21:213:2133:2133

21.3.3 Danger of a disimprovement by repair

Repaired parts have from experience an increased potential of unexpected operation problems and failures. At least partially it is caused by the phenomenon „disimprovement“ (Ill. 21.3.3-1 and volume 1, Ill. 3-2). This can be explained with, for aeroengines typical , unconcious operation loads and influences which lay in the limit range and elude an analytical consideration. It causes an effect which can be made ostensive with the model of an unstable equilibrium (Ill. 21.3.3-1). Changes that can be evaluated as negligible (Ill. 21.3.3-3.1 up to Ill. 21.2.2-3.4) can trigger a problem which stands in no noticeable or identified connection. This is promoted by the seemingly small safety factors of the aeroengine design. They are realised not at least with a high test effort, meticulous quality insurance and closely monitored procuction. Additionally the design follows the „design lines”, which under statistical consideration specify minimum values. This concept meets the high safety demands for new parts, which have passed through the development phase of the aeroengine. But it can be a reason, that a decrease of parameters (e.g., in the region of a welding seam), caused by repair, brings those dangerously near the operation loads.

Some examples shall motivate a low-risk approach during the develpment of repairs.
To these belong tha manifold effects of base material changes during a repair (Ill. 21.3.3-2).

Coatings can unexperienced influence the operation behaviour in an unexpected manner. This gets understandable, if besides other problematic effects the consequence of a brittle erosion protective coating (Ill. 21.3.3-3.1 up to Ill. 21.3.3-3.4) on the fatigue strength of compressor blades under real operation influences is considered. Have been inevitable, uncoated bearable small FOD impacts not considered, it can come to fatigue failures with catastrophic consequences (Ill. 21.1-8, and volume 3, Ill. 12.6.3.4-19.2).

Also the strength of the base material is influenced by repair welds (Ill. 21.2.1-3) and repair brazings (Ill. 21.3.4-4 and Ill. 21.2.2-4). Not always this is realistically estimated (Ill. 21.1-2 and Ill. 21.1-6).

Illustration 21.3.3-1: With „disimproving“ in the following an action, respectively remedy is called, from which an improvement is expected, but which leads to new problems. These can occur at the same part or at an other part by the reason of an unconsidered effect. As a model different equilibriom conditions are suited (sketch). Why just aeroengines suffer under this occurence more than the usual mechanical engineering is explained more in detail in volume 1, Ill. 3-2.

Width and form of the ball supports in the sketches shall correlate the safety reserves, respectively a possible fail safe behaviour. Drops the ball over the edge, this will relate the uncontrolled failure of the part. The displacement of the ball equals a straining/damaging effect at the part. Parts where this condition is not „seen” and/or which underlie a interaction of several different, potential damaging operation influences (sketch in the middle), are especially endangreed by a disimprovement. Than finally only a sufficient testing with operation experience decides about the suitability. This principle: „The engine will tell us“ is also valid for the applicability of a repair.

To minimize the risk of a disimprovement, suitable approaches offer itself:
Determination of the problem relevant influences:
For this a systematic problemanalysis (volume 4, Ill. 17.1-11) is the right choice. This approach ensures the consideration and objective evaluation of the known facts. On these the causative influences are reviewed. With this connected documents enable in case of problems again at a later point, a reconstruction of the decision making in time.

The analysis of experiences is an especially effective instrument to identify the risk of a disimprovement in time (Ill. 21.3.3-5). To such experiences belong analyzable collections of many years of failure (investigation) reports or customer service documents. Here must be noted, that also seemingly as single cases evaluated events must taken very seriously. So it is still possible they hint, that the part operates near its load limit. Then statistical overloads must be expected. Also instructions in the overhaul manuals, respectively repair manuals can indicate flaws/weak points. For example certain part zones can be repaired not at all or only very limited (Ill. 21.2.6-2).

An extremely safe but frequently very extensive process is the reproduction of the failure, which should be repaired it succeeds at all (volume 1, chapter 4.4). With a successful test rig, as simple as possible (laboratory conditions) the causative influences can be verified and if necessary corrected.

Comparative proving/suitability testing: Was the reproduction of the failure which should be repaired in the laboratory test successful, the chance of quantitative conclusions about the lifetime of the repaired part exists. For this the comparison with the new part is convenient. Naturally there will be a quite markedly remainig risk, because it will possible only in a minimum of all cases simulate all acting operation loads. For exampple a vibration test of a rotor blade can be carried out to test the influence of a coating at the root. A simultaneously distribution of the centrifugal load, similar to the operation, is not possible because of the blade geometry/form with a fixed clamping.

Such load combinations must be realized in an aeroengine run at the test rig. This is usually very expensive. Therefore it will be tried to test several different repairs together. Unfortunately in many cases the relatively short test run time of maximum few hundred hours is not sufficient to identify certain long time effects (fretting, oxidation, corrosion, LCF).
The best is the aeroengine in typical series operation (fleetleader) over a sufficient time. The long time period (years), till a meanigful result can be expected, is a big disadvantage.

Illustration 21.3.3-2: The properties of a material can be changed failure relevant in many ways by a repair and/or remedy (Ill. 21-8).

That is applied to processes like welding, heat treatment or brazing. To this also belongs the change to an other raw part supplier or a seemingly more suitable material.

Material changes and/or the change of the raw part supplier is usually, at least for repairs in the civil field, excluded. This would demand extensive cost and time demanding proofs for whith the OEM is essential.

However in the military field material changes during a repair development are quite possible. This is mostly carried out at the OEM or in cooperation of OEM and the operator/military (page 21.3.1-2).

The chart should show the consequences of changed material properties for the reparability and the operation behaviour. With this the alertness for such changes shall be stimulated. The sketches above assign two typical part potential consequences of material changes. Concerned are turbine rotor blades (left) and a compressor disk.

The consequences don't apply the base material. Also admixtures for coatings, braze metal, welding consumables and so on are concerned. Already the aberrance of parameters from coating processes can trigger an unexpected unsuitable operation behaviour. To this belongs a shortened lifetime up to a catastrophic failure. For example, if the porosity of a thermal sprayed rub in coating is changed different to the proved original parts, this can have varying consequences. Either a too high porosity promotes the erosion, or a too dense coating triggers during rub to high abrasion of the blade tip and damages it. In the following, some typical consequences and effects are discussed in detail:
Shortening of the lifetime (long time effects): Increased process temperatures (heat treatment) can promote changes of the material structure like a „sensitising” (volume 1, Ill. 5.4.1.1-9) and with this intercrystalline corrosion (volume 4, Ill. 16.2.1.7-10).

Increased and/or unfavourable internal stresses, e.g., caused by deviating forging parameters during the change of the supplier, can superpose operation stresses and so shorten the LCF life.

Crack formation by LCF loads (also thermal fatigue) and creep: Changed grain orientation (grain boundaries) influences crack development at the grain boundaries which are evaluated as weak points. For the fatigue strength the grain size is unfortunately opposed to the creep strength (volume 3, Ill. 12.5-11 and Ill. 12.5-16). This means, always a compromise must be found.

Vibration fatigue can be in connection with a change in the natural frequencies (danger of resonance). For example the orientation of a single crystal of a turbine blade plays a role (volume 3, Ill. 12.6.3.4-7). Thereby often the influence of the thermal expansion will be ignored.

This shows the following example: Heavy corrosion prevented the repair of a thin walled inner radiation protector around the shafts, made of steel sheet. This was a part of an elder aeroengine type in military operation. Inspite the corrosion pits, at this old steel sheet version no fatigue cracks have been observed. Then a corrosion resistant alloy was used. However, after longer operation periods, at those new parts from sheet of a Ni alloy fatigue cracks and break-outs accumulated. The investigation showed, that this crack formation obviously must be seen in connection with a tensioning by thermal expansion during operation. This probably did not occur at the steel parts because of a different thermal expansion. This effect was not considered, because of the elastic design an adjustment of the tolerances and clearances seemed not necessary. So the assumption was obvious, that an increased mean stress decreased the usable fatigue strength. So also a vibration exitation by the neighbored combustion chambers could have more influence.

Danger of spontaneous fractures: Possibly weldings may behave in a case of containment (high deformation speed) not ductile enough. This worsens the punching/penetration strength.

A decreased resistance against crack propagation (decreased fracture toughness), e.g., caused by the formation of brittle phases during the repair or later during operation, can promote forced fractures (e.g., „haircut“ of turbine blades).

Problem of the non destructive testing (NDT): The transition to a little more coarse grained material can hinder the ultrasonic inspection. This can mean larger incipient flaws and with this, an increased failure risk.

Worsened suitability for repair: To this belongs the susceptibility for heat tears. The cause can be an unfavourable heat treatment or a long time operation. Such an effect makes a weld repair more complicated.

The failure risk, because of material faults and weak points, is among others influenced by the producer of the rough parts, the casting parameters, forging parameters and the non drestructive testing.

Illustrations 21.3.3-3.1 up to 21.3.3-3.4: To sensitize for disimprovement during a repair development, consequences of typical rework at selected parts will be discussed.

Dimensional changes and mechanical rework (Ill. 21.3.3-3.1): As an example serves a compressor rotor blade. The change of a profile can occur during processes like rework, blending, cutting of airfoil corners or an intense shot peening (volume 4, Ill. 16.2.1.6-19). Does this affect the aerodynamic behaviour, a deterioration of of the aeroengine must be expected with effects like

  • drop of the compressor efficiency with consequences like decrease of thrust and rise of the gas temperature.
  • Decrease of the surge margin with the danger of surge events (e,g., during unfavourable start conditions). Thereby components can be damaged (volume 3, Ill. 11.2.1.2-1).Increased vibration load of blades by
  • rotating stall (can hardly be identified from outside) with the danger of a vibration overload (volume 3, Ill. 11.2.1.2-6).
  • Reduction of the cross section (blending, rework at the airfoil) can change the natural frequency in a manner that resonances occur.
  • In the blending region the vibration stress can rise, caused by an, even if seemingly small, notch effect. Additionally it is thinkable, that by an unsuitable or imperfect blending technology, unnoticed damages occur like
  • smeared up grooves/scratches.
  • Drop of strength, caused from too high machining temperatures.
  • Internal tensile stresses.

With this the fatigue strength can dangerously decrease in the blending area.
Rework at the root section is frequently limited to smoothening/cleaning and regenerative shot peening. Deviations of the shot peening parameters, like impigne angle (Ill. 21.2.5-2) or the roundness of the shot (volume 4, Ill. 16.2.1.6-11), can prevent the expected regeneration effect. In an extreme case, an unfavourable peening process even acts damaging (volume 4, Ill. 16.2.1.6-14 and Ill. 16.2.1.6-15). Such a drop in fatigue strength can cause fatigue fractures in the root. In modern compressors the probability increases, that the normal root stresses are very near the bearable limit. For safety, possibly in such a case helps an extreme tight dimensional tolerance of the contact surface geometry. So can already smallest dimensional changes act dangerously at the load distribution on the root (volume 4, Ill. 16.2.2.9-4).

A build up of the blade tips by welding and the rework of a new profiling can lower the fatigue strength and promote fatigue cracks („lyra mode”; volume 3, Ill. 12.6.3.1-6 and Ill. 12.6.3.4-10).

Applying and removing coatings (Ill. 21.3.3-3.2) can deteriorate the operation behaviour of a part in an unexpected manner. Besides the effects, already discussed, under dimensional changes there are especial specific:
-A low fatigue strength of the coating transition to the substrate (notch effect, corrosion) and/or the coating itself can promote fatigue fractures.

  • Crack formation and chipping of brittle, hard coatings (e.g., hard facing erosion protection) caused by FOD can lower the fatigue strength dangerously (volume 3, Ill. 12.6.3.4-19.2).
  • Basically a later repair, which demands a stripping/removing of the coating, must be planned. Thereby the base material must not be intolerable influenced or the removal change the specified geometry.Welding (Ill. 21.3.3-3.3): The influences of a welding repair on the operation behaviour of a part (chapter 21.2.1) must be absolutely considered. They are discussed in detail in volume 4, chapter 16.2.1.3, concerning the production of new parts.

Brazing (Ill. 21.3.3-3.4): With the problems of repair brazings the chapter 21.2.2 is engaged. Further braze processes are discussed in chapter 16.2.1.4 of volume 4.

Illustration 21.3.3-4: Volume wear / removing wear at guidings, seats, adjustments and bearing bores is a failure seemingly easy to control. The experience teaches: There is hardly a problem that so insistent detracts from a remedy. With tis wear problems are predestined for disimprovements.

A peculiarity for many wear problems is, that they attract attention not before longer operation times. Have the wear processes been influenced by a repair, already many delivered and assembled aeroengines can be concerned. With this the consequences can get unexpected dimensions.

A typical wear system consists of three „elements“: The both contact surfaces and the acting surroundings/environment. This for example can be corrosive (volume 2, Ill. 6.2-13), oxidizing, abrasive, lubricatng or arrestive/sticking (volume 2, Ill. 6.2-13). Thereby parameters like the operation temperature, time lapse and mechanical loads (bearing pressure) play a role (volume 2, chapter 6.1).

Trying to protect the worn surface with a hard coating/hard facing as a remedy, it must be expected, that now the partner surface wears (volume 2, Ill. 6.2-15).

Seemingly little changes of a coating/plating like porosity in size and type (clused, open), can already change the wear behaviour essential. For this reason it must be urgently advised against the change to cheaper or just available prosesses, similar to the specified process without sufficient proofs and tests.

Lubrication media which prove at best in a dry atmosphere, can block in a short time sliding joints with wear products in sea atmosphere (volume 2, Ill. 6.2-15). This is a known effect for MoS2 containing lubrications (Ill. 22.4.1-1).

Operation temperature and oszillating movements (volume 2, Ill. 6.1-18) play an essential role for plug connections of hot parts like combustion chambers (volume 2, Ill. 6.2-11) and turbine stators.

The consequences of wear problems can be catastrophic.
Grows the clearance of the bore from a guiding slide bushing of a variable compressor stator blade too much (sketch above left), aberrances of the design determined angle of setting from the airfoil can occur. Just modern compressors with transsonic flow, are frequently especially at this susceptible (volume 4, Ill. 16.2.2.9-5). Is the whole adaption mechanism of a stage concerned by the wear, this will influence the efficiency and the operation behaviour (e.g., surge margin). Additionally stalls can excite dangerous vibrations of the airfoils.

The sketch above right shows a case of wear at the outer fixation of the compressor stator blades (volume 2, Ill. 6.2-8). As consequences it must be reckoned with a catastrophic compressor failure.

Especially turbine stators/nozzles are subject to high circumferential gas loads. To prevent a rotation of the stator form-fit latchings are used. Often bolts (sketch below left) or fixing lugs are concerned. Because on these joinings severe vibrations and large thermal expensions act, they are subject to fretting wear. This wear is accelerated by the cooperation of abrasion and oxidation. So securing crosss sections are weakened up to failing. Several cases emerged, when a released turbine stator rotated and milled through the casing. The result was the fracture of the damaged rotor blades behind the stator. So fragments exited (uncontained) which endangered the the aircraft fuselage (volume 2, Ill. 8.1-8.2).

A similar problem can occur at fixing lugs and guidings of other hot parts. The sketch below right shows the outer shroud of a turbine stator blade segment, mounted at the casing. Also here wear can release/loosen components like labyrinth supporters or stators with extensive consequences.

Illustration 21.3.3-5 (Lit. 21.3.3-2): On at least five helicopters, failures of one or both aeroengines got known (Lit. 21.3.3-2). Thereby the turbine wheels of the gas generator reached overspeed (normal rotorspeed 52000 rpm) and burst (Lit. 21.3.3-3). Cause was the fracture of the about 25 cm long shaft to the compressor. Between the shaft fracture (volume 1, Ill. 4.5-4) and the bursting of the turbine wheel with the exit of fragments (uncontained) laid only 0,02 seconds.
As cause of the failure, coking on the shaft was identified. This was the result of insufficient cooling. When the coke bridged the gap to the outer concentric hollow shaft, damage occurred.
As remedy a redesigned shaft was introduced. A smaller diameter increased the gap to the outer shaft. This enabled a better oil flow. Additionally one end of the shaft got little notches. At those fatigue cracks formed, which at least lead to two accidents. So this was a typical case of disimprovement.

Ill. 21.3.3-6: This repair took place at a 2nd stage turbine stator of the gas generator (sketch below right) from a small gasturbine (sketch above). The original was an integral casted part (sketch below middle). This showed operation damages in the cylindric part above the turbine wheel of the 2nd stage. A repair should make the relatively expensive part reusable. For this the deteriorated region was removed and replaced by a brazed in ring of sheet metal (sketch below left). During operation this sheet metal ring distorted and deteriorated the blade tips of the turbine wheel which sowed externally with a performance drop.

As cause of the distortion come into consideration not sufficient matched thermal expansions and a lower creep strength of the sheet metal, compared with the original casting material.

Ill. 21.3.3-7 (Lit. 21.3.3-2): The rotor blades of the rear low pressure turbine (LPT) stage of this old military turboprop aeroengine type (sketch above), showed at the blade surface sulfidation (sketch middle left, volume 1, Ill.5.4.5.2-3). To avoid for the future logistics problems with these difficult replaceable/supplied parts, the sulfdation should be stopped. The complete removal of the deterioration by sulfidation should avoid its acceleration (volume 1, Ill. 5.4.5.1-4). For this the concerned regions have been mechanically removed, accordant to an especially edited specification (middle sketch). After this the blade was preventative supplied with an Al diffusion coating.

The influence of the diffusion coating at the fatgue strength was investigated with vibration tests at blades. These showed, that about 10% drop of the fatigue strength must be expected (diagram below right). However this seemed bearable, because the deterioration of already occurred sulfidation was estimated of the same magnitude, but till then did not cause the failure of a part.
From time reasons and cost reasons in accord with the certifying authorities, an operation testing in fleet leaders was disclaimed (page 21.3.1-2). So at once the series repair started. After about one year, when already a higher number of aeroengines have been delivered with repaired blades and used in flight service, a blade fracture occurred. The investigation showed a HCF fatigue crack just above the root platform. The vibration crack started from the trailing edge in the region of the sulfidation removal (detail middle right). In shorter time intervals, further comparable cases at repaired blades followed.

A vibration analysis showed, that obviously in several cases the thinning of the blade airfoil was out of the specification limits. This caused a, even if seemingly little, drop of the natural bending frequency (diagramm below left). Unfortunately this was now excited to resonance vibrations at about 100% rotor speed. This dangerous closeness was not known respectively not aware. The increased vibration load in combination with the coating caused drop of fatiguw strength promoted the failure.

Detailed enquiries showed, that already in former times at not repaired blades, i.e. new parts, comparable blade fractures have occurred. The minute documentation of the failure investgations of these cases showed, that blades have been concerned with a production caused (forging) too short chord length. The analysis in time of the reports could have been interpreted as a warning hint, that the blades already always have been operated near a resonance.

Measures and remedies:

  • Because of logistic reasons not at once at all concerned aeroengines the blades could be inspected and if necessary exchanged. So meantime the two engined aircraft was only allowed to be operated with one suspect aeroengine and the other without repaired blades.Blessing in disguise was, that the most repaired blades could be reused after the following measures:
  • The diffusions coatig will be stripped and no more used.
  • All blades will be dimensional checked.
  • Replacement of suspect blades by new parts.
  • In the future, before a repair development, the failure documentation (former reports) must be critical analysed for potential problems with the concerned part.

Illustration 21.3.3-8 (Lit. 21.3.3-4):

Prehistory:

Already almost 20 years before this accident (frame above) fatigue failures by vibrations (HCF) at the disk of the 14th compressor stage (high pressure compressor, detail below left) had occurred. One happened during operation and one at a new aeroengine on the test bed. In both cases aeroengines with changes, which caused a looseness of the outer radial centering seat, are concerned between 12th and 14th stator stage. Additionally the seal surface of the labyrinth (inner airseal) of the 13th stage sator has been axially shifted.

An investigation at the OEM showed at that time as cause a vibration of the labyrinth during high aeroengine thrust. A loosened seat between 12th and 13th stator stage, caused by wear, permitted larger vibrations of the stator. These excited vibrations of the 13th stage rotor disk.

Measures and remedies: After this awareness, a design change which was made responsible for the loosening of the stator centering, was retrograded. Additionally the assembly specifications have been adjusted. This was applied for all aeroengines with the unfavourable configuration. Also the disk geometry was optimized.

These measures lowered the vibration load to a bearable limit. Unfortunately now at low aeroengine load, an exitation, even if low energetic, occurred.

Acute failure case:
A segment with 20 rotorblades was broken out at the rim above the snap centering (sketch below right). The fracture surface showed a crescent-shaped fatigue fracture zone which, correspondig to the pronounced lines of rest, started at the rear surface of the disk. The further investigation showed, that also a second comparably crack existed at the rear side of the disk.

At the fracture origin a marking (feed line) in the transition radius of the feed motion from the chipping tool of the rework could be identified (see measures and remedies in the prehistory). This rework obviously lead to an aberrance of the tolerance at the centering seat, which was outside the drawing instructions.

Further it was discovered, that on the centerings of the 12th and 13th stage a specified hard coating lacked. This was a fault of a former repair/rework during an ovverhaul.
The wall of the mounting flange at the inner shroud of the stator of the13th stage had 0.5 - 0.7 mm undersize.

Stator and rotor showed similar fatigue cracks.
The front edge of the 13th stage stator inner shroud lacked.
According to the experiences of the prehistory, these aberrances have been evaluated as sufficient for a dangerous vibration exitation of the compressor disk.
Also in the other parts of the aeroengine like the 12th stage stator and the diffusor casing (compressor exit casing), fatigue cracks have been found. Possibly the cracked diffusor promoted the exitation of the disk. Other cracks have been classifyed as secondary failures.
As causative for the remained or new (?) vibration exitation during low (?) aeroengine load, in the discussed case also the „improved” configuration is valid (see `measures and remedies of the prehistory). With this also in this case we deal with a disimprovement.

References

21.3.3-1 „S-76 Awaits Engine Shaft Modification“, Zeitschrift „Interavia AirLetter”, No. 10,653 - December 14, 1984 - 4.

21.3.3-2 „Allison Fix for S-76 Engines“, Zeitschrift „Interavia AirLetter”, No. 10,675 - January 23, 1985 - 5.

21.3.3-3 NTSB Identification :FTW84FA216, „Turbine Assembly Shaft Fatigue“, page 1.

21.3.3-4 TSB Canada, Aircraft Accident Report 84-H40003, „Uncontained Failure, 22. March 1984, Boeing 737-200”, page 1-27.

© 2022 ITTM & Axel Rossmann
21/213/2133/2133.txt · Last modified: 2020/06/25 22:43 (external edit)

Page Tools