High-speed high Reynolds number boundary layer flows with mechanical non-equilibrium effects have numerous practical applications; examples include access-to-space ascent, re-entry and descent, and military hypersonic systems. However, many of the basic turbulent flow processes in this regime are poorly understood and are beyond the realm of modern direct numerical simulations Previous studies have shown that curvature driven pressure gradients significantly alter the state of the turbulence in high-speed boundary layers; the turbulence levels have been shown to decrease by large amounts (up to 100 percent) and the Reynolds shear stress has been shown to change sign. However, most of our understanding is based on point measurement techniques such as hot-wire and Laser Doppler anemometry acquired at low to moderate supersonic Mach numbers (i.e., M = 2-3). After reviewing the available literature, the following scientific questions remain unanswered pertaining to the effect of favorable pressure gradients: (1) How is state of the mean flow and turbulence statistics altered? (2) How is the structure of wall turbulence; break-up, stretch or a combination? (3) How are the Reynolds stress component production mechanisms altered? (4) What is the effect of Mach number on the above processes? To answer these questions and to enhance the current database, an experimental analysis was performed to provide high fidelity documentation of the mean and turbulent flow properties using two-dimensional particle image velocimetry (PIV) along with flow visualizations of a high speed (M4.88=), high Reynolds number (Re36,000θ≈) supersonic turbulent boundary layer with curvature-driven favorable pressure gradients (a nominally zero, a weak, and a strong favorable pressure gradient). From these data, detailed turbulence analyses were performed including calculating classical mean flow and turbulence statistics, examining turbulent stress production, and performing quadrant decomposition of the Reynolds stress for each pressure gradient case. It was shown that the effect of curvature-driven favorable pressure gradients on the turbulent structure of a supersonic boundary layer was significant. For the strong pressure gradient model, the turbulent shear stress changed sign throughout the entire boundary layer; a phenomena was not observed to this magnitude in previous studies. Additionally, significant changes were seen in the turbulent structure of the boundary layer. It is believed that hairpin vortices organized within the boundary layer are stretched and then broken up over the favorable pressure gradient. Energy from these hairpin structures is transferred to smaller turbulent eddies as well as back into the mean flow creating a fuller mean velocity profile. It was determined that the effects of favorable pressure gradients on the basic structure of a turbulent Mach 5.0 boundary layer were significant, therefore increasing the complexity of computational modeling.

Previous studies have shown that favorable pressure gradients reduce the turbulence levels and length scales in supersonic flow. Wall roughness has been shown to reduce the large-scales in wall bounded flow. Based on these previous observations new questions have been raised. The fundamental questions this dissertation addressed are: (1) What are the effects of wall topology with sharp versus blunt leading edges? and (2) Is it possible that a further reduction of turbulent scales can occur if surface roughness and favorable pressure gradients are combined? To answer these questions and to enhance the current experimental database, an experimental analysis was performed to provide high fidelity documentation of the mean and turbulent flow properties along with surface and flow visualizations of a high-speed ( 2.86 M = ), high Reynolds number (Re 60,000 q » ) supersonic turbulent boundary layer distorted by curvature-induced favorable pressure gradients and large-scale ( 300 s k + » ) uniform surface roughness. Nine models were tested at three separate locations. Three pressure gradient models strengths (a nominally zero, a weak, and a strong favorable pressure gradient) and three roughness topologies (aerodynamically smooth, square, and diamond shaped roughness elements) were used. Highly resolved planar measurements of mean and fluctuating velocity components were accomplished using particle image velocimetry. Stagnation pressure profiles were acquired with a traversing Pitot probe. Surface pressure distributions were characterized using pressure sensitive paint. Finally flow visualization was accomplished using schlieren photographs. Roughness topology had a significant effect on the boundary layer mean and turbulent properties due to shock boundary layer interactions. Favorable pressure gradients had the expected stabilizing effect on turbulent properties, but the improvements were less significant for models with surface roughness near the wall due to increased tendency towards flow separation. It was documented that proper roughness selection coupled with a sufficiently strong favorable pressure gradient produced regions of “negative” production in the transport of turbulent stress. This led to localized areas of significant turbulence stress reduction. With proper roughness selection and sufficient favorable pressure gradient strength, it is believed that localized relaminarization of the boundary layer is possible.

Turbulent boundary layers approximating those found on the NASA Orion Multi-Purpose Crew Vehicle (MPCV) thermal protection system during atmospheric reentry from the International Space Station have been studied by direct numerical simulation, with the ultimate goal of reducing aerothermodynamic heating prediction uncertainty. Simulations were performed using a new, well-verified, openly available Fourier/B-spline pseudospectral code called Suzerain equipped with a ``slow growth'' spatiotemporal homogenization approximation recently developed by Topalian et al. A first study aimed to reduce turbulence-driven heating prediction uncertainty by providing high-quality data suitable for calibrating Reynolds-averaged Navier--Stokes turbulence models to address the atypical boundary layer characteristics found in such reentry problems. The two data sets generated were Ma[approximate symbol] 0.9 and 1.15 homogenized boundary layers possessing Re[subscript theta, approximate symbol] 382 and 531, respectively. Edge-to-wall temperature ratios, T[subscript e]/T[subscript w], were close to 4.15 and wall blowing velocities, v[subscript w, superscript plus symbol]= v[subscript w]/u[subscript tau], were about 8 x 10-3 . The favorable pressure gradients had Pohlhausen parameters between 25 and 42. Skin frictions coefficients around 6 x10-3 and Nusselt numbers under 22 were observed. Near-wall vorticity fluctuations show qualitatively different profiles than observed by Spalart (J. Fluid Mech. 187 (1988)) or Guarini et al. (J. Fluid Mech. 414 (2000)). Small or negative displacement effects are evident. Uncertainty estimates and Favre-averaged equation budgets are provided. A second study aimed to reduce transition-driven uncertainty by determining where on the thermal protection system surface the boundary layer could sustain turbulence. Local boundary layer conditions were extracted from a laminar flow solution over the MPCV which included the bow shock, aerothermochemistry, heat shield surface curvature, and ablation. That information, as a function of leeward distance from the stagnation point, was approximated by Re[subscript theta], Ma[subscript e], [mathematical equation], v[subscript w, superscript plus sign], and T[subscript e]/T[subscript w] along with perfect gas assumptions. Homogenized turbulent boundary layers were initialized at those local conditions and evolved until either stationarity, implying the conditions could sustain turbulence, or relaminarization, implying the conditions could not. Fully turbulent fields relaminarized subject to conditions 4.134 m and 3.199 m leeward of the stagnation point. However, different initial conditions produced long-lived fluctuations at leeward position 2.299 m. Locations more than 1.389 m leeward of the stagnation point are predicted to sustain turbulence in this scenario. / text