Monte Carlo-Based Reliability Estimation Methods for Power Devices in Power Electronics Systems

Abstract

Monte Carlo simulation has been widely used for reliability assessment of power electronic systems. In this approach, multiple simulations are carried out during the lifetime estimation of the components in power converter, e.g., power devices, where the parameter variations are considered. In the previous mission-profile based reliability assessment methods, the dynamic thermal stress profiles are usually converted into a set of static parameters. However, this simplification may introduce a certain uncertainty during the reliability assessment, since the static parameters may not be able to accurately represent the thermal stress under highly dynamic conditions. Moreover, the previous research did not take into account the correlation between the method of introducing the parameter variation and the required number of Monte Carlo simulations. This can significantly affect both the accuracy and computation burden of the Monte Carlo simulation. To address this issue, an in-depth analysis of Monte Carlo simulation applied to reliability assessment of power devices in power electronic systems is provided in this paper. Two additional Monte Carlo simulation approaches based on semi-dynamic and dynamic parameters are proposed, and their reliability evaluation results are compared with the traditional static parameter method. It is demonstrated in a case study of photovoltaic (PV) inverter application that the reliability of power converter can be overestimated up to 30% when using the static parameters.

Keywords

• Power converter
• reliability
• lifetime prediction
• mission profiles
• Monte Carlo methods