An empirical study for the correction of the thermal parameters of a Li-Po battery with carbon-graphene-silicone coatings

The large use of batteries as energy storage systems for stationary applications and mobility has increased the concern about some limitations and issues, particularly in thermal management. This work presents the effect of coatings for battery heat dissipation with different conductivity, varying from 0.67 to 6 W/mK, and thicknesses, from 0.5 to 1 mm. The investigation is performed during the discharge phase with two current ratings, 0.5C and 1C, that correspond to 1.65 A and 3.3 A, respectively. Measurements of 2D temperature distribution are performed with thermal imaging while local data are collected with a thermocouple. The influence of the battery coatings is analyzed in terms of the temperature gradient, highlighting a lower increment in surface temperature for coatings with high conductivity. Then, the correction of the thermal parameters as the convective heat transfer coefficient and the specific heat capacity is considered for a reliable temperature prediction. In particular, the evaluation of these unknown parameters is performed simultaneously and in non-stationary conditions using two theoretical tools. The convective heat transfer coefficient must be changed from 14 to 17 W/mK and the specific heat capacity from 2900 to 3500 J/kgK to model the battery behavior with the coatings. The methods applied for this evaluation are the finite-difference method and Newton's cooling law. A reverse approach is applied, using the experimental data to find the thermal parameters, only one method suits for this application. The predictions of the finite difference method have a lower standard deviation than the Newton's law model, 10% versus 50%. Moreover, it is more robust for the test cases with high noise-to-signal ratio and it is sensitive to the coating material and thickness. The presented results highlight that the method that best fits these applications requires characteristics such as good reliability, higher flexibility, and low sensitivity to experimental data fluctuations. The parameters estimated with the finite-difference method are also validated in real driving conditions, representative of a homologation test cycle. In this case, the battery temperature is still well predicted, showing that underestimating the model temperature of only 1.5% produces the best fit for the experimental data.