| Author | Message | ||
     Chris Walton (ccwalton) Member Username: ccwalton Post Number: 6 Registered: 01-2008 |
(Mistakenly posted this at bottom of an unrelated thread - reposting...) I've been running a heat conduction problem with radiative boundaries for a glass slab with a complex heat load on the top surface. I recently tried to add a second element: a thin aluminum plate beneath the glass, separated by a small gap. There is radiative heat exchange between them and with the walls. Though simple enough, this problem gives a "memory protection fault" error immediately upon starting, if the plate is included. The details are ---called from fitting::evalat (1)Called from fitting::validate_fits ---Called from fitting::validate_fits ---Called from timsolve::first_udot ---Called from timsolve::evolve Sometimes (maybe 1 time in 10) instead of a memory protection fault, the code will give an "unknown exception" and crash completely. The problem occurs identically on two different WinXP machines. I am running version 4.1.0a. I don't think there are problematic discontinuities in the problem - it still occurs when the time- and position-dependent heat load, normally looked up in a table, is replaced with a constant value. I have attached the .pde file by email to support@pdesolutions.com Thanks! Chris Walton | ||
| Marek Nelson (mgnelson) Moderator Username: mgnelson Post Number: 68 Registered: 07-2007 |
There appears to be trouble creating the implicit coupling in the VAL statements of your definitions of GlassBotTemp and AlTopTemp. We will try to track down the exact cause, but in the meantime, if you use EVAL instead of VAL, the script runs and appears to get a reasonable answer. There is one advantage of using EVAL, and that is you can have the induced temperature based on the value directly opposite each surface point. This is an explicit rather than implicit coupling, but it seems to work on your problem. I will email the modified script to you. PS - you may want to download the most recent version 4 update (v4.2.16) at www.pdesolutions.com/v4support.html . | ||
| Chris Walton (ccwalton) Member Username: ccwalton Post Number: 7 Registered: 01-2008 |
Received by email your modifications (with VAL switched to EVAL on the two temperatures, and those temperatures taken at generalized radius "r" rather than specific values). Ran this unaltered on the same two machines, and both still give a memory protection fault. The fault details are different now: ---Called from finder::build_finder ---Called form finder::find_cell2 ---Called from treeval::evalat (4)Called from treeval::eval ---Called from treeval:eval About the side benefit of using EVAL so the slab and plate exchange T information with points directly opposite, that's nice - thanks! I would have wanted to do that down the road for sure. Chris Walton | ||
| Marek Nelson (mgnelson) Moderator Username: mgnelson Post Number: 69 Registered: 07-2007 |
This script runs to completion on my machine using version 4.2.16. If you are still running version 4.1.0a, you should download the free update at www.pdesolutions.com/v4support.html and see if it still fails. | ||
| Chris Walton (ccwalton) Member Username: ccwalton Post Number: 8 Registered: 01-2008 |
That's got it - thanks! Chris Walton |