Table of Contents
8.1 Origins of and Damages Caused by Rotor Fragments
There are many causes of rotor part failure. Selected examples with varying causes and consenquences are given at the end of this chapter. Some primary damages are caused by a lack of strength in the cracked rotor part itself. Other common primary damages that result in rotor failure are shaft breakages (overspeed) and seal failure (overheating).
Causes of rotor disk and blade failures(the order is not related to frequency, nor does this list claim to be complete):
- Overload: For extreme overspeed to cause the spontaneous failure of a rotor disk (danger of runaway turbine discs), it must be greater than the overspeed levels determined to be safe (roughly 30% over the 100% rpm rating). This will more commonly result in a drop in rpm (regulator, blade damage) rather than a blade fracture or a release of blades from the disk. The danger of a runaway disk is not present in compression rotors. An extensive loss of fan blades could, however, cause the fan disk and possibly the booster to reach high overspeed. In most cases of fan damage, consequential damage to the compressor will immediately drop the engine`s performance levels.
Typical causes of rotor overspeed:
- Regulator failure results in dangerously high levels of fuel intake
- Fuel being sucked in by the compressor (e.g. from open tank nipples, left over fuel traces from refueling)
- Insufficient drainage of the combustion chamber after an aborted start
- Separation of the rotor shaft due to fracturing or failure of a connection (flange, clamp bolt)
- Fatigue fractures due to cyclical stress in the LCF (low cycle fatigue) and/or HCF regions (e.g. through high frequency oscillations). The causes could be unusually high performance demands and/or defective parts. LCF damages also include fractures due to TMF (thermo mechanical fatigue), typically annulus cracks in integral turbine disks (Fig. "Typical rotor fragments"). The burst-resistance of rotor components is dependent upon sufficient verification of their operational life span in real conditions and reliable technical and logistical means of monitoring this.
- Overheating: there are several ways by which rotor disks can reach dangerously high operating temperatures:
- Lack of cooling air
- The incursion of hot gases due to failing seals in the annulus area of turbine discs
- Oil fires due to bearing
- or seal damage or the failure of an oil line
- A stalled compressor (lack of cooling air and/or excessive fuel flow)
- Blade rubbing (e.g. near seals)
- Titanium fires
- Malfunctions in the combustion chamber and/or the fuel injection system
- Regulator failure
- Sectional weakening due to blade rubbing (e.g. near seals)
- Insufficient strength: common causes of damage are:
- Flaws in the early production process: e.g. mistakes during forging, scabs
- Flaws in workmanship: e.g. grinding tears, comma tears, dents and notches
- Flaws from the splicing process: e.g. welding cracks, bonding flaws
- Coating flaws: e.g. scribed coatings, embrittlement
- Damages from handling: e.g. scratches, dents
- Extraneous forces:
- Disk damage caused by broken screws from flange connections
- Shaft fractures due to powerful momentum changes after a bird impact
Types of Fragments:
The rotor`s operational demands put specific types of stress on typical structural parts. In the case of cracks, primarily this means dynamic stress (LCF and HCF) caused by changes in centrifugal forces, high frequency oscillations, and temperature gradients. Rotor component fragments can be grouped into the following categories (Fig. "Typical rotor fragments"):
- Individual blade fragments: these fragments are created through fatigue fractures in the blade or root areas (e.g. in the bearing surface), through overspeed and/or failure of the root (e.g. due to overheating in the disc`s annulus zone) resulting in the blade being flung away, and also through bird or ice impacts.
- Multiple blades: turbine blades after an overheating of the turbine. Consequential damage after a foreign body impact or a blade failure. Release of a blade pair, where the configuration is made up of two blades in a fir tree root.
- Failure of the entire blading (Haircut) after extreme overheating, high overspeed, or a haircut as consequential damage (e.g. after foreign object damage or failure of a single blade).
- Multiple blades with the attached part of the disk annulus: Fractures in integral turbine discs originating in annulus cracks. Disk fractures due to oscillations, flaws in the material, or overspeed.
- Disk sections: following an LCF- or HCF-tear, overspeed, material flaws, blade rubbing , overheating. In extreme cases the disk breaks in half. The most powerful penetrative force can be expected from 1/3 disc sections (Fig. "Hazards of disk fragments").
- Entire rotor disc: e.g. due to the failure of the shaft, a flange connection, or a central clamp bolt (in small engines). However, dangerously high stress levels impact the housing only when the disk is already out of balance or, due to uneven deceleration, has a great deal of translatory inertia.
Danger of penetration by rotor fragments:
Engines should be constructed, in accordance with regulations, to ensure that rotor blade fragments cannot escape the housing and that no uncontrollable damage results from a blade failure (e.g. imbalance-induced overstress or failure of the mounting due to the deceleration of the rotor). This is verifiable by means of a containment trial within the rating of the engine type (see chapter 8.2).
The penetrative power of a rotor fragment increases linearly with its mass and corresponding kinetic energy. Verification of containment capabilities in turbofan engines is concentrated on the event of a fan blade failure, because this results in the greatest strain on the housing (thin-walled housings, heavy blade fragments).
Fractured rotor disks can usually not be contained by the housing. Adequately massive housings would lower the power/weight ratio too far. Thus there are no containment demands made for rotor disk fragments. For this reason, the frequency of disk failures must be kept considerably lower than that of blade failure.
Blade fragments can lead to the failure of an entire stage of blading (Haircut) and a blade jam or a release of blades at high overspeed (e.g. in the turbine after a shaft failure), which can result in dangerous stress levels on the housing and an escape of fragments.
Shape of fragments:
The shape of fragments affects their penetrative force. Thin fragments made of hardened steels have a cutting effect if they impact with the narrow edge. This effect can increase the danger of lightweight rotors, even though the mass of the fragment is relatively small. Examples of this are spliced hollow centrifugal compressor disks made of steel sheet with a thin-walled, back-bearing disk.
Trials have shown, against expectations, that fragments with spherical or chamfered impact surfaces have greater penetrative force than sharp projectiles when shot against a ductile housing material. This can be understood by looking at the penetrated housing. With spherical impact surfaces and tough (ductile) housing materials, the hole is sheared without much plastic deformation around the impact area. Housings penetrated by sharp projectiles show a much greater degree of plastic deformation around the hole, meaning a relatively high loss of kinetic energy from the particle (Ref. 8.1-7). Thin, tough target plates can withstand a greater deforming energy than punching energy (Ref. 8.1-7). Thus, sharp projectiles are more likely to be slowed than rounded ones. Brittle housing materials exhibit the opposite effect, and are more likely to be penetrated by sharp projectiles than rounded ones.
Especially dangerous are fragments of high-density, impact-strength materials. However, it is relatively easy to control fragments made of materials that shatter on impact (defibrate, pulverize), because the energy transfer is across a larger area, which saps kinetic energy. These materials include fibre-based materials and ceramics. The relatively low expectation of consequential damage from fibre-based materials give weight to the argument that they should be used in compressors.
The impact speed of the fragments determines the deformation rate of the housing wall and thus the high speed embrittlement (Fig. "Brittle material behaviour at impact"). Additionally, increasing speed and kinetic energy also increase the “shear effect” (low energy transfer means a punching action occurred) at the housing breach. Thus the impact resistance statistics of a housing material at high impact speeds are the “stamp energy”, which can be determined through a punch trial.
In a punch trial, the load rating and stiffness of the housing wall is especially important. The lower the impact speed, the more extensive the plastic deformations around the impact zone become. The higher the fragment speed, the lower the amout of transferred energy at the penetration point becomes. There are a few reasons for this:
- Increase of the shear effect (punching)
- Greater mass force of the housing wall, therefore the housing wall reacts more stiffly
- Embrittlement of the housing material at high deformation rates
Trials have shown that, to a sufficient degree of accuracy, the strength of the penetrated housing wall is proportional to the kinetic energy of the fragment. Conversely, the “required” impact speed for penetration increases with the square root of the wall's thickness.
Behaviour of fragments after penetration of the housing wall:
It has been shown that rotor fragments can be diverted up to +-30° from the rotor plane upon escaping from the housing. The remaining kinetic energy drops sharply if the angle of divertion is more than 5°.
The angle of divertion after escaping from the housing is strongly related to the mass of the fragment. From a certain mass on, depending on the housing configuration, there is no noticeable divertion (Fig. "Fragments outside rotational plane").
The influence housing material has on the manner of penetration (Ref. 8.1-19):
Hardness alone is an insufficient determinant of the housing material's behaviour when penetrated. This is much more dependent upon an optimal relationship between the solidity and plastic deformation capability when stressed. Materials that show a pronounced high-speed embrittlement (e.g. low alloy steel) are not suitable housing materials due to their relative inability to accept transferred energy (Fig. "Brittle material behaviour at impact" ). It is possible to greatly improve penetration resistance; e.g. through heat treatment of quenched and tempered steel.
Both the Co-alloy “HS 25” (Fig. "Brittle material behaviour at impact" and 8.1-14) and the Ni-alloy Waspalloy are highly resistant to penetration, although Waspalloy is liable to intercrystalline crack initiation. If, however, one rates the penetration resistance of housing walls of the same mass, there are clear advantages to Al, Mg, and Ti-alloys (Fig. "Containment clamping influence"). These materials behave very well during penetration, taking a relatively high amount of kinetic energy from the fragment. It is known that fracture strength and yield strength in metallic materials normally increase along with rising stress speed (Fig. "Brittle material behaviour at impact"). At the same time, the capability for plastic deformation decreases, embrittling the material (Fig. "Brittle material behaviour at impact"). In low alloy steels, such as structural steel, this effect is more pronounced than in austenitic steels, nickel alloys, Co-alloys, and light metals. The style of the break can change along with embrittlement. For example, at a certain deformation speed, a typically tough surface will become embrittled. Unfortunately, in practice, this behaviour is not always consistent with foreign body impact observations. This indicates that the shape of the structural part and the stiffness at the impact point must be taken into consideration.
Surface curing (nitriding) has been shown to have a negative effect (embrittling) at steep angles of impact. At flatter angles, however, it may offer advantages. The temperature of the housing parts influences penetration resistance with regard to changes in hardness and toughness (Fig. "Containment clamping influence").
The influence of housing design:
The combined penetration energies of the individual housing walls in a fragment`s path are equal to the total penetration resistance. However, multiple housing walls are stronger than a single wall with the same total wall strength. From this, it can be concluded that an engine design with multiple concentric walls around the rotor is advantageous with regard to containment. However, one must consider that, as radial distance increases, the angle of impact also becomes steeper (Fig. "Features of disk burst-protection concepts"), thus making the penetrating power of the fragment greater than it would be at a flatter angle of impact (“grazing shot”). These conditions are also present in reverse-flow combustion chambers for the turbine rotor, for example. Even in turbofans, it has been observed that rotor blade fragments that had escaped from the engine core and thus lost a great deal of their kinetic energy were caught by the bypass housing. Special containment designs are covered in chapter 8.2.
Stiff and heavy housing walls are not optimal with regard to their penetration resistance, as they tend to be sheared upon penetration and thus drain less energy from the fragment than multiple thin housing walls that can give way to plastic deformations.
Figure "Rotor failures risk tree" (Ref. 8.1-1): This explanation concerns technical hazards and damage sequences. Naturally, human casualties can also result directly from fragments. If rotor fragments (Fig. "Typical fragments of rotor blades") are not contained in the engine, many different areas and components of the aircraft are threatened. Some cases can result in crashes.
If the Structure of the aircraft is damaged in supporting cross-sections, even normal operating stresses can cause failure of the affected parts. Beyond this, the drop in rigidity can result in natural oscillations fomented by aerodynamic forces (“fluttering”). This leads to dynamic overlstress of the weakened structure with immediate danger of fracturing. If the pressurized cabin is breached, a dangerous pressure loss can be expected (in extreme cases passengers have been torn from the cabin).
Systems essential for safe operation, such as flight regulators, hydraulics (e.g. for flap adjustment) and cables, can have their functions impaired, interrupted, or spring leaks. If the emergency power supply is cut, a crash can occur.
If the tank is punctured or affected by external fires, fuel loss and/or fires can be expected.
Damage to the fuel supply system results in a loss of thrust. This is especially hazardous if all engines are affected or the damage occurs in the crucial start-up phase.
Fragments from one engine can also directly damage other engines (Fig. "Rotation direction demands position of auxiliary components"). In twin-engined warplanes, where the engines are close together and parallel in the hull, this danger is especially high. This danger is further compounded if the engines are connected to ancilliary equipment via a common gearbox.
There have also been cases in passenger planes, where multiple engines failed due to fragment impacts. In one case, for example, the second engine on the other wing was struck by fragments that had bounced off the runway (Fig. "Production caused turbine disk failure").
If a system vital to engine operation is damaged, a loss of thrust can be expected even in the engines not directly affected by the damage.
Damage to the engine outlet can cause a loss in thrust and/or divert hot gases.
The continuation of the flight becomes endangered when the crew`s ability to control becomes impaired due to a failure of steering systems or fire, or if fuel supplies are no longer sufficient.
Figure "Typical rotor fragments" (Ref. 8.1-2 and 8.1-4): Normally, the area of the disc experiencing the most stress during LCF (Starting/shutdown cycles) is at the hub bore (top left diagram). This is caused by the superposition of gradients of centrifugal force and temperature. Other disc regions are susceptible to cracks due to geometric anomalies, such as stiffness cracks in shaft sockets or axial bores in the shaft connections (top right diagram; Example "Rejected takeoff due to uncontained failure"). In the hub area, a crack will quickly expand and, due to the high stress levels and low gradients, even small critical crack lengths will lead to the disc breaking in half.
Rim fractures (diagram at top right) are usually connected to high thermal stress levels and/or overheating of the rim area. However, rim fractures can also be caused by disc oscillations, especially if the mean stress levels are high. In extreme cases, the entire rim assembly can detach itself. A crack can be diverted into this zone if the periphery is experiencing high stress levels (e.g. at the transition point of a shaft socket).
The root of turbine rotor blades is usually fir tree shaped. A root fracture usually lies in the radially outermost tooth (second row of diagrams). In terms of containment, it must be taken into account, that a root fracture results in a significantly higher fragment weight than a leaf fracture.
Unlike in compressors, rotor shaft failures result in runaway turbine rotors that reach uncontrollable overspeed. Therefore the behaviour of the turbine rotors at overspeed after a failure is important with regard to the consequential damages to be expected (see examples in Volume 1).
A preemptive release of turbine blades can prevent the turbine disc from bursting upon reaching overspeed. This lowers the risk of extensive damages. This is a special trait of blades with “fine teeth” (third row of diagrams), as are common especially in older engines, depending on OEM. Due to the expansion of the disc, smaller teeth lose their grip before breaking. Thus the blades are released from the disc and the turbine`s rpm drops rapidly, at least in single-stage engines.
Depending on conditions, blade rubbing on labyrinth partitions can create cracks that lead to ring fractures (bottom diagram). These fragments have relatively little kinetic energy and are usually contained by the overlying static labyrinth ring.
Example "Rejected takeoff due to uncontained failure" (Ref. 8.1-10):
Excerpt: “The flight had an uncontained failure of the No. 4 engine during the takeoff roll. The takeoff was rejected and the aircraft stopped on the remaining runway …Examination of the aircraft showed the engine, pylon, right wing flaps, and aileron, right horizontal stabilizer, and fuselage sustained damage from engine debris released during failure of the engine. Examination of the engine indicated the low pressure turbine fifth stage hub ruptured. About 180 degrees of the hub rim had separated along with the blade attachment slots. The hub ruptured from an area of a well oxidized, intergranular fracture that originated at a tierod hole.”
Comment: This example shows, that bores in the disc can crack due to the relatively high stress levels. In this case the crack obviously went peripherally in the direction of a high stress zone, so that a large piece of the rim seperated along with the blading.
Figure "Typical fragments of rotor blades": The top row of diagrams shows typical fracture points of fan blades in older engine types. The leaves of these blades have a relatively short chord length and are supported by annular struts (clappers). The clappers stiffen the blade ring considerably in parts, so that a bird impact at the tip of the blade above the clapper usually results in a small fragment breaking off. The typical fracture points due to oscillation fatigue are in the area of the clapper and in the contact area of the dovetail root (fluttering oscillations). Due to the supporting and cushioning effect of the clapper, leaf fractures in the transition area to the root platform are uncommon.
It is true for all fan rotor blades, that the strong axial aerodynamic forces in the case of a fixing failure (Example "Rotor blade fixing failure") lead to the release of multiple blades from the disc. These blades can also escape from the containment area towards the front, posing a threat to the safety of other parts of the aircraft.
The middle row of diagrams show the most common fracture points in compressors with wide chord blades. Due to the lack of a clapper, leaf oscillations that threaten the transition area to the root platform are more likely to occur.
The bottom diagrams show fan blades in large modern engines. These blades are either hollow soldered titanium alloys or they are made of carbon fibre-reinforced resin (CFK). These blades have no root platform. The hub contour is made up of separate segments. When the blades are distorted (e.g. bird impact), they push against these segments. If these segments are stiff, this can result in increased stress on the leaf and thus a failure in this zone. Hollow titanium or synthetic fragments are good from a containment perspective due to their light weight and their particular behaviour upon failing.
For containment certification, CFK blades usually fracture at the hub contour, which results in lower fragment weights. However, it must be ensured that the rest of the blading is not overstressed by a blade failure, which would result in an uncontrollable multi-blade failure. This requires special tuning and configuration of the housing and containment.
The introduction of integrally bladed fan discs (Blisk=bladed disk) eliminates the danger of root failure at the dovetail connections. However, a increased risk of fatigue fractures at the connection point between blade and disc can be expected. The advantage of this configuration is that containment measures are only necessary for leaf failures, and not, like in other versions, for the considerably heavier fragment resulting from a root failure.
Example "Rotor blade fixing failure" (Ref. 8.1-20):
Excerpt: “…Two No. 3 engine blades hit and stuck into the upper right side of the fuselage but did not penetrate completely into the passenger cabin.
…Damage to the engine was reported as:
- Spinner cone, inlet cowl from the No. 3 engine are missing along with a 180-deg. segment ot fan stator case and compressor discharge pressure rig.
- First stage fan disk remained intact with only five of the 38 blades still in place. All blades had been dislodged from the second fan stage disk….
“Comment: In this type of large turbofan with high bypass rate, the rotor blades are axially fixed by the spinner. The failure of the threading lead to the blades detaching axially from the front of the disc and being flung out.
Figure "Fragments outside rotational plane" (Ref. 8.1-1 and 8.1-4): If a fragment penetrates a housing, it escapes within a certain angle. The flight paths lie within a cone (top left). The axial divertion is especially interesting. Since the probability of a fragment penetration is roughly the same around the circumference (without regard to varying wall strengths and attached parts), the axial divertion is primarily responsible for determining the endangered structural area. High-energy fragments are only diverted slightly (approx. 5 °), fragments with low energy up to 33° (above right). This is also the minimum axial extension necessary for containments. Naturally, as the distance from the housing wall increases, the axial length of external containments must also be increased accordingly.
Figure "Hazards of disk fragments" (Ref. 8.1-1 and 8.1-4): The kinetic energy of a released fragment is composed of two parts: the rotor fragment moves along its trajectory and thus has translatory energy. At the same time, it rotates around its center of gravity in the direction of rotation of the rotor, because the outer zones have a higher circumferential velocity than the areas located further in on the radius. The fragment also has rotatory energy (top diagram). The penetration ability is determined by the translatory energy of a fragment. An unbroken disc (sector angle 360 °) stores all its energy in rotary motion. This means that it is, as far as penetration ability is concerned, “harmless”. The energy proportions of burst disc fragments depends on the sector angle (bottom diagram). The most dangerous fragments are those with sector angle of about 133 °, because they have the highest proportion of translatory energy .
Figure "Housing deformations by disk failures" (Ref. 8.1-3): Rotor turbines with a relatively heavy hub and short blades have typical damage symptoms after a hub failure. The foremost fragment area in direction of rotation is decelerated when the blades make contact with the housing. The blades are overstressed at the contact points and usually fracture above the root. The housing becomes oval and rips open symmetrically.
Figure "Housing stress by blade fracture" (Ref. 8.1-7): Aside from the translatory motion in its trajectory, a fragment also rotates around its center of gravity (Fig. "Hazards of disk fragments"). Upon release from the rotor assembly, the fragment has the same direction of rotation as the rotor (Fig. "Hazards of disk fragments"). The fragment is also affected by rotatorily directed impulses from impact- and aerodynamic forces that can go against the original motion of the fragment and influence the penetrative ability of the fragment considerably. An example are long rotor blades, as are commonly found in fans. If the blade fails in the root, the tip is stopped by the housing (top diagram, facing page). The heavier, radially more closely positioned blade root has a lower circumferential velocity than the blade tip. It is accelerated by the following blades and thrown against the housing, in this case causing the fragment to rotate in the opposite direction from the rotor. In tests of structural parts in a testing rig then, the released fragment will have the same direction of rotation as the engine.
The bottom diagram shows the calculated relationships (Ref. 8.1-18) between the area of impact of a leaf fragment and the necessary thickness of a titanium alloy wall for containing the fragment. According to these calculations, a fragment striking the wall with the tip will penetrate a housing wall roughly three times thicker than one striking the wall with the section connecting to the root. This seems to contradict the effect described in the top diagram, which did not account for this calculation. This clearly shows how greatly the penetration ability can vary and how susceptible it is to conditions and coincidences.
Figure "Penetration potential of fragments" (Ref. 8.1-7): The shape of a fragment`s impact surface plays an important role concerning its penetration ability (compare with diagram 8.1-6). Fragments with equal mass and different contours can have extremely different minimum penetration speeds (top right). The toughness of the housing is also influential, and dependent on the impact speed (left diagram). It is obvious that this is a complex process, and not very approachable analytically. Since it is not possible to predict the shape of a rotor fragment`s impact surface, the appraisal or rating of a housing`s penetration resistance should always be done with the worst-case scenario in mind. When analysing technical trials, the limited ability to analyse containment behaviour should be remembered. This limited ability to give concrete figures also contributes to the fact that fragments often escape from engines where the containment rating based on available evidence makes this completely unexpected (Example "Engines of older design unable to meet new requirements").
Figure "Containment performance by design": In line with experience, the penetration resistance of multiple concentric housing walls is higher than that of a single wall with the same total thickness (Ref. 8.1-1). Thus, configurations where a rotor fragment must penetrate multiple housing walls in sequence provide good containment.
Example "Containment of debris" (Ref. 8.1-13):
Excerpt: ” The pilots reported that shortly after takeoff, there was a compressor stall in the No. 2 engine, followed by a loss of engine power, and an abnormal oil pressure reading. The engine was shut down and the pilot made an emergency landing…Debris came to rest in a residential area where a vehicle was struck. Examination of the engine revealed that turbine parts punctured through the turbine exhaust case in several locations but were contained by the fan exit duct. All the 4th stage turbine blades fractured across the airfoil just above the platform and were retained in the disk. Twelve of the twenty-five 4th stage turbine vane clusters were missing and the remaining clusters showed considerable trailing edge damage.”
Comment: This example shows that evidence of containment of individual blades does not include haircuts.
Example "Snowballing of minor failures" (Ref. 8.1-8):
Excerpt: “Shortly after takeoff, the flight crew heard a noise from the right side of the aircraft, which sounded like an engine compressor stall. The No.3 engine was shut down and the crew made an uneventful landing at the departure airport. Pieces of cowling from the right engine fell in a residential area, but did not damage. Teardown of the engine revealed that turbine blades on the first stage low pressure turbine wheel had broken off and had penetrated the engine case and core cowl doors.“
Comment: The damage was caused by a loosening of the guide fins in front of the broken blades, and their subsequent rubbing against the low-pressure turbine wheel. This resulted in the failure of multiple blades.
Example "Rotor blade exiting through engine cowling" (Ref. 8.1-12):
Excerpt: ”… (the aircraft) had an uncontained failure of the No. 2 engine during climb to cruise after departing ….The flightcrew reported vibration and loss of power in the No. 2 engine while climbing through 15, 000 feet after departure. The engine was shut down and the flight returned …without further incident. Postincident examination of the engine … showed a fan blade on the first stage of the fan had separated about one inch from the root. The blade exited through the engine cowling at about the 12 o'clock position.”
Comment: This is an older engine with a small bypass ratio. Even though the blade size was relatively small, containment was not guaranteed.
Example "Engines of older design unable to meet new requirements" (Ref. 8.1-9):
Excerpt: “While climbing through 38 000 feet, the aircraft experienced an uncontained fan blade separation in the No. 2 Engine. Investigation revealed that one of the first-stage compressor fan blades had separated about 8 inches above the blade platform. Fragments from the blade had exited through the left fan cowl in three places and had penetrated the vertical stabilizer.”
Comment: This is a larger turbofan of older design. Because the fracture was about 20 cm. above the root, this was far from an entire blade with root. The question arises, why modern containment requirements for entire fan blades are not being met in this engine type.
Figure "Escaping of rotor blades despite containment": This diagram shows a summary of causes in damage cases where blade fragments escaped. These cases are frequently reported, although it would be expected that containment regulations would prevent at least most of these housing penetrations (Examples 8.1-3, 8.1-4 and 8.1-5).
It is worth noticing, that especially in the low-pressure turbines, occurrences of blade fragments escaping are relatively frequent. This could be explained by typical peculiarities of these components:
- Relatively large rotor blades with corresponding mass.
- Thin housing walls. Functionally one-layered design (the internally located segmented fuel guide cannot withstand any tangential stress).
- Long blades with thin cross-sections tend to haircuts and thus a greater stress on the housing.
- Overspeed after failure of the fan shaft creates relatively high-energy fragments.
Figure "Centrifugal strain in different blade parts": In the case of static overstress, such as during high overspeed (tensile stress) or extreme blade rubbing (flexural stress), the toughness of the material and the presence of notching, i.e. tension gradients, determines whether or not a structural part will break. In the root, especially fir tree roots (top left diagram), and at the connection between leaf and root, there are notches which influence the behaviour during failure. Brittle materials are more sensitive to notches (bottom diagram) than tough materials, since the latter reduce tension and distribute it more evenly across the most heavily stressed cross-section.
Material changes must also be critically assessed with regard to overstress. If this isn`t done, there is a risk of the structural parts failing at significantly lower tension levels or of damages due to overstress becoming unacceptably serious.
Such cases can occur if, for example, a tough, multicrystalline turbine blade material is replaced by a single crystal material that reacts more brittly at high deformation rates. This increases the possibility of all blades failing; a so-called “Haircut” (Fig. "Uncontainment by blade 'haircut'").
Whether plastic deformations occur in the root or leaf first depends on the progress of stress in the stressed cross-section (top right diagram). In a notched cross-section, a relatively small plastic lengthening can be expected. Therefore, the corresponding curve above the RPM will be fairly flat until failure. The more evenly and less heavily stressed leaf cross-section will progressively plastically lengthen upon reaching the yield point. The safe amount of clearance needed to prevent a turbine blade failure at high overspeed depends heavily on the toughness of the material. If the material`s behaviour changes noticeably at different operating temperatures, then this should be taken into account when considering containment or at technical trials calculating the RPM levels at which bursting occurs.
Example "Broken off blade tip leading to blade jam" (Ref. 8.1-11):
Excerpt: “…The flightcrew reported that as the airplane was climbing through 1,000 feet, they heard a “soft thunk” and the No. 1 engine spooked down to about idle….The on-site examination of the engine showed that the core cowl had a 15-inch long x 6-inch wide hole at about 3:00 o'clock in the plane of the LPT Stage 4 rotor. The LPT case had several penetrations in the plane of the LPT Stage 4 rotor. The examination of the airplane revealed numerous dents and impact marks on the underside of the left inboard aileron and the outboard side of the inboard flap track fairing aft of the No. 1 exhaust….There was one LPT Stage 1 blade tip broken off and the remaining downstream LPT stages had damage to the blades that was progressively worse to the rear of the engine. The turbine case wall thickness, hardness, and material composition were found to conform to the requirements.”
Comment: It is worth noticing, that the failure of a blade tip could cause such heavy, uncontained damage to the following blading. This shows that containment trials in testing rigs (concerning the extent of consequential damages) cannot be expected to sufficiently cover real engine conditions. In this case, most of the blades in the fourth low-pressure turbine stage failed. This obviously exceeded the containement capacity of the housing and resulted in the escape of multiple blade fragments in a limited circumferential area. One can conclude that a blade jam took place in this zone.
Figure "Uncontainment by blade 'haircut'" (Example "Containment of debris" and Example "Broken off blade tip leading to blade jam"): Containment certifications are usually done for the failure of individual blades. However, if a blade fragment jams in certain locations, this can lead to a failure of the entire blading (haircut). If the fragments jam in a small area, the released energy is concentrated there, understandably creating significantly higher stress levels in the housing than would be present due to a single blade failure. The housing wall can be breached and a large number of fragments escape at the same spot, thus concentrating the concequential damages. It is also important, when choosing blading material, not to increase the possibility of a “haircut”. The toughness of the blade material at high deformation rates and normal operating temperatures plays an important part in this (see diagram 8.1-10).
Figure "Factors influencing 'haircuts'" (Ref. 8.1-4): When all the blades on a rotor fail, it is referred to as a “haircut”. In this instance, the housing experiences high stress levels. Therefore, one must endeavor to avoid haircuts. The susceptibility of blading to haircuts depends on many factors, which can be classified into three main groups:
Construction: To ensure that blade fragments are as light as possible with little energy, blade failures due to flexural overstressing should be kept as far outward on the blade as possible (bottom diagram, Ref. 8.1-4). In any case, failure of the entire blade along with the disk`s fir tree toothing must be avoided. This behaviour is influenced by the structural design of the blade. A general rule is that the stress in the dovetail of the disk should be less than in the root shaft of the blade, which should in turn be less than at the connection point between leaf and root platform. Apart from the blade properties, the surrounding parts, such as control devices, housings, rotors and their bearings are also important in determining behaviour during failure (e.g. stiffness, i.e. flexibility and hardness).
Operation: The temperature of a structural part is very important with regard to its behaviour in case of failure (Fig. "Containment clamping influence"). High leaf temperatures usually lower hardness and increase ductility. When temperatures approach those of the initial fusing (in the region of the solidus temperature), nickel alloys behave brittly and show fission-fracture like signs. These fractures are easily confused with fatigue fractures. If the disc ring overheated along with the blade roots, the fir tree toothing may fail. The high centrifugal and high circumferential speed typical of overspeed can change the manner in which the blading fails. This is due in part to the fact that the deformation rate rises, increasing embrittlement.
Material: The blade material is primarily responsible for the manner in which it fails. Typical coarse-grained cast materials tend to brittle cracks when subjected to solidus temperature and impact stess. This is especially true for single-crystal blades. Experientially, the danger of a haircut is greater in single-crystal bladings than in polycrystalline materials. This tendency must be remembered when changing a blading from polycrystalline to single-crystal in order to permit higher operating temperatures. In each case, the containment ability of the housing should be reassessed. Brittle blade fragments lose kinetic energy when they shatter, making the fragments easier to contain due to their relatively low energy. When changing structural materials or assessing blade materials, the density of the material must be taken into account. For example, there are single-crystal materials that contain wolfram, making them considerably (about 10%) heavier than the standard nickel alloys.
When considering the introduction of “alternative” brittle blade materials such as ceramic or intermetallic phases (IP), their susceptibility to sudden failures of the entire blading should be an important criteria. Intermetallic phases exhibit a brittle behaviour up to a few hundred °C before becoming ductile. They are also especially susceptible to failure at the low temperatures present during startup, increasing the danger of foreign object damage. If a fragment breaks off, it will most likely result in a avalanche-like self-increasing destruction of the blading.
Figure "Stresses during blade fragment containment": The failure of a typical large fan blade in a modern engine with high bypass ratio stresses the entire engine structure. Controlling this type of occurrence is a very demanding design task. A fan blade failure is followed by extreme imbalances, oscillations in the housing and rotors, large torsion moments due to rubbing against the housing and rotor deceleration, combustible abrasions, bending of the rotors, and high bearing loads. These influences on the engine components lead to typical overstresses and damages:
Penetration stress against the fan housing wall. When the failed leaf fragment is “run over” by the tips of the remaining blades (provided there is no nesting behavior), extreme housing oscillations (Fig. "Housing damage by rotor rubbing") occur, subjecting flange connections to high levels of dynamic stress (Fig. "Containment behavior and design characteristics").
The extreme imbalances that follow a blade failure can lead to an overstressing of the engine suspension and/or the airframe, i.e. the wing structure. Thus the release of the engine in extreme damage situations has been provided for (Fig. "Design of engine mounting bolts for controlled failing").
The radial forces and the flexing of shafts and housings can overstress the main bearings. The sudden appearance of powerful radial forces can induce high axial forces in the bearings and thus the failure of the bearing rings, to name an example. Seal failure in the bearing chamber due to deflections in the rotors and housings can result in heavy oil leakage and thus an oil fire.
Housing braces that support the bearings are stressed by radial forces from imbalance in the main bearings and by torsion moments in the case of a fan blade failure. Torsion moments occur, for example, during a canting of the bearings. This is true not only for housings near to the fan. Even housings near the engine`s output can be overstressed due to fan damage.
The engine type affects the stressing of mounted components, such as regulators, gearings, or components of the fuel and oil supply systems. If these devices are mounted on the bypass duct or fan casing, as is usual in warplanes, the stress of a blade impact and/or strong vibrations in the low pressure rotor can cause the mountings to fail. Consequential damages can also include failure of fuel or oil lines resulting in fires.
rubbing of flammable surfaces creates combustible dust and the possibility of a dust explosion. This type of explosion can overstress flange connections to the point that housings are torn open or even torn off.
Heavy rubbing in the compressor can result in a titanium fire.
High torsion moments caused by rubbing can lead to a failure of the rotor assembly. Even a rotor drum can be damaged to the point of fracturing by overheating and/or sectional weakening.
Figure "Brittle material behaviour at impact" (Ref. 8.1-7): During situations necessitating containment, fragments, housings, and parts affected by consequential damages are subjected to high deformation rates. In order to correctly assess containment-related damage sequences with high stress rates, it is necessary to know the behaviour of the materials involved.
The top diagram is of a low-alloy steel pipe with 20 mm thick walls, that reacts in a very ductile manner during normal damage trials. At 50,000 rpm in a testing rig, an integral turbine disc from a small helicopter engine caused this pipe to burst. The steel reacted in a very brittle manner. Multiple large pieces broke out of the pipe without noticeable deformation, meaning that the energy taken from the disc fragment was minimal. Thus, this type of material is not suitable for containment rings. In general, a forged version of the same material is tougher than a cast one.
Welds along containment rings have proven to be exceptionally problematic, because they are notched themselves (molded notches- e.g. undercut, stiffness notches, e.g. due to overly high seams and structural notching), and because the cast structure of the welded seam tends to fracture brittly.
The middle diagram illustrates the phenomenon described above. One can see that, as the deformation rate increases, the elastic limit rises more sharply than the tensile strength and meets it in the embrittlement zone. If the elastic limit is equal to the tensile strength, a spontaneous break occurs without noteworthy plastic deformation; the part breaks brittly. This is evidently not followed by an even decrease in the ultimate strain, but a sudden, steep drop at a certain expansion rate.
Many materials do not exhibit this pronounced embrittlement behaviour, at least at deformation rates normal in containment situations (bottom diagram), even though a rise in hardness is clearly observable. This makes these materials more suitable for containment rings. They include austenitic steels and superalloys. 13% Cr-steel doesn`t exhibit this steep drop, but its ultimate strain is comparitively low.
Figure "Containment clamping influence" (Ref. 8.1-7): At first glance, the minimum penetration speed increases linearly with the wall strength. The angle of this line can, however, vary depending on the materials (top left diagram). At first glance, the minimum penetration speed also increases with the tensile strength of the housing material.
It is interesting, then, that this does not apply to some superalloys at high temperatures (right diagram). This behaviour can be seen in connection with a hardness increase in the temperature zone in question and/or an increase in ductility.
A superficial observation could give the impression, that a completely built-in, “well-supported” housing wall would offer more resistance to a fragment than a flexibly hung wall. This is not true at high impact speeds.
The chart at the bottom shows the minimum penetration speeds of various structural materials with different fixings. The attachment conditions varied the flexibility of the sample. The less the sample could flex (white bar), the lower the penetration speed. The highest penetration resistance, i.e. the highest minimum penetration speed was exhibited by the samples affixed at only one side (black bar).
Figure "Housing damage by rotor rubbing": Even a blade fragment that upon impact does not place any serious strain on the housing can lead to a housing failure and escape.
Diagram 1 shows the moment a large blade fragment is thrown from a fan stage. It is then “run over” by the following blades. The untwisting of the leaves increases the radial length of the blades, causing them to bend and jam (right diagram). Thus, the housing is stressed by strong pulsing radial forces that ovalize it and jerk it forward.
The housing is subject to strong, self-strengthening vibrations (diagram 2) that increase until the rotor blades are subjected to extreme bending stress. This results in the failure of the entire rotor blading (diagram 3). In order to control this phenomenon, at the very least massive flange connections and a flex-resistant housing wall are necessary (Fig. "Containment behavior and design characteristics"). This damage sequence is avoided in nested housings (Fig. "Containment with 'nesting' in the fan area").
Figure "Cause of uncontrolled rotor speed": Evidently, the fragmentation of rotors as a result of uncontrolled overspeed is not completely avoidable, despite measures such as designing the rotor for safety at high overspeed and an accordingly designed regulator (Example "Misuse of engine leading to overspeed").
The case in point concerns a twin-shafted low-output helicopter engine with a gas generator and a free turbine. The mechanical regulator is powered by the free turbine through the main gearbox. When a gear “ahead of the regulator” breaks (bottom diagram), it registers a drop in RPM and tries to correct it by increasing fuel flow. Consequentially, the gas generator and free turbine reach overspeed, leading to a turbine disc failing in the gas generator.
This shows, how important to safe operation the location of the RPM measuring devices can be in exceptional cases.
Example "Misuse of engine leading to overspeed" (Ref. 8.1-14):
Excerpt: “…(the customer) also had reported an uncontrolled overspeed problem in the engine…(the OEM) resolved the problem by modyfying the engine, giving it a redundant overspeed protection system.
Comment: This case concerns a shaft-power engine that was designed for helicopters and here was used as a propeller engine.
Example "Tail-mounted engine debris damaging empanage" (Ref. 8.1-15, Diagram 8.1-17):
Excerpt: “A burst of shrapnel that followed failure of the tail-mounted…engine…caused 50 hits on the empanage….The first-stage fan disk of the No. 2 engine is the prime suspect in the investigation….'The desintegration of the fan was of such magnitude that it tore all the lines not just in one place, and even in the side walls of the aircraft,' one investigator said. Investigators are unsure whether any design of hydraulic systems could survive the desintegration that occurred on the …flight. Some aircraft are equipped with fuse links
- devices that detect changes in flow of hydraulic fluids
- which shut off and preserve fluids to retain some form of control. But such devices are not a perfect solution.”
Comment: The fan disc failure was due to a inhomogeneity of the titanium alloy. An LCF crack
Figure "Failing of 'redundant systems' by fragments" (Ref. 8.1-15, Example "Tail-mounted engine debris damaging empanage"): The hydraulic systems for setting the rudders and elevators were multiply redundant. However, they all failed due to the damage from the fan disk fragments. It is interesting, that the energy of fragments at a 33° angle of dispersion was sufficient to cause dangerous amounts of damage, although one would expect this to be relatively low (see Fig. "Fragments outside rotational plane").
Figure "Failing of thick cross sections by rubbing" (Ref. 8.1-16, see Example "Damage to disk hub caused by friction "):
In this shaft-power engine (top diagram), rubbing of a thin plate in the disk area (middle diagram) resulted in a disk failure.
Turbine disk failure as the result of a filigreed, thin (relative to the disk width) section of metal plating causing rubbing is a fairly typical damage sequence. This danger is often underestimated. Evidently, the plating behaves like a friction band saw, and is capable of splitting massive cross-sections. The bottom left diagram shows an integrally cast turbine disk from another engine that was caused to fail in a similar manner. In particular, membranous soldered constructions have been shown to cause this type of damage (Fig. "Damage of rotor disc by rubbing membrane"). In order to avoid this type of damage, appropriate constructive measures should be taken (Ref. 8.1-4, Ill. 8.2-21 and 8.2-22).
Excerpt: “Following a complex damage sequence, an aircraft engine broke apart after 5200 hours of operation… As shown in the principal diagram, the idler is fitted with a expansion unit made from two cambered metal plates to compensate for heat expansion. Due to the low-cycle stress during start up and shut down, the entire inner periphery of the intake side expansion plate, and 85% of the output side plate`s outer periphery had split (stiffness cracks at the soldering points). The shaft seal and the plate on the outtake side then tilted and rubbed against the 3rd stage disc, causing the plate`s camber to be reduced to about two thirds of its circumference. The jagged-edged metal (cobalt-based alloy) ring that formed at the shaft seal`s inner thrust ring dug into the neighboring rotor like a grinding tool and created a 3 mm deep furrow around the whole of the wheel. It is testimony to the quality of the rotor material (nickel-based alloy), that this sequence did not lead to immediate rotor failure. Not until a good number of cracks due to oscillation had formed at it base, did the rotor fall apart (sectional weakening and stress concentration). Instead of setting the inspection intervals at 6000 hours, the manufacturer should have used the number of start-ups as inspection criteria.”
Comment: This damage sequence strikingly demonstrates the typical effects of a thick cross-section being split by rubbing from a thinner metal sheet.
Excerpt: ”… It happened about 600 meters from the point of brake release…Aircraftspeed at that time was about 100 kt. The abort was initiated about 120 kt., and the aircraft came to a stop approximately 1,900 meters from the start of the takeoff roll. Leaking fuel from debris impact points on the lower right wing was ignited, and flames were fanned by a tailwind of about 8 kt. The fire spread below the fuselage to the forward part of the aircraft. Evacuation of the passengers and crew was made through the …rear left-hand door.
Damage from flying engine debris was incurred to the right wing and flaps and slats. Impacts also occurred on the left engine, which apparently resulted from pieces thrown by the right-hand powerplant, which bounced off the runway. All four tires on the right main landing gear were blown.
Comment: This is a case in which a rotor fragment also damages the second engine.
Figure "Production caused turbine disk failure" (Example "Flying engine debris bouncing off runway", Ref. 8.1-17): This case demonstrates, that a rotor fragment (bottom diagram) from an engine (“A”) can escape in such a way that it damages an engine on the other wing (“C”). In this case, the relatively small annulus fragment bounced off the runway (“B”). There have been other cases, however, in which engine failure during flight affected the other wing.
8.1-1 E.A. Wittmer, T.R. Stagliano, J.A. Rodal, “Engine Rotor Burst Containment/Control Studies”, Proceedings AGARD-CP-248 of the AGARD Conference “Stress, Vibrations, Structural Integration and Engine Integrity”, Pages 15-1 to 15-30.
8.1-2 “T.W. Combe, D.F. Vowles, “Structural Effects of Engine Burst Non Containment”, Proceedings AGARD-CP-248 of the AGARD Conference “Stress, Vibrations, Structural Integration and Engine Integrity”, Pages 9-1 to 9-10.
8.1-3 C.J. Mangano, “Studies of Engine Rotor Fragment Impact on Protective Structure”, Proceedings AGARD-CP-248 of the AGARD Conference “Stress, Vibrations, Structural Integration and Engine Integrity”, Pages 10-1 to 10-24.
8.1-4 D. Mc Carthy, “Definition of Engine Debris and Some Proposals for Reducing Potential Damage to Aircraft Structure”, Proceedings AGARD-CP-248 of the AGARD Conference “Stress, Vibrations, Structural Integration and Engine Integrity”, Pages 7-1 to 7-10.
8.1-5 J.J. DeLuca, B.C. Fenton, S.P. Petrie, J.T. Salvino, “Lightweight Aircraft Turbine Protection”, Paper AIAA 93-1815 of the “AIAA/SAE/ASME/ASEE 29th Joint Propulsion Conference and Exhibition”, June 28-30, 1993/Monterey, CA, Page 1-7.
8.1-6 J.T. Salvino, G.J. Mangano, R.A. DeLucia, “Rotor Burst Protection: Design Guidelines for Containment”, Proceedings AGARD-CP-248 of the AGARD Conference “Stress, Vibrations, Structural Integration and Engine Integrity”, Pages 19-1 to 19-16.
8.1-7 J Thiery, (German, translated from French) “Festigkeit von Triebwerksgehäusen bei Schaufelbrüchen” Proceedings AGARD-CP-248 of the AGARD Conference “Stress, Vibrations, Structural Integration and Engine Integrity”, Pages 7-1 to 7-10.
8.1-8 NTSB Identification: CHI851A032, microfiche number 27668
8.1-9 NTSB Identification: DCA891A066, microfiche number 39111A
8.1-10 NTSB Identification: MIA96FA013.
8.1-11 NTSB Identification: DCA96SA026.
8.1-12 NTSB Identification: MIA96SA157.
8.1-13 NTSB Identification: NYC96IA168.
8.1-14 R.W. Moormann, “Power degradation target of CT7 improvements”, periodical “Air Transport World”, 3 /88, Page 108.
8.1-15 J.Ott, “Investigators Find Reconstructed Tail of DC-10 Riddled With Damage”, periodical “Aviation Week & Space Technology”, August 7, 1989, Page 22 and 23.
8.1-16 G. Lange, “Systematische Beurteilung technischer Schadensfälle”, DGM Informationsgesellschaft, Publisher, Page 177.
8.1-17 “A300 Damaged by Engine Fire”, periodical “Aviation Week & Space Technology”, March 29, 1982.
8.1-18 A.D. Lane, Development of an Advanced Fan Blade Containment System”, DOT/FAA/CT-89/20, Final Report, Oct. 1988-Apr. 1989, Page 1-25.
8.1-19 Dr. S.S. Birley, Dr. R.S. Tracey, Dr. M. Kearns, “The Effect of Microstructure on the Penetration Resistance of Titanium Laminates”, Proceedings Volume 3 of the conference “Ballistics `92 1-3 June, 1992, Stockholm-Sweden. Page 145-152.
8.1-20 “DC-10 CF6 Engine Investigation Pressed”, periodical “Aviation Week & Space Technology”, November 12, 1973, Page 28 and 29.