Joint optimization of constrained well placement and control parameters using teaching-learning based optimization and an inter-distance algorithm

Abstract

Well placement and parameter optimization (WPPO), which is a complex problem, is important in the petroleum industry. Among derivative-free methods, population-based metaheuristics global optimization algorithms have proved to be more pertinent than other methods. One should note that population based methods have drawbacks of pre-mature convergence of the algorithm which is due to poor tuning of parameters. In this study, a population-based metaheuristics global optimization method independent of parameter tuning called teaching-learning-based optimization (TLBO), has been utilized for WPPO. Particle swarm optimization (PSO) and genetic algorithm (GA) were added for comparison. Furthermore, three joint optimization of well location and parameters were implemented over a synthetic black-oil reservoir and maximization of the net present value (NPV) as the fitness function. First, joint optimization of well location and bottom-hope pressure (BHP) over fixed number of wells, followed by joint optimization of well type and BHP over fixed location of first scenario is investigated. Finally, number of production and injection wells, their location and BHP values were jointly optimized. In addition, a new algorithm was proposed to maintain an inter-distance constraint for any number of points in an ordinary n-dimensional space to avoid well interference problem. Based on the results, TLBO conferred considerable higher rate of convergence and within a fixed number of iterations, achieved NPV, 1–14% and 1–52% higher than the GA and PSO, respectively.