[Arne Voß]

Hi everyone,

is it possible to use e.g. the rolling, pitching and/or yawing moment coefficients to determine cauchy convergence?

Example: I believe that this would be useful to determine a converged solution for e.g. an aileron deflection, which doesn't change the lift coefficient very much but should be clearly visible in the rolling moment.

After reading the documentation, I tried to set:

CONV_FIELD= ( LIFT, MOMENT-X, MOMENT-Y )

in the config file, but I see the following messages:

Ignoring Convergence Field(s): MOMENT-X MOMENT-Y
Convergence field(s): LIFT

Where is my mistake??

[pcg]

I think they are written with underscore instead of -

[Arne Voß]

Hi Pedro,

thank you very much for the quick reply. Could you please update the documentation?

https://su2code.github.io/docs_v7/Custom-Output/
MOMENT-X Total Moment around the x-axis.
MOMENT-Y Total Moment around the y-axis.
[...]


However, i have a follow-up question. I believe it would be helpful if the user can choose between relative or absolute values for cauchy convergence.


Example: For the clean aircraft, the rolling moment coefficient MOMENT_X is close to zero and thus will never reach a relative cauchy convergence ->> dividing tiny numbers is not a good idea. Using absolute cauchy convergence is more robust in this case.


Arne

[pcg]

Happy to help Arne.
If you could open a PR to update the docs it would be great https://github.com/su2code/su2code.github.io

That is a good point, maybe we can switch automatically to absolute if the value of the monitored quantity is below 1, or add a custom scaling factor, that when set in the config overrides the behavior of dividing by the absolute value of the quantity.

[Arne Voß]

Hi Pedro,
I corrected the documentation and created a pull request. Because this is the first time for me, I am not sure if I did this correctly?

I would suggest to have something like:
CONV_CAUCHY_TOTAL = CONV_CAUCHY_ABS + CONV_CAUCHY_RELATIVE.
In this way, the user could set CONV_CAUCHY_ABS=0.0 if he or she wanted to use only relative convergence or set CONV_CAUCHY_RELATIVE=0.0 to use only absolute convergence. Also, I would suggest individual parameters for each physical value similar to this:

CONV_FIELD= ( LIFT, DRAG, MOMENT_Y )
CONV_CAUCHY_ABS= ( 1E-4, 1E-5, 1E-4)
CONV_CAUCHY_RELATIVE = (0.0, 0.0, 0.0)

[pcg]

Thank you for the PR!

I see how that would work but unfortunately, it would not be a backward compatible change, and I think the relative + absolute logic would confuse folks.
Whatever we introduce must be simple and inactive by default to keep backward compatibility.