DocumentCode :
3732445
Title :
GPU acceleration of EBE method for 3-D linear steady eddy current field
Author :
Dongyang Wu;Xiuke Yan;Renyuan Tang;Dexin Xie;Zhiwei Chen
Author_Institution :
School of Electrical Engineering, Shenyang University of Technology, Shenyang Liaoning, China
fYear :
2015
Firstpage :
325
Lastpage :
328
Abstract :
Parallel computing of three dimension (3D) linear steady eddy current problem based on element by element (EBE) method is proposed in this paper. Compared with the classical finite element method (FEM), EBE technique shows a different way in storing matrix, solving equations and dealing with boundary condition. Then Jacobi preconditioned bi-conjugate gradient (BiCG) method is used to solve equations which are discritized from elements level. Furthermore, we can fulfill the parallelization at elements level. In order to testify the accuracy of proposed method, TEAM workshop problem 7 is adopted as a numerical model. The numerical experiment is carried out on multi-core parallel platform named graphic processing units (GPU) with compute unified device architecture (CUDA). And the results demonstrated that EBE method and GPU acceleration are effective in solving large scale eddy current problem.
Keywords :
"Finite element analysis","Graphics processing units","Decision support systems","Eddy currents","Jacobian matrices","Acceleration","Three-dimensional displays"
Publisher :
ieee
Conference_Titel :
Electrical Machines and Systems (ICEMS), 2015 18th International Conference on
Type :
conf
DOI :
10.1109/ICEMS.2015.7385052
Filename :
7385052
Link To Document :
بازگشت