Challenge 3: JLR

JLR: Predicting Electrical Resistance in Battery Cells via Machine Learning

JLR (Jaguar Land Rover) is a modern luxury automotive company built around its four distinct British brands: Range Rover, Defender, Discovery and Jaguar. JLR is actively reimaging its future through cutting-edge design, advanced technologies and exceptional customer care. As part of its transformation, JLR is pursuing electrification, offering electric vehicles, and targeting net-zero carbon emissions.

The battery pack and cells are the essential parts in an electric vehicle. The lifetime of a battery pack and cell is strongly influenced by how its voltage and temperature are managed, both of which depend on the cell’s internal resistance. In order to use the battery in the most efficient way, we need to understand how the material chemistry and service conditions impact the resistance. Traditionally, one would employ equivalent circuit or electrochemical models to better understand the electrical behaviour of the battery cell. However, the phenomenon is quite complex as it entails a number of phenomena that contribute to the voltage drop and it is governed by the respective timescales (instantaneous voltage drop due to pure Ohmic resistance; subsequent drop due to double layer capacitance and charge transfer resistance; drop due to polarisation resistance). Hence, a combination of ‘physics’-based models and machine learning approaches is needed to capture the phenomenon.

This challenge will utilise a vast training data set from open literature including both experimental and synthetic data to aim to:

• Decouple the battery resistance components and analyse how they relate to the testing and modelling methods;
• propose machine learning models to link the battery resistance with battery cell design;
• apply the selected machine learning model to predict the battery resistance of a pouch cell with NMC cathode and graphite-silicon anode at different material composition.

Approaches

The participants will be encouraged to consider neural network approaches and phenomenological battery cell models. JLR will provide research context into the topic and guidance on the specific material composition of interest along with experimental data.

Extra background

Lithium-ion battery (LIB) cells are the foundational energy-storage technology powering modern electrified vehicles. They operate by shuttling lithium ions between the anode and cathode through an electrolyte, enabling high energy density, strong cycle life, and

efficient charge-discharge performance. Their modular cell-to-pack design allows engineers to optimise performance, safety, and thermal management, making lithium-ion chemistry the current industry standard for delivering the range, power, and reliability required in next-generation electric vehicles.

Battery resistance is one of the key attributes that characterise the performance of battery cells. Resistance measurement methods, such as Direct Current Internal Resistance (DCIR) and Electrochemical Impedance Spectroscopy (EIS), play a crucial role in understanding the internal behaviour and health of LIB. Resistance modelling approaches for LIB provide deeper insight into how internal resistance evolves under different operating conditions and ageing mechanisms. Equivalent circuit model (ECM), electrochemical model, machine-learning model and hybrid model are among the most useful ones. Together, these methods offer complementary perspectives on cell performance, ageing, and material-level processes.