Applying Response Surface Method to Oil-Immersed Transformer Cooling System for Design Optimization

In this paper, the response surface method (RSM) is exploited in the cooling system design optimization of oil-immersed transformers. To generate the training data for RSM, a 3-D computational fluid dynamic model for the transformer under study is constructed and solved, followed by verification on the solutions. Based on the simulation data, two types of RSMs, i.e., the second-order polynomial method and the Kriging method, are employed to evaluate their performance for this specific problem. The response surface with the best performance is used in the optimization process. The solution is compared with that found by the direct optimization method to showcase the effectiveness of applying RSM to oil-immersed transformer cooling system design. Details of implementation of the studied transformer are also presented.