Glued Versus Solid Model in Modal Analysis

Will the glued model work?

Poll closed Jul 31, 2014.
  1. Yes, Gluing will be fine

    0 vote(s)
    0.0%
  2. No, Go with the solid piece

    0 vote(s)
    0.0%
  1. I am performing Modal analysis on a fairly simple structure (a square aluminum rod with some pillars attached to the top of it) and need to get some basic modal analysis results. Making a solid aluminum piece would be rather expensive and time consuming to design and then mill out, so the alternative would be to simply glue the pillars on to the rod. We just need to be able to see the basic vibration modes of this system (only up to about 8000 Hz).

    one unit cell of this rod is a cube of dimensions (in cm) 6x6x6 with a pillar of 1.5x1.5x3 on top, this structure would be repeated 15 times (based on current design specifications)

    How adversely would a glued structure affect the resultant FRF graph compared to the real solid piece FRF graph. Would it be worth the time and money to produce the solid piece? or will the glued version give us usable results?

    Note: We are only doing this for research purposes, the glue we select will be simply for its vibrational properties. We are NOT looking for strength variables of any kind. The point in this experiment is to observe the bending and torsional modes of this system as a function of system response amplitude (in Db) versus frequency (in Hz)
     
    Last edited: Jul 21, 2014
  2. jcsd
  3. jim hardy

    jim hardy 4,666
    Science Advisor
    Gold Member
    2014 Award

    I've used Loctite 324 structural adhesive. Its strength is amazing , you separate glued metal pieces with a cold chisel and hammer.

    Here's its datasheet -
    http://krayden.com/tds/henk_loctite_324_tds.pdf
    if a few thousand psi tensile and shear strength will hold your pieces together it just might work.

    My application involved lower frequency than yours, just a few hundred hz, attaching ~ 1 lb vibration sensors to rotating machinery. We used their 707 activator for fast cure.
     
  4. AlephZero

    AlephZero 7,300
    Science Advisor
    Homework Helper

    Joints of any sort are usually a pain in vibration analysis, because they introduce an unknown amount of extra flexibility and damping compared with a solid component.

    If the extra flexibility and damping are not consistent between nominally identical joints, that can get very messy to measure and model. Joints are usually places with high stress levels relative to the rest of the structure, so they have a disproportionate effect on the dynamic response, regardless of whether they are structurally strong enough.

    I would suggest you first assemble some of the "unit cells" and measure their individual FRFs. If you get consistent results, you can then consider tweaking your model of the structure to match them. If there is a lot of scatter, you may need to reconsider how to build the complete structure.
     
  5. AlephZero, These pillars are not going to be supporting anything, in fact in this line of research that we are doing, the goal is to get them to resonate at a frequency and phase that cancels out a specific wave mode of the bar as a whole to produce a band gap in the FRF. The application being that if you know what frequency your system is resonating at, and you can calibrate this system to cancel out most of that vibration, you can effectively eliminate vibrations at critical frequencies in your system (such as cruising frequencies of aircraft so they are MUCH quieter and smoother). So, all I am asking here is if the glued model should give us an FRF graph that would be acceptable as baseline data, Though I do really appreciate your thorough response.
     
  6. AlephZero

    AlephZero 7,300
    Science Advisor
    Homework Helper

    Whether or not they are supporting something isn't relevant. Presumably you want to be able to "tune" the resonance of the device somehow, so unless you are going to do that by trial and error, you need a model to predict the behaviour. The flexibility and added damping in the joints wlll be some of the big "unknowns" in that.

    As a simple example, take a metal bar and fix it in a clamp so it vibrates like a cantilever. If the clamp is clean, dry, and tight you will get some well defined resonances with low damping levels.

    Now repeat the experiment after putting a few drops of oil on the surfaces that are clamped together. The Q factor of the system will probably reduce by a factor of 10 or even 100 times.

    A thin film of your adhesive inside the joint might have the same effect as the oil.
     
    1 person likes this.
  7. That's kinda what we are afraid of here. Thank you for sharing your opinion, I am not sure of what we will do, but if we do the glued model, I will be sure to post the results of how it worked.

    In previous experiments when we just stayed in 1D (composite rods), the glue had little affect (according to my mentor). So if we add another dimension (and many more degrees of freedom), I can kinda see how it could cause some issues...
     
  8. jim hardy

    jim hardy 4,666
    Science Advisor
    Gold Member
    2014 Award

    The modulus (of elasticity?) for that adhesive is 89,000psi per data sheet
    while aluminum is more like ten million

    so the glue joint will be a 'soft spot' in your structure
    and
    unless it's got enough area that the stress in it is very low, yet thin enough it can't deform noticeably,
    Aleph is right...

    the dimensions you gave seemed to me suggesting the 'columns' are more just small masses added to your rod, not tuned stubs that are going to flex or gyrate of themselves
    so i think a thin glue joint could stand a chance
    but it would probably be wiser to bolt them on with considerable clamping force. (I'm a big fan of socket head cap screws and lubricated threads.)

    that would be an interesting demonstration to run, glued vs solid
    even if only on a finite element computer model

    you're out of my field for i'm no vibration guy - so i shoulda kept quiet ,,,,
    .... but that is one impressive adhesive.

    old jim
     
    1 person likes this.
  9. So unfortunately it looks like we are going to be "encouraged" by our overseeing professor to do the solid piece, so I will not be able to test this for you guys... But it is an interesting thought for how it would work, and should I get the time and money to test this, I might just have to try it (even if it is just with a beam cut in half and glued back together)

    Thank you all for the awesome replies!
     
  10. AlephZero

    AlephZero 7,300
    Science Advisor
    Homework Helper

    There is a lot of literature on damping effects in structural joints. It's an active research topic in its own right. I think your overseeing prof is wise to "encourage" you to avoid it if the objective of your project is something different.
     
  11. jim hardy

    jim hardy 4,666
    Science Advisor
    Gold Member
    2014 Award

    now that joint is where i'd be afraid of glue. . Your beam is 6X6 cm and quite long
    while the columns are only 1.5X1.5 and short,
    so the beam is massive compared to your columns.

    I'm only an amateur at things mechanical

    but i do know bending stress is in proportion to moment
    which for your short columns/pillars (3 cm) is small
    but for your long (15 X 6 = 90 cm) beam, substantial ..

    Perhaps a genuine mechanical type will chime in ?
    Aleph ? is above correct?

    old jim
     
    Last edited: Jul 23, 2014
  12. AlephZero

    AlephZero 7,300
    Science Advisor
    Homework Helper

    From the dimensions given the mass of the complete structure will be about 10 kg. It doesn't seem unreasonable that glued joints would support that weight.

    We don't know anything about how the structure is supported, or what orientation it is in (e.g. is the long axis vertical or horizontal?) so estimating any stresses is guesswork without more information.

    Just a thought for the OP - if you don't want to machine parts from solid, could you assemble them with shrink fits instead of adhesive? If you machine them accurately and assemble them with a temperature differential of about 100C, that should give you nice solid joints.
     
    1 person likes this.
  13. jim hardy

    jim hardy 4,666
    Science Advisor
    Gold Member
    2014 Award

    Thanks, A..Z...

    my limited vibration experience is at lower frequencies, on 60hz rotating machinery .

    old jim
     
  14. AlephZero, the bar will be horizontal with the pillars up (along the top side in a row along the bar), and will be supported by thin wire to create a free-free system for our tests. Unfortunately we dont have the equipment to do the shrink fitting because our shop is being moved at the moment so all we have access to is the mill, but that is a really good idea.

    The bar will weigh in at 8.75 kg after milling, so our usual fishing line for supporting it will not work in this case. I was thinking that we could just use some high strength fishing line, but with the sharp corners on the beam, I am afraid that it will shear right through it, so we might have to go with some thin metal cable.
     
  15. AlephZero

    AlephZero 7,300
    Science Advisor
    Homework Helper

    OK, so you are want to do a vibration test without any constraints.

    You don't want a stiff metal wire or similar, you want a soft spring, for example bungee cord (you can get small cords to about 2mm diameter). You want the vibration modes of the whole structure moving as a rigid body on the suspension to be as low as possible, compared with the frequencies you are interested in.

    Of course you might want to use a wire loop round the beam attached to the end of the bungee, rather than trying to wrap the bungee cords directly around the beam.
     
    Last edited: Jul 24, 2014
  16. Yes sir that is the plan. Right now is just testing a theory. If we can prove said theory, then we will move on to clamped edges, 2D planes of this model, and beyond. That bungee cord idea sounds good, ill have to look into that...

    I have seen videos where people perform these tests on egg-carton foam. I feel like that would cause a lot of dampening and wouldn't be desirable testing conditions. Any thoughts on that?
     
Know someone interested in this topic? Share a link to this question via email, Google+, Twitter, or Facebook

Have something to add?

0
Draft saved Draft deleted