Modeling Complex Processes
Enhanced Oil Recovery (EOR) processes usually involve complex phase behavior between the injected and in-situ rock-fluids system. In order to represent the flow dynamics of EOR recovery processes accurately, compositional formulations are required. However, the computational cost of compositional flow simulation increases dramatically with the number of components and model resolution. To improve the robustness and efficiency of compositional simulation, a negative-flash method in conjunction with tie-simplex parameterization of the compositional space was developed and implemented in our GPRS simulator. This negative-flash-based, tie-simplex technique leads to significant improvements in the stability, accuracy, and efficiency of compositional flow simulation in highly detailed reservoir models.
Unstructured grids allow for accurate representation of the complex geometry of subsurface formations. Our flow simulation framework (GPRS) makes no assumptions about the underlying grid structure. We work on advanced discretization methods for heterogeneous reservoir models with unstructured grids, including discrete fracture models, and we also develop advanced linear and nonlinear solvers for thermal-compositional simulation using highly detailed models with generally unstructured grids.
Convergence of the Fully Implicit Method (FIM) usually requires special nonlinear solvers. Here, we compare the number of nonlinear iterations needed to solve a compositional problem for four different methods: (1) using local chopping (damping), (2) flash-reduction of the unknowns, (3) under-relaxation (globally), and (4) using trust regions of a flux function.