Forcing of Separation Bubbles by Main Flow Unsteadiness or Pulsed Vortex Generating Jets—A Comparison

Abstract

Low pressure turbines typically operate in the low Reynolds number regime. Depending on the loading of the blade, they may exhibit detached flow with associated reattachment in the rear part of the suction surface. Additionally, the flow is highly timedependent due to the sequence of rotating and stationary blade rows. The work presented in this paper covers experimental efforts taken to investigate this type of flow in detail. Typical low pressure turbine flow conditions have been chosen as baseline for the experimental work. A pressure distribution has been created on a flat plate by means of a contoured upper wall in a low speed wind tunnel. The distribution matches the one of the PakB airfoil. Unsteadiness is then superimposed in two ways: A specific unsteadiness was created by using a rotating flap (RF) downstream of the test section. This results in almost sinusoidal periodic unsteady flow across the plate, simulating the interaction between stator and rotor of a turbine stage. Furthermore, pulsed blowing by vortex generating jets (VGJ) upstream of the suction peak was used to influence the transition process and development of the separation bubble. Measurements have been performed with hotwire anemometry. Experimental results are presented to compare both forcing mechanisms. In sinusoidal unsteady main flow, the transition occurs naturally by the breakdown of the shear layer instability, which is affected by periodic changes in the overall Reynolds number and thus pressure gradient. In opposition, active flow control (AFC) by VGJ triggers the transition process by impulse and vorticity injection into the boundary layer, while maintaining a constant Reynolds number. The flow fields are compared using phase averaged data of velocity und turbulence intensity as well as boundary layer parameters, namely shape factor and momentum thickness Reynolds number. Finally, a model to describe the time mean intermittency distribution is refined to fit the data.