A novel approach to multi-objective efficiency optimisation for a distribution transformer based on the Taguchi method

In this paper a new approach to multi-objective design optimisation of a distribution transformer is proposed. It is based on a combined pattern search optimisation technique and the Taguchi methodology. The objective was to increase the transformer efficiency by reducing the total transformer loss, coupled with the aim to minimise the iron and copper mass, and hence to attain a new, more efficient, lower cost product. The transformer that was selected for this research work was an existing oil immersed three-phase, 50 kVA, 20/0.4 kV unit from the production line of the Rade Koncar factory in Skopje. The Taguchi method and the Design of Experiments (DoE) were employed to achieve the optimisation objective and to minimise the total mass of the active materials. A non-linear mathematical model of the transformer was developed and an objective function, presenting the efficiency, was derived. Pattern search optimisation technique for higher efficiency, lower loss, and lower overall mass was invoked; and the outcome was an optimal transformer design. The results, both analytical and numerical using 3D FEM, have been validated through measurements from a new prototype. They show very good agreement, and prove the new approach was effective and accurate.