Title :
Mesh Decomposition for Efficient Parallel Computing of Electrical Machines by Means of FEM Accounting for Motion
Author :
Böhmer, Stefan ; Lange, Enno ; Hafner, Martin ; Cramer, Tim ; Bischof, Christian ; Hameyer, Kay
Author_Institution :
Inst. of Electr. Machines, RWTH Aachen Univ., Aachen, Germany
Abstract :
The relative motion between stator and rotor in electrical machines requires a flexible representation in 2-D and 3-D finite element (FE) models. Numerous approaches to incorporate the relative motion are available. Due to increasing problem size and accuracy, parallel computing becomes more desirable. The parallel simulation of sufficiently large problems often involves domain decomposition algorithms, especially if distributed memory systems are used for the parallelisation. Accounting for motion usually requires explicit domain decomposition at each simulation step. This paper proposes an approach avoiding the computationally expensive domain decomposition at each step by properly choosing an initial decomposition, which can be applied to all steps throughout the simulation.
Keywords :
distributed memory systems; electric machine analysis computing; finite element analysis; machine theory; parallel processing; rotors; stators; FEM accounting; distributed memory system; electrical machines; finite element model; mesh decomposition; parallel computing; relative motion; rotor; stator; Computational modeling; Iron; Parallel processing; Partitioning algorithms; Rotors; Runtime; Stators; Multithreading; nonconforming mesh; parallel algorithms; rigid motion;
Journal_Title :
Magnetics, IEEE Transactions on
DOI :
10.1109/TMAG.2011.2176472