Learning Forum

[Yannice]

Dear Ansys,

I have a 3D model including Energy Model, k-eps turbulence and DO Radiation model (Solar Loading). I indicated 2 radiation bands. However when I activate the radiation model the simulation diverges. Tuning the under relaxation values does not work, neither does decreasing the time step 50 times compared to the same simulation but without radiation that runs perfectly. I have a total of 1 million mesh elements.

See enclosed the residual plot of the version with and without radiation model. Now the relaxation value of do-intensity is set to 0.1, but also when set to 0.5 or 0.9 or 1, the solution converges (already at the beginning).

What could be the issue? And what different things I could try to solve this?

That would really be helpful, thank you in adv!

Yannice

[Yannice]

Extra info: the sun begins to shine only after 100000 iterations, thus not yet in the iterations plotted in the residual graph

[Rob]

Energy looks to be diverging too so check the radiation sources and temperature field in the domain before it goes wrong.

[Yannice]

Hi Rob Thank you for your suggestion, however it is not solved yet.
If the do-intensity under relaxation is higher than 0.5, it can be clearly seen that do-intensity residuals diverge first (but: no sunshine yet at that moment!)
What is the maximum cell size recommended when using the DO radiation model (and/or the Solar Load Model)?

Best regards

[Rob]

And how does the temperature field look as it starts to diverge?
Cell size depends on the model, we might have 25m cells in some cases which are fine and 0.25mm in others that aren't.

[Yannice]

Dear Rob Many thanks for further reviewing my question. See the images below for more details. Indeed the temperature diverges, it starts at the small water volume. This small volume can be seen in the center of the two mesh images. I tried both meshes and both diverge.
Time step size 3s, but also the case with 0.3 s, etcetera. The initialization is on a temperature of 293 K in all zones. Important notice: the simulation starts at 6 AM and there is no sun yet when the simulation already diverges. Further no other heat sources included in the model, so it is strange that the temperature increases out of itself.
Average skewness (mesh with bow): 0.24177; Average skewness (mesh without bow): 0.50577.
I also tweaked with the FLUENT Controls, that did not solve the problem.
The different 'air zones' is just to let the mesh go smaller more smoothly; in FLUENT I treat them as one air zone.
No heat/radiation sources yet, but simulation already diverges
In 2D the same mesh and settings (without radiation) does work. In 3D it works too, when disabling the radiation model but keeping all other settings the same
Also when changing the water column into a column filled with air or any other fluid substance, gives divergence.
Kind regards

[Rob]

Given the jumps in cell size I'm amazed it ran. Energy & radiation can be more sensitive to poor mesh so that may explain why a cold flow ran. Note, poor mesh can mean large changes in cell size, so even though the cell quality is OK the mesh may not be.

[Yannice]

Thank you for your advice! I tried some other options including inflation, but the meshing got stuck or did not finish.
Do you have any tips or examples for optimizing the mesh in my case? The thin wall is only a few centimeters while the building around is 1 hectare
Best wishes

[Rob]

I'd use a thin wall and then turn on shell conduction to ensure the radiation part was handled correctly. I'd also review the whole model to decide if the wall actually matters given the overall size of the domain.

[Yannice]

Thank you, Rob!