In the following text, labyrinth damage is understood to be all flaws and failures that occur directly in labyrinths, as well as any damages where the causes can be traced back to a labyrinth.
Damage-causing operating factors:
A labyrinth usually consists of two ring-shaped components: the rotating labyrinth carrier with the sealing fins, and the static labyrinth ring, usually a housing component that carries the rubbing surface, which is often a coating (e.g. thermal sprayed layer) or a soldered structure (e.g. honeycomb seal). Operating influences can cause labyrinth damage individually or in combination (Fig. "Labyrinth operating behaviour"). This can lead to complex damages that cannot be dealt with by purely theoretical means (e.g. measuring resonances in dynamic cracks). Experience is therefore an important aid for working hypotheses, reproducing part tests, and proving preventive measures.
Typical labyrinth damages:
The bulk of labyrinth damage can be classified into groups according to the main causes of damage (Fig. "Labyrinth seals damages"):
Overheating damage due to rubbing:
The most dangerous damage mechanism in labyrinths is self-increasing rubbing. It begins with local melting and/or burning up of a labyrinth fin, which can lead to total destruction of the labyrinth and surrounding parts (e.g. shafts; Fig. "Labyrinth failure by rubbing"). In titanium alloy labyrinths, sufficiently strong air flow through the gap combined with high rubbing temperatures can ignite titanium fires (Ills. 9.1.2-6, 9.1.2-7 and 9.1.2-8). This can be caused by various operating influences and be exacerbated by poor design and/or improper material combinations. Poor constructive and material conditions must not necessarily have been present in the new engine. For example, in modular engines there is the possibility that aged rub coatings are combined with new labyrinth fins, which increases the probability of the labyrinth failing (Fig. "Labyrinth exchange problems").
A typical type of damage which should be controllable through proper configuration of the labyrinth, is thermo-mechanical crack initiation in the rubbing zone ( ). Sufficiently high cyclical stressing from centrifugal forces and thermal strain can cause cracks to spread from the labyrinth fins into the labyrinth carrier. The constructive design, strength distribution, and eventual unexpected high-frequency vibrations of the part determine whether uncontrollable crack growth followed by part failure will occur ( ).
Damage due to thermal gradients:
The heated rubbing zone expands and bulges. This increases rubbing on an inner labyrinth ring (Fig. "Labyrinth rings bulging by local heating").
Even those temperature gradients in the labyrinth carrier not caused by rubbing can cause thermal fatigue cracks (Fig. "Thermal fatigue (TF) in labyrinths") or act as mean stress to promote fatigue cracks due to high-frequency vibrations (Fig. "Acoustic waves incite labyrinth shafts").
Damage due to high-frequency vibrations:
If labyrinths and their supports are made from thin-walled structures, the leakage air flow and/or rubbing can cause them to vibrate at high frequency. Poor constructive design can cause a resonator to form in the labyrinth, which promotes dangerous vibrations (
There are several ways in which high-frequency labyrinth vibrations can occur (see section "Labyrinth vibrations" and table "Labyrinth seal vibrations"):
Rub coating damage:
Damage to rub coatings is less likely to be caused by the rubbing process than by other factors in the labyrinth environment. These include primarily oxidation and erosion of the coatings/honeycomb structures (compare with
Damage related to the mounting of the engine (Fig. "Labyrinth exchange problems"):
Modular engines (e.g. modern civilian turbofan engines and military engines) have many advantages for mounting and overhaul. However, they also have certain design-specific characteristics that must not be overlooked. They usually have connection/separation points in the bearing area. Roller bearings are used as connection points if the shaft is to be removed along with the inner ring of the roller bearing. In these connection points of the module, especially in the bearing and labyrinths, special care must be taken during mounting and assembly. Labyrinth fins can be damaged by forceful contact with the seal surface during axial mounting motions. If, for example, a newer module is connected to one that has already run a long time, potential distortions in the old parts can cause rubbing damage. Fig. "Labyrinth exchange problems".1 shows another problem that can be traced back to changed rubbing behavior in a combination of old and new labyrinth components.
Damage to other components:
Damage to rub coatings can cause consequential damages to other engine components. For example, if coating particles are removed by rubbing or erosion and carried away by the gas flow, they can have an erosive or corrosive effect on other components. Reactions with hot-part surfaces (diffusion, depositing, crack initiation), as well as external and internal clogging of the hot parts, must be prevented. The tendency to clog hot parts depends in part on the toughness and temperature of the molten particles that are created in the combustion chamber. Particle melts that react with the base material or protective oxide layers can cause these to break off during cooling (thermal strain difference between the deposits and the base material) and thus considerably increase the emaciation process.
If hard spalled particles from rub systems enter into bearing chambers (Example "Aluminum oxide particle contamination") and/or into the oil circuit, they can damage the bearings. Protective pressure differences (e.g. barrier air) prevent this type of damage in the bearing chamber area.
Example "Aluminum oxide particle contamination" (Ref. 7.2.2-3):
Excerpt: “Combination of aluminium oxide contamination and bearing loads has been identified as the cause of bearing wear found in some …. engines in service…
… electron-microscope examination of No. 4 and No. 5 roller bearings found evidence of aluminium oxide entering the bearing area and causing premature spalling of bearing surfaces. Aluminium oxide, an abrasive material, had been used… to coat air seal teeth adjacent to the bearing area in the pressurized cavity in the oil system. The material facilitates mating of air seal surfaces during initial run-in of the engines.
…engines have been removed after showing evidence of bearing distress caused by spalling of bearing surfaces. The engines ranged 40-400 hr use.
…A four week investigation found that flakes of aluminium oxide averaging 0.020th of an inch (0.5 mm) in size were coming off the knife edges of the seals, which keep the oil system pressurized, and entering the bearing cavity area…
Magnetic chip detectors and one master chip detector on the engines were used by the airlines to check the aluminium oxide flakes in an inspection procedure developed by… (the OEM) and the Federal Aviation Administration.
To rectify the problem, …(the OEM) is stripping the aluminium oxide coating from the affected area in all … engines, and bearing and seal clearances are being increased to provide higher tolerance to contaminants. `This will eliminate aluminium oxide contamination and increase bearing life..”
Comments: Normally, the pressure in the bearing chamber (Fig. "Bearing chambers labyrinth seals") and the seals should be adjusted so that the higher pressure in the bearing chamber causes the leakage flow to exit the bearing chamber and prevent particles from entering.
Naturally, the non-magnetic aluminium oxide particles could not have been registered by the magnetic chip detectors, as was erroneously reported. Instead, what registered were the fatigue-spalled fragments of the bearing tracks, which were caused by the impact of the hard oxide particles.
It is interesting that the damage occurred relatively soon after the engine was delivered. After this short time, the bond strength of the armor should safely prevent spalling during rubbing. This indicates a manufacture-related bond strength problem in the Al-oxide armor on the labyrinth fins.
It is unclear how increasing the bearing tolerances increases their life. Evidently, there is another contributing problem that is solved simultaneously with the removal of the armor. The increase in the size of the labyrinth gaps must have been necessary after removing the armor, since the poor rubbing behavior of uncoated labyrinth fins tends to result in catastrophic labyrinth damage through overheating (Fig. "Labyrinth failure by rubbing").
Figure "Labyrinth operating behaviour" (see also Fig. "Labyrinth seals damages"): The simplicity of a labyrinth contrasts with the multiplicity of damaging factors, which act individually or in combination:
Type of labyrinth: Rubbing behavior and seal effectiveness are determined by the basic design concept of the seal.
Geometry of the labyrinth fins: During rubbing, the geometry of the labyrinth tips determines heat development and dissipation, i.e. the “heat balance” in the fin. The seal effectiveness (leakage flow) also influences the heat balance in the labyrinth and therefore also the sensitivity to overheating (see also “Type of labyrinth”). This has a significant influence on the probability of hot crack and thermal fatigue cracks initiating (Fig. "Operating properties by labyrinth fin geometry"). The tooth geometry additionally influences potential crack growth (LCF- and HCF-dynamic fatigue) out of the damage zone.
Labyrinth fin material: The selection of the labyrinth material determines the susceptibility to damage. For example, since a labyrinth is a tribo-system, heat creation during rubbing and sensitivity to overheating depend on the material combination fin/rub coating. If the fin tends to collect material deposits (
If the fin material of several high-temperature resistant powder metal alloys tends to hot crack initiation, dangerous cracks in the rubbing zone can be expected. The susceptibility to hot cracks makes them very difficult to repair during later part overhauls using the usual fusion welding process.
If the leakage flow is strong enough, labyrinth components made from titanium alloys can ignite during rubbing and the resulting titanium fire can cause extensive damage.
The tendency of a labyrinth material to oxidize is important in order to ensure that the thin fin cross-sections are not unallowably weakened or worn down in combination with erosion.
The tendency to overheat during rubbing depends on the created friction heat, which in turn depends on the cutting effect of the fins into the coating. Rough armors made of thermally sprayed hard material coatings create very little heat and thus minimize the damage risk. This positive effect can be lost through a reworking of the coating, e.g. for dimensional accuracy. The disadvantage of these coatings is their tendency to spall. Their bond strength must be ensured by adhering to proven application and manufacturing parameters.
Direction of the leakage air flow and fastening of the labyrinth rings: The susceptibility to aerodynamically induced labyrinth vibrations depends on the location of the fastening flange/labyrinth supports in relation to the pressure gradients in the labyrinth. Damage due to labyrinth vibrations can cause the labyrinth rings to fail (Fig. "Acoustic incited resonance of labyrinth seals").
Geometry of the labyrinth rings: The heat balance and the stiffness of the labyrinth components determine the thermal strain and deformation during normal operation and/or rubbing (Fig. "Labyrinth rings bulging by local heating"). The tendency of self-increasing rubbing to result from high thermal mean stresses and thermal cycles must be considered during the constructive design phase.
Rub coating: In the tribo-system of a labyrinth, the rub coating must be suited to the fin material in order to avoid unallowable heat development and material deposits. The bond strength of rub coatings consisting of thermal spray coatings can, in the new part (Fig. "Labyrinth exchange problems"). Erosion sensitivity also increases as coatings age.
Coating particles must not damage other components, such as ball bearings (damage to the tracks) or cooled turbine blades (blockage).
It is important that no unallowable sparks are created during rubbing in labyrinths in areas where there is fire danger (titanium parts in the air flow, bearing chambers).
If the ignition temperature (about 1500°C) is reached and air flow is sufficiently strong, a titanium fire can occur and cause extensive consequential damages.
Damping: Relative motions of labyrinth fasteners cause friction losses and have a damping effect on oscillating labyrinths. This effect can be optimized by targeted installation of friction dampers:
If wear or deformation decreases the lay-on pressure of a damping ring during operation and causes the damping to fail, dynamic cracks may occur.
Manufacturing and proving procedures: Naturally, the probability that a labyrinth will fail also depends on the quality of the new part as it has been manufactured and tested. This includes the failing of thermal spray coatings. The bond strength and rubbing properties of these coatings must usually be ensured by exact adherence to proven manufacturing parameters, since serially applicable, reliable, destructive testing of the coating is not sufficient on its own. Repaired fins (e.g. fusion welding process) must have suitable rubbing properties and not be susceptible to hot-crack initiation.
Damage to coatings (rub coatings and fin armor)):
Clearance increase: Erosion due to material removal or foreign particles (e.g. from the barrier air or in the leakage air flow, Ref. 7.2.2-25). Erosion can be caused or promoted by oxidation (aging) of the rub coating or individual components of the coating.
Coating spalling: High-frequency vibrations of the thin labyrinth ring fatigue the coating`s bond zone and cause local spalling. This process is caused or promoted by manufacturing flaws (Fig. "Labyrinth vibrations at high loaded parts") that lower the bond strength. Large-scale spalling is caused by insufficient bonding (Fig. "Shaft vibrations by labyrinth pressures") and/or thermal strain (Fig. "Dangers by delamination of rubcoating")
Fin armor spalling: Hard-material armor (usually ceramic) on labyrinth fins can spall due to manufacturing-specific bond weaknesses and/or overloading during rubbing (Fig. "Labyrinth fins damages").
Damage due to overheating:
Creeping: Air friction combined with insufficient leakage air flow can cause the rotating labyrinth ring, which carries the fins, to reach operating temperatures that, together with centrifugal forces and heat strain, lead to creeping deformations.
Bulges and distortion: If the rubbing zone of the inner labyrinth ring bulges out, it increases rubbing (
Material build-up: Under poor rubbing conditions, fusion of the rubbing products with the labyrinth fins and/or the rubbing surface promotes self-increasing rubbing (
Crack initiation and strength losses in the labyrinth fins: The friction heat of the rubbing process flows into a zone limited around the circumference. Thermal strain in the heated zone is blocked by the cooler surrounding material. This creates high compressive stress in the heated zone, and it is plastically compressed. When the compressed area cools, high tensile stress is created, which causes cracking of the weakened grain boundaries (hot cracks,
Fractures and cracks due to dynamic fatigue:
Various factors can cause labyrinths to vibrate at high frequency and fail ( ). These dynamic cracks are caused or promoted by increased mean stress due to temperature gradients (Fig. "Thermal fatigue (TF) in labyrinths").
Cyclical thermal gradients (e.g. during start-up/shut-down cycles) cause thermomechanical crack initiation. This damage mechanism is especially pronounced in conical labyrinth carriers.
Figure "Labyrinth fins damages": Characteristic, load-specific damage symptoms in labyrinth fins:
The depth of the wear and the circumferential length of the wear zone depend on the radial infeed motion. Infeed is caused by local distortion, shaft deflection, imbalances, or vibrations. Wear can occur along the entire circumference or only in a limited area. Typical symptoms in a rubbing zone are burring (“B”) and tarnishing or increased oxidation. These characteristics also indicate damaged sections of material (e.g. decreased hardness/strength; “D”). If the grain boundaries in these zones soften or melt, even small tensile stress can cause hot cracks. These cracks occur individually or in crack fields (see also “H”). These cracks advance (“C”) due to sufficiently high LCF- (thermal fatigue) or HCF-loads (labyrinth vibrations).
If material deposits remain on the fin (“E”), they can increase the rubbing process to dangerous levels.
Erosion particles in the shape of spalled rub coatings or removed material put erosive stress on the fins (“F”). The symptoms of this are similar to “peppering” (the impact points of the individual particles are still recognizable).
Spalling or local delamination of hard-material armor (e.g. Al2O3) is a common type of damage (“G”) and should be considered in connection with manufacturing flaws and/or handling mistakes (transport, mounting).
Not even armoring provides absolute protection against hot cracks (“H”). This is especially true if the armor is not sufficiently cuttable (which may be indicated by the coating roughness) which causes too much heat to be created in the rubbing zone (also see “C”).
The labyrinth fin materials that, for example, are exposed to the hot gas flow due to the leakage air in the turbine area, must be sufficiently resistant to oxidation. In oxidation-sensitive, highly heat-resistant materials or unsuitable welded deposits, oxidation damages have been observed that are similar to the orange-peel effect on the intake edges of overheated turbine blades (“I”). A typical symptom are many small cracks due to thermal fatigue caused by start-up/shut-down cycles.
Figure "Labyrinth thermal spray rub coatings damages": Damages to rub coatings in labyrinths are very similar to damages to rubbing blade tip surfaces (Fig. "Damages at housing rub coatings"). A special characteristic is the erosion of particles in the shape of a circumferential groove. Particles trapped in the labyrinth chamber run around the circumference and erode it until they are carried out with the leakage air flow.
Figure "Labyrinth failure by rubbing" (Ref. 7.2.2-7): An important, but infrequent, damage mechanism leads to catastrophic failure of labyrinth seals. The damage usually begins with light (not necessarily heavy) rubbing of one or more fins. If no rub-in due to abrasion or grinding-in occurs, material deposits are smeared onto the outer rubbing area and build up in fractions of a second. This causes the rubbing to increase reciprocally. The heating-up of the rubbing zones causes the weakened rotor part to bow and increases the damage. Often, a ring section will crack axially and break off in the direction of the labyrinth carrier circumference. The affected fins determine the width of the fragment. In connection with this, structures that resemble thermal spray coatings can form on the rubbing surface. This type of damage is most likely if the abradable behavior of the rubbing partners is poor, e.g. if the rub coating has aged to an unallowable degree throughout the operating time (Fig. "Labyrinth exchange problems").
Figure "Labyrinth damage by foreign object": If a foreign object enters into a labyrinth gap, it can cause the catastrophic damage mechanism depicted in Fig. "Labyrinth failure by rubbing". If the damage is not too advanced, the foreign object can be identified during inspection. Unaffected oxidized fins indicate that the labyrinth functioned for a long period of time without problems. Typical burring shows the rubbing zone of the affected labyrinth area (compare with Fig. "Labyrinth fins damages"). In a metallographical cross-cut of the “bridge” between the labyrinth fins and the coating, one could recognize layered deposited wear products. Proper location of the cross-cuts enables conclusions as to all or part of the foreign object.
Figure "Labyrinth exchange problems" : After replacing an old module with a new one, there is a potential typical problem if the labyrinth components were combined.
In a new machine, the good rubbing behavior of the undamaged seal surface ensures unproblematic rub-in of the seal ridges (above diagram). After long operating times, the worn-out seal gap is widened so much that no further bridging of the gap occurs. However, if a new seal ring with correspondingly long seal ridges rubs with an old embrittled seal surface, it can cause a self-increasing damage mechanism (Fig. "Labyrinth failure by rubbing") with extensive consequential damages.
Figure "Labyrinth rings bulging by local heating": A typical damage mechanism with the potential for catastrophic consequential damages is bulging of labyrinth rings in the rubbing zone (top diagram). The diagram at left can explain this process:
The rubbing process locally heats the ring. The heat strain in this zone is impeded by the cold surrounding parts (“A”), creating high compressive stress in the heated zone, which bulges outward corresponding to its ring-shape (“B”). This further bridges the clearance gap and increases rubbing. If the locally plastically compressed labyrinth ring begins to run freely again, high tensile stress is created by the cooling process and, if it exceeds the strength of the material, cracks are initiated. If, during the cooling phase, the grain boundaries are still doughy, they only have minimal strength. A hot crack is created. From this hot crack and/or the ring zone under high tensile stress, start-up/shut-down cycles and/or high-frequency vibrations can cause dynamic cracks that result in the part fracturing and failing (Ref. 7.2.2-10, Fig. "Crack zones in labyrinth carrier").
The bulge-sensitivity of labyrinth rings depends especially on the ridge width of the ring (“s”) and the height of the labyrinth fins (“h”; middle and bottom diagrams). The constructive characteristics that minimize hot cracks (middle diagrams) increase bulge-sensitivity (bottom diagram). Therefore, the designer must make a compromise.
Figure "Operation behaviour by labyrinth clearance": Gap size is an important factor for the operating behavior of a labyrinth:
In general, the probability of rubbing occurring increases as radial gap size is reduced. Heating-up of the usually thin, rotating inner labyrinth ring causes distortion with bulges (Fig. "Labyrinth rings bulging by local heating") followed by deflection due to imbalances. If the gap is large, the rubbing area is small. All rubbing, including wear, potential material build-up, and heat creation occur in this small circumferential zone. Therefore, large gaps promote catastrophic labyrinth failures due to self-increasing rubbing (Example "Disimprovement at labyrinth", ). With small gaps, rubbing occurs around the whole gap circumference, putting less thermal and mechanical stress on the labyrinth.
A small amount of leakage air with a small gap provides a good seal in the labyrinth that improves engine efficiency. Consistently high pressure gradients corresponding to the constructive design keep bearing loads low. This additionally avoids damage due to hot gas incursions in turbine labyrinths and ensures that the cooling air flow is not disrupted. However, the smaller volume of leakage air also means that less cooling occurs, i.e. the labyrinth becomes hotter due to agitation losses (air friction).
Larger gaps increase the possibility of the gap air causing aerodynamically-induced high-frequency labyrinth vibrations (see from Fig. "High frequency vibrations of labyrinth cones" on).
In Ref. 7.2.2-12, it is stated that: “the available energy in the leakage flow is considerable and has often caused aeroelastically excited seal vibration with subsequent fatigue failure.”
Erosion in labyrinths due to the circumferential chamber flow is promoted by small gap sizes, since erosive particles (e.g. wear products) are flushed out of the chamber by the leakage flow.
Figure "Low pressure turbine labyrinth damage" (Ref. 7.2.2-13): The depicted shaft-power helicopter engine evidently had several problems with labyrinths in the area of the power turbine (low-pressure turbine) during the design phase and early serial use.
The intermediate-stage labyrinth overheated during rubbing due to the gap being too small. This caused the inner rotating labyrinth ring to expand, leading to catastrophic damages (see also Fig. "Labyrinth rings bulging by local heating") where the rotating labyrinth ring was split.
The attached detail diagram shows the mentioned labyrinth configuration with a relatively small inner stiffening brace. However, engines in use today have considerably strengthened inner stiffening braces, as shown in the bottom diagram. It was also reported that the labyrinth gap was increased in the new engines. After these changes, no further damages occurred.
The exit labyrinth (see also Fig. "Crack zones in labyrinth carrier") failed due to vibrations of the static labyrinth carrier and was outfitted with a circumferential damping bandage (see Fig. "Low pressure turbine labyrinth damage") to remedy the problem (top detail).
Figure "Crack zones in labyrinth carrier": This diagram shows fatigue cracks on the static labyrinth ring of the exit labyrinth of the power turbine (low-pressure turbine). The cracks initiated in two exposed areas:
The labyrinth fins (“1”) in combination with hot cracks due to heavy rubbing, and the outer stiffening brace (“2”). The damping bandage shown in Fig. "Low pressure turbine labyrinth damage" has proven to be effective against dangerous high-frequency vibrations.
“The start of extended range twin-operations (ETOPS) testing of the …(civil big high bypass engine type) is to be delayed by `two or three weeks', says the engine maker, 'after a seal crack developed in the 90,000lb (400 kN) thrust' at which the engine is being certificated. A strip-down inspection after the test revealed a crack in a part of the seal arm which …(the OEM) says was already earmarked for modification…(the OEM admits) the test is currently operating to an N1 (fan)speed restriction. The restriction will be lifted when the modification, which involves thickening of the seal arm, is in place.”
Comments: The thickening of the labyrinth ring and the limiting of the rotor RPM indicate that crack initiation was caused by overloads due to operating stresses consisting of centrifugal force and/or heat strain. It is most likely not rubbing-induced crack initiation in the seal fins.
Figure "Crack initiation model at labyrinth tips": Grain boundaries usually have a lower melting point than the inside of the grain. Therefore, they are more likely to crack at the high temperatures typical in rubbing zones. These hot cracks occur primarily during cooling due to the high tensile stresses in the area of the previously plastically compressed heating zone.
Figure "Labyrinth ring dynamic fracture by fretting": During rubbing, hot cracks initiated in the fins of a rotating labyrinth carrier in a middle-performance class civilian high-bypass turbofan engine. Fatigue cracks spread from the hot cracks due to low-cycle fatigue during start-up/shut-down cycles. After the cross-section of the labyrinth carrier was split through, a circumferential crack formed in the membrane. This caused parts of the labyrinth carrier to fly off.
Cracks in rotating labyrinth rings do not usually lead to catastrophic damages. They do however, have the potential for causing serious damage and should be taken seriously. Therefore, remedies should be developed and implemented as early as possible.
Excerpt 1 (Ref. 7.2.2-14): “…The FAA has issued an airworthiness directive ordering the inspection of….(big high-bypass) engines for cracks in the high-pressure compressor (HPC). The FAA says 10 HPC spools have been found with cracks. 'The HPC stage 2-6 spool may develop cracks in the stage 3-4 and 4-5 interstage seal teeth that could propagate into the stage 3-4 and stage 4-5 spacer arms, aft the seal teeth', it adds.
The FAA itself has determined that the most likely cause of the cracks, which it says could cause an uncontained failure if not corrected, is the formation of zones of excessive heat. These are produced when the rotating seal teeth rub into the static honeycomb material, says the agency.”
Excerpt 2 (Ref. 7.2.2-19): “…(the OEM) believes that the problem is understood from an engineering perspective and that the cause of the HPC seal-tooth cracking is 'insufficient clearance between the rotating seal and its mating stationary honeycomb shroud '…(the OEM) has also confirmed metallurgically that the crack 'is stable and its propagation rate is not increasing with time.' Additionally, …(the OEM) believes that the seal cracking problem can be effectively monitored through periodic inspections until such time when revised hardware is installed during a normal engine shop visit.”
Comments: As can be seen in Fig. "Labyrinth fin crack by rubbing", the labyrinth rings are an integral part of the compressor disks/rotor. For this reason, it is possible that axial cracks could spread into the disks (see also Fig. "Labyrinth cracking in turbine disk" and Chapter 7.2.3) causing parts of them to break out. Inspection of the open crack surface did not show any accelerated crack growth, but also did not rule out consistent or slowed crack growth. The concept of a stable fracture only rules out spontaneous failure of the part. Evidently, on basis of a strength assessment of the affected area, the OEM does not feel it necessary to take immediate action.
Figure "Labyrinth fin crack by rubbing" (Example "Stable labyrinth fin crack"): The first six stages of the compressor rotor of the pictured large turbofan engine are made of titanium alloys. It is especially problematic that the labyrinth rings are integrally mated with the disks. This poses the danger of axial cracks spreading into the disks and creating uncontainable disk fragments.
According to Ref. 7.2.2-15, the forward part, “spools 2-4”, consists of the titanium alloy Ti17, matched to the operating temperatures. The following two stages, “spools 5-6”, are made from Ti6242.
Titanium alloys have a combination of especially negative properties with regard to labyrinth fins:
For these reasons, titanium-alloy fins of rotating labyrinth rings must be matched to the partner surface. To this end, armoring plays an important role in minimizing the amount of heat that is created and transferred into the labyrinth fins.
Example "Insufficient pressure gradients at the seal" (Ref. 7.2.2-17): “…The latest incident occurred on 2 October while engine No 900103 was around 100h into a certification block test…(the OEM) says that, during runs at redline core speed, '…a problem was encountered with the high pressure turbine midseal. Investigations following removal from the test stand show that the aft arm of the midseal was opening at temperatures exceeding design intent. Design changes have been identified to modify slightly the midseal arm configuration and cooling to bring the arm temperatures within design intent.'…(the OEM) declines to comment on the nature of the damage caused by the HPT-midseal problem, but program sources say that there was some damage to the casing. It is not believed that there was an uncontained failure of the turbine casing. It prevents recirculation of high pressure air inside the engine, which is at a higher pressure than the gas flow passing through the turbine blades. The midseal which was likely to be rotating at more than 1,000 RPM at the time of incident, is cooled by air bleed from the high-pressure compressor (HPC). It is though that this supply of HPC air will be modified to ensure that a positive pressure gradient is maintained on both sides of the midseal….”
Comments: The improvements to the cooling of the rotating seal ring and ensuring a sufficient pressure gradient, as well as the indication of housing damage enable the following conclusions to be made concerning the damage sequence: due to insufficient protective pressure gradients at the seal, hot gas invaded and overheated the rotating seal arm (principle corresponds to Fig. "Low pressure turbine labyrinth damage"). The arm must have become so hot that its strength was insufficient to resist the centrifugal forces and it failed (see Fig. "Labyrinth cracking in turbine disk"). The fragments were evidently contained by the housing. The given figure of 1,000 RPM at the time of damage does not seem very plausible. It is more likely to have been 10,000 RPM.
Figure "Labyrinth cracking in turbine disk": Integrated labyrinth rings on rotor disks, as are typically used in cast turbine disks in small gas turbines (top left diagram), but also find use as forged disks in large engines, are subjected to high centrifugal force-induced tangential tension. If rubbing causes cracks to initiate in the labyrinth fins, they tend to spread in an axial direction due to the cyclical loads from start-up/shut-down cycles (detail “A”). Labyrinth arms should be mounted on the rotor in a location that ensures the cracks will be guided in a direction parallel to the disk surface at the transition to the disk. This allows the arm to break off in small segments and prevents the axial crack from spreading into the disk. In this way, the extent of the damage remains controllable and does not result in catastrophic disk failure.
Figure "Labyrinth seal damage affects flight safety" (Refs. 7.2.2-13 and 7.2.2-19, Example "Disimprovement at labyrinth"): The seal is located between the high-pressure turbine (“A”) and the low-pressure turbine (“B”). Damage was found on the rotating inner seal ring, which sits on the rear shaft stub of the high-pressure turbine (“C”) and braces against the roller bearing “D”. The outer opposite part of this seal ring is mounted on the low-pressure turbine disk. It is also in the shape of a seal rotor with seal fins and runs opposite to the seal stator “E”, which is mounted on the inner shroud of the turbine stator vanes. It is not known if the low- and high-pressure shafts run in opposite directions. If they do, then very high relative speeds (rubbing speeds) can be expected. In any case, this configuration causes both seal rings to rotate at different RPM and the play and gap behavior to be controlled by two different shaft systems. These are then affected by heat strain, vibrations, and centrifugal force-induced expansions of both systems. Because the seal stator “E” is fastened to the low-pressure turbine stator assembly, elastic deformation, thermal distortion, and creeping strain affect the clearance behavior in this type of configuration. Additionally, the radially far inward location of the seal (probably to keep rubbing speeds controllably low) causes large ring-shaped chambers between the seal stator “E” (attached to the stator assembly) and disks, which can easily be made to vibrate due to aerodynamic forces (Fig. "Acoustic waves incite labyrinth shafts").
This intermediate stage seal resulted in several serious damages, which were especially dangerous since the affected engine type was also used in single-engine tactical aircraft, causing an immediate risk of crashes.
The damages were evidently caused by hard-to-control rubbing (see Fig. "Secondary damages by labyrinth cracking").
Excerpt 1 (Ref. 7.2.2-16): “…USAF will accelerate replacement of turbine seals in …(fighter) power plants following recent uncontained failures in two aircraft… Although those accidents are still under investigation, the Air Force accelerated the seal replacement because the failures may be linked to well-known 3-2 turbine-seal problems in (several derivate types of) the engines. These power …(fighters and bomber) aircraft, respectively.
The seal problem is limited to new engines and generally occurs within the first 200 hrs. of operation. The problem is related to engine clearances. They are tight in new powerplants and in some instances can promote overheating of the knife-edge seal's teeth. If this occurs, heat can be conducted by the ring to which the seals are attached, causing the ring to bulge and separate from the engine. After about 100 hrs. of operation, clearances increase and the potential for overheating diminishes.
…Besides the two incidents now under investigation, there have been only four confirmed…engine failures due to the seal. More than 2,800 of the powerplants have been manufactured. About 300…engines worldwide are affected by the seal problem…“
Comments: The increased clearance after about 100 hours is due to the fact that less intensive rubbing has already increased clearance through material removal. This is indicated by the low number of damage reports relative to the number of potentially endangered engines.
About 3 months later, the following report was published:
Excerpt 2 (Ref. 7.2.2-20): ”…Based on preliminary data, Air Force officials now theorize that increased clearances combined with potential imbalances in several components may cause cracking in rotating `teeth' of turbine seals found in the …(3 Types) family of engines.
The seal, which is located at the rear of the high pressure turbine, consists of three rotating knife-edge disk-like structures surrounding the aft shaft of the engine. The knife-edge structures, also called `teeth' rub or `bite' against a surrounding ring made of honeycomb material. Eventually they cut a channel in the honeycomb which rotates at slower speed than the knife edges. Engineers originally believed that the knife-edge structures or `teeth' failed because of thermal instability and high stresses. The answer to the problem was to increase the separation between knife-edge structures and the honeycomb. Clearances were increased twice since the problem first cropped up in 1988. Engineers investigating the seal problem have had difficulty diagnosing it for several reasons. First cracks in the knife-edges are virtually undetectable during inspections because honeycombmaterial transfers onto the `teeth' during the seal break-in period. Another difficulty is that the seal failure results in heavy damage to the powerplant. `We are never able to recover any suspect parts when these things failed,' an engineer investigating the failures said. Engineers got a break in mid-September though when a cracked knife edge was discovered during inspection of a powerplant…test of the engine with the cracked `tooth' have enabled the Air Force and …(the OEM) to develop a preliminary theory as to why the component cracks. The key to developing the theory was `we found that only half of one of the knife edge (structures), the one nearest the front of the engine, was rubbing the honeycomb' an Air Force official said. Using this finding, 'we determined that several conditions were necessary for the failure of the knife edge. First there needs to be an eccentric rub between the knife edge and the honeycomb. This can be caused by several things - a slight imbalance in the knife edge (structure) or a small imbalance in a nearby support bearing' the official said. Another condition necessary for a failure is increased clearance between the knife edge structure and the surrounding honeycomb. `When clearances are small, the knife edge makes 360-deg. contact with the honeycomb, even if there is a slight imbalance. And once the channel is cut it keeps the knife edge centered. `As clearances between the knife edge and the honeycomb are increased, a slightly unbalanced tooth could shift to one side and get off center. It would than fail to make 360-deg. contact with the (surrounding) honeycomb, and the resulting stresses could cause the tooth to
crack during the engine break-in cycle or during early operations,' the official said. If this failure theory is accurate, engineers reason that powerplants already having tighter clearances will not break a tooth. …Despite the belief that they may have discovered why three toothed turbine seals failed in low time … powerplants, officials are proceeding cautiously….'By delaying commitment on how to fix the seal they will have time to run tests that can verify there current failure theories,' an official said.”
Comments: This description partially contradicts that given in the first report three months earlier. It is especially interesting to note that the play increase was precisely the wrong corrective measure (Fig. "Operation behaviour by labyrinth clearance") and served to further increase the risk of catastrophic labyrinth damage. During rubbing, local overheating and bulging became more frequent. Unbalances were created and the intensity of the rubbing was accelerated. This indicates a special instability of the rotor shaft and its bearing supports. It is also noteworthy that the problem was known for about six years and the wrong measures had been taken up to this point. This also explains the caution in naming the damage causes before they were verified in engine tests.
The following article was published one month later:
Excerpt 2 (Ref. 7.2.2-21): “ The U:S. Air Force and …(the OEM) have reached agreement on a plan to return some 500 grounded powerplants to flight status….Two corrections will be made to return the engines to flight. First, maintenance teams will install new aft shafts with an improved damper. According to USAF officials, the new shaft will eliminate a `wire damper' between the first and the second tooth in the affected seal and substitute a `sleeve' damper.Tests have demonstrated that even if cracks appear in the teeth of the problematical seal, the sleeve damper will prevent propagation. This was not the case with the wire damper. Along with new shaft, maintenance personnel also will install new honeycomb rub strips in the seal area to reduce clearances between the strips and the seal's three teeth. These strips will return the three-toothed seal clearances to levels found in…(one engine type) family engines manufactured before early 1989. These earlier powerplants, of which there are about 1,000, are known as `population one engines.'… Other corrective actions at that time increased the cooling air to the seal area and modified the manufacturing process for the seal's honeycomb materials.
`In the retrospect, we now believe the seal failure… (in the first bomber) was caused only by thermal problems and by processing problems with the seal's honeycomb material,' an Air Force official said. As a result, one of the corrections implemented in the wake of the …engine loss - opening up the seal's clearances - created the current problem.”
Comments: This is a classic case of “disimprovement”. The first corrective measure, increasing the labyrinth gap, made the problem especially bad.
It is also an example that shows that it is easiest to discern by looking at the measures taken, how the causes were assessed and evaluated. Evidently the honeycomb structure of the rub-tolerant coating (material? wall thickness?) was not suitable for the special operating conditions in this labyrinth, especially rubbing. The existence of a suitable honeycomb is evident by the measure taken to return to the original small labyrinth gaps. This may have been necessary in order to prevent hot gas incursions and/or the strong leakage air flow causing aerodynamically-induced labyrinth vibrations.
The improved cooling of the labyrinth indicates that the temperature levels in the damaged area were generally seen as being too hot and that the probability of dangerous self-increasing rubbing (Fig. "Labyrinth failure by rubbing") was especially high.
The introduction of a damping bandage (see and Chapter 7.2.3) instead of a wire damping ring explains why the damages could not be caught early enough by crack inspections. Evidently, high-frequency vibrations of the labyrinth caused extremely fast crack growth from the labyrinth tooth that had been cracked during rubbing. Because the damage had already increased to such a degree, it is understandable that no analyzable fragments were found for such a long period. The change of the bearing at the aft shaft stub to an oil-damped configuration indicates, that the shaft system was also made to vibrate heavily and/or had a causal effect on the labyrinth vibrations and subsequent damage.
Figure "Uncontained labyrinth fracture": This damage is an example of the complexity of labyrinth loads and the multiplicity of factors relevant to the damage. As was later discovered, the main cause of the labyrinth failures was not a material flaw in a certain production batch, but aeroelastically-induced vibrations of the rotating labyrinth ring (compare with ). However, it is also plausible that the dynamic cracks originated in damage zones caused by rubbing. Reference 7.2.2-5 contains a diagram of the affected labyrinth, in which axial crack initiation in the third labyrinth tooth due to cyclical thermal gradients is depicted (detail “A”). The cracks described in Example "Disimprovement at labyrinth", however, occurred in labyrinth tooth four (detail “B”). It is understandable that a crack zone lying at the edge of the ring is subjected to high stress from flexural modes in the ring. It is worth noting that damage often resulted in fragments escaping, although the damage is far inside the engine, enclosed by the turbine guide apparatus, and the rotating ring is filigreed. It can be assumed that the ring failure in the almost completely enclosed duct was followed by extensive consequential damages that resulted in fragments escaping.
Excerpt (Ref. 7.2.2-8): “… Seal fractures between the first and second stages in the high pressure turbine section of ….(a high bypass civil fan engine type) engines were responsible for the redesign. Almost 280 of the 48,000-lb.-thrust engines that are in service….(on different two engine aircraft types) could be affected by the redesign. Since..(about one year ago) there have been four rotor seal fractures resulting in in-flight engine shutdowns. All incidents involved …(one special aircraft type).The latest (showed) …Damage to the engine and aircraft cowling was recorded in this incident. Aircraft damage also was reported in one other incident earlier this year. In all four failures the aircraft was able to land safely. Inspections have uncovered rotor seal cracks in seven additional engines, two of which were on …(the other aircraft type). …(the OEM's) investigation into the problem as centered on the relationship between the seal design and the metallurgical properties of the forging lot from which the seals were made. All of the fractured seals were produced from the same forging lot of a high-strength, nickel base alloy….However seals from other forgings also have developed cracks. This other cracking coupled with the inability of… (the OEM) to duplicate the failures in laboratory investigations prompted the redesign.
The current seal design consists of four arrays of knife-edge seals mounted between the first and second stage of the engine's high-pressure turbine. The failures have occurred in the rearmost array. The redesign would eliminate that array and require removal of some associated hardware…As of last week, seals had been replaced in approximately 36 of the 55 engines affected by the bulletin.
Comments: If, at first, the damage causes were assumed to be a combination of the seal design and a suspect material lot, even though evidently seals from other lots also had cracks, it can be seen as a fundamental violation of a basic rule of damage analysis: If a fact (even a seemingly secondary one) is contrary to a hypothesis, the hypothesis should be assumed to be refuted.
It is interesting, that attempts to reproduce the damage in a laboratory were unsuccessful. This is necessary in order to verify suitable corrective measures.
A constructive change as a remedy is loaded with risk if no technical tests were able to verify its effectiveness and instead, performance during subsequent operation is used to verify its suitability.
Excerpt 2 (Ref. 7.2.2-9): ”…The redesign eliminates the fourth knife-edge seal array in the second-stage high-pressure turbine seal (detail in Fig. "Thermal fatigue (TF) in labyrinths"). The fourth knife-edge had been cracking and was responsible for four in-flight engine shutdowns. Modifications to eliminate the fourth seal in existing engines will be done at maintenance centers normally used by the aircraft operators.“
Excerpt 3 (Ref. 7.2.2-10): ”…'The AD main detail reads as follows: “FAA has determined that fatigue cracks attributed to high cycle fatigue (HCF) can originate on the tip of the fourth knife edge of the second-stage HPTR airseal, and may propagate to fracture, which could result in an uncontained second-stage HPTR air seal failure…'
…Although the mechanism of failure is not established, analysis indicates the most probable cause of the HPTR seal rear knife-edge cracking is attributed to an aeroelastic instability. The actual work necessary is outlined thus: `This amendment adopts a new AD which requires the second-stage high pressure turbine Rotor (HPTR) four knife-edge air seal…A modification to the turbine cooling air system which reduces the number of metering bolts in the first stage turbine rotor assembly and increases the airflow to the HPT cooling air duct is required'.”
Comments: It can be assumed that weak points or cracks due to rubbing were present in the labyrinth fin from which the damage spread. High-frequency vibrations led to crack growth. The measure by which the temperatures in the affected engine area were influenced through cooling air admeasurements serves primarily to lower the risk of dangerous rubbing occurring.
Figure "Thermal fatigue (TF) in labyrinths" (Ref. 7.2.2-4): This compressor exit labyrinth of a first generation civilian high-bypass engine has several problems in the rotating labyrinth carrier and the thin-walled static parts. The damages were promoted by thermal fatigue cracks. Aside from the start-up/shut-down cycles, there were also sudden power increases in the engine from idling to full start thrust. Computer calculations proved to be extremely helpful aids to this damage analysis.
Figure "Acoustic waves incite labyrinth shafts": Air vibrations play an important role in the dynamic stressing of labyrinths due to resonance. In order to make this more understandable, the following section covers the fundamentals of air vibrations.
In gases, longitudinal vibrations/longitudinal waves occur, where regions of increased pressure alternate with low-pressure regions (bottom diagram). These pressure vibrations correspond to sound waves. For comparison, a transverse wave (e.g. water wave) is depicted below.
The axial vibrating movement of the elements of a longitudinal wave is the reason why the circumferential movement of the air in the labyrinth noticeably affects the vibration process. Depending on its direction, the speed at which the wave travels can be accelerated or decelerated by the rubbing forces in the ring-shaped labyrinth chambers.
Both traveling and standing waves can occur. The top diagram shows, with the aid of a coil spring, how an axial shock on the left side of the spring travels as a wave. No material is transported by the wave, only the vibrating movement is through the medium with the traveling speed of the wave. This vibrating condition is called the phase of the vibration. Its speed is referred to as the phase speed. In gases, it is usually the same as the speed of sound. If part of a labyrinth seal has a honeycomb, then the phase speed is lower than the speed of sound due to the effect of the open cell structure (Fig. "Honeycomb seals damping effect"). If pressure waves are created in a flowing medium, the spread out with the resulting speed from the phase speed and the flow speed.
Standing waves are created by the overlaying of two waves with identical phase speeds in opposite directions. In this case, the wave does not move as a whole, even though the individual wave elements are constantly vibrating. Also, the individual elements do not vibrate with the same amplitude as in an traveling wave. Instead, the amplitudes are enclosed by two sine curves. Between two neighboring nodes (wave at rest) all wave elements vibrate in the same phase.
For dangerous vibrations of labyrinth components to occur, it is important that the wave is opposite these components, i.e. a standing wave relative to the surface. Correspondingly, the static or rotating labyrinth part is then endangered by resonant vibrations (see ).
Figure "Labyrinth gas vibration systems": An air column can work in various directions, similar to a spring in a spring-mass system, and cause the system to vibrate at resonant frequency. The larger the chamber volume and the diameter of a connecting duct, the lower the spring stiffness of the air column. The vibrating mass is primarily limited to the connecting duct. Therefore, the more voluminous the air column is, the larger the vibrating mass. The example in shows that, both in a ring chamber between a disk and labyrinth, as well as in the main gas flow, various pressure waves are created and can reinforce one another.
A “sound room” consists of two small enclosed chambers that are connected by an air column. The air-filled chambers act as springs. In engines, these configurations appear around cooling-air lines, aeration vents, and pressure equalizing lines.
The “Helmholtz-Resonator” consists of a relatively large chamber that is openly connected with the mass of the opposite side, with an air column as a spring. These configurations appear in engines in the main air flow at support covers in the housing.
If a small chamber, i.e. a high spring stiffness and a large air column volume are connected by a narrow duct, then it is referred to as a “Boys-Resonator” (right diagrams, see Fig. "High frequency vibrations of labyrinth cones", Ref. 7.2.2-22). This configuration corresponds to ring ducts in intermediate-stage seals or end labyrinths of the compressor and turbines. The relatively small ring chamber is connected by a small ring gap with the main air flow, which acts as a vibrating air column.
Figure "High frequency vibrations of labyrinth cones" (Ref. 7.2.2-13): This free turbine in a civilian aft-fan engine experienced vibrations in the region of the turbine exit labyrinth. The vibrations occurred due to mechanical resonance caused by pressure gradients spinning around with the rotor RPM. These pressure differences at the circumference were caused by the eccentric spinning of the rotor due to the constantly present imbalances, as well as the flexibility of the rotor itself and in the bearings. The described effect was not related to acoustic resonance.
The spinning pressure differences caused the labyrinth to vibrate. This resulted in extensive crack initiation and failure of the static labyrinth cone (top right diagram). Although the support cone has a circumferential stiffening metal ring riveted along the middle, the fatigue cracks evidently oriented themselves along the rivet holes. Altering the design of the transition from the support cone to the labyrinth ring, as well as applying a “Z-stiffener” to the fastened end of the support cone (top detail diagram), prevented damage, but was very elaborate and expensive.
The final remedy against the dynamic fractures was a simple thickening of the wall of the support cone, which also inexpensive. Evidently this took the part out of the resonance zone.
Figure "Air pressure vibrations in labyrinths" (Ref. 7.2.2-22): This diagram gives an idea of the typical air vibrations in ring canals. The middle diagram illustrates the principle behind a resonator that is formed by the main gas flow, a ring canal that is limited by the disk and labyrinth, and a connecting ring gap (see Fig. "High frequency vibrations of labyrinth cones"). The bottom diagrams depict two examples of flow caused by air vibrations. There are intensive gas exchanges across the ring gap. The above diagrams show typical pressure distribution in ring chambers with vibrating air columns.
Figure "Acoustic incited resonance of labyrinth seals": In this schematic diagram of a labyrinth with static outer and rotating inner rings, high-frequency vibrations of the labyrinth rings are caused by air vibrations in the seal clearance gap.
In the rotating shell-like labyrinth ring, the waves run in or against the direction of rotation of the rotor, depending on gyroscopic forces. However, in the static ring-shaped labyrinth components, the wave is stationary. The top diagram uses an example of a vibration with three nodal diameters (see Fig. "Air pressure vibrations in labyrinths") to explain the resonance of the inner labyrinth ring with the pressure wave in the gap. In this case, the pressure wave and the inner ring are rotating in the same direction at the same speed, which means that the pressure wave stands relative to the surface of the rotating inner ring (see Fig. "Acoustic waves incite labyrinth shafts"). This fulfills the requirements for resonance between the pressure wave and the inner ring. The bottom diagram shows resonance with the outer labyrinth part. In this case, the pressure wave moves at the same circumferential speed against the rotating inner ring, so that the pressure wave stands relative to the static outer ring and causes it to vibrate correspondingly.
Figure "Mechanical labyrinth resonances with rotor": Labyrinth vibrations can leave typical periodic wear patterns of the rubbing surfaces along the circumference (bottom diagram, compare Fig. "Labyrinth vibrations at high loaded parts"). For this to happen, the antinodes of one ring must remain stationary relative to the other ring. This makes it possible to identify labyrinth vibrations on basis of the wear patterns. Depending on whether the vibration
moved in (top left diagram) or against (top right diagram) the rotating direction of the inner ring, wear marks can be found evenly spread around the circumference of the inner ring or the static outer ring (bottom diagrams, see Fig. "Acoustic incited resonance of labyrinth seals"). The number of wear marks around the circumference indicates the number of nodal diameters and therefore the type of damage-causing vibration (Fig. "Air pressure vibrations in labyrinths", Ref. 7.2.2-23).
The bottom left diagram depicts a typical labyrinth configuration in which the damage in the right diagram can occur. The causal mechanism is described in Fig. "High frequency vibrations of labyrinth cones". “A” is a softly suspended static labyrinth ring, and “B” is the corresponding rotating labyrinth ring.
Figure "Vibrations of labyrinths by aeroelastic instability" (Ref. 7.2.2-12): This is a small turbofan engine used in ground-combat support aircraft and business jets. The depicted seal system is located at the forward side of the first high-pressure turbine disk. The inner seal is fastened on the low-pressure side, the outer seal is fastened on the high-pressure side. If they are sufficiently stiff, seals attached to the low-pressure side are usually safe from vibrations aeroelastically caused by the leakage flow. The damping wire of the inner labyrinth should also serve to prevent these vibrations. However, during a testing-rig trial during engine development, the front tooth of the inner labyrinth seal had a dynamic crack initiate (bottom right diagram with detail). In the first variation (top right diagram with detail), the damping wire was attached at the very front. Since this also resulted in problems, a change was made to the variation in the bottom left diagram.
Tests of the variation in the bottom right diagram showed that, at various leakage rates, vibrations with three nodal diameters occurred in the inner seal (see Fig. "Air pressure vibrations in labyrinths"). The pressure vibrations were in phase with the deflections of the rotating labyrinth ring. The tests also showed that seal rings which are only affixed on the high-pressure side can only be made to vibrate by the leakage air flow if the mechanical frequency of the ring was greater than the acoustic frequency of the air vibrations.
Conversely, labyrinth rings affixed on the low-pressure side can only become instable if their mechanical frequency is lower than the acoustic frequency (see Fig. "Labyrinth design").
In order to stabilize the ring of the inner labyrinth, in addition to the damping ring, an inner damping bandage (held fast by centrifugal force) was also installed (bottom left diagram). This reduced the vibration stress by about 40%. This variant has proven itself over many years and millions of flight hours.
Figure "conical labyrinth carrier dynamic fracture" (Ref. 7.2.2-13): Thin-walled support cones of labyrinths are often subjected to high dynamic loads from pressure waves running around the circumference at the same speed and direction as the rotor RPM (“spinning modes”, also see Fig. "High frequency vibrations of labyrinth cones"). These are caused by uncentered rotor rotations, due to remaining imbalances, for example.
A typical example of this are the conical supports of labyrinth seals in main bearing housings. Vibrations of the parts cause typical, pronounced, arced, and branching dynamic crack initiation and spalling (right diagram). The depicted damage scenario occurred because the wall was thinner than the minimum allowable thickness. Strengthening the wall of the cone and/or stiffening the design of the transition to the labyrinth ring are proven measures for preventing this sort of damage.
Figure "Damage progress after vibrations by gas flow": In this case, the housing wall (“A”) which carries the supports of the main thrust bearings failed. The low-alloy steel wall was weakened by corrosion dimples.
The failure of the housing wall caused air to rush in (“1”), evidently changing the pressure ratios around the neighboring cone that braced the labyrinth on the housing of the main thrust bearing (“D”). The altered leakage flow probably provided conditions that caused the labyrinth to vibrate (“3”; Fig. "Labyrinth design"). It is not clear whether the vibrations were caused aeroelastically or by mechanical resonance (e.g. with the rotor RPM).
As a result, a dynamic crack initiated in the extremely thin wall (about 0.7 mm) of the labyrinth cone (“2”). The crack spread around a large section of the circumference along a constructive notch (spot-welding).
This created the danger of pressure changes in the housing (e.g. in the ring chamber “C”) overloading the main bearing axially.
Figure "Vibrations of stationary seal supports" (Ref. 7.2.2-23): This labyrinth at the compressor exit (stator assembly “A”, compressor disk “B”) of an early engine type had dynamic fatigue damage. The vibrations were caused by resonance of the flexure of the thin-walled stator part with the rotor RPM.
Circumferential crack initiation (“1”) along the stiffness jump to the flange occurred in a first version without riveted reinforcement (“3”).
Axial cracks in the riveting of a retro-fitted strengthening bandage (“4”). This emphasizes the large stored vibration energy and the minimal change to the natural frequency of the seal stator due to the strengthener “3” riveted to its free end. Otherwise, the friction between the riveted bandage and the part surface should be sufficient to dampen the vibrations.
Surface spalling (“2”, compare with Fig. "conical labyrinth carrier dynamic fracture") occurs following pronounced “shell vibrations”.
Using constructive changes, such as installing additional damping, to solve vibration problems has the advantage that it generally dampens all types of seal vibrations. On the other hand, using stiffeners alone (even if they were sufficiently strong, unlike the recommended local stiffener “2”) improves the vibration behavior with regard to only one vibration inciting mechanism (e.g. resonance with rotor RPM). Stiffeners must be used in the correct order of magnitude, and are only useful if the damaging vibration incitation mechanism, as well as other possible vibration incitation mechanisms, are safely determined.
Figure "Compressor damage by labyrinth vibrations": Dynamic fatigue in the soldered compressor stator assembly made from CrNi 18/8-steel and used in a small gas turbine. The source of the damaging vibrations should be sought in the labyrinth on the back side of the radial compressor disk of the second stage. Crack initiation occurred in three exposed part zones:
Transition from the labyrinth cone to the stator assembly.
Transition from the labyrinth cone to the labyrinth ring.
Transition from the soldered fastening bolt into the stator assembly wall.
Aside from the high dynamic loads, the cracks were promoted by the relatively low dynamic strength of CrNi18/8 steel, stiffness jumps, as well as the usually external damper being located in the labyrinth cone.
Example "Sensitivity to high-frequency vibrations of blisks" (Fig. "Vibration prone 'blisks'", Ref. 7.2.2-24): “The U.S. Air Force has halted testing of engineering and manufacturing development versions of the…powerplant for the…fighter, following an apparent compressor seal failure. Investigations are underway to determine why a seal in the first stage of an EMD test engine's compressor ( an integrally bladed rotor in the fourth-stage position of the engine) apparently failed, releasing two small pieces of the seal material into the powerplant's flowpath and damaging the compressor. The failed knife edge seal is on the backside of the first stage of the compressor. The seal design consists of an arm that extends reward from the compressor disk. At the end of the arm are two teeth that rub against the honeycomb shroud of the stage's stator, creating an interstage seal. The purpose of the seal is to prevent air leakage between stages. Initial inspection of the powerplant at AEDC showed damage to be limited to the blading of stages 5-9 of the engine- the high-pressure compressor. Inspectors also found that the engine was missing two pieces from the rearmost tooth of its fourth-stage knife edge seal, each measuring about 1 in. circumferentially by about 1/4 in. axially.
WSAF officials suspect the problem may have been caused by high-cycle fatigue cracking. If so, the problem could be remedied - at least enough to allow EMD tests to resume - by installing a damping ring on the backside of the seal. This is essentially a wire ring that would attenuate high vibratory stresses. `All the …(engine's) seals were manufactured with a groove in place to accommodate such a ring.'…Another possible fix could be open up the clearances between the honeycomb rubstrip and the seal's rearmost tooth.
Comments: The sensitivity of blisks to incitation of high-frequency vibrations is due to the lack of damping in the blade roots, which are usually set-in in other disk designs. It s not clear whether the vibrations were incited by the labyrinth itself. The application of a damping ring to titanium alloy parts, as in the above case, is problematic. The desired friction between the damping ring and the labyrinth arm can cause damage through friction corrosion, lowering the dynamic strength of the titanium material to 30% of its original value. Therefore, this type of ring is only suitable if appropriate surface treatments (strengthening, coating) of the ring and/or labyrinth groove can ensure that unallowable damage is prevented.
Figure "Vibration prone 'blisks'" (Example "Sensitivity to high-frequency vibrations of blisks", Ref. 7.2.2-24): Modern engines, especially in tactical aircraft, use blisks in the compressor rotor stages. The integral (i.e. forged with the disk, not connected via flange) rotating seal parts are especially sensitive to vibrations. Depending on the height of their natural frequency, vibrations can be incited by the natural frequency of the disk, combined blade-disk-vibrations of the neighboring rotor stage, or the natural frequency of the neighboring rotor blades. The integral design means that vibrations of the rotating seal parts are only minimally damped. The low damping of blisks relative to disks with set-in blades promotes the creation of high vibration amplitudes and correspondingly high incitatement levels. With blisks, therefore, the possibility of labyrinth vibrations being incited must be carefully checked as early as the design phase.
Figure "Labyrinth vibrations at high loaded parts": Rubbing causes high temperatures to be created in the labyrinth teeth and the supporting ring. These damage the material in various ways (strength loss, crack initiation, tension residual stress, embrittlement) and dangerously lower the dynamic strength. This makes normally tolerable labyrinth vibrations to initiate dynamic cracks. The distribution of the dynamic cracks indicates the type of vibration (compare with Fig. "Mechanical labyrinth resonances with rotor"). Recognizing the vibration type and its possible incitations is a prerequisite for designing specific remedies.
Figure "Shaft vibrations by labyrinth pressures" (Ref. 7.2.2-25): Rotors can be made to vibrate by pressure ratios in labyrinth chambers that are caused by stator vibrations and change periodically in circumferential direction over time. The intensity of the inciting forces increases with labyrinth diameter, flexibility of the stator, and the pressure at the labyrinth. These vibrations result in, for example, heavy material removal from the labyrinth with corresponding clearance increases and accelerated leakage rates. In the quoted reports, the given example is the intermediate-stage labyrinth of a cantilever-supported turbine rotor. The bearing was affixed to the compressor exit side by a thin support cone. The overhanging shaft was deflected by lateral forces from the uneven pressure distribution around the circumference. The remedy was stiffening of the bearing support.
The typical inciting mechanisms are mentioned in connection with labyrinth seals:
If the static labyrinth ring (top diagram) vibrates so that it becomes noticeably torqued and the gap changes periodically at the intake and exit, the leakage rates in these areas will vary. This causes zones of high and low pressure to form in the circumferential ring chamber around the fins (compare with Fig. "Labyrinth design to be avoided").
The bottom diagram depicts a case in which axial movement of the rotor relative to the stator led to uneven gaps at the ends of the seal with corresponding pressure changes. This can also result in radial forces that deflect the rotor.
These axial movements with uneven gaps around the circumference can be caused by vibrations of the rotor and/or the stator. This is the case if, for example, flexure of the entire rotor or stator is accompanied not only by radial movement, but also by varying axial movements of the labyrinth diameter opposite.
An additional influencing factor may be an uneven axial gap around the circumference, e.g. due to uneven coating erosion or distortion of the stator.
It has also been observed that the greater air friction in zones with higher pressure creates corresponding circumferential forces on the rotor, which can cause it to deflect. However, this deflection occurs offset from the radial deflection in direction of the circumference.
Figure "Dangers by delamination of rubcoating": This labyrinth in the high-pressure compressor of a propeller engine several times caused extensive and expensive damages. During an overhaul, someone mistakenly selected a poor material combination with regard to thermal strain. At operating temperatures, the rub-coating material expanded considerably more than the Cr-steel of the ring the coating was on. This resulted in high compressive-thermal strain that caused the coating to delaminate and destroy the labyrinth.
Figure "Production caused delamination of rubcoating": Sufficiently low operating temperatures allow the use of heat-resistant adhesives for fastening rub coatings.
For filled rubber coatings, silicon based adhesives are suitable. The bond strength of these adhesives is dependent upon strict adherence to the prescribed guidelines. Reusing adhesive residue or mixing earlier charges, even if they were stored in a refrigerator, can cause local delamination of the rubber coating. The same is true for poured coatings. Pouring rubber material into forms that are too hot (to eliminate the need for heat treatment and/or to shorten hardening time) can seriously reduce the bond strength and promote delamination. Large, delaminated chunks of a rubber coating can cause heavy damage to compressor blading during rubbing and/or after complete separation from the base material.
Synthetic resins (epoxide and polyester resins) can be used at temperatures up to about 200°C. Temperatures up to 400°C require inorganic adhesives such are used when applying metal felt.
This poses a similar problem as with solder, which is that the adhesive is absorbed by the porous felt and does not form a sufficient bond layer. This flaw is very difficult to discover without destroying the coating, since experience has shown that the beginning of the gap is filled (top diagram). During operation, the entire ring-shaped rub coating can delaminate (see Fig. "Delamination of rubcoating by poor bond").
Adhesives that set quickly in contact with air (e.g. adhesives with inorganic binders or silicon adhesives; air = e.g. air humidity or CO2) can form a skin in minutes, which can seriously impede bonding if the contact with the partner surface takes place after this short period of time. In this case, sufficient bond strength can only be ensured by strict adherence to the guidelines for use (time). Sufficiently certain, non-destructive testing of the bond strength is no longer possible after the adhesive has set.
Even if the adhesive layer has been applied properly, without any problems, it can be damaged by operating factors and delaminate. The bottom diagrams show several causes of adhesive failure. Usually, it is caused by different expansions of the adhesive, rub coating, and base material. These expansion differences can be due to thermal strain and/or shrinking of the adhesive.
Figure "Delamination of rubcoating by poor bond": Delamination of an affixed rub coating in a cast-steel static part of the labyrinth seal of a radial compressor disk. The delaminated coating was in a ring on the labyrinth fins on the rear side of the compressor disk. The damage was caused by faulty affixing of the porous metal felt (see Fig. "Production caused delamination of rubcoating", top diagram).
Figure "Poor bond strength of thermal spray coatings": Thermal spray coatings have a special bond strength problem. If dust (e.g. rebounded particles) is present before the application of the first layer or between the layers, the coating will not bond. The same effect can be observed with adhesive tape on dusty surfaces. This damage mechanism can be identified as a “beading effect” by microscopic inspection of the separated surfaces.
7.2.2-1 D.A. Lewis, C.E. Platt, E.B. Smith, “Aeroelastic Instability in F100 Labyrinth Air Seals”, Paper Nr. 78-1087 of the “AIAA/SAE 14th Joint Propulsion Conference”, Las Vegas, Nev. , July 25-27, 1978, pages 1-7.
7.2.2-2 D.C. Whitlock, “Oil Sealing of Aero Engine Bearing Compartments”, AGARD-CP-237, pages 7.1-7.11.
7.2.2-3 “Investigation Team Identifies Causes of CF6-80 Problem”, periodical “Aviation Week & Space Technology”, February 7, 1983, page 32.
7.2.2-4 J.S. Alford, “Labyrinth Seal Designs Have Benefitted from Development and Service Experience”, SAE Paper 710435 of the “National Air Transportation Meeting” Atlanta, Ga., May 10-13, 1971, page 1-10.
7.2.2-5 C.R. Smith, “American Airline's Operational and Maintenance Experience with Aerodynamic Seals and Oil Seals in Turbofan Engines”, AGARD-CP-237, pages 5.1-5.11.
7.2.2-6 J.T. Akin, V.S. Fehr, D.L. Evans, “Analysis and Solution of the Rotor Instability Problem in the Advanced Model TF30 P111+ Engine”, AIAA-88-3166, Paper of the “AIAA/SAE/ASME/ASEE 24 th Joint Propulsion Conference”, July 11-13, 1988, Boston, Mass. pages 1-9.
7.2.2-7 A.Rossmann, P.König, “Industrie Gasturbinen”, Axel Rossmann, Turbo Consult, 1998.
7.2.2-8 “Pratt to Test New Rotor Seal Design for JT9D-7R4 High-Pressure Turbine”, periodical “Aviation Week & Space Technology”, October 13, 1986, page 27.
7.2.2-9 “Service Bulletin Details JT9D Turbine Redesign”, periodical “Aviation Week & Space Technology”, November 3, 1886, page 44.
7.2.2-10 “FAA publishes awaited JT9D AD”, periodical “Flight International”, November 1986, page 8.
7.2.2-11 “Investigation Team Identifies Causes of CF6.80 Problem”, periodical “Aviation Week & Space Technology”, February 7, 1983, page 32.
7.2.2-12 D.R. Abott, “Advances in Labyrinth Seal Aeroelastic Instability Prediction & Prevention”, ASME-Paper 80-GT-151 December 1980, page 1-6.
7.2.2-13 J.S. Alford, G.W. Lawson, “Dimensional Stability and Structural Integrity of Labyrinth Seals”, SAE-Paper 660048 of the “Automotive Engineering Congress” Detroit, Mich. January 10-14, 1966, page 1-31.
7.2.2-14 “Headlines”, periodical “Flight International”, 22-28 July, 1998, page 8.
7.2.2-15 “Titanium `95: Science and Technology”. page 1600.
7.2.2-16 “Seal Problem Triggers GE Engine Modification”, periodical “Aviation Week & Space Technology”, August 8, 1994, page 29.
7.2.2-17 “GE90 hit by turbine-seal snag”, periodical “Flight International”, 19-25 October, 1994, page 4.
7.2.2-18 “Crack causes delay in Trent777 ETOPS tests”, periodical “Flight International”, 11-17 October 1995. page 6.
7.2.2-19 “B77 Engine Problem Update”, “World Aero-Engine Review”, October 1997, Volume 5 No. 10, page 8.
7.2.2-20 S.W. Kandebo, “USAF Proposes Theory for F110 Seal Failures”, periodical “Aviation Week & Space Technology”, November 7, 1994, page 33.
7.2.2-21 S.W. Kandebo, “WSAF, GE Develop F110 Engine Plan”, periodical “Aviation Week & Space Technology”, Dezember 1994, page 19.
7.2.2-22 J.S. Alford, “Protecting Turbomachinery From Unstable and Oscillatory Flows”, periodical “Journal of Emngineering for Power”, October 1967, page 513-528.
7.2.2-23 J.S. Alford, “Protection of Labyrinth Seals From Flexural Vibration”, periodical “Journal of Engineering for Power”, April 1964, page 141-148.
7.2.2-24 S.W. Kandebo, “Seal Failure Halts F119 EMD Testing”, periodical “Aviation Week & Space Technology”, March 16, 1998, page 29.
7.2.2-25 J.S.Alford, “Protecting Turbomachinery From Self-Excited Rotor Whirl”, periodical “Journal of Engineering for Power”, October 1965, page 333-335.