Power-Twin

In the field of power electronics, semiconductor-based power module manufacturers have developed highly advanced design strategies that incorporate a variety of functions. However, while reliability is considered from the design phase, it is still evaluated experimentally at the end of the process, after the creation of physical prototypes. Reliability is thus approached with an initial 'global' method based on a statistical analysis of accelerated test results performed on a large number of components of the same technology. Power cycling is the most relevant reliability test, but it requires long testing campaigns. These tests involve applying amplified multiphysical stresses compared to those present under normal operating conditions to accelerate degradation mechanisms. Extrapolation to lower stress levels allows for estimating the lifespan under real-world conditions. However, the gap between predictions made by this approach and field observations is significant. Additionally, there is no feedback into the design phase, limiting the ability to adjust and improve initial designs based on insights from reliability tests. The variability of operating conditions is not sufficiently accounted for, which further weakens lifespan predictions. Lastly, component durability is often overlooked, as current models and practices do not promote a proactive approach to designing for reliability and longevity. PowerTwin addresses the need to develop new methodologies and tools to improve reliability prediction, reduce manufacturing costs, and optimize power module design in a rapidly evolving technological context.