[Dicanic]

Hi all,



I am finding some strange results when testing OpenFOAM against analytical solutions. The scenario I am testing is 1D ss with generation. I am using gauss linear for my gradSchemes and gauss linear corrected for my laplacian schemes.


This is a snipet from my fvOptions where power is generated:


// ***** Power *****
const scalar powerlvl = 1e6; // Watts/m3


// Apply power *************************************************
forAll(V, i)
{
eSource[i] -= powerlvl*V[i]; // (W per cell)
}


My question is how does a small timestep produce the wrong solution?

[Dicanic]

Figure for reference: https://ibb.co/C13590S

[LuckyTran]

Are you sure it doesn't have anything to do with not using the same mesh?

[Dicanic]

I used three different meshes, but I can see how this figure would mislead only showing two with different timesteps. Suffice to say, even the coarsest mesh of 400 cells correctly matched the analytical solution when the diffusion number was greater than 5 or so. Still a little confused and worried over this but I'll update once I figure it out. It might have something to do with my residual control or tolerance.

[Dicanic]

So I figured out my solver tolerances were too low and a steady state wasn't actually achieved, the solver just stopped. Meaning, changing the tolerances from the tutorial cases is probably a good idea if you want to reach analytical level accuracy.



You can read about it here under 4.5.1.1 Solution tolerances:
https://cfd.direct/openfoam/user-guide/v6-fvsolution/