CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   ANSYS Meshing & Geometry (https://www.cfd-online.com/Forums/ansys-meshing/)
-   -   [ICEM] ICEM periodicity specification (https://www.cfd-online.com/Forums/ansys-meshing/94458-icem-periodicity-specification.html)

robboflea November 16, 2011 11:11

ICEM periodicity specification
 
1 Attachment(s)
Dear all,

I'm new in using ICEM and Fluent. I'm presently meshing a 2D turbine blade (attached). I generated the mesh using basically the instructions of the tutorial for airfol meshing on youtube.

(http://www.youtube.com/watch?v=6wQPe...eature=related)

So basically I:

1)created the pre-mesh
2)converted it to unstructured mesh
3)selected Fluent as solver
4)specified boundary conditions
5)wrote input (Output -> write input)

While specifying the boundary conditions I assigned to the upper and lower boundaries of the domain (belonging to the same "PERIODIC" part) a periodicity boundary condition.

When I try to open the mesh in fluent i got the following error:

"Warning: Inappropriate zone type (periodic) for one-sided face zone 17.
Changing to wall.
Error: Internal error at line 2995 in file '..\..\src\f_periodic.c'.
thread_periodic_default: null neighbor cell thread(s)"

Am I missing something in the definition of the periodicity? For example I read that there's a function in Mesh->Global Mesh Setup->Set up periodicity. I didn't use that command since it was not used in the tutorial. Is it mandatory to use it always? How should I define the translational periodicity in this occasion?
Also I saw in some other ICEM tutorials that in some cases a periodicity between block vertices is created. What's its effect? When should it be needed?

I'm so sorry to ask you so many questions and I hope that somebody can help me. Unfortunately I didn't find ICEM help, or even google, helpful for solving my doubts.

Thanks in advance!

Rob

robboflea November 17, 2011 11:59

Looking on the forum I noticed that very often the concept of using a "shifted periodic" blocking is adviced.
The problem is that, from the pictures I saw, it's not clear to me how such a blocking strategy can result in a good mesh and how to actually build such a topology.

Is there anybody that can help me?

Rob

eric_albin February 3, 2012 18:47

Hi,
I have exactly the same problem (internal error f_periodic.c thread_periodic_default: null neighbor cell thread) and I do not succeed to find the answer on the web. Did you finally find the solution? This would help me a lot.
Thanks,
Eric

robboflea February 3, 2012 19:16

Hi Eric,

yes I found the solution. The problem arise from the fact that Fluent cannot handle non periodic grids while I was trying to feed a periodic grid into it (if I remember correctly).
I solved the problem by making a shifted periodic domain as described in this tutorial

http://www.youtube.com/watch?v=tYrbScUH9RE

and by assigning periodicity in ICEM (note that there are two different places where to specify periodicity in ICEM as indicated in several posts in the forum).
I hope that it helps. I'm sorry but I cannot attach pictures at present since I cannot access to my mesh files at the moment.

Rob

Far February 3, 2012 23:06

This was the problem in older version of Fluent, however, in latest version you can handle non periodic grids by using the GGI interface.

eric_albin February 8, 2012 04:04

4 Attachment(s)
Hi Rob and Far,

Thank you very much to both of you for your remarks. It helped me! My fluent version was recent enough (12.1.2) but the problem was probably a stupid oversight in ICEM (maybe a forgotten declaration of a body or association of an edge) or sometimes it was because of the name I was giving to the periodical boundaries... I tried both successfully a direct periodicity or a shifted periodicity.

I post some of my pictures since it can help future readers. This first test did not work because of the names "periodicity1" and "periodicity2". You should better to give them a name like "inlet1" and "outlet1". If you name it like "wall1" and "wall2", it will then be declared as a wall and won't work...
I give you some other examples later.
+, Eric

Far February 8, 2012 04:09

in ANSYS Fluent 14 there are many options for defining periodic interfaces. In Fluent 6.3 GGI (general grid interface) was not available

eric_albin February 8, 2012 09:49

5 Attachment(s)
Yes there is some other ways to define it in fluent 12.1.2 also, without defining any periodicty in ICEM. In theory, it works even if grids are not aligned but I did not succeed to make it work on a complicated case. However, it worked on the simple test case (like Test1). For this, I directly wrote the following command in fluent terminal : "grid", "modify-zones" and "make-periodic". I chose inlet1 when asking for the periodic zone and outlet1 for the shadow zone...

Otherwise, to make a shifted periodicity using ICEM, I used commands "delete block" and "merge-vertices" in the blocking section (see Test2). These commands are also very usefull to mesh more complex geometries (see Test4).

Thank you again for your participation.
Regards, Eric

Far February 8, 2012 10:14

Quote:

Otherwise, to make a shifted periodicity using ICEM, I used commands "delete block" and "merge-vertices" in the blocking section (see Test2). These commands are also very usefull to mesh more complex geometries (see Test4).
Instead of using these two commands, you can simply use the quarter O-grid option to shift the index. I have two question to ask particularly:
1) Vertices on opposite sides should be on same axial location? : I faced problem with older gambit and Fluent that even if you have equal no of index (i.e. same no of vertices on both sides) it was also important that you must have same axial location for each vertex pair.
2. If they are not exactly aligned then ICEM or Fluent allow the specification of periodicity and how much tolerance they give you.

One another comment, not related to this post, your NGV looks very thick and on the other hand gap between them is too narrow. Are you using them as windtunnel corner vanes? I can not see any different type of flows for pressure and suctions sides. Why is this so?

PS: By periodicity you are actually enforcing one-one periodicity. In my case non-matching periodicity worked well for 3D Transonic axial compressor rotor simulation in Fluent and CFX.

robboflea February 8, 2012 10:35

Quote:

Originally Posted by Far (Post 343377)
1) Vertices on opposite sides should be on same axial location? : I faced problem with older gambit and Fluent that even if you have equal no of index (i.e. same no of vertices on both sides) it was also important that you must have same axial location for each vertex pair.

Yes they must be. And, once you set the correct translational periodicity in ICEM and pair the vertices in couples they are automatically moved in order to be in the same axial position (or however in order to be translated by the vector defined in the periodicity specifications).

Quote:

Originally Posted by Far (Post 343377)
2. If they are not exactly aligned then ICEM or Fluent allow the specification of periodicity and how much tolerance they give you.

If they are not aligned ICEM will generate a non periodic mesh of course and the mesh check on periodicity will fail. Fluent will not open the mesh and crash during opening (at least this is what happened to me).
I don't know how much tolerance Fluent allows on periodicity but I can tell you that every mesh that I did by specifying correctly the periodicity in ICEM was read without problems by Fluent. For the answer on exact values of the tolerances I'm sorry but I can't help you.


Quote:

Originally Posted by Far (Post 343377)
PS: By periodicity you are actually enforcing one-one periodicity. In my case non-matching periodicity worked well for 3D Transonic axial compressor rotor simulation in Fluent and CFX.

Do you know since what version fluent is able to deal with non-matching periodicity?

Rob

robboflea February 8, 2012 10:38

Quote:

Originally Posted by eric_albin (Post 343371)
Otherwise, to make a shifted periodicity using ICEM, I used commands "delete block" and "merge-vertices"

I agree with Far, from my experience I noticed that sometimes ICEM starts doing weird things when many deleting and merging operations are performed. I also suggest the use of the quarter O-grid

Far February 8, 2012 10:43

Fluent 12.0 I have tested my self on Rotor 37. Since Turbogrid meshes were non periodic (by choice to increase the quality and better topology) therefore I had to create the GGI in CFX and periodic interface in Fluent (both are similar).

Go to interface option and make your periodic boundaries as wall 1 and wall 2 and you can see them in interface panel. However in Fluent 13 and Fluent 14, there are many options.

robboflea February 8, 2012 10:53

Are you sure that Fluent 13 can handle non-periodic grids? I'm using it, using ICEM to generate grids (I don't have turbogrid) and if I feed it with a non periodic grid it crashes while opening it.
Is there something I'm missing in the import from ICEM to fluent?

Far February 8, 2012 11:08

do not specify periodic boundary conditions on periodic boundaries in ICEM, instead specify as wall1 and wall2

eric_albin February 8, 2012 12:31

Quote:

Originally Posted by Far (Post 343377)
I have two question to ask particularly:
1) Vertices on opposite sides should be on same axial location?
2. If they are not exactly aligned then ICEM or Fluent allow the specification of periodicity and how much tolerance they give you.

One another comment, not related to this post, your NGV looks very thick and on the other hand gap between them is too narrow. Are you using them as windtunnel corner vanes? I can not see any different type of flows for pressure and suctions sides. Why is this so?

Hi,
Yes, I found the opposite sides should be on the same axial location. That's why, I define periodic vertices in Fluent...

For the simulation (Test4), it is nothing really serious but is an investigation for the design of an axial swirl generator for a research gas turbine. We were wondering if we could modify the swirl using some fluidics to modify the attachment of boundary layers...

nasa55 August 5, 2012 14:38

This is another way to create periodic B.C directly in FLUENT. I encountere the the error message when I specified this boundary in ICEM. You can type it in FLUENT and easily change a boundary condition to periodic.

I have copied this text from FLUENT help, part 6.8.4:
---------------------------------------------------------------------------
/mesh/modify-zones>* make-periodic

Periodic zone [()] 1
Shadow zone [()] 4
Rotational periodic? (if no, translational) [yes] n
Create periodic zones? [yes] yes
Auto detect translation vector? [yes] yes

*computed*translation*deltas:*-2.000000*-2.000000
*all*10*faces*matched*for*zones*1*and*4.

*zone*4*deleted

*created*periodic*zones.

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

Hope it works for you..:)

Jenny_W May 17, 2013 18:41

Hi Everyone,
I have a problem creating periodic boundary conditions. If you know the solutions, please help.

I created a geometry in ICEM CFD and tried to create the rotational periodic boundary condition in FLUENT by typing at the interface
define>boundary-conditions>modify-zones>make-periodic

However, FLUENT shows
Warning: differing number of faces on these threads.
zone 36: matched 0 out of 4112 faces.
zone 35: matched 0 out of 4237 faces.
Error: Failed to make zones periodic.
Error Objects: #f

Can FLUENT create periodic boundary condition on unstructured mesh? It seems to fail because the number of faces on the periodic zone does not match that of the shadow zone.
Thank you in advance,
Jenny

Shamoon Jamshed November 11, 2014 14:50

No, Fluent cant create periodic BC for US mesh. My question is that if you want to create periodic in a pipe, then how could you define the inlet and outlet BC. If your left and right sides are periodic? I m looking the side of pipe, ( abig rectangle)

How, if the surface of the pipe has rings (like grooves)

Shamoon Jamshed November 11, 2014 14:51

sijal I have a question
that if you want to create periodic in a pipe, then how could you define the velocity inlet and outlet BC. If your left and right sides are periodic? I m looking the side of pipe, ( abig rectangle)

How, if the surface of the pipe has rings (like grooves)


All times are GMT -4. The time now is 14:05.