| Author |
Message |
    
baeva
Junior Member
Username: baeva
Post Number: 3
Registered: 11-2009
| | Posted on Friday, September 24, 2010 - 09:00 am: | |
Hello,
I am trying to solve a 2D-fluid model, including flow and heat transfer. Is it possible to set limits for the solver to avoid e.g. a solution with negative temperatures?
Thank you in advance
Margarita |
rgnelson
Moderator
Username: rgnelson
Post Number: 1409
Registered: 06-2003
| | Posted on Friday, September 24, 2010 - 02:27 pm: | |
FlexPDE tries to solve the PDE system as posed. There are no ad hoc patches to falsify the system.
There are two causes of negative results that I can think of:
1) the system is incorrectly posed, and the negative results are true solutions to the equations as posed.
2) there are steep gradients that cannot be adequately resolved by the finite element interpolators. This can result in undershoots and overshoots. These can usually be removed by using a denser mesh or by eliminating discontinuities from the initial conditions.
|
rgnelson
Moderator
Username: rgnelson
Post Number: 1410
Registered: 06-2003
| | Posted on Friday, September 24, 2010 - 02:31 pm: | |
A third cause:
3) The system as posed is indeterminate and the solver is not converging. This usually results in wild oscillations.
And a fourth:
4) Navier-Stokes equations applied to very low-viscosity fluids can be unstable. I believe this is due to the fact that the fluid can sustain a multiplicity of turbulent flow patterns and the finite element solution cannot distinguish between them.
|
oio14644
New member
Username: oio14644
Post Number: 2
Registered: 10-2010
| | Posted on Friday, October 22, 2010 - 03:01 pm: | |
baeva,
how did you solve this kinds of problem£¿ |