Stochastic Optimization Approach to Surfactant-Polymer Flooding

Abstract

Abstract The research presents an effective way for the optimization of one of the enhanced oil recovery mechanism; surfactant-polymer flooding by the application of two stochastic evolutionary algorithms namely Covariance Matrix Adaptation-Evolutionary Strategy (CMAES) and Invasive Weed Optimization (IWO). The optimized parameters include the well placement, time duration for water and chemical flooding, and chemical injection rates in injection wells while net present value (NPV) served as the objective function. Surfactant Polymer (SP) flooding has proved to be an efficient enhanced oil recovery (EOR) mechanism in recent times. Research has been done on the efficiency of SP flooding by optimizing different properties of surfactant and polymer such that the process results in an improved oil recovery. However, these optimizations are based on the sensitivity studies which limit the researchers to search the optimum solution within a specific domain without extensively exhausted the search space. Stochastic techniques, however, showed a way to efficiently optimizes the SP flooding process even with higher number of optimization parameters. Detailed optimization results for several cases considered in this research are presented. Channeled reservoir and fully heterogeneous reservoir are the two reservoir models used for the optimization of SP flooding process. The maximization of NPV for SP flooding using well placement optimization and without well placement optimization is also compared for both reservoirs utilizing CMAES and IWO. Furthermore, all cases are compared with the base case of simple waterflooding. Statistical analysis is done for all the cases for several realizations and the realizations are ranked according to the best, median and worst, based on the NPV values for each case. The consideration of having such a higher number of optimized parameters is to fully evaluate the potential of considered stochastic optimization techniques to converge to a global maximum for NPV as opposed to conventional sensitivity analysis. Results suggested that these stochastic evolutionary algorithms have a potential to be utilized as the optimization tool for field development of reservoirs having higher number of parameters to be optimized. The successful evaluation of considered stochastic evolutionary algorithms (CMAES and IWO) proved that these algorithms can successfully be used for the optimization of field development plan under higher number of parameters to be optimized that cannot be achieved using sensitivity analysis or gradient based optimization algorithms.