Size effects on residual stress and fatigue crack growth in friction stir welded 2195-T8 aluminium – Part I: Experiments

https://doi.org/10.1016/j.ijfatigue.2011.05.006Get rights and content

Abstract

Residual stress fields were measured in three different sizes of Compact-Tension (C(T)) and eccentrically loaded single edge notch (ESE(T)) specimens containing transverse or longitudinal welds. The effect of size on residual stress profiles was studied. Fatigue crack growth tests were carried out with cracks growing into or away from the weld line, as well as growing along the weld centre line. Effects of weld residual stresses on fatigue crack growth rates parallel and perpendicular to the friction stir welds were studied. It was found that compressive residual stresses around the sample notch had significant retarding effects on both crack initiation and crack growth rates for cracks growing towards the weld line. Effects of residual stress on crack growth rates declined with increasing crack length. When cracks grew parallel to the weld line in C(T) samples the crack growth rate was around 20% lower than in parent material.

Highlights

Fatigue crack growth measurements on friction stir welded aluminium–Lithium alloys. ► Detailed measurements of residual stress as a function of sample size and orientation with respect to the weld. ► End to end experiments and predictive model construction for fatigue crack growth in welds. ► Validation of Kresid based models for effect of residual stress in welds.

Introduction

The effects of residual stresses on fatigue crack growth rates in welded samples have been studied by many researchers in the past 20 years. As residual stress measurement techniques improve more comprehensive knowledge of the form of weld residual stress fields has become more widely appreciated. Models to predict crack growth rates in the presence of residual stresses have been available for some years, but only recently has knowledge of residual stress fields become sufficiently comprehensive to allow quantitative model validation. A particular difficulty is the mixture of different variables contained within a weld which makes it hard to distinguish effects of microstructure and hardness changes within the weld from those due to residual stress. A related question is whether there can be “intrinsic” crack growth resistance within a weld separate from effects of residual stress. The former is something that will be unchanging with sample geometry, the latter will depend on sample geometry and post weld machining as well as on the weld process itself.

In order to quantify residual stress effects in different sample sizes and geometries, in this research samples of C(T) and ESE(T) geometry in three sizes were manufactured from plates of 2195-T8 Al–Li alloy containing friction stir welds. The residual stress profiles were measured by the neutron diffraction method before fatigue testing. The weld crack growth rates and crack paths were compared with those found in parent material. Experimental results are reported in Part I; Part II reports on model development to predict growth rates using measured residual stresses based on the Kres approach.

Several methods have been used to measure the residual stress distribution in welded plates. These include: synchrotron X-ray scanning [1]; sample slitting [2]; neutron diffraction [3], [4], [5]; cut compliance [6], [7]; hole drilling [8]; magnetizing stress indication [9], and the contour method [10]. Stress fields measured by all techniques show similar features with a double-peak in longitudinal stress with the maxima in the heat affected zones on each side of the weld line, with a smaller stress in the central weld line (the nugget). Residual stresses are greatest on the tool advancing side of the weld than on the retreating.

Residual stress effects on fatigue crack growth of FSW aluminium have been studied in previous research [5], [6], [7], [11], [12], [13], [14], [15], [16], [17], [18], [19]. The major conclusions of this work are as follows.

  • (1)

    Residual stress is the most important parameter influencing fatigue crack growth rates, with tensile stresses increasing crack growth rates, and compressive stresses decreasing them [5], [6], [7], [11], [12], [13], [14], [15], [16], [17], [18], [19].

  • (2)

    Residual stresses are not the only significant influence; microstructure of the FSW nugget region together with its local toughness and ductility play a role for cracks in the nugget [6], [7].

  • (3)

    Redistribution and relaxation of residual stress as the crack grows has important effects on crack growth [5].

Despite substantial work in definition of the influencing parameters there are still relatively few examples of work where crack growth rates and residual stresses have both been measured and the results of model predictions can be validated across a range of geometries and orientations of the crack and weld [13], [14], [15], [16]. This is particularly true for sample size effects, where sample size will influence residual stresses which in turn will influence crack growth rates.

Section snippets

Parent material properties

The composition and mechanical properties of 2195-T8 parent material are shown in Table 1, Table 2 [20]. This “third generation” Al–Li alloy, 2195-T8 has a greater Cu/Li ratio than the second generation alloys 2090 and 2091 [21]. Compared with 2024 (0.2% proof strength of 350 MPa and UTS of 490 MPa), the mechanical properties of 2195 are improved significantly, with 0.2% proof strength of 580 MPa and UTS of 615 MPa. Young’s modulus is increased to 79 GPa. Plates of this alloy have an elongated grain

Residual stresses in ESE(T) samples; weld parallel to sample long axis

Residual stress profiles for three sizes of ESE(T) samples (148 × 40 mm, 185 × 50 mm and 370 × 100 mm) are shown in Fig. 6. In Fig. 6a, X-direction stresses parallel to the weld direction have a double peak tensile residual stress field of similar form in all three specimens. The maximum tensile residual stress parallel to the weld ranged from 120 MPa, for the largest sample to 47 MPa, in the smallest. Remote from the weld line at the notch tip, the minimum stress value varied from about −130 MPa to −20 MPa

Stress field topography

The two sample geometries with three sample sizes allow a comprehensive picture to be developed of changes in weld residual stress produced by changes in sample size.

In all samples stresses parallel to the weld line follow the previously reported [1], [2], [3], [4], [5], [6], [7], [8], [9], [10] generic form. Tensile stress maxima occur 10 mm on either side of the weld centre, with a subsidiary minimum in the weld centre. At distance >10–20 mm from the weld line stresses parallel to the weld move

Conclusions

  • (1)

    Residual stress fields in notched C(T) and ESE(T) samples of 2195-T8 aluminium containing friction stir welds have a double peak tensile stress distribution of up to 120 MPa parallel to weld direction, with reduced stress on the weld centreline. Residual stresses perpendicular to the weld are generally smaller than those parallel to the weld. Around the notch tips, both stress components are compressive with stresses approaching −120 MPa in the largest samples.

  • (2)

    Samples with weld line lengths

Acknowledgements

The project is funded by the European Commission through a Framework Programme 6 project entitled “cost effective integral structures (COINS)” under contract number AST5-CT-2006-030825. Grateful thanks to Rob Maziarz, Airbus UK for performing the friction stir welding.

Cited by (65)

View all citing articles on Scopus
View full text