Theoretical and Applied Fracture Mechanics 60 (2012) 1–9

Contents lists available at SciVerse ScienceDirect

Theoretical and Applied Fracture Mechanics

journal homepage: www.elsevier .com/locate / tafmec

Fatigue and fracture behaviour of friction stir welded aluminium–lithium 2195

P.M.G.P. Moreira a,⇑, A.M.P. de Jesus b, M.A.V. de Figueiredo c, M. Windisch d, G. Sinnema e,P.M.S.T. de Castro c

a INEGI, Laboratório de Óptica e Mecânica Experimental – LOME, Porto, Portugalb Universidade de Trás-os-Montes e Alto Douro, Departamento de Eng. Mecânica, Vila Real, Portugalc Universidade do Porto, Faculdade de Engenharia, Departamento de Eng. Mecânica, Porto, Portugald MT Aerospace AG, Department TEA, Fracture Mechanics and Damage Tolerance, Augsburg, Germanye European Space Agency (ESA), ESTEC, Noordwijk, The Netherlands

a r t i c l e i n f o

Article history:Available online 28 June 2012

Keywords:Aluminium–lithium alloyCrack growth testsFSW

0167-8442/$ - see front matter � 2012 Elsevier Ltd. Ahttp://dx.doi.org/10.1016/j.tafmec.2012.06.001

⇑ Corresponding author.E-mail addresses: pmgpm@fe.up.pt (P.M.G.P.

(P.M.S.T. de Castro).

a b s t r a c t

Aluminium–lithium (Al–Li) alloys offer attractive properties for lightweight aerospace structures, due totheir low density, high strength and fatigue crack growth resistance. Although there are many advantageswith Al–Li alloys, limitations remain while using conventional joining techniques.

Friction stir welding is a well-established solid-state joining process that is expected to reduce many ofthe concerns about Al–Li welding.

The work presented in this paper involves the characterisation of the fatigue performance of theAA2195-T8X at room temperature. SN and crack growth tests of base material and friction stir welded5 mm thick specimens were performed. During crack growth tests, three different R ratios (minimumremote stress/maximum remote stress), 0.1, 0.5 and 0.8, were used per each three different material con-ditions: base material, heat affected zone (HAZ), and weldment. M(T) specimens containing notches atthe centre of the weld, at the HAZ and at the base material, were tested. The fatigue crack growth spec-imens were left with an un-cracked ligament for final evaluation of fracture toughness.

Novel results are presented for fatigue crack growth and toughness on T–L orientation. The results forSN fatigue behaviour, fatigue crack growth and toughness of the studied alloy and its friction stir weld-ments present high values when compared with data found in the literature.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

In the aeronautics and space industries one of the most effectiveways to reduce weight is to reduce the density of the aluminiumalloys used. For purposes of reducing the alloy density, lithiumadditions have been used. The rapid increase in solid solubility oflithium in aluminium over the temperature range of 0–500 �Cresults in an alloy system achieving, through precipitation harden-ing, good strength levels. However, the addition of Li–Al alloyspresents problems, as possible decreases in ductility and fracturetoughness, delamination problems and poor stress corrosioncracking resistance. Increased strength with only minimal or nodecrease in toughness is therefore a major issue [1,2].

The interest in Al–Li alloys derives from the large effect thatlithium additions have on the modulus of aluminium, a 6% increasefor every weight% added, and the density, a 3% decrease for everyweight% added [3]. These changes apply for lithium additions upto 3 weight%. There have been three early generations of Al–Li

ll rights reserved.

Moreira), ptcastro@fe.up.pt

alloys, (i) those produced in the 50s–70s, including alloys 2020and 1420; these alloys experienced ductility and fracture tough-ness problems, or were of relatively low strength; (ii) those pro-duced in the 1980s, including alloy 2090, 2091, 8090, and 8091,with high modulus and low density, but displaying anisotropicmechanical properties, and (iii) more recent high-strength alloysas the 2195 and 2198 alloys [4–6].

Al–Li alloys offer attractive properties for lightweight aerospacestructures, due to their low density, high strength and fatigue crackgrowth resistance. Although there are many advantages with Al–Lialloys, limitations remain while using conventional joiningtechniques, as low joint efficiency (ratio of weld strength to basemetal strength) in the as welded condition [7]. Post-weld heattreatment enhances the yield strength, but no increase in fatiguestrength was observed by [8] in GTA welded AA2195 samples.Friction Stir Welding (FSW) is a well-established solid-state joiningprocess, comprehensively reviewed in [9–11], that is expected toreduce many of the concerns about Al–Li welding.

The present work was performed under the ESA TRP ‘DamageTolerance of Cryogenic Pressure Vessels’ aiming at defining poten-tial applications for state of art FSW techniques in cryogenic tanksfor Expendable Launch Vehicle and Reusable Launch Vehicle.

Fig. 1. Geometry of the specimen to be use in the fatigue tests (SN).

Table 2SN tests, definition of remote loads matrix.

R ratio Number of specimens rmax Maximum load (kN)

0.1 2 FSW ryield 18.002 85% FSW ryield 15.302 70% FSW ryield 12.60

0.8 1 FSW ryield 18.001 115% FSW ryield 20.701 130% FSW ryield 23.402 150% FSW ryield 27.00

2 P.M.G.P. Moreira et al. / Theoretical and Applied Fracture Mechanics 60 (2012) 1–9

2. SN fatigue tests

SN and fatigue crack growth tests of base material and frictionstir welded (FSW) 5 mm thick specimens of the AA2195-T8X wereperformed at room temperature (RT). The test results of each spec-imen were linked to the specimen designation of the cut-off plan[12].

The AA2195 alloy was received in the T3R78 condition with aninitial thickness of 6.35 mm and aged to T8X condition to obtainhigher strength. The plates were machined to 5 mm thickness asa typical weld thickness in cryogenic launcher applications. Moredetailed information can be found in [13].

FSW, performed by MT-Aerospace, was optimised in order toobtain high strength and good ductility. The main parametersanalysed in the optimisation procedure were the rotation speed,the travel speed, and the vertical down force. The selectioncriterion for the final weld was a flawless visual appearance ofthe weld, a high strength of the weld, a high angle of bending with-out cracking in bending tests and the lack of cleavage planes on thefracture surface of tensile and bending specimens. With the opti-mised parameters obtained, about 20 m of weld were preparedclose to application conditions. As post-weld treatment only handscraping technique was applied to remove sharp flash on the sideof the weld. Welding direction is parallel to the material rollingdirection, and loading is perpendicular to the weldment.

Preliminary tensile tests were performed in order to obtain thestrength values of the AA2195-T8X friction stir welded material.These values, presented in Table 1, are used for the definition ofthe load levels of the fatigue tests and to determine the maximumload to be used on the fatigue pre-cracking and crack propagationprocedures to prevent plasticity effects.

2.1. Test definitions

The stress life curves were evaluated according to ASTM [14,15]using integral specimens perpendicular to the weld. The specimengeometry is presented in Fig. 1 [12].

For the calculation of the maximum loads to be applied in SNtests, the initial section is considered to be 5 � 12 = 60 mm2, andthe FSW yield stress is the value presented in Table 1 (300 MPa).

The roughness of the base metal is characterised by Ra (rough-ness average) of 0.26 lm and Rz (peak-peak) of 1.84 lm (measur-ing range 80 lm and cut-off 0.800 mm).

Tests were carried out at two different R ratios, 0.1 and 0.8,using for each stress ratio three different maximum loads. The R ra-tio of 0.1 represents pressure cycling (proof test, leak test, etc.), andthe R ratio of 0.8 represents external loads during operation. Loadfrequencies of 8 Hz and 15 Hz were used for R ratios of 0.1 and 0.8,respectively. The tests were performed according to the matrixpresented in the following Table 2.

2.2. Test procedure and setup

These tests were carried out in a MTS 810 servo-hydraulicmachine with a 100kN load cell. A mechanical grip fixture wasdeveloped in order to comply with the specimen geometry andmaximum loads defined in the general test plan [16]. The geom-etry of each specimen was accurately measured before each test,

Table 1Strength values for test definition and performance.

Rp0.2 (ryield)

2195 T8X 510 MPa2195 T8X FSW 300 MPa

especially in the section of the material affected by the weldingprocess.

3. Fatigue crack growth

Fatigue crack growth curves were evaluated according to ASTM[17,18] considering the use of M(T) specimens 5 mm thick. The ini-tial notch is oriented in accordance with the material rolling direc-tion (T-L according to ASTM [19]). The specimen geometry ispresented in Fig. 2 [16]. For three different R ratios, R = 0.1, 0.5and 0.8, 3 specimens were tested per each three different materialconditions: base material, heat affected zone (HAZ), and weldment.The fatigue pre-cracking of the spark erosion notch was made inorder to achieve a minimum of 0.2 mm sharp crack extension. Atthe end of the test Kc values and K–R curves were evaluated witha minimum remaining ligament of 15 mm.

A crack length of 25 mm was obtained before loading to frac-ture. According to [16], the crack length has been chosen such thatmaximum possible length for crack growth evaluation and a rea-sonable ligament (15 mm) before fracture can be achieved.

All experimental care was used aiming at symmetrical crackgrowth, which should result into a simultaneous verification ofthe 2a = 50 mm and (W–2a)/2 = 15 mm requirements. If unsym-metrical fatigue crack growth occurs, the criterion was to stopthe fatigue crack propagation test as soon as the first ligament of15 mm was reached.

The specimens have notches introduced in the centre of theweld, in the HAZ and in the base material (BM), as presented inFig. 3 by lines 1, 2 and 3, respectively. The first type of specimenshave notches in the centre of the weld (line 1), which coincideswith the centre of the weld nugget. In the second type of speci-mens the crack is located in the HAZ (line 2). The positioning of

Fig. 2. Definition of M(T) specimen geometry.

P.M.G.P. Moreira et al. / Theoretical and Applied Fracture Mechanics 60 (2012) 1–9 3

the notch in the HAZ was done by slightly etching the sides of eachsample, which is expected to allow a positioning of the notchaccording to the microstructure. The location of the notch on theretreating side is based on the position of the fracture in the inte-gral tensile specimen at room temperature.

3.1. Calculation of test parameters

Table 3 presents the specimen geometry and crack size defini-tion according to ASTM [17].

According to the standard ASTM [17] for the selected type ofspecimen the following equation should be verified,

ðW � 2aÞP 1:25 � Pmax

B � rYSð1Þ

where (W – 2a) is the specimens un-cracked ligament; rYS consid-ered to be the yield stress presented in Table 1.

For defining the loads to be applied in these tests, the selectionwas based on considerations proposed by the project partners [16]and previous results presented by NASA [20] leading to an initialDK of approximately 6 MPa m1/2.

Taking into account this reference value and the stress intensityfactor calibration proposed in the standard ASTM [17],

DK ¼ ðPmax � R � PmaxÞB

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffip � aW2 sec

p � aW

rð2Þ

calculations were performed in order to assess the load to be usedfor each R ratio.

Fig. 3. Through crack

For all R ratios and considering the maximum loads used, the fi-nal DK (for the maximum crack size 2a = 50 mm) is 18.41 MPa m1/

2. For R = 0.1 a maximum load of 21.76 kN was used.A maximum load of 35 kN was considered for the specimens

tested with R = 0.5. If so, for this load ratio the tests will be per-formed for DK values between 5.36 MPa m1/2 and 16.45 MPa m1/2.For R = 0.8, and considering a maximum load of 35 kN, DK valueswould be between 2.14 MPa m1/2 and 6.58 MPa m1/2.

The test output of these tests includes: da/dN curve and itsrespective data points as well as the basic a–N data, K–R curveASTM [18] and the load–displacement curve on which it is based,description and photography of each broken specimen, fractogra-phy and some conclusions about the crack growth in the differentmaterial zones.

The fatigue pre-cracking was performed under constant ampli-tude loading and crack length was visually determined using atravelling microscope.

For the R-curve determination a displacement clip gage wasused with screw attached knife edges spanning the crack at a cer-tain span. This procedure to attach the clip gages to the specimen isin accordance to the standard ASTM [18].

3.2. Test procedure and setup

Crack growth tests were carried out in a MTS 321.21 with a250 kN load cell, at a frequency of 6 Hz. A mechanical grip fixturewas developed in order to comply with the specimen geometryand maximum loads defined in the general test plan [16].

The crack extension was measured using a travelling micro-scope in each side of the specimen. The geometry of each specimenwas accurately measured before each test, especially in the sectionof the material affected by the welding process.

4. Final fracture: test procedure and setup

After the fatigue crack growth tests brought the crack length(2a) to approximately 50 mm, the specimens were loaded to frac-ture at room temperature. These tests were carried out under dis-placement control, at a speed of 0.1 mm/min.

The output of each test is a curve that relates the load P remo-tely applied measured by the load cell and the displacement v mea-sured by a clip gage.

Two M3 holes were made in each M(T) specimen to support theclip gage, as shown in Fig. 4.

An attempt was made to analyse tests following the ASTM [18]in order to obtain the K–R curves. The displacement v is measuredover 2y (y is equal to 12.5 mm). The effective crack length (ae) isobtained from the value of the compliance v/P (the inverse of theslope of different straight lines drawn from the origin), which is re-lated to ae by a calibration equation defined in ASTM [18]. This

locations [16].

Fig. 5. SN test results for R = 0.1 and R = 0.8.

Fig. 4. M(T) specimen, holes location to support the clip gage.

Table 3Specimen dimension according to [17].

B (mm) 5W (mm) 802a, Initial crack size (mm) 9.42a, Crack size before loading to fracture (mm) 50

4 P.M.G.P. Moreira et al. / Theoretical and Applied Fracture Mechanics 60 (2012) 1–9

calibration is adjusted to verify the initial crack length, which isknown.

The value of Kr is obtained from ae and from the value of P of theP�v curves, using the Eq. (2), ASTM [18],

The data points of the P�v curves that can be considered for fur-ther calculations are the ones that verify the following Eq. (3) (netsection stress criterion Rv):

Rv ¼rnet

ryld¼ P

ryldðB� 2apÞt< 1 ð3Þ

where rnet is the average stress on the remaining section, andryld the material yield stress.

This criterion has the purpose of assuring that the remainingsection of the specimen at each point is large enough to avoid plas-tic collapse. The variable ap is the physical crack length, obtainedfor each points by iteration using the following equations.

ry ¼1

2pK

ryld

� �2

ð4Þ

ae ¼ ap þ ry ð5Þ

where ry is the plastic zone size.Since the crack growth tests were stopped when the first of the

four crack length measures (two tips, front and back) reached25 mm, the initial crack length of the K–R curve test (the averageof the four final measurements of the crack growth test) is notthe same for all the specimens. The crack length to one side ofthe notch is different from the crack length to the other, sometimessignificantly. This means that the cracked specimens are not sym-metrical before loading to fracture. The crack length is also differ-ent as measured in the two surfaces of the specimen, front andback.

For this specimen width (w = 80mm), the crack lengths beforeloading to fracture are too large, for all specimens, because they fall

out of the acceptable range for 2a0 (0.25w–0.4w), defined in ASTM[18].

5. Results

5.1. Fatigue life

For the specimens tested at 70% of ryield substantial fatigue lifescatter was found; the first specimen fractured at around 800,000cycles and the second specimen remained un-fractured with a fa-tigue life of 107 cycles. Taking into account this observation thethreshold of infinite life was considered to be 107 cycles.

The plot of the fatigue lives for R = 0.1 is presented in Fig. 5.In [21], it was verified that base material fatigue tests with

R = 0.1 have a maximal stress corresponding to 105 cycles of the or-der of 350–400 MPa, whereas in this work for FSW joints, testedunder R = 0.1 with the weldment perpendicular to the load, maxi-mal stress is of the order of 260–280 MPa for the same number ofcycles, a reduction of just approximately 30%. It is noted that [8,22]present data for AA2195 with a fatigue life of 105 corresponding toapproximately rmax = 240 MPa, however the test conditions weredifferent – different specimen orientation and load frequency50 Hz. For the sake of comparison purposes, other results for Al–Li alloys given by refs. [23,24] where used. Ref. [23] gives armax � 220 MPa at 105 cycles, for AA2050; Ref. [24] presents, forthe case of AA2198, rmax � 250 MPa and rmax � 225 MPa at 105 cy-cles, for longitudinal and transversal orientations respectively. Thisillustrates the good performance attained by friction stir welds inthe present paper.

The location of the fracture surface in each SN specimen is pre-sented in Table 4.

A macrograph of each type of fracture location found in the SNfatigue specimens is presented in Fig. 6. The specimens chosen tobe presented were: IT23, IT26 and IT28, see Fig. 6a–c respectively.

5.2. Fatigue crack growth

27 specimens were tested. The following paragraphs presentsequentially plots of a vs. N for R = 0.1, 0.5 and 0.8.

5.2.1. Specimens tested with R = 0.1Analysing the fatigue crack propagation data, Fig. 7, it was ver-

ified that the higher values of fatigue crack growth rates werefound for the specimens containing a notch in the centre of theweld, and the lower crack growth rates for base material speci-mens. The specimens with a notch in the HAZ present intermediateresults. Analysing the base material specimens it is verified that forthe lower DK values, between 6 and 8 MPa m1/2, there is a steadycrack growth rate. This zone corresponds to a fatigue crack

Fig. 6. Macrograph of each type of fracture location – specimens (a) IT23, (b) IT26,and (c) IT24.

Table 4Fracture surface identification in each SN specimen.

Specimen rmax

(MPa)R Advancing or

retreatingCrack initiationsurface

Weldzone

Notation

IT 23 300 0.1 R S T A – AdvancingIT 25 R S T R – RetreatingIT 26 210 R S T C – Centre (weld line)IT 27 Did not break S – Shoulder surface (crown surface)IT 28 255 R S TIT 29 R S T B – Bottom surface (root surface)IT 30 300 0.8 Did not breakIT 31 345 Did not break T – Transition between material affected and material not affected by the shoulder

(shoulder limit)IT 32 390 Did not breakIT 33 450 A S TIT 24 C B N N – Nugget

P.M.G.P. Moreira et al. / Theoretical and Applied Fracture Mechanics 60 (2012) 1–9 5

growing in a irregular surface. In this figure it is also verified thatfor DK values between 10 and 11 MPa m1/2, data for all specimensconverge for the same crack growth rate. As a future work, it wouldbe of interest to test similar specimens at DK values higher than18 MPa m1/2.

5.2.2. Specimens tested with R = 0.5Analysing the crack propagation data, Fig. 8, it was verified that

for R = 0.5 all types of specimens present similar crack growthresults.

5.2.3. Specimens tested with R = 0.8Analysing the crack propagation data, Fig. 9, it was verified that

for R = 0.8 all type of specimens present similar crack growthresults.

Specimen M(T)34 presented a unique and surprising behaviourconsisting of final fracture occurring during the fatigue crackgrowth tests at a crack length of approximately 21 mm. The frac-ture surface is quite irregular and suggests fatigue crack growthin the interior of the specimen not noticed at the surfaces.

5.2.4. Fatigue crack growth results comparisonIn order to make a direct comparison of the R effect on each type

of specimen tested, comparison plots of the crack propagation rateper specimen type are presented in Fig. 10. It should be empha-sised that, for R = 0.1 the crack growth in the first 9 mm of basematerial specimens present a particular behaviour, as mentionedin Section 5.2.1.

5.3. Toughness

5.3.1. P–v curvesFigs. 11–13 present the P–v curves obtained for the Weld, HAZ

and BM specimens, respectively. The graphs’ scales are equal to al-low the comparison of the different types of specimen curves.

It was verified that the present specimens’ geometry falls sub-stantially outside the range of applicability of ASTM E561; there-fore, the treatment of test data was done as presented in thenext section.

5.3.2. Fracture toughness determinationIn order to estimate the Kc fracture toughness of the base mate-

rial specimens Eq. (2), ASTM [18], was used. In this equation a istaken as the average right and left crack lengths of crack after fati-gue crack growth test and before final residual strength test. Loadcorresponds to first pop-in. In Table 5 the Kc toughness values forbase material specimens are presented.

A calculation of Jo according to the EFAM procedure described,e.g. in [25] was performed for the M(T) specimens numbers 13and 21, which are the most representative of Weld and HAZ spec-imens type in the sense that they present the greater similarity ofleft and right cracks. These specimens correspond to weld nuggetmaterial and heat affect zone respectively. The necessary elasticconstant was assumed as E = 70GPa and maximum load of P–vcurves was used.

To carry out the calculation the physical value of initial cracklength was used. The area beneath the P–v curves was determinedby numerical integration.

With this approximate estimation of KJ for the Weld and HAZspecimens a single toughness figure is obtained for each specimen.It is noticed that the consideration of the elastic plastic behaviourgives a substantial increase in the toughness estimation; for exam-ple for specimen M(T) 21 the elastic K is 55.3 MPa m1/2 whereasKJ = 93.6 MPa m1/2.

Fig. 7. Fatigue crack propagation data for specimens tested with R = 0.1; (a) basematerial specimens, (b) weld specimens, and (c) HAZ specimen. Fig. 8. Fatigue crack propagation data for specimens tested with R = 0.5; (a) base

material specimens, (b) weld specimens, and (c) HAZ specimen.

6 P.M.G.P. Moreira et al. / Theoretical and Applied Fracture Mechanics 60 (2012) 1–9

The proper toughness characterisation for these specimensshould include the description of the resistance curve behaviourof the material. However, due to the choice of specimen geometry,and scarcity of material, valid ASTM E561 K–R curves were deter-mined for just a small range of Da values. Therefore, the presentapproximation provides a single number for the toughness of theWeld and HAZ specimens, which may be compared with the Kc va-lue obtained at first pop-in for the BM specimen.

KJ values were calculated according to the already mentionedEFAM procedure [25] for all R curve tests performed.

Within the limits of an approximation, the present work con-firms that base material toughness (average value 42.3 MPa m1/2,obtained from data of Table 5) is lower than that of the friction stirwelded material. The average values of KJ are 99.6 MPa m1/2 forweld material, and 92.8 MPa m1/2 for HAZ material.

6. Discussion

Plots comparing the present results with data presented in [20]for the same material are shown in Figs. 14 and 15. It should be

Fig. 9. Fatigue crack propagation data for specimens tested with R = 0.8; (a) basematerial specimens, (b) weld specimens, and (c) HAZ specimen.

Fig. 10. Comparison of crack propagation for different type of specimens. (a) Basematerial specimens; (b) HAZ material specimens; and (c) weld material specimens.

P.M.G.P. Moreira et al. / Theoretical and Applied Fracture Mechanics 60 (2012) 1–9 7

noted that data of [20] for thickness of 0.25 and 0.50 inches at 75�Fonly was only used, although the NASA reference includes moredata for other thicknesses and temperatures. The present resultsshow a reasonable agreement with NASA results for R ratios 0.1and 0.5 and for both types of specimens (base and welded mate-rial). It is noted in Fig. 14 that base material fatigue crack propaga-tion in orientation L–T and T–L given by [20] are approximatelyidentical, and close to the present T–L result. Concerning FSW sam-ples, [20] only gives only data for L–T orientation whereas the pres-ent results are T–L.

[26] also presents fatigue crack growth results of FSW speci-mens of AA2195 alloy. The geometry is M(T) with 12.5 mm thick-ness and L–T orientation, with the crack perpendicular to theweldment. The results for FSW specimens are not directly compa-rable with the results of the present paper because in [26] speci-mens are L–T whereas the present work is on T–L specimens.However, [26] gives data for base material and it was found thatthe present results show substantial agreement with [26].

In [27] fatigue crack growth behaviour of AA2195-T8 base andfriction stir welded material was analysed. It was found that thebase material fatigue crack growth is similar to the one presented

Fig. 11. P–v curve of weld specimens.

Fig. 12. P–v curve of HAZ specimens.

Fig. 13. P–v curve of BM specimens.

Table 5Kc determination for BM specimens.

M(T) ap0 (mm) Fpop-in (kN) Fmax (kN) Kc (MPa m1/2)

31 21.33 57.9 64.6 46.032 23.38 49.8 56.1 43.235 21.12 53.3 68.1 42.033 24.65 47.5 59.7 43.536 24.21 43.8 62.6 39.437 23.16 49.3 63.8 42.429 24.44 45.0 62.3 40.930 24.66 45.1 55.5 41.3

Fig. 14. Comparison of results for base material.

Fig. 15. Comparison of results for weld specimens.

8 P.M.G.P. Moreira et al. / Theoretical and Applied Fracture Mechanics 60 (2012) 1–9

in this study. Nevertheless, in [27] lower crack growth rates thanthe present ones were found for the case of friction stir weldedspecimens. The tests of [27] were performed using C(T) 8 mm thickspecimens at R = 0.1 that were obtained by reduction of the initialplate thickness (12.7 mm). The lower crack growth rates found in[27] may however be a result of the residual stress field of theC(T) specimens, since it is known that this type of specimen pre-sents important compressive residual stresses in the crack frontarea, as was recently shown, albeit on different material in [28].

The above discussion indicates that the FSW samples tested forfatigue crack growth rate determination show good results as com-pared with the corresponding base material performance.

Toughness values for base material are lower than for weldedmaterial. The base material average value of 42.3 MPa m1/2 com-pares favourably with data from the literature, as [29] forAA2090 or [30] for AA2195, where toughness values are lowerfor equivalent yield stress.

7. Concluding remarks

Good SN fatigue behaviour was found testing at room tempera-ture FSW specimens of AA2195-T8X. When comparing base mate-rial and FSW SN data obtained with R = 0.1 and 105 cycles thereduction of maximum stress of FSW specimens is of the order ofjust 30%. A comparison with published SN data including other

P.M.G.P. Moreira et al. / Theoretical and Applied Fracture Mechanics 60 (2012) 1–9 9

Al–Li alloys indicates the very good performance of the presentFSW joints.

Base material and friction stir welded Al–Li 2195-T8X middletension specimens were subjected for fatigue crack growth testingat room temperature. Existing fatigue crack growth data for FSWAA2195 was obtained using specimens with L–T orientation. Thepresent work provides data for T–L specimens. Fatigue crack prop-agation data did not show significant stress ratio dependency forFSW material. Concerning the AA2195 base material, some stressratio dependency was verified. In general, the fatigue crack propa-gation data was consistent with data available in literature for sim-ilar materials.

It was verified that FSW specimens present generally substan-tially higher fracture toughness than base material specimens.

Acknowledgements

Dr. Pedro M.G.P. Moreira acknowledges ‘POPH – QREN-Tipolo-gia 4.2 – Promoção do Emprego Científico funded by ESF and byMCTES. The collaboration of Paulo C.M. Azevedo and Sérgio M.O.Tavares is acknowledged.

References

[1] E.J. Lavernia, T.S. Srivatsan, F.A. Mohamed, Review – strength, deformation,fracture behaviour and ductility of aluminium–lithium alloys, Journal ofMaterials Science 25 (1990) 1137–1158.

[2] E.A. Starke, J.T. Staley, Application of modern aluminum alloys to aircraft,Progress in Aerospace Sciences 32 (1996) 131–172.

[3] Committee on New Materials for Advanced Civil Aircraft, New materials fornext-generation commercial transports, National Academy Press, Washington,DC, 1996.

[4] D.L. Chen, M.C. Chaturvedi, Near-threshold fatigue crack growth behavior of2195 aluminum–lithium-alloy – prediction of crack propagation direction andinfluence of stress ratio, Metallurgical and Materials Transactions A 31A (2000)1531–1541.

[5] D.L. Chen, M.C. Chaturvedi, Effects of welding and weld heat-affected zonesimulation on the microstructure and mechanical behavior of a 2195aluminum–lithium alloy, Metallurgical and Materials Transactions A 32A(2001) 2729–2741.

[6] R.J. Rioja, Fabrication methods to manufacture isotropic Al–Li alloys andproducts for space and aerospace applications, Materials Science andEngineering A257 (1998) 100–107.

[7] A. Kostrivas, J.C. Lippold, Weldability of Li–bearing aluminium alloys,International Materials Reviews 44 (6) (1999) 217–237.

[8] M.C. Chaturvedi, D.L. Chen, Effect of specimen orientation and welding on thefracture and fatigue properties of 2195 Al–Li alloy, Materials Science andEngineering A 387–389 (2004) 465–469.

[9] R.S. Mishra, Z.Y. Ma, Friction stir welding and processing, Materials Science &Engineering R – Reports 50 (1–2) (2005) 1–78.

[10] P.L. Threadgill, A.J. Leonard, H.R. Shercliff, P.J. Withers, Friction stir welding ofaluminium alloys, International Materials Reviews 54 (2) (2009) 49–93.

[11] R. Nandan, T. DebRoy, H.K.D.H. Bhadeshia, Recent advances in friction-stirwelding – process, weldment structure and properties, Progress in MaterialsScience 53 (6) (2008) 980–1023.

[12] MT-Aerospace, Cut Off Plan, in Damage Tolerance of Cryogenic PressureVessels, ESA Technical Research Programme, 2008.

[13] N. Eigen, U. Glaser, G. Sinnema, P.M.G.P. Moreira, J. Schneider, E Sattler,Microstructure and properties of AA 2195 T8 material and friction stir welds,in: 11th European Spacecraft Structures, Materials and Mechanical TestingConference (ECSSMMT), Toulouse, France, 2009.

[14] ASTM E466, Constant Amplitude Axial Fatigue Tests of Metallic Materials,2007.

[15] ASTM E468, Standard Practice for Presentation of Constant Amplitude FatigueTest Results for Metallic Materials, 2004.

[16] MT-Aerospace, Fracture Mechanics Test Plan, in Damage Tolerance ofCryogenic Pressure Vessels, ESA Technical Research Programme, 2008.

[17] ASTM E647, Standard Test Method for Measurement of Fatigue Crack GrowthRates, 2007.

[18] ASTM E561, Standard Practice for R-Curve Determination, 2010.[19] ASTM E1823, Standard Terminology Relating to Fatigue and Fracture Testing,

2010.[20] R.A. Hafley, J.A. Wagner, M.S. Domack, Fatigue Crack Growth Rate Test Results

for Al–Li 2195 Parent Metal, Variable Polarity Plasma Arc Welds and FrictionStir Welds, NASA/TM-2000-210098, 2000.

[21] J.P. Bonnafé, D. Gabard, E. Grosjean. Aluminium lithium alloys use for reusablefuture launcher cryogenic metallic tanks. In: 5th Conference on AerospaceMaterials, Processes, and Environmental Technology, Huntsville, Alabama,NASA/CP-2003-212931, 2002.

[22] M.C. Chaturvedi, D.L. Chen, Microstructural characterization and fatigueproperties of 2195 Al–Li alloy, Materials Science Forum 519–521 (2006)147–152.

[23] Ph. Lequeu, K.P. Smith, A. Daniélou, Aluminum–copper–lithium alloy 2050developed for medium to thick plate, Journal of Materials Engineering andPerformance 19 (2010) 841–847.

[24] P. Cavaliere, M. Cabibbo, F. Panella, A. Squillace, 2198 Al–Li plates joined byfriction stir welding: mechanical and microstructural behavior, Materials andDesign 30 (2009) 3622–3631.

[25] K.H. Schwalbe et al., EFAM GTP 02 – the GKSS test procedure for determiningthe fracture behavior of materials, GKSS Report GKSS 2002/242002. pp. 22, 43,103.

[26] O. Hatamleh, A comprehensive investigation on the effects of laser and shotpeening on fatigue crack growth in friction stir welded AA 2195 joints,International Journal of Fatigue 31 (2009) 974–988.

[27] Yu.E. Ma, P. Staron, T. Fischer, P.E. Irving, Size effects on residual stress andfatigue crack growth in friction stir welded 2195–T8 aluminium – Part I:experiments, International Journal of Fatigue 33 (2011) 1417–1425.

[28] V. Richter-Trummer, P.M.S.T. de Castro, The through-the-thicknessmeasurement of residual stress in a thick welded steel compact tensionspecimen by the contour method, The Journal of Strain Analysis forEngineering Design 46 (2011) 315–322.

[29] C.G. Bennett, S.P. Lynch, R.B. Nethercott, M. Kerr, E.D. Sweet, Fracturetoughness of 2090 Al–Li–Cu extrusions with high and low hydrogencontents, Materials Science and Engineering A 247 (1998) 32–39.

[30] P.S. Chen, A.K. Kuruvilla, T.W. Malone, W.P. Stanton, The effects of artificialaging on the microstructure and fracture toughness of Al–Cu–Li alloy 2195,Journal of Materials Engineering and Performance 7 (5) (1998) 682–690.