2021 Volume 61 Issue 10 Pages 2666-2676
Cyclic fatigue, dwell fatigue and crack growth properties were evaluated in the axial direction (L) and transversal direction (T) of Ti-6Al-4V forged round bar. In the SN curve where the stress is normalized by 0.2% proof stress, the cyclic fatigue life in the L/T direction is almost the same, whereas the dwell fatigue life in the T direction is as short as 1/5. In dwell fatigue, ductile fracture occurred when the maximum stress was higher than 95% of 0.2% proof stress. At stresses below 870 MPa, the inelastic strain range and the strain increase rate in the T direction gradually decreased with decreasing stress, and the fracture mode transitioned to that with fatigue crack growth. The gradual change must have been caused by the mixture of anisotropic microtexture regions. At stresses below 825 MPa, the fracture mode transitioned rapidly in the L direction, where the soft oriented microtexture regions were dominant. In the low ΔK region (≤15 MPa√m), the crack growth rate in the axial direction was about twice that in the radial direction of the bar. The shorter dwell fatigue life in the T direction under stress conditions showing fatigue crack growth was explained by the significantly earlier crack initiation compared to that in cyclic fatigue and the faster crack growth along the microtexture in the axial direction of the bar.
Some titanium alloys used in fan disks of aircraft engines and so on are known that the fatigue life reduced by dwell fatigue at temperatures between room temperature and 200°C or so. Under such dwell fatigue loading, a high stress state continues for a certain period of time. Many studies have been conducted on the influence of microstructures, textures, and microtexture regions (MTR) on crack initiation sites and characteristic facet formation under dwell fatigue loading.1,2) Concerning the facet formation mechanism, Evans and Bache explained it by using a stress redistribution model and the modified Stroh model.3) Taking account of crystallographic plastic anisotropy and time-dependent plasticity, Hasija et al. conducted an analysis. They found that local stress concentration is generated by the load shedding mechanism when crystal grains with different orientations are adjacent. And they also revealed that that local stress concentration can result in crack initiation.4) By using a crystal plastic finite element model incorporating a quasi-cleavage fracture mechanism, Dunne and Rugg revealed that facets are formed by the load shedding mechanism with a loading method by which stress is retained.5) In addition, Sinha et al. revealed that the facet surface observed on the dwell fatigue fracture surface has a tilt of ~10° in relation to the basal plane of the alpha phase.6,7) About characteristic facets, Pilchak and Williams discussed in detail8) and they classified them into initiation facets and propagation facets. And Pilchak revealed that the crack propagation rate increases with facets being formed,9) and that cracks propagate rapidly in a microtexture region and that the size of the region has a great influence on the dwell fatigue life.10) In addition, Lang studied time-dependent crack propagation with stress retained.11) With regard to the influence of textures, Everaerts et al. revealed that when Ti-6Al-4V bars do not contain hard-oriented microtextures, no internal facets are formed even under dwell fatigue loading.12) Concerning the fracture surface morphology and crack propagation rate under dwell fatigue loading, low-cycle fatigue loading, and combined loading, Tympe et al. analyzed in detail. They used unidirectionally rolled plates that have both soft-oriented and hard-oriented microtextures, and reported that under dwell fatigue loading, propagation facets are formed in microtexture regions and the characteristics of the fracture morphology depend on those of the texture.13) In addition, with regard to local strain behavior and crack initiation in Ti-6Al-4V under dwell fatigue loading, reports have been made by Littlewood and Wilkinson,14) Hémery et al.,15) Wang et al.,16) and Lavogiez et al.17)
In order to control industrially dwell fatigue damage, it is important to understand the influences of macrotextures and microtextures, stress conditions, and other factors on fatigue crack initiation and crack propagation behavior, formulate guidelines for material microstructure control, and establish a quantitative life prediction technology. With regard to dwell fatigue life prediction, several reports have been made by Venkatesh et al.18) and Ota et al.19) The authors have studied Ti-6Al-4V forged round bars with a fine equiaxed structure in which textures have been formed. They have revealed the relationships between the stress and fracture life, and fracture surface morphology20,21,22) and the relationship between the stress and room-temperature creep behavior23) under cyclic fatigue loading and dwell fatigue loading with the loads applied in the axial direction and radial direction of the round bar. This article studies differences in strain accumulation behavior between cyclic fatigue and dwell fatigue and the influence of crack propagation behavior on life and discusses the life reduction mechanism and the influence of structural anisotropy on life under dwell fatigue loading.
This study uses Ti-6Al-4V (Ti-6.3Al-4.0V-0.17Fe-0.18O (mass%)) forged round bars. The round bars have an equiaxed structure with an alpha grain size of approximately 8 μm. And they also have a texture which (1010)α has a high density in the axial direction of the round bar and (0001)α and (1120)α have a high density in the radial direction of the round bar (Fig. 1). Tensile test specimens and fatigue test specimens were taken in the axial direction (L direction) and radial direction (T direction) of the round bar. We used tensile test specimens which have a parallel portion with a size of φ5 × 30 mm and a gauge length of 25 mm. The 0.2% proof stress was measured at a strain rate of 8.3 × 10−5 s−1, and the tensile strength was measured at strain rates of 2.5 × 10−3 s−1 (L direction) and 8.3 × 10−5 s−1 (T direction). Table 1 shows the tensile characteristics in each tensile direction.
(a) Inverse pole figure (IPF) map, (b) IPF for longitudinal orientation and (c) IPF for transverse orientation of the Ti-6Al-4V forged bar. (Online version in color.)
| Loading direction | 0.2% proof stress* (MPa) | Tensile Strength (MPa) | Elongation (%) | Reduction of area (%) | Young’s modulus (GPa) |
|---|---|---|---|---|---|
| L | 867 | 954 | 18.0 | 43.1 | 113 |
| T | 913 | 971 | 17.9 | 39.2 | 121 |
For the fatigue test, specimens having a parallel portion with a size of φ5.08 × 15.24 mm and a gauge length of 12 mm were used. The fatigue tests were carried out under stress control with a stress ratio of 0.05, where the maximum stress was 805 to 870 MPa in the L direction and 850 to 915 MPa in the T direction (both are approximately 93 to 100% of the 0.2% proof stress). With regard to stress waveforms, a triangle wave with 1 s loading and 1 s unloading cycles was used for cyclic fatigue, and a trapezoidal wave with 1 s loading, 120 s retention, and 1 s unloading cycles was used for dwell fatigue. For the crack propagation test, compact CT specimens of B = 4.5 mm, W = 18.0 mm, and a = 4.5 mm were taken in the C-L, C-R, and L-R directions. The letter to the left of “-” indicates the loading direction, and the letter to the right of “-” indicates the crack propagation direction. L indicates the axial direction, R indicates the radial direction, and C indicates the circumferential direction. After a fatigue precrack was introduced (approximately 1.0 mm for the K-decreasing test, and approximately 1.6 mm for the K-increasing test), the K-decreasing test was performed by load reduction method with a stress ratio of 0.1 and a sine wave of 20 Hz. After that, the K-increasing test was performed by load range fixed method with a sine wave of 3 to 10 Hz. The crack length was measured by the unloading elastic compliance method. All the tests were performed at room temperature. The fracture surface of each specimen was observed by SEM (IT-300, manufactured by JEOL Ltd.), and the fracture surface and cross section of each specimen were measured by SEM-EBSD (JSM-7001F, manufactured by JEOL Ltd. and EBSD detector, manufactured by EDAX) and analyzed by OIM-Analysis (manufactured by TSL Solutions K. K.).
Figure 2 shows the relationship between the maximum stress σmax and the number of cycles to failure Nf under cyclic fatigue loading and dwell fatigue loading. Figure 3 shows the relationship between the maximum stress normalized by the 0.2% proof stress σmax/σ0.2 (hereinafter, the normalized stress, represented in percentage (%)) and the number of cycles to failure Nf. Under dwell fatigue loading, the fatigue life is much shorter than that under cyclic fatigue loading. The ratio of cyclic fatigue life to dwell fatigue life (dwell debit) at 850 MPa is 4.7 in the L direction and 9.3 in the T direction, and it is higher in the T direction than in the L direction. When the normalized stress is the same, under cyclic fatigue loading, the fatigue life is almost the same in the L and T directions. But under dwell fatigue loading, the fatigue life in the T direction is about one-fifth that in the L direction and the dwell debit in the T direction is higher.
Relationship between the maximum stress and the number of cycles to failure. (Online version in color.)
Relationship between the maximum stress normalized by 0.2% proof stress and the number of cycles to failure.22) (Online version in color.)
Figure 4 shows the changes of the maximum strain in each cycle as the number of cycle increases in the fatigue test. Under cyclic fatigue loading, the increase timing and peak value of the strain differ depending on the stress but the behavior is almost the same in the L and T directions. In a high stress region, the strain starts to increase around 200 cycles and is 3 to 4% at the time of fracture. Under dwell fatigue loading, the influence of ratchet strain is greater. In a high stress region with a normalized stress of 98% or more (L direction: 850 MPa or more, T direction: 895 MPa or more), the strain at the time of fracture exceeds 15% in the L direction and 10% in the T direction. The following sections discuss such behavioral differences from the perspectives of strain change behavior (Section 3.2), crack propagation (Section 3.3), and structural anisotropy (Section 3.4).
Strain behaviors during fatigue tests for (a) L-direction and (b) T-direction. The numbers in the legend represent the maximum stress (MPa)/the ratio of maximum stress to 0.2% proof stress (%). (Online version in color.)
As a result of fracture surface observation, it was found that in a high stress region with a normalized stress of 98% or more, under cyclic loading, common fatigue fracture surfaces initiating near surface were observed but under dwell fatigue loading, dimple fracture surfaces were observed over the entire dwell fatigue fracture surface. When the normalized stress was 95% (L direction: 825 MPa, T direction: 870 MPa), fatigue fracture surfaces initiating near surface were observed under both cyclic fatigue loading and dwell fatigue loading (Fig. 5). Under cyclic fatigue loading, both in the L direction (Fig. 5(a)) and T direction (Fig. 5(b)), multiple facets were observed within about 100 μm from the surface. Each facet was almost as large as one alpha grain (approximately 8 μm). Under dwell fatigue loading, the initiation sites were a little unclear and formed in an area larger than under cyclic fatigue loading. The reason the initiation sites were formed near the surface in the L direction (Fig. 5(c)) is that there are textures where coarse internal facets are unlikely to be formed. It is same as a result of fracture surface observation by Everaerts et al. with Ti-6Al-4V bars.12) In the T direction (Fig. 5(d)), initiation facet which is almost as large as one alpha grain (approximately 8 μm) and has no specific patterns on the surface, and propagation facets with a river pattern (ridge) were observed. In addition, propagation facets formed from multiple initiation facets were observed.22) A propagation facet formed about 200 μm from the surface was closely examined by EBSD, and it was found that the facet surface was almost parallel to the basal (0001) plane of the alpha grain (Fig. 6).
SEM images of fracture surface for (a) cyclic fatigue for L-direction at σmax = 825 MPa, 95% of σ0.2, (b) cyclic fatigue for T-direction at σmax = 870 MPa, 95% of σ0.2, (c) dwell fatigue for L-direction at σmax = 825 MPa, 95% of σ0.2, (d) dwell fatigue for L-direction at σmax = 870 MPa, 95% of σ0.2.
(a) SEM image of the fracture surface for dwell fatigue for T-direction at σmax = 870 MPa, 95% of σ0.2 and (b) IPF maps for facet area marked by red rectangle in (a), analyzed parallel to loading axis.22)
Figure 7 is the result of analyzing part of the propagation facet in Fig. 6 from another direction. In this analysis, the crystal orientation was analyzed in the elongation direction of the alpha grain. This direction is almost parallel to <1010> and corresponds to the texture orientation of the round bar in the axial direction. The crack propagation direction not only corresponds to the elongation direction of the alpha grain but also almost corresponds to <1010> even when the crack propagates to the neighboring alpha grain.
IPF and IQ maps of the fracture surface for dwell fatigue for T-direction at σmax = 870 MPa, 95% of σ0.2.22) Arrows indicate the crack growth direction.
Using rolled bars where T-texture was formed, Bowen reported that the static fracture toughness in (0001) <1010> is lower than that in other orientations,24) and in this orientation, striation is less likely to be observed on the fracture surface when the crack propagates.25) This result cannot be applied as it is, but there is a possibility that a crack will propagate more easily in the (0001) <1010> orientation.
3.2. Strain Change BehaviorThis section summarizes the fatigue test results by using the relationship between the inelastic strain and fracture life used for analyses in a low-cycle fatigue test under strain control and discusses the influences of stress conditions and material anisotropy on strain accumulation behavior under cyclic fatigue loading and dwell fatigue loading. Here, the inelastic strain range Δεin was obtained based on the difference in the strain at the mean value σmean of the maximum stress and minimum stress in the stress-strain hysteresis loops (equivalent to 52.5% of the maximum stress since the stress ratio is 0.05 in this test) between the time of loading and the time of unloading (Fig. 8). As for the strain, the value obtained immediately before unloading is used as the maximum strain in each cycle, and the increase in the strain between cycles is used as the ratchet strain εr. Looking at the inelastic strain, under cyclic fatigue loading, plastic strain accompanying rapid stress change is dominating, and under dwell fatigue loading, creep strain is superposed while a high stress is retained. When the results are summarized with the inelastic strain range at the middle of fatigue life, as shown in Fig. 9, cyclic fatigue and dwell fatigue have different relationships between the inelastic strain range Δεin and the number of cycles to failure Nf. When the maximum stress is the same, the inelastic strain range under dwell fatigue loading is larger than that under cyclic fatigue loading, and the inelastic strain range in the L direction is larger than that in the T direction (Fig. 10(a)). At the same time, the inelastic strain range at the middle of fatigue life does not depend on the stress conditions and has a correlation with the maximum strain in the corresponding cycles (Fig. 10(b)). Therefore, the analysis was focused on strain change behavior before the middle of fatigue life.
Schematic diagram of stress-strain hysteresis loop showing the inelastic strain range (Δεin) and ratchet strain (εr).
Relationship between the inelastic strain range at the middle of fatigue life and the number of cycles to failure.22) (Online version in color.)
(a) Relationship between the maximum stress and the inelastic stress range (Δεin) at the middle of fatigue life. (b) Relationship between the nominal strain (ε) and Δεin at the middle of fatigue life. (Online version in color.)
Figure 11 shows how the inelastic strain range changes in the fatigue test. Under cyclic fatigue loading, at the early stage of the fatigue life, the value is small. But after around 100 cycles, it exceeds 1 × 10−3% due to cyclic softening. The inelastic strain range starts to increase earlier and has a higher peak value as the stress increases. At the same time, when the stress is the same, the inelastic strain range in the T direction starts to increase later and has a lower peak value than that in the L direction. Under dwell fatigue loading, however, in a high stress region with a normalized stress of 98% or more, the inelastic strain range is initially high at 5 × 10−3% or more, and even in a low stress region, the inelastic strain range starts to increase earlier than that under cyclic fatigue loading (cyclic softening has developed). In addition, when the stress is the same, the inelastic strain range in the T direction is smaller than that in the L direction, just as it is under cyclic fatigue loading.
Inelastic strain range behaviors during fatigue tests for (a) L-direction and (b) T-direction. The numbers in the legend represent the maximum stress (MPa)/the ratio of maximum stress to 0.2% proof stress (%). (Online version in color.)
Then, Fig. 12 shows how the ratchet strain εr changes. Under cyclic fatigue loading, after the number of cycles reaches 10, the ratchet strain stays at or below 2 × 10−3% both in the L and T directions, and starts to decrease after it increases slightly or stays in the stable region. Under dwell fatigue loading, however, in a high stress region with a normalized stress of 98% or more, the ratchet strain is initially high at 1 × 10−1% or more, and it increases again after it reaches the minimum value. The influences of work hardening and the decrease in the cross-sectional area of the specimen were observed as were done in creep deformation. This corresponds to the fact that when the normalized stress is 98% or more, a ductile fracture occurs both in the L and T directions where a dimple fracture surface is observed. Under dwell fatigue loading in a low stress region, the initial ratchet strain decreases as the stress decreases, and it decreases monotonically except at the time of final fracture.
Ratchet strain behaviors during fatigue tests for (a) L-direction and (b) T-direction. The numbers in the legend represent the maximum stress (MPa)/the ratio of maximum stress to 0.2% proof stress (%). (Online version in color.)
Under dwell fatigue loading, inelastic strain range and ratchet strain behavior differ between the L and T directions on occasion of decreasing the stress. In the L direction, they change rapidly when the normalized stress is 95 to 98%. When the normalized stress is 95%, the ratchet strain behavior is close to that under cyclic fatigue loading. In the T direction, however, the inelastic strain range and ratchet strain behavior change slowly when the normalized stress is 93 to 98%. This is presumed that in the T direction, plastic deformation is locally generated when the stress is 850 MPa or more even if it is 95% or less of the 0.2% proof stress.
In addition, the interior of the specimen taken at the middle of fatigue life was examined to study the condition of microcracks, which corresponds to strain accumulation (Fig. 13). Both in the L and T directions in the dwell fatigue test conducted with a maximum stress of 850 MPa, multiple cracks of 5 to 30 μm, which is equivalent in size to one to three alpha grains, were observed in the specimen. It was confirmed that microcracks of such a size were already generated at approximately 25% of the fatigue life in the dwell fatigue test conducted in the T direction with a maximum stress of 850 MPa. Under cyclic fatigue loading, however, microcracks were not observed with the specimen taken at the middle (approximately 50%) of the fatigue life with a maximum stress of 850 MPa.
IPF mars of the longitudinal cross section of the specimens interrupted at the middle of fatigue life during fatigue tests for (a) T-direction at σmax = 850 MPa, 93% of σ0.2 and (b) L-direction at σmax = 850 MPa, 98% of σ0.2. Internal cracks are indicated by circles. (Online version in color.)
From the above, it was found that under dwell fatigue loading, reducing fatigue life is caused by earlier crack initiation and increasing number of cracks. The causes of this phenomenon are as follows; retained stress yields the time-dependent effect of stress-plastic strain response and the creep strain, then accordingly the total dislocation displacement is large even in the early stage of fatigue. And superposing ratchet strain and increasing strain increase rate accelerate the strain accumulation. With regard to the relationship between the inelastic strain range and fatigue life (Fig. 9), the slope differs between cyclic fatigue and dwell fatigue because the influence of ratchet strain on fatigue life increases as the stress increases. With regard to differences between the L and T directions, both under dwell fatigue loading and cyclic fatigue loading, the fatigue life in the T direction is shorter than that in the L direction. This corresponds to the fact that under dwell fatigue loading, the strain at the time of fracture is lower in the T direction (Fig. 4) and the fracture ductility decreases more significantly in the T direction in a high stress region (with a normalized stress of 98% or more) under dwell fatigue loading. Even though the reduction of area at the tensile test is a slight difference of 43% in the L direction and 39% in the T direction. On the other hand, a crack propagation-type fracture occurs under dwell fatigue loading with a low stress and cyclic fatigue loading. Therefore, crack propagation behavior was studied. The results are described in Section 3.3.
3.3. Crack Propagation BehaviorFigure 14 and Table 2 show the results of the crack propagation test. Sub-number 1 indicates the result obtained by K-decreasing test and sub-number 2 indicates the result obtained by K-increasing test. C-R-1 is a reference value because the crack angle is beyond ±20°.
Relationship between the crack propagation rate (da/dN) and the stress intensity factor range (ΔK). The letters in the legend represent the direction normal to the crack plane - the direction of crack extension. Here, L: axis direction, C: circumferential direction and R: radial direction of the bar. (Online version in color.)
| Constant force amplitude tests | K decreasing tests | |||
|---|---|---|---|---|
| C | m | ΔK range for Paris law (MPa | ΔKth (MPa | |
| C-L | 5.95 × 10−13 | 4.31 | 8–32 | 3.85 |
| C-R | 2.05 × 10−13 | 4.59 | 8–33 | 5.40*1 |
| L-R | 3.31 × 10−13 | 4.39 | 8–39 | 4.20 |
In a region with ΔK greater than 15 MPa
In a region with ΔK of 15 MPa
SEM images of fracture surface after crack propagation tests for (a) C-L-2 , (b) C-R-2 and (c) L-R-2.
SEM images of fracture surface after crack propagation tests for (a) C-L-2, (b) C-R-2 and (c) L-R-2 in the region of ΔK between 10 and 12 MPa
On the fracture surfaces with around ΔK of 8 to 10 MPa
Facets observed in the fracture surfaces after crack propagation tests for (a) C-L-2, (b) C-R-2 and (c) L-R-2 in the region of ΔK between 8 and 10 MPa
The following discusses these results in comparison with the fatigue test results while assuming that the loading direction for C-L and C-R corresponds to the T direction in the fatigue test, the loading direction for L-R corresponds to the L direction in the fatigue test, and the crack propagation behavior is the same between the R direction and C direction. On the condition of the stress being 850 MPa, the K value is 4.8 MPa
Many of the facets observed at the crack propagation test were formed by one alpha grain and only a few of them were formed by multiple alpha grains as observed on the dwell fatigue fracture surface. Presuming the cause of this fact, it is concluded that there are not many chances for cracks to penetrate through coarse microtexture regions; in the crack propagation test, the regions where cracks can propagate are limited. Under dwell fatigue loading, as reported by Pilchak, the rate at which cracks propagate after facet formation is higher than the rate at which striation is formed by two orders of magnitude,9) and cracks propagate at high rates in microtexture regions.10) Therefore, under dwell fatigue loading, ratchet strain is generated by the stress controlled loading and the time-dependent effect of inelastic strain response with retained stress, and inelastic strain accumulates over the entire specimen, causing many microcracks. As a result, the weakest areas where cracks propagate most easily, or microtexture regions having their basal plane perpendicular to the stress axis and an area larger than other such regions are prone to appear on the fracture surface. Another possible factor is that facets may have grown due to the influence of retained stress on crack propagation.
3.4. Influence of Microstructural AnisotropyThis section discusses the influence of microstructural anisotropy on crack initiation and crack growth. Based on the observation results of the initiation points of the fracture surfaces and the cross-sectional observation results of the specimens taken at the middle of fatigue life, formation of microcracks, each of which is almost equal in size to one alpha grain (approximately 8 μm), are regarded as drack initiation in this study. The microtextures of the Ti-6Al-4V round bar used in this study elongated in the axial direction of the round bar, and relatively large microtexture regions are present along the axial direction of the round bar (L direction). Therefore, when the load is applied in the L direction, the proportion of microtexture regions (hereinafter, soft regions) is high and the soft regions extend parallel to the axial direction of the specimen. The soft regions consist of so-called soft-oriented grain4) (or weak grain1)), where the direction of the stress axis and basal (0001) plane of the alpha phase are parallel to each other and prismatic <a> slip occurs easily. When the load is applied in the T direction, the proportion of microtexture regions (hereinafter, hard regions) is higher than in the L direction and the hard and soft regions are alternating in the direction perpendicular to the axial direction of the specimen. The hard regions consist of so-called hard-oriented grain (or strong grain), where the direction of stress axis and basal (0001) plane of the alpha phase are perpendicular to each other and <a> slip does not occur easily.
Under cyclic fatigue loading, when the stress is the same, the inelastic strain range in the T direction is smaller than that in the L direction (Fig. 10(a)) and the fatigue life in the T direction is longer than that in the L direction (Fig. 3). When the inelastic strain range is the same, the fatigue life in the T direction is shorter than that in the L direction (Fig. 9). When the ΔK is the same, the crack propagation rate in the crack growth stage in the early stage of the fatigue life in the T direction is higher than that in the L direction (Fig. 14). First of all, concerning the fatigue life, under cyclic fatigue loading, crack initiation is dominating in the T direction. In the T direction, the energy used for plastic deformation is smaller and crack initiation takes longer. As a result, when the stress is the same, the fatigue life in the T direction is longer than that in the L direction. In addition, when the inelastic strain range is the same, the fatigue life in the T direction which the crack propagation rate is higher, is shorter than that in the L direction.
However, under dwell fatigue loading, in a high stress region with a normalized stress of 98% or more, ductile fracture occurs both in the L and T directions. And the fracture life, like the creep fracture life, can be expressed with the Monkman-Grant equation. In this study, the equation uses the strain rate which is obtained based on the stress retention time, and the increase in the strain at the end of stress retention in each cycle.21,22) Figure 12 shows how the ratchet strain changes. From Fig. 12, it can be seen that when the stress is the same, the strain rate in the T direction is lower because of the influence of microstructural anisotropy. This is presumed that in the T direction, due to a high proportion of hard regions, the internal stress is high and the effective stress is low. In Fig. 12, the strain behavior in the L direction (870 MPa) and the T direction (895 MPa) almost overlap with each other, and the difference in the internal stress is estimated to be 25 MPa or so. This value is about half the difference of the 0.2% proof stress obtained in the static tensile test (46 MPa). In addition, the ratchet strain decreases as the stress decreases, and the minimum stress at which the additional ratchet strain appear is estimated to be around 825 MPa based on the strain change behavior in the L direction where the soft region is dominating.
Under dwell fatigue loading, in the T direction, when the stress decreases to 870 MPa or less, the fracture morphology transitions to one accompanied by crack propagation. As a result, at the early stage of fatigue, microcracks (initiation facets) grow at a high rate while forming flat and smooth facet fracture surfaces (propagation facets) along the microtextures in the axial direction of the round bar. In the L direction, at a lower stress of around 825 MPa, the influence of strain accumulation with retained stress decreases and the fracture morphology transitions in the same manner as under cyclic fatigue loading. Next the damage mechanism is researched. In the T direction, where the proportion of hard regions is higher, local inelastic deformation makes the stress change gradually until it drops below 850 MPa (normalized stress 93%). On the other hand, in the L direction, where the proportion of hard regions is lower, the stress changes rapidly around 825 MPa (normalized stress 95%) because additional inelastic deformation is entirely suppressed.
In the relationship between the inelastic strain range and fatigue life (Fig. 9), under the stress conditions that exhibit a fracture morphology accompanied by fatigue crack propagation, the fatigue life is shorter in the T direction, where the crack growth rate is higher, and under the stress conditions that exhibit ductile fracture, the fatigue life is shorter in the T direction, where the fracture ductility is lower. In Fig. 9, there is a correlation between the strain rate and inelastic strain range mediated by the effective stress. As a result, it can look like the strain rate and inelastic strain range can be represented by the same approximate line.
From the above results, following consequence is elicited by analyzing the anisotropy in dwell fatigue life based on a normalized stress (Fig. 3). That is, the main factors that the fatigue life in the T direction is only about one-fifth that in the L direction are local inelastic deformation, the increase in the crack growth rate, and the decrease in the fracture ductility in the T direction. In addition, the anisotropy in the ratio of cyclic fatigue life to dwell fatigue life (dwell debit) can be explained as follows. In the T direction, the crack initiation life under cyclic fatigue loading is longer, and the influence of the decrease in the crack initiation life under dwell fatigue loading becomes greater, and furthermore, under dwell fatigue loading, after microcrack (initiation facet) formation, the cracks grow at a high rate while forming propagation facets along surrounding microtextures, causing the dwell debit to increase. In addition, in the T direction, soft and hard regions are present together and local inelastic deformation occurs even with a low normalized stress, and the normalized stress level, at which the dwell debit becomes 1, decreases.
In order to establish a dwell fatigue life prediction technology, there is a need to study internal stress and take account of time-dependent crack growth. Using Ti-6Al-2Sn-4Zr-2Mo alloys, Lang revealed that when a load is kept constant in the middle of the crack propagation test, the cracks propagate with time.11) Further studies are needed on the influence of microtextures on such behavior.
The cyclic fatigue and dwell fatigue characteristics in the axial direction (L) and radial direction (T) were evaluated with Ti-6Al-4V forged round bars, and the crack propagation test was conducted to study the differences in fatigue damage behavior and the influence of microstructural anisotropy. Below are the findings obtained.
(1) When summarized with a normalized stress (ratio of the maximum stress to the 0.2% proof stress), under cyclic fatigue loading, the fatigue life is almost the same in the L and T directions but under dwell fatigue loading, the fatigue life in the T direction is one-fifth that under cyclic fatigue loading, and the ratio of cyclic fatigue life to dwell fatigue life (dwell debit) is high.
(2) Under dwell fatigue loading, in a high stress region with a normalized stress of 98% or more (L direction: 850 MPa or more, T direction: 895 MPa or more), ductile fracture occurs both in the L and T directions and the fatigue life has a correlation with the strain rate as the creep fracture life does. When summarized with a normalized stress, in the T direction, where the 0.2% proof stress and applied stress are high, the ratchet strain and fracture ductility reduction are high; therefore, the life is shorter than that in the L direction.
(3) Under dwell fatigue loading, in the T direction, as the stress decreases to a normalized stress of 95% (870 MPa) or less, the inelastic strain range and strain increase rate decrease gradually due to local inelastic deformation and the fracture morphology transitions gradually to fatigue fracture. In the L direction, when the normalized stress is around 95% (825 MPa), additional inelastic deformation is entirely suppressed and the initial inelastic strain range and strain increase rate decrease significantly. As a result, the behavior approaches that under cyclic fatigue loading and the fracture morphology transitions rapidly to fatigue fracture.
(4) In the crack propagation test, in a low-ΔK region (ΔK ≤ 15 MPa
(5) Under cyclic fatigue loading and dwell fatigue loading, considering the relationship between the inelastic strain range and fatigue life, it is found the fatigue life in the T direction is shorter than that in the L direction. The primary reasons for this phenomenon are as follows; under the stress conditions where ductile fracture is caused by dwell fatigue, the fracture ductility decreases more significantly in the T direction. And under the stress conditions that exhibit a fracture morphology accompanied by crack propagation, cracks initiate earlier both in the L and T directions. In parallel with this, in the T direction, cracks grow along microtextures extending in the axial direction of the round bar, causing the crack propagation rate to increase at an early stage of fatigue.
(6) The fatigue life is decreased by dwell fatigue at a stress of around 825 MPa or more because of strain change behavior in the L direction, where the soft region is dominating. In the T direction, where the proportion of hard regions and the 0.2% proof stress are high, local inelastic deformation occurs, and the minimum normalized stress at which the fatigue life is decreased by dwell fatigue decreases.
