Variation of amplification factor - Vibration shaker testing

[jeyaselvan]

We were trying to do a vibration test of compressor assembly in a electrodynamic shaker at an external lab. Initially, the test lab personnel did a sine sweep from 5 to 1000Hz at an input acceleration of 1g and a sweep rate of 1 oct/min. Thereafter we need to do a resonance dwelling at the identified natural frequencies (first two).

On sweep testing, we identified the first two natural frequencies (less than 100Hz in this particular test direction) and the amplification factors. When we subsequently did a dwelling for an input acceleration of 4g at the resonant frequency, the amplification factors are all off (pretty low in a particular direction which we were testing). The positions of control accelreometer in the shaker table and the response acclr on the compressor assembly are the same.The fixture (in this case was just a solid plate) resonant frequencies are established to be much more than 1000Hz.

Kindly for your suggestions on possible points I need to check at the laboratory.

Regards

 

[Twoballcane]

It sounds like you may be experiencing some friction damping (some slipping btw the bolt and chassis or fixture). Did they re-torque the bolts after each run? You may have to check each degree of freedom to see where you are losing the input load. Try getting data before and after each fastening point and compare this to the data from the control transducer. Work your way up from the shaker head to the concerned part and see if you notice a drop in g load. At least you know where you're dampening.




Tobalcane
"If you avoid failure, you also avoid success."
“Luck is where preparation meets opportunity”

 

[IRstuff]

> what frequency were the resonances?
> what form factor is the UUT, i.e., is it tall and skinny?

TTFN
faq731-376

 

[jeyaselvan]

Thanks for your kind replies.

Tobalcane : In fact, in the initial trial, we had some slipping between the bolt and chasis. This was then torqued and was ensured through wrench for later tests.

TTFN : The frequencies ar 18.9Hz and 59Hz. The first freuency is very lighly damped and hence I can understand as to why amplification factor is more sensitive. But even for the second nat frequency, the amp factor is quite varying! Whether the form factor has role in this? In this case, it is pretty much like a cantilevered load. I have attached the schematic of the setup and the resposne data.

Kindly for your comments

Regards
Jeyaselvan

 

[IRstuff]

Is this a standard slip table?

Were any of these resonances predicted beforehand?

I had thought there might be a problem arising from the slip table itself, given the cantilever, but your structure appears to be ill-suited for a vibration environment.

TTFN
faq731-376

 

[BrianE22]

The two mounting areas look very small. Is it possible that you are losing preload at those joints when you vibrate at the higher (4 g) level?

 

[GregLocock]

I'm very glad that your test has identified a very poor piece of design. It seems astonishing that you didn't catch it before. My guess is that the real installation is better supported, if not, then it needs to be.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[jeyaselvan]

TTFN: Yes, the shaker (LDS) has a standard slip table. But what I have shown is the excitation in the vertical direction, where the armature is comnected to an expander and then the test fixture and UUT.The lab has informed the nat freq of expander with armature to be at 1800Hz.

BrianE22: Yes, that was a real concern. Unfortunately, that is how it is installed in the application.

Greg : You are right.In the real installation, there is another stage compressor(first stage), which is not shown,(only mounting holes are shown). And that is supported. But this stage (second stage, which is of concern) still remains a cantilever.

The drive end is coupled to the engine flywheel housing. This is a two stage engine driven compressor and is typically skid mounted and is used in off road conditions. Our concern is that while the vehicle is moving in offroad, the assembly probably gets into transient excitations (no vibration mounts at the moment) from the road terrain and excites the inherent cantilever type of installation. We are trying to see whether we can prove a failure mode of the bearing. The bearing failures corresponds to those of the bearings at the free end of the cantilever.The bearing failures are related to static failure (as analyzed by bearing manfr), hence not on running condition (operational excitations).

Regards

 

[GregLocock]

Long experience would suggest that you cannot successfully mount bits to engines using two bolts in bending. The same applies to IUT on a vibration table.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[vibefreq]

Many systems are not linear as amplitude increases. As amplitude increases there will be some reduction in response frequency. usually, once back to a low er amplitude the response frequency will return to it's previous higher frequency.
While frequency shift is not your problem in this case, the same nonlinear effects with different input amplitudes my be behind your problem

I like Twoballcane's explanation better though. I once ran into a chassis slipping on its vibration fixture causing the response amplitudes to just drop away after a certain amplitude was reached.

Have you found a cause yet?

 

[amanuensis]

The algorithm for controling the resonance frequency use the phase value instead of the amplitude value.
It would be interesting to look at the phase spectrum.