Speaker
Description
Quantitative infrared thermography (QIRT) is employed for non‑intrusive, area‑wide temperature measurements in high‑speed aerodynamic facilities, where accurate surface‑temperature data are essential for determining local heat‑flux densities. A series of wind‑tunnel experiments at the Mach 6 flow condition of the Ludwieg-Tube facility Göttingen (RWG) regarding a 2D shock wave impinging on a transitional flat-plate boundary layer revealed a systematic over-prediction of the heat-flux density (Stanton number). Upstream of the interaction region the experimentally derived Stanton numbers were approximately 13% higher compared to the reference case, a discrepancy exceeding the statistical uncertainty of the measurement and that could not be reconciled with theory or numerical simulations. A dedicated laboratory campaign was therefore launched to scrutinize the infrared camera to identify unconsidered sources of measurement bias.
Two distinct systematic errors were uncovered by observing an object with a fixed temperature and periodically covering a significantly hotter second object in the field of view (FOV). The first error was attributed to lens‑flare effects and manifests as a uniform dependency of the recorded count value, depending on the temperature composition in the FOV. Detailed analysis revealed that the error scales linearly with the mean count value of the image, indicating that multiple internal reflections within the camera optics redistribute radiant energy leading to false temperature recordings. A heuristic correction model was derived by correlating the change in pixel values with the change in the image‑wide mean count, yielding a proportionality factor that could be applied to correct the lens flare effect.
The second error, became apparent after rotating the camera so that the readout direction traversed both the calibrated surface and the hot area. Even after the lens‑flare correction, a residual variation persisted along each readout line, following a perceived parabolic pattern with respect to pixel position. This behavior indicated a coupling between pixels during the line‑wise digitization process. By fitting a simple parabolic function to the observed variation, a correction factor was obtained for each pixel position, effectively decoupling the readout bias from the true signal. Both correction steps are applied in reverse order of the camera’s signal‑processing chain (readout correction preceding lens‑flare correction) to ensure physical consistency.
With both corrections in place, a new temperature calibration was performed using a high‑emissivity black body emitter. The corrected count values were mapped to absolute surface temperatures through a calibrated relationship based on the Stefan‑Boltzmann law.
When the corrected temperature fields are fed into the QIRT heat‑flux algorithm, the previously observed 13 % excess in Stanton number upstream of the shock-wave / boundary-layer interaction disappears and matches the reference curve within the statistical noise. The correction has negligible impact in regions with high heat flux, indicating that the systematic biases are only significant when the temperature changes over time are comparable to the magnitude of the error.
