I am simulating (CFX) a pipe with a turbulence promoter in it. I am using RNG k-e turbulence model yplus is lower than 1. For mesh independency analysis I am using turbulence kinetic energy as a output. The problem is, as I increase the mesh size, relative difference is getting higher up to certain mesh size (solution is getting mesh dependent). After that mesh size the relative difference is decreasing. For example from 4 to 12 million, relative change increases and from 12 to 25 million, it decreases. This values are coming because of the region where is just after the turbulence promoters (dominant anisotropic flow).
What could be the reason?
I am not sure about the focus of your question but the grid-convergence analysis generally has no monotonic behavior for arbitrary grid size. You could expect oscillations. Then, using the RANS formulation, the magnitude of the turbulence model is prevalent over the local truncation error that depends on the grid size.
That means also that the total kinetic energy would increase when further high wavenumbers (that is a reduced grid size is used) are added, as happens in DNS/LES, but you have that the RANS model covers such effect.
Without more details and some plot of the variables you are observing, it is difficult addressing the issue.
What I see is a plot along the streamwise direction x. Is the kinetic energy integrated along the section at any position x? If I understand the plot, from 15 to 20Ml the curves are becoming similar
Filippo Maria Denaro I am dividing X streamwise into 100 cross section and I am calculating area average TKE for every cross section. I want to use actually 13m mesh because of the computational effort ( from 13.2 to 15m the relative difference is again under 5%). 13.2 m is wrong in that case? And which I could not understand is that according to the table why it is incerasing and decreasing after 13m mesh?
The flow problem has a tool in the pipe that causes turbulence to correlate behind and that is quite challenging while using RANS.
Could you detail al the geometry?
- Badges
- Science topic