|Ph.D Student||Joshua Golbert|
|Subject||Model-Based Control of Fuel Cells|
|Department||Department of Chemical Engineering||Supervisor||Full Professor Lewin Daniel|
Fuel cells are chemical engines that convert chemical potential into electrical power. Since they are not based on temperature differences, they are not subjected to Carnot’s limit of efficiency. In addition, common pollutants such as sulfur dioxide and nitrous oxides are avoided since the process does not involve combustion. These advantages, together with the reduction of greenhouse gases and fuel consumption due to higher efficiencies and the possibility of alternative energy sources, have generated enormous interest in fuel cells for stationary as well as mobile applications.
This seminar describes a model-based controller of a proton exchange membrane (PEM) fuel cell, and focuses on the crucial issue of how to control the fuel cell to ensure stability and acceptable response time for the power demand over the entire operating range. The model accounts for spatial dependencies of voltage, current, material flows, and temperatures in the fuel channel. Analysis of the process model shows that the effective gain of the process undergoes a sign change in the normal operating range of the fuel cell, indicating that a linear controller with integral action cannot stabilize a fuel cell. Consequently, a nonlinear model-predictive-controller based on a simplified model has been developed, enabling the use of optimal control to satisfy power demands robustly. The models and controller have been realized in the MATLAB and SIMULINK environment. Results demonstrate improved performance and robustness when using model-based control in comparison with that obtained using an adaptive controller.
In addition, this controller has been used to improve overall efficiency of a fuel cell operating under dynamic conditions. Using multiple control variables the controller takes advantage of all the degrees of freedom to simultaneously satisfy power demands while optimizing the fuel efficiency of the entire system. Different methods of defining the non-linear program (NLP) and their influence on the closed-loop performance will be discussed. Of special interest is an definition of the NLP, inspired by optimal control, that enables offset-free response of the system, while still improving system efficiency.