Residuals

Clémence · June 5, 2014, 08:29

Hi,
I am running a implicit transient simulation, wih a time step of 0.1 et maximum inner iterations of 10.
The residual is very weird, with a lot of oscillations. And change the time step doesn't help..
Do you think that I have to change the values of L and I (from k-epsilon model) or is it something else?
Thanks for your help
Clem

me3840 · June 6, 2014, 16:15

Residuals with oscillations are expected for transient calculations, but they should all have the same period.

Yours do not, which may indicate your simulation is not converging with every timestep.

Clémence · June 10, 2014, 03:22

Thanks Me3840! But chat should I do to make my simulation converge at each iteration?
Change the value of L and I, or change the inner iteration step?

me3840 · June 10, 2014, 10:32

There are three major controls on timestep convergence outside of remeshing and physics:

1. Timestep size (dt)
2. URFs/Courant number
3. Number of inner iterations

For a fast calculation, 1 should be large, 2 should be large, and 3 should be small. For a stable one, everything reverses.

Now if you find that no combinations or an unreasonably small timestep will make your simulation converge, you need to address physics settings and BCs and mesh. I should note that you should not rely completely on residuals to tell the whole picture. You need to monitor engineering quantities (drag, lift, pressure drop, etc) and ensure they too converge for each timestep.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
under-relaxation factors -> level of residuals	Zigainer	FLUENT	19	July 21, 2017 16:53
motorBike Residuals for SST k-omega... and mine	JR22	OpenFOAM Running, Solving & CFD	6	August 1, 2013 09:08
judging convergence through residuals	MachZero	Main CFD Forum	7	December 25, 2012 12:18
Strange residuals of the Density Based Solver	Pat84	FLUENT	0	October 22, 2012 15:59
Convergence - scaled vs unscaled residuals	HS	FLUENT	1	November 7, 2005 05:45

June 6, 2014, 16:15		#2
me3840 Senior Member Join Date: Nov 2010 Location: USA Posts: 1,232 Rep Power: 24	Residuals with oscillations are expected for transient calculations, but they should all have the same period. Yours do not, which may indicate your simulation is not converging with every timestep.

June 10, 2014, 03:22		#3
Clémence New Member Join Date: May 2014 Posts: 12 Rep Power: 11	Thanks Me3840! But chat should I do to make my simulation converge at each iteration? Change the value of L and I, or change the inner iteration step?

June 10, 2014, 10:32		#4
me3840 Senior Member Join Date: Nov 2010 Location: USA Posts: 1,232 Rep Power: 24	There are three major controls on timestep convergence outside of remeshing and physics: 1. Timestep size (dt) 2. URFs/Courant number 3. Number of inner iterations For a fast calculation, 1 should be large, 2 should be large, and 3 should be small. For a stable one, everything reverses. Now if you find that no combinations or an unreasonably small timestep will make your simulation converge, you need to address physics settings and BCs and mesh. I should note that you should not rely completely on residuals to tell the whole picture. You need to monitor engineering quantities (drag, lift, pressure drop, etc) and ensure they too converge for each timestep.