Title :
Optimization of magnets segmentation for eddy current losses reduction in permanent magnets electrical machines
Author :
Belli, Z. ; Mekideche, M.R.
Author_Institution :
Electr. Eng. Dept., Jijel Univ., Jijel, Algeria
Abstract :
Magnet segmentation is an effective and simple technique for eddy current losses reduction in high power permanent magnets electrical machines. This work aims to study the magnets segmentation effects on eddy current losses in magnets. First, an analysis of eddy currents and associated losses in magnets is made. After, the magnets segmentation effects on losses reduction in the range of the industrial frequencies from few Hz to dozen of kHz are investigated. To achieve these tasks, a 2D non linear model based on finite element analysis is developed under MATLAB environment. The developed model is applied to a synchronous machine with surface mounted permanent magnets. In a second part of the work, we have adopted an optimization process which consists to associate the finite element analysis to genetic algorithm. In this stage of the work, we aim to find the best parameters of magnets segmentation and avoid any segmentations anomaly that appears under some conditions such as skin effect, and which can leads to losses increases instead of their reduction.
Keywords :
eddy current losses; finite element analysis; genetic algorithms; permanent magnet machines; permanent magnets; synchronous machines; 2D nonlinear model; FEA; MATLAB environment; eddy current losses reduction; finite element analysis; genetic algorithm; magnets segmentation; optimization process; permanent magnets electrical machines; skin effect; surface mounted permanent magnets; synchronous machine; Conductivity; Equations; Frequency conversion; Least squares approximations; Nickel; Skin; Windings; eddy current losses; finite elements analysis; genetic algorithm; losses reduction; magnets segmentation; permanent magnets;
Conference_Titel :
Ecological Vehicles and Renewable Energies (EVER), 2013 8th International Conference and Exhibition on
Conference_Location :
Monte Carlo
Print_ISBN :
978-1-4673-5269-7
DOI :
10.1109/EVER.2013.6521595