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.
Problems With Converging due to a solution dependant material property
Posted 4 ott 2016, 20:48 GMT-4 2 Replies
Please login with a confirmed email address before reporting spam
Hi all,
So I have been trying to get COMSOL to solve me a system using the electrostatics module.
I have found that my model does not converge when I add my own permittivity model in. Usually Permittivty can be set as a constant which when I did it converges but part of my thesis involves using this more accurate permittivity model. Additionally if the geometry is very basic (simple cylinder) the system does converge with the permittivity model I input. Mainly i find that if I specify more than 1 boundary as a certain voltage it does not converge.
My model solves for the electric potential in the system but the permittivity relies on the electric potential so I have made a variable E = -(d(V,x)+d(V,y)+d(V,z)) in cartesian and transformed to cylindrical coordinates = -(d(V,r)+(1/r)*d(V,phi)+d(V,z))
Now the permittivity model is the booth model which I have put in as:
(n^2)+(3/(beta*E+delta))*(eps0-(n^2))*(coth(beta*E+delta)-(1/(beta*E+delta)))
Where everything is a constant apart from E which is my defined variable:
What do you think is preventing it from converging?
Attached is the paper I am basing this model off and my model in progress.
Thanks, Kevin
So I have been trying to get COMSOL to solve me a system using the electrostatics module.
I have found that my model does not converge when I add my own permittivity model in. Usually Permittivty can be set as a constant which when I did it converges but part of my thesis involves using this more accurate permittivity model. Additionally if the geometry is very basic (simple cylinder) the system does converge with the permittivity model I input. Mainly i find that if I specify more than 1 boundary as a certain voltage it does not converge.
My model solves for the electric potential in the system but the permittivity relies on the electric potential so I have made a variable E = -(d(V,x)+d(V,y)+d(V,z)) in cartesian and transformed to cylindrical coordinates = -(d(V,r)+(1/r)*d(V,phi)+d(V,z))
Now the permittivity model is the booth model which I have put in as:
(n^2)+(3/(beta*E+delta))*(eps0-(n^2))*(coth(beta*E+delta)-(1/(beta*E+delta)))
Where everything is a constant apart from E which is my defined variable:
What do you think is preventing it from converging?
Attached is the paper I am basing this model off and my model in progress.
Thanks, Kevin
Attachments:
2 Replies Last Post 10 ott 2016, 03:41 GMT-4
Please login with a confirmed email address before reporting spam
Posted:
8 years ago
I suggest the following:
1. Simulate with beta = 0 to be sure it converges.
2. Perform a parametric sweep increasing the value of beta starting at zero.
It is not surprising that making a problem nonlinear causes difficulty with convergence.
DW Greve
1. Simulate with beta = 0 to be sure it converges.
2. Perform a parametric sweep increasing the value of beta starting at zero.
It is not surprising that making a problem nonlinear causes difficulty with convergence.
DW Greve
Please login with a confirmed email address before reporting spam
Posted:
8 years ago
Thanks, D Greve
After some more testing and help from COMSOL staff I have found that using a derived variable instead of an internal variable such as the norm of the electric field (es.normE) was the main issue and some minor meshing changes.
Thanks,
Kevin Leung
After some more testing and help from COMSOL staff I have found that using a derived variable instead of an internal variable such as the norm of the electric field (es.normE) was the main issue and some minor meshing changes.
Thanks,
Kevin Leung