Impact of lifetime model selections on the reliability prediction of IGBT modules in modular multilevel converters

Abstract

Power cycling in semiconductor modules contributes to repetitive thermal-mechanical stresses, which in return accumulate as fatigue on the devices, and challenge the lifetime. Typically, lifetime models are expressed in number-of-cycles, within which the device can operate without failures under predefined conditions. In these lifetime models, thermal stresses (e.g., junction temperature variations) are commonly considered. However, the lifetime of power devices involves in cross-disciplinary knowledge. As a result, the lifetime prediction is affected by the selected lifetime model. In this regard, this paper benchmarks the most commonly-employed lifetime models of power semiconductor devices for offshore Modular Multilevel Converters (MMC) based wind farms. The benchmarking reveals that the lifetime model selection has a significant impact on the lifetime estimation. The use of analytical lifetime models should be justified in terms of applicability, limitations, and underlying statistical properties.