Author |
Message |
Brigitte Watzke (bwa)
New member Username: bwa
Post Number: 1 Registered: 05-2004
| Posted on Tuesday, May 18, 2004 - 04:26 am: | |
Hi, I would like to know how to express BC between two domains (1, gas & 2, solid) where concentration distribution between them is described by the partition coefficient K. (There is a discontinuity in the concentration profile at the phase boundaries). Instead of Natural(C) = normal (-D grad(C)), I need to express it like: Natural(C) = -D ( Cgas - K * Csolid ).(where Cgas and Csolid are the local concentration of solutes at surfaces 1 and 2, and D is the mass transfer coefficient in the gap). How to get the local nodal concentration of surface 2 in the case of the above BC? Thanks in advance ! Brigitte |
Robert G. Nelson (rgnelson)
Moderator Username: rgnelson
Post Number: 148 Registered: 06-2003
| Posted on Tuesday, May 18, 2004 - 04:11 pm: | |
FlexPDE assumes that the quantities declared as Variables are continuous across interfaces. It also assumes that the fluxes implied by your PDE are continuous across interfaces. So the best solution to your problem is to choose a different variable, one that conforms to the assumptions made by FlexPDE and from which you can compute C. For example, if you declare a variable C0, from which you define C=q*C0, then C will jump by q1/q2 at an interface. The PDE div(D*grad(C0))=0 will imply continuity of normal(D*grad(C/q)). The PDE div(D*grad(C))=0 implies continuity of normal(D*grad(C)). You must adjust D to maintain the validity of the PDE. Since you know more about the physics of your problem than I do, you can probably find a fundamental property of the system that satisfies these requirements. Incidentally, FlexPDE version 4.1 (under development) implements facilities to simplify the definition of discontinuous variables. These facilities are based on the concept of a "contact resistance", with the fundamental equation Flux=Jump(variable)/resistance. |
Brigitte Watzke (bwa)
New member Username: bwa
Post Number: 2 Registered: 05-2004
| Posted on Wednesday, May 19, 2004 - 05:37 am: | |
Thanks a lot for your fast answer! |
|