[mohammad-mmm]

My project is about external flow around tube bundles (2D) . I want to get the Nusselt number around the tubes. Reynolds number is between 6000 & 16000. The model I used is realizable k-e . I want to Know what’s the best mesh around tubes so I need y+ range for my model.

I did a lot of research to find this range, but I have not reached the definitive answer .for example I find a article which used y+<3 for enhance wall treatment and 3<y+<10 for non-equilibrium wall functions.

I have two questions:
1. Is this model good for my geometry.
2. What’s the range of y+ for my model.
My second question is so importand for me.

The y+ that I've got from one of my mesh is between 2 and 15.

Thank you for your suggestions.[/SIZE]

[fluid23]

The y+ range that is acceptable is a product of the near wall model that you are using and not the turbulence model per se. As you noted, there is no DEFINITIVE answer. The article that you mention that has the conditional y+ limits is probably some kind of blended function. The traditional wall model (u+=y+) technically requires y+=1 but is generally considered to be fairly accurate up to y+<5, some people will use y+<3 to give them that warm fuzzy feeling about their results. I would be cautious about having y+<<1 as well. That can cause all kinds of problems as well.

Beyond this value cutoff value, a log law is used to get u+ from y+ in order to close the turbulence equations.

What wall model are you using?

[mohammad-mmm]

Originally Posted by MBdonCFD

The y+ range that is acceptable is a product of the near wall model that you are using and not the turbulence model per se. As you noted, there is no DEFINITIVE answer. The article that you mention that has the conditional y+ limits is probably some kind of blended function. The traditional wall model (u+=y+) technically requires y+=1 but is generally considered to be fairly accurate up to y+<5, some people will use y+<3 to give them that warm fuzzy feeling about their results. I would be cautious about having y+<<1 as well. That can cause all kinds of problems as well.

Beyond this value cutoff value, a log law is used to get u+ from y+ in order to close the turbulence equations.

What wall model are you using?

The model is realizable k-e and near wall treatment is non-equilibrium wall function

[fluid23]

For non-equilibrium, I think that 3<y+<10 is not unreasonable.