Simulating Fluid Flow through a Geologically Realistic Permeable Medium

LAKE, LARRY W., University of Texas at Austin, Austin, TX

For the past five years we have been studying the Page Sandstone outcrop in Northern Arizona as a prototype for eolian oil reservoirs and aquifers. The work has encompassed detailed geological studies and highly intensive descriptions of the distribution of hydraulic conductivity or permeability. We are now at the point of using this information to evaluate predictions about fluid flow through naturally occurring sands. Work on this topic is the subject of this presentation.

The first step in the study was to establish a "truth case," a standard against which to measure the success of alternative procedures for flow prediction. Our truth case consisted of a highly detailed (more than 11,000 finite element nodes) numerical simulation of a miscible displacement through a portion of the outcrop wherein each nodal property was assigned according to the actual value existing at that point. This intensely deterministic simulation also allows us to evaluate the extent of geologic detail necessary for a good prediction; mobility ratio and the distribution of average permeability are the most important quantities governing fluid flow. Permeability anisotropy, fourth order-bounding surfaces, and dispersion are much less important.

With the truth case in hand, we redo the simulations based on the amount of data that would normally be available for subsurface conditions. In this case, of course, it is the generation of interwell properties that is being tested. The work investigates two basic procedures: pseudofunctions and conditional simulation. The pseudofunction approach generates effective relative permeabilities (a concept borrowed from immiscible flow) to model the effects of heterogeneities on a gross or flow unit scale. The effective relative permeabilities now depend on the size of the flow units, the mobility ratio, and the nature of the distribution of the permeability within the flow unit. The conditional simulation approach generates interwell properties statistically and evaluates each procedure ac ording to whether the distribution of the simulation results agrees with the results of the truth case.

Pseudofunctions rely almost exclusively on geologic information and conditional simulation. We should be able to infer the merits and drawbacks of both methods from this work.

AAPG Search and Discovery Article #91004 © 1991 AAPG Annual Convention Dallas, Texas, April 7-10, 1991 (2009)