Hello,
I'm modelling some magnetic components for power electronics applications in MgDyn2D using cartesian coordinates (no symmetry). My test case is a 'coil' modelled as a current density flowing parallel to the z axis in two regions, one into, and one out of, the page. There is a ferrite core modelled with a spline BH curve. Air surrounds everything. The core has a gap which is part of the air region. The outer boundary is an infinite approximation.
The test case provides physically sensible results. The numbers may not be right but there are fluxes where there should be fluxes and current densities where there should be current densities. Post processing is done with MgDynPost.
My question is how does one set the depth (into the page) of the problem?
This is needed to allow the calculation of for example, the magnetic flux density i.e. flux / crossectonal area where the cross sectional area is either an x or y dimension and the (unmodeled) depth into the page (the z dimension). The 2D solution is only for x and y but, a value of the distance into the page is required at least for post-processing.
Many thanks,
James
(Trivial) Question regarding MagnetoDynamics2D
-
greenlinux
- Posts: 9
- Joined: 15 Aug 2020, 18:11
- Antispam: Yes
-
raback
- Site Admin
- Posts: 4851
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: (Trivial) Question regarding MagnetoDynamics2D
Hi James,
For 2D cases it is assumed that the vector potential has only the z-component active and hence the derived fields operated by the curl operator are always in the plane. If you really want to model the finite length of the object that breaks the symmetry and you no longer can use the dimensional reduction. For that purpose there is the 3D solver.
-Peter
For 2D cases it is assumed that the vector potential has only the z-component active and hence the derived fields operated by the curl operator are always in the plane. If you really want to model the finite length of the object that breaks the symmetry and you no longer can use the dimensional reduction. For that purpose there is the 3D solver.
-Peter
-
greenlinux
- Posts: 9
- Joined: 15 Aug 2020, 18:11
- Antispam: Yes
Re: (Trivial) Question regarding MagnetoDynamics2D
Hi Peter,
Many thanks for taking the time to answer. If I could trespass on your time a little further, I have a few follow-up questions:
1. If the MgDyn2d is used with an axis-symmetric problem does what you've said still apply? Is the quasi-3d aspect of the 2d problem built into the revolution of the geometry around the axis?
2. In the cartesian geometry case, you are saying there is an area which is infinitely thin in one dimension, perhaps called dA. It is made up of a physically measurable x or y distance in the geometry and an infinitesimal z distance (call it z0). The current density flows into or out of the page for an infinitesimal distance which is also z0. The current density generates a magnetic potential over the region dA. The flux density that the magnetic potential creates is measured in the area dA so, as long as the z component of the flux density measurement area (dA) is the same as the z dimension of the current density that creates the magnetic potential, it doesn't matter what the actual physical depth I'm thinking about in my real world problem is. The numerical answers will be the same for any depth of problem because changing the value of z0 would change the distance over which the current density flowed and would change the size of the area used to determine the magnetic flux density. Right?
3. If I choose to replace the body force current density in MgDyn2d with a coupling to the Circuits and Dynamics solver as Pavel Ponomarev did in https://www.researchgate.net/publication/317012206 (page 41 of the PDF, Section 5.2 of the document). Is it safe to presume that it will be possible to include the effects of finite depth in the circuit aspect of the simulation and observe the effects of this in post processing on lumped variables such as potential difference between the coil terminals?
Many thanks,
James
Many thanks for taking the time to answer. If I could trespass on your time a little further, I have a few follow-up questions:
1. If the MgDyn2d is used with an axis-symmetric problem does what you've said still apply? Is the quasi-3d aspect of the 2d problem built into the revolution of the geometry around the axis?
2. In the cartesian geometry case, you are saying there is an area which is infinitely thin in one dimension, perhaps called dA. It is made up of a physically measurable x or y distance in the geometry and an infinitesimal z distance (call it z0). The current density flows into or out of the page for an infinitesimal distance which is also z0. The current density generates a magnetic potential over the region dA. The flux density that the magnetic potential creates is measured in the area dA so, as long as the z component of the flux density measurement area (dA) is the same as the z dimension of the current density that creates the magnetic potential, it doesn't matter what the actual physical depth I'm thinking about in my real world problem is. The numerical answers will be the same for any depth of problem because changing the value of z0 would change the distance over which the current density flowed and would change the size of the area used to determine the magnetic flux density. Right?
3. If I choose to replace the body force current density in MgDyn2d with a coupling to the Circuits and Dynamics solver as Pavel Ponomarev did in https://www.researchgate.net/publication/317012206 (page 41 of the PDF, Section 5.2 of the document). Is it safe to presume that it will be possible to include the effects of finite depth in the circuit aspect of the simulation and observe the effects of this in post processing on lumped variables such as potential difference between the coil terminals?
Many thanks,
James
-
raback
- Site Admin
- Posts: 4851
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: (Trivial) Question regarding MagnetoDynamics2D
Hi James,
1) Yes. In axis-symm the vector potential only has the azimuthal (phi) component and the derived fields are in (r,z) plane. See the Elmer Models manual for details.
2) I would rather see this as infinitely long. It would be difficult to apply current to a infinitely thin conductor.
3) An infinitely long device would have infinite resistances etc. So it is convenient to have the length as a parameter when modeling with external circuits. So it that sense the true length is considered but the resulting fields are of course still not function of z. Considering end-effects would require 3D modeling.
-Peter
1) Yes. In axis-symm the vector potential only has the azimuthal (phi) component and the derived fields are in (r,z) plane. See the Elmer Models manual for details.
2) I would rather see this as infinitely long. It would be difficult to apply current to a infinitely thin conductor.
3) An infinitely long device would have infinite resistances etc. So it is convenient to have the length as a parameter when modeling with external circuits. So it that sense the true length is considered but the resulting fields are of course still not function of z. Considering end-effects would require 3D modeling.
-Peter