Professional Documents
Culture Documents
201208
Abstract
Aluminium alloy 5083 is used in the fabrication of lightweight, high-speed marine vessels. The high cyclic service stresses on
such structures render welded joints in them fatigue-critical. This paper presents experimental investigations on fatigue crack
behaviour of welded 5083-H321 aluminium alloy plates. Crack propagation in the heat-affected zone of welded specimens was
determined from tests conducted on single edge-notched tension specimens. Three-dimensional finite-element analysis was
employed to determine stress intensity factors for cracks in the weld line. The crack closure effects of weld residual stresses were
evaluated. The residual stresses significantly influence the crack growth rates. Fatigue behaviour of weld repairs in cracked plates
was also investigated. The weld-repair process significantly increases the grain size and the size of defects in the heat-affected
zone. The results indicate that weld repair of cracks in welded joints provides little improvement in residual life. 2002 Elsevier
Science Ltd. All rights reserved.
0261-3069r02r$ - see front matter 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 2 6 1 - 3 0 6 9 0 1 . 0 0 0 5 9 - 0
202 K. Shankar, W. Wu r Materials and Design 23 (2002) 201208
When cracks occur in the vicinity of weldments in usually derived from two-dimensional plane strain
structures, weld repairs are frequently considered for models containing edge cracks. Since the geometry of
crack repair in most cases to extend service life w3x. For welded plates differs from that of unwelded plates,
instance, in the construction of pipelines in the oil and traditional relations for stress-intensity factor calcula-
natural gas industry, when sections of a newly welded tions based on flat plate geometry may not be valid for
joint fail to pass inspection, it is common practice to welded joints. On the other hand, removing the weld
repair the weld by gouging out the initial weldments in reinforcement completely to make the geometry similar
the sub-standard sections and re-welding them. How- to that of a flat plate may be impractical, uneconomi-
ever, the effectiveness of weld-repairing cracks in an cal, or even impossible in practical engineering applica-
already welded joint is unclear, because of the increase tions. Using flat-plate equations for predicting fatigue
in residual stresses, distortion in geometry and the life for as-welded components may lead to overesti-
deterioration of the microstructure caused by the addi- mated results. Recent research employing three-dimen-
tional weld process. There is a considerable body of sional modelling has indicated that fatigue cracks on
literature on the effects of a single weld on the fatigue the weld toe cannot be accurately represented by two-
life of steel and aluminium alloys, yet data on weld dimensional models which ignore the weld profile
repair of weldments are scarce. However, just as crack geometry w8,9x. Hence, in the present study, a three-di-
growth data in weldments is very limited, quantitative mensional finite-element model of the butt weld joint
information on the effects of a second weld in a welded was carried out to compute the stress-intensity factor
material, either on overall fatigue performance or on SIF. solutions for weld toe cracks. These were then
crack growth behaviour, is not available. There have utilised to determine effective stress-intensity factors
been few previous investigations on the effects of weld by taking into account crack closure effects.
repair of cracks in welded structures. A section of the three-dimensional finite-element
The aim of the present work was to investigate crack model showing the weld reinforcement with the crack
growth behaviour in welded and weld-repaired alu- at the weld toe is shown in Fig. 1a. The notch intro-
minium alloy materials, specifically that of 5083-H321 duced on one side of the specimen to initiate the crack
plate material. This was achieved in two phases. In the at the toe of the reinforcement may be noted. The tip
first phase, crack growth rates in welded 5083 of the crack is evident in the figure as a darker square
aluminium alloy plates and in the unwelded material of finer mesh further inwards along the toe. Fig. 1b
were experimentally determined and compared, taking shows the arrangement of the three-dimensional singu-
into account aspects such as weld geometry and resid- lar elements around the crack tip. The stress intensity
ual stresses in the heat-affected zone. In the second factor K 1 is computed from the relative displacements
phase, the propagation behaviour of weld repairs of of the crack faces with respect to one another using a
cracks in existing weldments in the 5083 material were linear elastic-fracture mechanics approach w10x. The
examined, with a view to establishing whether weld geometry factors for the stress intensity solutions stress
repair of cracks in weldments is a viable means of intensity values per unit stress . at the toe of the rein-
extending service life. forcement on the outer surface and mid-plane of the
Previous work by many researchers w49x has indi- plate, obtained from the three-dimensional finite-ele-
cated that the weld geometry and weld residual stresses ment analysis, are plotted against the normalised crack
WRS. can significantly affect the fatigue crack-growth length in Fig. 2, along with the two-dimensional solu-
results in welded joints. The influence of residual tions. It is evident that the stress intensity values at the
stresses is usually considered by replacing the normal top of surface where the reinforcement meets the
stress-intensity factor range, K, by the effective plates top surface are approximately 10% higher than
stress-intensity factor range, K eff , which accounts for those given by the two-dimensional solution, while the
the influence of crack closure due to the weld residual stress intensities at the mid-plane are somewhat lower.
stresses w4,7x. The influence of weld residual stresses on
crack closure was experimentally evaluated by mea-
surement of crack opening displacements. Metallurgi-
3. Specimen material and geometry
cal examination of welded joints and fractography of
fracture surfaces were conducted using an optical mi-
3.1. Material
croscope and SEM.
Fig. 1. a. Three-dimensional finite-element model of the weld reinforcement; and b. singular elements around the tip of the crack.
Table 1
Chemical composition of 5083-H321 plate
Element Wt.%
Mg 4.04.9
Mn 0.41.0
Cr 0.050.25
Si 0.4 max
Fe 0.4 max
Ti 0.15 max
Cu 0.1 max
Fig. 2. Geometry factors f arw . for stress intensity solutions from Zn 0.25 max
the three-dimensional FE model top surface and mid-plane. and Others 0.15 max
2-D solution. Al Balance
204 K. Shankar, W. Wu r Materials and Design 23 (2002) 201208
Fig. 4. Stress strain behaviour of unwelded, welded and weld-repaired joints under static tensile load.
K. Shankar, W. Wu r Materials and Design 23 (2002) 201208 205
Fig. 6. Effective stress intensity factors taking into account crack closure effects.
206 K. Shankar, W. Wu r Materials and Design 23 (2002) 201208
Fig. 7. Crack growth rates of unwelded and as-welded material plotted against nominal stress intensity factors.
largely due to neglecting the crack closure effects of double-sided repairs have a much higher crack-growth
the residual stresses in the welded specimens. When rate, especially during the early part of crack growth.
plotted against the effective stress-intensity factor range
K eff Fig. 8., the d ard N curve for the welded plate
shifts to the left and decreases in gradient, coming 7. Fractography
close to and falling within the scatter band of the
unwelded base material. These results indicate that As shown in Fig. 5b, the weld-repaired material
fatigue-crack growth behaviour for the 5083-H321 base exhibits a much higher level of grain growth and
plate and that of the welded plate in the HAZ is quite porosity compared to the as-welded material. Apart
similar. from having coarse grains, the microvoids and macrop-
The fatigue-crack growth behaviour of samples with orosity at the secondary welds in the weld-repaired
single- and double-sided repairs is plotted against the specimens are significantly larger, both in number and
effective stress-intensity factors in Fig. 9, along with size. These defects interconnect, becoming micro-
those of the as-welded plates for comparison. It is cracks, and subsequently form large cracks when sub-
evident that the crack propagation rates of joints with jected to cyclic fatigue loads. Therefore, the weld-
single-sided repair are quite close to those of the repair process appears to increase crack growth rate,
as-welded plates which are again similar to those of and consequently to decrease fatigue life. Fig. 10 shows
the parent material.. In contrast, the samples with the fatigue fracture-surface examination for unwelded,
Fig. 8. Crack growth rates of unwelded and as-welded material plotted against effective stress intensity factors.
K. Shankar, W. Wu r Materials and Design 23 (2002) 201208 207
Fig. 9. Comparison of crack growth rates for single- and double-sided repairs with that of the as-welded plates against effective SIF range..
welded and weld-repaired plates by SEM. The fracture reinforcement are repaired by welding only on one
surface of welded plate appears quite similar to that of side, the crack growth rates of the repaired specimens
the unwelded plate, whereas that of the weld-repaired are quite similar in magnitude to those of the as-welded
plate Fig. 10c. exhibits indications of grain pull-out, plates and the parent material, when plotted against
owing to extreme porosity. the effective stress-intensity factor ranges. On the other
hand, when the repairs are effected by welding on both
sides of the crack in the weldment, the propagation
rates for cracks in the fusion zone seem to be much
8. Conclusion
higher, especially in the early stages of crack growth. It
is possible that this may have resulted from the dif-
Crack growth behaviour in welded aluminium-alloy ficulty in establishing the effective stress-intensity fac-
5083-H321 plates was experimentally investigated. It tors for the double-sided repair specimens, due to the
was found that when effective stress-intensity factors difficulty of accurately measuring the crack-opening
are computed, taking into account the crack closure displacements. In any case, it should be noted that the
due to plasticity and weld residual stresses, the propa- residual life of the weld-repaired joints is even lower
gation rates of cracks in the weld toe are comparable to than that of the as-welded plates. This is attributed to
those in the parent material. The lower fatigue lives of the significantly larger size and greater number of
the welded joints are therefore mainly due to the defects introduced by the secondary welding process.
porosity and internal defects introduced by the weld This leads to the conclusion that repair of cracks in
process. It was found that when cracks in the weld weldments by subsequent welding is a poor means of
Fig. 10. Fracture surfaces of: a. unwelded; b. as-welded; and c. weld-repaired specimens.
208 K. Shankar, W. Wu r Materials and Design 23 (2002) 201208
enhancing fatigue life. While it may be possible to Asian-Pacific conference, Sydney, August 1993, published by
reduce the size and number of defects introduced by A.A. Balkema. Rotterdam, 1993:883888.
w6x Kang KJ, Song JH, Earmm YY. Behaviour of fatigue crack
the weld-repair process by pre- and post-treatment of
growth and closure through tensile residual stress field under
the material, this aspect has not been studied in this
compressive applied loading. Fatigue Fract Eng Mater Struct
preliminary investigation. The effectiveness of pre- and 1989;12:363376.
post-treatments in improving the quality of welds and w7 x Itoh YZ, Suruga S, Kashiwaya H. Eng Fract Mech
in ameliorating weld residual stresses needs to be in- 1989;333.:397407.
vestigated before secondary welding can be accepted as w8x Nguyen TN, Wahab MA. The effect of weld geometry and
a viable means of restoring residual fatigue life of residual stress on the fatigue of welding joints under combined
cracked welded joints. loading. J Mater Process Technol 1998;77:201208.
w9x Nguyen TN, Wahab MA. A mathematical model to predict
geometry magnification factor for stress intensity factor of butt
References welded joints. Advanced Jointing Technologies for New Mate-
rials, American Welding Society Conference Proceedings,
w1x Ghose DJ, Nappi NS, Wiernecki CJ. Residual Strength of 1994:142154.
Damaged Marine Structures. Springfield, VA: Ship Structural w10x Anderson TL. Fracture Mechanics, Fundamentals and Applica-
Committee, NTIS, 1995. NTIS Report SSC-381.
tions. 2nd CRC Press Inc, 1995.
w2x Beg RG. Marine Structural Integrity Programs MSIP.. Spring-
w11x Elber W. The significance of fatigue crack closure. Damage
field, VA: Ship Structural Committee, NTIS, 1992.
tolerance in aircraft structure. Philadelphia: ASTM, 1971:
NTIS Report SSC-365.
w3x Slater BYG. The effect of repair welds on service performance. 230242. ASTM STP 486.
w12x Miki C, Mori T, Tajima J. Struct Eng Earthquake Eng
Weld J 1985;643.:2229.
w4x Link LR. Fatigue crack growth of weldments. In: McHenry P, 1986;31.:187194.
editor. Fatigue Fracture Testing of Weldments. Philadelphia: w13x Maddox SJ, Webber D. In: Hoeppner DW, editor. Fatigue
ASTM, 1990:1633. ASTM STP 1058. Testing of Weldments. Philadelphia: ASTM, 1978.
w5x Nguyen TN, Wahab MA. The effect of butt weld geometry ASTM STP 648.
parameters on stress intensity factor and fatigue life. In: Valli- w14x Donald JK. In: Newman JC, Elber W, editors. Mechanics of
appan S, Pulman VA, Tin-Loi E, editors. Computational me- fatigue crack closure. Philadelphia: ASTM, 1988:222229.
chanics from concepts to computations Proceedings of the ASTM STP 982.