Dynamic modeling and analysis of proton exchange membrane fuel cells for control design

This dissertation seeks to address a number of issues facing the advancement of Proton Exchange Membrane (PEM) fuel cell technology by improving control-oriented modeling strategies for these systems. Real-time control is a major ongoing challenge for PEM fuel cell technologies, particularly with regards to water and temperature dynamics. This can lead to a number of operational concerns, such as membrane flooding and dehydration, which can seriously diminish the efficiency, reliability, and long term health of the system. To combat these issues, comprehensive models that are capable of capturing the dynamics of the key operating conditions and can be processed in real time are needed. Also, given the inherently distributed nature of the system, such a model would ideally account for the changes in the conditions from cell-to-cell in the stack, which can be very significant. With this goal in mind, the main focus of this dissertation is the development and experimental validation of control-oriented modeling techniques for PEM fuel cell stacks. The first major work in this study was the verification of a relative humidity model in response to varying loads. Through this work, a multiple control volume (CV) approach was developed and experimentally validated to model the distribution of operating conditions more accurately while keeping the computational expense sufficiently low. To optimize the modeling efforts, further analysis of the temperature and vapor distribution was performed starting from first principles. This led to the creation of various techniques to optimally size CVs based on the parameters and operating conditions of the system in question. Finally, it was noted throughout the testing that the performance of the membrane electrolyte assemblies in the test stack declined significantly from their initial state. To compensate for this, a Kalman filter was implemented to quantify the membrane degradation. SEM analysis of membranes from the test stack confirmed the validity of this technique. This work can be used to significantly improve real-time models for PEM fuel cells for model-based control applications.