&EFA
United States
Environmental Protection
Agency
Municipal Environmental Research
Laboratory
Health Effects Research Laboratory
Cincinnati OH 45268
EPA-600/9-83-009
July 1983
Research and Development
Municipal Wastewater
Disinfection
Proceedings of Second
National Symposium
ci
uv
-------
EPA-600/9-83-009
July 1983
MUNICIPAL WASTEWATER
DISINFECTION
Proceedings of Second National Symposium
Orlando, Florida
January 26-28, 1982
Sponsored by the
Municipal Environmental Research Laboratory
and the
Health Effects Research Laboratory
Edited by
A. D. Venosa and E. W. Akin
MUNICIPAL ENVIRONMENTAL RESEARCH LABORATORY
HEALTH EFFECTS RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
-------
DISCLAIMER
The following papers have been reviewed in accordance with the U.S. Envi
ronmental Protection Agency's peer and administrative review policies and
approved for presentation and publication.
Infective Dose of Waterborne Pathogens
Elmer W. Akin
Viral Gastorenteritis Caused by Norwalk-
Like Agents
Raphael Do!in
Risk Assessment of Wastewater Disinfection
David W. Hubly
Wastewater Health Effects Studies and the
Need for Disinfection
Walter Jakubowski
Fresh Recreational Water Quality and
Swimming-Associated Illness
Alfred P. Dufour
Ultraviolet Dose Measurement in
Wastewater Disinfection
J. Donald Johnson
Pilot Investigation of Ultraviolet Wastewater
Disinfection at the New York City Port Richmond
Plant
0. Karl Scheible
A Comparison of Analytical Methods for
Residual Ozone
Gilbert Gordon
Control of Ozone Disinfection by Exhaust Gas
Monitoring
Albert D. Venosa
Ozone-Mass Transfer Coefficients
Edward J. Opatken
-------
The Effects of Operation and Maintenance
on the Performance of Selected Ozone Systems
Randy Junkins
The work described in the remaining papers was not funded by the U.S.
Environmental Protection Agency and therefore the contents do not necessarily
reflect the views of the Agency and no official endorsement should be inferred
-------
FOREWORD
The Environmental Protection Agency was created because of increasing
public and government concern about the dangers of pollution to the health and
welfare of the American people. Noxious air, foul water, and spoiled land are
tragic testimony to the deterioration of our natural environment. The com-
plexity of that environment and the interplay between its components require a
concentrated and integrated attack on the problem.
Research and development is that necessary first step in problem solution
and it involves defining the problem, measuring its impact, and searching for
solutions.
Two major functions of the EPA research and development program are (1)
to develop control technologies and systems to protect people from unnecessary
and harmful exposure to wastewater pollutants and (2) to determine the health
effects of waste treatment and disposal practices. To these ends, the Muni-
cipal Environmental Research Laboratory and the Health Effects Research Labo-
ratory in Cincinnati, Ohio have supported research studies in the respective
areas.
This report is the result of a combined effort of the two laboratories to
transfer relevant information obtained from recent research studies, most of
which were funded by EPA. The holding of a research symposium and the publi-
cation of the proceedings is a viable mechanism for disseminating the latest
results in a research area. This proceedings provides a comprehensive report
on what is known concerning the health and technological aspects of wastewater
disinfection.
F. Gordon Hueter, Director Francis T. Mayo, Director
Health Effects Research Municipal Environmental Research
Laboratory Laboratory
-------
PREFACE
This symposium was the sequel to a similar one on the same topic held in
Cincinnati, Ohio, in September 1978. It was designed to address many of the
questions raised and deficiencies in knowledge identified at the prior meeting
and to address an additional subject area, health aspects. The sessions were
organized into three scientifically related but topically separate research
areas: (1) health effects and epidemiology, (2) alternative disinfection
technology, and (3) design and operation/maintenance considerations.
A brief comment concerning organization of the proceedings' contents is
in order. The papers are printed in exactly the same order they were presented.
Most of the printed material, however, appears in much greater detail than was
presented orally. Those papers requiring peer review according to EPA's
publication regulations were so treated. All extemporaneous discussions were
tape recorded on site. Unfortunately, however, technical difficulties with
the microphone and recording equipment were experienced early in the meeting,
and consequently the questions and answers from the audience could not be
included in the written proceedings herein. This was truly a disappointing
development and the editors wish to apologize for their inability to provide a
written record of this valuable informal dialog.
-------
ACKNOWLEDGEMENTS
Appreciation is expressed to the speakers and authors of the papers for
their many hours of labor and preparation, to the session chairman, and to the
general registrants whose lively participation in the panel discussions con-
tributed greatly to the success of the symposium. We also wish to thank the
session moderators for the orderly progression of the sessions. Special thanks
is expressed to the banquet speaker, Dr. Arthur Lane, Jet Propulsion Laboratory,
whose banquet presentation entitled "The Voyager Odyssey to Jupiter and Saturn -
The Legacy of a Master Storyteller" roused the fascination of all who attended.
The editors also acknowledge the perseverance and efforts of Ms. Sheri
Marshall of the Dynamac Corporation and Mr. Denis Lussier of EPA's Center for
Environmental Research Information for arranging for the hotel and banquet
accommodations and coordinating registration and other administrative activities.
VI
-------
CONTENTS
Foreword i v
Preface v
Acknowledgements vi
SESSION 1: HEALTH 1
V. C. Cabelli and E. W. Akin, Session Chairmen
1. Don't Chiorinate Sewage 1
James B. Coulter
2. Pathogens? In Sewage?! 13
Henry J. Ongerth
3. Infective Dose of Waterborne Pathogens 24
Elmer W. Akin
4. Viral Gastroenteritis Caused by Norwalk-like Agents 40
Raphael Do!in
5. Risk Assessment of Wastewater Disinfection 55
David W. Hubly
6. Wastewater Aerosol Health Effects Studies and the Need for
Disinfection 68
Walter Jakubowski
7. Requirements for Wastewater Disinfection as Seen from the Results
of Epidemiological-Microbiological Studies 83
Victor J. Cabelli
8. Fresh Recreational Water Quality and Swimming-Associated Illness .. 99
Alfred P. Dufour
SESSION 2: TECHNOLOGY 120
A. D. Venosa and C. N. Haas, Session Chairmen
1. Optimization of Mixing for Disinfection 120
Karl E. Longley
2. Upgrading Existing Chlorine Contact Chambers 137
Frederick L. Hart
vii
-------
3. Problems of Disinfecting Nitrified Effluents
George C. White, et. al .
4. Operating Experience Disinfecting Secondary Effluent with Pilot
Scale Ultraviolet Units ........................................ 167
Paul H. Nehm
5. UV Disinfection of Secondary Effluent: Dose Measurement and
Filtration Effects ............................................. 184
J. Donald Johnson, et. al .
6. Pilot Investigation of Ultraviolet Wastewater Disinfection at the
New York City Port Richmond Plant .............................. 202
0. Karl Scheible, et. al.
7. Comparison of Analytical Methods for Residual Ozone .............. 226
Gilbert Gordon, and Joyce Grunwell
8. Control of Ozone Disinfection by Exhaust Gas Monitoring .......... 246
Albert D. Venosa and Mark Meckes
9. Optimizing Operational Control of Ozone Disinfection ............. 260
Enos L. Stover
10. Pilot Studies of Ozone Disinfection and Transfer in Wastewater ... 277
Patrick W. Given and Daniel W. Smith
11. Ozone-Mass Transfer Coefficients ........ . ....................... . 293
Edward J. Opatken
12. Innovations in the Electrolytic Generation of Ozone .............. 310
Peter C. Roller
SESSION 3: DESIGN/O&M ..... ..... ____ .... ..... . .............. . ........ 329
Gilbert Gordon, Session Chairman
1. Practical Considerations in the Use of Halogen Disinfectants ..... 329
Charles N. Haas
2. Design and Operational Considerations for Wastewater Ozone
Disinfection Systems .................. . ........................ 339
3. The Effects of Operation and Maintenance Practices on Selected
Ozone and Ultraviolet Disinfection Systems ..................... 359
Randy Junkins
4. Second National Symposium on Municipal Wastewater Disinfection -
Summary and Closing Remarks .................................... 372
Charles C. Johnson, Jr.
VI 1 1
-------
1. DON'T CHLORINATE SEWAGE
James B. Coulter, Secretary
Maryland Department of Natural Resources
Tawes State Office Building
Annapolis, Maryland 21401
ABSTRACT
During the last decade, fisheries dependent on tributaries and fresh-
water reaches of Chesapeake Bay have declined significantly. The decline
took place in waters that should have benefited most by an unprecedented
investment in sewage treatment plant construction. In every case, inquiries
into the possible reasons for the losses implicated chlorine. Investigation
showed that the use of chlorine at sewage treatment plants discharging into
vital fish spawning areas had increased by several fold. More thorough study
shows that chlorine and its byproducts are toxic to aquatic life, repel
and thus deny spawning grounds to anadromous fish, and at barely detectable
concentrations, decimate fish larvae and other first emergent forms of life.
Furthermore, it is found that chlorination of ordinary sewage treatment
plant effluent provides no significant public health protection and to the
contrary, could result in public health hazards that might go undetected.
INTRODUCTION
Chesapeake Bay is the most productive estuary in the world. Under the
dual assault of increasing population and a rising standard of living, the
Bay has remained surprisingly beautiful and productive after three centuries
of civilization. Where the Bay is concerned, Maryland and Virginia have
practiced strong conservation measures for more than a hundred years.
However, during the nineteen seventies, aquatic life dependent on the
Bay's tributaries showed signs of unusual disturbance. It is in the tri-
butary streams that anadromous fish come to spawn, other fish reside year
round, and still others come to forage. For finfish, the struggle to pre-
serve the chemical, physical and biological integrity of Chesapeake Bay
will be won or lost in its tributaries and tidal freshwater reaches.
During the Seventies, shad runs almost ceased. The commercial catch
from the Susquehanna River and its flats at the head of the Bay dwindled
from 184,000 pounds in 1971 to 2,300 in 1979. The Maryland Department of
Natural Resources banned further harvesting of shad to protect the last
remaining brood stock. Striped bass, the famed rockfish of the East Coast,
-------
went from a condition of plenty to one of relative scarcity. In 1970, the
young-of-the-year averaged slightly more than thirty per seine haul during
the annual survey conducted by the Maryland Department of Natural Resources.
By 1981, the average was barely more than one per haul. Perch and other
resident fish showed a marked decline in some tributaries.
It was puzzling that this deterioration took place during the Seventies,
a decade of unprecedented expenditure for sewage treatment plants and other
water pollution control measures. One possible solution to the puzzle
began to emerge as the search for reasons for tributary crop failures pro-
gressed. In every case, chlorine was implicated. That led to a look at the
use of chlorine. It was found in six spawning rivers that chlorine discharge
increased 4.4 fold from 1974 to 1980.
An estimated 13,900 tons of residual chlorine per year are discharged by
Maryland sewage treatment plants. Health Department records reveal that 115
sewage treatment plants annually discharge about 300 tons of residual chlorine
into spawning rivers.
The practice of chlorinating sewage treatment plant effluent was
examined to find if it is a significant factor causing damage to Chesapeake
Bay's tributary dependent aquatic life. The public health aspects of the
practice were examined also.
DAMAGE TO AQUATIC LIFE
Literature has proliferated in recent years as the damage to aquatic
life caused by chlorinated sewage effluent has become more and more apparent.
Space will not permit citation of all of the reports and publications re-
viewed. Instead, a small number have been selected to illustrate conclusions
drawn from a far greater volume of literature.
Collins and Deaner (3) quoted literature (9) (10) to show that when
wastewater is chlorinated, toxic compounds such as cyanogen chloride can
be formed. Questions regarding the formation and nature of the various
toxic compounds and their effect on aquatic life remain unanswered because
of the complexity of sewage and chlorine reactions.
Work of Michigan's Department of Natural Resources was described which
proved that chlorinated sewage is toxic to fish. Fathead minnows and rain-
bow trout xrere exposed to chlorinated and unchlorinated sewage effluents.
Survival \
-------
Collins and Deaner reported also on chlorine-induced fish kills in
California's Sacramento River. To test the thesis that chlorinated effluent
was the culprit, king salmon fry were exposed to river water taken upstream,
at the discharge point, 100 feet downstream, and 200 feet downstream. The
upstream water caused no adverse effects. Water from the discharge point
killed all of the fish in 12 minutes. In less than an hour, all of the fish
in the water taken 100 feet downstream from the discharge point were dead
and, in less than an hour and a half, all were dead in the 200 feet down-
stream water. In a companion test, salmon fry were suspended in the
Sacramento River. All fish below the outfall were dead within 14 hours
while all above survived. Downstream chlorine residuals ranged from 0.2 mg/1
to 0.3 mg/1 during the test period.
Osborne, et al, (17) studied the effects of chlorinated sewage efflu-
ents on fish in the Sheep River, Alberta, Canada. They found no mortality
when caged fish were subjected to unchlorinated effluent but 100 percent
mortality occurred when exposed to chlorinated effluent. They concluded
that chlorination of effluent was the principal factor in fish death. Quan-
titative sampling of fish populations supported the contention that fish
avoid chlorinated effluents.
Giattina, et al, 1,6) also investigated the avoidance of fish to chlorine
at a power plant on the New River in southwestern Virginia. They reported
that laboratory determined avoidance concentrations generally predicted
the total residual chlorine concentrations that would elicit avoidance be-
havior under natural field conditions. In general, fish avoid chlorine
residuals that are 50 percent or less of the median lethal concentration.
Tsai (21) studied fish life below 149 sewage treatment plants and con-
cluded that turbidity and chlorine caused species diversity reduction below
the outfalls. In the upper Patuxent River, (22) chlorinated sewage acts
as a toxic material which seriously reduces fish abundance below outfalls,
and chlorinated sewage will trigger fish to avoid the outfall water.
Chronic physiological responses to chlorine include delayed mortality, de-
pressed activity, decreased growth, and decreased spawning success.
Freshwater reaches of upper Chesapeake Bay are important spawning
grounds for many fish species including striped bass. Annual surveys showed
that by the end of the Seventies, egg-laden female rockfish still returned
to their spaxming areas each Spring in great numbers. Eggs were released
and found fertilized in the water but few survived to become small fish.
It has been shown (12) that chlorine in concentration as low as 0.01 mg/1
greatly reduces the percentage of rockfish eggs that are hatched. To com-
pound the problem it has been found (12) that the larvae once hatched
continue to be decimated by chlorine. A total residual of only 0.04 mg/1
is lethal in one hour to 50 percent of two day old larvae. Chlorine is
equally toxic to 30 day old juvenile fish.
-------
Chlorine in the saltwater .portion of Chesapeake Bay produces toxic
oxidants, chlorine-produced oxidants, from naturally occurring bromine.
Eggs and larvae of oysters and clams are very sensitive to chlorine-produced
oxidants. Roberts and Gleeson (18) demonstrated that 50 percent of four
hour old oyster larvae are killed by only 0.026 mg/1 of such oxidants.
Rosenburg and co-workers (19) found that chlorine-produced oxidants were
lethal to 50 percent of 96 hour old oyster larvae at concentrations of
0.06 mg/1 and 16 hour old clams at 0.27 mg/1.
PUBLIC HEALTH JUSTIFICATION
Attention turned to alternatives as evidence began to demonstrate that
sewage treatment plants chlorinating their effluent are a major source of
toxic pollutants. Alternatives under consideration include: better control
of chlorine; detoxification of the effluent; substitution of biocides that
produce less toxic residuals; and use of a chemical or radiation that will
produce a residual-free effluent. Unfortunately, each alternative has its
own set of costly difficulties, and may damage aquatic life. Each may pose
some danger to sewage treatment plant operators and perhaps to the surround-
ing community.
For instance, better control of chlorine application may seem to be a
simple inexpensive matter, but it isn't. Much improvement can be obtained
by eliminating wasteful, almost promiscuous, misuse of chlorine, but that
is not enough. There are very few sewage treatment plants that have been
built so that precise control of effluent residual in the part per billion
range is possible. To meet an effluent standard that low, drastic changes
have to be made in the capability of the sewage treatment plant and in its
operation. The orthotolidine color comparitor is useless. Instead, the most
precise method of analytical measurement must be used. Automatic chlorine
residual monitoring and feedback control units are necessary. Only four
percent of the sewage treatment plants that were surveyed (7) have feedback
control. In contrast, 60 percent use a manual method to feed chlorine.
Before blindly accepting the proposition that there is a need to find
a substitute for chlorine, the possibility that disinfection of sewage ef-
fluent is not necessary in most cases should be examined. The public health
necessity of disinfecting sewage effluent under ordinary circumstances
must be justified for the practice to continue in any form.
Some disagree (11) claiming that: "The cornerstone of public health
is preventive medicine and to require the justification for wastewater
disinfection is a giant step backward." The fault in that assertion is that
the alleged "public health" and "preventive medicine" benefits of effluent
chlorination are what need to be justified. As for requiring justification,
the health of the human race was improved dramatically as soon as public
health practitioners were required to justify their strongly held beliefs.
-------
There is an assumption that the act of chlorinating sewage will de-
crease the danger of disease, but for all practical purposes, that assumption
is not valid. Food or water contaminated with sewage will cause disease
and remains dangerous whether it is chlorinated or not. After a decade of
nationwide chlorination of sewage, there is no evidence to demonstrate that
the incidence of any illness has decreased as a result of that practice.
The United States chlorinates its sewage - England doesn't. There is no
credible evidence to show that any related illness occurs more frequently
in England than it does in the United States.
The U.S. Public Health Service with its Center for Disease Control
in Atlanta, Georgia, is the world's outstanding authority on the causes
of disease and how to prevent them. The Comptroller General reported to
Congress (4) that "The Center for Disease Control has taken the official
position that disinfection of sewage provides little public health benefits".
In correspondence, G. F. Mallison of the Bacterial Diseases Division of the
Center for Disease Control, wrote "I see, with rare exceptions, absolutely
no need with respect to health in attempting to control microbial contami-
nation after secondary sewage treatment".
Health Hazard to Workers
An examination of the health effects of chlorinating sewage might
start with its effect on sewerage workers. In the debate over the public
health benefit or lack thereof that comes from chlorinating effluent, the
health of the sewage treatment plant operator is largely ignored. That is
a mistake because chlorine creates an occupational hazard and there have been
a significant number of incapacitating accidents. Chlorine in the air
is almost as toxic to humans as chlorine in the water is to aquatic life.
A concentration of 0.1 percent of chlorine in the air is likely to be fatal
after a few breaths and almost certain to cause death within ten minutes.
A safe allowable concentration of one part per million has been established
by the Occupational Safety and Health Administration.
In a survey (7) conducted and reported by the Water Pollution Control
Federation in 1980, it was found that over 11 percent of the sewage treat-
ment plants surveyed reported chlorine accidents in which people required
medical treatment.
Debate Over Recreation Hater
Protection of the health of people using water for recreation is a
frequently used justification for sewage chlorination even though epidemio-
logical evidence of its value in that regard is nonexistent. In fact, no
study has examined the proposition that recreation waters shown to cause
disease can be made safe by chlorinating sewage effluent. Instead, the
effort to date has been to demonstrate, if indeed it is possible to demon-
strate, that swimming in polluted water causes a higher incidence of disease
-------
and, if so. to find an indicator bacterium that correlates with risk. For
thirty years the aim has been to establish a number for a particular indi-
cator organism that will give assurance against disease contracted from
swimming in sewage polluted water.
That there is a safe threshold of pollution for swimming, and that
such a threshold can be identified through an allowable number of easily
measured indicator bacteria, is a strongly held belief, but it is not
shared by all. Stevenson (2) pioneered studies in Lake Michigan and the
Ohio River. Though the studies were far from conclusive, he arrived at
a concentration of total coliform bacteria as the best practical standard.
Geldreich (5) related Salmonella detection to fecal coliform densities and
recommended a standard based on fecal coliform detection. Cabelli (2) found
an increase in gastrointestinal disturbances among those swimmers who im-
mersed their heads in water. Based on a correlation with fecal enterococci ,
a mathematical expression of the risk of increased incidence of disease was
developed.
A higher incidence of disease caused by swimming in polluted waters is
not a universal finding. The National Technical Advisory Committee found
Public Health Service studies on which the coliform standards are based
to be far from definitive. They expressed an urgent need to find if there
is a correlation between the various indicator organisms and disease attri-
butable to water recreation. In Sydney, Australia, many years of epidemio-
logical study in connection with Sydney's world famous bathing beaches
produced no evidence of water-borne diseases caused by unchlorinated sewage
effluent.
In the United Kingdom, a committee which Moore (14) headed did research
for six years in the 1950's and failed to establish any significant bacterial
hazard from sea bathing. Later work by the Water Pollution Research Labora-
tory also failed to find a satisfactory method for establishing bacterial
standards for bathing waters. It is Moore's contention that no shred of
evidence has been produced in Europe during the past 20-30 years that indi-
cates that human health has been endangered in the absence of bathing water
standards.
From a realistic public health perspective, the incidence of sewage
pollution related diseases contracted through recreational use of water is
trivial. Competent persons have searched for such a relationship. Some
claim that it does exist and others find that it does not. Even if it does
exist, the effort required to ferret out the relationship is strong testimony
that swimming in polluted waters accounts for a miniscule fraction of the
total incidence of serious disease. Most of the minor irritations that do
occur are of the eye, ear, nose, and skin variety- making it likely that
transmission is person to person and not sewage to person. It is highly
unlikely that an enteric disease indicator bacterium will ever be found that
correlates with those ailments.
-------
Even if a sewage treatment plant discharge to swimming water disease
relationship does exist, effluent chlorination would be the wrong thing
to do. In fact, health receives better protection if sewage effluent is
not chlorinated. Chlorination of ordinary sewage treatment plant effluent
kills more of any of the various indicator bacteria than it does of the
virus in sewage effluent, and virus as well as other chlorine resistant
organisms are the main cause of concern. That being the case, chlorination
of sewage effluent diminishes the indicators of pollution in relation to the
prevalence of the real danger, thus, creating a false sense of security.
A safer course of action is to provide better sewage treatment and greater
separation between outfalls and bathing beaches.
Shellfish
Like bathing beaches, chlorinating effluent gives the illusion of
public health protection, but the real protection of shellfish growing
waters is provided by good sewage treatment and safe separation between
outfalls and shellfish beds. Consumption of raw oysters harvested from
sewage polluted waters caused a high incidence of disease prior to the
shellfish sanitation program initiated by the U.S. Public Health Service
in the late 20's. Since the time that the program became effective, not
one case of illness has been traced to oysters harvested from approved
waters in Maryland.
The principal elements of this effective program are separation be-
tween pollution discharge and shellfish harvesting beds coupled with a
bacteriological standard applied at the place of harvest. The bacterial
standard for shellfish harvest water was derived from empirical observations
at a time when the discharge of untreated sewage was commonplace and many
people became ill from eating oysters taken from polluted water. Unlike
recreational waters, it was clearly demonstrated that when people ate
oysters taken from polluted water with an indicator bacterial density
higher than the standard, they got sick. When they ate oysters from waters
cleaner than that indicated by the standard, they did not get sick.
The shellfish harvesting bacterial standard works because of the general
relationship that exists between the density of indicator bacteria and the
density of disease agents. Chlorination of ordinary sewage treatment plant
effluent alters the indicator/disease producing organism ratio in a dangerous
fashion. It is disconcerting that virus can persist even after indicator bac-
terial organisms have been killed, because shellfish contamination by virus
has replaced bacteria as the disease agent of major concern.
Olivieri, et al, produced data that strongly supports the hypothesis
that free chlorine is required for significant viral reductions (16). Free
chlorine for the required contact time calls for break-point chlorination,
rapid mixing, and precise hydraulic control, things that are rarely
achieved in conventional sewage treatment plant operation.
-------
Recognizing that chlorine can disrupt the traditional indicator-
pathogen ratio, Bisson and Cabelli (1) have looked for alternatives.
They have examined the feasibility of using a spore former, Clostridium
perfringens, as an indicator for the potential for infectious disease from
fecal pollution because the spores of C. perfringens are much more resistant
to chlorination that H.coli. For specific applications against the potential
for infectious disease arising from fecal pollution of the aquatic environ-
ment, they suggest that there is no universal microbial indicator.
Destruction of the Natural Barrier
The argument is sometimes advanced that chlorination of ordinary
sewage treatment plant effluent provides another barrier in a multiple
barrier concept of public health protection. The strategy is to provide
as many barriers between a source of disease organisms and the public as
opportunity and cost will permit. The idea is sound but chlorination of
sewage treatment plant effluent does not impose a dependable barrier. In-
stead, it destroys one of the most effective barriers in existence. That
barrier is nature's relentless antagonism to the disease producing bacteria
and virus found in sewage.
Mitchell (13) studied the destruction of sewage bacteria and virus
that were discharged into seawater. He found that enteric bacteria are
destroyed by a specific antagonistic microflora that develops. Mitchell
was able to classify three groups of native seawater organisms associated
with the accomplishment of this destruction: native bacteria that destroy
by enzymatically lysing enteric bacteria cell walls; obligatory parasitic
bacteria; and, amebae which attack and consume bacterial cells. Of these,
the amebae are the most active. With respect to virus, native marine micro-
flora are involved in a manner similar to that observed with enteric bacteria
but a chemical component of seawater was also shown to be involved in the
virus destruction.
The specialized culture that develops in biological sewage treatment
processes exhibits similar antagonism to disease producing organisms. Un-
fortunately, chlorination of sewage effluent kills the predators as well
as the prey. The culture of specialized organisms that started their attack
on sewage-borne pathogens within the sewage treatment plant are disrupted
and the disruption carries over to the organisms of natural purification
in the receiving waters. Walsh and Mitchell (23) found that chlorination
of effluent produced hydrocarbons which can cause damage to the natural
predators responsible for self purification in the vicinty of sewage out-
falls.
In most situations the barrier imposed by nature's system is far more
important to the protection of shellfish beds than the superficial protec-
tion gained by the mere reduction of indicator bacteria that occurs when
chlorine is added to ordinary sewage treatment plant effluent.
-------
Disinfection
Contrary to repetitive misuse of the word in water pollution con-
trol literature, the conventional practice of chlorination at sewage treat-
ment plants does not produce a disinfected effluent. The term "disinfection"
is used to describe a process that removes all organisms capable of pro-
ducing a disease. In every other field of endeavor, including milk, food,
drinking water, and hospital care, "disinfection" has that meaning. It
does not imply sterilization where all forms of life are destroyed, but it
does mean that a disinfected material will no longer produce infectious
diseases.
Water pollution control workers are quick to point out that in the
general case,, they don't mean that kind of disinfection when they use the
word. No matter what the professional means, it is what administrators,
the press, and the informed public believe that counts. The public wrongly
perceives that chlorinated sewage is disinfected because water pollution
control workers continually tell them that it is.
No knowledgeable person would contend that chlorination of ordinary
sewage treatment plant effluent would render it disinfected, incapable of
producing disease. The reverse is true; chlorinated sewage treatment plant
effluents are highly infectious and should be treated with appropriate
caution. The use of the word, disinfection, is in itself dangerous in this
situation because it promotes a false sense of security and that could lead
to relaxation of the basic principles of sanitation that are, after all,
the main bulwark of public health protection.
It is well established that stringent conditions must be met before
chlorine or any chemical that acts in a related fashion can disinfect. Those
conditions include the removal of essentially all suspended solids, turbidity,
and interfering substances including BOD. Sewage effluent requires filter-
ing and break-point chlorination to produce on the order of 1.0 mg/1 of
hypochlorous acid (HOCL) for 30 minutes to achieve disinfection. Chlorine
must be completely and uniformly mixed as rapidly as possible. Careful en-
gineering of a holding and contact chamber is a necessity. Morris (15) has
pointed out that any measurable degree of short circuiting is ruinous. Only
0.01 percent of raw fluid may cause the water to fall below hygienic stan-
dards .
Obviously disinfection is not accomplished when chlorine is added to
the solids laden, organic rich effluent from an ordinary secondary sewage
treatment plant. Only in a very few instances where sewage is being con-
ditioned for direct reuse in specifically designed and operated purification
works is true disinfection practiced.
Chlorinated Hydrocarbons
While some persons within the U.S. Environmental Protection Agency
-------
continue to support the chlorination of effluent as the best practicable
measure, others in the agency are calling attention to the possible public
health problem that chlorination of sewage effluent is creating. In a state-
ment on the effects of chlorine on Chesapeake Say organisms, the EPA pointed
out that recently an unforeseen chlorine problem surfaced. Chlorine intro-
duced into sewage effluent can form a large variety of daughter compounds
of concern to drinking water supplies. Hunter and Sabatino (8) searched
out the sources of halogenated hydrocarbons in an urban water supply from
the Passaic River in New Jersey. The project which covered only the usual^
identifiable chlorinated compounds indicated that during the summer, chlori-
nation practices account for the predominant volatile halogenated hydrocar-
bons observed.
DISCUSSIOII
It should come as no surprise that the chlorine in sewage effluent is
killing valuable aquatic life. Chlorine has been used for seventy years
to kill a wide variety of unwanted aquatic organisms. Pollution control
experts use chlorine to kill bacteria in wastewater, to kill fouling organ-
isms in cooling water, in fact, to kill many things for many reasons.
When sewage treatment plant effluent is chlorinated, the killing effect
continues to be exerted on a host of organisms in the aquatic environment.
The effects fall into three categories: toxicity to fish and other mature
forms of life; fish avoidance of chlorinated effluent; and, destruction of
larvae and other first emergent forms of aquatic life.
Fish kills are likely to occur where there is an excessive use of
chlorine. Fish kills are spectacular and receive immediate attention in the
form of field surveys and bioassays. But even though they go largely un-
noticed, the deadly subtle effects on fish migration and reproduction are
far more devastating to many forms of aquatic life. Unlike fish kills, the
disruption of the reproductive process is unseen, but it is of fundamental
importance because it strikes at the ability of a species to sustain itself
through seasonal reproduction.
In the Maryland portion of Chesapeake Bay, there are more than a hundred
sewage treatment plants that discharge into tributary streams where fish
come to spawn. The discharge from a single sewage treatment plant is often
a sizable fraction of the total stream flow and many tributaries have multi-
ple points of discharge. Because spawning fish retreat from the slightest
trace of chlorinated effluent, chlorination creates an impenetrable barrier
that prevents the fish from reaching their spawning grounds. Should fish
be able to find a place to spawn in a stream below a sewage treatment plant
outfall,, the killing effect of chlorine first on the eggs, then on the larvae,
and then on the immature fish makes survival to adulthood very unlikely.
Oysters and clams have been shown to be susceptible to very low levels
10
-------
of chlorine produced oxidants. As with fish, damage to oysters and clams
is far greater to the first emergent forms of life during reproduction than
it is to the adult.
To offset the damage being done to the aquatic environment, there
would need to be an overriding public health benefit derived from the wide-
spread chlorination of sewage treatment plant effluent. Instead of bene-
fiting public health, chlorination of effluent produces unwanted chlorinated
hydrocarbons, creates a hazard to sewerage workers, could create a hazard
at bathing beaches, gives a false signal at shellfish harvesting grounds,
destroys a natural barrier to transmission of disease, and fails completely
to disinfect ordinary effluent,
Chlorination of ordinary sewage treatment effluent provides no appre-
ciable public health benefit to offset the major damage that it causes.
No other industry would be allowed to discharge a toxic pollutant capable of
causing damage like that of chlorinated effluent. The practice should be
stopped.
LITERATURE CITED
1. Bisson, J.W. and Cabelli, V.J., 1980, Clostridium perfringens as a
water pollution indicator, Journ.WPCF, 52:241-248^
2 Cabelli, V.J., 1980, Health effects criteria for marine recreational
waters, Report to the U.S. EPA, EPA-600/1-80-31 •
3. Collins, C.F. and Deaner, D.G., 1973, Sewage chlorination versus
toxicity - a dilemma?, Journ. EED, ASCE, 99:761-772.
4. Comptroller General, 1977, Report to Congress on the excessive use
of chlorine in sewage treatment plant effluents.
5. Geldreich, E.G., 1970, Applying bacteriological parameters to
recreational water quality, Journ. Am. Water Works Assn. 62:113-120.
6. Giattina, J.D., Cherry, D.S., Cairns, J. and Larrick, S.R., Comparison
of laboratory and field avoidance behavior of fish in heated chlori-
nated water, 1981, Trans. Am. Fish. Soc. 110:526-535.
7- Highlights, 1980, WPCF, July.
8. Hunter, J.V. and Sabatino, J., 1981, Sources of halogenated hydrocarbons
in an urban water supply, Report to EPA, NT5.
9. Ingols, R.S., Gaffney, P.E. and Stevenson, P.C., 1966, Biological
activity of halophenols, Journ. WPCF, 38:629-635.
11
-------
10. Katz, M. and Gaufin, A.R., 1952, The effects of sewage pollution
on the fish population of a midwestern stream, Trans. Am.Fish
Society, 82:156-165,
11. Kazuyoski, K., Olivieri, V.P. and Kruse, C.W., 1979, Discussion,
Wastewater disinfection - toward a rational policy, Ross, S.A.,
Journ. WPCF, 51:2023.
12. Middaugh, D.P., Couch, J.A. and Grove, A.M., 1977, Responses of
early life history stages of the striped bass, Morone saxatilis,
to chlorination, Ches. Sci.s 18:141-153.
13. Mitchell, R., 1971, Destruction of bacteria and viruses in seawater,
Journ. San. Eng. Div., ASCE, 97:425-432.
14. Moore, B., 1959, Sewage contamination of coastal bathing waters in
England and Wales. A bacteriological and epidemiclogical study,
Journ. Hyg. 57:435-472.
15. Morris, J.C., 1971, Chlorination and disinfection - state of the art,
Journ. AWWA, 63:769-774.
16. Olivieri, V.P., Donavan, T.K., and Kawata, K. , (1971), Inactivation
of virus in sewage, Journ. San. Eng. Div., ASCE, 97:661-673.
17. Osborne, L.L., Iredale, D.R., Wrona, F.J., and Davis, R.W., 1981,
Effects of chlorinated sewage effluents on fish in Sheep River,
Alberta, Trans. Am. Fish Soc., 110:536-540.
18. Roberts, M.H. and Gleeson, R.A., 1978, Acute toxicity of bromochlori-
nated seawater to selected estuarine species with a comparison to
chlorinated seawater, Marine Environmental Research 1:19-30.
19. Rosenburg, W.H., Rhoderick, J., Block, Kennedy, S,, Gullans, S,,
Vreengoor, S., Rosenkranz, A., and Collette, C., 1980, Effect of
chlorine produced oxidants on survival of larvae of oysters,
Crassotrea virginica, Marine Ecology, 3:93-96.
20. Stevenson, A.H., 1953, Studies of bathing water quality and health,
Am. Journ. Pub. Health, 43:529.
21. Tsai, Chu-Fa, 1968, Effects of chlorinated sewage effluents on fishes
in upper Patuxent River, Maryland,Ches. Sci. 9:83-93.
22. Tsai, Chu-Fa, 1973, Water quality and fish life below sewage outfalls,
Trans. Am. Fish Soc.,, 102:281-292.
23. Walsh, D. and Mitchell, R., 1974, Inhibition of intermicrobial pre-
dation by chlorinated hydrocarbons, Nature, 249.
12
-------
2. PATHOGENS? IN SEWAGE?!
Henry J. Ongerth, Consulting Engineer
Retired - California Department of Health Services
Berkeley, California
ABSTRACT
Domestic sewage is an infectious material carrying human pathogens shed
in the fecal discharges of infected individuals. Sewage effluents may affect
shellfish growing areas, sources of domestic water supply, recreational waters,
the ultimate users of rec.1 aimed sewage and others. For most of these dis-
infection is necessary to prevent infectious disease transmission. Chlorine
is almost universally used as a disinfectant for sewage. As to the suit-
ability of two other disinfectants, ozone and ultraviolet light, questions
must be answered concerning effectiveness, energy and dollar costs, and
practicability. The need for disinfection to meet bathing water quality
.standards is more extensively discussed. Sewage discharged to recreational
waters in significant concentrations will cause disease. How much disease,
and what are significant concentrations, what are infectious doses, what is
the best indicator organism, and what standards are to be used, are discussed.
It is concluded that fecal coliform is the best available indicator organism,
though not entirely satisfactory, but that sewage effluents must be
disinfected to protect users of recreational waters.
INTRODUCTION
This will not be an exhaustive discussion of the need for sewage dis-
infection. A brief commentary should suffice. Domestic sewage is an
infectious material - carrying human pathogens shed in the fecal discharges of
infected individuals. The concentration of the pathogens depends largely upon
the extent of infection in the tributary population. At the turn of the
century domestic sewage may still have carried some cholera organisms and
certainly carried significant numbers of typhoid organisms. Cholera is long
gone today and typhoid organisms must be at an exceedingly low level. Does
this mean that sewage is no longer hazardous? Not so. It is certain that
sewage in this ninth decade of the century is infectious, containing bacterial
pathogens, enteroviruses and parasites. Can there be any question, therefore,
about the necessity for disinfecting sewage?
DISCUSSION
In order to assess the consequence of not disinfecting sewage let us
consider the various modes of disposal of sewage effluents - to oceans and
13
-------
estuarine waters, to streams and lakes, to land, and for planned sewage reuse
projects. Discharged effluents nay affect shellfish growing areas, sources of
domestic water supply, waters used for food crop irrigation, recreational
waters, salt or fresh, or may be reused in planned projects for an array of
purposes. These include golf course - urban landscape, and crop irrigation,
recreational lakes, process and cooling water for industry, groundwater re-
charge and others.
There is no question that undisinfected sewage effluents reaching shell-
fish growing areas in any substantial concentration will infect shellfish
and that these shellfish will transmit disease. Through the first half of
this century many outbreaks of typhoid fever and paratyphoid fever and cases
of illness have been reported as shellfish associated diseases. Since the
1950's when raw shellfish was first shown to be a route of transmission for
infectious hepatitis there have been about 17 outbreaks involving some 1339
persons in the U.S. (19). In a recent outbreak of some 268 cases, the oysters
were traced to beds in Louisiana which earlier in the year had been closed due
to pollution associated with sewage polluted waters (20).
There is no doubt that large doses of undisinfected sewage may overwhelm
water treatment facilities, in extreme cases even those with filtration
facilities. The hepatitis epidemic in New Delhi in 1955-195G is evidence
of that (3). Moreover, good practice dictates that for an adequate level of
protection of domestic water supply, multiple factors of safety must be pro-
vided. Thus, effective disinfection of sewage discharges upstream from
domestic water supply intakes is essential. There can be no doubt, either,
that for most types of planned sewage reuse such as those cited above, a high
degree of treatment including reliable and effective disinfection is essential
to meet the appropriate water quality standards in order to prevent disease
transmission (5) .
Sewage chlorination also has been practiced for many years to delay bac-
terial action and to stretch out in time and distance the impact of BOD in
receiving waters - to modulate the oxygen sag curve. The most recently
recognized reason for sewage disinfection relates to identification of "R"
factors in bacteria. These factors are nucleic acid elements in bacteria
causing resistance to antimicrobial drugs. Coliforms may act as reservoirs
of "R" factors and transfer them to pathogens. There is evidence that sewage
polluted water may play a role in spread of coliform and other bacteria carry-
ing "R" factors, This supports providing effective disinfection of sewage
effluents (11).
There are of course, some situations where sewage disinfection may not be
necessary. Land-disposal projects and planned groudwater recharge projects
generally may be carried out without disinfection - though there may be some
situation where disinfection is warranted. Long-outfall ocean discharge,
particularly to deep-water sites may be accomplished without disinfection, if
the effluent does not significantly degrade adjacent recreational waters.
Determination of "significant" degradation depends upon guiding standards and
will be discussed in some detail.
14
-------
In a related situation, the recognition that chlorination results in
creation of trihalonethanes (TIIMs) at some domestic water treatment fa-
cilities, lead to questions about continuing the use of chlorine for this
treatment. Response of public health authorities to this question has been
unequivical; in the balance the value of chlorination for public health pro-
tection far exceeds the possible adverse effects of TIIMs, and there is no
comparable substitute. The same may be concluded for sewage treatment
practice, particularly where discharge is to groundwater basins as discussed
below. The central point of the forgoing is that for most applications
sewage discharges must be disinfected for a satisfactory level of public
health protection.
Now, a brief commentary on the question, should chlorine be used as the
disinfectant? This question arises primarily because the addition of chlorine
for sewage disinfection produces some biotoxicity in effluents (14,11). Bio-
toxicity has two different kinds of impact depending upon whether effluents
are discharged to surface waters or to groundwaters. Regarding surface water
the concern is about impact on fish ecology. With groundwater the potential
impact is a health effect. This biotoxicity increases with increaseu levels
of chlorine addition. Recent work by the Sanitary Engineering Section of the
California Department of Health Services indicates that these Liotoxic effects
can be minimized by well engineered sewage chlorination facilities - rapid mix
of the chlorine and plug flow contact basins. Also SC>2 dechlorination follow-
ing chlorination will remove from effluents chlorine-induced biotoxicity to
aquatic life (22). If the choice must be between no disinfection or chlori-
nation, chlorination it must be.
As to the suitability of other means of disinfection, ozone or ultra-
violet light, questions must be answered concerning effectiveness, energy and
dollar costs, practicability, and perhaps others. Papers scheduled for later
in this conference should provide some insights into these issues. My pre-
diction is that chlorination of sewage effluents will be practiced as the
primary method of disinfection for many more years.
Considering the scope of this conference the subject of recreational
waters requires more extensive discussion. It is emphasized, however, that
even for this category of affected environment there is no doubt that dis-
infection is necessary for protection of bathers. The issue instead is about
the nature and extent of illness attributable to recreation in contaminated
waters, which organism or organisms should be used for monitoring, the corre-
lation of these two factors, what should be limiting values for an acceptable
level of risk, and even whether or not any microbiological limits should be
set.
Authoritative leadership concerning public health control of natural
bathing places (as contrasted to artifical swimming pools) has been provided
by two series of publications. The first of these were a set of ten reports
from 1926 to 1957 of the "Joint Committee on Bathing Places" of the Conference
of State Sanitary Engineers and the Engineering and Sanitation Section of the
American Public Health Association. The tenth Edition - 1957 (12) reports
15
-------
efforts in 1921, 1939 and 1955 to secure authentic information on reported
cases of illness attributable to bathing places. Regarding the 1955 inquiry
the committee states "It is striking that the returns from 45 states and one
territory stated that they could report no authoritative cases of illness
attributable to swimming pools and bathing places... Until new developments
take place to warrant different conclusions, the summary of replies to earlier
questionnaires and the recent survey of data obtained from state health
departments, considered in the light of known epidemiological evidence, leaves
this committee unconvinced that bathing places are a major public health
problem even though bathing place sanitation because of the health consider-
ations involved should be under careful surveillance of the public health
authorities and proper sanitary control of bathing places should be exercised
... It is realized that new epidemiologic evidence may come forth in the
future. It is agreed that common sense public health programs must recognize
that bathing in polluted water is a potential danger, that unsanitary con-
ditions surrounding public bathing places are a hazard, and that common
decency as well as health considerations dictates that reasonable steps should
be taken to secure bathing in clean environments..."
This report notes there is a wide divergence of opinion as to standards
of acceptable bacteriologic quality for outdoor bathing places in streams,
rivers, lakes and tidal waters. It emphasizes that final classification of
bathing waters should depend largely upon sanitary survey information, and
that bacteriologic analyses should be used as a guide. Further, pollution
may be present in many waters where treatment of sewage removes visable evi-
dence of sewage but does not eliminate dangerous concentrations of bacteria.
The second series of reports are three documents developed under Federal
auspices. These three documents have served as a basis for water quality
standards for the Federal regulatory water pollution control program: Water
Quality Criteria - 1963 (17), Water Quality Criteria - 1972 (7), and Quality
Criteria For Water - 1976 (24).
Selected excerpts from the 1968 Report are as follows: "The establish-
ment of public health requirements for the protection of the primary contact
recreation users has been a major problem for the sub-committee. Moreover,
in recommending specific water quality criteria the sub-committee is faced
with a sharp dilemma - that of balancing reasonable safeguards for the public
health against possible undue restrictions on the availability of waters for
contact recreation. The problem is further complicated by the inadequacy of
studies correlating epidemiological data on waterborne diseases with degrees
of pollution in recreational waters... There is an urgent need for research
to refine correlations of various indicator organisms including fecal
coliforms to waterborne disease. The sub-committee feels that the Public
Health Service's three epidemiological studies on bathing water quality and
health are the only base available for setting criteria. These studies were
far from definitive and were conducted before the acceptance of the fecal
coliform as a more realistic measure of health hazard... The sub-committee
recognizes that localized bacterial standards may be justified, if based on
sufficient experience, sanitary surveys, or other control in monitoring
16
-------
systems..." In conclusion this report recommended that "...the fecal coliform
content of primary contact waters shall not exceed a log mean of 200/100 ml,
nor shall more than 10 percent of total samples during any 30-day period
exceed 400/100 ml."
Selected excerpts from the 1972 Report are as follows: "... All recre-
ational waters should be sufficiently free of pathogenic bacteria so as not to
pose hazards to health through infection. This is a particularly important
requirement for planned bathing in swimming areas. There have been several
attempts to determine specific hazards to health from swimming in sewage
contaminated water. Three related studies have been conducted in this country
demonstrating that an appreciably higher overall illness incidence may be
expected among swimmers than among nonswimmers (24). In evaluating micro-
biological indicators of recreational water quality it should be remembered
that many of the diseases that seem to be causally related to swimming and
bathing in polluted water are not enteric diseases or are not caused by
enteric organisms. Hence, the presence of fecal coliform bacteria in recre-
ational waters is less meaningful than in drinking water... When used to
supplement other evaluative measurements the fecal coliform index may be of
value in determining the sanitary quality of recreational water intended for
bathing and swimming. The index is a measure of the sanitary cleanliness of
the water and may denote the possible presence of untreated or inadequately
treated human waste but it is an index that should be used only in conjunction
with other evaluative parameters of water quality such as sanitary surveys..."
In conclusion this Report states "No specific recommendation is made concern-
ing the presence or concentrations of microorganisms in bathing water because
of the paucity of valid epidemiological data."
The 1976 Report includes the following: "...Pollution of aquatic systems
by the excreta of warm blooded animals creates public health problems for man
and animals... The number of fecal coliforms present is indicative of the
degree of health risks associated with using the water for drinking, swimming,
or shellfish harvesting. Arguments against the use of fecal coliform bacteria
to define swimming quality in water have noted the paucity of epidemiological
evidence linking fecal coliform levels in bathing waters and the incidence of
disease (15,16). The lack of epidemiological correlation between fecal
coliform levels in coastal swimming waters and the incidence of disease may
not have validity in fresh waters and it does not take into account non-
reported diseases that may develop as an unrecognized result of swimming in
polluted waters. Epidemiological evidence is but one consideration in set-
ting microbiological criteria. The presence of fecal coliform bacteria
indicates degradation of water quality and a relative risk of disease trans-
mission." In conclusion, this Report states that evaluation of microbio-
logical suitability of marine and fresh waters should be based on fecal
coliform levels, and reiterates the 200/400 fecal coliform limit recommended
in the 1968 Report.
In the aggregate these two series of documents express the need for
protecting water quality in natural bathing places, support the use of a
coliform-fecal coliform index of water quality, refer to four epidemiological
17
-------
studies, note the paucity of evidence linking fecal coliform levels in bath-
ing water and the incidence of bathing-associated disease, and indicate the
need for more definitive epidemiological data. A careful review of the
investigations of Moore (13,14) and those of the Public Health Service (23)
lead to a conclusion that these epidemiological studies are flawed in pro-
cedural methodology and the resulting conclusions have limited significance.
A much sounder but still limited epidemiological investigation has now been
made by Cabelli and Associates (2,3). This work represents a three-year
(1973-75) study of epidemiological-microbiological study conducted at New York
City beaches as part of the U.S. Environmental Protection Agency (EPA) pro-
gram to develop health effects-recreational water quality criteria. Symptom-
atology rates among swimmers relative to non-swimming but beach-going controls
at a barely acceptable beach and a relatively unpolluted beach were examined.
It was observed that the symptom rates categorized as gastrointestinal, respi-
ratory, "other" and "disabling" were higher among swimmers than non-swimmers.
The rate of G.I. symptoms was significantly higher among swimmers relative to
non-swimmers at the barely acceptable but not the relatively unpolluted beach.
I assume Cabelli x
-------
Mississippi River below Dubuque, Iowa (21). The workshop proceedings conclude
with the following comments: "... Swimming per se carries with it an in-
creased risk of infections and irritations of the skin, ears, nose, and upper
respiratory tract. This risk appears to be infrequently associated with
pollution of the bathing waters with human or animal fecal waste... Except
under conditions of heavy contamination with human waste or during epidemic
conditions among the population whose waste reach the bathing waters, the risk
of contracting any of the severe, well recognized, well defined enteric
diseases such as salmonellosis, infectious hepatitis, poliomyelitis, typhoid
fever, etc. is minimal. Sporatic swimming-associated cases of these diseases
possibly do occur. Even with moderately polluted waters, there is a signif-
icant risk of contracting a gastroenteritis which appears to be acute in its
onset but benign in its course..."
The most recent report of significance is that of a National Research
Council Committee (6) which deals solely with the subject of microbiological
measures of recreational water quality. In summary, this Report states,
"in essence, the curreuc recreational water quality criterion is an indicator
system for water that is contaminated by the feces of warm blooded animals.
It is helpful only in the prediction of health hazards of recreational water
where the fecal-oral route of transmission is involved. Fecal coliform tests
detect mostly E. coli, which is not consistently pathogenic..." This Report
notes also that fecal coliform tests detect organisms such as Klebsiella,
Enterobacter, and Citrobacter, whose precise health significance remains to
be resolved. Further, the fecal coliform test appears to be of little, if
any, significance in the control of the many external ear, eye, and skin
infections that can be traced to contact with contaminated water... The
fecal coliform test is a reasonable indicator system for Shigella spp.,
Salmonella typhosa, _S_. typhimurium, ji. coli, and other unidentified agents of
the varied gastrointestinal symptomotology that appear to be associated with
ingestion of swimming water. In a less direct manner, fecal coliforms may
indicate the presence of viruses that could be transmitted by the fecal-oral
route. There is evidence, however, that viruses may occur where fecal coli-
form counts are low or are not detectable.
Further this Report states, "One important epidemiological factor in
fecal-oral transmission, which has not been adequately addressed is the volume
of water unintentionally taken into the digestive tract by a swimmer or an
individual that had been immersed in water at recreational sites. The volume
of ingested water must be important in determining the numbers or dose of a
pathogenic agent to which an individual has been exposed. The intake may vary
with the individual's age, level of swimming proficiency, time of exposure,
quality or salinity of water, etc. These variables have not yet been measured
and it is unknown how they relate to the threshold dose for enteric infection
... The fecal coliform criterion remains a reasonable predictor for gastro-
enteric illness and possibly infections from non-coliform agents. Evidence
indicates an increased risk of gastrointestinal illness when the fecal coli-
form criterion is exceeded."
19
-------
"No criterion or guideline that is based on a single microbial indicator
species will serve as a measure of the health risk from the wide variety of
diseases that can be contracted from recreational water. For any indicator
used, however, sampling ancj laboratory testing must proceed in conjunction
with epidemiologic surveillance, public health engineering, sanitary surveys,
and monitoring..." "Absolute protection of the public health by relying on a
single water quality criterion is not feasible and, in fact, is not pos-
sible." This Committee concludes that "The fecal coliform test is acceptable
for protecting the public health until additional epidemiologic data, im-
proved laboratory procedures, and a better understanding of aquatic microbial
ecology are obtained... Its use certainly is better than abandoning micro-
biological criteria altogether. The Committee recommends that the fecal
coliform tests be replaced eventually by a test or series of tests that di-
rectly assess the health hazard posed by the presence of pathogens in
recreational waters."
To recapitulate, sewage effluents discharged to recreational waters in
significant concentrations will cause disease. How much disease, what are
significant concentrations, and what are infectious doses, are not defi-
nitely established. A search continues for more suitable parameters. Until
these are established, the fecal coliform test is considered the best avail-
able. Finally, on the basis of fecal coliforms, what should be limiting
values? The answer to this last question depends upon a value judgment
relating to acceptable risk. This value judgment cannot intelligently be
made at this time because the risk cannot be measured with sufficient ac-
curacy.
Epidemiolog}? is the tool that is used, essentially the only one avail-
able. Unfortunately it is a blunt instrument.
Two things seem certain. One, very little epidemic disease has been
associated with swimming in sewage polluted waters in this country, only
acute disease has been detected by the retrospective epidemiology, and even
then, only where the pollution has been gross. Two, disease unquestionably
results from swimming in polluted water, not only gastrointestinal illness,
but infections of the skin, eyes, ears, and upper respiratory tract; but
apparently not at epidemic levels. Much of this disease is subacute, is not
seen by physicians, and can best (perhaps only) be measured by prospective
epidemiology. Cabelli (4), in commenting on this phenomenon states, "It
would appear that data derived from published care and outbreak reports
markedly understate the rates of recreational waterborne disease." It is
also likely that some of the non-G.I. illness is more directly associated with
the act of swimming than from pollution in the water.
Considering the fact that this subject has been discussed in public
health circles for over 60 years, astonishingly little progress has been made
in establishing a sound basis for standards. On the other hand, the crudely
developed "standards" in use are probably not too far from the mark. They
have a sort of common sense ring to them. Furthermore, when one considers
the billions of construction grant dollars spent annually, some to meet
20
-------
arbitrary requirements for secondary treatment, and the big-ticket cost of
operating the facilities, it is tempting to wonder why so much issue is taken
with the 200/400 coliform limits. The cost for meeting these limits is
"peanuts" compared with the rest of the bill, and it buys a certain, though
unquantifiable, amount of public health protection.
In conclusion, do disinfect sewage effluents to protect recreational
water.
21
-------
LITERATURE CITED
1. Bryan, J.A., Lehmann, J.D., Setiady, E.F., and Hatch, M.H., 1974. An
outbreak of hepatitis A associated with recreational lake water. Am.
Jour. Epidemiol. 99:145.
2. Cabelli, V.J., Dufour, A.P., Levin, M.A., and Haberman, P.W. 1976. The
impact of pollution on marine bathing beaches. Am. Soc. Limnol.
Oceanogr. Spec. Symp, 2:424.
3. Cabelli, V.J., Dufour, A.P-, Levin, M.A., McCabe, L.J. and Haberman, P.W.
1979. Relationship of microbial indicators to health effects at marine
bathing beaches. Amer. Jour. Public Health 69:7:690.
4. Cabelli, V.J., 1978. Swimming associated disease outbreaks. J_. Water
Pollu. Control Fed. 50:6:1374.
5. California Administrative Code, Title 22, Environmental Health, Chap. 3.
Reclamation Criteria. Sect. 60301-60355.
6. Committee on Microbiological Standards for Recreational Water, 1979.
Microbiological Measures of Recreational Water Quality. National
Research Council. Washington, D.C.
7. Committee on Water Quality Criteria, 1973. Water Quality Criteria, 1972.
Environmental Studies Board. Nat. Acad. Sci, Nat. Acad. Eng.
Washington, D.C.
8. Dennis, J.M., 1959. 1955-56 infectious hepatitis epidemic in Delhi,
India. J.A.W.W.A. 51:10:288.
9. Flynn, M.J., and Thistlewayte, D.K.B., 1964. Sewage pollution and sea
bathing. Second Nat'1 Con, on Water Pollution Res.
10. Garrison, W.E., Nellor, M.H., and Baird, R.B., 1979. A study on the
health aspects of groundwater recharge in Southern California. County
Sanitation Districts of Los Angeles County.
11. Grabow, W.O.K., Prozesky, O.W., and Smith, L.S., 1974. Review paper.
Drug resistant coliforms call for review of water quality standards.
Water Research, 8:1.
12. Joint Committee on Bathing Places of the CSSE and the Engineering and
Sanitation Section of the APHA, 1957. Recommended practice for design,
equipment and operation of swimming pools and other public bathing
places. Tenth Edition. APHA. N.Y.
13. Ktsanes, V.K., Anderson, A.C., and Diem, J.E., 1981. Health effects of
swimming at Lake Pontchartrain at New Orleans. Project Summary. EPA -
600/S1-81-027. U.S. EPA.
22
-------
14. McCarty, P.L., Reinhard, M., Graydon, J., Schreiner, J., Sutherland, K.,
Everhart, T., and Argo, D.G., 1930. Advanced treatment for wastewater
reclamation at Water Factory 21. Technical Report No. 236.
15. Moore, B. , 1959. Sewage contamination of coastal bathing water in
England and Wales. Jour. Hyg. 57:435.
16. Moore, B., 1971. The health hazards of pollution in microbial aspects
of pollution. Sykes and Skinner, eds. Academ. Press. London pp. 11-32.
17. National Technical Advisory Committee to the Secretary of the Interior,
1968. Water Quality Criteria. Federal Water Pollution Control Admin-
istration. U.S. Government Printing Office, Washington, B.C.
18. Northrop, R. L. , Brenniman, G.R., Byington, R. B. , Hesse, C. S. , and
Rosenberg, S.H. , 1981. Recreational water quality and health. Project
Summary. EPA - 600/S1-81-059. U.S. EPA.
19. Pipes, W.O., ed. 1978. Water quality and health significance of bacte-
rial indicators of pollution. Proceedings of a National Science
Foundation Workshop. Drexel University, Philadelphia.
20. Portnoy, B.L. , Mackowiak, P. A., and Karaway, C.T., 1975. Oyster
associated hepatitis. Failure of shellfish certification programs to
prevent outbreaks. Jour. Amer. Med. Assoc. 233:1065.
21. Rosenberg, M.L., Hazlet, K. K. , Schaefer, U. , Wells, J.G. , and Pruneda,
R.C., 1976. Shigellosis from swimming. Jour. Am. Med. Assoc.,
236:1849.
22. Sepp, E. , Bao, P. , 1980. Design optimization of the chlorination
process. California Department of Health Services, Sanitary Engineering
Section, Berkeley, CA.
23. Smith, R. S. , and Woolsey, T.D. , 1961. Bathing water quality and public
health. III. Coastal waters. U.S. Public Health Service, Cincinnati,
OH.
24. Stevenson, A. H. , 1953. Studies of bathing water quality and health.
Amer. Jour. Public Health, 43:529.
25. U.S. EPA, 1976. Quality criteria for water. U.S. Government Printing
Office, Washington, D.C.
23
-------
3. INFECTIVE DOSE OF WATERSORNE PATHOGENS
Elmer W. Akin, Chief
Microbiology Branch
Toxicology & Microbiology Division
Health Effects Research Laboratory
Cincinnati, Ohio
ABSTRACT
Infective dose studies with a variety of enteric organisms have been
conducted over the past 30 years in human volunteers. The widest dose range
required to produce a response was found with the bacterial agents. Salmonella
spp. required the largest dose with the ingestion of 1(P to 10° cells needed to
produce a 50 percent attack rate. In contrast, three species of Shigella
produced illness in a significant percent of dosed subjects with 10 to 100
cells. Protozoan infections have been produced with Entamoeba coli and Giardia
lamblia dosed in gelatin capsules at the level of 1 to 10 cysts. Enteric
viruses have produced infection at low dosage levels via oral ingestion,
inhalation, and conjunctival exposure. These data produced with healthy human
subjects show that members of all three categories of enteric pathogens can
produce infection and/or illness at concentrations found in wastewater.
INTRODUCTION
The return to the soil of chemical nutrients and moisture existing in
wastewater for more productive cultivation of desirable plant life is an
ancient custom that still seems appropriate for our time. Wastewater that has
received little or no treatment has the advantage of less cost and greater
nutrient value, but it also has a greater number of pathogens and therein lies
a health concern.
A wide variety of enteric pathogens including viruses, bacteria and
parasites is known to occur in all community-derived wastewaters. The
concentration of pathogens, especially viruses, is debated primarily due to
limitations of currently available recovery techniques. However, the ex-
istence of any concentration of pathogens in the environment does not
necessarily pose a health hazard. A mechanism of transmission back to man must
exist and in order to warrant exposure-control expenditures, e.g., dis-
infection, the occurrence must be more likely than a rare event.
Three modes of transmission are perceived as the most likely potential
routes for reintroducing wastewater pathogens to man: (a) direct exposure to
wastewater aerosols, (b) ingestion of drinking water contaminated with waste-
water seepage or runoff, and (c) ingestion of contaminated animal or plant
foods produced on wastewater-amended soil. The hazard evaluation of these
potential exposure routes may be approached basically in two ways: (a) by
24
-------
epidemiological studies of exposed populations and (b) by a modeling approach
which seeks to determine the likelihood of infection by considering a number of
human and environmental variables. Considerable effort has been applied to the
former approach and some of this work will be discussed in detail in other
papers in these Proceedings.
A major variable important to utilization of the second approach is the
infective dose of an agent. Information must be available on infection rates
in populations exposed to diminishing numbers of a specific microorganism. The
often used phrase "minimum infective dose" is a misnomer in that it does not
acknowledge the concept of a changing probability of infection with exposure to
varying dosage levels. For infective dose data to have meaning in this
context, subjects must be exposed to multiple concentrations of an agent so
that dose response curves may be developed. Several studies of this type have
been conducted with animal models and human volunteers using a variety of
enteric microorganisms. This review of the data will be limited to the human
volunteer studies since they will pose fewer interpretative questions and
thereby be of more practical significance to the subject of this symposium.
In considering host responses to microorganisms, a distinction is to be
made between the two most common end points measured: infection and illness.
Infection may be defined as multiplication of a microbial agent within a host
with or without the production of disease. The occurrence of illness may be
determined by the manifestation of a single pathogenic effect or a group of
symptoms normally associated with an etiological agent. Both end points have
been utilized to study host-parasite interactions in human populations.
However, investigators, institutions, and the public in general are becoming
increasingly reluctant to support studies designed to produce pathogenic
effects in humans. Therefore, the more recent data, obtained with viruses,
have determined the asymptomatic infectivity end point as indicated by fecal
shedding of the test organism or by the detection of a specific antibody
response. Of course, the asymptomatic infective dose of a pathogen for one
individual may produce disease in another. However, it is generally assumed
that severity of response is proportional to the degree of exposure and in
feeding studies that administer relatively low doses, to determine the minimum
infectivity level, adverse effects in healthy subjects are rare. Nonetheless,
infective dose is an important parameter in hazard evaluations because
infected persons may transmit viable organisms to others who may experience
clinical disease. As added precautions, virus studies are normally conducted
with vaccine or very mildly-pathogenic strains and in studies with bacteria or
protozoa the termination of infection can usually be insured by the adminis-
tration of antibiotics or other antimicrobial drugs.
25
-------
DOSE RESPONSE TO ENTERIC BACTERIA
The organisms of greatest concern in exposure to wastewater-contaminated
environments are the enteric bacteria and viruses and the intestinal para-
sites. In the United States, Salmonella and Shigella species are essentially
the only enteric bacteria that have a recognized prevalence level sufficient to
be of concern. Hornick et al. (9) have conducted dose-response studies in
healthy adult male volunteers with Salmonella typhi, the etiological agent of
typhoid fever. This study determined the number of organisms required to
produce an illness end point. A positive response was determined when toxic
symptoms of typhoid fever occurred, e.g., headache, malaise, anorexia, and
temperature of 103°F for 24 to 36 hours. At this point, the infection was
interrupted by the administration of an antibiotic. Table 1 shows the dose
response obtained with the oral administration of various densities of
organisms suspended in 30 ml of milk. No symptoms of typhoid fever were
observed in 14 volunteers who received 103 organisms. Half of the volunteers
who ingested 10' organisms became ill.
Table 1. Number of Salmonella typhi (Quailes Strain)
Organisms Required to Produce Typhoid Fever
in Healthy Adult Male Volunteers (9)
Number of Viable Dose Response
Organisms Administered" No. Ill/No. Challenged (%)
103 0/14 (0)
105 32/116 (28)
107 16/32 (50)
109 40/42 (95)
"Organisms suspended in 30 ml of milk
Similar studies were conducted by McCullough and Eisele (19,20) to
determine a salmonellosis end point for several other Salmonella strains and
species. Serial dilutions of the suspension of organisms were plated on
trypticase soy agar in duplicate for final bacterial count. Table 2 shows the
results of these studies and indicates that a wide variation in cell numbers
was required to produce illness. Six adult male volunteers ingested various
densities of the organisms suspended in a glass of eggnog. Subjects were
selected who had no indication of Salmonella infections, i.e., no organisms in
their stools and absence of high serum agglutination titers. The natural
course of the illness was followed without specific treatment except when
medically indicated. Symptoms generally included abdominal cramping, nausea,
diarrhea, and low grade fever. Two strains of S. meleagridis did not produce
illness in the 12 volunteers who received 5.5 million organisms. However, the
ingestion of 1.7 million cells of S^. bareilly produced illness in four of six
(67 percent) subjects. The lowest pathogenic dose observed with eight
organisms studied was 125,000 cells of S. bareilly. One of six subjects became
ill after ingesting this number of organisms.
26
-------
Table 2. Number of Salmonella Organisms Required to Produce
Clinical Illness in Healthy Adult Male Volunteers (19,20)
Dose x 1(T6
Organism
S.
S.
S.
S.
S.
S.
S.
S.
anatum, Strain I
anatum, Strain II
meleagridis, Strain I
meleagridis, Strain II
meleagridis, Strain III
bareilly
newport
darby
0%*
.26
24
5.5
5.5
1.5
6.4
15-20%
45
24
10
7 7
.13
.15
30-50%
.58
67
20
10
.70
1.4
15
65-85%
50
41
1.7
"Percentages are number of volunteers ill/total number dosed x
100; six volunteers normally exposed/dosage level
Shigella has been found to be a more virulent genus of enteric bacteria
than Salmonella. Symtoms of shigellosis include diarrheal stools containing
blood and mucus, abdominal cramps and high fever. A group comprised of federal
and university investigators has conducted a series of Shigella feeding
studies in inmates of a correction institution (7,16) Adult male volunteers
were fed various types and strains of Shigella organisms suspended in milk.
Inocula were prepared by diluting 24-hour agar plate cultures and the cell
number confirmed by making pour plates of the inocula before and after each
experiment was conducted. Illness was terminated in the volunteers by the
administration of an antibiotic.
Results of studies with S. flexneri 2a and two strains of S^. dysenteriae
1 are shown in Table 3. All three organisms produced illness in a significant
number of subjects fed 200 cells. One of 10 subjects fed 10 cells of S^.
dysenteriae I, strain M131 became ill. This strain had been responsible for a
dysentery pandemic in Central America during 1968-1970. The virulence and
multi-drug resistance of this organism actually led to work on a live
immunizing agent for this disease. Although not reported in detail, these
investigators mentioned in an additional report the production of disease in a
significant percentage of adult volunteers fed 10 to 100 viable cells of £.
flexneri 2a and EL sonnei as well as S^. dysenteriae 1 (8)
A limited amount of dose-response data has been reported for three
additional enteric bacteria that, on recent occasions, have been associated
with waterborne disease in the United States and could be of health concern in
wastewater exposure (4,5). Yersinia enterocolitica can produce in man an
enteritis similar to salmonellosis. Szita e_t al. (25) have reported that
illness was produced in a single volunteer fed 3.5 x 10^ cells of the organism.
Enteritis also has been recently associated with the ingestion of Campy-
lobacter jejuni. One investigator experienced abdominal cramps and mild
diarrhea after ingesting 500 organisms, in 180 ml of milk, of a strain that had
27
-------
been isolated from a milk-borne outbreak (23). Pathogenic and toxigenic
strains of the common intestinal bacterium, Escherichia coli , have been
associated with waterborne enteritis. Koya et_ al. (14) studied the illness
response of E. coli enterotoxigenic strain 0-111 B4 in four male volunteers.
No illness resulted in the subject fed 2.7 x 107 organisms. Mild diarrhea,
abdominal pain, and fever resulted in three subjects fed 5 x 107 to 10y
organisms. DuPont e_t aJL. (6) studied two nontoxigenic invasive strains of E.
coli and observed illness in eight of 13 adult volunteers fed 10° cells. No
clinical disease was apparent on ingestion of 10" cells unless an antacid was
also administered. An additional enteric bacterium Vibrio cholerae, although
no longer a significant pathogen in this country, remains as a major
etiological agent in the developing countries. Studies with the Inaba 569B
strain of V. cholerae have indicated that a dose of 10° cells is required to
induce diarrhea in the absence of concomitant antacid administration (2). None
of these studies except Koya e_t al. (14) and DuPont e_t _al. (6) were designed to
determine a minimal dose response. Of the enteric bacteria that have received
considerable study, Shigella has been found to be the only genus that produces
illness in healthy adults at relatively low exposure levels, i.e., < 200
organisms. However, the very limited data from one exposure to 500 cells of £.
j e j un i indicate that this genus also may have a low infective dose.
Table 3. Response in Healthy Adult Male Volunteers to
Various Doses of Virulent Strains of Shigella
s.
s.
s.
Organism
. flexneri 2a
dysenteriae 1
(Strain A 1)
dysenteriae 1
(Strain M 131)
Dose
180
5,000
10,000
200
10,000
10
200
2,000
10,000
No. Ill/No. Fed
8/36
28/49
52/88
1/4
2/6
1/10
2/4
7/10
5/6
% 111 Ref.
22 7
57
59
25 16
33
10 16
50
70
83 .
DOSE RESPONSE TO ANIMAL PARASITES
Historically, amebiasis caused by the protozoan Entamoeba histolytica has
been the most important environmentally transmitted disease of animal parasite
origin in the United States. Most infections are asymptomatic; however, its
pathogenicity is well documented and the occurrence of cysts in the stool is
always of concern. Rendtorff was interested in determining the dose-response
of this organism in nan, but felt it improper to purposely expose prisoner
volunteers to this pathogen. He chose instead to determine the infective dose
of a non-pathogenic amoebae: Entamoeba coli (21). Adult male volunteers that
28
-------
showed no amoebic infections on stool examinations were selected for this
study. The volunteers were individually housed under specifically controlled
environmental conditions during the 10-week study period to minimize extra-
neous infections. The cyst inoculum was obtained from volunteer donors and was
separated from debris and other organisms by a flotation procedure and
micromanipulation. Individual dosage levels were obtained by micromanipulator
isolation of the cyst, a very accurate technique for obtaining low-dose
numbers. Appropriate volumes of the cyst suspensions were placed in gelatin
capsules and swallowed with 120 to 180 ml of tap water.
Table 4 shows the results of this study Infection, as determined by
repeated fecal shedding of cysts, was achieved in one of eight volunteers who
supposedly ingested only one cyst. The increasing infectivity rate with doses
of 10 (30 percent) and 100 (50 percent) cysts lends support to the authenticity
of the single positive response (12.5 percent) on ingestion of one cyst. In
addition the rigorous experimental design that included isolation of the
volunteers, negative controls, and a highly sensitive cyst counting procedure
tends to lend credibility to the remarkable finding.
The reluctance of Rendtorff to conduct human feeding studies with E.
histolytica was apparently not shared by Beaver et al. (1) They obtained
cysts from an asymptomatic donor and fed adult volunteers a single dose ranging
from 2,000 to 1 million cysts. Unfortunately, the lowest dose used in this
investigation produced infection in all volunteers (42 of 42) which precluded
the determination of a minimal infective dose (Table 4). It should be noted
that the authors concluded that no disease symptoms could be attributed to E.
histolytica infections in these subjects.
Table 4. Response in Adult Males to Various Doses of Amoebic Cysts
Number of No. Infected/ % Method
Cyst Type Cysts Given No. Fed Infected of Adm. Ref .
E . coli
1
10
100
1/8
3/10
2/4
12.5
30
50
gelatin
capsule
21
E. histolytica
G . 1 amb 1 i a
2000 to
4000
1
10
100
42/42
0/5
2/2
2/2
100
0
100
100
beverage
suspension
gelatin
capsule
1
22
Although Giardia lamblia was not the important pathogen in the 1950s that
it became in the 1970s, Rendtorff included this protozoan in his study of the
E. coli model of amoebic infection (22). This early work represents the only
human dose-response study conducted to date with this organism. Volunteers
were fed either Giardia cysts alone or Giardia plus E_. coli cysts in gelatin
29
-------
capsules. The inocula counts were determined by micromanipulator isolation as
stated above. A single volunteer received one of three dosage levels or
Giardia alone or the same dosage levels of Giardia and E. coli cysts. Table 4
shows the results of this study and indicates that a dose of 10 cysts or less
is sufficient to produce infection in a high percentage of exposed susceptible
individuals. Even though none of the five volunteers was shown to be infected
on ingestion of a single cyst, it is reasonable to assume, based on the
infection of two of two with ten cysts, that with a larger number of exposed
subjects, infection of a significant percentage would have occurred on
ingestion of one viable cyst. Perhaps the apparently more virulent strain
occurring in the United States today would be more infectious. However, the
direct correlation of parasite virulence (as indicated by disease frequency or
severity) with the exposure level required for a host response has not been
adequately demonstrated. Interestingly, these data indicate that these animal
parasites have about the same degree of infectivity as Shigella bacteria.
DOSE RESPONSE TO ENTEROVIRUSES BY ORAL INGESTION
The last group of enteric pathogens to be considered, viruses, represents
a somewhat more difficult agent to be safely studied in humans. Since viruses
are obligate intracellular parasites, even the infectivity end-point requires
the occurrence of the somewhat uncontrollable process of cell destruction,
i.e., virucidal drugs are not available to interrupt the infectious process.
The development of avirulent strains of polioviruses in the 1950s provided
opportunity to conduct dose response studies with live vaccine strains with
minimal risk of adverse responses. It also allowed such studies to be
conducted with a subset of the population that appears to be more susceptible
to natural infections, i.e., infants and children.
Koprowski and his colleagues at Lederle Laboratories conducted the first
reported dose response studies with attenuated strains of polioviruses.
Strain SM of poliovirus type 1 had been attenuated by rodent adaptation
followed by successive passages of the virus in chick embryo and monkey kidney
tissue culture. The virus was non-pathogenic for monkeys on intracerebral
injection. As a component of the field trials of this potential live vaccine
virus, Koprowski et al. (13) conducted a dose response study in children at a
state institution. A total of nine children, who showed no antibodies to type
1 poliovirus, were given various doses of the virus suspended in polyethylene
glycol 400 (0.5 ml) within a hard gelatin capsule. Each subject swallowed two
capsules consuming at the same time 8 ml of milk. The dosage was determined by
making 10-fold dilutions of a virus preparation titrated by the plaque
technique using monkey kidney cell monolayers. Infection was determined by
fecal shedding of the virus or by a specific antibody response.
The results of the study are shown in Table 5. Of course, the dose of 0.2
plaque-forming particle (PFP) is a dilution-determined average value and
indicated that a single PFP would not occur in most doses. The calculated dose
of 2 PFP produced virus shedding and an antibody response in two of three
subjects. The investigators, aware of the implications of this remarkable
30
-------
finding, suggested that the following factors should be clarified before
assuming that these data represented a practical occurrence (12): (a)
interference in virus titration by the diluent thereby giving an inaccurately
low value, (b) insensitive assay system that did not detect all the PFP that
would be infectious for the subjects, and (c) artifically high sensitivity due
to delivery of the encapsulated virus directly to susceptible cells in the
intestinal tract. Concurrent studies with an attenuated type 2 poliovirus
indicated that 300 units of this virus were required for infection (13).
However, titration of the rodent-adapted virus was performed in intact mice
yielding 50 percent mouse paralytic doses, a less sensitive quantitative
procedure.
Poliovirus vaccine feeding studies were subsequently performed by others
over the next few years (17). However, significant infection levels (> 50
percent) were not reported with doses <1Q3.5 tissue culture infective dose 50
percent (TCID5Q) until the studies of Katz and Plotkin (11) in the mid-sixties.
These investigators studied the dose response of poliovirus type 3, Fox strain
in 22 premature infants in the nurseries of a general hospital. Within the
first 48 hours of life, each infant was given a low dosage of the virus,
suspended in 5 ml of Hanks' solution, directly into the stomach by gavage using
a rubber oro-gastric tube. The tube was flushed with 10 ml of saline solution
before removal. Infectivity was determined by fecal shedding of the specific
virus type administered.
Table 5. Response of Infants and Children to Low
Doses of Poliovirus Live Vaccine
Dose
PFP"
0.2
2
20
1
2.5
10
16
50
90
160
No. Infected
No. Fed
0/2
2/3
4/4
3/10
3/9
2/3
0/2
3/6
3/4
3/3
%
Infected
0
67
100
30
33
67
0
50
75
100
Method Virus
of Adm. Type (Strain)
Gelatin 1 (SM)
Capsule
Gavage 3 (Fox)
Tube
Aqueous 1 (Sabin)
Susp .
Ref .
13
11
18
plaque-forming particles
tissue culture infective dose
Results of this study are also shown in Table 5. At the lowest dose
administered, 1 TCID5Q, three of 10 subjects were infected. The TCID5Q
quantitation procedure is thought to be less precise than the plaque procedure.
31
-------
However, the investigators demonstrated a high degree of accuracy in dose
titration. The lower dosage levels (2.5 and 1 TCID50) were titered before
administration to the infants by adding 0.1 ml aliquots of each inoculum to 50
cell culture tubes. The titration results yielded positive findings within 0.5
tubes of the statistically predicted number for each titer. A line fitted to
a log probability plot of the dose-response data indicated that the 50 percent
infective dose for the subjects was 4 TCID5Q. These findings and the authors'
conclusion that "a dose of any pathogenic virus sufficient to infect tissue
culture would also be infectious for man" have been extensively cited.
Recent Virus Dose Response Studies
In an effort to obtain virus infectivity data by more normal exposure
routes, the U.S. Environmental Protection Agency (EPA) has funded an addi-
tional infant poliovirus feeding study at the University of Wisconsin (18).
Infant patients of a private pediatric practice were recruited to receive a
reduced dose of the commercial poliovaccine type 1 (Sabin) two weeks prior to
the scheduled receipt of the full dosage. The commercial vaccine was diluted
in sterile distilled water to the desired dosage and administered in 0.5 ml
volumes to the oral cavity with a 1-ml syringe. Each infant was observed to
detect expectoration so as to insure that the entire dose had been swallowed.
Infection was determined by the shedding of virus in the stool within 10 days
post inoculation. Viruses isolated in the stool were identified as polio 1 by
specific neutralization test.
Thirty-two 2-month-old infants were fed doses of 7 to 280 TCID5Q °f virus.
Few infants were given the identical dose since experiments conducted at
different periods with freshly diluted virus gave slightly different titers.
Results obtained with multiple feedings of the same dose are shown in Table 5.
In addition to the two infants fed 16 TCID5Q, three more yielded negative
findings at doses of 42, 27, and 7 TCID50 (data not shown). Statistical
analysis of all the data yielded a 50 percent infective dose of 72 TCID5Q,
considerably higher than the findings of Koprowski and Katz and Plotkin.
An additional study has been supported by EPA to obtain data on the
infective dose in adults of a "wild" enteric virus. The virus, echovirus 12,
used in the study had been isolated from an 8-year-old girl with erythema
infectiosum (fifth disease). Previous volunteer studies had shown the virus to
be a very mild pathogen normally producing asymptomatic infections with
common-cold type symptoms occurring infrequently.
Healthy male students having no evidence of echovirus 12 infection were
recruited from local colleges. Selected subjects were isolated, 2 to a room,
from outside contact for 8 days, one day prior and 7 days after ingestion of the
virus. Appropriate dilutions of the purified virus were suspended in 100 ml of
distilled water. Volunteers ingested 100 ml of the virus suspension or 100 ml
of sterile water (negative control) under a double blind experimental design.
Neither the volunteers nor the investigative staff knew the contents of the
inoculum. Health status of dosed subjects was monitored twice a day by a nurse
and physician. Throat and rectal swabs were collected daily during the
32
-------
isolation period for virus assay. Blood specimens were collected on the day
prior to inoculation and on days 6 and 26 for echovirus 12 antibody tests.
Preliminary results from this study have recently been reported (24).
None of the subjects became ill. Table 6 shows the infection response. At the
lowest dose administered, 10 plaque-forming units (PFU) , an infection rate of
19 percent was observed. Subsequent to this report, additional data have been
obtained for a total of 108 subjects fed one of four concentrations of virus:
10, 30, 100, and 300 PFU. Fecal shedding of the virus was found to be a more
sensitive indicator of infection than humoral antibody response (Schiff,
personal communication).
Table 6. Infection Response in Healthy Young Adult
Male Volunteers to Various Oral Doses of
Echovirus 12 (24)
Dose
(PFU)"
10
30
100
No. Infected/No. Fed
6/32
2/7
14/21
% Infected
19
29
67
"Plaque-forming units
Statistical analyses of these data and the data of Minor (18) are shown in
Table 7 From these analyses, an estimate of the infective dose can be made at
exposure levels lower than can be practically obtained by experimentation. It
should be noted that viruses assume a Poisson distribution in very dilute
suspensions. At a mean virus concentration of one PFU/unit volume (dose), the
probability that a dosage volume will contain zero PFU is 37 percent.
Therefore, it becomes impossible to distinguish between a non-response in a
large percentage of the subjects due to administration of no virus versus
administration of sub-infective numbers, a major objective of a dose-response
study Under this condition, the experimental error may be greater than the
extrapolation error from a statistical analysis. Nonetheless, it seemed
appropriate to estimate the dose required to infect 1 percent of exposed
subjects. Table 7 shows that 20 TCID5Q (7-52) of polio type 1 and 0.4 PFU (<1-
2) of echovirus 12 would be required. The echovirus 12 data appear to be the
strongest data yet available indicating that the oral ingestion of a single
detectable unit of virus may be infectious for a certain portion of a
susceptible population.
33
-------
Table 7. Estimates of the Number of Cell-Culture-Infective Doses
of Two Enteric Viruses Required to Produce Infection
in Humans When Ingested Orally
Virus
polio 1
echo 12
%
Infected
50
10
1
50
10
1
Virus
Dose
TCID50
72
39
20
PFU
35
3
.4
95% Confidence Limits Ref.
55-93 18
24-63
7-52
21-64 24
-------
of this study is also shown in Table 8. At the two lowest dosage levels
administered, three of three subjects inoculated with five TCID5Q and one of
three inoculated with one TCID5Q were infected which indicated that the HID->0
approximated the TCID5Q. All infected subjects experienced lower respiratory
tract illness. These studies showed that aerosols in the 0.2 to 3.0 )jm range
penetrate into the lower respiratory tract and that cells here are highly
susceptible to at least these two enteric viruses.
Table 8. Response of Antibody-Free Volunteers to Small-Particle
Aerosol Inoculation of Enteric Viruses (3)
Virus
Coxsackie A21
Adenovirus 4
Inhaled
Dose
(TCID50*)
71
47
18
6
11
5
1
50% Human
Dose Response Infective Dose
No. Ill/No.
5/5
3/4
1/4
0/6
3/3
3/3
1/3
Exposed (%)
(100)
(75)
(25)
(0)
(100)
(100)
(33)
(TCID50)
28
~ 1
"Tissue culture infective dose 50%
The same investigators also showed that larger diameter aerosols (ap-
proximately 15 Mm) would transport these viruses to the upper respiratory tract
and produce infection there (3). A similar infective dose, i.e. , approximately
30 TCID^Q, was obtained with coxsackievirus A21 suspended in particles of
either size range. The lowest infective dose was obtained by administering the
virus via nasal drops. The instillation of 0.25 ml of diluted virus
suspensions into the nostrils of 14 volunteers produced an HID^Q of six TCID5Q-
These data suggest that the nasal mucosa is the preferred respiratory site for
infection. The HID5Q for adenovirus suspended in large diameter aerosols was
not given. However, data were presented that showed 1000 TCID5Q produced
infection in six of six exposed volunteers.
Extensive minimal-infective-dose studies of enteric viruses via con-
junctival inoculation have not been reported. However, in a broader study of
adenovirus infections, Kasel, et al. (10) produced conjunctivitis in two
volunteers who were swabbed in the lower conjunctival sac with approximately
0.1 ml of media containing 30 and 300 TCID5Q of adenovirus 26. Rectal shedding
of the virus was detected a few days after eye exposure.
DISCUSSION AND CONCLUSION
Man's relationship with microorganisms is dynamic and generally unpre-
dictable in healthy hosts. Organisms are classified as human pathogens with the
35
-------
understanding that a disease outcome of a host-microorganism interaction is
dependent upon many variables. This understanding has been given greater
expression recently with the increasing use of the term "opportunistic
pathogens". Host susceptibility is influenced by age, sex, nutritional and
health status. Pathogenicity and virulence of the organism is altered by
mutation, selection and gene transfer. Natural or man-made environmental
factors may determine the route of exposure, persistence of the organism, and
the level of exposure. In the study of dose response, an infectious end point
as indicated by in vivo multiplication of the agent and/or antibody response is
perhaps the more reproducible outcome of exposure to enteric organisms since
many of these variables affect primarily the occurrence of a pathogenic result.
Enteric pathogens are obviously able to infect the gastrointestinal
system. Therefore, they must possess characteristics that allow them to
compete and multiply in this environment. Theoretically, one particle should
be capable of initiating infection. However, for this to occur- a viable
particle must be transported to the site of multiplication. For the bacterial
and parasitic pathogens, this is normally the lumen of the small intestine.
Stomach passage actually activates some of the parasitic agents. However,
survival in this acidic and enzymatic environment may be the major limitation
of bacterial colonization of the intestine. It has been shown that an altered
gastric function produced by buffering agents or disease process can increase
susceptibility to enteric bacterial infection (6,15). Infection by enteric
viruses is a completely different process from that of other enteric pathogens.
Before a virus can enter a cell and establish infection, specific attachment to
the cell surface must take place. The existence of complementary receptor
sites on the host cell and the virion surface is the major factor in host
specificity to these agents. Apparently a limited number of cell types exist
in the intact host that are receptive to virus attachment. Therefore,
infection appears to be mainly controlled by the probability of the contact-
adsorption process occurring.
From the data of Minor, e_t al. (18) one may conclude that there is a
probability of 1:100 that on the ingestion of 20 units of poliovirus 1, at least
one particle will make specific attachment to and replicate within a
susceptible cell of a human infant. The data of Stefanovic, et al. (24) are
more remarkable in that they indicate that the ingestion of a single~detectable
unit of echovirus 12 could produce the same probability of infection in young
adults. The observation of an even more likely infectious outcome (1:2) with
the direct nasal application of one unit of coxsackievirus A21 (3) is
consistent with the view that cell contact is the limiting factor. The shorter
distance from virus entry to susceptible cells would favor a lower infective
dose for infection of the nasopharynx by nose drops versus intestinal infection
via oral ingestion.
The studies reviewed here have clearly shown that specific enteric
organisms of all three classifications, i.e., bacteria, animal parasites, and
viruses, can produce infections at relatively low exposure levels. It should
be mentioned that these data, with the possible exception of Rendtorff's
Giardia work (21), do not provide data on the number of particles producing
36
-------
infection. Quantitation of the agents is limited by the sensitivity of the ^Ln
assay used. With viruses and bacteria, stable clumps of particles may
represent single counts in an assay system. It is also possible that some
particles infectious for an intact host may not multiply in a specific in vitro
assay system. The quantitation error produced by these opposing limitations of
detection techniques is unknown. It is only mentioned here to indicate that
each infective unit detected by these assays does not necessarily represent a
single particle of a given microbial agent.
The high infectivity of echovirus 12 via oral ingestion suggested a high
transmission rate for this virus. This assumption was not supported by the
occurrence of specific antibody in young adult men. Of 385 volunteers screened
for the study, only 46 (12 percent) demonstrated antibody as determined by
hemagglutination- inhibition test conducted with 1:5 dilutions of sera. How-
ever, the importance of this observation is obscured by the finding that
antibody response is an insensitive indicator of echovirus 12 infection when
produced by low virus exposure levels (Schiff, personal communication)
The data summarized in this report indicate that enteric pathogens can
cause infections at exposure concentrations that typically occur in raw
wastewater. However, available data are insufficeint to evaluate the actual
health hazards that exist for individuals exposed to wastewater subjected to
varying degrees of treatment and dilution. Obviously, important factors in the
environmental transmission of infectious agents are not adequately understood.
The identification and quantitation of these factors must be accomplished
before the modeling approach can provide a realistic assessment of potential
environmental health hazards and the importance of wastewater disinfection.
LITERATURE CITED
1. Beaver, P.C., R.C. Jung, H.J. Sherman, T.R. Read and T.A. Robinson. 1956.
Experimental Entamoeba histolytica Infections in Man. Am. J. Trop.
Med. Hyg., 5:1000-1009.
2. Cash, R.A., S.I. Music, J.P Libonati, M.J Snyder, R.P. Wenzel and R.B.
Hornick. 1974. Response of Man to Infection with Vibrio cholerae. I.
Clinical, Serologic, and Bacteriologic Responses to a Known Inoculum.
J. Infect. Pis., Vol. 129, 1:45-52.
3. Couch, R.B., T.R. Gate, R.G. Douglas, Jr., P.J. Gerone and V. Knight.
1966. Effect of Route of Inoculation on Experimental Respiratory
Viral Disease in Volunteers and Evidence for Airborne Transmission.
Bacterial. Rev., 30:517-529.
4. Craun, G.F 1977 Waterborne Outbreaks. J. Water Pollut. Control Fed.,
Vol. 49, 6:1268-1279.
5. Craun, G.F 1981. Disease Outbreaks Caused by Drinking Water. J. Water
Pollut. Control Fed., Vol. 53, 6:1134-1138.
37
-------
6. DuPont, H.L., S.B. Formal, R.B. Hornick, M.J. Snyder , J.P. Libonati , D.G.
Sheahan, E.H. LaBrec and J.P. Kalas . 1971. Pathogenesis of
Escherichia coli Diarrhea. N. Engl. J. Med . , Vol. 285, 1:1-9
7. DuPont, H.L. , R.B. Hornick, M.J. Snyder, J.P. Libonati, S.B. Formal and
E.J. Gangarosa. 1972. Immunity in Shigellosis. II. Protection
Induced by Oral Live Vaccine or Primary Infection. J . Infect . Pis . ,
Vol. 125, 1:12-16.
8. DuPont, H.L. and R.B. Hornick. 1973. Clinical Approach to Infectious
Diarrheas. Medicine, Vol. 52, 4:265-270.
9. Hornick, R.B., S.E. Greisman, T.E. Woodward, H.L. DuPont, A.T. Dawkins
and M.J. Snyder. 1970. Typhoid Fever: Pathogenesis and Immunologic
Control. N. Engl. J. Med. , Vol. 283, 13:686-691.
10. Kasel, J.A. , H.E. Evans, S. Anderson and V. Knight. 1963. Conjunctivitis
and Enteric Infection with Adenovirus Types 26 and 27: Responses to
Primary, Secondary and Reciprocal Cross-Challenges. Am. J. Hyg. ,
77:265-282.
11. Katz, M. and S.A. Plotkin. 1967. Minimal Infective Dose of Attenuated
Poliovirus for Man. Am. J. Public Health, Vol. 57, 10:1837-1840.
12. Koprowski, H. 1956. Immunization Against Poliomyelitis With Living At-
tentuated Virus. Am. J. Trop. Med. Hyg. , Vol. 5, 3:440-452.
13. Koprowski, H. , T.W. Norton, G.A, Jervis , T.L. Nelson, D.L. Chadwick, D.J.
Nelson and K.F. Meyer, 1956. Clinical Investigations on Attenuated
Strains of Poliomyelitis Virus. J. Amer. Med. Assoc., Vol. 160,
11:954-966.
14. Koya, G., N. Kosakai, M. Kono, M. Mori and Y. Fukasawa. 1954. Observa-
tions on the Multiplication of Escherichia coli 0-111 B4 in the
Intestinal Tract of Adult Volunteers in Feeding Experiments. Japan J .
Med. Sci. Biol . , 7:197-202.
15. Lang, D.J., L.J. Kunz , A.R. Martin, S.A. Schroeder and L.A. Thomson.
1967. Carmine as a Source of Nosocomial Salmonellosis . N. Engl. J.
Med. , Vol. 276, 15:829-832.
16. Levine, M.M. , H.L. DuPont, S.B. Formal, R.B. Hornick, A. Takeuchi, E.J.
Gangarosa, M.J. Snyder and J.P. Libonati. 1973. Pathogenesis of
ShigeJUa dysenteriae 1 (Shiga) Dysentery. J. Infect. Pis. , Vol. 127,
'
17. Microbiology of Drinking Water. 1977. In: Drinking Water and Health,
Safe Drinking Water Committee, NRC, National Academy of Sciences,
Washington, D.C.
38
-------
18. Minor, I.E., C.I. Allen, A.A. Tsiatis, D.B. Nelson and D.J. D'Alessio.
1981. Human Infective Dose Determination for Oral Poliovirus Type 1
Vaccine in Infants. J. Clin. Micro., Vol. 13, 2:388-389
19. McCullough, N.B. and C.W. Eisele. 1951. Experimental Human Salmonel-
losis. I. Pathogenicity of Strains of Salmonella meleagridis and
Salmonella anatum Obtained From Spray-Dried Whole Egg. J. Infect.
Pis., 88:278-289.
20. McCullough, N.B. and C.W. Eisele. 1951. Experimental Human Salmonel-
losis. III. Pathogenicity of Strains of Salmonella newport,
Salmonella derby and Salmonella bareilly Obtained From Spray-Dried
Whole Egg. J. Infect. Pis., Vol. 89, 3:209-213.
21. Rendtorff, R.C. 1954. The Experimental Transmission of Human Intestinal
Protozoan Parasites. I. Endamoeba coli Cysts Given in Capsules. Am.
J. Hug., Vol. 59, 2:196-20£L
22. Rendtorff, R.C. 1954. The Experimental Transmission of Human Intestinal
Protozoan Parasites. II. Giardia lamblia Cysts Given in Capsules.
Am. J. Hyg., 59:209-220.
23. Robinson, D.A. 1981. Infective Dose of Campylobacter jejuni in Milk.
British Medical Journal, 282:1584.
24. Stefanovic, G.M., B. Young, J.K. Pennekamp, E.W. Akin and G.M. Schiff.
1981. Determination of Minimal Infectious Dose of an Enterovirus in
Non-Chlorinated Drinking Water in Human Volunteers. Abstracts of the
Annual Meeting of the ASM.
25. Szita, J. e_t al. 1972. Incidence of Yersinia enterocolitica Infection
in Hungary. In: Winblad, S., ed. Proceedings of the International
Symposium on Yersinia, Pasteurella and Francisella, Malmo, 1972.
Basel, Karger, 1973; pp. 106-110 (Contributions to microbiology and
immunology, Vol. 2).
39
-------
4. VIRAL GASTROENTERITIS CAUSED BY THE SNOW MOUNTAIN AGENT,
A NEWLY RECOGNIZED NORWALK-LIKE VIRUS
Raphael Dolin, M.D.
Professor of Medicine
University of Vermont College of Medicine
Burlington, VT 05405
ABSTRACT
The Norwalk-like agents are common causes of acute viral gastro-
enteritis which undergo waterborne and person-to-person spread. Outbreaks
are often explosive in nature., with high attack rates and short incubation
periods. Disease manifestations last 24 to 72 hours, and generally remit
spontaneously. Recent studies of an extensive waterborne outbreak at a
mountain resort, indicate that the etiologic agent (the Snow Mountain agent
or "SMA") was a 26-32 nm virus which is morphologically similar to, but
antigenically distinct from previously described Norwalk-like agents.
Challenge of normal volunteers with orally administered SMA resulted in
induction of acute gastrointestinal illness in 9 of 12 volunteers. Illness
was similar to that seen in the naturally-occurring outbreak. Virus
particles were detected in the stools of 2 of 5 naturally-occurring and in 3
of 9 experimentally-induced cases by immune electron microscopy (IEM).
Serum antibody rises were detected by IEM in 8/9 volunteers who became ill,
in 0/3 volunteers who did not become ill, and in 3 of the naturally-
occurring cases. A highly sensitive and specific solid phase radioimmuno-
assay (RIA) was developed which detects SMA antigen and anti-SMA antibody.
This RIA should enable assessment of the epidemiologic significance of SMA
and may be useful in the consideration of control measures for waterborne
spread.
INTRODUCTION
The Norwalk-like agents are common causes of acute viral gastroenteritis
which undergo waterborne and person-to-person spread (3). Outbreaks are
often explosive in nature, with high attack rates and short incubation
periods. Disease manifestations last 24 to 72 hours, and generally remit
spontaneously. The Nprwalk-like agents are found in diarrheal stools, are
approximately 27 nm in diameter, and thus far have not been successfully
cultivated in vitro (6). At least four antigenically distinct agents have
been described: the Norwalk agent (5); the Hawaii agent (15); the
W-Ditchling agent (1); and the Marin agent (13). Because of the lack of
suitable methods of detection, little is known about the prevalence, mode of
transmission, and spread of most of the Norwalk-like agents. An exception
to this is the "Norwalk agent itself, for which a sensitive radioimmunoassay
40
-------
has been developed (7). Employing this assay, infection with the Norwalk
agent appears to be world-wide in distribution, and in one study, accounted
for nearly one third of the outbreaks of gastroenteritis which were examined
(8). A waterborne source of infection is suspected for the original out-
break in Norwalk, Ohio, from which the agent was derived, although no common
source was detected (2). The Norwalk agent, however, has subsequently been
associated with multiple waterborne outbreaks of gastroenteritis in the
United States and in Austrialia (4,12,14).
The current studies were carried out with material obtained from an
outbreak of acute gastroenteritis at a resort, Snow Mountain, near Granby,
Colorado, in December of 1976, which has been reported previously (11). In
brief, the characteristics of illness were typical of those of acute viral
gastroenteritis: namely, fever, vomiting, and diarrhea, which lasted 24 to
48 hours. The attack rate was high, involving 418 of 762 individuals at
risk, and person-to-person spread was noted. Illness was associated with
consumption of water or ice containing beverages in a dose response manner
(p <0.0001). Stool specimens were negative for conventional bacterial
pathogens, and virus cultures employing standard tissue culture systems were
similarly negative. Because a limited amount of material (stools and sera)
was available from the naturally-occurring outbreak, we performed a series
of studies in normal volunteers at the National Intitutes of Health to
determine the infectivity of the preparations and to generate additional
reagents for in vitro studies. These studies were performed prior to the
laboratory studies which were supported by the Environmental Protection
Agency and are described below. Gastrointestinal illness, similar to that
observed in the natural outbreak, was transmitted to 9 of 12 normal volun-
teers who were challenged with bacteria-free stool filtrates from one of the
naturally-occurring cases (Dolin R, Reichman RC, Roessner KD, Tralka TS,
Schooley RT, Gary W, and Morens D, Detection by immune electron microscopy of
Snow Mountain agent of acute viral gastroenteritis, submitted for publi-
cation, 1982). The current report describes the detection of the viral agent
of this outbreak (the Snow Mountain agent or SMA) employing immune electron
microscopy, and reports the development of a sensitive solid phase radio-
immunoassay for the detection of both antigen and antibody to the Snow
Mountain agent.
MATERIAL AND METHODS
Preparation of stool filtrates. The stool filtrates are prepared as 2%
suspensions of diarrhea! stool in veal infusion broth supplemented with 0.5%
bovine serum albumin. After low speed centrifugation, the suspensions are
filtered through nitrocellulose filters of decreasing pore size to a final
filtration through filters of 0.45 u in size. The filtrates are free of
detectable bacterial, viral, and mycoplasmal agents by conventional
techniques (10).
41
-------
Immune Electron Microscopy (IEM). The techniques employed for I EM are
those previously described for visualization of the Norwalk agent (9). 0.8
ml of the stool filtrate is reacted with 0.2 ml of convalescent serum over-
night at 4°C. The reaction mixture is centrifuged at 17,000 rpm for 90 min.
in an RC58 Sorvall centrifuge with a fixed angle rotor (SS34). The supernatant
is discarded, and the pellet is resuspended in 1 drop of distilled water.
The suspension is then placed on a 400 mesh Formvar grid, stained with 2%
phosphotungstic acid, and examined under a Philips 300 electron microscope.
The agent is detected as individual particles or aggregates coated with
antibody. Conversely, employing a filtrate which contains a known concen-
tration of antigen, antibody content in serum specimens can be assayed in a
semi-quantitative manner, on a scale of 1 to 4+.
Radioi'rnmunoas_s_ay_(RIA) for SMA Antigen. The RIA for SMA antigen is
performed as previously described (7). PuTified anti-SMA IgG is prepared
by (NH.)2SO, precipitation of convalescent (post-challenge) serum and dia-
lysis with 0.005 M phosphate buffered saline (pH 8) at 4°C for 5 days. The
globulins are purified by passage through a DEAE cellulose ion exchange
column (Whatman-DESZ) at pH 8.0 with 0.005 M PBS. The purified IgG is
labeled with I by chloramine-T reaction. Wells in polyvinyl microtiter
plates are coated with a 1:10,000 dilution of either pre- or post-challenge
serum overnight at room temperature. After washing out the wells, varying
dilutions of stool filtrates are incubated in each well overnight at room
temperature. After additional washing, 50 ul of purified IgG containing
200,000 cpm of I are added to each well, and incubated for 4 hours at
37°C. The plates are washed again, and individual wells are counted in a
gamma counting system. Differences in binding of greater than 2 (P/N >_2)
when wells coated with pre- and post-serum are compared, indicate the
presence of antigen.
RIA for antibody to SMA. The methods are as follows: wells in poly-
vinyl microtiter plates are coated with a 1:10,000 dilution of a post-
challenge serum specimen derived from a case of SMA-induced illness. 25 ul
of a standard stool preparation containing SMA is added to the wells,
incubated overnight at room temperature, and subsequently washed. This
stool preparation results in a P/N >3 in the above assay. Ten-fold
dilutions of the serum to be tested, in 40 ul aliquots, are added to?each
well and again incubated over night at room temperature. 10 ul of I
labeled anti-SMA IgG is added to each well and incubated for 4 hours at
37°C. The plates are washed and counted as above. The titer of serum
obtained is the reciprocal of the highest dilution which results in 50% or
greater reduction in counts when compared to a PBS control.
42
-------
RESULTS
Detection of Snow Mountain agent by Immune electron microscopy.
Examination of stool filtrates from five of the naturally-occurring cases
from the Snow Mountain outbreak revealed 27 nm virus-like particles in two
of the specimens. Examination of the stools from nine cases of experi-
mentally-induced illness revealed similar virus-like particles in stools
from three of these cases (Fig. 1). Particles appeared to have cubic
symmetry, were non-enveloped, and were morphologically indistinguishable
from the previously described Norwalk and Hawaii agents. The vast majority
of particles were 27 nm in diameter, but occasionally particles as large as
32 nm were also seen. The particles were seen most frequently 24 to 48
hours after the onset of illness, but were not detected more than 72 hours
after illness had begun. The 27 nm particles were not observed in stools
obtained prior to challenge nor in volunteers who were challenged and did
not become ill. Virus particles were observed either as single particles or
aggregates heavily coated with antibody.
Detection of antibody to Snow Mountain agent by immune electron
microscopy. The results of antibody determination in serum specimens from
subjects challenged with the Snow Mountain agent as determined by immune
electron microscopy are presented in Table 1. Eight of nine volunteers with
experimentally-induced illness demonstrated serum antibody rises to the 27
nm particle, along with three of the three naturally-occurring cases (Nos
11, 12, and 13) which were tested. None of the three volunteers who were
challenged with the Snow Mountain agent but did not become ill manifested
serum antibody rises.
Antigenic relatedness of Snow Mountain, Norwalk and Hawaii agents.
Analysis of serum antibody responses to the Norwalk, Hawaii, and Snow
Mountain agents as determined by immune electron microscopy is presented in
Table 2. Serum specimens were available from previous volunteer studies
which had been carried out. Significant antibody rises to the homologous
antigens were demonstrated with all three agents. No heterologous rises
were seen among any of the agents, suggesting that the three agents are
antigenically distinct as determined by this technique.
Radioimmunoassay for the Snow Mountain agent (SMA). Employing serum
specimens which had marked differences in antibody ratings by IEM, a radio-
immunoassay for the Snow Mountain agent was developed. The results of the
radioimmunoassay for SMA antigen in stool filtrates from the three volun-
teers challenged with SMA who shed virus in stools are presented in Tables
3, 4, and 5. The pattern of shedding of SMA antigen was similar to that
previously observed for the Norwalk antigen. SMA antigen was generally
detected 24 to 52 hours after challenge and was no longer detectable beyond
five days after challenge. Antigen was not detected in stools prior to
43
-------
challenge, nor in stools from volunteers who did not become ill. SMA
antigen was detected in 11 separate stool specimens from the previously
described volunteer studies. RIA antigen was always detected in stools in
which virus particles were seen by immune electron microscopy, as well as in
a number of lEM-negative stools. The latter observation suggests that the
RIA is more sensitive than IEMS or that alternatively, the RIA may detect
soluble as well as vi rion-associated antigens. The radioimmunoassay did not
cross-react with stools which contained Norwalk or Hawaii agents.
Radioimmunoassay for antibody to the Snow Mountain agent. Emp 1 oy ing
the above stool filtrates as a source of antigen, a solid phase radioimmuno-
assay for antibody to SMA was developed. The results of RIA antibody deter-
minations are presented in Table 6. Serum antibody rises were observed in 8
of 9 volunteers with experimentally-induced illness after challenge with
SMA, and in 3 of the 3 naturally-occurring cases. No antibody rises were
detected in the three volunteers who were challenged and did not become ill.
Serum antibody rises determined by RIA correlated well with those observed
by IEM.
DISCUSSION
The current studies have clearly identified a new water-borne viral
agent of gastroenteritis, the Snow Mountain agent. This agent is present in
the stools of naturally-occurring cases in the Snow Mountain outbreak and
has induced gastroenteritis after oral administration to normal volunteers
in the form of 2% stool filtrates. Additional lines of evidence which
support the etiologic significance of the virus particle are serum antibody
rises detected in subjects with naturally-occurring and experimentally-
induced cases of gastroenteritis, and the absence of such rises in
individuals who were challenged and did not become ill. In addition, the
particle is shed during acute illness but is not detected prior to challenge
or when illness is not present. The particle is approximately 27 nm in
diameters has cubic symmetry, and morphologically resembles the Norwalk and
Hawaii agents. However, SMA appears to be antigenically distinct frcm the
Norwalk and Hawaii agents by immune electron microscopy as well as by radio-
immunoassay.
In vitro detection of SMA was accomplised despite the inability to
cultivate the agent in conventional tissue culture systems. Initially, the
virus particles were detected in stool specimens from the naturally-
occurring outbreak, employing immune electron microscopy. This cumbersome
although powerful technique, relies on the ability of virus particles to be
aggregated by specific antibody which can be readily recognized under the
electron microscope. The etiologic significance of such particles, however,
remains uncertain until specific antibody rises in acute and convalescent
44
-------
serum specimens have been demonstrated. Immune electron microscopy was then
utilized to identify stools with high concentrations of particles and serum
specimens with high titers of anti-SMA antibody.
Because of the laborious nature of immune electron microscopy, only a
small number of specimens can be examined by this technique. Therefore,
progress in the field depends on the development of an efficient, yet
sensitive method for the detection of SMA, such as a radioimmunoassay.
Employing reagents known to contain antigen and antibody to SMA as identi-
fied by IEM, a solid-phase radioimmunoassay was established. This assay
detected SMA antigen in the stools of volunteers following challenge with
the infectious inoculum, and demonstrated a shedding pattern similar to that
previously shown for the Norwalk agent. Serum antibody responses were also
demonstrated in both experimentally-induced and naturally- occurring
illness, and correlated well with antibody responses as determined by immune
electron microscopy. While the RIA appeared to be equally sensitive to IEM
in the detection of serum antibody rises, RIA was significantly more
sensitive in the detection of antigen in stools than IEM. The latter
phenomenon may reflect either an increased sensitivity of the radioimmuno-
assay for virus particles, or alternatively, the detection of soluble
(non-virion associated) antigen which is not detected by IEM. Additional
studies are required to resolve this question.
These studies again document the development of waterborne illness
caused by viral agents. Data concerning the overall impact of waterborne
disease in the United States are fragmentary, but the most recent CDC
summary indicates that the total number of cases associated with reported
outbreaks in 1978 numbered 11,435, which is a three fold increase from those
reported in 1977 (4). Both of these figures likely represent gross under-
reporting of the problem. Particularly interesting is the fact that no
etiology has been established in more than 50% of outbreaks of gastro-
enteritis reported to the CDC during 1978, despite analyses of samples for a
variety of bacterial, parasitic and viral pathogens, including the Norwalk
agent by radioimmunoassay. Of outbreaks in which an etiologic agent was
determined, the Norwalk agent accounted for approximately 20% (4). Since
the radioimmunoassay for the Norwalk agent does not detect antigenically
distinct agents such as SMA or the Hawaii agent, it is conceivable that the
Norwalk-like agents as a group may account for a much greater proportion of
water related illness. It should also be noted that in contrast to several
other waterborne pathogens, the Norwalk-like agents undergo rapid person-to-
person spread once infection has occurred, so that the impact of waterborne
transmission may be multiplied many-fold.
The establishment of the radioimmunoassay for SMA now provides a
powerful tool with which to pursue studies of this new agent. Clearly a
major requirement for advances in this field would be the establishment of
in vitro culture systems with which to detect and study SMA, as well as
45
-------
other Norwalk-like agents. Since a large number of samples can now be
analyzed efficiently for the presence of SMA, intensive investigation of
promising in vitro culture systems can be carried out. Similarly, the
epidemic!ogic impact of disease produced by this agent can now be evaluated
in studies of both acute outbreaks and seroprevalence. Enviromental
sampling for this agent can also now take place, along with evaluation of
procedures for decontamination of drinking water. Because of the documented
waterborne spread of these agents, effective methods of decontamination of
water sources may represent an important control measure for diseases caused
by Norwalk-like viruses.
Acknowledgement
These studies were supported by EPA Grant No. R806546. We wish to
thank Dr. Elmer P. Akin for his valuable support and suggestions.
46
-------
1. Appleton, H., Buckely, M., Thorn, B.T., Cotton, J.L., Henderson, S.
1977. Virus-like particles in winter vomiting disease. Lancet
1:409-411.
2. Adler, J.L., Zickl, R. 1969. Winter vomiting disease. J. Infect.
Dis. 119:668-693.
3. Blacklow, N.R., Cukor, G. 1981. Viral gastroenteritis. N. Engl. J.
Med. 304:397-406.
4. Centers for Disease Control: Water-related disease outbreaks -
Annual Summary 1978, HHS Publication NO (CDC) 80-8385, U.S.
Government Printing Office, pp. 1-26, 1980.
5. Dolin, R., Blacklow, N.R., Dupont, H., Buscho, R.F., Wyatt, R.G.,
Kasel, J.A., Chames, R.P-, Hornick, R., Chanock, R.M. 1971.
Transmission of acute infectious nonbacterial gastroenteritis to
volunteers bv administration of stool filtrates. J. Infect. Dis.
123:307-312.
6. Dolin, R., Blacklow, N.R., DuPont, H., Buscho, R.F., Wyatt, R.G.,
Kasel, J.A., Hornick, R., Chanock, R.M. 1972. Biological
properties of Norwalk agent of acute infectious nonbacterial
gastroenteritis. Proc. Soc. Exp. Biol. Med. 140:578-583.
7. Greenberg, H.B., Wyatt, R.G., Valdesuso, J., Kalica, A.R., London,
W.T., Chanock, R.M., Kapikian, A.Z. 1978. Sol id-phase microtiter
radioimmunoassay for detection of the Norwalk strain of acute
nonbacterial, epidemic gastroenteritis virus and its antibodies.
J. Med. Virol . 2:97-108.
8. Greenberg, H.B., Valdesuso, J., Yolken, R.H., Gangarosa, E. , Gary,
W., Wyatt, R.G., Konno, T., Suzuki, H., Chanock, R.M., Kapikian,
A.Z. 1979. Role of Norwalk virus in outbreaks of nonbacterial
gastroenteritis. J. Infect. Dis. 139:564-568.
9. Kapikian, A.Z., Wyatt, R.G., Dolin, R., Thornhill, T.S., Kalica,
A.R. Chanock, R.M. 1972. Visualization by immune electron
microscopy of a 27-nm particle associated with acute infectious
nonbacterial gastroenteritis. J. Virol. 10:1075-1081.
10. Knight, V. The use of volunteers in medical virology. 1974. Prog.
Med. Vi rol. 66:1-26.
11. Morens, D.M., Zweighaft, R.M., Versnon, T.M., Gary, G.W., Eslien,
J., Wood, B.T., Holman, R.C., Dolin, R. 1979. A waterborne
outbreak of gastroenteritis with secondary person-to-person
spread: Association with a viral agent. Lancet 1:964-966.
47
-------
12. Murphy, A.M., Grohmann, G.S., Christopher, P.J., Lopez, W.A.,
Davey, G.R., Millsom, R.H. An Australia-wide outbreak of gastro-
enteritis from oysters caused by Norwalk virus. Med. J. Aust.
2:329-3339 1979-
13. Oshiro, L.S., Haley, C.E., Roberto, R.R., Riggs, J.L., Croughan,
M., Greenberg, H.B., Kapikian, A.Z. A 27-nm virus isolated during
an outbreak of acute infectious nonbacterial gastroenteritis in a
convalescent hospital: A possible new serotype. J. Infect. Dis.
143:791-796, 1981.
14. Taylor, J.W., Gary, G.W., Greenberg, H.B. Norwalk-related viral
gastroenteritis due to contaminated drinking water. Am. J.
Epidemic!. 114:584-592, 1981.
15. Thornhill, T.S., Wyatt, R.G., Kalica, A.R., Dolin, R., Chanock,
R.M., Kapikian, A.Z. Detection by immune electron microscopy of
26-to 27-nm viruslike particles associated with two family
outbreaks of gastroenteritis. J. Infect. Dis. 135:20-27, 1977.
48
-------
TABLE 1. ANTIBODY RATINGS IN SERUM SPECIMENS FROM SUBJECTS
CHALLENGED WITH SNOW MOUNTAIN AGENT (SMA)
AS DETERMINED BY IMMUNE ELECTRON MICROSCOPY
Subject
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
11 Iness Fol lowing
Challenge
yes
yes
yes
no
yes
yes
yes
yes
yes
yes
*
yes
*
yes
*
yes
no
no
Pre-challenge Post-challenge
Serum Serum#
<1 4+
2+ 3+
<1 2+
3+ 3+
<1 3+
<1 2+
1+ 1+
2+ 4+
<1 2+
2+ 4+
1+ 2+
<1 2+
<1 3+
<1 <1
2+ 2+
*
Naturally-occurring challenge during outbreak
+Ratings determined on a scale of 0 to 4+ employing a stool filtrate as a
source of antigen
#3 to 6 weeks after illness
49
-------
Table 2. ANALYSIS OF SERUM ANTIBODY RESPONSES TO NORWALK,
HAWAII, AND SNOW MOUNTAIN AGENTS*(SMA) BY
IMMUNE ELECTRON MICROSCOPY (IEM)
Rating of serum specimens to homol
Agent which Norwalk Antigen
induced illness
in challenge study
Norwalk (1)
Norwalk (2)
Hawaii (1)
Hawaii (2)
SMA (1)
SMA (2)
pre
1+
1+
2+
1+
2+
1+
post
4+
3+
2+
1+
2+
1+
ogous and
Hawa i i
pre
2+
2+
1+
1+
1+
2+
heterl ogous antigens
Antigen SMA Antigen
post
2+
1-2+
3+
3+
1+
2+
pre post
1+ 1+
1+ 1+
2+ 2+
<1 3+
2+ 4+
Challenge studies performed previously
TABLE 3. RADIOIMMUNOASSAY FOR SMA ANTIGEN IN STOOL FILTRATES FROM
VOLUNTEER #5 CHALLENGED WITH SMA
Time at which stool was passed P/N
24 hrs pre-challenge 0.80
6 hrs post-challenge 0.82
13 hrs post-challenge 1.01
24 hrs post-challenge 2.01+
120 hrs post-challenge 0.83
*
P/N ratio determined by solid phase radioimmunoassay
+P/N _^_2 indicates the presence of SMA Antigen
50
-------
TABLE 4. RADIOIMMUNOASSAY FOR SMA ANTIGEN IN STOOL FILTRATE
FROM VOLUNTEER #9 CHALLENGE WITH SMA
Time at which stool was passed
24 hrs pre-challenge
4 hrs post-challenge
30 hrs post-challenge
50 hrs post-challenge
52 hrs post-challenge
70 hrs post-challenge
72 hrs post-challenge
77 hrs post-challenge
96 hrs post-challenge
122 hrs post-challenge
P/N*
0.90
0.93
1.47
1.51
3.17+
3.79
2.53
3.61
1.10
1.19
P/N ratio determined by solid phase radioimmunoassay
P/N _^_ 2 indicates the presence of SMA antigen
51
-------
TABLE 5. RADIOIMMUNOASSAY FOR SMA ANTIGEN IN STOOL FILTRATES
FROM VOLUNTEER #10 CHALLENGED WITH SMA
*
Time at which stool was passed P/N
48 hrs pre-chal lenge 1.05
24 hrs pre-challenge 0.90
6 hrs post-challenge 0.96
24 hrs post-challenge 0.71
31 hrs post-challenge 2.29+
34 hrs post-challenge 5.26
52 hrs post-challenge 12.69
72 hrs post-challenge 10.57
96 hrs post-challenge 6.66
120 hrs post-challenge 4.05
144 hrs post-challenge 1.32
164 hrs post-challenge 1.20
P/N ratio determined by solid phase radioimmunoassay
+P/N _^_ 2 indicates the presence of SMA antigen
52
-------
TABLE 6. RADIOIMMUNOASSAY FOR ANTIBODY TO SNOW MOUNTAIN AGENT (SMA)
Ul
U)
Subject
1
2
3
4
5
6
7
8
9
10
*
11
*
12
*
13
14
15
Illness following
challenge with SMA
yes
yes
yes
no
yes
yes
yes
yes
yes
yes
yes
yes
yes
no
no
Dilution
of SMA inoculum
10°
10°
10°
10°
10°
10°
10- l
10" 1
10"2
10"2
-
-
-
10" 3
10"3
Pre Challenge
Serum antibody titer
100
200
100
3200
100
<100
200
<100
<100
100
400
100
100
100
200
Post Challenge
Serum antibody titer
6400
800
800
1600
1600
200
200
1600
800
3200
1600
800
1600
200
200
Naturally-occurring illness during the original Snow Mountain outbreak - acute and
convalescent serum specimens are compared
-------
Figure 1. The Snow Mountain Agent as detected by immune electron microscopy.
Virus particles are 27 nm in diameter and are heavily coated with antibody.
.
-------
5. RISK ASSESSMENT OF WASTEWATER DISINFECTION
David U. Hubly
Associate Professor
University of Colorado at Denver
Denver, Colorado
ABSTRACT
An interdisciplinary team of University of Colorado at Denver faculty
have performed a limited risk assessment of wastewater disinfection alterna-
tives. The objective of the assessment was to provide policy makers with
another tool to use in choosing among the alternatives of chlorination,
chlorination/dechlorination, ozonation, ultraviolet radiation, and no disin-
fection. This paper summarizes some of the results of that study.
INTRODUCTION
What is risk assessment? I wish I knew. When we began our study months
ago I thought I knew, but the experience of developing our risk assessment
has convinced me that each risk assessment is a unique creation. Some have
dubbed risk assessment an emerging science: others have called it an art. I
have also heard it characterized as "jumping to conclusions from skimpy data".
A veteran risk assessor, who advised us during our work, described the risk
assessment process as gathering all the data you can find and then trying to
make something out of it. The moral of this preamble is that you must ap-
proach any risk assessment with flexible objectives. Trying to fit a risk
assessment into a preconceived structure will most likely lead to disappoint-
ment. On the other hand, viewing the results of a risk assessment within the
constraints of a data base and of the assessment resources may produce a
sense of accomplishment.
However, you cannot approach a risk assessment without objectives. In
an earlier paper (1), I described our overall objective as the development of
tools that decision makers, without expertise in the disinfection field,
could use in the adoption of public policy relating to wastewater disinfection
practices. I then subdivided this large objective into seven more specific
objectives. The study results that I am presenting today approach the overall
objective; however, not all of the seven more specific objectives proved to
be reasonable goals.
Early in the project proposal stage we decided to narrow the focus of
this project to only a few of the potential disinfection alternatives because
the available funds would not support a study of all disinfectants. The cri-
teria used to select the disinfection alternatives studied were: (1) the
55
-------
alternative was not subject to constraints that would make its use unlikely,
and (2) there was a good possibility that sufficient data existed to permit
a risk assessment of the alternative. The second criterionwas used because
the project sponsors had asked us to confine our work to the available data
base. The disinfection alternatives selected for study were: (1) chlorina-
tion, (2) chlorination followed by dechlorination, (3) ozonation, (4) radia-
tion with ultraviolet light, and (5) no disinfection at all.
METHODS
Risk assessments often consist of four parts. The first part is the
identification of the hazards. Each hazard is then investigated to identify
all of the consequences that can arise if the hazard occurs. These conse-
quences can usually be ranked or grouped according to severity. The third
part is the development of a probability of occurrence (or frequency) state-
ment for each consequence. Finally, the assessment interrelates the severity
and frequency elements for all the hazards/consequences.
The hazards associated with each disinfection alternative were divided
into three groups: (1) on-site use, (2) transportation, and (3) reaction
product hazards. Hazards indirectly associated with the use of the disinfect-
ants were not included in the study scope. For instance, hazards associated
with manufacturing a disinfectant or constructing disinfection facilities
were not studied.
The identification of some hazards was simple. For example, gaseous
chlorine and ozone releases were obvious on-site and/or transportation haz-
ards. On the other hand, the identification of the reaction product hazards
required lengthy and sometimes fruitless literature searches.
The consequences of some hazards were also easy to identify and describe.
For example, the effect of chlorine gas on humans is- well known; and the ulti-
mate consequences of death, physical impairment, and lost productive time were
easily identified. Some consequences, such as the effects of chlorine gas on
vegetation and inanimate objects required a little more literature searching
effort for adequate identification. At the other end of the spectrum, the
consequences associated with residual disinfectants and reaction products re-
quired massive literature searches. This latter set of tasks received a
major portion of the study's resources.
Estimating the probability of occurrence or frequency of the consequen-
ces is usually accomplished using one of three methods. The simplest and
most straight forward method is to collect the available data regarding the
occurrence of the consequence in situations where the potential hazard exists.
This method results in a quantitative statement of the consequence's probabil-
ity of occurrence in terms of a common measuring unit: for example, deaths
per man-year of exposure to the potential hazard.
A second method using available data can be used if a consequence can be
quantitatively linked to a specific event, and the available data can be used
56
-------
to estimate the probability of occurrence of the specified event. For exam-
ple, there are no data relating exposure to chlorine gas to the occurrence of
the expected health effects. However, there are data on lost time and total
manhours worked in treatment plants which can be used to indirectly estimate
the probability of the health effects occurring.
The third method of estimating the probability of a consequence occurring
is used whenever the available data will not permit using either of the first
two methods described above, i.e., there are no available data regarding the
occurrence of the consequence or a directly related event. In this situation,
the sequence of events leading to the occurrence of the consequence is defined
in ever increasing detail until a level of detail is reached that will permit
the estimation of the probability of occurrence of each subsequent event given
that the previous event has occurred. For example, suppose that consequence
C can result from event B, and event B is a result of event A. Also assume
that we can estimate the probability of C occurring given that event B has
occurred and the probability of event B occurring given that event A has oc-
curred. Finally, assume that we can estimate the probability of event A
occurring using available data. The probability of C occurring is then esti-
mated using the following probability equation.
< C> = (< C/B>) () (< A>) (1)
where (< >) denotes probability of occurrence.
It is important to note that this simple example requires the estimation of
three probabilities. Usually situations this simple are not found in risk
assessments.
This third method becomes even more complicated and cumbersome when the
consequence can result from two or more parallel sequences of events. When
the event sequences and the consequence are described pictorially (see Figure
1) they resemble an upside down tree, which has led to this analytical method
being labelled "fault tree analysis".
CONSEQUENCE C
EVENT B
EVENT E
EVENT A
EVENT D
Figure 1. Simple Fault Tree
57
-------
This simple fault tree results in the following probability equation for
the estimation of the frequency of consequence C.
C> = (< C/B>)(< B/A>) (< A>) + (< C/E>)(< E/D>) (< D> ) (2)
This fault tree method becomes even more complicated and cumbersome when
Boolean logic and/or feed forward or feed back relationships are added to the
fault tree.
Our original study proposal included extensive use of this fault tree
method which accounted for over half the proposed budget. The use of this
fault tree method was eliminated from the study to reconcile the needed re-
sources with the available resources. A combination of the first two methods
was, therefore, used in this study.
RESULTS
Chlorination
On-Site Use
The on-site use hazards of chlorination are human and vegetation expo-
sure to liquid or gaseous chlorine. Liquid chlorine vaporizes so rapidly that
exposure to liquid chlorine under normal working conditions is highly improb-
able. The consequences of exposure to gaseous chlorine are a function of ex-
posure dose and length of exposure, and are summarized in Table 1.
Table 1. Consequences of Exposure to Gaseous Chlorine
Exposure Exposure Consequences
Dose Time
Human Exposure
<1 ppm chronic No consequences.
5 ppm chronic Respiratory problems, nausea,
susceptibility to tuberculosis,
corrosion of teeth.
7 ppm 1 hour Mucous membrane irritation.
>7 ppm 1 hour Cough, conjunctivitis, pulmonary
edema, death.
100 ppm seconds Death.
Vegetation Exposure
.5-1 ppm 1 hour Spotting.
>1 ppm 1 hour Death.
Note - pprn = part per million by volume
No data relating exposure dose or exposure time to illness, lost work
time, or death could be found; however, large data bases relating death and/
or lost time to manhours worked were found. The sources of these data banks
58
-------
are shown in Table 2. The annual accident statistics publications were not
useful because the accident data were not broken down sufficiently. The OSHA
data were also not useful because municipalities are not required to submit
reports to OSHA which eliminates too large a portion of the data sources.
The OSHA data did report four deaths at wastewater treatment plants, but the
reports did not identify the hazards causing the deaths. The SDS data pro-
vide adequate detail; however, two basic deficiencies of this data bank depre-
ciate its usefulness. First, the data are collected at the state level, and
not all states collect SDS data, which means the data are not based on a
national sample. Second, the states do not require the reporting of lost
time accidents when the lost time is less than a given minimum ranging from
one to seven days. Thus, the minimal consequence accidents are not included
in the data banks, and the amount of such data lost varies from state to
state. We did, however, examine SDS data banks from several states and found
no death reports due to chlorine exposure.
Table 2. On-Site Accident Data Sources
United States Department of Labor
Bureau of Labor Statistics
Annual Accident Statistics
Supplementary Data System (SDS)
Occupational Safety and Health Administration (OSHA)
Safety Programs Office
Office of Management Data Systems
National Safety Council (NSC)
Water Pollution Control Federation (WPCF)
American Water Works Association (AWWA)
Most of the NSC data do not provide sufficient detail; however, the NSC
has reported one study of 156 treatment plants showing an accident rate of 40
lost workday cases per million manhours. Total lost time was 575 man-days per
million manhours worked; however, these data include both collection systems
and treatment plants.
The WPCF data are drawn from larger bases (7 to 10 percent of the plants
in North America) and are separated into collection system and treatment
plant groups. The most recent WPCF data for treatment plants are summarized
in Table 3.
These data are in agreement with the NSC data shown above which indicate
the accident rates for collection system employees and treatment plant employ-
ees are about the same. Comparing these rates with rates reported for other
industries indicates that wastewater treatment plant work is about as hazard-
ous as mineral mining. However, these data are still not sufficient for our
risk assessment because the chlorine accident rate cannot be separated from
our totals.
Data were collected for the broader based chemical and chlorine industries
in an attempt to separate the chlorine accident rates from the totals by
analogy. Ultimately, the AWWA data base was selected as the best analogy
59
-------
because the treatment processes are similar. Furthermore, the AWWA injury
frequency and severity rates were similar to the data shown above. And most
important of all, the AWWA data base contains a grouping that is mostly
chlorine accidents. According to the AWWA data chlorine related accidents
represent about 4 percent of the total accidents reported. This is comparable
to the accident rate reported for insect bites.
Table 3. 1979 WPCF Accident Data
Plant Size Man-hr per Injury Severity Rate
MGD Employee Freq. Lost Man-Days Fatalities
<1.0 1861 22.16 252.9 0
1.0- 2.5 1985 38.99 210.9 0
2.5-10.0 1945 48.23 436.9 0
>10.0 1958 61.95 749.3 0
Average 1952 52.48 566.1 0
Note - MGD = million gallons per day
Injury frequency is cases per million man-hours
Severity rate is per million man-hours
Finally, we concluded that the only two consequences associated with the
on-site use of chlorine that might be found in the data are death and lost
work time. Furthermore, the data are not sufficient to permit the estimation
of the probability of a death occurrence due to exposure to chlorine result-
ing from on-site use. However, the probability of lost work time can be esti-
mated by applying the four percent figure found in the AWWA data to the WPCF
data shown above. The resulting probabilities of occurrence are shown in
Table 4.
Table 4. Probable Lost Work Time Resulting
from On-Site Use of Chlorine
Plant Size Lost Work Time
MGD Man-hours Lost per Man-hour Worked
<1.0 0.00008
1.0- 2.5 0.00006
2.5-10.0 0.00014
>10.0 0.00024
Transportation
The consequences of the hazards associated with the transportation of
chlorine are identical to the hazards associated with on-site use described
above. The data on transportation accidents are also adequate for risk as-
sessment purposes . They came from two sources, the United States Department of
Transportation, and the Bureau of Census. These data banks permit the devel-
opment of frequency estimates for deaths, injuries, and property damage; and
those frequency estimates are summarized in Table 5. Regrettably, the injury
60
-------
data do not include any measure of severity so the conversion of those data
into lost work time is not possible. On the other hand, there is enough data
to permit the disaggregation of the probability estimates for trucking into
estimates for small cylinders, large cylinders, and tanker trucks.
Table 5. Probable Occurrence of Chlorine
Transportation Consequences
Transportation
Mode Deaths
Railroad 0.00063
Railroad
excluding
Youngs town 0.0
Barge 0.0
Truck
Cylinders < 250# 0.0
Cylinders 1 ton 0.0
Tankers 0.0
Consequences
Injuries
0.02
0.0068
0.068
4.0
0.02
0.047
Property
Damage
$87.00
$1.80
$0.00
$530.00
$31.40
$10.00
Note - Units are events or dollars per million-ton miles
The rail data included a catastrophic accident near Youngstown, Florida that
included all of the recorded deaths and substantial amounts of injury and
property damage. Therefore, two railroad entries are included in Table 5 to
show the impact of that single accident.
The results shown in Table 5 can be used to estimate the probable con-
sequences of using chlorine for wastewater disinfection for a given region if
you can estimate the amount of chlorine being shipped into the region and the
source of the chlorine shipments. Our final report will contain information
allowing the development of these latter estimates.
Reaction Products
The identification of total residual chlorine as a hazard was obvious;
however, the identification of chlorinated reaction product hazards was a
problem. Numerous potentially hazardous chlorinated compounds have been iden-
tified, and it was not possible to determine the consequences resulting from
the occurrence of each reaction product. Therefore, a subset of reaction
products was selected for study. An interim report of an EPA study of prior-
ity pollutants in wastewater effluent provided an estimate of the chlorinated
reaction products most likely to be found in a wastewater effluent. The
availability of toxicity data and the use of some compounds as models for
groups of compounds were also considered in the selection process. The reac-
tion product hazards selected for study were: Total residual chlorine, chlo-
roform, trichloroethylene, tetrachloroethylene, dichlorobenzene, chlorophe-
nols, and 5-chlorouracil.
Exposure to the selected reaction products can produce consequences
61
-------
primarily affecting humans, fish, and aquatic invertebrates. Human exposure
to the reaction products requires ingestion via our water supply, water-based
recreation, or consumption of aquatic flora or fauna. Available literature
indicates the fraction of the chlorinated compounds in our water supplies that
can be attributed to wastewater disinfection is negligible. Therefore, the
probability of a consequence occurring via this route is almost nil. Similar
analyses indicated that the probabilities of human consequences occurring via
the other routes are also very small providing there is no bioaccumulation.
Therefore, our study concentrated on the consequences occurring in the aquatic
systems.
The consequences of exposing aquatic organisms, particularly fish, to the
reaction product hazards cover a broad spectrum ranging from no effect to
acute toxicity. Consequences falling between the two extremes include:
avoidance, reduced spawning activity, reproductive dysfunction, and minor to
severe physiological changes (e.g., decreased size, mutagenesis, carcinogen-
esis, etc.). In some cases, these consequences are further complicated by
synergism among the reactants and by bioaccumulation. The literature relating
these consequences to exposure doses and length of exposure is massive. Our
biologists found over 400 pieces of data for residual chlorine alone. As a
result, a major portion of our time was spent in this area.
Summarizing the results of this part of our study is not possible within
the limits of this presentation. Therefore, I will show you only two of our
more important results. The minimum reported acute toxic effects concentra-
tion and the minimum reported LC-50 divided by 100 are shown in Table 6 for
each of the reaction products studied. The maximum reported effluent concen-
trations for each reaction product studied are also shown in Table 6.
Table 6. Reported Maximum Effluent, Minimum Acute Toxic
Effects, and Minimum LC-50x0.01 Concentrations
for Reaction Products Studied
Reaction Product
Residual chlorine
Chloroform
Trichloroethylene
Tetrachloroethylene
Chlorophenols
Dichlorobenzenes
5-chlorouracil
Maximum
Reported
Effluent
Cone .
rag/I
8.0
0.02
0.04
0.004
0.03
0.01
0.004
Minimum
Reported
Acute Toxic
Effects Cone.
mg/1
0.001
1.0
1.0
10.0
0.01
1.0
0.01
Minimum
Reported
LC-50x0.01
Cone.
mg/1
0.00014
0.018
0.36
0.13
0.0003
0.006
No LC-50s
reported
The results summarized in Table 6 indicate that residual chlorine will
probably have acute toxic effects unless the effluent is well diluted in the
receiving water body (not exactly a new finding). The summarized results
62
-------
also indicate that all of the chlorinated reaction products, with the excep-
tion of the chlorophenols, will probably not cause any acute toxic consequen-
ces even with zero dilution in the receiving water body. And the worst case
chlorophenol condition requires only a 3:1 dilution to reduce potential acute
toxic effects to a negligible level assuming that the available data include
the lower acute toxicity limit.
Assuming that the minimum LC-50x0.01 values are a reasonable estimate of
the no effects threshold, the data summarized in Table 6 indicate reported
effluent levels of chloroform, trichloroethylene, and tetrachloroethylene will
probably have no effect on stream organisms even with zero dilution in the
receiving water body. Chlorophenols may require dilutions up to 100:1, and
dichlorobenzenes may require dilution up to 2:1 to reach the no effects con-
centration.
The residual chlorine data bank is so large that it permits a more de-
tailed summary presentation. A simplified version of the method used in our
study report to present the residual chlorine summary is shown in Figure 2.
The thick bars shown in the upper half of Figure 2 enclose large concentra-
tions of reported data for that consequence, and the thin lines reach to the
outer limits of the reported data but do not enclose many data points.
Figure 2 illustrates some aberrations found in the chlorine data. For
example, most of the mortality threshold data show the threshold occurs at
concentrations much greater than many concentrations reported as LC-50's or
LC-100's. Furthermore, many reported LC-50's are greater than the bulk of
the reported LC-100's. These data help identify lower boundary conditions
(i.e., worst case), but they are not much help in analyzing specific dis-
charge and stream conditions.
Figure 2 also can serve as a fast method of estimating the impact of a
specific discharge on a receiving stream. For example, a discharge of 1 mg/1
residual chlorine with a Qe/Qs ratio of 0.01 results in a stream concentra-
tion near the left edge of the avoidance bar and just above the lowest values
shown for mortality threshold.
Probability estimates can also be used with Figure 2. An example prob-
lem will illustrate this point. Assume the values shown in Table 7 can be
developed from data available for a specific discharge. The first four
columns can be used to calculate the probability of Qe/Qs exceeding 0.01 for
each of the four flow combinations. The maximum probability is 0.338, and
this is taken as the limiting condition. Suppose we want to estimate the
probability of exceeding the 0.003 limit. At a Qe/Qs ratio of 0.01, the
effluent residual should not exceed 0.3. Based on the data in Table 7, the
probability of exceeding this value is 0.810 which means the overall prob-
ability of exceedance is 0.274 (0.338 x 0.810) or, in other words, the
stream concentrations of chlorine will exceed 0.003 about once every 3.6 years
on the average over all time. The same method can be used with Figure 2 to
predict the probable onset of mortality, 50 percent mortality, and 100 percent
mortality.
63
-------
CO
lu
O
2
Lu
Uj
to
O
/O
CM
-j
o
to'
Lu
-j
u.
u.
UJ
IQQ96 MORTALITY
5096 MORTALITY
MORTALITY THRESHOLD
AVOIDANCE
0.1
O.OOI
tO
10.0
0.01 O.I
/^STREAM RES. CL2,
ITote - Oe = effluent flow rate
Qs = stream flow rate
Figure 2. Summary of Residual Chlorine Data
Chlorination/Dechlorination
The consequences resulting from discharging residual chlorine as present-
ed above can be eliminated by adding an effective dechlorination process
after the chlorination process. Usually the most cost-effective and, there-
fore, the chosen process is the addition of sulfur dioxide. Our literature
search found that exposure to liquid or gaseous sulfur dioxide results in
consequences essentially the same as those shown for chlorine in Table 1.
Furthermore, the sulfonation process is very similar to the chlorination
64
-------
Table 7. Assumed Data for Example Problem
Qs Qe ER
1000
2000
3000
4000
0.100
0.450
0.700
0.845
10
20
30
40
0.998
0.750
0.270
0.065
0.1
0.3
0.5
1.0
0.997
0.810
0.500
0.110
= Probability of non-exceedance of
Qs shown
< Qe> = Probability of exceedance of Qe
shown
ER = Effluent chlorine residual
= Probability of exceedance of the
ER shown
process. Therefore, since no useful on-site use or transportation data were
found for sulfur dioxide, we concluded that the frequency estimates developed
above for on-site use and transportation of chlorine can also be used for the
dechlorination process using sulfur dioxide.
Since the addition of sulfur dioxide removes the chlorine residual haz-
ards and consequences, the reaction product hazards and consequences associ-
ated with the chlorination/dechlorination process will be the remaining chlo-
rination reaction products plus any additional products or effects of the
sulfur dioxide addition. The reaction products resulting from the addition of
sulfur dioxide to wastewater are not discussed in detail in the literature,
but available information indicates those products are mostly chlorides, sul-
fates, and sulfites. These products are not considered hazards in the aquatic
ecosystem; however, the addition of sulfur dioxide may create low dissolved
oxygen and pH conditions that can be hazardous. We could not identify the
probability of these conditions occurring.
Ozonation
The on-site use hazards associated with the use of ozone for disinfection
also create risks for both humans and nearby vegetation. The consequences of
human exposure to ozone include: no effect; minor irritation to eyes, skin,
and mucous membranes; headaches; respiratory distress; and death. Exposure
of vegetation to ozone can include: stunting, defoliation, and death. The
no effects threshold concentration for ozone is much lower than the chlorine
threshold with some sensitive humans experiencing effects at concentrations
as low as 0.02 ppm by volume. OSHA recommends a maximum exposure for an eight
hour period of 0.1 ppm by volume (as opposed to a 5 ppm limit for gaseous
chlorine). Since ozone is generated on-site using very high electrical volt-
ages the ozonation process also includes electrocution as a potential on-site
hazard.
No accident data were found that could be used to estimate the probabil-
ity of realizing the consequences associated with the on-site use of ozone.
65
-------
Therefore, we decided to survey the recently constructed ozonation plants to
see if their limited experiences would provide a qualitative assessment of the
risk. Vie hoped that we could then provide an intuitive assessment of the
probability of realizing the consequences of on-site ozone use in terms of the
on-site chlorine use estimates presented above. The probability of realizing
the ozone consequences should be greater than the chlorine probabilities be-
cause (1) humans and vegetation are more sensitive to ozone, and (2) ozone is
more likely to leak from the reactor because of its low solubility in water
and its almost negligible vapor pressure in the atmosphere. Our telephone
survey found several problems involving high ozone levels around the treat-
ment plant. In our opinion, the probability of realizing the on-site use
consequences associated with ozonation should be assumed to be several times
greater than the same probabilities for chlorine.
Since we are not considering the hazards of transmitting power in this
analysis there are no transportation hazards or consequences associated with
ozonation.
The reaction product hazards include ozone and a massive group of low
molecular weight alkanes, aldehydes, organic acids, and heterocyclics. The
literature indicates consequences of ozone exposure for fish range from:
locomotion and respiration impairment to death. These effects are also re-
ported to occur at relatively low concentrations. The literature also in-
cludes data on ozone residuals; however, most of these data were observed in
the ozone reactor effluent. Since ozone off-gases so readily, it is unlikely
much of the reported ozone residuals would be found in the receiving water
bodies. Therefore, we believe the potential consequences of aquatic expo-
sure to ozone are unlikely to occur.
A subset of six low molecular weight organic compounds were selected
from the mass of reported ozonation reaction products for toxicity analysis.
The compounds studies were: Heptane, n-octane, n-hexanol, m-xylene,
n-heptanal, and n-nonanal. No toxicity data were found for n-hexanal,
n-heptanal, and n-nonanal. The literature did contain a few studies that
show n-heptane and m-xylene are toxic in the mg/1 range, and n-octane was
found to be non-toxic at concentrations up to 100 mg/1. Even though no data
were found regarding expected effluent concentrations it is unlikely any of
these compounds will occur in hazardous amounts.
Ultraviolet Radiation
The hazards associated with the on-site use of UV radiation are: human
exposure to radiation, electrocution, and human exposure to ozone. Human
exposure can adversely affect both skin and eyes. The consequences of expo-
sure include reddening, blistering, and peeling of skin and corneal damage,
loss of visual acuity, and eye fatigue. Some literature was found linking
skin cancer with UV radiation. UV disinfection processes operate at levels
well above recommended human exposure limits so the probability of a conse-
quence occurring is certainly greater than zero. However, the estimation of
that probability was not possible with the available data.
66
-------
UV radiation can produce ozone if oxygen is present in the exposure area.
The ozone hazards and consequences have been discussed above; however, the
available data indicate this is not a significant hazard.
Since UV radiation does not produce a residual the reaction product
hazards are limited to changes in compounds existing in the wastewater. The
limited amount of literature dealing with this phenomenon prevented any risk
analysis of these hazards.
ACKNOWLEDGEMENTS
Several people have made major contributions to this study. First, our
Project Officer, Dr. AlbertD. Venosa, of the EPA Municipal Environmental Re-
search Laboratory has contributed a great deal of insight, direction, and
patience. My co-principal investigator has been Dr. Willard Chappell, Direc-
tor of the Center for Environmental Sciences at the University of Colorado at
Denver. Most of the investigative work for this project was accomplished by
our senior investigators: Dr. John Lanning, Dr. Martin Maltempo, Dr. Daniel
Chiras, and Dr. John Morris; who are all faculty members at the University of
Colorado at Denver. Dr. Chiras, our biologist, was assisted by two graduate
assistants, Mr. David Shugarts and Mr. Robert Williams. Ms. Betty Lepthien
assisted with the preparation and editing of the text.
LITERATURE CITED
1. Hubly, David W. 1979. Evaluation of Risks, Energy Costs, and Associated
Economic Factors of Wastewater Disinfection Alternatives. Proceedings
of Wastewater Disinfection Alternatives - State-of-the-Art Workshop,
October 7, 1979. Water Pollution Control Federation, Washington, D.C.
2. Hubly, David W.; Lanning, John; Maltempo, Martin; Chiras, Daniel;
Chappell, Willard; and Morris, John. Risk Assessment of Wastewater
Disinfection. EPA report to be completed in 1982.
67
-------
WASTEWATER AEROSOL HEALTH EFFECTS STUDIES AND THE NEED FOR DISINFECTION
Walter Jakubowski, Chief
Parasitology and Immunology
Health Effects Research Laboratory
U.S. Environmental Protection Agency
Cincinnati, Ohio
ABSTRACT
A series of epidemiological studies of community exposure to aerosols
from wastewater treatment plants, and a study of worker exposure to aerosols
and sewage liquids/solids contact, were reviewed and evaluated. Gastrointes-
tinal symptoms were reported in three of the studies and there was some sero-
logical evidence for viral infection, although a causal relationship to
sewage exposure could not be established. The preponderance of data was
negative and no definitive conclusions could be drawn. All of the studies
were limited by having low numbers of exposed individuals and by being unable
to adequately and quantitatively characterize exposure. Investigators have
been unable to confirm the results of a 1976 report which indicated two to four
times higher incidence of certain infectious diseases in agricultural
communities using wastewater for irrigation. Two additional health effects
studies on spray irrigation of wastewater are in progress.
INTRODUCTION
The principal potential routes for transmission of pathogens from sewage
to the population are through contamination of drinking water and recrea-
tional water, or through aerosols from sewage treatment plants and spray ir-
rigation practices. Contaminated food may also serve as a vehicle, but this
review will focus on the potential for direct aerosol transmission of
wastewater pathogens. The need for, and the efficacy of, disinfection of
drinking water have been adequately demonstrated throughout this century.
Typhoid and cholera have been dramatically reduced worldwide as a result of
the combined effects of improved sanitation and drinking water disinfection.
However, waterborne outbreaks of infectious disease still occur in the United
States, and they can often be traced to absent or inadequate disinfection.
Many of our surface water supplies and an increasing number of our groundwater
supplies are subject to sewage contamination. Although the species of
disinfectant and the point and manner of application may be subject to
modification due to trihalomethane considerations, disinfection of drinking
water appears to be a necessary practice that will continue to be widely used
in this country.
The second significant route for exposure to pathogens from sewage is
through recreational contact with contaminated fresh and marine waters. The
evidence for health effects as a result of this contact, and consequently,
68
-------
whether or not disinfection of wastewater discharged to receiving waters is
indicated, is presented elsewhere in these proceedings.
The first two routes of exposure involve dilution of wastewater in an
aqueous medium, and generally, oral ingestion of the contaminated water.
However, it should be'kept in mind that inhalation of aerosolized pathogens
from contaminated drinking and recreational waters is also possible. The
third route for potential exposure to pathogens from sewage is by
dilution and transport in air at sewage treatment plants and spray irrigation
sites. Over the last decade, the U.S. Environmental Protection Agency (USEPA)
sewage construction grants program provided the impetus for the development
of numerous activated sludge treatment plants. In the course of implementing
this program, several communities questioned the health hazards associated
with these plants, primarily from aerosols produced by the aeration basins.
It soon became apparent that there was a lack of data on health effects of
sewage aerosols.
A similar situation prevailed concerning aerosol hazards at land appli-
cation and spray irrigation sites. Land application and agricultural reuse of
municipal wastewater may have certain advantages over traditional treatment
and disposal practices employed in the United States. Municipal wastewater
can be considered as a resource rather than as an unwanted end product. It is
possible to recover and utilize some of the nutrient value contained in the
wastewater and to supplement water resources in water-poor areas. The
acceptance and implementation of land application and reuse, however, depends
upon resolving a variety of social, economic, and health effects issues.
Finding significant health effects associated with aerosols from acti-
vated sludge plants would most likely result in corrective actions other than
disinfection, e.g., instituting procedures to minimize aerosol formation; the
covering of aeration basins, or the erection of other physical barriers to
aerosol transport. Even so, health effects data from sewage treatment plant
studies could be useful in assessing the need for disinfection of wastewater
effluents used at spray irrigation sites or discharged to recreational
waters. Disinfection could be considered as an alternative or additional
treatment process for effluents used at spray irrigation sites if significant
health effects are demonstrated.
The purpose of this paper is to summarize the conclusions, limitations,
and relevance to the question of wastewater disinfection, of several studies
on the health effects of exposure to wastewater treatment plants and spray
irrigation practices. All but one of the studies have been funded by the U.S.
EPA. Two of the investigations are in progress and the rest were completed
within the last four years. The results of most of these studies appear in the
proceedings of a recent symposium (9).
STUDIES OF POPULATIONS NEAR SEWAGE TREATMENT PLANTS
Four epidemiological studies of populations exposed to activated sludge
69
-------
treatment plants were completed in 1978 and 1979. The plant locations and
type of exposed population are indicated in Table 1. Three of the studies
involved heterogeneous community populations, and the fourth was of children
at an elementary school near the surge and aeration basins of a treatment
plant. Details and findings of each of these studies will be presented
separately.
Tecumseh, MI
Tecumseh, Michigan was chosen as the site for this preliminary investi-
gation because it was part of a comprehensive community health study conducted
by the University of Michigan (3,9). Consequently, a considerable amount of
retrospective health and demographic data were available. The study
population was divided into five concentric rings radiating outward in
multiples of about 600 m from an activated sludge treatment plant. In 1965,
the plant was converted from a trickling filter facility to activated sludge
treatment. The study period was from 1965 to 1971. Average monthly sewage
flow rates during this time were from 0.64 to 1.18 MGD, although some of the
data for this period are missing. Self-reported acute illnesses and symptoms
from 4,889 participants during the 7-year period were grouped into total,
respiratory, and gastrointestinal illness (Gl) categories. Age-sex-dis-
tance-specific incidence and illness rates were analyzed using the minimum
discrimination information statistic. Data on income and education were also
used in the analysis.
The results indicated that there was a greater than expected occurrence
of total respiratory and GI illnesses in those living within 600 m of the
plant. However, this portion of the population also had lower income and
education than the rest of the study group. The investigators suggested that
high densities of lower socioeconomic families might be a more important
factor in excess illness than would be proximity to a small wastewater treat-
ment plant. Some excess illness (total, respiratory, and Gl) was also
reported for those living within the 2,400 m concentric ring. These people
had both higher incomes and education than any of the other groups and there
was no known source of exposure for acute illnesses in this area.
This retrospective study was obviously inconclusive and was undertaken
because of the availability of the data and the relatively small expense in-
volved. Limitations on interpretation of these results include: the presence
of a confounding demographically heterogeneous population, the lack of expo-
sure and meteorologic data, and the relatively low volume of the exposure
source. Differences by distance were detected, but determining the causes of
the differences was outside the scope of the project.
Schaumburg, IL
In this study, an opportunity was available to follow a community popu-
lation before and after an activated sludge treatment plant went on line
(5,9). The study period was from 1974 to 1976 and the John E. Egan plant
became operational in December, 1975, with average daily flow rates of 10-15
70
-------
MGD. The study design included a health survey of about 4,300 individuals (of
a total population of about 100,000) living within 5 km of the plant. The
residential area began 350 m from the plant. Clinical specimens from a subset
of this population (226 individuals) living within 3.5 km of the plant were
examined for bacteria, viruses, parasites, and viral antibodies in serum.
Wind speed and direction, and relative humidity, were monitored on-site.
Large-volume aerosol samples and wastewater samples were examined for
indicators and pathogens.
The results of the health survey indicated no change in asthma-hay fever
symptoms, decreases in worm infections, and decreases in sore throats after
the plant went into operation. Statistically significant (p < .01) increases
in the incidence of six diseases or symptoms were reported by those within 2
km of the plant (Table 2). Furthermore, all showed a relationship with
direction from the plant, i.e , increases were at the close distances in the
north and south directions, the predominant downwind quadrants. However, the
results for at least five of the six show considerable variability with
distance before the plant went into operation or show significant decreases in
those living more than 2 km from the plant. Nevertheless, one symptom,
diarrhea, showed remarkable uniformity in reporting throughout the study
population and increased from 4.1 to 7.6 percent in those living 0-2 km from
the plant. This finding is also interesting in that the post-operational
survey included a lower proportion of young children who would be more likely
to experience diarrhea.
These results must be considered in the context of the survey methodology
employed. Participants were asked on two occasions to report all acute
diseases occurring in the family in the previous year and to list all symptoms
occurring in the previous three months. The accuracy of surveys requiring
that much recall is open to serious question even if the respondent is
reporting only on himself. In addition, survey participants knew that the
study involved possible health effects of the sewage treatment plant.
Recognizing these limitations on survey information, the study included
objective measures of infection as well, i.e., serology and isolation of
pathogens from clinical specimens. Proteus, Pseudomonas, and Salmonella were
the only pathogen isolates from fecal samples. There was a significant de-
crease in Proteus isolations during the operational period and no significant
differences in Pseudomonas and Salmonella isolations. Streptococcus and
Staphylococcus isolations from throat swabs increased during the operational
period. However, the increases were not related to the treatment plant as
shown by regression analysis of the incidence pattern with distance and
direction from the plant. There were no significant differences in isolation
of parasites from fecal specimens. No viruses were found in throat swabs but
twenty viruses were found in fecal samples. There was a significant increase
in virus isolations during the operational period but the increase could not
be related to plant exposure. Antibody tests for 31 enteric viruses yielded
no serologic evidence of an adverse wastewater treatment plant effect. The
results of the aerosol monitoring indicated that levels of microorganisms in
the air in residential areas were indistinguishable from background concen-
trations .
71
-------
The investigators concluded that, at the exposure levels studied, sewage
treatment aerosols from well-operated American plants do not appear to pose
significant health hazards. They also indicated that there was insufficient
evidence to determine if minor effects such as gastrointestinal symptoms and
skin disease, were associated with aerosol exposure.
Tigard, OR
A retrospective study conducted in Oregon combined the rapidity and cost
advantages of the Tecumseh study with the exposure categorization experience
gained in the Schaumburg project. Anew activated sludge treatment plant had
been placed into operation in 1976 within 400 m of an elementary school and
local public health officials expressed concern about possible health
hazards. A preliminary study was performed to determine the types and numbers
of microorganisms in air upwind and downwind from the treatment plant and to
compare absenteeism rates, as a measure of possible health effects, at the
affected school and control schools (6,9).
The study design involved collecting seven years of attendance data
prior to initiation of plant operations and for two years afterwards. Data
were collected for the exposed school (Durham) and for five control schools in
the Tigard district. The plant had a design capacity of 20 MGD but averaged
9-13 MGD during the study period. There were two possible sources of exposure
to the wastewater aerosols: from a surge basin located within 50 m of the
school playground, and from an aeration basin about 400 m from the school
building. The Durham Elementary School had six classrooms (one for each
grade), open-window ventilation, and an enrollment in June, 1978, of 123 stu-
dents. On-site measurements of wind speed and direction, temperature,
humidity, solar radiation, and cloud cover were made. Composite wastewater
samples and large volume air samples were collected from the aeration and
surge basins and examined for indicator and pathogenic microorganisms.
The absenteeism results are shown in Table 3. As can be seen, the ab-
senteeism at the exposed school actually decreased during plant operation.
These negative data must also be considered in the context of the results ob-
tained from the environmental monitoring and exposure calculations. The
aeration basins were found to be a much stronger source of microorganisms than
the surge basin. The geometric mean aerosol concentrations at 30 to 50 m
downwind of the aeration basin were 12 colony forming units (cfu)/m3 of total
coliforms, 4.2 cfu/m3 of fecal streptococci, 19 cfu/m3 of mycobacteria, 1.5
plaque forming units (pfu)/m3 of coliphage, and less than 0.0002 pfu/m3 of
enteroviruses . However, the exposure calculations based upon the meteoro-
logical observations indicated that the classroom area was steadily downwind
of the aeration basin for only 10 days in the two operational years. In
addition, because of rainfall, the playground may have been in use on only 40
percent of the days when it was steadily exposed to aerosols. The
investigators concluded that the wastewater aerosols had no effect on
infectious disease incidence as determined through absenteeism for this level
of exposure.
Skokie, IL
72
-------
Probably the most thorough and the last of the treatment plant health
effects studies (1,9) was conducted by the University of Illinois near an
activated sludge facility with an average daily flow of 290 MGD. A subset of
the population living within a 1.6 km radius of the plant was studied for an
month period. A comprehensive health questionnaire survey was conducted of
2,378 persons at the beginning of the study to gather demographic and
historical health information on chronic and acute diseases. A subset of this
population (724 persons) was included in a health watch program where health
diaries were collected on family members every two weeks throughout the study
period. Although the keeping of health diaries is not without problems, this
procedure was felt to be a significant improvement over study designs relying
on recall over a 3-month or 1-year period. A subset of this population (161
persons) provided a total of 1,298 throat and stool specimens for bacterial
and viral analyses. In addition, 318 persons provided paired blood specimens
obtained at the beginning and end of the study period and these were used to
determine prevalence and incidence of infections with enteroviruses . It
should be emphasized that the participants knew only that this was a study of
the possible health effects of air pollution--they were not told of plans to
correlate effects with the sewage treatment plant.
The project also included microbial aerosol monitoring and meteoro-
logical data collection. These data were used to generate personal exposure
indices for each household. The environmental data were then integrated with
the health data to determine any associations with the treatment plant source.
Regression analyses were performed between total viable particle exposure
indices and self-reported illness rates, pathogenic bacteria isolation rates,
prevalence rates of virus antibody, and virus antibody titers. An attempt was
also made to determine if various subpopulations were at risk to infection.
Regression analyses between illness rates and exposure indices were run with
reference to length of residence, age, smoking, presence of young children in
the family, chronic respiratory disease, and chronic gastrointestinal prob-
lems .
The results from all of these analyses were negative. No associations
were found between any of the health factors and the treatment plant as an
exposure source. However, the investigators cautioned that the overall con-
clusion that the plant had no obvious health effect on residents must be
tempered by the small number of people who were exposed to the highest pol-
lution levels.
Summary of Wastewater Treatment Plant Aerosol Studies
The primary health effect findings of the above four studies are sum-
marized in Table 4. All of these studies had good designs and were performed
by competent investigators. The positive findings in Tecumseh were indeter-
minate—they correlated with socioeconomic status as well as with distance
from the sewage treatment plant. In Schaumburg, higher gastrointestinal
symptom and skin disease rates occurred in those nearest the plant. However,
significant decreases of some symptoms and diseases also occurred in all three
zones. The interpretation of these results is clouded by obtaining the self-
reported illness information through a long-recall survey instrument and the
73
-------
inherent variability and inaccuracy of that technique. However, objective
measurements, through pathogen isolation from clinical specimens and through
serology, were negative. In the Tigard study, the exposure was low, the
population at risk was small, and absenteeism is not necessarily indicative of
symptoms or illness. The investigators in the Skokie study cautioned against
overinterpretation of their negative results because of low exposure and
small population.
HEALTH EFFECTS STUDY OF SEWER AND SEWAGE TREATMENT PLANT WORKERS
One limitation of the wastewater treatment plant studies of community
populations has been low exposure. Residential areas in these studies were
generally 400 m or more from the aerosol source and were not necessarily in a
predominant downwind direction. Also, it is difficult to estimate the amount
of exposure residents are subjected to in their homes near a treatment plant.
Presumably, the population x^ith greatest direct exposure to wastewater
pathogens would be sewer maintenance and sewage treatment plant workers. With
this idea in mind, the University of Cincinnati initiated a study (2,9) in
1974 of wastewater workers. The study subsequently continued for more than
five years and additional analyses are still being done.
More than 500 workers in three cities (Cincinnati, Chicago, and Memphis)
were recruited. The workers were divided into three broad categories: in-
experienced and experienced wastewater exposed, and controls. A total of 336
workers remained with the study for the minimum 12-month requirement (Table
5). Inexperienced workers were those just beginning employment. To be placed
in the experienced category, a worker had to have been on the job for a minimum
of two years. The control groups consisted of highway maintenance workers in
Cincinnati, water treatment plant workers in Chicago, and utility workers in
Memphis.
Health monitoring included maintenance of an illness diary, examination
of employer absentee and illness records, annual multiphasic physical
examinations, and pathogen isolation attempts from stool specimens and throat
swabs. Blood specimens were collected quarterly for subsequent serological
analyses. A serologic survey was also conducted on the families of 82
wastewater and 41 control workers to determine possible associations with
transmission of infectious agents to the home from the job. Limited aerosol
and wastewater monitoring for indicators and pathogens was conducted in an
attempt to refine exposure categorization.
The results of the illness analyses indicated no significant difference
in illness rates by worker group or by city although gastrointestinal illness
rates were two to four times higher in the inexperienced worker group. Com-
bining the worker groups from the three cities did result in a statistically
significant difference (p = .004) in gastrointestinal illness rates (Table
6). Rates were higher in the inexperienced group and there was no difference
between experienced workers and controls. A seasonal peak during April-June
was observed. The GI illness rates for the inexperienced group were analyzed
74
-------
on the basis of time on the job and age of workers but no significant
differences were detected.
There were no significant differences in virus or bacterial isolation
rates among workers in the three cities although Salmonellae were isolated" from
sewage-exposed workers on six occasions (Table 7). One isolate was from an
inexperienced worker at the time employment began; one was from another worker
after one year on the job, and the remaining four were from experienced sewage
treatment plant workers. One Shigella isolate was obtained from a control
worker. There was a significant difference in parasite isolation rates—10
isolations were made, all from unexposed individuals.
The serologic analysis included a determination of immunoglobulin levels
on the hypothesis that individuals exposed to low levels of microorganisms may
develop higher levels of immunoglobulins. However, they were not found to be
consistently higher in the sewage-exposed workers in any of the three cities.
The virus serologic analysis involved comparing the geometric mean antibody
titers, titer level changes (increases and decreases), and cumulative sero-
conversions among the worker groups in the three cities. A total of 594 com-
parisons were made, and based on chance alone, one might expect about 30 of
these to be significant at the p = .05 level. Twenty-nine significant dif-
ferences were found and they were distributed evenly among the exposed and
control groups.
To improve the chances of detecting an effect, the inexperienced and
control workers were further subdivided into low and high exposure categories
on the basis of job observation and environmental monitoring. The virus sero-
logical results were then analyzed on a city-by-city and on a combined basis.
A total of 510 comparisons were made and 23 of these were found significant.
Nine out of 10 for the city-by-city comparison, and 10/13 for the combined
analysis were in the direction indicating a sewage exposure effect and these
were about equally divided between aerosol-exposed workers and sewage
liquid/solids-exposed workers.
To summarize this study, inexperienced workers reported higher rates of
gastrointestinal symptoms than did experienced workers or controls. These
rates could not be related to a specific agent or exposure. The symptoms were
mild and transitory and did not result in time lost from work. Pathogen
isolations did not indicate any increased risk from sewage exposure.
STUDIES OF WASTEWATER SPRAY IRRIGATION HEALTH EFFECTS
The transmission of sewage pathogens through aerosols at spray irri-
gation sites is a potential route of exposure where effluent disinfection may
be considered as a treatment. In the United States, the lack of suitable
exposed populations at such sites has prevented conducting health effects
studies of spray irrigation. However, studies have been and are being
conducted in Israel where this practice has been in use for many years, and a
study is now in progress in Texas (Table 8).
75
-------
In Che Israeli study reported in 1976 (7), the investigators compared
Ministry of Health communicable disease data from 77 kibbutzim (agricultural
communities) using partially treated nondisinfected oxidation pond effluent
with that from 130 kibbutzim not practicing wastewater irrigation. For cer-
tain infectious diseases, they found incidence rates 2 to 4.3 times higher in
the kibbutzim utilizing wastewater for irrigation (Table 9). The agents of
these diseases are found in wastewater and transmission by this route is
logical. No significant differences were found for diseases not considered to
be transmitted by wastewater, such as streptococcal infections, tuberculosis,
and laboratory-confirmed influenza. In addition, there were no significant
differences in enteric disease rates among kibbutzim during the nonirrigation
season. The investigators recommended disinfection of sewage effluent used
for irrigation near residential areas because of the potential public health
risks.
This study did not provide any evidence for an aerosol route of trans-
mission. The irrigated fields were located 100 to 3,000 m from the
residential areas. It was indicated that pathogens could reach the community
on the bodies and clothes of the field workers when they returned at mealtime
and at the end of the day. The quality of the drinking water was reportedly
good and a food-borne route was discounted because regulations did not permit
use of sewage to irrigate vegetables or other crops for raw consumption.
In an attempt to get more detailed information, another retrospective
study of kibbutzim examined age-illness distribution, the quality of re-
porting, crop types and irrigation schedules, distance of fields to resi-
dences and dining halls, and length of irrigation season. One group of 13
kibbutzim was in a switch category, i. e., they used effluent irrigation for
two consecutive years and then switched to non-effluent sources for another
consecutive two years, or vice-versa. A second group of 68 kibbutzim was
divided into effluent irrigating, effluent use in fish ponds, and non-
effluent irrigating categories.
Two preliminary analyses of the results have been reported thus far
(4,9). In the switch category kibbutzim, a significant increase was found in
the relative risk of enteric disease during effluent-irrigation years only in
the 0-4 age group. In the group of 68 kibbutzim, a slight excess of enteric
diseases was found in kibbutzim using effluent in fish ponds. Although there
was no difference in annual enteric disease rates between effluent and non-
effluent irrigating kibbutzim, there were increased seasonal rates (May-
July), coinciding with the irrigation period, in effluent irrigating kib-
butzim. These rates fell below those in the non-effluent irrigating kibbutzim
in the fall, thus accounting for the similar annual rates. Significant
increases were noted for shigellosis and streptococcal sore throats in
effluent irrigating kibbutzim. Streptococcal sore throats are not considered
to be associated with a wastewater mode of transmission. There was no
relationship of the enteric disease rate to source of effluent (own or
others), size of the irrigated tract, or distance from residences, although an
excess of enteric disease was noted for kibbutzim irrigating with effluent
volumes >5600 m^/year. These investigators also found that numerous kibbutzim
76
-------
in the 1976 study (7) were incorrectly classified as to effluent utilization.
They suggested that no firm conclusions on the degree of health risk should be
based on either that study or the 1981 retrospective study because of the poor
quality of the data (4).
A third study in'progress in Israel is scheduled for completion in 1983.
The quality of data in this prospective study is expected to be much improved
over the previous two retrospective studies, especially for illness re-
porting. In addition, high risk sub-populations such as field workers and
visiting volunteer groups, will be specifically followed serologically and
through illness monitoring.
The only spray irrigation health effects study presently under way in the
United States is being conducted near Lubbock, Texas. About 7.4 MGD of un-
chlorinated secondary effluent from Lubbock is being piped to 3,000 acres of
farmland 18 miles southeast near the town of Wilson. Construction of the
pipeline and installation of 22 center-pivot spray rigs was completed in 1981.
About 450 people, including about 40 persons in the farm families living on-
site, have been participating in a health watch. Baseline environmental and
health data have been collected over a two-year period. Spray irrigation has
started and one year of the same types of data will be collected. The project
is scheduled for completion in 1984.
DISCUSSION
In the studies described above, conscientious efforts were made at site
selection and study design. The projects were run by competent investigators
from respectable institutions. In the wastewater treatment plant studies,
some effects were noted, but they could not be conclusively associated with
the treatment plant source. These studies all had two major limitations that
make it difficult to attach any significance to either the positive or the
negative findings: (a) low numbers of highly exposed persons, and (b) the
inability to adequately and quantitatively determine that exposure. There is
presently no suitable indicator for airborne pathogens from a sewage source
and populations are subject to exposure to the same pathogens through other
routes in the community.
The Israeli report in 1976 (7) appeared, at first glance, to produce some
clear evidence of health effects associated with spray irrigation of waste-
water. The results led the investigators to recommend disinfection of
wastewater applied near residential areas. Subsequent investigations by one
of the original authors have not been able to confirm those findings. In fact,
many of the kibbutzim were incorrectly classified with regard to wastewater
usage. Also, it has been discovered that a number of the kibbutzim may exceed
even the liberal Israeli drinking water standard of 10 coliforms/100 ml. If
there is a wastewater-related effect, it may be due to contamination of
drinking water or to person-to-person transmission peculiar to the kibbutz
communal way of life.
The two current studies of spray irrigation health effects should
77
-------
provide useful additional information to impact the decision on whether or not
to disinfect wastewater. However, these results will not be available for up
to two and a half years. In a discussion of his paper on bacterial aerosols
at a spray irrigation site, Sorber (8) indicated that terminal disinfection
would be more effective and economical than buffer zones if such were
considered necessary. He concluded that a safeguard of some type would be
prudent until an adequate public health risk assessment can be made. Until
such time, it may be necessary to consider each particular situation on a
case-by-case basis.
LITERATURE CITED
1. Carnow, B., et al. 1979. Health effects of aerosols emitted from an
activated sludge plant. EPA-600/1-79-019, U.S. EPA, Cincinnati, Ohio.
2. Clark, C.S., et al. 1981. Health risks of human exposure to wastewater.
EPA-600/1-81-069, U.S. EPA, Cincinnati, Ohio.
3. Fannin, K.F., et al. 1978. Health effects of a wastewater treatment
system. EPA-600/1-78-062, U.S. EPA, Cincinnati, Ohio.
4. Fattal, B., et al. 1981. Study of enteric disease transmission
associated with wastewater utilization in agricultural communities in
Israel. In: Proceedings, Water Reuse Symposium II. AWWA Research
Foundation, Denver, Colorado.
5. Johnson, D.E., et al. 1978. Health implications of sewage treatment
facilities. EPA-600/1-78-032, U.S. EPA, Cincinnati, Ohio.
6. Johnson, D.E., et al. 1979 Environmental monitoring of a wastewater
treatment plant. EPA 600/1-79-027, U.S. EPA, Cincinnati, Ohio.
7. Katzenelson, E., I. Brium and H.I. Shuval. 1976. Risk of communicable
disease associated with wastewater irrigation in agricultural settle-
ments. Science, 194:944-946.
8. Sorber, C.A. 1977 Author's response to discussion of "A study of
bacterial aerosols at a wastewater irrigation site." JWPCF, 49:1919-
20.
9. U.S. EPA. 1980. Wastewater aerosols and disease. H.R. Pahren and W.
Jakubowski (eds.). EPA-600/9-80-028, U.S. EPA, Cincinnati, Ohio.
78
-------
Table 1. Health Effects Studies of Populations Near
Activated Sludge Treatment Plants
Plant Location
Tecumseh, MI
Schaumberg, IL
Tigard, OR
Skokie, IL
Exposed
Population
Community
Community
Grade School
Community
Reference
3,9
5,9
6,9
1,9
Table 2. Partial Listing of Health Survey Results
From the Schaumburg, IL Study
Percentage Incidence
Disease or Symptom
Skin disease
0-2 kma
2-3.5 km
3.5-5 km
Chest pain on deep breathing
0-2 km
2-3.5 km
3.5-5 km
Diarrhea
0-2 km
2-3.5 km
3.5-5 km
General Weakness
0-2 km
3.5-5 km
Nausea
0-2 km
3.5-5 km
Vomiting
0-2 km
2-3.5 km
Baseline
0.
1.
1.
0,
1.
1
4
4
4.
0
1
1
3
1
3
.5
.6
.4
.5
.6
.4
.1
.3
.8
.7
.5
.2
.0
.3
.0
Operational
1
1
1
1
1
1
7
4
4
1
0
3
1
3
1
.7
.3
.4
.9
.5
.1
.6
.8
.3
.9
.6
.0
.7
.1
.4
Participants lived within the indicated distance from the
plant
79
-------
Table 3. Absenteeism at Durham Elementary
and Control Schools
School
Durham
Controls
Absenteeism (%)
Preoperational3 Operational'-'
5.36 4.67
4.96 4.64
Change
-0.69
-0.32
aLast 2 years prior to plant operation
2 years of plant operation
Table 4 Summary of Health Effects From Four
WWTP Epidemiological Studies
Study
Health Effect
Comment
Techumseh, MI
Schaumberg, IL
Tigard, OR
Skokie, IL
Higher respiratory,
GI illness within
300 m of plant
Higher GI & skin
disease rates within
1 km of plant
Negative
Negative
Indeterminate cause;
socioeconomic confounders
Long-recall survey;
objective measurements
negative
Small population; low
exposure; absenteeism,
not illness
Small population; low
exposure
Table
City
Cincinnati
Chicago
Memph i s
Total
5 . Number o
Study a
Inexperienced
35
38
27
100
f Workers Remaining in the
Minimum of 12 Months
Worker Group
WWEa Experienced WWE
94
35
0
129
Controls
61
27
19
107
aWWE = wastewater exposed
80
-------
Table 6. Seasonal Comparison of Gastrointestinal Illness
Rates for Combined Three-City Groups
Illness/100
Season Inexperienced WWEa
Jan . -Mar .
Apr . -June
July-Sep .
Oct .-Dec .
3.
5.
2.
2.
6
7
9
6
Worker-Months Exposure
Experienced WWE
1.8
2.0
2.0
1.7
Controls
1.3
1.6
1.9
1.4
aWWE = wastewater exposed
Table 7. Salmonella and Shigella Isolations
From Workers
Worker Group
Inexperienced WWEa
Experienced WWE
Controls
No. of
Salmonella
2b
4
0
Isolations
Shigella
0
0
1
aWWE = wastewater exposed
"One isolate from initial employment specimen; one
isolate from another worker after 1 year on job
£1
-------
Table 8. Health Effects Studies of Populations Near
Wastewater Spray Irrigation Sites
Location
Israel
Israel
Israel
(1976)a
(1981)
(1983)
Exposed
Population
Community
Community
Community; workers;
Reference
7
4,9
-
Lubbock, TX (1984)
volunteer groups
Farmers; rural and
town populations
aDate of completion or expected completion of study
Table 9. Summer Incidence of Infectious Diseases in
Kibbutzim With and Without Spray Irrigation
Disease
Shigellosis
Salmonellosis
Infectious hepatitis
Typhoid fever
WW Irrigation
1002
234
88
11.6
Incidence/100,000
(A) No WW Irrigation (B)
455
63
44
2.7
A/B
2.2
3.7
2.0
4.3
82
-------
REQUIREMENTS FOR WASTEWATER DISINFECTION AS SEEN FROM THE RESULTS OF
EPIDEMIOLOGICAL-MICROBIOLOGICAL STUDIES
Victor J Cabelli, Ph.D.
Department of Microbiology
University of Rhode Island
Kingston, Rhode Island 02881
ABSTRACT
The United States Environmental Protection Agency in 1976 abandoned its
policy requiring universal microbial standards for municipal wastewater
effluents discharged into fresh and marine waters and, hence, the requirement
for universal disinfection of these effluents. It was replaced by a policy
in which the microbial limits and the need for and level of disinfection
are determined on a case by case basis. A flow diagram, with feed-back
loops, of the informational needs in making such decisions is presented.
It starts with a target area criterion and ends with the balance between
treatment and disinfection and outfall location.
One of the needs (also a feed-back for risk acceptability), a site-
specific, cumulative frequency distribution of swimming-associated illness
rates, was obtained for beaches along the New York Bight. The inputs to the
model used in making these predictions were the illness (gastroenteritis) -
indicator (enterococcus) regression equations obtained from the bathing
beach epidemiological program and the frequency distribution of enterococcus
densities at sampling stations near the beaches. The rates for "posted"
and "open" beaches were then compared to the predicted enterococcus densities
and illness rates (calculated by the application of the regression equation
to the enterococcus densities in primary and secondary sewage effluent)
following various treatment, initial dilution and subsequent transport decay
options. This preliminary analysis indicated that, with primary treated
effluents, disinfection would probably be required in the absence of the
option for long, deep ocean outfalls and that, in many, if not most, situ-
ations, this would also be true of secondary effluents.
INTRODUCTION
There would be no argument against universal disinfection of waste-
water effluents to levels which virtually eliminate all pathogenic micro-
organisms therein if there were a relatively inexpensive, energy-efficient,
facile, reliable and effective disinfection system which produces minimal
or no adverse ecological or human health effects. In fact, there was a
time in the early 1970's when universal disinfection (specifically chlorina-
tion) of wastewater effluents was considered a reasonable requirement by
the United States Environmental Protection Agency (USEPA) as a means of
attaining one of its objectives, to make all waters "swimmable and
83
-------
fishable1' (8). The corollary to this requirement was that the coliform
and fecal coliform limits for the target (stream standards) would be applied
at the source (effluent). This objective, as applied to the microbial
target area standards or guidelines most commonly used by the various states
or recommended by the Federal Government (16) could be achieved even in
primary treated effluents by chlorination to "reasonable" levels ( 18).
Moreover, the need for relatively restrictive microbial standards for
shellfish growing waters and hence for sewage disinfection, at least in
certain circumstances, was firmly supported by the history of shellfish-
associated outbreaks of disease during the preceding several decades (22)
although the epidemiological information in support of microbial standards
for bathing waters was more limited and less compelling (12). Finally, it
was generally accepted that coliforms were a reasonable surrogate for
salmonellae and the other "important" pathogens as regards the effectiveness
of disinfection, although there was some evidence that at least one of
the agents in question was viral (22)and that viruses were generally more
resistant to chlorination than the coliforms (13).
However, by 1975, it had become clear that, for a number of reasons,
the requirement for universal disinfection was no longer realistic. First,
it was shown that adverse ecological effects could be, and presumably were,
produced from the chlorination of sewage effluents (l9) although the
quantitative relationship of the levels producing adverse ecological effects
to those required for "adequate" disinfection were not defined. Second,
the demonstration of the carcinogenicity of some chlorinated organics
produced during the chlorination of municipal wastewaters raised the
possibility of adverse human health effects from the movement of these
compounds up through aquatic food chains. Third, the energy crisis of 1973
increased the awareness that both the quantity of energy required to produce
the chlorine and its cost could not be ignored in decisions on how,'when,
and where to chlorinate. Fourth, additional data were obtained showing
that coliforms were not a good surrogate for viruses with regard to
chlorination (20). Finally, although the early findings from an epidemi-
ological program conducted by the USEPA clearly showed health effects (a
gastroenteritis) consequent to swimming in waters having relatively low
indicator densities, they also suggested that the etiological agent was
viral and indicated that total and fecal coliforms were defective as
recreational water quality indicators (6). In effect, chlorination did not
meet the requirements of a wastewater disinfection process which could be
applied universally; and, at that time, there was no practical alternative.
Because of the considerations noted above, in 1976 the USEPA reversed
its position and abandoned the requirement for universal effluent standards
and, hence, universal disinfection. It was replaced by a policy whereby
the requirement for and extent of disinfection would be made on a case by
case basis with regard to all the factors involved. Some of us within the
Agency who advocated the change also did so on the grounds that the
existing policy was conceptually deficient. First, there was no provision
for the effects of dilution, sedimentation or biological decay (die-off)
in reducing the levels of pathogenic or indicator microorganisms between
84
-------
the effluent source (outfall) and the potential targets (bathing beaches,
shellfish growing areas, and raw drinking water inlets). Second, as noted
above, the requirement for universal disinfection derives from uniform source
and, hence, target indicator standards. The promulgation of a single standard
applied on a nationwide basis does not provide for local input on risk
acceptability.
INFORMATIONAL NEEDS FOR WASTEWATER DISINFECTION
The achievement of the balance between the need for wastewater dis-
infection and its undesirable consequences, along with the change in USEPA
policy, made it even more important to obtain the data bases needed in
determining the required level of wastewater disinfection on a case by case
basis. The informational requirements for doing so are illustrated in
Figure 1. The starting points are health effects criteria. They are
mathematically expressible relationships between the predicted rates of
infectious disease among the users of the sewage impacted aquatic resources
and some measures of the qualities of the resources. The resources in
question are bathing beaches (including areas used for water skiing,, surfing
and other direct contact activities), shellfish growing areas, and the raw
sources of drinking water. Of necessity, the criteria are generalizations
which should be reasonably applicable over extended periods of time to large
geographic areas since the cost of their development is considerable.
However, both temporal and spacial variability in the relationships can
occur due to several factors (e.g. the incidence of illness in the "dis-
charging" population, the immune status of the users)(3). Moreover,
relationships based upon fecal indicator densities in waters impacted by
small wastewater discharges are not reliable (3). Only one such criterion
is currently available, that for saltwater bathing beaches, although a
similar one for fresh water bathing beaches will be described in the next
paper. The marine recreational water quality criterion was developed from
a series of prospective epidemiological studies conducted over multiple
years at several locations in the United States. It relates the incidence of
swimming-associated gastroenteritis to the enterococcus density in the
bathing water. The importance of these two specific inputs with regard to
wastewater disinfection will be considered later in this paper (4).
Guidelines and standards can be derived from a criterion once a
decision is made as to incidence of illness which is considered acceptable
("acceptable risk") as illustrated in Figure 2. This decision has economic,
sociological and political inputs at both the national and local levels.
The guidelines and standards for all the potentially impacted targets in the
area can then be translated into effluent standards using as inputs estimates
of the physical and biological decay of the pathogenic microorganisms or
appropriate surrogates during transport between the source and the potential
targets. The final decision concerns the trade-off between treatment and
disinfection and outfall location needed to achieve the effluent standard
for that specific pollution source.
85
-------
There are three feedback loops in the system (indicated by broken lines
in Figure 1). First, the choice of the outfall location will influence the
physical and biological decay inputs needed in the translation of target to
source standards. Second, the costs and consequences of wastewater treatment
and disinfection and outfall location can be inputs towards determining the
acceptable risk of disease among the users of the impacted resources. Third,
once the wastewater treatment, disinfection and disposal system is in opera-
tion, the decisions on the acceptable risk can be reexamined and modified
from information on the frequency distribution of indicator densities at
the target, resource usage, and the illness-indicator relationship.
OUTPUT FROM USEPA EPIDEMIOLOGICAL STUDIES
Four necessary pieces of information were obtained from the USEPA
epidemiological-microbiological program to develop recreational water
quality criteria (4,7) — the next paper will describe a fifth. The
first is the illness (swimming-associated gastroenteritis) — water quality
(enterococcus density in the bathing water) regression line. It predicts
the former (Y) from the latter (X). The formula for the regression line
shown in Figure 3 is Y= 12.25 loginX + 0.073. The second was information
on the "best" indicator of those examined. It was defined as the one whose
mean densities in the bathing water correlated the best with the swimming-
associated rates of gastroenteritis. Table 1 shows the correlation co-
efficients (r) for four of the most commonly considered indicator systems.
By this criterion enterococci was the best indicator. The third was the
criterion (regression line and its confidence limits) itself. It predicts
the mean enterococcus density in the water (X) from the "acceptable"
swimming-associated gastroenteritis rate (Y) . The formula is
log.._X = 0.0456 Y + 0.677. The fourth was a membrane filter method for
enumerating the enterococci which does not require the picking of colonies
for identification (14). It was subsequently simplified even further (9).
SITE-SPECIFIC ILLNESS RATE PREDICTIVE MODEL AND ITS USE AT NEW YORK
BIGHT BEACHES
A model for predicting the swimming-associated gastroenteritis rates
(the second feedback loop noted above) was developed and applied in a study
of bathing beaches along the New York Bight sponsored by Marine Ecosystems
Analysis, National Oceanic and Atmospheric Administration. The detailed
findings are being prepared for publication. Two of the three inputs to the
model as noted earlier were obtained as follows. The illness-indicator
regression line was obtained from the USEPA epidemiological study. The
distribution of enterococcus densities E. coli data also were obtained
by the mTEC method (10) was obtained from assays performed in the
author's laboratory from water samples collected by the USEPA. They were
collected by helicopter during the summers of 1980 and 1981 from just
beyond the surf zone at 78 sampling stations located from Cape May, New
Jersey around to the Shinnicock Inlet, Long Island, N.Y.
86
-------
Indicator Density Frequency Distributions
The cumulative frequency distributions of enterococcus densities at
some New Jersey, New York City (Staten Island and Coney Island), and Long
Island sampling stations are shown in Figure 4. The number of values for
each station varied from 17-26. However, at most of the New Jersey and Long
Island stations, many of the enterococcus densities were below the sensi-
tivity of the assay method, 0.5 per 100 ml. This made it difficult to fit
straight lines to the distributions as shown. The E. coli densities were
generally higher than those for the enterococci, especially at the New York
City stations (data not shown).
Illness Rate Frequency Distributions
Each of the enterococcus density estimates per 100 ml (X) was used
to predict a swimming-associated gastroenteritis rate/1000 persons (Y)
using the formula given earlier. The cumulative frequency distributions
of Y corresponding to the indicator distributions for some selected
stations are shown in Figure 5. Enterococcus densities <. 0.5/100 ml
yielded negative gastroenteritis rates; these were recorded as Os. There
were 27 New Jersey and 8 Long Island stations where no more than one
positive Y value was obtained. With rare exceptions, the distributions of
illness rates predicted from the E. coli densities were higher than those
predicted by the enterococci, although the slopes of the latter generally
were greater than those of the former (data not shown).
Three percentile values for the predicted illness rates were selected
as being especially informative and useful (75, 90, and 95) in that they
could provide an individual some idea of the risk of gastroenteritis
incurred while swimming at a particular beach. The rates are presented
for the stations already considered and a few others in Table 2. For
example, the prediction is that, at station J-93 near Wildwood, the
gastroenteritis rate will not exceed 11.3/1000 swimmers more than 5 per-
cent of the time, 6/1000 more than 10 percent of the time, and 0.0 more
than 25 percent of the time. The comparison of the 75 and 95 percentile
values provides some idea of the relative slopes of the distributions
and hence the constancy of the risk from time to time. This can be seen
from the comparison of the values fox stations J-93 to J-97, LI-2 to
LI-4 and Si-Sou to CI-MB. It is of interest that the 75 percentile rates
exceeded 0.1/1000 at only seven stations J-97, Si-Sou, CI-35, CI-29,
CI-20 LI-4, and LI-16. South Beach on Staten Island and W. 35th Street
on Coney Island are posted as unsafe for swimming. The data presented
would suggest no greater justification for closing South Beach than
that for those beaches at 29th and 20th Streets on Coney Island. How-
ever, as noted earlier the acceptability of risk has other than illness
inputs. LI-16 is at an inlet to Great South Bay, and there apparently
are some marginally treated discharges near Cape May, N.J. (J-97).
The disparity in the illness rates predicted from the enterococcus
87
-------
and E. coli densities can be seen from Table 3, which compares the predicted
rates at the 35 (27 + 8) stations noted earlier.
Beach usage data were obtained from the project and are still being
analyzed. Once the output, the seasonal number of swimmers at each beach,
is obtained, the more useful information for the managers of the bathing
resources will become available, i.e. the predicted annual number of cases
of swimming-associated cases of gastroenteritis at the three percentile
levels for each beach area.
PREDICTED DISCHARGE OPTION ILLNESS RATES
Not all the necessary inputs for making the decisions on the required
level of wastewater disinfection are available and some must be determined
from site specific data (e.g. decay coefficients). However, some insight
can be obtained by an examination of the information presented against one
further input, the enterococcus densities in primary and secondary treated
sewage. The mean log densities for the influents and the primary and
secondary treated effluents as determined at a number of sewage treatment
plants in Rhode Island were 5.45, 5.32, and 3.94, respectively. Table 4
shows the expected densities following initial dilutions at the "boil" of
1:10, 1:50, and 1:100 followed by reductions of 90, 99, and 99.9 percent
(1,2, and 3 orders of magnitude, respectively) during transport between
the boil and the target. The residual densities then were used to predict
the mean swimming-associated gastroenteritis rates (Table 5) using the
appropriate illness-indicator regression equation.
Comparison of Predicted Discharge Option and Bight Beach Illness Rates
These mean rates predicted from the treatment-discharge options can
then be compared to the 75 percentile values (much less the 50 percentile
values) for those stations associated with beaches which are and are not
posted as being unsafe for swimming according to the local guidelines and
standards. The beaches and "associated" 75 percentile values for those
beaches which are posted are South Beach, Staten Island (1.69) and W. 35th
Street, Coney Island (3.76). The beaches with high 75 percentile values
which are not posted are Wildwood (3.38), W. 29th Street, Coney Island (1.69),
W. 20th Street, Coney Island (4.27), 67th Street, Rockaways (2.23), and
Cedar Island Beach, Long Island (2.23). If a 75 percentile value of 2.7 is
used as the "break-point" (the average of the two posted and of the five un-
posted beaches/stations), then one could make the following inferences. With
primary sewage effluent, it would appear that initial dilutions slightly in
excess of 1:100 and/or subsequent reductions slightly in excess of 99.9 per-
cent would be required as an alternative to disinfection. These can only be
obtained from the discharge through long distance, deep outfalls such as
those in place along the Pacific Coast. With rare, if any, exceptions, this
alternative is not available along the Eastern Seaboard or the interior.
The alternative would be the sacrifice of some nearby and not so near
resources. With secondary treated effluents, subsequent reductions of
88
-------
slightly less than three logs, more than two, and less than two would be
required with initial dilutions of 1:10, 1:50, and 1 :100 respectively.
DISCUSSION
The predicted, beach specific, swimming-associated illness rates
presented herein were meant to demonstrate the use of the model as one means
of evaluating the water quality at bathing beaches under existing conditions.
As noted earlier, the output information can then be used in risk assess-
ment and, as required, the modification of the specific treatment, dis-
infection, and disposal strategies for the wastewater discharges reaching
the beaches. The accuracy of the specific predictions made herein was
limited by the quality of the input data; and this was due to logistic con-
straints on the intensity of the sampling effort. First, the sampling
stations should be chosen with regard to the spacial distribution of the
swimmers at the beach; this was neither logistically feasible nor necessary
in the present study. Second, depending on the length of the beach, two
or more samples should be collected; in some instances, at least, the
samples could be pooled prior to assay. Third, at least 25 samples should
be collected from each sampling station. However, all the above deficiencies
can be rather easily corrected in a local effort of more limited geographic
scope, especially since the membrane filter assay method for enterococci
is relatively facile. Moreover, because of the nature of the gastroenteritis'
enterococcus regression line, the assay sensitivity need not exceed 1/100 ml.
There also are a number of conceptual limitations on the use of any fecal
indicator in predicting water-related health effects; and these were
considered in an earlier publication (3).
The adequacy of even the enterococci as a health effects water quality
indicator also needs to be addressed. There can be little doubt that it was
the best of those indicators examined in USEPA epidemiological—microbio-
logical studies with regard to an illness (gastroenteritis) whose etiological
agent(s), in all probability, was viral (5). Furthermore, there is increas-
ing evidence that enterococci better simulate the survival characteristics
of certain viruses than do the coliforms, at least in sludge (1) and during
transport in marine waters (21); and coliforms are much more sensitive to
chlorination than most animal and bacterial viruses (19). First} in the
one epidemiological study in which the presumed source of the etiological
agent was at the greatest distance away (in time), the swimming-associated
rates of gastroenteritis were disproportionately high relative to the entero-
coccus densities (4,5). Second, one explanation for the relatively high
illness rates associated with very low enterococcus densities (Figure 3)
is differential survival of the etiological agent and the indicator. The
third cause for concern is related to the second and will, become apparent
from the next paper. Fourth, coliphages (notably male-specific phages
such as f-2 and Fd) were not examined as possible viral surrogates of the
viral pathogens — they are not called a fecal indicator because they are
consistently found in sewage but not feces (15) -- in the epidemiological
studies because of methodological problems. This was unfortunate since
89
-------
field data collected in our laboratory (15) showed that the coliphages were
much more resistant to chlorination than coliforms, chlorination kinetic
studies with phage stocks conducted by Scarpino (20) showed coliphages,
especially the RNA male-specific phages f-2 and MS-2, were more resistant
than E^ coli, and some additional studies conducted in our laboratory (17)
with phage stocks and the viruses as found in sewage showed that the DNA
male-specific phages were even more resistant than the RNA male-specific
phages.
Five additional data inputs are needed in determining the level of
wastewater disinfection required on a case by case basis. The first, a
criterion for fresh recreational waters has been developed and will be
described in the next paper. The second is a similar criterion for shell-
fish growing waters. The third is the development and evaluation of the
technology needed for obtaining biological decay coefficients at specific
locations (i.e. with consideration to so-called "backyard effects") and
without recourse to open water, in situ studies. These would be used as
inputs to physical transport models which have been or are being developed.
Incidentally; Clostridium perfringens spores appear to be an excellent
conservative tracer for the conduct of such studies both for the examination
of the water column (2) and the underlying sediments (11). The fourth is
more information on the killing kinetics for enterococci in sewage during
chlorination and other disinfection procedures. The fifth is a cost-
benefit or cost-effectiveness model to be used in determining the accept-
able risk of the water-related diseases.
LITERATURE CITED
1. Berg, G. and D. Berman. 1980. Destruction by anaerobic mesophilic
and thermophilic digestion of viruses and indicator bacteria
indigenous to domestic sludges. Appl. Environ. Microbiol. 39:361-368.
2. Bisson, J.W. and V.J. Cabelli. 1980. Clostridium perfringens as an
indicator of water pollution. Jour. Water Poll. Control Fed.
52:241-248.
3. Cabelli, V.J. 1978. New standards for enteric bacteria. In: Water
Pollution Microbiology. Ed. R. Mitchell, Wiley, New York. p. 233-271.
4. Cabelli, V.J. 1980. Health Effects Criteria for Marine Recreational
Waters. EPA- 600/1-80-031, U.S. Environmental Protection Agency,
Washington, D.C., September, 132 pages.
5. Cabelli, V.J. 1980. Epidemiology of enteric viral infections. In:
M. Goddard and M. Butler eds. International Symposium on Viruses and
Wastewater Treatment. Pergamon, London, p. 291-304.
90
-------
6. Cabelli, V.J., A.P. Dufour, M.A. Levin, L.J. McCabe, and P.W. Haberman.
1979. Relationship of microbial indicators to health effects at
marine bathing beaches. Am. J. Publ. Hlth., 69:690-696.
7. Cabelli, V.J., A.P. Dufour, L.J. McCabe, and M.A. Levin. 1980. Swim-
ming-associated gastroenteritis and water quality. J. Epidemiol. 115:
(in press).
8. Congress of the United States. 1972. Amendment to Federal Water
Pollution Control Act: Public Law 92-500. Federal Register, 86 Stat.
816 p. 1 Oct. 17.
9. Dufour, A.P. 1980. A twenty-four hour membrane filter procedure
for enumerating enterococci. Abs. Ann. Meet. Amer. Soc.
Microbiol. p. 205
10. Dufour, A.P-, E.R. Strickland, and V.J.Cabelli. 1981. Membrane filter
method for enumerating Escherichia coli. Appl. Environ. Microbiol.
41:1152-1158. ~
11. Emerson, D.J. and V.J. Cabelli. 1981. Use of Clostridium perfringens
in marine sediments to monitor the deposition and movement of sewage
particulates. Third International Ocean Disposal Symposium, (in press)
12. Henderson, J.M. 1968. Enteric disease criteria for recreational
waters. J. San. Eng. Div. 94:1253-
13. Kelly, S. and W.W. Sanderson. 1958. The effect of chlorine in water
on enteric viruses. Am. J. Publ. Hlth. 48:1323-1334.
14. Levin, M.A., J.R. Fischer, and V.J. Cabelli. 1975. Membrane filter
technique for enumeration of enterococci in marine waters. Appl.
Microbiol. 30:66-71.
15. Lupo, L.B. 1979. Bacteriophage as Indicators of Fecal
Pollution. M.S. Thesis, Department of Microbiology, University
of Rhode Island.
16. Mechalas, B.J., K.K. Hekimian, L.A. Schinazi, and R.H. Dudley. 1972.
An Investigation into Recreational Water Quality. Water Quality
Criteria Data Book. 4 vol. 18040 DAZ 04/72 Environmental Protection
Agency, Washington, D.C.
17. McBride, G. 1979. A Bacteriophage Simulant for Enteric Virus Behavior
in Water Systems. MS. Thesis, University of Rhode Island.
18. Miescier, J.J. and V.J. Cabelli. 1982. Enterococcus and other
microbial indicators in municipal sewage effluents. Jour. Water
Poll. Control Fed. (in press).
91
-------
19. Roberts, M.H., Jr., R.J. Diaz, M.E. Bender, and R.J. Haggett. 1975.
Acute toxicity of chlorine to selected estuarine species. J. Fish.
Res. Board Can., 32:2525-2528.
20. Scarpino, P.V., G. Berg, S.L. Chang, D. Hahling, and M. Lucas. 1972.
A comparative study of the inactivation of viruses in water by
chlorine. Water Res. 6:959-965.
21. Vasl, Robert. 1978. The Isolation and Identification of Enteric
Viruses from Coastal Waters in Israel. M.S. Thesis, Department of
Human Environmental Sciences, Hebrew University, Jerusalem, Israel,
September, 32 pages.
22. Verber, J.L. 1981. Shellfish Borne Disease Outbreaks Internal Report,
Northeast Technical Services Unit, Food and Drug Administration,
Davisville. Rhode Island.
Table 1. Correlation Coefficients (r) for Gastro-
enteritis Against Mean Indicator Density
New York City Study 1973-1975
r-Values by
Indicator Summ. Density
Enterococci
E . coli
Fecal Coli forms
Total Coliforms
.75
.52
-.01
.19
.96
.56
.51
.65
Trial days grouped by summer.
2
Trial days grouped by indicator density.
92
-------
Table 2. Predicted Rates of Swimming-Associated Gastroenteritis at
Some New York Bight Beaches
Station
Location
Rate/1000 Swim, at Perc.1
75 90 95
J-24
J-75
J-81
J-93
J-97
SI-SB
CI-35
CI-29
CI-20
CI-MB
LI-2
LI-4
LI-8
LI-12
LI-16
LI-18
LI-26
Ocean Grove
Atlantic City
Ocean City
Wildwood, N. Wildwood
Cape May City
South Beach
Coney I si, W. 35th
Coney Isl, W. 29th
Coney Isl, W. 20th
Coney Isl, Manh. B.
Riis. Pk. , Rockaways
92nd St, Rockaways
Long Beach
Jones Beach
Cedar Island Beach
Great South Beach
Tiana Beach
0.00
0.00
0.00
0.06
3.38
1.69
3.76
1.69
4.27
0.00
0.00
2.23
0.00
0.00
2.23
0.00
0.00
0.07
4.95
0.07
5.98
9.39
6.5.1
7.05
8.04
9.34
2.26
4.84
6.33
2.23
0.07
5.63
0.05
3.76
4.71
5.92
1.03
11.23
11.19
12.15
17.01
16.57
16.98
12.34
11.11
7.27
4.27
0.07
5.92
4.22
10.43
•'-Swimming-associated gastroenteritis rate (Y) for given percentiles;
calculated from applying regression equation Y= 12.25 log^g x + 0.073
where X is the observed distribution (N= 17-26) of enterococcus
densities/100 ml at indicated sampling station.
Table 3. Comparison of Swimming-Associated Gastroenteritis Rates Pre-
dicted from the Distributions of Enterococcus and E. coli
Densities
General Area
New Jersey
Long Island
No.1
Stns
27
8
Indicator
Used
Entero.
E. coli
Entero.
E. coli
Rate/1000
75
0.0
6.6
0.0
5.8
Pers. at
90
0.01
9.3
0.01
8.3
Perc.2
95
0.05
10.5
0.04
9.0
•'-Only includes stations where 95th percentile, swimming-associated
gastroenteritis rate predicted from enterococcus densities did not exceed
0.05/1000 persons.
Table 2 for calculation of rates: formula for calculating
E. coli rates, Y= 6.32 log^Q X + 5.71; values given are the
averages from all the stations.
93
-------
Table 4. Calculated Residual Enterococcus Following
Hypothetical Reductions Due to Initial Dilution
and Decay During Transport
Treatment
Initial
Dilution
Log1Q Residual/100
Transport Reduction
90%
Primary
(5. 3D1
Secondary
(3.94)1
1:
1:
1:
1:
1:
1:
10
50
100
10
50
100
3.
2.
2.
1.
1.
0.
31
61
31
94
24
94
99%
2
1
1
0
0
-0
.31
.61
.31
.94
.24
.06
ml after
of
99.9%
1
0
0
-0
-0
_i
.31
.61
.31
.06
.76
.06
Mean log..,, enterococcus density/100 ml in sewage (18);
influent density 5.45.
Table 5. Predicted Mean Swimming-Associated Gastroenteritis
Rates/1000 Persons for Residual Enterococcus
Densities Given in Table 4
Treatment
Primary
Se condary
Initial
Dilution
1:10
1:50
1:100
1:10
1:50
1:100
Predicted Illness Rate (Y)1
After
90%
40.6
32.0
28.3
23.8
15.3
11.6
Transport
99%
28.3
19.8
16.1
11.6
3.0
0.0
Reduction of
99.9%
16.1
7.5
3.9
0.0
0.0
0.0
Gastroenteritis predicted from the formula Y= 12.25 X + 0.073,
where X= log.., enterococcus density/100 ml and Y is in cases
per 1000 swimmers.
94
-------
HEALTH EFFECTS
CRITERION ,
^FT
J C_ 1
.— - -
N i
<* — ACCEPTABLE -*-
RISK
. — — — — .
" SOCIAL
ECONOMIC 1
. POLITICAL
i
1
1
1
1
1
BEACH USAGE |
^
1
STANDARD INDICATOR
OTHFR ^
TARGETS
RCE
>
, ..„ ^ HFAITH IMPAPT
(SEASONAL NO.
OF CASES)
DISTRIBUTION
i
I
1
1
1
1
1
1
i
DILUTION
-X Of— Ptlklf— KITATI^KI .^_ . . ^« ««
-I 1
DIE-OFF . ,
rn^T
J
4
ECOL.ft HEALTH
EFFECTS
OUTFALL LOCATION
TREATMENT
DISINFECTION
1 I
1 1
| 1
1 '
_l '
._ 1
Figure 1. Information flow scheme for case-by-case decision making on the
need for wastewater disinfection.
95
-------
t
LU
(T
CO
CO
LU
Q
UJ
<
O
O
CO
CO
O
co WATER QUALITY INDICATOR DENSITY
Figure 2. Relationship of criteria to guidelines and standards,
96
-------
HCGI
STUDY
YR.
A NEW YORK CITY 1973
4 NEW YORK CITY 1974
A NEW YORK CITY 1975
A LAKE PONTCHARTRAIN 1977
LAKE PONTCHARTRAIN 1978
BOSTON HARBOR 1978
MEAN ENTEROCOCCUS
DENSITY/IOOml
Figure 3. Relationship of the rates
credible gastrointestinal
enterococcus densities in
more details).
for swimming-associated, highly
symptoms (gastroenteritis) to the mean
the water (see references 4 and 7 for
97
-------
>-
55
Ui
O
O
O
O
O
O
EC
UJ
i \ T i r ] i i
• JC 75
o JC 97
& Staten Is. - So. Beach
a Staten Is -Midland Beach
a Coney Is. - W 35th St.
+ LI-2
18 -
S ff '4
^ O 12
82
« ."^ '°
IS '
• JC 75
° JC97
* Stolen Is -So Beach
A Staten to.-Midland Beach
o Coney Is.-W 35th St. * ~
+ LI-2
9 10 2O3O4O5O«OTOeO «O»S
CUMULATIVE FREQUENCY (•/•)
2O 304O50607080 9O 95 96
CUMULATIVE FREQUENCY (%)
Figure 4. Cumulative frequency distributions for
enterococcus densities at some New York
Bight sampling stations.
Figure 5. Cumulative frequency distribution of
swimming-associated gastroenteritis
rates predicted from enterococcus density
distributions at some New York Bight
sampling stations.
-------
FRESH RECREATIONAL WATER QUALITY AND SWIMMING-ASSOCIATED ILLNESS
Alfred P. Dufour
Toxicology & Microbiology Division
Health Effects Research Laboratory
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
ABSTRACT
Prospective epidemiological-microbiological studies were carried out at
two freshwater bathing beaches, one at Erie, Pennsylvania and the other at
Tulsa, Oklahoma. The purpose of these studies, which covered a two-year
period, was to: 1) Examine the relationship between swimming-associated
gastrointestinal illness and freshwater quality; 2) Determine if the water
quality criteria established for marine bathing beaches are applicable to
freshwater beaches; 3) Investigate the relationship between waterborne,
microbe-bearing particulates and swimming-associated health effects.
Swimming-associated gastrointestinal illness was observed at freshwater
beaches. In general, significantly greater illness rates occurred at barely
acceptable beaches than at the relatively unpolluted beaches. The swimming-
associated rate of gastrointestinal illness observed in freshwater swimmers
was found to be appreciably lower than that observed in marine water swimmers.
Finally, the preliminary evidence indicates that there may be a relationship
between microbe bearing particulates and gastrointestinal illness.
Freshwater epidemiological-microbiological studies indicate that water
quality criteria established for marine recreational waters may not be
applicable to fresh recreational waters.
INTRODUCTION
Discussions about the need to regulate recreational water quality or
chlorinated wastewater effluents that may ultimately reach recreational waters
invariably lead to the question of whether or not wastewater-contaminated
surface waters have the potential to cause illness in swimmers (23,28,34).
This uncertainty about recreational waters being the vehicle of transmission
for pathogens that cause enteric disease in swimmers has persisted because much
of the evidence supporting the relationship was far from conclusive. The basis
of this doubt can be found in the list of some of the most frequently referenced
swimming-associated disease outbreaks shown in Table 1. Only two of the
outbreaks present a reasonably strong case supporting the premise that the
observed illnesses were due to pathogens from wastewater effluents. One was
the Walmer outbreak in 1909 in England where young recruits swam in a pool
filled with seawater contaminated with effluents from a nearby sewage
treatment plant (29). The other was the Dubuque, Iowa outbreak which occurred
99
-------
in 1979 (32). Thirty-one individuals were reported ill with shigellosis and
the only common factor to all the illnesses was swimming in the Mississippi
River. Water samples from the river, examined some days after the peak of the
outbreak, were found to contain high densities of coliforms. The suspected
pathogen was also isolated from the water. However, it was not unequivocably
established that a sewage treatment plant 17 miles upstream was the source of
the causative agent. In the other outbreaks the linkage between sewage-
contaminated water and swimming-associated illness was quite tenuous. The
United States reports of illness in swimmers were not very well documented,
especially for the early outbreaks (18,19,29,30). Similarly, the evidence in
the Australian (2,14) and French (12) outbreaks was not conclusive.
Since the study of outbreaks was found to be an unsatisfactory means to
show that swimming in polluted water is a health hazard, the epidemiological
approach was examined. Table 2 lists four epidemiological studies that have
been conducted since 1950 The conclusion of the 1959 retrospective study
reported by Moore (27) was that an association between poliomyelitis and
swimming in poor quality water as a causal factor could not be found. This
finding has frequently been used to justify the case against regulating
recreational water quality and disinfecting wastewaters. However, the
proponents of this position seldom take into account that negative findings in
retrospective studies should not be interpreted to mean that the relationship
does not exist, but rather that the case is not proven. The results of the 1981
retrospective study conducted by D'Alessio et al. (11) clearly show the
correctness of this interpretation. They found an increased risk of
enterovirus-caused illness in children who swam in lake water. The risk of
illness due to swimming in wastewater-contaminated waters was further sub-
stantiated in the two prospective studies listed. The United States Public
Health Service studies reported by Stevenson (37) in 1953 concluded there was
a risk of enteric illness associated with swimming in polluted fresh waters.
However, these studies have been criticized on a number of issues, such as the
adequacy of study design and the way in which the data were analyzed. As a
result of these criticisms the United States Environmental Protection Agency
(EPA) initiated a series of epidemiological-microbiological studies at marine
bathing beaches that were designed to correct the major deficiencies of the
studies reported by Stevenson. The results of the EPA studies have been
reported by Cabelli (8). These studies have unequivocably established that
there is a risk of gastrointestinal illness associated with swimming in
polluted recreational waters and that this risk increases as the water quality
decreases. Furthermore, the studies showed that the enterococcus group was the
most efficient indicator of marine recreational water quality from which a
prediction of the rates of swimming-associated illness can be made.
Although the criterion or model established by the Cabelli studies offers
conclusive evidence that marine recreational waters contaminated by sewage
effluents are a vehicle for transmitting enteric illness to swimmers, some
questions remain to be answered. For instance, it is not known whether the
model established with marine bathing beach data is applicable to freshwater
beaches. Another obvious question is why do statistically significant
swimming-associated illness rates occur in apparent high quality water. An
100
-------
example of this is the significant swimming-associated rates of highly
credible gastrointestinal illness observed when the density of Escherichia
coli in the water at New York City beaches was only 14 per 100 ml (8). The data
to be presented here will attempt to shed some light on both of these questions.
Freshwater Epidemiological Studies
Freshwater studies similar to those conducted at marine beaches were
initiated in 1978. Several freshwater sites were surveyed to determine their
potential for full-scale epidemiological-microbiological studies. Two sites
were found to be suitable. One at Keystone Lake, a man-made lake 15 miles from
Tulsa, Oklahoma and the other on Lake Erie at Erie, Pennsylvania. Each site had
two beaches whose beach-going populations were demographically similar, but
whose water quality was significantly different. Two groups of investigators,
one at the University of Oklahoma led by Dr. James Robertson and the other at
Gannon University led by Mr. Stan Zagorski, were supported through grants from
the Environmental Protection Agency in 1979 and 1980 to carry out the epidemi-
ological-microbiological studies. The data presented here were supplied by
the respective principal investigators who are preparing manuscripts for
publication that will describe the studies in detail.
Experimental Procedures
Although the epidemiological protocol has been described elsewhere
(8,9,10), a brief summary of the illness inquiry sequence of events is given in
Table 3. The freshwater trials closely followed the procedures used during the
marine beach studies to insure comparability of the data. However, the method
of data analysis had to be modified because the swimming activity of freshwater
swimmers differed from that of marine swimmers. Freshwater beach goers, unlike
those at marine beaches, had a tendency to do a great deal of swimming and
therefore only a limited number of non-swimming beach goers at any one beach
were available to serve as control subjects. Since the non-swimming beach
goers at each study site were demographically similar, the non-swimmers from
both beaches at each study site were combined to form a single control group.
The follow-up telephone survey obtained information on a number of
symptoms that might have occurred during the 9- to 10-day interval between the
swimming activity and inquiry Gastrointestinal symptoms and disabling
information were used to generate two variates reported in the marine water
studies. Definitions of the variates are shown in Table 4. Since symptoms were
self-diagnosed, multiple symptoms or unmistakably recognized symptoms were
used to establish the credibility of the gastrointestinal illness. These were
called "highly credible" symptoms. Swimming-associated illness rates were
determined by subtracting either the total or "highly credible" symptom rates
observed in the non-swimming control groups from the respective symptom rates
observed in the swimming groups. These swimming-associated symptom rates were
used to establish health effects/water quality relationships from the fresh-
water data.
101
-------
Microbiological Methods
Multiple indicators were examined at both sites, but only three, .E. coli,
enterococci, and fecal coliforms, will be considered here. The methods for
enumerating E. coli and enterococci have been described earlier (13,22). Fecal
coliforms were monitored using a standard method (1).
Particulate Study Procedures
In the second year of the bathing beach studies at Lake Erie, the group at
Gannon University was asked to conduct a small pilot study to determine if an
association exists between particles larger than three microns and the
incidence of gastrointestinal symptoms in individuals swimming on the day the
measurements were made. The usual epidemiological and microbiological
variates were measured during the course of each trial and two new micro-
biological measurements also were determined. The first of these was the
density of particles three microns or larger that were associated with 15. coli
colonies. The second was the average number of E. coli per particle. Figure
1 is a flow-chart diagram of the procedure used to determine the two
characteristics. Each water sample was divided into two parts. One part was
treated in the usual manner. The other was filtered through a three micron pore
size Nuclepore filter. The bacteria on the particles retained by the filter
were desorbed and dispersed by blending in a buffered surfactant solution
(24,33). The E. coli in the desorbed bacterial suspension and the filtrate
were enumerated on MTEC Medium (13) after refiltering each through a 0.45
micron filter (Gelman, GN6). The number of particles associated with E. coli
colonies was determined by subtracting the density of E_. coli found in the
filtrate from the density obtained using the customary technique. It was
assumed that the E. coli in the filtrate were non-particle associated cells.
The number of E. coli per particle was obtained by dividing the total number of
E_. coli desorbed from the particles by the total number of particles associated
with E. coli colonies.
RESULTS
The number of participants in the Oklahoma and Pennsylvania recreational
water quality studies and their "highly credible" G.I. symptom rates are shown
in Table 5. The mean indicator densities per 100 ml for enterococci, E. coli
and fecal coliforms for each swimming season are included in the table. The
average swimming-associated illness rate in freshwater swimmers for all of the
trials was 6.2 per 1000 individuals. In contrast, the average "highly
credible" swimming-associated illness rate in marine water swimmers was 14.8
per 1000 individuals (data obtained from reference 8). The difference between
these two swimming rates was shown to be statistically significant (p<0.05)
using the Wilcoxon Rank Sum Test (36). Correlation coefficients and lines of
best fit were calculated using freshwater indicator densities as the in-
dependent variable and swimming-associated illness as the dependent variable
and these were compared to similar statistics from the marine studies (8). The
relationship of enterococci density to "highly credible" G.I. illness for
fresh and marine waters is shown in Figure 2. The slope of the line of best fit
102
-------
for the marine bathing beach data is about twice that observed with the
freshwater data (11.6 versus 6.1). The correlation coefficients (r), on the
other hand, are similar in magnitude (0.71 versus 0.65). The regression lines
describing the relationship of "highly credible" G.I. symptoms to E_. coli
densities are shown in Figure 3. The slope of the line for the marine data is
again greater than that observed for the freshwater data; however, the
difference is much less than that for enterococci (7.3 versus 4.7). The
correlation coefficient for the freshwater points is approximately equal to
that obtained for the marine water points (0.514 versus 0.513). The regression
lines for highly credible G.I. illness on fecal coliform densities present an
interesting contrast to the relationships observed with enterococci and E.
coli. Figure 4 shows that the slopes of the lines calculated from marine (7)
and freshwater data are very flat (3.2 and 2.0, respectively). Since the
correlation coefficients are in part a function of the magnitude of the slope,
they too have small values (marine = 0.15, fresh = 0.23).
It was shown in the marine recreational water quality studies that of all
the bacteriological water quality indicators examined, enterococci had the
best correlation to the health effects observed in swimmers (8). E. coli were
ranked second and fecal coliforms ranked eighth among eleven indicators
studied. A similar ranking was observed in the freshwater studies. The three
freshwater health effects/water quality indicator regression lines shown in
the previous figures are compared directly in Figure 5. If the three health
effect indicator relationships are ordered according to the correlation
coefficients of their regression lines, enterococci would clearly rank first,
E. coli second, and fecal coliforms would rank third.
The results of the pilot study conducted to determine if particles were
related to swimming-associated illness are shown in Table 6. The health
effects data are given in terms of total gastrointestinal symptoms rather than
"highly credible" G.I. symptoms because the frequency of occurrence of the
latter was less than one on many of the trial days. The risk attributed to
swimming was calculated as described by Rimm, et al. and these are shown in the
first column of Table 7 (3). These values are ranked in increasing order, and
the companion water quality indicator and particle data collected on the same
trial day are shown in columns 2, 3, and 4. The relatedness of the attributable
risk to the E. coli density per 100 ml, to the density of E. coli-associated
particles at least three microns in size, and to the density of E. coli per
particle was determined using Spearman's rank-difference correlation co-
efficient (31). A comparison of the correlation coefficients is shown in Table
8. The number of E. coli associated particles per 100 ml had the highest degree
of relatedness to the swimming-associated risk. The E_. coli per 100 ml also had
a positive correlation to swimming-associated risk but are about one-third the
magnitude of that found with particle density. The correlation coefficient for
attributable risk relative to E. coli density per particle was a relatively
large negative value, indicating an inverse relationship between these two
variables.
103
-------
DISCUSSION
The results of the freshwater bathing beach studies are significant in at
least three respects, all of which may be important to those interested in
wastewater disinfection. The first aspect is that the direct relationship
between swimming-associated gastrointestinal illness and water quality ob-
served at marine bathing beaches was also found at freshwater beaches. This
finding was not unexpected since Stevenson (37) observed a detectable risk for
gastrointestinal illness in freshwater swimmers. However, the new data
confirm the fact that as the quality of bathing water deteriorates, the risk of
gastrointestinal illness increases. This information will be very useful for
establishing water quality criteria. The second notable aspect of the
freshwater studies is that, as in the marine studies, the enterococci correlate
best with gastrointestinal illness. The superiority of enterococci over E_.
coli and fecal coliforms as an indicator of recreational water quality is most
likely a reflection of their ability to survive better in aquatic environments
(4) and also because of their greater resistance to the effects of chlorination
(25). Enterococci also have been shown to be less sensitive than EL coli to the
effects of solar radiation (15,35). The attributes of this indicator
frequently have been overlooked because of methodological considerations and
its lower density in fecal wastes relative to coliforms or fecal coliforms.
However, it has been proposed in the past as a water quality indicator for
recreational waters (16,21) and perhaps the time has come that its use be given
serious consideration.
The most conspicuous aspect of the freshwater bathing beach studies is the
low swimming-associated gastrointestinal illness rates relative to those ob-
served in marine water swimmers at equivalent indicator densities. This
difference in illness rates is probably a function of dissimilar indicator die-
off patterns in the two swimming environments. Mitchell and Chamberlain (26)
have pointed out that the time interval for 90 percent die-off of coliforms was
approximately 52 hours in freshwater and only about two hours in seawater.
This appreciable die-off rate difference between coliforms in marine and
freshwater environments may well account for the observed differences in
swimming-associated illness rates. It is assumed that the similarity in the
symptoms of marine and freshwater swimmers is due to the same or closely
related enteric pathogens. The difference in illness rates is probably the
most significant finding of the freshwater study, since it will preclude the
use of a single criterion for marine and fresh recreational waters.
Although the swimming-associated illness data may prove useful for
establishing effluent guidelines, it is the data dealing with microbial laden
particles and swimming-associated risk that may hold the greatest interest
relative to wastewater disinfection processes. The rationale for the particle
experiments was based on the conceptual hypothesis that a particle of fecal
material may contain thousands or millions of bacterial or viral pathogens and
therefore the swallowing of a single particle by a swimmer would be sufficient
to initiate an infection. This hypothesis is supported by data indicating that
polioviruses encapsulated in fecal material are more resistant to chlorination
than nonencapsulated viruses (20). It also is known that particle-associated
bacteria and viruses survive longer than non-particle-associated bacteria and
104
-------
viruses (5,6,17). Therefore, if it could be shown that a health effect was
directly related to either the number of E. coli per particle or the density of
particles associated with E. coli, then it would not be difficult to infer that
pathogens behave similarly and are responsible for the effect. The results of
the small pilot study conducted at the Lake Erie beach indicated that the
density of particles containing E. coli appeared to be more closely related to
the observed health effects than either the density of E_. coli per 100 ml or the
number of E. coli per particle. This result implies that a high probability of
ingesting a single particle is more important than the average number of viable
pathogens per particle. This factor may provide an answer to the question,
"Why does significant swimming-associated illness occur in good quality
water?" If E. coli are valid surrogates for pathogens in feces, then the
ingestion of a single particle containing multiple infectious units could
easily account for the observed effects. The findings also suggest that the
swimming-associated illness rate could be lowered by some type of intervention
at the treatment level. Further studies, to confirm these preliminary data,
are being planned by the investigators at Gannon University.
CONCLUSIONS
The Environmental Protection Agency-supported studies at freshwater
bathing beaches during 1979 and 1980 have produced a great deal of data. The
small part of that data which has been presented here leads to the following
conclusions.
1. There is a risk associated with swimming in freshwater and this risk
is proportional to the quality of the water.
2. The bacterial water quality indicator that correlates best with
swimming-associated gastrointestinal illness is the enterococcus
group.
3. The swimming-associated gastrointestinal illness rate in freshwater
swimmers is significantly lower than that observed in marine water
swimmers at equivalent bacterial indicator densities. This dif-
ference rules out the establishment of a single water quality
criterion for both fresh and marine bathing beach waters.
4. The appreciable swimming-associated gastrointestinal illness rate
that occurs in good quality water may be due to the presence of
particles which contain high densities of pathogens.
LITERATURE CITED
1. American Public Health Association, Standard Methods for the Examination,
of Water and Wastewater, 14th Ed. 1976. Am. Public Health Assoc.,
Washington, D.C.
105
-------
2. Anonymous. 1961. Typhoid Traced to Bathing at a Polluted Beach. Public
Works, 9_2, 182-183.
3. Basic Biostatistics in Medicine and Epidemiology. 1980. A.A. Rimm, A.J.
Hartz, J.H. Kalbfleisch, A.J. Anderson and R.G. Hoffmann. Appleton-
Century-Crofts , New York.
4. Bianchi, A.J.M. and M.G. Bensoussan. 1977. Non-Marine Bacteria in Dialy-
sis Bags in Seawater. Marine Pollution Bulletin, J5, 282-284.
5. Bitton, G. and R. Mitchell. 1973. Effect of Colloids on the Survival of
Bacteriophages in Seawater. Water Res. , 8^, 227-229.
6. Bitton, G. and R. Mitchell. 1974. Protection of E. coli by Montmoril-
lonite in Seawater. J. Environ. Eng. Div., ASCE, 100, 1310-1320.
7. Cabelli, V.J. 1979. Recreational Water Route of Disease Transmission:
United States Studies. International Symposium on Health of Liquid
Waste Disposal, High Institute of Public Health, Alexandria, Egypt,
June 4-7.
8. Cabelli, V.J. 1980. Health Effects Criteria for Marine Recreational
Waters. Environmental Protection Agency., EPA-600/1-80-031, Cincin-
nati, Ohio.
9. Cabelli, V.J., A.P. Dufour, M.A. Levin, L.J. McCabe and P.W. Haberman.
1979. Relationship of Microbial Indicators to Health Effects at
Marine Bathing Beaches. Am. J. Public Health, 69, 690.
10. Cabelli, V.J , M.A. Levin, A.P. Dufour and L.J. McCabe. 1974. The Dev-
elopment of Criteria for Recreational Waters. In: International
Symposium on Discharge of Sewage from Sea Outfalls, H. Gameson (Ed.),
Pergamon, London, England, pp. 63-73.
11. D'Alessio, D.J., T.E. Minor, C.I. Allen, A.A. Tsiatis and D.B. Nelson.
1981. A Study of the Proportions of Swimmers Among Well Controls and
Children With Enterovirus-like Illness Shedding or not Shedding an
Enterovirus. Am. J. Epidemiology. 113, 533-541.
12. Denis, F.A., E. Blanchouin, A. DeLignieres and P. Flamen. 1974. Cox-
sackie AK, Infection From Lake Water. J. Amer. Med. Assoc., 228, 1370-
1371.
13. Dufour, A.P., E.R. Strickland and V.J. Cabelli. 1981. Membrane Filter
Method for Enumerating Escherichia coli. Applied and Environmental
Microbiology, 4J., 1152-1158. '
14. Flynn, M.J. and D.K.B. Thistlethwayte. 1964. Sewage Pollution and Sea
Bathing Advances in Water Pollution Research. Proc. 2nd Intl. Conf.,
Vol. 3, pp. 1-25.
106
-------
15. Fujioka, R.S., H.H. Hashimoto, E.B. Siwak and H.F. Reginald. 1981. Ef-
fect of Sunlight on Survival of Indicator Bacteria in Seawater.
Applied and Environmental Microbiology, 41, 690-696.
16. Garber, W.F. 1956. Bacteriologic Standards for Bathing Waters. Sew-
age and Indust. Wastes, 28, 795-808.
17. Gerba, C.P. and G.E. Schaiberger. 1975. Effect of Particulates on Virus
Survival in Seawater. J. Wat. Poll. Cont. Fed. , 47, 93-103.
18. Gorman, A.E. and A. Wolman. 1939. Water-borne Outbreaks in the Unit-
ed States and Canada, and Their Significance. J. Amer. Water Wks., 31,
225-275.
19. Hawley, H.B., D.P. Morin, M.E. Geraghty, J. TomkowandA. Phillips. 1973.
Coxsackievirus B Epidemic at a Boys' Summer Camp. J. Amer. Med.
Assoc. , 226, 33-36.
20. Hejkal, T.W., F.M. Wellings, P.A. LaRock and A.L. Lewis. 1979 Survival
of Poliovirus Within Organic Solids During Chlorination. Appl. and
Environmental Microbiol., 38, 114-118.
21. Lattanzi, W.E. and E.W. Mood. 1951. A Comparison of Enterococci and E.
coli as Indices of Water Pollution. Sewage and Industrial Wastes,
^3:1154-1160.
22. Levin, M.A., J.R. Fischer and V.J. Cabelli. 1975. Membrane Filter Tech-
nique for Enumeration of Enterococci in Marine Waters. J. Applied
Microbiol. , _30, 66-77.
23. Levin, M.A., A.P. Dufour and W.D. Watkins. 1980. Significance of Waste-
water Disinfection to Health Effects Observed in Swimmers. In: Water
Chlorination, Environmental Impact and Health Effects, Jolley, R.L.,
Brungs, W.A., Gumming, R.B. and Jacobs, V.A. (Eds.), Ann Arbor Sci.,
Ann Arbor, Michigan, Vol. 3.
24. Lockman, H.A., M. Meskill and C.D. Litchfield. 1980. Comparison of Tech-
niques for Enumerating Bacteria in Polluted Coastal Sediments. Abst.
Ann. Meeting, Amer. Soc. Microbiol., p. 192.
25. Ludovici, P.P., R.A. Phillips andW.S. Jeter. 1975. Comparative Inacti-
vation of Bacteria and Viruses in Tertiary-Treated Wastewater by
Chlorination. In: Disinfection Water and Wastewater, Johnson, J.D.
(Ed.), Ann Arbor Science, Ann Arbor, Michigan.
26. Mitchell, R. and C. Chamberlain. 1978. Survival of Indicator Organisms.
In: Indicators of Viruses in Water and Food, Berg, G. (Ed.), Ann Arbor
Sci. Publ., Inc., Ann Arbor, Michigan.
27. Moore, B. 1959. The Risk of Infection Through Bathing in Sewage-Polluted
107
-------
Water. In: Proc. 1st Intl. Conf. on Waste Disposal in the Marine
Environment, Pearson, E.A. (Ed.), Pergamon Press, N.Y., pp. 29-37
28. Moore, B. 1975. The Case Against Microbial Standards for Bathing Bea-
ches. In: Discharge of Sewage from Sea Outfalls, Gameson, H. (Ed.),
Pergamon, London, pp. 103-109.
29. Moore, B. 1954. Sewage Contamination of Coastal Bathing Waters. Bull.
of Hygiene, _29_. 689-704.
30. Morbidity and Mortality Weekly Reports. 1979. Gastroenteritis Associ-
ated with Lake Swimming. Center for Disease Control, 28, 413-416.
31. Nonparametric and Shortcut statistics. 1957. M.W. Tate and R.C. Clel-
lend. Danville, Illinois, Interstate.
32. Rosenberg, M.L., K.K. Hazlet, J. Schaefer, J.G. Wells and R.C. Pruneda.
1976. Shigellosis from Swimming. J. Amer. Med. Assoc., 236, 1849-
1852.
33. Scheraga, M., M. Meskill and C.D. Litchfield. 1979. Analysis of Methods
for the Quantitative Recovery of Bacteria Sorbed Onto Marine Sedi-
ments. In: Methodology of Biomass Determinations and Microbial
Activities in Sediments, Litchfield, C.D. and Seyfried , P.L. (Eds.),
ASTM STP 673, American Society for Testing and Materials, pp. 21-39.
34. Shuval, H.I. 1975. The Case for Microbial Standards for Bathing Beaches
In: Discharge of Sewage from Sea Outfalls. Gameson, H. (Ed.),
Pergamon, London, p. 95.
35. Sieracki, M. 1980. The Effects of Short Exposures of Natural Sunlight on
the Decay Rates of Enteric Bacteria and a Coliphage in a Simulated
Sewage Outfall Microcosm. Masters Thesis, University of Rhode Island.
36. Some Rapid Approximate Statistical Procedures. 1964. F. Wilcoxon and
R.A. Wilcox. Lederle Laboratories, Pearl River, N.Y.
37. Stevenson, A.H. 1953. Studies of Bathing Water Quality and Health. Amer.
J. Public Health, 43, 529-538.
108
-------
Table 1. Swimming-Associated Enteric Disease Outbreaks
Year
1909
1921
1932
1936
1958
1973
1974
1978
1979
Country
Type of
Water
England (29)* Sea
U.S.A. (29)
U.S.A. (29)
U.S.A. (18)
Australia
(1,14)
U.S.A. (19)
France (12)
U.S.A. (32)
U.S.A. (30)
"Reference number in
Table 2.
Type of
Study
Epidemiological
Year
Sea
Sea
Fresh
Sea
Fresh
Fresh
Fresh
Fresh
parenthesis
Disease
or Agent
Typhoid
Typhoid
Typhoid
Typhoid
Typhoid
Coxsackie B
Coxsackie A
Shigellosis
Enteritis
Studies of Swimming-Assoc
Etiologic
Agent
Retrospective 1959 Poliovirus
Swimming
Illness
No
Water
Quality
Poor
Poor
Poor
Unknown
Poor
Unknown
Poor
Poor
Unknown
iated Illness
Water
Quality
Variable
(27)x
Retrospective
(11)
1981
Enterovirus
Yes
Good
Prospective 1951 Unknown
(37)
Yes
Variable
Prospective 1972 Unknown
(8)
Yes
Variable
"Reference number in parenthesis
109
-------
Table 3. Sequence of Events for Epidemiological-Microbiological Trials
Day of Week
Saturday
Sunday
Monday
Monday
Day
1
2
3
10
Activity
Beach interview,
sample water
(same as above)
Reminder letter
Phone interview
a.
b.
c.
d.
a.
a.
Function
Obtain Personal Data
Reject Pre-Trial Midweek
swimmers
Query on beach activity
Assay of water samples
(same as above)
Reminder to note illness
Obtain illness information
b. Reject post-trial midweek swim-
mers
c. Obtain remainder of demographic
information
-------
Table 4. Definition of Total and Highly Credible G.I. Health Effects
Health Effects Variates Definition
Total G.I. Symptoms Any one of the following:
vomiting, nausea, diar-
rhea or stomachache
Highly Credible G.I. Any one of the following:
Symptoms , . .
1. vomiting
2. diarrhea with fever or
disabling condition"
3. stomachache or nausea
accompanied by a fever
-'-
"indicates individual remained at home, remained in bed or sought
medical advice
-------
Table 5. Highly Credible Gastrointestinal Illness
Rates Among Swimmers and Non-Swimmers at
Freshwater Bathing Beaches
Oklahoma
Beach A Beach B
Pennsylvania
Beach A Beach B
1979
Swimmers
Total No.
Illness Rate^
Non-Swimmers^
Total No.
Illness Rate
1980
Swimmers
Total No.
Illness Rate
Non-Swimmers
Total No.
Illness Rate
2491
25.29
1864
20.92
787
18.53
4503
15.32
3085
12.96
1063
9.11
3248
17.24
2139
14.49
1854
9.17
Enterococci
E . coli
Fecal Coliform
39
138
436
7
19
51
11
23
16
47
2383
13.42
1995
22.06
1532
9.30
Enterococci
E . coli
Fecal Coliform
23
52
230
20
71
234
38
139
37
85
246
104
1000 participants
2Non-Swimmers from Beaches A and B combined to form single
control group
^Density per 100 ml
112
-------
Table 6.
Total Gastrointestinal Symptom Rates
in Swimmers and Non-Swimmers by
Individual Day
Swimmers
Non-Swimmers
Trial Day
1
2
3
4
5
6
7
Number of
Participants
292
126
244
105
140
269
172
Number of
% 111 Participants
9.
5.
7.
9.
7.
8.
2.
2
6
8
5
7
2
9
103
77
76
129
63
88
105
% 111
8.7
3.9
5.3
3.9
3.2
3.4
0.9
113
-------
Table 7. Summary of G.I. Illness, Water Quality and Particle-Related Variates
Risk Attributable
to
Swimming Exposure
4.
21
28.
41 .
47
53.
56,
? "'
7
-1
0
9
1
0
E. coli/
Particle
1406
49
4]9
20
309
157
47
E. coli/
100 ml
141
253
110
567
127
308
200
Particles With
E. coli/100 ml
32
124
65
420
88
172
130
'Percent of illness in swimmers due to swimming exposure
Table 8. Correlation of Swimmer-Associated G.I. Illness
With Water Quality Indicator and Particle-
Related Illness
Comparison
Correlation
Coefficient
G.I. Illness vs. E. coli density
per particle
G.I. Illness vs. E_. coli density
per 100 ml
G.I. Illness vs. density of E. coli
associated particles
per 100 ml
-0.5
.21
.61
-------
FILTER THRU
3 MICRON
MEMBRANE
FILTRATE
FILTRATE THRU
.45 MICRON
MEMBRANE
PLACE MEMBRANE
ON MEDIUM &
INCUBATE
COUNT COLONIES
WATER SAMPLE
(split sample)
FILTER THRU
.45 MICRON
MEMBRANE
PLACE MEMBRANE
ON MEDIUM &
INCUBATE
PARTICLES
COUNT COLONIES
REMOVE PARTICLES
FROM FILTER
DESORB AND DISPERSE
CELLS FROM PARTICLES
FILTER THRU
.45 MICRON
MEMBRANE
PLACE MEMBRANE ON
MEDIUM & INCUBATE
COUNT COLONIES
FILTRATE COUNT
DESORBED COUNT
USUAL COUNT
Figure 1. Sample Treatment Protocol
115
-------
LLI
a
UJ
DC
o «>
>• o
^ e«
I CC
(3 UJ
^ Q-
30
27
24
2 1
018
UJ
cc
rf UJ
ff °-
o I
iu *
H O
< I-
— &,
O s
O x
(I) V)
us
O
z
to
15
1 2
9
marine water —«
fresh water •
1 ' ' 10 50 100
MEAN ENTEROCOCCUS DENSITY PER 100 m!
Figure 2. Comparison of Gastrointestinal Symptom Rates at Marine and
Freshwater Bathing Breaches Using Mean Enterococcus Density
as the Index of Water Quality
500
-------
uu
_i
2
O
UJ
oc
O
O
30
27
24
O uj 2 1
— Q.
tt O
0 O
u. ,_
£s
2 "•
«
S
O
I-
a
S
tn
w _
< o
O
z
Q
UJ
H
<
O
O
1 8
15
12
6
marine water —
fresh water—
10
Figure 3.
50 100
coli DENSITY PER 100 ml
500
MEAN E.
Comparison of Gastrointestinal Symptom Rates at Marine and
Freshwater Beaches Using Mean E. coli Density as the Index
of Water Quality
-------
00
111
_J
OQ
Q
LU
o
a
x
o
u.
L'J
O
O
o
if,
C
O
O
< o
o
30
27
24
2 1
1 8
iu 15
a
w
O
l-
Q.
s
(fi
12
0
20
marine water
fresh water
0.23
100 200
MEAN FECAL COLIFORM DENSITY PER 100 ml
1000
2000
Figure 4.
Comparison of Gastrointestinal Symptom Rates at Marine and
and Freshwater Bathing Beaches Using Mean Fecal Coliform
Density as the Index of Water Quality
-------
15
LU
_l
GO
Q
LU
en w . n
O z 12
5:
LU
a
O
x
LU
<
tr
Q
LU
O
o
w
(A
<
o
z
5
5
(A
cn
5
O
I-
0.
a
9
10
50 100
MEAN INDICATOR DENSITY PER 100 ml
500
Figure 5. Relationship of Enterococcus, E. coli and Fecal Coliform Mean
Densities to Gastrointestinal Symptom Rates at Freshwater
Bathing Beaches
-------
1. OPTIMIZATION OF MIXING FOR DISINFECTION
Karl E. Longley
Consulting Civil Engineer
4106 Nicholas Drive
Visalia, California 93291
ABSTRACT
Rapid bulk diffusion of chlorine solution introduced into a wastewater
stream has been demonstrated to markedly improve disinfection efficiency and
decrease chlorine dose requirement which, in turn, decreases the formation of
deleterious chlorinated by-products. Rapid mixing of chlorine with a waste-
water stream initially provides increased contact between the bacteria and
virus with chlorine before the chlorine is dissipated in other reaction
pathways. Controversy exists concerning the identification of the initial
active disinfecting chlorine species in a well mixed system and further work
is required to identify whether this chlorine species is free chlorine
(hypochlorous acid), a chloroorganic, or some other chlorine species. The
design of chlorine mixing is generally accomplished using empirical procedures
and disinfection models which do not account for system geometry or energy
input into the system. Rational optimization of the mixing of chlorine into
a wastewater stream for disinfection purposes requires consideration of waste-
water quality and flow rate, the flow rate of the chlorine stream and its
chlorine concentration, and the geometry of the mixing process. The Prandtl
eddy frequency and the mean velocity gradient have properties which make them
useful for disinfection process design. Though the Reynolds number quantifies
the amount of turbulence in a mixing system, it is not a universal descriptor
for the rapid mixing, disinfection process. The Collins' model, though not
containing mixing descriptors, was modified on an individual case basis to
describe bacterial inactivation for a disinfection system having greatly
improved mixing.
INTRODUCTION
Effective mixing of chlorine with a wastewater stream is recognized as an
important factor for optimizing a chlorine disinfection process. However,
most wastewater treatment plants employing chlorination add the chlorine as
an aqueous solution through a diffuser at the head of a chlorine contact
basin with little or no effective mixing. Under these transport and reaction
conditions free chlorine is not mixed throughout the mass of the incoming
wastewater. Since the formation of chloramines and other chlorinated by-
120
-------
products at wastewater pH values of 6 to 9 is very rapid, essentially
complete in a few seconds, normal methods of adding chlorine to a wastewater
stream do not optimally mix the chlorine into the wastewater stream, thereby
impairing the disinfection process. For the designer and operator to
rationally optimize the chlorination process, a model must be available for
their use which ideally incorporates parameters describing intensity of
mixing relative to the system's physical geometry, chemical reactions, and
the resultant inactivation of bacteria and other organisms of relevant
health significance. This paper describes work by the author and others to
develop the practice of mixing chlorine into a wastewater stream and to
rationally describe the disinfection process for coliform bacteria.
Stenquist and Kaufman (17) mixed an aqueous chlorine stream at bench-
scale into a wastewater stream by means of multiple source grid placed in a
pipe. The purpose of the grid was to achieve rapid mixing of the chlorine
solution with the wastewater stream. As a control for laboratory studies,
chlorine was introduced through a single inlet in the direction of flow so
that the primary source of turbulence generation for the control was
wall friction. Other conditions were similar. After 0.32 minutes of contact
time, coliform inactivation was 50 - 55 percent for the control (single in-
let) and 97.4 percent for the grid mixer. However, for a given chlorine
dose, detention time varied inversely with chlorine residual yielding similar
amounts of coliform inactivation for both types of reactors. Subsequent
field studies conducted on a plant of approximately 1.7 mgd (6,440 cu m/day)
demonstrated no improvement of a grid chlorine diffuser relative to a diffu-
ser placed in-line. White (18, 19) reported on a survey of the chlorination
facilities of several wastewater treatment plants discharging into San
Francisco Bay. Plants introducing chlorine at a point of turbulence demon-
strated consistently higher coliform removals. Kruse'j3t_ al_. (12) studied
the chlorine disinfection of the secondary effluent (trickling filters) from
a 1.5 mgd (5,680 cu m/day) wastewater treatment plant. Under normal plant
conditions an aqueous chlorine solution was introduced through a diffuser at
the head of the contact basin. Improved mixing was achieved by introducing
the aqueous chlorine feed stream at a point of turbulence in the wastewater
line upstream from the contact basin. Coliform inactivation after 2 and 10
minutes contact time showed no significant increased coliform inactivation
attributable to improved mixing.
Sepp and Bao (15), in a study of seven California wastewater treatment
plants, passed unchlorinated wastewater effluent through an optimized pilot
disinfection system employing turbulent mixing. In each instance they
compared the results of bacterial inactivation achieved with the plant full-
scale disinfection system to the bacterial inactivation achieved with the
optimized pilot plant. Sepp and Bao (15) found that better bacterial
inactivation resulted from better mixing.
Collins and Selleck (3) and Collins et_ al. (4) described the below model
specific for coliform inactivation in a wastewater stream.
N_ = (\ + 0.23 ct)~3 (<])
No
121
-------
where NQ = coliform bacteria density at time zero
N = coliform bacteria density at time t
t = mean detention time (reactor volume divided by the flow
rate), minutes
c = combined amperometric chlorine residual, mg/1.
Collins and Selleck (3) and Collins et al_. (4) also reported that backmixing
appreciably decreased the germicidal potential of the chlorine residual.
They found the effect of initial turbulent mixing on the bactericidal
effectiveness of chlorine introduced into a wastewater stream to be highly
significant.
Haas (8) enumerated the overall sequence of events during chemical dis-
infection as follows:
"1) The disinfectant, either as a solution, or as a gas must be brought
into intimate contact with the wastewater, and mass transfer into the bulk
solution must be allowed to occur.
"2) The disinfectant, entering the bulk liquid, must be transported to
the exterior of the microorganism which is to be inactivated.
"3) The active species, located at the microbial exterior, must be
transported or bound to or at the lethal site.
"4) The microorganism is inactivated at a rate proportional to the con-
centration of the disinfectant species which is in an 'active1 form at the
lethal site.
"5) Simultaneously to the above events, liquid phase decomposition of
disinfectant may occur via the exertion of demand or the formation of less
active species (such as chloramines)."
Longley (13), reporting on disinfection studies carried out at a waste-
water treatment plant located at Fort Meade, MD, found that the mean velocity
gradient and the Prandtl eddy Frequency are descriptors for disinfection
employing rapid mixing.
The Reynolds number (Re) is a dimensionless number relating inertial and
viscous forces. While the intensity of turbulence for a particular system is
directly related to Re, severe limitations exist for using Re as the
criterion to classify in different mixing systems the degree of material
homogeneity (completeness of mixing) which can be attained as a function of
time. A prime limitation is that for a pipe flow system the temporal and
mixing relationships are inversely related to the pipe diameter, whereas Re
is directly related to the pipe diameter.
122
-------
Brodkey (1) observed that the statistical theory of turbulent mixing has
been developed parallel to turbulent motion theory. The basic linear
equation for turbulent mixing is that of mass (or heat) conservation, which
is the counterpart of the nonlinear Navier-Stokes equation for turbulent
motion. The problem of turbulent mixing presents all the difficulties that
turbulent motion does because of the nonlinearity of the governing physical
equations when expressed in terms of averages. Hinze (10) discusses a
phenomenological theory describing the distribution of mean values of a
quantity, such as momentum or mass, by the effect of turbulence. One of
these, Prandtl's theory, has its analogy in the kinetic theory treating the
molecular transport processes of gas which describes the mean free path of
a gas as the average distance a gas molecule travels before striking another.
Davies (5) reports extensively on the development of the Prandtl mixing
length theory and its application to experimental data. Over the core of a
pipe and away from its wall it has been shown experimentally that an empirical
approximation of the velocity profile for Reynolds numbers to 105 is
Vy (center)
(2)
where
Vx (center)
= time-average axial velocity of the flowing fluid
at any point away from the wall
= time-average axial velocity of the flowing fluid
at the pipe center
y - distance from the pipe wall
a = pipe radius .
The above equation may be developed to define the Prandtl eddy frequency, f,
as
V
f = 0.33 m ? where Vm = mean flow velocity.
(3)
The analogy between mass and momentum fluxes is sufficient that the
effective mean eddy length may be approximated as being the same for both
momentum transfer and mass transfer-
Camp and Stein (2) stated that the concepts concerning the mean velocity
gradient, dv/dy, are applicable to all phenomena involving fluid friction
loss. The mean velocity gradient may be determined from the expression
G =
123
-------
where G = mean velocity gradient (dv/dy)
P = power input
•tf- = volume of system through which power is dissipated
jj - dynamic (absolute) viscosity.
Glover (7) has related observations of coliform disinfection by
Collins et al. (3) to the product of the velocity gradient and time of con-
tact (GlT!He credits the GT product as being a good parameter to describe
mixing intensity in a chlorine contact system.
McKee e_t al. (14) found that data obtained from bacterial inactivation
by chlorination, plotted as a function of time, best fit a line described by
N =
No
where N
N
(5)
o
m
= number of organisms surviving at time, t
= number of organisms at initial time, t0
= exponent characteristic of disinfection system
After fitting the data of several sources, they found that the value of m had
a range of -0.8 to -3.8 with an approximate mean of -2. When plotting data
obtained from bacterial inactivation as a function of chlorine dosage they
found two straight line relationships, one fitting data for chlorine dosages
less than about 11 mg/1, and the other fitting data for chlorine dosages
greater than about 11 mg/1. This disparity was explained as possibly re-
lating to the most susceptible coliforms which are not protected by solids.
Horn (11) studied the inactivation of coliforms in stabilization pond
effluent chlorinated at selected doses between 0.25 and 2.0 mg/1. He found
that the reaction kinetic is a complex m and n order reaction which is
dependent on chlorine dosage, contact time and the number of surviving
organisms. He postulated and developed the general model shown below.
dN_ = -KNtmCn
dt
where K ~ first order rate constant
N = number of coliform organisms per unit volume
t = time
C = concentration of applied chlorine dosage
(6)
124
-------
m = reaction rate constant
n = coefficient of dilution
Setting m = n i- 0 results in the below equation which is a form of
Chick's Law.
dN = -KN (6.1)
dt
Setting m = 0 and n = 0, an n-order model results.
dN = -KNCn (6.2)
dt
An inspection of equation (6.2) shows that it is a variant of the C t
relationship used by many investigators where the product of this relation-
ship equals a constant for a given percentage of organism inactivation.
Setting C t = constant = k', then
Cn = k'/t (6.3)
Substituting, setting limits, integrating, converting to log.n, and
developing the above model yields,
log (jj ) = -Kn log (« ) (6.4)
o o
K is determined by plotting log (N/N ) as a function of log t.
Letting m - 0 and n = 0, an m-order and n-order reaction results.
Substituting into equation (6) the expression C = k'/t,
dN_ = -K NtV (6.5)
dt t
Further development of Horn's model yields,
log ({J ) = -kk'tm (6.6)
o m
The rate constant, k, is for log-]g, and m is the reaction kinetic constant
for the m-order and n-order reaction where m -i 0 and n 4 0. The constant
"m" is determined by plotting log [log(N/N x 10' )] as a function of log
(t). °
Eliassen e_t_ al_. (6) and Hess _e_t_ al. (9) have proposed a model relating
coliform densities in sewage as expressed by most probable number, and the
125
-------
chlorine residual as determined by the orthotolidine test. The model is
1 a + bR (7)
log (MPN)
where MPN = most probable number of coliform organisms in 100 ml
a = constant
b = constant
R = the orthotolidine chlorine residual
Data for evolution of the model were obtained from over 100 sewage treatment
plants and represented 5,000 sets of data. A high correlation for the model
was achieved using the available data.
MATERIALS AND METHODS
Studies were carried out at the Fort Meade Sewage Treatment Plant
No. 2. The plant is a conventional trickling filter plant. The chlorine
stream was produced by passing tap water and chlorine gas through an ejector.
The flow rate varied from an approximate minimum of 0.9 mgd (3,410 cu m/day),
which was attained during the late morning or early afternoon hours. Waste-
water streams investigated during the study were primarily those occurring
between the hours of 0900 to 1800 during week days. During these hours 6005
of the secondary effluent was 20-25 mg/1, and the organic nitrogen and
ammonia concentrations were 4-6 mg/1 and 11-15 mg/1, respectively. Optimiza-
tion of mixing in the pilot plant was accomplished using the pipe and Venturi
mixers which were mounted and operated in a trailer located near the chlorine
contact chamber. The trailer was equipped for the conduct of all chlorine
and pH determinations and all coliform assay procedures.
Indigenous coliforms having a median density of 350,000 per 100 ml prior
to disinfection were used as indicator organisms for the bacterial inactiva-
tion studies. The multiple tube fermentation technique given in Standard
Methods (16) was used for determination of total coliform densities. Results
were confirmed using brilliant green bile lactose broth.
For the plant condition studies, composited samples in replicate were
taken in sterile bottles containing sodium thiosulfate. For mixer studies
samples of the mixed stream for bacterial analyses were withdrawn immediately
downstream from the mixer by means of a Cornwall syringe equipped with a
three-way valve. At least two 5-milliliter portions were withdrawn for each
sample and injected directly into a vial containing sodium thiosulfate. The
chlorine contact time from chlorine introduction into the mixer until in-
jection of the sample into the vial was about 2 to 4 seconds. The syringe
was flushed several times with the sewage-chlorine mixture between samplings.
Samples for contact periods of about 15 seconds or greater were collected at
126
-------
the discharge point into the contact basin and were held for the required
time period before neutralization of the disinfectant with sodium thio-
sulfate. Total chlorine and free chlorine residuals were determined, the
latter qualitatively, using modifications of the leucocrystal violet proce-
dure of Black and Whittle (16).
The sewage stream was pumped from the secondary effluent stream into the
trailer and through a rotameter prior to introduction into the mixer. The
chlorine stream likewise passed through a rotameter prior to introduction
into the mixer as the disinfectant stream. The rotameters were calibrated by
a positive displacement technique.
DISCUSSION
McKee's e_t_ al_. (14) proposed mathematical model, previously discussed, is
as follows,
N = (t}
N V
o o
m
The m values have been calculated for different plant and mixing conditions
and are tabulated in Table 1. The value of m, a function of the amount of
inactivation within a designated time period, is therefore also a function of
other variables. A listing of the more important parameters includes the
chlorine stream pH, the mixed stream pH, the chlorine species, the chlorine
dose, and mixing. It is observed that the calculated m values differ over a
wide range, presumableydue to differing disinfection conditions. Thus, this
is a poor model to be used for all but very well defined conditions.
For the case where m = 0, development of Horn's model was shown to yield,
log ( ) = -K log (
o o
(6.4)
K is determined by plotting log (N/N0) as a function of log (t). An
evaluation of Kn for representative mixing data was performed as is shown in
Table 1. Where multiple observations were made as a function of time, the
calculated values of Kn were quite dissimilar between each consecutive time
interval. Thus, it is evident that Horn's model, when m = 0, for the mixing
data is not linear as a function of log (t). The values of Kn vary with a
number of system parameters not included in the model, the most important
being mixing intensity and the pH of the chlorine stream. Therefore, as a
general model this model is inoperative.
Similar conclusions may be drawn for the model where m
which was shown to be expressed as
0, and n ^ 0
log
= -kk't
(6.6)
127
-------
TABLE 1
EVALUATION OF McKEE'S AND HOM'S MODELS EOR COLIFORM DISINFECTION
Test Condition
I. Plant Conditions8' b
17,4 mg/1 dose
II. Plant Conditions8" b
4.5 mg/1 dose
M III. Venturi Mixer8' b
££ 17 mg/1 dose
IV. Venturi Mixer8' b
4.3 mg/1 dose
V, Pipe Mixer 11 in. diam)a» c
17 mg/1 dose
VI. Pipe Mixer (1 in. diam)a» b
4.3 mg/1 dose
Time,t
(min)
2
11
16
2
13
18
.03
.37
.60
15.6
.03
.37
.60
15.6
1.0
15.0
1.0
15.0
Log (t)
0.30
1. 11
1.26
0.30
1.04
1.20
-1.52
-0.43
-0.22
1.19
-1.52
-0.43
-0.22
1.19
0.00
1.18
0.00
1.18
Inactivation
1.6
9.8
1.2
1.4
2.2
3.2
7.9
2.5
7.2
£7.2
2.3
1.7
1.2
1.4
6.5
1.0
1.7
3.8
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
lO-2
10~6
10~5
10-1
lO-2
10~3
10-"
10-*
10~5
10~6
10-1
10-2
10"2
lO-2
lo-1
10°
lo-1
10~3
Log (N/N0)
-1.80
-5.01
-4.92
-0.85
-1.66
-2.50
-3.10
-3.60
-4.14
£5.14
-0.64
-1.77
-1.92
-1.85
-0.19
0.00
-0.77
-2.42
t/to
5.50
8.00
6.50
9.00
12.3
20.0
520.
12.3
20.0
520.
15.0
15.0
McKee's m
-4.32
-3.46
-0.98
-1.72
-0.46
-0.80
-0.75
-1.04
-0.99
-4.48
0.16
-1.41
Horn's KJJ
m°05 riVO
-3
0
-1
-5
-0
-0
-0
-1
-0
-0
0
-0
.97
.60
.10
.25
.46
.26
.71
.04
.72
.05
.16
.62
Horn ' B m
mXO, nXO
-0.55
0.01
-0.39
-1.11
-0.06
-0.29
-0.07
-0.05
-0.17
-0.11
0.61
-0.91
a, Coliform results confirmed.
b. Chlorine stream pH and mixed stream pH were both 7.0.
c. Chlorine stream pH and mixed stream pH were 2.1 and 6.8 ± 0.2, respectively.
-------
Equation (6.6) may be verified by a linear relationship when a plot is
made for log log (N/NQ x 10'1-1) as a function of log (t). The rate constant,
m, is determined from the slope of the relationship. The rate constant, m,
is not linear as a function of log (t) as shown in Table 1. Intuitively
this model has some validity for it expresses the coliform inactivation as a
function of both contact time and chlorine residual. However, once again
the model does not include the important system parameters of mixing
intensity and chlorine stream phi, and therefore it is not ideal as a general
model.
The mathematical model originally proposed by Eliassen e_t a_^. (6) was
shown to be,
1
= a + bR
(7)
log (MPN)
Table 2 contains an analysis of data together with constants reported by
Eliassen et_ _al_. Eliassen's model gives a reasonable estimate of the expected
coliform inactivation if the constants a and b are evaluated for the system
under field operating conditions. As a change in the sewage or chlorine
characteristics will effect a change in the attainable coliform inactivation,
such a change will also change the constants. It should also be noted that
exceptional mixing increases the attainable coliform inactivation and thus
affects the constants and consequently the resulting curve. Eliassen's model
through careful use and evaluation of constants offered a good approximation
of the Et. Meade coliform data.
The mathematical model proposed by Collins et al. (3, 4) was shown to be,
N_ = (1 + 0.23 ct)"
N
:D
The model was applied to data for Et. Meade plant conditions and the Venturi
mixer, and the results are presented in Eigure 1. The Et. Meade No. 2
TABLE 2 DATA ANALYSIS EOR MATHEMATICAL MODEL OE ELIASSEN AND COWORKERS
Contact
Eliassen's Data
Et. Meade Data Analysis
Time, min
2
5
10
11
15
16
16 (Venturi Mixer)
20
Constants
a
.17
.21
.28 1.
.65 1.
b
32
74
04
02
Constants
a b
.20
.27
.38
.28
.014
.14
.22
.72
No. of
Observations
12
ia
11
10
Correlation
Coefficient
.818
.722
.826
.893
129
-------
0
10
10
10
•B O
9
z
I
I Id3
ft:
o
8io4
10
Id6
T~nr
TTTI ! I I I IIII I
SYMBOL SYSTEM
• PLANT CONDITIONS
A VENTURI MIXER
COLLINS et ol. MODEL:
-T3
" [l + 0.23(Ct)]
VENTURI MIXER:
N/No = 0.014 FI + 0,23(Ct)l
PLANT CONDITIONS:
,-3.25
N/N0 = 0.44 |i+0.23(Ct)J
i I I J _LULLLJ
I 111 JJ^ I I
r
I 2 3 4 6 8 10 20 40 6080100
| + 0.23(Ct)
FIGURE 1. Plot of mathematical model for confirmed coliform disin-
fection according to Collins and Selleck (3) and Collins et al. (4),
130
-------
conventional disinfection facilities generally attained a slightly greater
degree of coliform removal than that shown by the Collins' model as
represented in Figure 1.
For conventional disinfection facilities having hydraulic characteristics
approaching plug flow and treating domestic sewage, the Collins1 model may
present a good approximation of the amount of expected coliform removal. The
model has the additional advantage that it contains a term representing con-
tact time. However, the model is very conservative when mixing is optimized
as is shown by the results for Venturi mixer experiments. The Venturi mixer,
with improved mixing of the chlorine stream, achieved significantly greater
initial coliform inactivation of approximately 1.5 logs. The subseguent
disinfection rate achieved using the Venturi mixer was slightly less than the
disinfection rate achieved using conventional plant conditions. The experi-
mental models are shown below.
Venturi model: N_ = 0.014 Pi + .23 (ct)l ~2'77 (fl)
N L J
o
Plant model: N_ = 0.44 fl + .23 (ct)l ~3'25 (9)
N L J
o
Regression coefficients for the Venturi model and the plant model were -0.694
and -0.886, both significant at the 99 percent level.
In order to describe rapid mixing guantitatively for both design and
operational considerations, a good descriptor of the mixing process must be
identified and evaluated with inactivation data. Accordingly, bacterial
inactivation data were evaluated as a function of mean velocity gradient,
Prandtl eddy freguency, and Reynolds number. Generally, log-log transform of
the data yielded the best fit. Data fit, evidenced by the statistics in
Table 3, are best for the higher chlorine dose, 17 mg/1, and for the data
evaluated as a function of mean velocity gradient and Prandtl eddy freguency.
The non-significant correlation coefficients and unremarkable t statistics
for 4.3 mg/1 chlorine dose data expressed as a function of Reynolds number
may be due, in part, to the fact that Reynolds number is a direct function of
the mixer diameter, whereas, the time reguired for chlorine transport across
a transverse section of the mixer is inversely related to the mixer diameter.
Both mean velocity gradient and Prandtl eddy freguency show promise of being
adeguate rapid mixing descriptors to be used in conjunction with the design
and evaluation of disinfection facilities though considerable additional
disinfection data must be evaluated to establish firmly any relationship
which may exist. The mean energy gradient is an easily calculable guantity
which incorporates the design parameters of power, flow rate, and head loss,
the knowledge of which are essential to the designer. However, through the
development of the Prandtl eddy freguency theory and related concepts, re-
lationships may be developed which will incorporate material transport fac-
tors, rather than momentum transfer, and the decay of the free chlorine
species as a function of sewage characteristics. This type of relationship
is necessary to describe adeguately a rapid mix, disinfection system.
131
-------
TABLE 3
REGRESSION ANALYSIS*** OF COLIFORM AND f2 VIRUS INACTIVATION FOR MIXER STUDIES
AS A FUNCTION OF MIXING DESCRIPTORS
Independent Dependent Chlorine Number of Intercept, Regression Coefficient Correlation
Variable, X Variable, Y Dose, mg/1 Observations ao of Y on X, ai Coefficient
Mean Velocity
Gradient
Prandtl Eddy
Frequency
Reynolds
Number
Coliform
Inactivation
Coliform
Inactivation
Coliform
Inactivation
4.3
17
4.3
17
4.3
17
19
17
19
17
19
17
0.57
1.92
0.71
2.26
1.27
6.25
-0.31*
-1.10**
-0.37*
-1.27**
-0.37
-1.68**
0.32
0.85
0.34
0.82
0.11
0.55
* Significant at 95% level
** Significant at 99?o level
*** log Y = a0 + a1 log X
-------
Reynolds number appears to have limited value as a descriptor for mixing
conditions necessary to achieve a required degree of bacterial inactivation.
The high correlation coefficient of 0.98 and t statistics significant at
the 99 percent level were achieved when Prandtl eddy frequency was regressed
as a function of mean velocity gradient as shown in Table 4. These statistics
require further evaluation. With the assumption of a direct relationship be-
tween these two turbulence descriptors based on the statistics the following
expression can be derived where the subscripts G and f denote those terms
attributable to mean velocity gradient and Prandtl eddy frequency, respectively
(10)
JLJ
where y - density of water
6 = Moody friction factor
and the other terms have been previously described.
The density and viscosity of water can be closely approximated with a
constant over the range of water temperatures encountered during the study.
Therefore, the above relationship may be further simplified as
8 V,
m
1
(11)
Application of the continuity equation shows that for a given plug flow
mixing system and constant flow rate, Vm will vary inversely with a^. Under
the same condition 9 will decrease slowly with increasing Vm. This relation-
ship is, therefore, expected but significant since the use of phenomenological
relationships of mean velocity gradient and Prandtl eddy frequency allow a
close correlation to be developed between disinfection efficiency, material
transport parameters, and energy input to the system.
Reynolds number as a function of mean velocity gradient and Prandtl eddy
frequency, respectively, is shown in Table 4. The bivariate, linear regres-
sion analysis of the log-log transformed data yielded non-significant corre-
lation coefficients of 0.53 and 0.39 for the regression of Reynolds number on
mean velocity gradient and Prandtl eddy frequency, respectively. The only
practical use to which the Reynolds number may be applied for the evaluation
of disinfection data, as discussed herein, is the determination of the fric-
tion factor necessary for the derivation of the mean velocity gradient for a
given mixing system.
133
-------
TABLE 4
REGRESSION ANALYSIS*** OF MIXING DESCRIPTORS
Independent
Variable,
X
Mean Velocity
Gradient
Mean Velocity
Gradient
Prandtl Eddy
Dependent
Variable,
Y
Prandtl Eddy
Frequency
Reynolds
Number
Reynolds
Number of
Observa-
tions
11
11
11
Intercept,
0.29
3.46
3.53
Regression
Coefficient
of Y on X,
0.86**
0.36
0.35*
Correlation
Coefficient
0.98
0.53
0.39
Frequency Number
* Significant at 95?o level
** Significant at 99% level
*** log Y = ao + a-^ log X
CONCLUSIONS
Analysis of data collected in studies for improving disinfection of
sewage effluent from the Fort Meade Sewage Treatment Plant No. 2 justifies
the following conclusions:
1. Rapid and substantial bacterial inactivation may be achieved by
chlorination of wastewater under highly turbulent, plug flow conditions.
2. Rapid mixing of chlorine with wastewater may achieve a required
degree of disinfection by using less chlorine. Added benefits would be
material (chlorine) savings and possible decreased formation of chloro-
organics.
3. Mean velocity gradient and Prandtl eddy frequency are highly corre-
lated parameters for coliform inactivation, and they require further investi-
gation and development as descriptors for the rapid mixing, disinfection
process.
4. Reynolds number is not a universal descriptor for the rapid mixing,
disinfection process.
5. The Collins' model adequately predicted coliform disinfection using
conventional disinfection practices. Modification of the coefficients used
in the Collins1 model permitted use of the model to accurately predict
disinfection for an improved mixing system.
134
-------
ACKNOWLEDGEMENTS
The original work from which the data were extracted was performed under
the generous encouragement and guidance of Dr. Cornelius W. Kruse and
Dr. Kazuyoshi Kawata of the Environmental Health Department, The Johns
Hopkins University. The typing and proof reading of the paper was ably and
efficiently carried out by Corlyn Abbeduto.
LITERATURE CITED
1. Brodkey, R.S., 1960. "Eluid Motion and Mixing," in Mixing, Vol. 1, p. 7,
V.W. Uhl and J.B. Gray (Ed.), Academic Press, New York.
2. Camp, T.R., and Stein, P.C., 1943. "Velocity Gradients and Internal
Work in Fluid Motion," J. Boston Soc. Civil Engrs., 3PJ:219.
3. Collins, H.F., ejt al_. 1971. "Problems in Obtaining Adequate Sewage Dis-
infection," Jour. San. Engr. Div., Proc. Amer. Soc. Civil Engrs.,
97_:549.
4. Collins, H.F., and Selleck, R.E., 1972. "Process Kenetics of Wastewater
Chlorination," University of California, Sanitary Engineering Research
Laboratory Report No. 72-5, Berkeley, California, pp. 32-73.
5. Davies, J.T., 1972. Turbulence Phenomena, Academic Press, New York.
6. Eliassen, R., _et_ al_. , 1948. "A Statistical Approach to Sewage Chlorina-
tion," Sew. Works Jour., 20:1000.
7. Glover, G.E., 1972, discussion of "Problems in Obtaining Adequate Sewage
Disinfection," by H.F. Collins et_ _al_. , Jour. San. Engr. Div., Proc.
Amer. Soc. Civil Engrs., 98_:671.
8. Haas, C.N., 1980. "A Mechanistic Kinetic Model for Chlorine Disinfec-
tion," Environmental Science and Technology, 14:339.
9. Hess, S.G., e_t al^. , 1953. "Bactericidal Effects of Sewage Chlorination,"
Sew, and Ind.~Wastes, 25:751.
10. Hinze, J.O., 1959. Turbulence, McGraw-Hill Book Company, New York.
11. Horn, L.W., 1972. "Kinetics of Chlorine Disinfection in an Ecosystem,"
Jour. San Engr. Div, Proc. Amer. Soc. Civil Engr., 98:183.
12. Kruse, C.W. et_ al., 1973. "Improvement in Terminal Disinfection of
Sewage Effluents," Water and Sewage Works, 1_2J3:57.
13. Longley, K.E., 1978. "Turbulence Factors in Chlorine Disinfection of
Wastewater," Water Research, 12:813.
135
-------
14. McKee, J.E., e_t al., 1960. "Chemical and Colicidal Effects of Halogens
in Sewage," Jour. Water Poll. Control Fed., 32:195.
15. Sepp E., and Bao P., 1980. "Comparison of Optimized Pilot System with
Existing Full-Scale Systems," in Design Optimization of The Chlorina-
tion Process, Vol 1, U.S.E.P.A. Grant No. S803459, Municipal
Environmental Research Laboratory, Cincinnati, Ohio.
16. Standard Methods for the Examination of Water and Wastewater, 1971,
13th ed., Amer. Pub. Hlth. Assoc., New York.
17. Stenquist, R.J., and Kaufman, W.J., 1972. "Initial Mixing in Coagula-
tion Processes," US Environmental Protection Agency Report EPA-72-053,
Univ. of California, Berkeley, CA.
18. White, G.C., 1972. Handbook of Chlorination, Van Nostrand Reinhold Co.,
New York.
19. White, G.C., 1974. "Disinfection Practices in the San Francisco Bay
Area," Jour. Water Poll. Control Fed., 46:89.
136
-------
2. UPGRADING EXISTING CHLORINE CONTACT CHAMBERS
Frederick L. Hart
Associate Professor of Civil Engineering
Worcester Polytechnic Institute
Worcester, Massachusetts 01609
ABSTRACT
Most wastewater treatment facilities in the United States use chlorine
for disinfection. It has been estimated that over 200,000 tons of chlorine
are discharged through municipal wastewater treatment plant effluents each
year.
Recent studies strongly suggest that chlorine disinfection represents a
potential environmental health threat because of unwanted chlorinated organic
synthesis. Clearly, an alternate means of wastewater disinfection should be
made available. It must be realized, however, that significant time will be
needed to adequately develop, design and then install alternate disinfection
systems in municipal wastewater facilities throughout the United States.
During that time period, methods of improving existing chlorine disinfection
system efficiencies should be considered. An upgraded system will require
less chlorine and will therefore lessen the total burden of chlorine pollu-
tion.
An inexpensive method of increasing the contact period in a serpentine
flow chlorine contact chamber was developed through hydraulic model studies.
This modification scheme (a series of perforated baffles), when installed
into existing full scale units with a length of flow to width ratio of 8/1,
was found to parallel the performance of a unit with a length of flow to
width ratio of 25/1. Disinfection efficiency analysis of the modified and
unmodified unit operating at equal conditions at a facility in Maynard,
Massachusetts demonstrated that approximately eight percent less chlorine is
needed for the modified chamber.
Because savings from this chlorine dose reduction compares favorably to
the material cost for these perforated baffles, it may be concluded that the
modification scheme is cost effective. In addition to obvious financial ben-
efits, this proposed modification scheme represents the potential for signi-
ficant pollution reduction and should be considered a practicable interim
solution to dangers resulting from present wastewater disinfection practices.
137
-------
INTRODUCTION
Wastewater disinfection with chlorine is a well established practice
that traditionally enjoys favor with design engineers because it is proven in
terms of hardware technology and operations manageability. Existing waste-
water treatment plants designed and installed over the past couple of decades
are typically equipped with a chlorine feed system and a chlorine contact
chamber for final effluent disinfection. In effect, most municipalities are
committed to the chlorine disinfection process. Unfortunately, many studies
(11) indicate that residual chlorine and chlorinated organic compounds re-
leased in wastewater effluents represent a potential threat to water quality.
It seems inevitable, therefore, that alternative disinfection methods such as
ozone or ultraviolet light will replace chorine disinfection systems at
wastewater treatment facilities. This replacement process will not only re-
quire financial commitments for research and development projects, engineer-
ing design activities and installation, but will also require time. Although
progress in these areas is being made, the total time required to accomplish
widespread replacement of alternate disinfection processes is significant.
At present, more than 200,000 tons of chlorine are discharged each year
through municipal wastewater effluents in the United States (3). Numerous
studies have noted that efficiency of a chlorine disinfection system largely
depends on the chlorine contact chamber's hydraulic character (2,7,9,10). A
chlorine contact chamber (referred to as CCC in this paper) with short cir-
cuiting currents will not provide the necessary time for disinfection reac-
tions to approach completion. Consequently, a higher chlorine dose is used
to obtain the necessary degree of disinfection (7). If a method of improving
the efficiency of existing CCC units is available, hazard from wastewater
chlorination could be alleviated until more permanent solutions are imple-
mented. Because improvements to existing CCC units are an interim solution,
they must be relatively cheap, easy to install and versatile.
Model studies conducted by the author (4) developed a method for im-
proving the hydraulic character of the commonly used cross baffled serpentine
flow CCC unit by installing a combination of perforated baffles. These modi-
fications were specifically designed to meet the above mentioned requirements
of low cost, simplicity and versatility. This paper reports on observations
made at two wastewater treatment facilities in Massachusetts after installing
these baffles.
MATERIALS AND METHODS
Figures 1 and 2 illustrate the baffle schemes installed at the Marlboro,
MA Easterly Wastewater Treatment Plant and the Maynard, MA Wastewater Treat-
ment Plant respectively. The modification scheme installed at Marlboro, MA
is exactly similar to the model modification scheme while the modification
scheme installed at Maynard, MA varies from the original model scheme because
of differences in tank geometry.
138
-------
V
B
\_
A
V
-A
Figure 1. Modified CCC Unit at Marlboro, MA
V
•B
F
V
£
Figure 2. Modified CCC Unit at Maynard, MA
139
-------
Further description of the baffle configuration and design may be found in
other papers (4,5,6). Table 1 lists specifications for both field units.
Because both treatment facilities were equipped with dual CCC units, si-
multaneous evaluation of modified and unmodified systems were possible as baf-
fles were only placed in one side. This method of field testing helped elimi-
nate the inclusion of many uncontrolled variables (particularly wastewater
characteristics) because both modified and unmodified systems were subjected
to those variations during the experiment period.
Table 1. CCC Unit and Baffle Specifications
Specification Marlboro, MA Maynard, MA
CCC
Length 15.50 m 6.25 m
Width 7.30 m 3.05 m
Depth 3.00 m 2.44 m
L/W (flow) 8/1 8/1
Baffles B,D,E,I
Length 3.65 m 1.40 m
Depth 3.05 m 2.44 m
Hole Diameter 15.24 cm 6.35 cm
% Open Space 17% 17%
(except B=3.4%) (except B=3.4%)
Baffles C,F,H
Length 3.76 m 1.70 m
Depth 3.05 m 2.44 m
Hole Diameter 15.24 cm 6.35 cm
% Open Space 17% 17%
Baffle G
Length 6.10m
Depth 3.05 m
Hole Diameter 15.24 m
% Open Space 17%
Material Cost $599.00 $250.00
Tracer experiments using Rhodamine l-JT fluorescent dye were conducted
when chlorine was not in use. A single pulse of dye was injected directly up-
stream before entrance into the CCC unit. Measurements of the effluent dye
were made in the field with a Turner Model 111 Fluorometer until all dye was
recovered. C-curves (C/C vs t/T), and dispersion index values (d) were
140
-------
generated from the tracer data. Equations used to calculate the dispersion
index are as follows:
2 _ Zt2c /Etc \ 2 .
at ~ iF~~ -(TT> ]
where: t = time
c = tracer concentration at time = t
Z_tc
9 £c
2 22
a = aj:/e 3
2d + 8d2 = a2 . 4
Expression 4 identifies the relationship of the C-curve variance to the dis-
persion coefficient for an open vessel.
A further development of the dispersion index expression to describe the
C-curve for an open vessel is as follows (8):
E = _ J __ exp -
9 ^ ' 49d
where: EQ = C/C
D 0
6 = t/T .
Expression 5 is derived on the assumption that flow is minimally dis-
turbed at the inlet and outlet zones. Use of this expression is explained
later in this section.
Total col i form populations were measured by the Membrane Filter test as
described in Standard Methods (1). Samples were collected in sterilized bot-
tles containing a 10 percent sodium thiosulfate solution and were immediately
placed on ice. Initial MF screening tests were made in the laboratory to pre-
dict col i form numbers in order to increase the likelihood of successfully
bracketing the required dilutions. This step was found necessary because the
coliform population numbers varied considerably. Final incubation was always
conducted within 24 hours of sample collection.
Disinfection response data generated from field units and laboratory
batch reactors were fitted to an expression introduced by Collins (2) as fol-
lows:
n
N
6
N ~ct
o
where: FJ = coliform population leaving the CCC unit or at time t in a batch
reactor
141
-------
N0 = coliform population entering the CCC unit, or at initial time in
a batch reactor
b = lag coefficient (rug x min/L)
n - velocity coefficient
t = time (detention time of a CCC unit)
c = chlorine dose
Coliform population values entering and leaving the CCC unit, chlorine dose
levels and hydraulic retention times were applied to this expression to quan-
tify the disinfection capability of both modified and unmodified systems.
After solving these expressions for both systems, a comparison of their rela-
tive efficiency difference was calculated by setting both expressions to iden-
tical conditions of N/N .
A simulation of disinfection response under ideal plug flow conditions
was obtained through batch reactor experiments as illustrated in Figure 3.
These data were fitted to expression 6 for subsequent prediction of disinfec-
tion efficiency in a CCC unit defined through tracer experiment data. Such a
method of CCC unit examination was presented by Trussell and Chao (10) with
the following expression:
N/No -J (N/Vbatch Eed6 7
0
where: (N/N ) batch = expression 6 fitted to batch reactor data
£„ = expression 5
A value of N/N0 for a given chlorine dose (c) can be calculated with
equation 7 if the dispersion index (d), batch reactor data (b,n) and C-curve
point (E0,d) are known. As with expression 6, expression 7 is used in this
study to quantify the relative difference in disinfection capability of a mod-
ified and unmodified system.
PRESENTATION AND DISCUSSION
Because wastewater flow received at the Marlboro, MA facility during the
field test period (nine months) was unusually low, simultaneous operation of
the modified and unmodified CCC units could not be accomplished without allow-
ing hydraulic retention times beyond the typical range. In addition, coliform
population concentrations entering the CCC process were low. Consequently,
disinfection efficiency data obtained at this facility were not considered re-
liable and will therefore not be presented here. Information obtained at this
facility regarding the cost, durability and handling of the modification baf-
fles, however, was valuable to this study and will be discussed later.
Tracer Response
C-curve plots for the modified and unmodified CCC units are presented in
142
-------
SAMPLING:
T'- 0
DOSE:
SAMPLING I
7=2 MIN.
7 = 5 MIN.
T= 10 MIN.
7 = 20 MIN.
7=30 MIN.
NO
2MG/L
N I
NI
Nl
Nl
Nl
No
4MG/L
N 2
N 2
N2
N2
N2
NO
10 MG/L
N 3
N 3
N3
N3
N3
Figure 3. Batch Reactor Experiment
, an improvement in the chamber's hydraulic charac-
baffle modification scheme. The dispersion index
Figure 4. As can be seen.
ter was obtained from the
(d) and C-curve variance (a?) for these curves are 0.082 and 0.218 for the un-
modified unit and 0.035 and 0.076 for the modified unit. A comparison of a2
data for various 1/w configurations as^presented by Marske and Boyle (9) indi-
cates that the unmodified unit responds very closely to the expected performance
of a unit with an 8/1 1/w configuration, while the modified unit responds
closely to the performance of a unit with a ;
of hydraulic characteristics, therefore, the
ficiently.
!5/l 1/w configuration. In terms
modified unit responds more ef-
Disinfection Response
Log-log plots of N/N0 vs ct from field experiments at the unmodified and
143
-------
Modified
Unmodified
2.0
Figure 4. C-curves for Modified and Unmodified Units
144
-------
modified units are presented in Figures 5 and 6. Each data point represents
the mean of a two hour field experiment. Tests were conducted during mid-day
periods when influent rates were relatively constant. These points when ap-
plied to a regression analysis fitting expression 6 yield:
Unmodified Unit
N/NQ = (22.75/ct)
2.96
= 0.82
Modified Unit
N/NQ = (21.26/ct)
2.98
r = 0.86
Setting expressions 8 and 9 to an equal degree of disinfection,
modified = N/N0 unmodified) demonstrates that a seven percent decrease
in required chlorine dose is needed for the modified unit.
Figure 7 presents
piug flow conditions.
to expression 6:
a plot of batch reactor data used to simulate ideal
These data yield the following coefficients when fitted
N/NQ = (19.00/ct)
3.04
r = 0.83
Applying coefficients obtained for expression 5 and 10 to equation 7 yields a
nine percent chlorine reduction requirement for the modified unit. Field
tracer data and laboratory batch reactor data, therefore, indicate that the
modified CCC unit should require nine percent less chlorine to obtain the same
degree of disinfection.
Field Observations
In addition to disinfection performance evaluations, these field studies
were conducted to evaluate the cost, durability and versatility of the baf-
fles. Table 1 indicates that the material costs are relatively low and, as
will be noted in the conclusion, the costs are reasonable when compared to po-
tential savings from reduced chlorine use. Baffles at both facilities were
constructed from standard size corrugated plastic sheets attached together
with an epoxy resin and braced with 1" x 3" wood strips. The baffles were
lightweight and could easily be placed into the chamber by-two people. During
the nine month experimental periods, no damage to the baffles was noted.
Higher accumulations of solids and floating materials were noted in the
modified units. Such a drawback was expected because short circuit currents
capable of carrying these materials through the CCC unit and out the effluent
145
-------
q
d
o
CVJ
I
b -22.75
n—2.97
r2*0.82
2.0 2.4 2.8
Figure 5, Unmodified Unit Disinfection Response
146
-------
o
6
m
6
o _j
in
cvi
q
ro
in
ro
b-21.26
n «-2.98
r2- 0.86
MODIFIED
1.2 1.6 2.0
Log ct
Figure 6. Modified Unit Disinfection Response
2.4
2.8
147
-------
o
6
80
6
o
IO
IO
9
iq
10
b • 19.00
n* -3.04
2.0
2.4
2.8
Log ct
Figure 7. Batch Reactor Disinfection Response
148
-------
were eliminated. Periodic cleaning of the modified unit was possible by re-
moving the baffles. This was possible because the baffles were lightweight
and were secured to the CCC wall by being placed between two guides. Perma-
nent connection of the baffles to the CCC wall is not recommended.
Versatility of this modification scheme was partially demonstrated by
these field tests because the two facilities are different sizes (see Table 1)
and have slightly different CCC unit configurations (see Figures 1 and 2).
Radically different configurations, however, may not respond similarly. It is
anticipated that a very poor hydraulic design should benefit significantly
from this type of baffle modification scheme. It should be noted that the
configuration illustrated in Figures 1 and 2 was chosen because it is very
common, not because extremely poor hydraulic conditions were expected.
CONCLUSION
Data presented in this paper indicates that the modified CCC unit per-
forms at a more efficient level than the unmodified unit and will require
about eight percent less chlorine to meet the same degree of disinfection.
Assuming that this drop in required chlorine dose is achievable under field
conditions, a decrease in chlorine cost savings could be realized. Using a
chlorine cost figure of $0.22 per kg, and an average chlorine dose of eight
mg/L, a 3800 m3/d WWTP will save $243.00 per year. A comparison of this
figure to the cost for constructing these baffles (see Table 1) indicates that
these simple modifications are economically justifiable.
A more significant effect of improving the efficiency of existing CCC
units, however, is the resulting potential for reduced impact to the aquatic
environment. A thorough analysis of this potential, however, was beyond the
scope of these projects. Specific areas that remain to be explored include
the influence on chlorinated organic synthesis resulting from more efficient
CCC units and the degree of efficiency improvement possible for different CCC
unit configurations.
LITERATURE CITED
1. APHA, Standard Methods for the Examination of Uater and Wastewater, 14th
Edition, 1975.
2. Collins, H. and R. Selleck, "Process Kinetics of Uastewater Chlorination",
SERL Report No. 72-5, Univ. of Calif. Berkeley, (Nov. 1972).
3. Comptroller General of the United States, "Unnecessary- and Harmful Levels
of Domestic Sewage Chlorination Should be Stopped", CED-77-108, (Aug.
30, 1977).
4. Hart, F.L., "Improved Hydraulic Performance of Chlorine Contact Chambers",
J. UPCF, Vol. 51, No. 12, December 1979, pp. 2868-2875.
149
-------
5. Hart, F.L., and Z. Vogiatzis, "Performance of a Modified Chlorine Contact
Chamber", J. ASCE, Env. Div., Vol. 108, No. EE3, June, 1982.
6. Heath, G., and F.L. Hart, "Evaluation of a Full-Scale Modified Chlorine
Contact Chamber," presented at the NEWPCA, 1980 Meeting, North Falmouth,
MA,
7. Kothandaraman, V., et al., "Performance Characteristics of Chlorine
Contact Tanks", J. U'PCF, 45, 611 (1973).
8. Levenspiel, O.s and Smith, "Notes on the Diffusion-Type Model for the
Longitudinal Mixing of Fluids in Flow," Chem. Engr. Scie., 6, 227
(1957).
9. Marsky, D.M., and Boyle, J.D., "Chlorine Contact Chamber Design - A Field
Evaluation", Jour. Water & Sewage Uorks, 120, p. 70 (Jan. 1973).
10. Trussell, R.R., and Chao, J.L., "Rational Design of Chlorine Contact
Facilities", J. UPCF, 49, 659, (1977).
11. Venosa, A.D. (editor), "Progress in Wastewater Disinfection Technology",
proceedings of the National Symposium, Cincinnati, Ohio, Sept. 1978,
EPA-600/9-79-018, June 1979.
150
-------
3. PROBLEMS OF DISINFECTING NITRIFIED EFFLUENTS
George Clifford White
Consulting Engineer
556 Spruce Street
San Francisco, California 94118
Robert D. Beebe
Principal Sanitary Engineer
San Jose/Santa Clara
Water Pollution Control Plant
Los Esteros Road
San Jose, California
Virginia F. Alford, Microbiologist
San Jose/Santa Clara Plant
H. A. Sanders, Senior Chemist
San Jose/Santa Clara Plant
ABSTRACT
Several wastewater treatment plants in California have experienced some
unexpected problems when trying to achieve the NPDES requirement of 2.2/100ml
MPN coliform concentration in nitrified and filtered effluents. These prob-
lems do not exist in non-nitrified effluents or those containing 1.5 to 2.5
mg/1 of ammonia nitrogen, or more. The nitrified effluents in question con-
tain only trace amounts of ammonia nitrogen and nitrites.
The San Jose/Santa Clara Water Pollution Control Plant conducted a six
month evaluation of the disinfection process. This included chlorine demand
studies and a coliform profile of the various unit treatment processes. The
chlorination - mixing - contact chamber system was designed to achieve a
2.2/100 ml MPN total coliform concentration in the plant effluent.
When the tertiary plant (nitrified and filtered) effluent went on line
the chlorine required to achieve the 2,2/100 ml MPN coliform in the effluent
was enormous compared to a non-nitrified effluent. Laboratory studies were
made adding ammonia-N to the nitrified effluent. Considerably less chlorine
was required to achieve the NPDES requirement of 2.2/100 ml MPN total coli-
form. Laboratory results were transferred to plant operation and were con-
firmed as follows: the nitrified and filtered effluent required a minimum
dosage of 17 mg/1 chlorine, which produced a residual of 9 mg/1 at the end of
49 min. at peak dry weather flow (PDWF) (this residual contained 50-60$ free
chlorine). This compared to the same effluent fortified with 2 mg/1 ammonia-N
151
-------
requiring only 12 mg/1 chlorine dose that resulted in a 7 mg/1 combined
chlorine residual at the same contact time.
The surprising factor in this investigation is that the combined
chlorine residual was extremely more reliable in its germicidal efficiency
than the free chlorine residual.
INTRODUCTION
This paper discusses the ramifications of the disinfection process in
use at wastewater treatment plants required to turn out an effluent contain-
ing a maximum of 0.1 mg/1 unionized NH^-N and a total coliform concentra-
tion not exceeding 2.2/100 ml MPN. These requirements are the result of the
guidelines formulated by the California State Department of Health, the State
Water Quality Board and the State Fish and Game Commission. There are about
fifteen treatment plants in California that are subjected to the 2.2 coliform
requirements. However, not all of these plants are required to produce a
completely nitrified effluent. Those that do not nitrify do not experience
difficulty with the disinfection process.
The data presented here were developed over a twelve month period at the
San Jose/Santa Clara Water Pollution Control Plant located on the southerly
edge of San Francisco Bay. This treatment plant was first constructed as a
primary plant in the early 1950's. About ten or so years later the plant was
expanded into a secondary plant. The secondary effluent disinfection re-
quirements were set at 240 MPN per 100 ml total coliforms. Nitrogen removal
was not required. Disinfection to meet this requirement was achieved with
about 12 mg/1 dosage and a residual of 5-6 mg/1 at the end of the contact
chamber. The detention time at PDWF was approximately thirty minutes.
A clue to solving the problems of the future tertiary effluent occurred
during the canning season when the secondary effluent lost its ammonia nitro-
gen content. During this period the disinfection process fell into disarray.
There was not enough chlorinator capacity to achieve the 240/100 ml coliform
NPDES requirement. Supernatant liquor from the digesters was added to the
raw sewage at the headworks in sufficient quantity to produce a predominantly
monochloramine residual. As soon as this was done the disinfection process
returned to normal.
THE TERTIARY PLANT
In February 1979, the San Jose tertiary plant was put into operation.
These additions to the secondary plant which composed the tertiary treatment
process consisted of a nitrification unit (suspended growth system) and dual
152
-------
media filters. A new chlorine contact chamber was also part of this con-
struction project. The chlorine control system was a combination of flow
pacing and chlorine residual control commonly described as compound loop
control. Adjacent to the chlorine diffusers were turbine mixers followed by
a specially designed serpentine chlorine contact chamber. The contact
chamber had a 49 minute contact time at peak flow established by the first
appearance of a dye entering at the chlorine diffusers (tj_). It was fully
expected that the tertiary effluent would be of such superior quality to
the secondary effluent that the chlorine required to achieve 2.2/100 ml MPN
coliform concentration in the tertiary effluent would be less than that
required to achieve 240/100 ml in the secondary effluent. This was based
upon the assumption of the presence of a free chlorine residual in the
nitrified effluent and a much lower coliform concentration (Yo) in the
filtered effluent.
It developed that the tertiary effluent exhibited an abnormal free
chlorine demand, for which there was no ready explanation. At first it was
thought that this was due to the presence of nitrites (one mg/1 nitrite-N
will consume 5 mg/1 HOC1). Combined chlorine (chloramines) will not oxidize
nitrites to nitrates within the time frame of wastewater treatment systems.
The quality of the tertiary effluent is shown in Table 1. Examination of
these data eliminates nitrites as the cause of high free chlorine demand.
Table 1. Tertiary Effluent Quality*
Parameter Concentration, mg/1
Hardness 245
TOC 11-14
TDS 800-900
Org. N 1.3-2.3
N02-N 0.02-0.03
NH3-N trace
spite of this high quality filtered effluent the disinfection system
was not able to meet the 2.2 coliform requirement. Table 2 illustrates
this dilemma.
153
-------
Table 2. Final Effluent Total Coliform Concentration
After 49 Minutes Contact Time
Total Cl2 Residual Total Coliform
MPN/100 ml
6.5
6.5
6.5
7.4
7.4
7.4
9.1
9.1
9.1
5.1
5.1
5.1
79
23
23
4
2
2
2
2
2
2
13
7
The residuals shown in Table 2 were measured by the forward amperometric
procedure. These residuals contain about 60 percent HOC1 and the rest
titrates as "dichloramine." The latter species is probably composed of a
variety of non-gerraicidal organochloramines as a result of the organic-N
present.
INVESTIGATION OF TERTIARY EFFLUENT
Owing to the poor performance of free chlorine residuals it was decided
to investigate the following characteristics of the disinfection process.
a) Verify the t^ contact time at PDWF (t.^ = first appearance of dye
at the exit of the contact chamber"! .
154
-------
b) Determine the chlorine demand for various chlorine dosages and con-
tact times. Compare with chloramine species.
c) Establish a coliform profile (without chlorination) beginning with
the secondary effluent and continuing to the final effluent.
d) Determine chlorine dosage required to achieve 2.2/100 ml MPN total
coliforms at contact time t^.
e) Compare germicidal efficiency of free chlorine versus combined
chlorine residuals.
f) Determine the benefit, if any, of applying the postchlorination dose
at two separate points in the treatment train.
RESULTS
Contact Time
The contact chamber dye test revealed a tj_ of 49 min. at PDWF. The con-
tact times used for all samples analyzed in the laboratory were 5, 30, 49
min. and 24 hours.
Chlorine Demand Studies
The secondary effluent residuals were examined for total chlorine resid-
ual. This was done by the back titration method using an amperometric
titrator.
The nitrified effluent was examined for free chlorine, mono, and di-
chloramines. This was done with a separate titrator using the forward ti-
tration procedure.
Figure 1 illustrates the chlorine demand of the secondary effluent which
contains only combined chlorine residual. Figure 2 illustrates the same for
the nitrified filtered effluent which contains about 60 percent free chlorine.
Figure 3 illustrates the same for a nitrified filtered effluent that has been
fortified with enough NHo-N to provide a 6:1 chlorine to nitrogen wt. ratio.
Coliform Profile - A summary of the coliform levels in various stages of the
San Jose plant is presented in Table 3-
155
-------
Table 3- Coliform Profile of San Jose Plant
Location
Total Coliform MPN/100 ml
Max Min Median
Secondary Effluent
Nitrified Effluent
Filtered Effluent
(no prechlor)
Filtered Effluent
(with 8 mg/1 prechlor ,
C12 res. 1.2 mg/1,
contact time 17 min)
9.2 x 106 49,000 1.7 x 10
1.6 x 10 23,000 110,000
160,000 200 23,000
1,300 <20 80
Coliform Kill Stud^
This study was performed concurrently with the chlorine demand studies.
Each sample was divided into three replicates. Each of these replicates was
then transferred to five tubes for four different dilutions: 10 ml., 1 ml
0.1 ml and 0,01 ml. This amounts to 20 tubes for each of three replicate
samples. The secondary effluent and the filtered effluent were all subjected
to this same examination. The discussion of these results follows below.
a) Secondary Effluent. The objective was to find if possible the
chlorine dosage to provide a 2.2/100 ml MPN effluent using 49
min. as the contact time. This study provided a most important
clue. Both 12 and 15 mg/1 chlorine dosages at 49 min. contact
time were investigated. Some of the 12 mg/1 dosages resulted
in 2.2/100 ml MPN coliform and some resulted in counts as high
as 33/100 ml MPN. The 15 mg/1 dose was more consistent owing
to a higher residual at 49 min. contact time. At this dosage
the residual that achieved 2.2/100 ml MPN coliform were on
the order of 8 mg/1. This fits the Collins model.
b) Nitrified Filtered Effluent. The quality of this effluent is
considerably superior to the same effluent without filtration.
The Y0 coliforms are much lower and the organic nitrogen is sig-
nificantly less. The latter means that the combined chlorine
residual will contain less non-germicidal organochloramines.
In spite of the superior quality of this filtered effluent
the germicidal efficiency of the free chlorine residual was
156
-------
disappointing to say the least. For example: the median YQ
from April through May 1980 was 23,000/100 ml MPN coliforms.
This calculates to a ct = 91 in the Collins model. So,for a
49 minute contact time, disinfection should be possible with
a 2 mg/1 total chlorine residual. Laboratory and plant res-
ults have shown that the total residual for the nitrified fil-
tered effluent must be on the order of 9-10 mg/1 for consistent
results. The San Jose plant does not have enough sulfonator
capacity to dechlorinate this much residual.
Comparison of Germicidal Efficiency of Free Versus Chloramine Residuals
Owing to the above dilemma it was decided to experiment with artificial
chloramine residuals and compare their efficiency against the free chlorine
residuals. This was done by adding ammonia nitrogen in various Cl to N ratios
to the nitrified effluent. Chlorine dosages used were 10, 12, and 15 mg/1.
Cl to N ratios investigated were 6 to 1, 8 to 1, and 10 to 1. All of the
dosages using chlorine to ammonia N at 6:1 produced an effluent coliform
concentration of 2.2/100 ml or less without exception. From these tests it
was patently clear that a chloramine induced residual can outperform a free
chlorine residual by a wide margin at the San Jose plant. The breakpoint
curve for this ratio is shown on Figure 4.
Effect of Mixing
The San Jose investigation has put the subject of mixing as it effects
disinfection efficiency into an entirely different perspective. It appears
that the most important reason for superior mixing in wastewater disinfection
is to convert as soon as possible the free chlorine in the chlorine solution
to chloramines. This minimizes formation of organic-N compounds which have
low disinfection efficiency.(3) Laboratory experiments demonstrated the dif-
ference between good mixing and poor mixing.
The results shown in Table 4 are average residuals of several experim-
ents with wastewater containing 2 mg/1 artificially added ammonia-N and subse-
quently dosed with 12 mg/1 Cl and a contact time of 60 minutes.
Table 4. Average Chloramine Residual in Wastewater
Effluent as a Function of Degree of Mixing
Monochloramine Dichloramine
Good Mixing
Poor Mixing
7.25
3.25
1.25
3.45
157
-------
In plant practice, good mixing is considered achieved when the velocity
gradient, G, in the mixing chamber approaches 1000 (3)-
DISCUSSION
Collins Model
The Collins mathematical model (3) is used to establish chlorine dosages
at given contact times for combined chlorine residuals (non-nitrified efflu-
ents). It is a good basis for comparison of free versus combined residuals.
The Collins equation (1) is as follows: y/yo = (1 +0.23 ct)~3
where: y = 2.2/100 ml MPN (NPDES limitation),
y0 the median coliform concentration before chlorination
c - chlorine residual (mg/1) at the end of time t (minutes)
t = tj_ first appearance of dye at the end of contact chamber.
This investigation was not an exercise to prove or disprove the Collins
model, which has served so well in evaluating the efficiency of disinfection
systems of non-nitrified effluents. It has been used here to compare the ef-
ficiency of combined chlorine residuals versus combined residuals that are
measured as predominantly free chlorine. The conclusion based upon the San
Jose study is that the Collins model is not applicable for combined residuals
that are measured as predominantly free chlorine, e.g., 50 percent or more.
The above example for the nitrified filtered effluent indicates that the Col-
lins model predicts a total chlorine residual of 2 mg/1 at 49 min. coritact
to achieve a 2.2/100 ml coliform MPN. However, in reality it was found that
the required total residual to achieve the 2.2 figure was closer to 9 mg/1.
See Table 2. These residuals contained about 60 percent free chlorine.
Obviously the Collins model does not fit nitrified effluents. This is
indeed a surprising development. The reason for this lies somewhere in the
chemistry of the higher reactivity of free chlorine, hence its higher consump-
tion. However5 analyzing the 9 mg/1 residual referred to above, this con-
tained about 5 mg/1 free chlorine. The remainder titrated as dichloramine.
Chloramine Residuals
Chloramine residuals occurring in wastewater always contain a mixture of
monochloramine and dichloramine. The dichloramine is most probably due to the
presence of significant concentrations of organic nitrogen (1-3 mg/1). It is
presumed the chlorine residual species that titrates as the dichloramine frac-
tion in a wastewater is probably a variety of organochloramines having little
or no germicidal efficiency (4). Therefore,the objective is to get a chloram-
ine residual with the highest percentage possible of the monochloramine frac-
tion. Fast and thorough mixing of the chlorine with the wastewater is the key
factor to achieve this result.
158
-------
Compared to the germicidal efficiency of free chlorine, chloramines have
been thought of as inferior. Some researchers in the 1970's, however, have dis-
covered that if given enough time (40-60 minutes), chloramines are nearly as
effective as free chlorine (4). Selleck et al (2).have shown that the most
germicidal combined chlorine residual (chloramine) appears to occur when the
chlorine to ammonia-N ratio is on the "breaking " side of the B-P curve.
This is between points A and B on Figure 4. At point A the ratio at the hump
of the curve is nominally 5 to 1 Cl to N by weight.
The coliform kill study revealed the maximum kill of coliform organisms
with the least chlorine dosage occurred at a Cl to N ratio of 6:1 for the av-
ailable contact time of 49 minutes at PDWF.
Comparison With Other Plants
An integral part of this investigation was to visit other plants with
similar effluent requirements for coliforms and ammonia nitrogen. Including
San Jose, a total of ten plants were visited. All of them were in California,
and all but three were in the San Francisco Bay area. The plant processes and
operation varied considerably, depending upon whether water reclamation was
involved and whether or not the receiving waters or the end use of the efflu-
ent could tolerate ammonia-N in the effluent. (The NPDES requirement for am-
monia-N is for the receiving waters and not the effluent.)
At one nearby plant, a 2.8x10^ m3/d capacity investigation was begun in
1980 to find out how energy might be saved if nitrification were not complete
(1). It was found that NH^-N concentration above 2'mg/l resulted in a 25 per-
cent reduction in the chlorine demand. The investigation did not support the
dogma that free chlorine is a better disinfectant than combined chlorine. More
over, it was revealed that the final effluent dosage could be reduced without
adversely affecting the bacteriological quality of the effluent. When the ef-
fluent contained 2-3 mg/1 ammonia-N, the 2.2/100 ml MPN coliform concentration
could be achieved on a consistent basis. When complete nitrification was prac-
ticed it was not uncommon to require final effluent chlorine doses from 14 to
20 mg/1. The contact time was one hour
Four plants with one hour contact times required between 18 and 25 mg/1
chlorine to achieve the 2.2. These dosages resulted in total chlorine resid-
uals of 9 to 14 mg/1. The free chlorine residual fraction varied from 45 to
85 percent of the total. One small plant required a 50 mg/1 dose which re-
sulted in a 35 mg/1 residual.
Another plant was experiencing similar high chlorine dosage requirements
but was further plagued by intermittent ammonia spikes in the effluent. This
resulted in the conversion of the free chlorine to combined chlorine. An at-
tempt has been made to try and control the nitrification process to leave a
2-4 mg/1 ammonia-N residual. This has proved to be difficult. Owing to a
limitation of chlorinator capacity a savings in chlorine consumption has not
been realized. However, where the ammonia peaked as the plant flow increased
the coliform kill increased so that compliance was achieved (2.2 MPN).
159
-------
In every investigation there is always an exception. One plant, with a
flow range of 34,000-53,000 m-'/d, using suspended growth reactors for nitrifi-
cation and dual media filters, produced a completely nitrified effluent and
achieved 2.2/100 ml MPN coliforms (7 day median) in the effluent with a dosage
of 7-8 mg/1 that resulted in a 3 mg/1 total chlorine residual after about 60
minutes contact time.
Another plant was found to be unique because the effluent was not fil-
tered and the chlorine dosage control was based upon a free residual (in the
presence of combined residual) and the plant consistently turned out a com-
pletely nitrified effluent that met a 2.2/100 ml MPN coliform concentration.
The chlorine dosage was 8 mg/1, contact time at peak flow was 49 minutes (as
determined by tracer studies) and the total chlorine residual at the end of
the contact chamber was about 3-4 mg/1. This was a well oxidized effluent
(activated sludge) and the coliform concentration before chlorination was on
the order of 140,000/100 ml MPN. The Collins model predicted a residual of
3.45 total chlorine residual. This plant was in a suburb so that effluent was
primarily domestic wastewater. All of the industrial discharges were pre-
treated before entering the collection system.
CONCLUSIONS
a) A nitrified effluent, in spite of filtration, demonstrates a
much higher chlorine demand than a non-nitrified, non-filtered
effluent.
b) The higher demand described above is probably due to the higher
reactivity of free chlorine compared to combined chlorines.
c) Both the laboratory and plant scale investigations determined
that a 6:1 Cl to N ratio with a 12 mg/1 chlorine dose proved
to be the most germicidal ratio.
d) Plant scale operation proved that the addition of a 12 mg/1
dose of chlorine added to the effluent containing 2 mg/1 of
ammonia-N can produce an effluent which will consistently
meet the NPDES requirement of 2.2/100 ml MPN total coliforms.
e) The most germicidal combined chlorine residual proved to be
one that is composed of about 75-80 percent monochloramine.
The remainder titrates as "dichloramine" which is considered
to be organochloramines of low germicidal efficiency. These
chloramines are probably a result of the organic nitrogen
present in wastewater effluents.
f) The laboratory experiments proved that good mixing was re-
quired to achieve residuals containing 75-80 percent mono-
chloramine.
g) Adequate mixing should occur when the velocity gradient G
approaches 1000.
160
-------
h) Where mixing is poor the monochloramine species drops to
about 50 percent of the total residual. This results in
lower germicidal efficiency together with a higher con-
sumption of chlorine.
i) Plant scale operation also proved that the 2 mg/1 addition
of ammonia-N to the effluent did not jeopardize the NPDES
requirement of 0.025 mg/1 un-ionized ammonia nitrogen
(NH/OH) in the receiving waters (lower San Francisco Bay).
j) The chlorine dosage and residual requirement to achieve an MPN
coliform concentration of 2.2/100 ml in the nitrified effluent
containing 2 mg/1 ammonla-N was demonstrated to be 5 mg/1 and
2 mg/1 respectively less than for the nitrified effluent, without
any ammonia-N. See Figure 5.
REFERENCES
(1) Dhaliwal, B. and Baker R.A. "Controlling Nitrification to Reduce Energy
and Treatment Costs" presented at Ann. Conf. Calif. Water Poll. Control
Assoc., Long Beach, Calif., June, 1981.
(2) Selleck, R.E., Saunier, B.M., and Collins, H.F., "Kinetics of Bacterial
Deactivation with Chlorine" J. Env. Eng. Div. ASCE, p. 1197 (Dec., 1978)
(3) White, G.C., "Handbook of Chlorination" Van Nostrand Reinhold, New York
(1972)
(4) White, G.C., "Disinfection of Wastewater and Water for Reuse" Van Nostrand
Reinhold, New York (1978)
161
-------
SAN JOSE CALIF- WATER POLLUTION CONTROL PLANT
j§
Q
OJ
o
LU
•^
QL
O
_J
X
O
15
10
8
7
6
5
4
2
SECONDARY EFFLUENT
Chlorine/Dosage (mg/l)
-15
N05-N =
Org.N
TOC
4-9 mg/l
4-6 mg/l
1-5-2-5 mg/l
2-3-5 mg/l
15-25 mg/l
15 30 49
CONTACT TIME (min.)
1440
Figure 1. Chlorine Demand Secondary Effluent
-------
SAN JOSE CALIF. WATER POLLUTION CONTROL PLANT
O>
Q
2
Ul
Q
UJ
Z
tr
o
-j
in
o
15
10
9
8
7
6
5
NITRIFIED
FILTERED EFFLUENT
CI2 dosage
(mg/l)
0-lm.g/l
N02N=0-02mg/l
N03N= 12 mg/l
Org. N = 1-6 2'3mg/l
TOC = 9 14 mg/l
o =Ti of CI2Contact
Chamber at PDWF
5 30 49
CONTACT TIME (min)
8-
7—
6—
5—
4-
3-
1440
Figure 2. Chlorine Demand Nitrified Filtered Effluent
-------
SAN JOSE CALIF WATER POLLUTION CONTROL PLANT
NITRIFIED FILTERED EFFLUENT WITH NH,-N ADDED
Chlorine Dosage
Chlorine Dosage
T: at PDWF
15 30 49
CONTACT TIME (mm)
1440
Figure 3. Chlorine Demand Nitrified Filtered Effluent with Ammonia-N Added
-------
SAN JOSE CALIF. WATER POLLUTION CON TROL PLANT
12
10
o>
E
8
6
CO
UJ
en
LU
2
QC
O
_J
X
o
0
I I I I I I I
NITRIFIED FILTERED EFFLUENT (NHj-N added)
— NHrN
N02-N
N03-N
Org.N
= < 0-1 mg/l
= < 0.12 mg/l
= 222 mg/l
1.9 mg/l
3-N added= l-67mg/l
Contact time = 60min.
Total Chlorine
Monochlorctmme
6 8 10
CHLORINE DOSAGE
2 14
(mg/l)
o
16
Figure 4. Breakpoint Curves for Chlorine to Nitrogen Ratio 6:1 by Wt.
-------
SAN JOSE CALIF. WATER POLLUTION CONTROL PLANT
SUMMARY
Chlorine dosage and residual requirements to achieve an
MPN colitorm cone, of 2.2/IOOml
ON
Nitrified and Filtered Effluent
Same as above except that 2mg/i NH^N
has been added
Dosage
17
12
Residual
9*
Note :
Residuals are those measuredat the end of the contact
chamber. This amountsto 49m contact time at PDWR
* These residuals are 50 to 60% free. About 90to95%
of the remainder titrates as dichloramine,therestas
monochloramine.
Figure 5. Chlorine Dosage and Residual Requirements to Achieve MPN Coliforms of 2.2/100 ml
-------
4. OPERATING EXPERIENCE DISINFECTING SECONDARY EFFLUENT WITH
PILOT SCALE ULTRAVIOLET UNITS
Paul H. Nehm, Director
Wastewater Treatment Operations
Madison Metropolitan Sewerage District
Madison, Wisconsin
ABSTRACT
The effectiveness of disinfection of secondary effluent from an
activated sludge wastewater treatment plant was tested using pilot scale
units from four different manufacturers. Each unit was onerated to maintain
an effluent fecal coliform concentration of less than 200 per 100 ml.
Each unit was capable of maintaining the coliform standard when it was
clean. However, keeping the units clean was the most serious problem
observed with each unit. Flushing with citric acid oroved to be an adequate
method of cleaning the units. Units employing an ultraviolet sensor
provided an early warning of impending failure. Results of this study
will be used for the possible design of a full scale system.
INTRODUCTION
The Madison Metropolitan Sewerage District (MMSD) onerates the 50MRD
Nine Springs Wastewater Treatment Plant in Madison, Wisconsin. This olant
provides primary and secondary treatment with anaerobic digestion and land
apolication of residual solids. Plans are currently being develoned to
upgrade the treatment at the olant to Advanced Secondary Standards. A
portion of this upgrading deals with the replacement of the obsolete
chlorination equipment.
In the original Environmental Impact Statement the Environmental
Protection Agency (EPA) commented adversely on the continued use of chlorine
for disinfection. Therefore, the District's consultants considered the
use of ozone and ultraviolet light in the update to the Facilities Plan.
Preliminary calculations 'indicated that ultraviolet linht (UV) would be
more cost effective. Since at this time there was very little documented
experience with the use of ultraviolet liaht to disinfect secondary effluent,
the District decided to pilot test at least one manufacturer's unit. This
first unit was put on line in December 1979. By the end of the test period,
Seotember 1981, units from four manufacturers had been tested.
METHODS
The purpose of the pilot tests was to determine what operational
and maintenance problems could be expected and how to design to overcome
these problems. Features offered on the various units were to be compared
167
-------
and evaluated for inclusion in a full scale facility. No attempt was made
to compare theories of operation or to measure the ultraviolet dose applied
or the power used. Values measured on these test units would have been
invalid for scale-up since each manufacturer was continuing to modify and
refine his design.
Secondary effluent was pumped into each unit. Because of lack of
sufficient ancillary eauipment all four units could not be evaluated
simultaneously. Each unit was initially operated at the flow rate recommended
by its manufacturer. Based on the disinfection results achieved, the flow
rate was varied accordingly. The effluent from the treatment nlant will
be required to attain a monthly geometric mean fecal coliform count of
less than 200 per 100 ml. As a measure of reliability it was the goal of
the tests to operate the pilot units so the value of 200 fecal coliforms
per 100 ml was never exceeded. The unit was defined as being in the
failure mode when this value was exceeded.
Samples of the influent and effluent of the units were analyzed daily
for fecal coliforms by the membrane filtration method. The total and
volatile suspended solids concentration of the influent were also analyzed
as was the absorbance of the influent at 254 nm wavelength.
RESULTS
Unit A
The first unit evaluated, Unit A, was a standard production model
rated at 100 gpm. Twenty-four UV lamps in an array four high and six
wide comorised the disinfection chamber. Each lamp was enclosed in a
quartz tube with an outside spacing of three-fourth inch between the tubes.
Flow through the unit was perpendicular to the longitudinal axis of the
lamns. The inlet to the unit was baffled as was the discharge with the
free water surface being controlled by the effluent baffle. Teflon discs
encircling but not touching the tubes were used as mechanical wipers.
These discs were attached to a rack which slid along the longitudinal
axis of the tubes at adjustable time frequencies. Light emitting diodes
(LED's) on the outside of the unit indicated lamps which were operating
properlv. Also included in this unit was a sensor which measured the amount
of light transmitted to it at 254nm. This sensor was housed in a quartz
tube similar to those housing the UV lamps. Three circumstances could be
responsible for a decreased reading from the sensor: 1) an increase in
the UV absorbance of the water, 2) reduced output from the UV lamps,
or 3) coating of the quartz tubes. A receiver on the side of the unit
indicated the sensor reading on a scale labeled "relative transmittance".
The results obtained from this unit are shown in Figures 1 and 2.
During the first four weeks of operation the disinfected effluent exceeded
200 fecal coliforms per 100 ml on only one occasion. After three weeks of
operation the relative transmittance reading began to decrease. By the
fourth week the transmittance reading was less than 50 percent of full scale,
168
-------
and the effluent fecal coliform reading was consistently above 200 per 100 ml.
As seen in Figure 1, the decline in the relative transmittance reading
seems to correspond to an increase in the UV absorbance of the water.
However, this was not the only cause of the higher effluent fecal
coliform counts. Upon draining the unit it was discovered that the tubes
were coated with a white substance. The tubes were removed from the unit
and washed with hydrochloric acid. Further investigation of the coating
indicated that it contained calcium, magnesium, and iron. Compared to the
concentrations in the water, the iron seemed to be depositing in a higher
ratio. Although it is known that iron readily absorbs UV light, no
explanation could be found for its deposition on the tubes.
After cleaning the tubes they were olaced back in the unit, and
adeouate disinfection was again attained. However, the tubes continued to
scale. The results of this situation are shown in Figure 2. When the
unit was returned to service after cleaning on February 23, March 7, and
March 15, the reading on the relative transmittance meter rose to
100 % and good disinfection results were obtained. A method to adequately
prevent the inhibitory coating of the tubes was not found.
Unit B
Two units were tested from manufacturer B. The first was a 10 gpm upflow
unit containing four lamps which were enclosed in quartz tubes. Flow
through this unit was parallel with the longitudinal axis of the tubes.
Cleaning was provided by an ultrasonic system. An ultrasonic transducer
was mounted in the bottom of the unit with the ultrasonic energy being
generated in parallel with the longitudinal axis of the tubes. Three
ports were spaced along the length of the unit to accent a removable
sensor. This sensor was similar in function to the one on Unit A.
On most occasions the resistance in ohms was read across the sensor
at each port. The sensor ports were spaced at varying distances from the
ultrasonic transducer to determine the effective range of the ultrasonics.
Results seemed to indicate that the ultrasonic cleaner was able to keep
the quartz tubes clean, but did not keep the sides of the unit or the
sensor ports clean. Figure 3 shows typical results for this unit.
The second unit to be tested by manufacturer B had six lamps and was
a fully enclosed unit. As in the first unit the flow pattern was
parallel to the longitudinal axis of the lamps. However, the flow through
the unit was horizontal. An ultrasonic transducer was placed along the
bottom of the unit so that the ultrasonic energy moved perpendicular
to both the quartz tubes and the flow pattern. Only one sensor oort
was built into this unit. Samoles were collected when the unit operated
at flow rates of 20 to 40 gpm.
Both units performed well when clean. However, the quartz tubes
became coated with a scale formation just as with Unit A. Whenever
the scaling had increased to the ooint that the target fecal coliform
level was not being achieved, the unit was cleaned with a citric acid
169
-------
Solution. Run times between chemical cleanings ranged from two to eight
weeks. Figure 4 shows the results of the longest run of eight weeks.
Unit C
A totally different design was presented by the manufacturer of
Unit C. Six teflon tubes connected in series conveyed the secondary
effluent through an array of UV lamps. Each 1-1/2 inch diameter teflon
tube was surrounded by four UV lamps. Reflectors were attached to the
lamps to direct the light to the teflon tubes. This unit did not have an
ultraviolet sensor, although it did have an amperage meter to show total
current draw by the lamps. Since the output of the UV lamps is dependent
on the lamn temperature, a thermometer was installed to measure the
temperature near one of the lamps. It was found that during the summer
one of the housing panels had to be removed from the unit to reduce
the lamp temperature to a satisfactory value.
The stated advantage of this unit was that the scale formation that
plagued the quartz tube units would not affect the teflon tubes.
Unfortunately, a coating also formed on the teflon tubes. As with the
other units this coating was easily removed by circulating a solution of
citric acid through the unit. The unit was operated at flow rates
between 15-40 gpm. Allowable run times between chemical cleanings
ranged from less than one week to nine weeks. Typical results are shown in
Figures 5,6, and 7.
Unit D
The last unit to be evaluated was operated for only a short period
of time. Unfortunately, the manufacturer supplied a unit designed for indus-
trial rather than municipal use. The cylindrical unit contained 66 UV lamps
enclosed in quartz tubes. Flow entered the unit and was split to allow it to
run perpendicular to the longitudinal axis of the lamps. A mechanical wiping
system composed of teflon washers around the quartz tubes was actuated on a
variable time frequency. This unit did not contain a UV light sensor. Normal
flow rate through the unit was 80 gpm.
Figure 8 presents the fecal coliform results obtained with Unit D.
Hhen the unit arrived the mechanical wining system was jammed. The first
set of data collected was obtained when the wiper system was not working.
After the wiper was replaced, the second and third sets of data were
obtained. The longest run in which reasonable results were achieved was
three weeks.
DISCUSSION
As in any comparison of equipment each unit had its favorable and
unfavorable features. Each unit was able to consistently achieve fecal
170
-------
coliform concentrations below the target level when it was clean. Keeping
the unit clean proved to be the major ooerational problem. Both units B
and C were able to operate for over two months at one point without
requiring cleaning. However, on other occasions both units only operated
properly for a week. The cause of the scale formation could not be
determined. Only water temperature correlated to any extent, with lower
temperatures seeming to favor longer run times. Because of the varying
frequency of scale formation a conclusion could not be drawn on whether
the mechanical winers, ultrasonics, or teflon tubes were effective in
extending the run time between chemical cleanings. The only conclusion
to be drawn was that citric acid was an adeouate cleaning solution. Any
full scale unit to be installed at Madison would be designed with a
chemical cleaning system. Included in this system would be a cleaning
solution mix tank, circulating pumps, piping connecting the solution tank
to the units, and a drain line. Since a unit would have to be removed
from service to clean it, the units would have to be built in modules to
allow for adequate disinfection in the remaining units while one unit was
being cleaned.
The UV sensors proved to be a valuable operational tool. Obviously
they can not be used to measure ultraviolet dose, but they can give an
indication of the results that can be expected. The fecal coliform
results for Unit A were grouped according to a combination of the absorbance
value of the wastewater and the relative transmittance reading of the
UV sensor. The geometric mean value of each group was then calculated as
shown in Figure 9. Although the resulting effluent fecal coliform values
were dependent on both the absorbance of the water and the sensor reading,
as long as the sensor was reading above 55% adequate results were obtained.
It seems reasonable that a similar relationship could be developed for
this type of unit at a different treatment facility. Since ultraviolet
disinfection does not result in a measureable residual as in chlorination,
some method is needed for the operator to determine if his unit is
performing adequately. Use of the UV sensors may be a satisfactory method.
The same approach was used with the sensor readings of the first
unit tested from manufacturer B. Fiaure 10 shows the increase in the
sensor readings during a typical run. Since a sensor reading was being taken
at three sites on the unit, they were added to obtain a "total resistance"
reading. The fecal coliform results for all samnles analyzed on this unit
were then plotted against the total resistance reading as shown in
Figure 11. This graph seems to indicate that when the total resistance
was less than 18,000 ohms effluent fecal coliform values less than 200 oer
100 ml were assured. When the total resistance was above 24,000 ohms
poor results were always achieved. Between these two values the effluent
fecal coliform number could not be assured. By using this type of
monitoring system the operator would be able to know when he was approaching
conditions which would require him to chemically clean his unit. He
would when be able to react to the situation before he discharged effluent
with fecal coliform values above his limit.
171
-------
Units A and D were equipped with LEDs to indicate proper operation
fo the UV lamps. A meter for each lamp performed this same function on
the units from manufacturer B. Unit C was equipped with a meter indicating
current draw by the unit. Although low current draw readings indicated
that one or more lamps were out, the ammeter was not capable of indicating
which specific lamp had failed. As a matter of practicality, the LEDs
provided the necessary information at a much lower cost than the meters
of Unit B. The meters showed the loss of efficiency of each lamp as
it aged, but this information could be obtained by recording the operating
time of each lamp.
CONCLUSIONS
1. Each unit was capable of consistently attaining an effluent
fecal coliform count of less than 200 per 100 ml when it was
clean. Each unit was plagued by a scale formation which
limited typical run times from one to eight weeks.
2. Citric acid proved to be an adequate chemical cleaner. Any
unit to be installed at the Nine Springs Wastewater Treatment
Plant would be equipped with a chemical cleaning system.
3. Since a unit has to be removed from service during cleaning,
an ultraviolet disinfection system should be designed in a
modular fashion. This will allow for continued disinfection
while one unit is being cleaned.
4. For the units which were equipped with ultraviolet sensors,
a relation could be developed between sensor reading and
effluent fecal coliform count. Each unit should be equipped
with at least one sensor to allow the operator to develop a
relationship which he could use to indicate if the unit was
operating pronerly.
5. LEDs indicating operation of each lamn were useful and should
be included on all units. Meters indicating lamp output are
useful but probably could not be justified on a large scale.
ACKNOWLEDGEMENTS
The Madison Metropolitan Sewerage District wishes to thank the
four manufacturers who provided units for the study. The author wishes
to express his appreciation to the analysts in the Nine Springs Laboratory
who performed the many analyses required to make these tests worthwhile.
172
-------
ABSORBANCE
,\ \
A-A
-2 \
I
V
80 +
H
LJ 60
o
(£
40 +
RELATIVE
TRANSMITTANCE
\
\
500
400+
EFFLUENT
FECAL
COLIFORMS
WEEK OF OPERATION
Figure 1. Unit A Results - Run 1
0'
o-o-o-o-o
-------
0.27
0.25
0.23
u.0.21
o
0.19 •-,
oc
o
m 0.17
ABSORBANCE
UJ
UJ
a,
IOO
80
60'
40
500-
•g 400'
O
2 300'
m 200-
.
§ too
o
RELATIVE
TRANSMITTANCE
\
\
EFFLUENT
FECAL
COLiFORMS
vv/-
WEEK OF OPERATION
Figure 2. Unit A - Runs 2,3,4
-------
10-
o-o-o-o-o-°*o
Ln
250-
200-
o
o
V)
Ul
o
o
150-
100+
50 +
EFFLUENT
FECAL
COLIFORMS
4-
WEEK OF OPERATION
Figure 3. Unit B Results - Run 1
-------
1000 ••
D
IV
D
800-.
600
500
450
EFFLUENT
FECAL
COL1FORMS
D, /
I D
WEEK OF OPERATION
Figure 4. Unit B Results - Run 3
-------
TOTAL
SUSPENDED
SOLIDS
o-o-o-o-o-o-o
250-
_j 200 •
2
o
2 150 •
^
Ul
z 100 •
3
o
0 50-
0
/EFFLUENT
FECAL
COLIFORMS
O
o Voo7 ^°
WEEKS OF OPERATION
Figure 5. Unit C Results - Run 1
177
-------
1000 ••
800 •-
SOO
500
450'
400- >
350-
o 300 • •
o
uj 250
2
O
O
o 200
150
100
50--
0
EFFLUENT
FECAL
15 GPM
'25 6PM
WEEK OF OPERATION
Figure 6. Unit C Results - Run 2
-------
O
o
tn
o
1000 ••
800-•
600-•
500- •
450-
400 ••
350- •
300- •
250 • •
200 • ••
150 • •
100' '
50
^^
•I
• *v
EFFLUENT
FECAL
COLIFORMS
J\r\ tfo
A
\/
\x'
5—1.
•10-
WEEK OF OPERATION
Figure 7. Unit C Results - Run 7
-------
1000- -
CO
o
EFFLUENT
FECAL
CGLIFORMS
WEEK OF OPERATION
Figure 8, Unit D Results
-------
GEOMETRIC MEAN
FECAL COLIFORM
RESULTS
RELATIVE
TRANSMITTANCE
< 55 %
>55%
ABSORBANCE
UNITS
<0.200 >0.200
410
35
552
80
Figure 9. Effect of Relative Transmittance
Readings and Hastewater Absorbance Values
on Fecal Coliform Results of Unit A
181
-------
CD
NJ
20-
19- •
I8< •
IT"
15-
14-
13
12 +
It'
10'
9-
o
o 7
o
6-
5
4-
3
I-
SENSOR RESISTANCE
READINGS
A-TOP PORT
O - MIDDLE PORT
8 - BOTTOM PORT
WEEK OF OPERATION
Figure 10. Unit B UV Sensor Readings
-------
iooa-
800-
600
400- •
350"
300- •
250' •
o
o
UJ
z
o
o
o
200
150- •
100- •
50
TOTAL
RESISTANCE
vs
EFFLUENT
FECAL
COLIFORMS
•O--
A-RUN |
O-RUN 2
H 1-
.cP
•4—I—I—I—I—I—I—I—I—I—»-
10 12 14 16 18 20 22 24 26 28
TOTAL RESISTANCE (1000 OHMS)
Figure 11. Unit B Sensor Reading vs
Effluent Fecal Coliform Counts
183
30 32
-------
5. UV DISINFECTION OF SECONDARY EFFLUENT: DOSE MEASUREMENT AND
FILTRATION EFFECTS
J. Donald Johnson, Robert G. Quails, Kent H. Aldrich and Michael P. Flynn
Department of Environmental Sciences & Engineering
University of North Carolina at Chapel Hill
ABSTRACT
The first phase of this study involved an ultraviolet (UV) disinfection
pilot plant study comparing: filtration, water quality parameters, and two
reactors. The pilot plant study directed us to laboratory experiments
involving: (1) the development of a method for in situ measurement of dose
rate using a calibrated bioassay, (2) experimental verification of a method
for calculating dose rates, (3) evaluation of the role of lamp spacing in
dose efficiency, and (4) simulation of UV disinfection.
A bioassay method was developed to measure average dose rate (i.e.,
intensity) within a UV reactor. The survival of spores of Bacillus subtili s
was determined as a function of UV dose in order to standardize the sensiti-
vity of the spores. Spores were added to unknown systems and the survival
could be used to determine the average dose rate. A modification was used
for flowthrough reactors, in which spores were injected as a spike and
collected at a known time from injection.
Spectrophotometric measurements were found to significantly overestimate
the UV absorbance in wastewater because of scattering. A method to correct
for scattering was tested. A point-source summation method for calculation
of dose rate was verified by bioassay measurements in a simple cylinder. This
calculation method was also applied to multiple lamp reactors. A method for
simulating survival in complex flowthrough reactors was presented and a
simulation of our pilot plant runs corresponded reasonably well with the
observed survival. Mixed media filtration significantly improved disinfection
in pilot plant media experiments. A laboratory experiment showed that a
relatively small number of coliforms were protected inside particles, but they
were the factor limiting disinfection at -3 or -4 logs survival.
INTRODUCTION
Environmental problems associated with chlorination have prompted
research into alternatives for disinfection of wastewater effluents. Resi-
184
-------
duals and by-products can be toxic to aquatic life in receiving waters (15)
and they may form carcinogenic by-products (8). In addition, chlorination is
less effective in killing viruses, spores and cysts than in killing bacteria.
One disinfection process which would not be expected to produce undesirable
by-products is ultraviolet light (UV).
The Environmental Protection Agency has funded several pilot or full
scale investigations of UV disinfection of wastewater (5,9,11,12). While
these pilot studies of UV disinfection have generally been successful at
meeting disinfection goals, comparison, both within and between these and
most other UV studies, has been limited because there has been no direct
method of measuring UV doses, nor has there been a substantiated method of
calculating doses in the complicated geometries of a practical reactor. In
addition, lack of dose measurement methods has prevented the controlled
evaluation of effects of variables such as UV absorbance of the water,
filtration, reactor design and the varying sensitivity of different organisms.
The first phase of this study was a pilot plant study comparing:
(1) the effects of mixed media filtration, (2) the effects of randomly
varying water quality parameters, and (3) two UV disinfection reactors
employing different lamp spacing. Experience from the pilot plant study
directed us to a laboratory experimental second phase involving: (1) develop-
ment of a method for in situ measurement of dose rate (i.e., intensity) using a
calibrated bioassay, (2) experimental verification of a method for calculating
dose rates or intensities, (3) separation of effects of absorbed and scattered
UV light and its relation to spectrophotometer measurement, (4) evaluation
of the role of lamp spacing in dose efficiency, and (5) simulation of UV
disinfection.
The following are several problems with the dose estimation in previous
studies of UV disinfection. (Studies exemplifying these problems are indi-
cated in parentheses.)
1. UV radiometer detectors measure intensity on a planar surface. Thus,
they don't correctly measure the 3-dimensional intensity (i.e., dose rate) to
which a cell may be exposed near a long tubular lamp (3,9).
2. A UV radiometer detector positioned in the wall of a disinfection
reactor can't be used to estimate the average dose rate within the entire
reactor (3,2).
3. Wastewater contains particles which scatter UV light so that spectro-
photometers tend to overestimate the UV absorbance (9).
4. Equations have been used which incorrectly calculate the dose rate
near a tubular lamp in an absorbing solution (9,12,13).
5. In flowthrough systems there is a distribution of exposure times
not simply related to volume and flow rate (9,12).
MATERIALS
All measurements of intensity at 254 nm were made with a calibrated
International Light 500 radiometer. Measurements of UV output at 254 nm were
185
-------
made by integrating intensity measurements3 made far from the lamp, over a
spherical surface centered on the lamp centroid (5). To obtain accurate
dose-survival data,suspensions of bacteria were irradiated in a collimated
beam apparatus (Fig. 1), To test calculations of UV intensity in a cylindri-
cal geometry, suspensions of spores were irradiated for a fixed time inside
the cylindrical apparatus shown in Fig, 2, A moveable paper tube was located
between the lamp and the quartz tube so that the lamp could be warmed up and
an exact exposure made. Cylinders of different radii were used. Suspensions
were well stirred. Fulvic acid was used to vary absorbance.
Bacillus subtilis (ATCC 6633) spores were used for bioassays of UV dose.
Preparation of spore stocks is described elsewhere (5). Spores were suspen-
ded in buffered water (1) and plated on Thermoacidurans agar. For laboratory
experiments total or fecal coliform density was determined by the membrane
filter technique (1) ; however,in pilot plant experiments the MPN procedure
(1) was used,and both total and fecal coliforms were carried to the confirmed
level. Methods used for water quality parameters are described elsewhere.
Spectrophotometric UV absorbance (254 nm) was measured with a Gary 219
spectrophotometer. For some experiments, a special quartz cuvette, ground so
as to be translucent on the side nearest the detector, was used to correct
for scattering of UV light (14). For pilot plant experiments, two disinfec-
tion units were used: an Aquafine CSL-6, and a Pure Water Systems (PWS) 1-75.
Both filtered and unfiltered secondary effluent were disinfected. Filters
were pressurized and contained sand-anthracite media.
RESULTS AND DISCUSSIONS
Bioassay Method for Measurement of Dose Rate
A bioassay method was developed to measure average dose rate in flow-
through reactors as well as to verify a method of dose rate calculation.
Dose is defined as:
Dose = (dose rate) (exposure time) (1)
or, in units:
9 2
mW-sec/cirT = (mW/cm ) (sec) (2)
The term "dose rate" has been used instead of the more familiar "intensity"
because of the ambiguities in definitions of intensity. The survival
(NS/NO) of organisms is usually a function of dose:
NS/N0 = fn(dose) (3)
where N0 and N,, are the density of organisms before and after irradiation,
respectively. Equations 1 and 3 imply that dose rate and exposure time may
be varied reciprocally to obtain the same survival.
The survival of spores of Bacillus subtilis was determined as a function
186
-------
of the UV dose in order to "calibrate" the sensitivity of the spores. Since
dose rate, as measured by a radiometer, was only applicable in a collimated
beam, the spores were exposed for varying periods of time to a collimated
beam of UV light in a stirred petri dish (Fig. 1). The dose rate at the
surface of the suspension was measured. Since fluid depth and absorbance
were minimal, the dose could be calculated based on the measured dose rate
and the exposure time. In cases where absorbance was significant, the
average dose rate was calculated using an integration of Beer's law over the
fluid depth. Calibration curves of log survival vs. dose were constructed
(Fig. 3) and found to be quite reproducible over several months. The dose
rate may be determined in an unknown system by: (1) determining the survival
(N /No); (2) reading the dose corresponding to the observed survival using
the calibration curve (Fig. 3); and (3) using the known exposure time in
eq. 1 to calculate average dose rate.
Separation of Effects of UV Absorbance and Scattering
Calculation of average UV dose rate requires an absorbance measurement.
Wastewater effluents contain particles which may scatter as well as absorb
the UV light. Bioassay experiments showed that scattered UV light was still
effective for killing bacteria. Since the usual spectrophotometric measure-
ments do not separate scattering and absorbance, we needed a way to separate
the two. An established method using a frosted cuvette for both the blank
and sample allowed a correction for most of the scatter (14). A piece of
oil saturated paper placed on the cuvette face may also be used.
We tested this technique against a bioassay method to separate absorbance
and scattering. A sample of tertiary effluent (14 NTU turbidity) was filtered
through a 0.45y filter. Suspensions of intermediate turbidity were made by
mixing portions of the filtered and scattered sample. Thus, the soluble
absorbing component was held constant and the particulate component varied.
Samples were spiked with Bacillus spores and irradiated in a petri dish in
the collimated beam apparatus. The average dose rate in the suspension was
assayed. By using the integrated form of Beer's law (7) we determined the
absorbance which would yield the observed assayed dose rate. The assayed
absorbance for the suspensions of varying particulate content is shown as a
function of the spectrophotometric absorbance (Fig. 4). The difference
between the spectrophotometric absorbance and the assayed absorbance was the
scattering component. The soluble absorbance, particulate absorbance, and
scattering were 47 percent, 41 percent, and 12 percent, respectively, of the
spectrophotometric absorbance. The frosted cuvette method showed a slightly
lower scattering component. The scattering component was estimated to have
averaged 9 percent in our pilot plant studies. The soluble absorbance was
60 percent to 80 percent of the spectrophotometric absorbance in most of the
secondary effluents measured.
Calculation of Dose Rate
Common radiometer detectors cannot be used to measure dose rate near a
tubular lamp because they measure energy flux on the planar surface of the
detector. Light received at angles other than 90° to the surface of the
187
-------
is attenuated since the surface of the detector intercepts a smaller cross
section of the rays. The detector "sees" primarily the portion of the lamp
directly in front of it. Biological cells in motion in a solution, however,
present a 3-dimensional target and they respond to the 3-dimensional dose
rate from all angles within a disinfection reactor (6) .
To calculate the UV dose rate at a point near a tubular lamp in an
absorbing solution, we used an equation which we call the point source
summation (PSS) calculation (4,6,10). This equation assumes that a line
segment source can be treated as the sum of a number of point sources. We
can consider a cylindrical coordinate system around a line segment light
source surrounded by a quartz sleeve (Fig. 5) . The total line source of UV
output OPT is divided into N point sources each of which has strength S
(units in Watts) .
A = OPT/N (4)
The dose rate at a point I, , due to one point source (Z^) can then be
treated as the product of the Spherical spreading times the attenuation due
to absorbance over a definite path length (P-P]_) .
Z(ZT), (R,Z ) = [S/4Tr(R2 + Z2 )]exp[-a(R-R.,)P/R] (5)
L C LiL- -L
where a is the absorbance of the medium and the other geometry is shown in
Figure 5. The total dose rate at point I,- . is the sum of the contribu-
tions of each point source (at each Z^) over cche source length
(6)
The use of this calculation requires two measurements: absorbance of the
water, and the lamp UV output (5).
To test the PSS calculation, we compared the calculated average dose
rate inside a cylinder (Fig. 2) to that measured by the spore bioassay. We
used the PSS calculation in a computer program to average the dose rates over
the volume of a cylinder around a lamp. We did this for a series of cylin-
ders of varying radii and for fluids of different absorbances. The survival
of the spores was measured and the assayed average dose rate determined as
outlined previously,
The PSS calculations were generally verified by the bioassay measure-
ments. Figure 6 shows a comparison between the calculated PSS curves (solid
lines) and the bioassay data (data points). The correspondence was good both
for cylinders of different radii and for fluids of varying absorbances. The
stirring device may have produced some shadowing loss in the 2.5 cm cylinder.
188
-------
We also performed the same experiment using spores spiked in a secondary efflu-
ent, and PSS calculations were within 10 percent of the bioassay dose rates.
We also applied the calculation methods which had been used in some previous
studies (9,12) to these cylinders and those methods gave results which differed
greatly from our experimental average dose rates (5).
Practical UV reactors are flowthrough systems and have a distribution
of exposure times. To use the bioassay of dose rate in a flowthrough system
we needed a way to determine a definite exposure time. To do this we used
the spores in a manner analogous to a tracer injection study. To demonstrate
this method we used a flowthrough tube surrounding a UV lamp. Spores were
injected into the flowstream of water at the entrance to the tube and the
outflow fractions were collected in a rotating sampling tray as a function of
time from injection. The injection was performed with the light on and
repeated with the light off. The density of the unirradiated spores (No) is
shown in Fig. 7. The distribution of unirradiated spores reflects the reten-
tion time distribution (RTD). The density of surviving irradiated spores
(Ns) is shown in Fig. 7. The survival (NS/NO) was calculated for each flow
fraction separately by comparing spore densities in the corresponding irra-
diated and unirradiated fractions at a given time from injection. The average
dose rate was then determined for each fraction by finding the corresponding
dose from the calibration curve and dividing by the time from injection. The
assayed dose for each flow fraction is also plotted in Fig. 7. The slope of
the regression line of the assayed dose vs. time from injection was equal to
the average of the assayed dose rates in the separate fraction. A modification
of the spore injection bioassay may be used to measure average dose rate in
full scale reactors.
The assayed average dose rates within the flowthrough tubes (Fig. 6,
"injection expts." data points) corresponded well with the calculations of
the PSS model (Fig. 6,lines). The distribution of unirradiated and irradia-
ted viable spores in Fig. 7 also showed that nearly all of the surviving
spores emerged from the tube before the average retention time. This illus-
trates the important effect that flow dispersion can have on the disinfection
efficiency.
Calculation of Dose Rate in Multiple Lamp Reactors
To calculate average dose rate in multiple lamp reactors we used the
following method: (1) dose rate at each point was considered to be the sum
of the contributions from each lamp calculated by the PSS model; (2) dose
rate was mapped at each point on a grid of the cross-sections of the reactor;
and (3)dose rates were averaged over the cross-sections and along the length
of the reactor.
We found that UV lamps transmit little of the UV light coming from
adjacent tubes (5) or absorb nearly all UV output striking them from neigh-
boring lamps. Thus, it was necessary to make calculations which took this
shadowing into account. Our calculations also made these simplifications:
that reflections from the reactor walls was negligible under actual operating
conditions, and that reflection and refraction by the quartz sleeves were
189
-------
negligible.
There are divergent views on the design of UV reactors. Some of these
viewe are based on improper equations or conventional wisdom rather than cal-
culation or experimental measurement. This is because of the lack of ade-
quate and comparable methods for measuring or calculating UV dose (e.g., 13).
Our models can be useful for research and development of reactor design. We
applied our calculations to contrast the efficiency of the different schemes
of lamp spacing in absorbing fluids. Any surface or object which absorbs UV
energy (e.g., walls, baffles, other lamps), in addition to the unavoidable
absorbance of the water itself, reduces its efficient use. The product
of dose rate times reactor volume is a factor which is directly proportional
to the effectiveness of the unit at treating fluid volumes of water at a
given flow rate and flow conditions. This factor isolates the effectiveness
of the dose rate regime or intensity distribution from the effects of flow
dispersion or hydraulic characteristics and can be used to compare reactors
of different lamp spacings and volume. At a given flow rate and number of
lamps, a close lamp spacing gives a higher average dose rate or intensity but
at the sacrifice of shorter detention time because of the smaller volume of
the unit. We showed with calculations how the distance the light was allowed
to penetrate,before being lost on a neighboring lamp or wall,affects the
efficiency of light use. Figure 8 shows the dose rate-volume product in
cylinders of radius R or fluid depth around a UV lamp. The point at which
the lines level out is the radius at which most UV light has been absorbed
and no further improvement in efficiency occurs. In other words, the decrease
in intensity just balances the increase in detention time as the radius or
fluid depth increases. For an absorbance representative of secondary
effluent, 0.16, it can be seen that walls or other obstructions within 5 cm
can absorb a significant amount of available UV light. Fluid depths less
than 5 cm are less efficient at this absorbance. Two reactors used in the
pilot plant experiments were compared on the basis of their dose rate-volume
products (Table 1)„ The reactor with lamps placed close to one another and
the walls (PWS unit) had an average dose rate or intensity almost twice as
high as the other reactor (Aquafine). However, the PWS reactor had a much
smaller volume (and shorter retention time) so the dose rate-volume products
were almost equal. However, the PWS reactor used a greater lamp wattage.
We used a term we called the dose rate-volume "efficiency" (dose rate-volume
product/input wattage) to compare the efficiency of the use of the lamp
wattage, The PWS was much less efficient because of the proximity of the
lamps to the walls and the wall and neighboring lamp absorption of the light.
The dose rate-volume product does not consider the effects of non-ideal
flow. Although the dose rate-volume products of the two reactors were nearly
equal, the PWS reactor gave from 0.6 to 2.1 log units greater survival of fecal
colifomis than the Aquafine at the same flow rate because the less ideal
hydraulic characteristics of the PWS unit gave severe short-circuiting of
flow in the PWS reactor. Thus, the effects of flow dispersion must be con-
sidered as well as the dose-rate or intensity regime in determining the ulti-
mate disinfection efficiency or total dose produced by a given lamp wattage
into a given volume of fluid.
190
-------
We also used simulation of a full-scale reactor, operated in NW Bergen
County, N.J. (12) to show the effect of varying lamp spacing on the UV light
use efficiency and an analysis of the relative costs (5).
Simulation of Dose and Disinfection in Flowthrough Reactors
The second factor in calculation of dose, exposure time, can lead to
as much error in calculations as dose rate or intensity. In flowthrough
reactors there will be a distribution of retention time. Figure 7 shows
clearly neither the retention time (RT) calculated from flow rate and volume
nor even the average RT determined from dye studies can be used to predict
the average survival. Since survival is not linearly related to dose, the
average dose is insufficient to predict the average survival over the RT
distribution,but the survivor density must be calculated for each flow
fraction and then summed.
The following equations will show how the density of survivors (N )
may be predicted from the following data: (1) coliform density in inflow
(N0), (2) average dose rate (DR), either measured or calculated, (3) reten-
tion time distribution, and (4) dose-survival curve (determined accurately,
e.g., in collimated beam apparatus).
For an aliquot of volume V^- entering the reactor at time to, the aliquot
will exit in n fractions of volumes Vj_ at times t^. Survival in each fraction
is some non-linear function (fn) of dose.
N /N = fn(dose)
so
(7)
Dose for the ith fraction = (DR)(t.) (8)
Survival in the ith fraction = N /N = fn[(DR)(t.)] (9)
Average density of survivors, Nc = N V (fn[(DR)(t )])/V (10)
^ Q J- It
Data from a dye study on the RTD may be put in a form to use in these equa-
tions. The area under a curve of dye concentration vs. time to set equal
to Vt (and may be thought of as a 1 ml aliquot entering the reactor). Then,
V. = (At)(relative dye concentration)/V (11)
191
-------
For a computer simulation of average survival, the RT distribution, and
dose-survival curve data pairs were fed into arrays and intermediate values
needed in eq. 10 were generated by linear interpolation.
As an example of simulation of survival in a flowthrough reactor, we
simulated runs with the Aquafine reactor. These simulations we then compared
to the observed survival in the pilot plant experiments. The average dose
rates calculated by the PSS model for two levels of applied voltage and the
input data in equation 10 were used. The RT distribution was measured with
dye injection and adjusted to the correct flow rate. We lacked the methods
at the time of the pilot plant runs to determine an accurate coliform dose-
survival curve, so one was determined some time later for a sample from the
same site.
The average log survival predicted by the simulation corresponded
reasonably well with that observed in the pilot plant runs (Table 2). Some
deviation might be expected since the dose-survival curve was based on one
sample taken at a later date. Further research should involve simulation
using data obtained simultaneously with full reactor runs.
Simulation takes into acount the factors of the dose rate-volume
characteristics as well as the effects of flow dispersion and sensitivity
of the target organisms. It can be useful tool for research and development
of reactor design. For example, it can be used to find optimum lamp confi-
gurations and tradeoffs with flow dispersion. It can be used to predict the
design parameters needed for a specific situation so that costly overdesign
is not necessary. The predicted survival of a standard coliform sample at
a given flow rate may be used to compare a number of different reactors.
The simulations may also be used to prepare empirical curves of predicted
survival vs. flow rate, operating voltage, water quality, etc., for a parti-
cular installation as a guide to continuous operation.
Protection of Cells Inside Particles and Effects of Filtration
In our pilot plant experiments, an extended aeration secondary effluent
was subjected to mixed media filtration. Both filtered and unfiltered efflu-
ents were subjected to UV disinfection in two UV reactors, at two different
flow rates and two levels of applied lamp voltage. The filtered effluents
showed significantly better disinfection (Table 3). Total coliform log survi-
val was 0.33 to 0.79 log units lower in the filtered treatments. The effect
of filtration on UV absorbance was small and did not account for the disin-
fection differences. The differences in suspended solids, turbidity, and UV
absorbances indicate that filtration tended to remove the larger particles
which had relatively little effect on the absorbance. Average dose rate cal-
culations and simulation supported the idea that the filtration effect was
not due to the lower absorbance after filtration. We concluded that a rela-
tively small number of coliforms were protected inside particles but that
these tended to be removed by filtration.
We performed a laboratory experiment to support the hypothesis that
particle protection is the major effect increasing disinfection after filtra-
192
-------
tion. We determined the dose-survival curves of an unfiltered effluent sample
and the same sample passed through a 70y and 8y pore size filter. Since
coliforms are about l-2y in size, the 8y filter allowed only single cells or
very small aggregates to pass. The survival curve of this fraction (Fig. 9)
shows disinfection continuing beyond -4.5 log units survival where survivors
were undetectable. Curves for the 70y filtered and unfiltered samples tend to
level out after -2 or -3 log units survival. The coliforms not passing the
8y filter were extremely resistant to UV. Since the curves were similar
until less than about 10, or 1 percent, of the coliforms were surviving, the
protected coliforms appeared to be a small minority but became the limiting
factor to disinfection at levels needed to meet legal standards.
Other Pilot Plant Results
The Aquafine reactor met the disinfection goal of 200 MPN/100 ml in
every case. However, the PWS reactor did not because of short circuiting
of flow. Changes in applied lamp voltage and flow rate produced relatively
small changes in survival because, as can be seen from the dose-survival
curve in Fig. 9 for example, the dose-survival curves level out at -3 or -4
log units survival.Stepwise multiple regression of randomly varying water
quality parameters on log survival of coliforms showed no consistant correla-
tions. This lack of correction was probably due to the relatively small
variation in UV absorbance and the lack of response of kill to dose increases
at -3 or -4 log units survival. The significant correlations were spectro-
photometric absorbance was predicted well by coliform densities, or if these
were not considered, by COD, turbidity and suspended solids together.
CONCLUSIONS
If the disinfection of single coliform cells in wastewater under ideal
flow conditions is considered as "ideal efficiency" then the results of this
report show the following to be the chief factors limiting ideal efficiency
in practice: (1) protection of cells inside particles, (2) flow dispersion
and poor mixing across dose rate gradients, and (3) shadowing and absorption
of UV light by walls within a reactor.
ACKNOWLEDGEMENTS
We appreciate the collaboration of Dr. Donald E. Francisco, Dr. Forrest
D. Mixon, Douglas W. Van Osdell, Marion Elliott Deerhake and Thomas S. Wolfe
on the pilot plant phase of this project. We are indebted to the project
officer Albert D. Venosa for his valuable support, advice and reviews. This
work was supported as grant no. R 804770010 by the Municipal Environmental
Research Laboratory, U.S. E.P.A., Cincinnati, Ohio.
193
-------
LITERATURE CITED
1. American Public Health Association. 1975. Standard Methods for the
Examination of Water and Wastewater. 14th ed. A.P.H.A.,
Washington, D.C.
2. Department of Health, Education and Welfare. 1966. Division of
Environmental Engineering and Food Protection. Policy statement on
the use of the ultraviolet process for disinfection of water.
Washington, B.C., April 1.
3. Huff. C.B., H.F. Smith, W.D. Boring, and N.A. Clarke. 1965. Study of
ultraviolet disinfection of water and factors in treatment
efficiency. Public Health Reports 80:695.
4. Jacob, S.M. and J.S. Dranoff. 1970. Light intensity profiles in a
perfectly mixed photoreactor. Am. Inst. Chem. Eng. J. 16: 359.
5. Johnson, J.D. and R.G. Quails. 1981. Ultraviolet disinfection of
secondary effluent: Measurement of dose and effects of filtration.
Report of EPA project R804770010, Municipal Environmental Research
Laboratory, Cincinnati, Ohio.
6. Kase, K.R. and W.R. Nelson. 1978. Concepts of Radiation Dosimetry.
Chap. 5. Pergamon Press, N.Y.
7. Morowitz, H.J. 1950. Absorption effects in volume irradiation of
microorganisms. Science 111: 229-230.
8. National Research Council. 1980. Drinking Water and Health. National
Academy Press, Washington, D.C. 393 pp.
9. Petrasek, A.C., H.W. Wolf, S.E. Edmond, D.C. Andrews. 1980. Ultraviolet
disinfection of municipal wastewater effluents. E.P.A.-600/2-80/102 ,
262 pp.
10. Rockwell, J. 1956. Reactor Shielding Manual. Van Nostrand, Princeton,
N.J.
11. Roeber, J.A. and P.M. Hoot. 1975. Ultraviolet disinfection of activated
sludge effluent discharging to shellfish waters. E.P.A.-600/2-75-060,
85 pp.
12. Scheible, O.K. and C.D. Bassel. 1981. Ultraviolet disinfection of a
secondary wastewater treatment plant effluent. E.P.A.-600/S2-81-152.
13. Severin, B.F. 1978. Disinfection of municipal wastewater effluents with
ultraviolet light. Paper presented at the annual meeting W.P.C.F.,
Anaheim, California.
194
-------
14. Shibata, K., A.A. Benson, and M. Calvin. 1954. The absorption spectra
of suspensions of living microoganisms. Biochem. et Biophys. Acta
15: 461.
15. Ward, R.W. and G.M. DeGrave. 1978. Residual toxicity of several
disinfectants in domestic wastewater. J. Water Poll. Control Fed.
50: 46.
195
-------
Table 1. Comparison of Aquafine and Pure Water System units.
CHARACTERISTIC
Input wattage
UV output, total W
1 2
Calculated average dose rate (mW/cm )
2
Dose rate-volume product (mW/cm ) (1)
Dose rate-volume "efficiency"
2
[(mW/cm )(1) /input wattage]
Aquafine
240
54.6
8.5
93.5
0.390
PWS
350
68.2
16.2
94.2
0.269
at absorbance = 0.17
Table 2. Actual vs. simulated survival (S) of total coliforms
in a Sandy Creek secondary effluent.
Lamp voltage
60
128
Ave. intensity
2
(mW/ cm )
5.1
8.5
Simulated log S
- 3.00
- 3.61
Pilot plant
- 3.29 (- .
- 3.69 (± .
log S
13)
16)
Table 3. Inactivation shown as mean -log survival of fecal
coliforms in unfiltered and filtered secondary
effluent, broken down by filtration status,
applied voltage and flow rate. Standard deviations
(of log units) shown in parentheses.
Aquafine
Flow rate (1/s) Voltage
4.92 60
128
2.27 60
128
Unfiltered
Fecal coliforms
3.08 (.20)
3.41 (.23)
3.91 (.23)
3.47 (.28)
Filtered
(-log survival)
3.88 (.19)
4.17 (.18)
4.29 (.17)
3.92 (.24)
196
-------
SHIELD-*
SUPPORT
STAND -1
HJV LAMP
72cm
ultraviolet
lamp
quartz tube —
plexiglass
cylinder ~~
COLLIMATING
TUBE
-PETRI DISH
I-MAGNETIC STIRRER
stirring
device
Fig. 1. Collimated beam apparatus
sliding black
paper tube
between lamp
and quartz
tube i
Fig. 2. Cylindrical batch
irradiation apparatus
0
> -2
rr
-3
o
-4
-5
a-
o«
A
16
24
32
40
DOSE (mW-sec/crr/.)
Fig. 3. Log survival of Bacillus subtilis vs. UV dose in a collimated
beam of known dose rate. Different symbols represent 5
different runs. Data from doses of 10-30.5 mtf/cm2 appeared
linear and fit the regression line Y = .167x + 1.01 (r = .98).
197
-------
.40
LU
O
.30
CD
tr
o
00
Q
LU
CO
CO
.10
0
Depth (cm
o 2
o 3
A 4
cc
Ul
t-
o
CO
CO
CD
ce
2
to
CD
t t
7.0 11.5 14.0
Turbidity
0 .10 .20 .30 .40
SPECTROPHOTOMETRIC ABSORBANCE
Fig. 4. Spectrophotometric absorbance vs. absorbance
measured by the bioassay method for a
Chapel Hill tertiary effluent sample. The
soluble UV absorbance was kept constant and
the particulate concentration varied by
diluting the unfiltered (14 NTU) sample with
filtered (.07 NTU) sample. The solid line
represents an exact correspondence between
the two methods. The dotted line is a
regression through the data points. The
soluble and particulate absorbance and
scatter components of the Spectrophotometric
absorbance of the unfiltered sample are
indicated.
198
-------
Ri
QUARTZ
LAMP WALL
Fig. 5. Cylindrical reactor geometry for point
source summation calculation. (Modified
from Jacob and Dranoff [4]).
345
RADIUS OF CYLINDER (cm)
Fig. 6. Average dose within a cylinder of radius R.
The solid lines were calculated by point
source summation for several different
absorbances. Data points represent
bioassayed average dose rate within the
cylinders of various sizes. Data points
for 1.32 and 1.59 cm radius were obtained
from flowtbrough tubes rather than batch.
-------
S(a)UNIRRADIATED
PORES
SURVIVING
SPORES
246
TIME FROM INJECTION (sec)
Fig. 7.
Assay of average dose rate in 1.32 radius flowthrough
tube by injection of spores and collection of separate
fractions over time after injection. Fig. A shows the
concentration of spores vs. retention time (time after
injection) with no irradiation. Fig. B. shows the
spore concentrations as a function of retention time
when irradiated at the same flow rate. Also shown is
the assayed dose calculated from the NO,NS of each
fraction collected and the calibration curve. For
clarity, the viable spore distribution curves are
shown for only one experiment but the assayed dose
rate for each point is, (the assayed dose)/(retention
time), and the average corresponds to the slope of the
regression line through all the points forced through
the origin.
200
-------
RADIUS (cm)
Fig. 8. Effectiveness of various fluid depths in cylinders
of radius R around UV lamps. Calculated values of
the product of average dose rate in a cylinder, of
radius R, times the volume of that cylinder are
shown vs. the other radius of the cylinder for fluids
of absorbances 0, .16 and .32.
o Filtered through 8/1 filter
A Filtered through 70/i filter
DOSE (mW-sec/cm2)
Fig. 9. Effect of filtration on survival of total coliforms
in Sandy Creek with arrows indicating limit of
detectibility for exposure in which no survivors
were found.
201
-------
6. PILOT INVESTIGATION OF ULTEAVIOLET WASTEWATER DISINFECTION AT THE
NEW YORK CITY PORT RICHMOND PLANT
0. Karl Scheible
Principal Engineer
HydroQual , Inc.
1 Lethbridge Plaza
Mahwah, New Jersey 07^30
Angelika Forndran, P.E.
Project Engineer/Manager
New York City Department of Environmental Protection
Bureau of Science and Technology
51 Astor Place
New York, New York 10003
William M. Leo
Senior Engineer
HydroQual , Inc .
1 Lethbridge Plaza
Mahwah, New Jersey 07430
ABSTRACT
A major EPA-NYC funded project investigating ultraviolet disinfection
of secondary effluent and of CSO wastewaters has been started and is
entering the experimental phase. This paper presents a progress report on
the study.
The pilot plant is operational with two 100 lamp submerged bulb systems
in place. A third unit, a non-contact teflon tube system will be in place
by the Spring of 1982. Each unit will receive wastewater flows between
0.95 and 4.5 ML/d (0.25 and 1.2 mgd) under a controlled experimental
program. The field evaluation was started in December 1981 and will
continue for a period of 15 to 18 months. The major efforts which are
currently underway are the development of a generalized mathematical model,
a detailed characterization of the hydraulics through each system, and a
direct comparison of the two submerged systems, which differ only in the
spacing of the lamps.
INTRODUCTION
A large scale pilot investigation of wastewater disinfection by
ultraviolet light irradiation (UV) is being conducted at New York City's
Port Richmond Water Pollution Control Plant, Staten Island, New York. The
project is jointly funded by the United States Environmental Protection
Agency (Municipal Environmental Research Laboratory, Cincinnati, Ohio) and
202
-------
the City of New York Department of Environmental Protection. HydroQual,
Inc. is the Principal Investigator for the program.
This is a progress report. The experimental program has been underway
for approximately three months of the anticipated 18 month schedule, and
elements still remain in the construction of the pilot facility. Thus,
little can be presented in the way of actual field data or conclusions
regarding the operation of the facility. Rather, this presentation will
center on a description of the facilities, the scope of work and a
discussion of tasks which have been completed to date. Where appropriate,
field data will be presented, although the reader is cautioned that these
are preliminary and cannot be rigorously interpreted.
SCOPE OF WORK
The major objective of this project is to establish and demonstrate a
rationally based protocol for the design of ultraviolet disinfection
systems. This is an outgrowth of the conclusions and recommendations
reached in the recently completed study at the Northwest Bergen County
WPCP, Waldwick, NJ(3). Much of the investigative work in UV disinfection
to date has been empirically based, making it difficult to compare systems
or to test the sensitivity of a design to various operating variables. The
Port Richmond project will involve the development of a rational design
protocol and will then demonstrate the validity and application of the
method by collection of actual field performance data. Three systems will
be tested, each differing in their basic design configuration. The design
method considers water quality, system hydraulics, and system geometry
(lamp spacing), all of which will be study elements of the experimental
program.
Other objectives of the field program will involve the evaluation of
operation and maintenance requirements, photoreactivation, the impact of
wastewater variability, and the development of capital and O&M costs. The
experimental phase is expected to end in the Spring of 1983, with a formal
report to be issued by late Summer, 1983.
PILOT FACILITIES
The Pilot Facility is located at the Port Richmond Water Pollution
Control Plant (WPCP), one of New York City's twelve operating wastewater
treatment plants. Port Richmond, on the northern shore of Staten Island,
receives residential and industrial wastewater from a 62 square km drainage
area. The WPCP is a step aeration activated sludge facility which was
upgraded during the 1970s to provide secondary treatment. It is designed
to treat an average flow of 227 million liters per day, ML/d, (60 million
gallons per day, mgd) in the secondary system and a maximum of 454 ML/d
(120 mgd) through the primary system. Present flows average 151 ML/d (40
mgd) during dry weather and up to 378 ML/d (100 mgd) during storm events.
Flow in excess of the secondary design flow is bypassed directly from the
primary tanks to the final effluent channel.
203
-------
A schematic of the Port Richmond WPCP is shown in Figure 1. The UV
test facility is located north of the building containing the secondary
aeration tanks. A layout of the UV pilot plant is shown in Figure 2.
Secondary plant effluent is pumped from the effluent channel via an
existing spray water pump located in the sludge pump gallery. Primary
plant effluent, which will simulate the quality of settled combined sewer
overflow (CSO) wastewater, is pumped directly from the bypass channel
during storm events. Both types of flows are pumped into a constant head
tank just outside the temporary building (6.1 m x 7.6 m) housing the UV
systems. From the head tank, the effluent flows by gravity through the UV
units and is discharged to the bypass channel joining regular plant
effluent prior to the outfall. Each UV system can receive a flow between
0.76 ML/d (0.2 ragd) and 4.5 ML/d (1.2 mgd) . Palmer-Bowl us flumes have been
inserted into the effluent channels, which, in conjunction with ISCO Model
1700 meters, are used to monitor the flow of each system.
A description of the UV units is summarized in Table 1. There are two
UV systems inside the temporary building as shown in Figure 3. Each has an
influent and effluent tank attached to the lamp units. Overall dimensions
of each are 1.07 m wide by 2.74 m long and 3.05 m long, respectively. The
only difference between the units is the lamp spacing: 1.25 cm and 5 cm
(defined as the closest distance between the surfaces of two quartz
sleeves). The lamp battery dimensions (internal) are 0.74 m long by 0.69 m
high and 0.73 m wide for the widely spaced unit and 0.40 m long by 0.35 m
high and 0.73 m wide for the narrowly spaced unit. Each system contains
100 lamps in a symmetrical (10 x 10) array perpendicular to flow. The
lamps are Voltarc 40 Watt (nominal) G36T6VH units. Each is enclosed in 23
mm diameter quartz sleeves. The rated output at 253.7 ran for each lamp is
approximately 14 W.
The lamps are cleaned by a mechanical wiper system. The wiper blade is
cable driven at a variable stroke rate by a pneumatic cylinder. Each UV
system has a separate power panel containing shutoff switches for each of
three banks of lamps (divided into 30, 40, 30 lamps) and pilot indicators
for each lamp. UV intensity monitors for each bank of lamps, elapsed time
of operation totalizers, lamp ballasts and the wiper timing devices are
also mounted in the power panel. The remaining equipment within the
building includes an air compressor for the wiper mechanism, the variac
control for modifying lamp intensity, flow meters and additional power
distribution and lighting panels.
A third UV system is proposed for installation outside the temporary
building. The unit will connect to the same influent and effluent lines as
are currently used. It differs from the other two in that the wastewater
flows through teflon tubes. The ultraviolet bulbs are parallel to the flow
and are not immersed in the wastewater, The size of this unit is 4.9 m
long by 0.91 m high by 1. 52 m wide.
Construction of the UV facility was started in May, 1981. By October
the facility was complete, including electrical work and installation of
the two UV systems. A startup period followed during which various
operational difficulties were resolved. The pilot facility was ready for
204
-------
Effluent to
Kill Van Kull
O
Un
Chlorine
Contact
Tanks
Digesters)
^a ^
Thickeners^
Primary
Settling
Tcnks
Instailation
Aeration
Tank
Final
Settling
Tanks
Influent
ADM.
Aeration
Tank
Figure !.
Schematic Layout of Port Richmond WPCP
-------
Bypass eb*aml c«c««e
7
UTI Unit
u _
u.v. Pi let
Foeility
UV.
From
w@f«r
Piont Layout
206
-------
TABLE 1
SYSTEMS INSTALLED OR PLANNED
1
Contact
100
G36T6VH
40
-
5.0 cm
0.76-4.5
0.2-1.2
PWS(2)
378
100
2
Contact
100
G36T6VH
40
-
1.25 cm
0.76-4.5
0.2-1.2
PWS (2)
76
20
3
Non-Contact
72
G64T5
80
32
(1)
0.76-4.5
0.2-1.2
UTI (3)
(4)
(4)
No. Lamps
Type
Power (Watts)
Teflon Tubes (No.)
Spacing
Flow Range (ML/d)
(mgd)
Manufacturer
Void Volume (L)
(gal.)
(1) Lamps are 13 cm center to center.
Teflon tubes are 6.4 cm diameter, 13 cm center to center
(2) Pure Water Systems, Inc.
(3) Ultraviolet Technology, Inc.
(4) Liquid holding capacity is 9.8 liters/tube; total
of 314 liters.
207
-------
Pyturs UTI
•Pewsr Pan«l
Tank
to
LaJUu
Unit
I
§ Butterfly Valv«* — §
Int Tattfes
-icm =
. _
Unita
• Eft. Tonkt-
Pat* I
Unit
2
P§a«r Supply
Lighting Supply
Figure 3.
U.V. Pilot Facility
208
-------
continuous 24-hr operation and data collection by early December, 1981.
Laboratory facilities at the Port Richmond WPCP are used to support the
sampling and monitoring program. Bacterial density measurements (total and
fecal coliform, total plate count, fecal strep), suspended solids and
turbidity analyses are performed at Port Richmond by HydroQual personnel.
Other tests including UV absorbance, nitrogen (organic ammonia, nitrate,
nitrite), total organic carbon (TOC), chemical oxygen demand (COD), and pH
are performed at HydroQual1 s laboratory (General Testing Corporation),
located in Hackensack, New Jersey.
EXPERIMENTAL PROGRAM
There is little in the way of actual performance data to report. What
can be presented, however, is the basic model about which the experimental
program is being constructed and two tasks which have been or are nearly
completed which address two major elements of the proposed model.
Additionally, water quality data which has been collected to date will be
presented, indicating the characteristics of the secondary effluent
discharged by the Port Richmond plant.
The modeling framework and calculations will be described only in
general terms as they apply to Port Richmond. The reader is cautioned that
these discussions are preliminary and subject to modification as the study
progresses. The basis of the model and other calculations will be
presented in more detail in subsequent conference presentations and in the
final report.
Proposed Model
The proposed model to describe system performance for the disinfection
of wastewater is expressed as follows:
= ex p [ (•
ux
2E
4kE
- (1 +
1/2
) }]
(1)
where L = residual bacterial density (colonies/100 ml)
LQ = initial bacterial density (colonies/100 ml)
u = fluid velocity (cm/sec)
x = distance (forward direction) traveled during exposure (cm)
2
E = dispersion coefficient in forward direction (cm /sec)
k = rate coefficient (see" )
209
-------
The fluid velocity, u, is computed as
x(Q/Vy)
In the case of the PWS units at Port Richmond, x is the longitudinal
dimension of the lamp battery (cm). The term Vy is the void volume of the
lamp battery (cm ) and Q is the wastewater flow rate (cm /sec).
Fluid flow is assumed to be completely mixed in the plane perpendicular
to the direction of flow. The term EX describes dispersion only in the
forward direction. The rate coefficient (k) is a function of the intensity
of ultraviolet light, i.e. the rate at which energy is being delivered to
the wastewater. Intensity, in turn, is a function of the UV output of the
lamps, the placement (spacing) of the lamps and the absorptive properties
of the wastewater „
Hydraulic Evaluations
A series of tests has been performed to define the time distribution
and flow characteristics through each of the submerged systems at Port
Richmond. Prior to this each unit was evaluated to determine if there
existed an acceptable approach condition, i.e. no shortcircuiting or
significant velocity gradients across the front plane of the lamp battery.
Significant gradients were found, particularly on the widely spaced unit.
This was due (on both units) to the position of the inlet pipe and the
relatively small size of the influent tank. Sufficient time and volume
were not available to dissipate and equalize the velocity before entering
the lamp battery. An overflow weir and stilling wall were subsequently
installed in each influent tank to correct this problem. Further tests and
observations indicated a good flow distribution across the plane of the
lamp battery, with no evidence of shortcircuiting.
Mixing within each UV unit was evaluated using real time measurements
of conductivity. Salt (NaCl) was used as the tracer. The normal procedure
for developing detention time curves, i.e., taking discrete samples after
an instantaneous tracer injection, was not possible given the system design
at Port Richmond. The theoretical detention times (V /Q) were 2 to 30
seconds, leaving little time to practicably inject a sufficient quantity of
tracer and collect an adequate number of samples to construct the trace.
Additionally, the design essentially simulates an open channel with the
lamp battery inserted across the width of the channel, making it very
difficult to take representative discrete samples.
A new procedure was developed, as shown schematically on Figure 4. A
concentrated salt solution was injected at a constant rate at a selected
location immediately in front (approximately 3.8 cm in front of lamp plane)
of the lamp battery. A conductivity probe was used to search for the point
of maximum concentration on the exit side of the lamp battery. Note that
the probe was also used to scan the entire exit plane in order to define
the cross-section and location of the'plume as it exited the lamp battery.
210
-------
event signal
Wheatstone
Bridge
O Lamp Battery
O
O
OOOOOOOO
Salt
Solution
Oscillograph
Conductivity probe
Figure 4.
Experimental Setup for Retention Studies
211
-------
Once the probe was situated and fixed at the center of the plune, a
steady state condition was allowed to develop at fixed wastewater and salt
solution flow rates. The high frequency output from the conductivity meter
was amplified and continuously recorded using an oscillograph. Continuous
permanent tracings are made by a light beam onto light sensitive recording
paper advancing at a rate of 0.64 cm/second. The recorder/meter was first
calibrated by measuring the conductivity of known salt solutions.
Once steady state was indicated by the recorder, the salt solution pump
was shut-off. This event was automatically signalled to the oscillograph
and recorded. The die-away of salt was then monitored by the fixed probe
and continuously recorded. Readings (conductivity) were then taken off the
trace, converted to salt concentration (mg/1) and transposed to a plot of
concentration against time.
Figure 5 graphically presents the method used to analyze the resulting
trace. The upper plot shows concentration against time. The derivative of
this curve is taken by plotting the slope (dc/dt) of tangents drawn at
several locations along the curve. This is shown on the lower plot and
resembles the typical curve derived from an impulse release.
Assuming the flow through the lamp battery, as shown on Figure 6, is
completely mixed in the y and z plane and disperses only in the x
direction, the response to an impulse release may be described as
(2)
where
W
A
c
t
defined
= mass input (mg/sec)
= cross sectional area (cm )
= concentration (mg/1)
= time (seconds)
tQ, where t is the theoretical detention time, a constant, y, is
(3)
This constant is then substituted into Equation (2),
dc /_~ -(x-ut)'
dt = y ^o GXP C"^ET~
(4)
212
-------
Salt
Cone.
c
dc t
dt
Trace after shutoff
dc/dt
Derivative of Trace
Time
Figure 5.
Analysis of Hydraulics Data
213
-------
completely mii@d in y, i
rsion in %
Figure 6.
ing Assymptions
214
-------
Knowing the values of dc/dt, y, tQ , x, u, and t, the dispersion
coeff:"
data .
coefficient EX can be estimated by a trial and error procedure to fit the
An example of the procedure is presented on Figure 7. These data are
from a test run on Unit No. 2 (closely spaced lamp unit). The point of
injection is shown on the upper display. The lower display presents the
slope calculations as data xioints. The smooth line is a solution of
Equation (2) at an EX = 10 cm /sec.
The results of several runs on each unit, made at differing flows and
injection point locations, indicated an average E of 1.5 cm /sec for Unit
No. 1 (widely spaced unit) and 15 cm /sec for Unit No. 2 (closely spaced
unit). This implies that the flow characteristics of Unit 1 correspond
more closely to a plug flow condition (E approaches zero) than does Unit
2. It should be noted, however, that both units can be considered to
closely simulate a plug flow condition relative to the opposing condition
of complete mix, when E approaches infinity.
Figure 8 presents a series of solutions to Equation (1) which
demonstrates the sensitivity to E . The log of the survival ratio (L/L0)
is plotted as a function of the rate coefficient (k) for various values of
E . This is shown for both systems installed at Port Richmond at a flow of
1.9 ML/d (0.5 mgd) . It is evident that significant deviation from the
idealized flow (plug flow) condition can result in multilog increases in
the survival ratio. This emphasizes the importance which must be placed
upon the hydraulic characteristics of a system design.
ULTRAVIOLET INTENSITY
The rate of disinfection is directly related to the intensity of
ultraviolet light, i.e. the rate at which energy is delivered to the
wastewater medium by the UV source. Current system designs, which
generally involve lamps or lamp bundles immersed or surrounding the
receiving medium, have precluded any practical means to directly measure
the true intensity at any point within a system.
A mathematical model has been developed as an element of this study
which calculates the intensity at any point within a UV system and which
can estimate the average intensity emitted by a specific unit. The
calculations are based on the point source summation method described by
Jacob and Dranoff (1) and recently applied to disinfection systems by
Johnson and Quails (2) . The mathematical techniques rely on the basic
physical properties of the ultraviolet lamps, the configuration of the
multilamp chambers, and the properties of the aqueous medium.
UV energy emitted by a lamp is attenuated as the distance from the
energy source increases. This attenuation occurs via two mechanisms:
dissipation and absorption. Dissipation simply describes the dilution of
the energy as distance from the source increases. The surface area over
which the output of energy is projected increases with increasing distance,
215
-------
c
im/l)
0-
dt
y
t i I
V/Q« 6.9 »tc*
cm2/sec
Troctr Anolysss
216
-------
Compltt*
mixing
X (cm*/ B«C)
100
Q= 1.9 ML/d (0.5 mgd )
S= 1.25cm
100 lamps
Q= 1.9 ML/d (O.Smgd)
S= 5.0cm
100 lamps
5
Figure 8.
Sensitivity to Dispersion Coefficient
217
-------
thus there are fewer photons striking each unit of surface area. This
dissipation can be calculated by surrounding an energy source by a sphere
of radius R:
(5)
2
where I = intensity at distance R (cm) in watts/cm
S = output of UV energy source in watts
The second attenuation mechanism relates to the absorptive properties
of the medium through which the energy is transmitted. This is described
by Beer' s Law:
I « IQ e-°R (6)
2
where I = intensity at a given point (watts/cm )
a = absorbance coefficient (cm )
R = distance from the point of I (cm)
Combining equations (5) and (6) yields
T - S.. e-aR
i - 2 e
4irR
which describes the intensity at a given distance from a point source of
energy.
The tubular germicidal bulb is treated in this calculation as a series
of point sources. The intensity at a specific point is then the sum of the
intensities from the individual point sources:
n=N S/N 2 2 1/2
Hr.z) = I —-TL7- exp[-a(r + z^) ] (7)
n = 1 4ir(r +z^)
where z = L (•—.)
N-i
N is the number of point sources into which the line source is divided and
r and z describe the coordinates of the "receiver" at which the intensity
is being computed (R = r + z ). This is shown schematically on Figure 9.
To calculate intensity at point (r,z), Equation (7) must be applied N times
and all solutions summed.
An assumption inherent to Equation (7) is that the receiver located at
(r,z) is spherical and infinitely small. Thus, energy emitted from any
218
-------
Lamp
Division of
Lamp into
point sources
JV
J^
I
Y
Y
Y
— Z* L
Recevier location
Sample Lamp element
— Z» 0
Figure 9.
Lamp geometry for point source approximation
219
-------
point source element of the lamp will strike the receiver normal to its
surface .
The computer model uses Equation (7) as the basic element to compute
the intensity in a specific system. The model is capable of accounting for
. absorption of energy as it passes through various elements such as
wastewater, quartz sleeves, teflon, air and neighboring lamps
. any system lamp configuration including assymetrical arrays, as long as
the lamps are parallel to one another
. any lamp rating for UV output
. any lamp battery size (no. of lamps) and variation in output
Figure 10 displays preliminary workups for the two systems presently
installed at Port Richmond. Average computed intensity is plotted for each
system as a function of the UV absorbance coefficient (at 253-7 nm) of the
wastewater. Tne lamps in this instance have a rated output of 1W each at
the 253.7 nm wavelength. The reader is cautioned that these analyses are
preliminary at this point. More work is anticipated to refine the
calculation techniques and to experimentally confirm and/or modify the key
parameters which comprise the model.
The utility of the mathematical modeling technique is that it allows an
analysis of a system's sensitivity to the variables which impact its
design. As an example, a key parameter in certain system designs is the
spacing of the lamps. Spacing will affect the average intensity within the
system, the detention time and may influence the hydraulic characteristics
of a unit. The two units at Port Richmond differ in spacing by a factor of
four (1,25 cm and 5 cm); as Figure 10 shows, the model indicates that the
ratio of Igvg for Unit^Z to I for Unit 1 increases from approximately
2.3 at an a of 0.2 cm to approximately 4.1 at an a of 0.8 cm" . This
loss of efficiency in the closely spaced unit at the lower water
absorbances is attributed to the "shadowing" effect of neighboring lamps.
The lower the absorptive property of the wastewater, the further the UV can
penetrate. If the lamp spacing is such that this energy hits a neighboring
lamp, it will be absorbed by that lamp.
Figure 11 displays the effect of this phenomenon. The percent
reduction shown on the ordinate represents the loss of energy to
neighboring bulbs. This is shown as a function of bulb spacing for varying
wastewater absorbance coefficients. This analysis shows that the spacing
of bulbs is clearly a design parameter which will be dictated by the
quality of the water to be treated. Other design relationships are being
developed with the model, with field verification at Port Richmond.
220
-------
CS»
E
o
o
H
Rated Output (253.7nm)
14 W
S2" 1.25cm
I
0.2 0.4 0.6 0.8
Absorbance Coefficient ccw (cm"1)
Figure 10.
Computed Intensity
i.o
221
-------
o
H
c
o
o
3
•o
a?
ocw (cm-1)
234
Bulb Spacing (cm)
Figure 11.
Intensify Reduction vs. Bulb Spacing
222
-------
WATER QUALITY
Table 2 presents a summary of analyses on the influent to the UV
systems (secondary effluent). The data are limited, representing 14 days
of sampling (two to four samples/day) in December, 1981, and January, 1982.
The COD has been variable, with a mean of 45.4 mg/1 (total). The suspended
solids have averaged 20 mg/1, ranging as high as approximately 60 mg/1.
Two methods are employed to measure the absorbance (at 253.7 nm) of the
wastewater. In both cases a Perkin-Elmer Model 552 Double-Beam Scanning
UV/Visible Spectrophotometer is utilized. The first method is to simply
measure the absorbance of a direct beam through a 1 cm cell. This is
designated as the "direct" UV absorbance. The second method incorporates
an integrating sphere attachment to the spectrophotometer. This accounts
for light that may be scattered (and not measured by the direct beam
method) and is not absorbed. Thus the "sphere" absorbance more closely
corresponds to the true absorbance of the samples. The data on Table 2
indicate that the true absorbance is not significantly affected by
suspended/colloidal solids. Rather the energy is scattered by these
particles and remains available for disinfection purposes. Conversely
there appears to be little penetration or absorbance of the energy by the
particles, precluding the inactivation of organisms occluded by such
material.
SCHEDULED TASKS
A major fraction of the experimental program lies ahead. Specific
tasks anticipated for the project include the following:
evaluation of the kinetics associated with disinfection
confirm UV intensity parameters
verify the proposed disinfection model by operation of the units under
equivalent performance conditions
install and similarly evaluate the teflon system
evaluate the impact of photoreactivation under warm and cold
temperature conditions
monitor each system for cleaning, C&M needs, reliability
CLOSING
Ultraviolet light disinfection is a viable, cost effective process
which is quickly emerging as an alternative for wastewater disinfection.
The Port Richmond project will seek to develop needed information in the
area of design methods and the definition of critical process performance
parameters. Although we were unable to present a substantive store of data
223
-------
TABLE 2
INFLUENT ANALYSIS
14 Days
2-4 samples/day
mean
U.V. abs. Direct ( cm~1 ) , aw
U.V. abs. Sphere (cm~1)s aw
CODT (mg/1) 45.4 15.0
CODp (mg/1) 34.8 16.4
TOCT (mg/1) 16.4 6.0
TOCp (mg/1) 14.9 6.0
SS (mg/1) 20.0 16.5
Turbidity (NTU) 5.5 4.9
0.400 0.114
0.319 0.066
0.309 0.068
0.297 0.068
224
-------
in this paper, the participants will make an effort to disseminate, via
conference papers, the results of the study as it progresses.
ACKNOWLEDGEMENTS
The authors wish to express their thanks to Mr. Joseph McAllister,
Plant Superintendent, and the Port Richmond plant operators for their help
and interest in the project. We would also like to acknowledge the support
and contributions being made by Mr. William Pressman, Director of R & D,
New York City DEP; Mr. Gerard Cox, Engineer NYC DEP; Mr. Wilfred Dunne,
Field Technician, HydroQual; and Ms. Maureen Casey, Engineer, HydroQual.
REFERENCES
(1) Jacob, Solomon M. and Joshua S. Dranoff "Light Intensity Profiles in a
Perfectly Mixed Photoreactor, Journal, AIChE, Vol. 16, No. 3, pg. 359.
(2) Johnson, J. Donald and Robert G. Quails, "Ultraviolet Disinfection of a
Secondary Effluent" Draft Report to USEPA, Municipal Environmental
Research Laboratory, Cincinnati, Ohio, 1981.
(3) Scheible, 0. Karl and Carlene D. Bassell "Ultraviolet Disinfection of a
Secondary Wastewater Treatment Plant Effluent" USEPA, Municipal
Environmental Research Laboratory EPA-600/2-81-152; National Technical
Information Service, PB-81-242-125, September, 1981.
225
-------
COMPARISON OF ANALYTICAL METHODS FOR RESIDUAL OZONE
Gilbert Gordon and Joyce Grunwell
Department of Chemistry, Miami University
Oxford, Ohio 45056
ABSTRACT
Seven analytical methods for the determination of residual ozone in water
and waste water have been compared by measuring the decomposition of ozone in
water and waste water. This kinetic technique minimizes sampling errors and
allows a direct comparison of methods under conditions of rapidly changing
ozone concentration. Changes in the ozone-reductant reaction caused differ-
ences in ozone decay curves. Conditions which reduce ozone decay prior to
the ozone-reductant reaction reduced differences among methods.
The analytical methods are compared on the ease of calibration of the
reagent solutions, the stability of the reagent solutions, and the stability
of the titer of the ozonated reagent solution.
INTRODUCTION
Over the last eighty years, chlorine has been widely used for disinfec-
tion of municipal and industrial waste waters. Recent concerns, however,
over the toxic effects of chlorinated organic by-products produced during
chlorination of potable water and waste water have renewed interest in
ozone in water treatment. Ozone acts as an oxidant to remove taste, color,
odor, and organic matter from water as well as serving as an effective
disinfectant. The United States Environmental Protection Agency requires of
any disinfectant that its residual be measured accurately and conveniently.
In the case of ozone, the residual may vary from 0.05 mg/L to 30 mg/L depend-
ing on reaction time, sample contamination and dosage level. Residual ozone
levels below 5 mg/L are of most interest in water treatment (15).
With a standard reduction potential of 2.07V in acid solution and 1.24V
in basic solution, ozone will react with most oxidizable substances.
Ozone is usually generated by passing a stream of dry oxygen through an
electrical discharge which converts 2-5% of the oxygen to ozone. Therefore,
ozone, the species of interest, is only a small fraction of the gas mixture.
It would be ideal if methods for determining residual ozone could be
verified by the analysis of weighed samples of pure ozone. This is impos-
sible, however, due to the instability of pure ozone, the low solubility of
ozone in water, the high volatility of ozone, and the rapid decomposition of
ozone in water.
Most analytical methods for the determination of aqueous ozone take ad-
vantage of the property of ozone as a strong oxidizing agent. Some of the
most popular reductants are iodide ion (6), arsenic(III) (10), and indigo
22.6
-------
blue (2). An amperometric membrane electrode measures ozone in solution (13).
The objective of this paper is to evaluate and compare these analytical
methods and explain any differences.
The oxidation of iodide ion to iodine by ozone with subsequent titration
of the iodine formed is the classical method for the determination of residual
ozone (6). The reaction of ozone with iodide ion is described by
03 + I~ ->• 02 + I0~ fast
I0~ + H20 -»- HIO + OH~ fast
HIO + 2I~ -»• I3" + OH~ slow
3 HIO + 30H~ -»• I03~ + 2I~ + 3 H20 very slow
Upon acidification the species hypoiodite ion (I0~), hypoiodous acid (HIO),
and iodate ion (I03~) are all converted to triiodide ion (I3~) so that the
overall process is ideally
03 + 3I~ + H20 -»• I3~ + 02 + 20H~
and theoretically one molecule of ozone liberates one molecule of titratable
iodine. Thermodynamically, the ozone oxidation of iodide ion and iodine to
form iodate ion is favored at high pH.
303 + !-->• 302 + I03~
503 + I2 + H20 ->-502 + 2I03~ + 2H+
Furthermore, at a pH above 9, iodine is unstable, readily undergoing dispro-
portionation.
12 + 20H~ -v 10" + I~ + H20
310" -»• 21" + I03~
Regardless of whether iodate ion is formed by direct oxidation or dispropor-
tionation of iodine, for every three moles of ozone absorbed in the pH region
above 9, one mole of iodate ion should be formed.
In the standard iodometric method for analysis of ozone in water, ozone
is purged into potassium iodide solution and after acidification with sulfuric
acid the iodine is titrated with standard sodium thiosulfate (12). The ozone:
iodine stoichiometry has been extensively studied and found to range from 0.65
to 1.5 (3,4,5,8,14). The factors affecting the stoichiometry include pH,
buffer composition and concentration, iodide ion concentration, and sampling
techniques. Modifications in the iodine determination include changes in
endpoint detection, pH, and back titration techniques.
227
-------
Theoretically, both the pH during the initial ozone-iodide ion reaction
and the pH during the iodine determination can alter the ozone:iodine
stoichiometry. In acid, the ozone:iodine ratio could decrease due to
403 + 10 HI -" 512 + H202 + 4H20 + 302
Hydrogen peroxide could oxidize iodide ion also leading to excess iodine,
H202 + 21" + 2H+ -> 12 + 2H20
Air oxidation of iodide ion in acid also leads to a decrease in ozone:
iodine ratio
0-7 + 41" + 4H+ -> 212 + 2H20
Any errors in the assumption that in acid iodate ion, hypoiodous acid,
and hypoiodite ion are quantitatively reconverted back to iodine would lead
to an increase in the ozone:iodine ratio. In base, iodate ion formation and
hypoiodite ion formation lead to low iodine titers« Again, if the iodine
determination is carried out under conditions where iodate ion and hypoiodite
ion are not quantitatively reconverted back to iodine, the ozone:iodine ratio
is high,.
In summary,, as the pH decreases for the ozone oxidation of iodide ion,
the quantity of iodine should increase due to less iodate ion formation,
hydrogen peroxide formation, and air oxidation of iodide ion. As the pH for
the iodine determination decreases, the iodine titer should increase due to
reconversion of iodate ion to iodine and air oxidation of iodide ion. In the
lodometric method ozone is reacted with iodide ion in buffers of pH 3.5 to
9.0. A known excess of sodium thiosulfate is added, the pH adjusted to 2
with sulfuric acid, and the excess thiosulfate ion titrated with standard
iodine (12).
In the amperometric method, ozone oxidizes iodide ion at pH 405 in the
presence of a known excess of sodium thiosulfate, phenylarsineoxide (PAO) or
inorganic As(III). Without acidification these excess reagents are then
titrated rath standard iodine to an amperometric endpoint (12).
In the As(III) back titration method, ozone oxidizes iodide ion at pH 6.8
in the presence of a known excess of inorganic As(III). Without pH change
the excess As(III) is back titrated with standard iodine (14). The DPD
Method is an iodometric method carried out in phosphate buffer pH 6,4 (7).
Ozone oxidizes iodide ion to iodine which then oxidizes N,N-diethyl-p-phenyl-
enediamine cation (DPD) to a pink Wurster cation. The Wurster cation is
quantitated colorimetrically. In the direct oxidation of As(III), ozone
reacts with either inorganic As(III) or PAO at pH 4-7, the pH is adjusted to
6.5-7 and the excess As(III) species is back titrated with standard iodine
(10). The Indigo method is performed at pH 2 (2). Ozone adds across the
carbon-carbon double bond of a sulfonated indigo dye and decolorizes it.
228
-------
The change in absorbance is determined spectrophotometrically. The Delta
electrode (13) and the UV method (2), which measures ozone directly by its UV
absorption at 259 nm, involve no reagents and no pH restrictions.
The kinetic and mechanistic description of the decomposition of aqueous
ozone has been extensively investigated but no detailed mechanism is generally
accepted. Results indicate that decay leads to free radicals. The half -life
of dissolved ozone is readily affected by pH, UV light, concentration of
ozone, and concentration of radical scavengers (1,9,11). The experimental
results published prior to and during the 1950 's fit either a one-term or a
two-term rate law (9). Recent work of Hoigne'' (11) also supports a two-term
rate law.
A kinetic technique was developed for producing ozone solutions of known
concentration in the 24-0 mg/L range. By means of this kinetic technique, ad-
vantage is taken of the self -decomposition of residual ozone. A steady state
solution of ozone is prepared and allowed to decompose. At known time
intervals during the decomposition process, the ozone level of the solution
is determined by two or more different analytical methods.
The resulting time-concentration profile for each analytical method is
graphed and is fitted to the generalized rate law
dt
k2 [03]2
using FIT80. FIT80 is a computer program based on the method of Gauss which
allows the simultaneous least squares fitting of first and second order
parallel reactions. This rate law is a mathematical model which describes
time-concentration curves for ozone decay and does not relate directly to a
specific mechanism. The calculated rate constants are apparent rate constants
and not true rate constants. They are calculated and compared for each
kinetic run and are used for method comparisons within each run and not
between different runs. The kinetic parameters calculated for each method
should be identical if each method gives the same result.
Even though a mechanism is not necessary for the application of the
kinetic technique since comparisons among calculated kinetic parameters
indicate discrepancies between methods, a general mechanistic scheme aids
understanding of potential method differences. Based on a set of clever
experiments, Hoigne' (11) has recently proposed the mechanism shown below:
03 + OH~ ->- 02*~ + H02* k0H I70
+H+
°2'~ + °3 "* OH* + 2 02 k02*~ ~ 1.6 x 109 M~]
OH* + scavenger ->• products
229
-------
_
OH* + 03 -»• 02 + 02«~ kOH- ~ 4.7 x 108 M
02»- + 03 ->- OH- + 202
2 H02' ->• H202 + 02
H02» H
H202 -
H02~ H
h 02°
1- OH~ -
h 03 >
-*- HO 2"
^ H20 +
OH"
+ 02
HO 2
kHO ~ ~ 5 x 106 H~1sec~1
Ozone decay is initiated by hydroxide ion attack on ozone to form the
superoxide radical anion (02»~) and hydroperoxyl radical (H02»)« Then in an
almost diffusion controlled reaction, superoxide radical anion reacts with a
second ozone molecule to form the hydroxyl radical (OH« ) „ This hydroxyl
radical either can react with any radical scavenger present or can react with
an ozone molecule in an almost diffusion controlled reaction to generate
another superoxide radical anion which in turn reacts with ozone to generate
another hydroxyl radical.
The hydroperoxyl radical, a by-product of ozone decay, can dimerize to
hydrogen peroxide or can react with superoxide radical anion to form hydro-
peroxyl anion. Hydrogen peroxide reacts with ozone slowly. Hydroperoxyl
anion, however,, can catalyze ozone decay by attack on ozone to form the
hydroxyl radical in a very fast reaction. A hydrogen peroxide concentration
in excess of 10~' M will make the reaction of hydroperoxyl anion with ozone
as important as the reaction of hydroxide ion with ozone. The concentration
of hydrogen peroxide found as a reaction product increases with a decrease
in pH.
According to this mechanism, the lifetime of ozone in aqueous solution
depends on added solutes or Impurities. The hydroxyl radical, which forms
upon ozone decomposition, is a chain carrier for further ozone decomposition
and any solute or impurity which scavenges this radical will retard ozone
decomposition.
A simplified mechanism consistent with Hoigne" s model is shown below.
03 + OH~ ^ [03°OH~]
[03«OH~] ->- 02»~~, H02-3 OH', H202 + products
[03*OH~] + 03 ->- products
230
-------
Initially ozone complexes with hydroxide ion and this reactive inter-
mediate can undergo either a first order electron transfer reaction or a
second order reaction with ozone. The first order electron transfer reaction
could lead to the formation the superoxide radical anion, the hydroperoxyl
radical and the hydroxyl radical which in turn either act as chain propagators
and lead to further ozone decay or react with radical scavengers to form other
products. The second order reaction of ozone with the ozone-hydroxide inter-
mediate leads directly to products without formation of discrete short-lived
radical intermediates. This simple model is consistent with the two term rate
law for parallel first and second order reactions used by FIT80 to describe
the time-concentration curves for the kinetic technique.
RESULTS AND DISCUSSION
The use of PAO as a direct reductant for ozone is based on the assumption
that PAO (As(III)) is exclusively oxidized by ozone to phenylarsonic acid
(As(V)) and that oxidation of the arsenic-carbon bond and the carbon-carbon
double bonds are negligible. Since ozone has been used to digest organic
arsenicals and since any PAO decomposition would lower residual ozone measure-
ments, inorganic As(III) and PAO were compared by the kinetic technique. In
order to attribute any inconsistency to PAO decomposition, both reductants
were ozonized in acetate buffer pH 4.5. The pH was adjusted to 7 with 0.5 M
sodium bicarbonate solution and the excess reductant back titrated with iodine
to an amperometric endpoint.
Examination of the decay curves (Fig. l)(or the rate constants) calculated
by the FIT80 program reveals the similarity in behavior of PAO and inorganic
As(III) as reductants for ozone. Although the PAO seems to be consistently
high, the deviations are small enough to be within experimental error. There-
fore, it is concluded that no noticeable decomposition of PAO occurs. Fig. 1
also illustrates the recurring and worrisome observation that in a kinetic
comparison, one or two data points may fall significantly off the decay curve
calculated by the FIT80 program. This observation cannot be neglected.
As shown in Table I, PAO and sodium thiosulfate are equivalent when used
in the amperometric method. Here both function as reductants for iodine and
are not directly involved in an ozone reaction. When these amperometric
method results are compared with the results from As(III) direct oxidation,
however, the ozone concentration determined by As(III) direct oxidation is
almost nine percent low. Thus, when the ozone reductant changes from iodide
ion to arsenic(III), clearly, a difference occurs.
Table 1. Amperometric Comparison of PAO and Sodium
Thiosulfate vs_ the As (III) Direct Oxidation
Amperometric As(III) Direct
PAO Na2S203 pH 4 - 4.5
mg/L O-^ mg/L 0^ mg/L 0^
14.5 ± 0.3 14.8 ± 0.3 13.5 ± 0.1
231
-------
The DPD and the arsenic(III) back titration methods differ significantly
in two ways: first, in the excess reagent present when ozone oxidizes
iodide ion to iodine and second, in the iodine quantification. Since the DPD
calibration curve is based on standard iodine, the iodine quantification
should not be responsible for differences between the methods. In the
arsenic(III) back titration, ozone could be reduced by iodide ion or by
arsenic(III) although reduction by arsenic(III) occurs at a much slower rate
than reduction by iodide ion. However, the formation of arsenic(V) by the
direct oxidation by ozone or by the indirect oxidation through iodine, still
maintains the stoichiometry of one ozone per arsenic(V). In the DPD method,
ozone could oxidize iodide ion and oxidize DPD or the Wurster cation to the
diimine cation. Any direct attack of ozone on the indicator would lead to
low results. The decay curves in purified water determined by the DPD and
the arsenic(lll) back titration method (Figo 2) show considerable scatter
and in fact are good examples to illustrate the importance of the kinetic
technique. If conclusions had to be based on single point comparisons, then
the first, second, and third data point sets would lead to three different
conclusions. Similar ozone decay curves in purified water spiked with 5 mg/L
hydrogen peroxide prior to ozonation and in tap water show that scatter
decreases as the radical scavenger concentration increases. Our results
confirm the equivalence of the DPD method and the arsenic(III) back titration
method.
The DPD method is not equivalent to the Indigo method as shown by the
decay curves for ozone in purified water (Fig. 3). The DPD points are
scattered and the ozone concentrations are low compared to the smooth Indigo
plot. The critical differences between the methods are ozone reduction by a
carbon-carbon double bond at pH 2 in the Indigo method and ozone reduction by
iodide ion at pH 6.4 for the DPD method.
lodate ion formation could be responsible for the low DPD titer. lodate
ion formation is also indicated in comparisons made using the lodometric
method at pH 3,5, 5S 7, 9 and the arsenic(III) direct oxidation method in
bicarbonate ion solution at pH 7. Least squares analyses of plots of the
lodometric method results on the x-a.xis vs the arsenic(III) direct
oxidation results on the y-axis are shown~~in Table II,
Table 2,
Comparison of lodometric and Arsenic(III) Direct
Oxidation Methods.
3,5
5,0
7.0
9,0
Y-intercept
0.95
0.88
0.95
0.74
slope
1.00
0.89
0.70
0.81
correlation
0.9998
0.9979
0.9947
0.9986
The positive Y-intercept shows that the arsenic(III) direct oxidation
method consistently gives low results. The slope shows the trend that the
iodometric method gives lower results as the pH increases. This error is
232
-------
-------
consistent with iodate ion formation and with hydrogen peroxide formation.
It would j-iaply, however s that iodate ion is not quantitatively reconverted to
iodine under the acidic titration conditions.
Iodate ion formation was measured for the ozone oxidation of 2% iodide
ion solutions in 0.1 M phosphate buffers at pH 7.3, 6.8, 5.3, and 2.2. Immedi-
ately after the ozone injection, standard arsenlc(III) was added and two
aliquots were removed: one for iodate ion analysis and the other for ozone
analysis by the back titration of excess arsenic(III). The back titration
aliquots at pH 5.3 and 2.2 were brought to neutrality with sodium hydroxide.
All iodate ion aliquots were immediately made strongly basic for differential
pulse polarographic (DPP) analysis (16), The DPP is capable of detecting an
iodate ion concentration as low as 1 x 10™" M,, This corresponds to 0.001 mg/L
ozone taking into account the 3:1 ozoneiiodate ion stoichiometry. As expect-
ed, iodate ion formation tends to increase with ozone concentration and with
pH. The largest iodate ion concentration was found at pH 7.3 and correspond to
0.265 mg/L ozone (Table III). No iodate ion was detected at pH 2. The iodate
ion concentrations are too low to explain the observed differences in the
ozone decay curves traced by the arsenic(III) back titrations and by the
Indigo method (Fig. 4). When the above experiment was repeated for pH 7.0 and
pH 2.0 with arsenic(III) present in the iodide ion solution during the ozone
addition, iodide ion concentrations corresponding to less than 0.024 mg/L
ozone were found at pH 7.,0 and no iodate ion was detected for pH 2.0.
Table 3, Iodate Ion Formation (in ozone equivalents) with Time(sec).
pH 2,,, 2 PH 5.3 pH 6.8 pH 7.3
JL££. mg/L _0_3 .£££ EJI/!i_£3. sec ing/L Oo sec mg/L Oo
None 90 0.176 74 0.147 26 0.173
280 Ocl43 240 0,107 197 0.234
510 None 500 0.091 420 0.100
755 0.058 745 0,100 685 00265
1035 None 985 0.058 900 0.078
1565 None 1555 None 1485 0.109
Hydrogen peroxide can be formed by the ozone-iodide ion redox reaction
and by the decomposition of oz.one in water. Its concentration, measured by
DPF after ina hour ozonation of 0.07 M phosphate buffers, varied with pH and
with exposure to UV light (Ace Hanovia high pressure, mercury vapor lamp).
The detection limit of 1 x 10~7 M corresponds to 0.0034 mg/L. In the absence
of UV light, the hydrogen peroxide level increases with pH and in the presence
of UV light the hydrogen peroxide level decreases with an increase in pH
(Table IV). The hydrogen peroxide concentration varies dramatically under
234
-------
20 -I
18
16
o>
a
o
N
0 12
10
• DPD
• INDIGO
500
1000
2000
1500
Time (sec)
Figure 3. DPD vs. Indigo in Purified Water.
16-
14-
12'
0)
g 10-
N
O
•^ 8"
60
s
6-
4-
2.
• INDIGO
A
* A PH 6,80
A n PH 5.26
° • pH 2,12
A A
D
• •
A
ft
U *
A
n n
2500
Figure 4.
200400 600 800 1000 1200
Time (sec)
Indigo vs. As(III) Back Titration in Purified
Water at pH 7.21.
235
-------
Table 4, Hydrogen Peroxide Formation with Ozone Decay
pH [H202] x 1C)7 M [H202] x 107 M
with UV
2.1
7,1
12.0
7.0
11.8
7.8
11.8
4.8
7.1
10.0
2.9
4.0
9.5
8.7
77.9
4.7
seemingly similar conditions and all the hydrogen peroxide titers are too
small to account additively for method/pH differences. The hydrogen peroxide
probably catalyze ozone decomposition since at 10~7 M hydrogen peroxide the
hydroperoxyl anion is as important as hydroxide ion as a catalyst (11).
The scatter observed in kinetic comparisons of ozone decay in purified
water is reduced by working with ozone in buffered purified water. Because
this trend shows up with all methods, the significant difference must lie in
the ozone solution itself. A comparison of the Indigo method, the arsenic(III)
back titration and the arsenic(III) direct oxidation methods on ozone decay
in purified water buffered to pH 6.7 with perchloric acid-phosphate mixtures
shows the three methods to be alike.
When the Indigo method, the direct UV measurement, and the arsenic(III)
direct oxidation method are compared on acidified ozone solutions undergoing
minimal decay, they are also equivalent.
The ozone decay curves by the Indigo method, by direct UV measurement,
and by arsenic(III) direct oxidation method for ozone in purified water
buffered to pH 7.7 with a KH2P04-NaOH mixture, however, show the Indigo
method to give 10-15 percent higher residual ozone concentrations than the
arsenic(IH) direct oxidation method.
The kinetic technique has revealed differences among methods. These
variations could be caused by:
236
-------
1) the reaction of ozone or ozone decay products with the oxidized
indicator (e.g. iodate ion formation).
2) the reaction of ozone decay products with the reductant or
indicator.
3) the further decomposition of ozone prior to reaction with the
reductant. This could be caused by the pH of the reductant
solutions, solutes in the reductant solution, or by a relatively
slow reaction between ozone and reductant.
The effect of hydrogen peroxide on the ozone titer was examined. Oxford
tap water was ozonated, acidified for stabilization and analyzed for residual
ozone by UV analysis, the Indigo method, the direct oxidation of As(III), and
the back titration of As(III). Then, the residual ozone titer was determined
with the addition of 3 mg/L H2C>2 to the reductant solution immediately prior
to ozone sampling. The results are given in Table V.
Table 5. Effect of Hydrogen Peroxide on Residual Ozone
mg/L 03 mg/L 03
(No H202) (H202)
UV 8.98 ± 0.05 8.79 ± 0.12
Indigo 9.52 ± 0.45 9.11 ± 0.05
As(III) 9.42 ± 0.23 9.32 ± 0.04
direct
As(III) 11.62 ± 0.20 12.22 ± 0.27
back
The arsenic(III) back titration method gives a residual ozone level
2.64 mg/L or 29% higher than the UV method for analyses on the acid stabilized
ozone solution. Hydrogen peroxide increased this error to 3.43 mg/L or 39%
for the arsenic(III) back titration method. The Indigo and arsenic(III)
direct oxidation titers agreed within 6% of the UV titers.
Ozone decomposition prior to reaction with the reductant is most likely to
complicate the arsenic(III) direct oxidation method due to the relatively slow
reduction of ozone by arsenic(IIl). Iodide ion reduces ozone in a virtually
diffusion controlled reaction. The Indigo reductant solution is buffered at
pH 2 minimizing ozone decay prior to attack of the carbon-carbon bond. In
fact, when an arsenic(III) back titration reductant solution is dosed with
concentrated ozone, the amber iodine color appears immediately and then fades
as the iodine reacts with the arsenic(III). This clearly demonstrates that the
ozone reacts faster with iodide ion than with arsenic(III).
237
-------
When a dilute ozone solution undergoing minimal decay was directly
reduced by arsenic(III) in acetate buffer at pH 4.5 and in phosphate buffer
at pH 6.8, both arsenic determined decay curves were scattered compared to
the UV curve. The acetate curve had wide deviations from the calculated
curve.
The residual ozone concentrations for three steady state solutions were
also determined by these methods and compared with the concentrations deter-
mined by the direct oxidation of arsenic(III) in unbuffered solution at pH
7. (Table VI). The ozone titers do not consistently increase with a decrease
in pH and the ozone titers determined in unbuffered arsenic(III) are low
relative to buffered solutions and the UV method. The UV method provides
a convenient and rapid reference method when working with ozone solutions
free from other absorbing materials.
Table 6. Buffer Effect on Direct Arsenic(III) Method.
UV Acetate Phosphate No Buffer
pH 4.5 pH 6.8 pH 7
6.62 6.13 6.42 5.92
7.00 7.04 6.64 6.34
9.06 8.91 8.81 8.49
If ozone decay prior to reduction by arsenic(III) causes low ozone titers,
then any change in the reductant medium to slow decay should increase the
ozone titer. As generally accepted and as illustrated in Fig. 5, the rate of
ozone decay decreases with decreasing pH. These rates were determined by
direct UV measurement in 0.1 M phosphate buffers ranging in pH from 9.4 to
5.9.
Ozone decay, however, is a complex function of pH and solutes. The
relative rates for ozone decay in solutions at pH 7.0 - 7.2 containing vary-
ing concentrations of phosphate and carbonate ions are listed in Table VII.
The effect of these anions is enormous. The half-life (t^) for purified
water containing no phosphate ion or carbonate ion and adjusted to pH 7.0
with sodium hydroxide, is less than 500 sec. The t}/2 in 0.1 M phosphate,
0.1 M carbonate is longer than 12 hours.
238
-------
Table 7. Relative Rates for Ozone Decay (pH 7.0 - 7.2)
[carbonate]
0.0 M 0.01 M 0.1 M
0.0 M 200 2.2 2.4
[phosphate] 0.1 M 38 2.6 2.0
0.25 M 22 6.0 5.3
0.50 M 46 10 8.4
The Delta amperometric membrane electrode measures ozone concentration in
situ and should not be influenced by ozone instability or overoxidation. A
teflon membrane selectively transports gaseous molecules like ozone to the
cathode and prevents transfer of polar species and ions. The electrode
operates at a potential where only very strong oxidants are reduced. Thus,
the Delta electrode promises the ideal combination of chemical selectivity and
the capability for continuous monitoring. To be practical, the electrode must
be easily calibrated, must remain calibrated for a reasonable length of time,
perhaps a minimum of one working day, and the calibration must be valid over
the working range of the electrode, 0-10 mg/L. Direct UV measurement of ozone
was used to calibrate the electrode.
The kinetic comparison of ozone decomposition by direct UV measurement,
the Indigo method, and the Delta electrode shown in Fig. 6, illustrates two
recurring problems. First, ozone decay determined by the Delta electrode
followed its own rate law. Second, the electrode rarely maintained
calibration on switching from one solution to another. The initial points
measured by the electrode of Fig. 6 are low for this reason.
The change in rate law was traced to the lack of linear response of the
electrode over a large concentration range. The electrode was calibrated by
UV analyses on an acidified ozone solution. A volume of this solution was
removed and replaced with an equal volume of acidified water and the electrode
response and UV absorbance measured. This dilution was repeated several times
to obtain stable ozone solutions of varying concentration within a 2.5 hour
period.
The results in Table VIII show that as the ozone concentration decreases
relative to the calibration concentration, the error increases. For electrode
1, originally purchased from Delta Scientific, a 76 percent error is observed
over a concentration range of 4.8 mg/L 0-3. The error drops to 34 percent for
a concentration change of 3.0 mg/L 03. Electrode 2, a later model generously
supplied by Delta, measured residual ozone in the range 4.0 - 1.9 mg/L with a
2.5 percent error. The error increased to 23 percent at 0.5 mg/L.
239
-------
20-
18-
•
16"
M •
12"
0)
c
o
N
O i(J'
60
6 8-
6
"
2
n
A
D
M A^
V
• Q V
fV pH 5. 90
T
n v T
" V T
^ 0- v -
•
A o
A pH 7.68 Q a
•
A
* A A
O •
pH 8.22
• A
* * * 1 '
O * »
O
PH 9.39
0 0 n
° O n ^
200
800
400 600
Time (sec)
Figure 5. Effect of pH on Ozone Decay in Purified Water with
Phosphate Buffer.
o
N
o
60
e
7 •
fa •
5 -
2 '
1
i
A A A
i
A
UV
INDIGO
DELTA ELECTRODE
* 4
t ^
A
a
A
•
A A
A A
• *
2000
6000
TOO
Time (sec)
Figure 6. Comparison of Ozone Decay in Purified Water.
8000
240
-------
Table 8. Dilution Experiments for Linear Response
UV Electrode 1 % Error UV Electrode 2 % Error
mg/L 0-^ mg/L 0^ mg/L 0-^ mg/L 0^
7.31 7.38 4.03 4.00 -0.7
5.77 6.60 14.4 3.80 3.75 -1.2
4.33 5.81 34.2 2.65 2.55 -3.5
3.35 5.13 53.1 1.86 1.91 2.5
2.54 4.47 76.0 1.30 1.46 12.8
0.84 0.99 17.7
0.54 0.67 22.7
0.30 0.50 66.4
0.16 0.33 107.6
In a kinetic comparison, Oxford waste water treatment plant effluent was
ozonated for 10 minutes and the residual ozone concentrations were determined
over the next 15 minutes by the Indigo method, arsenic(III) back titration
method, arsenic(III) direct oxidation method, and the Delta electrode
(electrode 2). The waste water was then reozonated for 15 minutes and the
decay followed as above. Results are shown in Fig. 7. The decay curve traced
by electrode 2 differs from the other methods and the electrode again appears
to have lost calibration between runs. This instability and unpredictability
of the Delta electrode clearly emphasize the necessity for recalibration for
each run.
Notice that the second decay curve traced by the Indigo method and the
arsenic(III) back titration and arsenic(III) direct oxidation methods is very
similar to the first curve. This was a consistent observation in sequential
ozonation experiments with Oxford Sewage Treatment Plant effluent.
For example, the rate of ozone decay in waste water was determined
following an initial 30 minute ozone treatment. A two hour ozone treatment
followed the next day. On the third day, a 30 minute ozone treatment was
repeated and the ozone decay followed. The decay curves were superimposable.
The ozone decomposition rate was also similar after each of five minute con-
secutive ozone treatments on Oxford Sewage Treatment Plant effluent (Fig. 8).
These experiments imply that once ozone satisfies the initial ozone demand of
waste water, residual ozone levels are controlled by ozone self-decomposi-
tion and not by direct reaction with impurities. Making the reasonable
assumption that sufficient radical scavengers are present to quench the first
order decay process, the residual ozone decay should be pH controlled.
The pH controlled decay rate can be measured by determining the ozone tj/2
at a given pH with increasing concentrations of radical scavengers. The *-\ji
should reach a limiting value for each pH. With knowledge of the pH of the
ozonated effluent and also the maximum tj/2 at this pH, demonstration of the
presence of ozone after a calculated time could be sufficient to assure dis-
infection.
241
-------
0)
ti
o
N
O 4
txO
e
ozone 10 min
© Inciigo
O Electrode
lAs(III) Back
AAs(III) Direct
O •*
O A
ozone W rnin
A
«
O
bUU
200
Time (sec)
Figure 7. Ozone Decay in Waste Water.
10
01
8 6
o
luoo
200U
Time (sec)
mi) boo
Time (sec)
3000
Figure 8.
Ozone Decay in Waste Water Following Sequential Ozonation.
242
-------
For water and waste water treatment, continuous monitoring of residual
ozone is ideal. The direct measurement of the absorbance of aqueous ozone at
259 nm is the most straightforward and simplest method. However, in waste
waters, many impurities absorb in this region producing a large background
absorption. A membrane electrode promises continuous, specific ozone
analysis. Unfortunately the technology is not yet available to provide the
requisite reliability and stability.
All volumetric methods occassionally give a point 30-50 percent removed
from that calculated on an otherwise smooth decay curve. This makes a single
point analysis for residual ozone untrustworthy. The titer or absorbance of
a solution of reductant or indicator should be sufficiently stable to allow
convenient laboratory analyses and ideally to allow field collection with
later laboratory analysis. With the DPD method, the ozone titer changed
rapidly with time for ozone in purified water and for ozone solutions with
added hydrogen peroxide. The arsenic(III) back titration titer steadily
increased for ozone solutions with added hydrogen peroxide (Fig. 9). The
ozone titer by the amperometric method with excess sodium thiosulfate
increased 4 percent in 9 minutes with ozone in purified water. The ozone
titer determined by the arsenic(III) direct oxidation method and the Indigo
method varied less than 3 percent over 3 hours even with added hydrogen
peroxide. The arsenic(III) solutions are stable standard solutions readily
prepared by weight. Stock Indigo trisulfonate would need replacement at least
every ten weeks. Calibration is time consuming. These problems could be
avoided if higher purity dye were readily available and calibration could be
based on weight. The arsenic(III) direct oxidation method shows variable and
significant blanks.
The ozone titers differed among methods only when changes in the ozone-
reductant reaction were involved. Conditions which reduce ozone decay prior
to reaction with reductant, reduced the scatter observed within a single
method and reduced the differences observed among the analytical methods.
This is understandable since direct oxidation by the ozone molecule is
selective and stoichiometric. Oxidations by ozone decay products such as
the hydroxyl radical are non-selective and non-stoichiometric.
The Indigo method minimizes ozone decay by operating at pH 2. Buffers
which slow ozone1 decay increase the ozone concentration determined by direct
arsenic(III) oxidation. Multiple analyses on waste water show few differences
because waste water impurities scavenge the hydroxyl radical and prevent its
reaction with reductant and its catalysis of ozone decay.
The experiments reported here also clearly demonstrate that the purging
technique—widely used to eliminate in situ interferences—is unreliable
because of ozone decomposition during the purge and readsorption steps.
In conclusion, we have found the Indigo method and the arsenic(III)
direct titration method to be the most reliable. Additional comparisons,
along with the recommended detailed experimental techniques will be publish-
ed separately.
243
-------
Id-
17-
16-
15
0) 14
C
O
N
O
bC
6
U
u
m 3 POD
20 1)0 60 80 100 120
Time (min)
140
Figure 9a. Stability of DPD.
20-
19J
0)
C
O
N
O
M
H
17.
16-
15-
14.
13
1 No \\fi
13 pcm
^U i)0 60 SO 100 120 140
Time (min)
Figure 9b. Stability of As(III) back titration.
244
-------
ACKNOWLEDGEMENTS
This work was conducted under Grant Number R 806302010 from the U.S. EPA
(MERL). The authors gratefully acknowledge the work of Dr. Joseph Benga,
Harry Cohen, Dr. Dwlght Emerich, Barbara Thomas, and Dr. Hlroshi Tomiyasu.
LITERATURE CITED
1. Bader, H.; Hoigne', J., Water Res. 10, 377-386 (1976).
2. Bader, H.; Hoigne", J., Water Res. 15, 449-456 (1981).
3. Boyd, A.W.; Willis, C.; Cyr, R., Anal. Chem., 42, 670 (1970).
4. Flamm, D.L., Envir. Sci. and Tech., 11, 978-983 (1977).
5. Kopszynski, S.L.; Bufalini, J.J., Anal. Chem., 43, 1126-1127 (1971).
6. Manley, T.C.; Niegouski, S.J. in "Kirk-Othmer: Encyclopedia of Chemical
Technology", Vol. 14, 2nd ed.; Mark, H.F.; McKetta, J.J. Jr.; Othmer, D.F.
Eds.; Interscience: New York, 1967; pp 410-432.
7. Palin, A.T., Water and Water Eng., July, 271-277 (1953).
8. Parry, E.P.; Hern, D.H., Envir. Sci. and Tech., 7, 65-66 (1973).
9. Peleg, M., Water Res., 10, 331-365 (1976).
10. Smart, R.B.; Lowery, J.H.; Mancy, K.H., Envir. Sci. and Tech., 13,
89-92 (1979).
11. Staehelin, J.; Hoigne', J., 5th World Congress International Ozone Assoc.
Proceedings in Press (1981).
12. "Standard Methods for the Examination of Water and Waste Water'1, American
Public Health Association, 14th ed., American Public Health Association,
Washington, D.C., 1975.
13. Stanley, J.H.; Johnson, J.D., Anal. Chem., 51, 2144-2147 (1979).
14. Sullivan, D.E.; Hall, L.C.; D'Ambrosi, M.; Roth, J.A., Ozone Sci. and Eng.
2_,_ 183-193 (1980).
15. Symons, J.M., "Ozone, Chlorine Dioxide and Chloramines as Alternatives to
Chlorine for Disinfection of Drinking Water", presented at the Second
Conference on Water Chlorination, Gatlinburg, Tenn., November, 1977.
16. Kolthoff, I.M., Lingane, J.J., "Polarography", 2nd ed., Interscience:
New York, 1952.
245
-------
8. CONTROL OF OZONE DISINFECTION BY EXHAUST GAS MONITORING
Albert D. Venosa, and
Mark C. Meckes
Wastewater Research Division
Municipal Environmental Research Laboratory
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
ABSTRACT
The on-site manufacture of ozone is energy intensive, because it is
generated from electric current. Any advance that is made to help reduce
the cost of generating ozone will accelerate its acceptance as a viable
alternative to chlorine.
In the field of chlorination, dose is routinely controlled by a combina-
tion of a flow proportional signal and a chlorine residual signal. No such
control mechanism exists for ozone, primarily because of the difficulty in
accurately measuring true ozone residual. Typically, ozone dose is con-
trolled by turning on and off entire generators as flow increases or de-
creases. This is wasteful and grossly inaccurate, and consequently operating
costs still remain relatively high.
This paper discusses pilot plant data gathered from six different
treatment plant effluents indicating a reasonably good correlation between
the concentration of ozone in the exhaust gas from the contactor and the
total and fecal coliform levels in the final effluent. The advantages of
this approach are: (1) true ozone is being measured (not total residual
oxidant); (2) the reaction is instantaneous and extremely simple to conduct;
(3) it is easily automated; (U) it is useful on a wide variety of secondary
effluents; (5) it is not subject to interferences; and (6) it is not adverse-
ly affected by sudden shifts in effluent quality. There is one underlying
restriction that must be observed if the method is to be successfully applied:
the gas-to-liquid flow ratio must always remain constant. Thus, gas flow
must be paced to liquid flow by a flow proportional signal. If this restric-
tion is met, then dose can be controlled by monitoring ozone in the exhaust
gas and automatically signalling changes in the power to the generator.
INTRODUCTION
The thrust of the Environmental Protection Agency's in-house research
effort on ozone has been directed towards optimizing ozone, contacting to
achieve the desired bacteriological quality with the least amount of ozone
applied. The reason is simply that ozone, which must be generated on-site,
is energy intensive and, therefore, any effort at reducing the use of ozone
will result in a substantial savings in operating costs.
246
-------
Having established the bubble diffuser to be the most efficient contact-
or of five generic types tested (8,9,10), we deemed it necessary to determine
how best to monitor and control applied dose as demand and flow fluctuated.
We hypothesized that there was no reason why a compound loop control mecha-
nism analogous to that used in the chlorination field could not be developed
for ozone. Such a control mechanism would involve a flow signal, which
would increase the rate of addition of ozone to the effluent, and a demand
signal, which would increase the concentration of ozone applied. The problem
confronting us was the lack of ability to measure true ozone residual in the
process water. Conventional measurement techniques do not differentiate
ozone from other oxidants, and without such differentiation a control mecha-
nism would be impossible.
In a brief summary of the literature we found very little research has
been done in this area. In an EPA report published in August 1978 (4),
Miller e_t al. conducted a global survey of all treatment plants (mostly
drinking water) using ozone for disinfection and other uses. They reported
that French plants incorporate a closed loop control system by which the
residual ozone level in the ozonated water is used to control the amount of
ozone supplied to maintain that residual. In a symposium sponsored by the
International Ozone Association in 1975 (6), several speakers discussed the
need for measuring ozone in both the gas phase and the liquid phase. Nebel
and Forde (5), in discussing the principles of industrial and municipal odor
control with ozone, demonstrated that, by installing an ozone meter in the
exhaust gas stack coming off the contactor, it was possible to detect small
concentrations of ozone in the exhaust gas stream and then feed signals to
the generator to vary the ozone input to the contactor. Trussell (7) express-
ed a desire to be able to monitor ozone in the exhaust gas from the contactor
to determine the efficiency of consumption and the amount of ozone needed to
be destroyed before discharge to the ambient atmosphere. His primary in-
terest, however, was measuring ozone in water for dose control purposes.
From the foregoing it is clear that there are two variables that can be
measured for the purpose of establishing an automatic, real time monitoring
tool. They are: ozone residual in the liquid stream and ozone concentration
in the exhaust gas. Both of these variables represent unused ozone, the
former being that amount of ozone transferred to the liquid but not reacted
or decomposed, the latter being that amount of ozone not transferred to the
liquid. The magnitude of both will depend upon the demand of the liquid, the
ozone transfer efficiency of the contactor, the concentration of ozone in
the inlet gas, and the gas flow rate relative to the liquid flow rate. This
paper will discuss measurement of ozone residual and exhaust gas relative to
coliform destruction and attempt to demonstrate the superiority and relia-
bility of exhaust gas measurements as a dose monitoring and control technique.
MATERIALS AND METHODS
Sources of Secondary Effluent
Secondary effluent was obtained from six different sources: (1) the
247
-------
Fair field Wastewater Treatment Plant, Fairfield, Ohio; (2) the Indian Creek
Wastewater Treatment Plant, Cincinnati, Ohio; (3) the Loveland Wastewater
Treatment Plant, Loveland, Ohio; (4) the Mill Creek Wastewater Treatment
Plant, Cincinnati, Ohio; (5) the Muddy Creek Wastewater Treatment Plant,
Cincinnati, Ohio; and (6) the Sycamore Wastewater Treatment Plant, Cincinnati,
Ohio. Fairfield, Mill Creek, and Muddy Creek are conventional activated
sludge treatment plants. Loveland and Sycamore are contact stabilization
plants and Indian Creek uses rotating biological contactors to treat municipal
wastewater. Of the six treatment plants, five receive raw wastewater of
municipal origin. The 6th, Mill Creek, receives wastewater with a high
concentration of industrial wastes (about 50 percent of the organic loading
by weight).
Approximately 20 m3 of a given effluent was collected in a tank truck
on the day of an experiment or the day before and transported to the U.S.
EPA Test and Evaluation Facility, located adjacent to the Mill Creek treat-
ment plant. When an effluent was collected the day before an experiment, it
was recirculated in the tank truck for a minimum of one hour before initia-
tion of the experiment.
Ozone Generation
Ozone was generated from oxygen in a plate type corona discharge genera-
tor (Computerized Pollution Abatement Corporation Model OZ-180G). Oxygen
flow to the ozonator was maintained at a constant 33 L/min. Liquid flow
was a constant 75 L/min, resulting in a gas-to-liquid flow ratio of 0.44.
Changes in the applied dose were accomplished by varying the concentration
of ozone in the gas flow (i.e., by increasing or decreasing the power applied
to the generator).
Ozone Contactor
The ozone contactor used was a bubble diffuser with 3 columns connected
in series. Its design and operating characteristics are fully described
elsewhere (9) •
Sampling
All effluent samples were grab samples and were analyzed for total and
soluble chemical oxygen demand (TCOD and SCOD), total organic carbon (TOC),
total suspended solids (TSS), and turbidity by Standard Methods (1). Total
Kjeldahl nitrogen (TKN), ammonium nitrogen (NHJ-N), and nitrite-nitrogen (N0~
-N) were measured according to Methods for Chemical Analysis of Wastes (3)
and nitrate-nitrogen (NO^-N) according to Kamphake, Hannah, and Cohen (2).
Ozone concentration in the inlet and exhaust gases was periodically deter-
mined iodometrically (1) and continuously by ultraviolet adsorption analyzers
(Dasibi Environmental Corporation, Glendale, California). Ozone residual
(as total residual oxidant) was measured by the reverse titration standard
iodometric method for chlorine residual (1) using the amperometric end point.
248
-------
Samples collected for bacteriological analysis were assayed for total
and fecal coliforms by the standard Membrane Filtration (MF) method (1),
using 0.45 ym GN-6 membrane filters (Gelman Instrument Company). All chemi-
cal and bacteriological samples were collected from a sample tap located at
the bottom of the contactor's third column.
Procedure
The approach chosen was based on the hypothesis that an empirical rela-
tionship found previously predicts, with a reasonable degree of accuracy,
the total coliform density of a municipal wastewater effluent following
treatment with ozone. The relationship is:
log10TC 4.38 - 4.58 (logio T) + 0.040 TCOD (a)
where TC total coliforrns/100 ml after ozonation, and
T = ozone transferred, mg/L
If one knows the TCOD of an effluent, one may be able to predict the
final coliform density at various levels of absorbed ozone. To test this
hypothesis we chose six local treatment plants and grouped them according to
the mean TCOD concentrations in their effluents. Before the ozonation experi-
ments were conducted, nine effluent samples from each of the treatment plants
were collected over a 3-week period and measured for TCOD. The mean TCOD
levels were then compared and like plants were grouped accordingly. Three
groupings resulted: (i) a low TCOD group (three treatment plants), (ii) a
medium TCOD group (two treatment plants), and (iii) a high TCOD group (one
treatment plant).
By rearranging equation (a) and solving for T at 5 different total
coliform levels, we computed five absorbed ozone levels for each of the
three groupings. The applied dose levels needed to achieve the five absorbed
ozone levels were calculated by assuming an average transfer efficiency of
90 percent (previous data had indicated that 90 percent transfer efficiencies
were possible in the bubble diffuser contactor, with oxygen as the feed gas,
at gas-to-liquid flow ratios of <_ 0.5). The resulting five relative dose
levels for each of the treatment plant groupings are presented in Table 1.
The five applied doses (labeled A through E) are relative in the sense that
each one theoretically yields equivalent coliform densities consistent with
effluent quality as long as dose is varied by changing the ozone concen-
tration in the inlet gas stream and maintaining a constant gas-to-liquid
flow ratio. If the empirical model were a good predictor, dose A would
yield the same total coliform density in all effluents, dose B would yield a
lower number in all effluents, and so on.
By grouping the plant effluents in the above fashion, it would facilitate
further analysis of factors affecting ozone disinfection, should there be a
significant difference between effluent sources with respect to post-ozonation
249
-------
coliform densities. To minimize any trend in wastewater effluent quality in
a given day, the design was balanced so that each dose level occurred the
same number of times at each time of day. The five sets of observations
were taken over a period of approximately four hours. Plant effluents were
collected five times from each of the six plants according to a randomized
collection schedule.
Table 1. Relative Dose Scheme for the Three Treatment Plant
Groupings Used in the Study
Relative Calculated Resulting Log]_o
Actual dose, mg/L
dose Total Coliform Density group ia group iib group iiic
A 5
B 4
C 3
D 2
E 1
.0
.0
.0
.0
.0
1
3
5
6
8
-3
.3
.0
.7
.6
2
5
i'
9
12
.3
.1
.6
.8
.4
5
10
15
20
25
.1
.0
.0
.5
.1
aEffluent Sources Indian Creek Plant, Loveland Plant, and
Muddy Creek Plant
bEffluent Sources Fairfield Plant and Sycamore Plant
°Effluent Source Mill Creek Plant
RESULTS
Effluent Quality
Table 2 summarizes the physical-chemical and bacteriological character-
istics of the effluent sources prior to ozonation. The Mill Creek effluent
contained substantially higher amounts of TCOD and TOC than any of the other
effluents. This was due to the high proportion of industrial components
present in the raw wastewater entering the plant.
Effect o_f Absorbed Ozone on Coliform Numbers
We have shown previously (9,10) that total and fecal coliform levels in
a given effluent can be predicted if the demand properties of the effluent
and the absorbed ozone dose are known. Figure 1 is a graph of the log total
and fecal coliform numbers in the six effluents as a function of the amount
of ozone transferred to the effluents. The data were averaged over the five
replicate runs. Clearly, dose responses in five of the six effluents were
250
-------
similar. The coliform decline in the sixth effluent, the Mill Creek Treat-
ment Plant, deviated significantly from the others, reflecting the substanti-
ally higher demand characteristics of that effluent and indicating that such
responses are not universally predictable or applicable. Thus, attempts to
monitor disinfection efficiency by measuring ozone transfer may lead to
erroneous results.
Table 2. Secondary Effluent Characterization of the Six
Treatment Plants Prior to Ozonation
Effluent Source
Parameter
TCOD, mg/1
TOC, mg/1
TSS, mg/1
TKN, mg/1
NHj-N, mg/1
NOjj-N, mg/1
N03-N, mg/1
Turbidity, JTU
pH
log1QTC/100 ml
log1QFC/100 ml
Fairf ield
mean
(range)
39
(63-22)
5.6
(11.1-1.3)
4.1
(13.2-1.4)
5.3
(12.3-1.9)
4.3
(12.4-0.1)
0.6
7.5
(13.8-4.3)
2.5
(5.3-1.0)
(8.3-7.3)
5.41
(6.51-4.28)
4.69
(5.60-3.30)
Indian
Creek
mean
(range)
26
(48-16)
5.2
(9.0-2.2)
9.2
(36.8-1.8)
1.7
(7.2-0.5)
0.3
0.2
(1. !-<.!)
7.5
(14.1-3.2)
1.7
(7.2-0.5)
(8.4-7.9)
5.59
(6.11-5.08)
4.43
(5.00-3.76)
Lovel and
mean
(range)
39
(57-30)
6.8
(13.8-3.7)
8.3
(34.0-3.7)
13.8
(17.8-9.8)
11.7
(18.6-6.4)
0.5
(1.3-0.1)
0.1
13.8
(17.8-9.8)
(8.2-7.4)
6.56
(7.27-6.16)
5.80
(6.36-5.15)
Muddy
Creek
mean
(range )
29
(45-16)
5.0
(9.0-1.2)
3.8
(9.2-0.4)
1.5
(4.5-0.6)
0.8
0-2
6.0
(8.8-3.0)
1.5
(4.5-0.6)
(8.1-7.3)
5.41
(6.33-4.65)
4.59
(5.21-3.79)
Mill Sycamore
Creek
mean mean
(range) (range)
74
(103-53)
19.6
(29.9-13.9)
11.5
(25.0-2.8)
18.7
(31.0-8.1)
19.3
(31.8-7.3)
0.6
0.2
18.7
(31.0-8.1)
(7.9-7.4)
5.72
(6.84-4.68)
4.75
(6.18-3.85)
38
(56-26)
8.3
(12.7-4.6)
8.6
(18.8-3.0)
7.5
(17.6-4.4)
5.6
(7.3-4.5)
0.5
(1. !-<.!)
2.5
7.5
(17.6-4.4)
(7.9-7.2)
6.43
(6.86-6.08)
5.67
(5.94-5.46)
Effect £f Ozone Residual on Coliform Numbers
Figure 2 is a plot of total and fecal coliform numbers in the six ef-
fluents as a function of ozone residual in the liquid. Again, the data were
averaged over the 5 replicate runs. Response patterns are similar to those
shown in Figure 1, although the deviations in the Mill Creek effluent are
not as great. Use of ozone residual in the liquid as a real time monitoring
251
-------
tool may be appropriate if it is expected that fluctuations in wastewater
quality are relatively minor. However, as will be shown below, even if the
quality has not changed significantly, the presence of compounds or substances
which interfere with the measuring technique may argue against use of ozone
residual as the primary control technology.
Effect of Exhaust Gas Ozone on Coliform Numbers
Results of plotting log coliform numbers as a function of ozone concen-
tration in the exhaust gas are presented in Figure 3- Clearly, the data
from all six treatment plants fit the indicated curve quite well, suggesting
strongly that measurement of ozone in the exhaust gas from the contactor may
be an excellent control strategy. There is a very important restriction,
however, that must be incorporated when using this strategy: the gas-to-
liquid flow ratio must be held constant at all times. The reason is that the
mass transfer efficiency of the contactor decreases markedly as the gas flow
increases (9). Thus, an increase in gas flow relative to liquid flow may
result in a higher exhaust gas ozone concentration without any corresponding
increase in mass transfer or coliform reduction. In contrast, by maintaining
a constant gas-to-liquid ratio and varying the power (or frequency) to the
generator, the increase in ozone transferred will be almost in direct propor-
tion to the higher ozone concentration in the inlet gas, up to the limit
defined by Henry's Law. A higher exhaust gas level will occur also, but
only after more ozone has been transferred to the water- Thus, an increase
in coliform reduction will necessarily take place.
Interferences in Residual Measurement
To demonstrate further the inferiority of monitoring ozone residual in
the liquid to control disinfection, all 125 data points from five of the six
treatment plant effluents (the Mill Creek effluent was excluded) were used
to plot log total coliforms in the effluent as a function of ozone residual.
Results are shown in Figure 4. Although a clear trend is noted, 4 data
points (represented by open squares) stand out as significant outlyers.
Upon studying carefully the computer printout of the raw data, we discovered
that these high coliform and ozone residual values all occurred in the same
run with the same effluent (i.e., the Muddy Creek effluent). The only unusual
feature of this effluent on that date was a high concentration of manganese,
approximately 0.8 mg/L. This amount exceeded the normal levels measured on
all other days by more than 25-fold. According to the 15th edition of Stand-
ard Methods (1), oxidized forms of manganese give positive interferences in
all methods for total available chlorine (the method we were using for ozone),
including amperometric titration. Thus, if an operator is measuring ozone
residual by standard, state-of-the-art techniques, he could be misled on
days when positive interferences are present unexpectedly in the effluent.
On such days coliform discharge limitations could be exceeded.
We made a similar plot of log total coliform numbers versus exhaust gas
ozone (Figure 5). The Mill Creek effluent is included in this plot. Clearly,
no outlyers are observable on the graph because true ozone is being measured
252
-------
in the gas phase. This again confirms the usefulness and reliability of
exhaust gas monitoring for control of ozone disinfection at a secondary
treatment plant.
DISCUSSION
In this paper we have demonstrated empirically that disinfection with
ozone can be controlled by monitoring the exhaust gas ozone concentration
exiting the contactor. This method is more reliable than measuring dissolved
ozone because of the inherent difficulties and inadequacies of state-of-the-
art dissolved residual techniques. The advantages of measuring exhaust gas
ozone are summarized as follows: (1) true ozone is being measured, free of
interferences; (2) -ozone demand of the effluent and transfer efficiency of
the contactor are automatically accounted for in one measurement; (3) the
method is easily automated; (4) instruments are already available on the
market for measuring ozone in the gas phase with accuracy, precision, and
low level sensitivity; and (5) ozone is more stable in the gaseous phase
than in the liquid phase, and consequently the operator does not have to
concern himself with dissipation of the ozone from the time it leaves the
contactor to the time it arrives at the analyzer.
It must be emphasized that exhaust gas monitoring is only applicable if
the gas-to-liquid flow ratio is held constant. The control loop is then
envisioned as follows: (1) a flow proportional measurement signals a change
in the gas flow from the ozone generator to the contactor as liquid flow
changes, thereby keeping the ratio constant; (2) as ozone concentration in
the exhaust gas changes either as a result of a change in demand of the
effluent or a change in flow conditions, a signal is sent to the ozone genera-
tor to change the power or frequency input accordingly. Thus, disinfection
is controlled easily, reliably, and with confidence. The effect this control
strategy has on the cost of ozone production has yet to be evaluated. The
data presented in this paper were obtained using a plug flow bubble diffuser
contactor. There is no reason to believe, however, that the control strategy
would not be applicable to other types of ozone contactors as well.
ACKNOWLEDGMENTS
We thank Messrs. Harold P. Clark and Harld L. Sparks for enumeration of
coliforms in all samples. Ms. Rebecca McCutcheon and Mr. John Rogers assisted
in sampling, performance of ozone analyses, and operation of the ozone disin-
fection equipment. Chemical analyses were conducted by the Waste Identifica-
tion and Analysis Section, Wastewater Research Division, Municipal Environ-
mental Research Laboratory, U.S. EPA, Cincinnati, Ohio.
253
-------
LITERATURE CITED
1. American Public Health Association. 1981. Standard Methods for the Exami-
nation of Water and Wastewater , 15th ed . , Araer . Pub. Health Assoc . ,
Inc., Washington, B.C.
2. Kamphake, L. J., S. A. Hannah, and J. M. Cohen. 196?. "Automated Analysis
for Nitrate by Hydrazine Reduction," Water Research^: 205.
3. Methods Development and Quality Assurance Research Laboratory. 1974.
"Methods for Chemical Analysis of Water and Wastes," EPA-625/6-74-003,
U.S. Environmental Protection Agency, Cincinnati, Ohio.
4. Miller, G. W. , R. G. Rice, C. Michael Robson, R. L. Scullin, W. Kuhn ,
H. Wolf. 1978. "An Assessment of Ozone and Chlorine Dioxide Technologies
for Treatment of Municipal Water Supplies," EPA-600/2-78-147 , U.S.
Environmental Protection Agency, Cincinnati, Ohio.
5. Nebel, C. and N. Forde. 1976. "Principles of Deodorization with Ozone,"
in Ozone: Analytical Aspects and Odor Control , R. G. Rice and M. E.
Browning, editors. International Ozone Institute, Inc., Syracuse, N.Y.
pp. 52-64.
6. Rice, R. G., and M. E. Browning, ed . 1976. Ozone : Analytical Aspects and
Odor Control . International Ozone Institute, Inc., Syracuse, N. Y. ,
7. Trussell, R. 1976. "Ozone Analytical Methods Needs," in Ozone : Analytical
Aspects and Odor Control , R. G. Rice and M. E. Browning, editors.
International Ozone Institute, Syracuse, N.Y., pp. 12-17.
8. Venosa, A. D. , E. J. Opatken, and M. C. Meckes . 1979. "Comparison of
Ozone Contactors for Municipal Wastewater Effluent Disinfection." EPA-
600/2-79-098. U.S. Environmental Protection Agency, Cincinnati, Ohio.
9. Venosa, A. D. , M. C. Meckes, E. J. Opatken, and J. W. Evans. 1979.
"Comparative Efficiencies of Ozone Utilization and Microorganism
Reduction in Different Ozone Contactors." in Progress in Wastewater
Disinfection Technology. A. D. Venosa, ed . EPA-600/9-79^0l8, U.S.
Environmental Protection Agency, Cincinnati, Ohio, pp. 144-162.
10. Venosa, A. D. , M. C. Meckes, E. J. Opatken, and J. W. Evans. 1980.
"Disinfection of Filtered and Unfiltered Secondary Effluent in Two
Ozone Contactors." Environment International. 4: 299-311.
254
-------
O
o
E
o
"5
O
"<5
o
D)
O
64-
0 10 20 30
Ozone Transferred, mg/L Effluent
o
o
CO
o
a)
05
O
5--
4--
3.
o
O
2--
o Sycamore
A Loveland
a Fairfield
• Indian Creek
A Muddy Creek
• Mill Creek
0 10 20 30
Ozone Transferred, mg/L Effluent
Figure 1. Effect of Ozone Transferred on Coliform Densities in Secondary Effluents.
Each Point is Average of 5 Data Points.
-------
Ui
6--
0 0.5 1.0 1.5
Ozone Residual, mg/L Effuent
5--
4--
o
o
r-
\
i
| 3-
o
O
CO
o
0)
LL
CD
O
2--
Sycamore
Loveland
Fairfiefd
Indian Creek
Muddy Creek
Mill Creek
4-
h-
0 0.5 1.0 1.5
Ozone Residual, mg/L Effluent
Figure 2. Effect of Ozone Residual on Coliform Densities in Secondary Effluents.
Each Point is Average of 5 Data Points.
-------
t_n
0123
Ozone in Exhaust Gas, mg/L Gas
o
o
£
o
V£
"o
O
"co
o
0
O)
o
o
54-
2-1-
o Sycamore
A Loveland
n Fairfield
• Indian Creek
± Muddy Creek
• Mill Creek
14-
0123
Ozone in Exhaust Gas, mg/L Gas
Figure 3- Effect of Ozone Concentration in Exhaust Gas
on Coliform Densities in Secondary Effluents.
Each Point is Average of 5 Data Points.
-------
to
CO
O
a
6 --
5 -
4 --
o
(J
S 3
o
1 --
0
S
S
s
•I t
+
0.2 OA 0.6 0.8 1.0 1.2
Ozone Residual, mg/L
1.4
1.6
1.8
2.0
Figure 4. Effect of Ozone Residual on Total Coliforms
in 5 of the 6 Effluents (All Data).
-------
6 --
o
0
w
£
o
O
"5
*-
o
)
o
4 J-
3 --
2 --
. :
;
1 --
0
1.0
2.0
3.0
4.0
Ozone in Exhaust Gas, mg/L
Figure 5. Effect of Ozone Concentration in Exhaust Gas
on Total Coliforms in All 6 Effluents (All Data Points).
-------
9. OPTIMIZING OPERATIONAL CONTROL OF OZONE DISINFECTION
Enos L. Stover, Associate Professor
Bioenvironmental and Water Resources Engineering
Oklahoma State University
Stillwater, Oklahoma
ABSTRACT
Ozone is being developed and employed for disinfection of municipal waste-
water effluents in the United States as an effective alternative to chlorine.
The factors affecting ozone system performance, water quality, transfer effi-
ciency, and absorbed ozone concentration, are key issues to be addressed in
both design and operation of ozone systems for municipal wastewater disinfec-
tion. The ozone contacting system cannot be optimized independently of the
ozone generating equipment, because transfer efficiency in the contactors and
power requirements for ozone generation are both related to ozone concentration
in the carrier gas and the carrier gas flow rate.
Optimum operation of this equipment is required to minimize electric
power consumption and thus operating costs. However, cost optimization must
also consider achievement of the disinfection objectives. Therefore, the dis-
infection requirements in conjunction with both ozone production and ozone
transfer efficiency become the key factors in optimization of ozone disinfec-
tion systems. A relationship must be developed between the disinfection re-
quirements and the ozone system operating conditions to maximize disinfection
and minimize power consumption. Upon development of these relationships the
optimum operating conditions can be achieved.
Proper monitoring of these relationships such as ozone gas monitoring and
residual liquid ozone monitoring is required to maintain adequate disinfection
in the most efficient manner possible. This paper presents one approach for
optimizing operation of ozone disinfection systems, including definition of
disinfection requirements, ozone contacting and ozone generating equipment, as
well as instrumentation requirements for monitoring and control.
INTRODUCTION
Municipal wastewater disinfection by ozonation is a relatively new and
rapidly developing concept in the United States today, and is perceived by
many to be the most attractive alternative to chlorination. There is very
little information available describing design factors and operation criteria
260
-------
for ozone disinfection facilities. The primary objectives that must be con-
sidered include the design and operation of an effective, reliable, economic
and safe ozone disinfection system with minimal power consumption and main-
tenance requirements. Design and operation of such a system requires an under-
standing of the ozone generation equipment, ozone contacting equipment, fac-
tors affecting performance of this equipment, and ozone system instrumentation,
monitoring and controls.
Water quality, ozone transfer efficiency and absorbed ozone requirements
are key factors affecting ozone system performance that must be addressed for
efficient design and operation of ozone systems for municipal wastewater dis-
infection. Since water quality influences both the ozone dose requirements
and the ozone transfer efficiency, the ozone contacting system capabilities
must be defined. The ozone contacting system cannot be optimized independently
of the ozone generating equipment, because transfer efficiency in the contac-
tors and power requirements for ozone generation are both related to ozone
concentration in the carrier gas and the carrier gas flow rate.
An evaluation of ozone production efficiencies over the expected opera-
ting conditions must be considered for proper equipment selection and optimum
operation. This can be accomplished by monitoring or mapping the power con-
sumption versus ozone production over the available carrier gas flow range and
available applied voltage range. This type of information can then be used to
determine the proper size and number of ozone generators required to achieve
the most economical design and optimal operating conditions. The ozone con-
tacting equipment can also be evaluated to define the optimized operating con-
ditions by monitoring the ozone transferred into the wastewater at various
operating conditions. This concept for design of ozone disinfection systems
has been previously described (2,3). The purpose of this paper is to combine
these concepts of ozone equipment definition with disinfection requirements,
instrumentation and monitoring equipment and controls for optimizing opera-
tional control of ozone disinfection.
OPTIMIZING POWER CONSUMPTION
In order to optimize or minimize power consumption for ozone production,
it is necessary to evaluate the ozone generating equipment to define the
economics of disinfection with ozone produced from the appropriate carrier
gas (air or oxygen). This can be accomplished by mapping the ozone generator
by monitoring the power consumption versus ozone production, as shown in
Figures 1 and 2. In Figure 1 the ozone output is shown as a function of
power consumption for the production of ozone from air, while in Figure 2 the
ozone output is shown as a function of power consumption for the production
of ozone from oxygen by the same ozone generator. The differences in genera-
tor power requirements for air versus oxygen generation of ozone are signifi-
cant, as can be observed in these figures. The economics of oxygen supply at
a given site would have to be considered as a function of carrier gas prepa-
ration to evaluate the actual total difference in economics for oxygen versus
air operation. The total system power requirements (economics) for carrier
261
-------
600
500
400
o
Q
§300
a
UJ
Z
O
N
°200
100
3 l/s (30M3/HR)
6.9 l/s (25M3/HR)
5.6 I/s (20M3/HR)
4.2 l/s
I
• 2.8 l/s (10M3/HR)
2.1
12
14
18
20
22 24
W-HR/G
26
28
30
32
34
Figure 1. Ozone Generator Performance - Ozone Production Versus Power Consumption at
Various Gas Flow Rates Using Air Carrier Gas.
-------
o
o
1100
1000
900
800
fr
; 700
600
o
QC
Q.
O
N
O
500
400
300
200
100
2.5 5
Figure 2.
[8.3l/s|(30M3/HR)
5.9 l/s (25M3/HR)
/6
5.6 l/s (20M3/HRJ
2.8I/S(10M3/HR)
2.1 l/s (7.5M3/HR)
1.4 l/s
10
15
W-HR/G
20
25
Generator Performance - Ozone Production Versus Power
Consumption at Various Gas Flow Rates Using Oxygen
Carrier Gas.
263
-------
gas preparation and handling, ozone production, ozone contacting and ozone
destruction can then be evaluated for comparison of air versus oxygen opera-
tion.
Economics of ozone disinfection dictate that ozone be utilized very effi-
ciently due to the relatively expensive methods of ozone production available.
Thus, the ozone contacting system must be designed for optimal ozone transfer
or utilization by employing established principles of mass transfer and re-
action kinetics. The ozone contacting system capabilities must be defined
for optimization of the total ozone system (generation and contacting). The
optimum obtainable ozone transfer efficiency compatible with economic ozone
production required to achieve the disinfection objectives can then be deter-
mined. The ozone contacting equipment can be evaluated to define the opti-
mized operating conditions by monitoring the percent ozone transferred into
the wastewater versus the carrier gas ozone concentration at constant applied
ozone doses, as indicated in Figure 3 for filtered secondary effluent. The
shaded region represents the ozone gas concentration range where oxygen car-
rier gas operation starts becoming necessary to achieve the higher ozone con-
centrations.
As the applied ozone dose and subsequently the gas to liquid ratio in-
creases at a constant carrier gas concentration, the percent ozone transfer
into the effluent decreases even though the absorbed ozone concentration or
quantity of ozone added to the effluent increases. The percent ozone trans-
fer at a given applied ozone dose increases with increasing carrier gas con-
centration and corresponding decreasing gas to liquid ratio. Ozone contactor
transfer efficiencies are higher during oxygen operation due to the higher
ozone concentrations and lower gas to liquid ratios available to achieve the
same applied ozone dose requirements compared to air operation. Under iden-
tical operating conditions of applied ozone dose, carrier gas ozone concen-
tration, gas to liquid ratio and hydraulic flow rate, the percent ozone trans-
ferred into the effluent is independent of the type of carrier gas (air or
oxygen), as shown in Figure 3.
As observed in Figure 3, several different operating conditions and
applied ozone doses can be employed with a given effluent quality to achieve
the same absorbed ozone concentration. In combining the contactor evaluation
with the generator evaluation the ozone generating-contacting system can be
optimized to achieve the desired absorbed ozone concentration. The absorbed
ozone concentration(s) must then be correlated to the required disinfection
objective(s), and proper instrumentation, monitoring and control used to main-
tain the disinfection objective(s) while ensuring optimized operation of the
ozone generation and contacting equipment.
INSTRUMENTATION AND MONITORING REQUIREMENTS
Correlation of disinfection requirements with absorbed ozone concentra-
tions provides the opportunity to achieve the disinfection objectives while
maintaining optimal operation of both the ozone contactor and generator by
264
-------
100
K3
cr>
Ui
APPLIED OZONE DOSE
m9°3/'iiq (OXYGEN,)
APPLIED OZONE DOSE
CARRIER GAS OZONE CONCENTRATION, mgO3/|
gas
Figure 3.
Ozone Transfer Efficiency at Various Applied Doses for Filtered Secondary
Effluent at Average Effluent Flow Rate (2.5 1/s, 9 M3/HR) Using Both Air
and Oxygen Carrier Gas.
-------
the methods established in the previous section. The results of such an
analysis are presented in Figure 4 for the filtered secondary effluent quality
shown in Table 1. This analysis was conducted from the data collected over a
two year time period. Over the wide range of operating conditions evaluated
during this time period, good correlation was always observed between total
and fecal coliform reduction and absorbed ozone concentration.
Table 1. Summary of Filtered Secondary Effluent Characteristics
Parameter Mean Standard Minimum Maximum
Deviation
TSS, mg/£ 4.8 3.8 1.6 16.4
Turbidity, NTU 4.2 2.2 1.4 12.0
COD, mg/£ 40 6.5 21 52
PH — -- 6.9 7.9
Temperature, °C — — 6 21
Log10 total coliforms/100 ml 5.4 0.5 4.5 6.4
TOG, mg/£ 20 — 10 40
TKN, mg/£ 34 — 21 52
NH3-N, mg/£ 14 — 12 16
N02-N, mg/& 0.2
N03-N, mg/& 0.1
Color, Pt-Co 50 — 45 100
Determination of the absorbed ozone concentration requires monitoring of
the carrier gas ozone concentration and the contactor off-gas ozone concen-
tration. From these gas measurements the percent ozone transfer efficiency
can be determined and multiplied by the applied ozone dose to yield the
absorbed ozone concentration (4). Ozone concentration in the carrier and off-
gasses can be determined iodometrically by the method of Birdsall, Jenkins
and Spadinger (1). This procedure is a manual method requiring collection of
a gas sample in a gas washing bottle containing potassium iodide solution and
measurement of the gas volume sampled by a wet test meter. The amount of
ozone reacted with the potassium iodide is then determined by titration. This
procedure can be used to accurately determine the ozone gas concentration;
however, it is too cumbersome and time consuming to be used as a monitoring
tool for ozone production and requires manual feedback to the generator for
control of ozone production.
Instrumentation, such as the Dasibi Environmental Corporation Ozone
Analyzer (Model 1003-HC) used in this study, are also available for monitoring
ozone concentration in the gas streams. These instruments when properly
maintained and recalibrated on a daily basis by the previously described
iodometric procedure can provide reliable monitoring of system performance.
In order to provide system control by absorbed ozone concentration, both the
carrier gas and off-gas ozone concentrations would have to be monitored by
two analyzers or one analyzer with alternating gas streams. Next the
266
-------
10
1
o
o
z
1>
o
QC
O
O
o
g
10'
2.2
«P
LOG Y =-0.088X + 3.47
r = 0.8
I
I
I
14 21 28 35
ABSORBED OZONE CONCENTRATION , mg/l
42
Figure 4. Effluent Total Coliform Value Versus Absorbed Ozone
Concentration.
267
-------
absorbed concentration would have to be determined and a signal relayed to
the ozone generator to control the ozone output. The ozone output must be
determined in terms of the applied ozone dose to the contactor and the ab-
sorbed ozone concentration evaluated as a function of both generator produc-
tion and contactor transfer efficiency, as previously explained. Even with
reliable instrumentation this is a very complicated procedure which still
requires manual input to maintain optimized operating conditions and disin-
fection.
Excellent correlations of total and fecal coliform reduction were also
observed with effluent total residual oxidants or total residual ozone, as
shown in Figure 5 for total colifonus. A modification of the amperometric
titration method for total residual chlorine was employed throughout the two
year study period for determination of total residual oxidants and establish-
ment of the solid line relationship shown in Figure 5 (4). This test method
measures total residual oxidants, such as ozone,peroxides, etc., that may be
produced during the ozonation process. A Delta Scientific Continuous Auto-
matic Ozone Monitor Controller (Model 8340) was also used during the latter
stages of the project to determine dissolved ozone levels in the ozonated
effluent. Monitoring was performed continuously during this stage of the
project with the immersed Delta Scientific probe that is claimed to be speci-
fic for dissolved ozone. This Delta Scientific residual ozone monitor also
provided reliable instrumentation capabilities when properly maintained and
calibrated on a daily basis.
This instrument provided residual ozone readings that correlated well
with the total residual oxidant levels determined amperometrically, as shown
in Figure 6. The residual ozone levels were typically around 60 percent of
the total residual oxidant values. Total residual oxidants and residual
ozone both correlated well with the absorbed ozone concentration as shown in
Figure 7. Since these parameters correlate well with absorbed ozone concen-
tration, they both provide excellent potential as a process control parameter
for ozone disinfection.
Determination of total residual oxidants is a manual procedure, and thus,
presents the same disadvantages as manual determination of absorbed ozone
concentration for use as a process control parameter. However, instrumental
determination of residual ozone provides an excellent opportunity for process
control by providing a direct signal to the ozone generator. A simple feed-
back control loop from the residual analyzer to the ozone generator could be
used to provide process control by monitoring of a single parameter. The
excellent correlation of residual ozone to total residual oxidants (Figure 6)
allowed the development of the dashed line in Figure 5 which presents efflu-
ent total coliforms as a function of effluent residual ozone concentrations.
DISINFECTION SYSTEM OPTIMIZATION APPROACH
The filtered secondary effluent quality of this study required residual
ozone concentrations of around 0.5 mg/£, 2.5 mg/£ and 5.0 mg/£ to achieve
268
-------
10"
2
O
I10*
z
D
O
cc
O
O
O
I
10
2.2
10
-TOTAL RESIDUAL OXIDANTS
LOG Y = -0.387X +3.14
r= 0.7
\
\
~ RESIDUAL OZONE\
\
4 6
RESIDUALS, mg/l
8
10
12
Figure 5. Effluent Total Coliform Value Versus Total Residual
Oxidants and Residual Ozone.
269
-------
6
5 —
en
UJ
z
o
N
O
o
oi
ec 2
Y= 0.62 X-0.17
r = 0.85
1 2345 6
TOTAL RESIDUAL OXIDANTS, mg/I
Figure 6. Residual Ozone Versus Total Residual Oxidants
270
-------
en
£ 2
TOTAL RESIDUAL OXIDANTS
Y = 0.16X+0.33
r =0.75
RESIDUAL OZONE
Y=0.13X-0.42
r = 0.70
5 10 15 20 25 30
ABSORBED OZONE CONCENTRATION, mg/l
35 37
Figure 7. Total Residual Oxidants and Residual Ozone Versus Absorbed
Ozone Concentration.
271
-------
effluent total coliform values of 1000, 70 and less than 2.2 counts per 100
milliliters, respectively, (Figure 4). These residual ozone values can then
be correlated with the required absorbed ozone concentrations to provide the
required levels of disinfection, as shown in Table 2. These required ab-
sorbed ozone concentrations can then be evaluated in conjunction with the
contactor and generator mapping curves to determine the optimized operating
conditions for both ozone production and ozone contacting to achieve the
desired absorbed ozone concentrations. This economically optimized absorbed
ozone concentration thus provides the required residual ozone concentration
and the required disinfection objectives.
Table 2. Effluent Residuals and Absorbed Ozone Requirements to Obtain
Disinfection Objectives
Effluent Total Residual Absorbed
total Residual Ozone Ozone
coliforms/100 m£ Oxidants, mg/£ mg/£ mg/£
1000 0.5 0.5 5
70 3.5 2.5 20
<2.2 7.5 5.0 38
An evaluation of this ozonation system operating for achievement of the
disinfection objective of 70 total coliforms per 100 milliliters is shown in
Table 3. The numbers generated in Table 3 can be developed from the infor-
mation shown in Figures 1,2 and 3. As can be seen in Figure 3, this system
providing a 20 mg/£ absorbed ozone concentration is operating in the range of
ozone production requirements where oxygen operation starts becoming feasible
due to the high ozone gas concentrations required. Operation with oxygen
carrier gas would be more efficient than air operation because of the lower
power requirements, less wasted power and higher ozone transfer efficiencies
possible. As the ozone gas concentration increases, the gas flow rate re-
quirement decreases, the ozone output decreases, and the total power require-
ment for ozone production decreases. Of course the economics of air versus
oxygen preparation and handling would also have to be considered. The only
difference in total economics for this system analysis presented here during
air versus oxygen operation would be in gas handling and preparation costs.
The optimized operating condition during air carrier gas operation was
at an applied ozone dose of 25 mg/& with an ozone transfer efficiency of
80 percent and generator power requirement of 4275 watt power draw. This
operating condition corresponded to both optimum ozone production and maximum
ozone transfer efficiency. The optimized operating condition during oxygen
carrier gas operation was at an applied ozone dose of 30 mg/£ with an ozone
transfer efficiency of 67 percent and generator power requirement of 1897
watt power draw. This operating condition corresponded to optimum ozone pro-
duction (minimal power requirement) but not maximum ozone transfer efficiency.
Even though 89 grams per hour of ozone was wasted, the total generator power
272
-------
Table 3. Combined Ozone Generating-Contacting Evaluation to Achieve 70 Total Coliforms Per 100
Milliliters (20 MG/L Absorbed Ozone Concentration Required).
Wasted Ozone
Output
Applied
Ozone
Dose
ing/ A
25
30
35
40
50
20
25
30
35
40
50
Percent
Ozone
Transfer
Required
80
67
57
50
40
100
80
67
57
50
40
* Applied Ozone Dose =
Carrier Gas Carrier Gas* Required**
Ozone Flow Rate Generator
Generator
Power Requirements
Concentration Output
mg/£(g/m3)
(Figure
30 (30)
24 (24)
22 (22)
20 (20)
18 (18)
45 (45)
30 (30)
24 (24)
22 (22)
20 (20)
18 (18)
•a — pno c\r
r i/s (m3/hr) (g/hr)
3)
Operation with Air Carrier Gas
2.1 (7.5) 225
3.1 (11.3) 271
4.0 (14.3) 315
5.0 (18.0) 360
6.9 (25.0) 450
Operation with Oxygen Carrier
1.1 (4.0) 180
2.1 (7.5) 225
3.1 (11.3) 271
4.0 (14.3) 315
5.0 (18.0) 360
6.9 (25.0) 450
„, Gas Flow
W-hr/g
(Figure 1)
19.0
18.5
18.0
18.5
18.0
Gas (Figure 2)
13.0
9.0
7.0
7.0
7.0
6.5
Rate ,
Watts
4275
5013
5670
6660
8100
2340
2025
1897
2205
2520
2925
Percent
20
33
43
50
60
0
20
33
43
50
60
g/hr
45
89
135
180
270
0
45
89
135
180
270
Effluent Flow Rate
** Generator Output = Carrier Gas Ozone Concentration X(Carrier Gas Flow Rate).
Effluent Flow Rate = 2.5 if a (9.0 M /HR).
-------
consumption was lower when compared to lox^er applied ozone doses and corre-
sponding higher ozone transfer efficiencies. This evaluation shows the im-
portance of combining the ozone generating equipment with ozone contacting
to optimize the overall disinfection process to minimum power consumption.
The optimum operating condition with oxygen carrier gas required only 44 per-
cent of the generator power requirement compared to the optimum operating
condition with air carrier gas.
During this two year test period, continuous operating periods were con-
ducted to include night time and weekend testing for evaluation of the ozon-
ation system disinfection reliability, including process control, instrumen-
tation and equipment reliability, at changing water qualities due to diurnal
variations. One such operating period was conducted to evaluate disinfection
of the filtered secondary effluent to 70 total coliforms per 100 milliliters
during operation with air carrier gas. During this operating period the
generator output was paced to maintain a constant applied ozone dose of 30
mg/&. The maximum to minimum effluent flow rates varied by a four to one
ratio. The gas flow rate was varied manually to simulate an automatic gas
flow regulation since the research facility did not have automatic gas flow
rate controllers. During this test period the effluent quality varied as
shown in Table 4. With the applied ozone dose of 30 mg/£ a mean absorbed
ozone concentration of 20 mg/£ and effluent geometric mean total coliform
count per 100 milliliters of 22 was achieved. Successful disinfection to the
70 total coliform level was achieved in greater than 80 percent of the test
observations during this test period.
Table 4. Variability in Filtered Secondary Effluent Characteristics
During Continuous Operating Period (Disinfection
Objective of 70 Total Coliforms per 100 ML.)
Parameter Mean
TSSS mg/£ 4
Turbidity, NTU 5,4
COD, mg/£ 31
PH Q
Temperature, C 15.0
Log total coliforms/100 m& 5.0
Standard
Deviation
1.6
1.8
6.1
0.4
Minimum Maximum
2
2.9
12
6.4
14.5
4.0
11
9.3
53
7-9
15.5
5.9
The total power draw in watts of the complete ozone disinfection system
was monitored over this continuous operating period. The ozone thermal de-
struct unit required a constant 4200 watt power draw, and the two submerged
turbine type ozone contactors averaged 2600 watt power draw each. However,
each ozone contactor was oversized to allow the total research facility hy-
draulic flow to be disinfected in each contactor, as has been previously
described (5). This design was developed for the facility to provide the
operational flexibility required for evaluation of the feasibility of high
274
-------
level ozone disinfection. The power draw of the ozone generator and air
preparation equipment was variable during this test period. The average
power requirement for ozone generation during this test period was 22 W-hr/g,
while the average power requirement for both ozone generation and air hand-
ling and preparation was 24 W-hr/g. This power requirement was beyond the
optimum operating range for ozone production by this generator with air car-
rier gas, due to lack of optimized operational control during this test
period and the changing hydraulic flow rates and ozone demands (the primary
objective at the time of this test period was to demonstrate reliable disin-
fection to 70 total coliforms per 100 milliliters). The changing hydraulic
flow rates and corresponding relationships of ozone transfer and ozone pro-
duction must be evaluated for complete system optimization, as has been
previously presented (2,5).
DISCUSSION AND CONCLUSIONS
The importance of optimizing operational control of ozone disinfection
has been demonstrated during the continuous monitoring period described. The
disinfection objective was achieved; however, the overall system power re-
quirements were not optimized. This lack of ozone generator optimization
was due partially to the required operating conditions which approach oxygen
carrier gas ozone concentrations. Lower power requirements for ozone produc-
tion could be realized with a larger ozone generator or by operating the
existing generator with oxygen carrier gas. However, the optimization ap-
proach presented here can be used both in design or optimization of opera-
tions of an existing system.
Monitoring ozone residual with the proper instrumentation and providing
a direct feed back signal to the ozone generator allows the generator opera-
ting conditions to change in response to changing water quality or changing
ozone demand. The ozone output, ozone concentration and carrier gas flow
rate can be changed to maintain the proper conditions for both optimized
ozone production and ozone transfer. This requires mapping or monitoring
both the ozone generator power requirements and ozone contactor transfer cap-
abilities to combine this information into the most economical operating con-
dition.
Another possible optimization approach suggested by the results of the
continuous testing period would be monitoring of the contactor off-gas con-
centration as an indicator of the absorbed ozone concentration. During this
test period the applied ozone dose was maintained constant at 30 mg/&, while
the generator voltage and gas flow rate were varied with the changing hy-
draulic flow rate to maintain a constant gas to liquid ratio in the contactor.
With this operating strategy the mean absorbed ozone concentration was main-
tained at around 20 mg/£. Since the gas flow rate varied with the hydraulic
flow rate, the gas ozone concentration was maintained constant. Thus, both
the gas ozone concentration and the contactor transfer efficiency were ap-
proximately constant during this test period. With both a constant carrier
gas ozone concentration and transfer efficiency, the contactor off-gas
275
-------
concentration would also be constant. Under this operating strategy the off-
gas ozone concentration could be monitored with the proper instrumentation
and a direct feedback signal employed to control the generator output by main-
taining the constant off-gas concentration. This approach has also been sug-
gested by Venosa (6).
Both approaches presented here appear applicable for optimizing opera-
tional control of ozone disinfection. The off-gas monitoring approach re-
quires control of both the ozone generator gas concentration and flow rate
to maintain a constant gas to liquid ratio. The residual monitoring approach
only requires control of the ozone generator gas concentration to maintain
the desired residual ozone concentration. However, the data generated during
this research study indicates that optimized ozone disinfection, considering
both generation and contacting, appears to be obtained at a constant gas to
liquid ratio. Therefore, liquid residual monitoring would also require main-
tenance of a constant gas to liquid ratio in the contactor.not to accomplish
the disinfection objective but to accomplish economic optimization
ACKNOWLEDGEMENTS
The data used in this analysis was developed at the Marlborough, Massa-
chusetts ozone disinfection research project sponsored by the U.S. Environmen-
tal Protection Agency, the Commonwealth of Massachusetts and the City of Marl-
borough, Massachusetts. At the time of this study the author was Director of
Research and Development at Metcalf & Eddy, Inc. of Boston, Massachusetts.
REFERENCES
1. Birdsall, C.M., A.C. Jenkins, and E. Spadinger 1952. lodometric Deter-
mination of Ozone. Anal. Chem. 24, 662.
2. Stover, E.L. Engineering Requirements for Designing Ozone Systems. Pro-
ceedings of the 8th Annual Industrial Pollution Conference, Houston,
Texas, June 1980. 431-449.
3. Stover, E.L. 1981. Ozone for Municipal Wastewater Disinfection. Water
Engineering and Management. 128,10, 74-76.
4. Stover, E.L., and R.N. Jarnis 1981. Obtaining High-Level Wastewater
Disinfection with Ozone. Journal Water Pollution Control Federation.
53, 11, 1637-1647.
5. Stover, E.L. , R.N. Jarnis, and J.P. Long 1981. High-Level Ozone Disin-
fection of Municipal Wastewater Effluents. National Technical Infor-
mation Service, No. PB 81-172 272.
6. Venosa, A.D. 1982. Control of Ozone Disinfection by Exhaust Gas Moni-
toring. Paper presented at the Second National Symposium on Municipal
Wastewater Disinfection, Orlando, Florida.
276
-------
10. PILOT STUDIES OF OZONE DISINFECTION AND TRANSFER IN WASTEWATER
Patrick W. Given, and Daniel W. Smith, Professor
Senior Environmental Engineer Department of Civil Engineering
Underwood McLellan Ltd. University of Alberta
Edmonton, Alberta Edmonton, Alberta
ABSTRACT
Studies of ozone disinfection and transfer in wastewater, using counter-
current flow contactors, were undertaken in Whitehorse, Yukon. The studies
focussed on the effectiveness of ozone in the reduction of indicator organisms
and relevant factors influencing this reduction in screened, dilute waste-
water (BOD and suspended solids approximately 60 to 120 mg/L). The primary
factor influencing bacterial survival was the amount of ozone utilized in
the contactors. A log-log relationship was evident between bacterial
survival and ozone utilization. Other variables which demonstrated an
apparent effect on bacterial survival included wastewater strength, temper-
ature, and ozone residual. At fecal coliform reductions of 99.9%, approxi-
mately 20 mg/L of ozone utilization was required. Using a high quality
secondary effluent, only 4 mg/L of ozone was required to achieve the equi-
valent reduction. Other benefits of the ozonation system included substantial
wastewater strength reductions and high dissolved oxygen levels in the
effluent. Factors exhibiting an apparent effect on ozone transfer efficiency
included the ozone-oxygen gas flowrate, amount of ozone applied, wastewater
strength, and ozone residual.
INTRODUCTION
Ozonation of screened, dilute, cold, wastewater was studied at pilot
plant scale in the City of Whitehorse, Yukon Territory, Canada. The study
objectives were to evaluate ozone effectiveness in disinfection of the
wastewater and to assess relevant performance factors. Secondary benefits
of ozonation also were to be assessed.
Cold, dilute wastewater often results from the practice of discharging
cold tap water to sewers to prevent water pipe freezing. It can also
result from high in-flow and/or infiltration.
Review of this wastewater management problem in northern regions of
Canada indicated that the most important treatment requirement may be
proper disinfection. With this treatment requirement in mind, the screening-
ozonation pilot scale study was developed as one of a number of alternative
277
-------
treatment techniques for cold, dilute wastewater.
have been analyzed by Smith and Given ( 3) .
These treatment techniques
DESCRIPTION OF PILOT PLANT
The pilot plant consisted of a rotating screen, with slotted openings
(both 0.76 and 0.25 mm slot sizes tested), followed by two counter-current
flow ozone contact columns (5.2 m high and 150 mm diameter). Raw wastewater
was supplied to the rotating screen by a submersible pump from a wet well
in the main wastewater lift station for Whitehorse. The effluent from the
screen was collected in a mixed holding tank from which it was pumped to
the ozone columns at uniform rates.
Another holding tank was used to collect effluent from a rotating
biological contactor (RBC) during the first year of the study. The same
holding tank was used for lagoon effluent and high strength wastewater at
the end of the second study year. These wastes were tested separately with
the ozone disinfection system.
Figure 1. Schematic of Ozone Pilot Plant
278
-------
Ozone was produced by passing oxygen through a 10 g/h, air-cooled
ozone generator. The resulting mixture of oxygen and ozone was distributed
near the bottoms of the columns through 90 mm diameter porous glass diffusers.
The distribution of gas to the columns was controlled by metering valves
and measured using calibrated rotameters with attached pressure gauges.
Figure 1 presents a schematic of the pilot plant system.
METHODS
Pilot Plant Operation
A total of 300 runs of the ozone system were conducted over a two year
period in 1977 and 1978. In the first year, a 3 x 3 x 3 operating matrix
(three 02/03 gas flow rates, three gas distribution ratios to the contact
columns, and three wastewater flow rates) was set up. The operating matrix
is presented in Table 1.
Table 1. Summary of Operation Matrix for Ozone System*
(1)
O2/O3 Gas
Flowrate
L/min @ STP
1.6
3.0
6.0
Ozone
Concentration
( % by wt . )
(4.9)
(3.3)
(1.9)
(2)
Gas Distribution
% Column l/% Column 2
50/50
60/40
70/30
(3)
Wastewater
Flowrate
L/min
3.8
11.4
30.3
Average values presented.
In the second year of the study, a 1 x 1 x 3 matrix was used predominantly,
with gas flowrate at 3.0 L/min, gas distribution at 60% to the first column
and 40% to the second, and only wastewater flowrate was varied.
Normally a set of two to three runs of the ozone system was performed
during a given day and as many as five sets during a given week. Most of
the runs were performed during the January to April period of each year
with wastewater temperatures ranging from 6 to 8 °C. Some runs were also
performed in the summer when the wastewater was as warm as 14 °C.
Before sampling, the ozonation system was operated at the desired
wastewater and 02/03 gas flowrates for a minimum of six volume changes of
the columns. Uniform wastewater flowrates to the columns were set from 3.8
to 30.3 L/min, resulting in total ozone contact times from 48 to 6 minutes,
respectively. The wastewater flowrate variation was the primary method of
varying the ozone dosage (mg 03 per litre of wastewater). Ozone dosage was
also varied to a limited extent by changing the oxygen flowrate through the
ozone generator during different system tests. Gas flowrates were controlled
by calibrated rotameters. Ozone applied and ozone in off-gases were deter-
279
-------
mined by the iodometric procedure in Standard Methods (Procedure 423A) (1).
Ozone residuals in the liquid were determined with a Wallace and Tiernan
amperornetric titrator.
Analytical Procedures
Biochemical oxygen demand over five days (BOD), chemical oxygen demand
(COD), suspended solids (SS) and volatile suspended solids (VSS) were
determined according to Standard Methods (1), Procedures 507, 508, 208D,
208G, respectively. Wastewater turbidity was measured with a Hach turbidimeter.
Total coliforms, fecal coliforms, fecal streptococci and 35 °C standard
plate counts were enumerated by membrane filtration procedures according to
Standard Methods Procedures 909A, 909C, 910B, 907. Results of Salmonella
and virus determinations were reported elsewhere (2).
Data Analysis Methods
The disinfection data for screened wastewater were examined statistically
to evaluate the possible effects of operating procedure, year of sampling,
screen opening size, number of ozone contact columns, and wastewater charac-
teristics. BOD and suspended solids reductions were also examined.
Comparison of disinfection efficiency of ozone using RBC effluent,
lagoon effluent and a strong wastewater was made.
Because of the importance of ozone utilization on bacteria reduction,
additional analyses were focussed on factors affecting ozone transfer
efficiency for the contact columns.
ANALYSIS OF SYSTEM PERFORMANCE
Bacterial Numbers and Survival
The disinfection results for screened wastewater are shown for total
coliforms (TC), fecal coliforms (FC), fecal streptococci (FS), and standard
plant count at 35 °C (SPC) in Figures 2 and 3. The figures show actual
numbers and survival ratios respectively, plotted against ozone utilized
(Cu). Linear regression lines and equations indicate a log-log relationship
for each set of data points.
Of particular note with the indicator organism survival curves are the
intercepts of the regression lines with the abscissa (Cu axis). These
intercepts, which range from 1.8 to 2.5 mg/L of ozone utilized, may be
thought of as the initial amount of ozone which must be utilized in the
wastewater before significant bacterial reductions occur. The intercepts
can be used conveniently along with the slopes of the regression lines to
express bacterial survival in terms of a simple equation:
N/No = (Cu/Cuo)b
280
-------
N = 4.4x107CV26 ^
r = -0.79 n = 329
8 10
fc
• Initial conditions*
10'
10
10'
_ NO = 6.0x106perlOOmL • 4 ^\: ~
; = ° ' s :.^ ^i-
_
; Temp= 8.5 °C
: Turb = 41 FTU
• pH = 7.8
^ BOD = 95 mg/L
'- COD = 185 mg/L
: SS =100 mg/L
VSS = 57 mg/L
1 5 10
Ozone utilized, Cu mg/L
50
: Temp =
: Turb =
: pH = 7.8
! COD = 185 mg/L
: SS =102 mg/L
VSS = 57 mg/L
10
5 10
Ozone utilized, Cu mg/L
106
1.4 x
10 5
10"
103
102
10'
10°
'105
n** = 25
N = 5.7x105C ~2A
r =-0.89
n = 314
Initial conditions'
NO = 1.4 x 105perlOOmL
Temp= 8.3 °C
Turb = 40 FTU
pH = 7.8
BOD = 94 mg/L
COD = 182 mg/L
SS = 96 mg/L
VSS = 56 mg/L
5 10
Ozone utilized, Cu mg/L
50
1.8 x
105
: 104
103
| 102
: 10'
10°
10=
n**= 28
N = 1.8x106Cu"2'5
r = -0.80 n = 293 '
Initial conditions'
NO = 1.8 x 10s per mL
Temp= 7.8 °C
Turb = 39 FTU
pH = 7.8
BOD = 93 mg/L
COD = 178 mg/L
SS =92 mg/L
VSS = 54 mg/L
5 10
Ozone utilized, Cu mg/L
50
LEGEND
1977 data, column 1
1977 data, column 2
O 1978 data, column 1
A 1978 data, column 2
Figure 2. Total Coliform, Fecal Coliform, Fecal Streptococcus and
Standard Plate Count (35 °C) Numbers Versus Ozone Utilized
281
-------
Ozone utilized, Cu mg/L
3
.0
10"
10
,-5
Initial conditions*
= 6.0 x 106perlOOmL
Temp= 8.5°C
Turb = 41 FTU
pH = 7.8
BOD = 95 mg/L
COD = 185 mg/L
SS = 100 mg/L
VSS = 57 mg/L
10C
cq Ozone utilized, Cu mg/L
I 5 10
• ; i
NO = 1.3 x106 per 100 me-5'
Temp = 8.6 °C
Turb = 42 FTU
pH = 7.8
BOD = 95
COD = 185 mg/L
SS = 102 mg/L
VSS = 57 mg/L
Initial conditions*
10°
10"
¥
§
1 iff
r2
10
-3
10'
,-4
10
.-5
cq Ozone utilized, Cu mg/L
7 5 10
ii-/^\-2-4 .
N0 -11.8 / /
r =-0.89 n =314
Initial conditions*
NO = 1.4 x 10s per IOO mL
Temp = 8.3 °C
Turb = 40 FTU
pH = 7.8
BOD = 94 mg/L
COD = 182 mg/L
SS =96 mg/L
VSS = 56 mg/L
10°
10"
Ozone utilized, Cu mg/L
5 10
d
a
10"
1-2
Q 1(T
$2
~CL
| 10"
1
55
10
,-5
Ji-/-£uV2-5
Nn ~\Z5> "
NO = 1.8 x 105 permL
Temp = 7.8 °C
Turb = 39 FTU
pH = 7.8
BOD = 93 mg/L
COD = 178 mg/L
SS =92 mg/L
VSS = 54 mg/L
n** =28
50
LEGEND
1977 data, column 1
1977 data, column 2
O 1978 data, column 1
A 1978 data, column 2
Figure 3. Total Coliform, Fecal Coliform, Fecal Streptococcus and
Standard Plate Count (35 °C) Survival Versus Ozone Utilized
282
-------
where No = initial number of indicator organisms
N = number of indicator organisms after ozonation
N/N0 = bacterial survival ratio
Cu = the concentration of ozone utilized (mg/L)
Cuo = the initial concentration of ozone utilized (mg/L)
or intercept with Cu axis
b = slope of regression line (negative number)
It should be noted that data points for which ozone utilized values
were less than 2.5 mg/L were not included in the regression analyses because
the statistical residuals (observed values minus values predicted by the
regression lines) were generally negative. This indicated that the data
showed transition from the "lag phase" to the "rapid kill phase" when ozone
utilized increased beyond 2.5 mg/L, for the screened wastewater.
Additional data analyses of microorganism survival indicated that
there did not appear to be any significant effects of operating procedures,
year of operation, screen slot size, and number of contact columns.
Consequently, all of the screened wastewater data were pooled in the calcu-
lations for the regression lines.
Predictive Relationships for Bacterial Survival
Stepwise multiple linear regression techniques were used to determine
the statistical significance between bacterial survival as the dependent
variable, and several possible independent variables. Combinations of log
(base 10) transformed and non-transformed variables were investigated to
determine best predictive relationships for bacterial survival.
The log transformations for ozone utilized (log Cu) gave the best
initial correlation with log (N/NO) in all cases (Step 1). The next most
significant variable generally proved to be log BOD (Step 2). However,
both wastewater temperature (log T) and effluent ozone residual (log Cre)
were more significant for some of the indicator organisms. This is illus-
trated in Table 2 which presents the complete summary of the analyses for
each of the indicator organisms.
The tabulated values of r^ (square of the multiple correlation co-
efficient) show that it is of marginal benefit to include more than two
independent variables in the regression equations. The r^ value indicates
the fraction of the total variance of log (N/NO) which is contributed by
its regression on the independent variables.
The apparent effects of BOD and temperature as well as ozone utilized
on the survival of fecal coliforms for the screened wastewater are shown in
Figure 4. It was observed that increases in both initial BOD and temperature
resulted in higher initial ozone demand. This indicates that, for a given
level of ozone utilization, higher survival ratios would occur if either of
these parameters are increased. It should be noted that this is an apparent
effect as the project testing program did not allow absolute control of all
independent variables.
283
-------
Table 2. Summary of Regression Analyses for Bacterial Survival
in Screened Wastewater*
INDEPENDENT
VARIABLE
loeiFC/FCp)
log(TC/TC0)
Log(FS/FS0)
logfSPC/
SPC0)
NV13ER
01"
CQLUMfS
1
2
1 „ 2
1
2
162
1
2
1 & 2
1
2
1 S 2
STEP
1
2
3
4
1
2
3
1
2
3
<•
1
2
5
4
1
2
3
1
2
3
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
REGRESSION COEFFICIENTS FOR
bo
0.74
-3.2
-J.3
-J. 4
0.61
-4. 2
-4.6
_/, _ 2
log Cu
-2.8
-2.6
-2.3
-2.4
-2.8
-2.7
-2.8
-2.5
0.72 -2.3
-3.7 -2.7
-3.5 ; -2.4
-3.8 • -2.5
0.63 -2.2
-l.s -L.l
-2.6 -2.2
0.73 -2.5
-2.8 -2.5
-3.2
0.78
-2.5
-2.8
0.48
-0.37
-2.1
0.49
-2.6
-2.4
0.54
-0.26
-2.1
0.76
-0.70
-2.2
1.0
-2.9
-3.8
-3.5
0.98
-0.61
-2.9
-2.8
-2.6
-2.5
-2.4
-2.5
-2.2
-1.6
-1.6
-2.3
-z.3
-2.0
-2.3
-1.8
-1.9
-2.1
-2.2
-2.1
-2.5
-2.5
-2.6
-2.2
-2.4
-2.5
-2.4
-2.2
log BOD
1.9
1.7
1.5
2.4
2.1
1.7
2.2
1.3
1.7
1.5
1.3
log T
0.62
1.1
0.95
0.86
u.79
1.8
1.5
1.6
1.3
0.89
1.6
1.2
1.0
-0.80
1.9
1.7
1.3
1.3
1.1
1.1
1.1
1.7
1.5
1.6
1.5
1.9
1.6
1.4
log cre
-u.16
-0.13
-0.19
-u.19
-0.15
r*
0.750
0.814
0.833
u.838
0.805
0.872
0.885
0.893
0.793
0.854
0.867
0.875
0.616
0.667
0.679
0.713
0.751
0.766
1
-0.24
-0.21
-0.15
-0.25
-u.18
-0.19
-0.16
0.690
0.728
0.742
0.682
0.748
0.767
0.820
0.863
0.871
0.780
0.819
u.837
0.645
0.703
0.717
0.745
0.792
0.815
0.824
0.723
0.763
0.786
0.791
n
109
137
246
110
138
248
107
132
239
98
123
221
^Regression equation (with one to four independent variables):
log (S/N0) = b0 + bi log xj + b2 log xz + b3 log ^3 + b4 log x4, or
284
-------
IOC
10'
E
i_
o
o
CJ
o
o
,-z
10
ID'3
10"
10
,-5
40 50
Cuo= IO-'-6BOD075Ta39
V
J 1_J L
Figure 4. Apparent Effect of Ozone Utilized, BOD, and Temperature
on Fecal Coliform Survival in Screened Wastewater
Comparisons of Fecal Coliform Survival Ratios for Other Types of Wastes
Data on fecal coliform survival for RBC effluent, anaerobic lagoon
effluent and strong wastewater are shown together in Figure 5. The differences
between the individual regression lines indicate the effect of the differences
in wastewater characteristics.
The primary influence of wastewater characteristics on the FC survival
curves appears to be on the initial ozone demand of the wastewaters, varying
from 0.7 mg/L for the RBC effluent to 12.5 mg/L for the strong wastewater.
In all cases, the initial ozone demands, Cuo, were significantly different
at the 5% level.
The slopes of the regression lines varied from -2.9:1 for screened
wastewater to -4.6:1 for lagoon effluent. This difference was highly
significant (at the 0.1% level), although wastewater characteristics for
screened wastewater and lagoon effluent were similar with respect to BOD
and suspended solids. In comparing the screened wastewater with the RBC
effluent and with the high strength wastewater, no significant differences
in the slopes of the regression lines were observed. However, it should
285
-------
be noted that only 13 data points were obtained for the KBC effluent and 16
points for the high strength wastewater. With additional data, significant
differences between all of the slopes may have been found.
34 Ozone utilized mg/L 125
5 10
V3.4
n = 30
r =-0.92
Screened
Strong
N _/ Cu V3-1^
NO M2.5/
n = 16
r = -0.93
NO V0.7
n = 13
r =-0.82
Type of waste
Parameter RBC screened Lagoon Strong
6.5xl041.3x 106 3.8xl05 2.8x106
- 99.9
o
O
u_
-99.99
99.999
Figure 5. Fecal Coliform Survival for RBC Effluent, Anaerobic Lagoon
Effluent, and Strong Wastewater (Screened Wastewater
Shown for Comparison)
On-Line Operating Variables
The previously developed regression equations provide insight into the
effects of various factors on disinfection; however, they have little prac-
tical significance for on-line operation of a system. The,reason for this
is that a number of the relevant independent variables cannot be monitored
instantly. Therefore, the operator could not provide accurate on-line control
over the ozone disinfection process in response to changing wastewater
characteristics.
286
-------
To examine this problem from the operator's point of view, data analyses
were undertaken using only those variables which could be monitored on-
line. Log effluent fecal colifonn number (log FC) was selected as the
dependent variable. Log transformations of ozone utilized, effluent ozone
residual (log C ), wastewater turbidity (log TURB) and wastewater temperature
(log T) were selected as the independent variables. The results of these
data analyses are summarized in Table 3.
Table 3. Summary of Regression Analyses for log FC in Screened Wastewater*
1 Column
2 Columns
Ii2 Columns
1
2
3
'•<
1
2
3
4
1
2
3
4
REGRESSION COEFFICIENTS FOR
bo
7.0
5.1
4.1
3.6
7.0
5.3
3.7
2.9
7.0
4.8
3.4
3.2
log Cu
-2.8
-2.9
-2.9
-2.3
-/.9
-z.O
-z.2
-/.A
-2.9
-3.0
-3.0
-2.5
log T
2.0
1.7
1.4
l.i
2.3
1.9
1.5
log (TURB)
0.84
0.91
1.3
1.1
1.2
1.1
log Cre
-0.15
-0.46
-0.35
-0.26
-0.20
0.757
0.816
0.833
0.847
0.799
0.863
u.892
0.908
0.791
0.848
0.873
0.888
85
105
190
Regression equation (with one to four independent variables):
log FC = b0 + b± log xi + 62 log X2 + 03 log X3 + b^ log x^,
FC - 10bo. xxbl. X2b2. X3b3.x4b4
All four of the independent variables proved to be significant in the
equations; however, use of only two of these variables, Cu and Cre, would
result in reasonably good accuracy, indicated as follows:
FC (per 100 mL) = 105-3 Cu-2.0 Cr£-O.A6 (with r2 = 0.86)
It should be noted that this predictive equation is only applicable to
ozonation of screened wastewater in this study. The regression coefficients
would likely be different for other wastewater types.
BOD and Suspended Solids Reductions
BOD and suspended solids reductions were achieved by ozonation of the
screened wastewater. The results for BOD reduction are shown in Figure 6.
Suspended solids reductions were similar but had more scatter in the results.
BOD and suspended solids were reduced approximately 25% at an ozone utilization
of approximately 30 mg/L.
In a separate evaluation, screening resulted in negligible BOD reduction
and about 10% suspended solids reduction. Thus, overall BOD and suspended
solids reductions, achievable by screening followed by ozonation, would be
in the order of 25% and 30%, respectively, at an ozone utilization of
approximately 30 mg/L.
287
-------
20
Ozone utilized
Figure 6. BOD Reduction for Ozonated, Screened Wastewater
Analysis of Ozone Transfer Efficiency
Ozone transfer efficiency may be expressed as a ratio of ozone utilized
to ozone applied (Cu/Ca), or as a percent (100 Cu/Ca). Because of the
demonstrated effect of ozone utilization on bacterial reduction, ozone
transfer efficiency for a contactor is very important. Therefore, analyses
of relevant factors associated with ozone transfer efficiency (or utilization
efficiency) were undertaken for the 5 m high ozone contact columns.
Stepwise multiple linear regression analyses for ozone utilization
efficiency are summarized in Table 4. For the first contact column and the
over-all system, factors appearing to affect ozone utilization efficiency,
in order of importance, were the ozone-oxygen gas flowrate, Cv (or ozone-
oxygen gas loading rate or flowrate per unit contacting volume, Qg/V);
ozone applied, Ca; and wastewater strength, BOD. However, the variables
and order of importance changed for the second contact column due to the
ozone residual entering that column from the first column. The factors in
order of importance were then influent ozone residual, Cri; gas loading
rate, Qg/V; and wastewater strength, BOD. Certainly, other factors could
also have an effect on ozone transfer efficiency, particularly where conditions
differ from those examined in this study.
Figure 7 shows decreasing transfer efficiency with increasing gas
loading rate (data for one and both columns together). Considerable scatter
in the data points is apparent. Some of this scatter can be accounted for
by considering other factors affecting efficiency, as illustrated in Figure 8.
A similar type of plot is shown for the second contact column in Figure 9.
Analysis of the dissolved oxygen concentration in the effluent from the system
indicated high values as expected when oxygen is used as the feed gas for the
ozonator.
288
-------
Table 4. Summary of Regression Analyses for Ozone Utilization
Efficiency of Ozone Contact Columns with Screened Wastewater
CASE
1 Column
2 Columns
1 & 2
Columns
Column 2(2)
Column 2
STEP
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
REGRESSION COEFFICIENTS TOR
Q8/v
-n.O
-3.2
-4.5
-1.7
-J.7
-D.3
-4.3
-J.5
-4.8
-1.1
-4.8
-5.0
-o.:>
Ca
-0.0018
-0.0023
-0.0011
-u.0014
-u.0013
-0.0017
-0.0053
BOD
0.00039
0.00039
0.00038
0.00030
CriU>
-0.35
-u.26
-0.27
r2
0.554
0.666
0.724
0.603
0.741
0.798
0.554
0.692
0.749
0.121 0)
0.188 <3>
0.543
0.759
0.775
n
152
152
152
152
152
152
304
304
304
152
152
—
146
146
146
(1) Cri , ozone residual of column influent, only significant for Column 2
Cre , ozone residual of column effluent, not significant for any case
(2) Cri not included in analysis
(3) Poor correlation
100
95
90
I 85
80
75
70
Dor
e
0
A
4
^
a points
'"column's''-1
I
i
1
2
2
2
^w^e
30 3
II 4
3 8
303
II 4
3 e
Ozone
applied
16-34
47-98
143-248
34-51
9 2 - 14 3
27 I - 41 2
Wostewaw cforocrerisTics
BOD5
V5S
Tem0
PH
Average
95 mg/L
54 m'/L
Q 2'C
78
Range
4O - 222 mg/L
24 - 103 mg/L
55- II 5°C
71-81
0.01 0.02 0.03 0.04 0.05
Gas loading rate — QG/V* miri"'
* QG 03-02 gas flow rate at STP (25°C and 760mm Hg)
V = volume of column(s) (90 L for one column ; ISO L for two columns)
Figure 7- Ozone Utilization in Screened Wastewater
289
-------
100
s?
< 95
\
o
o
o
o
i 90
>.
I
o
o 85
c
o
o
! 80
o
o
75
70
Cu/Cfl - 1.0 -4.83 QG/V - 0.00173 CA + 0.00038 BOD
100 mg/L
ied
10 mg/L
ISO
90
90 L
Oxygen - o;
0 0.01 0.02 0.03 0.04 0.05
Gos loading rafe QG/V, min"'
Figure 8. Factors Affecting Ozone Utilization with
One or Two Contact Columns
100
CU/CA= I 0-027CR|-I65QG/V + 0.00030 BOD
70
Figure 9. Factors Affecting Ozone Utilization with
the Second Contact Column
290
-------
CONCLUSIONS
Based on the ozone pilot plant studies, the following conclusions are
applicable for screened, dilute wastewater, unless otherwise noted:
1. The principal factor influencing indicator organism survival was the
amount of ozone utilized in the wastewater.
2. A log-log relationship existed between bacterial survival and ozone
utilized. After transformation from the log to the power form of the
equation, the relationship was expressed as:
N/NO = (cu/cuo)b
3. Different indicator organisms demonstrated different degrees of sensi-
tivity to ozone. To achieve a three-log reduction (10-3 survival
ratio) of the indicator organisms, the following effluent numbers were
reached at the noted ozone utilization levels:
FC = 1,200 per 100 mL at Cu = 20 mg/L
TC = 5,800 per 100 mL at Cu = 31 mg/L
FS = 130 per 100 mL at Cu = 33 mg/L
SPC = 180 per mL at Cu = 40 mg/L
4. Adverse wastewater characteristics (high BOD, turbidity, etc.) adversely
influenced the effectiveness of the ozone disinfection system.
5. Other factors sometimes influencing bacterial survival in conjunction
with ozone disinfection appeared to be wastewater temperature and
ozone residual. Disinfection efficiency usually improved with decreasing
wastewater temperature and increasing ozone residual.
6. The predictive equations that were developed for ozone disinfection of
screened wastewater may not be directly applicable to other types of
wastewater. Nevertheless, it is hypothesized that the general approach
to data analysis presented in this paper would be applicable to other
systems with different wastewater characteristics. This could lead to
improved understanding of relevant factors influencing ozone disinfection
and to improved ozone system operations.
7. It was demonstrated that effluent fecal coliform numbers can be predicted
quite reliably with the assistance of on-line monitoring of certain
variables, for example:
FC (per 100 mL) = W5 •3 Cu~2'° Cre~0-46 (r2 = 0.86)
8. Ozonation of RBC secondary effluent was much more effective than
ozonation of screened wastewater. The 0.7 mg/L initial ozone utilized
(approximate value from the extrapolated curve) was less than half
that for screened wastewater. Also, a three-log reduction of FC
291
-------
(from 65,000 to 65 per 100 mL) was achieved at approximately 4 mg/L of
ozone utilized. This amount of ozone was 20% of that for the screened
wastewater at the same three-log reduction. The fact that the KBC
effluent initially was of superior bacterial quality (lower FC)
further served to accentuate the difference in effluent FC numbers
after ozone disinfection.
9. BOD and suspended solids reductions of approximately 25% were achieved
at an ozone utilization of 30 mg/L.
10. The ozone transfer or utilization efficiency with one contact column
and with the overall system appeared to be influenced by the ozone-
oxygen gas flowrate (Qg or Qg/V), the amount of ozone applied (Ca),
and the wastewater strength (BOD).
11. The ozone transfer efficiency of the second contactor in series
appeared to be influenced by the ozone residual in the wastewater from
the first column (Crl), the gas flowrate (Qg or Qg/V), and the x^aste-
water strength (BOD).
ACKNOWLEDGEMENTS
This project was funded by the Northern Technology Unit, Environmental
Protection Service (EPS), Environment Canada, Edmonton, Alberta.
The cooperation of the City of Whitehorse and the EPS Yukon District
Office are gratefully acknowledged.
In addition, a number of individuals provided technical support and
diligent work on various aspects of the project. These individuals include
J. Bell, J. Beaudoin, B. Burns, J. Dagg, D. Tilden, R. Orr and J. Vanderpost
Their efforts are highly appreciated.
REFERENCES
1. Am. Public Health Assoc,, 1976, Standard Methods for the Examination
of Water and Wastewater, 14th ed., Am. Public Health Assoc., Inc.,
Washington, D.C.
2. Given, P.W. and D.W, Smith, 1979. Disinfection of Dilute, Low Temper-
ature Wastewater Using Ozone. Ozone: Science and Engineering, 1_,
91-106,
3. Smith, D.W, and P.W, Given. 1981. Treatment Alternatives for Dilute,
Low-Temperature Wastewater. Design of Water and Wastewater Services
for Cold Climate Communities. Pergamon Press, Toronto, 165-179.
292
-------
11. OZONE-MASS TRANSFER COEFFICIENTS FOR BUBBLE DIFFUSER
Edward J. Opatken
Wastewater Research Division
Municipal Environmental Research Laboratory
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
ABSTRACT
A pilot plant project was conducted at the US EPA Test & Evaluation
Facility (T&E) to evaluate ozone as a disinfectant for wastewater treatment.
Various contacting devices were compared and the bubble diffuser proved to
be the most cost effective contactor (U). The data generated in conducting
the investigation on the effectiveness of the various contactors were used
to calculate the overall gas mass transfer coefficient for ozone. The cal-
culation consisted of dividing the overall ozone mass transfer rate by the
ozone gas concentration driving force across the bubble contactor. This
paper details the method that was used to calculate the overall ozone mass
transfer coefficient and the effect that gas flow rates had on the ozone
mass transfer coefficient. The paper also presents results that show the
enhancement of the mass transfer coefficient by the impurities in wastewater
MATERIALS AND METHODS
Secondary Effluent
Secondary effluents were trucked into the pilot plant from six secondary
treatment facilities located within a 32 km radius of the pilot plant. The
influent flow to the diffuser was normally controlled at 75 &/min, although
experimental runs were made at various liquid flow rates to determine the
effect of liquid flow on gas mass transfer coefficients.
Ozone Generation
The ozone generator had a maximum capacity of 10 kg/d using oxygen as
the feed gas. The concentration of ozone was controlled by power and flow
to obtain the dose specified for the experiments. The oxygen pressure was
reduced to 60 kPa before entry into the ozonator where two air cooled units,
that contained six plates per unit, were used to convert the oxygen to
ozone. The power was varied between 200 and 1500 watts to obtain the ozone
concentration specified for an experimental run. An ultraviolet absorption
analyzer (Dasibi Environmental Corp., Glendale, California) was used to
monitor the ozone concentration continuously and to establish steady state
conditions.
Bubble Diffuser Contactor
The bubble diffuser contactor (Figure 1) consisted of three aluminum
columns, each 3-7 m high and 300 mm in diameter, connected in series by PVC
piping. The three columns were arranged in steps so that secondary effluent
could flow by gravity from the first through the third column. The ozone
293
-------
gas stream was split and the flow controlled by rotameters leading to each
of the three columns. The ozone enriched gas was injected through a domed
ceramic diffuser (Norton Chemical Process Products Division) at the bottom
of each column. The flow was generally split with 50 percent of the gas
being fed to the first column and 25 percent each to the second and third
columns. The liquid residence time was approximately 3 minutes in each con-
tactor for a total contact time of 9 minutes.
Sampling
Gaseous ozone concentrations were determined at the inlet and outlet
from the diffuser columns using the lodometric method of Birdsall, Jenkins,
and Spadinger (2). A wet test meter was used to measure gas volumes. The
effluent leaving each column was measured for residual ozone using the
Amperornetric Back Titration Method.
RESULTS AND DISCUSSION
Ozone Transfer
The ozone transferred from the gas stream to the effluent was deter-
mined from a mass balance on gaseous ozone, as shown in Figure 2. The ozone
balance on the liquid stream includes a reaction factor to equate with the
ozone balance on the gas stream because the absorbed ozone can either react
with substances in the liquid or undergo decay. This assumption is shown as
AK.
Mass balance equations are given below:
N - G-|(y-|-y2) = L2 (x-|-x2) + AK
M = Ozone transferred, mg/min
y-| = Inlet ozone concentration, mg/&
y^ = Outlet ozone concentration, mg/&
GI to G^ Gas flow (Oxygen), &/min
x-| = Outlet residual ozone concentration, mg/&
x2 - Inlet residual ozone concentration, rng/&
1'1 L2 Liquid flow, 5,/mm
AK Ozone consumed, or decayed mg/min
The ozone transferred can readily be calculated by the mass balance in the
gas phase [i.e., G-j (y^-y^) ]. An ozone mass balance in the liquid phase can
be used to determine the quantity of ozone that reacted with the constituents
in the influent and/or underwent decay, AK.
294
-------
Experiments were conducted at various gas flow rates, G, to obtain the
effect of gas velocity on mass transfer coefficients. The ozone concentra-
tion was also varied to obtain various ozone dosages at a specific gas flow
rate. High ozone dosages were essential to calculate the gas mass transfer
coefficients accurately, because a low off-gas concentration may indicate
that the column is taller than required.
Gas Mass Transfer Coefficient Calculations
The method used to calculate mass transfer coefficients is summarized
in Figure 3•
The ozone transferred is equal to
N = G (y1-y2)
This in turn is equal to the product of the overall gas mass transfer
coefficient (K^Va) and the log mean concentration difference between the
inlet and outlet gas, (AClm). The overall ozone mass transfer coefficient
is a measure of the rate at which a contactor can transfer ozone from the
gas phase into the liquid phase. The log mean concentration difference is
the driving force necessary to transfer the ozone from the gas phase into
the liquid phase.
N
G (yi-y2) = (KgVa)(AClm)
The log mean concentration difference is calculated from the following
equation, which uses the inlet and outlet conditions of both the gas and
liquid phases.
AClm = Aylm = (yi-yi*) - (y2-y2*)/£n
Where
y-|s = HX-J = equilibrium concentration at x-| (bottom of the column)
and H is Henry's constant in (mg OgX&gVdng
The value for Henry's constants was obtained by converting the pub-
lished values in the International Critical Tables (ICT) into units that are
conducive for calculating overall gas mass transfer coefficients.
H20 °C = 2.86 x 10^ mm Hg/mol fraction of ozone = 2.63 mg
[To convert H into units that can be readily used, assume an ozone residual
and multiply by H obtained in the ICT at a specific temperature to obtain
the partial pressure (pp) of ozone in mm-Hg. Divide by 760 mrn-Hg (1 atm)
to obtain gaseous mol fraction. Convert mol fraction of ozone into mg O^X&g.
295
-------
Divide this value by the assumed ozone residual to obtain the H value in
(mg Oo/£g)/(mg 03/&]_j_q) at the specified temperature. Repeat at various
temperatures to obtain the curve shown in Figure 4.]
With the values of H obtained at several temperatures, a curve can be con-
structed covering the temperature range (2 to 40 °C} normally encountered at
wastewater treatment plants.
Example of Gas Mass Transfer Calculation
Below is an example showing the method employed to calculate the ozone mass
transfer coefficient from the data obtained at the T&E facility, using
effluent from the Loveland, Ohio wastewater treatment plant.
The ozone transferred to the liquid from the gas in columns A, B and C were:
Na Ga(yia-y2a) = 16-6 (27-9-2.1) - 428 mg 03/min
Nb Gb(yib-y2b) = 8-3 (27.9-6.0) 182 mg 03/min
Nc Gc(yic-y2c) = 8-3 (27.9-7.1) = 173 mg 03/min
The equilibrium partial pressure of ozone is calculated at the bottom of
column A in terms of mg 0^/lg rather than mm Hg. The ozone residual was 2.3
mg/2,liq. At 16.5 °C the H value is 2.2 (mg 03/£g)/(mg 03/&liq), which is
multiplied by the residual, 2.3 mg/Jlj_iq, to obtain 5.1 mg/&]_iq as the equi-
librium gas concentration of ozone at the bottom of the column. However, the
values of both the incoming ozone gas and the equilibrium gas concentration at
the bottom of the column need to be corrected because of the increased pressure
caused by the water head of 3-2 m. The incoming ozone gas concentration, in
reality the ozone partial pressure, is multiplied by the pressure correction
factor of (10.4 + 3.2)710.4 - 1.31. The inlet ozone partial pressure, or the
effective ozone concentration, is 1.31 x 27-9 or 36.5 mg Oo/fcg. The equilib-
rium ozone partial pressure for the ozone residual at the bottom of the column
(2.3 mg/£j_iq) is 5.1 mg/&g. This value also requires correction because of
the increased pressure at the bottom of the column. However, in this case the
correction factor is less than one because the partial pressure of ozone is
determined by the ozone residual. The ratio for the equilibrium concentration
is 10.47(10.4 + 3-2) = 0.765. The corrected equilibrium ozone concentration
y-l*c is 0.765 x 5-1 = 3-9 rag/Jig. The concentration or partial pressure
driving force at the bottom of column A is y-|-yic* = 36.5 - 3-9 = 32.6 mg/5-g.
At the top of the column the liquid influent has no residual ozone so the
driving force is only the concentration of the effluent gas, 2.1 mg/£g.
Since the top of the column is at ambient pressure there is no correction
factor applied to the ozone concentration leaving the column. The ozone con-
centration driving force over the entire column is the log mean concen-
tration difference of the two end conditions.
296
-------
AC- = Ay = ! _ ^ !i £
1m 1m Ay v -v
, y1 , Ylc Ylc*
In In
Ay2 y2-o
= 32.6-2.1 _ 30.5 _ 11.1 mg 0 /£g
In 32.6 ln 15_5
2. 1
and finally the ozone mass transfer coefficient is obtained by
N (KgVa) (Aylm)
KgVa = N = 428 mg Og/min = 38.6 mg O^/min/unit concentration difference
Aylm 11.1 mg 037£g
The calculation for columns B and C follow closely the method employed
for column A. The only difference is that the liquid influent to columns B
and C contain an ozone residual that must be taken into consideration at
the top of the column to obtain the ozone concentration driving force, or
partial pressure difference. The K^Va for column B was 20.8 mg Oo/min/unit
concentration difference and for column C the K^Va was 17.0 mg Oo/min/unit
D —'
concentration difference.
Mass Transfer Coefficients for Five Plants
The major objective of the pilot plant study with the bubble diffuser
was to establish the ozone requirements needed to disinfect secondary ef-
fluent from six treatment plants. During these experimental runs, data were
obtained that enabled the ozone mass transfer coefficients to be calculated.
The data used to determine the mass transfer coefficients included only
those results in which the effluent gas, y^, was greater than 0.7 mg/£g. The
Dasibi analyzer failed to show a measurable change when the effluent gas
concentration was below 0.7 mg/&g.
The mass transfer coefficients obtained for the five plants are shown
in Table 1.
297
-------
TABLE 1. Mass Transfer Coefficients at Column A
from Five Plants at a G of 17 &/min
Plants KcrVa n a
Indian Creek
Muddy Creek
Loveland
Sycamore
Fair field
Mill Creek
44
44
41
39
39
__
5
5
8
15
10
0
1
1
2
3
3
.5
-3
.3
.0
.7
-
Mean 40 43 2.1
The mean for the 43 samples was 40 rag O^/min/unit concentration differ-
ence. It should be noted that the mass transfer coefficient at Mill Creek
could not be calculated because the highest y^ value measured in 25 runs
was equal to or less than 0.1 mg/fcg.
Effect of Gas Flow on Mass Transfer Coefficients
Experimental runs were conducted with Indian Creek and Muddy Creek
secondary effluents to obtain mass transfer coefficients at various gas
flows. A plot of these data is shown in Figure 6. The rate of increase
declined as the gas flow increased and it appears that very little, if any,
improvement in mass transfer coefficient can be expected at gas flows above
30 5,/rain.
Enhancement of Mass Transfer by Secondary Effluents
The bubble diffuser was operated on tap water to obtain mass transfer
coefficients that were not influenced by ozone demanding material. Figure 6
shows the rate of increase was considerably less than that for secondary
effluent. The mass transfer coefficient leveled off at gas flow rates above
25 &/min. The difference in the mass transfer coefficients between tap water
and secondary effluent can be attributable to the ozone demand of the sec-
ondary effluent, which resulted in a signficantly higher mass transfer co-
efficient.
Enhancagent in Partially Satisfied Effluents
The bubble diffuser employed at the T&E Facility consisted of three
columns. The effluent exited at the bottom of the third column. The ozone
298
-------
enriched gas was split among the three columns, with 50% of the gas being
fed to column A and 25% of the gas being fed to columns B and C. The mass
transfer coefficient in column B was 3^ at a gas flow rate of 16 5,/min and
29 in column C as compared to 40 in column A at a gas flow of 17 £/min
(Table 1).
A summary of these results is shown in Table 2.
TABLE 2. Mass Transfer Coefficients for Columns B and C
Column B
Plant KgVa (n) KgVa (n)
@ 8.3 Jl/min (5 16 Jl/min
Sycamore
Fairfield
Love land
Muddy Creek
Indian Creek
Mean
Mill Creek
19
19
21
19
20
19
—
(15) 28
( 9)
( 3) 38
( 3)
( 1) 38
(31) 34
( 0)
(2)
(0)
(2)
(0)
(1)
(5)
Column C
KgVa (n)
@ 8.3 fc/min
19
19
20
17
20
19
26
(15)
(12)
( 5)
( 3)
( 2)
(37)
( 4)
KgVa
@ 16 8,/n
24
33
32
35
28
29
(n)
lin
(2)
(1)
(2)
(1)
(1)
(7)
A series of experimental runs was conducted specifically for obtaining
mass transfer coefficients in columns B and C at gas flow rates other than
the flow rates used in the disinfection study. Column A was operated at
the normal 17 2,/min to repeat the partial ozone satisfaction that prevailed
during the disinfection experimental runs. The mass transfer coefficients
for columns B and C were plotted against gas flow and again the coefficients
increased, up to a plateau region, with increasing gas flow (Figure 7).
When the coefficients are displayed with the coefficients from column A for
secondary effluent and for tap water (Figure 8), the curve for column B
lies under column A and the curve for column C lies slightly above tap
water. These relative positions for the mass transfer coefficient for
columns B and C, where they lie between column A and tap water, provide
further evidence on the enhancement of the mass transfer coefficient by
secondary effluent when compared with tap water.
Effect of Liquid Flow Rate
The liquid flow rate for secondary effluents and tap water was varied
from 56 to 94 l/min to determine the effect on the mass transfer coefficient
299
-------
Thp results displayed in Figure 9, showed that the mass transfer coefficients
increased with increasing liquid rates, up to a plateau region, and that
secondary effluent showed a similar enhancement in the mass transfer coeffi-
cient when compared with tap water, These results were expected since
increasing the liquid flow rate increased the system's capacity to absorb
additional ozone and thus improve ozone mass transfer.
Summary
The results from this study indicate that the secondary effluents from
5 of the 6 plants gave similar mass transfer coefficients. The mass transfer
coefficients for ozone that were obtained at these plants can be used for
the design and scale-up of ozone bubble diffuser contactors at plants where
the primary source of the wastewater is of domestic origin. The coefficients
should not be used where the source is primarily of industrial origin. The
high absorptive capacity obtained from the Mill Creek effluent indicates
that the enhancement of the transfer coefficient is considerably greater
than the coefficient obtained from effluents of domestic origin.
This paper presents a method for calculating ozone mass transfer coeffi-
cients using steady state conditions in a bubble diffuser contactor.
Although the calculations are tedious, the mass transfer coefficients provide
the designer with the necessary tools to determine the effect of variables
on ozone transfer, such as the effect of increasing the height of the con-
tactor, or operating the contactor under pressure. The mass transfer co-
efficient allows the designer to optimize the bubble diffuser contactor
without relying upon pilot plant operational data to determine contactor
sizing and performance.
The relationship on the ozone mass transfer coefficients with increases
in either gas or liquid rates shows a levelling off of the increase in the
mass transfer coefficient as the gas or liquid flow rate increases. The
designer can now employ these results to determine the rate of transfer at
various flow rates and calculate the efficiency of ozone transfer at any
specific rate.
Finally, the paper presents evidence on the enhancement of the ozone
mass transfer coefficient by secondary effluent when compared with tap
water, and the reduction in the enhancement when the effluent has undergone
partial ozone demand satisfaction.
LITERATURE CITED
1. American Puolic Health Association Inc.
Examination of Water and Wastewater.
1975. Standard Method for the
14th ed., APHA, Washington, DC.
2. Eirdsall, C. M,, A. C. Jenkins, and E. Spadinger. 1952. "lodometric
Determination of Ozone," Anal. Chem. 24, 662.
300
-------
3. U.S. Environmental Protection Agency. 1975. "Methods for Chemical
Analysis of Wastes," EPA-625/6-7-003, Methods Development and Quality
Assurance Research Laboratory, Cincinnati, Ohio.
4. Venosa, A. D., M. C. Meckes, E. J. Opatken, and J. W. Evans. 1979.
"Comparative Efficiencies of Ozone Utilization and Microorganism
Reduction in Different Ozone Contactors," in Progress _in_ Wastewater
Disinfection Technology, Proc. Nat. Symp., Sept 18-20, A. D. Venosa,
ed., EPA-600/9-79-018, U.S. Environmental Protection Agency,
Cincinnati, Ohio. 144-162.
301
-------
GAS SAMPLE
OJ
o
WASTEWATER
GAS
ww
SAMPLE
O3
GAS
SAMPLE
T
V
A
V
ww
SAMPLE
*- O3 OUT
GAS
SAMPLE
t
03
-ww
WW DRAIN
SAMPLE
FIGURE 1= BUBBLE DIFFUSER OZONE CONTACTOR
-------
G2,y2
1 X
L2' 2
AK
Gl'yi
Figure 2
Ozone Mass Balance
303
-------
L,x
304
-------
O)
O)
E
O)
E
n
j
II
c
to
c
O
u
CO
c
0)
0
<-*—
•u
O
2.0-
10
20
30
40
Temperature (°C)
Figure 4. Effect of Temperature on Henry's Constant
305
-------
LO
O
G=17 S/min
Y2=2.1 mg/
G=8.3 I/m
mg
B
X, =2.3 mg/l
=2.5 mg/i
Y, =27.9 mg/l
1
X1=2.6 mg/l
B "^
9
Figure 5. Ozone balance using Loveland effluent
-------
60r-
60
U)
o
u
Z
O
u
Z
5
\
n
O
O
S
o
>
O)
40
20
10 20 30
GAS FLOW L/MIN
FIGURE 6
40
40
20
COLUMN B
COLUMN C
10 20 30
GAS FLOW L/MIN
FIGURE 7
40
Figures 6 and 7. Effect of Gas Flow on Mass Transfer Coefficients
-------
OJ
o
-OO
C
o
20
D
>
O)
Liquid Fiow
56 L/min
76 L/min
94 L/min
20
Liquid Fiow
56 L/min
76 L/min
94 L/min
Gas Flow, L/mm
Figure 8. Effect of Gas Flow on Mass Transfer Coefficients
-------
60r-
U)
o
vo
C
O
v
± 40
c
D
E
'E
CO
O
O)
E
**
o
>
O)
20
Column A
Water
20 40
Gas Flow l/min
60
Figure 9. Effect of Liquid Flow on Mass Transfer Coefficients
-------
12. INNOVATIONS IN THE ELECTROLYTIC GENERATION OF OZONE
Peter C. Roller
Teknekron. Inc.
Applied Research Engineering Division
Berkeley, California
ABSTRACT
Though it has been known for well over a century that ozone may be
generated through the electrolysis of aqueous electrolytes at inert anodes,
only recently has research uncovered conditions that allow the process to be
considered seriously as an alternative to conventional generation techniques.
Innovations in anode material and electrolyte selection have resulted in ac-
ceptable current efficiencies at temperatures compatible with the use of
energy-saving air-depolarized cathodes. Thus, in the overall process, feed
air is reduced to water, which replaces that anodically decomposed (into ozone
and oxygen). Advanced electrolytic ozonizers will be able to produce ozone
concentrations all the way up to the limits of safety. Energy consumption is
projected to be nearly equivalent to that of conventional air-fed corona dis-
charge ozonizers, and will be independent of ozone concentration desired. The
initial cost of electrolytic ozonizers may be substantially under that of
conventional corona discharge equipment in that neither air pretreatment nor
compression are required, D.C. power supplies are used, and that non-noble
metal electrode materials can merely be stacked between injection-molded fram-
ing. Further engineering development is required before the technology can be
commercialized.
1. DESCRIPTION OF THE PROCESS
Significant improvements to the electrochemical route for ozone genera-
tion have been demonstrated in recent research studies. Unlike conventional
ozonators in which predried and compressed air (or oxygen) is passed through a
high-frequency corona discharge, ozone is formed by electrochemical oxidation
of water.
Certain aqueous fluoroanion electrolytes have been discovered, from which
water may be oxidized to ozone at high current efficiency near room temperature
( 4-6 ). Advances in anode material selection also have contributed to making
near ambient electrolysis temperatures possible ( 7 ). In previous work,
attractive current efficiencies for ozone generation had only been observed at
very low electrolyte (or anode surface) temperatures. Operation at such tem-
peratures (-20 to ~60°C)S in addition to requiring costly refrigeration, dis-
allowed the use of reduction of oxygen as the corresponding cathodic process
during electrolysis, The kinetics of oxygen reduction from air become very
poor as temperature is decreased. Hydrogen evolution had been considered as
the only available cathodic process, even though theoretically an additional
1.23V of cell potential is required. All economic projections for electro-
lytic production of ozone were most unfavorable. The electrolytic process
310
-------
considered in this review (Figure 1) is composed of the following half-cell
reactions. At the anode:
3H20 — 03. + 6H+ + 6e~, V° = +1.51V
and parasitical ly
2H20 - 02 + 4H+ + 4e", V° = +1.23V
At present there is no direct evidence for the two-electron reaction:
At the cathode not:
but
2H+ + 2e~, V° = +2. 07V
2H+ + 2e~ -* H, V° = 0.0V
4H+ + 4e~ — 2H0, V° = +1.23V
The theoretical cell voltage for the production of ozone is 0.28V.
Nothing even close to this voltage is achieved in practice, because one must
suppress oxygen evolution by employing anode materials that have very high
oxygen overvoltages. Similarly, the oxygen reduction reaction, which has been
studied extensively in the development of fuel-cells, is notoriously slow.
Cell voltages on the order of 1.8-2.IV are anticipated.
Overall, the process becomes: 02(air) —• 0.,, and immediately certain
inherent advantages may be pointed out. The air feed to the reactor need not
be pretreated in any way. It need not be dried; in fact, slight humidification
may be desirable to suppress water loss from the electrolyte. Compression also
is unnecessary. Atmospheric C02 is rejected by the acidic electrolytes
selected. On the anodic side, no NO is produced, only a mixture of ozone,
oxygen and air serving as a carrier gas. Carrier gas (air fed to the electro-
lysis cells in excess of the stoichiometric requirements of the cathodes) is
used to dilute the ozone formed as it evolves from the cells. Otherwise, ozone
concentrations well over the explosion limit would be formed.
Here again an inherent advantage of electrolytic technology can be seen:
the generated concentration of ozone is decoupled from power consumption,
unlike in corona discharge technology. Ozone concentrations are determined
first by current efficiency (which recent experiments ( 4,5 ) have shown may be
obtained in the range of 30-50 percent) and second by the flow rate of diluent
gas. Ozone concentrations of even 10 percent will be available using electro-
lytic technology. Air-fed corona discharge ozonators normally produce 2 per-
cent ozone at best, many times at an energy efficiency lower than that found at
concentrations approximating 1 percent.
311
-------
2. HISTORICAL DEVELOPMENT AND RECENT RESULTS
Sines ozone itself was first discovered by electrolysis of su If uric acid
in 1840 ( 18 )„ approximately 25 publications have appeared dealing with its
electrolytic generation. The field has developed slowly because until re-
cently, the results have been uniformly discouraging and of academic interest
only.
Work on electrolytic ozone generation may be characterized by electrolyte
composition and by choice of anode material. The electrolyte must engage in no
reactions other than oxygen and ozone evolution at the anode, and hydrogen
evolution or oxygen reduction at the cathode. Chemical reactions with the
ozone produced also must not occur. Such constraints led to the selection of
acids of oxyanions and f luoroanions, as well as their alkali metal salts, as
the most suitable electrolytes.
Very few anode materials are inert to ozone evolution conditions. Ex-
tremely high interfacial acid concentrations are produced during the anodic
decomposition of water. High anodic potentials led to dissolution or passiva-
tion in the case of most metals. Platinum has been used commonly, and proves
to be sufficiently inert. Certain of its noble metal alloys have been used,
although their oxygen overvoltages are reduced. Conductive oxides in their
highest oxidation states have been used (e.g., the alpha and beta forms of Pb(L
and Sn02) and show promise. Pyrolytic carbons also prove to be inert in
certain electrolyte compositions.
The platinum/sulfuric acid anode and electrolyte composition has been the
subject of intense effort in two electrolysis regimes. Early authors used
narrow filaments of platinum to achieve current densities on the order of 50-
100 A/ cm ( 3,15 ) „ Current efficiencies (the fraction of ozone anodically
evolved vs. oxygen) of up to 27 percent were reported from 0 C electrolyte;
however cell voltages of nearly 15V were observed. A glow discharge mechanism
seems likely due to the high electric field encountered and the gas-blanketing
that must occur.
The second ozone generation regime explored in the platinum/sulfuric acid
combination was the electrolysis of eutectic electrolyte compositions at the
lowest temperatures possible ( 2,19 ). Current efficiencies of up to 32 per-
cent were reported; however, refrigeration costs (calculated as 1/3 to 1/2 of
the energy consumed during the electrolysis itself) eliminated commercial con-
sideration of the technology.
The platinum anode/perchloric acid combination was studied extensively in
this same regime; however, maximum current efficiencies of 36 percent at -40°C
still were inadequate for scale-up ( 1,13,17 ).
A major advance in electrolytic ozone generation came with the use of o
anodes by three different, groups of workers. Semchenko et al. first electro-
lyzed phosphoric acid and found that yields of 13 percent current efficiency
can be obtained at temperatures of 10-15°C ( 20 ).
312
-------
Semchenko and co-workers next studied the use of perchloric acid, finding
yields of 32 percent current efficiency at temperatures of -15°C ( 21,22 ). In
conjunction with the use of PbCL anodes, a small quantity of fluoride ion was
added to the electrolyte with trie apparent effect of raising anode potential
(and therefore ozone current efficiency). As of 1975 these were the most
encouraging results yet obtained. However, with Pb02 anodes, some erosion is
observed during ozone evolution, following a combinecrchemical/electrochemical
mechanism advanced by Roller and Tobias ( 8 ).
Fritz et al. ( 10 ) continued the characterization of phosphoric acid-
based electrolyte systems, notably a neutrally buffered system in which PbO?
erosion is suppressed. Yields of 13 percent current efficiency were obtainea
at ambient temperatures.
Foller and Tobias ( 5 ) studied the use of fluoroanion electrolytes, and
continued to find yields using Pb02 anodes much greater than those obtained
with platinum electrodes. Further, n't was found that the electrolytes HBF. and
HPFg were particularly well suited to ozone evolution.
Figure 2 illustrates the current efficiencies obtained during the elec-
trolysis of various concentrations of HPFfi with beta-PbO? anodes at 0°C. Al-
though the circumstances of this electrolysis (rapid weight loss and high PFr
vapor pressure) are not compatible with commercial development, these experi-
ments illustrate that high current efficiencies for ozone generation may in-
deed be obtained. The research and development problem is to find alternative
conditions in which to run the oxidation of water so effectively.
The platinum anode was found to give very high ozone yields in HPFg as
well, which led Foller and Tobias to propose a rationale for electrolyte
selection based on anion electronegativity. Electrolyte anion adsorption on
anode materials also was found to correlate with ozone current efficiency
( 9,16 ).
Foller et al. ( 7 ) found that a certain form of carbon, known as glassy
carbon, also was capable of producing relatively high ozone current effi-
ciencies at temperatures above 0°C in fluoroanion electrolytes. Under condi-
tions ordinarily corresponding to ozone evolution, pressed carbon blacks (high-
surface-area carbons) rapidly degrade, exhibiting CCL evolution and structural
disintegration. Graphite also undergoes disintegration due to anion inter-
calation between its planes and consequent c-axis swelling.
Glassy carbon is much more resistant to oxidative processes and anion
penetration due to its random, yet fully coordinated structure. This form of
carbon is made by heat-treating certain resins under controlled inert atmos-
phere conditions. Attack is observed in oxyanion electrolytes and in low
concentration acids of the fluoroanions, however not at all in high concentra-
tion electrolytes. The phenomenon is as curious as it is fortuitous, in that
ozone yields reach their maximum at the highest concentrations of fluoranion
acid electrolytes.
313
-------
Figure 3 presents ozone current efficiencies as a function of current
density for the electrolysis of various concentrations of tetrafluoboric acid
electrolyte with glassy carbon anodes at 0 C. The highest yields are found at
the highest acid concentration commercially available (48 wt percent).
Figure 4 shows that, these yields are stable over the periods of time investi-
gated to date. No detectable weight loss is observed over 24 hours of accumu-
lated running time in acid concentrations higher than 5 M,
There is a certain amount of confusion over ex-situ versus in-situ elec-
trolytic ozone generation methods (when considering water treatment applica-
tions). What has been discussed to this point centers purely on gaseous ozone
generation irrespective of contacting and end-use. Methods have been ad-
vanced, however, that propose in-situ ozone generation as an explanation of the
efficacy of noble metal electrolysis as a treatment of potable water streams
containing the chloride ion at levels on the order of hundreds of parts-per-
million ( 23 )» Extraordinarily high voltages must be applied to pass minimal
currents (due to poor solution conductivity). Actual anode potentials (in-
dependent of solution I-R) sufficient to oxidize chloride ion to chlorine
(1.34V) and hypochlorite are achieved. These then, in conjunction with the
adsorption and oxidation of organic substances on the electrodes themselves
account for the levels of water sterilization observed.
From studies of ex-situ electrolytic ozone generation, it is clear that
levels of ozone production in dilute electrolytes are quite small, and indeed
may be attributable to analytical difficulties in separating the effects of the
other chlorooxidants, which most certainly are produced. In any event, elec-
trolysis at such high voltages (no matter what the assumed reaction products or
current efficiencies) cannot be economic in comparison to ex-situ optimized
ozone (or chlorine) generation processes.
3. PROJECTED COSTS
A accurate detailed cost estimate of electrolytic ozone generation tech-
nology is iiot yet possible. Projections, however, can be mades assuming that
certain development milestones will be reached. Projections such as the fol-
lowing demonstrate why interest in electrolytic technology remains high.
3_.JL Oge r at ij'. t^ Co s t
The operating cost of an electrolytic ozonator is almost entirely deter-
mined by the power consumption of the electrolysis cells. This power consump-
tion may be derived from the current efficiency and cell voltage. Figure 5 is
a plot of the aoiuiirit of ozone produced per direct current (dc) kilowatt-hour as
a function oi various current efficiency levels and cell voltages. Two regions
of operation are indicated,, which correspond to projected cell voltages for
either oxygen reduction or hydrogen evolution as the cathodic process. The
ranges of ceil voltage chosen as representative xjf the two process configura-
tions correspond to operator! at 0.35-0.40 A/cm (near the maximum of ozone
314
-------
current efficiency, but at the same time avoiding the higher levels of polari-
zation at higher current density). An anode potential of 2.2-2.4V vs. a
standard hydrogen electrode (SHE), and an air-cathode potential of 0.55-0.65V
vs. SHE were chosen for the purposes of this comparison. Electrolyte conducti-
vity and a projected interelectrode gap of 5 mm also were included in the
calculations.
Several current efficiency levels are indicated in Figure 5, which then
may be used to determine power consumption. Horizontal lines on the figure
indicate the power consumptions of conventional corona discharge ozonators. A
fairly broad range is defined when the power consumption of all auxiliaries
such as air drying and compression are added in, considering the entire spec-
trum of capacities commercially offered.
The energy efficiency of ozone production at a cell voltage of 2.0V (anode:
2.4V, cathode: 0.6V, heat disippation (I-R loss): 0.2V), and a current effi-
ciency of 50 percent exceeds that of the best air-fed corona discharge ozona-
tors. Similarly, a current efficiency of only 17 percent at 2.0V is required
to undercut the energy consumption of some of the smaller air-fed units on the
market today.
Projection of just where advanced electrolytic ozonators will lie within
this range of energy consumption when fully optimized is problematical. The
2.4V anode potential and 50 percent current efficiency necessary to develop a
75g/kWh ozone electrolyzer have been demonstrated with platinum anodes at
temperatures compatible with the flow of cooling water. It is possible to
achieve these performance levels under laboratory conditions. Stable current
efficiencies of 35-40 percent also have been achieved with the much less expen-
sive glassy carbon anodes in a less volatile electrolyte (HBFJ, however, at
somewhat higher anode potentials.
The optimization of energy efficiency in a commercially practical cell
design will include the selection of a current density (the trade-off is that
increasing current density increases current efficiency, but at the same time
increases electrode potentials, I-R losses and heat generation), selection of
an operating temperature (the trade-off is that increasing electrolyte tem-
perature decreases air-cathode polarization, and increases electrolyte conduc-
tivity, but at the same time diminishes ozone current efficiency), and selec-
tion of anode, air-cathode, and electrolyte compositions. It is very likely
that energy consumptions on the order of 45 to 50 g/AC kWh (95+ percent power
supply efficiencies are common) can be achieved with non-noble metal elec-
trodes and cooling water compatible anode temperatures.
Operating costs also will include maintenance. Both anodes and cathodes
probably will need replacement at certain intervals. Even platinum-clad
anodes probably will be subject to slow erosion. Glassy carbon anodes so-far
have appeared extremely stable during 12- to 24-hour testing. The air-cathodes
should exhibit lifetimes in excess of the 40,000 hours projected for high-
temperature (190 C) municipal power generation fuel cells. In these fuel
315
-------
cells, catalyst area loss through aggolmeration is a prime failure mode. At
ambient temperature,, longer lifetimes are expected, as migration is reduced.
Periodic electrolyte rebalance through water or acid addition may also prove
necessary,
3.2 Capital Costs
Electrolytic ozone generation should have initial cost advantages over
conventional air-fed corona discharge technology for three reasons. First,
the cell stack can be assembled from injection-molded polypropylene framework,
and non-noble metal electrodes. The power supply required is very unsophisti-
cated, a conventional dc source with minimal regulation. A 90-V, 3S500-A unit
for a 1,000-lb/day ozonator can be purchased for $19S000 (1981). High-
frequency and high-voltage power supplies for corona discharge ozonators are
much more expensive. Finally, contacting costs can be reduced as higher concen-
tration ozonizers can reduce contactor sizes and increase throughputs. How-
ever, mass transfer studies at the higher ozone concentrations available by
electrolysis must be conducted first, to prove this hypothesis.
The higher concentrations of ozone in air available by electrolysis imply
that a given quantity of ozone can be applied using a much lower volume of air.
This will provide savings, because of smaller gas-handling equipment.
The size and capital cost of electrolytic ozonators may readily be esti-
mated once some basic assumptions as to the progress of subsequent research are
made. Assuming that 40 percent current efficiency can be achieved at cooling
water temperature, and that a cell voltage of 2.0V will be encountered at
350 mA/cm , a 1,000-lb/day electrolytic ozonator may be sized.
A total current of 158,000 A is required. Therefore, if a 90-V power
supply is used, two parallel stacks of forty-five 1,750-A cells may be envi-
sioned. Each bi-cell would have an electrode area of 5,000 cm (50 x 100 cm)
and a thickness of approximately 3-4 cm, counting air and coolant flow provi-
sions. Thus cell stack dimensions of 1.5 x 1.5 x 2.0 m appear likely.
P
Costs may b$ calculated on the basis of anode material ($50/ft ), cathode
material ($20/ft"") and cell framing. A filter-press design seems most likely.
Electrolyte^, reservoirs, and auxiliaries such as monitoring equipment, air
blowers and filters also must be added in along with assembly, overhead costs,
and 40 percent mar-k-up. Figure 6 compares the projected cost of electrolytic
ozonators with the costs of conventional air-fed ozonators as determined in a
1979 study of the U.S. Municipal Environmental Research Laboratory ( 11 ). A
dramatic reduction in initial cost is forecast due to the basic simplicity of
electrolytic technology,, Whether this will be, in the end, the 75 percent
reduction exhaustively calculated in the preparation of Figure 6, or only a
50 percent reductons it is clear that significant advantages in cost are prom-
ised.
316
-------
The reduction of the capital cost of ozonators is extremely important in
that capital cost represents a very significant fraction of the total cost of
ozonation. Amortization of equipment costs can outweigh operating cost (power
consumption) for large installations. Figure 7, derived from the EPA-spon-
sored study of Gutmann and Clark ( 12 ), shows that even at 7 percent interest
rates and 20-year amortizations, the fraction of capital-related costs in the
total cost of ozonation (contacting costs included) rises rapidly. (An up-to-
date detailed analysis of ozone cost alone follows.)
Contactor costs may be reduced because mass transfer rates from the gas
phase to solution phase are inversely proportional to one minus mole fraction
of ozone in the gas phase ( 14 ). Therefore, at the higher ozone concentra-
tions that are produced by electrolytic technology (at no energy penalty),
contactor sizes might be reduced, and/or a greater volume of solution may be
treated per unit time. Such potential advantages of electrolytic technology
must be analyzed in greater detail with regard to specific applications of
ozone.
3.3 Total Amortized Cost of Ozone Produced
In order to more fully assess the economic impact of the development of an
advanced electrolytic ozonizer, the following analysis of the cost of ozone on
a per pound basis was performed.
Ozonizers of 1,000 Ib/day capacity were used as a basis. These were assumed to
have 20 year lives, and to be operated 24 hours per day 300 days a year.
Replacement of anodes and cathodes of the electrolytic cells was scheduled for
every five years. Twenty year financing at 15 percent interest, and main-
tenance cost of 5 percent the initial cost per year were assumed in each case.
An identical power consumption of 60g/kWh (7.5 kWh/lb) at a $0.04/kWh elec-
tricity cost was further assumed. An initial cost of $800K was taken for an
air-fed corona discharge ozonizer. An initial cost of $195K for the electro-
lytic ozonizer reflects a materials cost inclusive of applicable freight and
taxes with 25 percent contingency, labor overhead at 200 percent direct,
15 percent general and administrative costs, and a 40 percent profit.
The annualized costs are then computed as follows.
317
-------
Conventional Electrochemical
Cost Element Generator Generator
Capital Costs ($800,000) ($195,000)
Interest on debt $120,000 $ 29,300
Sinking fund for
debt retirement 7,800 1,900
Operating Costs
Maintenance 40,000 12,800
Electricity 90,900 90,900
Total Annualized Costs: $258,700 $134,000
Cost Per Pound of Ozone: 86.2£/lb 44.7
-------
such as Vulcan XC-72 bonded with approximately 20 wt percent Teflon-30 seems
likely. Techniques of cathode manufacture, however, are complex, and in many
instances proprietary.
The glassy carbon anodes also will have to be optimized for ozone evolu-
tion. Sensitivity to production methods, such as heat treatment temperature
and starting resin, has been noticed in ozone current efficiency data ( 7 ),
and to some extent accounts for differences in yield seen between Figures 3 and
4.
Most importantly, integrated cell testing to co-optimize operational tem-
perature and current efficiency for minimal power consumption must be per-
formed in practical cell designs. At this point, long-term testing of the
cells would be begun.
Mass transfer studies should be conducted, using ozone-air combinations
that contain higher concentrations of ozone, so that optimally sized ozone
contacting chambers can be designed.
In addition, higher concentrations of ozone in air should be tested for
compatibility with materials of construction. Higher concentrations of ozone
probably will result in shorter lifetime of certain components of ozone-handl-
ing equipment.
5. POSSIBLE APPLICATIONS
Electrolytic technology probably will find applications in certain spe-
cial-purpose fields well-suited to its particular characteristics in advance
of its full optimization. These may be applications in which high concentra-
tions of ozone are required (any concentration up to the limits of safety would
be available), or in which relatively small quantities of ozone are needed (say
0.2-1.0 Ib/day) at low initial cost. If 50 percent current efficiency at a
2.0-V cell voltage can be achieved in a commercial design at cooling water
temperature, electrolytic technology will, of course, find the widest possible
application.
Applications requiring very high concentrations of ozone are limited.
Many current large-scale applications should benefit from increased concentra-
tion during contacting, but would, at the same time, require a fully optimized
power consumption.
Hazardous waste treatment is a likely application, in that power cost is
not a central issue in disposal of certain highly toxic materials. The high
concentrations of ozone (previously unavailable) that electrolytic technology
can provide most certainly will improve the oxidation kinetics of organics. An
advantage of ozone in this field is that is is nonspecific; it can decompose
many unsaturated aliphatics and aromatic organics even when chlorinated.
Known pesticide, PCB, phenol, cyanide, surfactant, nitrocompound, dye waste,
319
-------
higher alcohol, and organophosphate decomposition process should be more rapid
at higher ozone concentrations.
Applications requiring very low initial cost may also be amenable to
unoptimized electrolytic technology. A low-maintenance, continuous-treatment
process for swimming pools may be devised, for example. The electrode area
required to treat a 20,000-gallon pool at 1-mg/l-day works out to less than
500 cm, and the power requirements lie in the range of 2-3 kWh/day (well below
filtration pumping costs).
6. CONCLUSIONS
Recent developments indicate that the electrochemical synthesis of ozone
may become an economically feasible alternative to corona discharge. Addi-
tional basic research is required, along with substantial engineering develop-
ment. However, the needed development centers on the available technologies of
fuel cell and water electrolyzer (products: Hp and 0?) design.
The possible outcome of continued efforts in electrolytic ozonator devel-
opment is that high-concentration ozonators of quite low cost may become avail-
able with power consumptions equal to those of the best air-fed corona dis-
charge technology. This new breed of ozone generators also may enable contact-
ing costs to be reduced. Further, the technology will scale-up and scale-down
with equal ease. Research in this field undoubtedly will continue.
7. REFERENCES
1. Boelter, E. D. PhD Dissertation, University of Washington (1952).
2. Briner, E., R. Haefeli, and H. Paillard. Helv. Chim. Acta, 20:1510-1523
(1937).
3. Fisher, F., and K. Massenez, Z. Anorg. Chem., 52:202-253 (1907).
4. Roller, P. C., PhD Dissertation, University of California, Berkeley
(1979). y
5. Foller, P. C., and C. W. Tobias. "The Anodic Evolution of Ozone," J.
ile£trochem._So£L, 129(3), (1982). —
6. Fo'ller, P. C., and C. W. Tobias, U.S. Patent Application #154,854.
7. Foller, P. C., M. L, Goodwin, and C. W. Tobias, U.S. Patent Application
#263,155.
320
-------
8. Roller, P. C., and C. W. Tobias. "The Mechanism of the Degradation of
Lead Dioxide Anodes under Conditions of Ozone Evolution in Strong Acid
Electrolytes," J. Electrochem. Soc., 129(3), (1982).
9. Roller, P. C., and C. W. Tobias. "The Effect of Electrolyte Anion
Adsorption on Current Efficiencies for the Evolution of Ozone," J. Phys.
Chem., 85(22):3238 (1981).
10. Fritz, H. P-, J. Thanos, and D. W. Wabner. Z. Naturforsch.,
34b:1617-1627 (1979).
11. Gumerman, R. C., R. L. Culp, and S. P. Hansen. Estimating Water
Treatment Costs, Vol. 2, U.S. Municipal Environmental Research
Laboratory, EPA-600/2-79-162b (1979).
12. Gutmann, D. L., and R. M. Clark. "Computer Cost Models for Potable
Water Treatment Plants," U.S. Municipal Environmental Research
Laboratory, EPA-600/2-78-181 (1978).
13. Lash, E. I., R. D. Hornbeck, G. L. Putnam, and E. D. Boelter. J.
Electrochem. Soc., 98(4):134-137 (1951).
14. McCabe, W. L., and J. C. Smith. Unit Operations of Chemical Engineering,
(New York: McGraw-Hill, 1976), p. 719.
15. McLeod. Chem. Soc. J., 49:591 (1886).
16. Potapova, N., A. Rakov, and V. Veselovskii. Elektrokhimiya,
5(11):1418-1420 (1969).
17- Putnam, G. L., R. W. Moulton, W. W. Fillmore, and L. Clark. J_._
Electrochem. Soc., 93(5):211-221 (1948).
18. Schonbein. Pogg. Ann., 50:616 (1840).
19. Seader, J. D., and C. W. Tobias. Ind. Eng. Chem., 44(9):2207-2211
(1952).
20. Semchenko, D. P., E. T. Lyubushkina, and V. Lyubushkin. Elektrokhimiya,
9(11):1744 (1973).
21. Semchenko, D. P., E. T. Lyubushkina, and V. Lyubushkin. Otkryitiya,
Izobret. Prom. Obraztsy. Tovarnye Znaki, 51(10):225 (19747!
22. Semchenko, D. P., E. T. Lyubushkina, and V. Lyubushkin. Izv. Sev.-Kauk.
Nauchn. Tsentra Vyssh. Shk. Ser. Tekn. Nauk, 3(1):98-100 (1975).
23. Wilk, I. J., Paper presented at the 157th National Meeting, American
Chemical Society, Minneapolis, MN, April 14-18, 1969.
321
-------
Air-Cathode
'Reduces Oxygen)
Ozone
Anode
Evolves Ozone and Oxygen)
Electrolyte
Figure 1. Schematic of ozonator cell design
322
-------
c
O)
(J
O)
S-
S-
3
O
O)
c
O
M
O
c
-------
c:
O)
01
S-
O)
c
O
IM
O
C
O)
O
s_
CD
Q_
Ozone Current Efficiencies Taken in
Ascending Steps of Current Density
GS V-10 Carbon Anodes, 0° C,
Electrolyte
20
10
1.0
Current Density (A/CNr)
Figure 3. Current efficiency of the glassy carbon/HBF. anode/electrolyte
combination as a function of current density and concentration at
0°C.
324
-------
35
30
c
OJ
u
c
(1)
S-
S-
3
o
0>
c
o
Nl
o
o
i-
0)
0.
25
20
0.6 A/CM^
0.4
0.2 A/CM"
Ozone Current Efficiencies Taken
at Single Current Densities
7.3 M HBF Electrolyte, 0° C
P.A.ff. Carbon Anode
30
60
Time, Minutes
90
120
Figure 4. Current efficiencies of the glassy carbon/HBF. anode/electrolyte
combination as a function of time at 0°C.
325
-------
140
130
120
110
100
90
70
50
40
30
70%
Current Efficiency
Power Consumption
Comparisons
60%
Air Depolarized CelIs
*—«*
50%
30%
20%
KL-Evolving CelIs
Conventional and SPE
Smal lest
1.5 1.7 1.9 2.1 2.3
Cell Voltage
2.5 2.7 2.9 3.1
Figure 5. Analysis of the power consumption of electrolytic ozonators,
326
-------
1000
500
Initial Cost of Conventional
Air-Feed Ozonizer
(EPA, 1979)
O
o-i
"O
100
50
Initial Cost of Proposed
Electorlytic Ozonizer
(1981 $)
10
J_
10
SO 100
Pounds of Ozone Per Day
1000
Figure 6. Comparison of capital costs (21).
-------
U)
NJ
CO
1,0
0.8
0.6
0.4
Fraction of Capital Cost in the Total
Expense of Ozonation of Potable Water
(20 Years, 7 Percent Interest)
0.2
Data of Guttman and Clark
EPA, 9/1978
1
10
50
100
Million of Gallons Per Day Treated
Figure 7. Fraction of capital cost in total expense of ozonation (22)
-------
1. PRACTICAL CONSIDERATIONS IN THE USE OF HALOGEN DISINFECTANTS
Charles N. Haas, Assistant Professor
Pritzker Department of Environmental Engineering
Illinois Institute of Technology
Chicago, IL 60616
ABSTRACT
The various issues to be faced when designing and operating wastewater
disinfection systems utilizing chlorine, hypochlorites, chlorine dioxide and
bromine chloride will be reviewed, and areas of continuing uncertainty will be
highlighted. These include dose estimation, contactor hydraulics, chlorine
process control systems, and mixing conditions at the point of application.
INTRODUCTION
Halogens have been employed as disinfectants of wastewater for at least
150 years, since Averill (2) reported "When it is desirable to destroy the ef-
fluvia from drains, sewers, etc., or to purify the water of a cistern—dissolve
about eight ounces of the chloride of lime in a pail full of water, and dis-
perse it into them. Repeat the operation until the object is effected."
Nevertheless, major issues relating to the design and operation of halogen
disinfection processes remain only partially understood. This paper will
review several of these as a preliminary step in the preparation of portions
of a design manual on the subject.
ISSUES IMPORTANT AT THE DESIGN STAGE
When a wastewater disinfection system is to be designed, numerous problems
present themselves, from those of dose estimation, to hydraulics, chemical
supply and safety. Rather th'an enumerating all possible issues, several
points of continuing uncertainty will be reviewed.
Chemical Dose Estimation
One of the basic questions in disinfection is how much of a given chemi-
cal is needed to attain a desired effluent standard. In wastewater, in parti-
cular, this question is complicated by the different upstream processes and
resulting inputs to a disinfection system, and by the existence of substan-
tial, competitive, demand-exerting reactions for all of the halogens employed.
While a number of references (18,35-38) present broad guidelines on the
dose requirements needed to disinfect wastewater using chlorine, information
on the analysis of microbial inactivation kinetics by halogens in wastewater
remains sparse. Only two authors have considered the estimation of such pro-
cess rates.
One model, of Selleck (24), which has been cited by other sources (18,35,
38), is of the form:
329
-------
N/NQ = (i + ct/b) (i)
In equation 1, c is the chlorine residual (generally, total residual) after a
contact time t, while a and b are empirical constants. In this model of batch
reactor wastewater chlorination kinetics, the empirical coefficients have been
shown to vary in a poorly understood manner with the degree of prior treatment,
and with the chlorine:ammonia-nitrogen dose ratio (24). The empirical Selleck
model has been verified by Roberts et al. (21) for modelling the inactivation
of coliforms in wastewater effluents by varying degrees of treatment using both
chlorine and chlorine dioxide.
There appear to be at least three major problems with the above dose esti-
mation procedure. The first, expressed by Roberts et al. (21), is that "...the
model...has no rational, mechanistic basis in describing disinfection by chem-
ical agents. Nonetheless, it does approximate empirically the behavior of the
real system and as such provides a useful design tool." The lack of theoreti-
cal justification for this model makes it difficult to incorporate knowledge
about contactor imperfections and mixing dynamics into the calculation proce-
dure.
As a corollary to the above, a second problem with the approach of
Selleck is the inability to extrapolate readily from kinetic parameters
obtained on one effluent to those of another effluent. For example, in the
studies by Roberts et al. (21), values for the a and b parameters were observed
to differ between the treatment plants examined in the case of chlorine and
chlorine dioxide, and the chlorine values differed from those reported by
Selleck (24) .
A third major problem with the Selleck approach is the need to estimate
the chlorine demand and thus to calculate the initial chlorine dose required.
Roberts et al. (21) have employed the empirical eauation initially developed by
Taras (33) to calculate the dose required for chlorine or chlorine dioxide dis-
infection in conjunction with the Selleck model. However, the Taras approach
to chlorine demand calculations appears to share some of the disadvantage of
the Selleck model in that it cannot readily be extrapolated to a different
wastewater.
A second approach to the problem of determining disinfectant doses is the
use of mass balance and reaction rate expressions for disinfection per se,
chlorine-demand reactions, and other simultaneous processes which might occur
(i.e., mixing of two fluid streams). The author (9) has described this process
elsewhere, for the particular case of wastewater chlorination, and has con-
trasted such mass balance models with experimental data. The major drawback
of this approach is the fact that the resulting mass balance models consist of
several simultaneous ordinary differential equations, which may be non-linear,
and thus are amenable only to numerical solution. In addition, the rate
constants for many of the chlorine-ammonia and chlorine-amine reactions are
not well characterized. The major advantage of this procedure, in principle,
is that the inherent sensitivities of microorganisms to various disinfectant
species and reaction rate constants with ammonia and amines might be expected
to remain relatively constant among various wastewaters.
330
-------
In the case of chlorine dioxide, the estimation of dose requirements for
wastewater disinfection remains clouded by lack of knowledge regarding the
chemical species responsible for chlorine dioxide demand. While White (38)
indicates that chlorine dioxide demand of wastewater should be greater than
that of chlorine demand, work of Roberts et al. (21) indicates that although
this is true in conventional secondary effluent, the chlorine demand in nitri-
fied filtered effluents may exceed the chlorine dioxide demand. The chemical
reactions leading to chlorine dioxide demand are not known.
Influence of Disinfection Pretreatment
A related issue to that of dose estimation is the effect of treatment prior
to disinfection. While efficient operation of secondary and tertiary treatment
can directly remove microorganisms from wastewater, and thus reduce the necessary
stringency of disinfection, several indirect effects upon this latter process
have also been uncovered.
It is well known that the degree of nitrification, if any, and the pre-
sence of ammonia, organic amines, and various reducing agents can affect the
efficiency of the chlorination process (37,38). Furthermore, the increase in
efficiency of disinfection by chlorine with reductions in pH has also been
reported (37,38).
More recent studies have indicated that the presence of certain cations
may affect chlorination efficiency, although the significance of these effects
in the field is unknown. For example, Kuzminski (16) , Reid and Carlson (20),
both working in laboratory demand-free systems, indicated that calcium concen-
trations could interfere with the chlorine inactivation of coliforms. In
other work, a number of studies (12,14,22,27,31) have indicated that sodium,
and perhaps potassium ions, can enhance the rate of inactivation of viruses as
well as coliforms in laboratory, demand-free systems, and that the formation
of a previously neglected ion-pair may explain this phenomenon (10,12). If
coagulants or neutralizing agents are added prior to disinfection by chlorine,
these indirect effects may be of significance, and may be amenable to manipula-
tion with the objective of chlorine dose minimization.
With respect to chlorine dioxide, virtually no information exists which
permits generalization regarding the effect of the surrounding menstruum on
disinfection efficiency. While a number of authors have indicated that in-
creasing pH increases the efficiency with which chlorine dioxide inactivates
microorganisms in laboratory studies (5,23), the mechanism of this effect, and
its applicability to full-scale wastewater treatment plants remain unknown. A
very recent paper indicates that, when applied as a potable water disinfectant,
chlorine dioxide inactivation efficiency also decreases as increasing amounts
of humic color material are present (7).
With respect to bromine, and presumably bromine chloride, increasing the
pH of a wastewater has been found to increase the efficiency of disinfection.
This effect, and its contrast to the behavior of chlorine, has been attributed
to the efficacy and stability of monobromamine, predominating at high pH, as
compared with dibromamine, predominant at low pH (30).
331
-------
A second type of pretreatment effect relates to the influence of prior
conditions upon the innate sensitivity of microorganisms to inactivation by
the halogens. These effects may relate to the selection of resistant strains
of microorganisms or the alteration of innate physiological conditions so as to
increase resistance. While the existence of these effects has rarely been
investigated at wastewater treatment plants, a variety of studies in labora-
tory systems or in potable water treatment plants have suggested that in
situ strains of microorganisms may be more resistant to chlorine than commonly
used laboratory strains (26), may develop altered resistance upon repeated
exposure and subculture (3,4,11), or that the antecedent growth conditions
may alter the sensitivity of coliform organisms to chlorine (19). In waste-
water, Aieta et al. (1) have shown that native populations of total coliform
organisms are more sensitive than pure cultures of E. coli exposed to chlorine
and chlorine dioxide under similar conditions; whether this is due to the
importance of resistance, encapsulated coliforms, or to an inherent or induced
strain resistance remains uncertain,
A peculiar example of this pretreatment effect appears to be emerging
with regard to the chlorination of nitrified effluents. While it is well
known that nitrites in such effluents may hinder chlorination due to exertion
of a chlorine demand, White (39) has reported on the San Jose, CA plant, in
which disinfection by chlorine was improved by addition of small amounts of
ammonia nitrogen. The mechanisms for this effect are still unresolved, but
it should be noted that in the vicinity of the breakpoint, the standard pro-
cedures for the analysis of chlorine forms may be subject to serious error
(29). The experience of White in regard to disinfection of nitrified efflu-
ents is also supported by unpublished observations recorded at the Metropoli-
tan Sanitary District of Greater Chicago (T.B.S. Prakasam, personal communi-
cation) .
Mixing and Contactor Hydraulics
The hydraulic conditions at the point of mixing between the solution of
disinfectant and wastewater, and in the subsequent contact chamber, have been
shown to have a substantial effect on process performance. However, particu-
larly with regard to the first effect, the mechanism of this phenomenon is
not well understood.
The enhancement of chlorine disinfection of microorganisms in wastewater
by intense mixing at the point of chemical addition has been documented by
Longley (17). Recently, it has been suggested that reductions in chlorine
dosage amounting to as much as 50 per cent can be achieved, in part, by
optimizing the flash mixing conditions (25), and White (38) has advocated the
use o| an rms velocity gradient at the point of mixing ("G") of up to 1000
sec „ However, theoretical modelling of the wastewater chlorination process
indicates that the observed enhancement is not due to the acceleration of con-
tact between microorganisms and the rapidly reacting free chlorine (9) , but may
be due to an as-yet poorly understood shearing of microorganisms from protec-
tive particulates.
Since the disinfection process is positive order in microorganism concen-
332
-------
tration, and since high performances are normally desired in such systems
(i.e., efficiencies will in excess of 90 per cent), classical theory predicts
that plug flow contactors should be vastly superior to complete-mix contac-
tors (13,28). A corollary of this principle is that any small deviations from
ideal plug flow behavior in a contactor will result in drastic deterioration
in observed performance of a real system. The empirical length to width
ratios resulting in close to plug flow conditions have been summarized by
White (37,38) and in the WPCF Manual of Practice (35,36). However, the only
quantitative synthesis of the effect of hydraulic imperfections upon chlorina-
tion contact chamber performance appears to have been that of Trussell and
Chao (34), who combined the theory of reaction with longitudinal dispersion,
under the assumption that inactivation is governed by the Selleck equation,
and that chlorine residual is constant, with the assumption of segregated flow,
to conclude that improvement in hydraulics which achieve a dispersion lower
than 0.01 have little practical effect. However, it should be noted that the
particular assumption of segregated flow used in the Trussell and Chao analy-
sis, as well as the neglect of residual decomposition during contact and the
limitations of the Selleck relationship introduce sources of error in this
conclusion.
A second hydraulic aspect which has been briefly mentioned by White (38)
is the ratio of volumetric flows of the halogen feed solution to the wastewater
and its influence upon inactivation efficiency. It has been suggested that
decreasing this ratio, i.e., using a low volume, highly concentrated, feed
solution will improve efficiency. While this has been supported by theoreti-
cal modelling of the disinfection process itself (9), no experimental data
appear to have been collected to elucidate this point.
ISSUES IMPORTANT FOR OPERATIONS
Following the start-up of a wastewater disinfection system, many opera-
tional factors become important in the performance of the process. Work is
now underway to enumerate these various factors. Two major issues have
received much attention, namely the control of chlorine dose and/or residual,
and the behavior of the contact chamber as a sedimentation tank.
Process Control
The design of halogen disinfection processes generally precedes utilizing
steady state assumptions and peak or average design flow conditions. In an
attempt to meet an effluent microbiological constraint while minimizing the
dose (and, in some cases, under regulation, the effluent residual) of halogen
during conditions where flow and influent composition vary, it is necessary to
introduce a control system. The most sophisticated version of this system is
compound flow and residual control (35).
This system is only as strong as its weakest link, which would appear to
be the chlorine analyser itself. Since there is a great difference in micro-
bial sensitivity between free and combined chlorine forms, and since there is
also a difference in sensitivity of microorganisms to mono- and di-chloramine
(8), it would seem desirable to employ an analyser which could differentiate
333
-------
among these distinct species and provide a sensitivity-weighted value of chlor-
ine residual present. No such analyser exists, with the possible exception of
the membrane polarographic electrode (15) which is sensitive primarily to HOC1,
Snead et al. (29) have noted that all commonly used methods for chlorine analy-
sis suffer from false positive indications of the presence of free chlorine
under certain circumstances.
While the use of amperometric and automated wet chemical analysers is
widespread in wastewater treatment plants practicing automatic control, there
would therefore seem to be some room for future improvements in this area.
Solids Sedimentation
To prevent solids deposition in disinfection contact chambers, various
sources recommend the use of a minimum horizontal flow-through velocity to pro-
mote scour (3,18,38). However, this approach has been questioned on the
grounds that the occurrence of additional sedimentation in contact basins may
promote the removal of microorganisms associated with the removed solids (32) .
There does not appear to have been a systematic study of this issue, which
also directly influences the operation of contact basins, in that if sedimen-
tation is promoted a means for solids collection must be provided.
SUMMARY
A number of issues associated with the design and operation of halogen
disinfection systems have been discussed, and various areas of continued un-
certainty highlighted. These include the following:
1) The estimation of halogen dose using procedures which are of a
rational nature is still not entirely possible.
2) The influence of pre-disinfection treatment on the efficiency of the
disinfection process, other than by alteration of pH or concentration
of ammonia, remain to be investigated. In particular, the signifi-
cance of cations in aiding or hindering wastewater chlorination
should be determined, and the effect of various types of biological
treatment on the inherent sensitivity of surviving microorganisms
should be addressed. With chlorine dioxide, the basis for the effect
of pH in altering disinfection efficiency should be explored.
3) The interaction of mixing at the point of chemical introduction and
the inactivation process is not well understood from a mechanistic
point of view, although the existence of this phenomenon is demon-
strated. Until such mechanisms are understood, it is difficult to
present any generalizations regarding the optimal amounts of such
mixing.
4) Further attempts to model the influence of contactor hydraulics upon
process efficiency should be made, and it is essential to obtain
field verification of these results, to confirm the many necessary
assumptions.
334
-------
5) Continuing efforts are needed to develop chlorine analysers which are
capable of distinguishing among the various forms of free and com-
bined chlorine, and to incorporate such analysers in process control
schemes. While the ideal analyser would be a rapid bioassay proce-
dure, this does not appear feasible at present.
335
-------
LITERATURE CITED
1. Aieta, E.M., J.D. Berg, and P.V. Roberts. 1980, "Comparison of Chlorine
Dioxide and Chlorine in Wastewater Disinfection." Jour. Water Poll.
Control Fed. 52:810-822.
2. Averill, C. 1832. "Facts Regarding the Disinfecting Powers of Chlorine."
Letter to the Mayor of the City of Schenectady (NY). S.S. Riggs
Printer, Schenectady.
3. Bates, R.C., P.T.B. Shaffer, and S.M. Sutherland. 1977. "Development
of Poliovirus Having Increased Resistance to Chlorine Inactivation."
Appl. Environ. Microbiol^ 34:849-853.
4. Bates, R.C., S. M. Sutherland, and P.T.B. Shaffer. 1978. "Development
of Resistant Poliovirus by Repetitive Sublethal Exposure to Chlorine,"
p. 471-482. In R. L. Jolley, H. Gorchev, and D.H. Hamilton, Jr. (ed.),
Water Chlorination: Environmental Impact and Health Effects, Volume
2_. Ann Arbor Science Publishers, Inc., Ann Arbor.
5. Benarde, M.A., B.M. Israel, V.P. Olivieri, and M.L. Granstrom. 1965.
"Efficiency of Chlorine Dioxide as a Bactericide." Appl. Microbiol.
13:776-780.
6. Berg, J.D., E.M. Aieta, and P.V. Roberts. 1980. "Comparison of Viral
and Bacterial Indicators of Disinfection in Wastewater with Chlorine
Dioxide and Chlorine," p. 711-722. In R. L. Jolley, W.A. Brungs,
and R.B. Gumming (ed.) , Water Chlorination: Environmental Impact and
Health Effects, Volume 3. Ann Arbor Science Publishers, Inc., Ann
Arbor.
7. Brett, R.W,, and J.W. Ridgeway. 1981. "Experiences with Chlorine Di-
oxide in Southern Water Authority and Water Research Center." Jour.
jnst. Water Eng. Sci. 35:135.
8. Chang, S.L. 1971. "Modern Concept of Disinfection." Proc. Amer. Soc.
Civil Engr., Jour. Sanit. Eng. Div. 97:689-707.
9. Haas, C.N. 1981. "Rational Approaches in the Analysis of Chemical Dis-
infection Kinetics," p. 381-399. In W.J. Cooper (ed.), Chemistry in
Water Re_use_, Volume 1. Ann Arbor Science Publishers, Inc., Ann Arbor.
10. Haas, C.N. 1981. "Sodium Alteration of Chlorine Equilibria: Quantita-
tive Description. " Environ. Sci. & Techno!. 15:1243-1244.
11. Haas, C.N., and E.G. Morrison. 1981. "Repeated Exposure of Escherichia
Coli to Free Chlorine: Production of Strains Possessing Altered Sensi-
tivity." Water;, Air, and Soil Poll. 16:233-242.
12. Haas, C.N., and M.A. Zapkin. In press. "Enhancement of Chlorine Inactiv-
336
-------
ation of E. Coli by Sodium Ions." In R. L. Jolley (ed.), Water Chlor-
ination: Environmental Impact and Health Effects, Volume 4. Ann
Arbor Science Publishers, Inc., Ann Arbor.
13. Holland, C.D., and R.G. Anthony. 1979. Fundamentals of Chemical Reac-
tion Engineering. Prentice-Hall, Inc., Engelwood Cliffs, NJ.
14. Jensen, H., K. Thomas, and D.G. Sharp. 1980. "Inactivation of Coxsack-
ieviruses B3 and B5 in Water by Chlorine." Appl. Environ. Microbiol.
40:633-640.
15. Johnson, J.D., J.W. Edwards, and F. Keeslar. 1978. "Chlorine Residual
Measurement Cell: The HOC1 Membrane Electrode." Jour. Amer. Water
Works Assn. 70:341-348. ~'~
16. Kuzminski, L.N. 1972. "Effect of Calcium Bicarbonate on Disinfection
by Halogens." Amer. Soc. Civil Engr., Proc. Jour. Sanit. Eng. Div.
98:229.
17. Longley, K.E. 1978. "Turbulence Factors in Chlorine Disinfection of
Wastewater." Water Res. 12:813-822.
18. Metcalf & Eddy, Inc. 1979. Wastewater Engineering: Treatment, Dispo-
sal, Reuse, 2nd Edition. McGraw Hill Book Co., NY.
19. Milbauer, R., and N. Grossowicz. 1959. "Effect of Growth Conditions on
Chlorine Sensitivity of Escherichia Coli." Appl. Microbiol. 7:71-74.
20. Reid, L.C., and D.A. Carlson. 1974. "Chlorine Disinfection of Low Tem-
perature Waters." Proc. Amer. Soc. Civil Engr., Jour. Environ. Eng.
Div. 100:339-351.
21. Roberts, P.V., E.M. Aieta, J.D. Berg, and B.M. Chow. 1980. "Chlorine
Dioxide for Wastewater Disinfection: A Feasibility Evaluation."
Stanford University, Department of Civil Engineering, Technical Report
#251.
22. Scarpino, P.V., G. Berg, S.L. Chang, D. Dahling, and M. Lucas. 1972.
"A Comparative Study of the Inactivation of Viruses in Water by Chlo-
rine." Water Res. 6:959-965.
23. Scarpino, P.V., F.A.O. Brigano, S. Cronier, and M.L. Zink. 1979.
"Effect of Particulates on Disinfection of Enteroviruses in Water by
Chlorine Dioxide." U.S. Environmental Protection Agency, Report EPA-
600/2-79-054.
24. Selleck, R.E., B.M. Saunier, and H.F. Collins. 1978. "Kinetics of Bac-
terial Deactivation with Chlorine." Proc. Amer. Soc. Civil Engr.,
Jour. Environ. Eng. Div. 104:1197-1212.
25. Sepp, E. 1981. "Optimization of Chlorine Disinfection Efficiency."
337
-------
Proc. Amer^ Soc. Civil Engr., Jour. Environ. Eng. Div. 107:139-152.
26. Shaffer, P.T.B., T.G. Metcalf, and O.J. Sproul. 1980. "Chlorine Resis-
tance of Poliovirus Isolants Recovered from Drinking Water." Appl.
Environ. Microbiol. 40:1115-1121.
27. Sharp, D.G., D.C. Young, R. Floyd, and J.D. Johnson. 1980. "Effect of
Ionic Environment on the Inactivation of Poliovirus in Water by Chlo-
rine." Appl. Environ. Microbiol. 39:530-534.
28. Smith, J.M. 1981. Chemical Engineering Kinetics, 3rd Edition. McGraw
Hill Book Co., NY.
29. Snead, M.C., V.P. Olivieri, and W.H. Dennis. 1981. "Biological Evalua-
tion of Methods for the Determination of Free Available Chlorine,"
p. 401-427. InW.J. Cooper (ed.) , Chemistry jLn Water Reuse, Volume _!.
Ann Arbor Science Publishers, Inc., Ann Arbor.
30. Sollo, F.W., H.F. Mueller, I.E. Larson, and J.D. Johnson. 1975 "Bromine
Disinfection of Wastewater Effluents," p. 163-177. In J.D. Johnson
(ed.), Disinfection—Water and Wastewater. Ann Arbor Science Publish-
ers, Inc., Ann Arbor.
31. Sproul, O.J., R.T. Thorup, D.F. Wentworth, and J.S. Atwell. 1970. "Salt
and Virus Inactivation by Chlorine and High pH." Conference on Disin-
fection. American Society of Civil Engineers, Washington, DC.
32. Thalhamer, M.G. 1981. "A Site-Specific Design of Chlorination Facili-
ties." Proc. Amer. Soc. Civil Engr., Jour. Environ. Eng. Div. 107:
473-480.
33. Taras, M.J. 1950. "Preliminary Studies on the Chlorine Demand of Speci-
fic Chemical Compounds." Jour. Amer. Water Works Assn. 42:462-472.
34. Trussell, R.R., and J. Chao. 1977. "Rational Design of Chlorine Contact
Facilities." Jour. Water Poll. Control Fed. 49:659-667.
35. Water Pollution Control Federation. 1976. Chlorinaton of Wastewater.
Manual of Practice #4. Washington, DC.
36. Water Pollution Control Federation. 1977. Municipal Wastewater Treat-
ment Plant Design. Manual of Practice #8. Washington, DC.
37. White, G.C. 1972. Handbook of Chlorination. Van Nostrand-Reinhold Co.,
NY.
38. White, G.C. 1978. Disinfection of Wastewater and Water for Reuse. Van
Nostrand-Reinhold Co., NY.
39. White, G.C., R.D. Bebbe, V.F. Alford, and H.A. Sanders. 1981. "Problems
of Disinfecting Nitrified Effluents." Proceedings of the National Con-
ference on Environmental Engineering. ASCE, Washington, D.C.
338
-------
2. DESIGN AND OPERATIONAL CONSIDERATIONS FOR WASTEWATER OZONE
DISINFECTION SYSTEMS
Kerwin L. Rakness, P.E.
M & I, INC., Consulting Engineers
Fort Collins, Colorado
ABSTRACT
Ozone systems are usually air fed; once through oxygen fed; or recycle
oxygen fed units. Air fed and recycle oxygen fed systems require dew point
treatment processes that are extremely sensitive, yet critical to ozone pro-
duction. Once through oxygen systems are desired, if the oxygen requirement
of the downstream oxygen process (e.g., oxygen activated sludge process) can
be balanced with the oxygen requirement of the ozone process.
A high dew point of the feed gas will decrease ozone production and may
damage generator components. Typically, more than 99.9 percent moisture
removal is required, and as little as 99.7 percent removal will cause prob-
lems. The design engineer and plant owner (as represented by the operator)
should consider maintaining tight control of this critical and sensitive
ozone system component.
Ozone systems are energy intensive, and energy consumption varies as
ozone production rate varies. Power usage rate at start-up may be as much
as 3 to 4 times the rate at design, unless system flexibility is provided.
Both start-up and design conditions should be analyzed during design. Auto-
matic control of ozone production and energy use may be employed, but the
extra capital cost, imprecise control abilities, and intensive maintenance
requirements for the control equipment may not be justified by the reduced
ozone production rate achieved. Each situation should be thoroughly evalu-
ated. Manual control may be quite complex or more simplified. A simple
approach reduces, but does not eliminate the need for some process monitor-
ing equipment.
Ozone transfer efficiency (T.E.) is proportional to absorbed ozone
dosage, which is proportional to the disinfection level achieved. Ozone may
be absorbed through chemical reaction or through ozone/liquid gas dissolu-
tion. The ozone chemical reaction must be satisfied before effective dis-
infection can occur. Municipal/industrial wastes which have known or sus-
pected ozone reacting pollutants should be analyzed using bench or pilot
scale studies to determine the required absorbed ozone dosage to achieve the
desired disinfection level. In all plants, the minimum acceptable T.E.
should be based upon ozone/liquid gas dissolution theory.
Ozone may be detected (smelled) at levels about 1/10 the typical 8-hour
human exposure standards. This constitutes a safety aspect of ozone sys-
tems. However, operators may become desensitized or careless; thus, ambient
ozone monitors with alarms should be provided. Ozone concentrations in the
contact basin feed and exhaust gas is several thousand times greater than
the human exposure standard, and a tiny leak can cause excessive ambient
ozone concentrations. System design and operation must address this fact.
339
-------
TYPES OF OZONE SYSTEMS
Ozone systems may be categorized in several ways, depending on the
topic to be emphasized. If the ozone feed stream is emphasized, three broad
types of ozone systems exist: 1) air fed, 2) once through oxygen fed, and
3) recycle oxygen fed. A flow schematic of each type is shown in Figure 1.
Oxygen fed systems generate about twice as much ozone per unit of electrical
energy used. However, oxygen fed units are typically not cost effective,
unless the oxygen can be used for another purpose, for example in the acti-
vated sludge process.
Each type of ozone wastewater disinfection system has an ozone genera-
tor, contact basin, and destruct unit. The air fed and oxygen recycle sys-
tems also have dew point treatment equipment. Dew point treatment for ozone
generation is very sensitive and will be discussed in more detail later.
Dew point treatment is typically not required for once through oxygen fed
systems, because direct feed high purity oxygen is normally much dryer than
required for ozone generation (-51°C dew point or dryer is desired for ozone
generation).
Once through oxygen fed systems can be used if the oxygen requirement
downstream of the ozone process is balanced with the oxygen requirement of
the ozone process. Figure 2 illustrates a balanced oxygen usage graph for a
once through oxygen fed ozone system and an oxygen activated sludge process.
Figure 2 shows that when the activated sludge oxygen requirement is 1.1 kg
02/kg (BODj)^ and the 6005 removal rate is 150 mg/1, then the ozone
concentration will be about 3 percent when the required dosage is 5 mg/1.
No dew point treatment equipment and correspondingly fewer operation and
maintenance tasks are required when a once through oxygen fed ozone system
is used. If the downstream oxygen consuming process requires about as much
oxygen as the ozone process, then a once through oxygen ozone process should
be considered. If significantly more oxygen is required downstream, then a
controlled amount of oxygen may be bypassed around the ozone system. If
significantly less oxygen is required downstream, an oxygen recycle system
may be considered.
DEW POINT TREATMENT
Ozone generation equipment must be supplied with dry, particle-free
gas. Filters are typically used to remove particles. Desiccant dryers plus
in some cases refrigerant dryers are used to attain dry gas. Feed gas
treatment is recommended if its dew point is -51°C or higher. A high dew
point will result in lower ozone production, as shown in Figure 3 (1).
Further, a high dew point will cause more rapid fouling and require more
frequent cleaning of the generator; nitric acid formation (air and oxygen
recycle systems) and damaged generator components; and may cause electrical
short circuiting.
340
-------
The feed gas dew point varies with its moisture content. The relation-
ship between moisture content and dew point is shown in Figure 4. For exam-
ple, when the moisture content of the feed gas is about 20 ppm by weight (at
1 atm pressure), the dew point is -51°C.
A relatively small change in moisture content will cause a significant
change in dew point, especially in the range of operating dew point levels
for ozone generators. An example for an air fed ozone system is shown in
Table 1. If 99.9 percent of the moisture is removed, the dew point is
satisfactory. However, if only 0.2 percent less moisture is removed (99.7
percent removal), the dew point is marginal to unsatisfactory! The impor-
tance of a well-designed and operated ozone feed gas dew point treatment
unit is apparent.
TABLE 1. DEW POINT TREATMENT SENSITIVITY TO MOISTURE CONTENT
Process Equipment
Compressor
Refrigerant Dryer
Desiccant Dryer
Moisture
( ppm b y
23,
5,
Content
Weight)*
000
000
20
80
Moisture
Removal
78.3
99.9
99.7
Dew Point
°C
27
4.5
-51
-40
*From Figure 4.
The ozone feed gas dew point treatment equipment is usually provided,
but not manufactured by the ozone generator equipment supplier. The ozone
equipment manufacturer will purchase the dew point treatment equipment from
other manufacturers, as needed. Limited design engineer control of the dew
point treatment aspect of the ozone system design will cause a myriad of
system and equipment options available to the ozone equipment suppliers.
Because of the important and sensitive nature of this process, as discussed
above, the plant owner (as represented by the operator) and design engineer
should consider maintaining tight control over this area of system design.
Air dew point treatment processes can be either low [103 kN/sq m (15
psig)], medium [206 kN/sq m (30 psig)j, or high pressure [688 kN/sq m (100
psig)] systems. Each has specific operation and maintenance advantages and
disadvantages, which should be evaluated on a case-by-case basis. Equipment
reliability, air flow control, power usage, turndown capability, and main-
tenance requirements are a few of the issues which should be evaluated.
Equipment duplication and system flexibility also should be provided,
because a small upset in dew point treatment can result in major problems
with ozone generation capability.
341
-------
Monitoring devices should be provided to measure and record the feed gas
dew point continuously. Alarms to indicate a high dew point level also
should be installed. However, care must be taken to insure that the sensi-
tive dew point measuring equipment is giving accurate results. A "dew point
cup" measuring device may be used to check and calibrate the in-line meter.
A procedure for using the dew point cup is described below. Refer to
Figure 5 for a schematic of the dew point cup.
A small stream of air is directed to the outside of a polished,
stainless steel cup. The cup is filled about half full with ace-
tone, the temperature of which is measured with a thermometer.
Dry ice is gradually added to the acetone to decrease the temper-
ature of the acetone. The temperature of the acetone and dry ice
mixture is then read, and that reading is the air dew point. This
dew point reading is at atmospheric pressure and must be adjusted
to the actual pressure dew point of the in-line dew point monitor
in order to calibrate the monitor properly.
OZONE GENERATION
Several different types of ozone generators are available including air
cooled, water cooled, or oil and water cooled; and voltage controlled or fre-
quency controlled units. Each manufacturer has prescribed advantages of his
brand, and the design engineer may decide to choose one type or consider all
types equally. The ozone generator, however, is only part of the ozone pro-
cess. Equally important is feed gas treatment, ozone contacting, and ozone
destruction. All units should be evaluated independently and also as they
interrelate, one to the other.
One consideration for ozone system design is power consumption of the
process. The relationship between power use rate and ozone production for
an air feed ozone process is shown in Figure 6 (l). The rate of power usage
for the ozone generator alone increases as the ozone production rate in-
creases. However, power use rate for the total system (generator, feed gas
treatment, and ozone destruction) decreases as ozone production rate in-
creases. The reason for this occurrence is the relatively high, constant
power demand of the feed gas treatment and ozone destruction equipment.
For the ozone system represented in Figure 6, the lowest power usage
rate occurs at the design point of the process. However, most ozone systems
used in wastewater disinfection probably will be operated at outputs much
less than the design output because: 1) conservative estimates of maximum
ozone dosage required may be used to size the ozone equipment, and 2) start-
up plant flow rate will probably be less than the plant design flow rate.
Both reasons cause the ozone production requirement to be less than design,
and will cause inefficient power consumption unless system flexibility is
provided to achieve lower power usage rates at lower ozone production rates.
Bpth the start-up and design power usage rate should be thoroughly evaluated.
342
-------
Power consumption of the ozone system increases as the ozone production
rate increases; thus, an energy savings is realized when the ozone produc-
tion requirement is decreased. The ozone production requirement is estab-
lished by the level of disinfection to be achieved (kill rate), the ozone
demand, and the ozone contact basin I.E. Ozone demand and contact basin
I.E. are discussed later. Venosa, et.al. (4)(5) and Stover, et.al. (3) have
shown that the kill rate is directly proportional to the absorbed ozone
dosage; thus, to reach a desired level of disinfection a certain absorbed
ozone dosage must be attained. The required absorbed ozone dosage may vary
for different plants, because of water quality, but in each plant the ozone
production rate would be used to adjust the amount of ozone absorbed.
Optimum process control for any given plant is achieved when the ozone
production rate is as low as needed to achieve the required disinfection
level. The two desired goals of good effluent quality and minimum energy
consumption are met. If ozone production is greater than necessary, good
effluent quality will still be achieved. The required level of disinfection
will be met, and the residual ozone caused by the overdose will decompose
back to oxygen fairly quickly (2). However, more energy will be consumed. To
minimize ozone production yet achieve good effluent quality, automatic con-
trol of the ozone supply rate is often a consideration in system design.
Some of the ways in which automatic control of ozone supply may be com-
pleted are:
• Effluent ozone residual control - Interloop between ozone resi-
dual analyzer and ozone production equipment.
• Ozone dosage with wastewater flow control - Interloop between
wastewater flow measurement and ozone production equipment.
• Ozone off-gas control - Interloop between ozone contact basin off-
gas residual analyzer and ozone production equipment.
• Combination off-gas and wastewater flow control - Compound inter-
loop between ozone contact basin off-gas residual analyzer plus
wastewater flow meter and ozone production equipment.
Each of the automatic control systems available have varying degrees of
equipment problems and somewhat imprecise control abilities. Also, they add
to the initial cost of an ozone system. This higher cost may not be recov-
ered if the overall ozone production level is not reduced by a substantial
amount. Thus, the cost of this additional control equipment and its intense
maintenance requirements may not be justified. Each situation should be
thoroughly evaluated.
The alternative to automatic process control of an ozone system is man-
ual control. Manual control requires that the operators adjust the ozone
supply rate as the wastewater flow rate varies to achieve the required level
of disinfection. This procedure is similar to simple chlorination system
control. However, manual control of ozone systems does not cause the prob-
343
-------
lem with water quality due to overdosing as chlorine systems cause. The
only drawback to manual control of ozone is the higher energy cost that may
occur because of overdosing.
Manual control of the ozone system may be fairly complex or quite sim-
plified. The degree of complexity is dictated by the number of parameters
measured and analyzed before an adjustment is made to the ozone production
rate. The operators may analyze ozone residual, wastewater flow rate, and
contact basin off-gas concentration data before adjusting the ozone supply
rate. These data provide information about the current operating condition
of the system that most directly relates to the disinfection kill rate, but
requires more complex and sensitive equipment which results in added mainte-
nance requirements,
A more simplified manual control procedure may be used to reduce both
the initial equipment costs and on-going maintenance costs. The approach
requires that the operators develop, for their system, a relationship
between ozone dosage to the wastewater and the desired level of disinfec-
tion. The ozone production rate can then be adjusted as the wastewater flow
rate varies, to achieve the prescribed ozone dosage. A procedure for a sim-
plified manual control approach for the Vail, Colorado, ozone process is as
follows:
When the required ozone dosage is established the ozone production
rate is varied to meet that dosage at various wastewater flow
rates. The required ozone production for various wastewater flow
rates is shown in Figure 7. The example shows that at a waste-
water flow rate of 10,200 up/day (2.7 mgd) and an ozone dosage of
4 mg/1, the required ozone production is 41 kg/day (90 Ib/day).
When the required ozone production is established the ozone system
must be adjusted to produce ozone at that rate. Two main factors
influence the production of ozone; the air flow rate and the power
supply (power supply controls the ozone concentration from the
generator), as shown in Figure 8. At a given air flow rate, for
example 1.98 m3/min (70 scfm),, the power supply adjustment will
cause the ozone concentration to vary and hence, ozone production
to vary. To achieve a given production rate, power supply is
adjusted and the air flow rate is left constant. The example in
Figure 8 shows that to reach 41 kg/day (90 Ib/day) production at
an air flow rate of 1.98 m^/min (70 scfm), the power should be
adjusted until the ozone concentration reaches about 7,100 to
7,200 ppm by volume.
The Vail ozone system has air flow meters and an ozone concentra-
tion meter that can be used to set the ozone production at the
desired rate. Several combinations of air flow and ozone concen-
tration can be used to achieve the desired ozone production rate.
The most economical operating point should be selected. This
point may be determined by conducting a special generator mapping
test, and then referring to the 'hap" each time the production
344
-------
rate is changed. An example "map" for an ozone system is shown in
Figure 9. The ozone production rate would be achieved using less
electrical energy at air flow rate "B" versus air flow rate "A".
Therefore, the operator would use the proper equipment in the
system to get air flow rate "B", then adjust the generator power
supply to achieve the required ozone concentration established
from Figure 8 .
It should be noted that the simplified manual control procedure has not
eliminated the use of all process measuring equipment. At least two instru-
ments are recommended; an in-line dew point monitor and an ozone concentra-
tion meter. The dew point cup is used to check and calibrate the dew point
monitor, as discussed earlier. Wet-chemistry testing is used to check and
calibrate the ozone concentration meter. The wet-chemistry procedure for
the Vail, Colorado, ozone system is presented below. The approach is
applicable to other systems. Note that the Vail system is at an elevation
of 2,470 m (8,100 feet) above sea level.
1. Set ozonator at desired power setting. Record generator informa-
tion on data sheet (see Figure 10).
2. Check High Concentration Ozone Meter zero, span, control, and
sample frequency readings and adjust to manufacturer's recommended
setting, if necessary.
3. Prepare wet test chemistry equipment (see Figure 11).
a. Add 400 ml of 2 percent KI solution to each of two 500 ml gas
washing bottles (Note: A fritted glass diffuser is not used on
ozone-air inlet tube).
b. Connect gas washing bottles in series and connect ozone supply
line and wet test meter.
c. Level wet test meter and adjust water level in the meter.
4. Open vent valve and vent test line for 2 minutes.
5. Read and record three consecutive Ozone Meter readings.
6. Set valve to direct ozone-air gas flow to the gas washing bottles
at a rate of 2 liters/minute.
7. Run approximately 3.0 liters of gas flow through the bottles and
record field data information on data sheet (see Figure 10).
8. Take gas washing bottles to laboratory immediately and have another
person read and record three more Ozone Meter readings.
9. Quantitatively transfer liquid from gas washing bottles to two
separate 1 liter Erlynmeyer flasks. Rinse tubes and bottles at
least three times.
10. Immediately add 10 ml of 2N Sulfuric Acid (l^SO^ .
11. Read initial buret volume which contains 0.1N Sodium Thiosulfate
solution (Na2S203). Note: Standardize Na2S203 using
the dichromate method. (Standard Methods Ed. 14, pp. 316.)
12. Quickly titrate the darker of the two flasks to a pale yellow color
with the Na2S203.
13. Add 5 ml starch indicator (see Standard Methods Ed. 14, pp. 314 for
starch preparation) and carefully titrate until clear.
14. Add 5 ml starch indicator to second flask and again carefully
titrate, dropwise, until clear.
345
-------
15. Record final buret reading and determine total volume of titrant
used. Record on data sheet (see Figure 10).
16. Complete calculations on data sheet (see Figure 10).
17, Adjust span setting on Ozone Meter by following calculation:
..Laboratory OT concentration^
New span = old span (- • '.__..__...•-_ >
Meter 03 concentration
OZONE CONTACT BASIN
The ozone contact basin plays a key role in achieving acceptable disin-
fection with ozone. Earlier it was mentioned that the level of disinfection
is related to the absorbed ozone dosage (3)(4). Contact basin I.E. is
directly proportional to absorbed ozone dosage, as shown below.
- (Mass of Absorbe_d_C^z_one_)_(jLOOj_
Mass of Applied Ozone
where: Mass of Absorbed Ozone = Mass of Applied Ozone minus
Mass of Ozone in Off-Gas
The relationship among applied ozone dosage, absorbed ozone dosage, and
T.E. is shown in Figure 12. The example lines 1, 2, and 3 show the level of
applied ozone dosage required to achieve the same level of absorbed ozone
dosage as the T.E. decreases. When the T.E. decreases from 90 percent to 80
percent to 70 percent, the applied ozone dosage is 111 percent greater, 125
percent greater and 143 percent of the absorbed ozone dosage, resepectively.
Indeed, if the T.E. is only 50 percent, a full 200 percent more applied
ozone dosage is needed. The point is that the level of applied ozone dosage
required, and resulting level of ozone production needed, to achieve a given
absorbed ozone dosage increases at a faster rate than the T.E. decreases.
Therefore, to minimize ozone production requirements, T.E. should be maxi-
mized.
For a given applied ozone dosage, the absorbed ozone dosage increases
as the ozone T.E. increases. Ozone absorption can occur through a direct
chemical reaction with the pollutants in the wastewater and through ozone
dissolution to the wastewater. Extremely high ozone T.E. can occur if the
chemical reaction predominates, for example as with potassium iodide (Kl) or
with certain kinds of industrial wastes. For these cases the T.E. and
absorbed ozone dosage may be high, but the disinfection kill rate will prob-
ably be low. The ozone chemical demand must be satisfied before effective
disinfection can occur.
When the ozone chemical demand is satisfied, the ozone dissolution rate
(gas to liquid transfer rate) will control the level of absorbed ozone at a
given applied ozone dosage. Based on this premise, the following recommen-
dations are made:
• The minimum acceptable ozone contact basin T.E. should be based
upon ozone/liquid gas transfer theory.
346
-------
• Wastewaters which contain known or suspected ozone chemical reac-
tants (i.e., municipal/industrial wastes) should be analyzed using
bench or pilot scale studies to determine the absorbed ozone dosage
required to achieve the desired disinfection level.
Venosa, et.al. (4)(5) have addressed ozone/liquid gas dissolution in
detail. Based on their findings, the following points should be considered
in design.
• Ozone T.E. is governed by Henry's law, like oxygen transfer to
water.
• Deep contactors using "fine" or intermediate bubble diffusers appear
to provide the best assurance for ozone dissolution.
• High applied ozone concentrations appear to yield better ozone dis-
solution efficiencies.
OPERATOR SAFETY AND OFF-GAS OZONE DESTRUCTION
Typical standards call for a maximum allowable atmospheric ozone concen-
tration for an 8-hour work day of 0.0002 mg/1 by weight/volume (0.1 ppm by
volume). Usually a person can smell ozone at a concentration of 1/10 this
level (2). Therein lies a built-in safety feature of ozone systems; the opera-
tors usually are not exposed to ozone concentration levels at or above the ac-
cepted standards when they do not detect (smell) ozone in the environment.
However, operators may become somewhat desensitized to ozone or somewhat less
careful when continuously around ozone systems; thus, all ozone systems should
have one or more ambient ozone monitoring devices to measure and record the
ambient ozone concentration, sound an alarm when concentrations exceed a pre-
determined level, and automatically shut-down the ozone system immediately,
or after a pre-set time alarm is not acknowledged within that time frame.
Note: The latter approach avoids unnecessary system shut-downs due to false
readings.
The concentration of ozone from the generator is typically between 12
to 24 mg/1 by weight/volume, or 60,000 to 120,000 times greater than the
typical 8-hour human exposure standard. As such, a tiny leak in the ozone
supply piping can cause excessive ambient ozone concentrations. Extreme
care should be used in design and installation of ozone pipe and equipment.
A remote location for the ozone system, stainless steel piping, and other
special precautions should be considered.
A good ozone contact basin may have a T.E. of 90 percent. At that T.E.
the off-gas ozone concentration would be 1.2 to 2.4 mg/1 by weight/volume.
This concentration is 6,000 to 12,000 times greater than the typical 8-hour
human exposure standard. An exceptionally high, probably unrealistic, T.E.
is 99 percent. Yet at 99 percent T.E. the off-gas ozone concentration is
still 600 to 1,200 times greater than the typical human exposure standard.
The ozone discharged in the off-gas will dissipate, in time, but the half-
347
-------
life of ozone in air is as long as 12 hours (2). The need for destruction
of ozone in the off-gas is apparent!
Some off-gas ozone destruction treatment options include heat destruct,
heat/catalyst destruct, activated carbon, recycle to sewage or sludge, and
discharge through a tall stack. Heat destruct provides positive control of
off-gas ozone concentration but requires a high power consumption. Heat/
catalyst destruct also provides positive control of off-gas ozone concentra-
tion but requires some power consumption and periodic catalyst replacement.
Activated carbon has an explosive potential when combined with ozone, which
should be thoroughly analyzed if considered. Recycle to sewage or sludge
does not provide positive control over off-gas ozone destruction and could
transfer the problem to another area of the plant. Discharge through a tall
stack also does not provide positive control over potential off-gas ozone
contamination of the work environment. The off-gas ozone treatment options
normally used are heat or heat/catalyst destruct units.
Special precautions should be employed to insure that the off-gas con-
taining ozone does not bypass the destruct unit. Also, foam suppression
equipment should be installed in the off-gas removal piping to keep the foam
from coating and contaminating the heating coils or catalyst equipment. The
foam suppression equipment should be simple to operate, routinely checked by
the operators, and easy to maintain.
The ozone concentration of the off-gas ozone destruction system should
be measured on a periodic basis (weekly or monthly) to monitor the perform-
ance of the process. The procedure to measure the off-gas ozone concentra-
tion gas streams to and from the ozone destruct equipment is similar. The
procedure for measuring the inlet off-gas ozone concentration for Vail,
Colorado, is described below. The approach is applicable to other systems.
Note that Vail is 2,470 m (8,100 feet) above sea level.
1. Prepare wet test chemistry equipment (see Figure 13).
a. Add 400 ml of 2 percent KI solution to one gas washing bottle.
b. Connect wash bottle to test line and wet test meter.
c. Connect vacuum line to wet test meter vent.
d. Level wet test meter and adjust water level in the meter.
e. .Open vacuum valve until moderate gas flow rate is established.
2. Run approximately 12 liters, or more if necessary, of gas flow
through the bottle and record field information on data sheet (see
Figure 14).
3. Take gas washing bottle to laboratory immediately.
4. Quantitatively transfer liquid from gas washing bottle to a 1 liter
Erlynrnyer flask. Rinse tube and bottle at least three times.
5. Immediately add 10 ml of 2N Sulfuric Acid (l^SC^) .
6. Read initial buret volume which contains 0.IN Sodium Thiosulfate
solution (N32S203). Note: Standardize ^28203 using
the dichromate method. (Standard Methods Ed. 14, pp. 316).
7. Quickly titrate to pale yellow with ^28203.
8. Add 5 ml starch indicator (see Standard Methods Ed. 14, pp. 317 for
starch preparation) and carefully titrate until clear.
348
-------
9. Record final buret reading and determine total volume of titrant
used. Record on data sheet (see Figure 14).
10. Complete calculations on data sheet (see Figure 14).
349
-------
LITERATURE CITED
1. Rakness, K. L., B. A. Hegg, L. A. Boehme, and B. B. Fairchild. Case
History: Ozone Disinfection of Wastewater with an Air/Ozone System.
Proceedings of Wastewater Disinfection Alternatives - State-of-'the-Art
Workshop, 52nd Annual Water Pollution Control Federation Convention,
Houston, Texas (October 1979).
2. Rice, R. C., C. M. Robson, C. W. Miller, and A. G. Hill. Uses of Ozone
in Drinking Water Treatment. Journal American Water Works Association
(January 1981) .
3. Stover, E. L. and R. W. Jarnis. Obtaining High Level Wastewater Disin-
fection with Ozone. Journal Water Pollution Control Federation, Vol.
53, pp. 1637 (November 1981).
4. Venosa, A. D. , M, C. Meckes,, E. J. Opatken, and J. W. Evans. Compara-
tive Efficiencies of Ozone Utilization and Microorganism Reduction in
Different Ozone Contractors. Progress in Wastewater Disinfection Tech-
nology, A. D. Venosa, ed., EPA-600/9-79-018. U.S. Environmental Protec-
tion Agency, Cincinnati, Ohio, pp. 287 (June 1979).
5. Venosa, A. D., M. C. Meckes, E. J. Opatken, and J. W. Evans. Disinfec-
tion of Filtered and Unfiltered Secondary Effluent in Two Ozone Contac-
tors. Paper presented at the 52nd Annual Conference of the Water Pollu-
tion Control Federation (October 7-11, 1979).
350
-------
AIR FED AIR / OXYGEN ENRICHED
Air (Dew Point) Treatment
Air
(Dew Point)
Treatment
-
Ozone
Generator
-
Ozone
Contacting
Vent
OXYGEN FED
OXYGEN RECYCLE
Vent
Figure 1. Line diagram for three types of ozone systems,
(O
o
a
in
z
o
M
O
7.O
4.0
5.0
4.0
SO 100 150 200
BO05 REMOVED (mg/l)
250 300
Figure 2. Required ozone concentration of various ozone dosages and
BOD5 removal rates for a once through oxygen/ozone and
oxygen/activated sludge system.
351
-------
z
o
I-
o
o
C
90
01
O
g 8*
80
a 73
u.
o
,- TO
ui
U
£ «5
a.
eo
-62-80 -6» -80 -48 -40 -88 -30 -28 -JO
DEW POIS4T TtfcSPeRATURt - 'C
-10 -s
Figure 3. Ozone production rate decreases as feed gas dew point
temperature increases.
O
U
UJ
a:
t-
III
DiW POINT TEMPERATURE (C°)
DEW POINT vs. MOISTURE CONTENT
Figure 4. Feed gas moisture content increases as its dew point increases
352
-------
THERMOMETER
POLISHED
STAINLESS
STEEL CUP
OBSERVATION
WINDOW
CONTAINER
Figure 5. Diagram of dew point cup feed gas dew point measuring device.
OZONE PRODUCTION- kg/day
. 0 4.5 9.1 13.6 iai 22.7 27.2 31.8 36.2
n 33
O
M
0 30
a
\ 27
Si
5
t
o M
?"
< 18
N
_J
POWER UT
u a
-------
Figure 7. Ozone production rate required
at various wastewater flow and
ozone dosages (Ib/day x 0.453 =
kg/day and mgd x 3785 = rn?/day
1.0 2.0 3.0 4.0
WASTEWATER FLOW IN (MGD)
O 400O 800O t2,OOO 13,OOO
OZONE CONCem?ATOM IN AIR
(PPM SY VOLUSSE)
Figure 8. Ozone concentration required
to achieve the desired ozone
dosage at various feed gas
flow rates (Ib/day x 0.453 =
kg/day).
a
ut
K
0
1U
ss
at
o
a,
CZOWE PRODUCTION (Ib/day)
Figure 9. Example ozone generator
'\nap" describing the
most efficient operat-
ing point.
354
-------
Ul
Ln
"ate Time of Analvgla
FIC.LD INFORMATION
Air Pretreatment : Volume cf» Presaure pa ig Temperature 'f Dewpoint ( *F)
Ozone Meter: Span Heading before Reading after Average
LAB INFORMATION
Wet Test Meter- Titration mis N of Na^SjO-i mole eq/L
GENERATOR OUTPUT AND APPLIED 07.0N1! DOSAGE CALCULATIONS
0 C t f
Calculate weight of orone trapped in Kl solution.
L mole eq gm ml
V - (v, )(PI)(T2) . ( ,w i" H70,,527.6°R,
Where: V, - Actual volume in L
.
Meter
ture CF)
42-47
47-53
54-58
59-63
6'<-66
67-69
70-72
73-75
76-77
78-79
79-80
(see Table)
Vapor
(in H70)
4
5
6
7
8
9
10
11
12
13
14
P| - Adjusted pressure - (Plant atmospheric preseure (8100 ft) of 301 in H20) - (water vapor prcisi.ro)
* (wet tent manotnol-er pressure - Note: suction Is negative). PI " _ +
_ in. H20
T2 - Standard temperature (absolute) =• 68'F + 459.6 - 527. 6*R
T| - Actual temperature (absolute) - _ 'F « 459.6 » 'R
ppm/vol - (
V,
(20'C)
p«i")(529.6'R[(|) .
20.7pnia R
14.7 p.i - 20 7 p.i.
Where: V| - Actual Volume in Ft'
P2 - Standard pressure (absolute) - Cauge pressure * atmospheric pressure •
P, - Actual pr.ssur, (absolute) - Gauge reading (psie) « plant atmospheric pressure (8100) of 10.SB p»i
psig + 10.RB psi - pala
Tj - Standard temperature for rotom.ter (absolute) - 70'F . 459.6 - 527.6'P
T| " Actual temperature for rotometer (absolute) - 'r t 459.6 - 'R
T, • Standard temperature of 20'r? (ab,ol.,te) - 68'F * 459.6 - 527. 6'R
:,,cu,a,: »„ ppl^rate. _ ^ ^^ , /f ,
__
Figure 10. Example wet-chemistry, ozone production concentration data sheet.
-------
OZONE GENERAlOR
SAMPLE CONTROL i '
VALVE
TO VENT DUCT
AIR
SUPPLY
L^sss— , 8
GAS WASHING
BOTTLES - W/
2% Kl SOLUTION
MANOMETER-
TEMPERATURE—
WET TEST
METER
Figure 11. Example wet-chemistry, ozone production concentration
testing equipment set-up.
O
z
UJ
li.
UJ
cc
UJ
It
(0
2
<
QC
H
z
O
N
O
100 120 140 160 180 200 220 240 200
APPLSED OZOME DOSAGE (% OF ABSORBED OZONE DOSAGE)
Figure 12. Applied ozone dosage increases significantly as
ozone transfer efficiency decreases.
356
-------
FROM BACKWASH
STORAGE BASIN "~[~~|
OZONE CONTACT BASIN
VACUUM SUPPLY
TO ATMOSPHERE
OZONE DESTRUCT UNIT
FLOW CONTROL
VALVE
WET TEST
METER
GAS WASHING
BOTTLE W/
2% Kl SOLUTION
Figure 13. Example wet-chemistry, off-gas concentration testing
equipment set-up.
357
-------
U3
ui
oo
FtFil.n ^H FORMATIHH
Off-Cap- Fln>
L<~>cn t i n
527.6
Wet Teal. HeLer; Vol.r
LAB INFORMATION
i, Temperntu
(20"C)
tn. H;0 (Suction is nefi.itive) Wnler Vapor Pt
in. II20 (nee TflMr)
« of N,12S2I>3 mote eq/L
CAI.rUi.ATlOHS
Ozone Concencration in Off r.aa:
of o/one trapped in KI uol.il ion.
t
1
Whrrp- V( - Actual volume in I,
Fy " RtanHarc) preosttre = ^07.8 in. II2O
PI - AdjiiRt^rf pr^nfiiirp =• (Tiflut at.tnonphpr (c precaure (RlOH ft) of 301 in ^0) -
*• (wpt teat manomplpr prf«tsnre - Note: suction is negative). Pj =•
in. M^O
T2 ° Standsrrt tPmpprHtirrp (flhnnlutp) - 6fi'F + 459.6 - 527.6*R
T( = Actus! tempernturf (nN0nlutp> = *F + 459.6 * °n
Cnlrtilntp o^oi.r^ concpnt ration In off
tnR/f, rtir
.
fi. ro5) (1,000, OnnX-iy) --
Rep id ii-i I 07;one Concentration:
Calritlavp weight of ozone trapped In KI point. Ion
, N'
I,
Catcvilare reflidufll o^one concentrnt ion.
1
""=/'• "7." = < "*>< SI nf B3
., .^^^Igni Oiw1.0 L
ote eq/L)( tnlX—*< J){
^ mole eq mR Rra
mg/1, H20
Otonp l.ojit in Vent:
Haler Vapor Pressure
Wet Te9t Wnfr
Meter Vnpor
Trmpera- TteRRure
ture ("F) (in HiO)
'.2-47
47-53
54-59
59-M
64-66
67-65
70-72
73-75
76-77
78-79
79-RO
5
6
7
8
9
10
II
12
13
r-,
Ozone Transfer:
(ozone supply rate (Ibe/day) - oeone lost Ln vent (Ibs/day) (100)
Percent o,.ooe trnn.Fer - **-> ^mf B,lppl/m,. (ib./d.y)
(Ihs/dny) - (IbB/dny) (100)
(Iha/dayl
Figure 14. Example data sheet for ozone off-gas concentration measurements.
-------
3. THE EFFECTS OF OPERATION AND MAINTENANCE PRACTICES ON SELECTED
OZONE AND ULTRAVIOLET DISINFECTION SYSTEMS
by: Randy Junkins, Manager O&M Section
WESTON, Designers-Consultants
West Chester, Pennsylvania
INTRODUCTION
Increased attention has been given in recent years to the
disinfection of municipal wastewaters via methods other than
conventional chlorination. Two alternative approaches that have
generated particular interest are ozone and ultraviolet light
(UV) disinfection. As part of EPA's efforts to compile and sub-
sequently promulgate design and operational information concern-
ing these two technologies, Roy F. Weston, Inc. was contracted
(EPA Contract No. 68-03-3019) to identify operations and main-
tenance factors affecting the performance of ozone and UV dis-
infection systems. This paper presents the study methodology
utilized by WESTON and discusses the project results to date.
The objective of the nine-month study is to determine, ana-
lyze, and prioritize those O&M factors that affect the opera-
tional efficiencies of ozone and UV disinfection systems. During
the study, on-site evaluations will be conducted at 15 municipal
wastewater treatment plants that utilize either ozone or UV dis-
infection. During these plant visits, operating personnel will
be interviewed, operational practices will be observed, and op-
erating data reviewed in order to establish O&M causative fac-
tors relating to poor and efficient process performance. This
information will be documented in individual plant evaluation
reports.
The project final report will integrate the data collected
and observations made at the 15 treatment plants into a compre-
hensive O&M document. The report will present the O&M problems
encountered, conclusions drawn concerning their cause, and rec-
ommendations made toward their resolution. Recommendations pre-
sented will address operating practices, process changes, moni-
toring and sampling techniques, staffing requirements, operator
training, and maintenance procedures.
STUDY METHODOLOGY
The study methodology formulated to accomplish the project
goals is shown in Figure 1. Initially, existing data concerning
the operation of ozone and UV disinfection systems and descrip-
tive information about the 15 treatment plants to be evaluated
will be collected and reviewed. Simultaneously, a preliminary
telephone survey of the treatment plants will be conducted in
359
-------
EPA
Authorizes
Project
Four Months
Review Existing
Data
EPA Reports
State Records
Vendor Material
Literature
Conduct Preliminary
Telephone Survey
Conduct Site
Visits
Two Months
• Review Plant
Operating Data
interview Plant
Personnel
• Conduct Sampling
As Needed
• Take Photographs of
Oa/UV Systems
Collect WWTP Design
Information
Determine Oa/UV
System Status
• Schedute Plant Trips
Three Months
Perform Plant
Evaluations
Prepare Comprehensive
O&M Document
• Prioritize O&M
Factors Identified
Analyze Cause/Effect
Relationships Between
O&M Factors and System
Performance
Prepare Plant Trip Reports
Documenting Observations,
Conclusions, and
Recommendations for
Improved Operation
• Integrate Information
Collected at Individual Sites
Discuss Study Conclusions
• Present Recommendations
to Resolve Problems
Encountered
• Prepare Cost Analysis
FIGURE 1 PROJECT METHODOLOGY
-------
order to gather design information and schedule plant trips.
Site visits will subsequently be made to gather plant operating
data and identify those O&M factors which affect the performance
of ozone and UV disinfection systems. Following completion of
the field trips, the data collected will be analyzed and cause/
effect relationships between the O&M factors identified and sys-
tem efficiencies will be formulated. Finally, the project re-
sults will be reported in a comprehensive O&M document that in-
cludes recommendations for optimizing process performance.
PROJECT STATUS
Originally the nine-month project was to be initiated during
the first few weeks in 1981. However, due to EPA budget cut-
backs, the project was delayed and WESTON was not authorized to
proceed until late December 1981.
The project is presently in the initial data collection and
review phase. The preliminary telephone survey has also been
started and is the primary source of the data presented in this
paper. It is anticipated that the treatment plant visits will
begin within the next two weeks. The project is scheduled for
completion in September 1982.
PRELIMINARY PROJECT RESULTS
Information collected to date concerning the treatment
plants to be visited, and specifically their disinfection unit
operations, is discussed below. The preliminary project results
presented will be confirmed, expanded, and refined following the
individual site visits.
PLANTS UTILIZING OZONE DISINFECTION
Descriptive information about the plants which use ozone
disinfection, and design data concerning their disinfection sys-
tems are presented in Tables 1 and 2. It can be seen that plant
hydraulic sizes ranged from 303 to 71,915 m3/d (0.08 to 19
mgd) and ozone system capacities varied from 6.35 to 1905 kg.d
(14 to 4,200 ppd) . It is also noted that fifty percent of the
plants contacted utilized a pure oxygen activated sludge process
which included second stage nitrification. The number of plants
which used air and oxygen as the ozone carrier gas was also
evenly divided.
Information concerning disinfection systems performance and
various operation and maintenance considerations is presented in
Tables 3 and 4. It can be seen that both operational efficien-
cies and costs varied greatly.
361
-------
TABLE 1. OZONE WWTP'S - PLANT DESCRIPTIONS
Plant
Plant age
no. yrs.
1 2
2 2
3 1.5
4 1.5
5 4
6 1
7 2
8 3
Avg . flow Eff. quality
Type*
RBC
02~A.S.
+ Nit.
A.S.
O.D.
02~A. S.
+ Nit.
RBC + A.S.
02-A. S.
+ Nit.
02-A. S.
+ Nit.
m-Vd
(mgd)
303
(0.08)
45,420
(12.0)
1,325
(0.35)
17,033
(4.5)
71,915
(19.0)
7,570
(2.0)
18,925
(5.0)
10,977
(2.9)
BOD
mg/1
12
20
3
20
2
10
9
3
SS
mg/1
3
30
2
20
2
12
8
3
Pre-
treatment
Filt.
None
Filt.
None
Filt.
Filt.
Filt.
Filt.
*RBC - Rotating Biological Contactor
A.S. - Activated Sludge
O^-A.S. - Pure Oxygen Activated Sludge
Nit. - Nitrification
O.D. - Oxidation Ditch
362
-------
TABLE 2. OZONE WWTP'S - OZONE SYSTEM DESCRIPTIONS
Plant Number
no. generators
1 2
2 17
3 2
4 3
5 13
6 2
7 3
8 3
Cells/gen.
8
90
16
109
90
72
72
90
Capacity
total
kg . d
(ppd)
6.35
(14)
1,905
(4,200)
77.1
(170)
272.2
(600)
1,562.7
(3,445)
113.4
(250)
340.2
(750)
571.5
(1,260)
Carrier Oz. transfer
gas efficiency
Air
o2
Air
Air 60%
02 86%
Air 67%
02 84%
02
363
-------
TABLE 3. OZONE WWTP'S - OZONE SYSTEM PERFORMANCE
Contact Ozone
Plant Operational mode time dosage Coliform count/100 ml
no. Auto. Manual min. mg/1 Inf. Eff.
1 x 15 9 3,800
2 x 15 22 --- 1,000
3 x 300 3.5 18,000 12
(60 min. design)
4 x 50 3.0 350,000 200
5 x 30 3.0 --- <1
6 x 60 10.0 30,000 3,000
7 x 40 7 --- 100
8 x 75 6 20,000 100
364
-------
TABLE 4. OZONE WWTP1 S - O&M CONSIDERATIONS
Plant
no .
1
2
3
4
5
6
7
8
Maintenance
Downtime Time reqts.
hr s/wk
Minimal
10% 8
Minimal 4
70% 8
High 8-12
30-50% 8
4
Power
requirements Capital
kwh/kg oz $
(kwh/lb oz)
74,000
22 500,000
(10)
11
(5)
26 200,000
(12)
48
(22)
Annual
(zVm3
(jzf/1,000 gal)
0.71
(2.7)
2.01
(7.6)
0.34
(1.3)
0.84
(3.2)
0 .92
(3.5)
365
-------
Typical O&M problems encounteredf and associated suggestions
for improved operation (determined during conversations with
plant personnel) are listed below. Most of the problems identi-
fied were maintenance-related items.
Ozone WWTP's - Typical O&M Problems
a Multiple and frequent Ozone Generator cell failures.
« Silicone control rectifier (SCR) failures.
* Severe foam problems with contact tank gas recovery system.
® Ozone system electronics are complicated, making it diffi-
cult for WWTP personnel to perform routine maintenance and
repair work.
• Corrosion problems with 03 analyzer valve components.
& Ozone contact tanks were constructed below the system con-
trol room and 03 leaks cause instrumentation rubber seals
to corrode.
o System equipment is very noisy.
€» The system includes much equipment which must be maintained.
« Much time is required to continually calibrate system in-
strumentation.
e Ozone generators are a high maintenance item and continual-
ly blow fuses.
© Catalyst poisoning in 03 destruction system.
® Dew point indicators are not reliable.
<& Excessive heat build-up in 03 generator room caused gener-
ator heaters to shut down units.
® A full-time instrumentation person is needed to monitor and
maintain 03 system.
366
-------
Ozone WWTP's - Operator Suggestions
• Ensure an adequate air filtration system is installed.
• An experienced and qualified instrumentation person must be
part of the WWTP's operating staff.
• Ozone contact tanks should be sited away from other build-
ings and equipment to avoid corrosion and safety problems.
• An 03 destruction unit may not be required if contact tank
off-gases can be vented in an isolated area away from WWTP
buildings.
• System should include a carrier gas 03 monitor to aid in
determining when unit needs to be cleaned.
• Provide a dry gas purge system for compressors.
• Provide a foam suppression system in contact tank and for
gas destruction system.
• Provide ample air circulation in 03 generator room to pre-
vent excessive heat build-up.
• Equipment should be housed inside a building for protection
and prevention of corrosion and freezing problems.
• The ozone generator room must be kept very clean, otherwise
the generator cells will short-out and blow fuses.
PLANTS UTILIZING UV DISINFECTION
Design information about the two plants contacted that util-
ize UV disinfection is presented in Tables 5 and 6. Data con-
cerning the performance of their UV systems is shown in Table 7.
It can be seen that both systems operate very effectively. Al-
though no cost data were available from the individual plants,
cost information presented in a previous EPA report on UV disin-
fection is indicated in Table 8. Typical O&M problems reported
during the telephone survey are listed.
367
-------
TABLE 5. ULTRAVIOLET WWTP'S - PLANT DESCRIPTIONS
Plant Age Type*
no . mo .
1 3 A.S.
2 4 Aero Lag.
Avg . flow
rrP/d
(mgd)
5,678
(1.5)
7,570
(2.0)
Eff. quality;
BOD
rng/1
15
2
SS
mg/1
15
2
Pre treatment
None
Filt.
A.S. - Activated sludge
Aer. Lag. - Aerated lagoon
TABLE 6, ULTRAVIOLET WWTP'S - UV SYSTEM DESCRIPTIONS
Lamp Flow/unit
Plant No, No. output Cleaning m^/d
no. sections lamps/sec. nw/cm2 mechanism (mgd)
90
32
190
Pneumatic
Scraper
Mechanical
Wiper
5,678
(1-5)
8,327
(2.2)
368
-------
TABLE 7. ULTRAVIOLET WWTP'S - UV SYSTEM PERFORMANCE
Contact
Plant time Dosage _ Coliforin count/100 ml.
no. sec.
1 3.5
2 1.5
yW-sec/cmz Inf. Eff.
70
30,000 5 <1
TABLE 8. ULTRAVIOLET WWTP'S -
PREVIOUS EPA STUDY COST DATA
Plant
flow
mgd
1
10
100
Capital Annual O&M
cost jzi/m
$ (<*/!, 000 gal)
80,000 0.66
(2.5)
700,000 0.53
(2.0)
5,200,000 0.48
(1.8)
369
-------
Ultraviolet WWTP's - Typical O&M Problems
9 Ballasts on UV lamps overheat and shut system down.
« Foam build-ups interfere with operation of cleaning mecha'
nism.
• Low flow rate caused unit to overheat.
« Algae accumulations on unit interfere with system operation.
COMPARISON OF OPERATING COSTS
Estimated operating cost for various alternative disinfec-
tion processes are compared in Table 9. These estimates were
prepared as part of a previous EPA study on wastewater disinfec-
tion. The data indicate that UV disinfection appears to be the
most cost-effective strategy for smaller treatment plants that
treat 1 mgd or less, while chlorination (even with de-chlorina-
tion) is the most economical approach for larger plants.
TABLE 9. COMPARISON OF ALTERNATIVE DISINFECTION
PROCESSES OPERATING COSTS
Flow
mgd
1
10
100
Ultraviolet
1.19
(4.5)
0.95
(3.6)
0 .82
(3.1)
Chlorination
1.69
(6.4)
0 .74
(2.8)
0 .58
(2.2)
Chlorination/
De- Chlorination
2.19
(8.3)
0 .82
(3.1)
0.61
(2.3)
Ozonation-air
3.41
(12.9)
2.01
(7.6)
1.45
(5.5)
Costs defined as jzf/cm3 (gf/I,000 gal.).
370
-------
GENERAL OBSERVATIONS
The following preliminary general observations are made:
Ozone Systems
1. Many of the ozone systems surveyed had only been in opera-
tion three to four months.
2. Minimal cost and maintenance requirements information is
currently available for full-scale installations.
3. Very few of the same problems were encountered at the treat-
ment plants surveyed.
4. The consensus of opinion among the plant operators inter-
viewed is that chlorine disinfection systems are more reli-
able, less expensive to operate, and require less mainte-
nance than ozone systems.
Ultraviolet Systems
1. Presently there are very few full-scale UV systems on-line.
2. Available O&M data concerning UV disinfection are minimal.
3. Those UV systems which are on-line were reported to be reli-
able unit operations.
REFERENCES
1. EPA Project Summary Report; EPA 600/52-81-152, Sept. 1981.
371
-------
4. SECOND NATIONAL SYMPOSIUM ON MUNICIPAL WASTEWATER DISINFECTION
SUMMARY AND CLOSING REMARKS
Mr. Charles C. Johnson, Jr.
C. C. Johnson & Associates, Inc.
11510 Georgia Avenue, S-220
Silver Spring, MD 20902
INTRODUCTION
Water, more perhaps than any other medium, illustrates the recycling
process that takes place in nature. All waste that is discharged to the
biosphere - biological,chemical,and physical-sooner or later finds its way into
the earth's water. That water must be cleansed by nature or by man before it
is again safe for human use and consumption. For 2-1/2 days some 250 persons
from 33 states and 2 foreign countries have been discussing this phenomenon as
it relates to wastewater treatment plant effluents. Fifty percent of these
persons are representatives of government agencies, 20 percent equipment manu-
facturers and suppliers, 20 percent consulting engineers and 10 percent academ-
icians. During the next few minutes I will try to capsulize the papers and
discussions of this conference and along the way add some comments that reflect
my own point of view on wastewater disinfection. To do this I will divide my
remarks into two general segments - health consideration associated with dis-
charge of wastewater and the concerns of this conference, and the technologi-
cal considerations related to disinfection practices.
HEALTH CONSIDERATIONS
Health considerations we have discussed embrace - arguments for and
against chlorination of sewage effluents. Jim Coulter presented strong
arguments and factual data against chlorination of Wastewater Treatment Plant
effluents citing the negative impact on fish populations in Maryland. Henry
Ongerth countered forcefully citing the generally recognized preventive health
arguments against the uncontrolled discharge to the environment of waste
potentially harmful to humans. The question is now raised as to whether there
must be this confrontation between fish and people? Before I attempt an answer
lets look at other discussions that shed light on the basic question of what
public health evidence supports the need for disinfection of these effluents.
Elmer Akin provided information on "Infective Dose of Waterborne
Pathogens." He reported on studies involving bacterial, protozoal and viral
pathogens. The data indicated that all these categories of enteric pathogens
can produce infection and/or illness at very low exposure levels. All the doses
were administered to the human volunteers by the oral route. As recognized by
one participant, the lesson learned in this case is clear - don't drink
contaminated water.
372
-------
Dr. Dolin related a case study on an outbreak of viral gastroenteritis
caused by a Norwalk-like virus. The outbreak accurred after a water supply
source was contaminated with sewage from a broken sewerline. Once again we
prove the point - don't drink contaminated water.
Now I am not being facetious in my comments on these two studies. Un-
fortunately, humans inadvertently do ingest water when swimming and these
waters when contaminated are capable of producing disease. The question would
appear to be what level of contamination can be considered reasonably safe under
these conditions?
Dr. Hubly provided some incite into risk assessment as related to the use
of chlorine in wastewater disinfection. His investigation uncovered very
limited to nonexistant historical data. We all know that risk assessment of
potentially harmful environmental impacts is an emerging science and in the
absence of background data little reliance can be placed on predictions of risk
associated with the use of chlorine in wastewater disinfection. All in all
though it would seem that the risk is minimal.
The luncheon speaker Dr. Arthur Lane of the Jet Propulsion Laboratory took
us on a most interesting and exciting photographic voyage to Jupiter and Saturn.
The pictures were simply fantastic.
Once we were back to earth, Walter Jakubowski reported on a series of
epidemiological studies of community exposures to aerosols from wastewater
treatment plants and a study of worker exposure to aerosols, sewage liquids, and
solid contacts. One could not conclude from these study results that the
communities were harmed by aerosol spray eminating from the wastewater treat-
ment plant. With respect to sewage treatment plant workers, inexperienced
workers evidenced higher rates of gastrointestinal symptoms than did experi
enced workers or controls. The symptoms were mild and transitory and did not
result in time lost from work. Pathogen isolation did not indicate any in-
crease risk from sewage exposure. The results of two studies from Israel re-
lated to spray irrigation of sewage were found to be inconclusive because of
the poor quality of the data. A third study is underway.
Dr. Cabelli's epidemiological studies enabled the development of a pre-
dictive model which is intended to aid in the determination of requirements for
wastewater disinfection of bathing beaches. While the model has not been
validated it offers a tool that is otherwise unavailable for this purpose.
Perhaps the validation should and can be obtained through its use.
While Dr. Cabelli's work was associated with marine waters, Dr. Dufour
investigated fresh recreational water quality and swimming associated ill-
nesses. General conclusions would suggest that as water quality deteriorates
the potential for disease increases. Further, as in marine waters, entercocci
probably represent the best indicator organism for recreational waters. Because
fresh water swimming associated gastrointestinal rates are lower than in marine
water, different water quality standards should apply. One additional thought
is worth recalling. The unanticipated illnesses associated with waters of rela-
tively high quality may be associated with particle ingestion. This observa-
tion certainly warrants further study.
373
-------
We should ask ourselves how is all this related to discussions by Coulter
and Ongerth? It tells me that the state of knowledge with respect to the
potential for harm to the public health from discharge of unchlorinated sewage
to the environment is a big question mark. On the other hand the discharge of
chlorinated sewage as presented by Coulter appears to be harmful to fish. Now
while we do not know what the harm to persons is^, no one argues that a benefit
to people is associated with such a practice. Further we will soon hear that
disinfection of Wastewater Treatment Plant effluents does not automatically
require chlorination. Finally when the economic value of protecting the fish,
or the accepted public health risk to the people, dictates a change in current
practices, these practices will be changed. Until then we will continue to
accommodate to the maximum extent possible the desires of the Coulters and the
Ongerths on a case by case basis .
TECHNOLOGY CONSIDERATIONS
Now we can't really get Jim Coulter out of his dilemma unless we have
useable alternatives to existing disinfection practices. Also economics and
improved efficiencies in current practices are always welcomed by operators of
Wastewater Treatment Plants. With this in mind chlorination, ultraviolet, and
ozone disinfection practices were discussed.
Chlorination
Better mixing means better disinfection. Dr. Longley presented a paper
which showed that rapid mixing of chlorine with the wastewater stream initially
provides increased contact between the bacteria and virus with chlorine before
the chlorine is dissipated in other reaction pathways. He studied several
models claimed to be useful in disinfection process design. He concluded that
the Prandt Eddy frequency and the mean velocity gradient have properties which
make them useful for disinfection process design. He offered a pipe mixer and a
venturi mixer as advances in technology for this purpose.
A simple, inexpensive, but effective modification to existing chlorination
contact chambers was presented by Fred Hart. Baffling to produce plug flow and
eliminate short circuiting was illustrated. The result was a 9 to 15 percent
savings in chlorine usage.
George White reported on problems of disinfecting nitrified effluents at
the San Jose/Santa Clara Water Pollution Control Plant. It was noted that the
nitrified effluent created an exceptionally high chlorine demand. The applica-
tion of amonia nitrogen reduced the chlorine demand and still enabled the plant
to meet its NPDES Permit requirement of 2.2/10CmL-MPN Total Coliform in the ef-
fluent. The surprising factor in this investigation was that the combined
chlorine residual was found to be much more reliable in its germicidal effi-
ciency than a free chlorine residual.
374
-------
Ultraviolet Disinfection
Many of us were hoping, even expecting, that discussions on ultraviolet
units would offer an immediate alternative to chlorination as a disinfectant.
While this may be true under limited circumstances there would appear to be
considerable work and study required before UV becomes fully competitive with
chlorination.
Mr. Nehm presented results of pilot plant studies of 4 manufacturers' UV
units. The test indicated that all units were capable of achieving the effluent
fecal coliform goal of 200/100 mL. However, the useful in-service time of all
units was limited by the formation of scale on the tubes. While this experience
is limited to one quality water, this has always been a problem with UV
disinfection, even with relatively cleaner drinking water. Until a solution is
found that extends the useful life of the tubes to an acceptable period of time,
it would seem that costs associated with operation and maintenance of the units
will limit their application in wastewater treatment.
Donald Johnson looked at UV disinfection of filtered and unfiltered
effluents under thick and thin film conditions. Given the proper dose (in-
tensity x time) the UV units tested produced a 200/lOCmL fecal coliform count
from both filtered and unfiltered effluents. I don't think anyone questions the
ability of UV to disinfect. The question is for how long and at what cost.
Perhaps the work now getting underway at the Port Richmond Plant in New
York will shed more light and provide more answers to these operation and
maintenance problems. Karl Scheible says they are in phase one of an 18 month
study. Their protocol provides for study of water quality, system geometry,
system hydraulics, and equipment specification.
Ozone
Discussion of ozone technology essentially concerned analysis and control
methodology. It is believed by some that analytical methods for determining
residual ozone require some attention. Gilbert Gordon's paper suggests that
electronic monitoring is simply not satisfactory and in wet analysis the
stability of some indicator 'solutions is a problem. Indigo and arsenic were
determined to be the most reliable. The effects of ozone decomposition and the
pH of the effluent also were discussed.
Control of the ozone disinfection process by monitoring of exhaust gas was
presented by Al Venosa. This approach is said to measure true ozone, require
only one measurement, capable of being automated, accurate, sensitive, and
stable. It requires a constant gas to liquid flow ratio. When this is done the
dose can be controlled by monitoring gas in the exhaust and automatically
signaling changes in the power to the generator.
Enos Stover told us how to optimize operational control of ozone dis-
infection. This should be considered in the design stages of the facility. The
disinfection process must be effective, reliable, economic, safe, and require
375
-------
minimal power, and maintenance. He noted that maximum dose applied may not
permit maximum ozone transfer and lowest power consumption. He offers 2 ways to
optimize your situation.
After investigating ozone disinfection and transfer in wastewater, Patrick
Given shared results of a Canadian experience with us. The study involved
screened, dilute wastewater. The results indicated not only a reduction of
99.9 percent in indicator organisms but a substantial wastewater strength
reduction produced high dissolved oxygen levels in the effluent. Wastewater
strength was an important factor in these studies.
Ed Opatken offered to convince us that it is possible to calculate ozone
mass transfer coefficients. Then he told us there is no such thing (it varies
with gas flow), and also that mass balance is enhanced by secondary effluent
when compared with tap water. He says that mass transfer coefficients for ozone
that were obtained from these studies can be used for the design and scale-up of
ozone bubble diffuser contactors of plants where the primary source of the
wastewater is of domestic origin.
Peter Foller presented a future new process for generation of ozone. It
promises reduced capital cost (reductions of 50 percent) and easier scale up and
scale down of equipment. Use is approximately 2 year away for a 10 pound per day
level of ozone use.
The last session, just completed, related some practical considerations in
the use of halogens, disinfectants, and in the design, operation, and
maintenance associated with ozone and ultraviolet disinfection. Charles Haas
reminded us that problems still exist in the design of disinfection systems
using halogens. Some of these problems are associated with dose estimation,
contactor hydraulics, process control, chemical supply, and safety.
Kerwin Rakness was concerned with the application of the correct ozone dose
and the reliable and economical production of ozone. The significance and
sensitivity of pretreatment of air was emphasized. Use of air in the ozone
generator with a dew point of less than-55°C is considered optimum. In some
situations manual control of ozone application may be preferable to use of
automated controls.
Randy Junkins reported on a study just getting underway that will evaluate
and document the effects of operation and maintenance practices on the
performance of ozone and UV systems. It is much too early to draw any con-
clusions from the effort, either as related to efficiency, reliability or costs.
Karl Scheible says that disinfection of treated wastewater by ultraviolet
irradiation has emerged as an accepted, feasible, cost-effective alternative.
In my opinion this statement is open to considerable question. Regardless,
initial steps to produce a UV design manual, sponsored by the EPA, are underway.
It is proposed to feature the latest developments in the state of the art as they
are recognized and practiced today.
376
-------
CONCLUSION
During the course of this symposium a great deal of information has been
presented and we now should be convinced that chlorination, ultraviolet, and
ozone processes can be used under varying degrees of difficulty to disinfect
wastewater effluents. Unfortunately no information has been supplied as to the
relative economics associated with the capital cost and operation and main-
tenance cost for these processes. Until this is done Jim Coulter is hard put to
press his desire to eliminate chlorination as the primary disinfectant for
wastewater effluents.
About the Authur
C. C. Johnson is President of C. C. Johnson & Associates, Inc. an en-
vironmental engineering consulting firm. He is a retired Assistant Surgeon
General of the PHS and former chairman of the National Drinking Water Advisory
Council.
377
SUSGPO: 1983 — 659-095/0713
------- |