Parallel processing of 3D rigid-viscoplastic finite element analysis using domain decomposition and modified block Jacobi preconditioning technique

Abstract

In three-dimensional finite element analysis of forging processes, a large number of elements is needed for improvement in the accuracy of solutions. This results in increased computation time and memory requirements, which still remain major problems for effective three-dimensional metal forming analyses. In the present study, a parallel computation technique was adopted for rigid-viscoplastic approach in order to increase the computational efficiency. The domain decomposition algorithm and preconditioned conjugate gradient (CG) iterative solver were used for parallel computation, and a new preconditioning technique was proposed. In the proposed preconditioning technique, the block Jacobi preconditioner was applied to interior degrees of freedom with the Jacobi preconditioner being applied to interface degrees of freedom of each sub-domain. Using the proposed parallel computation and preconditioning techniques, simple upsetting simulations of a cubic type workpiece were carried out. From simulation results, it was found that the proposed method is more efficient compared to both the conventional CG and the Jacobi preconditioned conjugate gradient (J-PCG) solving techniques. (C) 2002 Elsevier Science B.V. All rights reserved.