Summary
Accurately reproducing the coupling of fluid flow in porous media and rock mechanics is crucial for the modeling of CO2 geological storage to properly evaluate and prevent the risk of inducing fault instability during injection operations. As an alternative to using monolithic flow/mechanical suites, the process can be modeled by linking individual codes, that is i, a reservoir simulator for flow and a geomechanical package to account for fluid-induced stress change, which tackle the two problems sequentially. We developed a flexible numerical framework from which different coupling logics can be selected (i.e., one-way coupling, two-way iterative coupling, and two-way explicit coupling), which are characterized by different levels of accuracy and computational costs. A multirate, two-way coupling algorithm, which allows the flow and mechanical simulators to exchange information periodically rather than at every timestep, is also available to reduce the computational cost of two-way coupled simulations. In this work, we use this coupling infrastructure to perform numerical experiments aimed at defining whether sequential iterative coupling is strictly needed or not, and which less expensive logic can be used to attain a similar solution accuracy. First, a synthetic test case is used to illustrate the onset of fault instability during CO2 injection operations for different sets of coupling parameters (type and frequency), rock properties, and fault permeability. It is thus possible to evaluate, for a reasonable range of coupling strength, which depends on fluids and rock properties, the optimal level of coupling. Results are strongly influenced by the coupling strength, and two-way iterative coupling should be selected for tightly coupled systems to accurately reproduce the fault behavior. For a loosely coupled system, instead, the one-way approach should be the preferred choice due to its lower computational cost. Later, we consider CO2 injection into a realistic formation, and we analyze the impact of the coupling frequency on the computational performance. We show that, for complex cases, there is no one-to-one correspondence between the reduction in the number of coupling iterations and the reduction in computational time for increasing coupling period.
