The transition towards a more sustainable future has been more prevalent and necessary in recent years due to the increase of greenhouse gas emissions. With the increasing size of modern cities, major concern is risen in the transport sector, becoming one of the main contributors to greenhouse gas emissions. More specifically, road transport is the main contributor within this sector, causing as much as 70 % of the total emissions [3]. As a consequence, the adoption of Electric Vehicles (EVs) has become more prevalent in various countries as a replacement of Internal Combustion Energy Vehicles (ICEVs), being the main cause of pollution within road transport.
On the other hand, EVs are yet to be the perfect replacement due to 2 main bottlenecks, significantly slower charging times compared to refilling Internal Combustion Energy vehicle (ICEV), as well as Li-ion battery lifespan and deterioration over time, which is not an issue in ICEVs, thus a charging profile needs to be determined to mitigate these factors as much as possible. This brings the research proposal "Finding a model-based real-time control charging strategy that mitigates charging times as well as degradation".
To explore this research topic before developing a model-based control charging strategy, a battery model as well as a degradation model are needed to be implemented, to capture physical states of the battery such as State Of Charge (SOC) and of degrdation mechanisms such as Solid Electrolyte Interphase (SEI) and lithium plating. First of all, 3 electro-chemical Li-ion battery dynamical models have been implemented and/or simulated, the Pseudo 2-Dimensional (P2D) model, the Electrolyte Enhanced Single Particle Model (SPMe) and the Extended Single Particle Model (ESPM). The P2D model is considered a full order model, highly accurate for a wide range of charge/discharge currents, which is taken as the base for comparison. The SPMe is a simplified, less accurate model of the P2D, but faster in terms of computational time. Both models are computationally heavy, inapplicable for real time control applications, thus being infeasible for realtime control. The ESPM, a simplification of SPMe, has been developed to tackle the computational time issue, while still maintaining high accuracy performance. The P2D and SPMe are already available and simulation data was obtained using the Python library PyBamm, whereas ESPM was unavailable, thus implemented and simulated on MATLAB. Implementations of comparison between models on MATLAB showed that ESPM can be used for a current range of at least 2C, showing above 90 % similarity in all model variables compared. In order to include degradation effects in the control strategy, a detailed electro-chemical degradation model, incorporating the 2 main aging effects, SEI growth and lithium plating has been implemented and simulated on MATLAB where degradation parameters were adjusted to fit the capacity fade specifications for the LGM50 battery. Such degrdation effects were incorporated such that they could be minimized seperately in the charging strategy.
Lastly, based on the models developed, an Nonlinear Model Predictive Control (NMPC) control strategy has been developed with the aim to achieve a tradeoff between charging time, SEI growth and lithium plating as well as change of charge current with time while staying between the feasible model and cell constraints. A Health Conscious Fast charging strategy achieved a charging time of 34 minutes, while having a battery lifespan of approximately 800 charge/discharge cycles. This charging time matches typical DC Fast charging times, while saving approximately 100 cycles from the LGM50 battery lifespan