NaNs in the residuals history

AEMD · July 17, 2013, 11:47

Hi,
When trying second order schemes on SU2 v2.0.5, I observed a curious behavior in the reporting of the residuals for most of my cases (though it is not systematic).
I get all residuals changing suddenly from reasonable values to NaNs in the convergence history file while the calculation continues and exits when reaching the convergence criteria.
Moreover, although the residuals written on the command prompt and in the history files are reported as NaNs, the maximum residual value is still calculated (and is non-NaN). All the variables and residual values in the domain are non-NaN, according to the volume flow data file.
So, the computation seems not to diverge but the value of the residual in the convergence history file is sometimes reported as a NaN.
Did someone observe the same behavior?
Thanks.

rktchip · July 25, 2013, 13:49

nan's and inf's anywhere in the solution or history is generally a sign of a failed solution. when seeing these, check your mesh and numerics options in the config file.

July 17, 2013, 11:47	NaNs in the residuals history	#1
AEMD New Member AD Join Date: Feb 2013 Posts: 7 Rep Power: 13	Hi, When trying second order schemes on SU2 v2.0.5, I observed a curious behavior in the reporting of the residuals for most of my cases (though it is not systematic). I get all residuals changing suddenly from reasonable values to NaNs in the convergence history file while the calculation continues and exits when reaching the convergence criteria. Moreover, although the residuals written on the command prompt and in the history files are reported as NaNs, the maximum residual value is still calculated (and is non-NaN). All the variables and residual values in the domain are non-NaN, according to the volume flow data file. So, the computation seems not to diverge but the value of the residual in the convergence history file is sometimes reported as a NaN. Did someone observe the same behavior? Thanks.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
under-relaxation factors -> level of residuals	Zigainer	FLUENT	19	July 21, 2017 17:53
motorBike Residuals for SST k-omega... and mine	JR22	OpenFOAM Running, Solving & CFD	6	August 1, 2013 10:08
what to monitor besides residuals?	franzdrs	FLUENT	5	March 21, 2013 04:59
judging convergence through residuals	MachZero	Main CFD Forum	7	December 25, 2012 13:18
Convergence - scaled vs unscaled residuals	HS	FLUENT	1	November 7, 2005 06:45

July 25, 2013, 13:49		#2
rktchip Member Trent Lukaczyk Join Date: Feb 2011 Location: Stanford, CA Posts: 75 Rep Power: 15	nan's and inf's anywhere in the solution or history is generally a sign of a failed solution. when seeing these, check your mesh and numerics options in the config file.