A Novel Approach for Reservoir Forecasting under Uncertainty

Abstract

Abstract This paper presents a novel approach for developing statistical estimates of uncertainty in reservoir performance from only a limited number of multiphase flow simulations while using information from a full suite of geostatistical realizations. The proposed methodology involves ranking and weighting geostatistical models on the basis of a surrogate performance measure, and estimating the mean/variance of the desired performance measure via a weighted average of simulation results from a few selected realizations. The streamline time-of-flight based volumetric sweep efficiency is used as the surrogate performance measure to rank geostatistical models, and their weights are determined from a moment-matching algorithm which enables the approximation of a continuous distribution by a discrete distribution. A field example is used to demonstrate the applicability and computational efficiency of the proposed methodology.