Shock boundary layer interactions in a low aspect ratio duct

•We investigate shock boundary layer interactions using particle image velocimetry.•Depending on shock generating ramp size, regular or Mach reflection occurs.•Significant spanwise non-uniformity is observed in the low aspect ratio duct.•Detailed mean and turbulence quantities are provided for CFD validation.•Excursion length of reflected shock does not depend on strength of interaction. Experimental data acquired using high resolution two-component particle image velocimetry (PIV) are presented for shock boundary layer interactions (SBLIs) generated by a compression–expansion ramp geometry. The incident oblique shock wave is generated by a sub-boundary layer height ramp inclined 20° to the M∞=2.05 inflow. Results are presented for two different ramp sizes (hramp/δ0=0.56 and hramp/δ0=0.93), and compared to a previously documented hramp/δ0=0.20 case. For each case, mean velocitiy and turbulent statistics for both the SBLI at the foot of the compression ramp and the incident/reflected SBLI on the opposite wall are analyzed. Data are acquired in several streamwise-vertical planes across the span of the low aspect ratio duct in order to document spanwise non-uniformities and confinement effects, with the specific goal of producing a high quality experimental database for CFD validation. The angles of the incident and reflected shock waves become steeper as the side wall is approached, due to the lower velocities within the side wall boundary layer. Mean flow reversal is observed near the spanwise centerline of the duct, but only instantaneous flow reversal is seen closer to the side walls. These spanwise non-uniformities are more prominent for the stronger interactions caused by the larger ramp geometry. For this case there is no nominally two-dimensional region of the flow, and a Mach stem occurs in the core of the flow, causing a significant subsonic wake. The shock excursion length scale relative to the incoming boundary layer thickness, Lex/δ0, is measured for all of the shock features and found to be significantly lower than values reported in the literature for similar flows. Furthermore, Lex/δ0 for the reflected shock does not depend on the strength of the incident shock or the size of the separated zone.