These slides were presented at the NAFEMS World Congress 2025, held in Salzburg, Austria from May 19–22, 2025.

Abstract

Efficient thermal management in electric vehicle (EV) batteries is essential for maintaining performance, safety, and longevity of the EV battery. However, traditional approaches rely heavily on analytical methods to calculate heat transfer coefficients, which can be time consuming and inaccurate, given the dynamic nature of variables like the state of charge and ambient temperature. This paper proposes an approach that integrates cloud-based simulations and AI-based surrogate models, to enable predictive insights into battery performance. This paper focuses on a battery pack of an EV to study its thermal response at different operating conditions. High fidelity cloud-native simulations are used to create a design space for the battery temperature and coolant flow heat transfer coefficient at different ambient temperatures and battery state of charges (SOC). Each individual simulation took 45 to 60 minutes, the parallelization capabilities of cloud platform provided a training dataset of 24 parametric cases within 1 hour. The next step focuses on leveraging the data from these simulations to predict instantaneous heat transfer coefficients for electric vehicle (EV) coolant systems. So, the initial simulations, each requiring close to an hour, can be replaced by surrogate models that generate the same results in less than a minute. This speed is critical for enabling the dynamic adjustments required to manage rapid variations in EV battery conditions during operation. In real-world driving conditions, these models can deliver predictions in real-time for dynamic adjustments to manage rapid changes in battery conditions. Machine learning algorithms enable real-time prediction of thermal conditions, crucial for optimizing coolant power based on varying state of charge and ambient temperature, enhancing EV batteries'™ longevity and performance. Furthermore, the integration of cloud-based simulations and AI models ensures scalability, supporting the development of increasingly complex battery designs as well as enabling integration into digital twin environments. These advancements mark a significant step toward achieving safer, more efficient, and sustainable energy solutions for a decarbonized future.