CFD Online Logo CFD Online URL
www.cfd-online.com
[Sponsors]
Home > Forums > CFX

Adaptive Timesteps

Register Blogs Members List Search Today's Posts Mark Forums Read

Reply
 
LinkBack Thread Tools Display Modes
Old   July 22, 2011, 11:04
Default Adaptive Timesteps
  #1
New Member
 
Zhang Yang
Join Date: Jun 2011
Location: Zürich
Posts: 28
Rep Power: 7
swiss_zhang is on a distinguished road
Hello, Guys:

I am doing a transient fire simulation using CFX und have difficulties to define the timesteps.

Can anyone tell me the a rule of thumb how to define the timeteps with Adaptive method.

e.g.

initial Timestep
min Timesteps
max Timesteps
Timestep Decrease Factor
Timestep Increase Factor

Thanks a lot

Yang
swiss_zhang is offline   Reply With Quote

Old   July 23, 2011, 06:19
Default
  #2
Super Moderator
 
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 10,931
Rep Power: 85
ghorrocks has a spectacular aura aboutghorrocks has a spectacular aura aboutghorrocks has a spectacular aura about
Set the initial timestep to be small, in general. I don't generally use max/min timesteps but set a simulation time instead. The default increase/decrease factor should be fine.

As described in the documentation, adaptive time steps aiming for 3-5 coeff loops per time step (for simple models) or 6-10 loops per time step (for complex models, eg multiphase) is usually a good approach.
ghorrocks is offline   Reply With Quote

Old   August 26, 2011, 06:40
Default
  #3
Sal
New Member
 
Join Date: Jun 2011
Posts: 15
Rep Power: 6
Sal is on a distinguished road
Hi Glenn,

In my simulations Adaptive Timestepping has not varied with the time - direction is unchanged (ratio 1.000).
As you suggested previously, I set the 'dt range' wide enough to let the solver calculate the most suitable dt.

Could you please have a quick check to my timestep set up:

TIME DURATION:
Option = Total Time
Total Time = 0.006 [s]
END
TIME STEPS:
First Update Time = 0.0 [s]
Initial Timestep = 1e-06 [s]
Option = Adaptive
Timestep Update Frequency = 1
TIMESTEP ADAPTION:
Maximum Timestep = 1e+20 [s]
Minimum Timestep = 1e-20 [s]
Option = Number of Coefficient Loops
Target Maximum Coefficient Loops = 5
Target Minimum Coefficient Loops = 3
Timestep Decrease Factor = 0.8
Timestep Increase Factor = 1.06

CONVERGENCE CONTROL:
Maximum Number of Coefficient Loops = 4
Minimum Number of Coefficient Loops = 3
Timescale Control = Coefficient Loops


Regards,

Sal
Sal is offline   Reply With Quote

Old   August 26, 2011, 07:05
Default
  #4
Super Moderator
 
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 10,931
Rep Power: 85
ghorrocks has a spectacular aura aboutghorrocks has a spectacular aura aboutghorrocks has a spectacular aura about
Remove the max and min number of coeff loops in the convergence control section. This is forcing it to do either 3 or 4 loops per time step and that means your min3 max5 adaptive timestep critereon is never reached. Instead make the max number of coeff loops a number much larger, say about 10.
ghorrocks is offline   Reply With Quote

Old   August 26, 2011, 09:04
Default
  #5
Sal
New Member
 
Join Date: Jun 2011
Posts: 15
Rep Power: 6
Sal is on a distinguished road
Many thanks Glenn.
With max coeff loops = 10 in the control section, dt keeps decreasing with a ratio of 0.8.
Current timestep is 2e-10. Statistics are dropping down very sharply. RMS Co and Max Co = 0. Simulation takes forever.
Is it normal?
Sal is offline   Reply With Quote

Old   August 26, 2011, 13:22
Default
  #6
Sal
New Member
 
Join Date: Jun 2011
Posts: 15
Rep Power: 6
Sal is on a distinguished road
I have changed the residual target (convergence criteria) from 1e-8 [s] to 1e-6 [s]. Still costly but running definitely faster. Now, dt is approaching to 1e-07 [s].
Appreciated again your help Glenn.
Sal is offline   Reply With Quote

Old   August 26, 2011, 22:07
Default
  #7
Super Moderator
 
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 10,931
Rep Power: 85
ghorrocks has a spectacular aura aboutghorrocks has a spectacular aura aboutghorrocks has a spectacular aura about
Your convergence target is almost certainly far too tight. Do a residual sensitivity study so you know how tight you need to go - 1e-3 or 1e-4 is more normal for transient flows.
ghorrocks is offline   Reply With Quote

Old   August 30, 2011, 06:46
Default
  #8
Sal
New Member
 
Join Date: Jun 2011
Posts: 15
Rep Power: 6
Sal is on a distinguished road
Thanks Glenn, I have done that. Residual target of 1e-04 is ok for this case.
dt has now varied around 3e-06 [s].
Sal is offline   Reply With Quote

Old   August 30, 2011, 06:51
Default
  #9
Super Moderator
 
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 10,931
Rep Power: 85
ghorrocks has a spectacular aura aboutghorrocks has a spectacular aura aboutghorrocks has a spectacular aura about
Excellent. That sounds much more promising.
ghorrocks is offline   Reply With Quote

Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
adaptive meshing clifford bradford Main CFD Forum 13 August 11, 2014 09:10
Dynamic Adaptive Mesh Coarsening Will Anderson FLUENT 0 October 26, 2010 17:26
Adaptive Timestepping: Possible Glitch? Josh CFX 9 August 25, 2010 21:52
Loading timesteps comes to grinding halt EISTO CFX 7 August 9, 2010 06:35
unstructured grid sreekanth Main CFD Forum 1 August 6, 2001 15:09


All times are GMT -4. The time now is 13:16.