Long-Term Continuous Gas Diagnostics For Fuel Cell Systems: Development Of An Automatized Raman Metrology

Abstract

Knowledge of the composition of gas flows in fuel cell systems is crucial for their understanding. In current research, this data is rarely recorded inline and in real-time due to the lack of suitable metrology. Here, we present an optical gas sensor. Green light (532 nm) from a 450 mW laser excites Raman scattering. A spectrometer analyzes the scattered light and a Python software extracts gas composition data automatized and in real-time. We designed a measurement cell that allows optical inline access to gas flow systems and is easy to implement. Therefore, common signal enhancement methods with more complex setups can not be applied. Literature data on scattering cross sections does not offer a reliable basis for calibration what constitutes the need for a gas conditioning system that allows fast preparation of well defined mixtures. We used it to present two principles of calibration. First, calibration based on relative scattering cross sections is the more stable possibility, because it is still valid in cases where laser power and integration time are changed. However, in consequence the results for one species are not independent from the other species. A poor signal, e.g. following from a low mole fraction in a single component, will introduce uncertainty in all components. We avoid that problem through direct linear correlation of signal strength of one species to its partial pressure, which is the second calibration variant. In this case, there is an influence of laser power and integration time. The additional information on overall pressure can be used in model systems, where the presence of undetected components can be ruled out. We can compare that information to pressure measurements to do plausibility checks on the Raman results in real-time. We apply the sensor in fuel cell system research, where measurement intervals of one second yield results of sufficient precision. Increasing integration time can further decrease the uncertainties where favorable. The measurement and data evaluation routines work stable and constant enough to monitor gas compositions in longterm operation. To date, the sensor can measure mixtures of nitrogen, oxygen, hydrogen and water vapor.