Parallel electromagnetic-mechanical coupled analysis using combined domain decomposition method

Abstract

Since the electromagnetic-mechanical coupled analysis requires large computation time, development of the parallel processing techniques is inevitable. In this paper, the parallel computing technique with the combination of the domain decomposition method and the domain partitioned conjugate gradient method is proposed. The method to evaluate the parallel performance is presented and discussed for the coupled problem using a workstation cluster. This method is efficient for a large scale coupled problem of a magnetic fusion device component.

Electromagnetic Scattering Using the Iterative Multiregion Technique

“…A class of time and memory-efficient algorithms divides the computational domain into smaller subdomains and then combines the subdomain solutions after introducing the effect of interactions between these subdomains. A group of methods that decomposes the computational domain into subdomains is known as the domain decomposition method (DDM) [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17]. These methods in general require common boundaries between subdomains and boundary conditions are enforced on subdomain interfaces.…”

“…Therefore, the complexity of the problem can be reduced, and a time and memory efficiency algorithm can be achieved. Another advantage of the domain decomposition methods is that they are suitable to develop parallel processing techniques [14][15][16] and thus enable highly scalable algorithms.…”

The Iterative Multi-Region Algorithm Using a Hybrid Finite Difference Frequency Domain and Method of Moment Techniques

“…Assuming that the analysis for region 1 is performed using the MoM solution, while that of region 2 is performed using the FDFD method. Since in this paper the MoM is used to simulate conducting structures only, electric current is calculated due to an incident plane wave excitation based on equation (14). While the scattered electromagnetic fields due to an incident wave are calculated on region 2 (FDFD solution).…”

“…Therefore the complexity of the problem can be reduced, and a time and memory efficiency algorithm can be achieved. Another advantage of the DDM methods is that they are suitable to develop parallel processing techniques [14][15][16], and thus enable highly scalable algorithms.…”

“…A class of time and memory efficient algorithms is developed through which, the computational domain is divided into smaller sub-regions and then the sub-regions solutions, after introducing the effect of interactions between these sub-regions, are used to provide the entire domain solution. A group of methods that decomposes the computational domain into sub-domains is known as the domain decomposition methods (DDM) [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17]. These methods in general require common boundaries between sub-domains and boundary conditions are enforced on sub-domain interfaces.…”