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
------- |