Modeling and control of fuel cell systems and fuel processors.

Fuel cell systems offer clean and efficient energy production and are currently under intensive development by several manufacturers for both stationary and mobile applications. The viability, efficiency, and robustness of this technology depend on understanding, predicting, and controlling the unique transient behavior of the fuel cell system. In this thesis, we employ phenomenological modeling and multivariable control techniques to provide fast and consistent system dynamic behavior. Moreover, a framework for analyzing and evaluating different control architectures and sensor sets is provided. Two fuel cell related control problems are investigated in this study, namely, the control of the cathode oxygen supply for a high-pressure direct hydrogen Fuel Cell System (FCS) and control of the anode hydrogen supply from a natural gas Fuel Processor System (FPS). System dynamic analysis and control design is carried out using model-based linear control approaches. A system level dynamic model suitable for each control problem is developed from physics-based component models. The transient behavior captured in the model includes flow characteristics, inertia dynamics, lumped-volume manifold filling dynamics, time evolving spatially-homogeneous reactant pressure or mole fraction, membrane humidity, and the Catalytic Partial Oxidation (CPOX) reactor temperature. The goal of the FCS control problem is to effectively regulate the oxygen concentration in the cathode by quickly and accurately replenishing oxygen depleted during power generation. The features and limitations of different control configurations and the effect of various measurement on the control performance are examined. For example, an observability analysis suggests using the stack voltage measurement as feedback to the observer-based controller to improve the closed loop performance. The objective of the FPS control system is to regulate both the CPOX temperature and anode hydrogen concentration. Linear multivariable system analysis is used to identify the limitation of a decentralized controller and to design a model-based multivariable controller with significantly improved performance in CPOX temperature regulation. Further analysis unveils the critical controller cross-coupling term that contributes to the superior performance of the multivariable controller.