[ariorus]

Hello Oscar,

I looked at your code and I have a question about it.


Since the two equations are solved one after the other and not at the same time it is necessary to solve them many times such that the proper boundary conditions at the interface be satisfied, and you do this.
So you first solve the two solid equations and then reset the bc values of the coupled patches according to what is specified in the 0/T files.
But as far as I understood you reset the gradient value in solid2, if fixedGradient is specified in solid2 0/T file, and the temperature value in solid1 if fixedValue is specified in solid1 0/T file. What happens if you put for instance fixedGradient in both solid1 and solid2 coupled patch? Do you reset only the gradient value and not the temperature?

I may be wrong (if so please correct me) but I think that at every time>0 at the interface between cells of the 2 solids you should have always that the temperature and the heat flux be the same. Did you check this?

You could also try to see what happens playing with conductivity (if a conductivity for a solid is equal to zero you should recover a fixed temperature bc for the other).

I think that an analytical solution should be not too difficult to be found, for instance you could set up a 1D problem.

Ciao.

Rosario

[oscar_j]

Hi Rosario,
thanks for your comments.

My first idea in order to couple the two solids is to suposse no heat generation in the interface, so I want to couple the solids like Daniele propose:

T1=T2, at interface
dT2/dn=-k1/k2*dT1/dn, at interface

I think if you put fixedGradient in the two solids it could happen an error when the code reads a patch defined like fixedGradient initially and then it try to set fixedValue.

The solver hasnt been tested yet.
Any help and comment I appreciate so much.

Thanks a lot!
Oscar

[oscar_j]

Hi Rosario,

I post the first results I have obtained with twomeshes Solver. The case consists in two coupled cubes with 1 meter of side length and same material. The initial temperatures are T1=285K and T2=293K. With these considerations, the temperature of thermal balance would be T=289K. The image is a cross-sectional of the cubes.



Although the thermal balance is obtained, in the interface the temperature distribution has a rare variation around the thermal balance temperature, Any idea how could I make better the code?

Regards
Oscar

[oscar_j]

Sorry
This is the image:

I'm trying to use openFoam to simulate cooling of electronics by convection currents in a case as well as the heat dissipated by the case by convection as well.

Unfortunately, I can't seem to get any of these solvers posted to this thread to work with OpenFOAM 1.4.1 :-(

Also I was wondering about how one could do a continuous power input to a solid instead of specifying a temperature.

Forgive me I'm new to openfoam.

[sradl]

Dear Clark,

search the openfoamwiki for conjugate heat transfer - the solvers posted there should also work with OF 1.4.1 (not tested by myself).

Specifying a continous power input (e.g. W/m2) means that you will have a constant temperature gradient in the solid at the interface. Thus, dT/dx = q/lambda, where lambda is the heat conductivity of the solid. Specifying a constant gradient at the interface is a standard OF boundary condition :-)

br
Stefan

[sradl]

sorry, of course heat flows across the negative temperature gradient:

q=-dT/dx.lambda

cheers
stefan

Stefan,

I have tried to get the example of the conjugate heat transfer found here http://openfoamwiki.net/index.php/HeatTransfer
to work. I was able to hack the code so it compiled, however there are differences in the case file directory structure that I can't figure out how to solve. The main issue being how to combine the fluid region and the solid region in the constant directory so that the 1.4.1 based solver can read it.

Also note that the src tar ball for the pseudo heat solver doesn't contain the code for it.

[armin_h]

Dear All

I am a new user of OpenFoam as phd petroleum candidate, and i need to simulate fluid flow through the pipe , which should be cilynder.

I will be appreciate if someone let me know how can i create pipe shape in OpenFoam, and which part of OpenFoam Will help me to do so.

Many Thx in advance

[sradl]

Dear Clark,

it should be easy to find the code section in 1.4.1 where the source directory is defined for reading the mesh info. Just study the code - you should have a "solidRegion" and a "fluidRegion" directory under your case/constant directory; both solidRegion and fluidRegion should have a "polyMesh" dir.

In your case you possibly will need two liquid regions (the air inside and outside the case) as well as the metal. So I guess you need a major review of the solver.

br
Stefan Radl

[cedric_duprat]

Dear Forum,

My question is not a convection one but a coupling one and this thread is the more interresting one about coupling.
I'm doing a solver for coupling 2 fluids regions. In my case, I want to couple a channelOodles region and a Oodles one. It's usefull is you want to get turbulent flow without adding white noise.
I used a lot your upper convection solver to do that.

but, I still have one problem in the createFields.H files:

createFields.H:65: error: no matching function for call to 'Foam::dictionary::New(Foam::volVectorField&, Foam::surfaceScalarField&)'
/craya/big/duprat/OpenFOAM/OpenFOAM-1.4.1/src/OpenFOAM/lnInclude/dictionary.H:10 5: note: candidates are: static Foam::autoPtr<foam::dictionary> Foam::dictionary::New(Foam::Istream&)


it correspond to the following source:

autoPtr<transportmodel> laminarTransport1
(
transportModel::New
(
U1,
phi1
)
);

autoPtr<lesmodel> sgsModel1
(
LESmodel::New
(
U1,
phi1,
laminarTransport1()
)
);

in the Mattijs Janssens upper code, there is still line like that.
So my question is, Is there any constructor to create in the dictionnary files (.C and .H) or I made a mistake ther ?

Thank you for helping,
Regards,

Cedric

[eugene]

If you want fully developed inlet, it is much easier to use the directMappedFixedValue boundary to map internal fields to the boundary than to use a 2 mesh solution.

[panara]

Hi Eugene,
that is interesting, can you post an example of the blockmesh file with the internal field and ,for example, of the U file with the directMappedFixedValue BC?

I would like to simulate a long pipe with an oscillating flow and a section with heat transfer..

So my idea would be to have a first part of the channel with 'internal periodic BC' which will feed the second part with the heat transfer..

I oscillate the flow with a source term in the momentum equation.

I guess I am going to have problems when the flow reverse.. but I will think about that later on.. =)

For now, it is interesting for me just to try a pulsating channel (no reverse flow) and in this case it should work fine..

Daniele

[eugene]

Sure. In this example "inlet" refers to the patch that uses internal mapping:

1. In constant/polyMesh/boundary

inlet
{
type directMappedPatch;
nFaces #;
startFace #;
offset (<offset> 0 0 1)
}

2. Normally you need to map velocity and turbulent properties. Thus in U file boundary section:

inlet
{
type directMappedFixedValue;
value (<placeholder> 0 0 0);
average (<target> 0 0 2);
}

The "average" entry is not compulsory. For example a hypothetical k field inlet would look like this:

inlet
{
type directMappedFixedValue;
value (<placeholder> 1e-10);
}

Unfortunately, the "directMappedPatch" boundary mesh type is not yet supported by decomposePar and other mesh manipulation utilities. So if you want to run in parallel, you will have to manually edit the processor boundary files to add the directMappedPatch entries.

[cedric_duprat]

Hi Eugene and Daniele,

well, maybe you're write, I will try to use the directMappedFixedValue BC but, before trying, could you explain us what is it please. Because, I the forum, the "directMappedFixedValue" was cited only 3 times now : Eugene 1 today, Daniele 1 few minutes ago and me .... now :o)

more generaly, I think a coupling solver for two fluid will be also the first step for a RANS / LES coupling.

Cedric

[cedric_duprat]

oups ...
I'm a little bit late... :o)
Thank you Eugene

[oscar_j]

Hi everybody,

I would want to calculate the mean temperature in a fluid and solid, is there any function or way in order to do that?, I have worked on buoyantfoam for the fluid and laplacianfoam for the solid.

Thanks!
Oscar

[tehache]

Hi Oscar,

There is e.g.

fvc::domainIntegrate(T))

to integrate field T

[suredross]

hi,
i have a case with 2 inlets and an outlet,with 2 different liquids at the inlets.which equation and how do i couple it to work in openfoam?i tried the icoFoam solver but something seems wrong?
i am new to OF and need help,desperately!
thanks
david

[ngj]

Hi Davey

Have a look at the damBreak case in the User Guide. It is using the interFoam solver, which solves for a two-phase flow.

You say you are having two inlets, is it one inlet per fluid, or is there a mixture of the fluids in either inlet?
If the first is the case, you simply specify in the gamma file in /0/gamma that inlet1 has the value 1 and that inlet2 has the value 0. If you have a mixture, it might become somewhat more difficult - I don't know if there are any tools which you could use directly or if you need to program something yourself.

Best regards,

Niels