| Author | Message | ||
     Dawid Jankowski (gwigwi) New member Username: gwigwi Post Number: 1 Registered: 09-2006 |
I have to solve a problem. Normal component of electrical field on boundary between two materials that have different permittivity isn't continous. Previously I solved this problem in differ way. I created a thin layer of material. That materials' permittivity changed continously. Is function CONTACT proper for this case? How can I use it in this case? Thanks for all answers and sorry for my english  | ||
| Robert G. Nelson (rgnelson) Moderator Username: rgnelson Post Number: 681 Registered: 06-2003 |
This question is addressed in Chapter 1 of the Applications Manual (downloadable from www.pdesolutions.com/bookstore.html) and in several places in the documentation relating to Natural Boundary Conditions. You don't have to do any contortions to make this work. If you just write the equation as it appears in the usual form for Maxwell's equations, ie Div(eps*Grad(Phi)) + rho = 0 then all of the required physics will be observed. Integration by parts (equivalent in this case to the divergence theorem) will cause the normal component of eps*grad(phi) to be continuous across all cell interfaces. A nonzero Natural Boundary condition on either inner or outer boundaries will mean the surface charge. | ||
| Dawid Jankowski (gwigwi) New member Username: gwigwi Post Number: 2 Registered: 09-2006 |
Sorry I didn't write it before but I can,t use potential becaouse accuracy is important for me. I use flexPDE to solve microwave system with complicated geometry. I use 4 variables - components od eletric field in cylindrical coordinates. Imaginary and real part of Ero and imaginary and real part of Ez. Later conversion potential on electric field will introduce additional inaccuracy and will be difficult. So I renew my question about fukction CONTACT. | ||
| Robert G. Nelson (rgnelson) Moderator Username: rgnelson Post Number: 685 Registered: 06-2003 |
Have you actually experienced inaccuracies in using potentials, or are you merely anticipating? Mesh refinement and selectable accuracy should give you control over this. |