[lukas']

Hey there!


I'm new in Fluent and I have a general question:
How can I estimate how fine my mesh should be in general to get a proper simulation result? I red that there are some mesh quality criterias like orthogonal quality or skewness but I assume they are more or less independent from the element size. So is there a kind of general rule how fine my mesh should be?



And I have a second question: I have to do a simulation with a few components. I red that it is important that the nodes of the single components fit at its borders.

Therefore I found e.g. this video:
https://www.youtube.com/watch?v=y6xowX2UpR4



Is this a proper way to deal with that problem?


Thanks in advance!

[mCiFlDk]

Originally Posted by lukas'

Hey there!

I'm new in Fluent and I have a general question:
How can I estimate how fine my mesh should be in general to get a proper simulation result? I red that there are some mesh quality criterias like orthogonal quality or skewness but I assume they are more or less independent from the element size. So is there a kind of general rule how fine my mesh should be?

And I have a second question: I have to do a simulation with a few components. I red that it is important that the nodes of the single components fit at its borders.

Therefore I found e.g. this video:
https://www.youtube.com/watch?v=y6xowX2UpR4


Is this a proper way to deal with that problem?


Thanks in advance!

Hi again Lukas,


This is a trickier question than it looks. Meshes have several parameters that indicate you how numerical connectivity between them is, but they won't indicate if analyses are correct or incorrect. You can build a mesh up with skeweness of almost 1 and a great orthogonal quality, and never converge; or converge erroneously.


As a friend of mine says, meshes have an artisan point on them, and elements' first size approximation is given by the experience, but to "get a proper simulation result", a mesh independence study must be done. Don't worry if your first meshes look quite weird, because if performing a mesh indep. study your results seem to be ok, you shouldn't worry about that.


I'll leave here a link about what a mesh independence study is and how is it done, because it'll explain it much better than me: https://www.computationalfluiddynami...pendent-study/


As regards to your second question, you're completely right. Nodes must coincide between parts and this video is a good example for mechanical purposes. But if you're going to use it for a CFD domain with also several parts, I'd recommend you to think thoroughly about the meshing strategy.


Regards.