Abstract

Wake patterns behind a golf ball with and without back-spin ($ωD/2U=0.1$ where ω is the angular velocity, D is the diameter of the golf ball, and U is the freestream velocity) were investigated at $ReD=1.1×105$. To simulate wake regions, the in-house Cartesian mesh-based incompressible flow solver, namely ULSAN3D-Cart, was utilized. Navier-Stokes equations for incompressible flows are solved by the standard projection method with the unconditionally stable semi-Lagrangian scheme. The pressure Poisson equation is solved by the geometric multigrid method. The computational domain was set to $[−4D,28D]×[−8D,8D]2$. The number of grid cells was set to $4,096×2,0482(≅17.2×109)$. By using hybrid MPI/OpenMP parallelization, this large-scale flow was simulated on the KISTI Nurion supercomputer. For the flow simulation, total 32,768 cores of Intel Xeon Phi 7250 processors are utilized. For the visualization of the wake field, in-situ parallel processing was conducted on the Nurion machine using 17,408 CPU cores utilizing the open-source visualization software, Paraview. By the comparison of the stationary and the back-spinning golf balls, the differences in the wake patterns are clearly observable. For the stationary case, wake stretches in a symmetric manner with respect to the horizontal center line. On the other hands, for the back-spinning case, the wake bends downward direction, which indicates the generation of the upward lift.

This content is only available via PDF.