Optimization of Well Placement by PSO Assisted by Quality Map and Gompertz-Based Grey Model

Abstract

Determining optimal well locations is a challenging work because engineering and geologic variables affect reservoir performance which are often highly nonlinearly correlated and uncertain. Particle swarm optimization algorithm (PSO) could solve the well placement optimization problem, but it has the shortcomings such as low convergence speed and long computation time. This paper proposed a well placement optimization model based on PSO assisted by quality map (QM) and Gompertz-based grey model (GGM), and the well locations can be determined in the condition of use of numerical simulation in more fewer times. We conduct a numerical experiment which is developing an oil field by the four well placement arrangements namely the PSO+QM, PSO+QM+GGM, PSO, and original well placement plan (BASE) with the maximization of cumulative oil production as the objective function. It is shown that the PSO+QM method outperforms PSO+QM+GGM method, and PSO+QM+GGM method outperforms PSO and BASE method. The cumulative oil production of well placement generated by PSO+QM is 4.89 × 106 m3 which is 1.24% more than the PSO+QM+GGM based well placement plan (4.83 × 106 m3), but the cost time of PSO+QM is 11.25 h which is 47.64% more than the PSO+QM+GGM based well placement plan (7.62 h). Taken in total, the PSO+QM+GGM method has the superiority both in optimization results and computation time, which can provide a reliable basis for well placement optimization in real oil fields.