Carbon-Aware Operational Control of Fuel Cell Test Benches Using Adaptive Model Predictive Control

Proton exchange membrane fuel cell (PEMFC) test benches play a critical role in stack characterization and control validation, yet they consume substantial electricity for high-pressure air compression, thermal management, humidification, and other auxiliary systems during long-duration tests. Conventional control strategies primarily prioritize the precise regulation of air supply temperature, humidity, and stack thermal states, but typically neglect the time-varying carbon intensity of grid electricity, thereby limiting opportunities for operational emissions reduction.

This work proposes a carbon-aware operational control strategy based on adaptive model predictive control (AMPC). Unlike traditional approaches, real-time carbon intensity is embedded into the control objective via a time-varying weighting factor. This mechanism enables an explicit trade-off between tracking performance and electricity-related carbon impact under physical and safety constraints. The proposed approach is implemented at the device level and does not rely on system-level energy scheduling, allowing integration into existing bench control architectures with minimal hardware changes.

MATLAB/Simulink simulations under representative operating scenarios with load variations and fluctuating carbon intensity signals indicate that the carbon-aware controller effectively reduces carbon-weighted operating cost relative to a fixed-weight MPC baseline. Specifically, the controller achieves this by moderating the parasitic power demand of the air supply system during high-carbon periods while maintaining acceptable tracking performance and constraint satisfaction.

The results demonstrate that incorporating carbon intensity awareness into bench-level control is feasible and effective, supporting low-carbon operation of PEMFC test benches for sustainable fuel cell testing and validation.