A Computationally Efficient Iterative Solver for Reservoir Models with Faults/Pinchouts

摘要

An incomplete Gaussian elimination of a reduced system(RS) is one of the most efficient preconditioners of an iterative linear solver for reservoir simulation models. The reduced system is formed from eliminating half the unknowns from problems whose matrices have regular red-black ordering. This precludes the application of the RS method to reservoir simulation models with faults and pinchouts, which incur irregular connections. This paper extends the method to solve Jacobian matrix generated from such a reservoir simulation model. A generalized minimum residual method is used to accelerate the solution.