[cfdopenfoam]

dear all,

i want to solve scalarTransport in subdomain during runTime, after the flow fields of the whole domain are obtained by a flow model. is it possible to do it in this way:

……
#include "createSubMesh.H"
#include "createSubFields.H"  // the scalar fields are registered on the subMesh
……
// time marching, after U solved over the whole domain
solve
(
// scalar equations...
);
……

or

is there any other alternative to implement multiphysics modeling in subdomain? Two threads related to my problem but not fix it: http://www.cfd-online.com/Forums/ope...-equation.html
http://www.cfd-online.com/Forums/ope...subdomain.html

(i work with OF-2.3.0)

thanks for your interest. your suggestion will make my day.

/karekle

[cfdopenfoam]

greetings to all,

i think i should describe my problem more specifically, with the attached image. i hope this can help make myself clear.

after some searching, it seems that modeling multiphysics on subdomain is still a remained problem. there could be several reasons.

its difficult for one to only work with portion of the mesh to solve matrix equations.
to model physics on subdomain, one might re-create sub-mesh or split mesh into regions, and have to define new boundary conditions on the subdomain.
what makes things more difficult is that one may need to solve multiphysics on subdomain during runTime, which means one can not develop a new solver to handle the scalarTransport calculations after the flow simulation ends.

i am now still struggling on this way and, i think this problem deserves efforts. i will keep updating this thread based on my knowledge and put my solution here if i find one. also, i hope senior OF-users could kindly offer your comments, hints or suggestions here, which will receive highly appreciation!

thank you for your time.

/Karekle

[cfdopenfoam]

sorry that I missed sth important here.

the problem, both the flow and transport are transient, which means that we can not use mapfields to map flow fields from the whole domain to the subdomain and start another transport simulation. however, this can obviously be done in a steady case. we may need to do mapfields at every time step during runTime and solve the transport on the subdomain.

[arjun]

Solving in a subdomain or solving in some regions while not solving in others could be a big design change.
I am not familiar with openform so I can not really tell you how difficult or easy it is to do in openfoam.

If I had to do it in my solver FVUS wilkdkatze, it was as easy as setting up the case. Because this is base of design of FVUS.

In my case I would just create two regions and connect them by GGI (inteface). I would solve flow on two regions and energy on only 1.

If the region where T is solved changing with iterations or time, then FVUS has concept of partial region. Once should be able to define partion region and use this to create solver for Engery.

(Note FVUS at the moment only active on immersed boundary as partion region but other option for partial region creation like from volume of fraction are around the corner).

Good luck.

Originally Posted by cfdopenfoam

sorry that I missed sth important here.

the problem, both the flow and transport are transient, which means that we can not use mapfields to map flow fields from the whole domain to the subdomain and start another transport simulation. however, this can obviously be done in a steady case. we may need to do mapfields at every time step during runTime and solve the transport on the subdomain.

[cfdopenfoam]

Originally Posted by arjun

Solving in a subdomain or solving in some regions while not solving in others could be a big design change.

If I had to do it in my solver FVUS wilkdkatze, it was as easy as setting up the case. Because this is base of design of FVUS.

In my case I would just create two regions and connect them by GGI (inteface). I would solve flow on two regions and energy on only 1.

If the region where T is solved changing with iterations or time, then FVUS has concept of partial region. Once should be able to define partion region and use this to create solver for Engery.

(Note FVUS at the moment only active on immersed boundary as partion region but other option for partial region creation like from volume of fraction are around the corner).

Good luck.

thank you for your hints, Arjun.

I just googled your solver, FVUS wilkdkatze, but got little information on that. if you are happy to tell more, i am very much looking forward to it.

[arjun]

Originally Posted by cfdopenfoam

thank you for your hints, Arjun.

I just googled your solver, FVUS wilkdkatze, but got little information on that. if you are happy to tell more, i am very much looking forward to it.

On youtube,

https://www.youtube.com/channel/UCs6...M6JtRa8bLuVr6A

I am in process of releasing it. You could beta test it if you want.
Feedback help me, make it better.

[cfdopenfoam]

Originally Posted by arjun

On youtube,

https://www.youtube.com/channel/UCs6...M6JtRa8bLuVr6A

I am in process of releasing it. You could beta test it if you want.
Feedback help me, make it better.

thank you. i will have a try.

[cfdopenfoam]

hi all.

A little progress on the proposed problem.

I modified the scalarTransportFoam and developed a very simple solver subregionTransport, following the ideas mentioned above. I also setup a pitzDaily case to test the solver. unfortunately it did not work well. But I think we can work together to fix it. I do not know how to attach the solver and case here. if interested in my problem, plz leave a message. I will send you the files.

[ashish.svm]

Originally Posted by cfdopenfoam

hi all.

A little progress on the proposed problem.

I modified the scalarTransportFoam and developed a very simple solver subregionTransport, following the ideas mentioned above. I also setup a pitzDaily case to test the solver. unfortunately it did not work well. But I think we can work together to fix it. I do not know how to attach the solver and case here. if interested in my problem, plz leave a message. I will send you the files.

Hi,

I am facing a similar problem. I want to solve an equation of a sub-domain of the original mesh. If you have resolved the problem and if you can share the techniques used, then please post here.

Thanks a lot in advance.

[r2d2]

Hi, I am also interested in a correct setup/solver for solving some scalar transport on just a part of the domain using the global velocity field, as you mentioned in the two messages at the begin of the thread. Did you manage to get it work?