Gear rattle under full load?

[jeyaselvan]

I am referred to a gear (helical) rattle under both unloaded and loaded (?) condition. The gear box is that of a screw compressor (positive displacement). I can understand rattle under unloaded condition, due to a probable higher dynamic torque, which would aid in the separation of the gear contact. But rattling with full load is a bit misleading. The vibration levels measured in the gear housing shows higher amplitudes at gear meshing frequency. Is there any other possibilites for gears to rattle under load?

Kindly for your suggestions

 

[SomptingGuy]

Rattle of unsychronised gears in a constant mesh automotive manual transmission is very common. Most people wouldn't even recognise it as rattle - it is often a kind of "whooshing" sound. It is caused by through-the-cycle speed fluctuations, sometimes compounded by clutch damper resonances.

Loaded gear rattle is more common on the front-end-drive systems on bigger diesel engines. They tend to use idler gears to take power from the crank to the FIE & camshaft. Due to the high torque pulses created by the FIE and cam, you often see mesh separation. This sounds nasty.

Rattle is best analysed in the time (or cycle) domain. As it is impulsive, frequency domain analysis can be confusing.

- Steve

LinkedIn

 

[JJPellin]

Before I would assume that the rattle is a normal characteristic of your gearbox, I would check for all the usual suspects.

You said helical. I take that to mean single helical. We are just finishing the overhaul of a large single helical gear increaser that drives a multi-stage centrifugal compressor. It was howling and had high acceleration. The vibration was not at gear-mesh but was an unusual frequency (17 times run speed) that did not correspond to anything we could identify. The problem ended up being lack of oil to the gear mesh (plugged spray bar) and a badly racked (twisted) gear-case caused by a base that was not flat. While we were apart, we range the bull-gear and found it has a natural frequency at about 17 times run speed.

I suggest you verify tooth contact, backlash and lubrication. I am not convinced that tooth contact reveals all problems. I prefer to have coordinate mapping of the bearing bores to verify that they are round, concentric and parallel. A master level on the split line and at the bottoms of the bearing bores can reveal a problem that won’t be seen with tooth contact.

I would verify bearing clearance and configuration. Are the bearings correct with the correct dams at the correct angle? Is the clearance in spec? If it is single helical, the coupling spacing is also a major factor. Are the couplings pre-stretched properly? Depending on the types of couplings, they could be imposing an axial force on the gears or even causing an axial shuttling effect if they are running with excess axial stretch or crush.


Johnny Pellin

 

[Tmoose]

Among the symptoms attributed to torsional resonances are unloading of keys, gears, and splines. Impact wear and noise can be involved.

 

[jeyaselvan]

Thanks Stteva & Tmoose for the observations. Thanks Johnny for a deeper insight.

The gear is of single helical type. The problem is observed in majority of the production lot and not to specific machine and hence I am approaching more from a design view point. The torsional pulsation in the application is primarily due to the compressor, which is lot lesser compared to that in an engine. The drive speed is 1460rpm.
Attached the measured vibration data (for two revolutions of the drive train). As is evident, the response is predominantly at the gear mesh frequency. The rattle is clearly audible.

My drive train analysis shows that I was close to drive train torsional resonance at gear mesh frequency. Is there a quicker / dirtier (and accurate??) way to verify the predicted torsional resonance?

Regards
Jeyaselvan

 

[rob768]

quick and dirty: see if it appears when approaching operating speed. if it is a resonance, it will not be present at lower speeds.
accurate: not much but indicative.

 

[Strong]

Changing the type of coupling would change torsional stiffness, and may be easy and cheap depending upon size of the machine. What is the horsepower? I would not characterize gear tooth vibrations as a "rattle", unless there is some other structural component that is loose and is actually making a rattling sound in response to the gear vibrations. An impact test of the gearbox structure and torsional vibration measurements would be helpful to solve the problem.

Walt

 

[JJPellin]

You did not mention what the driver was or if this was an increaser or reducer. It sounds like you most likely have a motor driver and this is most likely a speed reducer. In any case, I like Strong's suggestion to change the torsional characteristics of the coupling. A coupling with lower torsional stiffness should reduce the torsional resonance away from gear mesh frequency. Using an elastomeric coupling would add torsional damping For diesel engine driven pumps, we often use rubber element couplings for this reason. For large synchronous motors, we use rubber block couplings to dampen the torsional excitation during start-up. One coupling could be selected which would accomplish both objectives; drop the torsional resonance frequency and increase the damping.

Johnny Pellin

 

[MikeyP]

jeyaselvan,

You and I have talked about these things before.

Do you know what the static torque through the gear mesh is at max load?

M

--
Dr Michael F Platten

 

[Tmoose]

Was the vibration measured radially? On a bearing housing, or mid panel?

 

[jeyaselvan]

Thanks to all for the responses.

Tmoose : The vibration was measured in the axial direction (helical gears)in the gear housing.

Johnny : The problem is with a drive train resonance as discussed below and hence any changes in coupling could not make significant changes in shifting the same.

Dr. Michael - Good to meet in you in person and hope you are back home after your visit here. I remember during the discussion that rattling as primarly (or commonly) a problem with lightly loaded gears. I remember you referring this as a system (torsional fluctuations) related issue than a gear related issue, which I am trying to attempt on at the moment.

The machine is a motor driven(small)screw compressor with motor rating of 18kW running at 1460rpm and hence with a mean torque at 118Nm at full load. During unload condition, the torque is typically close to 40Nm. The gearbox steps up the speed to 4614rpm (ratio of 3.16) and drive the rotor.

I was trying to predict the torsional natural frequencies of the drive train system. Have observed a torsional mode in the drive train closer to the gear meshing excitation (probably this may get excited due to transmission error). I was trying to verify my predictions, since at the moment I am short with instrumentation for torsional vibration measurements. It was interesting to see in reality that the same compressor with a much larger motor (55kW), meaning different motor inertia also results in gear impact/rattle. This is quite inline with my prediction that shows the concerned mode shape (fourth in the present case)is not sensitive to changes in motor inertia, but is does change the coupling mode and the second to a little extent. Have attached the drive train and the torsional analysis data.

Planning to have a loaded coast down tests and will try to fiqure out the excitation order of gear (79 in this case). I understand with gear whine this could work a lot better, but with rattle,I am not sure. Before proposing a modification in the drive train (or ?), I am willing to verify the predictions. I am equipped quite well with linear vibration measurements, but not for torsionals.

Kindly for your suggestions to verfy the predictions.

Thanks & Regards
Jeyaselvan

 

[btrueblood]

That looks/sounds like a v-twin motor driving a roots blower. I worked with a designer who had a similar problem with that configuration, and we traced some of the troubles to a) the motor was firing with "Harley timing", i.e. both cylinders fired on the same revolution, b) the motor was an uprated (bored) version of a smaller motor, and no change was made to the flywheel inertia.

The net result was a factor of ~2 greater peak torque for this beast than the next smaller engine in the manufacturer's series (which we were familiar with). We ended up doing much as JJPellin suggested, which was to fit an elastomeric-element coupling (which was a chore, since the high torque kept shredding the smaller sized coupler that the handbooks were suggesting based on motor power), which helped somewhat. What really helped was increasing the rotational inertia of the coupling, which was a side effect of increasing the size to keep the rubber parts from shredding in less than 100 hours.