Table of Contents
12.5.1 Fundamentals of Creep Behavior (Material Properties, Damage Mechanisms, Characteristics)
Metallic materials that are under mechanical loads at a constant, material-specific, high temperature (about 0.2-0.3 times the melting temperature; Fig. "Creep behavior of different metals"), experience time-dependent plastic deformation, or creep. This creep strain continues until it results in a creep fracture (Fig. "Life limiting 'creep fracture'").
Engineers do not orient the design of creep-stressed parts according to the hot tensile strength determined in an accelerated test. This short-term strength is better suited to the estimation of rotor-bursting RPM. The material properties that are relevant for determining engine part life are creep strength (e.g. 10,000 h) and/or the creep strain limit (e.g. 0.1%) at a certain temperature and load level. These characteristic values are illustrated by creep diagrams that include three parameters: temperature, loads, and time.
Naturally, in engine construction, the materials used will be creep-resistant at high temperatures. Usually, the high creep strength of these Ni-based superalloys is due to their g' hardening phase (Ni3Al). In general, cast materials have better creep strength at high operating temperatures due to their structural characteristics and/or their alloy composition. Cast materials with a typical high Al content are not forgeable for a large portion of the hardening phase. If strength with sufficient ductility is the primary objective, as with rotor disks subjected to LCF loads (Chapter 12.6.1), then forged materials are used. If high temperature resistance is required, such as for turbine blades, then cast materials are more suitable. The special weakness of these is the grain boundaries, which are usually oriented across the direction of the stress ( ). In general, as grain size increases, creep strength also rises (Fig. "Grain structure influencing metal properties"). Unfortunately, dynamic strength tends to decrease. Creep strength can be further increased through the use of directionally-solidified alloys and single-crystals ( ). Another approach is to prevent creep deformation through targeted application of reinforcements such as extremely fine particles (dispersion hardening).
Depending on the temperature and structural stresses, creep can occur with very different mechanisms (Fig. "Verification test of residual life spans"). If, during creep, pores are created on the grain boundaries, they are an indication of the extent of creep damage and can be used to estimate it. If one has sufficient part-specific experience, it may be possible to draw conclusions regarding the residual life span (Fig. "Scatter of creep life span data").
In order to minimize weight and lower costs, the materials used in engine construction are used at temperatures near their thermal limits, which depend on the mechanical loads. Because of this, the following assumption applies to all materials: a temperature increase of about 15 °C reduces the creep life of a part by half (Fig. "Life span decrease by overtemperature"). Therefore, seemingly minor changes in the thermal environment of the parts can have serious consequences for the cost and safety of an engine.
Creep strain can have a damaging influence on engine parts in many different ways (Fig. "Creep effects and part behavior"). For example, if labyrinth clearances are lost, then damaging rubbing may occur. The lengthening of the blade changes its resonant frequency and may result in resonance followed by dynamic fatigue fractures.
Figure "Creep behavior of different metals": An important question concerns the threshold temperature for a material, above which creep becomes damage-relevant. This question is not easy to answer. If one ignores the fact that relaxation, e.g. a certain reduction of shot peening-induced residual stresses, already occurs at room temperature in titanium alloys, the temperature above which damage-relevant creep occurs can be defined as 0.2-0.3 times the melting point of the material in question. Accordingly, the chart shows the reasonable operating limits of shot peening as a reference value for the beginning of creep (Ref. 12.5-9).
Figure "Creep resistance of Ni-based alloys": The 1000h creep strength rating can be used as an evaluation standard for turbine blades in military applications.
The curve N90 depicts forged alloys (product name: Nimonic) of the type found in older engines developed in a period of the 1950s and 60s.
IN713 is a standard cast material with a non-directional structure.
Single crystal alloys further increase the maximum operating temperature by another 30°C. This translates into two to four times longer part life than non-directional cast materials operating at the same temperatures.
Dispersion-hardened alloys have finely dispersed ceramic particles in their structure (Fig. "Reinforcing creep resistanc"), which minimizes creep. However, their brittleness and complex manufacturing process has prevented them from being widely used.
Figure "Life limiting 'creep fracture'": Metallic materials under static loads deform plastically (creep) during the duration of stress and fail upon reaching the fracture creep limit. This process occurs along the creep curve (constant temperature and static tensile stress), which is depicted here in its typical form. Depending on the material, load levels, and temperature levels, the curve can change considerably if the three typical phases develop very differently.
The shape of the curve is characteristic of a process in which two opposite effects act. In this case, hardening (healing) is marked by the black arrow and weakening (damage) by the white arrow. Depending on the stress phase, hardening will be greater (primary creep zone I), both effects will be roughly even (secondary creep zone II), or weakening will prevail (tertiary creep zone III) shortly before the ultimate failure through fracture. The total life under creep stress is called the stress rupture life. Zones I and II are used technically, and determine the design of the engine parts.
Figure "Life span decrease by overtemperature": Load levels and temperature have a special influence on the creep limit/stress rupture life. A part that is under creep stress, especially a cooled part, usually has pronounced temperature gradients (top diagram). The designer uses the temperature distribution as a basis for the configuration of the hot parts. In turbine rotor blades, the life-determining zone is usually near the center of the blade.
The middle diagram shows the typical temperature-dependence of the creep behavior of an Ni-based cast material, using a logarithmic standard on both axes. If one assumes an averaged temperature in the creep-critical blade cross-section (“1”), the point of intersection (“2”) with the curve for 1030°C will give the life span “L/2”. If the temperature of the engine part is 15 °C lower, the resulting life span is twice as long, i.e. “L” . Accordingly, a 15°C increase in temperature will reduce the life span by half. Because this process is exponential, every additional 15°C temperature increase again halves the life span. Therefore, an increase of 45 °C corresponds to a life span reduction of about a factor of 10. This behavior is observed espeically in all hot part materials being used near their operating limits (where they must be used due to their high cost). One can see the importance of preventing unplanned temperature increases. The causes of such temperature increases are extremely varied:
- Faulty sensors: fouled pyrometer windows, damaged thermal elements.
- Blocked hot parts: dust buildup in cooling air bores and cooling air vents.
- Internal oxidation
- Failure of thermal barriers
- Regulator problems
- Problems with the combustion chamber and injection systems: changes to the temperature distribution at the combustion chamber exit (see Chapter 11.2.2).
The bottom left diagram shows the changes in the creep curve as temperature changes, revealing the described influence on creep life. The creep strain becomes considerably faster and greater as the temperature rises. The curves show average values of the fracture creep strain.
Increasing the tensile stress has a similar effect (bottom right diagram). High mechanical stress leads to large creep strain and short part life.
Figure "Creep strain influenced by hot gases" (Ref. 12.5-1): The stress rupture life, i.e. the time until creep fracture, is also dependent on the surrounding atmosphere (also see ), which probably accelerates the growth rate of surface cracks, especially. Flowing hot air has a damaging effect when compared with still air. An especially drastic shortening of part life has been observed under the influence of sea salt (also see Fig. "Hot gas corrosion influenced by operation").
Figure "Creep mechanisms in hot parts": The damage mechanism of creep deformation occurs in various ways, depending on the material and loads. Minor creep strain (brittle behavior) either occurs with creep pore formation on the grain boundaries (which are oriented across the main tensile stress), or with large cracks forming in the gaps between the grains (left diagrams). Larger creep strain occurs with the formation of creep pores both inside the grains and on grain boundaries that are more oriented along the direction of the stress. Especially pronounced ductility develops if the creep damage is accompanied by an equally fast, opposite mechanism. This results in maximum deformation with pronounced constriction in the failure zone with no grain boundary damage or pore formation. This effect is used in superplastic forming, for example.
The bottom diagrams illustrate mechanisms of cracking and pore formation. Pore formation is based on deterioration in the region of grain boundary inhomogeneities (e.g. carbides). These areas (lattice widening) become points where the gaps in the atomic lattice collect “condense”. This mechanism is clearly detectable in SEM-analyzable fracture surfaces and metallographic sections (Fig. "Pore formation as creep damage"). Experience has shown that pronounced pore formation can be observed in forged materials, but is rather rare in cast alloys.
If materials tend to cracking open of the grain gaps (grain boundary triple points), micro-cracking will occur here.
Figure "Pore formation as creep damage": Creep damage (Fig. "Scatter of creep life span data") is accompanied in many materials (not materials without grain boundaries, such as single crystals) by the formation of pores (creep pores, Fig. "Creep mechanisms in hot parts") along the grain boundaries, which tend to run across the direction of the tensile stress. This effect is especially pronounced in Fe- and Ni-based forged alloys. The creep pores grow together over longer operating times and lead to a final stage with the jagged surfaces that are typical for creep fractures.
Estimating creep damage in order to determine the continued usability (residual life) of not yet visibly damaged engine parts is very difficult. It has many prerequisites, including extensive experience with the affected part type and the behavior of the specific material under the special operating conditions of the engine part (Fig. "Creep resistance by coarse grain structure").
Figure "Verification test of residual life spans": At first glance, it may seem plausible to recreate the degree of creep damage, i.e. the creep residual life, through a subsequent load test under laboratory conditions. Because long test times are very expensive and time-consuming, accelerated tests are often preferred. In order to achieve this, if the temperature is the same as the operating temperature, then the load levels must be increased. This procedure increases the speed of the creep deformation (Fig. "Life span decrease by overtemperature"). However, the creep rate determines the damage mechanism of the creep process. The diagram illustrates these relationships:
The following relationships are depicted in the diagram:
The creep mechanism which occurs under operating conditions is primarily dependent on the temperature and tensile stress levels. The value that steers the damage process is the strain under shearing loads (shear stress standardized with the shear modulus) over the test temperature standardized with the melting temperature. One can see that in the range of technical applications (creep rate < 10-12 s-1) displacement creep, grain boundary creep, and diffusion creep are all possible.
If the accelerated laboratory test results in an increased creep speed and therefore incites a different damage mechanism than during operation, the measured residual life spans will be closer to those of new parts. This leads to completely inaccurate estimates of the residual life span. Under these conditions, there is no alternative to accurately simulating the operating loads, which leads to elaborate tests. Structural analysis of parts that were damaged in operation can make estimates more accurate. Characteristics include creep pore formation, rafting of the g' phase, and changes in any coatings (Fig. "Metallographic findings of thermally stressed parts").
Figure "Scatter of creep life span data": The typical large scattering of the creep life creates a tendency to overestimate the importance of differences that seem to be relevant to damage. A scattering of 50% around the average value (middle diagram) is not unusual.
The bottom diagram depicts a turbine disk after a short-term, extreme overtemperature. Several blades show no noticeable lengthening, few are clearly lengthened, a few individual ones are fractured. It seemed correct to assume that the fractured and clearly lengthened blades were not made in accordance with specifications. However, subsequent inspection revealed that there were no unallowable deviations in the structure or in the strength values. Evidently, the variation in damage was due to normal scattering within specified values. Blade design uses specific minimum life span values that are considerably below the average values featured in brochures. This is also true for accelerated tests conducted as part of quality assurance.
The pore formation (Fig. "Creep mechanisms in hot parts") scatters widely over the creep time and creep strain (top diagrams). For this reason, creep pores only provide a reference value for the damage, i.e. the spent/remaining life span. If blade sets are evaluated according to the creep pore formation, then the implementation of a safe, conservatively estimated limit is reccommended. Only blade sets in which several blades (about 6 or more) that were evenly distributed around the circumference have been tested, and found to be considerably below the limit value, are suitable for reinstallation and/or rejuvenation. All other blade sets must be scrapped, even if they probably contain some blades that could be reused. Experience has shown that sufficiently reliable judgments can only be made in the case of forged materials with pronounced creep pore formation. However, these blades are only found in older engine types.
Figure "Creep deformations in turbine blades": Creep deformation on hot parts takes many different forms. The most typical is the bulging or bowing out of the thin rear edges of turbine stator vanes (airfoil bowing, left diagram, Refs. 12.5-5 and 12.5-9). When the rear edge heats up rapidly (e.g. start-up), it is prevented from deforming by the cooler surrounding blade and shroud zones. This creates high compressive stress levels and the bowing that serves as a sign of overheating (Fig. "Creep deformation at overheated turbine vanes").
If turbine rotor blades are overheated in the blade leaf area, it can result in distinctly noticeable constriction before fracture occurs (middle diagram, Fig. "Creep effects and part behavior"). On coated blades, a crack field oriented perpendicular to the centrifugal forces can be observed in the deformation zone (also see Fig. "Creep effects and part behavior").
Creep strain can cause blades to unwind, which causes rotor shrouds to do the same (right diagram). This can result in loss of the support necessary to suppress rotor vibrations (Fig. "Design of rotor blade tip shrouds"). If the blade tip is overheated (e.g. following radial movement of the temperature profile), this may be revealed by the shroud bending open.
Figure "Grain structure influencing metal properties": Experience has shown that new engine types require ever higher creep strengths and operating temperatures. This is related to rising circumferential speeds, higher gas temperatures, and longer life spans (Fig. "Historical trends of of fighter engine problems"). Because the blade materials in modern engines are still Ni-based alloys and will remain so for the foreseeable future (good combination of properties, availability), there have been attempts to optimize the structure to suit the specific operating characteristics. After cast blades with a multi-crystal, non-directional structure (multi-axial structure) replaced forged blades (top left diagram), the 1970s (Fig. "Development curve of thermal strength") saw the arrival of blades with directionally solidified structures for use in serial engines (top middle diagram).
In these latter blades, temperature gradients oriented along the length of the blade during the solidification phase were used to achieve a radial configuration of the grain boundaries. These weak points for creep stress now ran alongside the prevailing centripetal stresses. However, it turned out that the intensively internally-cooled, directionally solidified blades were also subject to powerful thermal stresses that were oriented perpendicular to the grain boundaries. This resulted in “wood-like” cracking of the longitudinal grain boundaries (Fig. "Grain boundaries influencing thermal fatigue"). This was the motivation for development of and, in the 1980s, introduction of blades made from single crystals, which was achieved by further refining the directional solidification process and optimizing the alloys to even greater thermal strength (Ref. 12.5-6).
The bottom diagram illustrates the influence of the orientation of the grain boundaries relative to the direction of stress.
As would be expected, at high temperatures, longitudinal grain boundaries (L) have the longest creep life, relative to transversal (T) and diagonally (D) oriented ones. The fact that the diagonally oriented grain boundaries have the worst life spans is understandable, due to the shear-initiated creep pore formation (Fig. "Creep mechanisms in hot parts"). This is also a quality characteristic of directionally solidified structures.
However, at low operating temperatures the results are completely different. Transversally oriented grain boundaries have the longest life spans. Diagonal grain boundaries are better than longitudinal ones, if only slightly. This behavior indicates that the damage mechanism at lower temperatures is different from that at higher temperatures (Fig. "Creep mechanisms in hot parts").
The table in the top diagram shows that none of the three named structural types is universally advantageous. For example, at smaller grain sizes the multi-axial variant is superior against dynamic stress in the HCF range, but its creep strength leaves much to be desired.
Figure "Reinforcing creep resistanc": Strengthening mechanisms can be used to increase the high-temperature strength of a material, i.e. reduce its creep strain.
Precipitation hardening: This strengthening method is used with Ni-based superalloys. The cobblestone-like g'-phase consisting of Ni3 Al (left diagram) will dissolve at temperatures around 1050°C and is secreted at a temperature range around 800°C. It is resistant to creep. At higher levels of Al content (about 5%) the alloys can no longer be forged due to their minimal creep deformability. For this reason, only cast alloys have such high Al content, making them highly creep-resistant.
Dispersion hardening: A fine, even distribution of hard particles in the material prevents creep (second diagram from left). These particles, which are usually ceramic, are applied through powder-metallurgy, since the varying thicknesses in the liquid (casting) state and the solidification process result in segregation of materials. This type of alloy (e.g. MA 6000, Fig. "Creep resistance of Ni-based alloys") is relatively brittle, expensive, and very difficult to machine due to the abrasive effect of the ceramic particles. Dispersion-hardened materials are rarely used. One area in which they do find use is thermally/mechanically highly-stressed turbine stator vanes in fighter engines.
In situ fiber-reinforcement: These are directionally solidified eutectic alloys (e.g. CoTaC alloys, Ref. 12.5-3). During directional solidification in the casting process, fiber-like carbides form in the direction of solidification (third diagram from left). These alloys have pronounced weaknesses and there have been no reports of their use in serial production. One weakness is the destruction of the fibers under thermal fatigue stress, due to the difference in thermal strain between the fiber and matrix. A further problem is workability. Electrochemical processes are especially problematic due to the different chemical properties of the fibers and matrix.
Fiber-reinforcement with high-temperature resistant wires: There were many attempts to reinforce Ni-based alloys through the insertion of tungsten wires with an HIP process (right diagram). Evidently there has not been any serial application to date. Reasons for this probably include the oxidation sensitivity of tungsten, thermal strain differences between it and the matrix, high costs, and high specific mass of the wires.
Figure "Cracks inside cooled turbine blades": Thermal fatigue cracks are often unavoidable in turbine stator vanes, and are therefore allowed at lengths that are not expected to result in dangerous crack growth that would cause the part to fail. Despite this, it may be necessary to limit the crack lengths beyond the strength requirements. This is the case (top diagrams) if, at least temporarily, the internal pressure in convection-cooled blades is not sufficiently greater than that in the outer gas flow. In this situation, cracks that penetrate the outer wall of the blade present a danger of serious hot gas encroachment into the inner cooling structure.
An additional case is illustrated in the bottom diagrams. If the cooling air is partially or completely used to cool the adjacent downstream turbine blading after it has passed through the inside of the stator blade, even with sufficient cooling air pressure, a gaping crack can result in an unallowably large amount of cooling air being lost. This may be the reason why, in some cases, maintenance manual guidelines for boroscope inspection results allow larger cracks in the stator vanes of front turbine stages than they do in the shrouds behind them.
Figure "Creep effects and part behavior": Creep not only limits the life span of hot parts through unallowable deformations, cracking, and fractures (top left diagram, Ref. 12.5-5). Creep deformations also have damaging effects on many other engine part properties:
Thermal fatigue: In the high temperature zones, compressive stresses arise which are in equilibrium with the tensile stresses in colder part areas. The compressive stresses can become so great that they cause plastic compression (
Dynamic part behavior: Length changes can lower the resonant frequency of the blades. If this is only slightly above an inciting frequency, there is a danger of resonance and dynamic fatigue fractures.
Unwinding of blades that are connected by a shroud (Fig. "Design of rotor blade tip shrouds") can also promote blade vibrations.
The resonant frequency can also change due to rubbing and material removal from the blade tips following creep strain. The rubbing can also have an additional damaging effect through local heat creation and vibration excitement.
High tension residual stresses that have built up through local creep and/or the reduction of protective compressive residual stresses both promote dynamic fatigue fractures. This effect is especially problematic where surface hardening (e.g. shot peening) has increased the dynamic strength (blade roots; in compressor blades, also the blade leaf).
Efficiency: Geometric changes of surfaces in the air flow, such as unwinding of blade leaves or shroud flexure (with the danger of shingling,
On the leaves of coated blades, an increase in roughness can be observed in the form of rippling due to thermal fatigue (Fig. "Coatings sensitive to thermal fatigue").
Coatings: There is an increased risk of cracking due to thermal fatigue in brittle coatings (Fig. "Coatings sensitive to thermal fatigue").
Material behavior: Structural changes also occur in connection with creep stress. Depending on the strength of the creep stress, creep pores can form along the grain boundaries, which are oriented tangentially to the direction of stress. As operating time increases, these pores can experience micro-cracking that eventually leads to creep fractures (Fig. "Pore formation as creep damage"). In single crystals, pore formation of the creep damage does not occur on the grain boundaries. Instead, it originates in cavities, carbides, and eutectics in the interdendritic zones. Under tensile stress at high temperatures, the g'-phase necessary for good creep behavior aligns itself in a preferred crystal direction (<001>) perpendicular to the tensile stress direction, and takes the form of plates in the shape of a raft (rafting, Ref. 12.5-7).
Creep deformation around notches can have a positive effect if it causes a reduction in notch stress peaks. This can increase both the dynamic strength and the creep life at notches.
Rubbing behavior: Lengthening of blades, overlapping of crooked shrouds, and the bending-open of shrouds leads to unpredicted rubbing and altered rubbing behavior. This creates a danger of inciting vibrations and/or a self-reinforcing process through additional heating of the shroud during rubbing.