Interpreting Accelerometer Data (Mass Participation)

[brysonc]

First, a little intro -- I design metal brackets that bolt to trucks that hold various types of equipment, all furnished by a third party. We perform structural simulation (FEA) on all of our brackets, but we don't have much information on the equipment that our brackets hold. We currently treat all of this equipment as rigid, but this is a poor assumption for some components (primarily antennas).

I am currently putting together a data acquisition system in order to determine the dominant natural frequencies and corresponding mass participation factors so that I can have a more accurate model to use in our FEA. The problem that I have run into is that I can easily extract the dominant natural frequencies and reproduce those as simple spring/mass systems, but I am not confident in my method of determining the modal participation of each mode.

Can this be determined by comparing the magnitudes of vibration at the first few frequencies in each direction? Or, is there some kind of normalization that needs to occur before doing this?

Any help is greatly appreciated. Thanks,

Bryson

 

[brysonc]

Thinking more about it, I think it may not be possible to determine modal participation with a single accelerometer. I would think that the magnitude of each frequency would depend greatly on location of the accelerometer with relation to that particular mode shape. A less dominant mode may appear to participate less if the accelerometer is placed in a location that (for that particular mode shape) does not accelerate much.

Any ideas? Perhaps placing the accelerometer as close to the mounting location as possible would give the most beneficial data for my studies...

 

[IRstuff]

As in the case of testing for vibration effects, one never only goes in with a single accel, or even one 3-axis accel. The key point to remember, I think, is that component resonances are described as "decoupling" from the main structure, so any part of the system that might oscillate on its own will need to be monitored.

The accel that might be mounted closest to the base or on the base is usually the control accel, since that's the one that most likely reflects the stimulus input.

TTFN

FAQ731-376

 

[brysonc]

The input will not be applied by a shaker table (though I wish it would be). Also, we plan on running two 2-axis accelerometers to capture all three directions. There are two ways that we are considering running the tests:

One possibility is to suspend the antenna (possibly using monofilament line) and to tap it with a hammer. In order to get enough data to determine the mass participation, I guess this would have to be repeated with the accelerometers in several different locations.

The other possibility is to bolt the antenna to a "ground" that is significantly more massive and rigid than the antenna, then applying the excitation either with a hammer or by simply moving the antenna out of place then releasing it.



Currently, we input PSD curves and time-history shock loading from MIL-STD-810 to the fixtures of our bracket, and we think that we may have some inaccuracies due to our assumption that the antennae behave as rigid bodies with simply an equivalent mass and CG (and an approximate MOI).

 

[IRstuff]

So, who's doing the analysis of the actual antennas? Don't they run simulations of their equipment mounted with your brackets?

To some degree, it looks like you're ostensibly duplicating someone else's work. Given that the antenna supplier has to make sure his antenna doesn't fall apart when mounted with your brackets on some vehicle, they have to verify that there's no decoupling of their equipment, and it's incumbent on them to design their stuff to survive the environment.

I can't see how you'd even begin to do any sort of testing without some representative equipment to mount to the brackets.


TTFN

FAQ731-376

 

[brysonc]

The antennas are more or less universal -- each truck gets several antennas, but each antenna is also used on a myriad of different trucks, and often times in different locations on the same truck.

What I am trying to do is validate our brackets, just to ensure that they will not yield during normal operation when mounted on the truck with its specific antenna mounted to it. I'm not interested in the stresses of the antenna.

 

[IRstuff]

I understand that, but if that were it, you'd simply have a lumped mass model and you'd be done. Resonances, however, interact with the rest of the system.

A simpler answer might simply be to design with a safety factor

TTFN

FAQ731-376

 

[mechengdude]

brysonc -
I would tend to agree with IR's advice.

Some additional comments, if you are analyzing using PSD data from MIL-STD-810 I assume your final answer or determination is based on a probablity of the structure surviving (i.e.) using the 1, 2, and 3 sigma stress results. This only provides a probablity of sucess.

Long story short, don't miss the forest for the trees. Most wheeled and tracked vehicles with the exception of HMMWVs have room for a little extra payload so make your bracket a little heavier.

 

[hacksaw]

with cantilever loads commonly produced by antennae the bracket is seeing bending loads.

The key variables are the stiffness of the bracket support, the stiffness of the bracket itself. Having those to quantities, the cantilever response is pretty well defined. Requires solution of the beam equation for the elastic support.

Having solved for the dynamic the loads imposed by the antenna and applied to your bracket, the loads can be incorporated into your fea model.

Measurementwise you want to look at the transversal displacements at the juncture of the antenna and the bracket, and the rotational movement of the bracket support.

Free body tests of the antenna are useless, unless you are trying to characterize its beam parameters for a free-free condition.