Aeroengine Safety

Figure "Low risk by rotor fragments engine positions" (Ref. 8.2-1): The top diagram shows the typical “technical high-risk zones” of a commercial aircraft (the conditions in supersonic aircraft are covered in Ref.. 8.2-3), where the impact of a rotor fragment must be avoided at all costs. These zones can be grouped as follows: areas with flammable materials (fuel system, especially tanks (Example "Rejected takeoff due to uncontained failure")), steering system components (e.g. actuators and checkback lines), and important supporting structural sections of the frame (e.g. the main wing spar). Since damage to persons should also be avoided, the cabin area is also classified as an unacceptable risk zone (bottom diagram).
For risk assessments, the “effective diameter” of a 1/3 rotor fragment is used. This is the diameter of the circle which encases the rotating fragment (detail sketch at left center, see also Fig. "Typical fragments of rotor blades").

Figure "Probability rotor fragments striking the nacelle" (Ref. 8.2-1): To specify the “hit probability” (probability of a catastrophic impact) of certain aircraft zones, the following approach can be used. The “effective diameter” (definition see Fig. "Low risk by rotor fragments engine positions"), the direction of rotation of the engine rotor in question, the zone in which impacts must be avoided (“taboo region” dark grey area), and the geometric conditions of the fuselage and engine determine the angle of the allowable “scatter segment”. In the three cases dealt with here, this angle was between 3° and 36°. This “scatter angle” as a fraction of 360° equals the hit probability.

Figure "Direction of rotation influences engine position" (Ref. 8.2-1): The layout of local containments (penetration protection, see Fig. "Suitable containment designs" and 8.2-10) depends upon the engine arrangement and the airframe, the direction of rotation of the failed rotor (detail diagram bottom right, note the differences between left and right engines), the “taboo zone” (definition Fig. "Probability rotor fragments striking the nacelle"), and the “effective diameter” (detail diagram at center) of the fragment.
The top diagram shows the conditions for a 1/3 disk segment and impact protection mounted on the side of the engines. The conditions in supersonic commercial aircraft are given in Lit 8.2-3.
The bottom diagram shows the conditions for impact protection mounted on the side of the frame (bumper, deflector, see also Fig. "Features of disk burst-protection concepts").

Figure "Rotation direction demands position of auxiliary components": In order to avoid failure of both engines the mounted components, such as electronic regulators and oil and fuel systems on the side of the engine, should if possible be mounted on the engine in such a way that the probability of them being damaged by fragments from the parallel engine is minimized (see ETOPS, Volume 1 Chapter 3). To this end, it is important that in twin-engine aircraft, only one of the two wing-mounted engines has the potential to damage the other due to possible fragment trajectories (top diagram). The fact that only one of the engines can damage the other is due to the direction of rotation of the shafts being the same and the parallel arrangement of the engines.
This makes it possible to position the ancillary equipment suitably (bottom diagram).

Figure "Features of disk burst-protection concepts": Various options for burst protection from usually uncontained rotor fragments are discussed here. These are usually affixed to housings and other components (e.g. combustion chamber). These burst protection measures can be divided into two groups: “engine side” (top left and center diagrams) and “frame side” ( deflector, top right diagram).

Engine side burst protection can be installed internally (top center diagram) and externally (top left diagram). This type of burst protection usually covers the entire circumference and is ring-shaped.

Internal burst protection is used in low output engines (helicopters, APU`s) to also contain disk fragments (Figs. "Absorbing rings containing disk fragments", Burst protections for turbines in APU's" and "Containments for turbine rotor fragments"). The internal burst protection is located closer to a potential rotor fragment`s orbit due to its smaller radial spacing (“a_i”). This means that relatively flat angles of impact a_i can be assumed (bottom diagram), putting correspondingly less stress on the burst ring (Fig. "Estimation of containment effectiveness"). Thus the burst ring can be relatively thin (“h”) and light yet still effective. Also, the ring can be relatively narrow (short axial length “b”), because the axial angles of deflection will not come into play. However the extremely limited space inside engines, makes it difficult to install sufficiently thick burst protection (Fig. "Absorbing rings containing disk fragments" and 8.2-13). High operating temperatures (e.g. for materials made of organic and/or oxidation sensitive fibres) in the rear compressor and hot part zones limit the selection of materials.

External burst protection has advantages and disadvantages compared to the internal varieties. Advantages include more space for installation and relatively low operating temperatures. The low operating temperatures allow the use of fibre materials even in the hot part zone (Fig. "Turbine disk aramid containment"). The housing and component walls in the fragment`s trajectory (sum of wall strengths “h”) have already drained energy from the fragment and possibly altered and “mitigated” its shape (e.g. rounded off corners, bent blades). This lowers the burden an external burst protection has to bear.
However, external burst protection is subject to steeper angles (“a_e”) of impact due to its greater radial spacing (“a_e”)and therefore the penetration stress placed upon it is increased (Fig. "Estimation of containment effectiveness"). Also, the axial angles of dispersion mean that the burst protection must be fairly wide (“b”), which can noticeably increase the space and weight required for its installation.

Frame side burst protection (top right diagram) is used in the form of a deflector, and can only protect a limited area (see Ref. 8.2-3 and 8.2-4). It can be installed at such an angle that potential fragments strike it at a flat angle. The geometric space for design is large and the choice of materials (low operating temperature) is relatively broad. The position of the deflector is dependent upon the direction of rotation of the relevant rotor and thus not symmetrical with the center of the hull (Ill. 8.2-3).

Figure "Fiber material containment for military aircraft?" : Containment rings made from fibre materials, especially aramdid fibres, are very light despite their outstanding resistance to penetration. This advantage can be used in turbo fan engines with a high bypass ratio and a large diameter. Additionally, a nesting arrangement (Fig. "Containment with 'nesting' in the fan area") limits potential consequential damages. However, in containment situations, this technology results in relatively large radial expansions, that can reach 10 cm in large turbo fan engines (bottom diagram). This degree of expansion is tolerable in wing-mounted nacelle engines (center diagram), provided no components are mounted in the fragment`s trajectory which would limit the necessary expansion.
Fibre containment rings are not (as far as is known) used in engines that are integrated into the aircraft fuselage (e.g. tactical aircraft, top diagram). The main reason for this is that the attractive weight advantages of fibre materials when compared to conventional metallic containments (e.g. locally reinforced ring-shaped housings) are only relevant if there is enough space (envelope) for expansion of the fibre ring (energy transfer, nesting). The extremely small space around an engine in the frame means that this is usually not the case (Ref. 8.2-5). If important mounted components whose failure would render the engine uncontrollable (e.g. regulators, gear boxes)or lead to flammable materials escaping (e.g. oil, fuel) are positioned in the plane of a fragment (bottom diagram, Fig. "Position of accessory equipment to containment"), then the space available for a containment expansion is further reduced. The maximum permissible expansion of the circumferential contour is thus usually, depending on location, within a centimeter. The fibre containment ring would therefore have to be so stiff (thick, heavy) that it would greatly hinder penetration (stress, clamping influence see Fig. "Penetration potential of fragments"), meaning that the weight potential of fibre materials could not be realized.

Figure "Suitable containment designs": An engine-side containment can be implemented in various styles and geometries depending on the specific requirements of the engine type.

“A”: Tactical aircraft and older civil engine types use a locally reinforced metal housing, usually made of a titanium alloy, in the fan region (Fig. "Housing damage by rotor rubbing", Ref. 8.2-6). The reinforcement can be designed as a ring-shaped rack (dotted contour, Ref. 8.1-6) in order to increase housing stiffness against oscillations resulting from a blade fragment being “run over” (Fig. "Housing damage by rotor rubbing").

“B”: A thin-walled housing can be brought up to a sufficient containment level with an aramdid fibre bandage (Fig. "Containment with 'nesting' in the fan area" and 8.2-16, and Lit 8.2-6). If the metal housing has a rounding or obtunding effect on sharp-edged fragments, the effectiveness of the fibre ring is increased. This type of configuration can only be implemented in regions where the typical operating temperatures are low, such as the fan. The assertion that it is not possible for a blade fragment to jam in such a way that several rotor blades are broken and/or overstress the housing, i.e. its flange connections, after contact with the fragment must be supported with evidence from technical trials. The fibre bandage must be sufficiently wide, i.e. its axial length must be considerably larger than the dispersion angle of possible fragments, in order to safely prevent the fragment from escaping laterally (Ill. 8.2-16). A stiffening frame limits the expansion and constriction of the bandage (Ill. 8.2-16).

“C”: The fibre containment is installed over the housing wall, at a height appropriate to the blade size. Between the containment and the wall, e.g. aluminum honeycomb structures can be inserted, thus letting the blade fragment break through the thin housing wall before being trapped between the wall and the fibre bandage (nesting, Fig. "Containment with 'nesting' in the fan area"). This minimizes consequential damages. The lateral constriction of the bandage prevents the blade fragment from slipping out (Ill. 8.2-16). This type of containment configuration is found in the fan region of large, modern turbofan engines.

“D”: Internal burst prevention rings (Figs. "Absorbing rings containing disk fragments", Burst protections for turbines in APU's" and "Containments for turbine rotor fragments") are often used in the turbine region of low-output engines (e.g. helicopters) as a re-fitting in cases where an unacceptable risk due to disc fragments has been determined. Due to the relatively high operating temperatures, only sufficiently heat-resistant metals such as superalloys or high-alloy steels are suitable for use in these rings.

“E”: If a housing`s acceleration, expansion, and/or the penetration resistance must be limited, especially locally at the periphery (Ill. 8.2-3), the use of brittle materials (e.g. ceramic tiles) can be considered (Fig. "Containment systems of fiber materials or ceramics" and 8.2-10). These materials absorb a fragment`s kinetic energy by shattering upon impact. This prevents extreme accelerations from affecting the mounted components. Additionally, the high degree of hardness prevents the housing wall from being sliced or cut open by sharp-edged fragments. The high heat-resistance of ceramic materials allows their use in hot part regions. There has been no word of a serial implementation.

“F”: The combination of variants “D” and “E” can be used in situations, where a local minimizing of expansion is required. This might be in the region of a mounted component that must not be damaged. It is also thinkable, that this might be applied to prevent the escape of fragments in a certain circumferential region (see Ill. 8.2-3). Here also, there are no records of a serial implementation.

Figure "Estimation of containment effectiveness" : The top diagram shows (Ref. 8.2-4) the housing wall strength necessary for containment according to the given formula (solid line), and the scatter range of actual operation experiences (grey zone). The characteristic penetrating power depends greatly on the angle of impact. A steep angle of impact qualifies a large penetrating power, i.e. penetrating ability of the fragment. Evidently, a high ultimate strain increases the penetration resistance of a housing wall considerably more than an increase in strength. According to this formula, a single massive wall would seem to be more penetration resistant than several thin housing walls of the same total thickness.
However, the opposite tendency can be observed (Ref. 8.2-8) if the fragment`s energy can be properly distributed among the individual walls and their deformability taken advantage of. In ballistics, laminates of many layers of metal plating and combinations with synthetic and ceramic layers are known to be highly resistant to penetration while also being lightweight (e.g. use in personal body armor).
The formulas at the bottom (Ref. 8.2-7) are used by large OEMs to estimate the strength of containment walls “t”. It is interesting that, with the exception of the effect of the translatory kinetic energy “KE”, the formulas are considerably different. The equalization must lie in the constants “K₁” and “K₂”, which include the know-how of the user. Apparently these constants take into account the influence of the angle of impact.
It is also interesting to note that these formulas do not take into consideration the properties of the fragment, such as strength, deformability, and geometry (e.g. stiffness, sharpness). However, in extreme cases concerning fan and rotor blades, fragments made of fibre-reinforced synthetics will have considerably less penetrative ability than fragments made of the usual high-hardness materials (e.g. titanium alloys) with the same kinetic energy. This indicates that these are only estimates, the correctness of which can be demonstrated in trials simulating operating conditions.

Figure "Aramid fibers fan blade containment rings": The top diagram (Ref. 8.2-7) shows the weight advantages of ceramic/synthetic fibre laminates (fibre material: ballistic polyethylene fibres) compared with an aramdid fibre containment. With high energy leaf fragments, the weight advantage can be up to 40%, and with low energy fragments, this advantage is about 10%. This may be due to ceramic`s ability to absorb large amounts of energy when shattering and its resistance to being sliced. The bottom diagram (Ref. 8.2-5) shows the advantage of layering even in synthetic webbings. A linear increase in the penetration energy is met with a quadratic increase in the number of layers in the containment ring.

Figure "Containment systems of fiber materials or ceramics": The top diagram shows typical strength ranges of several materials. It is worth noting that, evidently, the elasticity modulus is only of secondary value when determining penetration behaviour, since it does not appear in the formulas for estimating housing wall strength (Fig. "Estimation of containment effectiveness"). It can be assumed, however, that the influence of elasticity, i.e. stiffness, in fibre bandages and layered structures cannot be ignored. This is especially true in situations where minimizing the expansion of the bandage upon impact of a fragment is vital. The bottom left diagram (Ref. 8.2-7) can be used to estimate the thickness of an aramdid ring necessary for a certain fragment energy (see formula in Fig. "Estimation of containment effectiveness"). The bottom right diagram shows the grammage advantages ceramic/synthetic fibre laminates have over webbing made of pure aramdid fibre (compare Fig. "Aramid fibers fan blade containment rings").

Figure "Containment technologies": Ceramic plates (e.g. boron nitride or aluminum oxide) in combination with fibre webbing (Ref. 8.2-10 and 8.2-11) can absorb the kinetic energy of a fragment in an entirely different way than pure fibre layers. Fibre materials “kill” the energy through the friction of the fibres rubbing against one another during expansion and/or stretching until they fail. Thus, in order to absorb sufficient energy, a relatively large deformation is required (bottom diagram).
Ceramic (brittle) materials are completely different in that they absorb energy by shattering at the point of impact. Even if the ceramic tiles are affixed with silicon to a thin support made of aluminum sheeting and separated by partitions, sufficient energy is absorbed in shattering if a fragment strikes the rack. The tiles` support is deformed only slightly at the point of impact. This type of containment does not require much space to deflect a fragment and is suitable for use in fuselages of tactical aircraft. A further advantage is the relatively low acceleration the surrounding components are subjected to.

Figure "Absorbing rings containing disk fragments": This small shaft powering engine in a twin-engine helicopter has been fitted with an internal energy absorbing ring in the first stage of the gas generator (top diagram). While this ring is unable to prevent fragments escaping in the case of a hub failure (that would be a burst protection ring, Fig. "Burst protections for turbines in APU's"), it can contain annulus fragments. These fragments can be released due to thermal fatigue, especially in integrally cast turbine disks (bottom diagram). This type of fragment would threaten the parallel engine and put the craft in acute danger of crashing.

Figure "Burst protections for turbines in APU's" (Ref. 8.2-12): Concerns internal burst protection measures for overspeed-induced hub failures in engines with very low output (APUs). The left diagram shows a schematic example of a burst protection ring (black) for an axial turbine disc. The right diagram is of a burst protection device for a radial turbine disc. The protection ring is shaped like a thin disc, which is positive-fit to the hub. This layout is designed to accommodate radial disc failures due to overspeed-induced overstress (back side of hub).

Figure "Containments for turbine rotor fragments" (Ref. 8.2-13): Concerns an internal burst protection in the turbine region of a small helicopter engine. For economic reasons, type CrNi18/10 austenitic steel seems to have been selected as the ring material, although the cobalt alloy HS 25 would have been better. The ring was positioned as closely as possible to the turbine disc it was to protect (see Fig. "Features of disk burst-protection concepts"). A secure axial fastening of the protection ring is important. The inside of the ring has been circumferentially furrowed in order to prevent the axial escape of fragments and to direct them circumferentially. At the same time, the raised sections between the grooves serve to stiffen the ring (top right sketch), better protecting the relatively thin housing wall from impacts. The “windows” above the guide vanes both reduce weight and enable the uncoupling of protection rings from individual stages. The bottom diagram depicts the ring after a containment trial.

Figure "Containment with 'nesting' in the fan area" (Ref. 8.2-18): Containments can be designed in such a way that they not only trap fragments in the engine, but also increase the engine`s safety in case of damage and minimize consequential damages. These damages include the failure of the remaining blading of the affected or following stages, which can lead to an overstressing of the housing. Additionally, it is important to keep the damage-induced stresses to the housing and engine suspension within an acceptable range.
The large expansion of aramdid bandages when catching a fragment makes it suitable for minimizing damages. However, one consciously accepts that the fragment will penetrate the thin housing wall and will be trapped outside the gas duct between the bandage and the housing wall (nesting, top sketch).
A similar result can be attained by furnishing the inside of a purely metallic, thick-walled housing with a honeycomb structure. This can be sufficient for trapping a blade and its root platform, as the case may be (center diagram).
The bottom sketch shows a configuration (Ref. 8.2-15) that prevents subsequent blade failures and overstressing of the housing, if not damages, in the following stages of the engine (e.g. in a multi-stage fan in a tactical aircraft). In this case, the leaf tip region is constructively weakened (e.g. similar to a “feather edging”), so a “running over” of the fragment results in the tips of all blades shearing off and the rotor running freely.

Figure "Containment behavior and design characteristics": In turbofan housings with metallic containments (local ring-shaped wall thickenings and/or radial partitions running along the inside, Fig. "Suitable containment designs" version “A” ) that lack an inner lining for receiving fragments (Fig. "Containment with 'nesting' in the fan area"), special structural measures must be taken to prevent the housing from being overstressed in containment situations (Fig. "Housing damage by rotor rubbing").
The left diagram depicts a large, first-generation turbofan engine with a high bypass ratio. Imbalances following a blade failure are transferred from the outer flanged shaft to the load-bearing housing via the bearing. The forces of impact and rubbing are absorbed through the flange, putting high dynamic stress on the flange connection. The flange cross-sections are solidly built, the flange is outwardly centered to absorb the radial forces, especially shear stress. The connection consinsts of
tension screws with spacer sleeves (detail). Housing and flange deformation is limited by T-shaped racks running around the outside.
The right diagram depicts a fan in an older engine with a small bypass ratio. The fan rotor`s bearings are in the front, meaning that forces resulting from a containment situation must be directed forward. In this case, the housing flange is centered on both sides (detail) and braced with long tension screws.

Figure "Containment testing in spin test rig": At first glance, this diagram shows an astonishing trial result.
A large compressor rotor blade was made to fail in a testing rig. Even though the rotor was completely enclosed by the aramdid containment ring it was supposed to test and two cover plates, at the end of the trial, the typical rolled-up leaf fragment was lying outside of the unpenetrated containment ring.
It was discovered that the edge of the elastic containment ring bulges so much when a fragment strikes it off-center, that the fragment can escape between the cover plate and ring without penetrating the ring (Fig. "Features of fiber material containment rings").
This behaviour must be considered when designing fibre-technical containment rings. The ring must be considerably wider than the potential impact area as determined by the possible trajectories of energy-rich fragments and/or have its edges stiffened around the periphery (Fig. "Suitable containment designs").

Figure "Features of fiber material containment rings": The large radial expansion of fibre-technical containment bandages is accompanied by a lateral constriction. If the bandage is not sufficiently wide or its ends not reinforced and/or affixed to prevent axial shifting, there is a risk of the fragment slipping past the side of the bandage (bottom left diagram, Fig. "Containment testing in spin test rig", Ref. 8.2-5). Several design principles that take full advantage of fibre materials` properties can be referred to in order to implement a fibre bandage as lightly, safely, and with as little expansion as possible.
In principle, the volumetric content of the bandage involved in absorbing energy should be as large as possible (Fig. "Containment technologies"). Thus, a deformation that is strictly limited to the point of impact is disadvantageous in consideration of this aspect.
A design principle that results in high energy absorption makes use of the principles of a tennis racket: the fibre layers are fixed to a stiffening frame. This frame is made up of two rings. Cross-members prevent an unallowable axial contraction and thus also large radial expansion of the bandage. Affixing the rings to the housing would be advantageous, in case the cross-members were destroyed by a fragment. The volume of the stressed fibres can be optimized through suitable layering and directing of fibres.

Figure "Turbine disk aramid containment" (Ref. 8.2-16.): This diagram depicts a containment trial on a helicopter engine. It was demonstrated that an aramdid fibre bandage is capable of stopping rotor fragments (evidently turbine) with a kinetic energy of about 7500 mkp (about 750 J).

Figure "Position of accessory equipment to containment": The large radial deformations and high accelerations in the containment region endanger mounted components, such as regulators, oil tanks, pumps, fuel lines, and gear boxes. Therefore, when an engine is designed, these components should be positioned outside the potential trajectory of fragments. Even if there is no direct contact with the deformed containment ring or damage resulting from impact of a fragment, extremely high accelerations are created. These overstress fastenings and/or destroy components, which can result in an escape of flammable substances.
The positioning of sensitive components around the circumference is also dependent on the danger presented by possible fragments escaping from parallel engines (see Fig. "Rotation direction demands position of auxiliary components").

Figure "Designs avoiding extreme fan-rotor unbalances": Due its large mass and large radius of its center of gravity, a failed fan blade creates extreme imbalances, that can overstress the aircraft`s structure. To mitigate this effect, the supporting main bearing is “hung” in the housing as elastically as possible. This mitigating effect can also be achieved through a directed break of a load-bearing cross section (hyphenation point) in the bearing region (see Example "Dealing with fan-rotor unbalances"). This enables the rotor to rotate around the new center of inertia after the blade failure. A similar principle is used in spin dryers.

Figure "Design of engine mounting bolts for controlled failing" (see also Ill. 10-12): The dynamic loads created by a rotor (fan blade) failure increase with the size of the engine, especially the fan. Imbalances and high torsion moments due to rotor deceleration are typical. These put the structure of the aircraft in danger of failing. To prevent this, connection bolts/Fuse pins of the engine or pylon mounts can be designed to fail in a controlled manner and sequence at for the aircraft structure dangerous overstresses. In these cases the aircraft is not further endangered (e.g. through destruction of hydraulic and fuel lines).
Used are hollow fuse pins which can safely take the operation loads, especially dynamic loads , but will be sheared as forced fracture by extremely high mounting loads.

Figure "Preventing by design disk damage during rotor offset" (Ref. 8.2-4): The size of potential fragments created by extreme blade rubbing due to axial rotor offset (e.g. after a shaft failure) can be controlled to a degree. If the partition wall of the intermediate stage labyrinth comes into contact with the rotor disk first (top diagram, detail), a disk failure can occur (see Fig. "Damage of rotor disc by rubbing membrane").
If the rotor can be decelerated by its blade tips rubbing against the neighboring guide vanes (“intermesh principle”, bottom diagram), only blade fragments with relatively little energy can be expected. This can be accomplished by suitably designing the following turbine blade, the root region of which is drawn forward.

Figure "Soft metal sheets slicing by rubbing solid sections": The phenomenon that thick, hard metal cross sections are sliced through by a thin sheet of soft material can be explained with help of the depicted model:
In phase 1, the thin metal plate softens much quicker than the massive cross-section due to poor heat removal and its low heat capacity. A melt is created at the contact surface in which hard oxides are formed. In phase 2, these oxides embed themselves in the soft plate and function as a grinding disc. The thin plate itself now only plays a minor role in the cutting. The waste escapes through the cut as a melt. There are similarities between this and the mode of operation of friction band saws<U>.</U>
This process is greatly influenced by material properties such as the melting point (solidus) and the heat transfer capacity of the two surfaces. This does not apply, for example, to the often relatively low solidus temperature of Ni-based cast alloys (high Al content) used as turbine disc materials as compared to Ni-based forged materials and high alloy steels, or to the relatively high heat transfer capacity of brass plating.
For example, it has been observed that a thin brass plate sliced through a hollow shaft wall made of steel several centimeters thick.

Figure "Damage of rotor disc by rubbing membrane": This example shows a turbine disc failure caused by an axial offset of the membrane of the intermediate stage labyrinth (top diagram). A situation similar to that shown in Ill. 8.2-21 occurred. The membrane came loose due to the failure of a flawed soldering to the control device. The structural design of the solder connection allowed the membrane to become offset.
The bottom diagram contains suitable measures for improving the safety of the solder connection. Additionally, a dangerous axial offset caused by the membrane detaching is prevented.