Fuel Cell Diagnostics using Electrochemical Impedance Spectroscopy

Date created: 
2014-09-02
Identifier: 
etd8628
Keywords: 
PEM fuel cell
Hydrogen Crossover
Oxygen concentration
EIS
Neural network
Fuzzy logic
Abstract: 

When a proton exchange membrane (PEM) fuel cell runs short of hydrogen, it suffers from a reverse potential fault. This fault, driven by neighboring cells, can lead to anode catalyst degradation and, through cell reversal, to holes in the membrane due to local heat generation. As a result, hydrogen leaks through the electrically-shorted membrane-electrode assembly (MEA) without being reacted, and it recombines directly with air. This recombination results in a reduction in oxygen concentration on the cathode side of the MEA and a fuel cell voltage reduction. Such voltage reduction can be detected by using electrochemical impedance spectroscopy (EIS). In this research, in order to fully understand the effect of this oxygen reduction fault, the impedances of single and multi-cell stacks at different leak rates were measured. Then the impedance signatures were compared with the signatures of stacks having non-leaky cells at different oxygen concentrations with the same current densities. The signatures were analyzed by fitting the leaky stacks and oxygen concentrations impedance data sets with the parameters of a Randles circuit. The correlation between the parameters of the two data sets allows us to understand the change in impedance signatures with respect to a reduction of oxygen in the cathode side. Using the circuit parameters, a model that establishes a relationship between impedance and voltage was also considered. With the help of this model along with the impedance signatures, we are able to detect the reduction of oxygen concentrations at the cathode by using fuzzy logic (FL). However, resolution of detection was reduced with the reduction of leak rate and/or increases in the stack cell-count. The amount of hydrogen leak rates were quantified by simulating the resulting reduced amount of oxygen with the use of neural network (NN) method. Successful implementation of FL and NN methods in a fuel cell system can result in an on-board diagnostics system that can be used to detect and possibly prevent cell reversal failures, and to permit understanding the status of crossover or transfer leaks versus time in operation. Using such system will increase the reliability and performance of fuel cell stacks, where leaks can be detected online and appropriate mitigation criteria can be applied.

Document type: 
Thesis
Rights: 
Copyright remains with the author. The author granted permission for the file to be printed and for the text to be copied and pasted.
File(s): 
Supervisor(s): 
Farid Golnaraghi
Department: 
Applied Sciences:
Thesis type: 
(Thesis) Ph.D.
Statistics: