|
[Sponsors] |
June 24, 2020, 14:47 |
Getting very low Y+ for K-Omega SST
|
#1 |
Senior Member
MA
Join Date: Mar 2020
Posts: 163
Rep Power: 6 |
Hi Everyone,
I have generated a mesh (not the first though) in STARCCM+ where I observed that the Y+ value was low everywhere except the tip of the blade, so I took the geometry to surface repair and removed all the errors. Now, I kept the surface cell size almost the same for all the boundaries of my domain but the Y+ values appears to be very low i.e. 0.0000XX however I tried to bring it close to 1 by increasing the target surface size for blade mesh from 0.05 m to 1 m but still the y+ is very low. I have attached the figures for prism layers, and blade mesh. Why is that happening? In addition, I tried to reduce the Minimum Thickness Percentage from 10 (by-default) to 1% and Layer reduction Percentage to 2% from 50% (by-default) but it didn't work. Thanks for your time. Best regards, |
|
June 25, 2020, 04:05 |
|
#2 |
Senior Member
|
Very low Y+ values should not be of any trouble to a properly coded K-Omega model (whatever flavor, SST included). Of course, you are certainly wasting resources at that resolution, but that's another problem.
Still, I have never opened STARCCM+ in my whole life, so I have no idea about how it works. |
|
June 25, 2020, 04:54 |
|
#3 | |
Senior Member
MA
Join Date: Mar 2020
Posts: 163
Rep Power: 6 |
Quote:
Well. As far as computational effort is concerned, I have two meshes of 4.5 M and 9 M both have the Y+ of same order (very low). I am a little concerned about it because it is recommended to have Y+ closer to 1. |
||
June 25, 2020, 05:05 |
|
#4 |
Senior Member
|
I understand your concern, and these reccomendations are typically written in such bad formats that confusion is very common, so let me put it differently.
Working without wall functions (which is ok considering your y+ range), you should be able to do proper grid refinement and get down with the grid indefinitely. If that's not possible, a software manual should state that very clearly as it would be a very serious concern. Also, if you go trough the original (and followinf) papers of all the models that support low y+ values, none of them states that y+ should be higher than some threshold. In conclusion, I'm very confident that you don't have to worry about this when using the major commercial CFD codes out there. You can interpret the manual statement as: "you need 2k bucks monthly to live in this city". This doesn't mean that 10k bucks are going to hurt you. But I agree that using "at least" would make the statement clearer |
|
June 25, 2020, 05:19 |
|
#5 | |
Senior Member
MA
Join Date: Mar 2020
Posts: 163
Rep Power: 6 |
Quote:
Thanks. I noticed that when I just intialized the solution the Y+ values were in the range 0.0000XX but after running 500 iterations I see that Y+ values are between 3-7 everywhere on the blade and only at the tip Y+ > 10. I have attached the pictures here: https://drive.google.com/drive/folde...yp_3R7fksC_wHu Now, what do you think is a good way to confirm Y+, by running the simulation for a few hundred iterations or just initializing the solution? Secondly, do I need to refine the tip to bring the Y+ down or it's something related to curvature? Thanks for your time. |
||
June 25, 2020, 05:22 |
|
#6 |
Senior Member
|
Y+ always checked at complete convergence, there is no sense in doing it any early because, well, it will change. But I guess you could give some more density to that surface mesh near the tip.
|
|
June 25, 2020, 05:54 |
|
#7 | |
Senior Member
MA
Join Date: Mar 2020
Posts: 163
Rep Power: 6 |
Quote:
Alright. However, when I calculate the first layer thickness near the wall from Turbulent boundary Layer (Prandtl) I should use the target value for y+ = 1 right? |
||
June 25, 2020, 07:34 |
|
#8 |
Senior Member
|
I don't know if I understood you correctly. If you meant: what y+ should I use in analytical formulas that give me the correct wall spacing to achieve my desired y+? Then, it is just up to what value of y+ you want. If you want y+=1 (I guess so), then that's the value (but you might want, say, 50, for a number of reasons, so it's not fixed, otherwise you had no free y+ symbol in the formula in the first place).
|
|
June 25, 2020, 07:49 |
|
#9 | |
Senior Member
MA
Join Date: Mar 2020
Posts: 163
Rep Power: 6 |
Quote:
Yes exactly that's what I meant "using y+ value in analytical equations to get desired y+ which in my case is 1". Thanks a lot. |
||
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
RSM low Re Stress - Omega convergence issues | Jonathan | FLUENT | 1 | April 20, 2016 05:33 |
BC's for k-omega SST (low / hi Re setups) | Jonathan | OpenFOAM Running, Solving & CFD | 7 | February 10, 2014 15:36 |
SST Model for Low Re ... some minor settings | mrenergy | FLUENT | 7 | January 18, 2013 21:13 |
Wall function implementation K Omega SSt | cbarry | OpenFOAM | 3 | August 18, 2009 10:09 |
How to model Low Reynolds Turbulent in SST K-omega | asghari | FLUENT | 0 | August 18, 2007 03:31 |