| Author | Message | ||
     Bouke Tuinstra (bouke) Member Username: bouke Post Number: 4 Registered: 04-2006 |
I have a strange problem with a case where I have a rectangular 2-d domain in wich I solve the heat conduction equation div(-K grad(Temp))=0. Inside the domain are several regions with different values for K and JUMP boundaries as well. At the righhand domain boundary the temperature is given as Thot, at the lefthand boundary the temperature is Tcold, the top and bottom boundaries are symmetric. Now if I choose constant values for K, the problem converges and I get a reasonable temperature field. However, if I make the K value in (any) one of the several regions dependent on Temp the solution does not converge. For intermediate results a very strong temperature gradient is built up at the hot domain boundary (where this is not expected because it is a good conducting material). This happens even if the dependence is something like K=K0*(1+1E-6*Temp) so that K is practically constant, so apparently it has something to do with the structure of the equations, not the values of the solution. I've got another case with the same materials but a different geometry where this problem does not occur. I have tried SELECTing some solution parameters (Vandenberg, changelim), but to no avail. Any suggestions about what I could try to get it to work? |