If my solution is converged ?

jing113cn · September 8, 2009, 04:35

I want to use the time-dependant solver to get the steady state. And I also monitor some important parameters, such as temperature, vof...
After some iterations, I found these parameters almost do not change, but the residuals remains at a high value(10-2) and goes flat. I wonder if I my problem is converged ?
The residula and monitor look like this >
vof.JPG residual.JPG

srjp · September 9, 2009, 08:28

Residuals becoming flat may indicate:
1. Your solution has converged, or
2. This is the minimum residual possible with your current grid.

For further reduction in residuals, you may have to refine or adapt the grid.

delfel · September 9, 2009, 14:08

Yeah, I agree with srjp. You're going to need to refine your grid. I would recommend plotting contours of mass imbalance (since your continuity residuals didn't converge) and refining where they are high.

-sean

denizen · September 10, 2009, 01:39

you both wrong.
this is transient solution and if one decrease the time step the residuals will decrease too.
you execute 1 ineration on 1 time step?

-mAx- · September 10, 2009, 02:03

the continuity residuals are too high.
reduce your time step

srjp · September 10, 2009, 09:16

Sorry, I thought it was steady state simulation.
May be after frustration of seeing so many flat residuals !!!
Agree with mAx on the timestep reduction.

delfel · September 10, 2009, 11:00

ditto.

-sean

jing113cn · September 11, 2009, 02:51

Yes, 1 interation on 1 time step at beginning, because I want save computation time. And then I increase the interations each time step, and the residuals also remain flat.

Quote:

Originally Posted by denizen

you both wrong.
this is transient solution and if one decrease the time step the residuals will decrease too.
you execute 1 ineration on 1 time step?

-mAx- · September 11, 2009, 03:39

forget 1 iteration per time step
try the default settings, eg: 20 iterations per time step
I would also recommend you to start your unsteady computation from a well converged steady solution

ivanbuz · September 11, 2009, 17:46

Most likely you are using a too big time-step. The time in your pictures is 2.4e4. Imagine if time step=1e-6, how many steps it will be....

teguhtf · September 16, 2009, 09:21

Hello!!!
I'm a new user in FLUENT. I've read your Post. I wanna know what is the indication that we have success iteration??
I've thought that FLAT graph on the Residual graph is the indication!!!
Am I wrong???

jing113cn · September 17, 2009, 04:07

I think when the we get converged solution, the residual will go flat. But at the same time, the residual should fulfill the criterion, such as 10e-4,10e-5...

denizen · September 17, 2009, 06:59

Flat or not it is no matter. If iteration converged then it is. I think so.

Regards.

teguhtf · September 28, 2009, 03:29

Is there an equation to prove it?? just to know how much convergence value!!!

denizen · September 29, 2009, 01:33

there is navier-stokes equation.
and http://en.wikipedia.org/wiki/Residua...rical_analysis)

regards.

September 8, 2009, 04:35	If my solution is converged ?	#1
jing113cn New Member Join Date: Jun 2009 Posts: 22 Rep Power: 16	I want to use the time-dependant solver to get the steady state. And I also monitor some important parameters, such as temperature, vof... After some iterations, I found these parameters almost do not change, but the residuals remains at a high value(10-2) and goes flat. I wonder if I my problem is converged ? The residula and monitor look like this > vof.JPG residual.JPG

September 10, 2009, 02:03		#5
-mAx- Super Moderator Maxime Perelli Join Date: Mar 2009 Location: Switzerland Posts: 3,297 Rep Power: 41	the continuity residuals are too high. reduce your time step __________________ In memory of my friend Hervé: CFD engineer & freerider

September 11, 2009, 03:39		#9
-mAx- Super Moderator Maxime Perelli Join Date: Mar 2009 Location: Switzerland Posts: 3,297 Rep Power: 41	forget 1 iteration per time step try the default settings, eg: 20 iterations per time step I would also recommend you to start your unsteady computation from a well converged steady solution __________________ In memory of my friend Hervé: CFD engineer & freerider

September 16, 2009, 09:21	Character of success iteration???	#11
teguhtf Senior Member teguh hady Join Date: Aug 2009 Location: Saga, Japan Posts: 222 Rep Power: 17	Hello!!! I'm a new user in FLUENT. I've read your Post. I wanna know what is the indication that we have success iteration?? I've thought that FLAT graph on the Residual graph is the indication!!! Am I wrong???

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Solution converged, and then diverge again.	jrg	FLUENT	2	April 1, 2013 23:10
IcoFoam parallel woes	msrinath80	OpenFOAM Running, Solving & CFD	9	July 22, 2007 02:58
Calculate a converged flow solution	kumar	FLUENT	2	May 24, 2007 01:33
Could anybody help me see this error and give help	liugx212	OpenFOAM Running, Solving & CFD	3	January 4, 2006 18:07
converged solution or correct solution	SAM	FLUENT	1	October 12, 2004 15:38

September 9, 2009, 08:28		#2
srjp Member JP Join Date: Mar 2009 Posts: 57 Rep Power: 18	Residuals becoming flat may indicate: 1. Your solution has converged, or 2. This is the minimum residual possible with your current grid. For further reduction in residuals, you may have to refine or adapt the grid.

September 9, 2009, 14:08		#3
delfel New Member Sean Delfel Join Date: Aug 2009 Posts: 27 Rep Power: 16	Yeah, I agree with srjp. You're going to need to refine your grid. I would recommend plotting contours of mass imbalance (since your continuity residuals didn't converge) and refining where they are high. -sean

September 10, 2009, 01:39		#4
denizen New Member Join Date: Mar 2009 Posts: 25 Rep Power: 0	you both wrong. this is transient solution and if one decrease the time step the residuals will decrease too. you execute 1 ineration on 1 time step?

September 10, 2009, 09:16		#6
srjp Member JP Join Date: Mar 2009 Posts: 57 Rep Power: 18	Sorry, I thought it was steady state simulation. May be after frustration of seeing so many flat residuals !!! Agree with mAx on the timestep reduction.

September 10, 2009, 11:00		#7
delfel New Member Sean Delfel Join Date: Aug 2009 Posts: 27 Rep Power: 16	ditto. -sean

September 11, 2009, 17:46		#10
ivanbuz Member Ivan Join Date: May 2009 Posts: 85 Rep Power: 16	Most likely you are using a too big time-step. The time in your pictures is 2.4e4. Imagine if time step=1e-6, how many steps it will be....

September 17, 2009, 04:07		#12
jing113cn New Member Join Date: Jun 2009 Posts: 22 Rep Power: 16	I think when the we get converged solution, the residual will go flat. But at the same time, the residual should fulfill the criterion, such as 10e-4,10e-5...

September 17, 2009, 06:59		#13
denizen New Member Join Date: Mar 2009 Posts: 25 Rep Power: 0	Flat or not it is no matter. If iteration converged then it is. I think so. Regards.

September 28, 2009, 03:29		#14
teguhtf Senior Member teguh hady Join Date: Aug 2009 Location: Saga, Japan Posts: 222 Rep Power: 17	Is there an equation to prove it?? just to know how much convergence value!!!

September 29, 2009, 01:33		#15
denizen New Member Join Date: Mar 2009 Posts: 25 Rep Power: 0	there is navier-stokes equation. and http://en.wikipedia.org/wiki/Residua...rical_analysis) regards.