A Streamlined Test-Oriented Lifetime Assessment Framework for Automotive Power Electronics Converters

Automotive power electronics are critical components in electric vehicles (xEVs), forming the backbone of the powertrain, where their performance and durability directly impact overall vehicle reliability. Current state-of-the-art lifetime testing and modeling methods for Power Electronics Converter (PEC)s rely on extensive experimental testing and/or high-fidelity physics-based models to assess degradation over the vehicle's lifetime, which is both resource-intensive and time-consuming. Consequently, there is increasing demand from automotive Original Equipment Manufacturers (OEM)s to incorporate reliability and lifetime evaluation in the early design phases of PECs, such as DC/DC converters and power inverters, prior to entering series production. To address this challenge, this paper proposes a streamlined methodology for testing, modeling, and lifetime assessment of a 150 kW SiC-based traction inverter. The inverter is tested on an industry-grade automotive test bench across multiple driving profiles. Accurate multiphysics modeling is performed using the test data, focusing on electro-thermal models for bond-wire degradation and thermo-mechanical models for die-attach degradation. A reliability assessment model integrates these multiphysics modeling results with statistical lifetime estimation. Lifetime assessments across various driving profiles reveal the influence of profile characteristics on reliability, highlighting that bond-wire degradation accounts for 91% of total lifetime degradation, while die-attach failure contributes only 9%. This work demonstrates that a systematic, test-driven methodology, combined with a collaborative lifetime assessment framework, significantly enhances confidence in the design of automotive power electronics. The proposed framework can be seamlessly integrated into the early design phase to optimize design outcomes and accelerate the design-to-market process. Compared to a conventional high-fidelity model-based approach, the proposed framework reduces mission-profile-based lifetime assessment effort by approximately 90%.