As manufacturers of power modules, we at SEMIKRON use the ANSYS finite element platform in product development, since this software makes it possible to perform analysis in different physical domains, i.e. mechanical, thermal or electrical properties – including coupled analysis. At first glance, the progress made in terms of the processing power has reached a stage where realistic model simulation for power electronic modules is possible. However, taking a closer look, it becomes apparent that the processing power alone is not sufficient to simulate the complex interactions in real modules.
This begins with the issue that the losses dissipated in power components are themselves temperature-dependent. The widely held assumption that losses are distributed evenly between parallel chips in thermal simulations is rarely a realistic one. A fully coupled electro-thermal simulation, which correctly illustrates the current distribution on the chip and between parallel chips with a temperature-dependent exponential current-voltage curve (diodes, IGBTs), is only possible through significant modifications to the software and is extremely complex and time-consuming when creating models.
But even when you estimate an approximate temperature-independent loss density in the components, modelling the thermal-mechanical characteristics of power modules quickly leads to a fundamental problem, namely that of determining the stress-free temperature for a stack of layers with different coefficients of thermal expansion. In order to achieve this, the production process must be simulated and the plastic, or even visco-plastic, properties of materials and compound layers must be known. It is these very processes of determining the temperature-dependent material properties and performing validation on real samples which are extremely complex and time-consuming. However, these steps are indispensable if you want to progress from relative statements to realistic simulation models.
The expectations of simulation experts in Europe for the future of simulation in the packaging technology of power modules are the focus of an ECPE workshop ("The Future of Simulation in Power Electronics Packaging for Thermal and Stress Management" on 20th and 21st November 2018 in Nuremberg) that I am conducting together with Prof. Dr. Bernhard Wunderle from the Chemnitz University of Technology. In addition to the abovementioned problem areas, the workshop is also intended to discuss the possibility of forming a network to exchange validated temperature-dependent material parameters between experts from universities, research institutes and industry.