Author |
Message |
Mark (everest20)
New member Username: everest20
Post Number: 2 Registered: 01-2009
| Posted on Monday, January 19, 2009 - 12:42 pm: | |
Dear Mr Nelson! Can you tell me what is wrong with my file? I want to simulate moving bubble (spherically symmetric) in 1D. The calculation process begins, but it fails after some time. Why is that? Everest
|
Marek Nelson (mgnelson)
Moderator Username: mgnelson
Post Number: 104 Registered: 07-2007
| Posted on Monday, January 19, 2009 - 01:38 pm: | |
You are holding the end nodes fixed in space while the interior nodes are moving with a constant velocity. This causes the end cells to collapse to zero volume at which point the polynomial approximations fail. |
Mark (everest20)
Junior Member Username: everest20
Post Number: 3 Registered: 01-2009
| Posted on Friday, January 23, 2009 - 07:30 am: | |
Thank You for the answer. I've added "grid(r)" to the code to check your suggestion, and it seems to prove it. My problem is that, the phenomenon I am dealing with is spherically symmetric, therefore I cannot move the left-boundary point, this is also why natural BC is prescrbed there. I've tried to use the same velocity for the right-boundary point as for the interface (this is what I can eventually accept) and still the same error. In this situation, would be so kind, and tell me if the problem presented here can be solved using FlexPDE? |
Marek Nelson (mgnelson)
Moderator Username: mgnelson
Post Number: 107 Registered: 07-2007
| Posted on Saturday, January 24, 2009 - 12:44 am: | |
Now the left most cell is getting stretched until it can't fit a quadratic through the cell. This problem will work much better if you use velocities for the cell movements. Describe just the key point motions and let the rest of the cells diffuse to some equilibrium. I have attached a script that poses your problem that way. The right and left end points are fixed and the midpoint moves toward the right end. The rest of the nodes just diffuse to equilibrium.
|