Optimal Design of a 3-Phase Core Type Distribution Transformer Using Modified Hooke and Jeeves Method

Abstract

Hooke and Jeeves method is de facto a pattern search technique, which can be employed for getting an optimal solution. In this paper the method, in a modified form, has been applied for the design optimization of a distribution transformer. It is a constrained multi-variable optimization problem. The solution is obtained by choosing an initial point in the world map of the key variables and by making a local search (exploratory in all directions in the hyper surface formed by the variables. After recognizing the pattern, its advantage is taken by moving towards a lower cost point, using an acceleration factor for faster convergence. The step length is adjusted as we proceed to expedite improvement. The method has been applied to two different cost functions: the cost of production and the cost against production plus capitalized running losses. In both the cases, the problem has converged to a solution and the results are both interesting and illuminating.