CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   FLUENT (https://www.cfd-online.com/Forums/fluent/)
-   -   gambit (https://www.cfd-online.com/Forums/fluent/28667-gambit.html)

lucy July 3, 2001 13:07

gambit
 
I am a beginner using gambit and fluent. I want to make a model of three pipes intersection using gambit .The angles between the axises of the parent and two daughters are both 35 degree. The two daughters are 70 degrees.They are symmestry to the parent.The diameter of the parent is 24mm and the diameter of the two daughters are both 12mm.The intersection is a big problem to me.I can't make a smoothy intersection.There are so many very small faces and very sharp angles on those faces in the intersection that I couldn't mesh good engough to calculate.I used the TGRID mesh. Does anybody know how to make a smoothy intersection of the three pipes? Thanks a lot. lucy

Harry Qiu July 3, 2001 20:55

Re: gambit
 
Hi: After you create your geomitry model, can you start your iteration? Harry.

Alain July 4, 2001 06:49

Re: gambit
 
Hi lucy,

I had a similar problems few years ago. The intersection between pipe gave me headaches.

If you have the opportunity make your geometry with a real cad package which will give you a clean geometry and then import it into gambit.

Try to merge the small faces into the larger ones with face-merge

Try to heal your volume

Best regards.

lucy July 4, 2001 10:31

Re: gambit
 
Thanks a lot. Now I am calculating the model using fluent.But the results are not realiable. I have checked my mesh skewness.There are about 0.04% volume meshes above 0.8 skewness.My total mesh is 46434.There are 0.09% face meshes above 0.8 skewness.I can get large intersection faces but the very sharp angles of these intersection faces is a big problem when I doing calculate.I guess these sharp angles can't cause the results converge to 1e-6. I heard from other person that import the model from other software such as CAD and I-deas maybe rise other calculation problems,so I haven't tried this method.Maybe I can try it. Because I have a dirty mesh model,I am not sure of which is the true cause in calculation problems. How did you solve your problems? I also want to attach my picture in this letter,but I don't know how to do?Does anybody know? Best regards

Eric July 4, 2001 18:50

Re: gambit
 
I would not use CAD data for a simple pipe junction. Note that this is a tutorial example in the Fluent tutorial manual (available on their site as well) for a 90 degree junction.

Basically what I would do is:

Draw all 3 lines as separate volumes. Make sure the daughters overlap the big pipe and that the centerlines of the daughters cross each other in the centerline of the parent.

After this split the daughter volumes laying outside the parent on the wrong side (if there are undesired stick outs, depends on the how far they extend beyond the parent centerline ?), through 'split volume' daughter with parent outer face.

Delete the undesired volumes after the split.

After this unite the 3 volumes together to one volume and apply a tet mesh. Since your problem is only 46K cells, use of symmetry is not really necessary.

If you would have a large problem you would have to cut the model in 2 with a plane and apply a sym BC to halve the nr of cells.

If you solved this once, next time it is really peanuts to make a junction.

Best regards,

Eric

lucy July 4, 2001 19:56

Re: gambit
 
Thank you for your help.

But I don't understand what you mean exactly.

I can't split the daughters using the parent outer face because it has upper topology.Gambit would give me "Error".

Where is the volume laying outside of the parent on the wrong side? What the daughters overlap the big parent means? Do you mean that the two daughters intersect inside the parents and then split the inside parts of the two daughters,and then Boolean these three volumns?

Best regards

lucy

David July 6, 2001 06:19

Re: gambit
 
Hi,

I have made you a Fluent msh file and sent it to

ftp.icemcfd.com

username ftp password youremail address

cd outgoing/minns

get fluent.zip

It`s a structured HEX mesh with full O-MESH topology through all pipes. You may find it useful as a guide line for future models. The mesh was made in ICEM CFD HEXA and took 2-3 minutes to make. HEXA doesn`t care about small surfaces and gaps so there is never any CAD clean up to do.

Please let me know if you have any questions.

Regards

David www.icemcfd.co.uk

lucy July 6, 2001 11:34

Re: gambit
 
Hello,

Did you send it to my e-mail address x_lucy2001@yahoo.com or send it as a sharing file or using ws_ftp? If you are using ws_ftp,can you tell me the Profile name and other related name? I have tried once using your username and password, but I didn't find it.

Thank you very much.

lucy

Eric July 6, 2001 15:34

Re: gambit
 
Lucy, save the text posted below as test.jou (in a standard text editor) and then in Gambit do:

File--> Run journal --> test.jou (either in run or edit/run mode). After you run the journal it's a case of setting in and outlet planes and export it to a mesh file.

Save the lines below without the top and bottom dash line. ---------------------------------------------------

volume create height 200 radius1 12 radius2 12 radius3 12 offset 0 0 100 \ zaxis frustum volume create height 200 radius1 6 radius2 6 radius3 6 offset 0 0 100 zaxis frustum volume move "volume.2" dangle 35 vector 1 0 0 origin 0 0 0 volume copy "volume.2" to "volume.3" volume move "volume.3" dangle -70 vector 1 0 0 origin 0 0 0 volume move "volume.2" "volume.3" offset 0 0 200 volume create radius 12 sphere volume move "volume.4" offset 0 0 200 volume unite volumes "volume.4" "volume.1" "volume.3" "volume.2" face create wireframe "edge.7" real face create wireframe "edge.8" real face create wireframe "edge.9" real volume split "volume.4" faces "face.13" connected face delete "face.2" "face.3" "face.4" "face.9" "face.10" "face.11" "face.12" \ "face.13" "face.14" "face.15" lowertopology volume mesh "volume.4" cooper size 2 volume mesh "volume.6" "volume.5" cooper size 2 volume mesh "volume.7" tetrahedral size 2

----------------------------------------------------

Regards,

Eric

Eric July 6, 2001 15:36

Re: gambit
 
Better format:

volume create height 200 radius1 12 radius2 12 radius3 12 offset 0 0 100 \ zaxis frustum volume create height 200 radius1 6 radius2 6 radius3 6 offset 0 0 100 zaxis frustum volume move "volume.2" dangle 35 vector 1 0 0 origin 0 0 0 volume copy "volume.2" to "volume.3" volume move "volume.3" dangle -70 vector 1 0 0 origin 0 0 0 volume move "volume.2" "volume.3" offset 0 0 200 volume create radius 12 sphere volume move "volume.4" offset 0 0 200 volume unite volumes "volume.4" "volume.1" "volume.3" "volume.2" face create wireframe "edge.7" real face create wireframe "edge.8" real face create wireframe "edge.9" real volume split "volume.4" faces "face.13" connected face delete "face.2" "face.3" "face.4" "face.9" "face.10" "face.11" "face.12" \ "face.13" "face.14" "face.15" lowertopology volume mesh "volume.4" cooper size 2 volume mesh "volume.6" "volume.5" cooper size 2 volume mesh "volume.7" tetrahedral size 2

lucy July 6, 2001 20:27

Re: gambit
 
Thank you for your help.

I also can't delete the "face.2" "face.3" and so on because they have an upper topoly entity.

Now I have known what's your model like.

My model is made of glass,which can be deformed into any formation when heated,so the sides of the three pipes is very smoothy and there is no powders deposited in the bottom of the intersection.I want to simulate the real fluid trajectary, so the intersection shape is very important to me.

Could you give further help?

Thanks in advance

lucy

David July 7, 2001 07:38

Re: gambit
 
Hi,

If you have a PC then open a dos prompt and type,

ftp ftp.icemcfd.com

enter your user name as `ftp` enter your email address as the password

then type

cd outgoing/minns

then

get fluent.zip

regards

David


lucy July 7, 2001 11:13

Re: gambit
 
I am sorry that I can't get your file.

After I type ftp ftp.icemcfd.com , I get "ftp.icemcfd.com: Unkown host" . I don't know why.

I use the PC of my university.

Thanks a lot.

lucy

David July 9, 2001 03:55

Re: gambit
 
Can your email handle large files? About 5-10 megs? If so then I can email it to you.

Regards

David

lucy July 9, 2001 10:07

Re: gambit
 
Thank you Could you e-mail me the address xwang9@uwo.ca?

It's the address of my university.

Thanks again

Shyam Kishor July 11, 2001 10:00

Re: gambit
 
You do not need to delete these faces and can use the journal except this command to have a mesh.

There is a Gambit tutorial for creating hex mesh with boundary layers in three pipes junction. It is available through online help of Gambit. Go to Help then "Tutorial Guide" and see the tutorial "MODELING A THREE-PIPE INTERSECTION (3-D) ".

lucy July 11, 2001 12:53

Re: gambit
 
Thank you for your help.

I know what you mean. But my model is not exactly the same as the tutorial because the different diameter pipes and the 35 degree between the two daughters,not 90 degree.All of the difference can cause a lot of problems in the intersections when meshing and calculating.

Could you try it for once?You can see clearly.

Best regards

Shyam Kishor July 14, 2001 00:36

Re: gambit
 
I do not expect any problem in meshing it. Please contact your local Fluent Support for assistance.

Best Regards,

Shyam

lucy July 14, 2001 11:45

Re: gambit
 
It's great!

Could you give more details about your procedures?

Thanks you in advance lucy

lucy July 14, 2001 11:48

Re: gambit
 
I forgot to remind you that the diameters of the parent and two daughters are 24mm,12mm .They are not the same. Thank you


All times are GMT -4. The time now is 07:04.