Problem description
Large eddy simulation (LES) of the flow in a channel with a flameholder, or simplified the flow past a triangular cylinder bluff body. The Reynolds number based on the cylinder height is 47 000. The objective of this study of a realistic engineering application is to validate the second order discretization and compare statistics of the fluctuating quantities like the Reynolds stress against experimental data.
Software
OpenCruncher, version 3.0
Hardware
NVIDIA GeForce GTX TITAN X (2x)
Results
Structured hexahedral meshes, level 1 with 2.4 million cells and level 2 with 19.2 million cells were used with periodic boundary conditions in z direction. The Volvo Aero corporation triangular bluff body has been extensively researched both in terms of experiments and simulation, here we compare with experimental data and the results of a commercial CFD solver.
Comparison for the measurement location marked in green:
There is a very good agreement between the current results, the experimental data and as well the commercial solver. But we have to notice that we need a CPU based HPC cluster with hundred blades to get the results with the commercial solver. OpenCruncher is running on a local workstation using only two GTX graphic cards. The Mach number is 0.05, the flow is almost incompressible but due to the fact that the density based solver OpenCruncher is using an explicit time integration technology we do not run into the typical linearization or precondition issues.
References
A. Sjunnesson, C. Nelsson and E. Max, “LDA measurements of velocities and turbulence in a bluff body stabilized flame”, Laser Anemometry, vol. 3, pp. 83–93, 1991