United States
Environmental Protection
Agency
Municipal Environmental Research
Laboratory
Cincinnati OH 45268
Research and Development
EPA-600/S2-81-057 June 1981
Project Summary
Effectiveness of Activated
Carbon for Removal of Toxic
and/or Carcinogenic
Compounds from Water
Supplies
Walter J. Weber, Jr. and Massoud Pirbazari
This research investigated the dy-
namic performance of fixed-bed gran-
ular activated carbon processes for
treating public water supplies. The
adsorption of representative toxic
and/or carcinogenic trace compounds
of man-related origin was evaluated.
Adsorption of humic substances—
the predominant class of natural organic
matter and the trihalomethane (THM)
precursor—was also studied.
Adsorption efficiency tests and cost
effectiveness evaluations were per-
formed initially for a number of com-
mercial activated carbons. A bitumi-
nous-base carbon was then selected
for further use throughout the research.
Equilibrium, rate and continuous
long-term adsorption studies were
conducted for humic acids at raw
water background concentrations and
for carbon tetrachloride, benzene, two
commercial mixtures of PCB's, diel-
drin, and p-dichlorobenzene at repre-
sentative trace concentration levels.
Experimental breakthrough profiles
were developed for each of the trace
compounds alone and in the presence
of background humic acid solutions to
investigate competitive and/or chro-
matographic effects. The results indi-
cate that trace compound adsorption
is generally affected adversely by
background organic matter.
The adsorption capacity of activated
carbon for humic substances was
found to be markedly enhanced by the
presence and concentration of several
common constituents of water supplies,
including Mg2* and Ca2+ ions. There
was evidence from the work that
enhanced removal of humic sub-
stances might also provide for more
effective removal of associated organ-
ic molecules—such as PCB's and diel-
drin—in water treatment operations.
The Michigan Adsorption Design
and Applications Model (MADAM)
was generally able to simulate and
predict the performance of fixed-bed
adsorbers to remove the compounds
investigated in the absence of the
humic acids. With the strictly adsorp-
tion version of MADAM, reliable pre-
dictions were difficult because of the
prolific biological growth in the ad-
sorbers that received the humic acids.
For such cases, biological degradation
dynamics must be incorporated in the
model. The single-solute version of
MADAM was able, however, to simu-
late/predict independent pilot data
provided by the EPA about the adsorp-
tion of 1,2-dichloroethane in the
presence of naturally occurring back-
ground organics in Ohio River water.
This Project Summary was developed
by EPA's Municipal Environmental
-------
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
The widespread occurrence of unde-
sirable trace organic compounds in
public water sources is a matter of
record. Urban and suburban develop-
ments along major water courses have
affected both implicit and explicit multiple
reuse of water. In many areas of the
country, the need for successive reuse
of water—either planned or covert—has
led to accumulation of persistent organic
compounds of industrial and/or agricul-
tural origin in raw water sources and
supplies, frequently to concentration
levels that pose threats to human health
and well being. Further, a variety of
natural organic substances can be
transformed to potentially harmful
THM's in the course of water treatment
with chlorine.
Treatment with granular activated
carbon has evolved as a technically and
economically viable technology for
removing many of the compounds of
concern from public water supplies. The
efficiency and cost effectiveness of the
design and operation of this process,
however, depend markedly on optimum
use of the adsorption capacity of the
carbon, consistent with treatment ob-
jectives.
To accomplish efficiency and cost
effectiveness, appropriate data regard-
ing system dynamics, preferably from
relatively simple laboratory experiments
must be developed, and the body of
these data must be synthesized in such
a manner that they can be incorporated
in actual plant design and operation.
Mathematical modeling and simulation
of the dynamics of granular carbon
systems can be a particularly effective
means for synthesis of data regarding
adsorption dynamics, and resultant
models can be employed to good advan-
tage in predicting system performance
under a variety of conditions. Models of
this type can thus function as vehicles
for design, control, and operation of
adsorption systems.
The research described here focused
on developing rate and equilibrium data
on activated carbon adsorption of several
representative compounds of natural,
industrial, and agricultural origin from
water: specifically, humic acids, carbon
tetrachloride, benzene, polychlorinated
biphenyls, dieldrin, and p-dichloroben-
zene. The Michigan Adsorption Design
and Applications Model (MADAM),
which has demonstrated potential for
use in the design and operation of
granular carbon systems for wastewater
treatment, was used for simulation and
prediction of the dynamic performance
of carbon beds to remove trace quanti-
ties of these compounds under condi-
tions approximating those encountered
in water treatment practice. The MADAM
algorithms incorporate coupled mass
transfer resistances (external film and
internal surface diffusion) and accom-
modate nonlinear isotherm expressions.
Model predictions were predicated for
the most part on bench-scale equilibrium
and rate studies. Column experiments
were then conducted to experimentally
verify the model predictions and simula-
tions.
Experimental Approach
Adsorption Equilibria
Experimental Methodology
Adsorption equilibrium experiments
were conducted for each compound
using conventional static bottle-point
techniques, i.e., placing different amounts
of activated carbon in each of a series of
air-tight glass reactors. The reactors
were then filled with background solu-
tions spiked with the compound of
interest. Depending on the compound's
characteristics and the objective of a
specific experiment, the background
solution consisted of organic-free water
(OFW), deionized-distilled water (DDW),
tap water, or a solution of a 5 mg/l of
humic acid prepared in one of the
preceding background waters. The
reactors were then agitated at room
temperature to achieve equilibrium. The
residual concentrations were determined
using appropriate analytical techniques.
Isotherm Parameter Evaluations
Several theoretical and empirical
equations, including the Freundlich,
Langmuir, and three-parameter iso-
therms, were investigated for mathe-
matical description and quantification
of the adsorption equilibrium data. The
data were best described by the Freund-
lich and the three-parameter isotherms,
and the former was chosen for subse-
quent use because of its simpler form,
which facilitated parameter estimation.
Adsorption Rates
Experimental Methodology
Completely-mixed batch (CMB) re-
actor adsorption rate experiments with
the trace compounds were performed in
carefully sealed 2.6-L glass reactors;
3.5-L capacity open-top glass reactors
were used for the humic acid tests.
Weighed and prewetted quantities of
granular activated carbon were added to
appropriate experimental solutions in
the vapor-phase-free reactors. The
carbon was dispersed by a motor-driven
glass stirrer, and 5-ml samples were
withdrawn at fixed time intervals. A
displacement plunger built into the
reactors eliminated introduction of
headspace by sample volume displace-
ment.
Rate Parameter Evaluations
The MADAM model was used to
estimate mass transfer coefficients
simultaneously from the CMB rate data.
A two-parameter search routine was
used to determine values for the external
film transfer coefficient, kt, and the
intraparticle surface diffusion coeffi-
cient, D8. A CMB rate experiment that
employed a large amount of carbon to
nearly deplete the solute within the first
few minutes of the experiment was
used to estimate a close control value
for k(. These coefficients were then used
in the fixed-bed version of MADAM to
predict breakthrough profiles. Sensitivi-
ty analyses were also performed to
characterize the rate controlling mecha-
nism.
Fixed-Bed Adsorbers
Experimental Methodology
Experimental fixed-bed column studies
were designed and executed to investi-
gate dynamic adsorption behavior and
to obtain data about adsorption break-
through characteristics for prediction/
simulation modeling of each compound.
The studies used mini-column, or large
column experiments, or both. For mini-
column, 1 -ft. glass adsorbers with an
internal diameter (ID) of 1 cm were
employed; for large columns, 6-ft. glass
columns with a 3-cm ID. Feed tanks
were constructed of glass, or stainless
steel, or both. All transfer tubing was
stainless steel, or Teflon, or both.
Influent and effluent samples were
collected manually. Extreme care was
taken during sample transfer to minimize
losses by volatilization. When neces-
-------
sary, the feed solution was dosed with 2
mg/L silver to inhibit biological growth.
Humic Acids
To provide a reasonably reproducible
reference point and to ensure experi-
mental consistency, Aldrich Chemical
technical grade dried humic acid was
used. Different techniques for analysis
were investigated, and ultraviolet spec-
trophotometry at a wavelength of 250
nm was selected for routine use.
Equilibria
The data indicated (1) capacity was
enhanced with decreased carbon parti-
cle size; (2) the extent of adsorption of
the "humic acids" fraction was larger
than that of "fulvic acid" fraction; (3) the
amount of humic acid adsorbed generally
increased with decreasing pH; (4) initial
concentration hasa significant effect on
the position of the adsorption isotherm;
(5) carbon of a given particle size simply
sieved from a commercial lot demon-
strated a larger capacity for humic acids
than carbon of the same particle size
prepared by crushing larger particles
and sieving to the given size; and, (6)
several ionic species—including Ca2+,
Mg2+, and OCI"—substantially enhance
the adsorption capacity of humic acids
(Figure 1).
Rates
Model simulation analyses performed
on the CMB rate data for adsorption of
humic acids from DDW to estimate
values for the external film transfer and
internal surface diffusion coefficients
yielded kf = 1.44 x 10"4 cm/sec and Ds =
1.64 x 1CT10 cm2/sec, respectively.
A similar procedure used to analyze
data for adsorption from tap water
yielded k( = 7.04 x 10~4 cm/sec and Ds =
1.36 x 10~10 cmVsec. As noted, the
background solution has a substantial
effect on the'adsorption rate of humic
acid on carbon.
Fixed-Bed Adsorbers
The influence of carbon particle size,
hydraulic loading, and background
solution on the dynamic adsorption
behavior of humic acids was systemati-
cally characterized in these studies.
Typical breakthrough profiles for ad-
sorption of humic acids from tap water
and DDW indicate differences in the
dynamic adsorption characteristics in
the two different background solutions
are pronounced (Figure 2).
DDW + 10.0 mg/l Mgz*
DDW + 3.5 mg/l OCF
DDW +10.0 mg/l Caz+
DDW, control
V: DDW + 0.05 mg/l
I
•3
1 1 1 I 1 I I
•3 /I C C ~> 0 Q
1 1x10'
Equilibrium Solution Concentration—Humic Acid !mg/l)
Figure 1. Adsorption isotherms for humic acids; effect of selected inorganic
species on adsorption capacity. Co = 4.9 mg/l.
0.00
Figure 2.
40.00
80.00
120.00
Time (hours)
160.00
200.00
Column breakthrough profiles for humic acids in different background
solution. C0 = 5ppm;flow= 1.94 gpm/ff; 250 grams of 16.20 US sieve
size carbon.
Simulation-prediction analyses were
conducted for humic acids in DDW and
tap water for different bed lengths.
Extensive biological growth was noted,
and it was concluded that the strictly
physicochemical version of MADAM
could not be used as a truly predictive
model under these circumstances.
Rather, MADAM was used in simulation
mode with the values for the film trans-
fer coefficient, k(, and the intraparticle
diffusion coefficient, D8, being deter-
mined from the column data. Figure 3
depicts experimental data and the simu-
lated MADAM breakthrough profile for
adsorption of humic acid from DDW.
Mechanism of Removal
A notable characteristic of the column
data for humic acids was a relatively
rapid approach of the breakthrough
profile to an apparent plateau. The
-------
precise level of the plateau, however,
varied with system particulars and
operating conditions. This was true for
all adsorption column experiments
conducted with humic acids for this
investigation and has likewise been
observed in other studies. Various
mechanisms have been proposed for
the near steady-state removal observed
at the breakthrough plateaus; the two
most prominent suggestions being
catalytic oxidation and biological degra-
dation, although secondary micropore
diffusion is another possibility. Static
adsorption equilibrium measurements
were designed to gain some insight to
the magnitude of bio-oxidation and/or
catalytic oxidation effects. It was con-
cluded that a combination of biological
degradation and hindered diffusion of
the large rrtacromolecules within the
carbon micropores was responsible for
the observed plateau development.
Carbon Tetrachloride
Several different sample collection
and concentration techniques—dynamic
headspace, static headspace, liquid-
liquid extraction, and aqueous injec-
tion—were evaluated. All demonstrated
some degree of compatibility with the
methodologies used to evaluate the
adsorption properties and characteris-
tics of CCI4. The liquid-liquid extraction
technique followed by electron capture
gas chroma tog raphic analysis was
selected as most appropriate for this
work.
Equilibria
These data in'dicated that: 1) pH has
no significant effect on the equilibrium
adsorption characteristics of CCU; 2)
there is no dependency of CCU equili-
brium capacity on carbon particle size;
3) initial concentration has no signifi-
cant effect on the position of the isotherm
for CCU; and, 5) background organic
substances such as humic acids ad-
versely affect the capacity of activated
carbon for adsorption of CCU.
Rates
A simulation-modeling analysis was
performed on the CMS data for adsorp-
tion of CCU from OFW to evaluate
external and internal mass transfer
coefficient values of kf = 4.07 x 10~3
cm/sec and D8= 3.02 x 10~10 cm2/sec,
respectively. Values in the presence of 5
mg/Lof humicacid werekf = 3.78x10~3
cm/sec and D. = 1.86 x 10~10 cmVsec.
The intraparticle diffusion coefficient
for the system containing humic acids
was significantly lower than that for the
organic-free system, suggesting that
humic acid molecules retard the diffu-
sion of CCU molecules within the carbon
pore structure.
Fixed-Bed Adsorbers
These studies were designed and
executed in two phases. Phase I consist-
ed of mini-column tests to obtain pre-
liminary data for predictive model
(MADAM) calibration. Phase II consisted
of breakthrough runs with large fixed
bed adsorbers to develop data for mode
verification. Examples of mini-columi
and large column data and predictei
breakthrough profiles are illustrated ii
Figures 4 and 5, respectively.
Desorption and Competitive
Adsorption
Equilibrium adsorption-desorptior
experiments for CCU using the bottle
point technique indicated no significan
equilibrium hysterisis. Rates of desorp
**!
MADAM simulation T
0.00 0.50 1.00
1.50 2.00
Throughput
2.50 3.00 3.50 4.00
Figure 3. Effluent and influent concentration profiles for humic acids in DDW
Co — 5 ppm; empty bed contact time (EBCT) —15.6 min.
ly
0.00
0.20
0.40
0.60
Throughput
0.80
7.00
Figure 4.
Influent and effluent concentration-history profiles for CCU in fixed-bed
mini-columns. 2 Grams of 50/60 US sieve size carbon; C,n = 1 77fjg/l;
EBCT = 1. 1 min.
-------
tion of CCU from carbon were also
examined. The rate of approach to
equilibrium in the desorption experi-
ments was faster than that in the
adsorption tests. Desorption studies
with fixed-bed adsorber columns dem-
onstrated that a significant amount of
CCU was desorbed.
Fixed-bed adsorption studies with
CCU in the presence of 5 mg/L humic
acid were hampered by extensive bio-
logical growth, and it was impossible to
reach any quantitative conclusions
regarding the dynamic adsorption be-
havior of CCU in the presence of humic
acids. There was a suggestion, however,
that competitive adsorption plays a
significant role in the dynamics of
removal of CCU in typical adsorber
applications.
Benzene
As with CCU, several different sample
collection and concentration techniques
were evaluated for analysis of benzene
in the /ug/L range, and a dynamic
headspace/gaschromatographic analy-
sis technique was selected.
Equilibria
Studies of adsorption equilibria for
benzene in OFW and in a 5 mg/Lhumic
acid background solution demonstrated
no significant differences in extent of
adsorption between the two systems.
Rates
A simulation-modeling analysis of
CMB rate data for benzene in OFW
background solution gave values of kt
7.48 x 10~3 cm/sec and Ds 1.3 x 10~9
cmVsec for the mass transfer coeffi-
cients. A similar analysis of CMB rate
data for benzene in the presence of 5
mg/L humic acid yielded values of k/
6.38 x 1CT3 cm/sec and Ds 1.37 x 10~9
cmVsec. The humic acid system dem-
onstrated a slightly lower film transfer
coefficient, k(, than the OFW system,
although no significant change in sur-
face diffusion coefficient, Ds, was ob-
served.
Fixed-Bed Adsorbers
As with CCU, mini and large fixed-bed
adsorber studies were conducted for
benzene. Good agreement was demon-
strated between the experimental data
and MADAM's predicted profiles.
Fixed-bed adsorber experiments in
the presence of 5 mg/L humic acid were
hampered by biological growth, rendering
quantitative characterization of the
dynamic behavior of benzene impossible
in this situation.
Desorption and
Chromatographic Effects
Equilibrium adsorption-desorption
tests employing the bottle-point technique
demonstrated no equilibrium hystere-
sis. In fixed-bed desorption studies
using 3-cm ID -adsorber columns, a
substantial quantity of benzene was
desorbed.
To substantiate competitive adsorp-
tion, virgin and "humic acid pre-satu-
rated" carbon columns were fed with
benzene in OFW background solution.
Figure 6 demonstrates the correspond-
ing influent and effluent profiles, indi-
cating a substantial reduction in the
capacity of presaturated carbon for
adsorption of benzene.
7.00
1.20
0.60
Throughput
Influent and effluent concentration profiles for CCU in large fixed-bed
adsorbers. 250 Grams of 16/20 US sieve size carbon; Cm — 570 ug/l;
EBCT = 6.0 min
Influent profile
O- Effluent profile
(virgin carbon)
*~ Effluent profile
("humic acids-
saturated carbon)
0.00
2.00 4.00
6.00 g.OO 70.00 72.00 74.00
Time (days)
Figure 6. Effluent and influent profiles for adsorption of benzene on virgin carbon
and presaturated carbon. 1 Gram of 50/60 carbon; flow rate= 10 ml/
min; column diameter =7.2 cm.
-------
Polychlorinated Biphenyls
Two commercial polychlorinated bi-
phenyl (PCB) mixtures were selected;
namely, Aroclor 1254 (A-1254)* and
Aroclor 1016 (A-1016). Analysis was by
liquid-liquid extraction and gas chroma-
tography. Depending on the quantity of
PCB's in a given sample, either a vial
extraction technique or a multiple
extraction technique was employed.
Because of the extremely low levels of
PCB's found in water supplies, adsorp-
tion methodologies for these substances
required recognition of a number of
factors that markedly influence quanti-
fication of their adsorption parameters:
sorption on reactor walls, volatilization,
biological growth, interferences, and
potential adsorption of carrier solvents.
Equilibria
A comparison of equilibrium data for
adsorption of A-1254 and A-1016 from
OFW indicated that A-1016 is adsorbed
to a greater extent than A-1254, contrary
to the general observation that adsorp-
tion is inversely related to aqueous
solubility. The results of equilibrium
studies for adsorption from OFW and
humic acid background solutions dem-
onstrated that PCB's are adsorbed to a
lesser extent in the presence of humic
acids. In the case of PCB's, possibly both
the carbon and the humic acid molecules
compete for adsorption of these sub-
stances. This phenomenon advances
the possibility of enhanced PCB removal
via mechanisms that enhance the adsorp-
tion of humic acids.
Rates
A simulation-modeling analysis was
performed for adsorption of A-1016 and
A-1254 from OFW and humic acid
solution to evaluate the mass transfer
coefficients, k( and Da. Because the form
of the MADAM model used was the
single-solution form, the results, as
anticipated, demonstrated significant
disparity between experimental data
and fitted rate profiles. Because each
PCB mixture is composed of a large
number of different isomers, deviations
from single-solute behavior are ex-
pected.
To investigate differential adsorption
of PCB isomers, CMB reactor rate
studies were performed for A-1254 in
•Mention of trade names or commercial products
does not constitute endorsement or recommenda-
tion for use by the U.S. Environmental Protection
Agency.
OFW. The results indicated that lower
isomers are adsorbed more extensively.
Similar CMB studies with A-1254 in
humic acid solution also exhibited
differential adsorption of isomers. It was
further observed that the background
solution has a marked effect on the
adsorption of isomers.
Fixed-Bed Adsorbers
For the mini fixed-bed adsorbers, no
breakthrough of A-1016 was observed
within 244 days of operation, a duration
well beyond the theoretical carbon
capacity. Several plausible explanations
may be advanced, including biodegra-
dation of PCB's and filtration of PCB
aggregates by the column. Figure 7
depicts the influent and effluent con-
centration profiles for A-1016. Figure 7
also illustrates the results of dynamic
desorption studies, indicating that no
PCB was eluted or desorbed from the
column.
Competitive Adsorption
Two mini-adsorber columns, one
containing carbon saturated with humic
acids and the other virgin carbon, were
fed with a solution of A-1254 in OFW.
The effluent concentration (Figure 8)
indicated no significant reduction in the
capacity of the presaturated carbon.
This may be highly significant with
respect to the ability of carbon, which is
saturated with background organics, to
continue to remove trace quantities of
materials with strong adsorption poten
tial, such as PCB's.
Dieldrin
Determination of residual concentra-
tion of dieldrin in aqueous solution was
accomplished in a manner similar tc
that discussed for PCB's.
Equilibria
Equilibrium experiments were carriec
out in OFW and 5 mg/L humic acic
solutions. The results indicated lowei
dieldrin adsorption capacities in the
presence of humic acids. The difference
may be due to a preferential associatior
of dieldrin and humic species in solution
resulting in less dieldrin uptake by the
carbon.
Rates
Values of k(=1.75 x10~3andD,= 4.43
x 10~9 cm2/sec were obtained for ad-
sorption from OFW solutions; the cor-
responding values measured for ad-
sorption from 5 mg/L humic acid back-
ground solutions were kf = 1.29 x 10"s
cm/sec and Ds = 1.33 x 10~10 cm2/sec.
For systemscontaining humic acid, both
the external film transfer coefficient, k(,
and the intraparticle diffusion coeffi-
cient, Ds, are significantly lower than
those for OFW systems. Both mass
transport steps associated with the
adsorption of dieldrin are apparently
retarded by humic acids.
A-1O16
OFW
.o
c
-------
§
c>
o
o
o
§
^ o
§> ^
-§ o
;= o
CO O
S "*
0
S o
,°. o
§
ci
A-1254 in OFW
OFW
+
5 mg/l
humic acids
• Virgin Carbon
D Pre-saturated Carbon
Influent
concentration
profile
0.00
Figure 8.
4-
Effluent concentration data
3 » g88.\a? ft
10.00
40.00
50.00
20.00 30.00
Time (days)
Influent and effluent concentration data for A-1254 in fixed-bed
adsorbers. Flow = 10 ml/min; 1 gram of 50/60 US sieve size carbon.
Fixed-Bed Adsorbers
Large fixed-bed adsorber experi-
ments were designed and executed to
evaluate the effectiveness of carbon to
remove dieldrin in a "close-to-real"
situation. After 2 to 3 weeks of opera-
tion, profuse biological growth within
the column became evident. Influent
and effluent concentration data indicate
the carbon bed was able to reduce
influent levels of dieldrin as great as 30
A/g/L to effluent levels lower than the
detection limit (~ 0.01 /^g/LXFigure 9).
As evident, no breakthrough was reached
after 220 days of operation.
Mini-column adsorber experiments
for dieldrin were designed and run to
obtain data about adsorption break-
through characteristics for prediction/
simulation modeling investigations. No
breakthrough was observed after 206
days of operation, a duration well beyond
the isotherm capacity. Explanations
similar to those discussed for PCB's may
be advanced.
Desorption and Competitive
Adsorption
Dynamic desorption studies were
conducted in both the large and mini
fixed-bed adsorbers, and dieldrin was
observed to be eluted or desorbed from
the column.
p-Dich(orobenzene
Determination of p-dichlorobenzene
(PDB) in solution phase was by a vial
extraction technique and gas chroma-
tographic analysis. Extraction efficiency
was estimated to be 98%.
Equilibria
No significant differences in the
extent of adsorption of PDB from OFW
and 5 mg/L humic acid background
solutions were observed.
Dieldrin in DDW
Rates
Mass transfer coefficients for adsorp-
tion of PDB from OFW were determined
tobekfx4.3x10~3cm/secand D, = 8.0
x 10~8 cmVsec. Values for adsorption
from 5 mg/L humic acid solution were k(
= 5.3 x10'3 cm/sec and D,= 8.0 x10"8
cmVsec. It is reasonable to conclude
that humic acids have essentially no
effect on rates of adsorption of PDB.
Fixed-Bed Adsorbers
Relatively good agreement was dem-
onstrated between experimental ad-
sorber data and model predictions for
PDB (Figure 10).
As with dieldrin, mini expanded-bed
systems were employed to examine the
dynamic characteristics of PDB in the
presence of humic acids. The results
once more confirmed little if any adverse
effect of the humic acids on PDB adsorp-
tion mass transfer rates.
Case Study: Pilot-Plant
Investigations of the Adsorption
of 1,2-Dichloroethane
To test the MADAM model for gener-
ality, simulation-prediction analyses
were performed on independent pilot-
plant results obtained when 1,2-dichlo-
roethane (DCE) was removed from
settled Ohio River water (SRW). The
Municipal Environmental Research
Laboratory (MERL), USEPA, Cincinnati,
Ohio, conducted the pilot-plant studies
Dieldrin in tap water •§ §
I
0.00 40.00 80.00
200.00 240.00
Figure 9.
120.00 160.00
Time (days)
Influent and effluent concentration data for dieldrin in large fixed-bed
adsorbers. Flow = 40 ml/min; 250 grams of 16/20 carbon- EBCT =
min.
-------
10
.0
c
.o
c
0)
I
I
0.00
0.20
0.40
0.80
1.00
1.20
Figure 10.
0.60
Throughput
Effluent and influent concentration profiles for PDB in OFW-mini
fixed-bed adsorbers. Cm= 185 fig/1; 1 gram of 50/60 carbon; flow rate
= 15 ml/min; EBCT = 0.17 mm.
and provided the experimental batch
and column data used in these analyses.
The single-solute version of the
MADAM model was used for three
different sets of experimental conditions
to simulate and predict the adsorption
dynamics of DCE in fixed-bed adsorbers
in the presence of naturally occurring
background organics. The agreement
between MADAM'S predicted break-
through profiles and the EPA pilot-plant
verification data was generally very
good and was only slightly affected by
apparent biodegradation of DCE over
the course of the pilot studies. Figure 11
presents experimental data and predicted
breakthrough profiles for a typical fixed-
bed pilot-plant adsorber.
Conclusions
• Adsorption capacities of activated
carbon for background humic acids
are markedly enhanced by the
presence of several common ionic
constituents of water, including
Ca2+, Mg2+, and OCI".
• The effectiveness of activated carbon
to remove PCB's and dieldrin from
water is adversely affected by the
association of these compounds
with background humic substances.
Introducing certain inorganic spec-
ies such as Mg2+and Ca2+to remove
humic substances more effectively
might also remove associated or-
ganic molecules-such as PCB's
and dieldrin-more effectively.
• The Michigan Adsorption Design
and Applications Model (MADAM)
was generally able to simulate and
predict the performance of fixed-
bed adsorbers to remove carbon
tetrachloride, benzene, andp-
dichlorobenzene in organic-free-
water background solutions. Be-
cause it was not possible to obtain
experimental breakthrough profiles
for PCB's and dieldrin (no break-
through occurred in 7 to 8 months
of operation), no attempt was m
to model the behavior of thi
substances in fixed-bed adsor
systems.
Fixed-bed adsorber studies in WE
containing background humic ac
were hindered by extensive biol
ical growth and resulting headlc
Because of these operational pr
lems, the strictly adsorption vers
of MADAM could not be adequal
tested to forecast the dynar
performance of adsorber be
When adsorbers that exhibit
have a tendency toward prol
biological growth are modeli
using a version of MADAM t
incorporates biological degradat
terms would appear to be desirat
The single-solute, strictly adso
tion version of the MADAM mod
was, however, used successfuly
simulate-predict independent USE
pilot-plant data on the adsorpti
of DCE in the presence of natura
occurring background organics
Ohio River water.
The results of the long-term colur
studies with PCB's and dieldi
suggest the possibility of develc
ing specific microorganisms th
may be capable of biodegradi
these relatively resistant compoun
in activated carbon adsorbers.
lo
o*
c
o
c
-------
The full report was submitted in
fulfillment of Grant No. R804369bythe
University of Michigan under the spon-
sorship of the U.S. Environmental Pro-
tection Agency.
Walter J. Weber, Jr. and Massoud Pirbazari are with the University of Michigan
Ann Arbor, Ml 48109.
Alan A. Stevens is the EPA Project Officer (see below).
The complete report, entitled "Effectiveness of Activated Carbon for Removal of
Toxic and/or Carcinogenic Compounds from Water Supplies," (Order No.
PB 81-187 197; Cost: $27.50, 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:
Municipal Environmental Research Laboratory
U.S. Environmental Protection Agency
Cincinnati. OH 45268
* U,S GOVERNMENT PRINTING OFFICE 1981-757-012/7145
-------
3J
1
3)
z
s
C/J
CD
3
O
o
5'
o
5
0)
O
I
N)
O>
00
O
j
n
oc
'ZJUJ
73 HI
Pi O
tr
PJ
s
o
K,
m > TJ m
?! 11
u o o o
W< A 3
0! O g
D
------- |