Discussion Forum

Hi

I believe it's not only a "newbie" effect, you are also using the early 4.0 that was the first after the new approach of V4. Comsol is supposed to send out the 4.2 very soon, I have heard. You should absolutely get hands on the latest version, it will make you far less frustrated!

There is already something bizarre with your geometry one of the boundaries is not selectable as it should, could also be an early version issue.

Then you need to define a material (easiest way to get a mur) or at least also define mur somewhere, also for the air.

Finally, I'm not sure you have correct symmetry conditions, you might want to use a identity pair section symmetry there

In ACDC I find it often easier to run a full 3D model (its just that I often also get confused on the E, B and H symmetry conditions needed, I haven't found the "obvious" validation method yet ;)

--
Good luck
Ivar

Thank you very much Ivar. I will anxiously await the arrival of 4.2

Hi

we all are I believe, in the mean time the "V4.1 update 3" is already very good and stable

--
Good luck
Ivar

Hi,

The only obvious problem I could find with this model is that the relative permeability ("mu_r") is taken from the material but no material is defined. Either add a predefined material with a value of mu_r, add a custom material and define its value for mu_r, or use a user-defined value for mu_r directly in the Magnetic Flux Conservation node's Settings window.

Best regards,
Magnus Ringh, COMSOL

Ok I took some of the above advice and did a full 3D model for the time being. In the course of my work today i noticed a few very bizarre seeming bugs. I will include the latest version of my simulation to this post and also note here the strangeness I have encountered.

The order of certain things matters in a way that I don't understand:

Order in meshes:

Say I have two 'Free Tetrahedral' meshes
-One whose domain list includes only a large box that surrounds everything in my model (call this the 'Air Mesh')
-One for all of the other pieces (call this the 'Component mesh')

-Say that I have defined the air mesh to be 'Normal' and the component mesh to be 'fine'
-If the air mesh appears first in the list of meshes, the interior objects wind up with a 'normal' mesh instead of a fine mesh.
-if the air mesh appears second on the list, then the air mesh is 'normal' and the components are 'fine' (this is the behavior i would naively expected regardless of the ordering, unless the domain lists for the two meshes had common objects, which they do not in this case)

if anything I could understand if it worked the other way around, where finer meshes on interior parts took precedence unless explicitly overwritten, as it stands I am failing to understand the behavior. For the boundary conditions there seems to be a very simple way of seeing when earlier set parameters are overwritten by ones farther down the list. Meshes don't work this way I see.

You can see this behavior easily by switching the order of the two meshes in my attachment



Order in geometries:

When I built my model originally I first defined the outer box (by extruding a square from a workplane), then proceeded to make the components which are in its interior. When I make the solution I found that when I viewed 'slices' they never showed any data in the air space between components. However, I could see arrows in this region, so I assume it must have solved the problem there.

Recalling that order mattered for some reason for the meshed: I deleted the air box, and it's associated workplane, created a new workplane and extruded it. The only difference being that now the air box was at the bottom of the geometry list. After doing this I could see slices without a problem. This is doubly weird because since there doesn't seem to be a facility for changing the order of geometries like you can with meshes I don't imagine that the order should matter.

In my 3D and 1D plots, i can select my solution from the drop-down 'data set' list. But for 2D i don't see the solution on the drop down list. Why is this?

I am still waiting on our IT people to try and get me hooked up with a newer version, but I didn't even see a section labeled bug fixes in the release notes for the newer versions, I don't know how I can find out if there are known issues or not.

Hi

a few comments, as you have many questions ;)

In 3D you can generate these primitive solids as is, then "array" rotate them to the manyfold

the meshes adapt to the size of the objects, that is probably why you see them at different sizes, you can edit some of the many parameters in the "custom" tab to influence the meshing behaviour in more detail. The only thing important to learn is to set the "last" mesh node to Entity level : "remaining" so that you are sure ALL domains are meshed

Then because of the symmetry of your model, I would suggest to use a calinder of air around, this will influence less the B field lines that you will get now from the "square corners"

You can also define a "cylindrical coordinate", that would make your magnet Br field definition simpler for you see the Definition node

By the way are these soft iron with a coil around to generate the Br, or are these "magnets", if the latter is the case, I do not believe that true magnets have a mur of 4000 like iron, rather mur=1 or close to "1"

Working in the geometry section requires some exercise. To add anew item in the middle you must 1) select the item above , right click and do a build to selected" or select the object below and select build to preceding. You will see a small green square around the icon to the level of "build" geometry nodes. Then you might add a new item that will be inserted just after the node with the green box.

You have a 3D data set, so you can only generate 3D data from it. You must right click the Data Set Node and add a variant of your existing Data set, you can cut it with a plane , a line , select edges, surfaces ... This will generate 2D data sets or 1D data sets. Then you can find something in the 2D plot group selection list.

COMSOL takes some time to learn as there are many "buttons" but it's real fun and it is incredible powerful to illustrate and study physics. By the way, after some weeks training, it is absolutely worth to follow some of the COMSOL 1-2 day training courses, it will boost your productivity

--
Good luck
Ivar

Thanks for taking the time to reply Ivar.