Uncertainty quantification of turbulent statistics from hot-wire measurements applied in a high pressure turbine stage

The present work aims to characterize the flow field at the inlet of a high pressure (HP) turbine stage in terms of turbulence parameters such as turbulence intensity, time and length scales and turbulent spectra. Furthermore, a non-parametric statistical approach is proposed in order to confer all turbulent statistics with error bars representing 95% confidence intervals. In order to test aerodynamic and thermal performance in engine representative conditions, the Von Karman Institute (VKI) CT3 blow-down test rig is used, operating with the principle of isentropic light piston compression tube. Hot-wire anemometry (HWA) technique is employed to acquire velocity signal up to high frequency fluctuations representing the smallest scales of a turbulent flow. However, due to the step-wise evolution of temperature and pressure in blow-down facilities and due to the not feasibility of the flow to maintain a stable condition for a large time period (the test time in CT3 wind tunnel is around 0.15 seconds), the use of HWA is not simple. A non dimensional calibration technique (Cukurel 2012) allows to reduce HWA data also for strongly non-isothermal flows with a single curve fit. For measurements concerning turbulence, the parameters of interest are high order statistical moments, for which no conventional methods are available in order to quantify their uncertainty. The objective of this thesis is to present a standard method for the estimation of uncertainties of turbulent statistics by means of a re-sampling technique, the Moving Block Bootstrap (MBB). Thanks to this non-parametric approach, based on a straightforward algorithm, it is possible to perform statistical inference on high order moments without any assumption on their distribution. The blocks length is selected either by an automatic algorithm (Politis 2004) or by a sensitivity analysis. The latter method resulted to perform better for highly correlated structures and for flow presenting a high oscillating autocorrelation function around zero.