| Article ID | Journal | Published Year | Pages | File Type |
|---|---|---|---|---|
| 6918678 | Computer Methods in Applied Mechanics and Engineering | 2012 | 9 Pages |
Abstract
A major computational issue in the Finite Element (FE) integration of coupled consolidation equations is the repeated solution in time of the resulting discretized indefinite system. Because of ill-conditioning, the iterative solution, which is recommended in large size 3D settings, requires the computation of a suitable preconditioner to guarantee convergence. In this paper the coupled system is solved by a Krylov subspace method preconditioned by a Relaxed Mixed Constraint Preconditioner (RMCP) which is a generalization based on a parameter Ï of the Mixed Constraint Preconditioner (MCP) developed in [7]. Choice of optimal Ï is driven by the spectral distribution of suitable symmetric positive definite (SPD) matrices. Numerical tests performed on realistic 3D problems reveal that RMCP accelerates Krylov subspace solvers by a factor up to three with respect to MCP.
Related Topics
Physical Sciences and Engineering
Computer Science
Computer Science Applications
Authors
Luca Bergamaschi, Ángeles MartÃnez,