| Author | Message | ||
     Bouke Tuinstra (bouke) New member Username: bouke Post Number: 1 Registered: 04-2006 |
Hi, I just started to do some serious work with FlexPDE and I ran into something I don't understand and that I cannot find in the manual. I am using internal Contact/Jump boundaries in a thermal analysis. As in the user manual, I create a domain (region 1) with an embedded different material region 2. Then I specify the contact resistance for the perimeter of region 2. So far so good. Now if I have a third material in region 3 that is in contact with both region 1 and region 2. I need to specify a boundary condition for its perimeter (or leave it unspecified, implying NATURAL(Temp)=0). But now on the border between region 2 and 3 I have specified two boundary conditions (one in region 2 and one in region 3). What is the correct way to handle this? Which b.c. will be used? I could imagine that you would need to specify an identical (reverse signed?) b.c. at both sides and that you would get a heat source at the boundary if you specify different conditions, but I'm not sure. Can anyone clear this up for me? Thanks! | ||
| Robert G. Nelson (rgnelson) Moderator Username: rgnelson Post Number: 589 Registered: 06-2003 |
If regions 2 and 3 are in contact along some boundary, there is only one boundary, and it can have only one boundary condition. The nodes along the boundary are shared by the two adjoining regions. If you want to apply a contact resistance between regions 2 and 3, use the CONTACT condition. This causes two nodes to be generated at each nodal position to track the discontinuity. If you don't want a contact resistance, then leave it without a condition. This will result in a continuous solution through the shared nodes. You can use either Natural(var)=0 or NOBC(var) to turn off any other conditions as you walk the boundary. In general, FlexPDE tries to detect multiple BCs on a boundary and issue a diagnostic. It doesn't always succeed. | ||
| Bouke Tuinstra (bouke) New member Username: bouke Post Number: 2 Registered: 04-2006 |
Ok, I made a little sample problem defining the three regions. Now I want a contact resistance at each material interface as shown below. However, FlexPDE complains that "three way contacts are not supported". If I comment out one of contact conditions, e.g. for region 3, the problem runs, but then I have no contact resistance between material 1 and 3. How can this be done in FlexPDE? TITLE "Simple Problem" VARIABLES Temp DEFINITIONS R2 = 1 R3 = 2 K = 1 EQUATIONS div(K*grad(Temp))=0 BOUNDARIES REGION 1 START(0,0) VALUE(Temp) = 0 LINE TO (4,0) NATURAL(Temp) = 0 LINE TO (4,4) VALUE(Temp) = 1 LINE TO (0,4) NATURAL(Temp) = 0 LINE TO CLOSE REGION 2 K = 1/2 start(1,1) contact(Temp) = 1/R2*JUMP(Temp) line to (2,1) to (2,2) to (1,2) to close REGION 3 K=1/3 start(2,1) contact(Temp) = 1/R3*JUMP(Temp) line to (3,1) to (3,2) to (2,2) line to close PLOTS CONTOUR(Temp) END | ||
| Robert G. Nelson (rgnelson) Moderator Username: rgnelson Post Number: 590 Registered: 06-2003 |
A CONTACT boundary condition causes each node on the boundary to be duplicated, and the resistance link made between the two nodes. Each node can only have one resistance link. The intersection of two Contact boundaries would require multiple resistance links, and this is not supported. There are several ways to work around this restriction: 1) put a small gap in one of the boundaries at each 3-way joint, with a Natural()=0 BC in the gaps. This creates a small energy leak, but if it's small enough it should not significantly distort the result. 2) replace the region 2/3 interface with a very thin resistive layer instead of the contact resistance. Choose the K value to give the correct total resistance over the chosen thickness. | ||
| Bouke Tuinstra (bouke) Junior Member Username: bouke Post Number: 3 Registered: 04-2006 |
Ok, that's clear, thanks! I solved it the way you suggested, and it looks fine. It would be nice tho' if it were possible to specify a contact/jump boundary for region 2 and for three sides of region 3, but (e.g.) specify a nobc condition for the left edge of '3' (much the same as it works with regions that lie on the outer domain perimeter: there you can specify a contact/jump boundary for the internal sides and a natural(0) boundary for the external sides). Especially for thermal models with a larger number of different materials (regions), that would save a lot of planning. |