Hi all,

I would know if it is correct to use hexahedral mesh and 804 cells, 3220 Nodes to mesh a pipe 2 meter length.. because I have problem when I use moor cells, Fluent not recognise the mesh.

Cheers, Nicolás

One usually determine how fine the mesh should be by the requirement of the case.

What's the diameter of your pipe? You could use boundary layer on the face of one end first and then mesh the pipe. By this you could 'make the cells smaller'.

Hi bli,

thanks for your reply,

the diameter of the pipe is 2 mm..

I am meshing one face and than the complet volume, so I can obtain moore cells..but this needs few PCU time to mesh..

thanks,

Nicolás

Hi

The only way to know whether you have used the right number of elements is by comparing your results with experiment. You must refine the mesh till you get mesh-independent results that are validated with experiment. As a start, it is often advisable to perform "Gradient Adaption" on velocity and other parameters in the domain to obtain reasonable gradients. This will usually increase the number of cells. If you are analyzing near-wall aspects (boundary layer, etc.), you should have much more than 800 cells for your domain, especially when the Reynolds number is high. You should perform adaption for y+ in this case. I hope this helps!

Kind regards

Hi AAA, Thanks,

I have made different meshes until obtaining a similar result in Fluent, but the smaller mesh that fluent has allowed me to use is with 804… the obtained result look like normal, but the obtained velocity profile is very illogical.

What technical I can use in GAMBIT to make a mesh in the pipe? all the techniques that I have used need more than 2 hours processing..

my Re is smaller to 1000… in this way is not necessary to use y+.

thank you very much!

Nicolás

Hi

It is not clear to me when you say "the smaller mesh that fluent has allowed me to use is with 804". Do you mean that you can not perform simulations when you generate more elements when using adaption?

You should (and be able to) continue refining your mesh by performing adaption until you have a "grid-independent" solution that can be well compared with experiment.

Regards AAA

Hi AAA, thanks for your interesse

Yes, when I say that, I mean that fluent detects an error in the mesh and it not recognizes it. I have solved a part of my problem, now I can create 100000 cells throughout of a linear pipe, and fluent works well, but my problem is in the curvated pipes, the mesh that I do in gambit not adjust it in the curves of the pipe. You know how could to make a mesh in the curves of the pipe?

thank you very much Nicolás