Modelling of secondary batteries for state of charge evaluation

Energy storage is one of the biggest challenges the energy field is addressing. It is meant to represent the key solution for large penetration of renewables in the grid and for the decarbonisation of the transport sector. Among the different storage technologies, Lithiumion batteries have become the most reliable, competitive and efficient one for applications like electronic devices, electric vehicles or stationary electricity storage for the residential sector, and several large-scale battery plants are being installed for power system support. This thesis is focused on the modelling and simulation of Li-ion batteries for the state-ofcharge evaluation. Battery models allow to simulate their behaviour under different conditions and to predict their performance without the necessity of a real application. The state-ofcharge evaluation is a crucial feature of Battery Management Systems (BMS), as they permit the efficient and safe usage of batteries and they provide useful information about its current state. In this research, a deep literature review of secondary batteries, including their fundamentals, different chemistries and characteristics is presented, with a special focus on Li-ion batteries. After that, battery models and State-of-Charge (SoC) estimation techniques are also studied and compared. The case study of the thesis consists in the experimental testing of a maximum voltage of 4.2 V Li-ion battery, and its comparison with different simulated battery models. In order to perform the battery test, an Arduino platform is designed to collect the voltage and current measurements. The results of the test provide the charging and discharging voltage profile of the battery, as well as the state-of-charge profile as a function of the open circuit voltage. The energy efficiency of the battery is evaluated in 89.836 %. With the experimental test results, three different battery models are presented and implemented in Matlab-Simulink: the simple model, the zero-hysteresis model and the combined model. The off-line SoC estimation technique of Least Squares Estimation (LSE) has been used in all the cases to obtain the model parameters needed to run the simulations. The simple model simulation satisfactorily reproduces the battery behaviour. The energy efficiency calculated by this model is slightly higher than the real one: 91.279 % and the RootMean-Square Error (RMSE) between the test voltage and the model voltage is 0.215 V. The zero-hysteresis model simulation provides similar results than the simple model, with an energy efficiency of 91.278 % and an RMSE of 0.228 V. The combined model results to be the most accurate with respect to the test. However, this model is not able to reproduce the battery behaviour for a very low state-of-charge. The energy efficiency of the battery from this model is 89.989 % and its RMSE is 0.1879 V