Aeroengine Safety

Figure "Types of hollow material flaws": Experience has shown that persons who are not specialists in the field of materials science often confuse and misuse the terms cavity and pore, which refer to typical hollow spaces in cast parts.

Cavities, shrinkage cavities, or shrinkage porosity (left diagram) are hollow spaces caused by a lack of residual melt. They form between so-called dendrites when the melt solidifies. This formation process, combined with their position in the structure, makes their jagged shape understandable. Often, several cavities create a cavity field, in which the individual cavities can be connected through tight, crack-like interdendritic openings. Because cavities only form near the end of a local solidification process, it is understandable that they are often sealed off from the surface by a “casting skin”, which makes it difficult to detect them using penetrant inspection (Fig. "Limiting influences non-destructive testing"). In order to open pores, parts are etched. Abrasive blasting of cast parts does not guarantee that the cavities will be opened. It can even result in blockage and compression of small connections to the surface (Fig. "Limiting influences non-destructive testing"). In addition, a rough blasting surface can make flaw detection even more difficult. Thermal cycles have been shown to be effective in opening cavities near the surface (Fig. "Opening of cracks before penetrant testing"). In order to discover internal cavities, an additional X-ray inspection is conducted (Fig. "Limiting influences non-destructive testing"). Even very small cavities can be found with micro-focus X-rays (Ref. 15.1-1).
However, cavities may also open during operation due to thermally and/or mechanically induced strain. Therefore, it is not rare for cavities to be discovered only during operation.
Larger cavity fields can spread through thick cross-sections. In walls separating areas of different pressure, this can lead to a loss of seal effectiveness. This results, for example, in the typical
secretion of oil from gearboxes made of cast light metal. For this reason, gearbox housings are sealed with the aid of infiltration using hardening substances (synthetic resin, sodium silicate). A similar effect can be attained through shot peening with large aluminum pellets. These two procedures can also be combined.
Cavities can primarily be expected in thick cross-sections. This makes them especially dangerous as starting points for LCF cracks in the highly-stressed hubs of cast turbine disks in smaller engines and in the fir-tree roots of larger turbine blades.
Cavities can also form in thin cross-sections such as the walls of cooled turbine blades. This refers especially to so-called layer cavities. In this case, cavities have formed in a single plane and weaken the affected cross-section considerably (Ref. 15.1-1, Fig. "After-treatment flaws of casted turbine blades"). Cavities also often form at abrupt cross-section transitions. Typical examples are cavity formation at the transition of pedestals or cross-braces into the walls of cooled turbine blades (Fig. "After-treatment flaws of casted turbine blades").
Closed cavities can be eliminated using an HIP procedure (Fig. "HIP of cast parts"). Metallically smooth cavity surfaces fuse together and eliminate the flaw. The missing material in the cavity may appear as a dent on the surface of the part.
If, during a forging process following casting, cavities are re-fused and/or not closed completely, for example due to insufficient deformation, it is possible that cavities will be present in forged parts.
In the interest of completeness, cavity formation in weld seams should also be mentioned. Their formation corresponds to that in cast parts. Chapter 15.3 deals with the influence of cavities on part strength, especially under alternating thermal-mechanical loads (Ref. 15.1-18).
Gas pores in cast parts and welds are usually spherical hollow spaces resembling small bubbles (right diagram). They are caused by gas formation in the melt. Because they float, they collect in the upper region of the part or at surfaces that are closed towards the top. Gas pores usually have no connection to the surface, and detection therefore requires X-ray inspections. Pore walls are usually not oxidized, and can therefore be eliminated using an HIP process.
Even in cast parts, gas pores can be found in rare cases (e.g. in titanium alloys, Fig. "Pores causing cracks in titanium alloys"). They were created during casting and were not forged out. These pores are often related to hydrogen.
The same is true for many pores in arc-welding processes that rely on vaporization of the metal. The metal vapor can become trapped before sublimation and create gas pores. Affected processes include electron beam welding (Fig. "EB welding flaws"). The vaporization process is promoted by the surrounding vacuum.
Another form of pore formation is microscopic pores on grain boundaries (steel) in connection with hydrogen embrittlement. These pores are formed in the forged part (Volume 1, Ill. 5.4.4.1-2 ).
Pore formation has also been reported in PM materials. If a periodically leaking capsule allows the presure gas (usually argon) to penetrate into the still-porous part, and the gas remains trapped after densification, pores can form during and after the cooling process due to the high pressure.
In the interest of completeness, the formation of minute pores (creep pores) on the grain boundaries of hot parts should be mentioned (Volume 3, Ill. 12.5-7). Formation of these pores is especially pronounced in forged parts during the creep process during operation. Closed pores, i.e. those with no cracks to the surface (no oxidation, pressure difference can build up) can be eliminated in their early stages with the aid of a regenerative HIP treatment (also see Fig. "Safety of parts by manufacturing").

Figure "Mechanisms of hot cracking": In addition to operating influences (top center diagram), such as rubbing (Volume 2, Ill. 7.2.2-9.2), which are primarily associated with rapid local overheating, there are many procedures and finishing steps that can cause thermal or hot cracking:

• Casting (top right diagram)
• Forging (Fig. "Flaws in forged rotor disks")
• Heat treatment (example in explanation of Fig. "Failed detection of large cracks")
• Welding (top right diagram, Fig. "Hot cracks by heat treatment")
• Machining, i.e. grinding and separating (Ill. 16.2.1.1-2).

In these cases, the sufficiently heated material is primarily torn open at the softened or molten grain boundaries (middle diagrams) by residual stresses (thermal stress).
A combination of multiple factors promotes hot cracking (bottom chart). The following observations are limited to the production and shaping process of blank parts, i.e. casting:

Design: The design has a considerable effect on hot cracking in cast parts.
This is true for the trend towards highly complex, large integral cast parts such as turbine stators in segments or as whole parts. Sharp cross-section transitions combined with high stiffness, stress concentrations (sharp corners), and differences in heat dissipation promote hot cracking. A typical example is integral turbine stators (Fig. "Hot cracking sensitivity").

Metallurgical influence (see Fig. "Hip process problems"): There are, of course, alloy-specific differences in the tendency to hot cracking. Therefore, realization of a complex cast part is not guaranteed simply because it was successful with another material (with higher ductility, for example). Seemingly minor differences in the solidification process, shrinkage, a tendency to segregation (enrichment or depletion of alloy components), and small amounts of damaging contaminants (e.g. bismuth, Ref. 15.1-7, Fig. "Heattreatment batch as finishing problem indication") can have grave effects.

Casting conditions: A great deal of experience and the best casting equipment are prerequisites for maintaining deadlines and serial prices when producing the complex cast parts used in turbine engines. The ingate with the position and design of the sprue and riser, preheating and cooling of the casting mold, the casting rate, and the casting atmosphere must all be optimized in relation to one another.

Molding materials: In this case, the ceramic casting molds naturally play a decisive role during solidification and cooling due to their influence on the heat balance and stresses on the shrinking part (Fig. "After-treatment flaws of casted turbine blades").

Figure "Hot cracking sensitivity": Hot cracking is a major problem in integral turbine stators with diameters up to 50 cm, such as are used in smaller engines (Fig. "Mechanisms of hot cracking"). These parts combine several hot crack-promoting properties:

• Principle-conditional restricted thermal strain causes high thermal stress.

• Cross-section jumps and material buildup at the transition points of blades and shrouds (part diagram and detail at top right).

• Notches at the edges of the transitions.

• Disadvantageous grain orientation in the at-risk zones (bottom right detail).

These can lead to high reject rates and delivery problems, and even prevent realization of serial production. If a foundry has succeeded in delivering such a serial part after extensive development work (experimental casts), experience has shown that there is a high risk that this elaborate effort will not be considered worthwhile by other suppliers, making double-sourcing impossible. If the sole remaining supplier then drops out, the entire project will be delayed and perhaps even threatened.

It is interesting and important to note that hot crack-sensitive areas during blank part production are also susceptible to cracking during operation. However, these are usually thermal fatigue cracks (Volume 3, Ill. 12.6.2-4). Therefore, hot cracks in blank parts should direct the design engineer to optimize sufficiently early, especially when he or she considers the expected operating behavior.

Figure "Precision casting flaws": Many parts of turbine engines are produced as precision casts using the principles of the lost-wax technique. These are primarily hot parts made from superalloys, which are usually Ni-based, but sometimes have a Co basis. In general, it can be said that the more complex and integral a part is, the greater the expected casting problems, technological development costs, and reject rates.
Typical parts include turbine blades (Fig. "After-treatment flaws of casted turbine blades") in the form of single parts or segments, or even integral (one-piece) stators. Bladed turbine disks of smaller gas turbines are cast in one piece (blisks). Other parts are whole housings or housing parts. This type of precision cast part is found in the compressor and turbine. In small gas turbines, these are cast blisks made from steel, while in large gas turbines they are single blades or whole compressor stators. In these cases, dynamic strength is of primary importance. This makes the replacement of forged parts with cast parts even more difficult. Of course, titanium alloys, for compressor housings, for example, are also produced using similar casting methods. Because complex geometric surfaces such as blades attain the shape of the final part during the casting process, i.e. do not undergo significant machining, weaknesses and undetected flaws from the casting process are more likely to remain in the final part than would be the case in forged parts that are later shaped through chip removal, for example. These flaws determine operating safety to a large degree. Therefore, prerequisites for preventing or detecting casting flaws are understanding their formation, causes, distribution in the part (diagrams on left side), and the influence they have on the operating characteristics (diagrams on right side). In the following, the example of a cooled turbine rotor blade is used to explain the typical problems and flaws, and their locations (labeled A,B,C,D), that occur in each stage of the casting process:

“1” and “2” Wax model with ceramic core: Because deviations of outer masses and geometries are easier to detect, there is a greater risk from internal problems. The ceramic cores, which are often very filigreed and fragile, can be cracked, broken, or shifted from their required position. This can later be reflected in the internals, e.g. the cooling configuration, and cause unallowable changes (Fig. "After-treatment flaws of casted turbine blades"). Even if these flaws can be found with modern inspection methods, they can result in unallowable reject rates. This is especially true for single-crystal casting with high stress on the cores (Fig. "Single-crystal casting-flaws").

“3” and “4” casting mold: The ceramic mold for the pour determines the outer shape of the parts. Its stiffness and thermotechnical data are very important for the cast part. Through doping of the inner wall (nucleation), the grain growth on the surface of the cast part can be influenced. The external part geometry is easily inspectable, so parts with unallowable geometries can be rejected. Certain warping within specified limits can be corrected through trueing (e.g. blade twisting, Ill. 15.3-5; also see Point “11” and Fig. "Straighrening bent parts").

“5” Casting process: Ceramic particles from filters, cores, or the inner casting mold surface can be carried by the molten material. These impurities can weaken thin cross-sections (e.g. the wall of cooled turbine blades, Fig. "After-treatment flaws of casted turbine blades") considerably. However, these weak points should be sufficiently safely detectable with X-rays since, unlike cooling duct blockages, the low absorption of X-rays is an advantage. Contamination of the melt through tiny amounts of damaging metals such as bismuth (Fig. "Heattreatment batch as finishing problem indication") and lead, on the other hand, are far more difficult to detect and can make entire charges unusable. This is especially unpleasant when the damage is only recognized during later operation due to premature failure.

The structural properties, especially the grain size, grain boundary orientation, grain shape (globular, columnal), and grain structure (dentrites, phases, carbides), are determined by the temporal and spacial progression of the temperature gradients during solidification. Controlled cooling rates and temperature gradients can be used to produce directionally solidified structures (grain boundaries parallel to the life-determining loads) and single crystals (Fig. "Single-crystal casting-flaws"). This optimizes the operating properties of the part, including thermal fatigue and creep strength. Unlike the strengthening effect attained with phases and carbides, subsequent alteration of the grain size and geometry through heat treatments is not possible without certain drawbacks. If columnal crystals form in part zones under high thermal stress, such as the transition of blade corners into the shroud, the expected life span of the part will be shortened considerably.
During solidification of the part, cavities may form in certain cross-sections (Fig. "Types of hollow material flaws"). As long as these flaws are open to the surface or are sufficiently voluminous, penetrant testing and X-rays are sufficient for their safe detection. In comparison, small collections of cavities at cross-section jumps, such as at the transition of nubs into the wall of a cooling air duct or at layer porosity (Fig. "After-treatment flaws of casted turbine blades"), are difficult to detect and can require elaborate actions to be taken due to the risk of additional parts being affected during operation. If merging melt flows have already cooled down too far, they do not combine, and instead form a difficult-to-detect cold weld.
When the cast part cools, the shrinkage within the part and relative to the ceramic casting mold creates powerful thermal stresses. These can cause hot cracks (Fig. "Hot cracking sensitivity"). If sufficiently plastically deformed zones are created in a single-crystal material, then it can be expected that, at higher temperatures during the cooling phase, “recrystallization” with formation of small “flaw grains” will occur. These worsen the operating behavior and increase the risk of cracking during later machining (e.g. grinding blade roots).

“6” Demolding: Although the casting mold will have cracks after cooling, it usually requires considerable mechanical forces to remove it from the part. Removing the ceramic casting mold with vibrating tools (Ill. 16.2.2.5-7) can cause undetectable pre-damage due to dynamic fatigue and/or dynamic fatigue cracking.

“7” Separating: It is plausible for a similar problem to occur when separating the cast parts from the cluster. The excitement of vibrations should be seen in connection with poor clamping of the parts and a stick-slip effect of the cutting disk.

“8” Removing the cores (leaching): This is done with the aid of aggressive media due to the high chemical stability of the ceramic cores. Deviations of the procedural parameters can cause the base material to be damaged. Grain boundary attack is especially dangerous and difficult to detect. Cores that are not removed completely (core remnants) can constrict cooling air ducts and cause overheating and extreme life span reductions during later operation. The typical core material Al₂O₃ absorbs so few X-rays that it is very difficult to detect core remnants that are blocking cooling air ducts (Ref. 15.1-1).

“9” Abrasive blasting: In order to remove outer mold remnants, and to provide an optimal outer appearance, cast parts are blasted with ceramic particles, and with steel particles in extreme cases. The danger with this process is that flaws can be covered and cracks pressed shut, making subsequent penetrant inspection difficult (Fig. "Limiting influences non-destructive testing"). Remnants of blasting media in the blades should be preventable. If this does occur, however, it can cause problems during later finishing steps such as diffusion-coating (Fig. "Oxidation protecting Al diffusion coating problems"). Superalloys may react with blasting media (SiC, etc.). This danger is present in subsequent finishing processes with high temperatures, such as heat treatment, diffusion coating, and high-temperature soldering. Gradual damaging can also occur at high operating temperatures.

“10” Etching: in many cases, etching with aggressive media occurs before and/or after abrasive blasting. This makes flaws visible and/or opens cracks and splits. Macro-structures are developed and can then be evaluated (grain size, grain geometry, see “5 Casting process”). In this case, as well, it must be ensured that no unallowable corrosion occurs, especially in poorly controllable inner spaces, and that no remnants of the etching media remain, which could have a damaging effect later (Fig. "Damages by not approved processing baths").

“11” Heat treatment: heat-treatment of the cast parts serves to optimize the form and distribution of phases (g'-phases and carbides) that determine operating strength. The required annealing temperatures are very high, up to the range of the solidus temperature. If production steps such as blasting or straightening have created sufficiently plastically deformed zones (critical deformation) in single-crystal parts, high annealing temperatures can lead to recrystallization (Fig. "Single-crystal casting-flaws"). This causes new grains to form, which will act as weak points in subsequent finishing steps and during operation.

Figure "Flaws in ingots and forgings": Flaws and anomalies in forged parts can often be traced back to the ingot (Ills. 15.2-21 and 15.2-22). If the head piece that is cut off is too small (top diagram), residual flaws can enter into the forged part. This means that the forging process may no longer be capable of eliminating the flaw through forging (Ill. 15.1-37). Typical anomalies and flaws that can affect
forged parts are cavities (Fig. "Material separations in forgings"), gas pores, segregations, and structural anomalies such as columnar crystals. In these cases, the local degree of forging plays an important role (Fig. "Forging process caused characteristics")
Another common melting process is ESR (electroslag remelting; not shown). In this case, the electric arc is located in a fluid slag layer that floats on the bath.
Melting flaws (Refs. 15.1-4, 15.1-12, and 15.1-13; Fig. "White spots in Ni alloys") are formed in both the melting process in the (cast) ingot or in remelted billets ( or casting billets). These are later made into forging billets, which in turn become the reforged semi-finished parts (pancakes; Fig. "Terms of part production stages").
In high-alloy steels (e.g. A286) and superalloys (e.g. Waspaloy, IN718), flaws originating in the melting process are related to segregations, i.e. zones with enriched or depleted alloy components (Fig. "White spots in Ni alloys"). Segregations that are especially strength-reducing and therefore dangerous consist of oxides, carbides, nitrides, and carbonitrides. Forged parts made of titanium experience similar steps in the production process. However, the potential impurities and problems are different than those described here, and will therefore be treated separately in the context of documented damage cases (Ills. 15.2-13 to 15.2-20). In vacuum induction furnaces (“1”), the alloy is melted using the prematerial. The quality/origin of this prematerial can already have an effect on the later forged part. For example, if the prematerial is recycled material from a machining process, incompletely melted impurities from fragments of hard metal blades from machining tools may later be present in the cast part. This is also true for oxides from the prematerial, foreign materials such as blasting residue or fragments of ceramic molds. Bars are cast from the induction furnace into ingot molds. Their head (“H”) and foot (“F”) position is important for the location of flaws in the later forged parts (Fig. "Formation of forging flaws"). For this reason, the position is documented in a traceable manner for a prescribed period of time. In order to minimize the number of impurities in the following remelting process, oxides and any ingot mold residue is ground off (“2”). Here, there is a potential danger that fragments of the grinding disk stuck in the surface of the melting electrode can appear as segregations in the semi-finished parts.
In the following electric arc remelting process (“3”), the inverted bar (headfirst) is melted in argon or in a vacuum (Ill 15.2-17). It drops into the cooled mold below it, and reforms as a remelted bar. This remelting process can be repeated several times, and it can be assumed that the risk of segregations will decrease with more remelts. Therefore, this process is an important quality criterion for the semi-finished parts of forged rotor components (up to 3 remelts), and is required in the specifications. If forged raw parts with a deviating remelting process are used, an extremely extensive new rating with comprehensive cyclical overspeed tests may be necessary.
Especially important for the development of dangerous segregations is the area of the electric arc with the dripping upper electrode and the melt bath on the lower electrode. Light, unmelted impurities collect and float on top of the melt bath. Certain processes can lead to impurities being absorbed by the remelted bar. These processes are more closely dealt with in Fig. "White spots in Ni alloys").
After the remelting process, damaging characteristics can be expected in the head area, which solidifies last. Typical characteristics include a large central cavity, a special arrangement of the segregations from alloy components, and concentrations of impurities (Fig. "Terms of part production stages"). It is decisive for the quality of the later forged parts that a sufficiently large head section is cut off and not forged. This understandably affects costs. For this reason, unnecessarily large head sections are not cut off, if possible. Experience has shown that flaws from the area near the head section are found in the finished parts (Fig. "Segregation induced turbine disk failure"). In this case, the prescribed documentation of the remelting process must be consulted. This may enable one to identify additional parts that were located “below” the damaged part.

Figure "White spots in Ni alloys" (Refs. 15.1-4 and 15.1-14): Explanation of the melting process: The two most common melting processes are VAR (vacuum arc remelting) and ESR (electroslag remelting). Both processes use an electric arc for remelting. The difference between them is the medium around the arc. While VAR uses a vacuum, ESR uses a slag bath floating on the melt. In order to understand the types of flaws and their development during remelting, the following text examines the VAR process more closely. These flaws can also appear later in the forged part. The bottom diagram is a schematic depiction of a cross-section through the area of the melting process. Between the melting electrode above and the ingot that forms through solidification below, the electric arc that carries the melting energy “burns” in a vacuum. Various precipitated non-metallic impurities, such as oxides, nitrides, and carbides, float on top of the slag bath. Splashes from the bath can stick to the cooled ingot mold wall and later fall off. The top electrode forms a ridge-like edge (torus) that can break off and fall into the bath. The ingot has a crown around the surface level of the melt bath. Due to its high temperature and thermal expansion, it touches the cold ingot mold, and seals off the vacuum in the gap that forms below it due to cooling shrinkage. This gap is filled with a noble gas such as helium or argon and used to improve heat transmission, i.e. to control the ingot temperature and the size and shape of the melt bath (Fig. "HIP of cast parts"). Below the crown, there is a ring (shelf) that solidified against the cold ingot mold. The crown and shelf can have zones with enriched or depleted alloy components, giving them higher melting points than the melt. Fragments can break off both the crown and shelf and fall into the melt. These particles cause a specific type of flaw when they reach the mushy zone at the bottom of the bath (Fig. "Segregations influenced by melting temperature"). The shape of the melt bath is very important for the type and probability of potential flaws (Fig. "Segregations influenced by melting temperature"). There is pronounced dendrite development in the mushy zone, promoting certain local alloy irregularities (segregations).

Flaws and their origins: If a detectable ultrasonic signal develops at a flaw in a forged part, it generally indicates dangerous separations and/or collections of non-metallic particles (segregations). These are referred to as sonic defects.
Segregations can appear as sonic defects and/or as structural changes. They can be visually detected as changes in color and structure on etched surfaces and separation planes in forged parts made from Ni alloys (e.g. IN718 and Waspaloy) and high-alloy steels (Ills. 15.2-22 and 15.2-23). Segregations take the form of

• microsegregations
• macrosegregations, especially in accumulations in the Nb and Laves phase (“freckles”, also see Fig. "Melting segreations cause turbine disk failure").
  
• ring-like structures (“tree rings”, Fig. "Influences of structure features").

Tree rings evidently only have a small influence on the operating characteristics of the finished part. Depending on the configuration, they are tolerable as weak points.

An especially common occurrence is white spots with very poor etchability. Interpreting these etching findings to determine the type of segregation and their effects on the integrity of the part is difficult even for specialists. A white spot does not always mean that the dynamic strength of the part has been unallowably compromized (top right diagram). If the specifications permit it, these occurrences may be accepted. Because the affected parts are usually extremely expensive (rotor parts, top left diagram), costing the equivalent of one or more middle-class passenger cars, unnecessary rejection of parts is very costly. This makes decisions difficult in marginal cases.
“Discrete segregations” usually appear as a light, clearly outlined spot several millimeters in width. These flaws are usually found in a centered circular cross-section with a diameter roughly half that of the bar (billet). The grain size in this zone corresponds to that of the surrounding material. The strength usually decreases towards the center of the white spot. In the case of white spots with no particle impurities, one can observe carbon depletion with a low density of carbides, as well as a light depletion of alloy components such as Nb, Ti, and Al. This also often applies to the Mo content. The danger posed by white spots corresponds to a potential strength loss of roughly 20%.

Other white spot types without impurities, such as the especially common solidification white spots or dendritic white spots, do not exhibit a worrysome decrease in static and LCF strength or ductility even at normal operating temperatures. Dendritic white spots occur near the center of the billet (Fig. "Segregations influenced by melting temperature").

So-called tree ring structures are also categorized as solidification white spots with regard to their development and negligible influence on the material properties (Fig. "Influences of structure features").

A considerably more dangerous occurrence is fouling-induced segregation, or “dirty white spots”, which can be both discrete and dendritic. They contain concentrations of oxides, carbides, and nitrides (Ills. 15.2-11 and 15.2-23). If the flaw is sufficiently large, cracks can be expected to occur, although these should be reliably detectable in the forged part with optimized ultrasonic testing. Even crack-free fouled flaws will exhibit a considerable decrease in the LCF life. In comparison, evidently, no dangerous concentrations of impurities have been found to date in the solidification white spots discussed above.

Mechanisms by which white spots develop: Material from the shelf, crown, and torus (ridge at the melting electrode, sketch of bath area at bottom) falls into the turbulent melt bath during the remelting process.
Depleted particles have a higher melting point and greater density than the melt bath. Because they sink more quickly, they can reach the solidification front at the bottom of the melt bath and bond to this area without melting themselves. The same is true of impurities made from nitrides and oxides that have accumulated in the shelf and crown.
Deeper melt baths minimize the probability of this type of flaw. This illustrates the importance of exact adherence to optimized and specified procedural parameters (Fig. "Forging process modeling").
Dendritic white spots are most likely created by electrode material that falls into the melt. This can be a consequence of melting of the area of the central cavity of the melting (upper) electrode. Unstable electric arcs seem to increase the frequency of this type of segregation. Therefore, it is recommended that, in the case of intolerable flaws of this type, the remelting protocols be carefully examined to find indications of this type of problem (Fig. "HIP of cast parts").
Solidification white spots only occur in a circular plain between the billet surface and half of the billet diameter (Fig. "Segregations influenced by melting temperature"). They require a very flat melt bath. It is assumed that their development is related to a change in the solidification rate. Slow solidification rates promote the coarsening of the dendrites, which results in slight depletion of the alloy components in the spaces between.

Figure "Formation of forging flaws": Before becoming a semi-finished part, a forged part passes through several production steps during which specific flaws and weaknesses can develop. The problems of remelting are dealt with in Ills. 15.1-11 and 15.1-12. They are primarily

• segregations (“1-A”),
• accumulation or depletion of alloy components
• accumulations of non-metallic particles such as oxides, carbides, and nitrides.

The dangerous segregations develop at the head of the billet and enter into the blank when the head piece that is removed is too short (example 15.1-1, Ills. 15.2-19, 15.2-21, 15.2-22 ). These segregations can influence the forging and heat-treatment process due to their tendency to crack (e.g. warm/hot cracks, Fig. "Mechanisms of hot cracking").
Reforming the billets (Ills. 15.2-19 and 15.3-11) taken from the ingot can be done in various ways. Typical procedures are open-die forging and drop forging (Fig. "Forging process caused characteristics"), as well as rolling, extrusion (Ref. 15.1-11), and creep forming. Reforming transfers heat into the material through internal friction. For this reason, highly reformed zones can heat up to the point that grain boundaries soften and result in hot cracks (“2-C”, Fig. "Flaws in forged rotor disks"). Further reforming can reseal these cracks, provided no oxygen entered them. If an oxide coating formed, such as in the case of external cracks, there is a danger of the cracks being pressed shut but not rebonding. They then remain present as flaws in the semi-finished parts. A further problem is insufficient local reformation of the forged part (insufficient degree of reformation, “2-A,B”, Fig. "Forging process caused characteristics"). This can result in structures that do not conform to specifications. These flaws and weak points can result in grain size, grain boundary orientation, and undiminished casting-dependent flaws. During forging, especially in processes that include cold forming, high residual stresses can be induced. In order to obtain an optimal structure from the forging process, the use of heat treatments is limited. Dangerous residual stresses can remain in the semi-finished parts (“2-D”, Fig. "Residual process stresses in titanium rotor disk").
Heat treatments can be used to improve formability during forging, and also after forging to create optimal structural/strength properties and minimize undesirable residual stresses (Ills. 16.2.2.4-14, 16.2.2.4-15, and Volume 3, Ill. 12.6.1-16). Thick cross-sections and a relatively low thermal conductivity of the titanium and nickel alloys make it difficult to attain the time/temperature rate desired in all zones of the semi-finished part. This can have a negative influence in local areas of the structure (“3-A,B”, Fig. "Solution annealing and gamma phase"). If, during heat treatment, high tension residual stresses are created due to large temperature gradients, hot cracks or creep cracks may occur (“2-C”). Sudden cross-section jumps, such as when forged parts were overspun before the heat treatment, can have a crack-promoting effect (“3-C”, Fig. "Disk cracking during heat treatment"). Improper heat treatment can itself even induce dangerous new residual stresses (“3-D”).

Figure "Forging process caused characteristics" (Ref. 15.1-16): Optimal structures Ti- and superalloys require suitable heat treatment during and/or after the forming process. The poor thermal conductivity of titanium alloys especially limits the part cross-sections due to the required cooling rate. The top diagram shows the cooling rate relative to the diameter of a typical, torus-shaped disk blank (middle diagram). In the case of larger cross-sections, oil or water cooling may be required. Intensive heat removal can result in powerful thermal stress. This can promote high residual stresses and cracking. The residual stresses can dangerously overlap with the operating stresses (Volume 3, Ill. 12.6.1-16).
In the depicted case, the left side marks two temperature ranges that are important for structural development. They are necessary for the transition to the b-structure of a titanium alloy at the end of the forging process. The depicted temperature distribution can be calculated with the aid of a thermomechanical model (Fig. "Forging process modeling"). One can see that, inside the cross-section, a temperature range of about 60°C can be maintained around the b -transition. Zones of equal plastic deformation are marked on the right side.
Naturally, zones near the surface (forging contour) are especially influenced due to the contact conditions with the forging tools (friction, heat transmission). In life-determining part zones, sufficient reforming is desirable, yet without the onset of recrystallization. This allows high strength to be combined with high toughness.
Judging from the raw part contour (dotted line), it can be expected that a good temperature and reforming will be realized during forging, and will be usable in the especially highly stressed hub area of the part, for example.
The bottom diagrams regarding an Fe alloy show the location-dependent influence of the forging process on the temperature-dependent yield strength (at left) and the LCF strength (at right). The listed values were determined using specimens from actual parts. In both cases, one can see the tendency (estimated with about 10% margin of error) towards poorer values in the inner zones, relative to the areas near the surface. This corresponds closely to the expected operating loads (high hub loads) on a rotating part.

Figure "Influences of structure features" (Ref. 15.1-4): These ring-shaped structures are known as tree rings. They are created during the remelting process and remain in the forged part. In a cross-section, the concentric rings exhibit the profile of the melt bath. They are caused by periodic irregularities in the solidification rate (bottom frame). In this case, the dendrite spacing and micro-scale distribution of alloy components are changed. These slight differences in etchibility (comparable with “solidification
white spots”, Fig. "White spots in Ni alloys") appear as rings in cross-sections. Evidently, the possibility of an influence of rotary fields cannot be discounted. Dangerous segregations or accumulations of particles are not promoted by tree rings.
Tree rings, like solidification white spots, are structural effects that evidently have only a minor influence on the strength properties of the materials.

Figure "Safety of parts by manufacturing": Increasing performance levels and efficiencies of modern turbine engines lead to higher loads on the parts. This means that tolerable weak points (no unallowable cyclical crack growth) become ever smaller. Serially implementable non-destructive testing procedures (primarily ultrasonic inspections) have already reached their limits (Ills. 17.3.1-2 and 17.3.1-3). The conventional process of semi-finished part production through casting and forging is depicted at left. In order to prevent flaws in the head sections of billets, a relatively large piece must be cut off, which has a considerable influence on the costs of the raw parts (Fig. "Terms of part production stages"). Additional security is expected from the reforming process during forging. This can correct any flaws and weakpoints (detail “A”) by breaking them up (brittle particles) and/or realigning them so that they are parallel to the direction of the loads (detail “B”). This safety aspect is considerable when compared with processes that do not use specific reforming. When producing the finished parts, the relatively roughly shaped forging blank must have a large amount of material removed (up to 95% of the billet), which is a very expensive procedure.
For this reason, it is advantageous to attempt to safely limit the maximum flaw size in a closely contoured finished part through the use of suitable procedures even during the production process of the raw parts.
Powder metallurgy (PM) using the “as HIP” process now allows improved safety to be achieved by limiting flaw size, while lowering costs (procedure on right). The “as HIP-process” is a PM technology with no subsequent forging. In order to limit the potential flaw size, the evacuated metal capsule, the contour of which matches that of the finished part as closely as possible, is filled with sifted metal powder of the desired alloy. This capsule is welded shut and compressed in an autoclave with high gas pressures (10³ bar) and temperatures (about 1000°C). Through this process, the powder sinters into the shape of a dense, fine-grained billet. After possible heat treatment and the removal of the capsule, the raw part is ready for non-destructive testing. The fine grain provides optimal conditions for ultrasonic inspection (Fig. "Limiting influences non-destructive testing"). A problem experienced with this process is the possibility that fouling of the powder during handling and filling can cause dangerous flaws that are considerably larger than the mesh size of the sieve would allow. For this reason, raw PM parts are now subsequently forged (“HIP and forge”) in order to take advantage of the breaking up and realignment of potential flaws. However, this requires raw parts with larger dimensions, which results in higher costs for the forging process. As a result, any cost advantages of the already fairly expensive PM part (powder costs, capsules, contamination-free filling, autoclave) are usually more than negated.

Example 15.1-1:

Excerpt (Ref. 15.1-5): “…After one of its…turbofans suffered a severe surge and turbine failure during climbout, sending pieces of the turbine section into a residential area near the airport…airline officials said yesterday….that the damaged pieces are believed to be from the `fourth-stage turbine area', although the extent of the damage won't be known until the engine can be inspected…A fatigue crack in the same component led to an uncontained failure and aborted takeoff…prompting the NTSB to call for stepped-up inspections of the fourth-stage low-pressure turbine hub…The investigations focus on the metallurgy of the turbine hub in some older…(engines) and in June NTSB wanted immediate and periodic inspections of hubs made before 1989 from a single-piece machined forging of Incoloy 901 alloy.
The hub's alloy is initially cast as an ingot, and the safety board said that until 1989, cerium and lanthanum were added during the foundry process to deoxidise the alloy. NTSB linked cerium and lanthanum “inclusions” - areas where the material seeps into otherwise pure casting-to January`s engine failure….NTSB noted that dye-penetrant inspection had failed to uncover the crack that led to the uncontained failure in January.”

Excerpt (Ref. 15.1-6): “…The problem is created when the ingot is turned into billets. The cerium and lanthanum rise to the top when the ingot is first cast and waste - which should include those two elements-is lopped of the top. What's left is then chopped into 20 to 30 forging blanks called mults, and investigators believe that all of the suspect hubs were made from one of the first three mult layers at the top, just below where the waste was removed…NTSB wants to see immediate nondestructive inspections of all …engines whose components may have been made from one of the top three layers of metal billet…“

Comments: This case serves as an excellent example of many known cases in different engine types. Evidently, this concerns a single melting of the billet, not repeated remelting as is commonly used today (Fig. "Flaws in ingots and forgings"). Exact, retraceable documentation of the entire casting and forging process is of decisive importance for the limitation of suspect parts and the implementation of specific corrective measures. Despite this, similar flaws continue to be reported in parts currently in operation (Fig. "Melting segreations cause turbine disk failure").

Figure "Hip process problems": Not only the powder and the HIP process, but also the capsule production process, the filling process, handling, and transport of the capsules all have an influence on the probability of flaws in the finished part.

The probabilities of flaws occurring in the different production steps “A”, “B”, “C”, and “E” are described in greater detail in Fig. "Material flaws from HIP-process". For this reason, only “E” and “F” are examined more closely here.

The transport and storage of the storage containers (“A”) for the powder and/or filled capsules before the HIP process can lead to segregations and unevenly distributed grain sizes. Demixing can occur if capsules or storage containers are subjected to vibrations while in a stationary position. This can occur during storage (vibrating floors) or during transport (e.g. in a motor vehicle). Powder grains can demix according to size as well as according to any minor, alloy-specific differences in density.

In addition to possible effects on the part through unnoticed gas absorption before and/or during the HIP process (Fig. "Material flaws from HIP-process"), there is a special danger to personnel (“F”). Trapped gas under high pressure can rip open the capsule and/or throw capsule fragments during capsule removal through etching or machining, or during separation of the filling nozzle. For this reason, after a successful HIP process, the filling nozzle is removed first under strict adherence to the required safety measures.

Figure "Material flaws from HIP-process": Even in materials that are produced using the “as HIP” process, specific flaws can be expected (Example 15.2-2 and Fig. "Inspections of engines in installed state for flaws"). The special problem is contamination of the powder. It must be noted that it is entirely possible for considerably larger particles than would seem to fit through the mesh of the filtering sieve to enter into the capsule (bottom right diagram). This is the case when long thin particles fall through the sieve lengthwise, as experience has shown them to do.

Flaws resulting from solid, liquid, and gaseous contaminants can be created in various ways (top left diagram):

• during powder production (“1”): remnants of ceramic filters, oxidation, absorption of gases (dissolving or in pores). Later, during the HIP process, these can lead to thermally induced porosity (Ref. 15.1-15).
  
• in storage containers (canister, “2”): contaminants, oxidation of reactive powders such as titanium.
• during filling of the capsules (“3”):
  • wear products from pipes (elastomers)
  • fouling in the lines
  • wear products from seals in the pipes and valves
• fouling in the capsule (“4”):
  
  • remnants of etching and cleaning media
  • sweat
  • rust

Experience has shown that these can cause very different flaws in parts (top right diagrams):

• reactive metallic impurities that stabilize through diffusion. The affected volume can be considerably larger than the original foreign particle.
• reactive organic particles: the carbon in these compounds, but also other components such as sulfur, can react with the base metal. This can result in a large carbide segregation relative to the contaminating particle. * hard, brittle non-metallic particles: typical examples are particles that have broken out of ceramic filters through which the melt passes before atomization. Another possibility is powders (Al₂O₃ ) used in abrasive cleaning of the capsule or filling pipe. These can be loose remaining particles or particles that have become stuck in the surface (charging effect).
  
• soft metallic nonreactive particles: foreign powders or wear products and chips.
• gases: as mentioned before, the metal powder can absorb gas during melting and atomization. The gas expands under the high temperatures of the HIP process and creates pores. Similarly finely distributed gas pores (argon) can form (“5”) if the seal of the capsule has been broken at least temporarily during the HIP process (bottom left diagram, Ref. 15.1-15). If air reaches the reactive metal powder (e.g. titanium) in the storage containers or insufficiently evacuated capsule, powder surfaces can oxidize and form weak points (bonding flaws).
  

Figure "Temperature related cracking" (Ref. 15.1-18): In order to increase strength, iron-based alloys and superalloys with Ni and Co bases are usually given heat treatments (hardening, structure optimizing). Heat treatments are also used in many finishing processes as integral steps. Typical examples include welding, soldering, casting, diffusion procedures (e.g. coating), forging, HIP, hot forming, and hot straightening. The heating process in production steps such as machining and spark erosion is a related process.
Cracks can be caused by heat treatments in various ways (heat treatment cracks). These are referred to as “fire cracking”, “strain age cracking”, or “stress (relief) cracking” (also see Fig. "Location of welds and finishing cracks"). These terms evidently also include warm/hot cracks (Fig. "Mechanisms of hot cracking").
Crack initiation is intercrystalline, i.e. the grain boundaries break open. This often occurs during the heating phase (top diagram), but this is difficult to verify. Longer annealing times after cracking “wash away” clues in the structure. In order to take specific corrective measures, however, it is important to know the time at which the crack occurred in the heat treatment cycle. This allows important conclusions to be drawn regarding the damage-causing influences. The analysis and assessment of this type of cracking requires a great deal of experience, and should cover the heat treatment process together with the furnace, cooling gas supply, temperature monitoring, charging racks, as well as the affected parts and materials.
The formation of heat treatment cracks is primarily caused and promoted by three main factors:

• tension (residual) stresses, especially near the surface

• structures that act brittly

• notches (Volume 3, Ill. 13-18)

Tension (residual) stresses: In most cases, these are created through thermal stresses due to temperature gradients. It is clear that high tension residual stresses occur especially in the area near the surface during the cooling phase. The heating causes plastic compression in the hotter surface region. During cooling, high tensile stresses are created (thermal fatigue mechanism), insofar as they were not broken down by creep in the constant temperature phase. It is not so obvious why high tensile stresses evidently also occur in the surface region during the heating phase. This phenomenon has been proven to exist by typical metallographic findings (Details A, B, C) from weld seams. Especially in the case of highly concentrated energy input during laser and electron beam welding (Detail “A”, Ref. 15.1), high tensile stresses occur in the directly adjacent area, which is considerably cooler but also thermally weakened. The grain boundaries tear open and melt can enter the crack, which is a typical characteristic (Fig. "EB welding flaws").
Hardening and solution annealing of alloys such as Waspalloy and C263 influences the development of the g' phase and causes volumnal changes. The effects depend on the alloy composition. In the area of sudden cross-section jumps, for example at welds with different hardening conditions relative to the surrounding material, this can lead to high tensile stress.
If the entire part is not evenly heated, thermal stress can cause plastic deformations (e.g. forming, surface machining). Therefore, when determining the time of crack initiation, changes in these factors should be noted.
If, for example, the machining of an electron beam weld seam is changed to a procedure with greater hardening effects, this can be the cause of crack initiation. Temporal temperature progressions that do not permit stress reduction through significant creep or recrystallization can be expected to result in dangerously high residual stresses. The induction of tension residual stress through forming and straightening depends greatly on the process conditions. For example, the friction between the tools and the part (contact surface conditions, lubricants) is very important.
Parts made from different materials (e.g. coatings, welded wear-resistant layers) can build up tension stresses during heat treatment. This is also true of annealing and positioning fixtures (cross-sections and materials are different than those in the part) that exert force on the part. If the height, arrangement, or gradients of temperatures change relative to the verified specified procedural data during the heating or cooling process and/or during the static phase, unusual stresses are to be expected.
An indicator for the at least temporary presence of high tension residual stresses is deformation following heat treatment. This deformation can also indicate the type, size, and direction of the causal residual stresses.

Structures that behave brittly: These are structural components that considerably increase strength within the grains and weaken grain boundaries. In g'-hardening alloys, precipitations can develop more rapidly within the grains during heating. This increases the strength considerably relative to the grain boundary, at least temporarily.
In this way, the grain boundary becomes a weak point and susceptible to cracking. Strain hardening has an especially pronounced influence on the achievable strength/hardness (bottom right diagram). The effect of the strain hardening can be increased in meta-stable materials such as the Co alloy HS25 by a simultaneously occuring change in the crystal structure. The flow limit is raised, and plastic deformability decreases.
The formation of brittle phases can also increase the strength difference between the grain interior and the grain boundary, promoting cracks. Some precipitation types in the grains can cause them to harden. Others have a weakening effect on the grain boundary. One example is carbides that tend to precipitate onto the grain boundaries during the heating phase. In combination with strain hardening, the development of the carbides during heating can be accelerated until recrystallization only occurs afterward, preventing the effect from being defused (Detail “C”).
Additional brittle phases are the Laves phase and s phase. However, they require sufficient hold times at specific temperatures in order to develop. In metastable alloys, the m phase forms on the grain boundaries.
Notches: They increase stress levels locally. These are usually geometric notches such as bores, grooves, and internal corners. Welded connections and cross-section jumps have a comparable effect.
Experience has shown that even apparently safe, small machining grooves and overlaps can act as crack initiators (Detail “D”) and cause grain boundary attack through oxidation or etching.