Remedies for production damages (also see Chapter 17) can be classified into two main categories. Prevention uses experience to identify problems that are already understood. This is a prerequisite for the optimization of specifiable, stable production steps and reliable quality control (FMEA, Ill. 17-10). This can require a considerable portion of the manufacturing costs. Therefore, the costs of prevention of damages and flaws are minimized as much as is safely possible.
If flaws occur nonetheless, priorities may change. In these acute cases, in extreme cases involving damage in engines, very high costs with damaging loss of prestige are likely to occur. Therefore, it is essential to take safety measures immediately. This involves limiting and locating the potentially affected parts or engines. It must then be determined whether the problem can be detected sufficiently early through measures such as non-destructive testing, inspections (e.g. boroscopic), or the analysis of instrument data. However, the possibility of achieving this depends greatly on whether the flaw type, flaw origin, and possible flaw mechanisms/damage scenarios are understood and explained in a way that is sufficiently plausible for decision-makers, authorities, and customers (e.g. licensers).
Flaws and problems with semi-finished parts can also have costly and time-consuming effects on subsequent manufacturing steps. The earlier the problem is discovered, the more ability one should have to limit the (potential) damage. Therefore, it is important to report observations concerning problems and anomalies as early as possible within the framework of a quality assurance process. The prerequisite is experienced, sensitized personnel (Ill. 17-5) and a suitable corporate culture (Ill. 17.5-2).
Preventive measures and remedies for damage to blanks and semi-finished parts are extremely varied (Ill. 15.3-1). The following overview makes no claims to completeness. The temporal order and extent of the measures must be case-specific.
Measures in case of acute problems in production and engine operation (Ill. 17-11):
In addition to damage, there are other latent problems. Cast parts such as cooled turbine blades, especially directionally solidified or single crystal versions, often have unsatisfactory yield rates, at least during longer learning- and/or startup phases. Repairs or reworking with procedures such as welding or high-temperature soldering is usually not permissible in new parts such as rotor blades. These methods can clearly not be used in rotating forged parts due to the failure mechanisms. On the other hand, repair welding can be entirely necessary in static parts such as turbine stators and housings. In the discussed cases, problems such as overly high production costs and delivery delays may occur.
Insufficient yield rates, i.e. overly high reject rates, are an indicator of unsuitable constructive design (Ill. 15.1-9). Typical characteristics of this type of problem are large numbers of cross-sections, large thin surfaces that are more suited to sheet metal constructions, sharp internal edges, unusually tight tolerances, structural specifications too far removed from practice, and strength requirements above an economically realistic scatter zone (Ill. 15.3-2). If this type of situation results, considerable production delays and cost increases may occur. The risk of unexpectedly high costs increases additionally because such demanding raw parts usually require elaborate development work by the supplier, making it impossible to secure multiple sources. This results in the unfavorable situation of having a single remaining supplier, aka single sourcing.
Other complex cast parts such as integral stators or housings make it virtually impossible to produce a flawless raw part without reworking. In this case, a certified, serially applicable subsequent machining process must already be developed (Ill. 15.3-4) before serial operation can be considered. This means that flaws, i.e. zones that are outside the specified limits of the design (usually dimensional or strength requirements), are removed or transformed into weak points. Through consideration of the weak points in the design and configuration, it should be possible to ensure acceptable operating behavior.
Illustration 15.3-1 (Example 15.3-1 and Ref. 15.3-5): There have only been a few reported cases in which parts were installed with material flaws that had an unallowable effect on their operating capability. However, if this does occur, one can expect serious damage potential, high costs and delays due to the remedies, as well as serious loss of prestige and reputation. Probable reasons for this are:
In order to minimize these risks and implement solutions as quickly as possible,
This will usually reveal the corrective measures to be taken, such as
Example 15.3-1 (Ill. 15.3-1, Ref. 15.3-4):
Excerpt: “…an inspection/monitoring advisory covering 8,887 newly manufactured …turbine blades that were made in February and March in order to recover 173 blades which the company believes were improperly processed…and shipped between Feb. 26 and Mar. 31.
…the problem arose in a furnace which is used to cure blade coatings. To be effective, the curing process must take place in an inert atmosphere-in this case, argon. However, during processing of three batches of blades the curing furnace apparently leaked, allowing ambient air to enter. Properly processed blades should have a lifetime of at least 4,000 cycles. But tests have indicated that the 173 blades in question would probably make it only to about 2,500 cycles before showing excessive wear.
…(OEM) engineers have developed a flow chart inspection/screening technique to help users to identify the 173 potentially misprocessed blades. The flow chart calls for various blade markings, including shipment date markings, to be checked, because manufacturing steps performed after a blade's furnace curing can erase the critical, identifying batch numbers.
..The engine manufacturer also is advising airlines to monitor suspect blades already fitted into powerplants and to remove them before they accumulate 1,000 cycles. That is equivalent to about 9-12 months of service.
…(The OEM) discovered the blade problem during a routine quality inspection. During the next several days the company attempted to characterize the problem and to determine which blades were affected….The facility manufactures approximately 60,000 blades a month. As of late last week, the company said it has identified the location of about 95% of the blades…
`…powerplant's operator (of an engine, fitted with a suspect blade) has been notified, the blade has been inspected via boroscope and it looks good…'“
Comments: It can be assumed that the air encroachment damaged the diffusion coating (oxidation protection). It is not clear whether the influence on the blade life span was due to greater sensitivity of the damaged coating to thermal fatigue (in which case cycles would be a more likely criterion), due to reduced oxidation protection (in which case run time is more important), or due to a combination of these factors. The use of a boroscope inspection for installed blades indicates that the damage mechanism is one that can be externally recognized sufficiently early. It is possible that it is increased oxidation in the form of an orange peel effect (Volume 3, Ill. 126.96.36.199-10) in a coated blade zone under especially high thermal stress. Usually, this area is the inlet edge.
The extent of the problem can be seen in the fact that, out of thousands of blades, a few suspect parts must be identified, traced into engines, and possibly monitored in that location. The prerequisite for such measures is the best possible documentation of the entire production process and logistics up to installation (Chapter 17.4).
Illustration 15.3-2: The probability of damage and its effects are often determined by the design engineer and/or the strength design department. For example, it is not surprising to hear about raw part problems in cast parts with pronounced cross-section jumps, sharp notches, and large thin walls. Shrinkage cavities are more likely in the thick cross-sections. Notches near cross-section jumps are predestined for crack initiation. Thin, sheet-like walls promote problems with filling and decrease the yield rate, which is at least a problem with regard to costs and deadlines.
A typical example is thin-walled pressurized housings such as outer combustion chamber housings that are changed to a cast version. The trend to highly complex integral parts such as turbine stators (Ill. 15.3-4) and blisks increases the probability of flaws. The larger the part, the higher the probability of problems with dimensions and deformation. This will at least intensify the required subsequent machining work.
The selection of the material is also extremely important for the potential flaw frequency. In addition to strength and function demands, consideration must be given to problems such as castability, tendency to thermal cracking, and the development of the necessary structure for attaining the required strength properties.
An additional problem is quality assurance, which must include the serially applicable procedures and their verification limits (Ills. 17.3.1-2 and 17.3.1-3).
Illustration 15.3-3 (Refs. 15.3-7 and 15.3-8): The trend towards complex, integral-cast parts such as turbine stators (top diagram) places great demands on the caster. The realizability of such parts depends largely on tolerable flaws (= weak points). In cast parts, these weak points are micro-cavities and unfavorable structures. Cracks that can be detected using serially implementable detection procedures are usually categorized as flaws that must not appear in the part. The operating stresses, which are a fundamental part of the design, determine the size of tolerable weak points (Ill. 17.3.1-1). Therefore, the design has to orient itself according to the actually realizable quality level. Strengths must also be determined using specimens with part-typical flaws (Ill. 17.3.2-1).
Depending on the material, it is entirely possible that cracks shorten the cyclical life span more than notches or shrinkage cavities (Ill. 15.3-7). They are caused by missing residual melt during solidification, giving them a typical appearance (bottom detail, Ill. 15.1-7). A collection of them is called a cavity cluster.
An inspection (Ref. 15.3-7) of cast high-alloy CrNi steel showed how the life spans of comparably sized flaws behaved in the LCF range:
Crack: Notch: Shrinkage cavity = 1:100 : 1000.
The fatigue strength of cast nickel materials with a typical strength of 400 to 450 HV is generally +-180 MPa. Ref. 15.3-8 estimates that only cracks larger than 1mm have a noticeable effect on this fatigue strength (HCF; Ill. 15.3-12). This behavior is attributed to the material-specific inhomogeneity of the structure.
As shown, cracks in cast parts must be taken much more seriously than cavities. These are thermal cracks caused by shrinkage stress during solidification and thermal stress related to temperature gradients. Experience has shown that these cracks tend to appear in areas where dangerously high thermal stresses are expected in later operation. For this reason, they can drastically reduce the life span of the parts. On the other hand, cracks in cast raw parts are always an indicator of problem zones in engine operation. Even if uncracked parts can be realized with an acceptable yield rate, crack-sensitive weak points discovered during the casting process should be defused through appropriate measures. Appropriate measures include increased radii, more elastic design, and more gradual cross-section transitions.
Illustration 15.3-4: Independent of the complexity of the part, the material, and the strength requirements (Ill. 15.3-3), casting flaws can worsen the yield rate to the point that flawless serial production cannot be realized. This is also dependent on the reparability (subsequent machining) of the raw parts.
If this requirement for deliverability is overlooked, high costs and time investments may be required for an alternative. Therefore, especially when introducing new, complex part geometries and more demanding casting materials such as intermetallic phases, a proven (certified) repair process must be secured. If necessary, the repair should be developed in parallel to the development of the material.
Illustration 15.3-5: Especially in fine cast prototypes with tight tolerances, such as turbine blades, it is possible that a straightening process may be required in order to minimize time delays and costs in development projects. Although cast Ni alloys generally have a considerably lower fracture strain (toughness) than forged alloys, it is still possible to cold straighten them if one is sufficiently careful. Of course, the part-specific, non-damaging plastic deformation potential must be taken into consideration. This can be determined in iterative tests beforehand. First, the suitable deformation process must be determined. For example, the clamping position and the type and location of the force application must be determined. The criterion is the elimination of the flaw to be corrected, without causing unallowable deformation of other parts. This process can be assisted by the use of crinkle finish, laser-aided measuring methods, and strain gauges in the areas known to be critical. During an incremental loading process, the part is observed for signs of damage such as crack initiation or unexpected shifting of the deformation force/moment. Penetrant testing can be useful for this. During the load test, it may be possible to conduct an acoustic emission analysis in order to continually monitor the load limit. This is especially advantageous if, for example, internal damages are expected in thin-walled hollow parts.
A straightened part should also be subjected to a relaxation test with a warming that covers the operating temperatures. This is necessary in order to estimate the danger of possible unallowable deformation.
In addition, dynamically highly stressed parts should undergo an operation-relevant dynamic fatigue test. This allows one to estimate the influence of residual stresses and local plastic deformations, as well as changed mass distribution, on the frequency (resonance) and dynamic fatigue strength. This is especially recommended for parts with complex geometries (such as cooled turbine blades).
Parts under high thermal fatigue stress should be examined in this regard. It is possible that major plastic deformation in critical zones has a life-shortening effect in critical zones.
Understandably, parts to be straightened should not have any coatings. This includes especially diffusion coatings, which are typically brittle at room temperature. Due to their deformability and stiffness, which are different from the base material, they can suffer uncontrollable cracking.
Fundamentally, bending a part back and forth is not allowable in serial applications. Following the straightening process, suitable stress reduction annealing should be done.
Straightening processes are problematic in single crystal materials, since plastic deformation can cause recrystallization (Ill. 15.2-4) during subsequent annealing.
Illustration 15.3-6 (Ref. 15.3-8): The most common flaws in parts made from cast nickel materials are shrinkage cavities (Ill. 15.1-7) and thermal cracks (Ill. 15.1-8). For this reason, it is important to know the difference between a tolerable weak point and an intolerable flaw.
The following observations are primarily concerned with life spans, i.e. dynamic loads in the range of fatigue strength (HCF).
The fatigue strength of flawless cast nickel materials is 180 MPa. This value is based on structural inhomogeneities, such as casting structure (dendrites and grain size and orientation), which are typical for the material.
A fracture mechanical estimation (top right diagram) shows that, under dynamic loads corresponding to the fatigue strength of 180 MPa, no growth can be expected from flaws that correspond to cracks with a length of under 1 mm (Ill. 15.3-11). Therefore, flaws with sizes below 1 mm do not have a notable influence on fatigue strength in cast Ni alloys.
Stress gradients, such as occur near notches or under flexural loads, defuse the effect of a flaw (center right diagram). The dynamic strength is reduced less at notches than it would seem according to the notch factor ak , i.e. the stress increase in the notch base (gray area in the middle left diagram). Therefore, the notch sensitivity bk is smaller than ak. This effect can also be explained by the lower probability of a growth-capable flaw in the small, highly stressed volume of the notch area. However, this is contradicted by the example of cast iron, which certainly contains a sufficient number of flaws. The effect can more probably be attributed to the stress gradients. The middle right diagram shows that inhomogenous materials with lower fatigue strength, such as cast iron, exhibit an especially strong reduction of flaw influence as stress gradients increase. The low fatigue strength of gray cast iron, which is an absolute value, increases along with rising stress gradients to a much greater degree than is the case in more homogenous cast materials. In case of a flat stress gradient (tension, bending of thick cross-sections), volumnal flaws have an especially strong influence on fatigue strength, since higher stresses are acting. The steeper the stress gradient (bending of a thin cross-section, notch), the more likely it is that the flaw will already be located in a less stressed area, due to the stress reduction effect. This reduces the negative effect on fatigue strength. Therefore, the larger the flaw, the more its damaging effect will be minimized as stress gradients increase (bottom diagram).
In summary: the lower the fatigue strength that is determined by typical material inhomogeneities (weak points), the better a material will behave in case of flaws, especially if a decreasing stress gradient is present.
These relationships are usually required when suspect parts have been installed or important deadlines, such as testing runs, must be met. In this case, it is useful to estimate the risk of damage and to certify parts for a limited time. However, one must take into consideration the fact that crack growth may occur during operation, for example, through creep supported by effects such as hot gas corrosion. This means that it is possible that, after a certain period of operation, a flaw size is reached that reduces the fatigue strength by an unallowable degree.
Illustration 15.3-7 (Ref. 15.3-9): The typical life-determining loads of many, especially cooled, hot parts such as turbine blades and combustion chambers are thermal cycles. In this context, thermal fatigue (TF) refers to pure cyclical thermal stress, while thermomechanical fatigue (TMF) refers to an additional overlay of mechanical loads (e.g. centrifugal force, gas force; Volume 3, Chapter 12.6.2). These loads are in the LCF range, meaning that they occur with considerable plastic deformation.
The following text deals with the behavior of typical cast materials for turbine blades with shrinkage cavities (microporosity).
The frequency of cavities is grouped into three categories in the cited literature:
6-12 Vol % = high porosity, 1-4% = low porosity, and subsequently pressed material with no notable porosity = HIP.
The porosity is caused by shrinkage cavities that form around the dendrites during solidification due to a lack of melt (Ill. 15.1-7). This development gives them their typical jagged shape (top middle diagram). Groups of cavities, i.e. cavity fields (top left diagram), tend to occur in thicker cross-sections.
Cavity fields accelerate crack growth. Remaining material bridges are severed, and the cavities combine.
The following types of cracking have been observed (middle diagram):
In case of high porosity, internal cracks form and connect the microcavities (connecting cracks).
Low porosity leads to cracking at the surface.
In the HIP state, crack initiation in a fine-grained surface zone occurs at the grain boundaries. The coarse grain below then cracks in a transcrystalline and interdendritic manner.
The influence of the cavities on the TMF life has been found to be clearly material-dependent (bottom diagrams). However, there is a general tendency for high porosity to shorten life spans considerably. At low strain, i.e. high load cycle numbers, low porosity is in the range of HIP. Therefore, the influence of minor porosity is in the range of unavoidable material-specific weak points, and can be dismissed. In case of large cyclical strain, the HIP state (Ills. 15.3-8 and 15.3-9) can be expected to have the longest life spans.
In the top right diagram, the crack growth is shown relative to the stress intensity. In addition to the mechanical stress, the stress intensity takes into account the effective flaw size. “Short cracks”, which are below the threshold value for growth capability that defines “long cracks”, can still grow to become long cracks (Ill. 15.3-14). This behavior depends on micro-structural impediments. If local characteristics of the material structure, such as grain size and grain orientation, act as impediments, short cracks will not grow.
In the vicinity of notches, the stress gradients in the notch base will determine whether a crack will stop at a micro-structural impediment (Ill. 15.3-6).
Illustration 15.3-8: Hot isostatic pressing (HIP) is done in an autoclave (left diagram). The parts are put into the autoclave, which is closed and filled with a gas, usually argon. The pressure of the gas is adjusted so that the desired pressure is created when the HIP temperature is reached. For Ni alloys, material-specific temperatures in the range of 1120 °C to 1315 °C and pressures between 1020 bar and 1720 are used (Ref. 15.3-1.6). For titanium alloys, 845-970 °C and 1030 bar.
This process has performed excellently for removing enclosed porosity and has become standard for high-strength, safety-relevant parts. HIP is also used to minimize reject rates of complex cast parts (casting salvage).
Creep closes the pores due to the high pressure difference. However, in order to ensure fusion that prevents notch effects from the flaw, smooth metallic cavity/pore surfaces are necessary. If the surface has a coating, such as oxides resulting from contact with the atmosphere, a satisfactory result can no longer be expected.
Flaws will also fail to close if there is gas in the hollow space. In most cases, the gas will have entered through a connection with the surface. Typical “opening mechanisms” include cracks, cracked cavities, or the collapse of a thin “surface lid” (detail). In case of an opening to the outside, the pressurized gas can enter the space and prevent the necessary pressure difference from building up. If the hollow space is closed, but contains gas, the resulting internal pressure will prevent complete closure.
Subsequent HIP has several different positive effects on strength properties (Ill. 15.3-9). One must remember that during the HIP process, the increased temperature means that a simultaneous heat-treatment is taking place. For this reason, the HIP temperature must be selected material-specifically so that the material properties are not affected by structural changes. The relatively slow cooling rates (about 10 °C/min) at the end of the HIP treatment are not sufficient for optimal structures to develop in some alloys. In these cases, an additional heat treatment must follow.
In unfavorable conditions, the HIP process can have various undesirable effects on the parts.
Naturally, as with heat-treatments, deformation may occur, but this is not unique to HIP.
If the cast parts have alloy segregations, as are typical for Ni alloys, high HIP temperatures can cause local incipient melting. In these cases, the HIP process should be preceded by suitable homogenization annealing.
With large hollow spaces, the surface can collapse noticeably. Small distributed pores lead to a scarred surface.
The pressurized gas must be extremely pure. Impurities in the ppm range are sufficient for surface reactions or gas absorption. Typical gas impurities are hydrogen, nitrogen, oxygen, carbon monoxide, water vapor, and hydrocarbons. In Ni alloys, grain boundary oxides may develop. In titanium alloys, the development of an a-case at the surface is promoted. In order to ensure the required gas purity, the gas composition must be continually monitored.
If carbon has evaporated from a graphite heater, it can settle on the part surface and react with it, i.e. diffuse into it. This can result in coke formation with the development of carbides that may also compromise oxidation resistance. This should be prevented with suitable covers.
Illustration 15.3-9 (Ref. 15.3-1.6): If flaws were successfully removed through HIP, their notches disappear. As expected, depending on the original flaw size (Ill. 15.3-7), the dynamic fatigue strength is increased in both the LCF and HCF ranges (top left diagram). It is worth noting that both the median value and scattered values improve.
Even the static strength behavior at high temperature, such as short-time strength (resistance to thermal cracking) and fracture strain (toughness), can benefit from HIP (top right diagram).
An unexpected result is improvement of the notched impact strength (bottom diagram). This could be useful for parts with containment tasks. There is cause for concern if the HIP process is omitted, for example, due to cost reasons, even if verification tests were done on a HIPed part.
Illustration 15.3-10 (Refs. 15.3-10 and 15.3-11): Nickel and titanium alloys for aircraft engine rotors are created through multiple remelting. A typical process is remelting using an arc in a vacuum (Vacuum Arc Remelting =VAR). The quality of the material in this elaborate and complex process is especially dependent on the strict adherence to the process parameters and the prescribed functioning of the facility. Flaws that can be traced back to deviations in the process, have typical characteristics (Ill. 15.1-12). This makes it possible to trace problems back to the remelting process. For example, in case of operating damage that has been determined to have been caused by a material flaw, there is a possibility of obtaining clues to the origin of the flaw through examination of the remelting process, especially process documentation (notes, diagrams, protocols). If this can be used to establish the time of occurrence, it is possible to limit (e.g. charge, production lot) the potentially affected parts.
Precise documentation of the remelting process (e.g. written records of process parameters) enables later examinations to discover deviations. These are generally located around the flawed part of the casting ingot or the billets that were forged from it (Ills. 15.2-22, 15.3-11, 15.3-12).
The dependency of the quality of the raw part on the process parameters, i.e. deviations and malfunctions of the remelting facility, make this high sensitivity understandable. Even changes that are only intended for optimization may also demand new rating and acceptance of the affected part. Comprehensive cyclical centrifugal tests may also become necessary.
Illustration 15.3-11: The verifiable documentation of the path of a forged part in the total casting and forging process, and especially the location of the raw part, is of decisive importance for minimizing risks in case of damages and/or accidents.
One prerequisite is always the identification of the potential damage-causing flaw (Ill. 17-11). This includes determining the type of flaw and the mechanism of its development (Ills. 15.1-13, 15.2-14, and 15.3-12). The point of departure for tracing the flaw is the serial number of the damaged part (Ill. 17.4-1). On the basis of its position in the various intermediate stages, A, B, C, and D, it is possible to draw conclusions concerning the time of flaw creation, the flaw cause, and the number of potentially affected neighboring parts and their identification (Ills. 15.2-16, 15.2-18, 15.2-19, 15.2-21, 15.2-22, and 15.2-23).
Illustration 15.3-12: The casting/remelting process influences the distribution of flaws that originate here (Ill. 15.1-12). The distribution of the flaws created in this process can be grouped according to flaw type (Ref. 15.3-10, Ill. 15.2-14). For example, in the case of nickel alloys, solidification white spots can be expected in outer zones. Fouled or unfouled discrete white spots are located further inside. Dendritic white spots are concentrated in the center. The flaw distribution is transferred from the ingot (Ill. 15.3-11) to the billet by the forging process. This material is compressed into the semi-finished part (pancake). During this process, the flaws also maintain their positions relative to one another. Therefore, the type and location of typical flaws in the finished part indicate the mechanism by which they developed in the casting process (Ill. 15.3-8). They also indicate the probability of additional flaws, for example in neighboring raw parts (Ill. 15.3-16) such as the hub of a rotor disk.
The location of flaws in the part has an important influence on operational safety. For example, potential flaws that are located in a part-specific highly stressed zone are especially high-risk. In this case, immediate and extensive measures may be necessary.
Illustration 15.3-13 (Ref. 15.3-3): The depicted process serves to reduce the probability of flaws and weak points in a forged part (Ills. 15.1-13, 15.1-14, and 15.2-11), as well as optimizing its properties. The core of the process is modeling the forging process with regard to the quality demands. Individual steps of the process, such as heat treatment, are also taken into consideration. Advantages can be expected from this approach, especially when it is used during development. The schematic depiction is intended to provide an impression of the complexity of modeling and the large variety of influences. It is understandable that a trial-and-error approach is becoming less and less capable of dealing with the increasing part loads.
The above process is an iterative process combining hardware tests, FEM calculations, and material analysis (strength, structure). This approach also leads to a deeper understanding of their relationships, and increases the ability to respond quickly and in a targeted manner in case of damage.
Illustration 15.3-14: This Kitakawa diagram is based on fracture-mechanical examinations. This diagram can be used to determine the dependency of fatigue strength (HCF) on the crack size or corresponding flaw size. It does not apply to the LCF range, in which dangerous crack growth can be expected after the first load cycles.
Here, the behavior of a forged material with a mean grain size of about 0.1 mm is shown schematically. The connection between the relationship of the effective stress amplitude (Ds) to the square of the fatigue strength (sD) and the crack depth “a” of a corresponding flaw size is plotted in the diagram. Crack depths in the range of 0.001 mm do not exhibit crack growth leading to macro-cracking. Therefore, the fatigue strength is not influenced by flaws of this size. Crack depths of up to about 0.1 mm are usually below the size necessary for growth (threshold) of “long cracks” (Ill. 15.3-7). The top right diagram shows the behavior of short and long cracks. The load value is given as the amplitude of the stress concentration Dk. This takes into account the influence of the crack on the stress amplitude at the crack tip. The threshold value of long cracks with a depth of about 0.1 mm roughly corresponds to the mean grain size (Ill. 15.1-4). Long cracks, in this case above about 0.1 mm, should have accelerated growth and lower fatigue strength. Under unfavorable conditions at the crack tip (hardening, stress buildup), even short cracks below the threshold of long cracks can become capable of growth and turn into macro-cracks. Growth of short cracks is also possible if additional external influences such as corrosion or creep are also present.
Summary: Flaws within the size range of the normal, typical structural characteristics such as grain sizes and/or phases (short cracks) do not reduce fatigue strength. This is true provided that there are no additional influences such as corrosion or creep acting on the material.
Illustration 15.3-15 (Ref. 15.3-12): As with any material, PM materials can also have unavoidable flaws. The size of these flaws is determined either by the production process and/or by the limits of the available serially implementable non-destructive testing procedures (Ills. 17.3.1-2 and 17.3.1-3). These flaws must be taken into account when determining the design data for the part strength. This changes flaws into (specified) weak points. The left diagram shows that flaws near the surface have a considerably stronger effect on LCF dynamic strength than internal flaws do (also see Ill. 15.3-6). In the right diagram, the two dot/dash line curves show the (theoretically) fracture-mechanically determined life span reduction as flaw size increases. The solid line and the dashed line curves evidently show the behavior that was determined using specimens. The trajectory of the curves into a horizontal direction reveals a type of fatigue strength effect. At higher loads, specimens without flaws show the same cyclical life span as specimens with flaws of 0.15 mm. Here, it can also be assumed that this threshold value corresponds to the effective size of typical weak points in this material.
Illustration 15.3-16: The probability of a material flaw occurring and causing the fracture of an uncontainable rotor component depends on the combination of several unfavorable conditions. Experience supports this, since tests during inspections often find flaws (segregations) that did not lead to crack growth (Example 15.2-6). In order to ensure the desired failure probability of 10-9, the probability of a dangerous flaw occurring in the entire volume of a disk (“P1”) must already be very low through optimized manufacturing and testing processes. The damage-effective probability is the product of factors such as the probability of a flaw occurring in a highly stressed part zone (“P2”, also see Ill. 15.3-12) and unfavorable orientation relative to the loads (“P3”), resulting in the formula P1xP2xP3. However, even a very low probability of damage-causing flaws should not lead to complacency. Murphy`s Law is especially applicable to turbine engineering: if something can go wrong, it will.
The trend towards the greatest possible utilization of strength in the entire part volume increases the influence of weak points and flaws (Page 16.2-1) on the design data.
Illustration 15.3-17: In case of warping, the process involved is dimensional changes due to changes in residual stresses (breakdown, buildup, shifting; Ill. 188.8.131.52-15). Warping can have an unallowable influence on parts. This can be due to the dimensional changes themselves. However, it may also only cause temporary changes in the part. These include joining gaps at soldered connections and diffusion welds that depend on temperature gradients during the joining process. In these cases, the problems are bonding flaws (Ill. 184.108.40.206-37).
Heat treatment can cause warping in various ways.
Creep reduces residual stress (relaxation, Ill. 220.127.116.11-15). Typical examples of processes that induce high residual stresses with the risk of warping are welding (right diagram, Ill. 18.104.22.168-17), forming (deep drawing, etc.), and chip-removing machining processes (Ill. 22.214.171.124-10).
Heating or cooling can create temperature gradients with high thermal stress, which lead to plastic deformation. In addition to the permanent deformations, residual stresses may be induced in the cooled state. Parts with an especially high risk of warping are, for example, parts with large differences in their cross-sections, causing them to heat up and cool down at different rates. This also applies to parts with a geometry that only permits uneven heat application/removal. This may be related to blocking of the thermal radiation (shade), thermal conduction, and convection.
There are also residual stresses that result from structure-dependent volume changes. Hardening processes are typical of this.
The location and appearance of warping indicates its causes. The left diagrams show warping due to thermal stress caused by differences in temperature distribution.
In addition to parts, warping also especially affects equipment that is subjected to cyclical temperature changes. These include primarily charging racks and containers for diffusion coatings. For this type of equipment, elastic designs (Ill. 126.96.36.199-18 and Volume 3, Ill. 12.6.2-20) without cross-section jumps have proven effective, since they endure more thermal cycles without unallowable warping or cracking. Suitable equipment can minimize warping during heat treatment. This type of equipment requires extensive experience with constructive design and material selection (thermal strain behavior). For example, in the course of a straightening process, sufficient contact between the equipment and part during the heat treatment must be ensured.
Illustration 15.3-18: Various mechanisms promote crack initiation due to temperature cycles during heat treatment (Ill. 15.1-7) and welding. The combination of both influences, i.e. heat treatment of a weld (table 188.8.131.52-1), has a high potential susceptibility to cracking. Measures specifically matched to the problem can provide relief (Ref. 15.3-5) :
“1”, determining the time of crack initiation (Ill. 5.1-19): With the aid of metallographic sections (Ill. 17.3.2-5) and analysis of the characteristics of laboratory fractures (17.3.2-7), experienced specialists can draw conclusions regarding the crack initiation mechanism and causal influences. It is important to determine whether the cracks occurred during the heating phase, the dwell phase, or during the cooling phase. This makes it possible to minimize crack initiation through suitable temperature management.
“2”, chip-removing machining, such as for seam preparation, can be suspected to be a damage-causing influence even if only certain machined surface areas have cracks. This type of influence is likely if cracks form on only one flank of the weld seam, with no structural differences as an alternate explanation (Ill. 184.108.40.206-14). Previously unreported cracking indicates unfavorable changes to the machining process. In superalloys, which are typically difficult to machine with chip removal, high tensile stress can be induced, and can later tear open the grain boundaries when the material is heated.
“3”, reduction of tension residual stress: the mechanism described in “2” can be defused with the aid of shot peening in zones susceptible to cracking. The opposite approach of stress reduction annealing before welding can be problematic. In certain cases it can even promote crack initiation, if the structure is unfavorably changed, for example if brittle phases were created.
“4”, avoiding cross-section jumps: If the cracks are in areas with pronounced cross-section jumps (Ill. 15.2-13) with the usual dimensional notches, a better contour for heat treatment should be selected if possible.
“5”, optimizing structure: If the metallographic inspection (Ill. 17.3.2-5) shows that the crack initiation can be explained by structural differences (e.g. coarse grain, Ill. 220.127.116.11-14), optimization of the raw part production process may be appropriate. Fundamentally, the parts should not have clear structural differences, such as coarser grain, with the material used when testing the production processes.
15.3-1 “ASM Handbook”, -1.1, Volume 4 (“Heat treating”), -1.2, Volume 5 (“Surface Cleaning, Finishing, and Coating”), -1.3, Volume 6 (“Welding, Brazing, and Soldering”), -1.4, Volume 7 (“Powder Metallurgy”), -1.5, Volume 14 (Forming and Forging”), -1.6, Volume 15 (“Casting”), -1.7, Volume 16 (“Machining”).
15.3-2 Peter Adam, “Fertigungsverfahren von Turbotriebwerken” Birkhäuser Verlag,, 1998, ISBN 3-7643-5971-4.
15.3-3 A.Barussaud, Y. Desvallees, J.Y. Guedou, “Control of the Microstructure in Large Titanium Discs. Application to the High Pressure Compressor of the GE90 Aeroengine”, periodical “Titanium `95: Science and Technology”, pages 1599-1608.
15.3-4 S.W. Kandebo, “Pratt Plan Aims To Recover Misprocessed JT8D Blades”, periodical “Aviation Week & Space Technology”, May 11, 1998, Seite 67.
15.3-5 D.L. Klarstrom, “Heat Treat Cracking of Superalloys”, periodical “Advanced Materials & Progress”, 4/1996, pages 40EE - 40GG.
15.3-6 M. Beck, K.-H. Lang, “Feingussfehlstellen, Zulässigkeit von Fehlstellen in Feingussbau-teilen bei thermisch-mechanischer Wechselbeanspruchung”, FVV-Vorhaben 696, Issue 723, 2001.
15.3-7 P.Hausild, C.Berdin, P.Bompard, N.Verdière “Influence of Shrinkage Cavities on Fracture Behaviour of Duplex Stainless Steel”, Proceedings of the conference “Duplex 2000”, Venezia, Italy, on Oct. 17-20, 2000, pages 209-218.
15.3-8 H.Huff, “Die zulässige Beanspruchung bei Ermüdungsbeanspruchung”, periodical “Materialwissenschaft und Werkstofftechnik”, 32, pages 1-6, 2001.
15.3-9 M.Beck, K.H.Lang, “Zulässigkeit von Fehlstellen in Feingussbauteilen bei thermisch-mechanischer Wechselbeanspruchung”, Forschungsvereinigung Verbrennungskraftmaschinen (FVV), Issue 723-2001, “Feingussfehlstellen”, pages 1-138.
15.3-10 L.A.Jackman, G.E.Maurer, S.Widge, “New Knowledge About `White Spots' in Superalloys”, periodical “Advanced Materials & Processes”, 5/1993, pages 18-25.
15.3-11 L.G. Hosamani, W.E. Wood, J.H. Develetian, “Solidification of Alloy 718 During Vacuum Arc Remelting With Helium Gas Cooling Between Ingot and Crucible”, Proceedings of the “International Symposium on Metallurgy and Applications of Superalloy 718”, June 12-14, 1998, Pittburgh, Pennsylvania, pages 49-57.
15.3-12 R.L.Dreshfield, “Defects in Nickel-Base Superalloys”, periodical “Journal of Metals”, July 1987, pages 17-21.