[dak81]

Hi, All!

Please, answer my question. Is it possible to
solve the convection-diffusion equation with OpenFOAM when convection is dominated?

[hjasak]

Yes - in fact, it is quite easy.

[dak81]

Thanks for the answer. But could you point to my mistake? I was doing so:

solve(- fvm::laplacian(T) + (u & fvc::grad(T)) == f);

where T - temperature

u - predefined analytical velocity field with multiplier alpha, so u = alpha*b,
where b - some vector field with div b = 0

f - source term calculated from the equation by substitution some analytical function theta satisfying boundary condition for T

so when alpha is small enough (<< 1)
residual = T - theta is tolerable, but
when alpha is >> 1, i obtain something wrong
solution which differs from theta very much.

Be glad to hear any suggestions from you. Thanks.

[hjasak]

Did you every look at scalarTransportFoam? This is where you should start from:

solve
(
fvm::ddt(T)
+ fvm::div(phi, T)
- fvm::laplacian(DT, T)
);


Hrv

[dak81]

I've tested your suggestion. It has improved my solution but not enough. I've two problems now:
1) it is not clear how small deltaT need to be choosen to obtain stationary solution;
2) while modeling the solution become unstable on some time step i.e. after some time step an absolute residual become growing up.

Any ideas? And what is the best interpolation scheme for my case? I've noticed filteredLiear to be better than limitedLinear.