Inverse Modeling of Reservoirs with Tilted Fluid Contacts

Abstract

Summary In this paper, we present a new approach for simulating reservoirs with tilted fluid contacts produced by hydrodynamics. The proposed method solves a nonlinear inverse problem to determine the aquifer flow field that best reproduces the observed contact tilt. The computational effort required to solve this inverse problem is reduced by choosing a pressure-based objective function and applying gradient-based optimization. This approach is entirely automated, in contrast to previous works that have used laborious trial-and-error methods to estimate the aquifer flow field. In addition, the proposed method introduces no additional physics beyond hydrodynamics to model reservoirs with tilted contacts. The proposed method is integrated into a parallel reservoir simulator. A synthetic reservoir is constructed by introducing an artificial tilt, and the new approach is applied to estimate the aquifer flow field. The estimate produced by the proposed method matches the true flow field well and is able to prevent large fluid motions near the contact surface when simulating production from the reservoir. The proposed method is compared with an existing approach that uses capillary pressure adjustments to hold the tilted contact in place. The proposed method is shown to outperform the existing approach without significantly impacting the simulation results.