218
Accounting
for Model Uncertainties Using Reliability Methods - Application to Carbon Dioxide
Geologic Sequestration System--Geomatrix
Consultants, Inc., 2101 Webster Street, 12th Floor, Oakland, CA 94612-3027; 510-663-4100; www.geomatrix.com
Dr. Chin Man Mok, PhD, Principal Investigator, cmmok@geomatrix.com
Mr. Martin
Mullins, Business Official, mmullins@geomatrix.com
DOE Grant No.
DE-FG02-07ER84932
Amount: $99,976
Numerical
modeling is a critical tool for designing effective subsurface management
programs, including groundwater remediation and waste repositories. Because constitutive parameters and
subsurface-condition variables often are not known to the desired accuracy, the
results of model predictions are inherently uncertain; therefore, engineering
designs that account for model prediction variability are more reliable and are
more likely to achieve performance goals.
In recent years, stochastic models have been developed to incorporate
model input-output relationships and model input uncertainties into a
computational framework, in order to estimate the variability of model
predictions. Although these models focus
on capturing the mean, variance, and covariance of predicted values, the
computed probabilistic measures might not be sufficient for accurately
estimating the probability of extreme events.
This project will develop a distributed/parallel probabilistic modeling
code, CALRELTOUGH, which will utilize reliability methods to incorporate
parameter sensitivity and uncertainty analysis into subsurface flow and
transport models. The CALRELTOUGH code
will be based on the state-of-the-art reliability analysis code, CALREL, and the
parallel version of the state-of-the-art subsurface flow and transport code,
TOUGH2. For this project, we will apply
the CALRELTOUGH model to a carbon dioxide geologic sequestration system, in
order to benchmark performance.
Commercial
Applications and other Benefits as described by the awardee: The
CALRELTOUGH code should enable the efficient and accurate assessment of the
uncertainty associated with numerical modeling.
It can support the design of subsurface management systems, optimizing
costs and reducing the likelihood of negative impacts from extreme events at
thousands of government and private sites around the world. The current environmental remediation market
is $6.4B annually; the subsurface repository market (e.g., nuclear waste, CO2,
natural gas, petroleum, and water) has a similar or even greater market size.