CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   Definition of fluid-solid interaction (

syens December 21, 2011 11:13

Definition of fluid-solid interaction
Hi all,

I want to model the interaction between a fixed rigid body (e.g., a sphere) and a fluid.
From my weak experience, I identified two methods:

Englobing domain: you define in DesignModeler an englobing geometry around your sphere. (Ref:

Cutted domain: you define in DesignModeler, two superposed solids (sphere+box) and after you remove (or substract) the sphere from the box.
(Ref: see: Blunt Body)

What is the best way to model this kind of interaction for you ?
Have you some exemples or references to give me ?

Thank you

ghorrocks December 21, 2011 18:03

The answer to this question depends on what you are modelling.

How big is the sphere? How strong is the flow? What physics are you modelling - just the motion of the sphere or other stuff as well? Does the sphere have mass? Transient or steady state? If transient, then on what time scale?

As you can see there is lots we need to know before we can answer your question.

syens December 22, 2011 03:55

Hi Ghorrocks,

Thank you for your reply, I agree with your comments. I would have had to give more information about my simulation, I thought there was a general way to model the fluid-rigid body interaction.

Concerning my transient simulation (duration 40s), I would like to visualize the gas flux passing through a structure having holes with complex geometric shapes (rigid body).

Best regards

ghorrocks December 22, 2011 05:43

There are several methods of modelling this and which one is best depends on all the questions I asked.

If the flow is through some sort of structure then you have better show an image of the structure and an idea of what the gas velocities are.

syens December 22, 2011 06:11

3 Attachment(s)
I understand what you mean, but I cannot post this geometry on the web (sorry !!!). Otherwise I loose my job.

However, I am testing the method with the englobing domain. When I try to apply the boundary conditions, I see that the fixed rigid sphere has two layers as you can see on the 2 attached files.

Please, anyone can explain me why I have these two surfaces to define the sphere.

Thanks in advance

ghorrocks December 22, 2011 07:36

If the exact geometry is confidential then post something like it. Without context this conversation is pretty meaningless.

You have two surfaces because you meshed the sphere as well, so one mesh on the spherical face in the outer block and one mesh on the spheres outer face.

There are three main ways this can be modelled - rigid body solver with moving mesh, rigid body solver with immersed solid or with particle tracking. You still have not given me any information to put this in context so I cannot comment on which is most appropriate.

syens December 22, 2011 08:42

2 Attachment(s)
You can see the design of the holes and the thickness of the part in the attached files. I try to visualize the gas flowing through this part.

Concerning the rigid sphere simulation, in fact when you have a rigid body (i.e., a solid) and an englobing box, you have 2 solids in geometry section.
Therefore, you are obliged to mesh these 2 solids.

If you want to represent the rigid body geometry in the CFD simulation, do you always have to do a mark (i.e, a footmark by a boolean operation) in your domain ?

Best regards

ghorrocks December 22, 2011 17:01

The different rigid body approaches require different meshes.

Deforming mesh rigid body does not require the body to be meshed, it needs to be a cut-out in the fluid domain.

Immersed solid rigid body requires the fluid mesh to cover the entire domain (ie including the volume originally occupied by the body) and the body to be a surface mesh.

Particle tracking requires the fluid mesh to cover the entire domain and does not need a mesh for the body.

So there is different mesh requirements for each approach, and what you have done matches none of them.


If you want to represent the rigid body geometry in the CFD simulation, do you always have to do a mark (i.e, a footmark by a boolean operation) in your domain ?
I do not know what you are talking about.

All times are GMT -4. The time now is 06:08.