By Azita Soleymani, PhD
The thermal management of li-ion battery packs are quite crucial as it is directly related to the safety, reliability, performance and durability of battery packs. The thermal requirement of battery packs is quite specific. Not only the average battery pack temperature should be maintained within the given range but also the spatial gradient of temperature (at different directions) and the direction of heat transfer are key factors of consideration in order to deliver a robust and reliable thermal solution. The target range of temperatures for cells and battery pack depends on the charge/discharge modes and also on manufacturing of battery cells.The thermal design of battery packs should be evaluated in transient fashion at wide range of conditions to ensure all the thermal requirements are met. The design of experiment (DOE) includes conditions such as fast charging, cold start, charging at low temperature, discharging at low state of the charge and different drive cycles.
Numerical simulation is known as a powerful tool to evaluate and optimize a thermal design solution at the early stage of design and development. Later, it is used to conduct trouble-shooting. However, developing a reliable simulation tool to study temporal-spatial temperature profile of battery packs remains a challenge. Taking into account the large number of cases to be considered for a design validation, particularly at earlier development stages, conventional computational fluid dynamics (CFD) approaches are not practical. Therefore, majority of numerical analyses are limited to either transient 1D-system level type of simulations or steady-state three-dimensional CFD analyses (Fig.1). Either of the approaches is not sufficient and could lead to significant design problems.
|Fig. 1. 3D steady-state numerical simulations are used to estimate (a) contributions of each element (cells, wire-bonds, ribbon-bonds, etc.) on total heat generation of battery pack (b) spatial temperature profile.|
To overcome the limitations, a Digital Twin model of li-ion battery packs (operated in real-time) can be developed allowing evaluation of all possible inputs and operating conditions. The construction of a Digital Twin requires the generation of response curves, in this case real time temperatures as functions of SOC, pack temperature, coolant flow rate, etc. The response curves can be generated utilizing simulation data or measured test data. The Digital Twin model of battery pack thermal solution may be generated as follow.
1. Characterize battery cells
Considering the highly convoluted multi-physics nature of li-ion battery cells, the thermal load of each battery cell at one instant in time depends on the cell type (manufacturing parameters), State of the Charge (SOC), cell temperature, charge/discharge mode, magnitude of electric current extracted from, and aging. Typically, each cell is represented by a 2RC model (Fig. 2) with one resistance and voltage source in series. Hybrid Pulse Power Characterization (HPPC) tests are conducted to characterize and estimate the cell parameters.
|Fig. 2. The 2RC model representing the highly convoluted multi physics nature of li-ion battery cells|
2. Create equivalent circuit model (ECM) of battery cells
The estimated cell parameters obtained from HPPC test data are used to create an equivalent circuit model of battery cells. The model is then applied to estimate real-time heat generation of battery cells. Fig. 3 demonstrates the rate of heat generation for a given cell at four different cell temperatures.
|Fig. 3. The typical heat generated data from ECM of li-ion cells. Colors indicate the impact of operating temperature on the rate of heat generation.|
3. Create response curves
An Artificial Neural Network (ANN) or a Reduced Order Model (ROM) can be used to generate response curves either from detailed CFD tools or test data
4. Develop system level model of battery pack thermal solution
The final steps are to couple the ECM to the response curves and to build the system level representation of the battery pack. The resulting model is capable of capturing the transient highly-convoluted multi-physics behavior of a battery cell in a cost-efficient way.
Further live-sensor data can be integrated to the Digital Twin system level model of the battery pack to create a real-time environment. The generated tool can be utilized to monitor the temperature of a battery pack remotely and to have a predictive maintenance solution. Considering all the limitations on the installation of temperature sensors in battery packs (location, quantity, reliability), the above-mentioned feature of the developed tool can result in precise monitoring of the pack temperature in a real-operating environment.
To conclude, the Digital Twin model of battery packs can be developed to perform what-if scenarios, and to conduct in-depth root cause analyses, to further optimize the cooling system, to estimate life-time and to optimize operating parameters for thermal management.