[akalopsis]

Hey Guys,

I'm getting the following error when I try to run an analysis with a rotating solid inside of stationary fluid domain:

ERROR #555000003 has occurred in subroutine CS_COMPUTE.

Message:

GGI intersection error. For a particular boundary face on domain interface:

Fluid-Cylinder

The intersection failed to find any pixels. Likely causes are:

1) Very large differences in grid density between the two sides of the interface. In this case, increase the value of the expert parameter 'ggi bitmap resolution'.
2) An interface contains a degenerated face. In this case, also increase the values of the expert parameters 'ggi bitmap resolution' and/or 'ggi edge scale factor'.
3) A sliding interface has one or more nodes with zero radius. In this case, modify the grid to remove these nodes.
4) The interface contains some faces that are parallel to the rotation axis and others that are perpendicular. The transformation algorithm does not permit both of these face types to coexist in the same interface.

I've tried following some of the solutions that I have found online but none of them seem to really apply to my situation. I have attached an image of my setup. The small cylinder at the top is rotating about the z-axis. It is defined as a solid, while the surrounding fluid (air) is stationary. I have the interface between the two of them set to Rotational Periodicity about the z-axis again. Mesh connection is set to automatic.

Any help would be greatly appreciated.

Alex

EDIT:

For anyone who cares, I seem to have fixed the problem. Instead of using a rotating domain for the cylinder, I changed the cylinder from a solid to a fluid domain so that I could specify the surfaces at the interface as rotating walls (in both fluid domains). The solution has just begun to run, but I'm no longer getting any errors, so hopefully this will work.

[ghorrocks]

You do not seem to understand what a fluid and a solid domain should be, and what a rotating domain means.

You only model a solid domain if you want to model something (usually heat transfer) inside the solid. If you do not want to model heat transfer in the solid then you do not model the solid. It is modelled as wall boundaries on the outside of a fluid boundary.

Also this entire model appears to be able to be modelled in a single rotating frame of reference with no need for interfaces. You will need to know how to set up boundary conditions in rotating frames of reference to do this.

Have you done the tutorial examples on rotating machinery? It shows you how to set up rotating frame of reference simulations.

[akalopsis]

Yes, I believe you are correct. Since I'm still getting an error related to the rotation even after making my previous changes, I will look into the boundary conditions in rotating frames of reference. I may have more questions if my problem persists after trying the tutorial but, as always, thanks for the help Glenn.

[akalopsis]

I have looked at the tutorial examples dealing with rotating frames of reference and applying boundary conditions in a rotating domain, but I'm afraid I'm just not fully getting it. I just want the side and top of the small fluid domain to be rotating. It seems like it should be simple but everything I try gives me a different error when I run the solver.

Could you point me in the right direction in terms of setting up the boundary conditions in a "single rotating frame of reference"? What domain do I set as rotating: the surrounding fluid, the small fluid cap, or both? And if the domain(s) is set as rotating at a given angular velocity, how do I apply my desired angular velocity to just those two faces mentioned above?

Sorry for all the questions, I'm still trying to familiarize myself with CFX.

[ghorrocks]

Can you show on your image which bits rotate and about which axis, and which bits do not rotate?

[akalopsis]

Glenn,

I don't have access to CFX on my current computer, but I will try to get a better image for you. Basically, the heat sink is sitting on top of, and around, a shaft housing. I have defined the interior of the housing as a separate part so that I can apply the boundary conditions I want in CFX-Pre, namely a heat generation at the bearings. On top of this interior fluid, I have placed a small cylinder, which I want to rotate about the z-axis at a given angular velocity. Ideally, I'd only like the top face of cylinder and the side to be rotating, but if it's easier to make the entire cylinder rotate, that will work too. Lastly, the entire assembly is enclosed in the cylindrical fluid domain as shown in my first image. The top and side of the enclosure are openings and the bottom is a wall.

Nothing should be rotating except for that small cylindrical cap. My goal is to see how the surrounding fluid (air) behaves as a result of this rotation as well as the heat generation from inside the housing.

Thanks.

[akalopsis]

Here is an exploded view of the assembly, so that you can clearly see the individual components.

[ghorrocks]

If the central cap rotates then you have no need for rotating frames of reference. This can be modelled with a tangential wall velocity, with all the mesh in a single stationary frame of reference.

[akalopsis]

All domains are stationary and I have applied a wall velocity to the central cap (in the fluid cap domain as well as the surrounding fluid domain). I defined the top and side faces as rotating walls about the z-axis, which should run right through the middle of the cylinder. Now I'm getting an error along the lines of: "The specified velocity vector on the boundary patch back has a significant normal component at one or more faces."

Is this because the side face of the cylinder is parallel to the z-axis? If so, how do I define the rotation of the side face? I would like the cap to rotate clockwise about the z-axis.

[ghorrocks]

Can you post an image where you clearly show what is rotating and what is stationary? I do not know what your "cap" is. The exploded view shows to components, but it is not labelled and I do not know what is rotating.

[akalopsis]

Sorry for the confusion. When I refer to the "cap", I am talking about the small cylinder that I want to rotate (see attached figure). I also included the axis and direction of rotation in the figure. The axis coincides with the global z axis in CFX. I have left the boundary conditions that I refer to in my previous reply the same for the time being, as nothing I change seems to do any good. Please let me know if you need more information and thanks for taking the time to look into this.

[akalopsis]

It turns out the global z-axis was not perfectly aligned with the middle of the cylinder. I created a new coordinate system and specified rotation about the new z-axis. This appears to have done the trick.