[caohan]

Hi, all

I want to simulate single blade rotating by using MRF, in GAMBIT, I hard link the side face of computational domain and set is as periodic condition, when I import the mesh to fluent, it always said:

iter continuity x-velocity y-velocity z-velocity k omega time/iter
Error: divergence detected in AMG solver: x-momentum
Error Object: ()

it is werid because when I remove the periodic condition, Fluent can solve the model. Does someone can explain why? do I need to link the edge before I link the face?

By the way, for rotational simulation, can periodic condition use with multiple reference frame and sliding mesh, not only the SRF, sorry about my poor English and hope someone can give me a answer

Regrads
Han

[wildli]

hello, Han,
I am encountering similar problems as you. By simulating only one blade to represent the whole impeller, you will have to set periodic boundary conditions, otherwise you will not be able to use show the whole impeller region in FLUENT I think. You don't have to link edge before linking face in GAMBIT. link face directly and mesh those two periodic faces would be all right.
I have linked two faces and set them as periodic boundary conditions. but when I am checking mesh in FLUENT, it always comes up error such as zone 20 has inconsistent periodic setting.
Can any one give me any hint?
thanks.

[pepi33]

Hi both of you,

In fact, i had the same probleme an a solved it with this concept :

As you know, if you want a periodic model, the two faces (periodic-linked)must have exactly the same mesh! My advice is to not use a link with gambit.
In fact you mesh your two faces separately, but be sure that they have exactly the same mesh. It is quite difficult without "link", but if you use a thin mesh and grapping function you can succeed.

Now you have your both faces similar, you import your mesh in fluent and to make this two faces periodic you use the GUI command :
define=>boundary-conditions=>modify-zones=>make periodic
then you put the name or the number of your two faces. (don't change axes of symmetry, it is automatic)
If it works, faces become blue, and you have only one name for the two face.
If it doesn't work, your mesh is not exactly similar.

If you have any questions do not hesitate.
I made a report of that (in french), so if you need contact me

best regards,
pepi

[wildli]

hello, Pepi33,
I think I have solved my problem.
Your way is one of approaches to make periodic boundary conditions through FLUENT. That means we can define two EXACT faces as walls in GAMBIT (EXACT means two faces needs to have the same mesh, vortex, usually it can be achieved through linked function), then through FLUENT two walls can be made into periodic boundary conditions.
Another way is to define them in GAMBIT directly. While in my case I found something very interesting, it looks like there are some sequences for these two faces to be periodic boundary conditions. that means which one is going to be shadow matters. it looks like GAMBIT is not easy to make it and the way through FLUENT helped me solve this problem.
Inconsistent periodic setting error may result from different reasons from cases. hope my experience would be some help to some ones here.

[wildli]

can anybody else can help us understand what would be the reason of inconsistent periodic setting？
thanks

[teymourj]

I just wanted to add something that might be useful for some users that are getting this error.

"wildli" and "pepe33" are absolutely right about the procedure for setting the periodic Boundary Conditions (BCs). One more thing that might lead to the "Divergence in AMG solver" error is the type of periodic BC. By default after reading the mesh into FLUENT periodic BCs are set to be "translational". Make sure this type is the right type for your application. In my case I had to set my periodic BCs to be "rotational". The reason I was getting the error was that by default it was set to translational and I thought that this choice is done automatically in FLUENT.

Hope this helps.

[ghost82]

For rotational periodic boundary also take care at the rotational axis coordinates: sometimes, also a small difference in spatial coordinate (10^-4) may cause the inconsistency error.

Daniele