Gas turbine engines heavily rely on the durability of hot-section components to achieve the required levels of performance, reliability, and safety. While high pressure turbines are exposed to gas path temperatures approaching their melting points, features such as cooling systems and environmental coatings are used in combination to meet design goals. The performance of these parts is critically dependent on the temperatures, cycles, time, and stresses achieved during engine operation. Thermometry systems offer nonintrusive optical temperature monitoring for hot-section diagnostics. However, their potential is currently hindered by poor absolute temperature accuracy (large error bounds), as a result of ill-characterized uncertainty sources. Modern applications attempt to circumvent this issue by empirical corrections (target specific calibration), which is particularly problematic for surfaces with low and varying emissivity, as encountered in most metals. Unlike most common monochromatic pyrometers, we are focusing our efforts on multi-spectral thermography of unknown emissivity surfaces. Although the emissivity is typically a function of both wavelength and temperature, on sufficiently close spectral bands, per-scenario assumptions (such as graybody, linear change with wavelength, etc.) are valid, and provide direct solution to the system matrix. By acquiring multi-integration time images and conducting quantitative image fusion considering total exposure non-linearity compensation, the currently developing optimized multispectral radiation thermography technique is geared towards accurate 2-D temperature measurement of hot target objects, absent of any repeated calibration. Thereby, directly decoupling surface temperature could contribute to significant advances in online monitoring of gas turbines.
