CFD Online Discussion Forums

CFD Online Discussion Forums (
-   ANSYS Meshing & Geometry (
-   -   Connectivity problem in Gambit (

Xibalba March 31, 2009 19:35

Connectivity problem in Gambit
I am doing some simulations in FLUENT of heat transfer in electrical transformer windings. I have the mesh in Gambit of the sole winding, and I have done forced convection simulations. The winding consists in solid regions (paper and conductor) with several ducts through it.

The connectivity problem occurs when I want to make the natural convection case simulation, in wich the winding will be submerged in oil, I need to connect the volumes of the ducts with the volume of the oil surrounding the winding, and substract the conductor an paper volumes of the winding. If i substract the volumes of paper and conductor from the tankīs oil volume, the resultant oil volume of the oil isnīt connected with the winding. I tried to extrude the upper faces of the winding volume, and then, define them as fluid regions, and then, creating another volume that represents the remaining oil inside the tank (the tank is rectangular), but when I try to mesh the oil that surrounds the winding, I canīt do it with any mesh type.

I attach some pictures of the problem Iīm trying to analyze. I hope you can understand it and give me some tips to mesh it. The first one is the top view of the winding, i made the 2D geometry, and then extrude it to have the 3D volume. I have that mesh correctly made, the problem is to connect it with the tankīs oil. This winding will be submerged inside a rectangular tank full of oil. The second one is the scheme of the winding and the core around it. That would be submerged in oil.

ŋIs there an easy way to substract the winding volume from the tankīs oil volume mantaining the ducts of the winding (fluid entities) and having a good connectivity in the mesh? I hope i made myself clear with the whole problem.

Thanks a lot.

-mAx- April 1, 2009 02:09

can you provide the dbs file?

Xibalba April 1, 2009 11:33

Hi Max, thanks for the interest in my problem. I do have the dbs file but itīs quite large (>300 Mb). Right now, I solved the connectivity problems with my model, but the problem is that i can not mesh it properly.

I made the mesh for the winding, using a Hex/Wedge Cooper scheme (i first made the mesh for the upper faces). But when I try to mesh the surrounding volume of fluid, Gambit shows an error. I tried to mesh it with a TGrid Mesh and it doesnīt work. Gambit shows the following message:

"Initialization failed to mesh 8 nodes.
ERROR: TG_Mesh_Domain failed with error code 1.
ERROR: Tetrahedral meshing has failed for volume volume.1294
This is usually caused by problems in the face meshes.
Check the skewnesses of your face meshes and make sure
the face mesh sizes are not too large in areas of small gaps"

How can I solve this problem?, do I have to change the mesh of the winding to make sure the mesh of the surrounding fluid can be done with the required element size?

I made a mesh using the Staristep Scheme, but it creates faceted volumes which are not connected with the rest of the model.

Thanks for the help.

-mAx- April 1, 2009 11:51

complex geometries can't be meshed with hexa so simply.
You have to divide your domain into simply volumes, which can be easily meshed in hexa.
The message regarding skewness means that you have small edges or small angle in your domain.
The dbs is 300MB with ou without the mesh?

Xibalba April 1, 2009 12:18

Ok, I am working now in separating the main volume in simplified volumes. Thanks for the tip.

The dbs file is 17 mb without the mesh.

-mAx- April 2, 2009 01:40

ok, I think your winding is wrong constructed.
Thus, if I isolate alle the volumes included in this big "elbow", and if I try to unite all the volumes, gambit fused all the volumes and generates a big "elbow" without any crack.
But I assume there should be crack where fluid may flows (between conductor layers).
In other words you should modelled just the fluid domain, and substract the solid one.
I think you have both in your model.
Return and sketch your 2d profile with substracting the solid section (enable shade for checking the substract operation), then extrude, etc...
Maybe I didn't understand the geometry :o

Xibalba April 2, 2009 19:18

Hi Max. Thanks a lot for the tips. About the elbow you are talking about, thatīs the core of the transformer, which wonīt be meshed for this analysis.

I am working on simplifying the volume of the tank, extruding faces that I already meshed until I can form the total volume. I donīt know if thatīs the fastest or best method, but I am ensuring the connectivity in the whole mesh.

Thanks for the help!

-mAx- April 3, 2009 00:42

OK if you don't need to mesh it, then your model is quite simple to mesh.

All times are GMT -4. The time now is 21:36.