AI-Based Finite-Element Simulation Automation Workflow Applied to Automotive Electronics Field

Electronic component is one of the key design elements considered during the development of automotive ECU (Electronic Control Unit). As the current state-of-the-art method, FEM (Finite-Element Method) simulation is widely used for thermal design optimization and solder joint reliability assessment of electronics at early stage of ECU development. To address the challenges of modeling fidelity and efficiency, the AI (Artificial Intelligence) integrated automated simulation workflow platform is developed. Leveraging standard and Bosch in-house developed Python libraries called “pyact”, an Ansys-based automation toolset has been developed which accelerates the simulation workflow. With the assistance of AI methodologies, more precise electronic component models could be expected during calibration process. This paper presents 3 different industrial practices and examples focusing on the board level thermal management and solder joint reliability assessment. The first industrial example presents the standardization and automatization of the reliability simulation workflow of chip-resistor components. Utilizing Ansys ACT (Application Customization Toolkit), basic APIs (Application Programming Interface) of ACT are encapsulated into the Python library as pyact, enabling a fully automated simulation workflow to predicted solder joint lifetime. To further realize the fast assessment of reliability and component virtual release, an AI metamodel is derived from automated simulation-based DoE (Design of Experiments). For pursuit of accuracy for simulation models, the second industrial example presents the combination of Bayesian Optimization with BGA (Ball Grid Array) warpage calibration process to outperform the traditional manual methods. The third industrial example outlines another application of Bayesian optimization into the Zth (Transient Thermal Impedance) calibration process for component thermal models. Utilizing the multi-objective optimization algorithm, the Zth curves from different sources & thermal paths can be fitted with high-quality by the machine. In conclusion, the integration of AI with simulation automation significantly benefits thermal management and solder joint reliability evaluations in ECU development.