Modified Rainflow Algorithm for Lifetime Estimation of Semiconductors Devices

Reliable operation is essential for any power electronic system. Nevertheless, power electronic systems include a large number of fragile elements. Among these elements, the semiconductor devices are the most likely to fail. Recent reliability evaluation tools employ a cycle counting algorithm to compute the degradation in semiconductor devices, where the rainflow algorithm is the most popular. However, since the rainflow was first created for fatigue analysis, this algorithm presents some issues when applied for semiconductor devices. Indeed, the rainflow is unable to compute the effective heating time and time-dependent equivalent mean temperature of each thermal cycle. Besides, half-cycles are counted by conventional rainflow. However, apart from the half-cycles at the end of the data, these half-cycles should not be taken into consideration in the life consumption computation. Therefore, this work proposes a modified rainflow for the lifetime estimation of semiconductor devices. This methodology adapts the conventional rainflow to filter half-cycles and compute the effective heating time and the equivalent mean temperature. The impact of the methodology on lifetime estimation is verified in a modular multilevel converter. Besides, analyses with two distinct mission profiles are performed to evaluate the modified impact on lifetime estimation. The results demonstrated that the modified rainflow significantly affects the lifetime computation of all critical joints of an insulatedgate bipolar transistor (IGBT) module for both case studies considered.