Measurements are presented of the aero-optic distortion produced by a Mach 8 turbulent boundary layer in the Sandia Hypersonic Wind Tunnel. Flat optical inserts installed in the test section walls enabled a double-pass arrangement of a collimated laser beam. The distortion of this beam was imaged by a high-speed Shack-Hartmann sensor at a sampling rate of up to 1 MHz. Analysis is performed using two processing methods to extract the aero-optic distortion from the data. A novel de-aliasing algorithm is proposed to extract convective-only spectra and is demonstrated to correctly quantify the physical spectra even in case of relatively low sampling rates. The results are compared with an existing theoretical model, and it is shown that this model under-predicts the experimentally measured distortions regardless of the processing method used. Possible explanations for this discrepancy are presented. The presented results represent to-date the highest Mach number for which aero-optic boundary layer distortion measurements are available.
Previous efforts determined a set of calibrated model parameters for ReynoldsAveraged Navier Stokes (RANS) simulations of a compressible jet in crossflow (JIC) using a k-ɛ turbulence model. These coefficients were derived from Particle Image Velocimetry (PIV) data of a complementary experiment using a limited set of flow conditions. Here, k-ɛ models using conventional (nominal) and calibrated parameters are rigorously validated against PIV data acquired under a much wider variety of JIC cases, including a flight configuration. The results from the simulations using the calibrated model parameters showed considerable improvements over those using the nominal values, even for cases that were not used in defining the calibrated parameters. This improvement is demonstrated using quality metrics defined specifically to test the spatial alignment of the jet core as well as the magnitudes of flow variables on the PIV planes. These results suggest that the calibrated parameters have applicability well outside the specific flow case used in defining them and that with the right model parameters, RANS results can be improved significantly over the nominal.
Near-wall turbulence models in Large-Eddy Simulation (LES) typically approximate near-wall behavior using a solution to the mean flow equations. This approach inevitably leads to errors when the modeled flow does not satisfy the assumptions surrounding the use of a mean flow approximation for an unsteady boundary condition. Herein, modern machine learning (ML) techniques are utilized to implement a coordinate frame invariant model of the wall shear stress that is derived specifically for complex flows for which mean near-wall models are known to fail. The model operates on a set of scalar and vector invariants based on data taken from the first LES grid point off the wall. Neural networks were trained and validated on spatially filtered direct numerical simulation (DNS) data. The trained networks were then tested on data to which they were never previously exposed and comparisons of the accuracy of the networks’ predictions of wall-shear stress were made to both a standard mean wall model approach and to the true stress values taken from the DNS data. The ML approach showed considerable improvement in both the accuracy of individual shear stress predictions as well as produced a more accurate distribution of wall shear stress values than did the standard mean wall model. This result held both in regions where the standard mean approach typically performs satisfactorily as well as in regions where it is known to fail, and also in cases where the networks were trained and tested on data taken from the same flow type/region as well as when trained and tested on data from different respective flow topologies.
Turbulent viscosities have been calculated from stereoscopic particle image velocimetry (PIV) data for a supersonic jet exhausting into a transonic crossflow. Image interrogation must be optimized to produce useful turbulent viscosity fields. High-accuracy image reconstruction should be used for the final iteration, whereas efficient algorithms produce spatial artifacts in derivative fields. Mean strain rates should be calculated from large windows (128 pixel) with 75% overlap. Turbulent stresses are optimally computed using multiple (more than two) iterations of image interrogation and 75% overlap, both of which increase the signal bandwidth. However, the improvement is modest and may not justify the considerable increase in computational expense. The turbulent viscosity may be expressed in tensor notation to include all three axes of velocity data. In this formulation, a least-squares fit to the multiple equations comprising the tensor generated a scalar turbulent viscosity that eliminated many of the artifacts and noise present in the single-component formulation. The resulting experimental turbulent viscosity fields will be used to develop data-driven turbulence models that can improve the fidelity of predictive computations.
