Electromagnetic and thermal parameter identification method for best prediction of temperature distribution on transformer tank covers

• Patricia Penabad-Durán ^[1] ; Paolo Di Barba ^[2] ; Xose M. López-Fernández ^[1] ; Janusz Turowski ^[3]
  1. [1] Universidade de Vigo

     Universidade de Vigo

     Vigo, España

     
  2. [2] University of Pavia

     University of Pavia

     Pavía, Italia

     
  3. [3] Technical University of Lodz

  Mostrar afiliaciones +
• Localización: Compel: International journal for computation and mathematics in electrical and electronic engineering, ISSN 0332-1649, Vol. 34, Nº 2 (Special Issue: Electromagnetic Fields), 2015, págs. 485-495
• Idioma: inglés
• Enlaces
  • Texto completo
• Resumen

  • Purpose – The purpose of this paper is to describe a parameter identification method based on multiobjective (MO) deterministic and non-deterministic optimization algorithms to compute the temperature distribution on transformer tank covers.

    Design/methodology/approach – The strategy for implementing the parameter identification process consists of three main steps. The first step is to define the most appropriate objective function and the identification problem is solved for the chosen parameters using single-objective (SO) optimization algorithms. Then sensitivity to measurement error of the computational model is assessed and finally it is included as an additional objective function, making the identification problem a MO one.

    Findings – Computations with identified/optimal parameters yield accurate results for a wide range of current values and different conductor arrangements. From the numerical solution of the temperature field, decisions on dimensions and materials can be taken to avoid overheating on transformer covers.

    Research limitations/implications – The accuracy of the model depends on its parameters, such as heat exchange coefficients and material properties, which are difficult to determine from formulae or from the literature. Thus the goal of the presented technique is to achieve the best possible agreement between measured and numerically calculated temperature values.

    Originality/value – Differing from previous works found in the literature, sensitivity to measurement error is considered in the parameter identification technique as an additional objective function. Thus, solutions less sensitive to measurement errors at the expenses of a degradation in accuracy are identified by means of MO optimization algorithms.