Nonlinear observer designs for fuel cell power systems

A fuel cell is an electrochemical device that combines hydrogen and oxygen, with the aid of electro-catalysts, to produce electricity. A fuel cell consists of a negatively charged anode, a positively charged cathode and an electrolyte, which transports protons or ions. A low temperature fuel cell has an electrical potential of about 0.7 Volt when generating a current density of 300–500 mA/cm2. Practical fuel cell power systems will require a combination of several cells in series (a stack) to satisfy the voltage requirements of specific applications. Fuel cells are suitable for a potentially wide variety of applications, from stationary power generation in the range of hundreds of megawatts to portable electronics in the range of a couple of watts.; Efficient operation of a fuel cell system requires advanced feedback control designs. Reliable measurements from the system are necessary to implement such designs. However, most of the commercially available sensors do not operate properly in the reformate and humidified gas streams in fuel cell systems. Sensors working varying degrees of success are too big and costly, and sensors that are potentially low cost are not reliable or do not have the required life time [28]. Observer designs would eliminate sensor needs for measurements, and make feedback control implementable. Since the fuel cell system dynamics are highly nonlinear, observer design is not an easy task.; In this study we aim to develop nonlinear observer design methods applicable to fuel cell systems. In part I of the thesis we design an observer to estimate the hydrogen partial pressure in the anode channel. We treat inlet partial pressure as an unknown slowly varying parameter and develop an adaptive observer that employs a nonlinear voltage injection term. However in this design Fuel Processing System (FPS) dynamics are not modelled, and their effect on the anode dynamics are treated as plant uncertainty. In part II of the thesis we study the FPS dynamics, and estimate not only hydrogen but also all other species in its reactors.; We design nonlinear observers for the Catalytic Partial Oxidation (CPO), Water Gas Shift (WGS), and Preferential Oxidation (PROX), reactors in the FPS. The observers make use of temperature measurements (and possibly one more variable, such as pressure) to estimate the mole fractions of each species in the reactors. An advantage of these designs is that they are based on reaction invariants and do not rely on knowledge of reaction rate expressions. Finally, in part III, we illustrate how the designs of parts I and II can be incorporated in fault detection and estimation algorithms for common failures encountered in fuel cells, such as the cathode blower failure and the anode valve failure. For this task, we combine geometric tools with our observers.