Parallelization Methods for Implementation of Discharge Simulation along Resin Insulator Surfaces

Three parallelization approaches used in the program of discharge simulation along resin insulator surfaces in SF₆/N₂ gas mixture which initially consumes a great deal of computational time and suitable for implementation on computer systems with symmetric multiprocessor (SMP) architecture are developed and validated in this paper. The approaches include: (i) static domain decomposition using a distributed memory approach, (ii) particle decomposition using a distributed memory approach, (iii) hybrid dynamic domain decomposition and particle decomposition using multilevel distributed and shared memory approaches respectively. All three methods were implemented and tested on an IBM supercomputer system and finally achieved the goal of reducing execution time to a certain extent. Results show that the multi-lever hybrid approach provides an effective parallelization method than method (i) and method (ii) owing to its load balance considering, even arising additional computational time for dynamic domain decomposition.