SRS inversion for measurement of temperature profile in a combustion gas layer

Abstract

The spectral remote sensing (SRS) inversion technique to determine the complex, line-of-sight temperature distribution from the gas emission data at $CO_2$ 4.3 micron is analytically investigated and experimentally validated. Three inversion schemes are proposed: the Modified Constrained Inversion Method (MCIM), the Base Function based Inversion Method (BFIM) and the BFIM based Modified Constrained Inversion Method (BCIM). These three schemes are extensively tested with numerical experiments to evaluate their convergence characteristics for practical applications. In the calculation of intensity, the CK-based WNB model is employed throughout this study. The MCIM shows good applicability only to relatively simple profiles, while the BFIM gives reasonable estimations even for complicated multi-peak profiles provided that appropriate base functions are selected. However, MCIM and BFIM may give inaccurate results especially when applied to the complicated, asymmetric temperature profiles frequently found in practical applications. On the other hand, BCIM treats this problem properly. It gives reasonable accuracy even when handling asymmetric, complicated temperature profiles. Also, the effect of concentration error is examined with these three algorithms. MCIM is seen to be very sensitive to concentration error, while BFIM is robust to it. BCIM again gives reasonable accuracy in spite of some artificially imposed concentration error. Based on the numerical results, three inversion schemes are applied to retrievals of two-peak experimental gas temperature profiles. For this purpose, a 3.4m long stainless steel test furnace is prepared with two kerosene burners attached. Various test conditions are simulated to examine the practical performance of the three algorithms. The end wall boundary with the emissivity of 0.35 is employed. The spectral measurements between 2100 $cm^{-1}$ and 2460 $cm^{-1}$ are made, which later are inverted into the gas temperature profile. T...