CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   General grid/element size and time step questions (

stu May 24, 2007 02:34

General grid/element size and time step questions
Just a question regading grid size and time steps for transient analysis.

Question : In order for the fluid from one cell to reach the next cell, does the time step need to be long enough for the distance=velocity*time step relationship to be true? eg if my time step and input velocity is such that ts=0.1 secs, input vel = 10 mm/s and I have a gid/mesh/element size =10mm, hence in 0.1 secs, the fluid has 'travelled' 1 mm, it has not reached the next element by the end of the time step. Will this work? Can you run a simulation like this?

What happens if the reverse is true, the input vel was 800 mm/s, the fluid would have travelled across 8 elements in the time step. Would this work, can you run a sim like this, or would the results be useless?

Are there any general rules that govern velocity, mesh size, and domain size?

Thanks Stu

Joe May 24, 2007 05:37

Re: General grid/element size and time step questi

CFX is an implicit code so theoretically the Courant number (look this up) can be arbitrarily large.

Generally though try to keep the RMS Courant number under 5.

Ultimately time step selection must be governed by timestep independence studies analogous to mesh independence studies.

All times are GMT -4. The time now is 09:46.