Multiobjective Design Optimization of Stator for Synchronous Generator Using Bat Algorithm and Analysis of Magnetic Flux Density Distribution

Abstract

— In this study, we aimed to optimize 3000 kVA synchronous generator (SG) stator design to obtain the desired magnetic flux density distribution and efficiency. We used Maxwell simulations for experiments on some design parameters of stator (slot height and teeth width). Then second-order regression models are calculated that represent the relations between the factors (design parameters) and the measured performance criteria (called as the responses: stator-teeth flux density, stator-yoke flux density, and efficiency). These regression models are used at the multiobjective optimization phase. Bat algorithm (BA) is used for performing the multiobjective optimization. By combining Maxwell with regression modeling and BA, the efficiency of the SG is increased to 96.84% from 96.5% with a more acceptable magnetic flux density (between 1.65 and 1.70 T ranges). The stator-teeth flux density and stator-yoke flux density are calculated as 1.9 T and 2.07 T for the current SG, whereas these values are reduced to 1.647 and 1,634 T, respectively, for the optimized SG. Results of this study show how the numerical simulation can be successfully combined with the BA to improve the efficiency of the SG by providing the desired magnetic flux density distribution.