Learning Forum

[NimaNZ]

Hi everyone

When I model this building with only one opening, and prepare the domain for structured mesh using the Multizone method, and separating the domain into sub-domains. it's working well; but as soon as I add two other openings, although I added every detail of each length into the mesh, the MultiZone doesn't work; could you please let me know how I can fix this?

[Rob]

The blocking may have become confused so it's struggling to link the openings if there are windows on the other end of the model. Why are you modelling a wall thickness?

[NimaNZ]

So if there are too many guide lines in MultiZone it can be confusing? Should I add sub-domains separately then the data load would be less confusing?
The wall thickness has an effect on airflow passing through the window, but I suppress the body and it won't be solved or meshed.

[Rob]

It'll only have an effect if you can resolve the flow: how many cells have you got over the thickness?
Multizone needs to tie logical blocks together, as you add multiple holes etc that gets complex. ICEM CFD can be used to fix this, but it's often easier to use an unstructured mesh or spilt the volume up.

[NimaNZ]

I have two cells over the thickness, and IÔÇÖm using wall function in LES.
Thanks to what you mentioned as Multizone will be confused with openings, I split the volume into smaller volumes and now itÔÇÖs working. Thank you so much.

[Rob]

Try 5-10 cells over the thickness to resolve it: 2 is too coarse.

[NimaNZ]

I agree with you and it makes sense, but then the growth rate of the mesh would go higher than what is recommended for LES which is usually less than 10%. Otherwise, the number of grids will make the simulations infeasible. Another problem is that, close to the wall, by this much dense mesh keeping the Courant number less than 1 would make the time-step size again infeasible. I can send many studies with 2-4 grids at the wall thickness. (but still what you are saying is absolutely correct but makes LES infeasible for high Reynolds numbers)

[Rob]

Which is why we often use RANS or lots of really big computers. Welcome to the CFD trade off: accuracy, speed and resource. If you want one you need to trade off one of the other two.