Coupling Programming of an Adaptive Genetic Algorithm with Reservoir Flow Simulation for Infill Oil Well Locations Optimization

Abstract

AbstractPlanning of an optimal field development scenario in a mature oil field is a crucial decision since it requires an efficient design of integrated reservoir modeling. One of the most imperative field development scenario is determining the locations of infill wells. The reservoir under study is the main pay in the South Rumaila oil field. This is a mature oil field located in the South of Iraq with around 58 years of production. It has 40 producing wells and 20 injection wells. An efficient reservoir simulation model has been coupled in sequence with a stable and well-conditioned genetic algorithm to optimize the infill well locations. A Genetic Algorithm offers an efficient search method that can be used as a powerful optimization tool by randomly generating potential solutions in order to achieve increasingly better results by applying a set of operators: Selection, Crossover (Recombination), and Mutation.In this study, a black oil reservoir model has been used to evaluate the reservoir and predict its future performance. After attaining a considerable history matching, the simulator has been coupled with an Adaptive Genetic Algorithm (AGA). This algorithm is coupled with the simulator in order to re-evaluate the optimized wells at each iteration. Net Present Value (NPV) has been adopted here as an objective function. The initial population consists of a random number of chromosomes that have eight genes that represent the number of possible new proposed wells that have the highest values of permeability and oil saturation at the last time step. The genetic algorithm takes two parents randomly from the population and produces two children (offspring) by applying the operators such as crossover, mutation, and replacement on the two parents. Subsequently, the program sorts the population' chromosomes from best to worst. The resulting chromosome represents the optimal wells at this iteration. Then the AGA program re-inputs the optimized wells in the input file of the simulator to repeat the processes for many iterations until the optimal solution with the highest NPV is obtained. The entire procedure has led to optimize three infill wells that have the highest NPV among the other solutions.