ferrous ion can reduce chlorate, but
pH adjustment is required to minimize chlorate formation. The use of a reducing agent like thiosulfate
or ferrous ion can complicate the application of postdisinfectants.

Control of DBFs Using Biological Filtration

The potential for using biological filtration for controlling DBFs is discussed in Chapter 7, "Disinfec-
tion By-Product Control Through Biological Filtration." DBF control  through biofiltration is defined
as the removal of DBF precursor material  (PM) by bacteria attached to the filter media. Dissolved
organic matter (DOM), which is part of the PM, is utilized by the filter bacteria as a substrate for cell
maintenance, growth, and  replication. This effect makes  the PM utilized by bacteria unavailable to
react with chlorine to form DBFs. The prerequisite for maximizing bacterial substrate utilization in
filters is the absence of chlorine in the filter influent or backwash water.

Sand, anthracite, or GAC can be colonized by bacteria. Since anthracite and sand are considered inert
because neither interacts chemically with PM, removal of PM is due solely to biological activity. GAC
will initially remove DOM through adsorption and biological substrate utilization until its adsorptive
capacity has been exhausted. After that point, PM removal is achieved only through substrate utiliza-
tion, and the GAC is defined as biological activated carbon (BAG).  All drinking water filters will
become biologically active in the absence of applied disinfectant residuals. The process of biological
colonization and substrate utilization is enhanced by ozonating filter influent water.

Data collected, to date, indicate that biologically active filters remove  significant amounts of PM and
that preozonated biofilters  remove more PM than do non-ozonated filters. The resulting reductions in
trihalomethane formation potential (TFDVIFP) and haloacetic acid formation potential (HAAFP) should
help many drinking water utilities meet the 80 (TFDVI) and 60 (HAA) g/L limits mandated under the
Stage IID/DBP Rule.

Controlling Microbial Contaminants Using Filtration

An area of very active research for EPA has been the use of bench-, pilot-, and field-scale studies to
investigate various aspects  of surface water filtration. Many of these studies have been oriented toward
small drinking water system applications.  Chapter 8, "Microbiological Removal by Filtration Pro-
cesses," discusses research conducted by EPA which examines  various treatment techniques for re-
                                             1-9

-------
moving microorganisms from drinking water. Two of these technologies, Slow Sand Filtration (SSF)
and Diatomaceous Earth (DE) Filtration, are especially applicable to small systems. In addition, the
application of granular filtration, which is utilized by medium to large systems, has been investigated
for removal of Giardia and Cryptosporidium. These studies have considered the various operational
conditions that enhance removal efficiency as well as the conditions under which removal efficiency
deteriorates.

In the period between 1980 and 1990, Giardia  cysts and Cryptosporidium oocysts were the primary
organisms being considered. The effect of various water quality conditions and particle and pathogen
loadings were evaluated. It was shown that, in combination with effective chemical addition and co-
agulation, these processes are capable of efficiently removing high levels of both target organisms. It
was found that low water temperature reduced removal efficiency in SSF and conventional filtration,
but had little effect on DE performance. The studies conducted  by EPA concluded that good turbidity
reduction and good particle removal paralleled good microorganism removal.

EPA researchers were the first investigators to monitor filtration efficiency by measuring aerobic
endospore removal. Although not a direct surrogate for removal of a specific organism, they demon-
strated that endospore removal is an excellent measure of overall filtration performance. Their studies
have shown that endospore removal tracks particle and pathogen removal and turbidity reduction. En-
dospores were shown to be a conservative measure of both particle and pathogen removal.

Controlling DBFs and Microbes Using GAC and Membranes

Chapter 9, "Activated Carbon and Membrane Processes for Disinfection By-Product (DBF) and Mi-
crobial Control," provides a comprehensive review of activated  carbon and membrane research for the
control of DBFs and pathogens. Much of the work he cites was  conducted, or funded, by EPA's ORD.

GAC can be used as part of a multi-media filter to remove particulates (filter adsorber) or as a postfilter
to remove specific contaminants (postfilter adsorber). When used in a filter adsorber mode, the filters
are backwashed periodically to alleviate head loss, but the carbon itself is regenerated infrequently, if at
all. When used in a postfilter mode, the carbon bed is rarely backwashed and is regenerated as often as
needed to control for the contaminant(s) of interest.

Activated carbon in a filter adsorber application removes pathogens by the same mechanisms as any
other filter media. It does not remove particulates/pathogens to any greater degree than other filter
media types, so it is never recommended for particulate/pathogen removal alone.

Certain types of membranes can be very effective for controlling DBFs, while others are specifically
designed to remove particulates/pathogens. For example, reverse osmosis (RO) membranes are very
tight and are typically used to remove salts from seawater  and brackish waters. Due to their tight
membrane structure, they require high pressures to operate effectively.

Nanofiltration (NF) membranes are not as tight as RO membranes, but have been found to remove a large
percentage  of DBF precursors. Because they are not as tight, they can be operated at lower pressures
(typically 5 to 9  bar) than RO membranes while achieving the same, or greater, flux. Ultrafiltration (UF)
and microfiltration (MF) membranes are typically used for particulate/pathogen removal only.

Speth concludes that activated carbon is an effective process for removing DBF precursors, but is not
effective for pathogen removal. RO and NF are effective processes for removing DBF precursors, and
UF and MF membranes are excellent for removing pathogens and particulates and, under some condi-
tions, could be considered as a replacement for conventional treatment.
                                           1-10

-------
Removing DBF Precursors Using Enhanced Coagulation

Chapter 10, "Coagulation," discusses EPA's research on enhanced coagulation. He points out that co-
agulation has historically been used for the control of particulates in drinking water, while simulta-
neously controlling organic carbon. With the inclusion of DBF control in the Stage IDBPR, the role of
coagulation in organic carbon control was expanded to include the removal of DBF precursors. This
chapter presents recent studies conducted by the EPA that examined (1) conventional coagulation and
coagulation that was enhanced to better control organic carbon and DBF precursors, and (2) the effects
on other water quality parameters as enhanced coagulation was employed.

It is expected that many water systems will move from conventional to enhanced coagulation and
expand their coagulation objectives from removing turbidity to removing TOC as well. It is anticipated
that many systems will be able to meet  the requirements of enhanced coagulation for TOC removal
with only moderate changes in conventional coagulation. Conventional coagulation removes a greater
percentage of the humic fraction than of the non-humic fraction, but enhanced coagulation improves
the removal of both fractions. Making the change from conventional to TOC-optimized coagulation
generally results in improved removal of heterotrophic plate count (HPC) bacteria, total coliform (TC)
bacteria, C.parvum oocysts, Cryptosporidium  oocyst-sized particles, Giardia cyst-sized particles, total
plate count (TPC) organisms, and bacterial endospores.

A general concern associated with coagulation is that, although it lowers the concentrations of DBF
precursor, it shifts the distribution of the DBFs formed by chlorination toward the more brominated
species. This shift becomes more pronounced  with enhanced or optimized coagulation.

Systems switching from conventional to enhanced coagulation may achieve longer filter run times
(FRTs), but the tradeoff will  be greater amounts of sludge production. Systems practicing enhanced
coagulation should also consider pH adjustment ahead of the filter to achieve longer FRTs. One of the
concerns associated with  enhanced coagulation is that when alum is the  coagulant,  higher levels of
dissolved aluminum will enter the distribution system.

Controlling Microbes and DBFs  in Small Systems

In Chapter 11, "Controlling Disinfection By-Products (DBFs) and Microbial Contaminants in Small
Public Water Systems (PWSs)," the authors describe in-house and field research activities specifically
designed to evaluate alternative treatment technologies for small community and non-community wa-
ter systems. They discuss four major topics: (1) particulate removal, (2) disinfection/destruction, (3)
field-scale demonstration, and (4) small system remote monitoring and control.  Small systems have
many problems that make compliance with drinking water standards more difficult than for medium
and large systems. The pilot- and full-scale research efforts described in this chapter are intended to
address some of these needs. Because small systems often lack the financial, technical, and managerial
capabilities of larger systems and are responsible for the majority of the SDWA violations, they have
been targeted in several Federal Rules and Regulations.

EPA's in-house research has focused primarily on filtration and disinfection technologies that are consid-
ered to be viable alternatives to conventional package plants (flocculation, coagulation, media filtration,
post-chlorination). Conventional package plants require a high level of operator skill to properly maintain
appropriate chemical dosage and flow rates, especially when used to treat surface water. These difficul-
ties, in conjunction with the other small system problems mentioned previously, have resulted in EPA
focusing its research on technologies that are easy to operate and maintain and produce minimal residuals.
Field demonstration projects have been used to characterize some of the problems that can occur for even
the best technology when conditions are not optimal. The remote monitoring and control research efforts


                                            1-11

-------
resulted from the fact that many rural systems are located in topographically difficult areas or separated by
large distances from other systems, thus precluding any consolidation or regionalization efforts. The soft-
ware and sensing systems developed as a product of this research will allow individual treatment units to
be monitored and operated from a central location. This approach has come to be known as "the electronic
circuit rider" concept. One technique that has promise for improving the effectiveness of systems in the
field is the use of supervisory control and data acquisition systems.

Results from this research indicate that microfiltration, ultrafiltration, and reverse osmosis systems are
effective technologies for the removal of pathogens while still being affordable for small systems.
Their conclusions are very similar to those of Speth. New disinfection technologies appear to provide
improvements over current systems in handling chemicals and consistency of performance. This is an
area undergoing rapid change. Many organisms are readily removed and inactivated in the laboratory,
but under field conditions the same effectiveness cannot be taken for granted.

A very important "spin-off from this research is the EPA Environmental Technology Verification (ETV)
Program for Drinking Water Treatment Systems. The goal of this program is to develop  performance
standards and protocols than can be used to evaluate the performance of small systems technologies. The
ETV works in partnership with the private sector in order to provide test results and peer-reviewed data in
accelerating the acceptance and use of new, improved, and cost-effective technologies.

Modeling Chlorine Residuals and DBF Formation

Chapter 12, "Modeling Chlorine Decay  and the Formation of Disinfection By-Products (DBFs) in
Drinking Water,"  reviews current and historical research efforts related to the development of models
for predicating the decay of disinfectants and the formation of DBFs. It focuses on chlorine as a disin-
fectant and emphasizes EPA's research.

In the U. S., chlorine has been the final  disinfectant most often used before drinking water is discharged
into a drinking water distribution system. It is added to provide a disinfection residual and to protect
against microbial contamination. Even treated drinking water exerts chlorine demand due to the reac-
tions with NOM and other constituents in water. Therefore, the disinfectant dose must be enough to
meet the inherent demand in the treated water to provide sufficient protection against microbial infec-
tion, and at the same time minimize exposure to DBFs.

The conditions that govern the interaction of NOM and chlorine and the resulting formation of DBFs
are discussed. Research devoted to models for chlorine decay and the formation of DBFs are reviewed.
The factors that affect exposure to DBFs  are examined, and EPA field research studies that have pro-
vided the basis for current research on chlorine decay and DBF formation are presented. The develop-
ment of EPANET, a state-of-the-art, public sector water quality/hydraulic model, is reviewed, along
with the evolution of numerical modeling techniques. The topic of storage tanks and their impact on
water quality and the associated public policy issues are also discussed.  Models that predict the forma-
tion of MX, a potentially carcinogenic compound, are discussed.

The chapter reviews both the EPA research in this area as well as research done outside the Agency.
Clearly, much progress has been made in  developing realistic models to support risk management
goals. Of particular note is  the application of models to field conditions in water utilities. There is,
however, much research left to be done before these models are truly predictive.

Distribution  System Water Quality

Virtually anywhere a surface comes into contact with the water in a distribution system,  one can find
biofilms. Biofilms are formed in distribution  system pipelines  when  microbial cells attach to pipe
                                            1-12

-------
surfaces and multiply to form a film or slime layer on the pipe. Probably within seconds of entering the
distribution system, large particles, including microorganisms, adsorb to the clean pipe surface. Some
microorganisms can adhere directly to the pipe surface via appendages that extend from the cell mem-
brane; other bacteria form a capsular material of extracellular poly saccharides (EPS), sometimes called
a glycocalyx, that anchors the bacteria to the pipe surface (Geldreich 1988). The organisms take advan-
tage of the macromolecules attached to the pipe surface for protection and nourishment. The water
flowing past carries nutrients (carbon-containing molecules, as well as other elements) that are essen-
tial for the organisms' survival and growth (USEPA 1992).

The U.S. EPA has conducted a great deal of research into various aspects of biofilms and their impact
on water quality. In Chapter  13, "Biofilms in Drinking Water Distribution Systems," Meckes dis-
cusses the conditions that lead to biofilm formation in drinking water distribution systems and out-
lines past studies and the research that has been undertaken by EPA and its investigators.  According
to Meckes, biofilms are complex and dynamic microenvironments, encompassing processes such as
metabolism, growth, and product formation, and finally detachment, erosion, or "sloughing" of the
biofilm from the surface. The rate of biofilm formation and its release into a distribution system can
be affected by many factors, including surface characteristics, availability of nutrients, and flow ve-
locities. Biofilms appear to grow until the surface layers begin to slough off into the water (Geldreich
and Rice 1987). The pieces of biofilm released into the water may continue to provide protection for
the organisms until they can colonize a new section of the distribution system.

Few organisms living in distribution  system biofilms pose a threat to the average consumer. Bacteria,
viruses,  fungi, protozoa, and other invertebrates have been isolated from drinking water biofilms
(USEPA  1992). The fact that such organisms are present within distribution  system biofilms shows
that, although water treatment is intended to remove all pathogenic (disease-causing) bacteria, treat-
ment does not produce a sterile water. In fact, some otherwise harmless organisms (opportunistic patho-
gens) may survive the treatment process and cause disease in individuals with low immunity or com-
promised immune systems.

Additional work on biofilms within distribution systems is currently underway. This work is designed to
further assess the effect of water quality parameters and system operations on biofilm densities. Other
research efforts are focused on identification of specific organisms within biofilms and determining the
effectiveness of disinfecting agents on these organisms. These efforts are being conducted to determine if
biofilm contributions to delivered water may require treatment modifications or amendments.

Cost of Controlling  DBFs  and Microbial Contaminants

In  Chapter 14, "Control  of Microbial Contaminants and Disinfection By-Products (DBFs): Cost and
Performance," the authors review the current status of disinfection practices in the U.S., the conditions
that cause the formation of DBFs, and discuss the various treatment techniques and associated costs for
both controlling DBFs and ensuring  microbial safety. Making direct comparisons  among the various
alternatives is difficult. For example, moving the point-of-disinfection (chlorination) would seem to be
the lowest cost option. Nanofiltration, although the most expensive technology for precursor removal,
has the advantage of removing other contaminants such as total dissolved solids and various inorganics.
Therefore, it might be used for achieving other treatment goals in addition to  removing DBF precur-
sors. For example, NF effectively removed microorganisms, thus serving as an alternative for chemical
disinfection. Although the cost of enhanced  coagulation was not evaluated, it could be very effective if
a utility is only slightly  out of compliance. However, in addition to increased coagulation costs, an
additional cost may be associated with sludge handling. Clearly, modifying the disinfection process is
the lowest cost option for controlling DBFs but, as noted, there are by-products and problems associ-
                                            1-13

-------
ated with the use of some of the alternate disinfectants. For example, chloramination is not as good a
disinfectant as chlorine, and ozone may enhance regrowth of some organisms.

Retrofitting may be fairly easy with chloraminaton. For example, to switch from chlorine to chloramine
may only require the addition of ammonia feed equipment. However, the use of ozone will require
construction of expensive ozone contactors. The use of chlorine dioxide will probably require the use
of a reducing agent such as ferrous chloride, which was not included in this costing analysis.

The technologies discussed would normally be applied incrementally to a utility's existing treatment.
Therefore, the base cost associated with an assumed conventional treatment system and the incremen-
tal costs associated with various DBF control alternatives have been summarized. The unit processes
considered are those that are effective for precursor removal or for the use of alternate disinfectants.

More efficient treatment will be required to meet future regulations. Also, water treatment managers
will have to become more  knowledgeable about various treatment options that are cost-effective in
order for them to meet present and future regulations. Although essentially exempt in the past, small
water systems will be required to comply with future regulations.

EPA's Technology Research Program: Some Final Thoughts

After chloroform and other potentially harmful components were determined to be by-products of
disinfection, EPA mounted a major research effort to provide the scientific basis for identifying and
controlling these by-products. The Drinking Water Research Division of EPA's Municipal Environ-
mental Research Laboratory issued a report in 1981 titled "Treatment Techniques for Controlling
Trihalomethanes in Drinking Water" (Symons et al. 1981). This document summarized the technology
research which had been completed after the passage of the SDWA from 1974 through 1980. Perhaps
the  weakest aspect of this research effort was the minimal effort devoted to controlling microbial con-
taminants. As the importance of striking the proper balance between controlling DBFs and microbes in
drinking water was recognized, EPA promulgated the SWTR and the TCR in 1990. This promulgation
coincided with the development of a more balanced research program focusing on both DBF and mi-
crobe control. Subsequent regulations have reinforced the need to maintain this balance and the impor-
tance of the role that technology will play in achieving these regulatory goals. This document is in-
tended to summarize the research that has been conducted in technology research by EPA since the
publication of the 1981 treatment technology document.

The current EPA technology research program is supporting research in the following areas:

    •  The chemistry of DBF formation
    •   Source water protection
    •  Microbial pathogen disinfection
    •  Control of by-products from alternative disinfectants
    •  Control of by-products using biological filtration
    •  Control of microbes by filtration
    •  Control of microbes and DBFs using GAC and membranes
    •  Removal  of DBF precursors using enhanced coagulation
    •   Small systems technology
    •  Modeling chlorine residuals and DBF formation
    •  Distribution system water quality
    •  Cost of control technology

EPA's current drinking water research program is more sophisticated than it was twenty years ago. For
example, when the treatment technology manual was published in 1981, it reported primarily on treat-


                                           1-14

-------
ment-oriented research. Twenty years later, the technology research program includes source water
protection, treatment technology, and distribution system studies. The research also reflects a concern
over balancing the risks of potential carcinogenic exposure against the risks from microbial infection.

As has been discussed, the requirements of the various rules and regulations being promulgated under
the SDWA require a high level of expertise in both science and engineering. EPA's ORD is clearly in a
position to understand and solve the problems associated with producing safe drinking water.

References

Bellar, T. A. and Lichtenberg, J. J. (1974). "Determining volatile organics at microgram-per-liter
   levels by gas chromatography." Journal of the American Water Works Association, 66, 739-744.

Berman, D., Rice  E. W., and Hofr, J.  C. (1988). "Inactivation of particle-associated  coliforms by
   chlorine and monochloranfne." Applied and EnvironmentalMicrobiology, 54(2), 507-512.

Blair, K. (1994). "Cryptosporidium and public health." Health Environment Digest, 8(8), 61-63,
   December.

Bull, R. J. (1993). "Toxicology of disinfectants and disinfection by-products." Safety of water disin-
   fection: Balancing chemical and microbial risks, G. F. Craun, ed., ILSI Press, International Life
   Sciences Institute, Washington, D.C., 239-256.

Clark, R. M., Ehreth, D. J., and Convery, J. J. (1991). "Water legislation in the US:  An overview of
   the Safe Drinking Water Act." Journal of Toxicology and Industrial Health, 7(516), 43-52.

Clark, R. M., Goodrich, J. A., Lykins, Jr., B. W., and Neal, J. R. (1998). "Evaluating the effects of
   upstream  dischargers on downstream water suppliers: A source water protection model." Journal
   of Water Supply Research and Technology-Aqua, 47(5), 1-8.

Clark, R. M., Rizzo, G. S., Belknap, J. A., and Cochrane, C. (1999). "Water quality and the replace-
   ment and repair of drinking water infrastructure: The Washington, D.C. case study." Journal of
   Water Supply Research and Technology-Aqua, 48(3), 106-114.

Cox, W. E. (1997). "Evolution of the Safe Drinking Water Act:  A search for effective quality assur-
   ance strategies and workable concepts of federalism." William and Mary Environmental Law and
   Policy Review, 21, 69-164.

Federal Register.  (1979). 44(No. 231; November 29), 68624-68707.

Fox, K. R. and Lytle, D. A. (1996). "Milwaukee's crypto outbreak investigation and recommenda-
   tions." Journal of the American Water Works Association, 88(9), 87-94, September.

Geldreich, E. E. (1988). "Chapter 3: Coliform noncompliance nightmares in water supply distribu-
   tion systems." Water quality: A realistic perspective. University of Michigan, College of Engi-
   neering; Michigan Water Pollution Control Association; Michigan Department of Public Health,
   Lansing, MI.

Geldreich, E. E. and Rice, E.W. (1987). "Occurrence, significance, and detection oiklebsiella in
   water systems" Journal of the American Water Works Association,  79(5), 74-79.

Hoff, J. C. (1986). "Inactivation of microbial agents by chemical disinfectants." EPA/600/286/06 7,
   Environmental Protection Agency.

Howell, M. L. (1997). "SDWA then and now." Journal of the American Water Works Association,
   89(3), 11.


                                            1-15

-------
Miller, R. (1989). "Why use GAC—A utility viewpoint." Design and use of granular activated
   carbon—Predicted aspects. American Water Works Association Research Foundation, Denver,
   CO, 15-21.

Okun, D. A, Craun, G. E, Edzwald, J. K., Gilbert, J. B., and Rose, J. B. (1997). "New York City: To
   filter or not to filter?" Journal of the American Water Works Association, 89(3), 62-74.

Rook, J. J. (1974). "Formation of haloforms during chlorination of natural waters." Water Treatment
   Examination, 23, 234.
Safe Drinking Water Act (SDWA).  (1974). Pub. L. No. 93-523, 88 Stat. 1660.

Symons, J. M., Stevens, A. A., Clark, R. M., Geldreich, E. E., Love, Jr., O. T., and DeMarco, J.
   (1981). "Treatment techniques for controlling trihalomethanes in drinking water." EPA-600/2-81-
   56, U.S. Environmental Protection Agency, Drinking Water Research Division, Municipal
   Environmental Research Laboratory, Office of Research and Development, Cincinnati, OH,
   September.
Toranzos, G. A. and McFeters, G. A. (1997). "Detection of indicator microorganisms in environmen-
   tal freshwaters and drinking waters." Manual of environmental microbiology, C. J. Hurst, G. R.
   Knudsen, M. J. Mclnerny, L.  D. Stetzenback, and M. V. Walter, eds., American Society for
   Microbiology, ASM Press, Washington, D.C.
U.S. Environmental Protection Agency (USEPA). (1979). "National interim primary  drinking water
   regulations: Final rule" Federal Register, 44, 68624.

USEPA. (1992). "Seminar publication: Control of biofilm growth in drinking water distribution
   systems." EPA/625/R-92/001, Office of Research and Development, Washington, D.C.
USEPA. (1997a). "Research plan for microbial pathogens and disinfection by-products in drinking
   water." EPA 600-R-97-122, Office of Research and Development, Washington D.C., December.

USEPA. (1997b). "New York City filtration avoidance determination." ; .
USEPA. (1989a). "National primary drinking water regulations: Filtration, disinfection; Turbidity,
   Giardia lamblia, viruses, Legionella, and heterotrophic bacteria: Final rule." Federal Register,
   54(124), 27486.

USEPA. (1989b). "National primary drinking water regulations: Total coliforms (including fecal
   coliforms andE. colt): Final rule." Federal Register, 54, 27544.

USEPA. (1998a). "National primary drinking water regulations: Disinfectants and disinfection by-
   products: Final rule." Federal Register, 63, 69390.

USEPA. (1998b). "National primary drinking water regulations: Interim enhanced  surface water
   treatment: Final rule" Federal Register, 63(241), 69478-69521.

USEPA. (1998c). "Announcement of the drinking water contaminant candidate list: Notice." Federal
   Register, 63(40), 10273-10287.

USEPA. (1999a). "Microbial and disinfection by-product rules simultaneous compliance guidance
   manual." EPA 815-R-99-015.

USEPA. (1999b). "25 years of the Safe Drinking Water Act: History and trends." EPA 816-R-99-007,
   Office of Water, Washington, D.C.
                                           1-16

-------
                                      CHAPTER 2

           A Review of Federal Drinking Water Regulations in the U.S.1

Introduction
A Brief History of Safe Drinking Water in the U.S.

Approximately 400 major cities and towns in the U. S. were served by public water systems (PWSs) when
Dr. John Snow, an English epidemiologist, proved a cholera outbreak in England was caused by contami-
nated water supplies in 1855. More than 2500 PWSs would be placed into service in the U. S. over the next
30 to 40 years before scientists and engineers devised ways to remove or inactivate waterborne pathogenic
microorganisms with filters or disinfectants (Taras  1981). At the same time these important treatment
processes were being integrated into PWSs, some states had independently set drinking water standards to
improve drinking water quality and public health (Cox 1997). Despite this progress, there was concern
that, as long as the implementation of drinking water standards was left to each state's discretion, there
was no guarantee that everyone in the U.S. would have access to safe drinking water.

National drinking water standards for bacteriological contaminants were developed by the Public Health
Service (PHS) in 1914 to regulate drinking water provided on common  carriers (trains, buses, and
ships) engaged in interstate commerce to help prevent the spread of disease across state lines (Cox
1997). Although individual states voluntarily applied these standards to many PWSs, there was no
uniformity in their application or effectiveness (Cox 1997). The standards were revised and expanded
to include chemicals, as well as microbiological contaminants, in 1925. Further revisions, including
changes that actually made the standards more applicable to PWSs, were made in 1942 and 1946. They
were revised once more in 1962, setting limits for a total of 28 substances (USDHEW 1969). The
application of the PHS Standards to PWSs remained voluntary, and all 50 states eventually adopted
them as drinking water regulations or guidelines for their PWSs (Larson 1989).

As the application of the PHS Standards became increasingly common among states, health officials
expected the incidence of waterborne disease outbreaks to decrease. However,  such was not the case.
Confidence in drinking water quality began to wane by the late 1960s. Despite efforts to keep drinking
water microbiologically safe, waterborne disease outbreaks continued to plague public health. More-
over, chemicals used in agriculture and industry began to appear in water supplies. Concerns about
these conditions were confirmed by a PHS water system survey conducted in 1969 that revealed more
than 60 percent of participating treatment facilities had major deficiencies (USDHEW 1970). These
findings prompted Federal lawmakers to reconsider the need for Congress to regulate PWSs.

The Safe Drinking Water Act

As popular support buoyed Federal activism in the environmental movement during the 1970s, Con-
gress found constitutional authority in the Commerce Clause to regulate public water systems with the
Safe Drinking Water Act (SDWA) (USDHEW 1970). The Act was signed into  law in 1974 "to assure
that water supply systems serving the public met minimum national standards for protection of public
health" (H.R. Rep. 1974) by giving the U.S. Environmental Protection Agency (EPA) the authority to
(1) establish Federal drinking water standards for protection against all harmful contaminants in every
 1 James Owens: ORD/NRMRL/WSWRD, AWBERC Mailstop 689, 26 West Martin Luther King Dr.:
Cincinnati, OH 45268, 513-569-7235, owens.jim@epa.gov.
                                             2-1

-------
U.S. PWS; (2) establish a joint Federal-State system that would assure compliance with these stan-
dards; and (3) protect underground sources of drinking water (Cox 1997).

In its original form, the SDWA was designed to control drinking water quality with two types of regu-
lations: enforceable national primary drinking water regulations (NPDWRs) to cover substances with
potentially adverse human health impacts (SDWA 1974a) (current NPDWRs are listed in Table 2-1)
and non-enforceable national secondary drinking water regulations (NPSDWRs), in the form of perfor-
mance standards that were to control substances adversely affecting human welfare, i.e., taste and odor
(SDWA 1974b). The original SDWA also included a sole-source aquifer protection program that was
established to ensure that Federally funded activities did not cause harm to certain aquifers (SDWA
1974c). The Act changed considerably over the next 25 years (Figure 2-1).

Though the original SDWA helped to improve drinking water quality, Congress realized that only a
fraction of the potentially harmful contaminants in PWSs were addressed by the Act. Congress in-
tended that EPA promulgate drinking water standards expeditiously until a more comprehensive set of
Federal rules was enacted to govern drinking water in the U.S. However, standard-setting under the
original terms of SDWA was slow (Cox 1997). Consequently, Congress enacted the first substantive
SDWA amendments in 1986 to increase the rate at which EPA regulated contaminants. The 1986 amend-
ments were very  ambitious, establishing standard-setting deadlines, requiring the promulgation of
enforceable maximum contaminant levels (MCLs), and non-enforceable maximum contaminant level
goals (MCLGs).2  In sum, the  1986 amendments required EPA to regulate 85 contaminants within 3
years: 9 within  12 months, at  least 40 more within 24 months, and the remainder within 36 months
(Cox 1997). The  1986 amendments also required EPA to list additional contaminants every 3 years
thereafter and directed EPA to promulgate rules requiring PWSs supplied by surface water or ground
water under the influence of surface water (GWUDI) to disinfect and/or filter with variance systems
meeting certain criteria (SDWA 1974d).

EPA, states, and PWSs attempted to comply with the formidable demands of the 1986 amendments, but
ultimately found the requirements to be impossible to meet. EPA was criticized for having an inflexible
regulatory schedule, burdening states with unfunded mandates, and failing to incorporate a cost-benefit
analysis in the formulation of drinking water regulations. Consequently, Congress substantially revised
the SDWA again in 1996 (Pontius 1999). Key points of the  1996 amendments included (1) a revocation
of the requirement that EPA regulate 25  contaminants every 3 years;3 (2) an increase in EPA's authority
to consider costs  and overall risk reduction when setting standards; (3) the establishment of a state
revolving loan program to help communities meet compliance costs; and (4) an expanded source water
protection program (Feiner 1997).
2 The SDWA, as amended in 1986, requires EPA to publish a MCLG for each contaminant which, in the
judgement of the EPA Administrator, "may have any adverse effect on the health of persons and which
is known or anticipated to occur in public water systems" (Section 1412[b][3][A]). MCLGs are to be
set at a level at which "no known or anticipated adverse effect on the health of persons occur and which
allows an adequate margin of safety" (Section 1412[b][4]). The Act also requires that, at the same time
EPA publishes an MCLG, which is a non-enforceable health goal, it also must publish an NPDWR that
specifies either an MCL or treatment technique (Sections 1401 [1] and 1412[a][3]). EPA is authorized
to promulgate a NPDWR "that requires the use of a treatment technique in lieu of establishing a MCL,"
if the Agency finds that "it is not economically or technologically feasible to ascertain the level of the
contaminant."
                                             2-2

-------
                        Evolution of Federal Drinking Water Regulations
SDWA NIPDWR     TTHM                86SDWAA    TCR    LCR   ICR    96SDWAA   IESWTR     LTSWTR
                                               SWTR                          DBPR1      DBPR2
1974    1976   1978   1980   1982    1984   1986   1988    1990    1992   1994   1996    1998   2000   2002   2004    2006

SDWA - Safe Drinking Water Act, enacted in 1974.
NIPDWR - National Interim Primary Drinking Water Regulations enacted between 1975 and 1976.
TTHM - Total Trihalomethane Rule, promulgated November 29, 1979; effective November 29,1980 for PWSs serving 75,000 persons;
      effective November 29, 1981 for PWSs serving 10,000 to 75,000 persons.
86SDWAA - Safe Drinking Water Act Amendments of 1986, enacted June 16, 1986.
TCR - Total Coliform Rule, promulgated June 29, 1989; effective December 31, 1990.
SWTR - Surface Water Treatment Rule, promulgated June 29,1989; effective December 31, 1990.
LCR - Lead and Copper Rule, promulgated June 7, 1991; effective December 7, 1992.
ICR - Information Collection Rule, promulgated May 14, 1996; effective June 18,1996.
96SDWAA - Safe Drinking Water Act Amendments of 1996, enacted August 6, 1996.
IESWTR - Interim Enhanced Surface Water Treatment Rule, promulgated December 16,1998; effective February 16, 1999.
DBPR1 - Stage 1 Disinfection By-Product Rule, promulgated December 16,1998; effective February 16, 1999.
LTSWTR - Long-term Surface Water Treatment Rule, scheduled for promulgation in May, 2002.
DBPR2 - Stage 2 Disinfection Byproduct Rule, scheduled for promulgation in May, 2002.

Figure 2-1. Time line depicting when major Federal legislation became or will become effective.


EPA Regulations Promulgated Under the SDWA

The original set of NPDWRs was based on the 28 PHS Standards issued in 1962. Standards were set for
(1) six organic chemicals; (2) ten inorganic chemicals; (3) turbidity; and (4) total coliform bacteria in
1975; (5) radionuclides in 1976; and (6) trihalomethanes (THMs)  (volatile organic compounds that
form when disinfectants react with natural organic matter in water) in 1979. These  contaminants were
initially regulated by interim standards that would be revised, as necessary, following a comprehensive
review by the National  Academy of Sciences  (NAS). In addition to  establishing MCLs for listed con-
taminants, regulations also required PWSs to meet monitoring, reporting, record keeping, and public
notification requirements (USEPA 1979).

EPA was  unable to make much progress in establishing additional drinking water standards until the
SDWA was amended in 1986. To meet the requirements of the!986 amendments and to decrease what
was considered an unreasonably high risk of waterborne illness under existing rules, EPA promulgated
the Total  Coliform Rule (TCR) and  Surface Water Treatment Rule (SWTR) in 1989. The TCR was
implemented to (1) revise the MCL and monitoring requirements for total coliform bacteria; (2) require
small systems collecting fewer than five samples/month to have a periodic sanitary survey; (3) have
3The 1996 amendments have relaxed the  rulemaking process. EPA is currently required to list
contaminants that could potentially be regulated every 5 years. The Agency must publish the list after
consultation with the scientific community,  solicitation of public comment, and consideration of an
occurrence database. Then EPA shall use formal rulemaking to determine whether or not to regulate at
least five of the listed contaminants, beginning not later than 5 years after enactment and every 5 years
thereafter. Formal rulemaking generally requires Federal agencies to (a) state the time, place and nature
of the rulemaking (including the legal  authority for the proposed rule); (b) give interested parties an
opportunity to comment on the pending proposed rule; and (c) formulate a statement of the agency's
basis and purpose of the rule after the notice and comments.
                                                2-3

-------
         Table 2-1. Current NPDWRs
         Contaminants
MCLG1      MCL2or
(mg/L)4      TT3(mg/L)4
                Potential Health Effects from Ingestion of Water     Sources of Contaminant in Drinking Water
                                                                           Inorganic Chemicals
         Antimony
0.006
0.006
Increase in blood cholesterol; decrease in
blood glucose
Discharge from petroleum refineries; fire retardants;
ceramics; electronics; solder
         Arsenic
none5
0.05
Skin damage; circulatory system problems;
increased risk of cancer
Discharge from semiconductor manufacturing; petroleum
refining; wood preservatives; animal feed additives;
herbicides; erosion of natural deposits
         Asbestos (fiber
         10 micrometers)
7 million
fibers/L
7 million
fibers/L
Increased risk of developing benign intestinal polyps
Decay of asbestos cement in water mains; erosion of
natural deposits
         Barium
                             Increase in blood pressure
                                                                   Discharge of drilling wastes; discharge from metal
                                                                   refineries; erosion of natural deposits
         Beryllium
0.004
0.004
Intestinal lesions
Discharge from metal refineries and coal-burning
factories; discharge from electrical, aerospace, and
defense industries
to
         Cadmium
0.005
0.005
Kidney damage
Corrosion of galvanized pipes; erosion of natural
deposits; discharge from metal refineries; runoff from
waste batteries and paints
         Chromium (total)      0.1
             0.1
                Some people who use water containing chromium
                well in excess of the MCL over many years could
                experience allergic dermatitis
                                                   Discharge from steel and pulp mills; erosion of natural
                                                   deposits
         Copper
1.3
Action Level    Short term exposure: Gastrointestinal distress. Long
= 1.3; TT6       term exposure: Liver or kidney damage. Those with
                Wilson's Disease should consult their personal
                physician if their water systems exceed the copper
                action level
                                                    Source - reservoirs and plumbing
         Cyanide              0.2
         (as free cyanide)
             0.2
                Nerve damage or thyroid problems
                                                   Discharge from steel/metal factories; discharge from
                                                   plastic and fertilizer factories
         Fluoride
4.0
4.0
Bone disease (pain and tenderness of the bones);
Children may get mottled teeth
Water additive which promotes strong teeth; erosion of
natural deposits; discharge from fertilizer and aluminum
factories
         Lead
zero         Action Level    Infants and children: Delays in physical or mental
             =0.015; TT6     development. Adults: Kidney problems; high
                             blood pressure
                                                                   Corrosion of household plumbing systems; erosion of
                                                                   natural deposits

-------
Contaminants
Inorganic mercury
Nitrate
(measured
as nitrogen)
Nitrite
(measured
as nitrogen)
Selenium
Thallium
MCLG1
(mg/L)4
0.002
10
1
0.05
0.0005
MCL2or
TT3(mg/L)4
0.002
10
1
0.05
0.002
Potential Health Effects from Ingestion of Water
Kidney damage
"Blue baby syndrome" in infants under six months -
life threatening without immediate medical attention.
Symptoms: Infant looks blue and has shortness
of breath
"Blue baby syndrome" in infants under six months -
life threatening without immediate medical attention.
Symptoms: Infant looks blue and has shortness
of breath
Hair or fingernail loss; numbness in fingers or toes;
circulatory problems
Hair loss; changes in blood; kidney, intestine,
or liver problems
Sources of Contaminant in Drinking Water
Erosion of natural deposits; discharge from refineries and
factories; runoff from landfills and croplands
Runoff from fertilizer use; leaching from septic tanks,
sewage; erosion of natural deposits
Runoff from fertilizer use; leaching from septic tanks,
sewage; erosion of natural deposits
Discharge from petroleum refineries; erosion of natural
deposits; discharge from mines
Leaching from ore-processing sites; discharge from
electronics, glass, and pharmaceutical companies
Organic Chemicals
Acrylamide
Alachlor
Atrazine
Benzene
Benzo(a)pyrene
Carbofuran
Carbon tetrachloride
Chlordane
Chlorobenzene
zero
zero
0.003
zero
zero
0.04
zero
zero
0.1
TT7
0.002
0.003
0.005
0.0002
0.04
.005
0.002
0.1
Nervous system or blood problems; increased risk
of cancer
Eye, liver, kidney or spleen problems; anemia;
increased risk of cancer
Cardiovascular system problems; reproductive
difficulties
Anemia; decrease in blood platelets; increased
risk of cancer
Reproductive difficulties; increased risk of cancer
Problems with blood or nervous system; reproductive
difficulties.
Liver problems; increased risk of cancer
Liver or nervous system problems; increased
risk of cancer
Liver or kidney problems
Added to water during sewage/wastewater treatment
Runoff from herbicide used on row crops
Runoff from herbicide used on row crops
Discharge from factories; leaching from gas storage tanks
and landfills
Leaching from linings of water storage tanks and
distribution lines
Leaching of soil fumigant used on rice and alfalfa
Discharge from chemical plants and other industrial
activities
Residue of banned termiticide
Discharge from chemical and agricultural chemical factories

-------
to
Contaminants
2,4-D
Dalapon
l,2-Dibromo-3-
chloropropane
(DBCP)
o-Dichlorobenzene
p-Dichlorobenzene
1 ,2-Dichloroethane
1-1 -Dichloroethylene
cis-1, 2-
Dichloroethylene
trans- 1,2-
Dichloroethylene
Dichloromethane
1-2-
Dichloropropane
Di(2-ethylhexyl)
adipate
Di(2-ethylhexyl)
phthalate
Dinoseb
Dioxin
(2,3,7,8-TCDD)
Diquat
Endothall
Endrin
Epichlorohydrin
Ethylbenzene
Ethelyne dibromide
MCLG1
(mg/L)4
0.07
0.2
zero
0.6
0.075
zero
0.007
0.07
0.1
zero
zero
0.4
zero
0.007
zero
0.02
0.1
0.002
zero
0.7
zero
MCL2or
TT3(mg/L)4
0.07
0.2
0.0002
0.6
0.075
0.005
0.007
0.07
0.1
0.005
0.005
0.4
0.006
0.007
0.00000003
0.02
0.1
0.002
TT7
0.7
0.00005
Potential Health Effects from Ingestion of Water
Kidney, liver, or adrenal gland problems
Minor kidney changes
Reproductive difficulties; increased risk of cancer
Liver, kidney, or circulatory system problems
Anemia; liver, kidney or spleen damage;
changes in blood
Increased risk of cancer
Liver problems
Liver problems
Liver problems
Liver problems; increased risk of cancer
Increased risk of cancer
General toxic effects or reproductive difficulties
Reproductive difficulties; liver problems; increased
risk of cancer
Reproductive difficulties
Reproductive difficulties; increased risk of cancer
Cataracts
Stomach and intestinal problems
Nervous system effects
Stomach problems; reproductive difficulties;
increased risk of cancer
Liver or kidney problems
Stomach problems; reproductive difficulties;
increased risk of cancer
Sources of Contaminant in Drinking Water
Runoff from herbicide used on row crops
Runoff from herbicide used on rights of way
Runoff/leaching from soil fumigant used on soybeans,
cotton, pineapples, and orchards
Discharge from industrial chemical factories
Discharge from industrial chemical factories
Discharge from industrial chemical factories
Discharge from industrial chemical factories
Discharge from industrial chemical factories
Discharge from industrial chemical factories
Discharge from pharmaceutical and chemical factories
Discharge from industrial chemical factories
Leaching from PVC plumbing systems; discharge from
chemical factories
Discharge from rubber and chemical factories
Runoff from herbicide used on soybeans and vegetables
Emissions from waste incineration and other combustion;
discharge from chemical factories
Runoff from herbicide use
Runoff from herbicide use
Residue of banned insecticide
Discharge from industrial chemical factories; added to
water during treatment process
Discharge from petroleum refineries
Discharge from petroleum refineries

-------
to
Contaminants
Glyphosate
Heptachlor
Heptachlor epoxide
Hexachlorobenzene
Hexachloro-
cyclopentadiene
Lindane
Methoxychlor
Oxamyl (Vydate)
Polychlorinated
biphenyls (PCBs)
Pentachlorophenol
Picloram
Simazine
Styrene
Tetrachloroethylene
Toluene
Total trihalomethanes
(TTHMs)
Toxaphene
2,4,5-TP (Silvex)
1,2,4-Trichloro-
benzene
1,1,1 -Trichloroethane
1 , 1 ,2-Trichloroethane
Trichloroethylene
MCLG1
(mg/L)4
0.7
zero
zero
zero
0.05
0.0002
0.04
0.2
zero
zero
0.5
0.004
0.1
zero
1
none5
zero
0.05
0.07
0.20
0.003
zero
MCL2or
TT3(mg/L)4
0.7
0.0004
0.0002
0.001
0.05
0.0002
0.04
0.2
0.0005
0.001
0.5
0.004
0.1
0.005
1
0.10
0.003
0.05
0.07
0.2
0.005
0.005
Potential Health Effects from Ingestion of Water
Kidney problems; reproductive difficulties
Liver damage; increased risk of cancer
Liver damage; increased risk of cancer
Liver or kidney problems; reproductive difficulties;
increased risk of cancer
Kidney or stomach problems
Liver or kidney problems
Reproductive difficulties
Slight nervous system effects
Skin changes; thymus gland problems; immune
difficiencies; reproductive or nervous system
difficulties; increased risk of cancer
Liver or kidney problems; increased risk of cancer
Liver problems
Problems with blood
Liver, kidney, and circulatory problems
Liver problems; increased risk of cancer
Nervous system, kidney, or liver problems
Liver, kidney or central nervous system problems;
increased risk of cancer
Kidney, liver, or thyroid problems; increased
risk of cancer
Liver problems
Changes in adrenal glands
Liver, nervous system, or circulatory problems
Liver, kidney, or immune system problems
Liver problems; increased risk of cancer
Sources of Contaminant in Drinking Water
Runoff from herbicide use
Residue of banned termiticide
Breakdown of hepatachlor
Discharge from metal refineries and agricultural
chemical factories
Discharge from chemical factories
Runoff/leaching from insecticide used on catttle, lumber,
gardens
Runoff/leaching from insecticide used on fruits,
vegetables, alfalfa, and livestock
Runoff/leaching from insecticide used on apples,
potatoes, and tomatoes
Runoff from landfills; discharge of waste chemicals
Discharge from wood preserving factories
Herbicide runoff
Herbicide runoff
Discharge from rubber and plastic factories; leaching
from landfills
Discharge from factories and dry cleaners
Discharge from petroleum factories
Byproduct of drinking water disinfection
Runoff/leaching from insecticide used on cotton
and cattle
Residue of banned herbicide
Discharge from textile finishing factories
Discharge from metal degreasing sites and other factories
Discharge from industrial chemical factories
Discharge from petroleum refineries

-------
to
oo
Contaminants
Vinyl chloride
Xylenes (total)
MCLG1
(mg/L)4
zero
10
MCL2 or
TT3(mg/L)4
0.002
10
Potential Health Effects from Ingestion of Water
Increased risk of cancer
Nervous system damage
Sources of Contaminant in Drinking Water
Leaching from PVC pipes; discharge from plastic
factories
Discharge from petroleum factories; discharge from
chemical factories
Radionuclides
Beta particles and
photon emitters
Gross alpha
particle activity
Radium 226 and
radium 228
(combined)
Uranium
none5
none5
zero
zero
4 millirems
per year
15 picocuries
perL(pCi/L)
5pCi/L
30|j,g/L
Increased risk of cancer
Increased risk of cancer
Increased risk of cancer
Increased risk of cancer, kidney toxicity
Decay of natural and man-made deposits
Erosion of natural deposits
Erosion of natural deposits
Erosion of natural deposits
Microorganisms
Giardia lamblia
Heterotrophic
plate count
Legionella
Total coliforms
(including fecal
coliform and E. coli)
Turbidity
Viruses (enteric)
zero
N/A
zero
zero
N/A
zero
TT8
TT8
TT8
5.0%9
TT8
TT8
Giardiasis, a gastroenteric disease
HPC has no health effects, but can indicate how
effective treatment is at controlling microorganisms.
Legionnaire's Disease, commonly known as
pneumonia
Used as an indicator that other potentially harmful
bacteria may be presentlO
Turbidity has no health effects but can interfere with
disinfection and provide a medium for microbial
growth. It may indicate the presence of microbes.
Gastroenteric disease
Human and animal fecal waste
HPC bacteria have no known associated health risks;
HPC can be used to measure the treatment efficiency
Found naturally in water; multiplies in heating systems
Human and animal fecal waste used as an indicator for
the presence of potentially hazardous microorganisms
Soil runoff
Human and animal fecal waste
             *source of table: http://www.epa.gov/safewater/mcl.html
             1 Maximum Contaminant Level Goal (MCLG) - The maximum level of a contaminant in drinking water at which no known or anticipated adverse effect on the health
              effect of persons would occur, and which allows for an adequate margin of safety. MCLGs are non-enforceable public health goals.
             2 Maximum Contaminant Level (MCL) - The maximum permissible level of a contaminant in water which is delivered to any user of a public water system. MCLs are
              enforceable standards. The margins of safety in MCLGs ensure that exceeding the MCL slightly does not pose significant risk to public health.
             3 Treatment Technique (TT) - An enforceable procedure or level of technical performance which public water systems must follow to ensure control of a contaminant.
             4 Units are in milligrams per Liter (mg/L) unless otherwise noted.

-------
             5 MCLGs were not established before the 1986 amendments to the SDWA. Therefore, there is no MCLG for this contaminant.
             6 Lead and copper are regulated in a Treatment Technique which requires systems to take tap water samples at sites with lead pipes or copper pipes that have lead
              solder and/or are served by lead service lines. The action level, which triggers water systems into taking treatment steps if exceeded in more than 10% of tap water
              samples, for copper is 1.3 mg/L and for lead is 0.015mg/L.
             7 Each water system must certify,  in writing, to the state (using third-party or manufacturer's certification) that when acrylamide and epichlorohydrin are used in
              drinking water systems, the combination (or product) of dose and monomer level does not exceed the levels specified, as follows:
                 Acrylamide = 0.05% dosed at 1 mg/L (or equivalent)
                 Epichlorohydrin = 0.01% dosed at 20 mg/L (or equivalent)
             8 The Surface Water Treatment Rule requires systems using surface water or ground water under the direct influence of surface water to (1) disinfect their water, and
              (2) filter their water or meet criteria for avoiding filtration so that the following contaminants are controlled at the following levels:
                 Giardia lamblia: 99.9% killed/inactivated
                 Viruses: 99.99% killed/inactivated
                 Legionella: No limit, but EPA believes that if Giardia and viruses are inactivated, Legionella will also be controlled.
                 Turbidity:  At no time can turbidity (cloudiness of water) go above 5 NTU; systems that filter must ensure that the turbidity go no higher than 1 nephelometric
              turbidity unit (NTU) (0.5 NTU for conventional or direct filtration) in at least 95% of the daily samples in any month. HPC: No more than 500 bacterial colonies
              per milliliter.
             9 No more than 5.0% samples total coliform-positive in a month (for water systems that collect fewer than 40 routine samples per month, no more than one sample can
              be total coliform-positive). Every sample that has total coliforms must be analyzed for fecal conforms. There cannot be any fecal conforms.
            10 Fecal conform and E. coli are bacteria whose presence indicates that the water may be contaminated with human animal wastes. Microbes in these wastes can cause
              diarrhea, cramps, nausea, headaches, or other symptoms.
to

-------
states review sample-siting plans; and (4) require fecal coliform or E. coli testing (USEPA 1989a). The
SWTR established (1) MCLGs for Giardia lamblia, Legionella bacteria, and viruses; (2) disinfection re-
quirements; and (3) criteria under which filtration would be required, including limits on turbidity, and
procedures by which states are to determine which PWSs must filter source water (USEPA 1989b).

The 1986 amendments also set forth an aggressive plan to eliminate lead from PWSs by prohibiting its
use in the installation or repair of PWSs and plumbing that provides water for human consumption.
Furthermore, the amendments banned the sale of any drinking water cooler containing lead in interstate
commerce and declared drinking water coolers with lead-lined tanks imminently hazardous consumer
products that were to be repaired, replaced, or recalled. Ultimately, the 1986 amendments led to the
promulgation of the Lead and Copper Rule (LCR) in June of 1991. This rule established corrosion
control measures and source water treatment techniques for PWSs, and required PWSs to replace water
service lines containing lead with lead-free materials and inform consumers when lead concentrations
exceed action levels of 0.015 mg/L for lead and 1.3 mg/L for copper (USEPA 1991). EPA recently
made several minor revisions to the LCR to eliminate unnecessary requirements, streamline and reduce
the reporting burden, and promote consistent national implementation; this did not affect the lead or
copper MCLGs, the action levels, or the basic regulatory requirements of the rule (USEPA 2000).

The 1986 SDWA amendments also listed disinfectants and disinfection by-products (D/DBPs) among
the contaminants  that EPA must regulate. Considering the difficult issues associated with this re-
quirement, particularly the risk-risk paradox involving pathogenic microorganisms and the possible
toxicological impact of D/DBPs, EPA implemented the Negotiated Rulemaking Act of 1990 to pro-
vide stakeholders  in the drinking water industry with an opportunity to participate in the development
of a rule(s) that would balance health risks associated with waterborne pathogens and those associated
with D/DBPs. Thus, EPA initiated negotiated rulemaking among representatives from state and local
agencies, PWSs, elected officials, consumer groups, and environmental organizations to address this
public health issue. The Regulatory-Negotiation (Reg-Neg) Committee met from November 1992 through
June 1993 (Cox 1997).

Early in the process, negotiators agreed that a great deal of data would have to be collected before the
most effective plan for concurrently minimizing microbial and DBF risks could be developed. The
Reg-Neg Committee  concluded that, during the data collection period, EPA should issue a Stage I
Disinfection By-Product Rule (DBPR) in order to (1) reduce the current MCL for total trihalomethanes
(TTHMs); (2) regulate additional DBFs; (3) set limits for the use of disinfectants; and (4) reduce the
level of organic precursor compounds in the source water that may react with disinfectants to form
DBFs (USEPA 1998a). Among EPAs most significant concerns in developing regulations for D/DBPs
was the need to ensure that drinking water would be microbiologically safe at the limits set for D/
DBFs. Therefore, the Reg-Neg Committee considered a range of microbial issues and ultimately agreed
that EPA should also propose a companion microbial rule (USEPA 1998a).

Pursuant to the recommendations of the Reg-Neg Committee, in addition to a wide range of technical
comments from stakeholders and members of the public, EPA developed three sets of rules to control
microbial pathogens and D/DBPs: (1) the Information Collection Rule (ICR) to generate data required
to effectively regulate pathogens and DBFs; (2) a two-stage DBPR to minimize health risks attributed
to D/DBPs; (3) and a similarly staged Enhanced Surface Water Treatment Rule  (ESWTR) to maintain
or improve microbiological standards for drinking water as greater restrictions are placed on D/DBPs
(USEPA 1998b).

The ICR imposed  extensive monitoring requirements on certain categories of PWSs. The primary bur-
den for monitoring and testing under the  ICR falls on large  water systems—those serving at least
100,000 people from surface water sources and those serving at least 50,000 people from ground water
                                            2-10

-------
sources. The ICR focuses primarily on microbial contaminants and DBFs; however, some water
systems have also been required to generate data on alternative controls for DBFs and their precursors
(Cox 1997). The Stage I DBPR set maximum residual disinfectant level goals (MRDLGs) and maxi-
mum residual disinfectant levels (MRDLs) for chlorine, chloramines, and chlorine dioxide, in addition
to MCLGs and MCLs for four THMs (chloroform, bromodichloromethane, dibromochloromethane,
and bromoform), two haloacetic acids (HAAs) (dichloroacetic acid and trichloroacetic acid), bromate,
and chlorite (USEPA 1998a). Stage I DBPR also established monitoring, reporting, and public notifica-
tion requirements for these D/DBPs (USEPA 1998a).

A controversial issue concerning how chloroform is regulated under the Stage I DBPR has been subj ect
to judicial review. The Chlorine Chemistry Council criticized EPA for maintaining an MCLG of zero
for chloroform in the Stage I DBPR, despite the Agency's "repeated and unequivocal" affirmations that
the best available science supported a non-zero standard (Pontius 2000). In light of those findings, EPA
concluded that it could no longer defend its original decision to retain an MCLG of zero after the issue
was brought before the District of Columbia's U.S. Court of Appeals during the development of the
Stage II DBPR. Applying what was  considered a more appropriate, non-linear method for  assessing
chloroform health risks, EPA agreed to promulgate a non-zero MCLG using the best available peer-
reviewed science (Pontius 2000). EPA has scheduled promulgation of the Stage II DBPR for May of
2002 (USEPA 1999).

An Interim Enhanced Surface Water Treatment Rule (IESWTR) required filtered systems to ensure that
when filtration plants were deemed necessary to protect public health,  as specified in the SWTR, those
plants would afford sufficient protection against Cryptosporidium and other pathogenic microorgan-
isms (USEPA 1998b). It is important to note that  development of the IESWTR was based on the as-
sumption that all systems would fully comply with all SWTR requirements and adhere to filtration
avoidance criteria in the  SWTR. Thus, compliance with new  provisions in the IESWTR in no way
relieves a PWS of its obligation to comply fully with preexisting SWTR requirements (USEPA 1998b).

Key provisions of the IESWTR include an MCLG of zero for the protozoan genus Cryptosporidium;4
a requirement for PWSs that filter to achieve 2 log removal of  Cryptosporidium;5 more stringent per-
formance standards and individual filter requirements pertaining to turbidity;6 disinfection benchmarks
to keep PWSs from compromising microbial protection when they make system modifications to com-
ply  with DBF standards;7 the inclusion of Cryptosporidium in the definition  of GWUDI and in the
watershed control requirements for unfiltered public water systems; requirements for covers on new
finished water reservoirs; and requirements for states to conduct sanitary surveys for all surface water
and GWUDI systems, regardless of size (USEPA 1998b).8

EPA promulgated the first phase of the Long Term Enhanced Surface Water Treatment Rule (LTESWTR)
in November of 2000 to improve  control of microbial pathogens in drinking water, including
Cryptosporidium, for PWSs serving fewer than 10,000 people; prevent increases in microbial risk
4Genus rather than species because investigators have not been able to determine whether or not other
Cryptosporidium species are pathogenic in humans.
5Surface water systems serving 10,000 or more people, that are required to filter under the SWTR,
must achieve at least 2 log removal of Cryptosporidium. Systems that use conventional or direct filtration
meet this requirement if they comply with strengthened turbidity performance standards for combined
filter effluent and meet design and operating conditions as specified by states.
                                            2-11

-------
while PWSs serving fewer than 10,000 people control D/DBPs; and require certain PWSs to institute
changes to the return of recycle flows within the treatment process to help prevent recycle from com-
promising microbial controls (USEPA 1998b). Data from ICR and drinking water research will be used
to develop the second phase of the LTESWTR, scheduled for promulgation along with the Stage II
DBPR in May 2002 (USEPA 1999).

One of the major challenges in providing safe drinking water lies in adequately characterizing risks
associated with microbial pathogens and harmful DBFs and then reaching an appropriate balance of
those risks. The DBPR and the ESTWR have been crafted to achieve this goal. In accordance with their
development, the EPA's Office of Research  and  Development (ORD) Drinking Water Research
Program will provide a sound scientific basis for (1) the promulgation of the Stage II DBPR and the
LTESWTR by fiscal year (FY) 2002; (2) any revisions to these rules by FY 2004; and (3) a determina-
tion as to whether or not to propose  a Stage III DBPR by FY 2007. This will be accomplished by
employing the "Risk Assessment/Risk Management Paradigm," whereby research is conducted to first
better understand potential health risks of pathogens and  DBFs and then evaluate how technologies
may harmoniously mitigate those risks to provide safe, economical drinking water.

As EPA begins to use the Contaminant Candidate List (CCL) to protect public health against toxic or
pathogenic agents that have recently emerged as potentially harmful drinking water contaminants, ORD's
Drinking Water Research Program has made a concerted effort to provide a sound scientific basis for
(1) EPA to determine whether or not it should regulate at least five of the contaminants on the first CCL
(final list published March 2, 1998) by August 2001; (2) subsequent NPDWRs for any contaminants
selected for regulation by FY 2005; and (3) determining whether or not at least five contaminants on
the second CCL (required by 2003) should be regulated by FY 2006.

The CCL research process has been divided into two phases. During the first phase, an Implementation
Team will evaluate available data to determine if a CCL contaminant poses a public health hazard and
if the contaminant is treatable by current drinking water treatment practices. If the data set is inadequate
for making such an assessment,  the Team recommends that research be conducted to  provide more
6For all surface water or GWUDI systems that use conventional treatment or direct filtration, serve
10,000 or more people, and are required to filter: (a) The turbidity level of a system's combined filtered
water at each plant must be less than or equal 0.3 NTU in at least 95% of the measurements taken each
month; and (b) the turbidity level of a system's combined filtered water at each plant must at no time
exceed 1 nephelometric turbidity unit (NTU).
7The disinfection profiling requirement applies to surface water systems serving 10,000 or more people
and which have (1) measured TTHM levels of at least 80% of the MCL (0.064 mg/L); or (2) measured
HAAS levels of at least 80% of the MCL (0.048  mg/L), based on a 1-year running annual average of
representative samples taken in the distribution system. PWSs required to develop a disinfection profile
that subsequently decide to make a significant change in disinfection practice must consult with the
state prior to implementing such a change.
8Sanitary surveys are required no less frequently than every 3 years for community systems and no less
frequently than every 5 years for non-community systems. For community systems determined by the
state to have outstanding performance based on prior sanitary surveys, subsequent sanitary surveys
may be conducted no less frequently than every 5 years.
                                            2-12

-------
data. In the second phase, research needs for each contaminant identified as a potential hazard that is
difficult to treat will be prioritized based on the potential public health risk posed by the contaminant.
Once those determinations are made, human health risks and the risk management options will be
evaluated, in toto.

The CCL process will require several types of research. Methods for measuring or estimating the oc-
currence of CCL pathogens and chemicals in drinking water will be developed; their frequency of
occurrence and the concentration in source and finished waters, as well as water in distribution sys-
tems, will need to be determined;  and the extent to which human populations are exposed to CCL
contaminants will also have to be evaluated. Additional studies must be conducted to evaluate how
effective various treatment processes are in controlling or removing these contaminants from drinking
water. Other research will have to focus on how the quality of water may be maintained or improved in
distribution systems.

As research on specific contaminants is conducted, more comprehensive studies must be designed to:
explore how the  risks posed by pathogens in drinking water can be characterized; evaluate dose-re-
sponse relationship models; determine the impact of waterborne pathogens on human subpopulations;
and develop systems for characterizing the risks posed by exposure to specific and multiple or complex
mixtures of CCL compounds in drinking water.

At the  present time, "more than 90% of the [U.S.] population served by community water systems
receive water from systems with no reported violations of health-based standards. In the past decade,
the number of people served by public water systems meeting Federal health standards has increased by
more than 23 million. Although compliance with drinking water contaminant standards is good, public
health risks from drinking water can be further reduced" (Whitman 2001). EPAs Drinking Water Re-
search  Program is committed to making this possible—helping to achieve the goal of providing all
Americans with easy and affordable access to the safest drinking water possible.


References

Cox, W. E. (1997). "Evolution of the Safe Drinking Water Act: A search for effective quality assur-
   ance strategies and workable concepts of Federalism." William and Mary Environmental Law
   and Policy Review, 21, 69-164.

Feiner, W. P. (1997). "Just when you thought it was safe to go back in the water: A guide to comply-
   ing with the 1996 amendments to the Safe Drinking Water Act." Environmental Law, 4, 193-
   223.

H.R. Rep. (1974). No. 93-1185.

Larson, C. D.  (1989). "Historical development of the national primary drinking water regulations." Safe
   Drinking Water Act: Amendments, regulations and standards. E. J. Calabrese, C. E. Gilbert, andH.
   Pastides, eds., Lewis Publishers, Chelsea, MI, 3,10.

Pontius, F. W. (1999). History of the Safe Drinking Water Act. American Water Works Association,
   Denver, CO.

Pontius, F. W. (2000). "Chloroform: Science, policy,  and politics." Journal of the American Water
   Works Association, 92(5), 12.

Safe Drinking Water Act (SDWA). (1974a). Pub. L. No. 93-523, 88 Stat. 1660 (1974) §300g-l(e).

SDWA. (1974b). Pub. L. No.  93-523, 88 Stat. 1660 (1974) §300f(2).
                                             2-13

-------
SDWA. (1974c). Pub. L. No. 93-523, 88 Stat. 1660 (1974) §300h-3(e).
SDWA. (1974d). Pub. L. No. 93-523, 88 Stat. 1660 (1974) § 300g-l(b)(8).
SDWA. (1986). Pub. L. No. 93-523, 88 Stat. 1660 (1986).
Taras, M. J. (1981). The quest for pure water, 2nd Ed., Vol. 2, American Water Works Association,
   Denver, CO.
U.S. Department of Health, Education, and Welfare (USDHEW). (1969). "Public Health Service
   drinking water standards-1962." Public Health Service Publication No. 956, Reprinted Septem-
   ber 1969.
USDHEW. (1970). "Community water supply study: Analysis of national survey findings."
U.S. Environmental Protection Agency (USEPA). (1979). "National interim primary drinking water
   regulations; Final rule." Federal Register, 44, 68624.
USEPA. (1989a). "National primary drinking water regulations: Total coliforms (including fecal
   coliforms andE1. coli); Final rule." Federal Register, 54, 27544.
USEPA. (1989b). "National primary drinking water regulations: Filtration, disinfection; turbidity,
   Giardia lamblia, viruses, Legionella, and heterotrophic bacteria; Final rule." Federal Register,
   54(No. 124), 27486.
USEPA. (1991). "Maximum contaminant level goals and national primary drinking water regulations
   for lead and copper; Final rule." Federal Register, 56, 26460.
USEPA. (1998a). "National primary drinking water regulations: Disinfectants and disinfection
   byproducts; Final rule." Federal Register, 63, 69390.
USEPA. (1998b). "National primary drinking water regulations: Interim enhanced surface water
   treatment; Final rule." Federal Register, 63, 69478.
USEPA. (1999). "Microbial and disinfection byproduct rules simultaneous compliance guidance
   manual." EPA 815-R-99-015.
USEPA. (2000). "National primary drinking water regulations for lead and copper; Final rule."
   Federal Register, 65, 1950.
Whitman, C. T. (2001). Drinking Water Needs and Infrastructure. Prepared witness testimony
   delivered to the Subcommittee on Environment and Hazardous Materials, Committee on Energy
   and Commerce, March 28.
                                            2-14

-------
                                      CHAPTER 3

    Disinfection By-Product (DBP) Chemistry: Formation and Determination1

Introduction
The Need for Disinfection

In the mid 19th century, Chinese workers on the North American transcontinental railroad suffered less
illness than other groups. While generally mysterious at the time, today the reason is obvious. The
Chinese preference for tea required heating the water, thus killing many of the pathogenic microorgan-
isms. Today, the need to kill microorganisms in water is largely met through the addition of oxidizing
chemicals to the source water. The incidence of waterborne illness has decreased dramatically during
the 20th century, increasing human productivity and longevity. In addition to affecting the microorgan-
isms, however, the chemicals added to disinfect the water react with nonliving substances that occur
naturally in drinking water sources. These disinfection by-products (DBFs), some of which are carci-
nogenic, are the subject of human health concerns.

While the basic chemistry of disinfectants outlined in this chapter has been fairly well understood for
some time, the past  20 years have seen an incredible volume of scientific investigation into DBFs
resulting from the use of these substances. At the beginning of the 1980s, a great majority of the work
on DBFs was focused on the trihalom ethanes (THMs), and much of it was performed by U.S. Environ-
mental Protection Agency (EPA) Drinking Water Research facilities in support of the development of
regulation. As interest in the potential health effects of disinfection has dramatically increased, EPA's
direct contribution has become a smaller and smaller fraction of the work with each passing year. This
reflects not a lack of interest or effort on the part of EPA, but the growth in interest outside the Agency.
A perusal of university graduate schools shows the creation of environmental engineering departments
as well as divisions of environmental chemistry through this time period. EPA Offices solicit and fund
much research using contracts, cooperative agreements, and other vehicles. Most of the funding of
unsolicited research proposals is performed by the EPA Office of Research and Development's (ORD)
National Center for Environmental Research (NCER).  The  American Water Works Association Re-
search Foundation (AWWARF) is a research organization dedicated to the needs of water utilities and,
thanks to funding from EPA and AWWA members, has produced many results related to water utility
operation and disinfection practice.

This chapter addresses some of the major issues in DBP formation chemistry, but focuses mostly on
EPA-sponsored or in-house research. In addition to studies  that attempt to qualitatively identify by-
products, drinking water professionals have tried to understand the conditions that lead to the forma-
tion of DBFs and how these compounds are formed. In terms of monitoring and studying DBFs, it is
clear that monitoring DBP formation requires appropriate analytical tools. To meet this need, an entire
field of  analytical  chemistry has sprung up to support the study of DBP formation and regulation in
potable water.
'Matthew L. Magnuson, Edward T. Urbansky, Kathleen M. Schenck, and Michael S. Elovitz: ORD/
NRMRL/WSWRD, AWBERC Mailstop 681, 26 West Martin Luther King Drive, Cincinnati, OH
45268. Corresponding authors: Matthew L. Magnuson, 513-569-7321, magnuson.matthew@epa.gov
and Edward T. Urbansky, 513-569-7655, urbansky.edward@epa.gov.
                                              5-1

-------
Overview of Disinfection Issues
In the U.S., disinfection of drinking water is a common practice, although the choice of disinfectant
varies. These disinfectants have in common an ability to inactivate microorganisms. The disinfectants
destroy certain microscopic biochemical features of microorganisms, rendering them harmless to hu-
man health. Research into chemical treatment technologies has  focused on individual disinfectants,
although combinations of these disinfectants are often used. Table 3-1 lists the number of water sup-
plies in the U.S. by the type of disinfectants used. Attributes of these disinfectants will be discussed in
more detail in following sections of this chapter.
Table 3-1. Survey of Disinfectant Use (1997)
Type of Disinfectant
Number of Systems
Chlorine
Chlorine dioxide
Chloramines
Ozone
Potassium permanganate
22,307
313
135
30
1,122
The data in Table 3-1 are taken from a survey of disinfection practices published in 1997 (USEPA
1997). Of the disinfectants in Table 3-1, the use of ozone is increasing quickly, with 264 plants using
ozone as of May 1998 (Rice et al. 1998), primarily as a response to the regulatory requirements dis-
cussed in more detail in Chapter 2.
Table 3-2 lists  some  of the microorganisms targeted by disinfection practice and some of the more
appropriate disinfectants for each microorganism.
Table 3-2. Microorganisms and Disinfectants That Inactivate Them
Organism                        Chemical Disinfectant      Health Effects
Bacteria
   such as Legionella
   and coliform (Escherichia coli)
    Chlorine
    Chloramine
    Chlorine dioxide
    Ozone
Gastroenteric disease, Legionnaire's disease,
death
Giardia lamblia cysts
    Chlorine
    Chlorine dioxide
    Ozone
Gastroenteric disease, death
Cryptosporidium parvum oocysts
    Chlorine dioxide
    Ozone
Gastroenteric disease, death
Viruses
    Chlorine
    Chlorine dioxide
    Ozone
Gastroenteric disease, death
                                                3-2

-------
Like many technological improvements, disinfection has a downside. Namely, the disinfectants are
often so powerful that they nonselectively react with other substances in the water to form what are
known as DBFs. There  are three classes of DBFs listed in Table 3-3, which  also lists the residual
disinfectants, i.e., the forms of the disinfectant left in the water. There are actually thousands of DBFs,
and Table 3-3 lists some of the more common, more studied, and representative types. Some of the
detailed studies are discussed in following sections. The health effects of some of the compounds listed
in Table 3-3  (USEPA  1999a) have been investigated. Table 3-4 summarizes these health effects in
accordance with the classification scheme described by Table 3-5. Note that EPA is in the process of
revising the Cancer Guidelines (USEPA 1996) .
Table 3-3. List of DBFs and Disinfection Residuals

Disinfectant Residuals
Free chlorine
       Hypochlorous acid
       Hypochlorite ion
Chloramines
       Monochloramine
Chlorine dioxide
Inorganic By-Products
Chlorate ion
Chlorite ion
Bromate ion
Organic Oxygenated By-Products
Aldehydes3
       Formaldehyde (methanal)
       Acetaldehyde (ethanal)
       Glyoxal (ethanedial)
       Pyruvaldehyde (oxopropanal)
       Other aliphatic aldehydes
Carboxylic acids
       Acetic acid
       Other aliphatic monocarboxylic acids
       Oxalic (ethanedioic) acid
Ketoacidsa'b
       Glyoxylic (oxoethanoic) acid
       Pyruvic (oxopropanoic) acid
       Ketomalonic (oxopropanedioic) acid
Assimilable organic carbon
                                                      Halogenated Organic By-Products
                                                      Trihalomethanes
                                                             Chloroform
                                                             Bromodichloromethane
                                                             Dibromochloromethane
                                                             Bromoform
                                                      Haloacetic acidsb
                                                             Monochloroacetic acid
                                                             Dichloroacetic acid
                                                             Trichloroacetic acid
                                                             Monobromoacetic acid
                                                             Dibromoacetic acid
                                                      Haloacetonitriles
                                                             Dichloroacetonitrile
                                                             Bromochloroacetonitrile
                                                             Dibromoacetonitrile
                                                             Trichloroacetonitrile
                                                      Haloketones
                                                             1,1 -Dichloropropanone
                                                             1,1,1 -Trichloropropanone
                                                      Chlorophenols
                                                             2-Chlorophenol
                                                             2,4-Dichlorophenol
                                                             2,4,6-Trichlorophenol
                                                      Chloropicrin
                                                      Chloral hydrate
                                                      Cyanogen chloride
                                                      Organic chloramines
                                                      MX (3-Chloro-4-(dichloromethyl)-
                                                       5-hydroxy-2(5//)-furanone)

    a These carbonyl compounds are actually present as geminal diols even though their concentrations are reported in
     terms of the parent carbonyl compounds. See Urbansky 2000h for further explanation.
    b Although reported as acids, these species are actually present in water as the deprotonated anions.

-------
Table 3-4. Status of Health Information for Disinfectants and DBFs

Contaminant                          Cancer Classification
Chloroform                            B2
Bromodichloromethane                  B2
Dibromochloromethane                  C
Bromoform                            B2
Monochloroacetic acid
Dichloroacetic acid                      B2
Trichloroacetic acid                     C
Dichloroacetonitrile                     C
Bromochloroacetonitrile                  -
Dibromoacetonitrile                     C
Trichloroacetonitrile
1,1-Dichloropropanone
1,1,1 -Trichloropropanone
2-Chlorophenol                         D
2,4-Dichlorophenol                      D
2,4,6-Trichlorophenol                    B2
Chloropicrin
Chloral hydrate                         C
Cyanogen chloride
Formaldehyde                          BP
Chlorate
Chlorite                               D
Bromate                               B2
Hypochlorous acid
Hypochlorite
Monochloramine                        -
Chlorine dioxide                        D
  a Based on inhalation exposure.
Table 3-5. Scheme for Categorizing Chemicals According to Carcinogenic Potential
Group
A
B
C
D
E
Classification
Human carcinogen
Probable human carcinogen
Possible human carcinogen
Not classifiable
No evidence of human carcinogenicity
Definition
Sufficient evidence in epidemiologic studies to support
causal association between exposure and cancer.
Limited evidence in epidemiologic studies (Group Bl)
and/or sufficient evidence from animal studies (Group B2)
Limited evidence from animal studies and inadequate or
no data in humans
Inadequate or no human animal evidence of carcinogenicity
No evidence of carcinogenicity in at least two adequate
animal tests in different species or in adequate epidemiologic
and animal studies
                                                   3-4

-------
Because of concern over these DBFs over the past 25 years, some DBFs have been regulated and/or
subject to monitoring rules aimed at meeting the simultaneous goal of disinfecting water and control-
ling DBFs (USEPA 1999b). Table 3-6 lists these compounds along with important information about
them. It is important to remember that Table 3-6 is a small subset of Table 3-3, which itself is a subset
of the much larger list of substances sometimes identified as DBFs.

Regulatory issues were covered in more detail in Chapter 2, and a discussion of the Stage 1 DBF Rule
explains how the costs and benefits were utilized to  determine appropriate risk/exposure reduction
(Roberson et al. 1995). From a scientific standpoint, in chlorinated potable water supplies, two classes
of DBFs dominate the identifiable organic matter, the THMs and the haloacetates (haloacetic acids or
Table 3-6. National Primary Drinking Water Regulations Establishing Maximum Contaminant
          Levels (MCLs) and Maximum Contaminant Level Goals (MCLGs) Related to DBFs
Compound
Bromate
Bromodichloromethane
Bromoform
Chlorite
Chloroform
Dibromochloromethane
Dichloroacetic acid
MCLG
(mg/L)
Zero*
Zerob
Zero3
0.8a
Zero3
0.06a
Zero3
MCL
(mg/L)
0.010b
see
TTHMs
see
TTHMs
1.0b
see
TTHMs
see
TTHMs
see
HAAS
Potential Health
Effects
Cancer
Cancer, liver, kidney,
reproductive effects
Cancer, nervous system,
liver, kidney effects
Hemolytic anemia
Cancer, liver, kidney,
reproductive effects
Nervous system, liver,
kidney, reproductive
effects
Cancer and other effects
Sources of Drinking Water Contamination
Ozonation by-product
Drinking water chlorination and
chloramination by-product
Drinking water ozonation, chloramination,
and chlorination by -product
Chlorine dioxide disinfection by -product
Drinking water chlorination and
chloramination by -product
Drinking water chlorination and
chloramination by-product
Drinking water chlorination and
chloramination by-product
Haloacetic acids0
(HAAS)
Trichloroacetic acid
Total trihalomethanesd
(TTHMs)
                    N/A
                    N/A
0.060b    Cancer and other effects  Drinking water chlorination and
                              chloramination by-product
see       Possibly cancer and
HAA5    reproductive effects
 0.08b    Cancer and other effects
Drinking water chlorination
and chloramination by-product
Drinking water chlorination and
chloramination by-product
    Source: 63 Federal Register 69390
    a Finalized on December 16, 1998 (63 Federal Register 69390) as established in 40 CFR 141.53.
    b Finalized on December 16, 1998 (63 Federal Register 69390) as established in 40 CFR 141.64.
    0 HAAS is the sum of the concentrations of mono-,di-, and trichloroacetic acids and mono- and dibromoacetic acids
     expressed in mg/L.
    d Total Trihalomethanes are the sum of the concentrations of bromodichloromethane, dibromochloromethane, bro-
     moform, and chloroform expressed in mg/L.
HAAs) and hence are of regulatory interest. In Table 3-7, the THMs are a group of compounds with
three halogen atoms. Only the brominated and chlorinated ones are routinely found in potable water.
Occasionally, iodinated products are found, and fluorinated ones do not occur naturally and are not
formed during disinfection. The THMs are formed when individual carbon atoms are attacked by halo-
gen disinfectants. Small hydrocarbon chains are cleaved from natural organic matter (NOM)  mol-
                                                5-5

-------
ecules, and the reaction of the halogen species continues until THMs are formed. Small amounts of
tetrahalomethanes (carbon tetrahalides) are also formed in this fashion; however, THMs account for
some 20% of the halogenated organic carbon found after disinfection (Weinberg 1999).
Table 3-7. Trihalomethanes (THMs) Found in Potable Water
Name                          Formula
Trichloromethane (chloroform)        CHC13
Bromodichloromethane             CHBrCl2
Dibromochloromethane             CHBr2Cl
Tribromomethane (bromoform)        CHBr
HAAs are also formed during chlorination. These DBFs are listed in Table 3-8. Like the THMs, the
HAAs are also linked with increased incidence of cancer in laboratory animals (Xu et al. 1995; Herren-
Freund et al. 1987). Unlike the THMs, the HAAs are capable of dissociating in water. HAAs are >99%
ionized (deprotonated) to the haloacetate anions under drinking water conditions. However, they are
regulated and usually reported in terms of the parent acids rather than the carboxylate anions. HAAs
account for about 13% of the halogenated organic matter after disinfection (Weinberg 1999).

Table 3-8. Haloacetic acids (HAAs) Found in Potable Water
HAA
Chloroacetic
Dichloroacetic
Trichloroacetic
Bromoacetic
Dibromoacetic
Tribromoacetic
Bromochloroacetic
Bromodichloroacetic
Dibromochloroacetic
Formula
C1CH2CO2H
C12CHCO2H
C13CCO2H
BrCH2CO2H
Br2CHCO2H
Br3CCO2H
BrClCHCO2H
BrCl2CCO2H
Br2ClCCO2H
Grouping3
HAA5,6,9
HAA5,6,9
HAA5,6,9
HAA5,6,9
HAA5,6,9
HAA9
HAA6,9
HAA9
HAA9
   a HAAS concentrations (as the sum) are regulated under the Stage 1 DBF Rule. HAA6 data must be obtained and
     reported under the Information Collection Rule (ICR). HAA9 data are encouraged to be obtained and reported
     under the ICR, but not required.

Of the DBFs listed in Table 3-3, bromate is formed from the ozonation of source waters which contain
bromide. In ozonated water supplies, a variety of aldehydes and ketones abound as well as some car-
boxylic acids. In addition to these organic products, inorganic species are also found. These include
oxyanions of halogens, such as chlorite, chlorate, and bromate, which can be formed by a variety of
oxidizing disinfectants. Bromate is of particular interest since it is suspected of posing one of the
highest cancer risks of any DBF.

General Issues in Disinfection: Disinfectants and Source Material for DBFs

Many excellent reviews have been written (White 1999; USEPA 1999a) about the general chemistry of
the disinfectants in Table 3-1. The following sections discuss just a few of the relevant points of each.
The source material, with which the disinfectant may react to form DBFs, is also briefly discussed.
                                              3-6

-------
Disinfectants that Contain Chlorine: General Chemistry
Chlorine: Chlorine(I) and Chlorine(O) Compounds

Chlorine is the most widely used disinfectant in the U.S. It is U. S. practice that finished drinking water
leaves the treatment plant with a residual disinfectant. When surface water is used as the source for
drinking water, residual disinfectant is required by regulation. Therefore, chlorine is often added to
finished water, even if a different oxidant is used for primary disinfection. Chlorine is added to water in
a variety of forms, usually as a gas or in the solid hypochlorite form.

Chlorine Gas

Chlorine gas, properly  referred to as dichlorine (C12), is a greenish yellow gas that has a familiar and
pungent smell. Chlorine (oxidation state: 0) is modestly soluble in water. When added to water, chlo-
rine hydrolyzes, producing hypochlorous and hydrochloric acids:

                                 C12 + H2O^HOC1 + C1  +H+                      (3-1)

Hydrochloric acid is a strong acid and is completely dissociated into hydrogen and chloride  ions.
Hypochlorous acid (HOC1, chlorine oxidation state: +1) is a weak acid with a pKa of about 7.5, and it
dissociates into hydrogen and hypochlorite (OC1 ) ions:

                                      HOC1 ^ H+ + OC1                            (3-2)

It is believed that chlorine(O) and chlorine(I)  compounds work primarily by denaturing  enzymes or
proteins, thereby inactivating microorganisms. In some cases, physical disruption of cell  membranes
may also contribute. HOC1 is thought to be the more active species.

Hypochlorite

The equilibrium in Equation 3-1 can be driven forwards using strong base to deprotonate the hypochlo-
rous acid and to neutralize the hydrogen ion:
                               C12 + 2OH  ^OCl - + C1 +H2O                     (3-3)

When sodium hydroxide is used as the base,  the familiar sodium hypochlorite, found in household
bleach, is formed, which in turn undergoes the following reaction:

                              NaOCl + H2O ^ HOC1 + Na+ + OH                    (3-4)

Thus, the same active species, HOC1, is produced from both the reaction of chlorine gas and solid
hypochlorite.

Hypochlorous acid may also be produced by addition of solid calcium hypochlorite salt to water. The
choice of using chlorine gas or hypochlorite salts is a matter of preference by water utilities and is often
dictated by cost, safety concerns, and the availability of raw materials. The chemistry of chlorine has
practical considerations in this  regard: The chlorine(I)-cation transfer  step means that chlorine and
hypochlorous acid both undergo 2-electron reductions. If a reducing agent cannot offer 2 electrons,
reactions are generally slow or difficult. The 2-electron reduction can be expressed as follows:

                                HOCl + H+ + 2e ^ Cl +H2O                      (3-5)

                                       C12 + 2e- ^ 2C1-                            (3-6)
                                               5-7

-------
Chlorine(I) is unstable and disproportionates; thus, hypochlorite solutions are slowly converted to chlorate
and chloride, which are not disinfection by-products in the sense that no other reactant is required:

                                    3 CIO- ^ 2 Cl- + C1O3-                         (3-7)

Given enough time, solutions of sodium hypochlorite (e.g., chlorine laundry bleach) will be more than
99% converted to chlorate and chloride. Equilibrium is achieved faster at higher temperatures. Chlor-
ate is not a good disinfectant. Although the central chlorine atom  has a high oxidation  state (+V),
chlorate reacts much more slowly than hypochlorite and only in acidic conditions, which, in turn, reacts
more slowly than hypochlorous acid. This kinetic barrier precludes its use as an oxidizing disinfectant.
Unlike hypochlorous acid, which reacts  primarily by chlorine(I) cation transfers, chlorate must react
either by a reductant attacking the central chlorine atom or an oxygen atom transfer. Hypochlorite loss
via Equation 3-7 requires that a fresh supply of sodium hypochlorite solutions be available. As a rule,
most chlorination plants dissolve the chlorine in a small amount of water just before adding it to the
main stream, or they add the chlorine gas directly to the stream. Nonetheless, chlorate has been found
in these disinfection solutions (Bolyard et al. 1992; Bolyard et al. 1993). By contrast, C12 gas is stable
indefinitely if stored properly.

Chlorine Reaction with Inorganic Material

Chlorine and hypochlorous acid (or hypochlorite) react not only with organic matter, but with a number
of inorganic anions as well. In this way,  a number of inorganic by-products are  also produced. Chlo-
rine^) and chlorine(I) oxidize primarily by chlorine(I)-cation transfer. Although a net oxygen atom
transfer occurs, many reactions proceed  through the chlorine(I) transfer, followed by hydrolysis. For
example, nitrite is oxidized to nitrate as follows:

                                NO2  + HOC1 ^ C1NO2  + OH                       (3-8)

                               C1NO2 + H2O ^ NO3  + 2  H+ + Cl                     (3-9)

One beneficial reaction may occur when arsenic compounds, namely arsenite (As(III)), are present in
the source water. Reaction with chlorine oxidizes arsenite to arsenate, As(V), which is easier to remove
from the source water and is less toxic than arsenite:

                                  As(III) + chlorine ^ As(V)                      (3-10)


Chloramines

Another chlorine-containing disinfectant is chloramine, which is formed from the reaction of ammonia
with hypochlorous acid.

                                NH3  + HOC1 ^ NH2C1 + H2O                     (3-11)

The addition of the ammonia (NH3) ties up the "free" chlorine, available as HOC1.  It also slows down
undesirable reactions  of "free" chlorine which form DBFs. The chemistry of chloramines becomes
more complicated as  shown in  the following equations, in which the chloramine reacts with more
hypochlorous acid to tie up more chlorine.

                               NH2C1 + HOC1 ^ NHC12 + H2O                    (3-12)

                                NHC12 + HOC1 ^ NC13  + H2O                     (3-13)
                                              3-8

-------
                    o
 4    6     8    10    12
Chlorine Dose, mg CI2/mg N
14
                                                                       16
Figure 3-1. Speciation of free and combined chlorine species. When ammonia and chlorine are
reacted at various ratios, different concentrations of mono-, di-, and trichloramine are formed.
At C12/N (w/w) ratio of about 7, breakthrough occurs, producing NC13, which is not useful as a
disinfectant.

Together, the chloramines are referred to as combined chlorine. The equilibrium for the three reactions,
Equations 3-11 to 3-13, produces the distribution of species shown in Figure 3-1.

Figure 3-1 illustrates that, above a particular chlorine dose, the chlorine residual — and disinfection
ability — goes down almost to zero. In other words, the chlorine dose must be carefully controlled to
maintain a chlorine residual. If sufficient chlorine is added, another phenomenon known as breakpoint
chlorination occurs. In breakpoint chlorination, the nitrogen(-III) in ammonaceous (organic) and am-
moniacal (inorganic) species is oxidized to nitrogen(O). Superchlorination (shock treatment) of swim-
ming pools takes advantage of this phenomenon after organic amines and ammonia build up over the
winter. In  addition, the equilibrium is quite sensitive to the pH. Coupled with the breakthrough phe-
nomenon, the operation  of chloramine plants can be complicated because the pH and chlorine dose
must be carefully controlled. However, if used properly, chloramination is a tool for DBF control.

Chlorine Dioxide: a Chlorine(IV) Compound

The various oxidation states of chlorine make it useful in other disinfectants, such as chlorine dioxide
(C1O2), which is very much unlike chlorine and hypochlorous acid. This unusual oxide contains chlo-
rine in the +IV oxidation state. It is a moderately stable radical, C1O2», which does not undergo further
reaction with water after it dissolves. The mechanism by which chlorine dioxide reacts with most other
species is  believed to be a mixture of oxygen atom-transfer and electron-transfer steps. This allows
single-electron reductions along with multiple-electron pathway transfers:
                                    ClO.(aq) + e- ^ CIO.
                              CIO - + 2 FLO + 4e ^ Cl + 4OH
                              C1O3 + H2O + 2e ^ C1O2  + 2OH
                               CIO,  + 2FT + e  ^ CIO - + HO
               (3-14)

               (3-15)

               (3-16)

               (3-17)
Equations 3-14 to 3-17 illustrate how both chlorite (C1O2~) and chlorate (C1O3~) can be produced as a
result of the use of chlorine dioxide.
                                              3-9

-------
Disinfectants Not Containing Chlorine: General Chemistry
Potassium Permanganate

Nonchlorine-containing disinfectants require another species to act as oxidizing agent. In terms of
disinfection, this usually requires oxygen-containing species that are powerful enough to disrupt the
functioning of the microorganism. Potassium permanganate, KMnO4, is a weaker oxidant. It is used
mostly for iron and manganese removal, or taste and odor control, but can lower the level of DBF
precursor material. The reduction of Mnvn (oxidation state: +VII) to MnIV (oxidation state: +IV) pro-
vides the oxidation ability via the following half-reactions:
                             MnO - + 4H+ + 3e  ^ MnO9 + 2H O
                              MnOr + 8H+ + 5e ^Mn2+ + 4H O
                            MnO - + 2HO + 3e ^ MnO9 + 4OH
                          (3-18)

                          (3-19)
                          (3-20)
These reactions can oxidize some organic and inorganic DBF source material and inactivate some micro-
organisms, but are not powerful enough to inactivate many bacteria and protozoa. When used in conjunc-
tion with another oxidant, such as chlorine, potassium permanganate can be advantageous because it has
already oxidized some DBF source material. Thus, DBFs are controlled. If used improperly, the residual
permanganate anion renders the water pink. Potassium permanganate is a well-understood system and is
not the subject of much research investigation, as the by-products are regarded as innocuous.

Ozone

Like chlorine dioxide, ozone (O3) must be generated on site. However, this process is fairly straightfor-
ward and less susceptible to problems than chlorine dioxide. Ozone is generated by passing an electri-
cal discharge through dioxygen (O ):
                                      302(g)^203(g)
                          (3-21)
                 NH2Br
              NO3-
                  OH
                               NH3
DOM
                                             HOBr
       Brominated
       Organic Comp.
                                                             BrO-
                               Br
Figure 3-2. Scheme showing the production of bromate in ozonated systems.
                                             3-10

-------
Ozone is responsible for the familiar smell associated with lightning strikes. Ozone is a powerful
oxidant which engages in oxygen atom transfers. In addition to the direct action of O3 on living tissue,
ozone can cleave water molecules, producing hydroxyl radical (OH»), which also can act as a disin-
fectant. The contribution of each of the dual pathways, direct ozone and indirect hydroxyl, is highly
dependent on the source water quality because various chemicals, such as the ubiquitous carbonate,
tend to deactivate the hydroxyl pathways. The reaction of ozone with the bromide ion is important in
DBF formation,  and its complexities are illustrated in Figure 3-2.

Ozone and hydroxyl radical attack a variety of sites in organic molecules. Of particular interest is the
fact that ozone is far more effective than hypochlorite or chlorine for inactivating Cryptosporidium
oocysts. At the concentrations normally used for disinfecting drinking water, chlorination does not
affect cryptosporidians significantly, but ozone does. The reaction of ozone has a tendency to produce
many oxygenated compounds, such as carboxylic acids, aldehydes, and ketones, which are nutritious
compounds for microorganisms.

An Overview  of Disinfection By-Product Formation Source Material

The  source material for DBFs is important in understanding the chemistry and mechanism of DBF
formation, once the disinfectant reacts with the source material. Other chapters in this book deal with
the removal of this material to prevent DBF formation, and other facets of DBP/microbial issues relate
to the presence of source material.

Inorganic Sources

Source material for the formation of DBFs is inorganic and organic in nature. Inorganic components are
traced to various  minerals and other substances in the water derived from nonbiological sources. These
substances occur naturally in the water or may be anthropogenic in nature. One such naturally  occur-
ring  substance is the anion known as bromide, which is implicated in by-product formation, particu-
larly when used with ozone. Bromide in the water can also contribute, through a series of reactions, to
brominated products when chlorine is used. Bromide contamination in chlorine solutions is another
route through which bromide enters drinking water.

Natural Organic Matter (NOM)

Natural waters used as sources for drinking water supplies contain a variety of types of organic matter.
Some of this organic matter comes from natural sources. When organisms die, a mixture of biological
and chemical processes take place. These processes produce a mixture of compounds that are collec-
tively referred to  as NOM. NOM can be highly variable, depending on its source and extent of degrada-
tion. Many factors besides native flora and fauna influence NOM composition. These include tempera-
ture, rainfall/humidity, light, microbial populations, and geography. There is a complex interplay among
the native flora and fauna as well as the climate and season. There is much interest in understanding the
makeup of this material. The International Humic Substances Society (http://www.ihss.gatech.edu),
for instance, comprises  scientists interested in NOM.

A variety of schemes have been used to classify NOM. These categories are not mutually exclusive.
One  of the oldest and most respected (albeit generalized) methods is based on the solubility under
different pH conditions. Humic acid is the fraction of NOM in water not soluble at pH < 2, but soluble
at higher pH. Fulvic acid is soluble at all pHs. Humin is not soluble at any pH. When describing the
conjugate bases (e.g., the sodium  salts), the terms humate and fulvate,  respectively, are used.
                                              5-11

-------
Characterization of NOM

Typical soluble NOM has a molecular mass range of about 300 to 30,000 unified atomic mass units (or
daltons, Da). Common moieties include aromatic rings, alkyl chains, carboxylates, phenols, and other
alcohols. Polynuclear (polycyclic) aromatic compounds are not generally thought of as making up a
significant portion of NOM. A number of volumes have been dedicated to characterizing NOM (AWWA
1994; Barret and Krasner 2000; Minear and Amy 1996a; Owen et al. 1993; Croue et al. 1999).

Because NOM does not reflect a single compound or even a closely related group of compounds, it is
very difficult to characterize. Therefore, NOM is sometimes fractionated based on its physical proper-
ties, such as polarity, namely its relative retention on functionalized poly(styrene-divinylbenzene) res-
ins (e.g., Rohm & Haas XAD®). Other physical properties, such as ionizability, are also used. The U.S.
Geological Survey has developed elaborate techniques to fractionate NOM and characterize the indi-
vidual fractions. EPA currently is involved in multiple cooperative  efforts to relate NOM characteris-
tics to DBF formation.

Aside from fractionation, another avenue of NOM characterization is to study properties of the bulk
solution rather than individual chemical components. As a bulk source of organic carbon, NOM is
often measured in raw and finished water using total organic carbon (TOC) analyzers (Urbansky 2001).
Modern TOC analyzers convert the carbon in organic carbon compounds to carbon dioxide, which is
then measured with an infrared  detector. In  addition to TOC, which includes suspended particulate
matter, dissolved organic carbon (DOC) can also be reported. In practice, DOC is most often used, and
most TOC analyzers are more effective at determining DOC than TOC.

Techniques commonly used for characterization rely on identifying  individual functional groups, such
as amines, thiols, alcohols, carboxylates, and halides. In addition, NOM can be subjected to traditional
elemental analysis by combustion. Infrared spectroscopy is one of the instrumental techniques that can
assess some of the functional groups present since certain moieties are known to have distinct infrared
absorption bands that correspond to O-H stretch, C=O stretch, or other types of independent vibrations.
Nuclear magnetic resonance (NMR) spectroscopy is used to distinguish among aromatic, alkyl, and
alkenyl compounds. Relative contributions of these different types of carbon-carbon bonds  can be
estimated from the NMR spectra. Pyrolysis-GC/MS can fingerprint  NOM in terms of four biopolymer
groupings, namely, polysaccharides, proteins, aminosugars, and polyhydroxyaromatic compounds. The
complexity of the sample can produce difficulties in interpretation for whatever technique is used.

Factors Affecting DBF Formation from the Source Material

A number of factors in addition to the NOM composition  determine the composition of DBFs.  The
choice of oxidizing disinfectant is an obvious factor. The presence of other ions, such as bromide, can
have a profound impact on the nature  and distribution of the DBFs formed  during water treatment.
Temperature, pH, and oxidant dosing rates all can affect DBF formation. Hundreds or perhaps thou-
sands of papers have been written on small variations in conditions that affect DBF formation. A whole
series, Water Chlorination Volumes 1-6, edited by R.L. Jolley (Jolley  1976; Jolley et al. 1978, 1980,
1983, 1985, 1990) was dedicated to water chlorination chemistry. Several recent volumes have contin-
ued down this path (Symons 1997; Minear and Amy 1996b; Singer 1999).

More effort is focused on removing DBF precursors (i.e., NOM) (Shorney and Freeman 1999). Many
of EPA's surface water treatment rules emphasize this approach. The Stage 1 DBF Rule considers this
to be an important aspect because it is neither possible nor practical  to identify or monitor the plethora
of by-products that form during disinfection with oxidizing compounds. Certain classes of compounds
are monitored, but, to account for the many that cannot be, minimizing the amount of precursor mate-
rial is adjudged to be one of the best approaches.


                                             3-12

-------
EPA Research into DBF Formation and Chemistry
Measures of the Proclivity ofNOM To Form DBFs

By definition, NOM is a reducing agent. When an oxidant, such as chlorine or hypochlorous acid, is
exposed to NOM, a variety of oxidation-reduction reactions is possible. Every natural water has an
oxidant demand. For example, when chlorine is used, the chlorine demand is a measure of the ability of
dissolved organic matter to react with chlorine. Until the chlorine demand is satisfied, disinfection is a
compromise between the oxidant reacting with the NOM and the microorganism, so disinfection effi-
ciency decreases. Once the chlorine  demand is satisfied (essentially everything that can react with
chlorine has), additional chlorine goes to disinfection. As far as DBF formation is concerned, the chlo-
rine demand in and of itself is  not a measure of the tendency to form DBFs. Much of the chlorine added
to meet demand is reduced entirely to chloride rather than being incorporated into a halogenated by-
product.

To have some quantitative measure  of the proclivity of NOM to form DBFs, a test for the THM
formation potential or THMFP has been devised. The formation potential is determined by exposing
a raw (untreated) water sample to an excess of oxidizing disinfectant for a period of time at a specific
temperature. The change in THM concentration relative to time zero is the THMFP. The total concen-
tration of THMs at any time is expressible as

                   [CHX3]T = [CHC1J  + [CHBrClJ + [CHBr2Cl] + [CHBrJ        (3-22)

Thus, the THMFP(a) at time t = a is given by

                         THMFP(a) = [CHX3]T(r = a) - [CHX3]T(r = 0)             (3-23)

In practice, a quantity of oxidant is added to a fixed volume of water and an aliquot is drawn out at
defined time intervals. This aliquot is then analyzed to determine the concentrations of THMs in solu-
tion. The THMFP, expressed in concentration units, is an estimate of the maximal concentration of
DBFs that may be formed in  the presence of a large excess of oxidant. One of the problems with the
way the THMFP has been  applied is  that  the measurement conditions were not the same in different
investigations.  This makes it difficult to compare or contrast the values obtained. In order to standard-
ize the THMFP,  a set of uniform formation conditions (UFC) has been developed (Summers et al.
1996) under EPA sponsorship. These can be summarized as follows: pH = 8.0 ± 0.2 (borate buffer),
temperature = 20 ± 1°C, reaction time = 24 ± 1 hr, and active chlorine residual = 1.0 ± 0.4 mg L~! as C12
(28 joM), which is representative of routine operating conditions.  On the other hand, if a sample of
finished water with a typical  chlorine residual is monitored for THM concentration as a function of
time, this simulates the behavior of the water once it leaves the utility plant and makes its way into the
distribution system on its way to consumers. This procedure is referred to as a simulated  distribution
system (SDS)  THM test. In this case, it is possible for all the chlorine to be consumed, unlike the
THMFP test. Depending on the location, consumption rate, and water pipe size, treated or finished
water may linger for days in the distribution system.

Chlorination By-Products
Halogenation  of NOM

Halogenated (brominated and/or chlorinated) compounds are of greatest concern due to health effects
observed in laboratory animals. Total  organic halide (TOX), a concept largely developed/promoted by
EPA (Stevens 1984) is defined as the sum of the concentrations of all halogenated organic compounds.
The true value  of the TOX  concentration cannot be determined: the number and identities of the indi-
                                             > 1O
                                             5-13

-------
vidual halogenated compounds formed during disinfection are unknown. Therefore, in practice, the
TOX concentration is operationally defined with measurement by a TOX analyzer. TOX analyzers use
activated carbon to capture halogenated organic matter. The carbon is then combusted at about 800-
1000°C to convert all halogens to the hydrohalic acids (HX). The halide ion is then coulometrically
titrated with silver(I) and expressed as chloride. Halogenated organic matter that is not readily or strongly
adsorbed to activated carbon is routinely lost, negatively biasing the reported TOX value. Compounds
other than THMs and HAAs, such as 2,2,2-trichloroethanediol (chloral hydrate), haloacetonitriles, or
trichloronitromethane (chloropicrin), can also be found in chlorinated potable water supplies. Together,
the haloacetonitriles make up about 2% of the halogenated organic matter, and 2,2,2-trichloroethanediol
also makes up about 2% of the halogenated organic matter after  disinfection takes place (Weinberg
1999). These DBF species form regardless of the source of the NOM. It is believed that the same types
of structures are responsible for DBF formation on a molecular level. These structures are thought to be
duplicated throughtout NOM molecules regardless of the overall size of the molecule. This results in
fairly uniform distribution of baseline DBFs, such as THMs and HAAs when water is chlorinated.
Other by-products can also be formed.

Much of EPA's initial research focused directly on characterizing and exposing NOM to  oxidizing
disinfectants, especially active chlorine compounds. In this way, EPA identified a number of classes of
compounds that make up NOM and established procedures for extracting DBFs from solution using
XAD®  resins (Christman et al. 1980, 1983b). Because algae can be found  growing in finished water
reservoirs, concern over plant metabolic products led to studies in that area. Extracellular products
resulting from algal growth were shown to react with chlorine,  forming  chloroform in addition to
higher-molecular-mass (>1000 u) DBFs (Wachter and Andelman 1984). A number of chlorinated DBFs
were determined from the reaction with several NOM sources, including surface water and commercial
products isolated from soils (Seeger et al. 1984b, 1984b). XAD® resins were used to collect the DBFs,
which were  eluted with ethyl ether. Many chlorinated aromatic carboxylic acids were found by gas
chromatography-mass spectrometry (GC/MS), including some with ether linkages. Oxygenated DBFs
were also found, including some longer-chain carboxylic acids (Seeger et al. 1984a, 1984b). As should
be expected, chlorination of amino acids produced halonitriles; what was unexpected, perhaps, was the
formation of high levels of 2,2,2-trichloroethanediol (Trehy et al. 1986). Chlorination of NOM isolated
from a lake in North Carolina was demonstrated to produce a number of short-chain chlorinated car-
boxylic acids, including haloacetic  acids and some alkenyl species in addition  to THMs and 2,2,2-
trichloroethanediol (Christman et al. 1983a).  A variety of mutagenic compounds, including THMs,
HAAs, haloacetonitriles, and haloketones were demonstrated to form when NOM is chlorinated di-
rectly (Meier et al. 1985). The mutagenicity of some HAAs was  demonstrated by EPA (Meier et al.
1997). Accounting for the post-disinfection halogenated organic matter has  been continually problem-
atic. In general, studies have accounted for no more than 60% of the halogenated organic matter mea-
sured as TOX, and sometimes as little as 15% (Norwood et al. 1983). NOM was characterized by 13C
NMR to distinguish between aliphatic and aromatic portions as well as ultraviolet (UV) spectropho-
tometry (Reckhow et al. 1990). Chlorination of the NOM gave a mixture of DBFs, including several
HAAs and haloacetonitriles. This  study also attempted to link the various measurable characteristics of
the NOM (humate and fulvate) to  the distribution of DBFs. Another study (Fromme et al.  1995)
marginally linked the presence of biopolymeric groups quantitated by pyrolysis GC/MS with DBF
formation.

Other Sources of DBF Precursors

In addition to natural sources of NOM, anthropogenic (man-made)  sources of organic matter exist, too.
For example, water  treatment chemicals were shown to be a source of organic matter that led to the
                                             3-14

-------
formation of DBFs (Feige et al. 1980). The release of industrial chemicals and minerals is largely an
unknown contributor to DBF formation. In this case, the type of DBFs is highly site specific. Regulated
DBFs, on the other hand, tend to be formed regardless of source water.

Foodstuffs and, indeed, bodily fluids can also potentially be DBF precursors, considering that a quan-
tity of disinfectants are ingested. Because most tap water contains a chlorine residual, it is possible for
DBFs to form even after the water is consumed. As a model, when rats consumed sodium hypochlorite
(albeit at levels higher than would normally be found in potable water), THMs, HAAs,  and
haloacetonitriles were detected in both the gastric contents and the plasma (Mink et al. 1983). Oxidiz-
ing chlorine compounds can react with a variety of natural  compounds, including carboxylic acids
found in fruit juices. Such reactions have been shown to produce mutagenic organic compounds (Chang
et al. 1988). A recent study demonstrated that foods and beverages could provide an alternate exposure
route to DBFs (Raymer et al. 1999a, 1999b).

Influences on  and Mechanisms of DBF  Formation

As noted earlier, a number of factors can influence DBF  formation (Johnson et al. 1986). EPA has
funded or specifically worked on several of these. A significant advance in measuring the proclivity for
TFDVI formation was the establishment of the uniform formation conditions (Summers et al. 1996). The
location in the plant where chlorination occurs can affect DBF formation. Prechlorination is practiced
by many utility plants to oxidize iron(II) and manganese(II) as well as to minimize biological growth in
their agglutination-sedimentation facilities. However, agglutination-sedimentation removes a signifi-
cant fraction of NOM. Accordingly, prechlorination has been demonstrated to lead to additional DBF
formation (Solarik et al. 1997).

When waters contain bromide, chlorination produces a variety of brominated by-products. Bromide is
oxidized  by chlorine(I) to give  bromine(I). At drinking water pH, most chlorine(I) is in the form of
hypochlorite; however, hypobromite is a stronger base, and so the oxidation-reduction reaction is ac-
companied by hydrolysis:

                           CIO + Br + H2O  -^ Cl + HOBr + OH                 (3-24)

HOBr is  kinetically more labile than hypochlorous acid even though it is a weaker oxidant from a
thermodynamic standpoint. Thus, bromination reactions abound during chlorination. In this fashion, a
mixture of brominated, chlorinated, and bromochlorinated by-products are formed during disinfection.
Studies have attempted to evaluate the effect of bromide on the formation of mutagenic by-products;
for example, a study conducted with Jefferson Parish, LA, water considered the effect of bromide
(Coleman et al. 1992). Chlorination of source water containing bromide results in the formation of not
only chlorinated DBFs, but also brominated and bromochlorinated DBFs (Coleman et al. 1992). Other
studies have identified some of these brominated, chlorinated, and bromochlorinated by-products
(Caughran et al. 1999; Richardson  et al. 1999a).

The precise quantities of the specific brominated, chlorinated, and bromochlorinated by-products re-
quires further research. Some studies, however, have focused particularly on HAAs and THMs because
they are known to make up much of the identifiable DBFs and are the subject of regulation. As pH goes
down, the formation of brominated species increases (Pourmoghaddas et al. 1993; Pourmoghaddas
1991). This occurs because most reactions involving hypohalous acids proceed through a halogen(I)
cation transfer step (Equation 3-25). This elementary reaction proceeds faster in acidic solutions be-
cause a hydroxide leaving group is more favorable than an oxide leaving group (which would have to
be converted to hydroxide in water due to the leveling effect of the solvent).
                                             5-15

-------
                         RCH=CH2 + HOBr ^ [RCH-CH2Br]+ + OH             (3-25)*
    * In this case, bromine is shown adding to the less-substituted carbon atom. Regioselectivity of these reactions is a
     complicated subject and beyond the scope of this work.
The tendency to form brominated versus chlorinated species is also dependent on the DBF precursor
material (NOM). For example, some types of NOM tend to form brominated HAA species, while some
types of NOM tend to form chlorinated species (Magnuson and Kelty 2000).

In addition to more fundamental studies of chemical kinetics, attempts have been made to empirically
model DBF formation (Clark et al. 1996). Because NOM is an ill-defined material, it is not possible to
elucidate rigorously detailed reaction mechanisms. To help water utilities comply with the surface
water treatment rules and the disinfection by-product rules, the Office of Water has prepared a model-
ing program that can be used in conjunction with site-specific chemical and engineering data (USEPA
1992, 1994). More details on modeling developments may be found in Chapter 9.

Investigating DBFs with Genotoxicity Assays

The goal behind studying and regulating DBFs has been the protection of human health. There are
several measures of the effect of DBFs on human health. Some DBFs have been studied extensively
enough to be assigned a carcinogenicity rating (refer to Table 3-4). However, given the large number of
DBFs, many of which have not been identified, it has not been practical or economically possible to
study them all. Therefore, other measures of potential human health effects have been explored. One of
these is genotoxicity, which is a measure of the ability  of a substance to damage the genetic material of
an organism. The Ames Salmonella mutagenicity assay, which detects point mutations,  is one of the
most commonly used short-term tests for genotoxicity. It has been used extensively to detect the pres-
ence of genotoxicity in drinking water sample concentrates. There is substantial  evidence that most of
mutagenic activity in drinking water originates from the reaction of disinfectants, especially chlorine,
with the NOM present in source waters (Meier 1988). Because of the formation of mutagenic com-
pounds during disinfection, the Ames Salmonella assay has been used extensively to determine the
levels of mutagenicity in finished water concentrates from both chlorinated (Schenck Patterson and
Lykins 1993; DeMarini et al. 1995) and alternative disinfectants (Schenck Patterson and Lykins 1995;
Schenck Patterson et al. 1995; DeMarini et al. 1995) as well as wastewaters (Meier and Bishop 1985;
Doergeretal. 1992).

In addition to DBFs, source-specific contaminants from various industrial, agricultural, and municipal
sources may also contribute to the overall mutagenicity of some drinking waters. Mutagenic contami-
nants could be introduced during distribution by such things as leaching of mutagenic materials from
the inside of pipes or tanks.  Also, openings  in the distribution system may allow for the entry of con-
taminants from the outside. The level of mutagenicity  in a drinking water may also increase within the
distribution system, due to the continued formation of DBFs from the reaction of residual disinfectant
with organic matter in the water.

Mutagenic compounds have been concentrated from finished water by reverse osmosis and then sub-
jected to GC/MS  (Coleman et al. 1980). GC/MS was originally used to identify and quantify about
one-fourth of the TOX, including HAAs, haloacetonitriles, haloketones, and several other compounds
(Coleman et al. 1984).  GC/MS methods have been developed to measure mutagenic compounds in
studies where NOM was chlorinated directly (Meier et al. 1983; Meier and Bull 1984; Meier et al.
1985a; Stevens et al. 1989). These studies are ultimately aimed at providing a model for the formation
of mutagens during  chlorination of actual drinking water, i.e., to predict which mutagenic DBFs are
likely to be formed.
                                             3-16

-------
                           CI2CHv    Cl            CI2CHX    ,CI
                                         	^        c=c

                            HO' 'O- -°           °=\    NCO°H
                                                      H
                            Closed Form              Open Form
                                           MX
                         CI2CH     Cl             CI2CH     Cl
                         HOOC     COOH            HO"   °

                              ox-MX                   red-MX
                         CI2CH      COOH        CI2CH     COOH
                             \    /                  \    /
                              p—p                    c—c
                             /C~C\                  /~\
                         o=c      ci            o=c     ci
                              H                      OH
                               EMX                   ox-EMX
Figure 3-3. MX and its structural analogues (adapted from Richardson 1998a).

The discovery of the highly mutagenic compound, originally known only as Mutagen X (MX), prompted
considerable research in potable water. The genotoxic and toxic properties of MX and related com-
pounds have been reviewed elsewhere (Meier et al. 1990; Daniel et al. 1993). Many research papers
were subsequently devoted to assaying this species. Other studies were carried out to determine the
chemical properties of MX (Meier et al. 1987). MX is (Z)-2-chloro-3-(dichloromethyl)-4-oxobutenoic
acid. It engages in a cyclization equilibrium to  form  a chlorinated furanone (.R^-S-chloro^-
(dichloromethyl)-5-hydroxy-[5//]furan-2-one with the double bond still in the (Z)-configuration. Sev-
eral chlorinated furanones, including MX, were shown to form when NOM was chlorinated directly
(Meier et al. 1986). Figure 3-3 shows several of these forms. Despite small structural differences, MX
is by far the most mutagenic compound.

MX and related mutagenic compounds can also form when NOM is chloraminated (Kanniganati et al.
1992). MX has been found in U.S. potable water supplies (Munch et al. 1988). It can be recovered from
finished chlorinated water using XAD® resins (Schenck et al. 1990; Ringhand et al. 1988a, 1988b).
Moreover, stability studies suggested that MX could survive in the distribution system for days (Meier
etal. 1987).

Studies were made of MX and related compounds using GC with mass spectrometric and/or infrared
spectrophotometric detection; these studies helped to identify these species in drinking water matrices
(Collette et al.  1991). MX and related compounds have also been separated by liquid chromatography
(Meier et al. 1986). The studies on MX were reviewed, outlining its chemical, mutagenic, and toxico-
logical properties (Ringhand et al. 1989). Adverse effects on rats and mice were determined, but human
effects were not clear (Daniel et al. 1994). Later, it was determined that MX was substantially detoxi-
fied in vivo in rats and that very little was excreted in the urine (Meier et al. 1996). In addition, risk was
shown to be considerably lower than that from the  THMs because of the level of exposure. Further-
more, the animal studies used concentrations about 1000 times greater than those found in chlorinated
water (Melnick et al. 1997). Concentrations of MX in chlorinated water are in the low parts-per-trillion
range. By contrast, THM levels in the same waters are typically 1000 times higher.
                                              5-17

-------
Trace DBFs in Drinking Water

Aside from the regulated DBFs, there are hundreds and perhaps thousands of other compounds formed
from the reaction of disinfectant with substances in the water. In the strictest sense, products from the
reaction between oxidizing disinfectants and either NOM or naturally occurring inorganic constituents
are bona fide DBFs. On the other hand, some investigators classify all products formed from reactions
with substances in the raw water regardless of source (e.g., anthropogenic chemicals, microorganisms,
etc.) as DBFs. The observation of the plethora of chemicals  formed was made early on, and much
research went into trying to identify other DBFs, motivated by health concerns that trace levels may be
problematic for chemicals such as MX or bromate. The problem is that mass spectrometry, a powerful
method for identifying and quantifying DBFs, requires larger quantities of some DBFs than are natu-
rally formed in drinking waters. Therefore, early research used  concentrated solutions of NOM to
increase the amount of DBFs formed and provided early evidence  of the suspected link between NOM
in water and DBF formation in drinking water (Kopfler et al. 1984). Likewise, a library of DBFs was
built based on a natural water that contained an unusually high amount of NOM (Slocum et al. 1988).
In this manner, a library of over 780 DBFs was developed (Stevens et al. 1987), with particular regard
to the conditions required for formation. Because NOM differs greatly with source, later work was
aimed at concentrating the DBF formed from large volumes of water. Several methods for concentrat-
ing the water were  investigated and compared, and XAD® resins  were  determined to provide advan-
tages over Grob  closed loop stripping apparatus (CLSA)  and purge and  trap (Melton et al. 1981).
XAD® was used to study 580 compounds in several water supplies (Lin et al. 1981). Although initially
undertaken for mutagenicity studies, NOM extracts were  subjected to mass spectrometry and other
spectral techniques, resulting in the identification of hundreds of compounds (Richardson et al. 1994,
1996, 1999a, 1999b).

DBFs Formed from Alternative Disinfectants
DBF Formation from Alternative Disinfectants

Alternative disinfectants, namely disinfectants that are not chlorine gas or hypochlorite solutions, have
been under study for some time  in EPA, and they were the subject of an  early review (Stevens and
Symons 1984). The outside research community quickly picked up on DBF studies of alternative disin-
fectants. Within the EPA, the paradigm shift toward risk management (assessment and control)  meant
that more  emphasis was placed on the risks associated with the consumption of water rather than the
identification of all DBFs. To this end, several studies were performed to elucidate various issues that
were relevant to this effort.  One such issue involves ozone reaction pathways (ozone vs. hydroxyl),
which are fundamental to understanding how  to control the risks associated with ozone use, namely
bromate formation. Hydroxyl radicals form during ozonation; a  method was developed for rapidly
measuring hydroxyl radical concentrations (Ireland and Velinieks 1992). The modeling of ozone/hy-
droxyl radical behavior and the effect on ozonation was studied, and the  Rct concept was described
(Elovitz and von Gunten 2000), namely:

                                     Ra = [OH.]/[03]                           (3-26)

The formation of DBFs by these radicals was studied. In addition to ozonation, hydroxyl radicals are
made when titanium dioxide is exposed to UV light, electrons are promoted in energy. This allows
water to be cleaved to form hydroxyl radicals.  Thus, a number of  oxygenated DBFs were formed and
later identified by multispectral analysis (Richardson  et al.  1996).

Another research venture was preozonation, which, when  coupled with chlorination, can be used to
reduce DBF formation.  The ozone breaks down the NOM into smaller molecules and leaves fewer of
                                             3-18

-------
the highly reactive sites; thus, the chlorine has fewer places to react (Miltner et al. 1992). Ozone can
react with bromide to produce a variety of oxidized forms of bromine. These have been shown to react
with NOM to make bromohydrins (Collette et al. 1994; Cavanagh et al. 1992). Bicarbonate can affect
the efficacy of preozonation (Reckhow et al. 1986). Carbonate(l-) radical (CO3 •) formed by the action
of ozone on bicarbonate is a poor oxidant and would be expected to interfere in preozonation.

In order to better understand potential use of chloramine in reducing DBF formation, literature from the
period 1946 to 1984 was reviewed for THM formation from chlorine and chloramine, including in the
presence of bromide (Cooper et al. 1985). In summary, chloramine is a weaker oxidant than hypochlo-
rous acid from a thermodynamic standpoint. For this reason, it usually results in lower levels of DBF
formation, but it is not as good a disinfectant. The factors affecting DBF formation during chloramination
have been studied (Symons et al.  1998; EPA 1989). Cyanogen chloride is  one of the most recent
chloramination by-products to be identified and studied; it  can form when ozonated water is
chloraminated (Pedersen et al.  1999). This compound can be formed from the reaction of chloramine
and methanal (formaldehyde). The kinetics and mechanism of the reaction have been studied (Pedersen
et al. 1999). Methanal is ubiquitous from natural processes, but it can also be formed by the reaction of
hypochlorous acid with glycine, an amino acid that can be found in natural waters (Snyder and Margerum
1982).

DBF formation for chlorine dioxide was compared to that from ozone, chlorine, and chloramine (Koffskey
1993; Lykins et al. 1994). These studies found that no quick and  easy conclusion could be reached
regarding choice of disinfectant in terms of minimizing DBF formation, but that it was necessary to
strike a balance among competing needs. Chlorate formation from chlorine dioxide disinfection was
demonstrated when treated water is exposed to light, as is possible in coagulation-sedimentation basins
(Bolyardetal. 1993).

Analytical Chemistry of Alternative DBFs

Several analytical methods have been developed for chloraminated water. Purge and trap GC/MS was
used for cyanogen chloride analysis (Prakash et al. 1998), which compliments other methods of analy-
sis  of chloraminated water. Membrane introduction mass spectrometry was used to study the lifetime
of monochloramine in the human body. Human saliva and  stomach fluid  were examined for
monochloramine. Due to low time persistence, any toxic affects associated with chloramine were at-
tributed to DBFs rather than the disinfectant (Kotiaho et al. 1992).

Ozonation by-products have been identified using many of the same techniques and methods that work
for chlorination by-products (Richardson et al. 1999a, 1999b). EPA developed Method 556 to deter-
mine the aldehydes that form from ozonation (Munch et al. 1998a). The aldehydes make up a major
fraction of ozonation by-products. This was followed by the preparation of a user's guide to help labo-
ratories work around some known difficulties of the method (Munch et al. 1998b). Other by-products
form, too, such as carboxylic acids, including a number of 2-oxocarboxylates, commonly referred to as
ketoacids. A comparison of ion chromatography (1C) versus GC for the determination of the 2-
oxocarboxylates showed that the ion chromatographic method was more rugged and less susceptible to
problems during the analysis  compared to the double derivatization/GC experiment described (Urbansky
and Bashe 2000). The GC approach also suffered from interferences due to metal cations commonly
found in water supplies (Urbansky 2000d). As with  chlorination DBF formation studies, ozonation
DBF studies also require the use of a reducing agent to eliminate residual oxidant. Problems with a
variety of reagents were identified when applied to the determination of aldehydes (Urbansky et al.
2000a). It was later shown that indigo-5,5 ',7-trisulfonate and triphenylphosphine could be used as fast-
acting ozone-scavenging reagents (Urbansky et al. 2000b).
                                             5-19

-------
Many of the DBFs formed from ozonation experiments are highly polar in nature and therefore not
amenable to many conventional forms of analysis. The difficulty is that water in which the DBFs are
located is polar, and analytical techniques have difficulty separating the trace amounts of polar DBFs
out of the far more numerous polar water molecules (Weinberg 1999). These compounds have been
extracted from water through the use of solid phase microextraction (Shoemaker et al. 1999) or through
the use of derivatizing agents, which convert the polar molecules into less polar ones, which are easier
to extract. For example, aldehydes and ketones were analyzed following derivatization with 2,4-
dinitrophenylhydrazine (Guo et al. 1998).

The use of spectroscopy techniques in addition to mass spectrometry  has been used to help identify
DBFs. One of these is  infrared (IR) spectroscopy. This has been used in a number of studies with
chlorine and non-chlorine disinfectants. For instance,  multispectral analytical methods have been ap-
plied to determine DBFs in waters disinfected with chlorine and other disinfectants (Richardson et al.
1994,1995,1998a).  Multispectral techniques have also been applied to identify aldehydes (Richardson
et al. 1991). IR spectroscopy was a component of this multispectral analysis and is discussed in some
detail separately (Collette 1996).

Analytical Methods Development for Regulated DBFs

Mass spectrometry allows the study of molecules by, to put it colloquially, weighing them. To be more
precise, the mass/charge ration of ions resulting from the fragmentation of a molecule, as well as the
fragmentation pattern, is determined accurately. Mass spectrometry has long been the dominant means
to identify DBFs regardless of oxidizing agent. The quantification of DBFs through mass spectrometry
as well as other detectors forms the basis of many EPA methods to monitor regulated DBFs.

Analytical method development has taken an important role in EPA/ORD DBF strategy, since in order
to monitor, study, and regulate a DBF, a reliable method of analysis is necessary. Mass spectrometry is
often the recommended technique to identify and/or quantify DBFs, although other detectors are per-
missible. The use of mass spectrometry, because it produces such a definitive result, has gone far in
ensuring the quality  of data generated from compliance monitoring and risk management studies. En-
suring the quality is essential if decisions are to be based on those data. EPA has helped to define
practices for ensuring quality data (Budde and Eichelberger 1980; Boyd et al. 1996).

This effort has culminated in the development and promulgation of approved methods  of analyzing
DBFs in drinking water. Many of these methods can be used for determining regulated DBFs as well as
unregulated DBFs, which is useful for fundamental studies of these compounds. Table 3-9 summarizes
the methods for the regulated DBFs.

Table 3-9. EPA Methods for Regulatory Compliance Monitoring of Organic DBFs in
          Drinking Water
Method No.       Contaminant(s)
551             Halogenated hydrocarbons (including THMs), 2,2,2-trichloroethanediol, haloacetonitriles
502.2            THMs
524.2            THMs
552             HAA5 (see Table 2-2)
552.1            HAAS
552.2            HAA9
556             Aldehydes
300.x            Bromate, chlorite, chlorate
317.0            Bromate
321.8            Bromate
                                             3-20

-------
Trihalomethanes (THMs)

As shown by Table 3-9, there are often multiple methods for each DBF. Each method uses different
techniques and equipment because some compliance monitoring laboratories may be skilled in one
technique and/or may not have the equipment for another technique. Each method has been rigorously
evaluated to meet the requirements for compliance monitoring. These techniques are revised and up-
dated as new technology becomes available.

Closed loop stripping analysis, in which a large volume of water is effectively extracted into a small
volume of carbon disulfide, was used when DBF studies were first initiated. The carbon disulfide
would be injected into a gas chromatograph for detection with mass spectrometry or another suitable
detector (Coleman et al. 1981). With the development of purge and trap technology by EPA, analysis of
volatile DBFs was improved. Purge and trap methods are still effective and have been supplemented by
liquid-liquid microextraction techniques. The analysis of drinking water developments from 1996 through
1998 has been recently reviewed (Richardson 1999), in which the EPA developed many methods that
are not necessarily used in compliance monitoring, but are instead used for specific research purposes.

For the THMs, Methods 502.2 (Ho et al.  1995)  and 524.2 (Eichelberger 1995) are based on purge and
trap technology. In the purge and trap procedure, the water sample is placed in a specially designed
vessel and an inert gas is bubbled through the water sample through a frit, which causes the bubbles to
be small. The analytes (THMs) are purged by the inert gas and trapped on an adsorbent material. This
adsorbent material is then heated rapidly to release the analytes. A gas chromatograph separates the
mixture of analytes more or less by their volatilities and their abilities to partition into the stationary
phase of the column. In Method 502.2, the analytes are detected by photoionization and electrolytic
conductivity detectors. Detection is by elution time only and can be partially confirmed by the use of a
dissimilar chromatography column. For more reliable identification, a mass spectrometer is used  in
Method 524.2.

Method 551.0 was designed originally for only DBFs, but was later expanded into Method 551.1  to
determine a variety of pesticides and halogenated solvents encountered in drinking water (Hautman
and Munch 1997). Method 551.1 (Munch and Hautman 1995) extracts the water sample with an or-
ganic liquid. The analytes (THMs) are more soluble in the organic liquid than they are in the water,  so
a portion of the analyte molecules partition into the organic liquid. This organic liquid is then injected
into the gas chromatograph and is detected by an electron capture detector, which is very sensitive to
the chlorine and bromine atoms in the analytes. Qualitative confirmation of the identity of the analyte is
recommended by mass spectrometry.

Aside from these compliance monitoring regulatory methods,  EPA has developed alternative methods
to analyze THMs for special, i.e., research, purposes. For instance, to investigate more rapid analysis,
THMs were purged directly into an electrolytic conductivity detector (Hodakievic and Ho 1990). Treat-
ment studies often have special analytical needs that cannot be met using methods developed for regu-
latory compliance monitoring. In particular, DBF formation studies require that residual oxidants be
quenched to fix the DBF  concentrations in time.  The EPA method specifies ammonium chloride  or
sodium sulfite. Recently,  ascorbic acid has been  used for this purpose for HAAs, haloacetonitriles,
THMs, and other analytes of Methods 551.1A/B and 552.2 (Urbansky 1999; Urbansky et al. 2000c) as
well  as 502.2 analytes (Ho 1995). Bromochloroacetate possesses a chiral carbon atom; thus, some
work has focused on determining the enantiomer ratios (Wong et al.  1999).
                                              5-21

-------
Haloacetic Acids (HAAs)

HAAs are more difficult to determine than THMs, and the analytical chemistry has been recently re-
viewed elsewhere (Urbansky 2000e). This is a result of the acidic nature of these contaminants, which
causes them to not be amenable to  direct GC analysis like the THMs. To solve this problem, EPA
Method 552.0 (Hodgeson et al. 1988) provides for the analysis of 5 HAAs using diazomethane to
esterify the analytes after extraction into fert-butyl methyl ether. The methyl esters are then injected
into a GC and detected by electron capture. Advice for using this procedure was provided (Ulmer et al.
1988). Method 552.1 followed, replacing the diazomethane with acidified methanol. In Method 552.1,
the analytes were extracted by running the tap water through a solid phase anion exchange resin. The
current version of the method, Method 552.2 (Munch et  al. 1995b), eliminates the use of explosive
diazomethane, which is the most carcinogenic substance  known to man (on a base pair methylation
basis). Method 552.2 was designed with the preferred steps from both 552 and 552.1. Method 552.2
combines  an MTBE extraction with acidified methanol esterification (Pawlecki-Vonderheide et al.
1997). Method 552.2 was verified for all 9 HAAs. Although EPA promulgated Method 552.2 to moni-
tor HAA9 under the Information Collection Rule, many laboratories have continued to use Method
552. More care is necessary with Method 552 because diazomethane used in Method 552 degrades the
brominated trihaloacetic acids, especially in white light (Rubio et al. 2000). Following the  promulga-
tion of the Information Collection Rule, EPA attempted to discern how well labs were doing using
EPA-approved methods for DBF quantification (Stultz et al. 1998). The performance of Method 552.2
is dependent on both the specific water used and the skill of the analyst, particularly for the brominated
trihaloacetic acids. As an alternative, complexation electrospray mass spectrometry was recently used
to  determine HAA9 in drinking water. Because it does not have the acidic methanol step, problems
with the brominated trihaloacetic acids are reduced (Magnuson and Kelty 2000).

Inorganic DBFs: Bromate and Chlorite

Inorganic anions, e.g., bromate and chlorite, are produced  as DBFs. They have been determined using
ion chromatography originally developed in EPA Method 300.0 (Pfaff 1993). Bromate has attracted the
most attention due to higher possible health  risk. Several 1C methods have been developed for this
purpose (Hautman and Bolyard 1992a, 1992b,  1992c; Wagner et al. 1998). Lowering the detection
limit has been the goal of this research. Several concentration techniques were proposed (Sorrell and
Hautman 1993; Hautman 1993). EPA developed a method for bromate based on a chromophoric reac-
tion; this lowered the detection limit  substantially (Wagner et al. 1998), but the method can be affected
by impurities in the 3,3-dimethoxybenzidine used as a prochromophore (Urbansky and Brown 2000).
A GC/MS method has been developed for bromate; bromate is used to produce a volatile brominated
organic molecule (Magnuson  1998).  1C coupled with Inductively Coupled Plasma Mass Spectrometry
(ICP-MS) has been extensively  investigated to determine bromate in potable water under a variety of
conditions (Creed et al. 1996, 1997a; Brockoff and  Creed 1997). IC-ICP-MS is the basis  of Method
321.8 (Creed et al. 1997b).  Through the use of IC-ICP-MS, it was determined that brominated HAAs
may interfere with the 1C analysis of bromate (Creed et al. 1997a). Isotope dilution IC-ICP-MS was
investigated for the determination of bromate (Creed and  Brockhoff 1999). Isotope dilution involves
adding a known amount of bromate labeled with a stable (non-radioactive) bromine isotope to the
water sample before analysis. Whatever chemically and physically happens to the analyte (bromate)
also happens to the isotopic addition. Therefore, isotopic addition is considered  a primary and truly
definitive form of measurement.
                                            3-22

-------
Directions in DBF Analytical Chemistry Research

Carcinogenicity has been the primary driving force behind drinking water regulations, and it is likely
that carcinogenicity will continue in this role, although other health effects end points may also be of
concern. Genotoxicity data, not limited just to mutagenicity assays, will probably continue to be used in
assessing health risks of DBFs. However, relatively little effort has been paid to assessing other types of
health effects, such as reproduction and sensitive populations (Bove et al. 1995; Waller et al. 1998).
Reproductive end points are the subject of current EPA/ORD investigation, and the area of other end
points for human health effects could be an interesting area of DBF research for the future. These end
points may be associated with biologically active compounds that remain unidentified. Should a DBF
be implicated in health risks associated  with a form of disinfection, analytical methods will be needed
for its analysis.

Another area of future DBF research is  in the 60% or so of the halogenated material that is not part of
the identifiable classes of compounds (i.e., HAAs, THMs, haloacetonitriles [HANs]). It is possible that
some other highly active compounds are present, especially since the nonvolatile polar compounds are
not well characterized. With the shift in the risk management paradigm, it is not known whether there
will be large-scale continued interest in the identification of new DBFs. In the past, a large effort has
been directed toward first identifying DBFs and then pursuing toxicology/pharmacokinetic studies.
Unquestionably, this has been successful in encouraging utilities to use treatment practices capable of
reducing the concentrations of several key DBFs, including the THMs and HAAs. Because the number
of DBFs is essentially limitless due to the wide range of compounds that make up NOM, the feasability
of large-scale DBF identification efforts is discussed (Urbansky 2000f) in light of more directed ap-
proaches towards  specific human health goals. One such approach is the use of structure-activity rela-
tionships (SARs)  (McKinney et al. 2000; Moudgal et al. 2000).  SARs are based on the presumption
that toxicity is not governed simply by the presence of a halogen, but rather that similar functional
groups are responsible for the  mechanisms of toxicity. There is no a priori basis for asserting that
halogenated organic compounds are necessarily toxic; indeed, many halogenated organic compounds
find use as pharmaceuticals. Likewise, advances in epidemiology and biostatistics can pinpoint human
disease end points for further elucidation (Calderon 2000). Combining SARs with epidemiologic stud-
ies can focus the analytical chemistry on specific classes of compounds rather than expending time and
resources on identifying benign spectator compounds.

New advances in analytical chemistry may complement the use of SARs, epidemiology, and biostatis-
tics. DNA microarray technology permits rapid assessment of individual compounds or groups of com-
pounds to evaluate not only additivity, but also synergy. These methods can be cheaper and faster than
traditional animal toxicology/pathology studies,  which consume considerable resources  and require
sacrificing many laboratory animals. Microarrays are currently being used to investigate DBFs and
endocrine disrupters (Betts 2000). The National Institute of Environmental and Health Sciences (NIEHS)
has created a Microarray Center to study and document genotypic changes (Cooney 2000). Like bio-
logical systems, these arrays can be exposed to complex mixtures in order to  measure additive and
synergistic effects. The arrays are already making headway  in pharmaceutical and biotechnology re-
search.

Research on compounds likely to adversely affect health can be further guided by judicious use of
fractionated, but unidentified materials (Mount and Anderson-Carnahan 1988). If compounds are sepa-
rated using chromatographic, electrophoretic, or other means, the individual fractions may be tested on
microarrays, using indicator organisms (e.g., helminths, cladocerans, amphipods, insect naiads, or cope-
                                              > O">
                                              5-23

-------
pods) that have well-known physiology, anatomy, and biology. Such organisms are routinely collected
from natural waterways as ecological indicators of water quality, serving to identify the presence of
pollutants. The advantage of using biological organisms is that additive effects can be observed even if
the active principles exist at concentrations below the detection limits offered by modern analytical
chemistry. Moreover, if the effects are synergistic rather than additive, a biological system can be used
to observe the interaction phenomena in ways that no current chemoanalytical method could. The
advantage of testing fractionated material before identifying its constituents is that chemicals in samples
shown to be devoid of toxicity need not be identified at all. Consequently, these in vitro biotoxicity tests
serve as a screening mechanism for weeding out countless harmless spectators, saving resources. This
approach has been applied to estrogenic materials in sewage plant effluents and other mixtures more
complex than finished drinking water (Desbrow et  al. 1998).

From a  practical standpoint, there are unresolved  issues  about how many DBFs reach the drinking
water consumers. There are often lengthy delays in the water distribution system, and it is not always
clear how DBF concentrations change after leaving the water plant and before the water reaches the
tap. The stability of DBFs may be affected by reaction with components of the distribution, i.e., pipes,
valves, tanks, etc. Kinetic studies of DBF chemistry under distribution system conditions may someday
elucidate this. In the case of HAAs, for example, the concentration profiles observed in the distribution
system show losses inconsistent with the known chemical kinetics (Urbansky 2000g). It has been specu-
lated that biodegradation is responsible for this loss, but there are many unresolved issues, such as the
potential for heterogeneous catalysis or homogeneous catalysis (general acid/base) (Urbansky 2000g).

From the standpoint of considering DBFs for regulation, research must consider whether existing regu-
lations are already sufficient to control a candidate compound for regulation. Suppose that THM regu-
lations require water treatment plants to be operated in such a manner that compound Y, a candidate for
regulation, is controlled at  the same  time. Promulgating a regulation specifically for compound Y
would then offer no additional benefit to public health. Accordingly, the  expense associated with the
development and support of such a regulation would not be warranted.

An additional direction  for DBF research may be provided through extramural projects. While the
primary focus of this chapter is research conducted or managed by EPAs research laboratories, EPA/
ORD's National Center for Environmental Research continues to fund a wide range of research propos-
als in the area of disinfection by-products, as mentioned previously. For completeness, a list of recent
and ongoing projects, along with the investigators'  institutions, appears in Table 3-10.
                                             3-24

-------
Table 3-10. DBF Research Funded Through NCER
Title
lon-Pair/Supercritical Fluid Extraction
and Derivatization for Polar Organic
Pollutant Analysis
Novel Method for DBF Removal
Development of a Novel Ferroelectric,
Cathode-Based Ozonator for Drinking
Water Treatment
A Comparison of the Effectiveness of
Reverse Osmosis and Ion Exchange
Technologies on the Removal of the
Bromide Ion
Investigation of Model Titania Surfaces
for Heterogeneous Photocatalytic
Oxidation of Chlorinated Organics
Development of Biomarkers for
Haloacetonitriles-Induced Cell Injury
in Peripheral Blood
Water Solubility and Henry's Law Constant
Novel Method for DBF Precursor Removal
Combined Ozonation and Biological
Treatment for the Removal of Humic
Substances from Drinking Waters
Analysis of Organic By -Products from
the Use of Ozone/Chlorine and Ozone/
Chloramines in Drinking Water Treatment
Kinetic -Based Models for Bromate
Formation in Natural Waters
Use of Differential Spectroscopy to Probe
Reactions between Natural Organic Matter
and Chlorinated Oxidants
Engineering of Oxidation and Granular
Activated Carbon Treatment Processes to
Meet New Objectives in Drinking
Water Treatment
Removal of Chlorine Dioxide By-
Products from Drinking Water
Singlet Oxygen Disinfection of
Drinking Water
Zeolite Membranes for Removal
of Contaminants in Drinking Water
Acoustic -Enhanced Ozone Drinking
Water Disinfection
The Particle Size Distribution of
Toxicity in Metal-Contaminated Sediments
Assessment of Human Dietary Ingestion
Exposures to Water Disinfection
By -Products via Food
Institution
Oregon State University
Universal Fuel Development Associates, Inc.
UHV Technologies, Inc.
University of Nevada, Reno
Arizona State University, Tempe
The University of Texas Medical Branch,
Galveston
Lamar University
Universal Fuel Development Associates, Inc.
Michigan State University
University of Massachusetts
Arizona State University
University of Washington, Seattle, WA
University of North Carolina
Novatek
Fayette Environmental Services, Inc.
TDA Research, Inc.
Montec Associates, Inc.
Colorado School of Mines,
Colorado State University
Research Triangle Institute, NC
Grant Number
R821195
68D50145
68D98149
GF9501942
R8 19286
R825955
084LUB5101
68D40043
GF9500518
R825364
R826835
R826645
R820184
68D00033
68D99049
68D50081
68D99059
R826651
R826836
                                         5-25

-------
Title
Institution
Grant Number
Molecular Weight Separation and
HPLC/MS/MS Characterization of
Previously Unidentified Drinking Water
Disinfection By-Products
University of Illinois at Urbana-Champaign
and Metropolitan Water District of
Southern California
R826834
Formation and Stability of Ozonation
By-Products in Drinking Water
University of North Carolina at Chapel Hill     R82683 3
Mechanisms and Kinetics of Chloramine
Loss and By-Product Formation in the
Presence of Reactive Drinking Water
Distribution System Constituents
University of Iowa
R826832
Mechanistic-Based Disinfectant and
Disinfectant By-Product Models for
Chlorine Decay and Regulated DBF
Formation Derived from Free Chlorination
Arizona State University, University of
Massachusetts, University of Colorado,
Malcolm Pirnie
R826831
Integrated Approach for the Control of
Cryptosporidium parvum Oocysts and
Disinfection By-Products in Drinking
Water Treated with Ozone and Chloramines
University of Illinois at Urbana-Champaign     R826830
Pilot Studies of the Ozonation/FBT Process
for the Control of Disinfection By-Products
in Drinking Water
Michigan State University
R826829
Inhalation and Dermal Exposure to
Disinfection By-Products of Chlorinated
Drinking Water
Environmental and Occupational Health
Sciences Institute, University of Medicine
and Dentistry of New Jersey
R825953
Development of a New, Simple,
Innovative Procedure for the Analysis of
Bromate and Other Oxy-Halides at Sub-
ppb Levels in Drinking Water
University of North Carolina at Chapel Hill     R825952
Genotoxicity and Occurrence Assessment
of Disinfection By-Product Mixtures
in Drinking Water
University of Illinois at Urbana-Champaign     R825956
Metabolic Fate of Halogenated Disinfection
By-Products In Vivo and Relation to
Biological Activity
University of North Carolina at Chapel Hill     R825957
The Secondary Structure of Humic Acid
and its Environmental Implications
University of Idaho
R822832
Fate of Bromate Ion and Bromine Compounds
in Water Treatment
Purdue University
R821245
References
American Water Works Association. (1994). Natural organic matter in drinking water: Origin,
    characterization, and removal. American Water Works Association, Denver, CO.
Barrett, S. E. and Krasner, S., eds. (2000). Natural organic matter and disinfection by-products:
    Characterization and control in drinking water (ACS Symposium Series), Vol. 761, American
    Chemical Society, Washington, D.C.
Berts, K.  (2000). "An array of possibilities for evaluating environmental health." Environmental
    Science and Technology, 34, 296A-297A.
                                                  3-26

-------
Bolyard, M, Fair, P. S., and Hautman, D. P. (1992). "Occurrence of chlorate in hypochlorite solutions
   used for drinking water disinfection." Environmental Science and Technology, 26, 1663-1665.

Bolyard, M., Fair, P. S., and Hautman, D. P. (1993). "Sources of chlorate in U.S. drinking water."
   Journal of the American Water Works Association, 85(9), 81-88.

Bove, F., Fulcomer, M., Klotz, J., Esmart, J., Dufficy, E., and Savrin, J. (1995). "Public drinking
   water contamination and birth outcomes." American Journal of Epidemiology, 141, 850-862.

Boyd, R. K., et al. (1996). "Mass spectrometry and good laboratory practices." Journal of the Ameri-
   can Society of Mass Spectrometry, 1, 211-218.

Brockhoff, C. A. and Creed, T. J. (1997). "The determination of bromate in ozonated drinking waters
   by ion chromatography with isotope dilution and inductively coupled plasma mass spectrometric
   detection," Book of abstracts. 214th American Chemical Society National Meeting, Las Vegas,
   NV, September 7-11.

Budde, W. L. and Eichelberger, J. W. (1980). "Performance tests for the evaluation of computerized
   gas chromatography/mass spectrometry equipment and laboratories."  EPA/600/4-80/025, Envi-
   ronmental Protection Agency.

Calderon, R. L. (2000). "The epidemiology of chemical contaminants of drinking water." Food and
   Chemical Toxicology, 38, S13-S20.

Caughran, T. V, Richardson, S. D., Thruston, Jr., A. D., Chen, P. H., Collette, T. W., and Floyd, T. L.
   (1999). "Identification of new drinking water disinfection byproducts formed in the presence of
   bromide," Book of abstracts. 217th American Chemical Society National Meeting, Anaheim, CA,
   March 21-25.

Cavanagh, J. E., et al. (1992). "Ozonation byproducts: Identification of bromohydrins from the
   ozonation of natural waters with enhanced bromide levels." Environmental Science and Technol-
   ogy, 26, 1658-1662.

Chang, T. L., Stretched, R. P., Zimmer, H., and Munch, J. W. (1988). "The interaction of aqueous
   solutions of chlorine with malic acid, tartaric acid, and various fruit juices: A source of mu-
   tagens." Analytical Letters, 21, 2049-2067.

Christman, R. F., et al. (1980). "Chlorination of aquatic humic substances." Performed and prepared
   under EPA grant no. R804430.

Christman, R. R, Norwood, D. L., Millington, D. S., Johnston, J. D., and Stevens,  A. A. (1983a).
   "Identity and yields of major halogenated products of aquatic fulvic acid chlorination." Environ-
   mental Science and Technology, 17, 625-628.

Christman, R. F., Johnson, J. D., Millington, D. S., and Stevens, A. A. (1983b). Chemical reactions
   of aquatic humic materials with selected oxidants. EPA/600/D-83/117, Environmental Protection
   Agency.

Clark, R. M., Pourmoghaddas, H., Wymer, L. J., and Dressman, R. C. (1996). "Modelling the kinet-
   ics of chlorination byproduct formation: The effects of bromide." Journal of Water Supply Re-
   search and Technology-Aqua, 45, 112-119.

Coleman, W. E., Melton, R. G., Kopfler, F. C., Barone, K. A., Aurand, T. A., and Jellison, M. G.
   (1980). "Identification of organic compounds in a mutagenic extract of a  surface drinking water
   by a computerized gas chromatography/mass spectrometry system (GC/MS/COM)." Environ-
   mental Science and Technology, 14, 576-588.
                                              5-27

-------
Coleman, W. E., et al. (1981). "Determination of organic contaminants by the Grob closed-loop-
   stripping technique." Journal of the American Water Works Association, 73(2), 119-125.

Coleman, W. E., Munch, J. W., Kaylor, W. H., Streicher, R. P., Ringhand, H. P., and Meier, J. R.
   (1984). "Gas chromatographic/mass spectroscopy analysis of mutagenic extracts of aqueous
   chlorinated humic acid: A comparison of the byproducts to drinking water contaminants." Envi-
   ronmental Science and Technology, 18, 674-681.

Coleman, W. E., Munch, J. W., Ringhand, H. P., Kaylor, W. H., and Mitchell,  D. E. (1992).
   "Ozonation/post-chlorination of humic acid: A model for predicting disinfection byproducts."
   Ozone: Science and Engineering, 14,  51-69.

Collette, T. W. (1996). "Detection and precise  identification of environmental contaminants by GC
   IR," Book of abstracts. 211th American Chemical Society National Meeting, New Orleans, LA,
   March 24-28.

Collette, T. W., Christman, R. E, McGuire, J. M., and Trusty, C. (1991). Multispectralidentification of
   potentially hazardous byproducts ofozonation and chlorination. Part 1. Studies ofchromato-
   graphic and spectroscopic properties of MX. EPA/600/4-91/004, Environmental Protection Agency.

Collette, T. W., Richardson, S. D., and Thruston, Jr., A. D. (1994). "Identification of bromohydrins in
   ozonated waters." Applied Spectroscopy, 48, 1181-1192.

Cooney, C. M. (2000). "Analyzing toxicants on a chip." Environmental Science and Technology, 34,
   244A-245A.

Cooper, W. J., Zika, R. G., and Steinhauer, M. S. (1985). "Bromide-oxidant interactions and THM
   formation: A literature review" Journal of the American Water Works Association, 77(4), 116-
   121.

Creed, J. T., Magnuson, M. L., Pfaff, J. D., and Brockhoff, C. (1996). "Determination of bromate in
   drinking waters by ion chromatography with inductively coupled plasma mass spectrometric
   detection." Journal of Chromatography A, 753, 261-267.

Creed, J. T., Magnuson, M. L., and Brockhoff, C. (1997a). "Determination of bromate in the pres-
   ence of brominated haloacetic acids by ion chromatography with inductively coupled plasma
   mass spectrometric detection." Environmental Science and Technology, 31, 2059-2063.

Creed, J. T., Brockhoff, C. A., and Martin, T. D. (1997b). "Method 321.8: Determination of bromate
   in drinking waters by ion chromatography inductively coupled plasma-mass spectrometry."
   Environmental Protection Agency.

Creed, J. T. and Brockhoff, C. (1999).  "Isotope dilution analysis of bromate in drinking water ma-
   trixes by ion chromatography with inductively coupled plasma mass spectrometric detection."
   Analytical Chemistry, 71, 722-726.

Croue, J.-P, Korshin, G.  V, and  Benjamin, M., eds. (1999). Characterization of natural organic
   matter in drinking water. American Water Works Association, Denver, CO.

Daniel, F. B., Meier, J. R., and Deangelo,  A. B. (1993). "Advances in research on carcinogenic and
   genotoxic byproducts of chlorine disinfection: Chlorinated hydroxyfuranones and chlorinated
   acetic acids." Annali dell'lnstituto Superiore de Sanita, 29, 279-291.

Daniel, F. B., Robinsson, M., Olson, G.R., Stober, J. A., and Page, N. P. (1994). "lexicological
   studies on MX, a disinfection byproduct." Journal of the American Water Works Association,
   86(3),  103-111.
                                             3-28

-------
DeMarini, D. M., Abu-Shakra, A., Felton, C. R, Schenck Patterson, K., and Shelton, M. L. (1995).
   "Mutation spectra in Salmonella of chlorinated, chloraminated, or ozonated drinking water
   extracts: Comparison to MX." Environmental andMolecular Mutagenesis, 26, 270-285.

Desbrow, C., Routledge, E. 1, Brighty, G. C., Sumpter, J. P., and Waldcock, M. (1998). "Identifica-
   tion of estrogenic chemicals in STW effluent. 1. Chemical fractionation and in vitro biological
   screening" Environmental Science and Technology, 32, 1549-1558.

Doerger, J. U., Meier, J. R., Dobbs, R. A., Johnson, R. D., and Ankley, G. T. (1992). "Toxicity
   reduction evaluation at a municipal wastewater treatment plant using mutagenicity as an end-
   point." Archives of Environmental Contamination and Toxicology, 22, 384-388.

Eichelberger, J. W. (1995). "Method 524.2: Measurement of purgeable organic compounds in water
   by capillary column gas chromatography/mass spectrometry." Methods for the determination of
   organic compounds in drinking water—Supplement III, EPA/600/R-95/131', Environmental
   Protection Agency.

Elovitz, M. S. and von Gunten, U. (2000). "Hydroxyl radical/ozone ratios during ozonation pro-
   cesses. I. The^ concept." Ozone:  Science and Engineering, 21, 239-260.

Feige, M. A., Glick, E. M., Munch, J. W., Munch, D. J., Noschang, R. L., and Brass, H. J. (1980).
   "Potential contaminants introduced into water supplies by the use of coagulant aids." Water
   chlorination: Environmental impact and health effects, R. L. Jolley, W. A. Brungs, and R. B.
   Cumming, eds., Ann Arbor Science, Ann Arbor, MI, 3(69).

Fromme, W. R., Speth, T. F., and Summers, R. S. (1995). "Evaluation of pyrolysis-GS/MS for
   characterization of DBF precursors," Book of abstracts. 210th American Chemical Society
   National Meeting, Chicago, IL, August 20-24.

von Gunten, U. and Hoigne, J. (1994).  "Bromate formation during ozonation of bromide-containing
   waters: Interaction of ozone and hydroxyl radical reactions." Environmental Science and Tech-
   nology, 28,1234-1242.

Guo, Y, Richardson, S. D., and Caughran, T. V. (1998). "Identification and mechanistic studies of
   drinking water disinfection byproducts," Book of abstracts. 216th American Chemical Society
   National Meeting, Boston, MA, August 23-27.

Hautman, D. P. (1993). "Analysis of trace bromate in drinking water using selective anion concentra-
   tion and ion chromatography." Proceedings, Annual Conference of the American Water Works
   Association, Water Quality, 993-1007.

Hautman, D. P. and Bolyard, M. (1992a). "Using ion chromatography to analyze inorganic disinfec-
   tion by products." Journal of the A merican Water Works Association, 84(11), 88-93.

Hautman, D. P. and Bolyard, M. (1992b). "Analysis of inorganic disinfection byproducts using ion
   chromatography." Proceedings, 1991 Water Quality Technology Conference, American Water
   Works Association, Part 2, 1043-1059.

Hautman, D. P. and Bolyard, M. (1992c). "Analysis of oxyhalide disinfection byproducts and other
   anions of interest in drinking water by ion chromatography." Journal of Chromatography A, 602,
   65-74.

Hautman, D. P. and Munch, D. J. (1997). "Development of U.S. EPA Method 551.1" Journal of
   Chromatographic Science, 35, 221-231.
                                             3-29

-------
Herren-Freund, S. L., Pereira, M. A., Khoury, M. D., and Olson, G. (1987). "The carcinogenicity of
   trichloroethylene and its metabolites, trichloracetic acid and dichloracetic acid, in mouse liver."
   Toxicology and Applied Pharmacology., 90, 183-189.

Ho, J. S. (1995). "Method 502.2: Volatile organic compounds in water by purge and trap capillary
   column gas chromatography with photoionization and electrolytic conductivity detectors in
   series." Methods for the determination of organic compounds in drinking water—Supplement III,
   EPA/600/R-95/131, Environmental Protection Agency.

Hodakievic, P. A. and Ho, J. S. (1990). "A direct purge method for determination of total
   trihalomethanes in drinking water with an electrolytic conductivity detector." Advances in Instru-
   mentation and Control, 45, 339-347.

Hodgeson, J. W., Collins, J., and Earth, R. E. (1988). "Method 552.0: Determination of haloacetic
   acids in drinking water by liquid-liquid extraction, derivatization and gas chromatography with
   eectron capture detection." Methods for the determination of organic compounds—Supplement
   II, EPA 600/4-88/039, Environmental Protection Agency.

Ireland, J.  C. and Valinieks, J. (1992). "Rapid measurement of aqueous hydroxyl radical concentra-
   tions in steady-state HO flux systems." Chemosphere, 25, 383-396.

Johnson, J. D. and Jensen, J. N. (1986). "THM and TOX formation: Routes, rates, and precursors."
   Journal of the American Water Works Association, 78(4), 156-162.

Jolley, R. L., ed. (1976). The environmental impact of water chlorination. Proceedings,  Conference
   on the  Environmental Impact of Water Chlorination,  Oak Ridge National Laboratory, Energy
   Research and Development Administration, Oak Ridge, TN, October 22-24, 1975.

Jolley, R. L., Gorcher, H., and Hamilton, Jr., D. H., eds. (1978). Water chlorination—Environmental
   impact and health effects, volume 2. Proceedings, Second Conference on the Environmental
   Impact of Water Chlorination, Gatlinburg, TN, October 31-November 4, 1977. Ann Arbor
   Science Publishers, Ann Arbor, MI.

Jolley, R. L., Brungs, W. A., Cumming, R. B., and Jacobs, V. A., eds.  (1980). Water chlorination—
   Environmental impact and health effects,  volume 3. Proceedings, Third Conference  on Water
   Chlorination—Environmental Impact and Health Effects, Colorado Springs, CO, October 28-
   November 2, 1979. Ann Arbor Science Publishers, Ann Arbor, MI.

Jolley, R. L., Brungs, W. A., Cotruro, J. A., Cumming, R. B., Mattice, J. S., and Jacobs, V. A., eds.
   (1983). Water chlorination—Environmental impact and health effects,  volume 4 (Parts 1 and 2).
   Proceedings, Fourth Conference on Water Chlorination—Environmental Impact and Health
   Effects, Pacific Grove, CA, October 18-23, 1981. Ann Arbor Science Publishers, Ann Arbor, MI.

Jolley, R. L., Bull, R. J., Davis, W. P., Katz, S., Roberts,  Jr., M. H., and Jacobs, V. A., eds. (1985).
   Water  chlorination: Chemistry, environmental impact and health effects, volume 5. Proceedings,
   Fifth Conference on Water Chlorination—Environmental Impact and Health Effects,
   Williamsburg, VA, June 3-8, 1984. Lewis Publishers, Chelsea, MI.

Jolley, R. L., Condie, L. W., Johnson, J. D., Katz, S., Mattice, J. S., and Jacobs, V. A., eds. (1990).
   Water  chlorination: Chemistry, environmental impact and health effects, volume 6. Proceedings,
   Sixth Conference on Water Chlorination—Environmental Impact and Health Effects, Oak Ridge,
   TN, May 3-8, 1987. Lewis Publishers, Chelsea, MI.

Kanniganti, R., Johnson, J. D., Ball, L. M., and Charles,  M. J. (1992). "Identification of compounds
   in mutagenic extracts of aqueous monochloraminated fulvic acid." Environmental Science and
   Technology, 26, 1998-2004.


                                            3-30

-------
Koffskey, W. E. (1993). "Project summary: Disinfection byproduct formation by alternative disinfec-
   tants and removal by granulated activated carbon." EPA/600/SR-93/136, Environmental Protec-
   tion Agency.

Kopfler, F. C., Ringhand, H. P, Coleman, W. E., and Meier, J. R. (1984). "Reactions of chlorine in
   drinking water, with humic acids and in vivo." EPA/600/D-84/196, Environmental Protection
   Agency.

Kotiaho,  T., et al. (1992). "Time persistence of monochloramine in human saliva and stomach fluid."
   Environmental Science and Technology., 26, 302-306.

Lin, D. C. K., Melton, R. G., Kopfler, R C., and Lucas, S. V. (1981). "Glass capillary gas chromato-
   graphic/mass spectrometric analysis of organic concentrates from drinking and advanced waste
   treatment waters." Advances in the identification and analysis of organic pollutants in water,
   Vol. 2, L. H. Keith, ed., Ann Arbor Science, Ann Arbor, MI, 861-906.

Lykins, Jr., B. W., Koffskey, W. E., and Schenck Patterson, K. (1994). "Alternative disinfectants for
   drinking water treatment." Journal of Environmental Engineering Division of the ASCE, 120,
   745-758.

Magnuson, M. L. (1998). "Determination of bromate at parts-per-trillion levels by gas chromato-
   graphy-mass spectrometry with negative chemical ionization." Analytica ChimicaActa, 377,
   53-60.

Magnuson, M. L. and Kelty, C.  A. (2000). "Microextraction of nine haloacetic acids in drinking
   water at microgram per liter with electrospray-mass spectrometry of stable association com-
   plexes. " Analytical Chemistry, 72, 2308-2312.

McKinney, J. D., Richard, A., Waller, C., Newman, M. C., and Gerberick, F. (2000). "The practice of
   structure activity relationships (SAR) in toxicology." Toxicological Sciences, 56, 8-17.

Meier, J.  R., Lingg, R. D., and Bull, R. J. (1983). "Formation of mutagens following chlorination of
   humic acid.  A model for mutagen formation during drinking water treatment." Mutation Re-
   search, 118,25-41.

Meier, J.  R. and Bull, R. J. (1984). "Mutagenic properties of drinking water disinfectants and
   byproducts." EPA/600/D-84/224, Environmental Protection Agency.

Meier, J.  R., et al.  (1985). "Identification of mutagenic compounds formed during chlorination of
   humic acid." Mutation Research, 157, 111-122.

Meier, J.  R. and Bishop, D. F. (1985). "Evaluation of conventional treatment processes for removal
   of mutagenic activity from municipal wastewaters." Journal of the Water Pollution Control
   Federation,  57, 999-1005.

Meier, J.  R., et al.  (1986). "Mutagenic byproducts from chlorination of humic acid." Environmental
   Health Perspectives, 69,101-107.

Meier, J.  R., et al.  (1987). "Studies on the potent bacterial mutagen, 3-chloro-4-(dichloromethyl)-5-
   hydroxy-2[5//]-furanone: Aqueous stability, XAD recovery and analytical determination in
   drinking water and in chlorinated humic acid solutions." Mutation Research, 189, 363-373.

Meier, J.  R. (1988). "Genotoxic activity of organic chemicals in drinking water." Mutation Research,
   196,211-245.

Meier, J.  R., et al.  (1990). "Genotoxic and carcinogenic properties of chlorinated furanones: Impor-
   tant byproducts of water chlorination." Environmental Science Research, 39, 185-195.
                                              5-31

-------
Meier, J. R., et al. (1996). "Urine mutagenicity and biochemical effects of the drinking water mu-
   tagen 3-chloro-4-(dichloromethyl)-5-hydroxy-2[5//]-furanone (MX), following repeated oral
   administration to mice and rats." Toxicology., 110, 59-70.

Meier, J. R., Stewart, B. E., and Blazak, W. F. (1997). "Genotoxicity studies of sodium
   dichlororacetate and sodium trichloroacetate." Environmental Science (Tokyo), 5, 95-108.

Melnick, R. L., Boorman, G. A., and Dellarco, V. (1997). "Water chlorination, 3-chloro-4-(dichloro-
   methyl)-5-hydroxy-2[5//]-furanone (MX), and potential cancer risk." Journal of the National
   Cancer Institute, 89,  832-833.

Melton, R.  G., et al. (1981). "Comparison of Grob closed-loop stripping analysis with other trace
   organic methods." Advances in the identification and analysis of organic pollutants in water, Vol.
   2, L. H. Keith, ed., Ann Arbor Science, Ann Arbor, MI, 597-673.

Miltner, R.  J., Shukairy, H. M., and Summers, R.  S. (1992). "Disinfection byproduct formation and
   control  by ozonation and biotreatment." Journal of the American Water Works Association,
   84(11),  53-62.

Minear, R. A. and Amy, G. L., eds. (1996a). Water disinfection and natural organic matter: Charac-
   terization and control (ACS  Symposium  Series), Vol. 649, American Chemical Society, Wash-
   ington,  D.C.

Minear, R. A. and Amy, G. L., eds. (1996b). Disinfection by-products in water treatment: The chem-
   istry of their formation and control. Lewis/CRC, Boca Raton, FL.

Mink, F. L., Coleman, W. E., Munch, J. W., Kaylor, W. H., and Ringhand, H. P. (1983). "In vivo
   formation of halogenated reaction products following peroral sodium hypochlorite." Bulletin of
   Environmental Contamination Toxicology, 30, 394-399.

Moudgal, C. J., Lipscomb, J. C., and Bruce, R. M. (2000). "Potential health  effects of drinking water
   disinfection byproducts using quantitative structure toxicity relationship." Toxicology, 147, 109-
   131.

Mount, D. I. and Anderson-Carnahan, L. (1988). "Methods for aquatic toxicity identification evalua-
   tions. Phase I. Toxicity characterization procedures." EPA/600/3-88/034, Environmental Protec-
   tion Agency, Duluth, MN.

Munch, D.  J. and Hautman, D. P. (1995). "Method 551.1: Determination of chlorination disinfection
   byproducts, chlorinated solvents, and halogenated pesticides/herbicides in drinking water by
   liquid-liquid extraction and gas chromatography with electron-capture detection." Methods for
   the determination of organic compounds in drinking water—Supplement III, EPA/600/R-95/131,
   Environmental Protection Agency, Cincinnati, OH.

Munch, D.  J., Munch, J. W., and Pawlecki, A. M. (1995). "Method 552.2: Determination  of
   haloacetic acids and dalapon in drinking water by liquid-liquid extraction, derivatization and gas
   chromatography with electron capture detection." Methods for the determination of organic
   compounds—Supplement III, EPA/600/R-95/131, Environmental Protection Agency, Cincinnati,
   OH.

Munch, J. W., Munch, D. J., Winslow, S. D.,  Wendelken, S.  C., and Pepich, B. V (1998a). "Method
   556. Determination of carbonyl compounds in drinking water by pentafluorobenzylhydroxyl-
   amine derivatization and capillary gas chromatography with electron capture detection." Rev. 1.0,
   Environmental Protection Agency, Cincinnati, OH.
                                             3-32

-------
Munch, J. W., Munch, D. J., and Winslow, S. D. (1998b). "A user's guide to aldehyde analysis using
   PFBHA derivatization and GC/ECD detections: Avoiding the pitfalls." Proceedings, 1998 Water
   Quality Technology Conference, American Water Works Association, 898-913.

Munch, J. W., Coleman, W. E., Hoadkievic, P. A., Kaylor, W. H., and Meier, J. R. (1988). "Determi-
   nation of a chlorinated furanone (MX) in treated water." Proceedings, 1987 Water Quality Tech-
   nology Conference, American Water Works Association, 15, 933-942.

Norwood, D. L., Thompson, G. P., St. Aubin, J. J., Millington, D. S., Christman, R. R, and Johnson,
   J. D. (1983). "Byproducts of chlorination: Specific compounds and their relationship to total
   organic halide." American Chemical Society National Meeting, Washington, D.C., August 28-
   September 2.

Owen, D. M., Amy, G. L., and Chowdhury, Z. K., eds. (1993). Characterization of natural organic
   matter and its relationship to treatability. American Water Works Association, Denver, CO.

Pawlecki-Vonderheide, A. M., Munch, D. J., and Munch, J. W. (1997). "Research associated with the
   development of EPA Method 552.2." Journal of Chromatographic Science, 35, 293-298.

Pedersen, E. J., Urbansky, E. T., Marinas, B. Y, and Margerum, D. W. (1999). "Formation of cyano-
   gen chloride from the reaction of monochloramine with formaldehyde." Environmental Science
   and Technology, 33, 4239-4249.

Pfaff, J. D. (1993). "Determination  of inorganic anions by ion chromatography" Methods for the
   determination of inorganic substances in environmental samples, EPA/600/R-93/100, Environ-
   mental Protection Agency.

Pourmoghaddas, H. (1991). "Project summary: Effect of bromide on chlorination byproducts in
   finished drinking water." EPA/600/S2-91/036, Environmental Protection Agency.

Pourmoghaddas, H., Stevens, A. A., Kinman, R. N., Dressman, R. C., Moore, L. A., and Ireland,
   J. C. (1993). "Effect of bromide ion of formation of HAAs during chlorination." Journal of the
   American Water Works Association, 85(1), 82-87.

Prakash, B., Pepich, B. V, Munch, D., and Frebis, C. (1998). "The performance of a purge and trap-
   gas chromatography/ion trap mass spectrometry method for the routine analysis of cyanogen
   chloride in several hundred municipal drinking water samples." Proceedings, Water Quality
   Technology Conference, American Water Works Association, 914-926.

Raymer, J. H., Pellizzari, E. D., and Shoemaker, J. A. (1999a). "Human exposure to water disinfec-
   tion byproducts via foods and beverages," Book of abstracts. 218th American Chemical Society
   National Meeting, New Orleans, August 22-26.

Raymer, J. H., Pellizzari, E. D., and Shoemaker, J. S. (1999b). "Analytical methods for water disin-
   fection byproducts in foods and beverages," Book of abstracts. 218th American Chemical Society
   National Meeting, New Orleans, August 22-26.

Reckhow, D. A., Singer, P. C., and Malcolm, R. L. (1990). "Chlorination of humic materials:
   Byproduct formation and chemical interpretations." Environmental Science and Technology, 24,
   1655-1664.

Reckhow, D. A., Legube, B., and Singer, P. C. (1986). "The ozonation of organic halide precursors:
   Effect of bicarbonate." Water Research, 20, 987-998.
                                             >  O O
                                             5-33

-------
Rice, R. G., Overbeck, P. K., and Larson, K. (1998). "Ozone treatment for small systems." First
   International Symposium on Safe Drinking Water in Small Systems., NSF Internal!onal/PAHP/
   WHO, Washington, D.C., May 10-13.

Richardson, S. D., Thruston, Jr., A. D., Collette, T. W., and McGuire, J. M. (1991). "Application of
   multispectral techniques to the precise identification of aldehydes in the environment." Environ-
   mental Toxicology and Chemistry, 10, 991-997.

Richardson, S. D., et al. (1994). "Multispectral identification of chlorine dioxide disinfection
   byproducts in drinking water." Environmental Science and Technology, 28, 592-599.

Richardson, S. D., Thruston, Jr., A. D., Collette, T. W., Patterson, K. S., Lykins, Jr., B. W. (1995).
   "Alternative disinfectants for drinking water," Book of abstracts. 210th American Chemical
   Society National Meeting,  Chicago, IL, August 20-24.

Richardson, S. D., Thruston, Jr., A. D., Collette, T. W., Patterson, K. S., Lykins, Jr., B. W., and
   Ireland, J. C.  (1996). "Identification of TiO /UV disinfection byproducts in drinking water."
   Environmental Science and Technology, 30^ 3327-3334.

Richardson, S. D. (1998a). "Drinking water disinfection byproducts." Encyclopedia of environmental
   analysis and remediation, R. A. Meyers, ed., John Wiley & Sons, New York, 1398-1421.

Richardson, S. D., et al. (1998b). "Identification of new disinfection byproducts using GC/MS, LC/
   MS, and GC/IR." Proceedings, Annual Conference of the American Water Works Association,
   Water Quality, 2223-2234.

Richardson, S. D. (1999).  "Water analysis." Analytical Chemistry, 71, 181R-215R.

Richardson, S. D., Caughran, T. V, Poiger, T, Guo, Y, and Crumbley, F. G. (1999a). "Identification
   of polar drinking water disinfection byproducts using LC/MS," Book of abstracts. 217th Ameri-
   can Chemical Society National Meeting, Anaheim, CA, March 21-25.

Richardson, S. D., et al. (1999b). "Identification of new drinking water disinfection byproducts
   formed in the presence of bromide." Environmental Science and Technology, 33, 3378-3383.

Ringhand, H. P., et al. (1988a). "Biological and chemical studies on 3-chloro-4-(dichloromethyl)-5-
   hydroxy-2(5//) furanone: A potent mutagen in kraft pulp chlorination effluent and chlorinated
   drinking water." Hazardous Industrial Waste, 20, 420-433.

Ringhand, H. P., Meier, J.  R., Coleman, W. E., Schenck, K. M., and Kaylor, W. H. (1988b). "Biologi-
   cal and chemical studies on 3-chloro-4-(dichloromethyl)-5-hydroxy-2(5//) furanone: A potent
   mutagen in kraft pulp chlorination effluent and chlorinated drinking water." EPA-600/D-88/159,
   Environmental Protection Agency.

Ringhand, H. P., et al. (1989).  "Studies on MX: A potent mutagenic byproduct of the chlorination of
   drinking water and kraft pulp mill effluents." Trace Substances in Environmental Health, 23,
   269-267.

Roberson, J. A., et al. (1995). "The D/DBP Rule: Where did the numbers come from?" Journal of
   the American Water Works Association,  87(10), 46-57.

Rubio, F. J., Urbansky, E.  T., and Magnuson, M. L.  (2000). "Avoiding pitfalls in the determination of
   halocarboxylic acids: The photochemistry of methylation" Journal of Environmental Monitor-
   ing, 2, 248-252.

Schenck, K. M., Meier, J.  R., Ringhand, H. P., and Kopfler, F. C. (1990). "Recovery of 3-chloro-4-
   (dichloromethyl)-5-hydroxy-2(5//) furanone from water samples on XAD resins and the effect of
   chlorine on its mutagenicity." Environmental Science Technology, 24, 863-867.


                                             3-34

-------
Schenck Patterson, K. and Lykins, Jr., B. W. (1993). "The role of mutagenicity in determining drink-
   ing water quality." Proceedings, Annual Conference of the American Water Works Association.,
   Water Quality, 809-830.

Schenck Patterson, K., Lykins, Jr., B. W. and Richardson, S. D. (1995). "Mutagenicity of drinking
   water following disinfection." Journal of Water Supply, Research, and Technology-Aqua (Lon-
   don), 44, 1-9.

Schenck Patterson, K. and Lykins, Jr., B. W. (1995). "Mutagenicity of drinking water samples treated
   with alternative disinfectants." Environmental and Molecular Mutagenesis, 25, 41-48.

Seeger, D. R., Moore, L. A., and Stevens, A. A. (1984a). "The formation of acidic trace organic
   byproducts from the chlorination of humic acids." Fifth Water Chlorination Conference: Envi-
   ronmental Impact and Health Effects., Williamsburg, VA, June 3-8.

Seeger, D. R., Moore, L. A., and Stevens, A. A. (1984b). "The formation of acidic trace organic
   byproducts from the chlorination of humic acids." EPA/600/D-84/159, Environmental Protection
   Agency.

Shoemaker, J. A., Munch, J. W., and Behymer, T. D. (1999). "Evaluation of solid phase
   microextraction for the analysis of hydrophilic compounds." Journal of Exposure Analysis and
   Environmental Epidemiology, 9, 181-191.

Shorney, H. L. and Freeman, S. (1999). "Control of disinfection byproduct precursors." .

Singer, P. C., ed. (1999). Formation and control of disinfection by-products in drinking water.
   American Water Works Association, Denver, CO.

Slocum, C. J., Moore, L. A., and Stevens, A. A. (1988). "Analytical criteria for detection of disinfec-
   tion byproducts." Proceedings,  1987 Water Quality Technology Conference, American Water
   Works Association (Issue Answers Today's Water Quality Profession),  829-838.

Snyder, M. P. and Margerum, D. W. (1982). "Kinetics of chlorine transfer from chloramine to
   amines, amino acids, and peptides." Inorganic Chemistry, 21, 2545-2550.

Solarik, G., Hatcher, V. A., Isabel, R. S., Stile, J. E, and Summers, R. S. (1997). "Prechlorination and
   DBF formation: The impact of chlorination point and enhanced coagulation." Proceedings, Water
   Quality Technology Conference, American Waterworks Association, 2B1/1-2B1/19.

Sorrell, R. K. and Hautman, D. P. (1993). "A simple concentration technique for the analysis of
   bromate at low levels in drinking water." Proceedings, Annual Conference  of the American Water
   Works Association, Water Quality, 281-290.

Stevens, A. A. (1984).  "TOX (total organic halogen). Is it the non-specific parameter of the future?"
   EPA/600/D-84/169, Environmental Protection Agency.

Stevens, A. A. and Symons, J. M. (1984b). "Alternative disinfection processes." EPA/600/D-84/221,
   Environmental Protection Agency.

Stevens, A. A., Moore, L. A., Slocum, C.  J., Smith, B. L., Seeger, D. R., and Ireland, J.  C. (1989).
   "Chlorinated humic acid mixtures: Criteria for detection of disinfection byproducts in drinking
   water." Advances in Chemistry Series, 219, 681-695.

Stevens, A. A., Moore, L. A., Slocum, C.  J., Smith, B. L., and Seeger, D. R. (1987). "Chlorinated
   humic acid mixtures establish criteria for detection of disinfection byproducts in drinking water."
   EPA/600/D-87/142, Environmental Protection Agency.
                                              5-35

-------
Stultz, M. L., et al. (1998). "DBF analyses for the ICR: How well are labs doing?" Proceedings,
    Water Quality Technology Conference., American Water Works Association, 1471-1479.

Summers, R. S., Hooper, S. M., Shukairy, H. M., Solarik, G., and Owen, D. (1996). "Assessing DBF
    yield: Uniform formation conditions." Journal of the American Water Works Association, 88(6),
    80-93.

Symons, J. M., ed. (1997). Factors affecting disinfection by-product formation during
    Moranimation. American Water Works Association Research Foundation, Denver, CO.

Symons, J. M., et al. (1998). Factors affecting disinfection byproduct formation during
    chloramination. American Water Works Association Research Foundation, Denver, CO.

Trehy M. L., Yost, R. A., and Miles, C. J. (1986). "Chlorination byproducts of amino acids in
    natural waters." Environmental Science Technology, 20, 1117-1122.

Ulmer, N. S., Dressman, R. C., Moore, L. A., and Stevens, A. A. (1988). "Problems in the determi-
    nation of haloacetic acids and chloral hydrate in drinking water." Proceedings, 1987 Water
    Quality Technology Conference, American Waterworks Association (Issue Answers Today's
    Water Quality Profession), 741-750.

Urbansky, E. T. (1999). "Ascorbic acid treatment to reduce residual halogen-based oxidants prior to
    the determination of halogenated disinfection byproducts in potable water." Journal of Environ-
    mental Monitoring, 1, 471-476.

Urbansky, E. T. and Bashe, W. J. (2000). "Comparative methodology in the determination of the oc-
    oxocarboxylates in aqueous solution: Ion chromatography versus gas chromatography after
    oximation extraction and esterification." Journal of Chromatography A, 867,  143-149.

Urbansky, E. T. and Brown,  S. K. (2000). "Influence of reagent purity on the ion chromatographic
    determination of bromate in water using the prochromophore 3,3-dimethoxybenzidine for photo-
    metric detection." Journal of Environmental Monitor ing, 2, 571-575.

Urbansky, E. T., Magnuson,  M.  L., Elovitz, M. S., Freeman, D., and Shauntee, J. (2000a). "Interfer-
    ences due to ozone-scavenging reagents in the GC-ECD determination of aldehydes and ketones
    as the O-(2,3,4,5,6-pentafluorobenzyl)oximes." Ozone: Science and Engineering, 2000A, 22,
    551-561.

Urbansky, E. T., Freeman, D. M., and Elovitz, M.  S. (2000b). "Ozone-scavenging reagents suitable
    for use the in the determination of aldehydes as the O-(2,3,4,5,6-pentafluorobenzyl) oximes by
    GC-ECD." Water Research, 34, 2610-2613.

Urbansky, E. T., Freeman, D. M., and Rubio, F. J.  (2000c). "Ascorbic acid reduction of residual
    active chlorine in potable water prior to halocarboxylate determination." Journal of Environmen-
    tal Monitoring, 2, 253-256.

Urbansky, E. T. (2000d). "Influences of metal cations on the determination of the oc-oxocarboxylates
    as the methyl esters of the O-(2,3,4,5,6-pentafluorobenzyl)oximes by gas chromatography: The
    importance of accounting for matrix effects." Journal of Environmental Monitoring, 2, 334-338.

Urbansky, E. T. (2000e). "Techniques and methods for determining haloacetic acids in potable
    water." Journal of Environmental Monitoring, 2, 285-291.

Urbansky, E. T. (2000f). "Disinfection byproducts in drinking water." Analytical Chemistry, 72,
    439A-440A.
                                             3-36

-------
Urbansky, E. T. (2000g). "Modeling the fate and transformation of the haloacetates—fundamental
   chemical kinetics." (Submitted for publication.)

Urbansky, E. T. (2000h). "Carbinolamines and geminal diols in environmental chemistry." Journal of
   Chemical Education, 77(12), 1644-1647.

Urbansky, E. T. (2001). "Total carbon analyzers as tools for measuring carbonaceous matter in
   natural waters." Journal of Environmental Monitoring, in press.

U.S. Environmental Protection Agency (USEPA). (1989). "Drinking water criteria document for
   chloramines." ECAO/600/D002.

USEPA. (1992). "Water treatment plant simulation program, Version 1.21." Office of Water.

USEPA. (1994). "Water treatment plant model for MS Windows 3.1, Version 1.55." EPA/811/B-94/
   001.

USEPA. (1996). "Drinking water regulations and health advisories." EPA/882/B-96/002.

USEPA. (1997). "Community water system survey—Volumes I and II; Overview." EPA/815/R-97-
   OOla, -OOlb.

USEPA. (1999a). "Alternative disinfectants and oxidants guidance manual." EPA/815/R-99/014.

USEPA. (1999b). "Microbial and disinfection byproduct rules simultaneous compliance guidance
   manual." EPA/815/R-99/011.

Wachter, J. K. and Andelman, J. B. (1984). "Organohalide formation on chlorination of algal extra-
   cellular products." Environmental Science Technology., 11, 811-817.

Wagner, H. P., Alig, A. E., Pepich, B. V, Frebis, P. C., Hautman, D. P., and Munch, D. (1998). "A
   study of ion chromatographic methods for trace level bromate analysis in drinking water compar-
   ing the selective anion concentration (SAC) method, U.S. EPA Method 300.1, and a postcolumn
   reagent procedure." Proceedings,  Water Quality Technology Conference, American Water Works
   Association, 872-876.

Wagner, H. P., Pepich, B.V, Hautman, D. P., and Munch, D. J. (1999). "Analysis of 500 ng/L levels
   of bromate in drinking water by direct-injection suppressed ion chromatography coupled with a
   single, pneumatically delivered postcolumn reagent." Journal of Chromatography A., 850, 119-
   129.

Waller, K., Swan S., DeLorenze G., and Hopkins B. (1998). "Trihalomethanes in drinking water and
   spontaneous abortion." Epidemiology, 9, 134-140.

Weinberg, H. S. (1999). "Disinfection byproducts in drinking water: The analytical challenge."
   Analytical Chemistry, 71, 801A-808A.

White,  G. C. (1999). Handbook of chlorination and alternative disinfectants, 4th Ed., John Wiley &
   Sons, New York.

Wong,  C. S., Garrison, A. W., and Richardson, S. D.  (1999). "Enantiomeric composition of chiral
   haloacetic acid and haloacetonitrile disinfection byproducts in drinking water," Book of abstracts.
   217th American Chemical Society National Meeting, Anaheim, CA, March 21-25.

Xu, G., Stevens, D. K., and Bull, R. J. (1995). "Metabolism of bromodichloroacetate in B6C3F1
   mice." Drug Metabolism and Disposition, 23, 1412-1416.
                                             5-37

-------
3-38

-------
                                      CHAPTER 4

                  Source Water Protection: Its Role in Controlling
          Disinfection By-Products (DBPs) and Microbial Contaminants1

Introduction

Passage of the 1996 Safe Drinking Water Act Amendments (SDWAA) has focused the attention of
water utility managers and public health and regulatory officials on source water protection (SWP) and
its role in protecting public water supplies. There is growing awareness that water treatment and/or
disinfection may not always be enough to ensure the provision of potable and safe water to the con-
sumer.  The 1993 cryptosporidiosis outbreak in Milwaukee, WI, has raised the  possibility that even
water suppliers which meet all of the Surface Water Treatment Rule (SWTR) requirements of the
SDWA are vulnerable (Okun et al. 1997; Fox and Lytle 1996).

Most utilities in the U.S. invest a great deal of time, energy, and capital in developing mechanisms for
protecting against the impact of sudden changes in influent water quality. Some of these mechanisms
include investment in excess capacity and development of emergency procedures (Miller 1989).

Concern  over source water protection is not limited to surface water supplies.  Many ground water
supplies have proven to be vulnerable as well, resulting in the various states implementing wellhead
protection programs. Based on the 1996 amendments, the states will have to implement programs to
decide if a system's source of supply is threatened as well as determine the means to prevent pollution.
Communities will be allowed to ask for state assistance, and a certain percentage of the State Revolv-
ing Loan Fund has been earmarked to assist with source water protection (Howell 1997).

Water supplies vary greatly in  the nature of the source water they use and in the  circumstances under
which they provide water to their customers. Nevertheless, there are some common elements that are
applicable to source water protection in general. For example, land-use planning can provide informa-
tion that is related to source water protection. Information on population densities, the ratio of pervious
to impervious land cover, and the location of point and non-point sources  of pollution can be important
in assessing problems associated with both ground and surface source water protection.

As part of the Clean Water Act (CWA),  Comprehensive River Basin Planning was initiated under
Section 208 of the CWA. A major effort was undertaken to bring to bear the existing art and science of
comprehensive planning in river basins with regard to minimizing the impact of point and non-point
pollution on water quality in streams, lakes, and ground water. Many of the approaches suggested in
studies developed under this program are very relevant to the issue of source water protection today.

Stream and contaminant transport models  provide a mechanism for identifying and assessing the pol-
lutants that are likely to be present in surface sources used for water supply. These models can be used
for (1) identification of communities whose water supplies could be vulnerable  to contamination re-
sulting from industrial  and municipal discharges or urban and agricultural runoff, (2) design of water
and wastewater treatment plants, (3) design and implementation of water quality monitoring programs,
and (4) other water resource planning efforts requiring information on the quality of surface waters
(Clark etal. 1998).
'Michael Borst, Maureen Krudner, Marie L. O'Shea, Joyce M. Perdek, Donald Reasoner, Michael D.
Royer: ORD/NRMRL/WSWRD, 2800 Woodbridge Ave., Edison, NJ, 08837-3679. Corresponding
Author: Michael Borst, 732-321-6631, borst.michael@epa.gov.


                                             4-1

-------
This chapter will explore SWP as outlined in the Safe Drinking Water Act (SDWA): the nature of
threats to source water quality; methods, monitoring, and assessment of pathogens; technologies for
control of water quality; the use of models to assess water utility vulnerability; and the relationship of
source water protection to watershed management.

The Safe Drinking Water Act and Source Water Protection

The SDWA was passed in  1974 and  amended in  1986 and  1996, but SWP under the SDWA
actually began with the SDWA Amendments of 1986. The 1986 amendments included
provisions for "Protection of Ground Water Sources of Water." Two programs were set up under
this requirement: the "Sole Source Aquifer Demonstration Program," to establish demonstration
programs to protect critical aquifer  areas from degradation; and the "Wellhead Protection Pro-
gram," which required states to develop programs for protecting areas around public water supply
wells to prevent contamination from residential, industrial, and farming activities.

The SWTR, published in June 29, 1989, and effective December 31, 1990,  was designed to prevent
waterborne diseases caused by viruses, Legionella, and Giardia lamblia. These disease-causing organ-
isms are present in varying concentrations in most surface waters. This rule requires water systems to
filter and disinfect water from surface water sources to reduce the occurrence of unsafe levels of these
microbes. Surface water is  particularly susceptible to microbial contamination from sewage treatment
plant discharges, storm water runoff,  and snow melt. The rule sets nonenforceable health goals and
maximum contaminant level goals (MCLGs) for viruses, Legionella, and Giardia lamblia at zero because
any amount of exposure to these contaminants represents some health risk. In establishing legal limits for
these contaminants in drinking water, the U.S. Environmental Protection Agency (EPA) can set either a
maximum contaminant level (MCL), which is a legal limit, and require monitoring for the contaminant in
drinking water, or, for those contaminants that are difficult to measure, EPA can establish a treatment
technique requirement. Since measuring disease-causing microbes in drinking water is not considered to
be feasible, EPA established a treatment technique in this rule.

The SWTR Guidance Manual (USEPA 1991) identifies both natural  and human-caused sources of
contamination to be controlled. These  sources include wild animal populations, wastewater treatment
plants, grazing animals, feedlots, and recreational activities. The Guidance Manual recommends that
grazing and sewage discharges not be permitted within the watershed of unfiltered systems. Both may
be permissible on a case-by-case basis where the watershed provides a long detention time and a high
dilution between the location  of the activity and the water intake. The nonfiltering utility is required to
develop state-approved techniques to eliminate or reduce the effect of the identified point and non-point
pathogenic contamination sources.

In the 1996 amendments to the SDWA, protection of source waters was given greater emphasis to
strengthen protection against microbial  contaminants, particularly  Cryptosporidium., while reducing
potential  health risks due to  disinfection by-products. This increased  protection is embodied in the
Interim Enhanced SWTR (TESWTR) (USEPA 1998). This rule applies to public water systems that use
surface water or ground water under the  direct influence of surface water (GWUDI) and serve at least
10,000 people. The final IESWTR (USEPA  1998), issued December 16, 1998, and effective February
16, 1999, includes  several requirements specific to finished drinking  water, and three that relate to
watershed protection. EPA is  to

    •   set a MCLG of zero for Cryptosporidium
    •   require a 2-log oocyst removal for drinking water systems that filter
    •   include Cryptosporidium in the watershed control requirements for unfiltered public water
       systems (Filtration Avoidance Criteria [FAC])
                                             4-2

-------
    •   require covers on new finished water reservoirs
    •   set other requirements that build upon the SWTR's treatment technique requirements

States are to

    •   conduct sanitary surveys for all surface water systems, regardless of size

The watershed control program for Cryptosporidium must identify watershed characteristics and ac-
tivities that may have an adverse effect on source water quality and monitor the occurrence of activities
that may have an adverse effect on source water quality. The state must determine whether the estab-
lished watershed control  program is adequate to limit  potential contamination by Cryptosporidium
oocysts.

The 1996 SDWA amendments also included four prevention approaches as part of establishing a new
charter for protecting the nation's public water systems: SWP, State Ground Water Protection, Capacity
Development, and Operator Certification. The SWP approach established a new Section 1453 for source
water quality assessments. States with public water supply (PWS) primacy were required to submit
source water assessment program plans for EPA approval. A state assessment program is required to
(1) delineate the boundaries of the areas providing source waters for public water systems, (2) identify,
to the extent practicable,  the origins of regulated and certain unregulated contaminants in the delin-
eated area, and (3) determine the susceptibility of public water systems to the identified contaminants.
Assessments are to be completed for all public water systems within two years after EPA approval of
the state's program. To avoid duplication, assessments may make use of sanitary surveys, state well-
head protection programs, pesticide state management plans, state watershed initiatives including ef-
forts under the SWTR, and efforts under the Federal Water Pollution Control Act, commonly  referred
to as the CWA. Section 1453 provides a number of additional features that may be used to assist the
state in promoting and developing SWP programs.

In support of the Microbial-Disinfection By-Products (M-DBP) rule-making process, the Information
Collection Rule (ICR) was promulgated (May 14, 1996; 61 FR 24354; effective June 18, 1996) to
collect occurrence and treatment information to help evaluate the need for possible changes to the
current SWTR and existing microbial treatment practices,  and to help evaluate the need for future
regulation for disinfectants and disinfection by-products (D/DBPs) (USEPA 1996a). The ICR  pertains
to large public water systems serving at least 100,000 people, and a more limited set of ICR  require-
ments pertain to ground water systems serving between  50,000 and 100,000 people. About 300 PWSs
operating 500 treatment plants were involved in the extensive data collection required under the rule.
Surface water systems were required to monitor for microbials, including bacteria, viruses, and proto-
zoa {Giardia and Cryptosporidium), and for disinfection by-products (DBFs), including trihalomethanes
(THMs) and haloacetic acids (HAAs). This rule is intended to provide EPA with information on the
occurrence in drinking water of microbial pathogens and DBFs. In addition, EPA collected engineering
data on how PWSs currently control such contaminants as part of the ICR.

Under the ICR, PWSs were required to monitor source  and treated water for the designated contami-
nants for a period of 18 months. The 18-month monitoring period started in July 1997. PWSs were
required to conduct finished water monitoring at any treatment plant at which  it detected, during the
first 12 months of monitoring, 10 or more Giardia cysts, 10 or more Cryptosporidium oocysts, or one
or more total culturable viruses per liter of water. The PWSs were to analyze finished water samples for
the same organisms analyzed in source water until 18 months of source water microbial  monitoring
were completed. The data were placed in the ICR Federal Database, available to the public at the
following Internet address: http://www.epa.gov/safewater/icr.html.

-------
Finally, consistent with the emphasis on source water protection, a rule to control public health risk
from contaminated ground water was included under the SDWA amendments of 1996. An informal
draft of the Ground Water Rule (GWR) preamble was posted on the Internet in February 1999. The
proposed GWR was published in May, 2000, for public comment (EPA 2000c). This rule specifies the
appropriate use of disinfection in ground water and addresses other components of ground water sys-
tems to assure public health protection. The GWR establishes multiple barriers to protect against
bacteria and viruses in drinking water from ground water sources and will establish a targeted strategy
to identify ground water systems at high risk for fecal contamination. The final GWR was scheduled to
be issued in November of 2000, but has not yet been promulgated.

The proposed GWR provides several requirements to assure public health protection. These are

    •   Sanitary surveys to be conducted by the state and identification of significant deficiencies
       (every 3 years for community water systems, 5 years for non-community water systems; this
       is consistent with the IESWTR).
    •   Hydrogeologic sensitivity assessments for undisinfected systems.
    •   Source water microbial monitoring by systems that do not disinfect and draw from
       hydrogeologically sensitive aquifers or have detected fecal indicators within the distribution
       system.
    •   Corrective action by any system with significant deficiencies or positive microbial samples
       indicating fecal contamination.
    •   Compliance testing for systems which disinfect to ensure that they reliably achieve 4-log
       (99.99%) inactivation or removal of viruses.

Full details of these requirements will be found in the final rule when published.

Threats to Source Water  Quality

Two major threats to source water quality with respect to DBF control and  microbial protection are
natural organic matter and microbial pathogens. The impacts, sources, and challenges to the manage-
ment of the former are discussed below. The remaining portions of this chapter will address microbial
contamination in more detail, in particular contamination by the pathogens Giardia and Cryptosporidium.

Natural Organic Matter and DBFs

DBFs occur due to the reaction of disinfectants with naturally occurring organic matter (NOM) that is
present in all surface waters. Under the Stage 1  Disinfectants/Disinfection By-Product Rule promul-
gated under the 1996  amendments to the SDWA, water utilities must reduce NOM concentrations,
expressed as total organic carbon (TOC), in their raw water to certain specified levels before chlorine is
applied for disinfection (Hoehn et al. 1994). Minimum TOC removal requirements vary according to
the source water TOC concentration and alkalinity, but range between 20-50% (Hoehn et  al.  1994).
Since the type and extent of required prechlorination treatment and the ability of a utility to meet the
maximum contaminant levels for THMs are dictated by the quality of the raw water, attention has
recently focused on understanding, characterizing, and controlling the sources of NOM (Hoehn et al.
1994; Stepczuk et al. 1998; Krasner et al. 1996; Minear and  Amy 1996).

Sources of NOM in receiving waters can be  broadly categorized as either allochthonous (originating
outside the receiving water) or autochthonous sources (originating within the receiving water).
Examples of the former include watershed sources such as soils, leaves, and plant remains that are
transported to the receiving water by runoff or by tributaries, while autochthonous sources include algal
matter, aquatic animals, and bacteria (Cooke and Carlson 1989; Cooke et al. 1988; Hoehn et al.  1994).
                                             4-4

-------
The relative importance of NOM sources to the total TOC and THM concentration will vary between
receiving waters; Hoehn et al. (1994) provide several examples for a variety of watersheds and receiv-
ing waters.  Past research has indicated that algae are as potent as humic and fulvic acids from
allochthonous sources (Graham et al. 1998; Hoehn et al. 1994) and suggests that, for eutrophic water
bodies subj ect to high nutrient loading from their watersheds, algae is likely to be the greatest source of
DBF precursors during the growing season (i.e., spring to fall). Recent modeling efforts by New York
City's Department of Environmental Protection for their Cannonsville Reservoir demonstrates the need
for nutrient  loading to be considered in the management control of THMs for eutrophic  reservoirs
(Stepczuk et al. 1998).

Anthropogenic loadings of nutrients into our nation's atmosphere  and aquatic and terrestrial
ecosystems have increased dramatically within the past few decades. Significant watershed loadings
are associated with both point and nonpoint sources. Examples of the former include municipal point
sources such as sewage treatment facilities offering secondary treatment that characteristically provide
minimal nutrient removal; storm water that is enriched from the wet and dry atmospheric deposition of
nutrients; combined sewers that discharge nutrient-enriched sanitary sewage  and rainwater; industrial
discharges; and particulate nutrients associated with runoff from construction sites. Nonpoint or dif-
fuse sources that can be locally important include runoff from overfertilized agricultural lands; animal
pastures and waste lagoons; storm water runoff from unsewered communities; septic tank and landfill
leachate; particulate nutrients from  sediment erosion; atmospheric deposition from mobile sources
(e.g., automobiles), power facilities, and confined animal-feeding operations; and nitrogen emissions
from receiving waters and terrestrial ecosystems.

Challenges to managing  the risk posed by  nutrients include the determination of which nutrient to
control and  by how much; the relative importance  of sources (i.e., the relative bioavailability of a
source's nutrient load); how the relative importance and abundance of these sources vary spatially and
seasonally; and the determination of where and when controls are most needed. Since it is typically the
dissolved form of nutrients that are most bioavailable (i.e., most capable  of fueling eutrophication),
many traditional point  and nonpoint source pollution controls that are aimed at removing solids and
solids-associated pollutants may be minimally effective at  controlling  nutrients. In addition, many
pollutant controls that remove selected pollutants (e.g., solids, metals) may inadvertently fuel eutrophi-
cation through the removal of non-nutrient growth factors (e.g., reduced turbidity removing light trans-
mission limitations). Prior to the successful management of nutrients from  both point and nonpoint
sources, information is required on the relative importance (i.e., bioavailability) of nutrient sources;
when (i.e., which season) controls need to be most  effective to prevent ecosystem overfertilization;
where in the watershed should controls be placed to maximize the cost-effective control; which pollu-
tion controls, best management practices (BMPs), and pollution prevention techniques are most effec-
tive at removing the bioavailable forms of nutrients during the critical periods when these loads make
their maximum contribution to overfertilization; and  the costs and cost effectiveness of these controls,
practices, and techniques.

Protocol presently exists for determining numeric nutrient loading targets for a given waterbody (USEPA
1999c); however, the process is not a  straightforward one. Research is currently planned that will
determine which nutrient(s)  to control  and by how  much for the nation's ecoregions (Garber et al.
1999). Once the nutrient(s) that limit eutrophication have been determined;  numeric targets defined;
continuous,  episodic, and seasonal inputs of natural and anthropogenic sources characterized; and cy-
cling processes identified and their relative importance understood, managers can develop waste load/
load allocations and a management plan aimed at achieving the desired reductions for the identified
sources. Management options for the control of nutrient sources include point-sources controls (e.g.,
upgrades at water pollution control plants, emissions controls, etc.); the use of structural and nonstructural
BMPs for the control or treatment of nonpoint and diffuse sources; land use controls aimed at decreas-

                                               4-5

-------
ing population density, protecting vulnerable areas, or maintaining the assimilation capacity of natural
ecosystems; and the restoration of ecosystems capable of intercepting and assimilating nitrogen loads
(e.g., riparian zones, forests, or wetlands).

Although BMPs are often employed to treat nonpoint sources of watershed pollutants, including nutri-
ents, significant uncertainty is associated with their ability to control this  stressor with removals
ranging from 10-90% for some of the more common structural BMPs (Griffen 1993). For this reason,
nutrient controls are often targeted at point sources where less uncertainty  is associated with both
expected removals and costs. Non-site specific factors that may influence the effectiveness of BMPs
includes their age, capacity, maintenance, and design specifications. Watershed-specific characteristics
that can influence effectiveness include soil characteristics; land use;  land cover; climate; site location
relative to receiving waters; soil processes and ground water hydrology that can  influence pollutant infil-
tration, decomposition, adsorption, and transport; and biogeochemical processes that may differ between
drainage basins (Fisher et al. 1992). When many BMPs are applied to different locations within a water-
shed, it is still more difficult to predict their integrated effects,  and few studies have examined BMP
effectiveness for nutrient control on a watershed scale (Edwards et al. 1997; Griffin 1995).

In watersheds where surface waters have been degraded by excessive nutrient inputs, land-use controls
are often recommended as a means by which to limit future point and nonpoint nutrient inputs (Minei
and Dawydiak  1997). Common controls include the purchasing of farmland  development rights; the
conservation of forests, wetlands, and riparian lands; and changes in zoning. Where available, water-
shed models calibrated to actual data or regional  or national estimates are often used to predict the
pollution potential of various land-use scenarios (Houlahan et al. 1992; Preston 1996; Corbett et al.
1997; Valiela et al. 1997).  However, as with BMPs, there may be considerable uncertainty associated
with these "alternative futures  analyses," in particular where models rely on national or nonlocal esti-
mates of export coefficients.

Restoration of natural features (e.g., riparian forests and wetlands) are often part of management
plans aimed at  controlling the  transport  of nutrients to receiving waters. However, the effectiveness
of these features at capturing nutrients from upland land uses can be influenced by a number of
factors including the  magnitude of loadings relative to ecosystem structure  (Hopkinson 1992); the
relative distribution of natural ecosystems, e.g., uphill versus downhill (Correll et al. 1992); and the
infiltration or contact between ground water and root systems (Peterjohn and Correll 1986).

Finally, as the focus of controls shift from point to nonpoint management, the behavior of urban and
suburban private land owners may often  determine the success of nonpoint and diffuse source control
efforts. Although there is recent awareness that economic and social considerations play an important
role in the success of nutrient management efforts, few studies have evaluated the role of values, knowl-
edge, income, or other circumstances in  an individual's nutrient use and disposal, or the effectiveness
of education and economic or other incentives aimed at reducing nutrient loads.

Pathogen  Contamination

The potential sources of pathogens in source water are many and varied including nonpoint source runoff,
discharges from treated and untreated sewage, and combined sewer overflows.  From a waterborne out-
break and public health viewpoint, both Giardia and Cryptosporidium  are of primary concern.

Cryptosporidium is ubiquitous in the environment. Runoff from unprotected watersheds and treated
and untreated sewage discharges transports these microorganisms to water bodies used as intake sites
for drinking water supplies. Oocysts resist inactivation by commonly used disinfection practices and
temperature extremes (Payer 1994; Payer and Nerad 1996). As indicated above, Cryptosporidium in
                                              4-6

-------
source water, particularly source water serving unfiltered surface water systems, requires special atten-
tion mandated by EPA's IESWTR (USEPA 1998).

In the U.S., Giardia is the most commonly identified pathogen in waterborne disease outbreaks (LADWP
1996). Contamination of a water supply by Giardia can occur in two ways: by the activity of animals,
particularly beavers, in a watershed or by the introduction of sewage into the water supply.

For many years, detection and enumeration methods for microbial agents in water focused largely on
sanitary indicator bacteria, primarily the total and fecal coliforms, E. coli and fecal enterococci. Bacte-
rial pathogens such as Salmonella, Shigella, E. coli 0157:H7, and Campylobacter have received some
attention due to waterborne illness outbreaks. However, other bacteria, viruses, or protozoan pathogens
received very little attention until waterborne outbreaks caused by them were documented.

The occurrence of waterborne disease outbreaks has been the key driver of sanitary microbiology re-
search throughout the past 100 years in the U. S. The most recent waterborne pathogens to arrive on the
scene have been the pathogenic protozoa, Giardia lamblia and Cryptosporidiumparvum. Contamina-
tion of the soil and aquatic environments occurs through shedding of Giardia cysts and Cryptosporidium
oocysts by infected animals, including man. Control of the occurrence of these pathogens in water-
sheds and their surface waters will be difficult since many animals have been  shown to be infected by
these organisms, and human sewage contains sizeable concentrations of cysts and oocysts depending
on the level of infection in the community. There may be other waterborne protozoan pathogens to be
concerned about as indicated by the research being stimulated by the Contaminant Candidate List,
finalized in  1998 (63 FR 10274) by the EPA Office of Ground Water and Drinking Water (USEPA
1998b). This section presents background information on waterborne outbreaks due to Giardia and
Cryptosporidium and on the occurrence of these organisms in surface waters used as drinking water
sources,  storm water run-off, sewage, and combined storm water-sewage overflows (CSOs).

Giardiasis outbreaks were gradually recognized during the period from  1961-1980 (Craun 1986). Di-
agnosis was by fecal examination  of patients, and there was no suitable method for detection of the
cysts in environmental water samples. The first identified cryptosporidiosis outbreak occurred in the
United Kingdom (U.K.)  in 1983, while the first U.S. outbreak of cryptosporidiosis occurred in Braun
Station, TX, in 1984 (Lisle and Rose 1995). Giardia and Cryptosporidium were both formerly thought
to be harmless  commensals, and it took some  time for sufficient information to be developed  that
showed them to be disease agents. Since 1984, there have been numerous outbreaks of waterborne
cryptosporidiosis, including the massive Milwaukee, WI, outbreak in 1993 that  affected 403,000 people.
In addition,  Giardia continues to  be one of the most frequently identified  etiologic agents of gas-
trointestinal  illness due to contaminated drinking water.

The number of waterborne giardiasis and cryptosporidiosis outbreaks that have occurred in the U.S.
since 1960 are shown in Table 4-1. The outbreak data in Table 4-1  include both drinking water and
recreational water outbreaks. Cryptosporidium oocysts, although much smaller (4-6 |im) than Giardia
cysts (10-14 |im), behave much the same as Giardia cysts with regard to physical  removal processes in
drinking water treatment, but are much more resistant to chemical disinfection than are Giardia cysts.
However, to assure maximum removal of Cryptosporidium oocysts during water treatment, the physi-
cal processes must be optimized and consistently operated. Cysts and oocysts  can be detected  in a
sample using the same methodology, although the detection methodology needs much improvement.
The number of reported  Giardia outbreaks tended to increase from year to year once it was acknowl-
edged as a waterborne pathogen. With Cryptosporidium., no such trend has been evident, but this may
be due to the lack of a good detection method or to other, as yet poorly understood, factors including
environmental survival of oocysts, viable and noninfective versus viable and infective oocysts, a high
rate of inapparent infections, and infective dose variation due to both the strain of Cryptosporidium
parvum and individual host susceptibility.

                                              4-7

-------
Table 4-1. Summary of U.S. Drinking Water and Recreational Waterborne Disease Outbreaks
          Due to Giardia and Cryptosporidium
Date
1961-1965
1966-1970
1971-1975
1976-1980
1981-1988

1989-1990

1991-1992

1993-1994

1995-1996

Parasite
Giardia
Giardia
Giardia
Giardia
Giardia
Cryptosporidium
Giardia
Cryptosporidium
Giardia
Cryptosporidium
Giardia
Cryptosporidium
Giardia
Cryptosporidium
No. Outbreaks
1
2
13
26
120
2
7
0
8
5
9
9
3
6
No. Cases
123
53
5,136
14,416
573
14,966
697
0
157
3,526
526
403,930
1,536
8,572
References
Craunetal. 1986
Craunetal. 1986
Craun et al. 1986
Craunetal. 1986
Herwaldt et al. 1991
Lisle and Rose 1995
Herwaldt et al. 1991
Herwaldt et al. 1991
Moore etal. 1994
Moore et al. 1994
Kramer etal. 1996
Kramer etal. 1996
Levy etal. 1998
Levy etal. 1998
Several factors likely account for the increases in the number of reported waterborne outbreaks of both
giardiasis and cryptosporidiosis, including (1) recognition that Giardia lamblia and Cryptosporidium
parvum are human pathogens; (2) recognition as waterborne pathogens (first recognized waterborne
giardiasis outbreak, 1964-65, first recognized cryptosporidiosis outbreak, 1983 in the U.K. and 1984
in the U.S.); (3) improved detection methodology; and (4) improved surveillance and reporting. Good
background articles on Giardia and Cryptosporidium in water were written by Lin (1985) and Rose
(1988), respectively. An extensive review of Cryptosporidium spp. and cryptosporidiosis in animals
was published by Payer and Ungar (1986) and Payer (1997). Marshall et al. (1997) published a review
of waterborne protozoan pathogens that includes Cryptosporidum parvum, Giardia lamblia, and six
other protozoans as well as a section on water quality protozoan testing and monitoring. Craun et al.
(1998) reviewed 35 waterborne cryptosporidiosis outbreaks associated with contaminated drinking
water and recreational activities,  provided recommendations for prevention of  such outbreaks, and
assessed the need for epidemiological data.

Pathogens in the Environment and in  Wet Weather Flow

Cysts and oocysts are common in surface water, and the concentration appears to vary with watershed use
characteristics (Hansen and Ongerth 1991). It has been established that oocysts are found in most surface
waters as shown in Table 4-2. Hansen and Ongerth (1991) found oocysts in 34 of 35 river samples, using
a method with a detection limit ranging from 0.04 to 0.14 oocysts per liter and a recovery efficiency of
18.6 to 34.3%. The watersheds examined had a variety of nonurban land uses. In a study conducted on the
Allegheny River, Cryptosporidium was detected in 50% or more of all samples collected (NRCS 1997).
Samples were collected over a 31/2 year period, with a recovery efficiency of just 25%. Roughly 22% of the
samples collected in the New York  City watershed showed Cryptosporidium oocysts, with a slightly
greater fraction showing Giardia cysts (Stern 1996). The background concentration in one drinking water
reservoir was estimated at 0.36 oocyst/100 L (Stewart et al. 1998).

Research to determine correlations between Giardia and Cryptosporidium and other parameters  has
been inconclusive.  LeChevallier et al. (199la) measured  Giardia and Cryptosporidium in the source
waters of 66 surface water treatment plants in the U. S. and Canada. They identified oocysts in 87% of
                                             4-8

-------
Table 4-2. Occurrence of Giardia Cysts and Cryptosporidium Oocysts in Surface Waters
No. of
Water Type Samples
Surface 5 1
Rivers/lakes 181
Allegheny R. 24
Youghiogheny R. 24
Stream, 24
dairy farm
River diversion 1 9
Lake outlet 20
Stream/river 1 1
Surface 107
Reservoir inlet 60
Reservoir outlet 60
Surface water 85
River/stream 6
canal water
Raw source 262
waters
% Samples
Positive
Giardia
39
15
63
54
54
21
40
—
-
13.3
15
81
—
45
% Samples
Positive
Crypto.
39
51
63
63
82
50
50
77.6
77
5
11.7
87
ngA
51.5
Cone. Range
Per L (GM)#
Giardia
—
O.01-1.4
(0.03)*
0^.2 (0.34)
0-5.3 (1.2)
0-15.7
(0.82)
0-6.25 (0.22)
0-2.22 (0.08)
—
-
0.007-0.24
(0.19)
0.012-1.07
(0.061)
0.04-66
(2.77 )
—
0.02-43.8
(2.0)
Cone. Range
Per L (GM)#
Crypto.
—
O.01-44
(0.43)*
0-22.3 (0.31)
0-14.7 (0.58)
0-11.05
(0.42)
0-240 (1.09)
0-22 (0.58)
2-112(25.1)*
0.04-18 (0.91)
0.007-0.024
(0.012)
0.017-0.31
(0.081)
0.07-484
(270)
0.8-5,800 (ng)
0.065-65.1
(2.4)
Reference
Barthe and Brassard
1996
Roseetal. 1991
States etal. 1997
States etal. 1997
States etal. 1997
Roseetal. 1988
Roseetal. 1988
Ongerth and Stibbs
1987
Rose 1988
LeChevallier
etal. 1997
LeChevallier
etal. 1991a
LeChevallier
etal. 1991a
Madore
etal. 1987
LeChevallier
and Norton 1995
# GM = geometric mean
* = arithmetic mean
A ng = not given
sampled surface waters, reporting higher densities in waters receiving industrial or sewage effluents
and also significant correlations between Giardia and Cryptosporidium concentrations with turbidity
and fecal coliform concentrations. Giardia and Cryptosporidium concentrations reported in 39% of the
surface waters sampled in Canada showed no correlation with either total or fecal coliform concentra-
tions, heterotrophic plate count, pH, temperature, turbidity, or dissolved organic carbon (Barthe and
Brassard 1996). One factor to consider in explaining this inconsistency is that reported oocyst concen-
trations included both viable and nonviable organisms (LeChevallier et al. 1991b).

Anational study detected Giardia spp. in81%of source water samples from 66 surface water treatment
plants in 14 states and one Canadian province.  Cryptosporidium spp. were found in 87% of the raw
water locations. Higher cyst and oocyst densities were associated with source waters receiving indus-
trial or sewage effluents. Significant correlations were found between Giardia and Cryptosporidium
densities, turbidity, and total  and fecal coliform  levels. Statistical modeling suggests that cyst and oo-
cyst densities could be predicted on the basis of watershed and water quality characteristics (LeChevallier
etal. 1991a).
                                              4-9

-------
Concentrations of Cryptosporidium and Giardia in the Delaware River, a drinking water source for
several municipalities including New York, NY, Philadelphia, PA, and Trenton, NJ, increased after
rainfall events. The increased Cryptosporidium and Giardia concentrations correlated with increased
coliphage, total coliform, fecal coliform, E. coli, and C. perfringem concentrations. The increase was
attributed to transport through surface runoff, resuspended storm drain, and river bottom sediments
(Atherholt et al. 1998).

Giardia cysts were found in 94 (43%) of the 222 samples collected over a nine-month period from 17
sampling stations from three pristine rivers in the Pacific Northwest (Ongerth  1989). No statistically
supportable seasonal variations were found. Giardia cysts were continuously present, though at low
concentrations, even in relatively pristine rivers (Rose  et al. 1991; Rose 1997).

Giardia cysts and Cryptosporidium oocysts have been found at low levels in ground waters and springs,
as summarized in Table 4-3. In general, contamination of well waters  appears more likely for
Cryptosporidium oocysts than for Giardia cysts, but well depth, construction, and state of repair will
strongly influence the possibility of contamination. Regardless of the type of well or spring, cyst and
oocyst concentrations were usually found to be low.
Table 4-3. Giardia and Cryptosporidium in Springs and Ground Waters
Source
No. Samples
or Sites % Samples Pos.
Crypto. Giardia
Well waters
Ground waters
Spring, pristine
Vertical wells
Springs
Infiltration galleries
Horizontal wells
Total sites
Deep well, pristine
Well, coliform positive
20
18
7
149
35
4
11
199
288
138
0
-
0
1
14
25
36
12
-
-
0
5.5
57.1
5
20
50
45
12
0
5.8
Range Cyst/Oocysts/lOOL References
Giardia Crypto.
Barthe and Brassard
1996
O.25 (0.3)* Roseetal. 1991
O.25 <0.25-13(4) Rose etal. 1991
Hancock etal. 1998
Hancock etal. 1998
Hancock etal. 1998
Hancock etal. 1998
0.1-120(8) 0.2-45(5) Hancock et al. 1998
Bentonetal. 1991
4-92(23) Badenoch etal. 1990
- = data not given
* = arithmetic mean
Sources of Oocysts

Giardia cysts and Cryptosporidium oocysts are found at significant levels in domestic raw sewage,
treated sewage effluents, and CSOs. Table 4-4 summarizes data on Giardia and Cryptosporidium cysts/
oocysts in sewage and CSOs. Source identification and characterization play an important role in deter-
mining potential control measures. For example, SWP measures for oocysts from human sewage ver-
sus animal sources will be different. Even if the cysts and oocysts are known to be from human sewage,
there may still be considerable differences in control options available, depending on whether the cysts
and oocysts were discharged due to faulty septic systems, wastewater treatment plant effluent, treat-
ment plant bypass, sanitary sewer overflows (SSOs), CSOs, or storm water. Similarly, significant dif-
ferences in options occur if the animal source of the oocysts is from a dairy farm, cattle ranch, a concen-
trated feed, or wild animal populations.
                                             4-10

-------
Table 4-4. Giardia and Cryptosporidium Cysts/Oocysts in Sewage and Combined Sewer
          Overflows
% Samples Positive
Source
Raw sewage
Primary
effluent
Final effluent
Sewage4
influent
Return act.
sludge
STP trick.
filter
Raw sewage
STP effluents
Raw sewage
STP effluents
Sewage effluent
Raw sewage
Treated sewage
Combined
overflows, upper,
in stream, d.w.°
Lower, in stream,
d.w.
Upper, in stream,
w.w.°
Lower, in stream,
w.w.
End of pipe
CSO
No. of
Samples
29
37
33
24
8
8
-
-
3-36
—
15
4
9
6
6
3
3
11
5
Giardia
100
100
82
100
100
100
-
-
100
—
80
—
—
100
100
100
100
100
80
Crypto.
0.03
_b
0
100
100
100
-
-
14
—
27
100
—
100
100
67
100
100
"
Range Cyst/Oocysts per L(GM)a
Giardia
130-7900
(l,500)a
150-6,600
(l,100)a
4-130 (14)
200-3,200 (ng)
200-900 (ng)
4-44 (11)
11-397 (ng)
0.01-13.5 (ng)
26-3,
022 (ng)
2-3,
511 (ng)
0-4,
614 (42)
—
—
<0. 13-0.66
(36)
0.21-66
(3.43)
0.67-2.88
(1.15)
4.29-75
(26.5)
90-2,830
(354)
37-1,140
(287)d
Crypto.
0-28(ng)
—
—
—
—
—

-
0-74(ng)
0-333(ng)
0-4,
927 (43.2)
850-13,700
(51.8)b
140-3,960
(l,060)b
0.05-0.53
(0.18)
O.33-1.05
(0.78)
0.39-0.72
(0.70)
4.29-1.77
(7.5)
2.5-400
(60.4)
8.8-30
(20.1)d
Reference
Hirata and Hashimoto
1997
Hirata and Hashimoto
1997
Hirata and Hashimoto
1997
Cassonetal. 1990
Cassonetal. 1990
Casson et al. 1990
Sykoraetal. 1987
Sykora et al. 1987
Roach etal. 1993
Roach etal. 1993
States etal. 1996
Madore et al. 1987
Madore et al. 1987
Gibson etal. 1998
Gibson etal. 1998
Gibson etal. 1998
Gibson etal. 1998
Gibson etal. 1998
States etal. 1997
  a geometric mean number of cysts/oocysts/L; ng = not given
  b arithmetic mean
  0 d.w. = dry weather; w.w. = wet weather
  d 8-hour composite samples
                                              4-11

-------
There is little information on septic tanks as a potential source ofCryptosporidium. Septic tanks that
function poorly are possible sources of oocysts and need to be addressed for public health reasons,
including Cryptosporidium. The New York City Department of Environmental Protection is conduct-
ing a study on the transport of oocysts from functioning septic systems, and a report on its findings was
to be available in December 1999 (USEPA 1997).

A variety of mammals, particularly young ruminants, are sources of Giardia cysts and Cryptosporidium
parvum oocysts in the environment. Table 4-5 presents some information on concentrations
cysts and C. parvum oocysts in the feces from humans and some animals.
Table 4-5. Some Human and Animal Sources of Giardia Cysts and Cryptosporidium Oocysts
% Samples Pos
Source No. Samples Giardia Crypto.
Cysts Oocysts
Giardia Crypto.
Reference
Human
Infected - - -
AIDS patient, - - -
infected
3 x 103/
person/day
6 x 106-
1.2 xlO10
Erlandsen and Meyer
1984
Goodgame et al.
1993
Agricultural
Calves/lambs - - -
Calves, infected - - -
Cow, infected - - -
Cattle, infected 108 - 26.8
Swine, infected 90 - 34.4
1010/day
to 14 days
lOVg,
5-15 Kg
feces per day
104/g,
25-30 Kg
feces per day
-
-
Current and Garcia
1991
Breach et al. 1994
Breach et al. 1994
Quilezetal. 1996
Quilezetal. 1996
Parks/ Recreation al
Beaver - - -
— —
Erlandsen and Meyer
                                                                          1984
Muskrat
                                            Erlandsen and Meyer
                                            1984
Canada geese
100
77.7
75-7867
g feces
67-6867
g feces
Graczyketal. 1998
  *pooled sample
Eighty species of mammals have been shown to shed C. parvum oocysts (Barry et al.  1998). Most
measurements have been completed on domestic animals, with little information available regarding
the shedding of C. parvum by wildlife. In addition to humans, among the domestic and wild animals
found to be hosts for C. parvum are cattle (Atwill et al.  1998; Xiao and Herd 1994; Kuczynska and
Shelton 1999;  Garber et al. 1994), sheep, goats, deer, water buffalo, pigs (Atwill et al.1997), horses
(Forde et al. 1997; Haas and Rose 1994; Johnson et al. 1997), rabbits, opossum, rodents (rats, Webster
and McDonald 1995; mice, Klesius et al. 1986; Bajer et al. 1997), beaver and muskrats (Bajer et al.
1997), migratory water fowl, and primates (Graczyk et al. 1998b).
                                             4-12

-------
Oocyst Survival

The ability of oocysts to survive rather harsh environments (e.g., low temperatures, typical drinking
water chlorination) enhances their chances of successfully migrating to a treatment plant intake and
through the treatment process. Understanding conditions that oocysts can and cannot tolerate can be
instrumental in devising effective controls or in estimating when high levels of oocyst survival will
occur in source waters (Walker 1998; Graczyk et al. 1998a; Payer and Nerad 1996).

Payer and Nerad (1996) have shown that,  although freezing at very low temperatures (-70°C) inacti-
vated oocysts, freezing at higher temperatures (-10, -15, and -20°C) allowed oocysts to retain  some
level of infectivity. Oocysts frozen at -20°C for five hours or less remained infective. Oocysts frozen at
-10°C for 168 hours or less, as well  as those frozen at -15°C  for 24 hours or less, also remained
infective. Although freezing temperatures  are detrimental to oocyst survival, this study suggests that,
when the surface does not reach low temperatures  (below -10°C) for prolonged periods of time,  some
infective oocysts may survive for extended periods.

Jenkins et al. (1999) performed field studies of oocysts exposed to the environment of calf manure piles
and the surface of a field soil. Results of this study indicated that exposure to both manure and soil
environments significantly increased rates of oocyst inactivation compared to controls. Exposure  to
freeze-thaw cycles in soil were particularly deleterious to the oocysts. They concluded from their  study
that spreading manure contaminated with oocysts  on snow, in an absence of freeze-thaw cycles, may
contribute to sustained oocyst survival and increase the risk of surface water contamination during
spring melt and runoff.

Payer (1994) examined the effect of high temperatures on  oocyst infectivity. Oocysts were rendered
noninfective within one minute upon reaching a temperature of 72.4°C or higher. Oocysts held at 64.2°C
or higher for 2 minutes also lost their infectivity. This study was conducted on oocysts in distilled
water. It is possible that temperatures needed to inactivate oocysts on land or in compost may vary.
Jenkins et al. (1999) suggest that oocyst infectivity may be significantly reduced within 70 days  in
manure piles with temperatures between 35 and 50°C.

Jenkins et al. (1998) reported that low concentrations of ammonia associated with a barnyard environ-
ment inactivated oocysts and can inactivate a viable population in days. They also demonstrated that
the pH associated with the various levels of ammonia tested, pH 9 to 11, was not a factor associated
with their inactivation.

Results of one study (Chauret et al. 1998) examining the role of biological antagonism in the inactiva-
tion of oocysts suggest that biological antagonism may be a primary factor affecting oocyst survival in
natural waters. However, this process of natural interactions between organisms appears to be site
specific.

Measuring and Monitoring Pathogens in Source Waters

Except for the use of immunological and molecular methods (genetic probes, polymerase chain reac-
tion [PCR], ribotyping) for specific identification of isolates of pathogenic bacteria, cultural methods
for the detection, enumeration, and identification of waterborne pathogenic bacteria have not changed
significantly over the past three decades. Although somewhat dated, Singh and McFeters (1992) re-
viewed detection methods for pathogens in water. A comprehensive source of information on environ-
mental microbiology and microbial detection methods may be found in Hurst et al. (1997).

Although most water monitoring involves searching for indicators of fecal pollution, monitoring water
for the presence of pathogens is necessary under special circumstances, such as during and after water-
                                             4-13

-------
borne outbreaks, when dealing with a water source with a history of contamination, or where waste-
water reclamation is involved. The low densities of pathogens usually found in water require that large
volumes of water must be examined. For viruses and parasites, this is usually done by filtering a large
volume (10-1000 liters [2.6-264 gal]) of water through a filter cartridge to concentrate the target or-
ganisms. Sometimes volumes of one liter or more are concentrated by centrifugation or a combination
of filtration and centrifugation. The use of large volume samples limits the number of samples that can
be examined and increases the costs of testing. Overall, the costs for analysis of water samples for
specific bacterial pathogens and for enteric viruses and protozoan pathogens are quite high, with those
for viruses and protozoan cysts and oocysts being much higher than for bacteria. Given the limitations
of detection and enumeration methodologies and their complexities, a negative result for finding a
specific pathogen does not mean that no target pathogens were present, only that none were detected at
the detection limit of that method.

Giardia and Cryptosporidium

The methods currently in use for Giardia and Cryptosporidium detection in water have been developed
since the early 1980s. Since 1992, with the development of a series of regulations (D/DBP Rule, IESWTR,
and the ICR), water utilities have been in need of water quality testing laboratories, either in-house or
via contract, with the capability of analyzing for Giardia cysts and Cryptosporidium oocysts in finished
drinking water and in source waters. The method of choice in the U.S. for detection of Giardia and
Cryptosporidium in source waters was the proposed American Society for Testing and Materials (ASTM)
analytical procedure. The method is technically complex, labor intensive,  time consuming, and
requires good  laboratory quality-control procedures to provide maximum recoveries. The method
performance is  also affected by variations  in sample collection, water quality (turbidity) and analyst
training, experience, and competency.

The EPA method used for the ICR (USEPA 1996) differs from the ASTM method by requiring filtra-
tion of 100L (26.4  gal) of raw water or 1000 L (264 gal)  of finished water and the use of Hoffman
modulation or differential interference contrast (DIG) optics instead of phase-contrast optics for confir-
mation of morphological characteristics of the presumptive cysts and oocysts. Because of method per-
formance issues, modifications to the EPA  method resulted in the development of EPA Methods 1622
and 1623 (USEPA  1999a, 1999b), respectively. Each method uses sample concentration by filtration,
combined with  immunomagnetic separation and fluorescent antibody staining for recovery  and enu-
meration of cysts and oocysts. Method 1622 is a stand-alone method for Cryptosporidium, while Method
1623 is for simultaneous detection of Giardia and Cryptosporidium.

Enteric Viruses

Viruses are present in very low numbers in most environmental waters. Therefore, methods for the
detection of enteric viruses in water, as for the protozoan pathogens, also require concentration of the
viruses from a large volume water sample following a protocol involving filtration and centrifugation,
recovery of the viruses from the filtration medium, and assay of the concentrated sample for viruses by
inoculation into a mammalian cell culture line. The methods published in the USEPA Manual of Meth-
ods for Virology (Berg et al. 1983) and  in Standard Methods for the Examination of Water and
Wastewaters (APHA 1999) were probably the most commonly used prior to the ICR method. However,
the ICR virus monitoring protocol, developed by the EPA and modified by consensus agreements from
the scientific community (USEPA 1996), represents the methodology  most used during the 1990s for
detecting enteric viruses in water.
                                            4-14

-------
Protecting Source Waters

This section discusses protection of source water from microbial pathogens found in treated sanitary
sewage and wet weather flows (i.e., SSOs, CSOs, and storm water runoff).

Separate Sanitary Sewage Systems

Separate sewage systems require a dedicated infrastructure to carry waste to the treatment plant. Typi-
cally, these systems are largely gravity flow-augmented by pumping stations if needed. The system is
designed to  meet specified flow quantities, and balances flow from all influents with the treatment
plant throughput capacity. When the demand exceeds the flow capacity of the system, a surcharge, or
SSO, occurs. Under surcharge conditions, the system discharges through alternate escape routes, often
backing up into residences or streets, and eventually winding up in receiving waters. Separate sewage
systems are seldom leakproof. Connections between pipe sections along the length of the conveyance
system are not completely sealed. The connections and the privately owned laterals offer opportunities
for waste to escape and for subsurface water to infiltrate the  system. Inflow and infiltration can be
substantial during rain events, decreasing the flow capacity available for the wastewater.  Table 4-6
presents representative data on the  type and number of microorganisms found in untreated wastewater
(Metcalf & Eddy Inc. 1991). The table reports densities of both indicator and pathogenic microorganisms.

Table 4-6. Types and Numbers of Microorganisms Typically Found in Untreated Domestic
          Wastewater
Organism
Total coliform
Fecal coliform
Fecal streptococci
Enterococci
Shigella
Salmonella
Pseudomonas aeroginosa
Clostridium perfringens
Mycobacterium tuberculosis
Protozoan cysts
Giardia cysts
Cryptosporidium cysts
Concentration (number/mL)
105-106
104-105
103-104
102-103
Present3
10°-102
lO'-lO2
lO'-lO3
Present3
lO'-lO3
lo-'-io2
lo-'-io1
Helminth ova 1 0~2- 1 0 '
Enteric virus
lO'-lO2
  a Results for these tests are usually reported as positive or negative rather than being quantified.

Concentrations of microorganisms in sewage treatment plant effluent vary depending on the National
Pollutant Discharge Elimination System (NPDES) permit issued for the plant (King 1996). The efflu-
ent concentrations and level of treatment required are those necessary to achieve receiving water qual-
ity standards. Receiving water standards have been established pursuant to the Federal Water Pollution
Control Act Amendments of 1972 (Public Law 92-500) to protect beneficial uses of surface waters.
One goal is to eliminate pathogens to control transmission of waterborne diseases. In support of this
goal, wastewaters that pose a disease risk are disinfected prior to discharge. Generally, NPDES permits
require measuring the microbial indicator concentrations in the effluent rather than pathogen concen-
                                             4-15

-------
trations. Therefore, treating wastewater so that the permit standard is met does not guarantee an ab-
sence of pathogenic microorganisms. Indicators are more representative of some pathogens than oth-
ers. Olivieri et al. (1977) found that, in raw sanitary  sewage, there was a strong positive correlation
between the levels of total coliform (TC), fecal coliform (FC), fecal streptococci (FS), and enterococci
and the levels of pathogenic bacteria. Only the levels of TC and FC correlated well with the levels of
enteric viruses. Metcalf et al. (1995) reported that, on average, virus concentrations of about 50 plaque
forming units/liter (PFU/L) can be expected in wastewater treatment plant effluents.

Combined Sewer Systems

The design of combined sewer systems commingles sanitary and storm water flows in a single convey-
ance system that routes the entire flow to  the wastewater treatment plant. This system treats all
collected liquid, including storm water and sanitary sewage, before discharge to the receiving water. By
using a single conveyance to carry all flows to the treatment plant, this design requires a total pipe
length less than that required by separate storm and sanitary sewers.

At construction, the combined sewer system is sized to meet the existing and projected flows of sani-
tary (dry-weather flow [DWF]) and storm water flows. Design values are available for the sanitary
contribution from various types of structures  (e.g., hospital, school, or private home). The DWF vol-
ume varies over the course of a day, with morning and evening peaks. Weekend flow patterns differ
from the work-week flow patterns. Designers base the storm water flow contribution on regional rain-
fall statistics and the probability of a given storm-induced runoff volume. Whenever storms generate
runoff to create combined sewage flow volumes greater than the capacity of the system, the system
relieves the pressure by shunting flow to receiving waters, i.e., a CSO occurs. The total  flow of over-
flow events is often expressed as a multiple of the peak DWF.  As combined sewer systems age, the
number of sanitary users connected to the system increases. The increased sanitary flow volumes de-
plete capacity formerly used by storm water  and increase the frequency of overflows. Similarly, the
increased impervious area associated with the new connections increases the storm water runoff, which
also consumes conveyance capacity.

EPAs CSO Control Policy (USEPA 1994) limits the  number of annual  overflows for combined sys-
tems and requires disinfection  after primary  clarification, using the capacity of the publicly owned
treatment works, when it is required by local authorities. Systems are being modified to reduce the
number of overflows by  providing for in-system storage, on-lot storage (Milliken 1996), and discon-
necting inputs such as downspouts. Low-impact development is a method currently being evaluated for
reducing runoff volumes (Coffman et al. 1996). Its objectives include restoring the site hydrologic's
regime to reflect the natural or predevelopment condition and minimizing the generation and off-site
transport of pollutants via storm water runoff.

CSO disinfection is practiced to control the discharge of pathogens and indicator microorganisms into
receiving waters. Chlorination  is the  conventional approach to disinfection.  Due to concerns  about
chlorine's effects on aquatic life, alternative technologies are being investigated for CSO disinfection.
The New York City Department of Environmental Protection recently completed evaluations of high-
rate disinfection  technologies (Camp, Dresser, & McKee and Moffa & Associates 1997). Table 4-7
shows the disinfectant dosages associated with achieving effluent microbial indicator concentrations of
1000 colony forming units (cfu)/100 mL using chlorine, ultraviolet (UV) irradiation, ozonation, and
chlorine dioxide. Generally, all four technologies were able to provide 3- to 4-1 og bacterial reductions.
UV disinfection was found to provide reduced effectiveness at higher total suspended solids concentra-
tions (>150 mg/L). Chlorine dioxide disinfection requires doses significantly lower than those required
for chlorine disinfection, reducing the toxicity impacts on the receiving water aquatic life.
                                             4-16

-------
Table 4-7. Estimated Disinfection Dosages to Attain Microbial Indicator Concentrations of
           1000 cfu/100 mL (Camp, Dresser & McKee and Moffa & Associates 1997)
Estimated Dosages

Total coliform
Fecal coliform
E. coli
Enterococcus
Influent
Concentration
(cfu/100 mL)
106-107
105-106
105-10 7
104-106
Chlorine
Dose (mg/L)
>30a
18
17
22
UV Dose
(mW-s/cm2)
>100b
50
35
35
OzoneDose
(mg/L)
37
24
23
12
Chlorine Dioxide
Dose (mg/L)
>8C
6
5.5
5.5
   Target concentrations not achieved with highest applied dosage, i.e. 30 mg/L.
  b Target concentrations not achieved with highest applied dosage, i.e. 100 mW-s/cm2.
  0 Target concentrations not achieved with highest applied dosage, i.e. 8 mg/L.


Particles associated with or occluding microorganisms can reduce the effectiveness of wastewater dis-
infection by chlorination and by UV irradiation (Parker and Darby 1995). UV irradiation showed lower
effectiveness at suspended solids concentrations above 150 mg/L in the New York City studies (Stinson
et al. 1998). Understanding the effects of solids content on disinfection effectiveness is necessary for
designing treatment systems capable of achieving effluent requirements. Recent EPA research suggests
greater disinfection effectiveness is possible by removing solids before UV irradiation and chlorination
(Perdek and Borst 2000).

Lijklema et al. (1986) report that CSOs result in increased indicator organism concentrations in the
receiving water. They measured TC, FC, E. coli FS, somatic coliphages,  and F-specific coliphages.
Phage concentrations are  one to three orders of magnitude smaller than bacterial concentrations. The
ratio between the concentrations of different bacterial indicators varies between events, but is generally
within 1.5-log units. Ellis  and Yu (1995) report that CSOs serve as very effective generators of bacteria
and pathogens in urban receiving waters, particularly where available dilution volumes are restricted.
A recent EPA investigation showed a thirtyfold increase in FC and enterococci concentrations 28 hours
after disinfection by UV light (Wojtenko 1999). Enterococci regrowth after disinfection by chlorine or
chlorine dioxide was negligible over the period studied.

Municipal Separate Storm Sewer Systems

The design of separate sewer systems isolates the sanitary and storm water flows. The sanitary sewage
flows to the wastewater treatment plant. The second system routes storm water to nearby receiving
waters.  In this system design, the storm water runoff remains untreated carrying all the associated
contaminants, including microorganisms, directly to the receiving water. The  system storm water de-
sign capacity is based on  expected storm runoff volume under the proposed or existing development.
Because the system isolates the two flows, the design requires separate conveyance systems with longer
total pipe length and long-term maintenance costs than combined systems. With added development,
sanitary and storm systems are sometimes connected inappropriately, resulting in sanitary sewage be-
ing carried to the receiving water with no treatment.

The presence of microbial indicators and pathogens in storm water has been confirmed. Olivieri et al.
(1977) reported high densities of indicator organisms in urban streams in Baltimore, plus the presence
ofPseudomonasaemginosa, Staphylococcusaureus, Salmonella, and enteric viruses. Analyses of storm
water in the study area reported the occurrence of pathogenic bacteria and viruses in storm water run-
off. The bacterium P.  aeruginosa was found in all storm water samples taken from six sampling loca-
                                             4-17

-------
Table 4-8. Microbiological Concentrations of Storm Water
Contaminant
Concentrations (per 100 mL) in Storm Water   References for Storm Water
Total coliforms
7-1.8 xlO7
Dutka and Tobin 1978;
Dutka and Rybakowski 1978
Fecal coliform
0.2-1.9 x 106
Dutka and Tobin 1978;
Dutka and Rybakowski 1978
Fecal streptococci
3-1.4 x 106
Dutka and Tobin 1978;
Dutka and Rybakowski 1978
Enterococci
1.2xl02-3.4xl05
Gannon and Busse 1989
HPC (#/mL)
6.94 x 104-4.9 x 105
Dutka and Tobin 1978
Pseudomonas aeruginosa    1-1.1 x 107
                                        Dutka and Tobin 1978;
                                        Olivierietal. 1977
Escherichia coli
Salmonella
Shigella
Klebsiella
Enterobacter
Citrobacter
Yersinia enterocolitica
Staphylococcus aureus
(MPN/lOOmL)
Legionella
Streptococcus
Virases (enteric)
Giardia
Cryptosporidium
Fungi
Parasites-nematodes
Helminth ova
HPC = heterotrophic
1.2 x 10M.7 x 103
5.7-1.5 x 103
(MPN/10 L)
Not detected
4xl03-1.9x 105
Not detected
Not detected
Not detected
1-1.2 xlO2
Not detected
Detected
Detected
-
-
6 x 102-1.2 xlO7
Detected
-
plate count, NA = none available
Gannon and Busse 1989
Geldreich et al. 1968
Olivierietal. 1977
Schillinger and Gannon 1985
Dutka and Tobin 1978
Dutka and Tobin 1978
NA
Olivierietal. 1977
NA
Geldreich et al. 1968
Olivierietal. 1977
NA
NA
Dutka and Rybakowski 1978
Dutka and Rybakowski 1978
NA

tions. Staphylococcus aureus and Salmonella spp. were found at each of the six sampling locations in
a majority of the samples taken. Coxsackievirus B, animal virus, poliovirus, and echovirus were found
in storm water samples collected from all six of the sampling locations. Makepeace et al. (1995) sum-
marized concentrations of microbial indicators and pathogens found in storm water runoff reported by
others, which is presented as Table 4-8.

Sampling was conducted during the summer of 1985 to evaluate the impacts of discharges from storm
drains on bacteriological quality on the Huron River in the Ann Arbor, MI, area during both dry and wet
weather periods (Gannon and Busse 1989). Each river water sample was analyzed for FC, FS, E.  coli,
and enterococci. The investigators reported that wet weather bacterial densities were statistically sig-
nificantly higher than dry weather levels, and downstream densities were statistically significantly higher
than upstream densities. The FC/FS ratios for the storm drains were low, suggesting that sources were
more animal than human.

A 1999 study to determine the  source of unexpectedly high river and stream bacterial contaminations
near Nashville showed that FC densities were directly related to the density of housing, population,
development, percent impervious  area, and apparent domestic animal density. The data also showed
                                             4-18

-------
that FC counts were much higher in summer than winter, suggesting a possible seasonal variation. The
FC/FS ratios were generally low, suggesting primarily animal sources. Surface runoff samples from
more densely populated sewered areas generally showed higher bacterial counts than runoff from less
developed areas that utilized septic tanks. The investigators concluded that surface runoff from high
density urban areas may be a contributor to high fecal bacteria loadings (Young and Thackston 1999).
Consistent with these results are those presented by Mallin (1998), who reported patterns of increasing
coliform bacteria concentrations in stream samples with increased watershed development and imper-
vious surface in New Hanover County, NC.

Storm water discharges are regulated in selected communities through the NPDES program (USEPA
2000b). In response to the 1987 amendments to the CWA, EPA developed Phase I of the NPDES Storm
Water Program in 1990. Phase I requires NPDES permits for storm water discharges from

   •  Medium and large municipal separate storm sewer systems (MS4s), generally serving or
      located in incorporated places or counties with populations of 100,000 or more people.
   •  Eleven categories of industrial activity, one of which is construction activity that disturbs 5
      acres or greater of land.

The Final Rule for Phase II of the NPDES Storm Water Program was signed by the EPA Administrator
on October 29, 1999. The Phase II Rule requires NPDES permit coverage for discharges from

   •  Certain regulated small MS4s (primarily all those located in urbanized areas).
   •  Construction activity disturbing between 1 and 5 acres of land.

Sediment Resuspension

Pettibone and Irvine (1996) reported levels and sources of indicator bacteria in the Buffalo River, NY,
watershed and found that solids present in the water column may  offer a vehicle by  which bacteria are
kept in suspension and transported downstream. Additionally, the sediments provide an environment
that promotes microorganism growth and protects them from predators. Sherer et al. (1992) reported
longer survival of FC and FS in sediment-laden waters than in the  sediment's supernatant and in waters
without  sediment. When incubated with sediment, FC and fecal  Streptococcus half-lives were deter-
mined to be from 11 to 30 days and from 9 to 17 days, respectively. These are longer than when they are
incubated without sediment.

Best Management Practices

In addition  to the installation of sewage treatment and combined overflow systems, there are passive
pollution prevention and mitigation techniques called best management practices (BMPs). The tech-
niques vary dramatically in application, ranging from social practices to engineering applications. Even
the more heavily engineered solutions combine the practitioner's  art with traditional engineering tools
and rely on common sense approaches to what should work in a given  situation.

BMPs are often categorized according to the degree of structural intensity associated with the practice.
Low-structural intensity techniques include public education, emphasizing the consequences of spe-
cific actions. Many communities, for example, paint fish on storm sewer catch basins to emphasize the
link between potential waste disposal and receiving waters. Similarly, community master planning can
incorporate practices intended to prevent contaminant introduction. Mitigation techniques can range
from requirements for storm water controls during the development process, including leaving
designated areas undisturbed, to housing density controls through zoning. The effectiveness of these
pollution prevention techniques is, and is likely to remain, uncertain. Even when installed as part of a
remedial approach, it is unlikely that investigations can separate  the effects of these approaches. The


                                             4-19

-------
temporal scale similarly confounds investigations as source elimination will often require many years
of natural flushing and attenuation to become apparent in the receiving waters.

Well-planned and well-executed studies of more structurally intensive approaches are also limited and
questions of long-term cost and effectiveness remain.  A complete evaluation of a given technique
requires mass balances over several seasons to ensure that BMP effectiveness does not simply change
with timing, i.e., pollutants temporarily accumulate and are discharged in a later storm event. This
phenomenon can be identified in some event-specific and short-term evaluations when effluent con-
centrations, masses, or both exceed the influent (Kurz 1998). These studies are difficult to complete
and depend heavily on  flow measurement  and sample analysis. The inability to automate analytical
processes with data logging sensors makes  these evaluations expensive.

Although little well-documented research is available presenting the capabilities to control microor-
ganisms, watershed managers routinely install BMPs  for storm water treatment. There is strong
suggestive evidence that these installations preserve water quality and can reduce water treatment costs.
Kurz (1998) documented pathogen and indicator reductions in sand filtration, wet detention, and alum
coagulation treatment systems using simulated storm events. Each system produced significant reduc-
tions in TC and FC bacteria, male-specific coliphage 2, and beads (used as a protozoa surrogate) con-
centrations. Often, effluent samples showed greater concentrations of TC, turbidity, and total suspended
solids than influent samples. These increases show the incomplete understanding of the mechanisms,
processes, and temporal scales of BMP operation.

In 1999, EPA released fact sheets on the use of sand filters, wetlands, and vegetative swales for manag-
ing storm water (USEPA 2000a, 2000b, 2000c). Other BMPs include detention ponds, buffer strips,
and infiltration trenches. Sand filters are structurally intensive devices installed primarily to remove
particulate and particulate-associated contaminants. Sand beds block the migration of particulates as
water passes through the media bed. Some biological activity develops as biofilms develop within the
device. Augmenting the media with high organic matter such as peat increases sorption  within the
filter. Sand filters provide very limited flow modification and therefore provide little protection of
streambed or stream banks. Filter sizing is based on predicted runoff volume and is therefore based on
the  size and infiltration properties of the drainage area. These devices have impermeable bottoms to
prevent infiltration to ground water. The filters need to have the filter media replaced periodically
depending on loading. Typical replacement  periods range from three to five years, with expended filter
media suitable for landfill disposal (USEPA 1999a).

Storm water wetlands are incidental, natural, or intentionally constructed areas that are usually flooded.
Within these areas, physical, chemical, and biological processes trap or degrade entering contaminants.
Intentional use of naturally occurring wetlands to treat storm water runoff may be discouraged or pro-
hibited. Storm water wetlands are divided into subsurface and free water surface systems based on the
water flow pattern within the wetland.  The  selected location must have an adequate water supply and
appropriate soil characteristics. Sizing techniques vary and may be state regulated. Common approaches
include a designated design storm, fraction  of watershed area, and sizing to contain the runoff volume
generated by most rain events for the local area. Sources recommend an aspect (length to width) ratio
between 1 and less than 10 to reduce internal short circuiting. Wetlands are commonly augmented with
ponds (USEPA 1999b). Typical reported bacterial removal efficiencies for storm water wetlands are
70% to 80%. The heavy vegetation slows water flow, allowing particulate sedimentation and infiltra-
tion to ground water. The standing water promotes physical, chemical, and biological processes. Well-
designed and constructed wetlands are  long lasting.

Vegetated swales  are broad, shallow, terrestrial channels that often serve as substitutes for curb and
gutter drainage systems. To operate effectively, swales need a shallow slope with thick vegetation
growth. The underlying soil must provide adequate drainage to prevent accumulating standing waters.


                                             4-20

-------
There must be enough slope to promote water transport, but not so steep as to cause erosion and scour-
ing; typical values are 2% to 4%. There are no reported measurements of microbial reductions in swales.

Among the more common BMPs, a wet detention pond is an excavated volume designed to capture and
slowly release storm water runoff. The wet detention pond maintains a standing pool of water to pro-
mote physical, chemical, and biological processes to lower contaminant concentration in runoff. The
local rainfall, ground water, and geology must provide a standing water pool. The standing pool typi-
cally provides sufficient residence time to promote solids settling and removal of particle-associated
contaminants. The edges of the pond commonly have shallow  ledges to promote plant growth for
nutrient uptake, safety, and aesthetics. The pond design typically has an aspect ratio greater than about
two to reduce short-circuiting. The controlled flow discharge reduces the hydrograph peak (Botts et al.
1996; Frederick et al.  1996).  Pond sizing is based on the drained  area and effective runoff coefficient.
The runoff volume is typically modeled using simple hydrology techniques.

Installing buffer strips is a commonly prescribed BMP for protecting receiving waters from storm
water runoff in agricultural areas, with Cryptosporidium often being the primary concern. Buffer strips,
also called filter strips, are vegetated areas using single species or mixtures of grasses, legumes, or
other forbs with stem  spacing up to one inch installed parallel to the receiving water shore. Although
experts debate the minimum required width, a commonly recommended minimum is about ten meters.
The strip follows the contour, with variations less than 0.5%. The land slope immediately above the
filter is typically 1% to 10% to ensure flow through and control maximum velocities. The adequacy of
a buffer strip for protecting receiving waters is based on Natural Resources Conservation Service (NRCS)
criteria. The NRCS National Handbook of Conservation Practices (NRCS 1997; NRCS 1998) contains
the traditional filter strip design standards. Using these standards for pathogen control, NRCS expects
a slight decrease in surface water pathogen contamination.

Moore et al. (1988) cite several studies that show the effectiveness of buffer strips in reducing nutrients
and sediments in runoff. The mechanisms contributing to the effectiveness are reduction in volume
from increased infiltration, decrease in velocity resulting in increased sedimentation of particulates
with adsorbed pollutants, and increased pollutant adsorption to soil particles due to lower ionic
concentrations. For a vegetated filter strip to remove sediment-bound organisms, it must  provide an
appropriate mechanism for removing sediment. Design procedures (Dillaha and Hayes 1991) identify
several key considerations when selecting buffer strips. Filter strips are only effective under shallow
sheet  flow conditions. Sheet flow will occur if the filter strip can be installed approximately on the
contour. Fields with extensive internal drainage concentrate surface runoff. Excessive sediment inflow
to  an effective filter strip will clog and shorten the useful life. Routine maintenance, e.g.,  mowing to
encourage dense vegetation and weed control, inspection and repairs to fill  gullies, removing flow-
blocking sediment,  reseeding, and other measures, prevents concentrated flow. Excluding livestock
and vehicles reduces soil compaction and promotes infiltration. Walker et al. (1990) modeled the con-
centration of indicator bacteria in runoff resulting from a single storm event immediately after land
application of waste. The model predicted that a 30-meter filter  strip on a 3% slope could remove a
maximum of 75% bacteria. The model did not show if increased length would result in further reductions.

Infiltration trenches capture and hold the runoff volume for infiltration. These devices are typically one
to  four-meter-deep excavations filled with aggregate and gravel installed in well-drained, low-sloped
soils.  Sufficient underdrainage is critical for proper operation. Sand filters can capture up to 90% of
influent particulate matter (Botts et al. 1996). Functionally, infiltration trenches work as coarse-media
sand filters discharging to ground water. While the ground water discharge replenishes ground water,
there are often concerns  about the remaining contaminants and areas with deep water tables. Mainte-
nance is essential to prevent clogging as particulates accumulate in the filter media. More than half the
installed infiltration trenches fail after five years from inadequate maintenance (Botts et al.  1996).
                                              4-21

-------
Source Water Protection and Watershed Management

EPA's Office of Water has defined SWP as a common-sense approach to guarding public health by
protecting drinking water supplies.  SWP measures prevent contamination and reduce the need for
treatment of drinking water supplies. SWP includes managing potential contamination sources and
developing contingency plans that identify alternate drinking water sources. A community may decide
to develop  an SWP program based on the results of a source water assessment, which includes the
delineation of the area to be protected and an inventory of the potential contaminants within that area
(USEPA 2000a). SWP from quality degradation by microbial contaminants (i.e., bacteria, protozoa,
viruses, helminths, fungi) is any activity undertaken to minimize the frequency, magnitude, and dura-
tion of occurrence  of pathogens or indicators (e.g., indicator microorganisms or turbidity) in source
waters. SWP may also, by reducing the concentration of NOM, a DBF precursor, reduce the formation
of DBF.

SWP strategies comprise the first stage in the multiple-barrier approach to protecting the quality of
drinking water. Other major drinking water quality protection barriers include water quality monitoring
and selective source withdrawal, water treatment processes for removal or inactivation of pathogens
and control of DBF formation, water  distribution practices for preventing intrusion or regrowth of
pathogens,  and point-of-use treatment where required.

SWP strategies are a specific subset of a larger watershed protection strategy applied when the pro-
tected receiving water is used as a water supply. Conceptually, watershed protection is heavily linked to
pollution prevention, contaminant source identification, and  risk management. Although watershed
management does not have a universally accepted definition and connotes alternate approaches, each
interpretation has an underpinning of holistic approaches to prevent or mitigate threats to the receiving
water over a geographic region defined by a common hydrology.

Managing microbial contaminant risks in watersheds requires identification and quantification of or-
ganisms. Because of difficulties associated with assaying for specific pathogens, monitoring programs
have tested for indicator organisms, including PCs and TCs, to identify possible fecal contamination in
water. Monitoring regulations often specify indicators for determining water quality because the ana-
lytical methods are easier to complete, faster, and lower-cost than methods  for specific organisms.
Limitations of relying on indicators for determining the presence of pathogens include the occurrence
of false positives. The indicators measure bacteria that live not only in human enteric tracts, but also in
the enteric tracts of other animals (Toranzos and McFeters 1997).

Epidemiological studies in recreational waters (Dufour 1984) showed  no correlation between  mea-
sured FC densities and the occurrence of gastrointestinal illness in swimmers in fresh water, but a high
correlation between gastrointestinal illness and E. coli and Enterococcus concentrations. Based on
these results, EPA recommended that states adopt E. coli and Enterococcus as recreational water crite-
ria in 1986, but some feel that these new indicators are inadequate (Calderon et al. 1991).

Methods to identify and quantify pathogens in watersheds require filtering large volumes of water and
eluting the  organisms from the filter. Detection and quantification are accomplished by culturing or
molecular biology  methods.  Some organisms cannot be  identified through culturing techniques, so
molecular biology methods, based on nucleotides within nucleic acid sequences, are used. Low recov-
ery efficiencies commonly encountered with filtration recovery make it difficult to estimate original
concentrations with confidence. Methods for protozoa are cumbersome and do not indicate viability.
Infectivity studies can be done to determine viability, but are expensive and slow. When an outbreak of
a waterborne pathogen is  suspected and the water is tested, the pathogen may not be detected because
the contamination may have been temporary and been flushed out or died off (Moe 1997).
                                             4-22

-------
Modeling and Source Water Protection

Modeling can assist in identifying the vulnerability of a drinking water utility to threats from source
water contamination. These models can be used in assessing the impact of upstream point-source dis-
charges on downstream users as well as the potential for contamination from nonpoint sources (Clark
et al. 1998). For example, the Water Supply and Water Resources Division (WSWRD) has developed
two user-friendly modeling systems which include (!) a simplified model of the entire Ohio River, and
(2) a detailed model of the Ohio River mainstream that may be used under emergency spill situations.
Both models are built to interact with a Geographic Information System (GIS) for display and/or input
generation, and it is anticipated that this approach will be extended to other  source waters. The wide-
scale model uses representative steady state flow regimes and represents movement by simple travel
time relationship and transformations by dilution and decay mechanisms. Pollutants are routed through
the RF1 reach file representation of the basin (Clark et al. 1998).  The detailed mainstream model uses
actual dynamic flow patterns as  input to EPA's WASP4 water quality model (Ambrose et al. 1990).
WASP4 is a dynamic compartment model that can be used to analyze a number of water quality prob-
lems. The Ohio River mainstream is represented in the model from a series of segments ranging in size
from two to ten miles in length. The basic equation used in WASP4 governing decay of contaminants is
as follows:

                                 C, = (M/e,)exp(-frCr)                        (4-1)

where Qs (LIs) is the flow in the segment, M is the mass of the pollutant (mg/s) that enters the segment,
k is the decay coefficient with a typical value of 0.5/day, CTs is the cumulative time of travel (days),
"exp" denotes  the exponential function, and Cs is the concentration in mg/L at the end of the reach.
When the pollutant is stable and not reactive,  the value for k = 0.

The detailed model includes a hydraulic model (the Corps of Engineers FLOWSED model), which has
been combined with WASP4 to  make spill modeling predictions.  FLOWSED, which predicts daily
flow quantities along the mainstream and portions  of major tributaries near  their confluence with the
Ohio River, is applied daily by the Ohio  River Division of the Corps of Engineers. Five-day forecast of
stage and flow are generated for 400 mainstream  and tributary segments, and the  results were made
accessible to the Ohio River Valley Water Sanitation Commission (ORSANCO) via telephone lines.

A relational database management system was used to organize the various sources of data used in the
study. Individual data files included information on facilities, outfall, permit limits, monitoring data,
and codes used in the other files. The NPDES permit number was used as the primary key in each of the
files.

GIS modeling and Data Base Management System  (DBMS) techniques were integrated into  two tools
for use by ORSANCO for analyzing spills in the Ohio River. The NETWORK component of ARC/
INFO was used to provide a steady state contaminant routing capability. In addition, a C-based spatial
decision support system was developed  as a spill management system to serve as a quick response tool
for analyzing and displaying the results  of a pollutant spill into the Ohio River.

Research is underway to  extend this modeling approach to microbiological discharges from CSOs.
Research is also being conducted which  is intended to extend this modeling concept to nonpoint source
contamination.

Modeling Overland Migration of Pathogens

Another aspect of contamination modeling is the overland transport of pathogens. Although  efforts to
model overland transport  of Cryptosporidium oocysts have been limited, such models are needed to
                                             4-23

-------
predict oocysts loads and estimate the effectiveness of management practices. This information may
subsequently be used in reservoir models if accuracy requirements are met. Auer et al. (1998) identified
the need for developing pathogen loading data in order to support pathogen fate and transport modeling
within reservoirs. Several models exist that are capable of predicting soil loss, runoff, transport of
contaminants from animal waste, and bacterial die-off Whether these models, individually or combined,
are capable of accurately predicting Cryptosporidium loads or reductions achieved during transport
through buffer strips remains to be seen. Considerations in model selection include assumptions made
by the model, the size of the watershed, availability of data, and the desired level of accuracy. The
ability of a model to simulate oocyst transport is dependent on how well the model assumptions reflect
the actual characteristics of oocysts and the landscape over which they travel. EPA has sponsored
ongoing research to evaluate factors affecting overland migration of oocysts. A major goal of the re-
search is to determine the degree to which oocysts tend to stick to different materials and then to
evaluate their potential for runoff, either in attached or free-floating form. Key components of the
project include jar tests to determine partitioning of oocysts among water, clay, or other soil, fecal
matter, plant matter, etc.; flume tests to directly evaluate oocyst overland migration; evaluation and de-
velopment of a modeling framework; and evaluation of protocols for measuring oocysts in high turbid-
ity samples encountered in runoff samples.

Models used in predicting the transport of animal waste over land have typically utilized indicator
bacteria. This approach is useful in assessing risk due to fecal contamination since indicator organisms
are easily identified, while low levels of pathogens may not be discernable. Although FC is a common
indicator organism for fecal wastes, its physical characteristics differ significantly from those of oo-
cysts. This results in differences in die-off rates and soil retention. These differences result in inability
of existing models to predict oocyst transport. Identifying and quantifying the mechanisms which af-
fect die-off and retention of oocysts may facilitate the use of existing models for estimating oocyst
concentrations.

Crane and Moore (1986) found that, of the several patterns followed during enteric bacteria die-off, the
model for first-order die-off kinetics accurately described bacterial die-off under several conditions.
However,  the rate coefficient was highly variable due to differences in the effect of environmental
factors on the assorted types of bacteria. The authors identified pH, temperature,  solar radiation, mois-
ture, application method, and application medium as critical factors in determining microbial survival.
Information on  the effects of these factors on oocyst survival is necessary in order to develop a die-off
rate coefficient(s) for Cryptosporidium.

Reddy et al. (1981) combined an animal waste model with the Agricultural Runoff Management (ARM-
II) Model to simulate the effects on the quality of runoff from land receiving animal waste. The micro-
biological submodel was developed by simulating FC die-off and retention in the soil. Moore et al.
(1988) also developed a model, MWASTE, which follows indicator organisms from the animal waste
through leaving the land as surface runoff utilizing bacterial indicator organisms.  MWASTE is capable
of including data on the slope and width of buffer strips. The model  COLI (Walker et al. 1990) also
examines  the movement of indicator bacteria in runoff. Although these models contain a biological
component, they cannot be used to predict oocyst transport. It is possible that newer models,  with
improved capability to predict hydrology and sediment transport, may be adaptable to predicting oocyst
transport if mechanisms controlling overland flow were better understood. The New York City Depart-
ment of Environmental Protection has conducted an evaluation including pathogen loading in its ter-
restrial models and determined that improvements in identification and quantification of oocyst sources
was required (USEPA 1997).
                                             4-24

-------
Summary and Conclusions

Passage of the 1996 amendments to the SDWA has focused the attention of water utility managers and
public health and regulatory officials on SWP and its role in protecting public water supplies. There is
growing awareness that water treatment and/or disinfection may not always be enough to ensure the
provision of potable and safe water to the consumer.  The 1993 cryptosporidiosis outbreak in Milwau-
kee, WI, has raised the possibility that even water suppliers which meet all of the SWTR requirements
of the SDWA are vulnerable (Okun et al. 1997; Fox and Lytle 1996).

Most utilities in the U.S. invest a great deal of time, energy, and capital in developing mechanisms for
protecting against the impact of sudden changes in influent water quality.  Some of these mechanisms
include investment in excess capacity and development of emergency procedures (Miller 1989).

Concern over SWP is not limited to surface water supplies. Many ground water supplies have proven to
be vulnerable as well, resulting in the various states implementing wellhead protection programs. Based
on the 1996 amendments, the states will have to implement programs to decide if a system's source of
supply is threatened as well as determine the means to prevent pollution.  Communities will be allowed
to ask for  state assistance, and a certain percentage of the State Revolving Loan Fund has been ear-
marked to assist with SWP (Howell 1987).

The SDWA was passed in 1974 and amended in 1986  and 1996, but SWP under the SDWA actually
began with the SDWA Amendments of 1986. The 1986 amendments included provisions for "Protec-
tion of Ground Water Sources of Water." Two programs were set up under this requirement: the "Sole
Source Aquifer Demonstration Program," to establish demonstration programs to protect critical aqui-
fer areas from degradation; and the "Wellhead Protection Program," which required states to develop
programs for protecting areas around public water supply wells to prevent contamination from residen-
tial, industrial, and farming-use activities.

In the 1996  amendments to the  SDWA, protection of source waters was given greater emphasis to
strengthen protection against microbial contaminants,  particularly Cryptosporidium, while reducing
potential health risks due to disinfection by-products.  This  increased protection  is embodied in the
IESWTR (USEPA 1998). This rule applies to public water systems that  use surface water or GWUDI
and serve at least 10,000 people.

Two major threats to source water quality with respect to DBF control and microbial protection are
natural organic matter and pathogens. As reflected in the previous discussion, the two pathogens which
are currently of most concern are Giardia and Cryptosporidium.

Managing microbial risk requires identification and quantification of organisms. Because of difficul-
ties associated with assaying for specific pathogens, monitoring programs have tested  for indicator
organisms, including FC and TCs, to identify possible fecal contamination in water. The potential
sources  of pathogens in source water are many and varied, including nonpoint runoff and discharges
from treated and untreated sewage and combined sewer overflows. From a waterborne outbreak and
public health viewpoint, both Giardia and Cryptosporidium are of primary concern. Monitoring regu-
lations often specify indicators for determining water quality because the analytical methods are easier
to complete, faster, and lower-cost than methods for  specific organisms. Limitations of relying  on
indicators  for determining the presence of pathogens include the occurrence of false positives. The
indicators  measure bacteria that live not only in human enteric tracts, but also in the enteric tracts of
other animals (Toranzos and McFeters 1997).
                                             4-25

-------
Microbial pathogens are found in treated sanitary sewage and wet weather flows, i.e., SSOs, CSOs, and
storm water runoff. Many factors effect the types of organisms found and the concentrations at which
they are detected. These include watershed contributions, treatment plant efficiency, and length of
antecedent dry weather period. These treatment technologies can be both sources of contamination as
well as protective of source water quality. In addition to the  installation of sewage treatment and
combined overflow systems, there are  passive pollution prevention and mitigation techniques called
BMPs. The techniques vary dramatically in application, ranging from social practices to engineering
applications.

SWP strategies are a specific subset of a larger watershed protection strategy applied when the pro-
tected receiving water is used as a water supply. Conceptually, watershed protection is heavily linked to
pollution prevention, contaminant source identification, and risk management.

Modeling can assist in identifying the vulnerability of a drinking water  utility to threats from source
water contamination. These models can be used in assessing the impact of upstream point-source dis-
charges on downstream users as well as the potential for contamination from nonpoint sources (Clark
et al. 1998). Another aspect of contamination modeling is the overland transport of pathogens. Al-
though efforts to model  overland transport of Cryptosporidium oocysts have been limited, such models
are needed to predict oocyst loads and estimate the effectiveness of management practices.

Although SWP is currently more of a  collection of practices than a well-defined art  or science, it is
anticipated that it will become an integral part of water treatment practice in the future. As interest
grows in the concept of watershed management, it is likely that interest will grow in understanding the
factors that effect the quality of source  water for drinking water utilities as well.

References

Ambrose, R. B., Wool, T. A., Connoly,  J. P., and Schanz, R. W.  (1990). WASP4, A hydrodynamic and
   water quality model—Model theory, user's manual and programmers' guide. EPA Environmental
   Research Laboratory, Athens, Georgia.

American Public Health Association. (1998).  Standard methods for the examination of water and
   wastewater, 20th Ed., American Water Works Association, Water Environment Federation,
   Washington, D.C.

Atherholt, T. B., LeChevallier, M. W., Norton, W. D., and Rosen, J. S. (1998). "Effect of rainfall on
   Giardia and crypto." Journal of the American Water Works Association, 90(9), 66-80.

Atwill, E. R., et al. (1998). "Age, geographic, and temporal distribution of fecal shedding of
   Cryptosporidium parvum oocysts in cow-calf herds." American Journal of Veterinary Research,
   60(4), 420-425.

Atwill, E. R., Sweitzer,  R. A., Pereira, M. G. C., Gardner, I. A., Van Vuren, D., and Boyce, W. M.
   (1997).  "Prevalence of and associated risk factors for shedding Cryptosporidium parvum oocysts
   and Giardia cysts within feral pig populations in California." Applied and Environmental Micro-
   biology, 63(10), 3946-3949.

Auer, M. T, Bagley, S.  T., Stern, D. A., and Babiera, M. J. (1998). "A framework for modeling the
   fate and transport of Giardia and Cryptosporidium in surface waters." Journal of Lake and
   Reservoir Management, 14(2-3), 393-400.

Badenoch, Sir J. (Chairman). (1990). "Cryptosporidium in water supplies—Report of the group of
   experts." Department of the Environment, Department of Health, London, U.K., HMSO.
                                             4-26

-------
Bajer, A., Bednarska, M., and Sinski, E. (1997). "Wildlife rodents from different habitats as a reser-
   voir for Cryptosporidium parvum." Acta Parasitologica, 42(4), 192-194.

Barry, J. S., Atwill, E. R., Tate, K. W., Koopman, T. S., Cullor, J., and Huff, T. (1998). "Developing
   and implementing a HAACCP-based program to control Cryptosporidium and other waterborne
   pathogens in the Alameda Creek watershed: Case study." American Water Works Association
   Annual Conference, Dallas, TX, June.

Barthe, C. and Brassard, N. (1996). "Giardia and Cryptosporidium in river water, ground water and
   tap water in Quebec." Planning for Tomorrow: Proceedings, 6th National Conference on Drink-
   ing Water, Victoria, B.C., Oct. 16-18, 1994, 207-215.

Benton, C., Ives, K. J., Miller, D. G.,  and West, P. A. ( 1991).  "Cryptosporidium in water—progress
   in the United States of America." Report of an OSTEMS mission, 6-20. Water Research Centre,
   Medmenham, Marlow, Buckinghamshire, U.K., 19.

Berg, G., Safferman, R. S., Dahling, D. R, Berman, D., and Hurst, C. J. (1983). USEPA Manual of
   methods for virology, EPA-600/4-84-013. U.S. Environmental Protection Agency, Cincinnati, OH.

Botts, J., Allard, L., and Wheeler, J.  (1996). "Structural best management practices for storm water
   pollution control at industrial facilities." Proceedings, Watershed '9 6 Moving Ahead Together
   Technical Conference and Exposition, Water Environment Federation, Baltimore, MD, June 8-12.

Breach,  R. A., Howells, V. W., Porter, M.  J., Furness, M., Dawson, D., and Black, D. W. (1994). "An
   integrated water utility approach to minimize the risks associated with waterborne
   Cryptosporidiosis." Proceedings,  American Water  Works Association Annual Conference, New
   York, June 19-23. American Water Works Association, Denver, CO.

Calderon, R. L., Mood, E. W., and Dufour, A. P.  (1991). "Health effects of swimmers and nonpoint
   sources of contaminated water." InternationalJournal of Environmental Health Research, 1,
   21-31.

Camp, Dresser & McKee and Moffa & Associates. (1997). Draft CSO disinfection pilot study report.
   Submitted to New York City Department of Environmental Protection.

Casson, L. W., Sorber, C. A., Sykora, J. L., Gavaghan, P. D., Shapiro, M. A., and Jakubowski, W.
   (1990). "Giardia in wastewater—effect of treatment." Research Journal of the Water Pollution
   Control Federation, 62(5), 670-675.

Chauret, C.,  et al. (1998). "Effect of microbial antagonism on the survival of Cryptosporidium
   parvum oocysts in fresh natural waters." Proceedings, IAWQ 19th Biennial International Water
   Quality Conference, Vancouver, Canada, June 21-26.

Clark, R. M., Goodrich, J. A., Lykins, Jr.,  B. W., and Neal, J. R.  (1998). "Evaluating the effects of
   upstream dischargers on downstream water suppliers: A source water protection model." Journal
   of Water Supply Research and Technology-Aqua, 47(5), 1-8.

Coffman, L. S., Smith, J., and Lahlou M. (1996). "Environmentally sensitive low-impact develop-
   ment." Proceedings, Watershed '96 Moving Ahead Together Technical Conference and Exposi-
   tion, Water Environment Federation, Baltimore, MD, June 8-12.

Cooke, G. D. and Carlson, R. E. (1989). "Chaper 3: THM production and eutrophication." Reservoir
   management for water quality and THM precursor control. American Water Works Association
   Research Foundation, Denver, CO, 5-20.
                                             4-27

-------
Cooke, G. D., Palmstrom, N., Martin, A., Watkins, M., Beals, A., and Carlson, R. E. (1988). "Reser-
   voir management for THM precursor control." American Water Works Association seminar on
   water quality management in reservoirs., Orlando, FL. American Water Works Association,
   Denver, CO, 21-51.
Corbett, C. W., Wahl, M., Porter, D. E, Edwards, D., and Moise, C. (1997). "Nonpoint source runoff
   modeling: A comparison of forested watershed and an urban watershed on the South Carolina
   coast." Journal of Experimental Marine Biology and Ecology, 213, 133-149.
Correll, D. L., Jordan, T. E., and Weller, D. E. (1992). "Nutrient flux in a landscape: Effects of
   coastal land use and terrestrial community mosaic on nutrient transport to coastal waters." Estu-
   aries, 15(4), 431-442.
Crane, S. R. and Moore, J. A. ( 1986). "Modeling enteric bacterial die-off: A review."  Water, Air, and
   Soil Pollution, 27, 411-439.
Craun, G. R (1986). Waterborne diseases in the United States. CRC Press LLC, Boca Raton, FL.
Craun, G. F, Hubbs, S. A., Frost, F, Calderon, R. L., and Anad Via, S. H. (1998). "Waterborne out-
   breaks of Cryptosporidiosis." Journal of the American Water Works Association, 90(9), 81-91.
Current, W. L. and Garcia, L. S. (1991). "Cryptosporidiosis." ClinicalMicrobiological Review, 4(3),
   325-358.
Dillaha, T. A. and Hayes, J. C. (1991). "A procedure for the design of vegetated filter strips—Final
   report." United States Soil Conservation Service.
Dufour, A. P. (1984). "Health effects criteria for fresh recreational waters." EPA-600/1-84-004, U.S.
   Environmental Protection Agency, Research Triangle Park, NC, August.
Dutka, B. J. and Tobin,  S. (1978). "Monitoring of storm water runoffs for bacterial and viral patho-
   gens of man" Research Report No.  87, Project No. 76-8-4, Ministry of the Environment.
Dutka, B. J. and Rybakowski, I. (1978). "Microbiological study of some Canadian storm water
   runoffs at Burlington and Brucewood." Research Report No. 87, Project No.  75-3-24, Ministry of
   the Environment.
Edwards, D. R., Daniel, T. C., Scott, H. D., Moore, Jr., P. A., Murdoch, J. F., and Vendrell,  P. F.
   (1997). "Effect of BMP implementation on storm flow quality of two northwestern Arkansas
   streams." American Society of Agricultural Engineers, 40(5), 1311-1319.
Ellis, J. B. and Yu, W. (1995). "Bacteriology of urban runoff: The combined sewer as a bacterial
   reactor and generator." Water Science Technology, (31)7, 303-310.
Erlandsen, S. L. and Meyer, A. (\9%4)"Giardia and Giardiasis: Biology, pathogenesis, and epidemi-
   ology." Plenum Press, New York.
Payer, R. (1994). "Effect of high temperature on infectivity of Cryptosporidiumparvum oocysts in
   water." Applied and Environmental Microbiology, 60(8), 2732-3735.
Payer, R. and Nerad, T.  (1996). "Effects of low temperatures on viability of Cryptosporidium parvum
   oocysts." Applied and Environmental Microbiology, 62(4), 1431-1433.
Payer, R., Speer, C. A., and Dubey, J. P. (1997). "The general biology of Cryptosporidium."
   Cryptosporidium and Cryptosporidiosis, R. Payer, ed., CRC Press LLC,  Boca Raton, FL.
Payer, R. and Ungar, B. L. P. (1986).  "Cryptosporidium spp. and Cryptosporidiosis." Microbiologi-
   cal Review, 50(4), 458-483.
                                             4-28

-------
Payer, R., ed. (1997). "Cryptosporidium and Cryptosporidiosis." CRC Press, LLC, Boca Raton, FL.

Fisher, T. R., Peele, E. R., Ammerman, J. W., and Harding, Jr., L. W. (1992). "Nutrient limitation of
   phytoplankton in Chesapeake Bay." Marine Ecology Progress Series, 82, 51-63.

Forde, K. N., Swinker, A. M., Traub-Dargatz, J. L.,  and Cheney, J. M. (1997). "The prevalence of
   Cryptosporidium/Giardia in the trail horse population utilizing public lands." Proceedings, 15th
   Equine Nutrition and Physiology Symposium, Colorado State University, Fort Collins, CO, 223-237.

Fox, K. R. and Lytle, D. A. (1996). "Milwaukee's crypto outbreak investigation and recommenda-
   tions." Journal of the American Water Works  Association, 88(9), 87-94.

Frederick, R., Goo, R., Corrigan, M., Bartow, S.,  and Billingsley, M. (1996). "Economic benefits of
   urban runoff controls." Proceedings, Watershed '96 Moving Ahead Together Technical Confer-
   ence and Exposition, Water Environment Federation, Baltimore, MD, June 8-12.

Gannon, J. J. and Busse, M. K.  (1989). "E. Coli and Enterococci levels in urban stormwater, river
   water and chlorinated treatment plant effluent." Water Research, 23(9), 1167-1176.

Garber, L. P., Salman, M. D., Kurd, H.  S., Keefe, T. J., and Schlater, L. (1994). "Potential risk factors
   for Cryptosporidium infection in dairy calves." Journal of the American Veterinary Medical
   Association, 205, 86-91.

Garber, J. (1999). "Anthropogenic nitrogen: Too  much of a good thing?" Plan for USEPA research
   on ecological risks posed by increasing loads of anthropogenic nitrogen to the nation's air, land,
   and water, Draft version 5.1, August 28, 1999.

Geldreich, E. E., Best, L. C., Kenner, B. A., and Van Donsel, D. J. (1968). "The bacteriological
   aspects of stormwater pollution." Journal of the Water Pollution Control Federation, 40(11),
   1861-1872.

Gibson III, C. J., Staderman, K. L.,  States, S., and Sykora, J. (1998). "Combined sewer overflows: A
   source of Cryptosporidium and GiardiaT' Water Science Technology, 38(12), 67-72.

Goel, S.,  Hozalski, R. M., and Bouwer, E. J. (1995). "Biodegradation of NOM: Effect of NOM
   source and ozone dose." Journal of the American Water Works Association, 87(1), 90-105.

Goodgame, R. W., Genta, R. M., White, A. C.,  and Chappell, C. L. (1993). "Intensity of infection in
   AIDS-associated Cryptosporidiosis." Journal of 'Infectious Diseases,  167, 704-709.

Graczyk, T. K., Payer, R., Cranfield, M. R., and Conn., D.  B. (1998a). "Recovery of waterborne
   Cryptosporidiumparvum oocysts by freshwater benthic clams (Corbicula fluminea)." Applied
   and Environmental Microbiology, 64(2), 427-430.

Graczyk, T. K., et al. (1998b). "Giardia sp. cysts and infectious Cryptosporidium parvum oocysts in
   the feces of migratory Canada Geese (Branta  canadensis)." Applied and Environmental Microbi-
   ology, 64(7), 2736-2738.

Griffin, C. B. (1993). Effectiveness and feasibility of best management practices in reducing urban
   nonpoint sources of nitrogen to Long Island.  State University of New York—College of Environ-
   mental Science and Forestry, Syracuse, NY.

Graham, N. J. D., Wardlaw, V. E., Perry, R., and Jiang, J. (1998). "The significance of algae as
   trihalomethane precursors." Water Science Technology, 37(2),  83-89.

Griffin, C. B. (1995). "Uncertainty analysis of BMP effectiveness for controlling nitrogen from
   urban nonpoint sources." Water Resources Bulletin, 31(6),  1041-1050.
                                             4-29

-------
Haas, C. N. and Rose, J. B. (1994). "Reconciliation of microbial risk models and outbreak epidemi-
   ology: The case of the Milwaukee outbreak." Proceedings, American Water Works Association
   1994, 517-23.
Hancock, C. M., Rose, J. B., and Callahan, M. (1998). "Crypto and Giardia in U.S. groundwater."
   Journal of the American Water Works Association., 90(3), 58-61.
Hansen, J. S. and Ongerth, J. E. (1991). "Effects of time and watershed characteristics on the concen-
   tration of Cryptosporidium oocysts in river water." Applied and Environmental Microbiology, 57,
   2790-2795.
Herwaldt, B. L., Craun, G. E, Stokes, S. L., and Juranek, D. D. (1991). "Waterborne-disease out-
   breaks, 1989-1990." CDC Surveillance Summaries, December, 1991, Morbidity and Mortality
   Weekly Report, 40 (No.SS-3), 1-22.
Hirata, T. and Hashimoto, A. A. (1997). "Field survey on occurrence of Giardia cysts and
   Cryptosporidium oocysts in sewage treatment plants." Proceedings, 1997 International Sympo-
   sium on Waterborne Cryptosporidium, Newport Beach, CA, March 2-5. C. Flicker, J. L. Clancy,
   and P. A. Rochelle, eds., American Water Works Association, Denver, CO, 183-193.
Hoehn, R. C., Hargette, P. H., Knocke, W. R., Dietrich, A. M., Randtke, S. J., and Long, B. W.
   (1994). "The influence of raw water quality on the formation of disinfection by-products during
   drinking water treatment." Proceedings, 1994 Water Quality Technology  Conference; Part II, San
   Francisco, C A, November 6-10, 1947-1975.
Hopkinson, Jr., C. S. (1992). "A comparison of ecosystem dynamics in freshwater wetlands." Estu-
   aries, 15(4), 549-562.
Houlahan, J., Marcus, W. A., and Shirmohammadi, A. (1992). "Estimating Maryland critical area
   acts impact on future nonpoint pollution along the Rhode River Estuary." Water Resources
   Bulletin,  28(3), 553-567.
Howell, M. L. (1997). "SDWA then and now." Journal of the American Water Works Association,
   89(3),  11.
Hurst, C. J., Knudsen, G. R, Mclnerney, M. J., Stetzenbach, L. D., and Walter, M.  V. (1997). Manual of
   environmental microbiology. ASM Press, American Society for Microbiology, Washington, D.C.
Jenkins, M. B., Bowman, D. D., and Ghiorse, W. C. (1998). "Inactivation of Cryptosporidium
   parvum oocysts by ammonia" Applied and Environmental Microbiology, 64(2), 784-788.
Jenkins, M. B., Walker, M. J., Bowman, D. D., Anthony, L.  C., and Ghiorse,  W.  C. (1999). "Use of a
   sentinel system for field measurements of Cryptosporidium parvum oocysts in soil and animal
   waste." Applied and Environmental Microbiology, 65(5), 1998-2005.
Johnson, E., Atwill, E. R, Filkins, M. E., and Kalush, J. (1997). "The prevalence of shedding of
   Cryptosporidium and Giardia spp. based on a single fecal sample collection from each of 91 horses
   used for backcounrty recreation." Journal of Veterinary Diagnostics Investigation, 8, 56-60.
King, W. C. (1996). Environmental engineering: P.E. examination guide & handbook. American
   Academy of Environmental Engineers, Annapolis, MD.
Klesius, P. H., Haynes, T. B., and Malo, L.  K. (1986). "Infectivity of Cryptosporidium sp. isolated
   from wild mice for calves and mice." Journal of American Veterinarian Medical Association,
   189(2) 192-193.
Kramer, M. H., Herwaldt, B. L., Craun, G.  F., Calderon, R. L., and Juranek, D. D. (1996). "Water-
   borne disease: 1993 and 1994." Journal of the American Water Works Association, 88(3), 66-80.
                                            4-30

-------
Krasner, S. W., Croue, J-R, Buffle, J., and Perdue, E. M. (1996). "Three approaches for characteriz-
   ing NOM." Journal of the American Water Works Association., 88(6), 66-79.

Kuczynska, E. and Shelton, D. R. (1999).  "Method for detection and enumeration of
   Cryptosporidium parvum oocysts from feces, manure, and soils." Applied and Environmental
   Microbiology, 65, 2820-2826.

Kurz, R. C. (1998). "Removal  of microbial indicators from stormwater using sand filtration, wet
   detention, and alum treatment best management practices." Southwest Florida Water Manage-
   ment District.

LeChevallier, M. W., Norton, W. D., and Lee, R. G. (1991a). "Occurrence of Giardia and
   Cryptosporidium spp. in surface water supplies." Applied and Environmental Microbiology, 57,
   2610-2616.

LeChevallier, M. W., Norton, W. D., and Lee, R. G. (1991b). "Giardia and Cryptosporidium spp. in
   filtered drinking water supplies" Applied and Environmental Microbiology, 57, 2617-2621.

LeChevallier, M. W. and Norton, W. D. (1995). "Giardia and Cryptosporidium in raw and finished
   water" Journal of the American Water Works Association, 87(9), 45-68.

LeChevallier, M. W., Norton, W. D., and Atherholt, T. B. (1997). "Protozoa in open reservoirs."
   Journal of the American Water Works Association, 89(9), 84-96.

Levy, D. A., Bens, M. S., Craun, G. R, Calderon, R. L., and Herwaldt, B. L. (1998). "Surveillance
   for waterborne-disease outbreaks—United States, 1995-1996." CDC Surveillance Summaries,
   December, J 998, Morbidity and Mortality Weekly Report, 47 (No.  SS-5):l-33.

Lijklema, L., Habekotte, B., Hooijmans, C., Aalderind, R. H., and Havellar, A. H. (1986). "Survival
   of indicator organisms in a detention pond receiving combined sewer overflow." Water Science
   Technology, 18(10), 9-17.

Lin, S. D. (1985).  "Giardia lamblia and water." Journal of the American Water Works Association,
   77(1), 40-47.

Lisle, J. T. and Rose, J. B. (1995). "Cryptosporidium contamination of water in the USA and UK:  A
   mini-review." Journal of Water Supply Research and Technology-Aqua, 44(3), 103-117.

Los Angeles Department of Water and Power (LADWP). (1996). .

Madore, M. S., Rose, J. B., Gerba, C. P., Arrowood, M. J., and Sterling, C. R. (1987). "Occurrence
   of Cryptosporidium oocysts in sewage effluents and select surface waters." Journal of Parasitol-
   ogy, 73, 702-706.

Makepeace, D. K., Smith, D. W., and Stanley, S. J. (1995). "Urban stormwater quality: Summary of
   contaminant data," Critical reviews. Environmental Science and Technology, 25(2), 93-139.

Mallin, M. A. (1998). "Land-use practices and fecal coliform pollution of coastal water." Proceed-
   ings, 14th Annual North Carolina On-Site Wastewater Treatment Conference, North Carolina
   Department of Environment, Health, and Natural Resources. North Carolina State University,
   Raleigh, NC, October 27-29.

Marshall, M. M., Naumovitz, D., Ortega, I, and Sterling, C. R. (1997). "Waterborne protozoan
   pathogens." Clinical Microbiology, 10(1), 67-85.
                                             4-31

-------
Metcalf & Eddy, Inc. (1991). Wastewater engineering: Treatment, disposal, and reuse. McGraw-Hill,
   Inc., New York.

Metcalf, T. G., Melnick, J. L., and Estes, M. K. (1995). "Environmental virology: From detection of
   virus in sewage and water by isolation to identification by molecular biology—A trip of over 50
   years." Annual Review of Microbiology, 49, 461-487.

Miller, R. (1989). "Why use GAC—A utility viewpoint." Design and Use of Granular Activated
   Carbon—Predicted Aspects. American Water Works Association Research Foundation, Denver,
   CO, 15-21.

Milliken, B. (1996). "Summary of proposed stormwater management techniques for the village of
   Woodsong as of January 31, 1996." Proceedings, Watershed '96 Moving Ahead Together Techni-
   cal Conference and Exposition, Water Environment Federation, Baltimore, MD, June 8-12.

Minear, R. A. and Amy, G. L., eds. (1996). Water disinfection and natural organic matter, character-
   ization and control. American Chemical Society, Washington, D.C.

Minei, V. and Dawydiak, W. (1997). Proceedings, Second Marine andEstuarine Shallow Water
   Science and Management Conference, EPA/90 3/R-97 009. U.S. EPA Region III, Environmental
   Protection and Assessment Division, Philadelphia, PA.

Moe, C. L. (1997). "Waterborne transmission of infectious agents." Manual of Environmental Micro-
   biology,"  C. J. Hurst, G. R. Knudsen,  M. J. Mclnerny, L. D. Stetzenback, and M. V. Walter,
   eds., American Society for Microbiology, ASM Press, Washington, D.C.

Moore, J. A.,  Smyth, J., Baker, S., and Miner, J. R. (1988). "Evaluating coliform concentrations in
   runoff from various animal waste management systems." Special Report 817. Agricultural
   Experiment Stations, Oregon State University, Corvallis, OR.

Moore, A. C., Herwaldt, B. L., Craun, G. F., Calderon, R. L., Highsmith, A. K., and Juranek, D. D.
   (1994). "Waterborne disease in the United States, 1991 and 1992." Journal of 'the American
   Water Works Association, 86(2), 87-99.

Natural Resource Conservation Service (NRCS). (1998). National Handbook of Conservation
   Practices, Filter Strips, Code 393, September 1998.

NRCS. (1997). National Handbook of Conservation Practices, Contour Buffer Strips, Code 332,
   July 1997.

New York City Department of Environmental Protection. (1999). Final report for the Septic Siting
   Study Project.

Norton, W. D. and LeChevallier, M. W. (1997). "Survey of open finished water reservoirs for Giar-
   dia cysts and Cryptosporidium oocysts." New Jersey Department of Environmental Protection,
   Division of Science and Research.

Okun, D. A, Craun,  G. F., Edzwald, J. K., Gilbert, J. B., and Rose, J. B. (1997). "New York City: To
   filter or not to filter?" Journal of the American  Water Works Association, 89(3), 62-74.

Olivieri, P. W., Kruse, C. W., Kawata, K., and  Smith, J. S. (1977). "Microorganisms in urban
   stormwater." EPA-600/2-77-087, U.S. Environmental Protection Agency, Cincinnati, OH.

Ongerth, J. E. (1989). "Giardia cyst concentrations in river water." Journal of the American Water
   Works Association, 81(9), 81-86.
                                            4-32

-------
Ongerth, J. E. and Stibbs, H. H. (1987). "Identification of Cryptosporidium oocysts in river water."
   Applied and Environmental Microbiology., 53, 672-676.

Parker, J. A. and J. L. Darby (1995). "Particle-associated coliform in secondary effluents: Shielding
   from ultraviolet disinfection." Water Environment Research, 67(7), 1065-1075.

Perdek, J. M. and Borst, M. (2000). "Microbial particle association and combined sewer overflow
   disinfection." Disinfection 2000. Water Environment Federation, New Orleans, LA, March.

Peterjohn, W. T. and Correll, D. L. (1986). "The effect of riparian forest on the volume and chemical
   composition of baseflow in an agricultural watershed." Watershed Research Perspectives, D. L.
   Correll, ed., Smithsonian Press, Washington, D.C., 244-262.

Pettibone, G. W. and Irvine, K. N. (1996). "Levels and sources of indicator bacteria associated with
   the Buffalo River 'area of concern,' Buffalo, New York." Great Lakes Research, 22(4), 896-905.

Preston, S. D. (1996). "Study of nonpoint source nutrient loading in the Patuxent river basin, Mary-
   land." Water Resources Investigations Report No. WRIR-96-4273. U.S. Department of the Inte-
   rior, U.S. Geological Survey.

Quilez, J., Ares-Mazas, E., Sanchez-Acedo, C., del Cacho, E., Clavel, A., and Causape, A. C.
   (1996). "Comparison of oocyst shedding and the serum immune response to Cryptosporidium
   parvum in cattle and pigs." Parasitology Research, 82, 529-534.

Reddy, K. R., Khaleel, R., and Overcash, M. R. (1981). "Modeling nonpoint pollution due to animal
   waste with ARM II." Modeling Components ofHydrologic Cycle. Proceedings, International
   Symposium on Rainfall Runoff Modeling, V. P. Singh, ed., Water Resources Publications,
   Littleton, CO,  427-508.

Roach, P., Olson, M., Whitely, G, and Wallis, P. (1993). "Waterborne Giardia cysts and
   Cryptosporidium oocysts in the Yukon, Canada," Applied Environmental Microbiology, 59, 67-73.

Rose, J. B., Kayed, D., Madore, M. S., Gerba, C. P., Arrowood, M. J., and Sterling, C. R. (1988)
   "Methods for the recovery of Giardia and Cryptosporidium from environmental waters and their
   comparative occurrence." Advances in Giardia Research, P. Wallis and B. Hammond, eds.,
   University of Calgary Press, Calgary, Canada.

Rose, J. B., Gerba, C. P., and Jakubowski, W. (1991). "Survey of potable water supplies for
   Cryptosporidium and Giardia" Environmental Science and Technology, 25, 1393-1400.

Rose, J. B., Darbin, H, and Gerba,  C. P. (1988). "Correlations of the protozoa, Cryptosporidium and
   Giardia with water quality variables in a watershed." Proceedings,  International Conference on
   Water Wastewater Microbiology, Newport Beach, CA, 2, 43-1-43-6.

Rose, J. B. (1988). "Occurrence and significance of Cryptosporidium in water." Journal of the
   American Water Works Association, 80(1), 53-58.

Rose, J. B. (1997). "Environmental ecology of Cryptosporidium and public health implications."
   Annual Review of Public Health,  18, 135-61.

Rose, J. B., Lisle, J. T., and LeChevallier, M. (1997). "Waterborne Cryptosporidiosis: Incidence,
   outbreaks, and treatment strategies." Cryptosporidium and Cryptosporidiosis, R. Payer, ed.,
   CRC Press LLC, Boca Raton, FL.

Schillinger, J. E. and Gannon,  J. J.  (1985). "Bacterial adsorption and suspended particles in urban
   storm water." Jo urnalofthe Water Pollution Control Federation, 57(5), 384-389.
                                             4-33

-------
Sherer, B. M., Miner, J. R., Moore, J. A, and Buckhouse, J. C. (1992). "Indicator bacterial survival in
   stream sediments." Journal of Environmental Quality., 21, 591-595.

Singh, A. and McFeters, G. A. (1992). "Detection methods for waterborne pathogens." Environmen-
   tal Microbiology, Wiley-Liss, Inc., New York, 126-156.

States, S., et al. (1997). "Protozoa in river water: Sources, occurrence and treatment." Journal of the
   American Water Works Association, 89(9), 74-83.

States, S., Sykora, J., Stadterman, K., Wright, D., Baldizer, J., and Conley, L. (1996). "Sources,
   occurrence, and drinking water treatment removal of Cryptosporidium and Giardia in the Alle-
   gheny River." Proceedings, American Water Works Association Water Quality Technology Con-
   ference, New Orleans, LA, November 12-16, 1995. American Water Works Association, Denver,
   CO, Part II, 1587-1601.

Stepczuk, C., Owens, E. M., Effler, S. W., Auer, M. T., and Bloomfield, J.  A. (1998). "A modeling
   analysis of THM precursors for a eutrophic reservoir." Journal of Lake and Reservoir Manage-
   ment, 14(2-3), 367-378.

Stern, D. A. (1996).  "Monitoring for Cryptosporidium spp. and Giardia spp. and human enteric
   viruses in the watersheds of the New York City water supply system." Proceedings, Watershed
    '96 Moving Ahead Together Technical Conference and Exposition, Water Environment Federa-
   tion, Baltimore, MD, June 8-12.

Stewart, M. H., Yates, M.  V, Anderson, M. A., DeLeon, R., and Wolfe, R. L. (1998). "Modeling the
   effects of body-contact recreation on a drinking water reservoir." Proceedings, Source Water
   Protection Symposium: A focus on Waterborne Pathogens, San Francisco, CA, October 28-31.
   American Water Works Association, Denver, CO.

Stinson, M. K., Field, R., Moffa, P. E., Goldstein, S. L., and Smith K. J. (1998). "High-rate disinfec-
   tion technologies for wet weather flow (WWF)." Advances in urban wet weather pollution
   reduction. Water Environment Federation, Cleveland, OH, June 28-July 1.

Sykora, J., States, S. J., Bancroft, W. D., Boutros, S. N., Shapiro, M. A., and Conley, L. F. (1987).
   "Monitoring of water and wastewater for Giardia'' Proceedings, American Water Works Associa-
   tion Water Quality Technology Conference, Portland, OR, November 16-20, 1986. American
   Waterworks Association, Denver, CO, 1043-1054.

Toranzos, G. A.  and McFeters, G.  A.  (1997). "Detection of indicator microorganisms in environmen-
   tal freshwaters and drinking waters." Manual of Environmental Microbiology, C. J. Hurst, G. R.
   Knudsen, M. J. Mclnerny, L. D. Stetzenback, and M. V. Walter, eds., American Society for
   Microbiology, ASM Press, Washington, D.C.

U.S. Environmental Protection Agency (USEPA). (1991). "Guidance manual for compliance with
   the filtration and disinfection requirements for public water systems using surface water
   sources." NTIS Accession No. PB93-222933, Criteria and Standards Division, Office of Water.
   March 1991  Edition.

USEPA. (1994). "CSO Control Policy 59." Federal Register, (April 19), 18688.

USEPA. (1995). "Watershed protection: A project focus." EPA 841-R-95-003, August 1995.

USEPA. (1996a). "National primary drinking water regulations:  Monitoring requirements for public
   drinking water supplies: Cryptosporidium, Giardia, viruses,  disinfection by-products, water
   treatment plant data and other information requirements." Federal Register, 61(94; May 14),
   24354-24388.
                                            4-34

-------
USEPA. (1996b). ICRMicrobial Laboratory Manual, EPA/600/R-95/178, April, Washington, D.C.
USEPA. (1997). New York City filtration avoidance determination, .
USEPA. (1998a). "Announcement of the drinking water contaminant candidate list: Notice." Federal
   Register, 63(40; Mar. 2), 10273-10287.
USEPA. (1998b). "Interim enhanced surface water treatment rule." Federal Register., 63(241),
   69477-69521.
USEPA. (1999a). "Storm water technology fact sheet: Sand filters." EPA-F-99-007', September 1999.
USEPA. (1999b). "Storm water technology fact sheet: Storm water wetlands." EPA 832-F-99-025,
   September 1999.
USEPA. (1999c). "Storm water technology fact sheet: Vegetated swales." EPA 832-F-99-006, Sep-
   tember 1999.
USEPA. (1999d). "Method 1622: Cryptosporidium in water by filtration/IMS/FA." EPA-821-R-99-
   001, January, Washington, D.C.
USEPA. (1999e). "Method 1623: Cryptosporidium and Giardia in water by filtration/IMS/FA." EPA-
   821-R-99-006, April, Washington, D.C.
USEPA. (1999f). "Protocol for developing nutrient TMDLs," First edition. EPA Report No. 841-B-
   99-007, USEPA Office of Water, Washington, D.C., November 1999.
USEPA. (2000a). .
USEPA. (2000b). .
USEPA. (2000c). "National primary drinking water regulations, ground water rule, proposed rules."
   Federal Register, 65(91; May 10), 30193-30274.
Valiela, I, et al.  (1997). "Nitrogen loading from coastal watersheds to receiving estuaries: New
   method and application." Ecological Applications, 7(2), 358-380.
Walker, M. J., Montemagno, C. D., and Jenkins, M. B. (1998). "Source water assessment and
   nonpoint sources of acutely toxic contaminants: A review of research related to survival and
   transport of Cryptosporidium parvum" Water Resources Research, 34(12), 3383-3392.
Walker, S. E., Mostaghimi, S., Dillaha, T. A., and Woeste, F. E. (1990). "Modeling animal waste
   management practices: Impacts on bacteria levels in runoff from agricultural lands." Transac-
   tions of the American Society of Agricultural Engineers, 33(3), 807-817.
Webster, J. P. and Me Donald, D. W. (1995). "Cryptosporidiosis reservoir in wild brown rats (Ratus
   norvegicus) in the U.K." Epidemiology and Infection, 115,  207-209.
Wojtenko, I. (1999). Unpublished data.
Xiao, L. and Herd, R. P. (1994). "Infection patterns of Cryptosporidium and Giardia in calves."
   Veterinary Parasitology, 55, 257-262.
Young, K. D. and Thackston, E. L. (1999). "Housing density and bacterial loading in urban streams."
   Journal of Environmental Engineering Division oftheASCE, 125(12), 1177-1180.
                                            4-35

-------
4-36

-------
                                      CHAPTER 5

                                      Disinfection1
Introduction
The primary goal of the disinfection process in drinking water treatment is the inactivation of microbial
pathogens. These pathogens comprise a diverse group of organisms which serve as the etiological
agents of waterborne disease. Included in this group are bacterial, viral, and protozoan species. The
disinfection of potable water supplies was first initiated in the early part of the 20th century, and there
have been few developments in the area of public health which have been more effective in the control
of infectious diseases. While other unit processes, such as coagulation, clarification, and filtration, may
dramatically reduce the number of microbial pathogens, disinfection serves as the final and,  in some
cases, the only barrier to the entry of these organisms into the finished product water.

The disinfection process may be affected by a variety of both physical and biological factors. Tempera-
ture and pH are two physical factors which are known to play an important role in  the inactivation
process for most commonly used disinfectants (Hoff 1986). In actual practice, turbidity and particle
protection are two other physical parameters which influence disinfection efficiency, as well as clump-
ing of individual microorganisms (Berman et al. 1988). Resistance to chemical disinfection may vary
greatly between the various microorganisms of interest and also between different life-stages of indi-
vidual species, such as is seen with bacterial endospores or encysted  forms of protozoa.

Studies of microbial inactivation are often difficult to compare with one another owing  to differences in
methodological approaches. The role of mixing, the type of bioassays employed to determine viability,
the volume of sample analyzed, and the reporting of residual versus initial dosing concentrations of the
disinfectant are all factors which may vary greatly from one study to another. Often these parameters
are not described in sufficient detail in scientific manuscripts of these studies. Further, data collected
from field or pilot-scale conditions may show marked differences from the results of laboratory experi-
ments conducted under oxidant demand-free conditions. These discrepancies, along with the need to
determine the efficacy of disinfection for new and emerging waterborne pathogens, have spearheaded
the U.S. Environmental Protection Agency (EPA) research program on microbial inactivation. The
following discussion on potable water disinfection, categorized by individual oxidants, summarizes
the microbial inactivation research which has  been conducted or sponsored by EPA  during the time
period from 1980 to 1999.

Chlorine

Chlorination is the most frequently used form of halogen disinfection for treating drinking water in the
U.S. The use of chlorine has a long history in water treatment, and it has been successfully used in both
drinking water and wastewater applications. Data for chlorine inactivation of various  organisms have
often been used as a baseline measurement for determining a specific microorganism's resistance to
disinfection.
'Eugene Rice: ORD/NRMRL/WSWRD, AWBERC Mailstop 387, 26 West Martin Luther King Dr.:
Cincinnati, OH 45268, 513-569-7204, rice.gene@epa.gov.
                                              5-1

-------
Bacterial Inactivation

Research on chlorine inactivation of bacterial species represents the first research studies conducted on
potable water disinfection. Research in this area was initiated by predecessor organizations (U.S. Pub-
lic Health Service) of EPA and is notably represented by the pioneering work published by Butterfield
(1943) and colleagues on chlorine disinfection of enteric pathogens and sanitary indicator organisms
represented by coliform bacteria.
More recent studies on chlorine inactivation of coliform bacteria have centered on naturally occurring
organisms associated with particulate material and the protective effect provided by these particles
(Berman et al. 1988). In this study, sieves and nylon screens were used to separate primary sewage
effluent solids into particle fractions of less than seven micrometers (<7 jim) or greater than seven
micrometers (>7 \\.rs\) in size. The efficiency of separation was determined by electronic particle counting.
Indigenous coliforms associated with the particle fractions were tested to determine their resistance to
chlorine disinfection. Assays were conducted using the multiple tube  fermentation procedure,  and
levels of organisms were determined by the most probable number technique. Coliform bacteria asso-
ciated with the <7-|im fraction were inactivated more rapidly than the >7-|im fraction when exposed to
0.5 mg/L free chlorine, at pH 7.0 and 5°C (see Figure 5-1). Homogenization of the >7-|im fraction and
exposure to the disinfectant resulted in a rate of inactivation similar to that observed for the <7-|im
fraction. It is noteworthy that all of these  experiments were conducted in waters with turbidity levels
less than 1 NTU. These results indicate that particle association may play an important role in protect-
ing microorganisms from inactivation by chemical oxidation. Such findings support the importance of
water clarity in the disinfection process.
Survival of coliform bacteria in association with other biological organisms after chlorination was the
subject of two extramural research projects. Levy et  al.  (1984)  reported that Escherichia coli  and
Enterobacter cloacae were readily afforded protection from the effects of free-chlorine inactivation
when these organisms were associated with the amphipod Hyalella azteca. At a free available chlorine
level  of 1 mg/L, unassociated E. coli decreased to  less than 1% of the initial level at one minute of
exposure,  whereas more than 2% of the associated E. coli remained viable after 60 minutes of exposure.
This phenomenon was also noted for the E. cloacae culture. These findings support the contention that
bacteria associated with macroinvertebrates, which might commonly be found in water, are more
resistant to disinfection. In a similar study, it was demonstrated that bacteria which were ingested by
protozoa also exhibited an increased resistance to inactivation by chlorine. Examining both coliform
bacteria (E. coli, Citrobacterfreundii., Enterobacter agglomerans,  E. cloacae, Klebsiellapneumoniae,
and K. oxytocd) and several waterborne pathogens (Salmonella typhimurium, Yersenia enterocolitica,
Shigella sonnei, Legionella gormanii, and Campylobacter jejunf), King et al. (1988) reported that in-
gestion of these organisms by the protozoa Acanthamoeba castellanii  and Tetrahymena pyriformis
increased resistance to  chlorination by over 50-fold.
Escherichia coli O157:H7, a causative organism  of hemorrhagic colitis, has emerged as an important
waterborne pathogen in both drinking and recreational waters. Chlorination studies were conducted
using seven strains of this pathogen and were compared with four nonpathogenic, wild-type E.  coli
strains (Rice et al. 1999a). At a level of 1.1 mg/L of free chlorine, pH 7.0, 5°C, both the pathogenic and
nonpathogenic strains were inactivated by over four and half orders of magnitude within 120 seconds
under chlorine demand-free conditions (refer to Table 5-1). Results indicated that both the pathogen and
wild-type  strains were sensitive to chlorination and exhibited essentially the same rates of inactivation.
The outbreak of epidemic cholera in Peru  in 1991 prompted renewed interest in this classical bacterial
pathogen. At the request of the U.S. Centers of Disease Control, EPA conducted studies regarding the
efficacy of water treatment procedures for controlling the specific strains of  Vibrio cholerae Ol
isolated from the South American epidemic. During the course of these investigations it was found that


                                              5-2

-------
   10"-T
                                                               B
                10
                           20
                   MINUTES
                                                              MINUTES
                           20
                                      30
                   MINUTES
Figure 5-1. Effect of 0.5 mg of chlorine per liter at pH 7 and 5°C on inactivation of coliforms
           associated with sewage effluent particles. (A) Particles <7 um in size. (B) Particles
           >7 um in size. (C) Particles >7 um in size and homogenized. (Six experiments in
           each panel are shown, each represented by a different symbol.)
                                             5O

-------
Table 5-1. Chlorine Inactivation of Escherichia coli 0157:H7 and Wild-Type E. coli
Log10CFU/ml
Exposure Time
Isolate
Initial Inoculation
30 sec
60 sec
120 sec
Inactivation Rate (sec-1)
R2
E. coli 0157:H7
N009-6-1
N6001-8-10
N6021-5-1
N60049-26-1
N6059-7-2
N6 104-5-9
N6 114-7-2
Mean
5.63
5.78
5.78
5.68
5.72
5.62
5.63
5.69
2.60
2.52
2.54
2.35
2.42
2.40
2.52
2.48
1.88
1.44
1.52
1.40
1.74
1.69
1.66
1.62
0.82
0.72
0.66
0.54
0.86
0.72
0.89
0.74
-2.96
-3.06
-3.06
-3.00
-3.02
-2.96
-2.96
-2.93
0.82
0.68
0.54
0.86
0.72
0.89
0.82
0.82
E. coli (wild type)
A
B
C
D
Mean
5.53
5.79
5.68
5.52
5.63
2.66
2.60
2.48
2.34
2.52
1.80
1.48
0.92
0.95
1.28
1.52
0.81
0.84
0.39
0.89
-2.51
-2.68
-2.61
-2.50
-2.93
0.61
0.60
0.61
0.61
0.71
V. cholerae was capable of shifting to a rugose phenotype that exhibited increased resistance to chlori-
nation (Rice et al. 1993; Clark et al. 1994). Viable organisms were able to be recovered after 30 minutes
of exposure to 2.0 mg/L free chlorine, pH 7 at 20°C. The rugose variant displayed a deviation from
first-order kinetics, with a persistence of a subpopulation of organisms after an initial inactivation of
two to three orders of magnitude (see Figure 5-2). Subsequently, it was demonstrated that the rugose
form of V. cholerae retained its virulence and was capable of infecting human volunteers (Morris et al.
1996). While the natural occurrence of the rugose V. cholerae in nature is unknown, these studies
emphasize the dramatic differences which may be observed in disinfection kinetics between different
phenotypes of the same bacterial species.

Another bacterial pathogen which has been placed on the EPA Contaminant Candidate List (CCL) as a
potential waterborne pathogen is Helicobacterpylori. Chlorine inactivation experiments conducted at
5°C, at pH 6, 7, 8 (refer to Table  5-2) indicated that, under oxidant demand-free conditions, this organ-
ism was very sensitive to disinfection (Johnson et al. 1997). Inactivation studies such as these, coupled
with occurrence data, provide useful information to regulatory authorities regarding the potential pub-
lic health threat posed by emerging pathogens.

Two closely related bacterial pathogens, Campylobacter jejuni  and Arcobacter butzleri, have been
implicated as waterborne pathogens. Under EPA funded research, Blaser et al. (1986) examined the
chlorine susceptibility of three strains of C. Jejuni. A mean level of inactivation in excess of four orders
of magnitude occurred after 1 minute exposure at pH 6 to 0.1 mg/L free chlorine at 4°C in buffer. Rice
et al. (1999b), reporting on the results of an in-house research project, demonstrated that at 5°C in a
well water source (pH 7.1), three strains of A. butzleri were inactivated at a mean level of greater than
five orders of magnitude after exposure to 0.61 mg/L of total available chlorine residual (refer to Table
5-3). These findings indicate that these organisms are also sensitive to chlorine.

In a study designed to evaluate  the role of aerobic endospores as surrogate organisms for evaluating
water treatment plant performance,  indigenous  aerobic endospores from a river water source were
                                              5-4

-------
                       100 -
                        10 -
                         1 -
                        .10 -
                        .01 -
                                                          Peruvian Rugose
                                    , Gulf Coast Smooth
                                3—Peruvian Smooth
                            0.0
                                       0.5
                                                  1.0
                                              Time in Minutes
                                                             1.5
                                                                        2.0
Figure 5-2. Comparison of inactivation of smooth strain of cholera to rugose variant by free
           chlorine (2.0 mg/L free chlorine, pH 7.0, 20°C).


exposed to chlorination (Rice et al. 1996). The experiments using the river water (pH 6.9) were con-
ducted at 23°C. An exposure time of 180 minutes was required at a total available chlorine level of 1.75
mg/L to achieve a three order of magnitude inactivation.  These indigenous endospores, existing in
different stages of maturity and metabolic dormancy and representing distinct life forms (endospores)
of aerobic spore-forming bacteria, exhibited increased resistance to chlorination  as opposed to most
other vegetative bacterial cells.
Table 5-2. Chlorine Inactivation ofH. pylori in Chlorine Demand-Free Buffer, 5°C, 0.5 mg of
          Free Chlorine per Liter
LogmCFU/ml
Exposure Time
Isolate
43504


CVD33


CP41


MEAN


pH
6
7
8
6
7
8
6
7
8
6
7
8
Initial Inoculum
4.91±0.02
4.63± 0.03
4.93± 0.02
4.74±0.03
4.74±0.03
4.74±0.02
4.04±0.03
4.19±0.03
5.48±0.01
4.56
4.52
5.05
10 sec
2.64±0.05
3.90±0.05
3.89±0.03
2.65±0.03
2.78±0.02
3.49±0.06
3.12±0.03
2.87±0.04
3.86±0.02
2.80
3.18
3.75
20 sec
2.36±0.04
3.60±0.04
3.54±0.03
2.43±0.04
2.56±0.03
2.86±0.04
1.48±0.07
2.62±0.01
3.52±0.02
2.09
2.93
3.31
40 sec
1.30±0.10
1.78±0.04
2.18±0.02
1.95±0.01
2.00±0.08
2.34±0.06
O.20
2.15±0.02
2.57±0.05
1.15
1.98
2.36
80 sec
<0.20
<0.20
0.20
0.20
0.20
O.20
O.20
O.20
1.70±0.07
0.20
0.20
0.70
Inactivation
Rate (sec-1) R2
-
-
-
-
-
-
-
-
-
-3.90 0.82
-3.48 0.94
-3.92 0.87
                                               5-5

-------
Table 5-3. Chlorine Inactivation ofArcobacter butzleri in Well Water, 5°C, Mean Chlorine
          Concentrations: 0.46 mg I'1 Free Chlorine, 0.61 mg-1 Total Chlorine at pH 7.06
Log^CFUml-1
Exposure Time
Organisms Tested
Field isolate No. 1
Field isolate No. 2
Mean
Arcobacter butzleri
(ATCC49616)
TimeO
6.07±0.00
5.98±0.00
6.02
5.65±0.00
15 sec
2.46±0.01
3.16±0.06
2.83
2.90±0.08
30 sec
2.04±0.06
1.72±0.12
1.88
1.63±0.15
60 sec
1.09±0.09
0.70±0.00
0.58
0.74±0.04
Inactivation Rate (sec -1) R2
-
-
6.09 0.81
5.80 0.84
Viral Inactivation

In terms of resistance to chlorine inactivation, animal viruses and bacteriophage are generally consid-
ered to be more resistant than vegetative bacterial cells. Hoff (1986) summarized EPA-sponsored
research on chlorine inactivation of polio virus under oxidant demand-free conditions. At 5°C, a two
order of magnitude reduction occurred over a free-chlorine residual between 0.6 to 2.5 mg/L over a
time range of 0.7 to 2.4 minutes of exposure.

In an extensive research project, Berman et al. (1984) studied the inactivation of simian rotavirus SA-
11 using several disinfectants. Studies were conducted at 5°C with purified preparations of single viri-
ons and with cell-associated virions. A residual available chlorine concentration of 0.5 mg/L under
oxidant demand-free conditions yielded a four order of magnitude inactivation at pH 6 in less than 15
seconds of exposure. Under similar conditions at pH 10, a two order of magnitude reduction required
approximately 1.5 minutes of exposure. In all instances, the cell-associated virus was more resistant to
inactivation than were preparations of single virions.

Berman et al. (1992) also conducted studies on the inactivation of the bacterial virus MS2 coliphage,
which uses E. coli as its bacterial host. For free available chlorine, it was reported that a 2 mg/L residual
yielded inactivation greater than four orders of magnitude in pH 7 oxidant demand-free buffer at 5°C
after an exposure time of 1 minute. It was concluded that, under these experimental conditions, MS2
coliphage, like the animal virus rotavirus SA11, was very sensitive to inactivation by free chlorine.

Protozoan Inactivation

EPA, in response to a congressional mandate, developed the concept of target  pathogens. The target
pathogen concept was based upon the likelihood of the presence of a pathogen in  a given water type and
that organism's innate resistance to inactivation by chemical  disinfectants. Under the Surface Water
Treatment Rule (Federal Register 1989), the cyst stage of the protozoan parasite Giardia lamblia was
chosen as the target pathogen for surface waters used as a drinking water source.  Consequently, several
research studies, both internal and extramural,  were devoted to determining the inactivation kinetics
for Giardia (Clark et al. 1989).

Using in vitro excystation to determine cyst viability, Jarroll et al. (1981) concluded that Giardia cysts
exhibited resistance to free chlorine. Inactivation levels were determined after exposures to chlorine for
10, 30, or 60 minutes at different pH levels and temperatures. Inactivation greater than two orders  of
magnitude was reported after 10 minutes of exposure to 1.5 mg/L chlorine at pH 6, 7 at 25°C. Similar
inactivation occurred after exposure to 1.5 mg/L of chlorine at pH 6 at 15°C. An exposure time  of 60
minutes at 5°C, 2 mg/L of chlorine at pH 6 and 7 was required to achieve a greater than two order  of
magnitude inactivation. This report confirmed that inactivation kinetics occur at a slower rate with
decreasing temperature.


                                              5-6

-------
In studies conducted on cysts of G. lamblia from both symptomatic and asymptomatic human donors,
Rice et al. (1981), also using in vitro excystation, reported similar findings for chlorine inactivation. In
this study, cysts of G. muris, derived from a murine model, were also exposed to chlorination and were
shown to exhibit a somewhat greater resistance to chlorination than that observed for G. lamblia. Using
the murine model, Hoff et al. (1985) compared animal infectivity with in vitro excystation for quantita-
tively determining the viability of G. muris cysts before and after exposure to free residual chlorine. It
was concluded that in vitro excystation was an adequate indication of cyst infectivity for the host and
could thus be used to determine the effects of chemical disinfection on cyst viability.

The development of an animal model using Mongolian  gerbils for the propagation of G. lamblia af-
forded a readily available source of cysts of this human parasite. Rubin et al. (1989) used cysts derived
from the gerbil to conduct further chlorine inactivation experiments.  The gerbil-derived cysts were
reported to be somewhat more resistant than cysts obtained from  human donors, suggesting that the
host source of cysts may effect cyst resistance to disinfection. The microsporidian parasite Encephali-
tozoon intestinalis has been placed on the EPA's CCL. This parasite is transmitted in the environment
in a resistant-spore stage. The small size of the infective spore (circa 2  jim) presents special challenges
for physical removal and thus heightens the need for information on chemical inactivation. Preliminary
studies (Rice et al. 1999c) were conducted using spores of E. intestinalis produced in tissue culture from
a rabbit kidney cell line (RK13). It was found that exposure to 2 mg/L of chlorine for 8 to 16 minutes was
sufficient to achieve two orders of magnitude inactivation. Further research is planned to further define
inactivation parameters at decreased exposure times.

Chloramine

The use of chloramination in drinking water treatment has gained increasing popularity as concerns
have grown regarding adverse health effects attributed to chlorine disinfection by-products. Chemi-
cally, chloramines are a complex group of disinfectants; however, only the monochloramine form is of
major interest for drinking water treatment. When chlorine and ammonia are mixed in equimolar pro-
portions, nearly all free available chlorine is converted to monochloramine. This is an equilibrium
reaction and is affected by the chlorine-to-ammonia ratio. The rate formation is also pH dependent.
Monochloramine  is considered a weak biocide in comparison to free available chlorine, requiring ex-
posure times of 25 to 100 times greater than chlorine to achieve comparable inactivation. The efficacy
of disinfection for monochloramine is pH  dependent and increases with decreasing pH values.  Tradi-
tionally, laboratory studies have concentrated on the use of preformed chloramine,  and while these
results yield conservative values for inactivation, they are not representative of the disinfectant's effec-
tiveness under field conditions.  The chlorine-to-ammonia ratios, pH,  and method of application are
crucial parameters to consider in chloramine inactivation and must be clearly delineated to determine
biocidal effectiveness under various experimental conditions (Hoff 1986).

Bacterial Inactivation

As part of the study on particle protection from  disinfection,  Berman et al.  (1988) looked at
monochloramine inactivation of coliform bacteria. At pH 8, the coliforms associated with the smaller-
sized particle fraction were inactivated more rapidly than the organisms associated with the larger-
sized particle fractions, thus mimicking the results observed with free available chlorine.  The time
required to achieve two orders of magnitude inactivation with monochloramine was approximately 50-
fold greater than the time required for the same amount of inactivation with free chlorine.

In a study designed to determine the effect of the method of preparing monochloramine, Berman et al.
(1992) reported on the inactivation of E. coli andKlebsiellapneumoniae. The organisms were exposed
                                               5-7

-------
to free chlorine followed by the rapid addition of ammonia to determine inactivation for "forming
monochloramine." This type of inactivation was compared to inactivation using preformed
monochloramine. Their results showed that E. coli and K. pneumonias were rapidly inactivated by
"forming monochloramine," which was attributed to the brief presence of free chlorine during the
formation reaction. Preformed chloramine was also capable of inactivating the bacteria, but at a slower
rate.

Campylobacter also exhibited greater resistance to inactivation by monochloramine compared to free
chlorine (Blaser et al. 1986). Three strains ofC.jejuni were inactivated more than two orders of mag-
nitude after 15 minutes of exposure to 1.0 mg/L preformed monochloramine at pH 8.0, 5°C. It was also
noted in this study that E. coli was also more readily inactivated by free chlorine than by monochloramine.

Viral Inactivation

Preformed monochloramine was used to inactivate simian rotavirus SA11 (Berman et al. 1984). At pH
8, 5°C, a monochloramine level of 10 mg/L was required for over 6 hours of exposure to achieve two
orders of magnitude inactivation. As was noted with free available chlorine, the cell-associated SA-11
exhibited increased resistance to monochloramine when compared with the preparation composed of
single virions.

MS2  coliphage was used  in experiments to determine the effect of the method of preparing
monochloramine on inactivation of the bacteriophage (Berman et al. 1992). Monochloramine prepared
in situ by initial addition of chlorine to a suspension containing MS2, followed by subsequent addition
of ammonia, inactivated the  coliphage more rapidly than preformed monochloramine (see Figures 5-3
and 5-4). The exposure to free chlorine was given for the rapid viral inactivation observed in the in situ
experiments. Inactivation was more rapid at 15°C than at 5°C and when the chlorine-to-nitrogen weight
ratio was 5:1 compared to 3:1.
                        100
                         10-
                          1 -
                        0.1
                                 0.25
                                          0.5      0.75
                                             MINUTES
                                                          1      1.25
Figure 5-3. MS2 inactivation by 2 mg/L chlorine at pH 7, 5°C.
                                             5-8

-------
                         100
                          10 =
D

c£
                                                    MONOCHLORAMINE

                                                    0.25 MIN FREE CHLORINE

                                                    0.50 MIN FREE CHLORINE

                                                    0.75 MIN FREE CHLORINE

                                                    1.00 MIN FREE CHLORINE
                        0.01
                                                                /s
                                             3     4
                                             MINUTES
                                                              10
                                                                    240
Figure 5-4. MS2 inactivation by monochloramine and combined chlorine.


Protozoan Inactivation

There has been relatively little research on the role of chloramination for the inactivation of protozoa.
Korich et al. (1990) showed that monochloramine was the least effective disinfectant compared to free
available chlorine, chlorine dioxide, and ozone for inactivating Cryptosporidium parvum oocysts. At
5°C, an exposure to 80 mg/L of monochloramine for 90 minutes was required to achieve a one order of
magnitude inactivation.

Chlorine Dioxide

Chlorine dioxide exists as an undissociated gas dissolved in water in the pH range from 6 to 9, and the
disinfection efficiency increases within this range with increasing pH. Existing as an undissociated gas
makes this oxidant more vulnerable to volatilization than free chlorine or monochloramine. Chlorine
dioxide is a relatively stable disinfectant and is less likely to react with oxidant demand substances than
free chlorine. Chlorine dioxide is a potent disinfectant and is generally  considered to have a biocidal
efficiency greater than free chlorine or monochloramine (Hoff 1986).

Bacterial Inactivation

An extramural project was conducted to determine the role of antecedent growth conditions on the
inactivation of E. coll andLegionellapneumophilaby chlorine dioxide (Berg et al. 1988). In this study,
chemostat-grown bacteria were compared to batch-culture bacteria after being dosed with an initial
concentration of chlorine dioxide of 0.75 mg/L. A resistant subpopulation of each organism survived in
the presence of a constant chlorine dioxide residual. The observed resistance was attributed to a pheno-
typic trait which could be manipulated by altering the antecedent growth conditions in the chemostat
cultures.

Viral Inactivation

Studies were conducted on chlorine  dioxide inactivation of simian rotavirus SA11 (Berman et al. 1984).
More than two orders of magnitude  inactivation was achieved when the virus was exposed to 0.5 mg/L
chlorine dioxide at pH 6, 5°C in less than 1 minute. At the same temperature and disinfectant concentra-
tion, a greater than two order of magnitude inactivation was achieved in less than 15 seconds at pH 10.
                                               5-9

-------
Protozoan Inactivation

Owens et al. (1999) compared three bioassay procedures for determining the viability of Cryptosporidium
parvum oocysts exposed to chlorine dioxide. As with other reported studies, it was found that the
biocidal activity of chlorine dioxide was pH dependent, with better inactivation occurring at higher pH
levels. Differences in inactivation were observed between different lots of oocysts. Differences were
also observed between the bioassay procedures. The in vivo neonatal mouse model demonstrated a
higher level of inactivation compared to the in vitro tissue  culture infectivity assay and a modified
excystation procedure. These results led the authors to suggest that care must be taken in evaluating and
comparing laboratory inactivation data for this parasite.

Ozone

Ozone represents the most potent biocide examined in disinfection studies sponsored or conducted by
EPA. While ozone is a potent biocide, it is also very unstable and highly  reactive. Like chlorine diox-
ide,  ozone exists in water as a dissolved gas. It is subject  to losses due  both to volatilization and
reactions with demand substances present in the water. It would appear from most studies that ozone is
relatively unaffected by pH within the range normally encountered in water treatment. Maintaining
stable ozone levels in either laboratory or field conditions is very difficult, and factors such as applied
dose, as well as final residual, must be addressed when describing ozone inactivation studies (Hoff
1986). Concentration of disinfectant, especially in pilot-plant studies, is often determined by averaging
dissolved ozone residual measurements collected at various points in the ozone contactor. The average
ozone concentration, Cavg, is then multiplied by the mean exposure time, T2,to determine the product of
the concentration (C) and exposure time (T) (CT) value (mg  min/L).

Bacterial Inactivation

Bacteria are very sensitive to ozone inactivation. Bacterial endospores are the only life-stage of bacteria
which have been shown to exhibit resistance to ozonation. In a pilot-plant study, Miltner et al. (1997)
noted that indigenous aerobic endospores in filtered Ohio River water (ORW) required a CT value of
19 to achieve a three order of magnitude inactivation (see Figure 5-5).

Viral Inactivation

In the same pilot-plant study, Miltner et al. (1997) reported an inactivation of greater than two orders of
magnitude for polio virus exposed to a CT of 1.2 mg min/L at  pH 7.6 in a temperature range between 23
and 24°C. A tissue culture plaque assay utilizing BGM cells was used to determine virus viability (see
Figure 5-5).

Protozoan Inactivation

Encysted  forms of protozoa represent of the most  difficult forms of microorganisms to inactivate by
disinfection. Consequently, the use of ozonation has gained popularity as a means of inactivating these
organisms. Ozone was found to be effective in the inactivation ofGiardia cysts (Wickramanayake et al.
1985). Using in vitro excystation to determine viability, these investigators noted that the murine surro-
gate G. muris was consistently more resistant to ozonation than G. lamblia. The average CT values for
a two order of magnitude inactivation at 5°C, pH 7, was 1.9 mg min/L for G. muris and 0.55 mg min/L
for G. lamblia cysts. For the same conditions at 25°C, the mean CT value  for G.  muris was 0.25 mg
min/L and 0.17 mg min/L for G. lamblia. In a pilot-scale ozonation study, Miltner et al. (1997) reported
that at pH 7.65, 23 to 24°C in ORW, a CT value of 0.75 mg min/L was required to achieve two orders
of magnitude inactivation of G.  muris cysts (see Figure 5-5).
                                              5-10

-------
Figure 5-5. Comparison of inactivation of microbial populations exposed to ozone in filtered
           ORW. Temperatures = 23.6 to 25.2°C.
               100
              0.01
                                                         • Indigenous Endospores

                                                         A C. parvum oocysts

                                                         • C. muris oocysts

                                                           G. muris cysts

                                                           Poliovirus 1
                                       10         15
                                         CT (mg min/L)
                                                             20
                                                                       25
Ozone is  one of the few chemical oxidants that has been shown to be capable of inactivating
Cryptosporidium spp. under normal water treatment conditions. Korich et al. (1990) demonstrated that
ozone was able to inactivate C. parvum oocysts. In this study, using a neonatal mouse assay to deter-
mine infectivity, an inactivation of greater than two orders of magnitude was observed after 5 minutes
of exposure to  1 mg/L of ozone (circa CT value of 5 mg min/L). Similar results for a 2.5 order of
magnitude inactivation (CT value of 6.56 mg min/L) at 20°C were reported by Rennecker et al. (1999)
for C. parvum using a modified in vitro excystation procedure for determining oocyst viability. In a
pilot-scale study using ORW, a CT value of 4.0 mg min/L  inactivated approximately 1.4 orders of
magnitude of C. parvum oocysts  at pH 7.6, 23 to 24°C (Miltner et al. 1997) using the neonatal mouse
model to determine infectivity.

Ultraviolet (UV) Irradiation

Microbial inactivation by UV irradiation has not been  a major area of research within the Agency's
drinking water program. In the 1980s, an extramural research project was sponsored to examine the
potential for using UV disinfection for small drinking water systems (Carlson et al. 1985). It was noted
that cysts  of Giardia muris were  significantly more resistant to UV treatment than E. coli or Yersinia
spp. Results from this project also suggested that hydraulic short circuiting and entrapped air in UV
reactors may decrease the efficiency of inactivation.
                                              5-11

-------
An in-house project was also conducted during this time period to study the effect of UV irradiation on
cysts of the human pathogen Giardia lamblia (Rice and Hoff 1981). In this study, G. lamblia cysts were
found to be resistant to high doses of germicidal UV irradiation. There has been renewed interest in the
use of UV light for inactivating waterborne protozoan parasites. Recent studies suggest that UV irra-
diation is an effective treatment option for inactivating oocysts of Cryptosporidium (Clancy et al. 1998).
This finding, in contrast to previous studies regarding the  effect of UV light on the inactivation of
protozoa cysts, has been related to the method by which viability was determined. Earlier studies relied
upon in vitro excystation as a bioassay for determining inactivation. Current research suggests that
excystation is not a reliable method for determining inactivation of cysts and oocysts after exposure to
UV light. It appears that animal infectivity is necessary to adequately determine the biocidal activity of
UV light for the encysted protozoa. Current in-house studies are re-evaluating earlier Giardia inactiva-
tion studies in light of this development.

Summary

Studies on microbial inactivation have been a major part of the Agency's research efforts in the area of
drinking water treatment over the past 20 years. Disinfection research currently is focused on organ-
isms found on the Agency's CCL. Research on the inactivation of emerging microbial agents coupled
with the evaluation of new treatment methodologies remains a priority in the Agency's drinking water
program.

References

Berg, J. D., Hoff, J. C., Roberts, P. V, and Matin, A. (1988). "Resistance of bacterial subpopulations
   to disinfection by chlorine dioxide." Journal of the American Water Works Association, 80(9),
    115-119.

Berman, D. and Hoff, J. C. (1984). "Inactivation of simian rotavirus SA11 by  chlorine, chlorine
   dioxide,  and monochloramine." Applied and Environmental Microbiology, 48(2), 317-323.

Berman, D., Rice, E. W., and Hoff, J. C. (1988). "Inactivation of particle-associated coliforms by
   chlorine and monochloramine." Applied and Environmental Microbiology, 54(2), 507-512.

Berman, D., Sullivan, R.,  and Hurst, C. J. (1992). "Effect of the method of preparing
   monochloramine upon inactivation of MS2 coliphage, Escherichia coli, and Klebsiella
   pneumoniae" Canadian Journal of Microbiology, 38, 28-33.

Blaser, M. J., Smith, P. E, Wang, W. L., and Hoff, J. C. (1986). "Inactivation of Campylobacter
   jejuni by chlorine and monochloramine." Applied and Environmental Microbiology, 51(2),
   307-311.

Butterfield, C. T, Wattie,  E., Megregian, S.,  and Chambers, C. W. (1943). "Influence of pH and
   temperature on the survival of coliforms and enteric pathogens when exposed to free chlorine."
   Public Health Reports, 58(51), 1-30.

Carlson, D. A., et al. (1985). "Ultraviolet disinfection of water for small water supplies." EPA/600/2-
   85/092, Environmental Protection Agency.

Clancy, J. L., Hargy, T. M., Marshall, M. M., and Dyksen, J. E. (1998). "UV light inactivation of
   Cryptosporidium oocysts." Journal of the American Water Works Association, 90(9), 92-102.

Clark, R. M., Read, E. J.,  and Hoff, J. C. (1989). "Analysis of inactivation of Giardia lamblia by
   chlorine." Journal of Environmental Engineering Division of the ASCE, 115(1), 80-90.
                                             5-12

-------
Clark, R. M., Rice, E. W., Pierce, B. K., Johnson, C. H., and Fox, K. R. (1994). "Effect of aggrega-
   tion on Vibrio choleras inactivation." Journal of Environmental Engineer ing Division of the
   ASCE, 120(4), 875-887.

Federal Register. (1989). "Drinking water; National primary drinking water regulations; Filtration,
   disinfection; Turbidity, Giardia lamblia, viruses, Legionella, and heterotrophic bacteria; Final
   Rule." 54(No. 124), 40 CFR, Parts 141  and 142.

Hoff, J. C., Rice, E. W., and Schaefer, III, F. W. (1985). "Comparison of animal infectivity and
   exystation as measures ofGiairdia muris inactivation by chlorine." Applied and Environmental
   Microbiology, 50(4), 1115-1117.

Hoff, J.  C. (1986). "Inactivation of Microbial Agents by Chemical Disinfectants." EPA/600/2-86/067,
   Environmental Protection Agency.

Jarroll, E. L., Bingham, A. K., and Meyer, E. A. (1981). "Effect of chlorine on Giardia lamblia cyst
   viability." Applied and Environmental Microbiology, 41(2), 483-487.

Johnson, C. H., Rice, E. W., and Reasoner, D. J. (1997). "Inactivation of Helicobacterpylori by
   chlorination." Applied and Environmental Microbiology, 63(12), 4969-4970.

King, C. H., Shotts, Jr., E. B., Wooley, R. E., and Porter, K. G. (1988). "Survival of coliforms and
   bacterial pathogens within protozoa during chlorination." Applied and'Environmental'Microbiol-
   ogy, 54(12), 3023-3033.

Korich, D. G., Mead, J. R., Madore, M. S.,  Sinclair, N. A., and Sterling, C. R. (1990). "Effects of
   ozone, chlorine dioxide, chlorine,  and monochloramine on Cryptosporidium parvum oocysts
   viability." Applied and Environmental Microbiology, 56(5), 1423-1428.

Levy, R. V, Cheetham, R. D., Davis, J., Winer, G.,  and Hart, F. L. (1984). "Novel method for study-
   ing the public health significance of macroinvertebrates occurring in potable water." Applied and
   Environmental Microbiology, 47(5), 889-894.

Miltner, R. J., Shukairy, H. M., Rice, E. W., Owens, J. H., Schaefer, III, F. W., and Dahling, D. R.
   (1997). "Comparative ozone inactivation of Cryptosporidium and other microorganisms." Pro-
   ceedings, International Symposium on Waterborne Cryptosporidium,  New Port Beach, CA,
   March 2-5, 229-241.

Morris, Jr., J. G., et al. (1996). "Vibrio cholerae Ol can assume a chlorine-resistant rugose survival
   form that is virulent in humans." Journal of Infectious Diseases, 174, 1364-1368.

Owens, J. H., Miltner, R. J., Slifko, T. R., and Rose, J. B. (1999). "In vitro excystation and infectivity
   in mice and cell culture to assess chlorine dioxide inactivation of Cryptosporidium oocysts."
   Proceedings, Water Quality Technology Conference, Tampa, FL, October 31-November 3.

Rennecker, J. L., Marinas, B. J., Owens, J. W. and Rice, E. W. (1999). "Inactivation of
   Cryptosporidium parvum oocysts with ozone."  Water Research, 33(11), 2481-2488.

Rice, E. W. and Hoff, J. C. (1981). "Inactivation of Giardia lamblia cysts by ultraviolet irradiation."
   Applied and Environmental Microbiology,  42(3), 546-547.

Rice, E. W., Hoff, J. C., and Schaefer, III, F. W. (1982). "Inactivation of Giardia cysts by chlorine."
   Applied and Environmental Microbiology, 43(1), 250-251.

Rice, E. W., et al. (1993). "Vibrio cholerae Ol can assume a rugose survival form that resists killing by
   chlorine, yet retains virulence." InternationalJournal of Environmental Health Research, 3, 89-98.
                                              5-13

-------
Rice, E. W., Fox, K. R., Miltner, R. J., Lytle, D. A., and Johnson, C. H. (1996). "Evaluating plant
   performance with endosporess." Journal of the American Water Works Association, 88(9),
   122-130.

Rice, E. W., Clark, R. M., and Johnson, C. H. (1999a). "Chlorine inactivation of Escherichia coli
   O157:H7." Emerging Infectious Diseases, 5(3), 461-463.

Rice, E. W., Rodgers, M. R., Wesley, I. V, Johnson, C. H., and Tanner, S. A. (1999b). "Isolation of
   Arcobacter butzleri from ground water." Letters in Applied Microbiology, 28, 31-35.

Rice, E. W., Johnson, C. H., Naumovitz, D. W., Marshall, M. M., Plummer, C. B., and Sterling,  C. R.
   (1999c). "Chlorine disinfection studies of Encephalitozoon (Septata) intestinalis" Poster presented
   at Annual Meeting of'the American Society for Microbiology, Chicago, IL, May 30-June 3.

Rubin, A. J., Evers, D. P., Eyman, C. M., and Jarroll, E. L. (1989). "Inactivation of gerbil-cultured
   Giardia lamblia cysts by free chlorine." Applied and Environmental Microbiology, 55(10),
   2592-2594.

Wickramanayake, G. B., Rubin, A. J., and Sproul, O. J. (1985). "Effects of ozone and storage tem-
   perature on Giardia cysts." Journal of the American Water Works Association, 77(8), 74-77.
                                             5-14

-------
                                      CHAPTER 6

                               Alternative Disinfectants1
Introduction
Chlorination of drinking water results in the formation of numerous disinfection byproducts (DBFs),
several of which are regulated. Water systems seeking to meet maximum contaminant levels (MCLs)
of regulated DBFs may consider various approaches to limiting DBFs: removing the precursor com-
pounds early in the treatment train before the disinfectant is applied, using less chlorine, using alterna-
tive disinfectants to chlorine, and removing DBFs after their formation. Combinations of these ap-
proaches may also be considered. Removing DBFs after their formation is  a method that is generally
not employed. Whatever approach is selected, the system must be certain that the effectiveness of the
disinfection is  not jeopardized. This chapter presents recent  studies conducted by, or funded by, the
U.S. Environmental Protection Agency's (EPA's) Office of Research and Development (ORD) in Cin-
cinnati that examine the use of three alternative oxidants: chloramine, chlorine dioxide, and ozone.

Models for Assessing Halogenated DBF Precursors

The precursors for halogenated DBF formation are not well known. In the Disinfectants/Disinfection
By-Product (D/DBP) Rule, in which enhanced coagulation  is used as a treatment technique to control
identified and unidentified DBFs, total organic carbon (TOC) is the surrogate for DBF precursors.
While TOC or dissolved organic carbon (DOC) may be used as a surrogate, they will not well represent
the precursors  of specific DBFs. While the precursors for the specific DBFs are not well known, an
indirect means  of quantitating the control of specific DBF precursors is to sample the water influent to
and effluent from a treatment process, chlorinate both waters under a specific set of conditions (pH,
temperature, time, etc.), and examine the concentrations of the specific DBFs. Differences in these
concentrations  may be attributed to the effectiveness of the treatment process. For example, a raw
water may form 200 mg/L of total trihalomethane (TTHM), while an ozonated water may form 150 mg/
L. Ozonation may then be considered to have oxidized 25% of the TTHM precursors. The set of chlo-
rination conditions driving the DBF reaction is very important. Three models  for DBF precursor were
employed in the studies discussed in this chapter.

In the formation potential (FP) model, a relatively large dose of chlorine is used, and the reaction time
is typically  long, e.g., one week. This is assumed to drive the DBF reaction to completion, thus utilizing
all the precursor. The fate of precursors can be assessed across treatment processes, but, as conditions
are relatively extreme, the resulting DBF concentrations are rarely representative of a system's finished
water.

Systems may therefore choose to chlorinate under conditions unique to their distribution system. In the
simulated distribution system (SDS) model, the fate of precursors can be assessed; the resulting DBF
concentrations  are representative of the system's finished water. For example, SDS TTHM concentra-
tions before and after biological filtration may be 85 and 70 |ig/L, respectively. The biofilter is shown
to remove 17% of the TTHM precursor. The 70 |lg/L is meaningful, as the chlorination conditions were
'Richard J. Miltner: ORD/NRMRL/WSWRD, AWBERC Mailstop 689, 26 West Martin Luther King
Dr., Cincinnati, OH 45268, 513-569-7403, miltner.richard@epa.gov.
                                             6-1

-------
representative. The 70 |ig/L is below the Stage 1 D/DBP Rule MCL of 80 |ig/L. If FP conditions were
used, the TTHM formation potential (TTHMFP) concentrations before and after biological filtration
might have been 140 and 116 |lg/L, respectively. The biofilter would show 17% removal of the TTHM
precursor, but as chlorination conditions were not representative, the 116 |lg/L would incorrectly imply
that the MCL was exceeded. Additionally, the use of the FP model might result in a skewed distribution
of bromo- and chloro-trihalomethanes (TFDVIs). The chlorine-to-bromide ratio impacts DBF speciation
(Shukairy et al. 1994), and this ratio is typically higher when the FP model is employed.

Because chlorination conditions are unique to the systems employing them, the SDS model does not
allow for comparison of results from different waters.  The uniform formation condition (UFC) model
was developed to address this issue (Summers et al. 1996). In this model, the chlorination conditions of
the mean national distribution system are targeted, i.e., 1  mg/L free-chlorine residual at 24 hours at pH
8 at  20°C. Thus, DBF precursor  control can be assessed, the resulting DBF  concentrations can be
considered relative to MCLs, and  results can be compared from one water to another.

Chloramines

Chloramines are the second most commonly used final disinfectant in drinking water treatment after
free  chlorine. Although generally not as effective a  disinfectant as free chlorine, an advantage of
chloramination is minimization of the formation of DBFs.

Halogenated DBF Formation

The  formation of DBFs by chloramines is significantly lower than by free chlorine. Stevens et al.
(1989) treated humic acid solutions with free chlorine, monochloramine, and chlorine dioxide at the
bench-scale. Monochloramine was dosed as preformed chloramines without free chlorine. The solu-
tions contained no bromide, so only chloro-DBPs resulted. Figure 6-1 shows the relative formation of
CHC13 and nonpurgable organic halide (NPOX), a subset of the surrogate total organic halide (TOX).
The data show that CHC13 formation and NPOX formation by monochloramine is small compared to
that formed by free chlorine,  confirming that a treatment strategy for the control of DBFs is the use of
chloramines as an alternative final disinfectant to free  chlorine.

Parallel  oxidants were studied  at the pilot scale in Jefferson Parish, LA, treating  Mississippi River
(MR) water  (Lykins and Koffskey 1986; Lykins et al. 1989). Coagulated, settled, and  filtered waters
were treated with free chlorine, monochloramine, chlorine dioxide, and ozone and compared to a par-
allel nondisinfected water. Results for TOX formation representing the mean of one year of sampling
are listed in Table 6-1. Disinfectant contact times averaged 31 minutes. No details were given regarding
how monochloramine was prepared or dosed. The TOX formation by monochloramine, however, was
low compared to that by free chlorine.

At the pilot-scale, Miltner (1990)  studied chlorination and chloramination of Ohio River (OR) water.
Parallel  plants were predisinfected, alum-coagulated to  control turbidity, settled, and  filtered. In the
Table 6-1. Oxidation of MR Water (Lykins and Koffskey 1986)
Parameter
Residual, mg/L
TOX, |^g C1-/L

None

25

Free Chlorine
1
263
Disinfectant
Monochloramine
2.1
117

Chlorine Dioxide*
0.5
85

Ozone
0.5
15
  * Chlorite = 0.6 mg/L, free chlorine = 0.1 mg/L
                                             6-2

-------
                     600
                     500
                   O
                   c» 400
                   co
                   O)
                   g
                   "co
                     300
                   o
                   O
                   X
                   O
                   CL
                     200
                     100
                   O   0
                   CO
                   CO
                   O)
                     100
                   cu
                   o
                   c
                   o
                   O
                     200
                                     Chlorine
                                     20 mg/L
                                                        Reaction Time Scale - h
                                                                0
                                                             Chlorine Dioxide
                                                             20.7 mg/L
Ill
Q. Q. Q.
Figure 6-1. Formation of DBFs by alternative oxidants (Stevens et al. 1989).

chloramine plant, NH4OH was added in stoichiometric excess before rapid mix; no free chlorine was
present. Chlorine was added to the rapid mix in both plants. Chlorine was dosed on both plants so that
residuals carried through the filters and clear wells and met the Ten-State Standards (TSS) (Recom-
mended Standards for Water Works  1992) of 0.2 to 0.5 mg/L free chlorine and  1.0 to 2.0 mg/L
combined chlorine "at distant points in the distribution system." In this study, the distribution system
was simulated as clear well waters held 3 days. Table 6-2 shows results for mean sampling of clear well
effluents. HAN4 represents four haloacetonitriles (HANs): trichloro- (TCAN), dichloro- (DCAN),
bromochloro- (BCAN) and dibromo- (DBAN). Chloropicrin (CP) was not detected. Cyanogen chlo-
ride, chloral hydrate (CH), and the haloacetic acids (HAAs) were not analyzed.

THMs, HANs and 1,1,1-trichloropropanone (Ill-TCP) were detected in the finished water on the
prechlorinated  plant.  As expected, these DBFs were  not  detected in the finished water  on  the
prechloraminated plant. The TOX concentration was appreciably  lower in the effluent of the
prechloraminated plant and similar in concentration to the TOX concentration in the OR water influent
to the plant.

-------
Table 6-2. DBF Formation in Finished OR Water (Miltner 1990)
Concentration, Hg/L
Parameter
TOX
TTHM
CHC13
CHBrCl2
CHBr2Cl
CHBr3
HAN4
TCAN
DCAN
BCAN
DBAN
Ill-TCP
CP
Prechlorinated
115
15.6
10.0
4.6
0.7
0.3
3.1
3.1
<0.1
ND
ND
2.8
ND
Prechloraminated
20
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND = not detected
Although bacteria penetrated farther into the chloraminated plant, heterotrophic plate count (HPC) and
total coliform (TC) densities were comparable in the two finished waters, indicating that chloramination
following the TSS was sufficient for bacterial control.

On the same pilot plant, Miltner et al. (1990) studied parallel post-chlorination and post-chloramination
of OR water following  preozonation, alum coagulation to control turbidity, and settling. Following
settling, the stream was split for parallel filtration. Following filtration, the post-disinfectants were
applied to the clear well influents. For monochloramine, NH4OH was added in stoichiometric excess
prior to chlorine; no free chlorine was present. A parallel plant without preozonation was similarly
treated with post-chlorination at the clear well's influent. Finished waters collected from clear wells
and held 3 days were used to simulate distribution system waters. Residuals targeted recommendations
of the TSS (Recommended Standards for Water Works 1992); residuals after 3 days were near 0.2 mg/L
free chlorine and 0.7 mg/L monochloramine. Results are given in Table 6-3. HAA6 represents six
HAAs: trichloro- (TCAA), dichloro- (DCAA), chloro- (CAA), bromochloro- (BCAA), bromo- (BAA),
and dibromoacetic acid  (DBAA).

Comparing ozone/chloramine and ozone/chlorine, the concentrations of 3-day DBFs were significantly
lower with  use of monochloramine, as expected. TOX was also formed upon  chloramination, but at
significantly lower concentrations than in the chlorinated waters and only near double the raw water
TOX concentration of 24.4 |lg C1~/L. The only noted exception in DBF formation was the formation of
cyanogen chloride upon chloramination.

The THMs, HAAs, HANs, CH, CP, Ill-TCP and CNC1 concentrations in Table 6-3 were converted to
their TOX equivalents and compared to their TOX concentrations. In the two chlorinated waters, these
DBFs accounted for nearly 40% of the TOX, leaving nearly 60% of the TOX unaccounted for, i.e., 60%
of the TOX was comprised of compounds other than these DBFs. In the chloraminated water, however,
these DBFs made up only 23% of the TOX. Thus, the use of chloramine resulted in significantly lower
DBF formation than the use of free chlorine (refer to Table 6-3), but a larger percentage of what was
formed was unaccounted for by the measured DBFs. This unaccounted-for, halogenated material may
be nitrogenous.
                                             6-4

-------
Table 6-3. DBF Formation in Simulated Distribution OR Water (Miltner et al. 1990)

Parameter
TOX, ng C1-/L
TTHM
CHC13
CHBrCl2
CHBr2Cl
CHBr3
HAA6
TCAA
DCAA
BCAA
CAA
BAA
DBAA
HAN4
TCAN
DCAN
BCAN
DBAN
CH
CP
Ill-TCP
CNC1
Concentration,
p,g/L Unless
O3 Chloramine O3 Chlorine
51.5
5.6
4.5
0.8
0.2
ND
6.1
1.5
3.9
0.3
0.5
0.1
ND
2.9
ND
2.4
0.4
0.2
0.8
0.1
0.4
2.5
207
75.1
39.6
21.1
13.0
1.5
39.7
10.0
19.2
6.8
1.5
0.3
2.0
4.8
ND
2.6
1.7
0.6
5.8
1.6
1.1
ND
Noted T-Test* for Chlorinated Waters
Better Better
Post Chlorine With O3 Same Without O3
259 x
90.4 x
55.5 x
24.4 x
10.2 x
0.3 x
62.6 x
20.1 x
30.9 x
8.5 x
1.4 x x
0.3 x
1.5 x
5.7 x
0.2 x
3.5 x
1.9 x
0.1 x
4.2 x
0.5 x
0.8 x
ND
* at 95% confidence level
ND = not detected
Comparing ozone/chloramine and ozone/chlorine, the densities of HPC in the two clear wells were
similar. TC bacteria were not detected in any clear wells. These data suggest that chloramination fol-
lowing the recommendations of the TSS was sufficient for bacterial control. With this pilot-scale study
and the pilot-scale study noted previously, water distribution system materials could not be simulated
during the 3-day storage of chlorinated water in clean glassware; therefore, the question of bacterial
regrowth in the presence of the weaker chloramine disinfectant during distribution remains.

Nonhalogenated DBF Formation

The formation of nonhalogenated DBFs by chloramines is also significantly lower than by free chlo-
rine. Miltner (1993) reported on OR water at the bench-scale with several oxidants. With
monochloramine, the formation of formaldehyde and the P17 strain of assimilable organic carbon (AOC-
P17) was negligible  and  similar to the background concentrations. With free chlorine, however,
formaldehyde and AOC-P17 formation was evident (see Table 6-4). The data suggest that systems
employing monochloramine will experience lower concentrations of these bacterial nutrients in their
distribution systems than those employing free chlorine.
                                             6-5

-------
Table 6-4. Oxidation of OR Water (Miltner 1993)
Oxidant
None
Monochloramine
C102
KMnO4
Free chlorine
Ozone
Dose, mg/L

2
1
1
o
J
2
Time, min

15
15
15
15
7.5
Formaldehyde, Jlg/L
0.8 ±0.15
0.9
2.0
2.2 ±0.85
2.8
17.1
AOC-P17,ngCeq/L*
95
96
129
132
158
202
*as acetate
Chlorine Dioxide

Chlorine dioxide is a widely used disinfectant in drinking water treatment. It has long been used for
taste and odor control and for iron and manganese control and has gained in acceptance as an effective
disinfectant. An advantage of C1O2 treatment is minimization of the formation of DBFs; it does this by
oxidation of DBF precursors and by relatively minimal formation of DBFs themselves. A disadvantage
is the presence of chlorite and chlorate resulting from C1O2 treatment. The former is regulated under the
D/DBP Rule and the latter is of health concern.

Halogenated DBF Formation

The formation of DBFs by chlorine dioxide is significantly lower than by free chlorine.  Stevens et al.
(1989) treated humic acid solutions with free chlorine, monochloramine, and chlorine dioxide at the
bench-scale. The solutions contained  no bromide, so only chloro-DBPs resulted. Figure 6-1 shows the
relative formation of CHC13 and NPOX. The data show no CHC13 formation and little NPOX formation
by chlorine dioxide compared to that formed by free chlorine. Thus, a treatment strategy to control
DBFs is the use of chlorine dioxide, an alternative oxidant to free chlorine.

Table 6-1 shows that TOX formation  by C1O2 on the Jefferson Parish pilot plant was low compared to
that of free chlorine. Some of the TOX in the ClO2-treated water may be a result of inefficient C1O2
generation, as a yearly average free chlorine residual of 0.1 mg/L was detected following C1O2 contact.

The effect of C1O2  on TTHM control was observed by Lykins and Griese (1986) at Evansville, IN.
Pilot-plant effluents (no prior disinfection) were treated with chlorine and C1O2 and held 3  days to
simulate distribution  system conditions. Results are presented in Table 6-5. Even with a high C1O2
residual and 3 days' reaction time, TTHM formation was similar to the background TTHM concentra-
tion in the raw water and very low compared to the formation by free chlorine.

Based on the success of piloting, a full-scale switch to C1O2 was made at Evansville. Evansville has two
parallel full-scale plants. One was treated with C1O2 as a preoxidant, with an average dose of 1.4 mg/L
C1O2. Both plants were chlorinated ahead of the filters. No details were given on the free-chlorine doses
to the two plants or whether the free chlorine dose on the ClO2-treated plant may have been lower as a

Table 6-5. TTHM Formation at Evansville (Lykins and Griese 1986)
Simulated Distribution Concentrations
Parameter
TTHM, |^g/L*
Chlorine residual, mg/L
Chlorine dioxide residual, mg/L
Chlorine
141
2.5
—
Chlorine Dioxide
1.4
—
1.9
  * TTHM in raw water = 1.2 |j,g/L.
                                             6-6

-------
result of C1O2 satisfying some of the chlorine demand. Nevertheless, pretreatment with C1O2 was effec-
tive in lowering TTHM formation (Lykins and Griese 1986). Finished water TTHM averaged 37.3 mg/
L without C1O2 and 25.5 mg/L with C1O2. There are two explanations for this improvement. Lykins and
Griese (1986) hypothesized that lower TTHM concentrations were a result of ClO2's oxidation of DBF
precursor prior to downstream chlorination. Miltner (1976) showed that C1O2 oxidized DBF precursors
to the extent that lower concentrations of DBFs were formed with subsequent chlorination. The second
explanation contends that, if the free chlorine level was lower on the ClO2-treated plant as a result of
ClO2's oxidation of chlorine demand, lower TTHM may also result.

The effect of C1O2 on DBF precursors was also studied by Lykins and Koffskey (1986) at the pilot scale
at Jefferson Parish. Coagulated, settled, and filtered waters were treated with free chlorine, chlorine
dioxide, and ozone. TTHM and TOX precursors were assessed by FP. Table 6-6 shows C1O2 oxidized
34% and 17%, respectively, of TTHM and TOX precursors.

Table 6-6. Oxidation of DBF Precursors at Jefferson Parish (Lykins and Koffskey 1986)
Percent Removal
Parameter
TTHMFP
TOXFP
Chlorine Dioxide
34
17
Ozone
44
31
Nonhalogenated DBF Formation

While the formation of halogenated DBFs by chlorine dioxide may be minimal compared to free
chlorine, C1O2 can form nonhalogenated by-products. Miltner (1993) reported on OR water at the
bench-scale with several oxidants. With chlorine dioxide, the formation of formaldehyde and AOC-
P17 approached that of free chlorine (see Table 6-4).

EPA (unpublished data) sampled a full-scale plant treating OR water with KMnO4 and C1O2. The C1O2
dose was near 1.0 mg/L. Results in Table 6-7 show the presence, confirmed by chromatograph/mass
spectroscopy (GC/MS), of aldehydes and ketones in ClO2-treated water. While the presence of these
compounds in the ClO2-treated water may also be a result of their presence in the source water and/or
the result of KMnO4's ability to form them (refer to Table 6-4), the concentration of several of them was
enhanced by C1O2 treatment.

Table 6-7. Aldehyde/Ketone Formation in OR Water
Concentration, Hg/L
Parameter
Formaldehyde
Acetaldehyde
Propanal
2-Butanone
Butanal
Pentanal
2-Hexanone
Hexanal
Octanal
Benzaldehyde
C = GC/MS
KMnO4 Treated
Raw
10.1
11.2
1.5

2.6


1.6


confirmed
C1O2 Treated
Mixed
10.8 C
24. 1C
25. 1C
25. 1C
40.6 C
C
C
16.2 C
C
C

                                             6-7

-------
Richardson et al. (1994) utilized XAD® resin extraction and GC/MS to qualitatively search for by-
products in waters taken from a pilot plant in Evansville, IN, employing C1O2. They identified 20
compounds in ClO2-treated water that were not identified in the raw water. Most were carboxylic acids
in the C4 through C16 range. A few ketones were also identified.

Controlling Concentrations of Chlorine Dioxide, Chlorite, and Chlorate

Chlorite and chlorate are found in ClO2-treated waters. They may result from unreacted C1O2 generator
products, the reduction of C1O2, or the disproportionation of C1O2 and its related products. Table 6-1
shows chlorite measured in ClO2-treated waters at Jefferson Parish. As both chlorite and chlorate have
toxicological implications, as the D/DBP Rule regulates chlorite in drinking water, and as the D/DBP
Rule limits the allowable concentration of C1O2 in finished waters, the control of all three species is
important to systems employing C1O2.

Granular activated carbon (GAC) with an empty bed contact time of 9.6 minutes was studied  at the
Evansville pilot plant for control of C1O2 and chlorite (Lykins et al. 1990; Lykins et al. 1989). Results
are given in Table 6-8. They  show that much of the C1O2 is reduced to chlorite downstream  of its
application, that C1O2 was completely reduced before entering the GAC bed so its  control by  GAC
could not be evaluated, and that a substantial percentage of the chlorite was controlled by GAC. Chlo-
rite control by GAC would very likely be time dependent; no details were given as to GAC age, bed
volumes treated, etc.

Table 6-8. Control of CIO  and CIO at Evansville (Lykins et al. 1990)

Dose
Settled
GAC influent
GAC effluent
Chlorine Dioxide, mg/L
4.2
0.5
ND
ND
Chlorite, mg/L

2.3
3.0
0.3
ND = not detected
Griese et al. (1991) studied the use of reducing agents to control C1O2 and chlorite at the bench scale at
Evansville. Applying excess sulfur dioxide and sulfite was found to remove both C1O2 and chlorite.
SO2/SO2 was most efficiently applied after the oxidant demand for C1O2 had been met. The reaction, in
part, depended on the dissolved oxygen (DO) concentration. Using this means  of control, unreacted
SO2/SO2 would complicate post-disinfection. They assumed unreacted SO2/SO2 would be removed
by post-chlorination, but require a higher post-chlorine dose than would otherwise be required. This
means of control was not pursued, however, since unacceptable concentrations (exceeding 1 mg/L) of
chlorate were formed. They found similar results with the application of excess metabisulfite. Results
with excess thiosulfate were more promising. It controlled both C1O2 and chlorite, was pH and time
dependent, was not affected by DO, and did not form complicating concentrations of chlorate. But it
would also pose a problem for finished waters, as unreacted thiosulfate would complicate post-disinfection.

Griese et al. (1991) also studied ferrous chloride at the pilot scale at Evansville. They found this to be
the most promising reducing agent as it controlled both C1O2 and chlorite and formed only very low
concentrations of chlorate. Residual iron was controlled with prefilter chlorination. Other studies by
Griese et al. (1992) at the pilot scale expanded on ferrous iron as a means of control and focused on
chlorate. They found chlorate could be present as a product of the C1O2 generation process, as a result
of ClO2's reaction with sunlight, and as a result of uncontrolled C1O2 and chlorite reacting with post-
chlorine. They found that chlorate formation during ferrous iron treatment was higher at lower pH and
that adding lime to a pH range of 7.0 to 7.5 minimized chlorate formation.
                                             6-8

-------
Ozone

Ozone is a less commonly used disinfectant in drinking water treatment. Among the many benefits of
ozonation of drinking water are effective inactivation of microbes, taste and odor control, iron and
manganese control, oxidation of DBF precursors, and the enhancement of biological oxidation in fil-
ters. However, ozone results in the formation of bromate and of biodegradable organic matter (BOM).
Bromate is regulated under the D/DBP Rule. BOM includes ozone by-products (OBPs) like aldehydes,
keto acids, carboxylic acids, AOC, and biodegradable dissolved organic carbon (BDOC). These OBPs
may be responsible for regrowth of bacteria in distribution systems and can be controlled in-plant if
biological  oxidation is allowed to occur in downstream filters. (Refer to Chapter 7, "DBF Control
Through Biological Filtration.")

Halogenated DBF Formation

The formation of halogenated DBFs as a result of ozonation is minimal. EPA unpublished data
showed  the low-level formation of brominated DBFs by ozone in the conduct of pilot-scale studies
(Miltner et al. 1990; Miltner and  Summers 1992) of OR water. CHBr3, BAA, and DBAA were occa-
sionally detected at concentrations below 2 |ig/L, presumably through the reaction of molecular ozone,
bromide, and natural organic matter (NOM). Downstream chlorination significantly increased the con-
centrations of these DBFs.

Table 6-1 describes TOX concentrations as a result of ozonation of MR water at the Jefferson Parish
pilot plant.  A 31-minute contact time resulting in a 0.5-mg/L ozone residual did not increase TOX
concentrations beyond those in the background water.

Oxidation of Halogenated DBF Precursor

DBF precursors tend to be more humic than non-humic and of higher rather than lower molecular
weight. In Chapter 10, Coagulation, Dryfuse et al. (1995) describe TOX, TTHM, and HAA6 precursors
located predominantly in the humic and higher-molecular-weight fractions of East Fork Lake (EFL)
water. Koechling et al. (1996), studying the reaction of ozone with NOM, found that ozone converted
portions of the humic fraction to non-humic compounds and converted portions of the higher-molecu-
lar-weight fraction to lower-molecular-weight compounds. Therefore, ozone reacts with NOM to oxi-
dize a portion of the DBF precursors; this results in lower concentrations of DBFs formed by down-
stream chlorination. Coupled with the low-level formation of bromo-DBPs by ozone itself, this finding
supports ozone's role as an alternative oxidant for halogenated DBF control.

Tables 6-9 and 6-10 describe ozone's oxidation of DBF precursor in pilot-scale studies of OR water and
EFL water, respectively. In the OR water study (Table 6-9),  ozone was applied to raw OR water at a
transferred ozone/TOC ratio near 0.8 mg/mg. Miltner et al. (1992) studied ozone dose dependency and
demonstrated with pilot-scale ozonation of OR water that, at transferred ozone/TOC ratios above 0.7
mg/mg,  no further oxidation of TTHM,  HAA6, and TOX precursors occurred. While ozone changed
the nature of the DOC (to more non-humic and to  smaller-molecular-weight compounds),  it did not
significantly change its concentration, as Table 6-9 demonstrates. Ozone significantly oxidized com-
pounds that absorb at 254nm (UV254) and consequently lowered the water's specific ultraviolet (UV)
absorbance (SUVA), or UV254 divided by DOC. Using FP as a means of assessing DBF precursors,
removal of TTHM, HAA6, and TOX precursors by ozone was within the 14% to 19% range. However,
ozone altered the nature of CH and CP precursors to the extent that they increased.

In the EFL water study, ozone was applied to coagulated and settled EFL water at a transferred ozone/
TOC ratio near 0.9 mg/mg. Table 6-10 shows the removal of DBF precursors first by coagulation and
                                             6-9

-------
Table 6-9. Mean Changes in DBF Precursors in Ozonated OR Water (Miltner 1993)
Parameter
DOC, mg/L
UV254, cm-1
SUVA,L/mg-m
TTHMFP, ng/L
HAA6FP, ng/L
TOXFP, ng C1-/L
CHFP, ng/L
CPFP, ng/L
Raw
2.28
0.051
2.23
190
155
449
33
1.5
Ozonated
2.24
0.027
1.20
164
126
367
40
2.9
Percent

47
46
14
19
18
+21
+93
Removal








Table 6-10. Mean Changes in DBF Precursors in Ozonated EFL Water (Miltner et al. 1996)
Parameter
DOC, mg/L
UV254, cm-1
SUVA, L/mg-m
UFC TTHM, ng/L
UFCHAA6,^g/L
UFC TOX, ng C1-/L
UFC CH, ng/L
UFCHAN4,^g/L
Chlorine demand, mg/L
Raw
5.83
0.204
3.50
311
332
984
32.2
12.6
9.27
Coagulated Settled
2.77
0.068
2.12
83.6
79.1
300
11.4
6.9
3.04
Coagulated Settled,
Ozonated
2.74
0.023
0.84
48.7
52.5
210
18.3
3.6
2.86
Percent Removal
by Ozonation

66
60
42
34
30
+61
48
6
then by ozonation. Precursors were assessed by UFC. Again, ozone did not affect the DOC concentra-
tion, but removed SUVA, chlorine demand, and TTHM, HAA6, HAN4, and TOX precursors. Precur-
sors for CH were increased by ozone oxidation.

Ozone oxidation of TTHM and TOX precursors was also observed by Lykins and Koffskey (1986) in
pilot-scale Jefferson Parish waters (see Table 6-6).

Results of another pilot-scale study of ozonation of raw OR water (Miltner et al. 1990) are presented in
Table 6-3. A preozonated/post-chlorinated stream was compared to a stream that was post-chlorinated
only. Post-chlorination was conducted under SDS conditions of 3-day chlorination targeting TSS chlo-
rine residuals (Recommended Standards for Water Works 1992). The data show lower concentrations
of finished water TTHM, HAA6, HAN4, and TOX in the preozonated stream. Two factors account for
this: (1) preozonation removed a portion of the precursors by oxidation, and (2) preozonation lowered
the  chlorine demand so that less  chlorine was applied and was present to drive DBF formation. Excep-
tions were finished water concentrations of CH, CP, and Ill-TCP,  the precursors of which were
increased by preozonation.

Table 6-3 shows a statistical test  of the two chlorinated finished waters,  showing 95% confidence in the
lower concentrations of TOX, TTHM, CHC13, HAA6, TCAA, DCAA, and BCAA when preozonated,
and showing 95% confidence in the higher concentrations of CHBrCl2, CHBr3, DB AA, DB AN, CH,
and CP when preozonated. Ozone's increase of CH  and CP precursors has been previously  noted.
Ozone's effect on the bromochlorospeciation of halogenated DBFs is just as, if not more, important.
                                            6-10

-------
Shukairy et al. (1994) studied ozone's oxidation of DBF precursors in OR water and the resulting
bromochlorospeciation. DBF precursors were assessed by chlorinating under FP conditions. Figure
6-2 shows representative results for individual HAAs. Oxidation of precursors for TCAA, DCAA, and
BCAA occurred to the extent that, over the range of transferred ozone/DOC ratios up to 2.54 mg/mg,
their concentrations decreased upon chlorination. This is consistent with the behavior of HAA6 precur-
sors also observed by Miltner et al. (1992), noted previously.

At transferred ozone/DOC ratios at and below 1.11 mg/mg, however, concentrations of DBAA in-
creased. This behavior is consistent with the statistically significant increases in DBAA concentrations
presented  in Table 6-3 in which ozonation took place at 0.8 mg/mg transferred ozone/TOC. As the
ozone increased, however, there was a significant decrease in DBAA formation. At transferred ozone/
DOC ratios at and below 1.11 mg/mg, changes in bromide were small. At higher ozone doses, more
bromate was formed from bromide (refer to Table 6-15), thus bromide concentrations fell. As a result,
the bromide/DOC ratios and the bromide/free-chlorine ratios decreased. Decreases in either ratio favor
the formation of more chlorinated species as was observed above  1.11 mg/mg.

Figure 6-2 also shows the bromide incorporation factor n\ which is defined as the molar ratio of the
brominated HAAs to the total HAA6 (Shukairy et al. 1994). At lower ozone doses, when DBAA con-
centrations increased, «' increased; at higher ozone doses, when DBAA concentrations decreased, «'
decreased. Behavior for TFDVIs was similar with increased concentrations of CF£Br2Cl and CF£Br3 and
increased «, the bromide incorporation factor for TTHM, at lower ozone doses, and decreased concen-
trations of these two THMs and n at higher ozone doses. It is important  to remember that, although
brominated HAAs and THMs in chlorinated waters increase over the range of transferred ozone/DOC
ratios common to drinking water treatment, their concentrations are relatively  small (below a few
ug/L); TTHM and HAA6 concentrations, however, decrease, demonstrating ozone's overall benefit in
oxidizing DBF precursors.
          100
              A DCAA    O TCAA   O BCAA
              ri = 0.050    ri = 0.058                                      n' = 0.031

           -50 |	1	1	
             0123
                                 Transferred-Ozone-to-DOC Ratio - mg:mg
                Control concentrations: CAA - not detected, DCAA - 50.0 \ig/L, TCAA - 45.6 \ig/L,
                BCAA - 5.45\ig/L, BAA - not detected, DBAA - 0.36 \ig/L

Figure 6-2. Effect of ozone dose on HAA formation in OR water (Shukairy et al. 1994).
                                              6-11

-------
In the studies summarized in Tables 6-3 and Tables 6-9 through 6-13, microbial densities were moni-
tored. Ozonation was always observed to bring TC bacterial densities to less than one colony /100 mL.
Ozonation never destroyed all HPC bacteria; they flourished sufficiently in nutrient-rich downstream
waters to acclimate biological filters. Post-disinfection with chlorine or chloramine to achieve require-
ments of the TSS (Recommended Standards for Water Works 1992) lowered finished water HPC levels
typically to 2 to 3 log/mL densities.

Ozone By-Product Formation

Ozonation results in the formation of a number of OBPs. Several pilot-scale studies were conducted
with transferred ozone/TOC ratios in the 0.8 to 0.9 mg/mg range. Results are presented in Tables 6-11,
6-12, and 6-13. Ozonation of raw OR water over several months (Table 6-11) demonstrated the statis-
tically  significant formation of 9 aldehydes and a ketone,  principally formaldehyde, acetaldehyde,
glyoxal, and methyl glyoxal.

Table 6-11. Mean Formation of Aldehydes and Ketones in Ozonated OR Water (Miltner et al.
           1991; Miltner 1993)
Concentration, Jlg/L
Parameter
Formaldehyde
Methyl glyoxal
Glyoxal
Acetaldehyde
Propanal
Hexanal
Decanal
Nonanal
Pentanal
2-Butanone
Raw
1.1
0.1
1.3
1.1
1.2
0.4
0.4
0.5
ND@0.2
**ND@0.1
Ozonated *
11.7
11.4
7.6
3.5
2.1
1.4
1.1
1.0
0.3
0.1
  ND = not detected
  * Increase with ozone significant at >95% confidence level unless otherwise noted.
  ** Increase with ozone significant at 94% confidence level.

The same pilot study also demonstrated the formation of other OBPs (Table 6-12) over several months'
operation: two keto acids, AOC, and BDOC. The aldehydes, ketones and keto acids are small-molecu-
lar-weight compounds resulting from ozone's oxidation of the NOM. They are easily biodegradable
(assimilable) and are considered to make up portions of the AOC and BDOC. Note that most of these
measures of BOM are naturally present in the raw water and are enhanced upon ozonation. The BDOC
made up 17% of the DOC in the raw water and was enhanced to 32% following ozonation. Total AOC
(the P17 and the NOX fractions) made up 7% of the raw water DOC and was enhanced to 30% follow-
ing ozonation. Because these OBPs are, by definition, assimilable and biodegradable, they can serve as
substrates for bacterial regrowth in distribution systems if not controlled by biological filtration (refer
to Chapter 7, "DBF Control Through Biological Filtration").

The nature of the NOM can influence OBP formation. In OR water (Table 6-12) with a DOC concentra-
tion  of 2.28 mg/L, BDOC  and total AOC reached concentrations of 0.71 mg/L and 665 |ig Ceq/L,
respectively, with transferred ozone/TOC near 0.8 mg/mg. Following coagulation and sedimentation of
EFL water, DOC was lowered to a concentration of 2.77 mg/L. With transferred ozone/TOC near 0.9
                                            6-12

-------
Table 6-12. Mean Formation of AOC, BDOC, and Keto Acids in Ozonated OR Water (Miltner 1993)
Parameter
DOC, mg/L
BDOC, mg/L
AOC-P 17, |^g Ceq/L*
AOC-NOX,|^g Ceq/L**
Glyoxylic acid, [ig/L
Pyruvic acid, |J,g/L
Raw
2.28
0.39
37
129
0.2
0.4
Ozonated *
2.24
0.71
71
594
34.6
12.4
  * Increase with ozone significant at >95% confidence level except for DOC.
  * as acetate
  ** asoxylate
Table 6-13. Mean Formation of OBPs in EFL Water (Miltner et al. 1996)
Parameter
DOC, mg/L
BDOC, mg/L
Total AOC, \ig Ceq/L
Formaldehyde, |J,g/L
Methyl glyoxal, |j,g/L
Glyoxal, |J,g/L
Glyoxylic acid, |J,g/L
Pyruvic acid, [ig/L
Raw
5.83
1.27
399
7.7
0.2
0.1
0.2
0.7
Coagulated Settled
2.77
0.53
203
NA
NA
NA
NA
NA
Ozonated
2.74
1.21
1314
21.1
4.4
15.1
46.1
15.6
NA = not analyzed
mg/mg, BDOC and total AOC reached 1.21mg/L and 1314 |ig Ceq/L, respectively, or roughly twice
that of the OR water at approximately the same DOC and transferred ozone/TOC ratio (Table 6-13). A
biological filter treating EFL water would have to be more efficient than one treating OR water to
ensure similar distribution system loading of AOC and BDOC.

Ozone concentration can influence OBP formation. Several bench- and pilot-scale studies with ozonation
of different batches of OR water examined dose dependency (Shukairy et al. 1992; Miltner et al. 1992;
Miltner et al. 1998).  Maximum transferred ozone/TOC ratios were in the 2.5 to 2.8 mg/mg range.
Generally, much of the formation occurs at lower ozone doses. An example is given in Figure 6-3 for
total AOC; at a ratio of 2.5 mg/mg, formation was not yet maximized. Similar behavior of still-increas-
ing formation at higher ratios was also observed for BDOC, glyoxylic acid, and formaldehyde (Miltner
et al. 1992; Miltner et al. 1998). Maximized and level formation was observed for acetaldehyde and
propanal  (Miltner et al. 1992; Shukairy et al. 1992).

Pentanal  was found to reach a maximum near 1.8 mg/mg and then diminish in concentration as more
ozone was introduced (Shukairy et al. 1992).  It is possible that pentanal was converted to pentanoic
acid at higher ozone doses. Glyoxal and methyl  glyoxal exhibited different behaviors in different batches
of OR water. In one, it behaved like pentanal, i.e., observed at lower concentration at higher doses after
reaching  a maximum  (Shukairy et al. 1992). In another, they had not yet reached maximums at 2.8 mg/
mg (Miltner et al. 1992). At ozone doses more typical of drinking water treatment (0.5 to 1.5 mg/mg
transferred ozone/TOC), concentrations of OBPs are generally increasing; therefore, minimizing the
ozone dose can limit OBP formation.
                                             6-13

-------
                 250
                            0.5
2.5
3.0
                                     1.0       1.5       2.0
                                   Transferred Ozone/TOC (mg/mg)

Figure 6-3. Effect of ozone dose on total AOC in OR water (Miltner et al. 1998).
Ozone staging (when it is applied in the treatment plant) can influence OBP formation. Miltner et al.
(1998) studied bench-scale ozonation of raw, coagulated and settled OR waters. Raw water was ozonated
at a transferred ozone/TOC ratio of 1.4 mg/mg and settled water at a transferred ozone/TOC ratio of 1.1
mg/mg. These were based on oxidation of SUVA and DBF precursors in raw and settled waters and on
achieving the concentration and time (CT) required to inactivate approximately 2 logs Cryptosporidium
parvum oocysts. The inactivation studies treated C. parvum oocyst-spiked OR water (Owens  et al.
2000). Table 6-14 shows that ozonating settled water resulted in lower concentrations of OBPs than
ozonating raw water prior to coagulation and settling. Coagulation removed some of the ozone-reactive
NOM. Further, coagulation removed ozone demand such that the inactivation  CT requirements could
be met at a lower ozone dose (4.7 mg/L [1.4 x 3.35] in the raw water vs. 2.8 mg/L [1.1 x 2.59] in the
settled water). The ozone dose dependency of OBP formation was previously discussed. After forma-
tion of OBPs in the raw water, OBP removal by coagulation and settling was minimal. It must be noted,
however, that at the bench scale, no biological activity took place in the sedimentation process. Miltner
and Summers (1992) demonstrated removal of AOC in a pilot-scale, bioacclimated sedimentation ba-
sin at room temperature.
Table 6-14. Effect of Ozone Staging on OBPs in OR Water (Miltner et al. 1998)
Parameter
TOC, mg/L
Total AOC, \ig Ceq/L
Formaldehyde, |J,g/L
Glyoxal, |J,g/L
Methyl glyoxal, |J,g/L
Glyoxylic acid, |J,g/L
Pyruvic acid, |J,g/L
Raw
3.35
142
3.0
ND
ND
ND
ND
Settled, then Ozonated
(1.1 mg/mg O3/TOC)
2.59
297
11.9
3.9
9.2
64.8
36.3
Ozonated (1.4
then Settled
2.58
440
31.4
7.2
15.2
189
119
mg/mg O3/TOC),







                                            6-14

-------
Controlling Bromate

Ozone reacts with bromide to form bromate, and bromate is regulated at 10 |ig/L under the D/DBP
Rule. While this reaction, or series of reactions, is complex, hypobromite ion is an intermediate prod-
uct. Thus, minimizing pH to favor hypobromous acid over hypobromite ion is cited as a best available
technology (BAT) for bromate control in the D/DBP Rule. Other means of control include adding
ammonia to form bromamines in place of the free-bromine (hypobromite and hypobromous acid) spe-
cies and applying ozone in a manner that minimizes the presence of the dissolved ozone residual driv-
ing the reaction.

Bromate formation as a function of increasing bromide concentration was studied by Shukairy et al.
(1994) in a pilot-scale ozone contactor treating OR water. Owens et al. (2000) studied inactivation of
spiked Cryptosporidiumparvum oocysts in the same pilot-scale ozone contactor treating different batches
of OR water. The pH was in the 7.40 to 7.65 range for the bromate study. The results in Table 6-15 show
bromate concentration increasing with increasing  dissolved ozone residual and with increasing bro-
mide. With ambient bromide (50.7 |ig/L), the  bromate MCL was exceeded near a transferred ozone/
TOC ratio of 1.1 mg/mg, at which approximately 1.3-log inactivation of C. parvum oocysts would
occur. At relatively high bromide concentrations, prohibitive bromate concentrations occured at low
ozone doses. This pilot-scale contactor was a single, countercurrent chamber. In a full-scale, multi-
chamber contactor, the same ozone might be applied over several chambers, minimizing the dissolved
ozone driving the bromate reaction, but maintaining the dissolved ozone required for achieving CT.


Table 6-15. Effect of Ozone Dose and Bromide on Bromate Formation in OR Water (Shukairy
           et al. 1994; Owens et al.  2000)
Trans O3
mg/L
0
0.89
1.37
1.93
3.02
4.32
Trans O3/DOC
mg/mg
0
0.53
0.81
1.11
1.78
2.54
Residual O3
mg/L
0
0.28
0.66
1.16
2.15
3.27
CT*
mg min/L
0
0.96
2.15
3.85
7.18
10.9
Log Inact
C. parvum
Oocysts
0
0.30
0.72
1.31
2.48
3.79
Bromate
Br
50.7 jlg/L
0.2
1.1
4.1
10.5
24.1
40.7
Concentration, Hg/L
Br
258 jlg/L
0.2
7.6
25.4
45.2
103
198
Br
SSOjlg/L
0.2
14.2
24.4
58.8
145
303
Summary

The formation of halogen-containing DBFs by chloramines is significantly lower than by free chlorine.
An exception is the formation of cyanogen chloride with chloramination. The formation of non-haloge-
nated DBFs like aldehydes and AOC is minimal with chloramination.

The formation of halogen-containing DBFs by chlorine dioxide is significantly lower than by free
chlorine. C1O2 oxidizes DBF precursors to the  extent that lower concentrations of DBFs are formed
with subsequent chlorination. C1O2 forms non-halogenated DBFs like aldehydes, ketones, and AOC.

Chlorite and chlorate can result from the use of C1O2. Chlorite can be controlled by GAC and by
reducing agents. Sulfite and metabisulfite can reduce chlorite, but may form chlorate. Thiosulfate can
reduce chlorite without forming chlorate. Ferrous ion can also reduce chlorate, but pH adjustment is
                                             6-15

-------
required to minimize chlorate formation. The use of a reducing agent like thiosulfate or ferrous ion can
complicate the application of post-disinfectants.

The formation of halogen-containing DBFs by ozone is significantly lower than by free chlorine. Ozone
can form bromo-DBPs like CHBr3, BAA, and DBAA, but at relatively low concentrations. Ozone
oxidizes DBF precursors such that lower concentrations of TTHM, HAA6, HAN4, and TOX are formed
with subsequent chlorination. However, ozone alters  the nature of the precursors to the extent that
higher concentrations of CH, CP, and Ill-TCP are formed with subsequent chlorination.

Ozone converts portions of the humic fraction to non-humic compounds and converts portions of the
higher-molecular-weight fraction to lower-molecular-weight compounds. Examples of lower-molecu-
lar-weight materials formed by ozone are aldehydes, keto acids, AOC, and BDOC. Concentrations of
these may be appreciable and necessitate control to ensure distribution system biostability. Generally,
much of the formation of these OBPs occurs at lower ozone doses. Ozone staging (when it is applied in
the treatment plant) can influence OBP formation. Ozonation of raw water results  in higher OBP for-
mation than ozonation of downstream waters in which some ozone demand has been removed.

Bromate can result from the use of ozone. Bromate concentration increases with increasing dissolved
ozone residual and with increasing  bromide.

References

Dryfuse, M. J., Miltner, R. J., and Summers, R.  S. (1995). "The removal of molecular size and
   humic/non-humic fractions of DBF precursors by optimized coagulation." Proceedings, Ameri-
   can Water Works Association, Annual Conference, Anaheim, C A, June.

Griese, M. H., Kaczur, J. J.,  and Gordon, G. (1992). "Combining methods for the reduction of
   oxychlorine residuals in drinking water." Journal of the American Water Works Association,
   84(11), 69

Griese, M. H., Hauser, K., Berkemeier, M., and Gordon, G. (1991). "Using reducing agents to elimi-
   nate chlorine dioxide and chlorite ion residuals in drinking water." Journal of the American
   Water Works Association, 83(5), 56.

Koechling, M. T.,  Shukairy, H. M., and Summers, R. S. (1996). "Chapter 13: Effect of ozonation and
   biotreatment on molecular size  and hydrophylic fractions of natural organic matter." Water
   disinfection andNOM: Characterization and control, Minear and Amy, eds., ACS Symposium
   Series 649, ACS, Washington, D.C.

Lykins, B. W., Goodrich, J. A., and Hoff, J. C. (1990). "Concerns with using  chlorine dioxide disin-
   fection in the USA." Journal of Water Supply Research and Technology-Aqua, 39(6), 376.

Lykins, B. W., Goodrich, J. A., Hoff, J. C., and Kothari, N. ( 1989). "Chlorine dioxide for drinking
   water treatment." Proceedings, Montana Section, American Water Works Association/WPCA
   Joint Conference, Great Falls, MT, March 29-3 1 .

Lykins, B. W. and Koffskey, W. (1986). "Products identified at an alternative disinfection pilot
        " Environmental Health Perspectives,  69, 119.
Lykins, B. W. and Griese, M. H. (1986). "Using chlorine dioxide for trihalomethane control." Jour-
   nal of the American Water Works Association, 78(6), 88.

Miltner, R. J., Oilier, L. L., Brown, T. A., and Summers, R. S. (1998). "Assessing the point of appli-
   cation of ozone during conventional treatment." Proceedings,  Water Research, American Water
   Works Association, Annual Conference , Dallas, TX, June 21-25.
                                            6-16

-------
Miltner, R. I, Summers, R. S., Dugan, N. R., Koechling, M., and Moll, D. M. (1996). "Comparative
    evaluation of biological filters." Proceedings, American Water Works Association, Water Quality
    Technology Conference., Boston, MA, November 17-20.

Miltner, R. J. (1993). "Transformation of NOM during water treatment." Proceedings, American
    Water Works Association Research Foundation Workshop on NOM, Chamonix, France,
    September.

Miltner, R. J., Shukairy, H. M., and Summers, R. S. (1992). "Disinfection by-product formation and
    control by ozonation and biological treatment." Journal of the American Water Works Associa-
    tion, 84(11), 53.

Miltner, R. J. and Summers, R. S. (1992). "A pilot-scale study of biological treatment." Proceedings,
    Water Quality, American Water Works Association, Annual Conference, Vancouver, BC,
    June 18-22.

Miltner, R. J., Rice, E. W., and Smith, B. L. (1991). "Ozone's effect on AOC, DBFs and DBF precur-
    sors." Proceedings, American Water Works Association,  Water Quality Technology Conference,
    Orlando, FL, November 10-14.

Miltner, R. J. (1990). "Treatment for the control of disinfection byproducts." Proceedings, Seminar
    on Current Research Activities in Support of USEPA's Regulatory Agenda, American Water
    Works Association, Annual Conference, Cincinnati, OH, June 17.

Miltner, R. J., Rice, E. W., and Stevens,  A. A. (1990). "Pilot-scale investigation of the formation and
    control of disinfect!on byproducts." Proceedings, American Water Works Association, Annual
    Conference, Cincinnati, OH, June 17-21.

Miltner, R. J. (1976). "The effect of chlorine dioxide on trihalomethanes in drinking water." M.S.
    thesis, University of Cincinnati, Cincinnati, OH.

Owens, J. H., et al. (2000). "Pilot-scale inactivation of Cryptosporidium and other microorganisms in
    natural water."  Accepted for publication in Ozone Science and Engineering, 22(5), 501.

Recommended standards for water works. (1992). Great Lakes Upper Mississippi River Board of
    State Public Health and Environmental Managers, Health Education Services, Albany, NY.

Richardson, S. D.,  et al. (1994). "Multispectral identification of chlorine dioxide disinfection
    byproducts in drinking water." Environmental Science and Technology, 28(4), 592.

Shukairy, H. M., Miltner, R. J., and Summers, R. J. (1994). "The effect of bromide on disinfection
    by-product formation, speciation and control: Part  1, ozonation." Journal of the American Water
    Works Association, 86(6), 72.

Shukairy, H. M., Summers, R. S., and Miltner, R. J. (1992). "Control and speciation of disinfection
    by-products by biological treatment." Proceedings,  Water Research, American Water Works
    Association, Annual Conference, Vancouver, BC, June 18-22.

Summers, R. S., Hooper, S. M., Shukairy, H. M., Solarik, G., and Owen, D. (1996). "Assessing DBF
    yield: Uniform formation conditions." Journal of the American Water Works Association,
    88(6), 80.

Stevens, A. A., Moore, L. A., and Miltner, R. J. (1989). "Formation and control of non-THM disin-
    fection by-products." Journal of the American Water Works Association, 81(8), 54.
                                              6-17

-------
6-18

-------
                                      CHAPTER 7

          Disinfection By-Product Control Through Biological Filtration1

Introduction

Disinfection by-product (DBF) control through biofiltration is defined as the removal of DBF precur-
sor material (PM) by bacteria attached to the filter media. The PM consists of dissolved organic matter
(DOM) and is utilized by the filter bacteria as a substrate for cell maintenance, growth, and replication.
The PM utilized by bacteria is no longer available to react with chlorine to form DBFs. All other things
being equal, a water  with lower PM concentration will yield lower DBF concentrations,  at a given
chlorine dose,  after a given time period. The biological filtration process is cost effective, since the
bacteria are naturally present in the water supply, can colonize existing filter media, do not produce a
residual that needs disposal, and require almost no modification of ambient conditions. The only pre-
requisite for maximizing bacterial substrate utilization in  filters is the absence of disinfectant in the
filter influent or backwash water. The filter media colonized by bacteria can  be sand, anthracite, or
granular activated carbon (GAC). Anthracite and sand are considered inert because neither interacts
chemically with PM. Thus, any removal of PM would be due solely to biological activity. GAC that has
been colonized by bacteria will initially remove DOM through adsorption and biological substrate
utilization. After the  GAC's  adsorptive capacity has been exhausted, PM removal is achieved only
through substrate utilization, and the GAC is defined as biological activated carbon (BAG). All drink-
ing water filters will become biologically active in the absence of applied disinfectant residuals. The
process of biological colonization and substrate utilization  is enhanced by ozonating filter influent
water.

In the  U.S., preozonation  is practiced to remove color, taste,  and odor, to inactivate Giardia and
Cryptosporidium, and to serve as an alternative to chlorine  disinfection. Ozonation decreases the aver-
age molecular size and weight of the PM, allowing indigenous bacteria to utilize more of it as substrate
in a given amount of contact time. Some fraction of microbes will always survive ozonation. As long as
no liquid phase ozone residual is present in the filter influent, the surviving microbes will eventually
colonize the filter media. A preozonated biological filter will achieve greater PM removals than one
with influent that has not been preozonated. PM is measured as total organic carbon (TOC), dissolved
organic carbon (DOC), trihalomethane (THM) formation potential (THMFP), or haloacetic acid (HAA)
formation  potential (HAAFP). All of the studies discussed in this chapter measure DBF control using
one or  some combination of the parameters TOC, DOC, THMFP, and HAAFP. Because of its potential
to control  DBF precursors and its economic advantages, the U.S. Environmental Protection Agency
(EPA)  Water Supply and Water Resources Division (WSWRD) has performed or funded a number of
research studies to characterize the impact of biological filtration on the control of DBFs.

EPA-Funded  Studies
Pilot-Scale Study, Shreveport, Louisiana, 1982

A pilot-scale study was executed to investigate the combination of ozone and GAC for THM precursor
removal (Glaze et al. 1982).  Following conventional treatment (coagulation,  flocculation,  and sedi-
mentation), the process water was split to a GAC filter, and to an ozone contactor followed by a GAC
'Nick Dugan: ORD/NRMRL/WSWRD, AWBERC Mailstop 689, 26 West Martin Luther King Dr.,
Cincinnati, OH 45268, 513-569-7239, dugan.nicholas@epa.gov.
                                              7-1

-------
Table 7-1. TOC and THMFP Removal in Pilot-Scale GAC Columns, Shreveport, LA, 1982
Week
0-21
22-52
53-62
65-83
%TOC
GAC
82
55
10
24
Removal
O3-GAC
83
55
19
27
% THMFP
GAC
83
56
19
24
Removal
O3-GAC
79
53
26
30
  NOTE: Average filter influent TOC = .30 mg/L.

filter (O3-GAC) . The results are summarized in Table 7-1. Analysis of the data showed microbial
activity to be a significant contributor to the removal process for TOC and THMFP over the long term.
THMFP was defined as THM formation after chlorination at 20 mg/L, at a pH of 6.5, a temperature of
26°C ± 2°C, and a 3-day incubation time.

The study was run for 83 weeks. At the beginning of the study (weeks 0-21), both sets of columns
achieved approximately 80 percent TOC and THMFP removal. These removals were due primarily to
physical adsorption of the PM. In the final phase of the project (weeks 65-83), TOC removals averaged
24 and 27 percent in the GAC and O3-GAC columns. During the same time period, THMFP removals
averaged 24 and 30 percent, respectively, in the GAC and O3-GAC columns. By this  time, the GAC
columns had each passed more than 50,000 bed volumes, and their adsorptive capacity was exhausted.
In the absence of adsorption, TOC and THMFP removals during weeks 65-83 were deemed to be due
to biological activity.

Bench-Scale Studies, 1991

A series of three bench-scale studies was performed to examine the impacts of ozone dose, water type,
and attached versus suspended bacteria on the biological removal of DOC, THMFP, and HAAFP. The
first study investigated the effect of ozone dose and biological treatment on PM control (Shukairy et al.
1992a). THMFP and HAAFP were measured by chlorinating at 12 mg/L, at  a pH of 7.5-8.0, and
holding for 7 days at 20 °C. Bench-scale biological treatment was performed in batch recycle tests (see
Figure 7-1). The water sample was circulated continuously through a bed of bioacclimated sand for 5
days.  Oxygen was provided by applying a vacuum to the head space of the sample chamber and
drawing in  air through  a water trap, which removed foreign bacteria. The 5-day contact time was
considerably longer than the 5 to 10 minutes typical of pilot- or full-scale biological filters. As a result,
estimates of biological removal in the batch test should be viewed as "ultimate" or "potential" remov-
als, which will be higher than flow-through removals at equivalent organic matter concentrations,
compositions, and ozone pretreatment.

Water samples were ozonated at transferred doses of 0.5, 0.8, 1.1, 1.8, and 2.5 mg O3/mg DOC. Bio-
logical treatment alone resulted in 13 to 14 percent removal of DOC. Ozonation followed by biological
treatment yielded 20 to 30 percent DOC removals. Biological treatment alone yielded  a 28 percent
reduction in THMFP. Ozonation followed by biological treatment yielded a 40 to 47 percent reduction
in THMFP. Ozonation followed by biotreatment yielded a 75 to  80 percent reduction in HAAFP. The
reduction in THMFP with  respect to ozone dose is shown  in Figure 7-2. THMFP formation after
biotreatment,  but without Ozonation, was approximately 220 |ig/L. At the lowest transferred ozone
dose of 0.8 mg O3/mg DOC, followed by biotreatment, THMFP dropped to 160 |ig/L. THMFP was not
reduced appreciably  at higher O3/DOC ratios of 1.1, 1.8, and 2.5. The removal behavior of HAAFP
through biotreatment as a function of applied ozone dose was similar to that observed for THMFP.
These results imply that THMFP reduction through biotreatment is sensitive to the presence or absence
of ozonation, but insensitive to the ozone dose after a minimum ozone dose (< 0.8mg/mg) is achieved.
                                            7-2

-------
                          Vacuum
                                  Sample
                                   Sand
                                                            Air
Figure 7-1. Bench-scale biofiltration apparatus.

It is possible that the greatest reduction in PM molecular weight, and corresponding increase in PM
bioavailability, occurs when ozonation is introduced, and that further increases in ozone dose do not
yield proportional decreases in PM molecular weight.

The second study examined DOC removal through ozonation and biofiltration as a function of water
type (Shukairy et al. 1992b). The first water was raw surface water from the Ohio River (ORW). The
second was an artificial water, produced using a solution of humic substances isolated and concentrated
from ground water. The humic substances were mixed with dechlorinated tap water to the desired DOC
concentration. Dechlorinated tap water was used to provide the non-carbonaceous nutrients and min-
eral matrix required for the growth of microorganisms. Humic compounds are one of the major catego-
ries of organic substances that make up DOC and which act as PM. The bench-scale biological  filter
was the one shown in Figure 7-1. DOC removals through biological treatment and ozonation followed
by biological treatment are summarized for both water types in Table 7-2. For both waters, the percent-
age DOC removal  approximately doubled when ozonation preceded biofiltration. The percentage of
DOC removals observed for ORW were similar to those observed during the previous study. However,
DOC removals reported for the artificial water are significantly higher than those  observed for any of
the previously described studies that used natural waters. It is possible that a water with an organic
fraction consisting only of humic substances would have a higher fraction of biodegradable compo-
nents relative to natural water.

The third bench-scale study assessed the impact on PM removal of attached versus suspended bacteria
(Miltner et al. 1992b). Three bench-scale bioreactors were filled with equal volumes of preozonated
(0.7 mg O3/mg DOC) ORW. The first reactor contained bio-acclimated sand. The  second reactor con-

Table 7-2. DOC Removals Through O3 and Biotreatment, ORW and Artificial Water, 1992
                     % DOC Removal
Water
Biofiltration    O-Biofiltration
ORW
12
25
Artificial water    34-48
             64-72

-------
                                           Ozonation + Biotreatment
                                           1             2
                                      Transferred Ozone/DOC (mg/mg)
Figure 7-2. Impact of ozone dose on THMFP removal.

tained only water and no sand. Reactor 3 was the same as reactor 2, except that the water contained
mercuric chloride to suppress any biological activity. After 5 days  of operation, DOC reductions in
reactors 1,2, and 3 were 23,10, and 0 percent, respectively. These results imply that bacteria suspended
in the water column can account for a significant fraction of PM reduction, but that bacteria attached to
some type of fixed surface are necessary to achieve the full potential of the process.

Pilot-Scale Study,  Cincinnati,  OH, 1991-1992

A year-long pilot-scale study (Miltner et al. 1992a; Miltner 1993) was carried out to assess: the impacts
of filter disinfection, filter media and filter biomass on biological PM control (Wang et al. 1995b),
biological PM control during a filter cycle as a function of backwash disinfection (Miltner et al. 1995),
and the performance of biofilters with respect to depth (Swertfeger et al. 1993). The plant schematic is
shown in Figure 7-3.  Raw water  was ozonated and then  subjected to conventional treatment (alum
                                                     Chlorine   Fj|ters
                        Ozone   Alum
                             i - Denotes Sample Point
                                                          1 Anth/sand 66  [-•-
Clear
Well
                                                          2 Anth/sand 6O
                                                                               • Chlorine
                                                          3 Anth/sand 6O
Clear
Well
Clear
Well
                                                          4 Anth/sand 1.5O
                                                          5 Sand 1.5O
                                                          6 F400 1.5O
                                                          7 HD4000 1.5O
                                                          8 PICABIOLlSCf*-
Figure 7-3. Biological filtration pilot plant, Cincinnati, OH, 1991-1992.
                                              7-4

-------
Table 7-3. Filter Configuration, Operation and Performance, 1991-1992,12-Month Pilot-Scale
          Study
Biomass
(nmol lipid PO4/
gram dry media)
Filter Media
1

2

3

5
6

7

8

20" Anth./
10" Sand4
20" Anth./
10" Sand
20" Anth./
10" Sand5
30" Sand
26" GAC1/
4" Sand
26" GAC2/
4" Sand
26" GAC3/
4" Sand
Chlorination
Pre + BW

BW

No

No
No

No

No

Mean
2.0

6.0

55

91
310

470

380

Std. Dev
0.50

0.60

1.7

1.3
9.0

9.0

11

TOC
Removal (%)6
Mean
8.0

16

20

20
29

27

21

Std. Dev
5.0

9.0

6.0

9.0
8.0

8.0

7.0

THMFP HAAFP
Removal (%) Removal (%)
Mean
1.0

13

21

23
40

34

27

Std. Dev Mean Std. Dev
11 9 8

6.0 28 7

8.0 37 4

7.0
5.0

5.0

3.0

1Filtrasorb400
2Hydrodarco4000
  3 Picabiol
  "Based on 12 months of data.
  5 Based on the last 10 months of data.
  6Filter influent TOC =1.1-2.2 mg/L.
coagulation, flocculation, and settling). After sedimentation, the water was distributed to eight filters,
the design, operation, and performance of which are summarized in Table 7-3.

Of special interest during the study were the impacts of prechlorination and media type on PM removal
in inert media filters. The filters used in the comparison were Filters 1, 2, 3, and 5. Filters 1 through 3
contained 20 inches of anthracite over 10 inches of sand, while Filter 5 contained 30 inches of sand.
Filter 1 received chlorine in the influent and backwash water,  Filter 2 received chlorine only in the
backwash water, and Filters 3 and 5 received no chlorine. Over the 1-year study period, all four filters
were examined for TOC, TFDVIFP, and HAAFP removals, which are summarized in Table 7-3.

TFDVIFP and HAAFP were determined by chlorinating at 12 to 15 mg/L and incubating the samples for
7 days at 25°C.  The non-chlorinated (NC) filters (Filters 3  and 5) removed equivalent or larger
fractions of influent TOC, THMFP, and HAAFP than did prechlorinated (PC) Filter 1 and backwash
chlorinated (BWC) Filter 2. There were no statistically significant differences (95% confidence level)
in PM removals between Filters 3 and 5, implying that the choice of inert filter media did not affect PM
removals. PM removals in Filter 1 were not significant, but Filter 2 removed measurable fractions of
TOC, THMFP, and HAAFP. In fact, TOC removals in Filter 2 were close to those observed in Filters 3
and 5. The results indicate that PC  and BWC combined (Filter 1) will suppress most biological PM
control, while BWC chlorination alone (Filter 2) allows most PM removal to proceed.

The application of chlorine in the filter influent or filter backwash can affect biological substrate utili-
zation by altering the nature and concentration of bacterial colonization. As a result, the pilot-scale
study investigated the correlations between biomass development and PM removal in the biologically
active pilot-scale filters (Wang et al. 1995b). Biomass was defined as the total  assemblage of microbial
                                              7-5

-------
cells that have colonized the filter media and was quantified using phospholipids (Findlay et al. 1989)
as a proxy. Phospholipids are common to all viable bacterial cell membranes, but are broken down
quickly in dead cells. Phospholipid concentrations during the pilot-scale study are summarized in Table
7-3. The biomass concentrations are the averages of triplicate samples collected from the tops of filters
after 3 months of plant operation. The three GAC filters represented three significantly different types
of commercially available activated carbon. None of the GAC filters were chlorinated. After this point
in time, biomass levels at the tops of the filters did not change significantly throughout the remainder of
the year-long pilot-plant run. The TOC and THMFP removal estimates are averages of samples col-
lected between 155 and 330 days of plant operation. Based on analysis of breakthrough curves, the
GAC filters were considered exhausted by this time, and as a result, any observed PM removal would
have been due to biological activity.

The application of disinfectant in the filter influent or filter backwash water significantly suppressed
biomass development relative to NC  inert  media  filters. The PC and BWC dual media filters had
biomass levels of 2.0 and 6.0 nmol PO4/gram dry media, respectively. In contrast, the NC sand and dual
media filters developed 55 to 91 nmol  PO4/gram dry media. All of the GAC filters accumulated more
biomass than did the inert media filters. Biomass levels in the three GAC filters ranged from 310 to 470
nmol PO4/gram dry media. Increased biomass development in  GAC filters was attributed to its porous
structure and to the presence of adsorbed organic material. The large number of pore spaces protected
biomass from fluid shear forces exerted during filter operation and backwashing. The adsorbed organic
materials potentially provide extra utilizable substrate that is not available to bacteria on inert media.
THMFP removals in the PC filter  were not significantly different from zero. The THMFP fraction
removed in the BWC filter was about an order of magnitude higher than the PC filter, despite similar
biomass concentrations. THMFP removals  in the  NC dual media and sand filters were  essentially
identical, at 21 to 23  percent. THMFP removals in the three GAC filters were the highest of all the
filters, ranging from 27 to 40 percent. With the exception of the PC filter, TOC removals did not appear
to be correlated to biomass development. There appeared to be higher correlations between biomass
development and THMFP removals, but differences in THMFP removals were not proportionate to
differences in biomass concentrations.

In a final portion of the 1-year pilot-scale study, BWC and NC dual media filters were sampled for the
control of PM over the course of several filter cycles (Miltner et al. 1995). A filter cycle was the interval
between startup following backwash  to shutdown prior to the next backwash event. Filters were
backwashed at 60 inches of headloss or every 48 hours, whichever came first. Both filters received NC
influent and were backwashed for 10 minutes at 50 percent bed expansion. The free chlorine residual in
the backwash water for the BWC filter  was about 1.0 mg/L. The filter effluents were sampled for TOC,
THMFP, and HAAFP immediately  prior to backwashing, as well as at selected times (1, 4, 12, 24, 48,
and 72 hours) during the next filter cycle. Biomass concentrations in the filter media were also evalu-
ated. Statistically significant declines  in biomass levels of approximately 25 percent were observed
after backwashing in the BWC filter. The BWC filter biomass concentrations then increased steadily to
prebackwash levels by the end  of the  filter  cycle.  No significant changes in biomass concentrations
were observed in the NC filter with respect to backwashing. Backwashing with NC or BWC water had
no detrimental effects on the biological control of TOC, THMFP, or HAAFP at any time during the
three filter cycles in  which backwash effects were examined. Figure  7-4 illustrates the impact of
backwashing on HAAFP control during a filter cycle. The effluent HAAFP concentrations remained
constant over the filter cycle, despite the increase in filter influent HAAFP levels. These results imply
that HAAFP removal, as a fraction  of the influent concentration, actually improved over the course of
the filter cycle.
                                             7-6

-------
             70
             60
                    Settled
                          Filter 2 (BWC)
Filter 3 (NC)
             20
             10
                                \
                               20
                                   I            \
                                  40           60
                                    Run Time - h
              80
         I
        100
Figure 7-4. HAAFP control during a filter cycle, pilot-scale, Cincinnati, OH, 1991-1992.

At the conclusion of the 1-year pilot-scale study, Filter 5 (30" sand, no PC or BWC) was cored along its
full depth to examine the impact of filter depth on biological removal of DOC and THMFP (Swertfeger
et al. 1993). The core of biologically acclimated media was divided along its length into four sections,
each representing approximately equal increments of contact time. The segmented filters were then run
in series, using ORW that had been treated with potassium permanganate addition, followed by con-
ventional treatment and biological filtration. The pretreated water was then ozonated at the bench-scale
prior to passage through the segmented filters. The experimental setup is shown in Figure 7-5.  The
segmented filter was operated for 3 days, and sampling for PM removal was performed on each  day.
Total DOC removals across the segmented filters averaged 13 percent, with approximately 50 percent
                      03
        T
                        Overflow
                                  EBCT=
                                  (min)
        5.3
9.1
12.9
- Denotes Sample Point
Figure 7-5. Impact of contact time on biofiltration performance, segmented filter.
                                              7-7

-------
        T
                                          EBCT=
                                          (min)
                                                             C2
                                                   3.7
                                                             C3
3.7
         C4
3.7
- Denotes Sample Point
Figure 7-6. Impact of contact time on biofiltration performance, parallel operation.
of the removal observed in the first 1.6 minutes of empty bed contact time (EBCT). THMFP removals
across the segmented filters also averaged 13 percent. All of the observed THMFP removal occurred by
5.3 minutes of contact time.

After 3 days of sampling, the segmented filters were switched to parallel operation. This arrangement
allowed water to directly enter each column segment without contact with previous filter sections. This
setup yielded information about the capacity of the biomass in each section to remove compounds at
concentrations not previously encountered by that filter section. The new arrangement was operated for
2 days and is detailed in Figure 7-6. DOC removals in segments 1 and 2 were each less than 10 percent.
No DOC removals were observed in segments 3 and 4, which comprised the lower half of the original
filter. TFDVIFP removals in segments 1 and 2 were each about 12 percent of the influent values. No
TFDVIFP removals were observed in columns 3 and 4. The results from the series  phase of the study
confirmed that the bulk of PM  removal tends to occur in the top portion of a biological filter. This
implies that the chemical composition of the PM reaching the lower portion of the biological filter may
be fundamentally changed and  that biomass in the lower level of the filter develops to utilize  this
remaining fraction. The results  from the parallel phase of the study indicated that the biomass in the
lower half of the filter might need a certain amount of acclimation time before it can effectively utilize
a new type of substrate.

Five-Month Pilot-Scale Study, 1996

The  1991 to 1992 year-long EPA  pilot-scale  study did not evaluate the impact of preozonation or
disinfectant type in an integrated pilot-scale environment. To fill in some of these gaps, a 5-month long
pilot-scale biofiltration study was initiated  at EPA in 1996 (Miltner et al. 1996). This pilot plant  was
supplied with water from a local lake, the raw DOC of 5 . 8 mg/L of which was significantly higher than
that of the ORW used during 1991 to 1992 (1-2 mg/L). Filter design and performance are summarized
in Table 7-4. All plant influent water received conventional pretreatment, consisting of alum coagula-
tion, followed by flocculation and sedimentation. Following sedimentation, some of the water  was
ozonated prior to filtration (Filters 1-5), while the remainder went directly to filtration (Filters 7 and 8).
The average transferred ozone dose was 5.6 mg/L. None  of the filters were PC. Filters  1 and 8 were

-------
Table 7-4. Pilot-Scale Biological Filters,3 Cincinnati, OH, 1996
Percentage Removals


Backwash
Filter Preozonated Disinfectant
1
2
o
3
4
5
7
8
Yes
Yes
Yes
Yes
Yes
No
No
Chlorine
Chloramine
None
None
None
None
Chlorine
Biomass
(nmol lipid
PO4/g DOCC THMFP HAAFP
Filter Media
20" Anthracite/10" Sand
20" Anthracite/10" Sand
20" Anthracite/10" Sand
20" GACb /10" Sand
30" Sand
20" Anthracite/10" Sand
20" Anthracite/10" Sand
dry media)
28
120
130
288
96
66
16
12
19
23
42
23
8
0
8
17
23
53
21
20
24
26
33
39
62
39
27
23
  a Loading rate for all filters = 2 gal/min per ft2 (5 m/hr).
  bFiltrasorb400
  c Average filter influent DOC = 2.8 mg/L.
backwashed with chlorinated water (1.6 mg/L), Filter 2 was backwashed with chloraminated water (2.1
mg/L), and the remaining filters received no disinfectant in the backwash. Once the plant reached
steady-state operation, biomass development, and removals of DOC, THMFP, and HAAFP across the
filters were measured.

Formation potentials were measured under uniform formation conditions (UFC). These consisted of
chlorinating to yield 1 mg/L free chlorine residual after 24 hours of holding time at pH 8. Preozonation
impacted biomass development and DBPFP removal. Preozonated filters typically developed higher
biomass levels and removed larger fractions of DOC. However, the impact on THMFP and HAAFP
removals were mixed. The preozonated NC inert media filters (Filters 3 and 5) developed higher levels
of biomass and removed more PM, as measured by DOC and HAAFP, than their non-ozonated coun-
terpart Filter  7.  However, all three filters removed roughly equivalent amounts of THMFP. The
preozonated, chlorinated inert media Filter 1  developed more biomass and  removed more DOC than
did the non-ozonated, chlorinated, inert media filter (Filter 8). However, both filters removed equiva-
lent fractions of HAAFP, and Filter 8 removed a larger fraction of THMFP. The presence or absence of
BWC impacted biomass development, DOC, THMFP, and HAAFP removals in preozonated filters.
NC, preozonated Filters 3 and 5 developed more biomass and removed more DOC, THMFP, and HAAFP
than did BWC, preozonated Filter 1. BWC had less of an impact on non-ozonated filters. Non-ozonated,
NC Filter 7 developed more biomass and removed  more  DOC than  did  non-ozonated,
chlorinated Filter 8. However, both filters removed equivalent amounts of THMFP and HAAFP.

The type of backwash disinfectant had an impact on biofiltration performance. Preozonated,
chloraminated Filter 2 developed more biomass and removed more DOC, THMFP, and HAAFP than
did preozonated, chlorinated Filter 1. Filter 2 developed as much biomass and removed as much DOC
and HAAFP as preozonated, NC Filter 3. As in the 1991-1992 pilot-scale study, the use of anthracite/
sand versus sand only as a filter  material  had no impact on biological filtration performance.
Preozonated,  NC Filters  3 and 5 developed equivalent amounts  of biomass and removed almost
identical  fractions of DOC, THMFP, and HAAFP. However, the use of exhausted GAC significantly
improved biological filtration performance. Preozonated, NC Filter 4 (GAC) developed more biom-
ass and removed roughly twice as much DOC, THMFP, and HAAFP than  did Filters 3 and 5. All of
the preozonated filters in the current study removed larger DOC fractions than did equivalent filters
during the 1991 to 1992 study. This was most likely due to the higher filter influent DOCs.
                                            7-9

-------
Table 7-5. Precursor Control as a Function of Temperature, Cincinnati, OH, 1999

                                 Percent Removal
                                 (mean, std. deviation)
Parameter
DOC
UFC HAA

15
14
5°C



3
o
J

24
31
20°C



o
J
3

24
34
35°C



2
2
Impact of Temperature

Three biological filters were acclimated at 5,20, and 35°C during a 73-day study in order to investigate
biomass development and PM control as a function of temperature (Moll et al. 1999). For the first 44
days of the study, the filters were acclimated with ozonated ORW. During days 45 through 73, the
filters were acclimated with an ozonated, isolated  solution of NOM. The NOM was isolated by
nanofiltration from a Florida ground water low in particulates and high in DOC. Prior to ozonation, the
isolated NOM solution was diluted to a DOC concentration of 4.2 mg/L with dechlorinated tap water.
The temperature in the 5 and 35°C filters was regulated using water-jacketed columns. The cooling or
heating fluid was recirculated from the columns through constant temperature water baths. The tem-
perature in the 20°C filter was maintained by the ambient internal building temperature. The tops of the
filter media were sampled for biomass development on a regular basis. The filters were defined to have
reached a biological steady-state when consecutive biomass samples did not vary by more than 20
percent. During the last 2 weeks of the study, the filters were sampled intensively for DOC and HAAFP
control. Formation potentials were measured under uniform formation conditions. At the conclusion of
the study, the filters were sacrificed and sampled for biomass development with respect to depth. Bio-
mass profiles are summarized  in Figure  7-7,  and PM control is summarized in  Table 7-5. All filters
were operated at 3.6 m/hr loading rates. Top of filter biomass development was more extensive in the
               250  -
Figure 7-7. Biomass development as a function of filter temperature. Cincinnati, OH, 1999.
                                            7-10

-------
20 and 35°C filters than in the 5°C unit. However, middle of filter biomass levels were higher at 5°C
than at the two higher temperatures. DOC removal at 5°C was lower than at the two higher tempera-
tures. There were no significant differences in DOC removals between 20 and 35°C. HAAFP removals
at 5°C were about half the level reported for 20 and 35°C. HAAFP removals did not vary significantly
between 20 and 35°C.

Full-Scale Evaluation of Temperature Effects

A study was performed to assess the impact of temperature on biomass growth and PM control in eight
full-scale drinking water filters (Fonseca et al. 1999). Filters in these plants were sampled in the sum-
mer and winter to capture the extremes in water temperature. All of the filters in these plants received
preozonated influent. Filters in three of the plants contained exhausted GAC. Details on filter construc-
tion and performance are provided in Tables 7-6 and 7-7, respectively. The average winter and summer
temperatures were 6.1 and 24°C, respectively. Top of filter biomass levels remained the same  or de-
creased in all of the filters from winter to summer. On average, DOC removals decreased, while UFC
Table 7-6. Filter Construction, Full-Scale Temperature Study, 1999
Filter Loading Rate
(m/hr)
Plant
Celina, OH
Berea, OH
Lake Bluff, IL
Somerset, NJ
Millwood, NY
Andover, MA
Santa Clarita, CA
Sylmar, CA
GAC
(in.)
-
-
48
43
-
43
-
-
Anthracite
(in.)
-
-
-
-
24
-
72
72
Sand
(in.)
18
24
12
9
12
6
-
-
Garnet Chlorinated
(in.) Backwash
Yes
Yes
Yes
4.5 Yes
No
Yes
No
No
Winter
4.6
3.9
4.6
3.7
2.9
5.4
9.8
11
Summer
3.9
-
7.3
4.9
4.4
4.6
15
24
Table 7-7. Full-Scale Filter Performance and Biomass Development as a Function of
          Temperature, 1999
Winter
Summer
Percent Removal
Plant
Celina
Berea
Lake Bluff
Somerset
Millwood
Andover
S. Clarita
Sylmar
Temp
(C)
3
3
2
7
5
7
13
9
Biomass
(nmol PO4)
115
7
127
127
31
106
47
50
DOC
4
3
48
20
16
25
15
22
THMFP
9
10
8
28
-
49
8
24
HAAFP
14
9
30
47
3
81
41
46
Temp
(C)
28
-
21
28
22
26
17
24
Biomass
(nmol PO4)
70
-
78
71
31
47
40
38
Percent Removal
DOC
12
-
30
11
2
28
10
16
THMFP
-
-
32
37
26
40
5
12
HAAFP
17
-
39
55
23
67
19
41
                                            7-11

-------
THMFP and HAAFP removals did not increase significantly from winter to summer. The authors
attributed these phenomena to increases in filter loading rates. On average, filter loading rates were 61
percent higher in the summer than in the winter. The corresponding decreases in EBCT may have
mitigated the impacts of temperature increases on the rates of biological activity. As in the bench-scale
temperature study, however, the data indicated that significant biological utilization of PM occurs even
during the winter time.

Modeling Biological PM Control

A method was developed to predict DOC removal with respect to EBCT in biological filters (Wang and
Summers 1995,1996). The model was based on a steady-state mass balance around a plug-flow reactor
(PFR):
                           ac        ac,     3
                        V°^T~~  ^L^T + 7(1" £)^(Cl" Cir) = °


The symbol z is depth within the filter (m), VQ is the filter loading rate (m/hr), C. is the bulk DOC
concentration (mg/L), Cir is the DOC concentration at the surface of the filter media (mg/L) where the
biomass is attached, r is the radius of the filter media (m), e is the filter porosity (dimensionless), EL is
the substrate diffusivity in the bulk solution (m2/s) , and kL is the overall mass transfer coefficient (m/s)
that describes movement of the DOC from the bulk solution to the biomass colonies on the filter media
surface. The first term in the above equation describes convection of DOC in the bulk solution. The
second term describes diffusion of the DOC in the bulk solution. The third term describes transfer of
DOC from the bulk solution to the biomass.

The hydrodynamic conditions in the filter were such that bulk diffusion could be neglected. As a result,
the second term in the equation was dropped. DOC was divided into  biodegradable (BDOC) and
nonbiodegradable (NBDOC) fractions. The BDOC fraction, in turn, was divided into fast (FBDOC)
and slowly (SBDOC) biodegradable fractions. The BDOC fraction was defined as the fraction of BDOC
that could be biologically utilized in 5  days of contact with acclimated media. The FBDOC fraction
was defined as that fraction of substrate that could be utilized in three minutes of EBCT with accli-
mated media. The SBDOC fraction was the difference between BDOC and FBDOC fractions. FBDOC
and SBDOC utilization rates were described mathematically using Monod and first-order kinetics:
                                       dC,    ____
                                      ~W = K« + Ci                           (7-2)
                                         dC,                                     (?_3)
                                         dt   '
Vmax (mg/mg cells-s) and Ksl (mg/L) are Monod kinetic coefficients, and K2 (L/mg cells-s) is the first-
order kinetic coefficient. Xis the concentration of biomass on the surface of the filter media (mg cells/
L). The variation of biomass with respect to filter depth was expressed mathematically as:

                                      X=A(l+BeFz)                             (7-4)

A, B, and F are regression coefficients, based on fitting the equation to an observed biomass profile.
The two kinetic equations and the biomass profile equation were then combined with the mass balance
                                             7-12

-------
         D)
g

1
"c
o
o
O
O
O
Q
        ~  3.0
            2.5
            2.0
                         Observed
                         Predicted
              0.0
                 0.2
0.4
    0.6

Filter Depth (m)
0.8
1.0
1.2
Figure 7-8. Observed and predicted DOC utilization, 5 m/hr (2.0 gal/min»ft2) loading rate.

to yield a first-order, linear, homogenous differential equation that could be solved numerically to
predict bulk DOC concentration at any depth in the filter. The DOC utilization kinetic coefficients were
assumed to be constant with respect to filter depth, and they were determined using a method originally
developed by Rittman and McCarty (1980a, b). Acclimated filter media were placed in small biological
filters and exposed to influent water with varying concentrations of DOC. Utilization rates of DOC
were observed by measuring DOC removals at varying influent DOC concentrations. The observed
utilization rates in turn were used to estimate the kinetic parameters V  ,K., and K,. The model was
                                                    r            max'   s\'      2
used to predict DOC utilization with respect to filter depth in sand-only biological filters that had been
acclimated using a solution of ozonated ground water humic substances. A segmented filter arrangement
was used, which permitted the measurement of biomass and DOC concentrations at various filter depths.

The observed DOC concentrations and modeling results are shown in Figure 7-8. The individual points
are the observed concentrations, and the solid line is the  model prediction.  The model was able to
closely predict the biological utilization of DOC with respect to depth in the filter.

The applicability of the model would be enhanced if it could accurately predict substrate utilization at
the pilot-scale and in the presence or absence of ozonation and backwash disinfection. In order to
address these issues, the model was evaluated during the 5-month pilot-scale  study (Dugan and Sum-
mers 1997). Kinetic parameters were estimated for media acclimated with ozonated and non-ozonated
water. All of the filters listed in Table 7-4 were sacrificed at the end of the study in order to determine
the biomass development with respect to filter depth. The model predictions and deviations from ob-
served values are summarized in Table 7-8. The model predictions of effluent DOC were compared
with the observed average effluent DOCs during the last 4 weeks of the pilot plant study, when DOC
removals and biomass levels were at an approximate steady state. With one exception, model predic-
tions fell within ten percent of observed effluent DOCs.
                                              7-13

-------
Table 7-8. Summary of Model Predictions, Pilot-Scale, Cincinnati, OH, 1996
Filter
1
2
3
5
7
8
Observed Effluent
DOC (mg/L)
2.3
2.1
2.0
2.1
2.5
2.7
95% Confidence
Interval (mg/L)
0.7
0.4
0.7
0.7
0.9
0.7
Predicted Effluent
DOC (mg/L)
2.1
1.9
1.8
1.8
2.7
2.7
Absolute Deviation
(mg/L)
-0.2
-0.2
-0.2
-0.3
+0.1
0.0
Relative Deviation
(%)
-8.7
-9.5
-10
-14
+6.0
0.0
Discussion

The data collected so far indicate that biologically active filters remove significant amounts of PM and
that preozonated biofilters remove more PM than do non-ozonated filters. The resulting reductions in
THMFP and HAAFP could help many drinking water utilities meet the 80 (THM) and 60 (HAA) |ig/L
limits mandated under the Stage 2 Disinfection/Disinfection By-Products Rule. The data also raise
questions that indicate potential directions for future research.  One area is the relationship  between
biomass levels and PM removals. In general, PM removals increased with increasing biomass levels.
However, the relationship was not proportional. Biomass concentrations in PC or BWC filters were
often close to an order of magnitude lower than for NC filters. However, PM removals in the PC and
BWC filters, for some categories, decreased by less than that amount. Were the bacteria in disinfected
filters able to consume more substrate  due to decreased competition, or did surviving disinfectant
exposure confer some intrinsic metabolic advantage?

Another direction for research would be the performance of biofilters under transient conditions. Only
one paper (Miltner et al. 1995) investigated PM control over the course of a single filter cycle.  It would
be worthwhile to examine biofiltration performance after extended periods of shutdown, rapid changes
in loading rate, rapid  changes in influent temperature or pH, and as a function of different filter
backwashing methods.

References
Dugan, N. R. and Summers, R.  S. (1997). "A biomass based model to predict substrate utilization in
   field-scale drinking water biofilters." Proceedings, American Water Works Association Annual
   Conference, Atlanta, GA, June.
Findlay, R. H., King, G.  M., and Watling, L. (1989). "Efficacy of phospholipid analysis in determining
   microbial biomass in sediments." Applied and Environmental Microbiology, 55(11), 2888-2893.
Fonseca, A. C. et al. (1999). "Impact of temperature on biofilter performance and microbial commu-
   nity structure." Proceedings, American Water Works Association Annual Conference, Chicago,
   IL, June.
Glaze, W. H. et al. (1982). "Pilot-scale evaluation of biological activated carbon for the removal of
   THM precursors." EPA-600/S2-82-046, Environmental Protection Agency.
Miltner, R. J. and Summers, R.  S. (1992a). "A pilot-scale study  of biological treatment." Proceed-
   ings, American Water Works Association Annual Conference, Vancouver, British Columbia,
   Canada, June.
Miltner, R. J., Shukairy, H. M.,  and Summers, R. S. ( 1992b). "Disinfection by-product formation
   and  control by ozonation and biotreatment." Journal of the American Water Works Association,
   84(11), 53-62.
                                             7-14

-------
Miltner, R. J. (1993). "Transformations of NOM during water treatment: Oxidation, formation of
   disinfection by-products, biodegradation." Proceedings, American Water Works Association
   Research Foundation NOM Workshop, Chamonix, France.

Miltner, R. J., Summers, R. S., and Wang, J. Z. (1995). "Biofiltration performance: Part 2, effects of
   backwashing." Journal of the American Water Works Association, 87(12), 64-70.

Miltner, R. J. et al. (1996). "A comparative evaluation of biological filters." Proceedings, American
   Water Works Association Water Quality Technology Conference, Boston, MA, November.

Moll, D. M., Wang, J. Z., and Summers, R. S. (1995). "NOM removal by distinct microbial popula-
   tions in biofiltration processes." Proceedings, American Water Works Association Annual Con-
   ference, Anaheim, CA,  June.

Moll, D. M.  and Summers,  R. S. (1996). "Performance, biomass and community structure profiles of
   biological rapid media filters." Advances in slow sand and alternative biological filtration. John
   Wiley and Sons, Chichester, England.

Moll, D. M.  et al. (1999). "Impact of temperature on drinking water biofilm performance and micro-
   bial community structure." Environmental and Science Technology, 33, 2377-2382.

Rittman, B. E. and McCarty, P. L.  (1980a). "Model of steady-state biofilm kinetics." Biotechnology
   andBioengineering, 22, 2343-2357.

Rittman, B. E. and McCarty, P. L.  ( 1980b). "Evaluation of steady-state biofilm kinetics." Biotech-
   nology and Bioengineering, 22, 2359-2373.

Shukairy, H. M., Miltner, R. J., and Summers, R.  S. (1992a). "Control of disinfection by-products
   and biodegradable organic matter through biological treatment." Revue Des Sciences De L 'Eau,
   February, 1992, 1-15.

Shukairy, H. M. and Summers, R. S. (1992). "The impact of preozonation and biodegradation on
   disinfection by-product formation." Water Research, 26(9), 1217-1227.

Shukairy, H. M., Summers, R. S.,  and Miltner, R. J. (1992b). "The impact of ozonation and biologi-
   cal treatment on disinfection by-products." Proceedings, 4th Drinking Water Workshop, Govern-
   ment of Quebec, Montreal, Canada, November.

Shukairy, H. M., Summers, R. S.,  and Miltner, R. J. (1992c). "Control and speciation of disinfection
   by-products by biological treatment." Proceedings, American Water Works Association Annual
   Conference, Vancouver, British Columbia, Canada, June.

Swertfeger, J. et al. (1993).  "The control of ozonation by-products by biological filtration." Proceed-
   ings, American Water Works Association Annual Conference, San Antonio, TX, June.

Wang,  J. Z. and Summers, R. S. (1995). "Biomass growth and distribution in drinking water
   biofilters and its impact on NOM removal." Proceedings, American Water Works Association
   Annual Conference, Anaheim, CA, June.

Wang,  J. Z.,  Summers, R. S., and Miltner, R.  J. (1995). "Biofiltration performance: Part 1, relation-
   ship to biomass." Journal of the American Water Works Association, 87(12), 55-63.

Wang,  J. Z. and Summers, R. S. (1996). "Biodegradation behavior of ozonated NOM in sand filters."
   Revue Des Sciences de L 'Eau.
                                             7-15

-------
7-16

-------
                                     CHAPTER 8

                 Microbiological Removal by Filtration Processes1
Introduction
Filtration was originally used to remove contaminants that affect the appearance, odor, and taste of
drinking water (i.e., to improve drinking water aesthetics). Records show that James Simpson de-
signed and constructed a slow sand filter used by the Chelsea Water Works Company in London in
the 1820s.  This filtration system was used on  the polluted and turbid water of the River Thames.
Aesthetic considerations were the primary reason for construction of this filtration facility. It was not
until 50 years later that Robert Koch demonstrated that bacteria in drinking water were causative
agents of disease. Koch,  a health official, examined intestinal disease morbidity and mortality in
relation to use/non-use of filtration in Germany. In Altoona, Germany, unfiltered water was delivered
to residents; the rates of intestinal illnesses increased dramatically. As the causes of typhoid fever,
cholera, and other diseases were better understood through improved microbiology techniques, ma-
jor emphasis was placed on the operation of filtration plants to remove bacteria as well as to provide
water that was pleasant to the senses  (Logsdon and Lippy  1982).

Water treatment technology continued to improve with the addition of disinfection, coagulation, and
sedimentation to the treatment process and with further refinement  of the filtration process. Public
health agencies and water utilities also recognized that water treatment should be accompanied by
other measures that will protect the quality of water, such as protecting watersheds from contamina-
tion, treating upstream wastewater discharges adequately,  certifying operators, and adopting source
water quality standards. The multiple barrier concept evolved from the incorporation of protective
measures into water treatment technology. This concept is based on the idea that, in the event one
barrier fails, the remaining barriers will reduce the impact of the failure.

Even with the advances made in water treatment technology, outbreaks of waterborne disease con-
tinue to occur. In fact, the occurrence of reported outbreaks is increasing, as are the number of cases
of illness associated with those outbreaks. Outbreaks continue to occur because the multiple barrier
concept is  not properly applied, and in many  cases, filtration is not properly provided or applied
(Craun 1991). Recent outbreaks demonstrate the  importance of these principles and indicate that
current technology is not being properly applied to prevent waterborne disease (Fox and Lytle 1996).

Recent large waterborne  disease outbreaks have focused much of the nation's attention on large
water systems. When 403,000 people in Milwaukee fall ill, it is headline news, nationally (MacKenzie
et al. 1994). Although the numbers  of people affected by waterborne disease outbreaks in small
systems are smaller than those  affected by waterborne disease outbreaks in large systems, many
individuals are impacted (Centers for Disease Control and Prevention 2000). The sheer numbers of
small systems in the U.S. overwhelm the number of larger systems. In the U.S., there are 186,822
public water systems,  and 178,911 (greater than 95%) of those systems serve 3,300 individuals or
fewer (Office of Water 1995).

The problems of treating and distributing safe water to prevent waterborne disease in small systems
may parallel those of larger systems.  These problems are often exacerbated in small systems by the
lack of the economies of scale that larger systems enjoy. Smaller systems tend to deal with source
!Kim Fox: ORD/NRMRL/WSWRD, AWBERC Mailstop 689, 26 West Martin Luther King Dr.:
Cincinnati, OH 45268, 513-569-7820, fox.kim@epa.gov.
                                              8-1

-------
water of worse quality (there may be only one water source available), some deal with antiquated water
systems, and many or most may not be financially viable. Based on 1994 compliance data, 1204 public
water systems were in significant noncompliance with microbiological and turbidity requirements (even
excluding monitoring and reporting violations). Of that number,  1145 of the significantly noncompliant
systems were classified as small systems (Office of Water 1995).

Noncompliance issues place communities in situations in which unsafe water is distributed to custom-
ers. Long-term boil water orders are known to exist in several small communities. One small commu-
nity in the mid-western U.S. was on a boil water order for over one year and remained subject to that
order until new treatment facilities were built. Other small systems are known to issue boil orders
whenever it rains; the unfiltered source water turbidity rises (above 5.0 nephelometric turbidity units,
or NTUs) to the point at which a boil order is needed to ensure  safe drinking water (Fox 1996).

In the years 1993 and 1994, there were 30 outbreaks associated with water intended for drinking. Of the
30 recorded outbreaks, 21 were in  small water supply systems (seven were in private homes, and two
were in large systems) (Centers for Disease Control and Prevention 1996). The large system outbreaks
made the national news, but the outbreaks in the smaller systems were just as important to those indi-
viduals living in the smaller communities. The small system outbreaks consisted of both inorganic
contamination (e.g., lead, nitrate,  and fluoride) and microbial  contamination (e.g., Cryptosporidium
parvum, Giardia lamblia, Shigella sonnet, etc.) The small system outbreaks were relatively evenly
spread across the U.S. (Centers for Disease Control and Prevention 1996).

The U.S.  Environmental Protection Agency  (EPA) Water Supply  and Water Resources Division
(WSWRD) staff have been on-site during several waterborne disease investigations and have investi-
gated outbreaks in small systems that have occurred in untreated ground waters  and fully treated sur-
face waters. At one outbreak, the untreated distribution water became contaminated with surface runoff
when pipe breaks were being repaired, and in another, a storage tank was contaminated with bird feces.
In both of these cases, a number of individuals became ill and several people died. At one small system's
treatment plant, improper coagulant dosages caused poor particulate removal and allowed pathogens to
pass through the plant, causing illness (Swerdlow et al. 1992) (Clark et al. 1996). No one criterion has
been responsible for all waterborne disease outbreaks.

Reacting to changes in source water quality can be difficult in the context of the relatively short operat-
ing times characteristic of many small water systems. In one small system in West Virginia, during one
rainfall event, for example, the source water turbidity increased from a background of around 20 NTUs
to more than 2000 NTUs in a matter of 4 hours.  The rapid increase in turbidity required constant
changing of coagulant dosages. In this case, the plant's effluent turbidity exceeded 50 NTUs, and the
customers were asked to boil the water (or not consume it) until the system was cleaned. The water's
residence time in the treatment plant was less than 2 hours, and the operator could not keep up with the
changing source water conditions.  This is an extreme example, but less extreme situations occur often.

In several of the small system outbreaks, flyers notifying consumers to boil their water were hand
delivered  to each door (Swerdlow et al. 1992; Clark et al. 1996). In these cases, the small area of
concern allowed for quick notification.

What can be done to prevent waterborne disease outbreaks in small systems? Small systems (like their
large system counterparts) must aggressively treat their drinking water. This aggressive treatment may
mean maintaining proper disinfection (both concentration and contact time), or  it may mean control-
ling coagulant dosages and filtration rates at all times.  The integrity of the distribution systems and
storage tanks must also be maintained to ensure that properly treated water remains safe prior to deliv-
ery to the consumer. One of the ways to help encourage the need for aggressive treatment is to ensure
continuing education of the system operators and the local water boards or councils.
                                              8-2

-------
The potential for an outbreak exists with every water utility, whether that utility is large or small. The
likelihood of an outbreak occurring at any one utility depends on the integrity of the water treatment
(both the quality and the methodology) and distribution system at the utility and the utility's dedication
to providing safe water.

This chapter  discusses water treatment  technologies that are especially applicable to small systems,
such as Slow Sand Filtration (SSF) and  Diatomaceous Earth (DE) Filtration. In addition, the applica-
tion of granular filtration for Giardia and Cryptosporidium control is discussed. Granular filtration has
applications in systems of all sizes.

Slow  Sand Filtration

SSF is a filtration process in which water (typically untreated) is slowly passed through a sand filter.
The sand filter is constructed so that the  system remains aerobic, has fine sand grains, and  stays wet at
all times. The water passes through the sand, and a biomass (schmutzdecke) is formed both on top and
in the top layers of the sand. This schmutzdecke is composed of clays and biological matter. It helps
remove both clays and biological material from the water. Physical attachment, screening,  and deposi-
tion of matter also occur within the sand  filter as the filter run proceeds. SSF effluent is then disinfected
prior to sending it to the consumers. As the schmutzdecke builds up, head loss within the filter in-
creases; when it reaches a predetermined point, the system is taken offline and the top layer of sand is
removed. Water is once again  sent through  the sand, and once the filter has re-ripened,  the effluent
becomes acceptable (Fox et al. 1984).

Slow sand filters are used extensively in Europe and were widely used in the U.S. prior to  1910. In the
years following 1910, rapid sand filtration became  the system of choice in the U.S. to treat surface
waters. In the early 1980s, WSWRD began  to reconsider SSF as a viable option for treating surface
waters in small communities.  The amount of land required for SSF would likely prohibit its use by
large communities. The land issue is not usually considered a problem for small communities. Opera-
tion of a slow sand filter is not difficult, and construction of a slow sand filter requires few components.
Thus, WSWRD decided to re-evaluate SSF and to determine its capabilities for removing microorgan-
isms from drinking water (Fox et al. 1984).

One of the first projects that WSWRD undertook was in-house testing of a slow sand filter (Fox et al.
1984).  Two pilot-scale slow sand filters were constructed and operated in-house. Both filters had a
surface area of 0.21 m2 and a depth of media of 0.76 m. The first filter (Filter A) contained sand as its
media, while the  second  filter (Filter B) had a vertical  divider and contained sand on one side  and
granular activated carbon (GAC) on the  other. The filters were challenged with various surface waters
from the Cincinnati, OH,  area.

Filter A was challenged with a surface water that was collected from a lake located in a Milford, OH,
gravel  pit. The lake's water resulted from water that migrated from a river and through the gravel beds
and then into the lake. The result was a  low  turbidity water. Spikes  of turbidity over 1 NTU were the
result of sludge retained in the storage systems at the Andrew W. Breidenbach Environmental Research
Center (AWBERC) research facilities.  Small amounts of raw municipal sewage were added to the
source water (5 gal of raw sewage to 5000 gal of source water) to drastically increase the bacterial load
coming into the filter. This filter was routinely challenged with water that contained over 1000 colony
forming units (CFU)AOO mL  of total coliforms and, at times, up to 10,000 CFU/100 mL. After the
initial ripening period, Filter As effluent had an average turbidity of 0.3 NTU. Except  for a minor
excursion at about 200 days of operation, when the effluent coliform count reached 10 CFU/100 mL,
routine samples showed less than 1 CFU/100 mL of total coliforms. In the first 250 days of operation,
only 8  effluent water samples (effluent  sampling was once a day) exceeded 1 CFU/100 mL for total
                                               8-3

-------
colifonns. Filter A was operated continuously for 2 years and consistently produced high quality water,
even when it was challenged with extremely poor quality source water (Fox et al. 1984). This slow sand
filter was routinely challenged with influent total coliforms levels ranging from 1,000 CFU to 10,000
CFU/lOOmL.

Filter B was constructed with both sand and carbon so that SSF could be evaluated for organic removal.
Filter B was operated for about 300 days and was monitored for trihalomethane (TFDVI) formation
potential (THMFP) and total organic carbon (TOC) removal in addition to turbidity reduction and
coliform removal. THMFP reduction averaged 95%, and TOC removal averaged 90% (Fox et al.  1984).
Although Filter B achieved good removal of both THMFP and TOC, a full-scale operating slow sand
filter with GAC media would not likely be built due to cost constraints.

The promising results for turbidity reduction and bacterial removal by SSF led WSWRD to fund addi-
tional proj ects to further investigate the capabilities of SSF. A proj ect conducted at Syracuse University
(Letterman and Cullen 1985) evaluated SSF maintenance.  The project evaluated full-scale SSF  sys-
tems and related filter maintenance to effluent water quality. The cost of maintenance was also  evalu-
ated. The  project determined that filter ripening is necessary after filter scraping to ensure that the
effluent water is acceptable. Ten ripening periods were followed in six different SSF systems. In four of
those ripening periods, it took between 0.25  and 10.00 days for the filter effluent turbidity to drop to
levels similar to that of a control filter at the same site. During the other six ripening periods, the test
filters were producing effluent turbidities that matched control filters within 5 hours. These filters were
monitored for turbidity, particle counts, and standard plate count bacteria. The factor that seemed to
have the most effect on filtrate quality was the nature of the paniculate matter in the raw water.

Filter maintenance time and cost were also evaluated at the six locations. Under typical conditions of
filter scraping (removal of 1  inch of dirty sand with shovels and conveyance of this sand from the filter
with motorized or hydraulic transport), the labor requirement was approximately 5 man-hours/1000 ft2
of filter plant area. A resanding operation that applies 6 to 12 inches of clean sand to the depleted bed
requires approximately 50 man-hours/1000 ft2 (Letterman and Cullen 1985).

Another study funded by WSWRD looked at bacteria,  Giardia, and THM removal using SSF (Pyper
1985).  In this study, a full-scale slow sand filter in Mclndoe Falls, Vermont, was monitored and chal-
lenged. This unit was no longer being used as a public water supply, thus the researchers were able to
challenge  the systems with  Giardia. During the summer operation of this filter system, the filter's
influent waters were spiked with high levels of bacteria. In addition to bacterial spiking, eight spikes of
Giardia lamblia were applied to the top  of the slow sand filter.

The slow sand filter produced a finished water that contained an average of 7 CFU/100 mL of total
coliforms for the first 8 days after scraping. The average effluent coliforms levels then dropped to an
average of 2 CFU/100 mL for the duration of the filter runs. The influent total coliforms ranged from
1000 to 10,000 CFU/100 mL, and thus the SSF routinely achieved greater than 3 log removal of total
coliforms.

This slow sand filter achieved  99.9% removal of the Giardia during the summer months and warm
water temperatures. Removals dropped to 99.5% (and  even 93.7% on one sample) during cold water
testing. The lowest removal  was observed when the influent water was at 0.5°C  (other cold
water studies were conducted at approximately 2.0°C).

In an effort to look at hydraulic loading rates and efficiency of slow sand filters, a project was spon-
sored with Colorado State University (CSU) (Bellamy et al.  1985). In this study, three slow sand filters
were operated at various flow rates (0.04, 0.12, 0.40 m/hr), and various filter effluent parameters were
monitored. In addition to the rate studies, other SSF parameters such as depth of media, sand grain size,
nutrient addition, and influent water temperature were evaluated on separate pilot slow sand filters.
                                              8-4

-------
In the hydraulic rate studies, Giardia cyst removals exceeded 99.9% for all three loading rates. It was
noted that the removal did depend on the establishment of a biopopulation within the filter bed. When
the sand was new/clean, Giardia removals were 99.0%, but the additional 0.9% removal occurred once
the filter was considered ripe.

In the temperature studies, there was no difference in Giardia removal for waters that were 17°C versus
waters that were 5°C. However, total  coliform removal was affected by the temperature difference.
Percent removal declined from 97 to 87% as the temperature was dropped.

Removals of Giardia were 99.9% for all sand sizes tested. Removal of total coliform bacteria declined
from 99.4% for 0.128 mm sand to 96.0% for 0.615 mm sand. In the media depth studies, total coliform
removals declined from 97 to 95% when the media depth (1.1  meters) was lowered to 0.5 m (0.5 m is
considered  the lowest depth of media that should be used in a slow sand filter).  Giardia removal was
not investigated in the depth of media  studies.

The studies conducted both in-house and by EPA-sponsored research have shown that SSF can be very
effective for removing both Giardia and total coliform bacteria.

Diatomaceous Earth  Filtration

DE filters are in the class of precoat filtration systems. In this class, DE is precoated on a screen, and the
water to be treated is passed through the DE coat. The coating of the screen takes place by recirculating
a DE slurry mixture through the screen until the desired thickness of precoat is achieved. After precoating,
raw water is pumped through the DE cake and paniculate material is filtered out. A small amount of DE
(body feed) is added to the raw water to build up the DE cake continuously as the filtration process
continues. Once the cake becomes too thick, the system is shut down, the cake is washed off, and the
process is restarted. Under most conditions, no chemicals are used in the DE process (in some cases,
the precoat DE is coated with alum to increase particulate attachment) (Logsdon et al. 1981).

The WSWRD DE pilot system was a  0.1 m2 DE filter (Logsdon et al. 1981). This filter system was
challenged  with Giardia cysts and radioactive beads to simulate Giardia. For these studies, Giardia
muris was used because its availability was better than Giardia lamblia. Giardia muris cysts are ellip-
soidal in shape (5-13 |im), and the radioactive beads were spherical (9 jim diameter). Giardia lamblia
cysts are 7-15 jim in diameter and are spherical.

During the bead runs, the pilot DE system demonstrated greater than 99.9% bead removal when the DE
was precoated at a concentration of 1 kg/m2 or higher. When the precoat concentration was dropped to
0.5 kg/m2, bead removal was reduced to approximately 90%. The addition of body feed to the incoming
source water was not critical to bead removal. However, the body feed was critical in the length of run.
The addition of body feed is used to maintain adequate open pore structures. Without body feed, the
precoat cake clogs rapidly and requires cleaning more often. Body feed addition increases  the length of
filter runs and thus increases the time between filter cleaning cycles.

Eleven DE  filter runs were spiked with Giardia muris cysts in the raw water just prior to entering the
DE filter chamber.  The concentration of cysts used was in the order of 107 cysts per dose. The lowest
removal of cysts observed was 99.36%. This removal was observed in one run only. The other 10 runs
achieved greater than 99.8% removal of the Giardia muris cysts.

WSWRD also established a cooperative agreement with CSU to further investigate Giardia removal
via DE filtration. In this project (Lange et al. 1984), a pilot DE system was  challenged with Giardia
lamblia cysts and other particles. In these challenge studies,  hydraulic loading rate, DE grade, and
source water temperature were evaluated to determine their effect on DE filtration  efficiency.
                                              8-5

-------
Hydraulic loading rates investigated for this study were 2.44, 4.88, and 9.76 m/hr, which covers the
range of hydraulic rates typically used by drinking water DE filtration systems. For Giardia removal,
the hydraulic loading rate did not have an impact on removal.  Giardia removal was observed to be
99.9% (or higher) for all runs. Removal of total coliform bacteria and total particles were affected by
the hydraulic loading rates. Better removal was observed at the lower flow rates than at high flow rates
(90% versus 60%, respectively).

Water temperature did not have an effect on removal of any of the parameters tested. DE filtration is a
physical screening method and does not rely on chemical  reactions or attachment for the bulk of the
particulate removal.  Thus, temperature will not show a major effect on particle removal by DE.

DE grade did have an impact on bacterial removal. Standard plate count bacteria were removed on an
average of 80 to 90% with the finer grades of DE. Removals decreased and ranged from 30 to 40% with
the coarser grades of DE. For all of the grades tested (water  treatment grades only), there was no impact
on Giardia cyst removal. Even the coarsest water treatment grade DE removed greater than 99.8% of
the Giardia lamblia  cysts.

Granular  Media Filtration

Granular media filtration in drinking water is typically referred to as conventional filtration (chemical
addition, rapid mix,  slow mix, settling, and then filtration)  or direct filtration (chemical addition, rapid
mix, possible slow mix, and then filtration). Typically, direct filtration is used to treat surface waters
with normal turbidities less than 10 NTUs (USEPA  1989). WSWRD  has conducted both in-house
studies and sponsored projects to investigate the ability of both types of filtration to remove various
pathogens.

In the early 1980s, WSWRD conducted an in-house study  to assess Giardia cyst removal via granular
media filtration (Logsdon et al. 1981). In this study, a pilot direct filtration system was set up. This pilot
system consisted of inline mixers to mix the chemicals, followed by a 3 chamber flocculation basin and
then a filter column. The filter column (3.8 cm diameter) held 46 cm of 1.27 mm effective size (e.s.)
anthracite on top of 15 cm of 0.36 mm  (e.s.) sand. Water was pumped through the system at 0.2 L/min.

One  of the first tests performed on the  system was to look at  Giardia removal through the  system
without any  coagulant being added. In this case, the feed water contained 68,000 cysts/L. Effluent
samples from this run were collected and analyzed for Giardia.  The effluent cyst counts ranged from
4,100 cysts to 28,000 cysts/L, resulting in removal ranging  from  59 to 94%. Most direct filtration
facilities for drinking water treat low turbidity source waters. This means  that many of the facilities
would use low levels of coagulants to treat their water. Alum was used for the WSWRD test runs,  and
dosages were approximately 1.4 to 3.0 mg/L. The alum dosages were adjusted to produce a filter efflu-
ent turbidity near 1.0 NTU (alum dose was not set to achieve the lowest effluent turbidity possible). In
these runs, filtered water with higher turbidity was associated with higher cyst concentrations (lower
cyst removal). Refer to Table 8-1. Giardia removal in these studies ranged from 23 to 93%.

Filtration rate changes were also investigated with this system. The flow through the system was inten-
tionally increased by 50, 100, and 150% to see the impact on Giardia removal and retention within the
filter (initial filtration rate was 2.0 gallons per square foot per minute).  In a run at 20°C in which the
floe was formed using a combination of alum and nonionic polymer, the flow rate changes had no effect
on removal. In runs where alum was the sole coagulant or when the  temperature was dropped to 10°C,
decreased cyst removal was observed. In one run, the turbidity increased by 400% and the effluent cyst
concentration increased by 2500%.  It  was concluded  that  rate changes  should be controlled to mini-
mize impacts on filtered water quality  (Logsdon et al.  1985).
                                              8-6

-------
Table 8-1. Low Alum Doses with Granular Media Filtration
Run & Length
(Hr)
A 0.5-1.5 hr
A 3. 0-4.0 hr
A 4.8-5.7 hr
B 0.9-1.4 hr
B 2.3-2.7 hr
Alum Dose
mg/L
2.2
2.1
1.9
1.4
1.8
Filtered Turbidity
(NTU)
0.58-0.51
0.63-0.46
0.79-0.77
1.0-0.95
0.65-0.60
G. muris Cysts/L
Influent
8500*


5900*

G. muris
Effluent
1000
1900
4000
1500
420
Cysts/L





  * Influent cysts concentration done once each run.

A project sponsored with CSU was funded to further evaluate Giardia removal using granular media
filtration (Al-Ani et al. 1985). This research project used both laboratory-scale and pilot-scale rapid-
rate filtration  systems. The laboratory-scale pilot plant was a dual train, conventional, rapid-rate
filtration plant built to operate under pressure. There were four filter columns: two were single media
(sand only), and the other two were dual media (anthracite over sand). The field-scale pilot plant was a
1.3 L/sec (20 gallons per minute or gpm) trailer-mounted package water treatment plant. This unit had
one dual media filter. Both treatment units could be operated in three modes of filtration: conventional,
direct (rapid mix, flocculation, and filtration), or inline (rapid mix and filtration). Most studies for this
project utilized the inline mode of operation because the water used was less than 1 NTU.

Table 8-2 shows some of the results from the laboratory-scale treatment experiments. In these experi-
ments, Giardia cysts were spiked into a 1400 L storage tank. Water from the Cache La Poudre River or
from the Horsetooth Reservoir (both low turbidity waters) were the source waters used for this study.
The spiked water was subsequently pumped through the laboratory-scale system. The effects of chemi-
cal pretreatment can be seen in Table 8-2 (this table shows  a subset of all of the runs conducted). For the
runs in which no coagulant was used, removals of all parameters were normally low, although Run 48
did achieve good Giardia removals. High Giardia removals were observed in some of the no-coagulant
runs, but the majority of these runs showed poor Giardia removal. For runs in which chemical pretreat-
ment was used, good removals of all parameters were observed.


Table 8-2. Laboratory-Scale Inline Filtration Results (Typical)
Run No.
48
120
69
52
104
70
Filter
Loading
Rate (cm/min)
8.26
32.00
22.69
8.45
8.26
22.2
Pretreatment
Chemical
None
None
alum/573c
alum/572c
alum/572c
alum/573c
Dosage (mg/L)
0
0
15/1.1
2.1/1.2
13.4/0.6
7.6/1.3
Turbidity
Reduction (%)
-18.2
18.8
73.6
91.7
82.4
88.7
Total Coliform
Removed (%)
38.4
-7.5
99.9
99.9
79.8
99.5
Giardia Cysts
Removed (%)
99.9
36.4
99.2
99.7
98.7
99.4
  573c and 572c are cationic polymers.

Table 8-3 shows a sample of the results from the field-scale pilot system. In all field-scale studies,
the water temperature was 1°C or less. Both Giardia cysts and coliform bacteria were injected into
the source water as it entered into the pilot-scale system. Without a coagulant, coliform bacteria
removals were typically 20% or less. Giardia removal was a little higher, but never exceeded 30%
with no coagulant. The effluent turbidity was higher than the influent for these test runs. When the
                                               8-7

-------
Table 8-3. Field-Scale Inline Filtration Results (Typical)
Run No.
117
125
138
Pretreatment
Chemical
none
alum
alum/572c
Turbidity Total Coliform Giardia Cysts
Dosage (mg/L) Reduction (%) Removed (%) Removed (%)
0 <1
0.4 <1
7.0/2.0 42
20
10
98
30
35
95
  572c is a cationic polymer.
coagulant conditions were optimized (Run 138), removals of all parameters improved. For these
runs, turbidity reductions exceeded 42% reducing the 0.7 NTU influent water to less than 0.4 NTU.
Giardia removals averaged 95%, while total coliform removals were greater than 98%.

This project reaffirmed that proper chemical pretreatment is imperative if rapid rate filtration is to be
effective when using low turbidity waters. The range of chemical pretreatment dosages must be correct
to achieve high reductions of turbidity, coliforms, and Giardia.  For these studies, proper chemical
pretreatment resulted in greater than 70% reduction of turbidity, 99% removal of total coliform bacte-
ria, and 95% removal of Giardia cysts.

A project was sponsored in Utah to examine both Giardia and Cryptosporidium removal through con-
ventional and direct filtration treatment systems (Nieminski and Ongreth 1995; Nieminski 1997). This
project was unique, because in addition to pilot-scale testing, the influent to a full-scale treatment plant
was spiked with Giardia and Cryptosporidium. The effluent from the full-scale plant was wasted dur-
ing this study. The pilot-scale system was a 0.5 gpm treatment plant; the full-scale plant was a 900 gpm
plant. Both systems were operated in both direct and conventional treatment modes.

Table 8-4 shows data from the pilot-scale system. This system was operated at the Jordan Valley Water
Treatment Plant.

Table 8-4. Pilot-Scale Results from the Jordan Valley Pilot Plant
Run/Mode
2-c (conv)
6-c (conv)
9-c (conv)
3-d (direct)
7-d (direct)
10-d (direct)
Giardia Removal (%)
99.16
99.95
99.91
99.78
99.90
99.99
Cryptosporidium Removal (%)
98.66
99.88
99.69
92.06
99.80
99.84
  Data presented represents the average values for each run.


Table 8-5 shows results from the Huntington, UT, full-scale seeding studies. Several factors impacted
the results of the full-scale seeding trials, which make the comparison between conventional treatment
and direct filtration more dependent on uncontrolled variables. Changes in raw water quality were
observed from the time the plant was operated in the conventional mode to the time it was operated in
the direct filtration mode. This influenced removal rates more than mode  of treatment. The water was
treated in the conventional plant during the summer when treatability was more difficult, while direct
filtration was used in the fall when the water was easier to treat. The presence of algal blooms in
samples collected  during the summer runs also created a problem in processing the Giardia and
Cryptosporidium samples.

-------
Table 8-5. Full-Scale Results from Huntington, UT, Treatment Plant
Run/Mode
1-c (conv)
2-c (conv)
4-c (conv)
1-d (direct)
2-d (direct)
3-d (direct)
Giardia Removal (%)
99.95
**
99.66
99.97
**
99.97
Cryptosporidium Removal (%)
99.60
99.05
**
99.75
99.82
99.37
  **No organisms detected. (Data presented represents the average values for each run.)
This project demonstrated that, in a properly operated treatment plant reducing turbidity to 0.1 to 0.2
NTUs, 99.9% removal of Giardia can be expected. Cryptosporidium is harder to remove than Giardia
and, under the same conditions, only 99% removal of the Cryptosporidium was observed. Both effec-
tiveness and consistency of the removal of seeded Giardia and Cryptosporidium cysts depend prima-
rily on the effectiveness and consistency of the turbidity reduction.

As part of in-house laboratory and pilot-plant studies conducted by WSWRD, methods to evaluate
treatment plant performance were evaluated. One such promising method was to use endospores to
assess treatment (Rice et al. 1994; Rice et al. 1996). Endospores (often referred to as spores) of meso-
philic, aerobic, spore-forming bacteria were suggested for use since they are not considered a public
health risk and are often found in surface waters at fairly high concentrations. These spores are ellipsoi-
dal to spherical in shape and on the average measure approximately 0.5 x 1.0 x 2.0 jam. The spores are
noted for their resistance to various environmental conditions and are resistant to disinfection. They are
easy to culture and may be present throughout most drinking water treatment trains.

A water sample to be analyzed for spores is pasteurized by heat to 80°C for  10 minutes. This heating
inactivates vegetative bacteria, but the  endospores survive. The water  sample is membrane-filtered
(0.24 |im membrane filter), and the membranes are incubated on a nutrient agar. Surviving colonies are
counted as endospore colonies.

WSWRD conducted both jar test studies and pilot-scale runs, and removals of endospores were com-
pared to reduction of turbidity and removal  of particles. Table 8-6 shows results from several pilot-
plant runs, and spore removal tracked particle removal. Table 8-7 shows seasonal log removals from
samples collected at an utility.


Table 8-6. Removals of Aerobic Spores and Particles Across Pilot Plant
Log Reduction
Location
Conventional coagulation Settled
Conventional coagulation
Chlorinated-filtered
Enhanced coagulation
Settled
Enhanced coagulation
Chlorinated-filtered
Spores
0.85
2.12
1.51
2.42
Particles
3-5 p,m
1.11
2.1
1.86
2.91
Total Particles
0.91
1.7
1.39
2.87
                                              8-9

-------
Table 8-7. Seasonal Cumulative Log Removals Through Unit Process at a Utility
Log Removals
Particles
Season
Winter
Winter
Winter
Spring
Spring
Spring
Unit Process
Settling
Sand filtration
GAC filtration
Settling
Sand filtration
GAC filtration
Spores
1.19
2.19
2.89
1.39
2.57
2.70
3-5 [im
1.08
2.12
2.92
0.99
2.00
2.19
Total
1.22
2.02
2.96
1.32
2.19
2.30
Monitoring for indigenous spores of aerobic spore-forming bacteria represents a viable method for
determining water treatment plant performance. Unlike many microbiological parameters, spore con-
centrations can be detected throughout the treatment process and do not propagate in the water plant.
These organisms do not present a public health threat, and because they originate primarily from soil,
they would tend to be in surface waters that receive any type of runoff. The analytical technique is
straightforward and can be performed by most microbiological laboratories. Thus, these endospores
can be used to monitor treatment plant performance.

In an effort to better understand Cryptosporidium removal, WSWRD built a small pilot plant (SPP)
with theoretical residence time of 10 hours  from rapid mix to the weir. This plant was designed to
minimize flow rates and permit steady-state feeding of Cryptosporidium (Dugan et al. 2000). In each of
14 studies, target concentrations of 106 Cryptosporidium oocysts/L were seeded into the plant influent
for 30 to 71  hours. The purpose of these  studies was to evaluate the resulting log removals of
Cryptosporidium oocysts, turbidity, total particle counts (TPCs,  1 to 150 |im diameter), and  aerobic
endospores (spores). Log removals were evaluated as a function of coagulant type, coagulant dose, raw
water quality, filter loading rates, and filter media.

Figure 8-1  is a schematic of the SPP. Two identical SPPs were available for these studies. For this
discussion, runs  in which both  SPPs operated together carry "A" or "B"  subscripts to account for
parallel operation. The treatment process for each SPP consisted of Cryptosporidium addition, inline
mixing, coagulant addition at the rapid mix, flocculation, sedimentation, and filtration. Each SPP was
designed to run at a flow rate of 450 mL/min. The construction and dimensions of the SPP have been
described in detail by Lytle (Lytle and Fox  1998). Raw water from the Ohio River (ORW) was used as
source water for all SPP runs. Water quality parameters for the 14 runs are summarized in Table 8-8.
The raw water was stored in a 5000 gallon tank, which was equipped with  submersible recirculating
pumps in order to minimize settling of particulates. The water temperature for all 14 SPP runs was 19
to2FC(66to70°F).

Prior to the initiation of formal  pilot testing, Cryptosporidium oocysts were fed to the SPP without
coagulant addition. The purpose of this run was to examine oocyst removal through attachment to floe
tank, sedimentation basin and filter walls, and to filter media. Approximately 3.5  x 105 oocysts/L were
fed into the raw water for 13 hours. The rapid mix and floe paddles were run at their design speeds, and
two parallel dual media filters were run at loading rates of 5 and  10 m/hr. Raw, settled, and filter
effluent Cryptosporidium samples were collected at 10, 12, and 13 hours. From raw to settled water,
the average removal of Cryptosporidium was 0.25 Iog10 (o = 0.29). For low and high loading rate filters,
Cryptosporidium removals were -0.04 and -0.05 Iog10, (o = 0.31 and 0.38 Iog10), respectively. As a
result, Cryptosporidium attachment to pilot-plant surfaces was not considered a statistically significant
fraction of the log removals observed during  the 14 pilot-scale runs.
                                             8-10

-------
                                         Coagulant
                                         Reservoir
                                         Peristaltic
                                         Pump
                          D
                           Oocyst
                           Suspension
        Peristaltic
        Pump
      Raw       Static
     Water      Mixer
     Pump
                  Flocculation
Sedimentation
        Dual Media Filter
>

V



1
20" Anthracite
30" Sand
10" Sand
12" Support Media
> '
V


1
                                                                  o Sample Point
                                                                    (Raw, Settled, Filter Effluent)
                                           Single Media Filter
Pump
                                                            6—1
                                                           Pump
Figure 8-1. Mini pilot-plant schematic (SSP).

Runs 1 through 9 used a single pilot plant operating a single dual media filter at a 5 m/hr loading rate.
The primary goal of the first 9 runs was to examine the impact of under-coagulation on the downstream
removal ofCryptosporidium. To achieve this objective, coagulant in Runs 1 through 5 was deliberately
underdosed  relative to jar test  predictions. Runs 6 through 9 were  performed with the optimum
coagulant concentrations predicted by jar testing. The optimum coagulant dose was defined as the
concentration necessary to reach the bottom of the dose versus the settled turbidity curve in a given jar
test.  In Runs 10 through 14, the focus shifted to examining the impact of filter media, filter loading
rates, and coagulant type. In Run 10, a dual media and a single media filter were run in parallel, at 5 m/
hr loading rates, to examine the impact of filter media on oocyst removal. In Runs 11 and 12, two pilot
plants were run in parallel, with  one dual media filter per plant, each operating at 5 m/hr. The goals of
                                              8-11

-------
        Table 8-8. Summary of Small Pilot-Plant Runs with Cryptosporidium

-------
Runs 11 and 12 were to investigate the relative impacts of alum vs. ferric chloride and alum vs. polymer
coagulation, respectively, on the filtration removal ofCryptosporidium.

In each run, the coagulant doses for plants A and B were adjusted so that settled turbidities and particle
counts, and hence filter influent particle loadings, were as close as possible. As a result, any differences
in filtration removal of Cryptosporidium would have been due to  intrinsic chemical differences be-
tween the two coagulants. In Run 13, two dual media filters were run in parallel, at loading rates of 5
and 10 m/hr, to examine the impact of filter loading rates on oocyst removal. The goal of Run 14 was to
examine the impact of coagulant dose on the filtration removal of Cryptosporidium in high and low
loading rate filters. Alum doses for Runs 14A and B were 20 and 60 mg/L, respectively. These repre-
sented conventional and enhanced coagulant doses.

Average sedimentation removals of TPCs, spores, and Cryptosporidium and reduction of turbidity for
all 14 SPP runs are summarized in Table 8-9. The average turbidity and Cryptosporidium removals
through sedimentation are plotted against each other in Table 8-9 as a function of raw turbidity and
coagulant type. The division of raw turbidities into low (less than 10 NTU), medium (10 to 50 NTU),
and high (greater than 50 NTU) ranges was based on an analysis of raw ORW turbidity data collected
by the United States Geological Survey (USGS), as part of its National Stream Quality Accounting
Network. The data were collected at the USGS sampling station closest to the point where raw water
was collected for pilot plant runs, approximately 50 river miles downstream. The 25th and 75th percen-
tiles of the USGS turbidity data fell at approximately 10 and 50 NTU, respectively.

The results from suboptimal coagulation (runs 1 to 5) have been identified in Table 8-9. Sedimentation
Cryptosporidium removals  during these runs averaged 0.2 Iog10. Low sedimentation removals of
Cryptosporidium during sub-optimal runs were observed regardless of raw water turbidity. Sedimenta-
tion  removals of Cryptosporidium averaged 1.3 Iog10 in all of the remaining runs (6 to 14). In these

Table 8-9. Sedimentation Performance
Turbidity (Alog10)
Suboptimal

"3
"S.
O

Run
1
2
3
4
5
Mean:
6
7
8
9
10
11A
11B
12A
12B
13
14A
14Ba
Mean:
X
0.53
0.51
0.23
0.10
0.093
0.29
1.5
0.92
1.2
0.83
1.7
1.1
1.2
0.35
0.42
1.4
0.98
1.2
1.1
O
0.49
0.12
0.13
0.046
0.037

0.53
0.12
0.25
0.022
0.076
0.13
0.19
0.11
0.082
0.065
0.084
0.11

TPC (Alog10)
x~
0.
0.
0.
0.

71
52
43
15
-0.13
0.
1.
1.
1.
0.
1.
1.
1.
0.
0.
1.
1.
1.
1.
34
5
1
4
91
8
2
2
53
62
4
1
3
2
O
0.20
0.061
0.14
0.062
1.6

0.46
0.14
0.093
0.29
0.082
0.15
0.12
0.21
0.20
0.090
0.25
0.39

X
0
0
0
0
0
0
1
1
1
0
1
1
1
0
0
1
1
1
1
Spores (Alog10)

.42
.27
.22
.070
.10
.22
.4
.0
.1
.76
.6
.1
.1
.69
.75
.5
.3
.5
.2
O
0.14
0.42
0.10
0.37
0.46

0.32
0.44
0.44
0.18
0.16
0.37
0.52
0.32
0.34
0.20
0.32
0.53

Oocysts (Alog10)
x"
0.36
0.26
0.56
-0.020
-0.16
0.20
1.6
0.79
o
.5
.2
.8
.4
.6
0.62
0.72
1.3
1.2
1.6
1.3
a
0.38
0.23
0.20
0.38
0.39

0.62
0.81
0.45
0.48
0.44
0.26
0.20
0.34
0.57
0.56
0.43
0.37

  1 Enhanced coagulation
                                             8-13

-------
       o
       E
       O)
      .3 1 -I
       |
       o
       Q.
        50 NTU)
          0                                     1                                     2
                                       Turbidity Log Reduction

Figure 8-2. Turbidity reduction vs. Cryptosporidium removal (raw to settled).

runs, Cryptosporidium removals were positively and linearly correlated with turbidity reduction. The
magnitudes of the observed log removals also tended to correlate with raw water quality in runs 6
through 14. The lowest Cryptosporidium and turbidity removals were observed in the low turbidity raw
waters, and the highest removals were observed in the high turbidity waters. The relationships between
TPC, spore,  and Cryptosporidium removals were similar to the one shown for  turbidity and
Cryptosporidium in Figure 8-2.

Log removals of turbidity, TPCs, spores, and Cryptosporidium across filters operating at 5 m/hr load-
ing rates are summarized in Table 8-10. Suboptimal coagulation had a dramatic impact on filtration
removals of Cryptosporidium. Oocyst removals averaged 1.5 Iog10  in the suboptimal runs (1 to 5). In
contrast, Cryptosporidium removals averaged greater than 3.7 Iog10 in all other runs (6 to 14). The
relatively poor filtration performances observed in the  suboptimal runs were constant with respect to
time. The low log removals were not the result of breakthrough behavior. Suboptimal coagulation also
had a significant impact on the differences between Cryptosporidium and surrogate removals. Average
log removals of Cryptosporidium and all three surrogate parameters were within 0.5 Iog10 of each other
for the suboptimal runs. In the remaining runs, average Cryptosporidium removals were  at least 1.7
Iog10 higher than TPC and spore removals, and at least  2.4 Iog10 higher than turbidity removals. Rela-
tively poor Cryptosporidium removals during suboptimal runs were associated with higher filter efflu-
ent turbidities. Filter effluent turbidities during the suboptimal runs averaged 0.31 NTU  (o = 0.24).
During the remaining runs, effluent turbidities averaged 0.08 NTU (o = 0.03).

Run 10 examined the impact of filter  media on Cryptosporidium  and surrogate removals at 5 m/hr
loading rates. The sand filter had to be backwashed about halfway through the run. However, sand filter
                                              8-14

-------
Table 8-10. Summary of Filtration Performance (5 m/hr)

Suboptimal
Coagulation

Optimal
Coagulation

Run
1
2
o
J
4
5
Mean:
6
7
8
9
10 (A/S)
10 (sand)
11A
11B
12A
12B
13
14A
14B
Mean:
Average Turbidity
Eff.Turbidity (Alog10)
(NTU) x o
0.66
0.13
0.13
0.45
0.18
0.31
0.02
0.08
0.14
0.15
0.06
0.06
0.06
0.08
0.09
0.09
0.08
0.10
0.08
0.08
1.4
1.7
1.4
0.6
0.3
1.1
2.3
1.4
0.68
1.4
1.6
1.5
1.3
1.1
0.56
0.49
1.4
1.4
1.2
1.3
0.17
0.10
0.14
0.10
0.12

0.66
0.39
0.29
0.33
0.17
0.23
0.18
0.24
0.14
0.20
0.50
0.27
0.39

TPC
(Alog10)
x 0
2.0
2.3
1.7
0.78
2.0
1.8
2.9
2.5
1.4
2.1
2.3
2.1
2.1
2.0
1.4
1.2
2.3
2.0
1.6
2.0
0.034
0.096
0.50
0.71
1.8

0.63
0.44
0.19
0.57
0.25
0.091
0.14
0.19
0.29
0.49
0.42
0.87
0.48

Spores
(Alog10)
x 0
1.2
1.2
1.3
0.61
0.93
1.0
2.7
2.0
0.73
1.6
2.6
1.7
3.4
1.9
1.5
1.3
3.0
2.1
1.9
2.0
0.16
0.39
0.14
0.38
0.55

1.5
0.43
0.44
0.15
0.82
0.25
0.94
0.47
0.53
0.63
0.77
0.78
0.43

Oocysts Oocyst Nondetects
(Alog10) (% of samples)
x 0
1.2
1.5
3.6
0.18
0.8
1.5
2.9
4.4
>3.2
3.7
3.5
>3.6
>3.6
>3.3
>4.3
>4.4
>3.6
>4.1
3.7
>3.7
1
0
0
0
0

1
1
1
0
0
0
0
0
0
0
0
0
1

.1
.41
.68
.19
.36

.1
.1
.9
.62
.54
.48
.80
.88
.39
.66
.65
.44
.0

-
-
-
-
-

-
-
75
-
-
22
38
33
13
13
33
17
-

Cryptosporidium removals did not appear to decrease over the course of the filter cycle. There were no
significant differences in Cryptosporidium, turbidity, or TPC log removals with respect to filter media.
Spore removals (Table 8-10), however, differed by more than 1 Iog10. It is not known why spore remov-
als in the sand filter were consistently lower relative to the dual media filter.

Runs 11 and 12 evaluated the impact of coagulant type on Cryptosporidium control through filtration.
Runs 11A and B compared alum and ferric chloride, while runs 12A and B compared alum and poly-
mer. The alum dose in each of the runs was set at the optimum concentration. The alternative coagulant
dose in each case was adjusted to yield equivalent settled turbidities. Weir turbidities in runs 11A and B
averaged 1.1 NTU (o = 0.1) and 1.2 NTU (o = 0.2), respectively. Settled TPCs in runs 11A and B
averaged 4.6 x 10V10 mL (o = 0.7 x 105) and 4.2 x 105/10 mL (o = 0.5 x 105), respectively. Settled
turbidities in runs 12A and B both averaged 0.3 NTU (o  = 0.04 and 0.03 NTU, respectively). Settled
TPCs in runs 12A and B averaged 1.0 x 105/10 mL (o = 0.1 x 105) and 0.8 x 105/10 mL (o= 0.1 x 105),
respectively.

These results indicate that the filters in each run were loaded with water of equivalent particulate
concentrations. Consequently, any differences in Cryptosporidium removal through filtration should
have resulted from intrinsic chemical differences between the two coagulants. However, no signifi-
cant differences in Cryptosporidium control with respect to coagulant type were observed. Filtration
removals of Cryptosporidium in runs 11A and B averaged greater than 3.6 Iog10 (o = 0.80) and 3.3
Iog10 (o = 0.88), respectively. Filtration removals of Cryptosporidium in  runs 12A and B averaged
greater than 4.3 Iog10 (o = 0.39) and greater than 4.4 Iog10 (o = 0.66), respectively. With the exception
of spores in run 1 IB, relative surrogate log removals also did not vary with respect to coagulant type.
                                              8-15

-------
Run 13 evaluated the impact of filter loading rates on Cryptosporidium and surrogate removals. Runs
14A and B investigated the effects of coagulant dose and filter loading rates on Cryptosporidium and
surrogate removals. Two dual media filters were run in each plant at 5 and 10 m/hr, respectively, in each
of these three runs. Removals of Cryptosporidium and surrogates were stable at low (5 m/hr) filter
loading rates in all three runs. In runs 14A and B, coagulant dose did not significantly affect
Cryptosporidium removals at low loading rates. Average low-rate filtration removals of Cryptosporidium
in runs 14A and B were greater than 4.1 Iog10 (o = 0.44) and 3.7 Iog10 (o = 1.0), respectively. Differ-
ences in average low-rate surrogate removals as a function  of coagulant dose were also not significant
(refer to Table 8-10).

Summary

Over the years, WSWRD has continued to evaluate various treatment techniques for removing micro-
organisms from drinking water. This evaluation has included not only evaluating under what condi-
tions removals occur, but also under what conditions organism removals deteriorate. The primary fil-
tration processes that were part of WSWRD research include SSF, DE filtration, direct filtration, and
conventional filtration.

Giardia cysts and Cryptosporidium oocysts were the primary organisms investigated during the 1980s
and 1990s. The filtration processes were challenged with various water qualities based on particle and
pathogen loads. It was shown in these studies that filtration processes are capable  of removing high
levels of those organisms, but that removal is dependent  on the operation of those processes. Poor
chemical addition and coagulation reduced the treatment efficiency for removing microorganism and
other particles. Temperature was another factor that affected removal. Low water temperature reduced
removals in SSF and conventional systems, but had little  effect of DE systems. Most of the studies
conducted by the WSWRD showed that good turbidity reduction and good  particle removal paralleled
good organism removal.

One measure of filtration efficiency developed by WSWRD, and now widely used, was that of measur-
ing aerobic endospore removal in a water treatment process. Although not a surrogate for removal of a
specific organism, endospore removal was seen as a measure of the overall filtration  performance.
Studies conducted by the WSWRD showed that endospore removal tracked particle removal, turbidity
reduction, and pathogen removal. In studies in which good removals  of endospores were achieved,
good removals of particle and pathogens were observed. In some cases, good removals of pathogens
occurred when poor removals of endospores were demonstrated, but no studies showed poor removals
of pathogens when good removals of spores were achieved. Thus, removals of spores may be consid-
ered a conservative measure of particle and pathogen removal.

References

Al-Ani, M., McElroy, J. M., Hibler, C. P., and Hendricks, D. W. (1985). "Filtration of Giardia cysts
    and other substances: Volume 3." Rapid rate filtration.  PB 85-194 645/AS, U.S. Environmental
    Protection Agency, Cincinnati, OH.

Bellamy, W. D., Silverman, G. P., and Hendricks, D. W. (1985). "Filtration of Giardia cysts and
    other substances: Volume 2." Slow sand filtration. PB 85 191 633/AS, U.S. Environmental
    Protection Agency, Cincinnati, OH.

Centers for Disease Control and Prevention. (1996). "Surveillance for waterborne-disease out-
    breaks—United States, 1993-1994." Morbidity & Mortality Weekly Report, 45(SS-3).
                                             8-16

-------
Centers for Disease Control and Prevention. (2000). "Surveillance for waterborne-disease out-
   breaks—United States, 1997-1998." Morbidity & Mortality Weekly Report, 49(SS-4).
Clark, R. M., et al. (1996). "A waterborne Salmonella typhimurium outbreak in Gideon, Missouri:
   Results from a field investigation." InternationalJournal of Environmental Health Research, 6,
   187-193.

Craun, G. R (1991). "Causes of waterborne outbreaks in the United States." Water Science Technol-
   ogy, 24(3), 17-20.
Dugan, N., Fox, K. R., Miltner, R. J., and Owens, J. H. (2000). "Control of Cryptosporidium oocysts
   by conventional treatment." Accepted for publication, Journal of the American Water Works
   Association, March 2001.

Fox, K. R. (1996). "Waterborne disease outbreaks and small systems." Journal of the American
   Water Works Association, 88(9), 8.
Fox, K. R. and Lytle, D. A. (1996). "The Cryptosporidiosis outbreak in Milwaukee: Investigation
   and recommendations." Journal of the American Water Works Association, 88(9).

Fox, K. R., Miltner, R. J., Gary, S. L., Dicks, D. L., and Drolet, L. F. (1984). "Pilot-plant studies of
   slow-rate filtration." Journal of the American Water Works Association, 76(12), 62-68.
Lange, K. P., Bellamy, W. D., and Hendricks, D. W. (1984). "Filtration ofGiardia cysts and other
   substances, Volume 1." Diatomaceous Earth Filtration. PB 84-212 703, U.S. Environmental
   Protection Agency, Cincinnati, OH.
Letterman, R. D. and Cullen, J. T. R. (1985). "Slow sand filter maintenance: Costs and effects on
   water quality." PB 85-199 669/AS, U.S. Environmental Protection Agency, Cincinnati, OH.
Logsdon, G. S. and Lippy, E. C. (1982). "The role of filtration in preventing waterborne disease."
   Journal of the American Water Works Association, 74(12), 649-655.

Logsdon, G. S., Symons, J. M., Hoye, R. L., and Arozarena, M. M. (1981). "Alternative filtration
   methods for removal of Giardia cysts and cysts models." Journal of the American Water Works
   Association, 73 (2), 111 -118.
Logsdon, G. S., Thurman, V. C., Frindt, E. S., and Stoecker, J. G. (1985). "Evaluating sedimentation
   and various filter media for removing Giardia cysts." Journal of the American Water Works
   Association, 77(2), 61.
Lytle, D.  A. and Fox, K. R. (1998). "The design of a "mini-scale" conventional/filtration water
   treatment plant for Cryptosporidium research." American Water Works Association Annual
   Conference, Dallas, TX, June 21-25.
MacKenzie, W. R., et al. (1994). "A massive outbreak in Milwaukee of Cryptosporidium infection
   transmitted through the public water supply." New England Journal of Medicine, 331(3),
   161-167.
Nieminski, E. C. (1997). "Removal of Cryptosporidium and Giardia through conventional water
   treatment and direct filtration." PB97-162507, U.S. Environmental Protection Agency, Cincin-
   nati, Ohio.
Nieminski, E. C. and Ongreth, J. E. (1995). "Removing Giardia and Cryptosporidium by conven-
   tional treatment and direct filtration." Journal of the American Water Works Association, 87(9),
   96-106.
                                              8-17

-------
Pyper, G. R. (1985). "Slow sand filter and package treatment plant evaluation: Operating costs and
   removal of bacteria, Giardia, and trihalomethanes." PB 85-197 051/AS, U.S. Environmental
   Protection Agency, Cincinnati, OH.

Rice, E. W., Fox, K. R., Miltner, R. 1, Lytle, D. A., and Johnson, C. H. (1994). "A microbiological
   surrogate for evaluating treatment efficiency." American Water Works Association Water Quality
   Technology Conference, San Francisco, CA, November 6-9.

Rice, E. W., Fox, K. R., Miltner, R. 1, Lytle, D. A., and Johnson, C. H. (1996). "Evaluating plant
   performance with endospores." Journal of the American Water Works Association, 99(9),
   122-130.

Swerdlow, D. L., et al. (1992). "A waterborne outbreak in Missouri of Escherichia coli O157:H7
   associated with bloody diarrhea and death." Annals of 'InternalMedicine, 117(10), 812-819.

U.S. Environmental Protection Agency (USEPA). (1989). "Guidance manual for the compliance
   with the filtration and disinfection requirements for public water systems using surface water
   sources." NTISPB93-222933, Washington, D.C.

USEPA. (1995). "The national public water system supervision program FY 1994 compliance
   report." EPA 812-R-95-001, Office of Water, Washington, D.C.
                                            8-18

-------
                                      CHAPTER 9

 Activated Carbon and Membrane Processes for Disinfection By-Product (DBP)
                                and Microbial Control1

Introduction

It is likely that many utilities will be able to meet current and upcoming drinking water regulations for
DBFs by implementing one of the following relatively low-cost options: changing coagulation condi-
tions, changing the point of chlorination, or switching to an alternative disinfectant (Symons et al.
1981). However, some utilities may wish to utilize activated carbon or membranes either because a
lower-cost solution is not practical or because they wish to take advantage of the unique properties of
activated carbon or membrane processes. Activated carbon and membrane processes are considered
higher-price options (compared to enhanced coagulation) for DBP precursor removal and would most
likely require major plant construction, hence they are considered together in this chapter. For point-of-
use or point-of-entry discussions, the reader is referred to Chapter 11, "Controlling Disinfection  By-
Products (DBFs) and Microbial Contaminants in Small Public Water Systems (PWSs)."

For both GAC and membrane processes, it is more economical to remove the DBP precursor material
than the formed DBFs. DBP precursors, as a whole, are more readily adsorbed onto activated carbon
than DBFs (Symons et al. 1981). Precursor materials have larger molecular sizes than DBFs; therefore,
it is  easier for membranes to reject precursor material. Also, both activated carbon and membranes
have problems handling chlorinated water. Activated carbon quickly reduces free chlorine. This lowers
the capacity of the carbon, makes the carbon more brittle, and increases the amount of dioxins formed
upon regeneration (Lykins et al. 1988b). Also, because activated carbon reduces the disinfectant, postfilter
chlorination will be needed, which will form additional DBFs from the precursor material that was not
adsorbed onto the column. Free chlorine attacks membrane material through oxidation pathways,  and
failure quickly occurs for many of these chlorine-sensitive thin-film membranes. Thin-film membranes
are commonly used today because they have better flux and biodegradation characteristics than chlo-
rine-resistant membranes.

Activated carbon is  commonly applied as powdered activated carbon (PAC) or in granular activated
carbon (GAC) form. PAC is often applied at, or before, the coagulation/flocculation step. The pow-
dered carbon adsorbs contaminants and natural organic matter (NOM) until it is removed downstream
in the sedimentation and filtration processes. Unless specific changes are made to the water treatment
train (floe blanket clarifier or membranes), the typical adsorption residence time is too short to remove
a significant amount of the NOM (DBP precursors), which generally adsorb slowly as compared to
synthetic organic chemicals (SOCs). With regard to prechlorinated waters, the PAC adsorption capaci-
ties for DBFs are too low for economical removal (Symons et al. 1981). Therefore, PAC is most often
used for SOC or taste and odor control.

GAC is utilized in a filter mode. It can be used as part of a multi-media filter to remove particulates (filter
adsorber) or as a postfilter to remove  specific contaminants (postfilter adsorber).  Filter adsorbers are
operated as typical inert-media filters. They are backwashed periodically to alleviate head loss,  and the
carbon is regenerated very infrequently, if at all. When used in postfilter mode, the bed is rarely backwashed,
and the GAC is regenerated as often as needed to control for the contaminant(s) of interest.
Thomas F. Speth: ORD/NRMRL/WSWRD, AWBERC Mailstop 689, 26 West Martin Luther King
Dr., Cincinnati, OH 45268, 513-569-7208, speth.thomas@epa.gov.
                                             9-1

-------
Activated carbon has no specific ability to remove microbial pathogens unless it is used as GAC in a
filter adsorber application, where it removes pathogens by the same mechanisms as any other filter
media. GAC, due to its typically larger size, does not remove particulates/pathogens to any greater
degree than other filter media types, so its use is never recommended if particulate/pathogen removal is
the only goal. Therefore, this chapter will only cover activated carbon processes for DBF control.

Certain types of membranes can be very effective for controlling DBFs, while others are specifically
designed to remove particulates/pathogens. Reverse osmosis (RO) membranes are very tight  mem-
branes that have molecular-weight cutoffs (MWCOs) below 200 daltons. They are typically used to
remove salts from seawater and brackish waters. Due to their tight membrane structure, they are oper-
ated at very high pressures (10 to 100 bar).

Nanofiltration (NF) membranes are not as tight as RO membranes. The MWCOs for NF membranes
are generally considered to range between 200 and 1,000 daltons. They are designed to remove divalent
cations,  hence they are often referred to as softening membranes, although they have been found to
remove a large percentage of DBF precursors. Because they are not as tight as RO membranes, they can
be operated at lower pressures (typically 5 to 9 bar) while achieving fluxes that are the same, or greater,
than RO membranes. These lower pressures make NF membranes less expensive than RO membranes
for a given design flow. Research at the U.S. Environmental Protection Agency (EPA) Office of Re-
search and Development (ORD) has therefore concentrated on NF membranes.

Ultrafiltration (UF) and microfiltration (MF) membranes are typically used only for particulate/patho-
gen removal. UF  membranes have MWCOs that range from 1,000 to 500,000 daltons. While some of
the UF membranes that have MWCOs near 1,000 daltons may remove significant amounts of DBF
precursor material, the MWCO ranges are arbitrarily set, and therefore, the membrane could be consid-
ered  a loose NF membrane. MF membranes have an order-of-magnitude larger pore sizes  than UF
membranes. Typically, MF pores sizes are designated in micrometers and normally range from 0.05 to
5 jim. A rough rule of thumb is that UF membranes can reject viruses, while MF membranes cannot.

This chapter provides a comprehensive review of activated carbon and membrane research for the
control of DBFs and pathogens. Much of the work was conducted,  or funded, by ORD.  Outputs
include: peer-reviewed papers, proceedings papers, EPA reports, Master's theses, and Ph.D. disserta-
tions. It also includes other papers that were written under non-ORD projects. Some of these were
co-authored by ORD researchers, but many were not. Although the intent is to highlight ORD re-
search, non-ORD projects are included to make the discussion complete.

This work is a follow-up to the EPA work published by Symons et al. (1981). For other recent compre-
hensive scientific discussions of GAC and membrane technologies, the reader is referred to Jacangelo
(1999), Snoeyink et al. (1999), Snoeyink and Summers (1999), and Taylor and Wiesner (1999). Also,
the Information Collection Rule (ICR) treatment studies have been recently compiled.  The EPA re-
quired water utilities of a certain size and water quality to complete GAC or membrane studies so as to
create a national data base for advanced DBF removal technologies. The ICR treatment studies were
conducted through the auspices of the Office of Ground Water and Drinking Water with very limited
ORD involvement.

Activated Carbon
Filter Adsorbers vs. Postfllter Adsorbers

As mentioned previously, the removal of NOM and DBF precursors by PAC is not very efficient (Symons
et al. 1981). Therefore, the following discussion will only pertain to GAC filtration technologies. The
major limitation for filter adsorbers is their depth. The Ten-State Standards require sand depths of at
                                            9-2

-------
least 4 feet. Therefore, this leaves a limited depth of GAC that can be placed above the sand. The lack
of depth, or empty bed contact time (EBCT), is crucial for waters that contain moderate- to poor-
adsorbing NOM. Early breakthrough would require frequent replacement or regeneration of the car-
bon. Hartman et al. (1991) determined that filter adsorbers were not cost effective for NOM control.
However, due to GAC's ability to maintain a biological community that would remove DBF precur-
sors, systems that need only limited DBF precursor removal may find filter adsorbers to be practical.
Wiesner et al. (1987) concluded that filter adsorbers are generally more cost effective than postfilter
adsorbers if the desired total organic carbon (TOC) (precursor) removal is less than 55 percent. Under
more unique conditions, Wiesner et al. (1987) found that filter adsorbers are cost effective for removals
up to 75 percent. Because limited removals can generally be obtained by other means, ORD research
has concentrated on postfilter adsorbers which can be highly effective for controlling disinfection by-
products under a wide range of conditions.

Breakthrough Profiles

Figure 9-1 contains a typical NOM, as TOC, breakthrough profile for a postfilter GAC column (Miltner
et al. 1996).  There is immediate TOC breakthrough because a certain portion of the TOC is nonadsorbable.
The effluent TOC concentration then slowly increases toward the influent concentration. The break-
through profile eventually plateaus as the biodegradable or slowly adsorbing fractions are removed.

These results are different than for SOCs that  generally show complete  removal for a period of time,
followed by a sharp "S"-shaped breakthrough to the influent level as shown in Figure 9-2 (Speth and
Miltner 1989). Singer (1994) found that three-to six-month run times between regenerations were most
common  for NOM removal.
         O
         O
                                                   Ozonated Marsha Lake Water
                                                   F-400 Carbon
             0.5 -
             0.0
                                                                            250
                                       Operation Time (days)
Figure 9-1. TOC breakthrough profile for F-400 carbon treating ozonated Harsha Lake water
            (Miltner et al. 1996).

-------
          500 r
  400

°B>
 I
g 300
Q
"5
g
2200
         o
        O
          100
                Microcolumn 2 Influent
                Pilot Column Influent
 0
00
                Microcolumn 1 Influent
                                                
-------
Even though DBF formation potential (DBPFP) breakthrough can be predicted by TOC breakthrough,
a GAC column's performance for controlling DBFs is not always straightforward. The GAC adsorp-
tion performance is predicated on an interwoven matrix of system design, water quality, and chlorina-
tion conditions.

GAC Adsorption Performance

There are many factors that affect GAC adsorption performance. Assuming instantaneous kinetics, an
equation to calculate the time to breakthrough in days is:

                      Time to breakthrough (days) = (K Co1/n Wt) / (Q Co)            (9-1)

where K and 1/n are the Freundlich constants in |ig/g (L/|ig)1/n, Co is the influent concentration in |ig/L,
Wt is the weight of carbon in g, and Q is the volumetric flow rate in L/day.

Clark et al. (1986) and Clark (1987) developed the logistic function which describes the TOC effluent
concentration in terms of time, initial concentration, EBCT, and adsorption capacity as represented by
the Freundlich equation.

                             TOCf = ([TOCJ"-1) / (1 + A e-rt)1/n~l                   (9-2)

where A and r are fitting parameters, TOCf is the effluent TOC concentration, TOCo is the influent TOC
concentration, t is time, and n is the Freundlich parameter. The fitting parameters A and r were further
related to EBCT. The exact equations were determined by  fitting the following equations to break-
through profiles from GAC columns at Jefferson Parish (Clark 1987)

                                    A = 0.757(EBCT)135                          (9-3)

                                   r = 0.0743(EBCT)-°429                         (9-4)

where EBCT is the empty bed contact time (volume of the bed divided by the volumetric flow rate).

The preceding equations demonstrate the importance of TOC adsorbability, initial concentration, and
EBCT for a given carbon. Many other models have been developed for GAC adsorption; however, they
have limited applicability to DBF precursor adsorption because of the undeterminable heterogeneity of
the natural organics in the influent water. Luft (1984), Crittenden et al. (1987b),  Speth (1986), Warta et
al. (1995), and Hong (1985) have worked with limited success in developing and applying hypothetical
components for predicting TOC adsorption.

Adsorbability

Different carbons will adsorb DBF precursors to varying degrees. Figure  9-4 shows the breakthrough
profiles for three different activated carbons that treated ozonated Harsha Lake Water (Miltner et al.
1996). Wang et al. (1995) found that trihalomethane formation potential  (TFDVIFP) removals ranged
from 27 to 40 percent, while TOXFP removals ranged from 31 to 52 percent depending on what type of
carbon was used.

The NOM and DBF precursors of different waters  will adsorb to different extents. Figure 9-5 shows an
atypical TOC breakthrough profile for a ground water from Fairfield, OH, using F-400 carbon (Speth
and Miltner 1989). In this case, approximately 80 percent of the NOM in the Fairfield water  was
nonadsorbable. These results can be compared to those in Figure 9-1 for Harsha Lake water. The NOM
in the Harsha Lake water was adsorbed to a much greater  extent.
                                             9-5

-------
                                                        Ozonated Marsha Lake Water
                                                                 Pica Carbon
                                                                 Norit Carbon
                                                                 F-400 Carbon
            0.0
                                          Operation Time (Days)
                                                                                  250
Figure 9-4. TOC breakthrough profiles for different activated carbons treating ozonated
           Harsha Lake water (Miltner et al. 1996).
               O)


               O
               o
                       —• Pilot-Column Influent
                       	• Pilot-Column Effluent, 4.54-min EBCT
                       - * Pilot-Column Effluent, 7.95-min EBCT
                            10        20        30       40
                                           Time - Days
50
60
Figure 9-5. TOC Breakthrough profile for ground water from Fairfield, OH (Speth and
           Miltner 1989).
                                               9-6

-------
Generally, smaller NOM molecules adsorb more strongly than larger ones because pore blockage of
large molecules can limit the accessibility to adsorption sites. Also, the greater the hydrophobicity of
the NOM, the greater the adsorption. Aromatic compounds are generally more strongly adsorbed than
nonaromatic compounds. Therefore, terrestrial-derived organic compounds which tend to have greater
aromatic character are expected to adsorb to a greater extent than aquatic-derived ones. Proteinaceous
compounds are also known to be strongly adsorbed.

The Polanyi potential model can predict adsorption capacities based on the following equation (Speth
1986; Speth and Adams 1991). The Polanyi theory assumes there is a fixed volume close to the adsorp-
tion surface where adsorption occurs. The volume is  defined by equipotential surfaces which describe
the amount of work needed to move any molecule from the bulk solution to the adsorption space. A
final form of this development is:

                            K = p W exp((-p B R T In (Cs)) / Mw)                  (9-5)

where K is the Fruendlich parameter,  p is the compound density, W and B are fitting constants, R is the
ideal gas constant, T is temperature,  Cs is the compound's water solubility, and MW is its molecular
weight. The Freundlich K can be used to judge the relative strength of adsorption for a specific compound.
As can be seen, if a water has constituents with low solubilities, the capacity of the carbon for that con-
stituent will be high. Also, a lower temperature and higher Mw will result in greater adsorption (assuming
no pore blockage). Crittenden et al. (1999) has further evaluated adsorption correlations.

Isotherm data for specific contaminants are valuable because kinetic models have been developed to
predict full-scale results from isotherm data. Speth and Miltner (1990, 1998) contain isotherm data for
numerous DBFs. However, kinetic models are not useful for complex mixtures such as NOM or DBF
precursors. Although attempts have been made (Luft 1984; Crittenden et al. 1987b; Speth 1986; Warta
et al. 1995; Hong 1985), it is very difficult to predict NOM adsorption.

The logistic function (Equation 9-2) may be the best  way to predict GAC performance for the adsorp-
tion of NOM. The logistic function's adsorption parameters were determined by fitting the break-
through profiles  from Jefferson Parrish,  LA. These NOM breakthrough profiles had moderate
adsorbability as compared to  other available data.

Initial Concentration

Initial concentration is a function of the water's  source, although it  can be affected by pretreatment
processes. Different pretreatments will also affect the adsorbability and biodegradability of the NOM.
As mentioned earlier, chlorination has a detrimental effect on the control of DBFs primarily due to the
lower adsorption capacity of GAC for formed DBFs and detrimental surface reactions with the acti-
vated carbon.  Semmens et al. (1986b) showed that alum coagulation resulted in improved GAC run
times for TOC and THM precursor removal because of reduced initial concentration and the removal of
poorly adsorbed high-molecular-weight organics. Hooper et al. (1996) found that enhanced coagula-
tion reduced the concentration of the NOM, but also  increased its adsorbability due to the reduced pH
imparted from the increased alum dose.

Ozonation makes the NOM more polar,  and hence less adsorbable, but it also increases its biodegrad-
ability (Sontheimer and Hubele  1987). Therefore,  for short beds that are  regenerated frequently,
preozonation would not be  helpful, whereas for longer beds that are infrequently regenerated,
preozonation would be a benefit. The reader is referred to other chapters which discuss preozonation's
effect on subsequent chlorination reactions (Chapter 5) and preozonation's effect on biodegradation
(Chapter 6).
                                              9-7

-------
Benz et al. (1992) showed that anion exchange pretreatment can significantly improve the performance
of GAC columns. The anion exchange columns extended the GAC performance by a factor of two to
three for NOM and DBF precursors. The resin removed the hydrophilic weakly adsorbing NOM that
was not amenable to GAC treatment.

Finally, inorganic precipitation can foul GAC. Coagulants can result in an oversaturation of calcium
carbonate or iron hydroxides. Also, if air is introduced into an anaerobic ground water, iron and manga-
nese precipitation can coat the carbon, reducing its capacity and slowing film transfer kinetics (Speth
1991).

EBCT

As can be seen in Equation 9-1, EBCT is an important parameter for adsorption performance in GAC
columns. The greater the EBCT, the longer the column will remain operational. Figure 9-2 shows the
cis-dichloroethene breakthrough profiles for two different EBCTs. As can be seen, the breakthrough
profile for the longer EBCT occurs later than the shorter EBCT, as expected. However, doubling the
EBCT generally does not necessarily double the run-time length. Summers et al. (1995) showed that
columns with EBCTs of 10 and 20 minutes had similar carbon use rates for removing NOM. This
suggested that utilities should attempt to minimize the EBCT to reduce capital costs. Wiesner et al.
(1987) found that 6 to 12 minutes of EBCT was optimal for TOC removal.

Lykins et al. (1988a) showed that the most cost-effective EBCT will depend on the source water. Figure
9-6 shows the plot of costs versus EBCT. The costs for Jefferson Parish and Cincinnati water did not
vary greatly with EBCT, whereas Miami water shows a clear advantage of longer EBCTs. Because of
the tradeoff between cost of replacement/regeneration and the increased capital cost of the larger col-
umns, EBCTs typically range somewhere between 10 and 20 minutes. As with the other design factors,
the total cost will be a function of how the column effluents are blended together. This issue will be
discussed later in this chapter.
               140
               120
             "ro
             O)
             § 100
                80
             CO
             O  60
                ,0
                20
Miami - 60 mgd (227 ML/d)
Jefferson Parish - 70 mgd (265 ML/d)
Cincinnati - 235 mgd (889 ML/d)
                              10    15     20     25     30    35     40
                                          EBCT- min.
                                45
Figure 9-6. Costs versus EBCT for three waters (Lykins et al. 1988a).
                                             9-8

-------
                                 Diffusion Mechanisms   Model Mechanisms
                                    Local Equilibrium
                                 between Fluid Phase and
                                    Adsorbent Phase
                    Bulk
                   Solution
                        Fluid Phase     Adsorbent Phase
                    Mass Flux = kf(Cb-Cs) Mass Flux =-DsCa ~ -Rj
                                               or   *p
Figure 9-7. Adsorption kinetics (Crittenden et al. 1987c).

Adsorption Kinetics

Adsorption kinetics also influence the shape of the adsorption profile. Mechanisms of adsorption
kinetics include film diffusion, pore diffusion, and surface diffusion. Figure 9-7 shows the adsorp-
tion kinetics around and within a carbon particle (Crittenden et al. 1987c). Kinetic models must
account for external (film transfer) and internal (pore and surface diffusion) effects. Slow kinetics
will result in a flat breakthrough profile, whereas fast kinetics will result in a steep profile. If the
kinetics of adsorption are instantaneous, the breakthrough profile for a single compound will be a
step function as shown in Figure 9-8  (Speth  1990). For NOM breakthrough, the breakthrough
profile will be flattened even if the kinetics of adsorption are instantaneous. This is because NOM
is made up of countless different types of compounds that will breakthrough at different times. In
this case, the breakthrough profile will be a series of infinitely small  steps.
                 o.o
                                              Time
Figure 9-8. SOC Breakthrough profiles assuming slow and instantaneous kinetics.
                                               9-9

-------
The size of GAC particles used in full-scale columns varies within standard ranges. Postfilter carbon is
typically 12 x 40 mesh (1.68 by 0.42 mm), although other sizes are also on the market. These ranges
will not, however, produce markedly different adsorption results in a postfilter adsorber. The choice of
carbon size is often determined by the smallest size that does not create head loss problems because
smaller carbon particles will have faster kinetics, resulting in more efficient bed adsorption. For postfilter
adsorbers, this is typically 12 x 40-mesh carbon.

Crittenden et al. (1986, 1987a) developed the Rapid Small Scale Column Test (RSSCT) to help predict
full-scale adsorption data. The RSSCT uses smaller particle sizes to obtain full-scale results in a frac-
tion of the time needed to run a pilot column. For example, a RSSCT can be completed in 1 month and
be representative of a year-long full-scale study. Therefore, seasonal studies can be conducted without
having to account for major changes in influent water quality. Scale-up approaches were developed for
a number of diffusivity  assumptions.  Summers et al. (1995) found that the proportional diffusivity
approach developed by Crittenden et al. (1987a) was the most appropriate for TOC breakthrough pro-
files. However,  although useful for planning, the RSSCT is only for pilot-scale design and does not
hold practicality for full-scale systems (Speth and Miltner 1989; Speth et al. 1989).
NOM is predominantly negatively charged. Therefore, decreasing the pH renders the predominantly
negatively charged organic molecule more neutral. A neutral compound is inherently less soluble in
water than a charged molecule and, therefore, more adsorbable. Also, at low pHs, NOM is more coiled
due to a less negative-charge repulsion. This may allow for greater access to GAC pores.  Semmens et
al. (1986a) showed that a lower pH will result in better adsorption. Figure 9-9 shows TTFDVIFP break-
through profiles for the same water at different pHs. The lower pH system had a longer bed life. Hooper
et al. (1996) also concluded that lower pHs improved NOM/precursor adsorbability.
                      o pH 5.0
                      A pH6.1
                        pH7.0
                      n pH 8.7
                                        100
200
                                               Run Time - h
Figure 9-9. Effect of pH on TTHMFP adsorption (Semmens et al. 1986a).
                                             9-10

-------
Biogrowth and Dissolved Oxygen

As mentioned previously, biogrowth can improve GAC bed performance for removing DBF precur-
sors. Therefore, introducing oxygen to encourage biogrowth can result in greater biological removal of
DBF precursors.  Dissolved oxygen also has shown to increase the adsorption of NOM  due to their
polymerization onto the carbon surface (Warta 1993; Warta et al. 1995). Figure 9-10 shows an example
of greater NOM  adsorption at high dissolved oxygen levels (Warta et al. 1995). This utility that is
drawing anoxic water might consider aerating the water prior to adsorption for maximum NOM re-
moval, although secondary effects such as iron precipitation would have to be taken into account. This
increased NOM removal lowers the performance of a column that is treating a specific organic com-
pound that does not polymerize on the GAC surface because the polymerizing NOM competes against
the organic compound more effectively for adsorption sites (Vidic et al. 1992; Serial et al. 1994a; Serial
et al. 1994b; Cerminara et al. 1995). Interestingly, although oxic polymerization significantly increases
the removal of dissolved oxygen content (DOC), Warta (1993) showed that oxic polymerization only
slightly increases the removal of THMFP and TOXFP. Apparently, the fractions of NOM that polymer-
ize on the carbon surface are not precursor material. Warta (1993) did show that air-saturated oxygen
levels can significantly increase the removal of THMFP and TOXFP over anoxic and super-saturated
oxygen systems. This is presumably due to microbiological processes.

Biological growth occurs  in every GAC column that treats drinking water.  This  is even true for
prechlorinated waters because disinfectants are reduced in the top few centimeters of a  carbon bed.
There are  advantages to maintaining biological activity in GAC beds. As mentioned, bioactivity can
improve the DBP-precursor-removal performance of GAC columns (Miltner et al. 1994; Miltner et
al. 1996; Wang et al. 1995; Warta 1993). Because GAC columns  are biologically active, there is
typically an increase in the concentration of microbes in the effluent as compared to  the influent
(Symons et al. 1981). Cold-water systems were found to be the exception, most likely due to the
       o
       O
       "33
       O
         1.2
         1.0
         0.8
0.6
         0.4
         0.2
         0.0

                                                                    v Anoxic
                                                                    o Oxic
                                                                    n Ambient
Doo°
                    00
             Oooo_nn°
        V      o°
            ° o
       '  #ao
     V  D
   , V  n   O
   V rP CP
       o
   ,D 0°
   JO
            0246
                                10     12     14    16    18    20    22    24
                                  Time, Days
Figure 9-10. TOC breakthrough profiles under anoxic, oxic, and ambient conditions
            (Warta 1993).
                                             9-11

-------
inhibitory effect of low temperatures on microbial growth. Symons et al. (1981) found that adequate
post-GAC disinfection produced waters of acceptable quality. Camper et al. (1987) found that GAC
columns released disinfection-protected particles that were highly populated with heterotrophic plate
counts (HPCs) and coliform bacteria. Camper et al. (1987) concluded that increased bed depth, higher
applied-water turbidity, and increased filtration rate exacerbated the bacteria-laden particle problem.

In cases where there is no postfilter disinfectant, such as in-home treatment devices, biological activity
is definitely not desired. Silver impregnation has been used to reduce biogrowth, however, Reasoner et
al. (1987) found that point-of-use GAC devices that contained silver had concentrations of heterotrophic
bacteria as high as units that did not contain silver. Also, it is  expected that fouling will further reduce
the biocidal properties of the silver. Aside from using preoxidants and adjusting the rate of replace-
ment, or regeneration, a water utility has little control over biodegradation other than to increase it. As
previously mentioned, oxidants  such as ozone can increase the biodegradability of the NOM in the
water. A carbon column that is replaced, or regenerated, frequently (every 1 to 2 months) will not show
significant biodegradation. Therefore, it is generally beneficial to lengthen the run time  of each col-
umn, such as by blending the effluents of multiple columns. The reader is referred to Chapter 6, "Alter-
native Disinfectants," for a more in-depth discussion of biological filtration.

Other Effects

Backwashing is sometimes required to remove particulates from GAC filters. Hong (1985)  showed
that backwashing had little impact on the removal of TOC and THM precursors for five natural waters.

Because adsorption is an exothermic reaction, greater adsorption will occur at lower temperatures. This
can be seen in Equation 9-5. Therefore, utilities find better TOC removal in the winter months. This is
only one example showing that seasonal effects will change the adsorption characteristics of DBF
precursors. Another seasonal change includes changes in the NOM characteristics because of algal and
microbial production during the  summer months. Neither temperature nor NOM quality  is an adjust-
able parameter for a water utility.

Bromide Issues

The prediction of brominated  DBF breakthrough from  GAC columns is more complicated than for
certain chlorinated DBFs. Brominated THMs are often found to be elevated in GAC effluent water
(Symons et al. 1981). The formation of brominated DBFs is  a function of the bromide-to-NOM ratio
and the bromide-to-chlorine ratio (Summers et al. 1993). (Refer to Chapter 2 for a speciation discus-
sion.) Because GAC columns remove NOM but do not remove bromide, the bromide-to-NOM ratio in
the column effluent will be constantly changing. Early in the column  run, the bromide-to-NOM con-
centration will be very high. This will favor the formation of brominated DBFs over chlorinated DBFs.
Eventually, the NOM concentration in the effluent will increase, resulting in higher chlorinated DBFs.
Because of this effect, the precursors to the chlorinated DBFs will appear to be removed better than if
no bromide was present. Also, and more importantly, the concentration of brominated DBFs may in-
crease quickly and then later drop. Figure 9-11 shows the breakthrough of SDS-bromoform (USEPA
1999). The effluent peaks were much higher than the influent SDS-bromoform concentration due to
excess bromide to TOC in the  effluent. Eventually, the remaining TOC breaks through, and the SDS-
bromoform returns to influent  levels.

Blending

Greater  GAC capacity can be realized if columns are run in parallel or  in series. If the start times of the
columns are staggered, the first column can be run beyond the maximum total effluent concentration
because its effluent is being blended with columns in the start of their breakthrough profiles. Hence, the


                                            9-12

-------
                          SDS-BF
                          EBCT = 20 min
                             = 3.3|ig/L
                                                 D  Single Contactor Effluent
                                                 	Logistic Function Best Fit
                                                       = 0.978)
                                                 O  Blended Effluent
                                                 	 Dl Prediction (RSS = 3.47)
                                       100     150     200     250
                                     Scaled Operation Time (days)
300
Figure 9-11. SDS-Bromoform breakthrough with model predictions (USEPA 1999).
total blended effluent still remains below the design effluent concentration. Roberts and Summers
(1982) developed a strategy to calculate the benefit of a parallel approach.

For columns in series, the first column can be run to exhaustion, while the TOC mass transfer zone, or
breakthrough profile, is contained in the latter columns. When the first column reaches its maximum
operational capacity, it can be taken out of series, and the column can be restarted as the last column
with fresh GAC. This approach may be best for SOC removal when the breakthrough profile is rela-
tively sharp. The results of Summers et al. (1995) seem to suggest that the NOM breakthrough profiles
are too broad for series operation.

Membranes
Membrane Type

Membranes remove organic compounds, inorganic compounds, and particulates by creating a barrier
through which water is preferentially  passed. The rejection can be physical or electrochemical. There
are many types of membranes that are manufactured, with each having different stability, flux, and
rejection characteristics. The type of membranes studied at the EPA were strictly pressure-driven pro-
cesses such as RO, NF, UF, and MF because these types of membranes have the most applicability to
drinking water systems. Membranes with other driving forces such as electric potential and tempera-
ture have not been studied. As stated earlier, both RO andNF membranes are fully capable of rejecting
DBF precursors; however, higher fluxes from the NF systems make them generally more economical
than RO membranes. Although there may be instances where RO membranes have site-specific advan-
tages over NF membranes (i.e., brackish and sea waters), the following discussion will concentrate on
NF. The EPA has conducted the majority of its membrane research with NF membranes because of the
need to determine the potential for lowering the current maximum DBF levels in drinking water in a cost-
effective manner.
                                             9-13

-------
Although a NF membrane is inherently a barrier process, membrane manufacturers are reluctant to
claim log-removal credit for pathogens due to the potential of glue-line failure. Also, many plants blend
pretreated feed waters with the membrane-treated waters to create a less corrosive product water. This
practice results in limiting the microbial removal for NF and RO membranes. Therefore, although
particulate/microbial removals may be given forNF membranes, this should not be construed as a basis
for log removal credit.
Both UF and MF systems are capable of excellent particulate/pathogen removal; however, their DBF
removal performance is more consistent with that seen for conventional treatment (10 to 50 percent). Due
to the applicability and the economies of scale that favor using UF and MF for small systems, refer to
Chapter 11, "Controlling Disinfection By-Products (DBFs) and Microbial Contaminants in Small Public
Water Systems (PWSs)," for an additional discussion of these technologies.
Membranes are made of many materials. Common membrane materials are polysulfone, cellulose  ac-
etate, polycarbonate, polypropylene, polytetrafluoroethylene, and polyacrylonitrile. These materials can
be classified as hydrophobic or hydrophilic. Membranes made of polyamide, polysulfone, polypropy-
lene, or polytetrafluoroethylene are generally considered hydrophobic membranes. Membranes made
of cellulose acetate or polyacrylonitrile are generally considered hydrophilic. Hydrophilic membranes
do not foul as fast as hydrophobic membranes (Laine 1989; Bonner and O'Melia 1991; Eykamp 1978).
However, the anti-fouling behavior of hydrophilic membranes is offset by their limitations with regard
to pH, chemical, and temperature resistance. This is important for drinking water applications because
membrane cleaning is often completed at extreme pHs.
Membranes can be of homogeneous, asymmetric, or composite construction. Composite membranes
are most popular due to the thin separation layer that allows for maximization of water flux. The Desal-
5, Film-Tec NF-70, Film-Tec NF90, and Fluid Systems' TFCS membranes are of composite thin film
design. The film layers on these membranes are of polyamide construction. Reiss et al. (1999a) found
that for NF membranes of comparable productivity, NOM rejection was greater for the polyamide
versus cellulose acetate membranes. Membranes typically carry a charge. Polyamide, polysulfone, cel-
lulose acetate, and ceramic membranes carry a negative charge (Lahoussine-Turcaud et al. 1990). The
charge of a membrane can have implications regarding the rejection of charged, dissolved species.
Along with hydrophobicity and charge, surface roughness, porosity, pore size, and membrane consis-
tency can also affect  the extent of fouling (Marshall 1993). The  greater the surface roughness,  the
greater the fouling. Also, the greater the pore size, the greater the paniculate fouling. Given the same
operating conditions, MF membranes with large pores will foul to a greater extent than MF membranes
with smaller pores.

System Configuration

Full-scale membrane elements are designed in a number of ways to optimize membrane area to element
size. The types of membrane vessels include spiral wound, hollow fiber, shell in tube, and rotating disk.
Spiral-wound and hollow-fiber systems provide better packing densities compared to shell-in-tube and
rotating-disk units. RO  and NF elements are commonly of spiral-wound configuration, whereas MF
and UF are typically hollow fiber. Figure 9-12 shows a diagram of a spiral-wound vessel (Speth 1998).
The membrane material is glued into envelopes with the active film to the outside. Inside the envelope,
a permeate spacer sheet is added to allow for the permeate water to flow. The open edge of the envelope
is attached to a permeate collection tube. Therefore, any permeate water inside the  envelope will, by
small pressure gradients, be transferred to the permeate collection tube.  The number of envelopes
attached to the permeate collection tube varies by element size and manufacturer. A feed-spacer sheet
is placed in between the envelopes. The envelopes and the feed-spacer sheets are rolled into a cylinder.
The permeate collection tube is plugged at one end. The other end of the permeate collection tube is
                                            9-14

-------
                                               	  Concentrate
                   Feed Flow -

                     Permeate
                            Permeate
                            Collection               Pressure
                              Tube                  Vessel
                                \
                        Glue Line •
                                 j/X                  It
                                         	             Feed Spacer
                  Membrane

                                                          Permeate Spacer
Figure 9-12. Typical spiral-wound membrane element (Speth 1998).
plumbed into a permeate collection line. The feed line to the element is introduced on one end of the
membrane cylinder. The feed water runs through the feed-spacer sheet, past the membrane material on
either side. At the end of the element, the concentrated feed water, typically referred to as the concen-
trate, or retentate, is plumbed into a line that is either sent to the next element in series, sent to waste, or
recycled back to the feed stream.

Each NF element typically passes 10 to 15 percent of the feed water through the membrane as perme-
ate. The remaining 85 to 90 percent of the feed water leaves the element as reject water. Percent recov-
ery is used to give an indication of the percentage of the water that is produced.

                         % Recovery = (permeate flow/feed flow) x 100              (9-6)

Because the amount of concentrate in a single element is typically 85 percent of the feed stream flow,
wasting the  entire concentrate is not cost-effective. Also, in water plants, the amount of permeate
water needed for consumers  is large. Therefore, many elements are required to provide the requisite
membrane surface area to  generate the needed permeate flow. Often the membrane pressure vessels
are arrayed in formations in which the concentrate of the first row of elements is fed to the second row
of pressure vessels. Because of the loss of permeate water, the second and further rows of pressure
vessels are fewer in number than the previous row. This maintains the cross-flow velocity across the
membrane surface. Figure 9-13 shows a typical staged array (Speth 1998). Also, to  save money on
pressure  vessels, three elements are usually placed in a single pressure vessel. For testing and re-
search purposes, often only one element is used, with the concentrate stream being recycled to the
feed stream  to increase the recovery level to that associated with a staged array (typically 70 to 90
percent).

Allgeier and Summers (1995) developed a bench-scale technique using a 4x6-inch flat-plate system to
predict the performance of full-scale systems. Although limited in long-term performance, researchers
have shown that it can predict rej ection, initial flux, and cleaning intervals for full-scale systems (Allgeier
et al. 1996; Allgeier et al. 1997; Gusses et al. 1996; Speth et  al. 1996; Speth et al. 1997; Gusses et al.
1999).
                                              9-15

-------
                                             By-pass
                     Qb
           Pretreatment
cb
                         Qf
                                                     Permeate
           Q = flow rate
           C = concentration
           f = feed
           p = permeate
           w = waste
           b = by-pass
                                               Waste
Figure 9-13. Typical membrane staged array (Speth 1998).
Performance
Rejection

Fronk and Lykins (1998) and Lykins and Clark (1994) showed that RO membranes rejected between 0
and 95 percent of the individual THM species. The thin-film composite membranes had superior rejec-
tion properties compared to the other membranes tested. Because precursor material is easier to reject
than DBFs, NF membranes are typically only used for precursor removal.

Taylor et al. (1987) found thatNF membranes rejected over 90 percent of the DOC in natural waters.
Allgeier and Summers (1995) showed that for five natural waters,  67 to 94 percent of the TOC was
rejected.  Taylor et al. (1987) found that membranes with nominal molecular-weight cutoffs of 400
daltons or less were needed to control DBF precursors. This was confirmed by Taylor et al. (1989).
Blau et al. (1992), Watson and Hornburg (1989), and Reiss et al. (1999a) also found that THM precur-
sors could be controlled with NF. Allgeier and Summers (1995) showed that NF membranes removed
TFDVI precursors by 66 to 93 percent, haloacetic acid (HAA) six (HAA6) precursors by 67 to 97 per-
cent, as well as precursors for other DBFs. In general, TOC rejection was a good surrogate  for DBF
precursor control. Speth (1998) also saw good rejection of DBF precursors for a membrane system fed
conventionally treated Ohio River water (CT-ORW) for 15 months, as shown in Table 9-1.

Fouling and Flux Curves

The loss of membrane efficiency due to fouling is one of the main impediments to the development of
membrane processes for use in drinking water treatment. Membrane fouling is dependent on the water
quality as well as the membrane's properties and construction. In  general, fouling is defined as the
accumulation of material on the surface, or in the pores, of a membrane that decreases the water flux
through the membrane. Durham (1993) further distinguishes between membrane  fouling and spacer
fouling. The consequences of fouling can be severe; fouling can reduce the water flux through a mem-
brane up to 90 percent (Belfort 1977).
                                            9-16

-------
Table 9-1. Organic and Particulate Bulk Rejections for Pilot Systems Fed CT-ORW
          (Speth 1998)
Bulk Rejections (%)
Parameter
TOC, mg/L
UVA, 1/m
Term. THM, ug/L
SDS THM, ug/L
Term. TOX, ug/L
SDS TOX, ug/L
Turbidity, NTU
Particle counts, #/ml
HPC, CFU/ul
Aerobic spores, CFU/L
Mean Feed Cone.
1.99 (0.41,43)
4.54 (1.39,46)
224 (76,6)
97 (49,6)
265 (101,6)
153 (116,6)
0.23 (0.08,64)
882 (265,18)
33.5 (27.4,25)
100(NA,1)
4" x 40" System
94.8 (3.6,43)
97.9 (1.2,46)
95.6 (0.6,6)
97.1 (1.1,6)
95.8 (0.8,6)
95.9 (3.1,6)
72.7 (11.8,64)
99.5 (0.4,18)
96.9 (4.7,25)
81.8 (NA,1)
1.75" x 12" System
95.9 (3.3,43)
98.5 (1.4,46)
96.1 (0.7,6)
96.1 (4.9,6)
96.5 (1.4,6)
96.5 (4.9,6)
84.8 (10.9,64)
99.4 (0.5,18)
98.9(3.1,25)
80.0 (NA,1)
  Standard deviations and number of samples are listed in parentheses, respectively.
  NA = Not applicable
There are five broad fouling categories: microorganisms, colloidal or particulate matter, dissolved or-
ganics, sparingly soluble inorganics, and chemical reactants (Belfort 1977; Matthiasson and Sivik 1980;
Potts et al. 1981; Barger and Carnahan 1991; Suratt 1993). Dissolved organics and colloidal matter are
considered to be serious foulants due to the difficulty in removing them with pretreatment processes.
Biofouling is also a serious and common fouling problem (Paul 1991). Inorganic fouling can often be
controlled by acid addition. For instance, the pH needed to control calcium carbonate precipitation can
be determined by a Langelier saturation index. Often, the type of foulant is operationally defined by the
type of membrane cleaning agent that is effective in recovering water flux. Inorganic foulants are typi-
cally removed with acid solutions, whereas organic and biological foulants are typically removed with
alkaline/detergent solutions.

Direct methods for determining the nature  of the foulant include optical microscopy, scanning
electron microscopy, X-ray fluorescence,  atomic absorption, transmission and reflection infrared
spectroscopy, energy  dispersive X-ray, electron  spectroscopy for chemical analysis (ESCA),
pyrolysis-gas chromatograph/mass spectroscopy (GC/MS), and phospholipid analyses. Optical
microscopy is  useful for identifying color, size, crystalline structure, and  refractive  indices of
foulants. Scanning electron microscopy can further refine the results determined with optical mi-
croscopy. Energy dispersive X-ray analysis and ESCA will give information regarding the
elemental composition of the foulant. Infrared spectroscopy can qualitatively fingerprint the foulant
in terms of functional groups and chemical structure, while pyrolysis-GC/MS can fingerprint the
foulant in terms of biopolymer groupings.

For NF membranes, foulant precipitation is exacerbated by concentration polarization. Concentration
polarization occurs when the convective flow of foulants toward the membrane  surface is greater than
the diffusional flow of foulants to the bulk solution.  This only  occurs before steady state is achieved.
The concentration of foulants will therefore increase near the membrane until steady state is reached.
Figure 9-14 shows a schematic of membrane  permeation and rejection for RO, NF, and MF systems
(Speth 1998). For RO and NF, the elevated concentration  of foulants near the membrane surface is
often the causative agent for fouling due to precipitation. In dead-end cells, where the concentrate is not
removed from contact with the membrane, steady state is not achieved. The continuous accumulation
of the components that are rejected by the membrane in the dead-end cells maximizes precipitation and
fouling.
                                             9-17

-------
             Reverse Osmosis           ,   x „     _  _ 0   ,   «	v permeate

                             Feed l=!^oy^roVr<^""0 /V^^ry :^"0V"o^H^S> Concentrate
                                       °/0°0  0 OQ  o O  °.
             Ultrafiltration               ,  Ox-0-  -  _-_ ^-0/   r-t	v Permeate

                                                                     Concentrate
                                  Particulates
Figure 9-14. Schematic of membrane permeation and rejection for RO, NF, and MF systems
            (Speth 1998).

MF and UF membranes do not have the same type of concentration polarization that NF and RO mem-
branes experience because of the lower rejection of dissolved constituents, as shown in Figure 9-14.
However, UF membranes have a sparse membrane porosity that results in a local polarization phenomena
that may be much greater than the average polarization (Fane and Fell 1987).

Figure 9-15 shows a flux decline curve for NF membrane elements fed CT-ORW (Speth et al. 1998).
The  fluxes show a  distinct seasonal trend, with the greatest overall flux decline occurring in the
summer and fall seasons. During these times, weekly cleaning with acid only partially offset the lost
in flux due to fouling. This frequent cleaning would be difficult for a water utility to justify, but it
points to the difficulty of using NF on a surface water, even a surface water such as this one with a low
organic  concentration (average TOC = 2.0 mg/L). Taylor et al. (1992) also have evaluated fouling
from surface- and ground-water sources. They found that membranes used to treat surface waters
have much higher fouling rates than membranes used to treat ground waters.

Speth et al. (1998) found that the foulant from pilot NF membrane elements fed CT-ORW was a film
layer 20 to 80 urn thick, with the greatest depth in the first of three elements in series. Heterotrophic
plate counts, phospholipid analyses, and pyrolysis-GC/MS analysis of the foulant showed it to be domi-
nated by biological growth. This explains the flux declines in the summer months.

Flux Loss Mitigation

Improvements of fluxes can  be related to flow and solution characteristics, membrane material, and
pretreatment (Van den Berg  and Smolders  1990). With regard to equipment design, there are many
things that can be done to reduce fouling (Marshall et al. 1993). Membrane manufacturers incorporate
many fouling-reducing features into their element construction.

Membranes can be fouled either by adsorption or cake-layer formation. Adsorption involves materials
attaching within the pores, or matrix, of the membrane. This type of fouling is generally considered irre-
versible, although chemical cleaning can be effective in some cases (Belfort 1977). Cake-layer fouling, or
                                             9-18

-------
               6-
               5-
             CO
            €  4H
             E
             CD
             Q.
            CO
               3-
                1 -
                                                               •  Element #1
                                                               n  Element #2
                                                                  Element #3
LT>
O>
•O
                                (0
CD
JD
E
CD
                                                      .a
                                                      CD
                                             >s
                                             co
CD
O)
O
"to
^
o)
                               I   '   '  '   '  I   '   '  '   '  I
                              100          200          300
                                        Time (Calendar Days)
                                                 400
Figure 9-15. Flux decline curve for NF membrane elements fed CT-ORW (Speth et al. 1998).
reversible colmatage, involves the accumulation of material on the membrane surface. Cake-layer fouling
is thought to be controlled by adjusting hydraulic operating parameters, such as cross-flow velocity or
recovery of the systems. Cross-flow velocity is the velocity of the water in the feed channel, whereas
recovery is the percentage of permeate flow to the total feed flow.

System recovery will dictate the concentration of dissolved species on the feed side of the membrane.
The increased concentration of dissolved species on the feed side of the membrane increases the like-
lihood of precipitation of inorganic foulants such as calcium carbonate  or barium sulfate. Organic
contaminants also may be destabilized in the concentration layer. Reducing the system recovery can
lead to less fouling.

The fluid pressure can also be manipulated to limit fouling. A higher pressure above the membrane will
result in a greater flux of water through the membrane. Many researchers have found that the higher the
initial flux, the greater the flux decline (Gutman  1977; Wilkes et al. 1996; Hong and Elimelech 1997;
Chellametal. 1998).

Changing the operating conditions can limit the amount of irreversible fouling. Crozes et al. (1997)
found that increasing the cross-flow velocity, reducing the flux, and increasing the backwash frequency
limited the short-term reversible fouling. This also controlled the rate of flux loss due to irreversible
fouling.

As mentioned previously, the choice of membrane type will also determine the amount of fouling.
Elimelech et al. (1997) and Zhu and Elimelech (1998) found that there was a significantly higher
fouling rate for a thin-film composite membrane compared to that for a cellulose acetate membrane.
The higher fouling rate for the thin-film composite membrane was attributed to greater surface rough-
ness, which is inherent in interfacially polymerized aromatic polyamide composite membranes. Atomic-
force microscopy and scanning-electron microscopy images supported the conclusions. The greater
                                              9-19

-------
surface roughness would allow for greater depositions of colloids. Ridgway and Safarik (1991) found
that biofilms attached more strongly to polyamide surfaces than to cellulose acetate surfaces. They
attribute the greater fouling rate of polyamide membranes to biological growth.

Along with fluid flow characteristics and membrane type, fouling can be reduced by optimizing feed-
water pretreatment. This may entail adjusting the pH, adding anti-sealant agents, or using a pretreatment
unit process such as coagulation, filtration, MF, and/or GAC. Typically, acid addition and anti-sealant
agents are used to limit inorganic fouling. Numerous researchers have looked into pretreatment processes
to reduce membrane fouling (Laine 1989; Taylor et al. 1989; Wiesner et al. 1989; Reiss and Taylor 1991;
Taylor et al. 1992; Amy et al. 1993; Solomon et al. 1993;  Chellam et al. 1997; Gusses et al. 1997).
Processes used by these researchers were MF, coagulation/sedimentation, filtration, GAC, ozonation, and
biologically active sand columns. Solomon et al. (1993) and  Gusses et al. (1997) demonstrated that bio-
logical-filtration pretreatment showed the most promise with regard to reducing flux decline during UF.
Chellam et al. (1997) showed that MF and UF pretreatment of a conventionally treated water gave signifi-
cantly lower NF fouling rates than just conventional treatment.

For organic fouling of NF membranes, the greater the amount of hydrophobic compounds in a feed
water, the greater the degree of hydrophobic adsorption, or fouling. Also, the amount of high-molecu-
lar-weight organic matter correlates to greater amounts of cake-layer formation. Carbon adsorption can
be an effective, but expensive way, to control hydrophobic organic fouling.

In a follow-up study to the 15-month  membrane project shown in Figure 9-15, Speth et al. (2000)
evaluated five different additional pretreatments to CT-ORW. The chosen additional pretreatments
were intended to produce waters with varying biological-fouling potential. Five parallel membranes
were fed CT-ORW, ozonated CT-ORW, ozonated/biofiltered CT-ORW, CT-ORW reduced to 7°C,
and chloraminated CT-ORW. All  systems showed significant flux decline, indicating that methods
beyond those needed for just biogrowth control are required for NF systems treating conventionally
treated surface waters. The NF systems fed ozonated, ozonated/biofiltered, and untreated CT-ORW
had the least amount of flux decline  over the course of  the study;  however, they  had  significant
amounts of biological growth. Fouling in these systems was attributed to the deposition of extracellu-
lar material (polysaccharides) in the cake layer, either from the biogrowth on the membrane or carryover
from the pretreatment. The low-temperature system had greater flux decline, but it had lower biogrowth
than the ozonated, ozonated/biofiltered, and untreated CT-ORW systems. Although lower in biogrowth,
the deposited organic material in the low-temperature system still showed a strong biological signa-
ture (polysaccharides and aminosugars). The chloraminated system had the greatest flux decline, but
the least amount of biogrowth. The organic material deposited in the chloraminated system showed a
high level of proteinaceous character.

As mentioned previously, particulates and colloids can be major foulants. While ground waters gener-
ally do not cause concern because of the low parti culates found  in ground waters, surface waters re-
quire at least conventional treatment to remove parti culates to acceptable levels. Typically, if the tur-
bidity is less than one NTU, it is acceptable for membrane feed water (Potts et al. 1981).

In a comparison of conventionally pretreated surface water to riverbank-filtered water, Merkel et al.
(1998) and Speth et al. (1999) found that riverbank-filtered water had significantly less flux decline
than conventionally pretreated river waters. In essence, riverbank filtration changes a surface water to
a ground water. Therefore, because ground waters are better source waters for NF treatment, riverbank
filtration should be an effective pretreatment for a utility that wishes to utilize membrane technologies.
Table 9-2 shows the percent of flux lost after 62 days for two watersheds (Speth et al. 1999). In both
cases, the flux lost was much lower in the bank-filtered systems.
                                             9-20

-------
Table 9-2. Flux Loss Comparison for Conventionally Treated and Riverbank-Filtration
           Pretreated Water (Speth et al. 1999)


Location
Louisville
Bank filtered
Louisville
Conv. treated
Cincinnati
Conv. treated
Southwestern Ohio
Bank filtered
Southwestern Ohio
Conv. treated


Water
Ohio River

Ohio River

Ohio River

Little Miami
River
Harsha Lake

Length of
Operation
(days)
62

79

460

79

70

% Flux
Mean Calculated Lost After
Number of Cleaning/ Cleaning Frequency Approx.
Flux Cycles
7

8

51

1

12

(days)
75*

36*

8

62

8

62 Days
4

46

36

12

50

  Includes initial cycle.
  * Not arithmetic mean, projected from slope of entire run.
Flux Recovery Through Cleaning

There are a number of techniques that can be used to recover flux from a fouled membrane. Typically,
acid cleaning is used to remove inorganic foulants. If the membrane is fouled with organic molecules
(extracellular material and microbes), a detergent has been found to be successful (Ridgway et al.
1985). Hong and Elimelech (1997) found that a strong chelating agent is effective for removing free
and NOM-complexed calcium ions from fouled membranes.

The type of membrane cleaner will be dependent on the membrane material. Some membranes, such as
cellulose acetate membranes, are not stable over wide pH ranges. Therefore, acid/base cleaning is not
practical for these membranes. Aromatic cross-linked polyamide membranes are generally incompat-
ible with nonionic polyoxyethylene w-oxide detergents  and cationic surfactants such as quaternary
ammonium compounds. In these cases, the cleaner causes rapid and irreversible flux loss.

NF membranes are typically cleaned whenever the permeate flux decreases to 85 percent of the initial
permeate flux or when the feed  pressure increases to 115 percent of the initial feed pressure for ele-
ments operated at a constant flux (Fu et al. 1994). The degree of membrane fouling, which dictates the
frequency of cleaning, will have a significant impact on the cost, design, and operation of full-scale
facilities. Wetterau et al. (1996) showed that the efficiency of overall water production was not affected
by either constant pressure or constant flux operation.

The cleaning of NF  and RO membranes is completed in normal flow modes. However, backflushing
the membrane has been evaluated (Breslau et al., 1980). Backflushing involves passing water from the
permeate side of the membrane to the concentrate side. Backflushing has been found to increase flux,
but it does not work well for removing proteins that are  strongly adsorbed. Also, backflushing is not
recommended for composite membranes because of the potential to destroy the thin film due to the lack
of a support layer on the feed side of the membrane.

Most flux decline occurs immediately after startup or chemical cleaning. Although difficult to discern in
Figure 9-15, each weekly  specific flux decline curve for the membrane-fed CT-ORW was nonlinear in
nature with a rapid initial specific flux decline followed by a less rapid specific flux decline. It was found
                                             9-21

-------
that the flux curves could be accurately fit by two linear slope segments covering the initial slope and the
final slope for a typical weekly flux decline curve (Speth et al. 1998). The initial flux declines were greater
than the final flux declines. The final flux declines are used to determine cleaning frequency.

Other Issues
The use of coagulants and powdered adsorbents within the membrane element can improve the final
water-quality of MF and UF membrane systems (Jacangelo, 1999). Coagulants and powdered adsorbents
can remove DBF precursors while being rejected by the membrane. The membrane system can increase
the mean residence time of the coagulant or powdered adsorbent, increasing their effectiveness. This is
not an issue for RO and NF membranes because they already have excellent rejection characteristics.

Bromide Issues
NF membranes reject bromide to a lesser extent than NOM or DBF precursors are rejected. Therefore,
as with GAC treatment, the effluent/permeate stream will have a higher ratio of bromide to DOC than
the feed stream. This will result in a greater percent of brominated DBFs than would normally occur.
However, as mentioned previously, typical NF membranes reject such  a high percentage of DBF pre-
cursors that the absolute amounts of brominated species reaching the consumer is very low.  The bro-
moform to chloroform ratio is high for the membrane treated water; however, there is very little of
each.

Summers et al. (1993) found that NF was effective for controlling the formation of chloroform, but it
increased the relative percentage of bromo-substituted compounds to chloro-substituted ones.  This was
the result of higher bromide-to-organic precursor levels in the permeate water. Jacangelo et al. (1993)
also found that NF reduced the amount of TFDVIs formed, but altered the speciation of THMs and
HAAs. Allgeier and Summers  (1995) found that the DBFs of membrane-treated water shifted to the
bromo-substituted compounds because of the preferential rejection of organic matter over bromide.

Blending

As mentioned previously, blending membrane product water with pretreated membrane feed water
often occurs because utilities with NF membranes wish to lower costs. Because NF can remove such a
high degree of precursors, significant blending is possible, even with strict DBF limits. Not  only is it
more economical to blend, but it helps make the product water less corrosive (Lytle et al. 1996).

Integrated Membrane Systems

MF and UF membranes can be effective surface-water pretreatments for NF and RO membranes (Tay-
lor et  al.  1992; Robert et al. 1999; Reiss  et al.  1999b). The combination of MF/UF membranes for
particulate pretreatment control and NF/RO membranes for organic control is referred to as integrated
membrane systems (IMS).  In some hard-to-treat waters, a coagulant  is needed prior to the MF/UF
pretreatment (Robert et al. 1999).

The advantage of using MF or UF membranes for NF/RO pretreatment is that the MF/UF membranes
will remove a sizeable percentage of pathogens. Reiss et al. (1999b) showed that IMS systems removed
from 5.4 to  10.7 logs of pathogen surrogate Bacillus subtilis. Robert et al. (1999) showed greater than
4 logs of Bacillus subtilis removal.  The disadvantage of using MF or UF membranes for NF/RO pre-
treatment is the high cost of installing two different types of membrane systems. Owen et al. (1999)
found that MF removed from 4 to 6 logs of spores, while MF/NF systems together removed 8 to 11 logs
of spores. Kruithof et al. (1998) found that UF membranes could remove greater than 5 logs of MS-2
phages.
                                            9-22

-------
Conclusions

Activated carbon is an effective process for removing DBF precursors. It is not designed for pathogen
removal. The effectiveness for precursor removal is dependent on a number of design issues such as
carbon type, filter location, filter depth, filter flow rate, and blending choices. It is also dependent on a
number of water quality issues such as initial precursor concentration, precursor adsorbability, precur-
sor adsorption kinetics, temperature, dissolved oxygen levels, pH, and bromide concentration. Although
much progress has been made on the modeling of GAC system performance, the applicability of this
technology  to any drinking water utility will need to be determined by pilot testing.

Membrane technologies are effective processes for removing DBF precursors. MF and UF membranes
are excellent for removing pathogens and particulates and can be used as a replacement for conven-
tional treatment. NF and RO membranes are excellent for removing DBF precursors, but are not counted
on as a pathogen barrier due to blending and glue-line-failure issues. Reducing the effect of membrane
foulants is a main consideration in the design of NF and RO plants, especially for plants treating sur-
face  waters. Although much progress has been made on modeling and scaling up small-membrane
results, the applicability of this technology will need to be determined by pilot testing.

References
Allgeier, S. C. and Summers, R.  S. (1995). "Evaluating NF for DBF control with the RBSMT."
   Journal of the American Water Works Association, 87(3), 87-99.

Allgeier S. C., Gusses, A. M., Speth, T. F., Westrick, J. 1, and Summers, R. S. (1996). "Verification
   of the rapid bench-scale membrane test and the ICR requirements." Proceedings, GAC, Mem-
   branes and the ICR: A Workshop on Bench-Scale and Pilot-Scale Evaluations, Cincinnati,  OH,
   March 3-5.

Allgeier, S. C., Gusses, A. M., Speth, T. F., and Summers, R. S. (1997). "Verification of the rapid
   bench-scale membrane test for use in the ICR." Proceedings, American Water Works Association
   Membrane Technology Conference, New Orleans, LA, February 23-26.

Amy, G., et al. (1993). "Membrane separation of DBF precursors form low-turbidity surface waters."
   Proceedings, American Water Works Association Membrane Technology Conference, Baltimore,
   MD, August 1-5, 651-664.

Barger, M. and Carnahan, R. P. (1991). "Fouling prediction in reverse osmosis processes." Desalina-
   tion, 83, 3-33.

Belfort, G. (1977). "Pretreatment and cleaning of hyperfiltration (reverse osmosis) membranes in
   municipal wastewater renovation." Desalination, 21, 285-300.

Benz, M. A., Summers, R. S., Haist-Guide, B., Koching, M., and Sontheimer, H. (1992). "Combined
   use of ion exchange resins and granular activated carbon for the control of organic matter and
   disinfection by-products. "Proceedings, American Water Works Association Annual Conference,
   Vancouver, B.C., June 18-22.

Blau, T. J., Taylor, J. S., Morris, K. E., and Mulford, L. A. (1992). "DBF control by nanofiltration:
   Cost and performance." Journal of the American Water Works Association, 84(12), 104-116.

Bonner, F. A and O'Melia, C. R. (1991). "Some aspects of the fouling of ultrafiltration membranes
   by natural organic matter in water treatment." Proceedings, American Water Works Association
   Membrane Technology Conference, Orlando, FL, March 10-13.
                                             9-23

-------
Breslau, B. R., Testa, A. J., Milnes, B. A., and Medjanis, G. (1980). Ultrafiltration membranes and
   applications., A. R. Cooper, ed., Plenum, New York, 109-128.

Camper, A. K., Broadaway, S. C., LeChevallier, M. W., and McFeters, G. A. (1987). "Operational
   variables and the release of colonized granular activated carbon particles in drinking water."
   Journal of the American Water Works Association, 79(5), 70-74.

Cerminara, P. J., Sonial, G. A., Papadimas, S. P., Suidan, M. T., Moteleb, M. A., and Speth, T. F.
   (1995). "Effect of influent oxygen concentration on the GAC adsorption of VOCs in the presence
   of BOM." Water Research, 29(2), 409-419.

Chellam, S., Jacangelo, J. G, Bonacquisti, T. P., and Schauer, B. A. (1997). "Effect of pretreatment on
   surface water nanofiltrati on." Journal of the American Water Works Association, 89(10), 77-89.

Chellam, S., Jacangelo, J. G., and Bonaquisti,  T. P. (1998). "Modeling and experimental verification
   of pilot-scale hollow fiber, direct flow microfiltration with periodic backwashing." Environmen-
   tal Science and Technology, 32, 75-81.

Clark, R.M., Symons, J. M., and Ireland, J. C.  (1986). "Evaluating field scale GAC systems for
   drinking water." Journal of Environmental Engineering Division of the ASCE, 112(4), 744-756.

Clark, R. M. (1987). "Modeling TOC removal by GAC: The general logistic function." Journal of
   the American Water Works Association, 79(1), 33-37.

Crittenden, J. C., Berrigan, J. K., and Hand, D. W.  (1986). "Design of rapid small-scale adsorption
   tests for a constant diffusivity." Journal of the Water Pollution Control Federation, 58, 312-319.

Crittenden, J. C., Berrigan, J. K., Hand, D. W., and Lykins, Jr., B. W. (1987a). "Design of rapid
   fixed-bed adsorption test for nonconstant diffusivities."'Journal of Environmental Engineering
   Division of the ASCE, 113(2), 243-258.

Crittenden, J. C., Luft, P.,  and Hand, D. W. (1987b). "Prediction of fixed-bed adsorber removal of
   organics in unknown mixtures." Journal of Environmental Engineering Division of the ASCE,
   113(3), 486-498.

Crittenden, J. C., Hand, D. W., Arora, H., and Lykins, Jr., B. W. (1987c). "Design considerations for
   GAC treatment of organic chemicals." Journal of the American Water Works Association, 79(1),
   74-82.

Crittenden, J. C., et al. (1999). "Correlation of aqueous-phase adsorption isotherms." Environmental
   Science and Technology, 33, 2926-2933.

Crozes, G. E, Jacangelo, J. G, Anselme, C., and Laine, J. M. (1997). "Impact of ultrafiltration operat-
   ing conditions on membrane irreversible fouling." Journal of Membrane Science, 124, 63-76.

Durham, L. A. (1993). "Factors that influence cleaning of RO systems." Proceedings, American
   Water Works Association Membrane Technology Conference, Baltimore, MD, August 1-5, 139-
   148.

Elimelech, M., Zhu,  X., Childress, A. E., and Hong, S. (1997). "Role of membrane surface morphol-
   ogy in colloidal fouling of cellulose acetate and composite aromatic polyamide reverse osmosis
   membranes." Journal of Membrane Science, 127, 101-109.

Eykamp, W. (1978). American institute of chemical engineers symposium series, 74, 233-235.

Fane, A. G. and Fell,  C. J. D. (1987).  "A review of fouling and fouling control in ultrafiltration."
   Desalination, 62, 117-136.
                                             9-24

-------
Fronk, C. A. and Lykins, Jr., B. W. (1998). "Removal of chlorinated and brominated alkanes from
   drinking water using reverse osmosis." Journal of Water Supply Research and Technology—
   Aqua, 47(4), 183-195.

Fu, P. , Ruiz, H., Thompson, K., and Spangenberg, C. (1994). "Selecting membranes for removing
   NOM and DBF precursors." Journal of the American Water Works Association, 86(12), 55-72.

Gusses, A. M., Allgeier S. C., Speth, T. F., and Summers, R. S. (1996). "Verification and use of the
   rapid bench-scale membrane test." Proceedings, American Water Works Association Annual
   Conference, Toronto, Canada, June 23-27.

Gusses, A. M., Speth, T. F., Allgeier, S. C., and Summers, R. S. (1997). "Evaluation of surface water
   pretreatment processes using the rapid bench-scale membrane test." Proceedings, 1997 American
   Water Works Association Membrane Technology Conference, New Orleans, February 23-26.

Gusses, A. M., Allgeier, S. C., Speth, T. F., and Summers, R. S. (1999). "Impact of membrane
   sample variability on the performance of the rapid bench-scale membrane test." Proceedings,
   American Water Works Association Membrane Technology Conference, Long Beach, CA, Febru-
   ary 2 8-March 3.

Gutman, R. G. (1977). "The design of membrane separation plant." The Chemical Engineer, July,
   510-513,521-523.

Hartman, D. J., DeMarco, J., Metz, D. H., Himmelstein, K. W., and Miller, R. (1991). "DBF precur-
   sor removal by GAC and alum coagulation." Proceedings, 1991 American Water Works Associa-
   tion Annual Conference, Denver, CO, June 7, 1995.

Hong, S. (1995). "Activated carbon adsorption of organic matter: Backwashing, desorption, and
   attenuation." Ph.D. dissertation, University of Cincinnati, Cincinnati, OH.

Hong, S. and Elimelech, M. (1997). "Chemical and physical aspects of natural organic matter
   (NOM) fouling of nanofiltration membranes." Journal of Membrane Science, 132, 159-181.

Hooper, S. M., Summers, R. S., Solarik, G.,  and Owen, D. M. (1996). "Improving GAC performance
   by optimized coagulation." Journal of the American Water Works Association, 88(8), 107-120.

Jacangelo, J. G., Cummings, E. W., P Cote', P.,  and Mallevialle, J. (1993).  "Influence of feedwater
   characteristics on ultrafiltration and nanofiltration permeate water quality." Proceedings, Ameri-
   can Water Works Association Membrane Technology Conference, Baltimore, MD, August 1-5,
   293-322.

Jacangelo, J. G. (1999). "Control of disinfection by-products by pressure-driven membrane pro-
   cesses." Formation and control of disinfection by-products in drinking water. P. C. Singer, ed.,
   American Waterworks Association, Denver, CO, 305-328.

Kruithof, J. C., Schippers, J. C., Kamp, P.  C., Folmer, H. C., and Hofman, J. A. M. H. (1998). "Inte-
   grated multi-objective membrane systems for surface water treatment: Pretreatment of reverse
   osmosis by conventional treatment and ultrafiltration." Desalination, 111, 37-48.

Lahoussine-Turcaud, V, Wiesner M. R., Bottero, J. Y, and Mallevialle, J. (1990). "Coagulation
   pretreatment for ultrafiltration of a surface water." Journal of the American Water Works Associa-
   tion, 82(12), 76-81.

Laine, J. M.  (1989). "Optimization of organic removal in ultrafiltration of a natural water." Master's
   thesis, University of Illinois at Urbana-Champaign.
                                             9-25

-------
Laine, J. M., Hagstrom, J. P., Clark, M. M., and Mallevialle, J. (1989). "Effects of ultrafiltration
   membrane composition." Journal of the American Water Works Association,  81(11), 61-67.

Luft, P. J. (1984). "Modeling of multicomponent adsorption onto granulated activated carbon in
   mixtures of known and unknown composition." Master's thesis, Michigan Technological Univer-
   sity, Houghton, MI.

Lykins, Jr., B. W., Clark, R. M., and Adams, J. Q. (1988a). "Granular activated carbon for controlling
   THMs." Journal of the American Water Works Association, 80(5), 85-92.

Lykins, Jr., B. W., Clark, R. M., and Cleverly, D. H. (1988b). "Polychlorinated dioxin and furan
   discharge during carbon reactivation." Journal of Environmental Engineering Division of the
   ASCE, 114(2), 300-316.

Lykins, Jr., B. W. and Clark, R. M. (1994). "U.S. drinking-water regulations: Treatment technologies
   and cost." Journal of Environmental Engineering Division of the ASCE, 120(4), 783-802.

Lytle, D. A., Schock, M. R., Speth, T. R, and Swertfeger, J. (1996). "Evaluation of corrosion issues
   for nanofiltration treated water." Proceedings, GAC, Membranes and the ICR: A Workshop on
   Bench-Scale and Pilot-Scale Evaluations, Cincinnati, OH, March 3-5.

Marshall, A. D., Munro, P. A., and Tragardh, G. (1993). "The effect of protein fouling in
   microfiltration and ultrafiltration on permeate flux, protein retention and selectivity: A literature
   review." Desalination, 91, 65-108.

Matthiasson, E. and Sivik, B. (1980). "Concentration polarization and fouling." Desalination,  35,
   59-103.

Merkel, T. H., Speth, T. H., Wang, J. Z., and Summers, R. S. (1998). "The performance of
   nanofiltration in the treatment of bank-filtered and conventionally treated surface waters."  Pro-
   ceedings, American Water Works Association Water Quality Technology Conference, San Diego,
   CA, November
Miltner, R. J., Nolan, S. A., Dryfuse, M. U., and Summers, R. S. (1994). "The control of DBFs by
   enhanced coagulation." Proceedings, 1994 American Water Works Association Water Quality
   Annual Conference, New York, June 19-23, 721-735.

Miltner, R. J., Summers, R. S., Dugan, N. R., Koechling, M., Moll, D. M. (1996). "A comparative
   evaluation of biological filters." Proceedings, 1996 American Water Works Association Water
   Quality Technology Conference, New York, November.

Owen, C., Taylor, J. S., Robert, C., and Reiss, C. R. (1999). "Microbial challenge of integrated
   membrane system large scale pilot plants treating a highly organic surface water." Proceedings,
   American Water Works Association Water Quality Technology Conference, Tampa, FL, October
   31 -November 4.

Paul, D. H. (1991). "Reverse osmosis: Scaling, fouling, and chemical attack." Desalinization and
   Water Reuse, 1, 8-11.

Potts, D. E., Ahlert, R. C., and Wang, S. S. (1981). "A critical review of fouling of reverse osmosis
   membranes." Desalination, 36, 235-264.

Reasoner, D. J., Blannon, J. C., and Geldreich, E. E. (1987). "Microbial characteristics of third-
   faucet point-of-use devices." Journal of the American Water Works Association, 79(10), 60-66.
                                             9-26

-------
Reiss, C. R. and Taylor, J. S. (1991). "Membrane pretreatment of a surface watQt"Proceedings, Ameri-
   can Water Works Association Membrane Technology Conference., Orlando, FL, March 10-13.

Reiss, C. R., Taylor, J. S., Owen, C., and Robert, C. (1999a). "Diffusion-controlled organic solute
   mass transport in nanofiltration systems." Proceedings, American Water Works Association Water
   Quality Technology Conference, Tampa, FL, October 31-November 4.

Reiss, C. R., Taylor, J. S., Robert, C., and Hong, S. (1999b). "Successful nanofiltration of 20 mg/L
   TOC river water." Proceedings, American Water Works Association Membrane Technology
   Conference, Long Beach, CA, February 28-March 3.

Ridgway, H. R, Rigby, M. G., and Argo, D. G. (1985). "Bacterial adhesion and fouling of reverse
   osmosis membranes." Journal of the American Water Works Association, 77, 97-106.

Ridgway, H. F. and Safarik, J. (1991) "Biofouling of reverse osmosis membranes." Biofouling and
   biocorrosion in industrial water systems, H. C. Flemming and G. G. Geesey, eds., Springer-
   Verlag, Berlin, 81-111.

Robert, C., Taylor, J. S., Reiss, C. R., and Langenderfer, N. (1999). "Surface water treatment by
   integrated membrane systems with in-line coagulation." Proceedings, American Water Works
   Association Membrane Technology Conference, Long Beach, CA, February 28-March 3.

Roberts, P. V. and Summers, R. S. (1982). "Performance of granular activated carbon for total or-
   ganic carbon removal." Journal of the American Water Works Association, 74(2), 113-118.

Semmens, M. J., Norgaard, G.  E., Hohenstein, G., and Staples, A. B. (1986a). "Influence of pH on
   the removal of organics by granular activated carbon." Journal of the American Water Works
   Association, 78(5), 89-93.

Semmens, M. J., Staples, A. B., Hohenstein, G., and Norgaard, G. E. (1986b). "Influence of coagula-
   tion on removal of organics by granular activated carbon." Journal of the American Water Works
   Association, 78(8), 80-84.

Singer, P. C. (1994). "Control of DBFs in drinking water." Journal of Environmental Engineering
   Division of the ASCE, 120(4), 727-744.

Snoeyink, V. L., Kirisits, M. J., and Pelekani, C. (1999). "Adsorption of disinfection by-product
   precursors." Formation and control of disinfection by-products in drinking water. P. C. Singer,
   ed., American Water Works Association, Denver, CO, 259-284.

Snoeyink, V. L. and Summers, R. S. (1999). "Membranes." Water quality and treatment, 5th Ed., R.
   D. Letterman, ed., McGraw-Hill, Inc., New York, NY, 13.1-13.83.

Solomon, T., Summers, R. S., and Speth, T. F. (1993). "The reduction of membrane fouling by
   physical, chemical and biological pretreatment." Proceedings, American Water Works Associa-
   tion Membrane Conference, Baltimore, MD, August 1-5, 159-186.

Sontheimer, H. and Hubele, C. (1987). "The use of ozone and granulated activated carbon in drink-
   ing water treatment." Treatment of drinking water for organic contaminants. P. M. Huck and P.
   Toft, eds., Pergamon Press, Oxford.

Serial, G. A., Papadimas, S. P., Suidan, M. T., and Speth, T. F. (1994a). "Competitive adsorption of
   VOCs and BOM—Oxic and anoxic environments." Water Research, 28(9), 1907-1919.
                                             9-27

-------
Serial, G. A., Cerminara, P. I, Papadimas, S. P., Suidan, M. T., and Speth, T. F. (1994b). "Competi-
   tive adsorption of VOCs and BOM—Role of molecular oxygen." Journal of the American Water
   Works Association, 86(3), 80-90.

Speth, T. F. (1986). "Predicting equilibria for single and multicomponent aqueous phase adsorption
   onto activated carbon." Master's thesis, Michigan Technological University, Houghton, MI.

Speth, T. F. and Miltner, R. J. (1989). "The effect of preloading on the scale-up of GAC
   microcolumns." Journal of the American Water Works Association, 81(4), 141-148.

Speth, T. F., Lykins, Jr., B. W., and Miltner, R. J. (1989). "The use of pilot columns for predicting
   full-scale GAC performance." Proceedings, Design and Use of Granular Activated
   Carbon-Practical Aspects, Cincinnati, OH, May 9-10, 139-143.

Speth, T. F. and Miltner, R. J. (1990). "Adsorption capacity of GAC for synthetic organics." Journal
   of the American Water Works Association, 82(2), 72-75.

Speth, T.F. (1990). "Removal of VOCs from drinking water by adsorption." Significance and treat-
   ment of volatile organic compounds in water supplies, N. Ram, K. Cantor, and R. Christman,
   eds., Lewis Publishers, Inc., Ann Arbor, MI, 229-260

Speth, T. F. and Adams J. Q. (1991). "GAC and air stripping design support for the safe drinking
   water act." Strategies and technologies for meeting SDWA requirements, R. Clark and S. Sum-
   mers, eds., Lewis Publishers, Inc., Ann Arbor, MI, 47-89.

Speth, T. F. (1991). "Evaluating capacities of GAC preloaded with natural water.'''Journal of 'Envi-
   ronmental Engineering Division of the ASCE, 117(1), 66-79.

Speth, T. F., Allgeier, S. C., Summers, R.  S., Swertfeger, J., andMylin, M. L.  (1996). "Evaluating
   bench-scale nanofiltration studies for predicting full-scale performance." Proceedings, GAC,
   Membranes and the ICR: A Workshop on Bench-Scale and Pilot-Scale Evaluations, Cincinnati,
   OH, March 3-5.

Speth, T. R, Gusses, A. M., Allgeier, S. C., Summers, R. S., Swertfeger, J. W., and MacLeod, B. W.
   (1997).  "Predicting pilot-scale membrane performance with bench-scale studies." Proceedings,
   American Water Works Association Membrane Technology Conference, New Orleans, LA,
   February 23-26.

Speth, T. F. (1998). "Evaluation of nanofiltration foulants from treated surface waters." Ph.D. Disser-
   tation, University of Cincinnati, Cincinnati, OH.

Speth, T. F. and Miltner, R. J. (1998). "Adsorption capacity of GAC for synthetic organics—Part II."
   Journal of the American Water Works Association, 90(4), 171-174.

Speth, T. F., Summers, R. S., and Gusses, A. M.  (1998). "Nanofiltration foulants from a treated
   surface  water." Environmental Science and Technology, 32, 3612-3617.

Speth, T. F., Merkel, T., Parrett, C. J, Gusses, A. M., Grandin, A. M., and Summers, R. S. (1999).
   "Riverbank filtration as a pretreatment for nanofiltration membranes." InternationalRiverbank
   Filtration Conference, Louisville, KY, November 4-6.

Speth, T. F., Summers, R. S., and Gusses, A. M.  (2000). "Evaluation of nanofiltration pretreatments
   for flux loss control." Desalination, 130, 31-44.
                                             9-28

-------
Summers, R. S., Benz, M. A., Shukairy, H. M., and Cummings, L. (1993). "Effect of separation
   processes on the formation of brominated THMs." Journal of the American Water Works Asso-
   ciation, 85(1), 88-95.

Summers, R. S., Hooper,  S. M., Solarik, G., Owen, D. M., and Hong, S. (1995). "Bench-scale evalua-
   tion of GAC for NOM control." Journal of the American Water Works Association, 87(8), 69-80.

Suratt, W. B. (1993). "The reduction of membrane fouling by physical, chemical and biological
   pretreatment." Proceedings, American Water Works Association Membrane Technology Confer-
   ence, Baltimore, MD, August 1-5.

Symons,  J. M., Stevens, A. A., Clark, R. M., Geldreich, E. E., Love, Jr., O. T., and DeMarco, J.
   (1981). "Treatment techniques for controlling trihalomethanes in drinking water." EPA-600/2-81-
   156, Environmental Protection Agency, Cincinnati, OH.

Taylor, J. S., Thompson,  D. M., and Carswell, J. K. (1987). "Applying membrane processes to
   groundwater sources  for trihalomethane precursor control." Journal of the American Water Works
   Association, 79(8), 72-82.

Taylor, J. S., Mulford, L. A., Duranceau, S. J., and Barrett, W. M. (1989). "Cost and performance of
   a membrane pilot plant." Journal of the American Water Works Association, 81(11), 52-60.

Taylor, J. S., et al. (1992). "Reduction of disinfection by-product precursors by nanofiltration." EPA
   Project Summary, EPA/600/SR-92/023, Office of Research and Development, April.

Taylor, J. S. and Wiesner, M. (1999). "Membranes." Water quality and treatment, 5th Ed., R. D.
   Letterman, ed., McGraw-Hill, Inc., New York, NY, 11.1-11.71.

United States Environmental Protection Agency. (1999). "Analysis of GAC effluent blending during
   the ICR treatment studies." EPA 815-C-99-002, September.

van den Berg, G. B.  and C. A. Smolders (1990). "Flux decline in ultrafiltration processes." Desalina-
   tion, 77,  101-133.

Vidic, R. D., Serial,  G. A., Papadimas, S. P., Suidan, M. T., and Speth, T. R (1992). "Effect of mo-
   lecular oxygen on the scaleup of GAC adsorbers." Journal of the American Water Works Asso-
   ciation, 84(8), 98-105.

Wang, J.  Z., Summers, R. S., and Miltner, R. J. (1995). "Biofiltration performance: Part 1, relation-
   ship to biomass." Journal of the American Water Works Association, 87(12), 55-63.

Warta, C. L. (1993). "Effect of oxygen tension on the activated carbon adsorption of natural organic
   matter in Ohio River  water." Master's thesis, University of Cincinnati, Cincinnati, OH.

Warta, C. L., Papadimas, S. P., Sorial, G. A., Suidan, M. T., and Speth, T. R (1995).  "The effect of
   molecular oxygen on the activated carbon adsorption of natural organic matter in Ohio River
   water." Water Research, 29(2), 551-562.

Watson, B. M. and Hornburg, C. D. (1989). "Low-energy membrane nanofiltration for removal of
   color, organics, and hardness from drinking water supplies." Desalination, 72, 11-22.

Wetterau, G. E., Clark, M. M., and Anselme, C. (1996). "A dynamic model for predicting fouling
   effects during the ultrafiltration of a groundwater." Journal of Membrane Science, 109, 185-204.
                                             9-29

-------
Wiesner, M. R., Rook, J. J., and Fiessinger, F. (1987). "Optimizing the placement of GAC filtration
   units." Journal of American Water Works Association, 79(12), 39-49.

Wiesner, M. R., Clark, M. M., and Mallevialle, J. (1989). "Membrane filtration of coagulated sus-
   pensions." Journal of Environmental Engineering Division oftheASCE, 115(1), 20-40.

Wilkes, D. R., Chellam, S., Bonacquisti, T. P., Long, B.W., and Jacangelo, J. G. (1996). "Conven-
   tional treatment as a pretreatment for nanofiltration: Impact on DBF precursor removal, fouling,
   and facilities design." Proceedings, American Water Works Association Annual Conference,
   Toronto, Canada.

Zhu, X. and Elimelech, M. (1997). "Colloidal fouling of reverse osmosis membranes: Measurements
   and fouling mechanisms." Environmental Science Technology, 31, 3654-3662.
                                            9-30

-------
                                     CHAPTER 10

                                      Coagulation1
Introduction
Coagulation has historically been used for the control of particulates in drinking water, and its role in
the simultaneous control of organic carbon is well known. With the inclusion of disinfection by-prod-
uct (DBF) control as part of the Environmental Protection Agency's (EPA's) drinking water regulatory
philosophy, the role of coagulation has expanded to include control of DBF precursors. This chapter
presents recent studies conducted by the EPA's Office of Research and Development (ORD) in Cincin-
nati that (1) examined conventional coagulation and coagulation enhanced to more effectively control
organic carbon and DBF precursors, and (2) examined the effects of enhanced coagulation on other
water quality parameters.

Background

Water systems treating particulate-laden surface waters conventionally coagulate their waters to re-
move turbidity. Their goal is to achieve  sufficiently low levels that downstream filters operate without
excessive buildup of head loss (HL) and achieve cost-effective filter run times (FRTs). During conven-
tional coagulation, the concentration of natural organic matter (NOM) is lowered. Since DBF percursors
are part of the NOM, a strategy for control of DBF formation is removal of the NOM by coagulation
prior to disinfection. Because the NOM is largely  unidentified and not directly measurable, total or-
ganic carbon (TOC) serves as a surrogate for the DBF precursors. Typically, about 90 percent of the
TOC  is dissolved organic carbon (DOC); the other 10 percent is sorbed onto parti culates. There are
surface waters, however, for which the DOC is a lower percentage of the total.

Aluminum and iron salts are typically used for coagulation. For metal salts, two mechanisms for re-
moval of NOM are accepted (Singer and Harrington 1993; Krasner and Amy 1995; Owen et al. 1993).
In the first, negatively charged NOM is neutralized by positively charged metals forming insoluble
complexes (Al or Fe humates and fulvates), followed by precipitation of NOM with the floe. In the
second, NOM adsorbs onto metal hydroxide (A1(OH)3 floe or Fe(OH)3 floe) precipitates. The effective-
ness of coagulation is strongly dependent on pH and the dose of the coagulant. At higher coagulant
doses, more  metal for floe or complex formation is available.  Typically, coagulation of NOM is most
effective in the pH range of 5 to 6, as charge neutralization tends to be more effective at lower pH. At
lower pH, the charge density of humic and fulvic acids is reduced, making them more hydrophobic and
adsorbable. Lower pH can be achieved by acidification and/or by higher coagulant dosing. More metal
hydroxide (A1(OH)3 or Fe(OH)3) is formed at higher coagulant doses, therefore more FT in solution
lowers the pH. Thus, TOC removal and DBF precursor removal can be enhanced by decreasing pH
and/or by increasing coagulants doses.

NOM can be divided into hydrophobic and hydrophilic fractions (Singer and Harrington 1993). The
hydrophobic fraction tends to be less soluble, higher molecular weight, and more aromatic and is de-
scribed as humic. The hydrophilic fraction is described as non-humic. The humic (hydrophobic) frac-
tion is that retained on XAD® resin.  The non-humic (hydrophilic) fraction passes XAD® resin. The
humic fraction is more readily coagulated by aluminum and iron salts than the non-humic fraction.
'Richard J. Miltner: ORD/NRMRL/WSWRD, AWBERC Mailstop 689, 26 West Martin Luther King
Dr., Cincinnati, OH 45268, 513-569-7403, miltner.richard@epa.gov.
                                             10-1

-------
Table 10-1. Required Percent Removal of TOC by Enhanced Coagulation (Federal Register
           1998)
Source Water Alkalinity, mg CaCO,/L
Source Water TOC, mg/L
>2.0 to 4.0
>4.0 to 8.0
>8.0
Oto60
35
45
50
>60 to 120
25
35
40
>120
15
25
30
Ultraviolet absorbance (UVA) is caused by aromatic and unsaturated double bonds in the NOM and is
commonly measured at 254 nm (UV254). Specific UVA (SUVA), or UV254 divided by DOC, is an
indicator of the DBF precursor removal treatability (Federal Register 1998). Edzwald (1993) has shown
that SUVA in the 4 to 5 L/mg-m range is characteristic of waters with humic (hydrophobic) carbon and
is more easily coagulable for DOC control, whereas SUVA in the <3 L/mg-m range is characteristic of
waters with non-humic (hydrophilic) carbon, which is less susceptible to DOC removal by coagulation.

Enhanced  coagulation has two places in the Disinfectant/Disinfection By-Product (D/DBP) Rule. One
is as a treatment technique for the control of precursors for identified and nonidentified DBFs. Another
is as a best available technology (BAT) for the control of regulated total trihalomethanes (TTHMs) and
five regulated haloacetic acids (HAAS).

As a treatment technique, water systems are not expected to optimize, or maximize, the  removal of
DBF precursors. Whether coagulation is enhanced or optimized for the control of DBF precursors is a
matter of degree. So as not to be cost prohibitive, systems must meet target percent removals of TOC
where TOC serves as a surrogate for the identified and nonidentified DBF precursors. The targets are
based on two factors, the source water's TOC concentration and the source water's alkalinity. A 3 x 3
matrix results and is shown in Table 10-1 (Federal Register 1998). The table reflects two observed
phenomena:  relatively more TOC can be removed from higher-TOC waters than from lower-TOC
waters, and lower percent removal of TOC is expected in higher-alkalinity waters as higher alkalinity
makes depressing the pH more difficult.

Meeting the requirements of the 3x3 matrix is termed Step  1 in the D/DBP Rule. Systems may meet
these requirements however they choose; enhancing coagulation is an option, as is the use of granular
activated carbon, powdered activated carbon, etc.

As stated,  the purpose of the TOC removal in the Step 1 matrix is control of precursors for identified
and nonidentified DBFs.  Systems with good control of DBFs do not have to meet the matrix require-
ments, i.e., there are alternative compliance criteria to Step 1. These are based on other water quality
measures indicative of control of DBF precursors and include: source water TOC < 2.0 mg/L, treated
water TOC < 2.0 mg/L, TTHM not > 40 ug/L and HAAS not > 30 ug/L with the use of chlorine, source
water SUVA < 2.0 L/mg-m, and finished water SUVA < 2.0 L/mg-m.

Meeting the requirements of the  3x3 matrix is thought to be achievable by 90 percent of drinking
water systems, and these systems would realize an incremental improvement in DBF precursor over
their conventional practices (Federal Register 1998). Some systems are likely to already meet Step 1
requirements with conventional coagulation (Krasner and Amy 1995).

If systems  can not meet the Step 1  criteria because of the nature of their precursor, they must perform j ar
tests to determine how much TOC removal can be achieved, i.e., they must define alternative perfor-
mance criteria. This is termed Step 2. The D/DBP Rule will  force many water systems to  move from
conventional to enhanced coagulation and to expand their coagulation objectives to  include TOC
removal.
                                            10-2

-------
Conventional vs. Enhanced vs. Optimized Coagulation

Figure 10-1 describes the results of adding alum to move from conventional coagulation to enhanced
coagulation (achieving the target percent removal of TOC) to coagulation optimized for the removal of
TOC. It represents jar testing of 4 waters: Great Miami River (GMR), East Fork Lake (EFL), Miami
Whitewater Lake (MWL), and Stonelick Lake (SL) (Miltner et al. 1994a). Alum doses to control tur-
bidity ranged from as low as 15 mg/L to as high as 45 mg/L as indicated by the conventional box in the
figure. Control of turbidity in j ar testing was defined as achieving 1 nephelometric turbidity unit (NTU)
in settled waters. Jar tested waters consistently resulted in lower settled turbidities than pilot-treated
waters; based on these and other studies (Miltner et al.  1994b), jar test settled turbidities near 1 NTU
resulted in pilot plant settled turbidities near 2 NTU. At the other extreme, doses to achieve optimized
removal of TOC ranged from 75 to 130 mg/L  as indicated by the optimum box in Figure 10-1. Opti-
mized coagulation was defined as the lowest coagulant dose resulting in the best removal of TOC. An
                 o~
                 o
                 O
  3


  2

  1

  0

160 f-

140

120

100

 80

 60

 40

 20

  0
                            \ Conventional
                           • |\
                              ^  Target % Removal
                                N
                                            Optimum
                            H	1	1	1	1	1	1	h
                             Conventional
                                   Target % R TOC

                                           Optimum
                                                            Stage 1
                       0    25   50   75  100  125  150  175   200  225  250

                                         Alum Dose, mg/L
Figure 10-1. Control of TOC and UFC TTHM in GMR, EFL, MWL, and SL waters by
            coagulation (Miltner et al. 1994a).
                                             10-3

-------
                    0
                     0    20   40   60   80   100   120  140  160  180  200
                                        Alum Dose, mg/L

Figure 10-2. Control of TOC in EFL water by alum coagulation (Miltner et al. 1994a).

example is given in Figure 10-2 where 75 mg/L alum resulted in 51 percent removal of TOC; at higher
doses, TOC concentrations could not be differentiated. Alum doses required to meet the requirements
of enhanced coagulation ranged from 30 to 60 mg/L as indicated by the target % R TOC box in Figure
10-1. These results indicate that enhanced coagulation could be achieved at doses below those required
to optimize TOC removal and at doses not significantly greater than those required to control turbidity.
In some cases, conventional treatment alone was sufficient to meet the requirements of enhanced co-
agulation as indicated by the overlap of the conventional and target % R TOC boxes. These data sup-
port the intentions of the D/DBP Rule, wherein 90 percent of systems would be able to meet the re-
quirements of enhanced coagulation with moderate changes to conventional coagulation.

Figure 10-1 also shows similar data for control of TTHM precursors. Precursors in this chapter were
assessed by chlorination under uniform formation conditions (UFCs) (Summers et al. 1996). UFCs
represent national, mean, finished-water, distribution-system conditions of 1 mg/L free chlorine  after
24 hours at pH 8 at 20 C. Because they  are distribution-system conditions, the DBF concentrations
represent those reaching the consumer. When control of TTFDVI is considered for these four waters,
conventional coagulation rarely produced TTFDVI concentrations below the D/DBP Rule Stage 1 maxi-
mum contaminant level (MCL)  of 0.080 mg/L. Enhancing coagulation to meet the target percent
removals of TOC generally resulted in TTFDVI concentrations below the Stage 1 MCL. In Figure 10-1,
30 mg/L represents only the lowest alum dose that achieved the target percent removal of TOC, i.e., it
represents only one water on one date. Optimizing coagulation for TOC control always resulted in
Stage 1 MCL compliance. Similar results were obtained for HAAS and its Stage 1 MCL of 0.060 mg/L.
This supports the basis of enhanced coagulation's place in the D/DBP Rule. As a BAT, enhanced co-
agulation can result in the control of TTHM and HAAS.

Enhanced  Coagulation's Role in Water Quality

Ohio River (OR), Green Swamp (GS) (Dryfuse et al. 1995), MWL, and EFL waters (Miltner et al.
1994a) were jar tested to assess the relationships between TOC, SUVA,  chlorine demand, and precur-
sors for total organic halide (TOX), TTHM, six haloacetic acids (HAA6), chloral hydrate (CH), four
haloacetonitriles (HAN4) and chloropicrin (CP). Results are given in Table 10-2. The data show that
conventional coagulation would not meet the requirements of the 3x3 matrix for TOC removal in OR
and MWL waters and that additional treatment processes, enhanced coagulation or others, would be
                                            10-4

-------
required. Conventional coagulation would meet the requirements, however, in EFL and GS waters.
Several trends are apparent in the results presented in Table 10-2. These trends are also apparent in
Table 10-3.

Improved Water Quality with Enhanced Coagulation
Moving from conventional coagulation to coagulation optimized for the removal of TOC results in
improved removal of precursors for TTHM, HAA6, CH, HAN4, CP, and the  surrogate TOX and in
reduction in chlorine demand. Using coagulation of MWL water as an example, optimized coagulation
was better than conventional coagulation for all parameters. Based on the relationship between conven-
tional, enhanced, and optimized coagulation described in Figure 10-2, moving from conventional co-
agulation to enhanced coagulation would result in improved removal of DBF precursors and reduction
in chlorine demand, but to a lesser degree than moving to optimized coagulation.

Table 10-2. Coagulation of OR, MWL, EFL and GS Waters with Alum

                                         Percent Removal
Parameter
       OR
(Dryfuse et al. 1995)
DOC = 2.47 mg/L
Alkalinity = 59 mg/L
TOC Target = 35% R
UV254 = 0.05/cm
SUVA = 2.02 L/mg-m
       MWL
(Miltner et al. 1994a)
TOC = 4.79 mg/L
Alkalinity = 104 mg/L
TOC Target = 35% R
UV254 = 0.092/cm
SUVA = 1.92 L/mg-m
       EFL
(Miltner et al. 1994a)
TOC = 5.16 mg/L
Alkalinity = 80 mg/L
TOC Target = 35% R
UV254 = 0.19/cm
SUVA = 3.70 L/mg-m
       GS
(Dryfuse et al. 1995)
DOC = 15.3 mg/L
Alkalinity = 88 mg/L
TOC Target = 40% R
UV254 = 0.73/cm
SUVA = 4.77 L/mg-m
TOC, DOC
UV254
SUVA
UFC TOX
UFC TTHM
UFC HAA6
UFCCH
UFC HAN4
UFCCP
C12 demand
Conv
20
20
0
30
18
36




Opt
46
60
26
63
59
67




Conv
18
32
16
36
16
44
22
30
34
37
Opt
45
54
17
60
49
71
56
46
61
61
Conv
43
68
43
57
53
66
59
45
43
53
Opt
51
74
47
73
74
79
72
52
61
50
Conv
47
67
38
70
60
70




Opt
71
86
53
86
83
90




Table 10-3. Coagulation of GMR Water with Alum and Ferric Chloride (Miltner et al. 1994a)
Percent Removal
Parameter
TOC
UFC TOX
UFC TTHM
UFCHAA6
UFCCH
UFC HAN4
UFCCP
C12 demand
pH
Coagulant dose, mg/L
Alum
Conventional
22
9
13
10
30
25
39
50
7.92
15

Optimized
45
44
46
61
66
40
44
67
7.04
120
Ferric
Conventional
25
23
25
31
41
26
61
49
7.85
20
Chloride
Optimized
54
51
54
64
73
44
78
58
6.91
125
                                             10-5

-------
TOC as an Indicator of DBF Precursor Control

Coagulation to remove TOC results in the removal of the precursors for THMs, HAAs, CH, HANs, CP,
and the surrogate TOX, and the removal of TOC is generally a conservative indicator of the removal of
these DBF precursors. Using conventional coagulation of EFL water as an example, TOC removal was
43 percent, whereas removal of TTHM, HAA6, CH, HAN4, CP, and TOX precursors ranged from 43
to 66 percent. This supports the basis of enhanced coagulation's place in the D/DBP Rule. As a treat-
ment technique, enhanced coagulation to remove TOC can result in the removal of precursors for
numerous DBFs. As a BAT, enhanced coagulation can result in the control of TTHM and HAAS.

Better Removal of TOC in Higher-TOC Waters

Raw waters higher in TOC tend to have higher percent removals of TOC. OR water with a raw TOC
of 2.47 mg/L achieved an optimum TOC removal of 46 percent, whereas GS water with a raw TOC of
15.3 mg/L achieved 71 percent removal. This is consistent with the pattern in the 3x3 enhanced
coagulation matrix (Table 10-1) in which higher-TOC waters are targeted for higher percent removal of
TOC.

SUVA as an Indicator of DBF Precursor Control

Raw waters higher in SUVA tend to have higher percent removals of DBF precursors. OR and MWL
waters with SUVAs near 2 L/mg-m achieved 60 to 63 percent removal of precursors for the DBF
surrogate TOX when coagulation was optimized for TOC removal. EFL water with a SUVA of 3.70 L/
mg-m achieved 73 percent and GS water with a SUVA of 4.77 L/mg-m achieved 86 percent. Precursors
for TTHM and HAA6 generally followed the same trend. This is consistent with SUVAs use in the D/
DBF Rule as an indicator of DBF precursor treatability.
Coagulation With and Without Acid Addition

Systems may opt to achieve the lower pH of enhanced coagulation by increasing the coagulant dose
and/or by adding an acid. Tryby et al. (1993) studied acid addition in jar testing of OR water. Figure
10-3 shows TOC control in an alum-treated water of ambient pH and an alum-treated water with pH
               60
               50
               40
              o
              
-------
adjusted to 7 with hydrochloric acid. In each case, about 34 percent optimal removal of TOC was
achieved. Without acid addition, 120 mg/L alum was required to achieve optimum TOC removal at pH
5.9. With acid addition, only 30 mg/L alum was required to achieve optimum TOC removal at pH 6.7.
Each system attempting to enhance coagulation could decide on acid addition vs. additional coagulant
based on factors such as safety, chemical handling, sludge production and handling, and costs.

Comparing Alum and Iron Coagulation

Table 10-3 and Figure 10-4 show jar testing results in which GMR water coagulated with ferric chlo-
ride (FeCl3»6H2O) gave  somewhat better removal of TOC and DBF precursors than when alum
(A12(SO4)3»14H2O) coagulated (Miltner et al. 1994a). The exception was in the control of chlorine
demand. This may be affected by the reaction between iron and chlorine in meeting the UFC require-
ment of a 1 mg/L free chlorine residual.
                              	
                          0   20  40  60  80  100  120  140  160  180  200
                                        Coagulant Dose, mg/L
Figure 10-4. Control of TOC and DBF Precursors in GMR Water by Alum and Iron
            Coagulation (Miltner et al. 1994a).
                                            10-7

-------
The lower pH that occurs with iron at a given coagulant dose may explain, in part, the better perfor-
mance of iron compared to alum. These data must be viewed with caution, however, as they are pre-
sented on a weight basis. When doses are compared on an equivalence basis, the advantage of treating
with ferric chloride is offset as the ferric chloride curves in Figure 10-4  shift to the right about 9
percent, i.e.,  an 81.9 mg/L dose of FeQ3»6H20 is equivalent to a 90 mg/L dose of A12(SO4)3»14H20. A
system's choice of a coagulant should be based not only on treatment effectiveness but also based on
factors such as cost, chemical handling, and sludge production and handling.

Table 10-3 also shows data similar to that in Table 10-2, i.e., improving coagulation to better control TOC
results in improved control of precursors for TTHM, HAA6, CH, HAN4, CP, and the surrogate TOX.

Fractionation

Dryfuse et al. (1995) studied OR, EFL, and GS waters for the effects of coagulation on humic and non-
humic fractions of DOC and on molecular-sized fractions of DOC. Raw and coagulated jar tested
waters were examined before (in bulk) and after fractionation. Both conventional coagulation and co-
agulation optimized for DOC removal were employed. Waters were fractionated by XAD-8 resin to
isolate the non-humic (hydrophilic) fraction in the resin column effluent. The humic (hydrophobic)
fraction was determined by difference. Waters were separated by two parallel ultrafiltration membranes
into <0.5K and <3K molecular size (MS) ranges. Thus, by difference, <0.5K, 0.5K-3K, and >3K MS
ranges were determined.

Results for OR and GS bulk (unfractionated) waters are presented in Table 10-2 and show that DBF
precursor removal, as measured by TOX, TTFDVI, and HAA6, was improved as the coagulant dose was
increased from conventional to that optimized for DOC control.
Most of the DOC and the DBF precursors were in the larger molecular size (>0.5K)  range. This is
described in  Table 10-4 in which the fractions represent the mean of the three raw waters. Similarly,
most of the DOC and the DBF precursors were in the humic fraction.

Table 10-4. Molecular Size and H/NH Fractions of OR, EFL, and GS Raw Waters (Taken from
           Dryfuse et al. 1995; Dryfuse 1995)
Percent of Total
Parameter
DOC
UFC TOX
UFC TTHM
UFCHAA6
>3K
34
53
33
44
0.5K-3K
45
42
53
46
<0.5K
20
5
14
10
Percent of Total
H
53
68
67
73
NH
47
32
33
27
  H = humic; NH = non-humic.

The effect of coagulation on MS fractions is seen in Table 10-5 and represents the mean of the three
waters. The behavior of the DOC fractions was similar to that of the DBF precursor fractions. Conven-
tional coagulation removed a greater percentage of the >3K MS fraction (61 to 86 percent) than of the
smaller-sized fractions (9 to 45 percent). Optimizing coagulation brought about a small improvement
in the >3K MS range, i.e., a 0 to 14 percent increase. The greatest improvement with optimized coagu-
lation, however, was in the 0.5K-3K MS range where increases ranged 39 to 49 percent. Improvement
in the <0.5 MS range was small, i.e., 0 to 13 percent.
Table 10-6 describes the effect of coagulation on humic and non-humic fractions for the mean of the
three waters. The behavior of the DOC fractions was similar to that of the DBF precursor fractions.
Conventional coagulation removed a greater percentage of the humic fraction (51 to 62 percent) than of
                                            10-8

-------
Table 10-5. Effect of Coagulation on Molecular-Sized Fractions of OR, EFL, and GS Waters
           (Taken from Dryfuse et al. 1995; Dryfuse 1995)
Percent Removal
MS Fraction
>3K
0.5K-3K
<0.5K

>3K
0.5K-3K
<0.5K

>3K
0.5K-3K
<0.5K

>3K
0.5K-3K
<0.5K
DOC Conventional
81
9
16
UFC TOX Conventional
85
33
40
UFC TTHM Conventional
61
22
45
UFC HAA6 Conventional
86
26
31
DOC Optimized
86
49
26
UFC TOX Optimized
99
72
41
UFC TTHM Optimized
61
63
45
UFC HAA6 Optimized
98
75
44
Increase
5
40
10
Increase
14
39
1
Increase
0
41
0
Increase
12
49
13
Table 10-6. Effect of Coagulation on H/NH Fractions of OR, EFL and GS Waters (Taken from
           Dryfuse et al. 1995; Dryfuse 1995)
Percent Removal
H/NH Fraction
H
NH

H
NH

H
NH
DOC Conventional
51
18
UFC TOX Conventional
52
41
UFC HAA6 Conventional
62
42
DOC Optimized
72
40
UFC TOX Optimized
83
62
UFC HAA6 Optimized
86
66
Increase
21
22
Increase
31
21
Increase
20
24
  H = humic; NH = non-humic.
the non-humic fraction (18 to 42 percent). Optimizing coagulation brought about similar improvement
in the removal of both fractions, i.e., 20 to 31 percent for the humic fraction and 21 to 24 percent for the
non-humic fraction.

Speciation

Coagulation shifts the distribution of DBFs toward the more brominated species when enhanced or
optimized coagulation is practiced. Table 10-7 shows TOC concentrations in parallel pilot plants treat-
ing EFL water and employing conventional coagulation and coagulation optimized for TOC control
(Miltner  1994b). The precursor concentrations for TTHM and HAA6 formation are also shown in
Table 10-7. Comparing conventional and optimized coagulation indicates improved removal of these
precursors with optimized coagulation. Although the concentrations of THMs and HAAs that would
form in post-disinfected water decreased with coagulation, both on a weight basis and on a molar basis,
the percentage of brominated species increased as indicated by the ratios of brominated-to-total DBFs.
                                             10-9

-------
Table 10-7. Effect of Alum Coagulation of Pilot Plant-Treated EFL Water on DBF Speciation
            (Miltner et al. 1994b)

Raw
Conv. settled
Opt. settled

Raw
Conv. settled
Opt. settled
TOC
mg/L
4.81
3.42
2.21
TOC
mg/L
4.81
3.42
2.21
UFC TTHM
ug/L
102
79
53
UFCHAA6
ug/L
115
65
30
UFC TTHM
umole/L
0.814
0.620
0.424
UFC HAA6
umole/L
0.794
0.447
0.216
UFC TTHM-Br
umole/L
0.079
0.072
0.064
UFC HAA6-Br
umole/L
0.0133
0.0126
0.0111
Ratio
TTHM-BnTTHM
0.097
0.117
0.150
Ratio
HAA6-Br:HAA6
0.017
0.028
0.051
Table 10-8. Comparison of Jar Testing and Pilot-Plant Treatment of EFL Water (Miltner et al.
           1994b)
Percent Removal
Conventional Coagulation

TOC
UV254
UFC TOX
UFC TTHM
UFCHAA6
Alum dose, mg/L
Turbidity, NTU
pH
Jar Test3
26
42
22
46
38
40
0.47
7.23
Pilot Plant"
29
39
24
44
36
44
0.99
7.34
Optimized Coagulation
Jar Test3
52
65
41
73
61
130
0.37
6.50
Pilot Plant"
54
64
47
73
63
152
0.73
6.57
   Based on September 14 jar test.
  b Mean of six sample days from September 2 to September 14.

For example, optimized coagulation lowered TTHM precursors from 79 to 53 ug/L compared to con-
ventional coagulation. With post-disinfection, however, about 15 percent, on a molar basis, of the 53
ug/L resulting from optimized treatment was brominated THMs, while only about 12 percent of the 79
ug/L resulting from conventional treatment was brominated THMs. As the precursor compounds were
removed by  coagulation, the bromide was unaffected. Thus, the bromide-to-organic carbon ratio in-
creased, resulting in a higher percentage of the DBFs as brominated compounds when chlorinated
under UFCs. Tryby et al. (1993) reported the similar trends with jar testing of OR water.

Scale Up

A concern of the Step 2 jar testing procedure, to define alternate performance criteria for enhanced
coagulation,  is scale-up reliability. Table 10-8 compares conventional and TOC-optimized coagulation
for both j ar testing and pilot-scale treatment of EFL water (Miltner et al.  1994b). The j ar test was a good
predictor of pilot-scale performance, with the exception of turbidity; jar test settled NTUs were about
half the pilot-plant settled NTUs. These data suggest that if the coagulant mixing intensities (GT val-
ues) for the larger and smaller systems are similar (as they were in this study) and if the tests at both
scales are conducted at close to the same time (the water qualities are similar), then jar tests will predict
results from the pilot-scale system.
                                            10-10

-------
Secondary Effects

Two concerns with modifying coagulation to better control DBFs are (1) loss of control of microbial,
particulate, or inorganic water quality parameters, and (2) operational problems. Several studies of the
secondary effects of enhanced or optimized coagulation suggest no detrimental tradeoffs with regard to
water quality, but increased sludge and resulting increases in cost.

Bacteria

Lytle et al. (1994) monitored heterotrophic plate count (HPC) bacteria (using R2A media) and total
coliform (TC) bacteria through pilot plants treating EFL water using conventional and TOC-optimized
coagulation. TOC-optimized coagulation resulted in a 0.54 log increase in removal of HPC bacteria
and a 0.34 log increase in removal of TC bacteria.

Cryptosporidium Oocysts and Giardia  Cysts

In the same study, Lytle et al. (1994) monitored particle counts in the size range of Cryptosporidium
oocysts (3.1 to 7.0 um) and Giardia cysts (8.2 to 13.2 um). TOC-optimized coagulation resulted in 0.77
log better removal of Cryptosporidium  oocyst-sized particles (1.88 vs. 1.11 logs) and 0.80 log better
removal of Giardia cyst-sized particles  (1.79 vs. 0.99 logs) as compared to conventional coagulation.

Oilier et al. (1997) studied the secondary effects of conventional, enhanced, and optimized coagulation
in Cryptosporidiumparvum oocyst-spiked OR, EFL, and Mississippi River (MR) waters. In addition to
turbidity and oocysts, they monitored  indigenous bacterial endospores (Rice et al. 1996) and total
particle count (TPC) as indicators of oocyst removal efficiency. Several coagulation conditions were
examined: conventional targeting 5 NTU, conventional targeting 2 NTU, enhanced following the re-
quirements of the regulatory 3x3  matrix, optimized  for TPC removal, and optimized for TOC
removal. Table 10-9 summarizes results for the three waters.

These data suggest that systems that increase the coagulant dose to move from conventional to en-
hanced coagulation would realize improved removal of C. parvum oocysts and that endospores and
TPC would be reasonable and conservative indicators of oocyst removal.
Table 10-9. Control of Microbes and Particulates by Coagulation of OR, EFL, and MR Waters
           (Oilier et al. 1997)
Mean Log Removal
Parameter
C. parvum Oocysts
Endospores
TPC
Turbidity
n
Mean alum dose, mg/L
Conv-5
1.3
1.5
1.4
1.5
2
20
Conv-2
2.0
2.1
1.8
1.6
5
36
Enhanced
2.6
2.4
2.1
1.8
5
45
Opt-TPC
3.3
2.8
2.5
1.8
5
67
Opt-TOC
3.0
2.7
2.4
1.5
5
112
Turbidity and Particles

Systems that increase the coagulant dose beyond that required by the 3x3 regulatory matrix would
realize improved removal of C. parvum oocysts, endospores, and TPC through the coagulant dose
range that optimizes TPC control. As shown in Table 10-9, optimizing for TOC control resulted in
                                            10-11

-------
Table 10-10. Effects of Treatment on Pilot-Scale Filter Operation and Mean Water Quality of
            EFL Water (Lytle et al. 1994)
Conventional Coagulation
Parameter
Change in pH
Change in TPC/mL
Change in AP+, mg/L
HL buildup, cm/hr
FRT
Filter 1 Ambient pH
7.37 to 7.47
+ 0.10
18,207 to 2938
15,269
0.648 to 0.080
0.568
8.89
shortest
Optimized Coagulation
Filter 2 Ambient pH
6.88 to 6.90
+ 0.22
6027 to 205
5822
0.472 to 0.025a
0.447
3.60

Filter 3 pH 8
6.68 to 7.95
+ 1.27
6027 to 300
5727
0.472 to 0.291
0.181
1.45
longest
  1 dissolved Al detection level = 0.025 mg/L


better removal of oocysts, endospores, and TPC than practicing enhanced coagulation. Optimizing for
TOC control appeared to be destabilizing for particles since removal of oocysts and the other indicator
parameters deteriorated beyond the coagulant dose range usually selected for optimized TPC control.

Aluminum, Particles, Head Loss, and Filter Run Time

Lytle  et al. (1994) and Miltner et al. (1994b) reported on pilot-scale filter operation during parallel
conventional and optimized coagulation of EFL water. Table 10-10 summarizes the results. The pH of
coagulation and clarification  in the optimized plant was lower (6.88) than in the conventional plant
(7.37) because of the additional alum dose required in the optimized plant (see Table 10-8). Two paral-
lel filters were operated in the optimized plant. Chlorine was applied ahead of all three filters in both
plants to target UFCs. pH 8 was targeted in all three finished waters for UFCs. In Filters 1 and 2, pH
was adjusted at the clear wells. In Filter 3, pH was adjusted to 8 at the filter. In this manner, the effects
of pH and aluminum solubility on filter operation could be studied. The moderate pH changes in Filters
1 and 2 occurred because the  liquid chlorine was basic.

Optimized coagulation removed more TPC than did conventional coagulation. As a result, the particle
loading to Filters 2 and 3 was lower (6027/mL vs. 18,207/mL) and, consequently, the sludge produc-
tion was higher in the optimized plant. The TPCs in the filter effluents in the optimized plant were
approximately one log lower than in the conventional filter effluent (205/mL and 300/ml vs. 2938/mL).

Even  though more aluminum was utilized in the optimized plant, more was precipitated. Thus, the
dissolved aluminum loading  to Filters 2 and 3  was lower (0.472 mg/L vs. 0.648 mg/L). Aluminum
solubility increases with pH. Thus, the highest filter effluent dissolved aluminum was in Filter 3, which
was adjusted to pH 8. In Filters 1 and 2, where pH was ambient, lower filter effluent dissolved alumi-
num occurred. Consequently,  less aluminum precipitated in the optimized plant filters compared to the
conventional plant filter (0.181 mg/L and 0.447 mg/L vs. 0.568 mg/L).

Table 10-10 shows that Filter  1, following conventional treatment, removed more particles and precipi-
tated more aluminum  than the other two filters. Consequently, it built up HL faster and had the shortest
FRT. Conversely, with optimized treatment, HL buildup was slower and FRTs were longer.

This suggests that systems  switching from conventional to enhanced coagulation may achieve more
efficient filter operation. The  tradeoff will be more sludge production because of the addition of more
coagulant, or possibly a different type of sludge if enhanced coagulation was achieved by the addition
                                            10-12

-------
of an acid with the coagulant. These data also suggest that systems practicing enhanced coagulation
should consider pH adjustment ahead of the filter to improve filter operation. The tradeoff will be
higher dissolved aluminum entering their distribution systems.

Costs are always site specific. Because of the cost of additional coagulant or acid, the cost of sludge
handling, and the cost of raising the pH to distribute water from plants practicing enhanced coag-
ulation, the cost of water will likely be more expensive than from plants practicing conventional coagu-
lation.

Summary

With the D/DBP Rule, many water systems will move from conventional to enhanced coagulation,
expanding their coagulation objectives from turbidity removal to include TOC removal. Moving from
conventional coagulation to enhanced coagulation to coagulation optimized for the removal of TOC
results in improved removal of precursors for TTHM, HAA6, CH, HAN4, CP, the surrogate TOX, and
chlorine demand. When coagulated, raw waters higher in SUVA and in TOC tend to have higher per-
cent removals of DBF precursors. As a treatment technique, enhanced coagulation to remove TOC can
result in the removal of precursors for these DBFs. As a BAT, enhanced coagulation can result in the
control of TTHM and HAAS. Many systems should be able to meet the requirements of enhanced
coagulation for TOC removal with moderate changes to conventional coagulation.

Although coagulation lowers the concentrations of DBF precursors, coagulation shifts the distribution
of the DBFs formed by chlorination toward the more brominated species. This shift is even greater
when enhanced or optimized coagulation is practiced.

Most DOC and DBF precursors are in the larger molecular size range, and most DOC and DBF precur-
sors are in the humic fraction. Conventional coagulation removes a greater percentage of the >3K MS
fraction than the smaller-sized fractions. Enhancing coagulation brings about small improvements in
the >3K MS range and the <0.5 MS range; the greatest improvement with enhanced coagulation is in
the 0.5K-3K MS range. Conventional coagulation removes a greater percentage of the humic fraction
than the non-humic fraction. Enhancing coagulation brings about similar improvement in the removal
of both fractions.

Moving from conventional to enhanced to TOC-optimized coagulation generally results in better con-
trol of HFC bacteria, TC bacteria, C. parvum oocysts, Cryptosporidium oocyst-sized particles, Giardia
cyst-sized particles, TPC, and bacterial endospores.

Systems switching from conventional to enhanced coagulation may achieve longer FRTs. The tradeoff
will be more sludge production. Systems practicing enhanced coagulation should consider pH adjust-
ment ahead of the filter to achieve longer FRTs. The disadvantage of this practice when alum is the
coagulant will be higher dissolved aluminum entering their distribution systems.

References

Dryfuse, M. J., Miltner, R. J., and Summers, R.  S. (1995). "The removal of molecular size and
    humic/non-humic fractions of DBF precursors by optimized coagulation." Proceedings, Ameri-
    can Water Works Association, Annual Conference, Anaheim, C A, June.

Dryfuse, M. J. (1995). "An evaluation of conventional and optimized coagulation for TOC removal
    and DBF control in bulk and fractionated waters." M.S. Thesis, Department of Civil and Envi-
    ronmental Engineering, University of Cincinnati, Cincinnati, OH.
                                            10-13

-------
Edzwald, J. K. (1993). "Coagulation in drinking water treatment." Water Science and Technology,
   27(11), 21.

Federal Register. (1998). "National primary drinking water regulations: Disinfectants and disinfec-
   tion by-products; Final rule." 63(No. 241; December 16), 69389.

Krasner, S. W. and Amy, G.  (1995). "Jar-test evaluations of enhanced coagulation." Journal of the
   American Water Works Association, 87(10), 93.

Lytle, D. A., Dryfuse, M. J., Miltner, R. J., Schock, M. R., and Summers, R. S. (1994). "An evalua-
   tion of the secondary effects of enhanced coagulation." Proceedings, American Water Works
   Association, Enhanced Coagulation Research Workshop, Charleston, SC, December 4-5.

Miltner, R. J., Nolan, S. A., Dryfuse, M. J., and Summers, R. S. (1994a). "The control of DBFs by
   enhanced coagulation." Proceedings, Water Quality, American Water Works Association, Annual
   Conference, New York, NY, June 19-23.

Miltner, R. J., Nolan, S. A., Dryfuse, M. J., and Summers, R. S. (1994b). "Evaluation of enhanced
   coagulation for DBF control." Proceedings, American Society of Chemical Engineers, National
   Conference on Environmental Engineering, Boulder, CO, July 11-13.

Oilier, L. L. (1998).  "Microbial and particulate control under conventional and enhanced coagula-
   tion." M.S. thesis, Department of Civil and Environmental Engineering, University of Cincinnati,
   Cincinnati, OH.

Oilier, L. L., Miltner, R. J., and Summers, R. S. (1997). "Microbial and particulate control under
   conventional and enhanced coagulation." Proceedings, American  Water Works Association,
   Water  Quality Technology Conference, Denver, CO, November 9-12.

Owen, D.  M., Amy,  G. L., and Chowdhury, Z. K. (1993). "Characterization of natural organic matter
   and its relationship to treatability." American Water Works Association Research Foundation,
   Denver, CO.

Rice, E. W., Fox, K. R., Miltner, R. J., Lytle, D. A., and Johnson, C. H. (1996). "Evaluating plant
   performance with endospores." Journal of the American Water Works Association, 88(9), 122.

Singer, P. C. and Harrington, G. W. (1993). "Coagulation of DBF precursors: Theoretical and practi-
   cal considerations." Proceedings, American Water Works Association, Water Quality Technology
   Conference, Miami, FL, November 7-11.

Summers, R.  S., Hooper, S. M., Shukairy, H. M., Solarik, G., and Owen, D. (1996). "Assessing DBF
   yield: Uniform formation conditions." Journal of the American Water Works Association, 88(6),
   80.

Tryby, M.  E., Miltner, R. J.,  and Summers, R.  S. (1993). "TOC removal as a predictor of DBF
   control with enhanced coagulation." Proceedings, American Water Works Association, Water
   Quality Technology Conference, Miami, FL, November 7-11.
                                            10-14

-------
                                     CHAPTER 11

  Controlling Disinfection By-Products (DBPs) and Microbial Contaminants in
                        Small Public Water Systems (PWSs)1

Introduction

The purpose of this chapter is to describe the in-house and field research activities specifically designed
to evaluate alternative treatment technologies for small community and non-community water systems.
The discussion is comprised of four major sections: (1) Particulate Removal, (2) Disinfection/Destruc-
tion, (3) Field-Scale Demonstration, and (4) Small System Remote Monitoring and Control. Small
systems face a myriad of problems above and beyond those of the medium and large systems. The pilot-
and full-scale research efforts described in this paper are intended to address a subset of these needs.
Because small systems (defined below) often lack the financial, technical, and managerial capabilities
of larger systems and are responsible for the majority of the Safe Drinking Water Act (SDWA) viola-
tions, they have been the focus of specific portions of various Federal Regulations and Rules.

The U.S. Environmental Protection Agency (EPA) Water  Supply  and Water Resources Division's
(WSWRD's) in-house research described in this chapter primarily focuses on filtration and disinfec-
tion technologies that are considered to be viable alternatives to conventional package plants (floccula-
tion, coagulation, media filtration, post-chlorination). Conventional package plants require a high
level of operator skill to properly maintain appropriate chemical dose and flow rates, especially for
surface water supplies. These difficulties, in conjunction with the other small system problems men-
tioned previously, have resulted in the EPA focusing its research  on technologies that are easy to
operate and maintain and produce minimal residuals. An objective of this research is to provide
technology at the lowest possible cost while producing a finished water that is robust with respect to
operator skill and raw water quality. The field demonstration projects describe some of the issues that
can result when even the best available technology is not able to produce water of sufficient quantity
or quality. The remote monitoring  and control  research efforts described resulted from the fact that
many rural systems are located in topographically difficult areas or separated by large distances from
other systems, thus precluding any consolidation or regionalization efforts. The software and sensing
systems developed as a product of this research will allow individual treatment units to be monitored
and operated from a central location.  This approach has come to be known as "the electronic circuit
rider" concept.

A very important "spin-off" from this research, but which will not be discussed here, is the EPA Envi-
ronmental Technology Verification (ETV) Program for Drinking Water Treatment Systems. The goal
of this program is to develop performance standards and protocols that can be used to evaluate the
performance of small systems technologies. The ETV works in partnership with testing organizations,
stakeholder groups, and individual technology developers to provide high quality, peer-reviewed data
in order to accelerate the acceptance and use of improved and cost-effective technologies.
1 James A. Goodrich and Roy C. Haught: Water Quality Management Branch, WSWRD, NRMRL,
ORD, EPA, 26 W. M. L. King Drive, Cincinnati, Ohio 45268. Corresponding Author: James A. Goodrich,
513-569-7605, goodrich.james@epa.gov.
                                             11-1

-------
SDWA Coverage

There are two main categories of public water systems (PWSs):

Community Water Systems provide drinking water to the same people year-round. Today, there are
approximately 54,000 community water systems serving more than 250 million Americans. All Fed-
eral drinking water regulations apply to these systems.

Non-Community Water Systems serve customers on less than a year-round basis. Non-community
systems are, in turn, divided into two categories:

   •   Nontransient: Those that serve at least 25 of the same people for more than six months in a
       year but not year-round (e.g., schools or factories that have their own water source); most
       drinking water regulations apply to the 20,000 systems in this category.
   •   Transient: Those that provide water to places like gas stations and campgrounds where
       people do not remain for long periods of time; only regulations that control contaminants
       posing immediate health risks apply to the 96,000 systems in this category.

There are approximately 170,000 community and non-community drinking water systems in the U.S.
serving transient and nontransient populations of 10,000 or fewer people (Goodrich et al. 1999). Table
11-1 describes the size categories of PWS that provided water to at least 25 people or 15 service
connections for at least 60 days per year.

Table 11-1. Size Categories of Public Water Systems (USEPA 1999)
Percent of Community Water Systems
System Size (population served)
Very small (25-500)
Small (501-3,300
Medium (3,001-10,000)
Large (10,001-100,000)
Very large (over 100,000)
1980
67
22
6
4
1
1985
63
24
7
5
1
1990
63
24
7
5
1
1995
61
25
8
6
1
According to a 1993 survey (MacDonald 1997), small treatment systems serving fewer than 10,000
people represent 94% of all water supply systems in the U.S. and serve 21% of the national population.
Tens of thousands of these small PWSs are having difficulty complying with the ever increasing num-
ber of regulated contaminants. Small drinking water systems account for 93 percent of the maximum
contaminant level (MCL) violations and 94 percent of the monitoring and reporting (M/R) violations.
M/R violations are a result of no sampling being performed or data recorded. The majority of the MCL
violations are for microbial parameters. According to the National Research Council, the most com-
mon monitoring and reporting violation is for total coliform bacteria (NRC 1997). Tables 11-2 and

Table 11-2. Total Coliform Bacteria Violations for the Period October 1,1992, through
            December 31,1994 (Pollack et al. 1999)
Population Served
Number of
Consumers
<500
501-3,300
3,301-10,000
>10,000
Systems with Violations
Number of Percent of
Systems Total (%)
10,509 29.5
1,938 13.4
592 14.4
487 14.4
Violations by Source Water
Ground Water Surface Water
Systems (%) Systems (%)
95.0 5.0
84.8 15.2
71.8 28.2
59.1 40.9
                                            11-2

-------
Table 11-3. Chemical Contamination Violations for the Period October 1,1992, through
            December 31,1994 (Pollack et al. 1999)
Population Served
Number of
Consumers
<500
501-3,300
3,301-10,000
>10,000
Systems with Violations
Number of Percent of
Systems Total (%)
531 1.5
162 1.1
25 0.6
15 0.4
Violations by
Ground Water
Systems (%)
96.4
73.5
60.0
33.3
Source Water
Surface Water
Systems (%)
3.6
26.5
40.0
66.7
11-3 present a summary of small system violations in the U.S. for total coliform bacteria and chemical
contaminants, respectively, for the period of October 1, 1992, through December 31, 1994.

Impact of the SDWA on Small Communities

The 1996 Safe Drinking Water Act Amendments (SDWAA) required EPA to

    •   Assess technologies for three categories of systems:
        10,000-3,301 persons
        3,300-501 persons
        500-25 persons
    •   List affordable technologies for each category to achieve current and anticipated MCLs
    •   Identify "variance technologies" if MCLs cannot be met to maximize contaminant reduction
       and protect public health
    •   Provide assumptions to be used in determining affordability

The 1996  SDWAA required EPA to produce a variety of outputs in a very short time period. The
outputs are
    •   A final technology list for the Enhanced Surface Water Treatment Rule
    •   A report on applicability of Point-Of-Use/Point-Of-Entry devices
    •   A final technology list for National Primary Drinking Water Regulations (NPDWRs)
    •   A final "variance" technology list for NPDWRs

Future regulations may require a significant reduction in the usage of free chlorine for small systems
treating surface water or ground water with high organic content because of the potentially restrictive
disinfection by-product (DBF) levels mandated by the amendments. Alternative filtration and/or disin-
fection systems may need to be added to current small system treatment trains in order to meet en-
hanced filtration requirements. Alternately, some small systems will have to completely replace or
significantly upgrade systems in order to be in compliance.

The practice of chlorination for preoxidation or for disinfection purposes can result in the formation of
chlorinated organic by-products. Currently, the regulated DBFs in the U.S. are trihalomethanes (THMs)
with a maximum contaminant level of 0.080 mg/L. However, the recently promulgated Disinfectant
and Disinfection By-Products (D/DBP) Rule will result in the regulation of several other by-products
of chlorination such as haloacetic acids (HAAS) of 0.060 mg/L, along with a potential reduction in the
current THM standard of 0.080 mg/L (Federal Register 1998). In some cases, this might result in a
change to an alternative preoxidant, or disinfectant, use of membranes or elimination of the use of free
chlorine entirely (Pollack et al. 1999).
                                             11-3

-------
Microbial contaminant regulations for surface water disinfection are those promulgated under the Sur-
face Water Treatment Rule (SWTR) (EPA 1989) or proposed in the Enhanced Surface Water Treatment
Rule (ESWTR) (EPA 1994). The SWTR requires the practice of disinfection for all utilities treating
surface water and ground water under the influence of surface water. Most utilities are also required to
filter their water unless the following conditions are met in the surface water prior to disinfection:

    •   Fecal coliform bacteria <20/100 mL in 90% of samples
    •   Total coliform bacteria <100/100 mL in 90% of samples
    •   Turbidity <5 nephelometric turbidity units (NTUs)
    •   Other MCLs met

Treatment plants exempted from filtration must disinfect to achieve 99.99% inactivation of viruses and
99.9% inactivation ofGiardia lamblia cysts. Compliance with these requirements must be demonstrated
with the CT approach (the product of the average disinfectant concentration and t10 contact time). CT
values estimated for actual disinfection systems must be equal to or greater than those published in the
SWTR Guidance Manual for viruses and G. Lamblia cysts, respectively (Pollack et al. 1999).

The two-stage Enhanced Surface Water Treatment Rule and the Ground Water Rule essentially bring
all  small systems serving less than 10,000 under regulations requiring  at least 2-log removal of
Cryptosporidium, a filtered-water turbidity of less than 0.3 NTU in at least 95% of the monthly mea-
surements, and disinfection of ground water while managing by-product formation. A key issue for all
systems to consider under the two-stage Long Term Enhanced Surface Water Treatment Rule is whether
or not criteria can be developed for estimating removal efficiencies for Cryptosporidium. Filtration
issues include

    •   Pretreatment efficiencies
    •   How variable are filter performances?
    •   Source water quality impacts
    •   Surrogate performance indicators

Issues surrounding disinfection efficiencies and by-product formation include

    •   Can Cryptosporidium inactivation criteria be developed?
    •   Source water quality impacts
    •   Treatment technique impacts


Drinking Water State Revolving Fund and Small Communities

In order to improve  small drinking water systems, EPA is required to enter into agreements with eli-
gible states to make capitalization grants to further the health protection objectives of the SDWAA. The
Nation's water systems must make significant investments to install, upgrade, or replace infrastructure
to continue to ensure the provision of safe drinking water to their 250 million customers. Installation of
new treatment facilities can improve the quality of drinking water and better protect public health.
Improvements are also needed to help those water systems experiencing a threat of contamination due
to aging infrastructure systems.

The 1996 SDWAA established the Drinking Water State Revolving Fund (DWSRF) to make funds
available to drinking water systems to finance infrastructure improvements. The program also em-
phasized providing funds to small and disadvantaged communities and to programs that encourage
pollution prevention as a tool for ensuring safe drinking water.
                                             11-4

-------
The DWSRF program required that states develop a priority system for funding infrastructure projects
based on three criteria from the SDWAA. States are required to solicit and consider public comment
when developing their priority systems. Projects are ranked and funding is offered to the highest ranked
projects that are ready to proceed. Priority is given to those eligible projects that

   1.  Address the most serious risk to human health
   2.  Are necessary to ensure compliance with the requirements of the SDWAA
   3.  Assist systems most in need, on a per household basis, according to state-determined
       affordability criteria

A minimum grant amount of 1% of the appropriated funds are available to all states. States that lose
primacy will not be eligible for DWSRF funds. Receipt of the funds is linked to states having an EPA-
approved capacity development  program and operator certification program. Funds can be used for
loans, loan guarantees,  and as a  source of reserve and security for leveraged funds. States must also
contribute an amount equal to 20% of the total Federal contribution.

Congress appropriated $1,275,000,000 for the program in FY1997. This figure included amounts ear-
marked for the program in FY1994-1996.  The appropriations for FY1998 and FY1999 were
$725,000,000 and $775,000,000, respectively. The appropriation for FY2000 is $820,000,000. Annual
state grants ranged from $7 to $80 million.

The regulatory drivers described above have provided a focus for identifying small system solutions.
As a consequence, the WSWRD of the EPA National Risk Management Research Laboratory
(NRMRL) has been actively demonstrating and evaluating alternative and innovative  small system
drinking water treatment technologies for several years. In anticipation of the states' needs for inno-
vative and cost-effective small system treatment technology, the WSWRD has focused on the small-
est of these systems in the 25-500 population range.  Because of the availability of Federal funds
through the DWSRF to purchase equipment, research has emphasized those technologies that are
easy to operate and maintain. Even though alternative treatment systems are perceived as "high tech"
or more expensive to purchase  than  conventional technologies, in  many cases they  are easier to
operate and less expensive to maintain over the long term. This fact also  led to the consideration of
remote telemetry and control technology to improve monitoring/reporting and reduce operation and
maintenance (O&M) costs. Although such equipment could double the  capital costs  of a package
plant, the O&M paybacks can  be quickly realized through lower use of chemicals,  low residual
production (disposal), and increased  reliability. It is believed that technology  more appropriately
designed for small systems will not only produce higher quality drinking water, but better utilize the
resources of the utilities and encourage more timely monitoring and reporting.

Research Approach

The WSWRD conducts small system research at EPAs Test and Evaluation (T&E) Facility located in
Cincinnati, OH. These technologies can be packaged into treatment systems for field testing at remote
sites. The T&E Facility  is described below; some of the technologies that are under evaluation or have
been tested are also listed.

Drinking Water Package Plant T&E Facility

The T&E Facility is a flexible research facility where a wide variety of water treatment and other
environmental protection technologies can be evaluated. The facility is equipped with 10,000 gallons
                                             11-5

-------
of permanent stainless steel tank storage capacity consisting of two 2,500 gallon tanks and one 5,000
gallon tank. The tanks are used as blending/holding tanks for experimental studies. The drinking water
work area also has two  1,000 gallon stainless steel and two 1,500 gallon polyethylene reservoirs.

The following is a list of the technologies installed and in operation as of FY 2000:

    •   Filtration Systems
       •   Slow sand
       •   Rapid sand
       •   Ultrafiltration
       •   Bag
       •   Cartridge
    •   Disinfection Systems
       •   On-site electrochemical chlorine generator
       •   Ozone, UV, and hydrogen peroxide alone and in combination

The facility is also equipped with a complete machine shop that provides total support in the engineer-
ing and fabrication of the bench- and pilot-scale research studies. The shop supports plumbing, ma-
chining, and electrical and electronic services (Table 11-4).
Table 11-4. Amenities/Facilities of T&E
Experimental High-Bay Area
Supply Water for Treatment
24,000 ft2
Potable
30-ft-tall high-bay area
  Chlorinated
16 experimental work area bays
  Dechlorinated
Two 5-ton bridge cranes
Non-Potable
552-ft2 greenhouse
 Mill Creek (surface water)
Fully heated and lighted
Waste water
Remote monitoring and control
 Primary influent
Ventilation system
 Primary effluent
720-ft2 machine shop
 Raw waste water
Low pressure air (15 psi)
  Secondary effluent
High pressure air (130 psi)
Laboratory
Electrical supply (110, 240, 480 volt)
  Deionize
Three 16-ft overhead doors
  Super-Q
Storage Capacity
Remote Telemetry
Tank truck
24 hours per day
Two 2,500-gallon (stainless steel)
Experimental monitoring
One 5,000-gallon (stainless steel)
Experimental control
Two 1,100-gallon (stainless steel)
Facility security
Two 1,500-gallon (PVC)
Facility fire
Test Methodology

All of the filtration and disinfection technology research described in this chapter focused on removal/
inactivation of Cryptosporidium and surrogate parameters. The experimental procedures for acquiring,
spiking,  collecting, and analyzing Cryptosporidium follows. Cryptosporidium parvum oocysts were
isolated from sieved (10-, 20-, 60-, and 100-mesh sieves) feces of Holstein bull calves by centrifuga-
tion (1100 xg) through a step gradient of Sheather's sucrose (Finch et al. 1993). Purified Cryptosporidium
                                                11-6

-------
                       Bag
                       Filter
                                                      Pump
                Drain-
, Raw
 Water
                     Flow'
                    Meter
                                                   Manifold
                                                    Filter
Figure 11-1. Contaminant injection method.

oocysts were stored in phosphate buffered saline (PBS) with penicillin (100 unit/mL) and streptomycin
(100 |ig/mL) at 4°C for up to 6-8 weeks. These oocysts  are referred to as antibiotic-preserved
Cryptosporidium oocysts (APO) in this research. A second source of Cryptosporidium oocysts was the
University of Arizona. These oocysts, which were stored in potassium dichromate at 4°C for approxi-
mately one month, are referred to as dichromate-preserved Cryptosporidium oocysts (DPO).

Figure 11-1 shows the typical setup for technology evaluations for the injection of Cryptosporidium or
other surrogate parameters  such as polystyrene 4.5 Jim-sized  microspheres. At the initiation of the
research described, there were no accurate and precise methods for determination of Cryptosporidium
removal rates in filtration systems. Furthermore, the recovery of Cryptosporidium varies significantly
between different experimental methodologies and procedures (Chapman and Rush 1990), and the direct
use of Cryptosporidium in treatment studies would pose a potential health risk. Therefore, finding more
reliable, non-hazardous surrogates was necessary, such as microspheres, turbidity, and particle counting.
These surrogates allowed for more experiments to be conducted, with only a few Cryptosporidium experi-
ments being necessary for verification purposes. Polystyrene microspheres are used because they are not
hazardous, are easily handled, and are representative of the size of the  Cryptosporidium oocyst
(Figure 11-2). Because of the difficulty and expense of testing directly using Cryptosporidium., most
experiments  either spiked or monitored surrogates such as the microspheres, particle counts, and turbid-
ity. Thus, Cryptosporidium oocysts were only used sporadically to verify surrogate results.

A predetermined number of fluorescent 4.5-|im polystyrene microspheres or Cryptosporidium oocysts
were suspended in a fixed volume of 0.01% (v/v)  Tween 20 solution and fed into the  raw  water up-
stream of the inlet pump. Approximately 5% of the total effluent volume was filtered with a l-|im pore
size polycarbonate membrane (diameter = 293 mm), supported in a stainless steel filter  manifold. The
filter was removed from the manifold and eluted with  a  squeegee using approximately  200  ml 0.01%
(v/v) Tween 20 solution. The eluant was concentrated to 0.5-7.5 ml via centrifugation at  1200 x g.
Cryptosporidium parvum oocysts in both influent and effluent samples were stained with an indirect
fluorescent monoclonal antibody (IFA). Polystyrene microspheres and oocysts were enumerated using
a hemacytometer under an ultraviolet (UV) microscope using epifluorescence at 400x total magnifica-
tion. Turbidity measurements were conducted with a Hach 21 OOP portable turbidimeter. The particle
count analyses  were conducted with  a Met One particle counter using a light scattering liquid 211
sensor over a range of 1 to 25 |im  (Li et al. 1997). The types of technologies examined using this test
methodology are described in the following sections.
                                              11-7

-------
            1e+08 R
                        Cryptosporidum

                        Microspheres
              100
                    3.5-4.0    4.01-4.5   4.51-5.0   5.01-5.5    5.51-6.0   6.01-6.5    6.51-7.0
Figure 11-2. Cryptosporidium and microsphere size distribution.
Particle Filtration

Filtration is likely to be the most practical treatment technology used for Cryptosporidium removal in
the near future (Chapman and Rush 1990). Investigations have shown that drinking water disinfectants
such as chlorine or monochloramine at typical dosages have virtually no effect on the inactivation of
Cryptosporidium oocysts (Clark et al. 1994). High-dose ozonation appears to be effective for inactivat-
ing Cryptosporidium oocysts in drinking water (Goodrich et al. 1992; Sabran et al. 1992), but its appli-
cation may result in generation of disinfectant by-products that exceed proposed MCLs.

Studies developed to  determine  the pliability of Cryptosporidium oocysts indicated that during the
filtration process, Cryptosporidium oocysts may pass through filtration membrane pores which are
smaller than the diameter of the organism, and a fraction of these oocysts remain viable (Li et al. 1995).
An understanding  of these characteristics is important for evaluation and optimization of filtration-
based physical treatment systems and is critical in the development of the experimental methodology
described above. Figure 11-3 outlines the filtration technologies described in this section.
Microns
Approx.
Mol. Wt.
Filtration
Technology
.0005
100
RO/POU
Devices
.05
10,000
Ultrafiltration
.5
500,000
Cartridge
Filtration
1-10

Bag
Filtration
Figure 11-3. Filtration technologies studied.
                                              11-8

-------
Bag Filtration

Three different bag filtration systems were initially studied over the course of one year. Bag filter #1
was a multi-layer fabric made of polypropylene with an average pore size of 1 |im. Bag filter #2 was a
single layer of polypropylene with an average pore size of 1 |im. Bag filter # 3 was a thick multi-layer
polypropylene bag that incorporated airborne particle testing to develop ratings of 99% efficiency in
removing material of 2.5-|im and 95% removal of 1.5-|im particles. The average inlet pressure was 50
psi (3.52kg/cm2) with influent flow rates of 12.5, 25, and 40 gpm, the recommended maximum (47, 95,
and 151 LPM). Microsphere, turbidity, and particle-count measurements were made at various pressure
drops as the bags became fouled. Bag filter performance data were collected at pressure drops of 0, 5,
10,  15, and 25 psi (0, 0.35, 0.70, 1.05, and 1.76 kg/cm2) for bag filter #2. Data for bag filters #1 and #3
were collected at 0, 7, 15, and 25 psi, respectively (0, 0.49, 1.05, and 1.76kg/cm2). The influent concen-
tration ranged from 9.17 x 103 to 1.50 x  105 per liter, with an average of 5.53 x  104 per liter for the
polystyrene microspheres which were spiked into the raw water during the course of the experiment
typically lasting between one and  two hours. For Cryptosporidium oocysts, the concentrations ranged
from 7.04 x 103 to 1.34 x 105 per liter, averaged 3.24 x 104 per liter, and were tested at only one flow
rate (25 gpm/95 LPM).

Particle-count analyses were conducted for particle size ranging from 1 to 25 \\.rs\. Unlike the polystyrene
microspheres and Cryptosporidium oocysts, turbidity and particle counts in the raw water were subject to
natural variations due to changes in raw water quality. The measured raw water turbidity during the study
period averaged 10.59 NTU, with a maximum of 59.4 NTU and a minimum of 1.81  NTU.

Performance Results

Removal studies comparing APO and DPO Cryptosporidium oocysts for the bag  filters were only
conducted at one flow  rate of 25 gpm (95  LPM) and with new filters because of the expense and time
required  for Cryptosporidium testing  (Table 11-5). The duration of each Cryptosporidium test was
approximately 70 minutes. Three blank tests yielded consistent APO Cryptosporidium log removal of
0.12, 0.13, and 0.19 for the vessels  only (no bag inserted) for filters #1, #2, and  #3, respectively. In
comparison, a blank test using the DPO Cryptosporidium shows a lower removal rate. In these studies,
2.70 x  108 DPO Cryptosporidium  oocysts were spiked into the influent and 2.57 x 10s were recovered
in the system effluent for a log removal of 0.02, which was lower than that of the APO  Cryptosporidium
(0.12) for the same filter vessel. The difference was likely to reflect the effect of oocysts' adhesion to
surface walls of the system. This  phenomenon is significantly decreased when oocysts are stored in
potassium dichromate. Previous studies conducted by WSWRD also show that potassium dichromate
may change surface characteristics  of the oocysts, such as  zeta potential (Lytle and Fox 1994). The
corrected log removal (accounting for the vessel-only removal effects) of bag filter #1  for APO
Cryptosporidium ranges from 1.35  to  1.48, with an average of 1.41 ± 0.066. Bag filter #3 has the
highest log removal, ranging from 3.00 to 3.63 with an average of 3.29 ± 0.32, while bag filter #2 yields
the  lowest log removal in the range of 0.26 to 0.64.

Table 11-6 shows the observed log  removal for turbidity, 4.5-|im microspheres and 4-6-|im particle
counts at the various pressure drops. In order to determine if operational parameters influenced filter
efficiency, the removal rates were compared to the various pressure drops across the bag filter, the flow
rates, and the Cryptosporidium and  surrogate influent spike levels. The reduction of 4.5-|im polysty-
rene microspheres at a given flow rate exhibits no significant dependence on pressure drop for bag
filter #1 and #2 (Figure 11-4 A, B), but decreases with pressure drop for bag filter #3 (Figure 11-4 C).
The log removal ranges from 1.14 to 1.88  with an average of 1.39 ± 0.19 (lo, n = 23) for bag filter #1,
and from 0.14 to 0.72 with an average of 0.46 ± 0.17 (lo, n = 7) for bag filter #2. In contrast, bag filter
                                             11-9

-------
Table 11-5. Bag Filtration Removal of Cryptosporidium

Blanks
Vessel # 1
Vessel # 2
Vessel # 3
Inlet Pressure
kg/cm2 (psi)
3.52 (50)
3.52 (50)
3.52 (50)
3.52 (50)
Pressure Drop
kg/cm2 (psi)
0(0)
0(0)
0(0)
0(0)
System Flow Rate Inlet Concentration
L/min (gpm)
95 (25)
95 (25)
95 (25)
95 (25)
APO/L DPO/L
4.7E+04
1.90E+04
135E+04
1.49E+04
Cryptosporidium Oocysts
Influent
2.70E+08
1.25E+08
9.55E+07
1.06E+08
Effluent
2.57E+08
9.52E+07
7.13E+07
6.77E+07
Log Reduction
Apparent
0.02
0.12
0.13
0.19
Corrected
NA
NA
NA
NA
Bag filter 1
Testl
Test 2
Tests
3.52 (50)
3.52 (50)
3.52 (50)
0(0)
0.35 (5)
0(0)
95 (25)
95 (25)
95 (25)
3.50E+04
1.60E+04
1.38E+04
2.37E+08
1.15E+08
9.77E+07
5.92E+06
3.50E+06
93.32+06
1.60
1.52
1.47
1.48
1.40
1.35
Bag filter 2
Testl
Test 2
Tests
3.52 (50)
3.52 (50)
3.52 (50)
0(0)
1.40 (20)
0(0)
95 (25)
95 (25)
95 (25)
7.40E+03
8.07E+03
7.04E+03
5.00E+07
5.50E+07
5.00E+07
8.47E+06
2.30E+07
1.65E+07
0.77
0.38
0.48
0.64
0.26
0.37
Bag filter 3
Testl
Test 2
Tests
3.52 (50)
3.52 (50)
3.52 (50)
0(0)
0.35 (5)
0.35 (5)
95 (25)
95 (25)
95 (25)
3.80E+04
6.78E+04
1.34E+05
2.59E+08
4.75E+08
9.77E+08
3.87E+04
1.73E+05
6.24E+05
3.83
3.44
3.19
3.63
3.25
3.00

-------
Table 11-6. Bag Filtration Results
Log Reduction
System
Bag filter #1



Bag filter #2




Bag filter #3



Pressure Drop (psi)
0
7
15
25
0
5
10
15
20
0
7
15
25
Microspheres
1.36
1.29
1.38
1.24
.33
.48
.72
.47
.40
3.20
1.88
2.03
1.09
Turbidity (NTU)
.84
.88
.87
.86
.16
.13
.08
.04
.08
1.68
1.53
1.09
0.75
4-6-p.m Particle Count
1.61
1.16
1.51
1.72
.34
.19
.10
.00
.06
2.65
2.49
0.76
0.88
#3 shows significant dependency of log removal on the pressure drop. Log removal for the 4.5-|im
polystyrene microspheres varies from 0.93 to 3.42 with an average of 2.08 ± 1.05. Log removal drops
significantly as the bag becomes fouled and pressure drop increases. The large standard deviation re-
flects the effect of pressure drop on performance of the bag filter #3.

Bag filters #1 and #2 displayed fairly constant log removals for the 4.5-|im polystyrene microspheres
and turbidity, as shown in Figure 11-5, for various spiking concentrations. It is apparent from the lack
of correlation that the log reduction for these filters is not dependent on the number of 4.5-|im polysty-
rene microspheres and turbidity loaded in the influent. Log removal for bag filter #3 varied substan-
tially. However, the variation appears to result from  significant dependance of bag filter #3's perfor-
mance on other operational parameters, including pressure drop and flow rate (Figure 11-4 C). There
was no statistically significant difference in removal rates at the various operating conditions, influent
turbidity levels, or variations in the total number of microspheres loaded onto the  filters (7 x 107 to
9.45 x 10s influent microspheres).

Controlled Turbidity Challenge Experimental Results

Different configurations of bag filtration systems were also challenged under controlled turbidity levels
and flow rates following the previously discussed experiments. The research was not intended to com-
pare systems but to identify the most important design and operational characteristics that provide for
the most economical application at various raw water situations. Important design considerations are
bag quality, gasket integrity, and hydraulic reliability. Operational factors include continuous vs. inter-
mittent operation, flow rate, and pressure differential. Turbidity challenges ranged from 1 NTU to 10
NTU. Average log reduction has ranged from 0.4 to 0.85. Of course, at higher influent turbidity levels,
greater removals can be demonstrated, but there seems to be a minimum NTU specific to each vendor
that can be reached regardless of the initial influent quality.

Figure 11-6 demonstrates three different experimental runs. During initial start-up, removal was better
and then settled into a fairly steady performance rate until  near the end of the bag's life. Flow rate did not
seem to be a major factor in filter performance. Figure 11-7 shows run time represented by pressure drop
over time for low- and moderate-turbidity challenges. Once a bag begins to foul at 5 to  10 psi differential,
                                              11-11

-------
                     _ --••-- Flow Rate 12.5 gpm
                                    — o — Flow Rate 25 gpm
               —                    — - —-Flow Rate 40 gpm
               § 2.5 -   ?
                 0.5
                     (A)
                                           10      15      20      25      30
                                         Pressure Drop - psi
                 2.0
               S                                                              CD
               o                                                              o
               E 1.0 	   ^
                                              ^                             Tl
                     (B)
                                       5        10       15       20       25
                                       Pressure Drop -psi
                 4.5
                                                                              S?
                                                                              to
                 1.5
                     (C)
                 0.5 ,
                            0       5      10      15      20      25      30
                                       Pressure Drop -psi

Figure 11-4. Log removal of 4.5-|im polystyrene microspheres at various pressure drops and
             flow rates for bag filter 1 (A), bag filter 2 (B), and bag filter 3 (C).

the time until the bag must be replaced prior to possible failure quickly decreases. High NTU scenarios
(>5 NTU) indicate the need for multiple filtration barriers for economical operation. Figure 11-8 reveals
the minimum turbidity achievable for a particular bag filter. Despite the influent turbidity range of 1 to 9
NTU, the effluent turbidity was consistently around .5 NTU. Figure 11-9 shows the results of running a
bag too long with a subsequent rupture. The treatment barrier is ineffective, with effluent water quality the
same or worse than influent. This can be seen readily because of the loss  of pressure differential. The
results indicate that in systems with little water storage or without on-site/automatic operator control to
interrupt operation at this point, it is critical to be conservative in estimating bag life.
                                               11-12

-------
                            o Bag Filter 1      • Bag Filter 2      • Bag Filter 3
          o
          E
          


-------
       1.40
o


£  0.80

D)
O


^  0.60

;5



(H  0.40





   0.20





   0.00
           I
\^
                                        40gpm, 6.0NTU, .77 log



                                        12gpm, 5.0 NTU, .86 log



                                        25gpm, 5.4 NTU, 1 log
10       15       20       25


         Pressure Drop (psi)
                                                                  30
                                                                       35
                                                                       40
Figure 11-6. Turbidity log removal vs. pressure drop.
                       6 NTU Influent
                                                                   1 NTU Influent
                                                        10
                                                              12
                                                               14
                                                        16
                                           Time (hour)
Figure 11-7. Pressure drop vs. time.
                                            11-14

-------
  10

   9

   8

   7
   3

   2

   1

   0
                 -•- Run #16 Influent

                 -o- Run #16 Effluent

                 -•- Run #19 Influent

                 -n- Run #19 Effluent
      0         5        10        15        20        25
                                      Pressure Drop (psi)

Figure 11-8. Raw and finished turbidity vs. pressure drop.
                       30
          35
        40
0.50 -r-
0.40 -1-
| »-| 0
o.so -5- m — -•-•-""" "\.
ro f ^» 	 — *\
> T \0 0 9
| °'2°+ ' " ~*^ ^^--

O)
° 0.10-

±^

-P r\ r\r\
^ U.UU
I—

-0.10 -



-0.20 -
n on
-U.oU
^
\
\
\
— \
\
\
\
\
\

\
\
\
— \
\
\
\



                               10
 15         20
Pressure Drop (psi)
25
30
35
Figure 11-9. Bag break.
                                               11-15

-------
                                                                             0 psi drop

                                                                            I   I Influent

                                                                            • Effluent
                                                                             7 psi drop
                                                                                Influent

                                                                                Effluent
                                         Particle Size (microns)


Figure 11-10. Influent vs. effluent particle counts.


One bag treated an average of 15,000 gallons of water before having to be replaced. Another bag
treated approximately 4,000 gallons of water, but exhibited a much higher effluent water quality. Thus,
bag selection is not a straightforward matching of pore size and the size of the particle to be removed.
The difficulty lies in the fact that, although bags are rated similarly, their performance can be very
different. The selection depends on the specific water quality characteristics and effluent objectives. At
a minimum, bag filtration should not be used as the only barrier to Cryptosporidium removal, but could
be used in conjunction with other technologies as a pretreatment step.

Microflltration

Various field evaluations have been conducted to assess the  operational performance of microfiltration
technology and provide information on the removal of physical and biological constituents under con-
ditions of continuous operation. Microfiltration membranes  normally have pore sizes 0.1 |im or greater
(Jacangelo et al. 1997). The water flow of the test system was 15 gpm (56.7 LPM) during experimental
run number one, two, and three.  The water flow of the test  system (Figure 11-11) was 34 gpm (128.5
LPM) for the final (test run four) test run. Polystyrene beads (4.5 |im) were injected into the raw,
untreated test water.  Treated effluent samples were collected from approximately  3.5% of the total
effluent volume for test run one, two, and three (15 gpm) and 1.5% of the total effluent volume for the
final test run (35 gpm). Water samples were collected using a l-|im polycarbonate membrane filter.
Experiments  one, two, and three demonstrated an average log removal of 3.71+/-0.19std. The log
                                            11-16

-------
reduction of turbidity was 1.33+/-0.38. Test run number four demonstrated a log reduction of 3.57 for
the polystyrene microspheres and 1.78 for the turbidity (Table 11-7). In addition to the data presented in
Table 11-7, particle counts were performed for test run four resulting in log removals of 3.85 for par-
ticle counts (4-6-|im range) and 3.14 for particle counts in the l-25-|im range. Collectively,  results
showed no influence due to the different flow rates. Similar to the bag filter research studies, the log
reduction of microspheres is about 2 times the log reduction of turbidity. The results indicate that
microfiltration technology is a feasible small system drinking water treatment technology for particle
removal (Li 1994).

Ultrafiltration

A spiral-wound ultrafiltration (UF) membrane package plant  has been installed at the T&E Facility
(Figure 11-12). Nominal pore size is  .05 |im with a molecular weight cutoff of 10,000 daltons. The
package plant can produce up to  15 gpm. During initial testing, 4.5-|im microspheres were observed in
the permeate resulting in only 2.5 log removal, suggesting a problem with the system. Upon inspection
of the membranes, a plastic adapter on the downstream end  of the permeate tube was found to be
broken, allowing raw water to pass directly into the permeate. This situation has also been observed in
field tests. An advantage that hollow-fiber membranes have compared to spiral-wound membranes is
that such a loss of filter integrity would be quickly observed for the hollow-fiber membrane system
                                                           MEMCOR
                                                           20-30M10
                                                          Self-Cleaning
                                                           Continuous
                                                       Microfiltration System
Figure 11-11. Microfiltration system.
                                             11-17

-------
Table 11-7. EPA Microfiltration Performance Evaluations
Parameter
Pressure in (psi)
Pressure out (psi)
Pressure drop (psi)
Membrane filter flow (gpm)
Duration of test (min)
Turbidity influent (MTU)
Turbidity effluent (MTU)
Turbidity (NTU) log reduction
Removal efficiency
Mass thru membrane (g)
Manifold flow rate (gpm)
Total mass through manifold (gal)
Concentrated effluent (mL)
No. beads in influent
No. beads in effluent
Log reduction of beads
Removal efficiency
Testl
23.3
10
13.3
15
71
11.7
0.63
1.27
94.62%
1079
0.41
33
0.5
6.91E+08
1.23E+05
3.75
99.98%
Test 2
23.2
10
13.2
15
60
4.63
0.49
0.98
89.42%
900
0.44
26.5
0.125
4.81E+08
1.49E+05
3.51
99.97%
Test3
23.3
10.3
13
15
87
5.46
0.1
1.74
98.17%
1305
0.40
39
0.23
5.78E+08
7.7E+04
3.88
99.99%
Test 4
22
9.3
12.7
33.6
55
2.39
0.04
1.78
98.33%
1848
0.48
25.7
1.3
2.10E+09
5.60E+05
3.57
99.97%
through changes in pressure readings. There was no indication from flow, pressure, or turbidity that the
spiral-wound system was not properly removing Cryptosporidium-sized particles.

Once the adapter was replaced, several experiments at the T&E Facility were conducted similarly to the
previously discussed technologies. The microsphere influent concentration averaged 1.94 x 105/L. The
Cryptosporidium oocyst influent concentration averaged  1.25 x  106/L. These  experiments used
Cryptosporidium parvum isolated from sieved feces of neonatal mice by centrifugation (1100 x g) through
a step gradient of Sheather's sucrose compared to the Holstein bull calves used in the previous bag filtra-
tion experiments. Cincinnati tap water was used as the raw water. The 24 studies were performed at an
average inlet pressure of 29 psi, and effluent permeate flow rate averaged 7.2 gpm. The sample collection
duration of each test ranged from 218 to 5,532 minutes with an average of 1,110 minutes. The system was
operated continuously and was purged at least 8 hours between each test run with tap water.

Results  indicated a 3-4 log removal range of microspheres from the influent to the permeate, with an
overall log removal average of 3.47 (Figure 11-13). As a comparison, Cryptosporidium filtration achieved
a log removal of 3.51 oocysts, which was very similar to the average log removal of the 4.5-|im polysty-
rene microspheres. However, the last data point shown in Figure 11-13 (Run 24) represents samples being
taken from the permeate over almost four days compared to just one day for the other data points exhibited
in the plot. After 5,532 minutes (approximately 3.84 days) of run time, microspheres were still found in
the permeate even though influent spiking had occurred over  a two-hour period at the beginning of the
experiment four days earlier (Figure 11-14). Log removal was 2.95 for this individual experiment, lower
than most of the previous experiments. The higher average removal rate achieved by the shorter experi-
ments could be the effect of insufficient sample collecting time, and suggests that particles may have long
residence times in membrane filters but are still capable of ultimately passing through.

Based upon the above technology investigations, it appears that there are alternative filtration technolo-
gies to conventional package plants. Depending on raw water characteristics, a likely configuration
could consist of filters in series with decreasingly small pore sizes that could in effect remove most
microbiological contaminants, reducing the need for chemical coagulants and disinfectants. Operation
and maintenance would be simplified, thus enhancing long-term compliance.
                                             11-18

-------
          1.  Raw Water Inlet
          2.  Bag Prefilter
          3.  Membrane Module
          4.  Chlorine Reservoir
5.  Control Panel
6.  Recirculation Loop
7.  Turbidimeter
8.  Chlorine Monitor
 9.  Cleaner Tank
10.  NaOH Reservoir
11.  NaOCI Reservoir
12.  Finished Water Outlet
13.  Reject Water Outlet
Figure 11-12. Ultrafiltration package plant.
Disinfection

As evidenced by the MCL and M/R violations of the SDWA and its amendments over the years, small
systems have difficulty in disinfecting their water and recording and submitting compliance data to the
appropriate state agency. In cases where good filtration is lacking or ground water is under the influ-
ence of surface water, there is also the potential for using too much disinfectant, resulting in water that
is unpalatable or resulting in the formation of by-products which exceed the MCL. As is the concern for
systems of all sizes in  selecting the most suitable disinfectant, small systems have to be even more
concerned with the safe and easy handling, shipping, storage, and the capital and O&M costs of disin-
fectants. In anticipation of small system needs in meeting the Stage 1 D/DBP Rule, the Ground Water
Rule, and the Stage 1 Enhanced Surface Water Treatment Rule, the WSWRD has investigated alterna-
tive technologies, focusing on their ability to inactivate Cryptosporidium while at the same time being
affordable and easy to operate and maintain. One of these technologies was an on-site oxidant genera-
tor, described below.

On-Site Electrochemical Oxidant Generator

Salt brine electrolysis generators are generally safer to handle and operate than chlorine gas, sodium
hypochlorite, and calcium hypochlorite systems. Salt brine, together with patented electrolytic cells,
                                              11-19

-------
     o
     O)
     O
                       4000
8000          12000

     Time (min)
16000
20000
Figure 11-13. Microsphere removal vs. membrane run time.
     LU


     '«  500000
     T3
     (0

     CO

     £  400000  -
        300000
        200000  -
        100000  -
                     1328       2848       4174       4870       5234


                                            Time (min)



Figure 11-14. Number of microspheres in effluent vs. run time, run 24.
                                    5532
                                            11-20

-------
                                       Electrolytic Cell

Salt + Raw
Water




~~ -^^~
Brine
Solution




                                                  Oxidant
                                                               Oxidant
1
         Raw Water
                                                                       Treated Water
Figure 11-15. Electrochemical oxidant generator system.


generate an anolyte (from the anode) liquor of primarily hypochlorous acid (Figure 11-15). Enhanced
pathogen inactivation has been hypothesized to involve the synergistic actions of ozone, free radicals,
chlorine dioxide, and oxyhalogen intermediates that form in the cell. It has also been assumed that this
mix of oxidants would facilitate inactivation while minimizing the formation of total trihalomethanes
(TTHMs). EPA initiated a project with the following objectives:

    •   Verify the presence or absence of intermediate oxidants that would facilitate inactivation
    •   Measure oxidant species and passive inorganic by-products
    •   Understand interactions and reaction rate

Results  of a multi-year study revealed that, because of the very high concentration of free chlorine
generated (as much as 400 mg/L) at the electrolytic cell anode, it was very difficult to measure for other
oxidant species (Table 11-8). A wide variety of analytical methods were investigated, and analysis of
the actual anode and cathode cells concluded that only free chlorine was produced. Table 11-9 lists the
typical anolyte liquor concentrations. Results show that the concentration of bromate ion (BrO3 ) formed
varies depending on the bromide ion (Br) concentration of the salt used in preparing the brine solu-
tions. The formation of chlorate ion (C1O3) is not a function of chlorine dioxide produced, but rather a
result of free available chlorine (FAC) decomposition. Likewise, BrO3 formation in electrolyzed salt
brine solutions  does  not require the presence of ozone.  Thus, ozone, chlorine dioxide, or hydrogen
peroxide were not  found either immediately at the electrolytic cell anode or in the disinfected water
(Gordon etal. 1999).

Mouse studies involving oocysts did not show any enhanced disinfection from using the electrolyzed
brine solutions compared to free chlorine (Table 11-10). Nor did the brine solution oxidant and chlorine
perform any better  than the "no treatment" control that indicated an average LogID50 reduction of 2.17.
LogID50 variability is most likely due to the initial health of the neonatal mice used in the infectivity
analysis. Although LogID50 values range from 1.9 to 3.12 between experiments, there is quite good
Table 11-8. Effect of 400 mg/L FAC on Common Analytical Methods
Method
DPD
Indigo
Horseradish peroxidase
Determination
C1O2
03
H2°2
Error Caused by
200-400 mg/L FAC
0.1-100 mg/L C1O2
0.2-20 mg/L O3
0.2-10 mg/L H2O2
                                              11-21

-------
Table 11-9. Typical Anolyte Liquor Concentrations Freshly Prepared Solutions
Species
FAC
C1O2
03
H202
pH
Cl-
C1O2-
C1O3-
BrO,-
Lab System (mg/L)
320-350
Not detected
Not detected
Not detected
o
5
9-10,000
0.05
4-5
O.05
Full-Scale System
230-250
Not detected
Not detected
Not detected
4.5
9-10,000
0.05
1-2
1-2
(mg/L)









Table 11-10. Mouse Infectivity Results
                 Positive Control   Brine Cell-30'    Brine Cell-180'    Cl-180'
Experiment
June 1998
July 1998
February 1999
June 1999
Combined results1"
1.92
1.92
2.70
a
2.17
1.90
N/P
1.75
1.85
1.83
1.90
1.98
3.04
2.32
2.31
1.90
1.55
3.12
2.27
2.21
  N/P-Not performed
  a Calculated logID50 values had negative values.
  b Calculated using the grand total of infected neonates and the grand total of neonates sacrificed in each experimental
       group.

consistency within an experiment. Table 11-11 describes total and individual  THM levels for both
chlorine and the oxidant produced by the electrolytic cell. Chloroform levels were lower for the brine
cell oxidant than for the chlorine, but TTHM formation levels were essentially the same, with chlorine
exhibiting lower TTHM levels in some cases because of the higher levels of brominated compounds
produced from the brine cell (Goodrich 1999). Variable bromide levels in the  salt used to make the
brine solution will most likely impact the formation of brominated compounds in the treated water.

Field-Scale Demonstration  Projects

Oftentimes, the difficulty of accepting new drinking water technologies is not related to experimental
results obtained under controlled  conditions, but in the  acceptance by the consumers and regulatory
agencies in real-world situations. This section describes  field studies incorporating both filtration and
disinfection technologies and the  "lessons learned" when attempting to balance microbiological and
disinfection water quality needs in small systems with limited resources.

Reverse Osmosis Home Membrane Systems

Unlike the filtration experiments  described earlier that  took place  in-house at  the T&E Facility, the
experiences presented below are from a field study focusing on removal of naturally occurring fluoride
at the tap by a point-of-use (POU) system. The water being supplied to the homes was provided by a
well located within the subdivision. However, the driving force in the ultimate acceptance by the State
of Virginia was the treatment devices' ability to provide finished water with acceptable levels of het-
                                            11-22

-------
Table 11-11. Electrolyzed Brine Cell/Chlorine TTHM Results

TOC (mg/L)
CHCL3 (Fg/L)
CHBRCL2 (Fg/L)
CHBR2CL (Fg/L)
CHBR3 (Fg/L)
TTHM (Fg/L)

TOC (mg/L)
CHCL3 (Fg/L)
CHBRCL2 (Fg/1)
CHBR2CL (Fg/L)
CHBR3 (Fg/L)
TTHM (Fg/L)
Ohio River
Concentration
2.62
0.5
0.1
O.I
O.I

Mill Creek
Concentration
5.07
O.I
O.I
O.I
0.1

Cell
30 min.
2.55
40.4
22.4
9.1
O.I
71.9
Cell
30 min.
5.14
84.7
31.9
15.7
0.1
132.3
Chlorine
30 min.
3.9
73.5
9.9
0.8
O.I
84.2
Chlorine
30 min.
5.71
108.5
16.6
O.I
0.1
125.1
Cell
180 min.
2.4
59.8
34.4
12.2
O.I
106.4
Cell
180 min.
5.2
99.9
42.2
35.4
0.1
177.5
Chlorine
180 min.
3.16
99.7
13.2
O.I
O.I
112.9
Chlorine
180 min.
5.56
131.1
17.5
O.I
0.1
148.6
erotrophic plate counts (HPC). A public-private partnership between the State of Virginia, EPA, and
three POU vendors demonstrated the use of reverse osmosis (RO) POU systems to reduce fluoride for
a subdivision as a lower cost alternative to abandoning the well and installing a large transmission line
to connect with a PWS several miles away. Prior to this project, no treatment existed at the subdivision's
well. These standard units were designed to treat only the water that would be used for drinking and
cooking, and in some homes, the ice-making units in refrigerators. They consisted of a sediment prefilter,
a high-flow, thin film (HFTF) reverse osmosis membrane, a storage tank, and an activated carbon post
filter (Figure 11-16). Basic parameters such as conductivity, fluoride, HPC, total coliform, chlorine
residual, pH, sodium, total dissolved solids (TDS), and turbidity were used to evaluate the performance
of the RO units.
                                            n_
                   To Drain
                 City Line
                                   I
r
         RO
                     _ Carbon
                       Postfilter
                                  Prefilter
                                                        Tank   I
Figure 11-16. RO/POU unit.
                                              11-23

-------
Fluoride reduction was easily achieved for the entire duration of the study, maintaining levels below the
secondary maximum contaminant level (SMCL) of 2.0 mg/L; however, HPC counts were elevated.
The decision was made to centrally chlorinate at the well and replace the HFTF membranes with chlo-
rine-resistant cellulose triacetate (CTA) membranes and remove the activated carbon postfilter. Subse-
quent sampling demonstrated satisfactory fluoride and HPC levels. Variances in fluoride and FIPC
concentrations from site to site is explained by membrane degradation and water usage. The life ex-
pectancy of the membrane depends on the environmental conditions. High temperatures, bacteria, and
high pH have an adverse affect on the membrane life and results in poor performance. Membranes were
replaced when the conductivity reduction decreased to 70 percent of the influent. It was observed that
conductivity reduction was generally lower than fluoride rejection, so this became a convenient, inex-
pensive, and conservative means of monitoring system efficacy. A correlation between HPC and chlo-
rine residual was also observed. In fact, much of the project focused on maintaining HPC levels below
500 cfu/mL.

Water quality sampling data indicated that the risk of exceeding  500 cfu/mL at the tap was inversely
proportional to the chlorine residual in the post-RO holding tank. Any time the residual exceeded 0.5
mg/L free chlorine, the HPC limit was maintained without exception. This is extremely relevant, be-
cause an RO membrane allows some chlorine to pass through, thus maintaining a residual at the tap in
this case. The water reaching each household typically exhibited chlorine residuals of 1 to 1.5 mg/L.
Concentrations in the holding tank between 0.5 and 1 mg/L were observed frequently, indicating 33 to
50 percent passage of the chlorine through the membrane. This concentration decreases over time in
the finished water holding tank as it is consumed through various  oxidation reactions. Because of this,
it can be presumed that negligible chlorine residuals indicate the unit has not been recently used.

It was concluded that the HPC concern could be eliminated by using  chlorine-tolerant membranes and
by continually chlorinating the subdivision's well. In most cases, the RO storage tank unit was continu-
ally refilled with chlorinated water. The HPC depended on the chlorine residual in the storage tank, and
usually the residual chlorine remained high enough to keep the unit clean. However, if the units were
not used daily, stagnant water in the tank caused a loss of residual chlorine, and the water was suscep-
tible to microbiological growth. It was found that one way to overcome this was to flush the tank daily.
This concept was demonstrated at a business  site during the study where the water was only  used
sporadically.

Public Acceptance

At least 1 gallon per day of RO water was consumed by 77% of the homeowners, corresponding to the
75% who used the system for all of their drinking and cooking needs. Just 6% of the participants
claimed to rarely use the RO water (Lykins et al. 1995). Although demonstrating fluoride reduction
with RO  has been done before, the challenge in this study was maintaining microbiological integrity
and gaining public and regulatory acceptance of POU treatment. This required an  entirely different
relationship between the state authorities and the customers. The initial and exit surveys confirmed not
only public acceptance, but showed an increase in customer satisfaction with the POU treatment. When
asked to rate the water quality on a scale of 1 to 4, 52% of participants in the initial survey rated the well
water (not chlorinated) quality as  "fair" or "poor," while 77% rated the RO water as "good" or "very
good" shortly after installation. In the exit survey one year later,  94% rated the RO water as good or
very good, showing a significant increase in the acceptance of the POU systems. This acceptance may
be due, in part, to the treated RO water being softer than the raw water.

The average RO water quality was rated 1.5 points higher than the  average tap water quality rating. The
average  rating was calculated by  summing the individual ratings and dividing by the number of re-
sponses. In the exit survey, RO water quality averaged 3.5 points  on a scale of 1 to 4, while well water
                                            11-24

-------
                           1 - Raw Water Inlet
                           2 - Media Pre-Filter
                           3 -Air Dryer
                           4 - Ozone Generator
5 - UV Light
6 - Contact Chamber
7 - Polishing Step (Optional)
8 - Finished Water Outlet
Figure 11-17. Ozone/UV POE unit.
quality averaged 2.1 points. Moreover, RO water quality was always rated at least as high or higher
than the well water quality, even when the nonchlorinated well water was compared to chlorinated RO
water. This is noteworthy because the switch to a chlorinated supply initially precipitated a number of
negative comments about taste. Microbiological integrity was not an issue for the consumers, whereas
it was the primary driver from the regulatory perspective.

Ozone Point-Of-Entry Field Application

Most ozone whole-house point-of-entry (POE) applications for drinking water in the past have been
utilized for oxidation of inorganic contaminants such as iron and manganese. Recent projects have
focused on the use of ozone in conjunction with UV light (Advanced Oxidation Process [AOP]) and
granular activated carbon for the destruction of synthetic organic contaminants in ground water and
disinfection of surface water supplies. Figure 11-17 describes such a POE unit installed in the cellar of
a sportsman's camp that served up to 30 hunters and fishermen daily in a lodge and four cabins. The
raw water is filtered through garnet, followed by ozone injection (0.4 mg/L), and then passes by a UV
light to a holding tank.

Disinfection results are shown in Table 11-12. Raw water quality was good. Finished and distributed
drinking water was negative for total coliform. HPC values varied somewhat, with one episode exceed-
ing 500 cfu/mL.  The variability could have been the result of biofilm in one of the cabins. The cabin
had not been occupied for days prior to sampling, thus resulting in old stagnant water in the plumbing
system's service  lines.

Ozonation by-products for the treated water were analyzed once during this brief study and indicated
lower levels of all but one of the low-molecular-weight aldehydes found in the raw water (Table 11-13).
This could have been the result of the overall good quality of the raw water and lack of ozone-demand-
ing compounds, allowing reduction of the by-products already formed in the raw water.
                                             11-25

-------
Table 11-12. Ozone/UV Disinfection Results
Date
06/18/91
07/07/91
07/21/91
08/01/91
08/14/91
08/25/91
09/03/91
09/11/91
11/02/91
Total Coliforms (MF) CFU/100 mL
Raw Post AOP Distributed Water
450 <1 <1
LA LA LA
40 <1 <1
40 <1 <1
60 <1 <1
100 <1 <1
30 <1 <1
240 <1 <1
100 <1 <1
HPC
Raw
56
1300
2400
260
76
50
80
470
180
(pour plate)
Post AOP
260
<1
5
38
<1
7
1
170
9
CFU/100 mL
Distributed Water
110
33
1
110
7
>500
1
73
30
  LA = Lab accident
Table 11-13. Ozone/UV By-Product Results
Compound
Formaldehyde
Acetaldehyde
Propanol
2-Butanone
Butanol
Pentanal
2-Hexanone
Benzaldehyde
Nonanal
Decanal
Glyoxal
Methyl glyoxal
Chloral hydrate
Raw
40.4
6.4
1.0
0.8




0.7
0.2
4.2


Post AOP
2.4
5.0
ND
ND
ND
ND
ND
ND
1.0
0.7
5.4
1.8
0.1
Distributed
1.4
48.3
ND
ND
ND
ND
ND
ND



0.1
0.1
ND = Not detected
In order to produce ozone, low-humidity oxygen is required. This POE unit utilized silica gel to dry the
ambient air in the cellar rather than install an expensive oxygen generator. Operational concerns cen-
tered around the frequency of reconditioning the air-drying material. Because of the high humidity in
the cellar of the lodge, the silica gel had to be reconditioned every few days. Although not expensive or
time consuming (30 minutes in an oven at 325°C), constant attention to this might not be maintained in
a household, and ozone generation, thus disinfection efficiency, could become highly variable (Goodrich
etal. 1993).

South America Water Treatment Technology Demonstration

Under the EPA Environmental Technology Initiative, several international technology demonstration
projects were conducted in order to introduce U.S. technology into foreign markets. One of these projects
involved systems to provide basic filtration and disinfection to three sites  in Ecuador (Gallo 1999).
Other small system projects have also been carried out in Mexico (still underway) and the People's
Republic of China (nitrate and synthetic organic removal) by the WSWRD.
                                            11-26

-------
This discussion will focus on the Ecuador studies. The three sites in Ecuador were as follows:

    •   Hospital Rodriguez Zambrano—a public hospital in the coastal city of Manta, Ecuador. The
       water treatment system consists of (1) primary disinfection via ozone; (2) filtration; and (3)
       residual disinfection via chlorination (Figure 11-18).
    •   Monteoscuro—a rural community of approximately 150 families located 45 minutes from
       Portoviejo, Ecuador, and fifty miles west of Manta. The treatment plant is composed of three
       treatment processes: (1) backwashable multimedia filtration; (2) primary disinfection via UV
       radiation: and (3) residual disinfection via chlorination (Figure 11-19).
    •   La America—a rural community of approximately 56 homes located near Jipijapa, Ecuador,
       eighty miles southwest of Manta. Four POU units employing either filtration and UV or
       iodine disinfection were tested (Figure 11-20). These units were installed at key locations in
       the community such as the health clinic and the elementary school.

The goal of the project was to demonstrate low-cost, reliable, easy-to-operate POU/POE devices for
treating drinking water in small communities. It was carried out cooperatively between WSWRD, the
United States Agency for International Development, and Hagler Bailly,  a private contractor. The units
were installed in Ecuador between May 1996 and July 1996 (Clark et al. 2000).





66,000
Gallon Tank






D
Oxygen
Concentrator
Ozone /
Generator/
\
3
Wa
,q
<(
/
^
Di im
r, 1
Recirculation Line
-O Float Switch



66,000
Gallon Tank
^Venturi
'Assembly
                 New 36 x 72
                 Backwashing
                   Filters
Pump-*
                                                     Existing Pressure Tanks
                                                Chlorine
                                       i__|ORP   Injection
                                       rej Sensor Pump and
                                                Reservoir
                                     -D
                                                                           Chlorine
                                                                           Analyzer
Figure 11-18. Hospital Rodiguez Zambrano treatment system.
                                             11-27

-------
                    Chlorine
                    Analyzer
               Chlorine
               Injection
               System
                                                                          Well
Figure 11-19. Monteoscuro treatment system.
                       30
                     Micron
                              Pump
Solenoid
5
Micron
L A

1
Micron
k A
                                                                         UV Light
Figure 11-20. 110V/120V filtration/UV system.



Results of Pilot Test

Some of the problems that affected the results of the pilot test were as follows:

    •   The country went through political uncertainty due to elections in August 1996. The presi-
       dent left the country in early 1997, and Congress appointed a new president for a period of 18
       months. There was no local government oversight.
    •   Spare parts for the system in Monteoscuro disappeared from the storage site.
                                             11-28

-------
   •   El Nino caused severe damage in the coastal areas of Ecuador. Monteoscuro was the hardest
       hit. Large mud slides isolated the town for several months.
   •   A large earthquake hit the Province of Manabi (location of all three sites), causing significant
       damage in Monteoscuro. Water pipes were damaged in the vicinity of the site, and the village
       was cut off for several weeks.

Over a period of 14 months (December 1996 to January 1998), water samples were collected twice a
month at the Hospital in Manta and the health clinic in La America. Samples at Monteoscuro were
collected only during the month of November 1997.

The package plant at the hospital Rodriguez Zambrano in Manta performed satisfactorily and was
relatively sustainable because the hospital had the basic financial and technical resources needed to
maintain and operate the system. In addition, problems with power supply and the impact of El Nino
were less severe in the  city than in the nearby rural areas. However, water shortages were frequent
between November and March, and therefore the water had to be supplied by trucks. During 1997, the
Province of Manta suffered a drought that severely restricted the city of Manta's water supply. Because
of resulting water shortages, the hospital was forced to shut off the water treatment system in order to
prevent losing water that was normally consumed in the daily filter backwash. Every time this occurred
and after the city water supply was back to normal, the water storage tanks were supposed to be cleaned
and the whole system chlorinated; however, this procedure was not always implemented. Water samples
were taken from specific faucets in the hospital.

When the water flow from the municipality was sufficient, the treatment system operated properly and
produced the expected result in most cases. Figure 11-21 illustrates the results from the study. The high
total coliform values are associated with periods when the hospital water distribution system was con-
taminated during municipal water shortages and when water was distributed to the hospital by truck
and the treatment system was bypassed.
       o
       o
       E
       ,0
       "o
       O
600 -
500 -
400 -
300 -
200 -
100 -










n_
n
O) O) O)
OM 00 to
d d "~
CD ra c
O -3 ra
—3







^
fo
c
ra
— j







£
^
5







n
•- r-
3) 0)
2 £
ft (0






1
~
h* h*
O) O)
CD T-
i ~"






•
r^ r^
U) O)
2 S
c c
— ) —3
^| Before Treatment
After Treatment





J) O) O) O) O) O)
m o m m o o
«- CO T- T- CO CO
T -^ D) Q. Q. -K
^ -^ 3 CD CD ><
< W W (J







3)
S
o
z







J)
in
o

Figure 11-21. Total coliforms in the water samples at the Manta Hospital.
                                             11-29

-------
As a result of the problems discussed previously, including periodic power outages, the system in
Monteoscuro was operated properly for only one month. When the system was operated properly, it
functioned effectively.

In La America, the POU units operated properly most of the time. However, as seen in Figure 11-22, the
total coliform removal rates were not as high as in Manta. This occurred because there was no residual
disinfection provided by the POU units, and sampling procedures were not properly followed by the
clinic staff. For example it was learned that the spigot at which the samples were taken was contami-
nated (Clark et al. 2000).

Small System Remote Monitoring and Control Technology

EPA is currently evaluating technologies at the T&E Facility that are related to remote monitoring and
control of small drinking water package plant systems and distribution systems. Regulations require all
conventional water treatment operators to provide constant monitoring to assure quality of the treat-
ment processes. Small system operators are under the same reporting and water quality requirements as
the large treatment operators. Constant monitoring of the water quality can provide substantial savings
in costs of time and travel for operation and maintenance. Various package plants at the T&E Facility
and in the field (West Virginia) have been equipped with remote telemetry units (RTUs). The distribu-
tion systems at the T&E Facility are also controlled  and monitored via remote telemetry. EPA is also
monitoring various portions of the distribution system in Washington, D.C. Remote telemetry can sup-
port regulatory reporting guidelines by providing real-time continuous monitoring of the water quality
and reporting the information electronically.

Supervisory Control and Data Acquisition (SCAD A) systems of the past were not always used to their
fullest potential by small systems due to complex operating systems and controls that usually required
specially trained computer programmers or technicians and costly service agreements.  Thus, large (fi-
nancially secure) treatment plant operators typically used the SCADA systems. Usually, the hardware
   700
   600 -
   500 -
_i

o  400 -
i  300 H
O
s
£  200 -
   100 -
Before Treatment

After Treatment
         O)  O)  C7)O)O)O)O)O)O)O)O)O)O)O)O)O)O)  O)  O)  O)  O)  O)  O)
Figure 11-22. Total coliforms in the water samples at La America.
                                            11-30

-------
vendor also provided the SCADA system software with his or her own proprietary-user interface.
Thus, it was almost impossible even for the average programmer or technician to integrate hardware
and software from different vendors. This led to the perception that SCADA systems were expensive
data loggers that did not add cost-saving value to the utility. However, in the last few years, SCADA
system vendors have changed the way they design and fabricate their systems.

The application of SCADA systems to operate, monitor, and control small systems from  a central
location (electronic circuit rider) is believed to be one mechanism that can reduce both MCL and M/R
violations under the SDWA. Filters could be operated more efficiently for particle removal, disinfec-
tant doses altered in real-time in response to varying raw water conditions, and routine maintenance
and chemical resupply scheduled more efficiently. Small independent systems  could contract with an
off-site O&M firm or join with other small systems to hire O&M services while maintaining owner-
ship. This could help small systems enjoy some economies-of-scale that the medium and larger sys-
tems have in purchasing supplies, equipment, and power while also perhaps receiving a more qualified
operator. Although some county-wide public utility  districts may be responsible for  several systems
and already benefit from centralized management, limited staff may be forced to be on the road daily to
simply check each system. In some states, it is required that an operator be on-site when water is being
treated, thus overextending already limited staff in small systems.

Through the use of ultra-fast computers with large memory bases and graphical operating system inter-
faces, the current SCADA technology can be integrated with very little effort and usually with off-the-
shelf, easy-to-learn software. Fairly inexpensive SCADA systems can now collect information and
monitor water quality continuously every few seconds. Several treatment systems spread over a wide
geographical area can be monitored and controlled from a central location by a single operator. The
system operator can monitor,  respond, and adjust the treatment system performance in a matter of
seconds. Automated "smart" systems can identify operating trends and adjust operating parameters to
accommodate most trends. Many of the smart systems  can remedy problems  before technicians are
aware that any problematic situation exists with the treatment system (Haught and Panguluri 1998).

Selection  of a Remote Telemetry System (RTS)

EPA has been evaluating a variety of "small" SCADA  systems that would allow a single qualified/
certified operator to monitor and control the operation of several small treatment systems from a central
location. The following factors should be carefully considered before purchasing an RTS system (Haught
and Panguluri 1998):

   •   Does the water treatment system justify the requirement for an RTS system (is it remotely
       located)?
   •   Is the treatment system amenable (can water quality instrumentation and operational  controls
       "send and receive" data in real-time) to automation?
   •   What types of communication media can be used (phone, radio, cellular, etc.)?
   •   How much automation and control is available on the treatment system?
   •   What type of SCADA system is needed (is the goal to monitor, control, or both)?
   •   How many parameters are going to be monitored and/or controlled?
   •   Are there any specific regulatory monitoring and reporting requirements?

The above factors will determine the need and the basic design of the RTS. Retrofitting a treatment
system for remote operations can be cost prohibitive; many of the small treatment systems currently in
use were not originally  designed for remote operations. Rural areas have little or no  electronic hard-
ware to communicate with a telemetry system. Thus, the cost of upgrading the treatment system for
remote operations could be significant. Therefore, it is essential that the treatment system be fairly
                                             11-31

-------
amenable to automation. Table 11-14 identifies the amenability of packaged treatment technologies
currently used by small water treatment systems to automation and remote control. Membrane tech-
nologies are extremely amenable to automation and remote control and also provide efficient removal
for a wide range of drinking water contaminants.
Table 11-14. Amenability of Treatment Technologies Used for Small Water Systems to
            Automation and Remote Control (Haught and Panguluri 1998)
Source Water
Technology
Amenability for Automation/Remote Control
Ground
Air stripping
Very good to poor
                     Oxidation/filtration
                           Poor
                     Ion exchange
                            Good to poor
                     Activated alumina
                           Poor
Surface
Coagulation /filtration
Poor
                     Dissolved air flotation
                           Poor
                     Diatomaceous earth filtration
                            Good to poor
                     Slow sand filtration
                            Good to poor
                     Bag and cartridge filtration
                            Good to poor
Ground and surface
Disinfection
Very good
                     Corrosion control
                            Good to poor
                     Membrane filtration systems
                            Good to poor
                     ReverseOsmos/nanofiltration
                            Very good to poor
                     Electrodialysis systems
                           Very good
                     Adsorption
                            Good to poor
                     Lime softening
                           Poor
Prior to the purchase of a remote telemetry system or SCADA system, it is also essential to understand
the variability that exists within such systems. Many telemetry and/or SCADA systems may not pro-
vide the  results needed  to justify their purchase. Therefore, once the need for a SCADA system is
justified  and the suitable monitoring and controlling technologies are identified, the following factors
should be carefully evaluated (Haught and Panguluri 1998):

    •   Cost (initial, training, service agreements, and operation and maintenance)
    •   Ease of operation (user-friendly to the operator)
    •   Ease of customization (programmability)
    •   Networking ability (connecting to several remote systems)
    •   Remote operability (ease of remote technical diagnosis)
    •   Scalability (ease of adding monitoring and control devices to the system)
    •   Vendor support (hardware and software upgrades and remote diagnosis)

Currently, there are several commercially available  small-scale SCADA systems in the market. These
small-scale SCADA systems should be further evaluated for the use of open standards. A SCADA
system must be "scalable" to allow for future growth with respect to the number of input and output
channels. These input and output channels are used to communicate with various monitoring and con-
trol devices. The  SCADA hardware must also contain sufficient memory to store the monitored data
for extended periods of time. In case of brownouts or blackouts, the system should normally self-boot
upon resumption of power supply. Along with these basic features, the selected system must have some
of the advanced features, which include
                                             11-32

-------
    •   Call-out feature—This feature allows the system's software to notify appropriate personnel if
       problems develop with a treatment system or water quality. This feature can greatly enhance
       operator response in emergency situations and prevent costly shutdowns and loss of water
       and/or water quality.
    •   Security feature—This feature allows the system to be able to implement security levels of
       access to the treatment plant. Security levels with passwords can deny or allow monitoring
       and/or control access.

Whenever possible, a microprocessor-based unit or a "smart" system should be implemented in most
cases. The  smart systems greatly reduce the cost of on-line communication. The smart systems are
typically more expensive, but the payoff is in savings associated with communication, operation and
maintenance, and travel/repair costs. All operating functionality should be available to be transferred to
the RTS from a remote site. The smart systems also eliminate the need for a dedicated on-line central
computer. In such an implementation, the main computer is used only for periodic monitoring, transfer
of monitoring and reporting data, troubleshooting, and modifications of control parameters.

Capital Costs of Remote Telemetry System  Components

RTSs primarily consist of four main components: hardware, software, communication media, and elec-
tronic instrumentation for control and monitoring. The capital costs of these components are provided
in Table  11-15. The total capital cost for the hardware and software for setting up a "smart" remote
telemetry SCADA system at an EPA test site was approximately $33,250.

    Computer, hardware, software, and upgrades:                     $6,000
    Communication modem and phone line:                          $1,000
    Data collection and transportation terminal:                       $5,200
    Instrumentation for monitoring and control:                      $21,000
    Total capital cost:                                              $33,250

Table 11-15.  Cost Estimates of SCADA  System Components (Haught and Panguluri 1998)
SCADA System Component
Hardware

Software


Communication medium



Instrumentation


Component Option
Main computer
SCADA unit
Operating system
Telemetry system
Data collection & loggers
Telephone
Cellular
Radio
Satellite
Valves
Switch
Sensor
Range of Costs
$1,000 to 3,500
$500 to 30,000
$250 to 750a
$500 to 30,000b
$250 to 8,000
$75 to 125C
$250 to 500d
$1,500 to 3,500e
$20,000 to 75,000f
$25 to 1,5008
$25 to 300«
$350 to 85,000h
  a Operating system software is usually included in the purchase price of a computer.
  b SCADA software is usually included in the purchase price of the hardware.
  0 Monthly service charges are estimated.
  d Activation, roaming, and monthly service are estimated and included.
  e Transmission cost of integrated phone, cellular, radio frequency, and satellite system.
  f Satellite systems cost for transmissions, monthly service, and activation charges are estimated.
  8 Cost per valve and/or switch.
  h Cost per individual sensor or sensor system.
                                              11-33

-------
Summary

Filtration and disinfection of water supplies are highly effective public health practices. In particular,
the WSWRD has conducted extensive pilot- and field-scale experiments on the filtration of particles
and disinfection of various microorganisms while controlling by-products geared specifically for small
community and non-community systems. When applied in the field, the effectiveness of a package
system is highly dependent on site-specific conditions and operator attention. This is a lesson that must
be remembered when designing and setting up a package plant treatment system.

Microfiltration, ultrafiltration, and reverse osmosis filtration systems have been shown to be effective
technologies for the removal of pathogens while being affordable for small systems. New disinfection
systems appear to provide improvements to current systems in handling of chemicals and consistency
of performance. This is an area that is undergoing rapid change. Many organisms are readily removed
and inactivated in the laboratory but, under field conditions, the same effectiveness cannot be taken for
granted. One approach to improve the effectiveness of systems in the field is the use of SCADA sys-
tems to operate, monitor, and even record data from remote locations. Advances in computer hardware
and software capabilities coupled with decreasing prices have brought what was once thought to be a
luxury only affordable by large systems to the point where it is fast becoming a small system necessity.

References

Campbell, S., Lykins, Jr., B. W., Goodrich, J. A., Post, D., and Lay, T. (1995). "Package plants for
   small systems: Afield study." Journal of the American Water Works Association, 87(11), 39-47.

Chapman, P. A. and Rush, B. A. (1990). "Efficiency of sand filtration for removing Cryptosporidium
   oocysts from water." Journal of Medical Microbiology,  32, 243.

Clark, R. M.,  Goodrich, J. A., and Lykins, Jr., B. W. (1994). "Package plants for small water sup-
   plies—The U.S. experience." Journal of Water Supply Research Technology, 43(1), 23-34.

Clark, R. M., Rice, E. W., Haught, R. C., Reasoner, D. J., and Goodrich, J. A. (2000). "EPA's small
   systems disinfection research program," submitted for publication to Journal of Water Supply
   Research Technology.

Federal Register. (1998). "National primary drinking water regulations: Disinfectants and disinfec-
   tion by-products." 63(No. 24;  Dec. 16), 40 CFR Parts 9, 141, and 142, 69390-69426.

Finch, G R., Daniels, C. W., Black, E. K., Schaefer III, F. W., and Belosevic M. (1993). "Dose
   response of Cryptosporidium parvum in outbred neonatal CD-I." Applied and Environmental
   Microbiology, 59, 3661.

Gallo, L. J. (1999). Drinking water treatment technology demonstration: Ecuador 1995  1998.
   Hagler Bailly Inc., April.

Goodrich, J. A. (1999). "Innovative technologies for  small drinking water systems." Proceedings,
   Japan—U.S.  Governmental Conference on Drinking Water Quality Management and Wastewater
   Control, Colorado Springs, CO, July 24-30.

Goodrich, J. A. and Moore, G. T.  (1998). Remote sensing/control for small systems: Literature
   review/assessment. U.S. Environmental Protection Agency, Cincinnati, Ohio, January.

Goodrich, J. A., Lykins, Jr., B. W., and Gordon, P. (1993). "Ozone point-of-entry field applications."
   Proceedings, llth Ozone World Congress, San Francisco, CA, September 1.
                                            11-34

-------
Goodrich, J. A., Adams, I Q., Lykins, Jr., B. W., and Clark, R. M. (1992). "Safe drinking water from small
   systems: Treatment options." Journal of the American Waterworks Association, 84(5),49-55.

Goodrich, J. A., Lykins, B. W., Haught, R. C., and Li, S. Y. (1999). "Bag filtration for small sys-
   tems." Providing safe drinking water in small systems., J. A. Cotruvo, G. F. Craun, and N.
   Hearne, eds., Pan American Health Organizations, World Health Organizations, ILSI Press,
   Washington, D.C., 265-272.

Gordon, G., Gauw, R., Walters, B., Goodrich, J. A., Krishnan, R. A., and Bubnis, B. (1999).  "Chemi-
   cal detail of electrolyzed salt brine solutions." Proceedings, American Water Works Association
   National Conference and Exposition., Chicago, IL, June.

Haught, R. C. (1998). "The use of remote telemetry to complement the operations and maintenance
   of a small treatment system." Proceedings, 1998 American Water Works Association Annual
   Conference, Dallas,  TX, June  20-24.

Haught, R. C. and Panguluri, S. (1998). "Innovations for management of remote telemetry systems
   for small drinking water treatment plants." Paper presented at the First International Symposium
   on Safe Drinking Water in Small Systems, Washington, D.C., May 10-13.

Jacangelo, J. G., Adham, S., andLaine, J. M. (1997). Membrane filtration for microbial removal.
   American Water Works Association Research Foundation, Denver, CO.

LeChevallier, M. W. and Norton, W. D. (1992). "Examining relationships between particle counts
   and Giardia, Cryptosporidium, and turbidity." Journal of the American  Water Works Association,
   84(12), 54-60.

Li, S. Y. (1994). "Cryptosporidium potential surrogate and compressibility investigations for evalua-
   tion filtration-based water treatment technologies." Master's thesis, Miami University, Oxford,
   Ohio, November.

Li, S. Y, Goodrich, J.A., Owens, J. H., Willeke, G.E., Schaefer, III, F.W., and Clark, R.M. (1997).
   "Reliability of non-hazardous surrogates for determination of Cryptosporidium removal." Jour-
   nal of the American Water Works Association, 89(5), 90-99.

Li, S. Y, Goodrich, J. A., Owens, J. H., Clark, R. M., Willeke, G. E., and Schaefer, III, F.  W. (1995).
   "Potential Cryptosporidium surrogates and evaluation of pliable oocysts." Proceedings, 21st
   Annual RREL Research Symposium, EPA/600/R-95/012, 305-308.

Lykins, B. W., Goodrich, J. A., and Hoff, J. C. (1990). "Concerns with using chlorine dioxide disin-
   fection in the USA." Journal of Water Supply Research and Technology-Aqua, 39(6), 376-386.

Lytle, D. A. and Fox, K. R. (1994). "Particle counting and zeta potential measurements for optimiz-
   ing filtration treatment performance." Proceedings, 1994 American Water Works Association
   Water Quality Technology Conference, San Francisco, CA, November 6-10.

MacDonald, J. A., Zander,  A. K.,  and Snoeyink, V. L. (1997). "Improving service to small communi-
   ties." Journal of the American Water Works Association, 89(1), 58-64.

Marinas, B. J., Rennecker,  J. L., Teffey, S., and Rice, E. W. (1999). "Assessing ozone disinfection
   with nonbiological surrogates" Journal of the American Water Works Association, 91(9), 79-89.

National Research Council. (1997). Safe water from every tap. Improving water service to small
   communities. Committee on Small Water Supply  Systems, Water Science and Technology Board,
   Commission on Geosciences,  Environment, and Resources, National Research Council, National
   Academic Press, Washington, D.C.
                                             11-35

-------
Pollack, A. J., Chen, A. S. C., Haught, R. C., and Goodrich, J. A. (1999). Options for remote moni-
   toring and control of small drinking water facilities. Battelle Press, Columbus, Ohio.

Sabran, I, Cross, R., Libby, D., Goodrich, J. A., and Lykins, Jr., B. W. (1992). "Ultrafiltration: One
   answer to SWRT?" Water Conditioning and Purification Magazine, 34(11), 22-26.

U.S. Environmental Protection Agency (USEPA). (1989). "Drinking water; National primary drink-
   ing water regulations; Filtration, disinfection; Turbidity, Giardia lamblia, viruses, Legionella,
   and heterotrophic bacteria; Final rule." Federal Register, 54(No. 124; June 19), 40 CFR Parts
    141 and 142, 27486-27541.

USEPA. (1994). "National primary drinking water regulations; Enhanced surface water treatment
   requirements; Proposed rule." Federal Register, 59(No. 145; July 29), 40 CFR Parts 141 and
    142,  38832-38858.

USEPA. (1999). "25 years of the Safe Drinking Water Act: History and trends." EPA 8J6-R-99-007,
   Office of Water, December 3.

Witt, V. M. and Reiff, F. M. (1966). "Water disinfection technologies for small communities and
   rural areas." Water quality in Latin America: Balancing the microbial and chemical risks in
   drinking water disinfection, G. F. Craun, ed., Pan American Health Organization, World Health
   Organization, ILSI Press, Washington, D.C., 139-168.
                                            11-36

-------
                                    CHAPTER 12

   Modeling Chlorine Decay and the Formation of Disinfection By-Products
                             (DBFs) in Drinking Water1

Introduction

A major objective of drinking water treatment is to provide microbiologically safe drinking water.
The combination of conventional drinking water treatment and disinfection has proved to be one of
the major public health advances in modern times.

In the U.S., chlorine is most often the final  disinfectant added to treated water for microbiological
protection before it is discharged into a drinking water distribution system. However, disinfectants,
especially chlorine, react with natural organic matter (NOM) to form disinfection by-products (DBFs),
which are considered to be of concern from  a chronic exposure point of view.

Drinking water disinfection, therefore, poses the dilemma of a risk tradeoff. Chemical disinfection
reduces risk of infectious disease, but the interaction between chemical disinfectants and precursor
materials in source water results in the formation of DBFs. Although disinfection of public drinking
water has dramatically reduced outbreaks of diseases attributable to waterborne pathogens, the iden-
tification of chloroform, a DBF, in drinking water (Rook 1974; Bellar and Lichtenberg 1974) raised
questions about possible health risks posed by these DBFs. Since 1974, additional DBFs have been
identified, and concerns have intensified about health risks resulting from exposures to DBFs.

All natural waters and even treated drinking water exerts disinfectant demand due to the reactions
with NOM and other constituents in water. Therefore, the applied disinfectant dose must be sufficient
to meet the inherent demand in the treated water, to provide sufficient protection against microbial
infection, and at the same time minimize exposure to DBFs.

Consequently, much research has been invested in attempting to characterize the nature of DBFs and
the conditions that govern their formation in drinking water. One aspect of this research is the devel-
opment of mathematical models for predicting the decay of chlorine and other disinfectants and for
predicting the formation of DBFs themselves.

This chapter reviews current and historical research efforts related to the development of models for
predicting the decay of disinfectants and the formation of DBFs. It focuses on chlorine as a disinfec-
tant and emphasizes U.S. Environmental Protection Agency (EPA) research efforts in this area. The
conditions that govern the interaction of NOM and chlorine and the resulting formation of DBFs are
discussed. Research devoted to models for chlorine decay and the formation of DBFs are reviewed.
The factors that affect exposure to DBFs are  examined,  and  EPA field research studies that have
driven the current research on chlorine decay and DBF formation are presented. The development of
EPANET, a state-of-the-art public sector water quality/hydraulic model, is reviewed, along with the
evolution of numerical modeling techniques. The topic of storage tanks and their impact on water
quality and the public policy issues associated with this research is also discussed.
'Robert M. Clark, Lewis A. Rossman, Mano Sivagesean, Kathleen Schenck: ORD/NRMRL/
WSWRD, AWBERC Mailstop 689, 26 West Martin Luther King Dr., Cincinnati, OH 45268.
Corresponding Author: Robert M. Clark, 513-569-7201, clark.robertm@epa.gov.
                                             12-1

-------
Chemistry of Disinfectants in Water

In order to understand the nature of the risk tradeoffs associated with the loss of chlorine residuals and
the formation of DBFs, it is necessary to review some of the chemistry involved in this process.

Chlorine dissolved in water yields the following (White 1999):

                               C12 + H2O -» HOC1 + H+ + Cl                     (12-1)

HOC1 generally reacts with the various components that make up chlorine demand as follows:

                                HOC1 + Cl.   ,-> products                      (12-2)
                                          demand    r                             ^    /

Consequently, chlorine is consumed and chlorine residuals dissipate, which may then result in micro-
biological regrowth and which, in turn, may increase the system's vulnerability to contamination. Three
factors which frequently influence chlorine consumption in drinking  water are (1) reaction with or-
ganic and inorganic chemicals (e.g., ammonia, sulfides, ferrous iron, manganous ion, humic material)
in the bulk aqueous phase; (2) reactions with biofilm at the pipe wall; and (3) consumption by the
corrosion process.

A by-product of chlorination is the formation of total trihalomethanes (TTHMs) and other DBFs in
waters containing organic precursor compounds, such as humic and fulvic acid substances. Generation
of TTHMs has been shown to be a function of various water quality parameters and chlorination con-
ditions including total organic carbon (TOC), the  type of organic precursor,  chlorination level, pH,
temperature, bromide level, reaction time, and UV-254 absorbance. TTHMs are also regulated under
the Safe Drinking Water Act (SDWA) and its amendments of 1986 and  1996 (Amy et al. 1987; Clark et
al. 1996b).  Chlorine decay in distribution systems is generally considered to consist of two compo-
nents. One component is wall demand, while the other is associated with decay in the bulk phase of the
water (Clark etal. 1993a).

Chlorine demand and the formation of DBFs are influenced by both the condition of treatment as well
as the constituents in the raw and treated water. Treatment processes change the concentration of drink-
ing water constituents and are likely to change the composition and characteristics of the water distrib-
uted to the consumer. Therefore, when modeling chlorine decay and the formation of DBFs, it is impor-
tant to find modeling-input parameters which can reflect these changes in water quality characteristics.
Some common surrogate parameters forNOM are TOC and spectral absorbance. TOC concentration is
indicative of the mass  of material, whereas spectral absorbance relates more to specific structure and
functional groups. DOC is the dissolved fraction of TOC. ForNOM, the most commonly used spectral
absorbance is the ultraviolet (UV) absorbance (UVA) at a wave length of 254 nm (UVA) which mea-
sures conjugated double bonds. The specific UVA (SUVA) is the ratio of UVA to the DOC. Its values
give an indication of the NOM's nature, with higher values indicating a more aromatic character (Amy
1993).

For most waters, the reactions of chlorine with NOM make up the majority of the chlorine demand.
Chlorine also reacts with various inorganic compounds. For example, chlorine reacts with ammonia to
form different species of chloramines. Aqueous chlorine converts Br into hypobromous acid (HOBr),
which then attacks organic compounds to form brominated DBFs.

Chlorine will oxidize soluble iron and manganese to the insoluble ferric and manganic forms, respec-
tively. Hydrogen sulfide reacts with chlorine, and successful removal of H2S by conventional treatment
with prechlorination has been reported.
                                            12-2

-------
Normally, the reactions between disinfectant and NOM make up the majority of the disinfectant de-
mand and the subsequent formation of DBFs. The demand caused by inorganic or microbial demand is
much less than the demand associated with NOM.

Some of the factors that influence both the formation of DBFs and the demand for chlorine are as
follows:

    •   Disinfectant dose: several studies have shown that the formation of DBFs increases with
       chlorine concentration.
    •   Reaction time: a longer reaction time generally leads to both higher disinfectant demand and
       greater DBF formation.
    •   pH: For chlorine reactions, a shift in pH has little effect on chlorine demand. However,
       increase in TTHM formation has been observed with increases in pH. The sum of HAA6 (the
       sum of six of the nine haloacetic acid [HAA] species) has been found to decrease when pH
       increases.
    •   Temperature: An increase in temperature has been shown to cause an increase in the rate of
       both DBF  formation and chlorine demand.

As mentioned previously, chlorine residuals are important to ensure the microbial safety of distributed
drinking water. Requirements for disinfection of drinking water are defined in the Surface Water Treat-
ment Rule (SWTR). According to the SWTR, treatment including disinfection must reliably achieve at
least a 3-log (99.9%) removal and/or inactivation ofGiardia lamblia cysts, and a 4-log (99.99%) reduc-
tion and /or inactivation of viruses prior to the delivery of water to the first consumer (Clark and Feige
1993). A control parameter frequently considered and specified in the SWTR is the CT (the product of
the  disinfectant residual concentration (mg/L) and contact time (min) measured at peak hourly flow)
concept. Contact time is measured from the point of disinfectant application to the first customer (Clark
and Feige 1993).  Different disinfectants require different CT values because of their variability of
action against different types of organisms. For example, chlorine is relatively ineffective against some
protozoan, such as Cryptosporidium, but is generally very effective against most bacteria and viruses.

The SWTR requires that a minimum disinfectant level be maintained in all parts of a distribution
system. Therefore, it is important that the factors that influence chlorine decay be identified  and that
models that can reliably predict chlorine residual levels in treated and distributed water be developed.

Modeling the Decay of Chlorine Residuals

A number of investigators have conducted research into the development of models to predict  chlorine
decay in drinking water. In one of the earliest attempts to model chlorine decay, Feben and Taras (1951)
developed the following model:

                                         D=Djn                              (12-3)

In Equation 12-3,  Dt = the chlorine consumed at time t (hr), Dl = the chlorine consumed after 1 hour,
and n is a constant characteristic of a given water. The 1-hour chlorine demand and n must be deter-
mined experimentally for a given water.

Haas and Karra (1984) investigated several models to describe chlorine decay, including
    •   First-order decay
    •   Power-law decay (wth order)
    •   First-order decay with stable components
                                             12-3

-------
    •   Power-law decay with stable components (/rth order)
    •   Parallel first-order decay

They found that the parallel first-order decay model yielded the best results. This model assumes that
there are two constituents in water that react with chlorine: (1) fast-reacting components, which exert
an initial decay; and (2) slow-reacting components, which are responsible for long-term chlorine de-
mand. This model was applied by Vasconcelos et al. (1997) in a project conducted jointly by EPA and
the American Water Works Association Research Foundation (AWWARF) and is discussed later in this
chapter.

A model by Quails and Johnson (1983) described the short-term chlorine consumption by fulvic acids
during the first 5 minutes of a reaction. This model (Equation 12-4) was originally developed for
cooling water systems, but was then applied to disinfection of natural waters:

                               -dCl/dt = ^[CIHFJ + £2[C1][F2]                    (12-4)

The chlorine decay is described by the sum of two first-order equations, in which the first part describes
a rapid decay within the first 30 seconds, and the second simulates a slower decay from 30 seconds to
5 minutes. In Equation 12-4, [Cl] is the  free residual chlorine, kl and k2 are rate constants for the fast
and slow reactions, respectively, and [FJ and [F2] are the concentrations of reactive sites on the fulvic
acids for the fast and slow reactions, respectively.

Hao et al. (1991) demonstrated that chlorine reacts with organic material as well as inorganic material
by evaluating the kinetics of Mn(II) oxidation with chlorine and examined the effects of chlorine dose
and the presence of complexing agents on chlorine demand. They found that Mn(II) oxidation is facili-
tated by excess chlorine at pH = 8. An autocatalytic model analogous to that for Mn(II) oxidation by O2
was developed in which the major mechanism for Mn(II) removal was the heterogeneous Mn(II) ad-
sorption onto newly precipitated MnO2. The same model also describes the  effects of pH and the exter-
nal addition of Mn02 on Mn(II) removal.

Ventresque et  al. (1990) conducted a study at the Choisy-le-Roi water treatment plant near Paris to
identify the organic components that react with chlorine. The plant consists of preozonation, coagula-
tion, sedimentation, sand filtration, ozonation, granular activated carbon (GAC) bio-adsorption, and
post-chlorination. The water is dechlorinated before being distributed to the consumer.  The authors
applied a second-order kinetic model to the long-term chlorine demand data. They analyzed the kinet-
ics at every stage of the plant and found that the initial chlorine demand and the kinetic constants of
ozone/GAC -treated water were always lower than those obtained from sand-filtered water. These re-
sults demonstrated the effect of activated carbon on the removal of organic matter.

Jedas-Hecart et al. (1992) attempted to identify the organic compounds that react with chlorine from
the Seine River entering the final treatment stage at the Choisy-le-Roi plant. They studied the chlorine
consumption kinetics of samples of water taken after overall treatment.  They divided the chlorine
decay into two phases. An initial phase of immediate consumption during the first 4 hours was called
the initial chlorine demand. The second, slower consumption phase  after the first 4 hours was defined
as the long-term demand (LTD). The LTD was interpreted with the following kinetic equation:
where x = chlorine consumption after 4 hours, k = the rate constant, a = the total residual chlorine at 4
hours, b = maximum potential chlorine demand, n = stoichiometry, and (3 = partial orders of reaction.
                                             12-4

-------
This model cannot be applied from time zero of chlorination because the LTD obeys a different reaction
order than the initial chlorine demand.

The Water Treatment Plant (USEPA 1992) model described chlorine decay by dividing the decay curve
into three components. These included an initial (t less than 5 minutes) reaction, a second-order reac-
tion (5 minutes less than t less than 5 hours), and a first-order reaction (t greater than 5 hours). These
equations apply only when the initial chlorine/TOC ratio is >1:1.

Zhang et al. (1992) conducted a study of chlorine modeling in sand-filtered water (before post-chlori-
nation) at the Macao treatment plant which draws water from the estuary of the West River (the main
stream of the Pearl River in South China). The study indicated that the chlorine consumption in sand-
filtered water can be divided into two phases: an initial chlorine consumption during the first hour
which corresponds to the contact time in the reservoir of the treatment plant, and a long-term chlorine
consumption after 1 hour in the network. This second component is interpreted in terms of an apparent
first-order equation. According to the  experiments performed on steel and asbestos cement pipes (di-
ameter greater than 250 mm), the chlorine consumption by pipes is negligible. The authors suggested
more experiments to be performed to verify the eventual influence of water velocity and diameter of
pipe on the chlorine consumption by pipe itself. The chlorine disappearance in the network of Macao
can be modeled as a first-order reaction.

To describe the entire disinfectant reaction for one ground water treated in a particular plant, Lyn and
Taylor (1993) calculated the chlorine residual (CLR) as a function of chlorine dose, DOC, temperature,
and time, using an empirical constant applicable only for that particular water.

Dugan et al. (1995) proposed a saturation model in order to predict the entire chlorine decay curve with
one equation.  TOC  was chosen as the predictive water quality parameter for the  saturation model
because it represents the compounds exhibiting chlorine demand.

Chambers et al. (1995) conducted a study to test the validity of the exponential decay expression for
free and total chlorine modeling using two sample networks and proprietary models. A sampling pro-
gram was devised to collect information to calibrate the models. Rate constants for free and total chlo-
rine were calculated. The results showed that the exponential decay model is appropriate for modeling
chlorine in distribution systems and that it is possible to produce useful water quality models. They also
showed that the rate constants for the water in the network were different from the rate constants
collected in bench-scale  experiments.

EPA Research Activities

The Water Supply and Water Resources Division (WSWRD) of EPA has been very active in conduct-
ing research into the factors that  affect chlorine decay in drinking water. One of the first  projects to
investigate the feasibility of modeling water quality and chlorine decay in drinking water distribution
systems was conducted under a cooperative agreement initiated between the North Penn Water Author-
ity in Lansdale, Pennsylvania, and EPA (Clark and Coyle 1990; Clark et al. 1988a). The project pro-
vided the basis for development of a water quality model called the Dynamic Water Quality Model
(DWQM) which was applied to several service areas in the South Central Connecticut Regional Water
Authority (SCCRWA). An extensive field sampling study was conducted as part of the model valida-
tion and verification. Chlorine demand was  calculated according to a first-order decay assumption,
which is defined as follows:

                                         C=C0e^                              (12-6)
                                             12-5

-------
where C = the concentration at time t, C0 = initial chlorine concentration, k = decay rate in min^1, and
t = time in min.

It is clear that, as dissolved chlorine travels through the pipes in the network, it reacts with NOM in the
bulk water and with biofilm and tubercles on the pipe walls or with the pipe wall material itself (Clark
et al. 1993a). This reaction results in a decrease in chlorine residual and a corresponding increase in
DBFs, depending on the residence time in the network and the holding time in storage facilities. An
early study designed to address these issues using a complete water quality and hydraulics model was
conducted by EPA in collaboration with the  North  Marin Water District in California (Clark  et al.
1994). As a follow-on to the North Marin study, Vasconcelos et al. (1997) investigated the factors
leading to loss of chlorine residual in several water distribution systems. Kinetic rate equations describ-
ing the decay of chlorine were developed, tested, and evaluated using data collected in field-sampling
studies conducted  at these water utility sites.  These  studies are discussed in more detail later in this
chapter.

Clark (1998) developed an equation for chlorine decay based on the concept of competing reacting
substances and on the assumption that the balanced reaction equation can be represented by

                                        aA + bB^pP                            (12-7)

In Equation 12-7, if A and B are the reacting  substances, a and b are the proportion  of reacting sub-
stances, and P is the product of the reaction, then the rate of reaction is given by
or

                                        d-^- =  -kBCACB                           (12-9)


or

                                        ^ = kpCACB                          (12-10)

Since both CA and CB are changing with time, a relation connecting them is written in order to integrate
the differential equation. If CA  and CB represent the initial concentrations of A and 5, respectively, at
t = 0 and x represents the concentration of A that has reacted, then the concentration of B that has
reacted is given by bx/a. Consequently,
                                         C=C. -x                             (12-11)
                                          A    AQ                                ^      '
and
From Equation 12-12

                                                                                (12-13)
                                              12-6

-------
and

                                        ~dCA= —                           (12-14)


By substituting in Equation 12-8 and rearranging:


                                  (CA-XHCAB-bx/a)  -  W                  <12-15)
Integrating Equation 12-15 and making the appropriate substitution yields
                                                                               (\2-\6)
                                                                               v      >
                                           ,   — _ r
                                          A   \-Re-ut

In Equation 12-16, CA is the concentration of free chlorine. Rewriting Equation 12-16 yields

                                        Cl(t}= ^^Re.ut                        (12-17)


where C\(f) = the chlorine concentration in mg/L at time t; R (dimensionless), K (mg/1), and u (min^1)
are parameters to be estimated; and t = the time of reaction in minutes. In Equation 12-17, the value for
the rate constants can be rewritten as follows:

                                        u=M(\-K)                            (12-18)

where
                                       M =   Aa  ^                            (12-19)


Equation 12-16 was applied to a series of data sets collected from the Vasconcelos et al. (1997) study.

Clark and Sivaganesan (1998) utilized the equation developed by Clark (1998) to predict chlorine
decay and TTHM formation in a number of field and laboratory data sets. The parameters for Equation
12-16 are estimated using regression analysis. Predictive equations were developed for the parameters
K., M, and R based on initial chlorine concentration (CA ), pH, TOC, and temperature (°C). The esti-
mated parameters for Equation 12-16 are as follows:

                          K = e°32(C. )-044(TOC)° 63(pH)-°29(Temp)°14              (12-20)
                          K = e149(C^)-°48(TOC)° 18(pH)° 96(Temp)°28              (12-21)

and

     Loge(M) = -2.46 - (0.19 TOC) - 0.14 pH) - (0.07 Temp) + (0.01 Temp*pH)      (12-22)

The estimated Model ,R2s are 0.71, 0.78, and 0.42, respectively, for these equations. The model was
validated against data collected from two field studies. Using standard statistical techniques, the upper
and lower 95% confidence intervals were calculated for each parameter.

Rossman et al. (1999) examined the factors that characterize the reaction conditions affecting treated
water in a distribution system. Some of these factors are the pipe wall material, such as  corrosion
                                              12-7

-------
products and biofilm slime, which can exert a significant chlorine demand. If chlorine is being con-
sumed within  a pipe by non-DBP-producing materials (such as ferrous corrosion products), then
there is less available to react with the water's NOM. However, iron tubercles have also been shown
to contain organic material that might include DBF precursors.  Much of the study was devoted to
studying the rate at which DBFs form in a pipe as compared to a glass bottle.  One of the conclusions
from the study was that the rate constants for chlorine decay in the pipe were an order of magnitude
higher than in the bottle.
As an extension to their previous research, Clark and Sivaganesan (in press) hypothesized that two
competitive reactions would adequately describe chlorine decay in raw and finished water.  One reac-
tion was assumed to represent the fast-reacting components associated with chlorine decay, and the
second reaction was assumed to represent the slower-reacting components. In order to test this hypoth-
esis, a model consisting of two competitive reactions was developed as described below:

                                                                               (12.23)
                                                                               (12.24)


where CA is the free chlorine residual reacting with the collection of rapidly reacting components CB ;
CA is the free chlorine residual reacting with the collection of more slowly reacting components
CB ; Pl and P2 are a collection of the by-products of the two reactions; and av brpr ay by andp2 are the
stoichiometric coefficients. These equations can be used to quantify the  fraction of initial chlorine
being utilized by the fast- and slow-reacting components.
                                            1       O
The expressions below describe the change in CA and CA with time (Clark and Sivaganesan 1998):

                                        l-
                                                                               (1225)
                             2       C12Q(\-R2}
                            CA(^~  l_R^-(i-R2)k2t                              (12-26)

where C/0 and C/0 are the initial concentrations of CA  and CA ; CA (f) and CA(t) represent the change
in the  concentration of CA and  Cj with time; kv &2, R^, and R2 are parameters in Equations
12-25 and 12-26; and t represents time. The total initial chlorine concentration at time = 0 is

                               C/o = C/o + C/o                                  (12-27)

                                              1        O
where C/0 is the total initial chlorine residual. IfC/0 and C/^are the initial concentrations of chlorine
reacting in Equations 12-23 and 12-24, then

                               C/o^C/o-C/o1                                  (12-28)

The equation for the complete reaction is assumed as the sum of Equations 12-25 and 12-26 or

                                               (C/o - C/J)(1 - R2)
                            1 - R e(l'R^klt         1 - R e(l'R2)k2t                 (12-29)
                                             12-8

-------
Given that C/0 is known and that C/Q , kr k2, Rr and ,R2 are unknowns, Equation 12-29 yields a five-
parameter equation as follows:
                       ()            -                    -
where C/(/fj is the residual chlorine at t hours, and Z = (Cln  /C10), kr kT Rr and R2 are unknown
parameters. All the model parameters are positive, and Z cannot be larger than 1 .

The SAS procedure NLIN was used (SAS 1990) to estimate the model parameters. Since there is more
than one solution to the above model and since the solutions depend on the initial values, model param-
eters were estimated in three steps to stabilize the solution. First, residual chlorine (Cl) values from the
first 60 minutes (generally 4-6 data points) were used to estimate the parameters Rl and kl in Equation
18. The next step was to fix kl at its estimated level and then  use all the residual chlorine values to
estimate Z, Rr R2, and k2 in Equation 12-30. Finally, the estimated Z, Ry and k2 values from Equation
12-30 were fixed, and the remaining model parameters Rl and kl were re-estimated using data from the
first hour of the experiment.

Thirty seven raw water data sets and twelve treated water data sets were used to develop a general
model. The five model parameters were then regressed against TOC level, initial UVA level (UVAQ),
initial chlorine level (C/0), pH, initial bromide level (Br~\ temperature (Temp) in °C, and alkalinity in
mg/L (ALK). The following general multiplicative model was used for each of the parameters: kr k2,
RrR2, andZ/(l-Z):

                7 = c(TOC + 1 )d(UVA + 1 )e(Cl0 + 1 /(pFTF(5r- +  1 )A(Temp)!( ALK>   (12-31)

where 7 = the parameter value, and c, J, e,/ g, h, /', andy are the exponents. Equation 12-31 can be
rewritten as

                log(7) = log(c) + dog(TOC + 1) + elog(UVA + 1) +/log(C/0+ 1) +
                      glog(pH) + h\og(Br-+ 1) + /log(Temp) +ylog (ALK)          (12-32)

The SAS procedure "REG" is used to estimate the model parameters in Equation 12-32 (SAS 1992).
This model guarantees that the predicted parameters are positive. Only parameters at the 5% level of
significance were included in the model. Influential analysis was used to identify the data points which
have a major impact on the parameter estimates. If the absolute value of the standardized residual of a
data point was larger than 3, then that point was not included in the statistical analysis for any of the five
model parameters (Equation 12-30). The resulting parameter-estimating equations are as follows:

               ^ = e658(TOC + I)266 (UVA+ 1)763(C/0+ I)-325 (pH)-145  (Bt-+ I)006   (12-33)

               Rl = e-356(TOC + I)168 (UVA + 1)394(C/0+ I)-1-68 (pH)105 (Temp)069   (12-34)

                  Z/(l - Z) = e4-94 (UVA + I)2 89 (C/0 + 1)-° -57 (pH)-1 16 (Temp)-079      (12-35)

    k2 = e^-S3 (TOC + I)-243 (UVA + I)-771 (C/0 + I)3 63 (Br+ l)-°32 (Temp)-°31(ALK)° 14 (12-36)

                          R2 = e°-48(TOC + I)1'81 (C/Q+ I)"1'82 gO.03.Temp              (12-37)

The model R2 are 0.60, 0.67, 0.57, 0.58, and 0.52, respectively. The model was applied to various data
sets in order to test its application, and it was found that, in general, the model fits the experimental data
well.


                                             12-9

-------
Modeling the Formation of DBFs

Trusell and Umphres (1978) reviewed the effect of preozonation, bromide, pH, and chlorine dose on
the formation of TTHMs in natural waters and proposed a kinetic model that describes their formation.
They concluded that some of the factors that might influence the rate of the TTHM reaction are pH,
temperature, the level of precursor, the level of chlorine, and the level of bromide ion before chlorine
addition. They proposed two equations — one describing the rate of chlorine (Cl) consumption and one
describing the rate of reduction of precursor or, conversely, the rate of TTHM production. Assuming
that the reaction between chlorine residual and aquatic humic material is related to the concentration of
each, a simple relation is obtained for the rate of chlorine consumption.

                                                                              O2-38)
If it is assumed that the action of the chlorine does not significantly reduce the total concentration of the
humic precursor, then the following equation represents the rate of TTHM production and is first order
with respect to chlorine residual:

                                dTHM     dC  .  _.   _.                     (12-39)
where m is the order of reaction with respect to the precursor concentration and C is the concentration
of the organic precursor. The authors concluded that there are a number of factors of importance in
describing the formation of TTHMs including the nature of aquatic humus, the influence of preozonation
on TTHM formation, the influence of bromide, the influence of pH, and the influence of chlorine dose.

Kavanaugh et al. (1980) developed a two-parameter kinetic model for predicting THM formation in the
distribution system following post-chlorination. They hypothesized that THM formation can be de-
scribed by the overall stoichiometric expression

                                       3A+B^C                           (12-40)

where A = HOC1, B = TOC, and C =TTHM, while the nth overall rate constant for the reaction is k .
               >          '              >                                                  «
Based on this expression, three moles of hypochlorous acid react with one mole of carbon in the or-
ganic precursor material to form one mole of TTHM. The rate expression for the formation of TTHM
is given by
assuming that the rate of formation is first order with respect to TOC and mth order with respect to
HOC1. The chlorine concentration A can then be related stoichiometrically to C by defining a TTHM
yield as the moles of TTHM formed per mole of C12 consumed. The rate expression then becomes

                                   dC            3C                          (12-42)


When the free chlorine is exhausted


                                            3C
                                            AQ
                                            12-10

-------
the yield/can be determined by measuring the THM concentration when A = 0. The equation is there-
fore a three-parameter kinetic model with kn as the rate constant, m the reaction order with respect to
C12, and/the TTHM yield which must be determined empirically for the particular system under inves-
tigation.

Amy et al. (1987) discussed the formulation and calibration of several models for predicting TTHMs in
untreated natural waters subjected to chlorination. Their general approach was to analyze specific por-
tions of a large database derived from several natural water sources in order to isolate the effects of a
given parameter on TTHM formation. Two general strategies were used in formulating the models:
multiple linear regression models using logarithmic transformations of both independent and depen-
dent variables and multiple nonlinear regression models, which were also developed. Both models
assume that a chlorine residual is maintained throughout a 168-hour reaction period and that TTHMs
continuously increase with time. Variables such as UV absorbency, chlorine dose, temperature, and
TOC were used to predict TTHMs.

McKnight and Reckhow (1992) investigated the reactions of specific ozone by-products (OBPs) with
chlorine and chloramines through evaluation of the kinetics and stoichiometry of chlorine demand and
total organic halide (TOX) formation as a function of pH. All of the compounds studied had a carbonyl
(C = O) functionality, causing the types of chemical reactions to be similar. Simple compounds were
chosen from among the carbonyl compounds, aldehydes, ketoaldehydes, and keto-acids in order to
study the relationships between structural characteristics  and reactivity towards chlorine and chlora-
mines. Three general  classifications of alpha substituents were made: compounds bearing (1) an alpha
methyl group; (2) methylene group; or (3) other moieties  (hydrogen, hydroxyl, carboxylic acid, etc.).
Correlations were developed between these characteristics and chlorine consumption behavior to help
in understanding the significance of OBP formation in distribution systems (i.e., persistence of OBPs,
formation of chloroform or other DBFs, etc.).

Little or no  TOX was produced upon chloramination of all the  model compounds studied in this
research. Results suggest that acetaldehyde, methyl glyoxal, and pyrovic acid could be important chlo-
roform precursors in  chlorinated systems under certain conditions. The higher aldehydes studied re-
acted slowly with free chlorine and produce minor amounts of TOX. The keto-acids reacted rapidly
with both free  and combined chlorine. Chloramination of the model OBPs studied resulted in little
measurable TOX formation. The rates for chlorination and chloramination at pH 7 were comparable for
all three compound classes, but the rates of chlorination increased dramatically with pH. Chloramination
rates appear to be only weakly dependent upon pH. These  results suggest that aldehydes may persist in
distribution systems at low (less than 2 mg/L) chlorine doses and neutral pH, but can undergo signifi-
cant decomposition at higher chlorine doses and pH. The keto-acids are likely to react rapidly at low
chlorine doses and pH 7 or greater. These compounds may also be generated in distribution systems by
the reactions of disinfectants with NOM, and their by-products may be sources of assimilable organic
carbon (AOC); thus, their presence in a distribution system may be of concern.

Harrington et al. (1992) developed a computer program to simulate DBF formation, removal of NOM,
inorganic water quality changes, and disinfectant decay in water treatment processes. Equations were
developed that simulate the formation of TTHMs and removal of TOC and UVA by alum coagulation,
as well as changes in alkalinity and pH. Model simulations were  compared with limited sets of ob-
served values. The central tendency of the model was to underpredict finished-water pH by  4 percent,
finished-water TOC by 7 percent, and simulated distribution system TTHMs by 20 to 30 percent.

Shukairy et al.  (1994a) conducted a study with the objective of better understanding the chlorination
reactions of organic matter by investigating the formation of halogenated DBFs from three molecular-
sized (MS) fractions as follows: greater than 3000, 1000-3000, and less than 1000 daltons.  The study
                                             12-11

-------
was conducted in two phases. In Phase I, the impact of ozonation and biotreatment on DBF precursors
under variable reaction conditions was evaluated, and in Phase II, the chlorination kinetics and reactiv-
ity of these fractions under constant organic and inorganic precursor concentrations was investigated.

The impact of organic and inorganic precursor limitations on the reactivity and kinetics of halogenated
DBFs was also evaluated. It was found that the less-than-lK MS fraction was the most biodegradable
fraction and the  1-3K MS fraction  the least. Preozonation resulted in chemical transformation of the
organic matter, resulting in an increase in biodegradability in all fractions and in a decrease in the
reactivity to subsequent chlorination. Biotreatment, with and without ozonation, resulted in equivalent
removal of the DOC and the precursor compounds, as no selectivity was observed. They found that,
under variable precursor concentrations, the speciation of the THMs was dependent on the bromide-
to-DOC ratio. Increases in this ratio resulted in a  shift in speciation to the bromo-substituted DBFs,
irrespective of the fraction  or the treatment. Precursor limitation did not affect the reaction kinetics
significantly. However, the yield and the reactivity were affected.  Speciation depended on the available
organic matter. The less-than-lK MS fraction, under constant precursor concentrations, exhibited the
fastest chlorination kinetics as measured by the higher chlorine demand, TTHM, and HAA6 kinetics.
The less-than-lK fraction also exhibited the highest reactivity to bromo-substitution.

Cowman and Singer (1996) investigated the effect of bromide ion on the distribution of HAA species
resulting from the chlorination and chloramination of waters containing aquatic humic substances. Aquatic
humic substances were extracted from both a surface water and a ground water. They were chlorinated
and chloraminated at pH levels of 8 and 6 in the presence of bromide concentrations ranging from 0 to 25
(ig/L. The samples were analyzed for all nine of the HAA species containing bromine and chlorine. Stan-
dards for bromodichloroacetic acid and dibromochloroacetic acid were synthesized for use in this study. It
was found that bromochloro-, bromodichloro-, and dibromochloroacetic acids formed easily and consti-
tuted at least 10% of the total HAA concentration  in waters  containing as little as  1.2 (*M (0.1  mg/L)
bromide. At concentrations normally found in raw drinking water, the mixed bromochloro HAA species
were major components of the total HAA concentration. Among the mono-, di-, and trihalogenated forms,
the distribution of HAAs appeared to be independent of bromide concentration.

Shukairy and Summers (1992) conducted a study to examine the impact of ozonation and biotreatment
on organic precursor characteristics established by evaluating DBF formation, speciation, and kinetics
under constant DOC, bromide, and chlorination conditions. The following observations were made:

   •   Use of ozone resulted in a significant change in the characteristics of the organic  matter.
   •   Treatment,  ozonation, or biotreatment decreased chlorine demand, indicating selective oxida-
       tion of the organic precursors.
   •   In nearly all cases, treatment resulted in a decrease in DBFs formed after chlorination. In
       most cases, TOX and total THM formation decreased more by ozonation and ozonation/
       biotreatment in comparison to biological treatment, indicating that chemical oxidation of
       organic matter decreased its reactivity.
   •   HAA6 formation decreased the most by biotreatment, with and without preozonation.
       Biotreatment appears to be selective for HAA precursors.
   •   Spectral absorption coefficient (SAC) decreased significantly after ozonation, resulting in
       less available reactive aromatic unsaturated organic precursors than in the control, even when
       the DOC concentration was held constant. The reactive aliphatic (acetyl-)containing precur-
       sors  created after ozonation seem to favor bromine substitution over chlorine substitution.
   •   Precursor limitation is very  important in determining speciation. In the case where the reac-
       tion is precursor limited, bromine substitution is faster than chlorine substitution  and will
       govern the DBF speciation. The formation of chloro-substituted DBFs continues  as long as
       there are available precursors and thus  controls DBF distribution. They found that, under
                                             12-12

-------
       conditions of constant bromide, DOC, chlorination conditions, and holding times,
       biotreatment did not show any selectivity for DBF precursors, that is, DOC and the precur-
       sors were removed to the same extent.

Shukairy (1994) and Shukairy and Summers (1996) conducted a study to examine the impact of ozonation
and biotreatment on the organic precursor characteristics by evaluating DBF formation, speciation, and
kinetics under constant DOC, bromide, and chlorination conditions. The following observations were
made:

    •   Ozonation resulted in a significant change in the characteristics of the organic matter. SAC
       decreased while DOC concentration was nearly unchanged, indicating a decrease in UV-
       absorbing functional groups and the formation of saturated and aliphatic acetyl compounds.
       Biotreatment resulted in equivalent removal of both SAC and DOC.
    •   Treatment, ozonation, or biotreatment decreased chlorine demand, indicating selective oxida-
       tion of the organic precursors, either chemical or biological.
    •   In nearly all cases, treatment resulted in a decrease in DBFs formed after chlorination.
    •   In most cases, TOX and total THM formation decreased more by ozonation and ozonation/
       biotreatment in comparison to biological treatment, indicating that chemical oxidation of
       organic matter decreased its reactivity.
    •   HAA6 formation decreased the most by biotreatment, with and without preozonation.
       Biotreatment appears to be selective for HAA precursors. However, such a conclusion is only
       tentative as the decrease in HAA concentration could be due to a shift to the other three more
       bromo-substituted HAA species that are not quantified.
    •   Speciation was affected most by ozonation. A shift to bromo-substituted species occurred
       after ozonation and after combined ozonation/biotreatment.
    •   No such shift was observed by biotreatment alone. Bromine incorporation factors increased
       significantly with ozonation. The  effect was most pronounced at 6 hours. As the holding time
       was increased, more TTHM, Cl and HAA6-C1 were formed, increasing DBF formation. The
       relative increase in DBP-Br with increasing holding time was much smaller.
    •   SAC decreased  significantly after ozonation, resulting in less available reactive aromatic
       unsaturated organic precursors than in the control, even when the DOC concentration was
       held constant. The reactive aliphatic (acetyl-)containing precursors created after ozonation
       seem to favor bromine substitution over chlorine substitution. For the THMs and quantifiable
       HAAS, DBP-Br formation increased while DBP-C1 formation was much less, relative to the
       control, because of decreases in the organic precursors.
    •   Precursor limitation is very important in determining speciation. In the case where the reac-
       tion is precursor limited, bromine substitution is faster than chlorine substitution and will
       govern the DBF speciation. The formation of chloro-substituted DBFs continues as long as
       there are available precursors and thus controls DBF distribution.

Shukairy and Summers (1996) found that, under conditions of constant bromide, DOC, chlorination con-
ditions, and holding times, biotreatment did not show any selectivity for DBF precursors, that is, DOC
and the precursors were removed to the same extent. Ozonation, however, had more of an impact on the
organic matter characteristics, as observed by decrease in DBF formation (reactivity) and a shift to the
bromo-substituted compounds.  Precursor limitations are important in assessing DBF speciation;  both
DOC and SAC are important parameters that should be considered in evaluating organic precursors.

Shukairy (1998) investigated ozonation and biological treatment as a means of controlling the forma-
tion of DBFs as measured by TOX and purgeable organic halides (POX). Organic matter from a ground
water and a river water source was used.  Chlorine or chloramines were used as the final disinfectant.
Chloramination produced significantly fewer organic halides, especially POX, compared to chlorina-
                                             12-13

-------
tion. With both disinfectants and for both sources of organic matter, the nonpurgeable organic halide
formation rate was found to be much faster than that of POX. Preozonation decreased the amount of
organic halide formation by 10 to 40% upon subsequent chlorination. With chloramines, preozonation
had no significant impact on the extent of the reaction. Preozonation followed by biotreatment resulted
in the least amount of organic halide formation, with a reduction of 50 to 80% when chlorine was used
and greater than 90% with chloramines. In all cases, the ratio of organic halides to DOC decreased after
biological treatment, indicating a selectivity for the potential reactive sites.

Shukairy (1998) found biodegradation to be very effective for controlling organic halide concentra-
tions. The precursor concentrations were decreased, and the microorganisms seemed to be selective in
biodegrading the moieties that are prone to substitution by the chlorine. Biodegradation of nonprecursor
compounds may have also occurred, but not to complete mineralization, i.e., no change in DOC. While
ozonation yields its own DBFs, e.g., aldehydes and ketones, subsequent biotreatment should help in the
removal of these highly oxidized lower-molecular-weight compounds, many of which are biodegrad-
able. In some cases, ozonation followed by biotreatment decreased purgeable organic halide formation
potential (POXFP) and total organic halide formation potential  (TOXFP) more than the sum of the
decrease by  the individual treatments, indicating a synergetic effect. Chloramines, which have been
found to be  as effective as free chlorine in killing attached bacteria in the distribution system, were
shown to yield the lowest level of organic halides. Using ozonation followed by biotreatment would
decrease the available substrate for microbial regrowth and provide primary disinfection. Smaller amounts
of chlorine or chloramines could then be applied to provide adequate post-disinfection and a residual
for the distribution system, therefore controlling the DBF concentration in finished water. The authors
concluded that further investigations into the impact of biotreatment on individual DBFs is warranted.
Similarly, the impact of ozone doses on biotreatment and DBF  formation should be investigated to
optimize the synergy of these processes. In addition, for both sources of organic matter, the formation
rate of nonpurgeable organic halide (NPOX) is much faster than that for POX after either chlorination
or chloramination.

   •   The majority of organic halide formation occurs at disinfectant doses less than 2 mg disinfec-
       tant per mg. DOC chloramination significantly  decreased the organic halide formation,
       especially POX, compared to that formed by chlorination or ozonation/chlorination.
   •   Preozonation significantly decreased the amount of organic halides formed after chlorination,
       but had no impact on the organic halides formed after chloramination.
   •   Biological treatment alone and, more effectively, preozonation followed by biological treat-
       ment selectively reduced the organic halide precursor compounds compared to the overall
       background organic matter as measured by DOC.
EPA Research Activities

EPA research into the factors that affect chlorine decay and the formation of DBFs has ranged from pilot-
plant studies that have investigated the role of ozone on brominated DBFs to the effect of corroded pipe
on the formation of TTHMs and HHA6. Also included in these efforts is research on mutagenicity as a
possible DBF and an attempt to define exposure research as it relates to the DBF problem.

Shukairy et al. (1994b, 1995) studied the effect of variable ozone dosage and bromide concentration on
the formation of organic DBFs and bromate. Low-ozone dosages resulted in oxidation of organic pre-
cursors, yielding decreases in the formation potential for TTHMs,  six HAAs, and TOX. Increasing the
ozone dosage oxidized bromide to bromate, decreasing the bromide for incorporation into DBFs. Bro-
mate concentrations were linearly correlated with ozone residuals. Changes in the bromine incorpora-
tion factors reflected differences in the resulting speciation of THMs and HAAs, respectively. Because
                                             12-14

-------
TOX measurements based on chloride equivalence may underestimate the halogenated DBF yield for
high-bromide waters, they describe a procedure whereby bromide and bromate concentrations were
used to correct the TOX measurement.

The effect of bromide concentration, ozone dosage, and biotreatment on the control of DBFs was also
evaluated. Although TTHM precursors were better controlled by ozonation and the precursors of six
HAAs were better controlled by biological treatment, the combined processes were effective for the
control of all halogenated DBF precursors. Ozone's conversion of bromide to bromate and the chemi-
cal or biological oxidation of organic matter changed the ratio of bromide to DOC. Increases in this
ratio increased the formation of some brominated DBFs, but these DBF increases were offset by the
precursor oxidation provided by the combination of the two processes.

Clark et al. (1996b) developed a first order model to characterize the formation of brominated DBFs.
Using a data set generated by Pourmoghaddas et al. (1993), models were developed that describe the
formation of THM and non-THM chlorination by-products and their speciation. The model which
considered pH, time, chlorine, and bromide concentration demonstrates the effect of bromide concen-
tration on the formation of CHC13 and CHBr3. The model shows that the concentrations of CHBr2Cl
and CHBrCl2 increase to a maximum for bromide concentrations of 2.5 mg/L and 0.5 mg/L, respec-
tively, and then decline with increasing bromide levels. Concentrations of CHC13 consistently decline
with increasing bromide concentration, while CHBr3 consistently increases with increasing bromide
concentration.

As a complement to his work on chlorine decay modeling, Clark (1998) developed a TTHM formation
model based on chlorine consumption as follows:
                                                    ±Lj)                     (12-44)


where

                                T= dimensionless parameter
                           CA = the initial chlorine residual in mg/L
                 R = dimensionless parameter from the chlorine decay equation
                                u = the reaction rate in time"1
                               TTHM = total trihalomethanes

Clark and Sivaganesan (1998) verified this equation using both field and laboratory data from the study
by Vasconcelos et al. (1997) and from laboratory data collected by the WSWRD of EPA.

Schenck et al. (1998) conducted a study to assess the applicability of the model developed by Vartiainen
and Liimatainen (1988) to source waters and water treatment practices in the U.S. The model is based
on data collected in Finland and relates mutagenicity, as determined in the Ames assay, to TOC concen-
tration of the water,  chlorine dose, and to a minor extent, the concentration of ammonia. It has been
used as the basis for recent epidemiological studies conducted in Finland that have reported a positive
correlation between the mutagenicity of chlorinated drinking waters and certain human cancers. In the
work by Schenck et al. (1998), water samples were collected from three full-scale treatment plants and
one pilot-scale plant in the U.S. All the plants used chlorine exclusively for disinfection. One full-scale
plant used ground water; surface water sources were used by the other plants. TOC and ammonia
concentrations were determined analytically, and chlorine doses were obtained from the treatment plants.
The water samples were concentrated by resin adsorption for testing in the Ames assay. The observed
levels of mutagenicity in the finished waters were 1.5 to 2.0-fold higher than those predicted using the
                                             12-15

-------
Vartiainen and Liimatainen (1988) model. The authors concluded that further validation was needed
before the Finnish model could be used to assess exposure to mutagenicity  in chlorinated drinking
waters in the U.S.

Rossman et al. (1999) examined the factors that characterize reaction conditions affecting treated water
in a distribution system versus reaction kinetics as determined in bench studies. Included among these
factors are the pipe wall material, such as corrosion products and biofilm slime, which can exert a
significant chlorine demand.  If chlorine is consumed within a pipe by non-DBP-producing materials
(such as ferrous corrosion products), then there is less available to react with the NOM in the water.
Some of the factors that may contribute to differences in the rate of DBF formation in a pipe versus
bench scale experiments are

   •   Iron tubercles which contain organic material that might include DBF precursors.
   •   The organic matrix supporting the growth of biofilm on pipe surfaces may also contain
       precursors.
   •   Certain DBFs, such as dichloroacetic acid, are biodegradable.
   •   The rate at which reactants are transported between the bulk flow and the reaction region near
       the pipe wall is affected by hydrodynamic conditions within the pipes.

The authors conducted a study to measure the rate of formation of two classes of DBFs in a simulated
pipe environment and compared it with rates observed for the same water held in glass bottles. The
DBFs studied were THMs and HAAs, both of which are currently regulated by EPA under its Disinfec-
tants/Disinfection By-Products Rule. The simulated pipe environment is located at the EPA Test and
Evaluation (T&E) Facility in  Cincinnati, OH, and is designed to replicate actual flow conditions within
a ductile iron pipe. The pipes used in this test had been subject to significant corrosion and biofilm
buildup. The authors found that the production of THM and HAA in the pipe kept  pace with that
formed in the bottle given that the rate constants of chlorine consumption in the pipe (by mainly non-
precursor material) were more than ten times higher than in the bottle. This suggests that chlorine is not
a rate-limiting factor in the reactions that produce these compounds for the waters tested. The fact that
a small, but consistently higher level of THM was produced in the pipe compared to the bottle for the
same reaction time could be due to two reasons. First, the metallic surface of the pipe wall could serve
as a  catalyst for the THM formation reaction. Second, there could be THM  precursors in the scale,
tubercles, or biofilm attached to the pipe wall.

To test the second of these hypotheses, another set of experiments was performed with the pipe loop.
This time the test water was chlorine demand-free, DBF precursor-free water derived from Cincinnati
tap water. This water receives activated carbon treatment at the Cincinnati Water Works and was treated
again on-site with a GAC canister to remove any residual chlorine, DBFs, and possible DBF precur-
sors. Experiments were conducted with this water chlorinated to three different levels, without any
initial holding time in the feed tanks. The results confirm that there must be a reservoir of precursor
material attached to the pipe wall that is available to contribute to the formation of THMs. Results from
their experiments led to the following conclusions:
   •   The rate constants for chlorine decay in the pipe were an order of magnitude higher than in
       the bottle.
   •   The high rate of chlorine loss in the pipe did not decrease the rate at which DBFs were
       produced when compared to the bottle.
   •   THM4 production in the pipe averaged 15 percent higher than in the bottle over a 24-hour
       period.
   •   The distribution of THM species over time, as reflected by the Bromine Incorporation Factor,
       remained similar between the pipe and the bottle.
                                             12-16

-------
   •   The rate of HAA6 production in the pipe was essentially the same as in the bottle.
   •   The test pipe contains a reservoir of precursor material on its walls that is available to form
       THMs.

As an extension of the research conducted by Clark et al. (1996b), Pourmoghaddas et al. (1993), Clark
(1998), and Clark and Sivaganesan (1998) discussed previously,  Clark et al. (2001) developed a gen-
eral DBF formation model which is given as follows:
                DBP.=A:
                                 \-KeM(l-K)t
02-45)
where DBF. = the specific disinfectant by-product subspecies being modeled in micromoles/liter; M, K,
t, and CA are as defined previously; and A. is the ratio in micromoles/liter of by-product / formed to
mg/L of chlorine consumed.

Several functional relationships were examined in an attempt to find a general model to predict K, M,
and A in Equation 12-45. The variables/?//, Br, CIQ, andP
                mBf
        p = _
(where       (mCL + mBr~\ mClQ = moles of initial chlorine, and mBr = moles of bromide ion) were
utilized in various combinations to develop predictive equations for these parameters. A general model
for A was developed for chlorinated, mixed species and brominated compounds.

A multiple regression analysis forMyielded the following model shown below:

                        M = eaieblBr~eCl(cl°*pH)ClQl eeelP/lpH                       (12-46)

where av bv cv dr ev and/: are parameters to be estimated. In Equation 12-46, ea^ is a constant, e^1^
accounts for the  impact of bromide  concentration, ec^cl°*pir> accounts for the interaction of chlorine
(mg/L) and pH, C0!  accounts for the impact of chlorine concentration alone in mg/L, e|ip accounts for
the ratio of bromide in moles to bromide plus chlorine in moles, and efiPH  accounts for the impact of
pH alone.

A multiple regression model for log(K) with log(p//), \og(Br^\ and log(C/0) as the predictor variables
yielded the equivalent multiplicative model as shown below:
                                 K = eai(pH)b>(Br + 1 )C
-------
compounds considered in this analysis. In Equation 12-48, d is set equal to 1 for chlorinated com-
pounds and is otherwise 0.0001.

Using data from Pourmoghaddas et al.  (1993), least squares estimates for the parameters were calcu-
lated. The estimated parameters for K and Mare given in Equations 12-49 and 12-50, respectively, as
shown below:
                                              0 * pH) . £y -2.32e8.46(P)e-0.231pH         (12-49)

                           K= el -89 • (pH)-° 13 • (Br-+ If10 • (C/0)^75               (12-50)

The estimated regression model R2 were 0.70 and 0.95, respectively. Equations were developed for
chloroform, dichlorobromomethane, dibromochloromethane,  bromoform, monochloroacetic acid
(MCAA), dichloroacetic acid (DCAA), trichloroacetic acid (TCAA), monobromoacetic acid (MBAA),
dibromoacetic acid  (DBAA), tribromoacetic acid (TBAA), bromochloroacetic acid (BCAA),
dibromochloroacetic acid (DBCAA), and dichlorobromoacetic acid (DCBAA).

Exposure to DBFs from Distribution Systems

Traditionally, EPA has established environmental regulatory programs for the protection of the outdoor
environment from industrial and commercial sources of contamination. It has become increasingly
clear, however, that indoor environments may pose a risk from contaminants that are found outside the
household. For example, drinking water can be a carrier of contaminants that volatilize when they enter
a household, thereby subjecting the consumer to inhalation, dermal, and ingestion exposure (Clark and
Goodrich 1992). An EPA study found that personal or indoor exposures to many toxic or carcinogenic
chemicals are greater than to outdoor concentrations. It also found that for all chemicals, except for
THMs, the air provided greater than 99% of the exposure. Water provided nearly all of the exposure to
the THMs and more than half of most exposure to chloroform.

Clark et al. (1992) demonstrated that the dynamics of multiple source mixing and system operation in a
water distribution system can affect lifetime exposures to volatile organic chemicals such as chloroform.
The dose received will depend on one's location within the system and the timing  of water use. It was
found that dynamic water quality models are valuable tools for conducting drinking water exposure as-
sessments. They help in interpreting the results from field monitoring sites so that a clearer picture of the
impact of distributed water emerges. They also provide time-varying estimates of drinking water quality
at all locations within a service area so that more accurate exposure assessments can be made. A case
study showed that for adults, the inhalation pathway for chloroform contributes more to total exposure
than does the ingestion pathway at most locations within the service area, although the opposite is true for
infants. Showering is the greatest potential contributor to household inhalation exposure. This analysis
raised the question of equity  in providing safe drinking water (Clark et al. 1992).

Modeling Chlorine Decay and TTHMs in Distribution Systems
EPA has conducted extensive research related to the application of models to chlorine decay and TTHM
formation in drinking water distribution  systems. This research has been a major contributor to an
understanding of the factors that influence water quality in drinking water distribution systems and is
discussed in this section.
                                            12-18

-------
North Penn Study

One of the first projects to investigate the feasibility of modeling water quality in drinking water distri-
bution systems was conducted under a cooperative agreement initiated between the North Penn Water
Authority (NPWA) in Lansdale, PA, and EPA. The project focused on the mixing of water from mul-
tiple sources and investigated the feasibility for development and application of a steady-state water-
quality model. As the study progressed, it became obvious that the dynamic nature of both patterns of
demand and variations in water quality required the development of a dynamic model. Techniques for
semicontinuous monitoring of volatile organic contaminants were also explored (Clark et al. 1988b).
At the time of the NPWA/EPA study, the authority served 14,500 customers in 10 municipalities and
supplied an average of 19,000 m3/day (5mgd) (Clark et al. 1988b). The NPWA distribution system was
modeled in a network representation consisting of 528 links and 456 nodes, and water demands for
modeling represented conditions from May to July 1984. The network hydraulic model used was de-
veloped by the U.S. Army Corps of Engineers and contained provisions for both steady-state and quasi-
dynamic hydraulic modeling for extended-period simulation (Gessler and Walski 1985).

The model revealed that significant portions of the system were subject to flow reversals. The veloc-
ity  for each link was known from the hydraulic solution for each time period, which was evenly
divided into an integer number of computational time steps. Each link was then divided into sublinks
by a series of evenly spaced subnodes (though the distance between subnodes varied from link to link
or for a link at different time periods), so that the travel time from a subnode (or node) to the adjacent
subnode (or node) was approximately  equal to a specified time step. A model called the Dynamic
Water Quality Model (DWQM) was developed. The solution algorithm used in the DWQM operated
sequentially by time period. During a time period, all external forces affecting water  quality were
assumed to remain  constant (e.g., demand, well pumpage, tank head). The  DWQM  was used to
simulate a 34-hour period corresponding to conditions present during the pilot-level sampling pro-
gram conducted on November 14 to 15, 1985. Parameters of the model were adjusted so that pre-
dicted tank levels and flows at selected sites represented those measured during the sampling period.
For chloroform, THM, and hardness, the predicted concentrations compared favorably with the ob-
served values at the sampling stations.

South Central Connecticut Study

The North Penn case study provided an excellent test-bed for development of a dynamic water-quality
model (Clark and Coyle 1990). To extend the North Penn application, EPA initiated another coopera-
tive agreement with the University of Michigan and the South Central Connecticut Regional  Water
Authority (SCCRWA). The purpose of the cooperative agreement was to test the previously developed
modeling concepts and to verify and calibrate the model through field investigations (Clark and Goodrich
1993; Clark et al. 1988a). At the time of the study, SCCRWA supplied water to  approximately 95,000
customers (380,000 individuals) in 12 municipalities in the Greater New Haven area. The service area
was divided into 16 separate pressure/distribution zones. Average production was  190,000 m3/day (50
mgd) with a safe yield of approximately 281,200 m3/day (74 mgd). Water sources included four surface
water sources (Lake Gaillard, Lake Saltonstall, Lake Whitney, and the West River System) and five
well fields (North Cheshire, South Cheshire, Mt. Carmel, North Sleeping  Giant, and South Sleeping
Giant). Approximately 80 percent of the water in use in the system came from surface sources; the
remaining 20 percent came from wells.  All surface water was treated with chlorination, filtration, and
addition of a phosphate-corrosion inhibitor. The system included 22 pumping stations, 23 storage tanks,
and approximately 2,091,700 m (1300 miles) of water mains. Preliminary efforts to develop and vali-
date a model for the South Central System were concentrated on the Cheshire  service area (Clark
1993b; Clark etal. 1993b).
                                            12-19

-------
This area was relatively isolated and provided a prototype for modeling the remainder of the system. To
validate the model, an extensive study was planned in which the fluoride feed at the North Well Field
was turned off and the propagation of the fluoride feed water was tracked through the system (Clark
et al. 1991). Prior to the water-quality modeling effort and the related field study, extensive hydraulic
analyses were conducted on the system. For the preliminary modeling effort, the full SCCRWA system
network was represented by approximately 520 nodes and 700 links. In most cases, the network was
represented by "skeletonizing" the system (i.e., selectively choosing pipes based on their size and
perceived impact as transmission mains). The DWQM developed in the North Penn study was applied
to the Cheshire system to simulate the propagation of fluoride feed water and also to select sampling
locations for  a field study. Based on the simulation results and the objectives of the proposed field
study, a sampling scheme was designed. Fluoride was selected  as the tracer because it was added
regularly to the water at a concentration of approximately 1 mg/L,  as required by the State Department
of Health Services. Tracing the changes in the fluoride concentration in the distribution system allowed
accurate travel times to be determined. The time the fluoride feed was shut down at the well fields was
compared to the time it dissipated at the sampling points (Skov et al. 1991). The model predictions and
the sampling results were extremely close.

The behavior of the storage tanks was of special interest.  During the early portion of the sampling
period, variations in tank levels were held to a minimum (less than 0.91 m [3 ft]). After 2 days, little
change in fluoride concentrations was found in the tank and,  as a result, the water level was then
allowed to vary approximately 2.8 m (8 ft). The wider range in tank levels had the effect of turning the
water over relatively rapidly. Even with the rapid turnover, it took nearly 10 days to completely replace
old water with new water in the tanks. It was clear from this analysis that storage tanks could have a
detrimental effect on water quality, particularly as water aged in the tank.

On August 13 to 15,  1991, another sampling program at the Cherry Hill/Brushy Plains Service Area
was initiated with the goal of validating the previously discussed simulation results. The purpose  of
this sampling program was to gather information to characterize the variation of water quality in the
service area and to  study the impact of tank operation on water quality. The Cherry Hill/Brushy Plains
Service Area covered approximately  5.18 x 106 m2 (2 mi2) in the Town of Branford in the eastern
portion of the SCCRWA (Clark et al. 1993b).  This service area was almost entirely residential, contain-
ing both single-family homes and apartment/condominium units.  Average water use during the sam-
pling period was 1700 m3 per day (0.46 mgd).

The water distribution system was composed of 20.3-cm (8-in) and 30.48-cm (12-in) mains. The ter-
rain in the Cherry Hill/Brushy Plains Service Area was generally moderately sloping, with elevations
varying from approximately 15.2 m (50 ft) mean sea level (MSL) to 70.1 m (230 ft) MSL.  Cherry Hill/
Brushy Plains received its water from the Saltonstall system. Water was pumped from the Saltonstall
system into Brushy Plains by the Cherry Hill Pump Station. Within the service area, storage was pro-
vided by the Brushy Plains tank. The pump station contained two 10.2-cm (4-in) centrifugal pumps
with a total capacity of 5300 m3/day (1.4 mgd). The operation of the pumps was controlled by water
elevation in the tank. Built in 1957, the tank had a capacity of 3800 m3 per day (1.0 mgd). It had a
diameter of 15.2 m (50 ft), a bottom elevation of 58.8 m (193 ft) MSL, and a height (to the overflow) of
80.2 m (263 ft) MSL. During normal operation, the pumps were set to go on when the water level in the
tank dropped to 15.2 m (56 ft) and to turn off when the water level reached 19.8 m (65 ft). As had
occurred during the study of the Cheshire service area, the hydraulic model developed by Gessler and
Wolski (1985) and the DWQM were applied to establish flow patterns within the service area.  In
addition, during the periods of May 21 to 22, July 1 to 3, July 8 to 10, and July 30 to August 1, 1991,
chlorine residuals were monitored at the tank and operational patterns (pump records), and variations
in tank water  level were studied.
                                             12-20

-------
On the basis of these model runs and field data, a sampling strategy was adopted that involved turning
the fluoride off at the Saltonstall Treatment Facility and sampling for both fluoride and chlorine in the
Cherry Hill/Brushy Plains Service Area. The intention was to use defluoridated water as a conservative
tracer for the movement of flow through the system and for calibration the DWQM. The DWQM and a
chlorine decay model based on hydrodynamic principles were used to model the dynamics of chlorine
decay in the system. Seven sampling sites in the distribution system, in addition to sampling sites at the
pump station and tank, were identified. A hydraulic model and the DWQM model were used to simu-
late the Cherry Hill/Brushy Plains Service Area for a 53-hour period from 9:00 am on August 13 to
3:00 pm on August 15, 1991. A skeletonization was developed representing the Cherry Hill/Brushy
Plains distribution system, which included all 30.4-cm (12-in) mains, major 20.3-cm (8-in) mains and
loops, and pipes that connected to the sampling sites. Pipe lengths were scaled from a map, actual pipe
diameters were used, and, in the absence of any other information, a Hazen-Williams roughness coef-
ficient of 100 was assumed for all pipes. From these results, it is clear that the modeling effort matched
the sampling efforts well, with the exception of dead-ends.

It was clear that the pump cycles influenced water quality heavily at several sampling points. For
example, at node 11, during the pumps-on cycle, the fluoridated water was pumped into the system.
When the  system was being fed from the tank (pumps-off), the system was receiving water that had
reached an equilibrium concentration of fluoride before the stoppage of the fluoride feeders. As men-
tioned previously, the two scenarios evaluated during the sampling study were with the pumps-on and
pumps-off condition. Using the upstream and downstream chlorine concentration and the residence
times in the link, the chlorine decay coefficient was calculated for each link. Chlorine demand was
calculated according to a first-order assumption, which is defined as follows:
                                         C = C0e-rt                            (12-51)

where C = the concentration at time t, C0 = initial chlorine concentration, r = decay rate in day"1, and t
= time in day. A bench study was conducted in which chlorine demand for the raw water was calculated
using Cherry Hill/Brushy Plains water, and the chlorine decay rate was calculated as 0.55 day"1. This
decay rate might be regarded  as the bulk decay rate or the decay rate of chlorine in the treated water.
The total system demand was found to be two to three times higher than the bulk decay rate alone. It
was concluded that this additional demand was caused by pipe wall demand, biofilm,  and tubercles.

Development of EPANET

As a follow-up to the development of the DWQM and as a follow-up to the North Penn and SCCRWA
studies, Rossman (1994) and Rossman et al. (1994) developed a mass transfer-based model for predict-
ing chlorine decay in drinking water distribution networks. The model considers first-order reactions of
chlorine to occur both in the bulk flow and at the pipe wall. The overall rate of the wall reaction is a
function of the rate of mass transfer of chlorine to the wall and is therefore dependent on pipe geometry
and flow regime. As observed in the SCCRWA study, the model can thus explain field observations
that show higher chlorine decay rates associated with smaller pipe sizes and higher flow velocities. It
has been incorporated into a computer program called EPANET that can perform dynamic water-qual-
ity simulations on complex pipe networks. It represents  a third generation of water-quality models
developed by the WSWRD to improve our understanding of the movement and fate of constituents
within water distribution systems.
Advances  contained in EPANET include a coordinated approach to modeling both network hydraulics
and water quality, consideration of both bulk flow and pipe wall reaction mechanisms, and a graphical
                                            12-21

-------
user interface to aid in visualizing network behavior. The model's bulk decay rate constant is deter-
mined independently in the laboratory. Its wall decay constant can be varied over a range of values that
include both reaction rate-limiting and mass transfer rate-limiting values. EPANET is based on the
extended-period simulation approach to solving hydraulic behavior of a network. It has proven to be a
very effective research tool for modeling the movement and fate of drinking water constituents within
distribution systems. EPANET calculates all flows in cubic feet per second (cfs) and has an option for
accepting flow units in gallons per minute (gpm), mgd, or liters per second (1/s). The Hazen-Williams
formula,  the Darcy-Weisbach formula, and the  Chezy Manning formula can be used to calculate the
head loss in pipes. It also models pumps, valves, and minor loss. To model water quality within distri-
bution systems, the concentration of a particular substance must be calculated as it moves through the
system from various points of entry (e.g., treatment plants) and on to water users. This movement is
based on three principles: (1) conservation of mass within differential lengths of pipe, (2) complete and
instantaneous mixing of the water entering pipe junctions, and (3) appropriate kinetic expressions for
the growth or decay of the substance as it flows through pipes and storage facilities.

The model has been validated using data from various field studies. Resulting predictions have been
compared with observed chlorine measurements at eight field sites. Good agreement was achieved at
locations where the hydraulic conditions were well characterized. Model predictions were less accurate
at sites where the hydraulic calibration was  less precise. These results underscore the need to obtain
accurate  hydraulic information before running a network water-quality model.

North Marin Study

It is clear that, as dissolved chlorine travels through the pipes in the network, it reacts with NOM in the
bulk water and with biofilm and tubercles on the pipe walls or with the pipe wall material itself (Clark
et al.  1993a). This reaction results in a decrease in chlorine residual and a corresponding increase in
DBFs, depending on the residence time in the  network pipes and holding time in storage facilities.
Understanding these reactions will help water-utility managers deliver high-quality drinking water and
meet regulatory requirements under the 1996 SDWA and amendments. Water-quality modeling has the
potential to provide insight into the factors that influence the variables affecting changes in water
quality in distribution systems. Understanding the factors that influence the formation of TTHMs and
maintenance of chlorine residuals is of particular interest (Clark et al. 1995).

EPANET has proved to be especially useful for modeling both formation of TTHMs and the propaga-
tion and maintenance of chlorine residuals. Among the first studies to address these issues using EPANET
was one conducted by EPA in collaboration with the North Marin Water District (NMWD) in Califor-
nia (Clark et al.  1994). Another recently completed study conducted jointly by EPA and the AWWARF
examined these same issues (Vasconcelos et al. 1997). This study evaluated various types of models to
describe  both the formation of TTHMs and loss of chlorine residual and will be discussed later in this
chapter.

At the time of the North Marin study, the district served a suburban population of 53,000 people who
live in or near Novato, California. It used two sources  of water: Stafford Lake and the North Marin
Aqueduct. The  aqueduct is a year-round  source, but Stafford Lake is in use only during the warm
summer months, when precipitation is virtually nonexistent and demand is high. Novato, the largest
population center in the  North Marin service area,  is located in a warm, inland coastal valley with a
mean annual rainfall of 68.6 cm (27 in). Virtually no precipitation occurs during the growing season
from May through September. Eighty-five percent of total water use is residential, and the service area
contains  13,200 single-family detached homes, which accounted for 65 percent of all water use (Clark
et al. 1994). The water quality of the two sources differs greatly. Stafford Lake water had a high humic
content and was treated with conventional treatment and prechlorination doses of between 5.5 and


                                             12-22

-------
6.0 mg/L. The treated water had a residual of 0.5 mg/L when it left the treatment plant clearwell. The
potential for formation of THMs in the Stafford Lake water was high. The North Marin Aqueduct
water was derived from a Raney Well Field along the Russian River. Technically classified as ground
water, the source water contained a high proportion of naturally filtered water. Aqueduct water was
disinfected only and was low in precursor material, with a correspondingly low potential for formation
of THMs. Both sources carried a residual chlorine level of approximately 0.5 mg/L when the water
entered the system.

The major focus of the study was Zone 1 of the NMWD distribution system. Depending on the time of
year and the time of day, water entered the system from either one source or both sources. The North
Marin Aqueduct source operated year-round, 24 hours per day. The Stafford Lake source operated only
during the peak demand period from 6:00 am to 10:00 pm and generally operated for 16 hours per day.
EPANET was used to model the system hydraulics, including the relative flow from each source, TTHMs,
and propagation of chlorine residual (Rossman et al. 1994). The model was based on an earlier repre-
sentation of the network made by Montgomery Watson, Inc., for North Marin and was calibrated based
on a comparison of simulated versus actual tank levels for the May 27-29, 1992, period of operation.
The  dynamic nature of the  system led to variable flow conditions and variable water quality in the
network. Flow directions frequently reverse within a given portion of the network during a typical
operating day. The consequences of these variable flow patterns for water quality are significant.

To characterize the water quality in the NMWD,  EPA designed a sampling protocol and sent a team of
investigators to work with the district staff for the period May 27-29,  1992. The water quality is highly
variable. For example, at the Eighth Street sampling point, chloroform levels varied from 38.4 to 120.1
|ig/L over the 2-day period. This variability was caused by the penetration of water from the two different
sources. A regression relationship between UV absorbance  and THMs was established using the data
from both sources. The assumptions of the model (constant variance and normality of error terms) were
checked and deemed to be reasonable. Both Stafford Lake water and the North Marin Aqueduct water
generally maintained a chlorine residual level of 0.5 mg/L as the treated water entered the system. As
mentioned, the Stafford Lake water had a much higher chlorine demand than did the Aqueduct water.

To predict chlorine demand at the various sampling points, a first-order decay relationship was as-
sumed (Clark et al. 1995). In EPANET, chlorine decay is represented by decay in the bulk phase and by
decay in the pipe wall. Based on bulk water calculations,  the first-order decay coefficients or bulk
demand for the Stafford Lake and the Aqueduct sources were 0.31 and 0.03 day"1, respectively. Using
EPANET and the previously assumed hydraulic conditions, the chlorine residuals were estimated. It
was  evident from the analysis  that the pipes in the distribution network exhibited a demand for chlo-
rine. This demand probably comes from tubercles, biofilm, and perhaps the pipe wall material itself
(Clark et al. 1995). A  comparison between chlorine residuals using the first-order assumptions pre-
dicted from EPANET versus actual chlorine residuals provided an excellent illustration of this point. It
became clear that the demand in the system went beyond just bulk-water decay. Because EPANET has
the capacity to incorporate a wall demand factor in addition to the bulk demand factors for chlorine, the
system was simulated again using the bulk demand for the two sources, and trial and error was used to
estimate wall demands for four sections of the network. When chlorine residuals were re-estimated at
the four sampling points, wall demand obviously played a major role in chlorine residual loss. This
pipe wall demand may be the result of the source or the age  of the system. For example, the maximum
wall demand was found in the areas served primarily by Stafford Lake. However, those pipes were also
the oldest in the system.

EPA/AAWARF Study

Vasconcelos et al. (1997), in a study sponsored jointly by EPA and AWWARF, investigated the factors
leading to loss of chlorine residuals in drinking water distribution systems. Kinetic rate equations were

                                             12-23

-------
developed, tested and evaluated based on data collected from five drinking water utilities. This inves-
tigation uncovered a number of findings concerning the mechanisms of chlorine decay and the kinetic
models that describe it. The most significant findings of this study included the following:
    •   Chlorine decay in distribution systems can occur because of reactions within the bulk fluid
       and with the pipe wall demand.
    •   The rate of reaction of chlorine at the pipe wall is inversely related to pipe diameter and is
       mass transfer limited.
    •   There is no established method for directly determining the kinetics of chlorine decay attrib-
       utable to pipe wall reactions. These values must be determined from field data.
    •   A well-calibrated hydraulic model is a prerequisite for attempting to model water quality in a
       distribution system.
    •   Calibration of network chlorine decay models can be based on first-order kinetic models for
       bulk reactions and either first-order or zero-order kinetics for wall reactions.
    •   The wall kinetic constant appears to be inversely related to the pipe roughness coefficient.
    •   A non-reacting chemical should be used when calibrating a model during field studies.

Based on the experience gained during this study, the following recommendations for future work are
offered:

    •   More direct methods of estimating pipe wall-related chlorine reaction constants are needed.
    •   Sensors in the distribution system coupled with remote telemetry may offer a way to perform
       continuous on-line calibration of network chlorine decay models.
    •   Water quality models should be enhanced to better accommodate different water sources
       when each source water exhibits different bulk reaction kinetics.
    •   As an increasing number of systems calibrate chlorine decay models, it may be possible to
       establish a database relating kinetic parameters to water chemistry and pipe characteristics.


Evolution of System Modeling

On the basis of the results from the case studies previously described and from other studies reported in
the literature, it is obvious that water-quality modeling has the potential to provide insight into the
factors that degrade water quality in networks. It has also become increasingly obvious that, despite the
treatment investments being forced by regulations in the 1996 SDWA and amendments, the potential
exists for deterioration of water quality in the network itself. This realization led to the development of
several public- and private-sector water-quality models.  Only the public-sector models will be dis-
cussed here. EPANET, developed by Rossman (1994) and Rossman et al. (1994), as discussed earlier,
was based on mass transfer concepts.

Another approach to the propagation of contaminants was developed by Biswas et al. (1993) using a
steady-state transport equation. It accounted for the simultaneous advective transport of chlorine in the
axial direction, diffusion in the radial direction, and consumption by first-order reaction in the bulk-liquid
phase. Islam (1995) and Islam et al. (1997) developed a  model called QUALNET, which predicted the
temporal and spatial distribution of chlorine in a pipe network under slowly varying unsteady-flow condi-
tions. Boulos et al. (1995) proposed a technique called the Event-Driven Method (EDM), which is based
on a next-event scheduling approach and can significantly reduce computing times.

Several different types  of numerical methods have been proposed to solve these types of models, in-
cluding the Eulerian Finite-Difference Method (FDM), the Eulerian Discrete-Volume Method (DVM),
the Lagrangian Time-Driven Method (TDM), and the Lagrangian Event-Driven Method (Rossman and
Boulos 1996). The FDM approximates derivatives with finite-difference equivalents along a fixed grid
                                             12-24

-------
of points in time and space. Islam et al. (1997) used this technique. The DVM divides each pipe into a
series of equal-sized, completely mixed volume segments. At the end of each successive water-quality
time step, the concentration within each volume segment is first reacted and then transferred to the
adjacent downstream segment. This approach was used in the models that were the basis for the early
DWQM studies. The TDM tracks the concentration and size of a non-overlapping segment of water
that fills each link of a network. As time progresses, the size of the most upstream segment in a link
increases as water enters the link, whereas an equal loss in size of the most downstream segment occurs
as water leaves the link. The size of these segments remains unchanged. The EDM is similar in nature
to the TDM, but rather than update an entire network at fixed time steps, individual link/node condi-
tions are updated only when the leading segment in a link disappears completely through this down-
stream node. The development of the EPANET hydraulic model has satisfied the need for a compre-
hensive public-sector model and has been a  key component in providing the basis for water-quality
modeling in many utilities throughout the U.S.

Advances in Numerical Modeling Techniques

In addition to the development and application of EPANET, other "spin-off' research has resulted in
models that locate monitoring stations in networks, predict the propagation of disinfectants, and deter-
mine the location of booster chlorination. This research is summarized in the following text.

Lee et al. (1991) examined the problem of selecting monitoring stations that will adequately monitor
the changes in water quality between the time water leaves the treatment plant and the time it reaches
the customer's tap. They found that there is no uniform schedule or framework for monitoring under
the SDWA. This lack of specificity poses both management and technical barriers to states and water
systems ultimately responsible for implementation of the regulations. The authors provided systematic
and quantitative guidelines for locating monitoring stations. Their guidelines are based on the concept
of pathways. The authors applied the concept of coverage and inferred the quality at an upstream node
from the quality at a downstream node.

Lu et al. (1993, 1995) developed mathematical models to predict disinfectant concentration profiles
under breakpoint chlorination conditions and to predict the growth of biofilm in drinking water distri-
bution systems. The breakpoint chlorination model accounts for concurrent mass transfer and a series
of chemical reactions under breakpoint chlorination conditions, and the other is developed to predict
disinfectant concentration profiles in the drinking water distribution pipe. The disinfection model is
validated by comparing its numerical solutions to experimental data in the literature.  The impact  of
important parameters on the model performance is examined by sensitivity analysis. Practical applica-
tions of the model to minimize water-quality deterioration in the distribution system are discussed. It is
intended to provide insight into the factors that influence all of the fundamental reactions and disinfec-
tant transport of the breakpoint reaction. Operational criteria for the chlorination of distributed water
are derived.

Rossman and Boulos (1996) compared the formulation and computational performance of four nu-
merical methods for modeling the transient behavior of water quality in drinking water distribution
systems. Two are Eulerian-based (the finite-difference and discrete-volume methods) and two are
Lagrangian-based (the time-driven and event-driven methods). In the Eulerian approaches, water moves
between fixed grid points or volume segments in pipes as time is advanced in uniform increments. The
Lagrangian methods update conditions in variable-sized segments of water at either uniform time in-
crements or only at times when a new segment reaches a downstream pipe junction. Each method was
encoded into an existing distribution system simulation model and run on several pipe networks  of
varying size under equal accuracy tolerances. Results show that the accuracies of the methods are
comparable. The Lagrangian methods are more efficient for simulating chemical transport. For model-
                                             12-25

-------
ing water age, the time-driven Lagrangian method is the most efficient, while the Eulerian methods are
more memory-efficient. Results of the study showed the following: the numerical accuracy of the methods
was by and large the same, with the exception that FDM occasionally smeared sharp concentration
fronts and DVM occasionally accelerated the arrival of concentration changes. All of the methods were
capable of adequately representing observed water-quality behavior in actual water distribution sys-
tems. Regardless of the method used, network size was not always a good predictor of the solution time
and amount of memory required.

Boccelli et al. (1998) studied disinfectants added at discrete locations in a water distribution system.
Such a strategy can reduce the mass of disinfectant required  to maintain a detectable residual at points
of consumption in the distribution system, which may lead to reduced formation of DBFs, in particular
THMs. An optimization model is formulated for the dynamic schedule of disinfectant injections, which
minimizes the total dose required to satisfy residual constraints over an infinite-time horizon. This
infinite-time problem is reduced to a solvable finite-time optimal scheduling model by assuming peri-
odicity of mass injections and network hydraulics. Furthermore, this principle of linear superposition is
shown to apply to disinfectant concentrations resulting from  multiple disinfectant injections over time.
A matrix generator code was developed to interface with the EPANET network water-quality model.
This code automatically generates the linear programing formulation of the optimal scheduling model,
which is then solved using the simplex algorithm. Results from application of the model suggest that
booster disinfection can reduce the amount of disinfectant required to satisfy concentration constraints
when compared to conventional disinfection practiced only at the source. The optimal booster schedule
reduced the average disinfectant concentration within the distribution system and, in some cases, the
variability of these concentrations. The number of booster stations, booster location,  and distribution
system hydraulics were shown to affect the optimal schedule.

Lu et al. (1995) developed a biofilm model that accounts  for simultaneous  transport of substrates,
disinfectants, and microorganisms and that predicts substantial changes in quality of distributed water.
The model  consists of a set of mass  balance  equations for organic substances, ammonia  nitrogen,
oxidized nitrogen, dissolved oxygen, alkalinity, biomass, and disinfectants in the bulk liquid phase and
within the biofilm under laminar and turbulent flow conditions. This model is validated by comparing
its solutions with numerical solutions  in the literature and is then applied to predict the behavior of a
typical water treatment plant effluent through a distribution pipe. The flow properties  and disinfectant
consumption rate at the pipe wall play a significant role in the determination of potable water quality in
the distribution system.

Tryby et al. (1999) examined the feasibility of using booster  chlorination in distribution systems. They
developed a conceptual model for bulk chlorine decay under  booster conditions. The conceptual analy-
sis demonstrated that the use of booster chlorination allows microbial inactivation and maintenance of
a detectable residual to be viewed as separate treatment issues. Booster chlorination could allow dos-
ages to be reduced at the water treatment plant without compromising treatment objectives. By reduc-
ing dosages, the DBF yields at the water treatment plant may also be reduced and the disinfectant mass
conserved can be more efficiently applied at points located in the distribution system. Their studies
suggest that increasing residual concentrations alone will not eliminate the risks associated with patho-
gens and biofilm regrowth in problem areas. Several barriers against the entrance of contaminants are
required. However, booster disinfection is the most efficient method of minimizing DBF formation.

Water Quality and Tanks

The issue of water quality as affected by storage tank design and operation has been the  object of
extensive study. Most of this research has been conducted by EPA and AAWARF. The major studies in
this area are presented in the following paragraphs.

                                             12-26

-------
Grayman and Clark (1993) conducted a series of studies demonstrating that water quality is degraded
as a result of long residence times in storage tanks, which in turn has highlighted the importance of tank
design, location, and operation. Computer models were developed to explain the effect of tank design
and operation on various water-quality parameters. The diversity of the effects and the wide range of
design and environmental conditions make general design specifications for tanks unlikely.  The au-
thors recommend that modeling be refined to facilitate site-specific analysis.

Mau et al. (1995) developed explicit analytical mathematical models for use in water-quality simula-
tion studies and management of distribution system storage. The proposed models can be used for
investigating the mixing characteristics of tanks and their subsequent effects on water quality. They can
directly supplement any of the existing distribution system water-quality simulators. These models are
formulated analytically from mass balance principles and based on hydrodynamic processes. The re-
sulting models are simple to understand and implement and are well suited to the needs of practicing
engineers. The performance models are validated by application to actual tank data.

Rossman et al.  (1995)  studied the factors leading to the  loss of disinfectant residual in well-mixed
drinking water storage tanks. Equations relating disinfectant residual to the disinfectant's reaction rate,
the tank volume, and the fill and drain rates were developed. The authors presented an analytical solu-
tion for the minimum disinfectant residual in the tank under constant inflow/outflow conditions. It
showed that significant disinfectant loss begins when the product of disinfectant decay constant and the
refill time for an empty  tank exceeds 0.1 and that disinfectant residuals are relatively insensitive to the
fraction of total volume devoted to emergency storage. A second, numerical solution to the model is
developed to account for the fact that tank fill and drain rates are constrained by  system demand pat-
terns, pump capacity, and pump scheduling. Results from their study showed that pulsed  or periodic
pumping during a portion of the day can maintain much higher disinfectant residuals than  continuous
pumping can.

Clark et al. (1996a) demonstrated the use of compartment models to characterize mixing in three tanks.
It was found  that the mixing regimes in these  tanks were well characterized by compartment-type
models and that these tanks were not completely mixed,  contrary to conventional  wisdom. For pur-
poses of this analysis, one-, two-, or three-compartment models were developed. It is clear that, in some
cases, compartment models provide a very good representation of the mixing and residence times in
tanks. It is also clear that additional research needs to be conducted in this area.

Boulos et al. (1996) conducted an extensive study of reservoir water quality at the Ed Heck reservoir in
Azusa, CA. Emphasis was placed on understanding the hydraulic mixing regime and the distribution of
the free chlorine residuals in the reservoir. A unique roof-mounted sampling device was developed for
the study that allowed samples to be extracted from various depths in the water  column.  The device
proved to be very effective and greatly enhanced the study's findings. The following conclusions were
drawn:

   •   Fluoride tracer studies are useful for determining residence time distributions and internal
       mixing dynamics
   •   In the Ed Heck reservoir, the flow pattern was a relatively fast rotational flow around a large
       outer annular ring coupled with a slower downward flow within a smaller central core.
   •   For the conditions governing this study, the reservoir behaves, on average, as a completely
       mixed reactor.
   •   The mean residence time of 9.7 hours and a negligible loss of free chlorine residual was
       experienced during the course of the  study.
   •   Short-circuiting between the inlet and outlet lines caused the T10 for the reservoir to be less
       than half that of a true continuous-flow stirred-tank reactor (CSTR).
                                             12-27

-------
Boulos et al. (1998) developed and verified an explicit mathematical model of distribution storage
water quality based on a compartmental representation of the reservoir continuum. It was formulated
analytically based on mass balance relationships which are applicable to separate inlet and outlet con-
figuration reservoirs with simultaneous dual-directional flow. Previous models considered a common
inlet/outlet reservoir configuration with unidirectional flow. The model was verified by application to
actual reservoir data taken from a storage reservoir in Azusa, CA. Both  conservative (fluoride) and
reactive (chlorine) species were considered. A four-compartment model  resulted in a reasonable fit
between observed and simulated concentrations and should be predictive under operating conditions
similar to those under which the model was calibrated.

Rossman and Grayman (1999) reported on experiments conducted on cylindrical scale-model tanks
designed to determine the effect of various factors on mixing in the tanks. It was found that the time
required to mix the contents of a tank with water introduced during the fill period was proportional to
the initial volume to the two-thirds power divided by the square root of the inflow momentum flux (the
product of flow rate and velocity). The time is insensitive to the orientation of the inlet (vertically or
horizontally). Complete mixing depends on the ratio of the momentum to buoyancy fluxes of the inlet
jet. This is similar to past findings for jet discharges to unconfmed bodies of water. The confined
geometry of the tank results in a narrower range of conditions that produce stratification. The  investiga-
tors derive a formula to estimate the minimum volume exchange required for a fill-and-draw cycle to
ensure complete mixing before the end of the filling period.

Policy Issues
There are many policy implications to the research discussed in this chapter.  Two papers (Clark et al.
1994, 1995) attempted to frame some of the issues that modeling results have raised with  respect to
water quality in drinking water distribution systems.

Clark et al. (1995) discuss the  SDWA and its amendments, which has focused interest on the factors
that cause the deterioration of water between the treatment plant and the consumer. The authors discuss
how the distribution system itself can contribute to this deterioration. Numerous examples of water-
borne outbreaks have demonstrated the  importance of the distribution system in preventing disease.
The authors discuss water-quality propagation models that can be used to study the factors that contrib-
ute to water-quality deterioration. These models have been used in many  locations to study contami-
nant propagation. This paper describes the application of contaminant  propagation models in the
SCCRWA. In this study, the fluoride feed was cut off at the water treatment plant to calibrate  the model
and determine residence times in the system. An extensive simulation of the system was conducted to
predict conservative contaminant propagation and chlorine decay. After completing the simulation
study, a sampling program was conducted to verify the results from the model. In general, the field
results verified the model predictions. Water quality varied widely over the service area. Long retention
times in storage tanks and pipe wall demand, especially in dead-end sections, caused significant losses
in chlorine residuals.

Clark et al. (1994) present models that  might be  used to evaluate the consequences of investing in
treatment  and/or investing in replacement, rehabilitation, or repair of the pipe network to improve
water quality. The authors point out that water high in humic and organic material that is transported in
the network can ultimately lead to pipes that have a high disinfectant demand. Investment in treatment
might not only meet water quality; it might also lead to a "cleaner" network.

Summary and Conclusions
Conventional treatment combined with disinfection has proven to be one of the major public health
advances of modern times. In the U.S., chlorine has been the final disinfectant most often used before
                                             12-28

-------
drinking water is discharged into a drinking water distribution system. It is added to provide a residual
and to protect against microbial contamination. However, disinfectants, especially chlorine, react with
NOM to form DBFs, which are considered to be of concern from a chronic exposure point of view.
Disinfection reduces risk of infectious disease, but the interaction between disinfectants and precursor
materials in source water result in the formation of DBFs.

Even treated drinking water exerts chlorine demand due to the reactions with NOM and other constitu-
ents in water. Therefore, the disinfectant dose must be sufficient to meet the inherent demand in the
treated water, to provide sufficient protection against microbial infection, and at the same time mini-
mize exposure to DBFs.

The factors that cause the disappearance of residuals and the subsequent formation of DBFs has been
the subject of extensive study. Much of this effort has  been devoted to models which are intended to
identify the factors that influence both residual decay and the formation of DBFs. This chapter has
reviewed both the EPA-supported research in this area as well as research conducted outside the Agency.
Clearly, much progress has been made in  developing realistic models to support risk management
goals. Of particular note is the application of models  to field conditions in water utilities. There is,
however, much research left to be done before these models are truly predictive.

References

Amy, G. L. (1993). "Using NOM characterization for the evaluation of treatment." Workshop Pro-
    ceedings, Natural Organic Matter in Drinking Water Origin, Characterization, and Removal,
    Chamonix, France,  September 19-22, 19.

Amy, G. L., Chadik, P. A., and Chowdhury, Z. K. (1987). "Developing models for predicting
    trihalomethane formation potential and kinetics." Journal of the American Water Works Associa-
    tion, 79(7), 89-97.

Bellar,  T A. and Lichtenberg, J. J. (1974).  "Determining volatile organics at microgram-per-liter levels
    by gas chromatography" Journal of the American Water Works Association, 66(12), 739-744.

Biswas, P., Lu, C., and Clark, R. M. (1993). "A model for chlorine concentration decay in drinking
    water distribution pipes." Water Research, 27(12),  1715-1724.

Boccelli, D. L., Tryby, M. E., Uber, J. G., Rossman, L.  A., Zierolf, M. L., and Polycarpu, M. M.
    (1998). "Optimal scheduling of booster disinfection in water distribution systems." Journal of
    Water Resources Planning and Management, ASCE,  124(2), 99-111.

Boulos, P. E, Altman, T., Jarrige, P. A., and Collevati, F. C. (1995). "Discreet simulation approach
    for network water quality models." Journal of Water Resources Planning and Management,
    ASCE, 121(1), 49-60.

Boulos, P. F., et al.  (1996). "Hydraulic mixing and free chlorine residual in reservoirs." Journal of
    the American Water Works Association, 48-59.

Boulos, P. F., Mau, R. E., and Clark, R. M. (1998). "Analytical equations of storage reservoir water
    quality." Civil Engineer ing Environmental Systems, 15, 171-186.

Chambers, V. K., Creasey, J. D., and Joy, J.  S. (1995). "Modelling free and total chlorine decay in
    potable water distribution systems." Journal of Water Supply, Research, and Technology—Aqua,
    44(2),  60-69.

Clark, R. M. (1993a). "Applying water quality models." Computer modeling of free-surface and pres-
    surized flow, M. H. Chaudry andL. M. Mays, eds., Kluwer Dordrecht, Netherlands, 581-612.

                                             12-29

-------
Clark, R. M. (1993b). "Development of water quality models." Computer modeling of free-surface
   and pressurized flow, M. H. Chaudry and L. M. Mays, eds., Kluwer Academic Publishers,
   Dordrecht, Netherlands, 553-580.

Clark, R. M. (1998). "Chlorine demand and TTHM formation kinetics: A second-order model."
   Journal of Environmental Engineering Division of the ASCE, 124(1), 16-24.

Clark, R. M. and Coyle, J. A. (1990). "Measuring and modeling variations in distribution system
   water quality." Journal of the American Water Works Association, 82(8), 46-53.

Clark, R. M. and Feige, W. A. (1993). "Meeting the requirements of the Safe Drinking Water Act."
   Strategies and Technologies for Meeting SDWA Requirements., R.  M. Clark and R. S. Summers,
   eds., Technomic Publishing Company, Inc., Lancaster, PA.

Clark, R. M. and Goodrich, J. A. (1992). "Modeling human exposure to contaminants from drinking
   water." Journal of Water Supply, Research, and Technology-Aqua, 41(4), 224-230.

Clark, R. M. and Goodrich, J. A. (1993). "Water quality modeling in distribution systems." Strate-
   gies and Technologies for Meeting SDWA Requirements, R. M. Clark and R. S. Sommers, eds.,
   Technomics Publishing, Lancaster, PA, 344-359.

Clark, R. M. and Sivaganesan,  M.  (1998). "Predicting chlorine residuals and the formation of
   TTHMs in drinking water." Journal of Environmental Engineering Division of the ASCE,
   124(12), 1203-1210.

Clark, R. M. and Sivaganesan,  M.  (in press). "Predicting chlorine residuals in drinking water: A
   second order model." Accepted for publication by the Journal of Water Resources, Planning, and
   Management.

Clark, R. M., Grayman, W. M., and Males, R. M. (1988a). "Contaminant propagation in distribution
   systems." Journal of Environmental Engineering Division of the ASCE, 114(4), 929-943.

Clark, R. M., Grayman, W. M., Males, R. M., and Coyle, J. (1988b). "Modeling contaminant propa-
   gation in drinking water distribution systems." Journal of Environmental Engineer ing Division of
   theASCE,3, 137-151.

Clark, R. M., Grayman, W. M., and Goodrich, J. A. (1991). "Water quality modeling: Its regulatory
   implications." Proceedings, American Water Works Association Research Foundation, Environ-
   mental Protection Agency Conference on Water Quality Modeling in Distribution Systems,
   Cincinnati, OH, February 4-5, 389-416.

Clark, R. M., Rossman, L. A., and Goodrich, J. A. (1992). "Modeling the variation in human expo-
   sure to contaminants from drinking water." Journal of Exposure Analysis and Environmental
   Epidemiology, 1, 59-175.

Clark, R. M., Goodrich, J. A., and  Wymer, L. J. (1993a). "Effect of the distribution system on drink-
   ing water quality." Journal of Water Supply, Research, and Technology-Aqua, 42(1), 30-38.

Clark, R. M., Grayman, W. M., Males, R. M., and Hess, A. F. (1993b). "Modeling contaminant
   propagation in drinking water distribution systems." Journal of Environmental Engineering
   Division of the ASCE, 119(2), 349-364.

Clark, R. M., et al. (1994). "Managing water quality in distribution systems: Simulating TTHM and
   chlorine residual propagation." Journal of Water Supply, Research, and Technology-Aqua, 43(4),
   182-191.
                                            12-30

-------
Clark, R. M., Rossman, L., and Wymer L. (1995). "Modeling distribution system water quality:
   Regulatory implications." Journal of Water Resources, Planning, and Management, ASCE,
   121(6), 423-428.

Clark, R. M., Abdesaken, E, Boulos, P. E, and Mau, R. (1996a). "Mixing in distribution system
   storage tanks: Its effect on water quality." Journal of Environmental Engineering Division of the
   ASCE, 122(9), 814-821.

Clark, R. M., Pourmoghaddas, H., Wymer, L. G., and Dressman, R.  C. (1996b). "Modeling the
   kinetics of chlorination by-product formation: The effects of bromide." Journal of Water Supply,
   Research,  and Technology-Aqua.,  45(1),  1-8.

Clark, R. M., Thurnau, R.  C., Sivaganesan, M., and Ringhand, P. (2001). "Predicting the formation
   of chlorined and brominated by-products." Journal of Environmental Engineering Division of the
   ASCE, 127(6), 495-501.

Cowman, G. A. and Singer, P. C. (1996). "Effect of bromide ion on haloacetic acid speciation result-
   ing from chlorination and chloramination of aquatic humic substances." Environmental Science
   and Technology, 30(1), 16-24.

Dugan, N. R., Summers, R. S., Miltner, R. J., and Shukairy, H. M. (1995). "An alternative approach
   to predicting chlorine residual decay." Proceedings, American Water Works Association Water
   Quality Technology Conference, New Orleans, LA, Nov. 12-16, 1317-1338.

Feben, D. and Taras, M. J. (1951). "Studies on chlorine demand constants." Journal of the American
   Water Works Association, 43(11),  922-932.

Freund, R. J. and Littell, R. C. (1991). SAS systems for regression, 2nd Ed., SAS Institute, Cary, NC,
   15-58.

Gessler, J. and Walski, T. M. (1985). "Water distribution system optimization."  TREL-85-11,  WES,
   U.S. Army Corps of Engineers, Vicksburg, MS.

Grayman, W. M. and Clark, R. M. (1993). "Using computers to determine the effect of storage on
   water quality." Journal of the American Water Works Association, 85(7), 67-77.

Haas, C. N. and Karra, S. B. (1984). "Kinetics of wastewater chlorine demand exertion." Journal of
   the Water Pollution Control Federation,  56(2), 170-173.

Hao, O. J., Davis, A. P., and Chang, P. H. (1991). "Kinetics of manganese(H) oxidation with chlo-
   rine." Journal of Environmental Engineering Division of the ASCE, 117(3),  359-374.

Harrington, G. W., Chowdhury, Z. K., and Owen, D. M. (1992). "Developing a computer model to
   simulate DBF formation during water treatment." Journal of the American Water Works Associa-
   tion, 84(11), 78-87.

Islam, M. R. (1995). "Modeling of chlorine concentration in unsteady flows in pipe networks." Ph.D.
   dissertation, Washington State University.

Islam, M. R., Chaudhry, M. H., and Clark, R. M. (1997). "Inverse modeling of chlorine concentra-
   tion in pipe networks under dynamic conditions." Journal of Environmental Engineering Divi-
   sion of the ASCE, 123(10), 1033-1040.

Jadas-Hecart,  A., El Morer, A., Stitou, M., Bouillot, P., and Legube, B. (1992). "Modelisation de la
   Demande en Chlore D'une Eau Traitee." Water Reserves, 26(8), 1073.
                                            12-31

-------
Kavanaugh, M. C., Trussell, A. R., Cromer, J., and Trussell, R. R. (1980). "An empirical kinetic
   model of trihalomethane formation: Applications to meet the proposed THM standard." Journal
   of the American Water Works Association., 72(10), 578-582.

Lee, B. H., Deininger, R. A., and Clark, R. M. (1991). "Locating monitoring stations in water distri-
   bution networks." Journal of the American Water Works Association, 83(7), 60-66.

Lu, C., Biswas, P., and Clark, R. M. (1993). "Modeling of breakpoint reaction in drinking water
   distribution pipes." Environmental International., 19(6),  543-560.

Lu, C., Biswas, P., and Clark, R. M. (1995). "Simultaneous transport of substrates, disinfectants and
   microorganisms in water pipes." Water Reserves, 29(3),  881-894.

Lyn, T. L. and Taylor, J. S. (1993). "Modeling compliance of chlorine residual and disinfection by-
   products." Proceedings, American Water Works Association Water Quality Technology Confer-
   ence, Miami, FL, Nov. 7-11, 513-523.

Mau, R., Boulos, P., Clark, R., Grayman, W., Tekippe, R., and Trussell R. (1995). "Explicit math-
   ematical models of distribution system storage water quality." Journal of Hydrolic Engineering,
   ASCE, 121(10), 699-709.

McKnight, A. and Reckhow, D. A. (1992). "Reactions of ozonation byproducts with chlorine and
   chloramines." Proceedings, American Water Works Association Annual Conference—Water
   Research, Vancouver, BC, June 18-22, 399-409.

Pourmoghaddas, H., Stevens, A. A., Kinman, R. N., Dressman, R. C., Moore, L. A., and Ireland,
   J. C. (1993). "Effect of bromide ion on the formation of HAAs during chlorination." Journal of
   the American Water Works Association, 85(1), 82-87.

Quails, R. G. and Johnson, J. D. (1983). "Kinetics of the short-term consumption of chlorine by
   fulvic acid." Environmental Science and Technology, 17(11), 692-698.

Rook, J. J. (1974). "Formation of haloforms during chlorination of natural waters." Water Treatment
   Examination, 23, 234.

Rossman, L. A. (1994). EPANET Users Manual. Drinking Water Research Division, Environmental
   Protection Agency, Cincinnati, OH.

Rossman, L. A. and Boulos, P. A. (1996). "Numerical methods for modeling water quality in distri-
   bution systems: A comparison" Journal of Water Resources Planning and Management,  ASCE,
    122(2), 137-146.

Rossman, L. A., Brown, R. A., Singer, P. C., and Nuckols, J. R. (1999). "Kinetics of TFEVI and HAA
   production in a simulated distribution system." Proceedings, American Water Works Association
    Water Quality  Technology Conference, Tampa, FL, October 31-November 3.

Rossman, L. A., Clark, R. M., and Grayman, W. M. (1994).  "Modeling chlorine residuals in drink-
   ing-water distribution systems." Journal of Environmental Engineering Division of the ASCE,
    120(4), 803-820.

Rossman, L. A. and Grayman, W. M. (1999). "Scale model studies of mixing in drinking water
   storage tanks." Journal of Environmental Engineering Division of the ASCE, 125(8), 755-761.

Rossman, L. A., Uber J. A., and Grayman, W. M. (1995). "Modeling disinfectant residuals in drink-
   ing water storage tanks." Journal of Environmental Engineering Division of the ASCE, 121(10),
   755-761.
                                            12-32

-------
Schenck, K. M, Wymer, L. 1, Lykins, Jr., B. W., and Clark, R. M. (1998). "Application of a Finnish
   mutagenicity model to drinking waters in the U.S." Chemosphere, 37(3), 451-464.

Shukairy, H. M. (1994). "The control of disinfection by-product formation by ozonation and
   biotreatment." Ph.D. dissertation, University of Cincinnati, Cincinnati, OH.

Shukairy, H. M. (1998). "Ozonation in drinking water treatment." Encyclopedia of Environmental
   Analysis and Remediation, R. A. Meyers, ed., John Wiley & Sons, Inc., New York.

Shukairy, H. M., Koechling, M. T., and Summers, R. S. (1994a). "Fractionated organic matter DBF
   formation: kinetics, reactivity and speciation." Proceedings, American Water Works Association
   Annual Conference., New York, NY, June 19-23, Water Research., 941-957.

Shukairy, H. M., Miltner, J. R., and Summers, R. S. (1994b). "Bromide's effect on DBF formation,
   speciation, and control: Part 1, Ozonation" Journal of the American Water Works Association,
   86(6), 72-87.

Shukairy, H. M., Miltner, R. J., and Summers, R. S. (1995). "Bromide's effect on DBF formation,
   speciation, and control: Part 2, Biotreatment." Journal of the American Water Works Association,
   87(10), 71-82.

Shukairy, H. M. and Summers, R. S. (1992). "The impact of preozonation  and biodegradation on
   disinfection by-product formation." Water Research, 26(9), 1217-1227.

Shukairy, H. M. and Summers, R. S. (1996). "DBF speciation and kinetics as affected by ozonation
   and biotreatment." Disinfection by-products in water treatment: The chemistry of their formation
   and control, R. A. Minear and G. L. Amy, eds., Lewis Publishers, Chelsea, MI, 311-335.

Skov, K.  R., Hess, A. F., and Smith, D. B. (1991). "Field sampling procedures for calibration  of a
   water distribution system hydraulic model." Proceedings, American Water Works Association
   Research Foundation, Environmental Protection Agency Conference on Water Quality Modeling
   in Distribution Systems, Cincinnati, OH, February 4-5, 193-200.

Trusell, R. R. and Umphres, M. D. (1978). "The formation of trihlomethanes." Journal of the Ameri-
   can Water Works Association, 70(10), 604-612.

Tryby, M. E., Boccelli, D. L., Koechling, M. T., Uber, J. G., Summers, R. S., and Rossman, L. A.
   (1999). "Booster chlorination for managing disinfectant residuals." Journal of the American
   Water Works Association, 92(1), 96-108.

United States Environmental Protection Agency (USEPA). (1992). Water treatment plant simulation
   program user's manual, Version 1.21. Drinking Water Technology Branch, Drinking Water
   Standards Division, Office of Ground Water and Drinking Water, Malcolm Pirnie, Inc.

Vartiainen, T. and Liimatainen, A. (1988). "Relations between drinking water mutagenicity and water
   quality parameters." Chemosphere, 17(1),  189-202.

Vasconcelos, J. J., Rossman, L. A., Grayman, W. M., Boulos, P. F., and Clark, R. M. (1997). "Kinet-
   ics of chlorine decay" Journal of the American Water Works Association, 89(7),  55-65.

Ventresque, C., Bablon, G., Legube, B., Jadas-Hecart, A., and Dore, M. (1990). "Development of
   chlorine demand kinetics in a drinking water treatment plant." Water chlorination: Chemistry,
   environmental impact and health effects, R. L. Jolley, et al., eds., Vol. 6, Lewis Publications, Inc.,
   Chelsea, MI, 715-728.
                                             12-33

-------
White, G. C. (1999). Handbook of chlorination and alternative disinfectants. J. Wiley & Sons, Inc.,
   New York, 257-258.

Zhang, G. R., Keine, L., Wable, O., Chan, U. S., and Duget, J. P. (1992). "Modeling of chlorine
   residual in the water distribution system network of Macao." Environmental Technology, 13,
   937-946.
                                            12-34

-------
                                    CHAPTER 13

                 Biofilms in Drinking Water Distribution Systems1
Introduction
Virtually anywhere a surface comes into contact with the water in a distribution system, one can find
biofilms. Biofilms are formed in distribution system pipelines when microbial cells attach to pipe
surfaces and multiply to form a film or slime layer on the pipe. Probably within seconds of entering
the distribution system, large particles, including microorganisms, adsorb to the clean pipe surface.
Some microorganisms can adhere directly to the pipe surface via appendages that extend from the
cell membrane; other bacteria form a capsular material of extracellular poly saccharides (EPS), some-
times called a glycocalyx, that anchors the bacteria to the pipe surface (Geldreich 1988). The organ-
isms take advantage of the macromolecules attached to the pipe surface for protection and  nourish-
ment. The water flowing past  carries nutrients (carbon-containing molecules, as well as other ele-
ments) that are essential for the organisms' survival and growth (USEPA 1992).

Biofilms are complex and dynamic microenvironments, encompassing processes such as  metabo-
lism, growth, and product formation, and finally detachment, erosion, or "sloughing" of the biofilm
from the surface. The rate of biofilm formation and its release into a distribution system can be
affected by many factors including surface characteristics, availability of nutrients, and flow veloci-
ties. Biofilms appear to grow until the surface layers begin to slough off into the water (Geldreich and
Rice 1987). The pieces of biofilm released into the water may continue to provide protection for the
organisms until they can colonize a new section of the distribution system.

Few organisms living in distribution system biofilms pose a threat to the average consumer. Bacteria,
viruses, fungi, protozoa, and other invertebrates have been isolated from drinking water  biofilms
(USEPA 1992). The fact that such organisms are present within distribution system biofilms shows
that, although water treatment is intended to remove all pathogenic (disease-causing) bacteria, treat-
ment does not produce a  sterile water. In fact, some otherwise harmless organisms (opportunistic
pathogens) may survive the treatment process and cause disease in individuals with low immunity or
compromised immune systems.

Bacteria comprise the largest portion  of the biofilm population. These organisms may survive the
disinfection process to colonize the distribution system at the time of installation, or they may be
introduced through cross connections,  backflow events, line breaks, or repair operations. The public
health risk from these organisms is not known (Geldreich 1990). Although biofilms may represent
the greatest concentration  of biological material (biomass) in the distribution system, health surveys
conducted in systems experiencing biofilm growth problems (New Haven, CT; Springfield, IL; and
Muncie, IN) have revealed no increase in illnesses due to contaminated drinking water (Geldreich
1988). Most bacteria survive in disinfected drinking water by finding or creating environments where
they are protected from the disinfectant residual. Factors related to increased survival of bacteria in
chlorinated water include attachment  to surfaces, encapsulation, aggregation, low-nutrient growth
conditions, and strain variation.
'Mark Meckes: ORD/NRMRL/WSWRD, AWBERC Mailstop 689, 26 West Martin Luther King Dr.:
Cincinnati, OH 45268, 513-569-7348, meckes.mark@epa.gov.
                                              13-1

-------
Corrosion provides a protective surface for microorganisms, slows water flow, and contributes to
backflow occurrences where iron pipe walls corrode. In iron pipes, electrochemical reactions at the
pipe surface dissolve the metal to form pits (releasing free ferrous ions) at one point while building a
tubercle or nodule (composed of ferric hydroxide) at a remote spot. The pits and nodules formed may
catch and concentrate nutrients and provide the organisms with protection from water shear (Allen and
Geldreich 1977). Free chlorine itself promotes the pitting type of corrosion by reacting with the ferrous
ions and precipitation of ferric hydroxide. This not only accelerates corrosion but also represents an-
other demand on the free chlorine residual (USEPA 1984).

Some heterotrophic bacteria that live in biofilms may cause esthetic problems with water quality, in-
cluding off-tastes, odors, and  colored water problems. Biofilm organisms that fall into this nuisance
category include Actinomyces, Streptomyces, Nocardia, andArthrobacter (Geldreich 1990). Complaints
about taste and odor have resulted from Streptomyces and Nocardia spp. at concentrations greater than
10 organisms per 100 mL of water. For pigmented bacteria, the degree of pigment formation observed
in cultured cells will depend on the media used for isolating the bacteria in the water sample. Many
heterotrophic plate count (HPC) bacteria isolated from distribution system biofilms will produce yel-
low, orange, or pink colonies when grown on R2 A agar (Geldreich 1990;Reasoneretal.  1989;Reasoner
and Geldreich 1990; Carter et al. 2000).

Previous Research

Organic carbon is utilized by heterotrophic bacteria for production of new cellular material  (assimila-
tion) and as an energy source (dissimilation). Because heterotrophic bacteria require carbon, nitrogen,
and phosphorus in a ratio of approximately 100:10:1 (C:N:P), organic carbon is often a growth-limiting
nutrient. Most organic carbon compounds in water supplies are natural in origin, derived from living
and decaying vegetation. These compounds may include humic and fulvic acids, polymeric carbohy-
drates, proteins, and carboxylic acids.

Kaplan and Bott (1990), in a study conducted for the U.S. Environmental Protection Agency (EPA),
evaluated the effect of nutrients on bacterial growth in  drinking water. Their evaluation showed that
incubation vessel surface-to-volume ratio influenced the assimilable organic carbon (AOC) value by
enhancing wall growth of reversibly attached cells. The authors noted that the underlying assumptions
for the AOC bioassay include (1) organic carbon limits growth of the bioassay organism, (2) the yield
of the bioassay organism on naturally occurring  AOC is constant and equal to the yield  on model
organic compounds, and (3) the bioassay organism is an appropriate surrogate for the native microflora
of distribution systems  in utilizing AOC. Their research showed that some test waters required the
addition of phosphorus in order to generate carbon limitation. They also showed that AOC concentra-
tions in a small  sampling of surface water sources ranged from 48 to 607 |lg/L, while a ground water
sample yielded  AOC values from 40 to 146 |ig/L. Kaplan et al. (1993) modified the AOC bioassay
procedure to minimize the potential for contamination and to simplify the procedure so that it could be
used routinely by water utilities. These researchers compared media, culture vessels, glassware clean-
ing methods, physiological condition of the inoculum, and treatments of the inoculation water and
concluded that the AOC method could be simplified by the use of precleaned vials as culture vessels.
They also noted that the organisms commonly used for the bioassay, Psuedomonasfluorescens P-17
and Spirillum sp. strain NOX, need not be in log growth phase when inoculated into the culture vessels.
They also noted that replication of incubation vessels is the most efficient way to reduce variance and
the cost of the AOC measurement.

Researchers at the Compagnie Generate des Eaux in France developed a method to measure biodegrad-
able dissolved organic carbon (BDOC) (Pascal et al.  1986; Hascoet et al. 1986; Servais et al. 1987),
                                             13-2

-------
which is the fraction of DOC which is biodegraded by naturally occurring flora under controlled condi-
tions. In the BDOC test, the concentration of DOC for a given water sample is determined. Indigenous
bacteria are then allowed to grow for a specified time  within another aliquot of the sample under
controlled conditions. Finally, these samples are then filtered through prewashed 0.22-|im membrane
filters, and an organic carbon analyzer is used to measure the DOC remaining in the water. The differ-
ence in the DOC concentration (initial-final DOC) is the  BDOC. If the bacteria are incubated in water
samples for 10 to 30 days, the test allows measurement of slowly degradable organic materials (Pascal
et al. 1986). This procedure has some disadvantages, including insensitivity at low DOC levels and the
relatively high cost of a total organic carbon (TOC) analyzer. There are no operational data to relate
specific BDOC levels to HPC or coliform problems; however, a level of less than 0.1 BDOC mg/L is
thought to produce biologically stable water (i.e., water that is unable to support bacterial growth).

Rice et al. (1990) tested the regrowth potential of three species of coliform bacteria in source, partially
treated, and finished water samples. These researchers selected a strain of Enterobacter cloacae as the
organism of choice for conducting a bioassay for determining the potential for coliform regrowth in
waters. In this bioassay, an acclimated culture of the organism is added to filter-sterilized water samples
and incubated in the dark for 5 days at 20°C. After 5 days,  the density of the organism is determined and
compared to the initial density. If the Iog10 growth is less than 0.5, the water is not considered to support
coliform  growth. If the Iog10 growth is equal to or greater than 0.5 and less  than 1.0, the water is
considered to be moderately supportive of coliform growth. If the Iog10 growth is equal to or greater
than 1.0, then the water is considered to be supportive of coliform growth. Results of the assay on three
water sources with multiple treatments showed that water treated with ozone or chlorine and, in one
case, coagulated and filtered water yielded a coliform growth response (CGR) of greater than one log,
indicating that these waters were capable of supporting growth of coliform organisms.

Shortly after this, Rice et al. (1991) tested the CGR in a variety of water types from different geographi-
cal areas and at different stages of water treatment. They found that coliform growth responses corre-
lated with assimilable organic carbon concentrations. They noted that the correlation, although signifi-
cant (p < 0.05), had a low coefficient of determination (8.5%). Using analysis of variance techniques,
the authors were able to demonstrate that mean CGR values increased with increasing values of AOC.
They also found that significantly higher coliform growth responses were associated with waters that
had been exposed to ozonation. This work was completed prior to the introduction of the use of Spiril-
lum sp. NOX as a second AOC bioassay organism. This strain is better able to determine the utilization
of substrates such as oxalate and  other carboxylic acid compounds formed after ozonation. The authors
speculate that, if both bioassay organisms had been used, the differences in AOC levels between ozonated
and unozonated samples would have been much greater than they reported. They go on to state that the
CGR assay does not measure the same parameter as the AOC procedure, and note that this is due to the
widely held opinion that the nutrient threshold level for members of the family Enterobacteriaceae
exceeds that of the family Pseudomonadaceae.  They also note that the lack of a significant positive
correlation between the CGR and any of the measured chemical parameters is  indicative of the com-
plex nutritional requirements of these coliforms and that this finding parallels the relationship between
measured chemical parameters and AOC.

Kaplan et al. (1994) compared three methods for determining biodegradable organic matter (BOM) in
a broad range of U.S. drinking waters and treatment processes. The AOC assay, the BDOC assay, and
the  CGR were used to determine the potential for bacterial regrowth in waters from 109 different
sources. These included 53 from surface sources, 26 from ground waters, and 40 collected from a
variety of treated waters. Specified utilities were sampled quarterly for a period of 1  year in order to
determine if BOM concentrations varied seasonally. Results showed that the source of water samples
(e.g., ground water versus surface water) more so than the treatment of the waters had a strong influ-
                                              1O "
                                              13-j

-------
ence on the AOC and BDOC concentrations. The relationship between AOC and BDOC appeared to be
best when AOC values were calculated on the basis of oxalate carbon rather than acetate. No conclu-
sion was presented regarding the CGR assay. This research suggests that the AOC  and BDOC are
complementary techniques that can be applied for measurement of BOM in water supplies.

The density of organisms entering a water distribution system also affects the biofilm within such
systems. Mathieu et al. (1993) used a distribution system simulator to determine the effect of bacterial
flux on biofilm formation. The simulator, which was made up of three 30-meter loops  of 10-cm diam-
eter cement-lined pipe was operated under conditions of low and high nutrients and low and high levels
of disinfecting agents. After several weeks of operating the simulator, they observed that biofilm did
accumulate on cement coupons in the presence of relatively high (0.43 mg/L chlorine  or 1.06 mg/L
chloramine disinfectant residuals). They also noted that the chloraminated system contained more than
3.8 times as many cells/cm2 than did the system treated with chlorine. A linear relationship was found
when the results  from the chloraminated system were used to compare the biofilm (log cells/cm2) to
number of cells in the influent to the simulator loop. This relationship showed that an increase in one
log of bacterial cells in system feed water yielded a 0.32 log increase in biofilm density. This demon-
strated that the flux of bacteria within a water distribution system can contribute to biofilm accumulation.

Block et al. (1995) briefly reviewed the literature available on biodiversity in drinking water distribu-
tion systems. They noted that different trophic levels have been identified within systems  of heterotrophic
bacteria, free-living protozoa,  and macroinvertebrates. They also found that the bacterial species seem
to be as diverse as those found in natural systems since more microorganisms develop  opportunistic
strategies  for occupying environments with different temperatures, nutrients, and  toxins. They also
noted that others have shown that relatively high densities of bacteria, protozoa, and macroinvertebrates
can be found within water distribution systems, and there is some indication that seasonal variations in
populations exist. Predation of bacteria was noted as having been demonstrated by a number of re-
searchers, but they also noted that the clearing rate has not been documented. They go on to speculate
that this may in part be due to the fact that such activity is highly variable. The authors did note that the
highest protozoan populations correspond with high bacterial populations. On the other hand, they also
noted that, when an experimental distribution system was colonized by E. coli under varying water
quality conditions, the E. coli  density remained high in low-nutrient water. This was attributed to low
protozoan populations in the low-nutrient content water.

Carter et al. (2000) examined water samples from four sites within a water distribution system. At two
of the sampling locations (a pump station and the terminal point in the system), they  installed monitor-
ing devices which continuously measured pH, conductivity, water temperature, and free chlorine. At
these same locations, they also installed biofilm  collection devices. Weekly water samples were as-
sayed for AOC and TOC. Heterotrophic bacteria were measured using three different assay techniques,
plate count agar pour plates (PCA), R2A agar spread plates (R2A), and tryptic soy agar with 5% sheep's
blood spread plates (TSA-SB). Similarly, they sampled biofilm collection devices on a weekly basis
and assayed these samples for heterotrophic  bacteria using the same three techniques. Their results
showed that plate counts from the three heterotrophic bacteria assays differed markedly between the
bulk fluid and biofilm samples. They noted that  samples from the bulk fluid generated plate counts
ranging from about 4 to 400 CFU/mL, while  samples from the biofilm collector yielded plate counts
ranging from 5 x 103 to 1.5 x 106 CFU/cm2. They  also found that biofilm densities measured using the
PCA and R2A assays yielded  similar results, while the PCA and TSA-SB techniques yielded compa-
rable results for analysis of fluid samples. These researchers also noted that the number of bacterial
colonies exhibiting pigmentation increased from 57%  in fluid samples to 76% in biofilm samples
which were taken from the terminus.
                                             13-4

-------
Pilot Systems Research

In situ sampling of distribution system biofilms presents numerous obstacles. Consequently, distribu-
tion system simulators are often used to evaluate the effects of varying water quality parameters on
distribution system components and biofilms. Clark et al. (1994) used a distribution system simulator
to evaluate the effect of various disinfectants on water quality parameters. In this study, source water
treated by one of four treatment trains was disinfected with chlorine or chloramine. These were inde-
pendently examined in two distribution system simulators which were constructed of three 31-meter
lengths of 100-mm diameter, cement-lined, cast iron pipe. Results showed that, as disinfectant residu-
als decreased within the simulator, biofilm densities increased, regardless of the disinfectant used. No
discernable difference was noted in the density of biofilm organisms due to the disinfecting agent used.
The biomass that did accumulate was heterogeneous, and growth of surface-associated cells correlated
with a decrease in BDOC. These authors also noted that, when a residual chlorine concentration was
maintained above 0.4 mg/L, the density of biofilm was much lower on the cement pipe material than on
polyethylene.

Clark and Sivaganesan (1999), using data from the previously cited study, developed models which
may be used to predict bacterial densities in the bulk water phase and wall biofilms. Data  from one of
the four treatment trains displayed a linear relationship between the log of the wall microbial density
(measured by microscopic identification of epifluorescent cells) and residual disinfectant.  The authors
used analysis of covariance techniques to demonstrate that chlorine was significantly more effective
than chloramine in reducing epifluorescent cell counts in bulk water samples. Similarly, they demon-
strated that, when the data from all treatments was considered, the effect of chlorine on log transforms
of the epifluorescent cell counts from biofilm samples was significantly greater (p < 0.05) than the
effect of chloramine.

Cement pipe and polyethylene pipe have surfaces which are considered to be non-reactive with respect
to oxidizing disinfectants such as chlorine or its combined forms. Unlined cast or ductile iron pipe
surfaces do react with oxidizing agents, which results in corrosion and reduction of residual oxidants.
Although no longer used for new installations, unlined cast iron piping remains in use for water distri-
bution throughout the U.S. Consequently, the effects on water quality by these reactive materials is of
concern. The EPA National Risk Management Research Laboratory (NRMRL) Distribution System
Simulator (DSS-1) was fabricated for the purpose of studying the impact of distribution systems on
drinking water quality. This simulator is housed at the NRMRL Test and Evaluation (T&E) Facility
located in Cincinnati, OH. The DSS-1 (see Figure  13-1) is made up of six individual 27-m lengths of
15-cm inside-diameter ductile iron pipe configured as independent loops. Stainless steel tanks are lo-
cated at the base of each loop for the purpose of providing a reservoir which feeds a centrifugal pump
used for recirculating water within a loop. Each loop is insulated and equipped with a shell and tube
heat exchanger which is used to control the temperature of the water  as it moves through a loop. All
loops are equipped with valved coupon assemblies. The assembly is designed so that the  surface of a
coupon fits flush with the interior pipe wall. The valved assembly allows for the removal of the coupon
without disruption of flow.

Recirculating flow rates are measured by in-line magnetic flow meters. Feed water flow is measured by in-
line rotometers. An instrumentation loop is fitted with pH, oxidation reduction potential (ORP), dissolved
oxygen, and pressure and temperature sensors  which measure the condition of the water  as it moves
through the loop. Residual free chlorine is measured using a flow-through monitor, and turbidity is mea-
sured using a flow-through nephlometer. Signals from these instruments are polled, recorded, and archived
using a supervisory controlled data acquisition system (SCADA). The frequency at which a given instru-
ment is polled may be varied to accommodate data collection requirements for a simulation.
                                              13-5

-------
                         I   I   I   I  I   I   I   I  I   I   I   I   I   I   I   I  I   I
                                              Coupons
                   Flow Meter
                                                          Ductile Iron Pipe
                                                          Length = 27m
                                                          Inside Diameter = 15cm
                   Heat Exchanger
                 Temperature -
                Chemical Feed
                   Tank
                                           Recirculation Tank
                            pH ORP Turbidity Chlorine
                                Meter Loop
Figure 13-1. Schematic diagram of a test loop.


Using the DSS-1, Meckes et al. (1999) observed that biofilm growth could be diminished when organ-
isms are subjected to variations in pH. They suggested that pH adjustments to water within distribution
systems could reduce or control biofilm growth. The authors used five of the DSS loops operating in
parallel under identical conditions for a period of 2 weeks. They noted that the initial density of biofilm
organisms varied between test loops; therefore, one loop was used as a control with no modification of
the feed water pH (8.0 standard units), while the other four loops were operated at pHs of 5, 6, 9, and
10. They noted that biofilm densities remained stable within the loops that received alkaline waters,
whereas the loops which received acidic water showed a marked reduction in biofilm densities. This
was most dramatic in the loop which received pH 5 water, which showed a thousand-fold reduction in
viable biofilm organisms. Following 3 weeks of operating at the adjusted pH levels, the pH additives
were turned off, and the test loops were permitted to return to their original operating conditions.
Within 1 week, the authors noted that biofilm levels increased within the loops which had previously
received acidic waters. They also noted that calcium concentrations in the water increased in delivered
water samples obtained from the loops which received acidic water and noted that the calcium concen-
tration returned to its original level when the pH of the water returned to its original operating condi-
tion. These results suggest that the reduction of biofilm that was observed was due to substrate destabi-
lization rather than growth inhibition.

In addition to the DSS-1, the NRMRL fabricated a second simulator (DSS-2). This simulator is con-
structed of two individual pipe systems: one is constructed of polyvinyl chloride (PVC), while the
second system is constructed of cement-lined steel pipe. Both of these pipe materials are considered to
provide non-reactive wetted surfaces. Each pipe has a 15-cm inside diameter and is approximately 110
m long. Several 2.5-cm diameter pipes extend vertically from the main pipe sections. These valved
lines are designed to simulate service connections and are used as sample collection points. Both loops
are equipped with valved coupon assemblies. These assemblies are designed so that up to 20 coupons
can be inserted into the pipe at one of three locations. The valved assembly allows for the removal of
one or more coupons without disruption of flow. Quality of the source water for these systems can be
                                              13-6

-------
adjusted by addition of agents to a mixed head tank which feeds both pipes. Flow rates through the
pipes are monitored by a magnetic flow meter and by rotometers to ensure accurate flow measurement
throughout the design range of 0 to 75 L/min. These simulators were designed as single-pass systems
without provision for water recycle.
Li et al. (1998) described the PVC pipe system. Using a fluoride  tracer, they compared the actual
hydraulic residence time of the system to the theoretical detention time under various flow conditions.
The results from  this  work demonstrated that the flow dispersion  models described by Levenspiel
(1972) closely matched  the observed dispersion of tracer. These data, collected under various flow
regimes, can be used to calibrate the dispersion model. This information defines the hydraulic condi-
tions of the system under various flow regimes. With this information in hand, these researchers will
conduct additional experiments designed to determine the effect of various flow regimes on biofilm
densities and residual chlorine within this simulator.

On-Going and Future Biofilm Research

In addition to the  planned research noted previously, the EPA DSS-1 is currently being used to deter-
mine the effectiveness of various disinfecting agents on biofilms. Biofilms are known to be resistant to
residual oxidants; however, combined forms of chlorine are less reactive than chlorine itself. This
study is designed to determine if chlorine, chloramines, or mixed oxidants (MIOX) residuals are equally
effective in controlling biofilm. Other studies are planned which will determine the effect of various
operating conditions and nutrient levels  on biofilm growth.

In a cooperative effort with the University of Montana's Biofilm Research Center, EPA is studying the
interactions among  factors that influence biofilms, bacterial regrowth, and corrosion in distribution
systems. The goal of this work is to generate information which can lead to a better understanding of
the interactions among those factors which influence microbial growth in water distribution systems
and the mitigating effects of chlorination and commonly used corrosion control techniques. This re-
search is also designed to address specific fundamental questions about the availability of sorbed hu-
mic substances for biofilm growth.

Much of the work on isolation and enumeration of biofilm organisms has been through the use of
microscopy or cultural methods. EPA, in a cooperative agreement with the University of Illinois,  is
developing molecular biology tools which can be used for characterizing the biofilm community. These
tools include fluorescent in situ hybridization (FISH) techniques which can be used to quantify the
abundance and activity of biofilm populations and to visualize the three-dimensional structure of the
biofilm community and the location of potential pathogens within these communities. Also to be evalu-
ated will be solution-based hybridization methods for the rapid and automated detection and quantifi-
cation  of selected candidate contaminant  list (CCL) organisms such as Aeromonas hydrophila and
Mycobacterium avium complex and indicator organisms such as E. coli. This work will also compare
new molecular methods with currently used bacterial detection methods.
In a similar but more focused effort, EPA is developing group-specific oligonucleotide probes for de-
tection of non-tuberculosis mycobacteria (NTM) such as Mycobacterium avium in drinking water and
biofilms. This  work is designed to determine if paraffin-baiting techniques can be effectively used to
capture NTM in biofilm and finished water samples. If successful, this technique will be used to deter-
mine the numbers, types, and location of planktonic and biofilm NTM in the EPA DSSes and can lead
to the development of reliable control strategies for NTM in distribution system biofilms.
                                              13-7

-------
Summary and Conclusions

Biofilms are found in virtually every water distribution system. The most common organisms found in
biofilms are nonpathogenic heterotrophic bacteria. Some bacteria that live in biofilms may cause es-
thetic problems with water quality, including off-tastes, odors, and colored water problems. The fact
that such organisms are present within distribution system biofilms shows that treatment does  not
produce a sterile water. Consequently, if opportunistic pathogens survive in biofilms, they could poten-
tially  cause disease  in individuals  with low-immunity or compromised immune systems. Factors re-
lated to increased survival of bacteria in chlorinated water include attachment to surfaces, encapsula-
tion, aggregation, low-nutrient growth conditions, and strain variation.

There are numerous factors which affect the growth of biofilms within distribution systems. One of the
most important factors is the availability of BOM. Much work has been done leading to the develop-
ment of methods which can be used for determination of BOM. These include: AOC, BDOC, and the
coliform growth response. Although each of these methods measures BOM, the results from compari-
son tests between methods show that they do not correlate well with one another. Such results indicate
that additional work in determining the best way of measuring BOM  in water may need to be con-
ducted.

Numerous problems exist in the development of experimental protocols which could be used to evalu-
ate biofilm growth in full-scale distribution systems. DSSs have been developed and used to determine
the effect that such systems have on water quality. Much of the work conducted using these simulators
has resulted in  determining the effect of water-quality changes on biofilms in the systems. This re-
search has indicated that biofilms are resistant to disinfection regardless of the agent used. Other work
has demonstrated that lowering system pH can reduce biofilm densities; however, such an effect is
transient if system pHs are returned to normal operating ranges.

Additional work on biofilms within distribution systems is currently underway. This work is designed to
further assess the  effect of water-quality parameters and system operations on biofilm densities. Other
research efforts  are focused on identification of specific organisms within biofilms and determining the
effectiveness of disinfecting agents  on these organisms. These efforts are being conducted to determine if
biofilm contributions to delivered water may require treatment modifications or amendments.

References

Allen, M. J.  and Geldreich, E. E. (1977). "Distribution line sediments and bacterial regrowth."
   Proceedings, American Water Works Association Water Quality Technology Conference, Kansas
   City, MO, December 4-7.

Block, J. C., Sibille, I, Gatel, D., Reasoner, D. J., Lykins, B., and Clark, R. M. (1995). "Biodiversity
   in drinking water distribution systems: A brief review." The microbiological quality of water.
   From the IWSA andFBA Specialized Conference, London, U.K., December.

Carter, J. T., Rice, E. W., Buchberger, S. G., and Lee, Y. (2000). "Relationships between levels of
   heterotrophic bacteria, and water quality parameters in a drinking water distribution system."
   Water Research, 34(5), 1495.

Clark, R. M., Lykins, B. W., Block, J. C., Wymer, L. J., and Reasoner, D. J. (1994). "Water quality
   changes  in a simulated distribution system." Journal of Water Supply, Research, and Technol-
   ogy-Aqua, 43(6), 263.
                                             13-8

-------
Clark, R. M. and Sivaganesan, M. (1999). "Characterizing the effect of chlorine and chloramines on
   the formation of biofilm in simulated drinking water distribution systems. EPA/600X-99/027,
   U.S. Environmental Protection Agency, Office of Research and Development, Cincinnati, OH.

Geldreich, E. E. (1988). "Chapter 3: Coliform noncompliance nightmares in water supply distribu-
   tion systems." Water quality: A realistic perspective. University of Michigan, College of Engi-
   neering; Michigan Water Pollution Control Association; Michigan Department of Public Health,
   Lansing, MI.

Geldreich, E. E. (1990). "Microbial quality control in distribution systems." Water quality and
   treatment, 4th Ed., F. W. Pontius, ed., American Water Works Association, McGraw-Hill, Inc.,
   New York.

Geldreich, E. E. and Rice, E. W. (1987). "Occurrence, significance, and detection ofKlebsiella in
   water systems." Journal of the American Water Works Association, 79(5), 74.

Hascoet, M. C., Servais, P., and Billen,  G. (1986). "Use of biological analytical methods to optimize
   ozonation and GAC filtration in surface water treatment." Proceedings, American Water Works
   Association Annual Conference, Denver, CO, June 22-26, 205.

Kaplan, L. A. and Bott, T. L. (1990). "Nutrients for bacterial growth in drinking water: Bioassay
   evaluation." EPA/600/S2-89/30, U.S. EPA Office of Research and Development, Risk Reduction
   Engineering Laboratory, Cincinnati, OH.

Kaplan, L. A., Bott, T. L., and Reasoner, D. J. (1993). "Evaluation and simplification of the assimi-
   lable organic carbon nutrient bioassay for bacterial growth in drinking water." Applied and
   Environmental Microbiology, 59(5), 1532.

Kaplan, L. A., Reasoner, D. J., and Rice, E. W. (1994). "A survey of BOM  in U.S. drinking waters."
   Journal of the American Water Works Association,  86(2),  121.

Levenspiel, O. (1972). Chemical reaction engineering, 2nd Ed., John Wiley & Sons, New York.

Li, S. Y, Biswas, P., Clark, R. M., Meckes, M. C., Dosani, M. A., and Krishnan, E. R. (1998).
   "Detection of chlorine decay patterns in water distribution dead-end flow regimes under various
   hydrodynamic conditions." Proceedings, American Water Works Association Water Quality
   Technology Conference, San Diego, CA., November 1-4.

Mathieu, L., Block, J. C., Dutang, M., Mailliard, J., and Reasoner, D. J. (1993). "Control of biofilm
   accumulation in drinking water distribution systems." Water Supply, 11, 365.

Meckes, M. C., Haught, R. C., Dosani,  M., Clark, R. M., and Sivaganesan,  M. (1999). "Effect of pH
   adjustments on biofilm in a simulated water distribution system." Proceedings, American Water
   Works Association Water Quality Technology Conference, Tampa, FL, October 31-November 3,
   1993.

Pascal, O., Joret, J. C., Levi, Y, and Dupin, T. (1986). "Bacterial aftergrowth in drinking water
   networks measuring biodegradable  organic carbon (BDOC)." Proceedings, Ministere de
   I'Environnement/U.S. Environmental Protection Agency Franco-American Seminar, Cincinnati,
   OH, October 13-17.

Reasoner, D. J. and Geldreich, E. E. (1990). "Distribution systems: Treated water quality versus
   coliform noncompliance problems." Methods for the investigation and prevention ofwaterborne
   disease outbreaks. EPA/600/l-90/005a, U.S. Environmental Protection Agency, Office of Re-
   search and Development, Washington,  D.C.
                                              13-9

-------
Reasoner, D. J., Blannon, J. C., Geldreich, E. E., and Barnick, J. (1989). "Nonphotosynthetic pig-
   mented bacteria in a potable water treatment and distribution system." Applied and Environmen-
   tal Microbiology, 55(4), 912.

Rice, E. W., Scarpino, P. V, Logsdon, G. S., Reasoner, D. J., Mason, P. J., and Blannon, J. C. (1990).
   "Bioassay procedure for predicting coliform bacterial growth in drinking water." Environmental
   Technology, 11,821.

Rice, E. W., Scarpino, P. V, Reasoner, D. J., Logsdon, G. S., and Wild, D. K. (1991). "Correlation of
   coliform growth response with other water quality parameters." Journal of the American Water
   Works Association, 83(7), 98.

Servais, P., Billen, G., and Hascoet, M. C. (1987). "Determination of the biodegradable fraction of
   dissolved organic matter in waters."  Water Research, 21, 445.

United States Environmental Protection  Agency (USEPA). (1992). "Seminar publication: Control of
   biofilm growth in drinking water distribution systems." EPA/625/R-92/001, Office of Research
   and Development, Washington, D.C.

USEPA. (1984). "Corrosion manual for internal corrosion of water distribution systems." EPA/570/9-
   84/001, Office of Drinking Water, Washington, D.C.
                                             13-10

-------
                                     CHAPTER 14

 Control of Microbial Contaminants and Disinfection By-Products (DBPs): Cost
                                   and Performance1

Introduction

The U.S. Environmental Protection Agency (EPA) is in the process of developing a sophisticated regu-
latory strategy in an attempt to balance the complex trade-offs in risks associated with controlling
disinfectants and disinfection by-products (D/DBPs) in drinking water. EPA first attempted to control
DBPs in 1974, when trihalomethane (THM) formation in drinking water was identified as a by-product
of chlorination. Based on the toxicologic data from the 1970s, chloroform (one of the THMs) was
labeled as a suspect carcinogen.

Epidemiological studies also suggested a human risk. Because of these  suspected health effects and the
potential that a large number of drinking water consumers would be exposed to these by-products, a
Total Trihalomethane (TTHM) Regulation was promulgated on November 29, 1979, at a level of 0.10
mg/L (Clark et al. 1994). Since that time, many other objectionable by-products of chlorination have
been identified as well.

This chapter will review the current status of disinfection practices in the U.S., the conditions that
cause the formation of DBPs, and discuss the various treatment techniques and associated costs for
both controlling DBPs and ensuring microbial safety.

The Role of Disinfection in  the U.S.

In the U.S., an estimated 220,000,000 people receive disinfected drinking water (Clark et al.  1994).
Chlorine has been the disinfectant  of choice for many utilities, and more than 50% of the systems using
surface water use chlorine prior to settling and filtration. Many utilities in the U.S. have explored the
use of disinfectants other than chlorine to lower their DBF levels below the 0.80 mg/L TTHM and 0.60
mg/L HAA limits. Some utilities have considered switching to chloramines as an alternate disinfectant
to chlorine. A survey conducted by  the American Water Works Association Research Foundation showed
that the vast majority of utilities  that changed disinfection practices  have switched to chloramines
(McGuire and Meadow 1988). In  these cases, chloramines are applied as the final disinfectant. Some
utilities are considering the possibility of using ozone as a disinfectant followed by chlorine or chloram-
ines. Ozone is  drawing increasing interest, but concern with using ozone includes the need for
biostabilization of the treated water and possible formation of by-products such as bromate, aldehydes,
ketones, and acids. Chlorine dioxide is an effective disinfectant, but there are concerns about its react-
ing to form the inorganic by-products, chlorite and chlorate.
'Jeffrey Q. Adams and Robert M. Clark: Water Supply and Water Resources Division, NRMRL, U.S.
Environmental Protection Agency, 26 W. Martin L. King Dr., Cincinnati, OH 45268. Corresponding
Author: Jeffrey Q. Adams, 513-569-7835, adams.jeff@epa.gov.
                                             14-1

-------
Microbiological Control

Prior to the discovery that protozoan cysts (Giardia and Cryptosporidium) were a prime cause of water-
borne disease outbreaks, the apparent attainment of adequate disinfection was considered to be rela-
tively simple. Many of the commonly available chemical disinfectants (chlorine, ozone, chlorine diox-
ide, and even chloramines) are successful in reducing coliform bacteria to acceptable levels, which was
generally accepted as an indication of safe water. However, it is now known that pathogens can exist
even in the presence of high levels  of free chlorine. The EPA has therefore evaluated the common
chemical disinfectants for their efficacy in inactivating Giardia cysts, viruses, and Cryptosporidium.

The EPA has adopted the CT concept (concentration in mg/L x time in minutes) in comparing the biocidal
effectiveness of disinfectants. Major considerations are the disinfectant concentration and the time needed
to attain inactivation of a certain microbial population exposed under specific conditions. The CT concept
can be expressed as an empirical equation as shown below (Chick 1908; Watson 1908):

                                         K=C"*t                              (14-1)

where
       C = disinfectant concentration in mg/L
       n = coefficient of dilution
       t = contact time in minutes required for a fixed percent of inactivation
       K = constant for a specific microorganism

CT values have been developed for inactivation of various microorganisms for the major disinfectants.
An example of these values is shown in Table 14-1 (Clark et al. 1994).

It is evident from Table 14-1 that  ozone  shows the highest disinfection efficiency, inactivating 99% of
most types of microorganisms at very low CT values.  Chloramine shows the lowest efficiency. For
these data, "«" has been shown to vary between 0.7-1.3; therefore, a value of n = 1 was chosen for the
referenced analysis  (Lykins et al. 1990). In Table 14-1, preformed chloramine was used because it is
conservative with respect to CT values.
Table 14-1. Summary of CT Value Ranges for Inactivation of Various Microorganisms by
            Disinfectants (mg/L-min) (Symons et al. 1981; Lykins et al. 1990; Hoff 1986;
            Lykins et al. 1986; Korich et al. 1990)
Microorganism
E. coli
Polio vims- 1
Rotavirus
Phage f2
G. lamblia cysts
G. muris cysts
Cryptosporidium parvum
Free Chlorine
pH 6 to 7
0.34-0.05
1.1-2.5
0.01-0.05
0.08-0.18
47->150
30-630
7200b
Preformed Chloramine
pH 8 to 9
95-180
768-3740
3806-6476
ND
2200"
1400
7200°
Chlorine Dioxide
pH 6 to 7
0.4-0.75
0.2-6.7
0.2-2.1
ND
26"
7.2-18.5
78C
Ozone
pH 6 to 7
0.02
0.1-0.2
0.006-0.06
ND
0.5-0.6
1.8-2.0
5-10b
  Note: All CT values are for 99% inactivation at 5°C except for Giardia lamblia and Cryptosporidium parvum.
  a Values for 99.9% inactivation at pH 6-9
  b 99% inactivation at pH 7 and 25°C
  c 90% inactivation at pH 7 and 25°C
  ND - No data
                                              14-2

-------
Other Treatment Goals

Oftentimes disinfection can also perform other treatment tasks in a drinking water treatment plant, for
example: oxidation of metals, taste and odor control, and enhancement of turbidity removal. Probably
the greatest need for an oxidant, other than disinfection, is for precipitation of metals such as iron and
manganese. These metals occur in the reduced inorganic state and in metal organic complexes, which
are the most difficult forms to remove.

Ozone and chlorine dioxide oxidize metals very well. Both are more effective at removing metals than
chlorine. However, chemical costs can be considerably higher with ozone and chlorine dioxide, par-
ticularly if there is a high oxidant demand.

Formation of DBFs

Shortly after THMs were identified in chlorinated drinking water, it was recognized that THMs were
only one of many halogenated DBFs produced by water chlorination. Compounds such as di-and trichlo-
roacetic acids, haloacetonitriles, haloketones, chloropicrin, cyanogen chloride, and chlorohydrate have
been identified in chlorinated drinking water. Several of these halogenated DBFs, such as dichloroacetic
acid, are suspected carcinogens and are believed to be more potent carcinogens than any of the THMs.
MX [3-chloro-4-(dichloromethyl)-5-hydroxy-2(5H)-furanone, a halogenated furanone which has been
identified in chlorinated drinking water, has been found to be extremely mutagenic. Some undesirable
DBFs may be produced with all disinfectants. Halogenated by-products are, of course, of special inter-
est (Singer 1994; Stevens et al. 1987; Bull 1993). Bench-scale studies of the chlorination of natural
water and humic acid-spiked water using extraction, capillary  column chromatography, and mass-spe-
cial analytical procedures detected more than 500 DBFs (Stevens et al. 1989). Many of these were
formed at microgram-per-liter concentrations, and the majority were not identified.

By-product formation at the bench scale and control at pilot and full scale have been evaluated by
examining specific by-products and  surrogate parameters such as total organic halides (TOX). The
TOX data suggest the formation of other DBFs whose total concentrations are likely to equal or exceed
those  of TTHMs (Stevens  et al. 1989). One bench-scale study showed the  concentrations of THMs
increased with time for each pH value. Pilot  studies have shown that the percentages of removal of
DBF formation potential from raw Ohio River water (ORW) to low pH, alum-coagulated, and filtered
effluent were within a range of 60 to 80 percent. Bench-scale studies have  also shown that bromide
heavily influences the nature of the chlorination DBFs formed (Pourmoghaddas et al. 1993; Minear and
Bird 1980; Cooper et al. 1983).

At a pilot plant located at the Jefferson Parish, LA, water utility, four major disinfectants (chlorine,
chlorine dioxide, ozone, chloramine) were applied in parallel to clarified and filtered lower Mississippi
River water during two studies (Lykins et  al. 1986). Chlorine produced the highest concentration of
TOX, indicating that several other halogenated by-products  were formed with chlorination. Ozone
produced the lowest concentration of TOX, with concentrations below the nondisinfected feed to the
pilot plant, suggesting that some TOX destruction occurred as  shown in Figure 14-1. Average instanta-
neous TOX concentrations were 25 |ig/L,  15 |ig/L, 85  |ig/L,  117 |ig/L, and 263 |ig/L for the
nondisinfected, ozone, chlorine dioxide, chloramine, and chlorine streams, respectively. Not all of the
organics produced by the disinfectants evaluated at Jefferson Parish were identified. Flame-ionization
detection (FID) and electron-capture  detection (BCD) gas chromatographic profiles gave evidence of
the extent of by-product formation for the disinfectants. The number of different products formed and
concentration of products formed by  the various disinfection  process streams followed the sequence:
chlorine > chloramine > chlorine dioxide > ozone for these  measures.
                                              14-3

-------
         400 -r
                       75
150       225
  Day of Run
300
375
Figure 14-1. TOX concentrations after 30-minute contact time (Jefferson Parish, LA,
            pilot plant).

Although most of the research conducted by the EPA has focused on halogenated by-products, ozone
and other oxidants, to a somewhat lesser degree, can also form by-products such as formaldehyde,
glyoxal, and acetone. In the EPAs pilot plant in Cincinnati, using ORW, ozonation produced formalde-
hyde concentrations of approximately 26  |ig/L (Miltner 1992). Concentrations of formaldehyde, ac-
etone, and glyoxal subsequently declined through conventional treatment, but increased again due to
either clear well chlorination or chloramination. The loss of formaldehyde, acetone, and glyoxal during
conventional treatment is assumed to result from biodegradation by heterotrophic plate count (HPC)
species, as evidenced by a decline in formaldehyde concentration  and an increase in HPC densities.

The occurrence of DBFs in U.S. drinking water was evaluated at 35 water treatment facilities that had
a broad range of source water qualities and treatment processes (Krasner et al. 1989). THMs were the
largest class of DBFs detected on a weight basis. Haloacetic acids were the next largest class of com-
pounds found. Formaldehyde and acetaldehyde, by-products of ozonation, were also produced by chlo-
rination. Cyanogen chloride was preferentially produced in chloraminated water. The median, TTHM
quarterly values in the study (Krasner 1989) were comparable to those seen in a survey of 727 utilities
by the American Water Works Association Research Foundation (McGuire and Meadow  1988).

Treatment Strategies for Controlling DBFs

DBFs are the result of the interaction of the disinfectant with natural organic matter in water as shown
by Equation 14-2.
                          Precursor Material + Disinfectant —» DBFs
                                              (14-2)
This concept indicates that moving the point of disinfection to the end of the treatment process follow-
ing precursor reduction will minimize by-product formation. Numerous experiences have verified the
effectiveness of this approach (Symons et al. 1981). Other options include the use of a disinfectant that
minimizes the formation of by-products, and removal of DBFs once they are formed (Lykins et al.
                                             14-4

-------
1990). Stage 1 of the D/DBP Rule will require drinking water utilities to achieve a maximum contami-
nant level (MCL) of 0.08 mg/L, and Stage 2 may require an MCL as low as 0.04 mg/L. The Interim
Enhanced Surface Water Treatment Rule (IESWTR) and the Long Term Enhanced Surface Water Treat-
ment Rule (LTESWTR) will require total organic carbon (TOC) targets as a measure of precursor
removal.

In the following section, the effects, technology, and costs associated with these options for controlling
DBFs will be discussed. Since the impact of THM control is similar to the control of other halogenated
by-products, the costs developed for THM control would be similar to those for removal of other
DBFs.

Treatment Alternatives for Controlling DBFs

Two alternatives for controlling DBFs are to switch to a disinfectant that minimizes by-product forma-
tion or to remove by-product precursors prior to disinfection (Lykins et al. 1990). The cost and perfor-
mance of various options for controlling DBFs as mentioned previously will be explored in this section
(Clark 1998). Table 14-2 contains the common cost assumptions that apply to all of the evaluations.
Table 14-2. Common Cost Assumptions
Items
Capital amortization
Engineering fees
Contractor overhead and profit
ENR construction cost index
Producers price index
Labor and fringe rate
Electric power rate
Fuel oil prices
Value
10% over 20 years
15% of construction
12% of construction
6130 (January 2000) (1913 = 100)
134.7 (January 2000) (1982 = 100)
$15/hour
$0.024 x 10-6 Joules ($0.086/kwh)
$0.235/liter ($0.889/gallon)
Alternate Disinfectants

As mentioned previously, the use of disinfectants other than chlorine is one option for controlling the
concentration of halogenated by-products. In a study at Jefferson Parish, LA, lower Mississippi River
water was clarified and filtered before being diverted to five parallel streams. The major objectives of
this study were to evaluate (1) the control of halogenated by-products, (2) the microbiological effec-
tiveness of the disinfectants, and (3) potential health effects associated with the use of these disinfec-
tants. Many of the halogenated by-products of interest were analyzed for each disinfectant stream (Lykins
et al. 1990). Of special interest was the potential concentration of these by-products when water was
delivered to the customer. This was evaluated by storing samples  for a specified time with a disinfec-
tant residual to simulate residence time in the distribution system. It was found that, when chlorine was
added prior to and after sand filtration, an average of 45 |ig/L of  dichloroacetic acid was  detected. If
ozone was added prior to sand filtration  and chlorine after sand filtration, the average concentration
was reduced to 32 |ig/L. If monochloramine was used prior to and after sand filtration, then  the average
concentration was reduced to about 8 |ig/L. Further reductions were seen for the combination of ozone
added before sand filtration and monochloramine after sand filtration (4.6 |ig/L average). This trend
was seen for other prevalent halogenated by-products such as trichloroacetic acid, bromochloracetic
acid, chloral hydrate, and trichloromethane.
                                              14-5

-------
Another potential disinfectant that minimizes halogenated by-products is chlorine dioxide. At a drinking
water utility on the Ohio River, a pilot plant was used to compare chlorine dioxide disinfection with
chlorine disinfection (Lykins et al. 1986). The addition of chlorine dioxide to the raw water with delayed
chlorination permitted coagulation/settling/filtration and oxidation to remove THM precursors, thereby
reducing the amount of THMs formed during post-treatment chlorination. A comparison of average THM
concentrations for the two disinfectant modes showed a reduction of approximately 60 percent when
chlorine dioxide was used. Although chlorine dioxide disinfection can reduce THM concentrations, con-
trol of the metabolites (chlorite and chlorate) is essential before chlorine dioxide can be considered a
viable disinfection alternative. Equipment is now available to produce chlorine dioxide that is virtually
free of chlorite and chlorate. Studies were conducted to minimize chlorite and chlorate concentrations by
a reducing agent (Lykins et al.  1990)

Disinfection Costs

For the  purpose of this analysis, the following assumptions were made: Pre-chlorine dose-3 mg/L,
post-chlorine dose-1.5 mg/L, pre-ozone dose-2 mg/L, post-chloramine dose (2 mg/L C12 and 0.5 mg/L
ammonia), and pre-chlorine dioxide-1 mg/L. The various disinfectants including capital, O&M, and
chemical cost  for disinfection only are  summarized in Table 14-3.  In column 1 of Table 14-3, the
primary disinfectant is listed first.
Table 14-3. Alternative Disinfection Costs In eVIOOO Gallons (eVcu m) (Jan. 2000)
Flow in MGD (thou cu m per day" )
Unit Process
Chlorine/Chlorine
Chlorine/Chloramine
Ozone/Chlorine
Ozone/Chloramine
Chlorine dioxide/Chlorine
0.1 (0.3785)
42.1 (11.1)
61.5 (16.3)
119.5 (31.6)
138.9 (36.7)
139.2 (36.8)
1 (3.785)
14.4 (3.80)
16.3 (4.32)
22.2 (5.87)
24.2 (6.38)
19.8 (5.24)
10 (37.85)
2.4 (0.634)
2.8 (0.746)
7.36 (1.95)
7.8 (2.06)
2.9 (0.778)
100 (378.5)
1.2 (0.325)
1.7 (0.457)
4.2(1.10)
4.5 (1.18)
1.4 (0.368)
   To convert from thou cu m per day to mgd, divide by 3.785.
Precursor Removal by Enhanced Coagulation

Three methods for removing precursor material will be discussed in this section: appropriate coagula-
tion, granular activated carbon, and nanofiltration. The performance and cost of each of these technolo-
gies will be discussed in that section. Performance is based on Total Trihalomethane Formation Poten-
tial (TTHMFP). Some removal is observed when alum was used as a coagulant for turbidity control,
but when the treatment was modified by adding additional alum and reducing the pH, further precursor
removal was noted (Table 14-4).

Cost of Enhanced Coagulation
The EPA has prepared several estimates for treatment optimization that might lead to enhanced
coagulation  (USEPA 1997). Some of these costs  are provided in Table 14-5.

Precursor Removal by GAC

Granular activated carbon (GAC) is an effective  means of removing DBF precursors from water at a
cost that varies widely according to water quality  and treatment goals. An example of the effectiveness
                                             14-6

-------
Table 14-4. Precursor Control by Coagulation (EPA Pilot Plant)
           Percent Removal of Precursor3
TOC
A
21
B
46
THMFP
A
28
B
48
TOXFP
A
4
B
56
TOC
C
15
D
40
DBPFP"
C
56
D
70
  1 Ohio River water
  b Based on mean formation potential for several halogenated by-products.
  A - alum = 20 mg/L, pH = 7.5; for turbidity control
  B - alum = 89 mg/L, pH = 6.0; for precursor control
  C - alum = 26 mg/L, pH = 7.0; for turbidity control
  D - alum = 40 mg/L, pH = 5.7; for precursor control
  TOC = Total organic carbon
  THMFP = Total trihalomethane formation potential
  TOXFP = Total organic halide formation potential
  DBPFP = Disinfection by-product formation potential
Table 14-5. Annual Cost for Enhanced Coagulation in ^/Thou cu ma (eVIOOO gal) (USEPA 1997)
Design Flow in Thou cu m Per Day" (average flow)
Unit Process
Modification
Chemical
addition
Coagulant
improvements
Rapid mix
Flocculation
improvements
Settling
improvements
Filtration
improvements
Hydraulic
improvements
18.168
(7.949)
0.713-3.406
(2.7-12.9)
2.059
(7.8)
0.733-0.898
(2.7-3.4)
0.845-1.690
(3.2-6.4)
0.290-0.581
(1.1-2.2)
0.079-5.148
(0.3-19.5)
0.317-0.739
(1.2-2.8)
41.635
(18.925)
0.290-3.247
(1.1-12.3)
1.320
(5.0)
1.716-2.086
(6.5-7.9)
0.607-1.346
(2.3-5.1)
0.185-0.449
(0.7-1.7)
0.079-2.666
(0.3-10.1)
0.211-0.581
(0.8-2.2)
68.130
(23.308)
0.158-3.168
(0.6-12.0)
1.056
(4.0)
1.795-2.138
(6.8-8.1)
0.554-1.188
(2.1^.5)
0.132-0.396
(0.5-1.5)
0.079-1.795
(0.3-6.8)
0.185-0.502
(0.7-1.9)
98.410
(49.205)
0.106-3.115
(0.4-11.8)
0.950
(3.6)
1.742-2.112
(6.6-8.0)
0.528-1.082
(2.0^.1)
0.132-0.370
(0.5-1.4)
0.079-1.637
(0.3-6.2)
0.185-0.475
(0.7-1.8)
193.035
(102.195)
0.079-3.036
(0.3-11.5)
0.818
(3.1)
1.002-1.320
(4.1-5.0)
0.317-0.660
(1.2-2.5)
0.132-0.343
(0.5-1.3)
0.079-1.135
(0.3^1.3)
0.132-0.396
(0.5-1.5)
794.850
(454.200)
0.026-2.930
(0.1-11.1)
0.713
(2.7)
0.317-0.370
(1.2-1.4)
0.079-0.238
(0.3-1.0)
0.079-0.185
(0.3-0.7)
0.053-0.766
(0.2-2.9)
0.079-0.317
(0.3-1.2)
1,627.580
(1,021.950)
0.008-2.904
(0.03-11.0)
0.686
(2.6)
0.158-0.185
(0.6-0.7)
0.079-0.264
(0.3-0.9)
0.053-0.106
(0.2-0.4)
0.026-0.660
(0.1-2.5)
0.053-0.238
(0.2-0.9)
  1 To convert from thou cu m per day to mgd, divide by 3.785.
Table 14-6. Precursor Removal by GAC (Clark et al. 1994)
% Removal
Utility
Cincinnati, OH
Manchester, NH
Jefferson Parish, LA
Influent THMFP (jig/L)
160
72
93
Initial
98
85
83
After 90 Days
63
35
40
of GAC for three water utilities is shown in Table 14-6. Removal is initially good, but diminishes as the
time in service increases (Clark et al. 1994).

In some cases, the use of GAC for precursor removal would be unreasonable. For example, based on
field tests in Miami, FL, in order to remove TTHMFP in the 15-100-|ig/L range, it was found that the
carbon would require reactivation every 20 days (Symons et al. 1981).
                                               14-7

-------
Cost of Precursor Removal Using GAC

As mentioned, THM formation is a function of the influent concentration and reactive characteristics
of the natural organic matter in the source water. The following cost analysis is based on field scale data
collected from studies in Cincinnati, OH, Jefferson Parish, LA, and Manchester, NH (Clark et al. 1994).
Influent TOC during the studies ranged from 1.5 to 3.5 mg/L, and TOC effluents ranged from 0.6 to 1.6
mg/L. These effluent levels correspond to THMFP effluent values of approximately 50 |ig/L. Because
GAC  bed life is dependent on influent concentration (C0),  the bed lives were adjusted to reflect the
target effluent concentrations (Ce) to be considered. Table 14-7 contains the assumed bed lives for each
of the utilities studied.

Table 14-7. Bed Life for GAC for Removal of THMFP at an Influent Level of 150 ug/L
           (Clark et al. 1994)

Cincinnati
Cincinnati
Jefferson Parish
Jefferson Parish
Manchester
Target Effluent THMFP (jig/L)
100
50
100
50
50
Bed Life (days)
225
175
103
63
80
The cost calculations are based on the data shown in Table 14-7 and an assumption of a 20-minute
empty bed contact time. These target values were used because they were taken from actual studies and
are in the range associated with the anticipated regulation. For systems of 0.3785, 3.785, and 37.85
thou cu. m/day (0.1, 1, 10 mgd), pressure contactors were assumed and for systems of 95,  190, and
3785 thou cu.  m/day (25, 50, 100 mgd), concrete gravity contactors were assumed. For systems of
0.3785 and 3.785 thou cu. m/day (0.1, 10 mgd), replacement of spent carbon with virgin carbon was
used in the calculation, and for 37.8, 95, 190, and 378.5 thou cu. m/day (10, 25,  50, 100 mgd) systems,
on-site multihearth reactivation was assumed. Table 14-2 contains the cost assumptions used in this
analysis. Additional assumptions included a virgin carbon cost of $2.38 kg (1.08/lb) for 45.4 kg (100,000
Ib) and a carbon loss rate of 15% due to handling. Cost values are indexed to January 2000.

Table 14-8 summarizes the costs for using GAC for removing TFJJVI precursor removal based on THMFP
to TTFDVIFP levels of 100 |ig/L, respectively.

Table 14-8. Annual Cost for TTHMFP Removal by GAC in j^/cu m (j^/lOOO gal)
Design Flow in Thou cu
Target Levels
Ce <100 ug/L
Ce <50 ug/L
0.3785 (0.189)
73.3 (277)
to 85.7 (324)
75.8 (287)
to 95.6 (362)
3.785 (1.89)
27.5 (104)
to 38.7 (147)
29.7 (112)
to 47.8 (181)
37.85 (26.495)
16.2 (61)
to 20.6 (78)
17.3 (66)
to 23.9 (90)
m Per Day" (average flow)
94.62 (64.392)
10.7 (40)
to 14.3 (54)
11.5 (44)
to 17.0 (64)
189.25 (132.475)
8.8 (33)
to 11.8(45)
9.3 (35)
to 14.3 (54)
378.5 (264.95)
7.4 (28)
to 11 (42)
7.9 (30)
to 14.3 (54)
  1 To convert from thou cu m per day to mgd, divide by 3.785.

Precursor Removal by Nanoflltration

Membrane processes are  also promising for removing DBF precursors. Studies in Florida waters have
demonstrated that, for ground waters, membranes are an excellent treatment alternative. These studies
showed that, with an average raw water concentration of 455  ng/L, trihalomethane formation potential
(THMFP) average concentrations of 20 (ig/L were being produced (95% rejection) in the product water.
                                             14-8

-------
For total organic halide formation potential (TOXFP), an average raw water concentration of 977 |ig/L
was reduced to 34 jig/L (96% rejection). Additional studies of specific chlorination by-products at
Daytona Beach, FL showed that, with a 4-2-1 pressure vessel array, total DBFs expressed as chloride
equivalents show a raw water concentration of 530 to 715 |ig/L, with an overall system reduction of
95% to 98% (Taylor et al. 1989).

When membranes are used on surface water, however, extensive pretreatment is usually required. Al-
though rejections are good, the membranes will likely require frequent cleaning. When a Florida sur-
face water (tributary of the Peace River) was treated, pretreatment consisted of (1) alum-coagulated
and settled water from the full-scale plant, and (2) pressure sand filtration in the pilot plant, membrane
filtration. Under these operating conditions, membrane cleaning was required about every 16 days to
avoid a production loss greater than 10 percent. Raw water TFDVI formation potential averaged 612
|ig/L, and a product water of 37 |ig/L was produced (94% rejection). For an average raw water TOX of
1,965 |ig/L, the product water was 53 |ig/L (97% rejection).

Cost of Nanofiltration

In this analysis, it is assumed that an 8" x 40" element removes organics with molecular weights greater
than 300 molecular weight units (Taylor et al. 1989). These organics are separated from product water
primarily by sieving (little is removed by diffusion). Based on field experience, greater than 95% of
dissolved organic carbon (DOC), TOX, and DBF  are removed. If the concentration in the influent is
greater than 150 |ig/L of THMFP, then 25 |ig/L of formation potential is expected in the effluent perme-
ate. For example, the reference design for a 37.85 thou cu m/day  (10 mgd) nanofiltration system is
assumed as raw feed water of 44-287 thou cu m/day (11-76 mgd) to yield a permeate flow of 37.8 thou
cu m/day (10 mgd) or 85% recovery. A 3-stage membrane configuration was assumed. Two types of
systems were considered: ground water with an average pressure of 732 kg/m2 (150 psi) and an average
flux of 0.6 m/day (15 gal/ft2/day),  with 13  membrane skids (12 on line and one on  standby).  The
reference design was estimated using an approach described by  Suratt (1991) and Clark et al. (1998).

The ground water system was assumed to require no advanced treatment, only a 5-|im cartridge, prefilter,
and H2SO4 addition for scale control. Two types of disposal were assumed: disposal to a surface pond
or stream and deep well concentrate disposal. Surface water treatment requires advanced pretreatment
to reduce fouling (alum coagulation, solids contact, rapid sand filtration). Deep well disposal of con-
centrate was assumed. Table 14-9 summarizes the costs associated with the use of nanofiltration sys-
tems at various treatment capacities based on estimates using the reference design and a scaling ap-
proach described by Eisenburg and Middlebrooks (1986).
Table 14-9. Nanofiltration Cost Summary in j^/cu ma (j^/lOOO gal)
Ground Water System
Permeate Capacity
in Thou cu ma
0.3785
3.785
8.925
37.85
94.625
189.25
378.50
Surface Concentrate
Disposal
63.2 (239)
41.2(156)
32.4 (123)
27.7 (105)
28.6 (108)
26.1(99)
23.6 (89)
Deep Well
Concentrate Disposal
74.5 (281)
48.1 (182)
37.6 (142)
34.9 (132)
33.2 (126)
30.2 (114)
27.4 (104)
Surface Water System
Alum Coagulation Pretreatment with
Deep Well Concentrate Disposal
142 (537)
93.4 (354)
73.6 (279)
68.1 (258)
64.8 (245)
59.4 (224)
53.8 (203)
  1 To convert from thou cu m to mgd, divide by 3.785.
                                              14-9

-------
Comparative Analysis

Making direct comparisons among the various alternatives is difficult. For example, moving the point
of disinfection (chlorination) would seem to be the lowest cost option. Nanofiltration, although most
expensive for precursor removal, has the advantage of removing other contaminants such as total dis-
solved solids and various inorganics. Therefore, it might be used for achieving other treatment goals in
addition to removing DBF precursors. For example, nanofiltration effectively removed microorgan-
isms, thus serving as an alternative for chemical disinfection. Although enhanced coagulation was not
evaluated for cost, it could be very effective if a utility is only slightly out of compliance; however, in
addition to increased coagulation costs, an  additional cost may be associated with sludge handling.
Clearly, changing the type of disinfection was the lowest cost option for controlling DBFs. However, as
noted, there are by-products and  problems  associated  with the use of some of the alternatives. For
example, chloramination is not as good a disinfectant as chlorine, while ozone may enhance regrowth
of some organisms. Retrofitting may be fairly easy with chloraminaton. For example, to switch from
chlorine to chloramine may only require the addition of ammonia feed equipment. However, the use of
ozone will incur cost for the construction and operation of ozone contactors. The use of chlorine diox-
ide will  probably require the use of a reducing agent such as ferrous chloride, which was not included
in this costing analysis.

Incremental Costs

The technologies discussed would normally be implemented incrementally to a utility's existing treat-
ment. Table 14-10 summarizes the base cost associated with an assumed conventional treatment sys-
tem and the incremental costs associated with various DBF control alternatives. The unit processes
considered are those that are effective for precursor removal or for the use of alternate disinfectants.

Summary and  Conclusions

Regulations to control contaminants  in drinking water in the U.S. are expected to become more and
more stringent. Forthcoming DBF regulations will effect virtually every community water system in
Table 14-10. Incremental Cost for Disinfection By-Product Control in eVcu ma (eVlOOO gal)
Design Flow in Thou
Item
Conventional treatment
Conventional treatment
and nanofiltration
Conventional treatment
plus GAC(Ce= 100 mg/L)
Conventional treatment
plus GAC (Ce = 50 mg/L)
Conventional treatment plus
-Chlorine/Chloramine
-Ozone/Chlorine
-Ozone/Chloramine
-Chlorine dioxide/Chlorine
-Chlorine dioxide/Chloramine
0.3785 (0.189)
142 (539)
205 (778)
203 (772)
to 216 (818)
206 (781)
to 226 (856)
147 (556)
163 (619)
168 (636)
167 (633)
172 (651)
3.785 (1.89)
47.5 (180)
88.7 (336)
70.8 (268)
to 82. 1(3 11)
73.1(276)
to 91.2 (345)
47.8(181)
49.7 (188)
50 (189)
48.9 (185)
49.1(186)
cu ma (average flow)
37.85 (26.495)
12.6 (48)
40.4 (152)
28.3 (107)
to 32.7 (124)
29.4(111)
to 36.0 (136)
12.6 (48)
14 (53)
14 (53)
12.6 (48)
12.9 (49)
378.5 (264.95)
8.51(32)
32.1 (122)
15.7 (59)
to 19.2 (73)
16.2 (61)
to 22.5 (85)
8.5 (32)
9.3 (35)
9.3 (35)
8.5 (32)
8.5 (32)
  1 To convert from thou cu m to mgd, divide by 3.785.
                                            14-10

-------
the U.S. There are various ways to control DBFs, one of which is to use an alternative to chlorine to
control halogenated by-products. However, when this is done, one has to consider the consequences.
Water treatment managers will have to become more knowledgeable about various treatment options
that are cost-effective in order for them to meet present and anticipated regulations.

References

Bull, R. J. (1993). "Toxicology of disinfectants and disinfection by-products." Safety of Chemical of
   MicrobialRisk, G. F. Craun,  ed., ILSI Press, International Life Sciences Institute, Washington,
   D.C., 239-256.

Chick, H. (1908). "An investigation of the laws of disinfection." Journal of Hygiene, 8, 92-158.

Clark, R. M. (1998). "Cost considerations." Treatment process selection for particle removal, J. B.
   McEwen, ed., American Water Works Association Research Foundation/International Water
   Supply Association, 265-302.

Clark, R. M., Adams, J. Q., and Lykins, B. W. (1994). "DBF control in drinking water: Cost and
   performance." Journal of Environmental Engineering Division oftheASCE,  120(4), 759-782.

Clark, R. M., Adams, J. Q., Sethi, V, and Siveganesan, G. (1998). "Control of microbial contami-
   nants and disinfection by-products for drinking water in the U.S.:Cost and performance." Journal
   of Water Supply Research and Technology-Aqua, 46(6), 255-265.

Cooper, W. J., Meyer, L. M., Bufill, C. C., and Cordal, E. (1983). "Quantitative effects of bromine on
   the formation and distribution of trihalomethane in groundwater with a high organic content."
   Water chlorination: Environmental impact and health effects, R. L. Jolley, W. A. Bross, J. A.
   Cotruvo, R. B. Cummings, J. S. Mettice, and V. A. Jacobs, eds., Vol. 4, Ann Arbor Science
   Publishers, Inc., Ann Arbor, MI, 285-296.

Eisenburg, T. and Middlebrooks, E. (1986). Reverse osmosis treatment of drinking water. Ann Arbor
   Science, Stoneham, MA, 221.

Hoff, J. C. (1986). "Inactivation  of microbial agents by chemical disinfectants." EPA/600/286/067,
   U.S. Environmental Protection Agency.

Korich, D. G., Mead, J. R., Madore, M. S., Sinclair, N. A., and Sterling, C. R. (1990). "Effects of
   ozone, chlorine dioxide, chlorine, and monochloramine on Cryptosporidium parvum oocyst
   viabililty." Applied and Environmental Microbiology, 56, 1423-1428.

Krasner,  S. W., McGuire, M. J., Jacangelo, J. G., Patania, N. L., Reagan, K. M., and Aieta, E.M.
   (1989). "The occurrence of disinfection by-products in U.S. drinking water." Journal of the
   American Water Works Association, 81(8), 41-53.

Lykins, Jr., B. W., Clark, R. M., and Westrick, J. J. (1990). "Treatment technologies for meeting U.S.
   drinking water regulations." Proceedings,  1990 Joint Conference, Ontario Section American
   Water Works Association/Ontario Municipal Water Association, Toronto, Ontario, Canada, May
   6-9.

Lykins, Jr., B. W., Goodrich, J. A., and Hoff, J.C. (1990). "Concerns with using chlorine-dioxide
   disinfection in the USA." Journal of Water Supply Research and Technology-Aqua, 39(6), 376-
   386.

Lykins, Jr., B. W., Koffskey, W. E., and Miller, R. G. (1986). "Chemical products and toxicologic
   effects of disinfection." Journal of the American Water Works Association, 78(11), 66-75.
                                             14-11

-------
McGuire, M. J. and Meadow, R. G. (1988). "AWWARF trihalomethane survey." Journal of the
   American Water Works Association, 80(1), 61-68.

Miltner, R. J. (1992). "Pilot scale treatment for control of disinfection by-products." Strategies and
   Technologies for Meeting SDWA Requirements, R. Clark and S. Summers, eds., Technomics,
   Inc., Lancaster, PA, 203-236.

Minear, R. A. and Bird, J. C. (1980). "Trihalomethane:  Impact of bromide ion concentration on
   yield, species distribution, rate of formation and influence of other variables." Water chlorina-
   tion: Environmental impact and health effects,  R. L. Jolley, W. A. Brungs, and R. B. Cummings,
   eds., Vol. 3, Ann Arbor Science Publisher, Inc., Ann Arbor, MI, 151-160.

Pourmoghaddas, H., Stevens, A. A., Kinman, R. N., Dressman, R. C, Moore, L. A., and Ireland, J.A.
   (1993). "Effect of bromide ion on formation of HAAs during chlorination." Journal of the
   American Water Works Association, 85(1), 82-87.

Singer, P. C. (1994). "Control of disinfection by-products in drinking water." Journal of Environmen-
   tal Engineer ing Division of the ASCE, 120(4), 727-744.

Stevens, A. A., Moore, L. A., and Miltner, R. J. (1989). "Formation and control of non-
   trihalomethane disinfection by-products." Journal of the American Water Works Association,
   81(8), 54-60.

Stevens, A. A., et al. (1987). "Detection and control of chlorination by-products in drinking water."
   Proceedings, Conference on Current Research in Drinking Water Treatment, Cincinnati, OH,
   March 24-26.

Suratt, W. (1991).  "Estimating the costs of membrane water treatment plants." Proceedings, Ameri-
   can Water Works Association Membrane Processes Conference, Orlando, FL, March 10-13.

Symons, J. M., Stevens, A. A., Clark, R. M., Geldreich, E. E., Love, Jr., O. T., and DeMarco, J.
   (1981). Treatment techniques for controlling trihalomethanes in drinking water. EPA 600/2-81-
   156, U.S. Environmental Protection Agency, Drinking Water Research Division, Municipal
   Environmental Research Laboratory, Office of Research and Development, Cincinnati, OH.

Taylor, J. S., Mulford, L. A., Barrett, W. M., Duranceau, S. J., and Smith, D. K. (1989). "Cost and
   performance of membranes for organic control in small systems." EPA 600/2-89/022, U.S.
   Environmental Protection Agency, May.

U.S. Environmental Protection Agency (USEPA). (1997). Technology costs for the interim enhanced
   surface water treatment rule. Office of Ground Water and Drinking Water, Washington, D.C.

Watson, H. E. (1908). "A note on the variation of the rate of disinfection with change in the concen-
   tration at the disinfectant." Journal of Hygiene, 8, 536-542.
                                            14-12

-------
                                      GLOSSARY

assay A test for a particular chemical biological agent to determine its properties or effect.

Ames test A test developed by Bruce Ames that is based on the assumption that any substance that is
mutagenic (for the Salmonella typhimurium bacteria) may also turn out to be a carcinogen,  that is,
cause cancer. However, not all known chemicals that cause cancer in animal labs give a positive Ames
test (and vice versa). The ease and low cost of the test make it invaluable for screening substances in
our environment for possible carcinogenicity.

Best Available Technology (BAT) The best technology treatment techniques or other means that are
available to remove a contaminant(s) to below the set MCL. BATs are designated by the EPAs Admin-
istrator after examination for efficacy under field conditions and not solely under laboratory conditions
(taking cost into consideration).

Best Management Practice (BMP)  Structural, nonstructural, and managerial techniques that are rec-
ognized to be the most effective and practical means to control non-point source pollution and are
compatible with the productive use of the resource to which they are applied. BMPs are used  in both
urban and agricultural areas.

biofilm A structured community of microorganisms (including protozoa) enclosed in a self-produced
polymeric matrix and adherent to an inert or living surface.

carcinogen Any substance that can produce cancer in an organism.

chemical oxygen demand (COD) An indirect measure of the amount of oxygen used by inorganic and
organic matter in water.

chloramines  Compounds formed by the reaction of aqueous chlorine (or hypochlorous acid) with
ammonia.

chlorine-contact chamber  That part of a water treatment plant where chlorine is applied to water for
disinfection purposes.

chlorine demand  The amount of chlorine  that must be applied to water before free chlorine can be
detected. Chlorine demand is the difference between the amount of chlorine added to water and the
amount of residual chlorine remaining after a given contact time. Chlorine demand varies with dosage,
time, temperature, pH, and nature and amount of the impurities in the water.

chronic Occurring over a long period of time, either  continuously  or intermittently; used to describe
ongoing exposures and effects that develop only after a long exposure.

chronic exposure  Long-term, low-level exposure to a toxic chemical.

clarifier A large circular or rectangular tank or basin in which water is held for a period of time, during
which the heavier suspended solids settle to the bottom. Clarifiers  are also called settling basins and
sedimentation basins.

clear well A reservoir for the storage of filtered water of sufficient capacity to prevent the need to vary
the filtration rate with variations in demand.

coagulants Chemicals that cause very fine particles to clump together into larger particles. These
chemicals help  (by changing the particles'  surface charge and  destabilizing them) in separating the
solids from the water by settling, skimming, draining, or filtering.
                                             G-l

-------
coagulation The process of clumping together of colloids and fine particles into larger particles caused
by the use of chemicals (coagulants). This clumping together makes it easier to separate the solids from
the water by settling, skimming, draining, or filtering.

coliform A group of bacteria found in the intestines of warm-blooded animals (including humans) and
also found in  plants, soil, air, and water. Coliforms are gas-producing bacteria. Fecal coliforms are a
specific class  of bacteria which only inhibit the intestines of warm-blooded animals. The presence of
coliform is  an indication that the water is polluted and may contain pathogenic organisms.

colloids  Very small, finely divided solids (particles that do not dissolve) that remain dispersed in a
liquid for a long time due to their large surface area-to-volume ratio and electrical charge. When most
of the particles in water have a negative electrical charge, they tend to repel each other. This repulsion
prevents the particles from clumping together, becoming heavier, and settling out.

combined available residual chlorine The concentration of residual chlorine which is combined with
ammonia (NH3) and/or organic nitrogen in water, such as chloramine (or another chloro-derivative),
yet is still available to  oxidize organic matter and has bactericidal properties.

combined residual chlorination The application of chlorine to water to produce combined available
residual chlorine.  This residual can be  made up of monochloramines, dichloramines,  and nitrogen
trichloride.

combined sewer A sewer that transports surface runoff, human domestic wastes (sewage), and some-
times industrial wastes. Wastewater and runoff in a combined sewer may occur in excess of the sewer
capacity and cannot be treated immediately. The excess is frequently discharged directly to a receiving
stream without treatment or to a holding basin for subsequent treatment and disposal.

composite (proportional) samples A composite sample is a collection of individual samples obtained
at regular intervals, usually every one or two hours during a 24-hour time span. Each individual sample
is combined with the  others in proportion to the rate of flow when  the sample was collected The
resulting mixture (composite sample) forms a representative sample and is analyzed to determine the
average conditions during the sampling period.

compound A substance composed of two or more elements whose composition is constant. For ex-
ample, table salt (sodium chloride-NACl) is a compound.

contaminant Any physical, chemical, biological, or radiological substance or matter that has an ad-
verse effect on air, water, or soil.

contamination The introduction into water of microorganisms,  chemicals, toxic substances, wastes,
or wastewater in a concentration that makes the water unfit for its next intended use.

continuous sample  A flow of water from a particular place in a plant to the location where samples are
collected for testing. This continuous stream may be used to obtain grab  or composite samples. Fre-
quently, several taps (faucets) will flow continuously in the laboratory to provide test samples from
various places in a water treatment plant.

conventional filtration A method of treating water to remove substantial amounts of particulates,
biological and chemical contaminants. The method consists of a series of processes including the addi-
tion of coagulant chemicals, flash mixing, coagulation, flocculation, sedimentation,  and filtration.

corrosion The gradual decomposition or destruction of a material by chemical action, often due to an
electrochemical reaction. Corrosion may be caused by 1) stray current electrolysis, 2) galvanic corro-
                                             G-2

-------
 sion caused by dissimilar metals, or 3) differential concentration cells. Corrosion starts at the surface of
 a material and moves inward.

 corrosive  A chemical (water or otherwise) that reacts with the surface of a material (pipe), causing it
 to deteriorate, decompose, or wear away.

 corrosivity An indication of the corrosiveness of water (or other chemical). The corrosiveness of a
 water is described by the water's pH, alkalinity, hardness, temperature, total dissolved solids, dissolved
 oxygen concentration, and the Langelier Index.

 CT or CTcalc  The product of "residual disinfectant concentration" (C) in mg/L determined before or
 at the first customer, and the corresponding "disinfectant contact time" (T) in minutes, i.e., "C" x "T".

 curie A measure of radioactivity. One curie of radioactivity is equivalent to 3.7 x 1010 or 37,000,000,000
 nuclear disintegrations per second.

 decomposition  The conversion of materials to more stable forms by chemical or biological action.

 diatomaceous earth A fine, siliceous (made of silica) "earth" composed mainly of the skeletal re-
 mains of diatoms, a type of free-floating, microscopic plant found in the ocean.

 diatomaceous earth filtration A filtration method resulting in substantial particulate removal that
 uses a process in which 1) a "precoat" cake of diatomaceous earth filter media is deposited on a support
 membrane (septum), and 2) while the  water is filtered by passing through the cake on the septum,
 additional filter media, known as "body feed," are continuously added to the feed water to maintain the
 permeability of the filter cake.

 direct filtration A filtration method of treating water which consists of the addition of coagulant
 chemicals, flash mixing, coagulation, minimal flocculation, and filtration. The flocculation facilities
 may be omitted, but the physical-chemical reactions will occur to some extent. Compared to conven-
 tional treatment, the sedimentation process is omitted from the treatment train. Also see conventional
filtration and in-line filtration.

 disinfectant Any oxidant, including but not limited to chlorine, chlorine dioxide, chloramines, and
 ozone, that is added to water in  any part of the treatment or distribution process and is intended to kill
 or inactivate pathogenic microorganisms.

 disinfectant contact time  The time in  minutes that it takes for water to move from the point of
 disinfectant application or the previous  point of disinfectant residual measurement to a point before or
 at the point where residual disinfectant  concentration (C) is measured. When only one C is measured,
 T is the time in minutes that it takes for water to move from the point of disinfectant application to a
 point before or at where residual disinfectant concentration (C) is measured. Disinfectant contact time
 in pipelines must be calculated  based on plug flow by dividing the internal volume of the pipe by the
 maximum hourly flow rate through that pipe. Disinfectant contact time within mixing basins and stor-
 age reservoirs must be determined by tracer studies or an equivalent demonstration.

 disinfection The process  designed to kill most microorganisms  in water, including essentially all
 pathogenic (disease-causing) bacteria. There are several chemical and physical ways to disinfect, with
 chlorine being the chemical most frequently used for disinfection in water treatment. Of the physical
 disinfection methods, UV is the most frequently used method.

 disinfection by-product A compound formed by the reaction of a disinfectant such  as chlorine with
 organic material in the water supply.
                                               G-3

-------
dissolved oxygen (DO)  Measure of water quality indicating free oxygen dissolved in water.

enhanced coagulation To more effectively control TOC and DBF.

epidemiologic study  Study of human populations to identify causes of disease. Such studies often
compare the health status of a group of persons who have been exposed to a suspect agent with that of
a comparable non-exposed group.

epidemiology A branch of medicine which studies epidemics (diseases which affect significant num-
bers of people during the same time period in the same locality). The objective of epidemiology is to
determine the factors that cause epidemic diseases and how to prevent them.

fecal coliform bacteria  Bacteria found in the intestinal tracts of animals. Their presence in water or
sludge is an indicator of pollution and possible contamination by pathogens.

filtration A process for removing particulate matter, microorganisms, and some chemical contami-
nants from water by passage through porous media.

finished water  Water that has passed through a water treatment plant; all the treatment processes are
completed or "finished."

floe Clumps of microorganisms and particulate impurities that have come together and formed a clus-
ter; found in flocculation tanks and settling or sedimentation basins.

flocculation The gathering together of particles and microorganisms in water to form larger particles
by gentle mixing after the addition of coagulant chemicals.

flushing A method used to  clean water distribution lines. In this method, hydrants are opened and
water is pumped at a high velocity through the pipes to remove deposits from the pipes that flow out the
hydrants.

free available residual chlorine  That portion of the total available residual chlorine composed of
dissolved chlorine gas C12 hypochlorous acid (HOC1) and/or hypochlorite ion (OC1 ) remaining in
water after chlorination. This does  not include chlorine that has combined with ammonia, nitrogen, or
other compounds.

free residual chlorination  The application of chlorine to water to produce a free available chlorine
residual equal to at least 80% of the total residual chlorine  (sum of free and combined available chlo-
rine residual).

garnet  A group of hard, reddish, glassy, mineral sands made up of silicates of base metals (calcium,
magnesium, iron,  and manganese). Garnet has a higher density than sand.

gastroenteritis An inflammation of the stomach and intestine resulting in diarrhea, with vomiting and
cramps when irritation is excessive. When caused by an infectious agent, it is often associated with
fever.

germicide  A substance formulated ( or a physical method  designed) to kill germs or microorganisms
such as chlorine and UV.

Giardia lamblia  Flagellate protozoan which is shed during its cyst stage into the feces of man and
animals. When water containing these cysts is ingested, the protozoan causes a severe gastrointestinal
disease called giardiasis.

giardiasis Intestinal disease caused by an infestation of Giardia with symptoms that include abdomi-
nal pains and explosive diarrhea.
                                             G-4

-------
ground water The supply of fresh water found beneath the Earth's surface, usually in aquifers.

ground water under the direct influence (UDI) of surface water Any water beneath the surface of
the ground with 1) significant occurrence of insects or other macroorganisms such as algae, or large-
diameter pathogens such as Giardia lamblia, or 2) significant and relatively rapid shifts in water charac-
teristics such as turbidity, temperature, conductivity, or pH which closely correlate to climatological or
surface water conditions. Direct influence must be determined for individual sources  in accordance
with criteria established by the State. The State determination of direct influence may be based on site-
specific measurements of water quality and/or documentation of well construction characteristics and
geology with field evaluation.

gross alpha particle activity  The  total radioactivity due to alpha particle emission  inferred from
measurements on a dry sample.

gross beta particle activity The total radioactivity due to beta particle emission inferred from mea-
surements on a dry sample.

hardness, water A characteristic of water caused mainly by the salts of calcium and magnesium, such
as bicarbonate, carbonate, sulfate, chloride, and nitrate. Excessive hardness in water  is undesirable
because it causes the formation of soap curds, increased use of soap, and deposition of scale in boilers
and pipes. Hardness also may cause damage in some industrial processes and sometimes causes objec-
tionable tastes in drinking water.

herbicide A compound, usually a man-made organic chemical, used to kill or control undesired plant
growth.

heterotrophic microorganisms Bacteria and other microorganisms that use organic matter as an en-
ergy source.

heterotrophic plate count (HPC) The number of colonies of heterotrophic bacteria grown on selected
solid media at a given temperature and incubation period, usually expressed in number of bacteria per
milliliter of sample.

humus Organic portion of the soil remaining after prolonged microbial decomposition.

hydrophilic Having a strong affinity (liking) for water; the opposite of hydrophobic.

hypochlorite  Chemical compounds containing available chlorine; used for disinfection available as
liquids (bleach) or solids (powder, granules, and pellets).

insecticide Any substance or chemical formulated to kill or control insects.

in vitro In glass;  a laboratory experiment performed in a test tube or other vessel.

in vitro studies Studies of chemical effects conducted in tissues, cells, or subcellular extracts from an
organism (i.e., not in the living organism).

in vivo Within a living organism; a laboratory experiment performed in which the substance under
study is inserted into a living organism.

in vivo studies Studies of chemical  effects conducted in intact living organisms.

ion An electrically charged atom, radical (such as SO4  2), or molecule formed by the loss or gain of one
or more electrons.

ionic concentration  The concentration of any ion in solution, usually expressed in moles per liter.
                                              G-5

-------
ionization  The splitting or dissociation (separation) of molecules into negatively and positively charged ions.

jar test A laboratory procedure that simulates a water treatment plant's coagulation/flocculation units
with differing chemical doses plus energy of rapid mix, energy of slow mix, and settling time. The
purpose of this procedure  is to estimate the minimum or ideal coagulant dose required to achieve
certain water quality goals. Samples of water to be treated are commonly placed in six jars. Various
amounts of chemicals are added to each jar, and the settling of solids is observed. The dose of chemi-
cals that provides satisfactory settling removal of turbidity and/or color is the dose used to treat the
water being taken into the plant at that time.

Langelier Index (L.L) An index reflecting the equilibrium pH of a water with respect to calcium and
alkalinity. This index is used in stabilizing water to control both corrosion and the deposition of scale.
Langelier Index = pH - pHs, where pH = actual pH of the water and pHs = pH at which the water
having the same alkalinity  and calcium content is just saturated with calcium carbonate.

legionella A genus of bacteria, some species of which cause a type of pneumonia called Legionnaires
Disease.

maximum  contaminant level (MCL)  The maximum permissible level of a contaminant in water
which is delivered to the free-flowing outlet of the ultimate user of a public water system, except in the
case of turbidity, where the maximum permissible level is measured at the point of entry to the distribu-
tion system. Contaminants  added to the water under circumstances controlled by the user are excluded
from this definition, except those contaminants (such as lead and copper) resulting from the corrosion
of piping and plumbing caused by water quality.

maximum contaminant level goal (MCLG)  The maximum level of a contaminant in drinking water
at which no known or anticipated adverse effect on the health of persons would occur and which allows
an adequate margin of safety. Maximum contaminant level goals are non-enforceable limits.

maximum total trihalomethane potential (MTTP)  The maximum concentration of total
trihalomethanes produced in a given water containing a disinfectant residual after 7  days at 25°C or
above.

microbial growth The activity and growth of microorganisms such as bacteria, algae, diatoms, plank-
ton, and fungi.

microgram (g) One-millionth of a gram (3.5  x 10"8 oz.  0.000000035 oz.).

micrograms per liter (|lg/L) One microgram of a substance dissolved in each liter of water. This unit
is equal to parts per billion  (ppb), since one liter of water is equal in weight to one billion micrograms.

micron A unit of length; one millionth of a meter or one thousandth of a millimeter. One micron equals
0.00004 of an inch.

microorganisms  Living organisms that can be seen individually only with the aid of a microscope.

milligrams per liter (mg/L) A measure of concentration of a dissolved substance. A concentration of
one mg/L means that one milligram of a substance is dissolved in each liter of water. For practical
purposes, this unit is equal to parts per million (ppm), since one liter of water is equal  in weight to one
million milligrams. Also see parts per million.

most  probable number (MPN) MPN is the most probable number of coliform-group organisms per
unit volume of sample water; expressed as the number of organisms per 100 mL of sample water.

mutagen An agent that causes a permanent genetic change in a cell other than that which occurs during
normal genetic recombination (e.g.,  mutagen MX).
                                             G-6

-------
mutagenicity The capacity of a chemical or physical agent to cause permanent alteration of the genetic
material within living cells.

National Pollutant Discharge  Elimination System (NPDES)  A system where the regulatory agency
(either federal or state) issues a document (permit) which is designed to control all discharges of pollut-
ants from point sources in U.S.  waterways. NPDES permits regulate discharges into navigable waters
from all point sources of pollution including industries, municipal treatment plants, large agricultural
feed lots, and return irrigation flows.

nephelometric  A means of measuring turbidity in a sample by using an instrument called a nephelom-
eter. A nephelometer passes light through a sample, and the amount of light deflected (usually at a 90°
angle) is then measured.

nephelometric turbidity unit (NTU) The unit of measure for  turbidity.

non-point source Pollution  sources which are diffuse and do not have a single point of origin or are
not introduced into a receiving stream or the environment from a specific outlet. The pollutants are
generally carried off the land by  stormwater runoff. The commonly used categories for non-point sources
are agriculture, forestry, urban,  mining, construction, land disposal, and saltwater intrusion.

non-potable  Water that may contain objectionable  pollution,  contamination, minerals,  or infective
agents and is considered unsafe and/or unpalatable for drinking.

oncology  Study of cancer.

oxic polymerization Polymerization of the organic compounds in an oxic environment.

particle count   The results  of a microscopic examination of  treated  water with a special "particle
counter" which classifies suspended particles by number and size.

particulate A very  small solid suspended in water which can vary widely in size, shape, density, and
electrical charge.

parts per million (PPM)  Parts per million parts,  a measurement of concentration on a weight or
volume basis. This term is  equivalent to milligrams per liter (mg/L), which is the preferred term.

pathogenic organisms Organisms, including bacteria, viruses, protozoa, or cysts, capable of causing
diseases (typhoid, cholera, dysentery) in a host (such as a person, plant, or animal). There are many
types  of organisms which do NOT cause disease, such as the bacteria used to process milk into cheese.
These organisms are called non-pathogenic.

pathogens  Microorganisms that can cause disease  in other organisms or in humans, animals, and
plants. They may be bacteria, viruses, or parasites and are found  in sewage runoff from animal farms or
rural areas populated with domestic and/or wild animals and in water used for swimming. Fish and
shellfish contaminated by pathogens, or the contaminated water itself, can cause serious illnesses.

pathology The study of disease.

pesticide  Any substance or  chemical designed or formulated  to kill or control undesired plants, in-
sects,  or animals. Pesticides include algicide, herbicide, insecticide, and rodenticide.

pico  A prefix used in the  metric system and other  scientific systems of measurement which means
10-12 or 0.000000000001.

picocurie (pCi) A measure of radioactivity. One picocurie of radioactivity  is equivalent to 0.037
nuclear disintegrations per second or about two disintegrations  per minute.
                                              G-7

-------
point of disinfectant application The point where disinfectant is applied. Water downstream of that
point is not subject to recontamination by surface water runoff.

point-of-entry treatment device A treatment device applied to the drinking water entering a house or
building for the purpose of treating the drinking water distributed throughout the house or building.

point-of-use treatment device A treatment device applied to a single tap used for the purpose of
reducing contaminants in drinking water at that individual tap.

point source A stationery location or fixed facility from which pollutants are discharged or emitted;
also, any single identifiable source of pollution (e.g., pipe, ditch, ship, ore pit, factory smokestack).
pollutant  Generally, any substance introduced into the environment that adversely affects the useful-
ness of a resource.
pollution  Generally, the presence of matter or energy whose  nature, location, or quantity produces
undesired  environmental  effects. Under the Clean Water Act, for example, the term is defined as the
man-made or man-induced alteration of the physical, biological, and radiological integrity of water.
polymer A chemical formed by the union of many monomers  (a molecule of low molecular weight).
Polymers are used with other chemical coagulants to aid in binding small suspended particles to form
larger and heavier aggregates than individual particles for their removal from  water. All polyelectro-
lytes are polymers, but not all polymers are polyelectrolytes.
prechlorination  The addition of chlorine at the headworks of the plant prior  to other treatment pro-
cesses, mainly for control of tastes, odors, and aquatic growths; also  applied to aid in coagulation and
settling.

precipitation  1) The process by which atmospheric moisture falls onto a land or water surface as rain,
snow, hail, or other forms of moisture, and 2) the chemical transformation of  a substance in solution
into an insoluble form (precipitate).
precursor Natural organic compounds found in all surface and groundwaters.  These compounds may
react with halogens (such as chlorine) to form trihalomethanes  (THMs) or other disinfectants to form