Machinery failure/ damage survey

Author; Brett Weintz. Updated 15 June 2021.

How to conduct practical machinery failure analysis?

Numerous repeated critical reviews of evidence and intermediate conclusions are a crucial element of any successful failure/ damage analysis survey investigation. Thorough and precise identification the causes of damage help to avoid further incidents and costs. Brabon Engineering Services can undertake root cause analysis investigations for damage to all types of mechanical rotating machinery.

---

You can find the following Brabon Engineering Services site pages and case studies regarding machinery failure/ damage analysis:

• Machinery failure/ damage investigation discussion (this page).
• Diesel engine failure discussion page.
• Metal fatigue fracture failure discussion page.

• A machinery failure investigation case study 1.
• A propulsion machinery failure investigation case study 2.
• A metal fatigue fracture failure case study.
• A sterntube bearing failure case study.

Benefits of machinery failure/ damage investigation

Safety, recurring costs and commercial disputes are the main imperatives for investigating incidents of machinery failure/ damage. Preventing further incidents is vital to avoid compounding the total costs associated with the failure. An investigation may identify potential single point sources of machinery damage. Pinpointing the (root) causes of machine problems/ failures can inform changes to procedures and operations.

A thorough investigation and analysis with supporting evidence is the basis for any strong case in a commercial dispute, guarantee claim or hull and machinery insurance claim. Superficial consideration of an incident may lead to a claim being challenged or ineffective corrective actions. Investigations may be compromised by insufficient information or poor analysis. Whereas investigations pursued to a satisfactory level of certainty, e.g. beyond reasonable doubt, also provide a competent reference for the future and an accurate measure of failure trends.

• 

• 

• 

• 

Not all failures require detailed investigation. However, where there are significant associated repair and/ or off-hire costs, then identifying the cause(s) is warranted. Successful investigation of mechanical machinery failures/ problems requires an engineer having practical experience as well as academic knowledge. Operational and maintenance staff rarely have sufficient time to undertake a detailed investigation as is possible with a dedicated contractor. Practical experience provides understanding of machinery operation/ maintenance. Academic knowledge allows an assessment of the range of possible damage mechanisms.

Have you suffered an unaccountable machinery damage problem?
Brabon Engineering Services will conduct a meticulous investigation and help prevent further incidents.

Call for a discussion: +353 87 383 5043
email for a proposal: info@brabon.org
Send a message via our contact page
Connect on LinkedIn

Preliminary actions in event of machinery damage

While there may be no initial intention to undertake a survey and root cause analysis of a machinery failure/ damage incident, it may be prudent to preserve as much evidence as possible should a subsequent decision be taken in favour of an investigation.
Brabon Engineering Services would suggest the following:

• Preserve and retain as many of the damaged components as possible.  All contact and fracture surfaces should be protected from corrosion with a liberal coating of oil/ grease.
• Record numerous photos of the scene and damaged components.
• Mark orientation and relative positions on dismantled components.
• Print/ download relevant operating parameters and alarms from all monitoring systems during the period prior to the event.
• Take samples of oil/ fuel from the machine.

Machinery failure/ damage survey case study 1

Vessel

The case involved a feeder container vessel of about 1,000 TEU having a medium speed diesel, geared reduction, controllable pitch, single screw propulsion plant. Main propulsion was a four-stroke, 7 cylinder, large bore, engine having a maximum continuous rating of about 9,700 kW at 425 rpm.

Engine lube oil system

The engine lube oil system comprised; a (double bottom) wet sump tank, suction pipework, two crankshaft driven oil pumps (in parallel) located at the forward end of the engine as well as a motor driven oil pump, a pilot operated pressure regulating valve (spill to sump), plate type cooler, automatic back-flushing filter, (back-up) duplex filter, then the on-engine distribution system.

The as-built suction pipework comprised; a shallow bell inlet, a single vertical suction pipe transiting the tank top to a ‘tee’ where the flow split to approximate mirror arrangements to the port and starboard pumps. Each branch of the ‘tee’ led horizontally and athwartships to a (spring actuated) piston type non-return valve, a suction strainer, then a series of bends to the pump inlet. The pipework branches either side of the ‘tee’ were higher than the pump suction inlet. The port and starboard branches each had venting and priming arrangements (pressure oil from the distribution system).

Engine driven lube oil pumps

The engine driven lube oil pumps were a positive displacement gear type having two shafts with single helical gears. The pumps were gear driven off a wheel at the crankshaft free end (at a step-up speed). Five-off, thin wall, steel backed, whitemetal lined bearings supported the shafts. The pump bearings were pressure fed with a supply of oil (from the engine lube oil distribution manifold) connected at the pump mounting flange. Oil was fed to each bearing via holes and grooves in the pump casing and the bearing housings, then into the bearing via a hole in the shell. Oil was supplied to the layshaft/ driven shaft bearings from the driving shaft via holes in adjoining bearing housings.

The (pump) helical gear teeth generated opposing axial reaction forces at the tooth contact point which in turn result in axial thrust between each gear and the pump casing. Thrust washers were not fitted, and the steel to steel (thrust) running surfaces were lubricated by oil in the pump.

The pump manufacturer advised a suction pressure limit of -0.4 barg (equal to an absolute head of 7.2 m of oil in a vacuum). The net positive suction head available at the pump inlet was estimated as 9.7 m (equal to -0.2 barg). This was based on a vertical height of 1.2 m from the oil surface to the pump inlet plus an assumed dynamic head of 1.0 m of oil.

Failure incidents

The vessel had been accepted from the construction shipyard and in service for about three months. During this time there had been a series of problems related to main engine lube oil pressure as follows:

• During shipbuilder’s sea trials there had been problems maintaining lube oil pressure.
• At about 860 running hours the engine shut-down due to high differential pressure over the main lube oil filter as well as the additional duplex filter. The port engine driven lube oil pump was found to have limited damage, while the starboard pump was severely damaged. Both pumps and the pressure regulating valve were replaced.
• A second incident occurred at about 1,020 (engine) running hours. The engine was stopped on multiple occasions due to decreasing lube oil pressure (the regulating valve was adjusted several times), high differential pressure over the lube oil filter and finally high temperature on the No 1 and No 3 large end bearing splash oil sensors. The port engine driven lube oil pump was found to be severely damaged while the starboard pump had suffered limited damage. Both pumps were replaced (second time).
• A third incident occurred at about 1,310 running hours. The engine lube oil pressure decreased triggering an automatic start of the motor driven pump indicating there had been a failure of the engine driven pumps. The port engine driven pump was found to have limited damage, while the starboard pump was severely damaged.

After the third incident the vessel was taken off-hire to investigate and rectify the problem(s).

Engine lube oil system modifications

Following the second incident the engine sump tank was modified with a baffle to guide the return oil aft in the tank and prevent an oil short-circuit from the return to the suction. The gap between the tank suction well floor and the entry to the suction pipe was also increased.

Following the third incident the suction pipework was modified as follows:

• The single vertical suction run was replace by individual port and starboard suction pipes. Space constraints did not allow the horizontal sections to be repositioned at a lower height. A weir was fitted between the suction inlets to prevent interaction.
• Both non-return valves were relocated to just above the tank top.
• Flexible bellows were fitted in the suction connections to the port and starboard pumps.
• The automatic filter (back) flushing line was connected to the main oil return into the wet sump. This was so any debris would settle out during the normal tank residence circuit.
• The pressure regulating valve spill pipe was extended to a point below the oil surface (to reduce foaming).

On completion of the system modifications a 20 μm flushing filter was used to progressively clean the lube oil system.

Investigation and findings

Each instance of pump damage/ wear involved damage to the layshaft (driven shaft) bearings/ shaft journals and was greater than any damage suffered on the driving shaft. That is, bearings remote from the supply were more susceptible to loss of lube oil supply pressure. On the driving shaft, the free end bearing tended to have greatest wear (of the three drive shaft bearings). This suggested oil starvation was the mechanism of failure/ damage. The axial thrust faces and the (pump) gear teeth also typically suffered damage. This was likely consequential to increased friction in the bearings and reduced lubrication.

The following pump failure scenarios were considered:

• The suction pipework arrangement to the engine driven pumps, (with the pump inlet lower than the horizontal tee section) resulted in a barometric loop/ air lock/ air pocket that starved supply to the pumps. The reduced flow through the pumps (and output), then did not provide sufficient output pressure (and flow) as well as internal lubrication. This had a consequence of pump bearing failures, thrust surface damage as well as engine bearing damage.
• The (as-built) location of the non-return valve (in the upper ‘tee’ section) would result in additional air volume from the vertical section supplying the pocket at each engine start. Small variations in vessel list angle could direct more of the air volume to either the port or starboard branch. This would lead to one pump becoming starved and suffering severe damage with a random port/ starboard failure preference.
• The (as-built) arrangements of baffles (not fitted) and return pipes in the sump tank would allow aerated oil to be drawn into the pump suction inlet which accumulated/ maintained the air volume in the pocket. While the third failure occurred after the sump tank had been modified to prevent aerated oil to the pump suction, the engine ran for a longer period from the second incident than the interval from the first to second incidents.
• The (as-built) arrangements of baffles and return pipes in the sump tank would allow dirty oil (recirculated filtrate or remnant dirt from apparently poor construction) to be drawn in the suction inlet for the pumps causing accelerated wear in the pumps.

The following additional scenarios were considered as unlikely to have caused/ contributed to the pump failures:

• Defective pumps assembled with incorrectly arranged internal lube ports. Given the number of repeat failures (six in total), with one pump suffering severe damage in each incident, then the cause was likely to be unique to the vessel, i.e. oil supply to the pumps.
• Misalignment of the pump mounting or pipework connections causing distortion of the casing(s) are unlikely to have resulted in misalignment between the pump shafts and bearings. The observed contact/ wear patterns on the pump gear teeth and bearings did not indicate this occurred.
• A blockage in the suction or distribution system.

Sea trial

A sea trial was conducted after completion of the modifications and repairs. A duration of 100 hours was selected as similar to the period between previous pump failures. The pumps were monitored for suction/ discharge pressure, vibration, temperature as well as oil level in the suction strainer.

Performance of both pumps during the trial was satisfactory. The mean suction pressure at each pump during the sea trial was steady at about -0.24 barg and was within the pump manufacturer’s limit (-0.4 barg). Discharge pressure varied between 7.4 barg and 7.0 barg.

On completion of sea trials the engine driven lube oil pumps were dismantled for inspection. Both pumps were found to be in satisfactory condition. The pressure regulating valve was also dismantled and readjusted. The automatic filter and duplex filter were both opened and found to be clean.

Conclusions

The survey found that the machinery failure/ damage likely had multiple contributory factors in each incident including:

• Air accumulation in a pocket in the suction pipework. This was likely the primary cause. The reduced flow through the pumps, then did not provide sufficient internal lubrication and output pressure/ flow resulting in bearing damage. The non-return valve (as-built) location would result in additional air to the pocket at each engine start.
• Aerated oil at the suction inlet in the sump tank (supplying the air pocket).
• Contaminated oil at the suction inlet in the sump tank. The relative significance of these two factors may have varied between failure incidents.

Given the satisfactory condition of both pumps after 100 hours operation, it is likely that the circumstances that caused the failures were rectified. The modified pipework still had an air trap, however, the modifications likely eliminated the accumulation of air in the system. Accordingly, the suction systems should be primed (using the electric motor driven pump) before each engine start.

Problems were experienced with the pressure control valve causing a small decrease in the system pressure during (and after) the sea trials. This is unlikely to have contributed to the pump failures, however, it was an additional complicating factor in the investigation.

Available case studies

• Auxiliary Engine Exhaust Valve Stem Damage, May2019
• Engine Crankshaft (Fatigue) Failures, July 2019
• Intermediate Shaft Fracture, June 2019
• Main Engine Large End Bearing Damage, May 2019
• Main Engine Lube Oil Pump Failure, May2019
• Propeller Shaft Fatigue Failure, August 2020
• Ship Rudder Stock Problems, July 2019
• Stern Seal and A-bracket Bearing Problems, February 2020
• Sterntube Aft Bearing Whitemetal Fatigue Damage, May 2019

Case studies available on request.

Machinery failure/ damage survey case study 2

Vessel and propulsion shafting

The case involved an oil tanker of about 193,000 tonne DWT.  The vessel had a diesel electric twin screw propulsion system.  Four off medium speed diesel generators supplied power for main propulsion, auxiliaries, cargo systems and hotel services.  Each propulsion shaftline was independent and driven by an AC, brushless, variable speed electric motor prime mover having a maximum continuous output of 10,000 kW at 85 rpm.  Each motor rotor was rigidly coupled to a shaftline with a four blade, fixed pitch propeller.  Propeller rotation was inward over the top, i.e. starboard shaft rotated counter-clockwise when viewed from aft.
Each shaftline comprised the following:

• A motor rotor shaft supported forward and aft by an oil lubricated white metal lined, plain journal type pedestal bearing. The bearings were fitted with continuously run, jacking/ circulating oil pumps.
• A thrust shaft with a nominal diameter of 460 mm having oil lubricated, white metal lined, tilting-pad type thrust bearing sets to react the ahead and astern propeller thrust. An oil lubricated white metal lined, plain sleeve bearing supported the shaft in the radial direction on the aft side of the thrust collar.
• Two intermediate shaft sections having a nominal diameter of 460 mm supported by an oil lubricated, white metal lined plain sleeve type intermediate bearing.
• A propeller shaft having a nominal diameter of 520 mm supported by sea water lubricated, non-metallic, composite, plain sleeve type sterntube bearings.

Failure incidents

The vessel operated for about two years after being commissioned without any thrust bearing high temperature problems.  The starboard ahead thrust bearing then suffered overheating and damage incidents as follows:

• Incident 1 occurred while motoring at 83 rpm with the vessel laden and immediately following a course change. The bearing pads had a severe thermal wipe with the loss of all lining.
• Incident 2 occurred after a further period of about three months while motoring at 70 rpm in heavy weather with the vessel partially laden. A partial wipe had occurred on the bearing pads.
• Incident 3 occurred after an additional period of about one month while motoring at 70 rpm and resulted in partial wiping of the bearing pads.
• There were further incidents where the starboard ahead thrust bearing reached a temperature of about 64°C, but, prompt slowing of the shaft speed produced a corresponding reduction in bearing temperature.

Following incident 1 and 2 the shaft speed was limited to 70 rpm and an automatic slowdown linked to thrust bearing temperature was applied in the machinery control and monitoring system.  The lube oil was also changed from VG 100 to VG 150 extreme pressure.

Thrust block

The ahead and astern bearing sets each had 11 pads with a tin-based whitemetal lining.  The pads pivoted about a radial line offset to give a longer leading edge lever.  The thrust collar had an outside diameter of about 1,000 mm.

The thrust block was a self-contained unit with an integral lube oil system.  A segment of the thrust collar extended down into the sump and lube oil adhering to the outside diameter was redirected forward and aft at the top by a scraper/ diverter.  On the forward (ahead) side lube oil was fed to the small diameter of the thrust collar and across the working surface by centrifugal force.  On the aft side lube oil was fed to the top of the radial bearing.  The astern thrust pad set was lubricated by oil spill from the radial bearing and excess oil around the top of the thrust collar.  An additional pumped lube oil circulating/ supply system had been fitted following the second damage incident.

The ahead and astern pads at the 12 o’clock position had top vertical clearance pockets for Resistance Temperature Detectors located adjacent to the pivot axis and about 18 mm from the bearing surface.  The bearing manufacturer suggested a normal thrust pad operating temperature of approximately 60°C and a limit of 75°C. The thrust block housing comprised symmetric upper and lower halves.  The two halves had horizontal flanges at each side.  Five off 50 mm diameter fitted bolts (total 10 off bolts) fastened the halves together and secured the thrust block to its structural seating.

Investigation and findings

A process of elimination was applied with the following measurements and inspections on the port and starboard propulsion shafts.  Dynamics measurements on the port and starboard shafts during run-up trials in the ballast and laden conditions were conducted in order to eliminate shaft vibratory motion as a contributory factor.

The starboard thrust block upper half was dismantled for a visual inspection of the thrust pads and journal bearing upper half.  The thrust bearing pads were all equally polished.  The radial bearing had no running contact marks in the upper half, suggesting the incident had not involved severe shaft dynamics.  During reassembly it was noted that gasket compound was applied to the mating flange faces of the lower and upper halves.

Port and starboard thrust block casing feet movement

Displacement of the starboard and port thrust block casing feet relative to the ship seating were recorded during sea trials.  Measurements were made on the upper and lower feet at the inboard and outboard sides using dial indicators as well as strain gauge displacement transducers.

Movement of the port and starboard thrust block feet during an incremental run-up of shaft speed in the laden condition is shown in Figure 1 and Figure 2 below.

• 

• 

On the port thrust block at speeds above 50 rpm there was an approximately linear movement of the feet up to about 0.13 mm at 84 rpm.  The difference between the upper and lower feet was less than 0.02 mm.  When the speed was reduced from 84 rpm to 30 rpm on completion, the feet returned to within 0.03 mm of the initial position.

On the starboard thrust block, outboard side, the upper half deflected 0.07 mm further than the lower half.  On reduction of the speed from 84 rpm to 30 rpm, the outboard upper and lower feet returned to within 0.03 mm of the initial position.

Notably, on the starboard thrust block, inboard side, the upper half deflected 0.24 mm at 84 rpm, while the lower half moved 0.21 mm aft between 70 rpm and 84 rpm, then moved 0.20 mm further aft when the speed was reduced from 84 rpm to 30 rpm, i.e. giving a permanent offset.  Relative movement of the upper and lower halves would result in changes in the load distribution between the thrust bearing pads in the respective halves.

Starboard thrust bearing pad temperatures

The starboard ahead thrust bearing pads at the top, bottom, port and starboard positions were fitted with K-type thermocouples to record temperature during sea trials.  Lube oil feed temperature at the scraper/ diverter was also recorded.  Each of the four pads had a radial hole approximately 2 mm from the whitemetal to steel interface located in the upper region between the pivot line and the trailing edge with the thermocouple being fixed in epoxy.

The starboard ahead thrust bearing temperature during an incremental run-up of shaft speed in the ballast condition is shown in Figure 3 below.

For speeds up to 40 rpm, temperature of the four pads remained within a range of about 1.5°C.  At higher speeds, the port/ starboard pad temperatures showed a greater increasing trend than the top/ bottom pads by up to about 4°C at 84 rpm.  This suggests greater load on the port/ starboard pads as well as slightly greater load on the bottom versus top pads.  There was a small unremarkable temperature difference between the port and starboard pads.  At 84 rpm the bottom pad temperature was also about 1°C greater than the top pad.

Port and starboard shaft static bearing loads

Shaft alignment, bearing load measurements were conducted on the port and starboard shafts while stopped in the ballast and laden conditions.  Bearing loads were derived by bearing jacking as well as strain gauge shaft bending measurements.  Port and starboard shaft bearing loads for the ballast and laden conditions are shown in Table 1 below.

The port and starboard thrust block radial bearings both had a consistent load of 11.0 tonne to 12.4 tonne for all vessel conditions.  Shaft jacking measurements at the port and starboard thrust radial bearings had a jacked rate of between 3.6 tonne/mm and 5.9 tonne/mm.  That is, an offset deviation of at least 2.0 mm would be required to unload the bearing.  It is unlikely the bearing damage had been a result of shaft dynamics due to the thrust block radial bearing becoming unloaded.

Starboard thrust block vibration

Vibration of the port and starboard thrust block in vertical, transverse and longitudinal directions was measured using triaxial strain gauge accelerometers connected to a data logging system.  The starboard thrust block overall maximum repetitive amplitude of vibration velocity, as described in ISO 4867:1984, during an incremental run-up of shaft speed in the ballast condition is shown in Figure 4 below.

[ISO 4867:1984, code for the measurement and reporting of shipboard vibration data, superseded.]

Vibration levels were typically less than 10 mm/s and above 40 rpm vibration was less than 6 mm/s, i.e. relatively low.  In addition, there were no significant peaks at particular speeds that would be typical of a vibratory response.  Given the shaft mass, it would be anticipated that an adverse shaft dynamics would generate a significant level of vibration.  Note that the main generators were located one deck directly above the shaft tunnel and would have contributed to the measured vibration.

Starboard shaft radial displacement in way of thrust block

Dynamic radial displacement of the port and starboard shafts at the aft and forward ends of the respective thrust block casing were measured during sea trials.  Pairs of non-contact proximity probes arranged 90° apart were used to record instantaneous shaft radial displacement in the vertical-transverse plane at each station.  Each probe was connected to a data logging system.  Recorded signals were analyzed for mean displacement to derive progression versus speed as well as for vibratory motion.

Mean shaft centre as well as the semi-circular envelope of maximum repetitive amplitude of vibratory shaft radial displacement at each speed step in the ballast condition run-up at the aft and forward ends of the starboard thrust block are shown in Figure 5 and Figure 6 below.  An assumed bearing clearance of 0.4 mm on diameter is also shown for comparison.

• 

• 

Progression of the mean centre at the aft end, adjacent to the radial bearing, was towards port by about 0.04 mm at 84 rpm and slightly up.  At the forward end, adjacent to the ahead thrust bearing, the mean centre advanced to starboard about 0.05 mm and down 0.04 mm at 84 rpm.

Maximum repetitive amplitude of vibratory shaft radial displacement, zero to peak single amplitude, at the aft and forward ends of the starboard thrust block versus shaft speed in the ballast condition run-up are shown in Figure 7 and Figure 8 below.

• 

• 

Vibratory motion at the aft end in the horizontal and vertical directions showed steadily increasing trends with shaft speed with a maximum zero to peak amplitude of about 0.15 mm.  At the forward end the vibratory motion in the horizontal and vertical directions showed a step increase from less than 0.05 mm to about 0.2 mm between 40 rpm and 50 rpm.  Shaft motion at the forward end would be greater than aft given that the forward end sensors were about 930 mm from the forward edge of the radial bearing.

Waterfall spectra plots of shaft radial displacement in the vertical and horizontal directions at the forward end of the starboard thrust block are shown in Figure 9 and Figure 10 below.

• 

• 

Spectra characteristics were different above 55 rpm with predominantly harmonic motion below 55 rpm, i.e. 1x, 2x, 4x, 6x, 8x, etc.  Above 55 rpm, an approx 0.5x order appeared, the 1x order diminished, while the 2x and 4x propeller blade pass orders became blurred with random content.

The change in vibratory displacement after 40 rpm and the frequency characteristics above 55 rpm suggest a change in shaft dynamic motion around this speed.  The reduction in the 1x order indicated shaft motion became less similar each revolution, i.e. less harmonic and more random.  The movement of the thrust block feet with increasing shaft speed, Figure 2, may have been a contributory factor in the increased shaft vibratory motion.

Starboard shaft axial displacement

Dynamic axial displacement of the port and starboard shafts was measured during sea trials using non-contact proximity probes at the motor output flange.  The sensors were connected to a data logging system.  Recorded signals were analyzed for mean and vibratory displacement.

The starboard shaft mean and maximum repetitive amplitude of vibratory axial displacement versus shaft speed in the ballast condition run-up is shown in Figure 11 below.

The mean displacement showed a consistent increasing trend against shaft speed and thrust, as would be anticipated.  Vibratory motion amplitude was generally less than 0.1 mm and did not show any significant peaks characteristic of a resonance response.

A waterfall spectra plot of starboard shaft axial displacement for the ballast condition run-up is shown in Figure 12 below.

Spectra characteristics were different above 55 rpm with a reduction in the 1x order, normally associated with slight flange run-out, while an approx 0.5x order and slightly less than 2x order both appeared.  The 4x and 8x propeller blade pass orders also became blurred with random content.  The changes in spectra versus speed were broadly similar to that of the shaft radial displacement measurements.  Accordingly, the movement of the thrust block feet with increasing shaft speed, Figure 2, may have been a contributory factor in the increased shaft vibratory motion.

Conclusions

The circumstances and measurements suggested the following:

• The starboard thrust bearing high temperature incidents was likely caused by movement of the thrust block feet, relative to the ship seating and between each other.  Use of gasket compound would also significantly reduce the shear stiffness of the joint between the thrust block casing halves.  Deflection of one casing half would result in an overload of the pads supported by the other half.
• The recorded starboard ahead thrust bearing pad temperatures showed an increasing difference between port/ starboard versus top/ bottom.  This was consistent with adverse changes in load distribution across the bearing pads versus shaft speed/ thrust load.
• The high temperature incidents occurred at between 70 rpm and 84 rpm in ballast and laden conditions.  Incident 1 followed a manoeuvre and incident 2 occurred during heavy weather.  Thus, thrust load was likely a factor in the incidents.  However, overheating incidents did not occur every time the vessel was motored at above 70 rpm, thus there was likely a time varying or random element in the incidents.
• A change in the starboard thrust block likely occurred after two years in operation (as yet unknown).

It was suggested that four of the holding down bolts be replaced with hydraulically tightened, expanding sleeve type bolts.  The remaining holding down bolts should be fitted with positive means of locking, e.g. stopper bars.  Gasket compound should not be used on the thrust casing main flange faces.

The following additional improbable/ trivial scenarios were considered as unlikely to have contributed to the failure:

• The port and starboard thrust block radial bearings had satisfactory ballast and laden condition loads.  The starboard ahead thrust bearing pad temperature differences port to starboard sides and (separately) top to bottom were 1°C or less.  Thus shaft alignment, or local alignment of the thrust bearing, were unlikely to have been a factor in the incidents.
• Starboard (and port) shaft dynamic motion amplitudes in the radial and axial directions were low and did not show any significant peaks characteristic of a resonance response.  Thus, lateral or axial vibration resonances were unlikely to have been a factor in the incidents.  However, the shaft dynamic motion was likely affected, and showed changes in characteristics, of the cause(s).
• Oil starvation due to a blockage in the oil gallery or loss of oil from the sump.  This would have affected the thrust radial bearing, however, no such damage occurred.
• Insufficient cooling of the thrust block lube oil.  The temperatures recorded using epoxy set thermocouples were within typical allowable limits.

Machinery damage modes

Seemingly intractable mechanical damage problems often involve multiple, commonly encountered, contributory factors/ phenomena. Unusual anomalies, by their nature, are less frequently encountered. For example, bearing failure due to oil starvation is more common than say shaft lateral vibration resonance (whirling). The succession/ timing of each event can be pivotal in the (overall) failure. The challenge is to correctly discover the root cause(s) from all the available evidence.

Considering various component damage mechanisms:

• Fatigue fracture. This can occur irrespective of a conservative margin between yield strength and mean loading. This margin may be indicated by the relative size of the final overload area. A fatigue fracture surface is characteristically different to that of an overload failure at microscopic and macro scale, e.g. beach, arrest and ratchet marks. The shape/ orientation of fracture face indicates the type of stress having caused propagation, e.g. a spiral fracture face would be associated with torsional fatigue. Note that the dominant stresses may change as the crack propagates through the component section.
• Material deficiency or change. This may involve metallurgical imperfections such as forging lap, non-metallic inclusions, casting voids or corrosion.
• Deformation, elastic or plastic. Subject components may include; structural elements, linkages, bolts/ fasteners and gear teeth.
• Elastic/ plastic deformation. This is generally readily identified although the cause of overload may not be clear. It is possible for heavy vibration to generate excessive stresses, although this tends to occur infrequently. In this scenario it is important to distinguish between high levels of (source) excitation forces, and resonance (response) conditions.
• Surface texture changes, e.g. wear, pitting, fretting, corrosion and cracking. Note there is some overlap between particular categories.
• Assembly condition problems. This can include fasteners not tightened, looseness/ seizure and surface changes like surface corrosion and wear.
• Leakage/ compromised fluid containment. Shaft seal leakage may be due to normal wear and tear or lack of maintenance. In some cases, shaft vibration can manifest with seal leakage.

Contributory factors may include; mean force/ load, vibratory/ fluctuating load, time/ cycles of service as well as extreme/ variations in environment temperature.

Machinery failure/ damage analysis and investigation procedure

(How Brabon Engineering Services carry out machinery root cause analysis and troubleshooting?)

Machinery failure/ damage survey investigations by Brabon Engineering Services follow a systematic process with three main stages. This process is based on experience of a wide variety of machinery failures.

Numerous published root cause analysis procedures are available. However, no procedure should be considered a magic panacea that will self-generate an answer. Notwithstanding, we would be pleased to follow any root cause analysis process our Clients may nominate.

1. Gather relevant information

• Collect observations and information of failure event (timeline), e.g. bearing temperature trends, running hours and maintenance history.
• Examine the components/ symptoms to identify the fundamental damage processes. Detailed elementary observation is the foundation of the investigation process.
• Conduct any necessary metallurgical examinations to confirm or further understand the characteristics of the damage and the component properties. Detailed metallurgical examinations are usually sub-contracted. However, the metallurgical analysis needs to be closely managed so as to reveal/ confirm the damage mechanism, rather than simply repeating the component (material) acceptance/ certification tests.
• Acquire understanding of the underlying principles of the machine. Every machine is a unique combination of detail design features resulting in differing failure tendencies. For example, a shaftline bearing may have different lube oil supply arrangements leading to variations in possible damage mechanisms.

The initial investigative process involves applying simple tools, such as examining components and observing the failure scene as well as taking notes and photos. An investigation should give equal consideration to the machine mechanics, the sequence of events and materials/ metallurgical factors. In cases involving interactions of different components, the failure mechanism is principally identified via a sufficiently thorough examination on-site of the damaged machinery.

2. Develop failure scenario hypotheses

List possible cause/ effect couples and sequences to develop hypotheses of failure scenario(s). This step requires considered thought, and is generally aided by experience. Failures may involve multiple contributory factors and the succession/ timing of each event can be pivotal in the (overall) failure. Further, seemingly intractable mechanical machinery problems often involve multiple, commonly encountered, contributory factors/ phenomena. Unusual anomalies, by their nature, are less frequently encountered. For example, bearing failure due to oil starvation is more common than say lateral vibration resonance (whirling).

Hypotheses should be based on the evidence found, but not bounded by preconceived assumptions (although this is human nature). All permutations of intermediate event sequences should be considered. Checklists also provide a cue to ensure that the damage event has been considered from every possible angle. Writing a narrative of each hypothesis scenario enables a disciplined review of the logical links in each cause/ effect/ sequence scenario.

3. Review failure scenario hypotheses

Critically review the scenario(s)/ intermediate conclusions and iterate the process.

Hypotheses are (initially) ranked by the probability of each failure mechanism in normal experience. (The medical fraternity have a saying, ‘if you hear hoofbeats, think of horses not zebras’.) For example, one possible scenario for a diesel engine connecting rod large end failure based on common experience is incorrect tightening of the bearing cap bolts/ studs, which may be more likely than say a fracture of the bearing cap.

Correlation between observations and conclusions provides corroboration, and hence lends weight to, the failure scenario hypothesis. (Although it is important to distinguish between evidence of a phenomena and correlation of events where there may be no causation link.)

Successful failure investigations typically require persistence with numerous iterative critical reviews of the data and intermediate conclusions.  The importance of perseverance in discovering the cause(s) cannot be underestimated, and applies irrespective of the analysis process.

Have you suffered machinery damage?
The above recipe may sound easy, but as with any task, practice and knowledge make a significant difference.
Successful investigations require practical experience, academic knowledge and extreme perseverance.

Call for a discussion: +353 87 383 5043
email for a proposal: info@brabon.org
Send a message via our contact page
Connect on LinkedIn

Further information

Suggested further reference information:

• Bloch H.P., Geitner F.K., Machinery failure analysis and troubleshooting – Practical machinery management for process plants Vol 2, Gulf Publishing Company, 1990. Heinz Bloch has also authored numerous books and papers on the subject of machinery reliability as well as failure/ damage investigation/ survey.
• Forensic engineering. The Wikipedia article (linked) on this subject, although the scope is greater is more general than mechanical engineering.
• Failure Modes and Effects Analysis (FMEA). The Wikipedia article (linked) considers FMEA in the context of the design review process, i.e. to minimise the likelihood of adverse events.
• Reliability engineering. The Wikipedia article (linked) on this subject provides a general discussion of component review, failure modelling and statistical process control methodologies.

~..~

The above notes are applicable to typical generic events. However, please note that every actual event is unique and requires individual consideration. Any action taken upon the information on this website is strictly at your own risk. Brabon Engineering Services are not responsible in any way whatsoever for your use of the information.