&EPA
EPA/600/R-16/175 | August 2016 | www.epa.gov/research
United States
Environmental Protection
Agency
Petroleum Vapor Intrusion
Modeling Assessment
with PVIScreen
DiV^WJ (•(
Residual
LNAPL
Vapor Plume
Dissolved Plume
Smear Zone
*
Oxygen I
Transport ~4l
~
FLOW
Aerobic
Biodegradation
Zones
Unsaturated
Zone
Saturated
Zone
Office of Research and Development
National Risk Management Research Laboratory | Groundwater, Watershed, and Ecosystem Restoration Division
-------
EPA/600/R-16/175
August 2016
Petroleum Vapor Intrusion Modeling
Assessment with PVIScreen
James W. Weaver
United States Environmental Protection Agency
Office of Research and Development
National Risk Management Research Laboratory
Groundwater, Watershed, and Ecosystem Restoration Division
Ada, Oklahoma 74820
Robin V. Davis
Utah Department of Environmental Quality
Salt Lake City, Utah 84116
Office of Research and Development
National Risk Management Research Laboratory | Groundwater, Watershed, and Ecosystem Restoration Division
-------
Table of Contents
ABSTRACT iv
DISCLAIMER iv
1. BACKGROUND 1
1.1 Outline 1
1.2 Environmental Models and Vapor Intrusion 1
1.3 Empirical Basis for Bioattenuation 2
1.4 The Biovapor Model 3
1.4.1 Uncertainty Analysis 4
1.5 PVIScreen Use in Context of a Petroleum Vapor Intrusion Assessment 6
2. GETTING STARTED 7
3. RUNNING THE MODEL 7
3.1 PVIScreen User Interface 7
3.1.1 File Selection 7
3.1.2 PVIScreen File Names 9
3.1.3 Editing an Existing Data Set 9
3.1.4 Editing the Data Screens 9
3.1.5 Treatment of Ground Water Samples 12
3.1.6 Vadose Zone 13
3.1.7 Preparing for Run 15
3.1.8 Running PVIScreen 16
3.1.9 Displaying and Understanding Statistics Output 17
3.1.10 Automated Report 18
4. EXAMPLES 19
4.1 Soil Gas Source Example 19
4.1.1 Site Investigation 19
4.1.2 PVIScreen Model Parameters 21
4.1.3 PVIScreen Model Results 21
4.2 Ground Water Data Example 24
4.3 Ground Water Example Indicating Possibility of Vapor Intrusion 27
5. THEORETICAL BACKGROUND 33
5.1 Oil Phase Weathering 33
5.2 BioVapor Equations 34
REFERENCES 35
APPENDIX A Running PVIScreen in Command Line and Batch Mode A1
A.l Batch Mode A1
APPENDIX B Microsoft Excel Comma Separated Value (.csv) Format B1
APPENDIX C Unit Conversions in PVIScreen CI
APPENDIX D Post-Processed Output File D1
D.l File Identification D1
D.2 Reprinting Output Results D1
D.3 Simple Statistical Results D1
D.4 Histograms D1
-------
APPENDIX E Entry of Deterministic and Stochastic (Monte Carlo) Data El
E.l For Deterministic Models El
E.2 For Stochastic (Monte Carlo) Models El
E.2.1 Constant Parameter Values in Stochastic Models El
E.2.2 Variable Parameters in Stochastic Models E3
E.3 Uniformly Distributed Parameter Values E3
E.4 Triangular Distribution E4
E.5 Truncated Normally Distributed Parameter Values E5
Figures
Figure 1. Comparison between the processes governing non-biodegrading solvent vapor intrusion and petroleum
vapor intrusion 1
Figure 2. Thickness of clean soil associated with attenuation of benzene vapors 2
Figure 3. Thickness of clean soil associated with attenuation of TPH vapors 2
Figure 4. Schematic illustration of petroleum hydrocarbon (PHC) flux & distribution and oxygen flux & distribution 3
Figure 5. Building, vadose zone, ground water contamination layout used in PVIScreen for a soil gas source 3
Figure 6. Building, vadose zone, petroleum (NAPL) contamination, and aquifer layout used in PVIScreen for a
ground water source 3
Figure 7. Vapor attenuation factors predicted by three-dimensional numerical model as a function of separation
distance from a residential building foundation to an underlying petroleum contaminant vapor source
concentration of 10 mg/L (10,000,000 ng/m3) with various first-order biodegradation rates 4
Figure 8. Example cumulative probability curve 5
Figure 9. Petroleum vapor intrusion site assessment flow chart 6
Figure 10. File selection in PVIScreen 7
Figure 11. Opening a template file in PVIScreen 7
Figure 12. New folder created for a PVIScreen project 8
Figure 13. Selection of existing input or previous results 8
Figure 14. Identification & Options input screen 9
Figure 15. Illustration of input distribution choices 10
Figure 16. Building & Foundation input screen 10
Figure 17. Chemicals input screen 11
Figure 18. Reduction in water phase TCE concentration across the capillary fringe from a laboratory experiment 12
Figure 19. Schematic illustration of capillary rise and the tension saturated zone 12
Figure 20. Vadose zone input screen 13
Figure 21. Average shallow ground water temperature in the U.S 13
Figure 22. Suggested values: air flow and oxygen input screen 14
Figure 23. Suggested value for the ground water concentration factor 14
-------
Figure 24. Suggested values: model control input screen 15
Figure 25. Schematic representation of spatial relationships in a PVIScreen run with a petroleum (NAPL) source 15
Figure 26. Running PVIScreen after an input file has been selected 16
Figure 27. Result for simulation of benzene where petroleum vapor intrusion is unlikely 16
Figure 28. List form of PVIScreen output 17
Figure 29. Graphical form of PVIScreen output for a case with high potential for vapor intrusion 17
Figure 30. PVIScreen options after simulation completed 18
Figure 31. Example PVIScreen-generated report displayed in the user's browser window 18
Figure 32. The highlighted html file (extension htm) contains an automated report summarizing the output from a run
of samplegroundwaterinput-commercial.pvi made on 9/13/2017 18
Figure 33. Site plan for the convenience store and off-site restaurant simulations 19
Figure 34. Cross section summarizing site data in the vicinity of the off-site restaurant 20
Figure 35. Identifications and options input screen for the off-site restaurant simulation 21
Figure 36. Building and foundation input screen for the off-site restaurant simulation 22
Figure 37. Vadose zone input screen for the off-site restaurant simulation 22
Figure 38. Chemicals input screen for the off-site restaurant simulation 23
Figure 39. Screening levels input screen for the off-site restaurant simulation 23
Figure 40. Off-site restaurant simulation of benzene showing no simulation results above the screening level
of 0.5 ng/m3 24
Figure 41. Schematic for the MW-9 simulation with sample depth of 6.9 ft 25
Figure 42. Simulation results for MW-9 using ground water data 26
Figure 43. Site plan indicating benzene concentration and free product thicknesses for April 28, 2015 27
Figure 44. Cross section summarizing site data in the vicinity of the convenience store 28
Figure 45. Building and foundation input parameters for convenience store simulation 29
Figure 46. Vadose zone inputs for convenience store simulation 30
Figure 47. Chemical inputs for convenience store simulation 30
Figure 48. Schematic for simulation of convenience store 31
Figure 49. Benzene results indicating high probability for petroleum vapor intrusion 31
Figure 50. Gasoline range organics (TPH-GRO) results indicating high probability for petroleum vapor intrusion 32
Figure 51. Diesel range organics (TPH-DRO) results indicating a strong possibility of vapor intrusion 32
Figure Al. Executable jar file (PVIScreenBatchRun.jar) used to execute the batch version of PVIScreen A1
Figure Bl. Microsoft Excel output ("Save As") dialog box showing choice of other formats to write a comma separated
value (*.csv) file Bl
Figure B2. Microsoft Excel output dialog showing "csv" file type selected for PVIScreen input file B2
Figure El. Cumulative frequency for a constant or deterministic parameter E2
Figure E2. Alternate specification of constant parameter by specifying a uniform probability distribution E2
Figure E3. Uniform cumulative probability distribution with a range from 10 ft/d to 20 ft/d E3
Figure E4. Triangular distribution (line) and its approximation by 11 points (squares) E4
Figure E5. Truncated cumulative normal distribution with mean of 15 ft/d and standard deviation of 1.5 ft/d E5
-------
Tables
Table 1. Chemicals in the PVIScreen database 11
Table 2. Field results, PVIScreen input concentrations and UTAH DEQ screening levels for the off-site restaurant
simulation 20
Table 3. Ground water data for MW-6, MW-9, and MW-20 of Figure 34 25
Table 4. Benzene simulation results for the restaurant with ground water concentration data as the source
of contamination 26
Table 5. Ground water sampling data from MW-3 and MW-4 when no free product was present 28
Table 6. Sub-slab, indoor air, and screening concentrations for the Utah convenience store 29
Table CI. Excerpt of file showing default set of unit conversion factors in file OlPVIScreenUnitConversion.csv CI
Table El. Cumulative probability curve example E3
Table E2. Approximate triangular distribution with minimum of 10, maximum of 20 and most likely value of 13 E4
Table E3. Cumulative normal distribution frequencies, symbolic values and example E5
Abstract
Vapor intrusion of petroleum compounds differs from that of chlorinated solvents because of the dominant effect of aero-
bic biodegradation on the concentration and distribution of petroleum vapors. To better understand the behavior of pe-
troleum compounds, a model called PVIScreen was developed that applies the theory developed for the BioVapor model
(DeVaull, 2007) to a lens of petroleum hydrocarbons in the subsurface that is capable of acting as a source of petroleum
vapors. The PVIScreen model automatically conducts an uncertainty analysis using Monte Carlo simulations. The model
is intended to make uncertainty analysis practical for application at petroleum vapor intrusion sites. The model can be
run in either a batch mode, using Microsoft Excel files for both input and model outputs, and an interactive mode using
a graphical user interface. Each of these is described, along with required inputs, example problems and the theoretical
background of the model. Model simulations are in agreement with an EPA-sponsored analysis of field data that illustrate
and document the attenuation of concentrations of petroleum compounds in soil gas with distance above the source of
the vapors.
Disclaimer
The U.S. Environmental Protection Agency through its Office of Research and Development funded and managed the re-
search described here. It has been subjected to the Agency's peer and administrative review and has been approved for
publication as an EPA document. Mention of trade names or commercial products does not constitute endorsement or
recommendation for use.
iv
-------
1. Background
1.1 Outline
This report describes the basis of the PVIScreen model,
the procedure to run the model using the PVIScreen user
interface, detailed description of input files, statistical in-
puts, example problems and the theoretical background of
the model.
1.2 Environmental Models and Vapor
Intrusion
Environmental models are based on the application of
mass conservation principles to transport and transfor-
mation of chemicals in the environment. Generally, all
environmental models are based on a two-part concep-
tualization: an empirically-determined principle relating
chemical, physical and biological quantities, and empirical
coefficients. Taken together these two components have
the potential for representing transport and transforma-
tion of petroleum vapors in the vadose zone below a build-
ing.
Although models may represent important processes, the
ability to determine definitively that there are no vapor
impacts to buildings ("screen for PVI") also depends on ap-
plication-related factors. These factors include the degree
to which the site conceptual model matches the structure
of the mathematical model, the inherent limitations im-
posed by the assumptions in the mathematical model,
the values chosen for input parameters, and the ability to
calibrate the mathematical model to site conditions. Many
of these factors are difficult to address at leaking under-
ground storage tank sites, so as will be noted below, model
results should be viewed as one line of evidence in a site
assessment.
Over ten years ago, vapor intrusion and its evaluation
through modeling approaches were identified as a poten-
tial problem at subsurface contamination sites (Obamas-
cik, 2002). Application of simplified models using mostly
generic default parameters has contributed to confusion
over appropriate assessment strategies for these sites.
One of the primary models in use, the Johnson-Ettinger
model (JEM) was presented as a heuristic screening model
(Johnson and Ettinger, 1991). Essentially, the model con-
sists of two completely-mixed compartments, one repre-
senting the interior of a building and the other the soil
below. This conceptualization reflects the potential for
both features of the building and the subsurface to con-
tribute to indoor air contamination. In its original form,
the model simply related the concentration in the soil gas
to the concentration in indoor air. No biodegradation of
the compound was included as the model conceptualiza-
tion only related concentration between the two compart-
ments. Later extension of the JEM included diffusive flux
from a deeper source zone to the bottom of the founda-
tion. Even though the JEM does not include biodegrada-
tion, the JEM could be a valid conceptualization for chlori-
nated solvents, because most of these compounds do not
undergo aerobic biodegradation.
Petroleum hydrocarbons, however, are readily degraded
under aerobic conditions so the JEM excludes a process
with the potential for greatly affecting petroleum vapor in-
trusion (Figure 1). Chlorinated solvents are not degraded
in the presence of oxygen, so dissolved contamination in
the aquifer (saturated zone) almost always has the poten-
tial to contaminate indoor air (Figure 1, left). In contrast,
petroleum hydrocarbons can be degraded under aerobic
conditions, so the prospect for vapor intrusion is more
limited, but also more dependent on the specific configu-
ration of a source, presence of light non-aqueous phase
liquid (LNAPL), and depth to water, among other factors
(Figure 1, right).
Less Penetrable Zone
Residual
DNAPL
Dissolved Plume
Potential Vapor Plume
Unsaturated
Zone
Residual
LNAPL
Dissolved Plume
Oxygen Transport
Aerobic
Biodegradation
Zones
Unsaturated
Zone
Saturated
Zone
Figure 1. Comparison between the processes governing non-biodegrading solvent vapor intrusion (left)
and petroleum vapor intrusion (right) (U.S. EPA, 2012).
1
-------
1.3 Empirical Basis for Bioattenuation
In 2013, EPA published a report that evaluated a
large set of empirical petroleum vapor data from
the United States, Canada and Australia (USEPA
2013). This evaluation confirmed model simula-
tion results, and that aerobic biodegradation of
petroleum hydrocarbons is a weli-documented,
widespread and robust process that has been
demonstrated under a wide range of environmen-
tal conditions. Important factors influencing aero-
bic biodegradation of petroleum vapors include:
• Vapor source concentration, flux, and
composition;
• Minimal oxygen concentration is required to
support aerobic biodegradation;
• Oxygen supply and demand;
• Distance between the vapor source and a
building foundation;
• Soil type and properties (e.g., porosity and
moisture); and
• Size and characteristics of the building and
adjacent land surface.
The empirical data show that the capacity for bio-
degradation of petroleum hydrocarbons is high.
In cases where uncontaminated soil overlies dis-
solved benzene and total petroleum hydrocarbon
concentrations in ground water, vapors have been
shown to be attenuated (Figure 2 and Figure 3).
The thickness of clean soil associated with attenu-
ation depends on the concentration in ground wa-
ter as higher levels in ground water require greater
thickness of clean soil. These data demonstrate
the ability of the vadose zone to attenuated ben-
zene and TPH vapor concentrations, and formed
the basis for the EPA approach to screening sites
that was published in 2015 (U.S EPA, 2015), which
is discussed in section 1.5 "PVIScreen Use in the
Context of a Petroleum Vapor Intrusion Assess-
ment".
10
>
<1>
V
<
P
o
to
~ Benzene: Soil Vapor & Dissolved Paired Measurements
Near-Slab & Sub-Slab
All Soil Types
~ ~ ~
~ ~
~ ~ ~
*
~ *
~ ~ ~ ~
~~ t ~
~ ~~~!• ~ ~~~
:t #7j *3
~ ~ * %~~~ ~
~ *t ** 4 ~
~ ~ * #» /~
A A. * A
~
V ~
~
10 100 1,000 10.000
Benzene, dissolved, ug/L
100,000
Figure 2. Thickness of clean soil associated with attenuation
of benzene vapors (Davis, 2009).
¦ TPH: Soil Vapor A Dissolved Paired Measurements
Near-Slab & Sub-Slab
All Soil Types
100 1,000 10,000 100,000 1,000,000
TPH, dissolved, ug/L
Figure 3. Thickness of clean soil associated with attenuation
of TPH vapors (Davis, 2009).
2
-------
1.4 The BioVapor Model
The BioVapor code was developed (DeVaull, 2007;
API, 2010) to account for:
• aerobic biodegradation in the vadose zone,
• limits on oxygen supply imposed by the diffusive flux
into the vadose zone,
• the oxygen demand of any number of compounds
present in soil gas, and
• oxygen consumption by native soil respiration.
Conceptually, oxygen from the atmosphere (Figure 4) per-
meates the soil gas providing the electron acceptor need-
ed for aerobic biodegradation of petroleum hydrocarbons.
Because of the typical large flux of oxygen from the atmo-
sphere, petroleum hydrocarbons react in a zone near their
source and consequently their concentrations may be re-
duced relatively deeply in the vadose zone.
• linking directly to a fuel leaching model,
• providing the capability to use a flexible unit
conversion system,
• displaying key outputs in an intuitive fashion.
In PVIScreen, the building, vadose zone and aquifer are
defined in a layout which relates the bottom of the foun-
dation to a zone of petroleum contamination. Input pa-
rameters describe the size and characteristics of each com-
ponent in the model. Data on vadose zone contamination
are needed to drive the simulation. These are expected to
be soil gas (Figure 5) or ground water data (Figure 6).
Bottom of Building to Sample
Depth of Sample
3 Zone of Soil Gas Contamination
Aquifer
if minium
Figure 5. Building, vadose zone, ground water
contamination layout used in PVIScreen for a soil gas
source.
Figure 4. Schematic illustration of petroleum hydrocarbon
(PHC) flux and distribution and oxygen flux and distribution
(US EPA, 2012).
BioVapor was developed as a Microsoft Excel spreadsheet
application. BioVapor balances the supply of oxygen from
the atmosphere with the degradation-driven demand for
oxygen in the soil gas. The outputs of the model include
the depth of the aerobic zone, indoor air concentration for
all chemicals included in the simulation, and the chemical
concentrations at other points in the soil profile.
PVIScreen is an object-oriented petroleum vapor intrusion
model, which extends the concepts of BioVapor by
Bottom of Building to Sample
Depth of Sample
Ground Water Contamination
Aquifer
I I I I I I I II I I I I I
Figure 6. Building, vadose zone, petroleum (NAPL)
contamination, and aquifer layout used in PVIScreen
for a ground water source.
implementing an automated uncertainty analysis,
Land Surface
A
Oxygenated
Soil
Impacted
Soil
Oxygen
Flux
PHC + CH,
Flux
PHCs + CH«
Increasing Concentration
3
-------
1.4.1 Uncertainty Analysis
1.4.1.1 The General Need for Uncertainty Analysis
Uncertainty analysis, as used here, uses the response of
the model to changes in parameter values to assess the un-
certainty in model output. The method used in PVIScreen
is to presume that some or all parameters of the model
are uncertain. The probability of a parameter taking on a
value is governed by a cumulative probability curve. The
graphical user interface allows for two choices: a constant
parameter, and a parameter defined by a minimum and
maximum value1.
If the model is used in batch mode more options are avail-
able. These include entering sets of points defining the
input probabilities. No assumptions of particular distribu-
tions (i.e., normal) are needed, although a normal distribu-
tion can be approximated empirically (See Appendix Entry
of Deterministic and Stochastic (Monte Carlo) Data").
1.4.1.2 Uncertainty in Biodegradation Rates
The need for uncertainty analysis is illustrated by numeri-
cal model simulations by Abreu et al. (2009). The results
were presented as attenuation factors, defined as the in-
door air concentration (Cbuj|dr ) divided by the source con-
centration (C ):
x source'
a = AF = Cb™Wng/
/ csource
When this factor is low, the vapor concentration in the
building is reduced over that in the subsurface, and the
potential for vapor intrusion is low. Briefly summarizing
the results, the attenuation factors are much lower when
aerobic biodegradation is included as compared to the
non-biodegradation case (TCE and other solvents). The
literature shows that degradation rates, however, are not
known with certainty; neither are they measured at typical
field sites. Thus, the degradation rate is one that should be
(and is) treated as uncertain in PVIScreen (Figure 7).
1.E-02 3
1.E-11
1.E-03 i
No Biodegradation
1.E-04 =
o 1.E-05 ,
X - 0.079 h"1
1 .E-06 :
3 1.E-07 =
>.= 0.79 h
< 1.E-08 =
1 .E-09 i
1.E-10 :
). = 2 h
-> 1 ' r
01 23456789 10 11
Vapor Source Depth below Foundation (m)
Figure 7. Vapor attenuation factors predicted by three-dimensional numerical model as a function of separation distance
from a residential building foundation to an underlying petroleum contaminant vapor source concentration of 10 mg/L
(10,000,000 ug/m3), with various first-order biodegradation rates (Abreu and Johnson, 2005; Abreu et al., 2009).
^se of the minimum and maximum values correspond to a uniform distribution where there is a straight-line (linear) relationship
between the two specified values.
-------
1.4.1.3 Uncertainty Analysis in PVIScreen
In the Monte Carlo procedure, the model is run a speci-
fied number of times and the uncertain parameters are
chosen randomly from the probability curves. A sampling
technique called Latin Hypercube Sampling (LHS) is used
to assure that parameter values are drawn from all parts
of the probability distributions. LHS generally reduces the
number of simulations required. After completing all re-
quired runs of the model, the results are processed into
output probability curves and a summary table for each
chemical specified in soil gas.
The main PVIScreen outputs are cumulative probability (or
frequency) curves. As an illustrative example, 1000 runs of
the model produced the aerobic zone thicknesses ranging
from 10.88 cm to 120.86 cm (Figure 8). No result had an
aerobic zone thickness less than 10.88 cm and no simula-
tion had one that exceeded 120.86 cm. There was a 100%
probability that the result was between these two values.
The probability that the result was between 10.88 cm and
60 cm (vertical blue arrow on Figure 8) was 31% (hori-
zontal blue arrow on Figure 8). The probability that the
aerobic zone thickness was between 60 cm and 120.86 cm
was 69%, as indicated by the upward-sloping, red-colored,
cross hatching on Figure 8.
The main display of PVIScreen output is based on the cu-
mulative probability curve (Figure 8). The probability that
simulations exceed a risk-based concentration is displayed
as output. This result is directly analogous to the cross-
hatched area in Figure 8.
Aerobic Zone Depth (cm)
0.2
0.1
0
0 20 40 60 80 100 120 140
Aerobic Zone Depth (cm)
Figure 8. Example cumulative probability curve. The probability that the aerobic zone depth is 60 cm is
31%, while the probability that the aerobic zone thickness is greater than 60 cm is 69%.
5
-------
1.5 PVIScreen Use in Context of a Petroleum
Vapor Intrusion Assessment
The use of models in general, and PVIScreen specifically,
should be integrated with site assessment. EPA's PVI guid-
ance document outlined steps for an assessment. Because
PVIScreen needs measured concentrations to represent
the source of contamination, its use comes after sufficient
characterization of the site and construction of a concep-
tual site model (Figure 9). Professional judgment is needed
to determine when the characterization data is sufficient.
Two possibilities for PVIScreen use follow:
An option is to use the mode! in parallel with determina-
tion of a vertical separation distance (box outlined with
dashed line in Figure 9). For this scenario, the model pro-
vides an additional line of evidence to support site deci-
sion making. Because the model is essentially based on
the same set of observations as the empirical database,
these two lines of evidence should provide similar results.
The model may, however, allow for unique site-specific
factors to be incorporated, such as commercial rather than
residential air exchange rates.
A second option is to use the model where the site fails the
vertical separation zone criteria, or is a marginal case (the
circled "NO" in Figure 9). Here the use of the model results
may provide an additional line of evidence to support a
site decision (which could be to conduct further sampling
or which could be that no further investigation is needed).
For any confirmed or
suspected release,
START HERE:
Do PHC
vapors pose an
immediate threat
to safety of
building
occupants?
Conduct an adequate
site characterization
and construct a
Conceptual Site
Model (CSM)
(including all factors
that may affect the
vapor intrusion
pathway—see Special
Considerations,
Section 1)
(a) Alert first
responders &
assess potential
threat of fire
and/or explosion
(b) Mitigate threats as
appropriate
Delineate a
Lateral
Inclusion Zone
(including all
factors that
may affect the
vapor intrusion
pathway)
present, do
preferential
pathways connect
vapor source and
building?
Are any
existing or planned\NO
buildings within the
lateral inclusion
zone?
Evaluate vapor
source(s) and
mitigate PVI as
appropriate
Community Engagement
Federal regulations under 40 CFR 280.67
require implementing agencies to provide
notice to those members of the public who
are directly affected by a release from a UST
arid the planned corrective action if such a
release requires a corrective action plan.
Implementing agencies are advised to tailor
community engagement activities based on
site-specific circumstances. Such activities
may occur at any point(s) in the assessment
and mitigation process. It is recognized that
earlier and more frequent communication
yields positive results.
Evaluate vapor source* and
attenuation of PHC vapors by
either:
(1) Measuring PHCs in near-slab
and deep (near source) soil
gas, or
Collectingindoorair samples
paired withsub-slabsoilgas
samples
*lfcontaminationisin direct
YES
Determine Vertical
Separation Distances for
each building (including all
factors that may affect the
vapor intrusion pathway)
Is the
thickness of
dean, biologically activeN^ES
soil greater than the
minimum vertical
separation
distance?
Option
(2)
Option
(1)
Is any
potential threat
of PVI indicated by
near-slab & deep
(source) soil gas
sampling?
Isany
otentialthrea
of PVI indicated by
indoorair& sub-
slab soil gas
sampling?
PVI not
likely to be
a concern
Figure 9. Petroleum vapor intrusion site assessment flow chart. PVIScreen could be used anytime after the site has
been characterized and a conceptual site model constructed (outside dark gray box). One scenario is to use the model
in parallel with determination of a vertical separation distance (box outlined with a dashed line). A second is to use the
model when the vertical separation criteria fails or is marginal (circled "NO").
6
-------
2. Getting Started
The PVIScreen distribution file (PVIScreen.zip) needs only
to be copied to C:\MyDocuments\PVIScreen, followed by
extraction of all the files to this location. PVIScreen will
open by double clicking on the PVIScreen.jar file2. All
needed data files and examples will be available in the SRC
\systemData and projects\examples directories.
3. Running the Model
3.1 PVIScreen User Interface
PVIScreen is incorporated into a user interface (Ul) that fa-
cilitates preparing and editing input files, running the mod-
el and displaying the results3. The interface is designed so
that the model options are selected from left to right. As
more options become available, more of the buttons turn
green (Figure 10).
* select Fits
Previous Results
Welcome to PVIScreen Select an input file to begin
3.1.1 File Selection
PVIScreen must begin with an existing data file. There are
two options:
1) For a new project, template files are provided for the
common types of inputs. Template files are available
for cases with soil gas sources and ground water
sources, and residential and commercial buildings.
These are found in the directory
PVIScreen/projects/templates (Figure ll)4.
After editing, these must be saved under new names in
a project directory. A new directory is created by right-
clicking on the directory pane and selecting "New", renam-
ing to match project, and then saving input file in the new
directory (Figure 12).
2) When iater coming back to a project, the original input
data can be accessed or previously-created results files
can be accessed. The input file can be run again or the
previous results displayed (Figure 13).
II r . •" r? About Exit
Figure 10. File selection in PVIScreen.
Id EPA PVIScreen
• Existing Input Setect Fi|e
Previous Resurcs
View/Edit Input
ti c Prepare to Run |
Run PVIScreen
Results
Write Report
About
Existing Input file named: null
PVIScreen File Input
•* Jim * My Documents - PVIScreen - projects » templates
•* g) | Search ten-
Organize ~ New folder
Favorites
K Desktop
£ Downloads
Recent Places
Lor ares
• Documents
Jl Miiv
Name *¦
Date modified
| Type
I ~ i SampleGroundWaterlnput-Commerdal.pvi
9/13/2017 10:17 AM
PVI Fie
SampteGrouidWaterlnput-Residenbal.pvi
9/11/2017 3:14PM
PVI File
[3 SampleSoiiGasInpu t-Commercial, pvi
9/13/2017 10:17 AM
PVI File
|s'i SampleSoiiGasInpu t-Residential. pvi
9/6/20179:35 AM
PVI File
Figure 11. Opening a Template File in PVIScreen.
2if the model fails to execute, make sure that the Java runtime environment is in the computer's path. Check with your system
administrator for guidance.
3PVIScreen can be run in a command line mode, which is described in Appendix A. Running PVIScreen in Command Line and Batch
Mode.
"PVIScreen regenerates all the template files each time it is run. When template files are used, they must be renamed and/or
moved to save data for future use.
-------
CI EPAPVISotct
• Existing Input
Previous Results
Press Run PVIScreen to run C:\UsersUim\DocumentsVPVIScreen\projects\templates\SampleGroundWaterlnput-Commercial pvi
Select File View/Edit Input View Schematic Prepare to Run
Abo
Save Changes To Input Data?
^ J
- Jim • My Documents - PVIScreen * projects
- m\ Search projects
Organize ~ New folder
Favorites
K Desktop
Downloads
% Recent Places
Libraries
Jl| Documents
Jp Must
k= Pictures
Q Videos
Name
examples
New folder
templates
Date modified
Type
9/13/2017 10:11AM
9/6/20179:48 AM
9/13/2017 10:10 AM
File folder
File folder
File folder
Figure 12. New folder created for a PVIScreen project.
CI EPA PVIScreen
• Ex&ting Input
Previous Results
Welcome to PVIScreen Select arynput file to begin
Select File
Existing Input
Prevtous Results
Select File
EPA PVIScreen
Figure 13. Selection of existing input or previous results.
PVIScreen uses Microsoft Excel spreadsheet files in a
comrna-separated-value format for both input and output
(Appendix B. Microsoft Excel Comma Separated Value (.csv)
Format). These are saved to the user's computer disk.
8
-------
3.1.2 PVIScreen File Names
Several types of files are used in PVIScreen. An input file
is named:
SampleSoilGasinput.pvi
The "pvi" extension is used to identify the file as a PVI
Screen input file. The format of this file is the Microsoft Ex-
cel comma-separated-value. Thus the file can be opened
and read by Microsoft Excel.
3.1.2.1 PVIScreen Output Files
PVIScreen produces an output file for each run of the mod-
el with a file name related to the input file. For example,
the input file:
SampleSoilGasinput.pvi
has a results file called:
SampleSoilGaslnput2016-May-31-16h-52m-13.0s.PVI
Screen. Results.csv
where the input file name is given a date/time stamp and
".PVIScreen.Results" is added to the file name. The final
".cvs" extension identifies the file as a comma-separated-
value and allows its editing in Microsoft Excel and other
spreadsheet software.
3.1.3 Editing an Existing Data Set
The first input screen lists general information on the sim-
ulation (Figure 14). This and each other tab of the input
data screens can be edited in turn. Detailed examples fol-
low in the soil gas and soil sample source examples below.
3.1.4 Editing the Data Screens
3.1.4.1 Identification & Options
The Identification & Options screen (Figure 14) contains
general information on the simulation and choices for can-
cer risk level, hazard quotient and oil distribution5.
CI EPA PVIScreen
# Ex^ng Input Se|ect Fj|e
Previous Results
View/Edit Input View Schematic Prepare to Run
Existing Input file named: SampleGroundWaterlnput-Commercial pvi
Identification & Options Building & Foundation VadoseZone Chemicals Screening Levels Suggested Values
Site
LUSTLine Restaur
SiteLocation
MW-9 1-8-15
City
XX
State
UT
Analyst
JWW
AfFiliabon
U5 EPA
Risk Level
1.00E-06
•V
Hazard Quotient
1
¦V
Oil Distribution
none
Figure 14. Identification & Options input screen.
5These are in the process of being removed as they are typically overridden elsewhere, especially through the use of predetermined
screening levels.
9
-------
3.1.4.2 General Input Parameter Information
3.1.4.3 Building & Foundation
Each numerical parameter of the model can be specified
as constant or uniformly distributed (Figure 15). Constant
parameters are chosen to be values that can be deter-
mined with a high degree of confidence. Typically, these
include the building width, length and ceiling height. Oth-
er parameters are either unmeasured or unlikely to be es-
timated very precisely. For these, a uniform distribution
is available, where a minimum and a maximum value are
specified. Parameters likely to be in this category are crack
width, air exchange rate, and porosity. More options in-
cluding standard statistical and empirical distributions are
available in the command line mode of running the model.
See Appendix A Running PVIScreen in Command Line and
Batch Mode.
Constant
* one value
Uniform
min
max
Figure 15. Illustration of input distribution choices.
The Building & Foundation screen contains the specifica-
tion of building and foundation parameters (Figure 16).
Dirt Floor: yes or no.
Only buildings with a true dirt floor should answer "yes."6
Width: From measurement of building.
Length: From measurement of building.
Ceiling Height: The jEM and other models include a
mixing zone height. Examination of example values show
that these correspond to ceiling heights of 8 to 12
(US EPA, 2004). The mixing zone may not correspond
exactly to ceiling height, but an independent evaluation
of mixing zone height is not expected to be available.
Foundation Depth Below Grade: This is the depth of the
bottom of the foundation below grade.
Foundation Thickness: Measurement of building.
Crack Width: US EPA (2003) lists crack widths from
hairline (<0.1 mm) to 5 mm.
Air Exchange Rate: Values of 0.1 hr1 to 1.5 hr1 were
developed by US EPA (2003) for residential buildings.
|Q EPA PVIScreen
• Existing Input
Previous Results
Select File
View/Edit Input
View Schematic
Prepare to Run
Write Report ]
Existing Input file named: SampleGroundWaterlnput-Commercial pvi
| Vadose Zone
Chemicals
Screening Levels
Suggested Values
dirt floor
Constant
~ one value
Width
60.00
ft
Constant
» one value
Length
80.00
ft
Constant
~ one value
CeilingHeight
9.000
ft
Constant
» one value
Fou ndationDepth BebwGrade
6.000
in
Uniform
- min
FoundationThickness
6.000
in
max
Fou ndationThickness
6.000
cm "
Uniform
» min
CrackWidth
0.5000
mm »
max
CrackWidth
5.000
mm *
Uniform
•» min
AirExchangeRate
3.000
1/hr *¦
max
AirExchangeRate
10.00
1/hr ~
Irtsert air exchange rate ranges:
Full
High (Drafty) Moderate Low (Tight)
Figure 16. Building & Foundation input screen.
6The only current option is a concrete floor.
10
-------
3.1.4.4 Chemicals
The Chemicals screen contains the specification
of chemical parameters (Figure 17). Chemicals
that appear in the PVIScreen database (Table 1)
can be selected as a component of the contami-
nant source. In the example in Figure 17, BTEX
and petroleum hydrocarbon equivalent carbon
number fractions have been chosen. The source
type given (Fuel Phase Concentration By Volume)
is determined in the template for the input data
set7. Templates are available for soil gas and wa-
ter sample input files.
Table 1. Chemicals in the PVIScreen database.
Benzene
Toluene
Ethylbenzene
xylenes
TPH-GRO
TPH-DRO
C8 To C9 Aromatic
C9 To CIO Aromatic
CIO To Cll Aromatic
Naphthalene
C5 To C6 Aliphatic
C6 To C7 Aliphatic
C7 To CS Aliphatic
C8 To C9 Aliphatic
MTBE
EDB
|Q FPA PVIScreen
• Existing Input
Previous Results
Select File
View/Edit Input
View Schematic
Prepare to Run
Run PVIScreen Results Write Report About
Exit
Existing Input file named SampleGroundWaterlnput-Commercial pvi
Identification & Options Budding & Foundation Vadose Zone | Chemicals Screening Levels Suggested Values
Add or Remove Chemical » Add/Remove
Constant
-
one value
benzene
WatefPhaseConcentration
0.03300
me/1
•
Constant
•
one value
tokiene
WaterPhaseConcentration
0.001000
mg/l
*
Constant
-
one value
ethylbenzene
WaterPhaseConcentration
0.001000
mg/1
-
Constant
-
one value
xylenes
WaterPhaseConcentration
0.001000
-
Constant
-
one value
naphthalene
WaterPhaseConcentration
0.001000
mg/l
-
Constant
-
one value
MTBE
WaterPhaseConcentration
0.001000
mg/I
-
Constant
-
one value
TPH-GRO
WaterPhaseConcentration
0.1400
rag/1
-
Figure 17. Chemicals input screen.
'In an updated version of the model these will be editable from the input data screen.
11
-------
3.1.5 Treatment of Ground Water Samples
A soil sample is used as a direct input to PVIScreen,
because soil gas is the media through which contami-
nants are transported in the vadose zone. No adjust-
ments to the concentration are needed. For ground
water samples, a paradigm is needed to simulate the
transition between water and soil gas. Once in the soil
gas, the calculation proceeds as for a soil gas sample.
The following describes the approach used in PVIS-
creen.
Transport across the capillary fringe follows a complex
process. Data from McCarthy and Johnson (1993) are
used to develop a simplified water table paradigm for
PVIScreen. Trichloroethene (TCE) in aqueous solution
was used to generate a distribution of concentration
through the capillary in a laboratory column. The con-
centrations in pore water were measured for both a
rising and falling water table. Because TCE would not
be subject to aerobic biodegradation in this experi-
ment as in the case of chlorinated solvent vapor in-
trusion (Figure 1), it permits study of the mechanical
aspects of transport, without additional impacts from
biodegradation. For ground water flow in a moder-
ate flow range (i.e., 0.1 m/d), McCarthy and johnson
(1993) found that the concentration of TCE in the cap-
illary fringe was reduced by a factor of 10 at the top of
the tension saturated zone (Figure 18 and Figure 19).
The factor is included in the Suggested Value tab of
PVIScreen, with the value set to 0.1.
Experimental results from McCarthy and johnson
demonstrate the reduction in water phase concentra-
tion occurring through the capillary fringe. By com-
paring the moisture content and relative concentra-
tion, these experiments showed a roughly 1/10-fold
to 1/100-foid reduction in water phase concentration
(Figure 18). These relationships held during both im-
bibition and drainage8.
Although factors greater than 10 may be justified, the
data from McCarthy and Johnson (1993) represent
only one set of laboratory conditions. Ground water
data, also, might have representational problems. A
primary concern is a sample that is obtained from a
depth below the water table. If that sample is drawn
from a screen that crosses a contaminant plume so
water with differing concentrations are mixed togeth-
er, then the concentration is not representative of the
concentration at the water table.
Legend
• Drainage
Imbibition
Water Table
Top of Tension Saturated Zone Imbibition
Top of Tension Saturated Zone Drainage
o —
0.2
0.4
\ 100 x Reduction
\ :
i
£ _ . _
CL
0)
Q
0.6 -
0.8
10 x Reduction
0.001 0.01 0,1 1
Relative TCE Concentration
0 0.1 0.2 0.3 0.4
Moisture Content
Figure 18. Reduction in water phase TCE concentration across
the capillary fringe from a laboratory experiment (McCarthy
and Johnson, 1993, used by permission, John Wiley and Sons).
Height
Tension Saturated Zone
Water Table
Moisture Content
Figure 19. Schematic illustrating capillary rise and the
tension saturated zone, which is an area above the water
table where the water fills the pore space and is held by
capillary suction. Air is admitted to the pore space above
the tension saturated zone.
imbibition is defined as the replacement of a non-wetting fluid by a wetting fluid,, and drainage is defined as the replacement of a
wetting fluid by a non-wetting fluid.
12
-------
3.1.6 Vadose Zone
The Vadose Zone screen contains the specification
of the vadose zone parameters (Figure 20).
Depth To Sample: Depth from the surface to the sample
that is treated as the source of contamination.
Depth To Water: Depth from the surface to the water table.
Depth To Historic Water Table: Depth to the deepest water
table depth (that contributed to a smear zone.)
Depth To Bottom: Depth from the surface to the bottom of
the aquifer.
Moisture Content: Volumetric moisture content of soil
must be less than the porosity.
Porosity: Fraction of void volume in vadose zone.
Fraction Organic Carbon: Typically below the root
zone, the fraction organic carbon is 0.0001 to 0.001, but
depends on soil type.
Soil Temperature: Historic ground water temperatures
in the U.S. range from 3°C (37°F) to 25°C (77°F) and give
a rough guide for soil temperatures (Figure 21).
~I EPA PVTScreen
• Existing Input
Previous Results
Existing Input file named: SampleGroundWaterlnput-CommerciaLpvi
Select File View/Edit Input View Schematic Prepare to Run
Ret
Abou
| Chemicals
Screening Levels
Suggested Values
Constant
» one value
DepthToSample
6.900
ft
Constant
» one value
DepthToWater
6.900
ft
-
Constant
- one value
DepthToHistoricWaterTaWe
6,900
ft
-
Constant
» one value
DepthToSottom
20.00
ft
-
Uniform
» min
MoistureContent
0.04900
dimensio...
- j
max
MoistureContent
0.2127
dimensio...
- J
Uniform
» min
Porosity
0.2900
dimensio...
-
max
Porosity
0.4340
dimensio...
-
Uniform
¦» min
Fraction OrganbcCarbcn
7.5E-4
dimensio...
max
Fraction Organ icCartron
0.001250
dimensio...
'J
Uniform
» min
SoiiTemperature
11.25
C
max
SoilTemperature
18.75
C
~
Figure 20. Vadose Zone input screen
ft
Aversge Temperature
¦of Shallow®
a o will Wator
0
Temperature in
Degrees F
Figure 21. Average shallow
ground water temperature in
82 the U.S. (Collins, 1925).
13
-------
3.1.6.1 Suggested Values
Practitioners are unlikely to have ready access to some
specialized parameters of the model, so suggested val-
ues are provided (Figure 22 through Figure 24).
3.1.6.2 Suggested Values: Air Flow and Oxygen
On Figure 22:
Qso[l: US EPA (2003) gives a "typical" range for houses
on coarse-grained soils is on the order of 1 to 10 L/
min. In PVIScreen a second air flow parameter, the Air
Flow Below Building, is set equal to Qs . based on the
suggestion by API (2010).
Soil Respiration Rate: Estimated by DeVaull (2007) as
1.69 mg/g-d.
Diffusion In Air: The diffusion coefficient of oxygen in
air, as estimated by the methods used in Lyman et al,,
1982.
Diffusion In Water: The diffusion coefficient of oxygen in
water, as estimated by the methods used in Lyman et al.,
1982.
Surface Concentration: Value from atmospheric
concentration of oxygen.
Minimum Biodegradation Concentration: The minimum
concentration allowing biodegradation is taken as 1%.
3.1.6.3 Suggested Values: Concentration Adjustment
The Suggested Values: Concentration Adjustment screen
holds the value of the ground water concentration factor
(Figure 23). This value is saved with the input data file and
its selection is described in Section 3.1.5, "Treatment of
Ground Water Samples".
3.1.6.4 Suggested Values: Model Control
The Suggested Values: Model Control screen contains miscel-
laneous parameters which control the simulation and presen-
tation of the results (Figure 24). Under normal circumstanc-
es, these should not be modified.
3 EPA PVIScreen
3 ;r,put Select File View/Edit Input View Schematic Prepare to Run
Previews Results '•—
Existing Input file named: SampleGroundWaterlnput-Commercial pvi
Identification & Options Building & Foundation Vadose Zone Chemicals Screening Levels Suggested Values
Air Flow and Oxygen Concentration Adjustment Model Control
Uniform
mm
Qsoil
max
Qsoil
Constant
one value
SoilRespiratSonRate
Constant
-
one value
DifFusiorvInAir
Constant
w
one value
Diffusionln Water
Constant
~
one value
SurfaceConcentration
Constant
¦W
one value
MinimumBiodegradattonConcentrabon
1.000
L/m
10.00
L/m
-
1.690
mg/g-d
W
0.1750
onFactor 0.1000
Figure 23. Suggested Value for the ground water concentration factor. This factor is applied to input ground water
sample concentrations to generate the source term for PVIScreen. Section "Treatment of Ground Water Samples"
discusses the rationale for selecting a value.
14
-------
|E1 EPA PVIScreen
• Existing Input
Previous Results
Select File
View/Edit Input
View Schematic
Prepare to Run
Run PVIScreen
Resul
Existing Input file named: SampleGroundWaterlnput-Commercial.pvi
Identification &. Options
Building & Foundation
Vadose Zone
Chemicals
Screening Levels
Suggested Values
Air Flow and Oxygen
Concentration Adjustment
| Model Contro
l[
Bisection
Bisection
Bisection
max alpha
Number of Simulations
Number of frequency distribution intervals
Lower Limit
Max iterations
toJerance
200.0
1000.0
1.0E-4
5000.0
1.000e-12
50.00
Figure 24. Suggested Values: Model Control input screen.
3.1.7 Preparing for Run
After editing and saving the in-
put file, the schematic should
be viewed (Figure 25). The
schematic shows the spatial
relationships between the
foundation, sample and water
table. Once the spatial rela-
tionships and other data have
been checked the "Prepare to
Run" button completes prepa-
ration for running PVIScreen.
If the input data have been
changed, a dialog will ask to
save changes to the input.
^1B*I
• Select Ffe VWBtt Input | VW Schema* | Prepaio to Run
Existing Input file named nuO
0.562 (ft;
16.55 (ft)
16.56 (ft)
10.00 ft)
18.25 (ft)
iiwwiiiiwwwwiiWHUiiiiniiniwwww wnwwmi
Figure 25. Schematic representation of spatial relationships in a PVIScreen run with a
petroleum (NAPL) source.
15
-------
3.1.8 Running PVIScreen
After editing and checking the input file the "Run PVIS-
creen" button is used to execute the model (Figure 26).
The interface automatically displays the Monte Carlo re-
sults when completed. A simulation is shown where none
of the benzene simulation results exceeded the screening
level (Figure 27).
1 CI EPA PVIScreen [
• Existing Input Select Pile View/Edit Input
Previous Results
View Schematic
Prepare to Run
Run PVIScreen
About
Exit
Press Run PVIScreen to run C:\UsersUim\Documents\PVIScreen\projects\templates\SampleGroundWaterlnput-Commercial.pvi
Figure 26. Running PVIScreen after an input file has been selected.
~I FPA PVISotcti
Exsbng Input
Previous Results
Select File Vtew/lEdit Input View Schematic Prepare to Run Run PVIScreen Results Write Report About
Statistics results plotted for C:\UsersUim\Documents\PVlScreen\projects\templates\SampleGroundWaterlnput-Commercial-2 pvi
benzene toluene ethyttwiaene xylenes Naphthalene MTBc TPH-GRQ
PVIScreen Result for benzene indoor air concentration benzene risks/hazards
Exit
JH]x|
L2 .
u I
1.0
03
M-
0.7
1 o,s
I
0.5
0.4
OJ
0.2
0.1
r~i
-4-
0.0
-M
AS
0.0 % Exceed the ScreeningleveJ of 0.5 (0.5 )
"L" indicates screening level
Probability That Chosen Risk Level(s) Are Exceeded
¦ High model probability of exceedence
»Low or moderate model probability of exceedence
Probability
"M" indicates maximum probability result
¦ Most Probable Individual Result: 3.55E-5
(which is exceeded by 10.31 % of simulations)
Averaged-Parameter Result: 4.14E-23
(which is exceeded by 81.35 % of simulations)
Probability Density
-15-341-15 -10 -15
Log Indoor Air Concentrator! (LoglO )
Figure 27. Result for simulation of benzene where petroleum vapor intrusion is unlikely. None of the simulations exceed
the screening level of 0.5 as indicated by the "L" on the chart being located at a frequency of 1.0 and the notation in the
righthand column that 0.0% of the simulations exceed the screening criteria.
16
-------
3.1.9 Displaying and Understanding Statistics
Output
benzene risks/hazards
The model results are presented on separate tabs for each
chemical in the simulation. Each tab has a list output (Fig-
ure 28) and a chart output (Figure 29), which summarize
the same information.
3.1.9.1 List Output
The list form of PVIScreen output gives first the percent-
age of simulations above the chosen screening level (Figure
28). Here the chemical is benzene, so there are cancer and
non-cancer screening levels.
18.58% of the simulations exceeded the lxlO"6 screening
level of 0.29 |ig/m3. All of the Monte Carlo simulations
fell below the non-cancer hazard level of 30 ng/m3, and
no vapor intrusion is indicated by the non-cancer screen-
ing. Given the benzene result, the model is indicating a
strong possibility of a vapor intrusion problem. Note that
this example has a petroleum (NAPL) source fairly close to
the bottom of the building (Figure 25).
The list output gives the result of maximum probability
("M") and the averaged-parameter result ("V"). The most
probable result is close to the cancer screening level of
0.29 (ig/m3, and is another indicator that vapor intrusion
could be a problem for this case. The averaged-parameter
result is several orders of magnitude lower in concentra-
tion (0.00605 ng/m3) and illustrates the necessity for the
uncertainty analysis (as it misleadingly indicates no possi-
bility for vapor intrusion).
3.1.9.2 Graphical Output
The graphical form of output consists of a cumulative
probability curve and contains all of the results from the
Monte Carlo simulation,9 ranked from smallest to largest.
Concentration is plotted on the horizontal axis and the cor-
responding cumulative probability (or frequency) is plot-
ted on the vertical axis (Figure 29). Because the calculated
indoor air concentrations range over many orders of mag-
nitude in typical problems, the horizontal axis uses a log
base 10 scale for concentration. The curve is marked for
the cancer and non-cancer screening level concentrations;
averaged-parameter and most probable results.
"C" indicates specified cancer risk level
18.58% Above Risk 1.0E-6 Level (0.29 ug/m3)
0.0 % Above Hazard Quotient of 1.0 (30 ug/m3)
"H" indicates specified hazard quotient
Probability That Chosen Risk Level(s) Are Exceeded
High model probability of exceedence
Low or moderate model probability of exceedence
Probability
"M" indicates maximum probability result
Most Probable Individual Result: 0.28 ug/m3
(which is exceeded by 19.12 % of simulations)
V indicates averaged-parameter solution
Averaged-Parameter Result; 6.05E-3 ug/m3
(which is exceeded by 64,93 % of simulations)
Probability Density
Figure 28. List Form of PVIScreen output.
| benrane j totuerse «hyft*n2ene xylenes C8ToC9Aromas* CTToClQAromabc ClQToCllAromatJC CSToC6A4phac< C
PVIScreen Result for benzene indoor air concentration
L2
u
1.0
0.7
OA
0.3
0J
0.1
AS
IS
14
1-5
Log Indocr Air Cone«ntrabor» (LoglO ug/m3)
Figure 29. Graphical Form of PVIScreen Output for a
case with high potential for vapor intrusion.
9Results with concentrations below 10 5 |ig/m3 are omitted as they are much less than measureable values.
17
-------
3.1.10 Automated Report
PVIScreen automatically generates a report (Figure 30
arid Figure 31). The report is written in HTML and is au-
tomatically displayed in a browser window. The report
summarizes the modei assumptions, specific run informa-
tion, tabular results and all choices of input parameters.
Alternatively,, the report can be viewed from a standard
browser by double-clicking the html form of the simula-
tion output (Figure 32).
• Ewstinc input Select File View/Edit Input View Schematic
Previous Reaufes ^^-—' < —1
Prepare to Run Run PVIScreen Results Write Report About Exit
5£
Figure 30. PVIScreen options after simulation completed. Writing the automatically-generated report is now an available
option.
^¦¦¦
Print PVIScreen Report
PVIScreen Model Report
PVIScreen Background
PVIScreen Is a model for assessing impacts from petroleum vapors on
residences. PVIScreen was designed for automatic uncertainty analysis using
Monte Carlo simulation. The main result from the model is a probabiliy curve for
indoor air concentration for each simulated chemical. Both cancer and non-
cancer risk levels are indicated on the probability curves.
PVIScreen is based on the B'toVapor model (Devaull, 2007; API, 2010).
PVIScreen extends the capabilities of BioVapor by including automatic
uncertainty analysis, flexible unit selection, and direct inclusion of liquid gasoline
(NAPL). Major assumptions of the model include:
• Oxygen supply permits/limits biodegradation of petroleum vapors
¦ Mi ilHrvIo rnrnrvrmonfc nf final ronf-rJhi ifo t-r* nv\mon HomanH
< r
Figure 31. Example PVIScreen-generated
report displayed in the User's browser
window.
£ templates
fflOP
- Computer - OS (C:) » Users - Jim » My Documents - PVIScreen - projects - templates
fen r Search templati
Organize •" (£ Open ~
.-if Favorites
¦ Desktop
jf, Downloads
Recent Pfaces
Libraries
|3 Documents
Jfl Music
lS.1 Pictures
13 Videos
;*• Computer
os (co
•i' DVD RW Drive 03:) JimWeaver
r t My Passport 0=:)
r * My Passport (6*)
Network
Share with »
Name
Bum New folder
Type
Date modified
Q Samp(eGroundWaterInput-Commerdal-2 2017-Sep-13-10h-36m-37.0s.PVIScreen.Results.csv.htm Rrefox HTML Document
SampleGroundWaterInput-Commeroal-2 2017-Sep-13-I0h-36m-37.0s.PVIScreen.Results.csv
SampieOilLensInput.PVIScreen.csv
SampleSoilGasInput. PVIScreen, csv
9* SampJeGroundWaterInput-Commeraal-2 2017-Sep-13- 10h-36m-37.0s, PVIScreen.Results.csv.be.,.
Qp SampleGround WaterInput-Commercial -2 2017-Sep-13* lOh -36m-37.0s. PVIScreen. Results. csv. ht.,.
SampJeGroundVVaterlnput-Commer oal. pvi
\£ .. SampieGround Water Inpu t-Commer oal -2. pvl
SampJeGroundWaterlnput-ResidentiaLpvi
SampleSodGaslnput-Commwaal.pvi
J*. SampleSodGasInput-ftesdential.pvi
Microsoft Office Excel...
Microsoft Office Excel.,,
Microsoft Office Excel...
PNG image
PNG image
PVI File
PVI File
PVI File
PVI File
PVI File
14 KB 9/13/2017.
5,264 KB 9/13/2017.
5 KB 9/13/2017.
3 KB 9/13/2017,
34 KB 9/13/2017.
12 KB 9/13/2017.
3 KB 9/13/2017.
3 KB 9/13/2017.
3 KB 9/11/2017.
3 KB 9/13/2017.
3 KB 9/6/20179:
Figure 32. The highlighted HTML file (extension htm) contains an automated report summarizing the output from a
run of SampleGroundWaterlnput-Commercial.pvi made on 9/13/2017. Double clicking this file displays the report in a
default browser.
18
-------
4. Examples
4.1 Soil Gas Source Example
4.1.1 Site Investigation
A site with a 2,000 ft2 convenience store and adjacent
4,800 ft2 restaurant (the "off-site restaurant") reported a
release in August 2010 (Figure 33). The gas station was
active with two 10,000 gallon and one 8,000 gallon tanks.
The release was assumed to be from spills and overfills.
Four field investigations were made between 2010 and
2015 with 22 monitoring wells and 7 borings made. The
groundwater ranged from 5.5 ft to 7.5 ft deep (Figure 34).
A 3 ft deep boring was made at the edge of the restaurant.
Soil gas data from this boring was used as the source of
contamination for the simulation (Table 2). Because the
observed concentrations are low (1st row of Table 2), the
model result is expected to show that there is a low pos-
sibility of vapor intrusion at this location. The complete in-
put file can be found at projects/examples/LUSTLineRes-
taurantExample.pvi.
Site Map
Groundwater Contour and
Petroleum Contamination
Distribution
Leaking Underground
StorageTankSite
Murray, Utah
January 2015
0.032
2i
92.35
North
0.0613
MW-22
Off-Site Restaurant
/
35
0.0683
\tw li
M " III
¦kfl
52 54
92 53
.HI**
o»ir - -
<0.02
mwn
9
92 43
92 63
<0.02
<0.02
jj-.y-
LEGEND
Line of Cross-Section
A'
Groundwater Flow Direction
Q* GroundwaterMonitoringWell
2.86TPH-g, mg/'L
93.06 Groundwater Elevation, feet
O Temporary Geoprobe Boring
O Vapor M onitoring Point
-,o
¦ >—G roundwater Elevation
/ Contour, feet
O Subsurface Soil Contamination
ExceedingUtahScreeningLevel
13
n
TPH-gin Groundwater, mg/L
— • - Sanitary Sewer
-NaturalGas
Water Line
-¦ '• Storm Drain
— Electrical Line
—— • Communications
• Fiberoptic Line
Overhead Power
30 60
Scale, feet
Figure 33. Site plan for the convenience store and off-site restaurant simulations.
19
-------
A
NW
Property Line
C ff-S te Re sta u ra n t
MW-20
MW-9
MW-6
2: :
ua.fnj <32
T3~-
-------
4.1.2 PVIScreen Model Parameters
The input parameters for simulation are shown in Figures
35 to 39. The width and length of the building was entered
as measured (60 ft x 80 ft), the ceiling height was assumed
to be 9 ft, and thickness of foundation 10 cm. The crack
width was considered as a variable parameter, using a US
EPA range of values (0.5 mm to 5.0 mm). The air exchange
rate was set to a range of 3 hr1 to 10 hr1 to represent com-
mercial buildings (Figure 36).
Site-specific values were entered for the depth to sample
(3 ft), and depth to water (7.5 ft). The other vadose zone
parameters were given wide ranges as site-specific values
were not available (Figure 37).
The source of contamination was taken to be the soil gas
data from the 3 ft deep boring at the edge of the building
(SVP-1 on Figure 33). Concentration values reported at less
than the reporting limit were set to half the reporting limit
(Table 2 and Figure 38). Site-specific screening levels were
calculated from Guidelines for Utah's Corrective Action
Process for Leaking Underground Storage Tank Sites (Utah,
2010) and used as input (Figure 39).
4.1.3 PVIScreen Model Results
The PVIScreen simulation showed that all runs of the mod-
el were below the screening level of 0.5 (ig/m3 for ben-
zene (Figure 40). Although not shown, the same result
was found for each of the other chemical constituents.
Therefore, the model results suggest that there is a very
low chance of vapor intrusion at the restaurant.
| EI EPA PVIScreen |
* Select File
PrAvinuc ilN:
View/Edit Input View Schematic Prepare to Run
Existing Input file named: LUSTLineRestaurantExample.pvi
Identification & Options'
Building & Foundation Vadose Zone Chemicals Screening Levels Suggested Values
Site
LUSTLine Restaur
SiteLocation
MW-9 1-18-15
City
XX
State
UT
Analyst
JWW
Affiliation
US EPA
Risk Level
1.00E-06
HazardQuotient
1
Oil Distribution
none
Figure 35. Identifications and Options input screen for the off-site restaurant simulation.
21
-------
~I EPA PVIScreen
- p , Select File View/Edit Input View Schematic Prepare to Run
Previous Results 1 1 '•—: ¦—1 1 : 1 1 - : 1
Existing Input file named: LUSTLineRestaurantExample.pvi
Identification & Options Building & Foundation Vadose Zone Chemicals Screening Levels Suggested Values
dirt floor no
Constant
one value
Width
60.00
ft
~
Constant
one vaiue
Length
80.00
ft
~
Constant
-
one value
CeilingHeight
9,000
1 | ft
•w
Constant
-
one value
FoundationDepth BelowGrade
6.000
in
-
Uniform
•v
min
FoundabonThickness
6.000
In
~
max
FbundationThickness
6.000
an
*
Uniform
w
min
CrackWidth
O.SOOO
mm
~
max
CrackWidth
5.000
mm
~
Uniform
-
min
AirExchangeRate
3.000
1/hr
~w
max
AirExcharigeRate
10.00
| 1/hr
Insert air exchange rate ranges: Full High (Drafty) Moderate Low (Tight)
Figure 36. Buiiding and Foundation input screen for the off-site restaurant simulation.
~I EPA PVIScreen
• Existing Input
Previous Results
Existing Input file named: LUSTLineRestaurantExample.pvi
Select File View/Edit Input View Schematic Prepare to Run
j Chemicals
Screening Levete
Suggested Values
Constant
one value
DepthToSample
3.000
ft
Constant
~
one value
DepthToWater
7.500
ft
~
Constant
~
one value
DepthToHistoriCWaterTa ble
7.500
ft
Constant
~
one value
DepthToBottom
20.00
ft
w
Uniform
min
MoiStureContent
0.04900
dimensto...
¦W
max
MoistureContent
0.2127
dimensio...
~
Uniform
^r
min
Porosity
0.2900
dimensio...
max
Poroaty
0.4840
dimensio...
*
Uniform
w
min
Fraction Organic Carbon
7.5E-4
dimensio...
max
Fraction OrganicCarbon
0.001250
dimensio...
¦w
Uniform
ir
min
SoiTTemperature
11.25
C
max
SoiiTemperature
18.75
C
~
Figure 37. Vadose Zone input screen for the off-site restaurant simulation.
22
-------
Id EPA PVIScreen
• Basting input select File View/Edit Input View Schematic Prepare to Run
Prmsimic
About Exit
Existing Input file named: LUSTLineRestaurantExample pvi
Identification & Options Building & Foundation Vadose Zone Chemicals
Screening Levels Suggested Values
Add or Remove Chemical
Add/Remove
Constant
-
one value
benzene
AirPhaseConcentrabon
1.600
ug/m3
Constant
-
one value
toluene
AirPhaseConcentrabon
10.00
ug/m3
-
Constant
-
one value
ethylben2ene
AirPhaseConcentrat»on
2.200
ug/m3
-
Constant
¦»
one value
xylenes
AirPhaseConeentrabon
41.00
ug/m3
-J
Constant
V
one value
naphthalene
AtfPhaseConeentrabon
2.850
ug/m3
*
Constant
-
one value
MTBE
AirPhaseConcentrabon
1.800
ug/m3
1
Constant
-
one value
TPH-GRO
AirPhaseConcentrabon
210.0
ug/m3
Figure 38. Chemicals input screen for the off-site restaurant simulation.
CS EPA PVIScreen
• Existing Input Select File View/Edit Input View Schematic Prepare to Run
Previous Results
Existing Input file named: LUSTLineRestaurantExample.pvi
Identification & Options Building & Foundation Vadose Zone Chemicals I Screening Levels 1 Suggested Values
benzene
0.5000
mg/cm3
toluene
7310.0
mg/cm3
ethylbenzene
143:.0
mg/cm3
xylenes
148.0
mg/cm3
' |
naphthalene
4.390
mg/an3
w
MTBE
4380.0
mg/cm3
~
TPH-GRO
307.0
mg/cm3
~
Figure 39. Screening Levels input screen for the off-site restaurant simulation.
View Schematic
23
-------
JUL
Select File View/Edit Input View Schemabc Prepare to Run RunPVIScreen j Results J Write Report About Exit
plotted for C \U5ers\Jim\Documents\PVIScreen\prQjects\exampies\LUSTLjneRestaurantExample pvi
benzene tdkienc sthylbenzene xylenes Naphthalenr MTBE TFH-GRO
PVIScreen Result for benzene indoor air concentration
• Basting Input
Previous Results
Statistics results
-IDlxl
SS
L
H
f
iT
fl.5
0.4
i -3j0 -IS -2.0
Log Indoor Air Concentration (LoglO )
benzene risks/ hazards
QuO % Exceed the Screeningtevel of o.5 (0.5 )
"I" indicates screening level
Probability That Chosen Risk Levd(s) Are Exceeded
¦ High model probabtlJEy of exteedence
¦ Low or moderate model probability of exteedence
Probability
Most Probable Individual Result; L41E-6
[which e exceeded by 21.22 % of simulations}
Averaged-Parameter Result: 8.28E-13
(which is exceeded by 79.22 % of simulations)
Probability Density
Figure 40. Off-site restaurant simulation of benzene showing no simulation results above the screening level of
0.5 j-ig/m3. The most probable and averaged-parameter results are below the minimum plotting concentration of
10 s ng/m3.
4.2 Ground Water Data Example
The restaurant simulation can also be run using ground
water data. The chemical concentrations, their type (i.e.,
water phase concentration), and depths differ from the
previous simulations; all the other parameters were un-
changed. The location of the monitoring wells relative to
the building guides the selection of the data to drive the
model simulation. MW-9 is the closest monitoring well
that is upgradient to the building. This well is close to the
building, but between the contamination and the building
and is the best choice of these for use in the model. MW-
20 is downgradient and might be less representative of the
potential exposure in the building. MW-6 is located just
on the edge of the source, with benzene concentration
characteristic approaching that of NAPL10. For comparison
purposes the results of simulation with all three wells is
presented below. The schematic and result for MW-9 with
10 x ground water adjustment factor are shown in Figure
41 and Figure 42.
For comparison purposes data from each of the monitoring
wells was used in simulation (highlighted in Table 3). The
results from MW-9 (Table 4) show that none of the simula-
tions indicated the possibility of vapor intrusion above the
screening level. This was the result for both ground water
reduction factors. (Note that if the results show no impact
for a reduction factor of 10, the result when using the re-
duction factor of 100 must also show no impact as the con-
dition is less stringent). Similar results were obtained in
this case for MW-20.
For MW-6, which lies on the edge of the NAPL zone (Fig-
ure 34), at the reduction factor of 10, a very small num-
ber of simulations (0.12%) exceed the screening level for
benzene. At the reduction factor of 100, none exceed the
screening level. Accepting the conceptualization of the
model, in this case the occurrence of an impact depends
on the nature of transport across the capillary fringe.
Since the true nature of that transport is unknown from
the available data, a policy decision is needed for dealing
with borderline cases as this. The case is borderline be-
cause only a very few simulations exceed the screening
level (0.12%), and because the ground water concentra-
tion reduction factor of 100 results in 0.0% exceedance.
10ln the EPA PVI benzene concentrations >5 mg/L are presumed to indicate NAPL. Peargin & Kolhatkar presume the presence of
NAPL for benzene concentration of 1 mg/L or higher.
24
-------
Table 3. Ground water data for MW-6, MW-9, and MW-20 of Figure 34.
Well/
Elev.
Date
DTW
(feet)
MTBE
(mg/L)
Benzene
(mg/L)
Toluene
(mg/L)
Ethylbenzene
(mg/L)
Xylenes
(mg/L)
Naphthalene
(mg/L)
TPH-GRO
(mg/L)
MW-6
99.57
2/18/11
5.35
<0.00200
2.73
0.181
0.183
0.433
0.138
4.58
7/11/11
5.29
<0.00200
1.58
0.0635
0.136
0.231
0.129
3.10
12/4/12
6.15
<0.00200
1.93
0.00455
0.0141
0.0276
0.130
2.67
12/19/13
6.61
<0.00200
0.590
<0.00200
0.00803
0.0129
0.0786
1.02
6/25/14
6.08
<0.00200
1.63
0.0254
0.0979
0.171
0.0974
2.76
1/8/15
6.51
< 0.00200
2.04
0.00404
0.0179
0.0296
0.180
2.86
MW-9
99.66
12/20/13
7.18
<0.00200
0.813
<0.00200
<0.00200
0.0298
0.0136
1.29
6/25/14
6.47
<0.00200
1.84
0.00260
0.00268
0.00515
0.0160
2.40
1/8/15
6.90
< 0.00200
0.0330
< 0.00200
< 0.00200
< 0.00200
< 0.00200
0.140
MW-20
98.90
1/9/15
6.43
< 0.00200
0.336
< 0.00200
< 0.00200
0.0126
0.0349
0.897
6.900 (ft)
6.400 (ft)
6.900 (ft)
13.10 (ft)
Figure 41. Schematic for the MW-9 simulation with sample depth of 6.9 ft.
25
-------
-iai *J
• Existing Input
Select File
View/Edit Input
Vtew Schematic
Prepare to Run Run PVIScreen
Results Write Report About Exit
Statistics results plotted for CAUsefs\Jim\Documents^PVlScreen\projects\examples\HJSTLineRestaurantExampleWaterSample-MW9-1-8-15 pvi
benzene toluene
elhylbenaene
xylenes Naphthalene
MTBE TPH-CRO
PVIScreen Result for benzene indoor air concentration
benzene risks/hazards
UH
1.0
L
1
0.0 % Exceed the Screenir>gLevel of 0.5 (0.5 )
V indicates screening level
Probability That Chosen Risk tevel(s) Are Exceeded
i
•a-
OJ
0.7
&
— High model probability of exceedence
low or moderate model probability of exceedence
Probability
§ Qj6
?
0.5
0.4
0J ;
0J -
-------
4.3 Ground Water Example Indicating
Possibility of Vapor Intrusion
The Utah Department of Environmental Quality (UDEQ)
investigated potential petroleum vapor intrusion at a con-
venience store, where leaking underground storage tanks
(USTs) released gasoline into soil and groundwater (Figure
43). The potential for PVI was determined highly likely due
to presence of free product less than 15 feet below ground
surface (bgs) under the site convenience store, and ex-
tremely high subslab vapor concentrations of total petro-
leum hydrocarbons (TPH) and benzene. PVI was ultimately
confirmed through collection of indoor air samples from
the convenience store in April 2015, which indicated the
presence of TPH, benzene, and other gasoline component
vapors at concentrations that exceeded risk-based screen-
ing levels (see Table 6, Figure 44, and Davis (2015) for site
details and a summary of the site investigation).
Analyte concentrations from MW-4 taken on 10/2/2013
were used to drive the simulation. Since these concentra-
tions were high the possibility exists for an impact to in-
door air. The input parameter values are found on Figure
45 to Figure 47, and the site schematic is shown on Figure
48. The model results are suggestive of petroleum va-
por intrusion with 69% of the benzene (Figure 49), 69% of
the TPH-GRO (Figure 50), and 21% of the TPH-DRO (Figure
51) simulations indicating the potential for petroleum va-
por intrusion. These results are consistent with the state's
findings (Table 6).
OMW-5
0.42
OUTSIDE SUHMfc
CANISTER
, AtfS LATEftAl
INSIDE SUMMA CAN I*
ISLAND &
/ MW-? \
!br> 0.00141
0.00799
1,35 INAPL thickness, feet
Benzene & LNAPL Thickness
4-28-15 Monitoring Event
Farm Land
Figure 43. Site plan indicating benzene concentration and free product thicknesses for April 28, 2015.
Groundwater Flow Direction
0 40
Horizontal Scale, feet
17.9 Benzene,mg/L
Q 2-inchdiameter GW monitoring well
£ 4-inch diameter GW monitoring well
^ Vapor Monitoring Point
OScil contamination
>U DEQ Screening Levels
<0.001
27
-------
Convenfence Store
imjoor Air
Outdoor Air
MW-3
Sut-S!3C
Bja Room
Sue-sac
Main Store
MW-16
0 —
MW-12
MW-S
lay. lean
sandy
Silt sandy
Silt elas
moist
o 10
Sand
well-
graded
Clay, fat
Horizontal Scale, feet
Figure 44. Cross section summarizing site data in the vicinity of the convenience store.
Table 5. Ground water sampling data from MW-3 and MW-4 when no free product was present.
Date
Benzene
(mg/L)
Toluene
(mg/L)
Ethylbenzene
(mg/L)
Xylenes
(mg/L)
Naphthalene
(mg/L)
MTBE
(mg/L)
TPH-GRO
(mg/L)
TPH-DRQ
(mg/L)
Depth to
Water
(ft)
MW-3
10/2/13
39.4
49.0
3.26
17.2
0.688
< 0.2
118
0.396
9.78
5/3/13
13.4
12.6
0.929
5.76
0.239
< 0.2
34.2
<0.2
10.09
MW-4
4/28/15
17.9
0.774
2,51
10.5
0.246
<0.02
33.1
< 2.0
11.16
10/2/13
26.1
53.1
3.76
17.4
0.426
<0.04
128
<0.4
9.88
5/3/13
28.6
42.9
1,92
11.1
< 1.0
< 1.0
84.4
< 10.0
10.2
28
-------
Table 6. Sub-slab, indoor air, and screening concentrations for the Utah convenience store.
Benzene
(Hg/m3)
Toluene
(Hg/m3)
Ethylbenzene
(Hg/m3)
Xylenes
(l-ig/m3)
Naphthalene
(Hg/m3)
MTBE
(|ig/m3)
TPH-GRO
(l-ig/m3)
TPH-DRO
(Hg/m3)
Results
8-hour
indoor air
55
13
2
12
<0.53
<0.73
2,200
24-hour
indoor air
210
14
4
25
< 2.7
< 3.6
6,400
Sub-slab
main store
333,000
< 20,000
< 10,000
< 10,000
< 2,000
< 10,000
8,700,000
Sub-slab
back room
690
< 4,000
< 2,000
< 2,000
<400
< 2,000
420,000
Screening Levels
Commercial
Indoor Air
0.5
7,154
1,482
148
4
4,395
307
307
Commercial
Sub-slab
16.4
243,667
49,333
4,933
146
146,000
10,233
10,233
Id EPA PVlScreen
• Existing input Select File View/Edit Input View Schematic Prepare to Run -
Previous Results
reen Results
Write Report
Ab
Existing Input file named: GroundWaterExampleMW-3.pvi
| Identification & Options | Building & Foundation | Vadose Zone Chemicals Screening Levels Suggested Values
dirt floor
no "
Constant
» one value
Width
50,00
feet
Constant
•» one value
Length
50.00
feet
'.J
Constant
» one value
CeillngHeight
9.000
'J
Constant
» one value
FoundattonDepthBelowGrade
0.
in
-
Uniform
¦» min
FoundationThidmess
6.000
in
max
FoundationThidcness
8.000
in
Uniform
¦» min
CrackWtdth
0.5000
mm
max
CrackWidth
5.000
mm
'J
Uniform
» min
AirExchangeRate
0.5000
1/hr
max
AirExchangeRate
1.500
1/hr
-
Insert air exchange rate ranges: Full High (Drafty) Moderate Low (Tight)
Figure 45. Building and foundation input parameters for convenience store simulation.
29
-------
Id EPA PVIScreen
• Existing Input
Previous Results
Select File
View/Edit Input
View Schematic
Prepare to Run
Existing Input file named: GroundWaterExampleMW-3 pvi
Identification & Options
Building & Foundation
| Vadose Zone |
Chemicals
Screening Levels
Suggested Values
Constant
one value
DepthToSample
9.780
ft
~
Constant
- |
one value
DepthTo Water
9.780
ft
-
Constant
~
one value
DepthT oHistoricWaterTable
6.048
m
~
Constant
~
one value
DepthToBottom
20.00
1 ft
~
Uniform
-J
min
MoistureContent
0.04900
dimensio...
max
MoistureContent
0.2127
dimensjo...
Uniform
~
min
Porosity
0.2900
dimensio...
~
max
Porosity
0.4840
dimensio...
~
Uniform
~
min
FractionOrganic Cartoon
7.5E-4
dimensio...
•w
max
FractionOrganicCartoon
0.001250
dimensio...
~
Uniform
min
SoilTemperature
11.25
C
max
SojlTemperature
' 18,75
; c
w
Figure 46. Vadose zone inputs for convenience store simulation.
CI tPA PVIScrecti
Ewlmg Irput Selecl File
Previous Results
View/Edrl Input View Schematic Prepare to Run
Existing Input fife named GroundWaterExampleMW-3 pvi
Identification & Opftkxis Budding & Foundation Vadose Zone Chemicals ! Screening Levels Suggested Values
Add or Remove Chemical
Add/ Remove
About Exit
JSJxj
m
Constant
-
one value
benzene
WaterPhaseConcentratwn
39.40
Constant
-
one value
toluene
WaterPKaseC oncer,tration
-W.00
mg/l
Constant
*
one value
ethyS benzene
WaterPhaseCcmcentration
3,260
mg/1
Constant
-
one value
xylenes
WaterPhaseConcsntration
17.20
mg/J
Constant
-
one value
naphthalene
WaterPhaseConcenlr&tfon
0.6880
mg/l
Constant
-
one value
MTBE
WatefPhaseConcenirabon
0.1000
mg/l
Constant
*
one value
TPH-GRO
WaterPhaseConcentration
118.0
mg/l
Constant
-
one value
TPH-ORO
WaterPhaseConcentration
0.9396
mg/l
-
Figure 47. Chemical inputs for convenience store simulation.
30
-------
9.780 (ft)
9.780 (ft)
9.780 (ft)
10.22 (ft)
Figure 48. Schematic for simulation of convenience store.
Q fPA PVlSoeen
-*S -44 '33 "3.0 -15 'U0 -14 -14) -0^ 0.0 5.5 l.Q 1.5 24 IS 3 JO
Log Indoor Air Concentration (LoglO )
,lal*i
D
• Existing Input
Previous Results
Statistics results
benzene toluene
1.1
Select File View/Edit Input View Schematic Prepare to Run Run PVIScreen Results Write Report About Exit
plotted for C \Users\Jim\Doc uments\PVI Sc reen\projectstexamples\GroundWaterExampteMW-3. pvi
ethylbenzene xylenes Naphthalene mtbe TPH-GRO TPH-DRO
PVIScreen Result for benzene indoor air concentration benzene risks/ hazards
69.32% Exceed the ScreeningLevel of 0.5 (0.5 )
V indicates screening level
Probability That Chosen Risk Level(s) Are exceeded
™ High model probability of exeeedence
Low or moderate model probability of exeeedence
Probability
"M" indicates maximum probability result
¦ Most Probable Individual Result: 79.51
(which ts exceeded by 15.52 % of simulations)
"V" indicates averaged-parameter solution
^"Averaged-Parameter Result: 31.88
(which ts exceeded by 39.54 % of simulations)
Probability Density
Figure 49. Benzene results indicating high probability for petroleum vapor intrusion. As indicated on the graph (left) and
the table (right), 69% of the simulations exceeded the site-specific screening level of 0.5 |ig/m3.
31
-------
d tPA PVIScreen
<-5-4-3-2 4 0 i 2 3 4 S fr
Log Indoor Air Concentration (Log 10 )
JDl *1
• Existing Input
Previous Results
Select File View/Edit Input View Schematic Prepare to Run RunPVIScreen Results Write Report About Exit
Statistics results plotted for C \Users\Jim\Documents\PVlScreen^pro]ects\examples\GroundWaterExampleMW-3 pvi
benzene totaene ethyfcouene xylenes Naphthalene KT8C ITPH-GRO | TPH-DRO
PVIScreen Result for TPH-GRO indoor air concentration
TPH-GRO risks/ hazards
68.93% Exceed the ScreenwvgLevel of 307.0 (307)
V indicates screensng level
Probability That Chosen Risk level(s) Are Exceeded
¦ High model probability of exceedence
¦ Low or moderate model probability of exceedence
Probability
*M" indicates maximum probability result
-Most Probable Individual Result: 6.57E4
(which is exceeded by 10.71 % of simulations}
V indicates averaged-para meter soJubon
¦ Averaged-Parameter Result: 2.6E4
(which is exceeded by 39.51 % of simulations}
Probability Density
Figure 50. Gasoline range organics (TPH-GRO) results indicating high probability for petroleum vapor intrusion. As
indicated on the graph (left) and the table (right), almost 69% of the simulations exceeded the site-specific screening
level of 307 ng/m3.
~2223
.101 "I
Select File View/Edit Input View Schematic Prepare to Run Run PVIScreen Results Write Report About
• Ejasftng Input
Previous Results
Statistics results plotted for C \UsersUim\Documents\PVIScreen\project5texamples\GroundWaterEJcarnpleMW-3 pvi
Exit
benwne to*u«w etMbenw* xyients Naphttotanc MTBE TPH-GftO [ TPH-ORO j
PVIScreen Result for TPH-DRO indoor air concentration
& o.t
§
•5i -M -4.5
-3 JO -2J5 -2,0 J-S -1.0 -0.5 0.0 0.5
Log Indoor Air Concentration (LoglO )
1.0 1.5 2.0 2.5 2.0 3.5
tph-dro risks/hazards
21.27% Exceed the ScreeningLevel of 307.0 (307)
V indicates screening level
Probability That Chosen Risk Level(s) Are Exceeded
¦ High model probability of exceedence
¦ Low or moderate model probability of exceedence
Probability
"M" Indicates maximum probability result
¦ Most Probable Individual Result: 345.72
(which is exceeded by 18.52 % of simulations)
"V" indicates averaged -parameter solution
¦Averaged-Parameter Result: 92.57
(which is exceeded by 45.22 % of simulations}
Probability Density
Figure 51. Diesel range organics (TPH-DRO) results indicating a strong possibility of vapor intrusion. As indicated on the
graph (left) and the table (right), 21% of the simulations exceeded the site-specific screening level of 307 |ig/m3.
32
-------
5. Theoretical Background
Over ten years ago vapor intrusion and its evaluation
through modeling approaches were identified as a poten-
tial problem at subsurface contamination sites (Obamas-
cik, 2002). Application of simplified models using mostly
generic default parameters has contributed to confusion
over appropriate assessment strategies for these sites.
One of the primary models in use, the Johnson-Ettinger
model was presented as a heuristic screening model (John-
son and Ettinger, 1991). Essentially the model consists of
two completely-mixed compartments, one representing
the interior of a building and the other the soil below. This
conceptualization reflects the potential for both features
of the building and the subsurface to contribute to in-
door air contamination. Although models may represent
important processes, the ability to determine definitively
that there are no vapor impacts to buildings ("screen for
PVI") also depends on application-related factors. These
factors include the degree to which the site conceptual
model matches the structure of the screening model, the
inherent limitations imposed by the model assumptions,
the values chosen for input parameters, and the ability to
calibrate the model to site conditions.
Analysis of the Johnson-Ettinger model (JEM) using a one-
at-a-time sensitivity analysis showed a moderate level of
uncertainty resulting from parameter uncertainty (John-
son 2005). With a bounding value analysis Tillman and
Weaver (2006) showed that synergistic effects dominate
uncertainty because parameters of the model do not act
independently but interact. For JEM, the interactions are
clearasthe model can be formulated in dimensionlessform
using three dimensionless parameter groups. Tillman and
Weaver (2006) showed that the uncertainties were one to
two orders of magnitude higher when parameter interac-
tions were considered, and constructed a generic ordering
of parameter importance (Tillman and Weaver, 2007).
The Johnson and Ettinger model does not account for bio-
degradation and so inherently over-predicts indoor air
concentrations for situations where biodegradation oc-
curs. A primary and geographically-extensive example oc-
curs at petroleum hydrocarbon release sites, where many
studies have shown extensive aerobic biodegradation. In
response to these problems, the BioVapor model (Devaull,
2007) was developed to include the effects of oxygen-lim-
ited biodegradation, native soil respiration, and multiple
hydrocarbon species on vapor intrusion, all within a simpli-
fied modeling context. The model is based on the assump-
tion of steady-state diffusive transport in a homogeneous
vadose zone. Consequently, an analytical solution was ob-
tained.
5.1 Oil Phase Weathering
If the oil is uniformly leached due to flowing water and vol-
atilization to the vadose zone, then the mass conservation
equation for each constituent / becomes
dMt ff .
~ J J ^ Ciw dy dz - ADeff]si
where / is the absolute value of the hydraulic gradient,
KJSJ is the effective conductivity at water saturation,
S . D ,, is the effective diffusion coefficient for the vadose
w eff
zone, and Jsj is the flux of chemical leaving the source
through the gaseous phase. If the source is a rectangular
block of width, Y, then the mass balance that accounts for
the variation of flow with water saturation is
dM: r
^ — ~ y J ' K,v (Sw) Cjw dz - A Deff]Si
Solution of this equation provides a boundary condition
based on emplacement of a fuel phase and consistent
changes in composition due to weathering. When coupled
with transport in the soil gas as calculated by BioVapor, the
changes are assumed to occur slowly enough so that the
transients in the vadose zone are negligible. In the sim-
plest application of the model, leaching is assumed to oc-
cur up to a specified time, and the composition of the oil
phase is then used as input to PVIScreen.
33
-------
5.2 BioVapor Equations
Using the Johnson-Ettinger equations as the basis for representing
transport from the subsurface into a building, the BioVapor equa-
tions extend the modeling approach to include an analytical solu-
tion of steady-state diffusion-driven transport in the vadose zone
(DeVaull, 2007 and API, 2010). Multiple constituents of gasoline
diffuse from a soil gas or ground water source. Oxygen is supplied
at the foundation bottom either limited by atmospheric concentra-
tion or the foundation flux. It is then transported downward by
diffusion and it is available to react with the upward-diffusing petro-
leum hydrocarbons. The sum of the oxygen demands determines
the extent of oxygen penetration into the vadose zone, and the con-
centration of intruding chemicals for the oxygen concentration lim-
ited solution, according to:
where cffi - ct0 is the difference in oxygen concentration between
the foundation and the transition point between the aerobic and
anaerobic zones, (pj is the stoichiometric utilization factor for com-
plete mineralization, and Deffi0, are the effective diffusion coef-
ficients for the chemical, and oxygen, cfi - ct, is the difference in
chemical concentration between the foundation and the transition
point, pb, is the bulk density, yltase 0the baseline soil respiration rate,
La, the aerobic zone depth, and/t, the chemical flux at the transition
point. The oxygen flux equation is
where the change in oxygen flux between the founda-
tion and the transition between aerobic and anaero-
bic is denoted Jfi0 - ]w, and is determined from the utiliza-
tion factor, (pi the change flux of each chemical constituent,
i(]fi - Jt,i) and the baseline soil respiration. The latter is calculat-
ed from the bulk density, pb, the aerobic zone depth, and the soil
baseline oxygen utilization rate, Ahasefi- The model is solved for the
aerobic zone depth by bisection iteration, using either an oxygen
concentration or an oxygen-flux limitation. See API 2010 for the
complete model details.
34
-------
References
American Petroleum Institute, 2010, User's Manual, BioVapor: A 1-D Vapor Intrusion Model with Oxygen-Limited Aerobic
Biodegradation, American Petroleum Institute.
Brakensiek, D. L., Engleman, R.L., and Rawls, W.J., 1981, Variation within texture classes of soil water parameters,
Transactions of the American Society of Agricultural Engineers, 335-339.
Collins, W.D., 1925, Temperature of Water Available for Industrial Use in the United States, United States Geological
Survey, Water Supply Paper 520-F.
Davis, R., 2015, Petroleum Vapor Intrusion Assessment: Multiple Lines of Evidence Lead to Mitigation at Utah Gasoline
Fueling Station, U.S. Environmental Protection Agency, https://cluin.org/products/newsltrs/tnandt/view_new.
cfm?issue=0815.cfm#3 2/6 .
DeVaull, G.E., 2007, Indoor Vapor Intrusion with Oxygen-Limited Biodegradation for a Subsurface Gasoline Source,
Environmental Science and Technology, 41, 3241-3248.
Gustafson, J.B., Tell, J.G., Orem, D., 1997, Selection of Representative TPH Fractions Based on Fate and Transport
Considerations, Volume 3 of Total Petroleum Hydrocarbons Criteria Working Group Series, Amherst Scientific Publishers,
102 pp.
Hers, I., and Truesdale, R.T., 2013, Evaluation Of Empirical Data To Support Soil Vapor Intrusion Screening Criteria For
Petroleum Hydrocarbon Compounds, United States Environmental Protection Agency, EPA 510-R-13-001.
Johnson P.C., Ettinger, R.A., 1991, Heuristic Model for Prediction the Intrusion Rate of Contaminant Vapors Into Buildings,
Environmental Science and Technology, 25, 1445-1452.
Johnson, P.C., 2005, Identification of application-specific critical inputs for the 1991 Johnson and Ettinger Vapor intrusion
algorithm, Ground Water Monitoring and Remediation, 25(10, 63-78.
Lyman, W.J., Reehl, W.F., Rosenblat, D. H., 1982, Handbook of Chemical Property Estimation Methods, American
Chemical Society.
McCarthy, K.A. and Johnson, R.L., 1993, Transport of volatile organic compounds across the capillary fringe, Water
Resources Research, 29(6) 1675-1683.
Obamascik, M., 2002, EPA home-toxins test "crude and limited"; Widely used computer model often wrong, Denver Post,
January 7, 2001, A-l.
Potter, T.L., Simmons, K.E., 1998, Volume 2 of Total Petroleum Hydrocarbons Criteria Working Group Series, Amherst
Scientific Publishers, 102 pp.
Tillman, Fred D and James W. Weaver, 2007, Parameter Sets for Upper and Lower Bounds on Soil-to-lndoor-Air
Contaminant Attenuation Predicted by the Johnson and Ettinger Vapor Intrusion Model, Atmospheric Environment, DOI
10.1016/j.atmosenv.2007.05.033, 41(27), pp 5797-5806.
Tillman, Fred D and James W. Weaver, 2006, Uncertainty from Synergistic Effects of Multiple Parameters in the Johnson
and Ettinger (1991) Vapor Intrusion Model, Atmospheric Environment, 40(22) 4098-4112.
US EPA, 2003, User's Guide for Evaluating Subsurface Vapor Intrusion into Buildings. EPA Contract Number: 68-W-02-033,
https://www.epa.gov/risk/users-guide-evaluating-subsurface-vapor-intrusion-buildings.
US EPA, 2007, Control of Hazardous Air Pollutants from Mobile Sources: Final Rule to Reduce Mobile Source Air Toxics,
United States Environmental Protection Agency, Office of Transportation and Air Quality, Washington, DC, EPA420/F-
07/017.
35
-------
US EPA, 2008, Fuel Trends Report: Gasoline 1995-2005, United States Environmental Protection Agency, Office of
Transportation and Air Quality, Washington, DC, EPA 420/R-08/002.
U.S. EPA, 2012, Petroleum Hydrocarbons And Chlorinated Hydrocarbons Differ In Their Potential For Vapor Intrusion,
U.S. Environmental Protection Agency, Washington, DC., March.
U.S. EPA, 2015, OSWER Technical Guide for Assessing and Mitigating the Vapor Intrusion Pathway from Subsurface Vapor
Sources to Indoor Air, U.S. Environmental Protection Agency, Office of Solid Waste and Emergency Response, OSWER
Publication 9200.2-154, https://www.epa.gov/sites/production/files/2015-09/documents/oswer-vapor-intrusion-
technical-guide-final.pdf
Utah Department of Environmental Quality, 2010, Guidelines for Utah's Corrective Action Process for Leaking
Underground Storage Tanks Sites, Third Edition Final Draft.
van Genuchten, M.T., 1980, A Closed-form equation for predicting the hydraulic conductivity of unsaturated soils, Soil
Science Society of America Journal, 44, 892-898.
Weaver, J.W., L. R. Exum, and L.M. Prieto, 2010, Gasoline Composition Regulations Affecting LUST Sites, United States
Environmental Protection Agency, Washington, DC, 20460, EPA 600/R-10/001.
Weaver, J.W., L. Jordan, and D.B. Hall, 2005, Predicted Ground Water, Soil and Soil Gas Impacts from US Gasolines, 2004
First Analysis of the Autumnal Data, U.S. Environmental Protection Agency, Washington, DC, EPA 600/R-05/032.
Wilson, J.T., Cho, J.S., Wilson, B.H., Vardy, J.A., 2000, Natural Attenuation of MTBE in the Surface under Methanogenic
Conditions, United States Environmental Protection Agency, Washington, DC, EPA/600/R-00/006.
36
-------
Appendix A
Running PVIScreen in Command Line and Batch Mode
A.1 Batch Mode
In command line mode, PVIScreen can be used to run multiple sets of input files. The names for these files are listed in a
control file named:
OOlnputSetsToRun.csv
Each complete PVIScreen input file is listed on a separate line of this file. Double clicking on the PVIScreen executable jar
file - PVIScreenBatchLibrary.jar (Figure Al) causes each input file to be run in turn.
| a || -E
ll-Qil
~ Computer ~ OS (O) ~ Users ~ Jim ~ workspace » PVIScreen ~
~
Search PVIScreen
P
Organize
- i^j Open * Share with ~ Burn New folder
Jp ~ 3
#
l. r .. Name
ravorrtes
K Dgrlrtop ^ RFC 25Ft NYC2003 Leached flOOO PVIScreen 2013-Sep-1016h-15m-48.C
£ Downloads RFG 25 Ft NYC2003 Leached flOOO PVIScreen 2013-Sep-16 7h-35m-24.0-
Gjj) RFG 25Ft NYC2003 Leached flOOO PVIScreen 2013-Sep-L6 7h-35m-24.0:
^ RFG 25Ft NYC2003 Leached flOOO PVIScreen 2013-Sep-16 22h-7m-44.0:
"13 RFG 25Ft NYC2003 Leached flOOO PVIScreen 2013-Sep-16 22h-7m-44.0!
RFG 25Ft NYC2003 Leached flOOO PVIScreen Output2013-Sep-16 22h-5i
¦i Computer ^ RFG 25,14 NYC2003 Leached flOOO PVIScreen Output2013-Sep-16 22h-5:
£, OS (C:) LJ 'cla"Path
LAC1E (L) .project
OOInputSetsToRun .CSV
% Network ^ 001nPutSetsToRun-complete.csv
'5^ OlChemicalProperties.csv
>AiJ OlChemicalProperties-Fixed.csv
OlPVIScreenUnitConversion.csv
02ChemicalProperties,csv
03ChemicalProperties.csv
PVIScreenBatchLibrary.jar
kc! PVIScreen Schematic.jpg
Recent Places
Libraries
4
\
PVIScreenBatchLibrary.jar Date modified: 9/22/2013 2:57 PM Date created: 9/22/2013 2:56 PM
Executable Jar File Size: 7.81 MB
Figure Al. Executable jar file (PVIScreenBatchRun.jar) used to execute the batch version of PVIScreen.
A1
-------
Appendix B
Microsoft Excel Comma Separated Value (.csv) Format
File types are selected from the Microsoft Excel "Save As" menu (Figure Bl). Then the Comma Separated Value (.CSV)
format is chosen from the selections (Figure B2).
~
H
Save As
Prepare
Save a copy of the document
Excel Workbook
| Save the workbook in the default file
format.
Excel Macro-Enabled Workbook
! | Save the workbook in the XMl-based and
macro-enabled file format.
Excel Binary Workbook
Save the workbook in a binary file format
optimized for fast loading and saving.
Excel 97-2003 Workbook
Save a copy of the workbook that is fully
compatible with Excel 97-2003,
Find addins for other file formats
w Pyblish
| Close
Other Formats
Open the Save As dialog box to elect from
all possible file types.
Excel Options X Exit Excel
15
16
17
Figure Bl. Microsoft Excel output ("Save As") dialog box showing choice of Other Formats to write a comma
separated value (*.csv) file.
B1
-------
~
Save As
UU lj
PVIScreen - projects - templates
* vol Search templates
Organize ~ New fbWer
C Desktop
£ Downloads
. Recent Places
Ljbranes
Doojr»ents
Music
W; P»ctires
y Videos
:*• Computer
& os (co
•: DVD RW Drive (D:) Audio CD
Name
k~*JSampleFuetensProblem 2016-Jul-5-8h-47m-..,
'jy SampleFuetLensProWem 2016-3un-l-8h-9m-3..
feyj)SampJeFuelLensProWem 2016-Jun-l-8h-10m-.,
'Sj] SampJeFueiertsProblem 2016-Jun-l-8h-15m-..
SampleFudLertsProblgfn 2016Oun-l-8h-28m-.,
SampteFuelensProWem 20l6-Xm-l-Sh-28m-.,
SampjeFuelLensProbtem 2016-Jun-l-Sh-31m-.,
*3$)SampleFuetensProblem 20l6-Xin-2-llh-l6...
•SjJSampleFueiensProWem 2016-Jun-2-llh-57...
djLl
J
File name:
3IW"
-31-16h-52m-13.0s.PVIScreen.Resulte.csv
Save as t" |CSV (Comma delimited) (".csv)
Authors.
Tags: Add a tag
X]
Dated
7/5 J
7/5/.
6/1/
6/1/
6/1/
6/1/
6/1/
6/1/
6/21
6/2#
'if1
3
Hide Folders |
Toob
Save
Cancel
A
Figure B2. Microsoft Excel output dialog showing "CSV" file type selected for PVIScreen input file.
B2
-------
Appendix C
Unit Conversions in PVIScreen
A special subset of the Control group is the unit conversion factors. The allowable unit conversions are summarized in
the file OlPVIScreenUnitConversion.csv. Most PVIScreen users will never have to view or change this file. Editing the
unit conversion file is needed only to add a new set of units.
The unit conversion system is based on the idea that there is an internal set of units that are used in the model. For the
PVIScreen the unit set is:
quantity
unit
time
second
length
cm
mass
kilogram
concentration
mg/L
Any of the optional units (Table CI) can be used for an input data set, the specified unit conversions to the model's
internal unit set are made when the input data are read. Unit choices can be made for outputs.
Although not necessary for running the model, the following describes the operation of the unit conversion system:
When a parameter is read by the model, the name of the unit is included in the input group. The unit appears in input
immediately after the numeric value. The unit name and type is used to identify the appropriate unit conversion factors.
As many as five variants on the name are allowed. For example, the area in square meters can be designated by sm,
squaremeter, square meters, squaremeters, or m2. The case is unimportant so SM, Square Meter, etc. are also accepted.
When the unit is identified as belonging to a type (say area), the unit conversion to the model unit is performed. The file
indicates which units are the model units mostly for convenience as all input values undergo unit conversions. A model
unit always has the unit conversion factor of 1.0. The unit conversion factors for an optional unit give conversion to the
model unit. For example, to convert square feet to square meters a unit conversion of 0.3048 m per ft is applied twice.
In the unit conversion file, the unit conversion factors are supplied as up to four values. This is done for clarity. To get
square meters from square feet multiply the value in square feet by 0.3048 twice.
If the specified unit does not appear in the list of units from OlPVIScreenUnitConversion.csv, additional unit conversions
can be added to the OlPVIScreenUnitConversion.csv file as needed.
Table CI. Excerpt of file showing default set of unit conversion factors in file OlPVIScreenUnitConversion.csv.
Control
heading
type
status
ucfO
ucfl
ucf2
ucf3
nameO
namel
name2
Control
unit
length
model
1
1
1
1
cm
centimeter
centimeters
Control
unit
length
optional
100
1
1
1
m
meter
meters
Control
unit
length
optional
12
2.54
1
1
ft
foot
feet
Control
unit
length
optional
2.54
1
1
1
in
inch
inches
C1
-------
Appendix D
Post-Processed Output File
After completing stochastic (Monte Carlo) simulation runs, the main result file is read back into the model and the results
post-processed. The purpose of the post-processing is to generate statistical characterization of the results, including
simple statistics and histograms.
D.1 File Identification
The post-processed output file begins by listing the input and output files upon which it is based. Example output files
are provided with the executable PVIScreen file.
D.2 Reprinting Output Results
All of the results from the main output file are read and reprinted (not shown here).
D.3 Simple Statistical Results
A set of simple statistical calculations is performed on the output. These include:
• Minimum
• Average
• Maximum
• Range
• Variance
• Standard Deviation
• 5th, 95th Percentiles
• First, third Quartiles
• Median
• Inter-quartile range
• Median + Vz inter-quartile range
As for the main output file results, two tables show major parts of the statistical results.
D.4 Histograms
Histograms are used to determine the distribution of the results. In the histogram, the results, say for peak
concentration, are placed in binned intervals of the total output range. From these the most likely - and other - bins can
be determined. For each bin, the histogram gives the:
• Interval mid-point
• Count (number of results within the bin)
• Frequency (fraction of results in this bin relative to the total)
• Cumulative frequency
The histogram ends with a sum of points, which should equal the number of Monte Carlo runs, and the sum of the
frequencies, which should equal 1.0.
D1
-------
Appendix E
Entry of Deterministic and Stochastic (Monte Carlo) Data
PVIScreen has the capability of running either a deterministic or a stochastic (Monte Carlo) model. Many of the input
parameters may be specified as being variable ("stochastic"), but to do so the input probability distribution must be
specified. To simplify the input of data, the input for constant ("deterministic") and variable parameters follow the same
pattern.
For other than constant and uniformly distributed parameters, the inputs must be specified using the command
line version of PVIScreen, because the Graphical User Interface only allows for constant and uniformly distributed
parameters.
E.1 For Deterministic Models
Each input parameter has only one value. A parameter is entered with a key word to indicate the parameter group (here
"building"), followed by the specific parameter ("width"). Next a key word indicates that the parameter has only one
value. The key word is "constant". The value and unit are entered followed by the second key word which indicates the
cumulative frequency is 1.0. The frequency value is not used in deterministic models, but is included for compatibility
with stochastic models. An example of this input is:
Building, Width, Constant, value, unit symbol, 1.0, comment
E.2 For Stochastic (Monte Carlo) Models
Each stochastic model parameter is described by a cumulative probability distribution. These are entered by a series
of values. These are essentially empirical distributions. If the use of a parametric distribution is needed, these can be
entered as described below. The following describes how constant and varying values are entered in stochastic models.
E.2.1 Constant Parameter Values in Stochastic Models
Although every parameter in PVIScreen can be treated as variable, typically stochastic models will have some fixed
parameters. In that case, the parameter is designated as "constant." The entry of these values is given in the previous
section.
A constant or deterministic parameter has a single value. The probability of this value is 1.0, so a single entry is made
for the value with the frequency set at 1.0 (Figure El). To use a constant value for hydraulic conductivity of 15 ft/d, the
line of input for this parameter would be:
Groundwater, Hydraulic Conductivity, constant, 15, ft/d, 1.0
Alternatively, the deterministic parameter can be entered as if it is stochastic. Why would you want to do this? If you are
testing the effect of parameter variability, you might want to begin with a parameter following a uniform distribution but
later assign a constant value. To avoid inserting and deleting lines from the input file, two input lines are specified:
Groundwater, Hydraulic Conductivity, stochastic, 15, ft/d, 0.0
Groundwater, Hydraulic Conductivity, stochastic, 15, ft/d, 1.0
These lines are interpreted as specifying an input probability distribution with a uniform distribution with no values less
than 15 (first line) and no values higher than 15 (second line). Effectively, the parameter is set at 15 ft/d (Figure E2) for
all simulations. The impact on efficiency of running the model is negligible.
E1
-------
1
><
u
c
OJ Deterministic Value
3
cr
cu
QJ
>
.2
D
E
D 0.8
U
10 15 20 25
Hydraulic Conductivity (ft/d)
Figure El. Cumulative frequency for a constant or deterministic parameter. Here the value is
15 ft/d which is used for every simulation.
1 ~i
>•
U
c
Qj 0.8
3
O"
QJ 0.6
¦Alternate Deterministic Value Specification
OJ
>
¦
+-<
.2
~
E
d
u
0.4
0.2
10
15
20
25
Hydraulic Conductivity (ft/d)
Figure E2. Alternate specification of constant parameter by specifying a uniform probability
distribution with no values less than 15 ft/d and no values higher than 15 ft/d. Effectively the
value is set at a constant 15 ft/d.
E2
-------
E.2.2 Variable Parameters in Stochastic Models
For parameters considered to be stochastic, cumulative
probability distributions are entered as input. These are
values paired with a cumulative frequency. The cumulative
probability distribution begins with the probability that no
value is less than and ends with the value that all values are
less than.
In the example (Table El), no value of the parameter is less
than 10.0 hence its cumulative probability is 0.0. Similarly, no
value is greater than 20.0 and its cumulative probability is 1.0.
The intermediate values represent cumulative probabilities
between the extremes.
Table El. Cumulative probability curve example.
Parameter Values
Cumulative Probability
10.0
0.0
10.5
0.00135
12.0
0.0228
13.5
0.1587
15.0
0.5
16.5
0.8413
18.0
0.9772
19.5
0.99865
20.0
1.0
E.3 Uniformly Distributed Parameter Values
If a parameter distribution is uniform (i.e., all values are equally probable between a minimum and maximum value),
two entry values are entered for a uniform distribution. The minimum value is assigned a frequency of "0.0" and the
maximum value is assigned a value of "1.0."
To specify a uniform distribution with, say, a minimum value of 10 ft/d and maximum value of 20 ft/d (Figure E3) two
input lines are used:
Groundwater, Hydraulic Conductivity, stochastic, 10, ft/d, 0.0
Groundwater, Hydraulic Conductivity, stochastic, 20, ft/d, 1.0
The first line gives the minimum parameter value (10 ft/d) and cumulative probability of 0.0. This value indicates that
there are no hydraulic conductivity values less than 10 ft/d. The second line gives the maximum parameter value of 20
ft/d and cumulative probability of 1.0, indicating that there are no hydraulic conductivity values greater than 20 ft/d.
1
> 0.8
Uniform
Distribution
0.6
0.4
0.2
0
3
E
0
5
10
15
20
25
Hydraulic Conductivity (ft/d)
Figure E3. Uniform cumulative probability distribution with a range from 10 ft/d to 20 ft/d.
Options for entering triangular distributions where the minimum, maximum and most likely value are specified, and
truncated normal distributions are given in the appendix.
E3
-------
E.4 Triangular Distribution
A triangular distribution is specified from a minimum, maximum, and most likely value of a parameter. The triangular
cumulative probability distribution is determined from:
/(*) =
0; x < a
(x - a)2
1 -
(b — a)(c — a)'
(ib — x)2
(ib — a)(b — c)
1; x > b
a < x < c
; c < x < b
where f(x) is the cumulative probability, a is the minimum, b is the maximum and c is the most likely value. For
a parameter with minimum of 10, maximum of 20 and most likely value of 13, the cumulative probability curve is
approximated by 11 points (Figure E4 and Table E2). The figure illustrates the cumulative probability curve determined
from 41 points (using an increment of 0.25 ft/d) and its approximation by 11 points (squares). A spreadsheet illustrating
these calculations is available from the author (weaver.jim@epa.gov).
— Triangular Distribution
¦ Approximate Triangular Distribution
Hydraulic Conductivity (ft/d)
Table E2. Approximate triangular distribution
with minimum of 10, maximum of 20 and
most iikely value of 13.
Hydraulic
Conductivity
(ft/d)
Cumulative
Probability
10
0
11
0.033333
12
0.133333
13
0.3
14
0.485714
15
0.642857
16
0.771429
17
0.871429
18
0.942857
19
0.985714
20
1
Figure E4. Triangular distribution (line) and its approximation by
11 points (squares).
E4
-------
E.5 Truncated Normally Distributed Parameter Values
Parameters with a truncated normal distribution can also be entered by specifying the cumulative distribution (Figure
E5). A specific set of nine cumulative frequencies are needed to specify the distribution (Table E3). The distribution is
truncated because the normal distribution ranges from negative to positive infinity. For practical calculation purposes,
range is truncated at a minimum and maximum value. For this example the input lines are
Groundwater,
Hydra u
ic
Conductiv
ty,
stochastic,
10.0,
ft/d,
0.0
Groundwater,
Hydra u
ic
Conductiv
ty,
stochastic,
10.5,
ft/d,
0.00135
Groundwater,
Hydra u
ic
Conductiv
ty,
stochastic,
12.0,
ft/d,
0.0228
Groundwater,
Hydra u
ic
Conductiv
ty,
stochastic,
13.5,
ft/d,
0.1587
Groundwater,
Hydra u
ic
Conductiv
ty,
stochastic,
15.0,
ft/d,
0.5
Groundwater,
Hydra u
ic
Conductiv
ty,
stochastic,
16.5,
ft/d,
0.8413
Groundwater,
Hydra u
ic
Conductiv
ty,
stochastic,
18.0,
ft/d,
0.9772
Groundwater,
Hydra u
ic
Conductiv
ty,
stochastic,
19.5,
ft/d,
0.99865
Groundwater,
Hydra u
ic
Conductiv
ty,
stochastic,
20.0,
ft/d,
1.0
The truncated normal distribution is contained between 10.0 ft/d and 20.0 ft/d. Note that for the distribution to
approximate the normal distribution, the mean value less three standard deviations (10.5) must be above the minimum
(10.0), and the mean value plus three standard deviations (19.5) must be below the maximum value (20.0).
Table E3. Cumulative normal distribution frequencies, symbolic values and example.
Normal distribution
cumulative frequency
Symbolic values mean (n) and
standard deviation (c)
Example with mean of 15,
standard deviation of 1.5
0.0
minimum
10.0
0.00135
|o, - 3c
10.5
0.0228
|o, - 2c
12.0
0.1587
|0, - c
13.5
0.5
V
15.0
0.8413
|J, + c
16.5
0.9772
|o, + 2c
18.0
0.99865
|o, + 3c
19.5
1.0
maximum
20.0
l
0.8
Normal Distribution
CT
0.6
LL.
0.2
~~
10
0
U
0
5
15
20
25
Hydraulic Conductivity (ft/d)
Figure E5. Truncated cumulative
normal distribution with mean
of 15 ft/d and standard deviation
of 1.5 ft/d. The distribution is
defined by seven points (Table E3).
E5
-------
Depth of Sample
) Zone of Soli Gas Contamination
Depth of Sample
Bottom of Building to Sample
Bottom of Building to Sample
Aquifer
1 Ground Water Contaminatior
Aquifer
II111111111111
11111111111111
4>EPA
United States
Environmental Protection
Agency
Office of Research and Development (8101R)
1200 Pennsylvania Ave. NW
Washington, DC 20460
epa.gov/research
EPA/600/R-16/175
August 2016
------- |