[riccardo.frezza]

Hi everyone,
I'm trying to simulate a flow over a 3d body but I can't get the convergence of the residuals with the realizable k-e model. The minimum quality mesh is 0.6 and the first element thickness is set to obtain an yplus of 50.
I uploaded some images of the mesh and the residuals. I also uploaded an image of the cd value during the iterations.
What could be the problem? Any suggestion? I've tried to modify the model hundred of times without improving the results. Do I have to refine further the mesh (it has 1.2 milion of elements)?
Thanks

[Roh]

Tell us about the boundary conditions(and their magnitudes), near wall treatment, solver, and discreatizations.

[riccardo.frezza]

Thanks for your reply.
The boundary conditions that I use are:
-velocity inlet : 30m/s
-pressure outlet
-wall : body
-wall with specified shear =0 and simmetry for the boundary
I use standar wall treatment and a pressure-based coupled solver. The courant number is set at 200 as default.

[Roh]

Are you sure about the mesh of boundary layer(size of the first cell next to the wall)?


For first off, Try your solution by S-A turbulence model.
If the solution converged, then check out the Y+.



I saw your residuals and I think you need to be more patient because I don't see any convergence or significant oscillation. Try your solution for more itration.

[riccardo.frezza]

Yes, I've already tried the S-A turbulence model but even in this case I am unable to achieve good convergence. I uploaded an image of the residuals and one of the cd's values. I also uploaded an image of the yplus values in a section of the body that seem ok. Do you agree?

[Roh]

If I assume that everything is ok then probably your mesh is coarse and you need to reduce the size of the mesh. What is the direction of the velocity? from left to right or left to right? I think from right to left. Could you show us the contour of velocity? and streamlines? How did you choose the size of the domain?


Also, tell us about the spatial discretizations that you've chosen. Have tried other discretizations and schemes?



If I were you, I would rather work with S-P and after solving the problem, I would try other models.

[riccardo.frezza]

Yes, the flow goes from right to left as I show in the contour velocity attached.
I choose the default spatial discretization (I've also attached a screenshot of the setting) and I've not tried other discretizations.
I choose the size the domain in the way that its sides were 20 and 10 times the height of the body. Maybe it's not sufficient. I will try to modify my model as you suggested, making a bigger domain.
Thank you for you quick anserwing, I'll let you know.

[Roh]

Originally Posted by riccardo.frezza

...I choose the default spatial discretization (I've also attached a screenshot of the setting) and I've not tried other discretizations...

At first you can use the lowest order of discretization. Then you can reduce the order for "Pressure" and "Momentum".

Originally Posted by riccardo.frezza

...I choose the size the domain in the way that its sides were 20 and 10 times the height of the body. Maybe it's not sufficient. I will try to modify my model as you suggested, making a bigger domain...

Then I suggest to lengthen it to the right and also reduce the size of the mesh behind the Ahmed(?) body where you see a recirculation area.

Originally Posted by riccardo.frezza

...Thank you for you quick anserwing, I'll let you know.

[riccardo.frezza]

I solved the problem coarsening the mesh. Maybe with the fine mesh the solution came into some unsteady fenomena like the wake vortices and the convergence got worse