CFD Online Discussion Forums

CFD Online Discussion Forums (
-   Main CFD Forum (
-   -   grid spacing question (

clifford bradford November 8, 1999 16:21

grid spacing question
i'd like to generate a structured grid for a straight cylindrical pipe (compressible viscous) where the first grid point away from the wall is to be at y+ = 1. how can i calculate the actual distance this represents. i have some ideas already but i'd like to see if anyone has better ones. thanks in advance

Steve Ciesla November 8, 1999 16:32

Re: grid spacing question
Check out this link

Alton J. Reich, P.E. November 8, 1999 16:52

Re: grid spacing question
I usually estimate a dy based on a correlation for a flat plate with good results for both cylindrical (pipe) and irregular geometries:

1. Calculate the Reynolds number = rho*V*D/mu 2. Estimate the coefficient of friction for a flat plate Cf=0.025/Re^(1/7) 3. Estimate dy from the definition of y+ dy=(y+)/(Re*sqrt(Cf/2))

John C. Chien November 9, 1999 11:00

Re: grid spacing question
(1). In the fully-developed pipe flow region, you can use the pipe flow solution to estimate the y value at Y+=1. That is you need a solution first in order to compute the y value. (2). In the pipe entrance region, which can be rather long for high Reynolds number flow, you have to use the estimated boundary layer thickness to guide your mesh. (3). Actually, this is only the begining, and you have to develope a series of meshes based on the actual computed flow field to get the Cf and the y value at Y+=1. So, the initial grid spacing does not have to be very accurate in Y+=1. The formula given in Steve and Alton's message give roughly the same value, but it does not address the pipe inlet mesh problem. But I guess, these are only for initial mesh distribution.

clifford bradford November 9, 1999 11:13

Re: grid spacing question
thanks for the tip John. i'm lucky in my case because the pipe inflow is fully developed

Long Hung Dong November 10, 1999 19:40

Re: grid spacing question
Most internal pipe flows can be considered as a flat plate for y+ estimations. For example, a wind tunnel (= square pipe) is generally considered to be a flat plate. Also, the initial guess for y+ of 1 may be wrong but initial simulations can give you a better value if your initial guess is based on something physical as opposed to a guess.

Thanks much, Long Hung Dong

All times are GMT -4. The time now is 20:05.