Information exchange for volume coupling

[lb_yuan]

Hello,

I just started to work on coupled simulation of two fluid domains. In my previous post, Yvan has already mentioned that there are basically two ways to do a coupled simulation (external coupling): one is the boundary coupling (face coupling) and the other is the cell coupling (volume coupling). The latter may require an overlapping section of the coupling domains to allow information exchange on the overlapped cells. Up to now, I only know the subroutine 'cs_user_coupling.c' in which one can specify some general coupling parameters to run a coupled simulation. But how can I specify the details of information exchange, e.g. source term exchange on the coupled cells and boundaries?

Kind regards,
Bo

P.S. I'm aware that the cell coupling can be used to couple different turbulence models, e.g. RANS and LES. Can anyone who has such experiences send me an example of this? Many thanks.

[Paul Brss]

Hello,

I'm also trying to realize coupled domains simulations but with no success for now...

Did you manage to make it run ?

Best regards,

Paul

[rodion]

Hello,

Do you know, that you can have two separated fluid domains in the same Code_Saturne instance?
With the function cs_user_internal_coupling_add_volumes (file cs_user_parameters.c) you can separate some certain groups of cells. These groups of cells will be separated from the rest of the domain using automatically defined thin walls. For these thin walls you will need to impose your custom boundary conditions for coupling. Through these boundary conditions you will be able to impose desired values or fluxes.

Best regards,
Rodion

[Paul Brss]

Hello,

Thanks for answering !

In fact, I'm want to make one domain move with respect to the other. As I described here:
viewtopic.php?f=2&t=2220&p=12236#p12236

Therefore, I'm not sure that what you suggest is suitable for what I plan to do.
I might have misunderstood what you said though.

Best regards,

Paul

[Yvan Fournier]

Hello,

Yes, the functionnality Rodion mentions (and uses) is quite different, as it is used for a cleaner (stronger) coupling of non-overlapping meshes. Both have the name "coupling", but the newer "internal coupling" couples not only boundary conditions but most linear algebra operators, while the coupling used for moving meshes or RANS/LES coupling (and also a variant of the turbomachinery modelling) only couples boundary conditions (and possibly, volumes using source terms).

Regards,

Yvan

[lb_yuan]

Hello,

I've tried both boundary coupling and volume coupling with a simple case (both are external coupling between different code_saturne instances) a couple of weeks ago and will try later.

In my boundary coupling test case, two separate domains are connected via an interface. I assume the result obtained from one domain will be extrapolated onto the interface which will act as a boundary condition for the other domain. It takes much longer to reach convergence with respect to the case without coupling. There is also an issue that the conservation constraint seems not to be satisfied for one of the coupled domains. Moreover, the performance of this coupling is sensitive to the location of the interface.

In my volume coupling, I embedded a small domain into a large domain, which means the smaller one is totally immerse in the larger one. Both domains are meshed separately with different grid resolutions. Hence, in the overlapping region, there are actually two meshes. In this case, the issue of continuity still exists. I've also no idea what kind of information is exchanges between volumes of the coupled domains. My initial idea is to make use of the volume coupling to allow information exchange (perhaps source term exchange) between coupled domains meshed with fine and coarse grid respectively. the large coarse-mesh domain will provide boundary conditions for the embedded small domain, whilst the fine-mesh small domain will pass source terms to the large domain to tune the models applied on the large domain.

Kind regards,

Bo

[Paul Brss]

Hello,

So you manage to make a volume coupling simulation work ? Is it possible, by any chance, that you load your case (or a simplified one) here ?
Or at least your SRC/directories of both domain and .log files.

It would be a great help !!

Best regards,
Paul