Tech Briefs
Solving Unsteady Separated Flow Using Large Eddy Simulation in ADINA CFD
Sufficiently accurate and efficient turbulence modeling is obviously most important in CFD. While Large Eddy Simulation (LES) is regarded as very accurate, the computational cost is known to be very high. Hence, LES is, for now, hardly used for complex industrial geometries (see for example Spalart, 2000, reference given below).
However, as computational power increases, and algorithms become more efficient, LES will be used to an increasing extent. To illustrate a solution obtained with ADINA, we revisit the flow past a wallmounted cubic obstacle problem presented in the July 30, 2006 News. The obstacle is simple but could of course be geometrically more complex, like encountered in the airplane or motor car industries. The simple obstacle is used because experimental data have been published to compare with.
The LES results presented here were obtained using the SmagorinskyLilly subgrid scale model. Random perturbations were superimposed on the fullydeveloped velocity profile at the inlet. A total of 1,500 time steps were used. The sampling to compute mean flow quantities was taken from the last 250 time steps. The space and time discretization schemes are both secondorder accurate. The mesh was relatively coarse, with approximately 650,000 3D elements.
The above movie shows the chaoticlike timedependent streamwise velocity contours in the symmetry plane of the block. The timeaveraged streamwise velocity contours corresponding to the same plane as in the movie are shown below. The timeaveraged results are also shown in a horizontal plane located at 0.05h from the block base, h = block height. The animation and the timeaveraged results reveal the main features of the flow: the upstream separation zone, two downstream separation zones and the recirculation zone above the block. Note that the instantaneous flow looks remarkably different from the timeaveraged flow in the primary downstream separation zone.
The table below shows separation length results obtained with ADINA, compared against some of the most accurate benchmarks
(experimental, and numerical using LES) found in the literature. The agreement is excellent for all three recirculation lengths.
Timeaveraged Streamwise Velocity Plot at Symmetry Plane
Timeaveraged Streamwise Velocity Plot, 0.05h from the Base of the Block
ADINA CFD (LES) 
Martinuzzi
& Tropea, 1993 (Experiments) 
Shah, 1998 (LES) >10^{6} cells 
Rodi et al., 1997 (LES) 

Upstream recirculation  1.140  1.040  1.080  0.998 
Primary downstream recirculation 
1.670  1.612  1.690  1.432 
Secondary downstream recirculation 
0.166    0.160  0.134 
References
Spalart P. R., Strategies for turbulence modeling and simulations, International Journal of Heat and Fluid Flow, 21 (2000), 252263.
Shah K.B., Large eddy simulation of the flow past a cubic obstacle, Ph. D. Thesis (1998)
Martinuzzi R., and Tropea, C., The flow around surface mounted prismatic obstacles placed in a fully developed channel flow, ASME Journal of Fluids Engineering, 115 (1993) 8593.
Rodi W., Comparison of LES and RANS calculations of the flow around bluff bodies, Journal of Wind
Engineering and Industrial Aerodynamics, 6971 (1997) 5575.