Thermal characterisation of lithium-ion batteries and optimised control strategies in electric vehicles

Lithium-ion batteries (LIBs) are widely used in electric vehicles (EVs) due to their high energy density and high charge/discharge rate. However, the high
charge/discharge rate also leads to significant heat generation under various conditions, leading to higher temperature rises and could even trigger thermal runaway of the LIBs under certain conditions. Therefore, battery thermal management systems (BTMS) have become particularly important in improving battery performance and ensuring the safe and reliable operation of EVs. In this thesis, a comprehensive study is carried out from the determination of the specific heat capacity of LIBs, through the modeling of the electrical battery behavior, to the optimization of preheating and charging strategies and preliminarily exploring the thermal safety of batteries in the event of coolant leakage

In determining the specific heat capacity of LIB, this thesis proposes a novel method that requires only regularly available equipment in the laboratory instead of using expensive equipment such as calorimeters. The specific heat capacity is determined under a forced convection environment by introducing a heat loss compensation model. With this thermal compensation model, the test results remain near a constant value (1044 J/(kg·°C)) regardless of the time change. To validate the method, an aluminum alloy block with the same shape as the cell is used. Results showed a specific heat capacity of 867 J/(kg·°C) with a 3.3% error compared to the reference value of 897 J/(kg·°C). In addition, to validate stability, 3 independent replicates of the same battery were performed. Experiments were also carried out on 3 different cells from the same production line. The maximum difference from the mean value is less than 2.5 % in all the experiments, indicating that this method has a good consistency. This novel methodology presents an effective way to determine the specific heat capacity at a lower cost and can be achieved quickly.

With the determined battery thermal parameters, the battery electric-thermal models are developed at the cell and module level. At the cell level, two different models with one considering the external heating method and another only considering natural convection are developed. At the module level, a lumped electrical-thermal model is developed for investigating complex hybrid cooling systems. The proposed model enables the cell-to-cell temperature variations analysis with the hybrid cooling thermal management system under the 1D level, which has considerable time savings compared to running a 3D numerical model simulation. Good agreement is observed between the experimental and simulation results. The maximum error for the voltage calculation is 1.64%. The maximum deviations between the modeled and experimental results for battery module temperatures are less than 5%. Specifically, the maximum deviation is 4.5% at the middle position of the battery module, and 1.7% at the end position. Compared with a similar 3D numerical model, this model sacrifices approximately 1% accuracy in temperature prediction but requires only 3.9% of the computational time of the CFD model. Therefore, the proposed model is computationally efficient and can be easily used to optimize algorithms to optimize temperature control during battery module operation.

Using the developed electro-thermal models, optimization algorithms are developed for both preheating and fast charging. In optimizing preheating, the
developed algorithm could dynamically adjust the external heater’s power and optimize heating time by incorporating battery heat generation to minimize
energy consumption while driving. The proposed algorithm incorporates both offline training and online operation. During the offline training, the developed
electro-thermal model is used to calculate the cost grid, which considers various initial conditions of state of charge (SoC), temperature, remaining travel time,
and heating power. Dynamic programming is then employed to determine the optimal heating power, generating a trained optimized track map that guides the heating power based on current conditions. During real-time operation, the battery management system (BMS) interpolates this map based on real-time conditions. Experiments demonstrate a 32.9% energy saving, reducing energy use from 2.65 kWh to 1.75 kWh when heating the battery from 10°C to 25°C. The robustness of the algorithm is confirmed through tests in different scenarios, including different driving times, speeds, and unexpected stops, results indicating that it not only preheats the battery on time but also saves energy effectively in each scenario.

In optimizing fast charging, an algorithm is developed and could be running on low-cost processors. Similarly to the optimization of preheating, this algorithm
integrates both offline training and online operation. During offline training, optimized maps of charging current are obtained based on the developed electrothermal models and dynamic programming. Different from optimized preheating, the optimized charging process divides the starting and terminating
SoC into several distinct sub-phases. In each sub-phase, varying the initial temperature and charging current is used to generate the cost grid under those
conditions. Then, similar to optimized preheating, dynamic programming is employed to generate the trained optimized track map that guides the charging current based on real-time conditions. Robust analysis indicates that the algorithm maintains good stability and performance across various conditions.


Meanwhile, this thesis includes a preliminary study on thermal safety in lithium-ion batteries, focusing on coolant leakage scenarios that could trigger external
short circuits. A modified lumped electric-thermal model evaluates the relationship between coolant conductivity, short-circuit resistance and battery temperature rise. Initial results highlight a non-linear correlation and provide insights into critical safety parameters. While experiments validate the model’s end temperatures, differences in temperature rise rates suggest areas for future research to refine the model and enhance thermal management strategies.

This thesis proposes and validates a novel and cost-effective method for determining the specific heat capacity of LIBs. The developed electro-thermal models at both cell and module levels provide computationally efficient tools for optimizing battery temperature control and performance. Finally, optimized algorithms for both pre-heating and fast charging demonstrate significant energy savings and robust performance in various real-time operating scenarios, while a preliminary thermal safety model explores the relationship between coolant conductivity and battery safety under external short-circuit conditions.