The open-circuit potential (OCP) of a li-ion cell is a function of both state-of-charge and temperature. Characterising its dependence on temperature is important for thermal simulations as the derivative of the OCP with temperature acts as a reversible heat source term in the model. Known as the entropy coefficient, this derivative is a thermodynamic quantity that varies over the cell’s state-of-charge (SoC). The potentiometric approach [1] is widely used to characterise the entropy coefficient for a given SoC. This involves applying step changes in the temperature and allowing the cell OCP to reach (or approach) a steady-state value, which can take several hours (>8 hours), for a given SoC of interest.Robust techniques are required that can significantly reduce the experimental time and quantify the entropy coefficient over the full SoC window, leading to higher fidelity battery thermal models. In this work, the dynamics between the OCP and cell temperature are exploited via system identification techniques to derive the entropy coefficient. This is achieved by estimating the underlying kernel function between the cell OCP and temperature dynamics. The kernel function is a non-parametric estimate of the dynamics from which the entropy coefficient can be determined.In this work temperature signals are designed with multiple frequency components and applied to a cell via Peltier elements. The corresponding OCP response is then analysed, in the frequency domain, to estimate the kernel function. Unlike the potentiometric method, which drives the battery to steady-state and necessitates long experimental durations, the kernel function can be estimated under transients, facilitating faster characterisation. The figure demonstrates comparable results of the entropy coefficient, over the full SoC interval, when estimated via the kernel function and the potentiometric approach. The approach, compared to the potentiometric method, however, brings around a two to three times reduction in experimental time per SoC and gives insight into the OCP and temperature dynamics. The signal design, kernel estimation and entropy coefficient estimation procedures are detailed in this talk. Schmidt, J. P., Weber, A., et al., Electrochimica Acta 2014, 137, 311-319. DOI 10.1016/j.electacta.2014.05.153 Figure 1
Read full abstract