Practical CO2—WAG Field Operational Designs Using Hybrid Numerical-Machine-Learning Approaches

Machine-learning technologies have exhibited robust competences in solving many petroleum engineering problems. The accurate predictivity and fast computational speed enable a large volume of time-consuming processes such as history-matching field development optimization. Southwest Regional Partnership on Carbon Sequestration (SWP) project desires rigorous multi-objective optimization processes, which fits the superiorities machine-learning approaches. Although proxy models are trained validated before imposing to solve practical problems, error margin would essentially introduce uncertainties results. In this paper, hybrid numerical workflow various problems is presented. By coupling expert proxies with global optimizer, successfully solves CO2 water alternative gas (WAG) design problem low overheads. work considers heterogeneities multiphase relative characteristics, CO2-WAG injection takes multiple techno-economic objective functions into accounts. This an response surface, support vector machine, multi-layer neural network effectively learn high-dimensional nonlinear data structure. proposed suggests revisiting high-fidelity simulator for validation purposes. experience gained from provide valuable guiding insights similar enhanced oil recovery (EOR) projects.