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 The information provided may be out of date.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.

individual mode contributions in Modal Analysis

Please login with a confirmed email address before reporting spam

Hello,

I would like to know if it is possible to obtain individual mode contributions in a modal analysis response graph?

I have some resonant peaks in my structure whose gains are low and they are barely visible on the overall response. It would have been nice to have the sdof responses of individual modes on one graph without the need of MatLab.

Thanks in advance,
Onur

5 Replies Last Post 2012年9月7日 GMT-4 09:48
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 2012年4月17日 GMT-4 09:50
Hi

open the lower level eigenfrequency solver node and replace the RMS normalisation by the "participation mass"
You can then extract Derived Variable Global Expressions Mass participation factors, the sum of the squares should sum up to the total mass minus the residual of not shown modes.

Note these are only the linear displacement mass particpation factors, there is not (yet) the rotation inertial
participation factors. I hope they will arrive once too ;)
--
Good luck
Ivar
Hi open the lower level eigenfrequency solver node and replace the RMS normalisation by the "participation mass" You can then extract Derived Variable Global Expressions Mass participation factors, the sum of the squares should sum up to the total mass minus the residual of not shown modes. Note these are only the linear displacement mass particpation factors, there is not (yet) the rotation inertial participation factors. I hope they will arrive once too ;) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2012年4月17日 GMT-4 10:27
Hello Ivar,

Thank you for your response.

The mass participation factor calculation is something that I am using at the moment. This gives me the dc gains of each mode. For the report generation and visualization of the results, I have to generate the frequency vector and modal matrices so that I can create sdof gain and frequency response of each resonance and display them on a single frequency response graph. I wondered if somebody have done this within Comsol without using any MatLab?

As you would see this is just to visualize the mass participation factor results in a different way.

Regards,
Onur
Hello Ivar, Thank you for your response. The mass participation factor calculation is something that I am using at the moment. This gives me the dc gains of each mode. For the report generation and visualization of the results, I have to generate the frequency vector and modal matrices so that I can create sdof gain and frequency response of each resonance and display them on a single frequency response graph. I wondered if somebody have done this within Comsol without using any MatLab? As you would see this is just to visualize the mass participation factor results in a different way. Regards, Onur

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 2012年4月17日 GMT-4 10:56
Hi

Indeed it would be of interest, but I have had no time for that so far, hopefully there are someone out there. There was/is the model reduction function but it takes only the first n modes, which mostly is not valid for me, I need a model reduction with selectable modes, as my systems has many, and only a few far apart oare of true interest.

just as the SOLID phyiscs has no PSd vibration/shock predefined load possibilities, you must write out everything by hand, clumy I find. Hopefully it will come once.

Would be great if someone had already soemthing nice to shear, to build upon further ;)

--
Good luck
Ivar
Hi Indeed it would be of interest, but I have had no time for that so far, hopefully there are someone out there. There was/is the model reduction function but it takes only the first n modes, which mostly is not valid for me, I need a model reduction with selectable modes, as my systems has many, and only a few far apart oare of true interest. just as the SOLID phyiscs has no PSd vibration/shock predefined load possibilities, you must write out everything by hand, clumy I find. Hopefully it will come once. Would be great if someone had already soemthing nice to shear, to build upon further ;) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2012年9月7日 GMT-4 00:24
Dear Ivar,

I am in a serious problem. I am looking a solution from you. Please help me.

When I am doing modal analysis, how many mode should I extract? For complex piping system with thinner wall and less diameter the no of modes to be consider is a difficult task. How can I define upto how many modes should I evaluate for the modal results and check it with external excitation.

I am surprise there is no clear cut answer to this till now, as far as I know and I got. There are very few documents where they have written the mass participation factor, and it's direction, external excitation direction and mode shape determine the no of modes we have to consider. As far as my understanding, I also think like that only and do the same.

But there is a little bit confusion with my friends, who told me only first few modes (for piping four modes) is need to examine for resonance, no need to go for higher modes even though mass participation factor is high enough!!!

You please tell me for complex 3d geometry, is that the nodes of vibration always lie on the geometry itself??

And is it possible that the amplitude at higher modal frequency can be higher than that of lower modal frequency??

And for continuous systems how many fundamental frequency are there??

I am seriously looking for it's answer. Please help me.
Thanks in advance...Lot's of love.....
Dear Ivar, I am in a serious problem. I am looking a solution from you. Please help me. When I am doing modal analysis, how many mode should I extract? For complex piping system with thinner wall and less diameter the no of modes to be consider is a difficult task. How can I define upto how many modes should I evaluate for the modal results and check it with external excitation. I am surprise there is no clear cut answer to this till now, as far as I know and I got. There are very few documents where they have written the mass participation factor, and it's direction, external excitation direction and mode shape determine the no of modes we have to consider. As far as my understanding, I also think like that only and do the same. But there is a little bit confusion with my friends, who told me only first few modes (for piping four modes) is need to examine for resonance, no need to go for higher modes even though mass participation factor is high enough!!! You please tell me for complex 3d geometry, is that the nodes of vibration always lie on the geometry itself?? And is it possible that the amplitude at higher modal frequency can be higher than that of lower modal frequency?? And for continuous systems how many fundamental frequency are there?? I am seriously looking for it's answer. Please help me. Thanks in advance...Lot's of love.....


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 2012年9月7日 GMT-4 09:48
Hi

I would not say you are "in trouble", at least not more than any of us enginners working with structural vibrations ;)

Well, from my experience, if you choose the modal mass normalisation the sum of the squares of the mas participation factors should add up to the total mass (for each direction) and respectively to the total inertia for the rotational ones, but these are not yet implemented in COMSOL representation of mass participation factors (that I regret ;)

So normally you should pick only the major ones. Depending on your model there could be many minors in between the major frequencies, and to get a simplified model you should ignore these minor, BUT today COMSOL does not really allow you to create easily (at least I have not found out the way yet) to make a simplified model based on discontinous eigenmodes that you pick at will, mostly you get the first "n" and that is all, even if you would like to pick from a list ... (and dump the M,K,D highly reduced matrices to your control engineers)

Another thing to consider is the damping. Normally eigenmodes are the energy representation of all possible modes, and damping is ignored, and normally, as higher you go in frequency, the better the modes are damped (in most engineering cases) so the true energy in each mode tends to decrease rapidly with freqency. But this is very enginnering model dependent, that is depends on fixation means, materials etc, but also on excitation possibilities.
An example: if you have some piping going to a high power transformer operating at 50 Hz, you should consider modes around your structural modes 50, 100, 150 etc, and any close to, depending on how your ingeneering sense tells you these will get some energy from the previously excited modes, but a "srong" mode at 75 or 125 might well be ignored (again engineering sense is required).

In anycase a model remains a simplified representaton of real life, so you need to validate it carefully by measurements and tests, and build up in this way your own experience to juge the validity of a model.
Just one important remark here, when you change domain or only scale, within a given engineering domain, be aware that your ingineering sense will ofte be fooled as many phyiscal phenomena scale very differently with size, I have burnt my finger a few times on this over the years, so ALWAYS take some time to validate trough some freah measurements, and if possible some hand calculations

--
Good luck
Ivar
Hi I would not say you are "in trouble", at least not more than any of us enginners working with structural vibrations ;) Well, from my experience, if you choose the modal mass normalisation the sum of the squares of the mas participation factors should add up to the total mass (for each direction) and respectively to the total inertia for the rotational ones, but these are not yet implemented in COMSOL representation of mass participation factors (that I regret ;) So normally you should pick only the major ones. Depending on your model there could be many minors in between the major frequencies, and to get a simplified model you should ignore these minor, BUT today COMSOL does not really allow you to create easily (at least I have not found out the way yet) to make a simplified model based on discontinous eigenmodes that you pick at will, mostly you get the first "n" and that is all, even if you would like to pick from a list ... (and dump the M,K,D highly reduced matrices to your control engineers) Another thing to consider is the damping. Normally eigenmodes are the energy representation of all possible modes, and damping is ignored, and normally, as higher you go in frequency, the better the modes are damped (in most engineering cases) so the true energy in each mode tends to decrease rapidly with freqency. But this is very enginnering model dependent, that is depends on fixation means, materials etc, but also on excitation possibilities. An example: if you have some piping going to a high power transformer operating at 50 Hz, you should consider modes around your structural modes 50, 100, 150 etc, and any close to, depending on how your ingeneering sense tells you these will get some energy from the previously excited modes, but a "srong" mode at 75 or 125 might well be ignored (again engineering sense is required). In anycase a model remains a simplified representaton of real life, so you need to validate it carefully by measurements and tests, and build up in this way your own experience to juge the validity of a model. Just one important remark here, when you change domain or only scale, within a given engineering domain, be aware that your ingineering sense will ofte be fooled as many phyiscal phenomena scale very differently with size, I have burnt my finger a few times on this over the years, so ALWAYS take some time to validate trough some freah measurements, and if possible some hand calculations -- Good luck Ivar

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.