Vibration of a recip. air compressor 2

[mdossaji]

I have a 2cy recip. air compressor directly coupled to a 1800 RPM/60 Hz motor. The air compressor is mounted on a frame which is suspended on an underfloor of a rail car. When the air compressor operates there is a significant vibration felt on the car floor.

The compressor is 350 lbs suspended on a frame with three lord isolators (283P-185) rated to 185 lbs max. and 740 lb/in spring rate. It seems that the disturbed frequency by the air compressor at 30 Hz is causing some of my other components mounted on the compressor frame to vibrate under its natural frequency. Can anyone suggest a better isolator or idea as to how I can improve the resonance of my compressor frame? Shall I reduce the stiffness of the isolator in order to reduce the natural frequency of the isolated compressor?

 

[GregLocock]

In an ideal world you've mounted your processor at 7.5 Hz, so you'd feel aggrieved that 30 Hz is getting through.

Life isn't that simple, since you do not have an SDOF system, you actually have 6 rigid body modes to worry about. It is anywhere between possible and likely that one of those is near 30 Hz, thereby ruining your isolation. Also your mounts may be stiffer than 740 in the off-load directions.

What you really need to do is to optimise the location of the mounts and their stiffness. You either need to know a great deal baout the system, or some test data.







Cheers

Greg Locock


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

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

 

[FeX32]

Could you attach a vibration absorber tuned to 30Hz?


Fe

 

[GregLocock]

They tend not to work especially well unless the path is well defined, I suspect in this case it'll be coming through all 3 mounts. - it's a very similar problem to mounting a 4 cylinder engine.

Cheers

Greg Locock


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

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

 

[rob768]

Would using rubber mounts be an option? You'd get some damping from them and inprove overall levels (along with an option to shift resonance frequencies). Also, 30 Hz in an 1800 rpm installation suggest unbalance. Improving balancing conditions may reduce levels as well.

 

[GregLocock]

Guessing he's got either 2IL or 2 boxer configuration, he has got big first order imbalance moments.



Cheers

Greg Locock


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

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

 

[Hatch]

In the past where there are multiple degrees of freedom, I have used wire rope isolators. I used it in reverse to protect a computer system on a locmotive engine and also to isolate a fan system which was shaking a roof.

See:

http://www.enidine.com/Industrial/WireRopeIsolators.html

Just make sure that you select the proper size isolator (low natural frequency).

 

[mdossaji]

Hatch,

My calculated natural frequency based on the disturbing frequency of 25-35 Hz is 8.3 - 11.7 Hz. so are you saying that the rope isolators I select should provide a natural frequency of less than 8.3 Hz?

See attached calculations I conducted for the HERM series mounts from Enidine. Based on my number, I selected HR12-400-BHMY.

 

[jeyaselvan]

Your calculations are done only for the vertical direction and has assumed uniform mass distribution at each mounting point, which is not ver common. Try to have a larger extreme deflection ratio of the system so that it aids decoupling. As suggested by Greg, atleast you need to calcuate the rigid body modes for which you would require the assembly's centre of mass, inertia & mount stiffness & dynamic/static stiffness ratio at the orthogonal directions.

I had a similar problem earlier with my rolling rigid body mode at 39Hz close to a second order excitation of a 4S,4cyl engine while idling at 1200rpm. This was solved merely by orientation of the mount so that the stiffness changes in the lateral & transverse direction help to shift the rolling mode down to 25Hz. With recip compressors, you always need to ensure your rolling mode is away from your low torsional orders. (this would vary based on the torque-crake angle behaviour of your configuration).

Jeyaselvan

 

[Tmoose]

Do you have detailed vibration measurements of the floor attachment and its surroundings, vs the frame, vs the compressor?

Is the "direct coupled motor" a C face motor, so the motor and compressor are both "isolated"? or, Is the compressor on the lord mounts, and the motor hard mounted?
Passing shaft Horse power to a flexibly mounted component is tricky.

What is the piping like near the compressor?

 

[mdossaji]

Please see attached view of my compressor. The motor is directly coupled to the air compressor crankshaft. Both the motor and the compressor are on LORD isolators. Based on suggestion from Jeyaselvan I assume that my isolators are in vertical mode and when the pistons are in opposed state at 180 degree then there could be a high rolling mode.

 

[Tmoose]

Lacking instrumentation ( a situation I consider much like the dream of being naked in front of the class ) I find that touching a vibrating component with a finger nail rather than my finger tip is more "directional" . Might help ID relative strength of vertical/horizontal/axial/tangential vibration direction.

If there's no budget for vibration analyst or vibration measurement gear capable of discerning frequencies it may be time to glue weights on a small loudspeaker and feed the output into the soundcard's microphone port for processing with a freeware FFT program. Could also be used for "bump test" to ID actual mounted frequencies

 

[FeX32]

Interesting.
"glue weights on a small loudspeaker and feed the output into the soundcard's microphone port"
I don't see how this can attain the vibratory frequencies in the structure. What if it makes almost no noise when it resonates?


Fe

 

[btrueblood]

Fex,

By gluing weights on the speaker's cone, you can create an accelerometer sensitive (tuned) to any range of frequencies, from the normal high tweet down to nearly seismometer levels. It is left as an excersize to the reader on how much weight to add, and what the calibration of such a device would entail...

 

[FeX32]

Oh. That makes sense. Thanks for the elaboration.


Fe

 

[GregLocock]

Agree with the screwdriver.

DO you know what rates the mounts are in the radial as ooposed to axial directions?

Cheers

Greg Locock


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

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

 

[mdossaji]

Attached are the result from the measurements using an accelerometer. The top graph is using the original isolators (220 lbs on the motor and 185 lb on the compressor). The new isolators (155 lbs on all mounts) results is thegraph called "softer" isolator. also attached is the curves for the 155 lbs isolator for radial and axial loads.

 

[GregLocock]

So the isolators are stiffer in the radial direction (this is typical). Your forcing vibration is horizontal. Therefore you are more likely to have success if you align the axial direction of the mount with the cylinder axis.

You only included rate curves for the isolators, no accelerometer plots.

Cheers

Greg Locock


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

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

 

[mdossaji]

Greg,

Attached are the accelerometer plots.

Murtaza

 

[Tmoose]

Are the pictures of a frame mounted on a test rig?
Were the "as built" measurements made while operating on the same rig?

With the drastic change in frequency content ( many more spikes at higher frequencies) my hunch is something other than just the isolator stiffness changed between tests.

I'd make some measurements while striking the (power off) compressor at the CG with a semi soft faced hammer. If my analyzer has a peak hold function I watch to see how the spectrum fills in during the striking sequence. Also I like to capture a time trace to see if the response to each strike resembles a decaying sinusoid.