United States
Environmental Protection
Agency
Risk Reduction
Engineering Laboratory
Cincinnati OH 45268
Research and Development
EPA/600/S2-88/053 Jan. 1989
x°/EPA Project Summary
Performance of Air
Stripping and GAG for
SOC and VOC Removal from
Ground water
David W. Hand, John C. Crittenden, James M. Miller, and Joseph L. Gehin
A liquid-phase GAC pilot-plant, a
full-scale GAC adsorber, and a full-
scale air stripping tower were
operated to evaluate process per-
formance for the removal of
trlchloroethene (TCE) and other
volatile organic chemicals (VOCs)
and synthetic organic chemicals
(SOCs) present in this water supply.
Extensive laboratory investigations,
including single and multi-solute
isotherms, batch rate experiments,
and film transfer studies, were
conducted to evaluate GAC capacity
and kinetics. Rapid small scale
column tests (RSSCTs), or mini-
columns, were developed and their
predictive ability tested by
comparison to GAC pilot data.
Mathematical models and
correlations for obtaining kinetic and
single solute isotherm parameters
were developed and tested by
comparing their results to those
obtained from the pilot plant.
Possible surrogate parameters such
as total organic halogen (TOX), total
organic carbon (TOC), trihalo-
methane formation potential
(THMFP), total organic halogen
formation potential (TOXFP), and UV
absorbance for use in monitoring
GAC performance were investigated.
Costs of an actual full-scale air
stripping process designed for
minimum tower volume (lowest
capital costs) and energy require-
ments (lowest operation and main-
tenance costs) and liquid-phase
GAC fixed-bed processes designed
from pilot-plant data are compared.
In addition, the costs for air stripping
with GAC off-gas are presented.
This Project Summary was devel-
oped by EPA's Risk Reduction Engi-
neering Laboratory, Cincinnati, OH, to
announce key findings of the research
project that is fully documented in a
separate report of the same title (see
Project Report ordering information at
back).
Introduction
Groundwater is the primary source of
raw water in Wausau, Wl. In 1981, the
City of Wausau discovered that several of
its wells adjacent to the Wisconsin River
were contaminated with SOCs and VOCs.
Among the VOCs identified and targeted
for study were cis-1,2-dichloroethene
(DCE), trichloroethene (TCE), tetra-
chloroethene (PCE), vinyl chloride, and
1,1,1-trichloroethane. Among the SOCs
targeted for study were toluene, ethyl-
benzene, and isomers of xylene.
The objectives of this project were to
compare the ability and cost effective-
ness of two treatment techniques,
granular activated carbon (GAC) and
packed tower aeration (air stripping), in
removing these SOCs and VOCs from
the same water matrix without any form
of pretreatment.
A GAC pilot plant, a full-scale GAC
adsorber, and a full-scale air stripping
tower provided the data used to evaluate
-------
process performance. In cooperation with
the American Water Works Association
Research Foundation, a gas-phase pilot
plant study evaluated the effectiveness of
GAG for removing the SOCs and VOCs
emitted from the off-gas of the full-
scale air stripping tower.
Extensive laboratory studies evaluated
the capacity of the GAC. Equilibrium
isotherm studies were conducted to
quantify the adsorption potential of the
SOCs and VOCs. Single solute, multi-
component, TOX, and TOC isotherms
were conducted with the Wausau water
matrix. An examination was made of the
effects of temperature, pH, and equi-
libration time on isotherm results.
Methods for estimating single solute
isotherm parameters and procedures to
characterize the adsorbing strength of
the unknown components of a water
were investigated. Competitive inter-
actions among the VOCs, SOCs, and
naturally occurring background organics
(MOM) for adsorption sites were also
considered.
Extensive laboratory studies were
performed to describe the kinetics of
adsorption. Batch rate studies were
conducted to examine intraparticle mass
transfer kinetics and a correlation was
developed for the estimation of intra-
particle mass transfer coefficients for
halogenated one and two carbon ali-
phatics and aromatic compounds. Short
column experiments were conducted to
determine mass transfer coefficients and
verify existing film transfer correlations.
RSSCTs were developed and tested
as a predictive method for examining
GAC performance. Mathematical models
capable of predicting solute breakthrough
and carbon usage rate from equilibrium
and kinetic parameters were developed
and compared to pilot plant and full-
scale breakthrough profiles.
Composition of the Wausau
Water Matrix
A summary of Well No. 4's raw water
characteristics observed during each
phase of the project is presented in Table
1. In addition to the compounds pre-
sented in Table 1, many other synthetic,
volatile, and non-volatile organic com-
pounds were found in trace quantities.
The concentrations reported represent
the time weighted averages observed
during each phase of the project. Air-
stripper values are based on the first 4
mo of operation (August-November
1984). GAC pilot-plant values are based
on 12 mo of operation beginning August
Table 1. Average Raw Water Characteristics of Wausau's Well No. 4 During Each Phase of
the Wausau Project
GAC GAS
Air Stripper Pilot-Plant Full-Scale
Compound or Parameter Influent Influent Influent
cis-1,2-Dichloroethene, jig/L
Trichloroethene, itg/L
Tetrachloroethene, itg/L
Toluene, pglL
Vinyl Chloride, pg/L
1,1,1-Trichloroethane, itg/L
Ethylbenzene, fig/L
Xylenes",
Manganese, mg/L
Iron, mg/L
Fluoride, mg/L
Dissolved Oxygen, mgIL
Alkalinity, mg/L
Hardness, mg/L
pH
TOC, mg/L
TOX, itg/L
TOXFPt, ttg/L
THMFPt, ng/L
Influent Temperature, °C
Effluent Temperature, °C
82.3
72.0
59.6
30.9
8.4
1.3
5.1
16.6
1.10
5.02
0.35
<1.0
81.
80.
6.8
8.34
173.0
846.0
na
11
11
70.9
47.9
37.6
19.3
8.2
0.9
4.5
14.5
1.36
4.94
0.32
<1.0
82.
80.
6.8
8.35
141.0
805.0
235.0
13.
17.
71.5
17.1
27.9
7.2
4.1
.61
3.6
15.0
1.70
4.21
.29
2.2
76.
79.
7.0
9.13
88.9
na
10.5
11.0
* Sum of m, o, and p isomers.
t Increase in TOX resulting from a 5-day incubation period at a chlorine dose of 20
mg/L and a chlorine residual of 0.2 mg/L
t Yield of THMs resulting from a 5-day incubation period at a chlorine dose of 20 mg/L
and a chlorine residual of 0.2 mg/L.
#na - not available.
28, 1984. GAC full-scale values are
based on 12 mo of operation beginning
November 26, 1985.
GAC Liquid-Phase Equilibrium
Studies
Single solute isotherms were con-
ducted on most of the compounds shown
in Table 1 with Calgons F-400* carbon.
TOX and TOC isotherms were also
conducted. Correlation for F-400 carbon
was developed to estimate the
Freundlich isotherm parameters, K and
l/n, for hydrophobic compounds in
organic free water. This correlation is
easy to use and requires only knowledge
of the liquid density and solubility of the
solute.
Ideal adsorbed solution theory (IAST)
calculations successfully predicted mul-
ticomponent adsorption equilibria from
single solute isotherms for DCE as
shown in Figure 1 and identical results
found on another date. The IAST
* Mention of trade names or commerical products
does not constitute endorsement or recommenda-
tion for use.
predictions assumed that only the VOCs
and SOCs displayed in Table 1 were
present in the water matrix to compete
with DCE for adsorption sites. IAS1
predictions for the other VOCs and SOCs
were not successful. IAST may have
failed due to many reasons. The mosl
likely reasons are the inaccuracy ol
chemical analyses of these VOCs and
SOCs at concentrations near theii
detection limits and the fact that the
VOCs and SOCs did not account for all
the organics in the background as the
NOM concentration was (measured as
TOC) 7.66 mg/L. Competition with the
NOM in an isotherm bottle may not have
been significant enough to slow the
much faster diffusing DCE. During
equilibration in an isotherm bottle, DCE
can easily diffuse into the GAC particles
ahead of the NOM and no competition is
observed. However, it has been shown
that the GAC capacity for a number ol
chlorinated aliphatics in an isotherm
bottle is significantly reduced if the GAC
that is used was first exposed to NOM.
The impact of the Wausau water NOM
on the GAC adsorption equilibrium foi
-------
i 2
|
<5
is N
Q) *-
O
C
o
o
CO
O. _
§ 2
"5 *""
V)
Mill-a Water; Temp. = 13.8°C
DCE: Wausau Water; C = .633 umol/L
DCE; I.A.S. T. Prediction
Raw Water Date 3/2/85
DCE
f.
Laboratory Parameters
Temperature =12-14°C
Equilibration Time = 5-7 days
Carbon Type: F-400
200 = Three Weeks of Exposure
Ğ = Five Weeks of Exposure
B = Eight Weeks of Exposure
o = Ten Weeks of Exposure
= Top of Full-Scale Adsorber (SO wks)
= Middle of Full-Scale Adsorber (SO wks)
= Bottom of Full-Scale Adsorber (50 wks)
Trichloroethene
'N-
10' 102
Liquid Phase Concentration, ug/l
105
Figure 2.
Adsorption isotherms for trichloroethene on GAC exposed to natural organic
matter.
-------
others obtained from the literature were
used to develop a correlation to estimate
surface diffusion coefficients. This cor-
relation was based on the determination
of the surface to pore diffusion flux ratio
(SPDFR).
An SPDFR of 3.72 was found adequate
for estimating effective surface diffusion
coefficients used in predicting SOC and
VOC breakthrough data in small
laboratory columns in which NOM was
absent (i.e., conducted in organic free
water (OFW)). However, these results in
OFW estimate Ds values that are too
large when NOM is present. A SPDFR
value of 0.4 was required to describe
DCE breakthrough profiles for the pilot
plant and full-scale columns. If a typical
value for the tortuosity is taken as 2.5, a
SPDFR of 0.4 would imply that pore
diffusion has become more important
than surface diffusion. In limited testing,
lower SPDFR values were required for
higher NOM concentrations. Accordingly,
fouling of the GAG by NOM seems to
slow down and in some cases appears to
eliminate surface diffusion.
The impact of NOM fouling on intra-
particle mass transfer was investigated.
The surface diffusivity of TCE was found
to decrease with exposure time ap-
proaching a constant value after about 4
wk. SPDFR's corresponding to this
decrease in diffusivity were evaluated;
after a 10 wk exposure to NOM, an initial
SPDFR of 8.0 in OFW was reduced to
1.97.
Rapid Small Scale Column
Studies
Small columns containing GAG, scaled
to ensure perfect similarity to full-scale
adsorbers containing larger carbon, were
evaluated as a method to predict GAG
performance. Advantages of using the
RSSCT for design include significant cost
and time savings. Unlike predictive
mathematical models, extensive isotherm
and kinetic studies are not requied when
RSSCTs are used to predict full-scale
performance. The choice of a smaller
particle size determines the time
reduction, proper hydraulic loading rate,
and empty bed contact time for the
RSSCT.
Two sets of scaling equations were
developed to relate the small column to
the large column while ensuring perfect
similarity. The scaling equations were
derived from the dimensionless groups
appearing in the dispersed flow, pore
surface diffusion model. RSSCT scaling
equations were developed for the case
when the surface diffusion coefficient is
assumed to be independent of adsorbent
particle radius (constant diffusivity) and
for the case when the value of the
surface diffusivity depends on particle
radius (proportional diffusivity).
Near perfect similarity was obtained
between the small and large column
results for all the components in a six
component mixture in OFW assuming a
constant diffusivity. Field testing of this
RSSCT based on constant diffusivity
failed at Wausau where NOM was
present. When the scaling equations for
proportional diffusivity were used,
satisfactory results were obtained for
TCE, DCE, PCE, and toluene.
The RSSCT procedure is a promising
predictive technique. Considerable time
and expense can be saved in deter-
mining full-scale adsorber performance
with a properly designed small column
study. However, more field testing is
required because the extent to which
surface diffusivity changes with particle
size and the impact of NOM on the
RSSCT procedure have yet to be fully
characterized.
GAG Pilot Plant Studies
A pilot plant consisting of 6 columns
containing varying amounts of GAG was
operated to obtain effluent profiles of the
SOCs and VOCs for nominal EBCTs of 1,
3, 5, 10, 20, and 30 min. Influent and
effluent concentration breakthrough
profiles were developed for TCE (Figure
3), and the other compounds listed in
Table 1. The profiles were used tc
evaluate GAG performance.
The specific volume, or number ol
liters of water treated per gram ol
carbon, was assessed for each com-
pound at a treatment objective of 5 jig/L
Figure 4 displays the liters of watei
treated per gram of carbon for TCE foi
the six EBCTs investigated in the pilo
study. For the range of EBCTs examinee
in this pilot study, the specific volume
increased as EBCT increased (see
Figure 5). The other VOCs and SOC:
showed similar results.
Comparison of Pilot and Full-
Scale GAC Studies
A single GAC contactor, 7 ft ir
diameter, was installed to treat 100 gpn
at a nominal EBCT of 10 min. Influent
mid-depth (7.4 min EBCT), and effluen
(12.7 min EBCT) concentration break
through profiles were developed for TCE
(Figure 6), and the other compound:
listed in Table 1. As in the pilot-plan
study, the effluent profiles in Figure I
were plotted in terms of the specifi
volume treated to evaluate GA(
performance. Figure 7 displays thi
specific volume treated for TCE, as ob
served in the full-scale adsorber.
*
8-
0 =
o =
A =
ğ =
M =
=
Influent
Effluent; EBCT
Effluent; EBCT
Effluent; EBCT
Effluent; EBCT
Effluent; EBCT
Effluent; EBCT
= 1.01 min
= 3.09 min
= 5.08 min
= 10.4 min
= 21.2 min
- 32.3 min
Trichloroethene
F-400 (12x40)
Loading Rate = 4.72 m/hr
Bulk Density = 457 kg/m3
Ave. Influent Temp. = J3°C
Detection Limit .2 ug/L
0 30 60 90 120 150 180 210 240
Elapsed Time of Operation (days)
Figure 3. Pilot plant trichloroethene profiles.
270 300 330 36
-------
§
o
CQ
O = Effluent; EBCT =1.01 min
o = Effluent; EBCT = 3.09 min
A = Effluent; EBCT = 5.08 min
= Effluent; EBCT = 10.4 min
* = Effluent; EBCT -21.2 min
= Effluent; EBCT = 32.3 min
Trichloroethene.
F-400 (12x40)
Loading Rate = 4.72 m/hr
Bulk Density = 457 kg/m3
Ave. Influent Temp. = 13°C
Detection Limit .i
Average Influent Concentration
0 25 50 75 JOO
Liters of Water Treated per Gram of Carbon
Figure 4. Specific volume treated in the pilot plant for trichloroethene.
125
150
i
I
J8'
-------
o
i Ğ5
C ĞM
c _
01 O
r
o
V
% 10
£~
TJ.
Is
5
Full-Scale Adsorber
Influent
Effluent; EBCT = 7.4 mm
Effluent; EBCT = 12.7 min
Trichloroethene
F-400 (12x40)
Loading Rate = 6.25 m/hr
Bulk Density = 422 kg/m3
Ave. Influent Temp. = 10.5°C
Detection Limit .2 ug/L
Backwash
4/15/86
Backwash
7/16/86
0 30 60 90 120 150 180 210 240 270 300 330
Elapsed Time of Operation (days)
Figure 6. Full-scale trichloroethene profiles.
o
ci
'c: O
2 <6
I
o
a
j
o
-------
that could be modeled at an EBCT of
10.4 min using the conditions successful
at an EBCT of 3.09 min.
The modeling efforts demonstrated
that current model parameter estimation
methods provide parameters that ade-
quately describe SOC and VOC profiles
in organic free water. However, the
model estimation methods do not always
provide parameters that adequately
describe SOC and VOC profiles when
MOM fouling occurs. NOM fouling
reduces intraparticle mass transfer and
GAG capacity for SOCs and VOCs.
Accordingly, more work is required to
improve methods to characterize the
impact of NOM fouling on GAC capacity
and kinetics.
Full-Scale Air Stripping
Studies
Following a procedure developed for a
design of a least-cost air stripping
tower, an 8 ft diameter packed tower was
built at the water treatment plant in
Wausau, Wl to treat the SOCs and VOCs
shown in Table 1. The air stripper was
designed for minimum tower volume and
energy requirements to obtain 95%
removal of TCE. Excellent removals were
observed for all the VOCs and also the
SOCs because of their low concentra-
tions.
Operational problems encountered
during this study were the formation of
poorly settling iron floe in the water
treatment plant, an increase in the
TOXFP in the effluent water from the air
stripping unit, and iron precipitation that
may reduce the performance of high-
efficiency packing media. Analyses of
the effluent from the air-stripping tower
showed that no pathogens are present
that may cause a health threat including
Legionella.
Treatment Costs
The studies at Wausau demonstrated
that both air stripping and aqueous phase
GAC adsorption are two treatment
techniques that can effectively remove
SOCs and VOCs from a contaminated
drinking water supply without pretreat-
ment.
The costs of these two processes were
evaluated for treating 1,500 gpm, the
normal flow rate for contaminated Well
No. 4, to below the 5 ng/L level set for
DCE and TCE by the Wisconsin DNR.
The aqueous phase GAC designs were
based on the behavior observed in the
pilot and full-scale studies. The capital
and operational costs for aqueous and
gas phase GAC treatment processes are
estimates based on manufacturer's price
quotations for installation in Wausau, Wl.
Air stripping costs are actual costs based
on several months operation of a full-
scale tower. Two methods of handling
spent carbon were investigated; 100%
carbon replacement with ultimate
disposal of the spent GAC by incin-
eration and off-site regeneration
assuming 10% attrition losses.
Many design alternatives exist for
removing DCE and TCE and
consequently the more strongly adsorb-
ing compounds that exist in the Wausau
water matrix. When off-gas control is
not required, air stripping at 4.3
cents/1,000 gal was found to be the least
cost process to remove DCE and TCE.
Aqueous phase GAC was found to cost
29.8 cents/1,000 gal for DCE and 20.0
cents /1.000 gal for TCE. In the event
that off-gas treatment is required, the
cost advantage of air stripping was found
to be significantly reduced with air
stripping plus GAC off-gas treatment
costing 20.4 cents/1,000 gal for DCE and
12.8 cents/1,000 gal for TCE.
For all the various adsorber con-
figurations investigated, the lowest cost
adsorber configuration for DCE and TCE
was found to be two 5-mm EBCT
adsorbers in-series. Costs for several
other EBCTs and configurations were
reported to demonstrate the costs of
improper design. The costs for aqueous
phase adsorption are comparable to
those found in West Germany, where a
number of full-scale systems have been
in operation for several years. For
example, the Pforzhiem water treatment
plant treats about 1,500 gpm. The water
contains about 40 jig/L of TCE and 20
pg/L of PCE and is treated using 8
aqueous phase GAC ad-sorbers that
are arranged 4 in parallel with 2 in series.
The spent GAC is regenerated off site
and about 10% is lost during
regeneration. Even after about 8
regeneration cycles the capacity of the
GAC was found to be slightly higher than
the virgin carbon. The cost of treatment
is about 24 cents/1,000 gal at Pforzhiem,
which compares well to costs determined
in this study for TCE.
Conclusions
Correlations were developed that
successfully predicted aqueous phase
Freundlich isotherm parameters of
hydrophobic liquid compounds for two
different GACs.
IAST was able to predict the
multicomponent competitive inter-
actions of DCE from single solute
isotherm data obtained in the field
when the competition from the NOM
was neglected.
TCE and toluene isotherms were
conducted on GACs that were exposed
for various times to natural organic
matter from Well No. 4. These results
showed the Freundlich isotherm
capacity to decrease with increasing
exposure times.
TCE and toluene isotherms conducted
on carbon taken from the top of the
full-scale GAC adsorber after 50 wk
of operation showed higher residual
capacities than those conducted on
GAC taken from the middle and
bottom. This was caused by strati-
fication of the GAC bed. A larger
fraction of smaller GAC particles,
which are not fouled by NOM as fast
as larger ones, were found at the top of
the GAC adsorber. These experimental
results show that the impact of
preadsorption of NOM in fixed-beds
is a function of preloading time,
adsorber length, and the degree of
GAC bed stratification.
A correlation developed for the
estimation of surface diffusion
coefficients was able to predict
intraparticle mass transfer behavior of
SOCs and VOCs in fixed-bed
adsorbers where NOM is absent.
However, when surface diffusivities
obtained from this correlation were
used to predict the pilot-plant results,
surface diffusivities were found to be
much larger than the observed
intraparticle mass transfer rate.
Small columns containing GAC
(RSSCTs), scaled to ensure perfect
similarity to full-scale adsorbers
containing larger carbon, were evalu-
ated as a method to predict GAC
performance.
When the PSDM was fit to the 3.09
min EBCT pilot-plant data, good
descriptions of most of the data could
be obtained if the surface diffusivities
were set equal to zero and the pore
diffusivities were calculated from the
liquid diffusivities using a GAC particle
tortuosity of 3.0. Good descriptions
were also obtained for the 5.08 and
10.4 min EBCTs.
The pilot-plant data showed that for
the EBCTs investigated, specific vol-
ume in terms of liters of water treated
per gram of carbon increased for
increasing EBCTs. By plotting the
pilot-plant data in terms of specific
volume as a function of EBCT, as
shown in Figure 5 for TCE a least cost
-------
operation can be designed when
combined with cost calculations.
Backwashing of GAG adsorbers can
mix-up the mass transfer zone and
stratify the GAC according to size and
density. These effects reduced the
specific volume of water treated by the
full-scale adsorber. Stratification of
the GAC in a fixed-bed adsorber can
cause NOM fouling to have a greater
impact on the specific volume of water
treated for a stratified bed as compared
to an unstratified one.
A simple procedure was developed for
the design of a least-cost air stripping
tower for removing SOCs and VOCs.
The results of the cost analysis
comparing air stripping with and
without off-gas treatment to aqueous
phase adsorption showed that air
stripping without off-gas treatment is
the least cost process. If off-gas
control is required, air stripping plus
off-gas treatment is the least cost
alternative when compared to aqueous
phase adsorption. The least cost
aqueous phase adsorption design was
found to be two beds in series using an
EBCT of 5 min for each bed.
The full report was submitted in
fulfillment of CR811150-01-0 by the
City of Wausau, Wl, under the sponsor-
ship of the U.S. Environmental Protection
Agency.
David W. Hand. John C. Critteodea. ant
Technological University, Houghton, Ml 4902
City of Wausau, Wl 54401.
Benjamin W. Lykins, Jr., is the EPA Project Officer (see below).
The complete report, entitled "Performance of Air Stripping and GAC for SOC and
VOC Removal from Groundwater," (Order No. PB 89-110 2741 AS; Cost:
$15.95, subject to change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Risk Reduction Engineering Laboratory
U.S. Environmental Protection Agency
Cincinnati, OH 45268
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
EPA/600/S2-88/053
0000329 PS
U S EWVIR PROTECTION AGCHCY
RiSION 5 LIS8ART
Z30 S DEARBORN s
CHICAGO
------- |