Fault diagnosis in polymer electrolyte membrane fuel cells based on patterns of tolerance

Several studies have been published considering fuel cell (FC) operation in normal conditions; but few of them addressed their operation under fault conditions. Among fault detection methods, a simple method based on the variable behaviour analysis is introduced. Basically, a fault supervisor is responsible for analysing unexpected changes in some variables, since each change in the variable behaviour can be attributed to a pre-classified fault cause, and then a fault diagnosis can be performed. The diagnosis considers several types of faults in polymer electrolyte membrane FC: faults in the reaction air feeding, rupture of the membrane-electrode assembly, low pressure in the hydrogen fuel line, and fault refrigeration system; those faults are detected based on the monitoring of a limited number of variables. Tests show that the methodology is reliable, flexible (once it permits the addition of other fault detection modules) and easy for implementation.