Inexact methods for sequential fully implicit (SFI) reservoir simulation

The sequential fully implicit (SFI) scheme was introduced (Jenny et al. 2006) for solving coupled flow and transport problems. Each time step SFI consists of an outer loop, in which there are inner Newton loops to implicitly sequentially solve the pressure sub-problems. In standard SFI, sub-problems usually solved with tight tolerances at every iteration. This can result wasted computations that contribute little progress towards solution. issue is known as `over-solving'. Our objective minimize cost solvers while maintaining convergence rate SFI. We first extended a nonlinear-acceleration (NA) framework (Jiang Tchelepi 2019) multi-component compositional models, ensuring robust outer-loop convergence. then developed inexact-type methods alleviate It found no need one sub-problem strive perfection, (outer) residual remains high due other sub-problem. new solver tested using several complex cases. problems involve multi-phase EoS-based fluid systems. compared different strategies such fixed relaxations on absolute relative solvers, well adaptive approach. results show basic method quite inefficient. Away from solution, additional accuracy achieved wasted, contributing or reduction overall residual. By comparison, inexact provides adequate current state across wide range conditions effectively resolve over-solving issue, thus greatly improve efficiency.