CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   Why double precision for Y+<1? (http://www.cfd-online.com/Forums/cfx/111270-why-double-precision-y-1-a.html)

 Omer January 2, 2013 07:46

Why double precision for Y+<1?

Hi ,

I am aerodynamic designer and I used CFD for some of my analysis. The Best practise guide http://www.cfd-online.com/Wiki/Best_...omachinery_CFD suggests I use double precision for Y+<1.

I would like to know why we need to do this. I know the aspect ratios when Y+<1 condition is met, can be very high. But I don't understand why we need to use double-precision. What are the underlying mathematics that needs double-precision.

Regards
Omer

 mjgraf January 2, 2013 12:09

http://www.cfd-online.com/Forums/cfx...html#post82819

all boils down to round off error for very small terms.

Quote:
 Originally Posted by Omer (Post 399728) Hi , I am aerodynamic designer and I used CFD for some of my analysis. The Best practise guide http://www.cfd-online.com/Wiki/Best_...omachinery_CFD suggests I use double precision for Y+<1. I would like to know why we need to do this. I know the aspect ratios when Y+<1 condition is met, can be very high. But I don't understand why we need to use double-precision. What are the underlying mathematics that needs double-precision. Regards Omer

 Omer January 2, 2013 12:34

Hi ,

Thanks mjgraf. I would appreciate a few more details to understand this in more dept.

I do understand it is to reduce the round-off errors. Single-precision gets us about 5 to 6 digits after the decimal, whereas double-precision gets us about 15 to 16 digits . But why does this become more pertinent when y+<1? Does this have to do with the very small cell volumes near the wall? And how does this affect our solution?

Regards
Omer

 mohw2002 January 2, 2013 17:24

Please can any body help me in post-cfx
I want to know how can I take values to draw the relationships yplus, k plus, epsilon plus.
can I find the values of friction factors directly from CFX or by analytical equations?
Thanks.

 ghorrocks January 3, 2013 19:04

Omer: This is all basic numerics. Whether double precision is required is problem dependant, but in general when you are resolving fine gradients in a large field you can run into round off error problems so double precision can help. Just try it with and without and find out the difference for yourself. Even better, write a little test program (doing the heat equation of somethign simple) and see for yourself the effects of numerical precision.

 Omer January 3, 2013 23:46

Thanks Glenn.

I will take up your suggestion and comeback with something. :)

Regards
Omer

 maaz January 21, 2014 01:22

hi,
i am working on fluent to obtain pressure coefficient on hip roof low rise building, i want to know how to obtain value of y plus above 30

 brunoc January 22, 2014 08:27

Quote:
 Originally Posted by maaz (Post 470986) hi, i am working on fluent to obtain pressure coefficient on hip roof low rise building, i want to know how to obtain value of y plus above 30
You should post this as a new thread on the Fluent forum.