Global optimization of steamflooding in complex heavy oil reservoirs

Heavy oil is a class of oil whose viscosity is above 10 cP. In recovering heavy oil, steam flooding has been the most successful Enhanced oil recovery (EOR) method. Generally, Injection and production wells are arranged in a specific pattern. However, reservoirs are not always conducive to the common patterns typically due to the heterogeneity that could hinder the movement of the front resulting in heat losses. Estimating the optimal values of the operating parameters for existing reservoir conditions are therefore challenging. Stochastic optimization methods capable of handling many variables globally were employed in this work. Two cases were considered: Case 1, where only the injection rate is optimized and Case 2, where well placements are optimized along with the injection and production rates. Three sample reservoir models (relatively homogenous, fully heterogeneous & discontinuous channeled reservoirs) were used. The objective function was the flood's net present value (NPV) and Covariance Matrix Adaptation Evolution Strategy (CMA-ES) was used as the optimizer. Results indicated that NPV could be increased significantly for Case 2 in all the models. Discontinuity in heavy oil reservoirs would not be a problem if the optimum well placement is considered.