(186c) 2D/3D Turbulent Flow Calculation with Lattice Boltzmann Method

Kim, D. - Presenter, Carnegie Mellon University
Vinay, III, S. - Presenter, Bechtel Marine Propulsion Corporation, Bettis Laboratory
Buchanan, J. - Presenter, Bechtel Marine Propulsion Corporation, Bettis Laboratory
Jhon, M. S. - Presenter, Carnegie Mellon University

In this paper, we constructed an alternative computational fluid dynamics (CFD) code based on lattice Boltzmann method (LBM) [1]. In general, the continuum based, Navier-Stokes equations can be derived through the Chapman-Enskog multi-scale expansion from the lattice Boltzmann equation (LBE) [1]. In general, LBM can play a critical role in continuum fluid flow simulations as well as in multi-scale modeling.

We developed and implemented an efficient algorithm for fluid flow within arbitrary geometries through LBM. This tool can handle complex geometries more easily than the conventional CFD. The most important ingredient of complex geometry treatment process is a point inclusion test module, which decides whether a point is inside a solid or not [2]. For a guaranteed user convenience, our module takes in the CAD STL format data as an input.

The Smagorinsky large eddy simulation was implemented for the description of turbulent flows. In previously applied methods, instability arises during the high Reynolds number flow simulation using LBM because inlet boundary conditions such as the equilibrium and bounce-back conditions make disturbances that reflect at the inlet boundary and therefore can not suppress the spuriously reflected disturbances. Therefore, we applied a novel non-reflecting boundary condition to compute high Reynolds number flows stably [3].

As a consequence, our LBM tool becomes an improved methodology for flow simulation with arbitrary Reynolds numbers. To demonstrate the capability of our LBM tool, we performed flow analysis around surface-mounted, prismatic obstacles placed in a fully developed flow channel [4]. The effects due to different dimensions are investigated to demonstrate the fundamental differences between the nominally 2D and fully 3D obstacle flows by computing the two results for different aspect ratios. The flow structures of separated regions in front of wider obstacles are examined to show if saddle and nodal points on the forward face of the obstacles are distributed quasi-regularly or not. We compared the complex vortex structures with experimental data for various aspect ratios. The interaction of the horseshoe vortex with the corner vortex behind the cube and with the impinging mixing-layer in the wake was also examined.


[1] S. Chen and G.D. Doolen, Annu. Rev. Fluid Mech., vol. 30, pp. 329, 1998.

[2] F. Feito and J. Torres, Comp. & Graph., Vol. 21, pp. 23, 1997.

[3] D. Kim, H. M. Kim, M. S. Jhon, S. J. Vinay III, and J. Buchanan, Chin. Phys. Lett., Vol. 25, pp. 1964, 2008.

[4] R. Martinuzzi and C. Tropea, J. Fluids Eng., Vol. 115, pp. 85, 1993.