Convergence analyzer

[manexi]

Hello

I'm new to this software and there are some aspects I don't quite get about the calculations.

Lately, fast calculations seems to finish before 100% (or so does the bar say), and the program asks me if I want to see the convergence analyser (well; of course I do! ).

If I'm lucky there's nothing relevant, and if I'm not, I get some convergence failures which I understand as "we couldn't meet the expected results after maximum number of allowed iterations". Nevertheless, I don't quite understand the "Iteration interrupted" and "rain iteration" infos. I mean, why does the iteration stop? Was there a problem or did it find the perfect solution and then stopped? And what does WUFI do when this happens?

As for now I was trying to model a simple insulated concrete wall with a real local climate from a .wac file and found that there were quite some convergence failures in the convergence analyser (which I suspect the high variability of the real climate to be responsible for)... should I accept these results? I wonder what WUFI does when it can't reach convergence. Does it diminish the time step and tries again or does it take the last available value and uses it for the next time step?

And last, but not least. I've been looking around to find some answers regarding this subject and the only recurring hint is "balance". What exactly is this "balance" (I guess it's the flow of a given quantity, but do you refer to the global one, or the one through a surface? or something else? ...and how do you get it?)

I know it's a long message, but I need to understand the control tools given to me so as to be sure if I can accept the simulation results or not.

Thank you.

manexi

[Christian Bludau]

Hello manexi,

you can also open the convergence analyzer by performing a right mouse click on the calculation window and choose analyze...

Iteration interrupted means that the residuals do not change during a couple of calculation steps, so the calculation is interrupted.

Rain Iteration is a iteration, which deals with the adaption of the rain water absorption coefficient at the surface. This iteration is performed max. 10 times. If there is more water available, than the material is able to absorb, a rain convergence error is shown. Usually this has no effect on the calculation result quality.

A convergence error can be caused by many things, the grid, the climate, the combination of materials and so on. So if there are convergence errors in your calculations, you have to check, if they cause a high change in the water content. The best way to see this is Wufi2DMotion (use 'goto' and watch the time steps around your error). If there is a abrupt high peek in the water content somewhere in the construction, you have to care. If you see nothing, the convergence error is usually acceptable. The convergence error means, that the convergence criterion is not reached. Wufi is using the last calculated value (the difference to the convergence criterion is given in the calculation window) to start the next step.

The quality of the result you also can see in the results of theta and phi in the calculation window. There the difference between two subsequent inner iterations is given as the biggest deviation over the whole construction.
E0 means an deviation of 100%. So for example, if you get a phi with an result of 1E-02, that means, that somewhere in your construction your relative humidity is 81 or 79% instead of 80%. So values with a deviations of E-2 and higher are bad. Same for the temperature.

The balances you mentioned are only available in Wufi 1D (Pro). In Wufi 2D they are not that easy to build because of the possible complex geometry.

Hope that helps,
Christian

[manexi]

Thank you! It does help a lot!

Manexi

[Foti]

Hello
Just found this issue, and i read this convergence calculator, is this in Wufi 6? if so how to I pull that menu up?
I ran some simulations and i get zero convergences but after some new wall simulations I get a high number of convergences, and saw this "calculator"

[Christian Bludau]

This issue discussed here is for WUFI 2D.