Hello!

In the user guide, it is written that a value of y-plus 30 to 100 is recommended. I have tried refining one of models and after refining my tetra model the y-plus value decreases into the range as recommended. Does this always mean that a decreasing y-plus value (by means of refining) mean a better calculation? And if so, why is there a lower limit of 30 in the recommendation? I have read published works that have y-plus of 0.1 or 2. How small should my y-plus be?

On another model which consists of super-fine all hexa mesh (hexa mesh cell size ~10 times smaller than my refined tetra mesh), the y-plus value is strangely high (max is around 100000) but the cell data for velocities (the results) look more logical compared to velocities from both of my tetra model.

Can someone help explain the meaning of y-plus caused by tetra/hexa mesh in relation with logical or actual results.

Look up any good fluid mechanics book for the definition/background for Y+. The range of valid Y+ values will depend on the type of turbulence model you are using and how the wall functions have been coded. There are now hybrid wall fuctions available in star as a beta feature, where the value of Y+ can be in the range from approx 1 to 300. Strictly speaking, tets are not a good idea next to boundaries especially where heat transfer is concerned. Get a hold of Best Practice Guidelines, published by ERCOFTAC, available from their website, to get a good grounding on all this stuff.

Allan

The ercoftac tip is a good one. I will try to acquire it. Saw some abstracts from ercoftac about y-plus being in the range of 30-500 as well, something like the StarCD user guide. In the meantime, just to share info for potential novices next time around, I came across a kind and informative forum on Fluent at

http://www.cfd-online.com/Forum/flue....cgi?read=1672

just in case anyone wants a quick scoop of the small y-plus (y+=1) that I was talking about earlier. I can't help but wonder this time if y-plus is software-dependent?

Ciao!

Hi Azmer,

Acceptible y+ values are turbulence model dependant, not software dependant. In STAR, models are either called "High Reynolds Number" or "Low Reynolds Number" models. If the model is a low Reynolds number type then the cell closest to the wall is assumed to be in the viscous sublayer. In which case y+ values should be ~ 1. High Reynolds number models assume that the cell nearest to the wall is outside of the viscous sublayer in the turbulent region, in which case y+ values should be in the 30 - 500 range.

As Alan said, there is now a Hybrid model which blends these two approaches, and hence y+ can be from 1 - 500.

For further details see the ercoftac best practices guide or any good CFD text book.

Hi.

I was very interested in Hybrid model. How can I switch on it?

Daria

it's not in v3150A, but in some patches. You need to select a low reynolds number turbulence model and set c125=1. Write geometry and problem files and then starlink. When the analysis starts it should output to the screen that the hydrid wall functions have been activated. If this does not appear then they are not included in your patch. I know they're in v3150a.008 and v3150a.028.

Allan

First, thank all for valuable inputs. I recently tried to activate the "hybrid" function on my version (as explained by allan) but I think it was not activated successfully. Don't know what patch I am using.

To progress, I am now trying a recommendation from the main forum (http://www.cfd-online.com/Forum/main....cgi?read=5136). I am now going to use a two-layer model and activate the respective walls. Will see how that works. Is two-layer model similar to a hybrid wall function?

At the same time I wonder, if I choose a Low-Re turbulence model instead, will the calculations in the log-law region outward be auto-compensated in some way? (I don't know the distance/extent of my viscous linear sublayer).

Azmir

You should try to persist with the hybrid wall function, as good results can be obtained for pactically no extra cpu cost. If you look in the run file the first line will tell you what version of the code you're running, something like:

STAR 3150A.028 [Bakerloo_202]

This is patch 28.

Cheers

Allan