Hi

I am now doing analysis on flow in a pipe by using k-epsilon RNG method. I understand that in normal practice, people always set turbulent length scale to about 10% of the inlet diameter of inlet.

My question is, let say if the pipe is symmetry and I only do calculation on half of the pipe, should I use diameter of inlet, or radius of radius in determining my turbulent length scale?

Use the real geometry data, so ther real diameter, to determine the lenght scale...you model just an half of the pipe for convenience but the model intrinsecally take account of the complete geometry (cyclic boundaries).
Bye!
Andrea