Interaction analysis and multi-response optimization of transformer winding design parameters

This work conducted a multi-response optimization on the design parameter of a transformer winding based on the response surface methodology and desirability function approach to find a design with better combined-thermal-hydraulic performance. A numerical experiment was performed with two-dimensional axisymmetric numerical heat transfer models. Response surface models were established based on experimental results and checked through analyses of variance. Regression equations were fitted and then verified. Interactions among factors were investigated in detail. A multi-response optimization was carried out on the winding design parameters and optimum design was identified using the desirability function approach. The results showed that the response surface models built up in this work were all significant and all the responses could be predicted accurately based on the regression equations. The design with the block washer number of 1, axial duct size of 8 mm and radial duct size of 3 mm was found as the optimum winding design of this work. Compared with the conventional design, the optimum design increased the Colburn-j factor and Darcy friction factor by 212% and 9%, respectively, simultaneously decreased the maximum and mean temperature rises by 67.1% and 70.5%, respectively.