Help for setting varying timesteps

darookie · September 10, 2009, 05:46

The simple solution is always hard to get. After having searched in the forum for a good solution to set varying timesteps with an CEL expression without sucess, here is my way to obtain:
first 300 iterations -> 0.7s
next 300 iterations -> 0.01s

CEL Expression:
0.7[s]-0.69*step(citern-301)[s]

I hope I can help somebody who has the same task.
andrej

ghorrocks · September 10, 2009, 07:32

You could also have done it with a 1D interpolation function. Same end effect, but possibly easier to understand.

darookie · September 11, 2009, 06:08

I wanted to change the time in a jump function. But is a linear change better in regard to convergence?

ghorrocks · September 11, 2009, 08:32

You can define step functions in the 1D interpolation function. The following series of points define a step function good enough for most people:
(0,0),
(0.999999999999,0),
(1,1),
(2,1)

Yes, a sudden change in time step size can cause convergence issues. It will cause accuracy issues as well. Ramping the time step change will help both problems.

September 10, 2009, 05:46	Help for setting varying timesteps	#1
darookie New Member Andrej Hoyningen Join Date: Jul 2009 Posts: 22 Rep Power: 17	The simple solution is always hard to get. After having searched in the forum for a good solution to set varying timesteps with an CEL expression without sucess, here is my way to obtain: first 300 iterations -> 0.7s next 300 iterations -> 0.01s CEL Expression: 0.7[s]-0.69*step(citern-301)[s] I hope I can help somebody who has the same task. andrej

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
time varying inlet velocity	Zhe Liu	CFX	0	July 31, 2008 07:56
Timesteps in steady state	Daniel	CFX	3	June 1, 2006 03:10
Time Varying Boundary Conditon	ashish	CFX	3	February 15, 2005 07:21
Time Varying BC	A.S.	Siemens	1	January 31, 2005 07:37
How to model varying food models in a Blender	Raymond	CFX	0	July 10, 2004 00:40

September 10, 2009, 07:32		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,870 Rep Power: 144	You could also have done it with a 1D interpolation function. Same end effect, but possibly easier to understand.

September 11, 2009, 06:08		#3
darookie New Member Andrej Hoyningen Join Date: Jul 2009 Posts: 22 Rep Power: 17	I wanted to change the time in a jump function. But is a linear change better in regard to convergence?

September 11, 2009, 08:32		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,870 Rep Power: 144	You can define step functions in the 1D interpolation function. The following series of points define a step function good enough for most people: (0,0), (0.999999999999,0), (1,1), (2,1) Yes, a sudden change in time step size can cause convergence issues. It will cause accuracy issues as well. Ramping the time step change will help both problems.