Axisymmetric 2D mesh
 

Hi,

I have been doing 2D simulations of a axisymmetric geometry, but the outer flow field is not symmetric, and furthermore, the directions of inlet & outlet are set perpendicular to the axis （x axis）.


I' d like to know whether I can use the axisymmetric model.

Thanks.

few pics would certainly help to pin-point problem...

Thanks for the replay. I've sent two pics with bc.:D

Attachment 20380

Attachment 20381

there is no outlet?

outlet bc is the right side

You mention that you want to use axisymmetric model. In this mesh, where is the axis of revolution?

If I am not missing something, it is not possible to use your mesh as axisymmetric since all the edges are given boundary conditions. You can't revolve around an edge with a boundary condition since that edge would be an axis (line) in the real geometry and a line can't be an inlet or outlet for most of the cases.

OJ

half model may serve as axi-symmetric...

I'm intrigued and a bit surprised. If you take a case of FLUENT, you can't have any domain below X-axis, in axisymmetric case. How would you create a half axisymmetric planar model with X-axis at the center, which would have half of the domain below X axis, generating negative volume when revolved around X axis?

More importantly, why would one bother to do that, since it would suffice to keep a single planar domain with axis at one of the edges?

Or are you talking about a half 3D model? That would be symmetric about the middle plane, but not axisymmetric since it may have gradients in theta-direction.

OJ

OJ

I was not talking about the this particular model.

If we can rotate this model -90 about Z-axis (Z out of screen) and translate so that this model is on + x-axis. and X-axis is crossing through centre of this model. After that we can delete lower part (-Y) and model it as axi-symetric.

what do you think?

OJ

I was not talking about the this particular model.

If we can rotate this model -90 about Z-axis (Z out of screen) and translate so that this model is on + x-axis. and X-axis is crossing through centre of this model. After that we can delete lower part (-Y) and model it as axi-symetric.

what do you think?

I notice one thing that flow is not axi-symtric! May be you can try with quarter model or half.

Neither was I. I was just commenting on your earlier comment that half planar model can be considered axisymmetric, which doesn't seem plausible :)

It is still not entirely clear where his X-axis is. If we assume his X-axis is vertical and passes through middle of his wallled voids, it is evident that it is not even geometrically symmetrical about X-axis.

Moreover, if you have a look at his boundary conditions, he has three inlets at three edges of rectangle (top, bottom, left) and the fourth edge (right) is outlet. This is strange. This can not even be a cylindrical representation as there can not be an axis and and an axial flow. Only 3D domain he can represented using this is a cuboidal one, with his current 2D plane specified as symmetry.

Sieben, can you confirm if boundary conditions are right and specify location of X, Y axis? Also is the final domain cylindrical or cuboidal?

OJ

OJ

You comments are logical and correct.

Hi, thanks for the reply.

Maybe I‘ve given an ambiguous description. I've sent two pics again.

I wonder whether I can use the 2d axisymmetric model when the inlet & outlet boundaries are vertical to the axis.

Thanks again.

Attachment 20514

Attachment 20515

It looks like a turbine secondary air passage...

I think you can try it in fluent...

Alright, something fishy. If your top condition is outlet, left and right being inlet, the axisymmetric assumption doesn't seem to hold.

Visualize this - your actual domain is a cylinder with an axis as the one you specified and the revolution section being your domain. Going by your current configuration, your cylinder will have flat surfaces as inlets and curved surfaces as outlet! It is not bounded. IF I am not missing something, I can't imagine how it would fit in the axisymmetric assumption.

OJ