This paper proposes a degradation-aware integrated sizing and energy management strategy (ISEMS) for hybrid multi-stack fuel cell systems (MFCSs) in heavy-duty vehicles (HDVs), aiming to minimize total cost while enhancing system durability. While ISEMS approaches have been explored for single-stack systems, their extension to MFCSs remains unexplored. The proposed method simultaneously determines the size of the fuel cell (FC) stacks and the battery configuration, and evaluates the performance of the system under different FC stack arrangements. Unlike conventional approaches, sizing and stack allocation are handled jointly, without assuming a fixed total FC capacity. The degradation behavior of both the FC stacks and the battery is embedded in both sizing and energy management stages to improve system longevity. A hybrid genetic algorithm–model predictive control (GA–MPC) framework is employed, where GA explores design configurations and MPC ensures degradation-aware power allocation. The method is validated in a long-haul truck case study, where ISEMS reduces total cost by 68% and 50% compared to single-stack and quad-stack configurations from previous studies. Experimental validation is also conducted using data from FCs installed on a multi-stack test bench, in order to identify their optimal arrangement with the proposed method. Sensitivity analyses also demonstrate that widening the battery state of charge (SOC) window leads to additional cost reductions of up to 67.4% by enhancing battery utilization. The proposed framework offers a scalable, degradation-aware solution for cost-effective MFCS design and control in HDV applications. • An integrated framework for addressing sizing and fuel cell stack allocation problems in heavy-duty vehicles is proposed. • Battery size and FC arrangements are co-optimized, considering state of health for durability. • A sensitivity analysis is conducted to assess the influence of stack number on optimal sizing, investment, and operational costs. • Experimental validation at the multi-stack scale is provided to demonstrate the performance of the proposed approach.
Yamchi et al. (Tue,) studied this question.