[ck3]

I want to investigate a fluid flow over a circular cylinder for Re=3900. I use k-ebsilon Realizable turbulent model and steady solver.
I construct geometry for circular domain (with at least 60xdiameter of smaller circle).
y+=1, so I use Enhanced wall treatment.
for boundary conditions (see attached pictures).
Could you please help me to model this case? (How to seting boundary conditions to run this simulation)

[ck3]

Any help please.

[Leonardo.flores]

It's not very common to see that circular pressure outlet bc. It's much common whith a wall up and down and a pressure outlet perpendicular to the streamwise velocity.... Check this paper https://www.researchgate.net/publica...and_Validation

That way you could move the pressure outlet further or closer, depending on your interest on the wake analysis.

On the inlet condition I use a turbulent intensity of 0.5% since I compare results with a wind tunnel with that specifications... 1% is ok. And length scale I consider 0.038*cylinder diameter.

I also set a 0 gauge pressure in the the outlet... And a backflow turbulent Intensity and length scale same as inlet. I am not sure about this backflow setup but have no problems by now.

[ck3]

Hi Leonardo.
First, I want to thank you for your replay.I need your help for these two questions:
1. What is the best type of mesh to use (C, H or O-grid)?

2. what model of turbulence you advise me to useRegards,

[Leonardo.flores]

1. Most papers I read use structured rectangular grid with a O-grid of the cylinder. Check this video tutorial of a mesh made in ICEM CFD Structered Grid on Cylinder in ICEMCFD. I will share you a kinda refined mesh I use with a Y+=1 for a Reynolds= 140 000. https://drive.google.com/file/d/1PJu...ew?usp=sharing

So I think this is the best mesh type. You can control the blockage ratio (diameter/Mesh height), downstream lenght, and many others.

2. I am using K-w SST, since it has k-w and k-epsilon features. Its very common for bluff bodies. But, it might take many computational resources for high Reynolds since you need a y+=1-5. If you dont have a good cpu you can start with k-epsilon y+=30. Check this paper, they perform simulations with 3 kind of k-epsilon models Tutar - Cylinder
After performing your initial simulations with k-epsilon or k-omega SST go for transitional models... and then for LES.

[ck3]

thank you very much, I will try to follow your advice.

[Leonardo.flores]

Originally Posted by ck3

thank you very much, I will try to follow your advice.

Np, let me know your results if you can, I am interested on Re=3900. What parameters are you monitoring?