May be it's a simple thing , but ı'm confused and seek your help.
I was trying to solve a 1D diffusion equation:
fvScalarMatrix CEqn
(
fvm::ddt(C)-fvm::laplacian(DT1,C)
);
CEqn.solve();
and when I look at the flux using:
volVectorField C_flux("C_flux",(fvc::reconstruct(CEqn.flux())));
if (runTime.outputTime())
{
C_flux.write();
}
I'm getting non uniform distribution.
I'm getting a Gaussian distribution where as the flux is supposed to be constant. Isn't it?
Can any one help me to understand it :(
Did you make sure you've choosen the BCs so that your stationary solution is uniform (something like constant temperature 0K on all boundaries, initial temperature 300K)?
Did you make enough time steps to achieve stationary solution?
Hello Anton,
Yes I used a very low time step and allowed to reach steady state.
I also saw the steepness in flux calculated using above equation is
increasing on mesh refinement
But physically the flux is supposed to be constant isn't it ?
I really doubt the way fvc::reconstruct is implemented in openfoam
Please suggest me how to go about it
sorry I replied too quickly - i assumed you have a bell-shaped concentration distribution, not heat flux.
diffusive flux is
q=-D*dc/dx
if D is constant and dc/dx is constant, then q is constant, too.
Now it seems your D is constant. Have you checked whether c(x) is linear?
I will just add a brief comment as Anton already answered the analytical part of your question, that is if c(x) is linear and D is constant, the flux q has to be constant too.
OpenFOAM's fvc:reconstruct() allows you to interpolate a field that exists in the face centres onto the cell centres. More exact, you are going to reverse reconstruct the field from the operation:
fvc::interpolate() & mesh.Sf();
For more details see the attached links.
Have a nice day!
https://github.com/wyldckat/reconstruct-interpolate-fields/wiki
https://www.cfd-online.com/Forums/openfoam-programming-development/77943-fvc-reconstruct-algorithm.html
- Badges
- Science topic