Optimization of the Steam Alternating Solvent Process Using Pareto-Based Multi-Objective Evolutionary Algorithms

Steam alternating solvent (SAS) process has been proposed as a more environmentally friendly alternative to traditional steam-based processes for heavy oil production. It consists of injecting steam and a non-condensable gas (solvent) alternatively to reduce the oil viscosity. However, optimizing multiple process design (decision) variables is not trivial since multiple conflicting objectives (i.e., maximize the recovery factor, reduce steam–oil ratio) must be considered. Three different multi-objective evolutionary algorithms (MOEAs) are employed to identify a set of Pareto-optimal operational parameters. A multi-objective optimization (MOO) workflow is developed: first, a 2D reservoir model is constructed based on the Fort McMurray formation. Second, a sensitivity analysis is performed to identify the most impactful decision parameters. Third, two response surface (proxy) models and three different MOEAs are employed and compared. This paper is the first to compare different MOEAs for optimizing a wide range of operational parameters for the SAS process. The results show that if more steam is injected, extending the steam cycle duration is preferable. Conversely, if more solvent is injected, it is recommended to start with injecting a solvent with high propane concentrations over short cycles and switch to lower propane concentrations over long cycles near the end.