[lam]

Hi,

I want to know what is the criterium of convergence for a solver in Fluent which solves something like simpleFoam (incompressible flow).
Because I want to compare Fluent's criterium and my criterium that I used in simpleFoam.

Thanks in advance,

Lam

p.s: and can Fluent accept a model viscosity like BirdCarreau?

[alberto]

The standard convergence criteria adopted in FLUENT require that all residuals drop under 10^-3.

In my experience this requirement is not always satisfactory, especially for steady calculations. Usually I just set very low or zero residuals and check when residuals flatten and double-check convergence monitoring some property of the flow.

Yes, that viscosity model should already be implemented in FLUENT.

Regards,
A.

[lam]

Hi Alberto,

Thanks for replying me.

Do you think that a criterium which is somthing like:
norm[U_(n+1) - U_(n)]/norm[U_(n)] < tol
is wrong for a steady computations?
Otherwise, why this criterium is not implemented?

[sek]

The residuals in FLUENT are defined as the errors of the discretized equations, i.e., a_P*U_P -{Sum (a_NB*U_NB) + S_U} i the case of U-momentum equation. They are often normalized (or scaled) by the factors of users' choice such as Sum(a_P*U_P}. I guess OPneFOAM uses the same definition.