Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.
Maxwell Stress Tensor ignoring imaginary component
Posted 8 déc. 2016, 12:49 UTC−5 Low-Frequency Electromagnetics Version 5.2 1 Reply
Please login with a confirmed email address before reporting spam
Hello,
I am running a model with the original intent to calculate the forces on a permanent magnet from a coil. The original issue was that when using the frequency domain solver it added the frequency to the magnetization of my magnet.
We have since broken the simulation into two aspects one where the magnet is solved for using a stationary solver with the magnetic fields physics and the other for the coil using the frequency domain solver and magnetic and electric fields physics. Before attempting anything else we wanted to verify the force calculations are correct as we have had issues with the tensor calculation in the past.
My fist step in doing this was to enter the expression for the tensor in the direction we care about and compare it the one COMSOL calculated itself. These seemed to match. We then did the global evaluation to view the force values COMSOL calculated. I then computed the integral to get the force using the expression I wrote out for the tensor and the one COMSOL came up with.
The results I got with the stationary solver was exactly the same; however, the results with the frequency domain solver did not match. The value from my expression was complex. The global evaluation was just the real part of result of my expression with differences starting in the third digit to the right of the decimal place.
To explore further we then plotted and compared the tensor values to my expression. It turned out the tensor's real part for each method matched, but the Tensor value as solved for by COMSOL has an imaginary component of zero.
Is there a reason for this? Am I mistaken to keep the tensor value as complex after computing it the long way? We do not understand how the real parts are matching when the imaginary part does not seem to be being used in any of the calculations.
Any help is appreciated.
Regards,
Kyle E. Laferty
I am running a model with the original intent to calculate the forces on a permanent magnet from a coil. The original issue was that when using the frequency domain solver it added the frequency to the magnetization of my magnet.
We have since broken the simulation into two aspects one where the magnet is solved for using a stationary solver with the magnetic fields physics and the other for the coil using the frequency domain solver and magnetic and electric fields physics. Before attempting anything else we wanted to verify the force calculations are correct as we have had issues with the tensor calculation in the past.
My fist step in doing this was to enter the expression for the tensor in the direction we care about and compare it the one COMSOL calculated itself. These seemed to match. We then did the global evaluation to view the force values COMSOL calculated. I then computed the integral to get the force using the expression I wrote out for the tensor and the one COMSOL came up with.
The results I got with the stationary solver was exactly the same; however, the results with the frequency domain solver did not match. The value from my expression was complex. The global evaluation was just the real part of result of my expression with differences starting in the third digit to the right of the decimal place.
To explore further we then plotted and compared the tensor values to my expression. It turned out the tensor's real part for each method matched, but the Tensor value as solved for by COMSOL has an imaginary component of zero.
Is there a reason for this? Am I mistaken to keep the tensor value as complex after computing it the long way? We do not understand how the real parts are matching when the imaginary part does not seem to be being used in any of the calculations.
Any help is appreciated.
Regards,
Kyle E. Laferty
1 Reply Last Post 9 déc. 2016, 11:14 UTC−5
Please login with a confirmed email address before reporting spam
Posted:
8 years ago
Hello Kyle,
You have a good point. The Maxwell stress tensor variables that we are providing are not complex-valued phasors but rather tailored as "instantaneous values" that give the correct value when plotting vs. phase angle as in the attached model (Axial Force vs. Phase).
For 2D and 3D plots, you can control the phase of the solution in the settings for the Data Set (Results->Data Sets->Study 1/Solution 1) or you can (in 1D plots of scalars/tensore elements) set the x-axis data to phase.
When animating at any single frequency, using the "Sequence type" set to "Dynamic data extension" you will also cycle the phase of the solution.
An important point here is that the instantaneous force/stress will oscillate around the cycle average at twice the frequency solved for. This is also captured in the plot "Axial Force vs. Phase".
--
Magnus
You have a good point. The Maxwell stress tensor variables that we are providing are not complex-valued phasors but rather tailored as "instantaneous values" that give the correct value when plotting vs. phase angle as in the attached model (Axial Force vs. Phase).
For 2D and 3D plots, you can control the phase of the solution in the settings for the Data Set (Results->Data Sets->Study 1/Solution 1) or you can (in 1D plots of scalars/tensore elements) set the x-axis data to phase.
When animating at any single frequency, using the "Sequence type" set to "Dynamic data extension" you will also cycle the phase of the solution.
An important point here is that the instantaneous force/stress will oscillate around the cycle average at twice the frequency solved for. This is also captured in the plot "Axial Force vs. Phase".
--
Magnus
Attachments: