The development of a novel multi-objective optimization framework for non-vertical well placement based on a modified non-dominated sorting genetic algorithm-II

Abstract

A single-objective well placement problem is one of the classical optimization problems in oilfield development and has been studied for many years, by researchers worldwide. However, the necessity to face practical applications and handle insufficient data in a single-objective optimization leads to the introduction of a multi-objective optimization framework, which consequently allows an engineer to manage more information. In this study, for the very first time, a multi-objective well placement optimization framework, based on a Non-dominated Sorting Genetic Algorithm-II (NSGA-II) is utilized with a similarity-based mating scheme. To represent the power of this mating procedure, it is compared with two conventional mating selection methods (tournament and roulette wheel selection). In this novel framework, the net present value (NPV) and the recovery factor are considered to be the objective functions while the well coordinates, well types, horizontal section length, orientation, and water injection rate are all assumed as the problem variables. Compared to the tournament and roulette wheel selection methods, the convergence speed analysis of this method indicates a substantial reduction in time, with the number of iterations reduced by 26 and 20%, respectively. Among the mating technique, which is implemented in this work, the final Pareto front, presented in this similarity-based selection method, has more members in the same solution range. Having more individuals in the final Pareto front provides more scenarios for decision makers, which helps them in choosing an optimal scenario based on the limitations and interests of a company.