The heavy-duty transportation sector is a major contributor to greenhouse gas emissions, highlighting the urgent need for zero-emission solutions. This research develops a multilevel control architecture that optimizes fuel economy and minimizes emissions in fuel cell hybrid heavy-duty vehicles, equipped with proton exchange membrane fuel cell and battery pack as main power sources. The detailed fuel cell system model incorporates reactants and thermal dynamics, including air supply, hydrogen flow, water management and their effects on reaction kinetics, membrane conductivity, water balance, performance and durability. The low-level control strategy is designed using a physics-based approach that accounts for critical constraints, including temperature, membrane water content and differential pressure between the cathode and anode. By identifying optimal setpoints for key control variables, this methodology enables the development of accurate control maps for actuator management, ensuring maximum efficiency and improved system lifetime. In the high-level layer, the strategy handles energy flow by determining the optimal power split between the fuel cell and battery. The integration of these two layers within a multilevel architecture enables a comprehensive optimization process considering both fuel cell dynamics and overall vehicle performance, while also mitigating degradation phenomena through the implementation of current rate limiters and predefined idling power levels. To evaluate performance, a benchmark based on dynamic programming is employed. In a real-world driving cycle and with a final state-of-charge target of 57%, the rule-based controller achieves a fuel economy within 1.6% of the globally optimal solution, demonstrating its robustness and near-optimal efficiency. Furthermore, the introduction of an efficiency-driven model is a core novelty that addresses the computational burden of high-fidelity simulations. This model accurately reproduces the system’s electrical behavior under steady-state and moderate transient conditions, making it a fast and reliable tool for control-oriented studies and large-scale simulations.
Bove et al. (Sun,) studied this question.