[swe704]

Hi all,

I am simulating a ventilation system in a room,but the main purpose in the case is studying the jet behavior,so this room only contains diffuser and exhaust.
I want to start from standard k-e and enhanced wall treatment,after several times modification on mesh file in Gambit, i reduce y plus in the range 1 to 5 or 6,but i want to control all them below 1 for enhanced wall treatment.

Because the dimension of the room is big,the total number of generated elements is around 4.8 million,so it takes pretty long time to converge.
My qustion is:Can i check y plus durning interation and go back gambit to correct or i have to wait untill the solution is converged to check?
Is it possible for me to have a "faster"way to know the relative correct y plus?

Thanks in advance.

[Ralf Schmidt]

Hi!

you can use a surface monitor of your yplus during iteration.
->solve -> monitor->surface

select the important walls and use the "area-weighted average" of yplus...
If that value does not change for a (larger) number of iterations, your solution is convergent anyway...

Ralf

[swe704]

Hi Ralf,

Thanks very much for you fast reply and quite uesfull advice.

For me,floor is the most important wall, since the jet comes out and impingings and spread over.

As you said,i can choose the monitor surface to see how does the y plus varies,before the converge is reached,can i judge the y plus is higer or lower.For example, after 4000 iterations(which is still not converged),i check the alr flow patten on the floor,it looks like the correct pattern,then i can take the moment y plus for refining mesh?

[Ralf Schmidt]

ok...

for judging convergence, you might use the residuals with "standard" Fluent convergence check... BUT, this standard criteria are not set properly depending on your problem, So turn the standard convergence criteria off.

For me, a sufficient convergence of the solution is given, when the parameters, I am interested in, are not changing during iteration any more.
Of cause, your residuals should not be to high... e.g. they should be below 10^-4 ....
But when your interesting parameter is (almost) constant, a change of the residuals from 10^-6 to 10^-7, will NOT give you any more accuracy!!!

So if you focus on flow field, create some point monitors (on points with high velocity gradients) and monitor the velocities on them. If they does not change any more, your solution is good enough!!!

One additionally remark: you don't have to go back to gambit to change the grid.
Use ->adapt->yplus , select the walls and the desired yplus range... Fluent will create some finer mesh to achieve the desired yplus. But, this is only in the areas, where your yplus is out of the specified range.
That can result in a high jump in your yplus.... You might get better results by adapt the whole region close to the wall, e.g. every 3 cells distance from the wall.

Ralf