Mechanical Characterization of Li-Ion Cells and the Calibration of Numerical Models Using Proper Generalized Decomposition

Abstract

Abstract The behavior of lithium-ion cells under mechanical load is determined by both experiments and simulations. Extensive experimental data are required for the calibration and validation of complex numerical models. The possible experiments may differ in dimension (component, cell, cell stack and module) and in boundary conditions during the experiments (e.g. loading rate, sample condition). This study covers two aspects: The characterization and the modeling: First an overview on innovative experimental procedures is given. Second, it is shown how to quickly calibrate a macroscopic cell model against the experimental data. The calibration effort increases strongly with the number of calibration parameters as with the number of test configurations against which the model is calibrated. A new calibration procedure is presented to overcome this hurdle: First, a meta-model is created, covering the calibration parameters and calibration load cases, employing Proper Generalized Decomposition. Once the meta-model is adequately trained, the optimization of calibration parameters is conducted in real-time employing the meta-model. So a macroscopic cell model can be calibrated very quickly against a given set of experimental data sets. This is important because there are constant improvements and changes to the batteries in terms of materials, dimensions and structure.