Oil recovery from heterogeneous porous media is a challenging problem. The current trend in reducing the cost of oil recovery is toward integrated large-scale reservoir modeling whereby fine-detailed geologic models (10 million grid-blocks) and coarse reservoir models (of order of 100,000 grid-blocks) are used to solve complex multiphase flow equations, understand reservoir flow mechanisms and design techniques to improve oil and gas recovery; and (d) predict future reservoir performance. This research project is a multidisciplinary effort by computer scientists and engineers to develop new and novel ways of tapping the ever increasing capabilities of parallel computers to develop a better understanding of multiphase, multi-component fluid flow in porous media. The objectives of this research project are (1) to develop algorithms for loop optimization and data partitioning/clustering for MPP applications; (2) to develop parallel programming tools for specific implementation of the data and computations distribution algorithms on the CRAY T3E; and (3) to evaluate performance of parallel algorithms in solving a large scale reservoir problem.
