Loading...

Presenting a Machine Learning-Based Approach for Predicting Thermal Runaway Phenomena in Lithium-Ion Battery Packs in Electric Vehicles

Movahedian Attar, Delaram | 2024

0 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 57514 (46)
  4. University: Sharif University of Technology
  5. Department: Energy Engineering
  6. Advisor(s): Sattari, Sourena; Moeini Aghtaie, Moein
  7. Abstract:
  8. Nowadays, lithium-ion batteries are widely used in electric vehicles due to their higher energy density and longer life cycle compared to other storage systems. However, safety concerns about catastrophic risks like fires exist, caused by thermal runaway in lithium-ion batteries. This study aims to provide an accurate electrochemical thermal model for predicting the phenomenon of thermal runaway. In recent years, researchers have conducted many numerical models that generally estimate input parameters. Accurate modeling of thermal runaway is challenging due to the wide range of physical properties and conditions. Therefore, understanding the importance of uncertainty in input parameters, including physical thermal parameters and heat transfer, is essential. In this research, the Sobol index for five input parameters was calculated using a Gaussian process surrogate model through global sensitivity analysis. The results show that researchers can confidently estimate some thermal properties, but precise characterization of diffusivity and convective coefficients is needed for robust predictions. In the next phase of the research, the propagation of thermal runaway in a module consisting of 32 cells in various arrangements and different locations of thermal triggering (corner, middle, and edge) was examined. Evaluation criteria for assessing thermal runaway propagation to neighboring cells were described to identify the best cell arrangement that maximizes thermal runaway delay. The results indicate that a staggered column arrangement is not suitable due to the high likelihood of multiple thermal runaway initiations in the module's cells. The aligned arrangement is identified as the best cell arrangement. Also, the probability of severe thermal runaway propagation when faults occur at edge and corner positions is minimal under calm and turbulent currents, respectively. These findings can be used to design sensor positions within the battery pack
  9. Keywords:
  10. Gaussian process ; Lithium Ion Batteries ; Sobols Indexes ; Machine Learning ; Thermal Runaway ; Thermal Runaway Propagation ; Electric Motor Vehicles

 Digital Object List

 Bookmark

No TOC