A computationally efficient thermal model of cylindrical battery cells for the estimation of radially distributed temperatures

This paper presents a computationally efficient thermal model of a cylindrical lithium ion battery for real-time applications. Such a model can be used for thermal management of the battery system in electrified vehicles. The thermal properties are modeled by volume averaged lumped values under the assumption of a homogeneous and isotropic volume. A polynomial approximation is then used to estimate the radial temperature distribution that arises from heat generation inside the cell during normal operation. Unlike previous control oriented models, which use discretization of the heat equation, this model formulation uses two states to represent the average value of temperature and its gradient. The model is parameterized using experimental data from a 2.3 Ah 26650 Lithium-Iron-Phosphate (LiFePO₄ or LFP) battery cell. Finally, a Kalman filter is applied based on the reduced order thermal model using measurements of current, voltage and surface temperature of the cell and ambient temperature. The effectiveness of the proposed approach is validated against core temperature measurements.