| Author | Message | ||
     bill zhao (billzhao) New member Username: billzhao Post Number: 1 Registered: 03-2005 |
I am a new user of FLEXPDE and have some basic questions. 1. Is it possible to simulate a thermal induced stress in the transient heat transfer case? 2. If yes, how to write the equation considering about the thermal expansion in each time iteration due to the thermal expansion depending on the temperature? 3. Is there a command to control the plots of deformations of the grid without showing the internal grid but just the outline of the domain? Thanks! | ||
| Robert G. Nelson (rgnelson) Moderator Username: rgnelson Post Number: 339 Registered: 06-2003 |
1. We don't have any examples that do that, but "Elasticity.pde" and "3d_Bimetal.pde" (in the "Samples | steady_state | Stress" folder) do steady-state thermo-elasticity, and "Samples | Time_Dependent | Stress | Vibrate.pde" does time dependent stress analysis. With these and a textbook, extension to the problem you want should not be difficult. 2. You don't worry about iterations. You write a time-dependent PDE. See a textbook on thermoelasticity. 3. No. You would have to export to a paint program and erase the grid. | ||
| bill zhao (billzhao) New member Username: billzhao Post Number: 2 Registered: 03-2005 |
Thanks for your answer! I have another question. Currently, I am simulating the resin curing in the mold. It quite similar to the chemical reaction or the reverse melting case. It works well for just simulating the temperature and cure degree distribution of the resin. When I include the deformation (stress/strain) equations in the scrip to deal with the shrinkage and thermal expansion strain/stress, it stops at the begin and give the error message “over-floating”. I guess it is because the mechanical properties in the early stage of un-cured resin are very small. Actually, I concern the latter stage such as when the cure degree reaches 0.3 or so, the resin becomes a little solid and the mechanical properties becomes larger. My question is “ Is it possible to include the strain equations just in the latter stage and the early stage just solve the temperature and curing degree?” I guess but I didn’t try, the export and import using the transfer commands could do the job, the problem is the temperature and cure degree still need to be computed in the latter stage. Thanks! | ||
| Robert G. Nelson (rgnelson) Moderator Username: rgnelson Post Number: 344 Registered: 06-2003 |
If your system is nonlinear, then it is crucial that you begin the solution with a reasonable starting estimate, or convergence may be slow or impossible. FlexPDE uses Newton's method to solve nonlinear systems, and as I'm sure you know, if you start Newton's method too far from the solution, it will never converge. One common technique is to use STAGES to start the system with linear parameters, so as to get a good starting estimate. Then in later stages turn on the nonlinearities, possibly a little at a time. It is not at present possible to specify two different equation sets at different phases of the solution process. But as you mention, you can solve the system in two scripts, using TRANSFER to pass the solution of the first system to the later full system. Or, as mentioned above, you could provide trivial parameters to the strain equations in early stages, effectively disabling them. The computational labor of the additional equations remains, but if the solution is easy, it might not cost you much. |