Coupling Problem

[TPL]

I have a flexible coupling installed between a steam turbine and a centrifugal compressor, which was found to have cracks in the flexible elements when inspected.

The inspection was carried out after the compressor had been inadvertently surged quite heavily during an exercise to redefine the surge controller curves.

We have enough data to establish that the steady state torque being transmitted through the coupling before and after the surge had occurred was approx 60,000in-lb. The coupling normal torque capacity is 300,000 in-lb and its peak torque capacity is 800,000 in-lb.

At first glance, it look as if the coupling is well within its rated capacity, but I am now wondering whether or not the steady state data is really representative of what happens during surge.

Surge is essentially flow reversal (typically occurring at 30 to 120 times per second). Based on this, it might be possible to infer that one complete flow reversal implies a complete torque reversal, giving a peak to peak torque change of 120,000 in-lb, which again appears to be within the rating of the coupling.

However, after a bit of thought (the dangerous bit), I began to think about an analogy of a car being brought to rest from a steady state speed of 50 mph in one of two ways: firstly by hitting a brick wall and secondly by applying the brakes and slowing down. In both cases, the car comes to rest but the end result is different, determined mainly by the rate of change of velocity (i.e acceleration).

Applying this line of thought to the coupling damage, I am wondering if the rate of change of torque during flow reversals is the key to the damage seen.

This is about as far as I can go without assistance - does anyone have any ideas as to how, if I know the steady state torque, I can estimate the loading on the coupling with 30 flow reversals/sec and again at 120 reversals/sec)?

I don't know how many reversals took place: I am looking to see if there is any possiblity that the coupling was overstressed when it was surged.

If torque can be described as rate of change of angular momentum, does anyone know if there is a term that describes rate of change of torque

All help gratefully received - thanks in advance

 

[JJPellin]

You didn't specify what type of coupling had failed. The first thing I would do would be a full analysis of the failed coupling. The location and orientation of the cracks should indicate if it failed from overload or fatigue. If it is fatigue, then the torque reversal during surge would not be the likely cause. Some coupling failures can be specifically tied to misalignment. I assume that you examined the coupling only because you had just done the surge testing. But the damage that you found could have been there from before the test. I need to pull out some books a read up on the topic, but I did not think that surge could occur at the high frequencies that you describe. I always thought of surge as a much lower frequency phenomenon.

 

[TPL]

It's a disc coupling and it is visually inspected at regular intervals simply by removing the coupling cover.

Following surge, we picked up an increase in vibration at the compressor Drive End which triggered the inspection where we found the discs to be arranged in an unusual pattern: we replaced the coupling and stripped down the one that had been installed to find the cracks.

The previous inspection was a routine one, carried out 3 months back and there was nothing unusual at that time.

Our materials specialist has looked at the crack pattern and feels that it is not fatigue induced.

We have checked the alignment and whilst it could be better, it is well within that specified by the coupling OEM.

I got the figures for 30-120 flow reversals from Forsthoffers Rotating Equipment Book Vol 3. The actual rate or number of flow reversals is never going to be known: what I am looking for is any way putting numbers together to indicate if surge might have caused some material property of the coupling to have been exceeded.

We redefine the surge control line each time we rebundle the compressor: we will need to revisit our procedures if we find that surge might damage the coupling.

 

[Barry1961]

I am not familiar with the dynamics of your system but just in general..

A torque reversal can generate an extremely high peak torque depending on the inertia of your system. The primary driver of your system probably cannot generate enough torque to harm the coupling.

I would guess that you are changing the velocity of some mass at a high rate. My totally uneducated guess would be that you are trying to stop a high velocity column of air suddenly and it is acting like a hammer. It is easy to calculate the forces generated by solid objects in motion changing velocity until the point where the objects begin to yield or deform. Your application sounds like it would be difficult to calculate.

If you have a fairly good idea of what mass in your system has enough kinetic energy to cause the damage to the coupling you may be better off trying to cushion the shock or add a relief vent.

Here is a link to an Excell workbook that has linear kinetic and flywheel calculators but I doubt it will be much help with gas.

http://www.skadrives.com/engineering.htm

Barry1961

 

[crjones]

A coupling is somethnig of a mechanical fuse if you will. It is the weakest link in the drive system and should fail to protect the connected equipment when usually the driven element stalls or seizes. So significant load reversals will push the coupling to failure.