[CFDMedic]

Hello

I'm running LES through a complex pipe-like geometry and my residuals look unusual - the continuity residual is a flat line. Does this mean that the continuity has converged or is there a problem?

Thank you for any help!

I've attached a screen shot of the residuals

[FMDenaro]

Originally Posted by CFDMedic

Hello

I'm running LES through a complex pipe-like geometry and my residuals look unusual - the continuity residual is a flat line. Does this mean that the continuity has converged or is there a problem?

Thank you for any help!

I've attached a screen shot of the residuals

Actually I see that the convergence problem appears in the momentum components, too.

[CFDMedic]

Is this because the residuals are oscillating between 1-0.05? i.e. the error is too large?

[FMDenaro]

Originally Posted by CFDMedic

Is this because the residuals are oscillating between 1-0.05? i.e. the error is too large?

From your figure I see O(10^-2), that is a too large magnitude

[arjun]

Originally Posted by FMDenaro

From your figure I see O(10^-2), that is a too large magnitude

i think they are scaled residuals.


/////






The way i see is that the momentum residuals mostly have contribution from temporal derivative that is wrong when time step is changed.



However the velocity change due to this is not big enough to disturb fluxes in a significant way so continuity residuals don't fluctuate as much as you think they should.

[FMDenaro]

Originally Posted by arjun

i think they are scaled residuals.


/////






The way i see is that the momentum residuals mostly have contribution from temporal derivative that is wrong when time step is changed.



However the velocity change due to this is not big enough to disturb fluxes in a significant way so continuity residuals don't fluctuate as much as you think they should.

Maybe they are normalized, the value starting from O(1) shows that the residual decreased only two order of magnitude. That is not a convergence.
If the problem is for incompressible flow I suggest to use the NITA procedure.