--> --> Abstract: Preserving Geologic Heterogeneity in Reservoir Simulation Models, by D. Stern and A. G. Dawson; #90933 (1998).
[First Hit]

Datapages, Inc.Print this page

Abstract: Preserving Geologic Heterogeneity in Reservoir Simulation Models

Stern, D. and Dawson, A. G. - Exxon Production Research Co.

Preserving geologic features that influence flow behavior, such as high permeability thief zones or barriers to vertical flow, is important in building reservoir simulation models. This poster describes results obtained using Exxon's current proprietary scale-up technology to produce a simulation Previous HitmodelNext Hit of a deepwater field with a thin oil rim and gas cap. The simulation Previous HitmodelNext Hit comprises gridblocks that conform to the initial fluid contacts and layers selected to conform to geologic heterogeneity. Figure 1 shows a plan view of top-of-structure depth contours used to construct the simulation grid. The simulation grid contains finer gridding in the oil rim, and coarse gridblocks in the aquifer. Figure 2 compares permeability cross-sections of the geologic Previous HitmodelNext Hit and the simulation Previous HitmodelNext Hit. The poster describes some of the procedures used to ensure that scale-up is accurate, some of the problems that can reduce the accuracy of scale-up, and the practical solutions to these problems.

Calculation of effective permeability (scale-up) is carried out using a pressure-solver method similar to those described elsewhere. This involves constructing a fine calculation mesh that corresponds to each gridblock, populating it with permeability values from the geologic Previous HitmodelNext Hit, and solving the equations of Darcy flow. Potential problems in applying this technology result from 1) the boundary conditions used to solve the flow equations and 2) inaccuracies in property look up from the geologic Previous HitmodelNext Hit. In solving the flow equations, two assumptions are made, both of which can introduce inaccuracies in scale-up. First, a no-flow condition is imposed on all gridblock boundaries transverse to flow. This can lead to inappropriately low permeabilities because the no-flow boundaries prevent fluid from moving around small barriers within the gridblock. Second, a constant pressure is applied at the inlet and outlet end of each gridblock. As a result of this assumption, flow barriers outside the gridblock are ignored, sometimes leading to poor preservation of flow obstructions in the Previous HitmodelNext Hit. Exxon's scale-up software reduces look-up errors by conforming the calculation mesh to the underlying geologic Previous HitmodelNext Hit rather than the coarse gridblock. Errors can still result, however, when the simulation Previous HitmodelNext Hit grid does not align with the geologic Previous HitmodelNext Hit grid, and when layers are relatively thin. The poster describes detection of these errors and Previous HitverificationNext Hit that they are small.

We have also applied a recently developed method for selecting simulation layering so as to preserve geologic Previous HitmodelNext Hit heterogeneity as much as possible. To do this, an objective function is defined in terms of differences in easily calculated flow properties (such as single-phase breakthrough time), and coarse-grid layering is selected to minimize this objective function while conforming to stratigraphic boundaries. Simple flow calculations are used to roughly estimate the error that is introduced by coarsening the layering. This is used as a first-pass indication of how many layers are needed. The layering for this deepwater example was selected using this method and modified so the Previous HitmodelTop can account for aquifer influx and gas coning, which are expected to dominate performance of the field.

AAPG Search and Discovery Article #90933©1998 ABGP/AAPG International Conference and Exhibition, Rio de Janeiro, Brazil