OPERATIONAL ASPECTS OF GRANULAR ACTIVATED CARBON ADSORPTION TREATMENT
by
0. Thomas Love, Jr. and James M. Symons
Water Supply Research Division
Municipal Environmental Research Laboratory
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
June 1978
-------
OPERATIONAL ASPECTS OF GRANULAR ACTIVATED CARBON ADSORPTION TREATMENT
by
0. Thomas Love, Jr. and James M. Symons
Respectively Research Sanitary Engineer and Chief, Physical and
Chemical Contaminant Removal Branch, Water Supply Research Division,
U. S. Environmental Protection Agency, Cincinnati, Ohio
Much has been written about the effectiveness of adsorption using
granular activated carbon as a drinking water treatment process for the
removal of organic contaminants. Although the effectiveness of this
process is generally recognized for organic control, questions have
arisen concerning possible disadvantages that might occur during the use
of granular activated carbon. The purpose of this report is to critic-
ally review these questions and summarize what is currently (June, 1978)
known concerning each. Seven questions will be covered:
1) Do excessive bacterial growths occur on granular activated carbon?
2) Are endotoxins created when granular activated carbon is used?
3) Is turbidity effectively removed by granular activated carbon
beds?
4) Do heavy metals leach from virgin granular activated carbon?
5) Do polynuclear aromatic hydrocarbons leach from virgin granular
activated carbon?
6) Do materials desorb or "slough" in a slug from granular activated
carbon?
7) Does thermal reactivation of granular activated carbon cause air
pollution?
-------
-2-
1) Do excessive bacterial growths occur on granular activated carbon?
Controlling bacterial populations (and particularly killing or in-
activating pathogenic microorganisms) is a primary goal of water treatment.
Some concern, therefore, has been expressed about the possibility of
bacteria proliferating within granular activated carbon beds. Activated
carbon removes residual disinfectant while concentrating bacteriological
nutrients. Both factors could contribute to biological growth. This section
of the report summarizes some of the bacteriological data from full scale
and pilot plant studies where granular activated carbon was or is being
used continuously. The intermittent use of activated carbon such as in
home treatment units or the use of an oxidant to stimulate biological
activity in activated carbon, are separate topics and not addressed in this
report.
European Experience
2
Ford in pilot plant studies at Foxcote (United Kingdom), frequently
found higher plate counts (22 C) in the granular activated carbon filtrate
than the sand filtrate. Table I, the summary of a 5-year comparison, shows
60 percent of the samples from the adsorption beds exceeded a bacterial count
of 10/m£ while the sand filter effluent only contained greater than 10
colonies/mi 20 percent of the time. The predominant organism in the
effluent from the activated carbon filter was tentatively identified as
"chlorine-damaged Flavo bacteria". Ford felt 3 days was insufficient
for the development of easily visible colonies at 22 C so he suggested
an incubation period of 7 days for future studies.
-------
Plate Count,
Colonies/m&
0-10
11-100
101-500
>500
No. of Samples
-3-
Table I
Frequency of Occurrence of Agar Plate Counts
3-Days, 22°C
Sand
Effluent
80%
14%
3%
3%
455
Activated Carbon
Effluent
40%
25%
10%
25%
446
Melbourne and Miller reported an interesting finding at Colwick
(United Kingdom). Bacterial growths did not always occur on the activated
carbon beds. For example, when appreciable numbers of bacteria were applied
to the activated carbon, bacterial concentrations were reduced. At times
when prechlorination or caustic soda softening was used, however, few
bacteria remained in the influent to the activated carbon, yet growth of
organisms within the bed was significant.
4
Knoppert and Rook (Rotterdam) studied the effects of frequent back-
washing for controlling bacteriological activity in pilot activated carbon
adsorbers receiving filtered water. One adsorber was backwashed daily and
the other twice a month. The bacterial count (22 C) increased from 10 /m£
4
to 10 /m£ within the first two months service then steadily declined. The
authors concluded the bacterial quality from the adsorber backwashed daily
was slightly better than the effluent from the less frequently washed bed.
The plate count incubated at 37 C never exceeded 10/m£ from either adsorber.
-------
-4-
In the full scale water treatment plant at Vigneux-sur-seine (France),
Richard sampled the effluent from several types of granular activated carbons
and found the agar plate count (24 hrs. at 37 C) slightly higher from the
activated carbon as compared to a sand filter control. Tests for total and
fecal coliforms were always negative.
Schalekamp studied the bacterial content of granular activated carbon
effluents at the Lengg Waterworks in Zurich, Switzerland. He found rather
o 24
high bacterial counts after three days at 20 C, between 10 to 10 /m£, in
effluents from pilot activated carbon columns. Similar data from the
operating waterworks, however, showed bacterial counts between 10 and 40/m£.
These operating filters were backwashed twice per week. Van Lier and co-
workers studied the bacterial count after three days at 22 C in the
effluent from activated carbon filters in Amsterdam. In this study, bacterial
/ Q
counts declined from approximately 10 /m£ in the applied water to 10 /m£
in the effluent from the activated carbon filters.
In the Federal Republic of Germany the drinking water requirement for
general bacterial populations as measured by plate count is 20/m£ for systems
Q
employing treatment with disinfection. Engels reports that at Dusseldorf,
employing granular activated carbon adsorption following filtration, the
effluent consistently meets this requirement even prior to final disinfection.
At the Dohne plant in Mulheim ozonation is employed preceding granular
activated carbon adsorption in an effort to increase the biological populations
within the adsorber thereby enhancing organic removal. Even in this case,
the general bacterial population in the effluent from the granular activated
9
carbon adsorbers was only 3,700/m£.
-------
-5-
United States Experience
During the summer of 1975, Sylvia and coworkers investigated the
extent of bacteriological regrowth in granular activated carbon filters.
Bacterial concentrations (Standard Plate Count at 35 C for 48 hours)
were routinely monitored in the applied water (coagulated, settled, and
chlorinated Merrimack River water in Lawrence, Mass.) and in the effluent
from both a 30-inch and 48-inch deep pilot granular activated carbon filter
(9 min. and 14 min. empty bed contact time, respectively).
Table II summarizes the mean plate count results for the summer and
fall test period.
Table II
Mean (G ) Standard Plate Count/m£
m
(Pilot Plant, Lawrence, MA)
Period Influent Effluent
9 min. EBCT 14 min. EBCT
June - Sept. 60 2800 4000
Sept. - Nov. 15 2000 55
Coliform analysis were also performed and out of 667 determinations, coliforms
were detected in the granular activated carbon filter effluents on 22 occasions,
The highest concentrations of coliforms occurred during a period when the
settling basin was not performing adequately and coliforms were similarly
detected in the effluent from the control (sand) filter.
The columns were backwashed approximately every 48 hours and bacterial
concentrations were examined with respect to time in service. The samples
-------
-6-
collected 1 hour after backwash consistently had higher standard plate counts
than either the 24 or 48 hour samples, but no attempt was made to pinpoint
exactly when the peak occurred.
Sylvia attempted to identify some of the gram negative bacterial species.
tm *
Using the Minitek system (a produce of BBL ) the following organisms were
identified:
Salmonella arizona
Providencia alcalifaciens
Citrobacter freundi
Yersinia enterocolitica
Enterobacter cloacae
Enterobacter agglomerans
Proteus rettgeri
Hansen reported bacteriological regrowth in his activated carbon filters
but felt they could be controlled through an improved backwashing schedule
along with increased chlorination to the applied water. Mr. Hansen, a water
treatment plant superintendent, has several years operational experience
with granular activated carbon and recommends never allowing a filter or
adsorber to stand idle for over an hour. If a granular activated carbon
filter must be out of service for a longer period, it should be diligently
backwashed before resuming operation.
The bacterial quality of untreated and treated Ohio River water from
the EPA Water Supply Research Division's (WSRD) organics removal pilot plant
was routinely monitored over a four-month period by the staff of the WSRD
^Mention of commercial products does not constitute endorsement by EPA
-------
-7-
Microbiological Treatment Branch using the Standard Plate Count (SPC)
test. Samples were also collected intermittently and analyzed for total and
fecal coliforms, however, these indicator organisms seldom survived the
coagulation and settling processes and were never detected in the filter or
adsorber effluents.
The monthly average SPC (expressed as the geometric mean) for the pilot
plant studies (see Table III), in general, show 99 percent reduction in the
bacterial count through the treatment plant. Note no disinfectant was added
anywhere in the treatment process. Although an increase in bacterial
populations was expected _a priori, the SPC of the effluent from the granular
activated carbon adsorber was slightly lower than the SPC in the companion
sample taken from the dual media filter effluent.
Table III
MONTHLY MEAN (G ) STANDARD PLATE COUNT
m
(Pilot Plant Studies - Ohio River Water)
All bacterial counts are No./m£
Time in
Operation
Months
1
2
3
4
Raw
19,600
12,000
7,170
6,680
Settled
1650
1000
790
700
Effluent from Effluent from
Dual Media Granular Activated
Filter Carbon Adsorber
(EBCT = 10 min)
137 63
270 72
80 29
100 37
-------
No attempt was made to measure attached growths. An attempt was made, however,
to isolate and identify the predominant populations in the pilot plant. In
the effluent from the granular activated carbon, five or six different types
of colonies could be recognized and two genera, Flavobacterium and Xanthomonas
were identified from smear plates.
Currently (1978), several ongoing studies include monitoring bacterio-
logical development within granular activated carbon beds. For example, at
Little Falls, New Jersey standard plate counts (2-day, 35 C) and coliform
analyses are being made daily on the effluent from three full-scale adsorbers.
The applied water has a mean (G ) plate count less than 10/m£ and after one
m
month in service, the mean SPC of the granular activated carbon effluent is
2
approximately 10 /m£.
13
In the pilot GAC column study at Kansas City, Missouri O'Connor is
examining attached bacterial growth and the effects of hot-water (co-current
flow) washing. Also, studies are under way in Miami, Florida examining
14
microbial flora in granular activated carbon colums. Parsons examined
the effluent from a pilot granular activated carbon by several bacterial
isolation methods and concluded most bacterial growth went undetected by
Standard Methods. Her work is continuing in the tepid climate of southern
Florida and some of the organisms thus far identified include:
Pseudomonas - like bacteria
Enterobacter agglomerans
Acineto bacter
Alcaligenes faecalis
Moraxella
Flavobacterium
-------
In summary, the purpose of water treatment is to produce a safe (potable)
and palatable product. Bacteria can multiply within granular activated carbon
beds, however, studies have shown the concentration of bacteria depends on:
o length of time the granular activated carbon sits idle
o the concentration (and possibly type) bacteria in the
applied water
o the procedure used to enumerate the bacteria
(22°C versus 35° or 37°C)
o the length of times after backwashing and total time in service
o applied total organic carbon
o depth of bed
o temperature
None of the investigators cited expressed any real concern for the
bacteriological activity observed and several commented on how easily the
populations were controlled by a small amount of post disinfectant. Other
studies have shown a reduced disinfectant demand in granular activated
carbon effluents, further simplifying the task of final disinfection.
Finally, indicator organisms (coliforms) or pathogens have not been shown
to increase through granular activated carbon beds.
-------
-10-
2) Are endotoxins created when granular activated carbon is used?
Endotoxins are lipopolysaccharide-protein complexes produced in the
cell walls of Gram-negative bacteria. Concern, therefore, has been
expressed regarding the possible formation of endotoxins in granular
activated carbon adsorbers because of bacteriological activity.
For a 6-month period in 1977, the U. S. EPA Health Effects Research
Laboratory (HERL) monitored bacterial endotoxin concentrations in untreated
and treated water from the WSRD organics removal pilot plant. These were
companion samples with those collected for Standard Plate Count analyses in
Table III. Using the Limulus lysate bioassay, HERL scientists observed a
marked reduction in pyrogenic activity as a result of chemical coagulation
(and settling) and a slight additional decrease through filtration by
either dual media or granular activated carbon (Table IV). The encouraging
finding was that no increase in pyrogenic activity occurred in the effluent
from the granular activated carbon bed.
Table IV.
MONTHLY MEAN (G ) ENDOTOXIN CONCENTRATIONS,
(Pilot Plant Studies - Ohio River Water)
Time Dual Granular Activated
in Operation, Media Carbon Adsorbent
Months Raw Coagulated/Settled Effl. Effluent
(EBCT = 10 min.)
2
3
4
5
6
7
158
236
205
500
45
35
16
63
36
66
20
11
16
7
41
16
5
11
9
6
11
15
4
11
-------
-11-
An extramural project entitled "Pyrogenic Activity of Carbon-Filtered
Waters" is underway at Texas A&M University. Samples for endotoxin
concentrations are being collected from about a dozen full-scale water
treatment plants utilizing granular activated carbon adsorption. The
empty bed contact time for these adsorption systems range from 4 to 13
minutes and the time in service ranges from slightly over 1 month to 9
years. This study is near completion and thus far no instances have
been found where endotoxin levels increased through the granular activated
carbon. In all cases, the concentrations were either unchanged or reduced
through the activated carbon. The companion standard plate counts have
also been low in this study.
In summary, studies to date (1978) have not shown increased endotoxin
concentration in effluent from granular activated carbon filters, some of
which have been in service for several years.
-------
-12-
3) Is turbidity effectively removed by granular activated carbon beds?
One decision facing a state regulatory agency when a water utility
requests permission to replace the sand in their filters with granular
activated carbon is whether or not some sand should remain in the filter
as a guard against floe penetration. This section of the report summarizes
some of the performance data available from both a full-scale water treat-
ment plant and pilot plant studies. The question of whether granular
activated carbon should be used alone (sand replacement) or only following
filtration (post-filter adsorption) is a separate topic and is not addressed
here.
Specifications for Filter Media
The manner in which particulates are removed during filtration has been
examined and reported by numerous investigators, and traditional design
criteria cover hydraulic loading rates, media size, and filter depth. For
example, until recently in the United States probably 90 to 95 percent of
all gravity filters were designed based on a hydraulic rate of 2 to 2.5
2
gallons per square foot per minute (gpm/ft ) through 24 to 28 inches of
granular medium having an effective size of 0.4 to 0.5 mm. That these
criteria have been proven effective should not preclude modifications or
changes that can be demonstrated to provide equal performance. Evidence
of changes can be seen in some newer water treatment plants designed
within the past 10 years that have a variety of loading rates and multiple-
type media in their filters.
In the AWWA Standard for Filtering Material the effective size for
filter sand is reported as ranging between 0.38 and 0.65 mm and for filter
-------
-13-
anthracite, between 0.45 and 1.60 mm. The only physical constraints other
than cleanliness are:
"Filter sand shall consist of hard, durable grains of siliceous
material less than 2.4 mm in greatest dimension ...."
"Filter anthracite shall consist of hard, durable anthracite
coal particles of various sizes; the hardness shall be not
less than 2.7 on the Moh scale; the specific gravity shall not
be less than 1.4 ...."
Granular activated carbon "shall be composed of hard durable grains;
specific gravity shall be between 1.3 and 1.6." (This is given in
Appendix A of Reference 17 as "information only".)
Table V compares some of the properties of materials used in granular filters
Table V
PHYSICAL PROPERTIES OF GRANULAR FILTERING MEDIA
Medium Granular Activated Carbon
Description
Effective
size(s) mm 0.
Uniformity
coefficient (UC)
Density (gm/cc)
Coal Base
55 to 0.65
5 1.9
1.4
Lignite Base
0.80 to 0.90
1 1.7
1.4
Sand
0.38 to 0.65
1.2 to 1.7
2.65
Anthracite
0.45 to 1.6
<_ 1.8
1.57
-------
-14-
Sand With Granular Activated Carbon in the United States
18
In 1975, Symons attempted to contact all the U. S. water utilities
using granular activated carbon and sent a questionnaire concerning adsorption
practices. One of the questions asked was, "Was sand required to be left
under the granular activated carbon?" Twenty-eight of the 37 facilities that
responded to the question said "Yes", and the amount of sand required averaged
8 inches. Survey results included those from utilities in 22 states.
Comparison of Sand to Granular Activated Carbon as a Filtering Media
Field Study
Studies conducted in 1970 by the Lawrence Experiment Station
examined the parallel operation of a 30-inch deep sand filter (es = = 0.46 mm;
UC1= 1.9 resting on 18 inches of graded gravel) and a 33-inch deep granular
*
activated carbon (Filtrasorb 200) filter (resting on 15 inches of graded
gravel with no sand) within the municipal water treatment plant at Lawrence,
2
Massachusetts. The hydraulic loading was 2.5 gpm/ft and Figure 1 shows the
results for turbidity removal. The turbidity in the effluent from the
granular activated carbon filter was as low or lower than the turbidity in
the sand filter effluent in all samples. Thus, granular activated carbon
was concluded to be as effective as sand for use as a filtration medium
(under the conditions of the experiment), and the State of Massachusetts
allows the use of granular activated carbon (without any sand) as both
a filtration and an adsorption medium.
Calgon Corp., Pittsburgh, Pa. See the Table (coal base) for physical
properties.
-------
-15-
Pilot Plant Studies
Further studies by personnel at the Lawrence Experiment Station
on a small scale showed that the greatest floe buildup in a granular
activated carbon filter adsorber occurs (as one would expect with a sand
19
filter) within the top six inches. Pilot scale studies by the WSRD
showed that granular activated carbon (Filtrasorb 200 and 400) was comparable
to dual media (anthracite/sand) for turbidity removal (see Figures 2 and 3).
Floe penetration, as indicated by headloss at various depths, is much
greater in a dual media filter, see Figure 4. Because it is a single
medium, granular activated carbon provides more surface than depth filtration
see Figure 4, and, consequently, filter runs may be shorter than those for
a dual medial filter, but comparable to those for sand filters.
Head loss and effluent turbidity data were also collected for a 30-inch
pilot column containing lignite base granular activated carbon treating
settled water, (larger effective size, see Table V). These data are compared
to similar data collected from a dual-media filter in Figure 5. Note that
along the "Total Headloss" curve for the lignite-base granular activated
carbon the effluent turbidity was 0.1 NTU after 40 hours and only 0.2 NTU
after nearly 90 hours. Turbidity breakthrough up to 0.7 NTU occurred after
almost 40 hours with the dual-media system. Also some depth filtration
occurred in the lignite-base granular activated carbon system.
In summary, studies in both the laboratory and the field demonstrate
granular activated carbon with an effective size of 0.90 mm or less and
an uniformity coefficient of 1.9 or less is as effective a filtering media
as sand or as an anthracite/sand (dual media) mixture. The experiences in
-------
-16-
(DATA FURNISHED BY THE LAWRENCE EXPERIMENT STATION, LAWRENCE, MASS)
SETTLED WATER
EFFLUENT FROM SAND FILTER
to
at
EFFLUENT FROM 33in GRANULAR
ACTIVATED CARBON FILTER
JAN FEB MAR APR MAY JUN JUL
COMPARISON OF SAND vs. GRANULAR ACTIVATED CARBON
FOR TURBIDITY REMOVAL
1
-------
-17-
12
11 -
10 -
9
(/) Q
t
z
=> 7
9 6
m
cc 5
^ **
4
3
2
1
A
SETTLED WATER
A
/\
v
i /
i i
/ \
v
EFFLUENT FROM DUAL
MEDIA FILTER
\
EFFLUENT FROM 30"
GAC FILTER
(FILTRASORB 400)
v
MARCH APRIL
MAY
JUNE
JULY AUGUST SEPTEMBER
COMPARISON OF DUAL MEDIA VS. GRANULAR
ACTIVATED CARBON FOR TURBIDITY REMOVAL
(1.5"x30" PILOT FILTERS)
FIG 2
-------
-18-
>
I-
2 3
m
cc
UJ
O
<
EC
LU
/ MTri,RACITE-es=l. 2;
J UC=1.7
HI SETTLED I
CZD DUALMEDSA I SAND-es=0 4;
U w—I J
EH GRANULAR ACTIVATED CARBON
(FILTRASORB 200)
WEEKLY COMPARISON OF DUALMEDIA VS.
( 4in x 30in PILOT FILTER )
FIG 3
-------
-19-
8
T 1 r
TOTAL HEADLOSS
HEADLOSS @ 18"
HEADLOSS @ 6"
V)
34
Q
<
UJ
X
G.A.C.
DUAL
MEDIA
10 15
20
25 30 35 40 45 50
HOURS OF FILTER RUN
HEADLOSS WITHIN VARIOUS DEPTHS FOR GRANULAR
ACTIVATED CARBON & DUAL MEDIA (PILOT PLANT)
FIG 4
-------
TOTAL HEADLOSS
HEADLOSS @ 18"
HEADLOSS @ 6"
EFFLUENT TURBIDITY
DURING FILTRATION
o =0.1
0=0.2
• =0.7
LIGNITE
BASE
G.A.C.
o
i
DUAL
MEDIA
10 20 30 40 50 60
HOURS OF FILTER RUN
70
80
90
COMPARISON OF GRANULAR ACTIVATED CARBON
AND DUAL MEDIA FILTER PERFORMANCE
(PILOT PLANT)
FIGURE 5
-------
-21-
the Lawrence, Massachusetts water works and the EPA pilot plants in
Cincinnati, Ohio have also shown that a graded gravel base (or an equivalent
means of providing an uniformly distributed backwash flow) is necessary
for proper filter cleansing; however, any sand layer below 24 or more inches
of granular activated carbon (es = 0.90 mm or less; UC = 1.9 or less) is
redundant as a safeguard to floe penetration, and space could better be
utilized with additional granular activated carbon.
Because adsorption is directly related to contact time, the depth of the
granular activated carbon is directly related to its effectiveness and
t
longevity of performance. Therefore, every inch of sand left in a filter
is one inch less adsorption medium that could be effectively adsorbing
organics. Another consideration is the handling of the granular activated
carbon if frequent reactivations are necessary. The sand-activated
carbon interface might make educting only activated carbon difficult, as
some mixing of media within the filter bed is likely. This would require
an additional sand separation step to avoid the formation of fused siliceous
material in the reactivation furnace.
-------
-22-
4) Do heavy metals leach from virgin granular activated carbon?
20
Directo, Chen, and Miele sampled virgin granular activated carbon
for metal content as part of a study for treating municipal wastes.
Twenty-five grams of Calgon Filtrasorb 300 (a bituminous coal base
granular activated carbon) were refluxed 24 hours with acid (6N HC1), then
rinsed with distilled water. Table VI shows the concentrations of metals
found in that washing.
Recently, (1978) approximately 2500 gm of Calgon Filtrasorb 400 were
flushed repeatedly with distilled water within the EPA Water Supply Research
Division laboratories. Table VII shows the concentration of several para-
meters expressed both as mg/£ (in the recycled water) and mg/kg in the
activated carbon. This also demonstrated under very anomalous and stressed
conditions, some materials can be leached from granular activated carbon.
More importantly, however, are the results obtained when an actual drinking
water was used to flush the adsorbent. Following the distilled water
refluxing, the granular activated carbon was exposed to approximately 1000
gallons of Cincinnati tap water (four days of continuous flow) then the
influent and effluent were resampled. No discernible increases (but some
decreases) in the same parameters were observed (Table VII).
-------
-23-
Table VI.
20
Heavy Metals in Granular Activated Carbon — Acid Refluxing
(mg/kg carbon)
Metal Virgin Filtrasorb 300
Hg 1.92
Se <0.96
Sb <7.20
Sn <48.00
Co <4.80
Tl <24.00
Mo <12.00
Ti 240.00
V 24.00
Be 2.40
Bi <24.00
Zn 1.20
Cr 7.20
Pb 16.80
Ni 12.00
Mn 26.40
Cu <1.20
Ba 50.40
Cd 1.20
As 108.00
Fe 617.00
Al 1536.00
Sr 72.00
Ca 391.00
-------
Parameter
-24-
Table VII
Leaching of Materials from Granular Activated Carbon
a
Distilled Water Reflux
it *
Before
After
mg/kg carbon
Tap Water Flush
* *
Influent Effluent
Hardness
Turbidity
Color
TDS
Cl
SO,
4
N03(N)
Na
Ba
As
Se
F
Ca
Mg
pH
Ag
Cu
Mn
Pb
Fe
Zn
Hg
TOG
ETHM
<1
0.04
1
<0.1
<10
<15
<0.3
<1
<0.2
<0.005
<0.005
<0. 1
<0. 1
<0.01
5.4
<0.03
0.03
<0.03
<0.005
<0.1
<0.02
<0.0005
0.12
0.007
38
0.1
3
48
<10
<15
<0.3
1.3
<0.2
0.015
<0.005
<0.1
10.3
0.73
8.8
<0.03
<0.02
0.04
<0.005
<0.1
<0.02
<0.0005
0.33
<0.0001
851
-
-
1075
<224
<336
<6.2
29.1
<4.5
0.3
<0.1
<2.2
231
16.3
-
<0.7
<0.4
0.9
<0.1
<2.2
<0.4
<0.01
4.7
<0.01
130
0.08
3
210
20
70
1.3
12.6
<0.2
<0.005
<0.005
0.1
34.0
8.28
7.5
<0.03
0.03
<0.03
<0.005
<0.1
<0.02
<0.0005
1.44
0.027
130
0.10
3
209
19
67
1.4
12.5
<0.2
<0.005
<0.005
0.1
32.4
8.91
7.4
<0.03
<0.02
<0.03
<0.005
<0.1
<0.02
<0.0005
0.28
<0.0001
56£ distilled water cycled approximately 100 times through approximately
2500 gm F-400 granular activated carbon
Approximately 1000 gallons of Cincinnati tap water flushed through activated
carbon column following step a.
Except for turbidity, pH, and,color concentrations are mg/£
-------
-25-
In summary, inorganic compounds are associated with fresh granular
activated carbon and under stressed conditions (i.e., acid refluxing
or continuously recycling distilled water) quantifiable amounts of
inorganics can be leached out of the adsorbent. From the available
data, however, the likelihood of this being a problem with natural
water is remote.
-------
-26-
5) Do polycyclic aromatic hydrocarbons leach from virgin grauluar activated
carbon?
Most of the granular activated carbon commercially available for water
treatment is made from bituminous coal or lignite. These base products may
>v
contain polycyclic aromatic hydrocarbons (PAH) and the question is do these
contaminants survive thermal activation (or are they possibly created during
the activation step) only to be released when exposed to water?
21
Andelman and Suess present a good overview of the literature concerning
22
PAH's in water. In 1961, Borneff and Fischer extracted 50 kg of activated
carbon with benzene and were unable to find typical PAH adsorption bands using
23
paper chromatography. More recently (1978) Zoldak used a high pressure
24
liquid chromatography procedure developed by Sorrell, et al. and analyzed
the influent and effluent of an activated carbon column that had been refluxed
with distilled water then exposed to Cincinnati tap water for four days (see
previous section on inorganics). Three of 14 PAHs were detected in
quantifiable concentrations (i.e., greater than 1 ng/£), however these
contaminants were in the influent as well as the effluent. For example, the
total PAH's in the distilled water before refluxing were 24 ng/£ (parts per
trillion) and 22 ng/£ afterwards. For perspective, the World Health
25
Organization (WHO) has a recommended limit of 200 ng/£ for six readily
detectable PAH's (see Table VIII).
In summary, few analytical studies have specifically addressed the
question of PAH's leaching from granular activated carbon, however, from
the available information this is not considered a problem.
*
Also called polynuclear aromatic hydrocarbons, (PNA)
-------
-27-
Table VIII
PAH Concentrations (ng/£) Leached from Fresh
o
Granular Activated Carbon
PAH Distilled Water
Before Reflux After Reflux
phenathrene
anthracene
f luoranthene
pyrene
1-methyl pyrene
chrysene
benzo (a) anthracene
perylene
benzo (a)pyrene
benzo (b ) f luoranthene
benzo (k)f luoranthene
dibenz (ah)anthracene
benzo(ghi) perylene
indeno (1,2, 3-cd)pyrene
ZPAH
6
NF
4
14
<1
NF
NF
NF
NF
NF
NF
NF
NF
NF
24
10
<1
7
5
<1
NF
NF
NF
NF
NF
NF
NF
NF
NF
22
Tap Water
Influent
2
NF
<1
<1
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
2
Tap Water
Effluent
2
NF
<1
<1
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
2
F-400 granular activated carbon
Cincinnati, Ohio tap water
C World Health Organization (WHO) indicator PAHs (ZPAH < 200 ng/£)
NF - not found
-------
-28-
6) Do materials desorb or "slough" in a slug from granular activated carbon?
The effectiveness of adsorption is influenced by the temperature and pH
of the water, but to a greater degree adsorption depends on:
o concentration of adsorbent and adsorbate
o contact or residence time
o competition for available adsorption sites
Concern, therefore, has been expressed about having contaminants concentrated
on the granular activated carbon and then subsequently released in a "slug"
because of changing conditions.
9 f\
Crittenden, in a bisolute adsorption study, showed that during a
column test, periods occurred when the effluent concentration of a weakly
adsorbed material exceeded its influent values in the presence of a more
strongly adsorbed compound. The "over shoots", as they were termed, were
explained by an adsorption equilibrium theory, however, they seldom exceeded
the influent by 20 percent. Some of this difference could be the result of
analytical varibility.
A good example of adsorption equilibrium phenomenon is shown in
Figure 6 (Symons et al.) Carbon tetrachloride in the tap water at
Cincinnati was removed effectively during periods of high influent con-
centrations. As the influent concentration of carbon tetrachloride declined,
the contaminant began to desorb. Although the influent concentrations
exceeded 50 yg/£, desorption concentrations were generally less than
10 yg/£. These high concentrations of carbon tetrachloride in the tap water
had no discernible effects on the trihalomethane concentrations in the
effluent from granular activated carbon columns that had been in service
-------
- 29 -
60
50
40
CARBON
TETRA
CHLORIDE
CONC.
Aig/l
30
20
10
UPFLOW OPERATION
11 MINUTES EBCT
-— INFLUENT (CINCINNATI TAP WATER)
^ GRANULAR ACTIVATED
CARBON COLUMN EFFLUENT
10 15 20 25 30 35 40
TIME IN SERVICE, WEEKS
45 50
FIGURE 6.
PERFORMANCE OF GRANULAR ACTIVATED
CARBON BED FOR CARBON TETRACHLORIDE
REMOVAL (Ref. i)
-------
-30-
for several weeks (see Figure 7). This at least demonstrates that trihalo-
methanes are not desorbed abruptly because of high carbon tetrachloride
concentrations.
27
In a laboratory study on competitive adsorption, Snoeyink noted little
difference in the adsorptive characteristics of anthracene (a polynuclear
aromatic hydrocarbon) both with and without humic acid. He concluded, there-
fore, that PAH will probably not associate with poorly adsorbable humic
substances and result in "leakage of PAH from [activated! carbon beds".
A recent (1978) study at EPA shows the effects of varying water quality
on inorganic desorption. Over a 2-1/2 month period, a lignite base granular
activated carbon (ICI-1030) was intermittently exposed (58 to 90 hours/wk)
to arsenic+ (0.11 mg/£ to 0.23 mg/£), chromium+ (0.03 mg/£ to 0.05 mg/£)
and methyl mercury (0.004 mg/£). These contaminants were "spiked" in a
good quality ground water. Cadmium (0.015 mg/£) and selenium (0.042 mg/£)
were then selected as candidate contaminants in the influent to the activated
carbon. Samples of the effluent analyzed twice daily for one week showed
undetectable concentrations of methyl mercury and chromium. Some arsenic
was detected in the effluent, however, it was also in the influent for some
unexplainable reason (perhaps leaching from the sedimentation tanks).
In summary, adsorption and desorption occurs within granular activated carbon,
The differential migration of contaminants is presently (1978) unpredictable
without on-site pilot scale adsorption studies; however, from the available
data no evidence exists to support the belief that organic or inorganic materials
are concentrated, only to be released in dangerous "slugs". The data show
desorption, if any, is gradual and not abrupt.
-------
u
o
en
01
a
I
o
•H
M
H
1.2
1.0
100
80
60
40
20
5 10 15 20 25 30 35
Time in Service, weeks
Figure 7. Influence of Carbon Tetrachloride Adsorption on Previously
Adsorbed Summation Trihalomethanes
60
u
•H
M
O
0)
H
ti
O
,0
O
-------
-32-
7) Does thermal reactivation of granular activated carbon cause air
pollution?
Two studies were found that discussed the quality of exhaust gases
from the air pollution control equipment associated with thermal reactivation
of granular activated carbon used for treating sewage From the first from
? Q
Kyoto, Japan , the properties and components of exhaust gas from the
reactivation furnace itself, the after-burner and the scrubber during the
reactivation of activated carbon are shown in Table IX. From these data,
the dust and odor strength are shown to be reduced by after-burning, but
sulfurous acid gas and NO increase some. The scrubber is also effective
x
for removing dust and sulfurous acid gas, but did not contribute to the
removal of NO . The increase of odor at the scrubber was caused by the use
x
of secondary effluent for scrubbing. In the second and third reactivation,
the concentrations of odorous substances and other chemicals contained in
the exhaust gas from the reactivation furnace and after-burner were measured.
These results in Table X show the concentration of odorous substances at the
outlet of reactivation furnace is high, but after after-burning, the
concentration was considerably reduced.
20
At Pomona, California flue gases discharged from the top hearth of
the multi-hearth furnace contained both particulate and obnoxious-smelling
substances. These air pollutants were controlled through anxair pollution
control system consisting of a baghouse for particulate removal and an
afterburner, operated in series with the baghouse for odor control. The
afterburner was operated at a temperature range of 719°C (1326°F) to
741 C (1368 F). The baghouse was operated at a temperature ranging from
149°C (300°F) to 163°C (325°F) Although maintaining a high temperature in
-------
-33-
Table IX. Exhaust Gas Component
28
Items
Gas Temperature °C
Dry Gas Volume Nm'/H
Moisture V/V %
Dust g/Nm3
SO, Ppni
NO ppm
NO\ ppm
Power of Odor (PO)
CO, V/V%
0, V/V 7,
CO V/V ™,
Method of Analysis
JIS Z8808
JISZ8808
JIS Z8808
Dust Tube Method
JIS K0103 Solution
Conductivity Method
JIS K0104 Chemical
Radiation Method
JIS K0104
Equilibrium Method
in Salt Water
Orsat Method
Orsat Method
Orsat Method
Outlet ofRe-
generation Furnace
1st
240
69
42.4
1.31
3
62
-74
65
-76
11.9
10.9
3.3
-
2nd
240
83
42.8
3.14
<5
66
-70
66
-70
11.9
10.4
3.3
3.3
3rd
266
88
39.4
3.84
<5
-
40
-60
10.4
9 6
0.9
2.9
Outlet of After
Burning
1st
620
675
22.5
0.16
235
-240
100
-110
105
-115
2.6
11.4
4.0
-
2nd
650
272
21.6
0 27
140
-170
85
-89
88
-94
1.0
7.6
9.3
0
3rd
643
185
35.9
026
Outlet of
Scrubber
1st
40
373
2.4
0.05
350 ' 6-9
-400! 6 V
-
125
-135
0
10.2
6 1
0
113
-123
118
-128
5.3
8 3
8 6
2nd
50
294
12.2*
0.05
<5
107
-111
107
-115
2.3
5.8
12 4
0
3rd
25
229
13.1
0.06
<5
-
125
-130
5.1
9.9
6.2
0.2
Table X. Concentration of Odorous and Other Substances in Exhaust Gas (ppm)
Items
Name
Hyarogen
Cyanide
Carbon
DUulfide
Acetic Acid
Formaldehyde
Acetaldehyde
Methane
Ethane
Ammonia
Tri-methyl
Amine
Hydrogen
Sulfide
Methyl
Mercaptan
Dimethyl
Sulhde
Chemical
Formula
HCN
cs,
CHjCOOH
I1CIIO
CM, -CIIO
Measuring Method
JIS KOI09 Pyridme
Pyrazolon Method
E.A. Notification No. 9
I'PD-GC Method
1 ID-GO Method
JIS K0102 Acetyleace-
tone Method
FID-GC Method
Cm ; FID-GC Method
C,H6 , FID-GC Method
NH,
(CHjh-N
H,S
CHj-SH
(CH,),-S
JIS K0099 Indo
Phenol Test
L'.A. Notification No. 9
E.A. Notification No 9
E.A. Notification No. 9
E.A. Notification No. 9
Threshold
Value
1.0
0.21
0.0021-
0.005
0.041
Outlet of
Furnace
2nd
4 2
120
280
<0.7
<30
noo
54
990
<0.05
<3
<1
<0.5
3rd
5.2
90
370
0.19
19
1.360
35
680
<008
780
<2
<0.1
Outlet of After
Burning
2nd
<0.08
<0 1
40
<05
<30
<10
<10
049
<005
<003
0002
<0.(JOI
3rd
<0 006
<002
<0.2
<0 01
<0 3
35
<0.3
0 25
-------
-34-
th e baghouse to prevent condensation problems is advantageous, precaution had
to be exercised to prevent the temperature from rising to within 10 C (50 F)
to 38°C (100° ) of the critical temperature of the fabric filter. Thus,
to minimize the danger of burning the filter bags, the baghouse inlet was
equipped with a valved side connection for dilution air addition. Under
normal furnace operating conditions, the dilution air inlet valve was
maintained in a closed position. When the baghouse temperature increased
beyond 163 C (325 F), however, which could have been triggered by a dis-
ruption of the activated carbon feed rate or plugging of the quench tank
screen or both, the dilution air valve was manually opened for such a
duration as needed to restore the temperature to about 163 C (325 F).
During reactivation, odors were detected and this was confirmed by the
relatively high odor number of 3 odor units/£ (90 odor units/SCF). Particu-
late emission was also high and averaged 0.30 kg/hr (0.64 Ib/hr). Based
on the particulate emission rate data, the baghouse removed only about 25
percent of the incoming dust load, which was significantly below the design
removal efficiency of 99 percent. The emission data only represented samples
samples collected over a 45 to 60 minute sampling period, however. The total
actual weight of dust collected from the baghouse over the 53-hour reactivation
period was only 14.7 kg (32.5 Ibs.), which represented about 56 percent
of the calculated dust removed.
In Table XI a summary of the emission data from the various components
of the air pollution control system is presented. For purposes of comparison,
the emission data obtained during the third regeneration of the lead con-
tactor of the the tertiary two-stage carbon adsorption system is also included
-------
20
TABLE 11. SUMMARY OF AIR POLLUTION CONTROL SYSTEM PERFORMANCE
Parameters
. Particulate Matter
Concentration, mg/1
Emission Rate, kg/hr
. Oxides of Nitrogen ,(NOV)
A
Concentration, mg/1 dry
Emission Rate, kg/hr
. Oxides of Sulfur (S02)
Concentration, mg/1 S02
Emission Rate, kg/hr
. Hydrocarbons
Concentration, mg/1 C
Emission Rate, kg/hr
. Carbon Monoxide (CO)
Concentration, % vol. dry
. Odor
Odor Units/1
Gas Flow
Temp. ,° C
Rate, I/sec
APCD
Emission
Limit
0.46
0.45
225.00
2000.00
1st Regeneration of
IPC Carbon Column
Baghouse
Inlet Outlet
7.82 3.57
2.11 1 . 44
49.00 120.00
0.016 0.068
Nil
Nil
5530.00 2800.00
0.74 0.56
5.00 1.7
777.00
271.00 121.00
75.00 112.00
Afterburner
Outlet
0.48
0.30
423.00
0.49
729.00
1.17
221.00
0.066
0.47
3.00
665.60
167.00
3rd Regeneration of Two-
Stage Carbon Column (III 3A)
Baghouse
Inlet "Outlet
4.16 1.08
0.98 0.36
40.00
0.012
Nil Nil
740.00 561.00
0.09 0.095
1.36 0.86
706.00
177.80 70.60
66.00 93.00
Afterburner
Outlet
0.017
0.11
180.00
0.18
149.00
0.26
Nil
0.11
0.70
620
177.4
U)
Ui
-------
-36-
in this table. The emission parameters evaluated were significantly
higher in the IPC activated carbon reactivation than in the activated carbon
column III 3A. This observation was expected, considering that the ICP
activated carbon column was subjected to a much heavier load of organic
matter than column III 3A. Because both of these studies related to sewage
treatment the air polluting is more difficult than would be the case by
drinking water.
The following is a quotation from an April 27, 1978 letter from
Dr. McGinnis III of the SHIRCO, Co. concerning this subject:
"As you know, the exhaust emissions which are of concern, fall into
two general catagories: (1) particulate emissions, and (2) gaseous
pollutant emissions, e.g. oxides of sulphur in nitrogen, hydrocarbons,
and certain specific organic compounds. I will address these two catagories
of emissions individually.
With regard to particulate emissions, we have made a number of
measurements during pilot regeneration studies of granular carbon from
industrial waste water treatment applications. These measurements have
indicated typical uncontrolled particle loadings of 0.10 grains per dry
standard cubic foot, or equivalently 0.001 pounds per hour of particulates
per pound per hour of carbon regenerated. These uncontrolled emission
levels approach compliance with existing codes. However, to assure a
considerable margin on actual particulate emissions versus allowable,
we typically supply a venturi scrubber which reduces the particulate
emissions from the stack to 0.02 grains per dry standard cubic foot or
equivalently 0.0002 pounds per hour of particulates per pound per hour
-------
-37-
carbon regenerated. The fundamental reason for these low particulate
emissions lies in the process which we employ. Since there is no stirring
of the carbon, entrainment of fines is minimized and particulate emissions
are quite low.
As to the emissions of gaseous pollutants, we have thus far made
specific measurements only during regenerations of granular carbons used
in industrial waste water treatment. In these cases, the carbon loadings
are much heavier than one would expect in a municipal potable water
application. However, following afterburning and scrubbing, concentrations
of gaseous pollutants have been reduced to acceptable levels.
During our pilot tests of potable water carbons which have been used
for taste and odor removal, we have made no specific measurements. However,
I can say thay qualitiatively there are no obvious odors emitted from the
stack after water scrubbing and without afterburning. This would tend to
indicate that the small quantities of organic compounds present on the
potable water carbons are being incinerated within the furnace itself.
Although the atmosphere in the furnace is tightly controlled to minimize
oxygen concentration, there are small leaks which provide sufficient oxygen
to combust the compounds volatilized from the carbon.
Although data were not available in the literature, personal communications
with the water utilities of Zurich, Switzerland and Dusseldorf, Red. Rep. of
Berman revealed that their reactivation furnaces had been operating for
about 2 years on granular activated carbon used for drinking water treatment
and that the air pollution central devices (after-burners and dust collectors)
were effective. Therefore, air pollution should not be a problem from thermal
reactivation of granular activated carbon.
-------
-38-
SUMMARY
Several questions have frequently been raised over the possible
operational problems that might occur should granular activated carbon
adsorption become wide-spread as an organic control unit process in the
United States. This paper has examined seven of these possible problems
and summarized the data available (much unpublished) on them. These data
show that these potential problems are either non-existent or minor and
should not be cause for preventing the use of this unit process in water
treatment.
-------
- 39 -
ACKNOWLEDGEMENT S
The authors wish to thank Mrs. Patricia Pierson who typed the
manuscript, Mrs. Maura M. Lilly who typed the references, and Mr.
Gordon Robeck who reviewed the report.
-------
- 40 -
REFERENCES
1. Symons, J.M., Interim Treatment Guide for Controlling Organic Contaminants
in Drinking Water Using Granular Activated Carbon. Water Supply Research
Division, MERL, Office of Research and Development, EPA, Cincinnati, Ohio
(January 1978).
2. Ford, D.B., "The Use of Granular Carbon Filtration for Taste and Odor
Control," In: Papers and Proceedings of a Water Research Association
Conference at the University of Reading, United Kingdom, Paper 12
(February 1974).
3. Melbourne, J.D. and Miller, D.G., "The Treatment of River Trent Water
Using Granular Activated Carbon Beds," In: Papers and Proceedings of
a Water Research Association Conference at the University of Reading,
United Kingdom, Paper 4 (February 1974) .
4. Kroppert, P.L. and Rook, J.J., "Treatment of River Rhine Water with
Activated Carbon," In: Papers and Proceedings of a Water Research
Association Conference at the University of Reading, United Kingdom,
Paper 5 (February 1974).
5. Richard Y., "Experience with Activated Carbon in France," In: Papers and
Proceedings of a Water Research Association Conference at the University
of Reading, United Kingdom, Paper 14 (February 1974).
6. Schalekamp, M., "Use of Activated Carbon in the Treatment of Lake Water,"
in Translations of Reports on Special Problems of Water Technology,
Vol. 9 - Adsorption, H. Sontheimer, Ed. EPA Report EPA 600/9-76-030,
(Dec. 1976).
7. Van Lier, W.C., Graveland, A., Rook, J.J. and Schultink, L.J., "Experience
with Pilot Plant Activated Carbon Filters in Dutch Waterworks," In
Translation of Reports on Special Problems of Water Technology,
Vol. 9 - Adsorption, H. Sontheimer, Ed., EPA Report EPA 600/9-76-030,
pp. 160-181 (Dec. 1976).
8. Engels, C., Dlisseldorf Waterworks, Federal Republic of Germany, Private
Communication (1978)-
9. Sontheimer, H., Heilker, E., Jekel, M., Nolte, H., and Vollmer, F.H.,
"The 'Mulheim Process1 - Experience with a New Processes Scheme for
Treating Polluted Surface Water," JAWWA, (In Press).
10. Sylvia, Albert E., et al. "Investigations of Bacterial Growth in
Granular Activated Carbon Filters," contained in Final Report,
"Determining the Organic Content of Drinking Water," EPA 68-03-0267,
Massachusetts Health Research Institute, Lawrence Experiment Station.
Under Review.
-------
- 41 -
11. Hansen, Robert E., "Seven-and-a-Half Years Experience with Granular
Carbon Filters at Mount Clements," Presented at the Michigan Section AWWA,
Lansing, Michigan, Sept. 11, 1975.
12. EPA No. 68-03-2496 "Evaluating the Use of Granular Activated Carbon in
the Treatment of Drinking Water," Passaic Valley Water Commission,
Wendell Inhoffer, Principal Investigator, Clifton, New Jersey,
In Progress (1978).
13. EPA R804433 - "Removal of Trace Organics from Water Using Activated Carbon
aad Polymeric Adsorbants," AWWA Research Foundation, Denver, Colorado.
Mike Taras, Project Manager, In Progress (1978).
14. Parsons, F., "Microbial Flora of Granular Activated Carbon Columns Used
in Water Treatment," Preliminary Report School of Technology, Florida
International University, Tamiami Campus, Miami, Florida (January 1978).
15. Jorgensen, J.H., Lee, J.C., and'Pahren, H.R., "Rapid Detection of Bacterial
Endotoxins in Drinking Water and Renovated Wastewater," Applied and
Environmental Microbiology, 32, No. 3 (Sept. 1976).
16. EPA R-804420010, "Pyrogenic Activity of Carbon-Filtered Water," Texas
A&M University, College Station, Texas. Harold Wolf, Principal Investigator,
In Progress. (1978).
17. AWWA Standard for Filtering Material. AWWA B100-72. Jan. 31, 1972.
American Water Works Association, Denver, Colorado.
18. Symons, James M., "Summary of Granular Activated Carbon Practice Data,"
AWWA Taste and Odor Seminar, 95th Annual Conference, Minneapolis, Minn.,
June 8 - 13, 1975. (Copies available from EPA, WSRD, 26 West St. Clair St.,
Cincinnati, Ohio 45268).
19. Sylvia, A.E., Brancroft, D.A., and Miller, J.D., "Are Two Filters
Necessary?" Water and Sewage Works, May 1975. pp. 66-68.
20. Directo, L.S., Chen, Ching-Lin and Miele, R.P., "Independent Physical-
Chemical Treatment of Raw Sewage," EPA-600/2-77-137 (August 1977)-
21 Andelman, J.B. and Suess, M.J., "Polynuclear Aromatic Hydrocarbons in
the Water Environment," Bulletin of the World Health Organization,
43_, 1970.
22. Borneff, J. and Fischer, R. "Carcinogenic Substances in Water and Soil.
Part V. Investigations on Filter Activated-Carbon." ATch. Hyg. (Berl),
145, 1-11, 1961.
23. Zoldak, J.J., EPA, MERL, WSRD, Cincinnati, Ohio, Private Communication
(May 1978) .
-------
- 42 -
24. Sorrell, R.K., Dressman, R.C. and McFarren, E.F., "High Pressure
Liquid Chromatography for the Measurement of Polynuclear Aromatic
Hydrocarbons in Water," Presented at the Water Quality Technology
Conference, Kansas City, Mo. (Dec. 1977).
25. International Standards for Drinking Water, Third Edition, World Health
Organization, Geneva (1971).
26. Crittenden, J.C., "Mathematic Modeling of Fixed Bed Adsorber Dynamics —
Single Component and Multi Component," Ph.D. Dissertation, University
of Michigan, 1976.
27. Snoeyink, V.L., McCreary, J.J. and Murin, C.J., "Activated Carbon
Adsorption of Trace Organic Compounds," EPA-600/2-77-223 (December 1977).
28. Proceedings 5th United States/Japan Conference on Sewage Treatment
Technology, Tokyo, Japan, pp. 325-344 (April 1977).
4 U.S. GOVERNMENT PRINTING OFFICE: 1978—7 57 -140 / 1310
------- |