Fuel Cell Longevity in Motion: Understanding and Controlling Degradation in Dynamic Environments

This NSF project aims to investigate modeling and control frameworks that can improve the durability and longevity of proton electrolyte membrane fuel cells. The project will bring transformative change to fuel cell control that can improve their longevity when used for transient power production in aircraft and heavy-duty transportation applications. This change will be achieved by examining novel methodologies that leverage an understanding of the underlying physical and chemical interactions along with artificial intelligence tools. The intellectual merit of the project includes the creation of models of fuel cell degradation that pinpoint the underlying source of degradation and control techniques that help avoid these degradation events. The broader impacts of the project include graduate and undergraduate training that prepares the future transportation workforce and outreach to a local high school with strategic activities designed to improve engagement and enrollment in their engineering and robotics courses. Hydrogen fuel cells could provide an efficient source of energy for transportation sectors that are not well suited to batteries. Those sectors include heavy-duty trucking, marine, rail, and aviation. However, their durability suffers under the changing power demands, temperature swings, and moisture levels of real-world use in these applications and these conditions are not as extensively studied in laboratory experiments. This project develops a real-time method for estimating and mitigating degradation in proton electrolyte membrane fuel cells. Unlike current control-oriented models that track only overall lifespan, this work builds a hybrid model that separately captures four degradation mechanisms—catalyst degradation, membrane thinning, gas diffusion layer failure, and thermal cycling—by pairing physics-based models with machine learning that learns nonlinear residual dynamics and corrects model predictions based on experimental and operational data. Estimates from this model framework are then embedded in a closed-loop control system that adjusts operation. This approach aims to extend fuel cell durability in trucks, aviation, unmanned aerial vehicles, and hybrid power systems, while offering a generalizable strategy for combining physics-based and machine learning models in other energy systems. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.