Numerical and experimental investigation of VG flow control for a low-boom inlet

The application of vortex generators (VGs) for shock/boundary layer interaction flow control in a novel external compression, axisymmetric, low-boom concept inlet was studied using numerical and experimental methods. The low-boom inlet design features a zero-angle cowl and relaxed isentropic compression centerbody spike, resulting in defocused oblique shocks and a weak terminating normal shock. This allows reduced external gas dynamic waves at high mass flow rates but suffers from flow separation near the throat and a large hub-side boundary layer at the Aerodynamic Interface Plane (AIP), which marks the inflow to the jet engine turbo-machinery. Supersonic VGs were investigated to reduce the shock-induced flow separation near the throat while subsonic VGs were investigated to reduce boundary layer radial distortion at the AIP. To guide large-scale inlet experiments, Reynolds-Averaged Navier-Stokes (RANS) simulations using three-dimensional, structured, chimera (overset) grids and the WIND-US code were conducted. Flow control cases included conventional and novel types of vortex generators at positions both upstream of the terminating normal shock (supersonic VGs) and downstream (subsonic VGs). The performance parameters included incompressible axisymmetric shape factor, post-shock separation area, inlet pressure recovery, and mass flow ratio. The design of experiments (DOE) methodology was used to select device size and location, analyze the resulting data, and determine the optimal choice of device geometry. The best performing upstream VGs, with a height of 0.8 of the incoming centerbody boundary layer, were found to reduce average shock-induced separation by as much as 84%. This effect is achieved by the streamwise vorticity-induced transfer of higher momentum fluid to the centerbody surface downstream of the device centerline. The resulting energized boundary layer is more resistant to separation through the post-shock adverse pressure gradient. Though the supersonic VGs did not significantly affect the AIP boundary layer profiles, the reduction and break-up of the separation region may have a stabilizing effect on streamwise shock oscillation (which can not be obtained through the RANS formulation). On the other hand, the downstream subsonic devices with a height of about one boundary layer thickness substantially reduced the AIP radial distortion, in a spanwise sense. This improvement by the VGs was achieved by entraining higher momentum fluid in the near-wall region and effectively re-distributing the radial boundary layer profile. The effect on overall total pressure at the AIP was less than 0.25%. Based on the above studies, a test matrix of supersonic and subsonic VGs was adapted for a large-scale inlet test to be conducted at the 8???x6??? supersonic wind tunnel at NASA Glenn Research Center (GRC). Comparisons of RANS simulations with data from the Fall 2010 8???x6??? inlet test showed that predicted VG performance trends and case rankings for both supersonic and subsonic devices were consistent with experimental results. For example, experimental surface oil flow visualization revealed a significant post-shock separation bubble with flow recirculation for the baseline (no VG) case that was substantially broken up in the micro-ramp VG case, consistent with simulations. Furthermore, the predicted subsonic VG performance with respect to a reduction in radial distortion (quantified in terms of axisymmetric incompressible shape factor) was found to be consistent with boundary layer rake measurements. To investigate the unsteady turbulent flow features associated with the shock-induced flow separation and the hub-side boundary layer, a detached eddy simulation (DES) approach using the WIND-US code was employed to model the baseline inlet flow field. This approach yielded improved agreement with experimental data for time-averaged diffuser stagnation pressure profiles and allowed insight into the pressure fluctuations and turbulent kinetic energy distributions which may be present at the AIP. In addition, streamwise shock position statistics were obtained and compared with experimental Schlieren results. The predicted shock oscillations were much weaker than those seen experimentally (by a factor of four), which indicates that the mechanism for the experimental shock oscillations was not captured. The primary frequency of the experimental shock oscillations (based on Power Spectral Densities) was found to be much lower than that based on flow separation or that based on flow spillage, and instead was consistent with acoustic instabilities between the shock and the downstream choke plane at the mass flow plug. Since the DES computations did not extend to the choke plane, they were not able to capture this acoustic mode. In addition, the novel supersonic vortex generator geometries were investigated experimentally (prior to the large-scale inlet 8???x6??? wind tunnel tests) in an inlet-relevant flow field containing a Mach 1.4 normal shock wave followed by a subsonic diffuser. A parametric study of device height and distance upstream of the normal shock was undertaken for split-ramp and ramped-vane geometries. Flow field diagnostics included high-speed Schlieren, oil flow visualization, and Pitot-static pressure measurements. Parameters including flow separation, pressure recovery, centerline incompressible boundary layer shape factor, and shock stability were analyzed and compared to the baseline uncontrolled case. While all vortex generators tested eliminated centerline flow separation, the presence of VGs also increased the significant three-dimensionality of the flow via increased side-wall interaction. The stronger streamwise vorticity generated by ramped-vanes also yielded improved pressure recovery and fuller boundary layer velocity profiles within the subsonic diffuser. The best case tested (a ramped-vane with height of 0.75 of the uncontrolled boundary layer thickness located 25 uncontrolled boundary layer thicknesses upstream of the shock) yielded the smallest centerline incompressible shape factor and the least streamwise oscillations of the normal shock. However, additional studies are needed to better understand the three-dimensional aspects of this flow since corner interaction effects were adversely impacted by the VG devices