Aeroengine Safety

Figure "Risks during development of repairs": From new developed or changed repair processes a higher risk can be expected than from new part development. Cause is the more unfavourable framework requirement.

Operation load: Usually the OEM possesses sufficient knowledges about the design philosophy and the calculation data. With this for the independent developer the danger exists, that flaws/faults are not identified or underestimeted in its significance (Ill. 21.3.2-3)

Frequently a repair is necessary without knowing sufficiently the relevant operation load (Ill. 21.3.2-4). Cause is the complex interaction of different failure mechanisms. So the repair the sufficient certain analytic processing revokes itself respectively the design of new parts.

Thr serviceability of a new part was prooved in extensive test runs and verifications, according to the principle „the engine will tell us“. This is especially important for dynamically high loaded parts which can fail if the vibration load increases only little. So a repair with removing rework can enable a resonance (Fig. "Vibration of turbine blades influenced by rework").

For old aeroengine types often at the OEM no more senior experts are existing. In such a case no sufficient support can be expected.

The service conditions can differ markedly depending from operator and the mission. With this the evaluation of the failure triggering influences change considerably.

Behaviour of the part: The repair itself can markedly influence the failure behaviour. So for example brazing seams behave for wider brazing gaps brittle (chapter 21.2.2, volume 4, Ill. 16.2.1.4-9). Welds can have markedly internal tension stresses, which promote the vibration fatigue. Its strength which can be expected lays under this of the base material (chapter 21.2.1). With this the thermal fatigue behaviour can come into the foreground, even when the repair was necessary because of oxidation or creep.

Especially malicious are repairs which unconscious influence the vibration behaviour (resonance, damping, Fig. "Cases of disimprovement by repair"). Repair coatings must be seen suspicious . Here also the vibration strength can drop unexpected. Typical example are brittle erosion protection coatings in airfoils of blades which split out by small FODs (volume 1, Ill. 5.3.1-3 and volume 3, Ill. 12.6.3.4-19.2). A further example is the increase of friction at dove tail roots from compressor blades (volume 2, Ill. 6.1-10). This can occur by an operation caused aging/oxidation (time, temperature). Depending from the formed oxide type a rise of the friction can increase dangerously the load/stress at the root (combination of bending stress and shear stress) and promote incipient vibration fatigue cracks (volume 2, Ill. 6.1-20).

Operation caused changes at a part which should be repaired:
Especially aggravating are changes which prevent a consistency of the part features, comparable to the production of new parts. To these belong influences of the base material like changes of the structure. They affect the welding behaviour. Deeper oxidation, e.g., in cracks, hinder a brazing repair. Also the removing (strippen) of oxidized coatings can get more risky because a local attack (e.g., corrosion cracking, chapter 21.2.4, volume 4, Ill. 16.2.1.7-10).

Changes at the part after repair:
Removal by influences like erosion, oxidation and rubbing can change the behaviour with the operation time promoting failures. To this belongs a change of the vibration behaviour by crack formation or weakening of the cross section which is otherwise controllable. Also a change of the surface (erosion, oxidation, deposits) can influence the part temperature with its reflection behaviour (e.g., at the
combustion chamber or turbine blades; volume 3, Ill. 11.2.2.1-7). Is such a behaviour changed by the repair, there can be after log operatuion times disagreeable surprises.

Process relevant flaws/faults can develop to failures and with this become unacceptable (volume 1, Ill. 3-1). This is the case, if these in their the consequence atn the behaviour of the repaired part are not sufficient known and defined/specified. Especially repair welds and repair brazings must surrender this evaluation.

Test processes: At first the definition of repair failures, which determine the lifetime respectively the safety, is necessary. An aggravating influence is, that different to a new parts, not seldom the limits for flaws/faults must be extended. This can be for example on brazings a process typical, i.e. under repair conditions unavoidable, not optimal adhesion/joining or a brittle material structure. A repair in the part zone, which at the new part is not analogical tested, can require a particular adapted or a new testing process. Its practice suitable effectiveness must be tested and proved together with the repair development.

Limits of the development effort for repairs: Compared with the development of new parts from experience the repair development takes place with a forced, by the limited effort in a markedly restricted `environment'. Frequently there are acute problems which markedly limit the available time. This can be intensified by the competition between the repair shops. It can incite to introduce the repair too early into the series application and to increase with this the risk of a disappointment with extensive costs.

A sufficient period of time up to the proof of the serviceability like for new parts during the aeroengine testing, often is not possible for repairs. Instead „fleetleaders” are used. They have the disadvantage to need a comparatively large time period (years) till the introduction into series operation.

Personnel and facilities: Compared with the development resources of an OEM, those of a repairshop are rather limited. This begins with the design documentations and design specialists, does not yet end with basic accented facilities for investigations and tests. With this the necessity for intelligent and resource saving solutions increases. A repair shop needs for this especially experienced, qualified and application orientated employees. Seen from the work spectrum rather higher demads are needed than in the development of new parts.
Compared with the machinery of the new parts production it must be especially flexible, accordant to the diverse part types and small series.

Illustration 21.3.2-2(Lit. 21.3.2-1 and Lit. 21.3.2-2): To minimize special risks during the development of a repair (Fig. "Risks during development of repairs"), there is a sequence of recommendable single steps. The determination of the causal failure mechanism is a requirement for the identification of failure relevant influences. Base is a systematic failure analysis/ problem analysis. Out of it starting points unfold a pointed remedie also as a repair (volume 4, Ill. 17.1-11).

The reproduction of the operation failure is the requirement for a comparative assessment and validatio nof the repair. A simple test rig for reproducible results , sufficient near the operation conditions, is a very demanding task. It guarantees an acceptable effort for proofs.

Development of the repair respectively of the repair process, should as far as possible carried out with original parts and not with samples. Thereby all specifications and approval requirements must be considered. Naturally the repair mist guarantee the necessary lifetime with sufficient certitude.

Validation by analyses like FMEA and `analysis of potential proplems' (volume 4, Ill. 17.1-10 and Ill. 17.1-11). The repair process should be controlled with ensuring analysts for its possible undesirable consequences.

A comparing proving/testing from experience is the appropriate instrument for the proof of the serviceability of a repair. Requirement is the suitable test setup (see above). The best but also most time and cost consuming test may perhaps be in the engine under realistic operation.

It shold be insistet upon the testing/proving in the aeroengine (fleetleader). This is from experience time consuming and expensive. But it may offer the highest safety against disagreeable surprises (`disimprovements'). A comparing evaluation should be aspired. This is possible for a high number of similar parts like blades in so called „rainbow tests/runs“.

Monitoring, documentation and analysis of the tests/proofs: Naturally after an informative operation, the repaired blades in comparison to not repaired parts, should be most closely evaluated and documented. This effort must be budgeted in time and costs already at the beginning of the development.

Illustration 21.3.2-3: It is assumed, that in many cases e.g., for repair welds (volume 3, Ill. 12.6.2-22) and repair brazings no, equal to the new part, operation behaviour can be expected. This can mean a shortened lifetime but also influence the failure mechanism. With this an adjustment of maintenance and inspection for such a behaviour is necessary.

Illustration 21.3.2-4: This example shows how important „background knowledge” of the OEM about design and load distribution of a part are.

These, in segments (e.g., 4 vanes in one part) arranged low pressure turbine (LPT) stator vanes have above the shroud a stiffening fillet (sketch in the middle). Here crack formation occurred (detail above right) with features of thermal fatigue (volume 3, Ill. 12.6.2-10).

As remedy a repair was introduced. Cracks whitch didn't exceed a certain depth, have been removed by grinding (detail middle right). This recess at the fillet increased the elastic flexibility. It was assumed, the thermal stresses could be better equalised and so the thermal fatigue load decreased. However after the introduction, during operation dangerous cracks occurred, which can in an extreme case trigger the crack of the front fixing nose (sketch in the middle). The then backwart tilting statorvane could at the following rotor (sketch left) cause heavy rotorblade damages. Aggravating accrued, that the parts could not be checked for cracks in the assembled conditiion.

It showed, that the material removal during the repair weakened the stiffening fillet and produced a notch, which obviously promoted further thermal fatigue cracks. The rearwards oriented high gas bending loads than lead to a creep overload of the remained cross section.
As successful remedy a design change with a markedly more massive stiffening fillet was introduced.

Fig. "Disimprovement by repair" (Lit. 21.3.2-3): In elder aeroengine types in which still frequently rusting steals are used for highly loaded components like rotor disks and casings, cadmium (Cd) is applied as corrosion protection. Cadmium can already trigger at service temperatures of few hundred °C, embrittlement and crack formation (volume 4, Ill. 16.2.1.8.3-8 and Ill. 16.2.2.3-11). This as LME (liquid metal reaction) and SMIE (solid metal reaction) known processes, are extremely dangerous (Fig. "Casing behaviour like apressure vessel" and Fig. "Danger of LME and LMIE by galvanic platings"). In the illustrated case the fracture of a compressor disk was caused.

The investigation showed, that there was against the specification no sufficient protecting Ni interlayer between the cadmium corrosion protection layer and the base material.

A check of th coating process at the overhaul shop showed as main cause an unsuitable process sequence..

Obviously the potential danger of this new coating process during an overhaul was not known or not aware. This may have been the reason for this quality fault.

As remedy, audits of the shops which make such coatings have been carried out. Additionally an instruction about the danger of a cadmium embrittlement took place.