Multiscale Simulation Framework for Coupled Fluid Flow and Mechanical Deformation
A multiscale linear-solver framework for the pressure equation associated with flow in highly heterogeneous porous formations was developed. The multiscale based approach is cast in a general algebraic form, which facilitates integration of the new scalable linear solver in existing flow simulators. The Algebraic Multiscale Solver (AMS) is employed as a preconditioner within a multi-stage strategy. The formulations investigated include the standard MultiScale Finite-Element (MSFE) andMultiScale Finite-Volume (MSFV) methods. The local-stage solvers include incomplete factorization and the so-called Correction Functions (CF) associated with the MSFV approach. Extensive testing of AMS, as an iterative linear solver, indicate excellent convergence rates and computational scalability. AMS compares favorably with advanced Algebraic MultiGrid (AMG) solvers for highly detailed three-dimensional heterogeneous models. Moreover, AMS is expected to be especially beneficial in solving time-dependent problems of coupled multiphase flow and transport in large-scale subsurface formations.
- Research Organization:
- Board of Trustees of the Leland Stanford Junior University
- Sponsoring Organization:
- USDOE Office of Science (SC)
- DOE Contract Number:
- FG02-06ER25726
- OSTI ID:
- 1164145
- Report Number(s):
- DOE-STANFORD-ER25726
- Country of Publication:
- United States
- Language:
- English
Similar Records
Linear and Nonlinear Solvers for Simulating Multiphase Flow within Large-Scale Engineered Subsurface Systems
On performance of Krylov smoothing for fully-coupled AMG preconditioners for VMS resistive MHD