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.
Issues with initializing a universe
Posted May 24, 2017, 5:12 p.m. EDT 1 Reply
Please login with a confirmed email address before reporting spam
Can anyone here clarify how COMSOL treats an empty space vs one filled with, say, air? I'm attempting to model the electric field lines through an axisymmetric silicon part and I need to know if surrounding it with a bubble of air is good practice or not. When I don't, the electric field lines are wholly contained within the device, and edge effects seem to be neglected. When I do surround it with air, the edge effects behave more realistically, but I have field lines that go from the bottom part and arc through the air like some sort of dipole, which I don't expect is physical given the low potentials I'm working with and the huge difference between the dielectric constant of air and silicon.
The documentation is a little vague about building simulations in air or not, and I'd just like to clarify which would be the best practice. Could the fact that I'm doing a 2D simulation rather than true 3D be the culprit? The part is completely axisymmetric, so my intuition tells me that it shouldn't quit matter, but I don't know if COMSOL treats these physics differently.
The documentation is a little vague about building simulations in air or not, and I'd just like to clarify which would be the best practice. Could the fact that I'm doing a 2D simulation rather than true 3D be the culprit? The part is completely axisymmetric, so my intuition tells me that it shouldn't quit matter, but I don't know if COMSOL treats these physics differently.
1 Reply Last Post May 24, 2017, 9:04 p.m. EDT
Please login with a confirmed email address before reporting spam
Posted:
7 years ago
I don't know any of the details of your model or device, but unless the boundaries on your hardware correspond to very well-defined well-known conditions (e.g., constant potentials), then you should very definitely include sufficient surrounding volume, so that the external boundaries of the computation space can ultimately be specified in a way that doesn't interfere seriously with computation of the correct fields (or charges, or currents, etc) in/on the thing you are actually trying to model. Whether that volume consists of air or vacuum probably doesn't matter, since the relative dielectric constant of air and vacuum are almost the same (i.e. 1), and the relative magnetic permeability is the same (i.e., 1), and the conductivity is nearly the same too (about 0).
Failure to include a sufficient surrounding volume of meshed computational space in which fields are computed properly is one of the most common novice-type errors made in the use of these kinds of modeling codes.
If you don't believe your results make good physical sense after you have included such a (substantial) volume, then start looking more closely for other problems with your model. And there may be more than one.
Failure to include a sufficient surrounding volume of meshed computational space in which fields are computed properly is one of the most common novice-type errors made in the use of these kinds of modeling codes.
If you don't believe your results make good physical sense after you have included such a (substantial) volume, then start looking more closely for other problems with your model. And there may be more than one.