Experimental Measurement and Computational Simulation of the Flow in the Queen’s University Low Speed Wind Tunnel

Loading...
Thumbnail Image
Date
2014-08-11
Authors
Wexler, Chloe
Keyword
Low Speed Wind Tunnel , Computational Fluid Dynamics
Abstract
The undisturbed flow behavior in the Queen’s University Low Speed Wind Tunnel was investigated experimentally and computationally. The velocity profiles, turbulent parameters, boundary layer and corner effects were detailed by traversing the lower left quadrant of the wind tunnel with a seven-hole pressure probe and uniaxial constant temperature hot-wire anemometer. The research goal was to provide a complete reference data set as a comparison standard for future experimental work. The flow in the wind tunnel was verified using computational fluid dynamics in order to present the model requirements and inlet conditions required to accurately simulate the flow. The experimental results indicated that the LSWT was capable of producing a mean axial flow speed of 30 – 31 m/s, dependent on ambient conditions and atmospheric wind speeds, in a 1 m2 test section. Measurements were collected under a variety of external weather conditions and it was determined that the average transient and spatial flow deviation from the mean was on the order of 1.5% and 0.5% respectively. The boundary layer thickness at the exit of the test section was found to be 10 cm above the floor and the corner effects were determined to extend 15 cm into the test section. The turbulent intensity reached 10% at the wall boundaries decaying to a constant uniform value of 3% in the center. Computational simulations were performed for three specific sets of experimental data. The results indicated that the mean axial velocity was predicted with an average deviation of 0.5%. Solution results were extrapolated 1 m downstream from the contraction exit plane, averaged and used to provide reference inlet conditions to a simplified test section model. The turbulence model applied to the solution was varied; the Spalart-Allmaras, k-ε, k-ω, and the blended k-ω SST models were able to resolve the turbulent distribution and flow behavior in a comparable manner. The flow was predicted within the bounds of engineering accuracy, the discrepancy in the simulated and experimental mean velocity ranged from 0.4% to 0.02%.
External DOI