EnginSoft - Conference Abstracts

EnginSoft International Conference 2009
CAE Technologies for Industry

Optimal Design of a Safety Isolating Transformer with modeFRONTIER

Gong Jinlin - Ecole Centrale de Lille “L2EP” (France)
Vu Tran Tuan - Ecole Centrale de Lille “L2EP” (France)
Gillon Frédéric - Ecole Centrale de Lille “L2EP” (France)
Brochet Pascal - Ecole Centrale de Lille “L2EP” (France)

Abstract

To design a system without seeking to draw the best part of the elements used is a pity. The current tools for optimization make it possible to do optimal design. For that, models, optimization methods and a methodology of design are needed [1]. The modeFRONTIER (mF) software is used to design a safety isolating transformer [2]. This paper focuses on the optimal design for improving the energetic efficiency and mass of an analytical and a finite elements model. Three parts are achieved to complete this work:
Firstly, different types of optimization problems are tested with analytical functions for single objective and multiobjective functions. The results e.g. the optimal point or the Pareto front are obtained by mF and compared with results obtained by other optimization approaches. Fig. 1 shows the optimization result of Tanaka test problem [3].
Secondly, the optimization of a safety isolating transformer is studied through an analytical model. The simplified model is based on a set of assumptions for the structure and the physical properties. This analytical model is built with MATLAB® and then connected with mF. The simulation progress with the proposed model is very fast (a few seconds) but it lacks precision in the magnetic and thermal part. Fig. 2 shows the structure of a safety isolating transformer. The design problem has 7 factors which can be continuous and discrete.
Thirdly, a more accurate model is used, which is a coupling of magnetic and thermal 3D finite element model. A simulation point takes about half an hour with this accurate model. The two models will be compared in terms of accuracy and CPU time.