FATIGUE PROPERTIES OF NICKEL-BASE SUPERALLOY INCONEL 792-5A AT 800 C

Transcription

1 FATIGUE PROPERTIES OF NICKEL-BASE SUPERALLOY INCONEL 792-5A AT 800 C Miroslav ŠMÍD a, Karel OBRTLÍK a, Martin PETRENEC a, Jaroslav POLÁK a, Karel HRBÁČEK b a Ústav yziky materiálů, Akademie věd České republiky, v.v.i., Žižkova 22, Brno b PBS Velká Bíteš a.s., Vlkovská 279, , Velká Bíteš Abstract Smooth specimens were cyclically strained under strain control with constant strain amplitude and constant strain rate. Low cycle atigue tests were conducted in servo-hydraulic pulsator MTS equipped with a three zone resistance urnace at temperature 800 C in air. Surace o the specimen gauge length was mechanically ground and polished to enable SEM observation. Fracture surace was studied in SEM ater atigue test termination. Selected specimens were used to prepare oils or the transmission electron microscope (TEM) observation o microstructure and dislocation arrangement. Hysteresis loops were recorded or selected numbers o cycles. They were used to obtain cyclic hardening/sotening curves, cyclic stress-strain curve and atigue lie curves in the representation o stress amplitude, total strain amplitude and plastic strain amplitude versus number o cycles to racture. Experimental points can be approximated with the Manson-Coin law and the Basquin law. Fracture surace examinations revealed atigue crack initiation sites. Keywords: Inconel 792-5A, low cycle atigue, high temperatures, atigue lie curves, dislocation structure. 1. INTRODUCTION Inconel 792-5A (In 792-5A) is a cast polycrystalline nickel base superalloy. The material is strengthened by precipitates γ (Ni 3 Al). Thereore, it exhibits excellent mechanical properties at elevated temperatures. Its other advantage is very good high temperature corrosion resistance and hence it is used or production o the most thermally and mechanically stressed components, or example discs and blades o gas turbines and other power-producing units. In critical parts o these components cyclic elastic-plastic deormation occurs especially during warming up and cooling, e.i. start up and shut down cycles respectively [1]. From this reason, it is important to study low cycle atigue properties at elevated temperatures and also to observe changes in the material microstructure ater cyclic straining with electron microscopy. The aim o the paper is to report results o the stressstrain response and atigue lie o In 792-5A at temperature 800 C supplemented with electron microscope observation. 2. EXPERIMENT Polycrystalline superalloy In 792-5A was provided by PBS Velká Bíteš, a.s. as conventionally cast rods. Cylindrical button-end specimens were machined parallel to the rod axis with gauge length and diameter o 15 and 6 mm, respectively. Chemical composition is presented in Table 1. The gauge length o specimens was mechanically ground and polished or urther surace relie observation. Structure o the superalloy consists o coarse dendritic grains, carbides, eutektics γ/γ and numerous shrinkage pores with diameter up to 0.7 mm. Typical structure is shown in Figure 1a. Microstructure o In 792-5A contain matrix γ and mostly cuboidal precipitates γ with the edge size approximately 600

2 nm. It is also obvious that numerous ine precipitates γ are homogenously distributed in the matrix (Figure 1b). The linear intercept method revealed the average grain size o 3 mm, thus the gauge length can comprise several grains. This act relects in signiicant scatter o elastic modulus o individual specimens. Typical values were in interval rom 151 to 193 GPa at 800 C. Table 1. Chemical composition o IN 792-5A superalloy (wt. %) Cr Co Ti Al Ta W Mo Nb Fe Zr C B Ni rest a) b) Fig. 1 Structure o IN 792-5A a) carbides, eutectics γ/γ in interdendritic areas b) TEM micrograph showing typical microstructure consisted o cuboidal γ precipitates and matrix γ with very ine γ precipitates. Low cycle atigue tests were conducted on an electro-hydraulic computer controlled testing system MTS 810 in the strain control regime at temperature 800 C in air. The strain rate was held constant (0.002 s -1 ) and a symmetric push-pull cycle (R ε = -1) was used. The strain was measured and subsequently controlled by a sensitive extensometer with 12 mm base. Fatigue tests with constant total strain amplitude ε a were run to racture. Heating on experimental temperature was provided by a 3-zone resistance urnace controlled by a 3-channel regulator. Actual temperature was measured by 3 thermocouples attached to both specimen ends and also to the gauge length. Scanning electron microscope JEOL JMS6460 was used or surace relie and racture surace observations. Dislocation structure observation was conducted in transmission electron microscope Philips CM-12 operating at 120 kv with a double tilt holder. 3. RESULTS 3.1 Stress-strain response Figure 2a and 2b shows the cyclic hardening/sotening curves o selected specimens in the representation o the stress amplitude versus the number o cycles and the plastic strain amplitude versus the number o cycles respectively. Each experiment was conducted with dierent total strain amplitude. Tests conducted at low amplitudes result in stable stress response with slight sotening

3 during the whole atigue lie. Specimens cycled at high total strain amplitudes show initial hardening ollowed by sotening which became more signiicant at the end o the atigue lie. It can be also seen that the specimen cycled at the highest total strain amplitude doesn t yield the biggest stress and plastic strain amplitude. This act is a result o modulus scatter o individual specimens. Cyclic stress-strain curve o the material is shown in Figure 3. The diagram was plotted using the stress and plastic strain amplitudes at hal-lie. Experimental data were itted by the power law log σ + a = log K n logε ap (1) where K is atigue hardening coeicient and n is atigue hardening exponent. Their values are 1120 MPa and respectively. 6.0x ε a = 0.51% ε a = 0.48% εa = 0.37% ε a = 0.31% ε a = 0.17% 4.0x10-4 ε a =0.51% ε a=0.48% ε a=0.37% ε a=0.31% ε a=0.17% σ a [MPa] 600 ε ap [ - ] x10-4 a) 200 Fig. 2 Cyclic hardening/sotening curves a) stress amplitude vs. number o cycles b) plastic strain amplitude vs. number o cycles 3.2 Fatigue lie cycles [ - ] b) 0.0x cycles [ - ] σ a [ MPa ] The representation o the plastic strain amplitude ε ap at hal lie vs. the number o cycles to racture N is shown in Figure 4a. Experimental data were approximated by the Manson-Coin law ap, ( 2N ) c ε = ε (2) 300 where ε is the atigue ductility coeicient and c is the atigue ductility exponent. Their values are and respectively. It is obvious that this law describes suiciently low ε ap [ - ] Fig. 3 Cyclic stress-strain curve cycle atigue behaviour o the material. The great scatter in the elastic modulus can contribute to the great scatter o experimental data in the low amplitude domain. Figure 4b shows a atigue lie curve in the representation o the stress amplitude σ a at hal lie vs. the number o cycles to racture N. Experimental data were itted by the Basquin law a, ( 2N ) b σ = σ (3)

4 where σ (1333 MPa) is the atigue strength coeicient and b (-0.140) is the atigue strength exponent. Their values are in Table 2. It can be seen rom Fig. 4b that the law describes the atigues lie satisactorily ε ap [ - ] 10-5 σ a [ MPa ] a) 10 b) -7 Fig. 4 a) Manson-Coin atigue lie curve b) Basquin atigue lie curve 3.3 Surace relie N [ cycles ] Observation o surace relie was conducted on selected specimens with polished gauge length by SEM. Cyclic strain localisation was ound in the vicinity o natural stress concentrators like shrinkage pores. A typical example is shown in Fig. 5a. The surace relie consists o short and wavy persistent slip markings (PSM) o the orientation which depends on the crystal orientation o the grain. Fatigue crack initiation along the markings is clearly visible in Fig. 5a. 3.5 Fracture surace N [ cycles ] Selected specimens cycled until atigue ailure were used or racture surace observation. Shrinkage pores that are the most requent sites o the atigue crack initiation are clearly visible in Fig. 5b. Those technological laws were up to 0.7 mm in diameter. The irst transgranular stage o the atigue crack growth was predominantly ound in the vicinity o shrinkage pores. Fields o striations (Figure 5b) were ound in areas where the atigue crack reached the second stage o the atigue crack propagation. Secondary cracks were also observed in those areas. Similar eatures were already described on specimens cycled at dierent elevated temperatures. [2]. a) b) Fig. 5 SEM micrographs a) surace relie with persistent slip markings and atigue cracks initiated along them b) ields o striation and shrinkage pore denoted with arrows

5 3.5 Dislocation structures a) b) Fig. 6 TEM micrographs a) stacking ault going through matrix and precipitate γ along ( 111 ) plane b) a precipitate cut by two stacking aults and dislocation net and ine precipitates in the matrix Thin oils were cut rom gauge length o a cycled specimen (ε a = 0,3%, N = 170) in the direction parallel with the stress axis. Typical eature o dislocation structure ater cyclic loading is nonhomogeneous distribution o dislocations. Cyclic plastic deormation concentrates both in the γ channels and in the orm o persistent slip bands (PSB). It was already documented that PSBs occur in superalloy In 792-5A ater cyclic loading at elevated temperatures [3,4]. Figure 6a shows dislocations predominantly at the interace between matrix and precipitates. In the middle o Fig. 6a, a stacking ault is apparent going along the ( 111 ) plane through the matrix and shearing precipitates γ. Similar situation is also in Figure 6b where two stacking aults in the plane ( 111 ) shear a precipitate γ. The observed grain is oriented or multiple slip. PSBs were observed mostly in grains oriented or single slip [3,4]. Figure 6b also shows ine precipitates γ in the matrix. They are obstacles or the dislocation movement in the matrix. Thus, dislocations can be ixed and bent in their vicinity. 4. DISCUSSION Cyclic hardening/sotening curves, cyclic stress-strain curves and atigue lie curves were evaluated rom the LCF tests o IN 792-5A. The LCF behaviour o the material is in agreement with the previous study [3]. However, a large scatter o experimental data maniest in the LCF parameters o In 792-5A. Both large casting deects and comparatively large grain size contribute to the scatter. Surace relie was investigated by SEM. PSMs were ound short and wavy. Further research is needed to reveal their properties. The atigue crack initiation was ound mostly in the vicinity o shrinkage pores both at the surace and in the bulk. This inding is in agreement with studies [2]. The irst stage o the atigue crack propagation is transgranular and parallel to the primary slip planes (111). When the plastic zone at the crack tip is big enough, i.e. the crack is long enough, additional slip systems are activated. It can result

6 i the ormation o striation ields on the racture surace with possible occurrence o secondary cracks see Fig. 5. Shearing o precipitates γ by stacking aults (see Fig. 6) ater cyclic loading and also ater creep straining o superalloys is in accord with earlier studies [5-7]. Fine precipitates γ in the matrix (Fig. 6) can hinder the movement o dislocations and thereore they beneicially contribute to strength properties o the material. 5. CONCLUSIONS Results o present study can be summarized as ollows: (i) (ii) (iii) (iv) High strain amplitude LCF tests are characterised by initial hardening ollowed by sotening which is more pronounced at the end o atigue lie. Low strain amplitude test showed stable behaviour. Cyclic stress-strain curve can be approximated by power law. Fatigue lie experimental data can be approximated using the Basquin and the Manson- Coin law. The strain localisation into short and wavy persistent slip markings is documented. The atigue cracks initiate along the markings later in the atigue lie. Shrinkage pores prove to be signiicant stress concentrators. The atigue cracks initiate predominantly in their vicinity. Dislocations are present both in γ channels and in γ precipitates. High dislocation density was ound in the γ channels particularly at the γ/γ interace. ACKNOWLEDGEMENTS This research was supported by the projects Nos. AV0Z o the Academy o Sciences o the Czech Republic and the grant 106/08/1631 o the Grant Agency o the Czech Republic. LITERATURE [1] DONACHIE, M.J., DONACHIE, S.J., Superalloys. A Technical Guide, Mater. Park OH : ASM Int., [2] ŠMÍD, M., OBRTLÍK, K., PETRENEC, M., POLÁK, J., HRBÁČEK, K.: Únavová životnost a únavový lom lité niklové superslitiny inconel 792-5A při pokojové teplotě a při zvýšených teplotách. In: METAL 2009, Hradec nad Moravicí, ISBN , 5.díl, str [3] OBRTLÍK, K., PETRENEC, M., MAN, J., POLÁK, J., HRBÁČEK, K. Low cycle atigue o superalloy Inconel 792-5A at 23 and 900 C. In Fatigue 2006 : 9 th Inter. Fatigue Congress : Atlanta, Georgia Inst. o Technology, US [CD- ROM]. London : Elsevier, 2006, paper No. FT307. [4] PETRENEC M, OBRTLÍK K, POLÁK J, MAN J, HRBÁČEK K: Fatigue behaviour o cast nickel based superalloy INCONEL 792-5A at room temperature. Materials Engineering 12 (2005) [5] YU, J. et al.: High temperature creep and low cycle atigue o a nickel-base superalloy. Material Science and Engineering A, 527(2010), pp [6] ZHOU, H. et al.: Deormation microstructures ater low-cycle atigue in a ourth-generation Ni-base SC superalloy TMS Materials Science and Engineering A, 381(2004), pp [7] BRIEN, V., DÉCAMPS, B.,: Low cycle atigue o a nickel based superalloy at high temperature: deormation microstructures. Materials Science and Engineering A, 316 (2001), pp