21.2.1 Repair welding.
Welding processes of the repair are usually those of the new parts production (volume 4, chapter 126.96.36.199). Complicating for repairs (Lit.21.2.1-9) are additional effects when the material changed during operation and so no optimal welding conditions exist. Also during welding cracks with oxidation or corrosion residues an annoyance of the seam must be expected. Repair weldings are especially applied for welding designs from sheet maerial and forgings (casings, combustors). For Ni alloys (Fig. "Expection of fatigue strength from repair weldings"), because of the proneness for micro-hot cracking (Fig. "Welding of creep resistant alloys"), welding repairs must be evaluated more problematic as for steels. In some cases wear surfaces at blade tips (Fig. "Examples of welding repaired blades") or at contact surfaces from the tip shrouds of turbine rotor blades (Ill. 21.2-5) are build up by welding. A peculiarity are welding repairs of fan blades for the repair of FODs (Fig. "FOD welding repair of blades"). This is notably remarkable because it is successfully used since years at these highly dynamic loaded components (Fig. "Proof of repair welding process").
Illustrtaion 21.2.1-1 (Lit. 21.2.1-1 up to Lit. 21.2.1-4): For a long time welding repairs are used at fan blades of big bypass aeroengines. Primarily concerned is the welding of patches in damaged zones which have been machined before. This repair process needs more experience and demanding proof tests before it can be applied as could be expected at the first glance (lower frame). First it must be kept in mind, that it the fatigue strength of the base material from a fusion welding can not be estimated because of its casting structure. In the region of the welding and the neighboring recrystallisation zones suspend the advantage of an oriented forging structure as well as a deformatuion caused strengthening. So the structure step/transitiion has a certain notch effect (`structure notch' , volume 3, Ill. 13-18). At the other hand a probability of a process caused weak point or of a fault exists, even if it is very small (Fig. "Extent of welding repair"). To design a welding repair at a dynamically high loaded component like a fan blade the part specific, potential dangerous vibration modes with its nodes must be known (volume 3, Ill. 188.8.131.52-5 and Ill. 184.108.40.206-6). Thereby it must be considered, that the high vibration loads must be expected in the neighboring `vibration loop zones'. The thus identifiable cross section depending lower vibration loaded zones of the nodes rather offer a location for a welding.
The weldings mostly are carried out by electron beam welding (volume 4, Ill. 220.127.116.11-24).
Today bigger fan blades are often carried out as hollow designs. Such blades allow the welding of a „patch“ only in the massive region of the edges.
Damaged „clappers” (snappers) of suitable blades respectively operation loads, e.g., with a deformation caused by an overlapping („shingling“, volume 1, Ill. 18.104.22.168-10) can be replaced by welding.
After the welding and a machining and if necessary straightening rework, the release of internal tension stresses is carried out. Therefore an application specific locally heat treatment with a following shot peening is used.
Of course a very exact non destructive testing of the welding is necessary. Porosity, inclusions and open incomplete fusions (volume 4, Ill. 22.214.171.124-25) can be found by X-ray. To identify kissing bonds (volume 4, Ill. 126.96.36.199-27), these are only partly fused regions (`cold weldings', Bild 23.2.1-3, volume 4, Ill. 188.8.131.52-24 and Ill. 184.108.40.206-32 ), an ultrasonic test is necessary. Also eddy current testing can be applied in such a case.
Once more should be refered: Here also the awareness is valid, that a 100% likelihood for the detection of all faults can not be assumed (Fig. "Proof of repair welding process").
Figure "Proof of repair welding process" (Lit. 21.2.1-1 and Lit. 21.2.1-2): During climb the fracture of a fan blade about 20 cm above the root platform (sketch below right) occurred. The aeroengine showed up to the low pressure turbine excessive secondary damages (middle sketch). The fragment of the blade escaped and damaged the fuselage dangerously.
The in the disk remaining part of the blade was investigated in the laboratory. It arose, that an about 8 mm long fatigue fracture had been originating from a repair weld at the trailing edge of the blade. Obviously the blade is to such an extent subject of high operation loads, that already a 8mm long crack gets critical, that means it triggers a forced fracture.
The blade had after a welding repair 107 operation hours with 32 start-shut down cycles before it failed.
A renowned, approved/certified repair shop had welded a half-moon shaped patch with electron beam at the blade. The vibration fatigue crack started ftom a 0,75 mm deep discoloured zone in this weld.
After the repair the later failed part was subject of four different non destructive tests in the prescribed sequence :
- Penetrant inspection.
- Ultrasonic inspection.
- Eddy current.
The liklihood of detection of the tests shows the diagram down left. From this it can be seen, that from the eddy current testing the highest detection sensitivity can be expected. The opinion of the investigating authority believes, that an existing crack of 0,75 mm should have been found with the eddy current inspection. Why this did not happen could not be resolved.
The repair shop had already repaired more than 1000 fan blades with different shapes and sizes of the patches. However these have not all been subject of the same test procedure. After the failure the OEM allocated as safeguarding measure a further eddy current test at the operator (flight line). Concerned are all electron beam welded repairparts. Thereby a crack at a further fanblade of an aeroengine from an other aieplane type of the same operator was found. This time the crack was located in a repair weld at the leading edge, about 10 cm above the root platform. This blade had already about 4000 operation hours with about 1200 start-shut down cycles.
The investigation by the OEM showed an etched surface around the crack. From this it was concluded, that the crack started in the welding seam or a patch at the leading edge. The broken open crack showed an about 0,6 mm deep dark blue discoloured zone at the leading edge. The OEM estimated it as repair process typical. From this crack a fatigue crack of 0,25 mm proceeded. The little crack growth can be so explained, that in this case the operation load was low. That may be explainable as well with the other aeroengine type as also with the different failure location.
Figure "Welding of creep resistant alloys" (Lit. 21.2.1-3 and Lit. 21.2.1-4): About the reparability of an aeroengine component and with this the total operation costs not at least decides the weldability of the hot parts materials (diagram above left, volume 4, Ill. 220.127.116.11-11).
Typical welding repairs for minimising the gaps at the tips of turbine rotor blades (Fig. "Examples of welding repaired blades") and labyrinths (Fig. "Welding repaired labyrinths"). Worn contact surfaces of the shrouds from low pressure turbine rotor blades are successful weld deposited. Thereby a limited number of micro hot tears in the base material (heat affected zone = HAZ) must be accepted. The welding correlates a stellite hard facing (Fig. "Welding repaired labyrinths").
For the age hardening respectively for creep strength, necessary Al and Ti content, but also Cr and Co influence the weldability. Typical welding flaws are hot tears (volume 4, Ill. 18.104.22.168-10). The addiction to the formation of hot tears additionally affects adverse the typical rub process during function. Thereby also hot tears are triggered by a locally heating (volume 2, Ill. 7.2.2-9.2).
Single crystals and directionally solidified materials (e.g., B1900) can demand special processes or limitations. The multicrystalline non-directional welding structure can also change the natural frequency with a change of the modulus of elasticity (volume 3, Ill. 22.214.171.124-7). Even a seemingly small effect near a resonance exitation can also change the natural frequency cause vibration fatigue. In case of doubt specialists must be consulted, possibly a component test can not be avoided.
The worsened thermal fatigue behaviour by unfavourable oriented grain boundaries can be a further problem.
As a special problem show labyrinth rings made from extremely creep resistent powder metallurgical materials (e.g., Merl). If necessary those can only be sufficiently welded with an especially adapted welding process with extremely little heat input (e.g., with a laser). In this connection it must be poited out, that already the designer must keep an eye at the reparability. So later, during operation, unsolvable problems, also in logistics can be avoided.
Indeed crack formation directly after the welding process can be held in acceptable limits. However during a following heat treatment to increase the material strength (age hardening) an unacceptable crack formation can occur (diagram middle right). At high annealing temperatures cracks can develop by so called strain age cracking (Lit. 21.2.1-5).
For as far as possible effective and universal welding repair new processes are tested as well as existing facilities and parameters are optimised. As an example the diagram below right shows the influence of the welding speed at the crack formation of a certain material. The benificial influence of an increased welding speed may be explainable by a smaller application of energy related to the seam length.
Because for the most hot parts materials it must be rekoned with the formation of micro cracks during welding, this is in limits to be feasible accepted/approved (Lit. 21.2-8). For this a suitable specification is necessary. These limits must be appproved with component relevant samples and suitable typical for the operation in the aeroengines. An approval/certification is carried out by the responsible authorities with the comprehension of the OEM and must then guarantee these limits which are no more non destructive detectable. The monitoring of the welding process and its parameters must then guarantee those flaw limits, which can no more proven by non destructive testing (volume 4, Ill. 126.96.36.199-8).
Figure "Examples of welding repaired blades" (Lit. 21.2.1-5 up to Lit. 21.2.1-7): Welding repairs are primarily used for a recovery of the component efficiencies and with this for the minimising of the specific fuel consumption (SFC). The diagram above left provides an impression about the importance of this task (volume 2, Ill. 7.0-4). Don't underestimate a 0,5 -1 % higher efficiency over a long operation time. The correspondent fuel consumption can decide about the profitability of an airline.
Concerned is primarily the buildup of abraded/worn seal surfaces, which add to the widening of the sealing gaps and increase the leakage losses. Blade tips of the compressor stator and rotor (sketch middle left) and of the high pressure turbine rotor blades (sketch middle right) are especially affected. The repair of the tips from blades and labyrinths will be outlined in Fig. "Welding repaired labyrinths".
An aeroengine must prove after the overhaul the recovered efficiency of consumption and performance measurements as well as a specified operation behaviour (e.g., surge margin and operating curve, volume 3, Ill. 188.8.131.52-8). This is carried out with a calibrated and certified test rig (mostly in the „shops”). Only blade tips with an optimal height guarantee satisfying results. The nececessary length of the blade is in most cases reached by deposit welding.
For compressor blades especial high demands are required for weld quality und optimal rework (profiling by machining). This is up to the high vibration load in the blade tip region. Just the thin profiles of blades from modern compressors with a long chord (wide chord) are prone for high frequency edgewise vibrations. This vibration mode is named corresponding its typical line of nodes as „lyra mode“ (volume 3, Ill. 184.108.40.206-6). In this case fatigue cracks start from the tip into the blade where they turn to the edge (sketch bottom left, volume 2, Ill. 7.1.3-4). Because a flexural mode is concerned, the distance between outer fibre and neutral axis, here the profile thickness, plays a role (volume 2, Ill. 7.1.4-13). The less a weld seam overlaps the profile (detail down right), the lower is the level at the tip edge. Therefore also for aerodynamic reasons it is recommended to align the seam bulges at the blade profile with a machining process (lower sketch far right). This is carried out with especial exact guided processes of grinding or milling.
So a filling weld on a compressor blade tip, known as under operation dynamically high loaded (prone for breaklouts of the edges), must have no flaws. These are not only the usual welding faults. For example embrittlement of titanium alloys can be the result of oxigen uptake (unsufficient shielding gas veil, volume 4, Ill. 220.127.116.11-18) and promote incipient cracks. Nickel alloys, as they are used in the rear compressor and the turbine, tend to micro hot tears (Fig. "Welding of creep resistant alloys").
Also the parameters of the mechanical rework must be optimised for a sufficient fatigue strength. The compliance of these process adjustments must be monitored. Overheatings can minimise the strength and/or embrittle titanium alloy. Changed process parameters or insufficient tools (e.g. dull) possibly induce instead of wanted internal compression stresses harmful internal tensile stresses.
A special case represent tips of turbine rotor blades. The very hot leakage gasstream heats the blade tip during operation extremely. Tereby certain zones are removed by oxidation in the range of up to a millimeter (volume 3, Ill. 18.104.22.168-4). In many cases such lacking cross sections can be build up by welding (Fig. "Welding repaired labyrinths"). However this process is especially demanding because of the bad weldability of the nickel casting base material (Fig. "Welding of creep resistant alloys"). Additonally the danger exists, that a deterioration by oxidation is not fully removed and triggers welding flaws (volume 4, Ill. 22.214.171.124-8 and Ill. 126.96.36.199-19). Therefore is must be basically assumed, that the strength features of a build up weld and of the heat affected zone are worse than the unaffected base material (Fig. "Expection of fatigue strength from repair weldings")
In some cases hard particles are applied with high temperature brazing on the tips of repaired turbine blades (volume 2, Ill. 7.1.4-14). This is necessary to guarantee an acceptable rub in behaviour at the ceramic rub coatings (zirconia) of the sealing segments (volume 2, Ill. 7.1.1-7 and Ill. 7.1.3-16).
Illustrtaion 21.2.1-5 (Lit. 21.2.1-7): The sealing effect of labyrinths plays similar as the blade tips an important role for the specific fuel consumption (Ill. 21.2-4). Not seldom blade tips itself form a labyrinth. An example are labyrinth tips on shrouds of turbine rotor blades (sketches middle right and below). Highly loaded labyrinth tips can be found at rotating closed rings as interstage seals (spacers) in rotors of compressors (upper sketch) and turbines (sketch middle left). Here act cyclic tangential/circumferential stresses out of centrifugal force and thermal stresses (volume 2, Ill. 7.2.2-10 and Ill. 7.2.2-11). So it is important, that the build up weld has a sufficient high strength and internal tensile stresses as low as possible. Also the rub behaviour must be favorable. At the same time the proneness for hot tears must be as low as possible that it is not unacceptable deteriorated/damaged by operation typical rub processes (volume 2, Ill. 7.2.2-9.2).
In many cases labyrinth tips need a protective hard facing. Thereby usually thermal spray coatings from TC(tungsten carbide) in a cobalt matrix or alumina (volume 2, Ill. 7.2.1-8) are concerned. This coating servs on one hand, that during rubbing as little friction heat as possible develops. This demands a sufficient roughness and „cutting ability” of the coating. At the other hand the heat input into the bearing base material and so its heating in the rubbing zone can be minimised by a lower heat conductivity (insulation effect).
Illustrtaion 21.2.1-6 (Lit. 21.2.1-8): This example shows, that strength losses can be expected from a weld, especially a repair weld at Ni-base alloys with bad weldability (Fig. "Strength of base material and welding").
At the trailing edge (sketch middle right) of a rotor blade from the 1st stage of the high pressure turbine (sketch middle left) near the tip a piece of 15mm x 20 mm broke out. This entered into the low pressure turbine and triggered there a „haircut“ („avalanche manner”, volume 2, Ill. 8.1-10). This caused in the next turbine stages similar catastrophic damages.
A detailed investigation of the primarily failed blade and the other blades of this stage followed. All showed similar radial cracks above the tip cap (Sketch down left). The crack mechanism was identified as thermal fatigue (TF, volume 3, chapter 12.6.2). One feature was the typical V-shape which could be explained from the oxidation during slow crack propagation and the chipping of the oxides (volume 3, Ill. 12.2.1-12). At the primarily failed blade those cacks are, different to the other blades, deeper. They penetrated an intermediate welding layer below the tip-cap-build up weld (pictures of the microsections below right). The crack propagation through the intermediate layer can be traced at at a cyclic bending load of the blade during operation.
The material of the build up welding of the tip is René 80. This is an age hardening Ni-alloy with a high content of Al and Ti (Fig. "Welding of creep resistant alloys"). In contrast the welding intermediate layer below the tip-cap-weld is from Inconel 625. This is a weld metal, different to the base material. It is not age hardening and therefore with lower strength (creep and cyclic fatigue, Fig. "Strength of base material and welding"). Its oxidation resistance is rather worse. However it is ductile and so can plastically compensate strain differences between tip-cap-build up weld and base material. So obviously the bad weldability of René 80/125 should be equalised, to control the hot tear problem. Anyway those materials are only weldable with tight parameters.
However the lower strength and worse oxidation behaviour may have promoted the operation cracks in the build up weld.
Comment: The failure triggers the impression, the repair variation was not sufficienly tested for this application and was still in the proof phase.
Illustrtaion 21.2.1-7 (Lit. 21.2.1-10): This comparison of the strength between base material and welding material let identify the following tendencies: In comparison with increasing strength of the base material the weld declines. In contrast this effect can not be identified at the not so strong but pronounced ductile materials like In 625. This characteristic of the In 625 is used at repair welds for a compensating intermediate layer at high strength casting alloys (Fig. "Expection of fatigue strength from repair weldings")
The higher the strength of the base material the more pronounced is the drop of strenth of the weld that must be expected.
Figure "Vibration fatigue of tack welds" (Lit. 21.2.1-11): During a training flight the incident occurred. About one hour after departure the low pressure turbine with the thrust reverser separated from the middle aeroengine. However the airplane could land safe.
Obviously since the introduction of this aeroengine type into series there have been frequently vibration problems in the region of the HP turbine bearing chamber (middle sketch). Through the hub bores of the HP turbine disks a with the rotor rotating turbine pressure tube is induced. It serves as duct for the cooling air. Also this tube was subject of dangerous vibrations. For this reason the tube was later stiffened by the OEM (stiffening rings, frame below). For this ringshaped sleeves slid on the tube. This change was carried out at all aeroengines. However the stiffening rings loosened during operation. Therefore they have been secured at the pressure tube („add on“ fix) with a circumferential weld. In several cases it came at this fixing welds to crack formation, but never to the fracture of the ring. The cracks could be identified in time before a catastrphic failure with a periodic borescope inspection.
However in the shown case a fatigue fracture of the front stiffening ring occurred. The ripped ring was centrifuged. Also the fixing welds of the other stiffening rings showed cracks but did not break.
As result of the ring fracture the centrifuged part laid at the hollow high pressure shaft. This produced because of the large diameter an especially high unbalance. This excited the shaft system into heavy vibrations. The consequence was the fracture of the main bearing and the bearing chamber. This started to twist and the oillines ruptured. The oil lekage triggered an intensive oil fire in the low pressure (LP) turbine rotor (volume 2, Ill. 9.2-12). After that the overheated LP turbine shaft broke. When the 1st rotorstage had been separated from the 2nd LPT rotor stage, obviously it came to an overspeed. Numerous rotorblades have been centrifuged (volume 1, Ill. 4.5-4) and the the low pressure turbine was cut off.
This incident demonstrates the problem of tack welds/fixing welds/securing welds. Those have the potential of a „disimprovement” (chapter 21.3.3) because it must be expected, that those weak points have a lower fatigue strength. To this belong effects like notches by the material structure and geometry as well as internal tensile stresses (volume 4, Ill. 188.8.131.52-1). Above that fixing welds can hinder elastic and plastic deformations. This leads to stresses by thermal expansions and elongation differences because of different elasticity of the joint cross sections.
Note: Later (fixing) welds, which have not been in the certification tests of the aeroengines must be watched especially critical. They itself can represent dangerous weak points and so have a high potential of a „disimprovement“.
Figure "Risk of combustion chamber repair" (Fig. "Errors due to not understood failure mechanisms", Lit. 21.2.1-12): Already in former cases this aeroengine type suffered the fracture of the combustor casing. Cause was a failure of the combustion chamber. At numerous tubular combustion chambers some centimeter long cracks had been observed. Some propagated around the full circumference. The cracks preferred the circumferential welds of certain gas duct rings (liners, sketch below). Usually these cracks are found during the hot parts inspection. A repair is carried out by welding a new liner. Leads crack formation to a gape of the combustor tube or a deflection of the fuel injection spray, this can trigger a local overheating of the combustor casing. The result is the ripping of the high loaded wall of this „pressure vessel”. Through this opening hot gases exits, and combutor tubes are ejected. With this the immediate danger of a damage of the fuselage and the wing tank exists (volume 3, Ill. 184.108.40.206-9).
At the OEM a program to avoid such failures was initiated.
The first step concerned the understanding of the failure mode/mechanism and the causative influences. Parallel the repair versions have been investigated. For this it was proceeded as follows:
- Evaluation of the (thermal fatigue) cycles up to the full circumferential crack.At a former point of time in some cases an inner ceramic coating (thermal barrier coating) was introduced, which lowered the wall temperature 10-40°C. However it showed, that even after 3000-5000 operation hours heavy oxidation (burning) and crack formation at the liners occurred. The OEM did not use this modification for new combustors.
- Investigation of the deteriorations by operation, especially of the remaining material strength (hardness, fatigue strength) of combustors with long operation.
- Investigation of improved technics to identify cracks during overhaul (in the shop).
- Testing of a reinforcing welding on the endangered circumferential weld.Already at an early point of time a repair called “as braze reinforcement” was introduced, which promised the double lifetime. However this version was canceled after a further failure.
- Investigation of a heat treatment (solution annealing) for the rgeneration of the combustion chamber wall material.
- Investigation of alternative welding processes, which let expect longer lifetime of the cumbustion chambers.
- Examination of the repair limits in the overhaul manual for combustion chambers.Based on the program results the following recommendations are given:
- Solution heat annealing of the combustor rings (liners) in an inert atmosphere before the repair welding, correspondent with the instructions in the overhaul manual. With this it is possible to „rejuvenate“ the wall material Hastelloy X for a longer lifetime and to clean the surfaces before welding.
- Especial attention is needed before the repair at cracks in the weld seams of the liners. Cracks must be totally removed before the welding repair by grinding.
- Replacement of buckled/bulged and heavy oxidised liners, which have been frequently repair welded.
- The investigations show,that the fully circumferential cracks developed by thermal fatigue. Long cracks only occurred in repair welds. Therefore a strengthening by a second weld on the circumferential seam between liner A and B was demanded. Corresponding instructions had been assumed into the overhaul manual. Tests of such weldings showed the positive effect of this measure.
Thereupon the following additional recommendation was given:
- Operator specific periodic inspection of the combusion chamber tubes according to the „on condition”
- principle. As example intervals of 6000 operation hours are specified.In an airworthiness directive (AD), which was edited after the here described incident is claimed:
X-rax inspektion inside the aeroengine (Ill. 220.127.116.11-4 and Ill. 18.104.22.168-5) or disassembly of the combution chamber region for a visual inspection for cracks in combustor tubes.
Further the pilot was briefed about signs in the behaviour of the aeroengine during the start phase which point at a failure. For the failure case measures are recommended.
Comment: The problem occurred since years. Obviously it was tried to solve it without a redesign, but with repair procedures like a new weld above the original and a thermal barrier coating. However it seems that this could not improve the long time behaviour sufficiently. Therefore considreably short term and periodic inspections have been established.
This example shows the complexity of such a failure. It gives an impression of the effort of development and proof for a successful repair method (chapter 21.3.2).
21.2.1-1 J.L.Kolstad, NTSB „Safety Recommendation“, January 26, 1990, page 1-3.
21.2.1-2 „Safety Board Links Failure of JT9D Blade to Faulty Repair”, Zeitschrift „Aviation Week & Space Technology“, February 5, 1990, page 40.
21.2.1-3 J.S.Dodgson, „Repair of Gas Turbine Hot Gas Path Components”, Zeitschrift „Diesel & Gas Turbine Worldwide“, December 1982, page 22-24.
21.2.1-4 J.Liburdi, „Enabling Technologies for Turbine Component Life Extension”, Paper des RTO AVT Workshops, Corfu, Greece, 5-6 October 1998,December 1982, page 11.1-11.7.
21.2.1-5 J.Liburdi, P.Lowden, C.Pilcher, „Automated Welding of Turbine Blades“, Zeitschrift: „Transactions of the ASME”, Vol 112, October 1990, page 550-554.
21.2.1-6 „Single crystal and DS blade repair“, Zeitschrift: „Gas Turbine World”, July/August 1999, page 26-27.
21.2.1-7 C.Stoll, „Thermal joining of high temperature resistant coating materials“, page 80-85.
21.2.1-8 Australian Transport Safety Bureau, Technical Analysis Report No. 3/01, „Examination of a Failed CFM-3C-1 Turbo Fan Engine - Boeing 737-476, VH-TJN”, Occurrence July 2000 , page 1-18.
21.2.1-9 P.Adam, „Fertigungsverfahren von Turboflugtriebwerken“, Birkhäuser Verlag, 1998, ISBN 3-7643-5971-4, page 161-191.
21.2.1-10 J.Liburdi, P.Lowden, „Repair Technique for Gas Turbine Components”, Proceeding der AGARD-CP-398 AGARD Conference „Advanced Joining of Aerospace Metallic Materials“, Oberammergau, Germany, 11-13 September 1985, page 22.1-22-12
21.2.1-11 NTSB Aircraft Accident Report NTSB-AAR-72-29, „Continental Airlines, INC, Mc.Donnell Douglas DC-10, N68041, Tucson, Arizona, May 2. 1972”, page 1-10.
21.2.1-12 AAIB (UK), Aircraft Incident Report No. 8/88, „Report on the accident to Boeing 737-236, G-BGJL at Manchester International Airport on 22 August 1985“, page 1-129.