Generalized lattice Boltzmann equation with forcing term for computation of wall-bounded turbulent flows with using CUDAprogramming

In this paper we implement generalized lattice Boltzmann equation (GLBE) usingmultiple relaxation times with forcing term. The implementation of LBM is on a GPU(Graphic Processing Unit) platform using CUDA (Compute Unified DeviceArchitecture). In the present paper we accurately using a CUDA code of a latticeBoltzmann solver for simulating turbulent flows. we present the combination of LBMand CUDA for computing of turbulent channel flow for which large eddy simulation(LES) was employed as a turbulence model and compare our results with the results ofdirect numerical simulation (DNS). The sub grid scale (SGS) turbulence effects weresimulated through a shear-improved Smagorinsky model (SISM), which is capable ofpredicting turbulent near wall region accurately without any wall function. Computationswere done for fully developed turbulent channel flow at Re 180   , based on channelhalf-width and average friction velocity. It was concluded that using GPUs decrease thecost and the time of computation which help us to solve many practical problems. Thedefect of nowadays GPUs is that they are single precision.

Keywords:

Turbulent flow , channel flow , Generalized lattice Boltzmann equation , Large eddysimulation , Shear improved smagorinsky model , CUDA