A Probabilistic Approach for Correction of Multi-Hole Directional Probes for Varying Flow Fields.

Multi-hole directional probes are commonly used in experimental fluid mechanics to assess flow angles, total and static pressure. They need to undergo calibration to relate these quantities to the pressure readings in the measurement holes. This relationship is however dependent on the flow Mach and Reynold's number. It is also influenced due to flow distortion induced by the probe or interaction between probe and test article, especially in confined flows. Since it is not always feasible to calibrate in a similar flow field as the test, a correction methodology is developed that combines flow features and trends obtained numerically with experimental calibration. RANS simulations are run for an ideal probe for the actual test condition and results are used to train a Bayesian model to reduce the number of CFD simulations required for the range of the calibrated angle. This is used to provide enough data to generate a correction model. The correction model is then mapped to an experimental calibration, done at another Mach and Reynolds number, and is used to predict the true probe calibration at the required flow conditions. This methodology is compared against experimental calibration done at different flow conditions.