Robust Fully-Implicit Coupled Multiphase-Flow and Geomechanics Simulation

Abstract

Summary Material nonlinearity, boundary and arching constraints, nonuniform reservoir flows, sliding along material interfaces, or faults are among the causes of shear deformation or changes in the total stresses and the resulting stress redistribution in hydrocarbon reservoirs. Previous studies have demonstrated that shear or nonuniform deformation and stress redistribution in subsurface formations may have significant effects on reservoir fluid flows. Thus, a two-way coupled analysis is the required approach under circumstances where the shear deformation or changes in total stresses in the reservoir cannot be neglected. A coupled multiphysics simulator is developed for the dynamic modeling of multiphase thermal/compositional flow, and poroelastoplastic geomechanical deformation. The equations that govern multiphase flow in permeable media, heat transport, and poroelastoplastic geomechanics together lead to a highly nonlinear system. Finite-volume and Galerkin finite-element methods are used for the numerical solution of thermal/compositional multiphase fluid-flow and geomechanics equations on general hexahedral grids, respectively. Because of its improved stability and rapid convergence characteristics, the resulting multiphysics system of equations is solved with a fully-implicit formulation by use of an effective implementation of the Newton-Raphson method in the default mode. The coupled simulator is by design maximally modular with self-contained flow and geomechanics modules that can be operated in a two-way coupled mode with explicit-, iterative-, and fully-implicit-coupling options. The coupled-modeling system lends itself naturally not only to near-wellbore coupled flow and geomechanical deformation problems where poroplasticity may play a more prominent role, but also to reservoir-scale simulations where both poroelasticity and poroplasticity are relevant. The coupled simulator is validated against analytical solutions for simple cases, by use of published data in the open literature. Validation results demonstrate the robust, fast, and accurate predictive capabilities of the multiphysics modeling protocol.