CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   about accuracy of the CFD (

Hannah liang August 30, 2002 12:24

about accuracy of the CFD
Hi, friends, do you compare the result of CFD with practical situation? I want to know how big difference between them because I often meet the situation in which I have to get the practical pressure drop of the products from CFD simulation and i always received some doubt from that.Thanks for some infomation.


Jan Rusaas August 30, 2002 14:40

Re: about accuracy of the CFD
I think you can get many different answers to that. It depends on how well you do your calculation, You actually have to do a calculation with three grids and make a grid convergence study. From the pressure drop for each case you have to check if they are within the asymptotic range. When that is fullfilled you can do a Richardson extrapolation of the pressure drop and get a realistic guess on the pressure drop. While you do that you also have to check that you use the correct turbulence model and check your Y+ values. When I compare the pressure drop I calculate I often get results that are say +-25% from values in handbooks - but on the other hand how accurate are the values in the handbooks. I think I could write another half page about inaccuracy in measurements.

I do think that people that are measuring almost make as many mistakes as people doing CFD. Remember that there are errors/deviations in a CFD calculation, but the user is the factor that probably introduces the larges errors :o))))))

Sorry, probably not very helpful answer, but interesting question. Jan

Hannah liang August 30, 2002 16:03

Re: about accuracy of the CFD
Hi, Jan, do you think, generally, Y+ should be within the range of 11 to 300? In cfx, if Y+ is much higher(I mean higher than 300), it seems that the mesh should be refined. Is it?


Dimitris September 3, 2002 01:34

Re: about accuracy of the CFD
Actually from what I''ve been reading, a Y+ value as high as 300 at some regions of the model is a bit too high I think.

Jan Rusås September 3, 2002 10:46

Re: about accuracy of the CFD
I think Y+ should be in between 30 and 100, but it depends on your problem. If you have a problem which is wall dominant then you might need y+ <=1 or other value depending on model. On the other hand if you have a system where the wall does not have any impact on the flow , at all - then I guess Y+ could be above 300. Jan

Robin September 3, 2002 11:26

Re: about accuracy of the CFD
Hi Hannah,

Y+ is a good general indicator and should be below 100 for the scalable wall functions or between 11 and 100 for standard wall functions.

For some very high Reynolds number flows, such as around a ship hull, a high Y+ value is acceptable. The most general rule is that you should have at least 5 nodes in your boundary layer for the k-e model wall functions. Thus, if you create a vector plot intersecting your boundary, you should see at least 5 vectors increasing in magnitude before you hit free-stream values. If this is not the case, you should increase your near-wall resolution.

Regards, Robin

All times are GMT -4. The time now is 18:48.