Flaws in semi-finished parts are usually also referred to as material flaws. Damage in engines that can be causally traced to material flaws is rare. However, when it does occur, experience has shown that extensive safety-relevant consequences are to be expected (Ills. 15.2-2 and 15.2-4). For example, there is a danger of uncontained fragments following cracking in rotor parts such as disks and rings (Ill. 15.2-8). Flaws in blades, especially rotor blades, can also cause extensive and extremely expensive consequential damages with sudden engine failure. In this regard, blade damages in the compressor must be viewed more critically than turbine blade damage. This is because the cross-sections and blade sizes in the compressor decrease downstream, while the opposite is true in the turbine. If a part that has been identified to have a material flaw is found in an engine, additional parallel cases can be expected. The mere fact that such a part was installed indicates an unstable manufacturing process with fundamental problems and deviations (Ill. 17-8). In particular, the sufficient detection reliability of the non-destructive testing method is called into question.
In order to provide the reader with a quick overview of part-specific damages, this chapter contains diagrammatic compositions of typical parts to explain important procedures (Ills. 15.2-1, 15.2-4, and 15.2-11). This is intended to make current problems easier to categorize. Additionally, the most important technical terms are depicted and explained. This is a prerequisite for any more intensive studies of reference works and/or consultation with specialists. There is no detailed treatment of flaws that are typical for more general machine building and the processes it uses. The general technical literature recommended in the appendix to the reference citations at the end of Chapter 15.1 can serve as a source of additional information, especially for process engineering. This chapter, on the other hand, deals with highly-stressed engine parts and the technologies they use.
Problems and their attendant risks are additionally explained with the aid of specific examples of damage.
Engine types of earlier generations used primarily rolled and forged parts. Complex parts were usually manufactured as welded constructions of these semi-finished parts (diagram on page 126.96.36.199-2). In older engine types, casting was primarily used for Al and Mg sand casting for the forward compressor housings. Housings for the accessory gearboxes (Ill. 15.2-9), regulators, and pumps (Ill. 15.2-10) are still made of light metal casts in modern engines. At first, the casting methods were not capable of producing the complex parts (diagram on page 15.1-1) and cooling air structures common today. Only in smaller engines in helicopters were integral cast parts made from high-strength steels and Co and Ni alloys found at an early stage. These were primarily bladed axial and radial disks (blisks) as well as stators in turbines and compressors. Single-crystal materials or directionally solidified materials were not yet in use. With the serial implementation of these technologies in military and civilian aircraft beginning in the early 1970s, their specific material flaws gained in importance.
The special advantages of casting technologies relative to forged and welded constructions are design freedom, lower production costs, and better creep resistance (important for hot parts). This has led to their increased use even in large engines. The pioneering engines in this field were those in fighter aircraft. These advantages were realizable not only in housings, but also in turbine bladings.
Turbine blades with complex cooling configurations, which were later directionally solidified and are today manufactured as technical single-crystals, were especially suited for realizing special properties at an acceptable cost. This type of blade has been selected as an example of a complex part, and it is used to illustrate the most frequently reported material flaws (Ills. 15.2-1 and 15.2-4).
An additional example is titanium casting. Now, it is used even for the serial production of large complex housings with bypass ducts and integrated bearing chambers (diagram on page 15.3-1).
Illustration 15.2-1: Cast parts, especially cooled turbine blades, are threatened by a multitude of potential flaws. The following deals with flaw types, their causes, and effects.
Dimensional deviations: Dimensional deviations can even originate in the production of the casting mold through flaws in the wax model or positioning problems with the ceramic core (core shifting). The filigreed cooling air structures of modern turbine blades with thin walls are highly sensitive to dimensional deviations. Even during the wax melting and casting process, the core can shift, deform, or even break. Dimensional flaws of this type can lead to local overstress through centrifugal force (rotor blades) or gas forces (stator vanes) in later operation. Differences in the wall thickness and constriction of the cooling air ducts influence operating temperatures and thermal stress. Deformation of blades and shrouds in new parts (prototypes) can be corrected through straightening within a verifiable framework (no unallowable damage; Ill. 15.3-5). This can be expected to induce residual stresses, which must not cause unallowable deformations through relaxation in later operation (Ill. 188.8.131.52-15).
High thermal stress in the part and between the part and casting mold during cooling can plastically deform part zones such as shrouds.
Other permanent deformations may be caused by overly high demolding forces, heavy blasting of thin cross-sections, and overly high clamping forces.
A special dimensional deviation is casting ligaments that form on broken cores when the melt enters into the crack in the core. These constrictions in the cooling air ducts can disrupt the cooling air flow during later operation, thereby causing overtemperatures to occur.
Cracking: Thermal stress is the most important cause of cracking in cast parts. If the cast part is locally overstressed during solidification, thermal/hot cracks (Ill. 15.1-8) can spontaneously occur. Creep cracks occur during further cooling of the already solidified cast part.
The cooling of the solidified cast raw part creates high thermal stress, especially at cross-section jumps. One cause is temperature gradients in the cast part, another is thermal strain differences between the stiff ceramic casting mold and the metallic cast part.
Common locations for crack initiation are sudden cross-section jumps. Common cracking zones are transverse struts in cooling air ducts (dimples) or the transition of a thin blade wall into a relatively thick shroud. Internal cracks and cavities can be detected with X-rays, depending on the wall thickness. Penetrant testing can be used for cracks and cavities that are open to the outside. The usual intensive abrasive blasting of the cast parts after demolding can worsen the detectability of cracks considerably. A subsequent temperature cycle can reopen the cracks and cavities.
Crack initiation, especially inside blades, can occur through a corrosive attack on the grain boundaries due to poor leaching of the cores or due to overly aggressive etching before penetrant testing.
Cracking can also be caused by dynamic fatigue resulting from the removal of the casting mold with vibrating tools or from the separation of the risers and sprues (Ill. 184.108.40.206-7).
Cold lap (cold shut): This occurs when melt fronts run into one another but do not combine. These separations are comparable with cracks with regard to their effect on the operating behavior of parts. The probable cause is a too low temperature during the casting process (Ill. 15.2-2). These flaws can be difficult to detect, depending on their location in the part and their orientation relative to the X-ray source. If the flaw is covered by casting skin, or if there is a sealing contact, penetrant testing may not be effective.
Cavities (Ill. 15.1-7): This flaw type forms during solidification through a lack of melt. Cavities often have sharp edges, branches, and are often interconnected. They can form cavity fields that have a pronounced weakening effect on the affected cross-sections. This is especially true of layered cavities (Ref. 15.2-1), sometimes confusingly referred to as layered porosity. Cavities tend to form in material accumulations and are closed at the surface by a thin skin. Material removal from the surface (chemical, abrasive, chipping) can make the cavities visible to penetrant tests. If the geometry of the part is suitable, cavities can usually be detected with a sufficiently safe degree of reliability through X-ray inspections (e.g. microfocus X-rays; Ref. 15.2-1).
Cavities have a considerable effect on static and dynamic strength. However, it has been observed that cavity fields do not necessarily have the same damaging effects of a comparably large crack (Ills. 15.3-6 and 15.3-7).
Reactions at the surface of cast parts:
Casting process: The cast surface can be damaged in areas where reactions occur between the aggresive melt and the mold or core material. If the cast material takes on components of the core material, it can later result in grain boundary damage during the leaching process. During operation, these impurities can lead to grain boundary cracks or reduced oxidation resistance. Subsequent diffusion coating can also become more difficult due to the effects on the surface.
During repressing: Cast parts are often repressed at high temperatures through HIP (especially single-crystal materials, see Ill. 15.2-4). If the argon atmosphere in the autoclave has been contaminated with evaporated carbon from the heater, there is a risk of the part surface becoming carbonized (Ref. 15.2-5). This can result in poorer oxidation resistance.
Poor grain orientation and grain boundary alignment. Under transverse creep loads, grain boundaries are the life span-determining weak point. In addition, the lattice orientation of the grain determines important characteristics such as the modulus of elasticity. A high modulus of elasticity in a certain crystal direction results in correspondingly high thermal stress. This means that the thermal fatigue behavior worsens depending on the direction (Volume 3, Ill. 12.6.2-7). If poor solidification conditions (direction and size of temperature gradients, solidification rate) create unfavorable grain, it can have an unallowably negative effect on the operating behavior of the part. Typical examples are columnal crystals that run from the inlet edge into the blade in the transition zone to the shroud.
Segregations: The accumulation or depletion of alloy components is a greater problem for single crystal materials (Ill. 15.2-4), which have a considerably slower solidification rate, than it is for conventional (equiaxed) cast materials. However, segregations can be observed in larger conventionally-cast parts with material accumulations, such as integral turbine disks and turbine stators. These can cause local melting during high-temperature heat treatments, such as high-temperature soldering or solution annealing. Depending on the local alloy composition, long-term exposure to temperatures during operation can promote the development of brittle phases (e.g. sigma phase).
Blocking of cooling air ducts with foreign material: Non-metallic, ceramic foreign objects such as core remnants and blasting residue are difficult to detect with X-rays. The reason for this is their greater permeability for X-rays, relative to nickel alloys. A more successful option might be through-flow measuremet, if the constriction of the cross-section is sufficiently pronounced. Reactive blasting residue (e.g. SiC) can have a damaging effect at high operating temperatures through diffusion, melting, and increased oxidation.
Oxide skins and ceramic inclusions: These act like cracks. Unlike ceramic inclusions, oxide skins are difficult to detect with X-rays (see Ills. 15.2-4 and 17.3.1-4).
Example 15.2-1 (Ref. 15.2-3):
Excerpt: “…(the OEM) is conducting eddy current tests on all fourth -stage turbine blades on the …(small fan engine) following three blade failures traced to metal flaws.
The first blade failure in the…engine occurred in February while the engine was undergoing ground testing in a test cell. At that time,…(the OEM) determined that the fourth-stage blade failure could have been traced to some blades that had been damaged during handling, and the failed blade had not been replaced. The containment ring on the turbine failed to hold the blades within the engine, and the failed blade was not found.
(The OEM) discovered that the weld on one joint of the engine nozzle was ineffective and quickly incorporated design fixes to that area after inspecting the containment rings of other…engines.
Earlier in May, a fourth-stage blade on a (similar) engine installed in …(a business jet) had failed. The single blade left the wheel while being contained within the engine. This incident was followed mid-May by an engine failure during takeoff…The containment ring in this latest engine failure was also able to limit the damage within the engine.
(The OEM) officials found that all three failed engines had less than 100 hrs of total time. One engine had accumulated only 7 hrs. and another 20 hrs of operation. The third engine had 62 hrs of operation…
The failures have been traced to what…(the OEM) officials believe is a material flaw in the original casting of the turbine blades.
`What we pretty well have determined is that as the molten metal comes into the blade casting from both ends, if you get a cooling, there is the possibility that the metal will not solidify together and you get a cold shut, or a casting defect.'“
Comments: This example is in several ways typical for a material flaw and its effects.
Illustration 15.2-2 (Example 15.2-1, Ref. 15.2-3): This case also shows the large damage potential of material flaws (see Ill. 15.2-4). It evidently concerns a cold lap that led to the fracture of turbine rotor blades of the last stage. A cold lap is a laminar separation created by the contact between two overly cool melt flows (bottom right diagram). The contacting surfaces are no longer able to bond. These separations tend to occur in thin or hollow blades. In these cases, a relatively small, insufficiently heated mass flow in these thin cross-sections is cooled too quickly by a cold casting mold.
Illustration 15.2-3 (Ref. 15.2-5): Directionally solidified Ni alloys have anisotropic strength. This is especially true of the modulus of elasticity (Volume 3, Ill. 220.127.116.11-7). The direction of solidification in turbine blades is usually aligned with the blade stacking axis. The crystal orientation <100> with the lowest E modulus is desired. The right diagram shows the influence of the load angle “a” (top right diagram) at room temperature and at 760 °C. One can see that the E modulus of the conventional casting (equiaxed) is the same as that of a directionally solidified material with an “a “of 30°. The stiffnesses typically decrease with temperature for all grain orientations.
Lower stiffnesss decreases the stress that occurs due to restricted strain (strain-controlled process). With regard to thermal strain, this leads to better LCF and thermal fatigue behavior (Ill. 15.2-5). This characteristic, combined with the grain boundaries parallel to the main load, provides better operating behavior compared with conventional casting. It is debatable which of the two positive effects is predominant in directionally solidified materials. The left diagram shows the influence of the grain boundary direction relative to the main stress under cyclical loads in the LCF range (thermal fatigue). The number of thermal cycles to crack initiation decreases exponentially from a very small angle a in the range of a= 0° to a= 30°. At 30° the E modulus of the directionally solidified material is the same as that of the conventionally solidified material. Therefore, from the perspective of LCF strength, at this point there is no longer an advantage to directional solidification.
Specimen tests evidently revealed that LCF cracks did not originate in the grain boundaries, nor did they travel along them. This indicates that improvements in LCF behavior can be attributed
to the low E modulus. Therefore, the favorable crystal orientation contributes to this effect. This means that:
The direction of the crystal orientation must be specified. Even minor deviations can decisively worsen the cyclical operating properties (especially thermal fatigue) and must be viewed as flaws. Grain boundaries that are aligned flat across the load direction seem to be less significant in this case. On the other hand, the local grain boundary alignment in conventional casting can serve as an indicator for the LCF properties in the area in question. This is true, for example, of columnal crystals in critical part zones (Volume 3, Ill. 12.6.2-7).
If there is primarily creep stress (high proportion of static loads), one can assume a dominant influence of the grain boundaries. In such a force-controlled case (e.g. under centrifugal force), the stress depends on the uniformly aligned elasticity of the grains. The influence of the grain boundaries can be seen in the typical failure of conventionally cast materials with creep pore formation on the grain boundaries (Volume 3, Ills. 12.5-5 and 12.5-7).
Illustration 15.2-4 (Refs. 15.2-5, 15.2-6,15.2-7 and 15.2-10): Single crystal parts (SC) are not comparable with the highly pure flawlessly constructed crystals used in semiconductors. They have typical weaknesses and structural characteristics that also occur within the grain of multi-crystal materials. They are merely missing the grain boundaries. For this reason, the term “monograin” might be more accurate (Ref. 15.2-6). The mechanism and damage symptoms of many casting flaws, such as dimensional deviations and cracking, correspond to conventional casting (Ill. 15.2-1). For this reason, they are not dealt with here. Single crystals can also have casting flaws such as cavities, non-metallic inclusions, and undesirable growth directions, which are also known in directionally solidified (DS) materials (Ref. 15.2-5). In the following, the flaws are classified into five main groups based on Ref. 15.2-10:
If the E modulus that acts lengthwise along the part changes, it can have an undesirable effect on the natural bending frequency and the thermal fatigue behavior of the part.
Cyclical crack growth is also influenced in its direction and speed by the crystal planes (Ref. 15.2-11).
Depending on the grain orientation relative to the load direction, the LCF strength, i.e. thermal fatigue, should also be affected by weakened grain boundaries and a higher E modulus.
The tendency for microcavity development is alloy-specific and depends on the solidification conditions. The relatively slow directional solidification promotes a special type of cavity formation, called inter-dentritic porosity. These flaws concentrate on upper horizontal cast surfaces such as shrouds (Detail “4”, Ref. 15.2-8). This type of cavity formation is promoted by a diagonal solidification front that acts to hinder the required melt supply (“4”). Locating these weak points in a less highly stressed part zone can make them acceptable.
The microcavities form between the fir-tree structures (dendrites, Detail “1”) when the solidification front reaches the casting mold and the remaining melt can no longer flow in. Therefore, it is understandable that the frequency of microcavities is related to the dendrite spacing, i.e. the solidification rate. The lower the solidification rate is, due to a small lowering rate and a large oven gradient, the less microcavities are to be expected.
The curved shape and closed development complicate the usual serially usable non-destructive testing methods such as X-rays and penetrant testing. This leaves only prevention through careful process monitoring and the use of ceramic filters at the ingate for the melt.
Dendrite development (Detail “2”): During the usual solidification process of single crystals, the crystal and the dendrites orient themselves in the  lattice plane parallel to the part (Detail “3”). Crystal growth occurs over the dendrites (fir tree structure). A structural characteristic is the spacing of the dendrite trunks, which is related to the solidification rate. The crystal orientation remains the same even if the size of the heat flow is changed.
Alloy components are incorporated into the solidification front between the dendrites at different rates. This leads to changes in the composition of the remaining melt between the dendrites, until it suddenly solidifies as a residual eutectic. The residual eutectic is enriched with elements that are required for development of the g'-phase in other areas. This means that the strength of the single crystal is related to the dendrite spacing. The amount of residual eutectic is alloy-specific.
Contrary to the situation in multi-crystalline solidified parts, depending on the specific solidification, the g'-phase in single crystals is “as cast”. In order to achieve optimal distribution and refinement of the g'-phase through later heat treatment, it must be as completely dissolved as possible during solution annealing. This requires “dangerous” proximity to the melting point (solidus temperature, Ill. 15.2-12) with subsequent rapid cooling (e.g. > 150°C/min, Ill. 15.2-7). The solution annealing occurs at very high temperatures above 1270°C and evens out microsegregations between the dendrites as well as dissolving the undesirable residual eutectic. The high annealing temperature is only about 10°C from the solidus temperature. It relies on the missing elements that harden grain boundaries in conventional alloys (Ref. 15.2-11). The extremely high annealing temperature causes almost all g'-phases to dissolve. However, it contains the risk that even minor temperature deviations can result in melting with porosity and undesirable structures.
The rapid cooling allows optimal hardening and minimizes the risk of the d-phase developing in operation (Ill. 15.2-6). Subsequently, a heat treatment is conducted in stages at appropriate temperatures (e.g. 4h 1080°C, 16h 870°C). This gives the g'-phase a structure for optimal single-crystal strength properties.
Illustration 15.2-5 (Ref. 15.2-12): With regard to the influence of the grain orientation on dynamic fatigue strength (HCF, LCF), single crystals behave similar to directionally solidified structures (Ill. 15.2-3). Fundamentally, due to the better-controlled casting conditions, single-crystals have smaller defects and a lower probability of flaws than other cast structures (Ref. 15.2-7). This explains their greater dynamic fatigue strength. Under strain-controlled cyclical loads (e.g. thermal fatigue, see Volume 3, Ill. 12.6.2-2) the lower E-modulus in the direction of solidification (-crystal direction) has a stress-reducing effect. This results in a longer cyclical life span, i.e. higher LCF strength compared to other crystal orientations (top diagram).
Under force-controlled loads, such as those placed on blades by centrifugal force (bottom diagram), the increased resiliency from the influence of the E-modulus does not come into effect due to the unlimited strain (Ref. 15.2-11). This behavior can be seen in the single scatter band that is valid for all the various crystal orientations.
Illustration 15.2-6 (Ref. 15.2-5): Intermetallic phases such as s-, d-, m- and Laves can considerably shorten the creep life span of cast and forged nickel alloys (top diagram). These typically needle-like and brittle phases are usually precipitated during longer operating times with increasing temperatures. Therefore, low creep loads, as a prerequisite for long life spans, promote phase development.
The PHACOMP method can be used to estimate whether an alloy will tend to form undesired phases (topological close packed = TCP). This approach takes into account the electron gaps in the atomic lattice. They are specific to every alloy element, and can be gleaned from tables (Ref. 15.2-31). Because this can involve a considerable calculation effort, and requires technical expertise, there are companies that offer to take on the task of performing these calculations.
The bottom diagram shows the effects of the d-phase on the creep life of a single crystal material. A comparable effect can also be observed in the conventionally cast standard blade material IN100. The tendency of the single crystal to form the d -phase during operation increases with decreasing cooling rates during heat treatment after casting (Ill. 15.2-4).
Illustration 15.2-7 (Ref. 15.2-5): Slower cooling rates from solution annealing lead to a less favorable, coarser g' development during the simultaneously occurring hardening process, especially in single-crystal materials(Ill. 15.2-4). This lowers the creep life span. In the left diagram, the creep life span of the single crystal with coarser segregations corresponds to the normal (dotted) -3s line, which indicates scatter. The normal fine segregations correspond to the solid line (best fit). The space between the lines represents a life span difference of roughly 20% at 138 MPa in the range of 900-1000°C.
The right diagram shows the hot cracking strength (short-term strength). In the range of 900-1000°C (gray area), one can see a decrease in material strength of over 10% with the coarser g' phase (circles).
It should be noted that the strength difference has an overproportional effect on creep life.
Illustration 15.2-8 (Example 15.2-2, Ref. 15.2-4): Usually, flaws that were not recognized in the production process are only noticed when operating damages occur after longer run times. By this point, a large number of delivered engines can be expected to be affected. This makes the cost of risk minimization and solutions very high. In order to control this type of problem as quickly as possible with an acceptable amount of risk, it is necessary to be able to trace the origin of the parts. This may make it possible to limit actions to specific charges that are connected with the damage cause. Additionally, a decisive factor is a damage symptom that can be externally recognized sufficiently early and can be identified and/or controlled on location. This demonstrates the importance of sufficiently thorough marking of parts (Chapter 17.4).
A critical problem is with unpackaged or loose parts with minimal markings on some repaired parts, which makes them very difficult to trace. In these cases, it may be impossible to determine the earlier operating conditions and times of the parts.
Example 15.2-2 (Ill. 15.2-7, Ref. 15.2-4):
Excerpt:”…The Third …(fighter) loss attributable to engine failure occurred in February of this year. Analysis indicates that a blade in the fourth stage of the powerplant's low-pressure turbine failed because of a “casting anomaly” during the manufacturing process. There are two sources for the blades, in this case, the failed blade came from…(identified source). As a result of this latest loss, the…(customer air force) grounded the entire fleet of…(about 160 fighters with the failure prone engine types).
…a series of inspections- and blade replacements- will be performed to return those aircraft to operational status. If all goes well, the first of the affected fighters should return to the air near the end of April…
…(the OEM's) recovery plan calls for on-wing inspection of the fourth-stage low-pressure turbine in affected engines. If the blades show signs of a manufacturing problem, engines will be removed and new blades installed.
…(The customer) plans in addition to inspecting and replacing to go one step further. It plans to reblade some engines, apparently using a time criterium.
The blade manufacturing problem is not isolated to…(one customer), however, a number of the suspect blades are also in use on (another fighter type) throughout the world….(The …(OEM) reckons that there are more than 700…(engines) in service, and it has not yet determined exactly how many are affected by the blade problem. The company does know, however, the number and location of suspected blades…“
Comments: An inspection of the installed engine is probably done visually with a boroscope or optically by looking into the exhaust duct. In order to find the “casting anomaly” in time, the damage mechanism must be easily observable even in its early stages. This indicates that the damage is in the blade, but not the root. The “time criteium” indicates sufficiently slow damage development. One can speculate that this does not concern an inclusion, crack, or large cavity field. These types of flaws usually result in uncontrollable crack growth with spontaneous failure. Because the “anomaly” is evidently limited to one production batch, the problem is probably not significantly affected by the configuration. In the fourth turbine rotor stage, no life-shortening thermal fatigue is expected. Creep deformation and/or creep cracking, originating in a local weak point, on the other hand, is entirely plausible even in the fourth turbine stage. This type of damage could be detected in time, even through visual inspection, albeit not with complete certainty. These considerations indicate a damage mechanism that shortened creep life considerably (Ills. 15.2-8 and 15.2-9).
Illustration 15.2-9: Especially in older engine types, many parts were made from Al and Mg sand casting. Examples include:
These cast parts were prone to the typical flaws of this casting technology (Ref. 15.2-13). A certain amount of cavity formation or flaws originating in fragments of the casting skins from the casting flow were unavoidable. The usable dynamic strength of these materials was minimal, even without considering the influence of their structures and the missing fatigue strength. The most important flaws that occurred during operation are:
Shrinkage cavities that locally spread through the entire cross-section (top left detail) are a typical problem in larger sand cast parts, especially gearbox housings. These leakages can be sealed through infiltration with organic (synthetic resins) and inorganic (sodium silicate) media in the new part. This sealing effect can be reinforced through shot peening of the surface (e.g. Al shot). Experience has shown that multiple overhauls carry the risk of reopening cavities and causing leakages. If oil has entered the cavities, it makes resealing difficult and can end the life of the part.
Larger oxide skins (bottom left detail): These represent a crack-like separation and are not easy to detect with X-rays and penetrant testing. In some cases, their size can cause spontaneous part fractures even after short operating times (Ill. 15.2-10).
Bubbles and gas pores (right detail): These are gas inclusions that either originate in the casting process or are related to common subsequent reparatory welding of new parts in the area around cavity fields. These flaws should be sufficiently reliably detectable with X-rays.
Illustration 15.2-10: During casting processes in air (Ill. 15.2-9), such as those used for Al and Mg sand casting, there are typical flaws which occur (Ref. 15.2-13). The depicted part is a fuel pump housing made from cast magnesium. A leak occurred after a short operating period. Destructive testing revealed a shimmering, darkly colored separation several centimeters long that ended very close to the surface. Evidently, the fracture of a large section of the part wall was imminent, and would have resulted in a flight accident. The flaw was a casting-dependent material separation that was classified as an oxide skin. Oxide skins act as material separations and have a crack-like effect on strength and sealing properties (top right detail). These high-melting oxide skins form when air contacts the outside of the melt stream as the casting mold is filled. They are carried into the cast part or form on rapidly solidifying surfaces in connection with cold lap.
Shrinkage cavities can often locally spread through the entire cross-section (top left detail) and can influence strength and sealing properties (top left detail).