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.
Eddy current simulation not converging when velocity increases
Posted Mar 15, 2017, 5:34 p.m. EDT Low-Frequency Electromagnetics, Mesh, Modeling Tools & Definitions, Parameters, Variables, & Functions, Studies & Solvers Version 5.2a 2 Replies
Please login with a confirmed email address before reporting spam
My simulation simulates an array of magnets moving over an aluminum plate at a fixed height and a given velocity, and while it computes and yields results similar to those of a numerical analysis, it won't compute at high velocities--which in this case refers to speeds greater than 50 m/s; I need to be able to simulate the system up to 115 m/s, so this is a big issue.
When I say that the simulation doesn't converge, I don't mean that I get a non-convergence error, I mean that the computation time approaches infinity and the convergence plot levels out and won't decrease any further. since skin depth becomes a more significant factor at higher velocity I've tried to make the mesh size a function of the velocity so that the mesh elements will always be smaller than the skin depth so that it can be resolved, however the number of mesh elements gets so large that the computation time is still excessively long, and since I'm using a parametric sweep the computation time becomes even longer.
I'm wondering if there are any meshing methods or techniques to make this work/work faster; this is under the assumption that the only source of this issue is that at high velocity the skin depth becomes smaller than the minimum element size and thus can't be resolved and COMSOL can't compute a solution, hopefully there isn't anything else wrong with the model that I haven't caught yet.
Does anyone have any insight into this type of simulation and can help me out with this issue? The file is too large to attach to this post (6.31 MB) but I can email it to anyone who's willing to take a look at it.
When I say that the simulation doesn't converge, I don't mean that I get a non-convergence error, I mean that the computation time approaches infinity and the convergence plot levels out and won't decrease any further. since skin depth becomes a more significant factor at higher velocity I've tried to make the mesh size a function of the velocity so that the mesh elements will always be smaller than the skin depth so that it can be resolved, however the number of mesh elements gets so large that the computation time is still excessively long, and since I'm using a parametric sweep the computation time becomes even longer.
I'm wondering if there are any meshing methods or techniques to make this work/work faster; this is under the assumption that the only source of this issue is that at high velocity the skin depth becomes smaller than the minimum element size and thus can't be resolved and COMSOL can't compute a solution, hopefully there isn't anything else wrong with the model that I haven't caught yet.
Does anyone have any insight into this type of simulation and can help me out with this issue? The file is too large to attach to this post (6.31 MB) but I can email it to anyone who's willing to take a look at it.
2 Replies Last Post Mar 15, 2017, 7:25 p.m. EDT
Please login with a confirmed email address before reporting spam
Posted:
8 years ago
Ben,
you don't tell us what approach you are using to model the eddy currents. You may want to check this thread: www.comsol.de/community/forums/general/thread/77311
You can clear the solution in your model to get it small enough to upload.
Cheers
Edgar
--
Edgar J. Kaiser
emPhys Physical Technology
www.emphys.com
you don't tell us what approach you are using to model the eddy currents. You may want to check this thread: www.comsol.de/community/forums/general/thread/77311
You can clear the solution in your model to get it small enough to upload.
Cheers
Edgar
--
Edgar J. Kaiser
emPhys Physical Technology
www.emphys.com
Please login with a confirmed email address before reporting spam
Posted:
8 years ago
I am using the Velocity (Lorentz Term) to model the eddy currents, and the velocity term is applied to the conductive plate; I have cleared the solution and the file is still 6.1 MB, so I may not be clearing it completely? I right-clicked on the solution stores and cleared the solution, and I also deleted the solution data sets from the Results node. Am I missing something?
I am going to use a boundary layer mesh on the top surface of my conductive plate to see if this will help resolve the skin depth better than what I was doing before; I was forced to convert the sides of the boundary layers from quadrilateral to triangular elements using the Convert node and inserting diagonal elements in order to mesh the control volume with a Free Tetrahedral mesh.
I am going to use a boundary layer mesh on the top surface of my conductive plate to see if this will help resolve the skin depth better than what I was doing before; I was forced to convert the sides of the boundary layers from quadrilateral to triangular elements using the Convert node and inserting diagonal elements in order to mesh the control volume with a Free Tetrahedral mesh.