Problem with usproj for Drag coefficient computation

[Yvan Fournier]

Hello,
It seems you added usproj.f90 to the list of user input files, but this list is for files in the DATA directory only, so there is no reason to add it there. Files in SRC are accounted for automatically.
  Yvan

[Jean-Marc Blanquies]

Hello,
 
Thanks Yvan you're right i added usproj.f90 in the user file list. I'm really new with subroutines but with your help i'll not do the same mistakes in the future.

By the way, now i have bad results for the drag coefficient, in my simulation i compute a 10⁴4 reynolds but i find a value near 0.4-0.5 instead of near 1.2 according experience.

I checked the value of Uref and ro0 and it's ok.

I check if it's an issue of the grid, but u+ seems ok and twice the number of cells don't change the results at all.

Results with k-epsilon and k-epsilon linear production with scalable wall function are closed.

So i don't understand what's wrong.

I attach my mesh, xml file and usproj.

Any idea ?
 
Best regards,
 
Jean-Marc

[Jean-Marc Blanquies]

It's seems to be an error in the drag and lift coefficient. There is a Pi ratio between experimental results and my numerical results.
 

[Yvan Fournier]

Hello,
Did you test the sensibility to mesh refinement or turbulence model ? In many cases, obtaining a good drag coefficient may require a rather fine mesh.
Also, your sum using forbr is a total force, so I would expect your coefficient calculation to include a surface in the denominator, but this does not seem to be the case. Is this normal ?
Best regards,
  Yvan

[Jean-Marc Blanquies]

Dear Yvan,
 
As i wrote in a previous message i have tested the mesh sensibility, and refine the mesh by a factor 3 or even 10 didn't change the results at all.
Same results with different turbulence model, gradient calculation method or numerical scheme.
Same results for steady or unsteady calculation.
The calculation converged.
I think that the problem was in my Cd formula.
My first formula for Cd was :
xcof(II) = xfor(II)/(0.5d0*ro0*UREF**2*0.025d0 * 0.025d0 * 3.141d0)
I thought that for an infinte cylinder the area was Pi*R² (where my radius is 0.025 m)
When i use the next formula using R² as area i obtain good results
xcof(II) = xfor(II)/(0.5d0*ro0*UREF**2*0.025d0 * 0.025d0)
By the way i'm not sure that it's the right formula formula as i think it should be :
xcof(II) = xfor(II)/(0.5d0*ro0*UREF**2*0.001d0 * 0.05d0)
where 0.001 m is my z thicknes and 0.05 my diameter.
 
Best regards,
 
Jean-Marc
 

[James McNaughton]

The reference area you should use is the bluff area, i.e. diameter x thickness. The third formula you list is correct.
James

[Jean-Marc Blanquies]

Hi James;
 
So finally there is a problem in my calculation. Indeed, with this formula i find a value of 15.5 at Re=10^4.
 
I will refine my mesh again and see the results.
 
Thanks a lot.
 
Kind regards,
 
Jean-Marc
 

[Jean-Marc Blanquies]

Hi,
 
I've tried to improve the mesh by refining it, but nothing changes. I still have wrong values for Cd (no change in the value given in my previous post).

Is anyone have an idea of what could be the origin of this problem ?

My refined mesh :
http://dl.free.fr/getfile.pl?file=/25aIZ6e2

Thanks.
 
Best regard.

Jean-Marc