Multi-variable Optimization Methodology for Medium-Frequency High-Power Transformer Design Employing Steepest Descent Method

Annoy Kumar Das,Zhongbao Wei,Baylon G. Fernandes,Haonan Tian,Madasamy P. Thevar,Shuyu Cao,Vaisambhayana B. Sriram,Anshuman Tripathi,Philip C. Kjaer

DOI: https://doi.org/10.1109/apec.2018.8341259

2018-01-01

Abstract:To find balance among multiple design objectives of a medium/high-frequency (MF/HF) high-power (HP) transformer is best addressed employing an optimization technique. In this paper, MF HP transformer design is formulated as a multi-variable optimization problem, where efficiency, power density and temperature rise are chosen as design objectives. Total loss, core volume and maximum temperature rise are modeled as respective cost functions and amalgamated using weighted-sum approach to derive objective function. It is minimized using Steepest descent method. Being a gradient-based search technique, it preserves correlation among design variables during minimization. Using the proposed methodology, optimal design of a 10 kW, 0.5/5 kV, 1 kHz natural oil-cooled transformer with amorphous core and concentric foil winding, is derived. It is estimated to have an efficiency of 99.55% at a power density of 19.79 and maximum node temperature of 52.92 °C. These merit of figures are validated using FEM and CFD studies. Cost-effectiveness of proposed methodology is discussed with the help of a hardware prototype, built from off-the-shelf amorphous core. Benefits like design flexibilities and plausible cost-effectiveness, are inherent to gradient-based optimization method, which augur well for its applicability for MF HP transformer design.