Author |
Message |
Fred Sachs (sachs)
Member Username: sachs
Post Number: 9 Registered: 01-2004
| Posted on Thursday, July 08, 2004 - 11:36 am: | |
I have a 2D problem and it runs fine. However, I would like to do a parameter study and htough that perhaps I could fake a 3D euation where one dimension is used for the varied parameter. Is that possible and how can I set it up? |
Robert G. Nelson (rgnelson)
Moderator Username: rgnelson
Post Number: 179 Registered: 06-2003
| Posted on Thursday, July 08, 2004 - 03:39 pm: | |
Is there a reason not to use Staging? That would seem to be the natural way to approach this. You can plot histories vs stage. You can use the VERSUS qualifier to change the abcissa value in the histories (see attached script). You can in principle do what you suggest, but you may pay a high price in computation time, since the mesh size will increase enormously, and FlexPDE will try to solve all the nodes simultaneously. It may not converge. Furthermore, FlexPDE will assume there is some kind of differential coupling in the third dimension, so you will have to concoct a differential equation linking your parameters.
|
Fred Sachs (sachs)
Member Username: sachs
Post Number: 11 Registered: 01-2004
| Posted on Monday, July 12, 2004 - 10:47 am: | |
You are right. Staging is better in reducing the dimensionality. |
Fred Sachs (sachs)
Member Username: sachs
Post Number: 12 Registered: 01-2004
| Posted on Monday, July 12, 2004 - 02:41 pm: | |
Is there a way, within FLex, to plot a derivative (successive differences) of a history file? |
Robert G. Nelson (rgnelson)
Moderator Username: rgnelson
Post Number: 184 Registered: 06-2003
| Posted on Monday, July 12, 2004 - 03:16 pm: | |
Applying the history plot facility to stages was an afterthought, and in a true time-dependent problem you can simply take time derivatives, so we have not implemented a mechanism for differencing in time. Sorry.
|