&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

-------