Development, Experimental Validation, and Uncertainty Quantification Analysis of a Multiphysics Digital Twin for Predicting Thermal Behavior in Automotive Lithium Batteries
Please login to view abstract download link
The lithium-ion battery is essential for advancing the next generation of electric transportation. A key factor in maximizing the efficiency of these batteries is optimizing their thermal management system. To achieve this, we have developed a numerical digital twin of the battery cell to study its thermal behavior throughout its lifecycle. This digital twin uses Computational Fluid Dynamics (CFD) simulations to evaluate the external cooling system, incorporating a model for internal heat generation from [1] and simulating Conjugate Heat Transfer (CHT) from the cell’s internal structure to its surface. The digital twin was validated against wind tunnel tests, both with and without wind. The experimental setup includes a battery cycler with a 60 V–250 A Ametek SPS60x250-K02D charger and a 60 V–500 A Zentro-Elektrik EL6000 discharger, remotely controlled through LabVIEW software. Voltage and current are measured by a National Instruments DAQ 9219, which has a 100 ms acquisition rate. At the same time, the temperature distribution is monitored using 30 K-type probes with a maximum uncertainty of 0.3 K, as described in [2]. The digital twin accurately replicates the temperature gradients within the battery, with differences from experimental measurements typically within a few percentage points. Additionally, a stochastic sensitivity analysis was performed to assess the confidence in the digital twin’s outputs in response to small variations in the model parameters. This analysis was conducted using the generalized polynomial chaos method to construct continuous response surfaces in the parameter space. This validated digital twin model will be used to optimize the thermal management systems for lithium-ion batteries in automotive applications, enhancing the efficiency of electric vehicle battery packs.