I am using Fluent to calculate the flow field of a circular cylinder in crossflow at 500<ReD<10000. In this range of Reynolds number, the boundary layer will be laminar before separation but the wake area will be turbulent. I tried both turbulent modeling and laminar modeling in a 2-D simulation.

The turbulent model I used is standard k-epsilon model, with enhanced wall treatment. This model works well at low Reynolds number and the drag coefficient is 1.13 at ReD=600. But when I increased my Reynolds number to 5400, the drag coefficinet decreased to 0.76.

If I do not use any turbulent model and simulate the flow as laminar, the drag coefficient is 1.46 at Re=5400. The strouhal number is about right (0.24)

Both values are seriously deviating from the experimental value of 1.0. Can anybody help me and tell me what to do?

hi, Can somebody give some comments on this question? I am also interested in what happens. I also like to know the following

Did you check where the flow separates and whether the location have good agreement with experiments or DNS? What boundary conditions are you using? Did you solve the boundary or use any wall function? what kind of grid are you using?

Thanks.

Jun

The flow separates around 90 degree Re beween 2500 and 5400. The separation angle reduces from 100 to 85 when Re increases from 600 to 5400. The DNS seems to give better separation angle than the tur modeling but I do notice that the friction coefficient is not symmetrical on the upper half and bottom half cylinder due to the asymmetrical vortex shedding.

The boundary conditions are uniform inlet flow, with 1% tur. intensity and 0.7D tur. length scale for tur. modeling. I selected the enhanced wall treatment, which incorporates both laminar and log. law of the wall functions.

Quadratic grids are applied in the area near the cylinder surface to resolve the velocity profile inside the boundary layer. At the other are, unstructured triangular meshes are applied.

Hope this can help and I am eager for some response. Jun, how are you modeling it? Thank you very much!

I use a UDF program of customing drag in the mixture model. It is a program from the FLUENTUSERS.com and everything is ok and I compiled it successfully. the error occures immediately after I begin to iterate,It shows in the control window that : Updating solution at time level N... done. iter continuity x-velocity y-velocity z-velocity k epsilon uu-stress vv-stress ww-stress uv-stress vw-stress uw-stress vf-bub time/iter 99573 5.2328e-02 9.0766e-05 5.3900e-05 3.6095e-05 1.1646e-03 3.2961e-03 4.0613e-04 9.8431e-04 3.6854e-04 1.1754e-03 2.6536e-03 1.8777e-04 1.0813e-04 0:00:00 50

reversed flow in 4 faces on pressure-outlet 8.

reversed flow in 1382 faces on pressure-inlet 10.

Error: get_udf_function: function mzslip has wrong type: 3 != 4 Error Object: ()

what's is the mean of 'get_udf_function'? mzslip is a udf function in my UDF. If someone interested in it, I will send you the program.