Flow in Porous Media Parallel Project

Mary F. Wheeler, Research Director; Todd Arbogast, Clint N. Dawson, Philip T. Keenan, Marcelo Rame, and Nai-Ying Zhang, Center for Research on Parallel Computation, Rice University

Oil and gas, resources which currently cannot be replaced, provide two- thirds of the energy consumed in the United States. Although oil can be imported, there are profound advantages to domestic production, such as promoting a healthy national economy and insuring a stable supply and price for the consumer. The United States has provided world leadership in energy research, but recently there have been dramatic reductions in aggressive research programs due to economic restraints.

Exploration for new fields is for the most part prohibitively expensive. Therefore, the only viable way to uncover new reserves is through the use of enhanced oil recovery (EOR) processes and reservoir characterization technologies for existing fields. EOR now accounts for approximately 11% of domestic production, with the potential for significant increases. Without EOR, viable wells in existing fields are capped and permanently abandoned.

High performance computing and computer simulation are playing increasingly vital roles in geophysical modeling, reservoir characterization, and EOR. Parallel reservoir simulators have the potential to quickly, economically, and accurately model large, complex field problems. This includes the ability to use an adequately refined discretization mesh, to incorporate complex chemical and physical effects, and to employ stochastic or conditional simulation. The latter is essential in simulating realistic reservoir fields, since much of the data cannot be quantified accurately and since often the chemical processes are not well understood. Conditional simulation is ideally suited to parallel computation.

The goal of the Flow in Porous Media Parallel Project, which is administered through the Geophysical Parallel Computation Project of the Center for Research on Parallel Computation, is to investigate ways to use high performance parallel processing as a tool to simulate flow behavior in petroleum reservoirs. Current objectives include the development of advanced numerical and parallel algorithms for modeling flow in porous media to simulate EOR. Much of this research also has applications to the remediation of toxic waste sites and the restoration of groundwater contaminated by organic chemicals. The group is also working to obtain realistic reservoir/aquifer characterization. These models are essential in predicting the response of reservoirs to complicated processes and for the understanding, designing, and testing of economically feasible recovery or decontamination strategies.

Wheeler's research group plans to develop a new-generation parallel reservoir simulator within the next five years, and to work on integrating reservoir simulation and geophysical inversion/imaging technology. Collaborations with researchers from several oil companies and universities contribute to the development of the algorithms and simulators. An active industrial affiliates program provides a testbed for project applications as well as funding, resources, and technology transfer that contribute to the success of the project.

Table of Contents