Author |
Message |
jane New member Username: jane
Post Number: 1 Registered: 04-2010
| Posted on Friday, April 30, 2010 - 12:34 am: | |
Dear all, I just start to use FlexPDE. Now I have some problem about the boundary condition, not sure exactly how to deal with. Could anyone here can help me? the boundary conditions as follow: Jane and my code for the boundary conditions are: BOUNDARIES { The domain definition } REGION 1 { For each material region } START(rw,0) line to (rw,H) value(us) =0 line to (rs,H) value(us)=0 line to (rw,0) value (us)=value(un) {error!!!!!} line to close natural (us)=0 region 2 start (rs,0) line to (rs, H) ks*dr(us) = kh*dr(un) line to (re,H) value(un) = 0 line to (re,0) natural(un)=0 line to close natural (un)=0 Can anybody help me about this? Many thanks, |
rgnelson Moderator Username: rgnelson
Post Number: 1354 Registered: 06-2003
| Posted on Friday, April 30, 2010 - 04:50 pm: | |
I don't know what your equations are, but the conditions in your BMP file would appear to be the defaults that you get if you say nothing. Your BMP shows a single variable, "U". I don't know what the two-variable system Us and Un are intended to mean. VALUE is a keyword defining a boundary condition. You cannot use it as a right-hand-side. The proper form would be VALUE(us)=un. "Line to Close" following (rw,0) does nothing, because (rw,0) is where you started. "Natural(us)=0" following "close" does nothing because no path segment follows it. "ks*dr(us)" is not a BC specifier. You need a VALUE or a NATURAL. BMP is a stupid Microsoft format that is extremely wasteful. You should avoid it whenever possible. PNG is about 100 times more efficient.
|
jane New member Username: jane
Post Number: 2 Registered: 04-2010
| Posted on Monday, May 03, 2010 - 12:39 am: | |
Thanks Nelson. The governing equations are as follow: 1/r*dr(r*dr(un))+dzz(un)=dt(un) 1/r*dr(r*dr(us))+dzz(us)=dt(us) The variabes are un and us. Follow for your instructions, I still cant totally understand, how should I exactly deal with The BC as: value (us)=value(un) and ks*dr(us) = kh*dr(un) Could you please give me more hints? Many thanks, Jane |
rgnelson Moderator Username: rgnelson
Post Number: 1356 Registered: 06-2003
| Posted on Monday, May 03, 2010 - 02:06 pm: | |
You have never told us what system you are attempting to model, so I can only guess. Why are there two variables? If you are trying to equate values and fluxes, it would appear that there is only one variable, some U, that appears in two regions. Don't declare two variables, declare one, with the appropriate parameters in the various regions. Then all your requirements are met automatically.
|
jane Junior Member Username: jane
Post Number: 3 Registered: 04-2010
| Posted on Wednesday, May 05, 2010 - 02:10 am: | |
Thanks Nelson. The model I am going to analysis is one cylinder is in the middle of another one. And the inside one's outside surface is totally contacted with the outside one's inside surface. The materials of these two are different. Therefore, there are two variables. Can I use one which as you informed to simulate them? Many thanks again. Jane |
rgnelson Moderator Username: rgnelson
Post Number: 1359 Registered: 06-2003
| Posted on Wednesday, May 05, 2010 - 01:15 pm: | |
Is it your contention, then, that a temperature of 100 degrees in steel is a different physical phenomenon than a temperature of 100 degrees in aluminum, and that there is some kind of arcane conversion process that converts steel temperature into aluminum temperature? I would have thought that heat is heat. See "Samples|Applications|Heatflow|3d_Bricks+Time.pde" ("samples|time_dependent|heatflow|3d_bricks_t.pde" in version 5).
|
|