Comparison of Numerical Methods for Thermal Performance Evaluation of Circuit Protection Devices in EV Application

Abstract

With the growing demand of electric vehicles, design of circuit protection devices is now an important consideration in automobile industry. Modern day circuit protection devices have been constantly undergoing miniaturization due to requirement of minimizing the foot print for use in electrical vehicles and aerospace applications. This size reduction makes thermal management one of the most important aspects of their design. Use of numerical model to predict heat transfer can significantly reduce the cost and time required in testing physical prototypes. In this paper, three different approaches for numerically predicting temperature rise of circuit breakers are discussed and compared from the point of view of accuracy and computational effort. The three methods are 1) Finite volume based analysis in which conjugate heat transfer inside and outside the breaker is modelled by solving Navier-Stokes equations 2) Finite element based heat conduction model in which convection is modelled as boundary condition instead of solving for fluid motion, and 3) Thermal network based model which uses electrical analogy of heat transfer to solve a thermal resistance network. In the first two iterative models mentioned above, heat generation from current-carrying parts is calculated by solving Maxwell’s equations of electromagnetics by Finite element method. Eddy current losses and temperature dependence of electrical conductivity is considered in the calculation of heat loss. In all three methods, electrical and thermal contact resistances are added at appropriate locations based on analytical calculations. All three methods have been validated with temperature rise test results. In this paper, the heat loss and temperature of a molded case circuit breaker have been predicted by all three methods discussed above. It is observed that the Finite volume-based method is the most accurate amongst the three methods. It can computationally predict air motion and air temperature at critical locations. However, this additional accuracy comes at the cost of added effort in terms of additional mesh count and computation. The Finite elementbased method gives good accuracy but does not predict air temperature. The analytical network-based model is less accurate compared to other methods and relies on product expertise and experience. Based on the study, the following recommendations are made:1) The finite element-based method is best suited to evaluate designs which do not alter flow pattern significantly 2) The finite volume method is recommended to evaluate effect of flow altering design changes 3) The network-based model is recommended for initial evaluation of correct cross sections of current carrying members.