Design of distribution transformers based on a knowledge-based system and 2D finite elements

This paper presents the development of a hybrid system for the design of distribution transformers. It is based on a knowledge-based system (KBS) and two-dimensional (2D) finite-element (FE) models, which are used to compute the electromagnetic fields within the transformer. A KBS is a software product based on established artificial intelligence techniques that has been successfully applied in many areas, whereas the FE method is a sound numerical technique for solving partial differential equations. Both methods are combined to offer transformer designers a computational tool for designing distribution transformers. First order elements are employed in two FE models, an axi-symmetric one is used to compute leakage reactance and eddy losses in transformer windings and a 2D cartesian model is used to calculate the iron loss and magnetizing current. The KBS mimics the designers experience through the representation of heuristic, procedural and graphical knowledge. The proposed design system is applied to the design of a 1500 kVA, 13.2 kV–220 V, delta-star, oil-filled distribution transformer.