Optimization of Gas-condensate Reservoir EOR Technology under Geological Uncertainties

on of the technology for a pilot field application, we are still highly uncertain about property distribution and even fluid properties and their true interaction with injected chemicals. The problem becomes even more complicated when we also have to optimize the implementation of the technologies based on technical or economic efficiency. The current paper proposes the workflow that addresses the problem described above. The classic approach for enhancement of condensate recovery is implementation of gas recycling using hydrocarbon or nonhydrocarbon gases. This proves to be efficient method in cases when reservoir pressure is maintained close to dewpoint, preventing in-situ condensation of liquid fractions. The problem of current study is a synthetic deep gas-condensate reservoir that was developed under depletion, resulting in significant decrease of reservoir pressure way beyond dewpoint with formation of the liquid phase, which is only mobile in the vicinity of the wells, where critical saturations were achieved. Being uncertain about geological description of the reservoir, facies distribution, porosity, permeability, and SCAL data, we want to identify the most economically feasible chemical EOR technology and optimize its parameters under uncertainties. Using a numerical compositional simulator with a synthetic reservoir model, we performed optimization of a field development project’s net present value (NPV) for different chemical EOR methods – surfactant (S), alkaline (A), polymer (P), AS, SP, and ASP for the duration of injection (slug volume) under geological uncertainties within different static reservoir property realizations, fault transmissibilities, aquifer strength, and relative permeability endpoints. Optimization was done using a simplex algorithm combined with risk assertion to account for geological uncertainties. Results indicate repeatability and applicability of the proposed approach on real full-field gas-condensate models.