Root Cause Analysis-Refinery Reciprocating Gas Compressor Shaft Coupling Failure
Root Cause Analysis-Refinery Reciprocating Gas Compressor Shaft Coupling Failure
Root Cause Analysis-Refinery Reciprocating Gas Compressor Shaft Coupling Failure
Page 1 of 6
The Problem:
There are two identical 100% size Net Gas Hydrogen reciprocating gas compressors, with two stages each of 750 KW rating directly driven by
360-RPM motors at a Red Sea refinery. They were commissioned in 2006 and nine years later in 2015, the B compressor suffered high vibration
and then shaft coupling failure. In 2016, the author was contacted to study the second coupling failure that occurred on A-compressor. The two
machines operate for approximately half the time so each had accumulated five years of actual operation. The machine coupling guard was
opened and the shim packs were found heavily cracked, with damage to spacer also at bolt hole locations and some sheared bolts were found.
The motor and compressor coupling hubs flanges suffered wear at the coupling bolt locations. Question was why did the two flexible couplings
fail, with one achieving 9 years and the second about 10 years MTBF?
The Report goes beyond a normal RCFA and investigates other defects which have not reached failure condition yet but will reduce reliability.
Reliability Improvements were added to ensure eliminating the coupling bolts looseness issue.
1. Reciprocating gas compressors even when well designed; can generate a crankshaft torsional force up to 2.0 x rated compressor torque as a
result of torsional vibration, gas pulsation, fluctuating crank forces, and other causes. Beyond the value of 2.0 x Rated Torque, the torsional
vibration or induced peak torques will be considered excessive and not acceptable.
2. However if design defects occur, torsional vibration can develop peak torques that are much higher such as 4.0 or 6.0 x normal rated torque.
This will cause major coupling damage and later, motor or compressor crankshaft failures.
3. Liquid slugging entry to reciprocating compressors cylinders can cause sudden over torque events. This will then be transmitted to coupling.
4. Flexible couplings use metallic shim packs, which can experience reduced fatigue strength in seashore humidity environments with salt and
moisture in the air.
Page 2 of 6
5. Flexible couplings must be aligned properly. For such 360 RPM flexible couplings with DBSE= 460 mm, it is acceptable to operate with offset
error up to 0.35 mm and less.
6. Flexible couplings, particularly on reciprocating machines will experience a loosening effect on fasteners due to the normal lateral and
torsional vibration that occurs. Loose bolting leads to fretting and premature failures.
7. Coupling failures due to manufacturing quality issues can arise, but in this case it can be ruled out due to coupling manufacturer reputation.
8. User overloading by having a higher discharge pressure in system or lower suction pressure, will raise the total power across coupling.
9. To determine if the coupling failed by coupling bolt looseness, corrosion fatigue, torsional fatigue, or lateral fatigue from misalignment, it is
necessary to investigate the failed parts.
10. The goal in root cause analysis is to find all possible contributing mechanisms that led to failure. This requires a detailed search for all
stressors. If this is not applied, then root cause analysis is not successful, as failures will return.
1. Design Error by coupling manufacturer or by Compressor manufacturer. Design errors by coupling OEM may be undersized coupling for the
power-torque rating, incorrect material selection, and regarding compressor OEM there may be a high torsional force or vibration.
2. Manufacturing errors in coupling. Such as sharp corners in hub keyways, wrong shim thickness, poor quality bolting, and incorrect machining
of hubs and spacer giving high radial or face runouts.
3. Installation Errors such as wrong grouting, poor shaft alignment, piping strain, excessive gas pulsation in piping system.
4. User Operational errors such as overloading the compressor.
5. User Maintenance errors such as lack of periodic torqueing of coupling bolts, and lack of coupling alignment checks.
1. By utilizing logic, it is clear that the machines have attained 9.0 and 10.0 years’ service before first coupling failed; this strongly indicates that
coupling design was correctly strength-torque rated to give such lifetime.
2. By utilizing logic, it is clear that compressor manufacturer did not make major errors in torsional analysis otherwise, the couplings would
have failed by torsional fatigue in a shorter time period of 24 months or less since installation.
3. By utilizing Logic, it is clear from reported ten years smooth operation of both units that the compressor installation was superior, including
compressor epoxy grouting, Concrete foundation, process piping alignment, and original shaft coupling alignment.
4. By utilizing logic and knowing that the maintenance history for both machines has been almost zero for the first nine years, it is possible to
generally conclude that the machines have not been overloaded by operation personnel or had excessive liquid entry slugging events.
5. However, from logical point of view it is also possible that certain time related events may have occurred recently, such as deterioration of
mechanical components from wear, loosening of parts, deterioration of compressor-motor foundations, changes in piping stress and
alignment, and possible changes in process leading to liquid slugs or gas power overload.
1. In troubleshooting, engineers are best advised to cast a wide net for all information. Lack of knowledge is counterproductive as all failure
mechanisms must be investigated. Multiple failure modes can occur at the same time.
2. Full maintenance history of compressors since original installation was requested to understand all past issues.
3. We requested five year historical vibration trends for both compressor-motor units, process data trends, historical maintenance activities,
preventative maintenance frequency, pictures of all failed parts.
4. Drawings of the machine and coupling were studied.
5. Pictures of high resolution were used to inspect the type of failure.
6. Actual field findings such as coupling as found alignment was recorded.
7. In this specific case, we saw no reason for a lab failure analysis of the broken shims or other coupling parts. Visual analysis sufficient.
Page 3 of 6
The Final Recommendations Based Upon All Data Analyzed
1. I do not see any evidence for high torsional vibration as a root cause failure mode in this coupling, although this was our first suspect.
2. The as found reverse dial coupling alignment showed TIR of about 1.90 mm in Horizontal direction and 1.20 mm vertical. This equals an
actual horizontal misalignment of 0.95 mm, and vertical of 0.60 mm. This is far above allowable at this speed and shaft separation gap.
3. The shaft coupling has been exposed to heavy misalignment in service, leading to bending fatigue failure of its flexible disk pack; notice the
cracking pattern. On reciprocating machinery significant foundation settlement occurs with the vertical forces causing a vibrational tamping
of the foundation soil under the concrete foundation, so it settles downward. This causes high misalignment.
4. There was a strong possibility of corrosion attack on the stainless steel disc pack shims as the refinery is near shore with high humidity and
salinity in air which leads to accelerated corrosion. We concluded that material is suitable based upon long life under high stress environment.
5. The coupling has a design-rating factor of 2.15 according to the coupling drawing data. This is fine.
6. If both the couplings on A, B machines are giving a 9 to 10 year minimum life in this service then they can be considered of acceptable design
since reciprocating service is hard on flexible couplings. If not meeting 10 years then improvements or a new design is recommended.
7. For the Shaft couplings present condition: These are considered heavily damaged and it is not allowed to reuse the hubs or spools so please
immediately order new couplings to be installed as soon as possible. If long lead time, you can fabricate new coupling hubs [four pieces total
to cover both A, B machines]. Instructions: Send a sample of both worn hubs to Fabrication shop, include clear drawing of shaft coupling,
and request new hub made from hardened AISI-4140 or 4340 steel, hardened to Rockwell-C 33 plus/minus 3 points hardness. Insure that
bore dimension of new hubs has shrink fit to shafts, with required 0.004 inch diametral interference fit. Do not use existing dimension, as it
is loose fit, which is not suitable.
8. Coupling Bolts: Only use OEM quality bolts. Do not reuse any existing bolts or nuts and washers. For the future on all reciprocating
compressors flexible couplings [not solid couplings], always clean threads with solvent and add Loctite locking compound to bolt-nut threads.
This eliminates looseness induced fatigue failures. Realign motor to compressor; do not move the compressor, only move motor.
9. Perform NDT MFPT check for cracks on motor and compressor shafts Keyways.
10. Check the motor inboard bearing for high wear and replace before any alignment is applied. Also perform lift check to test for compressor
frame bearing clearance check as it may be worn out also.
11. Please implement a one year PM cycle for all shaft couplings over 300 HP in the plant.
12. New Issue: Did the two compressors discharge pressure increase above the design rated pressure, according to the past three year PI trend?
Check the discharge pressure tag for three years data and compare to compressor data sheet rated pressure. Higher than design pressure,
will overload the couplings by torque overload leading to shorter life.
13. There is very high rod drop on these two machines. This is due to wear of piston rider bands or misalignment of cylinders to frame. Need to
shutdown machine, remove all the cylinder head covers, replace the piston rider bands for all cylinders. Them, realign the cylinders to the
frame. If not having proper skilled labor for this, use Hoerbiger or OEM to perform this job. Also need to check for oval wear of the cylinder
liner. If exceeds limits all liners to be replaced with new.
14. Vibration tags V-11-VI-1359 and V-11-VI-1360 on both compressors are showing higher vibration compared to 2014, its now 0.23 inch/s rms
compared to 0.15 inch/s rms in year 2014. This indicates wear and looseness plus possible foundation bolt or other issues. Possibilities of
cylinder misalignment to frame, looseness in piston-to-piston rod, looseness in Cylinder bolting to distance pieces, loose bolting to frame,
looseness in cylinder hold down bolts. This means it is time for a major overhaul.
Page 4 of 6
Pictures of Failed Compressor Hub Flange and Spacer Element Flange Bolt Holes.
Page 5 of 6
Compressor-Motor Shaft Coupling Root Cause Failure Analysis Conclusion:
1. The coupling failure in both cases was caused by multiple factors most important of which was the high coupling misalignment between
motor and compressor leading to fatigue of flex members. This occurred with time due to normal settlement.
2. The second factor was a lack of any preventative maintenance program for the shaft couplings, They had been neglected, so the bolts-nuts
were getting loose with time and started fretting in the spacer and hub flange holes.
3. No design or manufacturing defects were found in coupling, compressor or motor.
4. The shim pack material was judged suitable for the salt air environment based upon nine and ten year life operation.
5. No process or human operational related defects were found. No compressor installation errors were found.
6. Reliability improvements included adding thread locking compound to coupling fasteners, and a 12-month preventive maintenance check.
7. Results: When new coupling spacer and welded repaired hubs were used, plus very improved alignment, the compressors have operated
well with no failures up to year 2020. Vibration is now smooth.
Page 6 of 6