United States
            Environmental Protection
            Agency
             Water Engineering
             Research Laboratory
             Cincinnati OH 45268
EPA/600/9-88/004
March 1988
&EPA
            Research and Development
Proceedings:

Conference on Current
Research in Drinking
Water Treatment

-------

-------
                                DISCLAIMER


      The following papers have been reviewed in  accordance  with  the  U.S.
Environmental Protection Agency's peer and administrative  review  policies
and approved for presentation and publication:

           GAC for Removing Trihalomethanes
           Control of Trihalomethanes Using Alternative Oxidants  and
             Disinfectants
           THMFP Reduction by Low Pressure Membranes
           GAC and RO Treatment for the Removal  of Organic Contaminants
             from Ground water
           Point-of-Entry/Point-of-Use Treatment for Removal of
             Contaminants From Drinking Water
           Evaluation of Radium Removal and Radium Disposal  for a Small
             Community Water Supply System
           Radon Removal From Groundwater Using GAC
           Cost of Drinking Water Treatment
           Factors Affecting the Inactivation of Giardia Cysts by
             Monochloramine and Comparison With Other Disinfectants
           Inactivation of Hepatitis A Virus and Model Viruses in Water
             by Free Chlorine
           Detection and Control of Chlorination Byproducts in Drinking
             Water
      The following papers describe work that was funded by the AWWARF and
therefore the contents do not necessarily reflect the views of the Agency
and no official endorsement should be inferred:

           American Water Works Association Research Foundation
             Trihalomethane Survey—A Progress Report
           Development of Rapid Small-Scale Adsorption Tests
           Removal of Volatile Organic Chemicals From Air Stripping
             Tower Off-Gas Using Granular Activated Carbon
           Impacts of Regulatory Requirements on Handling Water Plant
             Wastes
           A Study of Water Treatment Practices for the Removal of
             Giardia  Iambiia Cysts
           Removal of Giardia in Low Turbidity Water by Rapid Rate
             Filtration
           GAC Substitution for Sand
           The Characteristics of Initial Effluent Quality and Its
             Implications for the Filter-to-Waste Procedure

-------
                                 FOREWORD

     The U.S. Environmental Protection Agency is charged by Congress with
protecting the Nation's land, air, and water systems.   Under a mandate of
national environmental laws, the Agency strives to formulate and implement
actions leading to a compatible balance between human  activities and the
ability of natural systems to support and nurture life.  The Clean Water
Act, the Safe Drinking Water Act, the Resource Conservation and Recovery
Act, the Federal Insecticide, Fungicide and Rodenticide Act, and the Toxic
Substances Control Act are five of the major congressional  laws that
provide the framework for restoring and maintaining the integrity of our
Nation's water, for preserving and enhancing the water we drink, and for
protecting the environment from hazardous and toxic substances.  These laws
direct the EPA to perform research to define our environmental problems,
measure the impacts, and search for solutions.

     The Water Engineering Research Laboratory is that component of EPA's
Research and Development Program concerned with preventing, treating, and
managing municipal and industrial wastewater discharges; establishing prac-
tices to control and remove contaminants from drinking water and to prevent
its deterioration during storage and distribution; and assessing the nature
and controllability of releases of toxic substances to the air, water, and
land from manufacturing processes and subsequent product uses.  This publi-
cation is one of the products of that research and provides a vital com-
munication link between the researcher and the user community.

     The Conference on Current Research in Drinking Water Treatment was
held to provide progress reports on several active research projects spon-
sored by either U.S. EPA Drinking Water Research Division or the American
Water Works Association Research Foundation.  The topics of the papers pre-
sented addressed three significant aspects of the 1986 amendments to the
Safe Drinking Water Act:  contaminant regulations (MCLs), filtration of
surface water sources, and disinfection of public water supplies.  Because
of the widespread interest in this Conference, these Proceedings were pre-
pared to provide information to the many individuals who have a special
interest in the research information presented.  It is hoped that the con-
tents of these Proceedings will assist the many Federal, state and local
officials, utility managers, and consultants who are impacted by the 1986
amendments to the Safe Drinking Water Act.


                         Francis T» Mayo, Director
                         Water Engineering Research Laboratory
                                    iri

-------
                                  PREFACE

     During each fiscal year of 1984, 1985 and 1986,  Congress  appropriated
$1 million to support a cooperative research agreement between U.S.  EPA and
American Water Works Association Research Foundation  (AWWARF).  With an
equal amount of matching monies, AWWARF has funded over 45 research  pro-
jects on drinking water topics.  Concurrent with this program, the Drinking
Water Research Division (DWRD) of the Office of Research and Development,
U.S. EPA, continued to operate its extramural and inhouse research program
supporting a large number of research projects on many aspects of drinking
water treatment.

     Since the initiation of the cooperative program between the AWWARF and
U.S. EPA, the two groups have coordinated their research planning efforts
to avoid duplication of effort and to establish research priorities for
their respective programs.  Many of the research efforts of both programs
have assisted the water supply industry in meeting the regulatory program
of the U.S. EPA.  Of major significance to the industry are the 1986 amend-
ments to the Safe Drinking Water Act.  These amendments are far reaching
and will require changes in operation for many water utilities.  Many of
the research projects  supported by the U.S.  EPA and AWWARF address
three of the most significant aspects of the new amendments:  con-
taminant regulations (MCLs), filtration of surface water sources, and
disinfection for public water supplies.  The objective of this Conference,
therefore, was the presentation of research  results from U.S. EPA/AWWARF
sponsored research efforts associated with these three aspects of the new
amendments.

     With the cooperation of the U.S. EPA Center for Environmental Research
Information (CERI), the Conference was planned and organized  by the U.S.
EPA and AWWARF and held at the U.S.  EPA Research Center  in Cincinnati, OH,
March 24-26, 1987.  The Conference was well  attended by many  Federal,
state, and local officials, utility managers, consultants, and equipment
manufacturers.  Because of the  interest  in the  information, both by those
in attendance and by many others who could not attend, these  Proceedings
were prepared.  We sincerely hope  that the information will be of lasting
value to the water supply industry and express our appreciation to the many
people who contributed to the success of  this Conference.

          Thomas J. Sorg                       James F. Manwaring
  Drinking Water Research Division       American Water Works Association
U.S. Environmental Protection Agency           Research Foundation
                                     IV

-------
                                 ABSTRACT
      A Conference on Current Research in Drinking Water Treatment was held
at the U.S. EPA Andrew W. Breidenbach Environmental Research Center in
Cincinnati, Ohio on March 24-26, 1987.  The speakers at this Conference
were principally researchers funded in part by the U.S. EPA (Drinking Water
Research Division) or the American Water Works Association Research
Foundation.  The purpose of the Conference was the presentation of research
results from current research projects having direct application to three
important aspects of the 1986 amendments to the Safe Drinking Water Act:

      o  Contaminant regulators (MCLs)
      o  Filtration of surface water sources
      o  Disinfection for public water supplies.

      This publication is a compilation of either extended abstracts or
full papers prepared by the speakers and their co-authors.

-------
                    METRIC CONVERSION CHART
 in 4- 39.37 = m                          oz  x 28.35  =  g
 in x 2.54 = cm                          Ib  x 0.45 = kg
 ft 4 3.28 = m                           tons x  907.18 =  kg
 ft x 30.48 = cm
 pm x 0.000001 - m                       psi 4- 0.14  =  kPa
 mi x 1.61 = km                          lb/in2  x 0.07 =  kg/cm2
                                        lb/ft2  x 4.88 =  kg/m2
 in2 x 645,16 = mm2
 ft2 4 10.76 = m2                        Ib/ft3  x  16.02  = kg/m3
 mi2 4- 0.39 = km2
                                        ft/s 4-  3.28 = m/s
 oz x 29.57 = ml
 gal x 3.78 = 1                          g/5 (oc)  +  3?_ =  op
 gal 4- 264.20 = m3
 1n3 * °'06 ' cm3                        ppb x  1 - yg/1
 ft3 x 28'32 = ]                         ppm x  1 . mg/1
 ft3 4- 35.31 = m3
                                        gpm/ft2 4- 0.41  = m3/m2/h
 gpm 4 15.85 = 1/s
                                        gpd/ft2 4  24.57  = m3/mz/d
 gpm 4- 15,852.00 = m3/s
 gpd 4 22.83 = 1/s                                        -
                                        pCi/1 x 37  •  Bq/mJ
mgd 4- 22.83 = m3/s
 cfs 4- 35.31 = m3/s
                               vi

-------
                                 CONTENTS


Foreword	   Hi
Preface	    1 v
Abstract	     v
Metric Conversion Table	    vi

American Water Works Association Research Foundation
Trihalomethane Survey -- A Progress Report
      Michael J. McGuire, Metropolitan Water District of Southern CA
      Robert G. Meadow, Decision Research	     1

GAC For Removing Trihalomethanes
      Benjamin W. Lykins, Jr. and Robert M. Clark, U.S. Environmental
      Protection Agency	    15

Control of Trihalomethanes Using Alternative Oxidants and
Disinfectants
      Philip C. Singer, University of North Carolina	    23

THMPF Reduction by Low Pressure Membranes
      J. S. Taylor, University of Central Florida	    27

Development of Rapid Small-Scale Adsorption Tests
      David W. Hand and John C. Crittenden, Michigan Technological
      University
      John K. Berrigan, Zimpro Inc.
      Benjamin W. Lykins, U.S. Environmental Protection Agency	    41

Removal of Volatile Organic Chemicals From Air Stripping
Tower Off-Gas Using Granular Activated Carbon
      John C. Crittenden, Shin-Ru Tank, David Perram and Tim Rigg,
      Michigan Technological University
      Randy D. Cortright, Universal Oil Products
      Brad Rick, Amway Corporation	    54

GAC and RO Treatment For the Removal of Organic Contaminants
from Ground Water
      Joseph H. Baier, Suffolk County Department of Health Services...    90

Point-of-Entry/Point-of-Use Treatment for Removal of Contaminants
From Drinking Water
      K.E. Longley, G.P. Hanna, Jr. and B.H. Cump, California State
      University - Fresno 	   110
                                     vi

-------
 Evaluation of Radium Removal and Radium Disposal  for a Small
 Community Water Supply System
      Kenneth A. Mangelson, Rocky Mountain Consultants, Inc...........   125

 Radon Removal from Ground Water Using GAC
      N.E. Kinner and C.E. Lessard, University of New Hampshire
      J. Lowry, University of Maine
      H. Stewart and R. Thayer, N.H. Dept. of Environmental
      Servi ces.	   134

 Impacts of Regulatory Requirements on Handling Water Plant Wastes
      David A. Cornwell, Environmental Engineering & Technology, Inc..   138

 Cost of Drinking Water Treatment
      Richard G. Eilers, U.S. Environmental Protection Agency.........   147

 Regulations on Filtration and Disinfection
      Stig Regli, U.S. Environmental Protection Agency................   151

 A Study of Water Treatment Practices for the Removal of Giardia
 lamb Ha Cysts
      Jerry Ongerth, University of Washington.........................   171

 Removal of Giardia in Low Turbidity Water by Rapid Rate Filtration
      Ron R. Mosher, Molzen-Corbin & Associates
      David W. Hendricks, Colorado State University...................   176

 GAC Substitution for Sand
      Sandra L. Graese and Vernon  L= Snoeyink, University  of  Illinois
      at Urbana-Champaign
      Ramon G. Lee, American Water Works Service Company..............   189

 The Characteristics of Initial Effluent Quality and Their
 Implications for the Filter-to-Waste Procedure
      Karen Bucklin and Kelly 0. Cranson, Montana State University
      Appiah Amirtharajah, Georgia  Institute of Technology	   217

 Factors Affecting the Inactivation of Giardia Cysts by
 Monochloramine and Comparison with Other Disinfectants
      Alan J. Rubin, The Ohio State University........	   224

 Inactivation of Hepatitis A Virus and Model Viruses in Water
by Free Chlorine
      Mark D. Sobsey and Taku Fuji, University of North Carolina
      Patricia Sheilds, University of North Carolina.......	   230

Detection and Control of Chlorination Byproducts in Drinking Water
      A.A.  Stevens, R.J. Miltner, L.A. Moore, C.J. Slocum,
      H.D.  Nash, D.J. Reasoner, and D. Berman, U.S. Environmental
      Protecti on Agency.	.................••••••	  242

-------
           AMERICAN WATER WORKS ASSOCIATION RESEARCH  FOUNDATION
                TRIHALOMETHANE SURVEY —  A PROGRESS REPORT

                       by:  Michael  J.  McGuire
                            Metropolitan  Water  District  of  Southern
                            California
                            Los Angeles,  CA  90054

                            Robert G. Meadow
                            Decision Research
                            San Diego,  CA  92101
                               INTRODUCTION

    On November 29, 1979, the trihalomethane (THM)  maximum contaminant
level (MCL) of 0.10 mg/1 (100 ppb) was promulgated  by the U.S.  Environmen-
tal Protection Agency (EPA).  Since the MCL was based primarily on tech-
nical feasibility, EPA has given clear indications  that it intends to
significantly reduce the MCL.

    On June 19, 1986, the Safe Drinking Water Act Amendments of 1986 were
signed into law.  As part of the regulatory timetable established by
Congress, EPA recently announced that it intends to establish new MCLs for
disinfection by-products, including THMs, as part of the first group of 25
standards that are due to be finalized by 1991.

    The purpose of this paper is to present preliminary results of a
national survey of trihalornethanes in drinking water.  At the time of this
writing (March 1987), several additional survey questionnaires have been
received from large cities, as well as from at least one utility with very
high THM levels.  While we believe that neither the statistical data nor
the conclusions presented in this paper will change significantly, the data
herein should be considered preliminary.  The final report and an article
to be submitted to the Journal^ of the American Water Works Association will
contain the finalized data.  It is anticipated that this survey will form
the basis for determining costs and the feasibility of the water utility
industry complying with a new THM standard.

    The idea for the THM survey was developed by a committee of the
Association of Metropolitan Water Agencies.  The survey was carried out by
the Metropolitan Water District of Southern California and Decision
Research under partial funding from a grant by the AWWA Research Foundation
(AWWARF).

-------
                              SURVEY METHODS

    Because all water utilities serving more than 10,000 people are
 required to monitor for THMs, the survey was designed to sample this infor-
 mation by means of a questionnaire sent directly to the water utilities
 (see Attachment A).  This study determined that there was no comprehensive
 national THM data base in existence at EPA, AWWA, or any national organiza-
 tion.  All 50 states have THM records for the water utilities they supervise
 for compliance, but obtaining the data from files or noncompatible data
 bases was deemed an inefficient survey method.

    The full report on this project describes how the utility questionnaire
 was developed, reviewed, and finally approved.  A total of 1,255 question-
 naires were sent out in January 1987 to sample THM data from the 3,081
 utilities serving more than 10,000 people.

    Because it is possible that the new THM standard could be applied to
 utilities serving fewer than 10,000 people, an attempt was made to gather
 representative data from the more than 55,000 utilities in this category.
 A simple questionnaire (Attachment B) was sent to all 50 states and terri-
 tories to determine whether THM monitoring and MCL compliance are required
 for the small utilities.  The states were asked to send summarized THM data
 on the smaller utilities.

                          RESULTS AND DISCUSSION

 UTILITY SURVEY

    Table 1 summarizes the number of utilities and population sampled by
 the AWWARF THM survey as compared to national statistics.  The AWWARF THM
 data are based on up to 12 quarterly averages during the period 1984 to
 1986.  Table 2 compares the results of the AWWARF survey with the results
 of the two previous surveys by EPA-NORS in 1975 and NOMS from 1976 to 1977-
 Comparing the AWWARF THM overall average of 42 ppb with the averages of the
 NORS and NOMS, all phases show a 40 to 50 percent reduction in national THM
 levels resulting from compliance with the THM standard.

    Figure 1 is a log THM-frequency distribution graph of the same data as
 represented on Table 2.  Figure 1 shows that  for utilities with THM levels
 of 50 ppb and lower, the occurrence of THMs is about the same for all the
 surveys.  However, compliance with the THM standard has clearly  reduced the
 higher levels of THMs found in NORS and NOMS.  Approximately 225 utilities
 installed a total of 543 treatment changes to come  into compliance with the
THM standard of 100 ppb.  Even with this overall THM reduction on a nation-
wide basis, 38 utilities reported that they violated the THM MCL one or
more times from 1984 to 1986.

    Cost data to achieve compliance with the  100 ppb limit are not
complete, but Table 3 shows that the 1981 estimate made by TBS/Malcolm Pir-
nie for EPA was quite good.  A total of $31 million in capital  costs was
reported by approximately 225 utilities that  estimated costs of ^bo  treat-

-------
                TABLE  1.  AWWARF TRIHALOMETHANE SURVEY
                    POPULATION AND UTILITIES SAMPLED
                                   AWWARF SURVEY

Utilities serving >10,000
Population Retail
Wholesale
Utilities Sampled
1,255

Returns
910 (73%)
105 million
42 million
United States
3,081
171 million
           TABLE 2.   COMPARISONS BETWEEN NATIONAL  THM  SURVEYS

NORSa
NOMS-Phase 1*
NOMS-Phase 2**
NOMS-Phase 3D***
NOMS-Phase 3T**
NOMS-A11 Phases
AWWARF***,****
No.
of Cities
80
111
113
106
105
105/113
727

Mean
68
68
117
53
100
84
42
Tribal omethanes,
Median
41
45
87
37
74
55
39
ppb
Ranget
ND-482
ND-457
ND-784
ND-295
ND-695
ND-784
ND-360
   *Samples shipped and stored at 2 to 8°C for one to two weeks  prior to
    analysis.

  **Samples stored at 20 to 25°C for three to six weeks prior to analysis.

 ***Sodium thiosulfate added.

****Sampled, collected, and analyzed in compliance with THM monitoring and
    analysis regulations.

   tND--None detected.  Detection levels differed significantly between
    the three s
-------
 merit  changes.   Projecting these capital costs to the total number of treat-
 ment  changes  (543)  and  to the population served by the larger utilities
 (>10,000)  results  in  a  total estimated capital cost of $102 million.


                   TABLE 3.  AWWARF TRIHALOMETHANE SURVEY
                      COMPLIANCE WITH  100 ppb STANDARD
                                    AWWARF SURVEY
 	Survey Data   Projections     TBS/MP Estimate

 Number of treatment  changes
   made by utilities               543           —                242

 Capital  Costs                 $31 million   $102 million      $47 million
                                  (268)*

 Operation and                 $ 8 million   $ 29 million      $17 million
   Maintenance  Costs               (241)


 AWWARF Survey;   Number of  utilities that made
                 one  or more  treatment changes  =  225

 *Number  in parenthesis is  the number of treatment changes for which
  responding utilities had  dollar estimates.


     Figure 1 indicates that  26 percent of the utilities could not meet a
 THM  standard of  50 ppb.  Similarly, 60 percent and 82 percent of utilities
 could  not meet THM standards of 25 ppb and 5 ppb, respectively.  Cost esti-
 mates  from water utilities to meet these more stringent standards range in
 the  billions of  dollars.   However, these estimates must be viewed with
 caution,  as they are not based on detailed engineering or feasibility
 studies.

 STATE  SURVEY

     Only  four states (Michigan, New Hampshire, New York, and Rhode Island)
 require utilities serving  fewer than 10,000 people to monitor for THMs, or
 the  states  themselves do the monitoring.  Only New York requires the
 smaller utilities to actually comply with the 100 ppb THM MCL.  Twenty-four
 states responded that they had THM data available for the smaller utili-
 ties,  but  Table  4 shows that to date we have received usable data from only
 12 states.  Table 4 also shows that the number of utilities (677) for which
we received THM  data represents only a small percentage of the total  number
of utilities serving fewer than 10,000 people.

-------
               TABLE 4.   AWWARF STATE TRIHALOMETHANE  SURVEY
                     POPULATION AND UTILITIES  SAMPLED
                                      AWWARF

Utilities serving <10,000
Population
Number of states
State Survey
677
1.6 million
12
United States
55,449
48 million
50
    Table 5 summarizes 2,594 THM data points for these 12 states.   The low
THM results from Wisconsin, with 204 cities sampled,  appear to markedly
affect the overall statistics.  Figure 2 shows that the data from  Table 5
are not representative of the NORS and NOMS data, nor are they represent-
ative of the AWWARF utility survey data from Figure 1.  This lack  of
agreement may be caused by nonrepresentative data in  this survey or by the
possibility that smaller systems use sources that are generally lower in
THM precursors.  Removing the Wisconsin data improves the agreement with
the NORS and NOMS surveys, but there are still significant differences.
More THM data on smaller utilities are certainly needed to construct a
representative picture of how a more restrictive, more widely applicable
THM standard would affect these utilities.

SUMMARY AND CONCLUSIONS

    o  The existing THM regulation of 100 ppb has resulted in a 40 to 50
       percent reduction in average THM levels in the United States.  This
       has cost consumers between $31 million and $102 million.

    o  Reduction of the THM standard to levels of 5 ppb to 50 ppb  will
       result in large numbers of utilities falling out of compliance and
       will potentially cost consumers billions of dollars for the utili-
       ties to come into compliance with a new, more  restrictive standard.

    o  More data are  needed on feasible treatment technologies and costs of
       achieving lower THM standards.

    o  More data are  also needed on smaller systems'  THM  levels.

ACKNOWLEDGEMENTS

    This work was partially supported by a grant from the AWWARF;  special
thanks are extended to Jon DeBoer of that organization for his guidance and
assistance.  Many people at EPA, AWWA, the Association of State Drinking
Water Administrators, and individual state regulatory agencies aided  this

-------
                TABLE  5.  AWWARF STATE TRIHALOMETHANE SURVEY
         TRIHALOMETHANE DATA  FOR UTILITIES SERVING LESS THAN 10,000
State
Alaska
Illinois
Iowa
Maryland
Michigan
Montana
New York
Pennsylvania
Rhode Island
South Carolina
West Virginia
Wisconsin
TOTAL
No. Cities
81
57
17
12
14
9
236
8
11
20
8
204
677

24,304
185,363
44,584
48,204
78,454
32,801
609,500
24,438
38,012
54,445
41,048
435,035
1,616,188
Total
No. THM
Data Points
119
57
65
126
408
60
900
15
11
632
48
THM
Mean
21
56
155
29
78
22
49
32
41
107
53
204 ____2
2,594
Mean =
Mediae ~
DATA, ppb
______Ran£e
ND-184
9-184
ND-292
2-104
23-189
ND-34
4-308
4-63
ND-115
34-313
33-80
ND-42
ND-764
36
18
project, and their help is appreciated.  Special thanks to the personnel in
the 910 water utilities who took time out from their busy schedules to
answer the questionnaires that served as the basis for this report.

-------
         1000-


          500'
CO
m
<
LU .0
2 Q-
g a
<
E
8
i
          50 H
           10-
           5-
                            NOMS-AII Phases
                               Average
                                                     NORS
                                                2^*

                                                 AWWARF
                                                Utility Survey
                 I    I   I  I  I  I   I   T
                10     30  50   70     90  95    99    99.9  99.99
             PERCENTAGE LESS THAN OR EQUAL TO GIVEN CONCENTRATION
                                   90 95
99
                                                       T
                                                            1
Figure 1.   Frequency  distributions of national THM  survey data.
       1000


       500
  CO
  LU
  3
  <
  £
        100-J
    50-J
        10 H
                         NOMS-AII Phases
                             Average
                           \    ''
                                   ^
                                               NORS
                                      AWWARF Survey
                                        (<10,000)
             ~1IIIIIIIIIIII
              10     30   50   70     90  95     99     99.9  99.99
          PERCENTAGE LESS THAN OR EQUAL TO GIVEN CONCENTRATION
Figure  2.  Frequency  distributions of NORS, MOMS,  and AWWARF
            smaller utility survey data.

-------
                                            ATTACHMENT A
                             TRIHALOMETHANE (THM) SURVEY
 ********* URGENT: PLEASE  RESPOND BY JANUARY 27, 1987 *********
 INSTRUCTIONS: FOR EACH QUESTION. PLEASE FILL IN THE BLANK SPACE OR CIRCLE THE NUMBER CORRESPONDING TO YOUR ANSWER
 PLEASE COMPLETE ALL QUESTIONS. IF AN ANSWER DOES NOT APPLY TO YOUR UTILITY. PLEASE MARK "NA" IN THE SPACE PROVIDED
 1. WATER SUPPLY AND TREATMENT FACTS
 For each source, please indicate the percentage of water supplied by each source and the method used to treat the water.
                                          Flowing       Lakes. Ponds  Wells         Purchased
                                          Stream       Reservoirs                  Water (Specify treatments provided
                                                                                     by you. not supplier)
 Percent of total supply                         	 %     	 %     	 %     	 %
 Number of sources

 Number of treatment plants

 Treatment plant capacities  mgd
 METHOD OF TREATMENT
 (Specify typical dosage where appropriate, otherwise place check mark)
Chlorine

Chlbrammes

Chlorine Dioxide

Ozone

Other Oxidant (Specify.
Powdered Activated Carbon

Granular Activated Carbon

Filtration (direct, slow sand, conventional or other)

Aeration

Softening (ion exchange, lime/soda ash)

No treatment
PPM
PPM
PPM
PPM
PPM
PPM
PPM
PPM
PPM
PPM
PPM
PPM
PPM
PPM
PPM
PPM
PPM
PPM
PPM
PPM
PPM
PPM
PPM
PPM
Please list suppliers of purchased water below:

-------
2. TRIHALOMETHANES (THMs)
A, Please provide information on THM levels in your distribution system (of the past 3 years. UKJ should report the data based on the method you
use to report to your regulatory agency. The data should be ad|usted to be reported in parts per BILLION (PPB).

                     1st Quarter                    2nd Quarter                   3rd Quarter                    4th Quarter

             Maximum  Minimum  Mean     Maximum   Minimum  Mean    Maximum  Minimum  Mean    Maximum   Minimum Mean

1984          	    	   	     	    	  	     	     	  	     	    	   	

1935	     	    	  	     	     	  	     	    	   	

19S6          	    	   	     	    	  	     	     	  		    	   	
B. Has your system ever been required to make a Public NoMication that it was in violation ot the THM sandard of 100 PPB7

   1. No

   2 Yes—Speot> number of required ncwcat.ons sines January 193-i	

           •
C For each of the toltewng. please macate 'I you had to change treatment procedures to meet the 10Q PPB THM standard Also indicate the aacfct.or.ai
cao' ace ca: c~ of c s.rv'ec:3-t

CNor amines

CMonne Dcx'oe  ...

Pcwcereo Aci.vaieo Ca-bc-

Cr; r^ Sicrage  .

ASernaie Source
Orone

Grsr-.u'ar V. \a;ec

Cl-er (Soec-iv)
                                                    Treatment
                                                    Change'
.  Yes    No

 . Yes    No

  Yes    No

.. Yes    No

.. Yes    No

 . Yes    No

 . Yes    No

 . Yes    No

  Yes    No

  les    No

  Yes    No

  Yes    No

.  Yes    No
                   Additional
                   CatMtal Expenatures
                   [inS]
Additional Msarty
Ooerating ana Maintenance
Costs (savings) [in S]
D  Have you c^a~ceo vou.- scarce o' s^cov lo^rcr--asec:. ur".-ea:ec. ere) to co.^c>N w.-th tr-e ThM s:araarcT

   1  No

   2. ^s—Arroai ooss '"Or rrvs o^.3^ce ^ sccTce 01 S^CPIY in CO'3"S  S	

-------
 E. Please indicate on a scale of 1 to 4 the nature ol any problems which developed as a result of modifying treatment procedures to comply with
 the 100 PPB THM standard. Circle the number corresponding to your answer.

                                            Major                           Minor          No             No
                                            Problem                         Problem        Problem        Data

 Taste and Odor	  1                2                 3              4            5

 Corrosion	  1                2                 3              4            5

 Color	  12                 345

 Microbiological Quality	  1                2                 3              4            5

 Biolilm Growth in Distnbution System	  1                2                 3              4            5


 F. There are several possible levels of future THM regulation. To the best of your ability, please estimate the capital expenditures and operating and
 maintenance costs to comply with each level.

                                              Capital                         Operation  & Maintenance        Predominant Method of
                                              Expenditures                   Expenses  (Annual)              Anticipated Additional
                                                                                                          Treatment
THM standard lowered to 50 ppb

THM standard lowered to 25 ppb

THM standard lowered to 5 DOb
On whal basis are these projected expenditures and costs based9
    1. Detailed engineering estimate
    2. Preliminary feasibility study
    3. Educated guess
    4. Wild guess
In your opinion, should the THM standard be reduced to less than 100 ppb? Why or why nof .
3 What are the typical values for treated water in your system of the following

Tnhalomethane Formation Potential (THMFP)   		PPB

Bromide		PPB

Total Organic Carbon (TOC)		PPM

Color   		Units

Total Organic Halogen (TOX)    		PP3

Dihaloacetomtnles (DHAN)		. PPB

Other Disinfection Byproducts  			PPB
   (tncholoroacelc acid.
    haloketones.  etc.)


                                                              10

-------
4 TURBIDITY

What have been filtration plant performances for 1986 (in IMTUs)7 Report for your four largest treatment plants if more than four

                                                                    1986 Monthly Averages

                            Jan.     Feb.      Mar.      Apr.     May    June      July     Aug.     Sept      Oct     Nov.     Dec.

Treatment Plant #1

Raw Water Turbidity				   	   	

Finished  Water Turbidity      	   	   	  	   	  	   		.	  „	   		
Treatment Plan) #2

Raw Water Turbidity

Finished Water Turbidity

Treatment Plant #3

Raw Water Turbidity

Finished Water Turbidity

Treatment Plart *4

Raw Water Turbidity

Finished Water Turbidity
The following questions provide background information on your utility.

Ownership
   1.  Investor
   2.  Government
Retail population served

Wholesale population served
Total treated water sold during most recent 12 month period . .	,	million gallons

Average treated water cost in dollars per 1000 gallons   . $	,	per  1000 gallons
Please identify those areas you believe will require additional research to help solve the problems identified above (Check as aporopnate)

                                              Yes            Mo

1 THM treatment                             	         	

2 Analytics.' Methods                         	         	

3. Filtration research                          __	         	

4 Other (Speafy	,	,	)	         	

This questionnaire  is confidential, but if we have questions, may we contact you lor further information'' If so please complete the following-

Name of utility		.	,	,	
Name and title of person completing this questionnaire ,

Telephone number (	)	
THANK YOU VERY MUCH FOR HELPING THE AMERICAN WATER WORKS ASSOCIATION RESEARCH FOUNDATION
RETURN TO AWWA RESEARCH FOUNDATION. C/0 DECISION RESEARCH SUITE 313 341 W  BROADWAY SAN  DIEGO. CA 92101-3882



                                                                11

-------
                     ATTACHMENT B
      TRIHALOMETHANE (THM) SURVEY FOR SMALLER UTILITIES--
      	STATE DRINKING WATER PROGRAMS	
 1.   Name  of  State:
 2.   Total  number of water utilities in state:
 3.  Number  of water utilities serving less than  10.000 people:
 4.  Do you require utilities serving less than 10.000 people
    to monitor for THMs?

    No 	  (Skip to Question 9)

    Yes 	  (Answer Questions 5 to 8)

 5.  Indicate which size utility must monitor for THMs and
    which ones must comply with a 100 ppb THM standard:

   Size of Utility—                Must        Must
 Size of Population Served         Monitor      Comply
5
3
1


,001 -
,301 -
,001 -
501 -
25 -
10,000
5,000
3.300
1,000
500
6.  For the utilities serving less than 10.000 people, do you
    require the same frequency (quarterly) and number of
    samples (minimum of four per source) as the EPA regulation?

    Yes 	           No
7.  Explain any differences with EPA monitoring requirements:
                               12

-------
                              -2-
     Is the THM monitoring data for utilities serving less than
     10.000 people available from your office or should we
     contact individual utilities?
     Available from your office

     Contact utilities
     Has your state agency on any other agency/organization in
     your state conducted a THM survey of utilities serving
     less than 10.000 people?   Yes	    No _____

     If so, please enclose a copy of the data or study report.
     or send it at a later time.
     Thank you for your time and cooperation.  So that we may
contact you later in case we have any questions,  please fill
out the following information.
Name:

Title:

Agency:
Mailing Address:
Telephone
COMMENTS:
Please return to:
Michael J. McGuire
Director of Water Quality
Metropolitan Water District of
  Southern California
Post Office Box 54153
Los Angeles. CA  90054
                              13

-------
                               BIBLIOGRAPHY

Brass, H.J.,  et al.   The National  Organic  Monitoring Survey:  Samplings and
    Analyses  for Purgeable Organic Compounds.   In:  Drinking Water Quality
    Enhancement Through Source Protection, R.B.  Pojasek  (Ed.), Ann Arbor
    Science,  Michigan,  1977.

Mead,ow, R.G.  American Water Works  Association  Research Foundation National
    Trihalomethane Survey Report.   Prepared  for AWWA Research Foundation,
    April 1987.

National Interim Primary Drinking  Water Regulations:  Control of Trihalo-
    methanes  in Drinking Water. Final  Rule, Federal Register, Vol.   44,
    No.  231, pp. 68624—68707, November 29, 1979.

Symons, J.M., et aJ.  National Organics Reconnaissance Survey for Halo-
    genated Organics.  Journal American Water  Works Association. 67,
    634-647,  November 1975.

Temple, Barker and Sloane, Inc., and Malcolm Pirnie, Inc.   Economic  Impact
    Analysis:  Implementation Guidance  for the Drinking  Water Trihalo-
    methane Regulation—Revised Draft.   Submitted  to Office of Drinking
    Water, U.S.  Environmental Protection  Agency,  September 21,  1981.

U.S. Environmental Protection Agency, Office of Water Supply, Technical
    Support Division.  The National  Organic  Monitoring Survey  (unpublished,
    no date).
                                     14

-------
                     GAC FOR REMOVING TRIHALOMETHANES

                       by:  Benjamin W.  Lykins,  Jr.
                            Robert M. Clark
                            U.S. Environmental  Protection  Agency
                            Drinking Water Research  Division
                            Water Engineering Research Laboratory
                            Cincinnati,  OH  45268


    Disinfection by-products are among those compounds being considered for
regulation under the Safe Drinking Water Act Amendments of 1986.  The most
significant disinfection by-products for those utilities that chlorinate
are total trihalomethanes (TTHMs).  Pressure is growing to reconsider the
existing TTHM Standard of 100 yg/1, and to lower it  to some as yet unspec-
ified level,  Trihalomethane levels as low as 10 ug/1  to 50 yg/1 may be
considered.  Utilities may be forced to consider disinfectants other than
chlorine, and to consider treatment modifications that might include new
options ranging from improved conventional treatment to granular activated
carbon (GAC) adsorption.

    Some water utilities may be able to meet a TTHM  level  of 0.10 mg/1 (100
ug/1) by using properly operated conventional treatment.  If, however, the
standard is reduced substantially, adding GAC to conventional treatment may
be the only acceptable treatment option.  The length of time that GAC can
remove THMs to meet a 10 yg/1, 25 ug/1, 50 ug/1, or  100 yg/1 standard will
determine its efficacy as a viable treatment option.

    EPA's Drinking Water Research Division has collected extensive treat-
ment data for removal of organics including TTHM and Total Organic Carbon
(TOC) at several water utilities under actual operating conditions.  In
these studies GAC was used at sites  including Cincinnati,  Ohio; Jefferson
Parish, Louisiana; Manchester, New Hampshire; and Evansville, Indiana to
determine its ability for removing those organic compounds present after
conventional treatment.

    In this paper, data from these studies will be analyzed in order to
assess the potential of conventional treatment and a combination of conven-
tional treatment and GAC for removal of TTHMs and TOC.

                     EFFECT OF CONVENTIONAL TREATMENT

    By removing humic substances through proper conventional treatment
(coagulation/settling/filtration), chlorination by-products can be reduced.
Also, proper pre-treatment appears to increase the effectiveness of acti-
                                     15

-------
vated carbon adsorption,  as noted by Randtke and  Jepsen(l),  Lee  and
co-workers(2), and Weber  and Jodellah(3).   These  investigators especially
noted the benefit of alum coagulation in enhancing carbon adsorption.

    Various conventional  treatment methods used at the research  sites  to
remove or reduce the mix  of compounds present in  the source  water were
evaluated.  The type of treatment (conventional and GAC)  used at these
utilities is presented below.

CINCINNATI, OHIO

    Primary source water  for the Cincinnati Water Works is the Ohio  River.
To aid settling, 17 mg/1  ©f alum was added to the raw water.  Prior  to
flocculation and clarification, 17 mg/1  lime and  ferric sulfate  (8.6 mg/1
for high turbidity and 3.4 mg/1 for low turbidity) and chlorine  (plant
effluent concentration of 1.8 mg/1 free chlorine) were added. Post-
filtration adsorption was evaluated by deep-bed GAC contactors with  an
ultimate EBCT of 15.2 minutes.

JEFFERSON PARISH, LOUISIANA

    The Mississippi River provides source water to the Jefferson Parish
treatment plant.  Potassium permanganate (0.5 to  1.0 mg/1) was added for
taste and odor control.  A cationic polyelectrolyte (diallyldimethyl diam-
monium chloride; 0.5 to 8.0 mg/1) was added as the primary coagulant with
lime (7 to 10 mg/1) fed for pH adjustment to 8,0  to 8.3.   Chlorine and
ammonia (3:1 ratio) were  added for chloramine disinfection (1.4  to 1.7 mg/1
residual after filtration).  A sand filter was converted to a post-filter
GAC adsorber with 18.8 minutes EBCT.

MANCHESTER, MEW HAMPSHIRE

    The principal water source for the Manchester Mater Works is Lake
Massabesic.  Alum and sodium aluminate were added for coagulation, pH
adjustment, and alkalinity control at dosage levels averaging about 12 mg/1
and 8 mg/1, respectively^  Chlorine was added prior to sand  filtration at
an average dose of 1 mg/1.  At the clearwell, chlorine was again added in
the range of 2 mg/1 to 3 mg/1  to produce an average distribution free
chlorine  residual of 005 mg/1.  A GAC filter normally used  for  taste and
odor control was used for post-filtration adsorption with 23 minutes EBCT.

EVANSVILLE. INDIANA
        EvansviHe Water Works uses Ohio River water as their source.
Chlorine and alum were added before primary settling with average con-
centrations of 6 mg/1 and 28 mg/1, respectively.  A free chlorine residual
of 1.5 mg/1 to 2*0 mg/1 was maintained after sand filtration.  Approxi-
mately 12 mg/1 of lime was added after primary settling for pH control  to
8.0.  A pilot plant operating parallel with the full-scale plant used
chlorine dioxide for disinfection.  Average alum and polymer dosages of 12
mg/1 and 0.8 mg/1, respectively were added to the raw water.  An average
                                     16

-------
lime dose of about 6 mg/1 was used for pH control to 8.0.  Post pilot plant
GAC contactors had an EBCT of 9.6 minutes.

    Table 1 illustrates the removal of TOC through conventional treatment.

              TABLE 1.  AVERAGE TOTAL ORGANIC CARBON REMOVAL
                       DURING CONVENTIONAL TREATMENT
Water Utility
Cincinnati, OH
Jefferson Parish, LA
Manchester, NH
Evansville. IN
Raw Water
mg/1
3.4
4.0
4.5
3.0
Sand Filter Eff.
mg/1
2.0
2.9
2.4
1.9
Percent
Removal
41
28
47
37
     Table  2  shows  removal  of terminal  trihalomethanes through various steps
 in  the  treatment process*,   In this  case,  terminal trihalomethanes are used
 because they represent the formation  potential of TTHMs  in the distribu-
 tion system  itself.   The time to  the  most distant customer In the distribu-
 tion system  is  represented by the terminal  day.
              TABLE 2.   AVERAGE TERMINAL  TRIHALOMETHANE REMOVAL
                        DURING CONVENTIONAL  TREATMENT
Water Utility
Cincinnati , OH
Jefferson Parish, LA
Manchester, NH
Evansville, IN
Terminal
Day
3
5
3
3
Raw Water
mg/1
146
281
151
140
Sand Filter E
mg/1
89
175
70
82
ffo Percent
Removal
39
38
54
41
 As can be seen from Tables 1 and 2, the utilities examii
 variable performance in average percent removal  of both TOC and terminal
 THM.  This variability may be due to source water quality.   For example,
 with a river water source, Cincinnati, Jefferson Parish, and Evansville had
 a lower removal efficiency for terminal THM than did Manchester with a lake
 source.
                                      17

-------
                         GAC TREATMENT  PERFORMANCE

    GAC performance for removing both TOC and terminal  THMs  also varied for
the different utilities evaluated.   Examples  of terminal  THM removal  by GAC
for these utilities are shown in Figures 1 through  4.

    These data show the performance of  GAC over various days of operation
and bed volume through the adsorbers.  Normalization of the  data using per-
cent removal  shows that the GAC adsorbers used at Cincinnati produced the
overall highest removal rate for terminal THM (Figure 5).  Conversely,
Evansville had the lowest percent removal.  This may be due, in part, to
the use of a coal-based carbon in Cincinnati  and a lignite carbon in
Evansville.

    Removal of TOC by GAC can give an indication of trihalomethane for-
mation potential (THMFP) removal.  In many cases, removal of TOC also means
removal of THMFP (Figure 6).

    Since terminal THM values can simulate concentrations in the distribu-
tion system, one can estimate the length of GAC operation for meeting THM
goals.  Table 3 gives an indication of  how long GAC can remove various con-
centrations of THMs.

       TABLE 3.  LENGTH OF GAC OPERATION BEFORE EXCEEDING THM GOALS
Location
10 yg/1
Inf,
Day yg/1
25 yg/1
Inf,
Day yg/1
50 yg/1
Inf,
Day yg/1
100 yg/1
Inf,
Day yg/1
Evansville, IN            -             6    96     56    53
(3-day term, 9.6
min. EBCT)

Cincinnati, OH            50    75    155    45    208    70    280   150
(3-day term, 15.2
min. EBCT)

Jefferson Parish, LA      -      -     20    80     63   170    103   220
(5-day term, 18.8
min. EBCT)

Manchester, NH             2    73     16    70     98    65
(3-day term, 23
min. EBCT)


    As can be seen from Table 39 establishing a trihalomethane standard of
10 yg/1 will probably negate the use of GAC.  Using GAC to meet a 25 yg/1
standard also may not be feasible.  However, at the 50 yg/1  trihalomethane
concentration, GAC may be more attractive.
                                     18

-------
     300 H
     200
   01
   3

   5
   X
  2
  £ ioo
   Q
   I
   n
                                                    RAW WATER

                                             •D-EJ--D FILTER EFF.

                                                 -•^ GAC EFF.
               —I    '	1	1	1	1	1	1	j	1	1	1	j	1	]	1

                10      20      30     40      50      60     70     80     RUNOAYS
                       2,894            5,788           8,682           11,576   BED VOLUME
    Figure 1.  Terminal  THM removal  by conventional treatment and  GAC
                adsorption - Evansville,  IN.
      300 H
   5
   cc
   ui
   t-

   <
   Q
   I
   n
   ;  200-
100-
        0-
                •CJ-&-E) RAW WATER

                      SA.NO FILTER EFF.

                     I. OAC EFF.
              A,
'1 I I-'-' '•
0

' ' ' ' 1 ' ' ' ' ' ' '
100
8,590
' ' 1 '
200
17,180
i — i — r — i — i — i — i — i — i — i — i — IT— i
300
25.770
                                                                          RUNOAYS
                                                                          BED VOLUMES
Figure  2.  Terminal THM after  conventional  treatment and GAC  adsorption
            Cincinnati,  OH.
                                           19

-------
  600 I
5
X
2
cc
LJ
                  O-B— EJ RAW WATER
                       SAND FILTER EFF.
                      , OAC EFF.
1
0

-r -j
20
1,560
1 I
40
3,120
i |
60
4,680
I 	 1 	
80
6,240
> — r^
100
7,800
	 1 	
120
9,360
1 1 	
140
10,920
T 	 1
160
12,480
1 	 1 	
180
14,040
-> — r
RUNOAYS
BED VOLUMES
 Figure 3.  Terminal  THM after conventional treatment and GAC adsorption -
          Jefferson Parish, LA.
Q
I
  200
  190
  180
  170
  160
  150
  140
  13(
  120
  110
  100
   90
   80
   70
   60
   50
   40
   30
   20
   10
    0
                         —B
                                            RAW WATER
                                            SANO FILTER EFF.
                                       •» » » OAC EFF.
\/      T*-*™A  V    *
/vv^A/^
 i
10
— r
 20
 1.612
 r
30
— i —
 40
 3,224
 i
50
" — i
 60
 4,836
 i
70
' — i
 00
 6,448
90
' — i
 100
 8.090
                110
                                                    — i
                                                     120
                                                     9.672
                                                          r
                                                         RUHOAYS
                                                         BED VOLUMES
Figure 4.  Terminal THM after conventional treatment and GAC adsorption -
         Manchester, NH.
                                20

-------
    100-1
                                               *H— — 1 — 1 — 1 — 1 — 1 — 1 — 1 — 1 — 1
300
r- 1 i i i i i'|
400
                                        RUNDAY5
           Figure  5.  Terminal  THM percent  removal  for GAC  effluent.
   100-
T   90:
M
F
P
    80-

    70-
    60-
P
E
R
C

N   40
T
   50-
R

M   20
0
v   10
A
L    0-J
       "r
       0
•Jf—4>—*• CINCINNATI

G-D-B EVANSVILLE

4*-«*-W JEFFERSON PARISH

4-~Q.-Q MANCHESTER
                     - EQUAL PERCENT REMOVAL
              10
     —T~

     20
—r~
 30
   40     50     60     70

TOC  PERCENT  REMOVAL
80
90
100
        Figure 6.   THMFP  versus  TOC percent  removal  for GAC  effluent.
                                        21

-------
                                REFERENCES

1.   Randtke,  S.J.  and Jepsen,  C.P.   Chemical  pretreatment  for  activated
    carbon adsorption.  JAWWA.   73:8,  411-419,  August  1981.

2.   Lee, M.C.,  Snoeylnk, V.L.,  and  Crittenden,  J.C,  Activated carbon
    adsorption  of  humic substances.  JAWWA.   73:8,  August  1981.

3.   Weber, W.J. and Jodellah,  A.M.   Removing  humic  substances  by chemical
    treatment and  adsorption.   JAWWA.   77:4,  132-137,  April  1985.
                                    22

-------
               CONTROL OF TRIHALOMETHANES  USING ALTERNATIVE
                        OXIDANTS AND DISINFECTANTS

                       by:  Philip C.  Singer
                            Department of  Environmental  Sciences and
                            Engineering
                            School of Public Health
                            University of  North Carolina
                            Chapel Hill, NC  27514


                               INTRODUCTION

    In order to comply with the maximum contaminant  level  (MCL)  for total
trihalomethanes (TTHM), many utilities have modified their preoxidation
and disinfection practices by switching to alternative oxidants  and disin-
fectants in place of free chlorine.  Examples of  such modifications include
the use of chlorine dioxide, ozone, or permanganate  as preoxidants  and disin-
fectants with free chlorine used as the final disinfectant,  and  the use  of
free chlorine as a preoxidant and disinfectant with  combined chlorine used
as the final disinfectant.  Researchers at the University of North  Carolina
have recently completed an EPA-sponsored research project, the objectives
of which were to determine the impact of such modifications on overall
treatment plant operations, finished water quality,  and treatment  costs.
Specific attention was directed at the impact of these changes on  iron and
manganese removal, color  reduction, microbiological  quality, and filter  per-
formance, in addition to  the control of THMs and other organic halides
(TOX).

    The study was performed at eight utilities, in five states (Florida,
Indiana, Virginia, and North and  South Carolina), all serving between
10,000 and 75,000 consumers.  One of the utilities uses ground water as  a
source of water supply;  the others use  river or lake water.  Two of the
utilities attempted to solve their THM problems using chloramination, three
switched to chlorine  dioxide as a preoxidant, two are using permanganate
as a preoxidant, and  one  is using ozone for  pretreatment.

    Members of the research team  visited each of the utilities on approxi-
mately a quarterly basis  over a two-year period in order  to review plant
operations and performance.  In most  cases,  visits were made before and
after the treatment modifications were  implemented so that a before-and-
after evaluation could be made  using  the  same criteria.   In addition  to
reviewing plant monitoring  records,  samples  were also collected from
various  locations in  the  treatment  plant  and distribution system and
returned to the University  of North  Carolina for measurement of total
                                      23

-------
organic carbon (TOC), TTHM, TOX, and THM and TOX Formation Potential.
These samples were taken to determine the extent of THM and TOX formation
through various processes in order to evaluate the impact of the treatment
modifications, and to determine the extent of TOC and THM Formation
Potential removal by the various processes.  In the case of chlorine
dioxide, measurements of residual chlorine dioxide, chlorite, and chlorine
were made at various locations in the water plants and distribution systems
in order to assess the stability of chlorine dioxide, and to determine the
distribution of residual oxidant species.

    The results of this study were reported at the 1986 American Water
Works Association Annual Conference in Denver, Colorado, and have been
published in the Proceedings of that conference.  The following is an
abbreviated summary of those results.

                                  RESULTS

    For the two utilities using free chlorine as a primary disinfectant and
chloramines as a secondary disinfectant, both were able to lower TTHM for-
mation to levels well below the 100 ug/1 MCL.  One of these utilities has
experienced no adverse impacts on operations or on finished water quality;
according to distribution system monitoring records, the microbiological
safety of the water has been maintained.  The second utility has observed a
significant deterioration in finished water color.  The color of the
finished water has exceeded the standard of 15 color units several days
each month since the practice of chloramination was adopted, causing
numerous customer complaints.

    Of the three utilities using chlorine dioxide as a preoxidant, only
one appears to have consistently reduced THM formation to achieve
compliance with the MCL.  This utility had suffered from high THM levels
due to prechlorination, and when the point of chlorination was moved from
the raw water flash mix basin to post-filtration, manganese problems were
encountered.  By applying chlorine dioxide to the raw water, at dosages of
0.7 to 1.3 mg/1, and free chlorine to the settled water, the utility has
achieved compliance with the MCL for TTHMs, has controlled the manganese
problem, and has experienced no adverse impacts on plant operations or
finished water quality.  The chlorite concentration in the finished water
is well below the EPA-recommended level of 1.0 mg/1 for residual chlorine
dioxide species.  It should be noted that the quarterly average TTHM con-
centrations have been reduced from a range of 200 to 300 ug/1 to 50 to 120
ug/1.  The running annual averages are in compliance with the THM regula-
tion, but not by a wide margin of safety.

    The second utility using chlorine dioxide for preoxidation and free
chlorine for post-disinfection has experienced serious manganese problems
and has had difficulty limiting THM formation sufficiently to achieve
compliance with the MCL.  Additionally, due to the high TOC concentration
of the raw water (12 to 15 mg/1), chlorine dioxide dosages have ranged from
1.5 to 6.0 mg/1, resulting in the presence of up to 3.0 mg/1 of chlorite in
                                     24

-------
the finished water.  This exceeds EPA's recommended limit of 1.0 mg/1  for
the sum of chlorite, chlorate, and chloride dioxide residuals.

    The third utility using chlorine dioxide has a flow sheet involving the
application of chlorine dioxide at the raw water intake (2 to 2.5 mg/1),
chlorine and chlorine dioxide on top of and beneath the filters, and air
stripping of the finished water prior to its passage into the distribution
system.  Due to the high TOC of the raw water (25 to 35 mg/1),  chlorine
dioxide doses are  excessive (up to a total of 6 mg/1), resulting in the
presence of high chlorite concentrations (1.2 to 1.4 mg/1) in the finished
water.  Furthermore, in view of the high residual TOC in the filtered
water, THM formation is still appreciable after the stripping towers,
causing TTHM concentrations in the distribution system to exceed the 100
Ug/1 MCL.  It should be noted that while the stripping towers do remove a
significant portion of the TTHM produced up to this point in the treatment
train, the non-volatile organic halide species which comprise about 70 per-
cent of the TOX are not removed by stripping.

    The two utilities using permanganate as a raw water preoxidant and
chlorine on top of the filters have lowered THM formation considerably with
no adverse impact  on plant operations, but have not achieved compliance
with the MCL for TTHM.  An alternative strategy needs to be adopted in each
of these cases.

    The utility using ozone had historically generated quarterly average
TTHM concentrations of 400 to 1,000 ug/1 due to the application of 15 to 20
mg/1 of chlorine to a raw water containing 15 to 30 mg/1 of TOC, followed
by precipitative softening at pH values of 9.5 to 10.  By applying 3 mg/1
of 03  to the raw  water ahead of the flash mix basin and 3 mg/1 of 03 ahead
of the filters, and 4.5 to 6 mg/1 of chlorine after filtration, THM for-
mation has been reduced dramatically, to levels of 50 to 170 ug/1.  It
should be noted that when the utility was operating with prechlorination,
chloroform constituted more than 85 percent of the TTHM concentration.
After switching to preozonation, chloroform constitutes only about 30 per-
cent of the TTHM concentration, the remainder being various brominated THM
species, reflecting the presence of bromide in the raw water.  Still, no
THM species are produced until the chlorine is applied post-filtration.  In
any case, while the utility is still not in compliance with the THM regula-
tion, the quality  of the finished water, from the standpoint of TTHM con-
centration, TOX concentration, and color, is far superior to what it was
prior to the modifications.

                       CONCLUSIONS AND IMPLICATIONS

    Alternative pretreatment  oxidants and disinfectants are depleted rela-
tively rapidly and, accordingly, have short residence  times, particularly
in waters having TOC concentrations greater than 5 mg/1.  Such waters have
a  relatively high  oxidant-demand.  The implications  of  this  rapid rate  of
depletion are:
                                     25

-------
    o   Residual  oxidants will  not be  able  to  be  carried  through  sedimen-
        tation basins.   This can  lead  to  problems with  aquatic  growths,
        e.g.,  algae,  in the sedimentation basins, and to  the  release  of
        reduced impurities, e.g.,  manganese, from the sludge  retained in
        the bottom of the basins;

    o   Disinfection  effectiveness will be  reduced  as a result  of the
        decrease in "C x t" for disinfection,  i.e.,  concentration of  disin-
        fectant (C) times contact  time (t).  The  reduction  in the C x t
        value  will also impact  the effectiveness  of the oxidant for oxi-
        dizing taste  and odor compounds,  organic  color, and iron  and
        manganese;

    o   Alternative oxidants can be used  most  successfully  for  controlling
        THM formation in waters with relatively low TOC concentrations.

    Accordingly, if the MCL for TTHMs  is  lowered  to 20  to 50  yg/1, most utili-
ties will not  be able to comply with the  MCL using  only alternative
oxidants/disinfectants and conventional treatment without sacrificing
overall finished water quality. The establishment  of such  an MCL will
require, in most cases, elimination of free chlorine as a secondary disin-
fectant for protection of the distribution  system.   Combined  chlorine
(chloramines)  will have to be used for this purpose.

    Except for waters with a very  low THM formation potential,  an alter-
native oxidant/disinfectant will  have to  be used  for primary  disinfection
and oxidative  pretreatment.  The  use of prechlorination will  produce
excessive concentrations of THMs  too rapidly.  Since these  alternative
oxidants/disinfectants are very reactive, they will  be  depleted quickly,
providing short contact times for  iron and  manganese oxidation, decoloriza-
tion, taste and odor destruction,  and, most importantly,  disinfection.
Accordingly, finished water quality will  suffer.

    Alternatively, preoxidants  and disinfectants  can be employed, perhaps
at several pretreatment locations  or in conjunction with  activated carbon,
to reduce the  TOC concentration,  thereby  lowering the  oxidant-demand  of  the
water so that  effective post-disinfection can  be  achieved.   This  will
obviously increase the cost of  water treatment significantly.  However,  the
use of only alternative oxidants  and disinfectants  with conventional  treat-
ment in order  to meet an MCL for  TTHMs of 20  to  50  yg/1 is  likely to  cause
the deterioration of overall finished water quality.

                             ACKNOWLEDGEMENTS

    The author acknowledges V.  Brooks, R. Brown,  D. Chang,  W. O'Neil, D.
Reckhow, C. Salmons,  D. Simmons,  K. Werdehoff, and  J.  Wiseman for their
assistance in  carrying out this research  project.  The  cooperation of the
utilities participating in this study is  also  acknowledged.

    This work  was supported by the Drinking Water Research  Division  of the
U.S. Environmental Protection Agency under  Cooperative Agreement  CR  811108;
the assistance of the project officer, Ben  Lykins,  is  also  appreciated.
                                      26

-------
                 THMFP REDUCTION BY LOW PRESSURE MEMBRANES

                       by:  J. S. Taylor
                            Civil Engineering and Environmental  Science
                            University of Central Florida
                            Orlando, FL  32816


    Bench-scale (1,000 gpd) and pilot-scale (25,000 gpd)  investigations of
membrane processes were conducted at two water utilities  near West Palm,
Florida that used ground water supplies — the Village of Golf (VOG)  and
Acme Improvement District  (AID).  Bench-scale investigations were also con-
ducted at Lee County (Florida) Utilities - Olga plant that used a surface
supply.  Initially, one reverse osmosis (RO) and six ultrafiltration  (UF)
low-pressure membranes were purchased and tested for product (permeate)
water quality on a bench-scale level.  The UF membranes are designed  for
operation up to 100 psi, whereas the RO membrane is intended to operate at
200 and 250 psi.  Test results of bench scale studies from all three  sites
demonstrated that only two membranes — the Filmtec UF (N-50) and the RO
membrane (BW 3030) could produce a water from these highly organic sources
that would meet the THM MCL and maintain a Cl residual.  The N-50 is
referred to as a nanofilter by Filmtec, because it exhibits lower operating
pressures than RO and better  rejection than UF.

    The normal molecular weights rejected by each membrane were supplied by
each manufacturer and ranged  from 40,000 to 100.  The results from membrane
testing are shown in Tables 1,2, and 3 for raw waters from the Village of
Golf, ACME Improvement District, and the Olga plant.  Bench-scale testing
indicated that a molecular weight rejection of 2,000 would typically pass
50 percent of the raw (DOC),  but only 20 percent of the color.  The
resulting product water THMFP was generally 800 ug/1 for the surface water
source and 400 ug/1 for either ground water source.  The ultrafilter with a
molecular weight rejection of 400, passed less than 10 percent of the DOC
and three percent of the color at any site, and  it typically produced pro-
duct water with a THMFP of less than 50 ug/1 at the ground water sites and
approximately 100 ug/1 at  the surface water sites.  The product water was
essentially colorless, and the DOC was 2 mg/1 or less at all three sites.
The color, DOC, and THMFP  of  the product water from the RO membrane were
approximately equal to the color, DOC, and THMFP of the product water from
the best UF membrane.  However, the RO membrane operated at 200 psi and
rejected species with a molecular weight of 100 or greater as opposed to
the UF membrane, which rejected species with a molecular weight of 400 or
greater at a pressure of 100  psi.
                                     27

-------
                                 TABLE 1.  VOG OPTIMUM MEMBRANE TESTING - THMFP FORMATION CURVE
                                                      (all values in ug/1)
PO
co
Site
VOG
VOG
VOG
VOG
VOG
VOG
Initial Cl2
Dose
Membrane (mg/1 Cl2)
BW30
BW30
FT-50
FT-50
U90-G10
U90-G10






TABLE 2. AID OPTIMUM
Sample and
Membrane
Acme U90-G10
Raw
% Removal
Acme BW30
Raw
% Removal
Acme FT-50
Raw
% Removal
PH
8.1
7.7
7.9
8.2
7.3
7.3
DOC
(mg/1)
7.42
12.46
40
0.62
13.93
96
1.41
14.64
90
5
10
5
10
5
10
MEMBRANE
Color
(cpu)
14
35
60
1
32
97
1
35
97
Time (hours)
1 3
14 13
11 11
4 16
11 16
48 61
85 133
TESTING DATA AND
Alkalinity
(mg/1 CaC03)
300
342
12
14
311
96
126
325
61
24 48
11 11
11 11
11 16
16 18
99 84
221 225
PRODUCT WATER
Calcium-
Hardness
(mg/1 CaC03)
288
328
12
10
309
97
93
304
69
72 96
28 21
32
35 31
39
91 90
326 451
PARAMETER REMOVALS
Total
Hardness
(mg/1 CaCOa)
302
348
13
14
322
96
97
319
70
192
18
28
31
23
—
430

Turbidity
(NTU)
0.13
0.26
50
0.13
0.25
48
0.09
0.25
64

-------
            TABLE 3.   OLGA MEMBRANE SELECTION STUDY RAW PRODUCT  PARAMETERS AND REMOVAL EFFICIENCIES
MW DOC Color
Membrane Cutoff (mg/1 as C) (cpu)
and Sample (mw) * *
DESAL U90-G10 2,000
Raw
% Removal
DESAL U90-G50 20,000
Raw
% Removal
OSMO PT-2 (411PS) 20,000
Raw
% Removal
OSMO PT- (411TPS) 40,000
Raw
% Removal
FILMTEC BW30 100
Raw
% Removal
FILMTEC N-50 400
Raw
% Removal
**FILMTEC UFP 4040 10,000
Raw
% Removal
7.08
15.93
55.6
10.70
23.71
54.90
17.50
22.04
21
17.88
18.12
1.3
1.19
21.84
94.6
1.24
21.80
94.3
18.96
21.59
12.2
12.5
70.0
82.1
14.0
80.0
82.5
67.0
75.0
11.0
51.0
68.0
25.0
2.5
67.0
96\2
3.0
70.0
95.7
70.0
80.0
12.5
Alk.
(mg/1
CaC03)
102
110
7.3
132
137
3.6
135
140
3.6
130
131
3.1
12
143
92.3
31
127
75.6
127
130
2.3
TH
(mg/1
CaCOa)
190
220
13.6
204
214
4.6
228
234
2.5
240
246
2.4
16
252
93.6
46
202
77.2
194
206
5.8
CaH
(mg/1
CaC03)
180
204
11.8
200
200
0.0
200
220
9.1
220
226
2.6
12
208
94.3
36
180
80.0
186
190
2.1
TDS
(mg/1)
402
484
16.9
358
384
6.7
372
424
12.3
436
470
7.2
56
605
90.7
86
334
83.5
332
336
1.2
Temp
CO
28
28
—
23
23
—
22
22
—
22
22
—
19
19
—
22
22
—
21
21
"*"•
PH
7.5
7.9
—
8.1
8.0
—
7.8
8.2
—
7.9
7.5
--
7.8
7.5
--
7.6
7.5
—
8.1
7.5
_ «
Turb.
(NTU)
0.4
5.0
—
2.3
0.2
2.3
1.0
0.3
—
0.2
0.7
—
0.08
0.8
—
0.2
0.7
—
.3
.6
— —
ci-
(mg/1)
87
90
3
60
60
0
73
75
2
117
117
0
16
175
91
30
63
52
71
71
6
Na+
(mg/1)
41
43
2
29
29
3
38
38
0
56
56
0
9
96
91
17
30
57
35
36
3
  TJCFData - All  other parameters  Lee  County Lab Data except UFP 4040'
** All  UFP 4040 Data 1s UCF Data.

-------
    A summary of product water quality corresponding  to molecular weight
cutoff (MWC) is shown in Table 4 for both ground water sources and in
Figure 1 for the surface water source.  (The RO membrane rejected a much
higher inorganic fraction than did the UF membrane.   The RO membrane
rejected more than 90 percent of the total dissolved  solids (TDS), total
hardness (TH), Cl, and Na at all sites, whereas the UF membrane inorganic
rejection varied from 50 percent to 70 percent for the various parameters.)
Since the N-.50 was the membrane that operated at the  lowest pressure and
still produced a water that met the THM MCL, it was  selected for extended
operation.

    The N-50 ultrafilter was installed at the Olga plant in a bench-scale
unit capable of producing 1,000 gpd, and it operated  for 740 hours over a
45-day period.  An operational percentage of recovery and feed pressure
matrix was developed to determine the extended study  operating conditions.
The operating pressure and percent recovery affected  the THM removal as
shown in Figure 2.  The isopleth in the upper left hand portion of Figure 2
shows the THM MCL.  The DOC of the product water was  observed to increase
with increasing pressure and decreasing percent recovery (Figure 3).  Over
matrix conditions of 60 to 120 psi feed pressure and  60 percent to 90 per-
cent recovery, product water quality improved at high pressure (105+ psi)
and lower recovery (60 percent).  The matrix results  indicated the THM MCL
could be met if the operational conditions were 105  psi with 60 percent
recovery.  At these conditions, product water quality and the percentage of
rejections were 1.6 mg/1 DOC as C (92 percent), 3 cpu color (93 percent),
172 mg/1 TDS (64 percent), 78 mg/1 as TH CaCOa (65 percent), 60 mg/1 Cl (40
percent), and 68 mg/1 alkalinity as CaC03 (59 percent) with pH 7.8.

    During the extended operation, pressure was varied from 60 to 100 psi
with recoveries of 60 percent to 90 percent.  The THM MCL was met for 105
psi and 60 percent recovery, and for 75 psi and 60 percent recovery imme-
diately after the membrane was chemically cleaned.  The flux decreased with
time during the extended study from 18 to 14 gpd/ft2  over 150 hours of
operation.  The effect of the flux decline with hours of operation at the
Olga plant is shown in Figures 4 and 5.  After cleaning with the pressure
at 95 psi, the flux declined from 22 to 16 gpd/ft2,  but the product water
quality remained constant.  Flux was independent of recovery during the
extended study.  The water quality and percentage of rejection for the con-
ditions meeting the THM MCL were 2.7 mg/1 DOC as C (88 percent), 156 yg/1
TOXFP (80 percent), and 3 cpu color (98 percent).  These conditions are
shown in Figures 6 and 7.  The inorganic water quality was 145 mg/1 TDS (65
percent), 68 mg/1 Cl (32 percent), 85 mg/1 TH as CaC03  (64 percent), and 7
mg/1 alkalinity as CaC03 (58 percent).  The pH was 7.8, and the water was
stable.

    A 25,000-gpd mobile UF pilot plant using the N-50 membrane was built in
a 30 foot trailer.  The UF plant was housed in the 20 by 8 foot rear sec-
tion of the trailer and equipped for antiscalant feed, acid feed, chlorina-
tion, stabilization, prefiltration, and storage as well as UF with variable
recovery (50 percent to 90 percent) and feed pressure (80 to 120 psi).
This plant was installed and operated at  VOG for 365 hours from January 2
                                     30

-------
                  TABLE 4.  PARAMETER REMOVAL AND RETENTION BASED ON MOLECULAR WEIGHT
DOC
Membrane
BW30
N50
V90-G10
UFP-4040
U90-G50
PT-2
PT-4
MW
Passage
< 100
< 400
< 2,000
< 10,000
< 20,000
< 20,000
< 40,000
% Removal
from raw
95.5
90.4
40.4
15.7
16.2
5.1
5.0
% Remaining
in product
water
4.5
9.6
59.6
84.3
83.8
94.9
95.0
Total
% Removal
from raw
95.7
69.6
13.2
0.0
6.6
0.9
0.3
Hardness
% Remaining
1n product
water
4.3
30.4
86.8
100
93.4
99.1
99.7
Color
% Removal
from raw
96.9
97.1
60.0
8.6
37.1
5.7
8.6
% Remaining
1n product
water
3.1
2.9
40.0
91.4
62.9
94.3
91.4
THMs
Value
yg/i
32*
39*
326*
780**
605**
929**
942**
% Removal
from high-
est value
96.6
95.9
65.4
17.2
35.8
1.4
—
  10 mg/1 Cl2 - Initial chlorine dose
**20 mg/1 Cl2 - Initial chlorine dose

-------
            100
          (O

          o  80
          E
          O)
          a:
          s-
          o
             60
o   40

«   20


     0



   100
(0
>




I

O
o
Q
    80

    60


    40


    20


     0
                   96.6   95.8
                                  82.8
                                Average Raw True Color =
                                        72.8 CPU
                                                 29.9
                                 40.3
                                          4
                                         (a)
                                                 log M.W.cutoff
                                              Average Raw DOC =
                                                  20.7 mg/L
                     o
                     CO
                     o
                     CO
                     2
                     CO
                   o  o o
                   LO  i—I i-H
                   I   <£> CD

                      O O
                       CT>
                               oo co
                               LU LU
                               Q Q
                                                          log M.W.cutoff
                               O O CM «*
                               «a- in |  i
                               CD CD I— I—

                               T i"
                               O- O1 o O

                               =>   CO CO
                                  —IOO

                                  CO
                                  UJ
                                  Q
Figure 1.  (a) Percent  removal  of true color and  (b)  percent removal of DOC.
           As a function  of molecular weight size  fractionization using
           membranes with different molecular weight  cutoffs, Olga, Florida.
                                       32

-------
                120
            SL   105
                 90
                 75
                 60
    89
'   83
'  80
                        50
                60

            % Recovery
                  70
Figure 2.  THMFP isopleth for N-50  based on 96 hour formation potential
           of 533 ug/1  at 25°C for  Feb. 28 and March 29 matrices.
                120
            L   105
                 90
                 75
                 60
   93
      94           9J

% DOCReroval


      93
   94
               87
               	I
                  87
                       50
               60
           % Recovery
                  70
    Figure 3.  DOC isopleth for N-50 based on  Feb.  28  and March 29
               matrices for Raw DOC of 18.8 mg/1.
                                 33

-------
CSJ
 O>
 4->
 <0
 0)
  O)
 O-
22


21


20

19


18


17


16

15


14


13


12


11

10
              A  Average Flux
              O  Maximum Flux
                 Minimum Flux
         - D
                                        I    I
          60       70   75   80   85   90  95  100    105  110  115  120

                                Feed Pressure  (psi)
     Figure 4.  Permeate flux as a function of feed pressure,  Olga
                N-50 ultrafiltration study.
                                    34

-------
X
9
IB
    22
    21
    20
    17
    16
    15
                20       40        60       80       100

             Elapsed Hours of Operation Since Cleaning (hr)
120
 Figure 5.   Permeate flux at 95 psi  feed pressure  as  a  function of
            elapsed hours of operation  since  chemical cleaning.
            Olga N-50 ultrafiltration  study.
                                 35

-------
•* I
8,
   105
    90
    75
    60
86
86
• 93
. 84
60


Arg. DOC * 21.4 ng/L 1AC
88 S.
fi 90
80 £ 75
1
« 60
l 1 «
OHMFP - 533 ug/L
. 81 Dose - 5 ng/L Cl?
66
• 69
• 78 73
•59 31
I 1 * *
70 80 90 60 70 80 90
% Recovery % Recovery
(a) (b)
    Figure 6.   (a)  Percent DOC removal  and (b)  percent THM formation
               potential  reduction,  as  a function of feed pressure and
               percent recovery.
               Olga N-50  ultrafiltration study.
~ 105
I
1 9°
to
fi 75
1
60
Raw True Color = 55 cpa , ne
93 ^ lu:>
DC 0)
O-
90 ^90
m
• 96 94 * 75
1
85 fc
. 92 60
• i • i
TOXt'P = 873 ug/L
81 Dose = 5 mg/L
69
• 69
' 83 69
• 68
60 70 80 90 60 70 80
% Recovery % Recovery
(a) (b)



32
90


   Figure 7.   (a) Percent color  removal and  (b) percent TOX  formation
               potential reduction, as a function of  feed pressure  and
               percent  recovery.
               Olga N-50 ultrafiltration study.
                                     36

-------
to March 3, 1985.  Initially, an operational test matrix was developed for
water quality and flux from varying percentages of recovery and pressure.
Product water DOC, color, and THMFP were independent of recovery and
pressure over the test conditions and averaged less than 2 mg/1 as C, 1
cpu, and 50 ug/1.

    The N-50 was tested at each site to determine the effect of operating
pressure and percent recovery on product water quality.  Both color and
THMFP of the product water were less than the MCL.  The variation of THMFP
of the product water at the VOG plant is shown in Figure 8.  Product water
TDS and TH increased with increasing recovery, were independent of
pressure, and varied from 25 percent to 75 percent of the raw water value.
Operating conditions at VOG were set at 90 to 105 psi and 75 percent reco-
very to produce a water with a TH of 150 mg/1 as CaCOs, essentially no
color, and THMFP of 50 ug/1 or less.  During the VOG operation, the product
water quality results  (and the percentage of rejection) were 1.9 mg/1 DOC as
C (88 percent), 3 cpu  color (97 percent), 27 ug/1 THMFP, and 47 ug/1 TOXFP
as Cl.  The raw water, TSD, TH, and alkalinity were reduced to 195 mg/1  (60
percent), 142 mg/1 as  CaC03  62 percent), and 135 mg/1 as CaC03 (60
percent), respectively.  The final pH was 7.5 and the water was stable.
The product water flux declined 32 percent during the VOG operation from 20
to 13.4 gpd/ft2-  The  water temperature was approximately 25°C and essen-
tially did not vary during the operation.

    On March 3, 1985,  the UF pilot plant was moved from VOG to AID and
operated until May 31, 1985, with an elapsed time of operation of 1,020
hours.  A second operational test matrix was developed and showed that
color, DOC, and THMFP  removal were independent of product recovery and feed
pressure over the test conditions (50 percent to 90 percent product reco-
very and 80 to 120 psi).  The permeate color and THMFP for VOG are not
shown for brevity; however, in each of the ground water sites the permeate
water quality was significantly lower than the MCL for recoveries of 50
percent to 90 percent  and feed pressures of 80 to 120 psi.  The variation
of AID THMFP is shown  in Figure 9.  Product water color, DOC, and THMFP
were typically less than 2 cpu, 2 mg/1 as C, and 30 ug/1.  TH and TDS re-
moval were independent of pressure and dependent on product recovery, and
they varied from 25 percent to 75 percent of the raw water value.  Long-
term operating conditions varied from 90 to  103 psi and 67 percent to 33
percent recovery.  The product water quality and percentage of rejection
from the raw water were  1.5 cpu color (97 percent), 2.0 mg/1 DOC as C (86
percent), 50 ug/1 THMFP, and 48 ug/1 TOXFP as Cl.  The product water TDS,
TH, and alkalinity values  (and percentage of rejection) were 282 mg/1 (43
percent), 187 mg/1 as  CaCOs  (40 percent), and 180 mg/1 as CaCOs  (37
percent). The pH averaged 7.5, and the product water was stable.

    The UF pilot plant was designed with the membranes  in four pressure
vessels connected two  each in  series.  The average pressure drop  in  the
first pressure vessel  was 32 psi  (11 psi/membrane) and 41 psi  (13 psi/
membrane) in the second  pressure vessel.  The flux at AID rose slowly from
14 to 15 gpd/ft2 during  the first 150 hours  of operation, and after  a che-
mical cleaning, it rose  to slightly more than 20 gpd/ft2 and remained there
for the duration of the  study.
                                     37

-------
w

O  —>
"•  *
*  «s
i
        150  -
        100  .
 o
 u
 •)
  «»*
 u
 •g
 t
 •n
 •n

 t
 &.


 I
 75

 70


 £5




105

100


 95
                100          200         300

                       •Hit* of 0ptr»t1on
                                                             400
        Figure  8.   VOG membrane pilot  plant operation.
                    Trihalomethane  formation potential  versus
                    time of operation,  product recovery rate,
                    and feed pressure.
                               38

-------
O
a.

c
o
      100-
O cr>
u. =>

o^"

•o
JC.


I
o


>
O
u
O)
  O>
  *J

  «B
        75-
O

CL.
O>
E~  10&1
  to
  CL
T»
V
0>
                             *~^—A_
       95-
                          400      600      600
                              Time of Operation (hr)
                                                     1000
	I
T?00
 Figure 9.  AID membrane pilot plant operation trihalomethane

           formation  potential  versus time of operation,
           product recovery rate, and feed pressure.
                                39

-------
    Cost of construction  of  a  UF plant should be slightly less than an
equivalent RO plant because  of less expensive membranes and a lower (50  to
75 percent operating  pressure).   The construction and O&M cost were esti-
mated at $0.29 and $0.47/1,000 gal, respectively for a 1-MGD UF plant.
These estimates are summarized in Tables 5 and 6.

         TABLE 5.   ESTIMATED CONSTRUCTION COSTS FOR ULTRAFILTRATION PLANT
                           IN APRIL  1985 DOLLARS
Cost Category
Manufactured Equipment
Labor
Electrical and Instrumentation
Housing
Subtotal
Miscellaneous and Contingency*
Total
Cost per 1,000 gals**
* 10 percent of Subtotal Value
**20 year amortization at eight
0.1 MGD
$110,600
23,400
15,400
8,400
$157,800
15,800
$173,600
$0.48
percent
Plant Capacity
1.0 MGD 10,
.0 MGD
$647,300 $4,720,500
102,500
95,000
84,400
$929,200 $6
92,900
$1,022,100 $7
$0.29

504,900
702,300
608,300
,536,600
653,700
,190,300
$0.20

TABLE 6. ESTIMATED OPERATIONS AND MAINTENANCE COSTS FOR
ULTRAFILTRATION PLANT IN APRIL 1985 DOLLARS
Cost Category
Energy Costs - Building
Process
Maintenance Materials
Labor
Total
Cost per 1,000 gpd
0.1 MGD
$800
3,400
13,200
2,000
$19,400
$0.53
Plant Capacity
1.0 MGD 10
$5,500
31,500
131,800 1
2,800
$171,600 $1
$0.47
.0 MGD
$43,900
288,500
,013,600
4,300
,350,300
$0.37
                                    40

-------
             DEVELOPMENT OF RAPID SMALL-SCALE ADSORPTION TESTS

                       by:  David W. Hand
                            Water and Waste Management Programs
                            Michigan Technological University
                            Houghton, MI

                            John C. Crittenden
                            Department of Civil Engineering
                            Michigan Technological University
                            Houghton, MI

                            John K. Berrigan
                            Zimpro  Inc.
                            Rothchild, WI

                            Benjamin W. Lykins
                            U.S. Environmental Protection Agency
                            Drinking Water Research Division
                            Water Engineering Research Laboratory
                            Cincinnati, OH  45268
                               INTRODUCTION

    Design of full-scale adsorption systems typically includes expensive
and time consuming pilot studies to simulate full-scale adsorber perfor-
mance.  Recently, a  rapid method for design of large-scale fixed-bed adsor-
bers from small column studies, known as the rapid small-scale column test
(RSSCT) was developed(l).  One of the advantages of using the RSSCT for
design is that the RSSCT may be conducted in a fraction of the time it
takes to conduct a pilot study; unlike predictive mathematical models,
extensive isotherm or kinetic studies are not required to obtain a full-
scale performance prediction from an RSSCT.  Since only a small volume of
water is required for the test, the water can be transported to a central
lab for evaluation.  Accordingly, the RSSCT would significantly reduce the
time and cost of a full-scale design.

                   DEVELOPMENT OF THE SCALING EQUATIONS

    The development  of the scaling equations for the RSSCT method were
based on the dispersed flow pore surface diffusion model (DFPSDM)(1), which
accounts for many of the mechanisms that occur in fixed-bed adsorbers(l-6).
These mechanisms are axial dispersion, axial advective flow, surface dif-
fusion, pore diffusion, liquid-phase mass transfer resistance, local
equilibrium at the exterior surface of the adsorbent, and competitive
                                     41

-------
equilibrium of the solutes upon the adsorbent surface.   Recent work by a
number of investigators(3,4,6-8,10,ll) has shown that the DFPSDM can pre-
dict fixed-bed removal for single as well  as multicomponent mixtures.

    Using the DFPSDM, these conditions of similarity can easily be applied
to the model equations in dimensionless form such that the RSSCT and the
full-scale process yield identical breakthrough profiles:  1) The same
dimensionless differential equations apply to both processes; 2) the boun-
dary conditions occur at the same dimensionless coordinate values; 3) the
dimensionless coefficients in the dimensionless differential equations are
equal; and 4) no change in mechanism occurs with an increase in process
size.

    There are four dimensionless groups in the DFPSDM which express the
relative importance of competing kinetic mechanisms of each component.
Table 1 displays the dimensionless groups which are present in the DFPSDM.
By setting the four dimensionless groups which represent a small column
equal to those of a large column, relationships between key design
variables were found.  Equating the surface diffusion modulus and the pore
diffusion modulus or the small column to that of a large column, the
following scaling equation for determining the empty bed contact time
(EBCT) of the small column from the large column is developed:
        EBCTsc

        EBCTIc"
RSC
tsc
                                               (1)
This equation can also be used to calculate the time required to conduct
the small column test if that of the large column is known.

    Equating the Stanton and Peclet Numbers of the small column to that of
the large column, the following scaling equation between the small column
and large column superficial velocities can be developed:
                                               (2)
        vsc  =
        VLC     RSC
 Equations  1 and 2 assume that:  1) the physical characteristics of the GAC
 in both  the small and  large columns are identical; 2) the equilibrium capa-
 city  is  assumed to be  the same for both the small and large columns; and 3)
 the surface and pore diffusivities are assumed to be the same.  More
 detailed discussions of the development of the scaling equations are pre-
 sented by  Berrigan(l)  and Crittenden et a7(5).

         VERIFICATION OF THE SCALING EQUATIONS WITH LABORATORY DATA

    The  scaling procedure was first verified in the laboratory by comparing
 performances  of a pilot-scale adsorber to an RSSCT result.  The design and
 operational parameters are shown  in Table 2.  Calgon's F-400 carbon was
                                    42

-------
    TABLE 1.  DIMENSIONLESS GROUPS WHICH CHARACTERIZE THE DFPSDM MODEL
Dimensionless    Mathematical
    GrouP         Expression                      Definition
                 kf iT(l-e)      rate of solute transport by film transfer
     Sti         		
                     Re            rate of solute transport by advection

                     Lv          rate of solute transport by advection
     EdP,i
   Dej         rate of solute transport by dispersion

tDs,i°9s,i      rate of solute transport by surface  diffusion
    R2              rate of solute transport by advection

lDp,iD9p,i      rate of solute transport by pore diffusion
    R2            rate of solute transport by advection
TABLE 2.   COMPARISON OF THE OPERATIONAL  PARAMETERS FOR THE SMALL AND LARGE
                       LABORATORY  COLUMN  EXPERIMENTS
Parameter
R, cm
vs, m/h
EBCT, sec
BVFmax>
Run Time, days
Large Column
0.0513
5.0
60.0
16,000
11.1
Small Column
0.0105
24.4
2.51
16,000
0.466
                                      43

-------
used in the study and 12 x 40 mesh size was used for the pilot plant
whereas 60 x 80 mesh size was chosen for the RSSCT.  A superficial  velocity
of 5 m/h was chosen for the pilot operation and from Equation 2, a loading
velocity of 24.4 m/h was calculated for the small column.  An EBCT of 1
minute was chosen for the pilot-scale adsorber and from Equation 1 an EBCT
of 2.51 seconds was calculated for the small column.  The required maximum
number of bed volumes of feed required for each unit is 16,000 which
translates to about 2,700 liters of water required for the pilot column as
compared to about 26 liters of water required for the RSSCT.  In other
words, the volume of influent required for the RSSCT is only about 1 per-
cent of that required for the pilot-scale adsorber.  In terms of run time,
for this case, the RSSCT can be run in about 4 percent of the time required
for the pilot-scale adsorber, which is a significant time savings.

    The solutes used in this study were chloroform, trichloroethene,
chlorodibromomethane, 1,2-dibromoethane, bromoform, and tetrachloroethene.
The average influent concentrations varied between 20 and 30 uMol/1.
Figure 1 displays the pilot-plant and small-scale results for chloroform.
Figure 2 displays the results for trichloroethene.  Excellent agreement was
obtained between the small and large column results for both chloroform and
trichloroethene.  The slight discrepancies between the effluent profiles
can be due to the differences in the influent concentrations to the adsor-
bers.  Similar results were obtained for the other four components(l,5).

           VERIFICATION OF THE SCALING EQUATIONS FOR FIELD DATA

    This same scaling procedure was field tested by comparing performances
of pilot-plant adsorbers to RSSCTs on contaminated groundwater obtained
from Wausau, Wisconsin.  Table 3 displays the solutes and their average
concentrations monitored during the study.  They are dichloroethene, trich-
lorethene, tetrachloroethene, vinyl chloride, 1,1,1 trichloroethane,
toluene, ethyl benzene, m-xylene, and o,p-xylene.  Among some of the other
parameters which were monitored are TOC concentration, which was around 8.0
mg/1, and a fairly high iron concentration of about 5 mg/1.  Presented in
Figure 3 are the pilot-plant and small-scale results for trichloroethene.
The small-scale column was designed assuming perfect similarity with the
5.4 minute EBCT pilot-plant column results which had been in operation for
a period of about 5 months.  As shown the small column results break
through much earlier than those of the pilot plant, even with the influent
to the small column being about 21 percent  lower than that of the pilot
column.  These results seem to contradict the previous results presented
for the six component laboratory study in which near perfect similarity was
obtained.  For the laboratory study, it turned out  that  the breakthrough
profiles were liquid-phase mass transfer controlled and  good results were
expected because the Stanton numbers were identical; whereas, for these
results, intraparticle mass transfer was controlling.  In this case the
scaling equations assuming constant diffusivity gave poor results.  Similar
results were obtained for the other VOCs that were  present  in the water
matrix(l,2).
                                    44

-------
        o
        o
        o
      _J a

      DJ o
      -J CO
      > <=
      < O

      DC S
      UJ
      O
      o
      o
        o

        d b*
                   • = Chloroform-Pilot
                   • = Chloroform-Small
                   a = Chloroform-Pilot Influent
                   o = Chloroform-Small Influent
           0.0
2.0.
 4.0         6.0

BED VOLUMES
                                  8.0
10.0

*103
Figure 1.   Comparison of the chloroform effluent profiles for the
            rapid small-scale column  test and the pilot-scale columns,
CONCENTRATION ug/L
0.0 1000.0 2000.0 3000.0 4000.0
D
Qi D
D a
^ rP ^
aDa u cPa
-a a
a Q
• = Trichloroethene-Pilot
• = Trichloroethene-Small
D = Trichloroethene-Pilot Influent
o = Trichloroethene-Small Influent
« *^"
* " l"'1
• * * * • i~i i r' " ' '
0.0 2.0 4.0 6.0
BED VOLUMES
0 0
•
•
•

8.0 10.0
*103
  Figure 2.  Comparison  of  the trichloroethene effluent  profiles for
             the rapid small-sale column test and the  pilot-scale
             columns.

                                   45

-------
TABLE 3.  AVERAGE INFLUENT VOLATILE ORGANIC CHEMICAL CONCENTRATIONS FOR
         THE FIRST FIVE MONTHS OF THE PILOT PLANT STUDY ON THE
        CONTAMINATED GROUNDWATER FROM WELL NO.  4 IN WAUSAU, WI
         Volatile Organic
             Chemical
Average Influent
 Concentration
     (P9/L)
      cis-l,2-Dichloroethene
      Trichloroethene
      Tetrachloroethene
      Toluene
      Ethyl Benzene
      m.o.p-Xylenes
      Vinyl Chloride
      1,1,1-Trichloroethane
      83.2
      72.0
      58.2
      30.9
       5.1
       7.6
       6.4
       1.3
                                  46

-------
    o
    C4
    O
    10
  D>
  n
  Z
  O
  uu
  o
  z
  o
  o
   BED PARAMETERS
     PILOT-SCALE
Loading Rate = 4.59 m/hr
Bulk Density = 520 kg/m3
    EBCT = 5.4 mln

    SMALL-SCALE
Loading Rate = 22.7 m/hr
Bulk Density = 481 kg/nv*
    EBCT== 0.203 mln
                                                « = Effluent - Pilot
                                                • = Effl - Small Column
                                                Q = Influent - Pilot
                                                o = infl - Small Column
I  TRICHLOROETHENE   |
            II •!!
      0.0
                5.0
                           10.0        is.o       ao.o
                               BED VOLUMES FED
                                                          25.0
                                                                    30.0
Figure 3.   Comparison of the trichloroethene breakthrough profiles
            for the rapid small-scale  column test (EBCT = 0.203 minute)
            and the pilot-scale  columns (EBCT = 5.4 minutes).  The  tests
            were conducted on contaminated groundwater in Wausau, WI.
                                    47

-------
    There are a number of possible reasons for the  discrepancies  between
the pilot plant results and the small  column results  such as  differences in
influent concentration, the impact of  the TOC, the  differences  in isotherm
capacity, and differences in surface diffusivities.  With respect to the
influent concentration, the VOC concentrations for  the RSSCT  were lower
than those observed in the pilot study,  which would cause the RSSCTs to
break through later than the pilot study.  However, the RSSCT data appears
before that of the pilot study.  Similar trends were  observed for the other
VOCs present in the water matrix.  The TOC background remained  constant
during the study.  VOC isotherms that  were conducted  in the raw water
demonstrated that TOC did not cause significant competition with  the
aliphatic VOCs.  However, recent evidence has shown that if TOC is pre-
adsorbed onto the GAC, significant competition with the aliphatic compounds
can occur causing a large reduction in the GAC capacity.  Also, with
respect to the solute capacity, it was shown that the single  solute
isotherm capacities for both the small and large size carbons were
identical(1).  The only other possible explanation  was that the surface
diffusivity of the small GAC used in the RSSCT may  be lower than  that of
the larger GAC used in the pilot study.

    Some work previously conducted in  Germany has indicated that  the sur-
face diffusivity is a function of particle size(9).  The data were corre-
lated and the results showed a linear  decrease in the diffusivity with
particle size.  This contradicts the assumption of  constant diffusivity
that was used in the development of Equations 1 and 2.

    Presented below are the scaling equations for non-perfect similarity
which emphasize intraparticle control.
        EBCTsc

        EBCT[c"
RSC
RLC
DS,LC
DS.SC
(3)
        VLC     RSC       ReLC-sc
 Equation 3  is the scaling equation for intraparticle transport where it
 takes  into  account the surface diffusivity as a function of particle size.
 This was derived by equating the surface diffusion modulus of the small
 column to that of the large column and assuming that pore diffusion is
 negligible.  Equation 4  is the scaling equation for reducing the effects of
 dispersion  and film transfer.  This equation was derived by equating the
 Stanton number of the small column to that of the large column.  To deter-
 mine the hydraulic loading of the small column, a minimum Reynolds number
 of  the small column must be chosen such that dispersion is not important
 and the Peclet number of the small column is equal to or greater than the
 Peclet number of the large column(2).  The development of Equations 3 and 4
                                     48

-------
is presented by Berrigan(l).  It turns out that if Equations 3 and 4 and
the correlation for surface diffusivity as a function of particle size are
used for scaling, the RSSCT results that were presented in Figure 3 for the
5.4 minute EBCT, should compare to the pilot plant results for an EBCT of
1.0 minute.

    Presented in Figure 4 are the results for trichloroethene.  Plotted in
terms of reduced concentration as a function of grams of TCE fed per kg of
adsorbent are the influent and effluent profiles for both the RSSCT and the
pilot plant data for an EBCT of 1.0 minute.  The data were plotted in terms
of grams of solute fed per kg of adsorbent to account for the differences
in the influent concentrations of the RSSCT and the pilot data.   As shown
the comparison is satisfactory.  Similar results were obtained for DCE,
PCE, and toluene.  Comparisons of the other solutes were not made because
analytical precision would not allow accurate comparisons due to their low
concentrations.

    In summary, the RSSCT procedure is promising for determining full-scale
adsorber performance and considerable time and expense can be saved in
determining full-scale adsorber performance with a properly designed small
column study using a smaller adsorbent particle size.  However,  more field
testing is required, because the extent to which surface diffusivity
changes with particle size has yet to be fully characterized and the impact
of humic material on the RSSCT procedure must also be investigated.

                              ACKNOWLEDGEMENTS

    This  research is based upon work supported by the EPA under a coopera-
tive agreement  from the Municipal Environmental Research Laboratory (No.
CR811150-01-0),  the City of Wausau, Wisconsin, and the Water and Waste
Management Group at Michigan Technological University.
                                     49

-------
     O o
     O ci

     O
     UJ
     o

     05
     o
     o
     UJ
     o
     13
     Q u,
     UJ a
     a:
   BED PARAMETERS
     PILOT-SCALE
Loading Rate = 4.59 m/hr
Bulk Density = 480 kg/m3
    EBCT = 1.0 mln

    SMALL-SCALE
Loading Rate = 22.7 m/hr
Bulk Density = 481 kg/m3
   EBCT= 0.203 mln
• •= Effluent - Pilot
• = Effl - Small Column
a = influent - Pilot
o = infl -  Small Column
                                                  TRICHLOROETHENE
                                                    CALGON F-400
         0.0
                   0.5        1.0
                           g SOLUTE FED
                            1.5         2.0
                            / kg ADSORBENT
                                                 2.5
                                                                       3.0
Figure 4.  Comparison of the  trichloroethene breakthrough profiles
           for the rapid small-scale column test  (EBCT = 0.203 minute)
           and the pilot-scale  columns (EBCT =  1.0 minute).  The  tests
           were conducted on  contaminated groundwater in Wausau,  WI.
                                   50

-------
                               NOMENCLATURE
ROMAN LETTERS
BVFmax        =  number of bed volumes that can be treated when the volume
                 of water fed equals the capacity of the GAC (dimension-
                 less).
C0ji          =  initial bulk phase concentration (M/L3).
Dei           =  axial dispersivity based on adsorber length and intersti-
                 tial velocity (L2/t),
Dgp i         =  pore solute distribution parameter (dimensionless) ;
                 ep(l-e)/e.
    i         =  surface solute distribution parameter (dimensionless);
Dp>i          =  pore diffusivity based on pore void fraction (L2/t).
DSsi          =  surface diffusivity (L2/t).
Dia           =  column diameter (L).
DL            =  free liquid diffusivity (L2/t).
EBCT          =  i/e, VB/Q or L/VS, fluid residence time in the bed which
                 is devoid of the adsorbent or empty bed contact time (t).
Edpj         =  pore diffusion modulus (dimensionless); DpjDgp ,-ft/R2.
Edsj         =  surface diffusion modulus  (dimensionless); Ds>iDgS}-ji/R2.
i             =  subscript denoting a solute i.
Ki            =  Freundlich  isotherm capacity constant (M/M)(L3/M)1/n.
kfj          =  film transfer coefficient  (L/t).
l/n^          =  Freundlich  isotherm intensity constant (dimensionless).
L             =  length of fixed-bed (L).
?Q]           =  Peclet number based on interstitial velocity and adsorber
                 length (dimensionless); Lv/Dej.
PejtD         =  2vR/De, Peclet number based on particle diameter
                 (dimensionless).
qe,i          s  adsorbent phase concentration in equilibrium with initial
                 bulk phase  concentration (M/M); KiC0>i1/ni.
                                     51

-------
Q             =  fluid flow rate (L3/t).
R             =  adsorbent radius (L).
Re            =  Vpi_2R/u, Reynolds number (dimensionless).
Sc            =  y/D[_p|_, Schmidt number (dimensionless).
Sti           =  modified Stanton number (dimensionless); kffit(l-e)/Re.
v             =  interstitial velocity (L/t); vs/e.
vs            =  superficial velocity (L/t).
VB            =  volume of the bed (L3).
Vp            =  pore volume per mass of adsorbent (L3/M).
GREEK LETTERS
e             =  fraction of volumetric space in reactor unoccupied by
                 adsorbent, or void fraction (dimensionless).
ep            =  fraction of volumetric space in adsorbent phase unoccupied
                 by adsorbent on the pore volume fraction (dimensionless).
Pa            =  adsorbent density which includes pore volume (M/L3).
Pb            =  adsorbent bulk density (M/L3).
PL            =  density of water (M/L3).
T             =  fluid residence time in packed bed, or packed bed contact
                 time  (t).
u             =  viscosity of water  (M/L-t).
                                    52

-------
                                REFERENCES

1.   Berrigan, O.K. Jr.  Scale-Up of rapid small-scale adsorption tests  to
    field-scale adsorbers:  theoretical  and experimental  basis.   Master's
    Thesis presented in partial  fulfillment of the requirements  for the
    degree of Masters of Science in Chemical  Engineering, Michigan
    Technological University, Houghton,  MI, 1985.

2.   Crittenden, J.C., Berrigan,  J.K., Hand, D.W.,  and Lykins,  B. Jr.
    Design of rapid fixed-bed adsorption tests for non-constant  dif-
    fusivities.  Environ. Eng. Div., Am, Soc. Civ. Enc^  113:2, 1987.

3.   Crittenden, J.C. and Weber,  W.J. Jr.  A predictive model  for design of
    fixed-bed adsorbers:  multicomponent model verification.   J^  Environ.
    Eng. Div., Am. Soc. Civ. Eng.. 104,  1978.

4.   Crittenden, J.C., Wong, B.W.C., Thacker, W.E., Snoeyink,  V.L. and
    Hinrichs R.L.  Mathematical  modeling of sequential loading fixed-bed
    adsorbers.  J^ Water Pollut. Control Fed.  52, 1980.

5.   Crittenden, J.C., Berrigan,  J.K, Jr. and Hand, D.W.   Design  of rapid
    small-scale adsorption tests for a constant surface diffusivity.  Jour.
    Wat. Poll. Con. Fed. 58:4, 1986.

6.   Crittenden, J.C., Hand, D.W., Friedman, G., Kato, S., Berrigan, J.K.,
    Luft, P.J., and Lykins, B.  Design of fixed-beds to remove multicom-
    ponent mixtures of  volatile organic chemicals.  J_._ AWWA.  (in press,
    1987).

7.   Friedman, G.  Mathematical modeling of multicomponent adsorption in
    natch and fixed-bed  reactors.  Thesis presented in partial fulfillment
    of Master of Science degree in Chemical Engineering,  Michigan
    Technological University, Hougton, Michigan, University Microfilms, Ann
    Arbor, 1984.

8.   Hand, D.W., Crittenden, J.C. and Thacker, W.E.  Simplified models for
    design of fixed-bed  adsorption systems.  3± Environ.  Eng.  Div., Am.
    Soc. Civ. Eng.  110,  1984.


9.   Schneider, R.  Bestimung der effektiven Korndiffusion-skoeffizienten Ds
    nach dem Overflachendiffusiosmodell fur Korn-und Pulver kohlen
    unterschiedlicher Partikelgrobe.  MS thesis, Univ. of Karlsruhe,
    Federal Republic  of Germany, 1982.

10. Thacker, W.E., Snoeyink, V.L., and Crittenden, J.C.  Desorption of com-
    pounds during operation of GAC adsorption systems.  J^ Am. Water Works
    Assoc. 75, 1983.

11. Thacker, W.E., Crittenden, J.C., and Snoeyink, V.L.  Modeling of
    adsorber performance:  variable  influent  concentration and  comparison
    of Adsorbents.  J.  Water Pollut. Control  Fed.  56, 1984.
                                     53

-------
         REMOVAL OF VOLATILE  ORGANIC  CHEMICALS  FROM  AIR  STRIPPING
               TOWER OFF-GAS  USING  GRANULAR  ACTIVATED  CARBON

                       by:  John  C. Crittenden
                           Department  of  Civil  Engineering
                           Michigan  Technological University
                           Houghton, MI   49931

                           Randy D.  Cortright
                           Universal Oil  Products
                           Chicago,  IL

                           Brad  Rick
                           Amway Corporation
                           Grand Rapids,  MI

                           Shin-Ru Tang
                           Department  of  Civil  Engineering
                           Michigan  Technological University
                           Houghton, MI   49931

                           David Perram
                           Water and Waste  Management Programs
                           Michigan  Technological University
                           Houghton, MI   49931

                           Tim Rigg
                           Water and Waste  Management Programs
                           Michigan  Technological University
                           Houghton, MI   49931


                               INTRODUCTION

    In recent years, air strippers  have been used to remove  volatile organic
chemicals (VOCs) from contaminated  ground  water.  The  cost of  air stripping
treatment without control  of  VOC  emissions is  considerably cheaper than the
use of liquid-phase granular  activated  carbon  (GAC).  However, recently
there have been concerns about the  resulting VOC air pollution.  In this
work, the treatment scheme shown  in Figure 1 which uses  a fixed-bed GAC
adsorber with on-site steam regeneration was evaluated.

    In this process, air from the top of  the air stripper is  first heated
to reduce the relative humidity and then  the VOCs are  removed  by GAC.   Once
the treatment objective is exceeded,  the GAC is taken  off-line and steam
                                     54

-------
    RAW WATER
TR[
w;
-*»•'
AIR FLOW
,. 	 t
"^ ^ STRIPPING T0>
WITH DEMIST!
TO PARTIALL'
REMOVE AER
TRE
OFF -6;
rr
IATED AIR
kTER IN
/
REGENERANT
OFF -GAS
RECYCLE
• 	 *— | CONDENSEF
WATER GRAVITY
PH!SE SEPARATOR
V 	 -**
Y ORGANIC
PHASE (S)
WER
DR
f
OSOL
ATEO
\S FLOW
DRYING
GAS
RECYCLE
^
-i
^X
3
o
r
i
i i
s
S
/

\ <
( J BLOWER
/^N
AIR
HEATER
X
GAS PHASE
ADSORBER
yBATCHWISE
GAC
REGENERATION
TEAM
:i
.^ DRYING
i' "^ GAS
'JX
LOW
TEMPERATURE
REGENERATION
r
        TO
  AQUEOUS-PHASE
    ADSORPTION
       UNIT
Figure  1.  Process flow for the air stripping solvent recovery process,
                                55

-------
regenerated.  For the low GAC loadings such as those found with off-gas
concentrations of 1 to 5 ug/1 (STP), TCE in the condensate must be treated
with aqueous-phase GAC because very little separate organic phase is
formed.  Once the gas-phase GAC is regenerated, it must be dried to remove
the water from its pores and cooled down.  Since the drying gas also repre-
sents a considerable source of pollution, It must be treated by mixing it
with the off-gas from the tower as shown in Figure 1.

    The important issues concerning the design of the integrated stripping
and GAC process which were evaluated in this study are:  (a) the GAC usage
rates and bed sizes for aqueous and gaseous phase treatment were compared
for the commonly occurring synthetic organic chemicals (SOCs) in ground water
by Cortright  (1); (b) a thermodynamic model which described the impact
of relative humidity on GAC equilibrium capacity for VOCs was developed and
verified by Tang (2); (c) thermodynamic models which predicted competi-
tive interactions for a binary mixture were developed and verified by Tang
(2); (d) the  feasibility of using steam and liquid carbon dioxide to
regenerate the VOC-laden GAC was evaluated by Rick (3); and (e) finally,
based on the  bench and pilot scale work, the cost of air stripping with and
without GAC treatment and aqueous phase GAC treatment were compared by
Rick (3).

    This paper focuses on the use of models for determining the cost of
treatment, GAC usage rate, bed design, and the feasibility of steam
regeneration.

                    EXPERIMENTAL METHODS AND MATERIALS

ANALYTICAL MEASUREMENTS

    Gas phase samples were analyzed according to the Environmental Protec-
tion Agency reference method 23 except an electron capture detector and a
10-foot packed column of 0.2 percent carbowax 1500 on 60/80 carbopack C
were used (4).

GRANULAR ACTIVATED CARBON

    Table 1 displays the physical properties of the GAC that was used in
the experimental phase of the study.

GAS PHASE PILOT PLANT

    Figure 2  displays a schematic diagram of the gas phase adsorption pilot
plant.  The source of contaminated air was a slip stream from an existing
full-scale stripping tower and the relative humidity was controlled by
passing the air through an electric heater.  The temperature, relative
humidity, and gas analysis were taken before and after each vessel.  An
accumulated air flow meter was used to determine the total amount of gas
flowing through the pilot plant.  The diameter of the pilot plant column
was 26.47 cm  such that channeling did not occur.
                                     56

-------
                    TABLE 1.  PROPERTIES OF THE CARBON
Carbon
BPL carbon (Calgon Co., Pittsburgh, PA)
size of carbon
average diameter
apparent density
density of carbon
particle void fraction
average bed density
bed void fraction
total surface area
(N2,BET)
    :dp
    'Pa
    :pc
    :ep
    'Pb
    :e
4x6 mesh
.3715 (cm)
.85 (g/cm3)
2.1 (g/cm3)
.595 ( - )
(gram GAC)/(cc of Bed volume)
1.0 - Pb/Pa
1050-1150 (mz/g)
                                    57

-------
H20
IN
\i i i i i i i\
          H20
          OUT
       AIR
       IN
                                   BLOWER
                                 4* INSULATED
                      (PRESSURE
                    X

                                  PRESSURE
                                SAMPLE
                                PORT
                               !—CXJ
                               TEMPERATURE
                               (. RH
                   -a
                              x
                                                                    TEMPERATURE
                                                            SAMPLE   t "H
                                                                  „
                                                                  ?      n
                                                                                      HEATER
                                                                                  4* INSULATE
                   \    /
                   ADSORBENT
                   CHAMBERS
                   (INSULATED)
SAMPLE
PORT
                                                           TEMPERATURE
                                                           b RH
                                                               ACCUMULATED AIR,
                                                               TLOM METER
             Figure  2.   Schematic of  the  gas-phase pilot  plant.
                                          58

-------
           DESCRIPTON OF THERMODYNAMIC AND MASS TRANSFER MODELS

CORRELATION OF SINGLE SOLUTE GAS PHASE ISOTHERMS

    In this study, the Dubinin-Radushkevich (D--R) isotherm was shown to
correlate the isotherms of a number of VOCs (2).  Based on that work and
the work of Reucroft et al. (5) and Rasmuson (6), the following form of the
D-R Equation was used to estimate the gas-phase capacity:
    q  =  pi  W = P! W0 exp
B en2
   D                                    (1)
    eD = RT In  (PS/P)                                                   (2)
 t

 in which, a, the polarizability, may be calculated from the Lorentz - Lorenz
 Equation if it  is not known:
             -  1]  M
    a  = —s	                                                    (3)
        [n2  +  2]  pL


 If  the refractive index  is  not  known,  it may be estimated from a summation
 of  atomic  and  structural  contributions  (7).

    Equations  1  and 2  were  found  to  correlate  isotherm data for compounds
 with  dipole  moments less than 2 debyes  (2 X 10"18 esu - cm).  W0 and B have
 been  found to  be dependent  on the nature of the adsorbent.  Accordingly,
 when  the  data  is plotted as W versus (e/a)2 the data conforms to essentially
 one curve  for  different  adsorbates and  temperatures.  For this study, the
 constants  W0 and B which were used for  BPL carbon were 0.46 and 3.22 x 10~5
 (ca!2/gm-mole2), respectively.

    There  are  several  important limitations of the  D-R which must  be con-
 sidered when using  it  to predict  gas phase capacity.  First, it accounts
 for only  physical adsorption by weak physical  forces; accordingly,  it can
 not be used  to account for  adsorption capacity when capillary condensation
 occurs.   If  the  pore  size distribution  is known the region of P/PS  where
 capillary  condensation occurs,  can be estimated from the Kelvin Equation:
                -2y V|_ cos o
                   Fr
                                      59

-------
Typically if P/PS is much less than 0.2 then capillary condensation is
unimportant.  Another limitation of the D-R equation is that the adsorbed
state is assumed to behave as a condensed liquid; consequently, the liquid
density which appears in Equation 1 could only apply to temperatures below
the critical temperature of the vapor.

CORRELATION OF SINGLE SOLUTE LIQUID-PHASE ISOTHERMS

    Crittenden et al. (8) demonstrated that the following correlation
could be used to calculate the Freundlich isotherm parameters for 10
hydrophobic compounds including halogenated aliphatic and aromatic
compounds.
    q = PL W = PL W0 exp
                                    RT In
                                       'mi
                                                  (5)
 An  average  error of about 10 percent was observed for the Freundlich para-
 meters.  Accordingly, this correlation was used to predict the single
 solute  capacity of liquid-phase GAC.  When the aqueous-phase GAC capacity
 is  compared to the gas-phase GAC, the parameters which were used in
 Equation 5  are those for F-400 carbon.  W0, B, and a were found to be
 0.6299  cm3/gm, 0.02766  (g-mole/cal)1-208 and 1.208 by Crittenden et al.
 (8)  and Speth  (9).

 PREDICTION  OF THE  IMPACT OF RELATIVE HUMIDITY ON THE ADSORPTION OF VOCs

     The thermodynamic model which was proposed by Okazaki et al. (10) was
 used to predict the impact of relative humidity (RH) on VOC adsorption.
 The three basic mechanisms which are included in the model are shown in
 Figure  3.   In  larger pores that have not been filled by capillary conden-
 sation  of water vapor,  VOCs adsorb onto essentially dry walls without com-
 petition by adsorbed water because there are very few hydrophillic sites on
 the GAC surface.  Accordingly, the VOC capacity on this surface area, Q°QI
 is  given by the D-R equation.

     In  smaller pores where capillary condensation of water has taken place,
 VOCs will be dissolved  into the condensed phase.  The amount of VOCs in the
 condensed phase, Qn,2 is given by the following equations:
          Pi
           -  exp
¥mi
       In
RH
TOO
(6)
     Q02  =  Vc
                                                  (7)
                                     60

-------
     QOI
           DRY PORES
D-R


>


EQUA
t'j
) C
> c
) C
> c
> c
0 C
TION HENRY'S LAW
| 1


QQ, — »-

t
\

•«

^~
-7-7
I

— Q02

- CONDENJ
                   AQUEOUS  ISOTHERM
                  = Q
                     oi
          MODEL OF ADSORPTION WITH

           CAPILLARY CONDENSATION
Figure 3.  Okazaki's model for adsorption which predicts the impact
        of relative humidity on VOC adsorption.
                        61

-------
Equation 6 contains a term which accounts for the impact of a curved
miniscus on the partitioning of solute into the pore.   A detailed deriva-
tion of this is given by Tang (2).   The Henry's constants for TCE were
predicted using the results of Cosset et al. (11).

    The amount of condensed volume, Vc, which is assumed to be mostly
water, was given by a water vapor isotherm (12).  Since water vapor
isotherms exhibit hysteresis when the adsorption isotherm is compared to
the desorption isotherm, more discussion on the evaluation of Vc is
required.  According to the data of Okazaki et al.  (10), hydrophobic com-
pounds do not affect the hysteresis of water vapor  isotherm; therefore, the
amount of condensed volume is given by the adsorption  isotherm, if relative
humidity is controlled and remains constant.  On the other hand, if the
adsorber has been exposed to high humidity and then lower humidity, the
condensed volume is given by the water vapor desorption curve.  This is
very important as far as the operation of the adsorber.  Because it implies
that if the adsorber has been exposed to high humidities, it will be
necessary to dry out the bed at much lower humidities  in order to reduce
the condensed water vapor volume.  The condensed volume for removing water
from these pores would be given by the water desorption isotherm.

    Another contribution to the GAC  loading in the condensed phase is
aqueous phase adsorption onto the pore walls, Q°o3«  An aqueous phase
isotherm was used to estimate the amount of adsorbed TCE onto the wet pore
surface in the condensed phase.


    Q03 = < C1/n                                                         (8)


The Freundlich parameters, K and 1/n,  are  893. yg/gm (l/ygj^and 0.3985
and are valid for a concentration  range  of  15 to 102 yg/1.

    In  order to  calculate  the aqueous  phase capacity at other temperatures,
the aqueous phase TCE  isotherm was fit to  Equation 5.   Itaya et aJ.  (13)
demonstrated that the  correlation  in Equation 5 was independent of tempera-
ture  for  several adsorbate-adsorbent systems; consequently,  it was felt
that  Equation 5  was accurate enough  to estimate the effect  of temperature.
Although  Equation 5  is  assumed  to  be independent of temperature,  it will
predict a decrease  in  adsorbility with increasing temperature because  the
solubility  is a  function of temperature.   The solubilities  for Equation  5
were  predicted using UN1FAC  (9).  The  remaining variable that must be  esti-
mated  is  the fraction  of wet surface area,  Sw/$t, and  the  fraction of  dry
surface area, S
-------
PREDICTION OF COMPETITIVE INTERACTIONS BETWEEN VOCs

    For relative humidities less than about 45 percent, capillary conden-
sation of water vapor did not occur for BPL Carbon.  Accordingly, only com-
petitive interactions between VOCs need to be considered for the case where
RH is controlled before GAC.  Tang (2) demonstrated experimentally that
competitive interactions between VOCs could be predicted using ideal
adsorbed solution theory and Polanyi potential theory.

MASS TRANSFER MODELS

    These four mass transfer models were compared in this study in order to
determine the necessary degree of complexity which is required to predict
fixed-bed behavior:  (a) the dispersed-flow pore-surface diffusion model
(DFPSDM), (b) the dispersed-flow homogeneous surface diffusion model
(DFHSDM), (c) the plug-flow homogeneous surface diffusion model (PFHSDM),
and  (d) the film transfer constant pattern model (FTCPM).

     The most complex model, DFPSDM, incorporates mathematic descriptions of
the  following processes:  (1) axial transport by advective and dispersive
flow,  (2) diffusion  resistance  in the gas phase surrounding the adsorbent
particle,  (3) local  equilibrium adjacent to the exterior of the adsorbent
surface and within the  pores, (4) pore and surface diffusion resistance
within  the adsorbent, and  (5) competitive equilibrium of solutes upon the
carbon  surface.  Crittenden et  aJ.  (15) and Friedman  (16) have presented
the  equations and  their solution.

     The three simpler models may be compared  to the DFPSDM in order to
describe  their  essential assumptions.  The DFHSDM contains the identical
mechanisms as the  DFPSDM except the contribution of the pore diffusion  is
dropped.  The PFHSDM ignores the contribution of axial dispersion and pore
diffusion.   The simplest model, FTCPM, includes only  film transfer resis-
tance.  Friedman  (16) demonstrated  that the pore diffusion contribution was
only necessary  when  multicomponents were present and  the mass transfer
zones overlapped.   In this  study, pore diffusion was  not needed to predict
the  breakthrough of  a binary mixture  of PCE and TCE and axial dispersion
was  not needed  for gas  phase adsorption (1).  Furthermore, FTCPM was found
to be adequate  for calculating  breakthrough profiles  for single components
with constant influent  concentrations because this is the major mass
transfer  resistance  as  long as  the  pores of the GAC are not filled with
condensed water vapor.

     The following  analytical solution to the  FTCPM for the case  in which
1/n  is  less  than 1.0 is reported here because it is not complicated to  use
and  is  valid for many design calculations  (17).
 T  =
     3St(Dg+l)
ln(C/C0)  - (n
- (C/C0)
+  Y
+ 1   (10)
                                     63

-------
where, Y is defined by the following series:
    Y =
            1
            n
         I
        k=l
r
k k
L
1 '
1 - -
n
1 '
+ -
n
                                        (1)
In order to use Equations 10 and 11, the bed must be long to establish a
mass transfer zone that remains constant in shape.  This condition is known
as constant pattern (18).

MASS TRANSFER PARAMETER ESTIMATION

    In order to use the mass transfer models, estimates of the following
parameters are needed:  the molecular diffusion coefficient, Dm, axial
dispersion coefficient, De; film transfer coefficient, kf; pore diffusion
coefficient, Dp; and the surface diffusion coefficient, Ds.  The Wilke-Lee
(19) modification of the Hirschfelder-Bird-Spotz correlation was used to
estimate gas diffusion coefficients.  The gas phase axial dispersion coef-
ficient was calculated using the following correlation which was proposed
by Miyauchi and Kikuchi (20):
       dh
Pe,
                      T Pe,
    x = 22 / Pem2/3
               ;m
            :m
                                            (l-e-2x)
x Pen
(12)



(13)
    The film transfer coefficient was estimated using the following
correlation which was proposed by Wakao and Funzukri (21):
    Sh = 2 + 1.1 ReO-6 ScO-33
                                                         (14)
This correlation is valid for Reynolds numbers between 3 and 10,000.  Since
film transfer was the most important diffusion resistance, short fixed bed
tests were conducted in order to compare film transfers coefficients with
Equation 14 (1).  In this study, it was found that Equation 14 was within
15 percent of the observed for Reynolds numbers between 50 and 200 and for
both TCE and PCE.  This Reynolds number regime is within the normal
operating limits of gas phase adsorbers.  For the liquid phase mass
transfer coefficient, the correlation which was proposed by Williamson et
a/- (22) was used.
                                     64

-------
    The pore diffusion flux was found to be negligible in liquid and gas
phases; consequently, the methods which were used to estimate it are not
discussed but are reported by Cortright (1).

    The surface diffusion coefficient was found from fitting effluent
breakthrough data from several pilot plant studies.  In an attempt to make
the results more general, it was compared to a correlation which was pro-
posed by Dobrzelewski (23).  The correlation is based on the observation
that the surface flux was approximately a constant factor times the pore
flux for a number of VOCs in the aqueous phase.  The factor known as the
surface to pore diffusion flux ratio, SPDFR, was used as the fitting param
eter as shown in the following equation:
    Ds -- - - —  x  (SPDFR)                                      (15)
         *D Pa K C01/n


    The best description of the data was with a SPDFR of 16 but the upper
bound of the 95 percent confidence limit could not be determined because
the mass transfer  rate was limited by film diffusion. For the aqueous
phase, the correlation which was proposed by Dobrzelewski (23) was used.

                           RESULTS AND DISCUSSION

PROCESS DESIGN

    The design of  the gas-phase adsorber is dependent on the design of the
air stripper. The  volumetric flow rate and gas phase concentrations will
depend on the air  to water ratio, ground water VOC concentration, the treat-
ment objective, and the Henry's constant of the VOC.  Hand et al . (24)
have discussed a design procedure for packed tower air strippers that mini-
mizes tower volume and energy  consumption.  In that paper and other work,
Hand et aJ.  (24) found that for most compounds, an air to water ratio of
approximately 3.5  times the minimum air to water ratio minimized tower
volume and energy  consumption.  A range of Henry's constants from 0.093 to
265 were examined. Consequently, in the design of the gas phase adsorbers
an air to water ratio of 3.5 was used to determine the volumetric air flow
rate and gas phase concentrations.  Table 2 reports the Henry's constants
and the heat of dissolution that was used to account for impact of tem-
perature on the Henry's constant.  For this study, it was assumed that the
ground water temperature was 10°C and the off gas was heated to 24°C to
eliminate the impact of humidity.  Table 3 reports the optimum designs for
air stripping towers that  remove some common ground water contaminants.

COMPARISON OF THE  ADSORPTIVE CAPACITY OF ADSORBENTS

    In Figure 4, the adsorption capacity of different adsorbents for TCE
are compared using their respective D-R characteristic curves.  BPL, KG-BAG,
and resin correlations were obtained from the investigators as indicated.
                                     65

-------
      TABLE  2.   HENRY'S  CONSTANTS  FOR  SYNTHETIC  ORGANIC  COMPOUNDS
Henry's Constant
[10 deg C] Reference
Compound [(ug/l)/(yg/l)] Henry's Constant
Trichloroethylene
Tetrachl oroethyl ene
Carbon
Tetrachloride
1,1,1 -
Trichlorethane
1,2 -
Dichloroethane
Vinyl Chloride
Dichloromethane
1,1 -
Dichloroethene
cis 1,2 -
Dichloroethene
Benzene
Toluene
M - Xylene
Chlorobenzene
1,2 -
0.116
0.295
0.556

0.172
0.023
265.0
0.0484
0.935
0.0934
0.106
0.117
0.093
0.069
0.0896
Hand et aJ. (24)
Hand et al. (24)
Kavanaugh et aJ. (26)

Kavanaugh et al. (26)
Solubility, Vapor
Pressure Data
Kavanaugh et aJ. (26)
Solubility, Vapor
Pressure Data
Singley et al. (27)
Hand et aJ. (24)
Kavanaugh et al . (26)
Singley et al . (27)
Solubility, Vapor
Pressure Data
Singley et al. (27)
Solubility, Vapor
Heat of
Dissolution
[kcal/kmol]
3.41 x 103
4.29 x 103
4.05 x 103

3.96 x 103
3.93 x 103
-
-
5.66 x 103
3.48 x 103
3.68 x 103
4.17 x 103
3.80 x 103
-
—
Dichlorobenzene
Pressure Data
                                  66

-------
TABLE 3.  AIR STRIPPING DESIGNS FOR REMOVAL OF COMMONLY OCCURRING
                   SYNTHETIC ORGANIC COMPOUNDS

             Inlet Water Concentration - 100.0 ug/1
              Water Treatment Objective - 1.0 yg/1
               Air Stripper Temperature - 10 deg C
    Air Stripper Packing Pressure Drop - 5.0 (N/M2)/M Packing
      Air Stripper Packing - 3-inch Plastic Intalox Saddles
Compound
Trichloroethylene
Tetrachl oroethyl ene
Carbon Tetrachloride
1,1,1 Trichlorethane
1,2 Dichloroethane
Dichloromethane
cis 1,2 Dichloroethene
Vinyl Chloride
Benzene
Toluene
M - Xylene
Chlorobenzene
1,2 Dichlorobenzene
Henry's
Constant
yg/1 Air
ug/1 Water
0.116
0.295
0.556
0.172
0.023
0.048
0.093
265.0
0.106
0.117
0.093
0.069
0.090
Air to
Water
Ratio
29.9
11.8
6.2
20.1
150.6
71.59
37.10
0.013
32.69
29.62
37.26
50.29
38.67
Air
Stripper
Length
(meters)
11.59
13.34
13.68
12.21
10.20
8.72
10.63
18.16
11.05
11.90
12.34
11.46
12.33
Dimensions
Diameter
Packing
(meters)
2.47
1.82
1.51
2.16
4.54
3.39
2.66
0.58
2.55
2.46
5.59
6.93
2.70
                                67

-------
                         CORRELATED DATA
                         o= BPL-MTU
                            BPL-RASMUSON
                            BPL-REUCROFT
                            SORBO-NORIT
                            GAC-410G CECA
                            WV-G WESTVACO
                            WV-W WESTVACO
                            KG-BAG NOLL
                            RESIN-NOLL
        0.0    20.0    40.Q    60.0    80.0    100:0

      (ADSORPTION POTENT!AL)**2 (MILLION)


                (RTLNPS/P)**2(CAL/G MOLE)**2


Figure 4.  Characteristic curves for TCE on various adsorbents.
                          68

-------
The remaining correlations were obtained from the manufacturers.  According
to Figure 4, no significant differences in GAC capacity is expected to TCE
or other VOCs for the commercially available GACs; consequently, BPL was
used in this study.

GAC USAGE RATES FOR VOCs FOUND IN GROUND WATER

    Figures 5 and 6 display the expected GAC usage rates for commonly
occurring synthetic organic chemicals (SOCs) that are found in ground water.
These were calculated assuming a treatment objective of 1 ug/1 and the GAC
was assumed to be totally exhausted.  Using the appropriate air to water as
discussed in the integrated design section, the usage rates are reported as
a function of the aqueous phase concentration.  With respect to the impact
of relative humidity, the usage rates which are displayed in Figures 5 and
6 would be for 24°C and RH less than about 45 percent.

COMPARSION BETWEEN AIR AND AQUEOUS PHASE GAC USAGE RATES

    Table 4 compares the GAC usage rates for aqueous and gaseous phase
adsorption.  To determine the gas phase concentration, the treatment objec-
tive was set equal to 1 ug/1; the air to water ratio which is reported in
Table 4 was used; and the GAC was assumed to be in equilibrium with the
inlet concentration in the air or water.  The aqueous and gaseous phase
isotherm parameters for the SOCs were calculated using Equations 5 and 1,
respectively.  The temperature of the water was assumed to be 10°C and the
air was heated to 24°C to lower the relative humidity to 40 percent.
According to Table 4 the gaseous phase usage rate is approximately two to
four times greater than is observed in the aqueous phase.

IMPACT OF RELATIVE HUMIDITY ON VOC CAPACITY

    The impact of RH was investigated by heating the off-gas from the air
stripping tower assuming it is at 100 percent RH at 10°C.  Figure 7 shows
the effects of controlling the RH on the adsorbed amount of TCE.  The RH
and corresponding temperatures are given in Figure 7.  At high RH values a
majority of the pores are filled with water and its capacity is substan-
tially reduced.  As the temperature increases, the relative humidity is
reduced allowing for more of the pores to be dry and increasing the capac-
ity.  However, as soon as the pores are mostly dry further heating reduces
the capacity.  At  relative humidities between 40 percent and 50 percent,
the effects of RH and temperature balance out and a maximum loading is
obtained.  Although the results in Figure 7 are just for TCE, it is
expected that  similar results would be expected except for those noted in
Table 4 where  some compounds have a lower gaseous phase capacity than
aqueous phase  capacity.

VERIFICATION OF THE MASS TRANSFER MODELS

    In order to verify the mass transfer models,  the data from  three of  the
pilot plant  runs were compared to the mass  transfer models.  Velocities
ranging from 25 to 75 cm/sec and bed depths of 5  to  11 cm were  used.   In
spite of the fact  that these beds were very thin, two  to  six weeks were
                                     69

-------
       E
       Ul

       It
       O
       <
       0
       ra?
       — -
       LU1
       CO
       ID
                 D-R PREDICTED ISOTHERMS
                 TEMPERATURE = 24.0 deg C
               TREATMENT OBJECTIVE = 1.0 ug/L
O = TOLUENE
A=BENZENE
+ = XYLENES
v = CHLOROBENZENE
 = 1,2 DICHLOROBEN2ENE
© = TETRACHLOROETHENE
B = 1,1,1 TRICHLOROETHANE
        'o I  .  i  . i 1111.1 1111   i i  , i . 111,i  1111  i  i i \ . i. i.i 111
          101            1Cf            103            104
                WATER CONCENTRATION (ug/L)

Figure 5.  Low range GAC usage rates for common  ground water SOCs,
                              70

-------
        EC
 o
 *-•

 O

 CD



CD

CO
ID
                  D-R PREDICTED ISOTHERMS
                  TEMPERATURE = 24.0 deg C
                TREATMENT OBJECTIVE = 1.0 ug/L
                * = TRICHLOROETHENE
                * = 1.1 DICHLOROETHENE
                a = CARBON TETRACHLORIDE
                8 = 1,2 DICHLOROETHANE
                a = DICHLOROMETHANE
                 = CIS 1,2 DICHLOROETHENE
                0= VINYL CHLORIDE
           101             1(f            103'            104
                 WATER  CONCENTRATION (ug/L)

Figure 6.   High range GAC usage  rates for common ground water SOCs.
                                71

-------
TABLE 4.  COMPARISON OF GAC USAGE RATES FOR
   AN AQUEOUS CONCENTRATION OF 100 ug/1
Compound
TCE
PCE
CC14
III-TCA
1,2-DCE
CH2 C12
II-DCE
Cis-1,2 DCE
Vinyl Chloride
Benzene
Toluene
m-Xylene
Chlorobenzene
1,2 Dichlorobenzene
Aqueous Phase
(mg/1 H20)
4.47
1.68
2.69
13.3
19.6
33.2
17.6
21.6
212.0
9.94
2.22
0.916
2.23
0.683
Gas Phase
(mg/1 H20)
1.89
0.459
0.980
1.93
15.5
838.0
7.61
17.8
2.20
2.53
0.968
0.593
0.729
0.316
Air to Water
Ratio
29.8
11.8
6.2
20.1
150.6
71.6
3.7
37.1
0.0130
32.7
29.6
37.3
50.3
38.7
                    72

-------
required to saturate the beds and the mass transfer zones were shorter than
the bed depth that was used.  For example, in Figure 8, 10 mil ion bed volu-
mes fed corresonds to 11.5 days.

    Figure 8 compares the pilot plant TCE data for a velocity of 70.2
cm/sec to the mass transfer models.  Since no kinetic studies were con-
ducted, the SPDFR which is used to calculate the surface diffusivity was
determined by comparing the DFHSDM to the pilot plant data which was con-
ducted at a velocity of 25.1 cm/sec.  For the pilot plant run in Figure 8,
IAST was used to describe competitive interactions between PCE and TCE (1).
Based on the results in Figure 8, the PFHSDM predicts the effluent data
very well and the competitition interactions from PCE must be considered.

    Sensitivity analyses and other comparisons demonstrated that the FTCPM,
Equations 10 and 11, were adequate to predict breakthrough of single com-
ponents.  The only mass transfer parameter that is required, kf, may be
calculated from Equation 14.  Accordingly, the mass transfer zone lengths
which appear in the next section were calculated using Equations 10 and 11.

COMPARISON OF GAC CONTRACTOR SIZES

    Table 5 compares the mass transfer zone  (MTZ) lengths that are expected
based the gaseous phase pilot plant.  The correlation provided by
Dobrzelewski (23) was used  to estimate the surface diffusivity in the
aqueous phase and the HSDM  solutions given by Hand et aJ. (18) were used
to  estimate the MTZ lengths.  The assumptions that are built into these
calculations are:   (a) the  SPDFR for the aqueous phase was 3.72; (b) the
SPDRFR for the gaseous phase was 16.0; (c) the water temperature is 10°C;
(d) the RH was lowered to 45 percent by heating the off-gas to 24°C; (e)
the mass transfer zone length is defined as  containing the concentration
range of C/C0 0.95  to 0.05;  (f) the GAC is in a fixed position; and (g)
single solute adsorption is  taking place.

    The assumptions of fixed bed, single solute adsorption and a liquid
phase SPDFR of 3.72 need further discussion.  With respect to the fixed-bed
assumption, backwashing of  the  liquid-phase  GAC can destroy the MTZ and
deeper and/or multiple beds  in  series may be required.  However, in a
related study there was no  evidence of this  for a field-scale unit treating
200 gpm.   In that study a hydraulic loading  of 5 m/hr was used to treat an
anaerobic ground water and  gentle backwashing was required only every two to
four months.  With  respect  to the use of a SPDFR of 3.72 and  single solute
adsorption  in the aqueous phase, these are optimistic assumptions because
in  a  related study, it found that a SPDFR of 0.4 was needed to describe the
breakthrough data when 8 mg/1 of TOC was present.  Furthermore, competitive
interactions have been observed between TOC  and SOCs which would not be
considered  by assuming single solute adsorption.  Accordingly, the results
for the aqueous phase must  be viewed as a best case situation, whereas the
gas phase  results should be viewed as fairly accurate.
                                     73

-------
                EFFECT OF RELATIVE HUMIDITIES
                 ON TCE IN OFF-GAS ADSORBED
                  PREDICTED BY OKAZAKI'S MODEL
           RH TEMP
           100% 10.OC
            80% 13.OC
            70% 15.0C
            60% 17.50
            50% 20.3C
            40% 24.00
            30% 28.9C
     Ul
     O
                  20.0  30.0   40.0   50.0   60.0   70.0   80.0
                     PERCENT RELATIVE HUMIDITY
                                                     90.0  100.0
Figure 7.  Impact of relative humidity on  GAC capacity for TCE.
         q
         CT
         in
       _oi
        D)
       g-
       5.
        o
        O ">
          o
DFH3DM SIMULATION IMUITICOMPOHENT)
PFH3DM SIMULATION JMULTICOMPONENT1
'"VFHSDMVIM"U"LATION ijiNaLYsoluii)"
O EXPERIMENTAL INLET CONC- TCE '
• EXPEHIMENTAL OUTLET CONC - TCE



TWO VOC SYSTEM - TCE i PCE
BED DEPTH - 7.00 cm
VELOCITY - 70.3 cm/no
TEMPERATURE - 24.2 dig O
RELATIVE HUMIDITY - 41.0 H
5.0     10.0     15.0     20.0     25.0
       BED VOLUMES FED
                                                        30.0
                                                        *1Gf.
Figure 8.   TCE concentration data versus  model  predictions
           for pilot plant run  No.  5.  Gas  concentrations
           are reported at STP.
                              74

-------
TABLE 5.  COMPARISON OF GAC MASS TRANSFER ZONE LENGTHS AND CROSS-SECTIONAL
   AREAS FOR AN AQUEOUS PHASE CONCENTRATION OF 100 ug/1 TREATMENT OBJECT
  OF 1 yg/1 AND A FLOW RATE OF 8,175 m3/DAY (2.16 MGD).  THE AQUEOUS PHASE
  AND GASEOUS PHASE VELOCITY WERE 12.2 M/HR AND 25 CM/SEC, RESPECTIVELY.*
Compound
TCE
PCE
CC14
III-TCA
1,2-DCE
CH2 C12
1,1-DCE
C1S-1.2-DCE
Vinyl Chloride
Benzene
Toluene
m-Xylene
Chlorobenzene
1,2-Dichlorobenzene
Aqueous
Phase MTZ
(m)
.5787
.5039
.5497
.5266
.5507
.4754
.7350
.5990
.6176
.5026
.4946
.4843
.5318
.5168
Gas
Phase MTZ
(m)
0.0433
0.0301
0.0358
0.0431
0.0985
—
0.0807
0.105
Oo0762
0.0388
0.0283
0.0258
0.0287
0.0261
Cross-sectional
Area
(m2)
11.3
4.45
2.36
7.62
57.0
27.09
1.404
14.00
0.0049
12.4
11.2
14.1
19.0
14.6
 *The  required  cross-sectional  area  for aqueous GAC is 27.9
                                     75

-------
    The MTZs which are reported in Table 4 should be compared with the
following considerations in mind.   The best two bed in-series design, as
far as saturating the GAC,  would use an individual  bed length equal  to the
MTZ.  However, since the MTZs are  so short, a more economical design would
involve a single adsorber which is approximately three to five times longer
than the MTZ.

    As may be seen in Table 5, the GAC bed sizes for gas phase adsorption
are considerably smaller in most cases than for aqueous phase treatment in
spite of the fact that more fluid  volume is treated.  Table 4 reports the
air to water ratios that were used to size the gaseous phase GAC contactors.

EFFECTIVENESS OF STEAM AND  LIQUID  CARBON DIOXIDE REGENERATION

    Rick (3) has reported the details of steam and liquid carbon dioxide
regeneration for this study.  With respect to liquid carbon dioxide regen-
eration, Rick (3) demonstrated that 83 percent to 96 percent recoveries
of TCE could be obtained using a laboratory liquid carbon dioxide soxhlet
extractor.  GAC loadings of 1.5 percent to 4.5 percent by weight which are
typical for TCE gas phase concentrations of 1 to 3 yg/1 (STP) were used.
Although liquid carbon dioxide was effective in removing TCE at low
loadings, the capital investment of makng the GAC vessels to withstand cri-
tical pressure make the process too expensive.

    Table 6  summarizes the steam regeneration results on pilot column which
was regenerated three times.  The virgin column results correspond to the
data displayed in Figure 8.  The influent concentrations were variable over
the course of the study; consequently, the DFHSDM was run to assess the
GAC's virgin capacity by comparing it to column data for the regenerated
GAC.

    As shown in Table 6, the fraction of original capacity decreased with
each successive regeneration.  Based on the analysis of the  steam conden-
sate and a model for  regeneration, the cause for the loss in capacity is a
build-up of  PCE on the GAC which 100°C steam could not drive off.  However,
model calculations were made to assess whether 170°C, 115 psia steam would
drive off the PCE.  According to these calculations, a 70 percent recovery
of  the PCE loading which corresponds to the pilot plant 0.0139 by weight
would require approximately 50 kg steam/kg of GAC.  Since steam costs were
only two percent of the total cost for a steam usage rate of 20 kg/kg, this
higher usage rate could still be economically viable.

    Two additional factors  which must be considered when using steam regen-
eration are the concentrations in the condensate and the amount and concen-
tration in the drying gas.   With respect to the condensate, TCE was present
at or near its solubility limit with traces of a separate organic phase
appearing in the condensate.  The condensate was acidic with a pH range of
4 to 5 which indicates some dechlorination.

    One important consideration for this sytem was the presence of TCE in
the noncondensable gases in the regeneration system.  In an actual process,
                                     76

-------
        TABLE 6.  REGENERATION CONDITIONS AND OBSERVED TCE CAPACITY
                 FOR THE STEAM REGENERATED PILOT PLANT BED
Virgin
GAC
Mass of carbon (grams) 2012.7
Bed height (cm) 7.0
TCE loading after 52.9 g
bed exhaustion
Expected TCE loading 52.9 g
for virgin GAC
Weighted inlet 1.4 ug/1
concentration for TCE
Weighted inlet 0.35 ug/1
concentration for PCE
Steam superficial
velocity (cm/sec)
Condenser temperature
Steam quantity
(kg steam/kg GAC)
Steam Temperature:
Top of Column
Bottom of Column
Steam Pressure
Percent of Virgin 100%
Capacity
GAC
Regenerated
One Time
2012.7
7.00
50.6 g
63.4 g
1.6 ug/1
0.29 ug/1
3.52
22
17.5

110°C
100°C
1 atm
80%
GAC
Regenerated
Two Times
2012.7
7.00
50.1 g
71.4 g
2.2 ug/1
0.26 ug/1
3.52
22
17.5

110°C
100°C
1 atm
70%
GAC
Regenerated
Three Times
2012.7
7.00
40.3 g
67.8 g
2.2 ug/1
2.23 ug/1
3.52
22
17.5

110°C
100°C
1 atm
60%
*Based on DFHSDM simulation
                                      77

-------
this would have to be considered such that all  the VOCs  did not leave the
process with the noncondensable gases.   As far  as  drying gas is concerned,
initial concentrations of TCE of 1.7 mg/1  were  noted.   In this  study,
excess drying gas was used;  accordingly, the temperature and quantity of
drying need further examination.

COMPARISON BETWEEN LIQUID AND GASEOUS GAC PROCESS  COSTS

    Three design cases were  investigated for treating  0.5 MGD of water with
trichloroethene concentrations of 100 yg/1, 300 yg/1,  and 1,000 yg/1. A TCE
concentration of 1.0 y9/l in the water leaving  the air stripper was speci-
fied as the treatment objectives.  Table 7 gives the design parameters for
each of the three cases.  Figure 1 shows the process flow diagram and mass
balance for the 100 yg/1 design case.

    The approach which was described by Hand et al. (24) was used to
design the air stripper.  The theoretical optimum air to water ratio was
approximately 3.5 times the minimum air to water ratio.   However, in order
to maintain reasonable tower lengths, the air to water ratio was increased
from the optimum value in all three cases.  Three-inch plastic Intalox
saddles were used as the packing media.

    Table 7 reports the  length of the mass transfer zone.  The mass trans-
fer zone was approximated as the section of bed where the solute concentra-
tion drops from 95 percent of the inlet concentration at one end to five
percent at the other end.  As shown in Table 6, the mass transfer zones
were approximately 4 cm  long and are short compared to the bed length of
30.48  cm.  This would enable a  single adsorber to treat the off-gas while
the second adsorber  is  regenerated.  As shown by the values in parentheses
in  Figure  1, the drying  gas and  noncondensable gas recycle streams would
significantly  increase  the load  on  the adsorber currently on-line during
the regeneration cycle.  Further study  is  required to determine  if the
30.48  cm  (1.0  foot)  of GAC in the adsorber can handle these additional
loads  during regeneration without breaking through.  One alternative would
be  to  compress these gases and  store them.  The gases could then be slowly
blended with the air stripper off-gas at an appropriate rate to  prevent
premature  breakthrough.  Another alternative would be to simply  vent the
gases  to  the atmosphere, since  they only account for approximately four
percent of the TCE exiting the  stripper.

    For cost calculations, the  exhausted and/or spent GAC was  landfilled
and replaced with virgin GAC.   For  the cases with on-site regeneration of
the GAC,  steam usage per regeneration was assumed to be 20 kg  steam/kg GAC.
The carbon was assumed to retain 80 percent of its virgin capacity for 20
cycles.  Condensate  from the steam  regeneration system was treated using
self-contained adsorption units.  The units are Department of Trans-
portation approved 55-gallon drums  filled with GAC.

    Aqueous-phase adsorption systems were also designed for the three inlet
TCE concentrations for comparison with the air stripping solvent recovery
systems.  The aqueous-phase adsorption system consists of two GAC vessels
in  series.  Each vessel  is 7 feet in height and 10 feet in diameter.
                                     78

-------
      TABLE 7.  SUMMARY OF DESIGN PARAMETERS FOR GAS PHASE ADSORPTION
                       OF AIR STRIPPING TOWER OFF-GAS
Parameter CASE:
Inlet Water Temperature (°C)
Water Flow rate to Air Stripper
(m^/sec)
Volumetric Air to Water Ratio
Pressure Drop per Unit Length
of Packing (N/m2/m)
Air Stripping Tower Diameter (m)
Air Stripping Tower Length (m)
Blower Brake Power (kW)
Gas Velocity to Bed (m/sec)
Cross-Sectional Area of GAC Bed (m2)
MTZ Length (m)
GAC Bed Length (m)
GAC Bed Volume (m3)
Weight of GAC (kg) (per vessel)
Condensor Area (m2)
GAC Regenerations per Year (360 days)
Mass of Regeneration0 Steam (kg/yr)
Regeneration Period (hours)
Interstitial Steam Velocity (cm/s)
Off-Gas Heating Requirements to
100 ug/1
10
.021908
60
50
1.52
(5.0 ft)
8.26
(27.0 ft)
1.21
.25
5.26
.0437
.3048
1.60
618
9.10
2.0
32,640
10-12
10
2,050,000
300 ug/1
10
.021908
60
50
1.52
(5.0 ft)
10.3
(34.0 ft)
1.44
.25
5.26
.0405
.3048
1.60
618
9.10
3.8
62,016
10-12
10
2,050,000
1000 ug/1
10
.021908
100
50
1.85
(6.0 ft)
10.4
(34.0 ft)
2.55
.25
8.76
.0396
.3048
2.67
1362
15.2
5.7
155,270
10-12
10
3,410,000
Lower Relative Humidity to 40%
(BTU/day)
                                      79

-------
Calgon's F-400 aqueous-phase GAC was used for these designs.   The exhausted
GAC was landfilled and replaced with virgin GAC.

    Table 8 is a summary of the capital  costs for the three design cases
which were estimated through contact with various vendors and by methods
described by Peters and Timmerhaus (25).  The precision with which the
capital costs were determined for the air stripping units is + 10 percent
for all cases.  For the gas phase adsorption units, the precision is
approximately + 20 percent.

    Table 9 displays the detailed operation, maintenance, and GAC costs for
the three design cases.  The unit prices for utilities and GAC were:  (a)
$0.055/kW-hr, (b) $5.0/1,000 Ib of steam, and (c) $2.4/lb of GAC which was
based on a purchase of between 500 and 2,000 Ib lots.  The maintenance
costs for the air stripping units were based on operational data gathered
by Hand et a/. (24).  The maintenance costs for the gas phase GAC adsorp-
tion units were estimated to be five percent of the total equipment costs.
The cost of landfill ing a 55-gallon drum of solvent-laden GAC was estimated
at $300/drum, which includes the cost of the drum and transportation to an
approved site.  The cost of a self-contained adsorption unit for cleaning
up the regeneration effluent was $575/unit.

    Table 10 summarizes the annual capital, operation, and maintenance
costs for the three design cases.  The annual capital costs were based on a
20 year, 20 payment, 10 percent bond interest rate (Capital Recovery Factor
=  .11746).  At present, this is a typical bond rate that a utility  in
Michigan or Wisconsin would obtain.  Air stripping without off-gas  treat-
ment is the cheapest system.  Table 10 shows that an ASSRP with steam rege-
neration is more economically favorable than a system without regeneration
where the GAC is landfilled and replaced with virgin GAC.  Table 10 also
shows air stripping with off-gas treatment and steam regeneration is also
more economical than aqueous-phase adsorption with carbon replacement,
although only marginally so for the 100 ug/1 case.  Since steam costs are
less than two percent of the treatment cost, the high steam to carbon
ratios which are required  (15 to 20 kg steam/kg GAC) for successful rege-
neration has little impact on the total treatment cost.

                                CONCLUSIONS

1.  The Dubinin-Radushkevich equation was shown to predict single solute
    adsorption equilibria  for VOCs found in air stripping off-gas from the
    molar volume vapor pressure, polarizability and the isotherm of a
    reference compound.

2.  Water-vapor adsorption provides little competition to gas-phase YOC
    adsorption onto GAC at relative humidities less than 45 percent.
    According to the Okazaki model, heating the off-gas stream to reduce
    the relative humidity  40 to 50 percent should give the highest  VOC sur-
    face loading.
                                     80

-------
     TABLE 8.   SUMMARY OF CAPITAL COSTS FOR THE THREE DESIGN CASES OF
                  AIR STRIPPING WITH GAS-PHASE ADSORPTION
ITEM                         CASE:   100 ug/1        300 yg/1        1000 yg/1


Total Capital Cost for Air           $ 88,640       $104,620       $119,450
Stripping Unit (Installed)

Total Capital Cost for                144,275        148,275        198,025
Gas-Phase GAC Unit (Installed)

Total Capital Cost of ASSRP           232,915        252,895        317,474
System (Installed)

Total Capital Cost of                 113,760        113,760        113,760
Aqueous-Phase GAC System
(Installed)
                                     81

-------
    TABLE  9.  DETAILED OPERATION, MAINTENANCE, AND  GAC  COSTS  FOR THE
      THREE DESIGN  CASES  FOR AIR STRIPPING WITH GAS-PHASE  ADSORPTION
ITEM CASE:
Blower Power Requirements
($71,000 gal)
Regeneration Steam ($71,000 gal)
Heating Steam ($71,000 gal)
Carbon Cost ($71,000 gal)
(GAC replacement with carbon
regeneration)
Carbon Cost ($71,000 gal)
(GAC replacement with no
carbon regeneration)
Maintenance Costs for Air
Stripping Unit ($71,000 gal)
Maintenance Costs for Carbon
Absorption Units ($71,000 gal)
GAC Costs for Regeneration
Effluent ($71,000 gal)
Landfill Costs for Exhausted
100 ug/1
.32
.20
1.86
.24
4.79
.50
4.01
0.80
0.35
300 ug/1
.38
.38
1.86
.45
9.10
.50
4.12
2.40
1.04
1000 ug/1
.67
.95
3.10
1.14
22.77
.50
5.50
7.99
3.47
GAC ($71,000 gal) (with steam
regeneration)

Landfil Costs for Exhausted         3.00           5.67           14.33
GAC ($71,000 gal) (GAC replace-
ment with no regeneration)

                      Aqueous-Phase Adsorption System

Operation and Maintenance          12.60          15.20           19.00
Costs ($71,000 gal)

Carbon cost ($71,000 gal)           4.80           9.00           17.70
(GAC replacement with no
carbon regeneration)

Landfill Costs for Exhausted        6.30          11.70           23.10
GAC ($71,000 gal) (GAC replace-
ment with no regeneration)
                                    82

-------
  TABLE 10.   SUMMARY OF ANNUAL CAPITAL, OPERATION,  AND MAINTENANCE COSTS
   FOR THE THREE DESIGN CASES OF AIR STRIPPING WITH GAS-PHASE ADSORPTION
ITEM
CASE:   100 yg/1
300 ug/1
1000 yg/1
Total Annual Cost of
Air-Stripping System
without Off-Gas
Greatment (c/1,000 gal)

Total Annual Cost of ASSRP
System with Regeneration
of Carbon Stainless
Steel (
-------
3.  Axial  dispersion,  intraparticle diffusion,  and  film transfer can
    determine the mass transfer zone length.  Dispersion affects the mass
    transfer rate for  shorter bed lengths  at  the  beginning  of bed opera-
    tion.   However, dispersion has little  effect  on the mass  transfer rate
    for longer beds.

4.  The PFHSDM which included IAST to account for competitive interactions
    was shown to predict the breakthrough  of  a  binary  mixture of PCE and
    TCE.  Simplified models such as the user-oriented  and Kirwan solutions
    effectively predict breakthrough curves for single-solute gas-phase
    adsorption.

5.  Gas-phase adsorption usage rates for VOCs were  about one  half of the
    GAC usage rates that were found for aqueous-phase  adsorption.  Since
    gas phase adsorption kinetics are much faster than aqueous-phase
    adsorption kinetics, the required bed  depth and diameter  are much
    smaller for gas-phase beds than for aqueous-phase  adsorption beds.

6.  Higher steam to carbon ratios than those  recommended for  the solvent
    recovery field are required for good removal  for GAC adsorption systems
    treating vapors with VOC concentrations in  the  low ppb  range.  Steam
    usage on the order of 15 to 20 kg steam/kg  GAC  was found  necessary to
    achieve a stable working capacity on the  GAC  for TCE.  However, the
    steam costs are less than 2 percent of the  total treatment cost.

7.  A pilot plant which received a mixture of TCE and  PCE was regenerated
    three times and the TCE capacity decreased  from 80 percent of the
    virgin capacity to 60 percent over the three  cycles. The reduction in
    TCE capacity with  successive adsorption/regeneration cycles was due to
    the buildup of a PCE heel on the GAC,  since PCE was not removed well
    under the conditions used (100°C, 1 atm).  Model calculations
    demonstrated that  this problem may be remedied  by  using saturated steam
    50°C above the boiling point of PCE (170°C) but 50 kg of steam/kg of
    GAC would be required.

8.  Economic analyses  were performed for treating 0.5  MGD of raw water with
    TCE concentrations of 100 yg/1, 300 yg/1, and 1,000 yg/1.  The analys.es
    showed that gas phase adsorption with on-site steam regeneration was
    approximately 20 percent to 30 percent more economical  than just GAC
    replacement with virgin GAC.  It was also found that air stripping with
    gas phase GAC adsorption and on-site steam regeneration was more econo-
    mical than aqueous phase GAC systems, although only marginally  so for
    the case with an inlet concentration of 100 yg/1.

9.  Bench-scale experiments demonstrated that liquid O>2 extraction appears
    to be a technically feasible means of GAC regeneration, although no
    conclusions can be made as to the economic viability of this method of
    regeneration without further study.
                                     84

-------
                              ACKNOWLEDGMENTS
    This research Is based upon work which was supported by the American
Water Works Research Foundation under Contract No. 83-84 and by the Water
and Waste Management Programs at Michigan Technological University.
                         APPENDIX 1.  NOMENCLATURE
ROMAN LETTERS
B             =  microporosity constant (molz/cal2).
BVF           =  bed volumes of feed  (dimensionless).
BV|rmax        =  maximum number of bed volumes that can be treated
                 (dimensionless).
Cj            =  fluid  phase conentration  (ymol/gm).
C0            =  inlet  bulk phase concentration  (M/L3).
dh            =  hydrodynamic diameter; 2edp/(3(l-e),(L).
dp            =  particle  diameter  (L).
De            =  axial  dispersivity based  on adsorber  length and
                 interstitial velocity (L  /t).
Dg            =  solute distribution  parameter (dimensionless);
                 paqe>i(l-e)/eC0ji.
DM            =  free gas  diffusivity (L2/t).
Dp            =  pore diffusivity based on pore  void fraction  (L2/t).
Ds            =  surface diffusivity  (L2/t).
EBCT          =  t/e, VB/Q or L/VS, fluid  residence time in the bed which
                 is devoid of the adsorbent or empty bed contact time  (t),
KI            =  Freundlich isotherm  capacity constant (M/M)(L3/M)1/n.
kfj          =  film transfer coefficient (L/t).
L             =  length of fixed=bed  (L).
M             =  molecular weight.
                                     85

-------
1/rii          =  Freundlich isotherm intensity contstant (dimensionless)
n             =  refractive index (dimensionless).
P             =  partial  pressure of solute in gas  (mm Hg).
Pet           =  Peclet number for hydrodynamic mixing; VT  dn/Dg
                 (dimensionless).
Pez           =  Peclet number based on interstitial  velocity and
                 adsorber length (dimensionless);
Pem           =  Peclet number based on interstitial  velocity, adsorber
                 length and thermolecular diffusion;  LV/Dm.
Ps            =  saturation vapor pressure of solute  at temperture T (mm Hg),
QO            =  Okazaki model total surface loading  (umol/gm).
QOI           =  Okazaki model dry pore surface loading (ymol/gm).

Q§2           =  Okazaki model wet pore condensed phase loading (ymol/gm).
0.03           =  Okazaki model wet pore surface loading (umol/gm).
R             =  gas constant (cal/mol °K).
Re            =  dp Vip/u, Reynolds number (dimensionless).
Sd            =  dry surface area of the adsorbent (L2/M).
Sw            =  wet surface area of the adsorbent (L2/M).
St            =  total surface area of the adsorbent (L2/M).
Sc            =  y/Dm p, Schmidt number (dimensionless).
Sni           =  kf-j2R/D[_, Sherwood number (dimensionless).
sti           =  modified Stanton number (dimensionless); kf }iT(l-
t             =  elapsed time (t).
T             =  reduced time (mass throughput (dimensionless);
                 t/i(Dgt + 1) (used in fixed-bed models).
(V/Q)         =  air stripper volumetric air to water ratio.
                                     86

-------
Vc            =  condensed volume In the pore (L3/M).

Vi            =  interstitial velocity (L/t); vs/e.

Vs            =  superficial velocity (L/t).

Vmi           =  molal volume of component i (L3/M).

W             =  adsorption space occupied by adsorbate (L.3/gm).

W0            =  maximum adsorption space (L3/gm).

GREEK LETTERS

a             =  polarizability.

e             =  fraction of volumetric space in  reactor unoccupied by
                 adsorbent, or void fraction (dimensionless).

6p            =  fraction of volumetric space in  adsorbent phase unoccupied
                 by adsorbent on the pore volume  fraction (dimensionless).

en            =  adsorption potential in D-R equation, RT In PS/P  (cal/mol).

pa            =  adsorbent density which includes pore volume (M/L3).

Pb            =  adsorbent bulk density (M/L3).

PL            =  liquid adsorbate density (M/L3).

V             =  contact angle between water and  the YAt (degrees).

C             =  tortuosity  of the flow path in bed  (1.4) (dimensionless).

Tp            =  tortuosity  of adsorbent  (dimensionless).



                                 REFERENCES

 1.   Cortright, R.D.   Gas phase adsorption of volatile  organic compounds  from
     air stripping  off-gas onto granulated activated  carbon.  Thesis  in partial
     fulfillment  of Master of Science  Degree  in Chemical  Engineering,  Michigan
     Technological  University, Houghton, Michigan, 1986.

 2.   Tang,  G.  Predicting equilibria  for gas-phase adsorption of volatile
     organic compounds from  air  stripping  off-gas  onto  granular  activated
     carbon.  Thesis in partial  fulfillment  of Master of  Science Degree  in
     Civil  Engineering, Michigan  Technological University, Houghton,  Michigan,
     1986.
                                       87

-------
3.  Rick, B.G.   The regeneration of granular activated  carbon  with steam.
    Thesis in partial  fulfillment of Master of Science  Degree  in  Chemical
    Engineering, Michigan Technological  University,  Houghton,  Michigan,
    1986.

4.  Scott Environmental  Technology, Inc.  Field validation of  EPA Reference
    Method 23,  method for determination  of halogenated  organics from sta-
    tionary sources.  EPA Contract 68-01-3405.  Research Triangle Park,  NC,
    U.S. Environmental Protection Agency.

5.  Reucroft, P.J., Simpson, W.H., and Jonas, L.A.   Sorption properties of
    activated carbon.  The Journal of Physical Chemistry.  75:23, 1971.

6.  Rasmuson, A.C.  Adsorption equilibria  on activated  carbon  of mixtures of
    solvent vapours.  In:  A. Meyers and G. Belfort (eds.), Fundamentals of
    Adsorption Processing of the Engineering Foundation Conference, May
    6-11, 1983.  Engineering Foundation, New York,  1984.

7.  Perry, R.H. and Chilton, C.H.  Chemical Engineers Handbook.  6th Ed.
    McGraw-Hill Co. New York, pp. 3-240, 1986.

8.  Crittenden, J.C., Speth, T.F., and Hand, D.W.  Correlation of aqueous
    adsorption  isotherms for hydrophobic compounds  using the polanyi poten-
    tial theory.  Submitted to Environmental Science and Technology.  1986.

9.  Speth, T.F-  Predicting equilibria for single and multicomponent
    aqueous-phase adsorption onto activated carbon.  Thesis in partial
    fulfillment of Master of Science Degree in Civil Engineering, Michigan
    Technological University, Houghton, Michigan, 1986.

10. Okazaki, M., Tamon, H., Toei, R.  Prediction of binary adsorption
    equilibria  of solvent and water vapor  activated carbon.  Journal of
    Chemical Engineering of Japan.  11:3,  1978.

11. Gossett, Camerun, Eckstrom, Goodman and Lincoff.  Mass transfer coef-
    ficient and Henry's constant for packed-tower air stripping of volatile
    organics measurement and correlation.   Final Report, AFESC, Tyndal Air
    Force Base, Panama City, FL, 1985.

12. Freeman, G.B. and Reucroft, P.J.  Adsorption of HCN and H20 vapor mix-
    tures by activated and impregnated carbons.  Carbon.  17:313, 1979.

13. Itaya, A., Kato, N., Yamamoto, J., Okamoto, K.   Liquid phase adsorption
    equilibrium of phenol and its derivatives on macroreticular adsorbents.
    Journal of Chem. Eng. of Japan.  17:4, 1984.

14. Calgon Corporation.  Type BPL granular carbon.   Manufacturers Bulletin.
    Pittsburgh, PA, 1984.                           ~

15. Crittenden, J.C., Hutzler, N.J., Geyer, D.G., Oravitz, J.L., Friedman,
    G. Transport of organic compounds with saturated groundwater flow: model
    development and parameter sensitivity.  Water Resources Research.  22:3.
    1986.                                   	
                                     88

-------
16. Friedman, G.  Mathematical modeling of multicomponent adsorption in
    batch and fixed-bed-reactors.  Thesis in partial  fulfillment of Master
    of Science Degree in Chemical Engineering, Michigan Technological
    University, Houghton, Michigan, University Microfilms, Ann Arbor,
    Michigan, 1984.

17- Fleck, R.D. Jr., Kirwan, D.J., and Hall, K.R.  Mixed-resistance dif-
    fusion kinetics in fixed-beds under constant pattern conditions.
    Industrial and Engineering Chemical Journal.   12, 1973.

18. Hand, D.W., Crittenden, J.C., and Thacker, W.E.  Simplified models for
    design of fixed-bed adsorption systems.  J_^ Env.  Eng.  10440, 1984.

19. Wilke, C.R. and Lee, C.Y.  Estimation of diffusion coefficients for
    gases and vapors.  Ind. Eng. Chem. 47:1253, 1955.

20. Miyauchi, T. and Kikuchi, T.  Axial dispersion in packed beds.  Chem.
    Eng. Sci. 30, 1975.

21. Wakao, N. and Funazukri, T.  Effect of fluid dispersion coefficient on
    particle  to fluid mass  transfer coefficient.  Chem. Eng. Sci.  33, 1973.

22. Williamson, J., Bazaire, K., and Geankopolis, C.   Liquid phase mass
    transfer  at low Reynolds number.   I_^ and EC. Fund.  2, 1963.

23. Dobrzelewski, M. et al.  Determination and prediction of surface dif-
    fusivities  of volatile  organic compounds found in drinking water.   Nat.
    Tech. Info. Svc.. Springfield, Virginia, 1985.

24. Hand, D.W., Crittenden, J.C. and Gehin, J.L.  Design and economic eval-
    uation of a full scale  air  stripping tower for treatment of VOCs from a
    contaminated groundwater.   Journal of  the American Water Works
    Association,   (in press,  1986).

25. Peters,  M.S. and Timmerhaus, K.D.  Plant Design  and Economics for
    Chemical  Engineers.  McGraw-Hill  Inc.,  1980.

26. Kavanaugh,  M.C., and Trussell, R.R.  Design  of aeration towers  to  strip
    volatile contaminants  for drinking water.  J^ AWWA. 72:12,  1980.

27. Singley,  J.E.  et al.   Trace organics removal by  air  stripping.
    Supplementary  Report to AWWA Research  Foundation, Denver, Colorado.
    April 1981.
                                     89

-------
                  GAG AND RO TREATMENT FOR THE REMOVAL OF
                   ORGANIC CONTAMINANTS FROM GROUND WATER

                       by:  Joseph H. Baier
                            Suffolk County Department of Health Services
                            Suffolk County, NY


                               INTRODUCTION

     Under  the Safe Drinking Water Act of 1974, the U.S. Environmental
 Protection Agency is required to establish recommended maximum contaminant
 levels  (RMCLs) for contaminants that may have an adverse health effect on
 those persons consuming water from public systems.  Also required is the
 establishment of a maximum contaminant level  (MCL).  The MCLs are enfor-
 ceable  standards that are required to be set as near as feasible to the
 RMCLs (health goals), taking treatment technologies and cost into
 consideration.

     A proposed rule was presented in the November 13, 1985 Federal  Register
 to  establish RMCLs for several synthetic organic chemicals (SOCs).   Most of
 these organics are primarily found in groundwaters.

     The groundwater of Suffolk County, New York is designated as a  sole
 source aquifer, and in recent years there have been increasing concerns
 about the  contamination of this ground water by agricultural  chemicals
 (fertilizers, insecticides,  herbicides, nematocides, and fungicides).  This
 concern expanded when specific chemicals were identified in homeowners'
 private drinking wells.

     Since  1977, Suffolk County has examined ground water for agricultural
 and  organic contaminants and their decay products.   During this testing,
             rai Or °*9anic  comP°unds were evaluated,  with 41 found in the
                 EV* th"e contaminants were present in trace quantities,
 non      ™ carboluran>  x'2 dichloropropane (DCP)  and 1,2,3 trichloro-
Ai?Pn!%JI  } W6re  fT  aflr1«ltural  compounds found at elevated levels.
from fert?H7.?mSnnvS  ^C6pt TCP  are On the  SOC  regulatory list.   Nitrates
orimarv  dr Iklna  Sli    1°^ Te  SlS° present in  1uant1ties  exceeding the
New Ynri, JiSi"8     ?  sjandards-   Federal  proposed  RMCLs and the present
these ch  '  ?  DePartment  of Health  guidelines are  shown  in Table 1  for
                                    90

-------
              TABLE 1.  PROPOSED RMCLs AND NYSDOH GUIDELINES
                                  Proposed RMCL       NYSDOH Guideline
                                     (yg/i)                (yg/i)
Aldicarb                                9                     7
  (aldicarb + aldicarb sulfone
  + aldicarb sulfoxide)

Carbofuran                             36                    15

1,2-dichloropropane (DCP)               6                    50

1,2,3-trichloropropane (TCP)            -                    50

Nitrate                                10*                   10

*primary drinking water standard  (mg/1)


                              PROJECT PURPOSE

    A cooperative agreement1 was  initiated by the U.S. EPA Drinking Water
Research Division (Cincinnati, Ohio) to examine the effectiveness of cer-
tain water  treatment  systems to remove the agricultural chemicals mentioned
above from  Suffolk County groundwater.  Two parallel treatment systems were
evaluated for a  one-year period:  granular activated carbon (GAC) plus ion .
exchange and reverse  osmosis (RO).

    The main emphasis of this paper will be to present data from one year
of  pilot plant operation evaluating GAC and RO treatment.  The RO portion
of  this presentation  has been submitted to the American Water Works
Association Journal.   In addition, data are presented and discussed on the
use of 2,700 point-of-use/point-of-entry (POU/POE) devices installed in pri-
vate  homes  for aldicarb  removal.

                                BACKGROUND

    Agriculture  has been a major  industry in Suffolk County for  over 200
years.  Fertilization practices,  with as much as 250  Ib-N/acre applied,  led
to  widespread nitrate contamination of the shallow aquifer(l).   The potato
plant (a principal crop) is  susceptible to a number of pests, most notably
the golden  nematode,  which attacks the roots, and the Colorado potato
beetle, which eats the leaves.  Since the early  1950s, pesticides  con-
taining 1,2 dichloropropane  have  been applied to fields  infested with
golden nematodes, particularly those fields quarantined  by the U.S. Depart-
ment  of Agriculture.   In 1974, the carbamate pesticide aldicarb  (trademark
 Iproject Officer:   Benjamin  Lykins,  Jr.
                                     91

-------
TEMIK, Union Carbide Corp.)  was registered for use on potatoes,  and by 1976
the chemical was being used  by all  growers at an application rate of 3
pounds of active aldicarb per acre(2).

    Aldicarb was used extensively for four growing seasons in Suffolk
County.  Its use was discontinued when  the manufacturer Union Carbide (UC)
first discovered the chemical in Long Island ground water.  UC requested and
received approval to modify  their labeling permit to prohibit sale in
Suffolk County.

    The New York State guideline and the proposed EPA RMCL for aldicarb are
based on the sum of parent aldicarb plus the two metabolites:  aldicarb
sulfone and aldicarb sulfoxide.  The parent compound has not been detected
in Suffolk ground water and the metabolite occurrence is as follows:

total aldicarb = parent aldicarb + aldicarb sulfone + aldicarb sulfoxide
     (100%)            (0%)             (40-60%)            (40-60%)

Some  representative data from community private wells are shown in Table 2
to illustrate the contamination.  Further discussion of the use of POU/POE
to address  this problem will appear later.


  TABLE 2.   REPRESENTATIVE ALDICARB RESULTS - SUFFOLK COUNTY GROUNDWATER(3)

Community
1
2
3
4
5
Wells
Sampled
222
434
2,161
1,832
3,160

>7 ppb
2
43
351
270
359

1-7 ppb
18
46
345
256
374
% Below
Detection
91
79.5
68.8
71.3
76.8
     1,2  Dichloropropane  (DCP) testing only began in 1980, and only a few
 agricultural  communities have been found to be contaminated.  It 1s
 suspected  that  the primary sources of this chemical are several pesticides
 (DD,  Vidden D,  Vorlex Telone)~fumigants used for golden nematode control-
 each  containing DCP.  The chemical is no longer used by the Department of
 Agriculture as  a  Long Island fumigant.

     In one community, DCP was found  in 17 of 33 wells, with two wells
 approaching or  exceeding the State Health guideline of 50 ppb.  A second
 community  had 2 of 9 samples contaminated at levels of 10-15 ppb, and a
 third area had  a  private well with a concentration of 49 ppb(3).
                                     92

-------
    Carbofuran was available for agricultural use before and after aldi-
carb.  The amounts found in groundwater have been much less in number of
wells and concentration.  As an example, county records show only 1.8 per-
cent of 2,000 wells sampled in 1985 exceeded state guidelines, compared to
11.7 percent exceeded for aldicarb.

                                PILOT PLANT

    The initial phase of the EPA cooperative agreement called for the
construction of a pilot plant  (see Figure 1).  A 30 gpm well was used to
feed both systems in parallel, and a 5-micron cartridge filter protected
both the membrane and resin from deposition.  The 5 gpm GAC system con-
sisted of 3 carbon units in series;  each contained 3 ft3 of Filtrasorb 300
(Calgon).  The units could be  operated in any series order, and the empty
bed contact time  (EBCT) was 5  minutes for each unit, with a maximum EBCT of
15 minutes.  During start-up operation, it was generally observed that one
unit (5 minutes)  could  remove  all the contaminants for a short period of
time (one month)  before breakthrough.  Bed exhaustion signalled a change in
carbon and only two units  (10  minutes was EBCT) actually needed to achieve
removal (see discussion below).

    The reverse osmosis unit  (RO) used a hollow-fiber polyamide membrane.
Approximately  67  percent recovery was observed (8 gpm influent, 5.3
effluent and 2.7  concentrate)  using 400 psi  feed pressure and a water tem-
perature of 55°F-  The  unit consisted of three membrane cells piped to give
parallel flow  to  Cells  1 and  2, with the concentrate from each passing
through Cell 3 before disposal.  The unit operated virtually unattended
except for monthly membrane cleaning.

    The raw water quality  changed during the year of operation.  Some typi-
cal data is shown in Table 3  and Figures 2,  3, and 4.  Although the screen
depth was chosen  to provide a  blend of all contaminants, the individual
values in Table 3 are typical  of agricultural ground water  quality in Suffolk
County.  Consistent decline in aldicarb, carbofuran, and to some extent
DCP, reflects  the actual cleaning up of the  ground water.   Nitrate is also
shown as a control  to indicate the  consistency and wide-spread ground water
contamination  resulting from  decades of nitrate use.
                                      93

-------
  WELL
                                GAG
                                         GAG
                                                  GAG
IU
o



5 65
                                                 PERMEATE

s
REGENERANT
STORAGE
TANK


                                                REJECT WATER
                                                                  DISCHARGE
      Figure 1.   Flow schematic for pilot plant  -  Suffolk County,  NY.
01
     50
    40
    30-
o
0   20
    10
\^x_x
3 100 200
TIME (DAYS)
300
400
                    Figure 2.   Aldicarb sulfone  -  raw.
                                     94

-------
    40 T
&   30
3.
O

I—I



$
     20-
o
o
     15-
    10
                       100
    200

 TIME (DAYS)
300
400
                    Figure 3.  Aldicarb sulfoxide -  raw.
    ,25-
 S   20-
o
I—I


<



I   151
O

O
     10-
                       100
    200

TIME  (DAYS)
300
400
                  Figure 4.   1,2, dichloropropane - raw.
                                      95

-------
                     TABLE 3.   RAW WATER  QUALITY  DATA*
Date
6/85
7/85
8/85
9/85
10/85
11/85
12/85
1/86
2/86
3/86
4/86
5/86
Aldicarb
Sulfone
48
31
27
23
22
20
19
17
17
16
15
14
Aldicarb
Sulfoxide
35
24
20
17
17
16
15
14
13
13
13
12
Dichloro-
Propane
22
23
22
22
11
18
19
20
21
20
20
18
Carbofuran
13
10
8
8
8
6
6
6
6
5
5
5
Nitrate
__
11
12
12
13
11
11
11
11
11
11
11
                       REVERSE OSMOSIS - DISCUSSION

    During the 12-month period reported in Table 4, 3.9 million gallons of
water were treated and 2.6 million gallons of drinking water were produced.
A review of the data in Table 4 shows a steadily declining raw water con-
centration for aldicarb (both metabolites) and carbofuran, with a similar
reduction in concentrate.   Table 4 presents arithmetic averages only, and
mass balance calculations  should not be attempted.   During bench-scale
testing(4), with aldicarb  sulfone at 47 ppb, some leakage of 2-3 ppb in the
permeate was noted.  At the start of the pilot plant,  similar leakage was
observed at lower concentrations (31, 27 and 22 ppb).   It was not until the
aldicarb concentrations dropped to almost one-half  of  the original  con-
centration (20 to 40 ppb)  that complete removal  was noted.  This suggests
that higher values of aldicarb (>50 ppb) may not reject as completely and
should be evaluated further(4).
                                    96

-------
                  TABLE 4.   SUMMARY  OF  PILOT  PLANT  DATA*
Aldicarb
Sulfone
Month
July
Aug.
Sept.
Oct.
Nov.
Dec.
Jan.
Feb.
Mar.
April
May
June
R
31.4
27.2
22.8
21.6
20.3
19.0
17.4
16.5
16.3
14.8
14.7
14.0
P
1.3
1.2
1.0
1.0
<1.0
<1.0
1.5
<1.0
<1.0
<1.0
<1.0
<1.0
C
79
79.7
68.4
54.6
45.8
42.3
42.2
42.5
43.1
42.8
42.0
39.6
Aldicarb
Sulfoxide
R
__
20.
17.
17.
16.
14.
14.
13.
13.
13.
11.
11.
P
._
0 <1
2 <1
4 <1
2 <1
7 <1
0 <1
4 <1
0 <1
0 <1
6 <1
2 <1
C
__
58.2
51.4
42.4
36
32.7
34.9
33.8
33.2
34.3
32.8
30.5
1,2
Dichloropropane
R P C

22.2
22.2
20.6
18.2
19.0
20.2
21.0
19.5
20.0
17.5
18.8

6.5
6.4
6.8
6.4
5.8
8.5
6.5
6.3
7.5
8.3
6.9

43.0
41.0
49.6
44.0
35.7
42.7
45.3
31.3
46.3
47.8
36.7
Carbofuran
R P
9.8 <1
8.0 <1
8.1 <1
7.8 <1
6.2 <1
5.8 <1
5.9 <1
5.7 <1
5.3 <1
4.9 <1
4.7 <1
4.3 <1
C
24.5
21.5
22.5
19.5
13.0
13.8
14.0
13.4
12.8
12.6
13.3
12.7
*Monthly averages (all values in ppb).

R = raw
P = permeate
C = concentrate

    Carbofuran data show a 50 percent reduction in raw water and a similar
reduction in concentrate.  Previous work(4) noted wide variation of com-
pound mass balances possibly caused by several  factors:  the sensitivity of
the mass balance calculations; the possibility of non-uniform rejection; or
adsorption of compound on the membrane.  If the latter is occurring, the
adsorption is a consistent percentage, regardless of influent quality.

    The DCP concentrations were generally consistent throughout the study
period.  Removal efficiency varied from 58 to 72 percent, which was similar
to the removal percentage observed in the low-pressure bench-scale work(4).

                  GRANULAR ACTIVATED CARBON - PERFORMANCE

    This discussion will cover one year of operation for the GAC units.
Three adsorbers, in series, with 5 minutes EBCT (15 minutes total) each
operated to remove aldicarb, carbofuran, DCP, and TCP.  Figures 5, 6, and 7
plot the effluent values of DCP, aldicarb sulfone, and aldicarb sulfoxide
together with the raw water quality.  Carbofuran did not break through any
of the adsorbers during the testing period.  TCP is not discussed at this
time since there is no EPA regulation.
                                    97

-------
               7/1/85   8/1
9/1
10/1     11/1     12/1     1/1/86    2/1      3/1      4/1       5/1
UD

00
      B
      O)
      z  15 H
      O

      5
      z
      IU
      O


      8  10
/ / BW ' UNIT#4 	 / / #1
/-' /' / / #2
J #3
/ /( / #4
#5
/ / /i '
ii / i i
'i > \ I
,' / / _/ 	 /
6/12/85
7/12/85
8/20/85
11/5/85
1/27/86



                   30      60
    90
    120     150      180      210     240      270      300     330      360
                                     Figure 5.   5 GAC -  performance 1,2 dichloropropane.

-------
        7/1/85
  50-
  40-
O
o
O
  10
                                      BW
                              '    \
         UNIT #1
                           UNIT#2
     UNIT

      #1
      #2
      #3
      #4
      #5
                                10/1    11/1     12/1    1/1/86   2/1
3/1
        START

       6/12/85
       7/12/85
       8/20/85
       11/5/85
       1/27/86
4/1     5/1
                                                                                              0/1
            30      60       90      120      150     180     210     240     270     300      330     360
                             Figure 6.   6  GAC - performance aldicarb sulfone.

-------
                7/1/85
o
o
                                                                                     UNIT       START
                                                                                               6/12/85
                                                                                               7/12/85
                                                                                               8/20/85
                                                                                               11/5/85
                                                                                               1/27/86


                                                                                                 •/i
           o-i-
                   30     60      90     120
150    180    210     240     270     300     330     3*0
                                 Figure  7.   7 GAC - performance  aldicarb sulfoxide.

-------
    When examining Figures 5, 6, and 7, the adsorption isotherm or wave
front seems to be consistent for each contaminant;  i.e.,  after initial
breakthrough, the lead adsorber effluent concentrations continue to
increase to 50 to 75 percent of influent and then an improvement in
effluent occurs.  The improvement is temporary since the  column then pro-
ceeds to total exhaustion.

    Correlations were performed on the curves developed for each carbon
unit; i.e., Unit #1 vs. #2; Unit #1 vs. #3; Unit #1 vs. #4, etc.  Six
correlations were performed for the three parameters.   The lowest correla-
tions were:  .82 (Unit #1 vs. #3 for aldicarb sulfone); .83 (Unit #1 vs.  #3
for aldicarb sulfoxide); and .65 (Unit #1 vs. #3 for DCP), with the next
lowest 0.82.  These strong correlations are indicative of consistent per-
formance by all the adsorbers.

    The rate of carbon use is principally a function of the contaminant
type, carbon type, degree of contaminant removal required, and EBCT.  The
breakthrough curves provide a convenient way to calculate carbon usage
rates as a function of each of the above variables(5).

    The carbon usage at various EBCTs and breakthrough criteria are sum-
marized below:

                       TABLE 5.  CARBON USE RATES(5)
                                                Usage Rate (lb/1,000 gal)

       Compound      Avg.  InfluentEffluent
Concentration (ug/L)              EBCT  (min)
                               5     10      15
     aldicarb              20          3           0.34    0.2   0.18
     sulfoxide                         9           0.2     0.13  0.12

     aldicarb              25          3           0.38    0.22  0.2
     sulfone                           9           0.24    0.18  0.14

     1,2 dichloro-         22          6           0.29    0.24  0.23
     propane
     This  information  enables  two design parameters—optimum EBCT and carbon
 replacement  rate—to  be  estimated.  From Table 5 it can be seen that for
 all  three contaminants,  the carbon  usage rate decreases with increasing
 EBCT.   However,  beyond an  EBCT of 10 minutes, the decrease is not signifi-
 cant;  therefore,  increasing the contact time beyond 10 minutes does not
 provide any  additional carbon utilization.  Figure 8 shows this much more
 vividly,  and although the  DCP curve is somewhat flatter, it is still
 obvious that an  EBCT  of  10 minutes  is still optimum.
                                    101

-------
0.5
                                   SULFOXIDE (3ug/L)
                                             SULFONE (3ug/U
                                   10
                              EBCT (mSn)
   NOTE: (  ) INDICATES  EFFLUENT
              CONCENTRATION
              Figure 8.  Carbon usage vs. EBCT  (5),
                               102

-------
    Empty bed contact time (EBCT) provides an indication of the quantity of
carbon that will be on-line at any one time, and can impact the cost of
any full-scale system.  The impact of increasing EBCTs on the overall per-
formance of the carbon system can be studied by comparing the relative
treatment characteristics of each unit.  Summarized below (Table 6) is the
service time for each EBCT at different breakthrough criteria(5).

                   TABLE 6.  SERVICE TIME FOR EACH EBCT
                              Effluent          Service Time (days)
                            Concentration               EBCT
          Compound               (yg/L)       5 min     10 min     15 min


     aldicarb sulfoxide           3           37        126         206
                                  9           60        186        >250

     aldicarb sulfone             3           33        110         186
                                  9           52        146        >220

     1,2 dichloropropane          6           42        106         161

 Note:">"  indicates service  time until unit shutdown.

    As  indicated,  increasing  the EBCT  from 5 to 10 minutes almost triples
 the service  time.   Increasing  the EBCT from 5 to  15 minutes only increases
 the service  time by a  factor  of  4 to 5.  This further agrees with the 10
 minute  optimum  EBCT observed  from the  carbon usage rate.

    Using the volume of water  treated  by each unit, it is possible to
 calculate the amount of contaminant adsorbed on the carbon.  Table 7 below
 presents the removal data  for  the four carbon beds tested from startup to
 breakthrough.   Table 8 presents  the total amount  of contaminant removed
 from startup through exhaustion. Table 7 also shows the percentage of the
 total usage  that the breakthrough loading represents.  The percentages do
 not show any consistency,  either with  compounds or a carbon unit.  Some
 consistency  appears in the exhaustion  levels of aldicarb sulfoxide and DCP,
 but not aldicarb sulfone.

                TABLE 7.  CONTAMINANT REMOVAL @ BREAKTHROUGH*
                          Unit 1         Unit  2        Unit  3         Unit  4
aldicarb sulfone
aldicarb sulfoxide
1,2 dichloropropane
0.06 (49)
0.05 (56)
0.036 (48)
0.045 (35)
0.048 (54)
0.056 (73)
0.029 (58)
0.019 (30)
0.044 (63)
0.009 (23)
0.008 (20)
0.02 (33)
 "All  values  in Ibs;  value in  (  )  is  % of total  removal  (Table 8).
                                     103

-------
                TABLE  8.   CONTAMINANT  REMOVAL  @  EXHAUSTION*
                           Unit 1       Unit  2        Unit  3        Unit 4
aldicarb sulfone
aldicarb sulfoxide
1,2 dichloropropane
0.122
0.088
0.074
0.128
0.089
0.076
0.05
0.063
0.07
0.04
0.04
0.06
*A11  values in Ibs.

    During the adsorption process,  a phenomenon often encountered is the
chromatographic effect, or the elution of adsorbed compounds.   This effect
would be noted by effluent contamination concentrations exceeding the
influent at some time during the units'  operation.  Usually the con-
taminants that are more poorly adsorbed would show this for some period of
time until a new equilibrium is reached in the carbon bed.   The sampling
techniques established for this study and the 3-unit series flow offer a
unique opportunity to observe this  effect.  A review of the raw and treated
data for each GAC unit did not show evidence of the chromatographic effect,
nor was there any significant adsorption affinity of one compound over
another (Figures 5,  6, and 7).

    Some other interesting observations on carbon performance  and organics
removal were made from TOC analyses done by EPA's Drinking  Water Research
Division (DWRD) in Cincinnati.  Results are shown in Table  9,  together with
the mean, standard deviation, and median values.  Sample results for raw
water and GAC effluent are presented.  The GAC results are  presented in
numerical sequencing of carbon column operations; i.e., GAC #2,3,4 sequence
was used until #2 was totally exhausted.  GAC #2 carbon was replaced, it
became GAC #5, and the flow sequence became GAC #3,4,5.  This  practice con-
tinued throughout the testing.  The designations A and B represent split
samples taken at the start of the program for quality control.

Four observations follow:

    1)  A consistent TOC (mean value) was found in all adsorber effluents,
even when the effluent samples for  organics and pesticides  were less than
detectable.  For example, GAC unit  #7 was placed into service  on 5/19/86, 2
days before the first TOC sample.  Laboratory results on 5/21/86 from unit
#7 showed less than 2 ppb for DCP,  TCP, aldicarb sulfoxide, aldicarb
sulfone, and carbofuran, yet a TOC  of 0.89 ppm was present.  This suggests
that a background TOC exists for the units, amounting to 50-60 percent of
the raw water TOC.  This background TOC was not identified  in  a priority
pollutant scan performed in raw and treated samples (EPA laboratory per-
formed the analyses).

    2)  The mean TOC value for each adsorber over the testing  period had a
limited variation (0.71 to 0.89 ppm).  Since all the carbon used in this
project came from the same batch, it appears that the raw carbon was uni-
form in its performance and did release some organic compounds.
                                    104

-------
                TABLE  9.   TOC  SAMPLING  RESULTS  (ppm)
Date
1/13/86(A)
(B)
l/26/86(A)
(B)
2/10/86
3/10/86
4/21/86
4/28/86
5/7/86
5/12/86
5/21/86
5/28/86
6/2/86
6/11/86
6/16/86
6/23/86
7/9/86
MEAN
STANDARD
DEVIATION
Raw
1.58
1.41
1.34
1.45
1.68
1.28
1.19
1.65
1.40
2.01
1.13
1.30
1.54
1.92
1.49
0.26
GAC #2 GAC#3 GAC#4
1.02 0.67 0.96
0.79 0.64 0.76
0.84 0.77
0.85 0.65
0.89 0.73
1.05
0.92
0.96
1.15
1.08


0.82 0.89
0.15 0.17
GAC#5

1.23
0.79
0.65
0.72
0.82
0.83
0.89
0.93
1.68
0.65
0.78
0.77
0.84
0.89
0.28
GAC#6



0.68
0.71
0.80
0.69
0.71
1.50
0.44
0.68
0.61
1.27
0.81
0.32
GAC#7




0.89
1.18
0.31
0.73
0.61
0.56
0.67
0.71
0.27
MEDIAN        1.43             0.85    0.92     0.82    0.70    0.67
                                105

-------
    3)  TOC values showed a tendency to  generally decrease with the series
flow of the carbon units, with little reduction  by the  third series column.
The TOC reduction is greatest in  the first column in  the series.

    4)  The effluent from the lead column  never  approached the raw TOC.

                      GAC IN POU/POE - SUFFOLK COUNTY

    The magnitude of groundwater  contamination caused by aldicarb is dif-
ficult to comprehend without appreciating  the groundwater hydrology.  This
chemical was used extensively for four growing seasons  in the study area.
Its use was discontinued by Union Carbide  by requesting a modification of
the labeling permit for New York  State to  prohibit sale in Suffolk County.
This action was taken as a direct result of the  identified aldicarb resi-
dues in homeowners' private wells(2). As  contamination was discovered,  a
direct relationship between the proximity  of a well  to  an agricultural
field and the presence of aldicarb developed. Even though the sale and use
stopped, aldicarb kept advancing  as the  groundwater moved and continued to
contaminate wells further downstream.

    Some 2,700 activated carbon adsorption units have been installed by
Union Carbide (aldicarb manufacturer) since 1979.  Over 22,000 samples have
been analyzed, and finally in 1985, some of the  initial wells closest to
the farm fields that were originally contaminated, were beginning to clear.
However, new wells located further downgradient  are now showing the pre-
sence of aldicarb.  Present estimates indicate that the number of filter
installations have peaked at 2,700, but  complete remediation of the problem
may still be decades away(6).

    Before embarking on  a full-scale program of  installing filters where
aldicarb exceeded guidelines, UC and SCDHS conducted trial laboratory and
field  investigations to  determine the effectiveness and estimate the life
of the carbon at varying concentrations^).  A theoretical filter life was
obtained from laboratory tests which used varying concentrations of aldi-
carb  on separate carbon  beds and monitored the effluent until 7 ppb was
reached.  This filter  life,  in gallons or throughput, was reduced by a
safety factor after 5  units were field tested.  Two nomographs were pro-
duced  (one for POE, one  for  POU)  which allowed estimation of unit life.

    Series AF-10 filters manufactured by the Bruner Corporation of
Milwaukee, Wisconsin were chosen for use.   The filter tank was 10 inches  in
diameter by 40 inches  high and contained approximately 1 cubic foot of car-
bon, weighing 27 pounds.   Type GW12x40  carbon manufactured by the Calgon
Corporation was used for the filter media and the unit had automatic back-
washing capability(7).   Installation was either  for whole house (POE) or
single kitchen faucet  (POU)  use,  depending on customer preference.

    To monitor the GAC performance, 25 units were selected for bimonthly
testing.  The homes varied in family size, whole house (POE) versus single
tap (POU) installation,  and  aldicarb concentration.  Each had a water meter
installed with the filter so consumption could be observed along with
                                    106

-------
water quality (raw and treated).  Results of this effort, which includes a
detailed discussion of filter problems, are reported elsewhere(7).  Con-
cerns such as competitive displacement; microbial activity on adsorbents;
varying influent concentrations; equilibrium adsorption factors; and opera-
tional difficulties are also discussed.

    Typical filter performance data are shown in Table 10.  When comparing
actual to theoretical life, a range of 37 to 158 percent appears.  Looking
at Oamesport #2 filter provides insight into the sensitivity of the theore-
tical and actual life at different aldicarb concentrations.  While the
actual filter life shows a  13,500 gallon increase, due to lower aldicarb
levels, the theoretical life anticipates a 40,000 gallon difference for the
two aldicarb values.  Yet,  the first  recharge performed better than the
second (133 percent vs. 66  percent),  although treating less water.  This
inconsistency between actual and theoretical occurred elsewhere.  Review of
raw water values showed almost a four-fold range of aldicarb, which
explains some of the  inconsistencies.  This factor, combined with the pre-
sence of other  competitive  contaminants  (vydate, dinoseb, carbofuran,
dacthal, DCP) caused  UC to  lower the  theoretical filter  life by 25 percent.

    The operation  of  2,700  filters did not proceed without having some
operational difficulties,  including  the  following:

    o Failure  to  have  treatment unit in automatic backwash mode  or not
       having the  unit  connected to  electrical  outlet.
    o Raw  untreated  water  bypassing  filter  resulting from piping arrange-
       ments which cannot  be  easily  determined;  i.e., buried pipe in
       concrete slabs or  concealed piping  in walls.
    o Failure  of  homeowners  to  place filter back  into treatment  mode after
       manually bypassing  system for lawn watering, etc.
    o  Inadequate  backwashing  cycle  resulting  in plugging or reduction of
       water  pressure through  the  filter caused by accumulation  of sediment
        (specific to Long  Island  because  of  iron and manganese).
    o Mechanical  failure  of  some  components of the treatment  unit and
       plumbing accessories caused by sediment  blockage  and/or corrosion of
        treatment unit materials(7).

    The above  information  points out the need  for adequate treatment unit
 design; an  active  maintenance  program to regularly inspect each  unit; and
monitoring  to verify  removal  effectiveness.

    The previous discussion was  not  presented  as a critique of POU/POE,
 since the  program  should  be considered  successful.  The  residents who
 received  filters are  satisfied.  The  success  is due to the cooperation and
 efforts of  UC and  the Suffolk  County personnel.

    POU/POE is  not a  casual solution  to  a water quality  problem.  A person
 requiring  a home treatment  device  should not think that  once installed,  the
 problem  is  over*   The unit  should  be  selected  with care,  and a continuous
 testing and maintenance program  must be  performed,
                                      107

-------
                                     TABLE 10.  SUMMARY OF FILTER PERFORMANCE
o
oo
Average Theoretical
Aldicarb Effective
Concentration Filter Life
Site (ppb) (gallons)
Aquebogue #1 21

Bridgehampton #1 105
Calverton #1 53
Jamesport #2 262
(first recharge)
Jamesport #2 122
(second recharge)
Mattituck #1 105
Orient #1 151







Orient #2 64
Water Mill #1 36
78,000

58,750
71,400
19,500

59,500

58,750
47,500







68,800
74,800
Actual
Filter
Life
(gallons)
28,000

53,500
112,500
26,000

39,500

82,700
21,000







74,000
51,500
Percentage
of Actual to
Theoretical
Filter
Life Remarks
37%

91%
158%
133%

66%

140%
44%







106%
69%
Unit never
backwashed.







Data represents
first recharge.
Possible plumb-
ing problem.
Treatment unit
replaced once
and recharged
three times.



-------
                                  SUMMARY

    A pilot plant was operated for one year to compare GAC and RO for remo-
val of organic pesticides.  GAC performance data enabled upgrading the
plant to 200 gpm and showed the value of in-series carbon units.  Effluent
monitoring of each column must be included in full-scale operation and
maintenance; however, repeated column operation began to show consistent
exhaustions (gallons treated) which should allow a reduction in monitoring.

    The hollow-fiber polyamide membrane was not successful for complete
removal of DCP or high values (>50 ppb) of aldicarb, but demonstrated a
range with which the unit can be operated.  RO should be considered as a
competitive alternative for organics removal when pilot plant studies are
undertaken.

    The use of POU/POE for individual aldicarb problems was successful and
proved to be a viable solution to providing potable water to consumers
where no community water supply exists.
                                REFERENCES

1.  Baier, J. and Rykbsot, K.  The contribution of fertilizers to ground-
    water of Long Island.  Journal of NWWA.  November-December, 1976.

2.  Guerrera, A.A..  Chemical contamination of aquifers on Long Island, New
    York.  Journal AWWA.  73:4, April 1981.

3.  Baier, J. and Robbins, S.  Groundwater contamination from agricultural
    chemicals, North Fork, Suffolk County.  In: Proceedings, ASCE National
    Conference on Environmental Engineering, 1983.

4.  Baier, J., Lykins, B. et al.  Removal of agricultural chemicals from
    groundwater by reverse osmosis.  Submitted for publication to AWWA
    (copies available from authors), 1987-

5.  Malcolm Pirnie et al.  Phase  II design report, EPA cooperative
    agreement CR-811109-02, 1987.

6.  Baier, J.  Long Island's home water  treatment district experience.
    Fourth Water Quality Symposium, Chicago, Illinois, 1985.

7.  Moran.  Report on granular activated carbon treatment units used for
    removal of aldicarb residues  in private wells of Suffolk County, 1983.
                                    109

-------
             POINT-OF-ENTRY/POINT-OF-USE TREATMENT  FOR  REMOVAL
                    OF  CONTAMINANTS  FROM DRINKING WATER

                 by:  K.E.  Longley,  and G.P. Hanna, Jr.
                     Civil  Engineering Department
                     California  State University - Fresno
                     Fresno, CA   93740

                     B.H.  Cump
                     Chemistry Department
                     California  State University - Fresno
                     Fresno, CA   93740


                               BACKGROUND

    An objective of this study  is to develop verifiable and  cost  effective
criteria for designing  granular activated carbon  (GAC)  systems  to be used
for the removal  of DBCP and other pesticides from water supplies.  The
design criteria  would be based  upon  data collected  during the following tasks:

    1. The collection of data  from pilot GAC mini-columns.   This  work will
       be reported at a later  date.

    2. The evaluation of a representative sampling  of existing, installed
       point-of-entry (POE) GAC  systems  to  determine  their  removal  effi-
       ciencies  over time for  DBCP and  other pesticides.

    Another important objective  of this  study  is  the  determination of
measures to improve and strengthen existing administrative  guidelines and
jurisdictional  responsibilities,  pertaining to both community water systems
and private wells containing DBCP and other pesticides.

    The evaluation of existing  responsibilities relative to  an ideal juris-
dictional setting will  provide  recommendations for  meaningful changes in
existing institutional  systems.   Further, the  identification of institu-
tional factors,  which either constrain  or motivate  the  orderly and effec-
tive removal of  toxic substances  from ground water, will provide  invaluable
and necessary information concerning those  factors  that must be considered
by individuals  responsible for  the administration of  control programs for
toxics in ground water.

    Ground water is an  important  source  of  water  in California, par-
ticularly for small public water  systems and private  residences having
individual  wells.  A recent report (1)  states  that  approximately  40 percent
of the state's population uses ground water as the  source of its  domestic
supply, and 93 percent  of the  small  water systems use ground water to
supply approximately five percent of the state's population.

                                     no

-------
    Reacting to evidence of possible widespread organic pollution of
California's ground water, the California legislature responded in 1983
with a bill, Assembly Bill 1803, that required the implementation of a
program for detecting and monitoring organic chemical contaminants in
public drinking water supplies.  The data collected (1) as a result of this
program are from larger utilities, which are defined as those serving over
200 connections.  As shown on Figure 1, 24 of California's 54 counties had
one or more wells found to be contaminated with one or more of the organics
reported as part of the study.  A total of 2,944 wells were reported as
being sampled, with one or more organic contaminants found in 538 (18.3
percent) of the wells.  The 538 wells were distributed among 184 of the 807
water systems in the study, and the sample results from 165 of the wells
exceeded an action level.  The four most commonly detected organics were
PCE, TCE, DBCP, and chloroform.

    Treatment for surface waters is generally provided at a central water
treatment facility.  However, for contaminated ground waters and for some
treated surface waters, a treatment alternative that has been proposed is
the use of point-of-entry (POE) or point-of-use (POU) treatment devices.
POE devices treat all water in a water line entering a structure such as a
domestic residence, and POU devices treat only the water for one tap.  The
former is generally installed outside of the structure and the latter is
generally installed under the sink or on the end of the tap.

    POE/POU devices have been used for years by many consumers for water
softening.  However, the use of water softeners generally has been optional
for the consumer willing to accept the increased cost and questions con-
cerning system  reliability.  While the economics and reliability provided
by a central treatment system are generally superior to that provided by
POE/POU devices, they do offer a valid treatment alternative for individ-
uals having a water well without access to a water treatment system.
POE/POU devices may also provide a treatment alternative for water treat-
ment systems having one or more contaminated water wells, when the water
transmission system is such that water could not be economically treated at
a central facility.

    With the consideration of POE/POU devices as a water treatment alter-
native, numerous institutional, jurisdictional, and technical questions
must be resolved.  These questions include determining what agency has
responsibility  for validating the effectiveness of the POE/POU devices;
identifying what agency has responsibility for monitoring the installation
and use of  the  POE/POU devices; determining what institutional arrangement
is desirable for ownership and operation of the POE/POU devices; and  iden-
tifying who has  responsibility for consumer related  issues  including  adver-
tising practices.

    An operational and maintenance problem often not considered  for  POE/POU
GAC units is the ultimate disposal of  the spent carbon cartridges.   This
operational problem must be addressed  so that means  exist for the  proper
handling and disposal of spent carbon  in accordance with applicable  hazard-
ous waste regulations, particularly when the homeowner is disposing  of  the
spent carbon.   The temptation faced by the homeowner is simply to  discard
the spent material onto the ground or  into the nearest container.


                                    Ill

-------
     so
                          Jtesulta for Cotmttse
$
     40-
     30 -
     20 -
      10 -
           D    D
                           ZOO                 400
                            No. of Walla Tested
                                                                 600
     Figure  1.   California's AB  - 1803 monitoring  program.
                                 112

-------
                              DESIGN CRITERIA

    Gaston (2) has set forth basic engineering concepts, design con-
siderations, and problem areas that should be considered for GAC POE/POU
devices.  Important points he considers concerning the flow rate include
its magnitude, variability, and interruption (on/off operation).  POE/POU
devices may be idle for extended periods of time (typically hours or even
days) and then subjected to high hydraulic loading.  High concentrations of
bacteria might establish themselves on the GAC and then be washed from the
GAC into the treated water delivered at the water tap when moderate to
heavy use follows periods of idleness.  Evaluation of unit design must also
consider this operation characteristic in view of the chemical matrix to be
applied to the GAC contractor.  Substantial competition for adsorption
sites by the contaminants of concern and other organics in the source
water, high hydraulic loading, moderate to poor adsorption to the GAC by
the contaminant(s) of interest, and significant exhaustion of the GAC bed
all contribute to penetration of the mass transfer zone into the carbon bed
and early breakthrough of detectable amounts of contaminants from the bed.
Bed volume and depth must be sufficient to contain the mass transfer zone
for the expected design life of the GAC unit.  This is a function of the
adsorptive characteristics and rate of application of the chemical matrix
to be applied to the GAC unit.

    The hydraulic loading rate must be constrained to a level allowing suf-
ficient time for adsorption of contaminants by the GAC.  This can be
accomplished by equipping each GAC unit with a flow constrictor.  Bacterial
growth on the GAC or the water quality of the source water may contribute
to rapid clogging of the GAC unit.  This requires consideration of the need
for pretreatment of the source water before its application to the GAC
unit, and the need for disinfecting the treated water.  Without a con-
tinuous monitoring program, exhaustion of a GAC bed cannot be expeditiously
determined.  Consequently,  isotherm data and pilot testing data are needed
that are representative of  the chemical matrix and the GAC specific to the
application of each POE unit.  With this knowledge the theoretical bed life
can be determined, a suitable safety factor applied to reduce the expected
life of the GAC unit  (in terms of  the volume of source water applied to the
unit), and an automatic cutoff valve installed to  inactivate the unit when
ths predetermined volume of water  has been applied to  it.

    A typical POE unit employing GAC technology for the  removal of organic
contaminants may contain a  carbon  volume of 7.5 to 8.0 cubic feet, a minimum
empty bed contact time  (EBCT) of six to  10 minutes, and  a maximum hydraulic
loading rate  of approximately 10 gallons per minute per  square  foot of bed
surface area.  The maximum  hydraulic loading  rate  is affected by  the use  of
a  flow  constrictor.  The Fresno office  of  the California Department of
Health  Services recommends  a minimum EBCT  of  10 minutes.  However, Clarke
(3)  reported  on a study where the  raw  source water contained  16 to 20
yg/1 DBCP.  He  reported that GAC adsorption of DBCP ranged  from 2.22 to
4.43 mg/g,  and  he concluded  that the most  efficient use  of  GAC  for  removing
DBCP to levels  of 1 yg/1 or  less  in the  treated water  was the  use of EBCTs
of 1.5  minutes  for POE  units and  1.2 minutes  for  POU  units.
                                     113

-------
    The National Sanitation Foundation (4)  has published guidelines for
management of POE drinking water treatment  systems.   These guidelines
address many types of POE units used for drinking water treatment including
reverse osmosis, ion exchange, activated alumina, sedimentation and filtra-
tion, and GAC.  They do not provide detailed engineering design criteria,
but they do provide numerous general recommendations including the recom-
mendation that POE units should be subjected to rigorous third party
testing for performance evaluation.

                   INSTITUTIONAL/JURISDICTIONAL SETTING

    The starting point for this part of the study was the evaluation of
current California institutional and jurisdictional  factors that pertain to
water treatment, including organics removal and water distribution.  A
public water system is defined by the California Health and Safety Code
(5,6) as a system that has "... five or more service connections or regu-
larly serves an average of 25 individuals daily at least 60 days out of the
year."  Particular attention is being given to those institutional and
jurisdictional factors which pertain to small  water systems.  A state small
water system is defined as "... a public water system which meets one of
the following criteria:

    1. Serves from 5 to 14 service connections and less than 25 individuals
       any part of the year.

    2. Serves 15 or more service connections and any number of nonresident
       individuals less than 60 days per year.

    3. Serves 5 to 14 service connections and 25 or more individuals less
       than 60 days per year."

    Private water systems are those individual water supply systems that do
not qualify as public water systems.  Currently there is little jurisdic-
tional and institutional criteria pertaining to the quality of water pro-
duced by private water systems  in California.

    California has recently experienced a flood of unscrupulous vendors
using scare tactics and other deceptive practices to market their products,
oftentimes in areas already safely  protected by well managed and monitored
public water supplies.  These tactics have bilked impressionable  residents
of considerable sums of money for unneeded or inappropriate items, and at
times the marketed device has potentially contributed to a health problem
(as when an individual served by a  hard water source and on a  low sodium
diet  is sold a water softening  unit employing ion exchange technology).
Recently, the California legislature addressed the regulation  of  POE/POU
units by passing two laws which were subsequently signed by the Governor.
The first bill, Senate Bill 2119  (7) introduced by Senator Torres, requires
the Department of Health Services  (DOHS) to adopt standards and establish  a
procedure for testing performance of POE/POU devices that vendors desire to
                                    114

-------
market in California.  The second bill, Senate Bill 2361 (8) introduced by
Senator McCorquodale, addresses the truth in advertising issue by
establishing measures that mandate the following (9):

   "1. Makes it unlawful to make false claims or statements about a public
       water system.

    2. Makes it unlawful to make false claims about the health benefits of
       a POE/POU device.

    3. Makes it unlawful to make any product performance claims unless such
       claims are based on actual, existing factual data*

    4. Prohibits any  other attempts to mislead or misrepresent."

    This law provides a method  for consumers to check on claims, for the
 filing of a criminal  misdemeanor action, and for recovery of damages.
 However, the two bills  passed by the 1986 California legislature do not
 address a myriad of problems associated with the operation and maintenance
 of POE/POU devices.  Hopefully, these  two pieces of  legislation are the
 forerunners of  future legislation  in California and most other states that
 is needed to define policy  toward  POE/POU treatment, and to establish means
 of assuring that such treatment, where permitteds will  provide the desired
 water quality.

    The State  of Washington's Department of Social and  Health Services has
 been  outspoken  in  its stand on  POE/POU systems, and  suggests that where
 feasible, alternative solutions to installation of POE/POU  units be
 explored  (10).   For example,  if a  water  system  has a problem which is the
 cause of water  contamination,  the  Department advises that  the problem be
 remedied.   If  there is  no  public water system  in a sparsely populated area,
 the  Department  advises  to  install  one  if feasible-   If  neither of the pre-
 ceding steps  can  be accomplished,  the  Department advises as a last resort
 that  a POE/POU  unit meeting National Sanitation Foundation  criteria be
 installed.

     While  the  State of  Washington  does not  endorse the  use  of POE/POU
 treatment  systems,  it recognizes  that  at times  this  may be  the only alter-
 native.  Therefore, the State  of Washington  is  developing  criteria for
 design, operation,  and  maintenance of  POE/POU  units.  Reportedly, the cri-
 teria will  require  that water  from the POE/POU  units be chlorinated prior
 to use, and  that  the entire water  supply serving the house  be treated.  The
 State of Washington has also developed guideline design specifications that
 could serve as  a model  for  other  regions  (11).  The  guidelines address unit
 and media  criteria, and the basic  system components  recommended  for good
 performance.  A minimum unit capacity  of 2  cubic feet of GAC media with an
 EBCT  of five minutes is recommended.   A  five-to-one  length  to diameter
 ratio of the GAC  unit is also  recommended.  The GAC  media  criteria relate
 to minimum  impurities,  moisture content, apparent  density,  particle size
 distribution,  abrasion  resistance,  and carbon  adsorption capacity.  The
 recommended  system  components  include  a water meter, a  prefilter to  reduce
                                     115

-------
participate loading to the GAC media,  pressure gauges to show head losses
through the GAC unit,  and a means of disinfection.

    As reported by Burke and Strasko (12),  the State of New York has
enacted legislation that enables the creation of special town or county
water districts to oversee the installation,  maintenance, and operation of
POE/POU devices used to treat private well  supplies when neither a private
nor public water utility is present to provide service.  An ad hoc commit-
tee appointed by the State to develop uniform criteria for activated carbon
treatment systems issued an interim report  in 1982 which set the following
basic requirements for whole house treatment  units:

   "1. Flow rate of 5 gpm.

    2. Maximum application rate of 10 gallons [per minute] per square foot
       of carbon media surface.

    3. Minimum empty bed contact time of three minutes.

    4. Only virgin carbon.

    5. Disinfection be provided after treatment.

    6. The following appurtenances be provided:

       a.  flow meter;
       b.  raw and treated sample taps;
       c.  adequate valving to isolate units;
       d.  non-toxic materials and coatings;
       e.  withstand pressure;
       f.  ease of access;
       g.  prefnitration where necessary; and
       h.  pressure gauges before and after unit.

    7- Adequate sampling program be developed and implemented."

    The installation and use of POE/POU units will  ultimately require sound
management and jurisdictional arrangements  to assure proper performance to
meet the desired treatment objectives.  Recognizing this need, the US EPA
published in the Federal Register their proposed rules, "Criteria and
Procedures for Public Water Systems Using Point-Of-Use Devices ..." (13).
The proposed rules place the responsibility for ownership, operation, and
maintenance of POE/POU devices with the public water system.  Further, the
utility must develop a monitoring plan approved by the state before ini-
tiating a POE/POU program.

                               STUDY RESULTS

    Several hundred GAC units have been installed by local water con-
ditioning firms on private water well supplies contaminated with DBCP.  The
units being marketed have been approved in concept by the Sanitary
                                   116

-------
Engineering Branch, California Department of Health Services, Fresno,
California.  Typical water well production values are 20 to 1,000 gallons
per day.  GAC units are equipped with flow totalizers, flow restrictors to
control the minimum empty bed contact time, pressure gauges at the inlet
and outlet of the units, and facilities to backwash the carbon to control
head losses.  These units are sold or leased to the users who may contract
with local water conditioning firms to service the units.

    Ten POE units were selected for study and their feed water and product
water are being sampled for DBCP analysis on a four- to eight-week basis.
Cumulative flow data, pH, and temperature readings are also obtained at the
time of sampling for DBCP analysis.  DBCP analyses for this part of the
study were performed by the California DOHS Laboratory.  This laboratory
has an internal quality assurance/quality control program, and this labora-
tory is the certifying authority for other laboratories in California.

    The adsorption phenomena in the GAC beds are affected both by flow rate
and water temperature.  These units operate in an intermittent mode with
periods of relatively intense hydraulic loading followed by idle periods
that may persist for hours.  The season variation of water temperature for
the study's 10 POE units was approximately 11°C, ranging from 12°C to 23°C
as shown in Figure 2.

    The data shown on Figure 3 are the average ratio for all sites of the
DBCP concentration found in the feedwater for a given sampling day divided
by the average DBCP  concentration for all sampling events conducted at the
site.  The variation of the average concentration of DBCP in the feedwater
to the study's POE units is due to the difficulty in obtaining high preci-
sion at the low  levels of DBCP present (0.01 to 3.13 ug/1), and climatic
conditions.  The  low ratio obtained for February 1986 may be the result of
low precision  in  the laboratory analyses  for the month.   (At these low con-
centrations which  are near the detection  limit, an apparent low precision
does not  indicate  that the analytical  results are out of  control.)
However,  the general decrease  in the  ratio beginning  in May 1986 probably
results from the  earlier cessation of  rainfall events and the persistence
of nearly  no rainfall into early 1987.  This ratio is expected to  recover
with the  onset of  normal precipitation events which may wash DBCP  from the
vadose zone down  into the ground water.   Figure 4 for Site 2 shows con-
tinuous removal  of DBCP  from  the feedwater to below the detection  limit  of
0.01 ug/1, which  is  the  performance achieved by  those units  receiving proper
O&M.

    At the  initiation of the  study  in  November  1985,  the  POE unit  at  Site
4, as  shown on Figure 5, had  a DBCP product water concentration of 3.64
ug/1,  significantly  greater than the  feedwater  DBCP concentration  of  2.66
ug/1.   The  product water DBCP  concentration  remained  high until early 1986,
when the  GAC in  the  POE  unit was replaced.  Thereafter,  the  POE unit  has
removed the DBCP  to  near or below  the  detection  unit.
                                     117

-------
                               WATER TEMPERATURE
I
I
24 -
23 -
22 -
21 -
20 -
19 -
18 -
17 -
1B -
16 -
14 -
13 -
12 -
11 -
      10
      Oat-SB
   Figure 2.
                    .Fob-as
                             jun-ee         oot-ee
          Fresno eastside POE  study, water  temperature
                               118

-------
 UJ
 I
 I
 10
 m
 o
1.5 -r
1.4 -
1.3 -
1.2 -
1.1 -
1.0 -
0.9 -
o.a -
0.7 -
0.6 -
0.5 -
0.4-
0.3 -
O.2 -
0.1 -
           FRESNO EASTSIDE  POE STUDY
                     FEEDWATER DBCP CONG. TREND
     0.0
      Oct
    -85
Fob—86        Jun—86
                         DBCP Cone. Ratio
                                            Oct-86
Feb-87
Figure 3.   Fresno  eastside POE study, feedwater DBCP concentration
           trend.
            FRESNO  EASTSIDE  POE  STUDY
                     SITE 2:  DBCP CONCENTRATION
     1.0
 o
 I
 Ul
 o
 o
 o
 Q.
 O
 m
 Q
0.9 -

0.8 -

0.7 -

0.6 -
0.5 -

0.4 -

0.3 -

0.2 -

0.1 -
     0.0
                                    Note:  No DBCP detected
                                           in product water
      Oct-85        Feb-86         Jun-86        Oct-86
              D   FEED WATER          + PRODUCT WATER
Figure 4.  Fresno eastside POE study, site 2:  DBCP  concentration.
                               119

-------
    Figures 6 and 7  show the  data  for  Sites  5 and 8,  respectively.  The
DBCP concentration in  the product  water at Site 5 varies from  not detec-
table to 0.07 ug/1 which is well below the established California action
limit of 1.0 yg/1.  The  GAC  in  the Site 8 POE unit may be  nearing
exhaustion, since a  product water  DBCP concentration  of 0.38 yg/1 was
attained in August 1986.

    The Site 9 data  shown on  Figure 8  appears to indicate  that the  POE
unit's GAC is exhausted.  And,  the Site 10 data shown on Figure 9 shows
initial high product water DBCP concentrations followed by mostly nondetec-
table DBCP concentrations resulting from a change of  the unit's GAC.

                                CONCLUSIONS

    POE and POU devices  have  been  designed to effectively  remove DBCP from
drinking water.  The results  obtained  from monitoring 10 GAC POE devices
show that the performance of these units can change markedly over short
periods of time.  Thus,  these units require  conscientious, periodic moni-
toring.  This unit operation requirement was not carried out by the owners
or the vendor; the owners generally lack the expertise, and  the vendor  has
no contractual authority or  responsibility to monitor the  POE  units.  The
monitoring of the operation  of the POE (and  POU) units appears to be  a
significant shortcoming in the application of this  technology  in many areas
of the United States.   This  can pose a significant  health  threat to many
individuals who unwittingly drink contaminated water  that  is  supposedly
treated using POE/POU technology.   In summary,  the  technology  of the  GAC
POE  units  that were studied seemed to be  very good.   The  primary problems
observed were with operation and maintenance of  the GAC  POE  units.

                 FRESNO  EASTSIDE  POE  STUDY
           4.0
                           SITE 4: DBCP CONCENTRATION
           3.5 -
           3.0 -
           2.5 -
           2.0 -
           1.5 -
           1.0 -
           0.5 -
           0.0
                                  •4-
            Oct-85         Feb-86        Jun-86        Oct-86


                   a  FEED WATER         +  PRODUCT WATER

     Figure 5.  Fresno eastside POE study,  site 4:   DBCP  concentration.

                                    120

-------
     4.0
            FRESNO  EASTSIDE  POE STUDY
                     SITE 5:  DBCP CONCENTRATION
 o
  o
  o
  o
  0.
  o
  m
  a
3.5 -

3.0

2.5

2.0

1.5

1.0

0.5
                   Feb-86

              D   FEED WATER
                          Jun—86        Oct—86

                            +  PRODUCT WATER
                                                         Feb-87
Figure 6.  Fresno eastside POE study,  site 5:  DBCP concentration.
  O
  F=
  LL!
  O
  O
  O
  Q.
  O
  m
  Q
 3.0
 2.8 -
 2.6 -
 2.4-
 2.2 -
 2.0 -
 1.8 -
 1.6 -
 1.4-
 1.2 -
 1.0 -
 0.8 -
 0.6 -
 0.4 -
 0.2 -
             FRESNO  EASTSIDE  POE STUDY
                      SITE 8:  DBCP CONCENTRATION
                    Feb-86
                           Jun—86
Oct-86
Feb~B7
               D   FEED WATER         +   PRODUCT WATER
Figure 7.   Fresno eastside POE  study, site 8:  DBCP concentration.
                              121

-------
              FRESNO EASTSIDE  POE  STUDY
                       SITE 9: DBCP CONCENTRATION
        1.5
    01
    3
    z
    o
    LJ
    O
    O
    O
    Q.
    O
    m
    a
1.4-
1.3 -
1.2 -
1.1 -
1.0 -
0.9 -
0.8 -
0.7 -
0.6 -
0.5 -
0.4-
0.3 -
0.2 -
0.1 -
0.0 -
          OCT-85
                      FEE-86
                                  JUN-86
                                              OCT-86
                a   FEED WATER        +   PRODUCT WATER
   Figure 8.  Fresno eastside POE study, site 9:  DBCP concentration,

             FRESNO  EASTSIDE  POE STUDY
                      SITE 10: DBCP CONCENTRATION
   O
   F
   O
   o
   o
   Q.
   o
   m
   Q
                    Feb-86
                                 Jun—86
                                             Oct-86
                                                         Feb-87
               D  FEED WATER         +  PRODUCT WATER
Figure 9.  Fresno eastside POE .tudy,  site 10:   DBCP concentration.
                                122

-------
                               REFERENCES

1. Anonymous.  Organic chemical contamination of large water systems in
   California.  Department of Health Services, State of California, April,
   1986.

2. Gaston, J.M.  Design aspects of granular activated carbon POU/POE devi-
   ces.  Paper presented at Workshop, Point-of-Entry/Point-of-Use
   Devices, California State University, Fresno, February, 1987.

3. Clarke, W.F.  Carbon adsorption of DBCP from domestic water.  Master of
   Science Thesis, California State University, Fresno, 1981.

4. Bellen, G.E., Anderson, M., and Gottler, R.A.  Point-of-use treatment
   to  control organic and  inorganic contaminants in drinking water.  A
   report submitted  to the Water Engineering Research Laboratory, U.S.
   Environmental Protection Agency, Cincinnati, Ohio, September, 1985.

5. California Health and Safety Code.  California Safe Drinking Water Act,
   laws  and  standards  relating to domestic water supplies.  Excerpted from
   the California  Health and  Safety Code and the California Water Code,
   Department of Health Services, Berkeley, California, 1979.

6. California Health and Safety Code.  California waterworks standards.
   Excerpted from  the  California Health and Safety Code and the California
   Administrative  Code, Title 22, Department of Health Services, Berkeley,
   California,  1980.

7. California Senate Bill  No.  2119.  An act to add Chapter 8.5  (commencing
   with  Section  4057)  to Division 5 of the Health and Safety Code.
   Approved  by  Governor and  filed with Secretary of State September 26,
    1986.  Sacramento.

8. California Senate Bill  No.  2361.  An act to add Article 6 (commencing
   with  Section  17577)  to  Chapter  1 of Part 3 of Division 7 of  the Busi-
    ness  and  Profession Code.   Approved by Governor September 26, 1986 and
    filed with Secretary of State September 29,  1986.  Sacramento.

9. Rogers,  P.A.  Regulation  of point-of-use devices.  Paper presented at
   Workshop, Point-of-Entry/Point-of-Use Devices, California State
   University,  Fresno,  February, 1987.

10.  Pluntze,  J.C.   Point-of-use treatment - another view.  U.S.  Water
    News, August,  1985.
                                    123

-------
11.  State of Washington.   Home  treatment  units  and  ethylene  dibromide (EDB)
    removal.  Department  of Social  and  Health Services  Letter,  Directive.
    June 18, 1985.

12.  Burke, M.E.,  and  Strasko, G.A.   Water quality districts  in  New York
    state.  Paper presented at  Annual Conference, American Water Works
    Association,  Denver,  Colorado,  June,  1986.

13.  U.S. Environmental  Protection Agency.   Proposed rules, criteria
    and procedures  for  public water systems  using point-of-use  devices.
    Federal  Register.   50:219,  Part 141,  November 13, 1985.
                                   124

-------
             EVALUATION OF RADIUM REMOVAL AND RADIUM DISPOSAL
                 FOR A SMALL COMMUNITY WATER SUPPLY SYSTEM

                       by:  Kenneth A. Mangel son
                            Rocky Mountain Consultants,  Inc.
                            Englewood, CO  80111


                               INTRODUCTION

    In 1984 a radium removal treatment plant was constructed  for the small
community of Redhill Forest located in the central  mountains  of Colorado.
The treatment plant consists of a process for removing iron and manganese
prior to the ion exchange process for the removal of radium.   The raw water
comes from deep wells and has naturally occurring radium and  iron con-
centrations of about 30 to 40 pCi/1 and 7 to 10 mg/1, respectively.  Also,
before the raw water enters the main treatment plant, it is aerated to
remove radon and carbon dioxide gases.

    The unique features of the Redhill Forest Treatment Plant are related
to the way in which the radium, removed from the raw water, is further
treated and eventually disposed as treatment plant waste.  A separate
system removes radium, only, from the backwash/regeneration water of the
ion exchange process and  it is permanently complexed on a Radium Selective
Complexer  (RSC) resin marketed by Dow Chemical.  The RSC resin will be
replaced with virgin resin as needed, and the radium containing resin will
be transported to a permanent final disposal site acceptable to the state
regulatory agencies.

    This paper presents a description of the radium removal treatment
system, and some of the results of an on-going EPA-sponsored monitoring
study of the processes and other factors relating to the overall operation
of the radium removal system.  Included are  the procedures for final dis-
posal of the RSC resin containing radium.

                             STUDY OBJECTIVES

    The overall study objectives were to monitor and evaluate the operation
of treatment plant processes to remove iron, manganese, and radium and to
determine appropriate methods for disposal of plant waste water and
complexed radium waste.  The following summarizes the processes which make
up the treatment plant, and identifies the areas where in-depth monitoring
is performed:
                                    125

-------
    o   Aeration  for  radon  and  carbon dioxide gas  removal.

    o   Chemical  clarification  including  settling  and  filtration  for  iron
       and  manganese removal.

    o   Ion  exchange  for radium and  hardness  removal.

    o   Chlorination  and water  stabilization.

    o   Removal  of radium from  ion exchange  regeneration  water by Radium
       Selective Complexer resin.

    The problem of radium in ground water, which  serves  as  the raw water
supply for  the development,  is common  to many  communities  in  the United
States.  If the development  of new  water sources  that do not  have a  radium
problem is  not possible or economically  feasible, then the  treatment proc-
esses  for radium removal evaluated  in  this  study  are  alternatives that need
to be  considered.  This report concerns  itself with the  treatment alter-
natives and is contrasted with locating  new  raw water sources void of
radium.

    The treatment of well  water for the  removal of radium is  not practiced
to any great extent in the water treatment  field.  However, the  ion
exchange process using standard water  softening-type  resins for  radium
removal is  well documented.   The Redhill Forest water treatment  system
incorporates a new process for concentrating the  radium  removed  by the ion
exchange process to simplify the final radium disposal problem.   The regen-
eration water from the ion exchange process  passes through  a  bed of  Radium
Selective Complexer (RSC) resin to  remove the high levels of  radium  before
the waste water is discharged to the infiltration/evaporation pond for
final  disposal.  There are no known water treatment systems like the
Redhill system.  The RSC resin has  been  used on a trial  basis at several
locations primarily in Texas,  and one  site in Wyoming.  But in all these
cases, raw water from the wells was passed directly through the RSC  bed
with radium levels  in excess of 100 pCi/1.

     GENERAL DISCUSSION OF THE TREATMENT PLANT PROCESSES AND EQUIPMENT

    Raw water from  the  two wells serving the development,  is pumped through
a countercurrent flow carbon dioxide  (C02) stripper tower  located at the
booster pump house.  The purpose of the stripper tower,  which is
constructed of  PVC, is  to remove dissolved gases (specifically  radon and
C02) from  the  raw water.  Following the stripper tower,  the water is
pumped to  the  treatment plant at a rate of about 90 to  100 gpm  for  further
water  treatment  to  remove iron, manganese, radium, and  hardness  prior  to
chlorination and discharge  into the water distribution  system.

    Upon entering the  treatment plant,  alum, potassium  permanganate, and  a
polyelectrolyte  are added to  the raw water to  remove  iron  and manganese by
chemical precipitation.  The  treatment  unit is a prefabricated  self con-
tained unit that  includes a mixing and  flocculation  chamber,  tube settlers,
                                    126

-------
and multi-media filtration.  The effluent from the iron and manganese
removal process is further treated in a system using a cation resin to
remove radium and hardness.

    The effluent from the ion exchange system is chlorinated, zinc hexame-
taphosphate added to control corrosion and sequester any residual iron, and
subsequently pumped to the treated water storage tank.  The radium removed
from the water supply in the ion exchange process is removed from the rege-
neration water by passing the ion exchange process waste water through a
separate treatment process.  The process involves the permanent complexing
of the radium on a Radium Selective Complexer (RSC) material.  The waste
water  from this process along with the backwash waste water from the iron
removal process is pumped to the final disposal infiltration/evaporation
(I/E)  pond located to the west of the plant.  The RSC resin in the RSC tank
is retained in PVC cartridges specifically constructed to facilitate
handling and disposal of the resin complexed with radium when replacement
is necessary.  Figure 1 is  included to show the plant flow diagram
including the processes presented above.

   ULTIMATE DISPOSAL OF WASTE WATER AND RADIUM REMOVAL FROM WATER SUPPLY

    The original concept and design approved by the Colorado State Health
Department for ultimate disposal of waste generated at the treatment plant
is described below.

PLANT  WASTE WATER

    All plant waste water  from the plant operation after treatment for
maximum radium removal  is  discharged  into a infiltration/evaporation (I/E)
pond  located west of the treatment plant.  The main purpose of the pond is
to allow for rapid  infiltration  of the waste water into the Morrison for-
mation, which dips  steeply  to the east and is located below the Dakota for-
mation in the area  of the  raw water supply wells.  The deep wells obtain
the  raw water from  the  Dakota formation to supply the development.

RADIUM WASTE

    Most of the radium  removed from the raw water entering the treatment
plant  eventually is complexed on the  RSC resin.  As needed, the cartridges
of RSC resin complexed  with  radium will be replaced and the RSC resin
transported to an approved  hazardous/radiological waste disposal facility
for  final disposal.

                            MATERIALS  AND METHODS

    The experimental procedures  for this project generally consisted of in-
depth  monitoring of the operation of  the full-scale Redhill Forest water
treatment plant over a  two-year  period from October 1985 through September
1987.  All water quality parameter concentrations were determined according
to Standard Methods for the  Examination of Water and  Wastewater  (15th
Edition).  Most of  the  water quality  analysis work was performed by  Hazen
                                    127

-------
Research Laboratory in Golden,  Colorado.   Some analysis work was performed
by the EPA Laboratory in Cincinnati,  Ohio and some radon gas analysis was
performed by Lowry Engineering  in Maine.

    In-depth monitoring included water quality sample collection and analy-
ses, field testing, flow measurement, and detailed plant operation and was
performed to evaluate the following components of the treatment plant
operation:

    o  Aeration system for radon removal.  Water samples were collected and
       analyzed for radon concentrations  before and after aeration.

    o  Treatment system for iron and manganese removal.  Water samples were
       collected and analyzed to assess the efficiency of operation.  Typi-
       cally, iron, manganese,  gross alpha and gross beta, and radium 226
       measurements were made.   Also, the process waste water from backwash
       operations was analyzed a number of times to determine the com-
       position of the waste water discharged to the I/E pond for final
       disposal.  Parameters of primary interest included total iron,
       manganese, solids, and radium.

    o  Ion exchange process for radium and hardness removal.  Water samples
       were collected for the inflow to and outflow of the system.  The
       water quality samples typically were analyzed for iron, manganese,
       sodium hardness, and radium.  On several occasions, water samples
       were collected of the backwash, regeneration, and guide reuse water
       on a frequent time basis and analyzed.  The purpose of the moni-
       toring of the process waste water was to determine its chemical
       make-up.

    o  Radium Selective Complexer process for radium removal.  This process
       was monitored on a frequent basis to determine the efficiency of
       radium removal from the ion exchange process waste water, the build-
       up of radium in the complexer resin, etc.  Environmental radiation
       monitoring of the area outside the RSC tank surface was done to
       determine the exposure and to relate the exposure to radium build-up
       on the complexer resin.

    o  I/E pond monitoring of the sand and soils was done to determine the
       extent of radium build-up due to the disposal of plant waste water
       containing small amounts of radium.

    o  General plant monitoring of plant flow rates, volumes of water proc-
       essed, waste water volumes, etc. was performed to be used along with
       water quality data in determining plant process efficiencies, plant
       operation and maintenance costs, etc.

    o  Some radon gas measurements were conducted on-site using a RDA-200
       Radon/Radon Daughter Detector unit manufactured by EDA Instruments,
       Inc.  Also, some samples were collected and sent to Lowry
       Engineering for additional radon gas analysis.
                                    128

-------
                          RESULTS AND CONCLUSIONS

    The pre-aeration system has proven to effectively remove radon and car-
bon dioxide gases from the raw water supplied by the deep wells.  Carbon
dioxide gas has been typically reduced from about 125 mg/1 to 25 mg/1 in
the aeration system.  The reduction of radon gas, based upon the measure-
ments made, has been about 85 percent from about 23,000 pCi/1 in the raw
water to about 3,400 pCi/1 in the effluent from the aeration system.  Addi-
tional measurements have indicated that the radon gas concentration in the
treated water from the main treatment plant is about 600 pCi/1.  The pur-
pose of the iron removal process is to remove iron and manganese from the
raw water before treatment in the ion exchange process for radium removal.
However, some radium is also removed in this process.  Based upon the moni-
toring results over the last two years, about 13 percent of the radium in
the inflow to this process has been removed.  When the iron removal system
is backwashed, the radium removed is wasted along with the backwash water
to the I/E final disposal pond.  Based upon the results of the monitoring
of the backwash water, the average concentration of radium in the
wastewater is about 60 pCi/1.

    The ion exchange system removes radium and hardness and residual iron
and manganese through the use of a standard cation exchange resin.  The
process has been very effective  in removing radium, hardness, residual
iron, and  polishing the effluent from the iron removal process as long as
the  ion exchange capacity is not exceeded.  The monitoring results
generally  indicate  radium levels of less than the Drinking Water Standard
of 3  pCi/1 and  iron levels below the  recommended level of less than 0.3
mg/1.  Frequent monitoring of the system operation has indicated that the
radium breakthrough occurs between 40,000 and 45,000 gallons  (i.e., 178 to
200  resin  bed volumes).  The quality  of the inflow (i.e., effluent from the
iron  removal  system) to and effluent  from the ion exchange process is shown
in Table  1.
                TABLE 1.   SUMMARY OF  WATER QUALITY  OF  WATER
                      TO  AND FROM ION EXCHANGE  PROCESS
     	Inflow	Effluent

     Flow rate            90 to 100 gpm            90  to  100  gpm
     Iron                0.15 to 2.7  mg/1          0.03 to  0.5 mg/1
     Manganese            0.4 to 1.3 mg/1           0.01 to  0.15 mg/1
     Sodium              7.4 to 12.5  mg/1          40  to  150  mg/1
     Hardness             212 to 350 mg/1           5 to 70  mg/1
     Radium 226           22 to 35 pCi/1            0 to 4 pCi/1
                                    129

-------
    Figure 1 shows the flow volumes for each part of the total  system
operation for an assumed raw water flow volume of 10,000 gallons into the
plant.  Also presented are the average water quality data for each com-
ponent that makes up the treatment plant.

    The RSC system is designed and operated to remove radium only from the
ion exchange process waste water and to permanently concentrate the radium
on the complexer resin.  On July 10, 1986, new RSC resin was placed in the
complexer tank and a detailed program of monitoring the flow rate and the
water quality of the inflow and outflow was initiated.  Table 2 presents a
summary of some of the results of the monitoring from July 10,  1986 up to
the middle of February, 1987.  It should be noted that the flow rate
through the column has been about 22 gpm, which is equivalent to a surface
loading rate of about 10 gpm/ft2.  Also, the RSC resin bed depth is 2 feet.

    As can be seen in reviewing the data in Table 2, the RSC resin is truly
radium selective, with generally over 99 percent removal of radium from the
influent waste water to the treatment system.  Average data for the water
quality parameters included in Table 2 are shown on the bottom of the
table.  The average inflow and outflow water quality data indicate that
iron, sodium, hardness, and total solids are virtually unchanged in passing
through the resin.  However, over 99 percent of the radium in the influent
is removed and concentrated on the RSC resin.  Also, shown on the bottom of
Table 2 is the total quantity of radium removed and concentrated on the
resin from July 10, 1986 up to the date of the last data entry in the
table.  Based upon the operation of the plant since July 10, 1986 and the
past water demands, the rate of radium build-up on the RSC resin is about
310 uCi/yr (310 x 106 pCi/yr).

    Further, it has been determined that the rate of radium removed from
the raw water and permanently complexed on the RSC resin is about 9.6 uci
(9.6 x 106 pCi) per 100,000 gallons of water treated at the plant.  After
some period of operation, the RSC resin containing radium will  be removed
from the RSC tank, replaced with new resin, and the old resin disposed of
probably at the Nevada waste disposal site.  It is anticipated that the RSC
resin will be replaced and the old complexer material disposed of when the
radium level reaches about 3,080 uCi (3,080 x 106 pCi).  The 4 cubic feet
of RSC resin can then be placed in a 55-gallon drum, 3.35 cubic feet of
concrete added, and the entire drum transported to Nevada for final dispos-
al.  This method of handling the radium waste will insure that the total
radium content of the container to be buried will not exceed 10 nonocuries
per gram  (i.e., 10,000 pCi/gram).  Proper handling procedures to avoid
radiation exposure would be required when removing and transporting the
spent RSC.

    Finally, plant operating costs have been determined or estimated as
shown in Table 3.
                                    130

-------
CO
                   r
                                  BOOSTER PUMP HOUSE
                                                                                                                                                              HOUSED  IN  TREATMENT
                                      RADON-29,OOO pci/l
                                                                                                       TOS > 900 mg/l
                                                                                                       RADIUM < 3
 NOTES:

 1.  Jon exchange tanks are assumed to be  backwashed after
     40,000 gallons of water have been treated.

 ?.  Flows  shown are average flows for every  10,000 gallons
     of raw water processed through the plant from the wells.

 3.  The treatment processes Include:

     a.   Aeration for carbon dioxide (ph adjustment) and
         radon gal rracvel.

     t>.   Chemical precipitation of Iron and eanganese In the
         Septune mcrofloe flocculator/settler/fllter unit.

     c.   Jon exchange for radius removal end  softening.

     d.   Radius ranoval process uilng  Radius  Selective
         tonpleser (RSC).  Removes radium  frora the (on
         exchange backwash wastewater  and  concentrates -
         radium on coaplexer r«s1n.

«.    The norwl treatwnt pUnt fiowrate It about 100 gp».
     The «ter froai th« 1on exchange process  for radtua
     rtnoval  and softenlos It  discharged Into the sjstewater
     holding Uftk fnw which th« wastwater Is punped through
     the Radlua Selective'Ctaplexer  at  * constant rate for
     radii* raoval.
                                                                                                                                                        IRON » TO m«/l
                                                                                                                                                        No • 1,730 .5/1
                                                                                                                                                        TDS >IO,SIO>ng/l
                                                                                                                                                        HwdMtl -l,6IO«(/i
                                                                                                                                                        RADIUM • «0 |«i/l
                                                                                                                                                                  REDHILL FOREST
                                                                                                                                                                   FVOW (XASR&M

                                                                                                                                                                 WATER TRE4TWENT
                                                                                                                                                                 PLANT PROCESSES
                                                              Figure  1.   Water  treatment  plant  flow volumes.

-------
                                             TABLE 2.  SUMMARY OF WATER QUALITY DATA FOR REGENERATION
                                           WASTEWATER FROM  ION EXCHANGE REGENERATION THROUGH RSC RESIN*
                                                         (Effluent Discharged to I/E Pond)
CO
ro
Accumulated
Volume
Treated Bed**
Date Gals. Volumes Sample
7/10/86 0
7/30/86 2,400
8/31/86 9,460
9/29/86 14,600
10/30/86 22,600
11/26/86 27,700
1/14/87 39,550
2/12/87 47,700
AVERAGES
0 Inflow
Outflow
77 Inflow
Outflow
305 Inflow
Outflow
471 Inflow
Outflow
729 Inflow
Outflow
894 Inflow
Outflow
1,276 Inflow
Outflow
1,539 Inflow
Outflow
Inflow
Outflow
Iron
mg/1
2.
0.
2.
1.
9.
8.
7.
7.
2.
2.
7.
6.
31.
27.
64.
67.
10.
9.
48
98
03
56
0
5
21
15
79
07
17
30
4
8
3
0
4
6
Manganese Sodium
mg/1 mg/1
23.8
16.7
31.8
32.2
33.1
33.1
30.5
31.5
33.1
33.5
28.2
26.9
19.2
18.1
27.0
29.7
26.4
26.4
11,600
13,300
11,000
11,000
12,600
12,700
11,400
11,500
8,170
8,640
13,400
13,300
9,350
9,000
10,900
10,800
10,340
10,270
Hardness
mg/1
476
245
9,850
10,200
11,500
11,600
8,350
8,420
9,380
10,100
9,620
9,520
7,260
7,740
12,400
13,800
8,450
8,640
Parameters
Total Total
Solids Radium
mg/1 pC1/l
34
34
41
41
54
55
37
37
35
35
45
45
31
30
42
42
37
37
,900
,600
,700
,800
,200
,200
,600
,600
,000
,300
,400
,500
,300
,400
,700
,110
,440
,590
860+30
16+11
1280+40
1.6+3.2
1400+40
9.4+3.5
920+30
4.1+2.4
860+50
5.3+Z.8
1040+30
8.1+3.3
1070+60
8.4+Z.3
1660+70
2.2+T.4
1,025
7.1
%
Radium
Removal
98
99
99
99
99
99
99
99
99
.1
.9
.3
.6
.4
.1
.2
.9
.3
                   From 7/10/86 to 2/12/87 (I.e., 217 days)

                   47,700 gallons of plant wastewater was treated 1n RSC tank.  The following 1s the amount of radium
                   removed and  deposited In the resin.
                   Radium removed » 47,700 (3.785) (1025-7.1)
                                  = 183.78 x 106 pC1
                                  = 183.78 uC1  or about 0.847  pC1/day
                   Estimate for year = 309 uCi

                    *A11  data not Included 1n Table
                   **Res1n bed  volume - 4.15 ft3 (31.0 gals)

-------
           TABLE  3.   SUMMARY TREATMENT PLANT OPERATING COSTS
                                           Cost/1000 Gallons
                Item                       of Water Treated
    1.  Plant Chemicals,  Alum,                     $0.137
       Permanganate,  Chlorine,  etc.
       Salt                                      $0.475
    2.  Energy Costs                              $0.206
    3.  RSC Resin Disposal                         $0.088
       (includes disposal  and  new resin)
       TOTAL                                     $0.906*
*0perator cost not included.
                                   133

-------
                RADON REMOVAL FROM GROUND WATER USING GAC

                      by:  N.E. Kinner
                           Dept. of Civil Engineering
                           University of New Hampshire
                           Durham, NH

                           C.E. Lessard
                           Dept. of Civil Engineering
                           University of New Hampshire
                           Durham, NH

                           J. Lowry
                           Dept. of Civil Engineering
                           University of Maine
                           Orono,  ME

                           H. Stewart
                           N.H. Dept. of Environmental  Services
                           Concord, NH

                           R. Thayer
                           N.H. Dept. of Environmental  Services
                           Concord, NH


    Radioactive elements occur naturally in  many geological  formations.
Ground water that comes  in contact  with  these  elements  may contain varying
amounts of radioactivity.  The main source  of  the radioactive contamination
is uranium and its progeny.  Unlike other uranium progeny, radon  is a
colorless, odorless, and tasteless  gas  that  may  be dissolved in ground
water.  Radon decays rapidly  (half-life  = 3.82 days)  to a series  of short-
lived progeny that emit  alpha,  beta, and gamma radiation.  Once radon
enters a building via the ground water  supply,  it is  released relatively
easily into the air, particularly when  using showers, washing machines,
dishwashers, or faucets.  When inhaled,  radon  and its progeny may cause  lung
cancer.

    The problem of radon contamination  in ground water  is well documented in
northern New England.  Levels in wells may  range from <1,000 pCi/1 to more
than 1,000,000 pCi/1. In New  Hampshire,  it  is  projected that 100  or more
community water supplies and  a high percentage of the 1,500 noncommunity
public water supplies could require treatment  for removal of radon.
                                     134

-------
    Although some research on radon removal from household wells has been
conducted, to date there has not been a concerted effort to evaluate radon
removal techniques for community water supplies.  The EPA, New Hampshire
Department of Environmental Services, and University of New Hampshire are
working together under a cooperative research agreement.  The study is
evaluating three treatment techniques (Granular Activated Carbon, Diffused
Bed Aeration, and Packed Tower Aeration) in terms of their removal effi-
ciency, economics, and safety.  In addition, several passive technology
techniques will be evaluated, with the possibility of retrofitting low-cost
alternatives to affected small public ground water supplies.

    The pilot studies are currently being conducted at two sites:  Rolling
Acres Mobile Home Park, Mont Vernon, New Hampshire and Amherst Gardens
Mobile Home Park, Amherst, New Hampshire.  The sites were selected because
1) their ground water supplies represent two ranges of radon contamination
(Mont Vernon = 150,000 - 300,000 pCi/1 and Amherst = 35,000 - 55,000
pCi/1); 2) they are good models of small community water supplies (Q =
5,000 - 16,000 gpd); and 3) they are easily accessible for monitoring.

    Initially, a series of tests were run to evaluate the radon sampling
and analytical technique.  Two methods of obtaining samples, an inverted
funnel and a direct syringe technique, were compared.  Neither the means
nor the variances of these methods were statistically different.  As a
result, the direct syringe technique is being used due to its greater
simplicity.  The radon analytical technique being used was developed by
Pritchard and Gesell (1978).  Although the standard procedure recommends
using 5 ml of scintillation cocktail with  10 ml of aqueous sample, experi-
ments conducted on field samples indicated that 10 ml of scintillator
yielded better precision.  As with the Pritchard and Gesell method, experi-
ments  indicated that the sample should be  held a minimum of 4 hours to
allow for development of secular equilibrium and a maximum of 12  hours
before decay and volatilization become significant.

    During the first phase of the evaluation (starting  in October 1986),
only the Granular Activated Carbon  (GAC) units have been operating at both
sites.  The Amherst filter contains  30 ft3 of GAC in a  single fiberglass
tank, while the Mont Vernon system consists of two  filters operating in
series that contain a total of 47 ft3 of GAC.  The  influent and effluent
and water samples from  intermediate  points within the filters were moni-
tored  intensively during startup (approximately six weeks) and each week
thereafter.  Water samples were monitored  for radon, alkalinity,  turbidity,
temperature, pH, dissolved oxygen, bacterial enumeration, iron, and manga-
nese.  Monthly water samples were taken for uranium and  radium.   In addi-
tion,  the gamma and beta emissions from the filters were monitored using a
GM survey meter.  Air monitoring will be performed  in the pumphouse and
mobile homes during operation of each system.  Over the  course of opera-
tion,  the units will be backwashed,  monitored during typical diurnal flow
variations, and subjected  to periods of high and  low flow.  The GAC will be
operated  for approximately one year.
                                    135

-------
    During startup, the units exhibited typical removal profiles
throughout as radon was exponentially removed.  The amount of radon that is
being removed by the GAC units has remained steady (Mont Vernon = 28.0 +
0.5 mCi, Amherst = 12.8 + 3.2 mCi) during the course of operation.  At both
sites, design specifications (10,000 pCi/1) for removal have been exceeded
as effluent values continue to rise.  The major reason for the higher radon
levels in the effluent appears to be increased flow (Mont Vernon design Q =
6,500 gpd, Actual Q = 5,400 to 9,000 gpd; Amherst design Tj = 9,100 gpd, Actual
0 = 15 000 to 17,000 gpd) and increased influent radon levels (Mont Vernon
design = 155,000 pCi/1, Actual = 200,000 to 250,000 pCi/1; Amherst design =
39,750 pCi/1, Actual = 35,000 to 55,000 pCi/1).

    Water analyses indicate that the GAC may be saturated with uranium
that  is found in the influent water at both sites  (Mont Vernon = 20 to 30
pCi/1  Amherst = 50 to 100 pCi/1).  The units  are also emitting substantial
amounts of gamma radiation (4 to 45 mR/hr).  Since the filters stabilized,
there has been no  significant change in the alkalinity, turbidity, pH, and
temperature  as the water passes through the units.

    The GAC  filters will be cored in the near  future to determine the
amount  of iron, manganese, bacteria, and uranium and its progeny adsorbed
at  various depths.  The  units will  also be backwashed  and subjected to
fluctuating  flow conditions to test the stability/efficiency of the pro-
cess.  Monitoring  will be  conducted to assess  the  impact of the GAC treat-
ment  on indoor air radon  levels.  Passive  treatment technologies, diffused
bubble,  and  packed tower  aeration systems will  be  started during  the  summer
of  1987.
      320000 -
      240000 -
      160000 -
       QOOOO -
                                                      I

                                                     07
08
 i

09
                                               06

                                        PORT NUMBER

               Figure 1.  Rolling Acres, Mont Vernon, NH GAC profile.
            10
                  EF
                                       136

-------
                           BIBLIOGRAPHY


Pritchard, H.M. and T.F- Gesell.  1977-  Rapid Measurements of 222Rn
Concentrations in Water with a Commercial Liquid Scintillation Counter.
Health Phys.  33:577.
                                   137

-------
                   IMPACTS OF REGULATORY REQUIREMENTS ON
                        HANDLING WATER PLANT WASTES

                       by:  David A.  Cornwall
                            Environmental  Engineering & Technology,
                              Inc.
                            Newport News,  VA  23606


    Considerations of regulatory requirements  on  the  handling  and disposal
of water plant wastes have traditionally centered around whether the waste
can be directly disposed of to a watercourse and  if so, what type of
pretreatment is necessary.  Most of the regulatory and subsequent treatment
emphasis has been placed on the commonly produced coagulant and lime
sludges.  However, treatment requirements  in response to discharge regula-
tions have dominated the waste handling field for the last 20  years  for all
types of wastes resulting from the production of  potable water.

    In Table 1, the primary wastes produced at water  treatment plants are
divided into solid/liquid wastes, liquid phase wastes, and gas phase wastes.
Solid/liquid wastes include the traditional  sludges,  as well as spent GAC,
slow sand filter wastes, spent media  and waste from precoat filtration
plants, and wastes from iron and manganese removal plants.  Liquid phase
wastes are normally produced in the removal  of trace  inorganic or organic
compounds, or in hardness removal and include spent ion exchange brine,
reject from reverse osmosis, and reject from activated aluminum adsorption.
Gas phase wastes involve the off-gases produced from  air stripping of vola-
tile organic compounds.

    Regulatory considerations in the  handling of  water plant wastes  now
go beyond simply whether a particular waste can or cannot be disposed of
into a receiving stream.  Considerations include:

    o  Regulations involving the handling  and disposal of water plant
       wastes;
    o  Impacts of the 1986 Safe Drinking Water (SOW)  Act on waste charac-
       teristics and ultimate handling;
    o  Impacts of wastes on meeting new water quality goals.

Each is discussed below.
                                    138

-------
               TABLE 1.  MAJOR WATER TREATMENT PLANT WASTES
SOLID/LIQUID WASTES
    1.   Alum Sludges
    2.   Iron Sludges
    3.   Polymeric Sludges
    4.   Softening Sludges
    5.   Backwash Wastes
    6.   Spent GAC or Discharge from Carbon Systems
    7.   Slow Sand Filter Wastes
    8.   Wastes from Iron and Manganese Removal Plants
    9.   Spent Pre-Coat  Filter Media
 LIQUID  PHASE WASTES
    10.   Ion-Exchange Regenerant  Brine
    11.   Waste Regenerant from Activated Alumina
    12.   Reverse  Osmosis Waste Streams
 GAS PHASE  WASTES
    13.   Air Stripping Off-Gases
                                     139

-------
                        WASTE DISPOSAL REGULATIONS

    Table 2 shows an overview of the applicable regulations for the dispos-
al of water plant wastes.  The first category is the area where most of the
emphasis has been placed — disposal to streams.  Regulations in this area
primarily involve meeting the in-stream water quality criteria.

      TABLE 2.  REGULATORY ACTS GOVERNING WATER PLANT WASTE DISPOSAL
    Disposal Option                  Applicable Regulations


    Stream                  NPDES (CWA)*
                            In-Stream Water Quality Criteria (CWA)
                            Discharge Guidance Documents

    Waste Water Plant       Pretreatment Standards (CWA)

    Landfill                RCRA
                            CERCLA**
                            State SW Requirements (RCRA)
                            Low Level Radioactive Waste Requirements

    Land Application        Sludge Disposal Regulations (CWA)
                            Low Level Radioactive Waste Requirements


  * CWA = Clean Water Act
 ** CERCLA  = Comprehensive Environmental Response, Compensation and Liabil-
             ity Act
 *** RCRA =  Resource Conservation and Recovery Act

    In-stream water quality criteria and standards are developed by indivi-
 dual states (with the use of some federally published guidelines).  Most
 states have classified each body of water for a designated use and set in-
 stream quality guidelines appropriately.  Table 3 shows sample in-stream
 water quality criteria and standards for several selected compounds.
 (Since standards vary from state to state, only examples can be
 illustrated.  The specific agency involved should be contacted.)  These
 quality criteria would apply to solid/liquid waste streams or liquid phase
 waste streams.  In addition to meeting in-stream water quality standards,
 some states have established maximum allowable concentrations in the
 discharge.  These limits generally apply if they are more stringent than
 the allowable discharge that will meet the in-stream water quality cri-
 teria.

    In addition to the compounds in Table 3, criteria will usually be
 established for suspended solids and pH which can affect disposal options.
                                    140

-------
           TABLE 3.   SAMPLE IN-STREAM WATER  QUALITY GUIDELINES AND STANDARDS






Arsenic (Dissolved)
Barium
Beryllium
Cadmium
Chloride
Chromium (hexavalent,
dissolved)
trivalent, active)
(TOTAL)
Copper

Cyanide, free
Fluoride
Hydrogen Sulfide
Iron, total soluble
Lead
Manganese, total
soluble
Mercury
Nickel (total)
Nitrate (as N)
Phenol
Selenium
Silver
Sulfate
TDS
Zinc
Aldrin
Chloride
Endrin
Heptachlor
Lindane
Methoxychlor
Toxaphene
DDT
Chloroform
Radioactivity
R/\226+228

Guidel
Aquatic Life
Chronic Criteria

Fresh ug/1
72

130
el.!6 (ln(hardness))-3.841


7.2
e0.819 (ln(hardness))+.537

2.0

4.2

2.0
1,000
e1.34 (ln(hardness))-5.245


0.00057
e0.76 (ln(hardness))+1.06

1.0
35
.01el'72(1n(hardness))-6.52


47
0.03
0.0043
0.0023
0.0038
0.08
0.03
0.013
0.001
1,240


ines


Salt
yg/i
63


12


54


4(2y)
23(A)
0.57

2.0

8.6
100

0.1
7.1

1.0
54
0.023


58
0.003
0.004
0.0023
0.0036
0.0016
0.03
0.0007
0.001




Human
Health
ID-6 Risk
yg/i
2.2 ng/1

3.7 ng/1
10.0



170



20.0



50


146 ng/1
13.4

3,500
10
50


5,000
0.074 ng/1
0.46 ng/1
1.0
0.28 ng/1

0.71 ng/1
0.024 ng/1
0.19

Sample
Standards
Stream Used
For Potable
Water
mg/1
0.05
1.0

0.01
250



0.05
1.0


1.4

0.3
0.05
0.05

0.002

10
0.001
0.01
0.05
250
500
5.0


0.0002

0.004
0.10
0.005
0.1

5 pCi/1
Gross Alpha Particle
  Activity (excl  radon
    and uranium)
15 pCi/1
                                         141

-------
    Land disposal regulations can apply to landfilling of solid wastes or
land application of solid or liquid phase wastes.  For landfilling of solid
waste, the waste needs to be classified in one of three categories:

    o    Safe for normal landfilling as an industrial  waste;
    o    Classifiable as a hazardous waste;
    o    Contains low level radioactivity.

If a waste does not fit into category two or three, it can be landfilled in
an industrial waste landfill.  Some states will allow the disposal of water
plant wastes in a general sanitary landfill rather than an industrial waste
landfill.  However*, very often in these states, requirements for the
construction of a general sanitary landfill are as stringent as those for
the construction of an industrial waste landfill.  These landfills are
governed by the individual state requirements.  At a minimum, these state
regulations will require that the waste cannot contain any free water
(water that will drain by gravity).  Some states also have specific regula-
tions dealing with water plant wastes.

    In order to classify a water plant waste as hazardous, the term
"hazardous" must be defined.  EPA has developed a definition by stating the
ways  in which a waste can be classified as hazardous:   1) by its presence
on the EPA - developed lists; or 2) by evidence that the waste exhibits
ignitable, corrosive, reactive, or toxic characteristics.  The regulations
governing these definitions and the subsequent handling requirements are
known as the Resource Conservation and Recovery Act of 1976, or RCRA.  RCRA
concerns the handling of wastes at currently operating facilities  (such as
water plants) and at facilities yet to be constructed.  It was designed in
a  large part to meet disposal needs resulting from the Clean Water Act and
the Clean Air Act.  The five major elements of RCRA are:

    o    Federal classification of hazardous wastes;
    o    Cradle  to grave manifest system;
    o    Federal standards to be followed by generators, treaters,
         disposers, and storers of hazardous wastes;
    o    Enforcement of Federal standards;
    o    Authorization of  states to obtain primacy for implementation of
         the regulations.

    So the major question  is, are water plant wastes hazardous  (as per
current EPA definitions)?  Water plant wastes are not on the developed list
of specifically  identified hazardous wastes, so  that part of the definition
does  not apply.  That leaves the properties of ignitability, reactivity,
corrosivity, or  toxicity as  a means of defining  the waste material as
hazard-ous.  It  is highly  unlikely that water plant wastes will fall within
either of  the  first two  criteria.  As applied  to water plant wastes,  corro-
sivity applies  to wastes with a pH less than or  equal to 2  or  greater than
or equal to  12.5.  It is possible  that coagulant recovery side  streams,
perhaps filtrate from lime conditioning of sludge  in a filter  press,  and
brines from acid regeneration of ion exchange  resins would  fall outside
                                    142

-------
these limits.  While it is important to address this if it occurs, the
situation can be handled with appropriate neutralization.

    Toxicity is evaluated by the EP toxicity test.  Basically, the test is
a measure of defined constituents that are present or will leach from the
water plant waste.  For a liquid waste the constituents are measured
directly.  For a solid waste, the waste is held at pH 5.5 for several hours
under defined procedures.  If the liquid or extract from the waste contains
concentrations greater than defined levels, then it is hazardous.  Table 4
shows the currently defined contaminants for the EP toxicity test and their
maximum allowed values.  These values are set at 10 times the drinking
water MCL value.  In essence, failure of the EP toxicity test is currently
the only way a water plant waste could be classified as hazardous.

    Another set of regulations that could affect land disposal of water
plant wastes is the Comprehensive Environmental Response, Compensation and
Liability Act of 1980  (CERCLA).  CERCLA provides authority for the removal
of hazardous substances from improperly constructed or operated sites not
in compliance with RCRA.  The most noteworthy part of these regulations is
that they allow clean  up costs to be assessed against the user of the land
disposal facility based on a volume use basis.  The waste itself need not
have directly caused the problem.  For example, if a water utility disposed
of its sludge at a private landfill that also accepted other industrial
wastes which contaminated the ground water, the water utility can be liable
for clean up based on  its volume use of the landfill, even if its sludge
did not cause the problem.  For this reason it is highly recommended that,
if possible, the utilities use only landfills within their own governing
jurisdiction.

    Water plant wastes containing radium could come under the authority of
three Federal agencies:  the Nuclear Regulatory Commission (NRC), the
Environmental Protection Agency  (EPA), and the Department of Transportation
 (DOT).  However, none  of these agencies directly  regulates this type of
waste.  Currently, the ultimate authority for  regulation of wastes con-
taining  radioactivity  rests with the individual states.

                     IMPACTS OF  1986 SOW ACT ON WASTES

    With the  implementation  of the  1986 SOW Act,  higher  levels of removal
will be  required for inorganic and  organic compounds.  Obviously waste
 streams will  be produced that contain  these compounds.   The answer as  to
whether water plant wastes are hazardous may be impacted.  In the June  16,
 1986 Federal  Register, EPA published a proposed rule  for the  Toxicity
Charcteristic Leaching Procedure  (TCLP) designed  to  replace the  EP Toxicity
 test as  a method  to classify wastes as hazardous.  This  test  increases  to
52  the  number of  compounds which have  threshold levels for toxicity.   Many
of  these compounds, when  removed from  drinking water, will be in  the
 sludge,  on  the  activated carbon  or  in  a  liquid waste  stream.  For example,
exceeding a Teachable  chloroform level of  0.07 mg/1 would  classify a waste
as  hazardous.   Spent GAC,  sludge containing PAC while practicing  prechlo-
 rination, and  some other sludges may certainly exceed TCLP  levels.
                                    143

-------
          TABLE 4.  MAXIMUM CONCENTRATION OF CONTAMINANTS FOR
                      CHARACTERISTIC OF EP TOXICITY
Contaminant
Arsenic
Barium
Cadmi urn
Chromium
Lead
Mercury
Selenium
Silver
Endrin (1,2,3,4,10,10-hexachloro-l,
Maximum
Concen.
(mg/1)
5.0
100.0
1.0
5.0
5.0
0.2
1.0
5.0
0.02
  7-epoxy-l,4,4a,5,6,7,8,8a-octahydro-l,
  4-endo,endo-5,8-dimethano naphthalene)

Lindane (1,2,3,4,5,6-hexachlorocyclohexane,
  gamma isomer)

Methoxychlor (l,l,l-trichloro-2,2bis
  [p-methoxyphenyll] ethane)

Toxaphene (CioHi0Cl8,technical chlorinated
  camphene, 67-69 percent chlorine)

2,4-D (2,4-dichlorophenoxyacetic acid)

2,4,5-TP Silvex (2,4,5-trichlorophenoxpropionic acid)
 0.4


10.0


 0.5


10.0

 1.0
                                     144

-------
    As naturally occurring radioactive compounds (such as radium) are more
completely removed, new regulations will be developed to deal with the
resulting wastes.  Some states have already developed criteria in this
regard.  For example, Wisconsin has set maximum discharge levels of soluble
radium in liquid wastes as follows:

                 Ra -226  +  Ra -228   <  1  pCi/1
                   30          30

These regulations apply to discharge from a water plant to a storm sewer or
to a surface body of water.  They have also limited discharge to a sewer at:

                 Ra -226  +  Ra -228   <  1  pCi/1
                   400         800

    Wisconsin and Illinois have both set regulations regarding landfilling
of solid wastes  (lime sludges) containing radium.  Land application of
lime sludges in  Illinois has been limited to the extent that the incremen-
tal increase of  the radium in the soil cannot exceed 0.1 pCi/g.

             IMPACTS OF WASTES ON MEETING WATER QUALITY GOALS

    A potentially new facet of waste management is the impact that the
wastes have on finished water quality.  These impacts could  involve
releasing of compounds from sludges stored in sedimentation  basins and com-
pounds recycled  along with various sidestream recycles from waste proc-
essing.  In order to meet new finished water turbidity standards an
important operational parameter may be  lower applied turbidity.  Generally
as sludge is stored  in basins, especially with manual cleaning, applied
turbidities will increase with time.  The release of compounds from sludges
in basins is a subject on which very little work has been conducted.  It is
known  that these wastes turn anaerobic  in sedimentation basins, and some
limited work has shown releases of manganese and iron from these wastes.
There  is also evidence that settled water TOC increases as the wastes go
anaerobic.  Much more data are needed  in this area in order  to determine
the total potential  for release of  inorganic and organic compounds.  Figure
1  shows one such controlled study being conducted by the author to help
define sludge storage  impacts.  Two full-scale basins will be evaluated,
one with the sludge  continuously  removed and in the other the sludge will
be allowed to build  up.   Various measurements will be made with time to
evaluate  inorganic and organic compound changes as a  result  of this
storage.

    Many plants  also practice  recycling of backwash water, thickening tank
decant, and even dewatering sidestreams.  These wastes  can contain very
high  iron and manganese levels.   In some cases  significant THM  recycle
occurs along with  the  return  of these  sidestreams.  Again, there  are  no
reported data on the extent of this problem.

    The  effect  of  wastes  on finished water quality  is  indeed a  new area  to
be  considered  in developing waste  handling strategies.   It may  prove  to  be
one of the more  important and  difficult aspects we  have had  to  deal with.
                                     145

-------
     SAMPLE POINTS

     o TURBIDITY
     o PARTICLE COUNT
     o TTHMFP
     o TOC
     o FE
     o Mn
     ° TOTAL METALS
FLOW
    SLUDGE  CONTINUOUSLY
          REMOVED
                             ,--EFFLUENT TROUGHS
                                                        FLOW
SLUDGE BUILD-UP
              Figure  1.  Sedimentation basins.
                            146

-------
                     COST OF  DRINKING  WATER  TREATMENT

                       by:  Richard G.  Eilers
                           U.S.  Environmental  Protection  Agency
                           Drinking Water Research  Division
                           Water Engineering  Research  Laboratory
                           Cincinnati, OH  45268
                               INTRODUCTION

    Over the past 10 years, the Drinking Water Research Division  (DWRD)  of
the U.S. Environmental  Protection Agency (EPA) has conducted various  cost
studies to provide or supplement cost data for water production  in  the
United States.  Under the Safe Drinking Water Act, EPA is responsible for
collecting and making available information pertaining to the demonstra-
tion, construction, and application of acceptable water supply practice.
DWRD's research and development activities are essential  to the  upgrading
of existing systems, planning and design of new systems,  and prediction  of
system performance and cost.  Cost data bases developed and maintained by
DWRD conveniently provide some of this capability through the use of
several computer programs.  The user of these programs can obtain both
construction and operation/maintenance cost estimates for drinking  water
treatment and distribution systems composed of various unit processes
operating under specified design conditions.

                           COST ESTIMATING MODEL

    A data base has been established that can be used to estimate the
capital investment requirements and operational costs for water  treatment
systems composed of individual unit processes as a function of unit process
size or capacity.  This data base is suitable for calculating preliminary
design costs for treatment sizes of 2S500 gallons/day and up.  EPA can use
this cost information to predict the economic impact of proposed water
quality regulations on the water utility industry.

    The cost of new treatment technologies can be investigated based upon
this cost data through the use of sensitivity analysis.  When multiple
treatment solutions exist for achieving a particular water quality goal,
the cost information can be used to identify the least cost alternative,
thereby promoting cost-effective decision making.  Outside of EPA, the cost
data is frequently referred to a being the "standard" as applied to prelim-
inary design applications by consulting engineers, municipal planners,
state agencies, etc.  The cost data base is often used as a teaching tool
in universities, where it is included as part of design and cost engi-
neering courses.  Water utilities have referred to the cost data to aid


                                    147

-------
them in making decisions regarding future expansion  plans,  holding down the
cost of water supply, determining future rate structures,  etc.   Thus it can
be seen that the water supply field has a definite interest in  the cost
data generated through DWRD research efforts.  Since certain decisions
affecting the price that the consumer pays for drinking water can be
influenced by this cost data, it is quite important  to establish and main-
tain the data base in order to reflect the true cost of water supply as
accurately as possible.  Very little other cost information of  a similar
nature is available in a convenient form.

    Unfortunately, as time passes, certain conditions arise that affect the
accuracy and/or applicability of the cost data.  Construction techniques
and building codes may change and thus affect capital investment estimates.
The impact of inflation is not always even, and some costs may  not be
accurately escalated by means of cost indices.  New treatment methods and
concepts are developed over time, and cost data must be obtained to repre-
sent these technologies.  The way in which contractors bid on construction
projects may also vary due to the economic climate,  interest rates,
workload, etc.  Various other factors, both known and unknown,  can alter
the reliability of the original cost data over time.  Therefore, it is
appropriate to occasionally investigate the accuracy and completeness of
the cost data and to make changes and updates where needed.

                     COST AND PERFORMANCE DESIGN MODEL

    The accuracy of  cost estimates for water production can be improved if
performance  (with respect to contaminant removal) is related to the system
cost of producing specific water quality.  With this purpose in mind, an
additional computer model has been developed for use in estimating the per-
formance and associated costs of proposed and existing water supply
systems.  Design procedures and cost estimating relationships for various
unit processes that  can be used for drinking water treatment are contained
within the model.  The unit processes were selected on the basis of their
applicability to the removal of contaminants included in the National
Interim Primary Drinking Water Regulations or to the treatment and disposal
of  sludges and brines produced by these processes.  The computer model can
be  used to calculate the expected contaminant removal performance and asso-
ciated construction  and operation/maintenance costs of drinking water
treatment systems consisting of various unit treatment processes arranged
in multiple configurations.  The technology used in sizing unit processes,
estimating removal efficiencies, and determining treatment cost is the best
that is known to be  currently available for preliminary design.  Since the
technology for each  process is contained in individual subroutines of the
computer model, improving and updating the technology when more information
becomes available can be easily accomplished.  The program structure allows
for the inclusion of additional unit process models if desired.  The
influent raw water quality is characterized by the concentration of 55 con-
taminants and other  parameters entering the treatment system.
                                    148

-------
    The computer model was developed to provide an efficient preliminary
process design tool for the water supply field.  The primary purpose is to
evaluate any proposed system of drinking water treatment processes with
respect to treatment effectiveness and cost with a minimum of engineering
effort.  Technology used in the development of individual  process models is
consistent with the state of the art.  However, in many cases, the need for
additional research is clear.  Updating the technology when improved per-
formance, cost, and design information becomes available can be done by the
user.  Cost estimating data used in the model came from technical litera-
ture, equipment manufacturer information, EPA research projects, etc.  The
computer model approach provides a more flexible preliminary cost estimating
tool than that provided by graphical or tabular cost data.  Cost estimating
procedures or data bases are often presented at a number of specific design
parameter levels, such as pumping heads or hauling distances for sludge,
thus limiting their general applicability.  When a cost estimating algorithm
is known, it can be used in the model in place of mathematical representa-
tions of tabular data, thus providing more accurate cost estimates.  Some
of the design procedures used in the computer program are limited by the
availability of certain cost information.

    This computer model represents a mathematical modeling effort that is a
significant improvement over the hand calculation method of process design
that is  still commonly used today.  The principal deterrents to better
process  design are usually the manual effort required in computing the
expected performance  and cost of alternative designs and the labor required
to accumulate and  correlate the large amount of experimental process design
performance data that is often available.  The computer model can minimize
the  computational work required for examining alternative designs, and
assuming that the model has been correctly developed, it will reflect the
best experimental  and scientific information obtainable.  This model pro-
vides  the process  designer with a tool  for quantitatively selecting the
most cost-effective  system of unit  processes to achieve any drinking water
treatment goal.  The  use of computer design  techniques is a significant aid
in achieving  better  treatment at a minimum cost.

                       WATER DISTRIBUTION COST MODEL

     The  cost  of distributing water  after  it  has been treated  is  of growing
concern  as well as  its quality.  There  are a large  number of distribution
systems  in the U.S.  that are aging  and, as a result, are a potential threat
to the  future quality of water supply;  corrective measures will  greatly
influence the cost of providing water.  Some water  distribution  pipe
systems  date  back  to  the year 1900  or earlier.  Many systems, both large
and  small, are on  the verge of disintegrating  because of age and/or other
physical  factors that influence the  useful life of  the distribution system.
When it  comes time to pay for the replacement  and/or rehabilitation of
these  older systems,  the effects of  inflation  on  cost will be considerable
and  possibly  prohibitive.  This has  not been a major problem  in  the past,
since most distribution systems have held up well enough for a  long time.
                                     149

-------
Now, a frequent decision that utilities face is to determine if it is more
economical to replace or repair a "problem"  area within the distribution
network.  If the problem area is not corrected, water quality will
deteriorate; if corrected, the cost of water supply will  increase.  It
would be quite useful to have a mechanism for examining the economics of
various alternative solutions for correcting a problem (such as corrosion
in distribution pipes) within the supply system.  Since a distribution
system is made up of various components (such as pipes, pumps, storage,
etc.) it would be necessary to have cost information on each subelement in
order to estimate a realistic cost for replacing or upgrading a water
distribution system.  Proposed and future Federal  regulations may require
additional performance demands, in addition  to merely maintaining the pre-
sent level of water service and quality provided by the utilities.

    The intent of a current research effort  funded by DWRD is to provide a
data base of cost information, and associated computer program for accessing
the data, that can be used to estimate the costs associated with the
various physical components that comprise a  water distribution system.  The
cost data would include estimates of capital  investment necessary to
upgrade an existing system by means of new construction,  expansion of
existing facilities, or rehabilitation of all or part of  the system.
Operation/maintenance cost data for estimating the ongoing expenses asso-
ciated with continuous water distribution will also be generated.  The cost
estimates from this project will hopefully serve as a guideline to further
research and development efforts by identifying those areas which strongly
influence the final cost of water delivery to the consumer.
                                    150

-------
                 REGULATIONS ON FILTRATION AND DISINFECTION

                       by:  Stig Regli
                            Environmental  Engineer
                            U.S. Environmental Protection Agency
                            Washington, D.C.  20460


    The 1986 Amendments to the Safe Drinking  Water Act (SDWA)  require EPA
to promulgate primary drinking water regulations:   a)  specifying criteria
under which filtration would be required;  b)  requiring disinfection as a
treatment technique for all public water systems;  and  c)  establishing
Maximum Contaminant Levels (MCLs) or treatment technique  requirements for
Giardia JambJia, viruses, LegioneJla, heterotrophic plate count bacteria,
and turbidity.  Shortly, EPA will propose surface  water treatment technique
requirements (SWTR) to fulfill the SDWA requirements for  systems using sur-
face waters.  EPA intends to propose and promulgate additional regulations
at a later date specifying disinfection requirements for  systems using
ground water sources.

    Under the proposed surface water treatment requirements, all community
and noncommunity public water systems would be required to treat their sur-
face water sources to control Giardia  Iambiia, enteric viruses, and
pathogenic bacteria.  The minimum required treatment for such surface water
would include disinfection.  In addition, unless the source water is well
protected and meets certain water quality criteria (total or fecal coli-
forms and turbidity limits), required  treatment would also include filtra-
tion.  The treatment provided,  in any  case, would be required to achieve at
least 99.9 percent removal and/or inactivation of Giardia cysts, and at
least 99.99 percent removal and/or  inactivation of enteric viruses.
Systems that met certain  turbidity  removal and disinfection performance
criteria, and which complied with design and  operating criteria specified
by the state, would be considered to be in compliance with these require-
ments.  This paper will focus on the performance criteria (i.e., turbidity
requirements) and disinfection  of the  SWTR sources which EPA  shortly
intends to propose.


               TURBIDITY  REQUIREMENTS  FOR UNFILTERED SYSTEMS

    To avoid filtration,  a system would be required to demonstrate that  the
turbidity of the water prior to disinfection  does not exceed  5 Nephelo-
metric Turbidity Units (NTU), based on the collection of grab samples taken
at least every four hours.  Continuous turbidity monitoring could be
                                    151

-------
substituted for grab sample monitoring if the continuous turbidity
measurements were validated for accuracy by regular grab sample measure-
ments in accordance with a protocol approved by the state.  If the public
water system uses continuous monitoring, the system would depend on tur-
bidity values taken every four hours to determine whether it met the tur-
bidity raw water limit.  A system would be allowed to exceed the 5 NTU
limit no more than two periods during twelve consecutive months, or five
periods during 120 consecutive months, provided: 1) that the system
informed its customers to boil their water before consumption during the
period the turbidity exceeds 5 NTU, and 2) that the state determined the
exceedance occurred due to unusual or unpredictable circumstances.  A
"period" would be defined as the number of consecutive days during which at
least one turbidity measurement each day exceeded 5 NTU.

    The proposed raw water turbidity requirement is related to the existing
turbidity MCL, which has been in effect since 1977.  Under the existing
MCL, a system is in violation if the turbidity of the water, at a
representative entry point to the distribution system, exceeds 1 NTU, as
determined by a monthly average (based on at least one sample per day), or
if the average turbidity for two consecutive days exceeds 5 NTU.  Under the
existing MCL the monthly average limit of 1 NTU may be exceeded up to 5 NTU
if the system demonstrates to the state that the higher turbidity does not:
1) interfere with disinfection; 2) prevent maintenance of a disinfectant
residual throughout the distribution system; or 3) interfere with micro-
biological determinations.

    EPA has not proposed an average monthly limit of 1 NTU in accordance
with the conditions of the existing turbidity MCL because:

    1)  The proposed rule would require systems to filter if they fail to
comply with the proposed long-term MCL for total coliforms (a revised total
coliform MCL with monthly and long-term requirements will be proposed at
the same time as the SWTR).  Under the proposed total coliform MCL, systems
would be required to monitor throughout the distribution system and
demonstrate that no more than five percent of the samples in the last 12
months are coliform-positive.  If there is evidence that the number of
heterotrophic bacteria is high enough to interfere with the coliform analy-
sis, repeat samples would be required, for determining both the presence or
absence of coliforms and the number of heterotrophic bacteria present.  If
the heterotrophic bacteria level exceeds 500 colonies/ml, the system would
be required to report that repeat sample as coliform-positive, even in the
absence of detectable coliforms.  In addition, for surface water systems
which do not filter, the proposed rule would require samples for coliforms
to be taken near the first customer each day the turbidity exceeds 1 NTU.
These measurements would be counted in determining whether the system is in
compliance with the total coliform MCL.

    2)  Under the proposed rule, systems would be required to filter unless
they met the following disinfection conditions:
                                    152

-------
        a)  Maintenance of disinfection conditions, determined each day,
that theoretically (with some margin of safety) achieve 99.9 percent inac-
tivation of Giardia cysts and 99.99 percent inactivation of enteric viru-
ses.

        b)  Maintenance of a disinfectant residual, of at least 0.2 mg/1,
both at the point of entry into the distribution system and throughout the
distribution system.

    EPA believes that the proposed raw water turbidity upper limit of 5
NTU, in conjunction with the other requirements of this rule, would provide
a greater margin of safety than the requirements of the existing turbidity
MCL for ensuring that raw water quality will not significantly interfere
with disinfection of Giardia cysts, bacteria, and enteric viruses.  Since
significant fluctuations in turbidity levels can occur during a 24-hour
period, the proposed rule would require more frequent monitoring than does
the current MCL for turbidity to ensure more representative turbidity
measurements.  Increases in turbidity levels from less than 1 NTU to
greater than 5 to 10 NTUs have been shown to correlate with decreases in
disinfection effectiveness in unfiltered source waters (1).  In addition,
high turbidity waters may be unaesthetic in appearance and cause consumers
to  avoid  use of the public water supply and possibly to choose less safe
sources.   Exceedances to the 5 NTU limit are allowed for a limited number
of  unusual and unpredictable circumstances such as avalanche, hurricane, or
10-year flood.  EPA believes that the boiled water notice required to be
issued at such times would prevent exposure to acute risks.

    The proposed turbidity limit for systems that do not filter (i.e., only
practice  disinfection)  is  less stringent than  the turbidity limits proposed
for systems that filter because:

    1)    The requirements  that systems which do not filter meet the raw
water  fecal or total coliform  limits, and maintain a watershed control
program to restrict human activities, ensure very high probabilities of
minimal or zero occurrence of  human viruses in the source water.  Although
watershed control will  not eliminate animal activity, no viruses excreted
by  animals have been shown to  be pathogenic to humans.

    2)  Giardia  lambJia cysts  are relatively large organisms compared to
bacteria  and viruses, so interference with their removal and/or
inactivation by turbidity  levels below 5 NTU is unlikely.


                TURBIDITY REQUIREMENTS FOR FILTERED SYSTEMS

    Under the proposed  rule, systems that used conventional treatment,
direct filtration, or diatomaceous earth filtration would be required to
monitor the turbidity of the representative filtered water by grab sample
taken every four hours  (or at  regular shorter  time intervals), when water
is  being  delivered to the distribution system. Similar to the requirements
                                    153

-------
for unfiltered systems, a system could substitute continuous turbidity
monitoring for grab sampling if it has validated this measurement for
accuracy with grab sample measurements on a regular basis, as specified by
the state.  If a system uses continuous monitoring, it would be required to
utilize the turbidity value for every four hours (or some shorter regular
time interval) to determine compliance with the turbidity performance cri-
terion.

    For those systems that use slow sand filtration and technologies other
than conventional treatment, direct filtration, or diatomaceous earth
filtration (e.g., cartridge filtration), if the state determined that the
aforementioned frequency is not necessary to indicate effective filtration
performance, such sampling could be reduced by the state to once per day.

    Turbidity performance criteria would vary depending upon the filtration
technology in place.  For systems using conventional treatment or direct
filtration, the proposed rule would require that filtered water turbidity
be less than or equal to 0.5 NTU in 95 percent of the measurements taken
every month.  If the state determined that on-site studies demonstrate
effective removal and/or inactivation of Giardia lamblla cysts, or effec-
tive removal of Giardia  lamblia cyst-sized particles at other filtered
water turbidity levels, the state could then specify that these levels
replace the usual performance criteria.  This provision would allow the
state to take disinfection performance into account in determining the
overall performance by the system.  For example, the state could allow less
stringent turbidity performance criteria for systems using ozonation that
achieve 99.9 percent inactivation of Giardia cysts  (and therefore much
greater than 99.99 percent inactivation of viruses).  However, the proposed
rule would require that, in all cases, the maximum filtered water turbidity
level must be less than or equal to 1 NTU in 95 percent of the measurements
taken each month and must at no time exceed 5 NTU.  All systems would be
expected to optimize their treatment so as to achieve the lowest tur-
bidities feasible at all times.  This would promote optimal removal of
Giardia cysts and other pathogens, and thus create optimum conditions for
disinfection.

    For systems using  slow sand filtration, the proposed  rule would require
that the filtered water turbidity be less than or equal to 1 NTU in 95 per-
cent of the measurements taken each month and at no time  exceed 5 NTU.
However, the state could allow a turbidity value greater  than  1 NTU, but
below 5 NTU, in 95 percent of the measurements if the filter effluent at
the plant prior to disinfection met the long-term MCL (to be proposed) for
total coliforms for  one year.

    For systems using  diatomaceous earth filtration, the  filtered water
turbidity would have to be less than or equal to 1  NTU  in 95 percent of  the
measurements taken each month and at no time exceed 5 NTU.  In systems
using other filtration technologies, the performance criteria would be the
same as for conventional treatment and direct filtration.  The state could
allow a turbidity value greater than 0.5 NTU in 95  percent of  the measure-
                                    154

-------
ments, but at no time exceeding 5 NTU, in the system were able to
demonstrate to the state effective performance at such levels.

    The proposed turbidity performance criteria for systems that filter are
more stringent than those of the existing MCL.  EPA has concluded that the
existing MCL turbidity criteria are not adequate performance criteria for
filtered systems because:

    1)  High turbidity levels can occur frequently in finished water when
passage of pathogens  is most likely to occur  (e.g., during storm events, at
the end of filter  runs, and following backwash cycles), yet the system
could still be in  compliance with the current MCL.  Continuous effective
filtration, demonstrated by continuous effective turbidity removal, is
essential for effective pathogen control.

    2)  Systems  using conventional treatment  and direct filtration can
easily meet the  current MCL while not optimizing coagulation and floc-
culation processes.   Effective pretreatment to filtration is essential for
effective virus  removal  (2), and Giardia  cyst removal (3, 4, 5).  Giardla
cysts have frequently been detected in finished waters of systems using
rapid granular filtration  (direct filtration  and conventional treatment)
that  have inadequate  pretreatment (6).

    Good correlations between turbidity removal and Giardia cyst removal
have  been demonstrated in  pilot plant studies (7, 5).  Although finished
water turbidity  goals of 0.1 NTU have long been advocated within the
drinking water industry, many systems have not taken the initiative to
optimize turbidity removal, despite the fact  that such treatment improve-
ments have relatively low  associated costs (8).

    The purpose  of the turbidity performance  criterion for conventional
treatment and direct  filtration is to ensure  that public water systems pro-
vide  adequate pretreatment to ensure effective Giardia cyst removal.  EPA
believes the proposed performance criterion of less than or equal to 0.5
NTU 95 percent of  the time, is the lowest turbidity level that is generally
achievable by these technologies.  The National Drinking Water Advisory
Council supports these criteria as being  achievable (9).  It  is recognized
that  the proposed  performance criterion may not be adequate for a system
whose source waters have a turbidity of less  than 1 NTU  (5).  Therefore,  in
such  cases,  it would  be  expected that the state would set more stringent
turbidity performance criteria as appropriate to the circumstances.

    For  removal  of Giardia cysts, the turbidity of the filtered water  in
systems  using diatomaceous earth and  slow sand filtration has been  shown  to
be relatively  less important, as  long as  the  mechanical  integrity of  the
filter  is preserved.   Since no  relationship between turbidity removal  and
Giardia  cyst  removal  has been demonstrated for diatomaceous  earth and  slow
sand  filtration  systems, the  proposed turbidity performance  criteria  are
higher  than  for  conventional  treatment  or direct  filtration.
                                     155

-------
    When diatomaceous earth filtration  is  used,  the  relationship  between
the turbidity and microbiological  quality  of the water produced depends on
the nature of the particles causing turbidity and the  microorganisms pres-
ent.  If the diatomaceous earth is not  treated with  a  polymer or  with salts
of aluminum or iron,  the removal  mechanism is straining;  raw water coagula-
tion is generally not practiced in diatomaceous  earth  filtration.
Turbidity removal increases as finer grades (smaller particle sizes) of
diatomaceous earth are used, but Giardia cyst removal  has been shown to be
very effective for all grades tested (10).  If turbidity-causing  particles
are very small, they can penetrate the  filter even when cysts are removed.

    Studies of slow sand filtration have shown that  this process  is very
effective for Giardia cyst removal.  Pilot plant studies (11, 12)
have demonstrated that cyst reductions  were almost always greater than 99.9
percent, even though turbidity removal  generally was only from 6  to 8 NTU
(raw) to 3 to 5 NTU (filtered).  The existing MCL of 1 NTU was seldom, if
ever, met in water treated by the slow  sand filtration.  The turbidity-
causing particles appeared to be fine clay.  Other slow sand filter
research  (13) indicates that slow sand  filter can effectively remove both
turbidity and microorganisms.  Turbidity removal effectiveness appeared to
be  influenced by the quantity of nutrients in the water; waters that are
low in  nutrients may not be as treatable with respect to turbidity removal.

    The upper turbidity limit of less than or equal  to 1 NTU in 95 percent
of  the  turbidity measurements for all the filtration technologies is to
ensure  a  high probability that there is no significant interference with
disinfection.  Slow sand filters can substantially reduce concentrations of
viruses,  bacteria, and protozoan cysts  in water, and tend to attain the
microbiological water quality achieved  by disinfection.  If substantial
reductions of microorganisms are attained by slow sand filters, disinfec-
tion need not be as stringent.  Therefore, under the proposed rule, water
treated by slow  sand filters could have turbidity above 1 NTU (up to 5
NTU), at  the state's discretion, if the system demonstrates that  the filter
effluent, prior  to disinfection, meets  the proposed  long-term MCL for total
coliforms for one year.


             DISINFECTION REQUIREMENTS  FOR UNFILTERED SYSTEMS

    Under the proposed rule, unfiltered systems would be required to
demonstrate by monitoring the disinfection operational conditions, that
inactivation of at least 99.9 percent of Giardia cysts and 99.99  percent of
enteric viruses  is being achieved at all times of the year.  Since
Giardia cysts are much more resistant to free chlorine, ozone, and chlorine
dioxide than are enteric viruses (14, 15, 16, 17),  it could be assumed
that if a system achieves 99.9 percent  inactivation  of Giardia cysts using
these disinfectants it will achieve much greater than a 99.99 percent
inacti-vation of enteric viruses.  To demonstrate that the system is
achieving the required percent inactivation, they would monitor and report
the disinfectant(s) used, disinfectant residual(s),  disinfectant  contact
                                    156

-------
time(s), pH, and water temperature, and apply these data to determine if
its "CT" product (the product of multiplying the disinfectant concentration
[mg/1] and disinfectant contact time [minutes]) equalled or exceeded the
product specified in the rule.  These determinations would be required each
day that the system is delivering water to its customers.  The CT products
necessary to achieve 99.9 percent inactivation of Giardia cysts (CTgg.g) by
various disinfectants and under various conditions are specified in the
proposed rule.  An example of some of these products appears in Table 1.
The basis for these products, which include safety factors applied to
laboratory data, is discussed in a later section of this paper.

    The CT products given for chloramines in Table 1 were determined under
laboratory conditions in which no chlorine was present, i.e., the chloram-
ines were preformed.  Systems using chloramines would probably not able to
provide these CT products.  Under field conditions, chloramination as a
treatment process involves the addition of free chlorine and ammonia either
concurrently, or sequentially, the order of addition and timing between
adding each component being determined by needs of the utility.  Regardless
of the process used, chloramination, as conducted in the field, is more
effective than using preformed chloramines.  The relative effectiveness
will be influence by the order of addition, the chlorine to ammonia ratio
and water pH, and temperature.

    The proposed rule would allow utilities using chloramines to conduct
studies to determine if  lower CT products than those indicated in the rule
would achieve the required inactivation of Giardia.  Such studies require a
high  level of expertise  to carry out, and it may be necessary for utilities
using chloramination to  hire  specialized independent (commercial) labora-
tories or university researchers to make such determinations.

    For the purpose of calculating CT products, disinfection contact time
is the time it  takes the water to move between the point of disinfectant
application and  the first customer under peak hourly flow conditions.
Residual disinfectant concentration  is the concentration of the disinfec-
tant  at the first customer where contact time  is determined.  Contact time
in pipelines must be calculated based on "plug flow" (i.e., where all water
moves homogeneously in time between  two points) by dividing the internal
volume of the pipeline by the peak hourly flow rate through that pipeline.
Contact time within mixing basins and storage  reservoirs must be determined
by tracer studies or an  equivalent demonstration.

    If disinfectants are applied at more than one point, the percent inac-
tivation of each disinfectant sequence prior to the first customer could be
considered as part of the determination.  The disinfectant residual of  each
disinfection sequence and corresponding contact time would be measured
before subsequent disinfection application point(s) to determine the per-
cent  inactivation for each sequence and the total percent inactivation
achieved.  The following recursive formula would be used for making this
determination:
                                   157

-------
    TABLE  1.  CT  PRODUCTS  FOR ACHIEVING 99.9  PERCENT  INACTIVATION  OF
                             GIARDIA  LAMB LIA
Temperature

Free Chlorine*



Ozone
Chlorine Dioxide
Chloramines
PH
6
7
8
9
6-9
6-9
6-9
0.5°C
170
260
380
520
4.5
81
3,800
5°C
120
190
270
370
3
54
2,200
10°C
90
130
190
260
2.3
40
1,850
15°C
60
100
140
190
1.5
27
1,500
*CT products will  vary depending  on  concentration  of  free  chlorine.
 Values indicated  are for 2.0 mg/1  free  chlorine.   CT products for dif-
 ferent free chlorine concentrations are specified in tables  in the  pro-
 posed rule.
                                   158

-------
               Gtn - Gtn-l + Gn (10° - Gtn-l)                  Equation (1)
                                     100

where:        n = number of points of disinfection application

                = total percent inactivation achieved by n disinfectants
          Gtn-l = percent inactivation for the disinfection sequence(s)
                  prior to the nth disinfection sequence

             Gn = percent inactivation achieved by nth disinfectant

    If a system achieved 99 percent inactivation by the first disinfection
sequence (Gtn-l = 99) and 90 percent inactivation by the second disinfec-
tion sequence  (Gn = 90), the combined percent inactivation would be deter-
mined as follows:


    Gtn = 99 + 9°(100"") = 99.9 percent inactivation          Equation (2)
                  100

    Also, under the proposed rule, systems would be required to demonstrate
by  continuous  monitoring that a disinfectant residual is maintained in the
water entering the distribution system and throughout the distribution.  A
disinfectant residual of at least 0.2 mg/1 would need to be maintained at
all times in the water entering the distribution system, and disinfectant
residuals could not be allowed to be less than 0.2 mg/1 at any location in
the system  in  more than five percent of the samples in a month, for any two
consecutive months, on an ongoing basis.  The public water system would be
required to monitor the disinfectant residual at the same frequency and
locations for  which total coliform measurements are taken pursuant to the
proposed coliform MCL regulation.


               DISINFECTION REQUIREMENTS FOR FILTERED SYSTEMS

    The disinfection requirements for filtered systems are the same as
those stated in the last paragraph for unfiltered systems.  Also, under the
proposed rule, systems would be required to disinfect, to some level, in
accordance with criteria specified by the state, to ensure that overall
removal and/or inactivation of at least 99.9 percent of Giardia cysts and
at  least 99.99 percent of enteric viruses is achieved.

    Filtration without disinfection, with proper pretreatment where
appropriate, can be expected to achieve 99 to 99.9 percent removal of
Giardia cysts  and 90 to 99.9 percent removal of viruses (18).  Disinfection
is  needed to supplement filtration so that the overall treatment achieves
greater than 99.9 percent removal and/or inactivation of Giardia cysts and
99.99 percent  removal and/or inactivation of viruses.
                                     159

-------
    The level  of disinfection effectiveness should be commensurate with the
degree of potential  pathogen contamination in the source water and the
extent and type of clarification processes in place.   In general,  as a
minimum, systems with filtration, and relatively clean source waters,
should be designed and operated so that disinfection  achieves at least a 90
percent inactivation of Giardia cysts and a 99.9 percent inactivation of
enteric viruses.  This would assure with high probabilities of confidence,
that the minimum overall  performance requirements, by filtration and disin-
fection, of 99.9 percent removal/inactivation of Giardia and 99.99 percent
removal/inactivation of viruses are being achieved.  More stringent disin-
fectant (e.g., 99 or 99.9 percent inactivation of Giardia cysts) should be
provided when source waters are contaminated with sewage.

     CT products necessary to achieve 90 percent inactivation of Giardia
Iambiia cysts (CT90) are indicated in Table 2.  The basis for these prod-
ucts are discussed in the following section of this paper.  With the
exception of chloramines, these conditions will achieve greater than a
99.99 percent inactivation of enteric viruses (for chloramines, higher CT
products than those indicated in Table 2 might be needed).  Table 3 indi-
cates the relative sensitivity of different microorganisms to different
disinfectants.

    Guidelines for defining "C" and "T" within the treatment plant are pro-
vided elsewhere  (19).  If multiple disinfectants are used, e.g., ozone
followed by chloramines or chlorine, the percent inactivation achieved by
each of the disinfectants is additive, in accordance with Equation 1,
previously discussed, and would apply in determining the overall disinfec-
tion performance provided.
                            BASIS FOR CT VALUES

    The basis for the CT products in the proposed rule are discussed for
each disinfectant below.
FREE CHLORINE

    The CTgg.g products for free chlorine in Table 1 are based on animal
infectivity data by Hibler et al. (15) and regression analysis of
Hibler's data by Clark et al. (20).  As a safety factor, CTgg g products
are defined as those needed to achieve 99.99 percent inactivation under
experimental conditions.  If this safety factor were not applied, the
CT99.9 products in Table 1 would be about 25 percent lower.

    Hibler's data were developed at temperatures of 0.5°C to 5°C. pH levels
from 6 to 8, and free chlorine concentrations between 0.44 mg/1 to 4.23
mg/1.  Clark's model equation:
                                    160

-------
        TABLE  2.   CT  PRODUCTS  FOR ACHIEVING  90  PERCENT  INACTIVATION
                              GIARDIA LAMBLIA
Temperature

Free Chlorine*



Ozone
Chlorine Dioxide
Chloramines
PH
6
7
8
9
6-9
6-9
6-9
0.5°C
60
90
130
170
1.5
27
1,270
5°C
40
60
90
120
1
17
730
10°C
30
40
60
90
0.8
13
620
15°C
20
30
50
60
0.5
9
500
*CT products will  vary depending on concentration of free chlorine.
 Values indicated are for 2.0 mg/1  free chlorine,  (for other free
 chlorine concentrations, see Reference 19).
   TABLE 3.  SUMMARY OF CT PRODUCT RANGES FOR 99 PERCENT INACTIVATION OF
            VARIOUS MICROORGANISMS BY DISINFECTANTS AT 5°C (14)
Disinfectant
Micro-
organism
E. coll
Polio 1
virus
Rotarvis
Phage f2
G. lamblia
cysts
G. muris
cysts
Free
Chlorine
pH 6 to 7
0.034-0.05
1.1-2.5
0.01-0.05
0.08-0.18
47->150
30-630
Preformed
Chloramine
pH 8 to 9
95-180
768-3,740
3,806-6,476
-
-
-
Chlorine
Dioxode
pH 6 to 7
0.4-0.75
0.2-6.7
0.2-2.1
-
-
7.2-18.5
Ozone
pH 6 to 7
0.02
0.1-0.2
0.006-0.06
-
0.5-0.6
1.8-2.0
                                    161

-------
                 CT = 0.9847 C°-1758 pH2.7519 temp -0.1467     Equation (3)

was applied to generate CT products for chlorine concentrations from 0.4
mg/1 to 3.0 mg/1, pH concentrations from 6 to 9, and temperatures from
0.5°C to 5°C.

    CT products for temperatures above 5°C were estimated assuming a two-
fold decrease for every 10°C.  CT products for temperatures at 0.5°C were
estimated assuming a 1.5 times increase to CT products at 5°C.  This
general principle is supported by Hoff (14).

    Application of Clark's model to pHs above 8, up to 9, was considered
reasonable because the model is substantially sensitive to pH (e.g., CTs at
pH 9 are over three times greater than CTs at pH 6 and over two times
greater than CTs at pH 7).  At a pH of 9 about four percent of free
chlorine is still present as hypochlorous acid (HOC1).  Recent data indi-
cate that in terms of only the HOC1 residual  (versus total free chlorine
residuals which include both HOC1 and -QC1),  the CT products required for
inactivation of Giardia muris and Giardia Iambiia cysts decrease with
increasing pH from 7 to 9 (21).  However, with increasing pH, the fraction
of  free chlorine existing as the weaker oxidant species (OC1~) increases.
In  terms of total free chlorine residuals (i.e., HOC1 and ~OC1) the CT
products required for inactivation of Giardia muris cysts increase with
increasing pH from 7 to 9 but less than a factor of 2 at concentrations of
less than 5.0 mg/1 (see Table 4; Reference 21).  Also, the significance of
pH  on  the product of CT products achieving 99 percent inactivation appears
to  decrease with decreasing temperature and free chlorine concentration.
The  relative effects of pH, temperature, and chlorine concentration, upon
inactivation of Giardia muris cysts appears to be the same for Giardia
 lamblia cysts  (21,22), although not as much data for Giardia  lamblia cysts
as  for Giardia muris cysts  is yet available for high pH and temperature
values.

     The CT products for free chlorine in Table 2 were determined by extra-
polation of CT 99.9 products using first order kinetics  (i.e., CT90 = 1/3
x CT 99.9).  This extrapolation appears reasonable based on comparison with
CTgo products determined by Jarroll et aJ. (23).

     Table 5 compares CTgo products obtained by Jarroll with the CTgo Pro~
ducts  of Table 2.  In comparing these values from Hibler's CTgg.gg data,
included the safety factor already mentioned, whereas the CTgo products
from Jarroll's data did not include a safety factor.


OZONE  AND CHLORINE DIOXIDE

    The CT products for ozone in Tables 1 and 2 were based on disinfection
studies using  in vitro excystation of Giardia  lamblia (24).  CTgg products
at 5°C and pH 7 for ozone ranged from 0.46 to 0.64 (disinfectant con-
centrations ranged from 0.11 to 0.48 mg/1).  No CT products were available
                                    162

-------
TABLE 4.  CT PRODUCTS TO ACHIEVE 99 PERCENT INACTIVATION OF GIARDIA MURIS
                       CYSTS BY FREE CHLORINE  (20)
Concentration (mg/1)
Temperature
PH
7

8

9

(°C)
1
15
1
15
1
15
0.2-0.5
500
200
510

440
310
0.5-1.0
760
290
820
220
1,100
420
1.0-2.0
1,460
360
1,580

1,300
620
2.0-5.0
1,200
290
1,300
320
2,200
760
      TABLE 5.   CT PRODUCTS FOR ACHIEVING 90  PERCENT  INACTIVATION  OF
                          GIARDIA LAMBLIA AT  5°C
pH Jarroll et al. (23)
Free Chlorine


6
7
8
20
30
60
Hibler et a/. (15)
40
60
90
*Application of Hibler's Data to Clark's Model,  with  safety factor of
 CTgg g = CTgg gg and extrapolation using first  order kinetics  (CTgg  =
 1/3 x CTgg.g)!
                                   163

-------
for other pHs.   The highest CTgg  product,  0.64, multiplied  by  a  safety fac-
tor of 3, was used as the basis for extrapolation,  using  first order kine-
tics, to obtain the CTgo and CTgg.g products  at 5°C in  Tables  1  and 2.  For
example:

    CTgg.g = CTgg x 3 x | or (.064 x 3 x | =  2.88)  = 3         Equation (4)


CT products at 0.5°C and 15°C were estimated, based on  the  same  rule of
thumb multipliers assumed for free chlorine,  as already discussed.

    The CT products for chlorine  dioxide in Tables  1 and 2  were  based on
disinfection studies using In vitro excystation of  Giardia  muris cysts
(25).  CTgg products at 5°C and pH 7 ranged from  7  to 18 (disinfectant con-
centrations ranging from 0.1 to 5 mg/1).  The highest CT product, 18, was
used as a basis for extrapolation to obtain the CTgo and CT99.9  products in
Tables 1 and 2, applying the same principles  as discussed for  ozone.

    A much larger safety factor was applied to the  ozone and chlorine
dioxide data than to the chlorine data because:

    1)   Less data were available for ozone and chlorine dioxide than for
         chlorine;

    2)   Data available for ozone and chlorine dioxide, because  of  the
         limitations of the excystation procedure,  only reflected up to or
         slightly beyond 99 percent inactivation.   Data for chlorine, based
         on animal infectivity studies rather than  excystation procedures,
         reflected inactivation of 99.99 percent.   Extrapolation of data to
         achieve CT products for  99.9 percent inactivation  with  ozone and
         chlorine dioxide, versus the direct  determination  of  CT products
         for achieving 99.99 percent inactivation using chlorine, involved
         greater uncertainty;

    3)   The CT products for ozone and chlorine dioxide to  achieve  99.9
         percent inactivation are feasible to achieve;  and

    4)   Use of ozone and chlorine dioxide is likely to occur  within the
         plant rather than in the distribution system (versus  chlorine and
         chloramines which are the likely  disinfectants for use  in  the
         distribution system). Measurement of contact  time within  the
         plant will involve greater uncertainty than measurement of contact
         time in pipelines.


CHLORAMINES

    The CT products for chloramines in Tables 1 and 2 were  based on disin-
fection studies using preformed chloramines and  in  vitro excystation of
Giardia muris (26).  Table 6 summarizes CT products for achieving 99 per-
                                    164

-------
cent inactivation of Glardia muris cysts.  The highest CTgg products at 1°C
(2,500) and 5°C (1,430) were each multiplied by 1.5 to estimate the CT 99.9
products at 0.5°C and 5°C, respectively, in Table 1.  The CTgg product of
970 at 15°C was multiplied by 1.5 to estimate the CTgg.g product.  The
highest CTgg product of 1,500 at 15°C and pH 6 was not'used because it
appeared anomalous to the other data.  Interesting to note is that among
the data of Table 6, the CT products in the lower residual concentration
range (<2 mg/1) are higher than those in the higher residual  concentration
range (2 to 10 mg/1).  This is opposite to the relationship between these
variables which exists for free chlorine, indicating that for chloramines,
within residual concentration practiced by water utilities (less than 10
mg/1), residual concentration may have greater influence than contact time
on the inactivation of Glardia cysts.  No safety factor was applied to
these data since chloramination, conducted in the field, is more effective
than using performed chloramines.  Also, Giardia muris cysts appear to be
more resistant than Giardia Iambiia cysts to chloramines (26).


              CT PRODUCTS FOR INACTIVATION OF ENTERIC VIRUSES

    CT products for achieving greater than a 99.99 percent inactivation of
enteric viruses for free chlorine, chlorine dioxide and ozone are indicated
in Table 7.  These values are based on CT products which would be expected
to achieve 0.5 log (68 percent) inactivation of Giardia lamblia cysts
(determined by multiplying CTgg products by 1/5).  Since all  unfiltered
systems must achieve at least a 99.9 percent inactivation of Giardia cysts,
and since filtered systems are recommended to achieve at least a 90 percent
inactivation (in some cases where source waters are clean and there is
conflict with controlling trihalomethanes it might be appropriate to allow
for lesser inactivation), CT products to achieve lower enteric virus inac-
tivation are not provided.  The literature supports that the CT products
provided herein achieve substantially greater than a 4-log inactivation for
enteric viruses (generally by a factor of greater than 3), for which data
exists (14, 26, 16, 17).  One exception to this is for Coxsackie B-5 virus
(27, 17); but this virus has never been associated with a waterborne
disease.

    CT products for achieving a 99.99 percent inactivation of enteric viru-
ses by chloramines are not yet available.  The literature indicates that CT
products of greater than 5,000, using preformed chloramines, are needed to
achieve at least 99 percent inactivation of rotaviruses (14).  Systems
using chloramination for primary disinfection would need to conduct studies
using seeded indicator organisms to demonstrate if adequate virus inac-
tivation is being achieved.


                                  SUMMARY

    EPA will soon propose surface water  treatment requirements to regulate
for Giardia  JambJia, viruses, Legionella, heterotrophic plate count bac-
                                    165

-------
   TABLE 6.  CT PRODUCTS FOR 99 PERCENT INACTIVATION OF GIARDIA MURIS
                      CYSTS BY MONOCHLORAMINE* (25)
Temperature
PH (°C)
6 15
5
1
7 15
5
1
8 15
5
1
9 15
5
1
Monochloramine Concentration (mg/1)
<0.2 2.0-10.0
1,500
>1,500
>1,500
>970
>970
2,500
1,000
>1,000
>1,000
890
>890
>890
880
>880
>880
970
1,400
>1,400
530
1,430
1,880
560
>560
>560
*CT products with ">" signs are extrapolated from the known data.

   TABLE 7.  CT PRODUCTS FOR GREATER THAN 99.99 PERCENT INACTIVATION OF
                              ENTERIC VIRUSES
Temperature
PH
Free Chlorine
6.0
6.5
7.0
7.5
8.0
8.5
9.0
Ozone
0.5
31
38
47
56
67
80
93

5
22
27
33
40
48
57
66

10
16
20
25
30
36
43
50

15
11
14
17
20
24
28
33

20
8
10
12
15
19
21
25

25
5
7
8
10
12
14
17

6-9                 0.8

Chlorine Dioxide

6-9	           13
0.5      0.4
0.3
                   4.5
0.25
0.2
                                   166

-------
teria, and turbidity.  These requirements include turbidity and disinfec-
tion criteria which all public water supplies using surface water sources
would be required to meet.  The turbidity and disinfection criteria, and
their basis, are discussed as they would apply to filtered and unfiltered
systems.  The CT concept is proposed as a means for regulating different
levels of disinfection that might be required.  The basis for the CT prod-
ucts and the associated safety factors in the proposed rule are also
discussed.
                              ACKNOWLEDGMENTS

    The author would like to acknowledge the following:  Dr. Robert Clark,
Dr. John Hoff, and Dr. Gary Logsdon for their helpful suggestions; Dr. Alan
Rubin for his suggestions and for allowing the release of CT data per-
taining to chloramines in this manuscript, soon to be published elsewhere.
                                   167

-------
                               REFERENCES

 1.  LeChevalier, M., Evans, T. and Seidler, R.  Effect of turbidity on
    chlorination efficiency and bacterial persistence in drinking water.
    Appl.  Envir. Microbiology.  42:159-167, 1981.

 2.  Robeck.  G. G., Clarke, N. A. and Postal, K. A.  Effectiveness of water
    treatment  in virus  removal.  Journal American Water Works Association.
    54:1275-1292,  1962.

 3.  DeWalle, F. B.,  Engeset, J. and Lawrence, W.  Removal of Giardia
    Iambiia  cysts  by drinking water treatment plants.  U.S. EPA Report
    52-84-069, Office of  Research and Development, Cincinnati, OH, 1984.

 4.  Logsdon, G. S.,  Thurman, V., Frindt, E. and Stoecker, J.  Evaluating
    sedimentation  and various filter media for removal of Giardia cysts.
    Journal  American Water Works Association.  77:2:61-66, Feb., 1985.

 5.  Al-Ani,  M. Y., Hendricks, D., Logsdon, G. and Hibler, C.  Removing
    Giardia  cysts  from  low turbidity waters by rapid rate filtration.
    Journal  American Water Works Association.  78:5:66-73, May, 1986.

 6.  Hibler,  C. P-  Analysis of municipal water samples for cysts of
    Giardia.   Report prepared for Office of Drinking Water, U.S. EPA,
    January, 1987.

 7.   Logsdon,  G. S., Symons, J. M., Hoge, R. L., Jr. and Arozarena, M. M.
     1981.   Alternative filtration methods for removal of Giardia cysts
     and cyst  models.   Journal American Water Works Association.
     73:111-117, Feb.,  1981.

 8.   US EPA, Office  of  Drinking Water, Criteria and Standards Division.
     Technologies  and costs for the removal of microbiological con-
     taminants from  potable water supplies.  (Draft).  1987.

 9.   National  Drinking  Water Advisory Council.  Minutes of November 19-20
     meeting,  1986.

10.   Lange,  K., Bellamy,  W., Hendricks, D. and Logsdon, G. Diatomaceous
     earth filtration of  Giardia cysts and other substances. Journal
     American  Water Works Association.  78:1:76-84, Jan., 1986.

11.   Bellamy,  W.,  Hendricks, D. and Logsdon, G.  Slow sand filtration:
     influences of selected process variables.  Journal American Water
     Works Association.   77:12:62-65, Dec., 1985.

12.  Bellamy, W., Silverman, G., Hendricks, D. and Logsdon, G.  Removing
    Giardia  cysts  with  slow sand filtration.  Journal American Water Works
    Association.   77:2:52-60, Feb., 1985.
                                    168

-------
13.   Cleasby, J. L., Hilmoe, D. 0. and Dimitracopoulos, C.  J.   Slow sand
     and direct in-line filtration of a surface water.  Journal  American
     Water Works Association.  Vol. 44, Dec., 1984.

14.   Hoff, J. C.  Inactivation of microbial agents by chemical  disinfec-
     tants.  EPA/600/52-86/067, U. S. Environmental Protection  Agency,
     Water Engineering Research Laboratory, Cincinnati, OH, September,
     1986.

15.   Hibler, C. P., Hancock, C. M., Perger, L. M. and Wegrzn,  K. D.
     Swabby inactivation of Giardia cysts with chlorine at  0.5°C to
     5.0°C.  American Water Works Association Research Foundation,
     1987.

16.   Liu, 0. C., Seraichekas, H. R., Akin, E. W., Brashear, D.  A., Katz,-
     E. L. and Hill, W. L., Jr.  Relative resistance of twenty  human
     viruses to free chlorine in Potomac water.  1971.

17-   Sobsey, M. D., Fuju, T. and Shields, P.  Inactivation  of  Hepatitis A
     virus and model viruses in water by free chlorine.  Presented at the
     US EPA/AWWARF Conference, Cincinnati, OH.  March, 1987.

18.   Logsdon, G.S.  Comparison of some filtration processes appropriate for
     Giardia cyst removal.  Presented at the Calgary Giardia Conference,
     Calgary, Canada.  February, 1987.

19.   US EPA, Office of Drinking Water, Criteria and Standards  Division.
     Draft guidance manual  for compliance with the filtration  and
     disinfection requirement for public water systems using surface
     water sources.  October 8, 1987.

20.  Clark, R. M.,  Read, E. J. and Hoff, J. C.  Inactivation of Giardia
      lamblia by chlorine: a mathematical and  statistical analysis.
     EPA/600/X-87/149, Cincinnati, OH, 1987.

21.  Rubin, A. J. Evers,  D. P.,  Eyman, C. M. and  Jarrell, E.  A.
     Inactivation of gerbil-cultured  Giardia  lamblia  cysts by free
     chlorine.  Unpublished, 1987.

22.  Leahy, J. G.,  Rubin, A. J. and  Sproul, 0. J.   Inactivation of Giardia
     muris cysts  by free chlorine.   Appl.  Environ.  Microbiol.
     51:1448-1453,  1987.

23.  Jarroll,  E.  L., Binham, A. K. and Meyer,  E. A.   Effect of  chlorine on
     Giardia  lamblia cyst viability.   Appl.  Environ.  Microbiol.
     41:483-487.

24.  Wickramanayake, G. B., Rubin, A.  J.  and  Sproul,  0. J.  Effects  of
     ozone and  storage temperature on  Giardia cysts.   JAWWA.
     77:8:74-77,  1985.
                                   169

-------
25.  Rubin, H.  A.M Leahy,  G.  T.  and  Sproul,  0.  J.   Inactivation  of  Giardia
     muris by free chlorine and  chlorine  dioxide.   Water  Resources  Center,
     Ohio State University, Columbus,  Ohio,  1986.

26.  Rubin, A.  J.   Factors affecting the  inactivation  of  Giardia cysts by
     monochloramine and comparison with other disinfectants.   Presented at
     the US EPA/AWWARF Conference, Cincinnati,  OH.   March,  1987.

27.  Payment, P.,  Trudel,  M.  and Plante,  P.  Elimination  of viruses and
     indicator bacteria at each  step of treatment during  preparation of
     drinking water at seven water treatment plants.   Appl.  Environ.
     Microbiol.  49:1418,  1985.
                                  170

-------
                   A STUDY OF WATER TREATMENT PRACTICES
                 FOR THE REMOVAL OF GIARDIA LAMBLIA CYSTS

                       by:  Jerry Ongerth
                            Department of Environmental  Health
                            University of Washington
                            Seattle, WA
                               INTRODUCTION

    An investigation of Giardia cyst removal  in a variety of water filtra-
tion plants was conducted in the spring/summer of 1985.   The project was
sponsored by the American Water Works Association-Research Foundation
(AWWA-RF).  The project was conducted by the  University  of Washington (UW)
and the California Department of Health Services (CDHS). each of which
contributed approximately 20 percent to project funding.

                                OBJECTIVES

    The project objective was to document the Giardia cyst removal charac-
teristics of full-scale water treatment plants under normal operating con-
ditions.  The information is intended to provide field scale verification
of previously reported laboratory and pilot studies and  to provide direc-
tion and guidance to water utilities for maximizing removal of
Giardia cysts.

                                 APPROACH

    The project approach focused on collecting information of greatest
relevance and usefulness for water treatment plant design and operation and
on conducting relatively costly field  investigations including Giardia cyst
analysis efficiently, within stringent budget  limits.  Three types of
treatment were studied;  conventional  complete treatment, direct  filtra-
tion, and diatomaceous earth filtration.  A slow sand filtration  plant was
examined  in a companion project sponsored by AWWA-RF.  Criteria set by
AWWA-RF for selecting water treatment  plants  included the  following:  1)
plants on water sources likely to have Giardia present;  2)  relatively small
treatment plant capacity; and 3) raw water conditions that  tend to make
removal of particulates/turbidity difficult,  i.e., cold  water,  low tur-
bidity, and low alkalinity.  Plants evaluated  in the study were typical
rather than unicue  in design and operating characteristics.  Although
plants were selected for the likely presence  of  Giardia  in  the  source
                                     171

-------
water, a mobile pilot plant, seeded with Giardia cysts, was operated in
parallel with each full-scale plant.  This was to insure that useful data
could be collected even if Giardia were not present in the source water
during field investigations.  The following treatment plants were selected
for study in consultation with the AWWA-RF Project Advisory Committee
(PAC):  1) a 0.57 mgd complete treatment plant at Community A in
Washington; 2) a 4 mgd pressure/gravity in-line filtration plant at
Community B in California; and 3) a 0.016 mgd diatomaceous earth filtration
plant at Community C in Washington.

    Two-week field studies were conducted at each of the three plants.
Field studies included:  1) monitoring the full-scale treatment plant to
characterize cyst removal and treatment performance; 2) operation and moni-
toring of the pilot plant (to supplement cyst removal and performance data
on the full-scale plant, and to determine optimum chemical treatment
conditions); and 3) a sanitary survey of the watershed and treatment plant.
Raw and treated water at each plant were monitored for Giardia cysts and
for related operational parameters including turbidity, particle counts,
pH, temperature, filter head-loss, and filter cycle duration.  Analysis for
Giardia cysts was performed by the CDHS using membrane filtration and immu-
nofluorescence assay (IFA) procedures.  The project budget provided for a
total of  125 Giardia  analyses.  These were divided between the raw and
treated water of the pilot and full-scale plants, with approximately 40
samples for each of the three field sites.  Sampling for Giardia was
focused on periods that are associated with turbidity passage through
filters:  1) during filter conditioning immediately after backwash; 2) imme-
diately before backwash as turbidity breakthrough or time limit criteria
are reached; and 3) during mid-cycle under normally peak efficiency opera-
tion, but as affected by flow rate changes or off/on cycles.

    Chemical treatment optimization was conducted using one of two parallel
pilot plant treatment trains.  Turbidity removal was examined as a function
of pH and of alum and polyelectrolyte concentrations.  Cyst removal and
turbidity were monitored after optimal conditioning scheme was established.

    A sanitary survey of the physical facility and watershed was made to
identify  features related to Giardia cyst presence and to identify design
and operational features related to cyst removal performance or possibly
cyst passage as a result of malfunction or other causes.

                                  RESULTS

    In this section, findings of each of the three treatment plants are
described separately with common observations and conclusions.

COMPLETE TREATMENT, COMMUNITY A, WASHINGTON

    The Community A plant is a prefabricated, welded steel, site assembled
plant consisting of two duplicate parallel 200 gpm trains including in-line
flash mix, baffled hydraulic flocculator, upflow sedimentation with tube
                                     172

-------
settlers, and dual media filtration.  Filters operate at a constant
5 gpm/ftS with production matched to demand by clearwell elevation-
controlled off/on cycles.  The plant is constructed on the concrete slab
cover of the clearwell, and is enclosed by a frame/fiberglass building for
protection from winter freezing.  Raw water comes from a 20 to 30 acre foot
impoundment on a creek draining a wooded upland watershed of approximately
300 acres.

    Operating conditions and treatment performance during the two-week
field investigation were typical according to operating records.  Water
temperature averaged 8°C.  Raw water turbidity averaged 0.3 NTU, ranging
from 0.25 to 0.6 NTU.  Filtered water turbidity averaged 0.13 NTU, ranging
from 0.09 to 0.20 NTU with the exception of peaks as high as 0.3 to 0.5 NTU
immediately following backwash.  Turbidity removal averaged 55 to 60 per-
cent except for periods immediately following backwash when removals were
initially zero, improving the near average over a one to two hour initial
operating period.  It is important to note that no coagulants were used
during the study period.  Because of the low raw water turbidity and
filtered water turbidity generally less than 0.2 NTU, the operator did not
consider chemical conditioning necessary.

    Giardia cysts were found in one of two raw water samples (about 6/gal
each) and in seven of nine filtered water samples (0.05 to 1/gal, average
about 0.4/gal).  Based on observed influent and effluent average con-
centrations, cyst removal efficiency was about 40 percent.  Highest
filtered water cyst  concentrations were found in samples immediately
following backwash.

    Pilot plant cyst removal was  comparable to the full-scale plant.
However, pilot plant performance  with optimal chemical conditioning (10
mg/1 alum, 0.02-0.03 mg/1 Calgon  233, pH 7.0) was significantly  improved
 resulting in filtered water  turbidity averaging 0.03 NTU.  With  this treat-
ment, cysts were  still found  in samples immediately following backwash.

    The  sanitary  survey  identified  several features of the Community A plant
 relevant  to Giardia  duct presence and control.  These  include:   1) evidence
 of beaver and muskrat in the  raw  water  reservoir; 2) backwash effluent sump
 location above the clearwell--common wall construction;  3) no chemical con-
 ditioning; 4)  intermittent filter operation with  numerous off/on cycles
 between  backwashes;  5) preset  automatic backwash, without benefit of opera-
 tor observation to verify effectiveness; and 6) no filter to waste cycle.

 DIRECT FILTRATION, COMMUNITY  B, CALIFORNIA

    Facilities at Community  B  consist of filters, chemical feed  equipment,
 pumps, and piping.   The  filters are  horizontal cylindrical steel  pressure
 vessels,  6 ft diameter x 24  ft  long.  Filters are dual media  (anthracite,
 sand) and operate at 3.75 gpm/ft2.   Raw water is  drawn directly  from a
 creek source draining a  high mountain watershed of several square miles,
 including a meadow/marsh of  about 200 acres.
                                    173

-------
    Operating conditions and treatment performance during the two-week
field investigations were typical  according to operating records.   Water
temperature averaged 15°C.  Raw water turbidity averaged 0.3 NTU,  ranging
from 0.25 to 0.4 NTU; filtered water turbidity averaged 0.1 NTU, ranging
from 0.08 to 0.12 NTU.  Turbidity removal  averaged 70 to 75 percent.  Remo-
val was lowest immediately following backwash, improving from 30 to 40 per-
cent to near normal within one to two hours following backwash.  Chemical
conditioning consisted of 0.15 mg/1 Nalco 8102 added as a filter aid.

    Giardia cysts were found in three of five raw water samples (average about
2.4/gal) and in seven of 13 filtered water samples (0.02 to 0.06 cysts/gal).
However, higher concentrations of cysts were found in three of four samples
taken in one immediate post backwash period (0.1 to 0.7 cysts/gal).  No
cysts were found in four samples taken in a second immediately post back-
wash period.  Cyst removal calculated from average influent and effluent
concentrations would be about 85 percent.  This was comparable to observed
turbidity removal.

    Pilot plant observations indicated cyst removals comparable to the
full-scale plant.  Pilot plant operation with optimal chemical conditioning
(15 mg/1 Alum, 0.05 mg/1 Calgon 233 filter aid, pH 7.0), resulted in
filtered water turbidity of about 0.05 NTU.  No cysts were found in two
samples during this period.

    The sanitary survey identified the following features relevant to
Giardia cyst presence and control:  1) a large watershed with several miles
of riparian habitat typical of aquatic mammal and rodent carriers of
Giardia; 2) intermittent filter operation with numerous off/on cycles bet-
ween backwashes; 3) totally enclosed pressure vessel filters making routine
observation of filter condition and backwash effectiveness virtually
impossible; 4) present automatic backwash incompatible with operator obser-
vation to verify backwash effectiveness; and 5) limited filter to waste
cycle  (5 minutes).

DIATOMACEOUS EARTH FILTER, CRYSTAL MOUNTAIN, WASHINGTON

    The Community C plant consists of a Durco 24DV60 pressure leaf filter
(Duriron Co., Angola, N.Y.) with tanks, pumps, piping, and controls to pro-
vide for precoat (0.2 lb/ftz), body feed (20 mg/1), and continuous recycle
flow for cake retention during periods of reduced demand.  The filter has
60 ft* area  (five  leaves at 15 ft2 each) and is designed to product 1
gpm/ft^  (60 gpm) at normal operating pressure of 60 psi, using Hyflo Super-
Cel diatomaceous earth  (Manville Corp., Denver, Co.)-  Water supply comes
from two creeks, each draining about 300 acre of steep, partly-wooded
alpine terrain.  Operating conditions and treatment performance during the
two, one-week field investigations were typical according to operating
records.  Water temperature ranged from 9 to 12°C, varying diurnally.  Raw
                                      174

-------
water turbidity averaged 0.3 NTU, ranging from 0.2 to 0.7 NTU; filtered
water turbidity averaged 0.08 NTU, ranging from 0.05 to 0.5 NTU.  Turbidity
removal averaged about 60 percent.  The range of conditions and operating
details observed was limited by plant shutdowns caused by an obstruction in
the soure piping.

    Giardia cysts were found in one of two raw water samples (1.3
cysts/gal) and in one of three filtered water samples (approximately 0.01
cysts/gal).  Based on these concentrations, cyst removal efficiency was
approximately 90 percent.  This is comparable to the observed turbidity
removal efficiency.

    Pilot filter operation indicated  increased opportunity for cyst passage
in the initial operating period following application of the precoat.
Also, when feeding cysts in the raw water at 2 to 5,000/gal, there was a
trend towards increased cyst concentration in the effluent with time.
Overall cyst removal was about two logs (99 percent) during this period.

    The sanitary survey identified few features relevant to Giardia cyst
presence and control:  1) both human  and animal potential cyst sources are
present in the watersheds relatively  near the intake areas; also, 2) poten-
tial for cake loss from the filter septum would permit direct cyst passage
in the event of  power  interruption or from a small range of unlikely system
malfunctions.

                                CONCLUSIONS

    Several general  conclusions are  possible based on data collected during
this study.  They  include:   1)  cyst  concentrations are  likely to be appre-
ciable even  in  relatively  remote  high quality sources;  2) overall
Giardia cyst  removal  in water  filtration  is  likely to be comparable to the
efficiency of turbidity  removal;  3)  observed cyst concentrations in
filtered water  were  typically  one to two  logs less than  raw water cyst con-
centrations,  except  in the  first  30  to  60 minutes following backwash when
observed  cyst concentrations were comparable to or greater than  raw water
cyst concentrations;  4)  turbidity and cyst  removals  can  be significantly
 improved  by  operating  with  optimal  chemical  conditioning.  Under such  con-
ditions  removal  will  likely be significantly greater than the turbidity
 removal efficiency;  and  5)  operation including  off/on cycles  between back-
washes appears  to  increase  the probability  of cyst passage through  filters.
                                     175

-------
                    REMOVAL OF GIARDIA IN LOW TURBIDITY
                      WATER BY RAPID RATE FILTRATION

                       by:   Ron R.  Mosher
                            Molzen-Corbin & Associates
                            Albuquerque,  NM  87106

                            David W. Hendricks
                            Colorado State University
                            Fort Collins, CO  80523
                               INTRODUCTION

    The work reported here has been built upon previous research conducted
at Colorado State University (1,2)  using  laboratory scale filter columns 2
inches (5 cm) and 4 inches (10 cm)  in diameter.   The previous research
showed that the rapid rate filtration process could remove virtually 100
percent of Giardia cysts when proper chemical pretreatment was practiced
and about 99.9 percent of coliform  bacteria.   Without proper chemical
pretreatment, removal ranged from zero to 50  percent.  This previous work
also established that "in-line" filtration (rapid mixing followed by
filtration) was an effective mode of filtration for low turbidity waters.

    Research by Al-Ani et aJ. (1,2) related to filtration of low tempera-
ture waters was limited to temperatures of about 3°C, due to ice formation
on copper cooling coils, whereas temperatures of nearly 0°C are observed at
many water treatment plants treating water from mountain streams.  Also,
questions remained as to whether the same results observed at laboratory
scale could be achieved at full-scale operation.

    The latter was the focus of this project.  Of specific concern was the
filtration of low turbidity, low temperature  waters using the rapid rate
filtration process.

                                OBJECTIVES

    The objective of the research was to evaluate the rapid rate filtration
process under ambient conditions of low turbidity and low temperature
waters, utilizing a field-scale pilot plant.   The specific objectives were
to:
                                    176

-------
    o     Ascertain removal  of Giardia cysts;
    o     Ascertain removal  of total  coliform  bacteria and turbidity;
    o     Ascertain the efficiency of rapid rate filtration at water tem-
         peratures approaching 0°C (32°F); and
    o     Determine empirical  relationships between filtered water turbidity
         and coagulant dose for low turbidity ambient water conditions.

                                   SCOPE

    Since it was not feasible to use a full-scale treatment plant for the
experimental work, a 2 foot x 2 foot (0.6 m x 0.6 m) field-scale pilot
system designed to operate in the "in-line" mode of filtration was used.
Most of the testing was conducted during the winter of 1985 with typical
raw water turbidities of 0.5 Ntu and water temperatures of 0.2°C to 0.8°C.
Coagulants used in this research were limited to those previously found
effective for low-turbidity water (i.e., Nalco 8109® and the combination of
alum and Magnifloc 572C®).

                                 PREMISES

    The first premise in this research was that, by using a field scale
pilot plant subject to the ambient water of a mountain stream, the process
behavior of a full-scale water treatment plant would be simulated.  A
second premise was that spiking the field-scale pilot plant with Giardia
cysts and pure cultures of E. coJi would provide means to evaluate removals
which would be indicative of full-scale performance.  A third premise was
that turbidity removal was a measure of filtration performance (1,2)
and could be used to ascertain proper dosages of coagulants.

                                PILOT PLANT

    The pilot plant system,  located in the chemical storage building at the
Fort Collins Water Treatment Plant No. 1  (FCWTP#1) was comprised of two
identical treatment trains using  "in-line" coagulation.  Flow to each
treatment train was split using a "splitter box" located on the second
floor of the chemical building.  The two  rapid mix basins were each
constructed of two 55-gallon steel barrels welded together, having 5.5
minutes detention time and 1/4 Hp propeller mixers.  Following the rapid
mix basins were two identical dual media  filters constructed of 1/2-inch
thick acrylic plastic.  Each filter column was packed with 11 inches  (28
cm) of silica sand  (dig = 0.50 mm, DC = 1.2) and 22 inches (56 cm) of
anthracite  (dig = 0.91 mm, UC = 1.3).  The filters were  13 feet (390  cm)
high and were instrumented with piezometers.

                               CHEMICAL FEED

    Provision was made to meter in two chemicals for each  treatment train
following the splitter box and prior to the  rapid mix basins.   In  this
research, chemicals were used with Filter #2 only, while  Filter #1 was
retained for testing without  chemicals.   The desired chemical coagulant
                                     177

-------
concentration in the raw water was  the  lowest to  give  the  minimum effluent
turbidity.  This dosage was determined  by a plot  of filter effluent tur-
bidity versus dosage of chemicals (the  plot is three dimensional  if two
coagulants are used).

                           CONTAMINANT  INJECTION

    Giardia  Iambiia cysts and pure  cultures of E. coJi were metered into
the headworks of the pilot plant, prior to the splitter box, for the con-
taminant removal investigations.   The Giardia cysts were obtained from a
gerbil colony maintained by Dr. C.P.  Hibler (Pathology Department, Colorado
State University), who provided by laboratory count, one to five million
cysts for each test run.  The cysts were formalin fixed prior to use to
render them  nonviable.  This was one of several safety precautions.

    The pure cultures of about 100 million E. coli bacteria were obtained
from Kirke L. Martin, Director of the Department  of Microbiology Water
Quality Laboratory at Colorado State University.   The  duration of the con-
taminant  removal experiments was usually about 60 minutes.

                                 SAMPLING

    Sampling of the raw water and the pilot filter effluent was conducted
for turbidity, total coliform bacteria, and Giardia cysts.  Turbidity
samples were taken from the raw water flow prior to contaminant injection,
after rapid  mix, and from the filter effluent flow.  Coliform grab samples
were obtained from the raw water flow prior to the splitter box, after the
rapid mix basin, and from the effluent flow after filtration.

    Sampling for Giardia cysts was accomplished by using 1 micron
polyproylene cartridge filters (Micro-Wynd II®, DPPPY, AMF Cuno Division,
Meriden,  Connecticut) and was continuous for the duration of the test run.
The influent sample was obtained just prior to the splitter box, using a
small centrifugal laboratory pump to withdraw about 10 percent of the raw
water flow.  Effluent from the influent sample fiber filter was directed
back into the system at the splitter box.  Sampling the pilot filter
effluent  was done using either a 1-1/2 Hp or a 2 Hp centrifugal pump to
pump the  entire filtered effluent flow through two cartridge filters placed
in series.

                        MEASUREMENTS OF PERFORMANCE

    To ascertain the efficiency of Giardia cyst removal under conditions of
effective chemical coagulation and without chemical coagulation, the per-
cent removal of Giardia cysts was calculated as the number of influent
cysts minus  the number of effluent cysts divided by the number of effluent
cysts times  100.  The number of influent cysts was measured by the
cartridge filter sampling of the approximately 10 percent raw water flow
after contaminant injection, scaled up proportionately to the total flow
through the  pilot system.  This was considered the unequivocal indicator of
performance; i.e., actually measuring  removal  of Giardia cysts rather than
indicators such as turbidity and total coliform bacteria.
                                    178

-------
    Percent removal of indicators such as total coliform bacteria and tur-
bidity were also determined, since previous research by Al-Ani et aJ-
(1,2) had shown strong relationships with Giardia cyst removal when raw
water turbidities were less than 1 Ntu.  Thus, either or both of these
parameters could be considered as valid indicators of filtration perfor-
mance, and their use permits more comprehensive investigation than when
using Giardia cysts alone.

                                  RESULTS

    Removal of turbidity, total coliform bacteria, and Giardia cysts for
the 2 foot x 2 foot (0.6 m x 0.6 m) dual media pilot filters and the con-
ditions of the tests are summarized in Table  1.  These results show that,
with proper chemical pretreatment, the dual media pilot filters were effec-
tive in removing 84 to 96 percent of the turbidity, 97 to 99.95 percent of
total coliform bacteria, and virtually 100 percent of Giardia cysts (vir-
tually 100 percent removal meaning no cysts were detected in the effluent).
Without chemical pretreatment,  removal was 35 to 57 percent for tur-
bidity, 60 percent for total coliform bacteria, and 80 to 91 percent for
Giardia cysts.

     For each of  the runs  in Table  1, samples  were taken to measure tur-
bidity and total coliform bacteria  (if injected), providing data for 25
plots of  turbidity versus time  and  seven plots of bacteria versus time.
Figure 1  is an example of a turbidity versus  time curve using no chemical
coagulation  (Filter #1) and using  proper chemical coagulation (Filter #2).
As  is shown  in Figure  1,  both  the  raw water turbidity and the influent to
Filter #1 are about 0.46  Ntu, whereas effluent from Filter #1 is approxi-
mately 0.2 Ntu without chemical  coagulation.   Due to addition of the coagu-
 lant for  Filter  #2, the  influent  turbidity was increased to 0.68 Ntu, while
 the effluent was reduced  to 0.03  Ntu.  Comparison of the results obtained
for Filters #1 and #2  indicates the  turbidity reductions that can be
 expected  for  improper  (i.e.,  no coagulation or insufficient coagulation)
and proper chemical coagulation,  respectively.

     The methodology for  the majority of  the testing was that  a  "run" began
when the  coagulant feed was begun.   Generally, water without  coagulants  had
already flowed through the  pilot  plant  for at least an hour after back-
washing and  prior  to  initiating the  chemical  feed  (to measure and adjust
 flow rates,  set  up contaminant  and coagulant  feed,  etc.).  As such, the
 response  of  the  pilot  filter  through  the chemical  conditioning  process was
 observed. The shapes  of  the  influent  and  effluent  curves for Filter #2  in
 Figure  1  illustrate the  chemical  preconditioning  of the filter.

     A concern of regulatory officials  is that a  filter to waste period may
be  necessary  following backwash.   As discussed above, the majority of this
work was  performed without  chemical  preconditioning of the filter, and
 corresponding turbidity  versus  time  plots  indicated an experimental decline
 in  effluent  turbidity  to  an equilibrium  level, defined by no  further
 change, taking 20  to 25 minutes after  initiating  chemical feed.  The time
 to  reach  equilibrium  is  designated as  the  transition time, te,  during which
                                     179

-------
                                                                TABLE 1."  SUMMARY OF TEST CONDITIONS AND RESULTS FOB TURBIDITY REDUCTION, TOTAL COLIFOBM BACTERIA
                                                                     REMOVAL, AIID CIAHOIA CYST REMOVAL USING THE DUAL MEDIA PILOT SYSTEM LOCATED AT THE FORT
                                                                    COLLINS HATER TREATMENT PLANT NO. 1 WITH THE CACHE IA POUDRE RIVER AS THE SOURCE OF WATER.
OS
O
Run
Number Date
1

2

3

4

5

6

7
B
9

10
11
n
13

14

15
20

21
22
23

24
25
2-16-85

2-17-85

2-22-85

2-23-85

2-24-85

2-26-85

2-27-85
2-27-65
3-01-85

3-03-85
3-05-85
3-D5-B5
3-09-85

3-12-85

3-13-85
3-24-85

3-26-8S
3-26-85
3-29-B5

4-02-85
4-02-65
Hater Hydraulic
Filter Temp. Loading Rate
Number (-C) (m/hr) (GPH/ft*)
1
2
1
2
1
2
1
2
1
2
1
2
2
1
j
2
2
2
1
1
2
1
2
2
1
2
2
2
1
2
4
2
0.3

0.2

0.3

0.3

0.3

0.3

0.3
0.3
0.3

0.3
D.2
0.2
0.8

1.9

1.9
7.3

8.2
8.3
2.7

7.6
7.6
12. OB
12.17
12.37
12.27
12.22
12.15
12.20
12.30
12.15
12.25
11.29
11.42
7.82
7.99
12.25
12.25
12.34
6.21
6.14
12.49
12.47
12.03
12.10
8.70
12.12
12.22
11.86
11. 88
12.25
12.17
12.54
12.10
4.94
4.98
5.06
5.02
5.00
4.97
4.99
5.03
4.97
5.01
4.62
4.67
3.20
3.27
5.01
5.01
5. OS
2.54
2.51
5.11
5.10
4.92
4.95
3.56
4.96
S.OO
4.85
4.66
5.01
4.98
S.13
4.95
Primary
Coag.
*•
None
B109
None
6109
None
8109
None
8109
Hone
8109
None
8109
8109
None
None
8109
B109
8109
None
None
8109
None
8109
8109
None
Alum
Alum
8109
None
8109
8109
8109
Primary
Coag.
Dose
(mg/1)
• **
o'
10
0
24
0
14
0
13
0
12
0
12
12
0
0
10
11
26
0
0
9
0
5
13
0
7.4
8.0
18.3
0
19
23.5
20.4
Secondary
Coag.
*•
None
Hone
None
Hone
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
None
572C
572C
Hone
None
None
None
Hone

Coag.
Dose
(mo/I)
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1.8
2.1
0
0
0
0
0

Runt
Duration
(Hours)
7.00

22.50

19.17

5.00

18.63

3.00

1.58
1.00
65.33
19.00
3. 25
2.50
0.67
20.00

15.50

12.67
14.00

2.08
•0.83
14.00

3. SB
7.17


Ran Effluent
Turbidity Turbidity
(NTU) (NTU)
0.59
0.60
0.47
0.4B
0.49
0.49
0.47
0.47
0.46
0.46
0.45
0.45
0.46
0.45
0.92
0.47
1.46
1.39
1.46
O.SS
O.S9
0.49
0.48
0.61
0.51
0.51
0.61
0.6S
0.46
0.46
0.43
0.43
0.26
0.05
0.23
0.02
D.24
0.03
0.20
0.03
0.21
0.04
0.21
0.03
0.06
0.22
0.41
0.03
0.1B
0.23
0.86
0.31
0.31
0.29
0.30
0.06
0.24
O.OB
0.07
0.07
0.30
0.10
0.02
0.03
Average
Turbidity Conform
Removal Cone.
(Percent) (Org. /100ml)
55.9
91.7
51.1
95.8
51.0
93.9
57.4
93.6
54.3
91.7
53.3 500
93.3 360
87.0
51.1
55.4
93.6
87.7
83.5
41.1
43.6
47.5
40.8
37.5
90.2 3,900
52.9
84.3
88.5 2,500
89.2 8,700
34.8
78.3
95.3 5.600
93.0 11,000
Average Influent Effluent
Effluent Average Glardla GUrdla GlardlatT
Conform Conform Cyst Cyst Cyst
, Cone- Removal Cone. Cone. Removal
(Org. /100ml) (Percent) (Cysts/1) (Cysts/1) (Percent)










2DO 59.6
9 97.5
170 0 100
340 69 79.5



1.4 0 100
5.0 0.5 90.6




580 85.1 270 0.4 99.8


460 81.6 3.5 0 100
190 97.8 3.2 0 100


3 99.95 410 0 100
6 99.95 0.2 0 100
           'Abstracted from Table A-l. Kosher and Hendrtcks, 1985.

          "8109 and 572C  refer to Malco 8109* and Magntfloc 572C». respectively.

         "'Alum dosage reported as mg/1 of AI2(504)]-14II20.

          touratlon  of contaminant Injection and sampling for Sttrdta and conform bacteria runs was between 40 and  60 minutes.

         "|00  percent removal of Slard la cysts means NO cysts were detected In the effluent sample.

-------
the filter system is becoming chemically preconditioned, and which must be
a filter to waste period.  The term "chemical conditioning" can be defined
as "the transitive process of attaining equilibrium with the filter bed,
first by transport of chemical coagulants to the filter bed and within it,
and second by physical and chemical interactions within the filter bed."

    Figure 2 illustrates these definitions for run 4 (also depicted in
Figure 1), in which te was about 23 minutes.  A limited amount of testing
was done with the filter chemically preconditioned and where the run was
started immediately after backwashing.  Figure 2 also shows filter effluent
versus time for run 22 (a run with chemical preconditioning), in which te
was about 7.5 minutes.  The first effluent turbidity sample for run 22,
taken at five minutes, was 0.09 Ntu and the second sample, at 10 minutes,
was at the equilibrium level of 0.07 Ntu.  Giardia cysts were injected for
run 22, beginning immediately after backwashing.  No cysts were detected in
the filter effluent.  The response for run 4 characterizes the majority of
the testing conducted with the field scale pilot filters, whereas run 22
characterizes the response of a full-scale treatment plant where, under
normal operation, the water would be chemically preconditioned.

    To determine the appropriate coagulant dose for low turbidity water,
effluent turbidity versus coagulant dose curves were generated.  Figure 3
shows the experimentally generated response of the dual media pilot filters
to dosages of Nalco 8109® ranging from zero to 24 mg/1.  As Figure 3 shows,
the optimum dose of Nalco 8109® with respect to turbidity removal was about
11 mg/1.  Dosages exceeding  11 mg/1 were equally efficient for turbidity
reduction, as is shown by the L-shaped curve in Figure 3.

    Figure 4 shows the response of the dual media pilot filters to dosages
of alum and Magnifloc 572C®.  The three-dimensional surface was generated
by measuring effluent turbidity while holding the Magnifloc 572C® dose
constant and varying  the alum dosage.  The dark line on the surface shown
 in Figure 4  identifies the  locus of points describing  the optimum com-
binations of alum and Magnifloc 572C® resulting in an  effluent turbidity of
0.05  NTU or  less.

    The turbidity response  curves depicted  in Figures  3 and 4  illustrate
the procedure that  should be followed to determine  if  a particular coagu-
 lant  is effective and what  dosages should  be used.  Although the  idea  is
not new,  it  has  not been commonly  practiced  when  raw water turbidities  are
 less  than  1  NTU.
              /
    A 10-minute  lapse  in chemical  feed was  experienced during  a Giardia run
using the 2  foot x 2  foot  (0.6 m x 0.6 m)  dual media pilot filters.  This
was the only  run using the  dual media pilot  filters with  proper chemical
pretreatment  in which Giardia cysts were detected  in the  effluent, with a
Giardia cyst  removal  of  99.8 percent.  Figure 5 shows  percent  removal  of
turbidity and total coliform bacteria versus time  for  this run.   Figure 5
shows a dramatic decrease in total coliform  bacteria removal  (99  percent
before  the  lapse compared to 64 percent afterwards) coinciding with  the
 lapse in  chemical feed.  A  decrease in turbidity  removal  was observed  after
                                    181

-------
      0.7
      0.6 -
      0.5 -
      0.4 -
   Q  0.3
   m
   IX
   I-
      0.2 -
      0.1 -
                         INFLUENT FILTER *2
                         RAW WATER
                        	•	•	

                         INFLUENT FILTER
EFFLUENT FILTER
                         EFFLUENT FILTER *2
                                    TEMP = 0.3° C
                                    13 mg/l 8109
                                    NO CHEMICALS
                                     TIME (HOURS)

Figure 1.   Typical turbidity versus time for  runs  without chemical
             preconditioning  (data  from  run 4).
              -TERMINATE BACKWASH


              -START CHEMICAL FEED
               TE TRANSITION TIME FOR
            CHEMICAL CONDITIONING - RUN 4
                                           EQUILIBRIUM FOR FILTRATION EFFLUENT - RUN 4
                                                 FILTER EFFLUENT - RUN 22


                                                  FILTER EFFLUENT - RUN 4
      0.0
                                                      —r~
                                                       40
                                   TIME - MIN
Figure 2.   Illustration of  chemical conditioning process  for  a rapid
             rate  filter with proper chemical  coagulation.
                                      182

-------
     0.9
                             8        12
                              8109 DOSAGE (mg/L)
                                                          20
                                                                    24
 Figure 3.  Effluent turbidity versus coagulant dose for  Nalco  8109®.
Figure 4.  Effluent turbidity versus dosages of alum and Magnifloc 572C®.
                                   183

-------
7 UU -

90 -
80 -

70 -

^
Of? r\
o U ~-
5
UJ
^ 50 -
H
UJ
0 40 -
DC
UJ
0_
30 -
20 -

10 -

n

I

!
T
j



i
i



START
:-« 	 CONTAMINANT
INJECTION

"
I
i

_
.
!
i i i i i






SHUT PILOT SYSTEM DOWN
DUE TO MECHANICAL
x PROBLEMS ^
(35 MIN.)














II I I I

n \
I — \ TMDmrHTV QPKAOVAI IV*}
— — - — 4^_^^ 1 UnblUI 1 I ntlvivjVML \7of
\
\ COLIFORM REMOVAL (%)
N.
^^^^-—^
^~O




_^ 	 .
10 MIN.
LAPSE IN
CHEMICAL FEED




TEMP = 1.9°C
13 mg/l 8109
99.8% GIARDIA REMOVAL

il i i i i i
11.2   11.4   11.6   11.8
12
12.2  , 12..4   12.6   12.8
13
                     TIME (HOURS)
 Figure  5.   Effect of lapse in chemical  feed.
                        184

-------
the lapse In chemical feed, but it was not as pronounced as the decrease in
total coliform bacteria removal.  This run demonstrates the need for con-
tinuous chemical feed.

    Twelve experimental runs were conducted with the dual media pilot
filters at water temperatures ranging from 0.2°C to 0.4°C.  As Table 1
shows, when proper chemical pretreatment was practiced, turbidity removal
from 0.45 Ntu raw water to 0.02 Ntu effluent was attained.  Giardia cyst
removal was virtually 10 percent and total coliform bacteria removal was 97
percent or greater when the low temperature water was properly chemically
treated.

    Al-Ani et al. (1) reported on treating waters with turbidities less
than 1 NTU that ". .  . if turbidity removal exceeds 70 percent and if
filtered water turbidity is lower than 0.10 NTU, the probability is 0.85
that removals of Giardia cysts would equal or exceed 99 percent."  Results
obtained here substantiate this statement  in that every contaminant run
conducted in which greater than 70 percent removal of turbidity was
attained, corresponding Giardia cyst removal was greater than 99 percent.
Compared to turbidity removal, total coliform bacteria removal appears to
have greater sensitivity for  indicating Giardia cyst removal.  However, due
to the fact that total coliform bacteria concentrations can be as low as 1
per  100 ml in mountain streams during the winter, and since total coliform
bacteria analysis is  more difficult and time consuming than turbidity
measurements, turbidity removal must still be viewed as the most attractive
surrogate indicator  of Giardia cyst removal.

     Ten piezometer taps, at approximately 4-inch (10 cm) spacing, were
installed on each of  the 2 foot x 2 foot  (0.6 m x 0.6 m) dual media pilot
filters to measure headless through the media.  The rate of headloss
increase was measured in the  two pilot filters, with one operated with
proper chemical pretreatment  and the other without chemical coagulation.
With proper chemical  pretreatment, approximately 19 hours were required to
reach the terminal headloss of 6.7 feet (2.0 m).  In contrast, for the
filter without chemical pretreatment, 65 hours were needed to reach about 3
feet  (1 m) of headloss.

     A  limited amount  of investigation was  conducted to determine the
headloss profile through the  dual media of the pilot filters.  Initially,
at a headloss of about 1.6 feet (0.5 m), six percent of the head occurred
in the gravel support, 68 percent occurred in the 11-inch  (28 cm) layer of
silica sand, and 26  percent was in the 22-inch (56 cm) layer of anthracite.
The  headloss profile  at the terminal headloss of 6.7 feet  (2.0 m) showed
two  percent of the head developed occurred in the gravel support, 53 per-
cent occurred in the  silica sand, and 45 percent of the total head deve-
loped  occurred  in the anthracite.

     The  2 foot x 2 foot (0.6  m x 0.6 m) dual media pilot filters were  used
to ascertain the effect of backwash flow  rate on bed expansion.  The
results  indicate that, at flow  rates above 13 gpm/ft2  (32  m/hr), there  is a
linear relationship  between backwash flow  rate and percent bed expansion.
                                     185

-------
At a backwash rate of 15 gpm/ft2 (37 m/hr),  the sand and anthracite bed was
expanded about 15 percent.   To achieve 50 percent expansion of the bed,
which is usually a design target,  a backwash rate of 25 gpm/ftz (61 m/hr)
was required for the dual media used in the  pilot filters.

                                CONCLUSIONS

    Although the findings of this  research were generated using the 2 foot
x 2 foot pilot filter, they should be applicable to operation of full-scale
plants since the only difference is in the bed area/perimeter relationship.
The conclusions apply to the treatment of low turbidity waters to achieve
high removal efficiencies,  which can reduce  the hazard of giardiasis
outbreaks to very low risk levels.

    The findings of this research  further verify that the rapid rate
filtration process, in treating low turbidity waters, can provide high per-
cent removals of microscopic particles when  operated using "proper" chemi-
cal coagulation.  When operated with no chemical coagulation or with
improper coagulation, the process  will not function as intended, with
significant amounts of microscopic particles passing through the filter.
Further, a lapse in chemical feed  will permit a high proportion of par-
ticles to pass through the filter.

    The term "proper" chemical coagulation for low turbidity waters may be
defined as "the particular selection of chemicals and dosages which will
result in a significant  reduction  in microscopic particles in the rapid
rate filtration process, as measured by percent reductions in turbidity,
bacteria, and Giardia cysts, and as measured by visual analysis of
microscopic organisms retained on  cartridge filters."

    In "in-line" mode of rapid rate filtration process, i.e., using rapid
mix then filtration,  is  effective for rapid rate filtration of low tur-
bidity waters.  These findings verify the work of Al-Ani et al. (2) who
found the same using  lab-scale pilot filters.  Also, it was verified that
percent reduction of  turbidity waters is an effective and useful indicator
of efficient filtration  of low turbidity waters.

    The turbidity versus time data  showed reductions to final equilibrium
levels of turbidity within five minutes after backwash, when the filter  has
been "chemically preconditioned."   Without chemical preconditioning, the
time to reach equilibrium turbidity was about the same as the detention
time, with dispersion,  through the  rapid-mix basins and the filters.
Filter to waste after backwash may  not be necessary  if the system has  been
chemically preconditioned prior to  backwash  (by normal plant operation with
proper chemical coagulation).

    The role of water temperature at  near 0°C  (32°F) when treating  low tur-
bidity waters has been  uncertain, causing speculation  that the  temperatures
cause treatment difficulties.  The  results of this  research show  that  rapid
rate filtration of  low  turbidity waters  can  be  as efficient at  near  0°C
(32°F) as at higher  temperatures, if  proper  chemical  coagulation  is
practiced.
                                    186

-------
    The findings of this research using 2 foot x 2 foot filter columns were
the same as those of Al-Ani et aJ. (1,2), who used 2 inch (5 cm)  and 4 inch
(10 cm) filter columns, indicating scale is not a factor in operating pilot
filter columns.  There is little doubt that the findings of this  research
are applicable to full-scale operation.

    The single most important finding of this research with respect to
full-scale operation is that the paramount role of "proper" chemical coagu-
lation has been further reinforced.  Without it, full-scale operation can
be much less efficient and will permit passage of up to 80 percent of
Giardia cysts and other microscopic particles.  With it, the probability of
cysts passing the filtration process can be reduced to less than  0.1 per-
cent of cysts applied by the raw water.  The latter should be attained if
finished water turbidity levels are 0.05 Ntu or less.
                                REFERENCES

 1.  Al-Ani,  M.Y.,  McElroy,  J.M., Hibler,  C.P.,  and  Hendricks D.W.
    Filtration of  Giardia cysts and  other substances, Volume 3:   rapid
    rate filtration.   Colorado State University Environmental Engineering
    Technical  Report  5847-85-1, February, 1985.

 2.  Al-Ani,  M.Y.,  Hendricks,  D.W.,  Logsdon,  G.S., and Hibler C.P.
    Removing Giardia  cysts from low turbidity water by  rapid rate
    filtration.  J.AWWA.   78:66-73,  May,  1986.
                                    187

-------
                               BIBLIOGRAPHY


Al-Ani, M.Y., McElroy, J.M., Hibler,  C.P.,  and Hendricks D.W.   February
    1985.  Filtration of Giardia Cysts and  Other Substances, Volume 3:
    Rapid Rate Filtration.  Colorado  State  University Environmental
    Engineering Technical Report 5847-85-1.

Al-Ani, M.Y., Hendricks, D.W., Logsdon, G.S., and Hibler C.P-   May, 1986.
    Removing Giardia Cysts from Low Turbidity Water by Rapid Rate Filtra-
    tion.  J.AWWA.  78:66-73.

Gertig, K. and Williamson-Jones, G.  1985.   Personal Communication.

Gertig, K., Alexander, B., and Williamson-Jones G.  June 23, 1986.
    Giardia  lamblia Cyst Removal by In-Line Direct Filtration.  Paper pre-
    sented at 1986 AWWA Annual Conference and Exposition, Denver, Colorado.

Herman, L.  June 25, 1986.  Coagulation:  A Generation of Process Control.
    Paper presented at 1986 AWWA Annual Conference and Exposition, Denver,
    Colorado.

Hibler, C.P.  1985.  Personal Communication.

Karl in, R.   1985.  Personal Communication.

Mosher, R.R., and Hendricks, D.W.  May, 1986.  Filtration of Giardia Cysts
    and Other Particles Under Treatment Plant Conditions, Volume 2:  Rapid
    Rate Filtration Using Field Scale Pilot Filters on the Cache La Poudre
    River -  Part 1.  Colorado State University Environmental Engineering
    Technical Report No. 86-5847-2.

Mosher, R.R. and Hendricks, D.W.  December 1986.  Rapid Rate Filtration of
    Low Turbidity Water Using Field-Scale Pilot Filters.  J.AWWA.
    78:42-51.

Saterdal, R., Blair, J., Alexander, B., and Hendricks D.W.  May 1986.
    Filtration of Giardia Cysts and Other Particles Under Treatment Plant
    Conditions, Volume 3:  Survey of Rapid Rate Full Scale Plants.
    Environmental Engineering Technical Report 5074-86-3, Department of
    Civil Engineering, Colorado State University, Fort Collins, Colorado.
                                    188

-------
                         GAG SUBSTITUTION FOR SAND

                       by:  Sandra L. Graese
                            Department of Civil Engineering
                            University of Illinois at Urbana-Champaign
                            Urbana, IL  61801
                            Present Address:  CH2M Hill
                            Milwaukee, WI  53201

                            Vernon L. Snoeyink
                            Department of Civil Engineering
                            University of Illinois at Urbana-Champaign
                            Urbana, IL  61801

                            Ramon G. Lee
                            Director, Research and Technology
                            American Water Works Service Company
                            Marl ton, NJ  08053
                               INTRODUCTION
    Granular activated carbon (GAC) is widely used in drinking water treat-
ment.  In the United States, it is commonly used in place of granular media
in conventional rapid filters (GAC filter-adsorbers) for removal  of both
organic compounds (primarily taste and odor) and turbidity.  This practice
is in contrast to that in European countries where GAC adsorbers  most often
are used after granular media filters (post-filter adsorbers) for the re-
moval of taste and odor and other trace organics, in addition to  total or-
ganic carbon.  GAC filter-adsorbers have been proven effective in removing
a variety of taste and odor compounds for periods typically ranging from
0.5 to five years. The new drinking water standards necessitate that the
performance of these systems for the removal of specific trace organics be
evaluated.  In addition, because turbidity standards are likely to be
reduced under pending filtration regulations, the performance of GAC as a
filtration media needs also to be evaluated.

                SCOPE OF RESEARCH AND EXPERIMENTAL APPROACH

    Design engineers and utility personnel require several types of infor-
mation:  1) to make a good decision whether to use GAC as a replacement
for conventional granular filter media or to design and build a new system
specifically for GAC, 2) to adequately design and specify the retrofit and
                                     189

-------
replacement, and 3) to effectively operate the system after replacement.
Knowledge of filtration and adsorption performance,  means of controlling
microbiological growths, and procedures for filter-adsorber cleaning,
condition monitoring, and filter-adsorber design have been examined in
detail in a report prepared for the American Water Works Association
Research Foundation (1).  In this paper, selected information from that
report is presented, including:

    o   The performance of GAC as a filter media, and discussion of the
        need for a sand layer below the GAC.

    o   The performance of GAC filter-adsorber for removal of taste and
        odor, total organic carbon, trihalomethanes, and other organics.
        The differences in performance that are anticipated for filter-
        adsorbers, especially sand replacement systems, compared to post-
        filter adsorbers are also presented.

    Data were collected from published literature and from operating plants.
A questionnaire was distributed to 15 plants of the American Water Works
Service Company (AWWSC) and four plants of the Connecticut Water Company;
personal contact with several other utilities provided additional informa-
tion.  Results from this survey documented the current design, operation,
and performance of GAC filter-adsorbers, and identified potential problems.
Selected site visits followed the collection of survey data to more closely
observe filter-adsorber operation.  A sampling program was also developed
to provide  a means of monitoring filter-adsorber condition, and was
implemented to examine GAC size and stratification and extent of mudball
formation.  A critical analysis of plant experience and information from
the literature identified the merits and liabilities of filter-adsorber
systems, and indicated where modifications  in design and operation could
lead  to  improved  performance.

                           GAC AS A FILTER  MEDIA

    The  design, operation, and performance  of GAC processes are  influenced
by  their placement within the  treatment process  sequence. GAC post-filter
adsorbers  frequently  use GAC with a small effective size and  large unifor-
mity  coefficient  to  promote  rapid adsorption of  organic compounds and
 restratification  to  maintain the adsorption front.  However,  GAC with  a
 large effective size  and small uniformity coefficient  allows  longer  filter
 runs  to  a  given terminal head  loss, facilitates  cleaning  of the  filter,  and
 reduces  carbon  loss.   Since  filter-adsorber media must satisfy both  adsorp-
 tion  and filtration  constraints, a tradeoff between adsorption and filtra-
 tion  efficiency in  terms of  particle  size and uniformity  coefficient  is
 necessary.   In addition, higher  solids  loadings  into filter-adsorbers,
 compared to post-filter adsorbers, necessitate more frequent  backwashing,
 typically  at the  same  frequency  as conventional  media  filters.   The
possible effect of solids  loading, existing design  (for retrofit  systems),
and more frequent backwashing  on the  adsorption  of  organics need  also  be
 considered.
                                    190

-------
    Filter-adsorbers may be designed specifically for GAC.  More frequently,
some or all of the media in existing filters have been replaced with GAC;
these processes are commonly referred to as sand replacement systems.  In an
attempt to satisfy both adsorption and filtration requirements, the current
sand replacement systems, as indicated from our AWWSC survey, use 12x40 or
8x30 U.S. standard mesh carbon over several inches of sand.  The 8x30 mesh
carbon is significantly larger than conventional filter sand and is compar-
able to the average size anthracite, while the 12x40 mesh is only slightly
larger than typical filter sand (see Table 1).  The GAC media is also con-
siderably less dense, has a larger uniformity coefficient, and has a more
irregular and angular surface.  Other specially designed filter-adsorber
systems, such as those used by the Connecticut Water Company, use deep beds
of coarse GAC media with a small uniformity coefficient, the media having
been carefully selected through pilot plant studies.

               TABLE 1.  TYPICAL FILTRATION MEDIA CHARACTERISTICS
                         Granular Activated Carbon*
                            8x30          12x40        Sand(5)   Anthracite(5)



Effective Size  (rrni)       0.80-1.05     0.55-0.75    0.38-0.65      0.45-1.6


Uniformity Coefficient             <  1.9              1.2-1.7        <  1.8


Particle Density                  1.30-1.55             2.65          1.57
Wetted  in Water (g/cm3)



*Carbon manufacturer  data.

     The GAC media  characteristics  influence head  loss  development,  filter
run  length, backwash  requirements, and  filtered water  quality.  These fac-
tors are considered  in  the  following  sections.

PERFORMANCE HISTORIES OF  FILTER-ADSORBERS:  SURVEY RESULTS

     Survey data from  plants operated  by the American Water Works Service
Company (AWWSC) and  the Connecticut Water Company, coupled with selected
site visits,  indicate that  with  proper  design  and operation, GAC filter-
adsorbers can consistently  and effectively yield  the water quality and water
production desired.   Design parameters  and operating characteristics  from
the  15  AWWSC  plants  and four Connecticut Water Company plants  are  given in
Tables  2 and  3.
                                      191

-------
TABLE 2.  FILTER-ADSORBER CHARACTERISTICS
Plant
Product

American Water Works Service Company
Chattanooga, TN 1-10
Chattanooga, TN 11-20
Chattanooga, TN 21-23,25
Chattanooga, TN 24,26
Chattanooga, TN AldMch 1,3-5,6,8
Chattanooga, TN AldMch 2,7
Davenport, IA
East St. Louis, IL
Granite City, IL
Hershey-Palmyra, PA
Hopewell, VA
Huntington, WV 11-15
Huntington, WV 17,19
Huntington, WV 16,18,20
Huntington, WV Perl 1,2
Montrose, PA
New Castle PA (old design)
New Cumberland, PA
New Cumberland, PA
Norristown, PA
Peoria, IL
PeoMa (AldMch units), IL
Pittsburgh (AldMch Station), PA
Pittsburgh (Hays Mine Station), PA
Princeton, WV
Washington/McDonald, PA
Connetlcut Water Company
Mackenzie, Clinton, CT
Stafford, Stafford Springs, CT
Rockvllle, Vernon, CT
Williams, Chester, CT
Other
STocEton East Water District
CECA GAC 30
ICI Hydrodarco 3000
CECA GAC 30
ICI Hydrodarco 3000
CECA GAC 30
Calgon FUtrasorb
CECA GAC 30
Calgon FUtrasorb
Calgon FUtrasorb
Calgon FUtrasorb
Calgon FUtrasorb
Calgon Flltrasorb
CECA GAC 30
CECA GAC 40
CECA GAC 40
Calgon FUtrasorb
CECA GAC 40
CECA GAC 40
Calgon FUtrasorb
Calgon
Calgon FUtrasorb
Calgon FUtrasorb
Calgon FUtrasorb
Calgon FUtrasorb
Calgon FUtrasorb
CECA GAC 40

Calgon
ICI
Calgon
Calgon

Calgon FUtrasorb

100

200
200
300
200




100


200

300
200
200
200
200







300
GAC
Mesh

8x30
8x30
8x30
8x30
8x30
8x30
8x30
12x40
12x40
8x30
12x40
12x40
8x30
12x40
12x40
8x30
12x40
12x40
12x40
8x30
8x30
12x40
12x40
12x40
12x40
12x40

8x16
8x16
8x16
8x16

5x30
GAC E.S.
(mm)

0
0

0

0
0
0
0
0



0


0
0
0
0
0
0
0








0.85
.80-1.00
0.85
.80-1.00
0.85
.80-0.90
0.85
.55-0.65
.55-0.65
.80-0.90
.55-0.65
.55-0.90
0.85
0.60
0.60
.80-0.90
0.60
0.60
.55-0.65
.80-0.90
.80-0.90
.55-0.65
.55-0.65
.55-0.65
.55-0.65
0.60

1.3
1.35
1.3
1.3

GAC
U.C.

2!o
2!o
<2.1
<1.9
<2.1
1.7
1.7
1.9-2.4
<1.9
<2.4
<2.0
<1.8
<1.8
1.6-2.4
<1.8
<1.8
1.6-2.1
1.6-2.4
2.0-2.4
1.7
"=1.7
"=1.7
NA
<1.9(1.5)

1.4
1.3
1.4
1.4

0.85-1.05 <1.8
GAC
Depth
(in.)

28
30
25
25
25
25
30
18
18
24
18-24
28,30
30
30
30
30
30
20
20
24
30
30
30
30
15
30

48
48
42
48

36
Sand E.S.
(mm)

0
0
0
0
0
0

0
0
0






0


0
0

0
0

0






0,

.45-0.55
.45-0.55
.45-0.55
.45-0.55
.45-0.55
.45-0.55
None
.40-0.50
.40-0.50
.40-0.60
None
NA
NA
NA
NA
NA
.45-0.55
NA
NA
.45-0.55
.40-0.50
None
.35-0.45
.35-0.45
0.60
.45-0.55






.43-0.50
Sand
U.C.

2.0
2.0
2.0
2.0
1.6
1.6
None
x 1 • D
x 1 • V
1.5
None
NA
NA
NA
NA
NA
1.5
NA
NA
1.5
1.8
None
<1.6
1.6
NA
<1.5







Sand
Depth
(in.)

4
7
4
4
4
4
None
12
12
3
None
<4
<4
<4
<4
6
3
3-7
3-7
4
9
None
2-4
2-4
12
4






10

-------
TABLE 3.  FILTRATION PERFORMANCE OF GAC FILTER-ADSORBERS
Plant
Filtration
Rate
(gpm/sqft)
American Water Works Service Company
Chattanooga, TN 1-10
Chattanooga, TN 11-20
Chattanooga, TN 21-23,25
Chattanooga, TN 24,26
Chattanooga, TN Aldrlch 1,3-5,6,8
Chattanooga, TN AldMch 2,7
Davenport, IA
East St. Louis, IL
Granite City, IL
Hershey-Palmyra, PA
Hopewell , VA
Huntington, WV 11-15
Huntington, WV Perl 1,2
Montrose, PA
New Castle PA (old design)
New Cumberland, PA
NorMstown, PA
PeoMa, IL
PeoMa (AldMch units), IL
Pittsburgh (Aldrlch Station), PA
Pittsburgh (Hays Mine Station), PA
Princeton, WV
Washington/McDonald, PA
Connetlcut Water Company
Mackenzie, Clinton, CT
Stafford, Stafford Springs, CT
Rockvllle, Vernon, CT
Williams, Chester, CT
Other
Stockton East Water District
2.4
2.3
1
1
1.4
1.4
2.0-2.5
2
1.8
2.8-3.6
2
2.8
2.1
1.5
1.8
4
3.5
3
1.97
1.5-2.9
0.75-2.0"
1.8'
2

1.8-3.5
0.8-1.65
1.0-1.75
0.7-2.3

4-8
EBCT
(min)

7.3
8.1
15.6
15.6
11.1
11.1
7.5-9.4
5.6
6.2
4.2-5.3
5.6-7.5
6.7
8.9
12.5
10.3
3.2
4.3
6.3
9.4
6.5-12.6
9.3-24.8
5.2
9.3

8.4-16.9
18-36
15-25
12.9-39

2.8-5.6
Run Terminal Terminal
Length Head Loss Turbidity
(hours) (ft) (Ntu)

60
60
45
45
90
90
30-50
24
30
34
48
24
24
48
60
48
25
48
168
72-96
72
72
48

96
96
72
96

20-45

8
8


8
8
6


8

5
6

6
8



6
4
6
7

8
8
8
8

8

0.5
0.5
0.5
0.5
0.5
0.5

0.9
0.9
0.40-0.50
0.50-0.70
1
1

0.6
1
0.4
0.5
0.3
0.5
0.5
0.9
1






0.25
Terminal
Time
(Ntu)



45
45
100
100

24
30
48

24
24
48
X

25
48
X
96
96



X
X
X
X


Applied
Turbidity
(Ntu)

3
3
3
3
2
2
1.1-4.0
7.6
7
2.0
2-5
5.0
5.0
0.5
2.5
<5
3.0
2-4
2-4
1.6
4.0
>1
1.5

0.3
0.3
1
0.5

2-3
Effluent
Turbidity
(Ntu)

0.2
0.2
0.25
0.25
0.18
0.18
0.5
0.5
0.41
0.20
0.2-0.5
0.2-0.5
0.2-0.5
0.2
0.38
0.2
0.20
0.1-0.2
0.1-0.2
0.12
0.16
0.2-0.3
0.25

<0.1
0.1
0.2
<0.2

0.07-0.12

-------
    An average of 26 inches  (minimum =  15  inches,  maximum =  30  inches)  of
12x40 or 8x30 mesh carbon over 2  to  12  inches  of  sand  and 6  to  15 inches
of graded gravel  is used in  AWWSC filter-adsorber  designs.   There are three
exceptions to this design where 8x30 and  12x40 carbons are used without an
underlying sand layer.   Filtration rates  used  at  the AWWSC plants range from
1 to 4 gpm/ft2-  These  rates give rise  to  an average empty bed  contact
time of 8.6 minutes, and a range  of  3.2 to 24.8 minutes.   The use of GAC as
a filter media yields an average  filter run length of  55  hours  and average
effluent turbidities of 0.3  Ntu.   Unit  filter  run  volumes (UFRVs) vary from
2,700 to 19,900 gal/run-ft2, with an average value of  6,900  gal/run-ft^.

    Connecticut Water Company plants use  42 to 48  inches  of  8x16 carbon
(e.s. = 1.3 to 1.35, u.c. =  1.3 to 1.4) without underlying sand or gravel
layers.  Filtration rates range from 0.7  to 3.5 gpm/ft2,   giving rise to
empty bed contact times of 8.4 to 39 minutes.   Average filter run lengths of
90 hours and average UFRVs of 9,400  gal/run-ft2 (range =  5,900  to 15,000
gal/run-ft2) are obtained.  Raw water turbidity is typically less than 6
Ntu at the plants.  Pretreatment  reduces  the turbidity to less  than 1 Ntu,
and GAC filtration produces  effluent turbidities  that  are typically less
than 0.2 Ntu.

    All of the AWWSC and Connecticut Water company plants surveyed meet
current regulatory standards for  effluent  quality.  Optimization of coagu-
lant and filter aid doses may yield  improvements  in filtered water quality.
Alternatively, filtered water quality may also be  enhanced through the
selection of a smaller  media size or the  use of a  sand layer below the GAC,
in designs where sand is not currently  used.

GAC MEDIA:  TURBIDITY REMOVAL AND HEAD  LOSS DEVELOPMENT

    If an appropriate particle size  and particle  size  distribution is
selected, GAC can produce run lengths comparable  to conventional filters,
while achieving similar or better effluent quality.

    Hyde et al. (2) in  a pilot plant study at  Church Wilne,  United Kingdom,
found 60 percent longer filter run lengths and slower  rates  of head loss
development (0.5 m/d versus 0.9 m/d) for a GAC filter  (Filtrasorb 400)
operated in parallel with a sand  filter.   In addition, the effluent tur-
bidity and residual coagulant from the GAC filter was  essentially the same
as that from the sand filter even though the GAC  filter was  operated at a
marginally higher filtration rate.  In a pilot plant study at Elsham, United
Kingdom, reported by Whitford and McCawley (3), parallel  operation of GAC
 (e.s. = 0.5 mm, u.c = 1.74)  and sand (e.s. = 0.4  mm, u.c. =  2)  pressure
filters yielded results similar to those of Hyde  et aJ.  (2).  The rate of
head  loss buildup for the GAC filter was less  than that for the  sand filter
and head loss versus depth profiles indicated that filtration was occurring
at a greater depth  in the GAC.  These effects  may be attributed  to the
larger effective size and smaller uniformity coefficient of the  GAC media.
No detrimental effect of the larger GAC particle size on filtered water
quality was observed; in fact, the GAC filter was superior (on average)  to
the sand filter in  terms of turbidity and residual iron  removal.
                                    194

-------
    Pilot work at the Contra Costa Water District showed that 2 mm GAC was
more effective for turbidity removal than anthracite of the same effective
size (4).  On average, the effluent turbidity from the GAC filter-adsorber
was one-half that from the anthracite filter.  For example, using a filtra-
tion rate of 4 gpm/ft2, the turbidity from the GAC filter was 0.08 Ntu,
while that from the anthracite was 0.18 Ntu.  The improved filtration per-
formance of the GAC can be attributed to the greater angularity and surface
roughness of the GAC particles.

ROLE OF EFFECTIVE SIZE

    Love et al. (5) found GAC  (e.s. 0.55 to 0.65 m, u.c., < 1.9) to be as
effective in removing turbidity as sand (e.s. 0.46 mm, u.c. 1.9) and dual
media (anthracite:  e.s. 1.2 mm, u.c. 1.7, sand:  e..s. 0.4 mm, u.c. 1.6)
filters.  In addition, the filtration efficiency of a GAC filter (e.s.
0.80 to 0.90 mm, u.c. <  1.7) was found to be comparable to that of a dual
media (anthracite/sand)  filter.  Based on these studies, Love et al. (5)
concluded that 24  inches of GAC with an effective size less than 0.90 mm and
a  uniformity coefficient less  than  1.9 is as effective as sand or dual media
(anthracite/sand).

    Caution should be exercised, however, when using monomedia GAC near the
upper effective size  limit  (0.80 to 0.90 mm) suggested by Love et al.  (5),
because other pilot  studies and field observations suggest that filters
using GAC of this  effective size may be more susceptible to turbidity
penetration when high solids  loadings are applied.  Pilot studies at Contra
Costa (4) showed that a  GAC monomedia filter (Filtrasorb 400, e.s. = 0.90,
depth = 34  inches) and a dual  media filter consisting of 20 inches of  1 mm
anthracite  over 10 inches of  0.5 mm sand, were both effective in producing
turbidities less than 0.1 Ntu  for 60 percent of the experimental runs  con-
ducted.  However,  in  an  additional  25 percent of the runs, the effluent
from the GAC filter  exceeded  the effluent goal of 0.1 Ntu, while the
effluent from the  dual media  filter remained between 0.06 and 0.1 Ntu.
These results indicated  that  the monomedia GAC filter was more vulnerable
to water quality changes, particularly peaks in turbidity  (4).

    Similarly, filter-adsorbers at  one AWWSC plant that uses 30 inches of
0.80 to 0.90 mm e.s.  GAC without an underlying sand layer, appear to be more
susceptible to turbidity breakthrough.  While the filters at this plant per-
form as designed to  meet an average effluent turbidity of 0.5 Ntu, filtered
water turbidities  average 0.65 Ntu  during the winter months when pretreat-
ment is  typically  more difficult.   A deeper  bed using this media size, a
finer GAC media, or  an underlying  layer of  sand could all be used to achieve
 lower effluent turbidities.

ROLE OF THE UNIFORMITY COEFFICIENT

    The  large GAC  uniformity  coefficient,  i.e.,  1.7 to 2.4, may have a
significant impact on head  loss development  and  filtered water  quality,  and
may force  the selection  of  GAC with a  larger effective size, particularly  in
deep filters.  For two media  with  the  same  effective  size,  the medium  with
                                      195

-------
the larger uniformity coefficient will  have a greater number of fine par-
ticles in the upper layers of the stratified filter bed,  and also larger
particles in the lower layers.   Filtered water quality is enhanced as small
suspended solids can be effectively removed by the upper  layer of fine par-
ticles.  Unfortunately any improvements in water quality  are accompanied by
increased rates of head loss development and shorter filter runs, because
the fine particles promote rapid plugging of the filter surface (6).

    Love et al. (5) found that a GAC filter (e.s. 0.55 to 0.65, u.c. < 1.9)
showed significantly higher rates of head loss development than a dual media
filter (anthracite:  e.s. 1.2, u.c. 1.7; sand:  e.s. 0.4, u.c. 1.6) operated
in parallel; however, the GAC filter was more effective in removing turbidity
than the dual media filter.  The increased rate of head loss development was
attributed to surface filtration in the single media GAC filter (5).

Connecticut Water Company Pilot Plant Studies (7,8)

    Pilot plant studies by technical Connecticut Water Company emphasize the
role of the large GAC uniformity coefficient.  For the particular water
being  treated, the selection of GAC with a large effective size and small
uniformity coefficient was required to control head loss development.
Initial pilot plant studies were performed at the Kelseytown Reservoir.
Three  6-inch diameter columns were used to compare the performance of
10x30  GAC with dual and multimedia filters (Table 4).  Effluent turbidity

       TABLE 4.  CONNECTICUT WATER COMPANY:  PILOT COLUMN CHARACTERISTICS(7,8)
 Type
Layer  Material
Specific  Depth    Uniformity  Effective
Gravity  (inches) Coefficient  Size (mm)
Dual Media


Multimedia



GAC:
Kelseytown
Reservoir
GAC:
Naugatuck

1
2
3
1
2
3
4
1
2
3
1
2
3
Coal
Sand Support
Gravel
Coal (Ms4)
Sand (Msl8)
Fine Garnet (Ms7)
Garnet Gravel (Msll)
ICI 10x30 GAC
Sand Support
Gravel
ICI 8x16 GAC
Sand Support
Gravel
1.57+
2.63+
2.65+
1.5+
2.6+
3.8+
3.8+
mm
2.6
2.6+
_
2.6
2.6+
26
10
3
16.5
9
4.5
3
48
6
3
48
6
3
<1.5
<1.35
N.A.
<1.7*
1.5*
<2.0
N.A.
<1.7**
<1.5
-
<1.3**
<1.5
N.A.
0.95
0.45
2.4 to 4.8
1.0 to 1.2
0.45 to 0.50
0.20 to 0.30
N.A.
0.70 to 0.85
0.4 to 0.5
2.4 to 4.8
1.35**
0.4 to 0.5
2.4 to 4.8
 *Data obtained from Neptune Microfloc
 **Data obtained from ICI Industries
                                      196

-------
from the 10x30 GAC filter (e.s. 0.70 mm, u.c. < 1.7) was generally com-
parable to or better than that from the corresponding dual and multimedia
filters.  Head loss gradient versus depth profiles indicated that the
majority of suspended solids removal was occurring in the top 2 inches of
the GAC.  Greater depth filtration and reduced head loss development in the
top 3 inches of the dual and multimedia filters indicated that signifi-
cantly longer filter runs could be obtained with the dual or multimedia
filters than with the 10x30 GAC.

    Further testing at the Connecticut Water Company Naugatuck treatment
plant included the same dual and multimedia pilot columns examined at
Kelseytown; however, the fine  10x30 GAC (e.s. 0.70 to 0.85 mm, u.c. < 1.7)
was replaced with a coarser and more uniform 8x16 GAC (e.s. 1.3 mm, u.c.
1.4) to reduce the rate of head loss development and extend GAC filter runs.
For both the 4 gpm/ft2 and 6 gpm/ft2 rates tested, the dual and multimedia
filters reached the terminal head  loss values much sooner than the filter
with 8x16 GAC.  The 4 gpm/ft2  runs for the dual and multimedia filters were
terminated after only 24 hours with head loss approaching 8.3 feet; the GAC
filter  run was also terminated at  24 hours, but the total head loss was only
3  feet.  Effluent turbidities  from the GAC filter remained below 0.2 Ntu
for the entire duration of the test, while turbidity breakthrough began in
the multimedia filter at 19 hours.  Similar  results were obtained for tests
with a  filtration rate  of 6 gpm/ft2.

    As  a result of pilot testing at the Connecticut Water Company, 48 inches
of 8x16 GAC  (e.s. 1.3,  u.c. 1.4) were used.  The selection of the larger and
more uniform media,  in  this case,  reduced head  loss development to acceptable
levels.  The potential  adverse effect of the large media size on effluent
quality was mitigated by the  selection of a  larger media depth.

    The large GAC uniformity  coefficient does not appear to significantly
affect  the performance  of AWWSC sand  replacement plants.  With adequate
pretreatment, operators  report no  trouble in achieving the desired filter
run lengths; the average filter run length at these plants is 55 hours.
However, losses of carbon fines during  initial  backwashings may lower the
uniformity coefficient  and  raise the  effective  size so that adequate
filter  run lengths are  attained.

ROLE OF A SAND LAYER BELOW GAC

    Monomedia GAC may be an effective filtration medium; however, a sand
layer  is often recommended below the  GAC as  an  added barrier to floe
penetration  (9), particularly  when a  GAC with a large effective size  is
used.   The use of a  sand layer below  GAC can improve filtered water quality,
as floe penetrating  the upper  layers  of the  GAC may be removed by the
underlying sand layer,  and may also lead to  an  increase  in filter run length
(10).   The use of sand  below  the GAC  may cause  additional problems in
replacing or regenerating the  GAC  and limit  the adsorptive capacity of  sand
                                     197

-------
replacement systems because the presence of sand reduces the total  carbon
depth.  These factors must also be considered.

    When used, proper sizing of the sand and GAC media is essential to
insure adequate cleaning.  Conventional  sizing  methods and common rules of
thumb ignore the large uniformity coefficient and the angularity of the GAC
media.  An approach to media sizing and  backwash rate selection that con-
siders these two factors has been developed by Cleasby (11).  To insure
complete expansion of both the sand and  GAC layers, Cleasby suggests that
the minimum fluidization velocity of Dgn sand and GAC particles be equal.
The equations of Wen and Yu (12) eliminate the sphericity (which is a
measure of media angularity) from the Ergun equation to obtain the
following:


         RMF = [(33.7)2 + 0.0408Ga]°'5 - 33.7                           (1)

where

               Deq vmf p
                                                                        (2)
                  H

and

               Deq3 P(Ps~P)9
          Ga = —	2~	                                            (3)


where

         RMF = Reynolds number at the minimum fluidization velocity
         Ga  = Galileo number
         Deq = equivalent diameter
         Vmf = minimum fluidization velocity
         ps  = particle density
         p   = fluid density
         u   = dynamic viscosity


     Dgo may be substituted for Deq in the above equations for determination
of  the minimum fluidization velocity (Vmf).  Backwashing at a rate equal to
1.3 times the minimum fluidization velocity should give a five to 10 percent
expansion of the coarse grains which will allow for an effective backwash
 (13).  Figure 1 is a plot of backwash velocity, equal to 1.3 x Vmf for Dgo,
that may be used to determine the appropriate backwash velocity and the
     for the GAC and sand for a backwash water temperature of 20°C.
 Filter  Cleaning

     The use  of the  low density and highly angular GAC media may accelerate
 problems with mud accumulations and carbon  loss.  Mudball formation was
                                    198

-------
cr
g 0>
CXQ
30

28 -

2G -

24 -

22 -

2O -
~g   18 H
1 6 -

14 -

12 -

1O -

 8 -

 6 -

 4 -

 2 -

 O
u  e
o >
H
(11 X
(0 i-l
(0
3 II
X •"
O
It)
                                Sand
                                              GAG  (Wetted  in  Water)
              ~i	1	1	1	1	1	1	1	1	1	1	1	r
         O        O.4      O.8      1.2       1.6       2        2.4


                          Equivalent Particle Diameter, Dgo  (mm)
                                                                   2.8
 Figure 1.  Proposed  method for determining  the  appropriate media sizing  and
            backwash  rate for a water temperature of 20°C.
                                       199

-------
reported in 53 percent of the AWWSC plants  surveyed,  although  in  most cases
a significant effect on filter performance  was not observed.   One plant,
however, had a severe mudball problem.   Carbon loss in excess  of  2 inches
per year was noted at a few plants.  Knowledge of factors affecting filter
bed expansion and filter cleaning may be used to develop operational
schemes which reduce carbon loss while permitting good filter  operation.

    GAC particle size distribution and GAC  wetted particle density vary
widely between carbon brands, and also between different batches  of the same
carbon.  Appropriate backwash rates, based  on manufacturer recommendations
for the carbon, need to be used to insure adequate cleaning.  In  addition,
backwash rates need also be adjusted for variations in backwash water
temperature to avoid overexpansion of the filter and media loss.   The use
of surface washing or air scouring is recommended as a means to provide
the auxiliary energy needed to properly clean the filter.

    Monitoring of filter condition is essential to insure that backwashing
procedures are effective for filter cleaning.  Mudball formation and
excessive  sand and GAC intermixing can be readily determined through visual
analysis of core samples.  Where the depth of the filter media is large so
that a  core may not be obtained, observations for mudballs may be made by
digging to obtain samples.

   ADSORPTION EFFICIENCY:  FILTER-ADSORBERS VERSUS POST-FILTER ADSORBERS

    The success of  filter-adsorbers for the removal of tastes and odors has
been well  documented, while  post-filter adsorbers have proven effective for
 the removal of total organic carbon (TOC), volatile organic compounds
 (VOCs), and many specific contaminants.  It is  reasonable to question
whether filter-adsorbers can be  used to meet  the  increasingly more
 stringent  water quality  standards  for specific  contaminants because  their
 capital cost  is  lower  than that  for post-filter adsorbers.  As with  any
 system, the merits  and  liabilities which exist  for both  filter-adsorber and
 post-filter adsorber must be considered to make a prudent process  selec-
 tion.

 Filter-adsorbers:

          o Can  be  installed  readily as  a retrofit of  existing  filters;
          o Have  lower capital  costs than post-filter  adsorbers;
          o Require  less  land for construction.

 However,  filter-adsorbers;

          o Must be  backwashed more frequently than post-filter adsorbers;
          o May be limited to short empty bed contact times  (especially sand
            replacement systems);
          o May incur greater carbon losses because of more  frequent
            backwashing;
          o May have greater operational  costs because less  organic matter
            can be adsorbed per unit weight of carbon.
                                     200

-------
Whereas post-filter adsorbers;

         o Have greater flexibility in media size that can be used, as head
           loss and filtration considerations no longer constrain effective
           size or uniformity coefficient;
         o Are backwashed less frequently, and possibly may maintain better
           stratification;
         o May provide an additional barrier against microbial
           penetration (14);
         o Are more compatible with other processes, such as ozonation (15);
         o Can be designed for easy replacement of carbon;
         o Utilize more adsorption capacity of the carbon.

    The feasibility of GAC filter-adsorbers depends greatly upon the system
design  (including the type of carbon and the empty bed contact time), the
type of organics to be removed, and the extent of removal that is needed, as
well as the availability of  other treatment alternatives that could the same
tasks.

EFFECTIVENESS OF GRANULAR ACTIVATED CARBON FOR TASTE AND ODOR CONTROL

    Experience with operating systems  has shown that GAC sand replacement
filter-adsorbers are effective for taste and odor control.  Strong musty-
earthy  odors at Mt. Clemens  (16) were  effectively controlled by GAC filtra-
tion, with GAC life ranging  from three to four years, although the odor
intensity entering the GAC beds was not documented.  At Granite City, musty-
moldy odors attributed to actinomycetes and algae metabolites (17) are
controlled through GAC filtration with GAC bed life up to three years.
Experience at other American Water Works Service company plants has
likewise been favorable.  Filtration through 15 to 30+ inches of GAC con-
sistently reduces tastes and odors to  acceptable levels, with GAC replace-
ment frequencies of one to five years. The plant operators are generally
satisfied with the consistency with which GAC filter-adsorbers control
taste and odor.  Some plants use powdered activated carbon  (PAC) preceding
GAC during extreme conditions.

    More frequent replacement or regeneration of GAC is required at a few
selected plants.  At Regina, Saskatchewan (18), sour, musty, grassy, and
septic  odors from algal blooms produce threshold odor numbers (TONs) as
high as 50 in the water applied to the GAC post-filter adsorbers that have
an EBCT of 20 minutes.  The  GAC is  regenerated after one taste and odor
season  which is typically five months  long; a shorter bed life has also
been observed when odors are intense.  At Nitro, West Virginia (19) GAC
filter-adsorbers were used in the  1960s to remove tastes and odors
resulting from industrial organic wastes.  The GAC  influent had threshold
odor numbers ranging from 40 to 400.   GAC (e.s. 0.80 to 0.90) produced odor-
free water for as long as 30 days depending on the  contact  time  (range 3.8
to 15 minutes).  GAC with a  mesh size  of  20x50 produced odor  free water  for
twice as  long.  The Nitro plant has since been replaced, and  a source water
not  requiring GAC is used.
                                     201

-------
    Compared to the GAC bed life for total  organic carbon,  trihalomethanes,
and other contaminants (which may be several  weeks to a few months), the
bed life for tastes and odors is typically  much longer (one to five years).
The extended bed life for tastes and odors  may be attributed to low
influent concentrations.  Degradation of taste and odor producing compounds
by microorganisms within the filter may also  remove a large fraction of
these compounds, thereby lengthening bed life.  In addition, taste and odor
compounds may be desorbed in undetectable concentrations over the life of
the filter, thereby leaving additional  capacity for periodic high concen-
trations of incoming species.  Alternatively, some taste and odor com-
pounds, such as geosmin and methylisoborneol  may be strongly adsorbed (20),
although the capacity of the carbon for these compounds is  significantly
reduced when humic acids were present.

FILTER-ADSORBERS FOR THE REMOVAL OF OTHER ORGANICS

    Filter-adsorbers can be designed with empty bed contact times (EBCTs)
similar to post-filter adsorbers, although  other differences in adsorption
performance may persist due to variations in  methods of operation.  However,
carbon depths  in sand replacement filter-adsorbers are often restricted by
the depth of the existing filter box and the  requirement in some states that
sand  remain below the GAC. In addition, filtration rates cannot be reduced
to increase the EBCT without adversely affecting water production.  As a
result, the EBCTs of most sand replacement filter-adsorbers are often con-
siderably less than those of post-filter adsorbers.  The average EBCT for
AWWSC  sand replacement plants is 8.6 minutes, with a range of 3.2 to 24.8
minutes.  For many taste and odor applications, a short EBCT is adequate,
thus  filter-adsorbers are more economical.   However, where short EBCTs and
contaminant characteristics lead to excessively frequent regeneration or
replacement of GAC, the use of post-filter adsorbers may be more economical.

    In general, the short EBCT associated with sand replacement filter-
adsorbers may  prohibit their use for large removals of TOC and weakly
adsorbed organics.  For example, with an EBCT of nine minutes, GAC may be
effective for  only several weeks to a few months for a typical goal of 50
percent TOC removal or 20 percent chloroform removal.  For filter-adsorbers,
replacement of GAC on such a frequency may be costly and difficult as
filter-adsorbers are often not well designed  for rapid GAC replacement.
The short EBCTs associated with sand replacement systems also may lead to a
higher carbon  usage rate.  Carbon systems designed with larger EBCTs allow
for the mass transfer zone to occupy smaller  and smaller percentages of the
total  carbon bed and thus lower technical carbon usage rate (see Tables 5
and 6).
                                    202

-------
                                  TABLE 5.  EFFECT OF EBCT ON 50 PERCENT  TOC  REMOVAL
Location
Cincinnati, OH


IX)
CD
00 Miami, FL


Manchester, NH
Philadelphia, PA
Column Type
PFA
PFA
PFA


PFA
PFA
PFA
PFA
FA
GAC
WV-G 12x40
WV-G 12x40
WV-G 12x40


Flltrasorb 400
Flltrasorb 400
Flltrasorb 400
WV-W 8x30
Flltrasorb 300
EBCT
4.4
7.2
17.8


6.2
12.4
18.6
22
15
Influent
TOC (mg/1)
1.3-3.0
1.3-3.0
1.3-3.0


5-6
5-6
5-6
1.8-2.8
1.6-2.8
T(C/CO=0.5)
(days)
10
30
110


8
22
50
24
35
Bed Volumes
Processed Ref.
3,273 30
6,000
8,899


1,858 29
2,555
3,871
1,571 31
3,360 19
Note:  PFA:  Post-Filter Adsorber
       FA:   Filter-Adsorber

-------
          TABLE 6.  EFFECT OF EBCT ON 50 PERCENT CHLOROFORM REMOVAL
Average Influent
CHC13 Cone.
Location (ug/1)
Cincinnati, Ohio 30



Miami, Florida 20




EBCT
(min.)
3.2
7.5
11.8
16
6.2
12.4
18.6
24.8

T(C/C0)=0.2
(Days)
7
25
49
70
14
35
59
84

Bed Volumes
Processed
3,150
4,800
5,980
6,300
3,250
4,060
4,570
4,880


Ref
27



28



FILTER-ADSORBERS:  EARLIER BREAKTHROUGH AND  LOWER ORGANIC LOADINGS

    Where filter-adsorber and post-filter adsorber designs have comparable
EBCTs, field- and pilot-scale studies have found reduced organic loadings and
earlier breakthrough.  These may lead to an  increased frequency of carbon
regeneration or replacement and a higher carbon  usage rate for filter-
adsorbers.  These additional operational  costs  should be considered when
selecting the mode of GAC installation.

Case Study;  Jefferson Parish (21)

    Brodtmann et al. (21) compared the performance of a filter-adsorber (EBCT
= 14 minutes) and an adsorber (EBCT = 26 minutes) at Jefferson Parish.  Both
units were operated at a filtration rate of  0.35 mgd.  The filter-adsorber
was backwashed 80 times over the course of the  experiment, whereas the post-
filter was backwashed only 45 times.

    A plot of the cumulative loading  (urn substance adsorbed/lb GAC) versus
time allows for a direct comparison of the carbon in the adsorber and
filter-adsorber performance to be made.  The maximum loading on the GAC is
not a function of bed depth or EBCT.   Figures 2  through 5 show the cumula-
tive loading capacity of the GAC for  total  trihalomethanes, chloroform,
total tnhalomethane formation potential  (TTHMFP), and dichloroethane.  At
saturation, the adsorber cumulative loading  is  significantly greater than
that of the filter-adsorber, indicating that the carbon usage rate will be
greater for the filter-adsorber if it is to  be   operated to control any of
these substances.  The maximum trihalomethane loading,  for example, is about
50 percent of that for the post-filter adsorber.   The dichloroethane loading
curves are somewhat different than the rest  in  that they show more adsorp-
tion on the filter-adsorber early in  the run, and significant desorption
near the end of the run.   At run termination, the filter-adsorber has about
                                     204

-------
   «J
   Q
   O
   4J ---
   * O
   o
   J £1
     i-H
   0) -..
   > s
   -M 3
   4J —
   <0
  CJ
                                             *--•* FIT. AO*
           o  w xo
                   I   I   I  I   I   I  I   I  I   I   I   I  I
                   tO  40M  40  70  !o»0  IOC i«5tfO410 HO  OO
                                 Days
Figure  2.  Loading capacity of  the Phase  IIA GAC beds at
            Jefferson  Parish for TTHM(21).
   0)
   4J


   Q


   O
   10 O
   o
   J XI
     .H
    s
   •H a
   •3
   u
                                                 -«°--°»"«»-o
           I   I   I  I   I  I   I   I  I	I	1	l_
           O  IO  to  *O
                        3O  «o TO  to
                                             _1	1	I   I	1_
                                           IZO I3O  HO  IX) I«O  ITO
                                 Days
Figure 3.   Loading capacity of the Phase IIA GAC  beds at
             Jefferson  Parish for chloroform(21).
                               205

-------
  a)
  4-1
  o
  4-1 —
    u
  •a <
  ro o
  o
  a> -^
  > S
  -H 3
  4-) —
  "3
  <-\
  a
  6
  a
  o
           I  I   I  I   I  I   I  1   I  I   I   I  I
                                                 IM ««o  ira
                               Days


Figure 4.   Loading capacity  of the Phase  IIA GAC beds at
            Jefferson Parish  for TTHMFP(21).
  01
  4J
  O
  4J ^-~
    O
  T) <
  (0 O
  O
  0) \
  > s
  -H D
  3

  =1
  O
                       »  «o  ro  tf3
                                   IOO  IK} IXO
                                                1»O  ICO I TO
                               Days


 Figure 5.  Loading  capacity of the  Phase IIA GAC  beds at
            Jefferson  Parish for  Dichlorethane(21).
                              206

-------
30 percent less capacity than the post-filter adsorber.  Therefore, for any
given EBCT, the filter-adsorber carbon needs to be regenerated or replaced
more frequently.  Brodtmann et al. (21) hypothesized that the more frequent
backwashing (nearly twice as frequent) of the filter-adsorber may have
contributed to the above effects, but more extensive competition may have
also been a factor.

Case Study;  Philadelphia (22)

    Cairo et aJ. (22) analyzed the performance of a pilot filter-adsorber
(EBCT = 15 minutes) and a post-filter adsorber (EBCT = 15 minutes) for TOC
and chloroform adsorption.  Chloroform breakthrough was observed in the
filter-adsorber beginning from the first week of operation.  In contrast, the
effluent concentrations of chloroform from the contractor did not become
significant until week six.  Comparison of the filter-adsorber and adsorber
on a mass loading basis (Figure 6) yields results similar to those obtained
by Brodtmann et al. (21).  At saturation, the adsorber removed approximately
20 percent more chloroform than the filter-adsorber; the adsorber also
removed approximately 11 percent more TOC than the filter-adsorber.  TOC
concentration in the effluent from the filter-adsorber column was also
significantly higher than that from the adsorber until saturation, although
influent TOC concentrations were comparable for both filter-adsorber and
post-filter adsorber.  Judgements must be made to determine if the addi-
tional removal efficiencies would be worth the additional capital cost of
the post-filter adsorber.

    Cairo et al. (22) hypothesized that blinding of the GAC macropores by
floe  in the filter-adsorber may have contributed to the decreased adsorptive
capacity; turbidities ranging from 1 to 6 Ntu were applied to the filter-
adsorber system, whereas  turbidities going onto the post-filter adsorber
ranged from 0.1 to 0.3 Ntu.   In addition, because a higher solids loading
was applied to the filter-adsorber, the filter-adsorber system had to be
backwashed every 2.5 days; the adsorber was never backwashed during the
experiment.  Redistribution of particles within the bed, and a subsequent
destratification and elongation of the mass transfer zone was deemed more
likely in  the filter-adsorber, although analyses of particle size versus
depth were not performed  to evaluate this claim.

IMPORTANCE OF FILTER STRATIFICATION

    GAC filter-adsorbers  are  often designed with a  large uniformity coef-
ficient  (greater than 1.9) to promote  stratification of the GAC after back-
washing and to maintain the adsorption front.  Good stratification promotes
sharp adsorption fronts and lower  carbon usage rates.  Poor stratification
results in the  redistribution of particles, equilibrated with higher con-
taminant concentrations in the upper  layers of the  bed, throughout the  lower
layers of  the filter.  This may  result in desorption of weakly adsorbed  con-
taminants, leading to earlier contaminant breakthrough and higher  carbon
usage rates.

    Limited data are available regarding the maintenance of stratification
within adsorbers,  although the general assumption has  been that  stratifica-
tion  is occurring.   Data  obtained  from an AWWSC, plant, however,  shows  that


                                     207

-------
good stratification of the filter media does not always  occur.   Figure 7 is
a graph of the mean particle diameter for samples taken  from various depths
within a filter-adsorber for a number of service times.   These  data suggest
that the filter-adsorber is poorly stratified.   For most of the service
times, there is no significant increase in mean particle diameter with
depth, and in a few instances, the mean particle diameter at a  given depth
is larger than that of an underlying layer.

    Core samples were taken from two filters at the Hays Mine plant
(Table 7) to obtain additional data on the extent of stratification.  Filter
31 contained three-year old GAC and Filter 35 contained  six-month old GAC.
Examination of the mean particle diameter versus depth shows that both
filters are generally stratified, although some mixing is taking place in
two of the samples from Filter 31.  It is believed that  the large fraction
of sand present in sample 31L-B biased the particle size distribution; the
sand was not separated from the GAC prior to analysis and hence gave a
somewhat lower mean particle diameter for the bottom third of the bed.
However, the presence of a large amount of sand in the bottom one-third
sample might be attributable to mixing as well.

    Although the filters at Hays Mine appear to be stratified based on the
mean particle diameters, the grain size analyses (Table  7) also show that a
significant fraction of fines are found even in the lowest layers of the
filters, suggesting that some partial mixing of GAC may  be occurring.  This
partial mixing may explain the disparities in the performance of filter-
adsorbers and post-filter adsorbers documented by Brodtmann (21) and
Cairo  (22).  Pilot post-filter adsorbers using 10 feet of 8x30 carbon at
Regina, Saskatchewan, Canada  (18), have fewer fines present in the lower
layers  (Figure 8).

    Additional studies, carefully documenting the particle size versus
depth  for filter-adsorbers and post-filter adsorbers operated in parallel
are required to clearly evaluate the effect of backwashing on mixing.  Mean
particle diameter versus depth, as well as the quantity of fines and sand
at each depth, needs to be evaluated.  The adsorption-desorption behavior
of both weakly and strongly adsorbed compounds should also be investigated
as a  function of mixing to determine the effect of degree of stratification
on effluent quality.

                      MICROBIOLOGICAL ACTIVITY ON GAC

    In Europe, microbial proliferation is often encouraged by ozonation
prior to GAC, as ozonation makes  some nondegradable compounds biodegradable
and thereby provides an additional means of organic removal  (7).
Biodegradation may also occur without preozonation depending on the  nature
of  the organic compounds  in the water.  Reduction of  the  concentration  of
biodegradable compounds  in  the treatment plant also reduces  the likelihood
of microorganism growth  in  the distribution system where  they are more  dif-
ficult to control  (23).  Although advantageous for organic  removal,  microbial
growth has  raised  concern about potential health  risks, because microorga-
nisms on GAC  could include  indicator organisms, microorganisms that  are
                                     208

-------

o
ft
o
6
Cn
\
•0
0)
>
0
6
0)
Pi
on
«— i
U
ac
o
CP


1
1
1
1



0
0
0

0

-•- Adsorber -°- Filt . /Ads .
2 T
.8 •
.6-
.4 •
.2 •

1 -

.8-
.6-
.4 •

.2 •
n .
•-•"**
..•-•-•-•-•— «'*^
.^*
^9 f\
>•'* ^00-0-0-0-0-0-0-0-0-^°
^r f\^^^
J* r»-O-°
•X ^


-------
      TABLE  7.   HAYS MINE PLANT, PITTSBURGH, PENNSYLVANIA:
                       SIZE DISTRIBUTION VERSUS DEPTH
GAC PARTICLE
                                 Mesh Size
     Mean Particle
Sample
Filter 31L-A
Top
Middle
Bottom
Filter 31L-B
Top
Middle
Bottom*
Filter 31L-C
Top
Middle
Bottom
Filter 31L-D
Top
Middle
Bottom
Filter 35L-A
Top
Middle
Bottom
Filter 35L-B
Top
Middle
Bottom
Filter 35L-C
Top
Middle
Bottom
Filter 35L-D
Top
Middle
Bottom
12

0.2
0.4
0.5

0.7
0.4
0.3

0.6
1.2
0.9

0.6
0.6
0.6

0.3
0.5
1.0

0.2
0.5
1.4

0.3
0.8
0.7

0.2
0.5
2.4
14

1.8
8.4
19.1

1.8
5.4
5.9

5.6
12.9
29.8

3.0
9.4
11.3

1.3
4.1
20.7

1.2
12.2
26.1

0.8
10.5
31.0

1.1
7.6
32.2
16

8.6
22.0
31.0

4.5
8.4
13.7

10.4
34.7
43.3

10.4
31.7
31.4

3.2
15.2
33.7

2.5
25.4
27.8

4.3
23.2
37.7

3.9
23.4
27.3
20

42.8
55.6
40.8

32.6
38.3
27.7

36.9
45.1
22.2

50.7
48.7
45.7

34.1
62.5
36.9

36.1
50.4
33.2

48.7
58.0
22.5

42.1
55.6
24.4
30

35.6
12.4
5.5

48.6
40.8
27.7

36.6
4.9
1.7

28.8
7.8
6.9

50
16.8
3.9

49.8
10.3
4.7

35.1
6.5
2.0

44.2
11.7
4.5
40

9.6
1.0
2.7

11.6
6.0
23.1

9.2
0.8
1.6

6.1
1.5
3.7

10.7
0.7
3.4

10.0
0.9
5.5

10.6
0.6
5.1

8.4
0.8
8.1
-40 Diameter (mm)

1.4
0.2
0.4

0.6
0.7
1.6

0.7
0.4
0.5

0.4
0.4
0.4

0.4
0.2
0.4

0.2
0.3
1.2

0.2
0.4
1.0

0.1
0.4
1.1

0.89
1.09
1.18

0.84
0.92
0.88

0.78
1.1
1.29

0.95
1.13
1.13

0.83
1.03
1.20

0.83
1.12
1.20

0.88
1.12
1.26

0.86
1.09
1.23
* Sample contained sand.
                                     210

-------
                 60
                   4  8 12 16 20 30 40
                 40
                 20
                 40-I
                 20
40
20
 0
40
20
 0
40
20
 0
                           n
        I®
                            ^ -
  Depth

8" (7% down)


21 (20% down)


3*4" (33% down)


6'8" (67% down)


10' (100% down)
                    4  8 12 1620 30 40
                       Sieve Sizes
Figure 8.  Regina,  Saskatchewan:   GAC particle size distribution
           versus depth(18).
                               2\\

-------
resistant to disinfection, opportunistic pathogens,  and antagonists to coli-
form detection (24).  These microorganisms may pass  into the filter effluent
if they are sheared from the GAC particles or if GAC fines with attached
microorganisms are released into the effluent (25).   The number of orga-
nisms in GAC effluent does decrease with increasing  bed depth, however
(26).  The reduction of organic matter and ammonia in the GAC adsorber
should reduce disinfectant demand in the filter effluent, but carbon fines
could interfere with disinfection if they are present (27).  Microbial
growths on filter-adsorbers can cause increased rates of head loss build-up
and short filter runs if the growth is excessive and the backwash system is
not able to control the growth.

    Chlorine is often in the water applied to GAC because the contact time
during sedimentation is needed for adequate disinfection.  There i^also the
desire to minimize bacterial growth on GAC, to minimize the number pf bac-
teria in the adsorber effluent, and to improve filter performance../ Post
disinfection is effective in reducing the number of  organisms to acceptable
levels before entry to the distribution system (24)  (as measured by conven-
tional methods), but in some plants adequate time for post disinfection is
not available and thus predisinfection is necessary.

    Rapid head loss development due to bacterial growth on the carbon surface
is often reduced by influent chlorination; however,  more frequent backwash-
ing can in some cases be equally as effective.  The  frequency of backwashing
required and its effect on water production will determine if the practice
of continuously applying a chlorinated influent to GAC for this purpose
could be eliminated in plants where sufficient contact time is available for
disinfection.  Intermittent application of chlorine, in the form of a
superchlorinated backwash, could be used in some instances to control head
loss development.

    Several important questions remain concerning the practice of continu-
ously applying chlorine to GAC, which can only be answered by further
research, including:  1)  If no chlorine is applied to GAC sand replacement
filter-adsorbers, will the concentration of microorganisms in the effluent
be too high to be adequately controlled by a good post disinfection proc-
ess?  The short EBCT of sand replacement processes may be a contributing
factor to high concentrations.  2) Does the oxidation-reduction reaction
between chlorine and GAC  result in the production of fines which pass into
the effluent and make the effluent difficult to disinfect?  3) Will the
chlorine react with adsorbed compounds to produce new contaminants that
can be leached into the product water, as has been shown to occur  in  the
laboratory  (28,29,30)?  In view of the potential benefits of microbial
activity on GAC and the possible adverse effects of applying chlorinated
water to GAC, these questions  should be resolved.

    Carbon fines have been detected in filter-adsorber effluents.  None of
the systems studied have  shown any significant bacterial regrowth  in  the
distribution system, but  bacteria attached to these particles are  very dif-
ficult to kill with chlorine (27).  Additional  research  is needed  to  determine
whether the problem of fines is more severe  in GAC adsorber effluent  than
in the effluent of  other  rapid filters, the  factors which cause the  produc-
tion of fines, and  the effect  of filter design  including type of GAC, EBCT,


                                    212

-------
backwash frequency, and underdraln design, on the passage of fines to the
effluent.
                                CONCLUSIONS

1.  GAC as a total or partial replacement for sand is as effective, or more
    effective, for turbidity removal than conventional filtration media
    (both single and dual media) provided that an appropriate media size
    has been selected.  When the GAC effective size is greater than
    approximately 0.80 to 0.90 mm, an increase in GAC depth or the use of
    another media below the GAC is probably required for good filtration
    performance.

    GAC typically has a larger uniformity coefficient (<2.4) than
    conventional filtration media  (<1.6).  The large uniformity coefficient
    results in a more rapid rate of head loss development because of the
    layer of fine carbon particles at the surface of the filter-adsorber
    which promote a  surface filtration.  Backwashing may remove some of
    these small particles and thus reduce the rate of head loss development.
    However, where filter run lengths are critical for successful operation
    it  is recommended that the filter design be carefully piloted to insure
    the media has been properly sized.

2.  GAC filter-adsorbers can consistently eliminate tastes and odors from
    drinking water supplies for extended periods of time, typically one to
    five years.  Where odors are intense or where extensive competition
    from other organics occurs, the bed life may be considerably shorter.
    Additional research examining  the effect of competition between com-
    pounds causing tastes and odors and other contaminants is desirable.

3.  Sand replacement filter-adsorbers do not function well for removing
    less strongly adsorbed compounds, such as trihalomethanes, volatile
    organics, and fractions of the total organic carbon.  The short empty
    bed contact times for sand replacement systems, typically nine minutes,
    would require very frequent regeneration or replacement of GAC.
    Replacement or regeneration at a high frequency would be operationally
    cumbersome, and  carbon loss during regeneration could be a significant
    operational cost.  Where filter-adsorbers are designed with empty bed
    contact times similar to post-filter adsorbers, earlier contaminant
    breakthrough, lower organic loadings, and a higher carbon usage rate
    are observed for the filter-adsorbers.  A thorough economic analysis
    would be needed  to determine whether the costs associated with a higher
    carbon usage rate in the filter-adsorber, would negate the savings in
    reduced capital  costs.

4.  More research is needed to determine whether microbial growth in sand
    replacement filter-adsorbers is beneficial or detrimental and whether
    chlorine should  be applied to  GAC filter-adsorbers to control growth.

                              ACKNOWLEDGMENT

    This work was supported by the American Water Works Association
Research Foundation, project number 109-85, the American Water Works
Service Company, and the University of Illinois.

                                    213

-------
                                 REFERENCES

1.  Graese,  S.L.,  Snoeyink,  V.L,  and  Lee,  R.G.   GAC  filter-adsorbers.
    Report to the  American  Water  Works  Association Research  Foundation,  Denver,
    Colorado, 1987.

2.  Hyde, R.A.,  Hill,  D.G.,  and label,  T.F.   Granular activated carbon as
    sand replacement in rapid gravity filters.   Water Research Centre,
    Stevenage, England, 1984.

3.  Whitford, C.F. and McCawley,  R.   The use of GAC  as a filtration medium
    in the TWA project.  Paper 14,  Symposium on the  Trent-Witham-Ancolme
    Potability Study,  Anglian Water Authority/Thomas Ness.   December 16-17,
    1981.

4.  James M. Montgomery, Consulting Engineers,  Inc.   Contra  Costa Water
    District:  preozonation/deep  bed  filtration pilot plant  study.
    September, 1986.

5.  Love, O.T., Jr. and Symons, J.M.   Operational  aspects of granular
    activated carbon adsorption treatment.  U.S. Environmental Protection
    Agency, Water Supply Research Division.   June,  1978.

6.  Cleasby, J.  |£:  W. J. Weber,  Jr., Physiochemical Processes for Water
    Quality Control.  John Wiley and Sons, Inc., 1972.

7.  Metcalf and Eddy,  Inc.  The Connecticut Water Company,  Clinton,
    Connecticut:  report on the pilot plant treatment studies at Kelseytown
    Reservoir.  October, 1975.

8.  Metcalf and Eddy,  Inc.  Report to the Connecticut Water Company upon
    evaluation of Naugatuck pilot filter tests.  April 27,  1986.   ,

9.  Great Lakes - Upper Mississippi River Board of State Sanitary Engineers.
    Recommended standards for water works.  Health Education Service, Albany,
    New  York, 1982.

10. Kornegay, B.H.  Control of synthetic organic chemicals by activated
    carbon -- theory,  application,  and regeneration  alternatives.  Presented at
    the  Seminar on Control of Organic Chemical  Contaminants in Drinking Water,
    Atlanta, Georgia.  February 13-14, 1979.

11. Cleasby, J.  Personal communication.  September, 1985.

12. Wen, C.Y. and Yu,  Y.H.  |n:  J. L. Cleasby and K. Fan,  Predicting
    fluidization and expansion of filter media.  J.  Envir.  Eng. Div.,
    ASCE, 107, EE3, 455.  June, 1981.            ~

13. Cleasby, J.L. and  Fan, K.  Predicting fluidization and expansion of
    filter media.  J.  Envir. Eng. Div., ASCE, 107, EE3, 455.  June, 1981.
                                     214

-------
14. Lykins, B.W., Geldreich, E.E., Adams, J.Q., Ireland, J.C., and
    Clark, R.M.  Granular activated carbon for removing nontrihalomethane
    organics from drinking water.  U.S. Environmental  Protection Agency,
    September, 1984.

15. Wiesner, M.R., Rook, J.J., and Fiessinger F.   Optimization of organic
    removal through the water treatment process:   sand replacement or
    post-adsorber for granular activated carbon filtration?  Laboratoire
    Central, Lyonnaise des Eaux, 1986.

16. Hansen, R.E.  Problems solved during 92 months of operation of
    activated granular carbon filters.  93rd Annual AWWA Water Qual.  Technol.
    Conf., Atlanta, Georgia.  December 7-9, 1975.

17. American Water Works Association Research Foundation.  Handbook of taste
    and odor control experiences in the U.S. and Canada.  American Water  Works
    Association, Denver, CO,  1976.

18. Gammie, L., and Giesbrecht, G.  Full-scale operation of granular
    activated  carbon contractors at Regina/Moose Jaw, Saskatchewan.  Paper
    presented  at the 1986 AWWA Annual Conference, Denver, Colorado.  June
    22-26,  1986.

19. Dostal, K.A.,  Pierson,  R.C., Hager, D.G., and Robeck G.G.  Carbon bed
    design  criteria study at  Nitro, West Virginia.  J. AWWA.  57:5, 1965.

20. Herzing, D.R., Snoeyink,  V.L., and Wood, N.F.  Activated carbon
    adsorption of  the  odorous compounds 2-Methylisoborneol and Geosmin.
    J. AWWA.   April, 1977.

21. Brodtmann,  N.V. Jr., DeMarco,  J.,  and  Greenberg, D.  Critical study
    of large-scale granular activated  carbon filter units for the removal
    of organic substances from  drinking water.  jji:  Activated Carbon
    Adsorption of  Organics  from the Aqueous Phase.  Ann  Arbor Science
    Publishers,  Inc.,  1980.

22. Cairo,  P.R.,  Radziul, J.V.,  Coyle, J.T., McKeon, W.R., Hannah, R.E.,
    Pence,  M.M.,  and Suffet,  I.H.  Development of  criteria for the design of
    full  scale carbon  adsorption  systems.   Proceedings  of the AWWA Water
    Quality Technol. Conf., Louisville,  Kentucky.   1978.

23. Bourbigot, M.M., Dodin, A.,  and Lheritier, R.   Limiting  bacterial
    aftergrowth in distribution system by  removing  biodegradable  organics.
    Paper presented at the  1982 AWWA  Annual Conference,  Miami, Florida.
    1982.

24. Symons, J.M.,  Stevens,  A.A.,  Clark,  R.M.,  Geldreich,  E.E., Love,  O.T.,
    and DeMarco,  J.  Treatment  techniques  for  controlling trihalomethanes in
    drinking water.  EPA-600/2-81-156, U.S. Environmental Protection  Agency,
    Cincinnati,  Ohio,  1981.
                                      215

-------
25. Camper, A.K., Broadaway,  S.C.,  LeChevallier,  M.W.,  and McFeters, G.A.
    Operational  variables and the release of colonized  granular activated
    carbon particles in drinking water.   Montana  State  University,  Department
    of Microbiology.  (Submitted to J. AWWA).

26. Topalian, P. In:  H. Sontheimer and  C.  Hubele,  The  Use of Ozone and
    Granular Activated Carbon in Drinking Water Treatment.  Engler-Bunte-
    Institute, University of  Karlsruhe,  West Germany,  1986.

27. LeChevallier, M.W., Hassenauer, T.S., Camper, A.K.,  and  McFeters, G.A.
    Disinfection of bacteria  attached to granular activated  carbon.  Appl.  and
    Environ. Microbiol.  48:5, November, 1984.

28. Snoeyink, V.L., Clark, R.R., McCreary,  J..J.,  and McHie,  W.F.  Organic
    compounds produced by the aqueous free-chlorine-activated carbon reaction.
    Environ. Sci. and Techno!. 15:188,  1981.

29. Voudrias, E.A.  Effects of activated carbon on  the  reactions  of free
    chlorine with phenols.  Environ. Sci. and Techno!.   19:441, 1985.

30. Voudrias, E.A., Snoeyink, V.U, and  Larson, R.A.  Desorption  of
    organics formed on activated carbon.  J.AWWA.   78;2, 1986.
                                     216

-------
            THE  CHARACTERISTICS OF  INITIAL  EFFLUENT  QUALITY  AND
            THEIR IMPLICATIONS FOR  THE FILTER-TO-WASTE  PROCEDURE

                       by:   Karen Bucklin
                            Dept. of Civil  and Agricultural  Engineering
                            Montana State University
                            Bozeman, MT  59717

                            Kelly 0. Cranson
                            Dept. of Civil  and Agricultural  Engineering
                            Montana State University
                            Bozeman, UT  59717

                            Appiah Amirtharajah
                            School  of Civil Engineering
                            Georgia Institute of Technology
                            Atlanta, GA  30332
                               INTRODUCTION

    It is well known that the initial effluent from a granular media filter
after backwash is of poorer quality with higher turbidity than the effluent
later in the filter run.  Various theories have been put forward to explain
this phenomenon, known as "filter ripening," and to determine the source of
these higher turbidity readings (1,2,3,4).  Several studies (5,6) have also
shown that during this period of higher turbidity, the numbers of
microorganisms including Giardia cysts, that pass through the filter
correspond with the turbidity readings, and are higher during this phase
than in the balance of the filter run.

    Two separate studies were carried out at Montana State University to
determine the characteristics of the post-backwash filter ripening.  The
first study, funded by the AWWA Research Foundation, was at plant-scale,
and was conducted at two municipal water treatment plants in Montana.  The
second was a pilot-scale study funded by the Montana State University
Engineering Experiment Station.
                                     217

-------
                          THE PLANT-SCALE STUDY

    The facilities included in this research were the Bozeman Water
Treatment Plant (BWTP)  in Bozeman,  Montana,  and the Missouri  River Water
Treatment Plant, located in Helena, Montana  (HWTP).  The two  facilities are
similar; both have eight dual media filters  and are 10 MGD capacity plants.
However, the plants vary in terms of raw water characteristics and treat-
ment schemes.  The BWTP uses a direct filtration treatment scheme, with the
raw sources being mountain streams  and reservoirs.  The HWTP  is a conven-
tional  treatment plant, with the Missouri River as its raw water source.

    The objectives of this study were:  1) to confirm the post-backwash
filter ripening stage characteristics of the filters in the two plants, 2)
to determine the turbidity and microbiological characteristics of this
period, and 3) to evaluate whether  any seasonal variations in these param-
eters could be detected.

    The turbidity data collected during the  post backwash period were both
from grab samples and from continuously recorded monitoring of the filters'
turbidity.  Samples were collected  from March 1986 through January 1987.
Each filter run had more than 40 data points.  Thirty-three filter runs
were completed at the BWTP, and data from 43 runs were collected at the
HWTP.  Microbiological  samples were taken at the same sampling locations
and were analyzed for total coliforms (including injured coliforms) and for
heterotrophic plate counts (HPCs).   In addition, the raw water charac-
teristics, the type of water treatment and chemical additions, and total
plant effluent characteristics were also monitored.

    The results of this study are summarized as follows.

    1.   The  incidence of higher than normal turbidity, during the post-
backwash filter ripening phase, was evident  at both plants and confirms
that shown by other pilot studies  (1,5).  The periods of high turbidity
readings were found to show a dual  peak characteristic.  This has been true
of all data collected, regardless of  season.  However, the winter turbidity
profile at the BWTP was significantly different from that of the charac-
teristic profile for this treatment plant's  filters during the balance of
the year (Figure 1).  This is due to:  1) the  lower municipal water demand,
reducing filter rates from up to 4.5  gpm/ft2 to an average of about 1.7
gpm/ftz, and  2) the much lower winter raw water turbidities, often as  low
as 0.30 to 0.40 Ntu.

    Tracer studies were done at the Bozeman plant  using fluoride as the
tracer to determine the origin of the turbidity peaks.  The fluoride tracer
was fed to the filter  influent continuously from  time zero (Figure 2).   It
can be seen that the second peak in turbidity appears to be due to influent
water, with some dispersion occurring as  the  influent mixes with the water
remaining above the media after backwash.
                                     218

-------
    A similar type of tracer study was done using the chlorinated backwash
water remnants as the tracer.  Filter effluent grab samples were analyzed
for residual chlorine over time (Figure 3).  As the figure indicates, the
first turbidity peak is associated with backwash water remnants.

    2.   In general HPCs were highly variable.  Occasionally, microbiolog-
ical sampling showed relationships between the higher turbidities and the
corresponding increase in heterotrophic plate counts (Figure 4).  The
chlorinated backwash water seems to inhibit bacterial numbers during the
initial filtration stage, and to suppress the correspondence reported in
pilot plant studies utilizing dechlorinated backwash water.  Some data, in
contrast to Figure 4, showed HPC values near zero in the turbidity peak
associated with the chlorinated backwash water remnants, and rose with the
turbidity peak associated with the influent water.

    3.   There were much higher recovery rates for chlorine injured coli-
forms on MT-7 agar (a media developed for detecting injured coliforms) as
compared with recovery rates on m-Endo LES agar, during the post-backwash
period  (Fig. 5).  Total coliform counts on MT-7 agar were two to 37.3 times
higher  than on m-Endo LES agar.  Thus, chlorine in the backwash water plays
a  significant role in the transmission of viable organisms during the ini-
tial stages of filtration.

                       THE PILOT-SCALE INVESTIGATION

    The objectives of this study were to define the mechanisms of the
filter  ripening events and to determine the effects of the addition of
coagulants  into the backwash water.

    Using a square section, dual media, in-line pilot plant filter unit
with 0.25 ft2 surface area, 200 filter runs were conducted during which
various coagulants, alum, CatFloc TL polymer, and a 20:1 alum/polymer
combination were  injected into the backwash water of four different filtra-
tion systems.  The systems were:  1) polymer as the primary coagulant with
bentonite as the  turbidity source, 2) alum as the primary coagulant, Min-u-
sil 30, a silica  clay, as the turbidity source, 3) alurn/polymer 20:1 as the
primary coagulant, Min-u-sil 30 as the turbidity source, and 4) influent
from the BWTP flocculators with CatFloc T polymer as the primary coagulant.
The term primary  coagulant is used to indicate the coagulant applied to the
influent water to the  filters, to distinguish it from the coagulants added
to the  backwash water.

    The results of the study follow.

     1.   The characteristics of the  initial effluent degradation were  found
to be  consistent  with  results obtained via  the plant-scale  study  (Figure  6).
                                     219

-------
          20    40    60    3 hr   5 hr
           Thm Attar Backwaah (mbi)
                                                         FUorkfa Concentration and
                                                             Turbidity v«Tbiw
                                                                     BWTP t2-!6-fl6
                                                                     Q - 3.3 opm/fla
                                                                     100* Fr ConcinlrillM
                                                                      of Trior - 1.27 ppm
         20    40    60    3 hr   5 hr
           Time After Backwaeh (rnln.)
Figure  1.  Winter profile vs.  previous
            seasonal  profiles:   BWTP.
Figure 2.   Fluoride tracer study.
  0.6J

 5 o.s
                         BWTP V17-87
                         Q-1.8 gpm/ft2
         20    40    60    3hr   5 hr
            TbM Attw Backwaah (mbi)
                                      o.s'
                                      0.0
       10  20   30  40   SO  60
               Time (minute*)
                                                                          2 hr 3 hr  4 hr
Figure 3.   Backwash water  remnants
             (Cl2 residual)
             and turbidity vs.  time.
Figure 4.  Heterotrophic plate count
            and turbidity vs. time.
                                      BWTP l-H-87
                                      0 » I.» gpm/ft2
                                      Influent Average Count
                             20     40    60^   3hr   5 hr
                              Thiw Aftmr Backwash (mbi)
                     Figure 5.   Total  coliform counts  on MT-7
                                  and m-ENDO agar with  turbidity
                                  vs. time.
                                           220

-------
    2.   Injection of a coagulant into the backwash water serves to lower
the zeta potential (the degree of chemical treatment of a particle) of the
influent particles (Figure 7).  As the influent particles disperse into the
backwash water remnants above the filter media under conditions in which no
coagulant is injected into the backwash water, the particles develop a
higher negative zeta potential.  The opposite is true for influent par-
ticles dispersing into backwash water remnants that contain coagulants.
The improving phase is due to the accumulation of particles within the
filter media, and confirms previous research (1,3).

    3.   The most effective coagulant for addition to the backwash water,
in terms of reduction of the magnitude and duration of the filter ripening
stage, is generally the primary coagulant.  Figure 8 shows strip chart
records from a series of filter runs using various concentrations of alum
in the backwash water of the alum primary coagulant system.  It is seen
that higher alum dosages (Ml.4 mg/1) suppress the filter ripening peak,
but cause a higher initial peak due to solid aluminum hydroxide in the
underdrain system.

    4.   The optimum time for injection of coagulants into the backwash
water is identical to the time required to disperse the backwash water into
the volume of the entire filter unit.  This includes the volume above the
filter media, up to the backwash water gutter (Figure 9).

    5.   The magnitude and duration of the filter ripening peak is reduced,
in cases where coagulant is injected into the backwash water, by increasing
the remnant volume above the media (Figure 10).

                                CONCLUSIONS

    A filter-to-waste procedure cannot be generalized as being useful for
all plants.  In some cases, such a procedure is impractical due to the long
filter ripening period.  The procedure may be used during the post-backwash
period for systems with high peaks, a short duration of ripening, and
significant correlation between higher post-backwash turbidities and high
microorganism counts.  In addition, this  study indicates that low coliform
counts during this period may be misleading, due to the high numbers of
injured coliforms present in the backwash water remnants.  Also, high num-
bers  of organisms (HPC) are often present in the turbidity peak associated
with  the filter influent.

    A practical approach to determining an appropriate filter-to-waste
period is to try  to minimize both the magnitude and the duration of the
filter ripening stage of the filter run.  This can be accomplished by
several means, including lower filtration rates, slow incremental filter
startup after backwash, improved backwashing procedures, proper coagulant
dosing of the raw water, injection of coagulants into the backwash water
using optimum injection time and dosage of the coagulant, and varying  the
volume of the backwash water  remnants standing above the filter media  after
backwash.
                                     221

-------
                   Influent Miilnj With
                   Above nedlaRemnanl
                    TbiM
Figure 6.   Proposed  characteristics  of
             initial effluent degradation,
                 Thiw(Mn.) 10
                                     IS
Figure 8.   Actual pilot  plant  effluent
             strip chart turbidity.
                              (137) control
                              138) control
                                                                       //   (I39)£l mg/l alum in B.W.

                                                                            (140) 23 mg/l llum infl.W.
                                                                       8   12   16  20
                                                                          Time(Min.)
                                                      Figure 7.   Zeta potential  of  influent
                                                                   particles compared to  effluent
                                                                   turbidity.
                                                                              Experiment A-4 Polymer Primary
                                                                        ••°-».x  Coagulant Variation of Time of Injection of
                                                                            X"D Polymer 0.4 mg/l into Backwash Water


                                                                              Experiment B-4 Alum Primary Coagulant
                                                                              Variation of Time of Injection of Alum 18 mg/l
                                                                              Into Backwash


                                                                           Experiment D-3 Backwash Dye Trace at 20 l/min
                                                              I   2   J   4    S   6   7
                                                             Tbiw From End of Back waih(Mn.)
                                                                                              T
Figure 9.   Summary of  optimization of
             backwash coagulant injection
             time.
•COT
njurr
Ml
(IIIAC2
(•II ACS
(«2I 2ACI
(•11 2AC7
I«4I ]«•
(•7IOACII
(WIOMI2
(1171 AA1I
(!!•! 2AA3S
(1201 IAAJ4
(I2IIOAAI5

TUHBnrri
«nt
i«
17
14
I*
It
II
II
17
14
17
20
III
BACXWAM
1 COACUANT
OCMEM!
00
00
00
on
00
00
00
00

If
II
II

ABOVE fcOU
217
2 17
4 17
4 17
«I7
617
0 17
017
211
4 17
• 17
017
                          Figure 10.  Variation of  remnant volume
                                       above filter  media with alum
                                       primary  coagulant.
                                                  222

-------
                                REFERENCES

1.  Amirtharajah, A., and Wetstein, D.P-, Initial degradation of effluent
    quality during filtration.  Jour. AWWA.  72:9:518, 1980.

2.  Amirtharajah, A.  The interface between filtration and backwashing.
    Water Research. 19:5:581, 1985.

3.  O'Melia, C.R., and Ali, W.  The role of retained particles in deep bed
    filtration.  Progress in Water Technology (Great Britain).  10:516:167,
    1978.

4.  Francois, R. 0., and Van Haute, A. A.  Backwashing and conditioning of
    a deep bed filter.  Water Research.  19:11:1357, 1985.

5.  Logsdon, G. S. et al.  Evaluating sedimentation and various filter
    media for removal of Giardia cysts.  Jour. AWWA.  77:2:61, Feb. 1985.

6.  Logsdon, G. S. et al.  Alternative filtration methods for removal  of
    Giardia cysts and cyst models.  Jour. AWWA.  73:2:111, Feb. 1981.
                             ACKNOWLEDGEMENTS

    The  studies  reported  above were  funded by the American Water Works
Association  Research  Foundation  and  the Montana State University
Engineering  Experiment  Station.   The assistance of the staff of the cities
of  Bozeman and Helena in  collecting  data during the studies is gratefully
acknowledged.
                                    223

-------
          FACTORS AFFECTING THE INACTIVATION OF 6IARDIA CYSTS BY
          MONOCHLORAMINE AND COMPARISON WITH OTHER DISINFECTANTS
                       by:  Alan J. Rubin
                            Water Resources Center
                            The Ohio State University
                            Columbus, OH 43210
                               INTRODUCTION

    Chloramination has been used in the past as an economical  alternative
to conventional chlorination.  The addition of ammonia to chlorine results
in a less active but more persistent disinfectant.  This approach has been
used to eliminate taste and odor problems and to reduce the need for the
additional application of disinfectant in long distribution systems.
Because of their lower activity, chloramines are also receiving attention
today as a cost-effective means of meeting the THM standards.   There are
questions, however, about their relative effectiveness against cysts of the
human pathogen Giardia Iambiia.  This protozoan is the causative agent of
giardiasis, now one of the most common water-borne diseases in the United
States.

    Chloramines are formed by the reaction between chlorine and ammonia or
organic amines.  Monochloramine is most stable in the pH range of 7 to 9
and a chlorine-to-ammonia molar ratio <1:1.  Dichloramine forms primarily
in the pH range of 5 to 7 and at chlorine-to-ammonia molar ratios approxi-
mating 2:1.  Nitrogen trichloride, which has no disinfecting ability, forms
at pH <4 at even higher chlorine-to-ammonia ratios.  The formation of each
species is dependent upon factors such as pH, temperature, contact time,
and the initial concentrations and ratios of chlorine and ammonia.  The
overall purpose of the research described in this paper was to obtain basic
data on the inactivation of protozoan cysts using preformed monochloramine.
Cysts of Giardia muris, a parasite of mice, was used as a model for Giardia
lamblia.  Studied were the effects of pH, temperature, and chlorine/ammonia
ratio under carefully controlled conditions in batch reactors.
Inactivation data were obtained as a function of time and C.t1 products
were calculated from these data.
                                    224

-------
                           EXPERIMENTAL METHODS

    Excystation following exposure to disinfectant was used as the cri-
terion for cyst survival.  The number of intact cysts was compared to the
total number of intact cysts, shells, and partially excysted cysts for quan-
titation purposes.  The excystation procedure, which was designed to simu-
late the conditions in the gastrointestinal tract of a mouse, and the
other experimental procedures used in this work have been described in more
detail elsewhere  (1,2).  Concentrations of free chlorine, monochloramine,
and dichloramine were determined using the standard DPD colorimetric
procedure.

    Monochloramine stock solutions were prepared at three different
chlorine-to-ammonia weight to weight ratios.  For calculation purposes, the
molecular weight  of chlorine and hypochlorite ion was taken as 71 g/mole,
and for ammonia as 17 g/mole.  Solutions at a 1:4 ratio were prepared by
dissolving 15.53  grams of ammonium sulfate (formula weight 132.15) in
chlorine demand-free water; adjusting the pH to a value between 8 and 9
with 2.0 M sodium hydroxide solution; and combining with sodium hypoch-
lorite solution equivalent to 1.0 g as chlorine.  Monochloramine formation
was found to be virtually instantaneous at this ratio.  Solutions with a
1:2 or 1:1 chlorine-to-ammonia ratio were prepared in a similar fashion but
were mixed on a magnetic stirrer for 45 to 60 minutes in order to ensure
complete reaction between the chlorine and ammonia.

                          RESULTS AND DISCUSSION

    Typical survival curves for the  inactivation of G. muris cysts by pre-
formed monochloramine are shown below in Figure 1.
                            50
                                  100     150
                                  TIME (min)
                                              2B0
                                                    250
Figure  1.   Inactivation  of  cysts  of  G. muris with monochloramine at
            pH  7,  5  °C, and  a  1:4  chlorine-to-ammonia  ratio.
                                      225

-------
     Similarly shaped curves  were  found  for  cyst  inactivation with  free
 chlorine (2).  A concave upward curve,  as was  found  for  the inactivation  of
 cysts of both G. muris and G.  Tamblla cysts by ozone (1),  is the most com-
 monly observed shape.   The difference in the shapes  implies that the mecha-
 nism of inactivation is the  same  for monochloramine  and  free chlorine,  but
 different than with ozone.

     The monochloramine concentrations necessary  to produce 99-percent kill
 for the different combinations of pH, temperature, and chlorine-to-ammonia
 ratios were determined from  data  such as shown in Figure 1.  Most
 experiments were run at pH 6, 7,  8, and 9 and  15°C with  a  1:4  chlorine-to-
 ammonia weight ratio.   Several additional studies were run at  1, 5, and
 30°C, and at other ratios.  Typical results, plotted log 99 percent kill
 time as a function of log applied chloramine concentration, are shown in
 Figure 2.
                1000
              c
              'E
                 100 -
                  10
                                  4         10

                               CONCENTRATION , mg/L
40
 Figure 2.  Inactivation of G. muris cysts by preformed monochloramine
           at 15°C.


    One experiment, designed to study toxicity due to excess ammonia, was
 run at pH 9 with 120 mg/1 of ammonia.  There was no effect on the ability
 of the cysts to excyst under these conditions, which was the most extreme
 pH and ammonia concentration examined.

    As found in all disinfection studies with other organisms and other
 cnsinrectants, C.t  products for monochloramine decreased with increasing
 temperature.  Most researchers report that monochloramine is more effective
 as the pH is reduced; the opposite was found in this investigation.  The
 reason, as was determined in a study with free chlorine (2), is that G.
muns is more sensitive to inactivation in alkaline solutions.  Similar
 results were found with ozone (1) and chlorine dioxide (3).   it should be
 emphasized that, in the ranges of practical interest, temperature is
more important than pH.
                                    226

-------
    The results at different chlorine/ammonia ratios were somewhat ambig-
uous.  Most studies in this work were run at ratios of 1:4 to ensure
complete and rapid reaction of chlorine with ammonia.  A series of studies
at pH 7 (5°C) and pH 9 (15°C) were also run at ratios of 1:1 and 1:2.  The
rates of reaction between chlorine and ammonia were less, but there was no
free chlorine in these solutions.  The smallest C.t1 products were obtained
at the 1:2 chlorine/ammonia ratio, with intermediate results obtained at
the ratio of 1:1.

    Figure 3 shows that G. muris cysts are most readily inactivated by
ozone, followed by chlorine dioxide, elemental iodine (4), free chlorine,
and then preformed monochloramine, respectively.  C.t' products for ozone
were on the order of 1.94 mg-min/1 whereas those for chlorine dioxide
ranged from 7.05 to 16.32 mg-min/1.
                    iea.
                   CO
                   E
                                             100 200
                            99% Ki I I  Time (min)
Figure 3.  Inactivation of G. muris cysts by chemical disinfectants,
           pH 7 and 5°C.

The difference in the inactivation rate of free chlorine compared to that
of monochloramine changes principally with concentration.  C.t1 products
for free chlorine ranged from 604 to 1,195 mg-min/1 whereas for monochlor-
amine the C.t' products ranged from 1,370 to 1,890 mg-min/1.  Chang (5)
showed that chloramines were slower than chlorine in penetrating the cyst
wall of Entamoeba histolytica at short contact times, whereas with contact
times of two or more hours, chloramines were shown to be just as cysticidal
as chlorine.  Perhaps, here also, the relative efficacy of chlorine and
chloramines is dependent upon the concentration range under comparison.

    Table 1 presents typical C.t1 products generated over the pH range of 6
to 9 and temperature range 1 to 15°C for preformed monochloramine as a
result of this study.  C.t1 products with "greater than" symbols are inter-
polated values.
                                    227

-------
     TABLE 1.   SUMMARY OF INACTIVATION DATA WITH GIARDIA MURIS CYSTS
                     C.t1  PRODUCTS FOR MONOCHLORAMINE
                                   Monochloramine Concentration
    pH        temp                            (mg/1)
                                     <2       2-10         >10
6 15
5
1
7 15
5
1
8 15
5
1
9 15
5
1
1,500
>1,500
>1,500
> 970
> 970
2,500
1,000
>1,000
>1,000
890
>890
>890
880
>880
>880
970
1,400
>1,400
530
1,430
1,880
560
>560
>560
1,300
>1,300
>1,300
960
1,900
3,200
700
>700
>700
400
>400
>400
                                CONCLUSIONS

    It has been shown with ozone that G. muris is slightly more
resistant than G. lamblia, cyst inactivation being parallel over a
broad concentration range (1).  However, it has been observed more recently
that G. muris has an unusual  pattern of inactivation with pH (2).  It has
not been established whether such a pattern is also true for G. lamblia,
although this possibility is currently under investigation.  Therefore, the
promulgation of concentration-time data for monochloramines based on
results with G. muris is premature until it can be verified that G. muris
is a valid model for G. lamblia.

    Preformed monochloramine was shown to be relatively ineffective against
Giardia cysts, requiring extremely long contact times and doses.  However,
in view of their current widespread use and favorable economics, there is
no reason not to allow conventional chloramination practice in well
operated chemical  treatment and filtration plants that are without a
history of giardiasis.  In fact, a case can be made for the promotion of
chloramines especially when pre-oxidation is a part of the treatment train.
                                    228

-------
                             ACKNOWLEDGEMENTS

    The research described In this paper would not have been possible
without the hard work and dedication of Deborah Teitz, David Evers, Joseph
Leahy, G.B. Wickramanayake, Ricky Chen, John Engel, and Otis J. Sproul.

    This research is being supported in part by a cooperative agreement
between The Ohio State University and the Environmental Protection Agency.
The project officer is John C. Hoff, whose valuable assistance is also
gratefully acknowledged.

                                REFERENCES

1.  Wickramanayaka, G.B., Rubin, A.J., and Sproul, O.J.  J. Am. Water
    Works Assoc.  77:74, 1985.

2.  Leahy, J.G., Rubin, A.J., and Sproul, O.J.  Appl. Environ. Microbiol.
    53, 1987.

3.  Rubin, A.J., Leahy, J.G., and Sproul, O.J.  Inactivation of Giardia
    muris cysts with chlorine dioxide,  (paper in preparation).

4.  Rubin, A.J., and Chen, R.Y.S.   Inactivation of Giardia muris cysts
    with elemental  iodine in water,  (paper in preparation).

5.  Chang, S.L.  J_^ Am. Water Works Assoc.  36:1192, 1944.
                                   229

-------
                   INACTIVATION OF HEPATITIS A VIRUS AND
                  MODEL VIRUSES IN WATER BY FREE CHLORINE

                       by:  Mark D. Sobsey
                            Taku Fuji
                            Patricia Shields
                            Department of Environmental  Sciences
                              and Engineering
                            School of Public Health
                            University of North Carolina
                            Chapel Hill, NC  27514


                               INTRODUCTION

    Hepatitis A virus (HAV) is probably the most important waterborne
enteric virus.  One reason for its importance is the severity of the
disease it causes.  Hepatitis A is an acute liver disease lasting several
weeks to months, which may include symptoms of malaise,  anorexia, vomiting,
diarrhea, fever, and jaundice.  Another reason for the importance of HAV is
the high level at which it is fecally excreted by infected individuals;
fecal shedding may be in the range of 107 to 109 infectious units per gram
of feces (1).  Additional evidence for the importance of waterborne HAV  is
its great persistence in environmental waters compared to other enteric
viruses and indicator bacteria.  Recent data show that HAV survives longer
in ground water and in primary and secondary sewage effluent than other
enteric viruses such as poliovirus 1 and echovirus 1 (2).

    Outbreaks of hepatitis A due to consumption of contaminated drinking
water in the United States further emphasize the importance of this enteric
virus (3).  Most waterborne outbreaks of hepatitis A have been attributed
to fecal contamination of untreated or inadequately treated water  or con-
tamination of treated drinking water during distribution (3).  The risks of
hepatitis A infection and illness from drinking untreated, undisinfected
water are demonstrated by a recent waterborne outbreak of hepatitis A in a
small, rural  community in western Maryland (4).  Fourteen cases of hepati-
tis A occurred in this community of 300 people who were  drinking untreated
ground water from household wells.  Ground water was fecally contaminated
probably by septic tank effluent, and contained high levels of total  and
fecal  conform bacteria.   HAV was detected and quantified in four of six
concentrate samples of incriminated ground water by inoculation of African
green monkey kidney cell  cultures and by experimental  infection of chimpan-
nfeM;tl;LrTrtnem0nS,trateS.the Msks of heP^tis A from consumption
of untreated,  fecal  y contaminated water and underscores the need to disin-
fect all  water supplies with  a disinfectant that is effective against HAV
                                   230

-------
    A recent waterborne outbreak of viral gastroenteritis and hepatitis A
in Georgetown, Texas, highlights the potential for transmission of HAV in
communities relying on limited water treatment practices primarily
involving disinfection (5).  Hepatitis A antigen was detected by radioim-
munoassay in the contaminated water, and viral disease transmission
occurred even though chlorinated ground water samples taken from the
distribution system were negative for coliform bacteria.  The occurrence of
36 reported cases of hepatitis during the outbreak demonstrated the failure
of conventional indicator bacteria to adequately predict viral con-
tamination of the water supply, as well as the ability of HAV to survive
the chlorination process.

    Conventional water treatment practices utilizing chemical disinfection,
primarily chlorination, are generally believed to be effective in producing
microbiologically safe drinking water.  However, the growing number of
reports on the isolation of viruses from treated drinking water (6) suggest
that viruses may survive treatment under certain conditions.  The
establishment of reliable water treatment practices and water quality stan-
dards to  insure the virological safety of water supplies can be achieved
only by fully understanding the response of HAV to water disinfectants.

    Despite the need  to determine  the  kinetics and extent of HAV inac-
tivation  by water disinfectants, the few investigations  reported to date on
HAV inactivation by chlorine  have  been inadequate due to technical limita-
tions.  Early  studies  by Neefe  et  al.  (7,8) provided indirect evidence that
HAV is  insensitive to  combined  chlorine.  Using human volunteers for virus
infectivity assay, they found that a total chlorine residual of 1 mg/1 did
not completely  inactivate  HAV in dilute  fecal suspensions after a contact
time of 30 minutes.   The addition  of sufficient chlorine to produce total
and free  chlorine concentrations of  1.1  and 0.4 mg/1, respectively, in
purified  effluent was  required  to  prevent clinical manifestations of infec-
tious hepatitis  in the volunteers.  More recently, Peterson et al. (9)
used marmosets  to assay for HAV infectivity after chlorination of a par-
tially  purified  preparation of  HAV.  The infectivity of  the preparation,
which contained  about  1,500 infectious units/ml, was only partially reduced
by treatment with up  to  1.5 mg/1 of free residual chlorine at neutral pH
for 30 minutes.  These experimental  results,  along with  observations made
during  the outbreak of hepatitis in Georgetown, Texas (5), suggest that HAV
is more resistant to  conventional  water  chlorination processes than other
enteroviruses  and indicator bacteria.

    In  contrast  to the results  of  the  HAV disinfection  studies described
above,  studies  by Grabow et al. (10) indicated that HAV  may be more sen-
sitive  to free  chlorine than  previous  studies and epidemiological evidence
have suggested.  Using serological  techniques for assay  of HAV infectivity
in cell culture, Grabow and co-workers found  that HAV was very sensitive to
low levels of  free chlorine relative to  selected  indicator viruses and bac-
teria.  However, other studies  by  this group  indicated  that HAV was rela-
tively  resistant to combined  forms of  chlorine  (11).
                                     231

-------
    In view of the limited data on HAV disinfection  in  general  and the
inconsistent findings of the few studies on its  disinfection by chlorine,  a
critical  evaluation of HAV inactivation by free  and  combined forms of
chlorine  and by other disinfectants such as chlorine dioxide, U.V. light,
and ozone is clearly warranted.

    The study of HAV inactivation kinetics by chlorine  and other disinfec-
tants under carefully controlled experimental conditions in the laboratory
is now feasible using new methods for the cultivation and enumeration of
HAV in cell cultures (12,13,14,15).  The purpose of  this study is to exam-
ine the kinetics and extent of HAV inactivation  by free chlorine, combined
chlorine in the form of chloramine, U.V. light,  and  chlorine dioxide.
Inactivation of HAV is compared to the inactivation  of  model viruses
including coxsackievirus B5 and bacteriophages MS2 and  0X174.  These
studies are still in progress, and only data on  virus inactivation by free
chlorine are available for this report.

                           METHODS AND MATERIALS

VIRUSES, CELL CULTURES, AND VIRUS PURIFICATION

HAV

    The HM175  (NIH prototype)  strain of HAV, originally isolated from feces
of an  infected human in Australia  (12,14,16), is produced in persistently
infected BS-C-1  cells grown in 850 cm^ roller bottles or 6,000 cm2,
10-tiered  cell factories  (NUNC) incubated at 37°C.  Prior to persistent
infection,  the virus had  been  serially passaged six times in marmosets, 10
times  in primary African  green monkey kidney (AGMK)  cells, and seven times
in BS-C-1  cells.

    HAV  infectivity  is assayed by  the radioimmunofocus assay (RIFA)  in
BS-C-1 cells  as  previously described  (14,17), except the incubation  period
was reduced to one week.  The  RIFA is an enumerative assay analogous to a
plaque assay,  except non-cytopathic, focal areas of infected cells are
visualized  by an  immune autoradiographic method.

    For  preparation  of purified, monodispersed HAV, persistently  infected
cells  are  passaged every  two to four weeks by trypsinization and  then
resuspension  of  some of the cells  in growth medium at a concentration of
about  1  x  105 cells/ml for re-inoculation into culture vessels.  At each
passage, some of  the persistently  infected cells and all of  the culture
fluids are  harvested as crude  virus stock.   Harvested,  infected cells are
centrifuged at low speed  (about 3,000 x g),  resuspended in small  volumes of
phosphate-buffered saline (PBS), pH 7.5, and extracted with  an equal volume
of chloroform.   The  HAV-containing PBS is  recovered by  low speed  centrifu-
gation to  remove  cell debris and chloroform.  The cell  debris and chloro-
form  are extracted four to  six more times with equal volumes of PBS  to
obtain additional virus,  and all PBS extracts are pooled as  crude virus
stock.
                                    232

-------
    The residual cell debris and chloroform are further extracted twice in
succession with volumes of 0.1 percent sodium dodecyl  sulfate (SDS) in PBS
equal to the volume of chloroform and cell debris.  SDS-PBS extracts are
recovered by low speed centrifugation at room temperature, and SDS is
removed by precipitation at 4°C followed by recentrifugation at 4°C.

    HAV in culture fluids is concentrated by precipitation with polyethyl-
ene glycol 6,000 (12 percent w/v, pH 7.2) overnight at 4°C.  Resulting pre-
cipitates are recovered by low speed centrifugation, resuspended in a small
volume of PBS, and extracted with a volume of chloroform equal to the PBS
volume in order to remove excess PEG.  The PBS extracts are cleared of
chloroform and PEG by low speed centrifugation.

    PBS extracts of cells (chloroform and SDS) and PEG concentrates from
culture fluids are pooled, and HAV is pelleted by ultracentrifugation at
30,000 RPM (105,000 x g) for four hours at 5°C.  Resulting pellets are
resuspended in small volumes of 0.05M phosphate-buffered distilled water
(PBDW) and supplemented with CsCl to give a density of 1.33 g/ml.  These
samples are ultracentrifuged to equilibrium in self-generated gradients at
25,000 RPM (90,000 x g) and 5°C for three days using the SW27 rotor
(Beckman  Instruments).  Gradients are harvested in fractions from the bot-
toms of the tubes and assayed for HAV infectivity by RIFA.  Peak fractions
of HAV infectivity are desalted by ultrafiltration and washed with PBDW
using Centricon 30 ultrafiltration tubes  (Amicon  Inc.).  Desalted fractions
are  layered onto 10  to 30 percent sucrose gradients in PBHDF water, pH 7.5,
and  ultracentrifuged in the SW27 rotor at 25,000  RPM (90,000 x g) and 5°C
for  five  hours.  Under these conditions,  single virions sediment about two-
thirds of the way down the gradient.  Gradient fractions are harvested from
the  top of the  tube  and assayed for HAV  infectivity by RIFA.  Gradient
fractions corresponding to single virions are  then pooled  and mixed with
appropriate amounts  of gradient fractions containing single virions of the
other  test viruses.  The  titer of each virus  in the mixture is about  1 to  5
x  108  infectious units/ml.  Virus mixtures are stored at 4 to 5°C  for sub-
sequent use in  disinfection experiments.

Coxsackievlrus  B5

     Coxsackievirus B5  (Faulkner Strain)  is grown  and assayed  by  the plaque
technique in  the BGM (African green monkey kidney-derived) continuous cell
line as previously described  (18).  Coxsackievirus  B5 was  first  plaque
purified  two  to three  times and  then  grown  in large quantities at  low
multiplicity  (0.01 to  0.1  PFU/cell).  Crude  virus stock  is harvested  from
infected  cell  lysates  several days  post-infection when  cytopathic  effects
are  4+.   Virus  is  liberated from cells and cell debris  by freezing and
thawing,  and  then  cell debris is removed by  centrifugation at low  speed
(10,000 x g)  for 15  to 30 minutes.  Viruses  in resulting  supernatants are
pelleted  by ultracentrifugation  (105,000 x g  and  5°C  for  four hours).
Resulting virus pellets are resuspended  in PBHDF  water,  homogenized one
minute,  and  in  some  cases, centrifuged at 10,000  x  g  and  5°C  for 20 minutes
to remove additional debris.  After  supplementing the  sample  with  CsCl  to
give a density of  1.33 g/ml, viruses  are banded  to  equilibrium as  for HAV.
                                     233

-------
Gradient fractions are harvested and assayed for virus infectivity, and
virus peak fractions are desalted using Centricon 30 ultrafilters.  These
fractions are pooled and subjected to rate-zonal centrifugation in five
percent (or 10 percent) to 30 percent sucrose gradients as for HAV.
Gradient fractions are harvested and assayed for virus infectivity, and
fractions corresponding to single virions are added to HAV samples to give
the desired virus titer.

Bacteriophages

    Bacteriophages MS2 (ATCC 15597-B1) and 0X174 (ATCC 13706-B1) are grown
and assayed by the top agar plaque technique (19) in E. coli C3000 (ATCC
15597) and E. coll C (ATCC 13706) hosts, respectively, using nutrient agar
#2 (nutrient agar with 0.5 percent NaCl) media.  Crude virus is harvested
from the top agar of plaque assay plates having confluent lysis by scraping
into small volumes (3 to 5 ml/plate) of PBS.  Harvests are extracted with
chloroform and centrifuged at 5,000 x g for 10 minutes to remove chloro-
form, cell debris, and agar.  The resulting supernatant is centrifuged at
10,000 x g for 10 minutes to remove additional cell debris, and viruses in
this supernatant are pelleted by ultracentrifugation for four hours at
105,000 x g and 5°C.  Pellets are resuspended in PBHDF water, supplemented
with CsCl to give a density of 1.44 to 1.45 g/ml, and the viruses are
banded to equilibrium in self-generating CsCl gradients for three days at
25,000 RPM and 5°C using the SW27 rotor.  Gradient fractions are assayed
for virus infectivity and virus peak fractions are desalted using Centricon
30 ultrafilters.  To remove virus aggregates, desalted fractions are
filtered successively through 0.2 and 0.8 urn pore size polycarbonate
filters  (Nuclepore) which had been pretreated with 0.1 percent Tween 80 and
then  rinsed with HDFW.  The filtrates are collected as stocks of almost
exclusively single virions.  Appropriate amounts of the single virion
stocks are combined with single virions of HAV and coxsackievirus B5.

GLASSWARE AND HALOGEN REAGENTS

    All  glassware for disinfection experiments and preparation of halogen
demand-free  (HDF) virus stocks  is soaked at  least  four hours  in a strong
chlorine  (10  to 50 mg/1) solution and  then  rinsed  thoroughly with HDF water
prior  to  use.  HDF water and buffer  solutions  for  disinfection experiments
are prepared  from twice deionized, activated  carbon-filtered water which  is
then  passed through a macroreticular  scavenging  resin bed  (Rohm and  Haas).
HDF,  phosphate-based  buffers, 0.01M, were used  to  prepare  chlorine test
solutions and buffered water for  disinfection  experiments.

    Household bleach  (5.25  percent  sodium hypochlorite; Clorox)  is used  to
prepare  solutions of  free chlorine at  pH 6,  7,  8,  9,  or  10.   Free  chlorine
solution  of about 100 mg/1  is prepared  by diluting bleach  in  HDF water.
Stock  solution  is then  diluted  in test  water  (PBHDF water,  pH 6  to  10)  to
give  the  target chlorine concentration  of 0.5 mg/1.   Chlorine concentration
is  verified by  chemical analysis.
                                    234

-------
HALOGEN ANALYSIS

    Chlorine concentrations are measured by DPD colorimetric methods as
described in Standard Methods for the Examination of Water and Wastewater.
16th edition (20).  Standardization of procedures for chlorine measurement
was by the DPD ferrous titration method.  The reliability of chlorine
measurements is checked regularly by analyses of chlorine standards pre-
pared by the U.S. EPA.

PROTOCOLS FOR DISINFECTION EXPERIMENTS

    For disinfection experiments, samples are placed in 25 mm diameter x
150 mm long test tubes and tubes are kept in a water bath to maintain a
temperature of 5°C.  For experiments with free chlorine at a concentration
of 0.5 mg/1, 0.24 ml of purified, monodispersed virus stock mixture (HAV,
coxsackievirus B5, MS2, and 0X174)  is added to 11.76 ml of a chlorine solu-
tion containing 0.51 mg/1 free  chlorine and then briefly mixed.  A second
test tube containing only chlorine  solution serves as a halogen control.  A
third tube containing a 1:50  dilution of stock virus in PBHDF water serves
as a virus control.  Samples  of 0.7 ml are withdrawn from the reaction tube
(chlorine solution plus added virus) for viral analysis at 0.33, 1.0, 3.0,
10, 30, and 60 minutes after  virus  addition.  These samples are immediately
diluted two-fold  in  virus diluent  (2X Eagle's MEM) containing one percent
Na2S203.  Diluted  samples are stored at 4°C for subsequent virus assays.
For virus assay,  samples are  further diluted five-fold  (10-fold overall),
followed by serial 10-fold dilutions in separate diluents for HAV, cox-
sackievirus B5 and the two phages.  After  the 60 minute  reaction period,
the remaining  reaction mixture  (halogen plus added virus) and the chlorine
control sample  (halogen only) are  re-analyzed for free  and combined
chlorine.  Samples from the  virus  control  tube  (virus  plus PBHDF water) are
diluted serially  10-fold at  the beginning  and the end  of  the 60 minute
reaction period  for  subsequent  virus assay.

ANALYSIS OF VIRUS  DISINFECTION  DATA

    Virus disinfection data  are obtained  In the  form of virus concentra-
tions; as plaque  forming units  (PFU) for  coxsackievirus B5, MS2, and 0X174,
or radioimmunofocus  forming  units  (RFU) for HAV, per ml  of test sample.
These data are average values from triplicate cell culture plates for  each
countable dilution of  each virus.   For  each experiment,  the virus con-
centrations of the virus control  sample  (buffered water + viruses) are  com-
puted  (time =  0).  These values are taken  as N0, the initial virus
concentration, in  PFU  or RFU  per ml.   For  each  test  sample  (samples  taken
from  the test mixture  at 0.33,  1,  3,  10,  30, and 60 minutes), the average
concentration  of  each  virus,  as PFU or  RFU per  ml,  is  computed.  The pro-
portion of  initial viruses remaining at each test time (t)  is computed  by
dividing the  virus concentration at each  test time  (Nt) by  the  initial
virus  concentration  (N0),  i.e., computing  Nt/N0 for  all  sample  times for
each  virus.   TSvsa values are then  transformed  to logio values  (logio
 [Nt/N0])-   For e-jch  virus, the  logio Nt/N0 values of duplicate  experiments
for the  same  test condition  of  chlorine concentration, temperature,  and pH
                                    235

-------
are averaged.  These mean data for logic Nt/N0 are then  paired with the
data for sampling time (t) and analyzed by linear regression using a Texas
Instruments TI-55-II calculator.  The correlation coefficient, slope of the
regression line, and time for 99.99 percent inactivation of the initial
viruses are computed.  The data are also stored in and analyzed similarly
using Lotus 1-2-3 on an IBM PC.

                          RESULTS AND DISCUSSION

    The mean results of duplicate virus disinfection experiments using 0.5
mg/1 free chlorine in PBHDF water, pH 6 to 9, at 5°C are summarized in
Table 1 as times for 99.99 percent inactivation of the initial viruses
(T-99.99).  The mean results of duplicate experiments at pH 6, 7, 8, and 9
are summarized in Figures 1, 2, 3, and 4, respectively,  where log^ Nt/N0 is
plotted versus contact time in minutes.  These results indicate that HAV is
inactivated  rapidly by 0.5 mg/1 free chlorine at pH 6 to 9, with T-99.99
values of <8 minutes (Table 1).  There was little difference in HAV inac-
tivation rates at the different pH levels, however, inactivation was
somewhat slower at pH 9 than at lower pH levels.  It should be noted that
HAV inactivation at pH 6  is not faster than at pH 7 to 9, despite the fact
that some of the free chlorine  is present as the presumably less biocidal
OC1- at the  higher pH levels.  This may mean that HAV is relatively sen-
sitive to inactivation by OCT as well as HOC1.

             TABLE  1.  INACTIVATION OF HAV, COXSACKIE B5, MS2,
                  AND 0X174 by 0.5 MG/L FREE CHLORINE AT
                    pH 6.0, 7.0, 8.0 AND 9.0 AND 5°C IN
                       BUFFERED, DEMAND-FREE WATER*
                pH              Win. for 99.99% Inactivation
                              HAV       C-B5      MS2    0X174
6.0
7.0
8.0
9.0
5.0
3.6
3.8
7.7
14
24
38
108
1.2
4.4
18
16
0.4
0.4
1.2
4.6
                * Purified,  monodispersed  viruses.

     In contrast to HAV,  coxsackievirus  B5 was  relatively  resistant  to  inac-
 tivation by free chlorine,  with  T-99.99 values  ranging  from  a  low of  14
 minutes at pH 6 to a high of 108 minutes  at  pH  9  (Table 1; Figures  1  to  4).
 Inactivation kinetics of coxsackievirus B5 by  free  chlorine  showed  the
 typical pattern of decreased inactivation rates at  progressively higher  pH
 levels.  Presumably, this is due to greater  resistance  to inactivation by
 OC1" than by HOC1.
                                     236

-------
                          468

                            TIME (min)
10
12
Figure 1.  Inactivation of HAV,  CBS, MS2, and 0X174 by free chlorine,
           0.5 mg/1,  pH 6.0,  5°C.
           -1-
          N-2
       L  t
       0  /
       G  N
          „ -3
            -4--
            -5
                           468
                             TIME (min)
                                               10
       12
 Figure 2.   Inactivation of HAV, CBS, MS2, and 0X174 by free chlorine,
            0.5 mg/1, pH 7.0, 5°C.
                                237

-------
                        10         20

                            TIME (min)
                                               30
Figure 3.  Inactivation of HAV,  CB5, MS2, and 0X174 by free chlorine,
           0.5 mg/1,  pH 8.0,  5°C.
U 1
-1 -

N-2-
L I
0 /
G N
— ^-
o *
-4-

[' '
V ~*\
l\ \ x\
• i\ '• x""'--
i \ '• *'*1"
^A "'•-..

} '"'"••-
"l
	 1 	 1 	 1 	 1 	 1 	 1 	


-«- HAV
-*- CBS
•• MS2
-e- 0X174

i DETECTION
' LIMIT


            0     10    20    30     40    50    60    70

                           TIME  (min)
Figure 4.   Inactivation of HAV,  CBS, MS2, and 0X174 by free chlorine
           0.5 mg/1,  pH 9.0,  5°C.
                                 238

-------
    The inactivation rates for coxsackievirus B5 reported here are reason-
ably consistent with rates reported by Engelbrecht et al. (21).  With about
0.5 mg/1 free chlorine at 5°C, Engelbrecht et al. (21) reported T-99 values
of 3.4, 4.6, and 66 minutes at pH 6, 7.8, and 10, respectively.  The data
of our study give T-99 values of 7, 19, and 54 minutes at pH 6, 8, and 9,
respectively.  Thus, coxsackievirus B5 inactivation rates at the same or
similar pH values differ by no more than a factor of four in these two
studies.

    Bacteriophage 0X174 was inactivated more rapidly than either HAV or
coxsackievirus B5, with T-99.99 values ranging from 0.4 minutes at pH 6 and
7 to 4.6 minutes at pH 9  (Table 1; Figures 1 to 4).  Like coxsackievirus
B5, inactivation rates were generally greater at lower pH levels.  Bacteri-
ophage MS2 was inactivated more rapidly than coxsackievirus B5, with
T-99.99 values ranging from 1.2 minutes at pH 6 to 18 minutes at pH 8
(Table  1; Figures 1 to 4).  Inactivation of MS2 was faster than HAV at pH
6, similar  to HAV at pH 7, and slower than HAV at pH 8 and 9  (Table 1;
Figures 1 to 4).  As for  coxsackievirus B5 and bacteriophage 0X174, there
was a general pattern of  decreased  inactivation of MS2 at higher pH levels.

    The finding  that HAV  inactivation rates at pH 7 and 8 were somewhat
greater than at  pH  6 is unusual, because concentrations of HOC1 are lower
at pH 7 and 8 than  at pH  6, with the balance of the free chlorine at pH 7
and 8  in  the form of OC1~.  However, results at pH 9, where OC1~ predomi-
nates,  suggest that OC1~  is highly  virucidal for HAV  (Table 1).

    Another factor  which  may  influence virus inactivation rates by free
chlorine  at different pH  levels is  the degree of virus aggregation.
Aggregation was  controlled  in  stock virus preparations by selecting from
rate-zonal  density  gradients  only  those fractions corresponding to single
particles.  However, the  addition  of monodispersed virus stocks to reaction
mixtures  at different pH  levels may have caused virus aggregation,
resulting in slower inactivation kinetics due to protection from the disin-
fectant.  Results of previous  studies have shown that acid pH  levels can
induce  virus aggregation  and  decrease virus  inactivation rates  (22).

    Yet another  factor which  may influence the  rate of virus  inactivation
at different pH  levels  is differences  in the conformational form of the
virus.  A form of the virus existing at one  pH may be more resistant to
disinfection and/or less  infectious than another form existing at another
pH.  Both poliovirus type 1 and echovirus type  1 can  exist in  at  least two
different,  pH-dependent conformational forms  (22).  The  existence of dif-
ferent  conformational forms of HAV  has not been  fully established.
However,  preliminary evidence from this  laboratory  indicates  the  existence
of possibly two  conformational forms of  the  HM175  strain of HAV
 (unpublished results).

    The results  of  this study indicate that  the  HM175 strain  of  HAV  is
 relatively  sensitive to free  chlorine  and much  more  sensitive than  cox-
 sackievirus B5.   Bacteriophages 0X174  and MS2 also were  relatively  sen-
 sitive to free  chlorine,  thus making  them poor  indicators  for free  chlorine
disinfection  of  enteric viruses  such  as  coxsackievirus  B5  and HAV.
                                     239

-------
    Studies on the disinfection of other strains of HAV by free chlorine
are needed in order to determine if the sensitivity of the HM175 strain is
typical or representative of other strains.   Studies are also needed on HAV
disinfection by combined chlorine, especially monochloramine, U.V.  light,
and chlorine dioxide.  These studies are now in progress in this laboratory
and the results will be reported in the near future.

                          SUMMARY AND CONCLUSIONS

    HAV is relatively sensitive to disinfection by 0.5 mg/1 free chlorine
at 5°C and pH 6 to 9, and it is considerably more sensitive than cox-
sackievirus B5.  Bacteriophages 0X174 and MS2 are relatively sensitive to
free chlorine, and therefore, are poor indicators of enteric virus  inac-
tivation by this disinfectant.

                                REFERENCES

 1. Purcell, R.H., Feinstone, S.M., Ticehurst, J.R., Daemer, R.J.,  and
    Baroudy, B.M.  Hepatitis A virus,  lr±:  G.N. Vyas, J.L. Dienstag and
    J.H. Hoofnagle (eds.), Viral Hepatitis and Liver Disease, Grune and
    Stratton, New York, 1984, pp. 9-22.

 2. Sobsey, M.D., Shields, P.A., Hauchman, F.H., Hazard, R.L., and  Caton,
    L.W.,  III.  Survival and transport of hepatitis A virus in soils,
    groundwater and wastewater.  Water Science and Technology (in press).

 3. Lippy, E.G. and Waltrip, S.C.  Waterborne disease outbreaks - 1946-1980:
    a  thirty-five year perspective.  J. Am. Water Works Assoc. 76:60-67,
    1984.                            ~  ~~

 4. Sobsey, M.D., Oglesbee, S.E., Wait, D.A., and Cuenca, A.I.  Detection of
    hepatitis A virus  (HAV) in water.  Wat. Sci. Tech. 17, 1984.

 5. Hejkal, T.W., Keswick, B., LaBelle, R.L., Gerba, C.P., Sanchez, Y.,
    Dreesman, G., Hafkin, B., and Melnick, J.L.  Viruses in a community
    water  supply associated with an outbreak of gastroenteritis and infec-
    tious  hepatitis.   J. Am. Water Works Assoc. 74:318-321,  1982.

 6. Bitton, G., Farrah, S.R., Montague, C.L., and Akin, E.W.  Viruses  in
    drinking water.   Environ. Sci. Tech. 20:216-222,  1986.

 7. Neefe, J.R., Stokes, J., Baty, J.B., and Reinhold, J.G.  Disinfection
    of water containing a causative agent of infectious hepatitis.
    J.A.M.A. 128:1076,  1945.

 8. Neefe, J.R., Baty,  J.B., Reinhold, J.G., and Stokes, J.   Inactivation
    of the virus of  infectious hepatitis  in  drinking water.  Am. J. Pub.
    Health. 37:365-372,  1947.                                ~~  ~  	

 9. Peterson,  D.A.,  Hurley, T.R., Hoff, J.C., and Wolfe, L.G.  Effect  of
    chlorine treatment on infectivity  of  hepatitis  A  virus.  Appl.  Environ.
    Microbiol. 45:223-227,  1983.                             ~^   	
                                     240

-------
10. Grabow, W.O.K., Gauss-Muller, V.,  Prozesky,  O.W.,  and  Deinhardt,  F.
    Inactivation of hepatitis A virus  and indicator organisms  in water by
    free chlorine residuals.  Appl. Environ.  Microbiol.  46:619-624,  1983.

11. Grabow, W.O.K., Coubrough,  P., Hilner, C.,  and Bateman,  B.W.
    Inactivation of hepatitis A virus, other  enteric viruses and  indicator
    organisms in water by chlorine.  Wat. Sci.  Tech. 17:657-664,  1984.

12. Daemer, R.J., Feinstone, S.M., Gust, I.D.,  and Purcell,  R.H.
    Propagation of human hepatitis A virus in African green  monkey kidney
    cell cultures:  primary isolation and serial passage.   Infect. Immun.
    32:388-393, 1981.

13. Frosner, G.G., Deinhardt, F., Scheid, R., Gauss-Muller,  V.,  Holmes,  N.,
    Messelberger, V., Siegl, G., and Alexander,  J.J.  Propagation  of human
    hepatitis A virus in a hepatoma cell line.   Infection.  7:1-3, 1979.

14. Lemon, S.M., Binn, L.N., and Marchwicki,  R.H.  Radioimmunofocus assay
    for quantitation of hepatitis A virus in  cell cultures.  J.  Clin.
    Microbiol. 17:834-839, 1983.

15. Provost, P.J. and Hilleman, M.R.  Propagation of human hepatitis A
    virus  in cell culture in vitro.  Proc. Soc.  Exp. Biol. Med.
    160:213-221, 1979.

16. Gust,  I.D., Lehmann, N.I., Crowe, S., McCrone, M., Locarnini,  S.A.,  and
    Lucas, C.R.  The origin of the HM175 strain of hepatitis A virus. 0.
    Infect. Pis. 151:365-367, 1985.

17. Sobsey, M.D., Oglesbee, S.E., and Wait, D.A.  Evaluation of methods  for
    concentrating hepatitis A virus from drinking water.  Appl.  Environ.
    Microbiol. 50:1457-1463, 1985.

18. Sobsey, M.D., Jensen, H.R.,  and Carrick, R.J.   Improved methods for
    detecting enteric viruses in  oysters.  Appl. Environ. Microbiol.
    36:121-128.

19. Adams, M.H.  Bacteriophages,  Interscience,  New  York,  1959.

20. American Public  Health Association.   Standard Methods for the
    Examination of Water and Wastewater,  Section 408, pp. 294-325 and
    Section 415, pp. 369-372, 16th  edition, American Public Health
    Association, Washington, D.C.,  1985.

21. Engelbrecht, R.S., Weber, M.N., Salter, B.L., and Schmidt, C.A.
    Comparative  inactivation of  viruses  by chlorine.  Appl. Environ.
    Microbiol.  40:249-256,  1980.

22. Young, D.C. and  Sharp,  D.G.   Virion  conformational  forms and  the complex
     inactivation  kinetics of echovirus  by chlorine  in water.  Appl.
    Environ.  Microbiol.  49:359-364, 1985.
                                     241

-------
             DETECTION AND CONTROL OF CHLORINATION BYPRODUCTS
                             IN DRINKING WATER

                       by:  A. A. Stevens
                            R. J. Miltner
                            L. A. Moore
                            C. J. Slocum
                            H. D. Nash
                            D. J. Reasoner
                            D. Berman
                            Drinking Water Research Division
                            Water Engineering Research Laboratory
                            U.S. Environmental  Protection Agency
                            Cincinnati, OH  45268
                                 ABSTRACT

    Studies in the authors1  laboratories and pilot plants  focus on treat-
ment for the control of byproducts of the disinfection process in finished
drinking water.  Early work focused on the easily measured trihalomethanes
that are now regulated at the 0.10 mg/1  maximum contaminant level (MCL).
Current work involves identification and control  measures  for many more
byproducts of chlorination.   Ten finished drinking waters  were examined for
the presence of organic byproducts of chlorination in order to focus
efforts of treatment research on relevant target compounds.

    Technologies that are known to control  trihalomethanes are being
explored on the pilot plant-scale to achieve trihalomethane treatment
levels much lower than the current MCL.   The potential effect on disinfec-
tion efficiency is being investigated.

                               INTRODUCTION

         Studies in the authors' laboratories have three general objectives:
1) the detection and identification of non-trihalomethane  (THM) byproducts
of disinfection, 2) improve knowledge of the means to control these byprod-
ucts, and 3) control of microbiological  quality while attempting to lower
THM concentrations in finished drinking  water below the current maximum
contaminant level of 0.10 mg/1.  These three research objectives arise from
the United States Environmental Protection Agency (US EPA) regulatory
agenda that calls for a review of the current THM regulation while also
considering the regulation of other byproducts of disinfection.
                                    242

-------
                      DISINFECTION AND IDENTIFICATION

    In order to narrow the focus of future regulatory attention,  a limited
investigation of the presence of probable byproducts of chlorination was
undertaken of 10 finished waters from around the United States.   Although
no attempt was made to obtain a true random sample, the locations sampled
did represent a variety of water sources and treatment operations.  The
major characteristics of the 10 locations are summarized in Table,1.  All
locations used free chlorine at some point in the treatment process.

    This investigation of finished water quality had two parts,  the details
of which are described elsewhere (1,2).  Briefly, two "lists" of compounds
were addressed.  The "Short List" of 22 compounds shown in Table 2 was
developed by the US EPA Office of Drinking Water (ODW) from information
available in the literature.  Half of the compounds had previously been
reported in eight or more sources of finished drinking water, but many of
the others had previously been reported only once or not at all.
Analytical standards were not available for five of the compounds.

    Table 3 summarizes the findings of the analyses for the short list
compounds, giving both frequency of occurrence and approximate con-
centration ranges according to analytical confidence.  Of the short list
compounds, the trihalomethanes, dihaloacetonitriles, chloroacetic acids,
chloral hydrate, chloropicrin, and 1,1,1-trichloropropanone appear to be of
the most significance.  The compounds listed in Table 3 were found to
account for approximately 30 to 60 percent of the total organic halogen.

    The second list investigated in this 10 location study was developed
from the detection of compounds generated by the laboratory chlorination of
humic substances.  The bench-scale chlorination reactions were carried out
in approximately 20-liter volumes at three pH values (nominally 5, 7, and
11) and in the presence and absence of the bromide  ion at pH 7.   Ten to  15
liters of the samples after a three-day  reaction time were chemically
reduced, acidified, and pumped through adsorption columns for concentration
of the byproducts.  The adsorption columns were extracted with diethyl
ether, derivatized with diazomethane, and analyzed  by GC/MS.  Unique mass
spectra, MS area counts normalized relative to the  internal standard
(n-chlorododecane), and relative GC retention times were entered  into a
library of byproducts to be compared  to  portions of the field samples.
These field samples were analyzed by  the same adsorption/elution  technique.
To be considered a chlorination byproduct,  the MS area  count for  a  compound
in the chlorinated humic acid sample was  required  to be at least  three
times that of the unchlorinated control.  Approximately 500 entries  in  the
library from the chlorinated humic acid  studies met this byproduct  defini-
tion.

     Of  the  compounds  not  on the short  list,  196  entries from  the  10 loca-
 tions sampled matched byproduct  library  entries  from the  humic acid stud-
 ies.   Of  these,  128 were  unknowns, 63  containing  chlorine.   Sixty-eight
 entries  had tentative  structure or functional  groups  other than  chlorine
 assigned.   Of  these  68, 44 were acids  of various  types.
                                    243

-------
                             TABLE 1.   CHARACTERISTICS  OF  PARTICIPATING UTILITIES
Utility
Code
A
B
C
D
E
F
G
H
I
J
Population
Served
(++)
(+)
(++)
(++)
(+++)
(++)
(-H-)
( + )
(+++)
(-H-+)
Source
Surface
Shallow
Ground
Shallow
Ground
Surface
Surface
Surface
Surface
Surface
Surface
Ground
Organic
Carbon
mg/1
20-25(R)(T)
5-10(F)(T)
12(R)(T)
9(F)(T)
10-12(R)(T)
8(F)(T)
2.5(AR)
2.6(AR)
2.4(AR)
4.2(R)(T)
2.3(AR)
3.9(AR)
<1(F)
Disinfectant
Free
Residual
mg/1
3.2(F)
0.2(AR)
3.6(OF)
1.5(F)
0.85(F)
0.15(F)
0.2(F)
2.0(F)
1.7(F)
(combined)
l.O(F)
PH
8.6(F)
7.2(F)(T)
8.8(F)(T)
6.8-7(R)(T)
8.4(F)
7.6(R)
8.5(F)
7.7(R)
7.8(F)
7i4(F)
6.5(R)(T)
7.0(R)
•5.9(F)
8.4(R)
8.3(F)
8.6-8.7(F)
Treatment
Alum, polymer, lime, filtration, GAC,
post C12
Lime softening/polymer, C12
@ ~ 20 mg/1, filtration, post
C12 ~ 1 mg/1
Lime softening, recarbonation,
C12 @ ~ 15 mg/1, filtration
Anionic polymer, sand filtration
Alum coagulation and lime soften-
ing, pH adjustment w/C02, dual media
fi 1 tration
Alum, dual media filtration
Coagulation, settling, sand filtra-
tion, GAC adsorption
Prechlorination, F~
Lime, aluminum sulfate, sand filtra-
tion, post chlorine ~ 8 mg/1
Caustic, polymer, dual media
filtration, ~ 5hr free Cl before
NH3 addition
Lime softening, recarbonation,
sand filtration, F", C12 at end
(R)  = Raw water(F) = Finished water(T) = Typical value(AR)
(+)  = <10K        (++) = MO-249K         (+++) = >250K

-------
             TABLE 2.  DISINFECTION BY-PRODUCTS  -  SHORT  LIST
                                  Analytical          Number  of  Prior
                                  Standard             Occurrence
        Class/Compound            Available          Data  Citations
TRIHALOMETHANES

 1  chloroform                                  Trlhalomethanes are known
 2  bromodichloromethane                        to be ubiquitous where
 3  chlorodibromomethane                        chlorlnation is practiced.
 4  bromoform

HALOACETONITRILES

 5  bromochloroacetonitrile                               29
 6  dibromoacetonitrile                                   15
 7  dichloroacetonitrile                                  38
 8  trichloroacetonitrile                                  1

HALOACIDS

 9  dichloroacetic acid                                    3
10  trichloroacetic  acid                                  14

HALOALDEHYDES

11  dichloroacetaldehyde              No                    1
12  trichloroacetaldehyde                                  8

HALOKETONES

13  1,1-dichloropropanone             No                    1
14  1,1,1-trichloropropanone                               3
15  l,l-dichloro-2-butanone           No                    1
16  3,3-dichloro-2-butanone           No                    1
17  l,l,l-trichloro-2-butanone        No                    1

CHLOROPHENOLS

18  2-chlorophenol                                         0
19  2,4-dichlorophenol                                     2
20  2,4,6-trichlorophenol                                  4

MISC.

21   chloropicrin                                         24
22   cyanogen chloride                                      8
                                     245

-------
TABLE 3.  SUMMARY OF RESULTS GROUPED ACCORDING
    TO ANALYTICAL CONFIDENCE - SHORT LIST




Compound
High Confidence
Chloroform
Bromodichloromethane
Chlorodibromomethane
Bromoform
Dichloroacetonitrile
Dibromoacetonitrile
Bromochloroacetonitri le
Chloropicrin
Low Confidence
Chloroacetic acid
Dichloroacetic acid
Trichloroacetic acid
Trichloroacetaldehyde
(as Chloral hydrate)
1 , 1 , 1-Tri chl oropropanone
2-Chlorophenol
2,4-Dichlorophenol
2,4,6-Trichlorophenol

Qualitative Onl^
1 , 1-di chl oropropanone
1 , 1-Di chl oro-2-butanone
3, 3-Di chl oro-2-butanone
1,1, 1-Tri chl oro-2-butanone
Cyanogen chloride
Dichloroacetaldehyde
*
+ = less than 10
++ = between 10 and 100
+++ = greater than 100
Number of
Locations
Where Found
of Those
Analyzed

10 of 10
10 of 10
10 of 10
6 of 10
10 of 10
3 of 7
7 of 7
8 of 10

6 of 10
10 of 10
6 of 10
10 of 10

10 of 10
10 of 10
0 of 10
0 of 10
0 of 10

0 of 8
0 of 8
1 of 8
0 of 8
1 of 7
0 of 10





Range of
Concentrations*
(M/l)

2.6 to 594
2.6 to 77
0.1 to 31
0.1 to 2.7
0.2 to 9.5
0.4 to 1.2
0.2 to 4.0
0.2 to 5.6

+
+ to +++
+ to ++
+ to ++

+ to ++
--
—
—
_-.


—
__
_ —
—
— —



                     246

-------
    Table 4 summarizes the frequency of occurrence of the 196 library
entries.  Count of locations is the number of locations where any entry was
observed.  For example, six compounds were found at all 10 locations while
51 compounds were found at only one of the locations.  Cumulatively, 51
compounds were found at more than half of the locations.

            TABLE 4.  FREQUENCY OF OCCURRENCE OF THE LONG LIST
                     ENTRIES FOUND IN 10 LOCATION STUDY


Number (Count) of
    Locations          10    987654321
Entries occurring       6   10    9   14   12   20   15    24   35   51
at this Count of
Locations

Cumulative              6   16   25   39   51   71   86   110  145  196
at Count or Greater
             TREATMENT-RELATED STUDIES FOR NON-THM BYPRODUCTS

    The survey work described above aids in focusing on fewer candidate
compounds in treatment studies.  These studies must include formation and
decay studies under the  reaction conditions expected at water treatment
plants.  Both precursor  and specific compound removal strategies need to be
examined, as well as the effect of disinfectants other than free chlorine.
Currently, analytical method requirements are severely hampering progress
in these byproduct control related areas, especially for the very important
haloacids and chloral hydrate on the short list.  For the longer list
entries, treatment studies theoretically could be carried out to some
extent even without identifying unknowns, and this will be pursued to the
extent possible as studies of the short list compounds progress.

             CONTROL OF  MICROBIOLOGICAL QUALITY AND LOWER THM

    The US EPA has stated that, in the future, the total trihalomethane
(TTHM) maximum contaminant level (MCL) of 0.1 mg/1 will likely be lowered.
The 1986 Amendments to the Safe Drinking Water Act call for the regulation
of 25 new contaminants in 1991, and a reconsideration of the TTHM MCL will
take place then.  In anticipation of this, pilot-scale studies were con-
ducted employing known control measures for TTHM control and monitoring for
microbiological indicators to ensure that treatment modifications to
improve organic water quality would not compromise microbiological water
quality.  Pilot-scale studies also provided an opportunity to study for-
mation and control of other disinfection byproducts observed in the 10
location study discussed above.
                                    247

-------
PILOT PLANT OPERATION

    Sufficient untreated Ohio River water to operate two pilot-scale water
treatment plants at 1.7 gpm, was collected from the  pump station at the
Cincinnati Water Works and trucked daily to the US EPA pilot plant.  Upon
receipt, it was transferred to a 5,000-gallon storage tank.   A submersible
pump kept sediment in suspension during storage.   Jar tests  were conducted
to determine the optimum coagulant dose for turbidity control.  Alum was
used as the coagulant in these studies.  Typically,  a dose that gave <1 Ntu
in jar test-settled water gave 1 to 3 Ntu in pilot plant-settled water.  A
diagram of treatment options is given in Figure 1.  Parallel.pilot plants
that provided mixing, flocculation, gravity sedimentation, sand/anthracite
dual-medial filtration, clear well storage, and filter backwash were
employed.  With the exception of some pump parts,  construction materials
were limited to stainless steel, teflon, and glass.   Chemicals not added at
the mix tanks were added ahead of in-line mixers between unit processes.
Treatment chemicals shown in parentheses in Figure 1 were optional
depending on study objectives.  Unless noted, disinfection followed prac-
tices recommended in the Ten State Standards (TSS)(3).  For each study, the
plants were operated continuously for five weekdays  in two consecutive
weeks.  Following start-up and the establishment of  steady-state operation,
samples were collected daily from the middle to the  end of the week.
Sample points were selected to isolate unit processes and are indicated by
open circles in Figure 1.  Samples were collected for five or six days over a
two-week period.  In this presentation, mean values  are reported.  Studies
were conducted at different times throughout the year.  In the tabled data
presented, temperature, pH, and the background organic and microbiological
quality of Ohio River water for each study is described.

    Unit process detention times were  constant for all studies and were:
two minutes mixing, 3/4 hour flocculation, 9-1/2 hours sedimentation,  15
minutes filtration, and 8-1/2  hours clear well storage.

    To  boost background levels of  indicators of microbiological quality,
municipal  primary sewage was blended (1 to 1,000 dilution)  in the  storage,
tank during raw water transfer from the truck.  This typically provided 4
logs bacterial density, as measured by  total coliform  (TC), standard plate
count  (SPC), and membrane filter  heterotrophic plate count  (m-HPC), and
allowed monitoring of bacterial penetration  through the unit  processes.
The differences,  if any,  in the treatment  resistance of municipal,
intestinal-oriented bacteria and  natural, aquatic bacteria  are not  known.
As a viral indicator, laboratory-cultured strains of either F2 or  the  more
chlorine-resistant MS2 bacteriophage were also added in 4-log densities, by
in-line mixer, to raw water before treatment.

TRIHALOMETHANE DEFINITION

     In  these studies, both  instantaneous and terminal TTHM  (inst  TTHM  and
term TTHM) were sampled;  these are described in detail elsewhere  (4).   This
allowed the measurement of THMs formed  during  treatment,  i.e.,  inst TTHM.' It
                                     248

-------
also allowed the measurement of reacted and unreacted THM precursor material,
i.e., term TTHM, so that unit-process control of precursor could be eval-
uated.  Term TTHM samples were prepared in these studies by the addition of
15 mg/1 free chlorine followed by seven day, room-temperature storage.
Previous studies had shown that the TTHM reaction in Ohio River water was
typically at or near completion under these conditions.  TOX was sampled
identically.

MOVING THE POINT OF CHLORINATION

    A commonly-used approach to TTHM control was studied first, i.e.,
moving the point of chlorination from raw water to a better quality water.
Chlorine was added to raw water at one plant and to settled water at the
parallel plant.  For this study, clear well storage of filtered water was
not employed.

    Results are given in Table 5.  Lower formation of both inst TTHM, 27
vs. 73 ug/1, and inst TOX, 132 vs. 287 ug Cl/1, was observed when settled
water was chlorinated.  This is consistent with earlier TTHM control stud-
ies (4).  By allowing clarification to occur prior to chlorination, a water
with less precursor material was chlorinated.  This can be seen in Figure
2.  When prechlorinating raw water, 276 ug/1 precursor material, as
measured by term TTHM, was present, but when chlorinating settled water,
only 191 ug/1 precursor material was present.  Further, less contact time
was available to drive chlorination reactions, i.e., 15 minutes vs. 10-1/2
hours.

         TABLE 5.  LOWERING THMs BY SHIFTING POINT OF CHLORINATION
Parameter
Temperature, °C
Raw term TTHM, ug/1
Raw term TOX, ygCl/1
Chlorine dose, mg/1
Filtered free chlorine, mg/1
Raw pH
Filtered pH
Filtered inst TTHM, ug/1
Filtered inst TOX ygCl/1
% removal term TTHM
% removal term TOX
Raw TC/100 ml
Raw F2 phage/ml
Raw m-HPC/ml
Filtered TC/100 ml
Filtered F2 phage/ml
Filtered m-HPC/ml

Raw Water
Quality3
28
276
794


8.4





280,000
65,700
107,500



Resul
Raw Water
Chlorination



4.6
0.85

7.6
73
287
24
14



<1
<0.5
<1
t*
Settled
Water
Chlorination



1.3
0.4

7.6
27
132
28
48



<1
<0.5
36

No
Disin-
fection



0
0

7.5







415
3,680
42,200
 •Wean of six sample days.
                                      249

-------
                        ALUM
                        
           SEHAGE
                  HUGE
FLOC
                                                             -o
                         T
                        ALUM
                        (CL2)
                       (ACID)
[| FILTER

O
CLEAR
WELL

                                                             -o
            (CL2)
                     (BASE)
Figure  1.   Treatment scheme  for pilot-scale organics control  studies.
                       o
                       s








^/O
216

69




^

1


191
73


X
^


^
199

27
                            RAW
     SETTLED     FILTERED
                                  « IHST

                                 | » TERM HHH

                               [] - TTHM FORHATIOH POTENTIAL
Figure  2.   Raw  (left bar)  vs.  settled  water  (right bar)  chlorination
            of Ohio River water.
                                    250

-------
    Microbiological data indicated comparable quality in both filtered
waters.  Although a 15-minute contact time was less than that recom-
mended by the TSS, it was sufficient for complete reduction of total coli-
form and F2 bacteriophage, and near-complete reduction of bacterial  density
as measured by m-HPC.  This lends support to movement of the chlon"nation
point of TTHM control without compromising microbiological control,  and is
consistent with earlier microbiological/TTHM studies (4).

IMPROVING PRECUSOR REMOVAL

    Settled water containing 191 ug/1 term TTHM represented 31 percent
removal of term TTHM.  In a follow-up study, the coagulation process was
modified in an attempt to improve precursor removal prior to chlorination
of settled water.  A series of jar tests were conducted on Ohio River water
in which the pH was lowered both by the addition of alum and by adding
hydrochloric acid.  Low-pH coagulation had been shown earlier to enhance
precursor removal  (5).  Term TTHM was determined for jar test-settled
water.  These studies indicated that pH 6 and 90 mg/1 alum could increase
term TTHM removal to 50 percent.  (20 to 35 mg/1 alum was typically
required simply for turbidity control.)  The pilot plants were run with
prechlorination and typical coagulation in parallel with settled-water
chlorination and modified coagulation as described.  In the modified plant,
sodium hydroxide was added at the clear well in an attempt to produce
parallel finished waters of comparable stability.

    Results of this experiment are shown in Table 6.  Low-pH coagulation
provided better control  (46 vs. 30 percent removal of term TTHM and 56 vs.
35 percent  removal of term TOX) than did coagulation designed simply for
turbidity control.  Although low-pH  conditions  produced a  less-dense,
poorer-settling floe, similar turbidities were  observed at both sedimen-
tation basin weirs.  With enhanced precursor  removal, lower TTHM formation
resulted  (21 vs.  54 ug/1  inst TTHM and 84 vs.  153 ug/1  inst TOX) in
finished waters.   In finished water  samples  stored  six  days without further
addition of chlorine, the 21 ug/1  inst TTHM  concentration  increased to  only
40 ug/1, suggesting  that  low-pH coagulation  and delayed  chlorination  can
result in  inst TTHM  levels  in Ohio River water  well  below the 0.1 rng/1  MCL.

    Comparable microbiological quality was observed  in  both  finished  waters
with  8-3/4  hours  contact  times available  in  the plant employing  settled water
chlorination.   In this  study, MS2  bacteriophage was  used.   Comparing  this
and the  previous  study,  no  difference was observed  in bacteriophage
control.   Neither the MS2 nor the  F2 strains  were  measurable  in  the first
sample points  following  chlorination.

CHLORAMINATION

     Early  studies had  shown  that monochloramine would  not drive  the THM
 reaction (4)  and  would  drive  the TOX reaction only to  a limited  extent
 relative to free  chlorine  (6).  Monochloramine was  studied at the  pilot-
 scale to examine  microbiological quality  and formation  of other  disinfec-
 tion  by-products.  Both  plants employed  coagulation designed only  for
                                      251

-------
        TABLE 6.  LOWERING THMs BY SHIFTING POINT OF CHLORINATION
                     AND OPTIMIZING PRECURSOR REMOVAL
                                                       Result*
Parameter
Raw Water
Quality*
Raw Water
Chlorination
Settled Water
Chlorination
Temperature, °C                    26
Raw Term TTHM, yg/1               292
Raw term TOX, ugCl/1              551
Chlorine dose, mg/1                              3.1               1.8
Finished free chlorine, mg/1                     0.4               0.3
Raw pH                              7.7
Filtered pH                                      7.3               6.15
Finished pH                                      7.2               7.3
Finished inst TTHM, yg/1                        54                21
Finished inst TOX, ugCl/1                      153                84
% removal term TTHM                             30                46
% removal term TOX                               35                56
Raw TC/100 ml                  18,000
Raw MS2 phage/ml               43,600
Raw SPC/ml                     39,400
Finished TC/100 ml                              <1                <1
Finished MS2 phage/ml                           <0.5              <0>5
Finished SPC/ml                                 <1                
-------
turbidity control.  In parallel, one plant employed prechlorination  and  the
other employed prechloramination.  To accomplish prechloramination,  ammo-
nium hydroxide was added in 100 percent stoichiometric excess  and  mixed
prior to the addition of chlorine at the coagulant mix tank.   No measurable
free chlorine of dichloramine resulted.

    As expected, no measurable THMs were formed in chloraminated water and
significantly lower concentrations of TOX were formed.  See Table  7.  The
low THM and TOX concentrations observed at the chlorinated plant were a
result of low temperature.  Finished water haloacetonitrile,  chloropicrin,
and trichloropropanone concentrations from both plants are given in  Table
8.  Dichloroacetonitrile and 1,1,1-trichloropropanone were the principal
species formed upon chlorination.  These data are in agreement with  data
produced during the 10 location study.  None of these compounds were found
in finished waters as a result of chloramination.

    Although bacteria and bacteriophage penetrated further into the  plant
that was treated with monochloramine, concentrations of these indicators
and total coliform densities were comparable in the finished waters  from
both plants.  These data indicated that chloramination following the TSS,
i.e., three hours contact time resulting in 1 to 2 mg/1 combined residual
chlorine, is sufficient for microbiological control.  These results  indi-
cate that proper use of chloramines will result in little or no formation  of
THMs and will provide proper, in-plant, microbiological control.  Because
these pilot studies could not simulate water distribution, the question  of
bacterial regrowth in the presence of a weaker disinfectant remains.  Many
utilities have switched from chlorine to chloramine as a disinfectant and
realized both THM control and proper disinfection.  However, some utilities
have lowered THMs but have encountered difficulty with the use of chlora-
mine because, having a relatively-lower oxidation potential, chloramines
are less effective than free chlorine for color or taste-and-odor
control  (7,8).

    The  chloramine study was repeated with the monochloramine concentration
lowered  by 50 percent.  Even with concentrations below those recommended by
the TSS, microbiological densities were aceptably low.  See Table 9.  With
monochloramination, disinfectant byproduct formation was identical  to that
reported in Table 8.  In this study the parallel plant was operated with no
disinfectant to observe microbiological control provided by pH  suppression,
clarification, and filtration.  Results are given in Table 9.   An undisin-
fected,  parallel  filter was operated  in an aforementioned study;  its micro-
biological results were given in Table 5  (last  column).  These  data  suggest
that the pilot plant, operated with no disinfection, provided bacterial
densities typical of those observed in full-scale plants treating Ohio
River or other surface waters prior to disinfection  (9) and was,  therefore,
adequate for predictive purposes.
                                    253

-------
           TABLE 7.   LOWERING THMS WITH THE USE OF CHLORAMINES
         Parameter
Raw Water
Quality*
                                                       Result*
                                              Raw Water
                                             Chlorination
Settled Water
Chlorination
Temperature, °C                    12
Raw Term TTHM, yg/1               238
Raw term TOX, ygCl/1              499
Chlorine dose, mg/1
Monochloramine dose, mg/1
Finished free chlorine, mg/1
Finished monochloramine, mg/1
Raw pH                              8.0
Finished pH
Finished inst TTHM, ug/1
Finished inst TOX, ygCl/1
%  removal term TTHM
%  removal term TOX
Raw TC/100 ml                  33,300
Raw m-HPC/ml                   15,800
Raw SPC/ml                     22,600
Finished TC/100 ml             29,100
Finished MS2  phage/ml
Finished m-HPC/ml
Finished SPC/ml
                  4.1
                  0.8
                  0.1
                  7.0
                 16
                115
                 54
                 53
                 <0.3
                 <0.01
                 <0.3
     2.3
    <0.1
     1.3

     7.0
    <0.1
    20
    <0.3
     0.1
     0.8
 *Mean  of five  sample  days.

                 TABLE 8.   DISINFECTION  BYPRODUCT FORMATION


                                                    Result*
Parameter
Temperature, °C
Chlorine dose, mg/1
Monochloramine dose, mg/1
Finished free chlorine, mg/1
Finished monochloramine, mg/1
Inst CC13CN, ug/1
Inst CHC12CN, ug/1
Inst CHBrClCN, yg/1
Inst CHBr2CN, yg/1
Inst CCl3N02, yg/l
Inst TCP, yg/1**
Raw Water
Chlorination
12
4.1

0.8
0.1
<0.1
3.1
ND
ND
ND
2.8
Raw Water
Chloramination
12

2.3
<0.1
1.3
ND
ND
ND
ND
ND
ND
  *Mean of five sample days.
 **!,!,1-trichloropropanone.
 ND = not detected @ 0.1 yg/1.
                                    254

-------
                  TABLE 9.  CHLORAMINATION OF RAW WATER
                                                       Result*
                               Raw Water      Raw Water       Settled Water
         Parameter             Quality*      Chlorination     Chlorination


Temperature, °C                     9
Raw term TTHM, yg/1               204
Raw term TOX, ygCl/1              432
Monochloramine dose, mg/1                        1.1
Finished free chlorine, mg/1                    ND
Finished monochloramine, mg/1                    0.6
Raw pH                              8.0
Finished pH                                      7.3               7.4
Finished inst TOX, ygCl/1                       44
Raw TC/100 ml                  29,300
Raw MS2 phage/ml               15,400
Raw m-HPC/ml                   15,800
Raw SPC/ml                     75,800
Finished TC/ml                                  <1                19
Finished MS2 phage/ml                            2.5               1.3
Finished m-HPC/ml                                0.1           1,020
Finished SPC/ml                                  1.1             280
*Mean of five sample days.
ND = not detected.
                                   255

-------
                               REFERENCES

1.  Stevens, A. A. et ah  By-products of chlorination at ten operating
   utilities.  Proceedings of the 6th Conference on Water Chlorination,
   Environmental Impact and Health Effect,  Oak Ridge, Tennessee, May 3-8,
   1987.
                *

2.  Stevens, A. A. et ah  Chlorinated humic acid mixtures establish cri-
   teria for detection of disinfection byproducts in drinking water.
   Proceedings of the Symposium on Influence of Aquatic Humic Substances
   on Fate and Treatment of Pollutants, American Chemical Society, Denver,
   Colorado, April 5-10, 1987.

3.  Recommended standards for water works:  a report of the committee of
   the Great Lakes/Upper Mississippi River Board of State Sanitary
   Engineers.  Health Education Service, Albany, 1982.

4.  Symons, J. M. et aJ.  Treatment techniques for controlling
   trihalomethanes in drinking water.  EPA-600/2-81-156, Drinking Water
   Research Division, U.S. EPA, Cincinnati, September, 1981.

5.  Semmens, M. J. et ah  Optimizing coagulation-adsorption for haloform
   and TOC reduction.  EPA-600/2-83-042, Drinking Water Research Division,
   U.S. EPA, Cincinnati, September, 1983.

6.  Stevens, A. A. et ah  Organic halogen measurements:  current uses
   and future prospects.  Jour. AWWA.  Vol. 77, No. 4, April, 1985.

7.  Singer, P.  Control of trihalomethanes using alternative oxidants and
   disinfectants.  Proceedings of the U.S. EPA/AWWARF Joint Conference on
   Current Research in Drinking Water Treatment, Cincinnati, Ohio, May,
   1987.

8. McGuire, M. J. and Meadow, R. E.  AWWARF trihalomethane survey - a
   progress  report.  Presented at U.S. EPA/AWWARF Joint Conference on
   Current Research in Drinking Water Treatment, Cincinnati, Ohio, May,
   1987.

9. Ohio River Valley Water Sanitation Commission.  Water treatment process
   modifications for trihalomethane control and organic substances in the
   Ohio River.   EPA-600/2-80-028, Drinking Water Research Division, U.S.
   EPA, Cincinnati, March, 1980.
    U.S. GOVERNMENT PRINTING OFFICE: 198^ 5 „ 8_
                                  5 „,  6 ? „ 8 «.
                                    256

-------