vvEPA
United States
Environmental Protection
Agency
Office of Research and
Development
Washington DC 20460
EPA/600/R-92/174
September 1992
Proceedings of the
Symposium on
Soil Venting
April 29 - May 1, 1991
Houston, Texas
-------
PROCEEDINGS OF THE
SYMPOSIUM ON SOIL VENTING
April 29-May 1, 1991
Houston, Texas
EPA/600/R-92/174
September 1992
Sponsored by
U.S. Environmental Protection Agency
Robert S. Kerr Environmental Research Laboratory
Ada, Oklahoma
in cooperation with
the National Center for Ground Water Research
(a consortium of Oklahoma, Oklahoma State and Rice Universities)
under
CR-812808
Project Officer
Dominic DiGiulio
U.S. Environmental Protection Agency
Region 5, Library (PL- 12J)
77 West Jackson Boulevard, 12th Floor
Chicago, IL 60604-3590
ROBERT S. KERR ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
ADA, OKLAHOMA 74820
Printed on Recycled Paper
-------
NOTICE
These Proceedings have been reviewed in accordance with the U.S. Environmental Protection Agency's peer
and administrative review policies and approved for presentation and publication. Mention of trade names
or commercial products does not constitute endorsement or recommendation for use.
11
-------
FOREWORD
EPA is charged by Congress to protect the Nation's land, air and water systems. Under a mandate of national
environmental laws focused on air and water quality, solid waste management and the control of toxic
substances, pesticides, noise and radiation, the Agency strives to formulate and implement actions which
lead to a compatible balance between human activities and the ability of natural systems to support and
nurture life.
The Robert S. Kerr Environmental Research Laboratory is the Agency's center of expertise for investigation
of the soil and subsurface environment. Personnel at the laboratory are responsible for management of
research programs to: (a) determine the fate, transport and transformation rates of pollutants in the soil, the
unsaturated and the saturated zones of the subsurface environment as a receptor of pollutants; (c) develop
techniques for predicting the effect of pollutants on ground water, soil, and indigenous organisms; and (d)
define and demonstrate the applicability and limitations of using natural processes, indigenous to the soil and
subsurface environment, for the protection of this resource.
This symposium proceedings provides information which can be used to improve the performance of soil
venting systems. The emphasis of the symposium was to describe subsurface physical, chemical, and
biological processes affecting soil venting performance and to introduce recently developed enhancements
to conventional soil venting application.
Clinton W. Hall
Director
Robert S. Kerr Environmental
Research Laboratory
111
-------
PREFACE
Soil venting has been used to remediate soil,s in the United States since the early 1980's. It is a recognized
standard technology in Germany and is widely utilized in the Netherlands. While its application can
sometimes be complex, the basic principle behind soil venting is simple. Air extraction or injection wells
are used to create a pressure differential which causes air circulation through contaminated soils or
consolidated geologic media.
Soil venting has many advantages over remediation techniques requiring excavation and above ground or
offsite treatment. Advantages of venting over excavation include the following.
- Venting can be implemented at actively operating facilities with minimal site disturbance and interruption
of business activities.
- Venting systems can be placed near existing buildings with little concern over structural stability.
Capture zones of vertical wells and angle and horizontal drilling technologies enable remediation of soils
under existing structures.
- Venting can be applied at depths limited only by drilling technology.
- Venting can be applied to consolidated geologic media.
- Venting offgas treatment eliminates uncontrolled release of vapors during operation.
Soil venting is often preferable to above ground onsite and offsite treatment of soils because of lower cost.
Venting involves the use of standard and readily obtainable commercial and industrial equipment (e.g.,
positive displacement blowers, liquid ring vacuum pumps, PVC or stainless steel pipe). Capital expenditures
and energy consumption are often far less than other remedial technologies such as thermal desorption.
Another distinct advantage of venting is that air is introduced into soils which are often deficient in oxygen
causing concomitant removal of VOCs and semivolatile organic compounds through biodegradation.
Given all of these advantages, it is easy to understand why venting is receiving such great interest from
industry, environmental consulting firms, and regulators. Because of this increased interest, a symposium
dedicated solely to this technology was organized to provide information in optimizing venting design and
operation.
The symposium was organized in a way in that would enable attendees to go through many of the steps
commonly followed in designing a soil venting system. The first session entitled' 'S ubsurface Physicochemical
and Microbial Processes" provided necessary background in better understanding complex subsurface
physical, chemical, and biological processes which often dictate the ultimate effectiveness of soil venting
application. Mistakes are sometimes made in venting design because of lack of appreciation of subsurface
processes.
-------
The second session was entitled "Mathematical Modeling". Modeling is often utilized as an interpretation
tool after collection of data. While this utilization of models is appropriate, papers presented in the
symposium illustrate how models can be used to help plan field studies and remedial design prior to entering
the field. Discussion in this session also shows how modeling can be used to help understand or
conceptualize the subsurface system when preliminary information from previous investigations is available.
The third session entitled "Site Characterization" described actual data collection techniques such as
pneumatic testing, soil sampling, and soil gas surveying. A rigorous method to determine radial and vertical
pneumatic permeability of soils was described in this session. Determination of pneumatic permeability is
necessary to evaluate air exchange rates and to space air extraction and injection wells. Presentations were
also given during this session on how soil gas surveying and sampling techniques can improve delineation
of areal and vertical gradients in contaminant concentration.
The fourth session entitled "Field Studies: Conventional Venting/Vapor Treatment" contained presentations
on actual venting application. Factors affecting remediation time and methods to optimize venting design
were discussed. Separate presentations were given on methods to enhance venting application such as hot
air injection, ground water sparging, and an in-situ air stripping remediation method utilized in Germany.
Vapor treatment methods were also presented in this session including discussion on development of a new
catalyst capable of thermally oxidizing chlorinated hydrocarbons.
The fifth and last session was entitled "bioventing". With the exception of a few field research projects, soil
vacuum extraction has been applied primarily for removal of volatile organic compounds from the vadose
zone. However, circulation of air in soils often enhances the aerobic biodegradation of both volatile and
semivolatile organic compounds. Presentations in this session describe how subsurface oxygen levels and
air flow rates can be manipulated to maximize in-situ biodegradation. Presenters also discuss how
bioventing can reduce vapor treatment costs and result in the remediation of semivolatile organic compounds
which cannot be removed by physical stripping alone.
The primary purpose of this symposium and the proceedings which follow are to provide information to
improve remedial design of venting systems. Optimization of soil venting application results in greater
utilization of the technology and decreased remedial costs. Minimizing the financial burden of subsurface
remediation is a goal shared by both industry and government.
VI
-------
TABLE OF CONTENTS
SYMPOSIUM ON SOIL VENTING
Foreword iii
Preface v
Venting Overview/Subsurface Physiochemical and Microbial Processes
Soil Vapor Extraction Technology Development Status and Trends 1
Tom A. Pederson, Camp Dresser & McKee Inc. and
Chi-Yuan Fan, U.S. EPA RREL-Edison, NJ
Soil, Site and Waste Characteristics Controlling the Volatilization of
Organic Contaminants in the Vadose Zone abstract only H
R. Ryan Dupont, Utah State University
The Effect of Moisture on Adsorption of
Trichloroethylene Vapor on Natural Soils 13
Karol J. Oja, Salt River Project and
David K. Kreamer, University of Nevada, Las Vegas
Use of In-Situ Ozonation for the
Removal of VOCs and PAHs from Soils 29
Susan J. Masten, Michigan State University
Habitat Conditions Affecting Bioventing Processes 55
Darwin L. Sorensen and Ronald C. Sims,
Utah State University
Opportunities for Biotreatment of Trichloroethylene in the Vadose Zone abstract only 71
John T. Wilson, Don H. Kampbell and Jong Cho, U.S. EPA, Ada, Oklahoma
Mathematical Modeling
Soil Venting Design: Models and Decision Analysis abstract only 73
Marian W. Kemblowski and Shyamal Chowdery,
Utah State University
Modeling Transport of Organic Chemicals in Gas and Liquid Phases 75
A. K. Katyal, P. K. Patel and J. C. Parker, Virginia Polytechnic Institute
vn
-------
Application of Computer Simulation Models to the Design
of a Large-Scale Soil Venting Systems and Bioremediation 91
J.C. Walton, R.G. Baca, J. B. Sisson,
A.J. Sondrup, and S.O. Magnusen,
Idaho National Engineering Laboratory
Mathematical Modeling of SVE: Effects of Diffusion
Kinetics and Variable Permeabilities 103
Jose M. Rodriquez-Maroto, Cesar Gomez-Lahoz
and David J. Wilson, Vanderbilt University
Development and Application of a Three-Dimensional Air Flow
Model in the Design of a Vapor Extraction System 125
Michael C. Marley, Vapex Environmental Technologies, Inc.
Modeling In-Situ Biodegradation in Unsaturated and Saturated Soils abstract only
Harold W. Bentley, HydroGeoChem, Inc. and
Byran Travis, Los Alamos National Laboratory
Site Characterization
Characterizing Permeability to Gas in the Vadose Zone abstract only 143
Michael Sully, University of Arizona
Use of Soil Gas Measurements in the Design of Soil Vapor Extraction Systems ... abstract only ••• 145
Gary R. Walter and Harold W. Bentley, HydroGeoChem, Inc.
Assessing the Performance of In situ Soil Venting Systems by Soil
Sampling and Volatile Organic Compound Measurements abstract only 147
Robert L. Siegrist, Oak Ridge National Laboratory
A Field Technique for Determining Unsaturated Zone Air-Permeability 149
Craig J. Joss, Drexel University,
Arthur L. Baehr and Jeffery M. Fischer, U.S. Geological Survey
Field Studies: Conventional Venting/Vapor Treatment
Optimization of the Vapor Extraction Process: Large Physical Model Studies abstract only
Richard L. Johnson, Oregon Graduate Institute
Field Test of Enhancement of Soil Venting By Heating abstract only 173
David W. DePaoli, Oak Ridge National Laboratory,
and Neil J. Hutzler, Michigan Technological University
vm
-------
Performance Characteristics of Vapor Extraction
Systems Operated in Europe 193
Dieter H. Miller, HPC Harress Pickel Consult GmbH, Germany
Vacuum-Vaporizer-Wells (UVB) for In Situ Remediation of
Volatile and Strippable Contaminants in the Unsaturated and Saturated Zone 203
B. Herrling and J. Stamm, University of Karlsruhe, Germany
EJ. Alesi, GfS MbH, Kirchheim/Teck, Germany and
P. Brinnel, Hydrodata GmbH, Oberursel, Germany
Final Results of a Year-Long USEPA SBIRP/Industry-Funded
In Situ Vapor Stripping Pilot Scale Study 229
Ann N. Clarke, Robert D. Mutch, Eckenfelder, Inc.,
and David J. Wilson, Vanderbilt University
Air Sparging-Extending Volatilization to Contaminated Aquifers 249
Richard A. Brown and Ricardo Fraxedas, Ground Water Technology, Inc.
Commercial Vapor Treatment Processes 271
F.A.M. Buck and E.L. Seider, King, Buck, and Associates, Inc.
Catalytic Destruction of Hazardous Halogenated Organic Chemicals abstract only 281
George R. Lester, Allied Signal, Inc.
Field Studies: Bioventing
Bioventing For In-Situ Remediation of Petroleum Hydrocarbons 283
Robert E. Hinchee, Battelle and Ross N. Miller, U.S. Air Force
A Field Scale Investigation of Soil Venting Enhanced Petroleum Hydrocarbon
Biodegradation in the Vadose-Zone at Tyndall AFB, Florida 293
Ross N. Miller, U.S. Air Force, Robert E. Hinchee,
Battelle, Catherine C. Vogel, U.S. Air Force
Subsurface Remediation at a Gasoline Spill Using a Bioventing Approach 309
Don Kampbell, U.S. EPA, Ada, Oklahoma
A Demo-Project for In Situ Subsoil and Aquifer Restoration
Following Hydrocarbon Spills at a Tankstation 317
J. van Eyk and C. Vreeken, Delft Geotechnics, The Netherlands
IX
-------
ACKNOWLEDGMENTS
This document could not have been completed without the assistance of Dr. Kathy Diddle and Ms. Carol
House. Dr. Biddle, formerly of Rice University, provided logistical support and administrative assistance
for the actual symposium. Ms. House of the Dynamac Corporation provided extensive page layout and
desktop publishing support for the proceedings.
-------
SOIL VAPOR EXTRACTION TECHNOLOGY
DEVELOPMENT STATUS AND TRENDS
TOM A. PEDERSEN
Camp Dresser & McKee Inc.
Ten Cambridge Center
Cambridge, MA 02142
CHI-YUAN FAN
USEPA RREL
2890 Woodbridge Avenue
Edison, NJ 08837
INTRODUCTION
Soil Vapor Extraction (SVE) is the process of removing gaseous contaminants from soil pore spaces by
causing air to flow through the subsurface environment. Air flow is usually induced by pumping from a well
screened in the unsaturated soil zone. SVE technologies have been widely applied for the treatment of
unsaturated soils contaminated with volatile and semivolatile organic compounds and has been shown to
provide cost effective and environmentally sound remediation when appropriately designed and operated.
Although extraction of natural gas from geologic deposits and venting of methane from landfills has been
practiced for decades, application of gas extraction techniques for the remediation of environmentally
impacted soils is a relatively recent development. Scientific and trade publications are replete with reports
of SVE experiments and case studies undertaken since 1985, however, only a few published reports pre-date
this period; most notably being those by the Texas Research Institute (1980), and Thornton and Wootan
(1982).
SVE system components have been outlined in a number of publications (3,4,13); leading to the widespread
use of this technology at diverse site types. Although these technologies have most frequently been used to
treat underground storage tank (UST) petroleum product releases, they are being applied increasingly at
chlorinated organic solvent release sites as-well as Superfund sites. SVE has been identified as a remedial
technology in 48 of the 749 U.S. Environmental Protection Agency (USEPA) Superfund Records of
Decisions (RODs) signed from fiscal year (FY) 1982 through FY1990 (17). The first ROD to include SVE
was signed in FY 1985 whereas 19 FY 1990 RODs included SVE treatment approaches.
This paper provides a brief description of system components and operation, as well as a brief review of soil/
contaminant principles important for assessing SVE applicability. SVE system enhancements, such as air
sparging, steam injection, and modifications, including bioventing and soil mounds and technology trends
are also discussed.
-------
SVE APPROACHES
SVE technologies are relatively simple in design and the equipment that comprises the systems consists of
commonly-used and widely available devices such as PVC piping, valves, and pumps. These factors impart
an advantage to SVE over other techniques that may require more complex design, or which require single-
purpose equipment. Simplicity of system components however, does not imply ease of design. Maximum
system efficiency and contaminant removal will occur only through a thorough understanding of site
conditions and SVE processes.
The objective of a well-thought out and reasoned design process is to construct a soil vapor extraction system
that removes the greatest degree of contamination from the site in the most efficient, timely, and cost-
effective manner. The attainment of these objectives requires an understanding of the determinants of system
effectiveness which include: the composition and characteristics of the contaminant; the vapor flow path and
flow rate; and the location of the contamination with respect to the vapor flow paths (6). Design of an SVE
system is basically a process to maximize the intersection of the contaminated zone with the vapor flow paths.
Operation of the system should be aimed to maximize the efficiency of the contaminant removal and reduce
costs.
The basic equipment for SVE systems consists of: pumps or blowers to provide the motive force for the
applied vacuum; piping, valves, and instrumentation to transmit the air from the wells through the system
and to measure the contaminant concentration and total air flow; vapor pretreatment to remove soil particles
and water from the vapors treat; and an emission control unit to concentrate or destroy the vapor phase
contaminants (Figure 1).
Modified SVE approaches included bioventing, soil mounds and air sparging.
Bioventing
Bioventing is a term which has been used to describe the in situ microbial degradation of contaminants
mediated by SVE. Many environmental contaminants are subject to degradation by naturally occurring,
introduced or engineered organisms. The rate at which degradation proceeds is related to soil/waste
characteristics and in many cases is controlled by an adequate oxygen, or other electron acceptor or nutrient
source. Bioventing provides a means of introducing added oxygen to the subsurface environment needed
for aerobic degradation of organic contaminants. Although SVE applications will invariably result in
stimulating biodegradation, bioventing attempts to engineer the vapor flow rates and nutrient balance in soils
to maximize microbial activity.
Soil Mounds
Extraction of vapors from stockpiled soil mounds is a modified form of SVE. Vapors are extracted from the
stockpiled soil by use of blowers and slotted or perforated extraction piping placed within the soil mound.
The mounds are generally covered with geomembranes to provide control of volatile release and moisture
content. The air in the mounded soil also provides oxygen needed to mediate bioremediation. The use of
mounds to degrade organic compounds has been widely applied for composting of wastewater treatment
-------
plant sludges. However, treatment of petroleum contaminants or xenobiotics does not necessarily require
achievement of the thermic temperature regimes needed to destroy pathogenic organisms present in sludge.
The key conditions needed for enhancement of biodegradation in soil mound systems are the exchange of
air to remove volatile components and provide oxygen required for microbial growth, adequate moisture and
nutrients for growth, and mesic temperature regimes.
Biofilters
Soil mounds in the form of biofilters can also be used to treat gases extracted using in situ SVE systems. The
extracted contaminant laden vapor are introduced into the soil mound which serves as a growth media for
microbial growth. The compounds introduced are sorbed to soil particles and acted upon by microbes in the
soil mound. The treated air flows from the soil mound which are uncovered in most biofilter applications.
Air Sparging
Air sparging, also referred to as "in situ air stripping", is a treatment technology applied to remove volatile
organic contaminants from the subsurface saturated soil and groundwater zones. Air sparging is the process
of introducing contaminant-free air into the groundwater of an affected aquifer to achieve the transfer of the
contaminants from the saturated soil and groundwater to the unsaturated soil pore space, from which they
can be removed by soil vapor extraction (8).
Air sparging systems are almost always coupled with SVE systems to allow for the capture of the volatile
contaminants stripped from the saturated zone. The use of an air sparging system without an SVE system
could result in a net positive pressure in the subsurface which may cause contaminated soil vapor to migrate
to previously uncontaminated areas, thus increasing the overall zone of contamination. Additionally,
without SVE, uncontrolled contaminated soil vapor flow could enter building structures or utility conduits
creating potential explosion or health hazard.
Marley et. al. (1990) suggest that the transport of immiscible contaminants from the saturated zone to the
vadose zone requires the dissolution into the aqueous phase followed by diffusion/ dispersion of the
dissolved contaminants through the aqueous phase to the air-water interface where they can be removed by
SVE. This rate of contaminant transport from the groundwater to soil gas phase has been shown to increase
by sparging.
The effectiveness of the air sparging/SVE system can be attributed to two major mechanisms: contaminant
mass transport and biodegradation. Depending on the configuration of the system, the operating parameters,
and the types of contaminants found at the site, one of these mechanisms usually predominates, or can be
enhanced to optimize contaminant removal. In both remediation mechanisms, oxygen transport in the
saturated zone plays a key role. Although the exact nature of the saturated zone vapor phase is not completely
understood, Ardito and Billings (1990), as well as Brown and Fraxedas (1991), theorize that sparging creates
air bubbles which move through the groundwater to the unsaturated soil analogous to bubbles in an aeration
basin whereas Middleton and Killer (1990) describe the movement of air through irregular pathways in the
saturated zone and ultimately to the surface of the water table as discrete pockets of air. The nature of the
air sparging physics will determine the effectiveness of mass transfer to and from the groundwater regime.
-------
Transfer of oxygen to the groundwater or dissolution of volatiles from groundwater would be expected to
be related to bubble volume to total air volume ratios than continuous irregular pathways of air flow.
SITE EVALUATION
The transport and fate of organic contaminants in the subsurface environment continues to be the subject of
major research efforts. Although the behavior of non-aqueous phase contaminants in the subsurface and
unsaturated flow is not yet fully understood, some very useful descriptions are available including those by
Mercer and Cohen (1990), USEPA (1989) and Lyman, Reidy and Levy (1991).
Contaminants released to the subsurface environment are acted upon by numerous forces that influence the
degree and rate at which they migrate from the source or point of release. The extent to which the released
products partition into the vapor phase is influenced by: the quantity of product released; time over which
the release took place, or since the release occurred; contaminant physicochemical properties; soil
characteristics; and the nature of the subsurface environment.
Physical Factors
Critical to the application of SVE technology is the ability to achieve adequate vapor flow through the
contaminated zone. Vapor flow rates are dependent upon soil characteristics such as porosity, moisture
content, and permeability, as well as the gas viscosity, density, and pressure gradients.
Coarse-textured, highly permeable soils are best suited to SVE application although SVE can work
successfully in fine textured soils, where interbedded permeable layers exist or macropores and secondary
structure exist. High vacuum systems have been used to remove volatiles and groundwater in very slowly
permeable soils. Soil water content has a significant effect on the air permeability. In general, higher water
contents reduces the air-filled porosity thereby decreasing the connected pores through which air can flow
by advection. SVE is generally more successful at lower moisture contents since high water content reduces
the air-filled porosity available for airflow.
Since air permeability will control the decision to implement SVE or bioventing technology at a contaminated
site to a large degree, the importance of the air permeability measurement or estimation technique is evident.
Numerous methods have been advanced for determining soil-air permeabilities for sites at which SVE
techniques might be applied. Some SVE technology vendors utilize proprietary air permeability test
methods for estimating cleanup times and establishing system design criteria.
Contaminant Factors
The physical and chemical properties of the contaminant affects its movement and ultimate fate soil
micropores. The degree to which the contaminant partitions into the vapor phase is described by the
contaminant's volatility, its tendency to become sorbed to soil particles, and its ability to dissolve in pore
water. The contaminants distribution between product, soil particles, pore water, and pore gases will vary
over time in response to weathering.
-------
SVE has the ability to remove soil vapors resulting from the spill, disposal, or leakage of non-aqueous phase
liquids (NAPLs). Those NAPLs with densities less than water are termed light LNAPLs whereas those
denser than water are termed DNAPLs. LNAPLs float on the water table and DNAPLs may sink through
groundwater and hence have also been described by the terms floaters and sinkers. Although the density of
the NAPL product will affect the redistribution in the subsurface environment, numerous other soil factors
enter into the behavior of these products. In many cases, the NAPLs will not behave as conceptually depicted
due to soil macropore flow or discontinuities. Floating DNAPLs may therefore be present under some
conditions.
Weathering changes the nature of a chemical mixture after its release into the environment. The product
composition will change over time and affect the ease with which that product may be removed via SVE.
The more volatile, soluble, and degradable compounds will be subject to removal from the mixture initially,
leaving the resultant mixture relatively richer in less-volatile, less-soluble, and more-refractory compounds.
As SVE progresses and volatile fractions are removed ganglia, or isolated globules of product may form in
soil pores. These globules of product may develop resistant skins which slow diffusion of vapors to the soil
pores. Highly weathered petroleum products may therefore require implementation of enhanced SVE
approaches.
Perhaps the most important contaminant characteristic affecting the applicability of soil vapor extraction is
its volatility. Vapor pressure and Henry's Law Constant are used to describe this tendency. Vapor pressure
is the force exerted by the vapor of the chemical in equilibrium with its pure solid or liquid form. Henry's
law governs the volatilization of a solvent from an aqueous solution, rather than from a pure product. Henry's
Law Constant is a more appropriate partitioning constant for evaluating partitioning outside of the free
product zone, where product is likely to exist in solution with pore water. Compounds with Henry's Law
constants above 0.01 (dimensionless) will tend to move from the aqueous to gaseous phase and will be acted
upon by SVE systems (4). Comprehensive discussion of Henry's Law constants vapor pressure as well as
other important physicochemical properties of compounds and methods for estimating chemical properties
can be found in Lyman, Reehl and Rosenblatt (1990).
Sorption of contaminants to soil particles and organic matter will influence distribution and movement of
released products. Organic carbon, being the most effective soil component with respect to organic sorption,
is generally used in equations used to predict partitioning of contaminants between soil and aqueous phases.
As the soil organic content increases, sorption of most organic product increases.
The product's solubility controls the degree to which a product dissolves into groundwater and pore water
present in the vadose zone. Soluble products are more likely to dissolve in infiltrating precipitation and
migrate from the source.
SVE APPLICABILITY ASSESSMENT
Although SVE and bioventing are often implemented without conducting pilot studies, the data obtained
through field piloting is invaluable in developing full system design. Pilot programs may be as simple as
conducting air permeability tests at a potential SVE site. Such a test provides critical site specific data which
allows for an effective full scale system design.
-------
The air permeability of the soil is perhaps the single most important soil parameter with respect to the success
of vapor extraction. The permeability incorporates the effects of several soil and vapor characteristics.
Among the important soil characteristics to be considered are the stratigraphy, air-filled porosity, particle size
distribution, water content, residual saturation, and the presence or absence of macropores or preferred
flowpaths; important product characteristics include the vapor viscosity and vapor density. Soil-air
permeability plays a key role in determining SVE applicability and also in system design. The soil-air
permeability incorporates many soil characteristics including porosity, structure, grain size distribution,
water content, and preferred flow paths.
Soil-air permeability may be estimated from known physical characteristics of the soil sample, such as the
grain size distribution or saturated hydraulic conductivity (14). Massman (1989) discusses the use of a
correlation between both soil grain size distribution analyses or saturated hydraulic conductivity and soil air
permeability. These methods do not account for decreases in the air permeability due to increased moisture
contents or vice versa, nor do they account for in situ bulk density, soil structure, or heterogeneity of the
subsurface soils.
The use of saturated hydraulic conductivity values to estimate air permeability is subject to several additional
sources of error including soil moisture content, swelling soils, and gas slippage. Both air permeability and
saturated hydraulic conductivity are influenced by the soil water content and hydraulic conductivity
generally increases while air permeability generally decreases as the water content of a soil increases.
Correlation of saturated hydraulic conductivity values and air permeability would not be expected to be valid
for soils with an appreciable expandable clay content. The "slippage" of gas as it passes along the soil pore
wall is commonly known as the "Klinkenberg effect". This phenomenon accounts for greater soil-air
permeabilities than the aqueous permeability at low flows in fine grain soils. Correlation methods provide
a quick means of assessing the relative permeability of soil but are generally not appropriate for SVE system
design criteria development.
In situ field soil-air permeability methods rely on measuring the difference between the ambient atmospheric
pressure and the soil-air pressure during vapor extraction. The methods used for SVE evaluations are
modifications of oil field natural gas pressure build-up and drawdown tests. Johnson et. al. (1990 a&b)
developed a soil-air permeability test for SVE evaluations which is similar to the oil field drawdown gas
permeability test. The drawdown, or vacuum pressure, is measured in a monitoring point at a known distance
from the vapor extraction well, while extracting vapors at a constant rate. Soil-air permeability is estimated
graphically from field data by calculating the slope of the regression line that relates gauge pressure,
measured at a sample probe well to the natural logarithm of the time, from the initiation of vapor extraction.
All of the parameters used to estimate permeability are measured in the field with the exception of the
dynamic viscosity of air which is estimated as a function of air temperature.
Soil-air permeability determination using the method suggested by Johnson et. al. (1990 a&b) assumes that
as the time from initiation of soil vapor extraction increases, the vacuum pressure in the subsurface increases
(i.e., the absolute pressure becomes more negative). The time interval over which air permeability
measurements are made should be long enough to extract at least one pore volume of air, yet short enough
not to be hampered by: variations in atmospheric pressure, and effective porosity changes that occur after
rainfall and when soil air moisture condenses and evaporates during diurnal temperature changes. Because
-------
it is often difficult to maintain a constant vapor extraction rate during S VE operation, variations in the vapor
extraction rate should be recorded and used when evaluating data. The sensitivity of the air permeability
measurement will be reduced as the variations in the vapor extraction rate increase. Permeability values
should be measured at a number of locations around the vapor extraction well and then averaged to provide
a reasonable estimate of the areal variability soil air permeability.
Air permeability tests are relatively simple procedures similar to groundwater pumping tests. A vacuum (or
sometimes positive pressure) is applied to a vapor extraction well screened in the vadose zone. The pressure
distribution, which is established in the subsurface due to the application of the vacuum, is measured by
collecting soil pressure data from probes located at various horizontal and vertical distances from the
extraction well. Knowledge of the soil pressure data is used along with the pressure at the extraction well
to calculate a soil-air permeability.
In addition, trailer mounted S VE pilot units may be used for full scale treatment unit at sites where the volume
of contaminated soil is small and the cleanup time is short. Many retail gas station sites fall into this category
and trailer mounted systems which allow everyday business activities to continue uninterrupted at the site.
The units are also useful on a pilot scale in cases where data concerning contaminant removal rates are needed
before proceeding with full scale design.
TRENDS
S VE has found widespread application for remediation of soils impacted through the release of gasoline and
other petroleum products from underground storage tanks. SVE systems will find wider application to
complex subsurface conditions as a better understanding of vapor flow physics and contaminant fate is
gained. Models for subsurface vapor flow prediction will continue to become more sophisticated and
provide needed tools for understanding soil/vapor/contaminant behavior. Advances in analytical and field
investigation approaches will allow for more accurate assessment of contaminant transport and clean up
attainment evaluation will be advanced through the use of innovative statistical techniques.
Horizontal well systems and high vacuum techniques will continue to expand SVE application to more
restrictive subsurface conditions. Subsurface modification by pneumatic fracturing and bioaugmentation
will also be applied at an increasing number of sites in the future. Steam injection, subsurface radio
frequency, or other radiation sources to effect increase volatilization, and the use of gases other than ambient
air will broaden the types of contaminants that will be amenable to removal by SVE. Advances in discharge
air treatment technologies will allow for SVE implementation in areas where regulatory discharge
limitations previously restricted SVE technologies.
The overall trend in SVE is towards increased sophistication in the assessment of subsurface conditions,
simplification of system design, coupling of treatment technologies and application to increasingly complex
situations.
-------
REFERENCES
1. Ardito,C.P. and J.F. Billings. 1990. Alternative Remediation Strategies: The Subsurface Volatilization
and Ventilation System.
2. Brown,R.A.andR.Fraxedas. 1991. Air Sparging Extended Volatilization to Contaminated Aquifer.
Pre-publication Draft. Presented at the USEPA Symposium on Soil Venting, Houston,TX. April 29-
May 1, 1991.
3. Crow, W.L.,E.P. Anderson and E.Minugh. 1985. Subsurface Venting of Hydrocarbon Vapors from
a Underground Aquifer. API Report No. 4410.
4. Hutzler, N.J., B.E. Murphy, and J.S. Gierke. 1988. Review of Soil Vapor Extraction System
Technology.
5. Johnson,P.C.,M.W.Kemblowski,andJ.D.Colthart. 1990(a). QuantitativeAnalysisfortheCleanup
of Hydrocarbon-Contaminated Soils by In Situ Venting. Groundwater 28(3):413-429.
6. Johnson, PC., M.W. Kemblowski, J.D. Colthart, D.L. Buyers, and C.C. Stanley. 1989. A Practical
Approach to the Design, Operation, and Monitoring of In situ Soil Venting Systems. Presented at
the Soil Vapor Extraction Technology Workshop, Office of Research and Development, Edison, NJ.
June 28 & 29,1989.
7. Johnson, PC, C.C. Stanley, M.W. Kemblowski, D.L. Byers, and J.D. Colthart. 1990(b). APractical
Approach to the Design, Operation, and Monitoring of In Situ Soil-Venting Systems. Ground Water
Monitoring Review, Spring, 1990, pp. 159-178.
8. Loden, M.E. and C. Y. Fan. 1992. Air Sparging Technology Evaluation. Presented at HMCRI R&D
'92 - National Research and Development Conference on the Control of Hazardous Materials,
February 4-6,1992, San Francisco, CA.
9. Lyman, W.J., W.F. Reehl, and D.H. Rosenblatt. 1990. Handbook of Chemical Property Estimation
Methods - Environmental Behavior of Organic Chemicals. American Chemical Society, Washington,
DC.
10. Marley, M.C., M.T. Walsh, and P.E. Nangeroni. 1990. Case study on the application of air sparging
as a complimentary technology to vapor extraction at a gasoline spill site in Rhode Island.
Proceedings of HMCRI's llth National Conference and Exposition Superfund '90.
11. Massman, J.W. 1989. Applying groundwater flow models in vapor extraction system design.
Journal of Environmental Engineering, 115(1): 129-149.
12. Mercer, J.W. and R.M. Cohen. 1990. A Review of Immiscible Fluids in the Subsurface: Properties,
Models, Characterization and Remediation. Journal of Contaminant Hydrologv, 6:107-163.
-------
13. Pedersen, T.A. and J.T. Curtis. 1991. Soil Vapor Extraction Technology - Reference Handbook.
EPA/540/2-91/003.
14. Sellers, K.L., T.A. Pedersen, and C.Y. Fan. 1991. Soil Vapor Extraction Air Permeability Testing
and Estimation Methods. Presented at USEPA ORD Research Symposium. Cincinnati, OH. April
9-12,1992.
15. Texas Research Institute. 1980. Examination of Venting for Removal of Gasoline Vapors from
Contaminated Soil. Reported by API, 1986.
16. Thornton, J.S. and W.L. Wootan, Jr. 1982. Venting for the Removal of Hydrocarbon Vapors from
Gasoline Contaminated Soil. Journal of Environmental Science & Health, A17(l):31-44.
17. USEPA. 1991. ROD Annual Report - FY 1990. Office of Emergency and Remedial Response.
Publication No. 9355.6-04.
-------
SECONDARY
;. EMISSIONS
VACUUM
BALL GAUGE
VALVE
HEADER
IMPERMEABLE
SURFACE SEAL
A
BENTONITE
CEMENT.
GROUT
SL
—7
SAND
PACK
XJ=
0.020 SLOT
SCREEN
WATER COOLED
HEAT EXCHANGER
VAPOR
TREATMENT
UNIT
PRESSURE
GAUGE
T* 9
ANNUBAR
WATER
AIR-WATER
SEPARATOR
"r
STRAINER
WKz
PUMP
*•
SUBMERSIBLE
PUMP
PRESSURE
RELEASE
VALVE
SILENCE
MUFFLER
BLOWER
TO WATER
TREATMENTSYSTEM
Figure 1. Soil Vapor Extraction System Schematic
-------
SOIL, SITE, AND WASTE CHARACTERISTICS CONTROLLING THE
VOLATILIZATION OF ORGANIC CONTAMINANTS IN THE VADOSE ZONE
R.RYANDUPONT
Utah State University
ABSTRACT
Soil vacuum extraction (SVE) has found wide application in the in-place treatment of fuels and solvent
contaminated soils, but it's effectiveness and limitations under a wide range of soil/site/waste conditions is
not well understood. The fate of hazardous contaminants subjected to this remediation alternative is
controlled by the complex interaction of physical, chemical and biological process taking place under
vacuum induced flow conditions. The overall effectiveness of conventional high rate SYE systems is
determined in large part by the tendency of soil contaminants to move into the vapor phase, a property which
is described by a contaminant's vapor pressure, aqueous solubility and distribution among the various
compartments existing within the vadose zone. The physical performance of an SVE system, i.e., it's ability
to provide contact between extracted soil gas and the contaminated soil zone, also impacts the overall
efficiency of site remediation using SVE, and is limited by subsurface heterogeneities in both soil and
contaminant distribution.
Fundamental soil, site and waste characteristics that impact both constituent volatilization and vapor flow
through soils will be reviewed. Particular emphasis will be placed on a discussion of mass transfer limitations
that may result from these subsurface heterogeneities. The potential magnitude of such transfer limitations
will be presented, and practical field methods for the identification of potential limitations to SVE system
performance will be reviewed as they relate to system design, and operation for optimal site remediation.
11
-------
THE EFFECT OF MOISTURE ON ADSORPTION OF
TRICHLOROETHYLENE VAPOR ON NATURAL SOILS
KAROLJ. OJA1, DAVID K. KREAMER2
1 Salt River Project, P.O.Box 52025, Phoenix, AZ 85072
2 Director, Water Resources Management Program,
University of Nevada, Las Vegas; Las Vegas, NV 89154-4029
ABSTRACT
The vapor sorption of trichloroethylene (TCE) on three recompacted natural soils at various moisture
contents ranging from air dried to 20% (gH20/gsoil) was determined at room temperature using a static
adsorption procedure. Vapor uptake on the soils was determined at a single initial concentration of 97.8 ppb
TCE in nitrogen. Soil-vapor partition coefficients for the air dried soils were observed to be a factor 2-4 times
greater than that for the moist soils. Above 5% water content, the TCE partition coefficient was found to be
relatively independent of the degree of saturation. The observed increased adsorption of TCE vapor on dry
natural soils suggests that optimal design of vapor leak detection systems in unsaturated subsurface
environments, such as those used around underground storage tanks should use models based on dry soils.
This is because the greater vapor adsorption exhibited in dry environments would constitute a conservative
estimate of vapor migration and the effectiveness of vapor monitors.
INTRODUCTION
An understanding of adsorptive processes in natural soil environments is an important parameter in
describing contaminant transport in soil and ground water. Although a considerable amount is known about
the sorptive behavior of organic contaminants in aqueous solutions or saturated soils (13, 24), and dry
adsorbents with no free liquid water content (5), much less is known about sorption reactions of gaseous
pollutants in unsaturated soils. Such reactions, however, are of considerable significance because a large
class of groundwater contaminants are volatile organic compounds (VOC's), which can readily distribute
between the liquid phase in saturated soils and the vapor phase in unsaturated soil regions (22). Understanding
the transport and fate of these chemicals in unsaturated zones has become increasingly important in recent
years.
Soil-gas surveys, designed to measure VOC vapor in the soil atmosphere in unsaturated zones, are gaining
wide acceptance as a tool for delineating subsurface contamination by VOC's (17). A knowledge of the
factors affecting vapor-soil adsorption in unsaturated zones is necessary to accurately describe the diffusive
flux of gaseous pollutants that can move into unsaturated porous media from, for example, contaminated
groundwater, leaking underground storage tanks, or hazardous waste sites. Determination of such vapor
13
-------
phase interactions may have significant implications for effectively detecting or monitoring chemical leaks
from potential sources of contamination.
In saturated soils or aqueous solutions laden with sediment, uptake of nonionic organic compounds has been
correlated to essentially two parameters: 1) degree of compound hydrophobicity (or aqueous solubility) and
2) the fraction of organic carbon or organic matter associated with the soil or sediment (22,14). Isotherm
linearity over a wide range of relative concentration is characteristic in this situation, and it is arguable as
to whether partitioning into soil organic matter or linear adsorption is the mechanism responsible for the
uptake from aqueous systems (3,18). In unsaturated soils, uptake of organic vapors may be controlled by
the mineral fractions and also by competitive effects between water and organic vapors for adsorption sites
on mineral surfaces. Recent experimental studies (2,4) which determined the organic vapor adsorption on
natural soils at various humidities indicate that at least two different uptake mechanisms are involved. For
example, isotherms determined for a natural soil under dry and low humidities are distinctly nonlinear,
however, at high relative humidities the isotherms are linear over a wide range of relative concentration.
Furthermore, isotherms of organic vapor on soil humic acid (i.e., pure humic acids with no mineral fraction
present), a major constituent of soil organic matter, are linear. These findings imply that adsorption of organic
vapors on mineral matter is important for dry soils or soils of relatively low moisture contents. These results
also provide more conclusive evidence of a partitioning process to explain uptake of hydrophobic
compounds in wet soils. Because of the large surface areas provided by soil minerals, markedly greater
sorptive capacities are observed for dry soils compared to that for saturated soils. Specifically, solid-vapor
partition coefficients have been found to be several orders of magnitude greater than corresponding solid-
liquid coefficients (22). This finding suggests that vapor-soil interactions can be of considerable significance
in describing contaminant transport in unsaturated zones.
The purpose of this work was to determine from a series of experiments, effective adsorption parameters of
a selected pollutant vapor, trichloroethylene, for three recompacted natural soils under varying degrees of
water content. Trichloroethylene was selected for this study because it is one of the most commonly found
groundwater contaminants in the U.S. It has been detected in highest concentration and greatest frequency
in several states surveyed (9). It is widely used as a degreaser in the semi-conductor industry. TCE is of
environmental concern because it is a suspected carcinogen.
THEORY
Understanding the effects of moisture on adsorption becomes critical when considering the movement of
gases through natural soils. Diffusive transport equations usually contain a "lumped parameter" called the
Effective Diffusion Coefficient, and this coefficient in turn contains, either explicitly or implicitly, a sorption
term.
Weeks, et al. (1982) described gaseous diffusion theory in a three-phase medium, in one dimension with a
modified form of Pick's second law by replacing the general diffusion coefficient D with an effective
(apparent) diffusion coefficient D':
D'=3C=)C (1)
14
-------
where C is the concentration of vapor (g/cm3), x is distance (cm) and t is time (sec). D' (cm2/sec) can be
described in a non-reactive media by (Weeks et al., 1982):
D'= _ l _ (2)
6D + (eT - 0ol Pw kw + (l - 6T) ps ks
where t = the tortuosity factor accounting for the added resistance to diffusion imposed by the structure of
the porous medium (dimensionless). 0D = drained or gas-filled porosity (dimensionless), 0T = total porosity
(dimensionless), pw = density of water (g/cm3), ps = particle density of granular material making up solid
matrix (g/cm3), kw = liquid-gas partitioning coefficient (often taken as the Henry's Law Coefficient) that
describes the ratio of the concentration of the gas under consideration in solution to its concentration in
overlying gas phase under equilibrium conditions, (moles/gm water-?- moles/cm3 gas), ks = kJCp = gas-liquid-
solid distribution product describing the ratio of the moles of the gas under consideration sorbed on the solid
phase per unit mass or solid phase to the concentration of the gas in the soil atmosphere (moles/g solid -s-
moles/cm3 gas), and kD = solid-liquid distribution coefficient describing the ratio of the moles of solute under
consideration sorbed on the solid phase per unit mass of solid phase to the concentration of the solute in the
water (moles/gm solid •*• moles/gm water).
In the more generalized form of Pick's second law appearing in equation (1), it is assumed that the liquid
phase is immobile and completely wets the solid phase, and rapid (immediate) equilibration occurs between
the gaseous phase and the dissolved and sorbed concentrations in the liquid and solid phases. This also means
that it is assumed the gaseous diffusion through the liquid film prior to sorption on the solid phase is
essentially instantaneous with respect to the overall diffusion process (26). Recent studies have shown
equilibration to take approximately ten minutes for the VOC's to absorb onto soil (12).
Kreamer (1982) analytically solved the three-dimensional form of equation (1) for a point source which
continuously emits a vapor at a constant rate. The solution for a given radial distance from a gaseous source,
and at a known time after vapor emission began is
where C is the concentration of vapor (g/cm3), q is a constant vapor release rate (g/sec), A is the denominator
of D' (dimensionless) called the Sorption-Corrected Porosity, r is the radial distance from the source (cm),
erfc is the complimentary error function, t is time (sec) since release began, and the other parameters are as
defined before.
A graphical curve match procedure for calculation of D' and A from a field tracer test was developed by
Kreamer (1982) and expanded upon by Kreamer et al. (1988). This technique is analogous to Theis curve
matching for the inverse solution of the basic groundwater equation. It provides a means of field testing for
determination of diffusive properties of a given site, just as the Theis equation provided, by means of a
pumping well test, a way to determine aquifer properties. Laboratory procedures, used in this study, were
developed (12), to verify the "lumped parameters" obtained by field tracer tests. Laboratory sorption tests
15
-------
for organic vapor adsorption on unsaturated soils provide a means of calculating the subcomponents of the
Effective Diffusion Coefficient.
The effect of adsorption of organic vapors on soils does not only impact transport equations, adsorption itself
can be modeled. Crittenden et al. (1989) described thermodynamic adsorption models for gas-phase
adsorption equilibria of solvent vapors in unsaturated media. These researchers used the models to predict
single component adsorption capacity, competitive interaction between VOC's and water vapor, and
competitive interaction between individual VOC's. Single component adsorption equilibria was described
using the Dubinin-Radushkevich equation (1947), which is based on the Polanyi potential theory. Crittenden
et al. (1989) extended the Polanyi potential theory to predict multicomponent adsorption equilibria. Ideal
absorbed solution theory has also been used to describe binary and ternary gas-phase adsorption system using
single solute isotherm data (19, 23).
In the experiments associated with this paper, adsorption of a single solvent vapor was measured on natural
soils at high moisture contents. In order to simulate real-world condition, these soils had free water contents
ranging from just under 2% to 20% by weight. Much previous work has been done on competition between
VOC's and water vapor at various humidities in the very dry range of free water content. While this approach
is applicable to a variety of chemical engineering problems, including the adsorption of gaseous compounds
on granular activated carbon, it has poor applicability to gaseous movement through unsaturated natural
soils.
It has been recognized the adsorption of VOC's in the presence of water vapor is unlike single component
or multicomponent (competitive) adsorption of several VOC's. Okazaki et al. (1978) proposed a model to
predict the effect of competition between VOC's and water vapor. The model was based on the idea that
organic gases adsorbed in an unsaturated medium, have three unique components. These components are:
(1) adsorption on "dry" surfaces, (2) absorption on "wet" surfaces, and (3) dissolution into liquid capillary
water. Total adsorptive capacity is then said to be the sum of these three mechanisms (5).
However, for very wet soils the analogy is questionable as VOC loading on "dry" surfaces may be minimal.
In Okazaki et al. (1978), the VOC loading on dry surfaces is determined using the Dubinin-Radushkevich
equation. This equation has the form:
ptl <4)
where Q = concentration of gas adsorbed onto the solid phase (mg/g solid), Wo = the maximum adsorption
space on the adsorbent (cm3/g solid), PJ = the liquid density of the pure adsorbate (g/cm3), B = the
microporosity constant of the adsorbent (mol2/cal2), B=the affinity coefficient of the adsorbate (dimensionless),
R = the gas constant (1.987 cal/mol °K), PQ = saturation vapor pressure of solute (atm), Pf = equilibrium
partial pressure of solute in the gaseous phase (atm), and T = temperature (°K). Dubinin (1965) has shown
that the affinity coefficient, B, and the maximum adsorption space, Wo, are properties of the absorbent (in
this case the soil), and are independent of temperature and adsorbate.
16
-------
Okazaki et al. (1978) were concerned with water vapor competition with organic vapors for adsorption sites.
Any regions of water condensation were treated with a Henry's law approximation of organic vapor
dissolution into the aqueous phase. While in dry adsorbents this factor has sometimes been found to be
negligible (5), the wetted soils of this study were projected to dissolve approximately half the vaporous
adsorbate in many cases (20).
The adsorptive capacity of organic vapors onto "wetted" pore walls is usually estimated from an aqueous
phase isotherm. The total adsorptive capacity is then the sum of: (A) the Dubinin-Radushkevich predicted
loading multiplied by the dry area fraction; (B) the Henry's prediction of the aqueous VOC concentration
times the water volume; and (C) the amount predicted from an aqueous-phase isotherm times the fraction
of wet surface area.
The partitioning of TCE in this study is described by a single partitioning coefficient, Ka, defined at a known
temperature as:
where Ka is the equilibrium soil-vapor adsorption (mg VOC adsorbed/g soil + mg of VOC in gaseous phase/
g of VOC in vapor phase at saturation) and the concentration of adsorbed gas on the solid phase, Q, and the
relative vapor concentration P/Po are as defined previously. The use of this single partitioning coefficient,
does not include the extensive computation (and associated error in estimation) of parameters required in the
more rigorous and descriptive approach by Okazaki et al. (1978).
The use of a single partitioning coefficient, Ka to describe both adsorption onto the solid-phase and
dissolution into liquid capillary water would reduce Equation (2) to
eD + (i-eT)pswka
where p sv is a weighted average density of the combined solid and liquid phases.
EXPERIMENTAL METHOD
Vapor Sorption Apparatus
The method used in this work to determine vapor phase adsorption coefficients was a static adsorption
method developed by Houston et al. (1989) in which the adsorbent was placed in a chamber, the adsorbate
gas introduced, and the chamber isolated to achieve a static equilibria between soil and vapors. The
equilibrium vapor phase concentration in the soil chamber was measured by a direct injection of the vapor
into a gas chromatograph.
17
-------
Measurement of the vapor phase sorption of a selected organic compound onto dry and partially saturated
soils was carried out by a static equilibrium adsorption apparatus at room temperature (Fig 1). The test gas,
a commercially prepared mixture of TCE in nitrogen, was connected by stainless steel tubing (0.635 cm o.d.)
to a series of 34mm x 60mm (inner dimensions) stainless steel solid chambers that housed the sample. The
entire system was evacuated (down to 10"3 Torr) by a vacuum pump connected to the system. The test
compound was brought to contact with the soil sample at a constant pressure, which was controlled by a
regulator at the vapor source. The sorption chambers were then isolated from the system by closing valves
designated V3, V5, etc. to attain a static equilibria between vapors and soil.
Several concentrations of the commercially prepared TCE vapor were obtained, so that when needed, the
concentration of vapor could be varied. To do this, the test gas source was disconnected form the system at
the regulator and replaced by a source with different vapor concentration. Amounts of the organic vapor
adsorbed by the soil samples were determined by the difference between initial and final TCE concentrations
in the gas phase of the soil chamber. Soil-vapor adsorption coefficients were calculated form the resulting
data.
Some of these important physicochemical properties for TCE are listed in Table 1.
Table 1. Physicochemical Properties of TCE
PROPERTY VALUE
Structural Formula C1HC=CC12
Molecular Weight 131.39
Boiling Point 87°C
Vapor Pressure 71 mm Hg"
Solubility HOOmg/lb
Henry's Law Constant 619 atm
Dipole Moment 0.77C
•Value at 24°C from Weast (1986).
"Value at 25°C from Dean (1987).
cValue at 30°C from Dean (1987).
The equilibrium vapor phase was measured by a direct injection procedure. This was accomplished by
allowing the equilibrated gas to expand and fill a Valco 10 port sample valve (connected to the soil chamber
by 0.159 cm o.d. stainless steel tubing) of known volume and temperature directly to a chromatograph
column. Once filled, the vapor was then swept by a flow of nitrogen carrier gas into a gas chromatograph
and analyzed. This sampling methodology has been previously developed and described elsewhere and was
only slightly modified in this study. Changes to the apparatus included the addition of a sample blank used
as a control in the sample analysis, addition of a cold trap to keep the vacuum system clean, and replacement
of filter paper by a metal filter gasket to contain soil particles during evacuations. The gas samples were
analyzed for TCE with a Perkin Elmer Sigma 300 gas chromatograph with a column of 10% SP21-- on 80/
100 Supelcoport (Supelco, Inc.) operated isothermally at 135°C. Both 63Ni electron capture and flame
ionization detectors were used for the analyses.
18
-------
The basic soil chamber design was previously developed (15) and an extensive study on the best choice of
materials to use for these chambers in adsorption measurements was conducted by Houston et. al. (1989).
To ensure containment of soil particles during evacuation of the soil chamber, filter paper was placed
between the gasket and the top of the soil. However, difficulties were encountered with the filter paper
"bulging" as a vacuum was drawn and was especially a problem with soils of finer particles (<50|i). The
problem was remedied by replacing the filter paper with a filter constructed from copper metal and copper
wire mesh.
All pressure measurements are made with either an Ashcroft or Matheson pressure gauge (designated as Gl
in Fig. 1). The valves were stainless steel. All fittings were stainless steel. A liquid nitrogen trap was added
to the vacuum system to reduce contamination from pump oil and grease.
The test compound was derived from a mixture of TCE vapor and nitrogen prepared and analyzed
gravimetrically by Scott Specialty Gases. The TCE concentration employed for this study was a standard
0.0978 ± 0.005 ppm by weight.
The Area 5 Alluvium, Desert Rock, and Clay soils (<40 mesh fraction) used in this study were from the same
batch employed in aprevious sorption study by Houston et. al. (1989). In the previous study, the mineralogic
content of the soils was obtained. Houston et. al. (1989) also estimated the surface areas of these soils. This
was done by computing the ratios of the surface areas of the soils from the ratios of the adsorption coefficients
determined for air-dried soils. In their experiments, adsorption coefficients for the above soils and a sand
were measured. The adsorption coefficient ratio was computed comparing these soils with that for the sand.
Then, from the surface area of the sand, which was estimated assuming a spherical shape, the surface areas
of the soils were computed. A list of all adsorbents used in this study and their estimated surface areas and
soil classifications are presented in Table 2.
A soil compaction procedure was performed on the three natural soils: Area 5 Alluvium, Desert Rock, and
Clay. Also, the water content was varied. The soils were compacted in the chambers to obtain a desired bulk
density to replicate a field condition. Aprocedure developed by Houston et al. (1989) to obtain the partially
saturated soils and bulk densities was followed. The desired volume of deionized water was added and mixed
with the soil prior to compacting the sample in the sorption chamber. A final water content was measured
immediately following each experiment to take into account evaporative losses during compaction of the
soil and evacuation of the sorption chamber. Air dried Areas 5 Alluvium, Desert Rock, and Clay soils
contained about 2.5, 1.5, and 2.4 percent moisture, respectively.
Table 2. Adsorbent Properties
ADSORBENT
MESH ESTIMATED
SIZE SURFACE AREA
(cm2/g)
SOIL
CLASSIFICATION
Area 5 Alluvium
Desert Rock
Clay
<40
<40
<40
620
900"
1360s
Silty Sand
Sandy Loam
Clay
•Houston et. al. (1989)
19
-------
Adsorption Measurements by Direct Injection Technique
The test compound was brought into contact with soil by the following operations (referring to the apparatus
in Fig. 1):
1) The test gas was regulated to a constant absolute pressure of 200 kPa (2.0 atm) according to pressure
gauge, Gl, and then valve VI was closed.
2) The system from valve VI to V8 was evacuated. The sorption chambers were evacuated separately
from the system with flow out of the chamber through valves, V2, V4, etc.
3) Valve V1 was opened which refilled the system with test gas to 200 Kpa of total pressure up to valve
V8. The gas was brought in contact with the soil by opening valve V3 for 15 seconds. A reduced
exposure time was kept to minimize diffusion of vapors into the soil chamber, which may occur as
a result of immediate adsorption. The system was then evacuated and refilled again with test vapor
and the next soil sample was exposed to the gas by opening V5, etc. The adsorbent and vapor were
equilibrated for 12 hours or more. To measure the vapor phase equilibrium TCE concentration in the
soil chamber, the gas was expended by opening valves V3 or V5 etc. into a previously evacuated
sample loop of known temperature and volume and then swept by a nitrogen carrier gas into the gas
chromatograph and analyzed. The pressure drop resulting form the volume expansion was recorded
by pressure gauge, Gl. The pressure drop was about 50 Kpa (0.5 atm) in normal runs. To determine
whether expanding the equilibrated gas into a relatively large volume (40-80 percent of chamber
volume) caused significant desorption, the expansion volume was reduced to a minimum allowed
by the system which resulted in a corresponding pressure drop of 10 kPa (0.1 atm).
Standard calibration curves were established before each set of soil gas analysis by directly injecting a given
concentration of test gas as a function of pressure and plotting the number of moles of test compound versus
peak area. This was done by evacuating the sample loop and filling with test gas at some known pressure
according to gauge Gl and then analyzing the solute at that pressure. The procedure was repeated for
different gas pressures. If the soil gas peak area was not within the range of the curve, a different test vapor
concentration was connected to the system and a new calibration curve was generated. Due to the large
concentration range studied in associated experimentation (6 orders of magnitude), it was necessary to
employ two detectors because it was not practically feasible to sufficiently dilute the gas in order to stay
within the dynamic range of one detector. As a result, lower concentrations (0.0978 and 1.2 ppm) were
analyzed with an electron capture detector and higher concentrations (10.5,109,1070, and 5010 ppm) were
analyzed using a flame ionization detector which has a lower sensitivity for the test compound. Standards
were run in duplicate. During all evacuations, the system was flushed several times with high purity nitrogen
to thoroughly clean sample loop and line. Also, blanks of pure nitrogen were run to ensure the sample loop
and line were clean.
To determine the amount of adsorption on the chambers walls, an empty soil chamber was evacuated, filled
with test gas, and then analyzed following the same procedure as described above.
20
-------
RESULTS AND DISCUSSION
Vapor Sorption on Natural Soils
Equilibrium soil-vapor partition coefficients were determined for the vapor phase uptake of TCE on Area
5 Alluvium, Desert Rock, and Clay soils at various water contents, ranging from air-dried to 20% (g H20/
g soil). For these experiments on natural soils, the direct injection technique (described earlier) was used
to determine the equilibrium TCE concentration in the soil gas. All sorption determinations were made from
a single, initial TCE concentration of 97.8 ppb. The experimental data obtained from these experiments are
tabulated in Oja, 1988. From these data, the equilibrium partition coefficient, Ka, as defined by Equation
(5) was computed. Values of Ka for the soils at each water content are given in Table 3.
The error in the calculation of Ka was estimated from standard error propagation methods. On average, the
standard deviations are about 20% of Ka. The difficulty associated with obtaining and measuring accurate
water contents is manifested by inconsistencies in the void volume, which are greater at higher water
contents. The error in the void volume was evaluated by comparing the calculated void volume, determined
from the difference in the total system volume and the combined solid and water volume with that computed
according to the pressure drop during expansion of the gas. This latter volume was calculated based on an
average value of the expansion volume determined for the dry soils and the pressure volume relationship for
an ideal gas. The average of these two volume determinations was used in computing Ka. The void volume
discrepancies do not affect trends in the partition coefficient, however, they do contribute substantially to
the uncertainty in Ka.
Table 3. Equilibrium Partition Coefficient K, Describing the Vapor Uptake of TCE on Three
Natural Soils as a Function of Water Content.
SOIL WATER CONTENT Ka(MG/G)
(wt%)
Area 5 Alluvium
Air-Dried 1.4
5 0.34
10 0.42
15 0.60
20 0.41
Desert Rock
Air-Dried 4.4
5 0.96
10 1.0
15 0.75
20 045
Clay
Air-Dried 3.2
5 0.71
10 0.60
20 0.45
21
-------
Upon introducing TCE vapor to soil in the sorption chamber, a minimum exposure time was maintained by
closing the valve that controlled the flow into the container after 30 seconds. This was done in order to
prevent additional vapor from entering by diffusion as a result of a concentration gradient. Such a condition
may exist if the adsorption rate is very rapid, however, the amount of TCE that may diffuse into the sorption
chamber after 30 seconds as a result immediate adsorption was calculated to be insignificant compared that
entering by advection.
A significant reduction in the initial TCE gas phase concentration was observed in these experiments,
indicating there is a large uptake by the soil. The data indicate greater than 80% sorption occurring when
the initial moles of TCE are compared to the number of moles adsorbed, suggesting that the number of
adsorbent sites is still very large relative to the amount of vapor present. Considering the very low partial
pressure of TCE (~10~7 atm) used in these experiments, the large change observed in the TCE concentration
might be expected. At these pressures, linear adsorption may be anticipated.
The effect of water content on the vapor sorption of TCE by Area 5 Alluvium, Desert Rock, and Clay soils
at 24°C is illustrated in Fig. 2. In general, the air-dried soils exhibit a greater sorptive capacity for vapor
compared to their moist counterparts. For example, the adsorption partition coefficient for TCE onto air-
dried Area 5 Alluvium, Desert Rock, and Clay soils was found to be 1.4 mg/g, 4.4 mg/g, and 2.4 mg/g,
respectively. Partition coefficients for all moist soils (5-20 percent saturations) were less than or equal to
1 mg/g. Figure 2 shows a plot of Ka as a function of water content for all soils.
The influence of moisture on sorption indicates that water successfully competes for the active adsorption
sites on minerals. This behavior is consistent with the fact that adsorption on minerals is enhanced by
adsorbate polarity (4). Water molecules may effectively displace the less polar TCE. The results are also
consistent with Peterson et. al.(1988) who found that TCE vapor sorption on a synthetic soil is highly
dependent on water content. In their study, the partition coefficient for TCE vapor onto an oven-dried porous
alumina oxide coated with humic acids was found to be 1-2 orders of magnitude greater than that for 8.2%
and 11.6% water contents; the value for 8.2% is 4 times greater than for 11.6%. Results from Figure 2 of
the present study reveal that the air-dried soils possess a distinctly greater sorptive capacity for TCE vapor
than do the bulk of the moist soils. However, the decrease in adsorption with increasing water content
assuming equal distribution is not as pronounced as that found by Peterson et. al. In fact, the variation in
Ka from 5 to 20 percent moisture contents is insignificant, given the error in Ka, and this suggests that uptake
by the moist soils is essentially independent of water content.
The much greater difference in uptake observed by Peterson et. al. between the dry and moist simulated soil
may be explained by the much greater surface area of the soil (206 m2/g) compared to that of soils in this study
(0.136 m2/g). For example, an average of a few monolayers of water were estimated to be present on the
synthetic surface at 8.2% and 11.6% moisture contents. In comparison, several hundred monolayers of water
are estimated to be present on the soils used here for a similar water content (20). These estimates are based
on 11.4 A2 as the surface area occupied by a water molecule (11,22). The latter estimate is also based on the
soil surface areas given in Table 3. If the ratio of the partition coefficients for the air-dried natural soils and
oven-dried simulated soil is compared to the ratio of the above soil surface areas, the surface areas estimated
for the natural soils appear reasonable. A further possibility for the large difference in uptake observed
between the synthetic soil and natural soils may be explained if equilibrium between the natural soils and
22
-------
TCE vapor in the present work was not fully attained. This would result in a lower (relative to that at true
equilibrium) measured value of K^
Gas-liquid partition coefficients for TCE in water were not experimentally determined in this laboratory,
however, the amount of TCE dissolved in water at 25°C was estimated based on Henry's Law
Pf = XfJKH (6)
where Pf is the equilibrium partial pressure of TCE, Xfl is the final mole fraction of TCE in water and K^
is Henry's Law constant equal to 619 atm at 25°C (Table I). The results suggest that a 20 weight percent water
content, about half of the measured uptake may be attributed to dissolution (or adsorption at the water
interface) in water (20). Since this estimate accounts for only a portion of that observed, and Ka appears to
be relatively independent of the degree of saturation above 5% water content, residual uptake may be a result
of partitioning into the soil organic phase or adsorption on mineral matter in the presence of water (site
specific adsorption).
If it is assumed that residual uptake is predominately a result of partition into the soil organic matter at
moisture contents above 5% and that relationships for partitioning in saturated media are analogous to
partitioning relationships in unsaturated media, then the fraction of organic carbon associated with these soils
may be estimated. For example, under the assumption of negligible mineral uptake, in saturated systems,
the soil-liquid partition coefficient, Kj, can be normalized to soil organic carbon (K^. = K/f where f is the
fraction of organic matter) (2, 13). The K^ value has been found to be fairly constant for a particular
adsorbate and valid for a wide range of organic matter content (24). Using as a value for Kt, the average value
of the partition coefficient measured in this study for the soils above 10% water content (assuming that Kg
is linear) and using a K^ value for TCE of 61 units as reported by Garbini and Lion (1986), the percent of
organic carbon contributing to the soil uptake is estimated to be less than or equal to 1% (g organic matter/
g soil) for all three soils. The validity of using specified assumptions to make this calculation, however,
remain unproven.
The uptake capacities of the air-dried soils appear to be controlled by soil minerals. It might be predicted
that the sorptive capacities of a (dry) soil would be different for different soil types largely because of
differences in mineralogy. For example, a clay-rich soil would be expected to have a higher uptake than a
sandy soil. In a study by Call (1957), the vapor sorption of the ethylene dibromide on two different soil types
(0% moisture content) was highly variable largely in response to the clay content. Desert Rock soil contains
17% clay, 19% silt, and 64% sand; Area 5 Alluvium contains 14% clay, 15% silt, and 71% sand; the Clay
soil contains 52% clay, 18% silt, and 30% sand by weight (Houston et al., 1989). The Clay soil has 3 times
more clay minerals than Desert Rock or Area 5 Alluvium, yet Desert Rock soil was found to have the greatest
sorptive capacity for TCE vapor relative to the other soils as Fig. 2 illustrates. Thus, it appears that although
the air-dried uptake is affected by soil minerals, a correlation of clay content and soil uptake cannot be made
from the samples studied here.
CONCLUSIONS
The sorption on natural soils was found to decrease with an increase in water content indicated by noticeably
greater soil-vapor partition coefficients for the air-dried soils compared to moist soils. However, compared
23
-------
to air-dried, above 5% moisture contents, the uptake was observed to be relatively independent of the degree
of saturation indicated by an nearly constant TCE partition coefficient. This effect suggests that above 5%
water contents, the surfaces of these soils are fully covered by water and that residual uptake may be
controlled by soil organic matter (or specific adsorption sites favorable to TCE in the presence of water). In
all cases, a very high uptake was observed by the soils at the low sub-parts per million level of TCE employed.
The greater adsorption of TCE vapor on dry soils suggests that optimal design of vapor leak detection systems
in unsaturated subsurface environments, such as those used around underground storage tanks, should use
models based on dry soils. This is because the greater vapor adsorption exhibited in dry environments would
constitute a conservative estimate of vapor migration. Designs of leak detection systems based on dry soil
would therefore underestimate the effectiveness of gaseous monitoring.
ACKNOWLEDGEMENTS
The authors wish to thank Reynolds Electric and Engineering Company for their help and financial support
of this project. Particular thanks is given to Gene Kendell, Paul Dickman, Eric Williams, Dan McGrath,
Lynn Ebeling, Mark Olson, Robert Straight and Dudley Emer.
REFERENCES
1. Call, E, 1957. The mechanism of sorption of ethyene dibromide on moist soils. J. Sci. Food Agric.,
Nov. 8: 630-639.
2. Chiou, C.T., Kile, D.E. and Malcolm, R.L., 1988. Sorption of vapors of some organic liquids on soil
humic acid and its relation to partitioning of organic compounds in soil organic matter. Environ. Sci.
Technol., 22:298-303.
3. Chiou, C.T., Peters, LJ. and Freed, V.H., 1979. A physical concept of soil-water equilibria for
nonionic compounds. Science, 206:831-832.
4. Chiou, C.T. and Shoup, T.D., 1985. Sorption of organic vapors and effects of humidity on sorptive
mechanism and capacity. Environ. Sci Technol., 19:1196-1200.
5. Crittenden, J.C., Rigg, T.J., Perram, D.L., Tang, S.R. and Hand, D.W., 1989. Predicting gas-phase
adsorption and humidity, J. Environ. Eng. ASCE, 114 (3): 560-572.
6. Dean, J.A., 1987. Handbook of Organic Chemistry. McGraw Hill,NewYork,NY.
7. Dubinin, M.M., 1965. Theory of the bulk saturation of microporous activated charcoals during
adsorption of gases and vapors. Russ. J. Physic. Chem., 39(6):697-704.
8. Dubinin, M.M. and Radushkevich, L.V., 1947. Dokl. Atad. Nauk. SSSR. 55:331-338 (in Russian).
9. Federal Register, 1984. National primary drinking water regulations: Volatile synthetic organic
compounds. 49:24330.
24
-------
10. Garbini, D.R. and Lion,L.W, 1986. Influence of the nature of soil organics on the sorption of toluene
and trichloroethylene. Environ.Sci.Technol., 20:1263-1269.
1 1 . Himenz, P.C., 198 1 . Principles of Colloid and Surface Chemistry, Dekker, New York.
12. Houston, S.L., Kreamer, D.K. and Marwig, R., 1989. A batch-type testing method for determination
of adsorption of gaseous compounds on partially saturated soils. ASTM. Geotechnical Testing
Journal, 12(1):3-10.
13. Karickhoff, S.W., 1984. Organic pollutant sorption in aquatic systems. J.of Hydrau. Engineer,
110:707-735.
14. Karickhoff, S.W., Brown, D.S. and Scott.T.A., 1979. Sorption of hydrophobic pollutants on natural
sediments. Water Res.,13:241-248.
15. Kreamer, D.K., 1982. Insitu measurements of gas diffusion characteristics in unsaturated porous
media by means of tracer experiments. Ph.D. dissertation, University of Arizona.
16. Kreamer, D.K., Weeks, E.P. and Thompson, G.M., 1988. A field technique to measure the tortuosity
and sorption-affected porosity for gaseous diffusion of materials in the unsaturated zone with
experimental results from near Barnwell, South Carolina. Water Res. Research, 24(3):331-341.
17. Martin, D.L., and Kerfoot, H.B., 1988. Soil-gas surveying techniques. Environ.Sci. Technol.
22(7):740-745.
18. Mingelgrin, U. and Gerstl, Z., 1983. Reevaluation of partitioning as a mechanism of nonionic
chemical adsorption in soils. J. of Environ. Qual., 12:1-11.
19. Myers and Prausnitz, 1965. Thermodynamics of missed gas adsorption. Am. Inst. Chem Eng.J.,
20. Oja, K.J., 1988. Vapor phase sorption of trichloroethylene on quartz and partially saturated soils.
Master's thesis, Arizona State University.
21. Okazaki,M.,Tamon,H. andToei.R., 197 8. Prediction of binary adsorption equilibria of solvent and
water vapor activated carbon. J. Chem. Eng. Jpn. 11(3):209-215.
22. Peterson, M.S., Lion, L.W. and Shoemaker, C.A., 1988. Influence of vapor-phase sorption and
diffusion of the fate of trichloroethylene in an unsaturated aquifer system. Environ. Sci. Technol.,
22:571-577.
23. Rasmuson, A.C., 1984. Adsorption equilibria on activated carbon of mixtures of solvent vapors.
Fundamentals of adsorption, proceedings of Engineering foundation conference. United Engineering
Trustees, Inc., NewYork, NY., 451-470.
25
-------
24. Voice, T.C. and Weber, W.J., 1983. Sorption of hydrophobic compounds by sediments, soils and
suspended solids, Water Res., 17:1433-1444.
25. Weast, R.C., 1988. Handbook of Chemistry and Physics, 67th ed. CRC Press, Cleveland.
26. Weeks, G.P., Earp, D.E., and Thompson, G.M., 1982. Use of atmospheric fluorocarbons F-12 and
F-13 to determine the diffusion parameters of the unsaturated zone in the southern high plains of
Texas. Wat. Res. Research 19(5): 1365-1378.
26
-------
Regulator
Vent
^ Ultra High
Purity Nitrogen
Pump
T' n»
vi o valv»
i [I w
"xllj
Cold Trap
^ V1
\t*l
\t e
"9
V7 '
Sample VaJve
(Gas
^hromatograph)
SorpDon Chambers
Test Vapor
Source
Figure 1. Schematic diagram of the static vapor sorption apparatus.
-------
CC
6
5
4
3
2
1
0
0
•
• Area 5 Alluvium
a Desert Rock
o Clay
5 10 15
Percent Water (gH20/gsoil)
20
Figure 2. Plot of the soil-vapor partition coefficients for TCE on Area 5 Alluvium, Desert
Rock, and Clay soils as a function of water content
28
-------
USE OF IN-SITU OZONATION
FOR THE REMOVAL OF VOCS AND PAHS FROM UNSATURATED SOILS
SUSAN J.MASTEN
Department of Civil and Environmental Engineering
Michigan State University
East Lansing, MI 48824-1326
The contamination of soils has resulted in a complex problem which necessitates novel engineering
approaches. In this project we have investigated the applicability of ozone for the oxidation of volatile
organic chemicals (VOCs) and polycyclic aromatic hydrocarbons (PAHs) when these chemicals are present
in aqueous solutions containing humic acid or in soils.
The oxidation of several PAHs in the presence of humic acid and soils is described. The oxidation of four
olefinic VOCs and naphthalene occurred in solutions containing up to 12.0 mg/L humic acid, however, the
extent to which each of the compounds reacted is very much compound specific. The effect of pH and ozone
dosage on these reactions was considered. The effects of pH were weak for all compounds except
trichloroethane. Ozone dosage had a significant effect on the extent to which each of the VOCs and
naphthalene was oxidized.
Experiments were also conducted by adding aqueous ozone solutions to soil slurries containing the
compound of interest. Complete oxidation of cjs-dichloroethylene by ozone (22 mg/L) occurred in solutions
containing 1.0 g of Eustis soil suspended in 10.0 mL water. However, only 40% oxidation of tetrachloroethylene
was achieved using an ozone dosage of 17.3 mg/L when 1.0 g of Eustis soil was present. Complete oxidation
of naphthalene was achieved in solutions of a sandy loam soil (0.4% organic matter) containing 1.0 g of soil
suspended in 10.0 mL water. Similar results were also obtained in slurries of Ottawa sand and alumina oxide.
The oxidation of PAHs by ozone in dry soils was studied. Complete removal of phenanthrene from a soil
using an ozone dosage of 80 mg/hr for one hour was observed. Chry sene and pyrene were removed to a lesser
extent. The work described here suggests that ozone may be applicable for the treatment of contaminated
soils and waters containing naturally occurring organic matter.
INTRODUCTION
Vapor extraction (also referred to as in situ air stripping and vacuum extraction) techniques have been
developed for the removal of vapor phase contaminants from an unsaturated aquifer (1-5). In order for this
technique to be effective, the contaminant must be volatile and not strongly sorbed by the soil particles. With
conventional treatment systems, remedial activities can result in gaseous effluents which have concentrations
of volatile organic chemicals (VOCs) that are well above that which can be safely discharged into the
29
-------
atmosphere. While activated carbon units can be added to treat the off-gases, this additional treatment
approximately doubles the cost of remediation (6).
To alleviate some of the problems of the conventional system, a modified vapor extraction system has been
proposed and is illustrated in Figure 1. The principal design features of conventional vapor extraction system
are maintained. However, the major change is that instead of pumping air into the unsaturated zone, ozonated
air or oxygen would be used. This would require the installation of an ozone generation system as shown in
this figure. Despite the expense involved in the generation of ozone, such a system would have some definite
benefits. Ozone, being a strong oxidant, would react with most olefinic and aromatic compounds which
might be present. This should alleviate the need for post-treatment of the off-gases, or if treatment were
necessary, the excess ozone generated could be used. Additionally, in transport-limited situations, where
pump-andtreat technologies might not be feasible, the need to remove the pollutants from the soil system
would be negated as the ozone would react with compounds in the subsurface environment. Finally, while
capital costs for the ozone generation system are high, since the generation system would outlast the life of
the remediation project, such a system could be transported from site to site. This would significantly reduce
the costs of using this technique for remediation.
THEORY
Among the types of compounds that could be treated by this modified vacuum extraction system are the
chlorinated olefins, such as cis-dichloroethylene, trichloroethylene and tetrachloroethylene, the chlorinated
aromatics such as chlorobenzene and o-chlorophenol, and the polycyclic aromatic hydrocarbons, such as
pyrene and chrysene. The results of work on the treatment of soils contaminated with chlorinated olefins and
aliphatic compounds (VOCs), and polycyclic aromatic hydrocarbons are presented here. The chlorinated
compounds are commonly used solvents while the polycyclic aromatic hydrocarbons (PAHs) are components
found in petroleum products. Both groups of compounds have been identified in many aquifers and in the
unsaturated soils in a variety of locations throughout the U.S. (7). The compounds studied are listed in Table
1.
The mechanism of the ozonolysis reaction involving olefins and aromatic compounds has been well studied
and is documented in a number of review articles (8-11). This mechanism, also referred to as the Criegee
Mechanism, involves a 1,3-dipolar cyclic addition of ozone across the double bond of the olefinic or aromatic
compound to yield an unstable intermediate, known as a molozonide. These intermediates rapidly
decompose to form aldehydes, ketones and carboxylic acids. In the case of the highly chlorinated olefinic
compounds, chlorinated carboxylic acids or ketones may result. These compounds are rapidly hydrolyzed
in water to form carbon dioxide and HC1 (12). Dreher and Klamberg (1988) studied the ozonation of PAHs.
Among the products identified were phthalic acids, carboxylic acids and methyl esters.
For volatile chemicals, the gas phase reactions involving ozone must be considered. However, since much
work has been done in this area, and extensive reviews are available (see for example, Atkinson and Carter
(1984)) these reactions were not studied. The availability of gas phase rate constants for the reaction of ozone
and VOCs permits the prediction of the efficiency of this process. Additionally, as these reactions would be
much more rapid than the oxidation of the same compounds in water, gas-phase reactions would not control
the removal rates or efficiencies of VOCs from unsaturated soils.
30
-------
In the system described here, the decomposition of ozone must be considered since the concentration of
ozone present at any time, t, will affect the decomposition of the particular VOC. If the oxidation of the VOC
occurs by a direct (Criegee) mechanism, then an increase in the ozone decomposition rate will result in a
decrease in the extent to which the organic chemical of interest reacts.
Understanding the mechanism of ozone decomposition is important from the standpoint of knowing what
oxidizing species are formed. A number of researchers (15-26) have studied these reactions and recently
Staehelin and Hoigne (1985) developed a schematic for the decomposition of ozone in "pure" water. Based
upon this schematic it is clear that the most important oxidizing species present in solution are ozone, OH
radicals and superoxide. Both superoxide and OH radicals enter into the cyclic reaction which then enhances
the decomposition of ozone. Additionally, hydroxide will react with ozone to form superoxide and, as such,
will result in an enhancement of the rate of ozone decomposition.
An understanding of the reactions involving ozone that may occur in the presence of a solute is important
in terms of designing and optimizing such a treatment system. As is the case in "pure" water, ozone,
superoxide and OH radicals are present as three of the oxidizing species, termed promoters, which can react
with OH to form carboxyl radicals. The carboxyl radicals react with oxygen to form organoperoxides.
Staehelin and Hoigne (1985) found that the naturally occurring matter obtained from Greifen Lake and from
the Dubendorf ground-water, acted as a chain promoter. In contrast, other species may react with OH radicals
in such a way that they do not enter into the cyclic reaction. The production of OH radicals in such systems
is significant since OH radicals are much more powerful and less selective oxidants than ozone, and as such,
their formation may result in the oxidation of compounds that would notreact appreciably with ozone. Since
solutes may act as either chain promoters or scavengers; in such a complex mixture as soil, the rate of ozone
decomposition cannot be predicted, but must be measured.
Figure 2 illustrates some of the reactions that would occur in a soil. These reactions include transfers of
chemical from one phase to another, for example, if free product is present, it may volatilize. Gas phase
oxidation reactions, as discussed previously, would occur; as would the oxidation of the solubilized natural
organic matter with dissolved ozone. Additionally, the reaction of ozone with organic matter and the organic
chemical that is sorbed onto the soil surface may play a significant role in ozone decomposition and in VOC
oxidation. Clearly such a system is extremely complex and would require a great deal of work to completely
elucidate the kinetics and mechanisms of these reactions. Since it was felt that the gas-phase reactions
involving ozone and the VOCs were well understood and not rate limiting, no gas-phase work was planned
in the study. The PAHs studied are sufficiently non-volatile that the gas-phase reactions are unimportant. In
an attempt to simplify the system, most of the work conducted with VOCs was performed in homogeneous
aqueous systems containing humic acid as a model of naturally occurring organic matter. Limited work was
conducted in heterogenous systems containing either Eustis soil or a simulated soil (28). The work conducted
with PAHs was done by treating contaminated soils with gaseous ozone.
EXPERIMENTAL PROCEDURES
Materials. Unless otherwise specified, all solutions were prepared from water which had been treated using
reverse osmosis (RO water). The RO water was boiled for 20 minutes to remove any volatile contaminants.
Ozone was prepared from high purity oxygen (>99.99%) which had been dried through a molecular sieve
trap (Molecular Sieve 5A, Supelco, Bellefonte, PA). A Polymetrics Ozone Generator (Model T-408, San
31
-------
Jose, CA) was used. The VOCs and PAHs used in these experiments were obtained from either Aldrich
Chemical Co. (Milwaukee, WI) or from Chem Service Inc. (West Chester, PA) and were at least 95% pure,
although most were >99% pure. Humic acid was obtained from Aldrich Chemical Co. as the sodium salt.
The simulated soil was prepared by the method described by Garbarini and Lion (1985). The TOC of the soil
was 0.70%. Prior to use, the simulated soil was dried overnight at 105°C. The other soil used in the VOC
study was the <250 |im fraction of a surface soil from Florida, Eustis sand, (Typic Quartzipsamments). Since
the moisture content of the air-dried soil was found to be <0.5%, no additional drying was required. The TOC
of the Eustis soil was 0.66%. The physical and chemical characteristics of this soil are presented elsewhere
(29). The soil used in the PAH experiments was a sandy soil containing 0.5% organic matter. The
characteristics of this soil are presented in Table 2.
Methods. Quantification of the ozone concentrations in the stock solutions of aqueous ozone were
determined spectrophotometrically at 258 nm, the peak maximum for ozone. An extinction coefficient of
3000 M"1 cm"1 (30) was used to convert the absorbance units to concentration.
To prepare aqueous ozone solutions, gaseous ozone (approximately 4% in oxygen) was bubbled into water
at a flow rate of approximately 1 mL/min. The water had been previously acidified to a pH of 2 (or 3,
depending on the experiment) and was chilled to approximately 4°C. During the bubbling of the gas, the
ozone solution was kept chilled on ice. After one hour, the system was shut down and the ozone solution
(approximately 30-35 mg/L) was ready for use.
Solutions of Aldrich humic acid (sodium salt) were prepared by dissolving the humic acid in water followed
by filtration through a O.45 |iM filter. Stock solutions were stored in the dark. A10 mg/L solution of Aldrich
humic acid had an absorbance of 0.227 at 258 nm using a 1 cm path length.
Stock solutions of VOCs were prepared according to the method described by Masten (1991). The stock
solution was kept at 4°C during storage and on ice while dispensing the solution into the samples. Stock
solutions were kept for no longer than 10 days. Naphthalene was prepared by dissolving the solid in
methanol, then, diluting the stock solution in distilled water (pH 3) to 10~4 M.
The soil samples used to study the reaction of ozone with VOCs were prepared according to Masten (1991).
The PAH contaminated soils were prepared in the following manner. The soils were air - dried, then sieved
through a 0.5 mm sieve. 50 mg of the target PAHs were dissolved in <35 ml methanol. A known volume of
PAH solution was added to a known amount of soil. The mixture was shaken by hand. The solution was added
until the total soil weight was 500.0 g and all of PAH solution was added. After rotating the soil sample at
30 rev/min for 1 hour, the soil containing 100 |ig/kg of PAHs was ready for use.
VOC concentrations were determined in the samples using a gas chromatograph equipped with an FID (31).
PAH concentrations were determined using direct injection onto a gas chromatograph equipped with an FID
(32).
Procedure. Details of the experimental procedure for the VOC experiments are presented elsewhere (31).
Experiments with naphthalene were conducted in a similar manner, except that the naphthalene was first
dissolved in a 0.66% methanol/water mixture, then diluted into water to achieve the desired concentration.
32
-------
Glass columns (21.2 mm I.D., 25 cm length) were packed with the contaminated soil. The treatment system
was set up as shown in Figure 3. Gaseous ozone (produced at a concentration of approximately 3% in oxygen)
was passed through the soil. Flows were measured using a rotameter calibrated using a bubble flow meter.
The concentration of ozone generated was monitored by trapping the ozone into potassium iodide traps (33).
After ozone passed through the column for a period of time at a certain flow rate (depending on experimental
design), the soil was removed from the column and well mixed. The concentration of PAHs remaining in the
soil was tested by conventional soxhlet extraction techniques (34, 35) followed by GC analysis.
RESULTS AND DISCUSSION
Effect of ozone dosage. The ozonation of cjs-dichloroethylene (c_-DCE) was studied at two ozone dosages.
As expected, increasing the ozone dose resulted in an increase in the extent to which the £-DCE reacted as
indicated in Figure 4. At 300 mg/L humic acid, a 2.6 fold increase in the ozone dosage resulted in almost a
5 fold increase in the amount of compound which reacted. If the disappearance of C.-DCE occurred purely
by a direct reaction with ozone, than one would expect a 1:1 stoichiometry. The complexity of the reactions
suggests that secondary oxidants are important and that £-DCE is able to successfully compete with the
humic acid for these oxidants. Similar results were observed with the other VOCs (31).
Effect of pH. The effect of pH can be observed by comparing results obtained for trans-dichloroethylene (i-
DCE) and tetrachloroethylene (PCE). t-DCE will react by both the direct and indirect reaction mechanisms
while PCE does not react appreciably with ozone, but only with OH radicals. With i-DCE, at pH 3 (see Figure
5), there appears to be a significant portion of the DCE which reacts by the direct mechanism. Increasing the
pH from 3 to 6 had a much greater effect then did raising the pH from 6 to 8. Increasing the pH of a 0.05 M
phosphate buffer solution (measured at 25 C) from 4 to 6 results in a 7-fold decrease in the ozone half-life;
whereas increasing the pH from 6 to 8, resulted in a 15-fold decrease in the half-life (36). Based upon this,
one would expect that increasing pH from 6 to 8 would have a greater effect on the extent to which J-DCE
is oxidized, since the rate at which ozone decomposes "autocatalytically" changes most dramatically in this
pH range. The fact that this is not observed is further evidence that some mechanism other than the direct
reaction involving ozone and the VOC is occurring.
As is illustrated in Figure 6, the effect of pH appears to be much less with tetrachloroethylene (PCE) than
observed with l-DCE. Since the rate constant for the reaction between ozone and PCE is over than three
orders of magnitude greater than the rate constant for the ozone - t-DCE reaction, one would expect that the
extent to which PCE reacts would be much less than that for j-DCE, in the presence of similar concentrations
of humic acid. Thus, the it can be postulated that the PCE which reacted, reacted via an indirect mechanism,
most probably involving OH radicals. Similar results were obtained with the other VOCs studied (31).
Oxidation of VQCs and Naphthalene in Soil Slurries. The oxidation of PCE and cis-DCE was studied in
Eustis and simulated soil. Neither PCE nor cis-DCE sorbed onto the soils. This is illustrated in Figures 7 and
8. When no ozone was added, the recovery of cis-DCE was 101 + 3.7%, while the recovery of PCE was 100
+ 3.6%. The Eustis soils were treated with 20.9 mg/Lozone while the simulated soils were treated with 23.1
mg/L ozone. In both cases, almost complete oxidation of the cis-DCE occurred with up to 1.0 g soil per 10.0
mL (total volume) of water. Thus, in the systems containing 300 mg TOC/L water, complete oxidation of
cis-DCE occurred. Alternatively written, 0.22 mg ozone added to 1 g of soil (0.66% TOC by weight) would
result in complete oxidation of cis-DCE.
33
-------
The oxidation of PCE was studied in both buffered and unbuffered Eustis soil. The extent of reaction of PCE
was less than that of cis-DCE. Nevertheless, at pH 6.6 the application of 0.17 mg of ozone did result in
approximately 40% decomposition of PCE in the presence of 1.0 g of soil. As in the case for the humic acid,
at low organic concentrations, pH did not appear to effect the extent of oxidation. It is interesting to note
though, that at higher organic contents, the extent to which the PCE reacted (at the pH's studied) varied
somewhat. It is also interesting to note that in the simulated soil system, increasing the pH resulted in a slight
decrease in the extent to which the PCE reacted. On the contrary, in the Eustis soil system, a greater extent
of reaction occurred in the system buffered at pH 6.6 than at the lower pH. Results such as these are indicative
of the complexity of such systems, and may imply that natural soils are less capable of terminating the radical
chain reaction than is Aldrich humic acid or that the reactions are surface-catalyzed.
Effect of OH radical scavengers. Several experiments were conducted to better understand the mechanism
by which the oxidations of the target compounds occurred. To determine the extent to which PCE and
naphthalene reacted via a free radical mechanism, Jen-butyl alcohol (0.5 mM) was added as a scavenger of
OH radicals. If the direct reaction predominated than the addition of ten-butyl alcohol should have no effect
on the extent to which these compounds reacted. The addition of Jen-butyl alcohol reduced the extent to
which target compound was degraded. From Figure 9, it can be observed that approximately 30% of the PCE
was oxidized even at a humic acid concentration of 100 mg/L. This suggests that some of the target compound
which reacts, does so by the direct reaction with ozone; or that Jen-butyl alcohol does not completely
scavenge the OH or other radicals. The amount of PCE that degraded when the OH radical mechanism was
insignificant was also much higher than predicted using rate constants available in the literature. As shown
in Figure lOa, the extent to which naphthalene reacted with ozone was reduced approximately 27%. The
difference in the amount of naphthalene which reacted in the presence and absence of Jen-butyl alcohol vs.
the initial humic acid concentration added is plotted in Figure 1 Ob. The linear relationship observed suggests
that the importance of the indirect mechanism over the direct mechanism increases with increasing humic
acid concentration.
Oxidation of PAHs Sorbed onto Soils. The oxidation of PAHs in soils was studied by passing gaseous ozone
through a soil column which had been packed with a sandy soil contaminated with the target PAHs. During
each experiment, the ozone flux was measured and calculated by two different methods: (1) based on gas
being 3.5% ozone in oxygen, and (2) based on empirical results using KI traps. Several experiments were
conducted during the study to ensure that the gas was 3.5% ozone in oxygen. The discrepancies in the results
are most likely due to poor pH control of the KI traps. It is important to maintain a pH of less than 3 in the
KI solutions and this was not always done. As a result, the ozone fluxes which were calculated based on the
empirical results were less than the ozone fluxes which were calculated based on the ozone being 3.5% in
oxygen. The following discussion is based on the ozone flux as calculated based upon the ozone gas being
3.5% in oxygen.
The experimental results for phenanthrene are presented in Table 3. Before treating the soil using ozone, a
control experiment was performed by passing air through a phenanthrene contaminated soil at a flow rate
of 2.64 IVhr for 7 hours. No phenanthrene was removed by the air. After passing ozone through the soil at
a flux of 253 mg O/hr for 2.3 hours, greater than 95% of the phenanthrene had disappeared. The final
concentration of phenanthrene in the soil was below detection limits. The gas chromatographic scans
revealed no degradation products. When the ozone dosage was reduced to a flux of 151 mg O^r and the
soil was treated for one hour, 60-84% removal was still obtained. It was also observed that 5.8 mg of ozone
degraded 1 mg of phenanthrene.
34
-------
The results of the ozonation experiments conducted on soils contaminated with chrysene and pyrene are
presented in Table 4. In this case, less than 20% removal of chrysene and pyrene occurred with an ozone
dosage of 177 mg 0,/hr for one hour. It appears that when the target compounds are present together, as they
were in this studies with pyrene and chrysene, the ozone dosage must be increased in order to achieve the
same levels of removal as that obtained when phenanthrene was the soil contaminant present in the soil.
When the ozone dosage was increased to a flux of 217 mg Og/hr and the soil was treated for one hour, 34%
removal of pyrene was obtained while no removal of chrysene was found. Increasing the ozone dosage to
549 mg Oj/hr for one hour, resulted in no removal of chrysene while 94% removal of pyrene was obtained.
These findings seem to suggest the importance of the direct ozone reaction in the dry soil, since the results
follow the order of reactivity of chrysene and pyrene with ozone (11).
The experimental results for chrysene oxidation are presented in Table 5. After passing ozone through the
soil at a flux of 500 mg O3/hr for one hour, the total ozone dosage was approximately that used when the
phenanthrene contaminated soil had been treated with a flow rate of 253 mg O3/hr for 2.3 hours. Only 40%
removal of chrysene was achieved as compared to greater than 95% removal of the phenanthrene (when
either PAH was present as the sole contaminant). When the ozone dosage was increased to a flux of 594 mg
Oj/hr and the soil was treated for four hours, only 50% removal of chrysene was achieved. Thus, chrysene
requires more than eight times the ozone dosage of phenanthrene. These results follow the order of reactivity
for chrysene and phenanthrene with ozone (11). The gas chromatographic scans showed peaks for what
appears to be two major degradation products.
The results for experiments with pyrene-contaminated soil are presented in Table 6. When these results are
compared to those obtained with chrysene, it is observed that pyrene is much more efficiently oxidized. For
example, the passage of ozone through the soil at a flux of 26.3 mg Oj/hr for one hour, resulted in 53%
removal of pyrene while the treatment of a chrysene contaminated soil at flow rate of 253 mg 0.,/hr for 2.3
hours resulted in only 40% removal of the chrysene. However, pyrene is not as efficiently oxidized as is
phenanthrene. For example, when the ozone dosage was 581 mg 0.,/hr and the soil was treated for one hour,
only 83% removal of pyrene was achieved, compared to greater than 95% removal of phenanthrene. These
inconsistencies with the order of reactivity as indicated by Bailey (1982) may because the soil absorbed
pyrene much strongly than it absorbed phenanthrene. The partition coefficients (log KJ for phenanthrene
and pyrene are 6.12 and 6.51, respectively (37). Thus, a higher ozone dosage is required for treated pyrene.
The gas chromatographic scans of the pyrene treated soils revealed no degradation products.
CONCLUSIONS
This work indicates the feasibility of using ozone as an oxidant in solutions containing high concentrations
of humic acid or soils. It is important to note that in both situations the oxidation reactions appear to occur
predominately by the indirect reaction involving some oxidant other than ozone. The effects of the addition
of ten-butyl alcohol support this hypothesis.
The compounds studied all reacted to a much greater extent than one would predict assuming that only the
direct ozone reaction occurred. Both £-DCE and PCE reacted to a much greater extent in soils slurries than
in aqueous solutions containing humic acid. There is very strong evidence that the indirect reaction involving
some free radical, such as the OH radical, plays a significant role in the oxidation of these compounds. This
finding is extremely significant since it suggests that iniSim treatment of contaminated soils may be feasible,
35
-------
despite the highly reactive nature of humic acid or naturally occurring soil organic matter with ozone. It
appears that the organic matter acts, essentially, as a reservoir of OH radicals which can then react with the
olefinic anthropogenic compounds found in the soil.
From the results of the soil column experiments, it is apparent that the removal efficiency of the PAH of
interest depends on the reactivity of the PAH. Based upon the results presented here, it appears reactivity
decreases with increasing bond-localization energy (11) and with increasing partition coefficients
ACKNOWLEDGEMENTS
The authors would like to acknowledge NSI Analytical Services for their assistance. Jehng-Jyun Yao, Donal
Brady and Jim Day are appreciated for their assistance with the work on PAHs.
The research described here has been supported by the U.S. Environmental Protection Agency through
contract 68-C8-0025 and through the REF Biotechnology Program (State of Michigan).
LITERATURE CITED
1. Bennedsen, M.B. Pollution Engineering 1987,2, 66-68.
2. Crow, W.L.; Anderson, E.P. Groundwater Monitoring Review 1987,1(4}. 51-57.
3. Hydro-Geo Chem, Inc. (1987) Conceptual level design and feasibility study for in-situ air stripping
of volatile organic contaminants from the unsaturated zone at the Seymour Recycling Corp.
Hazardous Waste Site, Seymour, Indiana. Prepared for Geraghty and Miller, Inc. Plainview, NY, 27
pages.
4. Jafek, B. Waste Age 1986,10,66-67.
5. Thorton, J.S.; Wootan, W.L. J. Environ. Sci. Health 1982, A 17(1). 31-44.
6. Hinchee, R.E.; Downey, D.C.; Coleman, E. J. Jn: Proceedings of the 1987 Conference on Petroleum
Hydrocarbons and Organic Chemicals in Ground Water-Prevention, Detection and Restoration.
November 1987, Houston, TX. 17 pp.
7. Patrick, R.; Ford, E.; Quarles, J. Groundwater Contamination In The United States. 2nd ed,
University of Pennsylvania Press: Philadelphia, PA, 1987, pp. 145-152.
8. Murray, R.W. Ace. Chem. Res. 1968,1,313-320.
9. Bailey, P.S. In: Ozone in Water and Wastewater Treatment; Evans, F.L., Ed.; Ann Arbor Science: Ann
Arbor, MI, 1972; pp. 29-59.
36
-------
10. Gilbert, E. In: Ozone/Chlorine Dioxide Oxidation Products of Organic Matter, Rice, R.G., Cotrovo,
J.A., Eds.; Ozone Press International: Cincinnati, OH, 1976; pp. 227-242.
11. Bailey. P.S. Ozonation in Organic Chemistry. Vol. II Non-olefinic Compounds. Academic Press:
New York, 1982.
12. Masten, S. J. Ph.D. Thesis, Harvard University, Cambridge, MA, 1986.
13. Dreher, W. and Klamberg,_H. Fresenius Z. Anal. Chem. 1988, 331. 290-294.
14. Atkinson, R.; Carter, W.P.L. Chem. Rev. 1984, £4,437-470.
15. Sennewald, K. Zeitschrift fur Phys. Chem. 1933,164. 305-317.
16. Weiss, J. Trans. Farad. Soc. 1935,3_1, 668-681.
17. Taube, H.; Bray, W.C. J. Am. Chem. Soc. 1940, £2, 3357-3373.
18. Taube, H. J. Am. Chem. Soc. 1942, £4, 2468-2474.
19. Alder, M.G.; Hill, G.R. J. Am. Chem. Soc. 1950,22,1884-1886.
20. Kilpatrick, M.L.; Herrick, C.C.; Kilpatrick, M. J. Am. Chem. Soc. 1956,7_&, 1784-1790.
21. Hewes, C.G.; Davidson, R.R. AIChE J. 1971, H, 141-147.
22. Staehelin, J.; Hoigne, J. Env. Sci. Technol. 1982,& 676-681.
23. Bhattacharyya, P.K.; Saini, R.D. Chem. Phvs. Letters. 1982,22, 560-563.
24. Buehler, R.; Staehelin, J.; Hoigne, J. J. Phys. Chem. 1984, £&, 2560-2564.
25. Peleg, M. Water Res. 1976, 10_, 361-365.
26. Gurol, M.D.; Singer, PC. Env. Sci. Technol. 1982, !£, 377-38 3.
27. Staehelin, J.; Hoigne, J. Env. Sci. Technol. 1985,12, 1206-1213.
28. Garbarini, D.R.; Lion, L.W. Env. Sci. Technol. 1985,12, 1122-1128.
29. Bouchard,D.C.;Wood.A.L.;Campbell,M.L.:Nkedi-Kizza.P.:Rao.P.S.C.J.Contam.Hvdrol.l988.
2,209-223.
30. Nowell, J.H.; Hoigne, J. In: the Proceedings of the 8th World Congress of the International Ozone
Association, Zurich, Switzerland, Sept. 14-16,1987.
37
-------
31. Hasten, SJ. Ozone: Sci. Eng. 1991, H 287-312.
32. Yao, JJ. Ozonation of Soils Contaminated with Polycyclic Aromatic Hydrocarbons. M.S. thesis,
Michigan State University, East Lansing, MI, Aug. 1991
33. "Standard Method For The Examination Of Water And Wastewater", APHA, AWWA, WPCF, 16th
ed. (1985).
34. "Method 8100 - Polynuclear Aromatic Hydrocarbons", EPA Method Report, Rev. O, Sep. (1986).
35. "Method 3540 - Soxhlet Extraction", EPA Method report, Rev. 1, Dec. (1987).
36. Hoigne, J.; Bader, H. Water Res. 1976,10_, 377-386.
37. Kayal, S.I.; Connell, D.W. Aust. J. Mar. Freshwater Res. 1990,41., 443-56.
38
-------
Table 1. Organic Chemicals Studied
VOCs
PAHs
1,1-trichloroethane (TCA)
trichloroethylene (TCE)
tetrachloroethylene (PCE)
cis-dichloroethylene (c-DCE)
trans-dichloroethylene (t-DCE)
naphthalene
phenanthrene
chrysene
pyrene
Table 2. Soil Characteristics
Soil Characteristic
Results
Organic Matter
Content
pH
Lime index
Sand
Silt content
Clay content
Residual PAHs
Specific gravity
0.5%
6.6
71.0
82.3%
10.0%
7.7%
negligible
2.55
39
-------
Table 3. Treatment of Phenanthrene In Soil Columns
(Initial Concentration = 100 ng PH/g soil)
Ozone Flux*
(mg/hr)
(1) (2)
0.00
940
882
253
170
151
0.00
879
1010
111
73.0
78.4
Gas
Flow
Rate
(L/hr)
2.64
13.6
12.8
3.66
2.46
2.35
Run
Time
(hr)
7.0
8.0
6.0
2.3
2.5
1.0
%
Removal
[-]
>95%
>95%
>95%
87%
*60%
** 84%
Phenanthrene Recovery Rate From Soil: 78.8%
* 12 extracts analyzed
** 6 extracts analyzed
1 Ozone Flux (1): flux based on 3.5% ozone in oxygen
(2): flux based on KI trap measurements
40
-------
Table 4. Treatment of Chrysene And Pyrene In Soil Columns
(Initial Concentration = 100 ng PY+100 fig C/g soil)
Ozone Flux"
(mg/hr)
(1)
549
217
177
(2)
78.4
121
107
Gas
Flow
Rate
(L/hr)
8.53
3.25
2.56
Run
Time
(hr)
1.0
1.0
1.0
% Removal
PY
94%
34%
3%
C
[-1
H
18%
PY Recovery Rate From Soil: 63.3%
C Recovery Rate From Soil: 80.1%
* Ozone Flux (1): flux based on 3.5% ozone in oxygen
(2): flux based on KI trap measurements
41
-------
Table 5. Treatment of Chrysene In Soil Columns
(Initial Concentration = 100 |ig Chr/g soil)
Ozone Flux8
(mg/hr)
(1) (2)
594
558
611
501
412
339
48.4
250
278
27.1
32.9
166
Gas
Flow
Rate
(L/hr)
8.65
8.68
8.97
7.45
6.00
4.94
Run
Time
(hr)
4.0
2.0
1.0
1.0
1.0
1.0
%
Removal
50%
43%
33%
40%
39%
*39
** 65%
Chr Recovery Rate from Soil: 82.7%
* 12 extracts analyzed
** 6 extracts analyzed
"Ozone Flux: (1): flux based on 3.5% ozone in oxygen
(2): flux based on KI trap measurements
42
-------
Table 6. Treatment of Pyrene In Soil Columns
(Initial Concentration = 100 p.g PY/g soil)
Ozone Flux'
(mg/hr)
(1)
598
581
315
218
119
26.3
(2)
238
266
20.7
8.91
8.79
1.40
Gas
Flow
Rate
(L/hr)
8.65
8.40
4.55
3.22
1.72
0.38
Run
Time
(hr)
4.0
1.0
1.0
1.0
1.0
1.0
%
Removal
91%
83%
79%
>95%
71%
53%
PY Recovery Rate from Soil: 105%
"Ozone Flux: (1): flux based on 3.5% ozone in oxygen
(2): flux based on KI trap measurements
43
-------
Figure 1. Schematic of a modified vapor stripping system.
o
o
o
.o'
o
...o
voc
source
-:'- "flow .
fli rough
Recycle oxygen or air^_
Ozone
i Generator
i
i
Make-up J
Oxygen —L '
or air
I I
I—1—l Oxygen
j^ or air dryer
Ozone '
inlet
well { Soil Gas
Extraction
^ Well
v'
•".-'•- -
o
o
o
o
o
o
o
&.
Capillary
fringe
Water table
(adapted from Bennedsen, 1987)
-------
°3(g)
HC(g)
DISCRETE
WATER
org+HC -~ ^ org-HC
HC(g)+°3(g) —HCoxidtg)+°2(g)
org+O3
Figure 2. Reactions involving ozone in a soil system.
45
-------
99%
O,
[water
trap
Cooler
i
I
i
Kl
Trao
flowmeter
Son
Column
OZONE GENERATOR
o o
o
Figure 3. Experimental setup used in soil column work.
46
-------
1.0
0.8
0.6--
Q
I
O
U-l
W
U
? 0.4
u
0.2-
0.0
-t r
low ozone dose
high ozone dose
.*
200 400 600
Humic acid, mg/L
800
Figure 4. Effect of ozone dosage on the ozonolysis of cjs-DCE in the presence of Aldrich humic acid
(pH 3.2,10°C). Initial concentrations were (n) [cJs-DCE] = 2.22 mg/L, [O3] = 16.5 mg/L; (•)
[cJS-DCE] = 2.12 mg/L, [O3] = 6.20 mg/L.
47
-------
0.8
u
u
Q
u
U
£ 0.4
0.2-
low
ozone
dose
high ozone dose
o.or-- r -r • T*
40 80 120 160
Humic acid, rag/L
200
240
Figure 5. Effect of pH on the ozonolysis of trans-DCE in solutions containing Aldrich humic acid
(10°C). Initial concentrations were (s) pH 3.2, [ttajas-DCE] = 2.07 mg/L, [O3] = 5.75 mg/L;
(n) pH 3.1, rtrans-DCEl = 1.71 mg/L, [O3] = 6.60 mg/L; (•) pH 8.O, [ttans-DCE] = 1.71 mg/
L, [O3] = 6.60 mg/L.
48
-------
1.0
0.8
0.6--
u
u
cu
\
u
o
^0.4-
0.2-
0.0
pH 7.77
pH 3.26
0 20 40 60 80 100
Humic acid (mg/L)
120 L40
Figure 6. Effect of pH on the ozonolysis of PCE in solutions containing Aldrich humic acid (10°C).
Initial concentrations were (n) [PCE] = 1.81 mg/L, [O3] = 18.8 mg/L; (•) [PCE] = 1.68 mg/
L,[03] = 21.7 mg/L.
49
-------
1«J
. /
1.0
0.8
u
o
Q
o 0.6
\
w
o
a
i 0.4
0.2-
0.0
no
ozone
added
22
rag/L
ozone
1 I 1
100 200 300
[Organic Matter), rag TOC/L water
400
Figure 7. Effect of two soils on the ozonolysis of cis-DCE (10°C). The soil, pH and initial concentrations
used were (•) Eustis soil, pH 3.4-4.8, [cjfi-DCE] = 2.05 mg/L, [O3] = 20.9 mg/L; (n) simulated
soil, pH 6.1-6.8, [cJa-DCE] = 1.89 mg/L, [O3] = 23.1 mg/L.
50
-------
1.0
0.8--
u
0,
\
u
u
* 0.4+
0.2--
0.0
1
200 400 600
[Organic Matter], mg TOC/L water
800
Figure 8. Ozonolysis of PCE in the presence of Eustis soil (10°C). The pH and initial concentrations
used were (n) pH 3.4-4.8, [PCE] = 2.05 mg/L, [O3] = 17.0 mg/L; (•) pH 6.6, [PCE] = 1.63 mg/
L, [O3] = 17.6 mg/L.
51
-------
1.0
0.8
0.6
U
u
a.
U
U
Z 0.4-
0.2-
0.0
a
0.5 mM t-butyl alcohol
o
8
D
no scavenger
H h
20 40 60 80 100
Humic acid, mg/L
120
140
Figure 9. Effect of OH radical scavengers on the ozonolysis of PCE in the presence of Aldrich humic
acid (pH 3.2,10°C). Initial concentrations were (rn) [PCE] = 2.14 mg/L, [O3] = 25.5 mg/
L; (m) [PCE] = 2.24 mg/L, [O3] = 25.0 mg/L; (s) [PCE] = 1.68 mg/L, [O3] = 20.6 mg/L; (•)
[PCE] = 2.24 mg/L, [O3] = 24.4 mg/L; (u) [PCE] = 1.97 mg/L, [O3] = 23.3 mg/L.
52
-------
0.5-
0.4-
< 0.3-
0.2-
0.1 -
0.0-
• with t-Butyl alcohol
A without t-Butyl alcohol
x
X
<*
/
/
/
X
0
10
20
30
40 50
60
70
0.30-
0.25-
=, 0.20-I
0.15-
0.10-
0.05-
0 00-
0
10
20 30 40 50
b)
60 70
[HA] mg/L
Figure 10a. Effect of OH radical scavengers on the ozonolysis of naphthalene in the presence of Aldrich humic
acid (pH 3.2, 10°C). Initial concentrations were [1-BuOH] = 2.8 \ 1O^ M, [NA] = 3.3 x \OS M, [O3]
Figure lOb. The effect of tert-butvl alcohol on the extent to which naphthalene reacts via the direct and
indirect mechanisms in the presence of humic acid. Conditions as given in Figure 10.
53
-------
HABITAT CONDITIONS AFFECTING BIO VENTING PROCESSES
DARWIN L. SORENSENAND RONALD C. SIMS
Bioprocess Engineering Program and Utah Water Research Laboratory
Utah State University
Logan, Utah
THE ROLE OF HABITAT CONDITIONS IN CONTROLLING MICROBIAL ACTIVITY
Natural Boundaries Control Microbial Degradative Activity
Soil venting may be viewed as a process to enhance microbial metabolic activity to degrade organic
contaminants. Microbial metabolism, including contaminant decomposition, is a complex network of
chemical reactions controlled by the same kinds of physical and thermodynamic limits that all chemical
reactions follow. Enzymes produced by microorganisms act as catalysts, lower reaction energy barriers
(Figure 1), and greatly accelerate reaction rates. It is this increase in rate of degradative reactions that makes
biodegradation attractive for soil remediation.
The Niche
The physical location of an organism is its habitat. The dimensions of a microorganism's habitat is on the
order of a cubic millimeter or less and may be best described as a microhabitat (Figure 2). The combination
of an organism's habitat and what it does there defines the concept described as a niche (2). In theory, every
population of microorganisms present in the microbial community of a given habitat fills a niche. Some
overlap in functions of organisms and their habitats may exist, but in general, each niche is unique. If a niche
is not available, microorganisms from other habitats (i.e., allocthonous organisms) are likely to face
competition and may not become part of the community. The introduction of contaminant organic matter
into a habitat may expand an existing niche or serve to create more new niches in the community. If the
contaminant is biodegradable, populations already in the microbial community will expand to fill the niche.
In some cases, physical or chemical (e.g., toxicity) boundaries may exclude organisms with the appropriate
biochemistry to degrade the contaminant(s). Over time, physical and chemical processes, including
activities of the microbial community, may change the niche including those affected by contamination
making new niches available. For example, rapid oxygen consumption by microorganisms degrading a soil
contaminant may change the habitat from aerobic to anaerobic allowing anaerobic or facultatively anaerobic
microbial communities to occupy the habitat.
General Requirements for Microbial Activity
Before bioventing is selected as a treatment process, physical and chemical conditions conducive to growth
and/or maintenance of bacterial and fungal populations in the contaminated soil must be assured. Although
microorganisms that inhabit extreme environmental are known, it is not likely that those organisms will
55
-------
100
w
Not Catalyzed
Change in Activation Energy
Enzyme Catalyzed
Reaction Course
100
Figure 1. Schematic illustration of the decrease in reaction activation energy facilitated by
enzymes (adapted from Rawn 1989).
Fungal
Mycelium
Bacterial
Colony
Silt
Bacterial
Colony
Organic
Matter
Figure 2. Conceptual illustration of the microbial soil habitat. Bacterial distribution tends to be
in microcolonies and organic matter may be closely associated with clays (adapted from
Brock 1979).
56
-------
become important in the degradation of contaminants in most soils. Certain soil conditions should be
available, either through natural occurrence or through engineered controls if biodegradation is likely to be
successful. Moisture must be available; temperature, pH, and redox potential must not be extreme; nutrients
and terminal respiratory electron acceptors must be available; and the habitat must be not toxic. It is
reasonable to assume that if conditions are appropriate for an organism(s) to grow in a habitat, the degradative
enzymes produced by the organism(s) will be active.
Factors Affecting Enzymatic Activity
Biodegradation activity is dependent on the presence of the appropriate enzyme(s) and enzyme activity.
Activity is dependant on enzyme stability or appropriate conformation and chemical conditions at the
catalytic site on the enzyme molecule(s). For biodegradation to occur, many conditions must be met.
Obviously, the microorganisms present must produce enzymes that will catalyze the reaction(s) of interest.
Assuming that the enzyme is produced, activity will be affected by environmental factors including soil
moisture, temperature, pH, redox potential, nutrients, and availability of terminal electron acceptor.
In general, enzyme activity increases with increasing temperature to the point where the structure of the
enzyme is affected by the temperature. At temperatures above the normal operating range for an enzyme,
theprotein(s) which makes up the enzyme may become denatured resulting in loss of catalytic activity. This
leads to the observation that all enzymatic activities may have an optimum temperature (Figure 3). Optimum
enzyme activities are often found at temperatures that are near optimum for the growth of the organism in
which they are produced. However, appreciable activity may exist at temperatures below optimum.
I
lOOn
80-
60-
40-
20-
0 10 20 30 40 50 60
Temperature (° C)
Figure 3. Schematic illustration of the effect of temperature on microbial activity.
57
-------
The proton activity, pH, of the enzyme environment also affects the conformation of the catalytic protein.
Since protonation of protein function groups (carboxyl, amino, and sulfhydryl) is affected by pH, the charge
distribution of the active region of the enzyme is affected by pH. These affects result in a pH optimum for
enzyme activity (Figure 4). Within the cell, the pH environment of the enzyme may be modified and
protected from the extracellular environment. For extracellular enzymes, however, the pH of the
microenvironment has control.
Salinity also affects protein conformation through the interactions of ions functions groups of the protein.
In the laboratory, proteins can be "salted out" of solution by increasing ionic strength causing them to
denature. Ionic affects may also alter the electronic charge distribution at the catalytic region of the enzyme
and cause activity to cease (Figure 5). Within limits, the cell can control the salinity of the environment for
intracellular enzymes. This is not the case for extracellular enzymes. If the salinity of the cell environment
significantly exceeds the salinity inside the cell, osmotic forces will cause the loss of water from the cell
(plasmolysis) and metabolic activity will cease.
Many reactions catalyzed by enzymes are oxidation-reduction reactions. An enzyme cannot catalyze
reactions if electron and proton flow are affected by the redox state of the environment. The redox
environment inside the cell is protected by extracellular enzymes encounter competition with redox couples
in microenvironments.
Subsurface Characterization Requirements for Bioventing
The vadose zone is the region extending from the ground surface of the earth to the upper surface of the
principal water-bearing formation, and is divided into three belts. The uppermost belt consists of soil and
other materials that lie near enough to the surface to discharge water into the atmosphere in perceptible
quantities by the action of plants or by soil evaporation and convection. The lowest belt, the capillary fringe,
is located immediately above the water table and contains water drawn up from the zone of saturation by
capillary action. The intermediate belt lies between the belt of soil water and the capillary fringe. This paper
addresses bioventing of the vadose zone and conditions where the saturated zone is engineered to become
unsaturated, e.g., when ground water is pumped out to create an unsaturated zone.
A promising aspect of soil vacuum extraction is the application for bioventing, i.e., enhancement of
biodegradation of volatile and semi-volatile organic chemicals in a natural or engineered vadose zone. In
the process of bioventing, SVE provides air to the vadose zone, and thus carries oxygen that can be used as
the terminal electron acceptor by soil microorganisms to biodegrade chemicals aerobically. Air has a much
greater potential than water for delivering oxygen to soil on a weight-to-weight basis and volume-to-volume
basis. Oxygen provided by air is more easily delivered since the fluid is less viscous than water; higher
oxygen concentration in air also provides a larger driving force for diffusion of oxygen into less permeability
areas within a soil formation.
Hinchee (1989) and Hinchee and Downey (1990) successfully applied SVE for enhancement of biodegradation
of petroleum hydrocarbons in JP-4 jet fuel at Hill Air Force Base, Ogden, Utah, by increasing subsurface
oxygen concentrations. Soil moisture was found to be a sensitive variable affecting biodegradation, with
increased soil moisture (from 20 percent to 75 percent field capacity) related to increased biodegradation.
58
-------
100 -,
80 -
60 -
I
40 -
20 -
Figure 4. Schematic illustration of the effect of pH on microbial activity.
100
1
Salinity
Figure 5. Schematic illustration of the effect of salinity on microbial enzyme activity.
Considerable variations in this response have been observed (Atlas and Bartha 1978).
Salinity expressed as electrical conductivity (EC) measured in units of mmhos/cm at
25°C.
59
-------
Monitoring of carbon dioxide production and oxygen utilization was used to monitor biodegradation under
field-scale conditions.
Microbial ecologists have identified ranges of critical environmental conditions that affect aerobic activity
of soil microorganisms (Table 1). Many of these conditions are controllable and can be modified to enhance
activity (10,18, 20, 22). Each of the critical environmental conditions identified in Table 1 is considered
in more detail below.
Table 1. Critical environmental factors for microbial activity (Sims et al. 1984, Huddleston et
al. 1986, Rochkind and Blackburn 1986, Paul and Clark 1989)
Environmental Factor
Optimum Levels
Available soil water
Oxygen
Redox potential
PH
Nutrients
Temperature
25 - 85% of water holding capacity; -0.01 MPa
Aerobic metabolism: Greater than 0.2 mg/1 dissolved
oxygen, minimum air-filled pore space of 10%;
Anaerobic metabolism: O2 concentrations less than 1%
Aerobes and facultative anaerobes: greater than 50
millivolts; Anaerobes: less than 50 millivolts
5.5 - 8.5
Sufficient nitrogen, phosphorus, and other nutrients
so not limiting to microbial growth
(Suggested C:N:P ratio of 120:10:1)
15 - 45° C (Mesophiles)
Soil Moisture and Oxygen Concentration
Water is necessary for microbial life, and the soil water matric potential against which microorganisms must
extract water from the soil regulates their activity. Soil matric potential is the energy required to extract water
from the soil pores to overcome capillary and adsorptive forces. Soil water also serves as the transport
medium through which many nutrients and organic chemicals diffuse to the microbial cell, and through
which metabolic waste products are removed. Soil water also affects soil aeration status, nature and amount
of soluble materials, soil water osmotic pressure, and the pH of the soil solution (18).
Microbial respiration, plant root respiration, and respiration of other organisms remove oxygen from the soil
atmosphere and enrich it with carbon dioxide. Gases diffuse into the soil from the air above it, and gases in
60
-------
the soil atmosphere diffuse into the air. However, oxygen concentration in a soil may be only half that in air
while carbon dioxide concentrations may be many times that of air. Even so, a large fraction of the microbial
population within the soil depends on oxygen as the terminal electron acceptor in metabolism. When soil
pores become filled with water, the diffusion of gases through the soil is restricted. Oxygen may be consumed
faster than it can be replaced by diffusion from the atmosphere, and the soil may become anaerobic. Clay
content of soil and the presence of organic matter also may affect oxygen content in soil. Clayey soils tend
to retain a higher moisture content, which restricts oxygen diffusion, while organic matter may increase
microbial activity and deplete available oxygen. Loss of oxygen as a metabolic electron acceptor induces
a change in the activity and composition of the soil microbial population. Facultative anaerobic organisms,
which can use oxygen when it is present or can switch to alternative electron acceptors such as nitrate or
sulfate in the absence of oxygen, and obligate anaerobic organisms become the dominant populations.
Redox Potential
Another soil parameter that describes the effect of the soil environment on metabolic processes is the redox
potential of the soil (18). Biological energy is obtained from the oxidation of reduced materials. Electrons
are removed from organic or inorganic substrates to capture the energy that is available during the oxidative
process. Electrons from reduced compounds are moved along respiratory or electron transport chains
composed of a series of compounds. In an aerobic process, O2 acts as the terminal electron acceptor. In some
cases where O2 is not available, nitrate (NO3~), iron (Fe3+), manganese (Mn2+), and sulfate (SO42~) can act as
electron acceptors if the organisms have the appropriate enzyme systems. A measurement of the oxidation-
reduction potential (redox potential) of a soil provides a measurement of the electron density of the system.
As a system becomes reduced, O2 is depleted, and other substances are used as terminal electron acceptors.
There is a corresponding increase in electron density, resulting in a progressively increased negative
potential. Redox potential is measured as Eh, expressed in millivolts, or as pe, which is equal to -log [e~],
where [e'] is the concentration of negatively charged electrons. Redox potential is measured in a subsurface
system with a reference electrode in combination with a metallic electrode, such as platinum, which is
sensitive and reversible to oxidation-reduction conditions. Very well oxidized soils have redox potentials
of 400 to 800 mV, while extremely reduced soils may have potentials of -100 to -500 mV (4).
Eh defined as follows (Dragun 1988):
Eh = pe (2.3 RT/F)
where Eh = redox potential, Volts
pe = negative log of electron concentration
R = gas constant, 0.001987 kcal/°K
T = temperature, °K
F = faraday constant, 23.06 kcal/V equivalent
thus
pe = 16.9 (Eh)
Since electrons neutralize protons, many reactions in the subsurface are both Eh and pH dependent.
Therefore, Eh-pH diagrams are often used to indicate the predominant dissolved and mineral species as
functions of the Eh and pH values a system.
61
-------
Soil pH
Soil pH also affects the activity of soil microorganisms. Fungi are generally more tolerant of acidic soil
conditions (below pH 5) than are bacteria. Near neutral pH values are most conducive to microbial
functioning in general. The solubility of phosphorus, an important nutrient in biological systems, is
maximized at a pH value of 6.5. A specific contaminated soil system may require management of soil pH
to achieve levels that maximize microbial activity. Control of pH to enhance microbial activity may also aid
in the immobilization of hazardous metals in a soil system (a pH level greater than 6 is recommended to
minimize metal transport).
Buffering capacity of soil reflects the ability of the soil components to hold large number of ions in adsorbed
or reserve form. This, adsorption or inactivation of H+ ions or the release of adsorbed ions to neutralize OH-
ions provides protection against abrupt changes in pH when acidic or basic constituents are added to soil.
Varying buffer capacities among soils reflects differences in the soil cation exchange capacities and will
directly affect the amount of material required for addition to change pH as well as the length of time before
additional material must be added to maintain pH at a desired level.
pH also affects the molecular structure for chemicals that ionize. Comparison of the soil pH value with
chemical Ka value will indicate the distribution of chemical between ionized and un-ionized form. Generally,
ionized forms of chemicals are more water soluble and therefore less available for volatilization, and for
anionic species, less available for sorption to the soil solid phase.
Nutrient Availability
Microbial metabolism and growth is dependent upon adequate supplies of essential macro- and micronutrients.
Required nutrients must be present and available to microorganisms in (1) a usable form; (2) appropriate
concentrations; and (3) proper ratios (4). If the wastes present at the site are high in carbonaceous materials
and low in N and P, the soils may become depleted of available nitrogen (N) and phosphorus (P) required
for biodegradation of the organic constituents. Fertilization may be required at some contaminated sites as
a management technique to enhance microbial degradation.
In agriculture, fertilizer is added to hasten the decomposition of crop residues (1). This procedure also has
been used in the treatment of soil contaminated with hazardous wastes as a result of an oil spill (25). Skujins
et al. (1983) studied the biodegradation of waste oils at a disposal site where soils were amended with calcium
hydroxide, phosphate, and urea; within four years, 90 percent of the applied oil was degraded.
Although most microorganisms can efficiently extract inorganic nutrients from their environment, their
activity may be limited by the availability of nutrient. This is especially true if available carbon is excessive
relative to the amount of nitrogen or phosphorus required to degrade the carbon. If soil organic carbon,
organic nitrogen, and organic phosphorus are determined, the C:N:P ratio can be determined and nutrient
availability can be evaluated. If the ratio of C:N:P is wider than approximately 300:15:1 (weight basis) and
available (extractable) inorganic forms of N and P do not narrow the ratio to within these limits, supplemental
nitrogen or phosphorus should be added (1).
62
-------
Temperature
Biodegradation of organic constituents declines with lowering of soil temperature due to reduced microbial
growth and metabolic activity. Biodegradation has been shown to essentially stop at a temperature of 0° C.
Soils exhibit a variation in the temperature of the surface layers, both diurnally and seasonally. Diurnal
changes of temperature decrease with depth of the soil profile. Due to the high specific heat of water, wet
soils are less subject to large diurnal changes than dry soils (18). Factors that affect soil temperature, and
therefore microbial activity at a site, include soil aspect (direction of slope), steepness of slope, degree of
shading, soil color, and surface cover.
Soil temperature also influences the rate of volatilization of compounds from soil. As the rate of
volatilization increase, the residence time of the volatilized organic chemical in soil may decrease and result
in reduction of the mass of chemical available for biodegradation. However, if volatilization is increased
due to increased temperature, engineering controls, such as reduction in vacuum extraction rate, can be
utilized to maintain volatilized chemicals in the soil environment (i.e., increase soil residence time) for a
longer period of time to allow biodegradation to occur.
The effect of temperature on the biodegradation rate often can be described quantitatively by the change in
the degradation constant term using the Arrhenius expression:
k = Aoe-(Ea/RT) (1)
where k is the rate constant, Ao is a frequency factor, E is the energy of activation, R is the universal gas
constant, and T is absolute temperature.
Rate of degradation, k in equation (1), is often expressed as a function of the concentration of one or more
of the constituents being degraded or the change in concentration of the terminal electron acceptor, or the
rate of generation of a product of biodegradation, e.g., CO2. This is termed the order of the reaction and is
the value of the exponential used to describe the reaction (18). Either zero or first order power rate models
are often used in environmental studies.
Zero order reactions are ones in which the rate of transformation of an organic constituent is unaffected by
changes in the constituent concentration because the reaction rate is determined by some other factor than
the constituent concentration. If a constituent C is transformed to X, the rate of change of C is:
dC/dt = -k (2)
On integration, the equation becomes:
Ct = Co-kt (3)
where Ct = concentration of constituent remaining at time t; and Co = initial concentration of constituent, and
k = zero order rate constant. A useful term to describe the reaction kinetics is the half-life, t which is the
time required to transform 50% of the initial constituent:
63
-------
Ct = Cy2, then t1/2=C/2k (4)
The first order rate model (Equation 5) is widely used because of its effectiveness in describing observed
results as well as its inherent simplicity. Its use also allows comparison of results obtained from different
studies. In a first order rate reaction, the rate of transformation of a constituent is proportional to the
constituent concentration:
dC/dt = -kC (5)
where C = contaminant concentration (mass/mass); t = time; and k = first order rate constant (I/time). After
integration of Equation 4 and rearrangement of the integrated equation, Equation 5 may be used to
graphically determine the rate constant, k:
ln(C/Co) = -kt (6)
where Ct = concentration of constituent remaining at time t; and Co = initial concentration of constituent. A
plot of ln(C/Co) versus t is linear with a slope of -k. The rate constant k is independent of the concentration
of constituent, since the slope is constant over time. To calculate the time required to transform one-half of
the initial constituent (Ct = C/2), the following equation is used:
In ((Co/2)/Co) = -kt1/2 (7)
which is equal to:
t]/2 = 0.693/k (8)
where t1/2= half-life of the constituent.
First order kinetics generally apply when the concentration of the compound being degraded is low relative
to the biological activity in the soil. However, very low concentrations may be insufficient to initiate enzyme
induction or support maintenance requirements necessary for microbial growth, even if the compound can
be used as an energy source. (21).
A second model used to describe degradation in soils is the hyperbolic rate model, which is similar to
Michaelis-Menten enzyme kinetics. This model is expressed as:
dC/dt^-kjC/^+C) (9)
where kj and k2 are constants. The constant kj represents the maximum rate of degradation that is approached
as the concentration increases. This model simulates a catalytic process in which degradation may be
catalyzed by microorganisms.
Toxicity Assessment
Bioassays to quantify toxicity measure the effect of a chemical on a test species under specified test
64
-------
response of the test organism(s). Often a battery of bioassays is utilized that may include measurements of
effects on general microbial activity (e.g., respiration, dehydrogenase activity) as well as assays relating to
activity of subgroups of the microbial community (e.g., nitrification, nitrogen fixation, cellulose
decomposition). Bioassays utilizing organisms from different ecological trophic levels may also be used to
determine toxicological effects. However, use of a single assay as a screening test to identify relative toxicity
assessment and toxicity reduction in the environment is a common procedure employed in treatability
studies. Assays utilizing microorganisms are often used due to their speed, simplicity, ease in handling, cost
effectiveness, and use of a statistically significant number of test organisms that is required to detect the
effects of potentially toxic materials in the environment (5,13).
The Microtox™ assay is an aqueous general toxicity assay that measures the reduction in light output
produced by a suspension of marine luminescent bacteria in response to an environmental sample.
Bioluminescence of the test organism depends on a complex chain of biochemical reactions. Chemical
inhibition of any of the biochemical reactions causes a reduction in bacterial luminescence. Therefore, the
Microtox™ test considers the physiological effect of a toxicant and not just mortality. Matthews and Bulich
(1984) have described a method of using the Microtox™ assay to predict the land treatability of hazardous
organic wastes. Matthews and Hastings (1987) described a method using the Microtox™ assay to determine
an appropriate range of waste application loading for soil-based treatment systems. Symons and Sims (1988)
utilized the assay to assess the detoxification of a complex petroleum waste in a soil environment. The assay
was also included as a recommended bioassay in the U. S. EPAPermit Guidance Manual on Hazardous Waste
Land Treatment Demonstrations (26).
Soil respiration is generally accepted as a measure of overall soil microbial activity and has been used as an
indicator of the toxicity or of the utilization of organic compounds added to the soil environment. Respiration
may also act as an indicator for microbial biomass in soil because the transformations of the important
organic elements (C,N, P, and S) occur through the biomass. Measurement of CO2 evolution form soil
samples is a commonly used indicator of soil respiration, although measurement of O2 uptake is also an
option for short-term evaluations.
Determination of soil respiration through CO2 evolution or O2 utilization are inexpensive and simple
approaches for indicating general soil microbial activity and acute effects of the presence of organic
chemicals on general microbial activity. This approach has been used for evaluation of the biological
component in removal of organic chemicals in bioventing systems (8, 9, 17).
Dehydrogenation is the general pathway of biological oxidation of organic compounds. Dehydrogenases
catalyze the oxidation of substrates which produce electrons able to enter the electron transport system of
a cell. The type and quantity of carbon substrates, present and introduced, will influence dehydrogenase
activity. This assay involves the incubation of soil with 2,3,5-triphenyltetrazolium chloride (TTC). The
water-soluble colorless TTC intercepts the flow of electrons produced by microbial dehydrogenase activity
and is reduced to the water-insoluble, red 2,3,5-triphenyltetrazeolium formazan (TTC-formazan). TTC-
formazan is extracted from the soil with methanol and quantified colorimetrically.
The soil dehydrogenase activity assay is simple and efficient, and is convenient. Major equipment required
include a spectrophotometer, centrifuge, and an incubator.
65
-------
Nitrification activity represents another assay for evaluating biological activity and toxicity in subsurface
soils using bioventing technology. Nitrification, the oxidation of ammonium nitrogen to nitrite and nitrate,
is accomplished by environmentally sensitive chemoautotrophic bacteria. Energy substrates as well as
oxidized products are easily extracted from soil and measured, and therefore the nitrification process may
be used as a bioassay for microbial activity, as well as toxicity, in the soil.
No single assay of soil microbial activity is likely to indicate the activity of the broad spectrum of soil
microorganism. Measurements of respiration and general toxicity (e.g., MicrotoxTm) may represent the
broadest community of microorganisms. Therefore a battery of bioassays, including an activity assay as well
as a toxicity assay, are useful in determining the potential for biodegradation as well as the intensity of
biological degradation of organic chemicals at a site and for determining the influence of environmental
variables on the rate and extent of biodegradation achieve through engineered bioventing at a site.
Interaction of Environmental Factors
The environmental factors presented in Table 1, as well as soil and waste characteristics, interact to affect
microbial activity at a specific contaminated site utilizing bioventing as a treatment technology. Computer
modeling techniques are useful to attempt to describe the interactions of environmental factors and their
effects on treatment of organic constituents in a site-specific bioventing situation.
RESEARCH EFFORTS TOWARD INCREASING BIOVENTING EFFICIENCY
Is There an Advantage to Alternating Aerobic and Anaerobic Processes?
Aerobic habitats are conducive to some microbial activities that may be advantageous in bioventing.
Reductive dechlorination of some compounds under anaerobic conditions may be advantageous where
chlorinated compounds are the contaminant or part of the contamination (6,7). The reduced products of these
transformations may be susceptible to mineralization under aerobic conditions. Nitrogen fixation by free-
living anaerobic bacteria may add to the available nitrogen pool and may be a low cost, effective way of
providing nutrients in systems that have become carbon rich because of contamination with organic
chemicals. Organic acids produced in anaerobic fermentations may aid in solubilizing phosphorus and other
mineral nutrients from soil minerals thus promoting microbial growth and activity.
We have observed nitrogen fixation activity from free living bacteria, as measured by the acetylene reduction
assay in jet fuel (JP-4) contaminated soil. Acetylene reduction activity at a contaminated site at Tyndall Air
Force Base, Florida, averaged 200 nmol kg'1 h'1 under anaerobic conditions at 20°C (Miller 1990). Assuming
a 3:1 ratio of acetylene reduced to nitrogen fixed, an average of up to 16mg N kg"1 yr1 could be added to this
soil from nitrogen fixation. Free living nitrogen fixing bacteria require relatively high amounts of carbon
and energy substrates to be available (1) and, for this reason, their activity is often limited in subsurface soil
systems. In contaminated environments where degradable organic matter may be abundant however,
nitrogen fixation activity may be significant.
66
-------
Alternating periods of anaerobiosis may provide opportunity for the following reactions in subsurface
contaminated soil: (1) fixation of significant amounts of nitrogen (2) dehalogenation of chlorinated organics
(3) solubilization of mineral nutrients, and (4) denitrification of excess nitrate. Based upon preliminary
results obtained by Utah State University researchers, these reactions are considered potential advantages
of "pulsed venting." Characterization and evaluation of "pulsed venting," as described above, at the field-
scale level represents a research need that may have immediate application to sites considering bioventing
for remediation.
REFERENCES
1. Alexander, M. 1977. Introduction to soil microbiology. John Wiley & Sons, New York, NY.
2. Atlas, R.M. and R. Bartha. 1987. Microbial ecology fundamentals and applications, second edition.
Benjamin/Cummings, Menlo Park, CA.
3. Brock, T.D. 1979. Biology of microorganisms, third edition. Prentice-Hall, Englewood Cliffs, NJ.
4. Dragun, J. 1988. The soil chemistry of hazardous materials. Hazardous Materials Control Research
Institute, Solver Spring, MD.
5. Dutka, B.J., and G. Bitton. 1986. Toxicity Testing using Microorganisms. CRC Press, Inc., Boca
Raton, FL.
6. Freedman, D.L., and J.M. Gossett. 1989. Biological reductive dechlorination of tetrachloroethylene
and trichloroethylene to ethylene under methanogenic conditions. Appl. Environ. Microbiol.
55:2144-2151.
7. Genthener, B.R.S., W.A. Price II, and P.H. Pritchard. 1989. Anaerobic degradation of chloroaromatic
compounds in aquitic sediments under a variety of enrichment conditions. Appl. Environ. Microbiol.
55:1466-1471.
8. Hinchee, R. 1989. Enhanced biodegradation through soil venting. Proceedings of the Workshop on
Soil Vacuum Extraction, U.S. Environmental Protection Agency, Robert S. Kerr Environmental
Research Laboratory, Ada, OK, April 27-28.
9. Hinchee, R., and D. Downey. 1990. In situ enhanced biodegradation of petroleum distillates in the
vadose zone. Proceedings of International Symposium on Hazardous Waste Treatment: Treatment
of Contaminated Soils, Air and Waste Management Association, U.S. Environmental Protection
Agency, Risk Reduction Engineering Laboratory, Cincinnati, OH, February 5-8.
10. Huddleston, R.L., C.A. Bleckmann, and J.R. Wolfe. 1986. Land treatment biological degradation
processes, pp. 41-61. In: R.C. Loehr and J.F. Malina, Jr. (eds.) Land Treatment: A Hazardous Waste
Management Alternative. Water Resources Symposium No. 13, Center for Research in Water
Resources, The University of Texas at Austin, Austin, TX.
67
-------
11. Keck, J., R.C. Sims, M. Coover,, K. Park, and B. Symons. 1989. Evidence for cooxidation of
polynuclear aromatic hydrocarbons in soil. Water Res. (In press).
12. Lehr, J.H. 1988. The misunderstood world of unsaturated flow. Ground Water Monitoring Review
(Spring) :4-6.
13. Liu, D., and B.J. Dutka (eds.). 1984. Toxicity Testing Procedures using Bacterial Systems. Marcel
Dekker, Inc., New York, Inc.
14. Loehr, R. 1989. Treatability Potential for EPA Listed Hazardous Wastes in Soil. EPA/600/2-89/011,
Robert S. Kerr Environmental Research Laboratory, U.S. Environmental Protection Agency, Ada,
OK.
15. Matthews, J.E. and A.A. Bulich. 1984. A toxicity reduction test system to assist predicting land
treatability of hazardous wastes, pp. 176-191. In: J.K. Petros, Jr., W. J. Lacy, and R. A. Conway, eds.,
Hazardous and Industrial Solid Waste Testing: Fourth Symposium, STP-886, American Society of
Testing and Materials, Philadelphia, PA,
16. Matthews, J.E. andL. Hastings. 1987. Evaluation of toxicity test procedure for screening treatability
potential of waste in soil. Toxicity Assessment: An Internatl. Quarterly 2: 265-281.
17. Miller, R. 1990. A field scale investigation of enhanced petroleum hydrocarbon biodegradation in
the vadose zone combining soil venting as an oxygen source with moisture and nutrient addition.
PhD dissertation. Utah State University, Logan, UT.
18. Paul, E.A., and F. E. Clark. 1989. Soil microbiology and biochemistry. Academic Press, Inc., San
Diego, CA.
19. Rawn, J.D. 1989. Biochemistry. Carolina Biological Supply, Burlington, NC.
20. Rochkind, M.L., J.W. Blackburn, and G. Sayler. 1986. Microbial decomposition of chlorinated
aromatic compounds. U.S. Environmental Protection Agency, Hazardous Waste Engineering Research
Laboratory, Cincinnati, OH, EPA/600/2-86/090.
21. Rittmann,B.E., andP.L.McCarty. 1980. Model of steady-state biofilm kinetics. Biotech. Bioeng. 22:
2343.
22. Sims, R.C., D.L. Sorensen, J.L. Sims, J.E. McLean, R. Mahmood, and R.R. Dupont. 1984. Review
of in-place treatment technologies for contaminated surface soils, Volumes land 2. U.S. Environmental
Protection Agency, Risk Reduction Engineering Laboratory, Cincinnati, OH, EPA-540/2-84-003a
and b.
23. Skujins,, J.S., O. MacDonald, and W.G. Knight. 1983. Metal ion availability during biodegradation
of waste oil in semi-arid soils. Environmental Biogeochemistry 35:341-350.
68
-------
24. Symons, B.D. and R.C. Sims. 1988. Assessing detoxification of a complex hazardous waste, using
the Microtox™ bioassay. Arch. Environ. Contamination Toxicol.17: 497-505.
25. Thibault, G.T., and N.W. Elliott. 1980. Biological detoxification of hazardous organic chemical
spills. In: Control of Hazardous Materials Spills. Vanderbilt University, Nashville, TN. pp. 398-402.
26. U.S. EPA. 1986. Permit Guidance Manual on Hazardous Waste Land Treatment Demonstrations.
EPA-530/S W-86-032, Office of Solid Waste and Emergency Response, U.S. Environmental Protection
Agency, Washington, DC.
69
-------
OPPORTUNITIES FOR BIOTREATMENT OF TRICHLOROETHYLENE IN
THE VADOSE ZONE
JOHNT. WILSON
DON H. KAMPBELL, AND
JONG CHO
U.S. EPA-Ada, Oklahoma
ABSTRACT
Trichloroethylene (TCE) vapors in air can be amended with gaseous alkanes, then injected into the
unsaturated zone. The unsaturated zone acts as a natural in-situ bioreactor. During natural biodegradation
of the alkanes, the TCE is co-oxidized to carbon dioxide and chloride. The technology is most appropriate
for sites with a deep and homogeneous unsaturated zone. If necessary, a fertilizer solution can be applied
to encourage rapid rates of hydrocarbon oxidation.
The technology is well established at the bench scale. Commercial propane or LPG gas can be used as the
feedstock; however, the best results have been obtained with apropellant mixture designed for use in aerosol
cans. Prior work by Kampbell et al. (Journal of the Air Pollution Control Association, 37(10)1236-1239,
1987) demonstrated removal of volatile aliphatic hydrocarbons in a soil bioreactor. In a laboratory system,
acclimated soil was exposed to air amended with the propellant mixture and TCE vapors. TCE removal was
rapid and extensive compared to controls constructed from soil that was not acclimated to the hydrocarbons.
Current work attempts to optimize the rate and extent of biodegradation of the feedstock at field scale.
71
-------
SOIL VENTING DESIGN: MODELS AND DECISION ANALYSIS
MARIAN W. KEMBLOWSKI AND
SHYAMAL CHOWDERY
Utah State University
ABSTRACT
Venting of hydrocarbon-contaminated soils is a frequently used and generally efficient technology for soil
cleanup. Proper design of soil venting requires two major components: (1) understanding of the processes
that control venting efficiency, and (2) information related to relevant chemical and physical soil and
contaminant parameters. In this paper we present several simple models that describe the principal venting
processes, including air flow, hydrocarbon-vapor partitioning, ground water upwelling, and diffusion-
limited hydrocarbon removal. These models are used to determine which of the physical and chemical soil
and contaminant parameters determine the venting efficiency.
Ideally, one would like to know the spatial distribution of such parameters (for example: soil permeability,
contaminant concentration, contaminant composition, etc.) as precisely as possible. Incomplete and/or
inaccurate information may result, under some circumstances, in significant losses related to improper
design of a venting system. On the other hand, obtaining complete and accurate information is hardly ever
possible, let alone practical, mostly due to the cost of subsurface investigation and chemical analyses. Thus,
the whole design process consists of several modeling and investigation phases.
At the end of each evaluation (modeling) phase, a decision(s) regarding the need for and expectations from
further field investigation has to be made. The problem is even more complicated by the fact that one cannot
predict the success of subsurface investigation in a deterministic way. Due to this fact, and the intrinsic
random character of subsurface parameters, the process of venting design and related investigation should
be performed in a stochastic framework. This can be done using the Bayesian decision analysis approach.
Applicability of this methodology to venting design is discussed. Some examples are presented.
73
-------
MODELING TRANSPORT OF ORGANIC CHEMICALS
IN GAS AND LIQUID PHASES
A. K. KATYAL, P. K. PATEL AND J. C. PARKER
Virginia Polytechnic Institute & State University
ABSTRACT
The basic principles of modeling fluid flow and transport in multiphase systems relevant to the analysis of
in situ gas venting are discussed. The assumption of equilibrium phase partitioning enables multiphase
transport equations to be reduced to a single phase-summed equation that has the form of the conventional
single phase transport equation with lumped coefficients which reflect contributions from various mobile
and immobile phases. Due to diffusive limitations on the rate of mass transfer between phases at the pore
scale and between regions of high and low permeability at the field scale, apparent nonequilibrium
interphase mass transfer may occur. A first-order kinetic formulation for interphase mass transfer is
described which can be incorporated into the phase-summed transport equation by defining apparent
partition coefficients, which are functions of the actual local mass transfer rate. A numerical model has been
developed which enables solution of water, NAPLand gas flow and transport equations with nonequilibrium
mass transfer. An example problem is presented involving a spill of chlorinated solvent which penetrates the
unsaturated and saturated zones. Two remediation scenarios are simulated involving 1) gas venting only, and
2) gas venting combined with water pumping. In the former case, water table upwelling occurs due to the
reduction in gas pressure near the vacuum well. Water pumping enables water upwelling to be controlled and
increases the gas phase recovery while also enabling aqueous phase recovery of solvent in the saturated zone.
Simulations with different mass transfer rate constants show that mass recovery rates decrease as
nonequilibrium effects become more pronounced.
INTRODUCTION
Organic liquids enter the subsurface environment through a variety of sources which include accidental
spills, leaking underground storage tanks, pipeline leaks and other releases. These nonaqueous phase liquids
(NAPLs), which may be complex mixtures of different organic compounds, may migrate downward and
spread laterally on the water table if less dense than water, or they may penetrate into the aquifer if more dense
than water. NAPLs pose a serious threat to water quality due to the gradual release of components to the
aqueous phase.
Anumber of researchers have developed multiphase models with different capabilities and limitations. Most
models have been restricted to two-dimensional vertical sections (1, 6, 13, 14, 16, 19) and vertically
integrated models for areal analyses of oil-water flow (9,15,20,21). Abriola and Pinder (1985) presented
a computationally intensive model, which was later modified by Reeves and Abriola (1988) to improve its
efficiency. Corapcioglu and Baehr (1987) presented a multicomponent transport model with steady water
flow and immobile oil and air phases including oxygen-limited biodegradation. Most of the above
75
-------
researchers assumed local equilibrium partitioning between various phases. Falta and Javandel (1987)
presented a three-dimensional model which considered nonequilibrium phase partitioning, and Sleep and
Sykes (1989) presented a similar two-dimensional model. Miller et al. (1990) presented a comparison
between equilibrium and nonequilibrium oil-water partitioning models.
Under conditions of high oil saturation and low flow velocities, the assumption of local equilibrium
partitioning may be valid (10, 22). On the other hand, nonequilibrium conditions have been reported by
Schwille (1975), Feenstra and Coburn (1986), Miller and Weber (1986), Hunt et al. (1988b) and Geller and
Hunt (1989) in systems with low residual oil saturations.
Various researchers have addressed the problem of in situ air venting for remediation of volatile compounds
from soil. Baehr et al. (1989) presented a model for air venting based on equilibrium partitioning. Johnson
et al. (1990) presented a screening model for designing air venting systems, also based on equilibrium
partitioning. Recently, Powers et al. (1991) compared the removal efficiency of pump and treat systems with
equilibrium and nonequilibriumpartitioning between water and NAPLs. Powers et al. (1991) concluded that
the important parameters that controlled the removal efficiency were the mass transfer coefficient and the
flow velocity.
This paper extends the multiphase model of Kaluarachchi and Parker (1990) to include nonequilibrium
interphase partitioning with a dynamic gas phase. Emphasis is placed on impacts of nonequilibrium
partitioning on the removal efficiency for air venting systems. A hypothetical spill of dense chlorinated
hydrocarbon is simulated and two different scenarios are studied involving air pumping only or combined
air and water pumping. Water pumping enables control of the water table upwelling at the vacuum well to
increase the air removal efficiency. In each scenario, removal efficiency is compared with different
nonequilibrium partitioning coefficients and with equilibrium partitioning.
MODEL DESCRIPTION
Flow model. The mass conservation equations for water (w), organic liquid (o) and air (a) phases, assuming
an incompressible porous medium, incompressible liquid phases and compressible gas phase, may be written
in summation convention for a two dimensional Cartesian domain as
0—=--^- + ^ (la)
(Ib)
76
-------
where 0 is porosity, S is the p-phase saturation, x. (and x) are Cartesian spatial coordinates (ij =1,2), qp.
is the Darcy velocity of phase p in the /-direction, p is the density of phase p, R is the net mass transfer per
unit porous media volume into ( + ) or out of (-) phase p, and t is the time. Darcy velocities in the p-phase
are defined by
where Kpiiis the p-phase conductivity tensor, h = P Igp* is the water height equivalent pressure head of phase
p where Pp is the p-phase pressure, g is gravitational acceleration and p* is the density of pure water, pp is
the density of phase p, p = p Ip* is the p-phase specific gravity, and u. = dz/dx. is a unit gravitational vector
measured positive upwards where z is elevation. Initial and boundary conditions for each phase may be
specified as a prescribed head or a prescribed flux.
Two-phase air- water saturation-capillary pressure relations are described by the van Genuchten (1980)
function
Sw = [l + (ahmyi]'m (3)
where a and n are porous medium parameters, m = 1-1 In and hm= ha — hw. Following the occurrence of oil
at a given location, the system is described by the three phase relations
+ (a pow how) (4a)
St = [\+(aftaohao)n\m (4b)
where how = ho — hw and hm = ha — hg, ft^ and pgw are fluid-dependent scaling factors and Sw and St are
"apparent" water and total liquid saturations which account for effects of irreducible water saturation, Sn and
the maximum trapped oil saturation, Sar, as described in detail by Kaluarachchi and Parker (1991). Phase
conductivities are described by
K-oij = kro f^swi I'Hro (5b)
Kaij = kraKgwi /'Hra (5c)
where k^ is the relative permeability of phase p, ^ is the viscosity ratio between phase p and water, and K^..
is the saturated conductivity tensor for water. We shall assume here that the coordinate system is oriented
with the conductivity tensor, or otherwise that off -diagonal components may be disregarded, so that K .. =
0 for / *j. Phase relative permeabilities are described by
77
-------
*rw=Sw1/2[l-(l-$;1/IB)]2 (6a)
kra = l-S,l-S (6C)
where m = 1-1 In is the van Genuchten parameter.
Transport model. Mass conservation of species a in the p-phase requires that
where C is the concentration of the noninert a component in p-phase expressed as the mass of a per phase
volume [M L'3] , / is the mass flux density of a in p-phase per porous media cross section in the i-direction
[M L'2 T •'], and R is the net mass transfer rate per porous medium volume of species a into ( + ) or out of
(-) the p-phase [ML'3].
The mass flux density of component a in phase p due to convection, diffusion and mechanical dispersion is
described by
Japi = Cap (Jpi -QSpDapij - (8)
where Dapij is a dispersion tensor given by
Dopy =8ap(Dd/P +D%) (9)
in which D q£is the molecular diffusion coefficient of a in the p-phase of the porous medium, D ?. is a
mechanical dispersion coefficient, and g is a nonadult solution correction factor. For the case of transport
of low solubility organic components, small volume fractions of organic components in water and gaseous
phases will occur and g^ = g^ = 1 is assumed. For nonaqueous phase liquids, the phase composition may
reach 100% (e.g., single component organic liquid) at which point dispersive transport becomes nonexistent.
We accommodate nondilute solution diffusion-dispersion by approximating the oil phase nonideal solution
factor by
goo = 1 - Coo Ip0 (10)
78
-------
Employing the tortuosity model of Millington and Quirk (1959) yields the expression for D ^ as
7/3
ap
(11)
where D&p is the diffusion coefficient of a in bulk p-phase. The mechanical dispersion coefficient is shown
by Bear (1972) to have the form
Dhyd =
(12)
where AL and AT are longitudinal and transverse dispersivities [L], qpi and qpj are p-phase Darcy velocities
in the i and j directions, qp = IIqpi2l112 is the absolute magnitude of the p-phase velocity, and Sy is
Kroonecker's delta.
Combining the phase continuity equation and the mass flux equation for transport of component a in the
p-phase yields
SpD
papij
R
ax,-
ap
(13)
Expanding the first and third terms in (13), employing the bulk p-phase continuity equation (1), and assuming
density derivative terms to be of second order importance within a given time step yields
.c
S
+ Rap-^
Pp
(14)
To accommodate adsorption of aby the solid phase, an additional continuity equation is required which may
be written as
'OS
= R,
(15)
where Cos is the solid phase concentration expressed as mass of adsorbed component a per porous medium
volume [ML'3] and/? ay is the mass transfer rate per porous medium volume to (+ ) or from (-) the solid phase
[M L'3 T1]. The total phase mass transfer rate, Rp, is related to the individual component mass transfer rates
by
n,
Rp = / , Rao
a=l
(16)
79
-------
where ns denotes the number of "noninert" or partitionable species. In the present context, we will use the
term "inert" to refer to components of the NAPL phase which are for practical purposes insoluble and
nonvolatile.
We introduce equilibrium phase partition relations of the form
Coo = iao COM, (1 7a)
Caa = laaCaw (I'D)
Cos = FOS Caw (17c)
where superscript e denotes an equilibrium concentration, Tao is the equilibrium partition coefficient for
species a between water and organic liquid (Raoult's constant), Yaa is the equilibrium partition coefficient
between water and gas (Henry's constant), and F^ is a dimensionless equilibrium partition coefficient
between water and solid phase.
Under transient field conditions actual phase concentration ratios may differ from the true equilibrium ratios
defined by [17]. With this in mind, we define apparent partition coefficients — which will vary in time and
space — by analogs of [17] as
~ *ao Caw
= Faa Caw ( 1 8b)
Cos — *as Caw
For any two phases which are in physical contact, the rate of mass transfer will be described by first order
mass transfer functions of the form
Ran = kau(Cai-Cal} (19a)
2-Ca2) (19b)
where R a\2 is the rate of mass transfer of a per porous medium volume from phase 1 to phase 2, Ca\ is the
actual concentration of a in phase 1 , Ca i is the concentration of a which would occur in phase 1 if it were
in equilibrium with phase 2,Ca2 is the actual concentration of a in phase 2, Ca2 is the concentration of a
that would occur in phase 2 if it were in equilibrium with phase 1 , and &cd2 is a mass transfer rate coefficient
[T1].
Consider first, the case of mass transfer between oil and water phases when both phases exist at a point in
time and space. Employing [19a] along with [18a] for the actual oil phase concentration and [17a] for the
equilibrium concentration gives
80
-------
(r r - r r
1 V OCO^CCW -* QJO^-'C
which may be solved in terms of TOO
which indicates that the apparent partition coefficient may be expressed in terms of the actual partition
coefficient plus a correction term, which depends on the sign and magnitude of the actual mass transfer rate
and on the concentration in the water phase.
If free oil is present in the system at a given point in time and space, so that air and oil are physically in contact,
mass transfer between oil and gas phases may be described in a similar fashion. Employing in this case [ 19b]
we obtain
Raoa = ~ i^aoa \^aa t-otw ~ A acf^aw }
which yields
r - r
1 aa - i
aa - i aa
In the absence of free oil in the porous medium at a given location, mass transfer will occur between water
and gas phases rather than between oil and gas phases. Proceeding in the same manner as for oil-gas mass
transfer yields
aa
— r
~ *
Note that either (23) or (24) will be relevant at a given location and time, depending on whether free oil
occurs. Finally, mass transfer between water and solid phase may be considered by employing [19b] to
obtain
r - r ROWS
1 as — l as ' T
Using (17) to eliminate oil, gas and solid phase concentrations from (14) and summing the equations noting
that
Raw +ROO+KOS +Raa = 0 (26)
leads to the a-component phase-summed transport equation in terms of water phase concentrations
81
-------
a * aCaw v * „ ,__ .
[£>a - ] -saraa +ras (2?b)
+qai T^ (27d)
(27e)
Pw Po Pa
Note that (27) has the same form as the simple single-phase equilibrium transport equation. However, the
coefficients represent pooled effects of transport in all phases and the equation is nonlinear due to dependence
of apparent partition coefficients on concentrations. Note also that interphase mass transfer terms occur in
the phase summed equation only as a sum over all components.
Solution approach. The basic approach for solving the coupled multiphase flow and multicomponent
transport problem with nonequilibrium mass transfer is as follows
1 . Solve the fluid flow equations simultaneously for the current time-step using time-lagged phase densities
and interphase mass transfer rates.
2. Solve the phase-summed transport equation using current values of apparent partition coefficients,
interphase mass transfer rates and phase densities.
3. Back-calculate interphase mass transfer rates, update phase densities and apparent partition coefficients
and repeat (2) until transport solution converges.
4. Proceed to the next time step.
An upstream- weighted finite element formulation for the coupled flow and transport problem was developed
and implemented in a computer code (details available from authors on request).
APPLICATION TO SOLVENT SPILL REMEDIATION
This problem involves a spill of tetrachloroethylene (PER) in a radial domain and remediation using
vacuum extraction. The simulation is performed in three stages with two restarts as follows:
82
-------
Stage 1: Oil infiltration event
Stage 2: Redistribution and transport under natural gradients
Stage 3: Remediation using gas vacuum extraction and/or water pumping
The first stage of the problem involves infiltration of PER on a circular area with a radius of 2.0 m at the soil
surface. A total of 3 m3 of PER is assumed to infiltrate under a water-equivalent oil head of-0.25 m. A water
table occurs at a depth of 3.2 m and an impermeable layer occurs 5 m below the surface. The soil is uniform
and has properties given in Table 1. Properties for PER are also given in the table. The problem is analyzed
as a 2-D radial section with an inner radius of 0.1 m (in order to facilitate subsequent analysis of pumping
from a well of the same radius) and an outer radius of 8 m, A mesh with 12 nodes in the vertical direction
and 16 nodes in the horizontal direction is employed.
Initial conditions for the water phase are assumed to correspond to equilibrium with the water table with no
oil. Boundary conditions for the water phase involve a type-1 condition on the right side below the water table
corresponding to vertical hydrostatic conditions (i.e., same as the initial conditions). All other boundaries
are no flow for the water phase. Boundary conditions for the oil phase are a type-1 condition with ho = 0.25
m on the 5 nodes on the upper surface between r = 0 and r = 2.0 m. All other nodes are no flow for oil. Gas
flow was not considered during infiltration. Termination of Stage 1 occurred at t = 6.48 d after infiltration
of 3.13 m3 of oil.
Stage 2 of the problem involves a continuation from the final conditions of Stage 1. Redistribution of PER
is permitted for a period of 25 d under no flow boundary conditions for the oil phase on all boundaries.
Boundary conditions for the water phase are maintained as in the infiltration stage. Gas phase flow is again
disregarded. Transport is considered with the initial aqueous concentration of PER at nodes where oil phase
is present set to the solubility of 150 g m"3. Boundary conditions for transport impose zero dispersive flux
on all boundaries which is equivalent to zero flux since there is essentially no flow on the boundaries during
Stage 2. By the end of the redistribution period, the NAPL plume has reached the lower aquifer boundary.
However, a substantial volume of the spill is still retained in the zone above the water table as "residual
saturation" retained by capillary forces (Figure 1).
In Stage 3, remediation is simulated with vacuum pumping in the unsaturated zone. A vacuum well with a
0.1 m radius is assumed to be placed at the left boundary screened from the surface to a depth of 3.2 m and
regulated at a pressure head of ha =-1.5 m. The gas boundary at the well is treated as a type-1 boundary
condition (specified pressure). Gas inflow is permitted along the upper surface from r = 3.5 m to the outer
perimeter under a type-1 condition for gas with a constant pressure of ha = O (atmospheric pressure). The
inner 3.5 m of the surface is assumed to be covered and is treated as no-flow. No flow conditions are imposed
for the oil phase on all boundaries. Boundary conditions for transport are type-3 with zero influent
concentration on the top and right side boundaries and zero dispersive flux elsewhere.
Two different scenarios are evaluated to assess alternative designs of the gas venting system. In the first
scenario, only gas pumping is considered. The water head is prescribed on the right boundary at the initial
conditions below the water table as in the earlier simulations and all other boundaries are treated as no flow
for water. During this simulation, the gas flow rate stabilizes at 319 m3 d1 after 2 days. The simulations were
analyzed with three different values of nonequilibrium partitioning coefficients: 0.2 d~\ 2.0 d'1 and 2
83
-------
The variation of PER mass remaining in the system versus time for all the three cases is presented in Figure
3. It is clear that the mass removed during venting is reduced due to nonequilibrium partitioning between
PER and air. From Figure 2, it is evident that constant gas pumping results in water table upwelling which
will affect recovery in the gas phase since the gas transmissivity, and hence the gas flow rate, is reduced, and
the amount of PER which is exposed to gas flow is also reduced.
The second scenario is performed to evaluate the effects of simultaneous pumping of air and water on
removal efficiency. The boundary conditions on the entire domain are the same as in the first scenario, except
on the left edge, where the water head is specified with a 50 cm drawdown to induce water pumping. As
before, the simulation is repeated for three different values of nonequilibrium partitioning coefficients. The
mass remaining in the system versus time is presented in Figure 5 for all three cases. The reduction of water
table upwelling resulting from simultaneous water and air pumping is shown in Figure 4. The mass removal
rate is approximately 15% greater with simultaneous water and gas pumping than gas pumping only for a
given mass transfer coefficient, reflecting the greater gas flow rate for the water and gas pumping scenario
(365 m3
In the above simulations, nonequilibrium mass transfer is not initiated until after 2.5 days to avoid numerical
instabilities. Thus, the results are identical up to 2.5 days for all three partitioning coefficients. Total PER
mass in the soil shows a small increase ( ~ 2.5%) during the first day, which reflects a numerical mass balance
error before the solution stabilizes and mass removal rates become nearly constant
Table 1. Input parameters for PER spill problem.
Soil properties: Bulk fluid properties:
= 2.3 /30w=l-77
swz= I.Qmd'1 riro =0.9 pro =1.62
$ = 0.4 Component properties (PER):
Sm=0.l Dgu = 0.85x10-4 m2^-1
Sor = 0.2 D%o = 0.95x10-4 m2^-1
a = 3.0m'1 D&, = 0.65m 2d-!
n = 2.8 TOO = 10,800.
A£ = O.lm r#0a =0.35
Aj = 0.02 m p« = 1.62 x 106 g m ~3
84
-------
CONCLUSIONS
Nonequilibrium effects were incorporated in a multiphase flow and transport model. A first order kinetic
model was employed with a phased-summed transport model using the concept of apparent partition
coefficients. The model was tested for two different scenarios to study the removal efficiency during soil
venting. Efficiency clearly decreases as the mass transfer coefficients decrease due to the increasing degree
of nonequilibrium. Removal efficiency can be enhanced with simultaneous pumping of air and water as the
latter enables control of water table upwelling which increases the gas flow rate and the volume of NAPL
exposed to gas flow.
At high velocities or low oil saturations, mass transfer coefficients may diminish sufficiently to result in
nonequilibrium behavior. Furthermore, in heterogeneous formations, apparent nonequilibrium behavior
may arise due to diffusive mass transfer limitations between high and low permeability zones. At the present
time, the kinetics of mass transfer in heterogeneous media is not well understood. Future efforts are needed
to better understand the efficiency of in situ gas venting systems.
REFERENCES
1. Abriola, L. M., 1984. Multiphase migration of organic compounds in a porous medium A mathematical
model, Lecture Notes in Engineering, Vol. 8, C. A. Brebbia and D. A. Orszag (eds), Springer-Verlag,
New York, 232 pp.
2. Abriola, L. M. and G. F. Finder, 1985. A multiphase approach to the modeling of porous media
contamination by organic compounds, 1. Equation development, Water Resourc. Res., 21,11-18.
3. Baehr, A. L., G. E. Hoag and M. C. Marley, 1989. Removing volatile contaminants from the unsaturated
zone by inducing advective air-phase transport, J. Contam. Hydrol., 4, 1-26.
4. Corapcioglu, M. Y. and A. L. Baehr, 1987. A composition multiphase model for groundwater
contamination by petroleum products, 1. Theoretical considerations, Water Resourc. Res., 23,191-200.
5. Falta, R. W. and I. Javandel, 1987. A numerical method for multiphase multicomponent contaminant
transport in groundwater systems, EOS (Trans. Am. Geophys. Union), 68(44), 1284 (abstract).
6. Faust, C. R., 1985. Transport of immiscible fluids within and below the unsaturated zone: Numerical
model, Water Resourc. Res., 21, 587-596.
7. Feenstra, S. and J. Coburn, 1986. Subsurface contamination from spills of denser than water:
chlorinated solvents, Calif. WPCA Bull, 23, 26-34.
8. Geller, J. T. and J. R. Hunt, 1989. Non-aqueous phase organic liquids in the subsurface: dissolution
kinetics in the saturated zone, paper presented at the Int. Symp. Processes Governing the Fate of
Contaminants in the Subsurface Environment, Int. Assoc. of Water Pollut. Res. and Control, Stanford,
Calif.
85
-------
9. Hochmuth, D. P. andD. K. Sunada, 1985. Groundwater model of two phase immiscible flow in coarse
material, Ground Water 23. 617-626
10. Hunt, J. R., N. Sittar and K. S. Udell, 1988a. Non-aqueous phase liquid transport and cleanup, 2,
Experimental studies, Water Resourc. Res., 24,1259-1269.
11. Hunt, J. R., N. Sittar and K. S. Udell, 1988b. Non-aqueous phase liquid transport and cleanup, 1,
Analysis of mechanisms, Water Resourc. Res., 24,1247-1258.
12. Johnson, P. C, M. W. Kemblowski and J. D. Colthart, 1990. Quantitative analysis for the cleanup of
hydrocarbon-contaminated soils by in-situ soil venting, Ground Water, 28,413-429.
13. Kaluarachchi, J. J., and J. C. Parker, 1989. An efficient finite element method for modeling multiphase
flow, Water Resourc. Res., 25,43-54.
14. Kaluarachchi, J. J., and J. C. Parker, 1990. Modeling multicomponent organic chemical transport in
three-fluid-phase porous media, J. Contam. Hydrol., 5, 349-374.
15. Kaluarachchi, J. J., J. C. Parker, and R. J. Lenhard, 1990. A numerical model for water and light
hydrocarbon migration in unconfined aquifers under vertical equilibrium, Adv. Water Resourc. (in
press).
16. Kuppusamy, T., J. Sheng, J. C. Parker, and R. J. Lenhard, 1987. Finite element analysis of multiphase
immiscible flow through soils, Water Resourc. Res., 23, 625-631.
17. Miller, C. T. and WJ. Weber, Jr., 1986. Sorption of hydrophobic organic pollutants in saturated soil
systems, J. Contam. Hydrol., 1,243-261.
18. Miller, C. T., M. M. Poirier-McNeil and A. S. Mayer, 1990. Dissolution of trapped nonaqueous phase
liquids: mass transfer characteristics, Water Resour. Res., 26, 2783-2796.
19. Osbome, M. and Sykes, J., 1986. Numerical Modeling of immiscible organic transport at the Hyde Park
landfill, Water Resource. Res., 22, 25-33.
20. Parker, J. C., J. J. Kaluarachchi, and A. K. Katyal, 1988. Areal simulation of free product recovery from
a gasoline storage tank leak; site. Proc. Conf. on Petroleum Hydrocarbons and Organic Chemicals in
Ground water: Prevention, Detection and Restoration, NWWA, Houston, TX.
21. Parker, J. C., and R. J. Lenhard, 1990. Vertical integration of three phase flow equations for analysis
of light hydrocarbon plume movement, Transp. Porous Media (in press).
22. Pfannkuch, H. O., 1987. Bulk and distributed parameter mass transfer models for determining the
source strength at all oil spill-groundwater interface, Proceedings 3rd Technical Meeting, B. J. Franks
(ed.), USGS Prog, on Toxic Wastes-Groundwater Contam., Pensacola, Fla., Open File Rep., 87-109.
86
-------
23. Powers, S. E., C. O. Loureiro, L. M. Abriola, and W. J. Weber, Jr., 1991. Theoretical study of the
significance of nonequilibrium dissolution of nonaqueous phase liquids in subsurface systems, Water
Resour. Res., 27,463-477.
24. Reeves, H. W. and L. M. Abriola, 1988. A decoupled approach to the simulation of flow and transport
of non-aqueous organic phase contaminants through porous media. Proc. 7th Int. Conf. on Computational
Methods in Water Resources, M.I.T. (Mass. Inst. Technol.), Cambridge, MA.
25. Schwille, R, 1975. Groundwater pollution by mineral oil products, Proceedings of the Moscow
Symposium, AISH Publ., 103, 226-240.
26. Sleep, B. E. and J. F. Sykes, 1989. Modeling the transport of volatile organics in variably saturated
media, Water Resourc. Res., 25, 81-92.
87
-------
900
1? 100 1M 280 370 4M SSO «40 730
1 1 1 1—
400 -
300 -
O
<^x
Nl
200 -
100 •
10
i I
500
400
300
200
100
100 190 260 370 460
R(cm)
550 C40 730
Figure 1. Oil saturation contours at the end of Stage 2.
Water Saturation with Air Pumping Only
500
400
O
i 500
- 400
- 300
- 200
- 100
10 110 210 310 410 S10 610 710
Radial Distance (cm)
Figure 2. Water saturation profiles at the end of Stage 3 with air pumping only.
-------
a
"3
5
4.9-
4.8-
4.7-
4.6-
4.5-
4.4-
4.3-
4.2-
4.1-
4
Air Pumping Only
K. 02 K- 2.0 K« EO.O
10 15 20
Time (days)
25 30
Figure 3. PER mass in the system at the end of Stage 3 with air pumping only.
Water Saturation with Air and Water Pumping
300
500
- 400
- 300
- 200
100
110 210 310 410 510 $10 710
Radial Distance (cm)
Figure 4. Water saturation profiles at the end of Stage 3 with air and water pumping.
89
-------
60
S
tl
n
ti
5
4.9
4.8
4.7-
4.6-
4.5-
4.4
4.3
4.2
4.1
4
Air & Water Pumping
K- 0.2
K- 2.0
—i—
5
10
—i—
15
~20~
Time (days)
K- zo.o
30
Figure 5. PER mass remaining in the system at the end of Stage 3 with air and water pumping.
90
-------
APPLYING COMPUTER SIMULATION MODELS TO
DESIGN LARGE-SCALE SOIL VENTING AND
BIOREMEDIATION SYSTEMS
/. C. WALTON, R. G. BACA, J. B. SISSON, A. J. SONDRUP, AND S. O. MAGNUSON
Idaho National Engineering Laboratory
Geosciences Group
P.O. Box 1625
Idaho Falls, Idaho 83415
ABSTRACT
Computer models have been developed to simulate the generation, release, transport, and cleanup of organic
contaminants from waste sites. The models for generation and release of organic vapors consider processes
such as deposition rates, surface coverage, container corrosion, and temperature and mixing effects upon
contaminant fugacity. Subsequent to release from source waste zones, contaminants move into the
environment in liquid, vapor, and/or aqueous solution form. Computer models have also been developed to
simulate subsurface migration of contaminants and in situ site remediation methods such as soil venting and
bioremediation. This paper discusses model theory and recent applications of the models to the design of a
large-scale soil venting system at the Idaho National Engineering Laboratory.
INTRODUCTION
Migration of organic contaminants from hazardous waste disposal sites is a significant environmental
concern. Organic contaminants, particularly in the vapor phase, can pose a health hazard to workers in the
vicinity of the disposal site and contaminate the underlying aquifer. Volatile organic compounds such as
carbon tetrachloride, chloroform, and trichloroethylene are frequently encountered at waste sites. These
chlorinated hydrocarbons are relatively common chemicals and widely used as industrial solvents.
Organic vapor plumes have been noted at waste disposal sites at a number of U. S. Department of Energy
(DOE) facilities. At the Idaho National Engineering Laboratory, for example, chlorinated solvent vapors
have been found at the Radioactive Waste Management Complex (RWMC). The primary source of the
solvents is thought to be drums of mixed waste from DOE's Rocky Flats Plant near Golden, Colorado that
were buried during the approximate period 1966 to 1971. The waste consists of plutonium-contaminated
chlorinated solvents including carbon tetrachloride mixed with machining oil and solidified with a
commercial absorbent. Over time the integrity of the drums has been lost from crushing and corrosion, and
the vapors have migrated downwards through the vadose zone toward the underlying aquifer and upwards
towards the atmosphere.
91
-------
As a result of past disposal practices at the RWMC, significant amounts of organic vapors have moved
through the vadose zone. Organic compounds have been detected at low levels in the Snake Plain Aquifer,
a major water source for Southeastern Idaho and the primary source of drinking water at the RWMC. In
addition to being used for irrigation, water from the aquifer is used to culture a large portion of the trout grown
commercially in the U.S. Because of concern for water quality, soil venting is being considered as a means
of removing volatile organic compounds from the vadose zone and reducing further contamination of the
Snake Plain Aquifer.
HISTORY
Low-level, transuranic, and hazardous wastes have been disposed of at the RWMC since the mid-1950s.
While the majority of the waste was buried in trenches, pits, and vaults, unspecified amounts of liquid wastes
were discharged into open pits. A significant portion of the waste (i.e., the portion from Rocky Flats) was
packaged in 208 liter (55 gal) drums and placed in pits and trenches dug in the surficial sediments. The drums
were randomly dumped into the pits or stacked. The pits were then backfilled and the soil compacted. This
past disposal practice may have lead to early failure and release of contaminants from a significant number
of drums.
Disposal records indicate that approximately 3.4 x 10s 1 (88,000 gal) of organic waste from the Rocky Flats
Plant were disposed at the SDA from 1966 to 1971. These wastes are particularly significant because they
consisted of approximately 9 x 1041 (24,000 gal) of carbon tetrachloride, 9.5 x 1041 (25,000 gal) of
miscellaneous organics consisting of chloroform, trichloroethylene, trichloroethane, and tetrachloroethlyene,
and 1.5 x 105! (39,000 gal) of Texaco Regal machining oil. These solvents, after being solidified with a
commercial absorbent, were placed and sealed in 208-1 (55-gal) drums.
Organic chemicals were first detected in the groundwater at the RWMC in September 1987 as the result of
sampling conducted by the U. S. Geological Survey. Analysis of samples from wells in the vicinity of the
RWMC showed concentrations above detection limits for a number of organic compounds (2). All
groundwater samples but one, however, were below the U.S. Environmental Protection Agency maximum
of 5 ppb for drinking water standards. Organic vapors were later detected during well drilling operations at
the RWMC.
A soil-gas survey, performed by Colder Associates, was conducted to identify the types and sources of
organic vapors near the surface of the RWMC. Gas samples were taken on a 40 x 30 grid on 61-m (200-ft)
centers. Holes were augured to a depth of 0.76 m (2.5 ft) at each grid point and a carbon steel pipe was driven
in the hole for gas extraction. Four chlorinated hydrocarbons were detected: carbon tetrachloride,
trichloroethylene, trichloroethane, and tetrachloroethlyene.
GEOLOGY
The RWMC is located on a thin layer of alluvial and eolian sediments of Quaternary age. Beneath the surface
alluvium is a thick sequence of basalt lava flows and sedimentary interbeds that extend to a depth of
approximately 610 m (2000 ft). Sedimentary interbeds are generally considered to represent quiescent
periods between volcanic episodes, when the top most lava flow was covered by accumulations of eolian and
alluvial sediments. A schematic geologic cross section through the RWMC is presented in Figure 1.
92
-------
The Snake Plain Aquifer is considered to be unconfmed, and the depth to water is about 180 m (580 ft).
Permeability of the Snake Plain Aquifer is controlled by the distribution of highly fractured basalt flow tops
and interflow zones; some permeability is contributed by fractures, vesicles, and intergranular pore spaces.
The variety and degree of interconnected water bearing zones complicates the direction of groundwater
movement locally throughout the aquifer. Sedimentary interbeds generally have lower permeability than the
surrounding basalt flows and impede the movement of groundwater. Small regions of perched water have
been encountered above major sedimentary interbeds during drilling activities at the RWMC. The perched
water zones are not continuous below the RWMC and appear to be associated with local infiltration
phenomena.
Depth (ft)
50
(164)
100.
(328)
150
(492)
200—
(656)
Surficial Sediments and Waste
Sedimentary Interbed
73-m (240-ft) Sedimentary Interbed Z
5L Basalt
-4-T-* ^—t~L^r~r*~ V-u r'.j'^r t,1 i .- i i V 1V i f \ r 11'
S Aquifer ^Qip^ffi-ffip£pS?ffiift
Figure 1. Schematic diagram of geology beneath the Radioactive Waste Management Complex
(RWMC). Of major significance are the sedimentary interbeds and the depth to the water
table.
93
-------
The clean-up of the site by soil venting will be complicated by the complex geology. In particular, perched
water zones and moist sediments are expected to respond poorly to induced airflow. The impermeable (to
air) nature of the moist regions will greatly slow contaminant removal rates. Vapor/liquid partitioning will
cause the moist regions to act as secondary sources of contaminants as constituents dissolved in the aqueous
phase are volatilized.
INITIAL MODELING WORK
Role of Modeling Analysis
The vadose zone plume of chlorinated solvents extends vertically to the aquifer at a depth of 180 m (591 ft)
and laterally over an area of greater than 60 ha (150 acres). The large size of the RWMC, combined with the
difficulty and expense of drilling through fractured, contaminated basalt rock, have led to the use of
mathematical models to assist with site characterization and remediation.
The modeling analyses are a means of better understanding and interpreting the available data. Through
applying numerical models it is hoped that fewer wells will be required to characterize the extent of
contamination and a more cost effective cleanup strategy can be planned. Modeling is not an exact predictive
tool or a replacement for monitoring data.
Four separate computer simulations or submodels have been developed and applied to the RWMC organics
problem: 1) source term (release rate from waste pits); 2) plume growth in the vadose zone and release into
the aquifer; 3) transport in the Snake River Plain aquifer; and 4) cleanup of the vadose zone plume by soil
venting. Additionally a model for bioremediation has been developed but not applied at the RWMC. Output
from the source term and plume growth models serves as input for the aquifer and soil venting calculations.
The source term, plume growth, soil venting, and bioremediation submodels are summarized below.
Source Term
The source term submodel covers the release of contaminants from waste drums and migration from the
disposal pits. This submodel is discussed in detail in Walton el al (1989). A major conclusion from the source
term model was that the long-term release rate of highly volatile constituents (e.g., carbon tetrachloride) is
controlled by drum failure rates. This conclusion was used to simplify the source term model to analytical
equations for the release of contaminants from waste drums. Other processes that exerted minor controls on
the release rate from the waste pits were ignored. The analytical solution provides a time variant source term
for the plume growth model that describes contaminant release to the vadose zone. The source term code also
gave strong evidence, based upon the physics of the problem, that liquid wastes would not migrate significant
distances from the disposal pits (3). Transport away from the pits is considered to be vapor dominated.
Retrieval studies at the RWMC were used to obtain estimates of drum failure rates. The Rocky Flats Plant
wastes were assumed to be disposed of at a uniform rate during the period 1966 to 1971. Quantitative
estimates of other amounts of chlorinated solvents disposed of at the RWMC were unavailable and assumed
to be zero.
94
-------
•5
E
CC
HI
280 -
260 -
240 -
220 -
200 -
180 -
160 -
140 -
120 -
100 -
80 -
60 -
40 -
20 -
10 12
TIME (yr)
Figure 2. Simulated release rate of carbon tetrachloride from the waste disposal pits. The solid line
represents a simplified analytical solution and the dotted line represents the full numerical
solution. The annual cycle is related to seasonal variations in soil temperature.
The source term evaluation is based upon limited data. All of the input data are highly uncertain and the basic
simplifying assumptions remain unvalidated and untested. Although a sensitivity analysis could be
performed, no quantitative basis exists for estimating total uncertainty. The only validation data is the shape
of the current subsurface plume which exists greatest concentrations around the 110-foot interbed. The
source term code predicts the peak subsurface concentrations in this area. Mathematically the phenomenon
results from a declining or dying release rate from the waste pits.
If a decision is made to leave the waste in place, then validation of the source term model would be warranted.
This could be accomplished by evaluating important processes and the actual inventory in the waste pits, not
a simple task. If the source is removed or destroyed by in situ vitrification then validation of the source term
model will not be necessary.
Plume Growth
The second submodel is the growth of the vadose zone plume. The plume growth model results are dependent
upon the source term calculations and geologic transport properties. As stated previously, the analytic
95
-------
simplification of the source term model is discretized to form a time variant source zone in the transport code.
The governing equations are solved in three dimensions with the finite difference flow and transport code
PORFLO (4).
The geologic setting and its geometry are defined for the model in terms of the strata type, thickness, and
textural characteristics. Available geologic information for the vadose zone beneath the RWMC has been
used to estimate the primary geologic parameters. In many cases important parameters, such as the effective
diffusion coefficient, were initially set with engineering judgment. As more data have become available,
some of the original simplifying assumptions have been reevaluated.
The contaminants of concern are highly volatile and only sparingly soluble. The dominant mode of mass
transport for these compounds in the vadose zone is vapor diffusion. Sinking of the plume caused by density
differences between the contaminant plume and uncontaminated soil air was initially thought to be
unimportant in the overall plume because of the relatively low concentrations of the contaminants, relative
to the permeability of the geologic media. However this simplifying assumption is currently being
reevaluated because experience with the soil venting equipment has demonstrated that air permeabilities in
the system are 1 to 2 orders of magnitude higher than initial estimates. For this reason, it is likely that density
effects are more important than previously thought.
The original simplifying assumptions were designed to simulate organic vapor transport in computer codes
designed for simulating of saturated groundwater flow and transport in geologic media. Transport modes
considered are diffusion/dispersion of organic vapors, advection of vapors dissolved in soil moisture, and
advection in soil air. The flux is expressed by diffusion in the vapor phase, advection in the vapor phase, and
advection in the liquid phase. Diffusion/dispersion in the liquid phase is not considered. Monitoring data
gathered to date on the nature and extent of the subsurface plume are qualitatively consistent with model
predictions of plume size and extent.
The most significant prediction of the plume growth model is that releases to the Snake Plain Aquifer will
increase if no remedial action is taken. This prediction is consistent with the shape of the measured subsurface
plume, but monitoring data from the aquifer are currently inadequate to either support or reject the prediction.
Mathematically, the prediction of increased release to the aquifer results from transport time lag. The leading
edge of the diffusional plume is just now beginning to reach the aquifer.
Because of gross simplifying assumptions in the source term and plume growth models, it is preferable to
calibrate the plume growth model using monitoring data from the vadose zone plume. The calibrated results
from the plume growth model can serve as initial conditions for modeling cleanup of the vadose zone plume.
Currently, six wells are instrumented with gas sampling ports. Gas samples are being taken from these wells
and analyzed for organic vapor concentrations. When the data collection is complete and quality assurance
requirements have been satisfied, the plume growth model will be calibrated to represent the current
subsurface plume, lessening the importance of information obtained from the source term submodel.
Bioremediation Modeling
The Geosciences Group at the Idaho National Engineering Laboratory has developed a computer code for
use in design and optimization of bioremediation systems. The code is primarily intended for use with in situ
96
-------
bioremediation activities, but it is equally applicable to optimizing operation of bioreactors. The model
considers fluid flow with allowances for a number of injection or extraction areas. A number of contaminants,
nutrients, and microbial populations can be considered. Microbial kinetics are isolated in a separate
subroutine which can be modified to reflect different forms of kinetic equations. The simultaneous equations
are solved in a two dimensional domain using the alternating direction implicit method. The finite difference
grid allows for variable node spacing. The current code is limited to simulating a single fluid phase
(presumably water, but air flow could be simulated) and is still in the format of a research code with no
published user's guide, independent verification, or benchmark testing.
ADDITIONAL DATA GATHERED AND EXPERIENCE
While this paper focuses on remediating the present subsurface organic vapor plume, data from and modeling
of the RWMC indicate that vapors are migrating directly from the source pits down to the Snake Plain
Aquifer. Because contaminant concentrations in the aquifer are below drinking water standards, the cleanup
strategy focuses on stopping further contamination of the aquifer. If the source of contaminants to the aquifer
is removed, remediation of the aquifer will not be necessary.
Soil Venting Strategy
One widely used method for remediating a site with organic vapor problems is soil venting. Potential
application of soil venting to the RWMC has been evaluated in a preliminary computer modeling study (1).
The effectiveness of two distinct pumping configurations were evaluated in the modeling study. These were:
(1) single well soil-gas pumping and (2) multiple well pumping with an impermeable cover placed on the
soil surface. Results of the preliminary study indicated that both configurations could be very effective in
removing a significant portion of the subsurface vapor plume. The rate of cleanup, however, depends greatly
on the average permeability and ratio of vertical to horizontal permeability (anisotropy) of the basalt, which
is currently uncertain because of lack of site specific data. Another large source of uncertainty comes from
vapor migration out of relatively impermeable zones which remain stagnant even with pumping.
A field demonstration of soil venting was conducted at the RWMC in the spring of 1990. The soil venting
demonstration consisted of a single extraction well that was pumped for a 4 month period and six monitoring
wells for gas pressure and contaminant concentration measurement. The extraction well was open between
the 110-ft and 240-ft interbeds, where the greatest level of contamination occurs. Before pumping, baseline
data were obtained at each sampling port for at least 5 weeks. Because of the variable nature of field gas
chromatograph results obtained to date, a long baseline period was required to differentiate analytical
uncertainties from actual changes in the subsurface plume.
The objectives of the field demonstration were two-fold: (1) to collect sufficient field data for use in
calibrating the organic vapor transport model and (2) to provide an interim remedial action and evaluate its
effectiveness. In addition, the soil venting demonstration was intended to provide a field-based method for
evaluating the potential effectiveness of a longer term soil venting program. During the 4 month test the soil
venting system operated for 2090 hours. 65 x 106 ft3 of gas was extracted which contained 430 kg of carbon
tetrachloride and 160 kg of trichloroethylene.
97
-------
Reconsidering Initial Modeling Assumptions
Although sophisticated three-dimensional models provide impressive graphic images and illusions of
knowledge, realistic understanding and prediction of large and complex subsurface phenomena is much
more difficult. In particular, several problems with the initial predictions were discovered as a result of the
4-month extraction test.
The first revelation was that initial predictions of the permeability of the fractured basalt layers were
approximately 1 or 2 orders of magnitude too low. This is actually favorable for an extraction program and
reflects an initial desire to make conservative projections concerning the likely success of vapor extraction
efforts. The increased permeability means that the extraction blower was undersized and could not provide
sufficient drawdowns. The small pressure drops at the extraction well did not produce measurable pressure
changes at the monitoring ports in the monitoring wells. Thus the test did not provide information on
important parameters such as the ratio of vertical to horizontal permeability in the fractured basalt.
The higher permeability also has implications for plume growth. Initial calculations (based upon the
erroneous initial estimates of basalt permeability) suggested that density-related vapor migration was not of
great significance for growth and migration of the organic vapor plume and that density effects were only
important in pockets of high vapor concentration. The increased permeability may indicate an increased
importance for density driven vapor migration during plume growth. Unfortunately because the extraction
test did not provide data on vertical versus horizontal permeability in the basalts, the likely importance of
density driven flow cannot yet be resolved with certainty.
Approximately 50% of the soil gas extracted from the production well originated from a basalt rubble zone
at a depth of 58 m (190 ft). The presence of the rubble zone increased the effective radius of the extraction
well by producing a more vertical soil gas movement pattern. Thus, the rubble zone at 190 feet will be an
important consideration in the design of future systems.
Diffusion from Stagnant Zones
Any modeling analysis of a complex system has a limited capability to deal with (a) geologic heterogeneities
in the subsurface and (b) complexities of multicomponent transport with non-ideal solutions, non-linear
sorption, poorly understood container failure, and multiple, unknown chemical reactions. These uncertainties
lead some analysts to conclude that computer modeling is a meaningless exercise. The usefulness of the
modeling program will be determined by how uncertainties are dealt with in the modeling exercises.
One factor that slows cleanup is the rate of vapor migration out of stagnant zones in the subsurface. Regions
of massive basalt with few fractures are sometimes found in the basalt flows of the Snake River Group. These
regions have lower permeability and may remain relatively stagnant during the soil venting. Cleanup times
for stagnant zones will be controlled by diffusion rates out of the blocks. Perched water is also found at several
locations in the subsurface beneath the RWMC. Because it is not feasible to characterize the entire area of
the site it is important to examine the potential importance of stagnant zones at least parametrically. Water
contained in the stagnant zones will slow the diffusion rates and serve as a secondary source of vapors
because of vapor/liquid partitioning. The diffusion time above assumes that vapor phase diffusion is
dominant. In perched water zones, liquid diffusion will predominate.
98
-------
The actual impact on stagnant zone diffusion at the RWMC may never be known in detail. Site characterization
at this scale is simply not practical. The potential importance of stagnant zone diffusion is examined with
a set of calculations which estimate under what conditions stagnant zone diffusion will control cleanup times.
An estimate of the potential impact of stagnant area diffusion at the RWMC is given in Figures 3 and 4. The
extraction flow rate is taken as 0.47 m3/s (1,000 cfm) over a thickness of 40 m (130 ft). The 40 m thickness
represents the distance between the two major sedimentary interbeds (Figure 1). The ratio of diffusion out
of stagnant zones is plotted relative to travel time to the well as a function of stagnant zone size and well
spacing. When the ratio exceeds one diffusion will control cleanup times. As long as diffusion out of stagnant
zones controls cleanup times, adding additional extraction wells and/or pumping at a higher rate will have
a limited effect on ultimate cleanup time. The calculations illustrate that, in the presence of great uncertainty,
a simple dimensional analysis of the situation may provide as much useful information as more sophisticated
models and large computer codes.
Stagnant Zone Diffusion
Vapor Dominated Blocks
0
block
size
(m)
0 300
100
200 well
spacing
(m)
Figure 3. Dimensionless time for diffusion of organic vapors out of large stagnant blocks of basalt
as a function of well spacing and block size. When the dimensionless time moves above 1.0
then diffusion from blocks will control cleanup times, not travel time to a well. In this
situation, closer well spacing and/or greater pumping rates do little to speed ultimate
cleanup of the site. Note that the graph is truncated at a dimensionless time of 2.
99
-------
Stagnant Zone Diffusion
from Perched Water Locations
td/tc
0.5
0.4
perched °-3
water 0.2
size o. 1
(m)
200
300
0 400
0
100
well
spacing
(m)
Figure 4. Ratio of diffusion time out of perched water zones relative to travel time to well. Numbers
greater than 1.0 indicate that diffusion will control ultimate cleanup time. The graph is
truncated at a dimensionless time of 3.
Although diffusion is likely to control cleanup time at well close well spacings, diffusion out of the vapor
dominated stagnant zones is relatively rapid compared to perched water zones.
Perched water zones are potentially the greatest source of problems and may take very long times for removal
of contaminants. Evaluations of the potential nature and extent of perched water may be required. The total
mass of contaminants in the perched water may be considered de minimus. Because most of the other organic
solvents have a lower Henry's Law constant than carbon tetrachloride their characteristic time for removal
will be longer. The increased difficulty in removal of the other contaminants is compensated by the much
lower concentrations in the subsurface and generally lower toxicity.
100
-------
SUMMARY
Modeling has been a useful tool for understanding the site and for design of potential cleanup strategies at
the Radioactive Waste Management Complex (RWMC). Models have been developed for all stages of the
problem including release rate of contaminants from the source, contaminant migration in the vadose and
saturated zones, and cleanup by soil venting. The usefulness of the modeling work is limited by the
complexity of the site and limited availability of characterization data. Despite several years of work on the
organic vapor problem, design and implementation of a final remediation system has not been accomplished
to date.
REFERENCES
1. Baca, R. G., Walton, J. C., Rood, A. S., and Otis, M. D., 1988, Organic contaminant release from a
mixed-waste disposal site: A computer simulation study of transport through the vadose zone and site
remediation, in Proceedings, Tenth Annual Department of Energy Low-Level Waste Management
Conference, CONF-880839-Ses. II, p. 113-125.
2. Mann,L. J., andKnobel,L.J, 1987, Purgeable organic compounds in ground water at the Idaho National
Engineering Laboratory, Idaho, U.S. Geological Survey, Open-File Report 87-766.
3. Rawson, S. A., Walton, J. C., and Baca, R. G., 1989, Modeling potential migration of petroleum
hydrocarbons from a mixed-waste disposal site in the vadose zone, Proceedings of: Petroleum
Hydrocarbons and Organic Chemicals in Ground Water: Prevention, Detection and Restoration,
National Well Water Association, Dublin, Ohio.
4. Runchal, A., Sagar, B., Baca, R. G., and Kline, N. W., PORFLO - a continuum model for fluid flow,
heat transfer, and mass transport in porous media: model theory, numerical methods, and computational
tests, Rockwell Hanford Operations, RHO-BW-CR-150-P, 1985.
5. Walton, J. C., Rood, A. S., Baca, R. G., and Otis, M. D., 1989, Model for estimation of chlorinated
solvent release from waste disposal sites, Journal of Hazardous Materials, Vol. 21, p. 15-34.
Acknowledgment
Work performed under the auspices of the U.S. Department of Energy, DOE Contract No. DE-AC07-
76ID01570.
101
-------
VOC CLEANUP BY IN SITU SOIL VAPOR EXTRACTION--
SOME RECENT DEVELOPMENTS IN MODELING
EPA SOIL VAPOR EXTRACTION WORKSHOP,
HOUSTON, TX, APRIL 29 - MAY 1,1991
JOSE M. RODRIGUEZ-MAROTO*, CESAR GOMEZ-LAHOZ*,
AND DAVID J. WILSON
Departments of Chemistry and of Civil and
Environmental Engineering
Vanderbilt University
Nashville, TN 37235 (615) 322-2633
*Present address Depto. de Enginieria Quimica, Universidad de Malaga, 29071, Malaga, Spain
INTRODUCTION
There are both field results and laboratory work indicating the importance, at some sites, of diffusion and/
or desorption kinetics limitations to soil vacuum extraction (SVE) (3,4,11), although the assumption that
the vapor and stationary phases are at local equilibrium with respect to volatile organic compound (VOC)
transport appears to be an adequate approximation at other sites (1,3). DiGiulio et al. (1990) have described
a field technique for ascertaining the extent to which diff usion/desorption kinetics may be operative. Hutzler
et al. (1989) have developed a lab column model which includes kinetics limitation. Wilson (1990) and Oma
et al. (1990) have published lab column and field vapor stripping well models which include kinetics
limitation. A major practical problem has been the large amount of computer time which these required. A
way around this difficulty was recently discovered in which a diffusion-limited SVE model required virtually
no more computer time per run than a typical local equilibrium model.
HIGH SPEED MODELING OF DIFFUSION KINETICS
For reasons of mathematical simplicity and usually for want of an adequate data base on a site, models for
SVE have generally assumed that the pneumatic permeability of the soil is constant and isotropic in the
domain of interest. However this is generally not the case. One usually finds strata of very different
permeabilities, clay lenses of quite low permeability, and other heterogeneities in the soil which may have
significant impact on the time required for cleanup. The first law of SVE application is that you must be able
to draw air through all of the contaminated soil. If well logs indicate that the soil permeability is likely to
be highly variable, models should be able to explore at least qualitatively the impact this may have on SVE
cleanup times, and should provide some sort of estimate as to the uncertainty in the calculated cleanup times
which this heterogeneity causes. We have extended a model for SVE with horizontal slotted lateral pipes to
include the possibility of low-permeability lenses, as well as strata of different permeabilities. Results with
these models suggest ways in which boring log data can be used to design an SVE system for a particular
site which is minimally affected by such heterogeneities.
103
-------
Our two-dimensional model for diffusion-limited SVE well operation makes use of a lumped parameter
approximation to handle diffusion kinetics and can be run on a microcomputer (12). Although the results of
the model appear quite reasonable, it suffers from the fact that one must use very small values of the time
increment when doing the numerical integrations required by the model. Mechanical engineers would say
that the differential equations used in the modeling are quite stiff. When the model is used on a 20 MHz PC-
AT clone microcomputer with a math coprocesser, times required for physically realistic simulations are
typically 60 hours or more per run. This severely limits the usefulness of that model for the simulation of
SVE operations which are diffusion controlled; one would like to have a model which ran at least ten times
faster than our first attempt. One of the principal reasons to do modeling is so that you can explore quickly
and cheaply a large number of designs and options. Obviously this cannot be done if each simulation takes
60 hours.
EFFECTS OF VARIABLE PERMEABILITIES AND SYSTEM GEOMETRY ON THE
TIME REQUIRED FOR SVE CLEANUPS
Here we present models for soil vapor stripping in lab columns and by means of field vapor extraction wells
which include diffusion/desorption limited kinetics and which utilize a steadystate approximation similar
to that used in chemical kinetics to simplify the rate equations arising in that field.
In the following models we make the steady state approximation for the vapor phase VOC concentrations.
That is, we assume that the vapor phase concentrations are sufficiently small that the mass of contaminant
VOC in the vapor phase is an almost negligible fraction of the total mass of VOC present in the system. If
this is the case, then the net rates of change with time of the vapor phase VOC concentrations with time will
be extremely small in comparison to their rates of change by diffusion (replenishment) and their rates of
change by advection (removal). The steady state approximation cosists in setting these very small net rates
of change of VOC concentrations in the vapor phase equal to zero. This type of approximation is very
commonly used to simplify the analysis of the kinetic mechanisms of complex chemical reactions; one sets
the net rates of change of highly active intermediate species present at extremely low concentrations (atoms,
free radicals, etc.) equal to zero. In chemical kinetics the approximation is well understood and well
established; our use of it in this new context will require justification by comparison with the results of
calculations in which it is not made.
One-Dimensional Lab Column Model
Our model for SVE in a lab column with diffusion kinetics is diagrammed in Figure 1, which also includes
notation. We use SI units throughout. As with our previous diffusion-controlled model, we use a lumped
parameter method for approximating diffusion and desorption kinetics; estimation of the rate parameter for
diffusion has been discussed earlier (12). The lumped parameter approach assumes that diffusion into and
out of the blocks of low-permeability porous medium is governed by Equation (1).
104
-------
Included here is the assumption that partitioning of VOC between the vapor and stationary phases is linear—
that the system obeys Henry's Law, with an effective Henry's constant which typically is substantially
smaller than the Henry's constant for the VOC in aqueous solution.
If we focus only on diffusion transport between the stationary and vapor phases in the ith compartment (a
conservative process), we have Equation (2), which, together with Equation (1), gives Equation (3).
vAAx
dc?
dtl
diff
dc?
+wAAx :ii!_ = 0
dt
[2]
vAAx
dt)
= -XwAAx
diff
KH
-c?
[3]
A mass balance on the vapor phase in the ith compartment gives Equation (4) for the rate of change of vapor
concentration in the ith compartment with time.
vAAx
dt
= Vj-i/2 VA CY_! - Vi+i/2 VA
1 „ S
4
[4]
This is set equal to zero by virtue of the steady state approximation for the vapor phase concentrations.
Solving Equation (4) for the vapor phase concentrations then yields Equations (5).
WA.
-
Vl/2 +
Ax
[5]
Vi+1/2 ,
\ Ax
Equation (1) is then integrated forward in time, with the vapor phase concentrations cvj being calculated at
every step from Equation (5). This can either by done by a standard predictor-corrector method or one can
rearrange Equations (1) to Equation (6), assuming that the cvj remains essentially constant during the time
increment in the integration, At. This yields Equations (7).
dci
~T~
dt
s
~^-
KH
[6]
105
-------
c? (At) = cf (o) exp( -AAt) +
-------
of low permeability. The parameters for these runs were given in Table 1; only the diffusion rate constant was
varied. A calculation was made using our local equilibrium model for comparison purposes. Three
calculations were made for each of the values of A, shown in Figure 2. These were (1) a calculation using our
earlier "exact" model (i.e., without the steady state approximation); (2) a calculation in which the predictor-
corrector method was used for the numerical integration, and (3) a calculation in which the integrated form
of the differential equations was used for the numerical integration. All three models which included
diffusion gave results which are virtually identical; plots at the scale of Figure 2 are completely
indistinguishable. These numerical results are in agreement with our analysis above of the lab column model,
which showed excellent agreement between the steady-state diffusion model with a very large diffusion rate
constant and the local equilibrium model. It was found that the value of 2At must be 0.01 or less if reliable
results are to be obtained with the steady state models; At was set equal to 100 sec for the runs with A = 104 sec'1.
For this system and at this gas flow rate the local equilibrium results were indistinguishable from those
obtained when "k = 10~4 sec"1. As decreases below 10'5 sec"1, however, the rate of VOC removal becomes
progressively slower.
Another set of runs was made in which the effective Henry's constant KH was reduced from 10"2 to a value
of 10"3; these are shown in Figure 3. (Note the change in scale of the time axis between Figures 2 and 3.) We
found that, if all of the system parameters except A, and KH are held constant and that KH is varied proportional
to A, then the cleanup time is proportional to X"1. A comparison of Figures 2 and 3 also shows that the
damaging effects of a small effective Henry's constant are not limited to local equilibrium and near-local
equilibrium conditions.
In Figures 4 and 5 the effects of overlying impermeable caps are investigated for a range of values of the
diffusion rate constant. Figure 4 compares runs having no cap with runs having a cap of radius 6 m. We see
that a cap results in greatest percentage reduction in cleanup time (to, say, the 99.9% level) if the diffusion
rate constant is large; for slow diffusion rates the cap has a negligible effect. Figure 5 compares runs made
with caps of 4 and 8 m radius for systems having a range of diffusion rate constants. The larger caps give
shorter cleanup times, but, as in Figure 4, the differences decrease as the diffusion rate constant becomes
smaller. These results are as one would intuitively expect, since the effect of a cap is to increase the efficiency
of the gas flow pattern by increasing gas flow rates in the relatively stagnant portions of the domain out near
the periphery of the cylinder; however, the efficiency of the gas flow pattern in advection becomes less and
less important as diffusion rates become more and more limiting.
The effects of the presence of a set of passive vent wells located around the periphery of the domain of interest
and screened along their entire length are shown in Figures 6 and 7. No impermeable cap is present in these
runs. For this particular geometry, the passive wells result in relatively little changes in removal efficiencies;
with the local equilibrium run and the run with the largest diffusion rate constant (10"s sec'1) the presence of
passive wells may actually reduce efficiency toward the end of the run. As the diffusion rate constant becomes
very small, the effect of the passive wells on cleanup time is seen to decrease on a percentage basis, as one
would expect.
The situation is rather different if these passive wells are combined with overlying impermeable caps. In
Figure 8 runs were made with or without impermeable caps of radius 10 m and passive vent wells screened
107
-------
along their entire lengths. The combination of caps and passive wells results in significant increases in
removal rates for diffusion rate constants of 5 X 10"7 sec"1 and larger. As before, as diffusion becomes slower
it overpowers the effects of anything one can do to the gas flow pattern.
In Figure 9 runs were made with and without caps and passive wells screened along their entire lengths; the
radius of the impermeable cap is 8 m. In this configuration the passive wells are highly beneficial, more than
doubling the rate of removal for the local equilibrium model, and resulting in very substantial increases in
removal rates for diffusion rate constants greater than 5 X 10"7 sec"1. If one is under some pressure to meet
a rather short deadline for cleanup of a site, these results suggest that money might well be spent on a
configuration involving both caps and passive wells.
The effects of gas flow rate are shown in Figures 10 and 11. In Figure 10 we see that cleanup rates for the
local equilibrium model are directly proportional to the soil gas flow rate, exactly as one would anticipate.
As diffusion processes become more limiting, however, the payoff achieved by increasing the gas flow rate
becomes less and less. In Figure 11 it is apparent that doubling the gas flow rate from 2.75 to 5.5 mol/sec
results in only a 20% increase in cleanup rate if the diffusion rate constant is 10"6 sec"1. If this parameter is
10"7 sec"1, increasing the gas flow rate from 2.75 to 5.5 mol/sec results in a negligible increase in removal rate.
One concludes that diffusion limitations should be explored in some depth during the pilot phase of a SVE
feasibility study as suggested by DiGiulio et al. (1990).
EFFECTS OF VARIABLE PERMEABILITIES AND HORIZONTAL WELLS ON THE
TIME REQUIRED FOR SVE CLEANUPS
In this section, the results obtained with a two-dimensional local equilibrium model for SVE with horizontal
lateral slotted pipes is presented. It is assumed that the horizontal laterals are long in comparison to the
spacing between them, so that we can use a two-dimensional Cartesian coordinate representation and ignore
end effects. We assume that the permeability is anisotropic and is a function of position, so the calculation
of soil gas pressures and velocities in the domain of interest must be done numerically; we use an over-
relaxation method which has done well for us previously (9). The model assumes a linear, Henry 's Law type
isotherm. The details of the model will be published shortly (5).
The molar gas flow rate per unit length of slotted pipe is given by Equation (10).
^ RT
Soil gas streamlines and their transit times are very helpful in quickly identifying regions in the domain of
interest which will clean up slowly. The streamlines are calculated by numerical integration of Equations
(11) and (12), where the pressure derivatives are calculated by use of finite differences and Taylor series
representations.
<£ = -Kx(x,y) — [11]
dt 9x
108
-------
The movement of VOCs is given by numerical integration of Equation (13).
9m _ Vvm
a* ~ v~+
[13]
This equation contains no dispersion term; this is represented by the effects of numerical dispersion in the
finite difference mesh used for the numerical integration.
Eight runs were made to explore the effects of a single low permeability domain on the soil gas streamlines.
The permeability function used to represent m such domains is given in Equation (14).
Kx (x,y) = Kxo - V Ai exp NP^H + ^ t14!
The standard parameter set is given in Table 2. Figure 12 shows the streamlines and gas transit times (in
thousands of seconds) for the soil gas to move from the surface of the ground to the horizontal vacuum pipe
forrun number 1, for which no low-permeability lenses were present. Figure 13, run no. 2, shows streamlines
and transit times for a case where the lens is centered over the extraction well and a short distance above it.
The transit times are markedly different from those obtained in the absence of a lens (Figure 12), but none
are appreciably larger than the maximum time seen in Figure 12, nor is there any part of the domain from
which gas flow is excluded. We therefore expect that the presence of a lens in this position will have little
effect on the overall rate of cleanup. Shortly we shall see that this is in fact the case.
Run 3, shown in Figure 14, shows the effect of a low-permeability lens above the well and somewhat to the
left. The somewhat increased gas transit times on the left side indicate that the cleanup time for Run 3 will
be somewhat longer than those for Runs 1 and 2.
Run 4, in Figure 15, illustrates poor placement of a vacuum well relative to a low-permeability lens. The lens,
far on the left side of the domain of influence, has markedly reduced the flow of soil gas through the lower
left portion of the domain, as indicated both by the long transit time of the leftmost streamline and the shapes
of the streamlines. We therefore expect a substantial decrease in cleanup rate for this run.
Run 5, shown in Figure 16, represents total disaster in the placement of a vapor stripping well. The well is
screened right in the middle of the low permeability lens. This has reduced the molar gas flow rate from a
value of 0.218 mol/sec (no lens) to a value of 0.0265 mol/sec, one-eighth of the reference flow rate. The
transit times of the soil gas have been correspondingly increased, and we expect a greatly reduced rate of
cleanup.
109
-------
Figure 17 shows plots of Iog10 (total contaminant mass) versus time for runs 1 through 5. We see that, as
expected, the lens in run 2 (Figure 13) has had little effect on the rate of cleanup of this system. The off-center
lens in run 3 (Figure 14) has slowed down the rate of cleanup slightly. In run 4 (Figure 15), the lens which
was located far on the left side of the domain has very markedly reduced the rate of cleanup; the slope of the
plot for run 1 is 2.25 times the slope of the plot for run 4 in the linear sections of the plots. The plot of Iog10
contaminant mass versus time for run 5 shows that the effect of screening the well in the middle of the low
permeability lens was indeed disastrous, due principally to the decreased gas flow rate through the well in
run 5 as compared to run 1. 99% cleanup in run 1 required 15.9 days; in run 5 it required 85 days.
Runs 6,7, and 8 were made with two low permeability lenses located out toward the sides of the domain of
influence of the well; these were placed low-low, low-high, and high-high. As seen in Figure 18, the rates
of cleanup in these runs are rather similar; all are substantially lower than the rate of cleanup of the system
without any low permeability lenses.
Another series of runs were made with the model parameters specified in Table 3. The parameters describing
the low permeability lenses are given in Table 4. The locations of the centers of the lenses are shown in Figure
19.
The main point of interest with these runs is the spread in the rates of cleanup which result from the various
locations of the lenses. This is shown in Figures 20 and 21. Again we find that lenses near the middle of the
domain cause relatively little decrease in cleanup rate, while lenses out at the edges are fairly damaging. Run
17, in which the lens is located directly underneath the vapor stripping well, has the smallest cleanup rate
of the set. The differences in the positions of the lenses in these runs result in a variation in 99.9% cleanup
time from a low of about 700 hours to a high of about 1200 hours with these parameters. Runs 10,14, and
16 show cleanup rates which are slightly greater than the cleanup rate of run 9, which has no lenses at all.
These runs indicate a measure of the uncertainty in cleanup times which may be associated with variations
in the soil permeability; the lenses have permeabilities which may be as little as 5% of the permeability of
the surrounding matrix. These typically cause variations of the order of -10 to + 50% in cleanup times, and
may, in particularly unfavorable cases, may cause increases in cleanup times by as much as 500%.
The runs also indicate the advantages to be gained by taking into account the inhomogeneities in the
permeability when designing a soil vapor extraction system, provided that the necessary data are available.
Qualitative data of the sort needed might readily be obtained from boring logs, cone penetrometer tests, etc.
The runs show that the damaging effects of such low permeability domains can often be reduced to an
acceptable level by proper placement and design of the well system. Low permeability lenses centered near
the border separating the zones of influence of two adjacent wells are likely to be damaging. The screening
of a well within or over a region of low permeability is likely to be very damaging, and should be avoided
if at all possible. The other configurations studied yielded cleanup rates only slightly different from the
cleanup rate found for a homogeneous domain of influence.
110
-------
REFERENCES
1. A. L. Baehr, G. E. Hoag, and M. C. Marley, 1989, "Removing Volatile Contaminants from the
Unsaturated Zone by Inducing Advective Air Phase Transport", J. Contain. Hydrology, 4,1.
2. D. C. DiGiulio, J. S. Cho, R. R. Dupont, and M. W. Kemblowski, 1990, "Conducting Field Tests for
Evaluation of Soil Vacuum Extraction Application", Proc., 4th Natl. Outdoor Action Conf. on Aquifer
Restoration, Ground Water Monitoring and Geophysical Methods, May 14-17, Las Vegas, NY
3. E. W. Fall, et al., 1988, "In Situ Hydrocarbon Extraction: A Case Study", Southwestern Ground Water
Focus Conference, Albuquerque, NM, Mar. 23-25; see also The Hazardous Waste Consultant, Jan/
Feb., 1989, p. 1-1.
4. J. S. Gierke, N. J. Hutzler, and J. C. Crittenden, 1990, "Modeling the Movement of Volatile Organic
Chemicals in Columns of Unsaturated Soil", Water Resources Res., 26,1529.
5. C. Gomez-Lahoz, J. M. Rodriguez-Maroto, andD. J. Wilson, 1991, "Soil Cleanup by In Situ Aeration.
VI. Effects of Variable Permeabilities", Separ. Sci. Technol., 26, 133.
6. N. J. Hutzler, D. B. McKenzie, and J. S. Gierke, 1989, "Vapor Extraction of Volatile Organic Chemicals
from Unsaturated Soil", in Abstracts, International Symposium on Processes Governing the Movement
and Fate of Contaminants in the Subsurface Environment, July 23-26, Stanford, CA.
7. N. J. Hutzler, B. E. Murphy, and J. S. Gierke, 1989, "Review of Soil Vapor Extraction System
Technology", Soil Vapor Extraction Technology Workshop, U.S. EPA RREL, June 28-29, Edison, NJ.
8. D. A. Keech, 1989, "Subsurface Venting Research and Venting Manual by the American Petroleum
Institute", Workshop on Soil Vacuum Extraction, April 27-28, RSKERL, Ada, OK.
9. R. D. Mutch, Jr., A. N. Clarke, and D. J. Wilson, 1989, "In Situ Vapor Stripping Research Project: A
Progress Report", Proc., 2nd Ann. Hazardous Materials Conf ./Central, Mar. 14-16, Rosemont, IL.
10. K. H. Oma, D. J. Wilson, and R. D. Mutch, Jr., 1990, "In Situ Vapor Stripping: The Importance of
NonequilibriumEffects in Predicting Cleanup Time and Cost", Proc., Hazardous Materials Management
Conf. and Exhibition/Intern., June 5-7, Atlantic City, NJ.
11. R. J. Sterrett, 1989, "Analysis of In Situ Soil Air Stripping Data", Workshop on Soil Vacuum
Extraction, Apr. 27-28, U.S. EPA RSKERL, Ada, OK.
12. D. J. Wilson, 1990, "Soil Cleanup by In Situ Aeration. V. Vapor Stripping from Fractured Bedrock",
Separ. Sci. Technol., 25, 243.
13. D. J. Wilson, A. N. Clarke, and J. H. Clarke, 1988, "Soil Cleanup by In Situ Aeration. I. Mathematical
Modeling", Separ. Sci. Technol., 23,991.
Ill
-------
Mo
va;
A
T
A =
= n
= n-1
|
T
T
A =
T
= i+l
: i
= i-1
4
T
A :
1
: 2
: 1
bile Stationary
)or Phase
Vapor advective
transport
= Diffusion/
desorption
Figure 1. Model for soil vapor stripping in a laboratory column; mathematical partitioning.
cy
M
P?
~
cross-sectional area of column
soil air-filled porosity
soil volumetric moisture content
linear gas velocity between the ith and (i + l)th compartments into which the column
is mathematically partitioned
vapor phase VOC concentration in the ith compartment
stationary phase(s) VOC concentration in the ith compartment
Table 1. Standard parameter set for simulations of a soil vapor stripping well
Parameter
Value
Radius of domain of influence
Depth of water table
Depth of well
Radius of impermeable cap
Gravel-Packed radius of well
Wellhead pressure
Temperature
Gas-filled porosity
Water-filled porosity
Pneumatic permeability. Kz
Pneumatic permeability, Kr
Effective Henry s constant
Initial contaminant concentration
Soil density
Molar gas flow rate
Volumetric gas flow rate
10m
8m
6 m
Om
0.12m
0.866 atm
12° C
0.2
0.2
0.6206 mVatm sec
0.6206 m2/atm sec
0.01 (dimensionless)
lOmg/kg
1.7 gm/cm3
1.102mol/sec
0.02579 mVsec
112
-------
"3
2
5
o
8
-i
-2
0
10
15
20
25xlO-6 sec
Time
Figure 2. Plots of Iog10 total contaminant mass versus time; effect of diffusion rate constant "L
From the top down, values of X are 10", 107,5 x 107,10 *, 105, and 104 sec1. The plot with X = 10 4
sec1 is indistinguishable at this scale from a plot with X = oo (the local equilibrium model). Other
system parameters are given in Table 1. No passive wells or impermeable caps are present.
2r
o
§
0
-1
-2
0
10
15
20 25x10-6 sec
Time
Figure 3. Plots of log^ Mtotal versus time; effect of diffusion rate constant X. KH for these runs has
been reduced to 0.001. From the top down, values of X are 10"8,10'7,10'6,10's sec'1 and (local
equilibrium).
113
-------
2r
a
Q
bfi
O
0
-1
-2
0
10 15
Time
20 25x10-6 sec
Figure 4. Plots of Iog10 Mtota| versus time; effect of an impermeable cap of 6 m radius. Runs made
without a cap are indicated with dashed lines. From the top down, values of A, are 10'7,5 x 10'7,10'*
sec'1, and °° (local equilibrium).
2r
1
O
o
0
-1
-2
0
10
15 20
Time
25 30x10-6 sec
Figure 5. Plots of Iog10 MtoU1 versus time; effects of impermeable caps of 4 m radius (dashed lines)
and 8 m radius (solid lines). Values of X are 5 x 107,106 sec'1, and °° (local equilibrium).
114
-------
20 25x10-6 sec
-2
0
Figure 6. Plots of Iog18 Mtotal versus time; effects of passive wells screened along their entire length
and located around the periphery of the domain. Runs made with passive wells present (solid
lines) and absent (dashed lines) have values of X of 10', 10s secx and °o from the top down.
2r
o
-i
-2
0
5 10 15 20 25xlO-6 sec
Time
Figure 7. Plots of Iog10 Mtotal versus time; effects of passive wells. Runs with passive wells present
(solid lines) and absent (dashed lines) have values of X of 104,107 and 105 sec *.
115
-------
2r
0
-1
-2
0
5 10 15 20 25x10-6 sec
Time
Figure 8. Plots of Iog10 Mtotal versus time; effects of passive wells combined with a 10 m
impermeable cap. Runs with passive wells and caps present are indicated with solid lines; runs
without wells and caps are shown with dashed lines. Values of A, are 10~7,5 x 107 sec"1, and 10"* sec"1
from the top down.
2r
0
-1
-2
0
5 10 15 20 25x10-6 sec
Time
Figure 9. Plots of Iog10 MtoUI versus time; effects of passive wells combined with an 8 m
impermeable cap. Runs with passive wells and caps present are shown with solid lines; runs
without wells and caps are shown with dashed lines. Values of X are 5 x 10"7,10"' sec1 and <» (local
equilibrium) from the top down.
116
-------
25x10-6 sec
Figure 10. Plots of logxo MtoUI versus time; effects of gas flow rate. No passive wells or caps are
present. Runs made with the local equilibrium assumption are shown with dashed lines. Runs
having A, = 5 x 10'7sec-1, are shown with solid lines. Gas flow rates in each of the two sets of runs
are 1.05,2.75, and 5.5 mol/sec from the top down.
o
o
0
-1
-2
0
10
15
20
25x10-6 sec
Time
Figure 11. Plots of Iog10 Mtota| versus time; effects of gas flow rate. No passive wells or caps are
present. Runs made with A, = 107 sec J are shown with dashed lines; runs made with A, = 10 * sec -1
are shown with solid lines. Gas flow rates are 1.05,2.75, and 5.5 mol/sec from the top down in each
set.
117
-------
Table 2. Standard Parameter Set for the Runs shown in Figures 12-21
Domain length
Domain depth
dx,dy'
Location of well
Packed radius of well
Wellhead pressure
Temperature
Soil gas-filled porosity
K K
r, z
Initial soil contaminant concentration
Soil density
Specific moisture content
Effective Henry's constant
13m
8m
1m
x -= 6.5 m, y = 0.5 m
0.2m
0.85 atm
14°C
0.3
1.100m2/atm»s
lOOmg/kg
1.7g/mL
0.2
0.005
18.3 13.7 11.1 9.5 8.3 7.6 7.6 8.3 9.5 11.1 13.7 18.3
Figure 12. Streamlines in the absence of low permeability lenses (Run 1). The numbers at the top
of the streamlines are the gas transit times in units of 1000 s. See Table 2 for model parameters for
Runs 1-8.
118
-------
18.7 14.7 12.7 12.2 13.1 17.2 17.2 13.1 12.2 12.7 14.6 18.6
Figure 13. Streamlines in the presence of a lens centered over the vapor extraction well. The lens is
centered at the point (6.6,3). A = B = 0.095 m2/atm s, r = 3 m, s = 1 m, n = 2. Run 2.
21.5 22.5 25.6 14.1 9.5 7.5 7.2 7.6 8.8 10.4 12.9 17.3
Figure 14. Streamlines in the presence of a lens in the upper left portion of the domain, centered at
(3,5). Other lens parameters are in Figure 13. Run 3.
119
-------
29.0 16.0 11.3 9.2 8.0 7.3 7.4 8.0 9.2 10.8 13.4 17.9
Figure 15. Streamlines in the presence of a lens in the far left portion of the domain, centered at (1,
4). rl = 4, Sj = 1 m, other parameters as in Figure 13. Run 4.
102 90 91 101 114 114 114 114 101 91 89 100
Figure 16. Streamlines in the presence of a lens surrounding the screened section of the well,
centered at (6.5, 2). rx = s1 = 4 m, other parameters as in Figure 13. Run 5.
120
-------
•a
Jo
O
-1
-2
0
5xl05sec 10
t
15
20
Figure 17. Plots of Iog10 (total contaminant mass, MtoU|) versus time for Runs 1 through 4.
1 0
-1
-2
786
0 Ixl06sec 2
t
Figure 18. Plots of Iog10 (M ) versus t for Runs 6,7,8.
121
-------
Table 3. Geometrical and Physical Parameters for the Model, Runs 9 -17.
Depth of domain 12 m
Width of domain 25 m
dx 1m
dy 0.6 m
Wellhead pressure 0.75 atm
Height of bottom of screened well
section above the water table 1.2m
Height of top of screened well
section above the water table 3.0 m
Effective Henry's constant 0.001
Initial total contaminant mass 10 kg
Permeabilities Kxo and K o 0.1 m2/atm • s
Temperature *° * 25°C
Table 4. Parameters Describing the Lenses, Runs 9-17
A 0.095 mVatm • s
B 0.095 m2/atm • s
r. 6.0 m
s. 1.5m
n 1
122
-------
12
11
10
15
14 and 16
13
16
run #9 - no lens
17
Figure 19. Positions of the centers of the low permeability lenses for Runs 10-16. Run 9, the
reference run, has no lenses. See Tables 3 and 4 for parameter values.
1000
Figure 20. Plots of log^ (MtoU1) versus t for Runs 9-12, as indicated by the numbers by the plots.
See Figure 19 and Tables 3 and 4 for parameter values.
123
-------
-i
-2
17
250hr
500
t
750
1000
Figure 21. Plots of log^ (Mtote|) versus t for Runs 9 and 13-17 as indicated by the numbers by the
plots. See Figure 19 and Tables 3 and 4 for parameter values.
124
-------
DEVELOPMENT AND APPLICATION OF A THREE DIMENSIONAL AIR
FLOW MODEL IN THE DESIGN OF A VAPOR EXTRACTION SYSTEM
MICHAEL C. MARLEY
Vapex Environmental Technologies, Inc.
Canton, Massachusetts
ABSTRACT
The long term environmental and public health threats posed at sites with contaminated vadose zone soils
are only now attaining the attention required. Innovative technologies are being developed to aid
remediation of these contaminated soils. Vapor extraction (soil venting) of volatile and semi-volatile
petroleum hydrocarbons is rapidly being recognized as a technically sound and economical remediation
alternative.
To date, the majority of vapor extraction system designs have been based on long term pilot studies or non-
engineered, less economical installations based largely on the principle of trial and error. Engineered system
design depends on knowledge of the site soil properties in relation to air flow. The relevant soil properties
includes the relative intrinsic permeabilities of each geologic unit and of the confining unit(s) within the soil
system.
This paper presents the development of a three dimensional compressible fluid (air) flow model, the
utilization of the model in the evaluation of a site's soil properties in relation to air flow using the data
collected during the conduct of a short term field test, and the application of the model in the design of an
optimal, cost effective, full scale vapor extraction system. The model provides management with a
scientifically based decision making and design tool.
INTRODUCTION
The fate and transport of volatile, semi-volatile, and gaseous contaminants released into the subsurface
environment has become a subject of primary concern the past two decades. It has been estimated that up
to 20 percent of the approximately 2 million federally regulated underground storage tanks in the United
States may be leaking (15). Spilled product migrates through the unsaturated soil zone, under the influence
of gravitational and capillary forces, to the water table. Corrective action generally includes an effort to
physically remove the product by bailing and pumping as well as pumping and treating contaminated ground
water. The product retained in the unsaturated zone, however, can be a significant portion of the total spill.
For example, Hoag andMarley (1986) determined residual saturations of acommercial gasoline to be 26 and
44 grams per kilogram of medium and fine sand, respectively, at field moisture conditions. Baehr (1987),
Sleep and Sykes (1989) have developed natural transport models and demonstrated the potential for long-
125
-------
term ground water contamination due to vapor and solute transport emanating from the trapped immiscible
plume in soils above the ground water table.
In the past few years, the need to remediate these contaminated unsaturated soils as part of a comprehensive
and cost effective site clean-up has been emphasized. There are a number of methodologies commonly
utilized in the remediation of contaminated unsaturated soils including:
• excavation and off-site disposal
• excavation and on-site treatment
• biodegradation
• in-situ soil washing
• in-situ vapor extraction (soil venting, air stripping, enhanced volatilization)
In general, it is recognized that where applicable, vapor extraction is the most cost effective alternative.
Vapor extraction entails the induction of an advective air phase through contaminated vadose zone soils. In
the unsaturated zone, the optimal air flow field can be established with combinations of withdrawal and
injection (if required) wells and/or trenches. The advective air flow induces intraphase transfer of the
contaminants from the immiscible and water phases into the air phase. Air laden with contaminant vapors
moves along induced flow paths toward the withdrawing system where it is analyzed, treated, and/or released
to the atmosphere. The success of the method depends on the rate of contaminant transfer from the
immiscible and water phases into the air phase, and in particular, the ability to establish an air flow field that
intersects the distributed contaminants.
The cost effectiveness of utilizing vapor extraction has been diminished by less than optimal employment
of the technology. Often, vapor extraction systems have been designed and implemented based on less than
a full understanding of the physical/chemical principles governing the process. The design of a vapor
extraction system is comparable to the design of a ground water resource or ground water extraction and
treatment system. Industry has learned that mathematical models are essential tools in the design of ground
water extraction and treatment systems. This is also the case in the design of vapor extraction systems.
Mathematical equations/models are utilized to approximate the physical conditions that exist in the
subsurface environment. They provide a scientific basis - not a black box - upon which design and manage-
ment decisions can be made.
A SCIENTIFIC DESIGN APPROACH
A general approach to vapor extraction system design follows the pattern where the information obtained
from the evaluation of the degree and extent of contamination is utilized to provide assumptions of uniform
contamination over a specified soil zone. Intraphase equilibrium is assumed between the contaminants and
the air and water phases. Uniform air flow fields are assumed and extrapolations of the remediation process
are performed. However, following immiscible fluid flow in porous media, the retained, immobilized,
immiscible fluids may exist as a few large globs of liquid, or a large number of smaller globs. The geometry
of the fluid distribution depends on the nature of the capillary forces between the fluids, the pore sizes and
geometry, and the history of fluid movement in the medium. Although intraphase equilibrium may exist at
the pore scale, the heterogeneous distribution of the immobilized immiscible organics within the pores may
126
-------
make the overall equilibrium assumptions inappropriate. Marley (1985) and Baehr, Hoag and Marley (1989)
demonstrated that where a relatively uniform distribution of residual contamination does exist, the
assumption of a dynamic equilibrium between the advective air phase and the immiscible contaminant
provides a good approximation of the governing physical/chemical processes. More generally, the
intraphase transfer of contaminants should be considered in terms of mass transfer limitations. At this time,
few transport models which consider potential mass transfer limitations exist.
Of greater importance than the potential mass transfer limitations or equilibrium assumptions is the
capability to control the airflow pathways in the vadose zone to optimize contact with the contaminants.
Without the aid of air flow models, it would be difficult to evaluate the airflow pathways for all but the most
simplistic of cases (homogeneous, isotropic, dry sands, impermeable surface boundary, with no subsurface
structures within the extraction system zone of vacuum influence).
AIR PERMEABILITY AND AIR FLOW MODELING
Compressible flow in porous media has been a subject of investigation for many years in petroleum reservoir
engineering. Mathematical models of air movement in unsaturated porous media have been calibrated with
air pressure data in previous investigations to provide determinations of in-situ air permeability. Muskat and
Botset (1931) developed a one-dimensional (radial) air flow model to evaluate the horizontal permeability
of gas reservoirs. Boardman and Skrove (1966) injected air into packed-off sections of drill holes and
observed radial pressure distributions to obtain horizontal fracture permeability of a granitic rock mass.
Stallman and Weeks (1969) and Weeks (1977) describe the use of depth dependent air pressure to calculate
vertical air permeability in the unsaturated zone. Rozsa and others (1975) document an application of this
technique to determine vertical air permeabilities of nuclear chimneys at the Nevada Test Site. As another
historical note, soil scientists (6,9,19) have utilized injected air and pressure measurements to evaluate soil
permeability but these techniques provide estimates over small regions of soil and are not directly applicable
for unsaturated zone evaluation.
The application of such models to aid in the design of a vapor extraction system is exactly analogous and
can be thought of in two steps:
1. Evaluate, in-situ, the air permeability tensor of the contaminated unsaturated zone soils by
calibrating a steady-state airflow model with pressure measurements obtained during short-
term air pump tests; and,
2. utilize the air permeability values and air-flow models to determine the well spacings,
screened intervals of wells in the unsaturated zone, and the size and type of pumps required
to generate the desired air movement.
An in-situ determination of the air-phase permeability tensor is preferred over laboratory determinations to
account for variations in prevailing soil-water conditions, the presence of the immiscible organic liquid, and
anisotropy and heterogeneity in air phase permeability. Further, permeability evaluations are sensitive to the
soils bulk density and structure, these properties are generally altered in soil samples taken for laboratory
analysis.
127
-------
The governing equation defining conservation of mass for compressible flow is given as:
= 0 (1)
where, p = air density
0 = air filled porosity
q = specific discharge vector
expressing density as a function of pressure in accordance with the ideal gas law and q by Darcy's law:
_ PaWa
Pa - (2)
RT ( '
ka
Pa= —VPa (3)
where, P = air pressure
Wa = molecular weight of air
R = universal gas constant
T = temperature
k = intrinsic permeability tensor
}i = air viscosity
yields a partial differential equation in terms of air phase pressure. The selection of a coordinate system and
appropriate boundary conditions, together with equation (1) defines the air-flow model. The steady state air
flow model is given by the following partial differential equation:
V'(kaVPa2) =
(4)
Commonly, hydraulic conductivity values are available from ground water studies performed prior to the
vapor extraction feasibility study, where applicable (e.g., uniform, dry, medium-coarse sands), these values
may provide a relatively accurate evaluation of the intrinsic horizontal air permeability that could be used
in the design process. However, in the majority of cases, this assumption is invalid for one or more of the
following reasons:
• gas slippage is ignored (Klinkenberg (1961)
• anisotropy is neglected,
• swelling soils are present,
• the variable water saturation in the unsaturated zone is ignored,
128
-------
• the presence of an oil phase is ignored,
• the groundwater test may be in a different strata,
• the scale of the ground water test may invalidate the parameter evaluations.
To determine air phase permeability, a radially symmetric region with a single well (Figure 1) is assumed.
This implies that the principal axes for the air-permeability tensor are in the radial (r) and vertical (z)
directions. Air is injected or withdrawn through the well screen. The unscreened portion of the well is a no
flow boundary. Referring to Figure 1 the boundary conditions at r = r (well diameter) are:
P = Ps for r = r0 (5)
and
= Oforr = r0 (6)
dr
where Ps is the steady state pressure along the well screen, located between elevations zwl and zw2. The
boundary at r = rf (far radius of influence) can be chosen to be unaffected by the well:
P = P forr = rf (7)
atm i
where P is atmospheric pressure. Pressure measurements (if available) can be used to define the boundary
condition at r=rr The lower boundary, formed by the water table, is impervious to air flow, and is as follows:
dP
= 0 forz = z and all r (8)
dz
wat
The land surface may be open to the atmosphere or impervious to air flow to simulate paved surfaces. This
boundary is:
dP
-5— = 0 for z = z ,. and r < r.
sjr< surf imp
(9)
where r^ is the radius of the impervious portion of the land surface.
^
Solutions to eq.(2) subject to boundary conditions (5) - (9) were developed. For the special case of radial
flow that is induced by fully screening the well in the unsaturated zone beneath an entirely impervious ground
129
-------
surface, the analytical solution to equation (1) presented by Muskat and Botset (1931) can be used to obtain
a horizontal air-phase permeability, averaged over the entire depth of the unsaturated zone. The solutions
derived for 2, however, can simulate flow to a partially screened well, and thus, allow for a determination
of vertical and horizontal air phase permeabilities, averaged over portions of the unsaturated zone. Further,
the solutions allow for evaluating heterogeneous unsaturated zones.
During the conducted field air pump test, the system is operated at a minimum of two air flow rates; this
allows for both the initial calibration of the model (i.e., parameter evaluations using the collected field data
at the primary air flow rate) and verification of the model (i.e., the model is set to simulate the system for
the secondary air flow rates using the parameters established in the calibration mode and comparison is made
between the predicted air flow rates and pressure distributions at the well/probes by the model, and the actual
pressure data measured at the well/probes at the secondary air flow rates), Figures 2 and 3.
After determining air-phase permeability, multi-dimensional air flow models based on eq. (3) are used to
determine the well/trench spacings, screen intervals, air flow rates, (injection and/or withdrawal) and system
equipment requirements.
MODEL APPLICATIONS
The following sections present applications of numerical and analytical solutions to the two-dimensional
(radially symmetric) and a numerical solution to the three-dimensional (cartesian coordinate) forms of
equation (3).
Figure 4 presents a plan for a gasoline spill site in Massachusetts. The three-dimensional air flow model was
used to predict the pressure distribution at several depths in the subsurface soils using the field permeabilities
as calculated. Figures 5 and 6 display a plan view and 3-Dimensional perspective on the observed pressure
distribution in the soil system and the pressure distribution predicted by the calibrated model respectively.
The close correlation observed between the model predictions and the observed field data validates the air
flow modelling approach.
Figure 7 and 8 present an application of the three dimensional (cartesian coordinate) air flow model in the
evaluation of a proposed full scale soil vapor extraction system design. The design consists the simultaneous
operation of two vapor extraction wells and a vapor extraction trench. Figure 7 shows the vacuum isopleths
(in atmospheres) across the site under full scale operation. Figure 8 presents a perspective view of the zone
of vacuum influence associated with the system design. Further model simulations are performed to attain
a site-specific optimal vapor extraction system design.
SUMMARY
To date, the majority of vapor extraction system designs have been based on long term pilot studies of non-
engineered, less economical installations based largely on the principle of trial and error. Engineered system
design requires the evaluation of the site's soil properties in relation to air flow. The relevant soil properties
include the relative intrinsic permeabilities of each geologic unit and of surface and sub-surface confining
unit(s). Laboratory evaluation of these parameters is difficult, time consuming, costly and subject to the
errors associated with extrapolation of data from discrete soil samples.
130
-------
The relative intrinsic permeability is the fundamental design parameter required to predict the air-flow field.
An in-situ determination of this parameter can be obtained by calibration of an air-flow model with air flow
rate and pressure distribution data collected during the conduct of a short term pilot test.
Analytical and numerical air flow models have been developed to evaluate the relevant soil properties.
Following calibration and verification from field air pump test data, the models can be utilized to size vacuum
pumps and to define flow rates, subsurface air pressures, and the flow fields associated with alternate vapor
extraction system configurations. The simulations performed allow for evaluation of the most effective and
economical system design.
BIBLIOGRAPHY
1. Baehr, A.L., Hoag, G.E., and Marley, M.C., 1989 "Removal of Volatile Contaminants from the
Unsaturated Zone by Inducing Advective Air Phase Transport": Journal of Contaminant Hydrology,
Vol. 4, Feb., Pages 1-26.
2. Baehr, A.L., 1987. "Selective Transport of Hydrocarbons in the Unsaturated Zone Due to Aqueous
and Vapor Phase Partitioning": Water Resources Research, Vol. 23, No. 101, Page 1926-1938.
3. Baehr, A.L., and Hull, M.F., 1988, "Determination of the Air-Phase Permeability Tensor of an
Unsaturated Zone at the Bemidji, Minnesota Research Site": Presented at the 4th Toxic Substances
Hydrology Technical Meeting.
4. Boardman, C.R., and Skrove, J.W., 1966, "Distribution in Fracture Permeability of a Granitic Rock
Mass Following a Contained Nuclear Explosion": Journal of Petroleum Technology, Vol. 181, No.
5, Pages 619-623.
5. Bruell, C.J., and Hoag, G.E., 1984 Capillary and packed column gas chromatography of gasoline
hydrocarbons and EDB. Proc. National Water Well Association/America Petroleum Institute
Conference on Petroleum Hydrocarbons and Organic Chemicals in Groundwater, Nov. 87, Houston,
TX, Pages 234-266.
6. Grover.B.L., 1955"SimplifiedAirPermeabilitiesforSoilinPlace": Soil Science Society of America
Proceedings, Volume 19, Pages 414-418.
7. Hoag, G.E., and Marley, M.C. 1986 "Gasoline Residual Saturation in Unsaturated Uniform Aquifer
Materials." ASCE, Environmental Eng. Division, Vol. 112, No. 3, Pages 586-604.
8. Hubbert, M.K., 1940. The theory of ground water motion. J. Geol. 48(8).
9. Kirkham, D., 1946, "Field Methods for Determination of Air Permeability of Soils in its Undisturbed
State": Soil Science Society of America Proceedings, Volume 11, Page 93-99.
10. Klinkenberg, L.J., 1941, The Permeability of Porous Media to Liquids and Gases: American
Petroleum Institute Drilling and Production Practice.
131
-------
11. MacKay, D. and Shiu, W.Y., 1981. A critical review of Henry's Law Constants for chemicals of
environmental interest. J. Phys. Chem. Ref. Data, 10(4): 1175-1199.
12. Marley, M.C., 1985, Quantitative and Qualitative Analysis of Gasoline Fractions Stripped by Air
from the Unsaturated Zone"; M.S. Thesis, University of Connecticut, Department of Civil Engineering,
Page 87.
13. Marley, M.C., and Hoag, G.E., 1984 "Induced Soil Venting for Recovery/restoration of Gasoline
Hydrocarbons in the Vadose Zone"; Proceedings of the National Water Well Association - American
Petroleum Institute Conference on Petroleum Hydrocarbons and Organic Chemicals in Groundwater,
Nov. 5-7, Houston, TX, Pages 473-503.
14. Muskat,M.,andBotset, H.G., 193 l,I"Flow of Gas through Porous Materials": Physics, Vol. 1,Pages
27-47.
15. Porter, J. Winston, 1989. "Superfund Progress": Hazardous Materials Control, Volume 2, No. 1,1
Page 48.
16. Rosza, R.B., Snoeberger, D.F., and Bauer, J., 1975, Permeability of a Nuclear Chimney and Surface
Alluvium: Lawrence Livermore Lab Report UCID-16722, 11 p.
17. Sleep, B.E., and Sykes, J.F., 1989, "Modelling the Transport of Volatile Organics in Variably
Saturated Media": Water Resources Research, Vol. 25, No. 1, Page 81-92.
18. Stallman, R.W., and Weeks, E.P., 1969, "The Use of Atmospherically Induced Gas Pressure
Fluctuations for Computing Hydraulic Conductivity of the Unsaturated Zone": Geological Society
of America Abstracts with Programs, Pt. 7, Pages 213.
19. Tanner, C.B., and Wengel R.W., 19578, "An Air Permeameter for Field and Laboratory Use": Soil
Science Society of America Proceedings, Volume 21, Pages 663-664.
20. Weeks, E.P., 1977, Field Determination of Vertical Permeability to Air in the Unsaturated Zone: U.S.
Geological Survey Open-file report 77-346 92 p.
132
-------
Typical Subsurface Test Configuration
rf
Bentonite Seal
Ottawa Sand
- Cement Grout
- Auger Cuttings
FIGURE 1
-------
VW1 @ 10.1 CFM - 3D Model Calibration
Relative Pressure (ATM)
1.0000
0.9990
0.9980
0.9970 -
VP3 VP5 VP4
— Simulated Data
D Field Data
20 40
Radial Distance from Well (feet)
FIGURE 2
-------
VW1 @ 23 CFM - Model Verification
Relative Pressure (ATM)
1.0000
0.9980
0.9960
0.9940
0.9920
VP3 VP5 VP4
Simulated Data
Field Data
0
Radial Distance from Well (feet)
FIGURE 3
-------
Site Plan
Equipmepf'Staging Area
A VPS
ON
Manifold Trench
D Vapor Extraction Well
A Vapor Probe
FIGURE 4
-------
FULL SCALE SVES: 2 WELL SYSTEM; YARMOUTH, MA.
100
80 -
60 -
S!
4O -
20
0 «—•
0
J_
20
40
60
FEET
80
_J
100
FIGURE 5: VACUUM ISOPLETHS
-------
FULL SCALE SVES: 2 WELLS; YARMOUTH, MA.
oo
6O GO 1 OO
FIGURE 6: VACUUM PERSPECTIVE
-------
FULL SCALE SVES: MAIN St. ORANGE. CT.
VO
BOO
600
4OO
200
£00
400 000
FEET
BOO
FIGURE?: VACUUM ISOPLETHS
-------
FULL SCALE SVES: MAIN ST. ORANGE, CT,
FIGURE 8: VACUUM PERSPECTIVE
-------
MODELING IN-SITU BIODEGRADATION IN UNSATURATED AND
SATURATED SOILS
HAROLD W. BENTLEY
HydroGeoChem, Inc.
BYRAN TRAVIS
Los Alamos National Laboratory
ABSTRACT
In-situ biodegradation of subsurface organic contamination requires delivering the chemical constituents
necessary for the biodegradation reactions to the contaminated zone and the removal of toxic or inhibitory
components produced either by the biodegradation process or present because of the contamination event.
Modeling these chemical and physical processes requires numerical simulation of multi-component
transport and reaction in the gas, liquid, and solid phases.
TRACR3D is a finite difference code capable of simulating gas and liquid flow, and multi-component solute
transport under saturated and unsaturated conditions in three dimensions. We have now modified
TRACR3D (and renamed it) TRAMP to include microbial reaction kinetics. The microbial reactions are
based on Michaelis-Menten kinetic equations whereby the rates of organic substrate decay vary from zero
order when nutrients, electron acceptors, and substrate are in abundance to first order when the concentration
of a necessary component becomes rate limiting. These equations also include inhibition terms that limit
the rate when the concentration of a toxic or inhibitive component becomes excessive. The equations are
balanced according to stoichiometry and reaction yield. These coupled equations are solved with a iterative
numerical algorithm. Both aerobic and anaerobic biodegradation can be simulated. Lag time, the poorly
understood acclimation time before the biodegradation reactions are initiated, is not accounted for in the
model.
TRAMP is capable of simulating such processes as soil venting, the effect of introducing cometabolites into
the contaminated zone, co-oxidation processes such as methanotrophic oxidation of chlorinated ethenes, the
enhancement of biodegradation by nutrient addition, and the effects of toxin removal on in-situ bioremediation.
Practical examples presented include a one-dimensional simulation of the biodegradation of benzene
introduced into an aerobic ground water system, simulations of the effect of changes in water saturation on
rates of SVE/biodegradation, and a two-dimensional simulation of SVE/biodegradation of soils containing
volatile and non-volatile organics leaking from a contaminated perched mound.
141
-------
CHARACTERIZING PERMEABILITY TO GAS IN THE VADOSE ZONE
MICHAEL SULLY
Department of Hydrology and Water Resources
University of Arizona
ABSTRACT
The permeability of the vadose zone is a key parameter in predicting the rate of vapor transport in partially
saturated media. It is highly relevant to the successful design of a soil venting system to recover volatile
hydrocarbon compounds. The permeability of soil to gas is dependent upon the portion of the soil void space
which is occupied by liquid. In this paper we review the theory of permeability measurement, the state of
the art in field and laboratory testing, and gaps in current technology. We also present indirect methods to
determine permeability such as through particle size analysis, as well as methods to determine the moisture
content which is associated with the permeability measurement. A case study will be presented in which two
different field methods to characterize air permeability were composed.
143
-------
USE OF SOIL GAS MEASUREMENTS IN THE DESIGN OF SOIL VAPOR
EXTRACTION SYSTEMS
GARYR. WALTER AND HAROLD W. BENTLEY
HydroGeoChem, Inc.
ABSTRACT
The design of a soil vapor extraction (S VE) system requires a knowledge of the mass and spatial distribution
of the contaminants, their concentrations in the soil vapor phase, the rates at which air can be circulated
through the soil, and the air flow patterns which will result from a particular extraction system design. The
measurements required to design an SVE system are exactly analogous to those needed to design a ground
water pump and treat system. Given the premise that sampling for remedial design should target the mobile
phase of the media being removed, soil gas sampling and analysis represents the appropriate technology for
SVE design investigations.
Because soil cleanup standards are defined in terms of total soil concentrations, the standard approach to soil
remedial design has been to collect and analyze soil samples directly. Direct soil analyses are generally
regarded as being more accurate than soil gas analyses. In principle, however, soil gas analyses are no less
accurate than other methods for measuring contaminant concentrations in environmental media. The results
of soil analyses are well known to be subject to high variability and to suffer from contaminant losses during
storage and processing. The variability of soil analyses also increases the cost of investigations because large
numbers of samples are required to define contaminant distributions.
Soil gas samples represent a superior method for quantitatively defining contaminant distribution in the
vadose zone because they are repeatable, represent relatively large volumes of soil, and have minimal
processing losses. Soil gas samples can also be analyzed relatively easily and accurately in a close support
laboratory environment to provide rapid guidance to the sampling team and to greatly reduce analytical costs.
Given measurements of total organic carbon, moisture content and porosity, soil gas concentrations can be
converted into total soil concentrations.
An additional benefit of soil gas sampling is that the same tools used to collect the soil gas samples can be
used to perform in situ measurements of soil air permeability. Such measurements are essential for
successfully designing the physical air circulation equipment for the site.
The practical application of these approaches is presented based on the remedial design investigation for an
SVE system at the Seymour Superfund site.
145
-------
ASSESSING THE PERFORMANCE OF INSITU SOIL VENTING
SYSTEMS BY SOIL SAMPLING AND VOLATILE ORGANIC
COMPOUND MEASUREMENTS
ROBERT L. SIEGRIST
Oak Ridge National Laboratory
Oak Ridge, Tennessee
ABSTRACT
Soil contamination by volatile organic compounds (VOCs) is often remediated by in situ soil venting
systems. Assessing the performance of these systems requires measurements of the soil VOC concentrations
present. VOC measurements in soil systems are subject to many sources of error. Random errors can
normally be effectively managed through statistical techniques, but systematic error or bias is far more
elusive. While positive bias (e.g. cross-contamination of samples) can be managed through quality
assurance provisions (e.g. trip and field blanks), negative bias is more difficult to delineate and control.
Negative bias in VOC measurements (i.e. measured value
-------
Research is continuing at Oak Ridge National Laboratory and at other laboratories in the USA. This
laboratory and field research is directed at evaluating different sampling devices (e.g. Shelby tube; standard
split-spoon), sample preservation conditions (e.g. preservation temperature; holding time), and analytical
techniques (e.g. field vs. lab).
Based on this research, it is clear that soil VOC measurements must account for the special properties and
behavior of these compounds. This paper presents a discussion of the research conducted by this author as
well as a synopsis of selected other work ongoing in the USA. The application and implications of the
findings to performance assessments of in situ soil venting systems will be discussed.
148
-------
A FIELD TECHNIQUE FOR DETERMINING UNSATURATED ZONE
AIR PERMEABILITY
CRAIGJ.JOSS
Drexel University
Dept. of Civil Engineering
32nd & Chestnut Street
Philadelphia PA 19104
ARTHUR L. BAEHR
JEFFERYM. FISCHER
US Geological Survey
810 Bear Tavern Road
West Trenton, NJ 08628
ABSTRACT
A field technique for determining air-phase permeability in the unsaturated zone has been developed to
provide a method for obtaining site specific data for use in models that simulate air flow induced by vapor
extraction systems. The apparatus used includes a pneumatic pump, a length of stainless steel tubing, a flow
meter, a water manometer, a thermometer, and a barometer. One end of the tubing is modified to form a
pressure probe by drilling holes to form a screen. The screened end of the probe is set approximately mid-
depth in a lithologic unit. The test consists of withdrawing (or injecting) air from (or into) the unit through
the probe at various flow rates and measuring the corresponding air pressure at the probe. This information
is used to calibrate a mathematical model of radially symmetric steady state air flow. This model is coupled
with a model of air flow through a tube to account for pressure losses not due to the porous media.
Applications of the technique at a research site in Galloway Township, New Jersey, are presented.
1.0 INTRODUCTION
Recent research into the behavior of volatile organic chemicals in the unsaturated zone has resulted in the
development of a number of models which include advective transport in the vapor phase. For example,
models describing the movement of organic vapors in the unsaturated zone have been presented by Sleep
and Sykes, (1989), Mendoza and Frind, (1989), Falta et al.,(1989), Thorstenson and Pollock, (1989), Baehr
and Bruell, (1990). Models that simulate air flow in the unsaturated zone induced by vapor extraction
systems are also emerging, for example Baehr et al. (1989). With these increasingly sophisticated models,
the need to obtain values for the parameters of these models has been defined.
149
-------
A key parameter in the vapor phase transport equations, is the air-phase permeability. The air-phase
permeability varies with the degree of moisture saturation of the porous medium. Thus, in order to obtain
values of permeability that reflect prevailing field conditions, some form of in-situ measurement is required.
Also because changes in unsaturated zone lithology can occur over small intervals, it is important to develop
measurement techniques over relevant spatial scales. Variation in air permeability as a function of
unsaturated zone lithology is an important consideration in the design of induced venting systems.
The test procedure involves injecting or withdrawing air through a probe located in the unsaturated zone. The
induced air flow is resisted by the porous media surrounding the probe. Pressure response in the domain at
different flow rates is measured using water manometers or pressure transducers attached to observation
probes surrounding the extraction probe. The test may also be performed using a single probe which acts
as both an extraction probe and an observation probe. In such instances, the pressure at the probe is
determined by correcting the surface pressure measurements for pressure losses due to flow in the tube. Air
flow rates in the tube connecting the probe to the land surface are measured using a rotometer. Atmospheric
pressure, temperature and sub-surface temperatures are measured for standardization purposes. Knowing
the pressure at the probe(s) and the mass flow rate at steady state, an estimate of the air permeability is
obtained by calibrating the model of radially symmetric air flow.
The number of observation probes used in the test defines the information that is generated on the air phase
permeability. If a single observation probe is used to record pressure response in the domain under different
flow rates, calibration of the model will only provide estimates of an averaged air permeability (k) of the
particular lithologic unit in which the probe is located. If however, multiple probes are used to measure
pressure response, calibration of the model will provide estimates of the radial (k^ and vertical permeabilities
(kz) of the domain. The procedure employing results from a single vapor probe is referred to as a small-scale
or mini-permeability test while the procedure using measurements from multiple probes is called a full-scale
permeability test.
2.0 FIELD METHODS
2.1 Equipment
Figure 1 gives a general schematic of the test equipment used at the Galloway research site. Basic
components of the field equipment necessary to perform the permeability evaluations in the unsaturated zone
include:
• length(s) of stainless steel tubing;
• a water manometer;
• a condensation chamber;
• a flow meter;
• a pneumatic pump (and power supply);
• a thermometer,
• a barometer.
The probes used in permeability tests at Galloway Township, New Jersey, are stainless steel tubing that
extends down from the land surface to the desired depth. The final 15 centimeter (cm) section of the tube
150
-------
is perforated with a number of small holes wrapped in 100 mesh stainless steel screen to form the probe. The
probes are set in hand augured bore holes approximately 7.6 cm in diameter. The annulus between a probe
and bore hole wall are filled with coarse 10-20 mesh silica sand, A typical sand pack measures approximately
45 cm in length. Given the high permeability of the sand pack relative to the in situ material, the pressure
drop between the probe and the bore hole wall is assumed to be small compared with that attributed to the
surrounding porous media. This assumption requires that the dimensions of the sand pack be used in place
of the dimensions of the actual probe in the analytical model. For example, input for the well radius should
reflect the radius of the sand pack (equivalent to the bore-hole radius). Similarly, the screen length should
be approximated by the length of the sand pack. To prevent vertical air flow through the bore hole, a bentonite
layer was installed at each end to seal the sand pack.
The internal tube diameter used in the tests performed at Galloway was 0.396 cm. Small diameter tubing
was used to minimize dead space for soil gas determinations, therefore the diameter chosen is not necessarily
recommended if the probe is to be used solely to determine air permeability. Tubes were labeled at the surface
so that individual probes within a nest of probes could be identified.
A water manometer with a capacity of approximately 50 inches of water was connected to the system. A
water trapping device was installed in line before the rotometer to capture any water entering the system.
Calibrated rotometers were used to measure air flows rates in the range of 0 through 80 liters per minute. A
pneumatic pump was used to induce air flow through the vapor probe.
2.2 Probe Installation
A hand auger, 7.6 cm in diameter, was used to drill the holes for the vapor probes. This hand procedure was
selected over power techniques to enable a more detailed log of unsaturated zone stratigraphy. Based on the
field logs, distinct geologic strata were identified. Probe locations were then selected to coincide with the
mid-depth of each separate lithologic strata. Since one boring typically intersected several distinct strata,
the vapor probes were nested in the bore holes. Probe construction is described in Section 2.1. This
installation procedure allowed specific strata to be isolated for air-permeability determination.
2.3 Test Procedure
Flexible tygon coated tubing is used to connect the steel tubing at the surface to the pump and pressure and
flow measuring devices. Prior to commencing the test a pressure and temperature measurement is made to
record the prevailing atmospheric conditions. This information is needed to correct flow readings for non-
standard flow conditions and to define the model boundary conditions at land surface. The pneumatic pump
is then turned on and adjusted to give the desired air flow rate through the probe. Air flow rates should be
selected to allow several data sets to be collected, each set corresponding to a different flow rate. The system
should be allowed to reach steady state prior to taking any measurements.
2.4 Test Measurements
Three general types of measurement must be taken during the test. Information is required on the test
parameters, equipment and on the domain geometry. A summary of the measurements required to implement
the analytical model is presented below.
151
-------
a. Test Measurements
• air flow rate at steady state (e.g., 10.0 liters/minute);
• air pressure at the surface if a single vapor probe is being used (e.g., 5.0 inches of water), or,
air pressure in the observation probes surrounding the extraction probe if a network of probes
is being used;
• prevailing atmospheric pressure (e.g., 761 mmHg);
• prevailing atmospheric temperature (e.g., 15 °C).
b. Equipment Measurements
• radius of bore hole (e.g., 3.8 cm);
• internal diameter of steel tube (e.g., 0.396 cm);
c. Domain Geometry
Definition of the domain geometry is based on an interpretation of the boring log. One of two analytical
models may then be used to analyze the permeability test data. The first model applies to a domain separated
from the atmosphere by a confining unit. The second model applies to a domain open to the atmosphere.
Depending on the type of model most suited to the geology at the probe location, various input information
must be collected. The following data apply :
• presence or absence of a confining unit;
* thickness of confining unit;
• thickness of domain, namely distance between ground surface (or bottom of confining unit) and
water table;
• depth to top of sand pack (screen) from ground surface (or bottom face of confining unit);
• depth to bottom of sand pack (screen) from ground surface (or bottom face of confining unit).
The confining unit permeability should be less than the permeability of the domain for the model solution
to apply because the model emulates Hantush's leaky aquifer theory whereby leakage through the confining
unit is assumed to be distributed across the domain. For domains with permeabilities that are similar to those
in adjacent lithologic units, the unconfined model should be selected.
3.0 AIR FLOW MODELS
Baehr and Hull (1991), present two solutions to the air flow equation assuming a radially symmetric domain.
A summary will be presented in this paper for reference purposes.
3.1 Steady State Air Flow Model
The steady state air flow equation describing radially symmetric air flow is as follows :
152
-------
where, r and z are cylindrical coordinates aligned along the major axes of the
permeability kr and kz;
0 = air pressure squared [(g/cm-sec2)2];
t = time [sec].
3.2 Analytical Solution for Domain Separated from Atmosphere by a Leaky Confining Unit
The domain simulated is sketched in Figure 2. The leaky confining unit can be, for example, a strata less
permeable to air than the domain (e.g., a silty-clay unit) or a slightly permeable paved surface. The bottom
boundary is formed by the water table which is impervious to air. This solution emulates Hantush's leaky
aquifer solution by assuming that leakage from the confining unit is distributed across the domain. The
analytical solution from Baehr and Hult (1991), is :
- P2
-
atm
Ko(M0T)
si"
snnl\ /n;rd
si"
aw
where, = air pressure squared in domain [(g/cm-sec2)2]
Patm = atmospheric pressure [g/cm-sec2]
a = square root of anisotropy ratio (k/kz)1/2 [ - ]
Q* = QURT
CO
Q = mass flow rate [g/sec]
^i = dynamic viscosity of air [g/cm-sec]
R = universal gas constant [g-cm2/sec2-mol-K]
T = absolute temperature [K]
co = average molecular weight of air phase [g/mol]
rw = radius of well (to filter/soil interface) [cm]
K0 = zero order modified Bessel function of
the second kind
Kj = first order modified Bessel function of
the second kind
153
-------
b =
d =
1
M =
n
vertical thickness of domain
distance from lower confining unit to
top of well screen
distance from lower confining unit to
bottom of well screen
/mc\ 2 + k'
\b ) bb'k,
1/2
[cm]
[cm]
[cm]
[ I/cm ]
k' = permeability of confining unit
b' = thickness of confining unit
[cm2]
[cm]
3.3 Analytical Solution for Domain with Land Surface as Upper Boundary
The domain simulated is sketched in Figure 3. In this case, the top of the unsaturated zone is assumed to be
in direct connection with land surface. The bottom boundary is formed by the water table. The analytical
solution from Baehr and Hult (1991), is :
P2
atm
(3)
2aQ
cos(JM-d) _cos
MmK^IVV-f)
where, m = n -1
2
and Mm =
In tests using a single vapor probe, the probe pressure is obtained by correcting surface pressure
measurements for pressure losses due to flow in the tube.
3.4 Equation for Pressure Loss Due to Flow in a Tube
In mini-permeability tests, a single probe is used to stress the domain and to determine the pressure response
in the vadose zone under the imposed stress. Air is injected into, or withdrawn from, the vapor probe via a
length of tubing. Pressure and air flow rates in the tubing are measured at the ground surface during the test.
The surface pressure measurement is then used to estimate the pressure in the domain at the probe screen-
154
-------
soil interface. The probe pressure is the value used to calibrate the air flow model and thereby estimate the
air phase permeability.
A pressure differential exists between the surface measurement and that prevailing at the probe due to
pressure losses resulting from air flow through the tube. The magnitude of the pressure differential is
quantified in equation (4). Equation (4) presents the pressure at the probe as the difference between the
measured surface pressure and the pressure losses in the tube due to air flow. Equation (4) is an expression
derived from elementary fluid mechanics. Application of equation (4) in mini-permeability tests introduces
an additional parameter that is not required in full scale tests, namely the friction factor (f) corresponding
to air flow in the tube.
P
2 +
1
\D}
^ (
p
f
)-
X
VRT
(vi
CO
•Pi)
\
^ /
(4)
where, + = positive for air withdrawal
= negative for air injected
0 = square of probe pressure
Pj = system pressure measured at the surface
= P +-P
atm man
Patm = atmospheric pressure
Pmsm = manometer pressure differential between
system and atmospheric pressure
f = friction factor along pipe
D = internal diameter of the pipe
B = constant (assumed to be unity)
0 = mean pressure squared estimate for tube
co = average molecular weight of air phase
R = universal gas constant
T = temperature of air in tube
Pj = density of the air in tube at Pt
Vj = velocity of air in tube at Pt
x = length of tube section between probe and
surface measurement
[(g/cm-sec2)2]
[ g/cm-sec2]
[ g/cm-sec2]
[ g/cm-sec2]
[-]
[cm]
[-]
[ g/cm-sec2]2
[ g/mol ]
[g-cm2/sec2-mol-K]
[K]
[ g/cm3 ]
[ cm/sec ]
[cm]
The estimate 0 is obtained by evaluating equation (4) at x/2 and assuming all pressure loss is due to friction
(ie. 6=0 for determining 0 only). The friction factor is obtained from experimental data (see below) or from
theoretical considerations.
155
-------
3.5 Experimental Determination of Friction Factor
The friction factor (f) as a function of Reynolds Number is determined experimentally by conducting flow
experiments on a length of tube with the probe end at atmospheric pressure (PItm), and calibrating equation
(5). Equation (5) is obtained by rearranging equation (4) and replacing Pt with Patm.
f = .
x
JL m i "
f RT (vi-p!)2
(5)
is the pressure in the pipe (squared), after the air has passed through a length x of piping. The Reynolds
Number for pipe flow is presented in equation (6).
r> VlPlD
Re = 1K1 (6)
4.0 APPLICATION AT GALLOWAY TOWNSHIP, NEW JERSEY
An investigation of induced venting is being conducted at a site of a gasoline spill from a leaking underground
storage tank at Galloway Township, New Jersey. An air flow model is being used to examine the effects of
unsaturated zone heterogeneity on induced flow patterns. The field technique described in this paper is being
used to estimate the air permeability of distinct lithologic units to provide input for the site air flow model.
Figure 4 provides a description of unsaturated zone lithology at the location of this illustrative pneumatic
test. Three probes are located within the sandy unit between the perched water table and the thin clay lens
located at 183 cm below land surface as illustrated in Figure 5. Another probe is located in the brown clayey
sand unit above the thin clay lens.
One set of tests was conducted on 6/4/1991 by injecting or withdrawing air at probe VW9-7.4 and pressure
was measured at probes VW9-6.0 and VW9-8.2. Six tests were conducted in this set, three using air
withdrawal and three using air injection. Data collected are summarized in Table 1.
The tests were first analyzed assuming the thin clay lens was pneumatically insignificant and that the brown
clayey sand unit above the probes had the same air-permeability as the sand unit in which the probes were
located. Under this hypothesis the domain illustrated by Figure 3 applies. Table 2 provides a summary of
domain geometry and the best fit air permeability under the conceptualization of the unsaturated zone
described above. The best fit air permeabilities indicate an unlikely physical situation, that vertical
permeability is roughly an order of magnitude higher than the horizontal permeability. Therefore this
conceptualization of the unsaturated zone permeability distribution was abandoned.
156
-------
A more feasible analysis of the pneumatic tests is obtained by assuming the thin clay lens is significantly less
permeable than the underlying sand. Under this hypothesis the domain illustrated in Figure 2 applies. Table
3 provides a summary of domain geometry and the best fit air permeabilities under the conceptualization of
the unsaturated zone described above. The horizontal and vertical air permeability estimates are similar in
magnitude. The clay lens therefore is a significant air flow barrier which should be considered in designing
unsaturated zone remediation. Further, air permeabilities k. and kz listed in Table 3 are the best estimates
representing the sand unit for input in a sitewide air flow model.
To estimate the permeability of the brown clayey sand unit another set of tests were conducted by
withdrawing air from probe VW9-3.0. These tests were conducted at an earlier date, 1/21/91. Probe VW9-
3.0 was the only probe available in this unit in the vicinity, therefore the adjustment for pressure loss in the
tube, given by equation (4), was particularly important Further, because only one observation is available,
the brown clayey sand unit was assumed to be isotropic (kr = kz). The top of the clay lens located at z = 155.4
cm was assumed to be an impervious unit. The medium white sand unit between the brown clayey sand and
the clay lens (Figure 4) was assumed to have the same air permeability as the brown clayey sand. The domain
illustrated by Figure 3 applies. Data collected for this set of tests are summarized in Table 4. Table 5 provides
a summary of domain geometry and best fit permeability estimates. Comparing these estimates to those
obtained for the sand unit (Table 3) indicates that the brown clayey sand unit is about three times less
permeable than the sand unit beneath it.
5.0 SUMMARY
A field technique has been developed to obtain air permeability estimates of lithologic units in the
unsaturated zone. Two analytical solutions are presented for the equation describing axisymmetric air flow
to a partially penetrating well screened in the unsaturated zone. One solution applies to a domain directly
in contact with land surface and the other applies to a domain separated from land surface by a confining unit
of lower permeability. Analysis of one-dimensional air flow in a cylindrical tube allows for use of the
injection/withdrawal well as an observation point as pressure losses in the tube should not be attributed to
porous media.
The technique was applied at a field site in Galloway Township, New Jersey. Analysis showed that a thin
clay lens was pneumatically significant and would affect site wide flows induced for vapor extraction
remediation. Estimates for air permeability for the sand unit between the lens and water table range between
4.3 x 10'8 cm2 and 8.4 x 10~8 cm2. Estimates of air permeability for the brown clayey sand unit above the clay
lens range between 1.9 x 10~8 cm2 and 2.1 x 10~8 cm2.
These estimates for the air permeability of lithologic units in the unsaturated zone will lead to better
predictions of flow paths for site venting experiments. The field techniques illustrated here can be applied
at other sites to obtain the distribution of air permeability in the unsaturated zone.
157
-------
REFERENCES
1. Baehr, A.L., and CJ. Bruell, Application of Stefan-Maxwell Equations to Determine Limitations of
Pick's Law when Modeling Organic Vapor Transport in Sand Columns, Water Resour. Res., 26(6,
1155-1163, (1990).
2. Baehr, A.L., and M.F. Hull, Evaluation of Unsaturated Zone Air-Permeability Through Pneumatic
Tests, Water Resour. Res., (Accepted for Publication).
3. Baehr, A.L., G.E. Hoag, and M.C. Marley, Removal of Volatile Contaminants from the Unsaturated
Zone by Inducing Advective Air Phase Transport, J. Contam. Hydrology, 4(1), 1-26, (1989).
4. Falta, R.W., I. Javandel, K. Pruess, and P.A. Witherspoon, Density-Driven Flow of Gas in the
Unsaturated Zone due to the Evaporation of Volatile Organic Compounds, Water Resour. Res.,
25(10), 2159-2169, (1989).
5. Mendoza, C.A., and E.O. Frind, Advective-Dispersive Transport of Dense Organic Vapors in the
Unsaturated Zone, Model Development, Water Resour. Res., 26(3), 379-387, (1990).
6. Sleep, B.E. and J.F. Sykes, Modeling the Transport of Volatile Organics in Variably Saturated Media,
Water Resour. Res., 25(1), 81-92, (1989).
7. Thorstenson, D.C., and D.W. Pollock, Gas Transport in Unsaturated Zones: Multicomponent
Systems and the Adequacy of Pick's Laws, Water Resour. Res., 25(3), 477-507, (1989).
158
-------
TABLES
Table Description
1 Pneumatic Test Data Collected
6/4/1991 for Sandy Unit
2 Domain Geometry and Best Fit Air
Permeability Under Hypothesis of
No Confining Unit Above Sand
3 Domain Geometry and Best Fit Air
Permeability Under Hypothesis of
Clay Lens Acting as a Confining
Unit Above Sand
4 Pneumatic Test Data Collected
1/21/1991 for Brown Clayey
Sandy Unit
5 Domain Geometry and Best Fit Air
Permeability for Brown Clayey
Sand
159
-------
TABLE 1. Pneumatic Test Data Collected 6/4/1991 For Sandy Unit
Test Number
Injection (I) or
Withdrawal(W)
Q, Mass Flow Rate
(g/sec)
Pilm, Atmospheric
Pressure (mmHg)
T.,Air
Temperature (°C)
T^,, Soil Temperature
at Injection/Withdrawal
Depth (°C)
Re, Reynolds Number
in Pipe
f, Pipe Friction
Factor
P,/Pltm, Normalized
Pressure Measured
at the Surface
P/P VW9-7.4<"
ton
Normalized VW9-6.0
Probe Pressure VW9-8.2
1
W
0.6041
754
22
17
11,045
0.0167
0.880944
0.985138
0.999132
0.999256
2
W
0.4620
754
22
17
8,448
0.0174
0.923729
0.985166
0.999380
0.999479
3
W
0.2623
754
22
17
4,796
0.0202
0.970856
0.993006
0.999603
0.999628
4 5
I I
0.7018 0.5069
754 755
25 23
17 17
12,799 9,261
0.0163 0.0172
1.138278 1.076580
1.019580 1.011210
1.000943 1.000620
1.000868 1.000546
6
I
0.2533
755
23
17
4,627
0.0205
1.024493
1.005600
1.000298
1.000248
NOTES : (1) Pressure at end of probe adjusted for pressure losses in pipe according to equation (4)
160
-------
TABLE 2. Domain Geometry and Best Fit Air-Permeability Under
Hypothesis of No Confining Unit Above Sand
Well Locations
(centimeters)
d = 210
1 = 241
b = 301
r=3.8
w
Observation Coordinates (r,z)
(centimeters)
VW9-7.4 (3.81,225.0)
VW9-6.0 (50.0,183.0)
VW9-8.2 (50.0,250.0)
Best Fit Air Permeability Estimates
Test kr (cm2) kz(cm2)
Number
1 3.29 x lO'8 6.54 x 10'7
2.15 xlO-8 7.46 xlO-7
3.02 x 10-8 6.02 x 10-7
2.63 x lO'8 6.97 x 10"7
3.51 x 10-8 7.65 x 10-7
3.40 x 10-8 8.36 x 10'7
161
-------
TABLE 3. Hypothesis of Clay Lens Acting As a Confining Unit Above Sand
Well Locations
(centimeters)
d = 27
1 = 58
b=118
r= 3.8
Observation Coordinates (r,z)
(centimeters)
VW9-7.4 (3.81,42.5)
VW9-6.0 (50.0,1.0)
VW9-8.2 (50.0,67.0)
Best Fit Air Permeability Estimates
Test k^cm2) kz(cm2) k'/b'(cm)
Number
1 6.07 x 10-* 4.70 x lO'8 3.38 x 10'9
4.26 x lO'8 4.38 x lO'8 4.01 x 10'9
5.48 x 10-8 4.74 x 10'8 3.16 x 10'9
4.48 x 10-8 8.36 x 10'8 3.73 x 10'9
6.81 x lO'8 4.30 x lO'8 3.88 x 10'9
6.53 x 10-8 5.29 x 10'8 4.32 x 10'9
162
-------
TABLE 4. Pneumatic Test Data Collected 1/21/1991 For Brown Clayey Sandy Unit
Test Number
Injection (I) or
Withdrawal(W)
Q, Mass Flow Rate
(g/sec)
Pttm, Atmospheric
Pressure (rnmHg)
T^ Air
Temperature (°C)
T , Soil Temperature
1
W
0.3840
767
9
5.3
2
W
0.5107
767
9
5.3
3
W
0.8225
7o7
9
5.3
at Injection/Withdrawal
Depth (°C)
Re, Reynolds Number 7,144
in Pipe
f, Pipe Friction 0.0183
Factor
Pj/P^, Normalized 0.962353
Pressure Measured
at the Surface
9,502
0.0171
0.942846
15,302
0.0157
0.880218
P/P <»
' »tm
Normalized
Probe Pressure
VW9-3.0 0.978850
0.970685
0.951382
NOTES : (1) Pressure at end of probe adjusted for pressure losses in pipe according to equation (4)
163
-------
TABLE 5. Domain Geometry and Best Fit Air-Permeability for Brown Clayey Sand
Well Locations
(centimeters)
d = 61
1 = 107
b = 155
r=3.8
Observation Coordinates (r,z)
(centimeters)
VW9-3.0 (3.81,84.0)
Best Fit Air Permeability Estimates
Test kr = kz (cm2)
Number
2.06 x lO'8
1.98xlO-8
1.94xlO-8
164
-------
FIGURES
Figure Description
1 Mini-Permeability Schematic
2 Domain Separated From Atmosphere
by a Leaky Confining Unit
3 Domain with Land Surface as Upper
Boundary
4 Description of Unsaturated Zone
Lithology at Location VW9 at Galloway
Township, New Jersey, Research Site
5 Location of Probes for Pneumatic Tests
165
-------
FLOW
METER
WATER
MANOMETER
CONDENSATION
CHAMBER
E
o
TO BLEEDER AND
PNEUMATIC PUMP
-7.62cm-
'_; Bentonite seal
Sand •
NOT TO SCALE
FIGURE 1
Mini-Permeability Schematic
166
-------
- p
atrn
b
z = Q
z = b
Confining Unit or Paved Surface k
d
r
Water Table
V
FIGURE 2
Domain Separated From Atmosphere by a
Leaky Confining Unit
167
-------
z = 0
b
z = b
d
r
= P
atm
V
4Water Table or Impervious Unit
FIGURE 3
Domain with Land Surface as Upper Boundary
168
-------
Depth
Be low
Land
Surface
0 cm
46 cm
Soi I
Descr fpt ion
125cm
155cm
183cm
301cm
Topso
Brown Clayey Sand
Medi urn White Sand
Clay
Med i urn White Sand
Water Table
Figure 4
Description of Unsaturated Zone Lithology of Location VW9 at Galloway Township, NJ
169
-------
50 cm
VW9
VW9T
Z = 0
Z = 46
Z = 125
Z = 155
Z = 163
Z = 301
VAPOR
PROBE
VW9-3.0
VW9-6.0
VW9-8.2
C DEPTH)
C61)
C1D7)
C161D
•^
C207D
VW9-7.4
Topsoi
Brown Clayey Sand
Medium White Sand
Clay
C210)
C227) !=!
C273)
Medium White Sand
Water Table
FIGURE 5
Location of Probes for Pneumatic Test
170
-------
OPTIMIZATION OF THE VAPOR EXTRACTION PROCESS: LARGE
PHYSICAL MODEL STUDIES
RICHARD L. JOHNSON
Oregon Graduate Institute
Beaverton, Oregon
ABSTRACT
A series of physical model experiments are currently underway at the Oregon Graduate Institute to evaluate
how the efficiency of soil vapor extraction (S VE) can be improved. To accomplish this, gasoline which has
"leaked" into avery-large, three-dimensional physical model is being removed by SVE under a variety of
well-controlled conditions. Prior to the initiation of each SVE experiment, the distribution of the gasoline
within the unsaturated and saturated zones is carefully characterized. For the spill discussed here, the bulk
of the gasoline was retained in the unsaturated soil. It was estimated that ~ 10-20% of the gasoline reached
the capillary fringe, and some of that gasoline became trapped below the water table as the result of
fluctuations in the ground water level within the aquifer. A series of injection and extraction wells has been
installed in the sand aquifer to produce a two-dimensional flow geometry. An impermeable plastic barrier
has been placed over the ground surface, and air is being extracted from the system at -100 scfm. The
composition and concentration of hydrocarbon vapors in the extracted air are being monitored to evaluate
remediation performance. A three-dimensional array of vapor sampling points are being used to determine
the effectiveness of vapor extraction as a function of depth within the system. The sampling points are also
being used to monitorpressure within the system, and to conduct tracer tests to directly measure air velocities.
Following removal of the gasoline in the unsaturated zone, conditions will be altered to enhance gasoline
recovery from near and below the water table. Data from these experiments are being compared to numerical
flow and transport models. The combination of detailed sampling and well-characterized subsurface
conditions are resulting in a qualitative evaluation of the vapor extraction process.
171
-------
FIELD TEST OF ENHANCEMENT OF SOIL
VENTING BY HEATING
DAVID W. DEPAOLI
Chemical Technology Division
Oak Ridge National Laboratory
Oak Ridge, Tennessee 37831-6224
DR.NEILJ.HUTZLER
Department of Civil and Environmental Engineering
Michigan Technological University
Houghton, Michigan 49931-1295
ABSTRACT
An apparent means of contaminant removal enhancement during soil venting is the elevation of contaminant
vapor pressures by raising the soil temperature. This paper describes a seven-week test of heated air injection
during a full-scale technology demonstration at a JP-4 jet fuel spill site. In this test, heated air from the stack
of a catalytic oxidizer emissions control device was pulled into the soil through a passive inlet vent. Soil
temperature profiles and hydrocarbon removal rates were monitored during the test. Hydrocarbon removal
from the system as a whole during the period of the test was increased by 9% due to heating, while removal
was enhanced by 60% within the heated zone. Limited economic projections indicate that the strategy may
be advantageous provided that heat can be more evenly distributed throughout contaminated soil zones.
Recommendations for future tests and/or implementations are made based upon the results of this study.
1. INTRODUCTION
In situ soil venting, also commonly known as soil vapor extraction, is a rapidly growing technology for
remediation of unsaturated zone soils contaminated with volatile chemicals. In this technique the soil is
decontaminated in place by pulling air through the soil. The air flow sweeps out the soil gas, disrupting the
equilibrium existing between contaminant species in four phases: sorbed on soil particles, dissolved in soil
pore water, condensed in a separate liquid phase, and existing as vapor. This causes volatilization of the
contaminants and removal in the gas stream. Several well-documented field studies (1,2,3,4,5) and reported
site closures (6) have indicated the effectiveness of the technology.
Generally accepted qualitative limits for applicability of the technology are contaminants having a vapor
pressure of greater than 0.5 mm Hg (7) (66 Pa) and a soil air permeability of greater than 10~10 cm2 (8).
However, since these limits are based upon the rate of removal of contaminants, the above limits may be
extended, and applications well within the limits hastened, if removal rates may be increased by either
173
-------
shifting equilibrium conditions to higher contaminant vapor concentrations or by accelerating transport rate
processes.
One apparent means for enhancing removal rates is elevation of soil temperature. Johnson and Sterrett (3)
noted increased removal rates of 1,3-dichloropropene in field conditions with higher ambient temperatures.
Higher temperature will affect equilibrium conditions by increasing contaminant vapor pressures and
Henry's Law coefficients and by generally causing desorption of contaminants from the soil. Diffusive/
convective transport may also be somewhat affected by an increase in diffusivity and by changes in air
permeability (mainly due to changes in soil moisture content).
Of these effects, it is expected that the variation of vapor pressure will provide the largest contribution to
enhancement by heating. Table 1 shows the profound effect temperature has on the vapor pressure of some
selected compounds, as predicted by the Antoine equation (9).
Table 1. Variation of Vapor Pressure (in mm Hg) with Temperature
for Selected Compounds
n-hexane benzene toluene m-xylene n-octane
50°F
75°F
100°F
Vapor
Pressure
Ratio
100°F:50°F
76
140
260
3.4
46
90
170
3.7
12
27
53
4.3
3.2
7.8
17
5.3
5.6
13
28
5.0
This vapor pressure effect could be used to accelerate the timetable of cleanup, as shown in Figure 1. The
curves in this figure, displaying the fractional amount of contaminant remaining as a function of cumulative
air contacted per mass of initial contaminant present, were derived from a Raoult's Law equilibrium model
for removal of JP-4 jet fuel at soil temperatures of 50,75, and 100°F (10,24, and 38 °C). For an ideal case
of homogeneous air/contaminant contact and equilibrium conditions, 80% removal of JP-4 would be
achieved through contact of approximately 50 standard liters of air per gram of JP-4 at a soil temperature
of 100°F, while approximately 430 liters of air per gram of JP-4 is necessary for the same removal at 50°F.
Therefore, the cleanup using heat enhancement could be performed in 10 percent of the time of the non-
heated case. Alternatively, greater than 99% removal could be reached at 100°F in the same time necessary
for 80% removal at 50°F.
Due to the potential enhancement of extraction rates with increased temperature, several authors have
suggested means for raising soil temperatures. Anastos et al. (2) abandoned plans to heat inlet air by electrical
means due to the higher energy requirements for appreciable enhancements. Johnson et al.(10) suggested
174
-------
radio frequency and conduction heating or injection of exhaust from combustion units. Steam injection has
also been mentioned for heating the soil, both in soil venting and in an agitated soil air stripping technique
(11). It should be noted that steam injection may be less attractive due to the detrimental effect of moisture
in reducing air permeability and the possibility of dissolution and contaminant transport. However, for very
dry soil, moisture may enhance desorption of chemicals from the soil particles.
Obstacles to overcome in implementing a heat enhancement strategy include supplying the vast quantity of
energy necessary to heat the soils containing the contaminants to the elevated temperature and developing
methods for delivering the heat evenly and inexpensively. As an illustration of the magnitude of the heating
load required, consider soil having a dry density of 100 lb/ft3 (1600 kg/m3) and heat capacity of 0.2 BTU/
lb-°F (837 J/kg-°C). The temperature of a cubic foot (0.028 m3) of dry soil would rise 1°F (0.56 °C) with a
heat input of 20 BTU (21,100 J). If the soil contained 5% moisture by weight, the same soil volume would
require 25 BTU (26,400 J) for the same temperature rise. One standard cubic foot (0.028 std m3) of air cooling
from 1000°F to 68°F (538°C to 20°C) supplies 18.7 BTU (19,700 J), whereas 1 pound (0.454 kg) of steam
condensing at 212° (100°C) and cooling to 68°F (20°C) supplies 1115 BTU (1.18 x 106 J).
For the site at which this study was conducted, a contaminated soil volume of approximately 720,000 ft3
(20,400 m3) required treatment, thus, 450 million BTU (4.74 x 109J) would be necessary to raise the entire
soil volume 25°F (13.9°C), if there are no heat losses. If this were supplied by air cooling from 1000°F, as
described above, air volume is required, which, at a 1000 scfm (0.47 std m3/s) injection rate corresponds to
16.7 days. Steam injection as described above would entail 400,000 Ibs (181,000 kg) of steam to raise the
soil temperature 25°F (13.9°C). In the process, the average soil moisture content would increase from 5%
to 5.5%. However, local moisture levels, particularly in the vicinity of injection points could be much higher,
possibly significantly decreasing air permeability. (The effect of permeability on injection flow may be
insignificant, however, due to the steam pressures achievable to drive flow.) Also, the additional water would
be available for dissolution of hydrocarbons and possible transport by percolation to the saturated zone.
The above rough calculations neglect heat losses, which may be quite substantial. Although considerable
energy demands are made, the increased removal rate may make heating of the soil an economical addition
to venting systems in many cases. In order to investigate this concept, a test of heat injection was devised
and conducted during an Air Force-sponsored full-scale field demonstration of soil venting at a JP-4 jet fuel
spill site. The heat source for this study was chosen to be the stack gas of a catalytic oxidation emissions
control unit, from which heat was normally wasted to the atmosphere.
2. TEST DESCRIPTION
The in situ soil venting field demonstration was conducted at the site of a 102,000 liter (27,000 gallon) JP-
4 jet fuel spill at Hill AFB, UT. Site characterization prior to system operation indicated that the hydrocarbon
contamination was limited to a soil zone of approximately 36.6 meters x 36.6 meters x 15.2 meters deep (120
x 120 x 50 ft). Total hydrocarbon soil concentration measurements ranged from below detection (20 mg/
kg) to 6400 mg/kg. The soil at the site generally consisted of unconsolidated to weakly consolidated sand,
with thin lenses of clay. Air permeability measurements ranged from3 x 10"12to6x 10'um2. Moisture content
was generally 3 to 7 weight percent, with higher values (10 to 25 weight percent) associated with clay lenses.
175
-------
9 LI
PUB SL 'OS 1« I^nj Jaf fr-Jf J»J SAJHJ jBAOtnaj mnuqni
PERCENT OF MASS REMAINING
CM
O
cn
o
-------
The heat injection test system was a small part of the demonstration system which consisted of three
subsystems: (1) an array of 15 vertical vents and 31 pressure monitoring wells in the area west of the location
of the spill source, (2) a set of six lateral vents and 30 pressure monitoring probes installed under a new
concrete pad and dike that were constructed for the tanks after the spill, and (3) a set of lateral vents in the
pile of soil that was retained after excavation of the tanks. A common blower/emissions control system was
installed for inducing airflow from the three vent arrays and for treating emissions as necessary to meet
regulatory requirements of the state of Utah. Two rotary-lobe blowers provided the capability for extraction
of up to 0.72 std m3/s (1500 stdftVmin) of gas from the three vent systems at vacuum levels up to
approximately 25,000 Pa (100 inches of water). In order to protect against potential hazards presented by
combustible gas mixtures, flame arresters were installed at the inlet to each blower. The blowers were
controlled by an automatic shutdown system based on a combustible gas detector. Two propane-fired
catalytic oxidation units were used for conversion of the jet fuel hydrocarbons to carbon dioxide and water
before discharge into the atmosphere. A vapor/liquid separator, flowmeters, and gas monitors were also
included in the system.
Prior to the commencement of the heat injection test, the full scale demonstration system was operated for
eightmonths. During this period, 3.60 x 106 standard cubic meters (127 x 106 standard cubic feet) of gas was
pulled through the soil, extracting 45,700 kilograms (101,000 Ibs) of hydrocarbons. Concentrations in the
extracted gas had decreased from an initial reading of approximately 45,000 ppmv hexane equivalent as
measured by an on-line FID-based total hydrocarbon analyzer (THA) to approximately 700 ppmv.
The heat injection test system was constructed as shown in Figure 2. The three vents of interest in the test
were vertical vents located in the center of the most contaminated soil zone. They formed the eastern end
of the central of three vent lines in a grid of 15 vertical vents on a 12.2-m (40-ft) grid spacing, with 6.1-m
(20-ft) spacings along the central line. The eastern half of the venting system was covered by a 24 m x 43
m (80 x 140 foot) surface barrier consisting of a 10 mil polyethylene sheet covered by one 6 mil polyethylene
sheet and 30.5 cm (12 in) of soil.
Each 11.4-cm (4.5-inch) outside diameter and 10.2-cm (4-inch) inside diameter vent was installed in a 24.4
cm (9-5/8 inch) augured hole. The extraction vents each consisted of a 12.2-m (40-foot) length of flush-joint
Schedule 40 PVC screen of slot width 0.51 mm (0.02 inches), installed at a depth of between 3.0 m and 15.2
m (10 feet and 50 feet) below ground surface and capped at the lower end (see Figure 3). Flush-joint Schedule
40 PVC was used for the riser pipe. PVC cement was used to join all PVC fittings. The inlet vent was
constructed in a like manner of 4-inch stainless steel well screen and riser pipe, since the PVC was rated to
withstand temperatures only up to 60°C (140°F) in the presence of JP-4. Stainless-steel centralizers were
installed to maintain the riser pipe in the center of the borehole. Each auger hole was backfilled with dry
coarse sand to one foot above the screen, using a tremie tube. A 30.5 cm (12-inch) layer of bentonite pellets
was placed on top of the sand, and the hole was grouted to the surface by tremie tube with cement-bentonite
(9:1) grout. A concrete collar approximately 0.6 m (2 feet) in diameter was installed to provide mechanical
stability for the vent and to divert rainfall from penetrating the backfilled hole. Each vent extended at least
30.5 cm (1 foot) above the concrete collar and was protected with 20-cm (8-inch) diameter steel pipe
extending to 7.6 cm (3 inches) below the top of the vent pipe.
A tee was installed on the stack of one of the catalytic incinerators, allowing the diversion of some of the stack
gas to the injection vent. Carbon steel tubing of 12.7-cm (5-inch) nominal size was run approximately 107
177
-------
m (350 feet) to the inlet vent. The tubing was covered with 2.54 cm (1 inch) of high temperature fiberglass
insulation protected by aluminum sheet. No blower was installed in the heated gas line; rather, the test was
designed to allow vacuum induced in the soil to pull the heated gas into the vent.
Thermocouples were installed in the heads of each of the vents and in the soil in the positions shown in Figure
4. The thermocouples were placed in hand-augered holes at depths ranging from 3.3 to 4.1 m (10.9 to 13.5
ft). Since this depth is near the top of the screened interval of the wells, the temperatures are likely to be less
influenced than in the center of the affected zone.
Temperatures and concentrations were read periodically from the thermocouples and the THA as well as
continuously recorded on a chart recorder and data logger. Samples were taken periodically in canisters for
analysis of the extracted gas by a gas chromatograph (GC). The effects of elevated temperatures upon
bioactivity, which had proven to be significant in the earlier portion of the demonstration (5), could not be
measured due to the low O2 and high CO2 concentration in the inlet gas.
3. RESULTS
Heat injection was begun on 16 August 1989 and continued to 7 October 1989 with a total extraction rate
of 0.307 std m3/s (650 scfrn), with measured flows of 0.193 std m3/s (410 scfm) from the west extraction vent
and 0.113 std m3/s (240 scfm) from the east. In this configuration, a vacuum of 11900 Pa (48 inches of water)
was induced under static conditions at the inlet vent. Opening the inlet vent to the heat injection line induced
an injection flow rate of 0.044 std m3/s (93 scfm) at 423 Pa (1.7 inches of water) vacuum.
Operationally, the heat injection system ran quite well. However, due to heat losses in the piping, the inlet
gas temperature was decreased from over 316°C (600°F) at the incinerator stack to between 93 and 102°C
(200 and 215°F) at the inlet vent. Water uptake in the extraction piping was also noted to be increased during
the test in comparison to earlier operation, due to a combination of greater vacuum levels at the extraction
vents (approximately 14900 Pa - 60 inches of water) and to moisture content of the heated input air. No major
changes in the flow rates or vacuum required were noted despite the increased moisture.
The results of temperature measurements from each of the thermocouples are presented graphically in Figure
5 as a function of time of the test. The temperature at each of the points in the soil appears to have reached
steady state during the test. Thermocouples farther from the injection vent reached steady state slower than
those closer to the injection vent, and reached a lower steady state temperature. This may be seen most clearly
by comparison of temperatures at thermocouples 2, 3, and 4, which were placed 1/4, 1/2, and 3/4 of the
distance along the line from the inlet vent, to the west extraction vent. TC2 reached a steady state temperature
of 36.7 to 37.8°C (98 to 100°F) in about 15 days, TC3 reached steady state temperature of approximately
(32.8°C)91°F in about23 days and TC4 reached a steady state temperature of about 31.1°C(88°F) in 30 days.
The progression of the temperature profile is also shown in the three frames of Figure 6, displaying the
temperature (in °F) at the thermocouple positions at the start of the test, after 13 days, and after 36 days. It
should be noted that all points except those corresponding to the vents indicate soil temperature, whereas the
vent points (those points labeled with temperature values of 207°F at the inlet and 73°F at the west extraction
vent in the bottom frame) indicate gas temperature.
178
-------
a
era
n
3
ta
3
5'
ore
o
^ i
CATALYTIC
INCINERATOR
n>
05
p^
5'
o
ri
ft
X = THERMOCOUPLE
V9
V10
V11
-------
2 ft. min.
6 in. Steel Casing
Concrete
Ground Surface
Cement-Bentonite
Grout
>\XX \ \^\\'V^x'\V\\
10 ft. Below
Ground Surface
Bentonite Pellets
Coarse Sand
Riser: 3}£
Schedule 40 PVC,
Flush-Joint
Centralizer
Slotted PVC,
0.02 in. Slot Width
8 in. Auger Hole
PVC Cap-
i
'/
40 ft.
Figure 3. Vent construction.
180
-------
00
a
-------
(Do) 3anJLVa3dN3i
CO CD
CN O
00
CN
CD
CN
Csl
T
o>
I ' ' ' I 1 I I—I—I—I—I
O
K>
UJ
O 2
CM r:
o
o
o
o
CD
CD
-------
M
>2..S
<* S£
S-a
-------
CO
o
Ld
LU
in
00
O
CD
P O
°
R °
O oq
K)
O
CD
rO
o
o
oo
1
I I
1
o
o
CD
o
o
o
o
CM
to
O
LO
CD
m
in
m
m
CM
o
ui
ID
O
LU
m ID
rO O
(|U9|DAinb9 9UDX9L1 Uidd)
NOIlVdlNBDNOD
Figure 7. Variationofhydrocarbonconcentrationinthecombinedextractedgasstream. Open squares- measurements
by total hydrocarbon analyzer, solid circles - gas chromatographic analysis of samples.
184
-------
23
crS'
I
00
i
o
a
I
«
F1^
•1
00 ,
rs
n
o
E.
I"
3-
X
O
n
n
n>
r+
5'
O
o
*
c
Z D
O X
CD ®
CUMULATIVE GAS EXTRACTED (m3)
3.80 4.00 4.20 4.40 4.60
800
< c
a: JD
1 600
400
O Q.
* o, 200
REMOVAL
ENHANCEMENT
DUE TO
HEATING
X 106
130
140
150
160
170 X 106
CUMULATIVE GAS EXTRACTED (ft3)
-------
Hydrocarbon concentrations in the extracted gas are displayed in Figure 7 as pprny hexane total hydrocarbons
as a function of cumulative standard cubic meters (and feet) of gas extracted during the entire demonstration
(the heat injection test commenced at the 3.60 x 106m3-127x 106 ft3 mark). The upper points are from THA
measurements, whereas the lower points were calculated from GC samples taken from each extraction vent.
The GC results remained lower than THA results. The two sets of results display a common trait in the
general shape of the curves. Each showed a consistent decrease in concentration until the 4.39 to 4.53 million
cubic meter (155 to 160 million cubic feet) range. At this point (approximately 12 September through 15
September, or 27 to 30 days into the test), roughly the point at which the farthest soil thermocouples reached
steady state temperatures, the concentration in the extracted gas is seen to markedly increase. This
concentration increase qualitatively correlates with the temperature front approaching the extraction vent.
The nearly constant measurement of concentration by the THA from 4.45 to 4.73 million m3 (157 to 167
million std ft3) is postulated to be due to a balance of two factors, one being higher gas concentration because
of increased temperature and the other being decreased gas concentration due to decreasing soil contaminant
concentration and change in composition to a heavier hydrocarbon mixture.
Estimates of the removal enhancement by heat injection are made possible by analysis of the data as shown
in Figure 8. In this analysis, the THA data is used because of a greater number of points and less scatter. The
12 data points from 3.60 to 4.36 million m3 (127 to 154 million ft3) were fitted with a straight line which
appears to be valid over this limited range. Another line was regressed to the four points from 4.45 to 4.73
x 106m3(157to 167x 106ft3). The integrated difference between these two lines, as shown by the highlighted
area in Figure 8 is the calculated removal enhancement due to heating. From 4.45 to 4.73 x 106 m3,245 kg
(540 Ibs) is the quantity that would be expected to be removed in the absence of heating by extrapolation of
the straight line trend. The upper line indicates that approximately 417 kg (920 Ibs) were removed during
the period. The difference is 172 kg (380 Ibs), or an increase of 70%. The removal increase over the entire
period of the test, from 3.60 to 4.73 x 106 m3 (127 to 167 x 106 ft3) extracted is 8.9%, 2127 kg (4690 Ibs) actual
and 1955 kg (4310 Ibs) expected without heating. It is obvious that this ratio would increase with further
operation. Only limited soil sampling was performed in the soil zone of the heat injection test after the
demonstration, so little can be inferred as to the transport of contaminant with the temperature profile.
4. DISCUSSION
The results of this test of enhancement of removal with soil heating must be treated with care in predicting
the effectiveness of the technique in other site applications. The most obvious measures of performance in
this test are the 8.9% and the 70% values obtained above. However, these values are artifacts of the length
of the test - for instance, if the test were only operated for a period during which the heat front had not reached
the extraction vents, the enhancement value would have been 0%. Likewise, if the test had been conducted
for a longer period, a value of much greater than 8.9% would have been obtained, perhaps on the order of
the 70% enhancement measured in the period after the development of the steady-state temperature profile.
Another test condition which had a significant effect upon the measured removal enhancement was the ratio
of the inlet air rate to the extraction rate. Hydrocarbons were extracted from soil zones which were not
contacted with the heated air as well as from the soil zone of interest. Indeed, since the inlet air amounted
to a small fraction of the extraction flow rate, the majority of the hydrocarbon extraction could be assumed
186
-------
to be relatively unaffected by the heated air flow. Upper bounds on the achievable enhancement for this test
could be estimated by assuming homogeneous contaminant distribution in the soil and negligible heat
conduction. With these assumptions, the hydrocarbon removal may be split proportionally by air flow ratios
into two portions: that affected by heating and that unaffected by heating. In this approach, of the 1955 kg
(4310 Ibs) that were expected to have been removed over the entire test period, 281 kg (620 Ibs), as calculated
by (4310 Ibs x 93 scfrn inlet rate/650 scfm extraction rate) would have been extracted from the soil zones
affected by the heated air. Therefore, 1674 kg (3690 Ibs, or 4310 - 620 Ibs) would be extracted from outside
the heated zone regardless of whether heat was applied or not. Therefore, the amount extracted from the
heated zone during the test was 454 kg (1000 Ibs, or 4690 - 3690 Ibs), resulting in an enhancement factor
for the entire course of the test to be 61%. Likewise, consideration of only the period after the temperature
front arrived at the extraction vents results in an enhancement factor of nearly 500% (207 kg removed, 35
kg expected) from the heated zone. Thus, if a temperature rise similar to that achieved during this test could
have been applied over the entire site for long-term operation, it is possible that the cleanup could have been
significantly accelerated. Given the measured temperatures and extraction rates, a cleanup period perhaps
as short as one-sixth the time required for unheated soil treatment would be necessary for a well-designed
and well-operated heated system.
Despite the apparent success of heating enhancement in this test, the results do not conclusively prove the
universal value of heat injection for optimization of venting system operation. This is due to several
shortcomings of the test and to the fact that the results of this test may not be applicable to systems with
different characteristics. Shortcomings of this test entailed limitations of heat input to the soil and
distribution of the heat in the soil. Although 0.307 mVs (650 scfm) of gas at or above 316 to 371°C (600 to
700°F) was available at the stack, only about 0.045 m3/s (95 scfm) at about 99°C (210°F) was delivered at
the inlet vent, due to limited vacuum at the vent and heat losses in the 107 m (350 ft) of piping. With the
limited heat input, only a modest but measurable and effective soil temperature increase was induced.
Certainly, large improvements could be made using forced injection with a high temperature fan and
increased insulation and/or a shorter piping run. More important in uncertainty for extrapolation is the fact
that the heat was obviously not evenly distributed in the soil because of the flow geometry.
Upon consideration of these test results, one could conceive of test designs that would allow for greater
heated air flow more uniformly distributed. Two such examples are shown in Figure 9. The first shows a
ring of inlet vents (either forced or passive) surrounding an extraction vent. With a large number of inlet vents
and a surface barrier, a nearly uniform radial flow distribution could be achieved. This design is attractive
in that a balance may be achieved between the tendencies for higher temperatures and lower flow rates in
outer zones, and lower temperatures (due to conductive heat losses) but higher flow rates near the extraction
vent. The other design, more suitable for larger sites, is based upon an attempt to develop one-dimensional
linear flow between lines of inlet and extraction vents. The flow patterns would simplify the monitoring of
progress of temperature and concentration fronts and would be much more amenable to modeling.
Future tests should also address the impact of heated air injection upon bioactivity. One may be concerned
that elevated soil levels may harm bioorganisms in the soil. However, two points may be made upon
consideration of the results of this test: 1) the temperatures achieved in this test were shown to have the
potential for significant increases in removal rate by volatilization, while they were certainly not in the range
of harm to most bioorganisms except in the direct vicinity of the inlet vent (in fact, bioactivity may have been
increased in much of the heated zone due to the temperature increase), and 2) temperatures high enough to
187
-------
o
o
HEATED AIR
INLET VENTS
EXTRACTION
VENT
O
o
HEATED
AIR
INLETS
EXTRACTION
VENTS
HEATED
AIR
INLETS
Figure 9. Suggested vent configurations for improved heat injection systems.
188
-------
adversely affect bioactivity are also high enough to significantly increase the vapor pressures of contaminants
for which biodegradation may be the major means of removal in an unheated case. As was noted earlier,
monitoring of bioactivity by measurement of carbon dioxide generation was not possible during this test
because of the low O2 and elevated carbon dioxide levels in the stack gas. One possible means to avoid this
complication would be to transfer the stack gas heat to an injection stream of atmospheric air using a heat
exchanger.
Despite the shortcomings of this test, some rough estimates of the economic s of heat injection at this site may
be made. In these estimates, an equilibrium removal model was used to provide an estimate of approximately
1000 liters of air per gram of initial spill material necessary for removal of the weathered JP-4 at 12.8°C
(55°F), or 8.07 x 107 std m3 (2.85 x 109 std ft3) of air would be required. Thus, at an extraction flow rate of
0.47 std m3/s (1000 scfm), 66 months would be required for removal by volatilization in the absence of heat
injection. It is with this base case that the heat injection cases are compared. Comparison of the cases would
include operating costs and any additional capital cost for the heat injection system components. The piping
and stainless steel vents for this demonstration were estimated to cost about $40,000. For application to the
entire site possibly all five central vents or the vents on the fringes could be installed as heat injection vents.
It would be preferable to use the latter strategy, since contaminants will be driven away from the heat inlet
points. A conservative estimate of additional capital cost for the heat injection system is $50,000. Operating
cost rates would be common to each case, with or without heat injection, with a blower cost of $ 1100/month.
An average catalytic oxidation cost of $1900/month was assumed.
Four cases of heat injection corresponding to ranges of removal enhancement deduced from this test are
compared with the base case as shown in Table 2. The estimated cost of remediation in the absence of heat
injection is $750,000 (12). It is projected that a removal enhancement of approximately 33% would be
necessary for cost-recovery of the heat injection system. The first case assumes that the approximately 9%
removal enhancement obtained during over the length of this test would be applied to the entire site for the
complete remediation. At this rate, a considerable cost is projected for a 10% faster clean-up time. The
intermediate value of 70% percent, which may be considered a reasonable estimate of long-term removal
enhancement results in significant savings of both time and cost. As would be expected, the optimistic
removal enhancement value of 500% would yield remarkable savings.
The calculations of Table 2 show that heat injection would be likely to provide savings at the Hill AFB site
if reasonable removal enhancement (greater than 33%) were achieved. The results of this study indicate that
during long-term operation in venting configurations such as those discussed above this would certainly be
achieved. Savings would be greater if the site were less permeable (increasing blower costs) or if emissions
control were more costly. Certainly, heat enhancement would be much less attractive if emissions control
were not required.
189
-------
Table 2. Comparison of heat injection cases
REMOVAL ENHANCEMENT
0%
9%
33%
70%
500%
Air needed
Time at 1000
scfrn
Additional
capital
Operating
Cost at
1000
Total
operating
& additional
capital
Savings due to
heating
2.85 x 109 ft3 2.61 x 109 ft3 2.14 x 109 ft3 1.68 x 109 ft3 4.75 x 108 ft3
66 months 60.5 months 49.6 months 38.8 months 11 months
0
$0
$50,000 $50,000
$50,000 $50,000
$3,000/mo. $3,000/mo. $3,000/mo. $3,000/mo. $3,000/mo.
$198,000 $232,000 $199,000 $166,000 $83,000
-$34,000 -$1,000
$32,000 $115,000
5. CONCLUSIONS
This field test has proven the feasibility of the enhancement of soil venting through heating with the stack
gas of a catalytic incinerator emissions control device. Despite the shortcomings of the test, a measurable
enhancement of removal due to heating was detected. Results suggest that a system designed with a uniform
flow field for more even heating could remediate a site several times faster than an unheated case.
In general the concept of heat injection appears attractive when waste heat is readily available. Heat
enhancement will become more economical for systems with higher operating costs, such as sites with soils
of low air permeability or costly emissions control. Problems of poor air flow and heat distribution may
extend the time required for cleanup and decrease the economical advantage.
Further work in this area, using improved field demonstration systems as described above and complemented
with heat and contaminant transport modeling, is urged. Such work would be valuable to further illustrate
the advantages of the technique, to define ranges of site variables for which the technique is applicable, and
to provide practitioners with a means of estimating soil venting system performance with heating.
190
-------
ACKNOWLEDGEMENT
This work was funded by the Air Force Engineering and Services Center, Tyndall AFB, FL, 32403
REFERENCES
1. Crow, W. L., Anderson, E. P., and Minugh, E. M., "Subsurface Venting of Vapors Emanating from
Hydrocarbon Product on Groundwater," Groundwater Monitoring Review, vol. 7, Winter 1987, pp.
51-57.
2. Anastos, G. J., Marks, P. J., Corbin, M. H., and Coia, M. F., In Situ Air Stripping of Soils Pilot Study,
Final Report., AMXTH-TE-TR-85026, October 1985.
3. Johnson, J. J. and Sterrett, R. J., "Analysis of In Situ Soil Air Stripping Data," Proc. Fifth National
Conf. on Hazardous Wastes and Hazardous Materials, Las Vegas, Nevada, 19-21 April 1988, pp.
451-455.
4. Foster Wheeler Enviresponse, Inc., Superfund Innovative Technology Evaluation Technology
Demonstration Summary - Terra-Vac® In Situ Vacuum Extraction System, Groveland, Massachusetts,
EPA/540/S5-89/003, May 1989.
5. DePaoli, D. W., Herbes, S. E., and Elliott, M. G., "Performance of In Situ Soil Venting System at Jet
Fuel Spill Site," Soil Vapor Extraction Technology Reference Handbook, USEPA Report, EPA/540/
2-91/003, pp. 260-272, February 1991.
6. Payne, F. C. and Lisiecki, J. B., "Enhanced Volatilization for Removal of Hazardous Waste from
Soil," Proc. Fifth National Conf. on Hazardous Wastes and Hazardous Materials, Las Vegas, Nevada,
19-21 April 1988, pp. 456-458.
7. Bennedsen, M. B., Scott, J. P., and Hartley, J. D., "Use of Vapor Extraction Systems for In Situ
Removal of Volatile Organic Compounds from Soil," Proc.- Nat. Conf. Hazardous Waste and
Hazardous Materials, Washington, D.C., 16-18 March 1987, pp. 92-95.
8. Roy, W. R. and Griffith, R. A., "In Situ Extraction of Organic Vapors from Unsatruated Media,"
Report #24, Environmental Institute for Waste Management Studies, University of Alabama -
Tuscaloosa, October 1989,29 pp.
9. Reid, R. C., Prausnitz, J. M., Sherwood, T. K., The Properties of Gases and Liquids, Third Edition,
McGraw-Hill, 1977.
10. Johnson, P. C, Kemblowski, M. W, Colthart, J. D., Byers, D. L., and Stanley, C. C., "A Practical
Approach to the Design, Operation, and Monitoring of In Sim Soil Venting Systems," Ground Water
Monitoring Review, Vol. 10, Spring 1990, pp 159-179.
191
-------
11. Ghassemi, M., "Innovative In Situ Treatment Technologies for Cleanup of Contaminated Sites,"
presented at Third Annual Hazardous Waste Law and Management Conference, Seattle, Washington,
and Portland, Oregon, October 1986.
12. DePaoli, D. W., Herbes, S. E., Wilson, J. H., Solomon, D. K., Jennings, H. L., Hylton, T. D., and
Nyquist, J. E., "Field Demonstration of In Situ Soil Venting at Hill Air Force Base JP-4 Jet Fuel Spill
Site," U. S. Air Force Report AFESC ESLTR 90-21, Volume 3 (in press).
192
-------
PERFORMANCE CHARACTERISTICS OF VAPOR EXTRACTION SYSTEMS
OPERATED IN EUROPE
DIETER H. MILLER, PHD.
HPC HARRESS PICKEL CONSULT GMBH
Marktplatz 1
8856 Harburg
Germany
ABSTRACT
Vapor extraction, an in-situ process to remove volatile organic compounds (VOC) from soils of the vadose
zone, has been applied in Europe since the early 1980s. With considerably more than 1,000 systems operating
under virtually all subsoil conditions, vapor extraction is considered to be a standard procedure in Germany.
In a vapor extraction well a negative differential pressure is created by a blower or similar device. This
generates a steady flow of soil gas towards the extraction well, providing a purging of the soil with air
undersaturated with contaminants. After removal of the vapors existing under equilibrium conditions, VOC
will partition into the gaseous phase both from the liquid phase (contaminated soil moisture and/or free
product) and from the soil matrix.
Differential pressures applied to the extraction well typically range from 15"-350" of water, creating an
effective radius of hydraulic influence of about 15 ft to 150 ft. Beneath sealed surfaces, even higher radii of
influence are observed in high permeability soils.
Parameters that influence the performance of the vapor extraction system (VES) are the contaminants
involved, stratigraphy, soil type, size and type of cover materials, length and position of the screened section
in the well, and the pressure differential applied.
Discharge data from case histories of sites with halogenated hydrocarbons contamination reveal common
results, if the systems were optimally designed. Those data show that contaminant concentrations and mass
removal are high during a first phase of about two weeks, merging into a short transition phase which is
followed by an asymptotic decrease in concentrations to background levels. While concentrations in the
extraction system's discharge are reduced to some 10% of their initial value after a few weeks, it takes several
months to as long as two years to achieve a satisfactory clean-up. Because of these reductions, a
discontinuous mode of operation is applied during the closure phase.
In Germany, action levels and cleanup goals are increasingly being based on measurements of contaminant
concentrations in the soil gas rather than concentrations measured in discrete soil samples, due to potential
193
-------
inaccuracies related to soil sampling or uncertainties related to the representativeness of data derived from
a limited number of samples.
INTRODUCTION
Contamination by volatile organic compounds (VOC) has turned out to be widespread due to the almost
ubiquitous presence of those substances in industrial processes. Specifically, VOC include halogenated
hydrocarbons like trichloroethylene (TCE), perchloroethylene (PCE), or 1,1,1-trichloroethane (TCA),
aromatic hydrocarbons or petroleum products such as gasoline. Halogenated hydrocarbons, in particular,
exhibit physical properties that enable them to penetrate even coated concrete pads and seep into the ground
rapidly, both as liquids and vapors. Significant contamination not only occurs at underground storage tanks,
but also originates from above ground storage, handling and application areas as well as along pipelines or
transport pathways.
Rather than the excavation and disposal or aboveground treatment of contaminated soils, regulatory agencies
increasingly favor remedial techniques operating insitu. In the case of VOC contamination, vapor extraction
has shown to be an effective and economically feasible insitu alternative (1), (5).
VOC retained in the vadose zone represent a continuous source of groundwater contamination, as even after
a complete volatilization of free product, vapors will continue to migrate. Remediation of the unsaturated
zone by vapor extraction intercepts this migration path and is cost effective, since it captures the
contaminants prior to their dissolution in the groundwater.
PURPOSE
Vapor extraction has become a standard technology in Europe, particularly in Germany, where several
thousand systems have been operating since 1981. This has created an extensive database, facilitating the
assessment of typical performance characteristics. The comparison of case histories is suitable to:
demonstrate the common discharge characteristics of vapor extraction systems operated under
different subsoil conditions and at varying contaminant concentrations
- predict the development and progress of the remediation
- allow for the design of a treatment system for the extracted vapors.
DESIGN PARAMETERS
Aside from the type of soil and the stratigraphy, the effective radius of hydraulic influence of a vapor
extraction system is determined by a number of factors, in particular by the length and position of the
screened interval of the extraction well, the thickness of the vadose zone, and the permeability of the surface,
respectively the size of a surface seal. Differential pressures decrease exponentially with increasing distance
from the extraction well. While higher differential pressures create considerably higher volume flow rates,
the effects on the range of influence are small. However, higher pressure differentials may be required in low
permeability soils (e.g. clayey silts). This is reflected in the application of different suction devices. While
194
-------
in high permeability soils centrifugal fans or regenerative blowers are preferred, vacuum pumps are used in
low permeability soils.
Given a constant differential pressure, variations of the screen length reveal a linear relationship between the
length and the volume flow rate. Longer screens also produce a distinct increase of the effective radius.
Accordingly, the position of the screened interval of identical length creates reciprocal effects: shifting it to
deeper portions of the vadose zone reduces the volume flow but increases the range of influence (2).
Under the assumption of a screened interval of constant length positioned in the middle portion of the vadose
zone, an increasing thickness of the vadose zone results in a distinct increase of the range of influence, while
it seems to have only minor effects on the volume flow (2).
The same effect can be observed if variably sized areas of surface sealing are regarded. The presence of an
impermeable cover, such as a concrete pad, gives rise to significantly increased ranges of influence at almost
constant volume flow rates. However, a sealed surface is not a prerequisite for the successful application of
vapor extraction systems, unless highly permeable soils without a silty topsoil cover are encountered.
Aside from the casing configuration, the screen type and structure of the gravel pack have to be selected
carefully to inhibit turbulent flow in the immediate vicinity of the well. This has significant effects on the
volume flow achievable and the moisture retrieved from the soil. A vapor extraction well should be
developed similar to a groundwater well prior to continuous operation.
Independent of theoretical model calculations which, in their accuracy are limited by the data available on
the three dimensional structure of the subsurface, the actual range of influence of single systems or arrays
should be determined in the field. This can be done in several ways, by repeatedly measuring the soil gas
concentrations in vapor monitoring points at varying distances from the well, by measuring the differential
pressure created in vapor monitoring points or wells, or just by qualitatively determining the existence of a
differential pressure through the observation of the behavior of smoke trails created by air current tubes at
such points.
PROBLEMS ENCOUNTERED
Numerous sources for a deviation from the ideal progression of vapor extraction exist, leading to either an
extended remediation or, in the worst case, to a failure in achieving the cleanup goal. The two most common
problems for an inefficient cleanup of the vadose zone are the placement of the vapor extraction well(s)
outside or at the periphery of the source area, and an ongoing recharge of contaminants into the subsurface.
Such recharge can result from leakage and spills or from evaporation of VOCs from considerably
contaminated groundwater. Misplacement of vapor extraction wells can be avoided by carefully defining the
extent and the center of a source area.
Aprecise source definition is, in shallower soils, most efficiently done with soil gas investigations, the results
of which also supply an appropriate basis for the calculation of projected contaminant concentrations in the
discharge air.
195
-------
Given this information, a vapor extraction system might still operate inefficiently if, in the presence of a very
shallow groundwater table, vertical extraction wells with short screens are used. In such cases, a horizontal
screen installation within a trench provides an effective solution.
RESULTS
Case Histories
Six case histories have been selected for demonstration. The specific cases were chosen on the basis of their
documentation, variety of settings, difference in contaminant concentration, and ideal performance of vapor
extraction. The justification for the latter lies in the intention of this paper to show the predictability of the
cleanup process, if all relevant parameters are adequately considered. Knowledge of what to expect in an
ideal case allows for a timely detection of deviations which can point to deficiencies in the design or operation
of the particular system, potentially impeding a successful remediation.
For the purpose of comparison, the discharge concentration data were compiled into a composite graph (Fig.
1). Table 1 summarizes geological and technical data. To eliminate the influence of significantly different
chemical or physical substance properties, only cases with either tetrachloroethene (PCE) or trichloroethene
(TCE) as the primary contaminant were selected.
The data appear to reflect two phases merging into each other. A steep decline lasting for about 20 days (Phase
1) is observed initially, eventually continuing in a gradual asymptotic decrease to background concentrations
for the remainder of the operation (Phase 2). In all cases, the discharge concentrations decreased by 80-90
% within the first 20 days of operation. Phase 1 is more pronounced if the discharge starts out at high
concentration levels. During Phase 2, it is conspicuous that the absolute concentrations vary within a
comparatively narrow range, mostly on a level of less than 20 ppm and, in the final stage, about 2 ppm. This
is even more expressed if mass flow rates are regarded.
The decline of contaminant concentrations with time appears to be independent of the particular type of soil
(provided the organic carbon content is low), initial concentrations or the specific characteristics of the
suction device and extraction well applied in each particular case.
Figure 2 shows calculated means of discharge concentration versus elapsed time for the case histories
discussed. The measured data can be approximated by two regression curves. One of the curves fits the steep
branch of the empirical contaminant concentrations curve, the second regression curve approximates the
asymptotic decline in contaminant concentrations. The empirically derived relations are
Phase 1: C = 530r05
Phase 2: C = 230-44 In t
C = discharge concentration t = elapsed time
While the regression curves do not allow a precise prediction of the ultimate time required for a complete
cleanup, they can serve as a guideline to monitor the progress of the remediation.
196
-------
Although concentrations decrease rapidly duringPhase 1, mass balance calculations illustrate the considerable
contribution of Phase 2 even at low concentrations. Typically, 50% of the total amount of removable
contaminants are discharged after a time period ranging from about two to eight weeks, which is consistent
with other published data (e.g., Lisiecki et al., 1988). The removal of the remaining 50% is only achieved
within an additional four to twenty months of operation. Historically, the total cleanup time has ranged from
some 100 days to approximately 2.5 years, depending on the mass of contaminants retained in the soil and
the type of substances involved.
Physical Processes in the Soil
Unless a continuous recharge of separate phase material at significant rates is encountered, the presence of
free product in soils is typically limited to small fluid particles trapped in soil pores (7). At the expense of
these fluid droplets as well as of compounds dissolved in the soil moisture or adsorbed to the soil matrix, a
contaminant vapor phase will develop and spread over time controlled by diffusion and advection.
After an initial exchange of the soil gas volume in the pores, ambient air will be drawn continuously from
outside the contaminated area. While passing through the subsurface the air will be charged with VOC
partitioning into the vapor phase and subsequently be discharged through the vapor extraction system (1).
The process resembles a continuous purging of the soil with clean air and will continue until volatilization
and desorption of contaminants is complete within the limits of equilibrium partitioning.
The two phases describing the system performance are attributable to different processes becoming
prevalent in the respective stages of the operation. During the first phase, the contaminant saturated soil gas
which is present in the pore space under equilibrium conditions is discharged. A rapid evaporation of free
product droplets due to disturbance of the equilibrium is presumably also in part occurring during the first
phase. According to field tests, the time required to develop the steady state flow pattern and effective radius
ranges from about 15 to 30 minutes. It may take days for soil gas to flow from the edge of the affected soil
volume to the extraction well. A short transitional period, when the contaminant concentration in the
discharged soil gas has already decreased by more than 80%, is most likely characterized by a shift in source
of the contaminants. Rather than from an evaporation of liquid particles, the contaminants in the discharge
vapors result from a desorption of contaminants from soil particles. A second process, which is believed to
become important in this phase, is the partitioning of contaminants previously dissolved in the soil moisture
into the gaseous phase (3).
The second phase represents the comparatively slow, diffusion controlled desorption process and a gradual
reduction of the contaminated soil volume. The total cleanup time depends on the physical properties of the
compounds involved, the mass of VOC retained in the soil, the soil porosity, and the soil moisture. As long
as sufficient supplies of contaminants are existing, a nearly constant extraction rate is accomplished over a
period of several months up to about two years.
To save energy during this low yield phase, the VES may be shut down periodically for several days or weeks,
allowing a state of equilibrium to reestablish. Upon turning on the system, a concentration "spike" is
observed lasting as long as it takes to build up the flow field and to draw soil gas from its perimeter to the
extraction well. This discontinuous mode of operation is typically applied during the final months. The
maximum concentration and duration of each spike may serve as parameters to determine the completion
of the cleanup effort.
197
-------
Data from air flow models indicate a rapid decrease of pressure differentials to very low levels with
increasing distance from the vapor extraction well (2). The differential pressures throughout the majority of
the range of influence are too low to be directly responsible for enhanced volatilization of the contaminants.
However, even very small differential pressure create a pressure gradient and thus induce the flow of air
towards the vapor extraction well. The equilibrium disturbance caused by this process, rather than enhanced
volatilization through large differential pressures, is considered to be responsible for the extraction of
contaminants.
SUMMARY
In order to achieve a successful remediation of the vadose zone by vapor extraction, a thorough assessment
of the contamination pattern is a prerequisite. Factors which inhibit or impede the successful operation of
a vapor extraction system are: an ongoing release of contaminants into the subsurface, evaporation of
contaminants from a plume of considerably contaminated groundwater, and the positioning of the vapor
extraction well outside the contamination center must be ruled out.
To properly design a VES, various system parameters must be considered. These include the length and
position of the screened well interval(s), the type of screen and gravel pack, as well as the selection of the
appropriate suction device.
The application of the results of previous experiences with vapor extraction as exemplified in the case
histories facilitate:
- a prediction of the development of contaminant concentrations in the discharge air
- the design of a vapor treatment system
a timely recognition of factors impeding a successful cleanup by comparing actual data with the
model curve
- an optimized design of vapor extraction wells.
REFERENCES
1. BRUCKNER, R, HARRESS, H.M. and KILLER, D. (1986): Die Absaugung der Bodenluft ein
Verfahren zur Sanierung von Bodenkontaminationen mit leichtflUchtigen chlorierten
Kohlenwasserstoffen. Brunnenbau, Bau von Wasserwerken, Rohrleitungsbau (bbr), 37, pp 38.
2. CROISE, J., KINZELBACH, W. and SCHMOLKE, J. (1989): Computation of Air Flows Induced
in the Zone of Aeration during In Situ Remediation of Volatile Hydrocarbon Spills. In: KOBUS &
KINZELBACH (Eds.): Contaminant Transport in Groundwater. Balkema, Rotterdam, pp 437-444.
3. GRATHWOHL, P., FARRELL, J. and REINHARD, M. (1990): Desorptionskinetik fluchtiger
organischer Verbindungen bei Aquifer Material. In: ARENDT, F. et al. (Eds.): Altlastensanierung
'90. Kluwer Academic Publishers, Dordrecht, Boston, London, pp 401-408
198
-------
4. EINSELE, G., EISELE, G. and GRATHWOHL, P. (1988): Verteilung und Ausbreitung von
leichtfluchtigen chlorierten Kohlenwasserstoffen (CKW) im System Boden Wasser Luft. Dt.
Gewasserkundl. Mitt., 32, H. 4, pp.
*
5. GUDEMANN, H. and KILLER, D. (1988): In Situ Remediation of VOC Contaminated Soil and
Groundwater by Vapor Extraction and Groundwater Aeration. Proceedings HAZTECH International
'88, Cleveland, Ohio, pp 2A-902-A-111.
6. LISIECKI, J. B. and PAYNE, F. C. (1988): Enhanced Volatilization for Removal of Hazardous Waste
from Soil. Proceedings "Second National Outdoor Action Conference on Aquifer Restoration,
Ground Water Monitoring and Geophysical Methods", Las Vegas, NV, pp 1137-1146
7. SCHWILLE, F. (1984): LeichtflUchtige Chlorkohlenwasserstoffe in porosen und klUftigen
MedienModellversuche. Besond. Mitt. Deutsch. Gewasserkundl. Jahrb., 46,72 + XIIpp
8. SILKA, L. R. (1988): Simulation of Vapor Transport Through the Unsaturated Zone -Interpretation
of Soil Gas Surveys. Ground Water Monitoring Review, 8, No. 2, pp 115-123.
199
-------
Case History Data
A
Soil: 3' fi"
1 6' silty, gravelly
sands
Contaminant: TCE
*i
f
£ Depth to
Water Table: 19'
Volume Row: 85 CFM
Range of
Influence: 65'
B
10' fill
2' gravelly
sands
TCE
12'
390 CFM
180'
C
20' interb.
sands,
gravels, fills
PCE
20'
110 CFM
83'
D
6' fill
7' sands,
gravels
PCE
13'
1 03 CFM
75'
E/F
13' fill
6' silty, gravelly
sands
PCE
19'
70 CFM
40'
G
31 ' weathered
claystone,
sands, silts
PCE
31'
75 CFM
50'
HPC HARRESS PICKEL CONSULT GMBH
-------
n
Vapor Extraction
Discharge Performance
1000
Discharge Concentration (ppm)
800-
600-
400
200
0
20
120
-*- C
-B- D
y^.^' EH
F
-&- G
140
40 60 80 100
Elapsed Time (Days)
HPC HARRESS PICKEL CONSULT GMBH
-------
to
8
2
w"
U)
VAPOR EXTRACTION
Discharge Performance
Discharge Concentration
600' -
4CO H
300 -
200
100
0
n
20
40
obs. mean
BO
120
calculated
calculated
140
HPC HARRESS PICKEL CONSULT QMBH
-------
VACUUM VAPORIZER WELLS (UVB) FOR IN SITU REMEDIATION OF
VOLATILE AND STRIPPABLE CONTAMINANTS IN THE UNSATURATED
AND SATURATED ZONE
B. Herrling*, J. Stamm*, EJ. Alesi**, P. Brinnel***
* Institute of Hydromechanics, University of Karlsruhe, Kaiserstrasse 12,
D-7500 Karlsruhe, Germany
** ffiG Technologies Corp., 1833D Crossbeam Dr., Charlotte, NC 28217 USA
*** PROTEC GmbH, An den drei Hasen 21, D-6370 Oberursel, Germany
INTRODUCTION
The contamination of groundwater by strippable substances is a significant problem in all industrial
countries. For remediating aquifers in situ technologies are favored to reduce the investment and operating
costs. The paper presents an in situ method that can remove strippable substances, e.g. volatile chlorinated
hydrocarbons, and BTEX, from the subsurface (groundwater zone, capillary fringe, and unsaturated zone);
it is currently being used at numerous locations in Germany. This technology is an alternative to conventional
hydraulic remediation measures (pumping, off-site cleaning, and reinfiltration of groundwater) for the
saturated groundwater zone and for soil air extraction methods to clean the vapor zone. The contaminated
groundwater is stripped in situ by air in a below atmospheric pressure field within a so-called "vacuum
vaporizer well" (German: y_nterdruck-V_erdampfer-Brunnen, UVB), The used air, charged with volatile
contaminants, is cleaned using activated carbon. The UVB technique produces a vertical circulation flow
in the area surrounding the well, which catches the total aquifer. The vertical velocity component yields a
desired flow through the horizontal structure of a native aquifer. Numerical results demonstrate the size of
the sphere of influence and the capture zone of a well or well field in the saturated groundwater zone;
extended field measurements have been and continue to be taken (Herrling et al. 199la). The UVB
technology has most successfully been used for vapor extraction at most sites together with remediating the
groundwater zone. Using a ventilator and a double-cased screen give a lot of advantages which will be
demonstrated. For cleaning the vapor zone and the capillary fringe alone some special circulation systems
have been developed and are represented in this paper.
IN SITU REMEDIATION OF VOLATILE CONTAMINANTS BY THE UVB METHOD
The UVB helps to remove volatile substances from the groundwater, the unsaturated zone, and the capillary
fringe. When using the UVB method, a special well with two screen sections is employed, one at the aquifer
bottom and one at the groundwater surface (Fig. 1) or below an aquitard in a confined aquifer. The borehole
reach between the two screen sections should be made impermeable. One well should be used to remediate
only one aquifer (phreatic or confined) and should not connect different aquifers.
203
-------
I
JJ
fresh air activated carbon
ventilator filter
used air
cleaned air
soil air removed
via suction
stripping zone
working water level
additional pump to
support the air buoble
effect
separation plate
extensive grounawater
circulation
«•'."•;. ^ ^ borenole filling •
*•'••;. filter gravel sealing material
'acuifer bottom
Figure 1. Vacuum vaporizer well (UVB) with additional pump and separating plate.
The upper, closed part of the well is maintained at below atmospheric pressure by a ventilator. This lifts the
water level within the well casing. The fresh air for the upper part of the well casing is introduced through
a fresh air pipe: the upper end is open to the atmosphere, and the lower end terminates in a pinhole plate. The
height of the pinhole plate is adjusted such that the water pressure is lower there than the atmospheric
pressure. Therefore, the fresh air is drawn into the system. The reach between the pinhole plate and the water
surface in the well casing is the stripping zone, in which an air bubble flow develops. The rising air bubbles
produce a pump effect, which moves the water up and causes a suction effect at the well bottom. In recent
wells, a separating plate and an additional pump (Fig. 1) are used to reinforce the pumping effect of the air
bubbles. Additionally, soil air is drawn from the surrounding contaminated unsaturated zone at many sites.
Stripped air and possibly soil air are transported through the ventilator and across activated carbon, onto
which the contamination is adsorbed. Thus, only clean air escapes into the atmosphere. The cleaning effect
204
-------
of the well is based on reduced pressure, which reinforces the escape of volatile contamination out of the
water, and as a result of the air intermixing, onto the considerable surface area of the air bubbles and onto
the concentration gradient between the contaminated groundwater and the clean air. In this sense, the
permanent vibration caused by the air bubbles is beneficial to the escape process of the contamination. This
vibration is transmitted as compression and shear waves into sediment and fluid, and presumably influences
the mobility of the contaminants, even outside the well.
The upward-streaming, stripped groundwater leaves the well casing through the upper screen section in the
reach of the groundwater surface, which is lifted in a phreatic aquifer by the previously explained pump
processes and the below-atmospheric pressure. It then returns in an extensive circulation to the well bottom.
In this way, the groundwater surrounding the well is also remediated. The expansion of groundwater
circulation is positively influenced by the anisotropy existing in each natural aquifer possessing greater
horizontal than vertical hydraulic conductivities. The artificial groundwater circulation determines the
sphere of influence of a well and is overlapped with the natural groundwater flow (as described below).
The pinhole plate and all the installations within the well casing are designed as a float so they can adjust
automatically to changing groundwater levels.
For special contaminants of lower density than water, a special installation within the well is available: the
contaminated water enters the well through the upper screen, is stripped there, and with help of the additional
pump, leaves the well through the lower screen. Both installations can be used within the same well casing.
At many remediation sites, the UVB is used without an additional pump and separating plate (see Fig. 2).
In this case, a circulation flow occurs within the well casing, which is produced by the strong pumping effect
of the rising air bubbles. For the most part, the stripped water follows the path of least resistance and flows
down to the end of the suction pipe. Thus, a water of uniform temperature and oxygen content appears in the
entire well casing. The water temperature is influenced by the withdrawn evaporation heat in the stripping
zone, by the sinking of the air temperature in consequence of the vacuum in the air bubble zone, and by the
temperature of the fresh air. Depending on the groundwater temperature around the well, the water leaves
the well casing through the upper screen section and contaminated water enters the UVB at the lower screen
section. This occurs when the groundwater is colder than the circulation water in the well casing. On the other
hand, when the water in the well is colder than the surrounding groundwater, an outer circulation occurs
which is opposite to that shown in Figure 2. The water leaves the well at the lower screen section and enters
it at the upper. Both cases, influenced by density differences of the involved water bodies, have been observed
at different sites.
SPHERE OF INFLUENCE AND CAPTURE ZONE OF A UVB OR UVB FIELD
The extended circulation field outside the well is of special interest. In this paper numerical results of only
UVB installations with additional pump and separation plate will be discussed (Fig. 1). The effect of the
above-mentioned permanent vibrations, caused by the air bubbles, will not be considered. In principle, two
different cases have been considered:
• When there is no (or negligible) natural groundwater flow, the sphere of influence (or the range, R)
of a UVB is of interest.
205
-------
• When natural groundwater flow is significant, the extent of the capture zone has to be determined
for locating the well installations at a remediation site.
The resulting flow field of one or several UVB installations differs from the natural groundwater flow field
only in a limited area around the UVB. This is because sinks and sources are located at the bottom and top
of the same aquifer, each at places with the same horizontal coordinates. The effected area can, therefore,
be limited to the sum of the areas of influence of all the UVBs. When only confined aquifer conditions are
considered to reduce the computational effort, the flow field of each UVB can be superimposed onto those
of other UVBs and of the natural groundwater flow field.
To estimate the sphere of influence and the capture zone of a UVB, numerical investigations have been
performed. To calculate the complex three-dimensional flow field of a single UVB or a UVB field with
minimal effort, the following simplifications and assumptions have been used:
I
fresh air activated carbon
ventilator ty'lter
used air
cleaned air
.soU air removed
via suction
stripping zone
v
-------
• The aquifer thickness is constant.
• Only confined aquifer conditions are considered in the calculation.
• The aquifer structure is assumed radially homogeneous to hydraulic conductivities. Horizontal layers,
each with different conductivities, can be used. The hydraulic conductivities may be anisotropic, but each
horizontal layer may have only one vertical and one horizontal conductivity.
• The local below-atmospheric pressure field near the wells is neglected.
• Density effects are neglected.
• The computations assume steady-state conditions.
• For estimating the capture zone, only convective transport is considered.
The three-dimensional flow field in the above-defined, limited aquifer region is obtained by super-imposition
of a horizontal uniform flow field, computed in a vertical cross section and representing the natural
groundwater flow, and of radially symmetric, vertical flow fields for each UVB. The superimposition of the
different flow fields with their own discretization is achieved by interpolating and adding the different flow
vectors at the various nodes of a simple rectangular grid with variable grid distances that are independently
chosen for each Cartesian coordinate. The rectangular grid can be quickly and simply set up and allows for
some refinements near the wells and their screen sections. More details of the numerical computations are
given in Herrling and Buermann (1990).
Resulting Flow System
Before going into more detail, the complex flow field near an individual UVB is clarified for a vertical
longitudinal section in the direction of the natural groundwater flow (symmetry plane of the flow problem).
In Figure 3, the streamlines of three case studies are illustrated with Darcy velocities (v) of natural
groundwater flow of 0.0 m/day, 0.3 m/day, and 1.0 m/day. All other parameters remain constant: the
discharge (Q) through the well casing is 20.16 m3/hr, the thickness (H) of the aquifer is 10 m, the anisotropic
hydraulic conductivities are KH= 0.001 m/sec (horizontal) and Kv = 0.0001 m/sec (vertical), and the lengths
of the screen sections are s^ = 1.2 m at the bottom and Oj. = 2.1 m at the top.
Figures 3b and 3c show that the groundwater, flowing from the left, dives downward to the lower screen
section and is transported upward within the well casing, and that the cleaned water flows out to all sides at
the upper screen section. The flow situation can only be calculated and plotted in such a simple way in this
longitudinal section, otherwise the complex three-dimensional flow field has to be considered.
For a deep aquifer contaminated only in the upper groundwater zone, a UVB installation can be used at a
hydraulically imperfect well. The resulting flow system is demonstrated in Figure 4, clarified for a vertical
longitudinal section in the symmetry plane (Fig. 4b). The used parameters are the same as for Figure 3b. The
only difference is that the aquifer thickness (H) is 30 m (well length = 10 m, as before).
At most of the UVB installation sites, a natural, nonnegligible groundwater flow will exist. For a normal
withdrawal well, a separating streamline can be determined: all the water within this line is captured by the
well, and all water outside of it passes the well. In principle, the situation is the same when using a UVB.
In contrast to a normal withdrawal well, where the flow can be considered horizontal, the flow around a UVB
must be regarded as three-dimensional. Thus, the water body, flowing toward the UVB from upstream and
being captured by the lower screen section, cannot be delimited by a simple separating streamline, but by
207
-------
Figure 3. Streamlines clarified for a vertical longitudinal section with natural velocities: (a) 0.0 m/day;
(b) 0.3 m/day; (c) 1.0 m/day.
Figure 4. Streamlines at a hydraulically imperfect well clarified for a vertical longitudinal section with
natural velocities: (a) O.Om/day; (b) 0.3 m/day.
208
-------
a curved separating stream surface. This can be calculated as described in Herrling and Buermann (1990):
on the basis of the three-dimensional flow field, a three-dimensional, particle-tracking method is used. The
water body within the separating stream surface is captured by the UVB, and that outside of it, which flows
from upstream, passes the well.
In Figure 5 the outer surface of the capture zone, calculated numerically, and the surrounding horizontal
aquifer bottom and aquifer top are plotted for two situations (the natural groundwater flows from the
background at the right side to the UVB, as shown by the vectors). Figures 5a and 5b were calculated for the
situation described for Figure 3b; the only difference is that for Figure 5a the vertical hydraulic conductivity
is Ky = 0.001 m/sec, which means the calculation is performed for isotropic conditions. The figures have a
visible basis area of 50 m • 50 m (Fig. 5a) and 100 m • 50 m (Fig. 5b).
(a)
UVB
KH= K^ 0.001 m/s
(b)
., KH= 0.001 m/s
= 0.0 001 m/s
Figure 5. Separating stream surface of the capture zone for the situation of Figure 3b: (a) KH = 0.001
m/sec (isotropic); (b) anisotropic KH/Ky = 10.
209
-------
The captured water is cleaned within the well and leaves it through the upper screen section in all directions
(not shown in Fig. 5). Parts of it are again captured by the lower screen section, and the rest flows directly
downstream.
If a wide plume of contaminated groundwater is to be cleaned, one UVB might not be enough to capture the
whole plume. Different UVB installations can be arranged, for example, in one line normal to the natural
flow. An important question concerns the maximum distance that allows no contaminated water to flow
between two neighbouring wells without being cleaned. Figure 6 demonstrates such an example for the
situation of Figure 5b where the maximum well distance is 46 m. The visible basis area of Figure 6 is to 150
m»150m.
UVB,
KH= 0.001 m/s
Kv= 0.0 001 m/s
Figure 6. Separating stream surface of the capture zone for the situation of Figure 5b, but for two UVB
installations at a maximum distance.
Figure 7 presents a view of the separating stream surfaces of all three water bodies in connection with the
flow around a UVB. The natural groundwater flow comes from the left side. (In Figure 7b the three water
bodies were artificially separated for clarification.)
At the left side of Figure 7, the separating stream surface of the contaminated groundwater captured by the
UVB can be seen. In the center a water body is shown which consists of cleaned groundwater and shows the
circulation flow around the UVB. At the right side of Figure 7, the separating stream surface of the cleaned
groundwater flowing downstream is displayed. The calculation has accounted for the following dimensionless
parameters: Q/OHPv) = 30, a/H = 0.25, and KH/KV = 5. The screen lengths at the bottom and top are the same:
210
-------
(a)
(b)
Figure 7. Separating stream surface of the different water bodies in the outside flow of a UVB:
captured, circulating and flowing downstream water in (a) a real situation, and (b) water bodies
separated for clarification.
Diagrams for the Dimensioning of UVB Installations
Absence of Natural Groundwater Flow. At sites without natural groundwater flow, the sphere of influence
(R) of a UVB is of special interest. R is dependent on the anisotropy (horizontal over vertical hydraulic
conductivity: Kj/Ky), on the thickness (H) of the aquifer, and on the length of the screen sections aT and ^
at the top and bottom of the aquifer (see Fig. 8) or the ratio a/H (when the same length of the screen section
is used for both, then only a is referred to). Although R is mathematically infinite, it is, in practice, defined
as the horizontal distance from the well axis to the farthest point at which circulation flow is still significant.
In a dimensionless description, R has been made dependent on the ratio QR/Q, where QR is that water quantity,
which circulates within the distance R from the well. The ratio QR/Q, which is prescribed for practical
reasons, describes the strength of a circulation flow at the distance R from the well.
In Figure 9a, results are presented for ratios Q^Q=0.98 and 0.8 and for a = a,. = \ in a dimensionless diagram.
The sphere of influence (R) is independent of the discharge through the well, but strongly dependent on the
anisotropy Kj/K^ Within usual proportions, the length of the screen sections has only a small influence. For
a UVB with separating plate and additional pump, a totally screened well casing should be avoided because
of hydraulic short-circuiting.
211
-------
Figure 9b presents a dimensionless diagram that describes the differences (Ah) of the hydraulic heads
between the top and bottom of a double-screened well through which a discharge (Q) is pumped. Ah is
dependent on the parameter Q/^K,,) and the ratios KH/Ky and a/H. Abiding by the above-described
assumptions, the rise of the hydraulic head at the top of the well amounts to Ah/2, and the decrease is -Ah/
2 at the bottom (both referring to the position of rest). When using the UVB for stripping, the falling, stripped
water in the reactor causes a dynamic effect that will influence the upper hydraulic head within the well.
For the dimensioning or examination of a site, Figure 9b is a valuable expedient. When K^ is known (e.g.,
by pump test) - along with H, Q, and a - Figure 9b and the measured Ah allow an estimate of the anisotropy
at a site.
Or =0 -2TrrJvrdz
z,h
p
UVB
Figure 8. Notation in a vertical cross section.
(a)
(b)
£
H
Ah
H
0.5
0.4.
0.1
^=0.1
01 2 3
56789 10
•ID-*
Q
'H
Figure 9. (a) Sphere of influence (R) for a site without natural groundwater flow, (b) differences (Ah)
of the hydraulic heads between the top and bottom of a well. '
212
-------
Presence of Natural Groundwater Flow. At most remediation sites a natural groundwater flow exists.
Figure 11 shows numerical results represented in dimensionless form for the dimensioning of UVB
installations under these conditions. Figure 10 introduces the notations for an upstream cross section through
the capture zone normal to the natural groundwater flow direction (comparable with the open influx region
to the left of the capture zone in Figure 7) for one and two UVB installations. It is often the case when
remediating a wide contamination plume, that several wells are used in a line normal to the direction of the
natural groundwater flow. The length (D) denotes the maximum well distance at which the contaminated
groundwater cannot pass between the wells without being cleaned. The results of Figure 11 have been
calculated for an upstream distance of 5H from the well and for a constant ratio of a/H = 0.25 (screen length
over aquifer thickness). The results are discussed for wells which pump upward.
The widths BT and BB of the upstream capture zone, measured at the aquifer top and bottom, are shown in
Figure 11 a. The ratios B^ and BB/H are dependent on the ratios Q/(H2v), K^K^,, and a/H. v denotes the
Darcy velocity of the natural groundwater flow; all other variables are explained above. For small values of
Q/(lPv), the upper part of the capture zone does not reach the top of the aquifer. This implies that for
remediating a plume, a minimum well discharge (Q) is required. Again, the results are quite sensitive to the
degree of the anisotropy (see Fig. 5, as well).
Figure lib shows the results for the influx area (A) of the upstream capture zone, and Figure lie the
maximum well distance (D) of two wells between which contaminated groundwater cannot pass without
being treated. The ratios A/H2 as well as D/H are dependent on the same parameters as the widths BT and BB.
When a plume of width W is to be cleaned, the number (n) of necessary UVB installations can be estimated
by n = (W-BT)/D+1.
H
UVB
UVB
Figure 10. Notations in an upstream cross section through the capture zone for one and two UVB
installations (for wells pumping upward).
213
-------
(a)
(b)
0 5 10 15 20 25 30 25 10 15 50 Q
0 51015202530354)1550 Q
(c)
(d)
0 5 10 15 20 25 30 35 CO 15 50 Q
H2V
0 5 10 15 20 25 30 35 40 15 50 Q
H2v
Figure 11. (a) Widths BT and BB of the upstream capture zone at the aquifer top and bottom; (b) Influx
area A of the upstream capture zone; (c) Maximum well distance (D) at which the contaminated
groundwater cannot pass between the wells without being treated; (d) Upstream discharge (Qo) in the
capture zone, which is diluted with the circulating water to the total well discharge (Q).
214
-------
When a plume is remediated, the contaminated water of quantity Qo, flowing into the capture zone of a U VB
(single well) from upstream, is diluted with treated water that has already flowed through the well and
circulates around the UVB. Thus, the contaminant concentration of the water within the well casing will be
lower than in the upstream plume; at wells near a contamination source the situation is contrary. Figure 1 Id
illustrates the expected portion Q^ which belongs to the total well discharge Q. The ratio Q^Q is again
dependent on the same parameters as the widths of the upstream capture zone. Figure lid can be used to
estimate the expected concentration value of the water within the well casing for the dimensioning of a UVB
installation. To evaluate the progress of remediation at a site it may help top determine the concentration data
of the upstream plume and of the water within the well.
In Figure 12 the upstream distance (S) of the stagnation point at the aquifer top from the well axis is described
(see Fig. 3b and 3c, as well). The ratio S/H is also dependent on the parameters Q/CHV), K^KV, and a/H. The
location of the stagnation point is highly sensitive to the anisotropy of the aquifer. The length of the screen
section is of small importance within usual proportions (as described above). The knowledge of the distance
(S) from the stagnation point can be used to determine the positions of measuring equipment. The operation
of a UVB can also be supervised using depth dependent measurements between the stagnation point and the
well.
The sphere of influence of the circulation around a UVB at sites with natural groundwater flow is of special
interest. This sphere of circulation is limited in a quite different way than at a site with absence of natural
flow (Fig. 9a) as can be seen in Figures 3b, 3c, 4b, and 7. In the direction of natural groundwater flow, this
sphere has a maximum expansion of (S) (see Fig. 12) to the upstream and downstream sides. Normal to this
direction, the maximum radius of the sphere of circulation is approximated by (BB+BT)/4 (Fig. 11 a), and, in
the case of several wells in one line, by D/2 (Fig. lie).
0 5 10 15 20 25 30 35 43 45 50 Q
Figure 12. Upstream distance (S) of the stagnation point from the well axis.
215
-------
Figures 9,11, and 12 can be used for the dimensioning of a UVB or UVB field when the parameters
and Q/CH^v) can be estimated, where Q depends on the well size and on the additional pump. For an irregular
well field, a layered aquifer, or special critical cases, numerical calculations can be performed.
SOIL VAPOR EXTRACTION WITH DIFFERENT TECHNIQUES: A COMPARISON STUDY
(Contribution by P. Brinnel)
Description of the Initial Situation
At a site polluted with chlorinated hydrocarbons (CHC) on the property of a metal treatment plant located
in the Rhine - Main area of Germany, the soil air in the unsaturated zone was found to be contaminated with
500-800 mg/m3 of volatile CHC. Tetrachloroethene was identified as the primary contaminant. This
contamination was found to exist over an area of approximately 230 m2 and down to a depth of 2 m under
the ground surface. On the same property evidence of additional, secondary contaminations was present.
Within this study, however, these secondary contaminations will be ignored. The soil air remediation
measures undertaken in this study were focused on the center of the primary contamination, which lies under
a production building. For the comparison of different venting techniques under similar conditions,
remediation wells were drilled in pairs with the well casings of each pair lying one meter apart from each
other. The distance between the position of each pair is 10 m.
Geological Situation
Under a 0.20-0.30 m thick cement floor lies a 0.30-0.40 m layer consisting of fine gravel middle to coarse
sands. The following layer is 0.60-0.80 m thick and contains siltly fine sands. Under these moist, fine sands
is deposited a partially fine sand silt layer with varying contents of clay which extends to 2.30 m below the
surface. Beyond the boundaries of the contaminated area, this silt layer is found to extend to depths of
approximately 7 m.
Design of the Remediation Wells
Two pairs of remediation wells were constructed on the site. At each position of a pair one well was outfitted
with PVC slotted screens (LB1S and LB2S) and the other well with double-cased screens (LB ID andLB2D).
The construction diagrams for the well are given in Figure 13. The slotted screen, a trade standard 2" PVC
pipe with a slot width of 0.5 mm, possesses an open screen area of 5-7% of the screen's total surface area.
The slot perforation value is given with 0.057. Using these values, the open screen area is calculated to be
89.5 cm2. The double-cased screen (diameter 61 mm) possesses an open screen area of almost 42% of its total
surface area. A slotted screen of a diameter of 61 mm with such an open screen area would have a perforation
value of 0.421. Using this value for the double-cased screen, its open screen area is calculated to be 806.4
cm2.
The double-cased screen is a double walled screen unit whose annulus is filled with a hydrophobic material.
It has been patented in Europe and the United States by the company IEG mbH, Germany.
216
-------
P 5
Jl
iJ C
M «
i?
i
D CM T
0 O
1J 1
/
/
«i fll
IU
O CO «- CM T
O O f -7
<<
f-
«
l?!
rs
s
II
If
li
T- •- OJ Cl^-
ooooooooo
I W//////////////W/W///A
ONISY3 OtTOS 1N3WS3S N33HOS
ONISV3 OHOS 1N3WD3S N33HOS
I5
^ *
> 2
>m
11
«a
Figure 13. Construction diagram of the soil ail wells and pressure distribution along the
well casing.
217
-------
Operation of the Remediation Wells
Due to the unknown influences from the adjacent secondary contaminations in this case, a stable, slowly
descending contaminant removal curve was not to be expected. Such a contaminant removal would be
necessary for examination and comparison at the effects of different techniques on contaminant removal.
Therefore, the contaminant removal data and the changes of contaminant distribution within the soil were
not incorporated into this study. During the ventilation of soil air, water is always additionally removed as
air humidity. This phenomenon and the effects arising from it are used as parameters in this study for
comparison and evaluation purposes. Specifically, the content of water within the remediation well itself and
in the extracted soil air were monitored.
Each pair of remediation wells was equipped with two different types of air suction pumps; a radial blower
(ventilator), which generates a large air flow volume while creating only a low vacuum, and a side channel
blower, which creates a high vacuum but generates only a small air flow volume being conditioned by design.
Each of the four combinations of pump/well construction was tested and results were subsequently recorded
for a remediation period of more than 4000 hours. The following conditions are valid for each pump/well
combination:
(1) The remediation wells of each pair are situated in the same distance from the
center of the primary contamination within the soil.
(2) The screened segments of all remediation wells are equally long (1 m), and the wells of each pair are
positioned at the same depth level.
(3) Neither groundwater, stratum water, nor infiltration water was encountered during the drilling of the
wells. The soil which was drilled through had a normal moisture content.
Data and Calculated Values
At the beginning and the end of each test run for a particular pump/well combination, the soil air temperature,
humidity, and flow rate were measured as was the vacuum within the well. Soil air samples were also taken
at these times. If any single test run lasted longer than 14 days, all measurements were repeated and average
values were determined. In addition, the accumulation of water in each remediation well was monitored and
if present, measured. During operation of the venting systems, a profile of the pressure distribution along the
screen sections was determined.
•
From existing data and the measured parameters, the following values were calculated and utilized for
evaluation purposes.
The air volume amounts are given in cubic meters, normalized to 0° C (273.15 K), 1013.25 mbar, and 0%
humidity (dry air). The absolute air humidity was calculated from the relative air'humidity and the air
temperature. The velocity of the soil air entering the well was determined with respect to the open screen area
and using the calculated air volume flow rate and the respective perforation values. The water transport rates
were calculated using the normalized air volume flow rates.
218
-------
Results from the Soil Air Venting Test Runs
The results of the individual test runs are graphically presented in Figures 14-16. Of particular note is that
over the entire course of the remediation operation, the absolute humidity of the extracted soil air continually
decreased. Large differences in water transport rate, water accumulation within the well, and vacuum existed
among the different pump/well combinations. These combinations also yielded very different distributions
of pressure along the screen segments of each well (see Figure 13). Comparing the air volume flow rate values
obtained using the same type of suction pump, it is seen that wells at location LB1 were slightly more
permeable than those at location LB2. The average values for the individual test runs are collectively
presented in Table 1.
Table 1. Average values for the individual test runs (YEN = ventilator; SKV = side channel blower;
SF = slotted screen; DMK = double-cased screen).
Technical combination:
water transport rate (g/hr)
air entrance velocity (cm/sec)
vacuum (mbar)
soil vapor volume (m3/hr)
operating period (days)/maximum
water level above well bottom (cm)
VEN/SF
420
81.5
43.5
26
16/0
SKV/SF
850
270
164.3
61
14/46
21/46
VEN/DMK
460
13.1
41.7
38
87/0
SKV/DMK
850
25.4
121.5
74
28/0
Analysis of Results
The results obtained from the test runs show that the soil - soil air system reacts to the different venting
techniques in quite different ways. The most significant result was that water accumulated only in the
remediation well which featured the highest vacuum and the screened segment having the smallest open
screen area, which in turn generated the highest air entry velocities. Soil air is by nature saturated with water
and shows during a venting process relative humidity of approximately 100%. During a soil air venting
remediation featuring a high air entry velocity through a slotted screen having sharp inner edges, the moisture
present as air humidity will most likely precipitate as water droplets, which will then collect in the well
casing. Over a longer period of time, this effect can cause a water accumulation within the remediation well,
and from this, a saturated zone with downward flowing water can develop. This zone can eventually come
into contact with groundwater lying below the well site. The analysis of soil samples taken from such a zone
219
-------
(a)
100-
90-
§ 80-j
£
I 70-
CJ
1 60-
0>
5 50-
Q.
? 40-
<
30,
on-
Comparison of Different b
Venting Techniques - Remediation Well 1 |
.
i
H
!\ P
: \ / •
'•'..
'W '
I.
""••^.i
l
(
\
l
""-•-•.
''H
\
1 |
;
pise r^ 1
Air Output Halo
^f
•160 Vacuum
-140
^^
-120 |
E
-100 3
o
80
60
0 500 1000 1500 2000 2500 3000 3500 4000 4500
Time [h]
SF
VEN
(b)
Comparison of Different L
Venting Techniques - Remediation Well 2 |
90-
S SO-
IL
T3~
S 70-
1 60-
(?
1 50-
Q_
0 40-
30-
m
1
im*^_ $ '
•
;
/
i
i
•
•
j
i
i
i
m
i.
"'*
-180 I 1
Air Output Rate
y£
- 1 60 vacuum
-140
-120 2
e
=3
-100 3
>
-80
60
0 500 1000 1500 2000 2500 3000 3500 4000 4500
Time [h]
SF
VEN
Figure 14. Comparison of different venting techniques (DMK = double-cased screen; SF = slotted
screen; SKV=side channel blower; VEN = ventilator): normalized air output rate (m3/hr) and vacuum
(mbar); (a) at position 1 (LB1); and (b) at position 2 (LB2).
220
-------
(a)
Comparison of Different L
Venting Techniques - Remediation Well 1 I
DMK
SF
SKV
VEN
1,4
1.2-
1-
0,8-
0.6-
0,4-
I
0,2-
0
0
1-12 E
s
•10 f
£
•8 |
3
O
6 I
4
2
500 1000 1500 2000 2500 3000 3500 4000 4500
time [h]
(b)
Comparison of Different k
Venting Techniques - Remediation Well 21
0.2
500 1000 1500 2000 2500 3000 3500 4000 4500
Time [h]
DMK
SF
SKV
VEN
Figure 15. Comparison of different venting techniques (DMK = double-cased screen; SF = slotted
screen; SKV = side channel blower; VEN = ventilator): water transport rate (kg/hr) and absolute
humidity (g/m3); (a) at position 1 (LB1); and (b) at position 2 (LB2).
221
-------
(a)
Comparison of Drfferent U
Venting Techniques - Remediation Well 1 1
400 ""^
350-
300-
t
~ 250-
S
> 200-
c
| 150-
LU
.j
100-
50-
•W
r
i
i
:
• I
' 1
T /
i
I
I
•* "•-• /
•-- :
; /
„!•-"--•« J '
•
_
-m-m 1 - "
0 500 1000 1506~ 2000 2500 3000 3500 4000 45
Air Entrance Vekx;.
-« **-
Water Level
o
-35 E
0
-30 m
1
-25 g
o
-20 5
S
-15 ^
O
10 5
5
DO
Time [h)
UMK
SF
VEN
(b)
Comparison of Different ^
Venting Techniques - Remediation Well 2 |
350
300-
V 250-
1
f 200-
>
c 150-
01
5 100-
I
50-
0-
I
r
1 ;
\!
r
/
../
//
; I
1
: f
JB ^ •!
• •' . i /
* « • 1
1
• -.-•- m ~" « - ---« •-*
-ou
Air Entrance Veloc
-«-
-25 Water Level
E ^™"
E
-20 £
Q
&
"o
-15 o
5
-10 »
1
-5
0 500 1000 1500 2000 2500 3000 3500 4000 4500
Time [h]
VEN
Figure 16. Comparison of different venting techniques (DMK = double-cased screen; SF = slotted
screen; SKV = side channel blower; VEN = ventilator): air entrance velocity into the well (cm/sec) and
water level above the well bottom (cm); (a) at position 1 (LB1); and (b) at position 2 (LB2).
222
-------
underneath the bottom of a soil air venting well being developed during a remediation case, where slotted
screen segments were utilized, showed the soil to contain high concentrations of volatile CHC. The
continuous decline in the humidity of the extracted soil air over the entire remediation period could be an
indication that the soil in the immediate vicinity of the remediation wells slowly lost moisture and became
drier. Due to the situation present in this case, this phenomenon could not be linked to any specific venting
technique or combination. High air entry velocities at the screen segments of a remediation well, however,
create high nonuniform air flow velocities within the soil. When this occurs, a slow drying out of the surfaces
of the fine-grained soil particles can take place. This can result in the vaporized contaminants being removed
from the soil air and will be absorbed onto these dry surfaces (see Pederson and Curtis, 1991, p. 29ff), and
thus only being displaced within the soil without being removed from the underground. Further studies
should demonstrate if these processes can be confirmed by taking soil samples near the screen section of a
remediation well throughout a remediation period.
The results obtained from these test runs suggest the venting techniques which produce high soil air velocities
yield significant negative side effects within the soil - soil air system. In order to better understand these
effects, more research is necessary. For the formulation and optimization of soil air remediation strategies,
venting techniques which do not cause the accumulation of water in the remediation well and do not promote
the drying out of the soil in the vicinity of the soil air venting well are preferable. For keeping these negative
effects to a minimum, a greater open screen area, low air flow velocities within the soil and at the screen
segment openings, screen openings free of sharp edges, and the establishment of an even vacuum distribution
along the length of the screen segments are required.
The pressure values obtained from the test runs, which used screen segments of 1 meter in length, show that
the achievement of an equal vacuum distribution occurred most likely and was mostly even with the
combination of the double-cased screen and the radial blower. All the data indicate that the flow conditions
at the entry surface of the screen and within the soil surrounding the casing promoted the development of
a laminar flow of the soil air to the greatest possible extent. When considering the major factors involved
in the development of a thorough, fast, and cost-effective remediation process for a contamination of the
unsaturated zone, namely screen segment's entry speed (slow), vapor volume flow rate (high), and water
transport rate (low), the combination of double-cased screen and radial blower offer the best solution of the
four combinations tested.
In order to create comparable operating conditions for the test runs carried out in this study, remediation wells
with small diameters and relatively short (1 m) screened segments were utilized. Further studies should show
if the results obtained here can also be achieved using wells of greater diameter and longer screened
segments.
SOIL AIR CIRCULATION FLOW AND COAXIAL GROUNDWATER VENTILATION
(Contribution by E.J. Alesi)
Soil Air Circulation Flow
Directed soil air circular flow systems (SACF) are employed for the remediation of soils polluted with
volatile contaminants (see Figure 17). In addition, it can be used to inject gas into the soil for the stimulation
of biological or chemical degradation. SACF is a process patented by IEG mbH, Reutlingen, Germany. The
223
-------
VQ
§
to
to
I
f?
CM
o
O
sr
^f-
o'
a
l
Soil Air Circular Flow
o
Activated
Carbon
O
Axial Ventilator
Water Separator
Venting Options
- One-Way Venting
- Circular Venting (in either direction)
- Capillary Water Remediation
IEG-Process
System Sketch GfS
Capillary Zone
Groundwater Table
Pressurized Air
Distributor
-------
filtercasing built into the bore-hole is separated into an upper and a lower section, each of which is connected
to the above-ground ventilator. This allows for the withdrawal of air from either segment individually or from
both simultaneously. The air extracted, after passing through a suitable remediation unit (i.e. activated carbon
filter), is reinfiltrated into the soil. Horizontal and vertical flow circulation are generated in the soil
surrounding the extraction well. The circulation direction is reversible and can be adjusted according to the
distribution in the soil.
The SACF, in contrast to conventional venting methods, is capable of generating a directed circulation
through the center of the contamination. No fresh air is added to the circulation system. Air passing through
the ventilator is heated, thereby enhancing desorption of contaminants adsorbed onto soil particles. This
leads to a more effective remediation of the site.
For stimulating the biological degradation of contaminants, nutrients, in liquid or gas form, can be introduced
into the circulation. Chemical conversion of toxic substances into harmless and/or immobile material can
be achieved in situ by introducing, for example, strongly reactive gases into the soil.
If only biodegradable substances are to be removed from the subsoil a SACF system (without an above
ground extraction unit) consisting of an axial ventilator in the screened well can be implemented.
Should the capillary fringe be remediated along with the unsaturated (vadose) zone, it is possible to rinse the
capillary fringe by creating a circulation directly around the well casing. This is achieved by a combination
of SACF and coaxial groundwater ventilation (see next section).
Coaxial Groundwater Ventilation
Coaxial groundwater ventilation (CGV) is used in the remediation of groundwater and of stratum water
contaminated with volatile pollutants, but can also be employed to inject gas into the groundwater for the
enhancement of microbiological degradation (see Figure 18). CGV is a method patented by IEG mbH,
Reutlingen, Germany.
CGV consists of a combination of soil air venting with in situ groundwater stripping ("push and pull
technique"). Clean compressed air is pumped into a pressurized air distributor located between the capillary
fringe and the aquifer base according to the vertical distribution of the pollutants. The air bubbles rise within
the well, causing the water inside the well casing to flow upward (air-lift effect).
The design of the pressurized air distributor regulates the air flow so that the air can only flow upward.
Consequently a continuous circulation of groundwater is generated in the area surrounding the remediation
well, delivering new contaminants to the stripping site.
In contrast to other in situ stripping methods, the clean water leaving the upper screen section of the well has
no air bubbles, therefore no air-water phases can impede the flow.
Volatile contaminants dissolved in the groundwater are transferred from the liquid to the gas phase in an
amount relative to their gas-liquid distribution coefficient and are extracted from the groundwater surface
via the double-cased screen. Soil air from the unsaturated zone is also extracted and remediated. By
225
-------
Soil Air Venting with a Double-Cased Screen
Coaxial Groundwater Aeration
Activated
Com-
pressor
n n
(
^aroon
Filter
i- -i
Pressure
Regulator^.
1
Cement Cap
\_n.
Vacuum
Activated Carbon
Filters
Unsaturated
Zone
Water Table Upwelling
(Caused by Vacuum)
o ft ft n
Pressurized Air Distributor
1
Ui
rump
n n
i i
i i
Double-Cased
Screen Filter
Groundwater
Circulation
IfcG-Process
System Sketch CIS
Figure 18. Principle sketch of the coaxial groundwater ventilation.
226
-------
monitoring the volume of exhaust air the amount of air extracted from the unsaturated zone can be calculated
(exhaust air - air injected into well). Small monitoring tubes located in the annular space of the remediation
well allow the concentration of the contaminants in the groundwater to be measured before and after passing
through the well.
Vertical water and air flows differentiates the CGV from other in situ stripping methods. Of special note is
that groundwater need not be removed to prevent contaminant propagation when using the method. Best
results can be achieved if fresh air is continuously sucked into the pressurized air distributor. Heating the
groundwater and consequently increasing the off-air moisture is thus avoided.
A special advantage of the CGV is its ability to effectively remediate the often highly contaminated capillary
fringe.
The difficulties that arise during conventional remediation procedures due to contaminated stratum water,
which collects in the remediation well do not occur with the CGV. Stratum water can be stripped directly in
the ground without having to pump it up to an above-ground treatment system.
CONCLUSION
The U VB technique can be used for in situ stripping of volatile contaminants from the groundwater zone and
to clean the unsaturated zone at the same time by soil vapor extraction. For the saturated groundwater zone,
the hydraulic circulation system of the UVB offers many advantages, particularly when compared with a
typical hydraulic remediation system of pumping, off-site treatment, and reinfiltration of the groundwater.
Such advantages include:
• No lowering of the groundwater level
• No groundwater extraction
• No waste water
• Less permeable, horizontal layers are penetrated vertically
• Remediation of the groundwater takes place down to the bottom of the aquifer
• Even at low well capacity, remediation operation is continuous
• Soil air extraction is possible at the same time
• Low space requirement
• Investment and operating costs will be considerably lower.
When the water discharge through the well casing is directed downward, the hydraulic head is lowered at
the well top (-Ah/2, Fig. 9b; neglecting the increasing groundwater level in consequences of the vacuum),
but this amount is much smaller than that caused by a normal withdrawal well.
The total aquifer is caught by the circulation flow of a UVB. When using different wells for extraction and
infiltration, only those areas of an aquifer which are more permeable are penetrated. The other areas are
reached mainly by diffusion. The groundwater flow system will only locally be influenced, there is no need
for large extended groundwater flow investigations. Further, a plume can directly be treated without
pumping lots of clean water as in case of using pump and treat methods.
227
-------
A layered aquifer enlarges the sphere of influence or the distance between the well and the stagnation point.
This has been found by numerical simulations and by comparison with field measurements of a tracer test.
On the other hand, the positive effect of a layered aquifer is limited when an aquitard is present. Here, several
remediation systems must be installed, one for each aquifer.
Different techniques of soil vapor extraction are compared by measured field data at a remediation site. It
has been clarified by this investigation that a combination of slotted screen and side channel blower can cause
water accumulation in the well casing which may contaminate the groundwater lying underneath the air well.
The latter is avoided using a double-cased screen and a radial blower (ventilator).
For the vapor zone, circulation systems can be used as well, e.g. to realize special treatment technologies.
A special remediation system allows for cleaning the capillary fringe.
ACKNOWLEDGMENTS
The first two authors thank IEG mbH, D-7410 Reutlingen, for financial support of the numerical
investigations. In particular, B. Bernhardt, IEG mbH, D-7410 Reutlingen, inventor and patent holder of the
UVB method; W. Buermann, Institute of Hydromechanics, University of Karlsruhe; W. Kaess, D-7801
Umkirch; and HJ. Lochte, UTB mbH, D-4020 Mettmann, are gratefully acknowledged for many helpful
discussions and contributions to the operation and development of the vacuum vaporizer well.
REFERENCES
1. Herrling, B.; Buermann, W. "A New Method for In-Situ Remediation of Volatile Contaminants in
Groundwater - Numerical Simulation of the Flow Regime." In Computational Methods in Subsurface
Hydrology: Gambolati, G.; Rinaldo, A.; Brebbia, C. A.; Gray, W. G.; and Finder, G. F.; Eds.; Springer:
Berlin, 1990; pp 299-304.
2. Herrling, B.; Buermann, W.; Stamm, J. "In-Situ Remediation of Volatile Contaminants in Groundwater
by a New System of 'Vacuum-Vaporizer-Wells'." In Subsurface Contamination by Immiscible Fluids:
Weyer, K.U., Ed.; A.A. Balkema: Rotterdam, 1991a; [in press].
3. Pedersen, T.A.; Curtis, J.T. "Soil Vapor Extraction Technology, Reference Handbook". Report: EPA/
540/2-91/003; Risk Reduction Engineering Laboratory, Office of Research and Development, U.S. EPA,
Cincinnati, Ohio 45268; 316 pages, 1991.
228
-------
RESULTS OF A YEAR LONG SIMULATION OF THE END OF A SITE
REMEDIATION USING IN SITU VAPOR STRIPPING
ANNN. CLARKE, PHD.
ECKENFELDER INC., Nashville, Tennessee
ROBERT D. MUTCH, JR., P.E., P.HG.
ECKENFELDER INC., Mahwah, New Jersey
DAVID J. WILSON, PH.D.
Vanderbilt University, Nashville, Tennessee
1.0 INTRODUCTION
Vapor extraction, alternatively called in situ vapor stripping (ISVS), soil vapor extraction (SVE), or soil
vacuuming, in the literature, has been used in treatment trains as well as being the single remedial technology
selected. While other technologies are available which target the same compounds, they are often more
expensive, pose increased exposure to workers and/or the public, or have other types of site specific
disadvantages. Given the positive characteristics of vapor stripping, it is not surprising that the technology
has been successfully applied to many sites but with varying knowledge of the scientific and engineering
bases behind it. Since the initial phase of vapor stripping usually performs "like gang busters", we thought
it would be valuable contribution to the field to evaluate what happens at the end of a site cleanup, when the
residual levels are low in the soil and low in the extracted vapor. While much of the behavior could be
predicted once the site specific kinetics were known, it was thought important to monitor behavior at an
actual site. The specific objectives of this test are given in Table 1-1.
In addition, a mathematical model which had already been developed would be tested using the data. Table
1-2 contains the objectives of the mathematical modeling. It was anticipated that from the site and modeling
data general conclusions and recommendations could be made regarding the end of a site cleanup by vapor
stripping.
An industrial site in New Jersey was selected as the location for these efforts. The specific location was the
site of recently demolished manufacturing buildings which had been constructed in the 1950's. This site
offered the opportunity to study removal characteristics of the technology at the lower soil concentrations,
levels more commonly seen at the end of a remediation program. Preliminary soil testing around the area
had indicated that residual volatile organics constituents were present but at relatively low concentrations.
The vapor extraction involved a pilot scale unit which is essentially capable of full-scale operation at a single
well. It was extensively instrumented and the site was extensively monitored, both levels being above those
usually employed during remediation. There was a total of 42 gas monitoring probes installed around the
229
-------
extraction well within the zone of evaluation. Each probe was capable of monitoring soil gas vacuum levels
and allowing the collection of soil gas samples. Thermistors were employed to monitor soil temperature
during the program. Some probes were constructed to monitor a location at three depths.
Table 1-1. Technical Objectives of Pilot Scale Study
Define the three-dimensional flow field established by a soil gas extraction well in an anisotropic,
alluvial sand aquifer;
Assess the extent of organic compounds removal from the test zone especially at low residual soil
concentrations;
Understand the spatial progression of the removal over time within the treatment zone;
Evaluate the residual, non-strippable levels of various organic constituents in the soil;
Evaluate the impact of variations in soil temperature within the test zone;
Evaluate the quantity and quality of the liquid collected by the on-line demister;
Assess the effectiveness of granulated activated carbon in adsorbing volatile species from the
extracted soil gas, especially low concentrations of volatile organic constituents in the vapor stream;
Evaluate the impact of surface soil conditions such as moisture blocking and snow cover on the flow
field; and
Evaluate the extent to which the treatment zone can be expanded by restricting air entry into the soil
around the extraction well by means of geomembranes or other techniques (mathematic model only).
Table 1-2. Objectives of Mathematical Modeling
Determine the importance of radius-of-curvature effects (the Kelvin Equation) on the vapor pressure
of a volatile compound present either near or in aqueous solution in a finely divided or porous
medium;
Estimate the temperature changes which will occur in soils during aeration as the result of
evaporative cooling, as the result of heat transport by the air itself, and as the result of the gas;
Model volatile compound removal by in situ aeration with a vent pipe;
Calculate gas velocity fields in porous soil when the soil surface is covered by an impermeable barrier
in the vicinity of the vent pipe; and
Calculate gas velocity fields in the vicinity of a vent pipe in porous soil when the Darcy's constant
is not isotropic.
These clusters took samples/data from 5,10, and 15 feet below surface level. See Figure 1-1 for the layout
of the extraction well and monitoring probes. Figure 1-2 is a schematic of the pilot scale unit employed.
230
-------
K
TYPICAL INSTRUMENT
CLUSTERING
§©©.© ©O©
CROSS-SECTION A-A
NTS
(VERTICAL EXAGGERATION)
FIGURE 1-1
MONITORING ARRAY
ECKENFELDER
IlNC. Rochester, New York
-------
Vapor Stripping Trailer
Demlster
Carbon
Vapor-Pac
Unit
Extraction
Well 1149
232
#1 Blower
#2 Blower
Exhaust
A
Pitot
Tube
Legend:
Extracted gas flow
Alternate flow for series blower operation
Gas Sample Port locations
Butterfly valves
FIGURE 1-2
PILOT-SCALE IN SITU
VAPOR STRIPPING SYSTEM
FLOW SCHEMATIC
ECKENFELDER
INC.
Rochester, New York
-------
The constituent concentrations in the soil within the study area were determined before beginning and after
completing the 12-month program. (See Figure 1-3.) The spatial progression of volatile and some semi-
volatile organic constituents and their removal from the soil were studied during this period using a photo
ionization detector and chemical specific analysis of the soil gas. The study also addressed the effectiveness
of granular activated carbon treatment in the collection of the vapors at the low concentrations extracted and
the amount of moisture collected within the demister.
2.0 SUMMARY OF THE TEST PROGRAM
The vapor stripping unit employed at the site was capable of extracting and processing up to 400 cfm of vapor
during various weather conditions. Vacuums up to 108 inches of water were capable of being pulled at a 200
cfm flow rate. The unit was equipped with sampling ports at five locations to permit the monitoring of vapor
composition, temperature, pressure, and flow. The system was designed to sequentially access each of the
ports for a predetermined time period. The unit was operated at one location for 227 nonconsecutive, 24 hour
days over a one year period. Down time was limited to routine maintenance, carbon unit breakthrough and
exchange, and planned pulsing experiments to assess nonequilibrium mechanisms, such as desorption and
diffusion.
Removal of Volatile Organic Compounds
The study location was selected in order to study VOC removal at relatively low soil concentr^ations. It was
thought that this scenario simulated the latter phases of a site remediation. During the year-long stufly, several
series of off gas analyses were performed using one-liter Tedlar® bags in which the vapors were collected.
The bags had not previously been used and had been cleaned prior to use. The bags were outfitted with
Teflon(®) and stainless steel fittings. The filled bags were shipped by air in pressurized compartments by
express carrier for overnight (non weekend) delivery. The samples were analyzed the next day for the volatile
organic constituents at the ECKENFELDER INC. laboratory in Nashville, Tennessee.
Soil gas samples were collected at the ports in the mobile unit previously described and at the monitoring
probes. Initial soil gas concentrations were monitored after the installation of probes using an HNu meter.
The readings at the probes were generally consistent with the compound-specific soil concentrations
identified by the laboratory analyses.
The qualitative/quantitative port data were taken to assist in evaluating the behavior and efficiency of the
activated carbon as an off gas treatment option at low concentrations of volatiles in the extracted air stream
as well as to assess the overall performance of the technology. It must be remembered that concentration in
the extracted air does not directly relate to the residual concentration in the soil.
Figure 2-1 shows the behavior of the concentration of VOCs in the extracted soil vapor with time. Table 2-
1 provides a qualitative summary of the volatile organic constituents identified in the soil before and after
treatment as well as the VOCs in the condensate collected in the demister and in the soil gas itself.
Quantitative results are provided later in this paper.
233
-------
LEGEND
SHALLOW BOREHOLE DRILLED
WITH HAND AUGER
DEEP BOREHOLE DRILLED BY
TRUCK-MOUNTED DRILL RIG
No. of shallow boreholes : 21
No. of deep boreholes : 7
D = Depth to groundwater table
234
FIGURE 1-3
INITIAL AND FINAL
SOIL SAMPLING ARRAY
Nashville, Tennessee
Mahwah>Newjersey
Rochester, New York
-------
200
150 -
O!
o
C/5
<
O
n
uu
i-
o
<
DC
H
X
LJU
100 -
100
150
200
DAYS
FIGURE 2-1 BEHAVIOR OF VOC CONCENTRATION IN EXTRACTED SOIL
GAS WITH TIME
235
-------
Soil Vacuum Behavior
Prior to the initiation of the continuous operation of the system, the site was tested to see how rapidly it
responded to cessation of blower operation (reduction to zero vacuum) and then how rapidly the soil
recovered to a steady state vacuum upon reintroduction of blower operation at the same setting. This test was
performed analogous to recovery tests performed in water wells. Figure 2-2 is a graphic representation of
the response of five monitoring probes to system shutdown. Figure 2-3 is a corresponding graphic
representation of the response at the same monitoring probes during the system startup. Each of the five
probes monitored recovered their initial vacuum readings within 11 minutes of restart. The zero vacuum
readings had been reached for all monitoring probes in 10 minutes or less after system shutdown.
The data collected at the most distant monitoring probes help to establish the radius of influence exerted by
the extraction well. This radius was determined by linear extrapolation to zero vacuum under the same
operating conditions. An effective radius of influence of 8D (eight times the depth of the well) was
determined. It was impossible to monitor the vacuum at this distance since much of the area was paved and
had buildings situated upon it.
Vacuum response correlated in a predictable way to weather events. There were uniform increases in the
vacuum levels monitored at all probes during snow cover. The vacuum level responses from the water cover
developed by the continuous irrigation of the hydroseeded plots were less uniform. This could be anticipated
from the nature of the irregular water cover that was generated by these operations, designed to relandscape
the site after the demolition of the buildings.
Table 2-1. Summary of Volatile Constituents Identified on Site
Identified In
Constituent
Soil before
stripping
Soil after
stripping
Soil gas
stream
Condensate
in demister
Trichloroethylene Y Y Y Y
1,2-Dichlorobenzene Y Y Y N
1,3-Dichlorobenzene Y Y N Y
1,4-Dichlorobenzene Y Y N N
1,1,2,2-tetrachloroethane Y Y Y N
Tetrachloroethylene Y N N N
1,1,1-Trichloroethane Y N N Y
Chloroethane Y N N N
Methyl chloride Y N N N
Chlorobenzene N Y N N
236
-------
£
«
CO
-------
CD
CO
CD
O
o
CO
CD
O
DL
Probe 1-1
Probe 9-1
Probe 14-S
Probe 14-1
Probe 14-D
456
Elapsed Time, min.
FIGURE 2-3 RESPONSE OF VACUUM IN SELECT MONITORING PROBES TO SYSTEM
START UP
238
-------
Temperature Behavior
Throughout the study, temperature was monitored in the ambient air, the extracted soil gas, and in the soil
itself. Because of the insulative properties associated with deep soil, "overturn", as is frequently experienced
in deep lakes, rivers, etc. was seen. The surface soil was most sensitive to the ambient temperatures and in
the cooler months, exhibited the lowest temperatures. The temperature of the surficial soils increased until
it exceeded the deeper soil temperatures in the warm months. There was also a general correlation of the
temperature peaks and valleys at the three soil depths with the sharp rises and declines in ambient
temperatures.
As could be anticipated given the heat capacity of soil, the soil gas exhibited only a 10 to 15°C variation over
the year while the ambient conditions ranged over 35°C. The data reflected a relatively smooth transition
over the period of study. (See Figure 2-4.)
3.0 IMPACT OF NON EQUILIBRIUM KINETICS
At best, the non equilibrium considerations e.g., desorption/diffusion, which normally extend the remediation
time, can be capitalized upon to increase activated carbon performance near the end of remediation. At worse,
non equilibrium behavior can greatly extend the time required to reach target VOC levels. This is especially
true if the system is poorly designed. It is also the non equilibrium factors that cause the VOC concentrations
in the vapor not to correlate well to the residual soil VOC concentrations.
A series of three pulsed experiments were performed after the continuous operations had been terminated.
System shutdowns times included: 1 hour, 48 hours, and 7 days. The response displayed in all three tests was
the same. The response was a relatively rapid increase in VOC concentrations subsequent to turning the
system back on followed by a leveling off of the VOC level. For most situations one would normally
anticipate a rapid decrease in the VOC level after the initial use in concentration. After the initial increase,
the rapid decline was not seen in this study area because of the presence of a hot spot located near the
periphery of the zone of influence of the extraction well.
The VOCs desorbed from the hotspot were drawn over this distance to the well. If a hot spot were located
close to the extraction well, spike behavior (the rapid increase followed by rapid decrease in VOC
concentration) would be seen. The data indicate behavior consistent with a site not totally cleaned up. This
was, indeed, confirmed by the existence of a hot spot remote from the extraction well but within the zone
of influence. (Remember that the actual zone of influence was twice what was initially anticipated, 8D vs.
4D.) This is an example of a problem that could arise if a hotspot were not originally detected and
accommodated for in the design (i.e., location of an extraction well close to the hotspot). The presence of
an unanticipated hotspot extends remediation time as well as increases costs.
4.0 PERFORMANCE OF ACTIVATED CARBON ON RELATIVELY LOW
CONCENTRATION AIR STREAMS
The use of granulated activated carbon for the treatment of vapor phase organics is a widely employed
technology. It is known that the carbon system performs more efficiently at higher VOC concentrations. The
presence of water in the extracted gas from an IS VS system further exacerbated the reduced efficiency of
239
-------
to
-p*
o
o
*
01
cr
DC
UJ
Q.
5
UJ
30
25
20
15
10
-10
I I
I i i i
Probe 1S
Probe 11
Probe 1D
External Ambient Temperature
* *
* System Shut Down For Maintenance
I I I I
J I I I
j I
J I I I
25-Aug 04-Oct 13-Nov 23-Dec 01-Feb 13-Mar 22-Apr 01-Jun 11-Jul
1988
1989
DATE
FIGURE 2-4 SOIL AND AMBIENT TEMPERATURES
-------
the carbon at the low VOC concentration. The water vapor successfully competes for sites on the carbon,
thus reducing the efficiency further. Thus, it would seem that acknowledging the already high cost of off
gas treatment for ISVS systems, the cost can be further increased as the cleanup progresses toward
completion. This would be true for other off gas or vapor treatment techniques such as thermal incineration
as the contribution of BTUs by the extracted gas steadily declines with declining VOC concentration.
Over the 227 days of operation, approximately 2,900 cubic feet of vapors were adsorbed. This is 0.053
percent of the total 55 million cubic feet of air that passed through the carbon during the test. Based upon
the results from the Kriging (described below), approximately 209 kilograms of mass were removed from
the zone of influence for an average concentration of 134 mg/L. The removal of 209 kilograms by 5 carbon
units containing 1,800 pounds each of the activated carbon represents an average of 5.1 percent efficiency.
This is a rather lower efficiency that can normally be realized during the early stages of vapor stripping. Other
operating conditions have yielded a percent efficiency in the 10 to 15 percent range, up to as high as 30 percent
in some applications.
Consistent with the fact that the gas initially extracted is usually the most concentrated, a smaller volume
of gas was treated by the first Vapor Pac unit before the permitted breakthrough level (50 ppm VOC) was
reached. The level of efficiency of the carbon for the first flush of VOCs from the soil is approximately one
and a half times the system average of the last four units. This behavior tends to support the use of a pulsed
approach to vapor stripping near the end of a remedial program. By shutting the system down and permitting
a desorption/diffusion controlled build up of the VOCs in the interstices of the soil to develop, a more highly
concentrated flow of air can be flushed from the soil when the ISVS unit is turned back on. This approach
is discussed in more detail in the next section.
There are several alternatives which can be attempted to extend the useful life of carbon by increasing the
efficiency for given site conditions. Efforts which need to be addressed include insulation of piping and
increasing the temperature of the gas stream; shifting location of the carbon beds relative to the blower; and
checking the internal bed design for short circuiting. Of course, the most obvious answer to this question
of increased carbon efficiency is removing more water mass from the vapor stream being treated by the
carbon. One needs to be careful, however, as this could create a disposal problem for the demister condensate
since the condensed water will contain dissolved VOCs and could be listed as a hazardous waste requiring
special handling and disposal. Cooling, of course, is a most reliable means of condensing the water vapor
from the gas stream. Other alternatives applicable to larger, full scale operations include on site steam
regeneration of the carbon. This too, creates a wastewater stream that could require special handling and
disposal.
The HNu is sensitive to water vapor. Therefore, it is recommended that, especially at low concentrations of
extracted gas, a flame ionization detector (which is insensitive to water vapor) be used rather than a
photoionization detector, PID. Another potential problem is that the PID does not "see" methane. The flame
ionization detector, FID, is sensitive to methane. It is not uncommon for areas that have served as landfills
that are now being remediated to contain levels of methane which could prove a potential health and safety
hazard. The FID under these conditions may "peg-out". A combination of both types of monitoring units
may then be warranted.
241
-------
5.0 VOCS REMOVED BY VAPOR STRIPPING
Estimates of the total amount of volatile organic compounds initially in the soil at the site and of the amount
remaining after the vapor stripping are shown in Tables 5-1 and 5-2, respectively. An estimated 200 kilograms
of volatile organics or about 65 percent of the amount originally present had been removed by the time the
year's testing was over. See Table 5-3. The amount of volatile organic compounds removed from the soil
within the measurable zone of influence of the extraction well within the 82 feet radius was estimated by the
following procedures:
First, the measured concentrations in the soil before and after vapor stripping were interpolated using Kriging
to produce estimated concentrations throughout the soil volume of interest. Kriging was used for the
interpolation since this procedure takes into the account the correlation between measured values and
produces both an unbiased estimate of the value at each point and a quantification of the estimation error.
All Kriging calculations were performed using the GEOBASE system of programs developed at the
Department of Mining and Geological Engineering of the University of Arizona.
The first step in Kriging is the estimation of the variogram for the system. The variogram is a measure of the
correlation between measurements at a distances of "h" apart. An isotropic system was assumed here since
there were insufficient data to establish any dependence of the variogram on the direction or distance from
the extraction well.
The variogram was used to Krig both data sets taken before and after treatment. Kriging was performed at
three different depths, 5,10, and 15 feet below the surface corresponding to the measurements obtained from
the clustered monitoring probes. The Kriging at each depth was three dimensional. The extrapolated value
at each point was the weighted average of the nearest measured value. A maximum of eight neighboring
points were used in the interpolation. The maximum distance for which a point was used in the interpolation
was 100 feet.
Table 5-1. Average Concentration and Total Mass of Volatile Organic Compounds in Soil
before Vapor Stripping
Layer
2
3
Depth
(ft)
5.0
10.0
15.0
Average
concentration
(ppm)
26.9
8.6
2.7
Standard
deviation
(ppm)
6.0
2.1
0.5
Total mass
(kg)
241.7
51.5
24.4
Standard
deviation
(kg)
53.8
12.9
4.8
Total soil volume 13.3 2.3 317.7 55.5
242
-------
Table 5-2. Average Concentration and Total Mass of Volatile Organic Compounds in Soils after
Vapor Stripping
Layer Depth
(ft)
5.0
2 10.0
3 15.0
Average
concentration
(ppm)
9.8
2.5
0.8
Standard
deviation
(ppm)
1.6
0.5
0.1
Total mass
(kg)
87.6
15.1
7.1
Standard
deviation
(kg)
14.4
2.7
0.9
Total soil volume
4.6
0.6
109.9
14.6
Table 5-3. Changes in Average Concentration and Total Mass of Volatile Organic Compounds in Soils
as a Result of Vapor Stripping
Layer
2
3
Depth
(ft)
5.0
10.0
15.0
Change in
average
concentration
(ppm)
17.2
6.1
1.9
Change in
total mass
(kg)
154.1
36.4
17.3
Percent loss
63.7
70.7
70.8
Total soil volume
8.7
207.8
65.4
6.0 MATHEMATICAL MODELING OF THE VAPOR STRIPPING
TESTING USING FIELD DATA
The mathematical model used in interpreting the data from the site assumed the system was cylindrically
symmetric about the axis of the vapor stripping well. A three-dimensional model would require computer
capabilities that exceed those of the IBM PC compatible units used in this study. The data that were used from
the site and the facility are provided in the list of model parameters in Table 6-1.
243
-------
Examination of a contour plot of log constituent concentration indicated the presence of hotspot between 10
and 13 meters southwest of the extraction well and a second hotspot having concentrations up to nearly 100
times the maximum concentration. This hotspot was located 23 meters due west of the well. This distance
was near the perimeter of the estimated zone of influence from the well as discussed previously. Preliminary
calculations demonstrated that the smaller hotspot was removed and that the second hotspot controlled the
rate of cleanup. Had the hotspot been located closer to the extraction well, the model indicated that removal
would have been substantially faster as one could anticipate. If one scaled so that the bulk of the constituents
was located in a domain 1 to 9 meters from the extraction well, the model, with the other parameters
remaining the same, showed a 98 percent removal of the constituent in 89 days.
Several runs were made with minor variations in the distribution of the constituents. It was apparent that
moderate uncertainty in the distribution of the constituents did not have a serious effect on the rate of
constituent removal.
The presence of an overlying layer of quite low permeability greatly extended the range of the effective zone
of influence of the vapor stripping well at the study site. Originally, a value of four times the depth of the well,
or approximately 40 feet, was estimated to be the zone of influence. The actual radius, per measurements,
was approximately 80 feet. This was illustrated in a modeling run in which a constant permeability of 1.245
m2/atm • sec was assumed for the domain. Using identical parameters except that the permeability of the
overlay was equal to that of the underlying material, the resultant rate of contaminant removal was far less
with 951 days being required to achieve the 65 percent removal effected in 227 days. The impact of the two
layers is further discussed in the next section.
Table 6-1. Site Parameters Used in ISVS Model
Depth of water table 7 m
Radius of domain 25 m
Thickness of low-permeability upper layer 1 m
Permeability of upper layer 6.5 x 10 3m2/atm sec
Thickness of high-permeability lower layer 6 m
Permeability of lower layer 1.245 m2/atm sec
Well depth 5.5 m
Gas flow rate 3.34 moles/sec
Packed radius of well 0.35m
Wellhead pressure 0.94 atm
Soil temperature 17°C
Soil voids fraction 0.25
Soil volumetric moisture fraction 0.1
Soil density 1.7 gm/cm3
Duration of ISVS operations 227 days
244
-------
7.0 MEASURED VS. PREDICTED IN SITU SOIL VACUUM LEVELS
It was evident from the previous numerical modeling of the IS VS system at this site that the observed vacuum
distribution in the soil could not be simulated assuming isotropic soil conditions. Therefore, initial modeling
efforts focused on determining whether the observed vacuum distribution could be simulated by employing
a vertically anisotropic soil. It was quickly discovered that no amount of vertical anisotropy, including
horizontal and vertical values as high as a thousand times different, could account for the observed soil
vacuum distribution. Table 7-1 lists the parameters used in the soil vacuum model. Subsequent modeling
efforts focused on the possibility that the observed vacuum distribution resulted from a layered heterogeneity
rather than vertical anisotropy. The numerical model was reconfigured to simulate a two-layered geologic
system with the upper layer assumed to be 1 meter in thickness and of a lower permeability than the
underlying layer. The underlying layer was assumed to be 6 meters in thickness and of a uniformly higher
permeability that the upper layer. A series of model runs determined that the best match of the modeling
results to the observed vacuum distribution in the soil was achieved with a permeability of 6.5 x 10-3 m/
atm s for the top layer and 1.2 m/atm s for the underlying layer. A comparison of the results of the numerical
model and the distribution is provided in Figure 7-1.
The calibrated model indicated that the upper soil layer has an air permeability of approximately 1/200 of
the underlying stratum. This was not surprising since 25 percent of the surface area of the study area was
covered with concrete footings upon which the recently demolished buildings had stood and the intervening
surficial soil was predominantly fill material of a somewhat finer texture than the underlying native soils.
It was also note-worthy that the model assumed each of the two layers as individually isotropic. Since air
flow in the upper stratum was essentially vertical, it is the vertical permeability of this unit that was critical.
Variation in the horizontal permeability of the upper stratum had little or no effect on the outcome of the
modeling. Similarly, since air flow was nearly horizontal in the lower stratum, it was the horizontal
permeability of this stratum that was of importance to the modeling effort. Variations in the vertical
permeability of the lower stratum had no perceptible effect on the outcome of the model.
Table 7-1. Parameters Used in Model of Soil Vacuum Level
Depth to well screen = 5.5 meters
Equivalent radius of spherical well intake = 0.35 meters
Soil porosity = 0.25
Soil gas temperature = 17°C
Extraction well pressure atmospheres = 0.94
Flux = 3.33 gram moles/second
245
-------
63
SO
s? Quo
30
40
30 <-
EXTRACTION
WELL
© ©0
1
o ©
-SOIL GAS PROSE
MEASURED IN SITU SOIL
VACUUM IN INCHES OF WATER.
SCREEN SETTING
'MEASURED CONTOURS Of
IN SITU SOIL VACUUM.
PREDICTED CONTOURS OF
IN SITU SOIL VACUUM.
IO J5 0 10 10
r—i 1 . . j
icon HOflUOfrtiL 11,1
FIGURE 7-1
COMPARISON OF MEASURED
AND PREDICTED IN SITU
SOIL VACUUM LEVELS
FPKFNFFI DER Nashville. Tennessee
CJV_.lYli,l>r C,L.LH1.IV Mahwah, New Jersey
INC. Rochester, New York
-------
8.0 SUMMARY AND CONCLUSIONS
The year-long in situ vapor stripping program showed that there are some special problems encountered near
the completion of a remedial program and that these problems should be anticipated as well as possible and
addressed in the beginning of the program in order to meet deadlines, budgets, and target cleanup goals.
Specifically, nonequilibrium conditions and the location of contaminant hot spots with respect to the vacuum
extraction well placement will drive the overall cleanup time. Because of these factors, the nonequilibrium
kinetics and the contaminant distribution, one cannot simply correlate the concentration in the off gas to the
residual constituents in the soil. Adequate detailed information on the spatial distribution of contaminants
and the major features of the permeability of the site must be available for use when pilot scale studies are
designed. If this is not done, the pilot scale testing might result in unnecessarily poor performance and
excessively pessimistic assessment of soil vacuum extraction.
The results of the testing further demonstrate the importance of good site characterization to the successful
remediation. If this had been an actual remediation program, an extraction well would have had to been
placed closer to the major hotspot for more efficient removal of VOCs—even if this were late in the program
after results were interpreted to indicate the presence of the hotspot.
Off gas treatment can be a large part of in situ vapor stripping budgets. As one nears the end of remediation,
the overall efficiency of the treatment is reduced by the lower concentration of volatile compounds in the
extracted air stream. This in turn results in even higher costs per unit weight VOC sorbed.
Another recommendation that can be made as the result of this year long program is the use of technology-
based criteria for developing soil cleanup targets using vapor stripping technology. Efforts should be made
to reduce the impact of nonequilibrium behavior through appropriate design, flow rates, control, and on and
off cycling. The combination of these two factors should lead to a more realistic set of clean up objectives.
247
-------
AIR SPARGING EXTENDING VOLATILIZATION TO
CONTAMINATED AQUIFERS
RICHARD A. BROWN, PH.D. AND RICARDO FRAXEDAS
Groundwater Technology, Inc.
310 Horizon Center Drive
Trenton, NJ 08691
INTRODUCTION
Traditionally, soil and groundwater contamination have been treated by excavation of the contaminated soils
and/or by pumping and treating the contaminated groundwater. Soil excavation is often neither a practical
nor cost effective solution, and groundwater treatment often is still required following soil excavation.
Groundwater pump and treat, while effective in containing contamination migration, is generally an
unacceptably slow remediation process.
If, as an alternative to groundwater extraction, the dissolved/adsorbed contamination can be removed in
place, accelerated remediation of the site, reduced costs, and long term protection of potential down gradient
receptors could potentially be achieved. Soil vapor extraction has proven to be one of the more applicable
technologies for treating contaminated soils. This approach is, however, applicable only to unsaturated soils.
Volatilization can also be accomplished in saturated zones by sparging air under pressure through soils below
the water table. This process removes volatiles form the sorbed and dissolved phases thereby providing
treatment to both the soils and groundwater, in saturated soils.
Approximately seven to ten years ago the use of vapor extraction technology was developed and applied to
the treatment of soils contaminated with VOCs. During this process, contaminants are physically removed
using a form of in-situ air stripping or volatilization. This approach is normally conducted using the
application of vacuum to induce air flow through soils. Because the key to its successful use is inducing air
flow, soil vapor extraction technology has been limited to unsaturated soils. Treatment of soil contamination
below the water table with soil vapor extraction is often unfeasible since it requires extensive dewatering of
the contaminated area to create unsaturated conditions and facilitate induced air flow.
A recent innovation in remedial technology has, however, extended the utility of soil vapor extraction to
water saturated soils. This technology is referred to as air sparging. In this process, air is injected under
pressure below the water table The air bubbles which have formed traverse horizontally and vertically
through the soil column creating transient air filled regimes in the saturated zone. Volatile compounds that
are exposed to this sparged air environment "evaporate" into the gas phase and are carried by the air
movement into the vadose zone where they can be captured by a vent system.
249
-------
Air sparging effectively creates a crude air stripper in the subsurface. The soil acts as the packing. Air is
injected and allowed to flow through the water column over the "packing" (Figure 1). Air bubbles that contact
dissolved/adsorbed phase contaminants in the aquifer cause the VOCs to volatilize. The entrained organics
are thus carried by the air bubbles into the vadose zone where they can be captured by a vapor extraction
system or, where permissible, allowed to escape through the ground surface. As an added bonus, the sparged
air maintains high dissolved oxygen, which enhances natural biodegradation.
Air sparging also creates turbulence and increased mixing in the saturated zone, which increases the contact
between groundwater and soil. This will result in higher concentrations of VOCs in the groundwater which
can be recovered by pumping or stripped by sparging.
Combining air sparging and soil venting is an effective means of treating volatile organic contaminants both
above and below the water table. There are two reasons why a vent system is combined with the sparge
system. First, in sites where volatile contaminants have reached the water table, soil contamination is likely
to occur above, at, and several feet below the water table. Addressing the soil contamination, therefore,
requires application of both traditional soil vapor extraction (venting) and air injection (sparging). This
combined air sparge/vent system (ASV system) is effective in treating volatile organic chemicals distributed
throughout the soil column. A second reason is that the chemicals mobilized by the air sparging system could
discharge, near or at ground surface if not effectively captured in the vadose zone. The vent system is the
mechanism which prevents such a discharge.
AIR SPARGING CASE HISTORY
Air sparging is an effective remedial technology. The following case history illustrates its effectiveness.
The site is the former location of a dry cleaning facility. Soil and groundwater contamination resulted from
leaking underground storage tanks which were located in the north-north western part of the property. The
tanks, which have beenremoved, were used to store dry cleaning solvents. Primary groundwater contaminants
have been identified as perchloroethylene (PCE), trichloroethylene (TCE), dichloroethylene (DCE) and
some total petroleum hydrocarbons (TPH) related to heating oil. PCE is the primary contaminant of concern.
TCE and DCE are present primarily due to biologically mediated reductive dehalogenation.
The subsurface environment at the property generally consists of miscellaneous occurrences of fill material
(coarse sands) sporadically overlying a continuous sheet of naturally occurring Quatenary sediments. Site-
specific geologic conditions are such that a natural barrier (clays of the Potomac Formation) exists which
locally minimizes the potential for vertical downward migration of dissolved-phase total petroleum
hydrocarbon and chlorinated volatile organic compounds (VOCs) present in the shallow water bearing zone
into deeper water bearing units. This is supported by water quality data obtained from the deep wells.
PCE, TCE, DCE and total petroleum hydrocarbon constituents were detected in groundwater samples
collected from shallow monitoring wells at various locations on the property. Depth to water was -15 feet.
The detection of dissolved-phase PCE and total TPH constituents is consistent with the previous use and
storage of PCE product and fuel oil in underground storage tanks at the property. The occurrence of other
dissolved phase chlorinated constituents such as TCE and DCE is likely a result of natural anaerobic
biodegradation of the perchloroethylene product.
250
-------
Based on data obtained from several pilot tests a pattern of additional vent and sparge points was developed
to provide overlapping influence (negative net pressure) and favorable site coverage for the treatment system
(Figure 2). A complete list of treatment points installed at the site is specified below.
7 combination vapor extraction/air sparge points (AS/VP1-AS/VP7); (see Figure 3 for typical
schematic construction diagram).
1 vapor extraction only point (VP1);
7 sparge only points (AS1-AS7).
The venting system used a 15 Hp, Oil Recovery Systems soil vent system having a capacity of 500 CFM at
40 inches of water column vacuum. Influent vacuum/flow rate is controlled with an ambient air intake valve.
A liquid knockout tank, paniculate filter and muffler will be placed on the influent line to eliminate or reduce
water generated during system operation, solids and noise respectively. An effluent muffler was specified
to further reduce noise levels to meet zoning regulations. Two 1800 pound granular activated carbon (G AC)
units were specified to be used in series on the vent effluent to remove contaminants from the vent air prior
to discharge. These units are capable of accepting air flow rates in excess of 500 CFM.
The sparge system used a 20 Hp rotary lobe type blower capable of delivering 270 CFM at 10 psi. As with
the vent system, a paniculate filter is provided on the inlet to protect the moving parts of the unit. System
pressure is controlled with a valve on the ambient air discharge line placed on the pressure side of the sparge
blower. Noise reduction was achieved with mufflers on the inlet, outlet and ambient discharge lines of the
system. The air sparge blower was also furnished with an overpressure relief valve set to open at 15 psi.
Figure 4 shows an isoconcentration map illustrating that dissolved-phase total petroleum hydrocarbon and
total PCE, TCE and DCE constituents were primarily isolated to the shallow saturated Quaternary sediments
with migration of the constituents occurring in a hydraulically downgradient direction from the former
underground storage tank field area(s). This is consistent with the direction of inferred groundwater flow
within the water table aquifer. Contamination on site ranged from 40,000 ppb near the former tank pit to less
than 10,000 ppb total VOC.
In the first phase of remediation, the vent system was initiated. Total flow from the eight vent wells was
approximately 450 CFM at a vacuum of 25-30 inches of water column. After about one month of operation,
the vent system VOC influent concentration had dropped to <10 units. At this point, the sparge system was
started beginning with the outer ring of sparge locations. The VOC levels rose in response to the sparge
system from < 10 OVA units to -60-70 units and remained at an elevated level for several months. This is
an indication of removal of VOCs below the water table.
Sparging was implemented in a two stage process. First, the combined vent/sparge points (ASVP1-7) were
utilized. Second, the centrally located sparge only point (AS-1) was added and finally all other sparge points
were added.
Two sets of groundwater results were taken after 54 and 125 days of sparging operation. As can be seen in
Table 1, there has been substantial and dramatic changes. The net reduction is greater than 98% overall. The
251
-------
125 day groundwater samples were taken one week after the system shut down to allow groundwater to re-
equilibrate.
TABLE 1
Change in VOC Level With Time
(VOC = PCE,TCE,DCE)
Well
MW-1S
MW-1D
MW-2S
MW-2D
MW-3S
MW-3D
MW-4S
MW-5S
MW-6S
MW-7S
Before
Start-up
2,108
14
41,000
BDL
2,161
BDL
4,328
6,940
166
134
54 Days
After Sparge
3.5
1.9
290
BDL
2.2
BDL
444
357
5
31
125 Days
After Sparge
4.9
BDL
897
1.5
1.9
12
240
124
BDL
5
Based on the ground water results, it may be concluded that the ASV system has been effective in
significantly reducing groundwater and soil contamination across the site. The only area showing even a
moderate residual is the vicinity of MW-2 near the source area.
It should also be noted that MW-7 has also shown a decrease in concentration This indicates that VOCs are
not being mobilized by the sparge system. The decrease in MW-7 is also significant in that there is no active
ground water pumping occurring on the site to control off site ground water movement.
Tedlar bag samples were taken of the vent effluent and analyzed in the laboratory. Only PCE and TCE were
found in any of the samples; PCE was the major constituent.
The tedlar bag sample results were also graphed versus time. As was seen with the OVA results, the
concentrations of PCE and TCE dropped during the vent start-up and rose again with the initiation of the
sparge system (Figure 5). These results indicate that the combined system is effective in removing PCE and
TCE from soils and groundwater.
Based on the tedlar bag sample results, a cumulative mass removal was calculated for the site. As shown in
Figure 5, in the first six weeks of operation, approximately 900 pounds of PCE and TCE were removed from
252
-------
the site. These results reflect removal of the bulk of the adsorbed phase unsaturated zone contamination. The
increased VOC levels after sparge start-up indicated that the sparge system had initiated removal of
contaminants contained in the saturated soils and groundwater.
VOC contamination both above and below the water table has been effectively treated by the use of a
combined air sparging - soil venting system. In a relatively short period of time (125 days) groundwater levels
have been reduced by over 98%.
The system has been operated without active groundwater pumping. The mounding created by the air sparge
system was designed to prevent off-site groundwater migration. This system has been effective as there has
been no downgradient increase in VOCs. In fact, VOC levels in the downgradient off-site well have
significantly decreased during the treatment.
QUESTIONS CONCERNING AIR SPARGING
The above case history illustrates the effectiveness of air sparging technology. From appearances air sparging
is a straightforward and effective means of treating volatile organics below the water table.
There are two potential concerns with the use of air sparging. The first is the spread of dissolved
contamination. The second is the acceleration of vapor phase transport and the subsequent accumulation of
vapors in buildings.
Air sparging has been used in Germany for several years. It has however, recently fallen into disfavor. The
primary cause of this disfavor has been the increase of off-site downgradient dissolved contamination
There are three probable causes for an increase of downgradient dissolved levels. The first is geological
conditions. The second is operating conditions. The third is changes in hydrogeology.
Geological conditions can impact air flow. With a sparge system air flow must be both horizontal and
vertical. The vertical travel is important for the ultimate removal of the volatilized contaminant. If the
geology constricts vertical air flow, then sparging can push the dissolved contamination downgradient as
shown in Figure 6. Any permeability differential (i.e. clay barrier) above the zone of air injection may
severely reduce the effectiveness of air sparging. The present or absence of such barriers should be
determined during a pflot test study. Therefore, in low permeable/heterogeneous formations sparging may
require a groundwater recovery system to prevent the spread of dissolved contamination.
A second cause of increased downgradient contaminant migration, is over pressurizing the sparge system.
Ostensibly the minimum sparge pressure is that which is required to overcome the water column (i.e., 1 psi
for every 2.3 feet of hydraulic head). As pressure is increased above this minimum, air is "injected" laterally
into the aquifer. At low sparge pressure there is a balance vertical/horizontal flow as shown in Figure 7. The
injected air "cones" up through the aquifer. At high pressure there is greater horizontal flow than vertical flow
(Figure 7).
As seen in Figure 8, there is an initial linear relationship between the sparge pressure and direction of air
travel. At low sparge pressure (injection pressure equal to hydraulic head) the air travels 1 -2 feet horizontally
253
-------
for every foot of vertical travel. As the sparge pressure increases, the degree of horizontal travel also
increases.
Normally, one would desire this enhanced horizontal travel as the injected air would be able to effectively
remove more contamination. Where pressure causes problems is where the air flow changes from smooth
flow to turbulent flow. Figure 8 shows a point where the increase in sparge pressure does not give a
corresponding increase in horizontal travel. This transition is also observed in venting systems where too
high a vacuum is used causing turbulent flow. Under such sparging conditions a dissolved plume could be
pushed downgradient (Figure 9).
The third potential cause of increased downgradient dissolved contamination is changes in site hydrogeology
due to sparging, specifically water table mounding. Air sparging does cause mounding. Figures 10 and 11
show the impact of sparging on the water table for the case history discussed above. Figure 10 shows the
normal water table with a northwest to southeast gradient Figure 11 shows contours prepared with data
collected after two weeks of sparge system operation. There is an apparent mounding of the water table across
the site, oriented along the northwest southeast axis. The mounding appears greatest downgradient of the
source area
Normally mounding of the water table would accelerate groundwater velocity. However, with sparging the
mounding is caused by the displacement of water with air. Flow may not be accelerated because the net
density of the water column is decreased thus counteracting the mounding. This lowered density is
dramatically seen by taking water table measurements fifteen minutes after the sparge system was shut off.
The water table collapsed as shown in Figure 12 as the air exited the water. This collapse shows the
displacement of water by air during sparging. Because of this density compensation, mounding may not
spread any contamination.
The second "danger" of sparging is accelerated vapor travel. This is of concern where there are receptors.
Since air sparging increases pressure in the vadose zone, any exhausted vapors can be drawn into building
basements. Basements are generally low pressure areas, and this can lead to preferential vapor migration and
accumulation in basements (Figure 13). As a result, in areas with potential vapor receptors, air sparging
should be done with a concurrent vent system. A vent system provides an effective means of capturing
sparged gases.
DESIGN GUIDELINES
Because of the potential dangers for enhanced contaminant transport, proper design is important to effective
sparging. For a sparge system the information that is needed for effective design is:
1) The location of potential groundwater and vapor receptors.
2) The geological conditions of the site - permeability, lithology, heterogeneity.
3) The contaminant mass distribution within the area to be treated - soil and groundwater. This
distribution should be "superimposed" on the lithology of the site.
254
-------
4) The radius of influence of the sparge well(s) at various flow rate/pressure.
The following parameters can be measured during field testing to determine design information.
Pressure vs. distances. This is an indication of radius.
VOC concentrations in groundwater. This is an indication of what is being removed and areas being
impacted, it should be done before, during (with and without the system running) and after test.
CO2 and O2 levels in soil vapor. This is an indication of biological activity. These measurements need
to be taken before, during and after pilot test under static as well as pumping conditions.
Dissolved oxygen levels in water. This is a good indicator of effect -may be slower to see than air flow.
Measurements need to have good base line to determine changes.
Water levels before and during test. Air flow will cause some mounding. This needs to be done before
test to determine background.
As shown in Figure 14, there is a fairly good correlation between parameters measured during sparging. This
allows for cross correlation during design. With this cross correlation it is possible to obtain effective air flow
through the area of contamination.
CONCLUSION
Sparging is an effective technology. It is, however, not without its dangers. Our experience on air sparging
has been favorable to date. This however, has been due in part to choosing very permeable sites and paying
attention to design.
To further expand the utility of sparging, there are a number of questions that need to be addressed. These
questions are:
• What are the limitations to air sparging technology.
• How does air sparging impact the site hydrogeology and contaminant transport.
• What are the most effective means of determining the radius of influence, pressure requirements,
and effectiveness of a sparge system to minimize and detrimental effects.
• With effective design and careful monitoring, air sparging can be an important remedial tool. If
however, it is over simplified it can be ineffectual at best or counter-productive at worse.
255
-------
o
o o
4-
\
°6 •&)
-° i>
-------
EQUIPMENT
COMPOUND ASVPI
Figure!
AIR SfiftRGE/VENT
SYSTEM LAYOUT
NOT TO SCALE
MW-6S
•
« MONITORING WELL
Q SOIL VAPOR PROBE NEST
A AIR SPARGE/VENT POINT COMBINATION
« AIR SPARGE POINT
4- VENT POINT
GROUNDWATER
TECHNOLOGY, INC.
257
-------
TYPICAL
AIR SPARGE/VENT POINT
CONSTRUCTION
(ASVPI-ASVP7)
SPARGE
CONTROL
VALVE
VENT
CONTROL
VALVE
REMOVABLE
PLUGS
CONCRETE PLUG
MANHOLE
COVER
FROM
SPARGE
BLOWER
TO VENT
BLOWER
GROUT
I* BENTONITE SEAL
6 CASING
(VENT)
8* SCREENED
INTERVAL (VENT)
WATER TABLE
APPROX. 13
S-29* CASING (SPARGE)
2" 10 THREADED PVC
SCHEDULE 40 PLUG
I BENTONITE SEAL
ANNULAR SPACE FILLED
WITH 2 MORIE GRAVEL
2* SCREENED INTERVAL (SPARGE)
32' TOTAL DEPTH
2~ ID THREADED PVC
SCHEDULE -40 020 SLOT
WELL SCREEN
2 10 THREADED PVC
SCHEDULE 40 PLUG
Figure 3
AIR SRARGE / VENT POINT CONSTRUCTION
GROUNDWATER
TECHNOLOGY, INC.
258
-------
ir
16t.4<^ — . — ' |
1
126.061
EAST MAIN STREET
(601 WIDE)
LEGEND \
VOC VOLATILE ORGANIC COMPOUND
Figure 4
ISOCONCENTRATION MAP
TOTAL VOC CONCENTRATIONS IN PARTS PER BILLION (ppb)
SHALLOW MONITORING WELLS (PRIOR TO SYSTEM START-UP)
259
GROUNDWATER
TECHNOLOGY, INC
-------
to
8
I200-,
VENT
ONLY « I » SPARGE/VENT
NOTE' BASED ON CONTINUOS OPERATION AT AN
AVG. FLOW RATE OF 450 CFM
SAMPLES WERE ALSO ANALYZED FOR BENZENE,
TOLUENE, ETHYL8ENZENE, XYLENES, MISC. ALIPHATIC
HYDROCARBONS, MISC, AROMATIC HYDROCARBONS AND
t-l,2.DCE. NONE OF THESE WERE DETECTED IN
ANY OF THE TEOLAR BAG SAMPLES.
—\—
180
T
T
T
200 220 240 260 280
DAYS FROM SYSTEM START-UP
Figures
TOTAL REMOVAL OF VOC
rni
GROUNDWATER
TECHNOLOGY, INC
-------
Figune 61
Inhibited Venticai Mignation due to Impenvious Bannien
Impenvious Barnien
O ^.-.-.• ^r-r^^Qi.rrx.• ^r^&j&fa
- ^XXXXX^XXX^XSXXXNXXXXXXXNX>XXX>
v\-..v.. . ,>x^\yo^%y>yvVo^c^vysy-SV^v^c^%y%yvN
VVXVS\X>X^ V^ V ^- v- v v v v- v v v- v- v v v v v v ^- x- ^ i i V^V^X,\S>XVS>XA>SSSV^X^X^XSX^%X>XV\
^\yvyo-\Vv^^x^\>\>\>X>xo^v^X>sv^%>\v%>x>XNx^
XNXsX>X>XxX\X\X\X\XsX>X^X>X>X^X>X>X^^^
-v ,-v -x -x xcv^^^syo^y^,^^^^^^^^^^,^^^^^^^^^ o> -^
^X^x>x\X^XXsX>^V^^ »X-
* v> v> \>
tO^ s\. -v
'0 ^o Q oo o o co
..VrTLW^rv^.^ O OO
Contaminated Soil
Ain / Contaminant Migration
-------
Figune 7; Effect of Sparge Pnessune on Ain Movement
Balanced Pressure
Over Pressurized
0 0
° n °C
°0 0°° O
o °o o°0 oo o
o o 00° o o
o o o o o
o ° o o o „
o o0 00°
O O o °
o oo u
0 o o
000
o°oo °
0
o o
o
o o
O n <3>
o°oJ
°ft
go0 Oo"
°
°o 0°
8
o.
D)
•l—
IS In
o c
LJ
CD
Horizonlal
Air Density
3)
'ro In
o c
r! o>
t Q
(D
>
Horizontal
Air Density
-------
9,0 -I
Smooth Flow
Turbulent Flow
2,0 4,0 6,0 8,0 10,0 12,0 14,0 16,0 18,0 20,0
Figure 8; Ratio of Horizontal Radius vs Sparge Depth
-------
e
S)
Figu
\_/*
Due
9; Accelerated Downgradient
to Turbulent Flow Caused
Over Pressurization
ion
= 000
\<<<<<<<'<<
-------
1
EAST MAIN STREET
(60' WIDE)
Figure 10
SHALLOW WATER TABLE CONTOUR MAP (IN FT.)
265-
Effl
GROUNDWATER
TECHNOLOGY, INC.
-------
APPROXIMATE SCALE
0 75
EAST MAIN STREET
« MONITORING WELL
o PROPOSED MONITORING WELL
O SOIL VAPOR PROBE NEST
* AIR SRARGE/VENT POINT COMBINATION
« AIR SPARGE POINT
+ VENT POINT
(105.51) RELATIVE WATER LEVEL
ELEVATION IN FEET
X*-*—INFERRED GROUNDWATER ELEVATION
CONTOUR IN FEET
Figure 11
WATER TABLE CONTOURS
WITH ASVPI-7 and AS'I OPERATING
FEET
_J
MW-7S
LJLJL
nan
GROUNDWATER
TECHNOLOGY, INC.
266
-------
Ir
J26.OS'
EAST MAIN STREET
(60' WIDE)
Figure 12
COLLAPSE IN WATER TABLE
15 MINUTES AFTER CENSATION OF SPARGING
267
GROUNDWATER
TECHNOLOGY. INC.
-------
to
o\
00
Figune 13: Enhanced Vapon Tnavel due to Pnessune Gnadlent
R
Atm
Pp
P
Atm
P
Sp
00
Building
° 80 § ° Jx
O „ O" „ 13 O^
§o°o
00
005
00
o
*
F =
dB0 dP
n dx
Basement Pressure (low)
Atmospheric Pressure
Sparge Pressure (high)
F = Vapor Flux
d = Vapor
Density
n = Vapor
Viscosity
B0 = Permeability
dP = Pressure
dx Gradient
-------
Figune H: Conrelation Between Sparge Parameters
and Radius of Influence
10 H
C
CD
CO
U)
X
O
0_
o
01
en
a
250 -1
cn
o>
c
T) X
CD Q_
CD Q_ 125
O
I — I
D_
Q)
tn
-I—I
ce:
Q)
s
CD
LL
Q)
-t—
CD
0.5 -
0.4 -
0.3 -
0,2
0.1
Ar
_ x-* ~ •" ~ *" ~
-- ^\
^ ~ — ^
\ \
\ \
\ \
\ \
*
^
1 1 1 1 1 1
5
x
\
s \
\\ /
\
x "x^ /
^ v /
^ Xv
\ / ^
"""•--/ "\
N \
N v
x s
^^~— • ""
1 1 1 1 I 1 1
5
\
V
\
\
\ s
y\ /
/ * \
' \ / \
-/ " \
i v x *•
/ --- \
i i i i i i
ea Area Ar
2 ,3
\
N
\
\
\
\
\
s
s
\
\
\
, \
1
1 1 1 1 1
! 10
X
'X
X
X
X
X
X
^ X
"^ \
^« X
^x
, ,_
1 1 1 1 1
10
I
N
'ixN
X
• N
x v
x
i X
J
i i i r i
ea
1
(•
~ — — — — — — p* l , f-r- ,-,1
bacKground C 1 WJ
1 1 1 1 1 1 1 1 1
15 20
x
i \ Backaround (T~0)
, - i i i i , , od^y'|
15 20
_^ •"*
^ "
__, "
k ^ " ^
1 1 t 1 1 1 I 1 1
5 10 15 20
Distance From Sparge Point; Ft,
269
-------
COMMERCIAL VAPOR TREATMENT PROCESSES
F.AM. BUCK AND EL. SEIDER
King, Buck & Associates, Inc.
2384 San Diego Avenue, Suite 2
San Diego, CA 92110
Vapor extraction, also referred to as soil venting, is a remedial technique that grew rapidly in the early- to
mid-1980s. In simplest application, the suction port of a vacuum blower is connected to one or more
extraction wells installed to penetrate the contaminant plume in the soil, and a mixture of air and the volatile
contaminant is sucked out of the ground and discharged into the air. This simple process is often the most
effective way of reducing soil contamination and protecting groundwater. However, as demonstrated by
several papers at this symposium, the design of a collection system that cleans up the contaminated site in
minimum time and at minimum cost is dependent upon the skills of earth scientists, hydrogeologists, or
others.
Vapor extraction equipment has been developed to handle a wide range of vapor extraction rates, to greater
than 28 m3/min (1000 scfm). Early equipment usually had low capacities of about 0.3 nm3/min (10 scfm),
but lengthy cleanup times inspired the development of equipment capable of higher extraction rates.
However, with the higher extraction rates come potentially higher emission rates of VOCs to the atmosphere.
This led to more concern over air quality effects.
Many agencies responsible for controlling air pollution have mandated treatment of the recovered vapors.
Air pollution control districts in southern California have been in the forefront with strict regulations
controlling the processes of soil venting. The South Coast Air Quality Management District, El Monte,
California, issued Rule 1166, "Volatile Organic Compound Emissions from Decontamination of Soil," in
August 1988. The applicability is to "limit the emissions of VOC" from soil contaminated with VOCs, and
the regulation contains requirements that control VOC emissions during any type of mitigation measure,
including extraction.
EMISSION CONTROL SYSTEMS
Several potential treatment processes are available to meet pollution control standards using vapor
extraction. These methods include the following:
1. Refrigerated condensation to recover VOCs as liquids.
2. Adsorption of VOCs activated charcoal.
3. Burning (oxidation) of VOCs by thermal oxidation, catalytic oxidation, or by two-
staged thermal and catalytic oxidation. Examples of two-stage oxidation are an
271
-------
internal combustion engine with a catalytic converter on the exhaust; or, a system that
uses thermal oxidation when the VOC content of the extracted vapor is high, and a
catalytic oxidizer when the VOC content is moderate to low (Figure 1).
If the VOCs are hydrocarbons with combustion products not containing appreciable amounts of corrosive
or noxious compounds such as HC1, burning of the extracted vapor (by thermal or catalytic oxidation) is
usually the process preferred economically.
A few process characteristics are offered supporting the general conclusion that oxidation processes are the
preferred choice. The product recovered by refrigerated condensation would be gummy, partially oxidized
VOCs that present a disposal problem would be better to burn them in a vapor form without using a costly
refrigeration process as an intermediate step.
As for adsorption on activated charcoal, the process has competitive economics only at very low VOC
concentrations in the extracted vapor (1).
Burning the VOC in an ICE, may seem to be attractive, but problems arise from some basic characteristics
of ICEs. First, the modern ICE is a complex, finely tuned machine that requires highly refined fuels and
lubricants that meet stringent specifications. Purity specifications rule out materials that would form gums
or lacquers in the carburetor, on valves, etc., but the soil vent vapor is just such a reactive, gummy fuel.
Moreover, the combustion process must be completed efficiently in a fraction of a second. Experience has
shown that extracted vapor has large variations in combustion properties. Not only does the VOC content
vary but, as shown by analyses, carbon dioxide content may range to 6 percent or more, and oxygen may
be much lower than the twenty-one percent by volume expected in fresh air. Extracted vapor with such
composition has a significantly different flame propagation speed than normal fuel air mixtures. Operating
conditions of thermal oxidizers have to be modified to burn process gas efficiently if it has high CO2 and
low O2 concentrations. The problems are magnified in an ICE, especially if power output is a consideration.
EMISSION CONTROL BY OXIDATION
VOCs in the extracted air can be oxidized efficiently in properly designed burners. With simple VOCs such
as gasoline components the process is generally called "thermal oxidation", to distinguish it from
"incineration," a term usually applied to the combustion of various noxious materials under carefully
controlled, high temperatures at a destruction efficiency of 99.9% and higher.
This discussion will refer to one type of thermal oxidizer, the pre-mix type (a laboratory Bunsen burner is
a pre-mix burner) because, though simple, the pre-mix burner illustrates several principles of combustion
technology. In a burner of this type, VOCs in the extracted air can be oxidized efficiently if the concentration
of combustibles is above their lower flammability limit (LFL) in air. At VOC concentrations below the LFL,
enrichment fuel (e.g. natural gas or propane) must be added to maintain efficient combustion. There is no
lower limit on VOC concentration below which thermal oxidation will not effectively destroy the VOC; the
requirement is merely to add enough enrichment fuel to maintain a stable, efficiently burning flame.
Experience has shown that a premix of air and propane or natural gas will burn efficiently at a fuel content
20 to 50 percent above the LFL. This permits an estimation of the maximum fuel consumption for a premix-
type thermal oxidizer in an idealized case. For example, the fuel required for an efficient flame with 100
272
-------
would be 2.8 scfm, or 4.6 gal/h. VOCs in the extracted vapor will reduce the fuel requirement below the
"fresh air" case but, as mentioned earlier, other factors such as the concentrations of oxygen, carbon dioxide,
and water vapor will affect the combustion process. It is not always accurate to estimate the supplementary
fuel consumption by considering only the VOC content of the extracted vapor.
At VOC levels at or below about 30 percent of the LFL, oxidation over catalysts can be used for efficient
VOC destruction. There are two boundary limits to the catalytic process: (1) if the VOC concentration is
too high, the heat released during oxidation will cause high-temperature destruction of the catalytic oxidizer,
or (2) if the VOC concentration is too low, not enough heat is released to maintain the catalyst at a temperature
needed for efficient oxidation.
In the first situation, where the exotherm (heat released during oxidation of the VOC) is too large for the
system, it is obvious that dilution air can be admitted to the suction of the vacuum/compressor unit to dilute
the VOC content and lower the exotherm. There is an unwanted result of this dilution, however. The system
capacity is set by one or more limitations on the system, e.g., by the capacity of the vacuum pump, or by
the design space velocity of the process gas across the catalyst. At the maximum flow rate of the process
gas, dilution of the extracted vapor with fresh air (to contain the exotherm) means that the rate of extraction
of vapor from the soil must be reduced. Therefore, this reduction will add to the time and cost required for
site remediation. To avoid reducing the extraction rate of vent gas, it is desirable to oxidize high-VOC-
content vapor In a thermal oxidizer and to use the catalytic oxidizer with vapor of lower VOC content that
has an exotherm compatible with the catalytic system.
The second operating limit on VOC concentration leads naturally to some type of preheater for the process
gas before it enters the catalyst bed(s). Usually, a preheat system is controlled by a process controller reading
the temperature of one or more thermocouples in the catalytic reactor. For efficient oxidation of most VOCs,
the minimum temperature at catalyst entry will be set between 315 and 370°C (600 to 700°F). Energy
conservation in a catalytic oxidizer is standard. Usually a product-to-feed gas heat exchanger recaptures 50
percent or more of the sensible heat of the effluent gas from the catalytic reactor before the effluent is
discharged to the atmosphere.
An isometric block diagram of a two-stage thermal/catalytic oxidation system is shown by Figure 1.
OPERATING DATA
Efficiency of the emission control system is calculated by measuring total hydrocarbons (THCs) in the
effluent from the treatment system and comparing the result to THCs in the extracted vapor. From the point
of view of the emission control system, the latter stream is called the influent. THCs can be determined by
laboratory analyses of gas samples taken at the operating site, or by continuous, on-line analysis of the
influent and effluent. A variety of flame ionization detectors (F.I.D.) are available commercially that have
proven to be satisfactory for measuring THCs.
Typically, when an emission control system has been permitted by a local air quality control agency, a
condition of the permit requires that a formal performance test be made, often by a third party testing
laboratory. The performance test must demonstrate that the equipment is controlling emissions below the
limits specified in the agency's permit.
273
-------
The data reported in Tables 1 and 2 are representative of the emission control efficiencies that are obtained
with thermal and catalytic oxidizers, respectively. The data were from tests by testing laboratories pre-
approved by the South Coast Air Quality Management District (AQMD) in 1987 and 1988.
The declines in the VOC concentrations at three soil venting sites are shown in Figure 2. In all cases, the
vapor extraction rates were about 100 scfm. The so-called "Site A" data were observed by the authors; the
data from the two other curves are from cited references. In our experience these decline curves are typical
of soil venting projects that do not have special features. To summarize, it is typical for VOC concentrations
to decline from about 20,000 ppmv (i.e. above the LFL for gasoline/air mixtures) to less than 1000 ppmv
in 200 days or so.
To estimate the amount of THE potentially emitted to the atmosphere during vapor extraction, refer to Figure
3. In this chart, the conversion from concentration of gasoline vapor to emissions in pounds per day is based
on computed physical properties of a surrogate gasoline. The surrogate gasoline was taken to be one-third
each of isooctane, methylcyclohexane, and toluene (i.e., a mixture of equal volumes of an alkane, a
cycloalkane, and an aromatic). It is assumed to be representative of the weathered gasolines typically
encountered in vapor extraction projects. The surrogate gasoline has a vapor density, under standard
conditions, of 0.25 lb/ft3.
From the lower curve of Figure 2, and the correlation of Figure 3, it is calculated that vapor extraction at
100 scfm would have emitted 15,800 Ib (2400 gal) of gasoline vapors over the 240-day period. Emission
control, at a minimum 95 percent destruction efficiency of THCs, therefore reduced atmospheric emissions
by 15,000 Ib or about 2300 gal. Again, in our opinion, this would be typical.
HALOGENATED VOC
All the preceding discussion has referred to treatment processes for VOC that are hydrocarbons (e.g.
gasoline components) or are oxygenates (e.g. alcohols, ethers). Until recently, there were no simple
acceptable oxidative treatment processes for halogenated VOC (e.g. trichloroethylene). Burning
trichloroethylene (TCE) in a thermal oxidizer would produce two unwanted problems: (15) it would be
expensive to neutralize the product HC1 in the high temperature thermal exhaust gas, and (2) there is great
concern that the interactive oxidation of TCE and hydrocarbons may form toxic PICs (products of
incomplete combustion) such as dioxins. Oxidizing TCE catalytically at lower temperature would solve the
preceding two problems, but until recently a suitable catalyst for a simple fixed bed oxidation process was
not available.
The development of an active, stable catalyst for fixed-bed processing of halogenated VOC has been
described by Lester (1989). In this paper data from the first commercial application of the new catalyst will
be given. The system is called HD CatOx™, a trade acronym for the term halohydrocarbon destruct catalytic
oxidation system.
The prototype HD CatOx system has a nominal capacity of 200 scfm and was initially treating a chlorinated
VOC concentration of 3500 ppm. The vapor extraction system, i.e., the vapor extraction wells, the vapor/
liquid separator, and the vacuum/compressor unit is similar to conventional systems used for hydrocarbon
274
-------
recovery. Electrical resistance heaters heat the process stream up to 750°F before the new "H.D." catalyst
converts the TCE to carbon dioxide, water, and hydrochloric acid. The hot gas from the catalytic reactor is
cooled in a special quench section and passes through a scrubber tower where the hydrochloric acid is
neutralized to sodium chloride. An ancillary refrigeration system maintains the system temperature. The
effluents discharged are carbon dioxide and an aqueous solution of sodium chloride and sodium bicarbonate.
Destruction efficiency of TCE, at the design operating conditions for the unit, was over 95%; a longer term
average destruction efficiency is not available because, as one would expect, the operating conditions of this
prototype have been changed frequently. Figure 4 shows the decline of TCE concentrations over time as
well as the cumulative amount of TCE treated. After 160 operating days, the weight of TCE recovered from
the contaminated site exceeds 20,000 pounds.
REFERENCES
1. Buck, F. A.M., and Hasselmann, D.E.M., "Control of Air Emissions from Soil Venting Systems,"
Paper 94a, AIChE National Meeting, April 2,1984.
2. Report by W.W. Irwin, Inc., to California Dept. Of Health Services, Public Hearing on Amendment to
Title 22, Los Angeles, CA, October 30.1989.
3. Fall, E.W. and Pickens, W.E., "In situ Hydrocarbon Extraction," Focus Conference on Southwest
Groundwater Issues, Albuquerque, NM, March 1988.
4. Trowbridge, B.E., and Malot J.J., "Soil Remediation and Free Product Removal Using In-Situ Vacuum
Extraction with Catalytic Oxidation," NWWA Outdoor Action Conference, May 1990.
5. Lester, George R., "Catalytic Destruction of Hazardous Halogenated Organic Chemicals," Air & Waste
Management Association, Anaheim, CA, June 25,1989.
275
-------
Table 1. Performance of a Thermal Oxidizer, during Soil Venting of a Gasoline-Contaminated Site.
(W.W. Irwin, Inc., 1989)
Influent Effluent
Flow rate, air+VOC, scfm 71 170
O2,% 7.2 7.6
CO2, % 7.5 8.8
NOX, ppm - 30
CO, ppm 3.0 2.7
Hydrocarbons, NMHC as Cl, ppm 92,410 3
NMHC, Ib/hr 16.6 0013
Benzene, ppb 82,100 ND (<10)
Benzene, Ib/hr 0.72 <.0003
Hydrocarbon destruction
efficiency, % 99.9+
Table 2. Performance of a Catalytic Oxidizer, during Soil Venting of a Gasoline-Contaminated Site.
(W.W. Irwin, Inc., 1989)
CatOx Inlet CatOx Outlet
Flow rate, dscfm 72
NOX, ppm <15 <15
CO, ppm <15 <10
Hydrocarbons:
CH4, ppm - - 30
Nonmethane, ppm as Cl 14,400 573
Hydrocarbon destruction
efficiency, % 95.8
276
-------
ciwct
»«« siciiu
/*, anunai ww
V«U KJOKK
FIGURE 1. TWO-STAGE THERMAL /CATALYTIC OXIDATION PROCESS
277
-------
FIGURE 2. Soil Venting VOC Decline Curves
100
00
10
1000
ppmv
THC
0.1
10 100
Operating Days
1000
• Site A
4 Trowbridge
and Malot (1990)
A Fall and Pickens
(1988)
-------
0 1000
X
K
s
Q 100
Q:
S
CO
1 10
O
CL
,
i
10
FIGURE 3
-
-
'•^s^
P**"
\^
^
^
^
^
^
^-
^
s*
^
*r
^
s*
s
f*
t*
•**
'
-'
"'
^'
::^
'•^
^
^
s"
^
^
S"
s"
s
S1
S*,
S*
s*
s*
s
S
**
'
^
S*
S"
^ s
*
.s
^
^^
^
^
s*
^
^
s*
-
^
^*
S
'*
''
•
0 1COO JOOOO 1000OO
PPMV THC IN VAPOR
POTENTIAL AIR EM/SS/OMS DURING SO/L VENTING
279
-------
oo
O
FIGURE 4. Soil Vapor Extraction System, SC AQMD Site
TCE Decline Rate
IQOOppmvTCE 1000 Ibs TCE extracted
4 , , r- -, , , r_ , ^ , = , 50
0.
37.5
- 25
12.5
50 100 150 200 250 300 350
Operating Days @ 200 SCFM
400 450
500
-------
CATALYTIC DESTRUCTION OF HAZARDOUS HALOGENATED
ORGANIC CHEMICALS
GEORGE R.LESTER
Allied Signal, Inc.
Research and Technology
Des Plains, Illinois
ABSTRACT
A family of catalysts has been developed which appear to be clearly superior to those previously described
for the destruction of highly toxic or hazardous volatile halogenated compounds including chlorohydrocarbons
such as polychlorinated aromatic hydrocarbons and C; and C2 chlorohydrocarbons and
chlorofluorohydrocarbons. Applications for these catalysts include the destruction of hazardous chlorinated
and/or fluorinated hydrocarbons in vent gases in the chemical process industry, and the purification of the
discharge air from ground water and soil stripping processes. The new catalysts have demonstrated stable
operation for over 1600 hours at greater than 99 percent destruction of carbon tetrachloride (CC14) in moist
air at 375 degrees Celsius and 15,000 hr1 GHSV(STP). The effectiveness of these catalysts for destroying
chloro- and fluoro-aromatics and other fluorohydrocarbons will also be described. It will be shown that the
new catalysts are capable of operating at lower temperatures and/or shorter residence times than previously
known catalysts. Results will be shown for the catalysts in the form of monolithic honeycombs and extruded
pellets.
281
-------
BIOVENTING FOR IN SITU REMEDIATION OF PETROLEUM
HYDROCARBONS
ROBERT E. HINCHEE
Battelle Memorial Institute
Columbus, Ohio 43201
ROSS N. MILLER
Major, USAF
HSD/YAQE
Brooks Air Force Base, Texas 78235
Biodegradation of hydrocarbons in unsaturated zones through forced aeration has been observed at several
field sites. Data are presented from two sites, one located at Hill Air Force Base in Utah and one at Tyndall
Air Force Base in Florida where field demonstrations of this technology are ongoing.
INTRODUCTION
As a result of regulations requiring investigation of underground storage tanks as well as other surface and
subsurface spills, literally thousands of sites have been identified as contaminated with petroleum
hydrocarbons. To date, much attention has been given to pump-and-treat remedial technologies, but this
technique leaves a substantial fuel residue in the capillary fringe or vadose zone that continues to contaminate
groundwater. Methods of uniformly degrading fuel hydrocarbons in situ, without excessive groundwater
pumping or toxic releases to the atmosphere, need to be developed. This paper focuses on one such emerging
technology.
Petroleum distillate fuel hydrocarbons are generally biodegradable if naturally occurring microorganisms
are provided an adequate supply of oxygen and basic nutrients (1). Natural biodegradation does occur and
at many sites may eventually mineralize most fuel contamination. However, the process is dependent upon
natural oxygen diffusion rates (11) and as a result is frequently too slow to prevent the spread of
contamination. Such sites may require remediation of the contaminant source to protect sensitive aquifers.
At these sites, an acceleration or enhancement of the natural biodegradation process may prove the most
effective remediation.
Important in any in situ remediation is an understanding of the distribution of contaminants. Much of the
residue of hydrocarbons at a fuel contaminated site is found in the unsaturated zone soils, in the capillary
fringe, and immediately below the water table. Typically, seasonal water table fluctuations spread residues
in the area immediately above and below the water table. Any successful bioremediation effort must treat
these areas.
CONVENTIONAL ENHANCED BIODEGRADATION
Over the past two decades the practice of enhanced biodegradation has increased, particularly
for treating soluble fuel components in groundwater (8). Less emphasis has been given to enhancing
283
-------
biodegradation in the unsaturated zone. The current conventional enhanced bioreclamation
process uses water to carry the oxygen or an alternative electron acceptor to the contamination,
whether it occurs in the groundwater or unsaturated zone.
A recent field experiment at a jet fuel contaminated site using infiltration galleries and spray
irrigation to introduce oxygen (as hydrogen peroxide), nitrogen, and phosphorus to unsaturated,
sandy soils was unsuccessful because of rapid hydrogen peroxide decomposition and resulting
poor oxygen distribution (6). A study being conducted by the U.S. Environmental Protection
Agency and the U.S. Coast Guard at Traverse City, Michigan, uses deep well injection to raise the
water table in order to supply oxygen-enriched water to the contaminated soils. Pure oxygen and
hydrogen peroxide have been used as oxygen sources, and recently nitrate has been added as an
alternative to oxygen. Although results indicate better hydrogen peroxide stability than achieved
by Hinchee et al. (6), it was concluded that much of the hydrogen peroxide decomposed rapidly
and was lost as bubbles (7). Some degradation of aromatic hydrocarbons appears to have
occurred; however, no change in total hydrocarbon contamination levels was detected in the
soils (14).
In most cases where water is used as the oxygen carrier, oxygen is the limiting factor for bio-
degradation. If pure oxygen is utilized and 40 mg/1 of dissolved oxygen is achieved, approximately
80,000 kg of water must be delivered to the formation to degrade a single kg of hydrocarbon. If
500 mg/1 of hydrogen peroxide is successfully delivered, then approximately 13,000 kg of water are
necessary. As a result, even if hydrogen peroxide can be successfully used, substantial volumes of
water must be pumped through the contaminated formation to deliver sufficient oxygen.
ENHANCED BIODEGRADATION THROUGH VACUUM EXTRACTION
A system engineered to increase the microbial biodegradation of fuel hydrocarbons in the vadose
zone using forced air as the oxygen source is a cost-effective alternative to conventional systems.
This process stimulates soil-indigenous microorganisms to aerobically metabolize fuel hydro-
carbons in unsaturated soils. Depending on air flow rates, volatile compounds may be simul-
taneously removed from contaminated soils.
By using air as an oxygen source, the minimum ratio (based on stoichiometry) of air to hydro-
carbon on a mass basis is approximately 13 to 1. This compares with over 10,000 to 1, water to
hydrocarbon, for a conventional waterborne enhanced bioreclamation process. An additional
advantage of using an airborne process is that gases have greater diffusivity than liquids. At many
sites, geological heterogeneities present an added problem with a waterborne oxygen source
because fluid pumped through the formation is channelled into the more permeable pathways.
For example, in an alluvial soil with interbedded sand and clay, virtually all of the fluid flow will
take place in the sand. As a result, oxygen must be delivered to the less permeable clay lenses
through diffusion. In a gaseous system this diffusion can be expected to take place at a rate
several orders of magnitude greater than in a liquid system. Although it is not realistic to expect
diffusion to aid significantly in water-based bioreclamation, in an air-based application, diffusion
may be a significant mechanism for oxygen delivery to less permeable zones.
The first documented evidence of unsaturated zone biodegradation resulting from forced aeration
was reported by the Texas Research Institute, Inc., in a study for the American Petroleum
284
-------
Institute. A large-scale model experiment was conducted to test the effectiveness of a surfactant
treatment to enhance recovery of spilled gasoline. The experiment accounted for only 30 liters of
the 250 liters originally spilled and raised questions about the fate of the gasoline. A subsequent
column study was conducted to determine a diffusion coefficient for soil venting. This column
study evolved into a biodegradation study in which it was concluded that as much as 38 percent of
the fuel hydrocarbon was biologically mineralized. Researchers concluded that venting would not
only remove gasoline by physical means, but also could enhance microbial activity (12,13).
Wilson and Ward (15) suggested that using air as a carrier for oxygen could be 1,000 times more
efficient than transferring it to water, especially in deep, hard-to-flood unsaturated zones. They
made the connection between soil venting and biodegradation by observing that "soil venting uses
the same principle to remove volatile components of the hydrocarbon." In a general overview of
the soil venting process, Bennedsen (2) concluded that soil venting provides large quantities of
oxygen to the vadose zone, possibly stimulating aerobic degradation. He suggested that water and
nutrients would also be required for significant degradation and encouraged additional
investigation into this area.
Biodegradation enhanced by soil venting has been observed at several field sites. Investigators at
a soil venting site for remediation of gasoline contaminated soil claim significant biodegradation
as measured by a temperature rise when air was supplied. Investigators pulsed pumped air
through a pile of excavated soil and observed a consistent temperature rise that they attributed to
biodegradation. They claimed that the pile was cleaned up during the summer primarily by bio-
degradation (3). However, they did not control for natural volatilization from the aboveground
pile, and not enough data were published to review the biodegradation claim critically.
Researchers at Traverse City, Michigan, measured toluene concentration in vadose zone soil gas
as an indicator of fuel contamination in the vadose zone. They assumed absence of advection and
attributed toluene loss to biodegradation. The investigators concluded that, because toluene
concentrations decayed near the oxygenated ground surface, soil venting is an attractive
remediation alternative for biodegrading light volatile hydrocarbon spills (11).
Ely and Heffner (5) of the Chevron Research Company patented a process for the in situ bio-
degradation of spilled hydrocarbons using soil venting. Experimental design and data are not
provided, but findings are presented graphically. At a gasoline and diesel oil contaminated site,
slightly higher removal through biodegradation than through evaporation was observed. At a
gasoline contaminated site, results indicated that about 2/3 of the hydrocarbon removal was by
volatilization and 1/3 by biodegradation. At a site containing only fuel oils, approximately
75 liters/well/day were biodegraded, while vapor pressures were too low for removal by volatiliza-
tion. Ely and Heffner (5) claimed that the process is more advantageous than strict soil venting
because removal is not dependent only on vapor pressure. In the examples stated in the patent,
CO2 was maintained between 6.8 percent and 11 percent and O2 between 2.3 percent and 11 per-
cent in vented air. The patent suggests that the addition of water and nutrients may not be
acceptable because of flushing to the water table, but nutrient addition is claimed as part of the
patent. The patent recommends flow rates between 50 and 420 m3/min per well and states that air
flows higher than those required for volatilization may be optimum for biodegradation.
285
-------
APPLICATIONS
The use of an air-based oxygen supply for enhancing biodegradation relies on air flow through
hydrocarbon-contaminated soils at rates and configurations that will ensure adequate oxygenation
of aerobic biodegradation and minimize or eliminate the production of a hydrocarbon-
contaminated offgas. The addition of nutrients and moisture may be desirable to increase
biodegradation rates. Figures 1 and 2 illustrate possible air injection/withdrawal configurations.
Dewatering is illustrated in Figure 1, but this may not always be necessary depending upon the
distribution of contaminants relative to the water table. However, it is required at many fuel
hydrocarbon-contaminated sites. A key feature not illustrated is the narrowly screened soil gas
monitoring points that sample only a short vertical section of the soil. These points are required
to determine local oxygen concentrations. Measurements of oxygen levels in the vent are not
representative of local conditions. Nutrient and moisture addition typically may take any of a
variety of configurations.
A conventional soil venting installation where air is drawn from a vent well in the area of greatest
contamination is possible. The advantage of this configuration is that it generally requires the
least air pumping; the disadvantage is that hydrocarbon offgas concentration is probably maxi-
mized and all of the capillary fringe contamination may not be treated. Figure 1 involves air
injection only. This is the lowest cost configuration; however, monitoring must assure that surface
emissions do not exceed acceptance levels, and monitoring and/or protection of subsurface struc-
tures may be required. Figure 2 illustrates a configuration in which air is injected into the
contaminated zone and withdrawn from clean soils. This configuration allows the more volatile
hydrocarbons to degrade prior to being withdrawn and thereby eliminates contaminated offgases.
The optimal configuration for any given site will, of course, depend upon site-specific conditions
and remedial objectives.
The significant features of this technology include the following:
• Optimizing air flow to reduce volatilization, while maintaining aerobic conditions for
biodegradation
• Monitoring local soil gas conditions to assure aerobic conditions, not just monitoring
vent gas composition
• Adding moisture and nutrients as required to increase biodegradation rates
• Manipulating the water table as required for air/contaminant contact.
HILL AIR FORCE BASE SITE
A spill of approximately 100,000 liters of JP-4 jet fuel occurred when an automatic overflow device
failed at Hill Air Force Base near Ogden, Utah. The contamination is primarily found in the
upper 20 meters of a delta outwash of the Weber River. This surficial formation extends from the
surface to a depth of approximately 20 meters and is composed of mixed sand and gravel with
occasional clay stingers. Depth to regional groundwater is approximately 200 meters, however,
water may occasionally be found in discontinuous perched zones. Soil moisture averaged less
than 6 percent in the contaminated soils.
286
-------
FIGURE L CONCEPTUAL LAYOUT OF LOW INTENSITY
BIORECLAMATION TECHNOLOGY
Low rate
air injection
Air Injection Designed to
Satisfy Biological Demand
Biodegradation
of vapors
Monitoring of Soil Gas
to assure vapor biodegradation
FIGURE 2. POTENTIAL CONFIGURATION FOR ENHANCED BIORECLAMATION
THROUGH SOIL VENTING (AIR WITHDRAWN FROM CLEAN SOIL)
Air Injection
(Optional)
Soil Gas Monitoring
Contaminated
Soils
287
-------
Soil samples that were collected indicated JP-4 concentrations up to 20,000 mg/kg, with an average
concentration of approximately 400 mg/kg (10). Contaminants were unevenly distributed to depths of 20
meters. Vent wells were drilled to approximately 20 meters below the ground surface and screened from 3
to 18 meters below the surface. A background vent was installed in an uncontaminated location in the same
geological formation approximately 200 meters north of the site.
Venting was initiated in December 1988 at a rate of approximately 45 mVhr. The offgas was treated by
catalytic incineration, and it was initially necessary to dilute the highly concentrated gas to remain below
explosive limits and within the incinerator's hydrocarbon operating limits. The venting rate was gradually
increased to approximately 2,500m3/hr as hydrocarbon concentration levels dropped (10). During the period
between December 1988 and November 1990, more than 10,000,000 m3 of soil gas were extracted from the
site. In November 1989, ventilation rates were reduced to approximately 500 to 1,000 m3/hr to provide
aeration for bioremediation but to reduce volatilization. This resulted in removal of the catalytic incinerator,
saving approximately $6,000 per month. Oxygen and hydrocarbon concentrations were measured in the
offgas during extraction. To quantify the extent of biodegradation at the site, the oxygen was converted to
an equivalent basis. This was based upon the stoichiometric oxygen requirement for hexane mineralization.
JP-4 was determined based on direct readings of a total hydrocarbon analyzer calibrated to hexane.
Based upon these calculations, the mass of JP-4 as carbon removed and degraded was approximately 53,000
kg volatilized and 42,000 kg biodegraded. Figures 3 and 4 illustrate these results. Further details of this study
are reported in Dupont et al. (4).
TYNDALL AIR FORCE BASE SITE
As a follow-up to the Hill Air Force Base (AFB) research, a more controlled study was designed. The
experimental area at Tyndall AFB was located at a site where past JP-4 storage had resulted in contaminated
soils. The nature and volume of fuel spilled or leaked is unknown. Initial site characterization indicated soil
hydrocarbon levels up to 20,000 mg/kg. The site soils are a fine to medium grained quartz sand and depth
to groundwater is 0.5 to 1.0 meters. Four test cells were constructed to allow control of gas flow, water flow,
and nutrient addition. Two of the test cells were installed in the hydrocarbon contaminated zone and two in
uncontaminated soils. The contaminated area was dewatered and hydraulic control was maintained to keep
the depth to water at approximately 1.6 meters. This exposed more of the contaminated soil to aeration.
During normal operation, air flow rates were maintained at approximately 1 air-filled void volume exchange
per day. Biodegradation and volatilization rates were much higher than those observed at Hill AFB; this was
likely due to higher average levels of contamination, warmer temperatures, and moisture conditions. As
illustrated in Figure 5, at 120 days of aeration in one treatment cell approximately 50 kg of hydrocarbons had
volatilized and 50 kg biodegraded. This represents an average hydrocarbon reduction of approximately
3,000 mg/kg in the treated soils. Biodegradation rates measured here appeared to be initially high and then
declined substantially. The causes of this are unclear; however, they did not appear to be related to
hydrocarbon disappearance-biodegradation rates dropped substantially before 20 percent of the hydrocarbon
was removed. This study concluded that at this site nutrient and moisture addition did not result in higher
rates of biodegradation. More complete results are found in Miller et al. (9).
288
-------
120
D
88
FIGURE 3. CUMULATIVE HYDROCARBON REMOVAL
FROM THE HILL AFB SOIL VENTING SITE
JFMAMJJASOND
1989
Date
JFMAMJJASON
1990
FIGURE 4. RESULTS OF SOIL ANALYSIS AT HILL AFB BEFORE AND AFTER VENTING
(EACH BAR IS THE AVERAGE OF 14 OR MORE SAMPLES)
Depth
(feet)
20
40
^MM/^MMMMMMMMM^MM^^
<"""•'"•'"•'•'"•'"•"•'""•'"••'" 1 | 362
!%^M^^ - |44y
mSS/ /////////V7/W//7/1 39 970
^^^^^^^ I
mff^//mfm^m//m/mwm^m//mm, ,,,,,,,,,,,,,,,, r,\ \
?W///^/////tW/W////W/////////////^^^ — — 1 -„_
— , ^ 1 37 1
""^ .,^,J^,^., ,»„,,, ,,57 75/1
'MMf/'r"fffffrr'("ffr^^^ ' -,
?m%=m^%^%?M^^^^^^^ 1 47O
'^mmmMmw//////////////A 34
I — — . I 422
tmq<5 61
' I I
Depth
(meters)
5
10
15
20 100
Hydrocarbon Concentration (mg/kg)
EH Before HH Intermediate ^ After
1000
289
-------
FIGURE 5. CUMULATIVE HYDROCARBON REMOVAL IN TREATMENT
PLOT VI AT TYNDALL AFB
50 -
40
K
n
20 -
10 -
Volatilized
N
Biodegraded
n
f
/ ff
D
20
40 60
Time (days)
i 1—
80
n 1—
100
120
RECOMMENDATIONS
Using air as the oxygen source is a cost-effective way to increase the microbial degradation of fuel
hydrocarbons in unsaturated soil. Soil venting alone, with no nutrient or moisture addition,
typically results in some stimulation of in situ biodegradation. The following recommendations
are made for documenting biodegradation when conducting conventional soil venting of fuel
hydrocarbon contaminated soils:
1. Prior to venting, determine soil gas hydrocarbon, CO2, and O2 profiles.
2. Measure hydrocarbon, CO2, and O2 in the offgas. This information can be used to
document biodegradation and may help determine the end point for venting. A mixed
hydrocarbon fuel such as JP-4 has a fraction too heavy to volatilize, and it is possible
that biodegradation may continue after the light end has volatilized. This information
is essential to complete a mass balance of spilled hydrocarbons and to estimate the
extent of cleanup.
3. Develop an estimate of noncontaminant respiration. This may be accomplished either
through background measurements of CO2 and O2 in an uncontaminated location or
by means of carbon isotopic analysis.
290
-------
REFERENCES
1. Atlas, R.M. (1986). "Microbial Degradation of Petroleum Hydrocarbons: An Environmental
Perspective," Microbiol. Rev. 45,180-209.
2. Bennedsen, M.B., Scott, J.P., andHartley, J.D. (1987). "Use of Vapor Extraction Systems for In Situ
Removal of Volatile Organic Compounds From Soil," Proceedings of National Conference on
Hazardous Wastes and Hazardous Materials, Washington, DC, 92-95.
3. Conner, J.R. (1988). "Case Study of Soil Venting," Poll. Eng. 7, 74-78.
4. Dupont, R.R., Doucette, W., and Hinchee, R.E. "Assessment of In Situ Bioremediation Potential and
the Application of Bioventing at a Fuel-Contaminated Site," In Situ and Qn-Site Bioreclamation.
R.E. Hinchee and R.F. Olfenbuttel, Eds., Vol. 1, Stoneham, MA. Butterworth Publishers.
5. Ely,D.L.,andHeffner, D.A. (1988)."ProcessforIn-SituBiodegradationofHydrocarbonContaniinated
Soil," U.S. Patent Number 4,765,902.
6. Hinchee, R.E., Downey, D.C., Slaughter, J.K., and Westray, M. (1989). "Enhanced Biorestoration
of Jet Fuels; A Full Scale Test at Eglin Air Force Base, Florida," Air Force Engineering and Services
Center Report ESI/TR/88-78.
7. Huling, S.G., Bledsoe, B.E., and White, M. V. (1990). "Enhanced Biodegradation Utilizing Hydrogen
Peroxide as a Supplemental Source of Oxygen: A Laboratory and Field Study." EPA/600-290-006,
48pp.
8. Lee, M.D., et al. (1988). "Biorestoration of Aquifers Contaminated With Organic Compounds," CRC
Critical Reviews in Env. Control, Vol. 18, 29-89.
9. Miller, R.N., Hinchee, R.E., and Vogel, C. "A Field-Scale Investigation of Petroleum Hydrocarbon
Biodegradation in the Vadose Zone Enhanced by Soil Venting at Tyndall AFB, Florida." In Situ and
On-S ite Bioreclamation. R.E. Hinchee and R.R Olfenbuttel. eds.,Vol 1., Stoneham, MA. Butterworth
Publishers.
10. DePaoli, D.W. , Herbes, S.E., Wilson, J.H., Solomon, O.K., Jennings, H.L., Hylton, T.D., and
Nyquist, J.E. "Field Demonstration of In Situ Soil Venting at Hill AFB JP-4 Jet Fuel Spill Site," U.S.
Air Force Report ESLTR 90-21, Vol. 3, (in press).
11. Ostendorf, D.W. and Kambell, D.H. (1989). "Vertical Profiles and Near Surface Traps for Field
Measurement of Volatile Pollution in the Subsurface Environment," Proceedings of NWWA
Conference on New Techniques for Quantifying the Physical and Chemical Properties of
Heterogeneous Aquifers, Dallas, TX.
12. Texas Research Institute (1980). "Laboratory Scale Gasoline Spill and Venting Experiment,"
American Petroleum Institute, Interim Report No. 7743-5:JST.
291
-------
13. Texas Research Institute (1984). "Forced Venting to Remove Gasoline Vapor from a
Large-Scale Model Aquifer," American Petroleum Institute, Final Report No. 82101-F:TAV.
14. Ward, C.H. (1988). "A Quantitative Demonstration of the Raymond Process for In-Situ
Biorestoration of Contaminated Aquifers," Proceedings of NWWA/API Conference on
Petroleum Hydrocarbons and Organic Chemicals in Groundwater, 723-746.
15. Wilson, J.T. and Ward, C.H. (1986). "Opportunities for Bioremediation of Aquifers
Contaminated with Petroleum Hydrocarbons," J. Ind. Microbiol., 27,109-116.
292
-------
A FIELD SCALE INVESTIGATION OF SOIL VENTING ENHANCED
PETROLEUM HYDROCARBON
BIODEGRADATION IN THE VADOSE-ZONE AT TYNDALL AFB, FLORIDA
MAJOR ROSS N. MILLER, PHD, PE, CIH
U. S. Air Force, HSD/YAQE
Brooks Air Force Base, Texas
ROBERT E. HINCHEE, PH.D, PE
Battelle Memorial Institute
Columbus, Ohio
CAPTAIN CATHERINE M. VOGEL
U. S. Air Force, AFESC/RDVW
Tyndall Air Force Base, Florida
ABSTRACT
Soil venting is effective for the physical removal of volatile hydrocarbons from unsaturated soils, and is also
effective as a source of oxygen for biological degradation of the volatile and non-volatile fractions of
hydrocarbons in contaminated soil. Treatment of soil venting off-gas is expensive, constituting a minimum
of 50% ,of soil venting remediation costs. In this research, methods for enhancing biodegradation through
soil venting were investigated, with the goal of eliminating the need for expensive off-gas treatment.
A seven-month field investigation was conducted at Tyndall Air Force Base (AFB), Florida, where past jet
fuel storage had resulted in contamination of a sandy soil. The contaminated area was dewatered to maintain
approximately 1.6 meters of unsaturated soil. Soil hydrocarbon concentrations ranged from 30 to 23,000 mg/
kg. Contaminated and uncontaminated test plots were vented for 188 days. Venting was interrupted five
times during operation to allow for measurement of biological activity (C02 production and O2 consumption)
under varying moisture and nutrient conditions.
Moisture addition had no significant effect on soil moisture content or biodegradation rate. Soil moisture
content ranged from 6.5 to 9.8%, by weight, throughout the field test. Nutrient addition was also shown to
have no statistically significant effect on biodegradation rate. Initial soil sampling results indicated that
naturally occurring nutrients were adequate for the amount of biodegradation observed. Acetylene reduction
studies, conducted in the laboratory, indicated a biological nitrogen fixation potential capable of fixing the
organic nitrogen, observed in initial soil samples, in five to eight years under anaerobic conditions.
Biodegradation rate constants were shown to be effected by soil temperature and followed predicted values
based on the van't Hoff-Arrhenius Equation.
293
-------
In one treatment cell, approximately 26 kg of hydrocarbons volatilized and 32 kg biodegraded over the
seven-month field test. Although this equates to 55% removal attributed to biodegradation, a series of flow
rate tests showed that biodegradation could be increased to 85% by managing air flow rate. Off-gas from one
treatment cell was injected into clean soil to assess the potential for complete biological remediation. Based
on biodegradation rate data collected at this field site, a soil volume ratio of approximately 4 to 1,
uncontaminated to contaminated soil, would have been required to completely biodegrade the off-gas from
the contaminated soil.
This research indicates that proper ratios of uncontaminated to contaminated soil and air flow management
are important factors in influencing total biodegradation of jet fuel, substantially reducing remediation costs
associated with treatment of soil venting off-gas.
MATERIALS AND METHODS
Site Description
An in situ field demonstration of enhanced biodegradation through soil venting was conducted at the site of
an abandoned tank farm located on Tyndall AFB, Florida. The site is contaminated with fuel, primarily JP-
4, and free product has been observed floating on the shallow ground water table. Tyndall AFB is located
on a peninsula that extends along the shoreline of the Gulf of Mexico in the central part of the Florida
Panhandle. The highest ground on the peninsula is 7.6 to 9.1 m (25 to 30 ft) above mean sea level. The
uppermost sediments, at Tyndall AFB, are sands and gravels of Pleistocene to Holocene age (2). Soils at the
site are best described by the Mandarin series consisting of somewhat poorly drained, moderately permeable
soils that formed in thick beds of sandy material (5).
The climate at the site is sub-tropical with an annual average temperature of 20.5° C (69°F). Average daily
maximum and minimum temperatures are 25°C and 16°C (77°F and 61°F), respectively. Temperatures of
32°C (90°F) or higher are frequently reached during summer months, but temperatures above 38°C (100°F)
are reached only rarely. Average annual rainfall at Tyndall AFB is 140 cm (55.2 inches) with approximately
125 days of recordable precipitation during the year. The depth to ground water on Tyndall AFB varies from
about 0.3 to 3.0 m (1 to 10 ft). The water-table elevation rises during periods of heavy rainfall and declines
during periods of low rainfall. Yearly fluctuations in ground water elevations of approximately 1.5 m (5 ft)
are typical (2). Prior to dewatering at the site, the water table was observed to be as shallow as 46 cm (1.5
ft).
Field Testing Objectives
A seven month field study (October, 1989, to May, 1990) was designed to address the following areas:
1. Does soil venting enhance biodegradation of JP-4 at this site?
2. Does moisture addition coupled with soil venting enhance biodegradation at this site?
294
-------
3. Does nutrient addition coupled with soil venting and moisture addition enhance biodegradation at
this site?
4. Will the hydrocarbons in the off-gas biodegrade when passed through uncontaminated soil?
5. Evaluation of ventilation rate manipulation to maximize biodegradation and minimize volatilization.
6. Calculation of specific biodegradation rate constants from a series of respiration tests conducted
during shutdown of the air extraction system.
7. Determination of the effects of biodegradation and volatilization on a subset of selected JP-4
components.
8. Determination of the potential for nitrogen fixation under aerobic and anaerobic conditions.
9. Evaluation of alternative vent placement and vent configuration to maximize biodegradation and
minimize volatilization.
Test Plot Design and Operation
In order to accomplish project objectives, two treatment plots and two background plots were constructed
and operated in the following manner:
1. Contaminated Treatment Plot 1 (VI)-Venting only for approximately 8 weeks, followed by moisture
addition for approximately 14 weeks, followed by moisture and nutrient addition for approximately
7 weeks.
2. Contaminated Treatment Plot 2 (V2) - Venting coupled with moisture and nutrient addition for 29
weeks.
3. Background Plot 3 (V3) - Venting with moisture and nutrient addition at rates similar to V2, with
injection of hydrocarbon contaminated off-gas from VI.
4. Background Plot 4 (V4) - Venting with moisture and nutrient addition at rates similar to Vent 2.
Air Flow
Air flow was maintained throughout the field test duration except during in situ respiration tests. Flow rates
were adjusted to maintain aerobic conditions in treatment plots, and background plots were operated at
similar air retention times. Off-gas treatment experiments in one background plot (V3) involved operation
at a series of flow rates and retention times. Soil gas was withdrawn from the center monitoring well in VI
and V2 and from the only monitoring well in V3 and V4. This configuration was selected to minimize leakage
of outside air observed when air was withdrawn from the ends of the plots. In all but one plot, V3, atmospheric
air was allowed to passively enter at both ends. Off-gas from VI was pumped back to the upstream ends of
V3. Flow rates through all test plots were measured with calibrated rotameters.
295
-------
Water Flow
To allow control of soil moisture, tap water was applied to the surface of the treatment plots. The design flow
rates allowed variation from 10 to 100 mL/min in the contaminated treatment plots, and 2.5 to 25 mL/min
in the background vents. This corresponds to an average annual surface application rate of 43 to 430 cm (17
to 170 in). Based on vacuum and oxygen measurements in the soil gas monitoring probes, it was determined
that a flow rate of 100 mL/min in the Treatment Plots inhibited air flow and oxygen transfer. Using the same
technique, a flow rate of 50 mL/min (215 cm/yr surface application rate) was selected as the final water
application rate. This rate did not appear to inhibit oxygen transfer to the soil gas monitoring points.
Nutrient Addition
The objective of nutrient addition was to apply sufficient inorganic nitrogen (N), phosphorus (P), and
potassium (K) to ensure, as far as possible, that these nutrients would not become limiting during the
biodegradation of fuel hydrocarbons in the test plots. Optimizing nutrient addition rates was not the primary
objective of this phase of the study. Sodium trimetaphosphate (Na-TMP), ammonium chloride (NH4C1), and
potassium nitrate (KNO3) were used as sources of P, N, and K, respectively.
RESULTS AND DISCUSSION
Operational Monitoring of Treatment Plots VI and V2
Treatment plots were operated for 188 days between October 4,1989 and April 24, 1990. Operation was
interrupted only for scheduled respiration tests. Discharge gases were monitored for oxygen, carbon dioxide,
and total hydrocarbons throughout the operational period. The biodegradation component was calculated
using the stoichiometric oxidation of hexane (Equation 1).
C6H14+9.5 02 R 6 CO2+7 H2O (1)
3.5 g 0/g C6H14
Oxygen consumption was calculated as the difference between oxygen in Background Plot V4 and oxygen
in the treatment plots. Using the oxygen concentration in the background plot, rather than atmospheric
oxygen concentration, the natural biodegradation of organic carbon in uncontaminated soil was accounted
for. This method ensured that the biodegradation of fuel hydrocarbons was not overestimated. Biodegradation
based on carbon dioxide production was similarly calculated. Hydrocarbon removal rates attributed to
volatilization and biodegradation are presented in Figures 1 and 2, respectively, for Treatment Plots VI and
V2. Removal rates are expressed in mg/(kg day) and are based on an estimated soil bulk density of 1440 kg/
m3 (90 lb/ft3) and a treatment volume of 20 m3 (704 ft3).
As the more volatile compounds are stripped from the soil, biodegradation becomes increasingly important
over time as the primary hydrocarbon removal mechanism as illustrated in Figures 3 and 4 for Treatment
Plots VI and V2, respectively. Percentages of combined volatilization and biodegradation removal rates
attributable to biodegradation are compared in Figure 5 for Treatment Plots VI and V2.
296
-------
CO
•o
O>
a-
lil
o
c
n
x
Biodegradation - Oxygen Basis - V1
Volatilization - THC - V1
30 60 90 120 150
Venting Time (days)
180 210
Figure 1. Hydrocarbon removal rate attributed to volatilization and biodegradation (oxygen basis)
in treatment plot VI during the field study.
n
•o
o>
or
Ul
(0
X
o
Biodegradation - Oxygen Basis - V2
Volatilization - THC - V2
30 60 90 120 150 180 210
Venting Time (days)
Figure 2. Hydrocarbon removal rate attributed to volatilization and biodegradation (oxygen basis)
in treatment plot V2 during the field study.
297
-------
c
9
o
k.
o
Q.
% Removal Attributed to Biodegradation - V1
Total Removal - Hexane Eq (VoU-Deg) - V1
30
60 90 120 150 180
Venting Time (days)
210
LU
9
C.
n
x
9
O>
Figure 3. Comparison of the combined volatilization and biodegradation removal rates and the
percent of removal rate attributed to biodegradation (oxygen basis) in Treatment Plot VI
during the field study.
c
9
0
9
a.
% Removal Attributed to Biodegradation - V2
Total Removal (Vol+Deg) - V2
30
60 90 120 150 180
Venting Time (days)
210
3
LU
9
ra
X
9
TO
•o
3
*r
O)
E
Figure 4. Comparison of the combined volatilization and biodegradation removal rates and the
percent of removal rate attributed to biodegradation (oxygen basis) in Treatment Plot V2
during the field study.
298
-------
-------
•O
05
•o
o
3
i
o»
n Total Hydrocarbon Removed (Vol+Deg)-VI
• % Removed by Biodegradation (02)- VI
90
85-
80-
75-
70-
65-
60
i i i i
02468
Air Flow Rate (L/min)
180
- 160
- 140
- 120
_ 100
- 80
r 60
10
I
"es
a1
td
-------
Figure 6 for Treatment Plot V1 and assuming that 100,000 g (3500 mg/kg) of hydrocarbons must be removed,
a hypothetical case can be evaluated. If 62% biodegradation is desired, then 81/min (two air void volumes
per day) would be selected with an expected operational time of 571 days. However, if 85% biodegradation
were desired, then 11/min (0.25 air void volumes per day) would be selected with an expected operational
time of 1370 days. Although operational time is increased by a factor of 2.4, total air requirement actually
decreases from 6.6 to 2.2 million L. Optimal air flow conditions in VI appear to be 2 1/min (0.5 air void
volumes per day) where 82% biodegradation is achieved. Although 85% biodegradation is achieved at 11/
min in V1, hydrocarbon removal rate is greatly reduced. Operating at 21/min in V1, expected operation time
is 820 days (1.4 times that required at 8 1/min) and the total air requirement is only 2.3 million L. It is
emphasized that operational times in this case are merely hypothetical as relationships between air flow and
removal rate are applicable only over the seven week field test period. However, it is likely that similar
relationships would exist throughout the remediation period although the magnitude of removal rates vary
widely.
This research has documented that decreasing air flow rates will increase the percent of hydrocarbon removal
by biodegradation and decrease the percent of hydrocarbon removal by volatilization.
Respiration Tests
Respiration Tests, 1 through 5, were conducted October 24 through 26; November 28 through December 1,
1989; January 3 through 8; March 3 through 11; and April 24 through 26,1990, respectively. In addition,
two limited respiration tests, 3A and 4A, were conducted from January 25 through 26, and March 9 through
12,1990. The respiration tests were designed to determine the order and rate of hydrocarbon biodegradation
kinetics under varying conditions of moisture and nutrient addition. Treatment Plot V2 received moisture
and nutrients throughout the experimental period and therefore served as a control for kinetic changes due
to soil temperature and other factors not related to moisture and nutrients. The respiration tests were
conducted by first shutting down the air delivery system to both the treatment and background plots, followed
by measurement of oxygen consumption and carbon dioxide production over time. Biological respiration
in Treatment Plots VI and V2 was most consistently modeled by zero order kinetics during all respiration
tests (Miller, 1990). In a system not limited by substrate, such as fuel contaminated soil, biodegradation is
likely to be best modeled by zero-order kinetics (4).
Oxygen and carbon dioxide concentrations, measured in the vapor monitoring wells prior to initiating the
respiration tests, were highly variable. Regardless of initial concentration, however, oxygen consumption
and carbon dioxide production rates were relatively consistent. For this reason, the data were normalized by
dividing oxygen concentration data measured in each vapor monitoring well by the initial oxygen
concentration at each location. A regression of the normalized data versus time for each plot and each
respiration test yielded a normalized zero order rate constant, that when multiplied by the initial average
oxygen concentration in the plot, yielded the actual zero order rate constant (k = %/min).
The normalized regression and 95% confidence interval band for Treatment Plot VI is illustrated in Figure
8 for Respiration Test 4. Figure 8 is typical of all respiration tests conducted in Treatment Plots VI and V2.
A summary of the zero order rate constant data obtained from the respiration tests is graphically illustrated
in Figure 9. In Treatment Plot VI the rate constant showed a significant drop between Test 1 and Test 2 and
between Test 2 and Test 3. The rate constant significantly increased between Test 3 and Test 4 in Treatment
301
-------
y = -2.6033E-4X + .979, R-squared: .87
1.1,
1.
.9.
5 -8-
"T .7-
CM -6-
O
8 -^
D .4.
.3-
.2-
.1
0
500
1000
1500
Time (min)
2000
2500
3000
FigureS. Regression of normalized data and 95% confidence band for Treatment Plot VI and
Respiration Test 4.
Test
I
8?
o
Test 2
Test
Test 4
Test 4A
Off-gas I
Background Plot V4
Treatment Plot V2
Treatment Plot V 1
V3
TestS
Figure 9. Average zero order rate constants determined by respiration tests.
302
-------
Plot V1 but did not significantly increase between Tests 4 and 5. Since moisture was added to Treatment Plot
V1 after Test 2 and nutrients after Test 4 their addition would seem without further analysis to be of no benefit
and even detrimental in the case of moisture addition. In Treatment Plot V2 there was a statistically significant
drop in the rate constant from Test 2 to Test 3 and a statistically significant increase in the rate constant
between Test 3 and Test 4. Although a depression appears in the rate constant data there were no other
statistically significant differences in Treatment Plot V2 rate constants.
Statistically significant differences in respiration rate between Treatment Plots VI and V2 and the
Background Plot V4 on all tests and between Off-Gas Treatment Plot V3 and Background Plot V4 on Tests
3,4A, and 5 are illustrated in Figure 9. From the data presented it was concluded that biodegradation of jet
fuel in contaminated soil and biodegradation of hydrocarbon off-gas resulted in statistically significant
increases in respiration over that observed in uncontaminated soil.
Potential Temperature Effects on Respiration Tests
As described above, hydrocarbon biodegradation rates appear to have been unaffected by moisture and
nutrient addition. This conclusion was based on insignificant differences in biodegradation rates in
Treatment Plots VI and V2 even though the treatment plots were operated under different moisture and
nutrient conditions. Although biodegradation rates in the treatment plots were similar there was a general
decline in hydrocarbon removal rates from initiation of the field study; reaching minimum values near the
middle of the experimental period; followed by a general increase in hydrocarbon removal rates through the
completion of the field study. Since the treatment plots appeared unaffected by moisture and nutrient addition
soil temperature was investigated as the potential cause of the depression in biodegradation rates.
Soil temperature at this field site was related to ambient air temperature because air was continually drawn
through the soil. More importantly, the moisture provided to the treatment plots affected soil temperature as
the applied water temperature was a function of air temperature because this water was temporarily stored
in the site building prior to delivery to the treatment plots. Local ambient temperature data were obtained
from a weather station located near the field site. The 10 day moving average above ground air temperature
data are compared with soil temperature in Figure 10. Soil temperature data before January 5,1990, were
not collected at the field site. Therefore, the relationship between ambient temperature and soil temperature
was used to estimate soil temperatures prior to this date. Comparison of average soil temperature (Figure 10)
to oxygen consumption rate (Figure 9) during respiration tests in Treatment Plots VI and V2 imply a
relationship between soil temperature and biological activity. It appears from the respiration data presented,
that soil temperature had a much more significant effect on the rate of hydrocarbon biodegradation than
moisture and nutrient addition.
To evaluate the effect of moisture and nutrient addition on biological activity in Treatment Plot V1, the effect
of temperature must be understood. Treatment Plot V2 received moisture and nutrients throughout the
experimental period and should be a control on temperature and other unmeasured variables. Therefore, a
model that eliminates the effect of temperature on the oxygen consumption rate constants (Figure 9) in
Treatment Plot V2 should be adequate for temperature correcting rate constants measured in Treatment Plot
VI, thereby allowing an assessment of the effect of moisture and nutrient addition in Treatment Plot VI.
303
-------
o
o
£
TO
e
a.
o
30 -i
20-
10-
Calculated from
measured relationship
10 day Avg Temp °C
Soil Temp °C
—r—
30
60 90 120 150
Venting Time (days)
180 210
Figure 10. Comparison of the 10 day moving average of the mean ambient above ground air
temperature and corresponding measured and estimated soil temperature.
In aquatic systems, the van't Hoff-Arrhenius equation predicts a doubling of the rate constant with each
temperature increase of 10°C, assuming typical activation energy values (1). Figure 11 is the Arrhenius Plot
for determining activation energy using measured soil temperature and rate constant relationships from Tests
3,4, and 5 for Treatment Plots V1 and V2. Using the Arrhenius constants determined from the plots in Figure
11, the rate constants for Treatment Plots V1 and V2 were corrected to 23°C, the soil temperature of Test 1
. The Arrhenius model for temperature correction resulted in insignificant rate constant differences between
Tests 2,3,4, and 5 in Treatment Plot V2 (Figure 12). Therefore, the Arrhenius equation adequately modeled
the effects of temperature on hydrocarbon biodegradation rate. Using the same model, the oxygen
consumption rate constants in Treatment Plot VI were corrected for temperature (Figure 13). Although
statistically significant differences in rate constants remained between Test 3 and Tests 2 and 4 in Treatment
Plot VI, the magnitude of the differences are not important from a practical application standpoint. Test 1
in both treatment plots was not considered because it was conducted when hydrocarbon concentrations in
the soil gas were still very high.
Moisture was added to Treatment Plot VI following Respiration Test 2 and nutrients were added following
Respiration Test 4. Temperature corrected rate constants (Figure 13), were not significantly increased
between Tests 2 and 3, and between Tests 4 and 5. Therefore, it can be concluded that moisture and nutrient
addition were of insignificant benefit to the rate of hydrocarbon biodegradation in Treatment Plot VI.
Although moisture and nutrient addition did not effect biodegradation rates, the data indicate that soil
temperature likely did.
304
-------
4.5-
4.0-
3.5-
3.0-
B 2.5-
2.0
1.5H
l.OH
D VI y=23.501 - 6483.3x R2 =0.864
• V2 y=15.383 - 4000.3x R2 =0.942
VI - Ea/r=6483 lnB=23.5
V2-Ea/r=4000 lnB=15.4
0.0032
0.0033
0.0034
0.0035
1/T
Figure 11. Arrhenius Plot for determining activation energy using measured soil temperature and
rate constant relationships from Tests 3,4, and 5 for Treatment Plots VI and V2.
0.008-
0 95% Confidence Interval V2
IB Arrhenius Corrected Minimum K-V2
0.006
Moisture and nutrients added
OJD
^> 0.004-1-
I*N
O
0.002--
0.
(N
cs
Location
Figure 12. Temperature corrected (23°C based on Arrhenius Plot) oxygen consumption rate constants
determined by respiration tests for Treatment Plot V2. Mean k is at the center of the 95 %
confidence interval.
305
-------
0 95% Confidence Interval VI
D Arrhenius Corrected Minimum k-Vl
on
U.UU8
0.006-
0(\r\A
0.002-
0.000
m
Moisture added Nutrients added
i
-
—
^^^
1
mm
in
Location/Test No
Figure 13. Temperature corrected (23°C based on Arrhenius Plot) oxygen consumption rate constants
(k) determined by respiration tests for Treatment Plot VI. Mean k is at the center of the
95% confidence interval.
CONCLUSIONS
This field scale investigation has demonstrated that soil venting is an effective source of oxygen for enhanced
aerobic biodegradation of petroleum hydrocarbons (jet fuel) in the vadose-zone. Specific conclusions are:
1. Operational data and respiration tests indicated that moisture (6.5 to 9.8% by weight) and nutrients
were not a limiting factor in hydrocarbon biodegradation. Oil and water samples indicated that
nutrients were delivered to the treatment plots and passed through the vadose-zone to the ground
water.
2. Air flow tests documented that decreasing flow rates increased the percent of hydrocarbon removal
by biodegradation and decreased the percent of hydrocarbon removal by volatilization. Under
optimal air flow conditions (0.5 air void volumes per day) 82% of hydrocarbon removal was
biodegraded and 18% volatilized. Biodegradation removal rates ranged from approximately 2 to 20
mg/(kg day), but stabilized values averaged about 5 mg/(kg day). The effect of soil temperature on
biodegradation rates was shown to approximate effects predicted by the van't Hoff-Arrhenius
equation.
3. Off-gas treatment studies (3) documented that uncontaminated soil at this test site could be
successfully used as a biological reactor for the mineralization of hydrocarbon vapors (off-gas)
generated during remediation of fuel contaminated soil using the enhanced biodegradation through
soil venting technology investigated in this field study. The average off-gas biodegradation rate was
1.34 (SD± 0.83) mg/(kg day), or 1.93 (SD ± 1.2) g/(m3 day). The percent of off-gas biodegradation
306
-------
was inversely related to air flow rate (retention time), and was directly related to hydrocarbon loading
rate, at the 95% confidence level. Based on data collected at the field site, a soil volume ratio of
approximately 4 to 1, uncontaminated to contaminated soil, would be required to completely
biodegrade the off-gas from a bioventing system operated similar to this field project. However, if
air flow rates in contaminated soil were designed to maximize biodegradation, the ratio of
uncontaminated to contaminated soil required would be proportionally less
4. Respiration tests documented that oxygen consumption rates followed zero-order kinetics, and that
rates were linear down to about 2 to 4 % oxygen. Therefore, air flow rates can be minimized to
maintain oxygen levels between 2 and 4% without inhibiting biodegradation of fuel, with the added
benefit that lower air flow rates will increase the percent of removal by biodegradation and decrease
the percent of removal by volatilization.
5. Initial soil samples indicated that naturally available nitrogen and phosphorus were adequate for the
amount of biodegradation measured, explaining the observation that nutrient addition had an
insignificant effect on the rate of biodegradation. Acetylene reduction studies (3) revealed an organic
nitrogen fixation potential that could fix the observed organic nitrogen, under anaerobic conditions,
in five to eight years.
6. Soil moisture levels did not significantly change during the field study. Soil moisture levels ranged
from 6.5 to 7.4%, and 8.5 to 9.8%, by weight, respectively, in Treatment Plots VI and V2. Neither
venting nor moisture addition had a statistically significant effect on soil moisture at this site.
RECOMMENDATIONS FOR FUTURE STUDY
To further pursue the development of an enhanced biodegradation of petroleum hydrocarbons through soil
venting technology, the following studies are recommended:
1. Further investigate the relationship between soil temperature and hydrocarbon biodegradation rate.
2. Investigate methods to increase hydrocarbon biodegradation rate by increasing soil temperature with
heated air, heated water, or low level radio frequency radiation.
3. Investigate the effect of soil moisture content on biodegradation rate in different soils with and
without nutrient addition.
4. Investigate nutrient recycling to determine maximum C:N:P ratios that do not limit biodegradation
rates.
5. Investigate different types of uncontaminated soil for use as a reactor for biodegradation of generated
hydrocarbon off-gas and determine off-gas biodegradation rates.
6. Investigate gas transport in the vadose-zone to allow adequate design of air delivery systems.
307
-------
REFERENCES
1. Benefield, L. D; Randall, C.W. Biological process design for wastewater treatment; Prentice-Hall, Inc.,
Englewood Cliffs, New Jersey, 1980; pp 11-13.
2. EnvironmentalScienceandEngineeringlnc/'Installationrestorationprogramconfirmation/quantification
Stage 2 Volume 1 Tyndall AFB, FL"; final report to Headquarters Tactical Air Command, Command
Surgeon's Office (HQTAC/SGPB), Bioenvironmental Engineering Division, Langley AFB, VA, 1988.
3. Miller, R. N. Ph.D. Dissertation, Utah State University, 1990.
4. Riser, E. "Technology review - In situ/on-site biodegradation of refined oils and fuel"; PO No. N68305-
6317-7115 to the Naval Civil Engineering Laboratory, Port Hueneme, CA, 1988.
5. Soil survey of Bay County Florida; 1984; U.S. Department of Agriculture. Soil Conservation Service.
U.S. Government Printing Office, Washington, DC, 1984.
308
-------
SUBSURFACE REMEDIATION AT A GASOLINE SPILL SITE USING A
BIO VENT APPROACH
DONH.KAMPBELL
U.S. Environmental Protection Agency
Robert S. Kerr Environmental Research Laboratory
RO. Box 1198
Ada, OK 74820
INTRODUCTION
Soil vapor extraction in combination with biodegradation is a promising remediation technology. Laboratory
treatability studies have shown that the process of bioventing should be adaptable to a considerable range
of conditions and volatile organics.
An aviation gasoline spill of about 35,000 gallons occurred at an air station in 1969. A major portion of the
spill still persists as oily phase residue in the capillary fringe of the subsurface. The vertical profile of the
subsurface is a relatively uniform beach sand to below the water table which was near 5 meters.
The objective of the project was to design, install, operate, and evaluate two pilot-scale bioventing systems.
Anticipated results were to demonstrate the enhanced feasibility of engineered biological remediation of a
subsurface containing retained oily phase gasoline. Performance of the two pilot-scale systems was to
demonstrate that surface emissions of fuel are minimum, total fuel hydrocarbons in remediated core material
will be less than 100 mg/Kg, final benzene in the underlying groundwater will not exceed 5 Jig/L,
remediation will be completed in a reasonable time, and the technique is applicable to full-scale reclamation.
EXPERIMENTAL
Prior to design of the bioventing units several laboratory soil microcosm treatability studies were conducted
using surface soil from the field site. Aviation gasoline vapor biodegradation was rapid and complete
showing curves typical of first-order kinetics.
Degradation occurred at all temperatures within a range of 4 to 37°C. A moisture range of 3.5 to 20 percent
did not limit degradation. Reaction rates and active biomass were increased at least four fold in test
microcosms receiving a nutrient addition of nitrogen, phosphorus, and potassium. The biodegradation rates
obtained from acclimated soil microcosms were more than adequate to consume all vapors in the unsaturated
zone at the pilot-scale demonstration site.
DESIGN
The design concept provided subsurface forced aeration to vaporize and transport oily phase components
upward to more amiable microbial degradation activity. A surface area above the plume was divided into two
adjacent equal plots of 13.7 X 22.9 meters (45 X 75 feet). The north plot shown in Figure 1 was an injection
only aeration system. The injection wells were placed in a 3 X 5 grid 3.05 meters apart to a depth of 5.5 meters
309
-------
which was one meter below the existing water table level. Each well had an outer 5 inch diameter PVC tube
with a well screen of 10 vertical slots 1.5 meters long. A depth adjustable inner 4 inch diameter open-end
PVC tube 5.8 meter long with a lower rubber packer and an upper rubber collar seal was inside the larger
well tube. The above ground portion of the smaller tube was connected to a main forced air transfer line. An
injection depth was set to emit air flow over 0.15 meter depth just above the water table. Seven vapor
withdrawal wells were similar in construction to the injection wells and placed alternately in the south plot.
Extracted air was reinjected to a depth of 4 meters midway between coupled units as shown in Figure 1. A
blower flow rate of 5 cfm per plot was used which was equivalent to a calculated forced air residence time
of 24 hours.
Anutrient solution of 25,5, and2 mg/Kg soil of nitrogen, phosphorus, and potassium was applied throughout
the unsaturated zone to sustain microbial activity. A turf grass cover was established and maintained to
provide a root zone rhizosphere to complete removal of fuel vapor constituents in surface emissions.
SYSTEM MONITORING
Core material, underlying groundwater, soil gas and air vented air streams were monitored to determine the
extent of remediation.
Vertical profile core samples were collected and analyzed for fuel carbon at three month intervals.
Groundwater was analyzed monthly for BTEX, dissolved oxygen, and nutrients. Soil gas probes were
installed to measure fuel vapor, oxygen, and carbon dioxide on a regularly scheduled basis. Subsurface
moisture and temperature was recorded from the meter readings. Soil pore water was collected from different
depths for nutrient analysis. Gauge readings were recorded for flow and pressure of the vented air streams.
Surface emission samples were collected by cartridge traps for analysis of vapor constituents.
RESULTS & DISCUSSION
The air blowers were turned on during October 1990. They were operated continuously for three months at
a subsurface volume flow calculated to be a 24 hour residence time. The systems were shut down for the
winter in January. The frost line was then below the turf root zone. Operation will commence again in April
1991.
Soil gas was monitored at depths of 1,2,3, and 4 meters. Fuel vapor concentrations increased to a maximum
the first few days at all depths in both the north and south plots then decreased rapidly until stabilizing in three
weeks (Figures 2,3). A combustible gas Threshold Limit Meter (TLV) was used. A response factor near 0.6
was obtained from aviation gasoline vapor when comparing meter readings with standard calibration by
butane in ppm (v/v). Initial carbon dioxide in the subsurface soil gas varied from 3.5 to 8.0 percent. During
venting, the levels remained at <0.1 percent. Oxygen initially was in the 10 to 14 percent range then increased
to 20 percent during venting.
Water from the three monitoring wells was collected at two depths and analyzed as shown in Table 1. Benzene
at depths nearest to the water table within the influence of forced air venting was much lower than at depths
further down.
310
-------
Averaged concentrations of fuel carbon over vertical profiles of both the north and south plots showed much
greater reductions than in adjacent control locations (Tables 2,3). Although fuel carbon had been reduced
considerable, the desired concentration of 100 mg/Kg has not yet been attained as of February 1991. Water
table levels at the core sampling times of September 1990 and February 1991 were 472 and 477 cm
respectively.
Surface emissions were less than 1 mg fuel hydrocarbons/L soil air (Tables 4). The concentration of fuel
hydrocarbons were reduced over 100 fold between the soil surface and at a one meter depth.
CONCLUSION
Operational performance of both venting systems during the first three months of an expected twelve month
field project was satisfactory. Oily phase gasoline was reduced in the core profile of both bioventing system
plots, particularly in the unsaturated subsurface above the water table. Water quality for benzene has been
improved that suggested a beneficial influence from venting aeration. Fuel emissions from the soil surface
were minimal indicative of active cleaning action in the turf root rhizoshpere.
ACKNOWLEDGEMENTS
Christopher J. Griffin, Project Engineer, with the TraverseGroup, Inc. was the on-site operations manager.
John T. Wilson, Bioremediation Team Leader, with the Robert S. Kerr Environmental Research Laboratory
has provided helpful guidance.
311
-------
Table 1. WATER QUALITY - FEBRUARY 1991
SAMPLE
Upgradient
Upgradient
South Plot
South Plot
North Plot
North Plot
DEPTH
METER
5.3
6.7
5.2
6.1
5.1
6.4
TOTAL AVGAS
1 1 frf
\^&
3090
415
686
1680
2790
2410
BENZENE
1
182
132
<1
19
7
202
Table 2. CORE PROFILE OF FUEL CARBON (mg/kg)
DEPTH CONTROL, NORTH PLOT
(CENTIMETERS) SEPT. 1990 - FEB. 1991 SEPT. 1990 - FEB. 1991
406 234 97 52 <8
444
462
488
503
523
538
X
461
1030
701
6500
5620
N.D.
2862
1080
1000
731
8240
3020
<8
2814
923
1253
926
6740
5780
39
3124
18
153
972
1420
2200
53
913
(mean value within dashed area)
312
-------
Table 3. CORE PROFILE OF FUEL CARBON (mg/kg)
DEPTH CONTROL^ SOUTH PLOT
(CENTIMETERS) SEPT. 1990 - FEB. 1991 SEPT. 1990 - FEB. 1991
406 <8 <8 11 <8
444
462
488
503
523
538
X
193
238
212
549
34
<8
245
84
85
140
639
10
<8
192
144
253
1970
1880
2830
<8
1415
16
70
946
1860
17
<8
582
(mean value within dashed area)
Table 4. SURFACE EMISSIONS Volatile Hydrocarbons
Nov. & Dec South Plot North Plot
Canopy 0.24 mg/L 0.52 mg/L
One meter probe 271 mg/L 174 mg/L
Per cent removed 99.9 99.7
313
-------
•*-—-
\
\
\
\ DG
\ +
\
\
\
\
\
\
\
\
\
V
\
Ui N
£ \
\
\
\
V
\
\
LEGEND \
\
0 MONITORING WELL - MW
• AIR INJECTION WELL - Al
« VAPOR EXTRACTION WELL - VW
A VAPOR REINJECTION WELL - VRI
•f PRESSURE AND GAS PROBE - Cl
a MOISTURE AND TEMPERATURE PROBE - MT
FENCE
+ EXISTING SOIL CLUSTER WELLS
10 0 10 20
FEET
1"=20'
o
SOUTH PLOT
DF
r •=£_= — r- — . — ^
(• i e/^i »i
i/ !/.
j Y i
' / '
rn s\
/ , '
+ '\ r'\ I
/ DH /
NORTH PLOT
>• 4" PVC BALL VALVE
^T*l +
1 L_J» _ 1 _ J
l . 1
l
l i
l t---*- J - ^
, 1
; \ J^ -fy •_ ^J
1
L. J 1 __J
1 i h O
+ DB
L J -4L t J
DE
VAPOR WITHDRAWAL LINE ^ j , \
1 IZT""1 \- OUTER BOUNDARIES OF PROJECT AREA
\\/Af>nr? PTIM irrTiOM i IMT L,
" \ ct/
\ REDUCED TO (. }///
X> 2" PVC LINE <<£&/
\ IN THIS AREA f TT
\ I-T
i !
V
BIO VtNl
BLOWER
BUILDING
^LL VALVE
FIGURE 1
NG FLOW LINE PLAN
-------
SIC
SI-TO
Sl-TO
3NJT NOU33PNQH HCWVA »
Sl-TO
SI-ID
-------
BIOVENTING
AN IN SITU REMEDIAL TECHNOLOGY
SCOPE AND LIMITATIONS
/. VANEYK
Delft Geotechnics
P.O. Box 69
2600 AB Delft, The Netherlands
INTRODUCTION
For a variety of reasons, in situ methodologies are not as popular as one might wish. One obvious reason
is, that when the decision has been taken that a site has to be cleaned up, the results of the in situ cleanup
technique will have to comply with the standards laid down by the authorities with respect to maximum
allowable residual levels for soil. If no guarantee can be given with respect to the minimum levels achievable,
excavation (if possible) will be the cleanup method of choice. As a result, the development of in situ
techniques advances only very slowly. To circumvent this problem and to give the development of in situ
methodologies a fair chance, the Dutch Government has decided to support research and development of in
situ techniques.
Another problem, which particularly applies to the application of biological cleanup methodologies, is the
general consensus that biological methods intrinsically require very long cleanup times, which when
compared to excavation is true. However, excavation is not always an option, particularly when whole
buildings have to be demolished. The only alternative in those instances would be the application of
measures of abatement. In most instances, however, it will then be possible to apply in situ methodologies
for restoration, which eventually will make pollution control measures superfluous. In addition, when large
scale aquifer restoration methods have to be carried out at high costs, the additional costs for applying an in
situ methodology like bioventing, are relatively insignificant when compared to the results achieved.
Another advantage of an in situ technique such as bioventing is, that it can also be considered in terms of the
goal of prevention. For, when soil cleanup has been achieved as a result of the application of bioventing,
the system, when left intact can always be reactivated at relatively low costs, whenever an accidental spill
reoccurs in the underground. It has the additional advantage, that the extra costs to remove the system can
be avoided.
To date, the technique of venting to remove petroleum vapors from the vadose zone, has been well
documented (2). The use of bioventing to enhance volatilization and to stimulate biodegradation in the
vadose zone, was first reported in 1986 (1, 4). An investigation into the possibilities for bioventing was
prompted by the results of lysimeter studies on microbially mediated gasoline removal from sand. In the
absence of bioventing, almost 100 percent removal was achieved in a period of 7 to 8 years (6).
317
-------
Laboratory studies carried out with sand columns which had been saturated with gasoline or diesel oil,
showed for gasoline, that 99 percent could be vented off in a period of 6 weeks, by passing water vapor
saturated air through these sand columns with a venting ratio, d, of 30. The venting ratio, d, equals air flow
per m3 soil per day, divided by the air filled porosity of the soil (pore volumes exchanged per day). Such high
ratios are unlikely to be achieved in the field and the contribution of biodegradation to the removal of
petroleum components at lower venting ratios, say 6, is therefore also important (7). On the basis of carbon
dioxide production measured in the gas phase, the vented columns containing diesel oil (d = 6) showed a
hydrocarbon biodegradation rate of approximately 100 grams per m3 soil per day which is equivalent to 60
mg/kg/day if soil bulk density equals 1.67 g/cm3.
A feasibility study carried out to evaluate the possibilities for venting under field conditions, showed that
for hydraulic conductivities in soil equal to or greater than 105 m/sec, venting can be used either to remove
volatile hydrocarbons from subsurface soil strata or to stimulate biodegradation of the less volatile petroleum
components (8, 12).
A large scale field experiment was subsequently carried out to validate the feasibility study (8,9). The results
of these studies are summarized in Tables 1,2, and 3, and show that gasoline is largely removed by venting
and to a lesser degree by bioventing. In contrast, for diesel oil, the contribution of enhanced volatilization
is negligible and enhanced biodegradation appears to be the main attenuation mechanism. Table 2 also
shows, that the benzene, toluene, xylene (BTX) trio is removed preferentially, and, except for the xylenes,
to almost background levels. The xylenes appear both more difficult to biodegrade and to evaporate and
either need more time or, supplementary treatment measures. This will be discussed later in more detail.
Asa follow up, a contaminated location at a retail gasoline station was selected to demonstrate the bioventing
approach in a real spill situation. This demonstration Project, supported by the Department for the
Environment (Grant no. MJZ 20 D 8037) has been reported before (10). Therefore only a cursory description
of the project will be given in this paper. The results, and the scope as well as the limitations of the bioventing
technique, however, will be discussed in more detail.
INSTALLATION OF AN IN SITU AND ON SITE CLEANUP SYSTEM
The installation of an in situ and on site cleanup system for the selected site (Figure 1), required the necessary
preparations, which are summarized as follows:
Design Parameters
Soil structure and composition
Data on soil structure and composition are required for the design of a remedial system both for soil and
ground water. Hydraulic conductivities for soil were calculated from grain size distribution curves and also
determined in the laboratory. An average, calculated hydraulic conductivity of 1 .OE-04 m/s was selected on
the basis of these two sets of data. Hydrogeological parameters were required for the installation of two
pumping wells, to contain, pump and treat the polluted ground water.
318
-------
Degree and extent of contamination
The degree and extent of contamination was determined very accurately, based on an extensive soil and
ground water sampling procedure (Figure 1, Figure 2 and Tables 4, 5, and 6).
Hydraulic conductivity for soil
Hydraulic conductivity data for soil were collected to carry out calculations on ground water transport and
ground water withdrawal. In addition this data is used to get an insight into soil air permeability.
Soil air permeability
Soil air permeability was derived from the hydraulic conductivities and the parameter used for the air filled
porosity was adjusted for soil moisture content. On the basis of hydraulic conductivities, a value of 8.7E-
04 m/s, was calculated for soil air permeability.
Model Calculations
Air transport calculations
Air transport calculations were carried out to determine the spacing of the air wells. Air transport was
calculated, using the Hydrology Contaminant Transport Model (12). Air permeability for moist soil was
calculated with the help of the Karman-Kozeny equation. Based on these calculations, a distance of three
meters was selected between the air wells, resulting in a calculated air flux of 68 m3/day, at a negative pressure
of - 0.05 bar.
Vapor transport calculations
Vapor transport calculations were carried out to get an insight into the time required to remove the bulk of
volatile contaminants, particularly the BTX components, from soil (Figure 3).
Calculations of pumping rates for ground water
Pumping rates for the cleanup of contaminated ground water were calculated with the computer program
called SLAEM, Single Layer Analytic Element Model (5). The model was verified with measured data on
ground water and surface water levels. For the numerical calculations, a Kd of 250 m2/day was assumed and
a net rainwater infiltration rate of 0.15 m/year. Apumping rate of 5 m3 per hour was shown to produce a radius
of influence sufficiently large, to contain and to treat the contaminant plume.
Evaluation of biodegradation rates
On the basis of an air flux of 68 m3 per day per air well, a minimum venting ratio of 6 can be calculated. This
will sustain a biodegradation rate of 100 g per m3 per day, which in view of rates determined before in the
field (9), is more than adequate.
319
-------
System Design
The complete remedial system was composed of the following parts:
Pumping wells
Because the aquifer appeared to be contaminated up to 10 meters below ground surface, two pumping wells
had to be installed, one with a filter at 4 m below ground surface and one with a filter at 11 m below ground
surface (Figure 4).
Pumping air wells
A total of 24 pumping air wells were installed as part of the soil bioventing system (Figure 4). Each individual
air well is connected to the main pipeline which is connected to the blower. As the ground surface has an
impermeable pavement, access of air is obtained via two horizontal drains installed below the pavement
(Figure 5).
Impermeable Pavement
An impermeable pavement is an integral part of this venting system for several reasons. The venting system
has to be protected from any form of adverse effect or damage, caused, for instance, by heavy trucks. In
addition, because the retail gasoline station has an on-going business during the period of remediation,
migration of fresh spilled products into the treated zone had to be excluded.
Air and ground water cleanup system
The cleanup system for air and ground water is shown in Figure 6. The ground water, which is mainly
contaminated with volatile petroleum components, is cleaned up in a stripping tower. An oil water separator
was installed, because boring B2 (Table 4, Figure 1) had indicated the presence of free product floating on
the ground water table. As the soil had been shown to contain a considerable amount of iron, which as a result
of aeration will produce iron hydroxide, a sand filter was installed after the cascade aerator.
Biofilter
A biofilter which was installed to remove gaseous petroleum components from air extracted from the soil,
was also used for the cleanup of air from the stripping tower. The biofilter contained 60 percent heather, 40
percent peat, and 0.1 percent activated sludge taken from a biotreater used for the cleanup of ground water
contaminated with petroleum. The total volume amounted to 18 m3 (called Biomix).
Monitoring and control
Progress of the cleanup process is monitored via vented air samples by determining the concentration of
carbon dioxide and total hydrocarbons. Ground water cleanup is monitored from ground water samples
obtained from observation wells. Soil samples will be taken in the final stages of the cleanup process.
320
-------
RESULTS
Venting
To avoid the possibility of overloading the biofilter, only air extracted from the soil was passed over the
biofilter when the in situ cleanup operation was started. When after several weeks of operation air sampling
showed that the biofilter was working effectively, the pumping wells for ground water were also activated.
Air produced by the air stripper was then fed into the biofilter too. At the start of the operation, the air flux
amounted to approximately 20 m3/h. After two weeks it increased to 32 m3/h. The average over a period
of 12 months amounted to around 36 m3/h. Air transport calculations predicted a value of 68 m3/d per air
well (24 air wells were installed). The removal of petroleum vapors from soil by venting, is shown in Table
7. Over a total period of 13 months, approximately 770 kg of petroleum vapors were extracted from the
subsurface. These results confirm results obtained before(7), which showed that components like benzene
can be removed quickly and effectively. After 3 months, the benzene concentration had been reduced to 4
percent of its starting value. For toluene and the xylenes, the figures are 9 percent and 24 percent respectively.
Bioventing
The removal of petroleum from soil as a result of bioventing is shown in Figure 7. The data was obtained
by dividing the weight of the CO2 produced by the weight of [CH2] oxidized, i.e. 44/14 or 3.14. In a period
of roughly one year, approximately 430 kg of petroleum components were removed as a result of bioventing.
The data represented in Figure 7 indicates that CO2 production rates are presently falling off. This, however,
may at least in part be the result of a seasonal temperature fluctuation. This is underlined by the CO2 data
represented in Figure 8, which clearly shows a steady increase in ppm produced throughout the season and
a decrease in September. It is noteworthy that the first and the last CO2 value determined around the same
time differ significantly, which may be caused by the presence and absence respectively of toxic petroleum
components, like BTX.
Biofilter performance
Unfortunately, it is not possible to produce actual data on biofilter performance. The reason is, that local
environmental authorities, as yet, do not impose strict quantitative restrictions on the amount of petroleum
vapors which can be vented off onto the atmosphere. All that is presently being required is, that the vented
air is odorless. To get an insight into the biofilter performance, air samples were taken before entering and
after passing the filter with Drager tubes. These tubes are designed, however, to determine BTX compounds
singly and not in combination. The results, therefore, are only indicative, but do show that BTX compounds,
after three weeks of operation can not longer be detected. Assuming a 50 percent removal rate during the
first three weeks and a 100 percent removal for the remaining period (see Table 7), an average removal rate
of 0.120 kg/m3/day can be calculated.
Ground water cleanup
Tables 8 and 9 show the hydrocarbon levels analyzed in the two influents discharged into the cleanup system.
The data show, that ground water extracted from the deep pumping well is only marginally polluted. Ground
water pumped from the shallow well is, in contrast to the deep well, considerably polluted.
321
-------
Approximately three months after the start of the demonstration project, hydrocarbon levels start to rise
significantly. It appeared to be caused by free product entering the pumping well. Two months later,
concentrations start to fall again, particularly those for benzene, toluene, and the xylenes. Hydrocarbon
concentrations in the effluent were determined to comply with requirements for discharge. Maximum
allowable levels for petroleum and BTX compounds, prior to discharge into the sewer are 10 mg/1 and 100
ug/1 respectively. Analysis show that these targets are being met by the ground water cleanup system.
DISCUSSION
If soil has a hydraulic conductivity equal to or greater than 10~5 m/s, the technique of venting is very useful
for the removal of volatiles from soil (8, 12). To date, the application of venting to remove volatile
hydrocarbons from soil has been well documented (2). In the field experiment (7), BTX components were
reduced in a period of 50 weeks to between 1 and 10 percent (Table 2) of the starting value (uncorrected for
biodegradation). Calculations show, however, that higher removal efficiencies can be obtained by venting
over extended periods of time. The technique of venting, therefore, is an effective method for the removal
of volatiles from soil. Enhanced biodegradation as a concomitant result of bioventing will mop up the less
volatile components.
For diesel oil, the situation is different. The field experiment mentioned before, showed that the volatilization
of diesel oil is insignificant. Enhanced biodegradation as a result of bioventing is for diesel oil the most
important process for the removal of the petroleum components. The question, however, is: will
biodegradation as a result of bioventing achieve complete cleanup of the soil? Because the more water
soluble components of petroleum products are removed more quickly as a result of leaching and subsequent
biodegradation, the viscosity of the residual fraction increases. This will progressively slow down the
biodegradation of the residual petroleum components. This could explain an observation reported before,
that the relative concentration of polycyclic aromatics (PCAs) in diesel oil during bioremediation, increases.
Except for the naphthalenes which sublimate easily (naphthalene > methyl-naphthalene > dimethyl-
naphthalene), the relative concentrations of compounds like fluoranthene, increases significantly with time
(see Table 10).
Complete biodegradation of diesel oil in soil may therefore require considerable longer cleanup times than
anticipated on the basis of the initial removal rates. In the Netherlands, a standard soil with respect to
organics, is assumed to have an organic matter content of 10 percent and the reference values (R) are based
on this assumption. If the organic matter content is smaller than or equal to 2 percent, the value of 2 percent
is maintained. This reduces the value of R five times. For petroleum, R = 50 mg/kg. If the organic matter
content is 2 percent, the value of R is obtained by dividing the standard value of R by 10 and multiply it by
2, which reduces R to 10 mg/kg. The average residual diesel oil concentration which we determined in the
field experiment at a sampling depth of 1.25 to 1.5 m below ground surface was 23,087 mg/kg at to.
On the basis of the percentage composition of PCAs, for this particular diesel oil, (Table 10), this results in
the following concentrations:
322
-------
COMPONENT RESIDUAL CONCENTRATION R(OM=29EA
(mg/kg at to) (mg/kg)
PETROLEUM 20,387 10
NAPHTHALENE 18.8 0.002
ANTHRACENE/PHENANTHRENE 42.8 0.020
FLUORANTHENE 5.1 0.020
PYRENE 10.0 0.020
This data, when compared to the final levels to be achieved as listed in the second column shows, that a
considerable reduction is to be achieved, which in view of the observed initial enrichment, woi'< 't be easy to
accomplish. A true limitation for bioventing with respect to the removal of diesel oil from soil may be caused
by the fact, that soil polluted by liquid petroleum products contains free products in places which are
inaccessible to micro-organisms. This is probably one reason for the observation, that hydrocarbon
concentrations in landfarming operations usually appear to remain unchanged at the level of one gram per
kg, which according to the Dutch Act on Soil Protection and Soil Remediation, is 100 times larger than the
cleanup values to be achieved. The decrease in the rate of biodegradation due to the increase in viscosity
may be acceptable, because it only prolongs the period required for remediation. However, inaccessibility,
if true, would make it eventually impossible for an in situ technique to achieve the levels required.
Fortunately, at this point, the microbe itself may solve the problem for us. Due to microbial action, the
solubility of hydrocarbons can increase significantly (up to a 100,000 times) for example (3). The solution
to the problem therefore may be to exploit this phenomenon by intermittent flushing of the soil. This idea
is supported by an observation made by Verstraete et al. (1975), that gasoil entrapped in soil columns could
be leached out (up to 15 times), by percolating water enriched in minerals through these columns.
The in situ flushing may have the same effect as tilling in landfarming operations, namely to expose
inaccessible residues to microbial action, in part as a result of microbial solubilizing factors. Experiments
carried out in our laboratory (to be reported) show, that intermittent flushing of vented columns, reduces
residual petroleum concentrations much faster and much more efficiently than bioventing alone .
REFERENCES
1. Anonymous. "In Situ Reclamation of Petroleum Contaminated Sub-soil by Subsurface Venting and
Enhanced Biodegradation. Research Disclosure February 1986, No. 26233,92-93.
2. Hutzler, N.J.; Murphy, B.E.; Gierke, J.S. "State of Technology Review: Soil Vapor Extraction Systems";
final report to the U.S. EPA, Hazardous Waste Engineering Research: Cincinnati, OH, 1988.
323
-------
3. Roy, P.K. et al., 1979. Characterization of hydrocarbon emulsification and solubilization during growth
of Endomycopsis lipolytica on hydrocarbons. Biotechnol. Bioeng. 21:955-74.
4. Staatsuitgeverij Den Haag. Proceedings of a workshop, 20-21 March, 1986, Bodembeschermingsreeks
No. 9: Biotechnologische Bodemsanering, pp 31-33, rapportnr. 851105002, ISBN 90-12-054133,
ordernr. 250-154-59; Staatsuitgeveri; Den Haag: The Netherlands, 1986.
5. Strack, O.T.L. Groundwater Mechanics; Prentice-Hall: Englewood Cliffs, NJ, 1989.
6. Tibbetts, P.J.C. "The Analysis of Oil in Sand from Four Lysimeters in Katwijk, The Netherlands; a
COOW-CONCAWE draft report, No. 2870/227/1/925; 1982.
7. van Eyk, J., Vreeken, C. "Venting-Mediated Removal of Petrol from Subsurface Soil Strata as a Result
of Stimulated Evaporation and Enhanced Biodegradation. Med. Fac. Landbouww. Rijksuniv. Gent
1988, 53 (4b), 1873-1884.
8. van Eyk, J., Vreeken, C. "Model of Petroleum Mineralisation Response to soil Aeration to Aid in Site-
Specific, In Situ Biological Remediation." In Groundwater Contamination: Use of Models in Decision-
Making, Proceedings of an International Conference on Groundwater Contamination; Jousma et al.,
Eds.; Kluwer: Boston/London, 1989a; pp 365-371.
9. van Eyk, J., Vreeken, C. "Venting-mediated Removal of Diesel Oil from Subsurface Soil strata as a
Result of Stimulated Evaporation and Enhanced Biodegradation. In Hazardous Waste and Contaminated
Sites, Envirotech Vienna, Vol. 2, Session 3, ISBN 389432-009-5; Westarp Wiss: Essen, 1989b;pp 475-
485.
10. van Eyk, J., Vreeken, C. "In Situ and on-site subsoil and aquifer restoration at a retail gasoline station:
In Situ Bioreclamation", R.E. Hinchee and R.F. Olfenbuttel (ed.) 1991, pp 303-320.
11. Verstraete,W., Vanloocke, R., 1975. Modelling of the breakdown and the mobilisation of hydrocarbons
in unsaturated soil layers. Proc. 3rd International Biodegradation symposium, J.M. Sharpley and A.M.
Kaplan (ed.). Appl. Science Publications, London, 1975.
12. Vreeken, C., Sman, H.T. "The Use of a Hydrology Contaminant Transport Model for the Prediction of
the Effect of Air- stripping on the In Situ Cleaning of Contaminated Soil. In Groundwater Contamination:
Use of Models Decision-Making, Proceedings of an International conference on Groundwater
Contamination; Jousma et al., Eds.; Kluwer: Boston/London, 1989; pp 329-327.
324
-------
Table 1. Mass balance for the removal of gasoline from soil by bioventing
(van Eyk and Vreeken, 1988)
Applied
Mobile gasoline lost via drains
Removed by venting *
Lost by uncontrolled evaporation **
Dissolved products lost via drains
Residual gasoline
Removed by biodegradation
(as determined from CaCO3 leaching)
119.8/125 x 100X = 96% can be
accounted for
125kg
2kg
72kg
6.8kg
7.5kg
4.5kg
27kg
119.8kg
* Preliminary results determined by gas chromatographic analysis
* * Lost to the air during application and during the periods of time when the sealing of the field had
to be removed
Table 2. Effect of venting on the percentage contribution of BTX components to gasoline in soil
samples after a venting period of 12 months
sample
depth
m
.50 - .75
.75 - 1.00
1.00- 1.25
petrol
ppm
25
31
402
gasoline components2 in ppm
benzene
<0.50
.20
<0.50
.20
<0.50
3.0
toluene
<0.10
1.8
<0.10
2.3
<0.50
29.3
xylene
<0.10
2.2
<0.10
2.7
3.6
35
found
calcd1
found
calcd
found
calcd
1 Calculated on the basis that the total hydrocarbon content of the soil sample represents genuine petrol
2The gasoline used for this particular experiment, containedO.75% benzene, 7.3% toluene, and 8.7% xylenes
325
-------
Table 3. Mass balance for the removal of diesel oil from soil by bioventing
(van Eyk and Vreeken, 1989b)
RESIDUAL DIESEL OIL AFTER 50 WEEKS
REMOVED BY BIODEGRADATION
TOTAL ACCOUNTED FOR
APPLIED
UNACCOUNTED FOR
BIODEGRADATION RATE:
8 MG/KG/DAY
84kg
32kg
116kg
125kg
9kg
(7%)
Table 4. Results of borings Bl to B4 with respect to mineral oil and BTX in mg/kg.
Component
Benzene
Toluene
Ethylbenzene
Xylenes
CH-total
Mineral oil
Bl
0.0-3.0
<.5
<.5
<.5
<.5
<50
<100
B2
0.0-3.0
130
860
280
2100
6900
15000
B3
0.0-3.0
39
250
96
740
2300
1300
B4
1.0-3.0
5.4
130
60
430
1300
320
326
-------
Table 5. Analysis results of handborings carried out under the garage floor in mg/kg.
Component
Benzene
Toluene
Ethyl-
benzene
Xylenes
Mineral oil
Boring number and
No. 11
1.0-1.5
0.15
0.56
0.09
0.75
<20
depth below
No. 12
1.5-2.0
3.7
5.6
1.1
4.9
<20
ground surface
No. 13
1.0-1.3
28
270
81
460
850
Table 6. Results of BTEX analyses for groundwater samples.
Component
ng/i
Benzene
Toluene
Ethylbenzene
Xylenes
CH-total
Monitoring well
noil*
40000
85000
8400
58000
200000
no36b
Groundwater sample probe number
and depth below GS
GW2
9.0
280 3.2
1400 1.7
150 < 0.5
980 0.8
3300 < 20
GW3
9.5
<0.2
<0.5
<0.5
<0.5
<20
GW4
7.5
<0.2
<0.5
<0.5
<0.5
<20
GW5
9.5
<0.2
<0.5
<0.5
<0.5
<20
a -filter at 1.60- 4.10 -GS
b- filter at 9.10- 11.10 -GS
327
-------
Table 7. Hydrocarbons removed by venting
WEEK
0
1
2
3
7
14
30
38
43
48
56
FLUX
m3/hr
19.2
24.8
32.0
35.0
29.8
33.0
30.2
47.7
39.0
41.5
39.8
AIR CONCENTRATION in mg/L
BENZENE
2.37
0.44
0.16
0.12
0.09
0.10
0.034
0.0423
0.016
0.006
0.007
TOLUENE
4.71
1.64
1.00
0.64
0.37
0.41
0.21
0.169
0.103
0.054
0.008
XYLENE
1.84
1.23
0.87
0.50
0.52
0.45
0.41
0.296
0.22
0.196
0.075
TOTAL HC
33.28
9.32
5.11
3.88
3.00
2.90
1.60
1.69
1.14
0.987
0.354
TOTAL HC
REMOVED
IN KG
117
190
227
253
331
436
573
677
718
75
77
Table 8. Deep ground water cleanup. Influent concentrations in mg/1 filter at 11 m - GS
DATE
10/12/90
11/13/90
12/11/90
1/14/91
3/5/91
4/3/91
5/2/91
MINERAL
OIL
53
<50
<50
97
<50
<100
<100
BENZENE
<0.2
<0.2
<0.2
<0.2
<0.2
<0.2
<0.2
TOLUENE
<0.5
<0.5
<0.5
<0.5
<0.5
<0.5
<0.5
XYLENE
<0.5
<0.5
<0.5
<0.5
<0.5
<0.5
<0.5
328
-------
Table 9. Shallow ground water cleanup. Influent concentrations in mg/1 Filter at 4 m BGS
B- value
Disposal Norm
10/12/90
11/13/90
12/10/90
1/14/91
3/5/91
4/4/91
5/1/91
5/30/91
6/27/91
7/22/91
8/2/91
8/6/91
9/6/91
9/11/91
9/17/91
Volatile
Aromatic s
30
100
5900
22000
16000
3800
2000
3200
—
—
348
125
300
300
—
270
120
Benzene
1
1600
2600
1400
510
200
300
—
—
40
16
32
33
—
16
14
Toluene
15
2100
7800
5300
1200
570
970
—
—
110
20
98
99
—
60
34
Ethyl-
benzene
20
1250
300
63
150
—
—
9
--
18
21
--
20
7
Xylene
20
2100
11000
8000
1750
1150
1760
—
—
189
87
143
141
—
54
38
Mineral
oil
200
10000
6900
91000
72000
88000
9100
2800
—
—
--
—
—
150
—
210
—
Chemical
Oxygen
Demand
116
60
46
41
31
31
35
31
20
29
—
—
- Background values
Table 10. Removal of PCAs from soil after a bioventing period of 18 months
(van Eyk and Vreeken, 1989b)
SAMPLING DEPTH AT 1.25- 1.50 M-GS
COMPONENT
NAPHTHALENE
METHYL-NAPHTHALENE
DI-METHYL-NAPHTHALENE
ANTHRACENE/PHENANTHRENE
FLUORANTHENE
PYRENE
TOTAL PCAs
CONCENTRATION
<0.1*
29
67
35
11
9
(MG/KG)
(8.8)**
(83)
(162)
(20)
(2.4)
(4.7)
DIESELOIL
%
0.092
0.87
1.7
0.21
0.025
0.049
3.5
* Determined
** Calculated on the basis of total oil after a bioventing period of 18 months.
(9535 mg/kgds, which amounts to approximately 46% of the concentration at t.
329
-------
GARAGE
»12 d11
-13
AUTOSHOP
-------
contour soil
contamination
ROAD
contour groundwater
"contamination
Figure 2. Contours for soil and ground water contamination.
331
-------
0 05m
1 70m
Pa = 0.5m we
T
0 75
plastic sheet
capillary zone
j ^ £. 4 ' * *• ^
' groundwatertable
Figure 3. Computational scheme to calculate air transport in soil.
1APAGE
pipeline
pavement
boundery impermeable
plume delineation qroundwater
AQ pumping well filter 00,2m(9-11m -gs
BO pumping well filter 00,2m(l.S-4m -gs
1t/m2'« • air pumping wells
contamination
Figure 4. Plan view of retail gasoline station outlining the installation of the venting system,
pumping wells, cleanup and monitoring system.
332
-------
liquid tight pavement
sand
I... .1
I... I
• to, air .e'xtractio'n pump "
'•• .'„ sang1 J- cement, •
contaminated soil
Figure 5. Outlining of the design of air extraction wells under the
liquid tight pavement.
contaminated
air
influent
water
aerator/oil-water
separator
air effluent
water
4 effluent
air
biofilter
for air
Figure 6. Schematic overview of the system for treatment of polluted
air and ground water.
333
-------
2500
2000
1500
1000
500
CO 2 PRODUCTION IN ppm
24.07.91
10 20 30
—f*- weeks after start
40
50
60
70
Figure 7.
GRAPHIC REPRENSENTATION OF
BIOVENTING RESULTS
(CUMULATIVE)
400
300
200
100
10 20 30
—*- Weeks after start
40
50
60
70
Figure 8.
334
• U.S.CX3VERNMENTPWNTINCOmC£:i992 -e
------- |