Uncertainty and Approximation of a Highly Heterogeneous Oil Reservoirs when Solving Problems of Automatic History Matching

Abstract

Summary The paper presents the results of the study of uncertainties and the quality of approximation of highly heterogeneous oil reservoirs. These studies are performed by solving the inverse problem. The inverse problem solution is based on modeling a multiphase flow. For the research, synthetic production data were used. These data are obtained for a highly heterogeneous reservoir model constructed on the basis of a well-known model from the SPE-10 project. The results of two experiments are presented. In the first experiment, automatic history matching was carried out over the entire (rather long) lifetime of a synthetic field. In the second experiment, automatic history matching was carried out for 2/3 of the field life, and a forecast was made for the remaining 1/3. It is shown that the use of regular structures in the form of Voronoi cells in automatic history matching makes it possible to obtain “regular” models of highly heterogeneous reservoirs, which give highly accurate forecasts of oil production. Uncertainty of reservoir properties in some of its zones, which arises when performing automatic history matching, is associated with low production from these zones due to their high water cut or low permeability.