On the application of artificial bee colony (ABC) algorithm for optimization of well placements in fractured reservoirs; efficiency comparison with the particle swarm optimization (PSO) methodology

The application of a recent optimization technique, the artificial bee colony (ABC), was investigated in the context of finding the optimal well locations. The ABC performance was compared with the corresponding results from the particle swarm optimization (PSO) algorithm, under essentially similar conditions. Treatment of out-of-boundary solution vectors was accomplished via the Periodic boundary condition (PBC), which presumably accelerates convergence towards the global optimum. Stochastic searches were initiated from several random staring points, to minimize starting-point dependency in the established results. The optimizations were aimed at maximizing the Net Present Value (NPV) objective function over the considered oilfield production durations. To deal with the issue of reservoir heterogeneity, random permeability was applied via normal/uniform distribution functions. In addition, the issue of increased number of optimization parameters was address, by considering scenarios with multiple injector and producer wells, and cases with deviated wells in a real reservoir model. The typical results prove ABC to excel PSO (in the cases studied) after relatively short optimization cycles, indicating the great premise of ABC methodology to be used for well-optimization purposes.