Connectionist model for predicting minimum gas miscibility pressure: Application to gas injection process

Abstract

Improving the recovery factor of conventional oil reservoirs is not a far-fetched target when injecting miscible gases is discussed in technical Enhanced Oil Recovery (EOR) plan. Considering the leading role that Minimum Miscible Pressure (MMP) factor plays in the scenario of a miscible gas injection, and the significant impact that it does have on the sweep efficiency of the injected gas is inevitable. Because of the expensive, difficult and time consuming laboratory techniques which are used to obtain the MMP, concluding a quick, robust and cheap solution to measure the MMP has been turned into petroleum researchers’ priorities. In the current study, Least Square Support Vector Machine (LS-SVM) and evolutionary algorithms (for example, Genetic Algorithm (GA) and Imperialist Competitive Algorithm), both addressed in previous literatures, have been employed to estimate the MMP. A set of laboratorial data accessible in the open literature was gained to test the reliability of the proposed HGAPSO-LSSVM model which its generated results have been compared with the other proposed intelligent approaches. Moreover, the performances of both implemented solutions certify statistically the strong potential of models in prediction of the MMP.