This study proposes an intelligent optimization technique for polymer flooding injection parameters by integrating genetic algorithms (GA) with numerical simulation. It aims to address issues inherent in traditional injection parameter design, such as reliance on experience, poor adaptability, and low computational efficiency. By establishing a closed-loop optimization framework of “generation-evaluation-evolution,” the improved multi-objective GA (DW-NSGA-II) is loosely coupled with reservoir numerical simulators to achieve intelligent optimization of key parameters such as polymer concentration, plug size, and injection rate. The algorithm incorporates a dynamic weight adjustment mechanism to enhance convergence capabilities on non-convex Pareto fronts, adapting to reservoir heterogeneity and dynamic changes during mid-to-late development phases. Experiments conducted on Block M of a mature oilfield in eastern China demonstrated that this method delivers optimized solutions within 3 days—outperforming both traditional manual design (14 days) and standard NSGA-II (8 days). Results show a 10.5% increase in recovery rate and a 10% improvement in cost-benefit ratio, significantly enhancing the technical-economic viability and formulation efficiency of polymer flooding schemes.
Xuewei Yang (Sun,) studied this question.