Hi everyone!
I am trying to simulate flow past a square cylinder at 2100 Reynolds number. I have tried very fine mesh and almost all the schemes and tried Laminar as well as k-w SST model but each time getting very high values of coefficient of drag around 6,7 or even 10 while it is expected to get 2.124. Can anybody please refer to some work or suggest something that which problem may get caused by this?
Anxiously waiting for answers
Best regards
Hi
Some time the CFD software can not estimate the drag coefficient correctly. This coefficient is a function of freestream velocity and other parameters. These parameters must be defined for the software.
You can export the velocity field and calculated the drag coefficient by formulas. For more explain call me on 989149180988
First, I'd suggest to start and compare with a laminar case, here're some publications regarding the drag coefficients at low Reynolds number (40 for example), you can use the exact computational domain and see if you're close to their results, may be it's just a post processing error:
- Article Flow past a square cylinder at low Reynolds numbers
- Article Heat and Fluid Flow across a Square Cylinder in the Two-Dime...
- Article Effects of Reynolds and Prandtl Numbers on Heat Transfer Acr...
Second, for high Re, I'd check the non-dimensional wall distance next to the square walls (y+), and make sure it's less than ~50 if you're using k-epsilon or k-w SST (Fluent recommends 10 cells at least for the boundary layer). If not, you can decrease mesh sizing through controlling the Inflation (you have to use it anyways if you use tetrahedrons, that you make sure that the mesh near the wall region is quad/prism not triangles/tetrahadron).
Hope it was useful.
Mohammad Mobarak Thanks for your kind reply. I have checked and validated with all the same conditions for Re=100. For the same conditions, I am applying the K-W SST model at Re=2100 and it is giving so much weird results for square as well as for all my chamfer ratios. I am implementing 0.005 first layer thickness and thus Y+ = 0.7 corresponding to Re=2100.
Hope I will get a good suggestion from you.
Muhammad Ahsan Nawaz Where do you evaluate your drag coefficient, is it on fluent or CFD post?
Some common mistake can happen that I forget to change the reference values on Fluent which is used in the Normalization (Non-dimensionalisation). May be this is the case.
Some deviation (though less than what you have) can be influenced by the inlet turbulent intensity. Which is also the case in experiments depending on the type of grid used in the upstream, may be you check that too.
Mohammad Mobarak I am reporting drag force and drag coefficients prior to give my simulation. I have calculated manually using drag force as Dr. Tohid Abidi suggested and also I took an average of Cd that is calculated by Fluent. For both cases, I am getting a huge error.
In regard to reference values, I give '"compute from the inlet" and change nothing.
Your last point I could not understand clearly, means I should cluster the grid on upstream? I am using an O-type grid around the body and coarse otherwise with the parameters I told earlier.
Muhammad Ahsan Nawaz
For the reference values, it's not enough to put the inlet conditions, you have to also put the area (S) of the square. CD=2D/rho*v2*S where S=height*depth
By grid, I meant the experimental one in the wind tunnel. So, in experiments, they can change the configuration in the upstream to control the turbulent intensity. In simulations, you can change the turbulent charactersitics of your inlet, in order to somehow mimic the experiment.
Other questions, do you solve with constant density/incompressible/compressible?
Do you have 3D or 2D domain? Are you far enough from the walls and have the same configuration as the experiment?
- Badges
- Science topic