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.
comparison between two models
Posted 8 janv. 2010, 07:40 UTC−5 1 Reply
Please login with a confirmed email address before reporting spam
Hi there,
i am trying to simulate a single layer volume conductor, in 2D (attached) as well as in 3D (attached). When i compare both the results there is a huge difference (as i am just using a single plane to compare the results). I also have the analytical solution (attached) to compare with, which is more close to the 3D one. the analytical solution considers angle phi as zero thus restricts itself in y-z plane, which is not a problem as the dipole is located along z-axis radially). the spherical coordinates as arranged as (r,theta,phi)
Since i am using a single dipole with source and sink at z=0.0005 and -0.0005m, respectively. with I=1A and radial dipole moment i.e dipole moment is along z-axis. I am bit concern as why there is such a huge disparity (almost 10 times). I know i am doing something wrong somewhere, but cannot get my hands onto it. This is a simple volume conductor problem and there are no nonlinearity issues involved.
Both the models and the response of analytical problem are attached .
Since i have an older version of comsol, the femlab, therefore i cannot upload them with their native extension. so after downloading just remove the . mph extension, .fl is the native extension for femlab.
to get close results change the mesh to finer or higher and then solve it.
With regards
Salman
i am trying to simulate a single layer volume conductor, in 2D (attached) as well as in 3D (attached). When i compare both the results there is a huge difference (as i am just using a single plane to compare the results). I also have the analytical solution (attached) to compare with, which is more close to the 3D one. the analytical solution considers angle phi as zero thus restricts itself in y-z plane, which is not a problem as the dipole is located along z-axis radially). the spherical coordinates as arranged as (r,theta,phi)
Since i am using a single dipole with source and sink at z=0.0005 and -0.0005m, respectively. with I=1A and radial dipole moment i.e dipole moment is along z-axis. I am bit concern as why there is such a huge disparity (almost 10 times). I know i am doing something wrong somewhere, but cannot get my hands onto it. This is a simple volume conductor problem and there are no nonlinearity issues involved.
Both the models and the response of analytical problem are attached .
Since i have an older version of comsol, the femlab, therefore i cannot upload them with their native extension. so after downloading just remove the . mph extension, .fl is the native extension for femlab.
to get close results change the mesh to finer or higher and then solve it.
With regards
Salman
1 Reply Last Post 14 janv. 2010, 00:30 UTC−5
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
First of all your physics are applied to different geometries, I believe James Freels said this already a month ago, in the 3D you have "point sources", around a sphere while in the 2D you have a long cylinder with "long wires" (normalised to 1[m]), for me this should already NOT give the same results.
Furthermore, you are still very mesh dependent, in EM (ACDC or RF or optical) its often the case.
As you know, by symmetry there is a virtual ground plane cutting your volume in two (x-y, Z=0), you can either cut your model and use GND BC conditions, or make an interiour boundary and enforce V=0, this helps a little.
Then in these cases with large difference of sizes ("point" vesus big "sphere") I would draw a couple of embeedin spheres with a size increase of 10-30x. i.e. radius of 2mm, and 30mm inside your spere /circle of 0.075 and mesh the inner one very fine, and then coarser as you go out. Normally to get cleaner results. Furthermore, I believe you should use a rather symmetric structured mesh, typically such you get from a 2D axi revolve into 3D
Hope this helps on the way
Good luck
Ivar
First of all your physics are applied to different geometries, I believe James Freels said this already a month ago, in the 3D you have "point sources", around a sphere while in the 2D you have a long cylinder with "long wires" (normalised to 1[m]), for me this should already NOT give the same results.
Furthermore, you are still very mesh dependent, in EM (ACDC or RF or optical) its often the case.
As you know, by symmetry there is a virtual ground plane cutting your volume in two (x-y, Z=0), you can either cut your model and use GND BC conditions, or make an interiour boundary and enforce V=0, this helps a little.
Then in these cases with large difference of sizes ("point" vesus big "sphere") I would draw a couple of embeedin spheres with a size increase of 10-30x. i.e. radius of 2mm, and 30mm inside your spere /circle of 0.075 and mesh the inner one very fine, and then coarser as you go out. Normally to get cleaner results. Furthermore, I believe you should use a rather symmetric structured mesh, typically such you get from a 2D axi revolve into 3D
Hope this helps on the way
Good luck
Ivar