Stairstep meshing in Gambit
 

Hello !

I have a quiet simple geometry, but i cannot find the way how to mesh it with stairstep type. There is a pipe (H=0.5, R=0.01) and rectangular hexahedron (W=0.5 D=0.3 H=0.3), and i want to mesh the pipe with stairstep, but i don't know how to connect them back after that. Than i'm trying to mesh both pipe and hexahedron with required precision (Interval size 0.001) it cannot satisfy memory request. So i need to refine mesh in hexahedron, but idon't know how. Stairstep meshing is necessarily in my case because pipe has a complex shape with corners and curves in general.

Thank you for help!

I am not familiar with stairstep (never used), but do you want only stairstep mesh because of your solver or whatever?
Or do you want staistep because you think this is the only wayto mesh your stuff with hexa?
Ca you display your pipe with complex shape?

Well, it should be something like that. Worse than pipe has - like shape, but i haven't even tried to make it. In case shown on pic leads of pipe could be meshed with cooper scheme, but middles meshes only with tetrahedra cells. I'm investigating shockwaves propagation in pipe and following jet simulation, so i need mesh as symmetrical as possible as asymetric influences on jet.

If you decompose your domain you won't have any problem to mesh it with hex or cooper.
Just split before and after elbows.
Then if you want symmetric mesh, you can work with one half of your domain (symmetry), mesh it and once your grid is ready, you can reflect your domain (with mesh)
You get the whole domain with reflected grid from symmetry (>> symmetric grid)

Thank you very much for your replies !

I made decomposition as you recommended, now direct parts are meshing with hex, but elbow not. Maybe i made something wrong ? While meshing Cooper error appears : regular side face (s) mesh for projecting volume could not found.

About symmetry, actually i need it about the pipe axis to avoid asymmetry on the pipe output.. so pave scheme is not good too, but it is only scheme that can mesh endings with hex. I made boundary condition on the end and it seems that it could not be meshed symmetrically with hex.


Best regards !

From your first picture you can see 3 edges . From here it seemed you generated a local axis-system.
So split elbow with surfaces made from edges. Elbow will be splitted into 3 subvolumes.
Then check this tutorial: http://my.fit.edu/itresources/manual...guide/tg03.htm
Remember you have to mesh first the elbow and then the other volumes (because of projection)

i found that the stairstep meshing is rather strange.
it creates a new boundary for me and also the result is not good.

any suggestions please.

my interval size was 12 and Hex and stairstep.

pictures may help

DEAR Max,

thanks for replying. i had not proceeded with the stairstep as a repeat of my buildings and domain have been done they were called f_285 etcc. this had been automatically generated as soon as i apply stairstep function..

can you please elaborate on the use of this function?

why do you want to use stairstep?

because no other option is meshing my domain and building..:(

May I see how your geometry looks like?

hi this is my domain and you can see two has been generated.

I meant the buildings...
;)

i do not have it on my pc right now as all my work is done at the university..

i had only this one which i had meshed with stairstep.
you can clearly see another domain has been generated..:(

yes I see it, but it is stairstep.
For sure you can mesh it with combination of tetra/hexcore and size-function.
But also for sure, you cannot mesh on the fly, and you will have to prepare your geometry

yes i understand.
but i have already tried all combinations only Hex and Staircase works..:(

did you split your volume?
did you merge some surfaces for deleting sharp edges?
did you apply size function on small entities for preventing cell skewness?
etc...
etc...
And finally did you read error messages from gambit?

error message from gambit was that high skewness...

and i did most of the above.

If you want you can provide me the dbs file

yes i want to.

only one size function................
:rolleyes:
Attachment 16975 Attachment 16976

The only problem is between faces 133 & 139 (a combination of split and connect should fix the issue)
Attachment 16977

hi..
did as you told and got that error
"mesh of face.334 contain 1 highly skewed element"

what are your parameters for your size fonction?

tet/hybrid, tgrid and 10

http://aerojet.engr.ucdavis.edu/gamb...ommands_create

I used tet/Hexcore
http://aerojet.engr.ucdavis.edu/gamb...chemes_hexcore

sorry i misread.

0.03
1.3
1

I used
1
1.2
20

hi
did you mesh the faces first?
and my PC is quad core with 16 Gb of ram..i still can not mesh it.

If I meshed the surface then it is only building's surfaces.
What kind of error do you receive?

volume cannt be meshed without faces sucessfully meshed

then you have skewed cells for sure, check where I pointed out (post #22)

i did of course as you recommended for in post#22...

still it says i have skewed elements...
sometimes it meshes then all freeze and i have to start again..

go to examine mesh and check where are those cells.

dear max,

after i split n connect faces. i then mesh the faces using tri pave

then i mesh the volume.
attachd is the error i found.

can you please enlighten me ?

cheers

hi
what to do when we have highly skewed elements? it say my mesh has 63 highly skewed elements...;(

thanks

fix those skewed elements.
First, check where they are, and try to find any topology which is responsible for the skewness

Gambit size function problem!
 

Hi,
I am trying to use two size function together but it seems doesnt work. my geometry is 3D I would like to have dense mesh y+=1 near the wall and at the same time to have coarser mesh in flow direction. it it seems like I am getting dense mesh near the wall but I got same mesh size in direction of flow. the other problem is gambit giving me problem with the memory!
thanks in advance!

what parameters did you give for size function?

Hi max,
thanks for quick answer!
my geometry is 170 1500 190 in mm. first size function is 1 to 3 with growthrate 1.2. and y direction 3 to 8 with same growthrate.

thanks in advance!!!