FEM-based Modeling and Optimization of Dry-Type Transformers with Metaheuristic Algorithms


Creative Commons License

Kul S., TEZCAN S. S. , Duysak H., Celtek S. A.

Tehnicki Vjesnik, vol.29, no.5, pp.1678-1685, 2022 (SCI-Expanded) identifier

  • Publication Type: Article / Article
  • Volume: 29 Issue: 5
  • Publication Date: 2022
  • Doi Number: 10.17559/tv-20220114203522
  • Journal Name: Tehnicki Vjesnik
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Aerospace Database, Communication Abstracts, Compendex, INSPEC, Metadex, Directory of Open Access Journals, Civil Engineering Abstracts
  • Page Numbers: pp.1678-1685
  • Keywords: dry-type transformer, energy efficiency, metaheuristic optimizations, optimization, SL-PSO
  • Gazi University Affiliated: Yes

Abstract

© 2022, Strojarski Facultet. All rights reserved.Transformer optimization is a programming problem with an objective function for calculating the characteristics of a transformer in detail according to user requirements. Especially in recent years, efficient and optimum design of the transformer has become increasingly important. This study presents a comparative analysis of the application of metaheuristic optimization algorithms in transformer design for maximum efficiency of a three-phase dry-type transformer. In addition, transformer modeling was done using FEM analysis, and its magnetic characteristics were shown. The main contribution of this paper is to optimize the determined basic design parameters with optimization methods that have not been used in transformer optimization before and to maximize efficiency by reducing losses. During this process, using the loss constants of the materials in the basic loss equation for core loss, this process was tried to be carried out without any change in the method in applications to be made with different materials. This situation also strengthens the accuracy of the applied method. Crow Search (CSA), Moth-Flame Optimization (MFO), Vortex Optimization (VOA), Particle Swarm Optimization (PSO), and Social Learning-Particle Swarm Optimization (SL-PSO) algorithms were used in the study. The loss values were obtained by performing loaded and unloaded FEM analysis with the ANSYS/Maxwell program. As a result, the best result was obtained with SL-PSO as max 99,05%.