Aeroengine Safety

Figure "Observation of parts before assembly": The assembly personnel is the „last chance” to identify abnormalities, alarming variations and damages. In this phase it is still possible to carry out a test and if necessary an exchange of suspect parts with relative little effort and without high risk of a failure.

The larger experience, motivation and specialist knowledge the more possible it is to identify such abnormalities.

Figure "Slarming effects during clamping of bolts": Typical example for a highly safety relevant component with indications of unsufficient operation behaviour are bolt connections (chapter 23.3). The experience shows that an experienced technician can intuitive recognise on the trend of the torque moment if critical deviations exist. Such a case emerged when bolts of the rotor connection from a fighter aeroengine attracted the attention of a technician. Obviously the bolts lengthened unnormal plastically during tightening. So it was not possible to reach the prescribed clamping torque. This observation was reported. So it was possible to identify in time a shipment of bolts, not conforming to the specifications.

Naturally there are a multitude of further influences as cause of similar effects like wrong lubrication grease or a change of the thread coating (Fig. "Influence of pretension at fatigue" and Fig. "Influence at the bolting durability"). They can also be identified by an inspection.

Figure "Risk by corrosion media during overhaul" (Lit. 20.1-1): During the assembly used auxilary media (chapter22.4) play for the safety of the aeroengines a more important role than this could be seen at the first glance. Basically should count, that only media are used which are sheduled/approved explicit for the particular application by the OEM in manuals or specifications. Also change over to other supply sources, respectively an other product in an case of doubt the agreement of the OEM must be obtained.

In the following marking media are understood as electrolytes which are used with chemical (etching) or electricity assistance. Remain rests on a component/part these can at operation temperatures attack the material damaging. Markings occur during assembly procedures for example to mark a casing position, an executed special activity or the start of a sequence of rotor blades (Fig. "Marking method and location").

Unsuitable sulfur containing (MoS₂ ) lubrication media/antiseize agents can trigger sulfidation at operation temperatures. The attack can be shaped plane, as pitting or in the extreme case (e.g., under absence of air) crack producing (stress corrosion cracking, chapter 22.4). Thereby it is not absolutely necessary that this lubricant was knowingly. Also smeared respectively during the fitting of a bolt at susceptible components transferred remains (Ni alloys, Fig. "Significance of using approved lubricants" and Fig. "Unsuitable lubricant leading to fracture") can act dangerously.
Similar effects can also be triggered by machine oils with ingredients of chlorine and/or sulfur.

Rust remover/penetrating oil can act aggressive at certain materials/metals during a longer period of time and/or at elevated operation temperatures. This is especially true for MoS₂ containing rust removers.
When cleaning agents/degreasing media with clorine content are (e.g., „PER“ and „TRI”) extraordinary precaution is demanded. Those media can produce on the base material an extremely thin clorine containing reaction film. Under operation temperature and sufficient high tension stresses at titanium alloys stress corrosion cracking can be triggered above about 450 °C (volume 1, Ill. 5.4.2.1-8).

Hydraulic fluids can decompose at elevated temperatures and discharge aggressive products. In such a case brittle crack formation can occur at titanium alloys which are under tension stresses (Fig. "Contaminated Hydraulic fluid").

Elastic sealing compounds/pastes with a distinctly percentage of acid can deteriorate susceptible materials/plastics/metals. An example are filled elasomers likt they are used as rub in coatings in the front compressor region. The influence of acetic acid, which gets effective during the curing/hardening of sealing medium can lead to the deterioration of the coatings.

To the corrosive media of the assembly process does not at least count also the hand perspiration/sweat. It can produce on steels, especially pitch surfaces (races of anti friction bearings, gear teethflanks) corrosion pittings of which later during operation fatigue failures can start. This possibility exists especially then, when the parts are stored without corrosion protection after handling (Fig. "Gloves to avoid swet by handling bearings").

At parts of titanium alloys hand perspiration can trigger cracks at high tensile stresses above 450°C (volume 1, Ill. 5.4.2.2-6).

Figure "Conducting cables during the assembly": Also small sparks between a conducting cable and a part can produce serious damages (volume 4, chapter16.2.2.6). The cause are several each other intensifying effects, which can be traced back to a locally fused surface. Thereby even a microscopic small damage already suffices for

• embrittlement,
• tnternal tensile stresses,
• notch effect of the weld puddle,
• drop of strength as consequence of the heating and/or formation of an alloy with the conductor material.

So especially attention is demanded when handling with conducting cables (e.g., welding, hand-guided metal-cutting machines), also in the low voltage region (e.g., galvanic etch marking). Attention must be targeted on using only undamaged cables so that it can not come to an unnoticed electric contact with a surface of the part. Attention must be payed, that no cable connectors ot earthenings (welding, marking with sparks) will be damaged. In case of doubt the use of such a process must be approved by the OEM.

Figure "Marking method and location": Also during the assembly components/parts must be permanent marked. This can get necessary, for example to mark part positions or to document executed special measures. Precautionary it should be assumed that a marking/labeling rather decreases the fatigue strength of a part (volume 4, Ill. 17.4-1 up to Ill. 17.4-3). Insofar the markings may not be positioned at lifetime determinant highly loaded areas. Therefore the location to apply a marking on a component is exactly prescribed and must be met under all circumstances. Also it must not deviated from the specified marking nethod and its parameters unless it is approved by the responsibles, in case of doubt the OEM.

Figure "Causes for transport damages" (Lit. 20.1-2): Transport and storage during the assembly is essential. By an improper approach they can produce failures. Especially concerned are labyrinth seals and anti friction bearings. Act on an aeroengine or a module impact loadings, then on the races of the anti friction bearings plastic indentations can occur (Fig. "True and false brinelling" and Fig. "Causes of brielling on bearings"). In labyrinth seals of rotors with oil damped bearings (Fig. "Damped main bearings") because the lacking oil film during stand still, the sealing gap can be bridged and the filigree tips can be damaged. Both damages endanger the operation safety and are from the outside mostly hardly to identify. Thin walled respectively elastic casing struts can spring. So a contact between blade tips and rotor respectively casing is possible. This can as well deform permanent single blades as also damage susceptible rub coatings in the casing or on the rotor. At bent blades it can early come unnoticed to a rotating stall (volume 3, Ill. 11.2.1.1-1 and Ill. 11.2.1.2-6). Is this connected with an icreased vibration load, fatigue fractures with catastrophic secondary damages can not be ruled out.

Is the damage very large the chance exists, that it becomes noticeable with rubbing noise or vibrations during the turning of the rotors by hand or starter. Arises the suspicion of bumpy loads, often remains only the disassembly and the inspection of each single component.

Vibrations during the stand still of an aeroengine can „forge“ small indentations into the races of anti friction bearings (brinelling, Fig. "True and false brinelling"). Especially at the typical high-speed small gas turbines (e.g., for helicopters) this causes in a short time the sudden breakdown of the bearing. At large aeroengines exists the realistic chance, that early enough before a catastrophic breakdown the failure announces itself with chips at magnetic plugs in the oil system. However this does not protect against high repair costs. Also fretting wear, triggered by vibrations can cause the carving of the rolling elements into the races (false brinelling) and so in an extremely case shorten the life of the bearing. Dangerous vibrations can already occur at not sufficient elastic supported aeroengines/components if they are moved over concrete slabs, devided by grooves as they can be quite normal at shop floors. False brinelling can also develop by vibrations of the hall/shop floor on which parts/components or assembly devices are deposited without a springy pad.

Figure "Probles by blowing with pressurized air": To remove loose particles, dust, „unproblematic” fluids and chips from surfaces, out of bore holes and gaps, in shops willingly blast pipes are used. However this habit is problematic. So the question arises: Where will stay the blowed particles? Can they deposit on other, unprotected surfaces and promote there failures, e.g., as foreign objects or corrosion medium? Possibly they also will be not removed from cavities, instead blown even deeper. This can trigger blockages in systems for oil, fuel or cooling air.
Attention must be given also at the air jet. Unsufficient separators for oil and water in the line network can do more harm than benefit. At susceptible components even deeper (e.g., bearings, mating faces) water can trigger corrosion. Oil can change the sliding behaviour of components. For example by changing the clamping torque of bolts with a specified initial tension.

Figure "Danger by particles between flanges": Contaminated contact surfaces of flanges can lead to damages during the later operation or promote unbalances. Get hard particles like blasting abrasives or metal chips between the contact surfaces (sketch above) or snug fits/centering seats a close fitting will be prevented. During the push together galling will occur. With this the mating forces strongly increase and also prevent the required contact (Fig. "Failures of shrink fitting"). Carves the particle later during operation into the surface or is leveled by wear, the flange respectively the fitting surface will set. As consequences unbalances occur which show by vibrations. During vibrations particles obviously can move/wander between the contact surfaces and so create radial, score alike damages (`tiretracks', detail). Later those will afford at least an expensive repair.

Figure "Assembly as FOD risk": Little causes with the potential for a big impact. Locking wires from experience can be found once again as foreign objects respectively their impact marks. An explanation is the procedure during the securing process. Are the wires after the twisting too long they will be shortened. This was usually carried out with the same nipper which is also used for twisting. According to instructions thereby the other hand holds the end to be cut (sketch right), elsewise the wire end can jump away. For example, stands the aeroengine during assembly upright (Fig. "Assembly caused foreign objects" below) and is not covered, the jumping off wire end can fall into the engine. By experience there is only a little chance that those ends will fall out when the engine is upturned and shaked. Not before at the test facility then the wire piece will come loose and damage the blading. Caused by the notch effect it must be reckoned with dangerous high dynamic stresses at the always slightly vibrating blades. In an extreme case it comes to a blade fracture with the breakdown of the areoengine.

Figure "Assembly caused foreign objects": Not least the bath tub curve (Fig. "Determination of overhaul intervals" and volume 1, Ill. 5.2.1.3-5) as statistic trend of failure incidents gets its high tip at the beginning („burn in“, „infant mortality”). It is based on the increased likelihood that from the assembly foreign objects will remain (Ill. 20.2-7).

In the above illustrated case after an about 1 hour flight a compressor failure occurred, because a remained loose nut fom the intake area during exchange of the aeroengine was ingested from (volume1, Ill. 5.2.1.3-6 and Ill. 5.2.1.3-7).

The frame below shows situations which lead to assembly caused foreign object damages (Fig. "Assembly as FOD risk"). Expecially attention should be directed at aeroengines with with an intake at the side (in the frame below right). If this zone is not finally checked because of bad visibility, by touching for foreign objects at the engine during stand still stand still these can stay unnoticed.

Figure "Avoiding problems by weighing moduls and engine": Weight variations are of importance for the flight stability of some aircraft types. If necessary they must be readjusted on the aircraft, correspondent with the after the assembly documented weight and balance point. Thereto the aeroengine will be weighed. To identify beyond that unnormal deviances it is naturally a requirement to know as exactly as possible the expected weight.

Surely, this can not be a sufficient protection against FODs and deviations (e.g., components), but anyway a chance.

Further alarming indications for foreign objects can be a rattling when the aeroengine is pivoting. Also rubbing/grinding noises can hint at jammed foreign objects. At the floor space attention must be paid at falling foreign objects, which can count as an indication for a suspect assembly procedure.

Figure "Deviation from specified assembly process": This case is an impressive example for apparently bagatelles during the assembly of an aeroengine, which can lead to extensive damages. Concerned is a small APU (sketch above). After the change of the personnel which is responsible following the assembly, at several successive test runs cases of catastrohic anti friction bearing failures occurred. Thereby the rotor was so heavy damaged that this approximated a total loss. Investigations showed that during the first/initial run up obviously lack of oil at the front bearing lead to its overheating. After extensive enquiry it showed, that the special design of the oil circuit required from the beginning sufficient oil in the bearing area. Only so the in this APU type „rudimental“ oil pump a sufficient suction effect could develop. It emerged that for this an optimal axial hight of the `pump wings' (similar to gear teeth) is of high importance. They govern a sealing gap as small as possible. Before the change of the personnel (the experienced retired) the APU was till during the test run (mostly over the week end) in horizontal postion. So a sufficient quantity of oil which was filled during assembly could remain in the critical area. The new personnel stored the APU after the assembly in a vertical position (sketch down left). So the oil could drain and
the oil supply of the bearing failed. The for small turboengines typical extremely high rotation speed lead to a very early and fast failure of the bearing. The heavy damage rarely permitted conclusions. Only after several failures the „riddle” could be solved.

Figure "Safety by understanding failure mechanisms" (Lit. 20.1-4): Obviously it came in several cases to the development of cracks in stiff fuel lines (upper sketch). It can be supposed that fatigue cracks by vibrtations are concerned. They are promoted by a tensioning of the line (Ill. 23.5.2-6). Thinkable is that the tensioning was traced back to an unsufficient fitting accuracy of the line/tube connections and with this to the assembly. Also thermal expansions between the casing (e.g., hot) and the pipe line (e.g., cold) can lead to tensioning. Dangerous vibrations can be avoided with a clamp. However this was in that case obiously not successful. Therefore a change to a flexible line was necessary.

Figure "Influences of the stick slip effect" (Lit 20.1-5 up to Lit. 20.1-8): Just the assembly is more frequently than assumed influenced by the „stick-slip-effect“ (Fig. "Stick slip effect in maintenance and assembly"). In most cases it is responsible for the development of noises during the assembly and cooling down of an aeroengine. Known effects of other fields is the clicking of radiators during temperature changes or the crackle of an exhauist pipe during cooling down. Also the squeaking of brakes is an example from the daily life. It correlates the squeaking of dry sliding bearings/hinges or during the force fitting of snug fits.

However of safety relevant importance are other effects of the stick-slip-effect. During the actuation of actuator cables and feeback cables, as they are frequently used in elder type aeroengines jerky movements are the cause for triggering malfunctions of a control system. This is also true for actuartor cylinders with sluggish seals.

Loosen bolt connections (Ill. 23.1.1-5) under vibrations there is also an influence of the stick-slip-effect.

During rubbing events the stick-slip-effect triggers vibrations. After the same principle a violin bow gets a string to ring. In this manner rotor blades are excited to dangerous vibrations when they rub at casings (volume 2, Ill. 7.1.3-4).

In contrast the vibration damping action of the stick-slip-effects can be used advantageous, also if this is not always aware. Typical is the damping in the contact surfaces (volume 3, Ill. 12.6.3.4-10).

Figure "Stick slip effect in maintenance and assembly" (Lit. 20.1-5): Just because the stick-slip-effect emerges so often in the assembly practice, the understanding of its mechanism is so important. In the end it depends on changes of the coefficient of friction respectively the friction (diagram below left). The developing adhesion during stand still is markedly higher than the slide friction of the relative to each other moving surfacres. A further precondition is the elastic behaviour of the system. It can be described with a mass-spring system (upper sketch right, low see„A”), like it is presented by a compressor blade (above left).

During a rub in process the mass (blade tip) is deflected, caused by the adhesion against the spring regidity of the blade. The reset force increases till the adhesion is overcome („B“). Now the mass slides and the low slide friction acts („C”). This leads to a reduction of the reset force and it comes again to stand still with adhesion („A“). This sequence can be repeated very fast in high frequencies. Thus it is possible to excite dangerous vibrations of the blades (volume 2, Ill. 7.1.3-4).

Figure "Strength drop by galling seizing" (Lit. 20.1-9 and Lit. 20.1-10): Galling (seizure, „cold welding”) is a damage which can occur especially during assembly procedures.

Increasing danger for galling marks exists in the assembly during pushing together snug fits (Fig. "Failures of shrink fitting" and volume 4, Ill. 16.2.2.5-1).

Just materials used in the aeroengines like alloys of aluminium and magnesium, titanium, aistenitic CrNi-steels and nickel (superalloys) especially tend to galling.

Does a protective lubrication and or an oxide layer collapse, there will be welding processes in the microregion. Are those surfaces moving further against each other the surface will be ripped. The fatigue strengch decreasing scores respectively notches are formed (upper sketch).

Typical galling marks develop during operation at poor lubricated and/or overloaded gear tooth flanks (Fig. "Gear wheels failure mechanisms" and Fig. "Operation marks and damages on gears 1"). Bolts and nuts from austenitix steels can seize if poor lubricated (Lit. 20.1-14, Fig. "Jamming causes of bolts and nuts"). With this a further tightening or loosening is only possible with overloading. The risk of a seizing for example can be decreased for bolts and nuts made of Ni alloys with an preoxidation.

Also during loosening of burned in bolts and nuts seizing can occur. This leads not seldom to such high loosening torques that cracks up to a fracture occur. This is the reason which limits the reuse at hot parts.

The addiction to gallind/seizing for example is the reason why gears and bolts/nuts did not become accepted for aeroengines.

Illustraton 20.1-16 (Lit. 20.1-24): The connection of components, normally shafts and hubs, by shrink fitting is a usual and established process. Usually the shaft is left at room temperature or cooled down. This can be carried out corresponding to the assembly specification with dry ice (CO₂ ) or liquid nitrogen. Usually the hub of the shrinking-on part will be heated. All temperatures should be prescribed by the OEM.

The particular problem of the shrink fitting process is to guarantee the necessary/specified temperatures of the components during the slide on procdure. This depends essential from the span of time to the joining of the components. If too much time passed, so that the temperature of the hub lowers too much respectively the temperature of the shaft increases too much, there will be clamping during the slide on process. This reinforces, accelerating with the thermal conduction, if there is a too narrow contact between shaft and hub and/or a too slow sliding. With this the force for the sliding increases. Allow the assembly specifications for the critical joining phase discretionary decision (Fig. "Influence of shrink fitting with spline toothing 1"), experience and specialist knowledge is essential to avoid or identify of problems.

To the problems of shrink fitting belong :

• Unsuitable temperature („A”): Is the shaft not enough cooled, it will be again heated by contact with the hub, for example during handling or during sliding on, the mating forces can get too high, with an corresponding failure potential.An intense cooling of the hub during handling or slide on also lets the mating forcees increase.
  But a damaging high heating is also dangerous. Causes are carelessness or faulty heating (e.g., malfunction of temperature probes). Exceeding the tempering temperature of a hardened or tempered steel part (e.g., gears), the strength/hardness will drop unacceptable.
  Skewed/tilted attatching („B“): This danger icreases with the joining diameter. Basically an inclined located hub bore means, that that the projected hub diameter will be too little. So a clamping gets probable. Exists a local contact between bore and shaft, the temperatures will equalise because of the intense thermal conduction. The clamping process intensifies. With this the tolerable mating forces can be exceeded. As result the parts get deformed and/or damaged by galling. Occures a undirectional contact of the bore edge, this can carve damaging into the shaft (volume 4, Ill. 16.2.2.5-1).
  

• Contaminations of the joining surfaces („C”) can hinder the joining process, trigger the formation of scores, impair after the joining the distribution of the contact pressure or impress notches.

• Production problems: To meet the joining gap a suitable dimensional accuracy (diameter „A“, geometry, circularity) of shaft and hub is necessary. Also hight and orientation of the roughness (e.g., machining scores), can be of importance, especially at repaired surfaces. Also a role can play type and properties (hardness, tribology) of a coating. It is important that in the region of the hub bore („D”) or the shaft chamfer no burrs hinder the joining process.The consequences of an unsuitable or faulty shrink fit process can be very manifold (frame below):
  

• Drop of the strength of the component (e.g., gear) by intense heating (see „A“). This can add to operation deteriorations like wear (e.g., at a gearing) or crack formation by fatigue

• Plastic deformation of the components. On gears can be expected a local overloading of the tooth flanks/bad tooth contact pattern (Fig. "Gear wheels failure mechanisms" and Fig. "Tooth contact patterns at gears").
  

• Inclination, misalignment can cause unbalances with dangerous vibrations, fatigue fractures (Fig. "Fracture by resonance vibration of grars", Fig. "Damages from exitation by tooth flanks"), increased rubbing wear (Fretting) and failures of the gearing.- Seizing/galling grooves (Fig. "Strength drop by galling seizing"), notches can significant lower the fatigue strength and so trigger cracks and fractures (Fig. "Influence of shrink fitting with spline toothing 3").
  

• Loosening: Does the necessity of a high joining force prevent unnoticed the saturated axial contact, it can come to a loosening of the joining respectively an axial tensioning during operation by setting and slackening (e.g., bolting). The danger of an accelerated deterioration in interaction with extreme fretting wear exists (Fig. "Influence of shrink fitting with spline toothing 1").
  

Similar also can be expected, if as result of dimensional problems, distortion and deformation of the hub a non-uniform contact of the faces exists.

Figure "Influence of shrink fitting with spline toothing 1" (Lit. 20.1-24): The failed aeroenginem was investigated at the OEM. It arose, that the radial drive/shaft (`tower shaft') of the accessory gear was destroyed. In the gear casing lay the angular bevel gear with the cup washer and the tightening nut.

For the investigation of the fragments also the pieces of an other parallel case were available, which occurred about half a year later at the same aeroengine type. As a reference existed a radial shaft from a comparable aeroengine with about 9000 operation hours which suffered <U>no</U> failure.

The failed engines had since the last overhaul about the same operation time of 7000 and 8000 hours. The point in time of the overhauls of both cases differed about 4 years..

History of similar failures: At the date of the current failure 34 further cases of different versions from the same aeroengine type became known. All aeroengines showed in the failure area the same design (frame below right). Common feature of the failures was the fracture of the shaft (frame below left) as well as the separation of cup washer and tightening nut.

The OEM associated the failures with the following causative influences:

• Faulty sliding on procedure and unsuitable pre stress.
• Use of the cup washer as force fitting tool.
• Damaging of the roller bearing by the contact with the bevel during the assembly.
• Faulty assembly, deviating from the specifications causes the damage of the seats on the shaft and hub bore.

In the past 7 years the OEM tried to solve the problem with several temporally distributed measures in repair/assembly/overhaul and changes of the parts. To these belong:

• Improved joining procedure of bevel gear and radial shaft (tower shaft) in the overhaul manual. Use of a new hydraulic device to guarantee the axial fit (Fig. "Failures of shrink fitting"). This measure was carried out at the failed parts.
• Close inspection of the seat of the tower shaft during sliding on of the bevel at the hot parts inspection. The referee part showed also this measure in contrary to the failed parts.
• Use of a suitable medium on the roller bearing to locate the rollers during the assembly procedure.
• About 1/2 year after the current failure it was demanded in the overhaul manual to exchange roller bearing and bevel gear as a combination.

Failure analysis: In two cases three fragments developed . The position of the fractures is in the transition radius of necks and behind the spline toothing (frame below left). In both cases the end of the shaft thread was still stuck in the bevel pinion. After the disassembly the spline toothing as well as also the shaft seat of the bevel showed heavy galling marks(Bild 20.1-15). From those, at the side of the shaftend, a fatigue fracture was propagating. The galling marks were axial orientated (Fig. "Influence of shrink fitting with spline toothing 3"). A corresponding failure mode could be seen at the hub bore of the destroyed bevel.This points that they occurred during the pressing of the bevel on the shaft.
The flanks of the spline tootening showed features of accelerated wear.
The reference shaft also showed at the bevel seat extreme fretting marks (Bild 20.1-17.2). Striking have been circumferential orientated, rough seeming grinding scores. Because they ran transverse to the joining direction it can not be ruled out that they promoted in this case extreme seizing already at the beginning of the seat (during the beginning of the slide on procedure).
Damages on gears amd roller bearings were identifies as secondary failures.
This is also true for cup washers. they had loosened with the tightening nut and showed circimferential fatigue cracks and fretting wear on contact surfaces .

Conclusions:
The main failure influence/cause is the seizing/galling damage at the shrink fitting seats of the bevel gear and the radial tower shaft. From these fatigue cracks started into the shaft. Such failures are typical for a too narrow joint play. Both failed parts were joint with a hydraulic press after cooling the shaft with dry ice (CO₂ ) and heating the bevel gear (Fig. "Influence of shrink fitting with spline toothing 3"). Every delay of the joining procedure after the tempering can lead to a dangerous tolerance overlap as result of a temperature adjustment.
Thereby the adjacent cited, in the overhaul manual not enough specified warning about the maximum time promotes dangerous situations:

Citation: „The spiral bevel gearshaft must be seated as quickly as possible to ensure proper seating of the gear on the towershaft.”

Figure "Influence of shrink fitting with spline toothing 2" (Lit 20.1-24): Shrink fitting of the bevel pinion on the radial shaft (see failure in Fig. "Influence of shrink fitting with spline toothing 1").

Figure "Influence of shrink fitting with spline toothing 3" (Lit. 20.1-24): Appearance and arrangement of galling marks can quite give hints at causative influences. This in Fig. "Influence of shrink fitting with spline toothing 1" described incident mentions a failure with a shaft fracture (sketch left) and a not broken comparison shaft with similar operation loads. Location and distribution of the galling/seizing marks at the circumference of the bevel gear seat allow the following conclusion: The bevel was inclined attached for sliding on. At the contact to the cold shaft the bevel cooled locally faster and shrinked ahead of time. So it came to clamping and seizing. The position of these marks resided in the region of a pronounced stress concentration. This results from the notch effect of the transition radius at the contact and the change in stiffness at the transition to the bevel hub.

In the comparative case shown at the righ, the seize mark can be explained by a too short heating of the bevel gear. The heated through thinner hub shoulder of the bevel may already adopted a sufficient diameter. The thicker, slower bevel gear cross section was possibly still too cold. With this it came here during the slide on to high clampung forces and seizing.

Figure "Hearing cracks and failures" (Lit. 20.1-11): At this accident it came during the start to the breaking off of a aeroengine nacelle. Extensive investigations showed amongst others, that at least in two further cases during the mounting process of the engines the ripping of the attachements could probably heared (upper sketch, example 20.1-1). The maintenance personnel obviously could not interpret and assess the criticality of the noise.

This shows how important it is, that technicians know sufficient about the formation of assembly noises and operation noises as well as its interpretation.

During crack propagation, deformation (also elastic) of cracked cross sections and forced fractures soud formation and emission occurs. This effect is called sound emission. The ultrasonic percentage of such noises can be recorded with suitable sensors/probes and afterwards analysed. So cracks can be veryfied and located under favourable conditions (no background noise).

In case of a spontaneous crack formation or during critical crack propagation (forced fracture), especially at brittle materials and/or when the critical fracture toughness (K_Ic ) is reached, a sort of bang can occur (sketch below right). Certain materials with a fibre structure like wood or fabric produce a „ripping“ noise.

Crack growth and elastic deformation of crack zones leads in the case of a hearable sound range to screaming that means high frequency noise. The sound is primarily produced by the rubbing of the crack surfaces against each other (sketch below right).

Are metallic components/parts (e.g., sheet structures) deformed and rub thereby against each other, squeaking high frequency noises are emitted.
In every case an, for the particular situation abnormal noise formation during a maintenance process or an acembly process is an alarm signal. In no case we should go back to normal, before the harmlessness is not clarified.

Figure "Changes of components identified by noise" (Lit. 20.1-12): Besides noises which can be traced back to influence of violence with deformations and/or crack formation, there is a multitude which indicate other problems. They should be beneficial used.

Unusual rubbing noises, not typical for the situation can announce failures (Ill. 20.1-20). In the shown case an axial movement of the blades in the turbine disk (blade walking, volume 2, Ill. 6.2-5.2). In the advanced stage the blade shift can trigger a catastrophic aeroengine failure. So together with a shortened run down time an unusual noise during idle and running down of the aeroengine developed. These symptoms allow to take measures in time. Naturally temporary rub effects caused by the normal cooling conditions must be distinguished. This surely requires a certain type specific experience. Further alarming noises can derive from foreign objects which tumble inside the rotor (drum) during slow rotation. Correspondent click noises must be expected. Naturally an observer must have sufficient experiance to distinguish usual uncritical noises from danerous ones. Usual noises develop for example by the rattling of the clappers of fan blades during a slow rotation. In contrast noises of foreigen objects like nuts, bolts or tools which occur during slow rotation of the rotor (Fig. "Foreign object remained by assembly"). The noises appear depending of the formation of the foreigne object directly after the assembly or first after longer operation times.

Bearing failures can announce itself with a rough running during the rotor is turned.

Figure "Intercepting failures rub noises during rotor turning" (Lit. 20.1-13): In this case it came to the rubbing of the 2nd stage gas generator turbine wheel at the turbine stator in front (sketch right). This caused a heavy turbine failure with extensive blade fractures. In instructions(AD notes) the responsible safety authority claims to watch after the last flight of every day for rubbing noises during run down of the gas generator. Those checks are safeguarded by regular borescope inspections.

Figure "Seeing first signs of an arising catastrophy" (Lit. 20.1-25): If an aeroengine breaks off, this will lead to an accident with catastrophic consequences (volume 5, Ill. 10-10.1). The day before a spectator noticed after the landing, that one engine hang tilted on the wing. Unfortunately the meaning of this observation was correct evaluated not before the accident. Obviously at this time the mounting of the pylon at the wing had already failed.

Figure "Disk fracture by an unsuitable disassembly tool" (Lit. 20.1-15 and Lit. 20.1-16): The use of a not scheduled assembly tool/disassembly tool can lead to serious failures. To the shown example the Lit. 20.1-16 tells:

Citation: „The FAA tells the inspectors:' NTSB concluded that a locally manufactured metal rod was used (during maintenance) to remove the thermal shield (lower detail right) attached to the rear face of the disk (lower detail left), damaging the blade slot bottoms and leading to cracking and subsequent disk failure'.”

As investigations show a multitude of turbine disks of other aeroengines were also damaged and had been reworked. In several cases obviously crack formation occurred, however without catastrophic failure.

This damage in a high loaded disk region with an unsuitable tool obviously must also be seen in connection with the rework (blending out, Lit. 20.1-15). It emerged, that such a rework was no guarantee against crack formation, because of the danger of smeared small cracks.

Figure "Error risks by different systems of units": The interoduction of metric threads, at least in european designed military aeroengines can promote mistakes. The risk of an unaware interchange of metric bolts respectively nuts with „inch measuring parts“ increases if there are very similar bolts/thread dimensions (Fig. "Assembly caused fuel leak a nd fire").

A possible result is the loosening and/or the overload of the connection (Fig. "Operation loads of bolts"). Does this fail it must be reckoned wich considerably secondary failures.

Figure "Dangerous rubbing contacts of components" (Lit. 20.1-17 up to Lit. 20.1-21): The often extremely complex variety of pipes and cables at the outside of an aeroengine increases the possibility to touch each other (example 23.5.2-2). Because it must generally be asumpted, that the components vibrate, it comes at the contyst surface to wear movements (fretting). Even microscopic little relative movements are dangerous (chapter 23.5). Besides a weakening of bearing cross sections, also occurs depending of the material a drop in fatigue strength. Thereby the main danger is not a locally rubbed through pipe wall but the spontaneous ripping open or breaking of the pipe (example 23.5.2-3). That is the case if under high inside pressure a sufficient large fatigue crack developed. Especially titanium alloys are concerned of this effect (Fig. "Risk of titanium pipe lines 1" and Fig. "Risk of titanium pipe lines 2"). Those are increasingly used in pipe lines of aeroengines. This is up to the high strength in the undamaged condition and the low specific weight, compared with Ni-alloys and steels.
So basically during assembly and maintenance it must be strictly payed attention, that no such touching takes place. This requires, that only the exact scheduled/specified parts will be assembled (example 23.5.2-2). Also seemingly little differences as they are possible between variations of the same aeroengine type (Lit. 20.1-19), don't allow a substituted assembly if not explicit approved by the OEM.

Depending of the function of a line, from contact different secondary failures must be expected:
Pipes which guide flammable fluids (fuel, lubrication oil, hydraulic fluid) often are under high inner pressure, however are relatively thin walled. They are mechanically high loaded and elastic. With this they are subject to vibrations. Under contact it can come to rub through and ripping open. The flammable content sprays and so a high danger of a fire exists with the potential of catastrophic failures (example 23.5.2-3). Leaking hydraulic fluid can also trigger dangerous failures, even if not burning. If it gets on hot components, it can decompose and form aggressive residues which trigger in titanium alloys dangerous corrosion (Fig. "Contaminated Hydraulic fluid").

Will the electric insulation of current-carrying cables worn throug it comes to a short circuit (volume 4, Ill. 16.2.2.6-7). If there are sufficient strog currents even small sparks are sufficient to damage the wall of a metallic pipe line (volume 4, Ill. 16.2.2.6-2). Than the normal vibration load is sufficient for a fatigue crack and fracture (Fig. "Conducting cables during the assembly"). Escapes a flammable medium, it must be rekoned with the ignition by sparks. Are cables for electric/electronic test pulses of low energy concerned, at least the connected analysis system (control unit, monitoring) will be hindered in its function.

Not to underestimate is the danger which comes from a leaking respectively broken bleed air pipe/duct (e.g., cooling air, cabin air, Fig. "Dangerous situations by air duct leaks"). Bleed air from the rear region of a modern compressor has temperatures up to more than 500 °C. This can heavy damage other components and ignite fammable media. Does the fire detector system register hot air, it must be reckoned at least with a `in flight shut down' .

In older aeroengine types frequently actuator cables and feedback cables to the control unit can be found. When they are damaged, touching an other component, important control functions and steering functions can fail and trigger a flight emergency case.

Figure "Critical joining of big engine modules 1" (Lit 20.1-22): The principle of the module design was in many aspects an advancement (logistic, costs and expenditure of time). But it also implies special dangers. Especially the masses of big fan engines which must be handled (Fig. "Dangerous rubbing contacts of components"), in case of a the contact of susceptible parts (main bearingws, labyrinths, blade tips) during the pushing together of the modules can be unnoticed damaged.

In the shown case the low pressure turbine (LPT) came in contact with the turbine stator, mounted in the turbine centre frame in front (detail below). Thereby a vane segment in the 12'o clock position was shifted/displaced. After about 3000 operation hours the segment loosened as far that it touched the rotor blades. The consequence was a catastrophic turbine failure during the start of the airplane. The exit of fragments caused a particular closing of the start und landing runway of this large airport. This triggered extensive problems. The assembly problem was obviously previous identified by the OEM. Accordingly the assembly procedure was changed for all aeroengines. Additional the aeroengines where borescoped for damaged blades. Obviously the aeroengine of the described failure case „slipped through” this check, this means the damage was not identified.

Figure "Critical joining of big engine modules 2": In this case it came shortly after the start to the damage of a main bearing in the fan region with considerably secondary failures and the outage of the radial driving shaft (tower shaft).

A following investigation showed, that the roller bearing was obviously unnoticed damaged during the push together of the fan module with the aeroengine (upper sketch). The joining region could only badly observed during the assembly process and the heavy masses could damage the comparatively filigree bearing by canting of the cage (lower sketch). At this primary cause of the failure pointed roller indentations at the edge of the inner race. Those could only be explained with the sliding in of the shaft, but not with the drawing off.

Figure "Assembly problems by design traps": This is an example how the design can influence errors and risks during the assembly. Concerned is a main bearing of a fan engine in the region of the bevel gear to the radial shaft (tower shaft). If the very similar inner rings/races are slided on inverted, the land to slide over a transition radius on the shaft seat is missing (right sketch). So the inner race ring of the bearing can not lay against bevel gear like in contrast the left sketch of a correct assembly shows. Obviously the axial gap between inner bearing race and the bevel gear was so unremarkable, that it could not avoid an assembly fault sufficiently safe. The consequence was the damage of the radial power take-off.

Figure "How unbalances can develiop" (Lit. 20.1-23): Rotating components must be balanced, because too heavy unbalances can act in manifold ways damaging.

They load the rotating system with a synchronous circumferential tensioning. Such unbalances lead to unsymmetrical failures. The deflection and with it the rubbing processes of the rotor keep their location. Examples are an overload of the bolting, limited at a zone of the circumference or one-sided rubbing marks. Contrary unbalance caused failures can be identified by this feature.

Occur rotor vibrations which are not synchronous with the rotational frequency, failures by the rubbing processes must be expected, which are widely steady distributed over the whole circumference.

From an unusual heavy unbalanced rotor, normally strong vibrations are transferred by the bearings and casings to the whole aeroengine. This can lead to vibration fatigue with cracks and fractures. Especially casing struts to the bearing chamber and the bearing chamber itself are endangered.

Are heavy vibration accelerations triggered at the outside of the aeroengine, pipe lines, accessories (e.g., mounting flanges of pumps and starters/generators) as well as their mountings itself are endangered by overloads.

The elastic-damped bearing mount (chapter 23.1) of modern aeroenginetypes avoids during unbalances heavy engine vibrations at the outside. But just this can aggravate the identification of dangerous unbalances for the rotor components. So it can happen after the failure of a rotorblade (fracture in the middle of the blade), that the acceleration sensors/probes, usually fixed at the outside of the aeroengine don't respond, because the trigger threshold is too high positoned to avoid false alarms. However, the rotor and/or the bearings suffer a dynamic overload with the potential of a catastrophic failure.

The vibrations during unbalances can lead to rubbing wear (fretting). Especially at titanium components it must be reckoned with a dangerous drop of the fatigue strength. The vibrations can be so heavy, that it comes to the formation of extreme friction heat and to the fusion welding of the rubbing surfaces.

Unbalances can have very different causes. The illustration shows a summary without claiming completeness. A great deal can be already traced back at shortcomings respectively deficits before the assembly. Therefore for several exists the chance, that they will be identified respectively avoided during the assembly. Precondition is a sufficient experience and specialist knowledge.

Design caused unbalances are often based on an effective balancing adverse design. Are those shortcomings known, the possibility exists to avoid them.

Production caused by unbalances (distortion, dimensional aberrances) which are larger than usual, can show during the assembly in the joining forces. At blades for example an unusual weight distribution is remarkable (Fig. "Balancing rotor components 2").

Not at least there are also assembly caused unbalances. To these belong the canting and seizing/galling (Fig. "Strength drop by galling seizing") of seats. If therefore facing surfaces unnoticed don't fit a balancing can be quite possible. But during operation the vibrations and micro movements can trigger a sudden setting. The result are heavy vibrations which prevent a further operation.
A rather creeping effect can be expected, if there are contaminations between rotor flanges which will be indented during operation with a drop of the prestressing of the flanges (Fig. "Danger by particles between flanges").
Unbalances by foreign objects (bolts/nuts, tools) from the assembly are not the primary danger. More dangerous in such a case is the swirling of the foreign objects in a rotor drum. So the whole bolting can be damaged and fail with catastrophic consequences (Fig. "Foreign object remained by assembly"/-7.2).

Illustrations 20.1-28 and 20.1-29 (Lit. 20.1-23): The balancing can be carried out separate at the rotor. At a module design frequently the rotor which is mounted in the module will be balanced. This happens at the component parts like disks and rotorstages in specified sections during the assembly. Fig. "Balancing rotor components 2" shows usual balancing possibilities.

For the balancing at aeroangines special methods are approved. The sketches A,B,C show the fixing of small balancing weights at flanges.