Modal vibration analysis of 2 beams in ansys

[vasanthkamath]

Hi,
My problem is that I have 2 cantilever beams (rectangular cross section beams) placed next to each other. I want to do the modal analysis for 15 modes using ansys for these 2 beams and I also want to know how each beam is intercting with each other.CAn anyone tell me how to do this.
you can send the ananlysis guide to my mail ID cvmk@in.com.
Regards,
Vasanth

 

[OldEngr63]

Are these two beams in physical contact with each other (rubbing)? Are these two beams connected with each other (welded, riveted, etc)?

 

[vasanthkamath]

Thanks for your query.

They are lying at a distance from each other(can be anywhere from 1mm to 5mm).

When I do the modal analysis of the 2 beams modeled in ansys,I want to show in ansys how the neighbouring beam effects the other beam by say using contact anaysis, but I do not know how to do it

 

[OldEngr63]

If they are not in contact and not connected, they they should vibrate independently of each other. Now, in reality, there will be some measure of cross coupling through the support structure which could be modeled through the use of FEA, but that involves making an FEA model of the supporting structure.

As long as they do not touch, there is no need for contact elements, and they will not touch if you have a 1 mm separation, assuming small amplitude vibrations.

 

[vasanthkamath]

Thanks again.
But my project involves the vibration contact analysis of 2 or more beams when subjected to different forces (harmonic, step, impulse) , the effect on neighbouring beams . The analysis of the behaviour with respect to time etc.

So I have to do the contact analysis

 

[OldEngr63]

Do you want to do a modal analysis simply for the natural frequencies, or are you intending to do a modal forced response calculation for these various input types you mention?

 

[vasanthkamath]

yes you are correct. I intend to do the second case also

 

[AlephZero]

If the beams contact each other, this is a nonlinear problem so you can't model it with linear procedures like normal modes or forced response. But it would be useful to run those types of analysis first without bothering about the impacts, to check the model behaves as you expect (i.e. the vibrations frequencies, and the amplitude of the reponse to the applied loads, are OK).

You will then need to run a dynamics analysis in the time domain including the impacts. If I undestand what you are modelling, you know whcih geometrical points on the two beams will contact each other (i.e. points at the same position along the length of the two beams). That is simpler to model than the general situation where any part of one object might contact any part of the other object. I don't use Ansys but it probably has some type of "gap" or "contact" element you can use to connect the pairs of mesh (grid) points that will contact each other.

You will also need to think about how to include some damping in the model, otherwise you will probably get a lot of the energy going into unrealstically high frequency modes and possibly the whole structure appearing to go "unstable", or unrealistically small time steps might be needed to run the model.

 

[vasanthkamath]

Thanks Alephzero,
You have understood me correctly.

The reply to your first paragraph is , I have already done that.

For 2nd paragraph, I have modeled 2 rectangular cross section cantilever beams and tried to do the modal analysis by using the contact option as you said, in Ansys.Ihave applied the contact options to the 'nodes of the faces of the beam', facing each other.But my problem is when I am doing the modal analysis in Ansys , the 2 faces are interacting as if there is no nearby beam , that is one beam is crossing the other.
Could you help me

 

[AlephZero]

You can't analyse this by a modal analysis, because the response is not linear. If you apply a sinusiodal force and the beams impact and separate again in each cycle of vibration, the beams are NOT doing simple harmonic motion. The modal analysis will probably just ignore any nonlinear features of your model.

That's why I said you need to do a time marching analiysis.

You can check your contact options are working by doing a nonlinear static analysis. Apply a load to one beam so it deflects enough to contact the other one. I would get that working before trying the dynamics analysis.

 

[OldEngr63]

@ AlephZero: I think you are mistaken. For the several inputs that were mentioned, modal methods can be used by convolution. It is very messy, but it can be made to work; I have done this sort of thing on other systems.

The nonlinearity does not occur until there are collisions between the beams. At that point, yes, the problem does become nonlinear. Before that point, for smaller amplitudes, the problem is linear.

 

[AlephZero]

@ AlephZero: I think you are mistaken. For the several inputs that were mentioned, modal methods can be used by convolution. It is very messy, but it can be made to work; I have done this sort of thing on other systems.

Sure, but you can't do that just by picking "analysis option X" in a program like Ansys. But if you want to explain in detail to the OP how to do it, I'm not going to complain.

To get a "nonlinear steady state" response for this sort of problem I would probably use a version of the harmonic balance method, but again you aren't likely to find that in a general purpose FE program. And you still need to do a time marching analysis to check you haven't made an assumption that excludes some possible solutions, like a periodic response that is asynchronous with the forcing function.

The nonlinearity does not occur until there are collisions between the beams. At that point, yes, the problem does become nonlinear. Before that point, for smaller amplitudes, the problem is linear.

Yup, it's kind of embarrassing if you spend a week setting up a complicated nonlinear model of something without stopping to check whether anything nonlinear will happen

 

[OldEngr63]

Such is life when you are dependent on a canned program such as ANSYS to do your analysis for you. If you write your own code, and if you know what you are doing, then those limitations go away. If these last two conditions are not true, then you are just up a creek. This is not my problem, and I'm not about to try to teach all of that in this context.

 

[vasanthkamath]

Thanks are for your educative comments,

Such is life when you are dependent on a canned program such as ANSYS to do your analysis for you. If you write your own code, and if you know what you are doing, then those limitations go away. If these last two conditions are not true, then you are just up a creek. This is not my problem, and I'm not about to try to teach all of that in this context.

@OldEngr63 , can you enlighten me on a) "writting your own code".Which software can we use to write these codes and how

b)"canned program such as ANSYS".Just for my information and future purposes , which would be the ideal software to do such analysis.

Thanks,
VK

You can check your contact options are working by doing a nonlinear static analysis. Apply a load to one beam so it deflects enough to contact the other one. I would get that working before trying the dynamics analysis.

@ alephzero, I have tried that and it has worked fine.

Basically my project is a nuclear reactor study project, in which the cantilever beams are fuel holders and say arranged in a closely packed circular cluster, inside the liquid coolant in a container.

We have imagined different type of forces(as I had earlier mentioned like harmonic, step, impulse,...etc) being generated inside the coolant, during the operation of the reactor. Any of these forces can act on any of the beams and at any time and any magnitude.
Primerarily a Failure Mode Effect Analysis(FMEA) approach.

What i intend to do is ,to randomly give the different forces as the time progresses and study the behaviour of the beams(basically its position) with respect to time.This I plan to do with Ansys as i am a bit familiar with it.

Kindly suggest how to go ahead and provide some valuable inputs

Thanks,
VK

 

[OldEngr63]

I usually used FORTRAN to write such code, but I don't want to suggest it is the only way to do it. There may be other, better options now. It is preferable to be able to use a language for which you can access a package of eigensolution solvers such as EISPAK rather than having to write your own code for that part as well, although that too is possible.

I don't really understand your second question about ANSYS being a "canned" program. It simply is. To use it, you do not write any analysis code, only input code, so that you are limited to the functionality that is built into it by the owner of the code.