Convergence of flow past finned body
I'm running a 2D simulation of flow pass a finned body similar to the classic problem of flow past a cylinder. I'm working with Re #'s starting at 5. x10^5. When I work with a laminar model i get 'convergence' as my residuals are all below 0.01. but that doesn't seem right.
When I try a turbulent model (primarily working with k-omega, though I feel as though I've tried them all) the Tke and Sdr ends up on the order of 100 by the end of a 25 second simulation.which is at times taking me several days to run:mad: I feel like I've looked everywhere for the issue: -I've checked the 2D mesh for cell quality, skewness etc. and everything seems find there -I've taken the time step down to as low as 1.0 x 10^-5 -The inner iterations up to 25 - I've tried lowering the under-relaxation factors for the k-omega turbulence solver to .2 and the k-omega turbulent viscosity to .2 the segregated velocity to 0.4 and pressure to 0.1 I can't seem to figure out what's going on and it's driving me nuts are there any other thoughts or suggestions here?:confused: |
I have a similar problem. It would be great if anyone can give recommendations!
|
You should describe your simulation more I think.
Does the residual stay there or keep increasing? What about other residual? Or maybe check something more physical such as drag coefficient, do they converge? What about the boundary layer mesh? y+? Any error during running such as limited turbulent viscosity ratio or any other error? |
3 Attachment(s)
Quote:
the residuals hover around a significant value by the time i'm done with 3 different Re values. they are ramped to minimize 'shock' on the model over 0.5 seconds. drag coefficent has stabilized, lift coefficient has stabilized. i'm expecting lower values of drag than i am getting, what kind of cell count would be reasonable for a simulation like this. i've gotten as high as 500k cells. that seems a bit excessive, though when comparing monitored values like C_d and C_L_rms ans St they don't seem to agree, which would suggest increasing mesh resolution, but these are taking me some serious time to run, more than iwould expect, but i am new so what do i know!:D as always help is much appreciated |
Quote:
|
the problem with high residuals, i've been able to narrow down to need for more inner iterations per timestep (up to 50). :o this has since solved my problem. i hope this helps others in the future
|
All times are GMT -4. The time now is 11:00. |