Aeroengine Safety

Illustration 16.2.1.1-1

Illustration 16.2.1.1-2: This is a composition of typical effects in a machined surface (Ref. 16.2.1.1-14).

“A” Residual stresses: Every machining process will cause plastic deformations, at least on a micro-scale, and therefore create residual stresses (Chapter 16.2.2.4). Their size and type (tension or compression), and therefore their influence on the characteristic of the part, are very sensitive to the machining parameters, the shape of the part, and the condition of the cutting edges of the tools. Fundamentally, tension residual stress reduces dynamic fatigue strength and increases the risk of corrosion cracking (Fig. "Finishing related corrosion fracture of pipe connections") and liquid metal embrittlement (Fig. "Surface fouling affecting finishing and assurance").

“B” Changes in strength and hardness due to structural changes (Ill. 16.2.1.1-3.1): The warmth from the machining process can have a damaging heating effect on the surface. If, for example, a boring machine becomes blue (Ills. 16.2.1.1-7.2 and 16.2.2.5-3), it is a clear indication that the work surface, i.e. the bore wall, has also been dangerously overheated. The thermal development is influenced by many process parameters, as well as the physical properties of the part. In addition, the thermal conditions of the cutting process are decisively dependent upon the inflow and properties of the cutting and cooling media. For example, hardened or heat-treated steels lose some of their strength if their temperature exceeds the tempering temperature. If the solution annealing temperature is exceeded in hardenable alloys, such as nickel alloys, the hardening phase (g'-phase), which determines the thermal strength, will solubilize as temperature and time increase.

“C” Crack initiation: This occurs when the cutting forces mechanically overload the material. Ni-based forged alloys such as Waspaloy experience so-called “comma cracks”. These are transcrystalline forced cracks that are limited to one grain. Similar cracks can occur in connection with chatter marks (Fig. "‘Plucking’ damages by a machining process"). Especially dangerous forced cracks can occur when tool holders scrape against the part (Ill. 16.2.1.1-3.1)
Another type of cracking is hot or thermal cracking (Ills. 15.1-8 and 16.2.2.7-6). This occurs at high temperatures that have already caused sufficient softening of the grain boundaries, so that thermal stress and/or the cutting forces lead to intercrystalline cracks. Typical examples are grinding cracks in super alloys.

“D” Worsened verifiability following mechanical closure of flaws and cracks: Flaws such as cracks and pores can be pressed shut through the machining process at the surface area (Fig. "Limiting influences non-destructive testing"). This effect can be especially pronounced during grinding. In this case, previously created grinding cracks are closed, preventing penetrant testing from yielding satisfactory results. As a result, a heat-treatment cycle is required to open the cracks (Fig. "Opening of cracks before penetrant testing").

“E” Structural changes, caused by plastic deformation and thermal increases (Ref. 16.2.3-5, Ill. 16.2.1.1-3.1): Extreme plastic deformation leads to poorly etchable structures, which are usually brittle in martensite steels (white etching layer, Ref. 16.2.1.1-17). By comparison, very thin (mm range), poorly etchable soft coatings on machining surfaces are usually unavoidable and are not necessarily damaging. The cutting direction of cutting tools with powerful pressing forces results in pronounced bending of the grain near the surface. Whether or not this phenomenon is negative depends on the residual stresses induced by the deformation process. Tension stresses reduce the dynamic strength. High temperatures during machining can change the structure in an undesirable manner, as is also described in “H”.

“F” Notches: Machining can create notches in the form of grooves (Ills. 16.2.2.1-5 and 16.2.2.1-6, Refs. 16.2.3-15 and 16.2.3-16) and chatter marks (Fig. "‘Plucking’ damages by a machining process") that act as origin points for dynamic fatigue fractures. An additional type of damage that can occur in Ni-based materials is the breaking-out of pronounced carbides during the machining process. This results in noticeable notches.

“G” Tool fractures can represent dangerous damage (Fig. "Dangers of tool-fractures"). This is the case if fragments of the broken cutting edge remain in the surface.

“H” Impressed chips can act as notches. Related phenomena are smeared foreign materials that cause extreme surface hardenings (“E”) and overheating with strength losses (“B”; Fig. "Dynamic fatigue strength influenced by fused chips").

“I” Smearing of foreign metal can occur in several different ways during the machining process. Possibilities include wear products from equipment (Ills. 16.2.1.1-3.1 and 16.2.2.3-1) or low-melting metallic sprue materials (Fig. "Fastening of parts during finishing"). These types of fouling can have various damaging effects, such as through element formation or liquid metal embrittlement (Chapter 16.2.2.3 and Ill. 16.2.2.5-10.1).

“K” Contaminants that will necessarily come into contact with the machined surface are cooling and cutting media. Silicon oils used to suppress foaming can considerably worsen the wettability of the machined parts and, through carryover (Fig. "Difficult crack detection by fouling on baths"), affect other parts, as well. This may compromise the effectiveness of penetrant testing. Experience has shown that the resulting uncertainty will ultimately result in extensive cleaning measures and generate high costs.
Aggressive contaminants from etching and cleaning baths can result in unallowably heavy oxidation and/or selective corrosion (e.g. intercrystalline attack) during subsequent heat treatments (Fig. "Problems by contaminated cleaning baths").

Illustration 16.2.1.1-3.1 (Ref. 16.2.1.1-29): During machining, the surface can be damaged in various ways. The effects of the most important damage types (Ill. 16.2.1.1-2) on the LCF strength and, therefore, cyclical life of rotor components (disks and rings) are shown in the diagram. The most serious influences are evidently from smeared material and crack-like damages caused by local mechanical overloads (middle diagram) resulting from contact with the tool holder (Fig. "Friction contact during clamping"). This also includes heat development with local melting (recast layers, bottom diagram) that occurs during the rubbing process. This is followed by impressed chips (top diagram, Fig. "Dynamic fatigue strength influenced by fused chips") and the influence of serious structural changes.

Illustration 16.2.1.1-3.2 (Ref. 16.2.1.1-1): This example of bores in a disk material made from a forged Ni alloy provides a sense of the influences on a machined surface. Naturally, these changes depend not only on the machining parameters, but are also decisively dependent on the material properties. The tendency to harden, the structure, and the crystal structure all play important roles.
There are four primary effects at work:
Grain boundary deformation can be seen in a bending of the grain in the direction of the machining due to considerable plastic deformation (right detail).
Slip bands are marked by closely neighboring parallel line fields. They are created through slipping of the material on the slip planes of the crystals during plastic deformation.
Compressive residual stresses develop due to the difference in plastic deformations at different distances from the surface. Understandably, they drop off quickly towards the inside of the part. One can see that, using the applied machining parameters, compressive residual stresses from (circular) grinding acted especially deeply. They clearly extend beyond the structural characteristics.
Changes in microstrength can be traced back to strain hardening due to the plastic deformation. An increase in microstrength towards the surface can be expected, corresponding to the deformation pattern.
Therefore, the recognizable structural changes in a metallographic cut do not necessarily reveal information regarding changes in hardness or the depth of compressive stresses, or vice versa.

Figure "Chipping surface metallographic section": Grinding is a necessary process due to the difficulty of machining the high-strength materials used in turbine construction and the special demands on dimensional exactness and surface quality. Grinding can cause various types and degrees of damages in the surface area. In Ni and Ti alloys (Refs. 16.2.1.1-2 and 16.2.1.1-14), the damages are primarily related to

• residual stresses
• losses in strength/hardness
• cracking

Martensite steels can show embrittlement and increases in hardness, depending on the temperature and cooling rate (Refs. 16.2.1.1-4,-6,-7,-8).
During grinding, various influences act in combination (top diagram). They depend especially on grinding parameters such as cutting rate, thrust, and infeed (Ill. 16.2.1.1-3). In addition, characteristics such as the type of grinding disk and the stiffness of the grinder are important.

Thermal stresses are a result of temperature gradients during grinding. If they exceed the flow limit, the result is plastic deformations. During the heating process, compressions are created that cause tensile stresses during cooling. This process can tear open the grain boundaries that have been especially weakened by the high temperature (hot cracks, thermal cracks, Fig. "Mechanisms of hot cracking").

Strength losses, especially in the grain boundary zone, due to high temperatures. Softening may occur, with the result being that minimal tensile stress is sufficient for crack initiation. Even if no cracking occurs, hardenable alloys may lose strength if the solution annealing temperature is exceeded.

Plastic deformation due to the cutting forces: This usually involves a compression process. Compressive stresses overlay with the thermal stresses.

Groove formation is of considerable importance for the attainable dynamic fatigue strength (Fig. "Roughness influencing dynamic fatigue"). The notch effect of grinding grooves is more pronounced in the HCF range than in the LCF range.

“Peppering”: this term refers to a type of surface damage in the form of many small impact craters, similar to the appearance of abrasive shot processes. This effect is caused by ricocheting particles or particles from the disk and chips that are carried by the coolant flow.

Cracks being smeared shut: Experience has shown that grinding cracks are likely to be smeared shut by the machining process, especially in nickel alloys. For this reason, sufficiently safe detectability can only be expected after a suitable heat treatment cycle (Fig. "Testing of high strength material").
Naturally, the effect of the process influences also depends on the properties of the specific materials. In addition to physical data such as thermal conductivity, thermal strain, and the modulus of elasticity, the structure plays an important role. This means that earlier production stages such as raw part production and heat treatment become important.

Depending on the grinding conditions, these material properties are of varying importance and can combine with one another. Depending on the material and grinding conditions, (grinding) cracks can occur during the machining process and/or afterward (delayed; bottom diagram). Typically, several grinding cracks will occur, oriented perpendicular to the machining direction.
This crack alignment can be explained by the temperature and deformation distribution (Ref.16.2.1.1-4) in the part surface. It is influenced by enfolded grinding contours such as edges or humps in shaped disks (middle diagram). In these cases, overheating and cracking are supported by the restricted coolant flow as well as the large machining surface relative to the heat-absorbing volume.

Figure "Fjnishing caused grinding cracks" : Grinding cracks are a typical problem (Refs. 16.2.1.1-1 and 16.2.1.1-26) of working on highly precise contact surfaces in the roots and shrouds of turbine blades (top left diagram). The typical damage symptoms are crack fields located primarily on convex contours and perpendicular to the cutting direction (bottom left diagram). The cracks tend to run along brittle phases such as carbides. The more material-specifically pronounced these are (detail), the greater the tendency to grinding crack initiation. Therefore, the structure is important when determining the grinding potential of a material/structure. Anomalies from the casting process may become apparent in certain charges. Grinding cracks also orient themselves near micro-cavities. These can be torn open and influence the penetrant testing results in such a way that cracking and porosity are confused. There is also a danger of the grinding cracks being smeared shut. For this reason, a heat-treatment cycle is usually used after the grinding process to open the cracks before penetrant testing (Fig. "Opening of cracks before penetrant testing").
The top right diagram shows an example of the limiting of the grinding parameters in order to avoid damages under consideration of economically realistic conditions. On the one hand, effort is made to work in an economic range above a minimum level of machining efficiency. On the other hand, thermal damage (Ill. 16.2.1.1-2) limits machining efficiency.

Figure "Problems of hand guided machining": Hand-guided deburring, smoothing, and rounding using hard-particle coated, flexible, flat carriers and brushes with bristles made from materials such as steel or brass wire can be problematic (Ills. 16.2.2.3-11 and 16.2.2.3-13). There are more possibilities for these processes to damage the parts than are apparent at first glance:

Smeared bristle material (“A”): there are many possible damage mechanisms, depending on the material of the part (Ills.16.2.2.3-1, 16.2.2.3-2, 16.2.2.3-11, and Ref. 16.2.1.1-18). Steel brushes made from carbon steel leave residue on titanium disks, for example, which can cause corrosion during later operation (Ref. 16.2.1.1-20). There is a special danger for surfaces with metallic smeared material on them that are subsequently heat-treated. In Ni and Ti alloys, as well as steels, liquid metal embrittlement cracking may occur, depending on the material composition of the residue (Fig. "Damages by metallic surface fouling from finishingfinishing").

In titanium alloys, there is an increased danger of unallowable changes (strength, toughness) through hydrogen absorption in etching and cleaning baths (Fig. "Operation behaviour of parts by surface fouling").
It has been observed that Fe fouling on the surface can promote cracking during the welding of Ti alloys through the formation of brittle phases (Fig. "Fouling of Titanium welds"). In addition, there is a potential danger that coatings may be influenced (e.g. diffusion coatings, lacquers, and galvanic coatings).

Reactions with abrasive particles: At high temperatures, such as those expected during heat treatment, welding, soldering, and diffusion coating, SiC particles may react with the base material, especially with Ni and Ti alloys (Volume 2, Ill. 7.1.4-14). Brittle phases form and melting occurs, causing dangerous strength losses. Therefore, it must be ensured that only abrasive material with safe hard particles such as Al₂O₃ (corundum) is used. This also applies to polishing felt (“F”).

Transferral of wear products from other parts: This risk occurs when the same smoothing tool (for example, “A”, “B”) is used on parts made from different materials. This can result in, for example, wear products from titanium alloys being transferred onto Ni alloys and vice versa. The dangers inherent in this are comparable with those of smeared materials discussed above (Fig. "Damaging by foreign materials on parts").

Burrs: burrs can considerably reduce the dynamic fatigue resistance (Chapter 16.2.2.2). Because abrasive processes are used especially to remove burrs, it must be ensured that they do not result in new burrs forming.

Grooves: these can promote dynamic fatigue cracking in highly-stressed part zones (Ills. 16.2.2.1-7 and 16.2.2.1-8). Unfortunately, the accessibility determined by the shape of a part often provokes machining in a direction perpendicular to the main operating loads. Typical examples include the rounding of edges on blades (“D”) and the smoothing of transitional radii in disks (“E”, Fig. "Safety of parts by deburring").

Overheating: structural changes up to crack initiation can unallowably reduce part strength. The danger of unnoticed overheating through the machining process cannot be underestimated (Fig. "Demands of manual reworking"). Warning discolorations (tarnishing) are the only external indicator of possible damage (Fig. "Causes of tarnishing and discoloration on parts") and can also be erased in the same machining process. This means that specialized observation of the process by experienced experts is especially important (Ill. 17-5).
Another possibility for dangerous overheating is through the impact of sparks during an intensive smoothing process, for example, using a belt sander (“C”; Fig. "Damage potential of hot metal particles").

Illustration 16.2.1.1-7: Although the material removal process during vibration polishing (Refs. 16.2.1.1-1 and 16.2.1.1-3, top diagram), is quite gentle, it can still damage parts. Possible causes for damage are increased local material removal from exposed part zones (bottom right diagram) and longer local action of single machining chips. More intensive damage may occur if there is unintended contact between the parts during the machining process (middle diagram). Problems can also arise due to insufficient material removal. The local part shape can prevent contact with the chip (bottom left diagram), making it possible that undesirable burrs and grooves are not removed.

Figure "Deviation from proved finishing processes" (Refs.16.2.1.1-9, -10, -11): This spectacular damage incident was promoted by a finishing-specific weak point in a very highly cyclically stressed (LCF) disk zone that was not understood. During a transitional period, segmented finishing processes had to be used for ECM finishing of the complex contour in the annulus of the turbine disk. This resulted in several thin, radially oriented ligaments forming along the circumference (right diagram). These ligaments had to be subsequently removed, which was done in a hand-controlled machining process during finishing (Fig. "Demands of manual reworking"). During this process, the highly-stressed radial contour was overheated in single cases, and this was not noticed. A slight decrease in LCF strength was sufficient to cause serious damages (Fig. "Demands of manual reworking").

Illustration 16.2.1,1-8.2 (Refs.16.2.1.1-9, -10, -11): During the starting of a passenger aircraft, uncontained disk damage occurred. The fragment bounced off the runway, destroying tires on the landing gear, and then damaged the second engine (top diagram). Despite a fire, it was possible to evacuate the aircraft without injuries.
The damage investigation revealed that the disk failed due to an LCF crack in a machined area. During new part finishing, reworking was done at the transition radius to a boss near the annulus contour (Fig. "Deviation from proved finishing processes"). The disk was evidently damaged in this zone (bottom detail). Damage mechanisms are described in Ills. 16.2.1.1-2 and 16.2.1.1-4. A large number of engines were potentially affected, and several disks that had already been run showed crack initiation during inspections. Extensive reworking and corrective measures were required.

Figure "Flnishing problems of bores":

This damage is very instructive, and is therefore shown in considerable detail in several pictures corresponding to information from the “Aircraft Accident Report” NTSB/AAR-98/01 (Ref. 16.2.1.1-13).
Chronology of the flight accident (Fig. "Flnishing problems of bores"): The engine damage occurred in a smaller passenger aircraft with two engines mounted at the rear (top diagram). The incident occurred at the beginning of the takeoff roll. Fragments of the fan disk (bottom left diagram) and blades escaped from the left engine and penetrated the neighboring side of the fuselage. Takeoff was aborted and the aircraft came to a stop on the runway. Two passengers were killed by the fragments, and two others were seriously injured.

Figure "Damage-relevant characteristics of a bore"

The damage investigation revealed that the cause of damage was an LCF dynamic fracture in the fan disk, which was made from a high-strength titanium alloy. The fracture originated in one of the bolt bores for the flange of the low-pressure shaft. It was found that the crack initiation zone was in the area of a bore wall that had evidently been damaged in the finishing process (top left diagram, Fig. "Damage-relevant characteristics of a bore").

Figure "Crack size and life span of a disk":

Damage investigation and findings (Fig. "Crack size and life span of a disk"): First, the type of damage at the crack initiation had to be determined (Ill. 17-11). The next question was why the damage had not been recognized during new part finishing.
The disk was made from the high-strength titanium alloy Ti-6Al-4V.
Around the flaw, the bore wall had machining grooves and discolorations about 23 mm in length (top right detail).
A metallographic cut in the crack initiation zone revealed structural changes with micro-cracking about 0.25 mm deep (ladder cracking). The hardness in the damaged area was 52 HRC, i.e. considerably above the maximum allowable 39 HRC. There were no macro-cracks. There were three distinct structural zones (middle right detail).

Zone 1: Recrystallized grains of a phase are created at temperatures above 650°C. Because this type of structural change is related to time-dependent diffusion processes, the following is true: the shorter the duration of the temperature, the higher it must have been. Therefore, for the relatively short heating periods of a boring process, it must be concluded that the temperatures were considerably greater than the threshold value for the observed structural change. In addition, an iron accumulation was found in the bore surface. On the basis of these findings, it can be speculated that the borer showed signs of tarnishing, and perhaps even glowed (see reproduction experiments, page 16.2.1.1-23).
At these temperatures, titanium alloys absorb oxygen from the air. This stabilizes the a phase. These temperature levels are not expected from a cutting borer with sufficient cooling, which the specified boring process should guarantee.

Zone 2: This zone, which is located below Zone 1, has grains that are highly deformed parallel to the surface. Below this is Zone 3, which has a micro-structure that is bent along the circumference.
Electron-microscopic analysis of the fracture surface (Fig. "Scanning electron microscopy (SEM)") revealed (Fig. "Crack size and life span of a disk") that, at the point of damage after 16,545 hours of operation with 13,835 startup/shutdown cycles, the LCF crack must have reached a critical length of 37.5 mm. It originated in two flaws that were attributed to the boring process and located about 7.5 and 12.5 mm from the disk surface.
The investigation results allowed the following conclusions regarding the damage cause: The damages in the crack initiation zone can be traced back to a failure of the boring process (Ill. 16.2.1-9.2). Micro-cracks acted as the initiator of the cracking. The depth of the damage was considerably greater than that expected from an undetected failure of the boring process.
It was also necessary to determine how such dangerous damage could avoid detection during or after the finishing process.

The new part was measured by the manufacturer and underwent a visual inspection, penetrant testing, as well as macro-etching (blue etch anodizing = BEA). In this process, the bores are inspected with regard to their path, centering, and diameter. BEA reacts to structural changes (e.g. a and b accumulations, unusual grain growth, forging laps) and shows these as discolorations (similar to tarnishing). The procedure can be used on all titanium alloys. During pre-boring, chatter marks were noticed on two bores by the worker. After subsequent boring and honing to measure, these machining marks were evidently no longer noticeable, and the part was released for use. During the following BEA test, finishing marks were found in one of the bores. The findings did not correspond to any of the reference diagrams for finishing damages available at the time, and was not categorized as a BEA finding. For this reason, the production records for the damaged part did not contain any indication of damage.
Attempts to reproduce the damage to the bore were only successful when the cooling fluid was completely absent, with excessively high RPM, and powerful infeed thrust. The test led to borer fractures or borer failure with chip jams (Fig. "Dynamic fatigue strength influenced by fused chips"). In addition to typical structural changes, a high iron concentration (borer material) corresponding to that in the damage incident was found on the bore wall. It was not possible to recreate comparable damages with the subsequent honing process that gave the bore its final shape.

Figure "Inspection of bores in disks during overhaul":

Controlling bores with penetrant testing and BEA occurs both on new parts and during overhauls (Fig. "Inspection of bores in disks during overhaul"). Both test findings were visually inspected using a magnifier. The problem here was that the magnifier that was used had the wrong depth of focus. This prevented a sufficient overview. The evaluation of the bore surface was further complicated by the fact that it was mirrored, making it difficult to recognize details.

Figure "Maintenance penetrant inspection":

Based on the investigation results, the damage process occurred as follows:
During the first overhauls, it can be assumed that there were 0.5 x 0.25 mm LCF cracks. The last overhaul (heavy maintenance) before the damage occurred took place at 12,693 cycles (Fig. "Maintenance penetrant inspection"). Based on the fracture surface analysis, there was probably already a crack with an axial length of 20 mm and a radial length of 10 mm present during the last overhaul.
How did a crack of this size escape detection? This is most likely due to the human factor (Fig. "Failed detection of large cracks"). Experience has shown that this usually involves going beyond the “experience horizon”, as well as a lack of successful experiences (i.e. frustration) because these flaws are very rare. Additional problems were less than optimal preparation work (fouling) and testing processes, such as time durations, distribution of the developer, and drying.

Figure "Maintenance penetrant inspection part two":

An audit of the penetrant testing (FPI) at the manufacturer and in the overhaul facilities revealed potential inadequacies (Fig. "Maintenance penetrant inspection"). There is a surprising number of possibilities to compromise the effectiveness of an apparently simple non-destructive testing process such as penetrant testing (Fig. "Maintenance penetrant inspection part two"). This is an exemplary demonstration of the necessity of experience-based expertise, adherence to specifications, and the critical observation of testing processes.

Figure "Boring tool fracture identified by eddy current" (Ref. 16.2.1.1-19): This example shows that the danger of damage occurring during boring is not limited to titanium alloys (Ill. 16.2.1.1-9). As far as one can tell from the literature, the affected area is the rim bolt holes of the HPTR fan disk. This is evidently an inducer or cover plate (right diagram) that is fastened to the front of the high pressure turbine disk to provide more intensive cooling for the rotor blades. Judging by the expected operating temperatures, this part should be a nickel alloy. A crack was discovered during a routine overhaul with eddy current testing. The inspection revealed that the crack originated in a pressure point in the bore wall. This was caused by the boring tool breaking during the finishing of the new part.

Figure "Disk fracture caused by smoothing procedure" (Ref. 16.2.1.1-13): The depicted case concerns the disk of the first fan stage made from the high-strength titanium alloy Ti-6Al-4V (top diagram, corresponding to Fig. "Flnishing problems of bores"). A disk lobe had broken and released three fan blades. Three more lobes broke as consequential damages, releasing four more blades. Evidently, the resulting major imbalance caused the following fan stage to be thrown off, resulting in serious damage to the aircraft. The fracture of the first lobe showed an LCF crack on the front side of the disk at the transition into the annulus (bottom right detail). The crack required about 800 load cycles to grow to critical length. The lobes that broke as consequential damage also already had several small cracks in the corresponding area. Three more cracks were found through non-destructive testing (disk diagram). Closer inspection of the crack initiation zones revealed that the LCF cracking was related to surface damage. This damage occurred during an attempt to polish the radius by hand. Sharp scratches traversed the surface and broke open. There was an SiC particle embedded at the end of each scratch, and the dynamic fatigue cracks originated at these particles. The suspect procedure occurred during new part finishing. Astonishingly, the prescribed arithmetic (average) roughness of 32 microinches (0.0008 mm) was only slightly exceeded, to 40 microinches (0.001 mm). This demonstrates the limited informational value of a roughness measurement when the issue is the effects on the dynamic fatigue strength (Fig. "Roughness data tell not enough"). It is notable that two other damages that were at least externally similar had already been reported. This indicates an underestimated, very powerful stress on the part zone in question.