A Statistical Approach to Electro-thermal Simulations with Contacts

All devices in the power grid have to withstand high short-time currents, that may occur as a consequence of short-circuits. The thermally critical areas during short-circuit testing of power devices are usually the electrical contacts between conductors. Connections and contacts with insufficient short-time current withstand capability may lead to local melting of the material and failure of the respective tests during the product development. When thermal performance is evaluated by simulations, the actual electric contact resistance is often unknown and its value is usually simply assumed or estimated. In this work, we consider the possibility of overcoming such uncertainty by making a step towards a statistical approach. We examine how simulating for a range of possible electric contact resistances can translate into a probability distribution of the maximum temperature and, hence, into an estimation of a likelihood of failure. The paper provides simulation results of industrial cases calculated and compared to experiments as well as example results of the proposed idea. The method will help optimize the use of conductor material, reduce the testing effort, and develop power devices with an increased short-circuit rating.