Model‐based multi‐objective particle swarm production optimization for efficient injection/production planning to improve reservoir recovery

Abstract

AbstractThis study employs an adjusted version of the multi‐objective particle swarm optimization (MOPSO) algorithm to plan an optimized reservoir's injection/production strategy. Three case studies, including two water‐flooding benchmark models and one gas‐condensate problem, are exercised as subjected problems to validate the MOPSO approach. The contradicting values of objectives, long‐term net present value (LNPV) versus short‐term net present value (SNPV), are obtained so that relying on a Pareto front improves decision‐making. In one water‐flooded case, inflow control valves of smart wells are considered to be adjusted within the optimization, while in the second case, the optimized well injection rates are control variables. The obtained results for water‐flooded reservoirs are shown to optimize the competitive objective functions more than the previous efforts in the literature. Moreover, 10 different permeability maps of the second case are implemented to obtain the optimum injection rates to perform optimization under uncertainty. The MOPSO robustly optimized the production/injection strategy in the presence of model uncertainty. In the gas‐condensate problem, the optimal gas injection rate in the SPE‐3 benchmark model is determined. The gas‐condensate case's outputs yield a decreased oil saturation result in the reservoir compared to non‐optimized production scenarios. Results illustrate that for all cases, MOPSO can provide optimal injection/production scenarios. Therefore, the proposed scheme gives the advantage of deciding between the set of results into a decision‐maker to optimize the production program by trading‐off within different strategies. Besides the Pareto fronts with adequate variety and steadiness, a great converging rate is the main advantage of this method.