Text

theses

Turbulence models require constant research and development due to the nature of the models themselves. This thesis investigates the fidelity of the Reynolds-averaged Navier-Stokes (RANS) based 2006 Wilcox k - w turbulence model. The commercial flow solver GASPex is utilized for simulations, along with MATLAB for grid generation and Tecplot for post-processing. Associated results obtained are subsequently compared to an experimental study done by CUBRC in 2014. In this study, CUBRC ran a series of supersonic flow experiments on multiple physical configurations. The data obtained from these experiments include surface pressure and surface heat transfer values in regions of shock-wave turbulent boundary-layer interaction (SBLI). The purpose of the study was to document this data for further blind code validation studies. This thesis focuses on the results obtained for the large cone flare configuration. Ten runs were completed on the large cone flare, where six of the ten runs were simulated for comparison. Corresponding Mach numbers for the experiment range from 5 to 8. A grid convergence study was done and documented to ensure solution independence of grid discretization. Computational results conclude that the Wilcox k - w model predicts surface pressure well for all cases. Average surface pressure is predicted reasonably upstream of SBLI and post-flare, and peak surface pressure is predicted within the experimental uncertainty. However, separation is found to be significantly over-predicted for most cases. The Wilcox k - w model is shown to predict surface heat transfer poorly throughout. In regions of SBLI, surface heat transfer is shown to be drastically over-predicted, especially peak magnitudes. Additionally, it can be seen that the Wilcox k - w model produces a large anomalous spike in surface heat transfer downstream of the cone-flare junction in all cases. This spike is shown to be directly correlated to a large spike in turbulent kinetic energy near the surface of the large cone, observed at the same location. Causes for this spike are currently unknown and have not been further investigated, however similar spikes have been seen in the computational results obtained for the hollow cylinder flare configuration. Future work encompasses further assessment of the Wilcox k - w model in similar flow regimes. Since over-prediction is a strong factor of error in regions of SBLI, modifications to the Wilcox k - w model are required for more accurate predictive capabilities. The anomalous spike that occurs in surface heat transfer also needs be fully investigated to determine possible causes and resolutions.

ETD_9257

doi:10.7282/T3XD1593

M.S.

Includes bibliographical references

by Robert A. Alviani

ETD graduate

The author owns the copyright to this work.