Description
The increased use of lithium-ion batteries in future electric vehicles presents new challenges to systems security, particularly in terms of high energy densities and increased performance demands. A particularly safety-relevant scenario is thermal runaway of individual battery cells. In this process, a highly energetic gas-particle stream is released, characterised by strong transient thermal, mechanical, and abrasive loads. These stresses can cause significant damage to adjacent components and lead to structural failure.
The goal of this work is to develop a predictive method for evaluating material performance under thermal runaway conditions due to escaping gas and particles at materials used in venting structures. To achieve this, a neural network will be designed, trained, and applied to new materials. The neural network will be trained and validated using existing experimental and simulation data. The generated evaluation data will be compared with results from classical substitute and cell tests to evaluate the performance and reliability of the developed approach. Based on the obtained data, characteristic parameters for evaluating material failure under thermal runaway conditions will be identified and derived.
In the first step, a systematic literature review will be conducted on existing experimental, analytical, and simulation methods for evaluating fire protection materials in the thermal runaway context. Based on this, a suitable model for evaluating material performance in the context of thermal runaway will be developed, trained, and implemented. The model will be validated using experimental data to assess its predictive accuracy and robustness. Finally, the applicability of the developed approach and potential opportunities for further development will be critically discussed.
Key tasks:
- Conduct a systematic literature review on existing experimental, analytical, and simulation methods for evaluating fire protection materials in the thermal runaway context.
- Design, train, and implement a neural network for predictive evaluation of material performance under thermal runaway conditions.
- Evaluate and analyse experimental and simulation data to validate the model.
- Compare generated evaluation data with results from classical substitute and cell tests.
- Identify and derive characteristic parameters for evaluating material failure under thermal runaway conditions.
- Critically discuss the applicability of the developed approach and potential opportunities for further development.
Requirements:
- Bachelor's or master's degree in computer science, mechanical engineering, electrical engineering, or a related field.
- Experience in machine learning and deep learning.
- Familiarity with Python programming language and relevant libraries.
- Good understanding of thermal runaway phenomena and fire protection materials.
- Excellent communication and teamwork skills.
- Ability to work independently and manage multiple tasks.
Preferred skills:
- Experience with neural networks and deep learning frameworks such as TensorFlow or PyTorch.
- Familiarity with simulation software such as ANSYS or COMSOL.
- Knowledge of thermal analysis and heat transfer.
- Experience with data analysis and visualisation tools such as Matplotlib or Seaborn.
- Familiarity with version control systems such as Git.