--> ABSTRACT: Implications of Upscaling Submarine Channel Reservoir Models, by Safian Atan, Michael H. Gardner, James M. Borer, and Roger Wagerle; #90906(2001)

Datapages, Inc.Print this page

Safian Atan1, Michael H. Gardner1, James M. Borer1, Roger Wagerle1

(1) Colorado School of Mines, Golden, CO

ABSTRACT: Implications of Upscaling Submarine Channel Reservoir Models

Reservoir upscaling is a standard step in converting a high-resolution geological model to a more CPU manageable fluid flow simulation model. Upscaling involves increasing the cell size in the model, either the areally (upgridding) or stratigraphically (uplayering).

To test the impact of upscaling on models of submarine channel reservoirs, a 2D-outcrop geologic model was converted to a two-million cell deterministic reservoir model. The modeled middle Brushy Canyon Formation consists of stacked submarine channel complexes and associated overbank deposits. Over 400 1D profiles and 140 surfaces characterize the 120-m thick and 1770-m wide cross-sectional area. A matrix of fifty-six realizations of the base-case outcrop model was constructed by systematically increasing the cell size, first in the horizontal direction (grid) then in the vertical direction (layers). This process identified the optimum combination of layer and cell manipulation required to produce a visual match to the base-case model. The simulation involves a simple injector-producer pressure maintenance scenario. Single-phase, black oil properties and water are injected to maintain a constant reservoir pressure.

Simulation results indicate recovery efficiency is over-estimated when upscaled realizations fail to preserve fine-scale heterogeneity. Upscaled models that produce a good visual match, but average out centimeter-scale shale layers tend to flow an average of 45-50 percent better than higher-resolution models that include these fine-scale heterogeneities. The loss of fine scale heterogeneity through upscaling produces an approximate 40 percent change in ultimate oil recovery, depending on the realization (poor to good visual match) compared to the base-case model.

AAPG Search and Discovery Article #90906©2001 AAPG Annual Convention, Denver, Colorado