United States
Environmental Protection
Agency
Risk Reduction
Engineering Laboratory
Cincinnati, OH 45268
Research and Development
EPA/600/SR-93/136 September 1993
EPA Project Summary
Disinfection By-Product
Formation by Alternative
Disinfectants and Removal by
Granular Activated Carbon
Wayne E. Koffskey
The effects of the use of alternative disin-
fection (chlorine, ozone, chtoramine, chlo-
rine dioxide) on the formation of hatoge-
nated by-products and the removal of these
by-products by granular activated carbon
(GAG) were evaluated over a 1-yr opera-
tional period. Disinfection by-products ex-
amined included trihalomethanes (THMs),
hafoacetic acids (HAAs), haloace-tonfofes
(HANs), hatoketones (HKs), chloral hydrate
(CH), and chtoropicrin (CP). Microbiological
information was also obtained on the oper-
ating systems and included heterotrophic
plate count, total coliform, and MS2 coliph-
age. Other parameters evaluated included
total organic carbon (TOC), total organic
halkte (TOX), and assimilable organic car-
bon (AOC).
Each of four disinfectant process streams
was composed of a 30 min contact cham-
ber followed by a sand column in series
with a GAG column, the latter having a 20
min empty bed contact time (EBCT). One of
the four disinfectants was applied at the
beginning of each process stream. A fifth
nondisinfected process stream, consisting
of only a sand column in series with a GAG
column, was used as a control.
The lowest levels of halogenated disin-
fection by-products resulted from the com-
bination of preozonation and post
chtoraminatton after sand filtration with an-
nual simulated distribution system averages
of 27 ug/L of TOX and 12 ug/L. for the sum of
18 disinfection by-products. These respec-
tive concentrations were further reduced to
13 ug/L and 7 ug/L after GAC treatment.
Although ozonation produced significant lev-
els of AOC, sand filtration reduced these
levels by an average of 77% to 39 u,g/L and
subsequent GAC treatment provided a fur-
ther reduction to 4 jjg/L.
This Project Summary was developed by
EPA's Risk Reduction Engineering Labora-
tory, Cincinnati, OH, to announce key fold-
ings of the research project that is fully
documentedin a separate report of the same
title (see Project Report ordering informa-
tion at back).
Introduction
Chlorine has been widely used throughout
the United States for disinfecting drinking water.
During the disinfection process, chlorine reacts
with naturally occurring organic matter to form a
number of halogenated disinfection by-prod-
ucts. Existing disinfectbn/disinfection by-prod-
uct regulations apply only to trihalomethanes
with a maximum contaminant level (MCL) of
0.10 mg/L. Anticipated future Federal regula- .
tions for disinfection/disinfection by-products will
potentially affect most water treatment plants in
the United States.
Traditionally, drinking water standards for con-
taminants are set at the lowest possible, techni-
cally and economically feasible number. There
must, however, be assurance that drinking wa-
ter is mbrobbbgically safe, which may mean
that a greater risk will have to be accepted from
disinfectants and disinfection by-products in the
water than for other contaminants. This re-
search project was devebped to evaluate the
formation of disinfection by-products by alterna-
tive disinfectants and their removal by GAC
while evaluating the microbbbgbal quality of
the treated water. Specific objectives of this
project are: •
Printed on Recycled Paper
-------
• measure the effects of alternative
disinfectants on the formation of
halogenated disinfection by-products
including the trihalomethanes, the
habacetic acids, the habacetonitriles, the
haloketones, chloral hydrate, and
cWoropfcrin.
* measure the effectiveness of GAG filtration
following sand filtration in removing
habgsnated disinfection by-products and
their precursors.
• measure the general microbiological quality
of water treated with the alternative
disinfectants.
• assess the levels of AOC formed during
disinfection with ozone and chlorine.
* evaluate the effectiveness of the alternative
disinfectants in the inactivation of MS2
coTphage.
To meet these objectives, a pilot column
system consisting of four disinfected process
streams (ozone, chlorine dioxide, chlorine, and
chforamine) and a nondtsinfected process stream
was continually operated for one year.
Treatment Plant
Lower Mississippi River water entering the
34 mgd Permutit* treatment plant at Jefferson
Parish was dosed with 1 to 6 mg/L
diallyldimethylammonium chloride, or di-
methylamlne polyetectrolyte polymers, or both
for darificatfon, 0.1 to 0.3 mg/L fluosilbb acid
(as fluoride) for fluoridation, and 2 mg/L pow-
dered activated carbon for organics control.
After clarification via Permutit upftaw precipita-
tofs, a smaB portion of clarified water was di-
verted to a pilot column system and was filtered
through one of two pressure sand fitters at a
hydraulic loading of 1.7 gpm/f?. Each sand filter
contained 30 in. of 0.45 mm-fitter sand and
provided an average nondisinfected sand fil-
tered water flow of 8.5 gpm to the rest of the
plot column system.
Pilot Columns
The filtered water was split into five process
streams, one for each of the four disinfectants
and another nondisinfected process stream used
as a control. Each disinfected process stream
consisted of a 30-min disinfectant contact cham-
ber foUowed by series filtration through a sand
column and a GAG column. The nondisinfected
process stream consisted of only a sand col-
umn in series with the GAG column.
Each disinfectant contact chamber was con-
structed with the use of a 12-in. diameter stain-
less steel pipe that was lOfthigh, exceptforthe
ozone contact chamberthat was 11 ft high. The
sand and GAG columns were constructed from
10-ft sections of 6-in. diameter glass pipe. All
plot column components were constructed from
stainless steel, glass, and teflon. The sand col-
umns were charged with 30 in. of 0.45-mm filter
sand, and the GAG columns were charged with
6.8 ft of 12 x 40 mesh GAG to achieve a 20-min
EBCT at a fbw of 0.5 gpm. Each column was
operated at a hydraulic loading of 2.5 gpm/ft2
and was backwashed only when necessary to
achieve the desired flow rate. No media loss
was observed during backwashing. The GAG
used in this study was Ceca GAG 40 and was
selected after a thorough evaluation of varbus
types of GAG.
During the course of the operational period,
water temperatures ranged from 3 to 29 °C.
After the addition of the varbus disinfectants,
sight variations in pH were observed. On the
average, the pH of the chtarine dbxide contact
chamber effluent decreased 0.6 units to pH 7.0
whereas that for the chbrinejand chbramine
contact chambers increased"671"and 0.2 units
to pH 7.7 and 7.8, respectively. No change in
pH was observed for the ozone process stream.
A 96% yield of chbrine dbx'de was gener-
ated by the in-line mixing of hypochbrite/chlorite
and sulfuric acid solutions before its injection
into the process stream. Chbramines were
formed within the process stream with the injec-
tion of hypochlorfte followed within a few sec-
onds by ammonium hydroxide. The average
30-min demands for each disinfectant deter-
mined during the operational period are com-
pared in Table 1 atang with their average disin-
fectant contact chamber effluent residual con-
centratbns. Although residual concentrations
were measured as chlorine using the DPD
titrimetric method, they are reported as the spe-
cific disinfectant indicated, not as free chlorine.
The ozone residual dissipated completely across
the sand column, but the other three disinfec-
tant residuals were only slightly reduced. No
residual species of any disinfectant were ob-
served in the effluent of the GAG columns.
Pilot Column Performance
Microbiological Effectiveness
At the dosages used, all of the disinfectants
reduced total coliforms to acceptable levels.
With heterotrophb bacteria, however, all disin-
fectants except chbramines reduced the levels
to bebw 100. Other organisms may not have
been as effectively controlled by all disinfec-
tants. For instance, during one study when
MS2 Coliphage was spiked into the pilot plant,
chbramines were ineffective for this virus indi-
cator. Future studies at Jefferson Parish will
apply the concentration xtime (C«t) concept for
determining disinfectant efficiency.
Halogenated By-Product Control
During the 1-yr operation of the pibt plant,
1wo surrogate parameters that give an indica-
tion of organb concentrations, including habge-
nated by-products, were evaluated: TOG and
TOX. The average TOG concentratbns in the
disinfectant contact chamber effluents were 3.1,
2.9,3.2,3.2 and 3.2 mg/L for the nondisinfected,
ozone, chtarine dioxide, chbramine, and chta-
rine process streams, respectively. A slight re-
duction in TOG averaging 0.3 mg/L was indi-
cated across the ozone contact chamber,
whereas the concentration of TOG remained
fairly constant for the other disinfectants. TOG
was further reduced across the ozone sand
column by an average of 0.8 mg/L when com-
pared with the nondisinfected influent. Based
on the levels of heterotrophb bacteria in the
effluents of the ozone contact chamber and
ozone sand column, the reduction of TOG across
the ozone contact chamber appears to have
resulted primarily from oxidation and that across
the ozone sand column can be attributed to
biodegradation.
With an average nondisinfected TOX influent
concentration of 25 pg/L, TOX concentrations
increased significantly after 30 min of disinfec-
tant contact time to 86 u,g/L for chlorine dbxide,
99 u,g/L for chtaramine, and 246 |ig/L for chb-
rine.
The same trend seen for TOG was also
observed for TOX where an average reduction
of 33% occurred in the ozone contact chamber
to produce an average effluent concentration of
16 |og/L with a further reduction to 11 u,g/L
across the ozone sand column.-Treatment.of
the sand filtered effluents with free chbrine
followed by a 5-day storage to simulate the
distribution system significantly increased TOX
Table 1. Average Disinfectant Contact Chamber Demands and Residuals
' Mention ol tfada names or commercial products does
not constitute endorsement or recommendation for
use.
Ozone
Chlorine dioxide
Chlorine
Monochloramine
Average
30-min
disinfectant
demand
(mg/L)
2.5
0.7
1.8
0
Process stream
average
disinfectant
contact time
(min)
30
30
30
30
Process stream
average
disinfectant
residuals
(mg/L)
0.5
0.5
1.0
2.2
-------
concentrations for all process streams (557,
540,339, and 379 ugA. for nondisinfected, chlo-
rine, ozone, and chlorine dioxide, respectively).
The THMs reacted as expected with no
significant concentrations (1 ug/L average) ob-
served in the disinfectant contact chamber and
sand column effluents for the nondisinfected,
ozone, and chlorine dioxide process streams.
An average THM concentration of 3 ug/L oc-
curred in the chbramine disinfectant contact
chamber and sand column effluents whereas
that in the chlorine contact chamber effluent
averaged 39 ug/L and increased to 49 ug/L
across the chlorine sand column. By maintain-
ing a chloramine residual for 5 days, the termi-
nal THM concentrations increased slightly to an
average of 8.5, 3.2, 4.2, and 9.4 u,g/L for the
nondisinfected, ozone, chlorine dioxide, and
chloramine process streams, respectively. Simi-
lar treatment and storage with free chbrine
produced relatively high terminal THM concen-
trations for the nondisinfected and chlorine pro-
cess streams with average concentrations of
236 and 225 ug/L. When compared with the
nondisinfected sand column effluent, terminal
THMs were 35% and 41% less when ozone
and chbrine dioxide were used during pretreat-
ment, i.e., average concentrations of 154 and
138 ug/L, respectively. As expected, GAC 're-
duced all concentrations until the columns be-
came saturated in 60 to 80 days.
The haloacetic acids folbwed the same trend
as seen with THMs except at a tower concen-
tration. The highest concentrations were formed
by using free chbrine, which mainly produced
dfchforoacetic acid (DCAA), trichbroacette acid
(TCAA), and bromochbroacetb acid (BCAA).
Chloroacetic acid (CAA), bromoacetic acid
(BAA), and dibromoacetfo acid (DBAA) were
also formed to some extent. Average instanta-
neous concentrations for DCAA were 0.9, 1.7,
3.7, 1.1, and 13 ug/L for the nondisinfected,
chlorine dbxide, chbramine, ozone, and chfo-
rine process streams after sand filtratbn, re-
spectively. Five-day terminal values for the same
process streams with free chbrine were 60,44,
38, and 60 ug/L for the nondisinfected, chbrine
dbxide, ozone, and chlorine streams, respec-
tively.
Similar treatment of the sand filter effluents
with chbramine and a 5-day storage period
resulted in only slightly elevated DCAA concen-
trations. The slight concentratbn increase is
similar to that seen for THMs during similar
treatment with chtaramine suggesting that both
were formed by free chlorine during in-situ for-
matbn of chbramine. GAC adsorption produced
continued removals of 80% or greater of DCAA
after steady-state was reached at about 150
days of operatfon. After 5-day storage with free
chbrine, GAC steady-state was reached in about
250 days for DCAA with average removals
after steady-state of 48%, 73%, 53%, 46%, and
51% for the nondisinfected, ozone, chbrine
dioxide, chloramine, and chlorine process
streams, respectively. Similar observations for
all unit processes were also seen for TCAA.
Relatively low concentrations of HANs were
formed across each process stream with chb-
rine producing the highest levels which aver-
aged 3.1 ug/L total HANs. An average of less
than 1 ug/L of HANs was observed across the
other process streams. The predominant HAN
was dbhbroacetonitrile (DCAN) folbwed by
bromochloroacetonitrile (SCAN), dtoromoaceto-
nitriie (DBAN), and trichbroacetonitrile (TCAN).
Treatment of the chbrine sand column effluent
with additional free chbrine and subsequent 5-
day storage produced an average concentra-
tion of 1.9 ug/L of DCAN, which was the same
as that of the sand column effluent suggesting
that all DCAN precursors had reacted. No con-
sistent breakthrough of the HANs was observed
through any GAC column except that of the
chlorine process stream, which was still remov-
ing over 95% of the influent HANs at the end of
the 1 -yr operational period.
Only two haloketones, 1, 1, 1-trichloro-
propanone (TCP) and 1, 1-dichtoropropanone
(DCP) were detected with the highest concen-
tratbn (1 to 2 ug/L) being observed in the
chlorine process stream. Post >chlorinatbn of
the sand column effluents folbwed by 5-day
storage produced similar TCP concentrations of
2.1,2.5,4.2, and 2.5 ug/L for the nondisinfected,
ozone, chtarine dbxide, and chlorine process
streams, respectively. Although consistent break-
throughs of these haloketones were observed
across the GAC columns, removals remained
above 85% throughout the one-year opera-
tional period.
CH was formed predominantly in the chb-
rine process stream with an average contact
chamber effluent concentration of 2.9 ug/L that
increased to 4.5 ug/L across the sand filter
because of an addftbnal 30 minutes chbrine
contact time. The CH concentratbns in the
effluent of the chbramine contact chamber and
sand column were identical and averaged 0.25
ug/L. CH was detected intermittently in the con-
tact chamber and sand column effluents of the
chlorine dbxide, ozone, and nondisinfected pro-
cess streams at average concentrations rang-
ing from 0.01 to 0.07 ug/L. Post-treatment of the
sand column effluents with free chbrine and
storage for 5 days produced CH concentrations
averaging 79, 55, 45, and 75 ug/L for the
nondisinfected, ozone, chbrine dbxide, and chta-
rine process streams, respectively. Similar treat-
ment with chbramine produced average con-
centratbns of 0.03 to 0.3 ug/L. GAC adsorption
removed all of the CH throughout the project in
all process streams.
The concentratbns of CP in the effluents of
the disinfectbn contact chambers, and the sand
columns were the same, averaging 0.004,0.004,
0.015, 0.038, and 0.43 ug/L for the
nondisinfected, ozone, chbrine dtoxide, chtaram-
ine, and chbrine process streams, respectively.
Chbrination and 5-day storage of the sarid
column effluents produced concentrations aver-
aging 1.3, 7.7, 1.4, and 1.3 jig/L for the
nondisinfected, ozone, chlorine dioxide, and chlo-
rine process streams, respectively. Preozonattan
appears to have produced an increase in CP
precursors. Only slight increases in CP concen-
trations were observed after similar chbramine
treatment and storage of the sand column efflu-
ents with average concentrations of 0.03, 0.04,
0.11, and 0.09 ug/Lforthe nondisinfected, ozone,
chlorine dioxide, and chloramine process
streams, respectively. No consistent break-
through of CP above 0.003 ug/L was observed
across the GAC column of any process through-
out the operational period.
Summary
From these data, one might conclude that
ozonation before sand filtration and
chloraminatbn after sand filtration will solve the
habgenated by-product problem. Although this
is true, the total effect of disinfecttan must be
evaluated. For instance, although ozone shows
promise and may be a viable disinfectbn alter-
native, the increased AOC produced must be
controlled before entering the distributbn sys-
tem. This can be done by biostabilizatbn of the
water during treatment before distribution. Aver-
age AOC concentratbns at Jefferson Parish
are shown in Table 2. After ozonattan, 166 ug/L
of AOC is present. Sand filtratbn provides some
bbstabilization by reducing the AOC to 39 ug/L;
in the GAC effluent AOC is reduced to 4 ug/L
which is comparable to the AOC concentration
(5 ug/L) in the chbrine contact chamber efflu-
ent. Ozonatbn by-products such as aldehydes,
ketones, and acids are also a concern. Al-
though chloramines may be the disinfectant of
choice to provide a distributbn system disinfec-
tion residual, it may not provide the desired
disinfectbn effectiveness in all cases.
The full report was submitted in fulfillment of
Cooperative Agreement NO. CR814033 by the
Jefferson Parish,iLouisLana;Departme,nt of Pub-;
lie Utilities under the sponsorship of the U.S.
Environmental Protectbn Agency
TableZ. Average AOC Concentration at
Jefferson Parish, LA
Treatment
Nondisinfection
O3Contact chamber
O3-Sand
Os-Sand-GAC
CI2Contact chamber
CI2-Sand-GAC
AOC,ua/L
10
166
39
4
5
3
•&V.S. GOVERNMENT PRINTING OFFICE: 1*93 - 750-071/80075
-------
4
Wayne E. Koffskey is with the Jefferson Parish, Louisiana Department of
Water, Jefferson Parish, LA 70121.
Benjamin W. Lyklns, Jr. is the EPA Project Officer (see below).
The complete report, entitled "Disinfection By-Product Formation by Alternative
Disinfectants and Removals by Granular Activated Carbon," (Order No.
PB93-222214AS; Cost: $36.50, subject to change) will be available only
from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Risk Reduction Engineering Laboratory
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
United States
Environmental Protection Agency
Center for Environmental Research Information
Cincinnati, OH 45268
Official Business
Penalty for Private Use
$300
EPA/600/SR-93/136
BULK RATE
POSTAGE & FEES PAID
EPA
PERMIT No. G-35
------- |