Coupled fluid flow and geomechanical deformation modeling

Abstract

Accurate prediction of reservoir production in structurally weak geologic areas requires both mechanical deformation and fluid flow modeling. Loose staggered-in-time coupling of two independent flow and mechanics simulators captures much of the complex physics at a substantially reduced cost. Two 3-D finite element simulators—Integrated Parallel Accurate Reservoir Simulator (IPARS) for flow and JAS3D for mechanics—together model multiphase fluid flow in reservoir rocks undergoing deformation ranging from linear elasticity to large, nonlinear inelastic compaction. The loose coupling algorithm uses a high-level driver to call the flow simulator for a set of time steps with fixed reservoir properties. Pore pressures from flow are used as loads for the geomechanics code in the determination of stresses, strains, and displacements. The mechanics-derived strain is used to calculate changes to the reservoir parameters (porosity and permeability) for the next set of flow time steps. Mass is conserved in the coupled code despite dynamically changing reservoir parameters via a modification to the Newton system for the flow equations, and an approximate rock compressibility becomes a useful preconditioner to help with convergence of the modified flow equations. Two numerical experiments illustrate the accuracy of the coupled code. The first example is a quarter-five-spot waterflood undergoing poroelastic deformation, which is validated against a fully coupled simulator. Vertical displacements at the well locations match to within 10%. Moreover, experimentation shows that 13 mechanics time steps (taken over the course of 5 years of simulation time) were sufficient to achieve this result (a substantial cost savings over full coupling in which both the mechanics and flow equations must be solved at each time step). The second numerical example is based on real data from the Belridge Field in California, which illustrates one of the complex plastic constitutive relationships available in the coupled code. The results mimic behavior which was observed in the field. The coupled code serves as a prototype for loosely coupling together any two preexisting simulators modeling diverse physics. This technique produces a coupled code relatively quickly and inexpensively and has the advantage of accurately modeling complex nonlinear phenomena often observed in a real field but difficult to capture with a fully coupled simulator. Further, the code has produced promising results when used for time-lapse studies of compactible reservoirs.