Investigation of Energy Management Strategies for Fuel Cell-Hybrid Propulsion Systems in the Railway Sector

Nowadays, the railway sector in the European Union represents the most environment friendly transport system, being the majority of the trainlines electrified. However, electrification is not economically viable in mountain areas or for secondary lines with limited traffic. Thus, diesel trains are still in operation, producing polluting emissions. For these reasons, Fuel Cell-hybrid trains have been developed, representing a promising technology to completely decarbonize the railway sector. Fuel Cell (FC) exploit the electro-chemical reactions between hydrogen and oxygen to produce electricity with harmless emissions. Moreover, batteries allow energy storage, in addition to supply traction load, thereby improving fuel economy. Even though FC-hybrid systems allow to reduce fuel consumption and eliminate polluting emissions, the presence of two power sources arises a common challenge for hybrid vehicles, that is the power allocation among them. In fact, several methods for optimal power splitting are currently being researched and tested, with the main objectives of minimizing the fuel consumption and preserving components longevity. This thesis aims to analyse the effects of two Energy Management Strategies (EMSs) on fuel economy, battery State of Charge (SOC) tracking and the degradation of both FC and battery in the Fuel Cell-hybrid “Coradia Stream H” train by Alstom. More in detail, a widely industrially used method and a promising real-time implementable one are considered, namely the Deterministic Rule-Based (RB) strategy and the Model Predictive Control (MPC). The former is adapted into a Health-Aware Rule-based strategy by considering factors causing FC and battery degradation. The latter considers a cost function in which are present the costs due to fuel consumption and deterioration of the components. The results show that both the MPC and RB strategies improve fuel efficiency and reduce components degradation compared to the one based solely on SOC tracking. Moreover, the MPC outperforms the RB only when deterioration weights are increased, confirming the importance of the cost share associated with FC aging. Furthermore, the effects of prediction horizon and weights on the MPC are investigated. Finally, due to the high hybridization ratio, it is demonstrated that it is not possible to significantly influence battery degradation.