Initialization issue

nikhil · June 25, 2010, 15:00

Hello,

I have came across one initialization issue in Fluent. My problem is explained as following:

I am running one simple 2D transient case. At first i initialized the domain from inlet and solved for five time steps. Solution was converged at each time step. At the end of five time steps i got one particular solution. (1)

Next time, i solved the same case (without a single change) by initializing it from 'All zones'. This time also, i solved it for five time steps and each time step was converged. When i compared the solution from this case to the solution of previous case (1), i was astonished to see that these two solutions were totally different. I don't know which one to trust.

I was under the impression that, even if u initilize from different domain, you should get fixed solution at the end of certain time step, if u achieve convergence at each time step.

Can anyone explain me why this is happening?

Thanks,
Nikhil

jack1980 · June 26, 2010, 06:29

Hi, probably five time steps is not enough to reach a steady solution. What you want to do is monitor a certain flow variable (perhaps the drag coefficient) versus flowtime. When your monitor reaches a constant value, this usually indicates you have reached a steady solution.

Good luck!

nikhil · June 26, 2010, 19:29

Hi Jack,

I know that i am not running the case till steady state. I am interested in what happens after 'x' time instant and my point is after 'x' time instant different initialization provides different results. I am trying to understand why it is so.
Also, till i reach 'x' time instant i make sure that i achieve convergence at each time step.

Hope this clarifies my question.

Nikhil

jack1980 · June 27, 2010, 05:16

Hi,

In most cases you can only expect the steady state to be the same. The path along which you get there generally depends on the initial condition.

Think of the temperature distribution of a simple rod, with temp 1 at the left hand side and temp 2 on the right hand side. Now either initialize the complete rod with temp 1 or temp 2. Surely the paths will differ but the final solution will be the same.

Convergence of each time step doesn't mean the solution is converged in time, only that you are progressing accurately in time.

Good luck!

June 25, 2010, 15:00	Initialization issue	#1
nikhil New Member Nik Join Date: Mar 2010 Location: Michigan Posts: 25 Rep Power: 16	Hello, I have came across one initialization issue in Fluent. My problem is explained as following: I am running one simple 2D transient case. At first i initialized the domain from inlet and solved for five time steps. Solution was converged at each time step. At the end of five time steps i got one particular solution. (1) Next time, i solved the same case (without a single change) by initializing it from 'All zones'. This time also, i solved it for five time steps and each time step was converged. When i compared the solution from this case to the solution of previous case (1), i was astonished to see that these two solutions were totally different. I don't know which one to trust. I was under the impression that, even if u initilize from different domain, you should get fixed solution at the end of certain time step, if u achieve convergence at each time step. Can anyone explain me why this is happening? Thanks, Nikhil

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Initialization problem when using Axisymmetry BC	Eduardo	FLUENT	0	August 4, 2008 09:23
Memory issue - Suse 10 - Opterons	Andy R	FLUENT	1	June 23, 2008 14:44
Initialization	a student	Siemens	2	May 19, 2006 12:55
compressor initialization	paglia	FLUENT	0	February 1, 2006 15:14
Mesh Initialization	shakked	Main CFD Forum	2	August 7, 2004 22:50

June 26, 2010, 06:29		#2
jack1980 Senior Member Jouke de Baar Join Date: Oct 2009 Posts: 127 Rep Power: 16	Hi, probably five time steps is not enough to reach a steady solution. What you want to do is monitor a certain flow variable (perhaps the drag coefficient) versus flowtime. When your monitor reaches a constant value, this usually indicates you have reached a steady solution. Good luck!

June 26, 2010, 19:29		#3
nikhil New Member Nik Join Date: Mar 2010 Location: Michigan Posts: 25 Rep Power: 16	Hi Jack, I know that i am not running the case till steady state. I am interested in what happens after 'x' time instant and my point is after 'x' time instant different initialization provides different results. I am trying to understand why it is so. Also, till i reach 'x' time instant i make sure that i achieve convergence at each time step. Hope this clarifies my question. Nikhil

June 27, 2010, 05:16		#4
jack1980 Senior Member Jouke de Baar Join Date: Oct 2009 Posts: 127 Rep Power: 16	Hi, In most cases you can only expect the steady state to be the same. The path along which you get there generally depends on the initial condition. Think of the temperature distribution of a simple rod, with temp 1 at the left hand side and temp 2 on the right hand side. Now either initialize the complete rod with temp 1 or temp 2. Surely the paths will differ but the final solution will be the same. Convergence of each time step doesn't mean the solution is converged in time, only that you are progressing accurately in time. Good luck!