Interaction and superposition of creep damage and high-cycle fatigue of coarse-grained nickel-base alloy 247

ABSTRACT

The cast coarse-grained polycrystalline Alloy 247 is investigated, which is typically used in the rear turbine blades of land-based gas turbines. High rotational speeds up to 10,000 rpm in combination with high temperatures induce creep deformation. Additionally, gas turbine blades are subjected to high-frequency cyclic stresses. This complex loading state requires the investigation of the interaction of creep damage and high-cycle fatigue (HCF). Therefore, HCF tests with a stress ratio of R_σ = −1 were carried out on as-received and pre-crept specimens. In comparison, an HCF test series with high mean stresses (R_σ = 0.5) was conducted to consider the superposition of creep and fatigue. Since nickel-base alloys exhibit a pronounced elastic anisotropy, the grain orientations were considered in the damage analysis. The results show that creep-induced grain boundary damage is a dominant structural parameter for the HCF behaviour of Alloy 247 in both, sequential and superpositioned creep-fatigue loading.

KEYWORDS:

• Creep
• creep-fatigue
• HCF
• pores
• cavitation
• EBSD
• Ni-base
• superalloy

Introduction

Nickel-base superalloys feature an excellent combination of material properties at high temperatures as microstructural stability, strength, ductility, fracture toughness even in long-term use, creep and corrosion resistance as well as good thermomechanical fatigue properties [1–3]. Therefore, these alloys are an outstanding material class essential to aerospace, automotive and power generation industries. Due to growing share of volatile renewable energy supply, innovative land-based gas turbines are primarily used to compensate the deficits between energy demand and generation which is associated with frequent start-stop processes. This results in a changed requirement profile for the components. Nickel-base superalloys are used as blade material in gas turbines, whereby polycrystalline blades are mainly used in the rear stages due to the significantly lower manufacturing costs. The grain boundaries in these components, especially those perpendicular to the loading direction, exhibit creep damage due to high centrifugal forces caused by high rotational speeds at high temperatures [4,5]. In addition, both inhomogeneous flow fields behind the guide vanes and vibration excitations during start-stop processes lead to high cycle fatigue (HCF) loadings. This requirement profile necessitates the investigation of the interaction of creep damage on the HCF behaviour. While creep and HCF are superimposed during continuous operation of a gas turbine, frequent start-stop processes cause the components to experience increased vibration excitation when creep-induced damage is already present. Therefore, two different series of tests were conducted in this study, representing both continuous operation (R_σ = 0.5) and start-stop processes (R_σ = −1) at different creep pre-strain levels.

Material and experimental details

The specimens used in this study were all eroded and finished from individually cast coarse-grained Alloy 247 LC CC plates. Table 1 shows the chemical composition of this nickel-base superalloy. Comparing this alloy with other nickel-base superalloys, the high content of W leads to a high density, resulting in an increase in mass and, hence, centrifugal forces in the turbomachinery application, which are mainly responsible for creep loading [6–8]. Worth mentioning is the high content of Al and Ta and the low content of Ti as γ’-forming and stabilising elements [9,10]. This alloying concept indicates a relatively strong coarsening inhibition of the γ’-precipitation structure.

Table 1. Chemical composition of tested nickel-base alloy 247 (wt%).

Ni	Cr	Co	Ti	Mo	W	Al	C	B	Zr	Ta	Hf
bal.	8.25	9.30	0.71	0.49	9.58	5.69	0.08	0.015	0.01	3.19	1.45

Since this study is concerned with the interaction and superposition of creep damage and HCF, typical fatigue specimens are used, of which the geometry is shown in Figure 1. The specimens have a total length of 140 mm with a shaft diameter of 12 mm and a measuring section length of 20 mm with a diameter of 7 mm. To avoid premature surface cracks during the fatigue tests, the measuring sections of the specimens were polished.

Figure 1. Specimen geometry.

The individual casting of the specimen plates results in a relatively homogeneous microstructure within the plate material with average grain diameters d ~2 mm. Figure 2 shows optical micrograph images of polished and subsequently etched cross and longitudinal sections which illustrate that the alloy segregates dendritically.

Figure 2. Optical microscope images of alloy 247 plate material: (a) cross-section (b) longitudinal section.

For the description of the influence of creep damage on the HCF behaviour, it is necessary to detect the occurring damages and microstructural changes during creep stress. Therefore, scanning electron micrograph (SEM) investigations were carried out in the as-received state and after pre-straining. In addition, fractographic analysis was carried out to determine the crack initiation sites and the defects which led to crack initiation. Electron backscatter diffraction (EBSD) analyses on longitudinal sections through the crack initiation point were performed on the upper and lower fracture halves of the specimens to evaluate the influence of grain orientation on crack initiation behaviour and the crack propagation path towards trans- or intercrystalline crack growth.

Creep testing

Due to the coarse-grained microstructure of the test material, the relatively large test section diameter of 7 mm was selected for creep tests to ensure a sufficient number of grains within the measuring section volume, which leads to high required tensile forces. The isothermal stress-controlled creep tests for specimen pre-straining are hence carried out on a pneumatic creep test rig developed at the institute, which allows tensile forces up to 30 kN and specimen temperatures above 1000°C by using an inductive heating system. The creep strains are continuously measured during the tests by a capacitive high-temperature extensometer and by a capacitive distance measurement with a measurement accuracy of about 700 nm. The reduction of the cross-sectional area due to increasing creep strain is calculated by the measurement software into account to enable stress control. Therefore, assuming a constant measurement section volume and a remaining cylindrical measurement section shape during the test, the reduction in cross-sectional area as creep strain progresses leads to an adjustment in the test load. That allows a detailed determination of the transition between the secondary and tertiary creep regime. All pre-strained specimens were loaded with σ_n/σ_UTS between 0.14 and 0.21 in short-time creep tests below 1000 h at 900°C. At these nominal stresses and since the Norton exponents are n > 1.5, the deformation mechanism of dislocation creep dominates. This indicates that the dislocation movement and thus the occurring shear stresses have a high impact on arising creep damage. The test rig has three shutdown criteria: time, creep strain and strain rate. The strain rate shutdown is used to prevent the specimen from being stressed far into the tertiary creep regime, which would have fatal consequences for the subsequent fatigue behaviour due to excessive, and for practical application unrealistic, creep damage. Figure 3 shows an exemplary 454 h creep test that was shut down because of entering the tertiary creep regime, and the associated increasing creep strain rate.

Figure 3. Creep test on the pneumatic creep test rig. Black: plastic creep strain in %. Red: temperature in °C. Blue: normalized tensile stress.

Creep damage mainly affects grain boundaries orientated perpendicular to the loading direction in the form of diffusion-controlled processes such as pore formation, growth, and coalescence [11]. High nominal stresses can also additionally result in tearing of grain boundary triple points, and grain boundary sliding [11,12].

High-temperature ageing, as in creep tests, generally results in coarsening of the precipitate structure. The finely distributed coherent γ’-precipitate structure embedded in the γ solid solution is one of the main factors responsible for the high strengths of this material class at high temperatures [13,14]. In creep tests, the γ’-precipitates mostly coarsen perpendicular to the loading direction, so-called rafting, since in uniaxial loading conditions the total strain energy transverse to the loading direction is higher than that in the loading direction [15]. The creep-induced damages and microstructural changes are schematically shown in Figure 4.

Figure 4. Schematic change of microstructure by pore nucleation, growth and coalescence on grain boundaries, tearing of grain boundary triple points and rafting as a result of creep stress [16].

Figure 5 illustrates an additional mechanism of creep damage, where cavities are generated around precipitates or particles through carbide decohesion at grain boundaries, attributed to grain boundary sliding [12].

Figure 5. Schematic pore nucleation at a precipitate by grain boundary sliding [12].

High-cycle fatigue testing

All fatigue tests were carried out on a servo-hydraulic fatigue testing system from MTS. An inductive heating system is also used on this test setup together with closed-loop control of the specimen temperature. A clamping system identical to that in the creep test rig is used. Two different series of tests on the plate material were carried out. First, the influence of creep-induced pre-strain on the HCF behaviour was investigated by fatigue of as-received and pre-damaged specimens at a stress ratio R_σ = −1 and different stress amplitudes. Second, as-received specimens are loaded at a stress ratio R_σ = 0.5. In this case, a superposition of creep damage and fatigue exists. While high mean stresses cause creep damage, the fatigue phenomena are caused by the superimposed cyclic stresses. The experiments in both test series are stress-controlled constant amplitude tests (CATs) at a temperature of 900°C and a frequency of 10 Hz.

Results and discussion

Creep damage

To investigate the influence of creep damage on the HCF behaviour, it is necessary to detect the creep-induced microstructural changes that occur during the pre-straining.

The following Figures 6 and 7, which show exemplary SEM images taken after pre-straining the specimens on the pneumatic creep test rig, verify the assumed microstructural damages, i.e. pore formation, growth, and coalescence, tearing of grain boundary triple points, and γ’-rafting due to creep stress. Figure 6a shows a longitudinal section with pores on grain boundaries lying approximately perpendicular to the loading direction at a specimen with 1.0% plastic creep strain. The pores are predominantly formed near carbide precipitates. It is assumed that the pores nucleate at carbides, and the grain boundary damage progresses through growth and coalescence of the individual pores. SEM examinations of Alloy 247 creep fracture surfaces, as shown in Figure 6b, illustrate that grain boundaries and grain boundary triple points can tear due to high applied creep stresses. Creep-induced porosity is also visible, especially near the triple points.

Figure 6. SEM image of (a) grain boundary damage by pore nucleation, growth, and coalescence at a specimen with 1.0% creep strain; (b) a teared grain boundary triple point and nucleated pores at a creep fracture surface [16].

Figure 7. SEM of (a) cube-shaped γ’-precipitate structure before creep loading; (b) elongated γ’-precipitate structure perpendicular to the loading direction (rafting) after 527 h at 900°C under tensile load.

Rafting has a negative effect on fatigue life of single-crystal nickel-base alloys, as proven, e.g. by Yongsheng et al. [17]. Dislocation movement is facilitated by the coarsening of the material, which highly favours the formation of persistent slip bands. In single crystal alloys, crack propagation can only be transcrystalline, which is why crack propagation is promoted when the precipitation structure is rafted [17]. The effect of rafting on fatigue life in polycrystalline nickel-based specimens is investigated in this study, mainly using the crack propagation path as the evaluation basis.

Figure 7a shows the γ- γ’ structure of the as-received material, where finely distributed cube-shaped γ’precipitates are evident. In Figure 7b there is a coarsened precipitate structure after a 454-h creep test at 900°C with a loading of σ_n/σ_UTS = 0.14. As expected, an elongated (raft-like) γ’-structure develops, which is perpendicular to the loading direction.

Interaction of creep and HCF – stress ratio R_σ = −1

To investigate the interaction of creep-induced pre-strain and HCF behaviour, a constant amplitude test series with a stress ratio R_σ = −1 was performed. For this purpose, specimens in the as-received condition and specimens that were pre-strained on the creep test rig were examined. The pre-strain is quantified by the mean percentage of plastic strain over the specimen measurement section. The S-N_f diagram in Figure 8 illustrates that pre-strain leads to a significant reduction in the HCF life. Since the creep strain distribution in the material is highly inhomogeneous due to the coarse-grained microstructure with different grain orientations and the elastically anisotropic material behaviour, the measured mean values of the creep strain do not reveal the level of local damage in the material. Nevertheless, it is clearly recognisable that an increase in averaged creep strain leads to a reduction in fatigue life.

Figure 8. S-N_f plot of as-received and pre-strained alloy 247 specimens at 900°C, a frequency of 10 hz and a stress ratio R_σ = 1. [plastic creep strain in %].

When observing the as-received specimens, it is noticeable that the test results show a strong scatter in fatigue life. This is most evident for the three as-received specimens that were tested at σ_a/σ_UTS of around 0.25. Here, the fatigue life of the undamaged specimen is shorter than that of the pre-damaged specimens with 0.75% plastic creep strain and only reaches about 12% of the lifetime of the as-received specimen with the maximum service life. Also, in the tests at σ_a/σ_UTS of around 0.29, the as-received specimen with the shortest lifetime only endures about 23% of the highest lifetime specimen. Nevertheless, this big scatter is typical for coarse-grained nickel-base alloys and is caused by the elastic anisotropic behaviour. The root cause of this scatter is highly inhomogeneous stress distributions in the individual grains, because of their different orientations. In the formation of local stress peaks the different orientations of the grains which determine both the Young’s modulus E in loading direction and the Schmid factor m, play a decisive role, as shown by Engel et al. [18]. The Schmid factor m considers the influence of grain orientation on the shear stress occurring in the slip systems of the fcc Ni crystal structure.

Despite this high scatter, the S-N_f diagram shows that the fatigue life of the material decreases significantly with increasing creep strain. The shortened fatigue life is most evident in the tests at σ_a/σ_UTS of around 0.27, although here the specimen with 0.5% pre-strain has a fatigue life that is almost a factor of 2 shorter than the specimen with 1.0% pre-strain. These deviations can be explained both by the scatter in fatigue of the material described above, and by the differences in the local creep damage at identical global creep strain applied to the specimen.

Superposition of creep and HCF – stress ratio R_σ  = 0.5

Since the creep-induced damage of the specimens should dominate in this test series, a high stress ratio was chosen. During the tests, creep and fatigue damage are superimposed, whereas the mean stress causes creep and the cyclic stress causes fatigue damage. This test series is also used to understand the influence of defects, especially porosity, on the cracking behaviour, as the pores nucleate, grow, and coalesce while the test progresses. Therefore, a critical defect size can be assumed at the time of fatigue crack initiation. The slope in the S-N_f data observed at R_σ = 0.5 (Figure 9) is clearly flatter than that for R_σ = −1. The tests between σ_a/σ_UTS of 0.05 to 0.08 with associated σ_m/σ_UTS between 0.15 and 0.24 do not fail until the ultimate number of load cycles of 10⁷, whereby a near-time failure could be predicted for the test at σ_a/σ_UTS = 0.08, which is why this test was continued until failure at around 1.02∙10⁷ cycles. The tests conducted at σ_a/σ_UTS around 0.086 and σ_m/σ_UTS around 0.026 show a high scatter in the fatigue life.

Figure 9. S-N_f plot of as-received alloy 247 specimens at 900°C, a frequency of 10 hz and a stress ratio R_σ = 0.5.

Even with this stress ratio, the test series implies a high fatigue life scatter, since the prematurely failed specimen at σ_a/σ_UTS around 0.086 and σ_m/σ_UTS around 0.026 reached only 20% of the service life of the longer test.

Fracture surface analysis

SEM fracture surface analyses were performed to investigate the influence of creep damage on the fatigue cracking behaviour in the HCF regime in more detail. The test series on as-received and pre-strained specimens at R_σ = −1 shows a clear difference in the fracture surfaces. Figure 10a illustrates the fracture surface of an as-received specimen loaded at σ_a/σ_UTS of around 0.29. A single fatigue crack can be seen that initiated in the material volume at a grain boundary. The fracture surface indicates different crack propagation rates. In the closer area around the crack initiation site, a smooth fatigue fracture surface develops (marked blue), which indicates slow crack propagation. Between the blue and red markings there is the fast propagation zone, because of increasing stress intensity at the crack tip with increasing crack length. The final rupture occurs abruptly and can be observed outside the fast propagation zone. As mentioned above, for this specimen, a single crack initiation at a grain boundary without a visualisable defect was observed which is shown in higher magnification in Figure 10b. Crack initiation at a grain boundary in the volume without a defect suggests a crack due to stress concentrations, as already described by Engel et al. [18]. There it was proven that in coarse-grained nickel-base alloys, the grain orientations play a crucial role in the cracking behaviour. The different grain orientations and the associated different Young’s moduli E and Schmid factors m lead to high stress concentrations at the grain boundary, where the product of E and m in the adjacent grains show high variations. Thus, crack initiation is favoured at such grain boundaries.

Figure 10. Fracture surface in case of pure fatigue stress: R_σ = −1, T = 900°C, σ_a/σ_UTS ~ 0.29, N_f = 254.993, ε_p = 0%.

When observing the fracture surface of a pre-strained (ε_p = 1,0%) specimen loaded at σ_a/σ_UTS of around 0.25, as shown in Figure 11a, it is obvious that the fracture pattern differs significantly compared to undamaged specimens. At this fracture surface, the localisation of the incipient cracks is much more difficult. Fissured areas develop in regions of advanced creep damage, wherein multiple cracks usually originate. At this fracture surface at least three fatigue crack initiation sites can be identified. Looking at one of these three crack initiation zones (Figure 11b), pores have formed on the grain boundaries as a result of the previous creep test. The marked pores have a maximum length of about 30 µm and are considered responsible for the crack initiation. Similar grain boundary damage like in this SEM image is also evident at the crack initiation zones of the other pre-strained specimens.

Figure 11. Fracture surface under fatigue loading of a pre-strained specimen: R_σ = −1, T = 900°C, σ_a/σ_UTS ~ 0.25, N_f = 330.174, ε_p = 1.0%.

The fracture surfaces of the fatigue test series with R_σ = 0.5 differ fundamentally from those already presented. Figure 12a shows a fracture surface of a specimen loaded with a relatively low amplitude and mean stress (σ_a/σ_UTS ~0.086, σ_m/σ_UTS ~0.26). Here, the ratio of the fatigue fracture area of the total fracture area is significant smaller and no distinction between slow and fast propagation zones is possible. Also in these specimens, creep damage caused by the mean stress leads to multiple cracking. The areas of crack initiation, as shown exemplarily in Figure 12b, contain several pores. Therefore, these pores seem to play a significant role in crack initiation. Moreover, much of the remaining fracture surface shows clear indications of creep damage, as shown in Figure 12c: Typical creep combs with internal pores and carbides surrounded by voids can be observed, suggesting the already mentioned creep mechanism of grain boundary sliding.

Figure 12. Fracture surface in case of superposition of creep and fatigue loading: R_σ = 0.5, T = 900°C, σ_a/σ_UTS ~ 0.086, σ_m/σ_UTS ~ 0.26, N_f = 7.551.429, t = 209,76 h.

This fracture surface shows that creep damage is dominant in the superposition of creep and HCF at low stresses. The longer test duration at low stresses, here around 210 h, allows the mean stress to be effective for a relatively long time, which generally leads to progressive creep damage. Low stress amplitudes are only able to form and propagate fatigue cracks at sufficiently high defect sizes, as described by Kitagawa and Takahashi [19]. This results in late crack initiation at low stress amplitudes, after widespread grain boundary damage due to pore formation, which leads to the observed creep-dominated fracture surfaces.

For tests at R_σ = 0.5 with higher stresses σ_a/σ_UTS ~0.093, σ_m/σ_UTS ~0.28 and thus shorter test times, the fracture surfaces of the specimens change significantly, as can be seen in Figure 13a. The shown specimen failed after about 52 h of testing. Here, the portion of the fatigue fracture area in the total fracture area increases significantly, although several crack initiation sites can usually be localised as well. The crack initiation zone in Figure 13b evidences that high mean stresses result in tearing of individual grain boundaries and/or grain boundary triple points. At the same time, as described by Kitagawa and Takahashi [19], smaller defects cause failure at higher stress amplitudes, which is why early crack initiation occurs at the teared grain boundaries before advanced creep damage can generally occur in the material due to pore formation, growth, and coalescence. As a result, these fracture surfaces significantly have fewer creep-like fracture zones.

Figure 13. Fracture surface in case of superposition of creep and fatigue loading: R_σ = 0.5, T = 900°C, σ_a/σ_UTS ~ 0.093, σ_m/σ_UTS ~ 0.28, N_f = 1.888.209, t = 52,45 h.

Electron backscatter diffraction

To examine the influence of grain orientations on the creep damage and cracking behaviour, EBSD analyses were carried out on longitudinal sections through the crack initiation point on both fracture surfaces. These analyses also allow the determination of the crack propagation path depending on the grain boundary angle with respect to the loading direction. These issues are illustrated by two exemplary EBSD investigations of a pre-strained (ε_p = 1,0%) sample loaded at R_σ = −1 and an as-received specimen fatigued at R_σ = 0.5 with low mean stress and stress amplitude. The analysis shown in Figure 14a corresponds to Figures 11 and 14b to Figure 12.

Figure 14. Longitudinal electron backscatter diffraction analysis of: a) a pre-strained R_σ = −1 specimen: σ_a/σ_UTS ~ 0.25, N_f = 330.174, ε_p = 1,0%; b) a R_σ = 0.5 specimen: σ_a/σ_UTS ~ 0.086, σ_m/σ_UTS ~ 0.26, N_f = 7.551.429, t = 209,76 h; marked red: crack initiation zone; [Young’s modulus E in GPa – schmid factor m – E∙m in GPa].

The pre-strained specimen from Figure 14a cracked in an area with a grain boundary oriented almost 90° to the loading direction and a grain boundary quadruple point. The involved grains show significant differences both in their stiffnesses and in their E∙m values. On the upper half of the fracture, the large, extremely stiff grain #1 with a Young’s modulus E₁ = 223 GPa and the small, soft grain #2 with E₂ = 92 GPa are involved in the cracking. On the lower half of the fracture, the relatively stiff grain #4 with E₄ = 180 GPa with a simultaneously high Schmidfactor m₄ = 0.45 and the soft grain #3 with E3 = 102 GPa, also with a high m value of 0.48, are involved in the cracking. Based on this constellation of high differences in the E∙m values high normal and shear stresses must have occurred in the mentioned grains during the previous creep test. In addition, it must be mentioned that especially the lower fracture half has a soft grain periphery, which leads to high shear stresses in stiff grains in soft periphery due to the interaction of the individual grains, as already proven by Lion et al. [20]. This indicates high shear stresses in grain #4, as it is a stiff grain with sliding systems favourably oriented to the loading direction in a soft grain periphery. These high shear stresses have a strong influence on the creep deformation by dislocation movement as sliding and climbing [21]. This combination of high normal and shear stresses leads to high local creep-induced preliminary damage in this region. In the subsequent fatigue test, stress peaks also occur in this heavily pre-damaged area, which finally lead to fatigue cracking.

Initially, the grain boundaries in the vicinity of the crack initiation point are directed nearly perpendicular to the direction of loading. The fatigue crack therefore propagates in both directions across the grain boundaries damaged from the preliminary creep test. Between the grains #5 and #6 there is a grain boundary with an angle of about 60° to the loading direction, followed by a large area with a 90° grain boundary where the crack propagates. Far left of the crack initiation site, the crack progresses along a 45° grain boundary of grain #7 until this grain boundary ends. At this point, transcrystalline crack growth occurs through grain #7. This trend is confirmed in the EBSD analyses of the remaining pre-strained specimens: Crack initiation mostly occurs at 90° grain boundaries, or at grain boundary triple points with grains that have high differences in their stiffnesses and E∙m values in the loading direction. The crack propagates from there over the damaged grain boundaries that have an angle >45° to the loading direction. As soon as the crack arrives at a grain boundary <45° inclined against the loading axis, the crack propagates transcrystalline until a favourably located grain boundary is reached again. The crack propagation in the pre-damaged specimens is thus strongly dependent on the grain boundary angle with respect to the loading direction. This is assumed since the creep damage mainly develops at grain boundaries perpendicular to the direction of loading and decreases sharply on grain boundaries with an orientation angle <45°. This assumption is also confirmed via SEM images of the grain boundaries on the longitudinal sections, as shown in Figure 15. Here, it can be seen that grain boundaries with an angle >45° (marked red) show advanced grain boundary damage due to pore formation, growth, and coalescence, whereas grain boundaries <45° (marked blue) show barely any, or no creep damage at all.

Figure 15. SEM of a longitudinal section of a pre-strained specimen with marked grain boundaries and the angle to the loading direction: blue) undamaged grain boundaries; red) grain boundaries with creep damage; R_σ = −1, σ_a/σ_UTS ~ 0.25, N_f = 330.174, ε_p = 1,0%.

Since crack propagation mainly takes place along the grain boundaries and transcrystalline crack growth only occurs in isolated cases, it can be assumed that the γ’-precipitation rafting has only a minor influence on the fatigue life in this polycrystalline material. However, during the phases of transcrystalline crack growth, rafting will lead to an increased crack propagation rate, similar to single-crystalline alloys [17], but this is likely to be negligible in the case of sufficiently creep-damaged polycrystalline alloys.

Considering the EBSD image of an as-received specimen fatigued at R_σ = 0.5 in Figure 14b, the fracture surface is more fissured than in the pre-strained specimens. This is consistent with the fracture surface analyses of the specimens at low mean stresses and stress amplitudes, as the share of creep fracture surface with respect to the specimen’s cross-section s is significantly higher. Crack initiation in this specimen occurred at a triple point between the relatively stiff grains #1 and #3, with Young’s moduli E₁ = 167 GPa and E₃ = 172 GPa, and the relatively soft grain #2 with E₂ = 143 GPa. Due to the two large soft grains #2 and #4 with E₄ = 139 GPa, the stiff grains #1 and #3 are in a soft periphery. This also results in high shear stresses in the stiff grains, with grain #1 having a high m₁ = 0.48, which further increases this shear stress exaggeration. Therefore, it is assumed that during the test there are high stress concentrations, which are responsible for locally increased formation of creep damage. This will support fatigue crack initiation at the grain boundary triple point. Since only a minor proportion of the fracture surface can be attributed to fatigue, a tracing of the fatigue crack propagation path is not expedient here.

Conclusions & outlook

This section summarises the results of the R_σ = −1 test series carried out on as-received and creep pre-strained specimens to consider the interaction of creep and HCF, the R_σ = 0.5 test series to consider the superposition of creep and HCF and the further investigations such as fracture surface analyses and EBSD analyses. The focus is on the influence of creep damage on the fatigue behaviour, crack initiation and crack propagation path of the coarse-grained polycrystalline nickel-base Alloy 247. The summary of the results is followed by a brief outlook on further investigations at the institute.

(1): Creep damage

• At high nominal, or mean stresses, tearing of grain boundaries and grain boundary triple points are more frequently detected.

• At low nominal or mean stresses, grain boundary damage due to pore formation, growth and coalescence is detected more frequently.

• Pore formation around carbides occurs in case of grain boundary sliding.

• γ’-rafting is observed in the creep loaded specimens.

(2): Interaction of creep and HCF: R_σ = −1

• A high scatter of fatigue life is seen in the as-received specimens due to local stress concentrations because of different grain orientations.

• Creep induced pre-strain leads to a significant reduction of fatigue life.

(3): Superposition of creep and HCF: R_σ = 0.5

• A flatter slope of the S-N_f data is observed compared to R_σ = −1.

• The portion of fatigue fracture area in fracture area increases with increasing mean stresses and stress amplitudes due to rupture of grain boundaries and grain boundary triple points which leads to early fatigue crack initiation while the high R ratio promotes fatigue crack growth.

• At low mean stresses and stress amplitudes, creep-induced damage becomes more dominant, as this diffusion-controlled process has more time to become effective. Hence, at the time of fatigue crack initiation, the grain boundaries are already damaged in a wider area.

(4): Crack initiation

• Creep-induced damage favours multiple cracking due to the accumulation of failure-relevant defects.

• Cracks favourably initiate in creep-damaged specimens at 90° grain boundaries and at grain boundary triple points.

• High differences in E∙m lead to local stress concentrations, resulting in both increased local creep and fatigue damage.

(5): Crack propagation path

• Crack propagation in creep-damaged specimens mainly occurs along grain boundaries with angles >45° to the loading direction, as these grain boundaries show the most pronounced creep damage.

• Rafting has no or only a slight influence on the service life, as mainly intercrystalline crack propagation occurs.

Subsequent studies will examine the influence of temperature on the superposition of creep and HCF, where the temperature is lowered to reduce the creep dominance of the R_σ = 0.5 test series. In addition, the influence of the grain orientations regarding Young’s modulus E, the Schmid factor m and their product E∙m is currently being simulatively investigated to gain a more detailed insight into the occurring shear and normal stresses with consideration of the grain boundary angle.

Acknowledgments

The investigations were conducted as part of the OptiSysKom joint research program of the AG Turbo. The work was supported by the Bundesministerium für Wirtschaft und Klimaschutz (BMWK) as per resolution of the German Federal Parliament under grant number 03EE5035F. We would like to thank our project partner Siemens Energy AG, especially Dr. Dirk Kulawinski, Lucas Mäde and Christian Amman for stimulating discussions and helpful suggestions.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The work was supported by the Bundesministerium für Wirtschaft und Klimaschutz, Germany [03EE5035F].