Date of Award
Doctor of Philosophy (PhD)
Dr. Hussain Shareef
Dr. Mohammad Shakeel
Dr. Yaser E. Greish
Proton exchange membrane fuel cell (PEMFC), as a source of electrical power, provides numerous benefits such as zero carbon emission and high reliability as compared to wind and solar energy. PEMFC operates at very low temperature, high power density, and has very high durability as compared to other fuel cells. Being a non-linear power source with high sensitivity to ambient conditions variation, the prediction of PEMFC voltage and temperature is a complicated issue. The most common PEMFC models are classified as mechanistic models, semi-empirical models, and purely empirical methods. The mechanistic models are complex and require differential equations to predict the voltage and temperature of PEMFC. However, the semi-empirical models are less complicated and can be used easily for the online prediction of PEMFC outputs. Therefore, the first part of this thesis attempt to model the voltage of PEMFC using simple and effective semi-empirical equations. The initial feature of the proposed technique is to incorporate the features of a mechanistic model with less complex equations. The model considers the internal currents and the internal voltage drop associated with the PEMFC. Besides, activation and concentration voltage drops are addressed based on theoretical functions. Thus, the proposed model provides an additional benefit that not only does the output voltage model satisfy the voltage for both loaded and unloaded conditions but also the component voltage drops waveforms match with the theoretical waveforms given in the mechanistic models. The second part of the thesis focuses on modelling the PEMFC temperature. Previously most temperature models use complex equations incorporating PEMFC output voltage which is not a good option as the temperature must be predicted using only load current and ambient temperature.
The model proposed in this thesis is developed through an algorithm that tracks the online changes in the load current and ambient temperature. It provides the accurate temperature of PEMFC by using a simple first-order equation with the help of a tracking algorithm. Quantum lig tening search algorithm (QLSA) is used for the optimization of constant parameters for both voltage and temperature models. The PEMFC performance is affected by factors such as variations in ambient temperature, pressure, and air relative humidity and thus they are vital for predicting PEMFC performance. The thesis also attempts to directly predict the variations in PEMFC voltage under varying ambient conditions at different load resistance. For this purpose, statistical analysis is used to propose empirical equations that can predict the variations in PEMFC voltage for varying ambient conditions. In this context of the model development, the parameters which are significantly varying with ambient changes are identified with the help of statistical regression analysis and represented as ambient temperature and air relative humidity dependent parameters. The enhanced semi-empirical voltage model is verified by performing experiments on both the Horizon and NEXA PEMFC systems under different conditions of ambient temperature and relative humidity with root mean square error (RMSE) less than 0.5. Results obtained using the enhanced model are found to closely approximate those obtained using PEMFCs under various operating conditions, and in both cases, the PEMFC voltage is observed to vary with changes in the ambient and load conditions. Inherent advantages of the proposed PEMFC model include its ability to determine membrane-water content and water pressure inside PEMFCs. The membrane-water content provides clear indications regarding the occurrence of drying and flooding faults. For normal conditions, this membrane water content ranges between 12.5 to 6.5 for the Horizon PEMFC system. Based on simulation results, a threshold membrane water- content level is suggested as a possible indicator of fault occurrence under extreme ambient conditions. Limits of the said threshold are observed to be useful for fault diagnosis within the PEMFC systems.
My thanks go to my advisor Dr. Hussain Shareef whose enthusiasm about and introduction to research methodology got me started and introduced me to the exciting field of the fuel cell. His endless ideas and encouragement led to this and most other studies in which I have been involved. I would like to thank my committee for their guidance and support, especially my advisor Dr. Hussain Shareef for his assistance throughout my preparation of this dissertation. My special thanks are extended to Dr. Chafik Bouhaddiou for helping me in statistical analysis. Special thanks go to my parents, wife, brothers, and sisters who supported me along the way. I am sure they suspected it was endless. In addition, special thanks are extended to my friend Dr. Haroon Usman and Rizwanulhaq Faraz for their assistance and friendship.
Khan, Saad Saleem, "MODELLING AND FAULT DIAGNOSIS APPROACH FOR PROTON EXCHANGE MEMBRANE FUEL CELL SYSTEMS INCORPORATING AMBIENT CONDITIONS" (2019). Dissertations. 91.