Influence of unsteady wake on turbulent separated flows

Abstract

An experimental study was made of a turbulent separation bubble over a blunt body, where unsteady wake was generated by a spoke wheel-type wake generator with 24 cylindrical rods. The influence of unsteady wake on the turbulent separation bubble was scrutinized by altering the rotating direction (CW and CCW) and the normalized passing frequency ($0≤St_H≤0.20). The Reynolds number based on the cylindrical rod was $Re_d=4,000$ and the Reynolds number based on the blunt body was $Re_H=6,000$. As an introductory study, time-averaged measurements were performed to portray the turbulent separated flow over the blunt body with unsteady wake. To delineate the wall-pressure fluctuations over the turbulent separation bubble, a 32-channel microphone array was installed inside the blunt body. As a detailed study of the flow structures, large-scale vortical structures of a turbulent separation bubble under the influence of an unsteady wake were investigated. The mechanism of vortex shedding from the separation bubble was analyzed in detail by taking a conditional average as well as a phase average. Spatial-box-filtering (SBF) was used to extract the large-scale vortical structure from the turbulent separation bubble affected by the unsteady wake. As an attempt to understand the flow field, the time scale-variant flow structure of a turbulent separation bubble under the influence of an unsteady wake was introduced and investigated. Maximal overlap discrete wavelet transform (MODWT) was used to extract scale-resolved vortical structure from the turbulent separation bubble affected by the unsteady wake. Another experimental study was made of a turbulent separated flow over a backward-facing step, where unsteady wake was generated by a spoke wheel-type wake generator with cylindrical rods. The influence of unsteady wake was scrutinized by rotating the wake generator by changing the normalized passing frequency ($0≤St_H≤0.40$). The Reynolds number based on the step height was $R...