Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

Laplace stationary equation & Neuman Conditions

Please login with a confirmed email address before reporting spam

Hi Sorry to disturb you but I can’t manage to solve a very simple problem and it drives me crazy. For a college project I have to code some finite element method about current density distribution on different geometry in python and I choose to use Comsol to “Check the results”. I already used Comsol, but I didn’t have so much training on it. While building my models I always had the same errors “- Feature: Solveur stationnaire 1 (sol1/s1) Failed to find a solution. The relative error (2.1) is greater than the relative tolerance. There was a warning message from the linear solver. Iterative refinement triggered. Returned solution is not converged.” when I tried to compute the stationary solution. So, I decided to build a very simple geometry where I would be able to solve by hand and see if this error was still here, and it was. I searched on the offline documentation, online help, YouTube tutorials (it looks like tis but with Neuman conditions: https://www.youtube.com/watch?v=zO05DaLoeBg) aaaand I didn’t find so I hope you will be able to ^^” So, the problem is: let’s imagine a square with a current density distribution, I have the equations => j=σE (ohm) => E=-grad(U) => div(j)+dρ/dt=div(j)=0 (charge conservation, stationary) I have the Neuman boundary condition => On left and right: j.n=(dU/dx)(1/σ)=+-j_0 => On top and bottom: j.n=0 ie dU/dy=0 And I want to find a solution of σLap(U)=0 So on this simple square, by hand it gives Lap(U)=0 ie d²U/dx²+d²U/dy²=0, with invariances U(x,y)=U(x) so d²U/dx²=0 ie U(x)=Ax+B . Now I tried to implement it on Comsol. I build the geometry, I added the parameters without physical qualities to not complicate the thing, everything’s is at 1 except the electrical conductivity σ witch is at 1e7 like in metals, then I added a physics => maths => classical EDP => Laplace, so Lap(U)=0, I added my conditions on fluxes, build the meshes, tried to solve aaaand doesn’t work. I don’t figure out why, maybe it’s because I didn’t put enough conditions or maybe because I didn’t know what do with “initial value” (what is the purpose of this in a stationary study ??) so hope you will know ^^” I’m gonna add some pictures Best Regards Comsol version: 5.3 https://www.dropbox.com/sh/0q70a1ccwh8rojz/AAB--FSpzoFnwbwtYVMrG4_2a?dl=0



1 Reply Last Post 10 févr. 2019, 13:03 UTC−5
Jeff Hiller COMSOL Employee

Please login with a confirmed email address before reporting spam

Posted: 5 years ago 10 févr. 2019, 13:03 UTC−5
Updated: 5 years ago 11 févr. 2019, 08:24 UTC−5

Hi Marc,

There is an infinite number of solutions (corresponding to various values of B in U(x)=A*x+B) unless/until you specify the value of u at one point.

Best,

Jeff

-------------------
Jeff Hiller
Hi Marc, There is an infinite number of solutions (corresponding to various values of B in U(x)=A*x+B) unless/until you specify the value of u at one point. Best, Jeff

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.