Measurement of Gas Thermal Properties Using the Parametric Reduced-Order Modeling Approach

This paper deals with a thermal gas property micro-sensor. The proposed modeling approach of the sensor was based on reduced order modeling, in contrast to the traditional analytical modelling approach, which is the standard for this kind of sensors. This sensor was deployed for the measurement of the thermal conductivity (k) and the volumetric heat capacity (ρc p ) of gases and works according to the temperature oscillation technique. A proper model is crucial for the measurement accuracy. The scope of this paper was to investigate the applicability of a sensor model based on a reduced-order modeling approach, intending to improve the performance of this sensor, as the behavior of the sensor can be modeled much more accurately than using an analytical model. For this reason, a parametric model-order reduction technique using proper orthogonal decomposition was applied. The main advantage of the reduced-order model is the high accuracy in the modeling of the conductive heat transfer problem, while it requires low computation effort. The approach was tested experimentally, where the model was calibrated in two pure gases and evaluated in 21 gases and gas mixtures. The sensor achieved an accuracy in the thermal conductivity of 6.5% and in the volumetric heat capacity of 3.2%.