[oci95]

I'm running a study of a simple curved plate, I've created a structured mesh with ICEM, imported in fluent.
I think the mesh is enough fine (see the attachment) and shouldn't be the problem. (The plate is not visible because it's a simple curved line and it coincides with an edge of the mesh, anyhow the foil is exactly one of the curved lines)

Mesh.PNG


Initially I've run some 2D, steady, laminar, pressure based simulation but the continuity residual converge at about 1.5e-3 and I think it's not enough.
I've looked for on the forum and I found out that I should run a transient simulation. I've done it with time step of 0.001s but the result is the same: the continuty (white line on the attachment) converge at about 1.5e-3 and it doesn't go down.


convergence.PNG


I don't know what else could I do.


Re is 20700, but also with Re=2133 the result is the same.


Bounday conditions are:

• Velocity inlet
• Pressure outlet
• Periodic (on top and bottom of the domain)
• Wall (for the foil)

I've try also to use a turbulent model but the result is the same.

[sufjanst]

Maybe you should use turbulence models.

If you have a turbulent flow due to high velocities, the laminar "model" will hardly be able to give you a stable solution.

If I remember correctly, people use the laminar "model" with a extremely fine mesh to do DNS because the turbulence is directly calculated and not modelled which is clearly not what you are trying to achieve.

[oci95]

Originally Posted by sufjanst

Maybe you should use turbulence models.

If you have a turbulent flow due to high velocities, the laminar "model" will hardly be able to give you a stable solution.

If I remember correctly, people use the laminar "model" with a extremely fine mesh to do DNS because the turbulence is directly calculated and not modelled which is clearly not what you are trying to achieve.

I forgot to write that the Re is 20700, which should be perfectly laminar flow.
Anyway i've try to use also a turbulent model but the result is the same.

[Janet]

Originally Posted by oci95

I'm running a study of a simple curved plate, I've created a structured mesh with ICEM, imported in fluent.
I think the mesh is enough fine (see the attachment) and shouldn't be the problem. (The plate is not visible because it's a simple curved line and it coincides with an edge of the mesh, anyhow the foil is exactly one of the curved lines)

Attachment 71788


Initially I've run some 2D, steady, laminar, pressure based simulation but the continuity residual converge at about 1.5e-3 and I think it's not enough.
I've looked for on the forum and I found out that I should run a transient simulation. I've done it with time step of 0.001s but the result is the same: the continuty (white line on the attachment) converge at about 1.5e-3 and it doesn't go down.


Attachment 71789


I don't know what else could I do.


Re is 20700, but also with Re=2133 the result is the same.


Bounday conditions are:

• Velocity inlet
• Pressure outlet
• Periodic (on top and bottom of the domain)
• Wall (for the foil)

I've try also to use a turbulent model but the result is the same.

Increase the time steps maybe? To 0.1? And try. Changing the Courant number may also help converge. For coupled-explicit solver, Stability constraints impose a maximum limit on Courant number, and it cannot be greater than 2. Default value is 1. Reduce Courant number when
having difficulty converging. But sometimes even this does not help, you can try. For implicit, it is 5, which can be reduced too.