A Bayesian Framework for Enabling Predictive Simulation and Uncertainty Quantification in History Matching Geological Models for CO₂ Injection

Research Group

Victor Ginting (PI), Jarret  Barber and Felipe Pereira (Co-PIs), Arunasalam Rahunanthan (Research Scientist) and Armad Jan (graduate student).

Funding

Center for Fundamentals of Subsurface Flow and School of Energy Resources

Project Cost

$233,861 (2010-2013)

Abstract

This proposal outlines a three-year research project to quantify uncertainty related to injection of CO₂ into saline aquifers. The activities proposed here are in immediate conjunction with a research proposal to be submitted before April 30, 2010 to the DOE-NETL (DE-FE0000250). Upon successful application, the funds requested in this proposal will be designated as matching funds for the externally funded project. Our main goal in this proposal is to develop a reliable framework for predictive simulation and uncertainty quantification of CO₂ injection. Our thesis is that, for a wide range of applications, the ideal situation is to integrate available static and dynamic data with sufficiently accurate mathematical models which is to be implemented in high performance computing simulations. This is best accomplished within a Bayesian framework using Markov Chain Monte Carlo methodology. To achieve this goal, successful application of this proposal will enable a strong collaboration between applied mathematicians and a statistician to attack various problems in connection with the modeling of CO₂ flow after injection and uncertainty quantification associated with it.

Recent Results

F. Pereira, "Multi-stage Markov Chain Monte Carlo methods for Porous media flows," Keynote lecture presented at the 1st International Symposium on Uncertainty Quantification and Stochastic Modeling (Uncertainties 2012), Maresias, São Sebastião, São Paulo, February 26, to March 2, 2012.

Download presentation