Proton exchange membrane hydrogen fuel cells generate electricity without direct carbon dioxide emissions. But their operation depends on water balance: excess liquid inside the cell leads to flooding, while drying out of the membrane reduces conductivity. To detect such failures in time, the state of the cell needs to be monitored directly during operation.
Usually electrochemical impedance spectroscopy is used for this - a method in which a weak alternating current of various frequencies is passed through the cell and the response is assessed. A standard measurement takes several tens of seconds or longer, and during this time the conditions inside the cell may change.
In an article in the International Journal of Hydrogen Energy, a group from the University of Tsukuba described a way to reduce both the frequency range and the number of measurement points. Instead of 41 frequencies, as in conventional spectroscopy, the researchers used a smaller number of frequencies. Comparison with the full data set showed that if the frequencies are selected so that they capture changes in impedance - the cell's total resistance to alternating current - diagnostic accuracy is higher than with the standard choice of frequencies.
In addition, the authors assembled a program combining impedance measurement, data analysis and control of the cell's operation. Resistance parameters calculated from impedance served as indicators of the internal state. The program automatically changed operating conditions, detecting and eliminating membrane drying and insufficient air supply.
According to the authors, the approach makes it possible to diagnose and correct the state of the fuel cell almost in real time without greatly complicating the system. A limitation of the work is that the method was tested on a reduced set of frequencies, rather than on fully arbitrary modes. Next, the researchers plan to further reduce the measurement time and optimize the choice of frequencies in order to bring the technology to practical application.