High-resolution numerical methods for compressible multi-phase flow in hierarchical porous media. Progress report

This is the first year in the proposed three-year effort to develop high-resolution numerical methods for multi-phase flow in hierarchical porous media. The issues being addressed in this research are: Computational efficiency: Field-scale simulation of enhanced oil recovery, whether for energy production or aquifer remediation, is typically highly under-resolved. This is because rock transport properties vary on many scales, and because current numerical methods have low resolution. Effective media properties: Since porous media are formed through complex geologic processes, they involve significant uncertainty and scale-dependence. Given this uncertainty, knowledge of ensemble averages of flow in porous media can be preferable to knowledge of flow in specific realizations of the reservoir. However, current models of effective properties do not represent the observed behavior very well. Relative permeability models present a good example of this problem. In practice, these models seldom provide realistic representations of hysteresis, interfacial tension effects or three-phase flow; there are no models that represent well all three effects simultaneously. Wave propagation: It is common in the petroleum industry to assume that the models have the same well-posedness properties as the physical system. An example of this fallacy is given by the three-phase relative permeability models; they were widely assumed by the petroleum community to produce hyperbolic systems for the Buckley-Leverett equations, but later the mathematics community proved that these models inherently produce local elliptic regions. Since numerical methods must use the models for computations, oscillations that develop could erroneously be attributed to numerical error rather than modeling difficulties. During this year, we have made significant progress on several tasks aimed at addressing these issues.