Parallel Computation Of Meshfree Methods For Extremely Large Deformation Analysis

Jf Zhang,Y Zheng
DOI: https://doi.org/10.1007/3-540-27912-1_84
2005-01-01
Abstract:Due to the heavier computation requirement than other competitive techniques and the essence of applications that are usually highly complex and computationally intensive, parallel computing is especially attractive for these meshfree methods. We only focus on the Reproducing Kernel Particle Method (RKPM), one of the meshfree methods for large strain elasto-plastic analysis of solid and structures, in considering with its ability to accurately model extremely large deformations without mesh distortion problems, and its ease of adaptive modeling by simply changing particle definitions for desired refinement regions. The parallel procedure primarily consists of a mesh partitioning pre-analysis phase, and a parallel analysis phase that includes explicit message passing among partitions on individual processors. With redefinition techniques applied to the shared zones of different geometrical parts, the graph-based procedure Metis, which is quite popular for mesh-based analysis, is used for partitioning in this meshfree analysis. Parallel simulations have been conducted on an SGI Onyx3900 supercomputer with MPI message passing statements. The effectiveness and performance with different partitions has then been compared, and a comparison of the meshfree method with finite element methods is also presented.
What problem does this paper attempt to address?