How to set up a wind tunnel with interior?
 

Hello!

I have some huge problems with the following task: (see attachment)

I have to set up a wind tunnel (face 2), a smaller quader which has to act as interior for a smaller meshing setp (face 4) and an aerofoil in the middle.

My first problem is the substraction of the three faces.
Which steps i have to follow?

1. substract face 1 from face 4 (without retain or something?) gives then Face A
2. substract the new face A from face 2 (also without retain or something?

i think i have to set up face 4 as interior, right? do i have to do this after the substracting?


kind regards!

-Substract face 4 with face 1, it generates face A
-Split face 2 with face A
You don't need to set any face as Interior

okay, i did it your way, but now gambit meshes the area between face 4 and face 2, and the aerofoil in the middle (face 1). the area between face 1 and face 4 remains empty.

hmm funny...
if you substract face4 with face1, what is the result?

then it works. but when i split the faces afterwards, i have the porblem i wrote in the post before.

now i can mesh every face how it should be, but when i export this file and import it into fluent, face 4 has a behaviour of a normal wall. do i have to change the boundary coniditons into fluent or gambit of face 4?

(i know how to set up the other boundary conditions like velocity inlet, walls and pressure outlet)

okay wait, now the mesh from face 2 is under the mesh from face 4 and face 1

ok I assume the face 4 and face 1 are right substracted (I call this face A).
Try to split face 4 with the outer edges from A (rectangle)
Normally you shoul have 2 faces on the same location (face A and the face resulting from the split)
Delete the face resulting from the split.
All the edges should be right connected, but you can go and connect all edges (as security)

okay, you mean i should use the "split face" option, select face 4 by "face", select "edges (virtual)" by "split with" and then i select the 4 edges from face 4, right?

when i do this, gambit shows the following error meessage:

face.4 could not be split with edge... which is coincident with the face boundary. Such a split may result in a degenerate geometry.



sorry, i know the "basics" form gambit, but for this project i have to to it with the inner face 4 and i have so huge problems with this assignment as you can see :(

wait... you must split face 2 with the 4 edges from face A (face 4 shouldn t exist anymore as you substrated it with face 1)
send me the dbs file

on your email adress?

upload it on www.rapidshare.com

okay, here it is:

http://rapidshare.com/files/219227546/working.rar.html

yes, this works! thanks a lot!!!

okay, now i have the SAME problem, just for 3D.

I tried the same to solve this issue, but seems like that something with the mesh IN the interior doesn't fix. when i say fluent to show me "just the velocity vectors of the interior", it shows me just vectors on the wall of the interior.
The interior is not filled with vectors. I guess that this is wrong, but i don't know what my mistake is.

any ideas/help for my problems?

Kind regards!

post a picture

okay, picture 1 shows my gambit work. the round hemisphere is the interior.
the two bricks in the middle are the walls.

picture 2 shows the bricks (lorry) in fluent. seems like everything is all right!

picture 3 shows now the interior and i have just vectors on the surface of it. is this wrong? i think so, but what could be wrong?


when i want to see the vectors of the standard interior then it shows me all vectors, i mean, then it is okay.

in gambit, click on the icon (next to examine mesh)
you will have blue colors instead of green (and pink for the hemisphere).
post the picture.
You should have difficulties to mesh your domain because of the small angle between the big brick and the hemisphere.
What should be th functionnality of the hemispherein comparison of the big brick?
Create a XY plane, and display any variable (velocity for instance), and post also the picture

okay, here is the picture with the specific color mode.

the task of the hemisphere is... its just that we had this more or less in the lesson. the teacher said to us we should create an interior (what sounds logical) and he did it with a hemisphere.

maybe its better to take a big box instead of the hemisphere?

but maybe this is also a problem:

when i mesh the big box, this error message appears (attachment)

when i mesh it a second time, gambit refines the mesh and everything seems like okay. is this a problem? i'm not sure with this.

it comes from the lines wich are tangential (between base of hemisphere and brick's edge)
Post a picture of the plane (just to be sure that your big brick is meshed and exported)
In your case the hemisphere will just have a mesh control functionality. Indeed you may refine your mesh inside it (eg: around the small bricks), and coarse the mesh outside it.

okay, here the mesh. seen on the picture are the velocity vectors of the lorry (the two small bricks) the interior and the plane.

then it is ok, all the domains are exported.
To your first question, I believe fluent prefers displaying vectors on a plane.
Display pathlines from your inlet and yoou will see If you have any interior problem

yes, by showing the pathlines of the inlet to the outlet it looks pretty good:

but now i ask my self if the interior is necessary? ^^

As I said, in your case the hemisphere isn't necessary (your fluid domain es the big brick containing your hemisphere).
Per Default, Gambit exports and creates the fluid domain called something like "interior".
SO you don't need to specify the interior domain, except if you need to handle more than one fluid domain (porous, multiphase etc....)

okay, so, interior is good when i use more than one fluid domain. yes, this makes obviously sense for me.

its just that our teacher told us we should create an interior to mesh/iterate faster.

Indeed you can have more control on your mesh refinement inside your hemisphere. But it is not a must

but i still think that there is something wrong with my interior. in my opintion the flow IN the interior should be also filled with velocity vectors ect. like in the standard-interior created from gambit.

i will explain you very accurate what i did to create this hemispheric interior. maybe you can see an error in my steps and i could solve this problem then.

I create all the geometries in gambit. my wind tunnel is A, the lorry (two small bricks, with the red circle) is B, and the interior is C.


First step i have done: substract the lorry B from the wind tunnel A in one step. This is the same like in 2D. This gives me a new geometry, lets call it D.

But then i have a problem, the until now spheric interior has to be splitted from D.
So when i go to split volume, i chose volume 1 (D) and split the sphere (C) from it.

I just activate the option "connected" and NOT bidirectional and then just the hemispheric geometry stays.

Thats all, now just set up the boundary conditions and export the file as mesh for fluent.


Is this okay? Its the way i did it.

It's look like ok from my side.
How did you set your Interior in gambit?
in BC Panel as Interior or in Continuum Panel as Fluid?

i set it in "specify boundary types" as interior.
i didn't specified something in the "specify continuum types" windows.

That's the reason, the BC is just available for surfaces
Define your hemisphere as fluid in "specify continuum types", and export it.
You will be able to get what you want.

Okay, the thing is now the following. I changed the things you told me and now the mesh in gambit looks pretty well!

but now i cant iterate in fluent. i dont know why but the iteration goes to infinity after approx. 120 iterations. at the beginning the line looks steady, but then it goes up and down and up and down and this over values of 1000000000000000000 ^^

*check the set up of the new fluid domain in the bc-panel (eg: material).
*check the results to be sure that the fluid flows well

But as I previously said, defining the hemisphere as a separate fluid domain, won't bring you anything (except viewing it separately)

okay, this problem is now solved. thanks a lot for your help.

next problem :(

now i try to solve this with the k-epsilon model and the teacher told us y+ should be between 50 and 500. My y+ is over 40000... i try now since many hours different meshs but there is no change.

Adapting the y+ value is also no solution. Any tip for my about this problem?

yes, that was a big help :)

do you know/can you tell me which formula the first calculator uses?

kind regards

check turbulence theory , especially log-law for wall treatment

What do you think?
 

-----------------------------------------
Please help guide development at ANSYS by filling in these meshing surveys

CFD Online Users Survey

This second one is for ICEM CFD users...

Public ANSYS ICEM CFD Users Survey