CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM Running, Solving & CFD (http://www.cfd-online.com/Forums/openfoam-solving/)
-   -   non-stationary scalarTransportFoam (http://www.cfd-online.com/Forums/openfoam-solving/105114-non-stationary-scalartransportfoam.html)

kpax July 23, 2012 13:04

non-stationary scalarTransportFoam
 
hi everyone,

i am trying to modify the scalarTransportFoam solver so that it can be applied to time-varying velocity fields (that i have computed with another solver).
My idea was to insert the following lines from createFields.H inside the main loop of scalarTransportFoam.C:

-------------------
volVectorField U
(
IOobject
(
"U",
runTime.timeName(),
mesh,
IOobject::MUST_READ,
IOobject::AUTO_WRITE
),
mesh
);
# include "createPhi.H"
-------------------------------

i thought this would update the U-field for every time step. The U-fields in the respective folders are, however, all changed to the initial U-field ( /0/U).

Does anyone have an idea what's the problem here?

Thx in advance.

kpax August 2, 2012 09:52

anyone has an idea?!

wyldckat August 2, 2012 16:48

Greetings kpax and welcome to the forum!

Perhaps you should think the other way around, namely use an existing solver as a basis for a non-stationary scalarTransportFoam. Better yet, you'll find at openfoamwiki.net the following two pages:
  1. http://openfoamwiki.net/index.php/Ho...ure_to_icoFoam
  2. http://openfoamwiki.net/index.php/ScalarTransportFoam
If you compare the two articles, you'll see that the information from the second one is used in the first one!

Best regards,
Bruno

kpax August 3, 2012 11:12

hey wyldckat,

the other way around works perfectly! :)

thx alot!

vishalsacharya October 30, 2013 18:59

Hi,

But if icoFoam is used and a scalar transport is added to it, then both velocity and T would be solved for simultaneously coupled.
What if I want to use scalarTransportFoam, but instead have a time-varying velocity field that is not solved but prescribed by the user?
For example, what if my velocity is such that u=A+B*cos(t) and I want to use this time-varying field as the velocity with scalarTransportFoam, then how is this done?

wyldckat November 1, 2013 21:12

Greetings Vishal,

Quote:

Originally Posted by vishalsacharya (Post 459934)
But if icoFoam is used and a scalar transport is added to it, then both velocity and T would be solved for simultaneously coupled.

I might be wrong, but my interpretation is that the scalar transport of the passive scalar T is completely segregated!
One example of coupled vs segregated flow: http://www.cfd-online.com/Wiki/Fluen...ated_solver.3F

Quote:

Originally Posted by vishalsacharya (Post 459934)
What if I want to use scalarTransportFoam, but instead have a time-varying velocity field that is not solved but prescribed by the user?
For example, what if my velocity is such that u=A+B*cos(t) and I want to use this time-varying field as the velocity with scalarTransportFoam, then how is this done?

:confused: Uhm... with funkySetFields you can set the U field and then simply run scalarTransportFoam? http://openfoamwiki.net/index.php/Co...funkySetFields

Best regards,
Bruno

vishalsacharya November 5, 2013 13:01

Bruno,

Thanks for the reply.
Indeed, for a passive scalar, the U and T are independent. However what I have is a case where my U is a function of space and time and I have a mathematical form for it such as A+B*cos(t). I wanted to hard-code this into the solver.

I looked at the code and U (and phi) are defined in createFields.H before the simple.loop(). Which means that the default scalarTransport can only use a U that is defined at the beginning and kept the same throughout. I changed this by removed the block of code from "volVectorField U ( " until #include "createPhi.H" and included it in my simple.loop().
But since I am not reading or writing U to a file, I need not define U as an IOobject. Instead, I need to have a loop over all cells that calculates U as A(x,y)+B(x,y)*cos(t).

If I use the following inside the simple.loop():

volVectorField U = (U0+uamp*sin(2.0*pi*f0*runTime))*vector(1, 0, 0) + (V0+vamp*sin(2.0*pi*f0*runTime))*vector(0, 1, 0);

It does not work ! :(
Here, U0, uamp, V0 and vamp are dimensionedScalars that have units m/s and f0 is a dimensionedScalar with unit 1/s, runTime is the time value I have used, pi is a const scalar. All of these values are read from the transportProperties dictionary file in the createFields.H file.

wyldckat November 23, 2013 06:19

Hi Vishal,

I've finally managed to get some time and clear mind to look into your post.

The first thing that comes to mind is potentialFoam: http://openfoamwiki.net/index.php/PotentialFoam - this solver is as simple or simpler than scalarTransportFoam and allows for you to do tests of modifying "U" more easily with this solver.

The second thing that comes to mind: can you please share the source code of the modified solver you're working on? This way I could test for myself and figure out what's wrong/missing.

Best regards,
Bruno


All times are GMT -4. The time now is 22:48.