•  EPA/540/R-95/513
                                             July 1995
        REVIEW OF MATHEMATICAL MODELING
  FOR EVALUATING SOIL VAPOR EXTRACTION SYSTEMS
                          by

                     David L. Jordan
                    James W. Mercer
                    Robert M. Cohen

                     GeoTrahs, Inc.
                  Sterling, Virginia 20166
                 Contract No. 68-C2-0108
                     Project Officer

                     Chi-Yuan Fan
        Land Remediation & Pollution Control Division
        National Risk Management ResearchLaboratory
                Edison, New Jersey. 08837
NATIONAL RISK MANAGEMENT RESEARCH LABORATORY
       OFFICE OF RESEARCH AND DEVELOPMENT
      U.S. ENVIRONMENTAL PROTECTION AGENCY
               CINCINNATI, OHIO 45628
                                         Printed on Recycled Paper

-------
                                       NOTICE

       The information in this document has been funded by the U.S. Environmental Protection
Agency (EPA) under Contract No. 68-C2-0108 to International Technology Corporation and its
subcontractor GeoTrans, Inc.  It has been subjected to the Agency's peer review and
administrative review, and has been approved for publication. Mention of trade names or
commercial products does not constitute endorsement or recommendation for use.
                                         n

-------
                                      FOREWORD
       The U.S. Environmental Protection Agency is charged by Congress with protecting the
Nation's land, air, and water resources. Under a mandate of national environmental laws, the Agency
strives to formulate and implement actions leading to a compatible balance between human activities
and the ability of natural systems to support and nurture life.  To meet these mandates, EPA's
research program is providing data and technical support for solving environmental problems today
and building a science knowledge  base necessary to manage our ecological resources wisely,
understand how pollutants affect our health, and prevent or reduce environmental risks in the future.

       The National Risk Management Research Laboratory is the Agency's center for investigation
of technological and management approaches for reducing risks from threats to human health and the
environment.  The focus of the Laboratory's research program is on methods for the prevention and
control of pollution to air, land, water  and subsurface resources; protection of water quality in public
water systems ; remediation of contaminated sites and groundwater; and prevention and control of
indoor air pollution. The goal of this  research effort is to catalyze development and implementation
of innovative,  cost-effective environmental  technologies;  develop scientific  and  engineering
information needed by EPA to support regulatory and policy decisions; and provide technical support
and information transfer  to ensure effective implementation  of environmental regulations and
strategies.

       This publication has been produced as part of the Laboratory's strategic long-term research
plan.  It is published and made available by EPA's Office of Research and Development to assist the
user community and to link researchers with their clients.
                                                E. Timothy Oppelt, Director
                                                National RiskManagementResearchLaboratory
                                            111

-------
                                     ABSTRACT

       Soil vapor extraction (SVE) is a commonly used remedial technology at sites
contaminated with volatile organic compounds (VOCs) such as chlorinated solvents and
hydrocarbon fuels. Modeling tools are available to help evaluate the feasibility, design, and
performance of SVE systems. These models provide a means by which to quantify some of the
important SVE operating processes.

       Modeling can provide estimated answers for numerous questions concerning the
feasibility and use of SVE.  Screening models can be used in conjunction with site
characterization data and best professional judgment to determine the potential feasibility of SVE
at a contaminated site.  Flow and transport models can then be used to enhance the system design
process and estimate performance.

       A number of screening, air flow, and compositional flow and transport models are
available commercially and in the public domain. For screening, these models include the
Hyperventilate and VENTING codes, as well as analytical solutions. Air flow models available
at this time include AIRFLOW, CSUGAS, and AIR3D. VENT2D/VENT3D, a compositional
flow and transport code, can be used to simulate the transport and removal of complex
contaminant mixtures via SVE.

       The selection and application of any model will ultimately be the responsibility of the
model user. This document provides information on SVE model selection, data requirements,
design, and application, describes the equations governing flow and transport processes, and
highlights model limitations.

       This report was submitted in fulfillment of Contract No. 68-C2-0118 by International
Technology and its subcontractor GeoTrans, Inc. under the sponsorship of the U.S.
Environmental Protection Agency.  This report covers a period from April 1994 to April 1995,
and work was completed as of April 1995.
                                         IV

-------
r
                                                CONTENTS
           Section
Page
           Disclaimer	  ii
           Foreword	  iii
           Abstract	 iv
           Figures	ix
           Tables	xv
           Symbols	  xvii
           Acknowledgements	xx

           1    Introduction	1
                1.1    SVE Modeling Issues			 ..3
                1.2    Objectives	4
                1.3    Report Organization	 5

           2    Overview of SVE Modeling	 7
                2.1    The Modeling Process	7
                      2.1.1   Mathematical Model Development	7
                      2.1.2   Determining Model Need	8
                      2.1.3   Establishing Model Goals	10
                     .2.1.4   Model Calibration, Sensitivity Analysis, and Prediction	10
                      2.1.5   Model Documentation and Quality Assurance	12
                2.2    SVE Questions that Modeling Can Address	 13
                      2.2.1   Is the air permeability at the site high enough to allow effective use
                             of SVE?	 ,16
                      2.2.2   Are the compounds present at the site volatile enough to be removed
                             by SVE?'	,	19
                      2.2.3   How quickly will vapor concentrations drop to levels that do not
                             yield adequate removal?	 19
                      2.2.4   What effect does well placement have on air flow and mass transport?  .. 19
                      2.2.5  .How does changing the extraction well screened interval change the
                             air flow field?	,	21
                      2.2.6   What is  the design radius of influence of each well, and how is it
                             affected by geologic conditions and air flow rate?	23
                      2.2.7   How will air injection affect air flow to extraction wells?	23
                      2.2.8   How do the air flow fields differ for systems using drains or vertical
                             wells?	26
                      2.2.9   What are the effects of boundary conditions (for example, surface
                             capping)?	26
                      2.2.10  What residual semivolatile or nonvolatile contaminants will be left
                             after volatiles are removed by SVE?	28

-------
Section
                                      CONTENTS
           2.2.11  What cleanup levels can be achieved, and will these be within
                   regulatory limits?  	34
           2.2.12  What is the estimated cleanup time?	.34
           2.2.13  What monitoring is necessary to verify that SVE is working?	35

3    Flow and Transport Processes and Equations	'.	37
     3.1   Advection and Diffusion during SVE  	38
     3.2   Basic Equations 	            42
           3.2.1   Equations of Air Flow	'.	42
           3.2.2   Flow Equation Solution	 48
           3.2.3   Equations of Vapor-Phase Contaminant Transport  	51
           3.2.4   Transport Equation Solution	,	54
           3.2.5   Equilibrium Partitioning and Mass-Balance Approach  	55

4    Model Data Requirements  			                60
     4.1   Air Flow Model Parameters	60
           4.1.1    Porosity	t	60
           4.1.2   Volumetric Fluid Content	63
           4.1.3    Fluid Saturation	.. 63
           4.1.4   Capillary Pressure Relationships  	64
           4.1.5    Permeability and Relative Permeability	68
           4.1.6    Gas Temperature	yg
           4.1.7    Gas Molecular Weight	82
           4.1.8    Gas Density	83
           4.1.9    Gas Viscosity   	34
           4.1.10   Spatial Distribution of Media Properties Affecting Air Flow  	87
     4.2   Vapor Contaminant Transport Model Parameters  	-.	87
           4.2.1    Mechanical Dispersion	90
           4.2.2    Molecular Diffusion	       91
           4.2.3    Vapor Pressure	94
           4.2.4    Henry's Law Constant	  95
           4.2.5   Aqueous Solubility	97
           4.2.6    Sorption	,..	99
           4.2.7   Biodegradation	         JQ5
           4.2.8    Contaminant Composition	105
     4.3    Model Boundary and Initial Conditions	107
     4.4   Model Grid Design	108
     4.5    Data Limitations	109
                                          VI

-------
                                     CONTENTS
Section
Page
5    Model Selection Process	112
     5.1   Code Selection Criteria	112
     5.2   What Type of Model Should Be Used?	116
     5.3   Using Screening Models  	.	118
          5.3.1   When to Use a Screening Model	118
          5.3.2   What Type of Results to Expect	118
          5.3.3   Using Screening Models for Initial Site Characterization .. '.	 118
          5.3.4   Available Codes			 120
     5.4   Using Air Flow Models	120
          5.4.1   When to Use an Air Flow Model	122
          5.4.2   Level of Complexity	122
          5.4.3   What Type of Results to Expect	122
          5.4.4   Available Codes	123
     5.5   Using Compositional Flow and Transport Models	123
          5.5.1   When to Use a Compositional Flow and Transport Model 	123
          5.5.2   Level of Complexity	124
          5.5.3   What Type of Results to Expect	.124
          5.5.4   Available Codes	124
     5.6   Other Models	124
          5.6.1   Models that Account for Nonideal Conditions	125
          5.6.2   Air Sparging and Bioventing Models	 125
          5.6.3   Complex Multiphase Flow and Transport Codes	126
     5.7   What Site Characterization Data Requirements are Associated with
          Different Types of Models?	128
     5.8   What Level of User Knowledge is Required for Different Model Types?	128
     5.9   Within What Time and Budgetary Constraints Must the Modeling
          Exercise be Completed?	129
     5.10 Is the Main Obj ective Data Organization and Analysis Rather
          than Modeling?			129

6    Evaluation of Readily Available SVE Models	132
     6.1   Data Organization Tools	132
          6.1.1   GIS for Data Organization and Graphical Analysis	132
          6.1.2   Linking GIS and Simulation Models  	135
     6.2   Screening Models	138
          6.2.1   Hyperventilate	138
          6.2.2   VENTING...	141
          6.2.3   Analytical Solutions	143
     6.3   Air Flow Models	143
                                         vn

-------
 Section
                                    CONTENTS
           6.3.1   AIRFLOW	                           144
           6.3.2   CSUGAS	       	147
           6.3.3   AIR3D	148
      6.4   Compositional Flow and Transport Models .                        	157
           6.4.1   VENT2D	.'.'.'.'.'.'.'.'".'.']].'.'.'"" 152

 7    Example Applications	                J55
      7.1   Predicting Flow Rates from Permeability with Hyperventilate	       155
      7.2   Using AIRFLOW to Assess SVE Well Screen Interval	'.'.'.'.'.'.'.'.'.'.'. 157
      7.3   Estimating Contaminant Removal Rates Using VENT2D	 . 157
      7.4   Assessing SVE Impact on NAPL Distribution Using VENT2D	161
      7.5   Spreadsheet Application to Estimate the kr/kz Ratio	164
      7.6   Pneumatic Test Analysis Using Aquifer Test Software	164
      7.7   Comparing Constituent Removal Rates Using VENTING	 166
      7.8   Evaluating Surface Boundary Effects Using AIRFLOW	..... 166
      7.9   Simulating Pressures Induced by SVE Using AIR3D	 168

 8    SVE Simulation Case Studies	                174
     8.1   Union Chemical Company Superfund Site, South Hope, Maine	           174
     8.2   California UST Site  	' " "	178
           8.2.1   Hyperventilate Screening Application		.178
           8.2.2   Flow and Transport Model (VENT3D) Application	! 188
     8.3   Verona Well Field Superfund Site, Battle Creek, Michigan	 '. 201

9    References	                  209

Appendix A	                    225

Glossary	                  256
                                       vm

-------
                                  LIST OF FIGURES
                                                                                  Page

1.    Schematic diagram showing the components of a typical S VE system (from
     USEPA, 1991a)	2
2.    Trial-and-error model calibration procedure. The conceptual site model is
     converted to a numerical model and calibration goals are determined. Model
     results are then compared to calibration goals.  If the model results are not
     acceptable, parameter values are adjusted and the model is re-run (after
     Anderson and Woessner, 1992)	11
3.    Relationship of relative air permeability to relative water permeability (USEPA,
     1991a)	17
4.    Decision nomograph for evaluation of relative  volatility and potential for
     cleanup via SVE.  Units of vapor pressure are mm Hg (USEPA, 1994)	18
5.    Placement of the well screen affects how air flows to the extraction well (from
     USEPA, 1991a)	20
6.    Placement of an injection well to eliminate air  flow stagnation zones (from
     USEPA, 1991a)	'.	22
7.    Surface boundary  effects on radius of influence (from USEPA, 199 la)	24
8.    Air injection can be used to (a) reduce short-circuiting of air flow between the
     ground surface and an extraction well and (b) create a pressure barrier to limit
     the ROI of an SVE well.	 -.	 25
9.    Common configurations for SVE wells and drains (from USEPA, 199la)	27
10.  Chromatograms of (a) regular gasoline, (b) unleaded gasoline, (c) kerosene
     heating fuel, and (d) diesel fuel. Note that the  more volatile fuels (regular
     and unleaded gasoline) contain a significant, percentage of compounds that
     elute early during  the GC scan (after Preslo and Stoner, 1991)	31
11.  Chromatograms of gasoline before and after venting. Note the decrease in
     concentration for many of the lighter compounds (after Baehr et al., 1989)	32
12.  Comparison of relative removal rates of compounds of different volatility
     from a fuel mixture. Note that light compounds such as benzene are removed
     fairly quickly, while semivolatile compounds such as naphthalene are fairly
     resistant to SVE (results are from the VENTING model)	 33
13.  Illustration of the  concept of stages of VOC removal during SVE. Stage I is
     advection-dominated and characterized by rapid removal of VOCs. Stage II
     is a transition period from an advection-dominated system to a diffusion-
     dominated system. Stage III is a diffusion-dominated state characterized
     by mass removal rates which asymptotically approach some limiting value
     (after Hiller and Gudemann, 1989)	40
14.  Conceptual mathematical modeling results that exhibit Stage I behavior
     during early time  and progress to diffusion-limited Stage III behavior during
     late time for two different simulated SVE rates (from Silka et al., 1989)	41
                                           IX

-------
 16.
 17.
 18.
                                    LIST OF FIGURES
 15.   Comparison of simulated TPH mass-in-place versus SVE duration in homogeneous
      and layered systems. Due to slow contaminant diffusion from the low-permeability
      layer, it takes much longer for SVE to reach the 90% cleanup goal in the layered
      system (from Benson et al., 1993)	            43
      An exponential decay curve fit to early time SVE data may underestimate later	
      concentrations (or mass removal rates) where contaminant transport is limited by diffusion
      (from Hutton, 1990)	                  44
      Typical curves showing the relationship between capillary pressure and volumetric	
      water content for various USDA soil classes (from USEPA, 1991c)	  65
      USDA soil textural triangle	'              	66
19.   Saturated hydraulic conductivity (K) in cm/hr as a function of grain size
      distribution. Intrinsic permeability (k) in darcies can be estimated by:
      k = 0.29 K where K is in cm/hr	           71
20.   In a two fluid phase system with NAPL and water, the water is typically the
      wetting fluid and occupies the smallest pores (a); in a three fluid phase system
     with NAPL, water, and air, the order of solid wetting is typically water, NAPL
     and then air (b) (after Parker, 1989)	'      '         73
     Air-water relative permeability (k,-^) curves calculated for five soil types
     determined using (a) Equations (56) and (57) (Parker et al.,  1987) and the
     van Genuchten parameters given in Table 7; and (b) the Brooks-Corey
     Equations (61) and (62) and pore-size distribution indices given in Table 8 (b)	75
     Air-water relative permeability curves calculated using several relative
     permeability relationships (a); and measurements of relative air permeability
     versus saturation for a sand containing different mixtures of gasoline and water
     (b) (after Stylianou and DeVantier, 1994)	              77
     Relationships between temperature and (a) the density of air and water, and
     (b) the dynamic viscosity of air and water	      79
     At depth, the temperature of soil gas is relatively stable throughout the year	80
     Average temperature of shallow groundwater in the United States (after USGS
     Water-Supply Paper 520F)	                 81
     Representation of equilibrium partitioning of chemical mass between phases  	88
     Effective gas diffusion coefficient as a function of volumetric air content
     calculating using the Millington-Quirk equation	                 93
     For several important classes of organic compounds, ranges of (a) saturation
     vapor pressure at 25° C,  (b) Henry's law constants, (c) aqueous solubility at 25° C,
     and (d) octanol-water partition constants (from Schwarzenbach et al., 1993)	'	96
     Results from Johnson et al. (1990c) that illustrate the concentration ranges over
     which Henry's law and Raoult's law are valid for a particular case study	98
     Temperature dependence of vapor pressure and aqueous solubility for several
     representative organic compounds (from Schwarzenbach et al., 1993)	ioo
 21
 22.
23.

24.
25.

26.
27.

28.
29.
30.

-------
                                  LIST OF FIGURES
31.   Illustration of VOC adsorption under different moisture conditions (after Valsaraj
     and Thibodeaux, 1988)	. . .	104
32.   Soil uptake of TCE as a function of relative humidity (after Smith et al., 1990)	106
33.   Areal grid (a) and vertical grid (b) after Istok (1989):  grid cells should coincide
     with point sources and sinks and be placed along boundaries (a); grid spacing
     should be reduced in .areas where .significant flow or concentration gradients
     are expected and soil type boundaries should coincide with grid boundaries (b).  ...... 110
34.   Decision flowchart for selecting which class of model (screening, air flow,
     compositional flow, and transport) to use	117
35.   Estimates of cleanup time and zone of capture produced from confined
     (Theis  solution) versus leaky analytical models. Depending on the spatial
     relationship between extraction wells and the zone of contamination,
     confined models may underestimate cleanup times and overestimate zones of
     capture because they assume no short-circuiting of air from the surface
     (after Beckett and Huntley, 1994).	 119
36.   Schematic diagram of the process of intersecting layers using a GIS.  The
     intersection process allows the user to  determine those areas which conform
     to a selected group of constraints (e.g., VOC contamination, high permeability, significant
     vadosezone). The area with most promise for cleanup via SVE is
     defined by the intersection of those areas characterized by VOG contamination,
     high permeability, and a significant vadose zone	131
37.   The concept of layering in a GIS (from ESRI, 1992)	137
38.   Example application of a GIS to identify and highlight graphically those areas       ;
     on a site where NAPL is likely to be located. ..	 139
39.   Radial-symmetric geometry and grid for the AIRFLOW model	 145
40.   Example output (pressure contours) from the AIR3D code	,	151
41.   Geometry and boundary configuration for the VENT2D model (from
     Benson, 1994)	 154
42.   Card eight from the Hyper Ventilate soil venting stack is used to calculate a range
     of well flow rates based on permeability, user-defined radius of influence, well
     radius, and screened interval or flow zone thickness	156
43.  Model domain geometry and boundary conditions input to AIRFLOW for the
     example described in Section 7.2	158
44.  AIRFLOW results showing  how air flow and flushing of the lower vadose zone
     (where LNAPL tends to accumulate within and above the capillary fringe)
     increases with depth of screen placement:  (a) uppermost screen placement,
     (b) intermediate placement,  and (c) placement just above the water table.  ........... 159
45.  VENT2D results showing simulated VOC concentration and removal rate in
     extracted soil gas during the first 14 days of SVE operation for the example
     described in Section 7.3	160
                                           XI

-------
                                   LIST OF FIGURES
 46.  VENT2D results showing simulated concentration of total VOCs in soil with
      time at the model cell where the extraction well is located for the example
      described in Section 7.3.  Note how the concentration trends asymptotically
      toward a limiting value	         162
 47.  VENT2D results showing simulated total VOC concentrations and extent of
      NAPL presence in soil (a) before S VE begins and (b) after 42 days of vapor
      extraction for the problem described in Section 7.4. The contours are blocky
      due to the coarseness of the finite-difference model grid	163
 48.  Results from a spreadsheet (CRTC,  1992) solution of the analytical method of
      Shan et al. (1992) showing a fit between simulated and measured vacuum-distance
      data for a k^k, ratio of 0.5	       155
 49.  Results from Beckett and Huntley (1994) showing that, for many field air
      pumping tests, the Hantush-Jacob (1955) leaky solution provides a better
      fit to field data than the confined solution of Theis (1935). The curves
      were fit using the AQTESOLV (Geraghty and Miller, 1989)	167
 50.  Model geometry and boundary conditions input to AIRFLOW to simulate
      the example described in Section 7.8 (with the top of the  model open to the
      atmosphere)	       159
 51.   Vapor pathlines and pressure distribution simulated using AIRFLOW showing
      significant leakage from the surface due to the open ground surface boundary
      condition	       17Q
 52.   Vapor pathlines and pressure distribution simulated using AIRFLOW for the
      confined case where the ground surface is modeled as an  impermeable, no-flow boundary.
      Note that flow is essentially horizontal towards the  extraction well,
      and that the radius of influence extends beyond that simulated for the leaky
      case shown in Figure 51	171
 53.   Results of AIR3D simulation showing a plan view of the  pressure
      distribution resulting from a horizontal SVE drain installed at depth as
      described in Section 7.9	      172
 54.   Results of AIR3D simulation showing a cross-sectional view of the pressure
      distribution resulting from a horizontal SVE drain installed at depth as
     described in Section 7.9. Each layer is two meters thick	 173
55.   Comparison of field-measured pressure data during an air pumping test to
     the results from a radial-symmetric air flow model (from Balsam, 1992). 	176
56.   Boundary conditions for a radial-symmetric model used to simulate an air
     pumping test at a single extraction well	177
57.   Sample site data. Values for TPH concentrations in soil (mg/kg) are posted
     along borings (from Johnson et al., 1992)		179
58.   Card 10 from the soil venting stack	181
59.   Card 12 from the soil venting stack	181
                                          XII

-------
                                  LIST OF FIGURES
                                                                                 Page
60.  Card 13 from the soil venting stack.	182
61.  Card 16 from the soil venting stack	186
62.  Card 17 from the soil venting stack	186
63.  Card 18 from the soil venting stack.	187
64.  Card APS of the air permeability stack.	187
65.  Card SD2 of the system design stack	.'	 189
66.  Card SD4 of the system design stack	189
67.  Total contaminant removed (expressed as an equivalent volume of gasoline)
     and mass removal rate over time for the California UST case study site (from
     Johnsonetal., 1992)	.. 190
68.  Extracted vapor composition over time (from Johnson et al.,  1992)	191
69.  Costa Mesa service station layout (Johnson et al., 1992) and initial concentrations
     used in the numerical simulations. TPH measurements were lognormally
     distributed, so the initial TPH grid in soil was generated by kriging of the log
     TPH values from 40 to 48 feet BGS.	193
70.  Simulated and measured decline of TPH concentrations in vapor extracted from
     the Costa Mesa site. Measured values are represented by connected points.
     Simulated values are represented by large open circles. The vapor stream was
     measured on site by on FID-equipped recorder, and by laboratory GC methods
     on collected samples. The extraction rate was increased after 275 days from 10
     to 30 scfin. Simulated results shown before 275 days are from a lower-flow
     simulation, while later results are from a higher flow simulation that uses an
     average flow rate over the life of the SVE project	 194
71.  Residual soil masses of selected compounds in the 25-scfm model.  Plotted
     values represent summed mass of the compound in all phases throughout the
     model domain.	196
72.  Residual soil masses of selected compounds in the 10-scfrn model.  Plotted
     values represent summed mass of the compound in all phases throughout the
     model domain	197
73.  A comparison of the simulated (upper figure) and measured (lower figure)
     change in the composition of extracted vapor. The concentration of all
     compounds in vapor extracted during the first 275 days are added according
     to boiling point ranges.  The percentage of each range is given by its width
     on the graph. Johnson et al. (1992) suggest that the decline of the percentage
     of more volatile range is indicative of mass removal, whereas the TPH
     concentration decline may result from nonideal conditions. In light of the
     declining TPH concentrations, a comparison of the two plots shown here
     indicates that some fraction of the initial TPH mass is located outside of the
     near-well advection-dominated zone	198
                                          xin

-------
                                 LIST OF FIGURES
74.
75.
78.
79.
80.
                                                                               Pas
                                                                                200
     Simulated residual benzene (mg/kg) in soil after the initial 275 days of SVE
     at a steady rate of 10 scfm ............... . .................
     Simulated residual TPH (mg/kg) in soil after 900 days of SVE at a steady '"'
     rate of 25 scfm ..... . ..................................
76.  Total mass of VOCs removed from the subsurface over time.  Note how the plot is reaching
     a limiting, asymptotic value (from USEPA, 1991d) ............... ____ ........... 202
77.  VOC removal rate versus time. Note how the removal rate is high during early
     time, but diminishes during later time to a limiting value ...................       203
     Chromatograms of extracted gas from the TCE/PCE column test. Note how
     the more volatile TCE is removed from the system more quickly (from
     USEPA, 1991d) ....................................... .  .....     ....... 205
     Raoult's law effects on the TCE/PCE system. As TCE is removed from the     ' .....
     system during early time, the mole fraction of PCE increases, causing an
     increase hi concentration. The concentration of PCE in extracted gas begins
     to stabilize after the TCE has been removed from the system and the mole
     fraction of PCE has stabilized. The concentration of PCE then decreases as
     it is removed from the system (from USEPA, 1991d) ............................ 206
     Results of a column experiment where a third constituent, decane, was added
     to the TCE/PCE mixture. The addition of a third constituent further reduced
     the mole fractions of TCE and PCE, and hence their vapor pressures were
     reduced, and volatilization and removal from the system occurred more
     slowly (from USEPA, 1991d) ....................... . ....................... 207
                                       xiv

-------
                                   LIST OF TABLES
                                                                                   age
1.    Location of major topics in the document	-.	6
2.    Summary of important SVE questions and some potential uses of modeling	14
3.    Composition of fresh and weathered gasolines (from USEPA, 1991 a; after
     Johnson et al., 1989).	29
4a.  One atmosphere, as represented in several commonly used unit systems	.52
4b.  Pressure unit conversion factors.	 52
5.    Listing of SVE model parameters	;	61
6.    Selected physical properties of rocks and soil (after USEPA, 1990a; Freeze
     and Cherry, 1979; Krinshnayya et al., 1988; Morris and Johnson, 1967).
     Additional soil porosity data (and dry bulk density values using Equation 97)
     are provided in Tables 7 and 8 (8 = 8W in Table 7)	62
7.    Mean values of the van Genuchten soil moisture retention and relative
     permeability parameters for the USDA soil types (modified from Carsel
     and Parrish, 1988).  Descriptive statistics (e.g., standard deviation and
     coefficient of variation) for each parameter listed below are provided by
     Carsel and Parrish (1988)	67
8.    Brooks-Corey soil moisture retention and relative permeability parameters
     for the USDA soil types (modified from Rawls et al., 1985).	69
9.    Calculated vapor pressure, saturated vapor concentration, and total gas
     density values for some common volatile organic contaminants
     (from Falta et al., 1989)	.'..	85
10.  Approximate viscosities for various gas components (from Weast, 1974)	 89
11.  Fraction of organiccarbon (foc) values measured in various soils (from
     Mercer and Waddell, 1993)	102
12.  Worksheet for evaluating the feasibility of soil venting (after USEPA, 1990c)	113
13.  A summary of the basic issues to be addressed in planning for the
     development of a vapor extraction system (from Bedient  et al., 1993;
     after Johnson et al., 1990b)	114
14.  A sampling of analytical solutions available for analysis of SVE	121
15.  General design and application considerations appropriate for conventual
     versus bioventing SVE systems (modified from Dupont,  1993 based on
     USEPA, 1994). ..	 . . .  . . 127
16.  Summary of the screening, air flow, and compositional flow and transport
     codes that were evaluated	133
17.  Example data types that can be utilized in a GIS	136
18.  General categories of input and output data for the Hyper Ventilate
     screening model	 141
19.  General classes of input and output data for the VENTING screening model.	142
20.  General classes of input and output data for the AIRFLOW model	 146
                                          xv

-------
                                  LIST OF TABLES
21.
22.
23.

24.
General classes of input data and output features for the CSUGAS air flow model.
General classes of input data and output features for the AIR3D air flow model.
General classes of input data and output features of the VENT2D compositional
flow and transport model	
Example boiling point distribution for gasoline (from USEPA, 1993)	
..148
..149
  153
  184
                                       xvi

-------
                                       SYMBOLS
a{
B,
Q
C
C
P
^
"ge
rfgi
foe
g
H
hg
K
Ka
Kd
KH
KH
K00
k
^gx
kx
kz
Ma
M,
M0
Mt
M,
     gyy '"•gzz
  wi
m
aqueous phase activity coefficient of compound i [dimensionless]
rate of degradation [M/T]
mass basis solid-adsorbed concentration [ M/M]       ,   ,
mass basis solid-adsorbed concentration of compound i [ M/M]
vapor phase concentration [M/L3]
vapor phase concentration of compound i [M/L3]
saturated vapor phase concentration of compound i [M/L3]
equilibrium molar concentration of compound i in the air entering an SVE
well [mol/L3]                                  '      ';-' '    ^   '
concentration of compound i [M/L3]                   •••••.,
concentration of compound i in NAPL [M/L3]
dissolved (soil water) concentration [M/L3]              ' r '    '     . ^
aqueous solubility of a pure compound [M/L3]                 .".'.',.'
dissolved (soil water) concentration of compound i [M/L3].
molar concentration of compound i in the aqueous phase [mol/L3]
air-phase dispersion tensor [L2/T]
free-air molecular diffusion coefficient [L2/T]
effective free-air molecular diffusion coefficient [L2/T]
free-air molecular diffusion coefficient of compound i [L2/T]
fraction organic carbon content [dimensionless]
gravitational acceleration [» 981  cm/s2] [L/T2]
length of extraction well screen open to the unsaturated zone [L]
head of air [L]
saturated hydraulic conductivity  [L/T]
air conductivity [L/T]
linear solid/liquid distribution coefficient [L3/M]
Henry's law constant (atm-m3/mol) [M-L/mol]
dimensionless form of Henry's law constant [dimensionless]
organic carbon partition coefficient [ L3/M]
octanol/water partition coefficient [dimensionless]
soil permeability (unless otherwise noted) [L2]
effective gas permeability (unless otherwise noted) [L2]
diagonal components of air permeability tensor [L2]
horizontal soil permeability [L2]
vertical soil permeability [L2]
asymptotic  mass removal rate [M/T]
total number of moles of component i in soil [dimensionless]
initial mass removal rate [M/T]
mass removal rate [M/T]
total mass of component i in soil [M]
stratum thickness [L]
                                          xvn

-------
                                  SYMBOLS (continued)
 n
 P
  • Atm
  g\V
 R
 R
 RI
 Rw
 r
 w
 T
 t
 V.ir
 V
Yi
z
aT
T(t)
 number of components in a gas mixture [dimensionless]
 pressure deviation (vacuum) [M/L-T2]
 atmospheric pressure [M/L-T2]
 air pressure [M/L-T2]
 air pressure at an extraction well [M/L-T2]
 partial vapor pressure of component i [M/L-T2]
 saturated pure-component vapor pressure of component i [M/L-T2]
 volumetric flow rate of gas to a vapor well [L3/T]
 specific discharge vector for soil gas (Darcy velocity) [L/T]
 universal gas constant [0.08206 L-atm/K-mol] [ML2/molKT2]
 retardation coefficient for the ith component [dimensionless]
 radius of influence where pressure is PAtm [L]
 radius of SVE well [L]
 radial coordinate [L]
 pneumatic equivalent of specific storage [L'1]
 air saturation (volume of air / volume of voids) [L3/L3]
 NAPL saturation (volume of NAPL / volume of voids) [L3/L3]
 water saturation (volume of water / volume of voids) [L3/L3]]
 absolute temperature in soil [K]
 reference temperature in soil [K]
 time[T]
 average molar volume of gases in air (-20.1 cmVmol) [L3/mol]
 total volume of the contaminated soil zone [L3]
 average air velocity [L/T]
 average molar volume of compound i [L3/mol]
 molar volume [LVmol]
 interstitial gas velocity (vg = 3/0J [L/T]
 mole fraction of compound i in a NAPL phase [dimensionless]
 mole fraction of compound i in soil moisture [dimensionless]
 height above datum [L]
 mole fraction of compound i in the vapor phase [dimensionless]
 longitudinal dispersivity [L]
 transverse dispersivity [L]
 sources and sinks that may be functions of time [M/L3-T]
 Kronecker delta [dimensionless]
 an efficiency factor that can be used to account for nonequilibrium effects
 in mass removal during SVE [dimensionless]
total porosity [L3/L3]
effective porosity [L3/L3]
volumetric gas content (air-filled porosity) [L3/L3]
volumetric NAPL content (NAPL-filled porosity) [L3/L3]
                                         xvm

-------
Ug
^gi
H'gmix
Pb

Pg
PgAtm

Pp
Pw
O

T
tb
CO
COj
wv
v-
V
gmix
             SYMBOLS (continued)

mass-basis NAPL-occupied fraction of the pore space [M/M]
volumetric water content (water-filled porosity) [L3/L3]
residual water content [L3/L3]
air permeability tensor [L2]
the pore-size distribution index [dimensionless]
dynamic viscosity of the soil gas [M/L-T]
dynamic viscosity of gas mixture component i [M/L-T]
dynamic viscosity of a gas mixture [M/L-T]
dynamic viscosity of the soil gas at a reference temperature [M/L-T]
dynamic viscosity of water [M/L-T]
kinetic energy correction factor [dimensionless]
bulk density [M/L3]                  '
density of soil gas [M/L3]
air density at PAtm [M/L3]
particle density [M/L3]
density of water [M/L3]
surface tensipn [M/L-T2]
tortuosity [dimensionless]
Pg2 linearization (see Equation 10)
capillary pressure head [L]
air-entry (bubbling) capillary pressure head [L]
molecular weight of soil gas [M/mol]
molecular weight of a gas mixture [M/mol]
molecular weight of component i [M/mol]
molecular weight of water (18 g/mol) [M/mol]
divergence operator [1/L]
gradient operator [1/L]
                                           xix

-------
                              ACKNOWLEDGEMENTS

       This document was prepared for the U.S. Environmental Protection Agency (EPA) Office
of Research and Development Risk Reduction Engineering Laboratory (RREL) under Contract
No. 68-C2-0108 by GeoTrans, Inc. (GeoTrans) under subcontract to IT Corporation. IT and
GeoTrans appreciate the guidance and helpful suggestions provided by Mr. Anthony Tafuri,
RREL's UST Research Program Manager, Mr. Michael Gruenfeld, RREL's Project Officer,'and
Mr. Chi-Yuan Fan, RREL's Technical Project Manager for this Work Assignment.

       IT and GeoTrans acknowledge the documentation and software provided by the
following: Dr. Paul Johnson of Shell Development Company/Arizona State University for
Hyperventilate; Dr. Jack Parker of Environmental Systems & Technologies, Inc., for
VENTING; Mr. Thomas Franz of Waterloo Hydrogeologic Software for AIRFLOW; Mr. Roger
Claff of the American Petroleum Institute for AIR3D; Dr. James Warner of Colorado State
University for CSUGAS; and Mr. David Benson of the University of Nevada, Reno for
VENT2D/VENT3D. We also acknowledge Margarette Guerriero, EPA's Region V Office of
Superfund, for providing information on the Verona Wellfield Superfund site; Mr. Terry
Connelly, EPA's Region I Waste Management Division, for providing information on the Union
Chemical Superfund site; and Dr. Paul Johnson and Mr. David Benson for information on an
UST site hi California.

       The review comments of the technical committee and the contributions of the work group
participants during the initiation of the project are also acknowledged. In particular, we
acknowledge the following individuals:

       Gilberto Alvarez, EPA, OUST, Region V, Chicago, IL
       Arthur Baehr, U.S. Geological Survey, West Trenton, NJ
       Milovan Beljin, University of Cincinnati, Cincinnati, OH
       David Benson, University of Nevada, Reno, NV
       Lyle Bruce, Amoco Corp., Tulsa, OK
       Roger Claff, API, Washington, DC
       R. Ryan DuPont, Utah Water Research Lab, Logan, UT
       Bill Faggart, EPA, Region II, New York, NY
       Frank Freestone, EPA, RREL, Edison, NJ
       Rich Griffiths, EPA, RREL, Edison, NJ
       Patrick E. Haas, U.S. Air Force CEE/Technology Transfer  Division
       Jack Hwang, EPA Region III, Philadelphia, PA
       Paul Johnson, Shell Development, Houston, TX/ Arizona State University, Tempe, AZ
      Tod Johnson, University of Nevada, Las Vegas, NV
      Richard Koustas, EPA, RREL, Edison, NJ
      David K. Kreamer, University of Nevada, Las Vegas, NV
       George Mickelson, Wisconsin DNR, Madison, WI
      Bill Mills, Tetra Tech, Palo Alto, CA
                                        xx

-------
       I.P. Murarka, EPRI Environmental Division, Palo Alto, CA
       Jack Parker, Environmental Systems & Technologies, Inc., Blacksburg, VA
       Tom Pedersen, COM, Inc., Cambridge, MA
       Bill Peterson, EPA, Region VII, Kansas City, KS
       Jim Rumbaugh, Geraghty & Miller, Reston, VA
       Gregory D. Sayles, EPA, RREL, Cincinnati, OH
       Michelle Simon, RREL, Cincinnati, OH
       Mary K. Stinson, EPA, RREL, Edison, NJ
       Jim Stumbar, Foster Wheeler, Edison, NJ
       Dan Sullivan, EPA, RREL, Edison, NJ
       Mete Talimcioglui, Stevens Institute of Technology, Hoboken, NJ
       James W. Weaver, EPA, Robert S. Kerr Environmental Research Lab, Ada, OK
       David J. Wilson, Vanderbilt University, Department of Chemistry, Nashville, TN

       This document was produced under the direction of Mr. Robert Amick, IT's Program
Director. Mr. Roy Chaudet and Dr. Harvey Dove of IT Corporation were the Work Assignment
Leaders.  Mr. David Jordan, Dr. James Mercer, and Mr. Robert Cohen of GeoTrans were the
principal authors and were assisted by Ms. Karen Benoit, Mr. Clay Brown, and Mr. David Ward
of GeoTrans. Ms. Jan Goodman, assisted by Ms. Joanne Elkins, Ms. Ellen Barr, and Ms. Phyllis
Konikow, also of GeoTrans, prepared portions of the manuscript. Mr. Raj Setty, Ms. Brenda
Cole, and Mr. Pat Kelly of GeoTrans helped prepare graphics.
                                        xxi

-------

-------
                                      SECTION 1
                                    INTRODUCTION

       Soil vapor extraction (SVE) is used to remove volatile chemicals from the vadose zone.
Components of a typical SVE system are shown in Figure 1. Application of a vacuum at an
extraction well creates negative differential gas pressures, thereby inducing soil vapor flow to the
well. Volatilization of chemicals in vadose zone water and nonaqueous phase liquid (NAPL) is
enhanced by the movement of fresh air into the contaminated zone from nearby areas and/or air
injection vents. Due to its low viscosity, pore volume flushing with air in the vadose zone can be
maintained at high rates (compared to groundwater flushing rates achievable using pump-and-
treat systems). SVE may also be applied to remove light NAPL (LNAPL) floating on the water
table or entrained in the capillary fringe if the chemicals of concern have high vapor pressures
(e.g., benzene). Extracted soil vapor is treated, as necessary, and then typically released to the
atmosphere.

       SVE has been widely used in the United States to address vadose zone contamination. It
has been specified for remedial action at 13% of Superfund sites (USEPA, 1991e) and
approximately 7% of leaking underground storage tank sites (Trimblay, 1993)

       Bioventing and air-sparging are variants of SVE technology. Bioventing refers to aerobic
biodegradation that is enhanced by increasing oxygen delivery in the subsurface. Air flow
induced by SVE increases aeration, thus providing oxygen for use by indigenous microorganisms
that degrade contaminants.  Air sparging generally involves using wells to inject gas (typically
air) into the saturated zone below and within contaminated zones.  Ideally, dissolved, separate-
phase, and sorbed contaminants will partition into the injected air,  effectively creating an in-situ
air-stripping system. This can take place within a single-well (Herrling et al, 1990) or the
stripped contaminants can be transported in the gas phase to the vadose zone and collected by
SVE wells (Marley et al., 1992a). Under favorable conditions, such a system may eliminate the
need to pump, treat, and dispose of groundwater. Air sparging also causes and benefits from
bioventing. Additional variations of SVE include injecting hot air or steam to enhance
contaminant removal.

       SVE utility is limited by subsurface media and contaminant properties.  The method is
particularly applicable to volatile compounds (e.g., vapor pressure > 0.5 mm Hg and Henry's law
constant >10'3 atm-m3/mole) in relatively uniform, permeable media. In general, contaminant
removal efficiency declines with decreasing media permeability, decreasing contaminant
volatility, increasing organic carbon content (due to sorption of organic compounds), increasing
subsurface heterogeneity, increasing moisture content (which reduces air permeability), and,
decreasing Henry's law constant (lower air-water partitioning ratio).

       Contaminant removal may be enhanced as SVE proceeds because air permeability will
rise if the soil moisture content is reduced by vapor extraction. The rate of contaminant removal
usually declines with time, however, due to many of the same factors that cause "tailing" and

                                           1

-------
            .- SECONDARY
          >.*. & EMISSIONS
                            VACUUM
                     BALL   GAUGE
                     VALVE
                HEADER
 WATER COOLED
HEAT EXCHANGER
PRESSURE
 GAUGE
                                                AIR-WATER
                                               SEPARATOR
SURFACE SEAL
\
BENTONITE J*
CEMENT/
GROUT
v /
•w /
SAND
PACK

s
h.





s.
>
j:
\
<•-,

1 M=
^ci==y
J'i'iO1 '"s","s","i","y >yyy ,
0.020 SLOT
--"^SCREEN

                                                               STRAINER •
                                            SUBMERSIBLE
                                               PUMP
                                                                PUMP
                                                                	»»
PRESSURE
 RELEASE
  VALVE
                                                                          SILENCE
                                                                          MUFFLER
                                                                          BLOWER
                                                                  TO WATER
                                                               TREATMENTSYSTEM
Figure 1.      Schematic diagram showing the components of a typical SVE system (from
              USEPA, 1991a).

-------
"rebound" phenomena during groundwater pump-and-treat remediation (National Research
Council, 1994).  For example, bypassing occurs where air flows preferentially through permeable
portions of heterogeneous media. As contaminants are flushed from the permeable zones, the
rate of contaminant removal declines and is limited by the slow diffusion of contaminants out
from the bypassed media. Other chemical processes also limit contaminant removal.  Where
multiple compounds are present, lower molecular weight components will volatilize (and
dissolve) preferentially compared to higher molecular weight components. This selective
partitioning (weathering) causes the contaminant mixture to become increasingly enriched in
high molecular weight compounds. Thus, the contaminant removal rate declines as the mixture
becomes less volatile. Continued SVE system performance can be optimized by modifying
induced air flow rates and patterns and other operational parameters based on monitoring data.

       The SVE design process involves evaluation of contaminant chemistry and distribution,
subsurface media properties, and contaminant transport processes.  Design variables include: the
number, location, spacing, depth, and screen placement of wells and trenches for air extraction
and injection; extraction flow rates and vacuum specifications; possible surface amendments
(e.g., sealing to restrict air leakage); vacuum pump and/or blower criteria; and selection of
emissions control and treatment systems.

       At many sites, a trial-and-error or "seat of the pants" approach has been applied to SVE
system design, Within the past few years,  several mathematical models have become available
to evaluate subsurface gas flow and contaminant transport.  These models provide a quantitative
tool to assess the feasibility, design, and performance of SVE systems.

1.1     SVE MODELING ISSUES

      An understanding of the mechanisms controlling fate and transport processes and the site
characteristics that affect them is required to  effectively design a SVE system and interpret
system performance. The major processes that .affect SVE are advection, diffusion, partitioning,
and abiotic and biological transformations. Therefore, modeling problems include describing:

      *      The contaminant source and heterogeneous distribution;

             Advective transport, especially in heterogeneous media;

      •      Intergranular and intragranular diffusion and diffusion-limited processes, such as
             those associated with low permeability material (clay lenses) and diffusion into
             micropore channels (O.001 microns) within grains;

      •      Mass transfer associated with  sorption/desorption;

             Abiotic and biological transformations, as necessary; and

-------
       •      Fractionation of multicomponent compounds (such as gasoline) as lighter
              compounds volatilize faster, leaving a heavier, less volatile mixture.

       The ability to model contaminant transport in the subsurface is limited by a wide array of
 subsurface complexities, including heterogeneous permeability distributions, multicomponent
 mixtures, contaminant partitioning phenomena, and multiphase permeability relationships. A
 particularly formidable issue facing SVE modelers is that of rate-limiting transport processes that
 cause nonideal behavior, resulting in lower actual contaminant removal rates than predicted by
 equilibrium mass transfer theory. Although flow modeling can provide reasonable estimates for
 air flow rates and pathways in the subsurface, it is difficult to quantify rate-limiting transport
 processes.

       This document addresses the following issues or questions concerning SVE modeling:

       •      How does one determine if modeling is necessary and appropriate for an
              application?

       •      What types of questions can modeling answer?

       •      How can models be used as decision-making tools?

       •      What type of model is necessary for a given site?

       •      What types of site characterization data are necessary for modeling studies?

              What are the major properties of the porous medium and contaminant that control
              flow and transport in the vapor phase?

       •       What are the major properties that control mass transfer among phases?

       •       What models are available commercially or in the public domain?

These are some of the major issues to consider before embarking on any modeling exercise.

 1.2    OBJECTIVES

       This document is intended to guide the user through the process of selecting and applying
models to SVE sites. Technical information is provided in order to:

       •       Determine the types of problems that can be addressed by modeling;

              Highlight the methods that are commonly used to solve such problems;

-------
1.3
•      Assist potential users (1) determine the presence or absence of a need for
       modeling at their site, and, if a need is shown to exist, (2) select a model for their
       site;

•      Identify and illustrate the major processes governing air flow and contaminant
       vapor transport in the vadose zone;

       Present a discussion of model data needs;

       Review available commercial and public domain codes; and

•      Present example model applications and case studies.

REPORT ORGANIZATION
       A general guide for locating major discussion topics is presented in Table: 1. The reader
should note that the terms "air", "vapor", and "gas" are used interchangeably throughout this
document to refer to gas in the vadose zone. Terms and symbols used to represent equation
parameters are listed near the front of this document, and are also defined upon first use in each
main section.

       An introduction to model use is provided in Section 2.1. This is followed in Section 2.2
by a presentation of SVE issues that can be addressed by quantitative analysis. Section 3
contains an overview of flow and contaminant transport processes that can be simulated, their
governing equations,  and their formulation in mathematical models.

       SVE model data requirements are addressed in Section 4. Media and chemical properties
that affect air flow and vapor-phase transport,  and are flow and transport model parameters, are
discussed in Sections 4.1 and 4.2. This is followed by discussions regarding model boundary
conditions (Section 4.3), SVE model assumptions (Section 4.4), and data interpretation
considerations (Section 4.5).                                   .......

       The reader is guided through selection of an appropriate SVE model type in Section 5.
Issues critical to  model selection are discussed in Section 5.1. The remainder of Section 5
contains descriptions of the different types of models that are available to analyze SVE systems
(i.e., screening models, air flow models, and compositional flow and transport models).

       Readily available commercial and public domain codes are reviewed in Section 6, and
Section 7 follows with several example applications utilizing codes that were reviewed in the
previous section. Three case studies of model application to assess the feasibility, design, and
performance of SVE systems at contamination sites are described in Section 8.

-------
Table 1.
Location of major topics in the document.
Issue
General introduction to modeling
Questions that modeling can address
Flow and transport processes and equations
Air flow model parameters
Vapor-phase transport model parameters
Model boundary and initial conditions
Model selection criteria
Descriptions of different types of SVE models
Descriptions of readily available models
Example model applications
Case studies
Location
Section 2.1
Section 2.2
Section 3
Section 4.1
Section 4.2
Section 4.3
Section 5.1
Sections 5.2 to 5.6
Section 6
Section 7
Section 8

-------
                                       SECTION 2
                            OVERVIEW OF SVE MODELING

       Mathematical modeling facilitates analysis of SVE feasibility, design, and performance.
A general introduction to the modeling process is provided in Section 2.1.  This is followed in
Section 2.2 by an introduction to specific issues that can be addressed by SVE modeling.

2.1    THE MODELING PROCESS

       An initial step in the modeling process is to determine if a model is necessary to achieve
the intended goal. For SVE, this begins with the development of a site conceptual model that
incorporates the physical and chemical processes and media properties that control air flow and
contaminant vapor transport. After developing a site conceptual model and determining the need
for quantitative analysis, the modeler selects an appropriate simulation code and begins the
process of model application.  Model application typically involves an effort to simulate
observed conditions (known as calibration), sensitivity analysis, evaluation of study issues, and
documentation.

2.1.1  Mathematical Model Development

       Simulation of SVE systems refers to the construction and operation of a model whose
behavior represents a simplified approximation of the actual vadose zone behavior. The
following general approach to constructing deterministic mathematical models of natural
processes is derived from Mercer and Faust (1992).

       A mathematical model is simply a set of equations which, subject to certain assumptions,
describes the physical processes active in the domain of interest (e.g., the vadose zone). The
mathematical models discussed in this document are deterministic; they define cause-and-effect
relationships based on an understanding of the physical system. The procedure for developing a
deterministic, mathematical model of any physical system can be generalized as follows.  The
first step is to understand the physical behavior of the system. Cause-effect relationships are
determined and a conceptual model of how the system operates is formulated.  The next step is to
translate the physics into mathematical terms by making appropriate simplifying assumptions
and developing the governing equations.  This constitutes the mathematical model.

       The mathematical model for air flow consists of a partial differential equation together
with appropriate boundary and initial conditions that express conservation of mass and describe
continuous variables  (for example, pressure) over the region of interest. In addition, it entails
various physical laws describing rate-limited processes in the vadose zone.  An example is
Darcy's law for fluid  flow through porous media; this is generally used to express conservation of
momentum. Finally, various assumptions such as those of one- or two-dimensional flow are
invoked. For volatile contaminant transport, additional partial differential equations with
appropriate boundary and initial conditions are required to express conservation of mass for the

-------
contaminants of interest.  The basic equations that describe air flow and contaminant vapor
transport are presented in Section 3.

       Once the mathematical model is formulated, the next step is to obtain a solution using
one of two general approaches. The air flow equation can be simplified further, for example, by
assuming radial flow and infinite aquifer extent, to form a subset of the general equation that is
amenable to analytical solution. The equations and solutions of this subset are referred to as
analytical models. The familiar Theis curve represents the solution of one such analytical model.

       Alternatively, for problems where the simplified analytical models no longer describe the
physics of the situation, the partial differential equations can be approximated numerically, for
example, with the finite-difference technique. By doing this, one replaces continuous variables
with discrete variables that are defined at grid blocks. This approach constitutes a numerical
model.  A numerical model is most appropriate for general problems involving systems having
irregular geometric boundaries, heterogeneities, or highly variable extraction rates.

       Both types of model have advantages and disadvantages that are reviewed in this
document.  Consequently, no single approach should be considered superior to others for all
applications. The selection of a particular approach should be based on the specific vadose zone
problem addressed. Whichever approach is taken, the final step in modeling a system is to
translate the mathematical results back to their physical meanings. In addition, these results must
be interpreted in terms of both their agreement with reality and their effectiveness in answering
the questions which motivated the modeling  study.

2.1.2  Determining Model Need

       The need for a model and the requisite level of analysis are determined based on (1)
interpretation of available site  data, (2) site complexity, and (3) remedial analysis objectives.
Site conditions and management objectives may obviate the  need for modeling.  It would be
overkill to use a model to determine SVE feasibility at a site with minimal vadose zone air
permeability and/or contaminants of interest that are not volatile.

       Mathematical modeling can be done at various levels of complexity, and the tools utilized
range from relatively  simple (e.g., analytical solutions and screening equations) to complex (e.g.,
compositional flow and transport models). For example, whereas a simple flow equation can be
employed to optimize vapor extraction well spacing in certain cases, a more complicated model
will be necessary to consider three-dimensional air flow in a heterogeneous vadose zone. If the
goal is to estimate cleanup time, it may be necessary to use a model that accounts for mass
transfer between chemical phases.

        Screening models that  oversimplify site conditions are often useful for making initial site
analyses, but may be inadequate to support design and predictive decisions. This is because
geologic heterogeneity and other field complexities are usually not well represented in these

-------
 models. Although more complex models allow a wider range of situations to be simulated, they
 often require input that are not sufficiently known (due to a lack of field data), and thus, may
 produce erroneous results. An effort should be made to avoid using models that are more
 complex than necessary.  SVE model selection is addressed more fully in Section 5.

        Determining if a mathematical model is needed and selecting an appropriate code are
 steps within the analytical process outlined below:

        1.     Define study objectives;
        2.     Collect and analyze site characterization data;
        3.     Formulate a site conceptual model;
        4.     Determine mathematical model necessity;
        5.     Identify process descriptive equations;
        6.     Select computer code;
        7.     Apply computer code using site data; and
        8.     Make decisions to meet objectives.

        Study objectives generally are to locate and remediate site contamination.  Attaining
these objectives usually begins with a site characterization effort that includes inventorying
potential sources and estimating their histories. A conceptual model of the site is formulated
during the characterization process.  Hypothesis testing occurs whereby site characterization data
are used to verify or modify the conceptual model. This hypothesis testing may  take the form of
a computer model. The quality of model results is directly correlated to the quality of the input
data. Complete and accurate input data are required for reliable model results. Although
modeling can not be used as a substitute for data collection, it can be used to direct the site
investigation program. Thus, it is necessary to determine if a model would be helpful for site
characterization and/or remedy assessment.

       Use of a computer model during site characterization will force consideration of the
parameters required during design, and help ensure the collection of critical field data. Such an
exercise also compels consideration of the important processes controlling contaminant
migration and remediation. As a result, the proper equations are more likely to be  solved with an
appropriate code. For example, to design a successful SVE system, consideration of air flow
may be all that is necessary. The air flow model is used to estimate potential air  flow regimes  '
and pinpoint areas where more site characterization data may be needed. During the latter steps,
the selected code(s) is used with site data to make remediation decisions that meet  the study
objectives.

       As noted by the National Research Council (1990), modeling results should always be
used in conjunction with other site data and good professional judgment. Those  who rely on
models should be aware that simulation results often appear deceptively correct. Furthermore,
model users need to be aware of the simplifying assumptions that are part of any mathematical
model, and realize the implications of these simplifications.

-------
2.1.3  Establishing Modeling Goals

       The purpose of modeling should be clearly defined at the outset in order to approach the
modeling problem in a reasonable and efficient manner.  The scope and goals of modeling can be
elucidated by answering several questions (Anderson and Woessner, 1992):

       •      Will the model be used for system interpretation, prediction, feasibility analysis,
              remedial design, or remedial performance assessment?

       •      What questions should the model answer? What do you hope to learn from the
              modeling exercise?

       •      Is modeling the best way to address the issues at hand?

       •      What type of model should be used meet the study goals? Can an analytical or
              screening model be used, rather than a more complex numerical model?

Establishing objectives facilitates determination of a reasonable modeling approach and the
associated level-of-effort.

2.1.4  Model Calibration. Sensitivity Analysis, and Prediction

       Model calibration is the process of adjusting model input parameters to obtain a
reasonable match between simulated and observed field conditions. As described by USEPA
(1988) and Anderson and Woessner (1992), calibration poses an inverse problem whereby the
modeler attempts to deduce values for unknown parameters by a trial-and-error process of history
matching to known field conditions.  A flowchart for this process is presented in Figure 2.

       Model calibration may be limited by inaccurate or unavailable field data. Ideally, model
results are compared to those of a well-defined field experiment. An example is attempting to
determine air permeability values by matching model results with the results  of an SVE pilot test.
In the absence of such a field experiment, available data are used to calibrate  a site model
through reasonable adjustment of parameter values.

       Calibration is both intuitive and subjective, because more than one set of parameters may
produce output that matches measured field conditions. Due to the absence of a unique set of
calibration parameters, it is up to the professional judgment of the modeler (and model
reviewers) to determine when and if the model is calibrated.

       Sensitivity analysis involves varying formation properties, stresses, and boundary
conditions input to the calibrated model to quantify the uncertainty in simulation results
associated with model input uncertainty.  Calibrated parameter estimates, such as air
permeability and contaminant properties, are systematically changed within a reasonable range to

                                           10

-------
              Field
             System
Measured
 Output
   Estimates of
   Parameters ~
       Error
     Analysis
           Numerical
             Model
Computed
                          Acceptable
                             Error
               Calibrated
                 Model
Unacceptable^
 Output
   Error
 Parameter
Adjustment
                                      New Parameter
                                        Estimates
Figure 2.     Trial-and-error model calibration procedure. The conceptual site model is
             converted to a numerical model and calibration goals are determined.  Model
             results are then compared to calibration goals. If the model results are not
             acceptable, parameter values are adjusted and the model is re-run (from Anderson
             and Woessner, 1992). [Reprinted by permission of Academic Press, Orlando,
             Florida.]
                                          11

-------
examine resultant changes in simulated conditions (such as air flow patterns and contaminant
removal rates). Sensitivity analysis can be used to help determine critical data needs and
monitoring requirements, and/or to develop safety factors for remedial design.

       Predictive simulations are made to evaluate S VE feasibility and design parameters,
typically after the model has been calibrated to match data derived from an S VE pilot test and
other site characterization efforts.  Predictive simulations using alternative vapor extraction well
locations and rates can be made at a fraction of the time and cost needed to perform analogous
field studies. Comparing the simulated performance of different SVE schemes provides insight
to SVE system feasibility and design.

2.1.5   Model Documentation and Quality Assurance

       Quality assurance (QA) procedures are an integral part of computer code development
and application.  QA is generally the responsibility of the modeler and other personnel who
develop and utilize the model.  Aspects of computer code and model QA may include the
following activities:

       •      Documentation of code characteristics, capabilities, and use;
       •      Verification of the computer program structure and coding;
       •      Documentation of the model application;
       •      Achieving model calibration criteria; and,
       •      Technical review.

       Clear communication of the purposes, methods, logic, relationships, capabilities and
limitations of a model code enables users to effectively apply it to a particular problem. A user
manual may include instructions for operating the code and preparing data files, example
problems with input and output, instructions for programmers and operators, and a report of the
code verification. Clear and concise documentation will diminish the potential for model code
misapplication.

       Code verification demonstrates that the mathematical model can accurately solve the
governing equations and assures that the computer code is fully operational. Codes are often
tested by running problems for which an analytical solution exists, so that their results can be
checked against analytical results. A step-by-step analysis of the program operation during
testing is referred to as a "code walk-through." While not required, it does provide a complete
review of the program structure.

       Documentation of the modeling process records the project objectives, technical
approach, activities performed, problems addressed, and results achieved.  Reports and files
should be maintained to document:
                                           12

-------
        •      Model assumptions;
        •--,     Model parameter, values and sources;
              Model boundary and initial conditions;
              Model grid design;
        •      Code modifications and verification;
              Output of model runs and interpretation; and
      ,  •      Model calibration.

 All data files, source codes, and executable versions of computer software used in the modeling
 process should be retained for auditing or post-project reuse.

        Quantitative criteria are frequently used to assess and assure £oodness-of-fit associated
 with model calibration.  For example, the root-fnean-squared (RMS) error between pressure
 changes simulated and observed during an SVE pilot test may be calculated and compared to
 calibration QA goals. Monitoring data acquired during SVE system operation can be used to
 further evaluate and improve model reliability.

        Model applications are generally subject to internal technical review as part of a QA
 program and subsequent external review by interested parties. The technical review should
 assess if the modeling is reasonable and appropriate considering the analysis objectives.

 2.2     SVE QUESTIONS THAT MODELING CAN ADDRESS

       Questions regarding SVE feasibility, design, and performance that can be examined by
 quantitative analysis are introduced below and summarized in Table 2. A model user might want
 to consider these questions prior to selecting a code to develop a particular model application.
 Many of the SVE questions are addressed in greater detail elsewhere in this document.

       SVE feasibility and performance issues are evaluated based on design goals. Similar to
 hydraulic containment and aquifer restoration objectives adopted for groundwater pump-and-
 treat, two common SVE design strategies are: (1) to contain and control the movement of
 volatile contaminants in the vadose using a radius of influence (ROI) approach; and (2) to
 remove volatile contaminants from the vadose zone using a zone of remediation (ZOR) approach
 (Johnson and Ettinger, 1994). Both strategies require site data and may involve the use of
 models.

       The ROI refers to the greatest distance of measurable vacuum propagation, often
 specified as 0.1 inches of water, from an SVE well (USEPA, 1994). It is estimated by
 extrapolating vacuum versus distance measurements taken during a vapor extraction test,
 empirical calculation, and/or air flow model analysis.  Successful implementation bf the ROI
approach to contain contaminated soil gas does not ensure timely cleanup of the vadose zone due
to low air flow and mass transport rates in portions of the ROI. The ZOR approach to SVE
                                          13

-------
Table 2.     Summary of important SVE questions and some potential uses of modeling.
Question
Is the air permeability at
the site high enough to
allow effective use of
SVE?

Are the compounds
present at the site
volatile enough to be
removed by SVE?
How quickly will vapor
concentrations drop to
levels that do not yield
adequate mass
removal?
What effect does well
placement (location and
configuration) have on
air flow and mass
transport?
How does changing the
extraction well screened
interval affect the air flow
field?
What is the design
radius of influence (ROI)
of each well, and how is
it affected by geologic
conditions and air flow
rates?
How will air injection
affect air flow to
extraction wells?
Possible Model Uses
Use screening models, for relatively simple problems, to analyze air
pumping tests, estimate air permeability, and to estimate the number of
wells and air flow rates necessary for the desired cleanup; if these
numbers are not reasonable, then SVE may be impractical
Use air flow models, for simple or complex problems, to analyze air
pumping tests, estimate air permeability, and to estimate the number of
extraction wells and flow rates needed for cleanup; if these numbers are
not reasonable, then SVE may be impractical
Perform simple screening calculations based on literature values for
Henry's Law constant and vapor pressure
Use transport models to estimate contaminant levels in soil over time as
SVE proceeds to determine if cleanup levels can be reached in a
reasonable time; if not, then SVE may be impractical
Perform screening calculations to estimate equilibrium concentrations for
basic scenarios; high equilibrium concentrations do not ensure high
contaminant removal rates
Use transport models to estimate equilibrium concentrations in extracted
vapor considering media and contaminant property complexities
Use screening models to estimate the effects of well placement on the air
flow regime and radius of influence
Use air flow models to estimate air flow pathlines and assess stagnation
zones; test different extraction well configurations and flow rate scenarios
Use transport models to study the effects of well placement on
contaminant removal patterns and rates
Use air flow models to analyze flow to a well under different screened
interval scenarios
Use screening models to test the effect of using different values for the
ROI; confirm with field measurements to determine the actual ROI
Use air flow models to estimate the ROI by studying the radial vacuum
distribution about an extraction well; calibrate air flow models by
comparing simulation results with radial vacuum measurements made in
the field
Use air flow models to estimate the effects of injection wells, or passive
air inlet wells, on the air flow regime
                                        14

-------
Table 2.    Summary of important SVE questions and potential uses of modeling (continued).
Question
How do the air flow fields
differ for systems using
drains or vertical wells?
What are the effects of
boundary
conditions(e.g., surface
capping) on the air flow
regime?
What residual semi-
volatile or nonvolatile
contaminants may be left
after volatiles are
removed via SVE?
What cleanup levels can
be achieved, and will
these be within
regulatory limits?
What is the estimated
cleanup time?
What monitoring is
necessary to verify that
SVE is working?
Possible Model Uses
Use air flow models to estimate the usefulness of vertical versus
horizontal wells or drains
Use air flow models to estimate the effects of various surface conditions
by varying the boundary conditions
Use screening models to make a rapid estimate of residual mass in the
soil after venting
Use transport models to estimate post-SVE residual contaminant mass in
the soil subject to heterogeneities, rate-limiting processes, etc.
Screening models can be used to examine potential cleanup levels by
estimating post-SVE residual contaminant levels; cleanup levels may be
limited by various processes (diffusion, NAPL dissolution, desorption
etc.) '
Transport models can be used to estimate cleanup levels taking into
account subsurface heterogeneities, variable SVE design and operation,
rate-limiting processes, and other complexities
Use screening models to make a first estimate of cleanup times; consider
potential effects of complex mass transport mechanisms
Use transport models for a more rigorous estimate that considers
geologic and contaminant transport complexities
Air flow models can be used to estimate the air flow regime, which in turn
can be used to determine where vacuum points should be located
Transport models can be used to estimate the distribution of contaminants
over time and can assist in determining where soil samples during and
after the venting operation should be taken
                                       15

-------
design recognizes that the degree of cleanup achieved is related to the cumulative volume of air
flow through the contaminated media. A ZOR analysis can be made to estimate the air flow rate
distribution required to reduce volatile contaminant concentrations in the vadose zone to an
acceptable level in an acceptable time period.  Appropriate design parameters (e.g., well
locations and rates) are then specified to produce the desired air flow regime. Thus, one might
consider both air flow and contaminant transport modeling to determine a ZOR, rather than
considering only air flow and the radius of influence.

2.2.1   Is the air permeability at the site high enough to allow effective use of SVE?-

       Air permeability controls the rate at which soil vapor can be extracted from the vadose
zone, and thus, the feasibility of SVE. It is primarily a function of intrinsic permeability and
moisture content. As shown in Figure 3, the air permeability of a medium decreases as pore
spaces become filled with water; thus, it is not a single number, but a range of values depending
on water saturation. The distribution of air permeability is affected by media heterogeneity and
anisotropy (i.e., variation in grain size distribution and stratification, respectively). Combined
with boundary conditions, such as air leakage at the ground surface, the distribution of air
permeability determines the pattern of vapor flow induced by SVE.

       Estimates of air permeability can be derived from field permeability tests, SVE pilot
studies, laboratory core sample tests,  and knowledge of medium characteristics (USEPA, 199 la).
Field test data can be interpreted using models of varying complexity, ranging from simple  well
hydraulics solutions (e.g., Theis, 1935; Hantush-Jacob,  1955) to numerical models that account
for complex permeability distributions and boundary conditions.  Care must be exercised to
select a model that reasonably accounts for field conditions.  At many sites, surface leakage
effects are pronounced during SVE tests.  Thus,  analytical solutions that account for air leakage
through open ground or low permeability surface layers have widespread application for analysis
of air permeability test data and SVE well-spacing (Beckett and Huntley, 1994; Nyer et al., 1994;
Shan et al., 1992; Baehr and Hult, 1991; Falta, 1993; Falta, 1995).

       Similarly, the feasibility of SVE also can be assessed using models of varying
complexity. Initial screening for SVE effectiveness is frequently made based on an estimate of
air permeability as illustrated in Figure 4. High permeability generally indicates the potential for
significant air flow to an extraction well; it also suggests the potential for relatively extensive
vacuum propagation from an extraction well.  Given an estimate of the air permeability
distribution, models can be applied to evaluate the effectiveness of specific numbers, locations,
and extraction rates of vadose zone wells or drains, and thereby help determine SVE feasibility.
High air flow rates, however, do not necessarily produce high contaminant removal rates (which
also depend on contaminant volatility and connection to the air flow paths).
                                            16

-------
               100-
               90 -
               7oH
           LJJ
           1
               50-
               40-
               30-1
               20-
                10-1
     AIR
 PERMEABIUTY
O  A Surface soil

•  D  Subsurface soil
                          I      II      II       I
                         10      20    30     40     50      60

                                               Water Saturation %
                                         70
                                                I
                                                80
                                                       90    100
Figure 3.      Relationship of relative air permeability to relative water permeability (USEPA
               1991a).
                                               17

-------
VAF
PRES
j
Butane — ^~
Pentane — ^~
Benzene — ^~
Toluene — *~
Xylene — ^~
Phenol — ^~
Naphthalene — ^~
Aldicarb — •*-
i
S\
LIKEL
'OR 0
SURE SUC(
-104
-103
-102
^~ ^
— I I
-ID'1
-1Cf2 |
-,0-3 |
-<°-4 j
r ^
I
/E
HOOD
F S
DESS PER
i
SUCCESS
VERY
LIKELY
~ — • — .
SUCCESS
SOMEWHAT
LIKELY
ij
! SUCCESS
I LESS
{ LIKELY
''(
>OIL AIR
MEABILITY
I
1
tr*^
^M

" "i %"
t
»
*
,»
i
j
t .
,t
,<
<
' » ;
1
HIGH
(gravel,
coarse
sand)

MEDIUM
(fine sand)
LOW
(clay)
TIME
SINCE
RELEASI
Weeks
Months
' Years
Weeks
Months
Years
Weeks
Months
Years

                                                  Match Point
Figure 4.      Decision nomograph for evaluation of relative volatility and potential for cleanup
              via SVE. Units of vapor pressure are mm Hg (after USEPA, 199 la).

                                           18

-------
2.2.2  Are the compounds present at the site volatile enough to be removed bv SVE (i.e.. are the
       Henry's law constant and vapor pressure great enough)?

       In many cases, this critical question can be answered by referring to chemical properties
data as shown in Figure 4 or by making simple calculations. The maximum concentration of a
chemical in soil gas is its theoretical equilibrium (or saturated) vapor-concentration. Equilibrium
vapor concentrations for a chemical or mixture of chemicals in the vadose zone can be calculated
using:  (1) Henry's law constant where the compound is present dissolved in water, but not as a
separate phase, or (2) vapor pressure data where the compound is present as a separate phase
liquid or solid (Section 4.2). The potential feasibility of SVE can be evaluated based on these
equilibrium vapor concentrations. Data on vapor pressure and Henry's law constants are
provided in Appendix A. Actual  removal rates, however, are also a function of complex
chemical partitioning, diffusion, extraction flow rate, air flow patterns, and contaminant mass
distribution; and usually decline significantly with continued SVE operation.

2.2.3  How quickly will vapor concentrations drop to levels that do not yield adequate removal?

       In order to evaluate if SVE is a viable remedial alternative, it is beneficial to project the
trend of contaminant concentrations in the extracted vapor over time.  SVE pilot studies, which
typically involve extracting soil vapor from a single well and measuring extraction flow rate,
vapor concentrations, and vacuum distribution for several hours to several weeks, provide
information on volatile compound concentrations that are likely to be extracted during the early
period of SVE operation. Modeling can be used to  extrapolate the pilot test data over time and
space, considering alternative well configurations and other operating parameters.

       Relatively simple analytical models can provide estimates of contaminant removal rate by
combining equilibrium (or adjusted) concentrations with estimated air flow rates. Equilibrium
concentrations will likely overestimate vapor concentration and, hence, mass removal. A more
complex analysis of contaminant  concentration and mass removal can be developed with a
compositional flow and transport  model.  For example, such a model can be used to project
extracted vapor concentrations at  different possible well locations throughout a site. Mass
removal trends predicted using complex models are often subject to significant parameter
uncertainty.

2.2.4  What effect does well placement have on air flow and mass transport?

       The subsurface air flow regime is mainly controlled by extraction well location,
subsurface permeability distribution, surface boundary conditions, extraction well screen
interval, and extraction rate. An air flow model can be used to study the effects of changing
these parameters. The output from an air flow model consists of simulated vacuums throughout
the model domain. Estimated air  flow pathlines are determined from the calculated vacuum
distribution (Figure 5). For relatively simple problems, it may be desirable to calculate the

                                           19

-------
                             EXTRACTION WELL
                                                   SURFACE SEAL
                  1"
        ...37....
      I      f    A"    20-WATER     A-    f
[      I      I      |      VACUUM       I
Figure 5.     Placement of the well screen affects how air flows to the extraction well (from
             USEPA, 1991a).

                                          20

-------
vacuum distribution using a simple analytical solution and superposition theory to account for
multiple wells and/or boundary conditions.  Use of a more complex analytical or numerical air
flow model will be required to consider nonidealities such as spatially varying vertical leakage
due to different surface conditions, subsurface heterogeneities (i.e., flow obstructions and
preferential pathways), variable well screen intervals, and a variable thickness vadose zone.
Contaminant transport models are used to examine the movement of chemical mass that is
distributed within the simulated air flow field. Uncertainties associated with model parameters
may be evaluated by performing sensitivity analyses.  Modeling, therefore, enables evaluation of
air flow patterns, potential stagnation zones, air flushing rates, mass removal, and clean-up
afforded by alternative well placement designs.

       S VE extraction wells must be strategically placed in contamination source areas to
maximize mass removal. Simple analytical or screening models are typically sufficient to
develop an initial estimate of well numbers and spacings needed to contain contaminated soil
gas using the ROI approach. Application of an oversimplified model, however, can lead to
inadequate well placement. For example, failure to account for enhanced radial vacuum
propagation from an extraction well due to anisotropy (where Kx» Ky) may result in system
overdesign (e.g., more SVE wells than needed).  Alternatively, a flawed analysis can'also result
in system underdesign. An assumption sometimes made to solve the equation of air flow in the
vadose zone is that flow is confined (i.e., there are no air leaks from the ground surface). This
assumption will tend to overestimate the radius of influence of an SVE well (and lead to a design
with too few wells) because any vacuum applied to the well is assumed to draw air only from a
lateral direction, and not from vertically above the well.  Significant vertical leakage will occur at
most sites, even those covered by pavement or buildings.  For clean-up based oh! the ZOR
approach, more detailed analyses are required to optimize pore-volume flushing rates and
minimize air stagnation (Figure 6). Transport models are particularly useful to study the effects
of well placement on contaminant distribution and removal trends.

2-2.5  How does changing the extraction well screened interval change the air flow field?

       Well screen placement affects the pattern of air flow to an SVE well (see Figure 5).
Extraction wells are usually screened only through the zone of contamination to maximize mass
removal (Johnson et al.,  1990b). Wells with larger screen intervals  may short-circuit air flow,
thereby limiting the effectiveness of SVE. Under certain conditions, however, alternative screen
placement will optimize flushing of the contaminated zone.

        Modeling can be used to examine vacuum distributions and vapor flow pathlines induced
by different well screen intervals.  Several analytical solutions are available to calculate two-
dimensional axisymmetric vacuum distributions  and air flow patterns  caused by SVE from a
partially-penetrating well screen under steady-state, transient, isotropic, anisotropic, leaky, and
open ground conditions (e.g., Falta, 1995; Shan et al.,  1992; Baehr and Hult, 1991; Hantush,
1964). More complex analytical or numerical models may be required to evaluate screen

                                           21                        ,    l

-------
             -I	 VAPOR FLOW


              O  EXTRACTION WELL
                                             I

                                                           AIR FLOW
                                                           STAGNATION
                                                           ZONE
                  EXTRACTION WELL


                  INJECTION WELL
                                             O
                                                            O
Figure 6.     Placement of an injection well to eliminate air flow stagnation zones (from
             USEPA, 199 la).
                                          22

-------
placement at sites with multiple layers, heterogeneous media, nonuniform contamination,
multiple wells, and irregular extraction schedules.

2.2.6  What is the design radius of influence of each well, and how is it affected by geologic
       conditions and air flow rates?

       The radius of influence (ROI) of an extraction well is one of the most important
considerations for SVE design and operation. Its definition is subjective. According to USEPA
(1994), the ROI is "the greatest distance from an extraction well at which a sufficient vacuum
and vapor flow can be induced to adequately enhance volatilization and extraction of the
contaminants in the soil [which as] a rule-of-thumb ... is considered to be the distance from the
extraction well at which a vacuum of at least 0.1 inches of water is observed." Kuo et al. (1990)
define the ROI as "the radial distance where there is sufficient air flow to reduce contaminant
concentrations below an acceptable level within a pre-specified time frame." Note the similarity
of this definition to that of the zone-of-remediation (ZOR) approach described previously.

       Radius of influence is significantly affected by parameters such as extraction rate and
vacuum, SVE well depth, surface boundary conditions (see Figure 7), heterogeneities, use of
injection wells, and the ratio of horizontal to vertical permeability. Reported ROI values range
from five to  a few hundred feet (Hutzler et al., 1989; Johnson and Ettinger, 1994; USEPA, 1994).
In general, the ROI increases with increasing extraction rate, increasing extraction vacuum,
increasing well depth, increasing media permeability, increasing K^K, anisotropy, decreasing
water saturation, decreasing surface permeability, and decreasing air injection.

       The ROI can be estimated by: extrapolating vacuum versus distance measurements taken
during a vapor extraction  test or SVE operation, empirical calculation (e.g., using ROI
determined at similar sites), and/or air flow model analysis. Screening models can be used to
estimate ROI and assess the significance of different ROI values given simple geologic
conditions. To consider complex systems and conduct more thorough analysis, analytical or
numerical air flow models can be used to study the vacuum distribution resulting from SVE at
one or more wells.

2.2.7  How will air injection affect air flow to extraction wells?

       To enhance and manipulate flow patterns, air can be (1) forced into the vadose zone
through injection wells and drains or (2) drawn into the vadose zone through passive inlet wells
and drains that are open to the atmosphere.  Both active and passive air injection can be used to
enhance contaminant volatilization, reduce  short-circuiting of the vapor flow path from the
ground surface to the SVE well, eliminate air flow stagnation zones (Figure 6), and/or create a
pressure barrier to limit an SVE ROI (e.g., to prevent inflow of contaminated soil gas from an
adjacent gas station site, Figure 8).
                                           23

-------
                                                          is radius of influence)
Figure 7.     Surface boundary effects on radius of influence (from USEPA, 1991 a).
                                          24

-------
                                        Passive Air
                                           Inlet    /
                                         Trench  /
                                                                      Pump Islands
                                                        USTs
                    SVE Wells -
                                           Plan View
                                        Passive Air



^

/ -\

—



—
iiiitsi i lenun
-•<



\ Vacuum
Distribution
Water Table

                                           Cross Section
Figure 8.      Air injection can be used to create a pressure barrier to limit the ROI of SVE.
                                            25

-------
       Air injection may enhance SVE by increasing pressure gradients and diluting soil gas,
which allows increased volatilization from the contaminant source.  The rate of volatilization is
related to the concentration gradient between the source and the air moving through the system.
Thus, injected clean air can increase the concentration gradient and hence the rate of
volatilization.

       The amount of air flow through the subsurface is often expressed in terms of the number
of pore volumes flushed per day. One pore volume represents the pore space volume in the zone
to be remediated. As an example, at the Verona Superfund Site (see case study in Section 8),
removal rates were on the order often pore volumes per day (USEPA, 1991d).  The ability to
flush many pore volumes through the vadose zone is an advantage of SVE compared to pump-
and-treat technology in the saturated zone. At some flushing rate, however, the incremental
benefit of increasing air flow begins to diminish (and ultimately disappears) due to mass transfer
limitations.

       In order to examine the potential effects of air injection wells (or drains), it will be
necessary to perform an analysis of the pressure distribution throughout the SVE domain.
Analytical or numerical air flow models can be used to evaluate flow fields effected by air
injection alternatives (e.g., different configurations of active or passive wells or drains)
considering complex permeability distributions and boundary conditions. Transport models can
be used to examine the effects of air injection on contaminant movement and removal.

2.2.8  How do the air flow fields differ for systems using drains or vertical wells?

       In some situations, SVE will be more effective using drains or horizontal wells than
vertical wells (see Figure 9). .For example, it may be desirable to excavate contaminated soils
and place them around horizontal drains in an above-ground pile.  Elsewhere, the shape and/or
permeability of the contamination zone may dictate that horizontal drains or wells be used
instead of vertical wells. Horizontal wells or drains tend to be particularly effective, compared to
vertical wells, for SVE operation at sites with shallow contamination or low-permeability media.

       Assessment of the comparative effectiveness of SVE using vertical wells versus
horizontal wells or drains is best made using a numerical air flow model. Numerical models
generally allow more flexibility in defining irregular SVE well and drain locations and
orientations than analytical models.

2.2.9  What are the effects  of boundary conditions (for example, surface capping)?

       Surface boundary conditions can significantly affect SVE air flow patterns (Figure 7). As
noted previously, analytical solutions are available to calculate air pressure distributions under
confined, leaky, and open ground conditions. Changes in air flow patterns due to changing the
ground surface condition, therefore, can be examined by comparing the results of analytical

                                           26

-------
        CONTAMINATED
           LAYER
CONFINING
   UNIT
      ^ >^^..>../.Y%JS...
              (a) Vertical Well Screened in
                 Contaminated Zone
                                                                  SURFACE SEAL
                                                           CONTAMINATION
                             (b) Trench
                                                  AIR INLET VENTS    SURFACE SEAL
                                                                  CLEAN BACKFILL
                 (c) Horizontal Drilling
                            (d) Soil Pile
Figure 9.     Common configurations for SVE wells and drains (from USEPA, 199la).


                                          27

-------
solutions that incorporate different representations of this boundary condition.  Again, however,
numerical models typically provide a more versatile tool for simulating varied and complex
boundary conditions. Modifying boundary conditions in most numerical models is relatively
simple and straightforward.

2.2.10 What residual semivolatile or nonvolatile contaminants will be left after volatiles are
       removed bv SVE?

       One of the questions that arises when modeling multicomponent mixtures such as
gasoline is how to determine the extraction rates for the each of the many components. The first
step is to develop a representative composition of the multicomponent mixture.  This can be
accomplished by sampling and analyzing free product at the site. Several of the screening
models evaluated hi this document include composition data for at least two representative
gasoline mixtures (Table 3): new gasoline to represent recent spills, and weathered gasoline to
represent old spills. Due to preferential dissolution and volatilization, the composition of
weathered gasoline is depleted of the more volatile components relative to new gasoline.

       Different fuels have varying percentages of volatile and semivolatile components. Gas
chromatograms for regular gasoline, unleaded gasoline, kerosene heating fuel, and diesel fuel are
shown in Figure 10. Volatile components elute faster from the GC column, and thus have a
shorter retention time, than semivolatile compounds. The composition of gasoline and other fuel
mixtures will vary over tune as SVE progresses due to preferential volatilization of the lighter
components. Column studies of SVE conducted by Baehr et al. (1989) evidence this selective
removal of compounds based on their volatility (Figure 11). As evidenced by Figure 10,
kerosene and diesel fuel are more difficult to remove using SVE than gasoline due to their
relatively high fraction of semivolatiles.

       Calculations of contaminant mass-in-place or mass-removed versus time can be plotted to
assess the removal rates provided by SVE for multiple contaminants within a mixture. For
example, relative rates of benzene, toluene, p-xylene, and naphthalene removal by SVE
simulated using a screening model (VENTING) for a fuel contamination site are presented in
Figure 12.  As shown, the volatile components (benzene, toluene, and p-xylene) are stripped
from the vadose zone much more quickly than naphthalene, which is semivolatile. This type of
analysis will usually indicate that SVE is not particularly feasible for removing semivolatile
compounds from the subsurface.,

       Traditional flow and transport codes typically can model only one constituent at a time,
and do not consider the presence of NAPL. As such, a representative indicator compound may
be selected for simulation, or the user may conduct multiple simulations to assess the fate and
transport of multiple compounds. More recently, an alternative solution of the mass transport
equation (Benson et al., 1993) has enabled development of compositional flow and transport
                                          28

-------
Table 3.   Composition of fresh and weathered gasolines (from USEPA, 1991 a; after
          Johnson et al., 1989).
Compound Name
propane
isobutane
n-butane
trans-2-butene
cis-2-butene
3-methy!-1-butene
isopentane
1-pentene
2-methyl-1-butene
2-methyl-1 ,3-butadiene
n-pentane
trans-2-pentene
2-methyl-2-butene
3-methyl-1 ,2-butadiene
3,3-dimethyM -butene
cyclopentane
3-methyl-1 -pentene
2,3-dimethylbutane
2-methylpentane
3-methylpentane
n-hexane
methylcyclopentane
2 , 2-d imethy Ipentane
benzene
cyclohexane
2,3-dimethylpentane
3-methylhexane
3-ethylpentane
n-heptane
2,2,4-trimethyIpentane
methylcyclohexane
2,2-dimethylhexane
toluene
MW(g)
44.1
58.1
58.1
56.1
56.1
70.1
72.2
70.1
70.1
68.1
72.2
70.1
70.1
68.1
84.2
70.1
84.2
86.2
86.2
86.2
86.2
84.2
100.2
78.1
84.2
100.2
100.2
100.2
100.2
114.2
98.2
114.2
92.1
Fresh Gasoline
0.0001
0.0122
0.0629
0.0007
0.0000
0.0006
0.1049
0.0000
,0.0000
0.0000
0.0586
• •' 0.0000
0.0044
0.0000
0.0049
0.0000
0.0000
0^0730
0.0273
0.0000
0.0283
0.0083
0.0076
0.0076
0.0000
0.0390
0.0000
0.0000
0.0063
0.0121
0.0000
0.0055
0.0550
Weathered Gasoline
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0069
0.0005
0.0008
0.0000
0.0095
0.0017 '
0.0021
0.0010
0.0000
0.0046
0.0000
0:0044
0.0207
0.0186
' 0.0207
0.0234
0.0064
0.0021
0.0137
0.0000
' 0.0355
0.0000
0.0447
6.0503
0.0393
0.0207
0.0359
                                     29

-------
Table 3.    Composition of fresh and weathered gasolines (from USEPA, 199la; after
           Johnson et al., 1989) (continued).
Compound Name
2,3,4-trimethylpentane
3-methylhetane
2-methylheptane
n-octane
2,4,4-trimethylhexane
2,2-dimethylheptane
ethylbenzene
p-xylene
m-xylene
3,3,4-trimethylhexane
o-xylene
2,2,4-trimethylheptane
n-nonane
3,3,5-trimethylheptane
n-propylbenzene
2,3,4-trimethylheptane
1 ,3,5-trimethylbenzene
1 ,2,4-trimethylbenzene
n-decane
methylpropylbenzene
dimethylethylbenzene
n-undecane
1 ,2,4,5-tetramethylbenzene
1 ,2,3,4-tetramethylbenzene
1 ,2,4-lrimethyl-5-ethylbenzene
n-dodecane
naphthalene
n-hexylbenzene
methylnaphthalene
TOTAL
MW(g)
114.2
114.2
114.2
114.2
128.3
128.3
106.2
106.2
106.2
128.3
106.2
142.3
128.3
142.3
120.2
142.3
120.2
120.2
142.3
134.2
134.2
156.3
134.2
134.2
148,2
170.3
128.2
162.3
142.2

Fresh Gasoline
0.0121
0.0000
0.0155
0.0013
0.0087
0.0000
0.0000
0.0957
0.0000
0.0281
0.0000
0.0105
0.0000
0.0000
0.0841
0.0000
0.0411
0.0213
0.0000
0.0351
0.0307
0.0000
0.0133
0.0129
0.0405
0.0230
0.0045
0.0000
0.0023
1.0000
Weathered Gasoline
0.0000
0.0343
0.0324
0.0300
0.0034
0.0226
0.0130
0.0151
0.0376
0.0056
0.0274
0.0012
0.0382
0.0000
0.0117
0.0000
0.0493
0.0705
0.0140
0.0170
0.0289
0.0075
0.0056
0.0704
0.0651
0.0000
0.0076
0.0147
0.0134
1.0000
                                      30

-------
                                         (a)
   c
  _o
  ,+-_/
   00
  •4—1
   c
   CD
   O
   C
   O
  O
                                         (b)
         Retention Time
Retention Time
Figure 10.    Chromatograms of (a) regular gasoline, (b) unleaded gasoline, (c) kerosene

             heating fuel, and (d) diesel fuel. Note that the more volatile fuels (regular and
             unleaded gasoline) contain a significant percentage of compounds that elute early
             during the GC scan.  [From Preslo and Stoner,  1991, in Practical Handbook of
             Ground Water Monitoring, D.M. Nielson, ed., Ch. 3; reprinted by permission of
             Lewis Publishers,  an imprint of CRC Press, Boca Raton, Florida.]
                                          31

-------
           t =  1 day
                                                      01
                                                      c
                                                      JS!
           t = 30 days
       c
       .0
       o

       o
       O
01
g
                              OI
                              m
          Retention Time
Figure 11.     Chromatograms of gasoline before and after venting. Note the decrease in

              concentration for many of the lighter compounds (after Baehr et al, 1989).

              [Reprinted by permission of Elsevier Science.]
                                           32

-------
                                                             Naphthalene
   LU
   cc
   Q
   til
   N
   CC
   O
            0  23  48  73  98123148173198223248273298323348373398

                                        TIME, days
Figure 12.    Comparison of relative removal rates of compounds of different volatility from a

             fuel mixture. Note that light compounds such as benzene are removed fairly

             quickly, while semivolatile compounds such as naphthalene are fairly resistant to

             SVE (results are from the VENTING model).


                                         33

-------
models (VENT2D and VENT3D) for simulating SVE with multicomponent mixtures at sites
with complex vadose zone properties and boundary conditions.  The transport equation is
coupled with phase equilibrium and vapor flow equations in these models and solved in two and
three dimensions using conventional, explicit finite-difference methods.

2.2.11 What cleanup levels can be achieved, and will these be within regulatory limits?

       Potential cleanup levels can be estimated by applying the methods discussed in Section
2.2.10.  Depending on site complexity, data availability, and analysis objectives, different levels
of modeling will be appropriate for cleanup level analysis, ranging from use of screening models
to compositional flow and transport simulation. Model results commonly suggest a faster rate of
cleanup than is achievable in the field when phenomena such as contaminant desorption, NAPL
dissolution, air flow velocity variation, and matrix diffusion are not adequately represented in the
model.  At many sites,  concentrations and mass removal will trend toward an asymptote that may
be above desired cleanup levels. This is particularly true at sites with NAPL presence, making
cleanup level predictions difficult.

       There are no regulatory limits specifically for VOCs in soil gas.  Regulatory limits are
typically set for soil concentrations to be protective of groundwater.  Monitoring SVE offgas
quality can provide an indication of system performance, but can not be used to demonstrate
attainment of soil cleanup goals. Rather, post-remediation soil samples should be collected to
determine the degree of soil cleanup provided by SVE.

2.2.12 What is the estimated cleanup time?

       Cleanup time is difficult to estimate due to phenomena such as NAPL dissolution, matrix
diffusion, air flow velocity variation, and contaminant desorption.  It also depends on system
design and operation (e.g. SVE well density and placement, surface control, extraction rates and
schedules). Even a well-designed SVE system can suffer from short-circuiting and dilution
effects.  Short-circuiting (also known as bypassing) occurs when air flows through preferential
permeable pathways to an extraction well, bypassing portions of the contaminated zone. This is
a particular problem in heterogeneous and relatively fine-grained media. Dilution frequently
limits SVE efficiency at sites where only a fraction of the air flow contacts contaminated media
due to an irregular contaminant distribution. These limitations to SVE performance can be
minimized by conducting an adequate site characterization, and then using models to assess ,(1)
optimum locations and screen intervals for extraction and injection wells, and (2) optimum
extraction/injection rates and schedules to maximize air flushing and minimize air bypassing and
stagnation.

       Contaminant removal will be limited by diffusion from dead-end pores and low-
permeability zones at sites with heterogeneous media. Typically, three stages of contaminant
removal behavior can be described: (1) rapid removal of readily-accessible contaminant mass by

                                           34

-------
 advection; (2) transition between advection-dominated and diffusion-dominated removal- and (3)
 relatively slow removal of contaminant mass diffused from inaccessible (to air flow) areas of the
 subsurface. Diffusion rates are difficult to estimate, especially for multicomponent mixtures
 because the rate of diffusion varies over time with the change in composition caused by
 preferential removal of the more volatile fractions. Rate-limited contaminant removal can be
 estimated using a numerical flow and transport model. A similarly difficult case to model is that
 of air flow parallel to, but not through, a zone of contamination (see, for example, Buscheck and
 Peargm, 1991; Johnson et al., 1990a; Rodriguez-Maroto and Wilson, 1988; Wilson et al, 1987).

        Another reason that cleanup time is difficult to estimate is that the initial contaminant
 mass in place is usually highly uncertain, particularly at sites where NAPL is present. Cleanup
 times may be severely underestimated by an analysis that underestimates or does not account for
 NAPL mass.

        Despite these difficulties, models remain a cost-effective tool for estimating cleanup
 times. Analyses can be used to estimate cleanup time by projecting mass remaining in place with
 time given a specific set of SVE operating parameters (e.g., well numbers and extraction rates)
 Alternatively, modeling can be conducted to determine combinations of SVE operating
 parameters that are needed to reach cleanup levels within a specified time period. Critical review
 should be made of the simplifying assumptions applied in any model used to assess cleanup time
 In many cases, these assumptions will lead to underestimation of cleanup times.  More complex
 analyses (e.g., using numerical flow and transport  models) generally incorporate fewer restrictive
 assumptions and,  therefore, usually provide a more powerful analytical tool than simpler models,

 2.2.13 What monitoring is necessary to verify that SVE is working?

       Monitoring, modeling, and remediation are part of the iterative  process of understanding
 and remediating a site.  Monitoring data are needed to build and validate a site model.
 Remediation stresses the site, allowing valuable data to be collected that should improve site
 understanding and thereby facilitate model refinement.

       Models can be used to help design several aspects of an SVE monitoring program.
 Contaminant concentration projections can help establish sampling schedules for evaluating SVE
performance. This may be particularly useful during the early period after SVE startup when
changes in contaminant concentrations occur relatively quickly.  Simulations can also be used to
estimate what semivolatile constituents may remain in soil after completing SVE.  Simulated
pressure and air flow distributions provide insight to determining where monitoring wells should
be located and how frequently vacuum measurements should be made to assess induced flow
patterns and rates.

      Guidance on monitoring the performance of SVE systems is provided by Johnson et al
(1990b) and USEPA (1991; 1994). SVE parameters recommended for  monitoring include:

                                          35

-------
Vapor flow rates at each injection and extraction well.

Pressures at each injection and extraction well, and at several radial distances
from extraction wells to evaluate the vacuum distribution and air flow patterns.

Extracted vapor concentrations and composition, which can be used with
extraction flow rates to determine mass removal rates and the trend of cumulative
mass removed.

Temperature, both ambient and soil, and other parameters (e.g., CO2in soil gas)
that may indicate enhanced biodegradation due to SVE.

Water-table elevation measurements to determine the extent of upwelling induced
by SVE. It may be desirable or necessary to depress the water table by pumping
so that extraction well screen intervals and target contamination zones remain in
the vadose zone.

Soil-gas concentrations and composition at several radial distances from the
extraction well.

Analyses of soil samples at the end of SVE to determine the nature and
distribution of contaminants left in place.
                               36

-------
                                       SECTION 3
                FLOW AND TRANSPORT PROCESSES AND EQUATIONS

       Air flow and vapor contaminant transport in the vadose zone are simulated to examine
 many aspects of SVE as discussed in Section 2.2. Processes that may significantly affect vapor
 flow and transport, depending on site-specific conditions, include advection, mechanical
 dispersion, diffusion, mass transfer between phases (partitioning between air, water, solid, and
 NAPL),  and chemical reactions (e.g., biotic and abiotic transformations). During recent years,
 many investigators have developed and applied models to examine the influence of these
 processes on air flow, vapor-phase contaminant transport, and SVE performance (e.g., Baehr,
 1987; Baehr et al., 1989; Sleep and Sykes, 1989; Falta et al., 1989; Massman, 1989; Johnson'et
 al., 1990a; Mendoza and Frind,  1990; Gierke et al., 1990; Brusseau, 1991; Rathfelder et al.,
 1991; Baehr and Hult, 1991; Gierke et al., 1992;  Silka and Jordan, 1993; Falta et al., 1993;'
 Benson et al., 1993).  The basic subsurface processes affecting SVE performance, and equations
 implemented in models to describe these processes, are presented in this section.

       Vapor transport occurs in response to pressure and concentration gradients.  Advection is
 the transport process whereby gas moves as a fluid continuum driven by pressure gradients.  It
 occurs as bulk flow in which a mixture of gases behaves as one gas with no tendency to migrate
 according to constituent concentrations. SVE imposes pressure gradients in the vadose zone,
 thereby inducing advective gas flow to extraction wells. Similarly, advective flow can be
 established by venting from manmade structures  such as basements and sewers. Gas pressure
 gradients also arise from density differences between vapor with a large concentration of high
 molecular weight contaminants (e.g., chlorinated solvents) and less contaminated vapor (Falta et
 al., 1989; Sleep and Sykes, 1989). Pressure gradients due to density differences decrease with
 decreasing VOC concentration and increasing distance from a contaminant source.  Density-
 driven gas flow is significant only in highly permeable media;  it causes contaminant vapors to
 sink and  then spread above the saturated zone or other barriers to gas flow. Barometric pressure
 fluctuations have little impact on air flow in the vadose zone. In the absence of artificial venting
 or density-induced advection, air flow may be negligible, in which case, vapor contaminant
 transport is dominated by concentration gradients (molecular diffusion).

       Advective laminar viscous fluid flow in porous media is usually described by Darcy's
 law, which relates the rate of fluid flow to a pressure gradient over distance multiplied by the
 media permeability. Permeability is primarily a function of the intrinsic permeability of the
porous medium and moisture content. Air permeability decreases as pore spaces become filled
with water.  The concept of relative permeability  is discussed in Section 4.1.  Darcy's law
 assumes  zero fluid velocity along pore walls.  With gases, however, a phenomenon known as gas
 slip flow, or the Klinkenberg effect, results in non-zero fluid velocities along pore walls and a
greater flow rate than predicted by the Darcy equation (Klinkenberg, 1941; Massman, 1989;
Baehr and Hult, 1991). The Klinkenberg effect is an enhancement of air-phase permeability that
increases with decreasing pressure and decreasing pore size. It can usually be ignored, however,
                                           37

-------
by models designed to simulate SVE due to its limited significance in coarse soils and under
normal SVE operating vacuums (Baehr and Hult, 1991; Brusseau, 1991).

       Mechanical dispersion and molecular diffusion, which result from air flow velocity
variations and concentration gradients, respectively, cause spreading and dilution of vapor-phase
contaminants. As with groundwater flow, velocity variations effecting dispersion of vapor
contaminants arise from differences in pore size, flow path length (which are scale-dependent),
and pore friction. Unlike dissolved contaminant transport,  however, it appears that molecular
diffusion predominates over mechanical dispersion for vapor-phase transport (Brusseau, 1991;
Benson et al., 1994), except perhaps where air velocities are very high (e.g., close to an SVE
well). The difference in relative importance of mechanical dispersion versus molecular diffusion
between  groundwater and air systems occurs because gas-phase diffusion coefficients are much
larger than aqueous-phase diffusion coefficients (e.g., 8x10'2 cm2/s for a typical VOC  in open air
compared to 8x10"6 cm2/s in water, see Appendix A).

       Diffusion is the process through which constituents in a single phase equilibrate; thus,
given sufficient time, it will thoroughly mix the components of any gas-filled space (Kreamer,
1986). Diffusion occurs due to random molecular motion whereby each molecule is constantly
colliding with other molecules and changing direction. This random motion effects a net transfer
of molecules from areas of high concentration to areas of low concentration.

       The rate of mass transfer during diffusion follows a predictable pattern that is governed
by the concentration gradient.  Pick's law relates diffusive flux to the concentration gradient over
distance  multiplied by an empirical coefficient called the effective diffusion coefficient. Many
vapor transport models employ the Millington and Quirk (1961) functional relationship to
calculate the effective diffusion coefficient based on a chemical-specific free-air diffusion
coefficient, media porosity, and fluid content as described in Section 4.2.2. The magnitude of
diffusion, therefore, depends on concentration gradients and properties of both the porous
medium  (e.g., porosity, moisture content) and the diffusing gas contaminant.  Although more
detailed equations are available to account for complexities of gas-phase diffusion (Youngquist,
1970; Thorstenson and Pollock, 1989), using Pick's law to  simulate diffusion in the vadose zone
is generally appropriate and adequate (Brusseau, 1991).

       The transport of vapor contaminants in the vadose zone may be greatly affected by mass
transfer between phases and transformation reactions. Partitioning processes and effects are
addressed in Section 4.2. A discussion of the control exerted by advection and diffusion on SVE
performance follows.

3.1    ADVECTION AND DIFFUSION DURING SVE

       Under ideal conditions, VOC removal from the subsurface via SVE is dominated by
advection.  In portions of most sites (e.g., low-permeability zones), however, the rate of VOC
removal  is diffusion-limited, thus slowing contaminant recovery. Heterogeneous soils contain

                                           38

-------
isolated pores and relatively fine-grained materials that tend to have high water saturations. Air
stagnates in these zones due to low air permeability. As such, advective vapor movement is
negligible and contaminant migration will be controlled by molecular diffusion; thus restricting
the effectiveness of SVE.

       In a soil where some regions allow transport of VOCs by advection and other regions
limit VOC transport to diffusion, the portions dominated by advection will be flushed of resident
VOCs much more quickly during SVE operation than those dominated by diffusion. Under these
conditions, the contaminant removal rate will be high during the advection-dominated early stage
of SVE operation. Once the advection-dominated region has been flushed by the inflow of
cleaner air, contaminant concentrations in the extracted gas will exhibit considerable tailing as
VOCs slowly diffuse from stagnant regions of the soil into advection-dominated regions. If an
SVE system is shut off, vapor concentrations can be expected to rise in the permeable zones due
to this diffusion.  Similar concentration tailing and rebound phenomena are observed during
groundwater pump-and-treat remediation (National Research Council, 1994).  As described
below, both field and modeling studies evidence the characteristic trend of contaminant
concentrations in  extracted soil vapor that results from the transition to diffusion-limited mass
transport.

       Hiller and Gudemann (1989) describe three phases of contaminant concentrations in
extracted gas during SVE:  the initial short-term, advection-dominated phase (Stage I); the
transition from advection to diffusion control (Stage II); and the ultimate, diffusion-dominated,
long-term period (Stage III). Figure 13 presents field data from several sites which exhibit these
stages. During Stage I (advection-dominated), contaminant-saturated soil gas is removed from
readily accessible pore spaces and evaporation may be occurring from NAPL. High contaminant
concentrations in the extracted gas decline quite rapidly during this stage. The transitional Stage
II period typically occurs when contaminant concentrations in the extracted gas have decreased
by more than 80%. Stage III is characterized by the relatively slow processes of contaminant
diffusion from dead-end pores and stagnation zones, evaporation from the dissolved phase,  and
desorption from the solid phase. Contaminant concentrations in the extracted gas asymptotically
approach some limiting value during Stage III. Based on data collected from several hundred
SVE systems, this general contaminant concentration trend is commonly observed regardless of
soil type and other parameters (Hiller and Gudemann, 1989).

       A simple mathematical model was developed by Silka et al. (1989) to perform a
conceptual  analysis of SVE from a permeable layer bounded above and below by low-
permeability layers. Simulated contaminant transport from the low-permeability layers was
limited to and by diffusion.  As reflected by the simulation results in Figure 14, although
contaminant mass resident in the permeable zone is quickly removed by SVE, slow diffusion
from the  bounding layers causes long-term tailing of the extracted vapor contaminant
concentration curve.
                                           39

-------
 E
 Q_
 Q_
O
2:
o
O
UJ
CD
GO
CD
       400
          0
            STAGE I
            ADVECTION-DOMINATED
                                            STAGE  III
                                       DIFEUSIQN-DOMINATED"
BIN
     0    30    60    90   120  150   180   210
                                     ELAPSED TIME,  days

                  ULM   o  RED5    4 STG5  . * WUE4    o WUE8
                                                                            300

SOIL
CONTAMINANT
VOLUME FLOW
RANGE OF INFLUENCE
BERLIN
(BLN)
medium & fine
grained sands
PCE
390 CFM
180ft
WURZBURG
(WUE4)
clayey silts
w/limestone
fragments
PCE
68 CFM
40ft
WURZBURG
(WUE8)
clayey silts
w/limestone
fragments
PCE
70 CFM
40ft
REDWITZ
(REDS)
sandy-clayey
silts
PCE
50 CFM
60ft
STUTTGART
(STG5)
weathered
claystone, silt
TCE
75 CFM
50ft
ULM
(ULM)
coarse
grained fill,
silty sands
PCE
103 CFM
75ft
 Figure 13.    Illustration of the concept of stages of VOC removal during SVE. Stage I is
             advection-dominated and characterized by rapid removal of VOCs. Stage II is a
             transition period from an advection-dominated system to a diffusion-dominated
             system. Stage III is a diffusion-dominated state characterized by mass removal
             rates which asymptotically approach some limiting value (after Hiller and
             Gudemann, 1989).
                                      40

-------
   CD

cc rt
oo
                          CONCENTRATION

-------
       Benson et al. (1993) present results from a modeling study that compares contaminant
removal from homogeneous and layered media. The layered system contains a low-permeability
zone. Simulated contaminant mass-in-place versus time curves for each system are similar for
the early period of SVE operation during which a large portion of contaminant mass is removed
(Figure 15). As shown, the decline of the mass-in-place curve for the layered system lags behind
that of homogeneous medium. This occurs because the slow diffusion of contaminants from the
low-permeability layer reduces contaminant concentrations in the extracted gas. As a result, the
low-permeability layer acts as a long-term source of volatile contaminants, roughly doubling the
time (from 500 to 1000 days) required to reach the 90% cleanup level in this diffusion-limited
scenario.

       An exponential decay equation is sometimes used to describe and/or project the trend
over time of declining SVE mass removal rates or contaminant concentrations (Buscheck and
Peargin, 1991; API, 1991). The equation used to match mass removal rate or contaminant
concentration data over time is, written for mass removal rate, of the form M, = (Mo-M-Je^ + Ma,
where Mt is the mass removal rate, M0 is the initial mass removal rate, Ma is the asymptotic mass
removal rate, k is a constant fit to the empirical data, and t is time. At sites where mass transport
is limited by diffusion, use of the exponential decay equation to predict cleanup time based on
early time data may result in substantial underestimation of cleanup time (Figure 16).
Concentration or mass removal rate tailing may indicate the need for long-term SVE operation or
that cleanup goal attainment is impracticable.

3.2    BASIC EQUATIONS  .

3.2.1   Equations of Air Flow

       In this section,  mathematical equations are presented that describe the processes
discussed above. Once developed, the equations must be solved to obtain meaningful results.
Two examples of mathematical solutions for the air flow equation are discussed below. The first
technique poses the air flow equation in the same form as the groundwater flow equation, then a
numerical groundwater flow model is used to solve the equation. The second approach discussed
is an analytical solution for the air flow equation, based on some simplifying assumptions.  Other
analytical solutions are available (e.g., Shan et al., 1992; Marley et al., 1990; and others), but are
not presented in detail  here.

      The following derivation is adapted from Baehr et al. (1993)  and outlines the equations
of air flow as  programmed in the AIR3D code (Baehr et al., 1993).  Their approach is to change
variables to cast the air flow equation into the same form as the groundwater flow equation.  The
equation can then be solved using an existing numerical groundwater flow model, such as
MODFLOW (McDonald and Harbaugh, 1984). A more detailed derivation, including discussion
of the assumptions, is presented in Baehr and Hult (1991).
                                          42

-------
                                         HOMOGENEOUS PERMEABILITY

                                                 LAYERED SYSTEM
                   10'
                                200
400      600
 TIME (days)
800
1000
Figure 15.    Comparison of simulated TPH mass-in-place versus SVE duration in
             homogeneous and layered systems.  Due to slow contaminant diffusion from the
             low-permeability layer, it takes much longer for SVE to reach the 90% cleanup
             goal in the layered system (from Benson et al., 1993).  [Reprinted by permission
             of Ground Water Publishing Company, Dublin, Ohio.]
                                         43

-------
     0.000
                  20
                           40
                                                            Field Data

                                                            Exponential Fit to Early Data
                                    60
                                             80       100
                                              TIME, days
                                                              120
140
         160
                 18'
Figure 16.    An exponential decay curve fit to early time SVE data may underestimate later
              concentrations (or mass removal rates) where contaminant transport is limited by
              diffusion (from Hutton, 1990).

                                            44

-------
       Note that the terms "gas", "air", and "vapor" are used interchangeably in this document;
and that the subscripts "g", "w", and "N" refer to the gas, water, and NAPL phases, respectively.

       The conservation of mass equation for air flow in a variably-saturated porous medium is
given by the following:
     !
                                                                                    (i)
where pg is the density of soil gas [M/L3], 6g is the volumetric air content [L3/L3], t is time [T],
and"qg is the specific discharge vector for air [L/T], or
        dp
      6 -^  + V


-------
                                                                                   (6).
 The ideal gas law is assumed as the equation of state relating pressure and density, thus
 providing a relationship for air compressibility:
                                          CO P
                                        =   g  8
                                           R T
 (7)
where cog is the molecular weight of soil gas [M/mol], R is the universal gas constant [0.08206 L-
atm/K-mol][ML2 /molKT2], and T is the absolute temperature [K]. Substituting Equation (7) into
Equation (6), and integrating at constant co and T yields:
                                                                                   (8)
assigning hg = 0 at Pg = 1 atm.  Substituting Equation (8) and the ideal gas law (Equation
(7)) into Equation (3), and assuming isothermal conditions, gives Darcy's law as a function of
pressure:
                                                                                   (9).
       Substituting Equations (7) and (9) into Equation (1), and then using the following
variable substitution for linearization suggested by Muskat and Botset (1931):
                                     *  = (PB):
(10)
yields the following three-dimensional air flow equation in Cartesian coordinates that is
analogous in form to the groundwater flow equation, which has been solved numerically in such
groundwater flow codes as MODFLOW:
                                                                                  (11)
where x, y, and z  are Cartesian coordinates aligned along the major axes of the permeability
tensor, with diagonal components kgxx, kgyy, and k^, and Sg is the pneumatic equivalent of
specific storage [I/1].

                                           46

-------
       The differential equation governing air flow in porous media is nonlinear because air
density increases with air pressure.  However, if the difference in air pressures within the flow
field does not exceed 0.2 atm, the linear differential equation used to simulate groundwater flow
provides a very close approximation of air flow (Massman, 1989).  Most SVE systems operate
within this range of pressure difference. Thus, by changing input variables (e.g., air conductivity
for hydraulic conductivity and air pressure for hydraulic head), groundwater flow models can be
employed without modification to simulate SVE. For greater pressure differences, the pressure
squared linearization given by Equation (10) can be implemented in groundwater models to
accurately approximate the nonlinear air transport equation (Massman, 1989).

       Air phase permeability is assumed to be independent of pressure. Therefore, the
Klinkenberg slip effect (Klinkenberg, 1941) can only be modeled as constant with respect to
pressure. If the air-filled porosity is constant with respect to time (i.e., water movement is
neglected), then:
                                     S8 =
                                                                                  (12).
The change of variable 4> = (Pg)2 results in a linear equation for steady air flow. The transient
equation is linearized by assuming /(j> = PAtm in the definition of Sg above, where PAtm is the
prevailing atmospheric pressure.

       Equation (11) can be directly compared to the linear groundwater flow equation. This
method of analysis was also previously described by Massman (1989).  The simplifying   !
assumptions needed to arrive at this linear air flow equation are summarized below:

       •     Darcy's law is valid for air flow where permeability variations due to changes in
             water content are neglected;

       •     Elevation component of pneumatic head is neglected;

       •     Isothermal conditions are assumed;

       •     Ideal gas law is assumed as model for compressibility;

       •     Klinkenberg effect is neglected;

       •     Water movement and consolidation are neglected, therefore, air-filled porosity is
             constant with respect to time; and

                 = PAtm in the definition of storage coefficient Sg.
                                           47

-------
 Baehr and Hult (1991) and Massman (1989) found that the linear air flow model given by
 Equation (11) provides a reasonable working model for SVE, based on the assumptions stated
 above.

 3.2.2  Flow Equation Solution

       MODFLOW solves the flow equation using a numerical method referred to as the finite
 difference technique.  Other numerical techniques include the finite element method and the
 method of characteristics. These methods are discussed in Mercer and Faust (1992), Anderson
 and Woessner (1992), and many other modeling textbooks. Alternatively, the flow equation may
 be simplified and solved analytically.  Just as numerical groundwater flow models can be used to
 solve the analogous air flow equation, analytical solutions for groundwater (e.g., Theis and
 Hantush) can also be used for SVE under certain flow conditions.

       Johnson et al. (1990a,b) present an analytical solution to the air flow equation for steady-
 state air flow to an extraction well from which the following discussion is adapted.  Fpr radial-
 symmetric air flow through a confined porous stratum of thickness m [L], the governing
 equations are:
-y	-V- (Pgm
-------
                                  Pg    PgAtm
                                              • Atm'
                                                                                    (15)
where pgAtm is the gas density [M/L3] at atmospheric pressure. Substituting Equations (14) and
(15) into Equation (13) produces:
  20 n \ 3P
  —£l 	£ = V2P
   v-/  a
                                                                                    (16).
       Pressure can be expressed in terms of atmospheric pressure, PAtm, and a deviation, P',
from this pressure.  P7 is equivalent to the vacuum that would be measured in the soil.  If this
conversion is made in Equation (16), and if the product Pg2 relative to the product PAtmP' is
neglected, then the resulting equation for radial flow is:
                                             r 3rV  5r
                                                                                    (17).
       The solution to Equation (17) for the following boundary conditions (constant flow rate):
                                P'  = 0
                             lim

                             r •-» 0 I  r
is given by Bear (1979, p.320):
                           P  =
47rm(kg/ng)
                                                    (18)
    .3P.M  _
      dr
              oo
              /   *
dx
                                                                                     (19)
                                                     x
where Qg is the volumetric flow rate to the vapor well [L3/T]. Solutions to the integral in
Equation (19) are also given in Bear (1979). The behavior of the integral is such that for
(r26g|j,g/4kgPAtmt) O.001, its value is very near to the asymptotic limit.

       Estimates for air flow rates, pressure distributions, and air velocities in unsaturated soils
can be calculated from the steady-state solution to Equation (16). Equation (16) can be solved
                                            49

-------
 subject to the assumptions that air flow is steady (i.e., the flow rate does not vary over time) and
 radial (i.e., all air flow is horizontal and inward toward the extraction well). Note that these
 simplifying assumptions are invoked to arrive at a closed-form analytical solution.  For the case
 of steady-state, radial flow subject to the boundary conditions:
                             P  = P
                             r    ^
                at
                at
 r  ~ R
r =R1
                                                                                   (20)
the solution to Equation (16) for homogeneous or layered soil systems is:
PVrt— P2  =
 gS'  rsw
                                                                                    (21)
where PgW is the pressure at the vacuum well [M/L-T2],which has a radius Rw [L].  Rx is the
radius of influence [L], where the pressure, by definition, is equal to the ambient atmospheric
pressure PAtm.  The corresponding solutions for the radial Darcian velocity distribution qg(r)
[L/T], and the volumetric air flow rate Qg, in homogeneous soil systems are (Johnson et al.,
1990a):
                        - -  il
                                            8W
                             /P '  \5

                           -  f^
                          	rgW  •
                                                                1/2
                                                                                    (22)
and
                                                                                    (23)
where H is the length of the extraction well screen that is open to the unsaturated zone [L].  For
layered soil strata, QE is equal to the sum of the air flow rates for all soil layers:
                           Qg =£
                                                         (24)
where"^[gi and Hj are the Darcian air velocity [L/T] and thickness of each soil layer [L] i through
which the extraction well is screened, respectively.

       To convert actual air flow rates to equivalent standard air flow rates, Qg* (P = 1 atm), the
following correction factor is used:
                                            50

-------
                                                                                   (25)
                                            " Atm
where PgW is the absolute pressure expressed in the same units as P
                                                             Atar
       Note that Equation (21) appears to imply that the pressure distribution is independent of
soil properties.  However, this is misleading because the radius of influence, RI5 is an empirical
parameter that is a function of soil properties.  The radius of influence is also a measure of the
validity of the radial air flow assumption. Smaller values of Rj may indicate increasing vertical
air flow, such as short-circuiting caused by air flow from the surface.  This could occur when an
extraction well is not properly sealed at the surface, or when the well is screened over a shallow
interval beneath an unpaved surface. Regardless, it turns out that the radial pressure distribution
is fairly insensitive to significant changes in Rj. Air flow rates are slightly more sensitive to
changes in Rl5 because Qg is a function of the pressure gradient.  For a 4-inch diameter vacuum
well, the air flow rate estimation decreases by 20% when Rt increases from 49 to 200 feet. For
most venting applications, typical Rr values will range from 30 to 100 feet (Johnson et al.,
1990a,b).

       Recently, a computer program called GASSOLVE (Falta, 1995) was developed to
analyze soil gas pump tests under the following conditions:  (1) transient compressible gas flow
to a partially penetrating well in which the ground surface is open to the atmosphere; and (2)
transient compressible gas flow to a partially penetrating well bounded above by a leaky
confining layer.  Pump tests involving steady-state conditions, fully penetrating wells, or tight
confining layers are considered as special cases of these solutions. In GASSOLVE, field gas
pressure data from multiple observation wells are inverted using a multidimensional nonlinear
optimization routine to provide a best fit of the data with the selected analytical solution.

       A number of different unit conventions are used for representing air pressure. Tables 4a
and 4b present conventions for one atmosphere (the standard condition at the ground surface),
and conversion factors for several common systems of pressure units.

3.2.3   Equations of Vapor-Phase Contaminant Transport

       The conservation of mass for an individual chemical component within a mobile vapor
phase in the presence of relatively immobile soil moisture, adsorbed, and NAPL phases is given
in vector format by Baehr and Hoag (1988):
V • (C g  -  D VC )  = — (0 C  + 6 C  + p,C .  + x.9M )
    v  g^g     g  g'    ai. v g  g     ww    ^bads     i Nm7
                                                                                     (26)
                                            51

-------
Table 4a. One atmosphere, as represented in several commonly used unit systems.
Units
Atmospheres (atm)
Millimeters of Mercury (mm Hg)
Inches of Water (in H2O)
Torr
Bar
Pascal (Pa)
Atmospheric
Pressure
! ' ' • '
760
406.78
760
1.013
1.01 3x1 05
Table 4b.  Pressure unit conversion factors.
M
U
L
T
1
P
L
Y
BY
atm
mm Hg
inH2O
Torr
Bar
TO GET
Pa
1.01 3x1 05
133.3
249.0289
133.3
105
Bar
1.013
1.33x10-3
2.49x1 0-3
1.33x10'3

Torr
760
1
1.868


inH2O
406.78
0.5352



mm
Hg
760




                                    52

-------
where Cg is the vapor phase concentration [M/L3], Cw is the dissolved concentration [M/L3], Cads
is the mass basis solid phase concentration [M/M], 6W  is the volumetric water content [L3/L3],
6Nm is mass based NAPL content [M3/M3], xf is the mole fraction of the compound in the NAPL
phase, pb is the soil bulk density [M/L3], Dg is the gas phase dispersion tensor [L2/T], and r(t)
represents sources and sinks that may vary with time [M/L3-T].

       Rectangular coordinates are a more general (and useful) form, so the left side of (26)
becomes:
                                                         = 1,2,3
(27)
where x refers distance rather than mole fraction in Equation (27) only. All the chemical and
transport models reviewed solve some portion of Equation (26). For example, VENTING
(ES&T, 1994) and Johnson et al. (1990) set the left-hand side and r(t) to zero, while Mendoza
and Frind (1990) set 0N and r(t) to zero to solve (26) in cylindrical coordinates. Almost all
models assume a linear relationship among Cg, Cw, and Cads, although this is not necessary.

       Retardation arises from chemical partitioning into multiple phases. The general
expression for equilibrium-based retardation of the ith component is given, through chain-rule
manipulation of the right-hand side of Equation (26), by:
                         =
                                                                                   (28)
where R^ is the dimensionless retardation coefficient for component i. If any terms in the right-
hand side of Equation (28) are not fully linear with Cg, then the retardation will vary in time and
space and Equation (28) must be solved explicitly. It also must be solved iteratively (subject to
SXj = 1) if a NAPL is present.  An intuitive look.at (28) indicates that sparingly soluble
chemicals will derive a significant portion of their retardation from the large relative mass in the
NAPL term when present. If linear sorption and liquid/gas partitioning are assumed (as is
common), then:
                        R,= =  1 +
                                            p.K
                                            "b i
                                             b d
(29)
where KHd is Henry's law constant [dimensionless], Kd is solid-liquid partition coefficient [L3/M],
R is the universal gas constant [0.08206 L-atm/K-mol][ML2 /molKT2], T is the absolute
temperature [K], and Pjv is the pure-component vapor pressure of compound i [M/L-T2].
                                           53

-------
The simplest (and numerically smallest) expression is found when the NAPL term disappears:
                            R  =  1 +
                                                     (30).
3.2.4  Transport Equation Solution

       Numerous approaches have been taken to solve the equation of mass transport by soil
vapor. Falta et al. (1993) use a semianalytical method for modeling advective transport of a
VOC to a soil vapor extraction well. Mendoza and Frind (1990) solve the full advective-
dispersive mass transport equation in radial coordinates. The application of this solution is
limited to radially symmetric problems. Their model considers retardation and dispersion of
contaminants in the vapor phase. The governing equation for advective-dispersive mass
transport in radial-symmetric coordinates is given by (Mendoza and Frind, 1990; Bear, 1972):
            lieb  S
            r  8r  Is •"  dr
                r  9z
    JL.S
    r  dr
                       T 8 •
dCg
~9zT
                 9C
5z
                                   -  6v
                                          BC
1  .1
r  3z
                    5C
                                                                                   (31)
                  a
                 f
where r and z are the radial and vertical coordinate directions, and vgr and vgz are the average
interstitial gas velocities [L/T] in the r and z directions, respectively. The dispersion tensor (Dgij)
is defined as (Mendoza and Frind, 1990; Bear, 1979):
                                                                                   (32)
where vg is the interstitial gas velocity [L/T], vgi and Vgj are velocity components in the r and z
directions [L/T], «L and KT are the longitudinal and transverse dispersivities [L], T is the
tortuosity of the medium [dimensionless], tfg is the free-air diffusion coefficient [L2/T], and 6^ is
the Kronecker delta [dimensionless].

       The Mendoza and Frind (1990) model is a single-component transport code; it cannot
simulate multicomponent mixtures such as gasoline.  Single-component models are appropriate
for simulating SVE where the contamination is comprised of a single contaminant and where no
NAPL is present. In order to apply a single-component model to a multicomponent spill, the
user must pick one or more indicator compounds and run the model once for each compound.
                                           54

-------
       Benson et al. (1993) solve the advective-dispersive equation by neglecting mechanical
dispersion in the vapor phase for the VENT2D model. This allows use of a simpler numerical
formulation to solve the transport equation, thus enabling simulation of multicomponent
mixtures subject to retardation and diffusion.  Inclusion of the mechanical dispersion term gives
rise to nonlinearities in the solution of the mass-transport equation, because mechanical
dispersion is a function of velocity. There is some disagreement, however, as to the degree of
error introduced into the calculations by neglecting mechanical dispersion (see Walter, 1994; and
Benson et al., 1994).  The approach of Benson et al. (1993) considers four-phase equilibrium
partitioning of contaminants into air, water, soil, andNAPL.  Approaches such as Mendoza and
Frind (1990) consider only three-phase partitioning (air, water, soil) and do not consider the
presence of NAPL. The four-phase approach is more robust for situations where NAPL is
present.  Several other authors (Baehr and Corapcioglu, 1987; Rathfelder et al., 1991; and others)
have also developed numerical models based on the four-phase partitioning approach.' As an
alternative to solving the general set of transport equations, solutions are available in the
literature for a number of idealized scenarios (e.g., see Johnson, 1990a).
3.2.5   Equilibrium Partitionir
ice Approach
       A third approach based on an equilibrium mass-balance analysis that considers the four
contaminant phases (air, water, soil, NAPL) identified hi Equation (26) is presented by Johnson
et al. (1990a,b). Portions of the VENTING and VENT2D models discussed in this document are
based on this method. While the mass-balance approach is not as rigorous as using the full
transport equations, it provides a means to quantify transport mechanisms during SVE.

       The following discussion is adapted from Johnson et al. (1990a,b). A total mole balance
on component i during SVE can be written as:
                                dM.
                                                                                  (33)
where: Mj is the total number of moles'of i in the soil in all phases (NAPL, soil moisture, sorbed
to soil, and soil gas); Qg is the volumetric flow rate of air into the extraction well [L3/T]; e is a
dimensionless efficiency factor that can be used to account for nonequilibrium effects; CgMi is the
equilibrium molar concentration of i in the air entering the vacuum well [mol/L3]; and t denotes
time.  The rate of degradation of i, whether due to biological or chemical processes, is lumped
into the term Bj [M/T] in Equation (33).

       Equation (33) can be solved by determining the relationships among M{, CgMi, and Bj.
The differential equation describing advection and diffusion can be solved to calculate a detailed
solution for Equation (33), however this demands significant information regarding the
subsurface structure and spatial distribution of the contaminant. Several models that solve the
equations of contaminant transport in a more rigorous manner are discussed briefly in Section 5.
For our purposes here, a simpler (but less accurate) one-dimensional approach is taken, which

                                         •  55

-------
assumes that the contaminant is uniformly distributed throughout a given amount of soil at all
times. In addition, the vapor, NAPL, sorbed, and dissolved phases are assumed to be in
equilibrium at all times. This is a one-dimensional approximation of a three-dimensional
problem. These assumptions allow explicit expressions that relate Mj and CgMi to be developed.
Soil column venting experiments by Marley and Hoag (1984) indicate that an equilibrium
distribution assumption may be valid for pore air velocities in the range 1 to 7 cm/s.

       A total mole balance on component i can be used to develop an equilibrium relationship
between C^ and M} at any time t:
       P0V
M. = z-a-£-2. +
       1  RT
                                            p 0 V
                                            "w w c
                                              0)
                                                          CO
                                                                                (34)
where: z,- is the mole fraction of compound i in the gas phase [dimensionless]; Vc is the total
volume of the contaminated soil zone [L3]; x; is the mole fraction of component i in the NAPL
[dimensionless]; MN is the total number of moles in the NAPL; y; is the mole fraction of i in soil
moisture [dimensionless]; cow is the molecular weight of water [18 g/mol]; and the other
parameters are defined above and in the symbols listing.

       Solid-liquid soil partition coefficients can be estimated using the relationship given by
Karickhoff(1980):
                                     = 0.63 K w.
                                                                (35)
where K,,W)i is the octanol-water partition coefficient [(mass i/mass octanol)/(mass i/mass H2O)],
and f,,,. is the organic carbon fraction in the soil [M/M]. Additional information on measurement
and typical values of Kow and foc, is presented in Lyman et al. (1982), Section 4.2, and Appendix
A.

       The right-hand side terms hi Equation (34) represent the number of moles of component i
in the vapor phase, in NAPL, dissolved in soil moisture, and sorbed to soil particles, respectively.
The total number of moles in the soil moisture [(pw6wVc)^-cow] is assumed to be approximately
equal to the number of moles of water because hydrocarbons are typically only slightly soluble in
water.

       To calculate the equilibrium distribution between phases, assume that the vapor phase
behaves as an ideal gas, the NAPL behaves as an ideal mixture, and the soil moisture phase is
nonideal.  Then x,-, yi5 and zi are related by:
                                =      =
                                                                                  (36)
                                           56

-------
where P;v is the pure component vapor pressure of compound i [M/L-T2], and ^ is the activity
coefficient for i in water.  Both P;v and a; are functions of temperature, which is taken to be a
known, constant parameter throughout the region of interest. Equation (36) can be substituted
into Equation (34) to obtain an expression that relates M; and either xi5 z;, or CgMi. One form of
this equation is:
                        P.V6 V           p 8 V
                    x. -I iLJVli + MN  + Pw w  "
                         RT
                                  M.
                                (37)
where 6W > 0 and MN > 0. Recall that xb the mole fraction of i in the NAPL, is equal to:
                                    x. = MjN/MN
                                                       (38)
where MjN is the number of moles of compound i in the NAPL, and MN is the total number of
moles in the NAPL. The equilibrium distribution between all phases can be determined for any
set of MJ by iteratively solving Equations (37) and (3.8) subject to the constraint that £xi = 1. The
solution is simplified somewhat by assuming that the total number of moles in the soil moisture
[(pw0wVc)-=-cow] is equal to the number of moles of water.

       When solving Equations (37) and (38), determine whether or not NAPL is present (i.e.,
that Mj is  great enough). Otherwise, when all the M; values are small, the contaminant is
distributed among the soil moisture, sorbed, and vapor phases, and no NAPL liquid is present.
When no NAPL is present, Equation (37) reduces to:
                                           M,
                                                    KUP.VC
                                 RT
                                                       (39).
If all the products a^ calculated from Equation (39) are less than unity, for a given set of Mi,
then the equilibrium distribution does not include a NAPL.               '

       Equation (33) can then be solved numerically.  The procedure is to solve for each time
step 6t explicitly:
M.(t  + 6t)  =
                                             - 8t
Q x.P.
^g >  '
 RT
(40).
 A new equilibrium distribution is then calculated by solving Equations (37), (38), and (39) given
 the new M; (t + 6t). The equilibrium distribution is calculated by first determining the number of
 phases that are present by using Equation (39) and the condition that all a^j < 1.  When NAPL is

                                           57

-------
present, Equations (37) and (38) can be solved iteratively subject to the condition that £x; = 1.
At each time step, the air-filled pore space can be calculated based on the total porosity, the soil
moisture content, and the amount of NAPL. A value for biological or other decay can be
included via the additional term, Bj, in Equation (33).

       For each constituent, the simple model requires known values for the pure component
vapor pressure (Pjv), sorption coefficient (Kdi), activity coefficient in water (a;), and molecular
weight (CD,-).

       The vapor pressure of each component at ambient operating temperature can be
calculated from values for vapor pressure at a reference temperature (e.g., 20° C) and the boiling
point at 1 atm (TB). The Claussius-Clapeyron equation (Barrow, 1961) predicts vapor pressure
dependence on temperature T,  assuming that the vapor behaves as an ideal gas with a relatively
constant enthalpy of vaporization over the temperature range between TB and TR:
Piv(T) =
                                 exp
 (41)
where TR is a reference temperature [K] at which the vapor pressure is known, T is the absolute
temperature [K] at which the vapor pressure is to be calculated, and PB is the pressure at which
the boiling (TB) is measured.

       Vapor pressures may be affected by capillary forces that trap residual NAPL in the soil
matrix.  Barrow (1961) presents an equation for predicting the change in vapor pressure across a
curved surface with radius of curvature r:
                           ^(pore)  = P.V(T) exp -
                                2QV,
                                rRT
(42)
where Py (pore) is the modified vapor pressure, o is the surface tension [M/L-T2], Vm is the molar
volume [L3/mol] of species i, and R is the gas constant. For benzene at 293 K (o = 28.9 dyne/cm
and Vm = 80 cnWmol), the ratio of modified to unaffected vapor pressures considering pore radii
of 0.01 cm and 1x10"5 cm are:
                       PiV(pore)/PiV(T) = 0.99998  for r = 0.01 cm
                                    = 0.98151  for r = 1  x 10-5cm
                                                                (43).
       Because the radius of curvature of a liquid surface between soil particles is not likely to
be as small as 1x10'5 cm, the influence of surface tension on vapor pressures is negligible.
However, if the contaminant is trapped within micropores the diameter of which is smaller than
IxlO'6 cm, then the vapor pressure reduction will be significant.
                                           58

-------
       Activity coefficients in water (a;) can be estimated from tabulated solubility values (see
Appendix A). For a pure compound at its solubility limit in water, a^ = 1.  For organic
compounds that are sparingly soluble, and (in pure form) are liquid or solid at atmospheric
pressure and ambient temperature:
—  = (55.55 moles/1) -^i
v                  s.
                                                                               (44)
where u; is the molecular weight of compound i, and Sj is the aqueous solubility of compound i.

       For gases, the corresponding equation is:
                     -UJ
= (55.55 moles/1)
                                               (45)
where o>w is the molecular weight [g/mole], and S; is the solubility [M/L3] of compound i in
water. Molecular weight and solubility values for numerous organic compounds are provided in
Appendix A.
                                         59

-------
                                       SECTION 4
                            MODEL DATA REQUIREMENTS

       Media, fluid, and contaminant properties control air flow and vapor-phase transport in the
vadose zone in accord with the physical and chemical processes described in Section 3 and
subject to boundary conditions imposed on the system. Parameter input needed to simulate SVE
depends on the processes, equations, and system features selected for analysis, and model-
specific data requirements. The following discussions address air flow and vapor-phase transport
model parameters (Sections 4.1 and 4.2), model boundary conditions (Section 4.3), model grid
design (Section 4.4), and data limitations (Section 4.5).

       SVE modeling can be conducted at various levels of complexity with increasing data
input requirements as described in Section 2.1.2.  Model input parameters and parameter
estimation methods are listed in Table 5 and discussed below. Also identified in Table 5 are the
data requirements of different types of SVE models. Additional information on model
parameters is provided by Mercer et al. (1982), USEPA (199la), and model user documents.

4.1    AIR FLOW MODEL PARAMETERS                                           /

4.1.1  Porosity

       Porosity (0) is a measure of the volumetric fraction of a porous medium that is occupied
by void space:                                         .
         _        Volume of Voids         _   _  Pb
           Bulk Volume of Porous Medium         p
(46)
where pb is the soil bulk density [M/L3] and pp is the soil particle density [M/L3]. Total porosity
is composed of primary and secondary porosity. Primary porosity is controlled by the shape,
sorting, and packing of grains and is independent of grain size. Secondary porosity refers to
openings (such as joints, fractures, rootholes, and animal burrows) that develop after a medium
has been created by sedimentary or igneous geologic processes. Representative porosity values
for different geologic materials are presented in Table 6 (also see Tables 7 and 8 in subsequent
sections) . Total soil porosity is commonly calculated from bulk density measurements and
particle density estimates using Equation (46).

       Effective porosity (0e) is the actual amount of interconnected pore space accessible to
fluid flow.  The effective porosity available to convey air, water, or NAPL in the vadose zone is
less than the total porosity due to the presence of dead-end pores and the other fluids. For
example, during SVE, the effective porosity for air flow will be reduced by the presence of
water, and possibly NAPL, in pore spaces. Even a dry soil will contain dead-end pores and
residual moisture bound to particles by surface tension. Thus, depending ori moisture conditions,
effective porosity for a given fluid may be much less than, or nearly equal to, total porosity.

                                          60

-------
Table 5.       Listing of SVE model parameters.
                               (*F=air flow, model, T=vapor transport model, M=multifluid flow and transport model).
Parameter
Porosity; volumetric fluid
content (saturation)
Capillary pressure v.
saturation
Permeability
Relative permeability v,
saturation
Gas temperature
Gas molecular weight
Gas density
Gas viscosity
Water table elevation
Spatial distribution of
formation properties
Mechanical dispersion
coefficient
Molecular diffusion
coefficient
Vapor pressure
Henry's law constant
Aqueous solubility
Sorption parameters
(K
-------
Table 6.
Selected physical properties of rocks and soil (after USEPA, 1990a; Freeze and
Cherry, 1979; Krinshnayya et al., 1988; Morris and Johnson, 1967). Additional
soil porosity data (and dry bulk density values using Equation 97) are provided in
Tables 7 and 8 (0 = 6W in Table 7).
Rock/Soil Type
Porosity
(%)
Particle
Density
(g/cm3)
Dry Bulk Density
(g/cm3)
Hydraulic
Conductivity
(cm/sec)
Unconsolidated
Gravel
Sand
Loam
Silt
Clay
25-40
25-50
42-50
35-50
30-60
Consolidated
Sandstone
Shale
Granite
Granite (fractured)
Limestone
Limestone (Karstic)
Basalt (permeable)
5-30
0-10
0-5
0-10
0-20
5-50
5-50
2.65
2.65
2.65
2.65
2.65

2.65
2.25
2.70
2.70
2.87
2.71
' 2.96
1.59-1.99
1.33-1.99
1.33-1.54
1.33-1.72
0.68-1.35
10-1-102
10'4-10
10'5-10-1
10-7-10'3
io-10-io-7
Intrinsic
Permeability
(cm2)

10-6-10'1
10-9-10'5
io-1°-io-6
10-12-10'8
10-15-10'12

1.86-2.52
1.98-2.25
2.57-2.70
2.43-2.70
2.30-2.87
1.36-2.57
1.48-2.81
lO-MO-4
10'11-10-7
io-11-io-8
1 0-6-1 Q-2
10-7-10'4
10'4-1
10'5-1
10-13-10'9
io-16-io-12
io-16-io-13
io-11-io-7
10-12-10'9
10-9-10'5
lo-10-^-5
                                          62

-------
       Effective air-filled, water-filled, NAPL-filled, and total porosity values may be used
during a modeling analysis to calculate fluid velocities, flushing rates, fluid storage terms (e.g.,
see Equation 12 for pneumatic specific storage), velocity-dependent mechanical dispersion,
contaminant mass distributions, media bulk density, retardation factors, and relative
permeabilities.  Portions of void space occupied by air, water, and NAPL are also described in
terms of volumetric content and saturation.

4.1.2  Volumetric Fluid Content
       The volumetric fractions of a porous medium occupied by air, water, and NAPL are
defined by the volumetric air content (0g), water content (8W), and NAPL content (0N),
respectively:
                           Volume of Air
             0  = 	
              8   Bulk  Volume of Porous Medium
                                                                   (47)
             0...  =
            0.  =
                          Volume of Water
Bulk Vohiihe of Porbus Medium

       Volume of NAPL
Bulk Volume of Porous Medium
 (48)


(49).
The volumetric content of a particular fluid in the vadose zone can vary from a residual content
(e.g., the irreducible water content) to near the total porosity value. Fluid content in the upper
few feet of soil varies much more than at depth, where it may be relatively stable in uniform soil
(Charbeneau and Daniel, 1993).  Volumetric water and air contents can be determined using a
variety of direct and indirect methods which are described by Rawls et al. (1993) and Gardner
(1986). Most frequently, volumetric, water content is determined in the laboratory by gravimetric
analysis of soil samples and in the field by neutron attenuation techniques. Methods for
determining NAPL content in soil are reviewed by Cohen and Mercer (1993).

4.1.3  Fluid Saturation

       The saturation of a fluid is the volumetric fraction of the total void volume occupied by
that fluid. Thus, for example, water saturation is given by:
                      _  Volume of Water
                   s  —  	
                    w    Volume of Voids
                                                                  (50).
Values of air saturation (sg) and NAPL saturation (SN) are calculated similarly. Fluid saturation
in the vadose zone ranges from residual saturation, which varies significantly depending on fluid
and media properties (Cohen and Mercer, 1993), to nearly 1.0.  Saturation and volumetric fluid
content are particularly important to S VE because effective air permeability is a function of air-
filled porosity.

                                            63

-------
4.1.4  Capillary Pressure Relationships

       In the vadose zone, water is subject to negative pressure (also termed suction, tension, or
capillary pressure) arising from capillary and adsorptive forces exerted by the soil matrix.
Laboratory experiments demonstrate that capillary pressure head O) can be represented as a
function of water saturation (sw) or volumetric water content (0W). As shown in Figure 17, this
relationship is described by nonlinear soil moisture retention O-0W) curves which are typically L-
shaped for sandy soils and slope more uniformly with finer-grained soils.

       Actual iK0w) relations are more complicated than revealed by the monotonic curves of
Figure 17 (Lenhard, 1992; Parker, 1989). Changes in capillary pressure with water saturation
depend on whether the medium is undergoing drainage (drying) or infiltration (wetting).  This
capillary hysteresis results from air entrapment and differences in fluid contact angles during
drainage and infiltration. These phenomena cause different i|;(0w) curves to be followed
depending on the prior saturation history. During drainage, water drains quickly from the larger
pores while the smaller pores drain slowly, if at all. Thus, capillary pressure corresponds to
higher water content on the drying curve. During wetting, the smaller pores fill first and the
larger pores are least likely to fill with water, thereby leading to a lower i|;(0w) curve.  Although
errors can occur by neglecting hysteresis for certain multiphase flow problems (Lenhard, 1992),
it is usually ignored due to the larger uncertainties associated with other parameter estimates
(Parker, 1989).

       Modeling conducted to evaluate S VE rarely includes simulation of immiscible water or
NAPL flow.  Several complex multiphase flow and transport codes, however, can be used to
simulate the simultaneous flow of gas, water, and NAPL. Such models require descriptions of
i}r(0w) relations for the different fluid phases for each modeled soil or rock type.  Capillary
pressure head curves are input to simulators as tabulated values or by specifying functional
relationship parameters.

       Laboratory ty(0w) measurements are often fitted by nonlinear regression to an empirical
parametric model such as the Brooks-Corey (1964) and van Genuchten (1980) power function
relationships (see also  Luckner et al., 1989; Parker, 1989; Hillel, 1980).  For example, the van
Genuchten relationship is
                0-0
                 w     wr
                 0-0
                                          M
(51)
where 6W is the residual water content [L3/L3], i|r is the capillary pressure head [L], a is an
empirical constant [L'1], N is an empirical constant [dimensionless], and M = 1 - 1/N. The i|/(6w)
curves for different USDA soil types (Figure 18) that are shown in Figure 17 were determined
using Equation (51) and the soil water retention parameters in Table 7.  These van Genuchten
                                           64

-------
                   *
                  £  10* -
                      JO'
                         o.o
                                      Legend

                                Clay
                                Clay Loam
                                Loam
                                Loamy Sand
                                Silt Loam
                                Silty Clay Loam
                                Sand
                                Sandy Clay      	r
                                Sandy Clay Loam  y	-Y
                                Sandy Loam     4	*.

                                	•
    X----X

    B----H
    o	-«,
                                    0.1
    0.2        0.3

Volumetric Water Content
Figure 17.     Typical curves showing the relationship between capillary pressure and

               volumetric water content for various USDA soil classes (from USEPA, 1991c).
                                              65

-------
  o
Figure 18.    USDA soil textural triangle.
                                          66

-------
Table 7.      Mean values of the van Genuchten soil moisture retention and relative
              permeability parameters for the USDA soil types (modified from Carsel and
              Parrish, 1988). Descriptive statistics (e.g., standard deviation and coefficient of
              variation) for each parameter listed below are provided by Carsel and Parrish
              (1988).
Soil Texture
(USDA)
Clayey soil*
Clay loam
Loam
Loamy sand
Silt
Silt loam
Silty clay
Silty clay loam
Sand
Sandy clay
Sandy clay
loam
Sandy loam
Saturated
Hydraulic
Conductivity
(cm/hr)
0.20
0.26
1.04
14.59
0.25
0.45
0.02
0.07
29.70
0.12
1.31
4.42
Saturated
Water
Content
0.38
0.41
0.43
0.41
0.46
0.45
0.26
0.43
0.43
0.38
0.39
0.41
Residual
Water
Content
0.068
0.095
0.078
0.057
0.034
0.067
0.070
0.089
0.045
0.100
0.100
0.065
van Genuchten
Parameters
d
(1/cm)
0.008
0.019
0.036
0.124
0.016
0.020
0.005
0.010
0.145
0.027
0.059
0.075
N
1.09
1.31
1.56
2.28
1.37
1.41
1.09
1.23
2.68
1.23
1.48
1.89
M
0.083
0.237
0.359
0.561
0.270
0.291
0.083
0.187
0.627
0.187
0.324
0.471
Number
of
Samples
.**
400
364
735
315
82
1093
374
641
246
46
214
1183
Notes:  *Clay soil refers to agricultural soil with <60% clay.
      **Number of samples as indicated with minor exceptions; see Carsel and Parrish (1988).
                                            67

-------
parameters are based on statistical analysis of thousands of soil samples taken from 42 States
(Carsel and Parrish, 1988).
       Alternatively, the Brooks-Corey soil water retention relationship is given by:
e  -  e
 w     w
e  - e.
                               *
                                      fon|/;>T|;.
(52)
where tyb is the air-entry (bubbling) capillary pressure head [L], K is the pore-size distribution
index [dimensionless], and the other parameters are defined in Equation (51). Water retention
and Brooks-Corey parameter data for eleven USDA soil types based on a compilation by Rawls
et al. (1985) are given in Table 8.

       Both the van Genuchten and Brooks-Corey parameters can be estimated by: (1)
calibrating a multiphase flow model to field data; (2) fitting >(0W) curves determined by
laboratory core or field tests; (3) calculation based on grain size data (Arya and Paris, 1981;
Mishra et al., 1989); and (4) using data provided in the literature for similar soils. Field and
laboratory methods used to determine soil moisture retention curves are described by Bruce and
Luxmoore (1986) and Klute (1986), respectively.

4.1.5   Permeability and Relative Permeability

       Permeability is a measure of the resistance to fluid flow in a porous medium; the greater
the permeability, the lower the resistance. It is uncertain, variable with space and time (due to
changing fluid saturation), and critical to SVE performance (see Section 2.2.1). Due to
anisotropy, horizontal permeability (kj is usually greater than vertical permeability (kj.
Permeability has dimensions of L2, and values are often given in units of darcies where:

  1 darcy = 9.87 x 10'9 cm2 = 1.062 x lO'11 ft2 ~ 9.66 x 10'4  cm/s (water at normal temp.)  (53).

       Methods for estimating air permeability, in order of generally decreasing reliability,
include:  (1) analysis of SVE operation data (e.g., measured vacuum distribution and flow rates);
(2) analysis of pneumatic tests; (3) calculation based on saturated permeability measurements;
(4) laboratory testing of core samples; (5) calculation based on grain-size analysis; and (6)
comparison to literature values. SVE pilot test and operation  data, and pneumatic test data, can
be analyzed by model calibration and/or using appropriate analytical solutions (e.g., Beckett and
Huntley, 1994; Shan et al., 1992; Baehr and Hult, 1991; Falta, 1995; Nyer et al., 1994; Johnson
et al., 1990a,b;  Massman,  1989; USEPA, 199la; USEPA, 1993). Equation (54) given below
allows estimation of air permeability from saturated permeability measurements. A new
laboratory apparatus for measuring air permeability in cohesive and cohesionless soils, and
determining relative permeability versus saturation curves, is described by Rodeck et al. (1994)
and Stylianou and DeVantier (1995). Massman (1989) presents equations to estimate air

                                            68

-------
 Table 8.
Brooks-Corey soil moisture retention and relative permeability parameters for the
USDA soil types (modified from Rawls et al, 1985).
Soil Texture
(USDA)
Sand
Loamy sand
Sandy loam
Loam
Silt loam
Sandy clay loam
Clay loam
Silty clay loam
Sandy clay
Silty clay
Clay
Total
Porosity
0.437*
(0.37-0.50)
0.437
(0.37-0.51)
0.453
(0.35-0.56)
0.463
(0.37-0.55)
0.501
(0.42-0.58)
0.398
(0.33-0.46)
0.464
(0.41-0.52)
0.471
(0.42-0.52)
0.430
(0.37-0.49)
0.479
(0.42-0.53)
0.475
(0.43-0.52)
Residual
Water Content
0.020
(0.001-0.039)
0.035
(0.003-0.067)
0.041
(-0.024-0.106)
0.027
(-0.020-0.074)
0.015
(-0.028-0.058)
0.068
(-0.001-0.137)
0.075
(-0.024-0.174)
0.040
(-0.038-0.118)
0.109
(0.013-0.205)
0.056
(-0.024-0.136)
0.090
(-0.015-0.195)
Effective
Porosity
0.417
(0.35-0.48)
0.401
(0.33-0.47)
0.412
(0.28-0.54)
0.434
(0.33-0.54)
0.486
(0.39-0.58)
0.330
(0.23-0.43)
0.390
(0.28-0.50)
0.432
(0.35-0.52)
0.321
(0.21-0.44)
0.423
(0.33-0.51)
0.385
(0.27-0.50)
Brooks-Corey Parameters
Air-Entry
(Bubbling)
Pressure
(cm)
7.26
(1.4-39)
8.69
(1.8-42)
14.66
(3.4-62)
11.15
(1.6-76)
20.76
(3.6-120)
28.08
(5.6-142)
25.89
(5.8-116)
32.56
(6.7-159)
29.17
(5.0-172)
34.19
(7.0-166)
37.30
(7.4-187)
Pore Size
Distribution
A
0.694
(0.30-1.09)
0.553
(0.23-0.87)
0.378
(0.14-0.62)
0.252 .
(0.09-0.42)
0.234
(0.10-0.36)
0,319
(0.08-0.56)
0.242
(0.07-0.41)
0.177
(0.04-0.32)
0.223
(0.05-0.40)
0.150
(0.04-0.26)
0.165
(0 04-0 29)
#of
Sam-
ples
762
338
666
383
1206
498
366
689
45
127
291
Note: *First line is mean value; second line is ± one standard deviation about the mean.
                                           69

-------
permeability for sands and gravels based on grain size data.  Literature values of permeability
associated with different media are given in Table 6 and plotted as a function of USD A soil type
in Figure 19.

       In general, field pneumatic tests and SVE-pilot tests  are preferred for evaluating air
permeability because these methods stress a relatively large  portion of the subsurface and
account for site-specific conditions (e.g., moisture content, stratification, boundary conditions,
etc.). Such tests, however, are typically only used to determine air permeability at one fluid
saturation condition. As discussed below, air permeability rises and falls significantly with air-
filled porosity. Thus, it is important to consider, and incorporate as necessary, the effects of
relative permeability when modeling SVE.  Functional relative permeability relationships
described below provide a means for adjusting effective air permeability to account for saturation
effects.

       Several related terms are used to quantify the ease of fluid flow through a porous
medium.

       Intrinsic Permeability (k) is a property  of the medium alone that varies with the size and
shape of connected pore openings. Intrinsic permeability can be estimated based on the saturated
hydraulic conductivity:

                                                                                      (54)
where k is intrinsic permeability [L2], K is saturated hydraulic conductivity [L/T], |j,w is the
dynamic viscosity of water [M/L-T], pw is the density of water [M/L3], and g is acceleration due
to gravity [L/T2].

       Hydraulic Conductivity (K) is the volume of water at the existing kinematic viscosity that
will flow in a unit time under a unit hydraulic gradient through a unit area of porous medium
measured at right angles to the direction of flow. It assumes that 6W = 6 and is defined by
rearranging Equation (54).

       Air conductivity (Kg), analogous to the hydraulic conductivity, is calculated using the
density and dynamic viscosity of soil gas:
K8 =
                                                                                   (55)
 where pg is the density of soil gas (Section 4.1.8) [M/L3], and ug is the dynamic viscosity of soil
 gas (Section 4.1.9) [M/L-T].
                                             70

-------
                    10
           10-
                   Sand (%)
20     30      40     50     60
        I   V	1   i   I   i	1-
          100-
Figure 19.     Saturated hydraulic conductivity (K) in cm/hr as a function of grain size
              distribution. Intrinsic permeability (k) in darcies can be estimated by:  k = 0.29 K
              where K is in cm/hr. [Reprinted from Rawls and Brakensiek, 1985; by permission
              of American Society of Civil Engineers, New York, NY.]

                                            71

-------
       The definitions of permeability and conductivity presented above assume constant and
complete saturation of pore space by air or water.  Both water and air (and sometimes NAPL),
however, are present in the vadose zone, and the saturation of each fluid changes with time. For
example, SVE will tend to decrease the soil moisture content if the incoming vapor is drier than
the pore vapor, but it may also induce water table upwelling, thereby increasing the water content
at the base of the vadose zone.

       In an air/water, or air/water/NAPL system, the fluids compete for pore space and each
fluid develops its own flow channels, thus reducing the cross-sectional area available for the flow
of any one fluid. Effective permeability is the permeability of the porous medium to a fluid when
more than one fluid is present; it is a function of saturation.  Relative permeability is defined as
effective permeability divided by intrinsic permeability, and can vary from zero to one. The
relative permeability of a given fluid increases with its saturation.  Thus,  in the vadose zone, the
permeability to air is greatest when soil moisture declines to residual saturation (Figure 3).

       Relative air permeability (krg) is the ratio of the effective air permeability at a fixed
saturation to the intrinsic permeability. Relative permeabilities at different fluid saturations can
be measured in a laboratory via column studies. The relationship of relative air permeability to
relative water permeability as a function of saturation is illustrated in Figure 3. General features
of relative permeability curves indicate that: (1) the relative permeabilities rarely sum to the
intrinsic permeability; (2) the relative permeability of air (the nonwetting fluid) generally
exceeds that of water  at the same saturation for each phase; and,  (3) the relative permeability of
each fluid goes to zero at residual saturation. The relative permeability of air exceeds that of
water or NAPL at equivalent saturations because it tends to  occupy the largest interconnected
pore spaces.  This occurs because, as depicted in Figure 20,  solid mineral surfaces, with few
exceptions, are preferentially wet by water, and then NAPL, compared to air.

       As with capillary pressure, due to the difficulty of direct measurement, several functional
relationships have been developed that relate relative permeability to fluid saturation. Parker et
al. (1987) extended the relative permeability model developed by van Genuchten (1980) to allow
estimation of relative permeabilities for air, water, and NAPL in a two- or three-fluid system:
                                                                                      (56)
                                         )M
(57)
                                            72

-------
                                                     NAPL
                                     WATER
                                                                        (a)
                               NAPL;
                                       WATER
                                                                        (b)

Figure 20.    In a two fluid phase system with NAPL and water, the water is typically the
             wetting fluid and occupies the smallest pores (a); in a three fluid phase system
             with NAPL, water, and air, the order of solid wetting is typically water, NAPL,
             and then air (b) (after Parker, 1989).
                                          73

-------
                                                                                    (58)
Stc -
                 Sw  + SN
                                                                                    (59)
   Swc -
                                                                                     (60)
where k^ is relative air permeability, k^ is relative water permeability, krN is relative NAPL
permeability, sm is the residual water saturation, swe is the effective water saturation, ste is the
effective total fluid saturation, the other variables are defined above and in the terms listing, and
all variables are dimensionless.  Air-water relative permeability (kr-sw) curves for five different
USDA soil types (Figure 18) determined using Equations (56) and (57) and van Genuchten
parameters given in Table 7 are shown in Figure 21.
       The Brooks-Corey relative permeability equations for an air-water system are:
                                                                                     (61)
                                   2+3JI
                              (O  x
                                                                          (62).
Air-water kj(sw) curves calculated using Equations (61) and (62) for five soil types are also
shown in Figure 21.  The residual and saturated water content values given in Table 7 that were
used to calculate the van Genuchten kr(sw) curves were also applied to the Brooks-Corey
calculations. Pore-size distribution indices were taken from Table 8. Both the van Genuchten
and Brooks-Corey kr(sw) parameters can be determined experimentally or estimated using data
provided in the literature for similar soils. A method for converting between the Brooks-Corey A,
parameter and the van Genuchten M parameter is presented by Lenhard et al. (1989).

       Simpler relative permeability equations are incorporated in some multiphase flow
models.  In a two-fluid system, relative permeabilities can usually be approximated in terms of
                                            74

-------
                                                  van Genuchten Model
                                                   (Parker et al., 1987)
                        Sandy Loam

                        Sandy Clay

                        Silt Loam

                        Silty Clay Loam
                                              Brooks-Corey Model
                      Sandy Loam


                      Sandy Clay


                      Silt Loam ,
                      Silty Clay Loam
                                          0.4         0.6
                                       Water Saturation
Figure 21.     Air-water relative permeability (kr-sw) curves calculated for five soil types
              determined using (a) Equations (56) and (57) (Parker et al., 1987) and the van
              Genuchten parameters given in Table 7; and (b) the Brooks-Corey Equations (61)
              and (62) and pore-size distribution indices given in Table 8 (b).
                                           75

-------
fluid saturation to a power (n) between 2 and 4 (Faust et al., 1989). For example, relative air and
water permeabilities in variably-saturated unconsolidated sands, with negligible NAPL
saturation, can be represented by the following relations (Frick, 1962; Bear and Verruijt, 1987):
                                                                                  (63)
                           1 -  s
                       _   -w
                            1 - s
                                                                                  (64).
       Relative permeability curves for air and water in sand calculated using several relative
permeability relationships including Equations (63) and (64) are plotted in Figure 22a. Actual
measurements of relative air permeability versus saturation for a sand (passing 60 mesh, but
retained on 100 mesh screen) containing different mixtures of gasoline and water are shown for
comparison in Figure 22b. Based on the data presented in Figure 22b, Stylianou and DeVantier
(1995) determined the following best-fit relationship between krg and sg for sand with sg greater
than 0.45:
            = 1.105  - 2.576(1 ~sg) +  1.577(1-sg)2
(65)
which is also plotted on Figure 22a.

       Three-phase relative permeabilities are needed to simulate the simultaneous, interrelated
movement of air, water, and NAPL in the subsurface. Due in part to cost considerations,
air/water/NAPL relative permeabilities are rarely determined at contamination sites.
Furthermore, it is uncertain whether relative permeability curves determined in a laboratory will
adequately represent field-scale conditions. A number of theoretical relationships in addition to
those proposed by Parker et al. (1987) have been developed to estimate three-phase relative
permeabilities (Stone, 1970; Stone, 1973; Dietrich and Bonder,* 1976; Fayes and Matthews,
1984; Delshad and Pope, 1989). Three-phase relative permeability relations, however, are only
incorporated into complex simulators (such as MAGNAS or MOTRANS, see Section 5) that are
rarely used in SVE modeling.

       Simplifying assumptions are usually invoked to reduce computational requirements
associated with modeling SVE in the presence of air, water, and NAPL.  Frequently, it is
assumed that air is the only fluid in motion and that fluid saturations remain constant.  Using this
approach, intrinsic permeability values input to a model can be reduced to account for the
presence of water and NAPL.

                                            76

-------
      JO

      & 0.2
           0 H_
                      0.2       0.4      0.6      0.8
                            Water Saturation
                                                                 Parker etal. (1987)
                                                                 Brooks-Corey
                                                                 Effective Saturation Cubed
                                                                 Saturation Cubed
     Equation (66)
  Sand:
  Total porosity = 0.43
  Residual water content = 0.045
  Pore-size distribution index = 0.694
* van Genuchten parameter M = 0.627

              (a)

i
i
1
i
a.
I
£


0.80
0.60

0.40

0.20
Onn
D • water
* D D gasoine
^•« + 20% water&gas
* ^ *
+^ * 30%water&gas
«»n
*+•
* * 1?4f» * *
* • .








0.00 0.20 0.40 0.60 0.80 1.00
                                 Degree of Liquid Saturation, fraction
                                                                           (b)
Figure 22.    Air-water relative permeability curves calculated using several relative
              permeability relationships (a); and measurements of relative air permeability
              versus saturation for a sand containing different mixtures of gasoline and water
              (b) (after Stylianou and DeVantier, 1995).  [(b) is reprinted by permission of
              American Society of Civil Engineers, New York, NY.]
                                             77

-------
       Effective air permeability (kg) is the permeability of the medium to air that accounts for
the presence of water and NAPL under prevailing or representative conditions of fluid saturation.
Effective air permeability is the product of intrinsic and relative permeability:
                                                                                   (66).
Effective air permeability or effective air conductivity, an analogous parameter which
incorporates the density and viscosity of soil gas in accord with Equation (55), is the parameter
actually measured in the field during a pneumatic test. A more detailed discussion of the concept
of effective permeability is presented in Bear and Verruijt (1987).

       As noted above, complex multiphase flow models incorporate relative permeability
effects due to variable and changing fluid saturation. Many air flow models used to simulate
SVE, however, assume constant air-filled porosity and permeability with time. When using such
a model, effective air permeability estimates derived from field pneumatic tests or relative
permeability equations should be entered instead of intrinsic permeability data. Additionally,
consideration should be given to the effects of changing fluid saturations (e.g., due to SVE
operation or infiltration rate changes) when determining an effective air permeability to represent
a particular simulation period. Of the readily available SVE codes examined herein, only
VENT2D updates the effective air permeability based on changing air-filled porosity during
simulation (using k^ = sg3).

4.1.6  Gas Temperature

       Several properties (e.g.,  air viscosity, density, and vapor pressure) relevant to vapor flow
and transport vary with temperature (Figure 23), and several models used to simulate SVE
require vapor temperature data.  The temperature  of soil gas varies with location and depth.
Seasonal shallow soil temperatures can be estimated by (Toy et al., 1978):
             Summer:      Y  = 16.115  + 0.856X
             Fall:          Y = 1.578 +  1.023X
             Winter:       Y  = 15.322  + 0.656X
             Spring:       Y  = 0.179 + 1.052X
(67)
where Y is the mean monthly shallow soil temperature (°F) and X is the meari monthly air
temperature (°F). At depth, the temperature of soil gas is relatively stable throughout the year
(Figure 24) and can be approximated by the average temperature of shallow groundwater (Figure
25) or the mean annual air temperature.
                                            78

-------
    -
   OQ



   1

   to
   OJ
                           Dynamic Viscosity
                                                                        Density
    .
«' sr
§  I
S2. ^
•t' g
 \i  hj

   CD
   f
     p

     en
     o
     o
         ki    b     N>     en     -*i     o
         en    o     en     o     en     o
         o    o     o     o     o     o
            CD
            CD


            3
            r-f
"cx
CD
(Q

cB
CD
OT

0

              o ~ ~
o - -


-------
                    0  -i
                                    Mean Annual Temperature
                    2  -
                o
                CD
                ~a  A
                c  4

                I
                CD
                    6  -
                
-------
                                                                                                              37°
oo.
                                                                                    Average Temperature
                                                                                         of Shallow
                                                                                       Ground Water
              Temperature in
                Degrees F.
                                                                                                   77°
                    Figure 25.     Average temperature of shallow groundwater in the United States (after USGS
                                 Water-Supply Paper 520F).

-------
4.1.7  Gas Molecular Weight

       Several models used to evaluate SVE require and/or compute gas molecular weight.  The
molecular weight (tog) of a pure gas is calculated from the weight of its molecular components.
For example, the molecular weight of methane, CH4 is:
  wCH4=oJc+4(DH=12.011g/mol+(4xl.0079g/mol)=16.0426 g/mol CH4
                                                          (68).
Molecular weights of many organic compounds are given in Appendix A.

       The molecular weight of a gas mixture is equal to the average of the individual gas •..
components weighted by mole fraction:
CO    =
 gmix
                                                                                   (69)
where togmix is the molecular weight of a gas mixture [M/mol], n is the number of components in
the gas mixture [dimensionless], z; is the mole fraction of component i in the gas
[dimensionless], and co; is molecular weight of component i [M/mol].

       Thus, the molecular weight of dry air at 1 atm is calculated as follows:
Major Components of
Air
Oxygen
Nitrogen
Argon
Carbon Dioxide
Total
Chemical
Formula
02
N2
Ar
CO2

Component Mole
Fraction (z, )
0.78102
0.20946
0.00916
0.00033
0.99997
Molecular Weight
(MI)
28.0134
31.9988
39.948
44.0098
Mole Fraction
Weight fc oj,)
21.87903
6.70247
0.36592
0.01452
wair = 28. 962 g/mol
The coair of moist air depends on humidity and is slightly less than dry air (-28.6 g/mol when
saturated with water at 25° C and 1 atm).  As noted in following section, soil gas mixtures
containing high concentrations of iiigh molecular weight compounds (e.g., chlorinated solvents
with high vapor pressures) can cause density-driven gas flow in highly permeable media.
                                           82

-------
4.1.8  Gas Density

       Soil gas density (pg) varies with temperature, vapor contaminant concentrations
(molecular weight), and pressure in close accord with the ideal gas law:
                p  = —s—s
                KB     RT
                                                                                  (70)
where o>g is the gas molecular weight [M/mole], Pg is the gas pressure [M/L-T2], R is the
universal gas constant [ML2/molKT2], and T is the absolute temperature [K]. Using Equation
(70), the density of moist air at 1 atm and 25° C can be calculated as:
P*  =
              RT
                         28.6g/mol (latm)
 0.08206
                      L-atm
                      K-mol
                                           = 1.17g/L
                                    (298K)
                                                             (71).
As shown in Figure 23, the density of moist, pure air at 1 atm pressure declines from 1.29 g/L at
0° C to 1.13 g/L at 40° C.

       Gas densities encountered during SVE generally are within a range of 0.66 g/L for 100%
methane to 1.3 g/L for air with a high concentration of volatile organic vapors (Massman, 1989).
NAPL volatilization in the vadose zone causes variations in gas density which depend on the
NAPL molecular weight and vapor pressure, gas temperature, and gas mixing (e.g., due to
advection, diffusion, and a heterogenous contaminant distribution). The saturated vapor
concentration of a single component NAPL can be calculated by (Falta et al., 1989):
                              R T
                                                                                  (72)
where C°; is the saturated vapor concentration of the compound i [M/L3], and P;v is the saturated
vapor pressure of compound i [M/L-T2].  For example, the saturated vapor concentration of
trichloroethene in soil gas adjacent to liquid TCE is:
     •VTCB
        RT
_  131.4g/mol (0.0977atm)  _

   0.08206 L~atm  (298K)
          K-mol
                                                ).52g/L
                                                                                  (73).
                                           83

-------
       Assuming the validity of Dalton's law of partial pressures and equilibrium with NAPL,
total gas densities can then be calculated based on the saturated vapor concentration (Falta et al.,
1989):
                 P« =
                              R T
                                                                             (74).
The total density of soil gas (p°) saturated with trichloroethene at 1 atm and 25° C, therefore, is
given by:
  _
8
                 B            RT

         0.0977atm(13 1 .4 -28.6g/mol) + latm(28.6g/mol)
                          T —a
              0.08206
                                                                             (75)
                          K-mol
                                 (298K)
Total gas densities calculated using Equation (74) for some common NAPLs are presented in
Table 9.

       Several recent modeling studies have examined the potential for soil gas flow to be
induced by gas density gradients that result from NAPL volatilization (Sleep and Sykes, 1989;
Falta et al., 1989; Mendoza and Frind, 1990). Density-driven gas flow may be significant where
contaminated gas density exceeds ambient gas density by >15% in coarse sand and gravel (e.g.,
where effective air permeability is at least IxlO'11 m2) (Falta et al., 1989; Mendoza and Frind,
1990). Dense gas emanating from NAPL in the vadose zone will generally sink to the water
table and then spread outward. Gas migration patterns will be strongly influenced by
heterogeneities.  Gas density gradients dissipate with declining contaminant concentrations and
distance from the contaminant source.

       Density-driven gas flow is rarely incorporated in simulations conducted to assess SVE.
Rather, it is usually assumed that variations in gas density due to temperature and chemical
composition are sufficiently small that gas density can be simulated as uniform in space and
constant with time. Gas density values are input to some models and calculated by others (e.g.,
based on gas molar weight input).

4.1.9  Gas Viscosity

       Smaller pressure gradients are needed to induce significant flow of gas compared to water
because the viscosity of gas is about two orders of magnitude less than that of water. Dynamic
viscosity (u), also known as absolute viscosity, is a measure of the internal friction within a fluid
that causes it to resist flow. Gas viscosity results from the transfer of momentum as gas

                                           84

-------
Table 9.
Calculated vapor pressure, saturated vapor concentration, and total gas density
values for some common volatile organic contaminants (from Falta et al., 1989).
Chemical
Trichloroethene
Toluene
Benzene
Chloroform
Tetrachloroethene
1,1,1 -trichloroethane
Ethylbenzene
Xylene
Methylene chloride
1,2-dichlorothene
1 ,2-dichloroethane
Chlorobenzene
1,1-dichloroethane
Carbon tetrachloride
Air at 1 atm, 25° C.
Molecular
Weight
(g/mole)
131.4
92.1
78.1
119.4
165.8
133.4
106.2
106.2
84.9
96.9
99.0
112.6
99.0
153.8
28.6
Vapor Pressure
(kPa @ 25° C)
9.9
3.8
12.7
25.6
2.5
16.5
1.3
1.2
58.4
43.5
10.9
1.6
20.1
15.1 .-
(101.3)
Saturated Vapor
Concentration
(9/L)
0.52
0.14
0.40
1.23
0.17
0.89
0.06
0.05
2.00
1.70
0.44
0.07
1.20
0.94

Total Gas
Density
(9/L)
1.58
1.27.
1.42
2,11
1.31
1.87
1.22
1.21
2.50
2.37
1.48
1.23
2.03
1.93
1.17
                                            85

-------
molecules collide. Liquid viscosity, however, is apparently derived from molecular cohesion
that opposes shearing stresses (Lyman et al., 1982). All other factors being equal, the rate of
fluid movement is inversely proportional to dynamic viscosity (e.g., Equation 55).  Kinematic
viscosity is defined as the dynamic viscosity of a fluid divided by its density.  Where required for
SVE model input, gas viscosity values are usually estimated based on literature values or
equations that relate viscosity to temperature and/or chemical  composition.

       Air and water viscosities vary with temperature as shown in Figure 23. The viscosity (ng)
of an ideal gas is independent of pressure, but varies with temperature as given by (Moore,
1972):
                                                                                   (76)
where ugR is the dynamic gas viscosity [M/L-T] at a reference temperature, TR [K].  Baehr et al.
(1989) specify u^ = 0.000176 g/cm-s at TR = 283.15 K in the AIR3D model.
       The viscosity of a gas mixture depends on its chemical composition.  If concentration
data are available, the viscosity of most gas mixtures encountered during SVE can be estimated
by (Massmann, 1989):
^gmbc
                                                                                    (77)
where |agmix is the dynamic viscosity of the gas mixture [M/L-T], z{ is the mole fraction of
component i [dimensionless], and |j,gi is the dynamic viscosity of component i [M/L-T].

       Alternatively, the viscosity of a multicomponent gas mixture can be estimated at system
temperature by Wilke's semiempirical viscosity formula (Bird et al., 1960):
                       * cnux

                                                                                    (78)
in which
                             -V4
                                1+1
                                           0),
                                           CO.
                                                           (79)
                                           86

-------
where i and j refer to gas mixture components i and j.  Equation (78) has been shown to estimate
measured gas mixture viscosities with an average deviation of only two percent (Bird et al.,
1960).  Approximate viscosities for some gas components that may be encountered during SVE
are given in Table 10.

4.1.10  Spatial Distribution of Media Properties Affecting Air Flow

       Media heterogeneity and anisotropy can be incorporated to varying degrees within
models used to simulate SVE. The k^ anisotropy ratio is usually controlled by the depositional
environment in sedimentary materials; it ranges from near 1 for manmade fill to > 100 for highly
stratified deposits.  The k^ ratio can be estimated by conducting pneumatic tests (e.g., Shan et
al., 1992) or based on knowledge of site conditions. Rapid movement of soil gas through
fractures, root holes, animal burrows, and other heterogeneities in the subsurface is referred to as
preferential flow. Preferential flow paths are typically abundant near the surface, and so are
important when considering air flow through the vadose zone.  Heterogeneity and anisotropy can
cause air flow induced by SVE to bypass low permeability materials and thereby impair SVE
effectiveness as described in Section 3.1. The reliability of any model will be constrained by
inadequate site characterization. Sensitivity analysis can help assess the significance of limited
field data.
4.2    VAPOR CONTAMINANT TRANSPORT MODEL PARAMETERS

       Chemicals may exist in and migrate between four phases in the vadose zone: dissolved in
NAPL (if present), dissolved in water, volatilized in soil gas, and adsorbed to soil particles.  A
schematic representation of this multiphase system is presented in Figure 26. The total
concentration (Q) and mass (Mwi) of chemical i in a volume of soil (Vc) are given by:
and
                  =. v. [0NcNi+6wcwi+0hc.+Pbcadj
(81).
Chemical transfer (partitioning) occurs across the interfaces between each phase and is governed
by thermodynamic equilibria and kinetic behavior.  Schwarzenbach et al. (1993) provide an
excellent text on interphase partitioning of organic chemicals in environmental media.

       Most vapor transport models developed to date have assumed local equilibrium between
phases to simplify calculation of chemical mass distribution, movement, and interphase transfer.
The local equilibrium assumption presumes that (1) the rate of chemical movement through any
particular phase in the vadose zone is slow compared to the rate of mass transfer between phases

                                           87

-------
     Equilibrium Partitioning: NAPL Mixture and Water
    o
    W
    §
    o

    I
    i
         .,..., Aqueous Solubility
Raoult's Law:




= Xi  Cwi
        0     0.2     0.4     0.6     0.8      1

         Mole Fraction of Compound in NAPL Mixture
                          Equilibrium Partitioning: NAPL Mixture and Air
                                                        o
                                                        O
                                                        fc.
                                                        o


                                                        I
                              ., -Saturated Vapor Concentration	
Raoult's Law:

i  =«, V
                             0     0.2     0.4     0.6     0.8      1

                              Mole Fraction of Compound in NAPL Mixture
                               Equilibrium Partitioning Among Phases


                                                         Air  -+—-*- Water

                                                          tNX   t
                     NAPL—?
                           •  x^N   .
                        NAPL^^  Soil
         Equilibrium Partitioning: Air and Water
           Saturated Vapor Concentration	
                Dissolved Concentration
                             Equilibrium Partitioning: Adsorption to Soil
                                                        8
                                                        'o
                                                        in
                                                                               K.  =  K   f
                                                                                 a      oc   oc
                                                                    Dissolved Concentration
Figure 26.     Representation of equilibrium partitioning of chemical mass between phases.
                                                88

-------
Table 10.     Approximate viscosities for various gas components (from Weast, 1974).
Gas or Vapor
Air


Benzene, vapor
Carbon dioxide
Chloroform, vapor
Ethane
Ethylene
Isopentane, vapor
Methane

Nitrogen
Oxygen
n-Pentane, vapor
Temperature ° C
0
18
40
14
0
15
30
0
14
0
17
50
0
14
20
25
0
20
11
27
0
19
25
Dynamic Viscosity (g/cm-s)
0.000171
0.000183
0.000190
0.000074
0.000139
0.000146
0.000153
0.000094
0.000098
0.000085
0.000090
0.000100
0.000091
0.000095
0.000101
,0.000070 v
0,000102
0.000109
0.000156
0.000171
0.000189
0.000202
0.000068
                                          89

-------
 in contact locally; (2) thus, chemical concentrations in all phases remain in thermodynamic
 equilibrium; (3) mass transfer is reversible; and (4) equilibrium between any two phases is
 independent of the presence of the other phases (Charbeneau and Daniel, 1993). Linear
 partitioning, whereby partition coefficients do not vary with chemical concentration, is also
 commonly assumed in vapor transport modeling.  Hence, in most models used to simulate vapor-
 phase transport during S VE, the total contaminant mass is shared and moves among all phases
 based on linear,  reversible relations such as those depicted in Figure 26. A linear, reversible
 mass-balance approach to simulate vapor transport is described in Section 3.2.5 and implemented
 in the VENTING (ES&T, 1994) and VENT2D codes (Benson, 1994).

       Concentration tailing is commonly observed during SVE as discussed in Section 3.1.
 Several researchers have concluded that the assumption of equilibrium partitioning between
 phases is invalid due to physical and/or chemical limitations to mass transfer (e.g., Armstrong et
 al., 1994; Gierke et al., 1992;  Brusseau, 1991; McClellan and Gillham, 1990; Smith et al., 1990;
 Steinberg et al.,  1987; Buxton and Green, 1987; Valocchi, 1985; Parker and Valocchi, 1986;
 Karickhoff, 1980).  Besides the gas flow bypass phenomena described in Section 3.1, chemical
 desorption from soil solids to  water and chemical volatilization from water to air have been
 identified as potential rate-limiting mass transfer steps during SVE by Brusseau (1991) and
 Armstrong et al. (1994), respectively.

       Recently, relatively complex models have been developed to examine nonideal mass
 transfer associated with vapor-phase transport in the vadose zone (Armstrong et al.,  1994; Gierke
 et al., 1992; Brusseau, 1991).  Models incorporating chemical disequilibrium and/or nonlinear
 sorption isotherms are rarely used, however, in SVE studies.  The benefits of assuming
 equilibrium partitioning for SVE simulation are reduced computational and data input effort. Of
 course, the modeler must account for the fact that the linear equilibrium model tends to predict
 more effective mass removal that can be achieved in the field.

       Transport models require input of various parameters describing media and contaminant
 properties in addition to those needed for gas flow modeling.  Many of these parameters are
 discussed below, including mechanical dispersion and molecular diffusion coefficients, vapor
 pressure, Henry's law constant, aqueous solubility, solid-water distribution coefficient, organic
 carbon content, bulk density, and chemical composition.

 4.2.1  Mechanical Dispersion

       Dispersion causes spreading and dilution of volatile contaminants in soil gas from the
 combined effects of mechanical dispersion and molecular diffusion. For an isotropic medium,
the dispersion tensor (D;j) is given by (Bear, 1972) as:
                                                                                    (82)
                                           90

-------
where Dy is the gas dispersion tensor [L2/T], aT is the transverse dispersivity [L], OCL is the
longitudinal dispersivity [L], v; is the absolute interstitial velocity [L/T], vi; and Vy are the
interstitial velocity components in the i and j directions [L/T], cfge is the effective molecular
diffusion coefficient [L2/T], and 8^ is the Kronecker delta. Unlike dissolved contaminant
transport, it appears that molecular diffusion predominates over mechanical dispersion for vapor-
phase transport (Brusseau, 1991; Benson et al., 1994), except perhaps in close proximity to an
SVE well where air velocities are very high.

       Mechanical dispersion is caused by velocity variations that arise from differences in pore
size, flow path length (which are scale-dependent), and pore friction. Longitudinal dispersivities
determined for solute transport in groundwater range from 1 m to 10,000 m and are strongly
correlated with scale of measurement (Waldrop et al., 1985). Field-scale dispersivities^
determined by tracer tests and model calibration reflect unresolved velocity variations associated
with formation heterogeneities. Thus, mechanical dispersion is recognized as a transport model
calibration tuning parameter.  Ratios of aL:aT in groundwater systems typically range from 2:1 to
10:1, but are larger (e.g., 10:1 to 200:1) when using «T to simulate mechanical dispersion in the
vertical dimension. Dispersivity values from groundwater studies are reviewed by  Gelhar et al.
(1992), Waldrop et al. (1985), Anderson (1979), and Isherwood (1981).

       Relatively little research has addressed mechanical dispersion of soil gas; few (if any)
field-scale values are reported in the literature. This is due in part to the reduced importance of
mechanical dispersion in vapor-phase transport compared to molecular diffusion, and compared
to its role in groundwater transport. Walter (1994) and Benson et al. (1994) discuss the potential
significance of mechanical dispersion to SVE simulation. Values of «L used in published soil
gas modeling  studies range from 0.0 to 1.0 m (Mendoza and Frind, 1990).

4.2.2  Molecular Diffusion

       Molecular diffusion occurs due to random molecular motion. It is a much more
important transport process in soil gas compared to water due to the relative lack of molecule
crowding in gas compared to water.  Hence, diffusion coefficients are approximately four orders
of magnitude  greater in gas than in water (e.g., 5x10'2 cm2/s in free air versus 5x10'6 cm2/s in
water; see  Appendix A). Molecule size (and mass with few exceptions) is inversely related to
diffusion because it is more difficult for large molecules  to move through a crowd of molecules
and the thermal energy of large molecules is less than that of small molecules (Schwarzenbach et
al., 1993).  Free air diffusion coefficients are given for various volatile compounds in Appendix
A and can be  estimated by (Fuller  et al., 1966; Schwarzenbach et al., 1993):
                             1'75
                    = m-3
                rf.  = 10
(82)
                                            91

-------
 where cfgi is the free-air diffusion coefficient for compound i [cm2/T], T is absolute temperature
 [K], to; is the molecular weight of compound i [g/mol], coair is the molecular weight of air [28.97
 g/mol], Pg  is gas phase pressure [1 aim], Vair is the average molar volume of gases in air [-20.1
 cnrVmol], and V; is the average molar volume of compound i [cmVmol]. The average molar
 volume of a liquid chemical can be calculated by dividing the its molecular weight by density.

        The free air diffusion coefficient is calculated for trichloroethene in air at 20° C using
 Equation (83) as:
             _ IP"3 2931-75 [(1/29) +(1/131.4)1 *
             "    1  [20.1*+(131.4/1.464)* f   =
(84).
 This nearly matches the cfgTCE of 0.081 cm2/s reported by Mendoza and Frind (1990). Diffusivity
 estimates calculated using Equation (83) were found to be within 10% of measured values bv
 Fuller etal. (1966).

        The rate of mass transfer by gaseous diffusion follows a predictable pattern that is
 governed by the concentration gradient in accord with Equation (27). Although more detailed
 equations are available to account for other complexities of gas-phase diffusion (Youngquist,
 1970; Thorstenson and Pollock, 1989), using Pick's law to simulate diffusion in the vadose zone
 is generally appropriate and adequate (Brusseau, 1991). Effective gas diffusion coefficients (rfge)
 in soil are less than the free air diffusion coefficients (
-------
         0.014
       q
       cr
         0.012
          0.01 --
       CD
       O
       O
       c 0.008
       o
       'to
       u

       ^ 0.006  - -
       CO
          0.004 - -
       CD
       jg 0.002
                0
                      Effective Gas Diffusion Coefficients
                      Millington-Quirk Model
Xylene


Toluene


Trichloroethene


Benzene


Vinyl'Chloride
     0.1          0.2          0.3
           Volumetric Air Content
0.4
Figure 27.     Effective gas diffusion coefficient as a function of volumetric air content
             calculating using the Millington-Quirk equation.
                                        93

-------
4.2.3  Vapor Pressure

       Volatilization refers to chemical mass transfer from liquids and solids to gas. Thus,
chemicals in soil gas may be derived from NAPL, water, or soil constituents. Chemical
properties affecting volatilization include vapor pressure, chemical mole fraction, and solubility
hi water. Other factors influencing vapor transport, and hence volatilization rate, include:
concentration in the soil; soil moisture content (e.g., Smith et al. 1990); soil air movement;
sorptive and diffusion characteristics of the soil; temperature; and bulk properties of the soil such
as organic carbon content, porosity, density and clay content (Lyman et al., 1982).

       Equilibrium vapor concentrations for a chemical or mixture of chemicals in the vadose
zone can be calculated using (1) vapor pressure data where the compound is present as a separate
phase liquid (NAPL) or solid, or (2) Henry's law constant where the compound is present
dissolved in water but not as a separate phase. As shown in Figure 20, preferential wetting of
media solids by water compared to NAPL causes equilibrium vapor concentrations to be
governed, where NAPL exists, by air-NAPL partitioning (vapor pressure) rather than air-water
partitioning (Henry's law constant).

       The vapor pressure of a substance is defined as the partial pressure exerted by the vapor
of a pure solid or liquid chemical when it is under equilibrium conditions.  For example,
trichloroethene confined in a glass container will evaporate from the liquid-air interface, diffuse
hi air, and reach an equilibrium (saturated) concentration.  The trichloroethene vapor exerts a
pressure (its vapor pressure) on the container that can be measured experimentally as the height
to which it lifts a column of mercury. In the vadose zone, net evaporation of separate-phase
trichloroethene proceeds until the NAPL supply is exhausted (which may take many years
depending on NAPL volume and volatilization parameters noted above) because there is no jar
lid to completely confine soil gas.  SVE speeds the removal process by inducing air flow and
increasing concentration gradients.

       Transport models use vapor pressure data to calculate equilibrium gas concentrations and
the rate of contaminant volatilization. The effective vapor pressure of a compound in a
multicomponent NAPL is reduced from its pure-phase vapor pressure in proportion to its mole
fraction in the NAPL mixture as described by Raoult's law:


                        ^ =  *i PSV                                                (86)

where Pf  is the contribution to the total vapor pressure of i in the solution [M/L-T2], Xj is the
mole fraction of chemical i in the NAPL mixture [dimensionless], and Py is the  vapor pressure of
the pure compound i [M/L-T2].  Thus, a NAPL mixture, such as gasoline, generates vapors in
accord with Raoult's law. The concentration of mixture component i (Cgi) in soil gas adjacent to
the NAPL mixture can be calculated by:
                                           94

-------
C  =
                              x P
                               '  '
                              R T
(87).
       As an example, consider trichloroethene with a mole fraction of 0.34 in a NAPL mixture,
a pure phase vapor pressure of 0.075 atm at 20° C, and a molecular weight of 131.4 g/mol. Using
Equation (87), the equilibrium concentration of TCE in soil gas is:
   BTCE      -    (0.075atm)] / [(0.08206 L-atm/K-mol) (293 K)]
       = (0.00106 mol/L) (131.4 g/mol)
       = 0.139 g/L  = 139 mg/L
                                                           (88)
The concentration of a NAPL in vapor generally decreases with distance from the source due to
dilution effects.

       Vapor pressure is inversely related to the boiling point of a chemical and rises with
temperature.  As described near the end of Section 3.2.5, the vapor pressure of each component
of a NAPL mixture can be calculated using the Classius-Clapeyron equation (Equation 41) from
a value for vapor pressure at a reference temperature (e.g., 20°C or 293 K) and the boiling point
at 1 atm. Reviews of vapor pressure estimation methods are provided by Reid et al. (1977),
Mackay et al. (1982), and Burkhard et al. (1985). Vapor pressures for numerous compounds are
given at a reference temperature in Appendix A, and ranges of vapor pressure are shown for
several classes of organic compounds in Figure 28a.

       In general, chemicals with vapor pressures above 0.5 mm Hg at ambient temperature are
considered to have a high potential for removal by SVE (USEPA, 1991a). Given that increasing
temperature will increase vapor pressure, field research and modeling has been conducted to
evaluate the effectiveness of hot air and steam injection and soil heating to increase volatilization
and SVE removal rates (e.g., Hunt 1988a,b; HWC, 1992; Lord et al., 1987; Lingineni and Dhir,
1992). Although the range of compounds susceptible to SVE remediation can be extended by
raising soil temperature, costs may be high and few field trials have been conducted to date.

4.2.4   Henry's Law Constant

       In the absence of NAPL, contaminant volatilization in the vadose zone will be controlled
by air-water partitioning. The equilibrium air-water distribution ratio is approximated by Henry's
law constant which is given by:

                               p;
                         ^  = 7^                                                 (89)
                                           95

-------
10-12 1Q-IO 1Q-8 ,0-6 1Q-4 ,0-2 ,
haiogenated cct2'CCi2-,CH2ci2
Ci-and C2- compounds ^^^
alkylated (§P^^ ©
benzenes .m*m^m
Cl Cl
chlorinated ci^ci ©
Cl Cl Cl Cl ci
polyehlorinaled «-@ — ?<&-ci ®-®
biphenyls ci ci cici
phthalate * .— • °
esters ^
polycyclic ^©3? ,"v~,
aromatic *&& H38 C=.IV C'°HM
hydrocarbons ••"•^™
ii,
IQ-4 iO'2 1 IO2 IO4
(b) KH,, Henry's law constant (atm Lmol"'
1 IO2 IO4 IO6 IO8 10'°
haiogenated
Grand Ci CH2ci2 cci2=cci2
compounds ™"^™1
alkylated © @~~^
Cl Cl
chlorinated © afspw

phthalate 6 •»— 6
esters
polycyclic ^.^ .^©$1
aromatic ISIS) ^SISJ
(PAHs)
aliphatic c,H,2 C8H,8 c,8H38
	 1 	 U 	 1 	 L 	 1 	 1
           (c)
                   i<
                    12
                                         (d)
            106  IO8   10'°
Kow,octanol-water partition constant
       (mol Li'mof'tw)
Figure 28.
        C5,,'1, water solubility (mol.L'M
For several important classes of organic compounds, ranges of (a) saturation
vapor pressure at 25° C, (b) Henry's law constants, (c) aqueous solubility at 25° C,
and (d) octanol-water partition coefficients [from Schwarzenbach et al.,  1993,
Environmental Organic Chemistry, Copyright © 1993 John Wiley & Sons, Inc.,
reprinted by permission of John Wiley & Sons, Inc.].
                              96

-------
where KHi is Henry's law constant for compound i (atm-m3 /mol) [M-L/mol], Py is the vapor
pressure of the pure compound i (atm) [M/L-T2], and Cwi is the aqueous solubility (or molar
concentration) of compound i (mol/m3) [mol/L3]. The greater the Henry's law constant of a
compound, the greater will be the tendency of the compound to volatilize from an aqueous
solution.
       The dimensionless form of Henry's law constant is:


                      T^/ -
                                                                                  (90)
where K^ is an alternate form of Henry's law constant [dimensionless], CgMi is the molar
concentration of compound i in soil gas [mol/L3], and CwMi is the molar concentration of
compound i in the aqueous phase [mol/L3].  Equations (89) and (90) are related using the ideal
gas law as follows:
                      RT
= 41.6
                                    at 20°C
(91).
       It is important to consider the range of concentrations over which Henry's law and
Raoult's law are valid. Henry's law is appropriate for describing volatilization from low,
dissolved chemical concentrations; Raoult's law describes volatilization from NAPL. Figure 29
presents results from Johnson et al. (1990c) that illustrate the relationship between Henry's law
and Raoult's law for one hypothetical case. As shown, the deviation from Henry's law occurs at a
concentration of approximately 500 mg/kg of fresh gasoline in a typical soil.

       Henry's law constants are provided for numerous compounds in Appendix A; ranges for
several classes of organic chemicals are shown in Figure 28b. The vapor-phase transport models
described in more detail in this document (VENTING and VENT2D) do not require input of
Henry's law constants.

4.2.5   Aqueous Solubility

       Aqueous, solubility refers to the maximum concentration of a pure chemical that will
dissolve in pure water at a given temperature and pressure.  Solubilities may be obtained from
literature, measured experimentally, or estimated using empirical relationships developed
between solubility and other chemical properties such as partition coefficients and molecular
structure.  For example, Lyman et al. (1982) and Kenaga and Goring (1980) present many
regression equations that correlate aqueous solubility with Kow (octanol/water) and Koc (organic
carbon/water) partition coefficients for various  chemical groups.  Kow and Koc values for many
organic chemicals are provided in Appendix A. Nirmalakhandan and Speece (1988) developed a
predictive equation for aqueous solubility based on correlations between molecular structure and
solubility of 200 environmentally relevant chemicals. For most chemicals of interest, multiple

                                           97               :

-------
                    100
      Benzene Vapor
      Concentration
          (mg/l)
                      10-
Henry's law
                               Raoult's law
                         100
                                           ;  Divergence from Henry's Law
                                             at about 500 mg/kg for fresh
                                           i  gasoline in a typical soil
                                                                T = 20°C
         1000

Residual Soil Concentration
   (mg-gasoline/kg-soil)
                         10000
Figure 29.    Results from Johnson et al. (1990c) that illustrate the concentration ranges over
             which Henry's law and Raoult's law are valid for a particular case study.
                                          98

-------
 aqueous solubilities are reported in the literature (Montgomery and Welkom, 1990;
 Montgomery, 1991; Lucius et al., 1990; Verschueren, 1983). Aqueous solubility values for
 numerous compounds are given in Appendix A, and ranges of aqueous solubility for selected
 classes of organic compounds are shown in Figure 28c.

       As with vapor pressure, the effective solubility of a compound in a multicomponent
 NAPL is reduced from its pure-phase aqueous solubility in proportion to its mole fraction in the
 NAPL mixture.  Thus, assuming an ideal solution, the effective solubility can be calculated using
 Raoult'slaw:
= x,  C,,
                                                                                  (92)
where Cwi is the effective aqueous solubility of NAPL mixture component i [M/L3], and
the aqueous solubility of the pure compound i [M/L3].                  "
                                                       is
       Factors affecting solubility include temperature, cosolvents, salinity, and dissolved
organic matter. As shown in Figure 30, the relationship between temperature and the aqueous
solubility of organic chemicals is less significant (and less uniform) than between temperature
and vapor pressure. The effect of cosolvents on aqueous solubility depends on the specific mix
of compounds and concentrations. Based on laboratory data and modeling, Rao et al. (1991)
conclude that solubility enhancement for most organic chemicals will be minor (<20%) unless
cosolvent concentrations exceed 2% by volume in pore water.  Banerjee (1984) and Groves
(1988) describe methods to predict the solubilities of organic chemical mixtures in water based
on activity coefficient equations.  The aqueous solubility of organic chemicals generally declines
with increasing salinity (Rossi and Thomas, 1981; Eganhouse and Calder, 1973).  Dissolved
organic matter, such as naturally occurring humic and fulvic acids, are known to enhance the
solubility of hydrophobic organic compounds in water (Chiou et al.,  1986; Lyman et al., 1982).

       The vapor-phase transport models reviewed herein (VENTING and VENT2D) follow the
mass balance approach described in Section 3.2.5 and require input of the pure component
aqueous solubility for each compound considered. The aqueous phase in these codes is
considered nonideal; thus, activity coefficients are estimated from solubility data using Equations
(44) and (45).

4.2.6  Sorption

       Sorption refers to chemical partitioning to solids from other phases that results from the
affinity of a solute for a solid phase but may also by due to the  lack of affinity of the solute for
the solvent phase. It includes adsorption onto particle surfaces and absorption into particle
matrices. Many of the chemicals of environmental concern which are addressed by SVE are
hydrophobic organic compounds that preferentially partition to nonpolar substances such as
NAPL and soil organic carbon relative to water or mineral surfaces.
                                           99

-------
                           T(°C)
            160 140 120 100  80   60   40 30 20 10  0
         0

         -1


         •2


         -3

      "e
      S  '*
      o
      Q.
      I"  -5

         -6


         -7


         -8


         -9


         -10
                           3.0 "
                                       3.5 MO'3
                            (a)
                                                                      T(°C)
                                                            40 35 30 25 20 15 10  5  0
-2
-3
-4
                          iCH,Br /superheated)
                               \  liquid  /
                          • CH3Br '(gas, I aim)

                           CH2CI2 (liquid)
                               (liquid)
                      	.   CCI2 - CHCI
                       0       (liquid)
                           C°Io)
                         (subcooled liquid)
C9l°J
(solid)
       3.3 xKT3    3.5 xlO'3    3.7xlO'3
                                                                          (b)
Figure 30.      Temperature dependence of vapor pressure and aqueous solubility for several
                representative organic compounds [from Schwarzenbach et al., 1993,
                Environmental Organic Chemistry, Copyright © 1993 John Wiley & Sons, Inc.,
                reprinted by permission of John Wiley & Sons, Inc.].
                                                100

-------
        The solid-water distribution coefficient, Kd, is a valid representation of the partitioning
 between the solution phase and the solid phase only if the partitioning is fast (compared to the
 flow velocity) and reversible, and the isotherm is linear. Although many organic compounds
 reportedly follow a linear adsorption isotherm (at least over a finite range of concentration),
 others do not and sorption is not always rapid enough to be in equilibrium. Karickhoff et al
 (1979) investigated the sorption and desorption of organic contaminants and found that a very
 rapid component of sorption was followed by a slower component that required days or weeks to
 complete. The slower component was visualized as diffusive transfer to interior sorption sites
 that were inaccessible to the fluid.

        The distribution coefficient, Kd [L3/M], is related to organic carbon in porous media
 according to:
                        K  = K  f
                          o      oc oc
(93)
 where Koc is the organic carbon partition coefficient [L3/M] based on organic carbon content, and
 foc is the mass fraction of nonvolatile total organic carbon in a porous medium [dimensionles's].
 Equation (93) is generally considered strictly valid only if foc > 0.001, which is not always the
 case for sandy soil. Values of foc are site and soil horizon dependent, and typically decrease with
 depth below the organic-rich upper soil horizon.  Several foc values reported in the literature are
 given in Table 11.

       Many studies of various chemical groups have been conducted to relate Koc to aqueous
 solubility, octanol-water partition coefficient (Kow ) [dimensionless], and various other
 parameters (Chiou et al,  1977;  Lyman et al, 1982). Karickhoff et al. (1979) determined the
 following relationship between Koc and Kow and Koc and Sw:
and
                       K   = 0.63 K.
               log KM  = -0.54 log Sw +  0.44
                                                                                    (94)
(95)
where Sw is the aqueous solubility of a compound expressed as a mole fraction. Equation (94) is
incorporated within the VENTING and VENT2D codes.  Values of Koc, Kow, and aqueous
solubility (expressed in terms of M/L3) are given in Appendix A. Ranges of Kow for several
classes of organic compounds are shown in Figure 28d.

       The distribution coefficient (Kd) is related to the retardation factor, Rf [dimensionless], for
the saturated zone by the following:
                                           101

-------
Table 11.      Fraction of organic carbon (foc) values measured in various soils (from
              Mercer and Waddell, 1993).
Formation
Sand
Silty clay
Sandy loam
Silty clay
Silt loam
Silt loam
Silty clay loam
Silty clay loam
Sand and gravel
Sand
Fine sand
Silt loam
foc Value
0.00028
0.162
0.108
0.017
0.010
0.019
0.026
0.018
0.00008
. 0.001
0.00087
0.016
Reference
Patrick etal. (1985)
Nathwani and Philips (1977)
Nathwani and Philips (1977)
Nathwani and Philips (1977)
Nathwani and Philips (1977)
Chiouetal. (1983)
Rogers etal. (1980)
Rogers etal. (1980)
Thomas etal. (1988)
Thomas etal. (1988)
Wilson etal. (1981)
Thurman (1985)
                                           102

-------
                      Rf = 1
= 1  +
                                                          (96)
where 6 is porosity [dimensionless], and pb is the bulk density of the porous media [M/L3].  Bulk
mass density is related to particle mass density, ps [M/L3], by:
                        Pb  =
                                                           (97)
where ps = 2.65 g/cm3 for most quartz-rich mineral soils. Values of pb for various soils are listed
in Table 6.

       Where NAPL is not present in the vadose zone, the equilibrium retardation factor for the
vapor phase, Rg [dimensionless] is defined as:
                 Rg = 1  +
                                                           (98)
where 0W is the water-filled porosity [dimensionless], 9g is the air-filled porosity [dimensionless],
and K'H is Henry's law constant (dimensionless form). Equation (98) is based on Henry's law,
which assumes low dissolved concentrations in soil pore water.  Thus, Equation (98) is invalid in
the presence of NAPL. For the case where NAPL is present, the relationship for a constant
retardation factor does not hold. As may be seen, R increases with increasing water, content and
increasing distribution,coefficient, and decreases with increasing air-filled porosity and
increasing Henry's law constant. More mobile chemicals have low retardation factors.

       For example, consider trichloroethene (TCE), which has a K00 = 126 mL/g. Using
Equation (93) and a foc value of 0.0017 percent, the distribution coefficient is 2.14 mL/g. If
porosity is 0.3, then pb is calculated to be 1.85 g/cm3, and the retardation in groundwater^using
Equation 96) is 14,2. For conservative chemical movement, Rf = i. Thus, for this example, TCE
moves only 7 percent (100/14.2) as fast as groundwater. Using the same data, but.assuming
6g=0.15 rather than 0=0.30, a retardation factor for the vapor phase can be calculated for TCE. If
K^ is 0.379, then using Equation (98), the R for TCE is 73.26.

       The soil moisture content can affect the adsorption process in several ways (Valsaraj and
Thibodeaux, 1988). In soils with very low water contents (below that needed to form a
monolayer) water molecules can effectively compete with nonpolar organic molecules for
adsorption sites on mineral and organic matter and can therefore reduce the adsorption of VOCs
(Figure  31).  In a dry soil, VOCs can be adsorbed without competition from water molecules. At
moisture content levels near those required to form a monolayer, mineral and organic matter

                                           103

-------
      a) DRY
                            0
                   Non Polar
                   Organic
                 O H20
      Vapor
      Phase

      Adsorbed
      Layer
  0

DoTJ
            0       o
           TO nnn "a
      Solid Surface
       b) DAMP
             0
                 0
      a°
OoOooooOObODoood
       c) WET
            0
          ooo
      0°00°°°o
       OOOOO
                                 OC5"
Figure 31.  . Illustration of VOC adsorption under different moisture conditions (after Valsaraj

       and Thibodeaux, 1988). Reprinted by permission of Elsevier Science.
                       104

-------
 surfaces in the soil may be wetted by water and hence restrict the adsorption of VOCs, because
 both water and mineral fractions are more polar than most VOCs. However, under these
 conditions, the nonpolar or slightly polar VOCs can partition between soil water and soil organic
 matter, for which VOCs have a greater affinity. Therefore, adsorption to soil solids can vary as a
 function of soil moisture content.

        Smith et al. (1990) present results of laboratory experiments that illustrate these concepts.
 Figure 32 presents plots of TCE vapor uptake versus concentration for several different relative
 humidities.  At 0% relative humidity, adsorption appears to be controlled mainly by adsorption
 onto mineral surfaces, as reflected by the similarity of the isotherm's shape to that of a typical
 Brunaur Type-II adsorption isotherm (Adamson, 1976). At higher relative humidity, TCE
 adsorption on mineral surfaces is reduced by competing water molecules.  At 100% relative
 humidity, TCE uptake is most likely dominated by partitioning onto the soil organic matter.
 Chiou and Shoup (1985) demonstrated similar moisture content effects on adsorption for
 benzene, m-dichlorobenzene, and 1,2,4-trichlorobenzene vapor.

 4.2.7  Biodegradation                                          ,

       Vapor extraction induces air flow and increases the oxygen supply in the vadose zone.
 As a result of enhanced oxygen delivery, SVE stimulates aerobic degradation of contaminants by
 indigenous microorganisms.  Enhanced biodegradation by SVE (bioventing) has been the subject
 of extensive research during the past decade (e.g., Norris et al., 1994). Degradation rate ranges
 under a variety of environmental conditions are estimated for 313 chemicals by Howard et al.
 (1991).  The SVE models examined herein are not designed to simulate degradation or chemical
 reactions. The VENTING code, however, contains a simple scheme to account for
 biodegradation through the dimensionless efficiency factor term (e) in Equation (33).

 4.2.8  Contaminant Composition

       Composition of multicomponent mixtures can complicate attempts to try to estimate fate
 and transport behavior.  Multicomponent mixtures  such as gasoline contain hundreds of
 chemicals, which span a range from very volatile (e.g., benzene) to semivolatile (e.g.,
 naphthalene)  compounds. The simultaneous presence in the vadose zone of compounds with
 such widely varying characteristics makes estimates of transport somewhat difficult. Some
 models which are, based;on a mass-balance approach to the transport equation (e.g., VENTING)
 are able to circumvent this problem somewhat.  For more rigorous transport modeling, the
number of components that can be simulated at once may be limited, because the model must
 solve the transport equation for each component.

      It is also important to be aware of all components of a multicomponent mixture when
considering SVE. While SVE is well suited to remove components with high vapor pressures
and Henry's law constants, a substantial mass of semivolatile compounds may remain in the
subsurface after SVE is completed.  The significance of such residual contamination will depend

                                          105

-------
UJ
Z
UJ


I-
O


°S
          ^     6
          dl
        o
        CO
                        Percent Relative Humidity

                              (from water)


                               0


                              40    A—A


                             100
                  0.0   0.1   0.2   0.3   0.4   0.5   0.6   0.7   0.8   0.9   1.0
                                             P/P°orC/S
Figure 32.    Soil uptake of TCE as a function of relative humidity (after Smith et al., 1990).

             [Reprinted with permission from Environmental Science and Technology, 1990,

             American Chemical Society.]
                                        106

-------
on potential exposure pathways, risk characteristics, and cross-media impacts. Transport
modeling should provide useful information for determining patterns of preferential removal of
VOCs.

       Vapor-phase transport models require chemical composition data to simulate
multicomponent problems and chemical property input for individual compounds.  Values for
many of the chemical properties required by vapor-phase transport models (e.g., boiling point
and molecular weight), including some of those properties described above, are provided for
numerous organic compounds in Appendix A.

4.3    MODEL BOUNDARY AND INITIAL CONDITIONS

       Boundary conditions are necessary in any mathematical model to define the relationship
between the model domain and the outside world. There are two main types of boundary
conditions implemented in SVE models that will be considered here. The "first type" boundary
condition (known as the Dirichlet condition) specifies constant pressure or constant contaminant
concentration along model cells or nodes that represent a model boundary. For air  flow
simulations, prescribed flow and no-flow boundaries are analogous to groundwater flow
conditions; however, constant pressure boundaries are prescribed rather than constant head
boundaries. Constant-pressure nodes are frequently specified to represent wells, peripheral
boundaries, and/or the ground surface. The "second type" (Nuemann) boundary condition
specifies a flux of air or contaminant mass along model cells or nodes that represent a model
boundary. A special case of the second type boundary condition is the no-flow boundary, which
specifies that no flux occurs across that boundary (e.g., at the water table). The only type of
constant air flow condition implemented in SVE models (e.g., AIR3D and AIRFLOW) is the no-
flow condition, in which the permeability of the no-flow cells is set to zero.

       Air flow model boundaries are typically represented as follows:

     •  When modeling groundwater flow, pumping wells are typically represented as constant
       flux nodes. For air flow modeling, however, SVE wells are specified as constant pressure
       nodes where the pressure is less than atmospheric and depends on the magnitude of the
       applied vacuum. Air flow rates are then calculated by the model and contained in .the
       model output. This is done because representing extraction wells as constant flux nodes
       introduces nonlinearities to the mathematical solution. If the SVE well diameter is
       reasonably approximated by its model cell size, then soil permeability can be calibrated to
       match the observed SVE well vacuum and flow rate.  Alternatively, the size of the cell
       representing a well can be adjusted to reach agreement between the simulated and
       observed flow rate.

     •  Passive air inlet trenches or wells are modeled as nodes at constant atmospheric pressure.
                                         107

-------
     •  The bottom of the model domain is usually specified as a no-flow boundary at the water
       table.

     •  Peripheral model boundaries are represented as no-flow or constant pressure nodes, the
       latter allows lateral air inflow from the surrounding domain to be induced by SVE into
       the rnodel domain. Alternatively, peripheral model boundaries can specified as no flow
       or leaky to represent low permeability flow barriers (e.g., manmade or geologic).

     •  When boundary conditions are unclear, they can be placed far enough away so as not to
       significantly affect results in the area of interest.

   ,  •  Ground surface is represented as no-flow, constant pressure (atmospheric), or leaky. The
   -    boundary at the ground surface varies depending on the surface conditions.  The ground
       surface may be fully or partially covered with  uncracked pavement or some other low
       permeability material, or it may be open to the atmosphere. The boundary condition at
       the ground surface has significant effects on the radius of influence of an extraction well
       (Figure 7). In AIR3D, an additional model layer is added above  the top (soil) layer
       specified by the user and set to atmospheric pressure to represent the interface between
       air and the ground. The ground surface layer will be no-flow if its permeability is set to
       zero, open air if its permeability is the same as the permeability of underlying soil, and
       leaky if its permeability is  less than that of the underlying soil (e.g., where SVE occurs in
       a lower unit). VENT2D allows input of the thickness and vertical permeability of a
       surface confining layer to simulate surface leakage.

       Initial conditions are necessary for transient air flow modeling and contaminant transport
modeling.  They describe the distribution of some state variable (e.g., pressure or contaminant
concentration) at some initial time (Bear and Verruijt, 1987). For example, the initial gas
pressure throughout the model domain prior to the  startup of SVE is generally  set equal to the
atmospheric pressure. For transport simulation, the user must initialize chemical concentrations
in the model region.  VENT2D and VENTING are limited to input of a uniform initial
contaminant distribution. For transport simulations, the accuracy of the  initial  conditions reflects
the accuracy of the characterization of source areas. Uncertainty in predicted removal time for
any compound resulting from SVE will be proportional to uncertainty associated with the
estimate of initial mass-in-place because removal time is proportional to the initial contaminant
mass. Initial condition input required by specific SVE models is described in Section 6.

4.4    MODEL GRID DESIGN

       A limitation of finite-difference and finite-element numerical models is the need to
discretize the model domain into cells. Discretization is a critical step in model development. A
fine grid produces more accurate results, but costs in terms of computational run time and
hardware requirements.  Generally it is desirable to use a variable grid spacing, with a finer mesh
in the areas of interest, extraction, and contrasting soil properties. Numerous authors (Voss,

                                           108

-------
1984; Istok, 1989; Anderson and Woessner, 1992; Mercer and Faust, 1992) have presented
guidelines for design finite-difference and finite-element grids. A few grid design
recommendations are presented below:

     •  Locate "well" nodes or cells as near as possible to the location of the extraction well or
       cluster of extraction wells (Figure 33a).

     •  Boundaries should be located accurately, especially in areas where significant
       concentration or air pressure gradients are expected.  The grid should be aligned with the
       model domain boundary (Figure 33a).

     •  Grid spacing should be reduced in areas where significant concentration or air pressure
       gradients are expected, such as in the vicinity of an extraction well or a contaminant
       source area (Figure 33a,b).

     •  Grid spacing should not be changed abruptly across the model region. A good rule of
       thumb is to increase grid spacing by no more than a factor of 1.5 at a time (Figure 33b).

     •  The grid should be aligned with the major directions of anisotropy (primary) and air flow
       (secondary). It is not always possible to align the grid with directions of both anisotropy
       and air flow.

     •  The boundary between two different soil types should coincide with a grid boundary
       (Figure 33b).

       An advantage of using analytical (as opposed to numerical) models is the ability to
calculate pressures without having to discretize the model domain (Nyer et al., 1994). Available
analytical models for groundwater flow (e.g., Fitts, 1989; Strack,  1989) can be modified to
simulate SVE.         ,

4.5    DATA LIMITATIONS

       Due to data limitations, there are inherent uncertainties in any model of the vadose zone.
National Research Council (1990) presents a comprehensive discussion of model uncertainly and
reliability.  They define modeling as the "art and science of collecting a set of discrete
observations (our incomplete knowledge of the real world) and producing predictions of the
behavior of a system."  The success of any modeling exercise depends on how well the real
world is represented by the model and how well the model limitations can be understood and
accounted for when making interpretations.  As discussed in Section 2.1.4, sensitivity analysis
can be conducted to identify the model input parameters that have the most influence on model
results, and thus should be most thoroughly characterized.                             '
                                           109

-------
                                       IMPERMEABLE BOUNDARY
                   (a)
                                                                        PUMPING WELL
                   (b)
               SCREENED
               INTERVAL
                                         IMPERMEABLE BOUNDARY

                                   AIR EXTRACTION WELL
                                                                         CLAY
SAND
Figure 33.     Area! grid (a) and vertical grid (b) after Istok (1989): grid cells should coincide
              with point sources and sinks and be placed along boundaries (a); grid spacing
              should be reduced in areas where significant flow or concentration gradients are
              expected and soil type boundaries should coincide with grid boundaries (b).
                                           110

-------
       Vadose zone properties affecting SVE, such as permeability, porosity, and water content,
can vary significantly within relatively short distances and with time. Approaches such as
stochastic modeling attempt to quantify subsurface variability (Gelhar, 1986; Sudicky 1986), but
these techniques are expensive and time-consuming. Contaminant transport data are
characteristically uncertain. The nature and distribution of contaminants is usually interpolated
based on relatively few measurements. Many contaminant parameter values are routinely taken
from the literature, and may not adequately represent field values, even when variations in
temperature and other parameters are taken into account. Nonequilibrium effects noted in
Sections 3.1 and 4.2 must be considered when interpreting model output. Finally, boundary
conditions at the surface and perimeter of the model domain are often ill-defined. Modeling
results, including sensitivity analysis, therefore, must be combined with good professional
judgment to determine the need for additional data collection and to interpret simulation results.
                                           Ill

-------
                                      SECTION 5
                            MODEL SELECTION PROCESS

       The model selection process logically follows the question of need for a model discussed
in Section 2.1.2. It entails choosing an available code to simulate processes and features of the
real world that are relevant to analyzing a particular S VE issue. The intended model use and the
availability of data to support modeling are key factors that influence model selection as well as
the level of detail and cost that are appropriate for a particular modeling effort.

       SVE models are most frequently developed to evaluate SVE feasibility, design, and/or
performance. Preliminary feasibility analysis may involve the use of worksheets, such as that
presented in Table 12, as a precursor to modeling. Once a decision has been made to further
consider SVE, relatively simple or complex models can be used to develop a more detailed
analysis of SVE feasibility and design. Basic issues which should be addressed when designing
an SVE system are listed in Table 13. Design optimization is facilitated by using models to test
hypotheses regarding well placement, extraction rates, and cleanup time.  Currently, most models
are either empirical or deterministic, thus design optimization is generally performed by
comparing simulation results for alternative designs. Performance assessment modeling
typically involves integrating improvements to the site conceptual model  gained by SVE
operation back into the mathematical model. This is followed by re-examination of vacuum
distribution and mass removal trends. Common goals of performance assessment modeling are
to improve system design and cleanup trend projections.

5.1    CODE SELECTION CRITERIA

       A model code should be selected based on the type of processes being  simulated and the
needs'of the model user. Whereas a screening model can be used to conduct an initial feasibility
evaluation, more complex models might be required for detailed design and predictive analysis.
The code selection process is accomplished by matching the desired model processes and the
intended use with the required model criteria. USEPA (1988) presents a methodology for code
selection, which is summarized below.

       In order to select a model in an effective manner, a number of considerations  should be
made, including costs associated with running the model and level of sophistication of the model
user. In addition, the following code selection criteria should be considered, and are  discussed
below:
       Availability
       User support
       Degree of user-friendliness
       Portability
       Hardware requirements, and
       Reliability, credibility and extent of use.
                                          112

-------
Table 12.   Worksheet for evaluating the feasibility of soil venting (after USEPA, 1990c).
Critical Success Factor
Units
Success
Less
Likely
Success
Somewhat
Likely
Success
More
Likely
Site Related '
Dominant Contaminant Phas.e
Soil Temperature
Intrinsic Permeability
Moisture Content
Geological Conditions
Soil Sorption Capacity
Surface Area
Typical Ranges
V. Coarse Sand (0.001-0.003)
Coarse Sand (0.003-0.005)
Medium Sand (0.005-0.01)
Fine Sand (0.01 -0.03)
Very Fine Sand (0.03-0.1)
Silt (0.1-1)
Clay (>0.1)
Depth to Groundwater
Phase
°C
cm2
% Void
Volume
—
m2/g
Meters
Sorbed to soil
Low
(<10)
Low
(<10-10)
Moist
(>30)
Heterogeneous
High
(>1)
Low
(<1)
Liquid
Medium
(10-20) :
Medium
(10-10-10-8) ,
Moderate
(10-30) , ,
— . - '•'

Medium
(1-5)
Vapor
High
(>20)
High
(MO-8)
Dry
(<10)
Homogeneous
Low
(<0.1)
High |
(>5) 1
Contaminant Related
Vapor Pressure
Water Solubility
mm Hg
mg/L
Low
(<10)
High
(>1000)
Medium
(10-100)
Medium
(100-1000)
High
(>100)
Low
(<100)
                                                                                                 f I
                                                                                                 [t
                                                                                           '<• *.F.
                                        113

-------
Table 13.
A summary of the basic issues to be addressed in planning for the development of
a vapor extraction system (from Bedient et al., 1993; after Johnson et al., 1990b).
             Major Activity/Issue
                                            Approach to Resolving Issue
 Feasibility analysis:
       What is the expected range in rates of
       air flow for a single well?
                                Estimate Q with the flow equation
       What is the estimated removal rate with
       a single well?
                                Estimate total concentration of vapor in the air flow Ctol

                                Estimate single-well removal rate (percent air flow
                                contacting spill x Q x C)
       Is the single-well removal rate
       acceptable?
                                Compare removal rate for a single-well with the rate
                                required (i.e., total mass of spill x target time for recovering
                                spill)
      What residual contamination will remain
      at the end, and will the level of cleanup
      meet regulatory requirements?
                                Given the air flow rates and a model describing the
                                removal rates as a function of time, determine soil
                                concentrations of key constituents at the end of the
                                remedial period

                                Compare these levels with requirements of the regulations
 5.    What negative effect may develop as a
      consequence of vapor extraction?
                               Evaluate the possibility of offsite contributions from other
                               sources

                               Evaluate the possibility of a water table rise accompanying
                               vapor extraction
 Testing with vapor extraction and groundwater
 wells:
                               Evaluation of pumping tests gives site-specific data
                               necessary for finalizing the design of the vapor extraction
                               and water control systems

                               Monitoring of vapor concentrations during the well test
                               helps to validate earlier calculations and establish whether
                               efficiencies are affected by heterogeneities
 Final system design:
                               Site-specific data is used in conjunction with flow equations
                               to determine the single-well removal rate

                               The required removal rate is divided by the single-well rate
                               to establish the number of wells
Monitoring:
                               Monitoring documents the progress of the cleanup and will
                               assist in determining whether changes to the system are
                               required

                               Monitoring helps to establish when the system can be shut
                               off
                                             114

-------
       Code availability affects the decision whether to use it, because not all codes are readily
available or easily accessible. A code that has limited distribution and is used by few people is
usually less desirable than a more well-known, popular code. For reasons of availability, all of
the codes reviewed in this document are readily available at present either from commercial
vendors or public sources. Models are often categorized as being either in the public domain or
proprietary. Public domain models are generally preferred; but other considerations, such as
code capabilities and the availability of technical support, may be more important.  Public codes
are inexpensive (typically less than $100) compared to proprietary codes (typically $100 to
$1000).                                        ,

       User support is an important factor in considering any code. Many potential difficulties
face the new user of any model, both during code installation on a particular computer, and
during initial code application. In these situations, it helps to have technical support to assist the
new user.

       Usability is a measure of the degree of ease with which a model can be operated by a
given user. The input/output structure of a code has great effect on its usability. Codes with a
graphical user interface (GUI) are typically easier to use, especially for beginners, than those
with a simple text interface. However, some experienced modelers prefer a text-based model
interface because it may allow the user a greater degree of flexibility in specifying input and in
presenting output. A GUI is no substitution for code knowledge, which is required to avoid code
misuse. Problems associated with code incompatibility with hardware or operating system
parameters degrade its usability, but these problems can be minimized given adequate code
support and maintenance.

       Portable codes can be readily transferred from one operating environment to another. A
portable code should be flexible enough to run under a wide range of operating systems and
hardware platforms, and be able to move seamlessly from machine to machine within a single
operating platform.

       Increasing desktop computing capacity allows simulation of larger problems (e.g., more
grid blocks and more complex equations).  Significant computer resources are needed to run the
more advanced air flow and compositional flow and transport models. For example, to simulate
a reasonably large air flow problem with several tens of thousands of grid blocks on a PC will
likely require 4 to 8 MB of RAM and 50 MB of disk space.

       Users should be aware of problems with hardware and software compatibility. More
complex flow and compositional flow and transport models typically require a significant
amount of machine capacity. Many of the models reviewed require a large amount of base
memory (the lower 640K of memory on a PC). Thus, the user may have to eliminate as much
extraneous memory usage as possible. This would typically involve closing the Windows
program, disabling network software, and removing any other memory-resident programs.
Often this process entails developing a "vanilla" boot configuration so that the lower 640K of

                                          115

-------
 memory is left as free as possible. Another issue is the use of extended memory managers.
 Many of the models reviewed use various extended memory managers to take advantage of
 memory beyond base memory. These memory managers can potentially conflict with some
 hardware and software configurations.  One solution to this problem is for the model vendor to
 provide users with the model source code, so that they can compile the code using an extended
 memory manager of their choice.

       The reliability and credibility of a model depends upon its ability to be applied to a wide
 range of problems and its general acceptance in the scientific and regulatory community. It
 generally takes several years before a model gains significant credibility. Almost all
 hydrogeologic models evolve over a period of several years before measurable confidence and
 dependability are achieved. Once a model has been well tested and shown to be reliable, people
 are generally more accepting of the model results.  This is particularly important with regard to
 regulatory acceptance.

 5.2    WHAT TYPE OF MODEL SHOULD BE USED (NO MODEL, SCREENING, AIR
       FLOW, COMPOSITIONAL FLOW AND TRANSPORT)?

       A key decision in the model selection process is what type of model should be applied to
 a particular problem. In many cases, a relatively simple analytical  solution may be appropriate to
 perform the necessary analysis and predictive tasks. The model user should avoid overkill (i.e.,
 performing a more complex analysis than needed). A flowchart depicting the model selection
 process is presented in Figure 34.  .

       Any modeling effort begins with development of a site conceptual model identifying
 controlling processes and problem geometry.- Screening models are appropriate to make
 preliminary determinations of SVE feasibility, extraction well parameters, and contaminant
 removal rates. Commonly used for detailed design of SVE systems, air flow models facilitate
 quantitative analysis of extraction well placement, flow rate, and air flow patterns within the
 contaminated region. The user can test different extraction well vacuums, well placements, well
 numbers, and screened intervals with an air flow model.  The user can also identify portions of
 the contaminated region that are unlikely to be flushed effectively by air flow due to low
 permeability, heterogeneities, or the development of pneumatic stagnation zones.

       Use of a compositional flow and transport model allows simulation of air flow and vapor-
phase contaminant transport resulting from SVE. The design and performance analysis of SVE
systems can be completed more accurately (but not necessarily more cost-effectively) by using
models that simulate both the physical and chemical processes which govern the movement of
multicomponent contaminant mixtures in the subsurface. Compositional flow and transport
models allow projection of changing contaminant distributions with time of SVE operation,
thereby facilitating development of cleanup time estimates. They can also be used to pinpoint
problem areas in the subsurface where contaminants are not being removed effectively due to
mass-transfer limitations associated with, for example, soil structure heterogeneities.

                                          116

-------
                   Feasibility Evaluation
                   Preliminary Design
                   Preliminary Costing
    -»OR
       1
      Increasing
      Complexity
                  System Design
                  Extraction Well Placement
                  Screened Interval
                  Extraction Rate
                  Number of Extraction Wells
                  System Design
                  Contaminant Removal Rates
                  Source Composition History
                  Post-Remediation Soil Residual |
  Compositional
Flow and Transport I
Figure 34.     Decision flowchart for selecting which class of model (screening, air flow,
              compositional flow, and transport) to use.
                                           117

-------
 Compositional transport analysis can also be used to help design a field monitoring plan to
 determine the preferential removal of volatile components, and concentration of less volatile
 components, in the subsurface as SVE proceeds. In general, complex models should be applied
 only where needed and where their use can be supported by available data.

 5.3    USING SCREENING MODELS

 5.3.1  When to Use a Screening Model

       Screening models are useful in making initial assessments of SVE feasibility at VOC-
 contamination sites.  The screening models discussed in this document are all based on analytical
 solutions that entail restrictive assumptions. Screening models can be applied with very limited
 site data, and thus can be used early in the remedial planning process. Depending on site-specific
 conditions and requirements, it may be inappropriate to utilize screening models for detailed
 SVE system design.

 5.3.2  What Type of Results to Expect   '   '                             '    "  ""'

       Screening models can provide estimates  for, or aid in determining, air permeability,
 extraction well radius of influence, extraction well flow rate, rate of mass removal, and relative
 removal rates of different constituents of multicompbnent mixtures.  Pneumatic tests can be
 interpreted using analytical solutions for radial air flow to an extraction well to estimate air
 permeability.  Likewise, analytical solutions can be used to project vacuum distributions that
 would result from various extraction rates considering a given permeability and assumed
 boundary conditions.

       It should be stressed that these solutions  may be subject to fairly limiting assumptions.
 For example, application of a confined-flow analytical solution to a leaky system will result in
 overestimation of the area influenced by an extraction well. Beckett and Huntley (1994)
 performed an analysis of the error introduced by assuming confined flow in a leaky system and
 found that zones of capture were overestimated,  and cleanup times were underestimated
 (Figure 35).

       The rate of mass removal can be estimated using relatively simple mass-balance
 equations that can be solved analytically or numerically.  Relative removal rates of different
 constituents can be estimated using these same mass-balance equations for individual
 constituents of multi-component mixtures.

 5.3.3   Using Screening Models for Initial Site Characterization

       Screening models can be used to initially characterize a site in terms of the potential
feasibility of SVE for remediating the site. This can be especially useful for managers who are
responsible for prioritizing large numbers of VOC-contaminated sites.  For a given group of

                                          118

-------
10 1
g
8
52
8 7
% 6

o
u. 4 •
0
Ul
S3 .
H
2 -

1
0 -















S













:




,

AP
CO
tn
RE, feet
CO
o
N>
cn
AL ZONE OF C
-».-». IO
ocno
                       Kv = Kh
Kv = 0.1 Kh
 Kv = 0.01 Kh     Theis Solution

                                                                H-
                        Kv = Kh
Kv = 0.1 Kh
Kv = 0.01 Kh   Theis Solution
Figure 35.     Estimates of cleanup time and zone of capture produced from confined (Theis
              solution) versus leaky analytical models. Depending on the spatial relationship
              between extraction wells and the zone of contamination, confined models may
              underestimate cleanup times and overestimate zones of capture because they.
              assume no short-circuiting of air from the surface (after Beckett and Huntley,
              1994).  [Reprinted by permission of the Ground Water Publishing Company,
              Dublin, Ohio.]
                                          119

-------
 sites, the use of screening models should allow the sites to be ranked against each other and
 ordered by the relative potential for successful cleanup using SVE. This initial screening will
 identify those sites that have the most potential for cleanup via SVE and should proceed to the
 next stage of data collection, modeling, and system design.

 5.3.4  Available Codes

       The two screening codes that have been evaluated as part of this document are
 Hyperventilate and VENTING. Hyperventilate is available in the public domain from the
 USEPA (1993) and is a byproduct of the work of Johnson et al. (1990a,b). Hyperventilate can
 be used both as an educational tool to acquaint the user with SVE and as a screening model to
 develop a preliminary conceptual design for an SVE system. The VENTING code is a screening
 model that allows estimation of the rate of removal of contaminant mass from the subsurface
 based on air flow rates input by the user.  Like Hyperventilate, VENTING is based on the air
 flow and equilibrium partitioning, mass balance equations presented by Johnson et al. (1990a,b).
 Both codes are described in greater detail in Section 6.2.

       Many analytical solutions have been published in the literature and can be solved on a
 personal computer either by using spreadsheet software, a numerical analysis package, or some
 straightforward programming.  These analytical solutions can be useful  screening tools, and are
 presented in Table 14. Marley et al. (1990) developed an analytical solution to evaluate soil
 properties based on results of pilot SVE tests. Analytical expressions developed by Walter et al.
 (1990) can be used to calculate drain spacings required to .circulate  a specified number of pore
 volumes in a given time for a given pressure drop. Massman (1989) developed an analytical
 expression for vapor flux that is similar in form to the transient groundwater flow equation,
 which was then used to analyze the results of a field air extraction test. Baehr (1987) presents an
 analytical solution to the one-dimensional transport equation where the retardation coefficient is
 dependent on phase-partitioning coefficients and moisture content.  Brasseau (1991) developed a
 one-dimensional transport model called MPNEG (multiprocess nonequlibrium transport by gas
 advection) which accounts for structured or heterogenous porous media and rate-limited sorption.
 Shan et al. (1992) developed analytical solutions for modeling steady state air flow to a single
vacuum extraction well in isotropic or anisotropic media in which the ground surface is open to
the atmosphere. Analytic expressions for the gas pressure field and stream function distribution
 allow calculation of air streamline travel times which can be used to optimize SVE design.  Falta
 (1993,1995) presents a suite of analytical solutions for analyzing air pumping test data under
different scenarios (confined, leaky, open ground). Beckett and Huntley (1994) show that the
Hantush-Jacob (1955) analytical model for groundwater flow in a leaky aquifer is generally more
appropriate for intrepreting SVE tests than the confined flow equation that is commonly used.

5.4   USING AIR FLOW MODELS

      Air flow models are used to assess the subsurface air flow field that would result from
alternative SVE well placements, completions, and vacuums. Zone of influence and air removal

                                           120

-------
Table 14.      A sampling of analytical solutions available for analysis of SVE.
Reference
Baehr(1987)
Massman (1989)
Marleyetal. (1990)
Walter etal. (1990)
Brusseau (1991)
Shan etal. (1992)
Falta(1993)
Functionality
One-dimensional transport with
retardation
Analyze the results of a field air extraction
test ••-...
Evaluates soil properties based on results
of pilot air pumping tests
Calculate drain spacings required to
circulate a given number of pore volumes
in a given time for a particular pressure
drop
One-dimensional transport which
considers multiprocess nonequilibrium
Analyze steady-state air flow to a single
extraction well
Suite of analytical solutions for analyzing
air pumping test data under a variety of
scenarios
                                    121

-------
 rate calculations derived from air flow models facilitate design optimization.  A main advantage
 of air flow models is their ability to consider complex geologic conditions (e.g., anisotropy and
 heterogeneity), irregular boundary conditions, and distributed well placement and operation.

 5.4.1  When to Use an Air Flow Model

       Air flow models should be considered for application once the potential feasibility of
 SVE has been established at a site. More simplistic screening models should be utilized to make
 the initial determination as to whether SVE is reasonable for serious consideration. Frequently,
 the main use of an air flow model is to be a predictive design tool. Modeling can help optimize
 all of the SVE flow design parameters, including well vacuum and extraction rate, well
 placement, screened interval, and boundary modification (e.g., surface sealing). Assuming that
 the subsurface geology is fairly well defined and can be represented in the model, modeling can
 predict potential problems with the design such as incorrect well spacing or short-circuiting.  Air
 flow models can also be combined with contaminant transfer/removal data derived from pilot
 tests (or other means) in order to estimate cleanup time.

 5.4.2  Level of Complexity

       The level of complexity in applying an air flow model typically exceeds that associated
 with a simple screening model.  Whereas most screening models assume a homogenous
 subsurface and a single extraction well, a numerical air flow model allows the user to account for
 heterogeneous subsurface conditions (e.g., nonuniform permeability distribution), irregular
 placement of multiple extraction and injection wells, and varied extraction and injection rates.
 Most numerical flow models allow representation of two or three dimensions in space, rather
 than the one-dimensional axisymmetric domain simulated by many screening models.  In order
 to take full advantage of an air flow model, it is necessary to adequately characterize subsurface
 contaminant and permeability distributions, boundary conditions of the region to be modeled,
 and the locations and completion details of air flow sources and sinks (e.g., injection and
 extraction wells or trenches).

 5.4.3  What Type of Results to Expect

       Numerical air flow models calculate the soil gas pressure (vacuum) distribution resulting
 from input conditions. Some models process the calculated pressure distribution to determine
 flow pathlines, flow rates, and/or well capture zones. Alternatively, post-processing of simulated
pressure data can be accomplished using separate programs such as the general particle tracking
module of the USEPA wellhead protection code (Blandford et al, 1993). Measured field
pressures, if available, can be compared to simulated results in order to calibrate the model
 (Section 2.1.4). Once a reasonable match has been obtained between observed and simulated
pressures induced by SVE, the model can be used with some confidence to calculate potential
 changes in the vacuum distribution that would accompany changes in well location, extraction
rates, etc.  Pressure gradients and air flow directions can be inferred from the pressure

                                           122

-------
distribution. In addition, for areas where no field data are available, the pressure drawdown and
radius of influence of each extraction well can be estimated. The pressure distribution can also
be used to determine contaminated areas in the model region which are not being affected by
SVE (i.e.,  where little or no air flow is occurring).

5.4.4   Available Codes

       Three readily available air flow codes were evaluated for this document:  AIRFLOW,
CSUGAS, and AIR3D. AIRFLOW allows two-dimensional, axisymmetric analysis of air flow
to a single well, or a multiwell cluster, in a heterogeneous soil unit. CSUGAS and AIR3D allow
two- or three-dimensionsal simulation of air flow induced by multiple extraction wells in
heterogeneous porous media. All three codes are discussed in more detail in Section 6.3.

5.5    USING COMPOSITIONAL FLOW AND TRANSPORT MODELS

       There are several approaches to modeling of mass transport by soil vapor. The focus of
this discussion will be on compositional flow and transport models which simulate the transport
of multicomponent mixtures such as petroleum fuels. Single-component models that solve the
full advective-dispersive transport equation may be difficult to apply to multicomponent
mixtures such as gasoline. This is because only four to five components, at most, can be
simulated  at a time, and thus one simulation must be run for each group of compounds. Single-
component models also are generally based on Henry's law, and do not consider free-phase
NAPL that may be present. Exceptions to this limitation include MOTRANS (Kaluarachchi and
Parker, 1989; Kaluarachchi and Parker, 1990) and MAGNAS (Huyakorn et al., 1994; Panday et
al, 1994), both of which consider the presence of NAPL for transport calculations.

       Compositional flow and transport models can be used to estimate both the subsurface air
flow regime and the transport and removal of contaminants via SVE.  These models can also be
used to estimate post-remedial soil concentrations in order to determine if cleanup goals can be
met. Flow and transport modeling can be used to generate estimates for most aspects of SVE
design, ranging from well placement to contaminant extraction rates,  However, the level of
complexity of the modeling and data requirements are also  substantially greater than that
associated with screening and air flow models.

5.5.1   When to Use a Compositional Flow and Transport Model

       A compositional flow and transport model should be used when it is necessary  to attempt
to estimate both the air flow regime  and the transport of contaminants in the subsurface. The
transport portion of the model allows more robust analysis of contaminant removal rates than
simplified screening models, and provides estimates of contaminant levels in soil over the
operating life of the SVE system.  Compositional flow and transport models can be applied to
estimate the concentration of a specified contaminant in the vadose zone for any  given point in
                                          123

-------
space and time. Thus, it may be advantageous to develop a transport model to examine mass
removal trends given sufficient site characterization data.

5.5.2  Level of Complexity

       Simulating vapor transport increases both the complexity and uncertainty associated with
a modeling analysis. In addition to the parameters which affect air flow (Section 4.1), transport
simulation also requires characterization of the contaminant distribution and the contaminant,
media, and fluid properties which affect transport (Section 4.2) but not flow.  Characterizing
contaminant distribution, transport properties, and transport processes is typically subject to
significantly greater uncertainty than characterizing flow properties and processes.

5.5.3  What Type of Results to Expect

       Results from a transport model consist of contaminant concentrations throughout the
model region over time in soil and soil vapor. As such, transport models can be used to predict
or evaluate contaminant removal rates and residual concentrations. These results can be used to
assess the overall  efficacy of an SVE system and identify problem areas where mass removal
rates are inadequate.  Areas where simulated contaminant Concentrations remain at unacceptable
levels for the duration of the planned SVE operation may require the installation of additional
extraction or injection wells. Contaminant removal rates can be used to develop a general
estimate of cleanup time. A determination of residual concentrations can be useful both in
predicting possible cleanup levels and determining which, if any, semivolatile compounds may
be remain in place after SVE has been completed.  Such an analysis requires characterization of
the existing contaminant distribution.

5.5.4  Available  Codes

       Only one readily available compositional flow and transport code, VENT2D, was
identified and reviewed as part of this study. However, a new version of the AIRFLOW code,
entitled AIRFLOW/SVE, that includes multicomponent contaminant transport has become
available recently. VENT2D allows two-dimensional (areal or cross-sectional) analysis of air
flow and multicomponent transport with multiple extraction wells. A three-dimensional
extension of this code called VENT3D has also become available recently. Some of the new
features incorporated within VENT3D include the ability for individual layers to be given unique
values of thickness, anisotropy, porosity, moisture content, foc content,  and distributions of
permeability and contamination.

5.6    OTHER MODELS

       Other simulation tools potentially applicable to SVE have or might be developed to
evaluate nonideal  flow and mass transfer conditions during SVE, air sparging in the saturated
zone, bioventing,  and multiphase fluid flow.

                                          124

-------
5.6.1   Models that Account for Nonideal Conditions

       As noted in Section 4.2, the local equilibrium assumption is the basis for many SVE
models. Several models have recently been developed to examine nonideal mass transfer
associated with vapor-phase transport in the vadose zone (Armstrong et al., 1994; Gierke et al.,
1992; and Brusseau, 1991). Brusseau (1991) questions whether local equilibrium can be assumed
in an SVE application.  He found that air velocities may be too high to allow equilibration
between dissolved contaminants and the vapor phase. In addition, some researchers (Clarice et
al.,  1992) have determined that during SVE, vapor flow velocities may be high enough such that
flow is no longer laminar, and thus Darcy's law no longer holds. As discussed in Sections 3.1
and 4.2, diffusion-limited processes hinder SVE performance.  Contaminants trapped in low
permeability zones may extend the cleanup times for many SVE applications.  Lastly,
multicdmponent mixtures, such as gasoline, will respond favorably to vapor extraction only at
first and then extraction efficiency will decline over time as the boiling point of the mixture
increases with the loss of the more volatile fractions.

5.6.2   Air Sparging and Bioventing Models

       Some research has been completed on modeling air sparging and bioventing.  Air
sparging is a technique that involves aerating contaminated groundwater by injecting air through
wells into the contaminated saturated zone. The injection of air effectively creates an in-situ air
stripper, whereby dissolved, separate-phase, and sprbed contaminants partition into the injected
air. Numerous air sparging studies have been published (e.g., Herrling et al., 1991; Gvirtzman
and Gorelick, 1992; Coyle, 1985; Marley,  1992; Brown and Fraxedas, 1991; Kresage and Dacey,
1991; and Johnson et al., 1992). Air sparging works best for homogenous aquifers with fairly
high permeability. Heterogenous aquifers cause problems because flow is channeled through
preferential pathways.  Changes in oxygen content in the system may cause precipitation of
dissolved minerals and clogging of the system.

       In order to mathematically model the flow of air through the saturated zone, additional
data associated with the physics of multiphase flow must be gathered.  Since air sparging
involves multiphase flow, many numerical difficulties are associated with modeling.  Important
parameters for air sparging system design include sparge rate, well spacing, and chemical
partitioning.  The model must be able to describe the hydrodynamics around the air sparging
wells, and the fate and transport of the VOCs. A numerical difficulty referred to in the petroleum
literature as the gas percolation problem (e.g., Peaceman, 1977; Thomas, 1982) is problematic
for simulation of air sparging. High gas velocities arise from the low gas viscosity and large
density difference between the air and oil phase.  Numerical oscillations arising from these high
velocities can result unless the time steps are decreased to very small values.  Turbulent (non-
Darcian) flow may occur in the near-well area.
                                           125

-------
       Only a few numerical models of air sparging have been developed.  Herrling and
 Buermann (1990) use a simplified approach by superposition of a uniform regional flow field
 calculated hi a vertical cross section, and of a radial-symmetric flow field for each air sparging
 well. This allows estimation of the groundwater flow field in the vicinity of an air sparging well.
 They use the finite-element method for their model, and no solute transport is considered.
 Marley et al. (1992) also use a groundwater flow model to predict air pressures in the saturated
 zone. They assume that air flow obeys Darcy's law and that groundwater flow is negligible.
 Their model SPARG1 was developed by modifying an existing boundary element groundwater
 flow code; it does consider transport.

       Bioventing is an in-situ bioremediation process for the vadose zone, where air flow is
 provided to vaporize VOCs and to deliver oxygen to the subsurface to stimulate in-situ
 biodegradation of organic contaminants.  Bioventing systems are operated at lower air flow rates
 than SVE systems to reduce vapor extraction quantities, and increase biodegradation associated
 with longer vapor retention.  Soil moisture levels necessary for biological activity are usually
 higher than those recommended for optimum SVE systems. A comparison of design and
 application considerations for SVE versus bio venting is provided in Table 15.  Experience
 indicates that bioventing is optimized when performed in cycles of oxygenation followed by
 shut-in.  The addition of nutrients may also enhance bioremediation. A number of field and
 experimental studies have shown that bioventing can be effective in cleaning up such relatively
 nonvolatile hydrocarbons as diesel fuel (Bulman et al., 1993) as well as a hydrocarbons ranging
 in composition from gasoline to heavy fuel oil (Lee and Swindell, 1993).

       No models are reported in the literature specifically for use in simulating bioventing. Air
 flow models discussed for SVE may be used to help design bioventing well spacing and flow
 rates. Critical parameters required to rigorously predict biodegradation include oxygen status,
 soil moisture, nutrient availability, and contaminant concentration; other important parameters
 include soil temperature and pH. Several of the  computer codes reviewed in this document
 (VENTING and VENT2D) include the addition  of a degradation rate parameter to the transport
 equation. If the biodegradation rate due to bioventing can be quantified by bench- or field-scale
 testing, then it could be included using the degradation rate parameter in these codes.

 5.6.3   Complex Multiphase Flow and Transport Codes

       Several complex multiphase flow and transport codes that will model the simultaneous
 flow of water, NAPL, and gas are commercially  available. These models typically demand a
high degree of computing power and data input.  In order to apply these models, it is necessary to
define relative permeability relationships between air, water, and NAPL.  Gathering these types
of data can prove to be quite difficult. Although conceptually and computationally difficult to
use, multiphase flow and transport models allow more detailed analysis of certain aspects of
SVE problems than other available codes.
                                          126

-------
Table 15.   General design and application considerations appropriate for conventual versus
           bioventing SVE systems (modified from USEPA, 1990c and 1994).
Parameter
Compound type
Vapor Pressure
KH (dimensionless)
Aqueous solubility
Soil concentration
Depth to groundwater
Intrinsic Permeability
Subsurface conditions
NAPL phase
Vent well placement
Operating mode
Operating flow rates
Pore volumes per day
Optimal soil moisture
Nutrient requirement
Soil gas O2 levels
Toxicants
Conventional SVE
Volatile @ room temperature
> 0.5 mm Hg
>0.01
< 100mg/L
> 1 mg/kg
>20ft
Bioventing
Biodegradable
„
__
MX
< 1%
..-
>1x10'8cm2
Little or no stratification "
Some or hone
Within contamination
Maximum soil gas exchange
rate
46 to 700+ actual L/s
(100)to1,500+acfm)
1to15
~ 25% field capacity
• , -
—
-
Biodegradable
Outside contamination
Maximum retention time &
aerobic conditions
4.6 to 23 actual L/s
(10to50acfm)
0.1 to 0.5
~ 75% field capacity
Carbon:Nitrogen:Phosphorus
~ 100:10:1
> 2 vol%
Little or none
                                      127

-------
       Two commercially available codes are MOTRANS (Kaluarachchi and Parker, 1989;
Kaluarachchi and Parker, 1990) and MAGNAS (Huyakorn et al., 1994; Panday et al., 1994).
MOTRANS is a two-dimensional, finite-element multiphase flow and transport model.  It can be
used to simulate the flow of LNAPL or DNAPL in three-phase systems, multiphase transport of
several chemical species, and dynamic gas flow in saturated or unsaturated media.  MAGNAS is
a two- or three-dimensional flow and transport model which simulates the flow of water, NAPL
(light or dense), and air in heterogenous and anisotropic subsurface media. Several other
available codes are reviewed in USEPA (1993c).

5.7    WHAT SITE CHARACTERIZATION DATA REQUIREMENTS ARE ASSOCIATED
       WITH DIFFERENT TYPES OF MODELS?

       The availability of site data will influence model type selection. For small leaking UST
sites with little site characterization data, application of a simple screening model may be all that
is appropriate and possible. For a large Superfund site where significant site characterization
data are available, it may be more reasonable to develop a complex model. In some cases,
additional data should be collected to support an SVE modeling effort.  Input parameters
required to simulate air flow and vapor transport are listed in Table 5 and discussed in Section 4.

5.8    WHAT LEVEL OF USER KNOWLEDGE IS REQUIRED FOR DIFFERENT MODEL
       TYPES?

       National Research Council (1990) presents an excellent discussion on the issues
concerning the necessary qualifications of persons utilizing and evaluating numerical models.
They state that:

       "...modeling...is not a trivial exercise. Ideally, a modeler should have a broad
       background in earth sciences with particular strengths in hydrogeology, low-
       temperature geochemistry, and analytical and numerical mathematics. This
       background will have developed through graduate and undergraduate studies and
       will have been tempered by relevant experience."

This statement of qualifications will hold true whether the modeler is applying an analytical
screening model or a more complex numerical model. A significant problem stemming from
inexperience is an excessive reliance on model results.  An experienced user can temper
interpretation of simulation results with an understanding of model assumptions, the physics of
flow and transport, chemical processes, and site conditions.  Graphical displays and output with
many significant digits can make simulation results  appear overly accurate.  Any model user
should understand and consider the assumptions on  which a model is based when evaluating
simulation results.

       The level of user expertise needs to increase with the complexity of the modeling
exercise. A screening model, such as Hyperventilate, can be used by persons with a relatively

                                         128

-------
 limited experience. However, it is important to understand the assumptions and limitations of a
 simplified model when applying their results to SVE design. For example, even though it is
 stated in the Hyperventilate manual (USEPA,  1993) that "Hyperventilate will not completely
 design your vapor extraction system, tell you exactly how many days it should be operated, or
 predict the future," chances are good that some inexperienced users will rely on the code to
 provide these results. A common example of model misapplication is utilizing an analytical
 solution that assumes confined flow at a site with significant surface leakage.

       Air flow and compositional flow and transport models generally incorporate more
 assumptions and require more data input than simplified screening models. The user must be
 able to select and apply increasingly complex boundary and initial conditions effectively. As the
 complexity of the model increases, so does the complexity of the results and their potential
 uncertainty.  Simulation results of a compositional flow and transport model of a
 multicomponent mixture depend on numerous properties of each component.  In order to
 effectively interpret the results of the model, the modeler must understand the physics, chemistry,
 and mathematics of the model application.

 5.9    WITHIN WHAT TIME AND BUDGETARY CONSTRAINTS MUST THE
       MODELING EXERCISE BE COMPLETED?

       Careful consideration should be given to the time and budget available to complete a
 modeling exercise. Budget and time requirements generally increase with model complexity. A
 simple screening analysis can typically be completed within hours to days. It may take weeks to
 develop an air flow model of moderate complexity. For a complex air flow or compositional
 flow and transport model, the time frame to complete a modeling study may stretch to several
 months.

       The modeler should weigh the projected cost and benefits of model development when
 selecting a simulation approach. The most simplistic level of modeling that is viable should be
 used in order to simplify the modeling process, reduce complexity in the results, and minimize
 cost. There is no need to apply a complex and expensive model to a basic problem that can be
 solved by a simple model. In addition, some sites that are so complex that an appropriate tact is
 to conduct uncertainty analysis using a relatively simple model.  A flexible, phased
 implementation is recommended for installation of multiwell SVE systems. This allows system
 design to benefit from conceptual model improvements derived by monitoring system
performance.

 5.10   IS THE MAIN OBJECTIVE DATA ORGANIZATION AND ANALYSIS RATHER
       THAN MODELING?

       For many applications, a program of data organization and graphical display can be very
effective in making decisions concerning remedial alternatives. A Geographical Information
 System (GIS) could be used, for example, at a large site contaminated by many different

                                         129

-------
constituents, to highlight areas where the majority of soil contamination is made up of VOCs
present in the vadose zone. The GIS could select these areas based on querying various
databases to determine what areas of the site are contaminated with VOCs. The query could be
further narrowed by selecting only those areas that meet certain permeability or porosity criteria.
The final step is to intersect these two queries with the area present at the site where there is a ,
significant vadose zone above the water table. In this manner, the areas of the site which can
potentially be cleaned up via SVE are quickly and easily selected. This procedure is presented
schematically in Figure 36.
                                            130

-------
                                         SITE FEATURES
                                         AREAS OF VOC CONTAMINATION
                                         AREA OF HIGH PERMEABILITY
                                         AREA OF SIGNIFICANT VADOSE ZONE
                                          AREA WITH MOST POTENTIAL FOR
                                          SVE: VOC CONTAMINATION, HIGH
                                          PERMEABILITY AND A SIGNIFICANT
                                          VADOSEZONE
Figure 36.    Schematic diagram of the process of intersecting layers using a GIS. The
            intersection process allows the user to determine those areas which conform to a
            selected group of constraints (e.g., VOC contamination, high permeability,
            significant vadose zone). The area with most promise for cleanup via SVE is
            defined by the intersection of those areas characterized by VOC contamination,
            high permeability, and a significant vadose zone.
                                    131

-------
                                     SECTION 6
                EVALUATION OF READILY AVAILABLE SVE MODELS

       Models capable of simulating SVE that are readily available from companies,
organizations, or government agencies were evaluated during this study. An attempt was made...
to discuss codes herein that were most widely used and commonly available at the time of
writing.

       The computer codes described below include GIS programs, screening models, air flow
models, and compositional flow and transport models. GIS are used for data organization,
display, and analysis.  Screening models are used primarily to examine SVE feasibility and
conceptual design.  The ability to simulate pressure and flow fields in two or three dimensions
resulting from alternative well configurations makes air flow modeling a popular tool for detailed
design of SVE systems.  Compositional flow and transport models are used to calculate the
movement and removal of contaminant mass in the subsurface over tune, and thereby provide
insight to potential removal rates, cleanup times, and residual concentrations. Table 16 presents
a summary of the screening, air flow, and transport codes that were evaluated during this study.
Each codes is discussed further hi Sections 6.2 (screening models), 6.3 (air flow models), or 6.4
(compositional flow and transport models).  A brief discussion of GIS follows in Section 6.1.

6.1     DATA ORGANIZATION TOOLS

       Data organization tools are available in the form of database management systems
(DBMS) or GIS. Both of these tools allow functional organization of voluminous data such as
chemical concentrations, water levels, well construction data, and soil properties.  A GIS takes
the approach one step further by allowing graphical manipulation, querying, and comparison of
multiple databases; it provides an integrated system to organize, process, and view site
characterization data which vary in space and time. In addition, a GIS conveniently stores data
for future access.

6.1.1   GIS for Data Organization and Graphical Analysis

       A GIS is defined as (ESRI,  1992):

       "An organized collection of computer hardware, software, geographic data, and
       personnel designed to efficiently capture, store, update, manipulate, analyze, and
       display all forms of geographically referenced information."

       In application to SVE, a GIS can be used to store field measurements such as vapor ,
concentrations, well locations, extraction rates, etc. What distinguishes a GIS from a
conventional database is the spatial reference of the measurements. The spatial reference is
generally the sample point or monitoring well location. The GIS allows the users, for example,
                                          132

-------
     Table 16.    Summary of the screening, air flow, and compositional flow and transport codes that were evaluated.

Model
Hyperventilate,
v2.0 (IBM PC)
V1.01 (Apple
Macintosh)













VENTING, ,
V3.01










AIRFLOW™
V2.07












Type
Screening
















Screening










Air flow













Functionality
Simplistic steady
state, radial-
.symmetric air flow
and transient one-
dimensional
multicomponent
contaminant
transport









Transient, one-
dimensional
multicomponent "
contaminant
transport







Steady-state, radial-
symmetric (two-
dimensional cross
section) air flow









Solution
Methodology
Analytical solution

Finite-difference
solution of a one-
dimensional mass
balance equation











Finite-difference
solution of a one-
dimensional mass
balance equation








Finite-element
solution of the air flow
equation











Assumptions
Two-dimensional,
radial, confined air
flow to a vapor
extraction well

One-dimensional,
mass-balance
approach,
volatilization based
on Raoult's law







Calculations based
on user-defined flow
rate, assumes
equilibrium
partitioning between
phases in a one-
dimensional volume
of soil




Based on Darcian
flow of an ideal
compressible gas in a
porous medium










Capabilities
Calculates air
permeability, well
flow rates, mass
removal rate, mass
removal from
several idealized
diffusion-limited
scenarios

Calculates
contaminant
concentrations
over time for
multiple
constituents


Calculates
contaminant
concentrations
over time for
multiple
constituents






Calculated
pressure
distribution in a
radial domain,
calculates air flow
pathlines and
velocities







Advantages
Providesjapid
estimates for
determination of
the potential
feasibility of SVE

Provides rapid
estimates of
contaminant
concentrations in
extracted gas,
allows comparison
of removal rates of
different
constituents


Provides rapid
estimates of
contaminant
concentrations in
extracted gas,
allows comparison
of removal rates of
different
constituents



Easy-to-use 'CAD-
type' graphical user
interface which
simplifies model
input and setup

Rapid setup for
simple problems
aids in hypothesis
testing
Many sample
problems included


Limitations
Analytical air flow
solution

Should not be
used to design
SVE systems











User supplies flow
rate to extraction
well

Simplistic one-
dimensional
representation of
mass transport

Should not be
used to design of
SVE systems
Only allows for
one extraction
well

No mass
transport







Hardware/Software
Requirements
IBM PC or Compatible:
80386/80387 coprocessor or
80486, 4 MB RAM, DOS 3.1
or higher, Microsoft Windows
3.x and runtime version of
Object PLUS

Apple Macintosh (Plus, SE,
SE/30, II, IIX, or portable): 1
MB RAM, Apple HyperCard
Software (v2.0 or greater)






IBM PC/AT or Compatible, •
DOS, 512 KB RAM, math
coprocessor









IBM PC or compatible,
80386/80486, 4 MB RAM,
DOS 2.0 or higher, mouse
and. math coprocessor for
80386-based machines
recommended








Availability
Available from EPA
as EPA/600/R-
93/028
(EPAORD
Publications,
513-569-7562)
Price: FREE

Object PLUS
available from
Object PLUS Corp.
125 Cambridge
Park Dr.
Cambridge, MA
02140
Price: $100
(runtime version)
Environmental
Systems &
Technologies, Inc.
2608 Sheffield
Drive,
Blacksburg, VA
24060-8270
703-552,0685
Price: $400.00



Waterloo
-lydrogeologic
Software
19McCauleyDrive
(RR#2)
Bolton, Ontario,
Canada L7E SR8
905-880-2886
Price: $650.00




u>

-------
Table 16.    Summary of screening, ak flow, and compositional flow and transport codes that were evaluated (continued).

Model
CSUGAS










AIR3D




















VENT2D/
VENT3D











Type
Airflow










Airflow




















Airflow
and
mutticom-
ponent
contamin-
ant
transport






Functionality
Transient, two- or
three-dimensional air
flow








Three-dimensional
airflow



















Steady-state, two- or
three-dimensional air
flow and transient
contaminant
transport







Solution
Methodology
[finite-difference
solution of the air flow
equation


*




•
Finite-difference
solution of the air flow
equation posed in
terms of the
groundwaterftow
equation and solved
bytheMODFLOW
code













Finite-difference
solution of the air flow
equation, finite-
difference solution of
the transport equation








Assumptions
Based on Darcian
flow of an ideal
compressible gas in a
porous medium







Based on Darcian
flow of an ideal
compressible gas in a
porous medium

















Transport equation is
simplified by ignoring
mechanical
dispersion









Capabilities
Calculates vacuum
distribution in the
subsurface, in
inches of water







Calculates
pressure
distribution in the
subsurface

















Calculates
pressure
distribution in the
subsurface,
multicomponent
contaminant
constituent
concentrations
over time in the
subsurface



Advantages
Allows full, three-
dimensional
analysis of
heterogeneous.
multi-well air flow
problems

Text-based
input/output is
flexible and up to
the user
Easy-to-use 'CAD-
type' graphical user
interface which
simplifies model
setup and input

Allows three-
dimensional
analysis of complex
problems











Only readily
available
compositional flow
and transport code

Source code is
available

Text-based
input/output is
flexible and up to
the user

Limitations
Lack of easy-to-
use input/output
interface may
intimidate
beginners

No steady-state
solution option

No mass
transport
Users need to an
awareness of the
operation and
limitations of the
MODFLOWcode

No mass
transport













Grid size limited
to 25x25 eels
(can be increased
with a different
version available
from the author)






Hardware/Software
Requirements
IBM PC AT/XT or compatible,
640 KB RAM, DOS ZO or
higher








IBM PC or compatible, DOS
3.3 or higher, 4 MB RAM,
VGA card and color monitor,
mouse is highly
recommended
















IBM PC or compatible, 80X86
with math coprocessor, DOS
3.0 or higher, 525 KB RAM










Availability
Dr. James W.
Warner
Department of Civil
Engineering
Colorado State
University
Fort Collins, CO
80523
303-491-5048
Price: $125

American
Petroleum InsL
1220 L Street
Northwest
Washington, DC
20005
Price: $500.00

The original
version of AIR3D
(without the GUI) is
available .free of
charge from the
USGS:
USGS
Book and Open
File Reports
BLDG810, Box
25425
Denver, CO 80225
Price: FREE
David A. Benson
524 Claremont
Street
Reno, NV 98502
702-322-2104
Price:$495.00







-------
to plot all vapor sampling points on a map in an interactive environment, and then graphically
select any sampling point to query the database for detailed information on analytical results.

       A GIS combines tabular data and computer maps.  Table 17 presents examples of tabular
data. These data may be single-event measurements or time-variant, recurrent events. Thus each
measurement or parameter can be a function of space and time.  This can be written as:
                          P(x,y,z,t)
(99)
where: P is the measurement or inferred value; x is the horizontal ordinate, typically the easting
in map coordinates; y is the other horizontal ordinate, typically the northing; z is the sample
measurement depth, typically the depth below ground surface or relative to an elevation datum;
and t is the time of the observation.

       Many GIS systems are two-dimensional and use the concept of layers. This is a logical
outcome of the evolution of GIS, as many systems were originally developed for land use
purposes (Figure 37).  The layers are combined through the intersection process to construct a
composite analysis. For application to SVE, there is a need to use three-dimensional GIS or two-
dimensional GIS using layers to represent geological strata, soil types, or depth zones.  For
example, one might overlay permeability and vacuum distribution maps to assess the effect of
permeability variation on vacuum distribution in a sand layer.

       There are numerous GIS systems for SVE data and models. While many of these systems
are designed for UNIX workstations, an increasing number are available for PCs. On the UNIX
platform, the common systems include Arc/Info (Environmental Research Institute, Inc.) and
MGE (Intergraph). On the PC platform, commercial products include: Maplnfo (Maplnfo
Corporation), Atlas/GIS, and GRASS. Each system has unique features. Almost all systems
require some level of customization. GIS software selection may be based on several factors
including ability to integrate SVE data with other types of data and the availability of map files
(i.e., digital line files from USGS, street maps such as ArcData from ESRI, or street maps from
ADC or other vendors).

6.1.2  Linking GIS and Simulation Models                        •.>••••'•  •••••    ;
       Models can be integrated within the GIS network of data tables and maps, these models
can range from simple analytical solutions to complex numerical models.  Several example
applications of linking a GIS to an analytical solution for use as a screening tool are introduced
below.

       The GIS environment can be used to determine and compare capture zones of extraction
wells with areas of known contamination.  A GIS containing database information on vadose
zone properties (e.g., permeability and contaminant distributions) can be linked to an analytical
solution that calculates the SVE radius of influence for input well locations and extraction rates.

                                          135

-------
Table 17.     Example data types that can be utilized in a GIS.
Measurement or
Parameter
Permeability
Porosity
Dispersivity
Specific storage
Moisture content
Soil bulk density
Viscosity
Vapor density
Vapor pressure
Sorption coefficient
Partition coefficient
Soil organic carbon
Clay mineralogy
Soil concentration
Vapor concentration
Activity coefficient
Water table elevation
Molecular weight
Extraction rate
Extraction concentration
Vacuum pressure
Classification
Geologic
Geologic
Geologic
Geologic
Hydraulic
Geologic
Hydraulic
Hydraulic
Chemical
Chemical
Chemical
Geologic
Geologic
Chemical
Chemical
Chemical
Hydraulic
Chemical
Operational
Operational
Operational
Point or
Regional
Regional
Regional
Regional
Regional
Regional
Regional
N/A
N/A
N/A
N/A
N/A
Regional
Regional
Point
Point
N/A
Point
N/A
Point
Point
Point
Time-
dependent
No
No
No
No
Yes
No
Yes
Yes
Yes
No
No
No
No
Yes
Yes
No
Yes
No
Yes
Yes
Yes
Depth-
dependent
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
No
No
Yes
Yes
Yes
Yes
No
No
No
Yes
Yes
No
                                   136

-------
                    A computer system capable of holding and using data describing places on
                    the earth's surface.
               A number of related data layers can represent the many geographies of the real
Figure 37.     The concept of layering in a GIS. [Figure supplied courtesy of Environmental
              Systems Research Institute, Inc., Copyright © 1990, 1991, 1992 ESRI, Inc.,
              Redlands, California.]


                                            137

-------
Parameter values required by the solution would be retrieved from the database. This type of
analysis is especially useful for large sites with multiple source areas.

       A second example is to link chemical concentration and physical property data in the GIS
database with equilibrium partitioning equations to determine areas of suspected NAPL presence
based on a method described by Feenstra et al. (1991). The GIS can be used to calculate the
theoretical pore water concentration, Cw, and the effective solubility, Se, for all constituents by
accessing soil concentration and chemical property values in its database. The system can then
query these results and highlight areas on the site map where NAPL presence is suspected (where
 Cw > Sc). This process is illustrated on the flowchart in Figure 38. A similar example would be
to query the dissolved groundwater concentration database for dissolved concentrations of NAPL
chemicals in excess of a specified fraction of their aqueous solubilities. To do this, the dissolved
concentration database would be linked to the contaminant solubility database and a comparison
would be made for each chemical of concern. These areas also could then be highlighted on the
site map  as areas of suspected NAPL presence.

6.2    SCREENING MODELS

       Screening models are used to determine the potential feasibility of SVE as a remedial
option at a contaminated site based on limited, preliminary data.  It is not the purpose of the
screening process to produce a detailed SVE design, although preliminary conceptual design
plans can be developed from screening model results. The two screening models that have been
evaluated as part of this study are HyperVentilate and VENTING. A brief discussion on
analytical solutions is also presented.

6.2.1   HyperVentilate

       The HyperVentilate code (Johnson, 1991) is an educational and decision-support tool  that
operates  within a hypertext framework. It is available free of charge from the USEPA, although
the user must have a copy of the Object PLUS (formerly called Spinnaker PLUS) software to run
the code on a Windows-based machine (see Table 16 for software sources). The code also runs
on the Apple Macintosh. HyperVentilate consists of a several "stacks" of graphical reference
"cards."  Hypertext describes a method of organizing information where highlighted words in the
text can be "expanded" at any time to provide additional information about the highlighted term.
The highlighted term provides links to other documents that can be text, files, pictures, or
calculation sheets (Krol, 1992). Many cards contain key words that reference other information
cards, so that by clicking on key terms with the mouse, the user is transferred to another card  in
the stack that provides additional information on a topic.  The user can move through the card
stack either sequentially or by hypertext reference. The program operates in the Windows
environment within the Object PLUS program, and is very graphically oriented, with many
figures.
                                           138

-------
                                       Select All Soil
                                    Sampling Locations
                                    Above Water Table
                                     (graphically or by
                                  querying the database) I
                                   Calculate Theoretical
                                        Pore Water
                                     Concentration, Cw
                                   Chemical Property
                                       Database
                Next Soil
           Sampling Location
Calculate Effective
   Solubility, Se
Chemical Property
    Database
                                           Yes
                                            i
                                    Highlight Map Area to
                                      Show Possible
                                     Presence of NAPL
Figure 38.     Example application of a GIS to identify and highlight graphically those areas on
              a site where NAPL is likely to be located.
                                            139

-------
       Some of the hypertext cards contain informative tips and diagrams on SVE operation and
design, while other cards prompt the user for input data to be used in analytical calculations. A
number of analytical calculations are included in the code. These include calculations of
extraction well flow rates from permeability and well radius of influence data, permeability
based on time versus pressure drawdown data derived from an air pumping test, and contaminant
removal rates based on the equilibrium partitioning, mass-balance  approach. Generally, for any
calculation, a help card is available to the user that presents the analytical formula and its
underlying assumptions. The Hyperventilate card stack also contains information cards on other
remedial actions, such as thermal desorption, incineration, and soil washing, so that these
measures can be compared to SVE.

       Table 18 presents the general categories of input and output data required by the
HyperVentilate code in a format that highlights each of the code's major capabilities (outputs).
Hyperventilate is especially useful for making a rapid determination of the potential feasibility
of SVE based only on the permeability and thickness of the vadose zone. Soil intrinsic
permeability can be calculated from  saturated hydraulic conductivity values derived from slug or
pumping tests.  HyperVentilate includes a calculation card that converts saturated hydraulic
conductivity to intrinsic permeability. The calculated permeability is a maximum value because
it does not account for the presence of soil moisture or NAPL which will reduce the pore space
available for air flow. For flow rate  calculation, the user need only input a range of soil
permeabilities, the desired radius of influence, the screened interval for the extraction well, and
the extraction well radius in order to determine air flow rates for different SVE well vacuums.
SVE feasibility can be assessed on the basis of these flow rate calculations.  For example, if the
calculations indicate a flow rate of 0.1 to 1  scfm (these are low values) at a vacuum of 60 inches
of water (a reasonable value for extraction well vacuum), then the user should consider other
remedial alternatives.                                                         ,

       Based in part on contaminant composition, HyperVentilate will calculate estimated
maximum removal rates. The analytical mass transfer equations used by the code are based on
Raoult's law, and assume the presence of NAPL in the vadose zone. The program is intended to
be used mainly to screen sites contaminated with gasoline and other fuels, but can also be applied
to sites contaminated with other immiscible volatile solvents. It is not appropriate to apply
HyperVentilate at sites where NAPL is absent, however, because Henry's law, rather than
Raoult's law, describes volatilization of a contaminant from the dissolved phase.  Chemical
composition and property files for gasoline and weathered gasoline are included with the
program.  These files include mole fraction, vapor pressure, boiling point, and molecular weight
data for the major constituents of each gasoline type. The user can input other compositional
files for any contaminant or multicomponent mixture of interest.

       HyperVentilate can also be used to analyze the results of air pumping tests. Given time
versus pressure drawdown measurements at several radial distances from the pumping well, an
analytical solution is used to calculate soil permeability.  The analytical solution used is a form
of the Theis equation, which assumes confined, radial flow to a single pumping well.  This

                                           140

-------
    Table 18.  General categories of input and output data for the Hyperventilate screening
              model.
Input
Soil permeability
Extraction well radius of influence
Screened interval
Well radius
Contaminant composition
(accepts multicomponent mixtures)
Contaminant mass in the subsurface
Desired remediation time
Minimum number of wells
Time versus pressure drawdown
(air pumping test data)
Soil permeability
Radius of influence
Well radius
Screened interval
Radius of free-phase NAPL layer
Well radius
Radius of contaminated zone
Contaminant concentration
Contaminant molecular weight
Contaminant vapor pressure
Minimum number of wells
Output
Extraction well flow rate
Maximum estimated contaminant
concentration removal rate
Desired removal rate versus maximum
estimated removal rate
Soil permeability ,
Relative efficiency of removal based on
diffusion from the free-phase NAPL layer
Removal rate of contaminant which must
pass through a low-permeability or
contaminant-free zone before reaching
the extraction Well
solution is not valid if there is significant leakage of air from the ground surface during the
pneumatic test.

       Two boundary-layer screening calculations are included that allow the user to calculate
theoretical removal rates of contaminants: (1) above a liquid layer of free-phase NAPL; and (2)
above a low-permeability or contaminant-free zone overlying a zone where residual
contamination is present.  These calculations permit the user to explore the effects of diffusion-
limited mass transfer and subsurface heterogeneities on SVE performance.

6.2.2   VENTING                 .

       VENTING (ES&T, 1994) is a mass transport screening code. It uses a DOS-based text
menu interface for data input and provides graphical output of the results in the form of mass
versus time plots for each constituent of a multicomponent mixture (see Figure 12). The output
                                          141

-------
is also stored in ASCII text format that can be input to other software for custom plotting of
results. VENTING calculates contaminant mass in the subsurface (averaged over the entire
specified contaminated zone) over time during extraction from a single well or a multiwell
system based on a user-defined SVE flow rate. The flow rate can be measured in the field,
determined using other models, or estimated byVENTING from permeability and pressure
gradient data.  Table 19 presents the primary input requirements for the VENTING code and
highlights the major output features of the code.

       If the SVE flow rate is unknown, VENTING can calculate a flow rate from an estimate of
air permeability and vacuum measurements made at the extraction well and an observation well.
If observation well data are unavailable, the user can assume an extraction well radius of
influence (typically 30 to 100 feet) and set the pressure at this point equal to one atmosphere to
calculate the extraction well flow rate.  The calculated flow rate is not highly sensitive to the
assumed radius of influence (Johnson et al., 1990a,b).  In addition, permeability can be estimated
within the code from distance versus pressure data obtained during a pneumatic test using the
analytical solution for steady-state, confined, radial flow to a single well.
 Table 19.     General classes of input and output data for the VENTING screening model.
                     Input
                 Output
  Extraction well flow rate
  Contaminant composition
  (accepts multicomponent mixtures)
  Contaminant mass in the subsurface
  Desired venting period
  Contaminated soil volume
  Soil physical properties
  Estimated venting efficiency factor (see below)
  Biodegradation rate (optional)	
Contaminant mass versus time
  Mass remaining in the soil
  Gas phase concentration in the well bore
  Equilibrium gas phase concentration
  Contaminant concentration in soil
  Soil physical properties
  Screened interval
  Extraction well pressure
  Well radius
  Extraction well radius of influence
Efficiency factor
  Air flow rate
  Vadose zone thickness
  Well radius
  Extraction well pressure
  Observation well pressure
  Radial distance to observation well1
Soil permeability
       1lf the radius of influence is selected, let the pressure at the observation well equal one
        atmosphere.
                                            142

-------
       The mass transport equation is based on an idealized, one-dimensional, mass-balance
formulation whereby during each time step of the simulation, the mass which is extracted at the
vacuum well is iteratively removed from the total mass in the system. The mass transport
equation assumes equilibrium partitioning between gas, soil water, sorbed contaminant, and
NAPL. This formulation does not explicitly account for heterogeneities or rate-limiting
processes in the system, but does incorporate a user-defined value for extraction efficiency that
can be set to less than 100% to represent subsurface conditions that hinder mass removal. The '
mass removal of each constituent in a multicomponent mixture is calculated as a function of time
(Figure 12). VENTING also allows the user to input a first-order degradation rate constant to
simulate biodegradation effects.

6.2.3  Analytical Solutions

       Analytical solutions, such as those listed in Table 14, can be solved to evaluate many
aspects of S VE. Two tools commonly used to solve analytical solutions by computer are
mathematical analysis and spreadsheet software. Mathematical analysis packages'such as
Mathcad (Mathsoft) and Macsyma (Symbolics) can be used, for example, to solve integral
equations that are applicable to SVE. However, analytical solutions for many of the commonly
occurring integrals that describe SVE have already been developed. Examples of analytical
solution programs include AQTESOLV (Geraghty and Miller, 1989), which contains the Theis
(1935) and Hantush-Jacob (1955) equations, and the CRTC (1992) spreadsheet program for the
Shan et al. (1992) solution.  Both of these applications are explored further in Section 7.  A suite
of analytical solutions for analysis of air pumping tests that consider air as a compressible fluid
are presented by Falta (1993).

6.3     AIR FLOW MODELS

       Air flow models are utilized for detailed SVE design because they can simulate the
effects of multiple extraction wells in heterogeneous media with various boundary conditions.
The model domain is typically represented in two or three dimensions. The two-dimensional
model geometries include plan view, cross-sectional, and radial axisymmetric.  Air flow models
are frequently relied upon to test hypotheses concerning parameters such as well placement,
screen interval,  and permeability distribution by performing sensitivity analysis. Given a
reasonably accurate characterization of the vadose zone, an air flow model can be utilized to test
alternative SVE designs. Simulation analysis thereby provides insight to pressure distributions,
air flow patterns and rates, and well capture zones associated with different SVE  designs. This
type of analysis can be particularly helpful for examining how SVE performance is affected by
heterogeneities, preferential flow paths, layering, and variable boundary conditions.

       Numerical simulators generally are more robust than analytical models due to the
incorporation of fewer restrictive assumptions. Analytical flow models, however, do not require
discretization of the model domain (Section 4.4). Three numerical air flow codes have been
evaluated as part of this study: AIRFLOW, CSUGAS, and AIR3D.

                                          143

-------
6.3.1  AIRFLOW

       AIRFLOW (Waterloo Hydrogeologic Software, 1993) is a finite-element code that allows
radial-symmetric analysis of steady-state air flow to a single extraction well (Figure 39). It
calculates the steady-state pressure distribution, air flow velocities, and air flow pathlines in
cross section.  The input and output functions of AIRFLOW operate within a DOS-based
Graphical User Interface (GUI) that enables the user to develop model grids and assign grid
block properties in a CAD environment.  These functions allow rapid development of simple
models for hypothesis testing. The code is ideal for determining radius of influence and
performing sensitivity analyses by varying the screened interval and permeability distribution in
the model. Simulation results are available quickly and graphically via the GUI.

       AIRFLOW solves the governing equation for compressible flow of an ideal gas in porous
media (see, for example, Huyakorn and Finder,  1983) in a radial-symmetric  domain by the
method of Galerkin  finite elements. The model domain is divided into a rectangular grid that is
subdivided into triangles. Using these triangular elements with linear shape functions,
AIRFLOW solves for the pressure at the vertex of each triangle. The finite-element solution
becomes more accurate for finer grids, and approaches the exact solution as the grid becomes
infinitely fine. Additional  discussion of the finite-element method and its application can be
found in Finder and Grey (1977) and Huyakorn and Finder (1983), among others. The model
also solves the two-dimensional equation of air flow pathlines from computed air velocities for
particles released at user-specified locations in the model domain.

       The boundary conditions that AIRFLOW allows are constant-pressure and no-flow. An
extraction well is defined by one or more constant-pressure nodes.  The model does not allow
specification of constant flux nodes (for example, an extraction well with a known flow rate), nor
does it provide any air mass balance information. According to the code authors,  future revisions
of the model will include the capability to determine air flow rates from an extraction well. The
reader is referred to  Section 8.1 for an example application where it is necessary to know both
the pressure and air flow rate at the extraction well in order to fit simulation results to field data.

       The finite-element grid, boundary conditions, and subsurface permeability distribution
are developed interactively using the CAD interface. Grid lines can be added, moved, or erased
interactively, and boundary conditions are set in the same manner, using a point-and-click
method to define constant pressure or no-flow at selected nodes. The default boundary condition
at all exterior nodes  is a no-flow condition. Extraction well nodes are set to the extraction well
pressure at each node within the screened interval of the well. Table 20 summarizes the input
data requirements and output features of AIRFLOW.

       AIRFLOW allows definition of up to 255 different permeability zones, although porosity
remains constant throughout the model domain. Permeability zones are defined interactively by
graphically selecting a window within the model domain and defining a permeability value
                                          144

-------
                VAOOSE
                ZONE -
             SATURATED
             ZONE
                                                                     MODEL
                                                                     SECTION
                                  FINITE-ELEMENT GRID
Figure 39.     Radial-symmetric geometry and grid for the AIRFLOW model.


                                          145

-------
      Table 20.      General classes of input and output data for the AIRFLOW model.
                           Input
            Output
        Finite-element grid
        Boundary conditions
        Extraction well location and pressure
        Permeability distribution
        Vapor properties
              Temperature
              Gas molar mass
              Viscosity
              Soil porosity
        Particle release locations for pathlines
Pressure distribution
Air flow pathlines
Air flow velocity vectors
associated with the highlighted area. Likewise, pathline starting points for particle tracking are
defined by selecting particle locations interactively.

       AIRFLOW output consists of pressure contours plotted across the model domain, air flow
pathlines, and air flow velocity vectors.  Pressure contours indicate the pressure distribution in
the subsurface, and can be used to define the radius of influence (based on a specified vacuum
criterion) and determine the pressure gradient throughout the model domain. The contouring
package included with AIRFLOW does not automatically label the contour lines, so some users
may wish to label these by hand, or use another contouring package. The pressures at all r,z
coordinates can be written to an ASCII text file and input to any contouring package.
AIRFLOW can also generate a plot of the finite-element grid.

       Air flow pathlines track the paths of theoretical particles released into the subsurface and
subject to the pressure distribution induced by SVE. Pathlines can be generated in either a
forward or reverse direction.  Reverse pathlines are useful because particles placed along the
screened interval of the extraction well will trace backwards into the areas that are swept by the
induced air flow. Forward tracking of particles can be useful, for example, to determine if the
induced flow velocity from an area of known contamination toward the extraction well is
sufficient. If particles released from a known area of contamination do not reach the extraction
well with adequate speed, this indicates a need for modification of the screened interval or
extraction rate (if viable).

       Velocity vectors are plotted as arrows on the model grid to indicate the relative
magnitude and direction of the air flow velocity at the centroid of each finite element triangle.
By studying the velocity vectors, the user can identify air stagnation zones in the subsurface.  For
the single well system simulated by AIRFLOW, air stagnation occurs at distance from the
                                            146

-------
extraction well and in isolated low-permeability zones that are not effectively swept by the
induced air flow.

       Pressure contours, pathlines, and velocity vectors can be output in several different
formats, including screen output, printer or plotter, DXF, HPGL, PostScript, and PC Arc/Info.
The pressure distribution can also be output in ASCII text for use in any contouring package. A
log file can be produced which echoes information about the finite-element grid (including node
and eleme.nt numbering), boundary conditions, and material properties.

       The major limitation of AIRFLOW is that its application is restricted to single well,
axisymmetric problems. As such, it has limited capability to evaluate heterogeneous subsurface
conditions and multiwell SVE systems.  As noted in Section 5.5.4, a new version of AIRFLOW,
entitled AIRFLOW/S VE, includes the capability to simulate multicomponent contaminant
transport and other enhancements.

6.3.2   CSUGAS

       The CSUGAS (Sabadell et al., 1988) finite-difference model can be used for transient
two- or three-dimensional simulations of air flow in heterogeneous porous media.  Although the
model does not explicitly simulate steady-state conditions, it would be straightforward to use it
in this fashion by running a transient simulation until the change in pressure distribution between
each time step is sufficiently small.  It calculates air pressures and flow rates in a two-
dimensional plan or cross-sectional domain, or a three-dimensional domain. The three-
dimensional geometry of the code allows a great deal of flexibility in simulating many different
field scenarios.

       The input functions of CSUGAS operate within a DOS-based, menu-driven text interface
where the user is prompted for each input item via a menu selection. The text interface allows
input of permeability, porosity, interior constant-pressure or constant-flux (extraction well) node,
and exterior boundary condition matrices in a spreadsheet format that is automatically displayed
by the menu system.  The user can also create input files by manually editing the input text file.
Table 21 presents the general categories of input data and output features of the CSUGAS model.

       CSUGAS solves the governing equation for compressible flow of an ideal gas in porous
media using a block-centered, finite-difference method.  The model is based on the Darcy flow
equation, the principle of continuity, and the ideal gas law for developing the compressible gas
flow equation (as described in Section 3). The user specifies the numerical solution iteration
parameters.  Numerical oscillations that occur during solution of some problems can often be
mitigated by varying  these parameters. The model domain is  divided into a rectangular grid
defined by rows, columns, and layers.  The minimum number of layers allowed is three, even for
areal simulations.  For a cross-sectional simulation, the user would set the number of rows to
one.
                                          147

-------
       The output from CSUGAS consists of pressure and change in pressure data at each finite-
difference block, and the results of mass balance calculations for air flow in and out of the entire
model domain, but not at individual wells. Thus, for any simulation with more than one well, it
is not possible to determine extraction rates for vacuum wells without making additional
calculations based on simulated pressures and media properties.  CSUGAS includes a simple
contouring program with which to plot the simulated pressure distribution, and also includes an
option to output the data in SURFER format. SURFER is a commonly used contouring program.
A separate output file is created for each model layer.
 Table 21.     General classes of input data and output features for the CSUGAS air flow model.
                              Input
                    Finite-difference grid ,,

                    Soil permeability matrix

                    Soil porosity matrix

                    Boundary conditions

                    Extraction well locations
                    and pressures

                    Initial pressures in the
                    model domain

                    Total time of analysis and
                    time step

                    Iteration parameters
        Output
Pressure distribution

Change in pressure at
each time step

Change in pressure over
total time

Air mass balance
 6.3.3  AIR3D

       AIR3D uses the finite-difference method to simulate steady-state or transient air flow in
 two- or three-dimensions and calculate air pressures in the vadose zone induced by extraction
 wells or trenches.  The model operates within a DOS-based GUI that enables the user to develop
 model grids, assign grid block properties, and locate extraction wells and trenches in a CAD
 environment (Baehr et al., 1993). Based on the ModelCad386 preprocessor developed by
 Rumbaugh (1993) for the MODFLOW groundwater flow code (McDonald and Harbaugh, 1984),
 the GUI facilitates rapid development of models for hypothesis testing. Simulation results are
                                           148

-------
processed easily with contouring and velocity vector plotting routines that operate within the
GUI program.

       The block-centered, finite-difference groundwater flow code MODFLOW is used by
AIR3D to solve the equation of air flow in the vadose zone. The air flow equation is cast in the
same form as the groundwater flow equation and solved by MODFLOW are described in Section
3.2." AIR3D automatically converts the input data in terms of pressures into the correct form
required by the MODFLOW solver. The MODFLOW solution of groundwater heads is then
converted back into pressure values which are output by AIR3D. As with CSUGAS, the
iteration parameters can be modified by the user to optimize the numerical solution:

       The finite-difference grid, boundary conditions, and subsurface .permeability distribution
are developed interactively by pointing-and-clicking using the CAD interface. Table 22 lists the
input data requirements and general output features of AIR3D. Basemap files in DXF format can
be imported to the GUI to use as an overlay for model development and output display.

       AIR3D allows designation of constant pressure and no-flow boundary conditions.
Extraction wells or trenches are defined as constant-pressure nodes within the model domain.
Cells representing extraction wells or trenches are set to the extraction well pressure in each cell
within the screened interval of the well or trench. The bottom of the lower layer in the model is
set to no-flow by default.  AIR3D adds an extra model layer above the top of the model that is set
to atmospheric pressure. The remaining exterior cells are set to atmospheric pressure by default.
The default specification of the exterior boundary of the model as constant pressure means that
all recharge of air to the model must ultimately come from the atmosphere, either at the
groundsurface or through the surrounding vadose zone. The ratio of air recharge from ground
surface versus from lateral boundaries reflects the significance of surface leakage to SVE  wells
and impacts the extent of well capture zones.
   Table 22.  General classes of input data and output features for the AIR3D air flow model.
                       Input
              Output
     Finite-difference grid

     Boundary conditions

     Extraction well or trench locations and pressures

     Soil permeability/porosity parameter zones

     Layer tops and bottoms

     Iteration parameters  	'
Pressure distribution

Velocity vectors

Mass balance in any selected region of
the model
                                          149

-------
       Flow rates to wells or trenches can be determined by using a mass balance option which
 allows the user to select an area in which to perform a mass-balance calculation. In order to
 obtain a specific extraction rate at a well or trench, an iterative approach must be employed
 whereby the pressure at the well is adjusted until the desired flow rate at the well is achieved. An
 alternate iterative approach can be used to match simulation results to field results from an air
 pumping test. Given the pressure and flow rate at the well, and pressure distribution data from
 observation wells, the model permeability can be adjusted until the simulated pressure
 distribution matches the field results.

       The AIR3D GUI uses the concept of zones to input values of permeability, porosity, and
 elevation for each cell in the model domain.  Parameter distributions are defined interactively by
 graphically selecting a window within the model domain containing one or more cells and
 associating a zone number with the highlighted area. For each parameter, each zone is assigned a
 number and a corresponding value. For example, porosity zone one may be assigned a value of
 0.4. Each parameter has its own distribution of zones.  A cell may be assigned, for instance,
 permeability zone two, porosity zone three, and bottom elevation zone one.  The'zone numbers
 do not relate to layer numbers because they are purely arbitrary and set by the user. Data for an
 parameter zone can be imported from a SURFER file (e.g., so that a layer bottom elevation'can  :
 be converted into a matrix of bottom  elevations using bilinear interpolation).' Database files can
 also be imported to define property zones.

       AIR3D output consists of pressure contours and air flow velocity vectors. These results
 can be output in several different formats including screen output, printer or plotter (see. Figure
 40), DXF, HPGL, and PostScript.  Velocity vectors are plotted as arrows on the model grid that
 indicate the relative magnitude and direction of the air flow velocity at the center of each finite-
 difference block.

       The AIR3D code includes an optimization module that helps select optimal well locations
 and extraction rates without using the trial-and-error approach of running the model numerous
times to determine preferred well locations and extraction rates. The optimization process poses
the design challenge in terms of a linear programming problem, which is then minimized to
determine the optimal solution. The linear programming problem is subject on constraints
supplied by the user. The constraints  should be defined such that the S VE system induces the
desired radius of influence and air flow pattern. AIR3D is runpnce for each potential well
location supplied by the user to determine a unit response (pressure  drawdown) at each well.  The
solution methodology assumes that the unit response can be scaled linearly by a simple
multiplication.  The linear problem is posed in terms of determining what combination of scaled
unit well responses will meet the extraction rate and pressure gradient constraints. Additional
discussion of the linear optimization problem can be found in GeoTrans (1992).
                                           150

-------
                    1.
                      10
                             .03
                                      DC
                                                               )C
                                                                      n
                                                                         f^
                                               ft
                                                        ss
                           tif
•?JT
                 .OD	1 OC	1.00	.1.0
                                                          .CO
Figure 40.    Example output (pressure contours) from the AIR3D code.




                                             151

-------
6.4    COMPOSITIONAL FLOW AND TRANSPORT MODELS

       As described in Section 5.5, compositional flow and transport models can be used to
simulate SVE-induced air flow and vapor-phase multicomponent transport in the vadose zone.
The only model readily available for review during this study was the VENT2D (Benson, 1994)
code which is described below. A three-dimensional extension of this code called VENT3D is
now available. An additional feature of VENT3D, beyond the increased dimensionality, is that it
includes a solution algorithm developed by Hills et al. (1994) that uses higher-order corrections
to counteract numerical dispersion, resulting in an accurate solution even with coarse grids. In
addition, as noted above, AIRFLOW/SVE is a recent enhancement of AIRFLOW that can
simulate air flow and multicomponent transport in an axisymmetric domain.

6.4.1   VENT2D

       VENT2D simulates steady-state air flow and transient vapor-phase transport of
multicomponent mixtures. Any number of vapor extraction or injection wells can be simulated,
and grid-variable permeability, and initial contaminant distribution can be specified.  An upper
confining layer can be included to simulate the effects of surface boundary condition effects.
The equilibrium partitioning distribution of contaminant in NAPL, adsorbed, dissolved, and
vapor phases is calculated, as are the effects of contaminant retardation.  The transport of soil
moisture is simulated, and permeability is a time- and space-dependant function of soil moisture
and NAPL saturation (actually air-filled porosity based on Equation 63).

       With the VENT2D model, Benson et al. (1993) take a convenient approach to the
solution of the advective-dispersive equation by neglecting mechanical dispersion in favor of
diffusion in the formulation of the dispersion tensor within the advective-dispersive equation (see
Sections 3.2.4, 4.2.1, and 4.2.2). This allows a simpler numerical formulation to solve the
transport equation, thus enabling transport modeling of multicomponent mixtures with
retardation. Equilibrium phase partitioning is assumed and the presence of NAPL is indicated
when all non-NAPL phases are saturated. Sorption to soil solids of individual constituents is
calculated based on octanol/water partition coefficients and the soil fraction of organic carbon.

       The model solves the steady-state equation of air flow based on the principle of
conservation of mass. The concept of relative permeability (Section 4.1.5) in which effective
permeability varies as a function of changing air-filled porosity is incorporated in VENT2D. The
other air flow codes reviewed herein assume a constant moisture content, and hence a constant
permeability over time. Permeability increases with decreasing soil moisture and NAPL
saturation due to drying induced by S VE. In order to save computational time, the permeability
distribution is only recalculated when 10% or more of the active cells' permeability values have
changed by 25% or more based on Equation (63) (the equation given on the right side).

       VENT2D operates in the DOS environment and does not provide any input or output
interface or menu.  The general categories of input data and output features are summarized in

                                          152

-------
Table 23. The code author suggests that an effective approach to developing a simulation is to
modify one of the example input files provided to suit the user's needs. A utility program is
included that allows conversion of SURFER grid files to ASCII .text so that site concentration
contour maps can be used as initial concentration input to VENT2D. The code includes several
composition files, including fresh and weathered gasoline, and common solvents.

       The outer boundary of the model domain is set by default as follows: no-flow at the,
bottom of the domain (water table); and constant, atmospheric pressure and zero concentration
elsewhere around the model perimeter. A surface confining layer is allowed, whereby the user
can control vertical leakage by varying the vertical permeability of the confining layer (Figure
41).  Extraction wells are represented by constant-flow cells. The pressure at an extraction  well
can be determined based on the pressure distribution output from the model. An additional
feature of VENT2D allows placement of zero-concentration nodes at the model surface to
simulate contaminant diffusion to the atmosphere.

       VENT2D produces a wide range of output data, including distributions for pressure,
permeability, air flow specific discharge (analogous to air flow velocity), soil moisture content,
total VOC  and individual component concentrations in soil and soil gas,  and NAPL.  Output data
from VENT2D are summarized in Table 23. A utility program included  with the software will
convert any of the parameter maps in the output file to a SURFER file format for contouring.
    Table 23.  General classes of input data and output features of the VENT2D compositional
              flow and transport model.
                         Input
     Output (at each time step)
      Finite-difference grid
      Soil porosity
      Soil moisture content
      Soil permeability distribution
      Soil fraction of organic carbon
      Relative humidity of atmospheric air
      Relative humidity of injected air
      Well locations and extraction/injection rates
      Thickness of leaky confining layer
      Vertical permeability of leaky surface layer
      Contaminant composition
      Contaminant physical properties
      Zero concentration nodes
         (allow diffusion at the ground surface)
      Initial contaminant distribution
      Total time of analysis and time step
      Iteration parameters
Pressure distribution
Relative permeability distribution
Air flow specific discharge        <
  (can be used to calculate velocities)
Soil moisture content distribution
Total VOC concentrations in soil
Total VOC concentrations in soil gas
NAPL distribution
Concentrations of selected constituents,
  in soil vapor
Concentrations of all VOCs in vapor
  at extraction well nodes
Contaminant mass balance
Residual contaminant mass
Extracted contaminant mass
                                            153

-------
VALUES COMMON TO  ENTIRE MODEL DOMAIN:

  INITIAL CONTAMINANT COMPOSITION
  FRACTION ORGANIC  CARBON
  INITIAL MOISTURE CONTENT
  OVERBURDEN DEPTH (z)
  COLUMN WIDTH  (Ax)
  ROW WIDTH (Ay)
  THICKNESS  (b)
  BULK DENSITY
  POROSITY
                                                        ACTIVE TRANSPORT AREA
                   ,vz




























                 i  - 1  2 345
                                           n
                                                  VALUES  UNIQUE TO EACH
                                                 NODAL VOLUME:
                                               • PERMEABILITY
                                              •CONTAMINANT CONCENTRATION
                                             • WATER CONTENT AT TIME > 0
                                             • CONTAMINANT COMPOSITION AT TIME >  0
                                            • WELL EXTRACTION  (OR INJECTION) RATE
                                           •CONFINING LAYER VERTICAL PERMEABILITY
                       BOUNDARY  CONDITIONS:
                       IN THE  NODES i=t. j=1.  i=n. i=
                       PRESSURE  = 1 ATMOSPHERE
                       CONCENTRATION »  0
Figure 41.     Geometry and boundary configuration for the VENT2D model (from Benson,
              1994).
                                            154

-------
                                     SECTION 7
                              EXAMPLE APPLICATIONS

       Several examples are presented below that highlight the application of models to the
design and evaluation of SVE systems. They are intended to provide the reader with a basic,
albeit incomplete, overview of the types of analyses that can be conducted using the SVE models
reviewed in this document.  The purpose of this section is not to compare the models to each
other, but rather to illustrate their capabilities.

7.1     PREDICTING FLOW RATES FROM PERMEABILITY WITH HYPERVENTILATE

       The goal of this first example is to apply Hyperventilate to estimate SVE flow rates that
might be achieved at a single well as a function of wellhead vacuum for specified values of
permeability, well radius^ interval thickness, and estimated radius of influence. Given a saturated
hydraulic conductivity (K) of 2.8xlO'3 cm/sec (7.9 ft/day) for a fine sand layer that extends above
the water table, Hyperventilate uses Equation (54) to calculate intrinsic permeability (k), After
accounting for fluid properties, k is calculated as 2.9x10'8 cm2. Hyperventilate requires that
permeability values be in darcys, and thus the value calculated above is converted to 2.9 darcys
(1 darcy = 9.87x10'9 cm2).

       The flow rate estimation card in Hyperventilate is card eight in the soil venting stack.
Assuming that the well radius is two inches, the screened interval is 6.6 feet, and the estimated
radius of influence is 40 feet, then the relationship between extraction well vacuum and flow rate
is determined for a range of permeabilities (1 to 10 darcys) typical of fine sand to account for
uncertainty in the permeability estimate.

       SVE well flow rates calculated for different well vacuums (5 to 200 inches of water) and
the permeability range of 1 to 10 darcys are presented in Figure 42. Wellhead vacuums for SVE
typically range from 5 to 100 inches of water, with higher vacuums applied in less permeable
media. As shown, Hyperventilate indicates that for a typical well vacuum of 60 inches of water,
extraction well flow rates can be expected to be in the range of 3.7 to 37 standard cubic feet per
minute (scfm). This compares with typical SVE well flow rates of 10 to 100 scfm. The
calculated range of flow rates for the fine sand suggest that it will be amenable to SVE operation.

       It is important to consider the underlying assumptions when interpreting these
calculations. The analytical solution upon which the results are based assumes confined flow.
Surface leakage, however, will probably effect a higher flow rate in a smaller zone of influence.
The value for permeability is a best-case estimate because it does not take into account pore
space occupied by soil moisture or NAPL (relative permeability effects).  However, these
calculations do provide a rapid initial estimate of SVE parameters and feasibility.
                                          155

-------
               Flowrate Estimation:

                 O
                 o
Medium Sand
Fine Sand
Siftj> Sand
1) Choose Soil Type, or
  Optional - Enter your own permeability values (darcy)
2)EnterWellBadius(in)
3) Enter Kadius of Influence (ft) & IntervalThickness'
4) Optional- Enter your own well vacuum (406" = maw)
5) Cliclt button to calculate Predicted Flowrate Banges
                                                                 Predicted Flonrate Eanges
O Input Your 0 wn Permeability Range
Permeability Range (darcy)
i I to | 10 |
Well Radios | 2 lin
Radius of Influence [ 40 Ift
Interval Thickness* 1 6.6 Ift

( --^Calculate Flonrate Ranges^- }
• thiclnoss of screened interval, or JJ 1
pum«ble :on« (whichever if smaller 1. LI

Well Flowrate
Vacuum (SCFM)
Pv, (single well)
(inB^O)
	 5
10
20
40
GO
120
200 1]
0.33
0.66
	 .yo 	
2.54
	 3.71
683
1007
to
to
to
to
to
tn
to

	 3.32
	 6.59
13.02
	 .25.38. . ..
37.09
6827
100.66



Figure 42.     Card eight from the Hyperventilate soil venting stack is used to calculate a range
                of well flow rates based on permeability, user-defined radius of influence, well
                radius, and screened interval or flow zone thickness.
                                                   156

-------
7.2    USING AIRFLOW TO ASSESS SVE WELL SCREEN INTERVAL

       AIRFLOW can be used to evaluate the extraction efficiency of alternative well screen
completion intervals in the vadose zone. For this example, we assume a vadose zone that is three
meters thick, a single extraction well with a screen length of 0.9 meters, and a homogeneous sand
with an effective permeability of 50 darcys. The SVE well vacuum is set to 107 inches of water.

       The geometry and boundary conditions of the modeled system are presented in Figure 43.
The geometry is radially symmetric, with the extraction well (constant pressure) at the left
boundary (axisymmetric center) of the model domain. The other boundary conditions are no-
flow on the bottom (water table), atmospheric pressure on the  domain right side (axisymmetric
perimeter) to simulate influx of air at distance from the extraction well, and atmospheric pressure
on the top of the model to simulate a ground surface open to the atmosphere.

       Figures 44a to 44c present results of air flow particle tracking for three different screened
intervals. In Figure 44a, the well is  screened in the middle of the vadose zone and does not
effectively draw air from the bottom of the model domain. By lowering the screened interval
somewhat (Figure 44b), more of the lower modeled area is swept by the air pathlines. The most
effective well screen placement is just above the water table as shown in Figure 44c.  In this case
the majority of the vadose zone is swept fairly effectively. Therefore, in the case of a
homogeneous vadose zone and depending on contaminant distribution, it may be advisable to
place the extraction well screen as close as possible to the water table. If the screen is set too
low, or is too short, however, a rising water table caused by seasonal variations or upwelling
induced by SVE may flood the screen interval and impair system performance. Neither of these
processes is considered in the model.

7.3    ESTIMATING CONTAMINANT REMOVAL RATES USING VENT2D

       This  simulation example is based on a sample data set included with VENT2D. The data
set describes the following scenario. A volume of soil 12 x 10 x 2 meters thick is contaminated
with relatively fresh gasoline at variable levels. The composition of the gasoline is approximated
by a 37-component mixture. Initial concentrations of total VOCs in soil range as high as 25,000
ppm.  The model domain has an intrinsic permeability of 50 darcys and is overlain by a five-
meter thick layer with a vertical permeability of 50 darcys. A single SVE well in the middle of
the model domain withdraws 25 scfm. This example is illustrative of a small UST spill site with
uniform sandy soil.

       The simulated total VOC concentration and removal rate in the extracted soil  gas during
the first 14 days of SVE operation are shown in Figure 45. Values for total VOCs in  soil gas at
each time step were taken from the model output file using a text editor and imported into a
spreadsheet for graphing. The total VOC removal rate is calculated  by multiplying the soil-gas
concentration in the well bore by the extraction well flow rate.
                                          157

-------
         3.CI-
          2.4'
     i/i
     u
          1.8-
          1.2
          0.6-
          0.0
              //
                7
                7
                7
                7
                7
                7
                7
                7
                7
                7
                7
                7
                7
                7
                7
                7
                7
                7
              //
            o.o
                            Simulation Domain and Boundary  Conditions


                                           P = 1 atm
                           1.0
                                         2.0
                                                        3.0
4,0
                                                                                     5.0
                                                 no-flow Boundary



                                            DISTANCE, meters
Figure 43.     Model domain geometry and boundary conditions input to AIRFLOW for the

              example described in Section 7.2.
                                            158

-------
               3.0
               Z.4-
               1.8-
               1.2-
               0.0
               3.0
                            1.0
                                        2.0
                                                   3.0
                                                               4.0
                                                                           5.0
                                                                           5.0
Figure 44.    AIRFLOW results showing how air flow and flushing of the lower vadose zone
             (where LNAPL tends to accumulate within and above the capillary fringe)
             increases with depth of screen placement: (a) uppermost screen placement, (b)
             intermediate placement, and (c) placement just above the water table.
                                          159

-------
       1200
       1000
        800
        600
900


800


roo


600


500


400


300


200


100
I
t
r
                                   6       8       10

                                         TIME, days
                                                            12
                                                                    14
                                                                            16
Figure 45.     VENT2D results showing simulated VOC concentration and removal rate in
              extracted soil gas during the first 14 days of SVE operation for the example
              described in Section 7.3.
                                           160

-------
       As shown in Figure 45, the VOC removal rate declines steeply during the first few days
of SVE operation.  A pattern of exponential decay of VOC concentration is simulated, in part,
because the modeled system is homogeneous. If the model domain were heterogeneous, the
concentration curve would tend to exhibit tailing at higher concentrations.

       Examination of the simulated total VOCs in soil with time is also insightful.  Figure 46
shows total VOC concentration in soil at the extraction well grid cell.  Although the simulated
VOC concentration in soil decreases from 25,000 mg/kg (or ppm) to just over 10,000 mg/kg
during the first 20 days of operation, it remains over 5,000 mg/kg after 50 days, and declines to
just below 1,000 mg/kg after 200 days of operation. Model output indicates that, as expected, an
increasing fraction of the total VOCs in soil are comprised of the less volatile components of
gasoline. These results indicate that although the more volatile constituents are removed from
the system fairly rapidly, it takes much longer to remove the less volatile constituents.

       Subsequent model runs indicate that even after an additional 100 days of SVE operation,
the total VOC concentration in soil remains just above 100 mg/kg, which is a typical regulatory
limit for total petroleum hydrocarbons. From a risk-based or groundwater impact potential
perspective, however, removal of the more volatile fraction (which contains the higher risk
components such as benzene) is more important than removing relatively immobile components
of a petroleum product mixture.

       Analyses such as these are useful to gain insight into time frames associated with SVE
remediation. While the more volatile gasoline constituents are removed fairly rapidly in the
example simulation, further  analysis of total VOC concentrations in soil reveals the presence of
residual contamination by less volatile components. Real-world heterogeneities tend to  diminish
the effectiveness of SVE.  These results suggest a need for monitoring of soil concentrations as
well as soil-gas  concentrations during and after remediation.  For less volatile constituents that
resist SVE, other remedial actions such as bioventing may be appropriate.

7.4    ASSESSING SVE IMPACT ON NAPL DISTRIBUTION USING VENT2D

       The VENT2D model can be used to examine the changing distribution of soil
contamination and NAPL presence over time during SVE. NAPL presence is inferred by
VENT2D when (and where) all other phases (dissolved, adsorbed, and vapor) are saturated in the
model. This example is based on the simulation described in the prior section.  Figure 47 shows
the distribution of total VOCs  and the extent of NAPL in soil prior to SVE and after 42 days of
SVE operation.  The plots in Figure 47 were produced using the SURFER contouring package,
and VENT2D's utility program to convert matrix data to SURFER grid file format. Initial VOC
concentrations range as high as 25,000 mg/kg and NAPL is present over a large portion  of the
model domain.  SVE is shown to be effective at reducing both the, simulated total VOC
concentrations in soil and the extent of NAPL presence (Figure 47). However, total VOC levels
in soil are still relatively high in the vicinity of the extraction well (the center of the model
                                          161

-------
                                  TOTAL VOCs IN SOIL
   30000
   25000
   20000
   15000
   10000
    5000
                20
                                                                              180
                                                                                     200
Figure 46.    VENT2D results showing simulated concentration of total VOCs in soil with time
             at the model cell where the extraction well is located for the example described in
             Section 7.3. Note how the concentration trends asymptotically toward a limiting
             value.
                                          162

-------
            80O
            600
           400
           200
                      Total  VOCs  in mg/kg
                       200
                                400       600
                                   X,  cm
 BOO
          1000
                                                          (a)
                        Initial  Condition
                                          Extent  of NAPL
                      200
                                400        600
                                  X, cm
aoo
         1000
                                                         (b)
                       After  42 days  of  s.VE
                                                                  Extent of  NAPL
Figure 47.    VENT2D results showing simulated total VOC concentrations and extent of
            NAPL presence in soil (a) before SVE begins and (b) after 42 days of vapor
            extraction for the problem described in Section 7.4. The contours are blocky due
            to the coarseness of the finite-difference model grid.
                                     163

-------
domain), and this area would require continued remediation. Recall that soil conditions in the
simulation are permeable and homogenous; such rapid results are not likely under actual field
conditions.  This analysis illustrates how a compositional flow and transport model can be used
to examine SVE effects on the areal extent of contamination in the vadose zone.

7.5     SPREADSHEET APPLICATION TO ESTIMATE THE k/kz RATIO

       Popular spreadsheet software (e.g., Excel and Quattro) can be programmed to solve
analytical solutions and present the results graphically. The analytical solution for steady air
flow to an SVE well based on the method of Shan et al. (1992) was programmed by CRTC
(1992) into a spreadsheet format. The Shan et al. (1992) solution allows air leakage through an
open ground surface in response to SVE vacuum propagation. Such leakage can greatly affect
SVE efficiency and radius of influence.

       The CRTC (1992) spreadsheet program can be used to estimate the ratio of horizontal to
vertical permeability (k/kj at an SVE site.  Using the program, the user inputs values of kr, well
screen interval, well flow rate, pressure measurements at several radial distances from the
extraction well, and a k/k.^ ratio.  The extraction well is assumed to be screened in a
homogeneous, isotropic medium, with air leakage from the ground surface, and confined below
by the water table or a less permeable formation. The normalized vacuum distribution calculated
by the program can then be compared to actual field data, and through a trial-and-error curve-
fitting process, the k/k., ratio can be estimated (Figure 48). The purpose of this spreadsheet
application is to facilitate analysis of the effects of the k/kj ratio on surface leakage and flow
patterns.

7.6     PNEUMATIC TEST ANALYSIS USING AQUIFER TEST SOFTWARE

       Pneumatic (air pumping) tests can be analyzed to determine soil intrinsic permeability
using analytical solutions and software developed to interpret groundwater pumping tests.  For
example, Beckett and Huntley (1994) apply AQTESOLV (Geraghty and Miller, 1989) to
determine permeability from pneumatic test data using the Theis (1935) and Hantush-Jacob
(1955) equations for radial flow in confined.and leaky formations, respectively. The pneumatic
test data consist of pressure drawdown versus time values.

       The purpose of the discussion presented by Beckett and Huntley (1994) is to highlight the
differences between the Theis and the Hantush-Jacob solutions.  Their work, however, provides a
good example of using aquifer test software (in this case AQTESOLV) to analyze pneumatic test
data.  For SVE, the major difference between the two solutions is that whereas the Theis equation
assumes confined flow, the Hantush-Jacob solution accounts for air leakage into the flow zone
through the ground surface in response to pressure gradients induced by SVE.
                                          164

-------
    O
        100
    -=    10
        0.1
        0.01
       0.001
                                       Example Site
                                               -f-
                                                      -f-
6      8      10      12

 RADIAL DISTANCE (ft)
                                                             14
16
                                                                             Zd = 0.50
                                                                             Kr/Kz = 0.5
Figure 48.     Results from a spreadsheet (CRTC, 1992) solution of the analytical method of
              Shan et al. (1992) showing a fit between simulated and measured vacuum-
              distance data for a k,.:!^ ratio of 0.5.

                                           165

-------
       The results of curve-fitting using the Theis and Hantush-Jacob methods on data from four
pneumatic tests are presented in Figures 49. As shown, the Hantush-Jacob leaky solution fit the
field data much better than the Theis curve, indicating that air leakage at the ground surface is
significant at the test site. Beckett and Huntley (1994) conclude that the Hantush-Jacob leaky
solution is generally more appropriate for evaluating pneumatic test data than the Theis equation
due to air leakage.

7.7    COMPARING CONSTITUENT REMOVAL RATES USING VENTING

       The VENTING screening model can by used to examine the relative removal rates of
compounds of varying volatility. Low molecular weight hydrocarbons with high vapor pressures
(e.g. benzene) are removed more quickly via SVE than less volatile constituents such as
naphthalene. The more volatile compounds are removed preferentially due to their greater
affinity for the vapor phase; this drives partitioning into soil gas that is being extracted via SVE.

       The test data set supplied with VENTING was utilized for this example. VENTING
includes a library of gasoline compounds, of which benzene, toluene, p-xylene, and naphthalene
were selected as indicator compounds to illustrate relative removal rates. The vapor pressures of
benzene, toluene, p-xylene, and naphthalene are 76, 10, 6.5, and 1.0 mm Hg, respectively and
represent the range of vapor pressures for gasoline constituents.

       The results of a 400-day model run  are presented in terms of normalized total mass
remaining in the system in Figure 12, which was plotted using data extracted from the
VENTING output file. The total mass includes constituent mass in soil gas, moisture,  solids,
and NAPL. The normalized mass is calculated for each constituent by dividing the mass at each
time step during the simulation by the initial mass of that constituent in the system. The
normalized mass is presented to illustrate the relative rates of removal for each of the indicator
compounds. As shown, benzene is removed relatively quickly due to its high vapor pressure. In
contrast, the mass of naphthalene present in the system is not significantly reduced  during the
simulated period of SVE. The removal rates for toluene and p-xylene are intermediate between
benzene and naphthalene.

7.8    EVALUATING SURFACE BOUNDARY EFFECTS USING AIRFLOW

       The ground surface boundary condition at an SVE site has a significant effect on the
trajectories of the vapor pathlines. Air leakage from the ground surface shortens vapor pathline
travel distances because air is drawn from the atmosphere, rather than laterally through the
formation toward the extraction well. An efficient SVE system will be designed to maximize air
flow through the contamination zone. In some cases, this will occur if air flow is predominantly
horizontal with minimal leakage through the ground surface. The SVE well radius  of influence
is maximized for this case due to the absence of vertical flow from the ground surface.  Air flow
simulations can be used to examine these relationships.
                                          166

-------
 -
OP
vo
                    Pressure Drawdown (ft)
           Pressure Drawdown (ft)
                 o
                ,- o
      O
       o
            o
             o
         H

         3"
         CD
         CD
 CD
                                                      CD
                                                      O)
                    Pressure Drawdown (ft)
           Pressure Drawdown (ft)
         3
         CD
         13

         «-*-
         CD
         CO
                 II
                 55"
  D
3-a>

-------
       Two different simulations were completed using AIRFLOW to illustrate the effects of a
surface cap. For the first simulation, the top boundary of the model was set to atmospheric
pressure to simulate a leaky open ground condition. The model grid and boundary conditions of
the uniform radial-symmetric model domain are presented in Figure 50, and include a constant
pressure of 1 atm at the ground surface,  a constant pressure of 1 atm at the lateral perimeter
boundary, a no-flow boundary at the model bottom to represent the water table flow barrier, and
a constant pressure of 0.26 atm at the SVE well.  The soil permeability is set to 50 darcys
throughout the model domain.

       Vapor pathline and pressure contour plots shown in Figure 51 reflect significant air
leakage from the ground surface. This leakage reduces the rate of lateral air inflow away from
the extraction well.  The extraction well radius of influence as defined by the 0.9 atm pressure
contour line is highlighted by the shaded area in Figure 51.                          .

       The second simulation is identical to the first, except that the top boundary is set to a no-
flow condition. However, as shown by comparing Figures 51 and 52, the results are quite
different.  Due to the absence of leakage through the ground surface, all extracted air must flow
laterally inward through the vadose zone from the outer constant pressure boundary to the
extraction well. As shown, this extends the simulated SVE well radius of influence appreciably
(as defined by the 0.9 atm pressure contour line).  Surface capping (e.g., using plastic, asphalt,
clay, etc.) is sometimes used to achieve this effect. In other cases, controlled surface leakage
and/or air inlet wells or trenches can be designed  to enhance flushing of a contaminated zone.

7.9    SIMULATING PRESSURES INDUCED BY SVE USING AIR3D

       AIR3D was used to construct a three-dimensional air flow model to evaluate the pressure
distribution induced by a horizontal SVE drain installed at depth. The CSUGAS code could also
be used to analyze this particular air flow problem.  A 50 x 50 x 5 model grid was designed to
represent a volume of soil with dimensions of 130 x 130 x 10 meters. Homogeneous porosity
and horizontal permeability values are specified as 0.3 and 10'8 cm/sec, respectively. The
vertical permeability of the model domain is set to 10'9 cm/sec. The extraction drain is
represented by seven cells with a constant pressure set to 0.92 atm in layer 3, row 30, columns 24
through 30. Boundary conditions are atmospheric pressure at all outer cells, except for a no-flow
condition at the bottom of model.

       For the purpose of this analysis, the area of influence of the extraction drain is (somewhat
arbitrarily) defined as the area inside of the 0.99 atm pressure contour.  The simulated pressure
distribution is shown in plan and cross-sectional views in Figures 53 and 54, respectively. These
perspectives allow estimation of the SVE drain area of influence in horizontal and vertical
dimensions. Alternatively, a three-dimensional block model analysis could be used to visualize
the SVE system's three-dimensional zone  of influence.. These types of analyses can be used to
design drain dimensions, spacings, and vacuums.   •
                                           168

-------
CO
(X
LU
CQ
01
UJ

O
CD
UJ
o


I
CO
Q
           SIMULATION  DOMAIN AND BOUNDARY  CONDITIONS
     3.0-e
£j    2.4
                                                              Q
O'-O         1.0         2.0         3.0          4.0


     RADIAL DISTANCE FROM EXTRACTION WELL,  METERS'


   D EXTRACTION WELL  NODES, P=200 TORR ( 0.26 ATM)

   O CONSTANT  - PRESSURE  NODES, P=760 TORR (1.0  ATM)
                                                                   -t)
                                                                   5.0
Figure 50.    Model geometry and boundary conditions input to AIRFLOW to simulate the

           example described in Section 7.8 (with the top of the model open to the
           atmosphere).
                                  169

-------
    o:
    UJ
    m
    of.
    UJ
    I
    $
    CD
    o
    1
    o
    in
    C£
    UJ
    UJ
    CD
    UJ
    UJ
    §
    UJ
    o
    1
         0.6-
          0.0
          1.2-
0.6-
                                        U\ AREA OF  INFLUENCE  OF
                                            THE  EXTRACTION WELL,
                                            P<710 TORR  .(0.93  ATM)
                                                        740
         I:; Iff tiSaK.^w"*145™"''1''"'"'''""™™^                  '
         PRESSURE  DISTRIBUTION  [TORR
         i:»,rt>i,n'i't-i.ti>.Vi;'.n'"irr:':'^-.:i'.Mi'iT!?'iV,Wii'iJI
          Q Q  pl^yii;t>Tir*^c.i-iW^rwo^t:,^v^;t)r.!v^ii:'r.-^yli^v.v.v^n^  	_

             0.0           1.0           2.0           3.0           4.0
                      RADIAL DISTANCE FROM  EXTRACTION WELL, METERS
                                                                        5.0
Figure 51.    Vapor pathlines and pressure distribution simulated using AIRFLOW showing
             significant leakage from the surface due to the open ground surface boundary
             condition.
                                          170

-------


CO —
o:
UJ
tj 2.4-
s

_i
CD
<
i5 1.8-
oc.
UJ
g
5
uj 1 .2-
g
m —
<
ft n fi
O (J.D —
<
oo —
o



* PAIHIINF^

_^_ 	
-{ 	 — _ 	 	 __
^_ 	 ___ 	
-< 	 	 - - - —
^ 	
-<- 	 	 —
-^ 	 _ 	 	
-£ 	 	

-£__ 	 __ 	
<
-C- 	 	
^ 	
^ 	
^ * -— .
^ 	 ._ 	
^- - 	 	 	 	
-f- 	 	 	 	 	 — • 	 — 	 	 	



i i i i i • i i — i 	
   CO
   o:
   UJ

   UJ
   UJ

   m
   o:
   UJ
   UJ

   o
   m
UJ
o

ii
oo
Q
           0.0
        3.0-
     2.4-
        1.8-
        1.2-
        0.6-
        0.0-
                     1.0
2.0
3.0
                                                             4.0
                                                                       5.0
                                                    AREA  OF INFLUENCE
              P<710  TORR (0.93.ATM)
                                                o
                                                             740
              PRESSURE  DISTRIBUTION  [TORR]
           0.0
                     1.0          2.0         .3.0          4.0          5.0


                  RADIAL DISTANCE FROM' EXTRACTION WELL, METERS
Figure 52.
          Vapor pathlines arid pressure distribution simulated using AIRFLOW for. the

          confined case where the ground surface is modeled as an impermeable, no-flow

          boundary. Note that flow is essentially horizontal towards the extraction well,

          and that the radius of influence extends beyond that simulated for the leaky case

          shown in Figure 51.
                                      171

-------
                            .03
                                     OC
                                             .CO
                                                     0(1
                                                                  m
                                                                        ftt
                                        A
                                                                             Q£
                                              0.9,
                                              IM
                                             Si
                           ^fl
W&
                 ^J
                        TJK
       1|.pp
                                                                 1 0
                                                                         1.00
Figure 53.    Results of AIR3D simulation showing a plan view of the pressure distribution
              resulting from a horizontal SVE drain installed at depth as described in Section

              7.9.
                                            172

-------
       West
Cross-Section along Row 31
                                                     J
East
Figure 54.    Results of AIR3D simulation showing a cross-sectional view of the pressure
             distribution resulting from a horizontal SVE drain installed at depth as described
             in Section 7.9. Each layer is two meters thick.
                                         173

-------
                                      SECTION 8
                           SVE SIMULATION CASE STUDIES

              Several case studies are discussed briefly below to present examples of ways in
 which models have been applied at SVE sites. Detailed case study data of SVE model
 application at field sites were difficult to obtain. The first case study is a Superfund site in South
 Hope, Maine, where a combination of air flow modeling and column tests was used to design the
 SVE system (Section 8.1). This is followed by examples of the use of a  screening model
 (Hyperventilate) to assess SVE feasibility (Section 8.2.1) and then a flow and transport model
 (VENT3D) to examine SVE performance (Section 8.2.2) at a UST spill site in California. The
 last case study is of another Superfund site, the Verona Well Field Site in Battle Creek,
 Michigan.

 8.1    UNION CHEMICAL COMPANY SUPERFUND SITE, SOUTH HOPE, MAINE

       The discussion presented below follows from Balsam Environmental Consultants (1992).
 The contaminants present at the site include mainly tetrachloroethene (PCE) and trichloroethene
 (TCE) at total concentrations ranging up to greater than 100 parts per million (ppm).  Other
 contaminants present include 1,1-dichloroethene (DCE) and xylenes. Site stratigraphy consists
 of localized fill overlying unconsolidated drift deposits or glacial till which extend to bedrock.
 The fill consists of silty sands and gravels with organic material such as tree stumps present in
 some areas.  The glacial till consists mainly of silt, although it may contain some silty clay and
 fine to coarse sand. Sandy till occurs in discontinuous zones throughout  and may contain
 cobbles, boulders, and gravel. The thickness of the till ranges from 25 to 70 feet.

       The proposed approach for SVE design at this site was based on a literature review of
 other case studies, numerical modeling studies, field-scale air pumping tests, and bench-scale
 column studies of mass transfer rates.

       Initial numerical modeling was performed in order to  estimate air flow rates and radii of
 influence based on an initial estimated permeability. A field-scale air pumping test was then
 designed to develop data on the permeability of the soil matrix, and to design well spacings and
 assess operating air flow rates. Results of the field test were used as input to a second round of
modeling hi order to develop more refined estimates of soil permeability. Bench-scale mass
transfer column studies were then completed using air flow rates developed from the field and
modeling studies. The column studies were used to evaluate the potential for reaching site-
specific soil cleanup levels and determine contaminant removal rates. The number of pore
volume exchanges required to reach soil cleanup levels was estimated based on the results of the
column studies. Mass transfer and removal rates gleaned from the column studies were then
extrapolated to make estimates of site cleanup times.

       Air flow modeling studies were completed in order to determine air permeability values
based on the results of field air pumping tests. Air permeability values can be determined by

                                         174

-------
matching model results with field results. A two-dimensional, radial-symmetric air flow model
was utilized to perform the calculations. Two field air pumping tests were completed in order to
obtain pressure data at more than one air flow rate. The model was used to fit data from two
separate air pumping tests at two different pumping rates in order to allow for calibration and
verification of the model.  The initial calibration stage consisted of fitting  the model results to the
field results from one of the pumping tests.  The verification stage was then performed to see if
the model results fit the field results at a second air flow rate.  Verification consists of testing the
match between model results and field data using data other than those that were used to calibrate
the model.

       A number of air pumping tests were  completed in several areas around the site.  Only the
results from one of these areas, Area A, will be discussed here. The stratigraphy of Area A
consists of a uniformly anisotropic distribution of fill from the ground surface to approximately
ten feet below grade, where the water table is encountered.  VOCs were detected in this area
down to a depth which corresponded approximately with the water table.

       An air pumping well was installed in the center of the test area and screened from a depth
of 2.5 to 7.5 feet BGS. A surface seal consisting of several inches of clay  overlain by a
geomembrane was placed on the ground in the vicinity of the test area. Passive air inlet wells
were placed around the perimeter of the test  area in order to supply a source of uncontaminated
air to the test area. However, test results subsequently indicated that the radius of influence of
the pumping well extended well beyond the  location of the air inlet wells.  Vacuum probes were
placed at a depth of five feet and radial distances of 3.75, 7.5,  11.75, and 16.2 feet from the air
pumping well.  These probes allowed development of a pressure versus radial distance curve in
the vicinity of the air pumping well (see Figure 55).

       The first step of the modeling was to make an initial determination of the intrinsic
permeability of the site.  A simple, one-dimensional air flow model was used to make an initial
estimate of permeability.  An initial permeability estimate could also be calculated based on the
saturated hydraulic conductivity. Hydraulic  conductivity data for the site were available from a
variety of pumping and slug tests which had been completed.  Once an initial determination of
the permeability is made, it can be used as an initial input to the air flow model. This serves as a
starting point from which to vary the model permeability in order to match the simulated results
with the field results.

       A two-dimensional, radial-symmetric model was used to simulate the results from the
field air pumping tests. The boundary conditions for the model are shown on Figure 56. The top
boundary of the model is assumed to be open to the atmosphere and the pressure at this
boundary is set to one atmosphere. A sealed boundary at the model top can be simulated by
adding a thin, low-permeability layer at the top of the model. The right-hand model boundary is
set at 15 feet, and the pressure here is taken to be one atmosphere (the field measurement at 16.2
feet was 0.999360 atm during the air pumping test).  The right boundary is set at the
approximate furthest point of influence from the extraction well (i.e., the radius of influence).

                                           175

-------
                                                           VP-A-03m
                                                                       VP-A-03m
                                                        Field Data
                                                        Simulated Data
                          RADIAL DISTANCE
Figure 55.    Comparison of field-measured pressure data during an air pumping test to the
             results from a radial-symmetric air flow model (from Balsam, 1992).
                                        176

-------
               r = 0 ft
       2.5 ft
 10 ft
       2.5 fl .
                           • VACUUM  EXTRACTION WELL
                    NO-FLOW
SCREENED INTERVAL
P = 0.9 ATM
                   NO-FLOW
                                                 P = 1  ATM
                                           NO-FLOW BOUNDARY

                                           	„	 ic ft	,_
                                                                                         r =  15 ft
                                                                                            P  = 1  ATM
Figure 56.     Boundary conditions for a radial-symmetric model used to simulate an air
               pumping test at a single extraction well.
                                              177

-------
The bottom boundary of the model is set to a no-flow boundary to simulate the presence of the
water table, because no air flows across the water table. On the left boundary, where the well is
located, the pressure over the screened interval of the well is set to the pressure which was
measured at the well in the field (0.9 ami).  The area above the well along the left boundary is
assigned as no-flow because this portion of the well has been sealed and grouted. The section of
the left boundary underneath the screened interval is set to no-flow because negligible flow to the
well will occur here.

       The air flow modeling consisted of trial-and-error calibration whereby given the
extraction rate, pressure at the well, and the screened interval, the values of radial and vertical
permeability were varied until the modeled pressure distribution curve matched the field-
measured pressure values. The air flow rate and pressure at the pumping well were already
known from field measurements. Values of pressure from the model at a depth of five feet (the
location of the field vacuum probes) were plotted against the pressure values measured in the
field at a depth of five feet.  Figure 55 presents a plot of the calibrated model results as compared
to the field results.  In this manner it was possible to determine the values for radial and vertical
permeability, based on the results of the air pumping tests.  The model was then verified by
matching the model results to a second set of field data that had been collected from a second air
pumping test performed at a different air flow rate than the first test. Simulating a second set of
field pressure values collected at a different air flow rate increases the confidence in the ability of
the numerical model to accurately simulate the system.

       During the treatability study, total VOC concentration in extracted gas typically declined
exponentially within the first few days of operation. When the system was shut down after 30 to
60 days of operation, VOC concentrations in soil gas exhibited negligible rebound, probably
indicating that residual contaminant mass in the subsurface had been significantly reduced.
Results from soil samples collected after completion of the treatability study were difficult to
interpret due to a variety of reasons:  CLP method detection limits, localized heterogeneities, and
soil samples collected from inappropriate areas. However, the conclusion was drawn that, based
on concentrations hi extracted vapor, contaminant mass was removed from the subsurface,
although attainment of soil cleanup levels could not be confirmed.

8.2    CALIFORNIA UST SITE

8.2.1   Hyperventilate Screening Application

       This case study was originally presented in Johnson (1991) and Johnson et al. (1992), and
the discussion presented below follows from Johnson (1991).  Data from a service station site in
Costa Mesa, California is used in conjunction with the Hyperventilate software to evaluate the
potential feasibility  of SVE as a remedial action at the  example site.

       The vadose zone stratigraphy is presented in Figure 57. It consists often feet of sandy
clay at the surface, overlying 20 feet of fine to coarse sand, 13 feet of silty clay and clayey silt,

                                           178

-------
              North
                                                                            South
     10'
      20,
   co  30,
    o
   O
    1*0.
      60'
•
' •
•
•
•
1
•




rt
i
0.3
•0.02
• 0.0
.0.0

• 0.0
-1.7
-24 t


-9.5
B-17
\ Tank
Sandy \ Backfill
day \ (fonnerunk
\ locKion) '
\
•
Fine to
Coarse Sand
•

Silty Clay
&.
Clayey Silt
Ml
1
Medium Sand
•
r
H
0.5 /
1.7 y — — •
512
-5.4
• 8577 .
•341- — — -
•653
•3267
.1237
• 23831

• 1.7
B-10 H
0.8
..4.6- _ _ __
. 0.3
-8.2
.214 •
-•31— — — — •
• 967
• 971
^28679 -
. 23167


B-5 H
0.31
— 1.2-
- 0.44
• 0.17
-8.8
rO.63
- 0.86
. 23
-1.6
. 3.2 T


EB-3
         Static Ground
         Water Table
                                                                     SCALE (ft)
                                                                         i	1
                                                                         10
                                                                                  20
Figure 57.     Sample site data.  Values for TPH concentrations in soil (mg/kg) are posted along
              borings (from Johnson et al., 1992). [Reprinted by permission of Lewis
              Publishers, an imprint of CRC Press, Boca Raton, Florida.]
                                           179

-------
 and seven feet of medium sand, respectively. This exercise will be used to evaluate the potential
 for utilizing SVE for cleanup of the lower soil zone (45 to 50 feet BGS), which is composed of
 fine to medium coarse sands, and contains hydrocarbon levels greater than 20,000 ppm.  This
 zone is probably a good candidate for SVE because the contaminant concentrations are high and
 the expected permeability is high due to the coarse sandy lithology.

       The first step in the evaluation is to estimate the potential air flow rate in the subsurface.
 Recall that the Hyperventilate air flow equation is an analytical solution based on confined,
 radial flow to one extraction well. Since the lower sand layer is overlain by a low-permeability
 layer, the assumption of confined flow is reasonable.  The following input parameters are used
 for Hyperventilate Card 8 (see Figure 42): well radius = 2 inches, radius of influence = 40 ft,
 and interval thickness = 6.6 ft. Based on these parameters, a range of potential flow rates are
 calculated (Figure 42) for a range of well vacuums. For a typical extraction well vacuum of 40 to
 60 inches of water, estimated air flow rates are 3 to 37 scfm.

       Next, an estimate of the contaminant vapor concentration that can be expected from an
 extraction well is calculated. Assume an ambient temperature of 18°C and a weathered gasoline
 contaminant composition. The weathered gasoline composition has been included with the
 Hyperventilate software. Card 10 (Figure 58) presents the maximum theoretical vapor
 concentration (in mg/1) based on the contaminant concentration and ambient temperature.

       Based on the maximum vapor concentration and the range of potential flow rates
 calculated previously, it is then possible to calculate a range of estimated maximum theoretical
 removal rates (Card 12, Figure 59). These rates can be used to determine if, based on the total
 estimated quantity  of contaminant at the site, the contaminant mass could be removed in a
 reasonable amount of time.  Card 13 (Figure 60) performs calculations, based on the estimated
 contaminant mass present and the desired remediation time, of vapor extraction well flow rates
 which would be required for mass removal based on a user-defined vacuum at the well.
 Estimated removal rates are also calculated. Note that the desired removal rate of 22 kg/day is
 less than the projected minimum removal rate of 165 kg/day.  The estimates for removal rate
 represent an idealized, best-case scenario.  The calculations assume that all of the free-phase
NAPL is contacted by the air flow and that none of the rate-limiting processes that we observe at
 real sites are operating.

       The next step in the screening process is  to estimate the changes in vapor concentrations
 and removal rates over time as the SVE operation progresses.  Typically, due to preferential
removal of the more volatile constituents, the rate of removal declines over time. The
multicomponent contaminant is represented in terms of boiling point ranges, rather than
concentrations of each specific contaminant. This  is a useful manner in which to deal with
complex, multicomponent mixtures because it depends only on an analysis of several ranges of
constituents, rather than many constituents themselves. Boiling point curves can be developed
based on knowledge of the gas chromatographic (GC) elution behavior of a known series of
constituents in a given boiling-point range.  Because constituents generally elute from the GC

                                          180

-------
                       Vapor Concentration Estimation - Calculation
              1] Typein Temperature (C) (hit )

                  Cfickto Enter Composition of Contaminant
             7) or
                  Choose one of the D ef auft Distribvrtions

              3J Qiclcto View Distributions, (optional)

              4J Cft&to Perform Calculations
                                         18
                               O  Enter Distribution
                               O  "Fresh" Gasoline
                               ®  "Weathered" Gasoline
C
                                   View Distributions
                                    Perform Calculations
                Results:
                    How Do I Measure a Distribution?
Sum of Mass Fractions
Cab. Vapor Pressure
Calc. Vapor Concentration
                   101
                                                                             1.00000
                      atrn
                      mg/l
     About Calculation JHK  Print Card
Figure 58.      Card 10 from the soil venting stack.
            Maximum Semeyal Hate
            Estimates
           select your unit preference below
                   ®   [Ib/d]
                   O   lkg/d]
           Note:
           These are "maximum removal
           rates", and should only be used as
           screening estimates to determine
           iE venting is even feasible at a
           given site. Continue on to the next
           card to assess if these rates are
           acceptable...
         Temperature [C)
         Soil Type   .     ;
         Soil Permeability Kange (darcy)
         Well Radius (hi)
         Eadius of Influence (ft)
         Contaminant Type
         Permeable Zone Thickness (ft)
Pw - Well Flowrate Estimates
Vacuum [SCFM]
(inH^O) (single well)
	 5.......
	 JL_
......IP......
	 «_
.60 	
120
200
0-33 ...
0.66
__jaL_
	 2,54 	
	 .371 	
	 £83 	
10.07

to
to
to
to
to
to
to
3.32
	 isa___
.........13,0.2 	
	 25,38 	
	 3.7,09.......
68.27
100.66







Ma». Removal Bate Estimates
[Ibfd]
(single well]
6
.. 	 12__
	 	 25.. 	
.. 	 52... 	
	 JO. 	
.......JZ8 	
364

to
to
to
to
to
to
to
62
„ 	 12.4 .......
	 .251 	 _
	 517. 	
	 .799.. 	
1778
3636

Figure 59.      Card 12 from the soil venting stack.

                                                         181

-------
               AtiWspoirrt. you cort^arethemaxTmum
               possibleremovalTate with your desired
               removalrate.

               If themaximurnrernovalrate does not exceed
               your desn:edremovalrate.then soifventingis
               not Jikelyto meet your needs, and you should
               consider another tceatmenttechnology. or
               males y our needs mote realistic.

               Infhenextcards.wewilrefinetheranoval
               rate estimates, m order to decide if venting can
               achieve your objectives.
                                              days
      Single Vertical Hell Results
Desired Removal Bate:
Gauge Vacuum (in H20):
Mm Flowrate @  200n E20
MaK Flowrale @ 20CmH2O
Man. Est. Removal Bate:
(lower estimate) -per well
(upper estimate) -per well
   200
 10.0?
ioo.6i>
[mH2O]
[SCFM]
[SCFM]
Figure 60.      Card 13 from the soil venting stack.

                                                          182

-------
column in order of increasing boiling point, a boiling point distribution can be developed by
grouping all of the unknowns that elute between two known peaks (Johnson etal., 1990b). Table
24 presents an example boiling point distribution list of compounds for gasoline.

       Card 16 (Figure 61) prompts the user to input a number of boiling point ranges to define
the contaminant composition. Estimated residual concentrations over time are then calculated
based on these boiling point ranges, and are presented on Card 17 (Figure 62). Note that as the
flow rate, Qt/M(0), increases, the vapor concentrations and residual soil levels decrease, and the
composition of the residual becomes enriched in the less volatile constituents.

       The initial saturated vapor concentration and volume of air which must be extracted per .
gram of initial contaminant mass are also calculated. These values are used later in the screening
process.

       The final card (Card 18, Figure 63) presents a range of the minimum number of
extraction wells which are required to achieve a 90% reduction in contaminant levels under ideal
conditions. The range of 0.7 to 7 wells is a reasonable number of extraction wells for a service
station site.  A range of, for instance, 10 to 100 wells should motivate the user to consider other
remedial alternatives for the site, because it would be untenable to install so many wells at a
small site.

       The results presented above have been calculated based on approximate, analytical
solutions to the equations of air flow and contaminant transport.  The model predictions should
be used as guidelines, rather than absolute predictions of what may happen under actual site
conditions. The screening model results should be used in conjunction with knowledge of site
conditions and best professional judgment in order to make an informed  decision as to the
potential feasibility of SVE as a remedial action.

       The Hyperventilate program also calculates permeabilities based on an analysis of
pressure versus time data from an air pumping test. Given a set of pressure versus time values
collected during a field air pumping test, a radial-symmetric, confined-flow analytical solution is
used to estimate the permeability of the soil. Two different methods (A and B, see Card APS,
Figure 64) are used depending on the availability of data.  The first method is used when both the
flow rate, Q, and the screened interval of the well are known. The second method is used when
Q or the screened interval are not known with confidence. Help screens  contained within
theHyperVentilate software provide additional explanation and present the  analytical equations
upon which these analyses are based.

       The next step in the SVE screening process, assuming that the site has passed the
screening steps presented previously, is to develop a preliminary design of the system.  The
Hyperventilate screening model should never be used as an actual design tool, but it is
appropriate for developing a preliminary conceptual design for scoping and costing purposes. A
                                           183

-------
Table 24.  Example boiling point distribution for gasoline (from USEPA, 1993).
B.P. Range
1






2















3









4









Compound
propane
isobutane
n-butane
trans-2-butene
cis-2-butene
3-methyl-1-butene
isopentane
1-pentene
2-methyl-1-butene
2-methyl-1 ,3-butadiene
n-pentane
trans-2-pentene
2-methyl-2-butene
3-methyl-2-butadiene
3,3-dimethy 1-1 -butene
cyclopentane
3-methyl-1-pentene
2,3-dimethylbutane
2-methylpentane
3-methylpentane
n-hexane
methylcyclopentane
2,2-dimethylpentane
benzene
cyclohexane
2,3-dimethylpentane
3-methylhexane
3-ethylpentane
2,2,4-trimethylpentane
n-heptane
methylcyclohexane
2,2-dimethylhexane
toluene
2,3,4-trimethylpentane
2-methylheptane
3-methylheptane
n-octane
2,4,4-trimethylhexane
2,2-dimethylheptane
ethyl benzene
p-xylene
m-xylene
3,3,4-trimethylhexane
Boiling Point
-42.07
-11.40
-0.05
0.88
3.70
20.00
27.85
29.97
31.16
34.00
36.07
36.35
38.57
40.00
41.20
49.26
51.14
58.00
60.27
63.28
68.95
71.80
79.20
80.10
80.74
89.80
92.00
93.50
99.24
98.42
100.90
106.84
110.60
113.47
117.70
118.00
125.66
126.00*
127.00*
136.20
138.35
139.10
141.00*
                                     184

-------
Table 24.  Example boiling point distribution for gasoline (from USEPA, 1993) (continued).
        B.P. Range
Compound
Boiling Point
                            o-xylene
                            2,2,4-trimethylheptane
                            ,n-nonane
                            3.3.5-trimethylheptane
                            n-propylbenzene
                            2,3,4-trimethylheptane
                            1,3,5-trimethylbenzene
                            1,2,4-trimethylbenzene
                            n-decane
                            methylpropylbenzene
                            dimethylethylbenzene
                            n-undecane
                            1,2,4,5-tetramethylbenzene
                            1,2,3,4-tetramethylbenzene
                            1,2,4-trimethyl-5-ethylbenzene
                            n-dodecane
                            naphthalene
                            n-hexylbenzene
                            methylnaphthalene
                                144.40
                                147.00*
                                150.80
                                152.00*
                                159.20
                                159.00*
                                164.70
                                169.35
                                174.10
                                185.00
                                189.75
                                195.90
                                196.80
                                205.00
                                208.10
                                216.30
                                218.00
                                230.00*
                                241.05
    *This is an approximate value.
                                         185

-------
             Model Predictions
             To the right is a summary of the data you
             have input. IE you wish to change any o£ the
             Wo. (hen click en the parameter name, and
             redo the calculations on the ca Set Default BP Eanges <-    J
SfflllflJMB.3nflSJ.3l 	
BoitrnqPoint Range 82 	
IiilQflPJolS.aB.geJ3 	
B oifinq,Poirrt Ranqe 84 	
Boffino Point Range 85 	
	 .-50 	
28 .
	 80 	
Ill
M
.J.o.
to
.19...
to
to
........?8 	
8Q
	 1.11 	
I'M ....
250
JL
JL
JL
JL
c
                                                Generate  Predictions J
Figure 61.       Card 16 from the soil venting stack.
                  C   ~> Import Data <—    J  Saturated Vapor
                                                    Concentration at lhne=0
                FIRST PRESSTHE IMPORT DATA
                BUTTON!
                                     Mm Volume to Bemoue
                                     >WX of Initial Hesidual
                                                           Temperature (oeC):
                                                           ContaminantType:
These are the results for the contaminant
time that uouhave specified. All of this
                     Weathered Gasoline  |
                                                                                               BP#5
                                                                                              Residual
                                                                                              [•/. total]
                                                          BP#3
                                                         Residual
                                                         [X total]
 BP#2
Residual
[x total]
Besidual
 Leuel
[X Initial]
 BP#1
Residual
[•/. total]
 Vapor
 Cone.
[y. Initial]
                                        100.00
                                        95.00
                                        90.02
                                        85.03
                                        30.03
                                        75.03
                                        70.04
                                        65.04
Figure 62.      Card 17 from the soil venting stack.

                                                          186

-------
           Thisis a complete summary of the data
           andicesdts. Basedipmtheseimim'bet's.a
           'Vininimumimimber of wells" has been
           calculated, which should gwe you some
           indication of how aipopriatevenitingisfot
           your application. Note that this is the
           number of wellsif circumstances are ideal
           Which they rarely are.

           The next card discusses some of the
           conditions that may limit the effectiveness
Temperature [o?C]:
ContammantTitpe:
Soil Type:
Hell Radius [hi]:
Est. Eadius o£ Influence [ft]:
Permeable Zone Thickness [ft]:
Flowrateper Well (120" Vac) [SCFM]
Flowrateper Well (120" Vac) [SCFM]
Mm. Vol. oE Air [L/g-residual]:
Estimated Spill Mass:
Desired KemediationTime [days]:

18
I Heathered Gasoline
Fine Sand
FM]
FM]
]:
2
40
G.G
6.83
68.27
128.48
4000
190

*g

                                             0.72
           Hmhnum t of Hells Based
           on Your Input Parameters
                                                                                               7.23
Figure 63.       Card  18 from the soil venting stack.
3
/
c
fi
) screened inte
(hiekness

6.6 |

(SCFM)
rual
(t)
^ K
--> Calculates- 1 ,.
I ^ ^fli






* *^
fest - ]
IP
Dal

ta Analys
| I
is (cont.)


•+ <= 53 |(&] f= | 32.4 |(Et)
(mm) (mH2O) (mm) (mH20)
9
> ' 11
15
23
30
40
100
0.1 ±
0.2 _
0.2
0.4
'0.7
1.3
2.3
4
7
9
12
16
24
30
39
52
77
1.2 I
3.0 "
4.3
5.5
6.9
9.9
11
13
'16
20 "
f clear ^ f clear J
= 18.694895
= 9.298663



mm
darcy(A) t= 2.858452 darcyU
darogfBl t= 7.767599 darcy(I



Return m
p— ,

I [ ; I

i

'= 1
$838338$8&
ss; 1 1

':

IfEl)
I


(mm) (in E20)
b
•
0 t=
3) 1=


C clear ^





t
i_

darcgfAl
darcg(B)
^ Ann 3

Figure 64.      Card APS of the air permeability stack.

                                                          187

-------
more rigorous analysis, which might include air flow modeling or compositional flow and
transport modeling, should be applied during the actual design process.

       Card SD2 (Figure 65) prompts the user for descriptions of different lithologic units at the
site and information concerning the amount of contamination present in each soil unit.  Then,
based on the contaminant composition input previously, the screening model calculates a range
of the minimum number of wells required for cleanup based on the desired cleanup tune and
extraction well vacuum. From the results (Card SD4,  Figure 66), it is apparent that the flow rates
and removal rates are high for the medium sand (minimum number of wells is less than or equal
to 0.09), while removal rates are quite low for the clayey silt layer. Based on the minimum
estimated number of extraction wells (64 to 643) for the clayey silt, SVE may not be an
appropriate remedial technology for cleanup of this zone. For the fine sand, the estimated
minimum number of wells (four to 35) lies in the middle ground between workable and
unworkable. If the actual number of extraction wells necessary were toward the lower end of this
range, then SVE could prove to be a workable remedial technology. However, if the number of
extraction wells necessary were toward the upper portion of this range, then it may be
appropriate to look into other remedial technologies, especially if the area of the site were
relatively small, such as a typical service station.

       Field results from the project are presented in Johnson et al. (1992).  Figures 67 and 68
summarize the results of the SVE operation.  By 370 days, a total of approximately 900 gallons
of product had been removed from the subsurface (Figure 67). The removal rate had dropped
from 50-60 kg/day to about 5 kg/day (Figure 67).  Figure 68 illustrates the change in vapor
composition over time. Compounds are grouped hi the following manner based on their boiling
points:
                    Methane - Isopentane (<28°C)
                    Isopentane - Benzene (28-80°C)
                    Benzene-Toluene  (80-111 ° C)
                    Toluene-Xylenes (111-144°C)

After the first 20 days, there was a shift hi composition to less volatile compounds. The
composition remained relatively unchanged until day 266 when dual vapor extraction/
groundwater recovery wells were added, which caused a decrease in the vapor concentration.
Although a significant amount of NAPL was removed at the site, after 370 days it was still
unclear what residual level of NAPL could reasonably be reached.

8.2.2.  Flow and Transport Model fVENTSD') Application

       This  application is provided by D. Benson (written communication, 1995).  To
summarize the findings of the field study, as presented by Johnson et al. (1992), the soil beneath
the site is generally found in four layers that differ in grain size and degree of contamination.
From zero to 10 feet BGS, a relatively uncontaminated sandy clay is present. From 10 to 30 feet
BGS, a generally coarse sand is present with small areas of somewhat  elevated total petroleum

                                          188

-------
             Design Input Parameters...
                                               *
                (soil stratigraphy & contaminant characteristics)
i Please enter the required information for each distinct soil layer, Please
j enter the required information for each distinct soil layer, elicit on the
| "Update"button, and then proceed to the newt card (i.e. elicit on right
arrow at bottom!.

...J...
2
.....?...
4
b
6
7
_£.
Description of
Soil Unit
	 Med.jum.Sand.
.Fine Sand ;._



	
Depth BBS'
	 1.9......
30
	 .43..;.




...to..
to
...to...
to
to
to
to
to
	 30 	
43
-Jfi 	





Description of
Contamination
	 52j 	 -
gasoline




mat youpreter
C Clear All Entries ^

Contaminant Distribution
radius
[ft]
	 20......
20
	 ...20.,...




interval
tniclcness
[ft] 	
......IP..™..
13
	 7 	




average
cone.
[mgfltg]
. 	 100. 	
1000
......igpog......




• Below Ground Surface
U [Ib]
Calc.
Total
Mass."
	 ... 	 120.9
786 U
4232.3
'"""""" 	 or]
	 or]
On

" 	 arj
f n_Jr,rn^
                                                     Retrain
                                                                                           SD2
Figure 65.     Card SD2 of the system design stack.
             Design Input Parameters...
                                                      Note: - click on any table heading to get more info
Please enter (1) the desired time period for remediation. (2)
the design gauge vacuum, and then (3) clioX the "update"
button.
Description of
Soil Unit
	 1...
....?....
....!..
.....i..
.....5...
e
.....?....
i 	 «.,
.^Sd.iHK'Jis!!!:!....:
.CJ.ay,ey.Sjjt 	
5M.Sand 	
	
	 •• 	
Time for
Clean-up
[days]
	 18fJ
	 180
180
	
	
Design
Vacuum
(mH2Q)
	 	 4.0
	 40
40
	
	 -,.--
- use arrow Ttey to wouetetween cells

Flowrateper Vapor
EKtraotion Well
tSCFM]
	 38.M
	 	 OJ02
1.92
	 	 	 MA
	 ...MA
MA
	 .....MA
----- NA
.J.o.,.
...«S...
to
...19...
...to...
tn
Jo..;
to
	 3.84.38
	 PJ?
19.22
	 MA
	 MA
MA
	 MA
MA
®

' Dpdate 1

Minimum Number o£ Wells
Sascd on An
	 0.3
	 .......0,3
0.3
	 	 MA
..; 	 MA
ISA
	 MA
HA
-, Based on Critical
Volume"
	 0.0
	 64.3
35
	 MA
	 MA
IfA
	 m
MA
:JS...
...to...
to
...IS...
...to...
rn
.to..
to
	 0.0
	 64.3.0
34.6
	 MA
	 	 MA
MA
	 MA
MA
                        • not enough input data
                                                     •• minimum volume oE vapor required to achieve remediation
                                                                                           SD4
Figure 66.     Card SD4 of the system design stack.



                                                  189

-------
     60
     50-
                                     1000
                    Total Removed
                      100
     200
TIME, days
300
                                                                      0
Figure 67.    Total contaminant removed (expressed as an equivalent volume of gasoline) and
            mass removal rate over time for the California UST case study site (from Johnson
            et al., 1992). [Reprinted by permission of Lewis Publishers, an imprint of CRC
            Press, Boca Raton, Florida.]

                                      190

-------
  -

  CfQ



  I

  Ov
  OO
  w
  o
  o


S- O


5s
o 1
CO
s S
W to
 ^
          VAPOR CONCENTRATION, mg/l
  o
  o
                 ^a.  NJ

                 m  s
Q.

03
                    CO
                    o
                    o
                    o
                    o
                               o
                               I
                       o
                       I
                                 en
                                 o
co
o
                                ?f CD 5T  A
                                — <" O  JT-

                                I g 1  I

                                ? s I  s
                                   i
                                * 1,

                                i t
                                ® §
                                » i

-------
hydrocarbon (TPH) concentrations. A silt and clay rich layer is present from approximately 30
to 40 feet BGS, and a well-sorted, fine sand is present beneath the silt and clay layer. The water
table is found within the well-sorted, fine sand layer at a depth of approximately 48 feet BGS.
The highest levels (isolated samples up to 30,000 mg/kg) of TPH were found just above the
water table in the fine sand. NAPL gasoline accumulated in several monitoring wells.  Several
samples (but a relatively small percentage of the total number) collected from the bottom of the
silt and clay layer also were shown to have TPH concentrations in excess of 10,000 mg/kg.
Because the majority of the contaminant mass was found in the lowest layer, a two-dimensional
areal layer has been used to simulate SVE at the site.

       Several vapor extraction wells were installed at the site (Figure 69), and existing
monitoring wells with screened intervals above the water table were also piped to the vacuum
system. During the field study, vapor was withdrawn from only one well for about 275 days
(Johnson et al., 1992). After this time, other wells and configurations were tested for various
durations, causing many disruptions in the quantity and location of SVE. As a result, only an
average centralized SVE rate has been used in simulations longer than 275 days. Additionally,
air sparging wells were used later in the project. The injection rates were low compared to
extraction rates; therefore, the injection is not modeled in this section. Certainly, a more-detailed
three-dimensional model could be constructed for the site, but the large number of alternative
SVE well configurations and extraction rates used precludes viable temporal simulations.

       TPH concentrations from all samples collected from the site are lognormally distributed.
The log-transformed concentrations from the layer of greatest contamination (40 to 48 feet BGS)
were kriged, giving the spatial distribution of initial TPH concentrations in soil shown in Figure
69. A great number of soil physical parameters  were measured at the site, and will not be listed
here (e.g., air permeability, porosity, organic carbon content, etc.).  These parameters are listed
by Johnson et al. (1992) and included in the data files used by VENT3D. The NAPL
contaminant was analyzed by GC methods, and  found to contain 47 identifiable compounds.
The simulations that follow use this reported composition.  Two relatively simple simulations
were constructed. The first uses the lower extraction rate of 10 scfm for 275 days, while the
second uses the average extraction rate of 25 scfm over the 950-day life of the project.

       A total-gas flame ionization detector was located on site during the operation of the vapor
extraction system. Vapor samples were also collected for GC analysis several times a week
during the first several months of operation and  slightly less frequently thereafter. The samples
were analyzed for identifiable hydrocarbon compounds in the gasoline, and also for a distribution
of mass with respect to boiling point ranges.

       VENT3D prints the composition of soil gas at extraction wells in order to provide one
measure of model performance.  A plot of the measured soil vapor TPH concentrations and the
simulated concentrations is shown in Figure 70. Included on the plot are simulated data for the
low flow rate (approximately 10 scfm) used for the first 275 days, and also from the second
simulation, which uses an extraction rate of 25 scfm over the 950-day SVE period.

                                          192

-------
Figure 69.
             0      10     20
                55as
                SCALE (ft)
                        +  Well used for vapor extraction

                        EH  Well or boring used for soil and vapor sample collection
Costa Mesa service station layout (Johnson et al., 1992) and.initial concentrations
used in the numerical simulations. TPH measurements were lognormally
distributed, so the initial TPH grid in soil was generated by kriging of the log TPH
values from 40 to 48 feet BGS.
                                           193

-------
     1000 n
                                   O Simulated values
                                                     On-site FID
                           GC analysis
                        i  i i i  i i i  i i i  i I i  i i i  i i i  i i  i i i  i i i  i i i  i | i  i i i «i i i  i i
       0.1
            0
200
   400          600
ELAPSED TIME (days)
1000
Figure 70.    Simulated and measured decline of TPH concentrations in vapor extracted from
             the Costa Mesa site. Measured values are represented by connected points.
             Simulated values are represented by large open circles.  The vapor stream was
             measured on site by on FID-equipped recorder, and by laboratory GC methods on
             collected samples.  The extraction rate was increased after 275 days from 10 to 30
             scfm.  Simulated results shown before 275 days are from a lower-flow simulation,
             while later results are from a higher flow simulation that uses an average flow rate
             over the life of the SVE project.
                                          194

-------
       Based on the large decrease in the TPH concentrations measured (and simulated) in
extracted vapor, one might expect that the total mass of hydrocarbons should be similarly
reduced. Figures 71 and 72 show the masses of selected compounds (and TPH) that are
predicted to remain in the soil by the two simulations. The plotted masses are summed for each
compound in all phases throughout the model domain.  Clearly, the more volatile compounds are
predicted to decrease sharply; however a large fraction of the TPH mass is predicted to remain in
the soil at this site. The amount of TPH mass simulated as removed in the 25 scfm model  at 900
days is approximately 84 percent.

       Johnson et al. (1992) assert that vapor composition is a more reliable indicator of
remediation completion than total vapor TPH concentrations.  A number of mechanisms can
cause lowered TPH concentrations in soil gas without being related to total mass removal.  An
example is TPH that is initially held in both clay layers and sand layers.  The coarse grained
material (and extracted soil gas) will become relatively clean without removing the clay-bound
TPH. Other examples are leakage and irregular plume geometry (Benson et al.,  1993). A total
mass decrease, however, should be accompanied by a stepwise depletion of lower boiling
compounds. As a result, the composition of extracted soil vapor should become progressively
"heavier" as a function of the percentage of initial mass removed.

       Figure 73 shows the measured and simulated soil vapor composition (by boiling point
ranges) for the first 275 days of venting. A comparison of the measured and predicted
compositions indicates that the two-dimensional model does not allocate a large enough fraction
of the TPH mass to non-advective areas. A three-dimensional model that more accurately
depicts the high concentrations  found in the silt and clay layer above the sand layer would make
more realistic predictions of eventual cleanup time.  It should be noted that after 950 days of
venting, two borings yielded samples from the lower sand that showed no detectable TPH, yet
samples from the silt and clay had TPH concentrations on the order of 10 to'1000s of mg/kg.

       The distribution of compounds left in soil may interest practitioners of S VE.
Confirmation borings are usually drilled in the area of highest initial concentration, even though
the process may leave pockets of contamination in unsuspected areas. Figures 74 and 75 show
maps of simulated residual benzene and TPH at different times in the remediation process. After
275 days of venting, the benzene concentrations are predicted to be near current detection limits
(approximately  0.005 mg/kg). After 950 days, the area near the extraction wells is relatively
devoid of TPH,  but the areas more  distant from the wells continue to have TPH concentrations in
the high 1000s of mg/kg.
                                          195

-------
          1E7
       §
       CO
       CO
                        200
     400         600
   ELAPSED TIME (days)
1000
                      TPH
-e- BENZENE -*- TOLUENE-"-XYLENES
Figure 71.    Residual soil masses of selected compounds in the 25-scfm model. Plotted values
            represent summed mass of the compound in all phases throughout the model
            domain.
                                       196

-------
            1E7
         §
         CO
            1EO
                        50
100       150       200
   ELAPSED TIME (days)
250
                                                                        300
                        TPH     -&-_ BENZENE-*"- TOLUENE-a-XYLENES |
Figure 72.    Residual soil masses of selected compounds in the 10-scfm model. Plotted values

            represent summed mass of the compound in all phases throughout the model
            domain.
                                       197

-------
               100%
                                      100       150        200
                                     ELAPSED TIME (days)
                                     250
50       100       150       200
         ELAPSED TIME (days)
                                                                   250
Figure 73.    A comparison of the simulated (upper figure) and measured (lower figure) change
             in the composition of extracted vapor. The concentration of all compounds in
             vapor extracted during the first 275 days are added according to boiling point
             ranges. The percentage of each range is given by its width on the graph. Johnson
             et al. (1992) suggest that the decline of the percentage of more volatile range is
             indicative of mass removal, whereas the TPH concentration decline may result
             from nonideal conditions. In light of the declining TPH concentrations, a
             comparison of the two plots shown here indicates that some fraction of the initial
             TPH mass is located outside of the near-well advection-dominated zone.

                                          198

-------
                                                                   O.OOL
                                NEW
                                TANK
                              LOCATIONS
                                              PUMP ISLANDS'
              0      10     20
                 351HH
                  SCALE (ft)
   Well used for vapor extraction

IB  Well or boring used for soil and vapor sample collection
Figure 74.    Simulated residual benzene (mg/kg) in soil after the initial 275 days of SVE at a
             steady rate of 10 scfm.

                                         199

-------
         PLANTER
                a as
   NEW
   TANK
LOCATIONS
                 MINI-MART
                                                                   PLANTER^
                                            \
                            7
                 PUMP ISLANDS
                                            PLANTER
             0      10      20
               5S555
                SCALE (ft)
         +  Well used for vapor extraction

         BJ  Well or boring used for soil and vapor sample collection
Figure 75.    Simulated residual TPH (mg/kg) in soil after 900 days of SVE at a steady rate of
             25 scfrn.
                                         200

-------
 8.3    VERONA WELL FIELD SUPERFUND SITE, BATTLE CREEK, MICHIGAN

       No information was available on any S VE modeling that may have occurred in
 conjunction with the design and implementation of the SVE system at the Verona Well Field
 site. However, a short discussion of the site has been included here because many interesting and
 useful topics are discussed in the Performance Evaluation Assessment Report (USEPA, 199Id),
 It discusses one of three source areas at the Verona Well Field Site, known as the Thomas
 Solvent Company's Raymond Road facility. The site was operated by the Thomas Solvent
 Company from approximately 1963 through 1984, where it purchased, stored, containerized,
 blended, transported, and sold virgin industrial solvents. They also transported, stored, and
 arranged for disposal or recycling of spent solvents.  The solvents handled included both
 chlorinated and non-chlorinated solvents.

       During the course of the remedial activities, a groundwater extraction system has
 removed approximately 14,000 pounds of priority pollutant VOCs from the aquifer Total VOC
 concentrations in the aquifer have decreased from a high of 19,000  ug/1 in 1987 to approximately
 800 ng/1 at the end of 1990.

       The SVE system was pilot tested in the fall of 1987, and full-scale operation began in the
 spring of 1988.  The SVE system removed greater than 45,000 pounds of VOCs from the
 subsurface through 1990.  It is of interest to note that the SVE system has removed a
 significantly larger mass of VOCs from the subsurface. The extraction rate of VOCs has
 decreased from an initial high of 1,000 pounds per day to less than five pounds per day. The
 SVE system operates over an area of approximately 36,000 square feet and throughout the entire
 vadose zone down to a depth of approximately 20 feet. Approximately 14 pore volumes of soil
 vapor are removed from the vadose zone each day.

       Based on information presented in USEPA (1991 d), the system was designed by
 performing pilot testing to determine extraction well flow rates and radii of influence.  During
 the pilot test, total VOC concentrations in the extracted gas ranged from 2,000 to 204,000 ug/l.
 The radius of influence at each extraction well was determined by measuring pressures in nearby
 piezometers. One of the largest radii of influence recorded was a vacuum of 1.25 inches of water
 (0.9969 atm) recorded at 60 feetfrom an extraction well. Approximately 3,000 pounds of VOCs
 were removed from the vadose zone during the 15-day pilot test.

       The SVE system had operated for almost 400 days by the end of 1990, and removed
 approximately 45,000 pounds of VOCs from the vadose zone. Figure 76 presents a plot of the
 cumulative mass of VOCs removed from the subsurface since startup. It is apparent from the
 figure that the cumulative mass  is asymptotically approaching a limiting value as the rate of
 removal from the subsurface declines with time.  The rate of removal of VOCs over time is
presented in Figure 77, and is shown to be  decreasing over time and approaching a small limiting
removal rate. The rate of removal has decreased from a high of over 1,000 pounds per day in the
 fall of 1987 to less than five pounds per day at the end of 1990. Contaminant concentrations in

                                         201

-------
          50000


          45000


          40000


          35000


          30000

  Cumulative
    VOCs   25000
   (pounds)

          20000


          15000


          10000 •


           5000
 I
  ^™
                                  TOTAL VOCS REMOVED THROUGH SOIL VAPOR EXTRACTION
                                        THOMAS SOLVENT RAYMOND ROAD SITE
                                            October 1987 to September 1990
	.£_
                        so        100       ISO        200        250
                                            Cumulative Days of Operation
                                                                        300
                                                                                  350
                                                                                           400
Figure 76.     Total mass of VOCs removed from the subsurface over time. Note how the plot
               is reaching a limiting, asymptotic value (from USEPA, 199 Id).
                                               202

-------
                                                 POUNDS OF VOCS REMOVED
                                              THROUGH SOIL VAPOR EXTRACTION
                                              THOMAS SOLVENT RAYMOND ROAD
                                                October 1987 lo September 1990
   Total VOCs Removed
        Per Day
      (pounds/day)
                    1200
                    1000
                     800
600
                    400
                    200
                       Ocl Nov Jan Mar May Jul  Sep Nov Jan Mar May Jul  Sep Nov Jan Mar May Jul
                       87  87  88  88  88  88  88  88  89  89  89  89  89  89  90   90   90  90
                                                                       Sep
                                                                        90
Figure 77.     VOC removal rate versus time.  Note how the removal rate is high during early
               time, but diminishes during later time to a limiting value.
                                             203

-------
the extracted gas have decreased from a high of 23,000 jig/1 in April of 1988 to 38 ug/1 at the end
of 1990.

       These figures present a classic example of rate-limited removal of VOCs from the
subsurface.  As is often common during SVE operations, removal rates are high initially (Stage I)
and tend to drop off over tune and tend toward some asymptotic value (Stage III). The reasons
for this are discussed in detail elsewhere in this document, but are related to the availability of
transportable contaminant mass, which declines over time during the removal operation.  During
the initial stages of removal, readily available VOC vapor present in large, accessible pores in the
subsurface is removed first (the Stage I portion of the removal operation). However, as time
goes on and the readily accessible portion of the contaminant mass is removed, there still remains
a large portion of the contaminant whose removal is limited by diffusive processes (the Stage III
portion of the removal operation). These rate-limiting processes include contaminant
vaporization from the dissolved phase and free-phase NAPL trapped in inaccessible, dead-end
pores which are not swept by the extraction air flow stream.

       Results from soil borings performed after  the SVE system had been in operation for some
time indicated that the areas with the highest levels of residual contamination were those areas
near the capillary fringe where free-phase NAPL had been observed. The method of kriging was
used to estimate the mass of contaminants left in the soil after SVE via dividing the site into
discrete blocks and statistically estimating the contaminant mass in each block. The results of the
kriging analysis indicated that approximately 90% of the initial contamination had been
removed.

       A number of column studies were completed in order to qualitatively study some of the
aspects of vapor transport. In the first experiment, a soil column was spiked with a  1:1 by
volume mixture of TCE and PCE. A vacuum was then applied to the column and samples of the
extracted gas were analyzed at intervals throughout the test. The results (Figure 78) indicated
that TCE was removed from the system more rapidly due to its higher vapor pressure. Figure 79
shows the effect of Raoult's law on the system. Raoult's law states that a constituent's vapor
pressure is directly proportional to its mole fraction in a solution. If the mole fraction of a
constituent is reduced, then the vapor pressure of that constituent is reduced and less of it is
available in the vapor phase for removal via vapor extraction.  Because mass of TCE is lost
initially from the system (at time < 40 minutes), the mole fraction of PCE increases, and thus its
vapor pressure and concentration in the vapor phase increase (see Figure 79). However, once the
mole fraction of PCE stabilizes, the concentration of PCE begins to decrease as well, as the PCE
is removed from the system.

       In a second experiment, the soil column was spiked with TCE and PCE, and a small
amount of decane, a low-volatility NAPL, was added. By adding decane to the mixture, the mole
fractions of TCE and PCE were further reduced.  The results (Figure 80) indicate that by
reducing the mole fractions of TCE and PCE in the system, their vapor pressures are decreased,
and hence their volatility and rate of removal are reduced. In this experiment, it takes much

                                          204

-------
              TCE PCE
                                               PCE
                                                                            PCE
                                           TCE
                                                                        TCE
                VJ
              1Hour
                                      1 Hour 30 Minutes
                                                                  2 Hours 15 Minutes
Figure 78.    Chromatograms of extracted gas from the TCE/PCE column test. Note how the
             more volatile TCE is removed from the system more quickly (from USEPA
             1991d).


                                       205

-------
400

350

sr 300
i .
j* 250
°
i 200
3 150
i
6 100
Q
§•
§ 50
0

y« — • 	 TCE
\ • -A 	 PCE
\.

t \
|_ \


f
\ A
- I A->--^
I XX "-.A
I X A ^
AJ\-A- "^ N

k .^ \ fc
r4 \ \%
i l V^^. -i '- m '_ ,' '. w **4^l A'
— .A. * ** AM A A JAM 4AA 4 Af\ H eft HOA '"'t*
                   20     40    60     80    100    120
                                     Time (minutes)
Figure 79.    Raoult's law effects on the TCE/PCE system. As TCE is removed from the
             system during early time, the mole fraction of PCE increases, causing an increase
             in concentration. The concentration of PCE in extracted gas begins to stabilize
             after the TCE has been removed from the system and the mole fraction of PCE
             has stabilized. The concentration of PCE then decreases as it is removed from the
             system (from USEPA, 199Id).
                                        206

-------


1
o
2
o
3
a
a
5
i
a




HU
70
60

50
40

30
20

10
rt

—• »•— - TCE
""•Vj • -A- - PGE
- \
\
k
-* \
^4. \ ^ ^ _ ^ 4^
*^L— % "^ """*"*•••. A A
*>JI •* A ***>>
\ X
\ \

i T"*"^ — •—•-*-•_.«_ ^ jt^j
0 100 200 300 400
                                         Time (minutes)
Figure 80.    Results of a column experiment where a third constituent, decane, was added to
             the TCE/PCE mixture. The addition of a third constituent further reduced the
             mole fractions of TCE and PCE, and hence their vapor pressures were reduced,
             and volatilization and removal from the system occurred more slowly (from
             USEPA, 199 Id).

                                        . 207

-------
longer for the same mass of TCE and PCE to be removed from the soil column. After 14 hours
detectable levels of TCE and PCE remained in the air stream, although over 90% of the original
mass had been removed. In the first experiment, after approximately three hours TCE and PCE
levels were below detectable limits. These results indicate that the presence of additional
constituents in a mixture reduces the volatility of each constituent in the mixture, as described by
Raoult's law, and hence reduces its removal rate. Additional discussion on the column studies,
column study methodology, and the effects of rate-limiting processes on the SVE operation can
be found in USEPA (1991d).
                                         208

-------
                                      SECTION 9   '
                                    REFERENCES

Adamson, A.W., 1976. Physical Chemistry of Surfaces, John Wiley & Sons, Inc., New York.

Anderson, M.P., 1979. Using models to simulate the movement of contaminants through
       ground-water flow systems, Critical Reviews in Environmental Control, 9:97-156

Anderson, M.D. and W.W. Woessner, 1992. Applied Groundwater Modeling, Simulation of
       Flow andAdvective Transport, Academic Press, 381 p.

API, American Petroleum Institute, 1991.  Technological limits of groundwater remediation: A
       statistical evaluation method, API Publication Number 4543, Washington, DC.

Armstrong, J.E., E.G. Frind, and R.D. McClellan, 1994. Nonequilibrium mass transfer between
       the vapor, aqueous, and solid phases in unsaturated soils during vapor extraction, Water
       Resources Research, 30(2):355-368.

Arya, L.M. and J.F. Parris, 1981, A physicoempirical model to predict soil moisture
       characteristics from particle size distribution and bulk density data, Soil Science Soc. Am.
       J., 45:1023-1030.

Baehr, A.L., 1987.  Selective transport of hydrocarbons  in the unsaturated zone due to aqueous
       and vapor phase partitioning, Water Resources Research, 23(10):1926-1938.

Baehr, A.L. and M.Y. Corapcioglu, 1987.  A compositional multiphase model for groundwater
       contamination by petroleum products, 2. Numerical solution, Water Resources Research,
       23(1):201-214.

Baehr, A.L. and G.E. Hoag, 1988. A modeling and experimental investigation of induced
       venting of gasoline-contaminated soils, in Soil Contaminated by Petroleum, John Wiley
       and Sons, New York, pp. 113 -123.

Baehr, A.L. and M.H. Hult, 1991. Evaluation of unsaturated zone air permeability through
       pneumatic tests, Water Resources Research, 27(10):2605-2617.

Baehr, A.L., C. Welty, C.J. Joss, J. Rumbaugh, and S. Potter, 1993.  Enhancements to AIR3D:
       A three-dimensional model of air flow in the unsaturated zone, American Petroleum
       Institute, Washington, DC.

Baehr, A.L., G.E. Hoag, and M.C. Marley, 1989. Removing volatile contaminants from the
       unsaturated zone by inducing advective air-phase transport, Journal of Contaminant
       Hydrology, 4:1-26.

                                         209

-------
 Balsam Environmental Consultants, 1992. Vapor extraction treatability study, Union Chemical
       Company site, South Hope, Maine, USEPA Region I, Boston, MA.

 Banerjee, S., 1984. Solubility of organic mixtures in water, Environmental Science &
       Technology, 18(8):587-591.

 Barrow, G.M., 1961.  Physical Chemistry, McGraw-Hill Book Co., New York.

 Bear, J., 1979. Hydraulics of Ground-water, McGraw-Hill Book Co., New York, 569 p.

 Bear, J., 1972. Dynamics of Fluids in Porous Media, Elsevier Science Publishers, The
       Netherlands, 764 p.

 Bear, J. and A. Verruijt, 1987. Modeling Groundwater Flow and Pollution, Kluwer Academic
       Publishers, Norwell, MA, 414 p.

 Beckett, G. and D. Huntley, 1994. Characterization of flow parameters controlling soil vapor
       extraction, Ground Water, 32(2):239-247.

 Bedient, P.B., F.W. Schwartz, and H.S. Rifai, 1993. Hydrologic design for groundwater
       pollution control, in Handbook of Hydrology, D.R. Maidment, ed., McGraw Hill, New
       York.

 Benson, D., 1995. Written communication.

 Benson, D.A., 1994. User's manual to VENT2D: A multi-compound vapor transport and phase
       distribution model, David A. Benson, Reno, NV.

 Benson, D.A., D. Huntley, and P.C. Johnson, 1994. Reply to the discussion by Gary R. Walter
       of "Modeling vapor extraction and general transport in the presence of NAPL  mixtures
       andnonideal conditions," Ground Water, 32(1):148-150.

 Benson, D.A., D. Huntley, and P.C. Johnson, 1993. Modeling vapor extraction and general
       transport in the presence of NAPL mixtures and nonideal conditions, Ground  Water,
       31(3):437-445.

Bird, R.B., W.E. Stewart, and E.N. Lightfoot, 1960. Transport Phenomena, John Wiley & Sons,
       Inc., New York, 780 p.

Blandford, T.N., P.S. Huyakorn, and Y.S. Wu, 1993. Well head protection area delineation code,
       FOS 41 PC, Version 2.2 (September), International Ground Water Modeling Center,
       Golden, CO.
                                         210

-------
Brooks, R.H. and A.T. Corey, 1964. Hydraulic properties of porous media, Hydrology Paper
       No. 3, Colorado State University, Fort Collins, CO, 27 p.

Brown, R.A. and R. Fraxedas, 1991. Air sparging-extending volatilization to contaminated
       aquifers, Symposium on Soil Venting (April 29-May 1), Robert S. Kerr Environmental
       Research Laboratory, Houston, XX.

Bruce, R.R. and RJ. Luxmo'ore, 1986. Water retention: Field methods, in Methods of Soil
       Analysis, Part 1-Physical and Mineralogical Methods, second edition, A. Klute, Ed.,
       Number 9 (Part 1) in the series Agronomy, American Society of Agronomy, Inc.,
       Madison, WI, p.663-686.

Bruell, C.J., and G.E. Hoag, 1986.  The diffusion of gasoline-range hydrocarbon vapores in
       porous media, experimental methodologies, in NWWA/API Conf. on Petroleum
       Hydrocarbons and Organic Chemicals in Ground Water, Prevention, Detection and
       Restoration, NWWA, OH, pp. 420-443.

Brusseau, M.L., 1991.  Transport of organic chemicals by gas advection in structured or
       heterogeneous porous media: Development of a model and application to column
       experiments, Water Resources Research, 27(12):3189-3199.

Bulman T.L., M. Newland, and A. Wester, 1993. In-situ bioventing of a diesel fuel spill,
       Hydrological Sciences Journal, 38(4):297-308.

Burkhard, L.P., A.W. Andren, and D.E. Armstrong, 1985. Estimation of vapor pressures for
       polychlorinated biphenyls:  A comparison of eleven predictive methods, Environmental
       Science and Technology, 19:500-507.

Buscheck, T.E. and T.R.  Peargin, 1991. Summary of a nation-wide vapor extraction system
       performance study, Proceedings of Petroleum Hydrocarbons and Organic Chemicals in
       Ground Water Conference (November), National Water Well Association, Houston, TX.

Buxton, D.S., and R.E. Green, 1987. Desorption and teachability of DBCP residues in soils, in
       Agronomy Abstracts, American Society of Agronomy, Madison, WI, p. 167.

Carsel, R.F. and R.S. Parrish, 1988. Developing joint probability distributions of soil and water
       retention characteristics, Water Resources Research, 24(5):755-769.

Charbeneau, R.J. and D.E. Daniels, 1993. Contaminant transport in unsaturated flow, in
       Handbook of Hydrology, McGraw-Hill Book Co., New York.
                                          211

-------
Chiou, C.T., R.L. Malcolm, T.I. Brinton, and D.E. Kile, 1986.  Water solubility enhancement of
       some organic pollutants and pesticides by dissolved humic and fulvic acids,
       Environmental Science & Technology, 20(5):502-508.

Chiou, C.T., V.H. Freed, D.W. Schmedding, and R.L. Kohnert, 1977. Partition coefficient and
       bioaccumulation of selected organic compounds, Environmental Science & Technology,
       11:475-478.

Chiou, C.T., andT.D. Shoup, 1985.  Environmental Science and Technology, 19:1196-2000.

Clark, A.N., M.M Megehee, and D.J. Wilson, 1992. Soil cleanup by in situ aeration: Effect of
       departures from Darcy's law on soil vapor extraction, manuscript.

Cohen, R.M. and J.W. Mercer, 1993. DNAPLSite Evaluation, C.K. Smoley, Boca Raton, FL,

Coyle, J.A., H.J. Borchers, Jr., and R.J. Miltner, 1985.  Control of volatile organic chemicals in
       groundwater by in-well aeration, Proceedings of the American Water Works Association
      Annual Conference, Washington, DC, p. 1101-1114.

CRTC, Chevron Research and Technology Company, 1992. Spreadsheet SVE model,
       Richmond, CA.                             '

Delshad, M. and G.A. Pope, 1989. Comparison of the three-phase oil relative permeability
       models, Transport in Porous Media, 4(l):59-83.

Dietrich, J.K. and P.B. Bonder, 1976. Three-phase oil relative permeability problem in reservoir
       simulation, SPE 6044, Presented at the 51st Annual Meeting of the SPE (October 3-6),
      New Orleans, LA.

Eganhouse, R.P. and J.A Calder, 1973. The solubility of medium molecular weight aromatic
      hydrocarbons and the effects of hydrocarbon col-solutes and salinity, Geochimica et
       CosmochimicaActa,VoL37.

Environmental Systems Research Institute, Inc. (ESRI), 1992.  Understanding GIS, The Arc/Info
      Method, Redlands, CA.

Environmental Systems and Technologies, Inc. (ES&T), 1994.  VENTING: A program for
       estimating hydrocarbon recovery from soil vacuum extraction systems, Version 3.01.
      ES&T,  Inc., Blacksburg, VA.

Falta, R.W., 1995. A program for analyzing transient and steady-state soil gas pump tests,
       submitted to Ground Water.
                                         212

-------
Falta, R. W., 1993.  Analysis of soil gas pump tests, Proceedings of Meeting the Challenge,
       Environmental Remediation Conference, Augusta, GA, United States Department of
       Energy, 1:441-447.

Falta, R.W., K. Pruess, and D.A. Chesnut, 1993. Modeling advective contaminant transport
       during soil vapor extraction, Ground Water, 31(6):1011-1020.

Falta, R.W., I. Javandel, K. Pruess, and P. A. Witherspoon, 1989. Density-driven flow of gas in
       the unsaturated zone due to the evaporation of volatile organic compounds, Water
       Resources Research, 28(2):323-335.

Farmer, W.J., M.S. Yang, J. Letey, and W.F. Spencer, 1980.  Hexachlorobenzene:  Its vapor
       pressure and vapor phase diffusion in soil, Soil Science Soc. Am. J., 44:676*680.

Faust, C.R., J.H. Guswa, and J.W. Mercer,  1989. Simulation of three-dimensional flow of
       immiscible fluids within and below the unsaturated zone, Water Resources Research,
       25(12):2449-2464.

Payers, F.J. and J.P. Matthews, 1984. Evaluation of normalized Stone's methods for estimating
       three-phase relative permeabilities, SPE Journal, 24:224-232.
                                   '* --;

Feenstra, S., D.M. Mackay, and J.A. Cherry, 1991. A method of assessing residual NAPL based
       on organic chemical concentrations in soil samples, Groundwater Monitoring Review,
       11(2): 128-136.

Fitts, C.R.,  1989. Simple  analytic functions for modeling three-dimensional flow in layered
       aquifers, Water Resources Research, 25(5):943-948.

Freeze, R.A. and J.A. Cherry, 1979.  Groundwater, Prentice-Hall, New York, 604 p.

Frick, T., Ed., 1962. Petroleum Production Handbook, 2nded., Society of Petroleum Engineers
       of AIME, Dallas, TX.

Fuller, E.N., Schettler, P.O., and J.C. Giddings, 1966.  A new method for prediction of binary
       gas-phase diffusion coefficient, Ind. Eng. Chem., 58:19-27.

Gardner, W.H., 1986. Water content, in Methods of Soil Analysis, Parti-Physical and
       Mineralogical Methods, second edition, A. Klute, Ed., Number 9 (Part 1) in the series
       Agronomy,  American Society of Agronomy, Inc., Madison, WI, p. 493-544.

Gelhar, L.W., 1986. Stochastic subsurface hydrology from theory to applications, Water
       Resources Research, 22:1358-1458.
                                          213

-------
 Gelhar, L.W., C. Welty, and K.R. Rehfeldt, 1992.  A critical review of data on field-scale
       dispersion in aquifers, Water Resources Research, 28(7): 1955-1974.

 GeoTrans, Inc., 1992. Documentation and user's guide:  MODMAN, an optimization module for
       MODFLOW, Version 2.1.  GeoTrans, Inc., Sterling, VA.

 Geraghty and Miller, Inc., 1989. AQTESOLV Aquifer Test Solver, Version 1.10, Geraghty and
       Miller Modeling Group, Reston, VA.

 Gierke, J.S., N.J. Hutzler, and D.B. McKenzie,  1992. Vapor transport in unsaturated soil
       columns: Implications for vapor extraction, Water Resources Research, 28(2):323-335.

 Gierke, J.S., N.J. Hutzler, and D.B. McKenzie,  1990. Experimental and model studies of the
       mechanisms influencing vapor extraction performance, Proceedings of Petroleum
      • Hydrocarbons and Organic Chemicals in Groundwater: Prevention, Detection, and
       Restoration, National Water Well Association, American Petroleum Institute, Houston,
       TX, p. 325-338.   •

 Groves, F.R., Jr., 1988. Effect of cosolvents on the solubility of hydrocarbons in water,
       Environmental Science & Technology, 22(3):282-286.                           i

 Gvirtzman, H. and S.M. Gorelick, 1992. The concept of in-situ vapor stripping for removing
       VOCs from groundwater, Transport in Porous Media, 8:71-92.

 Haider, C.A., G.S. Van Gorp, N.S. Hatoum, andT.M. Warne, 1986.  Gasoline vapor exposures,
       Part I: Characteristics of workplace exposures. American Industrial Hygiene Association
       Journal, 47(3): 164-172.

 Hantush, M.S., 1964. Hydraulics of wells, in Advances in Hydroscience, V.T. Chow, Ed.,
       Academic Press, New York, 1, p. 281-442.

 Hantush, M.S. and C.E. Jacob, 1955. Nonsteady radial flow in an infinite leaky aquifer,
       Transactions, American Geophysical Union, 36:95-100.

 Heiba, A. A., M. Sahimi, L.E. Scriven, and N.T. Davis, 1982. Percolation theory of two-phase
       relative permeability, SPE 11015.

Herrling, B. and W.  Buermann, 1990. A new method for in-situ remediation of volatile
       contaminants in groundwater - numerical simulation of the flow regime, in
       Computational Methods in Subsurface Hydrology, G. Gambolati, A. Rinaldo, C.A.
       Brebbia, W.G. Gray, and G.F. Finder, Eds., Springer, Berlin, p. 299-304.
                                         214

-------
Herrling, B., W. Buermann and J. Stamm, 1990. In-situ remediation of volatile contaminants in
       ground-water by a new system of "Underpressure-Vaporizer-Wells," Presented at the
       International Association of Hydrogeologists Conference on Subsurface Contamination
       by Immiscible Fluids (April 18-20), Calgary, Alberta.

Herrling, B., J. Stamm, and W. Buermann, 1991. Hydraulic circulation system for In-situ
       bioreclamation and/or in-situ remediation of strippable contamination, Proceedings ofln-
       Situ and On-Site Bioreclamation International Symposium (March 19-21), San Diego,
       CA.

Hillel, D., 1980.  Fundamentals of Soil Physics, Harcourt, Brace, and Jovanovich, New York.
                                                                               f
Hiller, D. and H. Gudemann, 1989. Analysis of vapor extraction data from applications in
       Europe, Proceedings of Third International Conference on New Frontiers for Hazardous
       Waste Management (September 10-13), EPA/600/8-89/072, Pittsburgh, PA.  '  " '

Hills, R.G., K.A. Fisher, M.R. Kirkland, and P.J. Wierenga, 1994.  Application of flux-;cori
-------
 Huyakorn, P.S. and G.F. Finder, 1983. Computational Methods in Subsurface Flow, Academic
        Press, New York.

 Hazardous Waste Consultant, 1992.  In situ steam-enhanced extraction process, in The
        Hazardous Waste Consultant, 10(4):1.16-1.17.

 Isherwood, D., 1981.  Geoscience Data Base Handbook for Modeling a Nuclear Waste
        Repository, Vol. 1, Lawrence Livermore Laboratory Report UCRL-52719, Livermore
        CA.

 Istok, J., 1989. Groundwater Modeling by the Finite Element Method, American Geophysical
        Union Water Resources Monograph 13,495 p.

 Johnson, P.C., 1991. Hyperventilate Users Manual, Shell Oil Company.

 Johnson, P.C., and R.A. Ettinger, 1994. Considerations for the design of in-situ vapor extraction
        systems:  Radius of influence vs. zone of remediation, Ground Water Monitoring Review
        14(3):123-128.

 Johnson, P.C., M.B. Hertz, and B.L. Byers, 1990c. Estimates for hydrocarbon vapor emissions
       resulting from service station remediations and buried gasoline-contaminated soils, in
       Petroleum Contaminated Soils, Vol. 3, P.T. Kusteck and E.J. Calabrese, Eds,, Lewis
       Publishers, Chelsea, MI.

 Johnson, P.C., M.W. Kemblowski, and J.D. Colthart, 1990a. Quantitative analysis for the
       cleanup of hydrocarbon-contaminated soils by in-situ soil venting, Ground Water
       28(3):413-429.

 Johnson, P.C., C.C. Stanley, D.L. Byers, D.A. Benson, and M.A. Acton,  1992. Soil venting at a
       California site: Field data reconciled with theory, Hydrocarbon Contaminated Soils and
       Ground-water, Vol. 1, Lewis Publishers, Chelsea, MI.

 Johnson, P.C., C.C. Stanley, M.W.  Kemblowski, D.L. Byers, and J.D. Colthart, 1990b.  A
       practical approach to the design, operation and monitoring of in-situ soil-venting systems,
       Ground Water Monitoring Review, Spring p. 159-178.

Kaluarachchi, J.J. and J.C. Parker, 1990.  Modeling multicomponent organic chemical transport  .
       in three phase porous media, Journal of Contaminant Hydrology, 5:349-379.

Kaluarachchi, J.J. and J.C. Parker, 1989.  An efficient finite element model for modeling
       multiphase flow in porous media, Water Resources Research, 25:43-54.
                                         216

-------
Karickhoff, S.W., 1980.  Sorption kinetics of hydrophobic pollutants in natural sediments, in
       Contaminants and Sediments, R.A. Baker, Ed., Vol. 2, Ann Arbor Science Publishers,
       Ann Arbor, MI, p. 193-205.

Karickhoff, S.W., 1980.  Sorption kinetics of hydrophobic pollutants in natural sediments, in
       Contaminants and Sediments, Vol. 2, R.A. Baker, ed., Ann Arbor Science Publishers,
       Ann Arbor, MI, pp. 193-205.

Karickhoff, S.W., D.S. Brown, and T.A. Scott, 1979. Sorption of hydrophobic pollutants on
       natural sediments, Water Res., 13:241-248.

Kenaga, E.E. and C.A.I. Goring, 1980. Relationship between water solubility, soil sorption,
       octanol-water partitioning, and concentration of chemicals in biota, Aquatic Toxicology,
       ASTM STP 707, American Society for Testing and Materials, Philadelphia, PA, p.
       78-115.

Klinkenberg, L.J., 1941. The permeability of porous media to liquids and gases, American
       Petroleum Institute Drilling and Production Practice, p. 200-213.       .

Klute, A., 1986.  Water retention: Laboratory methods, in Methods of Soil Analysis, Part 1, A.
      • Klute, Ed., American Society of Agronomy, Madison, WI, p. 635-686.

Kreamer, D.K., 1986. Chapter Three of Processes Affecting Subsurface Transport of Leaking
       Underground Tank Fluids, Desert Research Institute, University of Nevada System,
       Water Resources Center.

Kresge, M.W. and M.F. Dacey, 1991. An evaluation of in-situ groundwater aeration,
       Proceedings ofHazMat International '91 Conference, Atlantic City, NJ.

Krishnayya, A.V., M.J. O'Conner, J.G. Agar, and R.D. King, 1988. Vapor extraction system
       factors affecting their design and performance, Proceedings of Petroleum Hydrocarbons
       and Organic Chemicals in Groundwater:  Prevention, Detection and Restoration,
       National Water Well Association, Houston, TX, p. 547-567.

Krol, E., 1992. The Whole Internet Users Guide and Catalog, O'Reilly and Associates, 376 p.

Kuo, J.F., J.M. Montgomery, and P.J. Yang, 1990. A two-dimensional model for estimating
       radius of influence of a soil venting process, Presented at HAZMACON 90  (April 17-19),
       Anaheim, CA.

Lee, M.D. and C.M. Swindoll, 1993.  Bioventing for in-situ remediation, Hydrological Sciences
       Journal, 38(4):273-282.
                                          217

-------
 Lenhard, R.J., 1992. Measurement and modeling of three-phase saturation-pressure hysteresis,
       Journal of Contaminant Hydrology, 9:243-269.

 Lenhard, R.J., J.C. Parker, and S. Mishra, 1989. On the correspondence between Brooks-Corey
       and van Genuchten models, Journal of Irrigation and Drainage Engineering, 115(4);744-


 Lingineni, S., and V.K. Dhir, 1992. Modeling of soil venting processes to remediate unsaturated
       soils, J. Environmental Engineering, ASCE, 118:135-152,

 Lord, A.E., Jr., R.M. Koemer, V.P. Murphy, and J.E. Brugger, 1987. In-situ, vacuum-assisted
       steam stripping of contaminants from soil, Proceedings ofSuperfund '87, 8th National
       Conference on Management of Uncontrolled Hazardous Waste Sites, Hazardous
       Materials Control Research Institute, Silver Spring, MD, p. 390-395.

 Lucius, I.E., G.R. Olhoeft, P.L. Hill, and S.K. Duke, 1990. Properties and hazards of 108
       selected substances, USGS Open-File Report 90-408, 559 p.

 Luckner, C.A., M.T. van Genuchten, and D.R. Nielsen, 1989. A consistent set of parametric
       models for two-phase flow of immiscible fluids in the subsurface, Water Resources
       Research, 25(10):2187-2193.

 Lyman, W.J., W.F. Reehl, and D.H. Rosenblatt,  1982. Environmental behavior of organic
       compounds, in Handbook of Chemical Property Estimation Methods, McGraw-Hill Book
       Co., New York.

 Mackay, D., A. Bobra, W.Y. shiu, and S.H. Yalkowsky, 1982. Vapor pressure correlations for
       low-volatility environmental chemicals, Environmental Science and Tech., 16:645-649.

 Marley, M.C., P.E. Nangeroni, B.L. Cliff, and J.D.  Polonsky, 1990.  Air flow modeling for in-
       situ evaluation of soil properties and engineered vapor extraction system design,
       Presented at HAZMACON 90 (April 17-19), Anaheim, CA.

 Marley, M.C., D.J. Hazebrouck, and M.T. Walsh, 1992a. The application of in-situ air sparging
       as an innovative soils and ground water remediation technology, Ground Water
       Monitoring Review. Vol. 12, No. 2, p.  137-145.

Marley, M.C., F. Li, and S. Magee, 1992b.  The application of a 3-D model in the design of air
       sparging systems, Proceedings of the NWWA Focus on Eastern Ground Water Issues, p.
       377-392. Need 1992 - Are either one of these what we need?
                                         218

-------
Marley, M.C. and G.E. Hoag, 1984. Induced soil venting for recovery/restoration of gasoline
       hydrocarbons in the vadose zone, Proceedings of Petroleum Hydrocarbons and Organic
       Chemicals in Ground Water, National Water Well Association-American Petroleum
       Institute, Houston, TX, p. 473-503.

Massman, J.W., 1989.  Applying groundwater flow models in vapor extraction system design,
       Journal of Environmental Engineering, 115(1):129-149.

McClellan, R.D. and R.W. Gillham, 1990. Vacuum extraction of trichloroethene from the
       vadose zone, Presented at the International Association of Hydrogeologists Conference
       on Subsurface Contamination by Immiscible Fluids (April 18-20), Calgary, Alberta.

McDonald, M.G., and A.W. Harbaugh, 1984. A modular three-dimensional finite-difference
       ground-water flow model, U.S. Geological Survey, Reston, VA.

Mendoza, C.A. and E.G. Frind, 1990. Advective-dispersive transport of dense organic vapors in
       the unsaturated zone 1. Model development, and 2. Sensitivity analysis, Water
       Resources Research, 26(3):379-398.

Mercer, J.M., S.D. Thomas, and B. Ross, 1982.  Parameters and variables appearing in repository
       siting models, NUREG/CR-3066, U.S. Nuclear Regulatory Commission.

Mercer, J.W. and C.R.  Faust, 1992. Ground-Water Modeling, National Ground Water
       Association, Dublin, OH, 60 p.

Mercer, J.W. and R.K. Waddell,  1993. Contaminant transport in groundwater, in Handbook of
       Hydrology, D.R. Maidment, Ed., McGraw-Hill, Inc., New York.

Millington, R.J., and J.M. Quirk, 1961.  Permeability of porous solids, Trans. Faraday Society,
       57:1200-1207.
          *
Mishra, S., J.C. Parker, and N. Singhal, 1989. Estimation of soil hydraulic properties and their
       uncertainty from particle  size distribution data, J. Hydrology, 108:1-18.

Montgomery, J.H., 1991.  Groundwater Chemicals Desk Reference, Volume II, Lewis Publishers,
       Chelsea, MI, 944 p.

Montgomery, J.H. and L.M. Welkom, 1990. Groundwater Chemicals Desk Reference, Lewis
       Publishers, Chelsea, MI, 640 p.

Moore, W.S., 1972. Physical Chemistry, 4th ed., Prentice-Hall.
                                          219

-------
 Morris, D.A. and A.I. and Johnson, 1967. Summary of hydrologic and physical properties of
       rock and soil materials as analyzed by the hydrologic laboratory of the U.S. Geological
       Survey, U.S. Geological Survey Water Supply Paper 1839-D.

 Muskat, M. and H.G. Botset, 1931. Flow of gas through porous materials, Physics, Vol. 1, p. 27-
       47.

 National Research Council, 1990. Ground Water Models, Scientific and Regulatory
       Applications, National Academy of Sciences, Washington, DC, 302 p.

 National Research Council, 1994. Alternatives for Ground Water Cleanup, National Academy
       Press, Washington, DC, 315 p.

 Nirmalakhandan, N.N. and R.E. Speece, 1988. Prediction of aqueous solubility of organic
       chemicals based on molecular structure, Environmental Science & Technology,
       22(3):328-338.

 Norris, Hinchee, Brown, McCarty, Semprini, Wilson, Kampbell, Reinhard, Bouwer, Border,
       Vogel, Thomas, and Ward, 1994. Handbook of Bioremediation, Robert S. Kerr
       Environmental Research Laboratory, Lewis Publishers, Boca Raton, FL.
                                                   »
 Nyer, E.K., R.W. Macneal, and D.C. Schafer, 1994.  VES design:  Using simple or sophisticated
       design methods, Ground Water Monitoring & Remediation, Summer, p.  101-104.

 Panday, S., Y.S. Wu, and P.S. Huyakorn, 1994. A three-dimensional flow model for assessing
       NAPL contamination in porous and fractured media, 2. Porous medium simulation
       examples, Journal of Contaminant Hydrology, 16(2):131-156.

 Parker, J.C., 1989. Multiphase flow and transport in porous media, Reviews of Geophysics,
       27(3):311-328.
                                                                          &
 Parker, J.C., R.J. Lenhard, and T. Kuppusamy, 1987. A parametric model for constitutive
       properties governing multiphase flow in porous media, Water Resources Research,
       23(4):618-624.

Parker, J.C. and A.J. Valocchi, 1986.  Constraints on the validity of equilibrium and first-order
       kinetic transport models in structured soils, Water Resources Research, 22(3):399-407.

Peaceman, D.W., 1977. Fundamentals of Numerical Reservoir Simulation, Elsevier Science
       Publishers, The Netherlands, 176 p.

Perry, R.H. and D.W. Green, eds., 1984. Perry's Chemical Engineers'Handbook, Sixth Edition,
       McGraw-Hill Book Co., New York.

                                         220

-------
Finder, G.F. and W.G. Gray, 1977.  Finite Element Simulation in Surface and Subsurface
       Hydrology, Academic Press, New York.

Preslo, L.M. and D.W. Stoner, 1991. Overall philosophy of site investigation, in Practical
       Handbook of Ground-Water Monitoring, D.M. Neilson, Ed., Lewis Publishers, Chelsea,
       MI, p. 69-95.

Rao, P.S.C., L.S. Lee, and A.L. Wood, 1991. Solubility, sorption and transport of hydrophobic
       organic chemicals in complex mixtures, USEPA Environmental Research Brief, Robert
       S. Kerr Environmental Research Laboratory, Ada, OK, EPA/600/M-91/009, 14 p.

Rathfelder, K., W.W-G. Yeh, and D. Mackay, 1991.  Mathematical simulation of soil vapor
       extraction systems: Model development and numerical examples, Journal of      •• .  -
       Contaminant Hydrology, 8:263-297.

Rawls, W.J., L.R. Ahuja, D.L. Brakensiek, and A. Shirmohammadi, 1993.  Infiltration and soil
       water movement, in Handbook of Hydrology, D.R. Maidment, Ed., McGraw-Hill, Inc.,
       New York.             .             ,                  .

Rawls, W.J. and D.L. Brakensiek, 1982.  Estimating  soil water retention from soil properties, J.
       Irrig.Drain.Eng.,l08(IR2):166-l7l.

Reid, R.C., J.M. Prausnitz, and T.K. Sherwood, The Properties of Gases and Liquids, 3rd ed.,
       McGraw-Hill, New York, 688 p.

Rodeck, S.A., B.A. DeVantier, and B.M. Das, 1994.  Air permeability measurement for soil at
       low and high pressure, J. Envir. Eng., ASCE, 120(5): 1337-1343.

Rodriguez-Maroto, J.M. and D.J. Wilson, 1988. Soil cleanup by in-situ aeration, VII. High-
       speed modeling of diffusion kinetics, Separation Science and Technology, 26(6): 743-760.

Rossi, S.S. and W.H. Thomas, 1981. Solubility behavior of three aromatic hydrocarbons in
       distilled water and natural seawater, Environmental Science & Technology, 15:715-716.

Rumbaugh, J.O., 1993.  ModelCad user's documentation, Geraghty and Miller, Reston, VA.

Sabadell, G.P., J.J. Eisenbeis, D. van Zyl, and O.K. Sanada, 1988. CSUGAS: Flowfield model
       for in-situ volatilization of organic compounds in soils, Argonne National Laboratory,
       Argonne, IL.

Schwarzenbach, R.P., P.M. Gschwend, and D.M. Imboden, 1993. Environmental Organic
       Chemistry, John Wiley & Sons, Inc., New York, 681 p.
                                        ,  221

-------
 Shan, C.R., W. Falta, and I. Javandel, 1992. Analytical solutions for steady state gas flow to a
        soil vapor extraction well, Water Resources Research, 28(4): 1105-1120.

 Silka, L.R. and D.L. Jordan, 1993. Vapor analysis/extraction, in Geotechnical Practice for
        Waste Disposal p. 379-429, D.E. Daniel, Ed., Chapman & Hall, London.

 Silka, L.R., H.D. Cirpili, and D.L. Jordan, 1989. Modeling applications to vapor extraction
        systems, in Soil Vapor Extraction Technology, p. 252-259, T. A. Pederson and J. T.
        Curtis, Eds., Noyles Data Corporation.

 Sleep, B.E. and J.F. Sykes, 1989.  Modeling the transport of volatile organics in variably
        saturated media, Water Resources Research, 25(1):81-92.

 Smith, J.A., C.T. Chiou, J.A. Kammer, and D.E. Kile, 1990. Effect of soil moisture on the
        sorption of trichloroethene vapor to vadose-zone soil at Picatinny Arsenal, New Jersey,
       Environmental Science & Technology, 24(5):676-683.

 Smith, J.H., J.C. Harper, and H. Jaber, 1981. Analysis of environmental fate of Air Force
       distillate and high density fuelds, Engineering and Services Laboratory, Air Force
       Engineering and Services Center, Report ESL-TR-81-54, Tyndall Air Force Base,
       Florida.

 Steinberg, S.M., J.J. Pignatello, and B.L. Sawhney, 1987. Persistence of 1,2-dibromomethane in
       soils:  Entrapment in intraparticle micropores, Environmental Science and Technology
       21:1201-1208.

 Stelljes, M.E. and G.E. Watkin, 1993.  Comparison of environmental impacts posed bydifferent
       hydrocarbon mixtures: A need for site-specific compositional mixtures, in Hydrocarbon
       Contaminated Soils and Groundwater, Vol. 3, P.T. Kostecki and EJ. Calabrese, eds.,
       Lewis Publishers, FL, pp. 549-569.

 Stone, H.L., 1973. Estimation of three-phase relative permeability and residual oil data, Journal
       of Canadian Petroleum Technology, 12(4):53-61.

 Stone, H.L., 1970. Probability model for estimating three-phase relative permeability, Journal of
       Petroleum Technology, 20:214-218.

Strack, O.D.L., 1989.  Groundwater Mechanics, Prentice Hall, Englewood Cliffs, NJ, 732 p.

Stylianou, C., and B.A. DeVantier, 1995. Relative air permeability as  function of saturation in
       soil venting, J. of Environmental Engineering, ASCE, 121(4):337-347.
                                          222

-------
Sudicky, E.A., 1986.  A natural gradient experiment on solute transport in a sand aquifer: Spatial
       variability of hydraulic conductivity and its role in the dispersion process, Water
       Resources Research, 22(13):2069-2082.

Theis, C.V., 1935. The relation between the lowering of piezometric surface and the rate and
       duration of discharge of a well using ground water storage, American Geophysical Union
       Transactions, 16:519-524.

Thomas, G.W., 1982. Principles of Hydrocarbon Reservoir Simulation, International Human
       Resources Development Corporation, Boston, 207 p.

Thorstenson, D.C., and D.W. Pollock, 1989. Gas transport in unsaturated porous media:  The
       adequacy of Pick's Law, Rev. Geophysics, 27(l):61-78.

Toy, T.J., A.J. Kuhaida, Jr., and B.E. Munson,  1978. The prediction of mean monthly soil
       temperatures from mean monthly air temperature, Soil Science, 126:181r189.

Trimblay, D., 1993.  Overcoming barriers to the use of innovative technologies at underground
       storage tank sites, Presented at the 86th Annual Meeting and Exhibition, Air and Waste
       Management Association (June).

USEPA, 1993 c. Compilation of groundwater models, EPA/600/R-93/118.

USEPA, 1991a. Soil vapor extraction technology reference handbook, EPA/540/2-91/003.

USEPA, 1991c. Seminar publication: Site characterization for subsurface remediation,
       EPA/625/4-91/026.

USEPA, 199Id. Performance evaluation report, Thomas Solvent Raymond Road operable unit,
       Verona Well Field site, Region V, Battle Creek, ML

USEPA, l991e. Innovative treatment technologies, EPA/540/9-91/002, Office of Solid Waste
       and Emergency Response, Washington, DC.

USEPA, 1990a. Assessing UST corrective action technologies;  Early screening of clean-up
       technologies for the saturated zone, EPA/600/2-90/027.

USEPA, 1990c, Geostatistics for waste management: A user's manual for the GEOPACK
       (Version 1.0), Geostatistical Software System, EPA/600/8-90/004.

USEPA, 1988. Groundwater modeling: An overview and status report, EPA/600/2-89/028.
                                         223

-------
Valocchi, A., 1985. Validity of the local equilibrium assumption for modeling sorbing solute
       transport through homogeneous soils, Water Resources Research, 21(6):808-820.

Valsaraj, K.T. and LJ. Thibodeaux, 1988.  Equilibrium adsorption of chemical vapors on surface
       soils, landfills and landfarms-A review, Journal of Hazardous Materials, 19:79-99.

van Genuchten, M.Th., 1980.  A closed-form equation for predicting the hydraulic conductivity
       of unsaturated soils, Soil Science Society Am. Journal, 44:892-898.

Verschueren, K., 1983. Handbook of Environmental Data on Organic Chemicals, Van Nostrand
       Reinhold, New York, 1310 p.

Voss, C.I., 1984.  SUTRA: A finite-element simulation model for saturated-unsaturated, fluid-
       density dependent ground water flow with energy transport or chemically reactive single-
       species solute transport, U.S. Geological Survey Water-Resources Investigations Report
       84-4369.

Waldrop, W.R., L.W. Gelhar, A. Mantoglou, C. Weltry, and K.R. Rehfeldt, 1985.  A review of
       field-scale physical solute transport processes in saturated and unsaturated porous media,
       EPRIEA-4190, Electric Power Research Institute, Palo Alto, CA.

Walter, G.R., 1994. Discussion of "Modeling of vapor extraction and general transport in the
       presence of NAPL mixtures and nonideal conditions," by D.A. Benson, D. Huntley, and
       P.C. Johnson, Ground Water, 32(1):146-150.

Walter, G.R., A. Yiannakakis, and H.W. Bently, 1990. Gas in a hat, Proceedings of llth
       National Conference, Hazardous Materials Control Research Institute, Washington, DC,
       p. 557-64.'

Waterloo Hydrogeologic Software, 1993. Airflow User's Guide, Waterloo, Ontario, Canada.

Weast, R.C., ed., 1974.  Handbook of Chemistry and Physics, CRC Press, Boca Raton, Florida.

Weeks, E.P., D.E. Earp, and G.M. Thompson, 1982. Use of atmospheric fluorocarbons F-l 1 and
       F-12 to determine the diffusion parameters of the unsaturated zone in the southern high
       plains of Texas, Water Resources Research, 18(5):1365-1378.

Wilson, D.E., R.E. Montgomery, and M.R. Sheller, 1987. A mathematical model for removing
       volatile subsurface hydrocarbons by miscible displacement, Water, Air, and Soil
       Pollution, 33:231-255.

Youngquist, C.R., 1970. Diffusion and flow of gases in porous media, in Flow Through Porous
       Media, American Chemical Society, Washington, DC, pp. 58-69.

                                         224

-------
  APPENDIX A
CHEMICAL DATA
      225

-------
Table A-1. Unweathered composition of three common hydrocarbon products (USEPA, 1990d).
Hydrocarbon Group Representative
Hydrocarbon
n-Alkanes
C4
C5
C6
C7
C8
C9
C10-C14

n-Butane
n-Pentane
n-Hexane
n-Heptane
n-Octane
n-Nonane
n-Decane
Representative Concentrations (%w/w)
1
Automotive
Gasoline
10.8-29.6
4.8 - 7.0
1.9-4.5
2.0-12.9
0.2-2.3
1.3
0.4 - 0.8
0.2-0.8
2
#2 Fuel
Oil








3
Jet Fuel
JP-4

0.12
1.06
2.21
3.67
3.80
2.25
8.73

Branched Alkanes
C4
C5
C6
C7
C8
C9
C10-C14

Isobutane
Isopentane
2-Methylpentane
2-Methylhexane
2,4-Dimethylhexane
2,2,4-Trimethylhexane
2,2,5,5-Tetramethylhexane
18.18-59.5
0.7-2.2
8.6-17.3
4.6 - 9.7
1.4-8.3
1.8-16.7
1.2-2.7
0.5-2.6









0.66

2.27
5.48
8.82
3.36
1.35

Cycloalkanes
C6
C7
C8
C9
Others
Olefins
C4


Cyclohexane
Methylcyclohexane
1 ,2,4-Trimethylcyclopentane
1 , 1 ,3-Trimethylcyclohexane

1-Butene
3.2-13.7
0.2
1.0-3.9
0.2-1.4
0.2-0.7
1.6-7.5
5.5-13.5
0.9










2.40
3.77
1.35
3.21


                                      226

-------
Table A-1.  Unweathered composition of three common hydrocarbon products (continued).
Hydrocarbon Group Representative
Hydrocarbon
C5
C6
Others
1-Pentene
1-Hexene

Representative Concentrations (%w/w)
1
Automotive
Gasoline
1.3-3.3
0.8-1.8
2.5-7.5
2
#2 Fuel
Oil



3
Jet Fuel
JP-4




Mqno-aromatics
Benzene
Toluene
Xylenes
Ethyl benzene
C3-benzenes
C4-benzenes
Others

Benzene
Toluene
m-Xylene
Ethyl benzene
1 ,3,5-Trimethylbenzene
1 ,4-Diethylbenzene

19.3-40.9
0.9-4.4
4.0-6.5
5.6-8.8
1.2-1.4
3.2-11.3
2.1 T2.6
1.6-5.2



0.07
0.03
0.67
0.88


0.50
1.33
2.32
0.37
3.59
3.98


Phenols
Phenol
C1 -phenols
C2-phenols
C3-phenols
C4-phenols
Indanol
Poly-aromatics

Phenol
o-Cresol
2,4-Dimethylphenol
2,4,6-Trimethylphenol
m-Ethylphenoi
Indanol
Fluorene









0.001
0.01
0.02
0.02
0.01
0.001
0.57









Nitro-aromatics
C1 -anilines
C2-anilines
Complex anilines



Quinoline





0.003
0.004
0.002




                                      227

-------
  Table A-1.   Unweathered composition of three common hydrocarbon products (continued).
Hydrocarbon Group Representative
Hydrocarbon
Di-aromatics
Saturated
hydrocarbons
C8
C9
C10
C11
C12
C13
C14
C15
C16
C17
C18
C19
C20
C21
C22
C23
C24
Pristane
Phytane
Unknowns
Naphthalene

n-Octane
n-Nonane
n-Decane
n-Undecane
n-Dodecane
n-Tridecane
n-Tetradecane
n-Pentadecane
n-Hexadecane
n-Heptadecane
n-Octadecane
n-Nonadecane
n-Eicosane
n-Heneicosane
' n-Docosane
n-Tricosane
n-Tetracosane



Representative Concentrations (%w/w)
1
Automotive
Gasoline
0.7




















6.6-13.8
2
#2 Fuel
Oil
3.43

0.05 .
0.20
0.58
0.98
1.14
1.20
1.31
1.42
1.53
1.51
1.31
1.16
0.99
0.51
0.29
0.15
0.05
0.52
0.46

3
Jet Fuel
JP-4
1.59





















NOTE: Blanks indicate the unavailability of data and do not indicate the absence of a particular compound
       from the hydrocarbon product.
                                           228

-------
Table A-2.    Range of abundance of some of the constituents typically found in virgin
              mixtures of gasoline (from Stelljes and Watkin, 1993). Values are in
              percent by weight1.
Component
Maximum Abundance
Minimum Abundance
Aromatics
Benzene
Toluene
Ethylbenzene
Xylenes (total)
Naphthalene
2-Methylnaphthalene
Benzo(a)pyrene
3.5
21.8
2.86
8.31
0.49
3.85
2.8x1 Q-6
0.12
2.73
0.36
3.22
0.09
2.91
1.9X10'7
Branched Alkanes
Isopentane
10.17
6.07
n-Alkanes
n-Butane
n-Pentane
n-Hexane
4.70
10.92
3.5
3.93
5.75
0.24
Additives
Ethylene dibromide
1. 77X10"4
.
7x1 0-7
1 Percent by weight values can be converted to parts per million (milligrams per kilogram) by
multiplying the values shown above by 106.
                                    229

-------
Table A-3.    Range of abundance of some aromatic chemicals typically found in virgin
              mixtures of diesel fuel (from Stelljes and Watkin, 1993).  Values are in
              parts per million (mg/1) by weight1.
Component
Benzene
Toluene
Ethylbenzene
Xylenes (total)
Naphthalene
2-Methylnaphthalene
Benzo(a)pyrene
Benz(a)anthracene
Chrysene
Fluoranthene
Phenanthrene
Pyrene
Triphenylene
Cresol
Phenol
Quinoline
Maximum Abundance
82
800
800
800
2.7302
6.7002
0.6
1.2
2.22
37
1.5002
412
2.22
54.32
6.82
9.22
Minimum Abundance
6
100
100
100
2.7302
6.7002
0.006
0.001
2.22
ND3
1.5002
412
2.22
54. 32
6.82
9.22
1Parts per million can be converted to percent by weight by dividing the concentrations
  shown above by 106.
2Only one concentration was reported for this chemical.
3ND = Not detected.
                                     230

-------
Table A-4.    Major components of JP-4 (from Smith et al., 1989).
Fuel Component
n-Butane
Isobutane
n-Pentane
2,2-Dimethylbutane
2-MethyIpentane
3-Methylpentane
n-Hexane
Methylcyclopentane
2,2-Dimethylpentane
Benzene
Cyclohexane
2-Methylhexane
3-Methylhexane
trans-1 ,2-Dimethylcyclopentane
cis-1 ,3-Dimethylcyclopentane
cis-1 ,2-Dimethylcyclopentane
n-Heptane
Methylcyclohexane
2,2,3,3-Tetramethylbutane
Ethylcyclopentane
2,5-Dimethylhexane
2,4-Dimethylhexane
1 ,2,4-Trimethylcyclopentane
3,3-Dimethylhexane
1 ,2,3-Trimethylcyclopentane
Toluene
2,2-Dimethylhexane
Kovats
Index
400.0
466.3
500.0
527.7
562.4
578.7
600.0
622.0
629.1
644.5
653.6
669.5
677.3
679.6
681.9
684.4
700.0
715.1
720.5
729.8
737.3
738.4
740.8
743.3
748.1
753.0
764.2
Percent by
Weight
0.12
0.66
1.06
0.10
1.28
0.89
2.21
1.16
0.25
0.50
1.24
2.35
1.97
0.36
0.34
0.54
3.67
2.27
0.24
0.26
0.37
0.58
0.25
0.26
0.25
1.33
0.71
                          231

-------
Table A-4. Major components of JP-4 (continued).
Fuel Component
2-Methylheptane
4-Methylheptane
cis-1 ,3-Dimethylcyclohexane
3-MethyIheptane
1 -Methyl-3-ethylcyclohexane
1 -Methyl-2-ethylcyclohexane
Dimethylcyclohexane
n-Octane
1 ,3,5-Trimethylcyclohexane
1 ,1 ,3-Trimethylcyclohexane
2,5-Dimethylheptane
Unidentified
Ethylbenzene • ;
m'-Xylene ;
p-Xylene
3,4-Dimethylheptane
4-EthyIheptane
4-Methyloctane
2-Methyloctane
3-MethyIoctane
o-Xylene
1 -Methyl-4-ethyIcyclohexane
n-Nonane
Isopropylbenzene
n-Propylbenzene
1-MethyI-3-ethyIbenzene :
Kovats
Index
772.0
772.7
775.3
778.0 '
784.1
786.7
788.8
800.0
825.3
831.0
833.6
839.9
844.9
853.9
854.8
859.8
865.0
868.5
869.6
873.9
875.3
881.3
900.0
905.1
937.2
944.9
Percent by
Weight
2.70
0.92
0.42
3.04
0.17
0.39
0.43
3.80
0.99
0.48
0.52
0.98
0.37
0.96
0.35
0.43
0.18
0.86
0.88
0.79
1.01
0.48
2.25
0.30
0.71
0.49
                     232

-------
Table A-4. Major components of JP-4 (continued).
Fuel Component
1 -Methyl-4-ethylbenzene
1 ,3,5-Trimethylbenzene
1 -Methyl-3-ethylbenzene
1 ,2,4-Trimethylbenzene ,
n-Decane
n-Butylcylohexane
1 ,3-Diethylbenzene ,
1 -Methyl-4-propylbenzene
1 ,3-Dimethyl-5-ethylbenzene
, 1-Methyl-2-i-propylbenzene
1 ,4-Dimethyl-2-ethylbenzene
1 ,2-Dimethyl-4-ethylbenzene
n-Undecane
1,2,3,4-Tetramethylbenzene . .
Naphthalene :
2-Methylundecane
n-Dodecane
,2,6-Dimethylundecane
Unidentified ,
2-Methylnaphthalene
1 -Methylnaphthalene
.n-Tridecane
2,6-Dimethylnaphthalene
n-Tetradecane
Kovats
Index
946.8
952.8
961.0
975.6
1000.0
1025.6
1031.4
1034.7
1041.6
1049.1
1060.2
1067.1
1100.0
1 128.8
1156.5
1166.0
1200.0
1216.1
1262.3
1265.7
1276.4
1300.0
1379.4
1400.0
Percent by
Weight
0.43
0.42
0.23
1.01
2.16
0.70
0.46
0.40
0.61
0.29
0.70
0.77
2032
0.75 ;
0.50
0.64
2.00
0.71
0.68
0.56
0.78
1.52
0.25
073
                    233

-------
Table A-5.  Average composition of gasoline vapor exposures (from Haider et al., 1986).
Compound
Compositional Make-upA
Amoco Oil6
(wt %)
Amoco Oilc
(wt %)
Shell Oil0
(vol %)
APIE
(wt %)
Alkanes:
C3
C4

Cs


C6




C7




C8
n-propane
n-butane
isobutane
n-pentane
isopentane
cyclopentane
2,3-dimethylbutane
2-methylpentane
3-methyIpentane
methylcyclopentane
n-hexane
2,3-dimethyIpentane
2,4-dimethylpentane
2-methylhexane
3-methylhexane
n-heptane
2,2,4-trimethylpentane
_
33.7 (7.8)
4.1 (0.8)
8.1 (2.5)
21.6(3.7)
_
1.3(0.7)
3.4(1.3)
2.0 (0.7)
1.1 (0.6)
1.8(0.7)
0.6 (0.3)
_
0.6 (0.3)
0.7 (0.4)
-
0.7 (0.5)
_
21.2(10.4)
3.4(1.6)
9.4(1.5)
27.2 (6.7)
_
3.3(1.8)
4.9(1.4)
3.2 (0.9)
1.5(0.4)
3.1 (0.7)
0.9 (0.8)
0.8 (0.5)
1.1 (0.3)
1.1 (0.3)
0.7 (0.2)
1.8(1.2)
0.8(1.1)
38.1 (5.7)
5.2(1.9)
7.0(4.0)
22.9(6.1)
0.7 (0.7)
0.7 (0.5)
2.1 (1.3)
1.6(0.9)
1.3(0.4)
1.5(0.9)
0.7 (0.6)
_
_
_
-
0.5 (0.5)



10.9(4.2)






1.7(0.9)






Alkenes:
C4



cs


isobutylene
1-butene
trans-2 butene
cis-2-butene
2-methyl-1 -butene
2-methyl-2-butene
1-pentene
_
_
1.2(0.5)
0.9 (0.3)
0.9 (0.4)
.
_
_
1.0(0.7)
»
1.2(0.7)
-
1.5(0.7)
0.7 (0.4)
1.1 (1.5)
_
.
_
1.6(2.1)
1.7(1.8)
_







                                      234

-------
Table A-5.    Average composition of gasoline vapor exposures (continued).
Compound


trans-2-pentene
cis-2-pentene
Compositional Make-upA
Amoco Oil8
(wt %)
0.8 (0.6)
_
Amoco Oilc
(wt%)
_
.
Shell Oil0
(vol %)
.
1.2(1.7)
APIE
(wt %)


Aromatics:
C6
C7
C8

benzene
toluene
xylene (p, m, o,)
Total Percent
2.2(1.0)
3.1 (1.6)F
0.9 (0.7)
89.7
0.6 (0.3)
4.0(1.8)F
1.5(0.7)
94.1
0.7 (0.4)
1.8(1.3)
0.5 (0.6)
91.7
2.2(1.1)
2.2(1.8)
1.1 (1.5)

       A Components listed comprise at least 0.5% by wt. or vol. Composition
       less than 0.5% denoted by "-", composition presented as arithmetic mean
       (± standard deviation).
       B N = 12. Bulk terminal exposures.
       c N = 11. Marine loading exposures.
       DN = 95
       EN = 152
       F Toluene coeluted with 2,3,3-trimethylpentane on the analytical column;
       however, the major proportion is assumed to be toluene.
                                     235

-------
Table A-6.    Physicochemical properties of five common hydrocarbon mixtures (from USEPA,
              1990a).
Product
Automotive Gasoline
#2 Fuel Oil
#6 Fuel Oil
Jet Fuel (JP-4)
Mineral Base
Crankcase Oil
Air
Saturated Aqueous
Vapor ,
Liquid Density
(g/cm3)
(0.73)
0.72-0.76 [15.6]
(0.91)
0.87-0.95
(0.91)
0.87-0.95
0.75
0.84-0.96 [15]
	
1
Liquid Viscosity
(cP)
(0.45)
0.36-0.49 [15.6]
(1.56)
1.15-1.97 [21]
(254)
14.5-493.5 [38]
0.829 [21]
275 [38]
	
1
Water Solubility
(mg/l)
(158)
131-185 [13-25]
3-10 [20-23]
"O
10-20
insoluble
	
	
Vapor Pressure
(mm Hg)
(469)
263-675 [38]
(14.3)
2.12-26.4 [21]
(14.3)
2.12-26.4 [21]
91
N/A
760
17.5
N/A = Not Available

Notes:  All values for 20oC unless noted in brackets [].
              Values in parentheses are typical of the parameter ().
              Values for air and saturated aqueous vapor are included, where applicable, as a means of
              comparison.
                                           236

-------
Table A-7.  Chemical properties of selected organic chemicals (after Cohen and Mercer, 1993)
COMPOUND
Acetone '
Aniline
o-Anisidine
Benzene
Benzo(a)anthracene
Benzoic Acid
Benzo(a)pyrene
Benzyl alcohol
Benzyl chloride
Bis(2-chloroethyl)ether
Bis(2-chloroisopropyl)ether
Bromobenzene
Bromochlpromethane
Bromodichloromethane
Bromoethane
Bromoform
n-Butane
1-Butene
Butyl benzyl phthalate
Carbon disulfide
Carbon tetrachloride
Chlorobenzene
Chloroethane
2-Chloroethyl vinyl ether
Chloroform
Chloromethane
1 -Chloro-1 -nitropropane
2-Chlorophenol
4-Chlorophenyl phenyl ether
Chloropicrin
m-Chlorotoluene
o-Chlorotoluene
p-Chlorotoluene
Chrysene
o-Cresol
Cyclohexanone
Cyclohexane
n-Decane
Synonym
2-Propanone
Benzenamine
2-Methoxybenzenamine
Annulene
Benza(a)anthracene
Benzoate
1 ,2 Benzopyrene
Benzenemethanol
Chloromethylbenzene
Bis(B-chloroethyl)ether
Bis(S-chloroisopropyl)ether
Phenyl bromide
Chlorobromomethane -
Dichlorobromomethane •
Ethyl .bromide
Tribromomethane
Butane
Ethylethylene
Benzyl butyl phthalate
Carbon bisulfide
Tetrachloromethane
Benzene chloride
Aethylis
(2-Chloroethoxy)ethene
Trichloromethane
Methylchloride
Chloronitropropane
o-Chlorophenol
p-Chlorodiphenyl ether
Trichloronitromethane

2-Chloro-1 -methylbenzene
4-Chlorotoluene
Benz(a)phenanthrene

Anone
Benzene hexahydride
Decane
CAS#
67-64-1
62-53-3
90-04-0
71-43-2
56-55-3
65-85-0
50-32-8
100-51-6
100-44-7
111-44-4
108-60-1
108-86-1
74-97-5
75-27-4
74-96-4
75-25-2
106-97-8
106-98-9
85-68-7
75-15-0
56-23-5
108-90-7
75-00-3
110-75-8
67-66-3
74-87-3
600-25-9
95-57-8
7005-72-3
76-06-2
108-41-8
95-45-8
106-43-4
218.01-9
95-48-7
108-94-1
110-82-7
124-18-5
Empirical
Formula
C3H6O
C6H7N
C7H9NO
C6H6
C18H12
C7H6O2
C20H12
C7H80
C7H7CI
C4H8CI2O
C6H12CI2O
C6H5BR
CH2BrCI
CHBrCI2
C2H5Br
CHBrS
C4H10
C4H8
C19H20O4
CS2
CCI4
C6H5CI
C2H5CI
C4H7CIO
CHCI3
CH3CI
C3H6CINO2
C6H5C1O
C12H9C1O
CCI3NO2
C6H4CH3CI
C6H4CH3CI
C6H4CH3CI
C18H12

C6H10O
C6H12
C10H22
Formula
Weight
(9).
5.81 E+01
9.31 E+01
1.23E+02
7.81 E+01
2.28E+02
1.22E+02
2.52E+02
1.08E+02
1.27E+02
1.43E+02
1.71E+02
1.57E+02
1.29E+02
1.64E+02
1.09E+02
2.53E+02
5.81 E+01
5.61 E+01
3.12E+02
7.61 E+01
1.54E+02
1.13E+02
6.45E+01
1.07E+02
1.19E+02
5.05E+01
1.24E+02
1.29E+02
2.05E+02
1.64E+02
1.27E+02
1.27E+02
1.27E+02
2.28E+02
1.08E+02
9.82E+01
8.42E+01
1.42E+02
Ref.
f
b
b
a
f
f
a
a
a
b
a
b
b
a
b
a
f
f
a
a
a
a
a
a
a
a
b
a
a
b
f .
f
f
f
f
f
f
f
Specific
Density
(g/cc)
7.90E-01
1.02E+00
1.09E+00
8.77E-01
1.27E+00
1.27E+00
1.35E+00
1.05E+00
1.10E+00
1.22E+00
1.10E+00
1.50E+00
1.93E+00
1.98E+00
1.46E+00
2.89E+00
6.01 E-01
5.95E-01
1.12E+00
1.26E+00
1.59E+00
1.11E+00
8.99E-01
1.05E+00
1.48E+00

1.21E+00
1.26E+00
1.20E+00
1.66E+00
1.07E+00
1.08E+00
1.07E+00
1.27E+00
1.05E+00
9.48E-01
7.79E-01
7.30E-01
Ref.
a
b
b
a
a
a
a
a
a
b
a
b
b
a
-. b
a
b
b
a
a
a
a
a
a
a

b
a
a
b
f
f
f(25)
a
w
b
b
b
                                        237

-------
Table A-7.  Chemical properties of selected organic chemicals (after Cohen and Mercer, 1993).
COMPOUND
Dibromochloromethane
1 ,2-Dibromo-3-chloropropane
Dibromodifluoromethane
Dibutyl phthalate
1,2-Dichlorobenzene
1,3-Dlchlorobenzene
DIchlorodifluoromethane
1,1-Dlchloroethane
1,2-Dichloroethane
1,1-DIchloroethene
trans-1 ,2-Dichloroethene
1 ,2-Dichloropropane
cts-1 ,3-Dichloropropene
trans-1 ,3-Dlchloropropene
Dichlorvos
1,4-Diethylbenzene
DIethyl phthalate
2,3-Dimethylbutane
2,3-Otmethylpentane
2,4-DImethylpentane
2,4-Dimethylphenol
Dimethyl phthalate
1 ,4-Dioxane
n-Dodecane
Ethyl Acetate
Ethylbenzene
Ethylene dibromide
Fluoranthene
Fluorene
n-Heptane
Hexachlorobutadlene
Hexachlorocyctopentadiene
n-Hexadecane
n-Hexane
1-Hexene
lodomethane
1-lodopropane
Isobutane
Synonym
Chlorodibromomethane
DPCP
Freon 12-B2
Dlbutyl-n-phthalate; DBP
o-Diohlorobenzene
m-Dichlorobenzene
Freon 12
1,1 -DCA
Ethylene dichloride; 1,2-DCA
Vinylidene chloride; 1,1-DCE
trans-1 ,2-DCE
Propyiene dichloride
cis-1 ,3-Dichloropropylene
trans-1 ,3-Dichloropropylene
No-Pest Strip

DEP
Biisopropyl
3,4-DimethyIpentane
Diisopropylmethane
2,4-Xylenol
DMP
Dioxane
Dodecane
Acetic Acid Ethyl Ester

1 ,2-Dibromoethane; EDB
1,2 Benzacenaphthene
9H-Fluorene
Heptane
HCBD
HCCPD
Hexadecane
Hexane
Hexene
Methyl iodide
Propyl iodide

CAS#
124-48-1
96-12-8
75-61-6
84-74-2
95-50-1
541-73-1
75-71-8
75-34-3
107-06-2
75-35-4
156-60-5
78-87-5
10061-01-5
10061-02-6
62-73-7

84-66-2
79-29-8
565-59-3
108-08-7
1300-71-6
131-11-3
123-91-1
112-40-3
141-78-6
100-41-4
106-93-4
206-44-0
86-73-7
142-82-5
87-68-3
77-47-4
544-76-3
110-54-3
592-41-6
74-88-4
107-08-4
75-28-5
Empirical
Formula
CHBr2CI
C3H5Br2CI
CBr2F2
C16H22O4
C6H4CI2
C6H4CI2
CCI2F2
C2H4CI2
C2H4CI2
C2H2CI2
C2H2CI2
C3H6CI2
C3H4CI2
C3H4CI2
C4H7CI2)4P

C12H1404
C6H14
C7H16
C7H16
C8H10O
C10H10O4
C4H8O2
C12H26
C4H802
C8H10
C2H4Br2
C16H10
C13H10
C7H16
C4CI6
C5CI6
C16H34
C6H14
C6H12
CH3I
C3H7I
C4H10
Formula
Weight
(g)
2.08E+02
2.36E+02
2.10E+02
2.78E+02
1.47E+02
1.47E+02
1.21E+02
9.90E+01
9.90E+01
9.69E+01
9.69E+01
1.13E+02
1.11E+02
1.11E+02
2.21 E+02
1 .34E+02
2.22E+02
8.62E+01
1.00E+02
1.00E+02
1.22E+02
1.94E+02
8.82E+01
1 .70E+02
8.81 E+01
1.06E+02
1.88E+02
2.02E+02
1.66E+02
1.00E+02
2.61 E+02
2.73E+02
2.26E+02
8.62E+01
8.42E+01
1.42E+02
1.70E+02
5.81 E+01
Ref.
a
b
b
a
a
a
a
a
a
a
a
a
a
a
b
f
a
f
f
f
f
a
b
f
f
f
b
f
a
f
a
a
f
f
f
b
b
f
Specific
Density
(g/cc)
2.45E+00
2.05E+00
2.30E+00
1.05E+00
1.30E+00
1.29E+00
1.75E+00
1.18E+00
1.24E+00
1.22E+00
1.26E+00
1.56E+00
1.22E+00
1.18E+00
1.42E+00
8.62E-01
1.12E+00
6.62E-01
6.95E-01
6.73E-01
9.65E-01
1.19E+00
1.03E+00
7.49E-01
9.00E-01
8.67E-01
2.18E+00
1.25E+00
1.20E+00
6.84E-01
1 .55E+00
1 .70E+00
7.73E-01
6.60E-01
6.73E-01
2.28E+00
1.75E+00
5.57E-01
Ref.
a
b
b
a
a
a
a
a
a
a
a
a
a
a
b(25)
m
a
b
b
b
f
a
b
b
b
a
b
a
a
b
a
a
r
b
b
b
b
ii
                                        238

-------
Table A-7.  Chemical properties of selected organic chemicals (after Cohen and Mercer, 1993).
COMPOUND
sobutanol
Isopentane
Walathion
Methanol
Methylcyclohexane
Methylene chloride
Methyl Isobutyl Ketone
Methyl Ethyl Ketone
2-Methylpentane
2-Methylhexane
2-MethyInapthalene
Napthalene
Nitrobenzene
Nitroethane
1-Nitropropane
2-Nitrotoluene
3-Nitrotoluene
n-Nonane
n-Octadecane
n-Octane
Parathion
PCB-1016
PCB-1221
PCB-1232
PCB-1242
PCB-1248
PCB-1254
Pentachloroethane
n-Pentane
n-Pentadecane
1-Pentene
Phenanthrene
Phenol
Pyrene
Quinoline
Styrene
1 ,1 ,2,2-Tetrabromoethane
1 ,1 ,2,2-Tetrachloroethane
Synonym
sobutyl Alcohol
.


Cyclohexylmethane
Dichloromethane
MIBK
MEK
sohexane
soheptane

vlapthalin
Nitrobenzol
UN 2842
UN 2608
1 -Methyl-2-nitrobenzene
1 -Methyl-3-nitrobenzene
Nonane

Octane

Aroclor 1016
Aroclor 1221
Aroclor 1232
Aroclor 1242
Aroclor 1 248
Aroclor 1 254
Ethane pentachloride
Pentane

Propylethylene
Phenanthren
Benzenol
B-Pyrene

Ethyenyl
Acetylene tetrabromide
Acetylene tetrachloride
CAS#
78-83-1
78-78-4
121-75-5
67-56-1
108-87-2
75-09-2
108-10-1
78-93-3
107-83-5
591-76-4
91-57-6
91-20-3
98-95-3
79-24-3
108-03-2
88-72-2
99-08-1
111-84-2

111-65-9
56-38-2
12674-11-2
11104-28-2
11141-16-5
53469-21-9
12672-29-6
11097-69-1
76-01-7
109-66-0
629-62-9
109-67-1
85-01-8
108-95-2
129-00-0
91-22-5
100-42-5
79-27-6
79-34-5
Empirical
Formula
C4H100
C5H12
C10H19O6PS2
CH4O
C7H14
CH2CI2
C6H12O
C4H8O
C6H14
,.C7H16
^ CIWIQ
C10H8
C6H5NO2
C2H5NO2
C3H7NO2
C7H7NO2
C7H7NO2
C9H20
C18H32
C8H18
C10H14NO5P
varies
varies
varies
varies
varies
vanes
C2HCI5
C5H12
C15H32
C5H10
C14H10
C6H6O
C16H10

C8H8
C2H2Br4
C2H2CI4
Formula
Weight
(g)
7.41 E+01
7.22E+01
3.30E+02
3.20E+01
9.82E+01
8.49E+01
1.00E+02
7.21 E+01
8.62E+01
1.00E+02
1.42E+02
1.28E+02
1.23E+02
7.51 E+01
8.91 E+01
1.37E+02
1 .37E+02
1.28E+02
2.54E+02
1.14E+02
2.91 E+02
2.58E+02
1.92E+02
2.21 E+02
2.61 E+02
2.88E+02
3.27E+02
2.02E+02
7.22E+01
2.12E+02
7.01E+01
1.78E+02
9.41 E+01
2.02E+02
1 .29E+02
1.04E+02
3.46E+02
1 .68E+02
Ref.
f
f
b
f
f
a
f
f
f
f
a
f
a
b
b
b
b
f
f
f
b
a
a
a
a
a
a
b
f
f
f
f
f
f
w
f
b
a
Specific
Density
(g/cc)

6.20E-01
1.23E+00
8.10E-01
7.69E-01
1.33E+00
7.98E-01
8.05E-01
6.53E-01
6.79E-01
1.01E+00
9.63E-01
1.20E+00
1.04E+00
1.01E+00
1.16E+00
1.16E+00
7.18E-01

7.03E-01
1.26E+00
1.33E+00
1.18E+00
1.24E+00
1.39E+00
1.41E+00
1.51E+00
1.68E+00
6.26E-01
7.69E-01
6.41 E-01
9.80E-01
1 .06E+00
1.27E+00
1 .09E+00
9.06E-01
2.88E+00
1.60E+00
Ref.

y
b(25)
w
b
a
y
w
b
b
a
a
a
b(25)
b(24)
b
b
b

b
b
a(25)
a(25)
a(25)
a(15)
a(25)
a(15)
b
b
P
b
a
a
a
w
a
b
a
                                        239

-------
Table A-7.  Chemical properties of selected organic chemicals (after Cohen and Mercer, 1993)
COMPOUND
Tetrachloroethene
n-Tetradecane
Tetrahydrofuran
Thiophene
Toluene
1 ,2,4-Trichlorobenzene
1 .1 ,1-Trichloroethane
1 ,1 ,2-Trichloroethane
Trichloroethene
1 ,1 ,2-Trichlorofluoromethane
1 ,2,3-Trichloropropane
1 ,1 ,2-Trichlorotrifluoroethane
n-Tridecane
1 ,3,5-Trimethylbenzene
1,1 ,3-Trimtehylcyclohexane
2,2,5-Trimethylhexane
2,2,4-Trimethylpentane
2,3,4-Trimethylpentane
2,4,6-Trimethylphenol
Tri-o-cresyl phosphate
Triphenytene
n-Undecane
Vinyl Chloride
m-Xylene
o-Xylene
p-Xylene

Water
Synonym
Perchloroethylene; PCE

Hydrofuran
Thiacylopentadiene
Methylbenzene
1,2,4-TCB
Methyl chloroform; 1,1,1-TC
1,1,2-TCA
TCE
Freon 1 1
Allyl trichloride
Freon 113

Mesitylene
1 ,3,3-Trimethylcyclohexane
2,5,5-Trimethylhexane
Isooctane

Picric acid
o-Cresyl phosphate


Chlorethene
m-Xylol
o-Xylol
p-Xylol

Ice
CAS#
127-18-4
629-59-4
109-99-9
110-02-1
108-88-3
120-82-1
71-55-6
79-00-5
79-01-6
75-69-4
96-18-4
76-13-1
629-50-5
108-67-8
3073-66-3
3522-94-9
540-84-1
565-75-3
88-89-1
78-30-8


75-01-4
108-38-3
95-47-6
106-42-3

7732-18-5
Empirical
Formula
C2CI4
C14H30
C4H80
C4H4S
C7H8
C6H3CI3
C2H3CI3
C2H3CI3
C2HCI3
CCI3F
C3H5CI3
C2CI3F3
C13H28
C9H12
C9H18
C9H20
C8H18
C8H18

C21H21O4P

C11H24
C2H3CI
C8H10
C8H10
C8H10

H2O
Formula
Weight
(9)
1.66E+02
1.98E+02
7.21 E+01
8.41 E+01
9.21 E+01
1.81E+02
1.33E+02
1.33E+02
1.31E+02
1.37E+02
1.47E+02
1.87E+02
1.84E+02
1.20E+02
1.26E+02
1.28E+02
1.14E+02
1.14E+02
2.29E+02
3.68E+02
2.28E+02
1.56E+02
6.25E+01
1.06E+02
1.06E+02
1.06E+02

1.80E+01
Ref.
a
•f
f
b
f
a
a
a
a
a
b
b
0
f
b
b
f
b
f
b
f
n
f
f
f
f


Specific
Density
(g/cc)
1.62E+00
7.63E-01
8.89E-01
1.06E+00
8.67E-01
1.45E+00
1.34E+00
1.44E+00
1.46E+00
1.49E+00
1.39E+00
1.56E+00
7.56E-01
8.65E-01
7.66E-01
7.07E-01
6.92E-01
7.19E-01
1.76E+00
1.96E+00


9.11E-01
8.64E-01
8.80E-01
8.81 E-01

1.00E+00
Ref.
a
y
b
b
a
a
a
a
a
a
b
b
P
b
b
b
b
b
w
b


a
a
a
a


                                       240

-------
Table A-7. Chemical properties of selected organic chemicals (after Cohen and Mercer, 1993)
COMPOUND
Acetone
Aniline
o-Anisidine
Benzene
Benzo(a)anthracene
Benzoic Acid
Benzo(a)pyrene
Benzyl alcohol
Benzyl chloride
Bis(2-chloroethyl)ether
Bis(2-chloroisopropyl)ether
Bromobenzene
Bromochloromethane
Bromodichloromethane
Bromoethane
Bromoform
n-Butane
1-Butene
Butyl benzyl phthalate
Carbon disulfide
Carbon tetrachloride
Chlorobenzene
Chloroethane
2-Chloroethyl vinyl ether
Chloroform
Chloromethane-
1 -Chloro-1 -nitropropane
2-Chlorophenol
4-Chlorophenyl phenyl ether
Chloropicrin
m-Chlorotoluene
o-Chlorotoluene
p-Chlorotoluene
Chrysene
o-Cresol
Cyclohexanone
Cyclohexane
n-Decane
Absolute
Viscosity
(cP) (1)

4.40E+00

6.47E-01



7.76E+00

2.14E+00
2.30E+00
9.85E-01
5.70E-01
1.71E+00
4.18E-01
2.02E+00
7.00E-03


3.66E-01
9.70E-01
7.99E-01
2.79E-01

5.80E-01
1.83E-01

2.25E+00


7.50E-01
7.50E-01


4.49E+00

1 .02E+00
9.20E-01
Ref.

c

cc



d(15)

c(25)
JI
e(30)
h
e
d(15)
c
II


c
c
c
kk

c
X

e(45)


h(38)
h(38)


x(40)

nn
y
Boiling
Point
(deg.C)
5.62E+01
1 .84E+02
2.24E+02
8.01 E+01
4.37E+02
2.49E+02
4.95E+02
2.05E+02
1.79E+02
1 .79E+02
1 .87E+02
1.56E+02
6.80E+01
9.00E+01
3.80E+01
. 1 .49E+02
-1.00E+00
-6.00E+00
3.70E+02
4.60E+01
7.70E+01
1.32E+02
1.23E+01
1.08E+02
6.20E+01
-2.40E+01
1.42E+02
1.75E+02
2.84E+02
1.12E+02
1.60E+02
1.59E+02
1.62E+02
4.88E+02
1.91E+02
1.57E+02
8.10E+01
1.74E+02
Ref.
f
b
b
f
a
f
a
a
a
b
a
b
b
a
b
a
f
f
a
a
a
a
a
a
a
f
b
a
a
b
f
f
f
f
f
f
f
b
Aqueous
Solubility
(mg/L)(1)
misicible
3.50E+04
1.30E+04
1.78E+03
1.40E-02
2.70E+00
1.20E-03
3.50E+04
4.93E+02
1.02E+04
1.70E+03
5.00E+02
1.67E+04
4.50E+03
9.14E+03
3.01 E+03
6.10E+01
2.22E+02
2.82E+00
2.10E+03
8.00E+02
5.00E+02
5.74E+03
1.50E+04
8.00E+03
4.00E+03
6.00E+00
2.85E+04
3.30E+00
2.00E+03
4.80E+01
7.20E+01
4.40E+01
6.00E-03
3.10E+04
2.30E+04
5.50E+01
9.00E-03
Ref.
a
b
b
f
a(25)
f(18)
k
a
a
b
a
b
b(25)
a(0)
b
a
f
b(25)
a
a
a
a
k
a
a
a
b
a
a(25)
b
e
e
e
f(25)
f(40)
f
f
f
Vapor
Pressure
(mm Hg) (2)
8.90E+01
3.00E-01
<0.1
7.60E+01
2.20E-08
1.00E+00
5.60E-09
<1
9.00E-01
' 7.10E-01
8.50E-01
3.30E+00
1.41E+00
5.00E+01
3.75E+02
4.00E+00
1.82E+03
4.00E+02
8.60E-06
2.98E+02
9.00E+01
9.00E+00
1.00E+03
2.68E+01
1.60E+02
3.79E+03
5.80E+00
1.42E+00
2.70E-03
2.00E+01
4.60E+00
2.70E+00
4.50E+00
6.30E-09
2.40E-01
4.00E+00
7.70E+01
2.70E+00
Ref.
a(5)
b
b
f
k
a(96)
k
a
a
b
a
b
b(25)
a
b
a.
f(25)
f(-21.7)
a
a
a
a
k
a
a
a
b(25)
a(25)
a(25)
b
e
f
e
a
f(25)
f
f
f
-
241

-------
Table A-7.  Chemical properties of selected organic chemicals (after Cohen and Mercer, 1993).
COMPOUND
Dibromochloromethane
1 ,2-Dibromo-3-chloropropane
Dibromodifluoromethane
Dibuly! phthalate
1,2-Dichlorobenzene
1 ,3-DIchlorobenzene
DIchlorodifluoromethane
1,1-DIchloroethane
1 ,2-Dichloroethane
1,1-Dichloroethene
lrans-1 ,2-Dich!oroethene
1 ,2-Dichloropropane
cis-1 ,3-Dichloropropene
trans-1 ,3-Dichloropropene
Dichlorvos
1,4-DlethyIbenzene
Diethyl phthalate
2,3-Dimethylbutane
2,3-Dimethylpentane
2,4-Dimethylpentane
2,4-DimethylphenoI
Dimethyl phthalate
1,4-Dloxane
n-Dodecane
Ethyl Acetate
Ethylbenzene
Ethylene dibromide
Fluoranthene
-luorene
n-Heptane
Hexachlorobutadiene
•iexachlorocyclopentadiene
n-Hexadecane
n-Hexane
l-Hexene
lodome thane
1-lodopropane
Isobutane
Absolute
Viscosity
(cP) (1)



2.03E+01
1.32E+00
1.04E+00
2.62E-01
4.40E-01
8.00E-01
3.60E-01
4.00E-01
8.60E-01




3.50E+01



1.85E+00
1.72E+01
1.20E-02
1.37E+00
4.40E-01
6.40E-01
1.72E+00



2.45E+00




5.18E-01
8.37E-01

Ref.



c
c(25)
c(25)
PP
c
c
c
c
c




c



aa(70)
c(25)
w
V
y
jj
c



c(38)




d(15)
d(15)

Boiling
Point
(deg.C)
1.17E+02
1.96E+02
2.30E+01
3.35E+02
1.80E+02
1.73E+02
-2.98E+01
5.60E+01
8.30E+01
3.70E+01
4.70E+01
9.60E+01
1.04E+02
1.12E+02
1.40E+02
1.84E+02
2.98E+02
5.70E+01
8.98E+01
8.05E+01
2.12E+02
2.83E+02
1.01E+02
2.16E+02
7.70E+01
1.36E+02
1.31E+02
3.75E+02
2.98E+02
9.80E+01
2.15E+02
2.37E+02
2.87E+02
6.8VE+01
6.30E+01
4.24E+01
1.02E+02
-1.20E+01
Ref.
a
b
b
a
a
a
a
a
a
a
a
a
a
a
qq
f
a
f
f
f
f
a
f
f
f
a
b
a
a
f
a
a
f
f
f
b
b
f
Aqueous
Solubility
(mg/L)(1)
4.00E+03
1.00E+03

1.01E+01
1.00E+02
1.11E+02
2.80E+02
5.50E+03
8.69E+03
4.00E+02
6.00E+02
2.70E+03
2.70E+03
2.80E+03
1.00E+04
1.50E+01
9.28E+02
1.91E+01
5.25E+00
4.41 E+00
4.60E+03
4.29E+03
miscible
8.00E-03
7.90E+04
1.40E+02
4.32E+03
2.65E-01
1.69E+00
3.00E+00
2.55E+00
1.10E+00
9.00E-04
9.50E+00
5.00E+01
1.40E+04
1.06E+03
4.90E+01
Ref.
a
b

a
a
a
a(25)
a
a
a
a
a
a
a
b
m
a
b(25)
b(25)
b(25)
k(25)
a
b
b(25)
b
f(15)
b
a(25)
a(25)
f
a
a(22)
f(25)
f
f
b
b(23)
f
Vapor
Pressure
(mm Hg) (2)
7.60E+01
8.00E-01
6.88E+02 _j
1.40E-05
1.00E+00
2.30E+00
4.25E+03
1.82E+02
6.40E+01
4.95E+02
2.65E+02
4.20E+01
2.50E+01
2.50E+01
1.20E-02
6.97E-01
1.65E-03
2.00E+02
4.00E+01
1.00E+01
6.21 E-02
1.65E-03
3.00E+01
3.00E-01
7.28E+01
7.00E+00
1.10E+01
1.00E-02
1.00E+01
3.50E+01
1.50E-01
8.10E-02
1.00E+00
1.20E+02
1.00E+02
3.75E+02
4.00E+01
7.60E+02
Ref.
a
b
b
a(25)
a
a(25)
a
a
a
a
a
a
a
a
b
m
a(25)
f
b(13.9)
b(17.1)
k
a(25)
f
f
f
f
b
a
a(146)
f
a
a(25)
t(105)
f
f(13)
b
b(24)
f(-11.7)
                                       242

-------
Table A-7. Chemical properties of selected organic chemicals (after Cohen and Mercer, 1993)
COMPOUND
Isobutanol
Isopentane
Malathion
Methanol
Methylcyclohexane
Methylene chloride
Methyl Isobutyl Ketone
Methyl Ethyl Ketone
2-Methylpentane
2-Methylhexane
2-Methylnapthalene
Napthalene
Nitrobenzene
Nitroethane
1-Nitropropane
2-Nitrotoluene
3-Nitrotoluene
n-Nonane
n-Octadecane
n-Octane
Parathion
PCB-1016
PCB-1221
PCB-1232
PCB-1242
PCB-1248
PCB-1254
Pentachloroethane
n-Pentane
n-Pentadecane
1-Pentene
Phenanthrene
Phenol
Pyrene
Quinoline
Styrene
1 ,1 ,2,2-Tetrabromoethane
1 ,1 ,2,2-Tetrachloroethane
Absolute
Viscosity
(CP)(1)

2.33E-01

6.14E-01

4.30E-01

4.10E+01




2.01 E+00
6.61 E-01
J.98E-01
2.37E+00



5.42E-01
1.53E+01
1.93E+01
4.84E+00
8.20E+00
2.40E+01
6.50E+01
7.00E+02
2.75E+00
2.89E-01
2.55E+00


1.27E+01

3.00E+00
7.51 E-01
9.79E+00
1.75E+00
Ref.

y

ff

c

bb




c
d(25)
d(25)
d



y(o)
w
9(38)
9(38)
9(38)
9(38)
9(38)
9(38)
d(15)
Y(0)
V


rr

bb(30)
ss
d
c
Boiling
Point
(deg.C)
1.08E+02
2.80E+01
1.20E-I-02
6.50E+01
1.01 E+02
4.00E+01
1.16E+02
7.96E+01
6.00E+01
9.00E+01
2.41 E+02
2.18E+02
2.11 E+02
1.15E+02
1.30E+02
2.22E+02
2.33E+02
1.51 E+02
3.17E+02
1.26E+02
3.75E+02
3.25E+02
2.75E+02
2.90E+02
3.25E+02
3.40E+02
3.65E+02
1.59E+02
3.60E+01
2.70E+02
3.00E+01
3.40E+02.
1.82E+02
3.93E+02
2.38E+02
1.45E+02
2.39E+02
1.46E+02
Ref.
f
f
0
f
f
a
f
f
f
f
a
f
a
b
b
b
b.
f
f
f
b
a
a
a
a
a
a
b
f
f
f.
f
f
a
f
f
b
a
Aqueous
Solubility
(mg/L){1)
9.50E+04
4.80E+01
1.45E+02
misicible
1.40E+01
2.00E+04
1.90E+04
3.53E-01
1.30E+01
2.54E+00
2.46E+01
3.00E+01
1.90E+03
4.50E+04
1.40E+04
6.00E+02
5.00E+02
7.00E-02
7.00E-03
6.60E-01
1.20E+01
2.30E-01
5.90E-01
1.45E+00
2.00E-01
5.00E-02
5.00E-02
5.00E+02
3.95E+01

1.48E+02
1.60E+00
8.20E+04
1.60E-01
6.00E+04
2.80E+02
7.00E+02
2.90E+03
Ref.
f(18)
f
b
99
f
a
f
f(10)
b(25)
b(25)
a(25)
f(25)
a
b
b
b
b
f
f(25)
f
b
a
a(24)
a(25)
a
a
a
b
b(25)

b(25)
a(15)
a(15)
a(26)
f
f(15)
b
a
Vapor
Pressure
(mm Hg) (2)
1.00E+01
5.93E+02
1.25E-06
9.20E+01
1.44E+02
3.49E+02
6.00E+00
7.75E+01
4.00E+02
4.00E+01

1.00E+00
1.50E-01
1.56E+01
7.50E+00
1.50E-01
1.50E-01
3.22E+00

1.10E+01
4.00E-04
4.00E-04
6.70E-03
4.60E-03
1.00E-03
4.94E-04
6.00E-05
3.40E+00
4.30E+02
1.00E+00
1.00E+02
6.80E-04
2.00E-01
6.85E-07
1.00E+00
5.00E+00
1.00E-01
5.00E+00
Ref.
f(25)
f
b
f
f
a
f
f
f(42)
f(14.9)

f(53)
a
b
b
b
b
f

f
b
a(25)
a(25)
a(25)
a
a(25)
a
b
f
v(91.6)
f(-18)
a(25)
f
a(25)
f(59.7)
a
b
a
                                        243

-------
Table A-7.  Chemical properties of selected organic chemicals (after Cohen and Mercer, 1993).
COMPOUND
Tetrachloroethene
n-Tetradecane
Tetrahydrofuran
Thlophene
Toluene
1 ,2,4-Trichlorobenzene
1 ,1 ,1-Trichloroethane
1 ,1 ,2-Trichloroethane
Trichloroethene
1 ,1 ,2-Trichlorofluoromethane
1 ,2,3-Trichloropropane
1 ,1 ,2-Trichlorotrifluoroethane
n-Tridecane
1 ,3,5-Trimethylbenzene
1 ,1 ,3-Trimtehylcyclohexane
2,2,5-Trimethylhexane
2,2,4-Trimethylpentane
2,3.4-Trimethylpentane
2,4,6-TrimethyIphenol
Tri-o-cresyl phosphate
Triphenylene
n-Undecane
Vinyl Chloride
m-Xylene
o-Xylene
p-Xylene

Water
Absolute
Viscosity
(CP)(1)
8.90E-01
2.15E+00

6.54E-01

1.42E+00
1.20E+00
1.20E-01
5.70E-01
4.20E-01


1.78E+00



<1.26


8.00E+01


1.07E-02
6.20E-01
8.10E-01
6.48E-01

1.00E+00
Ref.
c
V

d

c
c
c
c
c(25)


V



V


d


uu
rr
rr
rr


Boiling
Point
(deg.C)
1.21E+02
2.52E+02
6.57E+01
8.40E+01
1.11E+02
2.10E+02
7.40E+01
1.14E+02
8.70E+01
2.40E+01
1.42E+02
4.80E+01
2.26E+02
1.65E+02
1.38E+02
1.24E+02
9.93E+01
1.13E+02
exp>300
4.10E+02
4.38E+02
1.96E+02
-1.39E+01
1.39E+02
1.44E+02
1.38E+02

1.00E+02
Ref.
a
f
f
b
f
a
a
a
a
a
b
b
f
f
b
b
f
b
f
b
f
f
f
f
a
f


Aqueous
Solubility
(mg/L) (1)
1.50E+02
2.20E-03
misicible
3.60E+03
4.70E+02
1.90E+01
1.36E+03
4.50E+03
1.10E+03
1.10E+03

2.00E+02
1.30E-02
4.82E+01
1.77E+00
5.40E-01
1.14E+00
1.36E+00
1.40E+04
3.00E-01
3.80E-02
7.00E-03
1.10E+03
1.73E+02
1.52E+02
2.00E+02


Ref.
a
f(25)
b
b(18)
f(16)
a(22)
a
a
a
a

b
f(25)
b(25)
b(25)
b(25)
b(25)
b(25)
f
b
f(26)
n
a(25)
a(25)
a
a(25)


Vapor
Pressure
(mm Hg) (2)
1.40E+01
1.00E+00
1.45E+00
6.00E+01
1.00E+01
4.00E-01
1.00E+02
1.90E+01
5.78E+01
6.87E+02
2.00E+00
2.84E+02
1.00E+00
1.00E+01

1.65E+01
4.00E+01
4.00E+01
<1
1.00E+01

1.00E+00
2.40E+02
6.00E+00
5.00E+00
6.50E+00

1.75E+01
Ref.
a
f(76)
b
b
f(6.4)
a(25)
a
a
a
a
b
b
v(59.4)
b(47.4)

b(25)
b(20.7)
b(32)
vv
v(265)

f(32.7)
a(-40)
f
f
a


                                       244

-------
Table A-7.  Chemical properties of selected organic chemicals (after Cohen and Mercer, 1993).
COMPOUND
Acetone
Aniline
o-Anisidine
Benzene - - -
Benzo(a)anthracene
Benzoic Acid
Benzo(a)pyrene
Benzyl alcohol
Benzyl chloride
3is(2-chloroethyl)ether
3is(2-chloroisopropyl)ether
Bromobenzene
Bromochloromethane
Bromodichloromethane
Bromoethane
Bromoform
n-Butane
1-Butene
Butyl benzyl phthalate
Carbon disulfide
Carbon tetrachloride
Chlorobenzene '•
Chloroethane
2-Chloroethyl vinyl ether
Chloroform
Chloromethane
1-Chloro-1 -nitroprppane
2-Chlorophenol
4-Chlorophenyl phenyl ether
Chloropicrin ;
m-Chlorotoluene
o-Chlorotoluene
p-Chlorotoluene
Chrysene
o-Cresol
Cyclohexanone
Cyclohexane
n-Decane
Henry's Law
Constant (2)
(atm-m3/mol)
3.97E-05
1.36E-01
1.25E-06
5.48E-03
6.60E-07
7.02E-08
< 2.4e-6

3.04E-04
1.30E-05
1.10E-04
2.40E-03
1.44E-03
2.12E-04
7.56E-03
5.32E-04
9.30E-01
2.50E-01
1.30E-06"
1.33E-02
3.02E-02
4.45E-03
1.11E-02
2.50E-04
3.20E-03
8.82E-03
1.57E-01
8.28E-06
2.20E-04
8.40E-02
1.60E-02
6.25E-03
1 .70E-02
7.26E-20
1.50E-06
1.20E-05
1.94E-01
1.87E-01
Ref.
a
b
b
a
a
a
a
a
a
b
a
b
b
a
b
a
b
b
a
a
a
a
a
a
a
a
b
a
a
b
e
e
e
a
k
b
b
b
Surface
Tension
dyn/cm) (1)
2.37E+01
4.29E+01

2.89E+01



3.90E+01
3.75E+01
3.79E+01

3.58E+01
3.33E+01

2.45E+0/I
4.55E+01
1.49E+01


3.23E+01
2.70E+01
3.32E+01
1.95E+01

2.72E+01


4.03E+01

3.23E+01
3.28E+01
3.29E+01
3.46E+01


3.40E+01
2.46E+01
2.57E+01
Ref.
bb
c

aa



aa
aa
c

e
h

h
c
II


c
c
c
aa

c


e

mm
e
h(25)
h(25)


aa
mm
II
Interfacial
Liquid
Tension
dyn/cm) (1)

5.77E+00









3.98E+01







4.84E+01
4.50E+01
3.74E+01


3.28E+01




3.00E+01








Ref.











j







j
j
j


j




mm








Log Kow
(mL/g)
-2.40E-01
9.00E-01
9.50E-01
2.13E+00
5.61 E+00
1.87E+00
5.99E+00
1.10E+00
2.30E+00
1.58E+00
2.58E+00
3.01 E+00
1.41 E+00
1.88E+00
1.57E+00
2.30E+00
2.89E+00
2.40E+00
4.78E+00
1.84E+00
2.83E+00
2.84E+00
1.43E+00
1.28E+00
1.95E+00
9.00E-01
4.25E+00
2.16E+00
4.08E+00
1.03E+00
3.28E+00
3.42E+00
3.30E+00
5.60E+00
8.91 E+02
8.10E-01
3.44E+00
6.69E+00
Ref.
a
b
b
a
a
a
a
a
a
b
a
b
b
a
b
a
b
b
a
a
a '
a
a
a
a
a *
b
a
a
b
e
f.
e
a
k
b
b-
b
                                        245

-------
Table A-7.  Chemical properties of selected organic chemicals (after Cohen and Mercer, 1993).
COMPOUND
Dibromochloromethane
1 ,2-Dibromo-3-chloropropane
Dibromodifluoromethane
Dibutyl phthalate
1 ,2-Dichlorobenzene
1 ,3-DIchlorobenzene
Dichlorodifluoromethane
1,1-Dichloroethane
1,2-DIchloroethane
1,1-D!ch!oroethene
trans-1 ,2-Dichloroethene
1,2-Dichloropropane
cis-1 ,3-Dichloropropene
trans-1 ,3-Dichloropropene
Dfchlorvos
1,4-Diethylbenzene
Diethyl phthalate
2,3-Dimethylbutane
2.3-Dimethylpentane
2,4-Dimethylpentane
2,4-DImethylphenol
Dimethyl phthalate
1 ,4-Dioxane
n-Dodecane
Ethyl Acetate
Ethylbenzene
Ethylene dibromlde
Fluoranthene
Fluorene
n-Heptane
Hexachlorobutadiene
Hexachlorocyctopentadiene
n-Hexadecane
n-Hexane
1-Hexene
lodomethane
1-lodopropane
Isobutane
Henry's Law
Constant (2)
(atm-m3/mol
9.90E-04
2.49E-04

6.30E-05
1.90E-03
3.60E-03
3.00E+00
4.30E-03
9.10E-04
2.10E-02
3.84E-01
2.30E-03
1.30E-03
1.30E-03
5.00E-03
4.56E-03
8.46E-07
1.94E+00
1.73E+00
3.15E+00
2.83E-06
4.20E-07
4.88E-06
2.42E+01
1.34E-04
6.60E-03
7.06E-04
1.69E-02
2.10E-04
2.04E+00
2.60E-02
1.60E-02
2.28E-01
1.18E+00
4.35E-01
5.48E-03
9.09E-03

Ref
a
b

a
a
a
a
a
a
a
a
a
a
a
b
m
a
b
b
b
k
a
b
b
b
a
b
a
a
b
a
a
u
b
b
b
b

Surface
Tension
(dyn/cm) (1



3.34E+01
3.70E+01
3.32E+01
9.00E+00
2.48E+01
3.22E+01
2.40E+01
2.50E+01
2.87E+01
3.12E+01



3.75E+01





3.69E+01

2.40E+01
3.15E+01
3.87E+01




3.75E+01



3.10E+01


Ref.



c
c
c
pp
c
c
c(15)
c
c
e



c





kk

oo
bb
c




e



e


Interfacial
Liquid
Tension
(dyn/cm) (1)




4.00E+01



3.00E+01
3.70E+01
3.00E+01

2.38E+01













3.65E+01











Ref.




e



e(30)
e(23)
e

e(27)













e











Log Kow
(mL/g)
2.08E+00
2.63E+00

4.57E+00
3.40E+00
3.38E+00
2.16E+00
1.78E+00
1.48E+00
2.13E+00
2.09E+00
2.28E+00
1.41E+00
1.41E+00
1.40E+00

2.35E+00
3.82E+00
3.26E+00
3.24E+00
2.63E+02
1.61E+00
4.20E-01
5.64E+00
6.60E-01
3.13E+00
1.76E+00
5.22E+00
4.12E+00
4.66E+00
4.78E+00
5.04E+00
8.25E+00
3.90E+00
2.25E+00
.69E+00
2.49E-f-00

Ref.
a
b

a
a
a
a
a
a
a
a
a
a
a
b

a
b
b
b
k
a
b
b
b
a
b
a
a
b
a
a
s
b
b
b
b

                                      246

-------
Table A-7.  Chemical properties of selected organic chemicals (after Cohen and Mercer, 1993).
COMPOUND
sobutanol
Isopentane
Malathion
Methanol
Methylcyclohexane
Methylene chloride
Methyl Isobutyl Ketone
Methyl Ethyl Ketone
2-Methylpentane
2-Methylhexane
2-Methylnapthalene
vlapthalene
Mitrobenzene
Nitroethane
1-Nitropropane
2-Nitrotoluene
3-Nitrotoluene
n-Nonane
n-Octadecane
n-Octane
Parathion
PCB-1016
PCB-1221
PCB-1232
PCB-1242
PCB-1248
PCB-1254
Pentachloroethane
n-Pentane
n-Pentadecane
1-Pentene
Phenanthrene
Phenol
Pyrene
Quinoline
Styrene
1 ,1 ,2,2-Tetrabromoethane
1 ,1 ,2,2-Tetrachloroethane
Henry's Law
Constant (2)
(atm-m3/mol)

1.40E+00
4.89E-09

4.35E-01
2.00E-03


1.73E+00
3.42E+00

4.60E-04
2.45E-05
4.66E-05
8.68E-05
4.51 E-05
5.41 E-05
5.95E+00

3.23E+00
8.56E-08

3.24E-04
4.64E+00
5.60E-04
3.50E-03
2.70E-03
2.45E-03
1.26E+00

4.06E-01
3.90E-05
2.70E-07
1.09E-05

2.61 E-03
6.40E-05
3.80E-04
Ref.

I
b

b
a


b
b

a
a
b
b
b
b
b

b
b

a
a
a
a
a
b
b

b
a
a
a

a
b
a
Surface
Tension
dyn/cm)(1)

1.37E+01
3.71 E+01
2.26E+01

2.79E+01
2.36E+01
2.46E+01



3.18E+01
4.30E+01
3.13E+01


3.91 E+01


2.18E+01
3.92E+01






3.47E+01
1 .60E+01



3.99E+01

4.50E+01
3.21 E+01
4.97E+01
3.60E+01
Ref.

y
aa
ff

c
aa
bb



aa
c
aa


aa


y
w






e
aa



rr

aa
aa
X
c
Intel-facial
Liquid
Tension
dyn/cm) (1)





2.83E+01






2.57E+01

























Ref.





j






j


V






















Log Kow
(mL/g)
7.60E-01
2.30E+00
2.89E+00
-7.70E-01
2.86E+00
1.30E+00
1.19E+00
2.60E-01
2.77E+00
3.30E+00
4.11E+00
3.36E+00
1.95E+00
1.80E-01
8.70E-01
2.30E+00
2.42E+00
4.67E+00

5.18E+00
3.81 E+00
5.88E+00
2.80E+00
3.20E+00
4. 11 E+00
6. 11 E+00
6.47E+00
2.89E+00
3.23E+00

2.26E+00
4.57E+00
1.48E+00
4.88E+00
2.03E+00
2.95E+00
2.91 E+00
2.56E+00
Ref.
ee
I
b
ee
b
a
hh
z
b
b
a
a
a
b
b
b
b
b

b
b
a
a
a
a
a
a
b
b

b
a
a
a
•z
a
b
a
                                        247

-------
Table A-7.  Chemical properties of selected organic chemicals (after Cohen and Mercer, 1993).
COMPOUND
Tetrachloroethene
n-Tetradecane
Tetrahydrofuran
Thfophene
Toluene
1 ,2,4-Trichlorobenzene
1,1,1-Trichloroethane
1 ,1 ,2-Trichloroethane
Trichloroethene
1,1 ,2-Trichlorofluoromethane
1 ,2,3-Trichloropropane
1 ,1 ,2-Trichlorotrifluoroethane
n-TrWecane
1 ,3,5-Trimethylbenzene
1,1,3-Trimtehyloyclohexane
2,2,5-Trimethylhexane
2,2,4-Trimethylpentane
2,3.4-Trimethylpentane
2,4,6-Trimethylphenol
Tri-o-cresyl phosphate
Triphenylene
n-Undecane
Vinyl Chloride
m-Xylene
o-Xytene
p-Xylene

Water
Henry's Law
Constant (2)
(atm-m3/mol)
1.53E-02

7.06E+05
2.93E-03
6.70E-03
2.32E-03
1.80E-02
7.40E-04
9.10E-03
1.10E-01
3.18E-04
3.33E-01

3.93E-03

2.42E+00
3.01 E+00
2.98E+00




2.78E+00
7.00E-03
5.27E-03
7.10E-03


Ref.
a

b
b
a
a
a
a
a
a
b
b

b

b
b
b




a
a
a
a


Surface
Tension
(dyn/cm) (1)
3.13E+01

2.80E+01

2.90E+01
3.91 E+01
2.54E+01
3.40E+01
2.93E+01
1.90E+01

1.73E+01










2.31 E+01
2.89E+01
3.01 E+01
2.83E+01

7.28E+01
Ref.
c

aa

aa
c
c
c
c
c

bb










uu
rr
rr
rr


Interfacial
Liquid
Tension
(dyn/cm) (1)
4.44E+01



3.61 E+01

4.50E+01

3.45E+01



















Ref.
e(25)



aa

e

e(24)



















Log Kow
(mL/g)
2.60E+00
7.20E+00
4.60E-01
1.81 E+00
2.65E+00
4.02E+00
2.47E+00
2.18E+00
2.53E+00
2.53E+00

2.57E+00

3.42E+00
N/A
3.88E+00
5.83E+00
3.78E+00

5. 11 E+00


3.90E-01
3.20E+00
2.95E+00
3.18E+00


Ref.
a
s
b
b
a
a
a
a
a
a

b

b
b
b
b
b

b


a
a
a
a


                                       248

-------
Table A-7.  Chemical properties of selected organic chemicals (after Cohen and Mercer, 1993).
COMPOUND
Acetone
Aniline
o-Anisidine
Benzene
Benzo(a)anthracene
BenzoicAcid
Benzo(a)pyrene
Benzyl alcohol
Benzyl chloride
Bis(2-chloroethyl)ether
Bis(2-chloroisopropyl)ether
Bromobenzene
Bromochloromethane
Bromodichloromethane'
Bromoethane
Bromoform
n-Butane
1-Butene
Butyl benzyl phthalate
Carbon disulfide
Carbon tetrachloride
Chlorobenzene
Chloroethane
2-Chloroethyl vinyl ether
Chloroform
Chloromethane
1-Chloro-1-nitropropane
2-Chlorophenol
4-ChlorophenyI phenyl ether
Chloropicrin
m-Chlorotoluene
o-Chlorotoluene
p-Chlorotoluene
Chrysene
o-Cresol
Cyclohexanone
Cyclohexane
n-Decane
Log Koc
(mL/g)
-4.30E-01
1.41E+00

1.69E+00
6.14E+00
2.26E+00
5.60E+00
1.98E+00
2.28E+00
1.15E+00
1.79E+00
2.33E+00
1.43E+00
1.79E+00
2.67E+00
2.45E+00
N/A
N/A
2.32E+00
2.47E+00
2.64E+00
1.68E+00
5.10E-01
8.20E-01
1.64E+00
1.40E+00
3.34E+00
2.56E+00
3.60E+00
8.20E-01
3.08E+00
3.20E+00
3.08E+00
5.39E+00

N/A
N/A
N/A
Ref.
a
b

a
a
a
a
a
a
b
a
b
b
a
b
a
b
b
a
a
a
a
a
a
a
a
b
a
a
b
e
e
e
a

b
b
b
Air
Diffusion
Coefficient
(sq.cm./sec)
9.89E-02
7.50E-02
7.52E-02
8.74E-02
5.86E-02
8.04E-02
5.72E-02
7.90E-02
7.43E-02
7.29E-02
6.37E-02
7.59E-02
9.28E-02
8.35E-02
9.03E-02
7.94E-02
8.83E-02
8.98E-02
4.71 E-02
8.92E-02
7.97E-02
7.50E-02
9.82E-02
7.97E-02
9.90E-02

7.83E-02
'7.82E-02
6.04E-02
7.74E-02
7.35E-02
7.38E-02
7.33E-02
5.86E-02
7.90E-02
7.98E-02
7.98E-02
5.83E-02
Ref.
ww
c(30)
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
h
i
c(30)
ww
ww
i

ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
Water
Diffusion
Coefficient
(sq.cm/sec)



















1.10E-05

7.90E-06


9.10E-06













Ref



















h

h


h













Est.
Half-life
in Soil
(days)


28-180





0.62-12
28-180
18-180




28-180


1-7

180-360
68-150


28-180













Est.
Half-life in
Groundwater
(days)


56-360





0.62-12
56-360
36-360




56-360


2-180

7-360
136-300


56-1800













                                        249

-------
Table A-7.  Chemical properties of selected organic chemicals (after Cohen and Mercer, 1993).
COMPOUND
Dlbromochloromethane
1 ,2-Dibromo-3-chloropropane
Dibromodifluoroniethane
Dibutyl phthalate
1,2-Dlchlorobenzene
1 ,3-Dichlorobenzene
Dlchlorodifluoromethane
1,1-Dichloroethane
1,2-Dichloroethane
1,1-Dichloroethene
trans-1 ,2-Dichloroethene
1,2-Dichloropropane
cis-1 ,3-Dichloropropene
trans-1 ,3-Dichloropropene
Dichlorvos
1 ,4-Diethy Ibenzene
Diethyl phthalate
2,3-Dimethylbutane
2,3-Dimethylpentane
2,4-Dimethylpentane
2,4-Dimethylphenol
Dimethyl phthalate
1,4-Dloxane
n-Dodecane
Ethyl Acetate
Ethylbenzene
Ethylene dibromide
"luoranthene
"luorene
n-Heptane
Hexachlorobutadiene
•{exachlorocyclopentadiene
n-Hexadecane
n-Hexane
1-Hexene
lodomethane
1-lodopropane
Isobutane
Log Koc
(mL/g)
1.92E+00
2.11E+00

3.14E+00
2.27E+00
2.23E+00
2.56E+00
1.48E+00
1.15E+00
1.81E+00
1.77E+00
1.71E+00
1.68E+00
1.68E+00
9.57E+00

1.84E+00
N/A
N/A
N/A
2.22E+02
1.63E+00
5.40E-01
N/A
N/A
1.98E+00
1.64E+00
4.62E+00
3.70E+00
N/A
3.67E+00
3.63E+00

N/A
N/A
1.36E+00
2.16E+00

Ref.
a
b

a
a
a
a
a
a
a
a
a
a
a
b

a
b
b
b
k
a
b
b
b
a
b
a
a
b
a
a

b
b
b
b

Air
Diffusion
Coefficient
(sq.cm./sec)
8.13E-02
7.10E-02
7.88E-02
4.20E-02
7.40E-02
7.36E-02
9.22E-02
8.90E-02
8.90E-02
9.11E-02
9.11E-02
9.11E-02
8.33E-02
8.20E-02
6.24E-02
6.48E-02
5.60E-02
7.35E-02
6.90E-02
6.81 E-02
7.16E-02
6.17E-02
8.76E-02
5.36E-02
8.28E-02
7.36E-02
8.13E-02
6.19E-02
6.71 E-02
6.85E-02
5.99E-02
6.10E-02
4.67E-02
7.34E-02
7.50E-02
9.47E-02
7.79E-02
8.56E-02
Ref.
ww
ww
ww
c(25)
ww
ww
ww
i
i
i
i
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
Water
Diffusion
Coefficient
(sq.cm/sec)



4.10E-05





9.50E-06
9.50E-06



























Ref



c





h
c



























Est.
Half-life
in Soil
(days)
28-180
28-180

2-23
28-180
28-180

32-154
100-180
28-180

167-1289
5-11
5-11


3-56




1-7




28-180



28-180
7-28



7-28


Est.
Half-life in
Groundwater
(days)
14-180
56-360

2-23
56-360
56-360

64-154
100-360
56-132

334-2592
5-11
5-11


6-112




2-14




20-120



56-360
7-56



14-56


                                       250

-------
Table A-7.  Chemical properties of selected organic chemicals (after Cohen and Mercer, 1993).
COMPOUND
sobutanol
sopentane
Malathion
\/Iethanol
Methylcyclohexane
Methylene chloride
Methyl Isobutyl Ketone
Methyl Ethyl Ketone
2-Methylpentane
2-Methylhexane
2-Methylnapthalene
Napthalene
Nitrobenzene
Nitroethane
1-Nitropropane
2-Nitrotoluene
3-Nitrotoluene
n-Nonane
n-Octadecane
n-Octane
3arathion
PCB-1016
PCB-1221
PCB-1232
PCB-1242
PCB-1248
PCB-1254
Pentachloroethane
n-Pentane
n-Pentadecane
1-Pentene
Phenanthrene
Phenol
Pyrene
Quinoline
Styrene
1 ,1 ,2,2-Tetrabromoethane
1 ,1 ,2,2-Tetrachloroethane
Log Koc
(mL/g)


2.46E+00

. N/A
9.40E-01


N/A
N/A
3.93E+00
2.74E+00
2.01 E+00




N/A

N/A
3.07E+00
4.70E+00
2.44E+00
2.83E+00
3.71 E+00
5.64E+00
5.61 E+00
3.28E+00
N/A

N/A
3.72E+00
1 .43E+00
4.66E+00

2.87E+00
2.45E+00
2.07E+00
Ref.


b

b
a


b
b
a
a
a




b

b
b
a
a
a
a
a
a
b
b

b
a
a
a

a
b
a
Air
Diffusion
Coefficient
sq.cm./sec)
ERR
7.90E-02
4.78E-02
1 .42E-01
7.30E-02
1.02E-01
7.33E-02
8.82E-02
7.31E-02
6.83E-02
6.73E-02
6.97E-02
7.20E-02
9.60E-02
8.62E-02
7.30E-02
7.29E-02
6.12E-02

6.46E-02
5.15E-02
5.62E-02
6.18E-02
5:89E-02
5.70E-02
5.46E-02
5.28E-02
7.03E-02
7.93E-02
4.82E-02
8.14E-02
5.91 E-02
8.54E-02
6.22E-02
7.33E-02
7.58E-02
6.85E-02
7.54E-02
Ref.
ww
ww
ww
ww
ww
i
ww
ww
ww
ww
ww
ww
h
ww
ww
ww
ww
ww

ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
Water
Diffusion
Coefficient
sq.cm/sec)





1.10E-06






7.60E-06

























Ref





c






c

























Est.
Half-life
in Soil
(days)


3-7


7-28






12-197

28-180






















0.45-45
Est.
Half-life in
Groundwater
(days)


8-103


14-56






2-197

56-360






















0.45-45
                                        251

-------
Table A-7. Chemical properties of selected organic chemicals (after Cohen and Mercer, 1993)
COMPOUND
Tetrachloroethene
n-Telradecane
Tetrahydrofuran
Thfophene
Toluene
1 ,2,4-Trichlorobenzene
1 ,1 ,1-Trichloroethane
1 , 1 ,2-Trich!oroethane
Trichloroethene
1 ,1 ,2-Trichlorofluoromethane
1 ,2,3-Trichloropropane
1 ,1 ,2-TrichlorotrifIuoroethane
n-Tridecane
1 ,3,5-Trimelhylbenzene
1 ,1 ,3-Trimtehylcyclohexane
2,2,5-Trimethylhexane
2,2,4-Trimethylpentane
2,3,4-Trimethylpentane
2,4,6-Trimethylphenol
Tri-o-cresyl phosphate
Triphenylene
n-Undecane
Vinyl Chloride
m-Xylene
o-Xylene
p-Xylene

Water
Log Koc
(mL/g)
2.42E+00

N/A
1.73E+00
2.06E+00
3.98E+00
2.18E+00
1.75E+00
2.10E+00
2.20E+00

2.59E+00

3.21 E+00
N/A
N/A
N/A
N/A

3.37E+00


6.00E-01
3.20E+00
2. 11 E+00
2.31 E+00


Ref
a

b
b
a
a
a
a
a
a

b

b
b
b
b
b

b


a
a
a
a


Air
Diffusion
Coefficient
(sq.cm./sec
7.40E-02
4.98E-02
9.19E-02
9.10E-02
7.95E-02
6.97E-02
7.96E-02
7.90E-02
8.11E-02
8.09E-02
7.59E-02
7.08E-02
5.16E-02
6.88E-02
6.36E-02
6.08E-02
6.41 E-02
6.52E-02
6.75E-02
5.60E-02


1.00E-01
7.36E-02
7.41 E-02
7.42E-02

2.18E-01
Ref.
i
ww
ww
ww
ww
ww
i
h
i
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww
ww


ww
ww
ww
ww

ww
Water
Diffusion
Coefficient
(sq.cm/sec
7.50E-06





8.00E-06
8.00E-06
8.30E-06



















Rft






h
h
c



















Est.
Half-life
in Soil
(days)
180-360




28-180
140-273
136-360
180-360
180-360
180-360
180-360
















Est.
Half-life in
Groundwate
(days)
360-720




56-360
140-546
136-720
321-1653
360-720
360-720
360-720
























252

-------
Table A-7.    Chemical properties of selected organic chemicals ~ references and notes.


a = Montgomery, J.H. and L.M. Welkom, 1990.  Groundwater Chemicals Desk Reference, Lewis
       Publishers, Chelsea, Michigan.
b = Montgomery, J.H., 1991.  Groundwater Chemicals Desk Reference, Lewis Publishers,
       Chelsea, Michigan.
c = Lucius, J.E., G.R. Olhoeft, P.L. Hill and S.K. Duke, 1990.  Properties and hazards of 108
       selected substances, USGS Open-File Report 90-408.
d = Mercer, J.W., D.C. Skipp and D. Giffm, 1990.  Basics of pump-and-treat ground-water
       remediation technology, USEPA-600/8-90/003, Robert S. Kerr Environmental Research
       Laboratory, Ada, Oklahoma.
e = Mercer, J.W. and R.M. Cohen, 1990. A review of immiscible fluids in the subsurface:
       Properties, models, characterization and remediation, Journal of Contaminant Hydrology.
f = Verschueren, K.,  1993.  Handbook of Environmental Data on Organic Chemicals, Van
       Nostrand Reinhold, New York.                                     ,
g = Monsanto, 1988. Polychlorinated biphenyls material safety data sheets.
h = Tetra Tech, Inc.,  1988.  Chemical data for predicting the fate of organic chemicals in water,
       Volume 2: Database, EPRIEA-5818, Volume 2, Electric Power Research Institute, Palo
       Alto, California.
i = Mendbza, C.A. and E.O. Frind, 1990b.  Advective-dispersive transport of dense organic
       vapors in the unsaturated zone, 2.  Sensitivity analysis, Water Resources Research,
       26(3).
j = Dean, J.A., ed., 1973. Lange's Handbook of Chemistry, Eleventh Edition, McGraw-Hill Book
       Co., New York.
k = USEPA, 1990b.  Basics of pump-and-treat ground-water remediation technology.
       EPA/600/8-90/003.. .
1 = Lyman, W.J., W.F. Reehl, and D.H. Rosenblatt, 1982. Handbook of Chemical Property
       Estimation Methods, Environmental Behavior of Organic Compounds, McGraw-Hill
       Book Co., New York.
m = USEPA, 1990d. Assessing UST corrective action technologies: Site assessment and
       selection of unsaturated zone treatment technologies. EPA/600/2-90/011.
n = Benson, D.A., D. Huntley, and P.C. Johnson, 1993. Modeling vapor extraction and general
       transport in the presence of NAPL mixtures and nonideal conditions, Ground Water.
       Vol. 31, No. 3, p. 437-445.
o = Waterloo Hydrogeologic Software, 1993. Airflow user's guide. Waterloo, Ontario, Canada.
p = Weast, R.C., Ed. CRC Handbook of Chemistry and Physics, 60th ed. Boca Raton, FL: CRC
       Press.,  1979).
q = TOXNET database
r = Hawley, G.G. The Condensed Chemical Dictionary, 12th ed. New York: Van Nostrand
       Reinhold Co., 1993. (From TOXNET database)
s = Coates, M. et. al., Environmental Science and Technology, Vol 19, 1985. (From TOXNET
       database)

                                         253

-------
Table A-7.   Chemical properties of selected organic chemicals -- references and notes.

t = Weast, R.C., Ed. CRC Handbook of Chemistry and Physics, 73rd ed. Boca Raton, FL: CRC
       Press., 1993).
u = Yaws, C. et. al., Chemical Engineering, November 1991. (From TOXNET database)
v = Patty, F.A., 1982. Patty's Industrial Hygiene and Toxicology, 3rd Edition, Volumes 2a, 2b,
       2c, Clayton, G.D. and F.E. Clayton, Eds. Wiley, New York.
w = Windholz, M., S. Budavari, R.F. Blumetti, and E.S. Otterbein, Eds, 1983. The Merck Index:
       An encyclopedia of chemicals,  drugs, andbiologicals, 10th Edition. Merck, Rahway,
       New Jersey, 2179 p.
x = Weast, R.C., Ed. CRC Handbook of Chemistry and Physics, 67th ed. Boca Raton, FL: CRC
       Press., 1987).
y = Weast, R.C., Ed. CRC Handbook of Chemistry and Physics, 69th ed. Boca Raton, FL: CRC
       Press., 1989).
z = Hansch, C. and A. J. Leo, Substitute Constants for Correlation Analysis in Chemistry and
       Biology, John Wiley, New York (1979).
aa = U.S. Coast Guard, 1985, CHRIS Hazardous Chemical Data.
bb = Kirk-Othmer, Encyclopedia of Chemistry and Technology, 3rd ed.1978. (From TOXNET
       database)
cc = Cheremisinoff, P.N. and A.C. Moresi, 1979. Benzene, Basic and Hazardous Properties.
       M. Dekker, New York, 250 p.
dd = NIOSH, 1981, Pocket Guide to Chemical Hazards, U.S. Government Printing Office,
       Washington, D.C.
ee = Hansch. Log P database, 1987. (From TOXNET database)
ff = Environment Canada,  1981. Styrene: Environmental and technical information for problem
       spills (Draft). Technical Services Branch, Environmental Protection Programs
       Directorate, Environmental Protection Services, Ottawa, Ontario.
gg = Flick, E.W., 1985.  Industrial Solvents Handbook,  3rd Edition. Noyes Publications, Park
       Ridge, New Jersey.
hh = Ginnings, P.M. et. al. Journal of American Chemistry Society, vol 62,1940. (From
       TOXNET database)
ii = Sax, N.I., 1975. Dangerous properties of industrial materials, 4th Edition, Van Nostrand
       Reinhold, New York, 1258 p.
jj = Sax, N.I. and R.J. Lewis, 1987. Hawley's Condensed chemical dictionary, 11th Edition.
       Van Nostrand Reinhold, New York, 1288 p.
kk = Flick, E.W., 1985.  Industrial Solvents Handbook,  3rd Edition. Noyes Publications, Park
       Ridge, New Jersey.
11 = Dean, J. A., 1987. Handbook of organic chemistry.  McGraw-Hill, New York.
mm = U.S. Coast Guard, 1978, CHRIS Hazardous Chemical Data.
nn = Weast, R.C., Ed. CRC Handbook of Chemistry and Physics, 65th ed.  Boca Raton, FL:  CRC
       Press., 1985).
oo = Sax, N.I. Dangerous Properties of Industrial Materials. New York: Van Nostrand Reinhold
       Co., 1984.
                                         254

-------
Table A-7.    Chemical properties of selected organic chemicals -- references and notes.

pp = Osol. Remington's Pharmacy Science, 16th ed, 1980. (From TOXNET database)
qq = Budavari, S., O'Neil, M.J., A. Smith, and P.E. Heckelman, EDS, 1989.  The Merck Index:
       An encyclopedia of chemicals, drugs, and biologicals, llth Edition. Merck, Inc. Rahway,
       New Jersey.
rr = Weast, R.C., Ed. CRC Handbook of Chemistry and Physics, 68th ed. Boca Raton, FL: CRC
       Press., 1988).
ss = Environment Canada, 1981. Styrene: Environmental and technical information for problem
       spills (Draft). Technical Services Branch, Environmental Protection Programs
       Directorate, Environmental Protection Services, Ottawa, Ontario.
tt = NFPA,  1986, Fire Protection Guide to Hazardous Materials, 9th ed. 1986. (From TOXNET
       database)
uu = Braker, W. and A.L. Mossman, 1980. Matheson Gas Data Book,  6th Edition.  Matheson,
       Lynhurst, New Jersey, 711 p.             i
vv = American Conference of Governmental Industrial Hygienists (ACGIH), 1986.
       Documentation of the Threshold Limit Values, 5th Edition. 486 p. ACGIH, Cincinnati,
       Ohio,
ww — Derived from equation 83.
(1) Values are given for 20 degrees C, unless noted in the references column.
(2) Values are given for 25 degrees C, unless noted in the references column.
(3) Most references are compilations, users should cqnsult these sources to determine the origin
of the data.
(4) References with letters p and greater (except for ww) were taken from TOXNET (1994), but
original references are listed where possible.
(5) Estimated half-lives in soil and groundwater are from Howard et al. (1991).
                                          255

-------
GLOSSARY
   256

-------
Adsorption refers to the adherence of ions or
molecules in solution to the surface of solids.

Advection refers to  the process  by which
contaminants  are transported  by bulk  fluid
flow, i.e., the movement of the contaminant is
controlled by the movement of the fluid.

Air  conductivity is calculated in the  same
manner as hydraulic conductivity, i.e.,  K =
kpag/|j,a, where: pa and |j,a are the density and
dynamic viscosity,  respectively, of air.

Airflow model is used to describe a numerical
model that mathematically describes the  flow
of air through the vadose zone.

Air flow pathlines  are lines that describe the
path which air takes  as it flows through the
subsurface.

Air flow stagnation zone is an area where no
air  flow  takes  place  due to, for example,
improper placement of extraction wells,  short
circuiting  to   the   surface,  or  subsurface
heterogeneity.

Air injection is the process by which air is
injected  into the subsurface to  increase the
efficiency  of   SVE  or,  in  the  case  of
bioventing, to promote biodegradation.

Air permeability is defined as the intrinsic
permeability divided by the dynamic viscosity
of air.

Air sparging  refers to  the injection  of air
below  the water  table  to   strip volatile
contaminants from  the saturated zone.

Biodegradation, a subset of biotransformation,
is the biologically  mediated conversion  of a
compound to more  simple products.
Bioventing  is  a  process  by  which air is
injected into  the subsurface  to  stimulate
biodegradation by native microbes.

Bulk density is  the  oven-dried  mass  of a
sample divided by its field volume.

Capillary forces are interfacial forces between
immiscible fluid phases, resulting in pressure
differences between the two phases.

Capillary fringe refers  to the saturated  zone
overlying the water table where fluid is under
tension.

Capillary hysteresis refers to variations in the •
capillary   pressure   versus    saturation
relationship  that  depend  on  whether  the
medium is undergoing imbibition or drainage.
Capillary hysteresis results  from  nonwetting
fluid entrapment and differences in contact
angles during imbibition and drainage that
cause different wetting  and drying curves to
be   followed  depending   on   the   prior
imbibition-drainage history.

Capillary pressure causes porous media to
draw in the wetting  fluid  and repel  the
nonwetting fluid due to the dominant adhesive
force between the wetting fluid and the media
solid surfaces.   For  a  water-NAPL system
with water being the wetting phase,  capillary
pressure equals the NAPL pressure minus the
water pressure.

Compositional flow and transport model is
used  to describe a numerical model  which
simulates  both  the  flow of air  and  the
transport of a multiconstituent contaminant in
the subsurface.

Concentration gradient refers to the change in
concentration with distance across a fluid
medium.
                                         257

-------
Confined flow refers to a flow system with no-
flow  boundaries  both above and below a
1ota»«r»1 -PlrttTf trtfot/arr*
lateral flow system
Cosolvency refers to the interaction of one or
more organic contaminants that may cause
them to behave differently in the.subsurface
than if they were present alone in their pure
form.

Darcy's lanv is the empirical law that relates
the hydraulic head gradient to the flowrate'in
a  porous medium  based  on  the  hydraulic
conductivity of the medium.  Darcy's law is
commonly used to quantify the flow of air
through porous media as well.

Density is the mass  per unit volume of a
substance.

Diffusion refers to mass transfer as a result of
random motion of molecules; it is described
by Pick's law of diffusion.

Dispersion is the spreading and  mixing of
chemical constituents in groundwater caused
by diffusion and mixing due to microscopic
variations in  velocities within  and between
pores.

Distribution coefficient refers to the quantity
of the solute sorbed by the solid per unit
weight of  solid  divided  by  the quantity
dissolved in the water per unit  volume of
water.

DNAPL  is an acronym for denser-than-water
nonaqueous rjhase liquid.  It is synonymous
with  denser-than-water   immiscible-phase
liquid.

Dynamic viscosity is  a proportionality factor
which relates  the shear stress in a fluid to the
time rate of  strain in  the fluid.   Dynamic
viscosity quantifies the tendency of a fluid to
move when a stress is applied.

Effective   air  permeability   (k^)  is   the
permeability of the medium to air when air is
not the only fluid present.  It is a function of
saturation,  and is  maximum  when air is the
only fluid present. This is the parameter that
is  actually measured  during  a  field  air
pumping test.

Effective porosity is the porosity of a porous
medium  that takes into account the presence
of one or more fluids in the pore space, which
effectively  reduces  the   total  pore   space
available for fluid flow.

Equilibrium vapor concentration refers to the
concentration  of a contaminant in the vapor
phase that  has  reached, its  maximum, and
remains  at that point  until some  stress is
applied to the  system.

Pick's law of  diffusion states that  the rate of
change of concentration in a stationary fluid is
directly related to the concentration gradient in
the fluid.

Fingering  refers to the formation of finger-
shaped irregularities at the leading edge of a
displacing  fluid in a porous medium which
move out ahead of the main body of fluid.

Free-phase NAPL refers to immiscible liquid
existing  in the subsurface with  a positive
pressure such that it  can flow into a well.
Also called mobile  DNAPL  or  continuous-
phase DNAPL.

Henry's law constant is the equilibrium ratio
of the partial pressure of a compound in air to
the concentration of the compound in water at
a reference temperature.   It  is  sometimes
referred   to  as   the  air-water  partition
                                         258

-------
coefficient.

Heterogeneity refers to a lack of uniformity in
porous media properties and conditions.

Hydraulic conductivity is a  measure of  the
volume of water  at  the  existing kinematic
viscosity that will  move in a unit time under
a unit hydraulic gradient through a unit area
of medium measured  at right angles to  the
direction of flow.

Hydraulic gradient is the change in head  per
unit distance  in a given direction, typically in
the principal  flow  direction.

Hydrophobic adsorption refers to the tendency
of a  hydrophobic  organic chemical to sorb
more readily  to dry soil solid surfaces than to
solid  surfaces where  water is present.

Ideal gas refers to  a  gas  whose  pressure-
volume-temperature behavior can be described
completely by the  ideal gas law, given by  PV
= nRT, where P is pressure, V is volume, n is
the number of moles  of gas, R is the universal
gas constant,  and T is temperature.

Interface  refers  to  the  thin surface  area
separating two immiscible fluids that are in
contact with  each other.

Interfacial tension is the strength of the film
separating two immiscible fluids (e.g., oil and
water)  measured  in  dynes  (force)   per
centimeter or millidynes per centimeter.

Interphase mass transfer is the net transfer of
chemical  compounds between two  or  more
phases.

Intrinsic permeability is  a measure of  the
relative ease with which a porous medium  can
transmit a liquid under a potential gradient.
Intrinsic permeability  is  a property of the
medium alone that is dependent on the shape
and size of the  openings  through which the
liquid moves.

Kinematic  viscosity  is  defined as  dynamic
viscosity divided by  the density of the fluid.

Klinkenberg effect is the effect caused by gas
slippage along , pore  walls.   Darcy's  law
assumes that the velocity of the fluid along
pore walls is zero, and the Klinkenberg effect
describes the  departure of a gas  from this
assumption.

Local   equilibrium  assumption,   or  LEA,
describes the assumption that in a four-phase
system, on a local (pore-scale)  basis,, the
contaminant is in equilibrium in each of the
four  phases  (vapor,  dissolved,   adsorbed,
NAPL).

LNAPL is  an acronym for  less-dense-than-
water  nonaqueous  phase  liquid.    It  is
synonymous  with   less-dense-than-water
immiscible-phase liquid.

Macropores are relatively large pore spaces
(e.g.,  fractures  and  worm  tubes)   that
characteristically    allow   the    enhanced
movement of liquid and gas in the subsurface.

Mass transport equation is the mathematical
equation  that  'describes   the  transport   of
contaminants in porous media by advection
and diffusion.

Moisture content refers to the amount of water
lost from the  soil upon drying to a constant
weight, expressed  as  the  weight  per  unit
weight of dry soil or as the volume of water
per unit bulk volume of the soil.  For a fully
saturated medium, moisture content equals the
porosity; in the vadose zone, moisture content
                                        259

-------
               ranges between zero and the porosity value for
               the medium.  See porosity, vadose zone.

               Mole is a standard unit of chemical measure,
               and refers to a quantity of a compound which
               contains an Avogadro's  number (6.022xl023)
               of atoms.

               NAPL is an acronym for nonaqueous phase
               liquid.

               Organic carbon content refers to a measure'of
               the organic carbon present in a soil.  Organic
               chemicals in  soil typically  adsorb to  soil
               organic carbon, and the amount of adsorption
               can be  related to  the  soil organic  carbon
               content.

               Partial pressure refers to the portion of total
               vapor pressure in a system due to one or more
               constituents in the vapor mixture.

               Partitioning refers to a chemical equilibrium
               condition where a chemical's concentration is
               apportioned  between  two  different  phases
               according to the partition coefficient, which is
               the ratio of a chemical's concentration in  one
               phase to its concentration in the other phase.

               Permeability or intrinsic permeability is  a
               measure of the relative ease with which  a
               porous medium can transmit a liquid under  a
               potential gradient.  Intrinsic permeability  is  a
               property  of  the  medium  alone  that  is
               dependent on  the  shape and  size of  the
               openings through which the liquid moves.
               Intrinsic permeability can be estimated based
               on saturated hydraulic conductivity.

               Phase refers to a separate fluid that co-exists
               with other fluids.

               Porosity is the volume  fraction of a rock or
               unconsolidated sediment not occupied by solid
material but usually occupied by water and/or
air.  Porosity is a dimensionless quantity.

Radius of influence is the area over which a
vacuum  extraction   well   exerts   enough
influence to cause an appreciable movement of
soil vapor towards the well.

Raoult's law relates the ideal vapor pressure
and  relative  concentration of a chemical  in
solution to  its  vapor  pressure  over the
solution:  PA = XAPA° where PA is the vapor
pressure of  the  solution, XA is  the  mole
fraction of the solvent, and PA° is the vapor
pressure of the pure solvent.  It can similarly
be used to estimate the effective solubility  of
individual DNAPL components in a DNAPL
mixture based on their mole fractions.
Relative  air  permeability  (kjj  is  the
permeability of the porous medium  to  air
when any two  or  more  fluids are present,
expressed  as  a fraction  of  the  intrinsic
permeability of the medium. This parameter
is typically measured  in  the laboratory via
column studies.

Relative permeability is the permeability of
the rock to air, NAPL, or  water, when any
two  or more  are  present,  expressed  as  a
fraction of the single phase permeability of the
rock.

Residual saturation is the  saturation  below
which fluid drainage will not occur.

Retardation  is  the  movement  of a  solute
through a geologic medium at a velocity less
than that of the flowing groundwater due to
sorption or other removal  of the solute or air.

Saturation is the ratio  of the  volume of a
single  fluid  in  the pores  to  pore volume
expressed as a percent and applied to water,
                                                         260
_

-------
DNAPL, or air separately.  The sum of the
saturations of each fluid in a pore volume is
100 percent.

Saturated zone is the zone  of the soil below
the water table  where all space between: the
soil particles is occupied by water.

Screening model is used to  describe a simple
model,  typically  based on   an  analytical
solution,  that  is  used   to   make  rapid
determinations as to the feasibility of SVE!

Soil gas refers to  vapors (gas) in soil above
the saturated zone.

Soil sorption coefficient is  a measure of the
preference of an organic chemical to leave the
dissolved aqueous  phase  in  the   soil and
become attached or adsorbed to soil particles
as organic carbon.

Soil tension describes the force by which
residual water is held in the vadose  zone by
capillary forces.

Solubility  refers  to  the   dissolution  of  a
chemical in a fluid, usually water.  Aqueous
solubility   refers    to   the   maximum
concentrations of a chemical that will dissolve
in pure water at a reference temperature.

Sorption  refers to  processes  that  remove
solutes  from the fluid phase and concentrate
them on the solid phase  of  a medium.

Theis curve, or Theis solution, is an analytical
solution which describes the drawdown in a
confined aquifer based On the pumping rate in
a single well.  This equation is also used to
describe the pressure drawdown associated
with a vacuum extraction well completed in a
confined aquifer.
Transmissivity is calculated for the saturated
zone by multiplying the hydraulic conductivity
by the saturated thickness of an aquifer.

Transpiration describes the process by which
moisture from the soil  is taken up by plants
and released to the atmosphere.

Unconfined  flow  occurs   when  the  top
boundary of a  flow  system is open to the
atmosphere, for example, a water-table aquifer
is an unconfmed system.

Unsaturated  zone refers to the portion of a
porous medium, usually above the water table
in an unconfined aquifer, within which the
moisture content is less than saturation and the
capillary pressure  is less than  atmospheric
pressure.   The unsaturated zone  does  not
include the capillary fringe.

Vacuum extraction  refers  to  the   forced
extraction of gas (with volatile contaminants)
from  the vadose zone,  typically to  prevent
uncontrolled  migration of contaminated soil
gas and augment a site  cleanup.

Vadose  zone is the  subsurface zone that
extends between ground surface and the water
table   and  includes  the  capillary  fringe
overlying the water table.

Vapor pressure is the partial pressure exerted
by  the  vapor   (gas) of a  liquid or solid
substance under equilibrium conditions.  A
relative measure of chemical volatility, vapor
pressure is used to calculate air-water partition
coefficients (i.e., Henry's law constants) and
volatilization rate constants.

Vapor short circuiting occurs when the ground
surface in the vicinity  of a  SVE well  is not
sealed and air  is allowed to enter from the
ground surface, rather than  flowing  laterally
                                           261

-------
 through the vadose zone to the well.

 VOC  is an  acronym  for  volatile organic
 contaminant.    Typically  these  are organic
 chemicals with a high vapor pressure  and a
 tendency   to   evaporate   rapidly   under
 environmental conditions.

 Volatilization  refers  to the  transfer  of  a
 chemical from liquid to the gas phase.

 Water table refers to the water surface in an
 unconfmed aquifer at which the fluid pressure
 in the voids is at atmospheric pressure.

 Wetting fluid refers to the immiscible fluid
 which spreads on (or coats) the solid surfaces
 of a porous medium preferentially relative to
 another  immiscible  fluid.    In NAPL-water
 systems, water is usually the wetting fluid.

Zone of remediation is the radial area over
which cleanup of soil is actually  expected to
occur, based on the  minimum  amount of air
flow per unit mass of contaminant required to
reach target cleanup  levels.
                    ftU.S. GOVERNMENT PRINTING OFFICE: 1 995-650-006/22052

                                        262

-------