||CMES: Computer Modeling in Engineering & Sciences, Vol. 19, No. 2, pp. 135-144, 2007
||Full length paper in PDF format. Size = 129,539 bytes
||Grid generation, Seamless connection, Finite element method, Speedup ratio, Parallel efficiency.
||A new parallel mechanism for Node-based Seamless Finite Element Method was proposed in this paper, which possessed the following three prominent points: realizing the workload balance for the parallel processes naturally, achieving synchronization of all the schedules under complex parallel environment, and filling up the gap between pre-processing and main processing. To support the scheme, three specific solutions of the parallel mechanism were proposed in this paper, all of which achieved the highly efficient parallel seamless connection between the FEM mesh generation process and structure analysis process. Two of the three schemes, i.e. dynamic scheme and self-adaptive scheme, enabled both schedules achieving synchronization of the whole process counting from FEM mesh generation to global stiffness matrix computation. The good tolerance of the two schemes to processor failure was also pointed out here. The parallel efficiency of the schemes was shown by both the theory analysis and initial parallel numerical examples results.