Author |
Message |
Mongi Mansouri (mongim_pde)
Member Username: mongim_pde
Post Number: 20 Registered: 11-2003
| Posted on Friday, March 12, 2004 - 12:28 pm: | |
Hi All... This is a classic feature and I hope someone has an answer to it. I want to run a transient problem (say for Temperature) till steady state is reached. For instance, assuming a steady state exists, I would like to run the transient problem till the solution ceases to change to within some tolerance, at which point the program is done, and simulation stops. Is there a way to do this in FlexPDE? I would appreciate any help with this question. Thank you. /Mongi M. |
Robert G. Nelson (rgnelson)
Moderator Username: rgnelson
Post Number: 124 Registered: 06-2003
| Posted on Friday, March 12, 2004 - 07:12 pm: | |
In principle, if the system has reached steady state, the timestep should grow large, and complete the final period quickly. There must be some component of your system that is introducing an instability at equilibrium. In version 3.10b we introduced a generic HALT condition, but I'm not sure what you would write to get the effect you want - maybe a time derivative of total energy or a spatial integral of time derivatives. But if something is oscillating, this may not go smoothly to zero.
|
Mongi Mansouri (mongim_pde)
Member Username: mongim_pde
Post Number: 21 Registered: 11-2003
| Posted on Friday, March 12, 2004 - 07:50 pm: | |
Dear Robert... In reality, as you guessed it, I am running an oscillating test. There is a steady state (ss) in average. The temperature oscillates about some ss average. So based on your answer, and the answer to the other User's question about Time integral (integrating over a number of cycles), I think the integral of a certian value, say the max. temperature (a scaler) could be used to determine a ss halt condition. The sad thing is I am still using FPDE 2.22.... Can we do it in this version? I have to think about this a bit... I would really appreciate any further suggestions... Thanks again for your Help. /Mongi M. |
Mongi Mansouri (mongim_pde)
Member Username: mongim_pde
Post Number: 22 Registered: 11-2003
| Posted on Friday, March 26, 2004 - 04:06 pm: | |
Dear Robert... You mentioned in previous response that Version 3.xx has a generic Halt condition. My question is: is it different than the one found in Version 2.22 (which is based on minimum time step)? Would Version 4 have a better Halt condition that would allow me to terminate a run based on the solution field itself? This is a very important feature which I think should be incorporated. Thank you.
|
Robert G. Nelson (rgnelson)
Moderator Username: rgnelson
Post Number: 129 Registered: 06-2003
| Posted on Friday, March 26, 2004 - 08:32 pm: | |
The version 3 (and 4) HALT statement takes an arbitrary relational argument, and halts if it evaluates to TRUE. For example: HALT integral(c)>4.2
|