Leveraging Multiphysics with AI/ML for Power Converter Thermal Design and Reliability

Maciej Tyniec, Product Specialist and Juan Rosales, Application Engineer Electronics at Altair present at the 2024 ATCx Electronics conference.

The well-known challenge for electronics manufacturing is to keep control of the cost, time, and quality of the product. A key design decision of a modern power converter is the thermal design where tradeoffs between device selection, switching speed, and thermal system are critical. More than half of modern electronics products face thermal failures on multiple fronts starting from the soldering process, down to component degradation and aging. There are laboratory procedures to cover these problems, which are reaching their limits. Using simulation - virtual validation approach - in place of hardware testing is very attractive for cost reduction and the ability to exhaustively test almost all scenarios. Converging multiphysics with AI/ML data analytics can make this process easy and effective.

This presentation highlights workflows that converge high-fidelity multiphysics simulations together with AI/ML data analytics - the best current solution to address this situation.