Aeroengine Safety

Fig. "Bearing failure by shaft unbalances" (Lit. 23.1.2-1): During the flight, there was a loud noise and the helicopter tilted to the left. The right aeroengine dropped out. The investigation showed a failure at the thrust bearing of the gas generator shaft. The failure was the result of an unbalance at the gas generator shaft. Cause was the fracture below the root platform of a turbine rotor blade. The track on the race of the roller bearing shows markedly signs of a forced overload. Concerned are extremely seizing marks and plastic deformations. This failure mode fits with the very short remaining life after the first failure symptom. Typical for a bearing overload, caused by an unbalance, is the intensified failure at a small circumference section of the bearing ring on the shaft (frame below). This failure mode explains itself with the steady unbalance, rotating with the shaft.
Does a bearing failure occurat high speed shafts of small aeroengines, in seconds a catastrophic failure must be expected.

Fig. "Bearing failure by electric continuity" (Lit. 23.1.2-2): The following investigation of the aeroengine at the OEM, after the inflight shut down showed, that the bearing No. 1 of the compressor had heavy damages and features of overheating like annealing colours. Balls and the race of the inner ring revealed locally heavy wear and deformations. They form during slipping of the balls. In spite of the progressed failure at the unloaded side of the race of this thrust bearing, indications of electrical continuity with the formation of arcs have been found (Fig. "Bearing failure by lightning strike" and Fig. "Electrically caused bearing failure"). That just the axial unloaded side of the ball bearing was affected, is typical for the failure mechanism of electrical continuity. Namely electrical arcs/sparks don't develop at metallic contact (direct current) of the balls to the race ring, but need the separating electrical insulating oil film.
Electrical arc weld puddles lead after load depending lifetimes to fatigue of the race with break outs (pits/pittings). Featuresm for electrical continuity could be traced back through the accessory gear up to the shaft of the startergenerator (frame below). At two or three neighbouring teeth of the gear on the starter shaft, the alignment of the craters repeated. This corresponded electrical pulses with the four times frequency of the starter shaft rotation speed. In contrast, the distribution at the mating gear was uniform. With this the startergenerator was identified as source of the current.
Tests at the testrig showed, that the problem was triggered by brush abrasion. This gathered in the casing of the startergenerator and caused pulsating electrical discharges. These curret pulses flowed through the gearwheels up to the bearing.
Four similar failures have been reported before to the OEM. In fact, in three cases electrical circuit could be identified, however because of the failure extent it was not posible to identiffy the power source. All together 17 cases got known.

In one case, the logistics showed a generator exchange before, let suppose, that from the electrical circuit till the drop out of the bearing a formation of the failure of some hundred operation hours can be needed.

Fig. "Secondary effects of bearing failues" (Lit. 23.1.2-3 and Lit. 23.1.2-4): In cabin and cockpit of a civil twin-jet in a short period of time dense smoke formed. In this situation a dangerous annoyance of the crew exists. The smoke obviously came from the bleed air of the high pressure compressor of one of the aeroengines. Thereon the pilot shut down the suspect aeroengine.
A later investigation unfolded, that the smoke was connected with a failure at the bearing number 3 (sketch berlow left). About a quarter of a hour before the entrance of the smoke, in the cockpit a warning appeared, that the associated oil filter is blocked. In the two years before already 20 up to 2006 all together 55 (!) such bearing failures occurred. In 12 cases it came to an in flight shut down (IFSD), 43 times the aeroengine had to be dismounted (unscheduled engine removal = UER).
Concerned have been fatigue break-outs (fatigue pittings) at the balls and in an extreme case, the fracture of the race ring. Especially after the fracture of the race ring, the described smoke formation formed.
Cause for the fatigue failures is a contamination of the supplied oil. Concerned is a chipped off hard coating at the shaft, made from a titanium alloy, which bears the inner ring (sketch right). Affected are thermal spray coatings by a low energetic plasma spray process. Obviously, parts with high energy plasma spray coatings have not been concerned.
As remedy the exchange of the suspect parts and frequent magnetic chip controls at several bearing chambers in specified intervals took place (125 operation hours).

Fig. "Dangerous particle sources in bearings" (Lit. 23.1.2-5 and Lit. 23.1.2-6,): Deteriorating particles can get into main bearings in different manners.

Particles from the labyrinths: It must be absolutely prevented, that oil leaks from the bearing chambers. This could trigger an oil loss and in an extreme case an oil fire (volume 2, Ill. 9.2-11). Therefore a certain pressure drop through the seal to the bearing chamber must be guaranteed (volume 2, Ill. 9.2-1). This means for a labyrinth seal, that leakage air flows through the seal into the bearing chamber. During rub developing hard particles like chipped off tip hard-facing (example 23.1.2-1), metallic abrasion or reaming rub from a rub in coating (e.g., age hardened, see volume 2, Ill. 7.2.2-1) these can be blown into the bearing.
Also contaminations of the sealing air, like abrasion or dust, can so get through the labyrinths into bearings.

Particles from the air conditioning like oil coke or dust during operation conditions with inward orientated flow.

Particles from the fresh oil are coke, hard crack products (Fig. "Formation of depositions in hot oil systems"), which form in the supply line. Possibly are also transported contaminations like not filtered out peening/blasting particles and abrasion.

In the bearing chambers itself, dangerous particles can develop. To these belong:
Abrasion/wear chips of loose bearing seats (fretting) or from elastic dampened configurations (Ill. 23.1-1).
Coke which forms in the bearing chamber itself. There are for this different causes, like too hot chamber walls (Fig. "Problems of bearing chambers near hot parts", Fig. "Problema by heat soaking" and Fig. "Importance of resting time at idle"), overheatings or limited oil fires.
Bearing own particles are frequently secondary failures of a failing bearing element. Typical are abrasion splints from cages (Fig. "Deterioration by cage slipping so called skidding", Fig. "Bearing fractures by shock loads" and Fig. "Oscillation of the rollers sc skewing weaving"). Further possibilities are fragments of the cage rivets (Fig. "Forces and moments inside a bearing") or fatigue break-outs (pittings) of the bearing races.

Fig. "Wear loaded sliding surfaces in bearings" (Lit. 23.1.2-7 and Lit. 23.1.2-8, example 23.1.2-2): For high speed rotating anti friction bearings (high D x n, Ill. 23.1-10.1 and Fig. "Tendency of bearing speeds") at aeroengine typical high service temperatures, the danger of overheating exists. Especially much heat during sliding of the rolling elements against the cage (sketch below right, Fig. "Forces and moments inside a bearing" and Fig. "Oscillation of the rollers sc skewing weaving") and of the cage against the bearing rings (sketches below and in the middle, Fig. "Bearing behaviour by guidance of the cage"). Especially for outbord guided cages (Fig. "Bearing behaviour by guidance of the cage"), because of the thermal expansion, self-energising friction must be expected. This rises the already high circumferential forces at the cage (Fig. "Forces and moments inside a bearing") even more. These are transferred by the rolling elements an can cause the fracture of the cage (sketch below).
With extreme friction heat must be reckoned, when the bearing runs under unnormal operation conditions. Such conditions are skidding (Fig. "Deterioration by cage slipping so called skidding") and roller weaving (Fig. "Oscillation of the rollers sc skewing weaving").

Fig. "Skidding by unsuitable bearing tolerances" (Lit. 32.1.2-9): Obviously a production caused dimensional deviation triggered failures of roller bearings. So the following failure mechanism could get effective: Arise too high friction forces at the rollers, lateral jamming between the bords of a bearing ring these can no more balanced from the kinematic driving forces. Cage slip (skidding) occurres (Fig. "Deterioration by cage slipping so called skidding"). This leads to the deterioration of the race tracks (Fig. "Skidding failure modes at races") and the failing of the bearing.

Fig. "Compressor surge by oil loss" (Lit. 32.1.2-10): After climb the right aeroengine of this two engined airliner surged two times. The pilot returned to the initial airport. During extending of the landing gear and the landing flaps, the aeroengien surged again at 60% power. After that, the power was reduced to idle. After the landing a 3 meter long flame at the exhaust of the aeroengine could be seen. An evacuation of the aircraft was carried out without problems.
The concerned aeroengine was investigated at the safety authority. Thereby it showed, that the bearing of the high pressure turbine (bearing No. 4, sketch below) failed, obviously because of lubrication oil starvation. The total running time of the bearing was about 20 000 hours, the last inspection was carried out about 2400 hours ago. The failure sequence could be reconstructed as follows:
The braking /deceleration effect of the failing bearing brought the high pressure shaft and the low pressure shaft to about the same rotoation speed. This led to a flow disturbance in the compressor and the surging. Not before a short term increased power during the approach for landing, the total failure of the bearing occurred. The result was an oscillating of the high pressure shaft, with the failure of a seal in the rear oil sump. The escaping oil ignited and an oil fire developed, whose flames could be seen during landing.
As cause of the bearing failure the investigating authority sees shortage of lubrication oil. Unfortunately the available literature gives no satisfying explanation for this conclusion. Obviously, former parallel cases allow this suggestion.
Problems with the bearing No. 4 of this aeroengine type is at the operators generally known. A material change is in the test phase. The introduction is scheduled after successful tests in about two years. However the aviation authority seems to be rather skeptical about the efficiency of this change if lubrication lacks.

Fig. "Oil deficiency shaft failure sequence 1" (Lit. 23.1.2-11): Prehistory: Since the introduction of this aeroengine type 1971, 259 (!) similar failures occurred at the low pressure location bearing (LPLB). This correlates 0,924 failures per 1000 operation hours. This is about the 24-fold, compared with the bearing failures of the two competing OEMs. With this the bearing can be termed as the weak point of the concerned aeroengine type. In 7 cases it came to an oil fire, 6 times the fan shaft was overheated and fractured, with the current case 3 times (volume 2, Ill. 9.2-9). Usually the failures occurred a short time (within 400 opertation hours) after an aeroengine overhaul, respectively module overhaul. However there are cases, at which the bearing region was not opened during overhaul. The investigations gave no evidence for causative assembly anomalies.

During climb, shortly after the start, the warning light of the vibration sensor from the middle aeroengine (Nr. 2) flashed. Little time later, the oil pressure warning light displayed. A short time later, the oil pressure warning light of this aeroengine displayed. This is a sign for a clogged oil filter, which then changes over into the bypass mode. The pilot reduced in about 250 m hight the power of the aeroengine to idle. After that the oil pressure and the oilflow normalised.
Afterwards the pilot selected at about 700 meters hight the level flight. Than he increased, accordant the operation manual, the power of the aeroengine, to continue the climb. At about 3000 meters there was without warning a loud and observable explosion with heavy shaking of the airplane. At once many warning lights displayed. To these belong also displays for larger sections of the fuselage hydraulic system. The concerned aeroengine was shut down immediately and as a precaution the fire extinguishing system activated. Because of the obviously damaged steering control of the aircraft, the approach and the landing have been carried out with help of the aeroengines thrust.
The check of the aircraft showed that the fan shaft was broken. The fan has climbed foreward about 3 meters out of the aeroengine into the intake duct (detail above left). To this belonged also an about 1 meter long shaft piece. In the intake duct, fragments punctured the fuselage. This produced an about 2 x 2 meter big hole. The horizontal stabilizer showed further penetrations. Several hydraulic lines of the control system have been damaged and affected its function.
An investigation took place (sketch below) with the following results:
Obviously the origin of the failure was an axial thrust bearing, the so called Low Pressure Location Bearing (LPLB). This is an intershaft bearing. It is located between intermediate shaft and and low pressure (fan) shaft. The race rings rotate, accordant to the shafts, in the same direction. This ball bearing has a diameter of about 350 mm with balls of 27 mm. The bearing takes as thrust bearing the thrust loads of the fan. During start the inner ring rotates with about 7000 rpm and the outer ring with about 3800 rpm. The material of bearing rings and balls is tool-steel with 18 % tungsten, 4% chromium and 1 % vanadium (Fig. "Improvement of bearing materials"). The hardness of Rockwell C 40 is relatively low. The normal service temperature is 240°C, which makes the use of this especially heat-resisting bearing material understandable. The cage, riveted from two parts, consists of a low alloyed steel and is silver plated.
The oil supply (frame below) is indirectly carried out by one oil jet. About 60% of this oil is gathered by an rotating oil catcher ring and centrifuged into the bearing. Obviously important for the problems is, that the oilflow is directed past two flange connections.
The investigation allows the suggestion about the failure sequence, shown at the Fig. "Oil deficiency shaft failure sequence 2".

Probable cause of the bearing failure: The investigating authority comes from an unsufficient oil supply. Thereby, oil leaks at the oil convaying flanges play a role. So heat could no more sufficiend dissipated. The increased bearing temperatures caused a thermal expansion of the balls and the bridging of the clearances to the bearing rings. With this, the heat production rose further and the heating up to the failing of the bearing (Fig. "Oil deficiency shaft failure sequence 2").

Fig. "Oil deficiency shaft failure sequence 2" (Lit. 23.1.2-11): This illustration shows the probable failure sequence after the bearing already failed from overheating. Sparking occurs („1“) inside the rubbing labyrinth seals of the bearing chambers. After this the hydraulic seal failed („2”). Ignition of the oil inside the failng labyrinth („3“, volume 2, Ill. 7.2.2-4). The following oil fire spreads after the seal failure through the cooling air ducts to the fan shaft. This softenes and fractures („4”, volume 2, Ill. 9.2-9).

Fig. "Bearuing failure by overheating" (Lit. 23.1.2-8): The D x n data of the aeroengine bearings with its normally already high operation temperatures, lay at the technical limit. They don't indicate in time, symptoms of a failing (vibrations, chips) during overheating. The self-energising failure process with exponential rising heat production caused from increasing friction forces at the rolling elements to them race rings and the cage (Fig. "Bearing behaviour by guidance of the cage"), is extremely short (seconds, Fig. "Oil deficiency shaft failure sequence 1"). Also the failure mode usually permits no sufficient certain conclusions at the real cause of the bearing failure. For thrust bearings, the failure mode correlates the acting axial forces (see above).

Fig. "Shaft fracrure by rotation on the bearing seat" : A special danger exists, if the inner ring of a bearing rotates on the seat. In this case, from experience, it comes to an intense heat production. With this there is the danger of the softening of the shaft till it fractures. The following influences can trigger this situation.

Too large seat clearances: Problems with the production tolerances. This is true not so for the bearing itself, as for the bearing seats. Especially if they are reworked for repair and coated as wear adjustment.
From experience, an especially dangerous situation exists for vibration wear (fretting) of a shaft from a titanium alloy. Here a combination with the bearing rings from steel has shown as especially prone (volume 2, Ill. 6.2-17). So in short time (hours) it can come to a dangerous loosening of the seat. To avoid this, contact surfaces of shafts for bearings and casings from titanium alloys must be supplied with a wear protection coating (mostly tungsten carbide = TC).

Assembly: Often the anit-twist locking of a bearing ring depends from an axial tensioning. Is this not sufficient, the ring can get loose. This is supported from unfavourable heat expansions, e.g., during transient operation. A similar situation arises during an assembly. In such a case the ring has contact during axial tension at the face, and during operation it comes to a settling.
Too high bearing loads: High unbalances with appropriate elastic deformations of the shaft and high shock like bearing loads (Fig. "Skidding failure modes at races") can cause bearing rings to rotate on the seat. These are especially prone, which already during loads in the designed operation are exposed to loosening heat expansions and/or markedly elastic deformartions. This is supported by vibration fatigue (fretting) at the seats.

Overheating, caused from unsufficient oil supply and/or a race track failure, can lead to the expansion and rotation of the inner ring at the seat. Supporting acts a temperature caused bridging of the bearing clearence up to jamming of the rolling elements (Fig. "Skidding by unsuitable bearing tolerances") and/or of the cage with high braking forces (Fig. "Wear loaded sliding surfaces in bearings").

Fig. "Bearing failure by Interaction with accessories" (Lit. 23.1.2-12): About 1 hour after the start of the fourengined airliner, the pilot noticed during cruise vibrations of the of the fuselage and one aeroengine (No. 1) shut down itself independent spontaneous. Obviously all displays in the cockpit have been normal. The pilot landed auf at an alternate aerodrome. Thereby obviously both main shafts of the concerned aeroengine didn't rotate. However, a rotation would have to be expected during windmilling.
The aeroengine in question, had till the drop out like its accessory devices (sketch middle right), about 15 500 operation hours with about 2600 startcycles. It has an electronic control (Full Authority Digital Engine Control = FADEC). This control unit consists of several components (scheme middle left). To these belong the Electronic Control Unit (ECU) and the generator (Permanent Magnetic Alternator = PMA).
After the landing the stored data have been printed out, but showed no cause for the sudden shut off. Also a borescope inspection showed nothing special. The rotors could be rotated without problems. The oil filter in the accessory gear showed no chips. A rev up of the engine without ignition showed, that the high pressure rotation speed was markedly lower than to be expected. From experience this points at a failure of PMA or ECU. After the demounting of the associated computer, at rotor and stator of the PMA rubbing traces and burns have been identified. This are features of a distroyed bearing of the driving shaft from the generator. Unfortunately, in the manuals have been no instructions about the measurement of the clearance or the rubbing of the rotor.
Thereupon PMA and ECU have been exchanged, The testrun was repeated and the required high pressure speed was reached. After this, a test of the aeroengine followed, during which after about 10 minutes, again without prewarning, the engine shut down spontaneous. Also this time the dismounted PMA showed similar damages like the previous. Then the whole PMA mechanism was disassembled.
The ball bearing of the drive shaft to the generator, which is located at the end of the accessory gear, showed a crack in the cage (sketch below right). It could not be seen from the outside in the assembled condition. Then the drive mechanism was changed and a third PMA mounted.
The also changed ECU was checked from the producer and sent back at the operator as qualified for service.
The dismounted drive mechanism of the PMA went back to the OEM for investigation. Here, everywhere on the Balls at on 90° of the race track from the inner ring fatigue, out-breaks (pittings) have been found (sketch left below). The pockets of the cage have been worn, one was broken. The cause for the bearing fatigue was not found.
Already formerly problems with such bearings have occurred. At least 26 cases emerged. It was believed that the cause for these problems is identified as „infant mortality effekt“ (volume 1, Ill. 5.2.1.3-5). This lead to instructions (service bulletin) from the OEM. Bearings of an other producer which seemingly had proven better, should be assembled as fast as possible. This action is more then doubtful in the light, that bearings of both producers had failed.
Rather the conclusion of the OEM is important, that the bearing, especially the inner ring gets overloaded by radial forces. In several cases also corrosion was found. Both influences can trigger the fatigue of the bearing racetracks. The ball bearing runs at 160 °C service temperature and a rotation speed of 20 000 rpm. A hypothesis for the overload is thus supported, that at other locations of the gear, the same type of bearing is used but no failures occurred.
If the electric power supply from the PMA for the ECU drops out, obviously the existing type was not able to use the necessary replacement because of a deficit of the software. With this the shut down of the aeroengine can be traced back to the bearing failure.
Spontaneous shut down of an aeroengine after the drop out of the ECU is caused by the lacking current supply from the PMA. This was also observed at other types of airplanes, only with electronic control (FADEC). The investigating aviation authority corresponding concluded as following:

• Radial overload causes the fatigue failure of the thrust bearing on the drive shaft of the generator (PMA).
• Oil supply, design or unsuitable use, possibly also in combination, triggered the fatigue failure.
• As consequence of the bearing failure, the rotor of the PMA touched the stator. This produced a repeating short circuit. The electric current supply of the ECU from the PMA failed.
• A deficit of the ECU software prevented, that these used the substitution electrical current supply (on bord power system). With this the ECU dropped out and triggered the shutdown of the aeroengine .
• Rubbing traces of the PMA rotor with a clearance at the drive shaft, are dependable indications of a bearing failure. Because there are no hints at this in the manuals, the maintenance personnel could not evaluate those.

Fig. "Overhaul caused bearing failure" (Lit. 23.1.2-13 uand Lit. 23.1.2-14): During the failure, the fragments penetrated directly from the high pressure turbine through four holes.
The aeroengine had alltogether about 20 000 operation hours with 16 000 start/shut down cycles. The failure occurred less than 80 operation hours and 56 cycles after the aeroengine had to be repaired because of an ice strike. The following investigation showed, that the shaft in the area of the supporting intershaft bearing (bearing „A” in sketch below) was separated. The fracture is located at a weak point from design with bores for the scavenge oil flow.

The roller bearing „B“ is positioned in front at the HP turbine rotor. The fixing nut of this bearing serves as oil catcher ring and has oil bores to the bearing. These bores have been blocked by a dark substance.
In the overhaul manual of the OEM it is claimed, that a new silver plating (SPOP) of the bearing „A” inner ring makes it necessary, to remove before the old silver coating with (abrasive) blasting. Thereby in the process specification for the bearing races aluminium oxide size 500 is required. Above this, the OEM recommends aluminium oxide with the grain size between 200 and 500 or glass beads. The numerical data apply for the number of meshes in a certain sieve area. So a smaller number means a more coase blasting abrasive. After the stripping of the coating the nut must be cleaned and rinsed. Enquiries unfolded however, that at the nut of the failure, a mesh size of 120 was used.
The work sheets for the assembly don't point directly at a check of the the oil bore in the nut. However the following hint was prefixed.

„NOTE: BEFORE ASSEMBLY BE SURE PARTS ARE CLEAN AND BLOW OUT OIL PASSAGES OF OIL MOVING PARTS BEFORE ASSEMBLY“.