Robust ASP Flooding Optimization Under Chemical Properties Uncertainty Using Markov Chain Monte Carlo Optimizer

Abstract

Abstract The paper presents Alkaline, Surfactant and Polymer (ASP) flooding optimization work under the uncertainties in chemical components’ properties in order to assess the risk in simulated incremental oil recovery value associated to uncertainties in chemical components’ properties. Uncertain chemical properties for ASP were identified from published coreflood works and were defined in range instead of as discrete value in the simulation model. 100 chemical properties realizations were generated in the study based on the range of these key uncertain chemical properties and nine representative chemical properties realizations were selected based on the total oil recovery. Robust optimization work was performed using Markov Chain Monte Carlo (MCMC) algorithm to determine the optimum ASP flooding design parameters, namely time to start ASP injection, main ASP slug size, post-flush polymer slug size and ASP concentration that gave highest Net Present Value (NPV) for all the nine selected chemical realizations. Optimized ASP design parameters was eventually run on 100 initially generated chemical properties realizations to generate NPV cumulative probability plot. Nominal optimization workflow that based on single chemical properties realization was used to generate another set of optimized ASP design parameters and the NPV cumulative probability plot generation was followed on the same 100 chemical properties realizations for comparison purpose. The sensitivity of identified uncertain chemical properties on incremental oil recovery is demonstrated in the paper. Both nominal & robust optimization workflows improve the project NPV value compared to base case ASP design, with robust optimization showing further improvement over nominal optimization in all chemical realizations as expected. The spread in NPV clearly illustrated the risk of ASP flooding design related to uncertainties in ASP chemical properties. In this project, the exclusion of chemical properties uncertainties in optimization work led to the underestimation of ASP oil recovery performance. The study is novel as while there were uncertainties in ASP chemical properties reported from laboratory core flood tests or core flood history matching simulations and presence of dynamic chemical adsorption behaviour under different chemical concentration combination, most of the published ASP optimization simulation studies has considered single chemical properties realization in their simulation models. The impact of uncertainties in chemical properties on simulated ASP oil recovery profile is demonstrated in this paper.