To optimize well pattern during miscible gas injection process via heuristic techniques

Abstract

To detect optimum well locations for new injection wells one must generally notice to existing well locations, reservoir shape, capital cost per well construction in each area as well as recovery increase relevant to new well layout. Designing injection process presents inverse optimization issue whereas adjusting exploitation wells prior to primary recovery exhibits classic optimization problem. Most important design factor namely injection well layout has no unique explicit solution to solve via analytical or numerical techniques thus one can propose meta or hyper heuristic approaches to solve such problems. Our main novelty in this work is that here we cooperate four heuristic optimization approaches with our self made fast simulation technique to optimize well locations for two exemplar miscible gas injection processes. We employ MATLAB software to construct artificial porous media prototypes with different uncertainties to investigate optimization technique performances in low rather than high risk media. Outcomes illustrate that approaches like artificial immune system or particle swarm optimization which exhibit strong local search can effectively detect optimal well locations in high risk porous media. Results also show that solution techniques such as Monte Carlo that have strong global search capabilities at expense of weak local search can discover optimum well locations only in media with low uncertainty.