Degradation-aware optimal control of grid-connected lithium-ion batteries

With the accelerating deployment of intermittent renewable energy sources, the need for balancing on the power system is ever increasing. Even though lithium-ion (Li-ion) batteries are not yet cost-competitive with other solutions for most applications, the number of grid-connected Li-ion batteries is expected to increase sharply in the coming years. Due to the sunk costs of the upfront investment, understanding the lifetime of a battery is crucial for the business case. Nevertheless, predicting the lifetime of a battery is not straightforward. Many degradation mechanisms reduce the total energy which can be stored in a battery, thus decreasing its lifetime. This thesis begins with an extensive literature review into the causes of this battery degradation, both qualitatively and quantitatively. Various physics-based degradation models are combined into one flexible battery degradation model which is used to study which mechanisms cause degradation under which operating conditions. The model is then carefully formulated for optimisation in a power system application. While a battery makes revenue from trading power on the day-ahead market, it degrades. The physics-based model is used to optimise the usage profile of the battery, determining when to buy power (charging the battery) and when to sell (discharging the battery), in order to maximise revenue while minimising degradation. These optimal profiles are compared to profiles which are obtained using state-of-the-art economic optimisation models. Finally, Li-ion batteries were cycled with different usage profiles in an experimental validation. The battery degradation was measured monthly, while the revenue each battery could make was also tracked. It is shown that the usage profiles determined by the physics-based model outperform the profiles from conventional optimisation models. The revenue can be increased by 17% while at the same time decreasing the degradation by 31%. This implies that such physics-based models can increase the revenue and the lifetime of grid-connected batteries, increasing the net profit by over 300%.