Difference between Circuit and MagnetoDynamics currents

[mr_monkey]

Hi,

I am trying to calculate currents induced in individual conductors in an electric motor (similar to this forum post).

But I am getting a mismatch between the Component currents calculated by the CircuitsAndDynamics solver and the currents I calculate by integrating `current density e` over the conductor bodies.

The Component currents look somewhat reasonable and add up to the total current I put into the model:

component_currents.png (78.79 KiB) Viewed 267 times

The currents calculated by integration are different and oscillate wildy:

fea_currents.png (89.73 KiB) Viewed 267 times

I have tried this on a simplified case with just a few conductors and no other parts of the motor and the currents calculated by both different methods were the same.

Are there maybe any convergence settings that need tweaking? Or is there something more fundamentally wrong with my model setup.

I have attached the case setup in the zip file. Any input would be much appreciated.

[mr_monkey]

This is a diagram of the circuit:

circuit.png

(368.18 KiB) Not downloaded yet

[raback]

Hi

Wow, what a circuit! Do you have some tool for building it?

pdf2 is not the most stable time-integration scheme. Maybe you can try 1st order.

-Peter

[mr_monkey]

Wow, what a circuit! Do you have some tool for building it?

I have used elmer_circuitbuilder to create the circuit. It is fairly straightforward using that. Just requires modifying the circuits.definition file afterwards to put the time varying currents in.

Plotting the circuit was done using pydot, which uses graphviz as a backend. Again, that's pretty straightforward once the circuit has been created.

pdf2 is not the most stable time-integration scheme. Maybe you can try 1st order.

Well, that was an easy fix. The currents are matching after changing the timestepping order to 1. Thanks a lot.

[raback]

Hi, Maybe you can share the improved comparison as well. -Peter

[mr_monkey]

Sure, it is a bit tricky to visualise, since there's so many lines and no visible difference anymore, but I gave it a shot:

Individual currents (Left: Circuit Components, Right: FEA surface integral):

currents_comparison.png

(341.42 KiB) Not downloaded yet

Sum of all currents:

currents_sum_comparison.png (34.23 KiB) Viewed 196 times

[vencels]

Hi,
It looks that we are having the same problem
https://www.elmerfem.org/forum/viewtopic.php?t=8238

How do you excite your coils - voltage or current?

[mr_monkey]

How do you excite your coils - voltage or current?

My coils are current excited. I haven't tried voltage excitation.

My problem was fixed by switching

Code: Select all

BDF Order

from 2 to 1. Have you tried using BDF with order 1 for your case?

[vencels]

This does not make a difference in my case. Changing time-stepping schemes worked for cases with lower frequency, but this high-freq still has these oscillations. Making smaller timesteps seems to make things even words.

Screenshot_1.png (108.09 KiB) Viewed 175 times