Validation of the Lattice Boltzmann Method for Direct Numerical Simulation of Wall-Bounded Turbulent Flows

Abstract

In this work, the lattice Boltzmann method (LBM) was validated for direct numerical simulation (DNS) of wall-bounded turbulent flows. The LBM is a discrete-particle-based method that numerically solves the Boltzmann equation as opposed to conventional DNS methods that are based on the Navier-Stokes (NS) equations. The advantages of the LBM are its simple implementation, its ability to handle complex geometries, and its scalability on modern high-performance computers.

An LBM code was developed and used to simulate fully-developed turbulent channel flow. In order to validate the results, the turbulence statistics were compared to those calculated from a conventional NS-based finite difference (FD) simulation. In the present study, special care was taken to make sure the computational domains for LBM and FD simulations were the same. Similar validation studies in the literature have used LBM simulations with smaller computational domains in order to reduce the computational cost. However, reducing the size of the computational domain affects the turbulence statistics and confounds the results of the validation.

The turbulence statistics calculated from the LBM and FD simulations were found to agree qualitatively; however, there were several significant deviations, particularly in the variance profiles. The largest discrepancy was in the variance of the pressure fluctuations, which differed by approximately 7%. Given that both the LBM and FD simulations resolved the full range of turbulent scales and no models were used, this error was deemed to be significant.

The cause of the discrepancy in the pressure variance was found to be the compressibility of the LBM. The LBM allows the density to vary, while the FD method does not since it solves the incompressible form of the NS equations. The effect of the compressibility could be reduced by lowering the Mach number, but this would come at the cost of significantly increasing the computational cost. Therefore, the conclusion of this work is that, while the LBM is capable of producing accurate solutions for incompressible turbulent flows, it is significantly more expensive than conventional methods for simple wall-bounded turbulent flows.