United States
Environmental Protection
Agency
Water Engineering
Research Laboratory
Cincinnati OH 45268
Research and Development
EPA-600/S2-84-165 Dec. 1984
Project Summary
RECEIVED
Granular Activated Carbon for
Removing Nontrihalomethane
Organics from Drinking Water
MAR 5 1985
WIRONMENTAL PROTECTION AGENCY
LIBRARY, REGION M
Benjamin W. Lykins, Jr., Edwin E. Geldreich, Jeffery Q. Adams,
John C. Ireland, and Robert M. Clark
A summary is presented of the vari-
ous aspects of using granular activated
carbon (GAG) to adsorb organics other
than trihalomethanes from drinking
water using surface water sources.
Areas of research considered were GAG
performance (service life, water quality,
and reactivation), microbiological con-
cerns, cost of GAG treatment, and GAG
adsorption capacity.
Several large, field-scale research
projects were undertaken to evaluate
the performance of GAG under varying
operating conditions and with different
source waters. Most of this GAG re-
search effort has been completed at
nine locations (Cincinnati, OH; Man-
chester, NH; Jefferson Parish, LA;
Evansville, IN; Miami, FL; Huntington,
WV; Beaver Falls, PA; Passaic, NJ; and
Thornton, CO).
Various carbons produced for or-
ganics removal were evaluated at the
nine locations. From these nine loca-
tions, more than 150 organics were de-
tected. In addition to these specific or-
ganics, surrogate parameters such as
total organic carbon (TOG) and total or-
ganic halide (TOX) were used in many
cases to determine GAG performance.
Most of these compounds were re-
moved by GAG. Low molecular weight
compounds such as 1,2-dichloroethane
were continuously removed for periods
up to 100 days. Higher molecular
weight compounds such as the chlori-
nated herbicide atrazine were removed
through 180 days of operation.
Both the post-sand filter adsorber
and the sand replacement adsorber
were evaluated. The performances of
these two systems were found to be
comparable. Also, pilot systems at
these sites generally predicted the per-
formance of their full-scale counter-
parts.
After exhaustion, GAG was reacti-
vated onsrte at three locations and
offsite at one location. Reactivation
systems included a fluidized bed fur-
nace and a infrared furnace. Several
properties were evaluated to ensure
that proper reactivation had occurred.
Actual cost data associated with the
construction and operation of full scale
research facilities were summarized. In
addition, cost estimates for various
GAG treatment scenarios were de-
veloped to examine the effect of
economies of scale and cost tradeoffs
between alternative GAC treatment
systems.
This Project Summary was de-
veloped by EPA's Water Engineering
Research Laboratory, Cincinnati, OH,
to announce key findings of the re-
search project that is fully documented
in a separate report of the same title
(see Project Report ordering informa-
tion at back).
Introduction
Granular activated carbon has long
been considered a broad-spectrum ad-
sorbent and has been used for many
years to remove tastes and odors from
drinking water. In the past 10 years,
however, some utilities have seriously
considered the use of GAC for remov-
ing trace organics from their drinking
water. With more sophisticated, afford-
able instrumentation rapidly becoming
-------
available to utilities, more organics are
being detected at trace concentrations.
Utility managers as well as the general
public are becoming more concerned
about the health effects of ingesting
these trace concentrations of various
organics over many years.
GAC was investigated under actual
operating conditions to provide adsorp-
tion efficiency data that considered the
mix of compounds in drinking water
sources and the competition for availa-
ble adsorption sites. Sampling schemes
were developed, and analytical meth-
ods were used to detect various organ-
ics, surrogates for organics, microbio-
logical contaminants, and general oper-
ating parameters. Several thousand
pieces of data have been generated
from this effort and stored in the U.S.
Environmental Protection Agency
North Carolina Computing Center at Re-
search Triangle Park, North Carolina.
These data were used to evaluate the
following items:
1. removal or reduction of specific
organics and their surrogates by
GAC,
2. the positive and negative aspects
of GAC reactivation, and
3. the microbiological effects of GAC
use and control measures re-
quired.
The general characteristics of the fol-
lowing items were also summarized
from the research projects and de-
veloped from various treatment
scenarios:
1. GAC adsorption costs for steel-
pressure and concrete gravity con-
tactors,
2. onsite reactivation using fluid bed,
infrared, and multihearth furnaces,
3. offsite reactivation costs,
4. behavior of GAC treatment costs
with variations in system size, and
5. the cost relationships associated
with several GAC treatment alter-
natives.
Presented here is a summary of the
various aspects to consider when using
GAC to adsorb organics from drinking
water using surface water sources.
GAC Performance
Service Life
Various GACs produced for organics
removal were evaluated. Generally, the
coal-based GACs resulted in the best re-
movals over the longest time period. Of
the carbons evaluated, one coal-based
GAC appeared to be the best choice for
removal of low-molecular-weight vol-
atile organics such as 1,2-dichloro-
thane. The higher-molecular-weight or-
ganics appeared to be equally well re-
moved by all of the coal-based GACs
evaluated.
With steady-state total organic car-
bon (TOC) as an evaluation criterion, 3
months would generally be expected to
be the service life of GAC before re-
placement or reactivation. Deviations
from this norm occurred in both direc-
tions, with service lives lasting both
more and less than 3 months. High
TOC/organic precursor waters can gen-
erally expect shorter GAC service times
before exhaustion, whereas other water
sources have experienced more than 3
months of GAC operation before
exhaustion.
Water Quality
Direct GAC adsorption after coagula-
tion and settling appeared to produce
water quality that was comparable with
that from post-sand filtration and GAC
adsorption with the same empty bed
contact time (EBCT). Water utilities
without the capital or available land can
replace their sand with GAC and effec-
tively remove turbidity and organics if
sufficient EBCT is available. However, if
a utility has a choice of GAC filtration/
adsorption or post-sand filtration and
GAC adsorption, the latter is recom-
mended. Sand filtration provides
another barrier for certain microbial
contaminants. Turbidity accumulation
on the GAC will also require more fre-
quent backwashing if prior sand filtra-
tion is not used.
Both GAC pilot columns and pilot
plants were used to predict the perfor-
mance of a full-scale system. EBCT is
important for efficient GAC operation,
but longer times are not necessarily
cost-effective. An optimum EBCT exists
for each situation after which a di-
minishing return can be expected. From
the sites evaluated, 15- to 25-minute
EBCTs were generally optimum, based
on a steady-state TOC removal.
Reactivation
GAC was effectively reactivated. Per-
formance of subsequently reactivated
GAC was comparable if not better than
virgin GAC. Thermal reactivations did
produce volume losses ranging from 5
to 12 percent, depending on the furnace
type. Within the 45.4-kg (100-lb)/hr in-
frared furnace, approximately 5-percent
losses can be expected. Ten to 12 per-
cent losses can be expected from the
227-kg (500-lb)/hr fluidized-bed furnace
An additional 3-percent loss can also be
expected for carbon transport, regard-
less of furnace type.
Concern always exists that organics
in water adsorbed on GAC will be trans-
ferred to the air during reactivation. The
intent during GAC reactivation was to
incinerate the adsorbate, but some or-
ganics were detected in the stack gases.
The detected organics of major con-
cern were tetrachlorodibenzo-p-dioxin
(TCDD) and tetrachlorodibenzo furan
(TCDF). Average total concentrations of
both chemicals were in the low ppt
range (0.14 ppt TCDD and 0.25 ppt
TCDF). A major portion (0.13 ppt TCDD
and 0.21 ppt TCDF) of the levels found
were attached to the particulates in the
stack gas. When a dispersion model
was applied to the stack discharge, no
health hazard was indicated. Dryer off-
gas carbon fines contained higher aver-
age concentrations than the stack gases
(2.2 ppb total TCDD and 1.7 ppb total
TCDF). Disposal of these fines is a con-
cern.
The TCDDs and TCDFs appear to be
produced during reactivation from pre-
cursor compounds, since no detectable
concentrations were found in the spent
carbon fed to the furnace. Some of
these precursors are suspected of being
formed during chlorination, and others
may be in the natural gas used to fire
the furnace. Future research is planned
to investigate systems that do not
chlorinate water before GAC adsorption
and thermal destruction of TCDDs and
TCDFs in a furnace afterburner.
Microbiological Concerns
Start-up operations using GAC in sev-
eral pilot plant and full-scale processes
revealed a surprising problem with ini-
tial coliform and heterotrophic bacterial
contamination. The origin of this con-
tamination may have been the filter
basin (which was converted from sand
filtration operation), the GAC supplied
by the manufacturer, or the water slurry
movement of GAC into the filter com-
partment.
In the first 2 months of operation, in-
fluent quality for new GAC beds
showed no detectable total coliforms in
100-ml samples containing 1.0 to 2.0
mg/L of free chlorine residual. Filter
effluent from each of three different
GAC materials contained 46 to 85 total
coliforms/100 ml during the first week
of operation. The total coliform densi-
ties peaked in about 3 weeks and then
declined below detectable levels by
-------
week 11 to 14. The heterotrophic bac-
terial population rose to density levels
of 104/ml within 4 to 7 weeks and de-
clined below 102/ml by the 11th week.
Reactivation of GAC may cause some
particle size reduction and change in
surface characteristics, but they appear
to have little effect on the ability of
heterotrophic bacteria to find sites for
attachment and colonization. A field
study of a full-scale operation using
GAC filters showed identical microbial
activity on virgin GAC and the same
GAC material after reactivation.
The higher velocity of process water
passing through pressure contactors or
adsorbers does minimize the continued
presence of high densities of hetero-
trophic bacteria in the GAC effluent
after these systems are stabilized. Total
coliform regrowth also appears to be
minimized in these closed devices. Fil-
ter adsorbers are affected by floe or silt
that coats the carbon particles. This
condition contributes to shorter service
life for the GAC and to potential coloni-
zation by coliforms and their sub-
sequent release into the GAC process
effluent. One pilot study with connect-
ing contactors in series indicated that
coliform passage from one contactor to
another can occur, with releases in the
final effluent. However, colonization
was not permanently established in any
of the four contactors in series.
The effects that GAC treatment has
on distribution water quality are largely
unknown from field operations. Pro-
vided the GAC is replaced or reactivated
frequently, the theoretical long-term
benefit would be to improve the micro-
bial quality of water in distribution and
thereby reduce available nutrients (or-
ganics) to support bacterial regrowth in
the pipe sediments and tubercles. How-
ever, investigation of bacterial quality in
a water system using GAC treatment
did on one occasion reveal evidence of
a protected pathway for coliform pas-
sage from carbon effluent, through final
disinfection, and out into the distribu-
tion system. These coliform occur-
rences were based on three replicate
examinations of 1-liter samples. They il-
lustrate how very low densities of coli-
forms go undetected when only 100-ml
sample volumes are analyzed.
Cost of GAC Treatment
Concrete gravity GAC contactor sys-
tems appear to exhibit lower water pro-
duction costs than steel-pressure GAC
contactor systems, given the same
water treatment goal. The concrete sys-
tems have greater economy of scale
since they use a few large contactors as
opposed to many small steel contac-
tors.
Research conducted at the Cincinnati
Water Works indicates, however, that a
full-scale, deep-bed, steel-pressure GAC
contactor system is more economical
than a conventional-depth, sand-re-
placement GAC filter adsorber system,
given the same water treatment goal.
The filter adsorber system with a short-
er EBCT would require more frequent
reactivation and incur significantly
higher costs.
Infrared reactivation is more econom-
ical than fluid bed and multi-hearth
reactivation for small quantities of car-
bon. However, for quantities greater
than about 908,000-kg (2 million lb)/yr,
fluid bed reactivation is the most cost-
effective alternative evaluated. Onsite
reactivation is more economical than
replacement of spent carbon with virgin
GAC for amounts of carbon as small as
136,200-kg (300,000-1 b)/yr using a 22.7-
kg (50-lb)/hr infrared furnace. For larger
quantities of carbon, onsite reactivation
becomes increasingly more cost effec-
tive. If the amounts of carbon are too
small for economical onsite reactiva-
tion, off-site regional reactivation may
be a cost-effective alternative for replac-
ing spent carbon with virgin GAC. Pro-
cess costs for GAC adsorption and reac-
tivation can exhibit significant econo-
mies of scale. Total water production
unit costs (adsorption plus reactivation)
for 113,555-m3/day (3-mgd) and 662,-
375-m3/day (175-mgd) plants ranged,
respectively, from about 13.2 0 to 6.1
0/m3 (500 to 230/1000 gal) for steel-
pressure contactor systems and from
about 10.80 to 4.50/m3 (410 to 170/1000
gal) for the concrete gravity contactor
systems.
GAC Adsorption Capacity Model
The adsorption capacity characteris-
tics of three full scale GAC systems
(Evansville, IN; Manchester, NH; and
Cincinnati, OH) were modeled over 3 to
4 months of operation. The GAC sys-
tems were reactivated several times so
that capacity renewal or reduction
could be investigated in addition to
variation in carbon type and physical
properties. The total GAC loading was
calculated overtime, and a model func-
tion was subsequently fit to the loading
curves by a nonlinear, least squares re-
gression procedure for each operation
phase of a particular GAC system. The
model function is in the following form:
Y(t) = £
1 + Ke-Rt
where:
Y(t) = mg of solute adsorbed/kg car-
bon
t = run time in days
c = limiting value for GAC capacity
(mg solute adsorbed/kg GAC
per day)
K,R = calculated values
Reasonably accurate estimates of
TOC loading capacity can be made in a
full-scale GAC system given iodine
number, molasses number, rate of TOC
applied, EBCT, mesh size, mean particle
diameter, and bed depth for a particular
GAC sample and contactor system. The
modeling procedure requires a straight-
forward calculation of the GAC loading
based on TOC influent and effluent con-
centrations, total mass and volume of
carbon, and flow data in conjunction
with regression-derived values of the
three parameters K, C, and R. The over-
all accuracy of the model equations will
no doubt improve as more physio-
chemical GAC data are collected to in-
crease the total number of model study
cases.
-------
The EPA authors Benjamin W. Lykins, Jr.. Edwin E. Geldreich, Jeffery Q.
A dams. John C. Ireland, and Robert M. Clark are with the Water Engineering
Research Laboratory. Cincinnati, OH 45268.
The complete report, entitled "Granular Activated Carbon for Removing Nontri-
halomethane Organics from Drinking Water. "(Order No. PB 85-120970; Cost:
$22.00. subject to change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield. VA22161
Telephone: 703-487-4650
The EPA authors can be contacted at:
Water Engineering Research Laboratory
U.S. Environmental Protection Agency
Cincinnati, OH 45268
* U S GOVERNMENT PRINTING OFFICE. 1985 — 559-016/7882
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
BULK RATE
POSTAGE & FEES PAID
EPA
PERMIT No. G-35
Official Business
Penalty for Private Use $300
•? f \j I ") N V P 3 c
L' c
------- |