An Abnormal Increase of Fatigue Life with Dwell Time during Creep-Fatigue Deformation for Directionally Solidified Ni-Based Superalloy DZ445

Introduction

DZ445 is a nickel-based superalloy that is often used for making turbine blade material for heavy-duty gas turbines owing to its high tensile strength and creep resistance at elevated temperatures [1]. During service, the high-temperature components are exposed to a range of loading conditions. It is the result of the start-up and shut-up processes that cause the happening of cyclic plasticity. Strain-controlled high temperature creep-fatigue can simulate the hostile environment subjected to temperature and load well. The creep-fatigue properties of elevated temperature are of great importance for the safe design of the components [2, 3].

A lot of previous researches on the influence of dwell time on the fatigue life of various alloys has been conducted. The incorporation of dwell time has a pronounced effect on the fatigue life of various alloys [3, 4, 11]. The creep-fatigue tests have been performed to investigate the dwell time effect for many alloys, such as precipitation hardened superalloy DZ125 [10], GTD-111 [5], René 80 [11], solid-solution-strengthened superalloys HAYNES 230 [4], Inconel 617 [6], oxide-dispersion-strengthened Ni-based superalloy PM 1000 [7], and stainless steels 316 [8], and 9Cr–1Mo ferritic steel [9] et al.

The fatigue life decreases with incorporation of dwell time for most kinds of alloy. The lifespan for DS GTD-111 and HAYNES 230 superalloy significantly reduces with increasing dwell times [4, 5]. Lu et al. [4] observed the accumulated strain in the crack-tip region. This induces intergranular crack growth and promotes the propagation of a fatigue crack, resulting in a decrease in the fatigue life. Gordon et al. [5] proposed that for a given set of experimental conditions the plastic strain ranges of dwell in tension are larger than those from creep-fatigue without dwell time, which is the reason for reduction of fatigue life. Meurer et al. [6] explained the cycles to failure for alloy 617 decreases under creep–fatigue by the large stress relaxation, especially at low strain ranges. Ngala et al. [7] reported the PM 1000 alloy exhibits the shortest cyclic life owing to the additional creep damage in the tensile-producing part of the cycle, whereas environmental attack plays only a minor role at and above 850 °C. Hormozi et al. [8] found the introduction of the dwell time reduces the life of 316 stainless steels at strain of ±0.4 %, but that is nearly unchanged where the imposed strain amplitude was ±0.8 %, which can be explained by the difference of the values for inelastic strain energy density. Fatigue life was found to decrease with increase in the duration of dwell time in both tension and compression for 9Cr–1Mo ferritic steel by Shankar et al. [9]. However, a saturation effect on the reduction of fatigue life with increasing dwell time was observed in GH4169 [3] and DZ125 [10]. Chen et al. [3] reported the elastic strain is partially converted into the inelastic strain during the process of stress relaxation and the inelastic strain reduces the fatigue resistance of the alloy. But the change amount of inelastic strain energy with a holding period of 120 s is similar to that of 300 s. Shi et al. [10] found the fatigue damage, creep damage, and creep-fatigue damage all contribute to the reduction of fatigue life. However, not all fatigue life reduces with increasing dwell time. Antolovich [11] found the dwell time of 90 s increases the life of René 80 at 871 °C. The fatigue life increase was all the more pronounced as the plastic strain decreases. He also explained it by structural coarsening. This necessitates a large value of spike depth in a crack initiation criterion and a correspondingly large number of cycles. The fatigue life decreases with incorporation of dwell time for most alloys, but that is opposite for few alloys.

From the above existing literatures, we can draw a conclusion. The reduction of fatigue life with dwell time can be attributed essentially to the accumulated strain in the crack-tip region, the accumulated inelastic strain energy, and the environmental damage during high temperature. Therefore, creep-fatigue behavior at elevated temperatures is an important property for various alloys in service. Unfortunately, systematic research on creep-fatigue interactions for the directionally solidified Ni-based superalloy DZ445 has not been reported so far. It is necessary for essential experimental investigation on the creep-fatigue in order to investigate the creep-fatigue damage mechanism of DZ445.

In this study, strain-controlling creep-fatigue tests were conducted to investigate the influence of dwell times on the fatigue life of DZ445 alloy at high temperatures. Various dwell times are introduced at the maximum tensile strain during the cyclic deformation. The abnormal fatigue life with the dwell time for DZ445 superalloy is found during the creep-fatigue deformation. It is explained from the area of the hysteresis loops at the initial steady cycle. The fractograph morphology further support the abnormal behavior of the fatigue life.

Experimental

Creep-fatigue test

The round bars were cut by wire-electrode cutting. Cylindrical specimens with a gage size of Φ9 mm × 26 mm and total length 101.6 mm were machined from round bars and the specimens were used for strain control tests. The geometry of the test specimen is shown in Figure 1. The gage surface was burnished by SiC sand paper with different mesh sizes in order to exclude interference from surface-machining defects.

Figure 1:

The geometry of the creep-fatigue specimen.

All creep-fatigue tests were performed on the RPL series of electronic equipment. They were conducted under a strain control of 1.6 % at a temperature of 900 °C. Specimens were heated with a resistance furnace and the test temperature was controlled to within ±2 °C over the specimen gauge length. Three thermocouples were mounted on the gauge length to monitor the test temperature. For the case without dwell times, a cyclic frequency with a triangular waveform of 0.08 Hz was used. The strain rate was held constant at 0.5 %/s during loading and unloading ramps in creep-fatigue tests with dwell times. The dwell times of 60, 120, 180, 300, and 480 s were introduced at the maximum tensile strain of each cycle to form trapezoidal waveforms. The axial strain was measured with a high temperature extensometer with a 25 mm gauge length. The strain ratio, R, of minimum to maximum strain was −1. As shown in Figure 2, triangular waveforms were used in creep-fatigue tests without dwell times and trapezoidal waveforms were used in creep-fatigue tests with dwell times. All specimens were run to failure. For each condition, three specimens were conducted.

Figure 2:

(a) Waveform of the test without dwell times. (b) Waveform of test with dwell times.

The microscope of DZ445 alloy and fracture surface were observed by a SU-1510 SEM and microscope Leica DM6000.

Results

Microstructure

The directionally solidified superalloys possess columnar crystals in which there are array dendrite-structures. The strengthening phase γ’ disperses in γ matrix both in the interdendritic region and dendritic stem. Other precipitate phases, such as carbides and γ/γ’ eutectic, distribute in the interdendritic region or along the grain boundary. The dendrite-structure and precipitate phase of Ni-based superalloy DZ445 are shown in Figure 3. The dendrite-structure profile before heat treatment [Figure 3(a)] is more obvious than that after heat treatment [Figure 3(a’)]. The strengthening phase γ’ before heat treatment [Figure 3(b)] is more regular than that after heat treatment [Figure 3(b’)]. The carbide morphology almost unchanges both in before heat treated and heat treated DZ445 [Figure 3(c) and (c’)]. The petaline γ/γ’ eutectic structures are presented along grain boundary in Figure 3(d) before standard heat treatment due to non-equilibrium solidification. The γ/γ’ eutectic structures disappear after heat treatment due to element diffusion [Figure 3(d’)].

Figure 3:

The dendrite-structure (a and a′), the cubic γ' (b and b′) in the matrix, the carbides (c and c′), and the petaline eutectic structure (d and d′) near the grain boundary of DZ445 superalloy. Figure a~d and a′~ d′ are the graphs from the before heat-treated and heat-treated samples, respectively. It should be noted that the eutectic disappears after heat treatment. Therefore, the eutectic is not observed near the grain boundary in Figure (d′).

Fatigue life behavior

The fatigue life with dwell time during the creep-fatigue loading for DZ445 alloy is shown in Figure 4. The fatigue life is defined as the number of cycles to specimen rupture. The fatigue life initially decreases with increasing the dwell time. However, when dwell time is more than 180 s, the fatigue life increases with dwell time. In order to clarify the ascending tendency when dwell time reaches 300 s, the upper-right enlarged picture in Figure 4 is shown. For the reported alloy, the fatigue life monotonously decreases with increasing the dwell time during the creep-fatigue deformation [4-9]. Consequently, the fatigue life shows an abnormal change with dwell time.

Figure 4:

The relationship between dwell times and the cycle of fatigue life (Nf) for DZ445 alloy at 900 °C.

Cyclic-stress response behavior

In order to clarify the abnormal change of fatigue life, dwell times at the maximum tensile strain were applied to investigate the effect of dwell time on the cyclic hardening and softening behaviors, thus the stress of cyclic deformation with different dwell times is shown in Figure 5(a). The tensile maximum stress and compressive maximum stress at a dwell time of 0 s are symmetrical, whereas the introduction of dwell time makes them asymmetric. Thus, the effect of dwell time was significant. Figure 5(b) presents the maximum tensile stress of every cycle with different dwell times. It could be seen that the maximum stress of every cycle results in a continuous softening behavior at 0 s dwell time. The application of dwell time hardens the alloy slightly after the initial softening. Both the softening and hardening velocity keep a constant value, which is defined as a steady cycle. We chose the turning point of the cyclic curve as the initial steady cycle, as indicated in Figure 5(b). The initial steady deformation under dwell times of 0, 60, 120, 180, 300, and 480 s are the 31st, 25th, 31st, 31st, 37th, and 46th cycle, respectively. At the point of initial steady cycle, compared to that in the absence of dwell time, the maximum stress in the presence of dwell times significantly decreases. This indicates that the failure rate of the internal material accelerates after introducing the dwell times.

Figure 5:

(a) The cyclic stress and (b) the maximum cyclic tensile stress with different dwell times (The intersection point between the initial rapid and steady decrease range is defined as the starting point of a steady cycle in the case of no dwell time).

Since the area of cycle hysteresis loops can indicate the consumed energy during the deformation, it is shown in Figure 6 at the initial steady cycle for different dwell times. The area for dwell times of 0, 60, 120, 180, 300, and 480 s was 19.55, 39.81, 42.82, 53.87, 39.94, and 33.57 cm², respectively, as shown in the lower-right enlarged picture in Figure 6. The area with dwell times is much larger than that without dwell time, and it goes through a maximum in the case of 180 s dwell time. This means that much more energy is consumed during the cycle of applying the dwell time, and the maximum energy is done in the case of 180 s dwell time. Therefore, the fatigue life shows the minimum during the creep-fatigue deformation in the case of 180 s dwell time.

Figure 6:

The hysteresis loop of the initial steady cycle of different dwell times at 900 °C.

Fractography

The failed samples were examined by fractographic observation in order to further clear the abnormal behavior of fatigue life from the damage mechanism. Figure 7(a) and (b) shows the fractograph morphology of the sample without dwell time. The fatigue source zone and crack propagation can be clearly seen. The crack initiation occurs from the specimen surface, and the crack propagation is wave-like. This is a typical transgranular fracture mode of low-cycle fatigue failure. Under the dwell time of 60 s [Figure 7(c) and (d)], the propagation route shows circuitous characteristics. So the propagation velocity is not so fast as that in the case of no dwell time. Under the dwell time of 120 s [Figure 7(e) and (f)], the circuitous characteristic is more evident. The propagation route develops into the tearing ridge, and the uneven fracture surface is formed. Furthermore, the dwell time reaches to 180 s [Figure 7(g) and (h)], the multi-source cracking occurs. The cracks initiate not only at the grain boundary but also around the specimen surface. So the fracture transforms into a mixture of cleavage transgranular and intergranular characteristics. When the dwell time increases to 300 s [Figure 7(i) and (j)] and 480 s [Figure 7(k) and (l)], the typical fatigue failure disappears, the intergranular fracture is still kept, and the ductile fracture appears. Thus, the fracture mode changes into the mixture of the ductile and intergranular creep mode. Therefore, the partial ductile fracture makes the alloy exhibit the longer fatigue life when the dwell time is more than 180 s.

Figure 7:

SEM images showing specimen surface during the creep-fatigue deformation at 1.6% total strain with various dwell times of (a) (b) 0, (c) (d) 60, (e) (f) 120, (g) (h) 180, (i) (j) 300, (k) (l) 480 s.

The change of sample fracture characteristics with different dwell times can also be seen from the fracture profile. As shown in Figure 8(a), in the case of without dwell time, the longitudinal macrostructure near the fracture surface is very smooth. When the dwell time reaches to 60 s [Figure 8(b)], the uneven extent increases from the macroscopic range. The enclosed area seems to be intergrain crack. These characteristics indicate that intergrain creep fracture begins to appear. When the dwell time increases to 120 s [Figure 8(c)], the uneven extent further propagates from the submacroscopic range. Under the dwell time of 180 s, multi-source cracking characteristics are shown in Figure 8(d), and the uneven extent continuously increases. When the dwell time increases again to 300 and 480 s, the variation of uneven extent in fracture surface keeps continuous propagation. In addition, the dashed area in Figure 8(e) appears to be the dimple-ductile characteristic. Consequently, the increase in the unevenness of the profile with the dwell time also demonstrates the fracture mode changes into the mixture of creep-fatigue and further into the mixture of creep-ductile from the typical fatigue.

Figure 8:

Optical longitudinal macrostructures near the fracture surface with dwell times of (a) 0, (b) 60, (c) 120, (d) 180, (e) 300, (f) 480 s.

Discussion

he above results show that the fatigue life of DZ445 alloy goes through a minimum when the dwell time increases. The area of cycle hysteresis loops under the different dwell time undertakes a maximum with the dwell time. As shown in Figure 9 (Symbols of Δεt, Δεin, Δεp, Δεpp, Δεcp, Δσr, and σT represent the total strain range, inelastic strain range, plastic strain range, plastic strain component, creep strain component, relaxed stress, and total stress range, respectively.), compared with the schematic of stress-strain hysteresis diagram of with dwell times and without dwell times, we can find that a creep strain component is added due to the application of dwell time. Therefore, the application of dwell time could result in a much larger area of hysteresis loop. As shown in the present experiment, the area of hysteresis loop with dwell times is much larger than that without dwell times. However, when the dwell time increases a certain value, the area of hysteresis decreases again. The percentage of plastic strain and creep strain of DZ445 under the different dwell times from the above experiment results was summarized in Table 2. It is found that the percentage of creep strain initially goes through a maximum in the case of 180 s dwell time and decreases again with increasing dwell times. Thus, the percentage of plastic strain presents the opposite trend. That indicates intergranular creep mode initially increases and then decreases when the dwell time exceeds a critical value. The ductile mode shows the opposite trend. Therefore, the change of cycle hysteresis loops can explain the variation of fatigue life of DZ445 with dwell time during creep-fatigue deformation.

Figure 9:

Schematic of stress-strain hysteresis diagram without dwell times (a) and with dwell times (b).

The change of fractography and fracture profile with different dwell times indicate that the fracture characteristics transforms into the creep-fatigue mode and further into creep-plastic mode from the typical fatigue mode. The appearance of creep fracture mode makes the fatigue life reduce. By contrast, when the plastic fracture begins to form, the fatigue life would be increased. Therefore, the transformation of fracture mode can explain the abnormal change in fatigue life under the different dwell times.

From the above analysis, it can be concluded that the fatigue life, deformation response, and fractograph exhibit the intrinsic relationship for the DZ445 superalloy during the creep-fatigue loading.

Conclusions

A series of creep-fatigue tests were conducted with directionally solidified Ni-based superalloy DZ445 at 900 °C to investigate the effect of dwell time. Based on the above experimental results and discussion, the following conclusions can be drawn.

1. The fatigue life initially decreases with increasing dwell time, it reaches the minimum under the dwell time of 180 s and then increases again with increasing dwell time during creep-fatigue deformation.

2. The introduction of dwell time makes the maximum tensile and compressive stresses asymmetric, remarkably softens the material, and further decreases the fatigue life.

3. The area of hysteresis loop goes through a maximum in the case of 180 s dwell time and decreases again with increasing dwell times. The fracture mode of a typical fatigue without dwell time changes into the mixture mode of the creep and fatigue when the dwell load initially is applied. When the dwell time further increases, the fracture mode exhibits the mixture characteristic of the ductile and creep failure. The change of the area of hysteresis loop and fracture mode can explain an abnormal increase of fatigue life with dwell time.