Subset of a volVectorField

em17 · October 5, 2018, 07:14

Hi all,

Does anyone know if it's possible to take a subset of the mesh of a volVectorField (in my case U) after the velocity has been solved? Ideally I want to create a subset of cells adjacent to a boundary patch.

I want to do this because I need to solve a separate differential equation that re-calculates the velocity at a boundary cell face using the velocity values of the boundary cell centre (i.e. adjacent internal field).

I know that subsetMesh can be used to split the mesh domain into two parts, but this would involve solving the UEqn on both sets, which I would like to avoid.

Many thanks,
Em

msaravia · October 5, 2018, 10:33

A simple way would be to create a volVectorField with reference to the current mesh and fill it with values only in the owner and neighbour cells of your patch using a forAll loop. You can loop, for example, through the owners of faces

Code:

const labelUList& owner = mesh.owner(); //Reference to owner 
forAll(owner, facei){ ... }

You could do the same only for your patch.

em17 · October 12, 2018, 09:06

Hi Martin,

Thanks for your reply.

I am trying to follow your method but I do not know how to specify the size of the new volVectorField. If I specify the size using 'mesh' (like it's done for the definition of the U field, below) then my new volVectorField will be the size of the entire mesh.

I can access the number of cells adjacent to the patch, but when I try to use this value (i.e. an integer number) it does not work. Do you know how I can do this? I am guessing the value of 'mesh' in the below example does not simply refer to the number of cells in the entire domain (which I initially thought it did) but something more complex - although I cannot figure out what this term represents?

Thank you,
Em

volVectorField U
(
IOobject
(
"U",
runTime.timeName(),
mesh,
IOobject::MUST_READ,
IOobject::AUTO_WRITE
),
mesh //this value
);

msaravia · October 12, 2018, 19:08

Yes, the volVectorField will have the size of the entire mesh, but I don't see why this would be a problem. Except it is mandatory to save memory. This is the simplest implementation that I can think of. Of course there are several alternatives, but all of them imply more code.

em17 · October 13, 2018, 05:23

Okay, so the clarify - I should fill the new volVectorField with the near patch data that I want (as you described previously) and leave the rest of the volVectorField empty? I.e. don’t even initialise it?

Would this mean that the differential equation I want to solve will only be solved for the given data in the volVectorField? Not the entire domain, which would be inefficient?

Thank you,
Em

msaravia · October 15, 2018, 10:10

No, with this approach you will be solving the equation in the entire domain (with no interface BCs). I think the only way to solve the equation in a subdomain is to instantiate a new fvMesh object from your desired element data (see the available constructors in the fvMesh class) and then set the appropriate boundary conditions.

October 5, 2018, 07:14	Subset of a volVectorField	#1
em17 New Member Emily Join Date: Nov 2017 Posts: 24 Rep Power: 8	Hi all, Does anyone know if it's possible to take a subset of the mesh of a volVectorField (in my case U) after the velocity has been solved? Ideally I want to create a subset of cells adjacent to a boundary patch. I want to do this because I need to solve a separate differential equation that re-calculates the velocity at a boundary cell face using the velocity values of the boundary cell centre (i.e. adjacent internal field). I know that subsetMesh can be used to split the mesh domain into two parts, but this would involve solving the UEqn on both sets, which I would like to avoid. Many thanks, Em

October 5, 2018, 10:33		#2
msaravia Member Martin Join Date: Dec 2011 Posts: 40 Rep Power: 14	A simple way would be to create a volVectorField with reference to the current mesh and fill it with values only in the owner and neighbour cells of your patch using a forAll loop. You can loop, for example, through the owners of faces Code: const labelUList& owner = mesh.owner(); //Reference to owner forAll(owner, facei){ ... } You could do the same only for your patch. Tobi likes this. Last edited by msaravia; October 8, 2018 at 10:48.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Multi subset mesh motion	hchen	OpenFOAM Running, Solving & CFD	3	March 5, 2022 23:41
the format of adding volVectorField term into equation	linyanx	OpenFOAM Programming & Development	1	March 16, 2017 09:27
Multiplication of two volVectorField	ChGr	OpenFOAM Programming & Development	1	May 30, 2016 11:08
Initialize a uniform 0 volVectorField	Schag	OpenFOAM	3	August 26, 2009 10:08
ICEMCFD: subset "smooth_show_map"	Andy	CFX	2	October 31, 2006 07:57

October 12, 2018, 09:06		#3
em17 New Member Emily Join Date: Nov 2017 Posts: 24 Rep Power: 8	Hi Martin, Thanks for your reply. I am trying to follow your method but I do not know how to specify the size of the new volVectorField. If I specify the size using 'mesh' (like it's done for the definition of the U field, below) then my new volVectorField will be the size of the entire mesh. I can access the number of cells adjacent to the patch, but when I try to use this value (i.e. an integer number) it does not work. Do you know how I can do this? I am guessing the value of 'mesh' in the below example does not simply refer to the number of cells in the entire domain (which I initially thought it did) but something more complex - although I cannot figure out what this term represents? Thank you, Em volVectorField U ( IOobject ( "U", runTime.timeName(), mesh, IOobject::MUST_READ, IOobject::AUTO_WRITE ), mesh //this value );

October 12, 2018, 19:08		#4
msaravia Member Martin Join Date: Dec 2011 Posts: 40 Rep Power: 14	Yes, the volVectorField will have the size of the entire mesh, but I don't see why this would be a problem. Except it is mandatory to save memory. This is the simplest implementation that I can think of. Of course there are several alternatives, but all of them imply more code.

October 13, 2018, 05:23		#5
em17 New Member Emily Join Date: Nov 2017 Posts: 24 Rep Power: 8	Okay, so the clarify - I should fill the new volVectorField with the near patch data that I want (as you described previously) and leave the rest of the volVectorField empty? I.e. don’t even initialise it? Would this mean that the differential equation I want to solve will only be solved for the given data in the volVectorField? Not the entire domain, which would be inefficient? Thank you, Em

October 15, 2018, 10:10		#6
msaravia Member Martin Join Date: Dec 2011 Posts: 40 Rep Power: 14	No, with this approach you will be solving the equation in the entire domain (with no interface BCs). I think the only way to solve the equation in a subdomain is to instantiate a new fvMesh object from your desired element data (see the available constructors in the fvMesh class) and then set the appropriate boundary conditions.