Optimal scenario design of steam-assisted gravity drainage to enhance oil recovery with temperature and rate control

Abstract

Steam-assisted gravity drainage (SAGD) is a commonly used thermal enhanced oil recovery (EOR) method in heavy oil reservoirs. Scenario optimizations are conducted with different optimization techniques to determine the optimal steam injection rate and temperature strategies. The performance of standard artificial bee colony (SABC), directed ABC (DABC), generalized pattern search (GPS) and mesh-adaptive direct search (MADS) algorithms were investigated. Also, the effect of initial guess and polling type on the performance of GPS and MADS were analyzed. DABC approaches the global optimum better than other employed algorithms, with a huge number of function evaluations. While, GPS is the fastest algorithm, likely to be trapped in local extrema. To eliminate this issue, the novel multi-region pattern search (MRPS) algorithm is proposed, in which the search space is divided into smaller subregions, each one is searched independently. Hence, search space is more efficiently explored and initial guess dependency is reduced. MRPS algorithm provided similar results to the DABC algorithm while lowering the computational costs up to 93%. MRPS algorithm is successfully applied for a 5-year SAGD scenario optimization. Furthermore, by scenario optimization, SAGD operation could be reduced for 1-year, providing the same NPV as that of the reference case operating for 4-years.