Aeroengine Safety

Fig. "Examples of rivets in aeroengines 1" und Fig. "Examples of rivets in aeroengines 2" (Lit. 23.3.2-23.3.2-2 up to Lit. 23.3.2-23.3.2-6): Rivets find especially in elder aeroengine types different applications. They use the specific properties and advantages of a riveting. In the aeroengine engineering typical applications of rivets are:

Assembly, respective connections of sheet metal parts (frame below): A fomer design principle was, to produce components as welding design from a multitude of single parts. This principle found, especially for turbine casings, an application (sketch below left). Aerodynamic contures in the hot gas stream have been realized with riveted sheet metal structures. This design allows, to disassembly them during overhaul. For this the rivets are removed. So the bearing inner structure gets accessible for an inspection, respectively non destructive testing (fluorescent penetrant inspection = FPI ). After this procedure, the sheet metal jacket can be again asembled with a riveting. This technique has proven as excellent. However it demands experience and extensive testing to prevent sufficient certain fatigue cracks by gas oscillations and thermal fatigue (volume 2, Ill. 7.2.2-18 and Ill. 7.2.2-19).
Advantages:

• Easy disassembly and assembly of sheet metal structures. Riveted wear parts like of a flame tube with damagey from overheating (sketch below right) or zones if combustion chambers, deteriorated by fretting can be easy exchanged, using a riveting.
• Vibration damping effect of the friction between the connectet sheet metal surfaces.
• Tolerance for cyclic thermal expansions of the elastic riveted sheet metal structures.

Disadvantages:

• Extensive mounting and dismounting, compared with one piece bearing structures (e.g., castings, milled forgings).
• Intense proving in operation.
• At several designs danger as resonator of a vibration exitation (volume 2, Ill. 7.2.2-20).
• During service, no direct visibility and testability of the jacketed structure.

Fixing of a position (above left): Some OEMs use rivets for the fixing/locking of the axial position of turbine rotor blades („bladewalking“) in the disk (example 23.3.2-1 and volume 2, Ill. 6.2-5.3). Obviously the axial forces at the rivet head are sufficient low to guarantee the demanded safety. Anyway, always again cases emerge, at which this locking by riveting failed (example 23.3.2-1, volume 2, Ill. 6.2-5.2). It came to extensive secondary failures:

• FOD/OOD from loosened rivet fragments.
• Rubbing of the shifted blade.
• Vibrations respectively unbalances at little blade displacement in the range of clearances/tolerances (examole 23.3.2-1). Danger of seal damages/failures (e.g., labyrinths) and bearing failures.

Cause for the failing of the rivets is usually the rivet head. Typical triggers are:

• Too short upsetting (length, diameter),
• During upsetting unnoticed crack formation below the head (Ill. 23.3.2-1),
• Weakening by fretting.
• Unusual high axial forces as secondary failures.

As remedy, obviously to get a better load distribution, in one case under the rivet head a washer was positioned (volume 2, Ill. 6.2-5.2). Because of the mentioned problems, it seems that at new aeroengine types this axial locking of the rotor bladses is already no more used.

One sided fixation of axis and levers (sketch above right): Typischcal application example is the actuating of variable compressor guide vanes (Fig. "Risk of a riveting deviations"). Thereby one axis at which a hinge is positioned one sided is riveted into a lever. In this manner the riveing transfers bending forces, in contrast to its usual function. At unusual high actuating forces,e.g., caused by corrosion and FOD, it can come to the overloading of the riveting. a changed blade position can trigger by flow distributions extensive compressor failures (Fig. "Risk of a riveting deviations").

Anti-twist device: Rivets offer itself,because of its relatively simple replaceability as mechanical locking of lugs against twisting. With this the secured parts will not be damaged. Must a higher number of nuts, e.g., of a flange connection be secured against twisting and positioned, suitable formed sheet metal ribbons which will be riveted, are used (Fig. "Examples of rivets in aeroengines 2" left).

The fixing of special versions of of nuts can take place by riveting (riveting nut/clinch nut).

Sometimes the closing of bores is needed, to seal for example stop-drilled cracks in a casing wall or to influence the airflow through bores. Naturally such measures are only allowed according the specifications/istructions of the OEM. Here is must be kept in mind, that thermal expansions and fretting can lead to the loosening/separating of the rivet. Naturally the rivet must not suchlike pressed into the bore, that it promotes a new crack formation.

Balancing of rotors can occur with rivets in different ways (Fig. "Examples of rivets in aeroengines 2" right). The rivet can be introduced directly in a radial circumferential collar. Also a balance weight can be fixed with an axial rivet.

Vibration damping/mistuning: This application should only be mentioned for completeness. It is not known in modern aeroengines and today rather a provisional solution. Concerned are loose-fitting rivets which as so called rattling/clatter rivets art placed in the region of the blade tip.

Fig. "Problems of riveting technology" (Lit. 23.3.2-2 and Lit. 23.3.2-5): Failing riveting connections trigger serious secondary failures, depending from application and function (Ill. 23.3.2-1). To these belong:

• The fragments of the rivet act as foreign objects and produce FODs.
• The outage of the rivet function leads to the offset of fixed/locked parts (example 23.3.2-1).
• Drop out of the function of parts (Fig. "Strength limits of pop rivets"). An example is the failing of an actuation of variable compressor guide vanes (Fig. "Risk of a riveting deviations").The failing of rivet connections or of components, weakened by the riveting, mostly is caused by flaws in the rivet and/or the riveting process. Rather seldom is the case of an unusual high operation load.

The riveting itself and with this, primarily the production of the closing head, can be cause of flaws which lead during service to a failure. These are in many cases process specific. Radial cracks develop, when the bore is overloaded by the expansion of the river shaft. This can trigger a failing of the riveted cross sections (sketch above, Fig. "Risk of a riveting deviations").

At conventionel solid rivets, primarily two processes are used:

Hot-riveting is suitable for the application of many rivets like at sheet metal structures of elder aeroengine types (Fig. "Examples of rivets in aeroengines 1"). This can be a converted spot welding machine. With time controlled contact pressure and circuit continuity (resistance heating) the closing head of the rivet is formed (Ill. 23.3.2-3). Appear problems during riveting, these show mostly in an, from the outside good visible radial crack formation at the closing head (sketch above left). Problematic are especially from the outside not visible flaws and cracks. These are mit not sufficient certain detectable with, for series application suitable, non destructive testing methods. Concerned are deformation folds, wrinkles, creep cracks respectively forced cracks or hot tears (volume 2, Ill. 7.2.2-9.2 and volume 4, Ill. 16.2.1.3-10). Typical crack positions are at the transition from the closing head to the shaft (sketch above right). But during unfavourable temperature distribution, it can also came to the formation of hot tears in other levels of the shaft cross section.

Cold riveting is used with hand or hand-guided tools at difficult accessible locations. Here especially much depends from the executing person. Thereby the behaviour and the appearance of the rivet head durig the forming is crucial for the evaluation of the final result (sketches above, seat/possible clearance). Requirement for this is experience. Even seemingly little abnormalities at the new rivet, before riveting, schould be checked for the cause and effect. To these belong (shaft) length, shape and surface. During forming of the head, deformation behaviour and surface should be observed. This is especially true for rivets, which serve an axial fixing/locking, e.g., of turbine rotor blades.
The closing head of solid rivets or one sided hollow rivets also forms with the expanding (volume 2, example 6.2-1.1). This is similar the pop rivet/blind rivet. However in this case the head is here at the accessible side. For a better contact/support sometimes washers below the head are used.

Fig. "What riveting quality needs" (Lit. 23.3.2-2 ): Riveting as one of the oldest connection procedure is simple and proven. Anyway just in a time, when then manual skills seem to have taken a back seat, it holds problems not to underestimate. Aso if these appear trivial, its prevention demands experience. That is as well true for the riveting process (Fig. "Problems of riveting technology") as also for the subsequent evaluation of the serviceability (pop rivets see Fig. "Problems with pop rivets").

The most problems (frame above) are visible from the outside or can attract attention during the riveting proces (frame above). Thereby also the clearance of the rivet shaft in the bore is important. This demands „feeling“ together with sufficient experience. Hints at not fitting rivet length must be taken seriously. If necessary, this must be reported to a person, which is authorized for an evaluation.

In the frame below it is pointed from experience at problems during hot-riveting with an adapted spot-welding machine. This processing is used for rivets from high temperature alloys like high alloyed age hardening steels (A 286). Here the exact adherence to optimized and confirmed process parameters of the machine, plays an important role for the security of the riveting. As well the tool, as also the condition of the rivet, must comply the condition, with which the proof was carried out. Deviations, which change the deformation process (wear of the head forming tool) or the heating, electric current (oxidation) must be seen critical. Appear deformation problems and crack formation, the rivet material must be checked for a suitable, respectively specified structure condition (e.g., age hardened. solution annealed).

Fig. "Problems with pop rivets" (Lit. 23.3.2-5 and Lit. 23.3.2-7): Riveting with pop rivets (blind rivets)has definitely advantages. However it is from experience in the surrounding of an aeroengine extremely problematic. These rivetings can be carried out easily, without higher effort/forces. Therefore they offer itself for smaller, fast repairs of sheet metal components. A special characteristic of these rivets is, that the closing head forms at the opposite side of the riveting tool (rivetting tongs, sketch above right). So it can be withdrawn a visual testing (Fig. "Problems with pop rivets"). With this if the bore diameter is too large, the danger exists, that of the opposite the mandrel head is pulled into the bore or even through the bore. This is promoted by `soft' material (e.g., light metal, plastic) compared with the rivet material. The rivet head will be formed corresponding bad (sketch below left). In such a case the acute danger of a slip out/pulling out, i.e. failing of the riveting exists. In an extreme case the back sheet metal will be not at all gripped by the rivet head (sketch below right, Fig. "Strength limits of pop rivets"). Such rivetings are only allowed when they are explicite specified in the manual.

Potential dangers with blind rivets:

• During riveting the rivet mandrel will be function necessary torn off. A remainder can by mistake stay in the aeroengine or the intake region. Then extensive foreign object damages are the result (volume 1, Ill. 5.2.1.2-11.2).Will the scheduled grip of rivet/shaft length exceeded, the possibility exists, that the mandrel head falls off or spalls off.
  The mandrel head can be also loosened by a false setted riveting tool (compressed air) . This is the case, if the rivet mandrel is pulled too fast, respectively with too much force. The result is, besides a loose rivet, also FOD danger.

• Bad seat, respectively loos seat of the rivet, can usually traced back to a too low compression force at the closing head during riveting. This can have several causes:
  A too small bore diameter can hinder the pshing through of the rivet up to the full closing head contact. Preferential, then the transition to the closing head contacts the shaft.

• Loosens the mandrel head in the rivet by wear (fretting) because of vibrations and or corrosion, mandrel head and rivet can get loose and separate and get into the aeroengine. Also in this case extensive FOD must be expected. This situation will be dangerously promoted by the use of a rivet with too short grip (clamping length).
  Is the grip too large for the riveting cross sections, the rivet can also sit loose.

Corrosion problems (galvanic corrosion/aqueous corrosion) are promoted by different noble materials of the rivet and the sheets to join. The necessary electrolyte develops under service conditions with the access of condensate, especially in sea atmosphere (table salt). The corrosion products like from Al

• and Mg-alloys, can, because if its enlarged volume, trigger a

blasting effect in the region of the rivet bore. Do the corrosion products crumb off, the rivet gets loose. Pits may promote with its notch effect in high loaded zones, the formation of cracks in rivet bores. Additionally corrosion can accelerate the growth of cracks.

Is the opposite of this, therefore also called blind riveting process, not visually controllable/testable,the danger of an unnoticed faulty closing head rises. For example this is the case in a closed interior space. An unsufficient formation can in service during vibrations and fretting, lead to the falling out of the rivet (sketch below left).

Remedy:

• Precondition for a faultless blind riveting is the right, as specified bore diameter.
• Correct selection of the rivet material. Is this, compared with the material of the bore too hard (e.g., stainless steel in plastic), the danger exists, that the mandrel head will be pulled through.
• Correct grip (-length) of the rivet. It is necessary to prevent, that the rivet shaft does buckle (too long) or the closing head is too small (too short). To avoid corrosion as possible, the rivet material should correspond the material of the sheets to join. Most save is an electric insulating intermediate layer, like lacquer.

Fig. "Strength limits of pop rivets" (Lit. 23.3.2-8): A first investigation of the failed aeroengine showed, that half of the cowl (`transcowl') from the (bypass-) thrust reverser at the outer side (middle frame) has separated. Thereby obviously four flaps (blocker doors), with elements of the actuating mechanism (mounting of the adjusting nut, frame, hinges) have been entrained. A detailed investigation of the failure area at the responsible aviation safety authority showed:
Preliminary remark: The description of the complex content in the available literature was generated correspondent the understanding of the author. Therefore it can deviate in some details.
The lower transcowl hinge had totally separated without signs of an overload. The heads of the four fixing rivets still resided at the scheduled place. Concerned are blind rivets (detail below right). Thereby they differ from all other hinges with solid rivets (detail above right). All rivets had countersunk heads. The countersunk heads of the obviously, during an exchange inserted blind rivets, have been deep pulled into the countersink. The solid original rivets for the mounting of the actuator connection at the casing wall (transcowl, sketch below left), still stuck in the flange (detail below right). They had failed by shear. The inside positioned, from the outside not visible, closing heads had emerged with fretting traces below the flange (sketch below left). They showed, that the flange was pressed against the rivet heads in an offset position, but not gripped. So it came at the, by the thrust reverser actuations higher loaded components, to fatigue fractures.

Conclusion: Obviously, the original solid rivets have been sheared some time ago by an overload. During the following maintenance, they have been exchanged against the specifications, with blind rivets. The flange with the rivet bores has slighly displaced adverse the mounting wall. So the blind rivets could not be introduced into the inside located flange bores, which can not be seen from the outside. Unnnoticed, the inside located flange of the actuator connection was not gripped by the riveting. The offset and the so caused developing fatigue cracks at other bearing parts lead to a tilting of the sensitive actuating system (spindle and adjusting nut). Extreme actuating forces occured, which caused the further failures. This sequence can be concluded from previous cases at which the actuating system failed due to adjustment shortcomings.
Obviously the function and load of the riveting was not known or aware to the performers of this repair. The blind rivets would have been merely suitable for a sealing. For the transfer of the mounting loads they have been inapplicably. Here the specified solid rivets should have been used.

It is interesting, that in the maintenance documents of the operator from the last three years, no hint at such a repair can be found.

Fig. "Risk of a riveting deviations" (Lit. 23.3.2-4 and Lit. 23.3.2-9): After the incident the failed aeroengine was subdued an outer inspection. There were no signs of an especial loading, e.g., of a bird strike, ice ingestion or an other FOD.
However a broken lever of the actuation from the variable stator vanes (= VSV, detail sketch right) from the 1st stage of the intermediate compressor (3 shaft aeroengine, sketch in the middle). Also during the investigation after the disassembly of the compressors from the core engine, the crack inspection showed no further cracks in the actuating levers. Several blades of the compressor showed FODs of small metallic fragments. In the 2nd compressor stage, two rotorblades without FOD notches showed fatigue breakouts at the tips. In the high pressure compressor a rotor blade of the 1st stage had failed from a fatigue fracture (volume 3, Ill. 12.6.3.4-10). This obviously started from such a FOD notch at the leading edge.
The fracture of the actuating lever propagared transverse to the arm through the contact surface of the bore, from the riveted axis (sketch below left). A fatigue crack was concerned, whose origin was at the bore surface of the rivet shaft.

Results of the laboratory investigation: Caused by the upsetting, the diameter of the axis expands during riveting. With this in the region of the rivet bore, high tension loads can develop. These promote as prestresses a crack by fatigue.
At further four actuating levers of the concerned stage, microscopic seizing (welding, fusion) with many microcracks in the millimeter range, could be observed. They were positioned in the transition region of the actuating lever to the riverting axis (detail in the middle). As known weldings/seizing between steels (axis) and titanium alloys (lever) form brittle crack susceptible phases. The appearance of observed cracks in the micro section correlates such an embrittlement. Obviously the cracks stay in an causative connection with the production of the levers and the riveting. Such notches can lower dangerously the fatigue strength. This may be enough to trigger a fatigue crack during normal operation loads.

History: Till now there was no fracture of a lever from the 1st compressor stage, in contrast there have been two cases at the 2nd stage. The cause have been unsuitable rivetings in the production. These lead to a dangerous fretting wear (titanium alloy!) at the levers. As result in both cases violent flow dusturbances triggered fatigue fractures of the rotor blades from following stages.

Conclusion: The fatigue crack of the actuating lever is seen in causative connection with high tensile stresses in the bore region from the riveting process. After the fracture, the vane took the „position shut”. From experience (see section `history'), thereby intense flow disturbances develop. They lead to vibrations with fatigue break outs at the blade tips of two rotor blades. These small fragments have been oscillating initially in the first stage. Then they passed the compressor and damaged several further compressor blades. At such a damage, also the fatigue crack of a high pressure compressor blade, can be traced back.