Well Placement Optimization Under Constraints Using Modified Cuckoo Search

Abstract

Abstract A new robust metaheuristic optimization method, namely modified cuckoo search (MCS), is presented in this paper. MCS is inspired by breeding behavior of cuckoo birds and combined with Lèvy flight approach to efficiently search for optimal solutions. MCS is coupled with a filtering technique to provide the ability to handle nonlinear constraints. The filter-based MCS is efficient insofar as it provides a bias toward exploration during early generations allowing for global search and then shifts that bias toward exploitation at final generations allowing to search promising areas of the solution. This helps in finding feasible solutions at earlier search stages and consequently improves convergence rate. Two example cases involving two-dimensional synthetic reservoir models are presented. The first case compares the performance of MCS to that of genetic algorithm (GA) to maximize oil recovery by optimizing the location of four injection wells. It is shown that MCS outperforms GA in terms of the optimal solution as well as the rate of convergence. The second case entails the use of filter-based MCS to maximize NPV under maximum water cut constraint at the production well. The results indicate the superior performance of the filter-based MCS as it was able to quickly find feasible solutions even though all previous initial solutions were infeasible. The incorporation of filtering technique allows to assess the sensitivity of the objective function to the constraint violation. This provides additional insights that can lead to better future planning.