[myaccount]

Hi
I have simulated 2-D flow applying velocity inlet and pressure outlet boundary conditions.Later I applied periodic boundary conditions at inlet making the outlet conditions as shadow zone.After convergence, I see the velocity vectors coming out of the outlet zone.Why is this happening?

[flotus1]

Why should it not happen?

[myaccount]

Thank you for the question that made me to think.I was initially in an impression that all the vectors will be represented in the closed domain.I came to that they need not be .
I have another problem now.
Iam simulating the flow over solids(squares).For this, I created a domain(big rectangle) with solids squares inside.I gave velocity inlet boundary condition at the entrance and pressure outlet at the exit.I have meshed it and transferred to fluent.I have applied periodic boundary conditions at the inlet making the outlet shadow. I have run the case for different grids with convergence criterion of 10-6.As I refine the grid the the curve goes flatter.Why is this happening? Which value is to be taken for further analysis?please find attached a pic of the grid independent study result.

[flotus1]

I have no idea which values you put into this figure, so it is impossible to comment on the results.
And please, attach figures in a file format that everybody can read without much effort, something like .png or .jpg would be nice.

[myaccount]

Thanks for the response.
Please find the following files attached
1.Model with boundary conditions with out mesh
2.Model after mesh generation


a.After mesh generation ,mesh is transferred to FLUENT.
b.Periodic boundary conditions are applied at inlet making the outlet shadow.Convergence criterion is kept at 10-6.
c.After convergence,a vertical rake exactly at the center of the model is created .Velocity magnitude along the rake is exported to excel sheet.
Then the mesh is refined and send to FLUENT and steps (b) and (c) are repeated.I have done this with many(6) times.The velocity magnitude curves go flatter with refinement. If I change my convergence criterion to 10-7,all curves coincide how coarser or how finer I change my grid.
What convergence should I take? Which curve value should I take for my study?
I am confused.Why is this happening?

Sorry,I could'nt send the pics in the format which you mentioned.
Thank you

[flotus1]

If the curves change as you lower the convergence criteria, the solution can not be considered converged.
Instead of comparing the errors introduced by coarse meshes, you compare the errors from insufficient iterations.

My tip for steady-state simulations in an academic context: Run the case until the residuals level out.
Then you can compare the errors caused by the discretization or any other influence you want to investigate.