Parallel Computing of Local Mesh Finite Element Method [chapter]

G. Yagawa
2001 Computational Mechanics–New Frontiers for the New Millennium  
Efficient parallel computational procedures for incompressible and compressible flows with the free mesh method (FMM) are presented. The method utilizes data structure based on node, and it is featured by the node-by-node procedure in local mesh generation and finite element assembling. The distribution of loads among multiple processors is fulfilled through the nodal identification number, and the operations for preparing the communication among processors and for keeping load balance among
more » ... m are handled in a quite simple manner. As a numerical example of parallel computing with the FMM, large-scale cavity problem of incompressible flow is computed. In the example, parallel efficiency of almost 100% is obtained on the construction of system equations, using 64 processors of Hitachi SR2201. Another numerical result of compressible flow analysis is also shown for the problem of two-dimensional transient supersonic flow over a step. In this computation, the parallel efficiency of more than 90% is achieved on time integration process, using a 101,381 DOF model and nine Alpha PCs.
doi:10.1016/b978-0-08-043981-5.50008-0 fatcat:hxgd7lklbnaa3asffxonfwdvte