Numerics problem related to roof with airlayer

[WEI LI]

Hi, I am trying to simulate the metal roof system with the air layer.
However, the numerics problem always exists.
I have attached the assembly: metal deck - air layer - polyisocyanurate board - metal deck

assembly.jpg (70.37 KiB) Viewed 3082 times

Most of the settings are kept as default, such as surface transfer coefficient, initial conditions.

sur.jpg (69.22 KiB) Viewed 3082 times

The weather of Auckland, NZ from the database is applied as the indoor and outdoor climate.
After the calculation, a lot of convergence failures showed up, and balance 1 and balance 2 are in incredible difference.

numerics.png (22.83 KiB) Viewed 3082 times

I have tried using the adaptive time step, doesn't work.
I have tried using air layer without moisture capacity, doesn't work.
I have tried using surface sd-Value instead of actually insert the metal deck, doesn't work as well.
It is driving me crazy, please help!
Thank you so much

[Christian Bludau]

Dear Wei Li,

that may happen due to dew water in the air layer. Try to split it as described here:
viewtopic.php?f=32&t=1777&p=5226#p5226

Further the convergence will be much better, if you do not model the metal sheets as a layer but as a sd-value in the boundary condition.

Christian