Time Resolved Stereo-PIV Measurements of the Horseshoe Vortex System in Low Aspect Ratio Pin-Fin Arrays

Open Access
Anderson, Corey David
Graduate Program:
Mechanical Engineering
Master of Science
Document Type:
Master Thesis
Date of Defense:
Committee Members:
  • Stephen P Lynch, Thesis Advisor
  • Pin-fin
  • PIV
  • Stereo Particle Image Velocimetry
  • Heat Transfer
  • Horseshoe Vortex
Pin-fin arrays are a type of internal cooling feature, often used in gas turbine engines, in which individual pin-fins act as turbulators to increase heat transfer to a cooling fluid. The horseshoe vortex (HSV) system is a flow structure which is associated with individual pin fins. The system forms upstream of the pin-fins, in the endwall junction region. The present study utilized time resolved, stereoscopic particle image velocimetry to examine the behavior of the HSV system within a pin-fin array at rows 1, 3, and 5; at Reynolds numbers of 1.0e4, 2.0e4, and 5.0e4; and at streamwise spacings of 1.73*Diameter, and 3.46*Diameter. For both spacings, the first row behavior was found to be the same. Increases in Reynolds number resulted in the time average HSV system moving closer to the pin, and both TKE and vorticity became more concentrated. In downstream rows, the shape of the time averaged HSV system showed reduced variation with both Reynolds number and row location. For the more loosely packed array spacing of 3.46*Diameter, vorticity contours in the downstream rows were very similar across all Reynolds numbers. For the more tightly packed spacing, vorticity was found to vary with Reynolds number and row location. Bimodal distributions of the streamwise component of velocity, representative of two modes, were found for each case. These distributions were used to perform conditional averaging on the flow. The HSV system’s oscillation between these two modes were found to cause regions of high streamwise and wall-normal components of turbulence. The spanwise component of turbulence in the HSV region was found to be dominated by upstream wake fluctuations instead of the HSV motion. The shape of the turbulent kinetic energy contours were driven by these regions of high turbulence. In downstream rows for both geometries, normalized turbulent kinetic energy was found to decrease with Reynolds number due to dissipation effects outpacing increases in mean velocity.