In an assessment of PIV measurement accuracy under practical experimental circumstances, grid-generated turbulence in its early decay (xiM =13.6-15.2, ReA z 25) has been studied. Since a realtime processor was used, it was practical to acquire a statistically significant sample size of 3000 vector maps so that accurate turbulence statistics could be calculated. Results include moments of velocity components and spatial structures in terms of auto-correlations, turbulence scales and spectra. Data fall within limits of standard error estimates. This study demonstrates the need for large sample sizes, particularly for higher order statistics. Displacement vector [L] dpiXe! Pitch of pixels on CCD chip [L] d,truct Diameter of smallest resolved velocity-structure [L] f Longitudinal correlation function g Lateral correlation function g' Theoretical correlation function G k ID power density spectrum [VIP] k Turbulent kinetic energy [VIP] k' Harmonic number M Mesh spacing [mm] N Number of samples N, Number of vector maps N« Side of interrogation area [pixels] N, Number of vectors in x-direction N, Number of vectors in y-direction ReM Reynolds number based on mesh spacing, MUiv ReA Reynolds number based on the Taylor microscale, A,

SUMMARYThe faster growth curves in the speed of graphics processing units (GPUs) relative to CPUs have spawned a new area of development in computational technology. There is much potential in utilizing GPUs for solving evolutionary partial differential equations and producing the attendant visualization. We are concerned with modeling tsunami waves, where computational time is of extreme essence in broadcasting warnings. We employed an NVIDIA board on a MacPro to test the efficacy of the GPU on the set of shallow-water equations, and compared the relative speeds between CPU and GPU for two types of spatial discretization based on second-order finite differences and radial basis functions (RBFs). We found that the GPU produced a speedup by a factor of 8 in favor of the finite difference method and a factor of 7 for the RBF scheme. We also studied the atmospheric dynamics problem of swirling flows over a spherical surface and found a speedup of 5.3 by the GPU. The time steps employed for the RBF method are larger than those used in finite differences, because of the fewer number of nodal points needed by RBF. Thus, RBF acting in concert with GPU would hold great promise for tsunami modeling because of the spectacular reduction in the computational time.

No abstract

No abstract