United States
                     Environmental Protection
                     Agency
Robert S. Kerr Environmental
Research Laboratory
Ada OK 74820
                     Research and Development
EPA/600/S2-91/016 July 1991
<3rEPA       Project Summary
                     Forced  Air Ventilation for
                     Remediation  of Unsaturated
                     Soils  Contaminated  by  VOC
                     Jong Soo Cho
                       Many cases of soil vacuum extrac-
                    tion (SVE) applications In the field have
                    been reported, but very few systematic
                    studies about physical and chemical
                    processes In soil air are found. Param-
                    eters which were expected to control
                    the removal process of VOCs from con-
                    taminated  soil during the SVE opera-
                    tion were studied by means of numerical
                    simulations and laboratory experiments
                    In this project
                       Experimental results of SVE with soil
                    columns In the laboratory Indicated that
                    the removal efficiency of VOCs from
                    soil columns was a complicated func-
                    tion of air flow and the hydrogeometry
                    Inside. The partition process  between
                    air and the immobile liquid was not an
                    equilibrium one, and the interfacial mass
                    transfer varied with the residual amount
                    of VOCs In the soil. Additional experi-
                    ments under various conditions should
                    be conducted to obtain further insight
                    Into the SVE process.
                       Two computer models were devel-
                    oped to study soil air and VOC move-
                    ment during the SVE process. The first
                    one was an analytical approximate
                    model which could be used for the simu-
                    lation of air movement in the  SVE op-
                    eration  with multiple  wells  In
                    homogeneous soil media. The second
                    one was a numerical model in three-
                    dimensional geometry which used a fi-
                    nite  difference solution  scheme.  A
                    simple pneumatic pump test was con-
                    ducted, and parts of test data were used
                    for the validation of the simple analyti-
                    cal model.
    This Project Summary was devel-
 oped by EPA'* Robert S. Kerr Environ-
 mental Research Laboratory, Ada, OK,
 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
    Fuel leakage and spills are the most
 frequent sources of soil  and ground-water
 contamination at service stations and un-
 derground storage tank areas. A large por-
 tion of released hydrocarbon infiltrates the
 subsurface and remains bound under cap-
 illary  pressure as a residual immiscible
 phase liquid. The residual  hydrocarbon
 serves as a continuous source for ground-
 water contamination. Therefore, the recla-
 mation of the contaminated aquifer should
 include removal of the long-term con-
 tamination source. The cleanup of soil
 contaminated by Volatile Organic Contami-
 nants (VOCs) is generally  an expensive
 operation due to the high cost associated
 with excavation, transportation, and dis-
 posal.  Classic methods such as soil re-
 moval, forced percolation, encapsulation,
 or trenching are frequently  impossible or
 prohibitively expensive,  especially in the
 midst of an industrial or  residential area.
    An alternative method to remediate soils
 is the use of the soil vacuum extraction
 (SVE) system. This process has proven to
 be inexpensive and effective for the cleanup
 of soil and ground water contaminated by
 solvents and volatile components of petro-
 leum products. The cost of installation and
                                                                    Printed on Recycled Paper

-------
operation  of  an SVE system is  usually
lower than the cost of other methods. An-
other major advantage is that this method
is an in-situ  process. The contaminated
soil remains in place and is not excavated
and  disposed of in other locations.  The
SVE system is also used for the control of
methane gas migration from landfills. Meth-
ane and carbon dioxide, which are gener-
ated by microbial decomposition of organic
materials, can migrate a long distance from
the landfill and build up to explosive levels.
The  SVE wells, sometimes with interdic-
tion walls, are installed  to prevent  the mi-
gration of  methane gas. In spite of many
field applications, very few scientific or sys-
tematic studies have been reported. There-
fore, the design of the  SVE  system  has
been mainly dependent on experience and
rough estimations.  Sometimes, prototype
or pilot-scale  systems are used to obtain
design parameters such as well depths,
well spacings, and extraction rates.
   The  objective  of  this project  was to
investigate the movement of VOCs in soil
air during the  SVE process. Several physi-
cal and chemical processes are involved
in the movement of VOCs in soil air, in-
cluding convective and diffusive transport,
interfacial  mass transfer between  immis-
cible phases, and biological/chemical trans-
formations.   Physical  and  chemical
properties of soil and VOCs are expected
to control  these processes. The ultimate
goal of this project was to obtain knowl-
edge on relationships among the various
properties and processes of  VOC trans-
port  in soil air. These relationships were
integrated in computer models. This report
includes laboratory experiments and field
tests of SVE  in relatively simple systems.
The  model development procedures are
also  included  in this report.

Process of Soil Vacuum
Extraction
   The  basic principle  of SVE is  very
simple. Air flow is induced in the subsur-
face by a pressure gradient applied through
vertical wells  or horizontal trenches. The
flowing air sweeps out VOCs by vaporizing
highly volatile components from soil pores,
and the contaminated soil air is collected
by extraction  wells.  Effluent air from ex-
traction  wells  is often  treated by off-air
treatment systems, i.e.,  an activated car-
bon tank or catalytic converters. A typical
SVE system consists of air pumps or blow-
ers connected to a series of wells located
in contaminated soil.  The bwer pressure
inside the extraction well generated  by
pumps causes soil air to move to the well.
Sometimes air injection wells are added
for the further control  of  air flow. VOC
movement in convective and diffusive
modes is considered as the main process
of chemical transport during  the SVE op-
eration. When there is an induced pres-
sure gradient, the bulk phase of soil air
moves and carries a large amount of VOCs
in the  convective transport mode. Espe-
cially in  the  close  vicinity of wells  and
trenches, a large pressure gradient is de-
veloped and the convective transport domi-
nates the movement of VOCs. At remote
areas from wells,  the  pressure  gradient
becomes very small. VOC transport in these
remote locations from the wells is expected
to be slow because of the diffusive trans-
port. In addition to the convective and dif-
fusive movements,  VOC transport in soil
air during SVE is expected  to be influ-
enced by several other processes includ-
ing the partition process among gas, liquid
and solid soil  matrices,  and biological/
chemical transformations. Biological/chemi-
cal transformation processes of VOCs were
not investigated in this project.

Soil Air Flow
   The VOC concentration  in soil air is
usually low, and the changes in thermody-
namic and transport properties of soil air
due to the concentration  change are not
significant. Therefore, air flow is consid-
ered to be independent of the VOC con-
centration in soil air and treated explicitly
from  VOC transport.  In cases where
changes of physical and chemical proper-
ties due to the high concentration of VOCs
in soil air are significant, iterative or updat-
ing computational  procedures should  be
used to solve the coupled problems.
   Three basic equations are considered
in the  description  of air  flow: the mass
balance equation of soil air, the flow equa-
tion  due to pressure gradients,  and the
equation of state. The mass balance of soil
air is expressed by  the equation of conti-
nuity.

        t
          = -VpaV
(1)
where the 
-------
   NA-cAV-4>.DVcA
                       (5)
where c is the concentration of the compo-
nent A in  soil air, V is the velocity vector of
air flow, $ is the air-filled  porosity in soil,
and D is the diff usivity.
   The mass  balance of the component A
in soil air becomes
                                   (6)
       dt
            -VcAV + V$a DVcA
where S, is the source/sink term of soil air.
   The mass balance of the component A
in the immobile phase becomes
     9CA
      dt
•S,
(7)
where q>/ is the pore volume occupied by
the immobile phase, CA is the concentra-
tion in the immobile phase, and S/ is the
source/sink term for the immobile phase.
The source/sink terms both in the mass
balance of component in soil air and the
immiscible  phase include the  interfacial
mass transfer and biological/chemical trans-
formations.
   Several parameters are used in the
description  of  the VOC  movement  and
therefore are needed to obtain solutions of
the air flow and VOC transport equations.
Properties of soil and VOCs should be
measured for the accurate  design of the
SVE operation  on the specific  site.  Most
thermodynamic and transport properties of
VOC components in  air can be obtained
from reported data. These properties may
change depending on the  operating condi-
tion of the SVE. The property changes
caused by operating conditions sometimes
give considerable variations in the  effi-
ciency  of the SVE.  Thermodynamic and
transport properties of VOCs in various
conditions can be estimated from theoreti-
cal and empirical relationships. These pa-
rameters are  the  soil air  permeability,
molecular weight of soil  air, viscosity of
gas mixture, diffusivity in gas mixture, va-
por pressures and aqueous solubilities of
VOCs.

Interfacial Mass Transfer
   One of  the  empirical  correlations for
the interfacial mass transfer between im-
miscible phases is the first  order kinetics
expression.  The rate  of mass  exchange
between  immiscible phases is expressed
by the mass transfer potential and  the
mass transfer coefficient. The difference of
the concentration at equilibrium and  the
actual concentration in the main body is
defined as the mass transfer potential. The
mass transfer rate is expressed as
   nA-KQ(CA-cA)                (8)

where CA  is the concentration of A at
equilibrium, CA is the actual concentration
in the main body of fluid, and KQ is  the
mass transfer coefficient. The mass trans-
fer coefficient, KQ  , is expected to be a
function of Reynolds' number, Schmidt's
number, and the air saturation, 6., which is
the ratio of the air-filled porosity to the total
porosity of soil.

   KQ - KQ (Re, Sc, 9. )             (9)

Effect of Parameters
   Analytical solutions in one-dimensional
systems have been obtained for the soil air
flow and contaminant transport equations
under simplifying assumptions. Those so-
lutions were applicable  for analysis of
simple soil column  operations. Effects of
parameters on the VOC movement have
been studied through simulations and com-
parisons with soil column experiments.

Analysis of SVE Processes
   The  pressure distribution calculated
from the analytical solution of the air flow
equation showed variations with location
inside a one-dimensional soil column, but
was not a function of the air permeability at
steady state. The pressure and correspond-
ing velocity distributions obtained from the
linearized and nonlinear equations showed
discrepancies between calculated values.
The maximum difference in pressure distri-
butions inside the column was about 20%
when the ratio of the inlet pressure, p^, ,
and the outlet pressure, p^ , was 0.6.  But
the corresponding air flow velocity showed
a significant difference between calculated
values from the  linearized  and nonlinear
equations.  When  the pressure ratio was
0.9, the maximum difference was about 10
%, and it was about 70 % at the location of
the lowest pressure when the pressure
ratio was 0.6. Errors in the air flow velocity
generate errors  in  the estimation of  the
convective movement term in the transport
equation. Therefore, one should  be very
careful when the  linearized equation is
used.
   Air and liquid saturations in soil pores
were expected to have significant effects
on both the air flow and VOC transport.
The  relative permeability of soil air was a
function of the air saturation. If the liquid
saturation, including the water and immis-
cible nonaqueous phases, was high, then
the relative permeability became so small
that a large pressure drop was expected.
The effect  of  the air saturation on the
transport of  the VOC  impacted  the
diffusivity, as did the interfacial mass trans-
fer from the immobile phase to the air flow.
The liquid saturation was  also considered
to determine the effective interfacial area
for the mass transfer. As the liquid satura-
tion increased above the  residual satura-
tion, the mass transfer  coefficient was
expected to increase due  to the increased
interfacial area between contacting phases.
After passing the maximum point, the in-
terfacial area decreased,  and so did the
mass transfer coefficient. The  change of
contacting area was considered  to  be a
complicated function of the hydrogeometry
inside soil pores and was expected to vary
continuously as the removal process of the
VOC and soil  water continues. Additional
efforts should be  made  in this  area of
research. Systematic studies in the labora-
tory and field should be conducted to gain
better knowledge of the process.

Temperature Effect on the
Removal Efficiency
    To  investigate the temperature effect
on the performance of the SVE process,
the soil column experiment was simulated
at three different temperatures. In  compar-
ing property changes at different tempera-
tures, the largest differences were found in
vapor pressures.  At a 15°C increase of
temperature in the ambient condition, the
vapor pressure of TCE doubled. The major
contributing factor to the  performance of
the SVE process at increased temperature
was the vapor pressure increase. The re-
moval  rate doubled when the operating
temperature increased 15°C. In future stud-
ies, it may be worthwhile to consider in-
creased temperature operation.

VOC Removal  Rate
Measurements
    One of the controlling processes in SVE
is the partitioning among  gas, liquid, and
solid soil phases. When  the convection
dominates the transport of VOCs and its
rate is fast, local equilibrium assumption is
not accurate; and a kinetics model seems
to be more appropriate to describe parti-
tion processes. Experimental investigations
of two parameters, the air flow rate and the
liquid distribution  including the nonaque-
ous phase liquid and water, on the re-
moval rate of VOCs from soil columns are
presented.

-------
So/7 Column Experiments
   Nine  soil columns made of 2-inch,
Schedule 80, PVC pipe were used. Each
column had a soil-packed section 28 to 30
cm bng and two additional empty sections
with  a cap on each end. Brass fittings were
attached on the top and bottom sections of
the columns. Both 1/4 inch I.D. plastic and
copper tubings were used; plastic tubings
were used where  VOC did  not contact.
The  Oil  Creek  sand was  packed in  the
columns. The sand was white  and very
uniformly sized with very small amounts of
organic content on the surface. These col-
umns were set up in a constant  tempera-
ture  room.  Pure toluene was used as a
VOC and was applied on top of the soil-
packed section of  each column drop by
drop through a hypodermic syringe to mini-
mize the disturbance of soil packing. The
same method was used for water applica-
tion to  control water content. After applica-
tion  of each liquid, a 24-hour equilibrium
period  was set to achieve uniform distribu-
tion of liquid. The air flow rate was moni-
tored with a rotameter and  adjusted as
necessary with two needle  valves.  The
vacuum  pressure and the  pressure drop
through the  column were measured with
manometers. An in-house vacuum line was
attached to  the  column for  the vacuum
source. The vacuum pressure and corre-
sponding air flow rate fluctuated  about 10
% from the  set point. Air was  saturated
with  moisture by passing it through water
baths  before it entered the columns  to
minimize the water content change in  the
soil.  Samples of  effluent air from  each
column were taken directly by a gas-tight
chromatograph  syringe and  injected into
an HP 5840A gas chromatograph  equipped
with  an FID. A 6 ft, 1/4 inch O.D. custom
packed stainless steel column  (Supelco,
EPA Method 602) was used for analysis of
toluene  concentration in  air.  Nitrogen
served as the carrier gas at a flow rate of
36 ml/min. The injector and FID  tempera-
tures were set at 110°C and 150°C,  re-
spectively. The column oven  temperature
was fixed at 90°C. A specialty gas mixture
(Blend 3, Scott  Specialty Gas) was used
as a  standard for calibration.

Results and  Discussion
   Two major parameters investigated in
this experiment were the air flow rate and
liquid phase  contents of the immiscible
nonaqueous VOC and water in  soil. The
effluent concentration  decreased as the
time  passed even in the  case where only
pure  toluene was applied. This decrease
of effluent concentration suggests that the
interfacial mass transfer is a function of
residual liquid contents inside soil pores.
Also these  decreasing  rates of effluent
concentrations varied with respect to the
air flow  rate.  At  a lower flow  rate, the
effluent concentration change was slower
than at the higher air flow rate. These
trends were maintained through  different
water content and distributions inside soil
columns. To verify the VOC and  water
distribution effects on the removal, the ef-
fluent concentration changes at  different
initial liquid VOC  and water were investi-
gated. Throughout these experiments, the
amount of toluene applied on each column
was far above the  water solubility, and,
therefore, toluene existed as an immiscible
nonaqueous liquid. The  vapor  pressure
was expected  to be that of pure  toluene.
Nevertheless,  the effluent rate was seri-
ously affected  by the water content in the
soil. Increasing water  contents  reduced
the removal rate.  This may be due to the
entrapment of residual VOCs inside soil
pores by water and thereby reducing con-
tacting  area between the air and  VOC
phase. However, this could not be verified.
    Even though the air flow rate and the
liquid distribution of VOCs and water were
distinctive  parameters  verified  through
these experiments, it was not possible to
obtain quantitative correlations among
them The main reason was that through-
out these experiments,  mass  balances of
VOCs  and water could  not be checked.
Another unexpected result was  the  non-
uniform distribution of  VOCs and  water
inside soil columns. It was found that liq-
uids, including toluene and water, moved
upward as air flowed up from the bottom of
the column and accumulated in the upper
part of the soil packed section of the col-
umn. Redistribution of liquids  and experi-
mental results suggest the necessity for a
new design of soil column and experimen-
tal  procedures which include  the control
and measurement of  liquid  content
changes.

Removal Rate Model
   A similarity of  physics involved in the
moisture removal  from a wet solid by dry
air and the VOC removal from soil pores
by uncontaminated air suggests the same
conceptual  model. The moisture  removal
model consists of the constant rate period
and the falling period. The constant rate
period  is at the first stage of the drying
process in which the moisture removal rate
remains constant and is mainly controlled
by external factors like air flow, tempera-
ture, and the moisture content in the air.
The falling rate period represents the sec-
ond stage of the removal process in which
the removal rate decreases as the mois-
ture content reduces after a critical point.
 In this stage, the moisture removal rate is
 controlled by internal factors such as the
 liquid diffusion, the capillary flow of liquid,
 or the flow by shrinkage of solid pores. The
 capillary movement of liquid water due to
 the change of the suction potential by
 evaporation of moisture  inside pores is
 expected to control the drying rate of  sand
 soil or granular materials by air flowing
 through pores. The moisture removal rate
 is approximated by first order kinetics. The
 removal  rate of VOCs from soil pores can
 be expressed with the  same concepts of
 physical  process. During  the first stage
 after air flow begins, the removal rate from
 soil pores remains constant until the  VOC
 content is reduced to a  critical value.
       dt
when *A
(critical amount)
                                  (10)
At the second stage, after the critical point
of the VOC content has been reached, the
removal rate decreases as the VOC con-
tents decreases. It is expressed by a first
order kinetics model  of  the  removal of
VOC from the soil pores.
       dt
where K-|
               -Si
        -, and
                                  (11)
                                  tow-
           (CA,c-CA,e)
est obtainable amount of liquid VOC under
the given operating conditions. The solu-
tion of the  above equation for the falling
rate period  can be obtained.


t_(CA.c-CA.e)|n(CA.c-CA,e)


                        (CA r - CA o)
The semilogarithmic plot of   Mf c   M> e/
vs. t should give a straight line and  the
slope of the curve is related to the constant
rate. The constant  rate S,, and the coeffi-
cient, K,, are expected to be complex func-
tions of the liquid saturation, 6^ and various
operating conditions.

   K,-K, (Re, Sc, 0X)            (13)

Soil Air  Flow Model with
Superpositions of Analytical
Solutions
   The development of a three-dimensional
analytical  approximate model to simulate
the air flow during the SVE operation and a
simple pneumatic pump test conducted on
an aviation gasoline-contaminated site  are

-------
presented. The model used a superposition
of the  analytical solutions obtained from
potential theory in a three-dimensional
space. This model is only applicable to
homogeneous media.  Pneumatic pump
tests were conducted to obtain soil air flow
characterization around an air injection and
vacuum extraction well in relation to a field
demonstration project for bioremediation.
The result revealed the importance of the
pneumatic pump test prior to the design of
a full scale operation. A part of the test
results was used for the validation of the
model.

Analytical Solutions
   At steady state, the equation describ-
ing the pressure distribution of soil air is
   V K. p Vp - 0
                               (14)
Three appropriate boundary conditions are
a constant pressure  boundary, like soil
surface exposed to the atmosphere, a zero
flux boundary for the  impermeable layer,
and the continuous flux condition at the
interface between different permeability lay-
ers.
   The superposition of exact solutions is
possible only for the linear equations. By
applying the Kirchoff transformation on the
above  equation as follows,
dm -KaPorm -*a
-------
   Very important  information on  design
of an SVE system can be obtained from
pneumatic pump tests. Design parameters
in the SVE system should include the zone
of influence in which the pumps can pro-
duce the sufficient airflow. Estimation from
the  measurable vacuum  pressure can
cause overestimation of the radius of the
zone of influence. The major factor in de-
termining the zone of influence should be
the pressure  gradient and corresponding
air flow. The  pumping efficiency could be
measured from the tests. A single high
vacuum pumping well  does not have a
larger zone  of influence than the  low
vacuum pumping well because  a large
pressure gradient exists only in the vicinity
of the well. The SVE system with multiple
low vacuum wells may be more efficient
than one with a single high vacuum well.
Combined operation of extraction and in-
jection wells  could  induce horizontal air
flow,  but still a large vertical flow  is  ex-
pected in  the vicinity of wells. Therefore,
the system should be carefully engineered
to obtain  reasonable efficiency.  A  tightly
covered soil surface may help to  induce
more horizontal flow. Additional field tests
are suggested to measure the  pressure
distribution of the operation with a tightly
covered soil surface and also with various
lengths and locations of well screens.
Three-Dimensional Finite
Difference Model
   A mathematical model was developed
for the simulation of the soil vacuum  ex-
traction process in field scale. This model
consisted  of a soil air flow equation,  the
contaminant transport equation  and  the
mass balance equation of residual hydro-
carbon inside soil matrices. The air flow
equation was transformed to a Laplace
type  equation to obtain the soil  air pres-
sure distribution  and the flow velocities.
The calculated air flow velocities were used
in the transport equation to describe  the
convective movement of VOC. In addition
to the convective movement, the transport
equation included the diffusive movement
and the interfacial mass transfer between
the air and the residual hydrocarbon con-
tacting the flowing air. The mass balance
equation for the residual hydrocarbon was
also  used. The numerical method to solve
these differential equations with boundary
conditions was the finite difference method
in a three-dimensional space domain and
the unsteady state time process.

Finite Difference Solutions
   The soil air flow arrives at steady state
quite rapidly after the SVE system  is initi-
ated  by pumping  the air into/out  of  the
subsurface. The  contaminant movement
 usually is a non- steady state problem until
 all the contaminant mass is removed from
 the system. Therefore, it is reasonable to
 assume that the air flow is at steady state
 during the operating period. In this project,
 a finite difference method with central spa-
 tial difference scheme was adopted to solve
 this air flow equation. Replacing the differ-
 ential equations with the difference opera-
 tors results in the difference equations for
 the air flow equation. By applying the above
 finite difference operator on all the interior
 nodal points, excluding the boundary points,
 w, where specific conditions are assigned,
 the total number of linear algebraic equa-
 tions  becomes  (M-1) x (N-1) x (L-1) -  w.
 The number of equations increases rapidly
 as the nodal points increase. For example,
 100x100x100 system generates 100,000
 simultaneous algebraic equations  which
 need  a tremendous amount of computa-
 tional time. In this project, the point Jacob!
 iterative method was selected to solve the
 equations because it uses considerably
 less CPU  memory than the direct solution
 method, allowing the large physical prob-
 lems  to be simulated. Additional refine-
 ment  adopting various preconditioner and
 accelerator schemes is needed to make
 the program faster and more stable.
   Several finite difference schemes have
 been  developed to solve these convective
 diffusion equations of the VOC transport.
 In this project period, two simple schemes
 have been tested, the explicit method and
 the alternating direction  implicit  (ADI)
 schemes;  only  the explicit scheme was
 implemented in the program. The  advan-
 tage of the explicit scheme over the implicit
 scheme is that each node is  computed
 explicitly and the computations need less
 memory and processor time. The disad-
 vantage is that  the selection of time step
 increment is severely dictated by the sta-
 bility conditions. Usually, the ADI scheme
 is unconditionally stable and has second-
 order convergence error. One would have
 to invert a set of three tridiagonal matrices
 for each time step with the  ADI scheme.

 Computer Implementation
   Very often, the modeling of fluid flow
 and contaminant transport in the  subsur-
 face is dictated by availability of computer
 resources.  Because  of limited  computa-
 tional  resources at Kerr Laboratory, the
 point Jacob! method for the air flow equa-
 tion and the explicit scheme for the VOC
 transport equation were selected. Both
 methods require less memory and compu-
 tational time; but because  of limited time
 step allowed, they may not be suitable for
the long period of simulation. Currently,
the algorithms and controlling program dis-
cussed above  are implemented in  FOR-
 TRAN 77 on an Apollo DN4500 using Unix
 system V as the primary operating system.
 The sampling array used by this model is
 101x101x51  which calculates to 520,251
 physical nodes.  Given  this  amount of
 nodes, the program requires approximately
 10 megabytes  (MB) of memory which is
 not suitable for personal computers.
 Through implication, a larger model would
 require even greater memory to function.
 Provided with  the  main program is a
 postprocessor program which is for graphi-
 cal display of data generated by the main
 program. It is specifically designed with an
 XWindows interface and will require  an
 XWindows server be available to run the
 postprocessor.
 Conclusions and
 Recommendations
    From analyses  of  experiments and
 model simulations, very valuable informa-
 tion could be obtained. The observation of
 simulated results based  on soil column
 experimental conditions  showed several
 physical  and  chemical properties influenc-
 ing the efficiency of the SVE operation.
 Among VOC chemical properties, the va-
 por pressure was the most sensitive factor
 that controlled the efficiency of total opera-
 tions. Air flow rate and liquid distribution
 were very important parameters which con-
 trolled the removal rate of VOCs from soil
 columns. A conceptual model  was pro-
 posed to describe the evaporative process
 of VOCs from  the residual liquid in  soil
 pores. The pneumatic pump test gave very
 important information for design of SVE
 systems, including the zone of  influence,
 soil characterization, and pumping efficien-
 cies. Two computer models for soil air flow
 and VOC transport in the SVE system
 were developed.  The analytical solution
 model developed was very simple and easy
 to use. Simulations of pneumatic  pump
 tests  with this  model  revealed that  the
 model generated  reasonable results and
 could be used as an initial design tool. A
 fully three-dimensional  finite difference
 model was  developed.  Various solution
 methods have  been  tried, and explicit
 schemes were selected to reduce the com-
 putational time and memory requirements.
 A graphical postprocessor was attached to
 enhance  the visualization of output results.
   The proposed future works are as fol-
 lows: A large number of studies on mass
transfer have  been reported in engineering
 literature, but very few pertain to soil sys-
tems. Soil particles and pore sizes are not
 uniform, and the Reynolds' number is usu-
 ally less  than 0.1  in  soil systems. The
extrapolation  of empirical correlations to
 unmeasured  operating  conditions  is  not
desirable, and further studies are required

-------
to obtain more precise  correlations suit-
able for soil systems. More experiments
are suggested with modified columns and
procedures for quantitative analyses. The
proposed model for VOC removals from
soil pores needs to be  verified through
additional experiments. More research on
the enhancement of  SVE  by increased
temperature is also needed to achieve bet-
ter efficiency of the SVE operations. Pneu-
matic pump tests are recommended under
various operating conditions before full
scale implementations of SVE systems.
Tracer gas tests will help further. A main
reason for the  lack of field scale  model
developments is the expense of the model
validation with field  scale data. It is very
costly to perform tests for the model vali-
dation, but this is a very necessary step. A
simple pneumatic pump test, like the one
reported in this project, will give very im-
portant information for model validation and
optimal design of the  SVE  system. Addi-
tional refinement and validation of the ana-
lytical solution  model are necessary for
further field applications.  The finite differ-
ence model is  still in the developmental
stage  and  needs a  rigorous validation
through numerical experimentations and
comparison with field  data. In addition to
the validation, alternative schemes should
be tested to accelerate the computation.
                                                                            .S. GOVERNMENT PRINTING OFFICE: I«W2 - 64H-080/402H

-------
  The EPA author, Jong Soo Cho (also the EPA Project Officer, see below), is with
    Robert S. Kerr Environmental Research Laboratory, U.S. EPA, Ada, Oklahoma 74820.
  The complete report, entitled "Forced Air Ventilation for Remediation of Unsaturated
    Soils Contaminated by VOC," (Order No. PB91-181750/AS; Cost: $17.00, 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:
          Robert S.  Kerr Environmental Research Laboratory
          U.S. Environmental Protection Agency
          Ada, OK 74820
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/S2-91/016

-------