Content area
Abstract
The tug-of-war between Computational Fluid Dynamics and wind tunnel is sometimes taken as a zero-sum game with time as wind tunnel will be replaced by CFD, which is far from reality. Effectively, symbiotic use of both tools edges the competition. The development in computational power elevated the synergy between CFD and experiments. This dissertation presents the CFD works done by the author in Turner-Fairbank Highway Research Center laboratory. The works include wind flow characterization in a test section of the TFHRC wind tunnel, simplification of an existing complex wind tunnel model, and study of wind-generic bridge deck interaction. Wind flow characteristics were assessed by velocity distributions in the test area of the empty full scale wind tunnel. The complexity in the computational domain was eliminated by reducing the size of the tunnel, greatly reducing the computational time and cost. The realizable k-ϵ turbulence model was used as a closure for the Reynolds Averaged Navier-Stokes Equations (RANS) and Unsteady Reynolds Averaged Navier-Stokes Equations (URANS) equations. For all the simulations, commercial numerical code, Star-ccm+, was used.
Simulations with and without screens were analyzed. The presence of the screens produced a fairly uniform velocity distribution across the test section, and patterned pressure distribution was also observed compared to screenless wind tunnel. A good agreement was obtained between the measured and calculated velocities. The open return wind tunnel was able to yield fairly uniform velocity distributions throughout the width of the test section. Ensuring removal of obstacles that can interfere the circulation of wind around the wind tunnel should not be undermined.
Floating number errors and geometry complexity were targeted during the simplification of the wind tunnel model. The computational size and time were reduced 45% and 40% respectively. Mesh sensitivity study was done to choose the correct discretized model. The computed velocity was found to be close to measured velocity.
Wind-bridge deck interaction was studied in the Reynolds number range of 4.25 x 104 -4.29 x 104 for seven angles of attack ranging from -15° to 15° in an interval of 5o. The aerodynamic response of the deck was numerically computed. URANS with implicit unsteady time steps was implemented. The calculated mean drag and lift coefficients were found to be close to the measured values. However, further work is recommended for the pitching moment coefficients, which aberrated from the wind tunnel testing results. The k-ϵ turbulence model was able to provide engineering results. The developed open circuit wind tunnel model lays the platform to studies such as the effect of aspect ratio of bridge decks and Reynolds number on aerodynamic performance.
