★ how to generate gird accoring to Y+?
 

Hi, in Fluent, before calculating the flow field, how to generate gird according to Y+?

for example, if Re=20,000, may I use "enhanced wall treatment" ? if so, how can I set the grid points near the wall? and how many points?

thanks

Re: ★ how to generate gird accoring to Y+?
 

for complicated geometry, it needs much time to generate grid, but I have no idea what is the Y+ when I place grid points. for example, if 50-50-1000 is used, the Y+ is 5 after the solution is converged. Now, what is the guidelines for generate gird again? or using "enhanced wall treatment" ?

Re: ★ how to generate gird accoring to Y+?
 

You can make a coarse grid, solve it initially, then adapt the grid according to y+ values. For example adapt all cells where y+ (or y*) is above 5.

EWT is fine when wall cells are very fine, i.e. y+=1.

May be worth starting off the solution with standard wall functions, then adapt the grid and switch to EWT once y+ values are alot smaller.

Hope this helps, Phil

Re: ★ how to generate gird accoring to Y+?
 

thank you Phil.

i first computed with use standard wall functions, and plot the Y+ along the wall, most of Y+ are less than 4, but some of them above 4,up to 12. In this case, what can I do ?

Re: ★ how to generate gird accoring to Y+?
 

conan,

go into the adapt menu and select yplus/ystar. This facility will adapt(refine) the grid at only the areas where y+ is above a certain range. You can then refine all grid points where y+ >2 say, but 12 isn't actually very high so you may not actually want to do this.

For standard wall functions you want y+>30, but as close to 30 as possible.

For EWT you want on the order of y+=1.

I think you only need to adapt your grid if the dense regions are of particular interest - it won't make a huge difference to the results.

Phil

Re: ★ how to generate gird accoring to Y+?
 

thank you, Phil.

in some papers, there mentioned "to resolve the near wall viscous region, 10-grid points were placed in boundary near wall", in FLUENT, as I known, to generate boundary grid, first distance to wall and the scaling ratio should be given. how did they do so that Y+ being around unity ?

Re: ★ how to generate gird accoring to Y+?
 

also, by the way.

if i adapt grid with standard wall function, in the adapt menu, should I choose ystar not yplus? and should the min be set 30, and max be 60 ?

if i adapt grid with enhanced wall treatment, in the adapt menu, should I choose yplus not ystar? and should the min be set 0.5, and max be 1 ?

thank you

Re: ★ how to generate gird accoring to Y+?
 

I don't think you can really make the initial grid according to y+ as its a flow dependent/solution dependant value. There's probably an equation for it but to be honest I don't think you need to worry so much about the initial grid with standard wall functions. They aren't all that accurate anyway and basically just model a boundary layer within the wall adjacent cells. They are valid for y+ of around 30 unpwards. I'd just make the original grid as good quality as possible, and hexahedral.

I always just use y+, with y+ of around 1 for EWT, where possible. I don't advise going to millions of cells just to resolve y+ of 1.

Re: ★ how to generate gird accoring to Y+?
 

thank you, Phil. what is your email? i want to discuss further with you.

Maybe my original grid is not so good. Even if I adapt the grid in adapr menu, the results is almost not improved.

Re: ★ how to generate gird accoring to Y+?
 

Results might not change because the Standard wall functions are not so bad for simple cases. But in one of your messages I saw y+ obtained is 4 while using standard wall function which is not a good thing to have, because condition to use SWF is that you should not have any node in viscous sub-layer. In fact, first node from wall should be placed where you may have wall y+ >30 (Phil has mesntioned it somewhere). y+<5 is good enough to explore any of the turbulence model. You can only try to get y+~1.

In your case, refinement based upon adaption suggested by Phil should work. But with current y+ <4, you may use near wall treatment option.