Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.
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.
Mismatch between time and frequency domain simulations
Posted 14 nov 2011, 08:26 GMT-5 Acoustics & Vibrations, Structural Mechanics Version 4.1, Version 4.2 6 Replies
Please login with a confirmed email address before reporting spam
Hi,
I've been experiencing a strange mismatch between time and frequency domain solid mechanics simulations. I have a tube in which I want to excite a torsional wave. I did it by adding 8 points on the circumference, in a ring configuration, where I specify point load boundary conditions in the tangential direction.
When running a time domain simulation, I clearly see that the resulting excitation is a torsional wave, by looking at the direction of the displacements. However, when I use exactly the same system with frequency domain simulation, at a given (any) arbitrary frequency, the resulting displacements are in the radial direction (i.e. breathing mode) instead of tangential displacements in torsional waves.
Attached is the COMSOL model. To view the results one has to run Study 1 and Study 2.
Any ideas?
Many Thanks,
Silviu.
I've been experiencing a strange mismatch between time and frequency domain solid mechanics simulations. I have a tube in which I want to excite a torsional wave. I did it by adding 8 points on the circumference, in a ring configuration, where I specify point load boundary conditions in the tangential direction.
When running a time domain simulation, I clearly see that the resulting excitation is a torsional wave, by looking at the direction of the displacements. However, when I use exactly the same system with frequency domain simulation, at a given (any) arbitrary frequency, the resulting displacements are in the radial direction (i.e. breathing mode) instead of tangential displacements in torsional waves.
Attached is the COMSOL model. To view the results one has to run Study 1 and Study 2.
Any ideas?
Many Thanks,
Silviu.
Attachments:
6 Replies Last Post 16 dic 2011, 05:22 GMT-5
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
as your tube is free I see a rotating tube there and not really a mode, or ?
--
Good luck
Ivar
as your tube is free I see a rotating tube there and not really a mode, or ?
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Dear Ivar,
Thank you so much for your comment. You are absolutely right, the way I defined it induces also a rigid rotations. I fixed it by fixing the two faces at the tube entrance/exit.
However, after fixing the problem the differences between time domain and frequency domain are left unchanged. You can see it from the displacement arrows in plots 1 and 4.
Do you see any other problems?
Many thanks,
Silviu.
Thank you so much for your comment. You are absolutely right, the way I defined it induces also a rigid rotations. I fixed it by fixing the two faces at the tube entrance/exit.
However, after fixing the problem the differences between time domain and frequency domain are left unchanged. You can see it from the displacement arrows in plots 1 and 4.
Do you see any other problems?
Many thanks,
Silviu.
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
try to change the mesh size and observe the pattern change, then compare the pattern to the mesh size, you will notice that you have a nice correlation (in frequency mode),
what kind of oscillation do you expect at that frequency ? then make an estimation of the standing wavelength hence the mesh density, as you need some five 2nd order elements per period of your standing wave
If you want to identify the rotations, you have the "small angle values" defined as Thx=0.5[rad]*solid.curlUX, respectively ...UY and UZ for Thy and Thz
--
Good luck
Ivar
try to change the mesh size and observe the pattern change, then compare the pattern to the mesh size, you will notice that you have a nice correlation (in frequency mode),
what kind of oscillation do you expect at that frequency ? then make an estimation of the standing wavelength hence the mesh density, as you need some five 2nd order elements per period of your standing wave
If you want to identify the rotations, you have the "small angle values" defined as Thx=0.5[rad]*solid.curlUX, respectively ...UY and UZ for Thy and Thz
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Dear Ivar,
Thanks for your reply.
I tried using denser grids (to the extent my PC allows) but still I get the same qualitative behavior. Moreover, if it is a grid issue, how come it seems to be qualitatively correct in the time domain and incorrect (or at least different) in the frequency domain, for the same mesh. Is there a different sensitivity to mesh density in the two computation modes?
Could you also elaborate some more about identifying rotations using "small angle values"?
Many thanks,
Silviu.
Thanks for your reply.
I tried using denser grids (to the extent my PC allows) but still I get the same qualitative behavior. Moreover, if it is a grid issue, how come it seems to be qualitatively correct in the time domain and incorrect (or at least different) in the frequency domain, for the same mesh. Is there a different sensitivity to mesh density in the two computation modes?
Could you also elaborate some more about identifying rotations using "small angle values"?
Many thanks,
Silviu.
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
an eigenfrequency analysis gives infinte energy to all modes, a time analysis is "softer" as you are driving it, and a frequency domin analysis is a harmonic development of the equations. All slightly different in the way modes and their energy might interact.
But have you anaysed by hand what you expect as twist modes, for these dimensions and material(s) ? It's always necesary to check and validate your model by some simple analytical checks.
And to make the model "lighter", if it's a "thin tube" you could consider to use a "shell tube" in "shell physics, or even a 2D-axi model, but then you restrict yourself only to 2d-axi symmetric modes, the other dissappear by design, which might not be what you want ;) be aware that shell physics has more and other dependent variables than solid
--
Good luck
Ivar
an eigenfrequency analysis gives infinte energy to all modes, a time analysis is "softer" as you are driving it, and a frequency domin analysis is a harmonic development of the equations. All slightly different in the way modes and their energy might interact.
But have you anaysed by hand what you expect as twist modes, for these dimensions and material(s) ? It's always necesary to check and validate your model by some simple analytical checks.
And to make the model "lighter", if it's a "thin tube" you could consider to use a "shell tube" in "shell physics, or even a 2D-axi model, but then you restrict yourself only to 2d-axi symmetric modes, the other dissappear by design, which might not be what you want ;) be aware that shell physics has more and other dependent variables than solid
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
an eigenfrequency analysis gives infinte energy to all modes, a time analysis is "softer" as you are driving it, and a frequency domin analysis is a harmonic development of the equations. All slightly different in the way modes and their energy might interact.
But have you anaysed by hand what you expect as twist modes, for these dimensions and material(s) ? It's always necesary to check and validate your model by some simple analytical checks.
And to make the model "lighter", if it's a "thin tube" you could consider to use a "shell tube" in "shell physics, or even a 2D-axi model, but then you restrict yourself only to 2d-axi symmetric modes, the other dissappear by design, which might not be what you want ;) be aware that shell physics has more and other dependent variables than solid
--
Good luck
Ivar
Hi Ivar,
I've just started working on the acoustic domain.
I'm stuck at a simple and elementary problem.
The model has a point source at the center with spherical wave radiation as a boundary condition.
I'm checking the polar plots for the sound levels at different distances from the point source.
Expected results would have concentric pressure level polar plots at different radii.
The model does not show that. I've attached the model in this post.
I'm quite confused and lost at this moment.
Am i missing a boundary condition or is my post processing at fault.
Any kind of response will be of great help.
Thanks a lot for your time and effort.
Regards
Glenston
Attachments: