(314b) Sensor and Online Diagnostic Needs in Automotive Fuel Cells

Considerable research and development efforts have been devoted to fuel cell operation and systems. In particular, in the area of polymer electrolyte membrane (PEM) fuel cells for mobile and stationary applications. The operating conditions and operational sensitivities of process parameters that determine the operational characteristics of any fuel cell dictate the accuracy and response requirements of sensors for fuel cell application. This paper will discuss merits and demerits of traditional automotive sensors for fuel cell application. The paper will also discuss need for laboratory sensors that can be effectively used to investigate the process dynamics and how such empirical data can be used to develop control-relevant models as online estimators. One of the challenges of meeting the reliability targets is to develop process monitoring strategies to detect onset of instability and take remedial action. In particular for any fuel cell stack, the stability is often caused by cell-to-cell variation. For example Wilkinson (2003) proposed that reliability of fuel cell stack is a product of reliability of individual cells:

R_stack = R1(t)*R2(t)*..*Rn(t)

where Ri(t) is the reliability of ith cell at time (t), and n is the number of cells.

The control system only monitors the average process condition for the entire stack, e.g. the temperature sensor measures the average temperature of coolant in the individual cells. To stimulate interest in needs for diagnosing cell-to-cell variation an example will be presented on how the response of individual voltage can be correlated to cell-to-cell variation in level of hydration.