Aeroengine Safety

Figure "Limited inserted number of repaired parts": Are components composed during the assembly, may be limitations must be considered. In the shown case the 1st high pressurt turbine stator (turbine nozzle) is concerned. It consists of several components (vane segments, detail right). It is supported by the vanes and the inner combustion chamber liner at the compressor exitcasing. The vanes are highly creep stressed by the gas loads.

During the flight the fracture of the vane occurred. The stator came loose, with catastrophic consequences. The investigation showed, that the procedure during an earlier overhaul did not correlate with the instructions of the overhaul manual. All vanes have been repaired by extensive brazings. In contrast the manual limits the number of the repaired vane segments for the assembly of a stator and demands an alternating combination with new parts.

Figure "Extent of welding repair": Concerned of the following problem are especially welding repairs on seemingly low loaded components of elder aeroengine types (upper sketch). Here because of logistic shortages the desire may exist to erxploit or to exceed the specified limits in the overhaul manual excessively. Even if this does not happen it must be anticipated, that the quality of the welding repair degrades with every overhaul respectively repair welding (Fig. "Risk of combustion chamber repair").

This case must be expected from an `aging' of the material and/or deteriorations like not sufficiently removable oxidation. Such deficits must not be sufficient sure detectable from the outside. So the question arises, how often such parts can be overhauled if the instructions are interpretable.

An especially dangerous situation arises, if such parts are subject to not aware high dynamic operation loads (high frequency vibrations). Is the part weakened at repair weldings by a higher number of cracks during operation, the natural frequency can drop till resonance will occur. In the extreme case the, by cracks „softened“ (elastic) region, can be excited by flutter vibrations.

An example are flame tubes of can-type combustion chambers, produced from sheet metal (lower frame left). In the shown case the possibility existed to exploit excessive the, for a repair limited length, number and distribution of typical thermal fatigue cracks in the corners of the combustion chamber gills (detail). It showed, that then the dynamic loads (usual combustion oscillations) lead to vibration fatigue cracks. If lager pieces of the combustor wall separated, it came after few hundred operation hours to dangerous turbine failures (volume 3, Ill. 11.2.2.2-2).

The sketch down right shows a turbine casing in welded design, made of sheet material. Typical weak points of such casings are the transitions of the support struts between the bearing chamber and the wall of the outer casing. Here from experience a high thermal fatigue load reigns in combination with high frequency vibrations. Is there for multiple overhauled parts no sufficient safety margin, compared to the repair limits in the manual, the danger of an early failure exists.

Figure "Repair of engine casings" (Lit 21.1-14): The special requirements for repairs of casings are often underestimated. Their high operation loads can be of differen type and cause. Also at the dimensional accuracy in certain zones like centering areas or tip gaps at the rotor blades there are high requirements. For the thin walled, big and complex structures the requirements can be extremely demanding. Often weld repairs make procedures like straightening and/or stress relief (heat treating, shot peening) necessary.

Frequently aeroengine casings bear main bearings. For this, bearing chambers are supported by struts (Fig. "Extent of welding repair"). From the bearings high radial forces and from fixed bearings high axial loads can act. Unblances introduce high frequency dynamic loads. High temperature gradients, especially in so called `hot strut' designs (struts through the hot gas stream) produce equivalent thermal stresses. So especially the transitions of the struts at casing and bearing chamber are so exposed to a high thremal fatigue load.

Pressure casings are highly stressed. Appropriate attention must be payed during welding repair. The stresses of such a casing is comparable with a pressure vessel. Concerned are rear compressor casings, cumbustor casings (Fig. "Operation load limiting repair of casings") and casings of the high pressure turbine. Frequently casings have also containment function in failure case, to catch fragments of the rotor blading (volume 2, chapter 8.2) or exiting fire (titanium fire, volume 2, chapter 9.1). All this functions must be considered during a repair and must not restricted. Thereby for example must kept in mind, that from the casting structure of a weld seam, not the same favorable energy absorption during a fragment impact like from a not influenced forged material can be estimated.

Figure "Operation load limiting repair of casings" (Lit. 21.1-1): Stresses in casings with high inner gas pressures during operation can be easily underestimated. The shown example gives an impression for this. It can apply as warning to pay especially attention during repairs in flange zones of such casings. In no case repair limits in manuals or instructions may be exceeded (Fig. "Casing behaviour like apressure vessel"). In case of doubt the OEM must be consulted.

That especially counts for the combustion chamber outer casing (CCOC). In the example the operation temperature is 550°C. Concerned is a repaired part. The transtition geometry with two radii of the casing wall to the flange of the incident aeroengine is an elder version (details below). That, already a changed version, was introduced, permits the conclusion at known problems in this component area. In the radius near the flange a circumferential LCF crack over about 195° developed which can be assigned to the start-shut down cycles. About several similar failures are reported.

Figure "Casing behaviour like apressure vessel" (Lit. 21-2): The cause for this aeroengine failure with the exit of fragments (see also Fig. "Interpretation of an instruction") was a dangerous metallic contamination of the surface. It was deposited during the repair during an overhaul process when a Ni-Cd corrosion protection coating plated also the flange bores. Such a plating violated the repair instructions of the OEM.

Citation from the overhaul manual:
„Cadmium Plate is not allowed in holes of front or rear flange or holes in any other area. Any cadmium plate in any hole must be removed.”

It was known that cadmium embrittles the steel of the casing/flange at operation temperatures of several 100°C. Thereby about a so called liquid metal embrittlement (LME, volume 4, chapter16.2.2.3) occurres.

Obviously the cracks in many flange bores of this critical zone have not been identified with eddy current tests during two intermediate inspections. So, after about 8000 start-shut down cycles it came to the rupture of the casing when the crack had reached the critical length.

Figure "Limits of welding and brazing repair" (Lit. 21.1-3, Lit. 21.1-4 and Lit. 21.1-5):The strength of repair brazings at hot parts can not estimated like the original base material. It can be supposed, that the hot strength and the creep characteristics are markedly lower (volume 4, page 16.2.1.4-50). This disadvantage compared with a not deteriorated base material, is increased by a pronounced brittleness (volume 3, Ill. 12.6.2-22 and volume 4, Ill. 16.2.1.4-9).

So the same quality respectively lifetime like new parts can not be expected from repair brazings, which meet contrary to brazings of the new part worse operation caused properties of the base material (oxidation, change of structure). This must be kept in mind for repairs (Fig. "Engine failure by repair test rig and manual"). Under this point of view the brazed `patch' in the upper sketch is for example extremely problematic. Its braze seam runs in a usually lifetime determinant high thermal fatigue loaded transition zone from the blade to the shrouds. With such a design an early failure in the braze seams may be preprogrammed.
Better seem the lower sketched versions. At the left the brazed piece extends over the whole leading edge of the blade. So the braze seam is also orientated in direction of the thermal fatigue stresses/loads and so is released. An additional positive effect is achievable, if an especially oxidation stable and thermal fatigue resistant material (directional solidified, single crytal) will be brazed.

Positive also the lower right version can be evaluated. Here, by means of a thermal barrier coating, the thermal stresses respectively the thermal fatigue load are reduced. This can be beneficial combined with the already discussed versions.

Similar considerations like here for repair brazings are also true in a transferred sense for repair welds (chapter 21.2.1).

Figure "Internal stresses of coatings as application criterion" (Lit. 21.1-6): For many coatings/platings, especially applied by galvanic processes (volume 4, Ill. 16.2.1.8.3-9), it must be reckoned with detrimental (tensile) internal stresses (diagramm right, volume 3, Ill. 12.2-13). Tensile internal stressses superimpose with operation stresses and so increase the mean stress. With this the usable fatigue strength drops. This effect is intensified fromfor such coatings typical high modulus of elasticty and their brittleness.

The hight of the tensile stresses depends heavily from process parameters. Therefore these must meet accurately the appropriate specifications. Also a change of the coating type, e.g., the switch from cromium plating to nickel plating is not allowed without the approval of the responsibles (OEM, authorities).

Especially for the repairs of worn centering /contact surfaces from rotor components highest attention is needed. The left sketch shows such a disk with a nickel plating. In every case the instructions in the manual and specification must be kept exactly. In no case, coatings may be used which are not explicit approved for this part but are allowed at similar parts or component zones.

Figure "Influence of a hard coating at fatigue strength": With a repair worn surfaces can be rebuild and protected as prevention. This is used as well against fretting wear (volume 2, Ill. 6.2-8) at contact/support surfaces as also against erosion (volume 1, Ill. 5.3.3-5) of blade surfaces in the air/gas stream or against abrasion by rubbing (blade tips). For this reason components will be frequently repaired by coatings/platings. However thereby it is absolutely necessary to consider the risks. Primarily concerned are effects which decrease the fatigue strength and so can lead to a catastrophic failure of the component.

The danger is to evaluate so much higher the harder, brittle and stiffer (high modulus of elasticity) the coating (volume 4, Ill. 16.2.1.8.2-5). Therefore coatings/platings may only applied where this is permitted/specified in the manual respectively instruction. In case of doubt it is absolutely necessary to consult the OEM and the resposible approving authority (e.g., also for military applications). For the developing of a repair coating, the production process with its parameters must be part specific selected and optimised. Its harmlessness must be proven with suitable certification tests.

The following potential damaging effects must be considered:
FOD-behaviour (detail above left) especially for the blading in compressor and turbine stands in the foreground. Primarily concerned are firm adhering hard but also brittle coatings which offer itself as erosion protechtion. In an undamaged condition such coatings can even increase the fatigue strength. But if even tiny foreigen objects like sand grains produce in this coatings sharp edged notches and cracks, highest attention is demanded. In such a case a dangerous drop of the fatigue strength must be expected.

The transition coating/base material (detail below left) can act as notch and decrease the fatigue strength. Therefore such a transition schould at least not be located in a markedly dynamic loaded region. Can a transition not be avoided at edges and chamfers it must be suitable shaped, respectively reworked.

Basically attention must be payed, that there is no „leaked coating“ (from the coating process if not sufficient masked) as this happens for example with brittle diffusion coatings. Typical example is a leaked coating at the fir tree root and shaft of turbine rotor blades with a diffusion coating of the blade.

Corrosion of the base material (detail above right) can emerge in different manners. Is the coating material more noble than the base material it must be reckoned with the formation of a galvanic cell with corrosion during operation (volume1, Ill. 5.4.1.1-8.1 and Ill. 5.4.1.1-8.2).

Develop thereby corrosion pits (volume 1, Ill. 5.4.1.1-3) they will act as notches and decrease the fatigue strength dangerously. Additional such an effect can be intensified by the notch effect of the transition of the coating. From experience blades of martensitic steels (Cr-steels) are particularly endangered. Those are used in elder aeroengine types .

Stripping of a coating can attack the base material. Thereby a surface removal is less serious than locally attack and/or formation of microcracks (intercrystalline corrosion, volume 1, Ill. 5.4.1.1-3).

Intrenal tension stresses (detail below right) increase during opreation the mean stress and lower the fatigue strength (Fig. "Pipe line crack by tensioning" and volume 4, Ill. 16.2.2.4-4).

Figure "Unsufficient crack detection during overhaul" (Lit. 21.1-14): This example (volume 4, Ill. 16.2.1.8.3-6) shows the potential dangerousness of production procedsses with electric contuiunity. During galvanic plating of the compressor disk (sketch below middle) it came because unsufficient contacting to electric sparking. The developed little fusing zone at a bore of the disk lowered the fatigue strength (here LCF) markedly. It came to crack development and fracture of the disk.

This case demonstrates the danger also of small fusion puddles (volume 4, chapter 16.2.2.6). Therefore during repairs especially attention is needed, that no electric sparks/arcs occur at the part. Such a damage may act deeper than it may seem from the outside. Therefore an assessment and rework demands special monitoring and the proof of harmlessness (volume 4, Ill. 17.5-1). In case of doubt the OEM must be consulted.

Fig. "Rework of firtree slots in disks" (Lit. 21.1-8 und Lit. 21.1-10): The bottom of disk slots is as well in the compressor as in the turbine an especially cyclic high loaded component region. Here damages (Ill. 20.1-20) but also rework are accordingly problematic (volume 4, Ill. 16.2.2.2-8). Therefore it is necessary to keep especial tight specifications and limitations in instructions, respectively manuals.

To the noteworthy influences belong because of the orientation of scratches/grooves the machining direction. Therefore our attention should especially at axial orientated marks. Also the process parameters must be kept to avoid for example a local overheating. This can after a rework like polishing almost no more verified with a following non destructive test (volume 4, Ill. 16.2.1.1-9.1 and further).

Figure "Favorable design for weld repair" (Lit. 21.1-11): How good a part can be repair welded depends, besides from the material not at least from the design/geometry. This is especially true for complex casings like exit casings of the compressor (lower sketch) and turbine from elder aeroengine types. Here weldig designs are concerned (Fig. "Extent of welding repair"). Widely closed hollow spaces like bearing chambers are an especial challenge. A further important aspect is the verifiability of the repair quality by non destructive testing. This possibly is carried out visual and/or by X-ray. However this demands a sufficient accessibility which is not always guaranteed.

If low alloy steels are concerned, attention must be payed, that during welding no moisture can interfere. Otherwise the great danger of a hydrogen embrittlement with delayed crack formation exists (volume 1, Ill. 5.4.4.2-4 and volume 4, Ill. 16.2.1.3-18). This demand is just for hollow spaces not easy to keep (condensate), even if shielding gas is used.

Also austenitic alloys need especially at the welding root sufficient shielding gas. So a critical oxidation of the welding can be avioded.

Figure "Avoiding corrosion during etching process" (Lit. 21.1-12): In many cases the overhaul respectively repair demands an etching of the components. The used etching media are extremely aggressive. Frequently chlorine containing acids (e.g. watery FeCl₃ solution) are concerned. Etching serves especially the cleaning and removing of oxides as well as the preparation for coating processes (e.g., chromium plating and nickel plating). Also for the stripping of coatings aggressive etching baths are used. Often those must act stronger as baths in the production of new parts when the longtime operation has changed the coating. Only after a sufficient removal of the oxides with a possible crack opening a successful penetrant inspection has a chance (Fig. "Influencing NDT by abrsive blasting and etching").

Just oxides are chemically extremely stable, more than the base material. This is the reason why during the removing of oxides by etching the danger of damaging/deterioration of the grond material exists. The cause can be a deviation of the structure from the base material compared with a new part. This is the result of an aging respectively change of the base material composition by oxidation and diffusion processes during operation (e.g., sensitising, volume 1, Ill. 5.4.1.1-9). However from experience rather process parameters probable are not met, like too long etching times or aged etching baths. These are able to attack preferential grain boundaries cracklike (intergranular corrosion = IGC, upper frame, volume 1, Ill. 5.4.1.1-9). Such damages can not be found non destructive. In an extreme case an unusual dull sound occurres during handling (volume 4, Ill. 16.2.1.7-10). This can be an indication for a damping by cracks.

A further feature, especially of nickel alloys, which can be identified visually, are small dark spots (frame below) which point at grain separations by attack (page 21.2.4-1 and Fig. "Problems by thr removal of coatings"). They form at far advanced, heavy grainboundary attack. Even if only the first grain layer at the surface should be concerned, it must still be reckoned with a dangerous drop of the fatigue strength.

Figure "Use of new overhaul instructions during repair" (Lit 21.1-13): Just (fretting) wear systems like spline shaft couplings must be seen as unities during changes. If one side is improved this does influence the behaviour of the mating surface. In such a case an improvement can demand measures on both sides. This can get for the overhaul of a module to an particular problem. This is the case if the module has only one element of a tribological system (e.g., only the coupling sleeve or the shaft) and only this element is recognised. This can lead to the premature failing of the system after an anew assembly. This is the case if the couterpart without an optimal surface now wears faster. This behaviour must also kept in mind during the development of repair processes of (fretting) wear systems.

In the shown case the wear behaviour of a spline coupling nitrided at both sides (sketch below left) should be improved. For this during the overhaul, both sides are chromium plated (sketch below middle). However during repair only the sleeve was chromium plated (sketch below right), the shaft remained nitrided. Tests showed, that this combination wears out the nitrided part extremely fast and lead to the failing of the connection.

Figure "Wear by a new combination of contact surfaces": From fretting wear (volume 2, chapter 6.1) contact/seat surfaces like they are typical at blade roots, centrings and spline shafts experience a local abrasion. Are these surfaces combined during an assembly with new or already worn surfaces of of exchanged parts, the worn profile of the contact surfaces does no more fit. This causes locally high contact pressure and adverse load combinations by the friction forces (equivalent stress, volume 2, Ill. 6.1-11). With this, fatigue cracks are promoted. For example attention must be payed especially at this effect during the rework of frettingsurfaces on blade roots and disk slots (volume 2, Ill. 6.2-12).

Also the wear rate of a new combination of used contact surfaces can get markedly increased (e.g., at spline couplings).

Figure "Reuse of parts with accepted wear" (Lit. 21.1-15): The disassembly of the failed aeroengine showed, that a turbine rotor blade of the 3rd stage (right sketch) was broken. This was already the third „parallel case” in the same airplane during one year. The blade was broken 12 mm above the root platform (frame left). Thereby it knocked off the shroud of the next blade following against the rotation.

The investigation of the blade fracture surface showed, that it was HCF (fatigue fracture). It started at the leading edge like usual for casting materials as a typical `1st stage crack/cleavage crack (volume 3, Ill. 12.6.3.2-4). A fault, which could be evaluated as causal like porosity was not found. The total operation time of the blades was about 4 100 hours with about 3 800 start-shut down cycles.

All three blade fractures occurred at already used blades of newly equipped stages. Those blades met at the point of time of the assembly the dimensional specifications. However the gaps at the tip shrouds increased faster than expected by fretting wear during operation. This was traced back by the OEM at unsuitable, punctual contact conditions between the neighboured already slightly tip worn shrouds. The large shroud gaps lowered the damping and stiffening effect. With this dangerous blade vibrations became possible, which in the first case triggered a fatigue fracture in the root platrform and the following next two blades.

As remedial measure the OEM developed a repair process for the contact surfaces of the tip shrouds.

Figure "Differences to older part types": Considering the total life of an aeroengine, the far longer service life compared with a motorcar is striking. Half a century is quite usual for aeroengines. Naturally in such time periods the technical evolution continues. Simultaneously old technologies, because of unsufficient effectivity of the process or the product for modern applicationsm, are dropped (e.g., forged turbine blades). This can cause, that later productions can no more benefit from formerly know how and experience based necessities. So it is quite possible, that at new parts, seemingly insignificant deviations from the original parts, approved by the OEM and the authorities emerge. Those differences can cause during operation unforeseen problems,which have not considered during the modification because they have been forgotten.

For example the process and the after treatment for the bending of tubes may have been changed because of cost reasons (sketch). Fatigue cracks can be prpomoted if in such tubes/pipelines undetected internal tensile stresses are induced or unknowigly a strain hardening like from the former used forming process lacks.

These problems are worsened if after a long time period since the development of the aeroengine the senior experts are retired long ago. The consequence just for military aeroengines is, that the development of repair processes devolved to the responsibility of the operator respectively licence holder. So also the knowledge of the operating loads, basis of the design, lacks.

Changes in material and processes belong to the especially problematic measures.
Changes of the material can get necessary, if the life of the component should be raised (e.g., for turbine blades). This is also true, if the availability of a former alloy (e.g., Ni forging alloy) no more exists. An example is the change from Ni forging alloys of turbine blades like they are usual in old aeroengine types to casting alloys with a higher creep life. However these have a lower fatigue strength what can cause fatigue cracks just for blade roots of an elder design (many small teeth).

Coupling nuts made of steel at connecions of pipelines have a higher strength than such of aluminium like used in elder aeroengine types. However causes a conversion because of a resonance due to the mass change. So the danger of a fatigue crack is increased in spite of the better strength. Also changes in stiffness and thermal expansion of the bolt connection have an influence. .

Costs are quite a reason for cheaper materials. Also the availability is an important criterion. For example, was formerly a gold alloy brazing material used for the joining of compressor stators, the wish to replace this with a Ni-brazing is uderstandable. But the higher brittleness and the higher brazing temperature can dangerously decrease the strength of the joint. Unfortunately, frequently such a problem will be identified not before a longer operation time what especially to increases the risk of a failure (Fig. "Vibration of turbine blades influenced by rework").

The environment compatibility gets for the material selection more and more into the foreground. But this can have the unattractive consequence, that problematic materials like asbestos with a combination of special, favorably properties are no more available. The experience shows the difficulty to find in such a case a satisfactory substitute. For example not before the operation it will be identified, that also at unconsidered features like the absorption of water (condensate), friction coefficient under operation influences or softness during overload (e,g., under unexpected thermal stresses) are important.

Figure "Vibrations of nearby very exact blades": A quite unexpected effect can lead to problems just in elder aeroengine types. Modern production processes often permit higher accuracy than the former produced „original components“. But this scatter established a positive operation experience according to the principle „the engine will tell us”. For example, if a new produced set of fan blades is assembled it can be more susceptible for vibrations. This can be possibly caused by the almost identical vibration frequencies of all blades what promotes a resonance (sketch right, volume 3, Ill. 12.6.3.4-6.2).

In case of blades from a former production with small, approved deviations or reworked blades, the scatter of the vibration frequencies (natural frequencies) may have helped to avoid fatigue failures. However this will get aware not before problems with the new production occur. So such effects must be considered also during the development of repair processes and improved production processes.