Learning Forum

[sebastian93]

Hi,

I'm trying to compare a simple simulation on a wire with the analytic approach. Unfortunately, no matter how fine the mesh in the simulation it doesn't match the analytic solution. I also checked the material properties in my simulation (resistance and magnetic permeability), but those seem to be correct. I used a cylindrical 50mm long wire with a radius of 1,3mm. Therefore I could use the Biot-Savart law for straight, finite wires for my analytic approach.

I used the magnetostatic tool in Ansys Workbench. I know that my simulation uses Solid117 elements for the calculation, but I couldn't find out how magnetic fields created by such elements are calculated. Pictures of the equation I used and the comparison of both magnetic fields are attached. Maybe someone has an idea and can help me out with that problem.

Best regards

Sebastian

 

[timlinNL]

Hi Sebastian,

 

If you get rid of the connection lines, then you'll see the data points you got from the simulator actually match quite well with your analytical results. This means that you can get a "more comparable" results by simply increase the resolution of your data point. I am attaching two curves here, both from Maxwell 2D simluation. You can see the one with 1001 points shows a good result while the one with 105 points show a similar result of yours.

Always be aware of your resolution in both data presenting and mesh. Hope that helps.

Tim

 

[sebastian93]

Hi Tim,

that's it! Thank you so much, I can't believe I didn't notice that. You're my hero of the day!

[sebastian93]

Hi,

sorry, I still have a problem concerning this issue. The peaks now match in both the simulation and the analytic calculation, but if you move further to the side on the x-axis it still doesn't match.

Also these are the values in the middle of the wire. According to Biot-Savart the magnetic field must become weaker at the ends of the wire, but in the simulation this is not the case. Here it is just as strong as in the middle as it would be with an infinite wire.

Do you also have an explanation for that? Thanks a lot for your help.

[timlinNL]

Hi Sebastian,

In this case you may have missed another important factor which also doesn't exist in real-life but significiant in FEA simulation: the simulation Region. 

Check your region settings, it sould be far enough from the region of interest (in this case, the entire wire) if you use the defult boundary condition for the "Region". Otherwise, it should be set as Bolloon so that the magnetic flux can naturally strech out of the bounding box.

 

You can also play with the size of the Region and see how it affects the results.

 

Tim

[sebastian93]

Hi Tim,

unfortunately, I'm not sure where to find those region settings. For the boundaries of my system I set the "magnetic flux parallel" condition, but I assume that's not what you're referring to.

Sebastian

[timlinNL]

Hi Sebastian,

I noticed that you are using 3D instead of 2D Maxwell for such a simple simulation. In fact, the way you model the wire in 3D is illogical since the current has to come from something. In Maxwell Magnetostatic Solver, the excitation face has to contact the edge of the boundary, suggesting a continuous condition outside of the simulation boundary. In this case your wire is no longer a 50-mm wire "float" in the air while carrying current. 

If you are using 3D simulation, the geometry has to make some sense, e.g. an entire close loop. Otherwise, you can go back to 2D and you can see the effect of setting improper boundary conditions as I described in the previous reply.

Tim

[sebastian93]

Hi Tim,

thank you. I tried a different way of modeling and now my solution looks almost identical to the Biot-Savart equation for infinite wires.

Sebastian