Model Uncertainty Analysis, Field Data
Collection and Analysis of Contaminated
Vapor Intrusion into Buddings

Fred Tillman1'2, Jim Weaver2, David Spidle2 and Sandra Bird2

''National Research Council
2ORD/NERL/ERD, Athens GA

Introduction

The intrusion of vapors into indoor air from soil and groundwater contaminated by organic compounds is of concern both for the short-term hazard of explosion as well as
for longer-term risks associated with lower-level exposures. Volatile organic contaminants (VOCs) are frequently associated with contamination from leaking fuel storage
tanks and releases from dry cleaners and industrial facilities. These contaminants may volatilize from soil or ground water and enter homes or businesses through cracks in
basement floors or slabs. The Ecosystems Research Division in Athens, Georgia has undertaken an extensive vapor intrusion research program that seeks to address several
of the current challenges associated with evaluating the subsurface-to-indoor air pathway. Three of these challenges involve providing vapor-intrusion model users with a
full uncertainty analysis given a range of viable model inputs, evaluating the movement of moisture beneath buildings and its effects on vapor intrusion, and using dogs for
screening of homes prior to indoor air sampling in order to identify possible indoor sources of contamination. The goal of this program is to provide regulators and other
interested parties with practical, cost-effective solutions to challenges in evaluating vapor intrusion.

• Automated Uncertainty Analysis of Vapor Intrusion Models

Numerical models are often used to make decisions as to whether or not a site or particular
building at a site has the potential to be impacted by vapor intrusion from the subsurface. A
popular model, called the "Johnson and Ettinger Model", is widely used for this application.
EPA's Office of Solid Waste and Emergency Response (OSWER) distributes several
spreadsheet-versions of the Johnson and Ettinger (J & E) Model for use at RCRA and
Superfund sites and uses the model to establish screening-level concentrations in its Draft
Guidance for Evaluating the Vapor Intrusion to Indoor Air Pathway. The documentation for
the OSWER-distributed model suggests that users run several simulations for differing
parameter sets to evaluate uncertainty. However, it is impracticable to run all combinations of
inputs by "hand." ORD has developed an automated software system for analysis of model
results. When complete this software will be available on the web at
http://www. epa.gov/ athens/ onsite.

\R

t\-

Ranges of input for temperature!
four building parameters and
sight soil parameters are allowefl.
These are intended to represent J
equally probable values.

Use of Canines for Indoor Source Detection

The ERD program is designed to demonstrate the utility of dogs as a tool in vapor intrusion
(VI) investigations with an emphasis on evaluating the cost-effectiveness of employing them
developing quality assurance strategies in support of their use, and protecting the dogs from
exposure to harmful levels of chemicals. Benzene, toluene, ethylene, and xylene (BTEX),
major constituents of gasoline and frequent culprits related to vapor intrusion into buildings
from contaminated groundwater, are the initial set of taiget compounds that the dogs are
trained to detect.

Initial tests indicate that air concentrations of BTEX components can be identified by the dogs
down to 1.0 ppb. The dogs discriminate easily between taiget and nontarget compounds (e.g.,
acetone, methanol, acetic acid) and detect taiget compounds in the presence of conflicting
scents (e.g. acetone) at much higher (lOOx) concentrations. A moderate size room or office
area (200 sq ft) can be searched with the dogs in a matter of 2 - 3 minutes.

Sammy examines the foundation of a building
for target scents.

Investigation of Vapor Intrusion Processes

The Johnson and Ettinger vapor intrusion model is widely
used for evaluating which contaminated sites may pose
unacceptable risk for vapor intrusion to overlying
structures. This model incorporates diffusion from the
contaminant source to the overlying building slab, with
advection and diffusion of vapors through the slab into the
building. As a one-dimensional solution, the Johnson and
Ettinger model assumes lateral constancy of parameters.
The effective diffusion coefficient required in the model is
often computed using the Milhngton and Quirk relation.
This relation is very sensitive to porosity, moisture and air-
filled porosity values. The lower the moisture content, the
more vapors will migrate towards a building. Site-specific
VI pathway assessment may include modeling that utilizes
site-measured subsurface moisture content. However,
little information is available regarding the movement of
water beneath and around structures and the subsequent
effect on vapor transport to a building.

A building was constructed on-site at EPA-
Athens and outfitted with instruments to
measure moisture content and soil gas
pressure at several depths beneath and
outside the footprint of the structure.

The unsaturated-zone flow and transport
model HYDRUS-2d was used with two-
years of daily precipitation data to
simulate infiltration beneath and near the
structure.

X

A range of attenuation o one order
of magnitude was produced from the
4096 simulation runs. With this
knowledge decision makers can
choose to use the results given their
uncertainty, measure J &:fe model
" input parameters or indoor air
concentrations.

Sammy is trained to sit as an alert when he
identifies a target compound.

Moisture content results were used in the Johnson and
Ettinger vapor intrusion model to predict allowable soil gas
concentrations over time for a hypothetical benzene
contamination at the site. Results show concentrations at a
location outside the building slab as high as 230 times
concentrations predicted using center-slab concentrations.

Although this work was reviewed by EPA and approved for presentation, it may not necessarily reflect official Agency policy.


-------