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 The information provided may be out of date.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.

random vibration analysis

Please login with a confirmed email address before reporting spam

Hello everybody,

I have PSD acceleration values of shaker [attached with this post ] and I am trying to apply it on my model, however I could not find a way to enter PSD directly as a boundary condition, so I converted it to frequency domain in order to use Frequency modal study.
The PSD load is applied in the reality as acceleration on the basement of the structure, where usually I have them as a fixed constraint in Comsol model. How can I apply this basement acceleration on my model for the modal study?
I tried to convert them to force by multiplying them with total mass of my structure (I am not sure if this step is correct)
Has someday experience with PSD analysis with Comsol?
Thanks in advance
Ghalib


2 Replies Last Post 28.06.2012, 09:02 GMT-4
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 27.06.2012, 10:00 GMT-4
Hi

that is an interesting issue, indeed you cannot input a PSD like that (you could try a random excitation with a LP filter but it would take ages ...

The best I can propose is how you do it, add a frequency sweep (then you are in harmonic mode) shape the input amplitude to follow your PSD shape (take care for the conversion between amplitude, rms and PSD (^2) values).

But then one comes to the issue: acceleration input or body load acceleration ? This is not strictly the same ! Here for the moment I have not found any other methods than the classical "big mass" approach, attaching your device to a large mass and run a body acceleration load case

There is also the frequency domain - modal that could be considered if your model is very heavy

There are a few threads on the issue

PS do not forget to put your eigenfrequency analysis to the solver modal mass normalisation to get the participation factors normalisation, check the COMSOL doc

--
Good luck
Ivar
Hi that is an interesting issue, indeed you cannot input a PSD like that (you could try a random excitation with a LP filter but it would take ages ... The best I can propose is how you do it, add a frequency sweep (then you are in harmonic mode) shape the input amplitude to follow your PSD shape (take care for the conversion between amplitude, rms and PSD (^2) values). But then one comes to the issue: acceleration input or body load acceleration ? This is not strictly the same ! Here for the moment I have not found any other methods than the classical "big mass" approach, attaching your device to a large mass and run a body acceleration load case There is also the frequency domain - modal that could be considered if your model is very heavy There are a few threads on the issue PS do not forget to put your eigenfrequency analysis to the solver modal mass normalisation to get the participation factors normalisation, check the COMSOL doc -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 28.06.2012, 09:02 GMT-4
Hi Ivar,

thank you very much for your answer, I was expecting that you will answer :), great thanks!
I have applied a harmonic force (body load) on "big mass" only, which fixed to ground and I attached my structure on it. Then I checked the acceleration on the connecting surface between the structure and the "big mass" but I did not get the acceleration values that I need. I think it depends on the geometry of the used "big mass".

Another issue I am looking for evaluating the maximum stress on my structure resulting from PSD load. if I succeed to make the frequency modal analysis, I still need to know the stress resulting from all frequencies at the same time. So I though to go further to time domain and make transit modal analysis, but here I do not have information about the phase shift of each frequency, maybe I try to take worst case scenario?
Does Comsol have a way to combine deformations and stresses resulting from different frequencies to gather?

One point regard converting PSD to frequency domain is that how I sample the PSD data, the more frequency points I take the lower the Grms values of each, but I still think its ok, right? I think the sampling must be related to the eigenfrequencies of the structure.

What I got from PSD is 1 sigma(standard deviation) which occurs for 68% of the time, if I need a higher probability I have to multiply frequency magnitude by 3? and then calculate again ?

I had a look on the MPF for the eigenfrequencies, man must be carefull while dealing with remote masses of rigid connectors, because they are not considered automatically in the the calculation of MPF, I had to make small cubes and assign to them material inroder to have them in MPF calculations.

many thanks in advance
Best Regards
Ghalib

Hi Ivar, thank you very much for your answer, I was expecting that you will answer :), great thanks! I have applied a harmonic force (body load) on "big mass" only, which fixed to ground and I attached my structure on it. Then I checked the acceleration on the connecting surface between the structure and the "big mass" but I did not get the acceleration values that I need. I think it depends on the geometry of the used "big mass". Another issue I am looking for evaluating the maximum stress on my structure resulting from PSD load. if I succeed to make the frequency modal analysis, I still need to know the stress resulting from all frequencies at the same time. So I though to go further to time domain and make transit modal analysis, but here I do not have information about the phase shift of each frequency, maybe I try to take worst case scenario? Does Comsol have a way to combine deformations and stresses resulting from different frequencies to gather? One point regard converting PSD to frequency domain is that how I sample the PSD data, the more frequency points I take the lower the Grms values of each, but I still think its ok, right? I think the sampling must be related to the eigenfrequencies of the structure. What I got from PSD is 1 sigma(standard deviation) which occurs for 68% of the time, if I need a higher probability I have to multiply frequency magnitude by 3? and then calculate again ? I had a look on the MPF for the eigenfrequencies, man must be carefull while dealing with remote masses of rigid connectors, because they are not considered automatically in the the calculation of MPF, I had to make small cubes and assign to them material inroder to have them in MPF calculations. many thanks in advance Best Regards Ghalib

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.