[Pavelko]

Hello All !

I want to simulate a flow in a rigid pipe using CFX .
I've created a solid pipe with some finite thickness .

When I export this geometry to ANSYS DesignMesh it makes the mesh for the solid (between the inner and the outer radius : Rin<r<Rout).

What should I define (surfaces/edges etc.) in order to mesh the inside of
the pipe (for r<Rin) , that is to say to mesh the fluid ?

Thanks ,
Pavelko .

[BrolY]

I would recommand to use two bodies : One for the solid part and another one for the fluid part.
If you are meshing a pipe, you should create one O-grid to better capture the geometry. I think there is a tutorial about that !
Good luck !

[Pavelko]

Can you please send me a link to thisw tutorial ? I can't find it .

[Pavelko]

And another question :
----------------------
When I create the inner full cylinder with radius r=Rin , which supposed to be a fluid ,
and another hollow cylinder with Rin<r<Rout , which supposed to be a solid ,
the DesignModeler defines it like ONE body . Do anyone have an idea how can I solve this problem ?

Thanks,
Pavelko .

[PSYMN]

You need to use "Add Frozen" when you create the second geometry... (instead of Add Material), or Freeze the first geometry before extruding the second sketch...

Then if you want the mesh to be conformal between the bodies, you need to combine them into a multibody part.

[PSYMN]

Or you could use the enclosure tools and use the pipe to enclose the fluid inside...

[Pavelko]

Thank you ! It works

[palofpinaki]

Hey!!

I am doing FSI simulations using ANSYS mechanical and CFX. My geometry is that of an axisymmetric abdominal aortic aneurysm which is not perfectly cylindrical(there is a bulge in the middle which is defined by a curve). I have created the geometry of the wall with a constant thickness in solidworks. But after importing it in ANSYS designmodeler, I am not able to define the fluid body inside the aorta. ......I would appreciate if somebody could help me in figuring this out. How do i get two different bodies, one for solid and another for fluid in ansys 12.1 designmodeler.

[PSYMN]

Palofpinaki, There are several ways to do this in DM...

One very simple way is to use Tools => Fill.

You can set the extraction type to "By Cavity" or "By Caps".

If you select "By Cavity", it is asking you to select the inner faces of your vessel... just select the inner wall. If you have multiple faces, hold down CTRL to multi-select. Then hit apply and then Generate...

If you select "By Caps", it is expecting that you have an assembly of parts that form a closed volume... It will extract that volume for you. You could use this if your pipe ends had pugs or something like that.

After you generate your fil, you may not notice the new body right away because it is a transparent fluid body, but look in the model tree and you will see it. Click on it and it will appear solid.

I assume you are keeping the thickness of the wall because you want to do some sort of FSI... If you leave the bodies as separate parts, you will get two separate meshes (one could be fine CFD mesh, the other a coarse FEA Mesh, etc.). If you would rather have the meshes share nodes, you must combine the two bodies into a single multi-body part. Just CTRL and select the bodies in the tree or on the screen and right click to "Form New Part"

[palofpinaki]

Thanks PSYMN ...... it works. I also found another method to do that by using 'FREEZE' (which is a bit more time-consuming as I'd need two separate geometries then)...so that the two bodies are different.
But didn't know about the "multibody part" thing...and yes I am doing two-way coupled FSI simulation for a 2-D axisymmetric case.

I do, however, have some more difficulties/doubts regarding my simulation which are as follows :
For my case, there are two boundary conditions in the fluid domain at the inlet and at the outlet respectively.
Inlet: time dependent fully developed turbulent velocity( i.e velocity is different at different radial locations except that it is time varying also)
Outlet: Time dependent normal "traction".
1) where should I input these b.c's: in the fluid part or the transient structural part ??
2) and how can these b.c's be incorporated for analysis?
3) Are there any special precautions(or better ways) that need to be followed while meshing the geometry for FSI ?
I would really appreciate some help.
Also, I wasn't able to find good examples of FSI in ansys 12.1 except the "oscillating plate" one. If somebody has better resources, please help....
...

[PSYMN]

Create named selections for the inlet and outlet to the fluid body...

Mesh as normal and then go to the solver...

In the solver, you can apply what ever bocos you need on these named selections...

As for special meshing tips for FSI... Will you be morphing or remeshing? Will the deformations be large or small? Basically, you just want a mesh that will be robust enough to handle the deformations...

Best regards,

Simon

[palofpinaki]

The deformations are gonna be large...
I am trying to use euler-lagrangian formulation to describe fluid motion in the FSI model....
Hey Simon.....
I've also created a solid-fluid interface which I've defined to be the inner surface of the solid geometry...
Should I also create a named selection for the interface and then put bocos for the interface....
in that case where do I need to put these bocos... in the transient structural(ANSYS) part or Fluid Flow(CFX) part??
Actually, I am confused about the steps to be followed for performing two-way coupled FSI....like where to start from:structural part to fluid or vice-versa.....how and where exactly to put the boundary conditions...
I am using ANSYS for the first time.

[PSYMN]

Sorry, I have not done a 2 way FSI for a long time and the method I last used is hazy in my mind and probably out of date. I do 1 way FSI regularly because it is so easy to setup in WB. Best to ask questions about that on the solver forum... Named selections are aways a good idea because they give you a handle to hang on to...

Hexa is great for deformations because you can generate the same mesh (same degrees of freedom) for each iteration, even though the shape is different. You may also want to look into rbf-morph (a 3rd party morpher that hooks up seamlessly in Fluent). Remeshing with Tetra is ok if you have no choice, but the frequently changing mesh does give some static in the results.

[Prakash.Paudel]

I want to model a solid pipe that is curved and i know the equation of that curve. I need this solid model to simulate fluid flow through this pipe. what is the best way to create this solid pipe.

This is what i did:

I imported "formatted point data" in design modeler and created a curve on a plane using these data points. I sketched a circle on a plane normal to the first plane and I tried to sweep this circle along the curve but design-modeler wouldn't let me select this curve as sweep path ( I suppose the reason is that the sweep path has to be a sketch and not a curve) .
Does anyone have solution to this problem?? Thanks for any help

Prakash

[sihaqqi]

Originally Posted by PSYMN

Palofpinaki, There are several ways to do this in DM...

One very simple way is to use Tools => Fill.

You can set the extraction type to "By Cavity" or "By Caps".

If you select "By Cavity", it is asking you to select the inner faces of your vessel... just select the inner wall. If you have multiple faces, hold down CTRL to multi-select. Then hit apply and then Generate...

If you select "By Caps", it is expecting that you have an assembly of parts that form a closed volume... It will extract that volume for you. You could use this if your pipe ends had pugs or something like that.

After you generate your fill, you may not notice the new body right away because it is a transparent fluid body, but look in the model tree and you will see it. Click on it and it will appear solid.

I assume you are keeping the thickness of the wall because you want to do some sort of FSI... If you leave the bodies as separate parts, you will get two separate meshes (one could be fine CFD mesh, the other a coarse FEA Mesh, etc.). If you would rather have the meshes share nodes, you must combine the two bodies into a single multi-body part. Just CTRL and select the bodies in the tree or on the screen and right click to "Form New Part"

Dear Simon

I am doing my MS by research in flow induced vibrations and as such I have to first perform LES to get results, then FSI to determine mode shapes. My project is "Determination of zones doownstream of a tee junction where cicumferential mode of vibrations change to lateral mode." For this, I have following queries:

1. Can you please advise why one can not do 1 way FSI by modelling the way you have described in your answer below in red. Can we expect some difference in results if we model the way you have descried below against the one in which we make a cylinder and use fill command. I am finding it easy compared to making a pipe with wall thickness as my geometry is a 90 degree mitre bend pipe configuration and at the intersection I am constantly having trouble. I have two errors as attached in fig 1 and fig 2. Graphically the second option is much better but from analysis perspective, do results change. I am very new to Ansys and this is my first time with FSI so please guide,

"You need to use "Add Frozen" when you create the second geometry... (instead of Add Material), or Freeze the first geometry before extruding the second sketch...
Then if you want the mesh to be conformal between the bodies, you need to combine them into a multibody part. "

2. For this what strategy should I adopt, once I finish LES, should I directly take the model in Modal Analysis or do I first use STructual Analysis and then join it with Modal Analysis. Also what benefit can I get if I use Structual analysis option instead of using Modal analysis directly. Since you have good experience modelling FSI, I shall be grateful for your guidance and patience towards advising me.

I shall be grateful for your response.

Best Regards