[TWIN]

Hi guys,

Here we are.
I was working on some simulation and apparently I have some issue with PERIODIC boundary conditions.

My domain is incredibly simple, just a cubical box, and i used PERIODIC boundary conditions coupling opposite faces! Up to this point, no issue!

Running the case, it's straightforard that It calculates surface of coupling boundary and their magnitude is equal.
However going ahead it comes out with an error that is something like:
ERROR: the slave nodes must be a PERIODIC status!.

This error appears several times and the number of this errors changes with mesh refinement. However no monotonic trend has been found, thus number of errors does not decrease along with mesh size..

Do you have any suggestion????
Thanks in advance! Have a nice day!

[ymao]

Andrea,

Can you email your case (*.in and *.dat) files to support@convergecfd.com as it would be easier to diagnose the issue with the case setup. It would also be helpful to know which version is in use. Thanks!

Best,

[ymao]

Andrea,

I will check it. Thanks!

[TWIN]

Finally I solved the problem.
it was about running in parallel. Dividing domain for parallelization it comes out that two points coupled by PERIODIC boundary condition are on two different processors! Thus they try to find one another without success! It runs using just one couple of periodic bc, as soon as another couple of surfaces have been added to the latter it blows up.

Hoping it helps for future issues.

[hmzha]

Hi Twin,
I am also facing the same problem. How do you solve this?

Thanks

[TWIN]

Hi,

Apparently you can run SERIAL instead of PARALLEL!
It will be slower but it should work properly!

[hmzha]

Ah yes. Of course you can do that. But that does not solve the problem of running periodic boundary conditions in parallel though.. Converge should be able to do that

[MFGT]

The example cases of engine sector with periodic boundaries run in parallel without problems, so it must be something else.

However, Andrea had a cubic box with periodic boundaries on opposite faces.
What exactly should be represented by that? Its basically a pipe where the outflow mass enters the inflow or what?

What does your case/geometry look like, Afiqah?

[hmzha]

I am using a sector of a nozzle and I also get the same error. I will have a look at the engine sector case in detail. Thanks for the tip!

[hmzha]

I managed to get help from Converge directly. It seems this problem will arise when the periodic angle and boundary is not specified correctly.

In my original setup of a 45deg sector, I defined the first Periodic face which is at -45deg angle with the second Periodic face if we follow the right hand rule.

The correct way is to switch the boundary condition definition so that the first Periodic face is always at positive angle with the second Periodic face based on the right hand rule.

Once the Periodic faces are switched, the case ran smoothly.

Hope this helps

[hmzha]

I managed to get help from Converge directly. It seems this problem will arise when the periodic angle and boundary is not specified correctly.

In my original setup of a 45deg sector, I defined the first Periodic face which is at -45deg angle with the second Periodic face if we follow the right hand rule.

The correct way is to switch the boundary condition definition so that the first Periodic face is always at positive angle with the second Periodic face based on the right hand rule.

Once the Periodic faces are switched, the case ran smoothly.

Hope this helps

[filipimfs]

Hello,


I'm facing the same problem for a cube with oposite periodic BC's: Converge (I'm using v2.4.19) does not allow me to run it in parallel. Any ideas?


BR,
Filipi

[ksrivast]

Hello Filipi,

As hmzha mentioned, this error is commonly seen when the periodic boundary information is not entered correctly. Please re-check your dx/dy/dz values provided for your Planar periodic boundary and that it corresponds to your geometry.

If your settings are correct, and you still face this issue, please contact support@convergecfd.com with your case setup and we shall have a look.

Sincerely,

[filipimfs]

Just sent the email.


Thank you,
Filipi