An efficient optimization framework of cyclic steam stimulation with experimental design in extra heavy oil reservoirs

The Jurassic sandstone reservoir of the Pre-Caspian basin has abundant heavy oil resources. It has characteristics of high porosity, high permeability, shallow depth, and high crude oil viscosity. So the cyclic steam stimulation (CSS) is necessary to produce oil in an effective and economic way. However, the optimization of injection-production parameters for CSS is the premise of the technical strategies and development schemes. The conventional method usually adopts the one-parameter-at-a-time approach to perform sensitivity studies. This method is lack of interactions between different factors and difficult to determine the optimal level, in some situations even the main controlling factor is ignored. In our work, the experimental design method was used to consider the interactions of multiple injection-production parameters. The quantitative ranking of important factors was also determined. Based on geological characteristics and pilot test of the extra heavy oil reservoirs in Mortuk oilfield, the numerical simulation model was established. Firstly the two-level Plackeett-Burman experimental design was used to screen seven injection-production parameters for the CSS process. The first five parameters, the sum of their influencing weight larger than 98%, are steam injection intensity, production liquid rate, production time, steam quality and injection rate, respectively. Then the three-level Box-Behnken experimental design was used to perform optimizing studies. Two regression equations of cumulative oil production and cumulative oil steam ratio (COSR) were obtained on the basis of the response surface method, which fits very well with the numerical simulation results. Finally, the optimal values of uncertain factors were determined to maximize cumulative oil with relatively high COSR. This optimization framework is helpful for petroleum engineers to provide a quick evaluation of CSS. (C) 2019 Elsevier Ltd. All rights reserved.