Need Guidance on Direct Magnetic Field Generation with Gmsh/Elmer

[Tom_B]

Hello everyone,

I am seeking some assistance with a challenging project that I've undertaken for a friend. The goal of the project is to measure the fluctuations of the magnetic field around a metallic object, in this case, a simple cuboid made of iron, copper...

I come from a mathematical background(statistics, ML…) , and although I've been trying to familiarize myself with this field, this has proven to be an overwhelming task. As a result, I'm reaching out in the hopes of finding some guidance here.

From the research I've done so far, I've found that the Gmsh/Elmer software could be a potential solution. My task is to generate a magnetic field directly, without constructing a magnet or a solenoid.

To start with, I want to create a homogeneous magnetic field, and then proceed to a magnetic field generated by a pure magnetic dipole. Ultimately, I wish to explore the creation of a random pseudo-static magnetic field.

I am particularly interested in examples where someone has directly generated a magnetic field using these methods and would greatly appreciate if anyone could provide me with a roadmap to accomplish this. Any advice, insights, or recommendations would be incredibly valuable to me in navigating this task.

I look forward to your replies and am thankful in advance for any assistance you may provide. This is a great community and I'm confident that I will find the help I need here.

[raback]

Hi

If you want to create an add'hoc magnetic field what do you need FEM for? I would see more natural problem setup, for example, to try to find a coil configuration that creates the desired magnetic field. If you already know what kind of field you want I don't see how Elmer can help.

If you have a pure magnetic dipole I think you could use simple analytical expression:
https://en.wikipedia.org/wiki/Magnetic_dipole

-Peter

[Tom_B]

Hi Peter,

Thank you for your thoughtful response. You're right that the magnetic dipole would be the first line of approximation in an ideal scenario. However, in our case, we're dealing with an additional complexity. The application we're working on involves a sensor (3-coil arrangement) that is positioned close to a metallic object. This metallic object is causing a perturbation in the magnetic field which results in a consistent error in our distance measurements - they're always 2/3 cm off (it should be less than 1mm).

The reason we are considering Finite Element Method (FEM) simulations, and more specifically a tool like Elmer, is to better understand and model this interaction. By modeling the magnetic field in the presence of this metal object, we hope to understand the pattern of perturbation it causes. This, in turn, could potentially help us adjust our measurements or algorithms to account for this specific scenario.

Would you have any thoughts or suggestions on how we might go about this?

[raback]

Hi,

Ok, got you. Sounds like a good application for WhitneyAVSolver.

It is difficult to "order" a certain type of field. Instead you could, for example, create a permanent magnet or a coil that results to the type of idealized field you're looking for. Then you can add objects to your simulation and study their effect.

I guess you could also have idealized farfield and give BCs for WhitneyAVSolver in terms of B. Then the assumption is that the far field is not affected by the local disturbance. For example, you would not like to include the earths iron core to your simulation setup.

-Peter

[Tom_B]

Hi,

I appreciate your response. A friend of mine and I were recently discussing the potential use of a permanent magnet or a coil in a different simulation software. However, he noticed that most of the "computational power" (to use his words) was consumed near the coil, and he had a tough time getting reasonable results.

This got me thinking - is this a common problem I might encounter as well? If it is, would it be possible to tweak the mesh density and decrease it in the area near the coil or magnet? Would such an adjustment influence the results significantly?

I should admit, my experience with Elmer or any other FEM tool is quite limited, so there's a good chance I may be misunderstanding some key elements here. But I'm eager to learn and get a grasp on it.

If you could provide an example or some sort of guide that would help me understand how to better handle my problem, I would greatly appreciate it.

Thank you in advance for your time and patience.

Best regards,

[raback]

Have you looked for potential starting points here:
https://github.com/ElmerCSC/elmer-elmag

The mesh around the source does not need to be too fine as long as you are just interested in the relative perturbations. Ultimately the source could perhaps be given in one element only.

-Peter