Surrogate-assisted evolutionary algorithm with dimensionality reduction method for water flooding production optimization

Abstract

The objective of oil reservoir production optimization is finding optimal scheme of each well to maximize the net present value (NPV) or the hydrocarbon production. Various optimization methods have been proposed to address production optimization problems. Surrogate-assisted evolutionary algorithms have gained increasing attention due to the great performance in reducing the computational resource. However, when dealing with high dimensional optimization problems, surrogate model requires a large number of memories and becomes computationally prohibitive and building a high-quality surrogate model become difficult because of the “curse of dimensionality”. In this paper, a new framework is proposed based on surrogate-assisted evolutionary algorithm and dimensionality reduction to maximize the expected NPV in life-cycle production optimization. To take advantage of surrogate model in low dimension, Sammon mapping is applied to map high-dimensional decision variables to lower dimensions. Then the search can focus on smaller area guided by surrogate model. The optimization results of two examples show the great performance of the proposed method in comparison with classical evolutionary algorithm and different popular dimensionality reduction methods.