CFD Online Discussion Forums

CFD Online Discussion Forums (
-   FLUENT (
-   -   Pipe flow with obstacle - HELP (

Min April 8, 2007 19:15

Pipe flow with obstacle - HELP

I have a question about modelling pipe flow with an object within the pipe.

My main question is this: I have been able to create an obstacle within the pipe but I am unable to mesh the pipe volume so that the fluid does not go within the object. Basically, I can mesh the object fine and the downstream flow reacts as if there was an object, but the pipe volume mesh goes THROUGH this object mesh and acts as if it is there but really not there. My grid shows the object as being there and as solid but when looking at flow in the cross-section, it appears that flow is somehow in the object area which is impossible if it is solid. How should I defeat this problem?

Phil April 8, 2007 21:11

Re: Pipe flow with obstacle - HELP
you need to go into boolean operations(volume operations) in gambit and subtract: MAIN VOLUME - OBJECT IN MIDDLE. This means gambit won't make a grid inside the object.

Min April 9, 2007 11:44

Re: Pipe flow with obstacle - HELP
I actually was just told this same advice by someone else and tried it out (the logic seems right) but now have a new problem.

I get two types of errors: ERROR: TG_Mesh_Domain failed with error code 1 ERROR: Tetrahedral meshing has failed for volume This is usually caused by problems in the face meshes.

It seems Gambit has trouble meshing this new volume (the pipe minus the object) and I'm not sure how to remedy this.

Phil April 9, 2007 12:37

Re: Pipe flow with obstacle - HELP
probably due to a complex geometry. you need to do some tutorials and get the basics of grid generation. It seems as if you have no experience of CFD but access to Gambit. If it is important to you then spending afew hours doing some Gambit tutorials will do you infinite good.

Once you have the basics, to make a good mesh in 3d you need to try and make good faces on a 2d plane then extrude it through the model. Then T-grid if this isn't possible as the elements are poor quality. THE TRICK TO ALL OF THIS IS LEARNING TO LINK FACE MESHES THEN USING COOPER MESHING. This is all much much easier to start learning if you do the Gambit tutorials and CONSTANTLY REFER TO THE DOCUMENTATION.

Hope this helps.

Min April 10, 2007 16:07

Re: Pipe flow with obstacle - HELP
Thanks for the help so far, the problem is pretty complicated. You are correct, the geometry is complicated and that's my main problem. It is a twisting fin within a pipe so it's not like a simple solid box that must be subtracted, it's a rotating solid object. I have used tutorials for Gambit before on previous projects and they've helped but in this case I'm a bit confused. I already know what the Cooper scheme does and have used it many times before to get the desired effect to extrude face meshing to volume meshing but this time Gambit can't seem to deal with this twisting geometry. I understand that what I have to do is: 1) subtract fin from pipe 2) mesh faces 3) extrude using cooper

and I have done this before. I can get up to step 2 but extruding using cooper, I keep getting error messages that says "Volume contains a void and thus cannot be meshed using the Cooper scheme". I'm pretty much at a loss at this point even after referring to Gambit tutorials which do not deal with something like this.

Phil April 10, 2007 17:23

Re: Pipe flow with obstacle - HELP
If the fin is rotating then it should be modeled as stationary with the mesh(hence the fluid) rotating. Multiple moving reference frames for steady and sliding meshes for unsteady. You should do the multiple moving reference frames tutorials in Fluent.

The documentation can tell you what boundary conditions, etc you need. The fin can then just be subtracted and modeled as a wall - you DO need to set the fin as a rotating wall but at 0m/s though!

For meshing, you need to decompose the complex geometry into lots of little simple geometries. There are lots of tutorials in Gambit to learn this. It would be worth splitting it up a little then meshing the parts using t-grid to see if you get an acceptable mesh.

Remember to keep a uniform spacing or have smooth transitions between dense and coarse mesh regions. Size functions are good for this.

All times are GMT -4. The time now is 01:19.