Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Scripta Materialia 54 (2006) 5155 www.actamat-journals.

com

Creep crack growth resistance of an age hardened aluminium alloy for supersonic applications
G. Odemer a, G. Hena
a

a,*

, B. Journet

canique et de Physique des Mate riaux, UMR CNRS n6617, ENSMA, 86961 Chasseneuil, France Laboratoire de Me b partement Inge nierie des Structures, EADS CCR, 92152 Suresnes, France De Received 21 July 2005; received in revised form 29 August 2005; accepted 2 September 2005 Available online 6 October 2005

Abstract This paper presents experimental results on the creep crack growth behaviour of an age hardened aluminium alloy. Inuence of a a given creep regime at dierent temperatures on the values in the exponent of the creep power law d / K b and on fracture modes is dt investigated. 2005 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
Keywords: Creep; Fracture; Aluminium alloys; Grain boundary diusion; Intergranular

1. Introduction The rst supersonic civil transport aircraft, Concorde, was originally designed to sustain 7000 ights, i.e. 15,000 h. The cruise speed of Mach 2.05 induced a maximum temperature of the fuselage skin of 130 C. From a design point of view, creep and fatigue were considered independently, although some experimental results suggested a possible creepfatigue interaction for the 2618A aluminium alloy used for the fuselage. Furthermore at that time, the damage tolerance philosophy was not mature yet and the life predictions were mainly based on safe life concepts, without specic consideration of crack growth. The future supersonic aircraft will have to sustain a total of 20,000 ights, i.e. 60,000 h, under the same temperature. The fuselage skin will still be made of aluminium alloy and the alloy selected is the 2650 T6 alloy. Moreover, the structure will have to meet damage tolerance requirements. Due to the elevated temperature, issues related to creepfatigue interactions during crack growth can be expected.

With this respect, the present work was undertaken within the framework of the French National Program on Supersonic Transport with the aim of developing a crack propagation model taking into account creepfatigue interactions. However, in order to evaluate possible creep fatigue interactions, one needs crack growth data not only under creepfatigue loading, but also under pure fatigue and creep loading. However the creep crack growth (CCG) behaviour of aluminium alloys has not been extensively studied so far. The present paper presents an experimental study on the CCG resistance of the 2650 T6 alloy in the 100175 C temperature range. 2. Experimental procedure 2.1. Material The 2650 alloy is a coppermagnesium aluminium alloy, nalu (thickness: 2.5 and 5 mm). provided in sheets by Rhe Chemical composition is presented in Table 1 and optical micrographs in Fig. 1a and b. The crack growth resistance of aluminium alloy is investigated after T6 articial ageing (192 C for 21 h). The precipitation hardening evolution during thermal exposure has

Corresponding author. E-mail address: hena@lmpm.ensma.fr (G. Hena).

1359-6462/$ - see front matter 2005 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.scriptamat.2005.09.006

52

G. Odemer et al. / Scripta Materialia 54 (2006) 5155

Table 1 Composition of constituent elements in aluminium 2650 T6 Element Min Wanted Max Si Fe Cu Mn Mg Cr Ni Zn Ti Zr

0.36 0.08 2.60 0.32 1.50 0.08 0.40 0.11 2.70 0.35 1.60 0.10 0.44 0.13 2.80 0.38 1.70 0.04 0.03 0.10 0.12 0.03

been studied by Lapasset et al. [1]. It was shown that the astreated alloy contains both S 0 precipitates and ne GPB (or S00 ) needles which are an obstacle to dislocations because of their high density. During a thermal exposure at 150 C during 5000 h, these needles tend to disappear caused by their consumption by S 0 particles. This consumption is enhanced during creep deformation. 2.2. Testing CCG testing was performed on compact tension specimens (W = 32 mm) of 5 mm thickness in the LT orientation. The samples were precracked by fatigue at room temperature using a servohydraulic machine. In order to avoid a possible interaction between the fatigue precracking plastic zone and the creep zone, the initial value of K at the beginning of the creep test has to be higher than the nal value of Kmax calculated at the end of the fatigue precracking. The CCG tests were carried out using deadweight lever-type creep machines to apply a constant load. The crack length was monitored by means of the potential drop technique. 3. Results and discussion 3.1. Incubation time During CCG experiments, crack extension was not detected immediately after application of the load. The time

during which no crack advance is measured is denoted incubation time, ti. Its value was experimentally determined as the time required to obtain a 1 mV variation in the potential drop, as proposed by Bensussan et al. [2], which corresponds to a crack advance of about 0.05 mm. The ti values obtained at dierent loads at 130 C are given in Table 2. This time can be interpreted as the time necessary to accumulate a critical amount of creep damage at the precrack tip to initiate propagation. According to Vitek [3], this initiation would be governed by a critical value of the crack opening displacement. Ewing [4] and Riedel [5], using a modied Dugdale model, suggests that at high stresses: ti / K i2n , where n is the creep power law exponent. 2 In the present case we have: ti / K i , while n  8 for the alloy studied at 130 C [6]. This discrepancy between theory and experimental results may be partly accounted for by the fact that the experimental incubation time may also include a propagation stage where the CCG rate is so slow that it cannot be detected by the potential drop method [2]. 3.2. Creep crack growth Fig. 2 presents the crack length measured as a function of time for dierent applied load P values at 130 C for nearly the same initial crack length. It appears that time to failure, and subsequently CCG rates, depend on P. This highlights the need for a crack growth parameter to correlate these data. Dierent parameters have been proposed in

Table 2 Incubation times as a function of the initial stress intensity factor Ki Incubation time (h) 192 96 72 K-level (MPam0.5) 20 29 33

Fig. 1. (a) Optical micrographs (EADS) and (b) optical micrographs (ENSMA).

G. Odemer et al. / Scripta Materialia 54 (2006) 5155

53

Fig. 2. Crack advance vs time.

the past to correlate CCG rates in various alloys as summarized by Saxena [7] and Riedel [8]. In creep ductile materials, creep deformation accumulation is rapid ahead of the crack tip while the crack advance is slow. It is possible to consider the crack as stationary within an expanding creep zone. Then, CCG rates can be correlated to a time-dependent crack tip parameter like C* or Ct [7,9]. However, in creep brittle materials such as age-hardened aluminium alloys, CCG rates are fast and crack growth signicantly perturbs the crack tip stress eld. A time-independent parameter such as the stress intensity factor K or the elasticplastic J-integral can be used to describe CCG. Previous studies suggest that K would be the most suitable parameter to correlate crack growth rates in aluminium alloys [2,10]. The CCG rates are plotted as a function of K in Fig. 3. It appears that the K parameter accounts for the dependency of growth rates upon applied load for crack growth rates higher than 5 109 m/s. However such a correlation is

not obtained in the slow crack growth rates regime. A similar inuence of the initial loading conditions on CCG curves has been previously observed for aluminium alloys [2,10]. The inuence of temperature on the CCG resistance of the 2650 alloy is presented in Fig. 4. Crack growth resistance is slightly aected by temperature in the range 100 130 C and is more pronounced in the range 130160 C probably due to an increase of creep contribution in local stressstrain response. The phenomenon seems to saturate between 160 C and 175 C. This inuence of temperature is consistent with previous CCG studies on 2618 [10], 2219 [2] alloys. Regardless of temperature the CCG curves present three stages, schematically represented in Fig. 5. In stage I, which is strongly inuenced by initial loading as previously mentioned, CCG rates rapidly increase with K. Stage II, which is less inuenced by loading, is characterised by a power a law relationship between CCG and K: d / K b2 . In stage dt III, an increase in the exponent of the power law b3 is noticed.

Fig. 4. Inuence of temperature on CCG rates.

Fig. 3. CCG rates vs factor K (130 C).

Fig. 5. Schematic representation of CCG curve.

54

G. Odemer et al. / Scripta Materialia 54 (2006) 5155

Table 3 Values of exponents in the steady state regime at dierent temperatures T (C) 100 130 160 175 [K2] (MPa m0.5) 2947 2235.5 1631 1942 [da/dt]2 (m/s) 5 10 3 10 1096 109 2 1093 108 2 1085 107
10 8

b2 8.47 3.88 4.04 4.08

[K3] (MPa m0.5) 4752 35.547.5 3550 4252

[da/dt]3 (m/s) 5 1083 107 6 1091 107 5 108106 5 1072 106

b3 22.36 10.17 8.33 8.03

tioned by Leng [10], one source of discrepancy is that the power exponent n is determined for bulk material, although CCG is governed by highly localised phenomena. Furthermore, the transition between stage II and III is shifted towards higher K values as temperature increases. Fig. 6 compares the CCG rates of the 2650 T6 alloy under study with the resistance of 2618 T651 [10] and 2219 T851 [2] aluminium alloys at 175 C. It can be seen that the 2650 alloy presents the highest resistance, which gives further condence in the alloy selection. However an ageing eect cannot be excluded at this temperature (175 C) which is close to the heat-treatment temperature (192 C for 21 h) of the 2650 alloy. However, such an eect when it exists would be benecial to crack growth resistance. 3.3. Creep crack growth mechanisms
Fig. 6. CCG resistance of the 2650 T6, 2618 T651 [10] and 2219 T851 [2] alloys.

The values of the exponents found in the indicated K and da/dt ranges of stage II and stage III are given in Table 3. It can be seen that, except at 100 C where the steady state regime is hardly reached, the exponent values are nearly temperature independent and are close to: b2  4 and b3  8. A fourth power law was previously observed by Bensussan et al. in stage II. Many theoretical models [4,8] predict a power law dependence between CCG rates and stress intensity factor. However, the exponent is generally related to the creep power law exponent n. Clearly such a dependence is not observed in the present case. As men-

CCG fracture surfaces exhibit two characteristic failure modes: an intergranular mode at low growth rates and a mixture of transgranular and ductile fractures before failure. Fig. 7 presents these dierent fracture modes. Intergranular decohesions are prevailing just after the fatigue precracking, in the slow CCG rates regime. The intergranular mode (Fig. 7a) occurs by cavitation along grain boundaries. A physical model of CCG, proposed by Wilkinson and Vitek [11], considers that cavities nucleate and grow on grain boundaries ahead of the crack tip. The exponent b depends on the mechanism controlling the cavity nucleation and growth, basically diusion or creep deformation. However, a K4 dependence of CCG is obtained when CCG is controlled by the growth of a crack-like cavity close to the crack tip [11] except for 100 C where diusion is probably

Fig. 7. (a) Intergranular fracture mode (K = 18 MPam1/2 175 C) and (b) intergranular and ductile rupture (K = 48 MPam1/2 175 C).

G. Odemer et al. / Scripta Materialia 54 (2006) 5155

55

3. Temperature has a deleterious eect on the crack growth rates between 100 C and 175 C. This eect is particularly marked in the range 130160 C but seems to saturate above 160 C. However, exponents of the CCG rates are nearly temperature independent. 4. Regardless of temperature, the fracture mode is predominantly intergranular at slow growth rates, while a mixture of a ductile rupture is noticed at the end of the propagation. This change in fracture modes is consistent with the transition from stage II to stage III in CCG curves. 5. The 2650 T6 alloy exhibits a higher CCG resistance than similar age-hardened aluminium alloys.

Acknowledgement
Fig. 8. Evolution of the intergranular decohesions during CCG.

negligible. At high growth rates, the fracture surfaces consist of a mixture of a dominant intergranular propagation and a ductile rupture with tearing which increases with the CCG rates (Fig. 7b). This type of fracture mode is also observed for the 2618 T651 [10] and the 2219 T851 [2] alloys. Intergranular damage slowly accumulates on grain boundaries until a critical stage where some of these grain boundary cracks coalesce and join the main crack by unstable ductile tearing. Fig. 8 presents the evolution of the intergranular decohesions percentage correlated with the stress intensity factor K for two temperatures: 130 C and 175 C. The relative importance of this fracture mode in the crack growth process is a function of the crack growth rates and the stress intensity factor. The transition to the third stage in CCG, where unstable ductile tearing begins, takes place when the intergranular decohesions percentage is equal to one third of its initial value. 4. Conclusions 1. The crack growth stage is preceded by an incubation time, which strongly depends on the applied load and on temperature. 2. Crack growth rates are correlated with the stress intensity factor K, consistent with the brittle creep fracture of the alloy 2560 T6.

This work was carried out within the framework of the French National Programme on Supersonic Aircraft and the nancial support by the French Ministry of Research is gratefully acknowledged. References
[1] Lapasset G, Denquin A, Casanove MJ, Coujou A, Majimel J, nat G, et al. Investigation of the evolution of hardening Mole precipitates during thermal exposure or creep of a 2650 aluminium alloy. Scripta Mater 2002;46(2):1139. [2] Bensussan P, Jablonski D, Pelloux R. A study of creep crack growth in 2219-T851 aluminium alloy using a computerized testing system. Metall Trans 1984;15A. [3] Vitek V. A theory of the initiation of creep crack growth. Int J Fract 1977;13:3950. [4] Ewing DJF. Strip yield models of creep crack initiation. Int J Fract 1978;14(1):10117. [5] Riedel H. A Dugdale model for crack opening and crack growth under creep conditions. Mater Sci Eng 1977;30:18796. [6] EADS. Unpublished data. [7] Saxena A. Mechanics and mechanisms of creep crack growth. In: Proc. fracture mechanics microstructure and michromechanism, in ASM Seminar, Cincinnati, (OH), October 911, 1987. [8] Riedel H. Fracture mechanics: perspectives and directions. 20th Symp, ASTM STP 1020. Philadelphia, (PA): ASTM; 1989. p. 10126. [9] Bassani JL, Hawk DE, Saxena A. Nonlinear fracture mechanics. ASTM STP 995, vol. I. Philadelphia, PA: ASTM; 1989. p. 726. [10] Leng Y. Study of creep crack growth in 2618 and 8009 aluminium alloys. Metall Mater Trans A 1995;26A(February):31528. [11] Wilkinson DS, Vitek V. The stress analysis and diusional growth of microvoids ahead of a crack. Cavities and cracks in creep and fatigue. Cavities and cracks in creep and fatigue. London: Applied Science Publishers; 1981. p. 24358.

You might also like