Available online at www.sciencedirect.

com
Procedia
Engineering
Procedia
Procedia Engineering
Engineering 00 (2009)
2 (2010) 000–000
1815–1821
www.elsevier.com/locate/procedia
www.elsevier.com/locate/procedia

Fatigue 2010

Fatigue behavior of friction stir spot welding and riveted joints in an

Al alloy
Malafaia, A. M. Sa *, Milan, M.T.b, Oliveira, M. F.a, Spinelli, D.a
a
Universidade de São Paulo, Av. Trabalhador São-carlense, 400, São Carlos CEP 13566-590, Brazil
b
Materials Institute of Brazil, R. Paulino Botelho de Abreu Sampaio, 881, São Carlos CEP 13561-060, Brazil

Received 8 March 2010; revised 10 March 2010; accepted 15 March 2010

Abstract

The main aim of this study was to compare the fatigue resistance of welded joints produced by FSSW process and riveted joints
of AA2024 alloy. The specimens welded with the best preliminary parameters determined by previous tensile shear tests were
tested in fatigue under load control, R=0.1, at room temperature. Two welding parameter sets were used, and P-N curves (load
versus cycles) were plotted, using 2x106 cycles as the fatigue life limit. A similar curve was obtained for riveted specimens. The
FSSW welding procedures were carried out in a CNC milling machining and the riveted specimens were produced in accordance
with aircraft industry parameters. Although the welded specimens presented almost the same results in the tensile shear tests, the
results were fairly lower than those observed for riveted joints in fatigue. The main failure mode observed in the welded joints
was shearing, besides some cases of crack propagation in the perpendicular load direction, while for riveted specimens occurred
mainly fretting nucleation followed by crack propagation in the perpendicular load direction. The evidences of shearing and
lower fatigue lives for welded specimens indicate that the joint geometry highly affects the joint properties, due probably to stress
concentrators presented locally.
c 2010 Published by Elsevier Ltd. Open access under CC BY-NC-ND license.

Keywords: FSSW; fatigue; riveted joints; failure modes.

1. Introduction

Friction Stir Spot Welding (FSSW) is a process developed recently and has been studied for applications in
automotive, aeronautic and other industries [1]. This welding technology is quite similar to Friction Stir Welding
process (FSW), and the main difference is the type of joint. In FSSW, the plates form a lap-joint and the tool
penetrates the plates only in a point. In the FSW process the plates are positioned in a butt-joint configuration and
the tool moves towards the joint direction. Therefore, the FSSW process can be explained in three distinct steps:
plunging, stirring and retracting. The process initiates with the tool plunging slowly, then the heat and the rotation
promote the stirring and finally the tool is retracted [2]. The retracting step is performed rapidly, just after the tool

* Corresponding author. Tel.: +55-16-8812-6534; fax: +55-16-3373-9590.

E-mail address: [email protected].

1877-7058 c 2010 Published by Elsevier Ltd. Open access under CC BY-NC-ND license.
doi:10.1016/j.proeng.2010.03.195
1816 A.M.S. Malafaia et al. / Procedia Engineering 2 (2010) 1815–1821

2 A. M. S Malafaia et al. / Procedia Engineering 00 (2010) 000–000

has reached the plunge depth or after a dwell time, that is a period in which the tool just turns in the maximum
plunge depth [3].
During the process the heating caused by the friction softens the material; the rotation of the tool pin is
responsible for the flow of material which occurs in the radial and axial directions, and the pressure applied by the
head of the tool allows the welding formation in the solid state. At the end of the process, after the tool’s retraction,
a characteristic hole is observed [3]. Due to the short period spent in the welding (2 to 5 s), the tool plunging motion
determines basically the heat generation, the joint formation, the weld mechanic properties, and the formation of the
plastically deformed material, helping the bond around the pin [4].
FSSW mimics the resistance spot welding process (RSW) and can be used to replace bond processes, such as
riveting, RSW, screwing and any other punctual method. FSSW overcomes the inherent difficulties found in other
methods, as high cost for fasteners, longer downtime caused by feeding issues and need for other operations (e.g.
drilling) [5].
The mechanical properties and microstructures of FSSW welded joints have been studied to evaluate the
parameter’s process effects. The main parameters studied were plunge rate, rotational speed, plunge depth and dwell
time. Another relevant parameter is the axial force during the welding. Some authors have measured the axial force
generated in FSSW [3, 4, 6]. On the other hand, it can be used axial force control with tool plunge rate fluctuation
[7]. Although both processes are possible, the first is mostly applied due to its easiness. The mechanical properties
measured in most works are restricted to tensile shear tests [3, 7, 8], however cross-tension tests have also been cited
[9].
Concerning fatigue tests, an interesting study using AA 6111-T4 alloy covered tensile shear and fatigue tests [10,
11]. Two different tool geometries, flat and concave, were evaluated and in both tests the concave tool generated the
best results. Other aspects discussed in the study were failure modes, microstructural analysis and a life prediction
model. Tran et al [12] also studied failure modes and a life prediction model using welding joints in tensile shear
and fatigue tests, and Chang et al [13] examined the failure modes in tensile shear tests.
Three distinct failure modes were observed for the samples tested in tensile shear: tear fracture, plug fracture and
shear fracture [13]. A tear fracture mode was observed in the most resistant joints followed by plug fracture. Both
failure modes occurred in the interface between the thermo-mechanically affected zone and the thermally affected
zone. Shear fracture mode was linked to the worst results, and in this case the weld ring suffered shear.
In the fatigue tests the failure modes occurs in a more complex way. Lin et al [10, 11] and Tran et al [12]
observed failure modes caused in both fatigue and tensile shear tests. Lin et al [10, 11], used a flat tool and verified a
shear fracture in the tensile shear tests and low-cycle fatigue tests. In the high-cycle fatigue tests two types of cracks
appeared, both starting at the tip of the bonded region. On the other hand, using the concave tool, no shear fracture
was observed, and in all tests the specimens presented cracks (in the stir zone for the tensile shear tests and in the
base material for the fatigue tests). These observations explained the best results yielded by the concave tool, due to
a different geometry profile.
The present study aimed to evaluate the fatigue behaviour of two sets of samples welded by FSSW process.
Furthermore, riveted joints were also tested for comparison, as riveting is the mostly used technique in the
aeronautic industry to join aluminum based plates, in which the AA2024 alloy is usual. The characterization of
failure modes was also performed for riveted and welded joints.

2. Materials and Methods

Lap shear specimens riveted and friction stir spot welding were prepared using AA2024-T3 alloy, 60x40x1,6 mm
plates and 40x20 mm overlap. The welding process was performed in a CNC milling cut using parameters based on
a previous study [14]. Two sample sets were welded changing only the plunge depth parameter (2.85 and 3.1mm). A
10 mm in diameter flat tool with a pin of 2.3 mm length and 4 mm in diameter were employed. The material used
was the H13 steel, and the pin was heat treated and carbonitrated. A concave profile tool was also tested in a
preliminary study. However, the results of tensile shear tests in the specimens welded with a concave profile tool
were poorer [14]. Table 1 shows the parameters used in each set of specimens, named WAxx and WBxx. The
riveted joints (RAxx) were fabricated at TAM Company, Brazil, using the same specifications of aircraft
maintenance: AA2117-T4 alloy with MIL-C-5541 superficial treatment, 4.0 mm diameter and 100º ± 30’
countersink.
 A.M.S. Malafaia et al. / Procedia Engineering 2 (2010) 1815–1821 1817

A. M. S Malafaia et al. / Procedia Engineering 00 (2010) 000–000 3

Table 1. FSSW parameters used to prepare the welded joints.

Specimens plunge depth plunge rate RPM dwell time

WA 2.85 mm 2 mm/s 4000 4s
WB 3.10 mm 2 mm/s 4000 4s

The fatigue tests were performed in an MTS test machine, in load control, R=0.1 and a frequency of 20Hz. End
tabs were added to the riveted and welded specimens, in order to avoid bending. A total of 2x106 cycles was
established as the fatigue limit and the highest load used was adjusted to generate a minimal life near 104 cycles.
The failure modes of the welded and riveted joints were identified and photographed. Analyses were performed to
verify the presence of cracks in the riveted joints that failed by the rivet shear.

3. Results and Discussion

Table 2 presents the maximum load used during the fatigue tests. The load values were determined based on a
previous tensile study and on the criteria explained above. The averages of the tensile shear failure load of the joints
in the preliminary study were 6.4 kN for riveted joints (RA), and 5.45 kN for WA and 4.95 for WB for FSSW
welded joints.

Table 2. Maximum loads applied during the fatigue tests.

Specimens Smax (kN)

RA 02 5.8
RA 01, 05 and 12 5.15
RA 03, 06 and 10 4.5
RA 04, 08 and 09 3.85
RA 07, 11 and 13 3.22
WA 05, 08 and 14; WB 02, 05 an 14 2.73
WA 04, 10 and 16; WB 01, 06 an 12 2.18
WA 02, 07 and 13; WB 03, 07 an 15 1.64
WB 04 and 10 1.36
WA 01, 12 and 15; WB 08 1.09

Figure 1 presents the curves of maximum load vs. fatigue life until specimen failure. The welded joints supported
proportionally lower loads than the riveted joints under the fatigue tests, considering the tensile shear results as a
reference. The maximum loads in the riveted joint tests were approximately twice that used in the welded joints to
achieve the same lives, as the rivet is homogeneous while the welded material presents only a heterogeneous
metallurgical bond. This heterogeneity is caused by microstructural changes with a consequent decrease in hardness
due to the overaging. Another analysis could also be performed concerning the stresses concentrator factor, as the
region where the welds end seems a crack, concentrating more stress than the hole used in the rivet process. Finally,
comparing the two types of FSSW welded joints the results were similar; however the WB joints presented longer
lives, contrary to the tensile tests, in which the WA joints presented the best results. This finding is probably due to
the mechanical anchoring caused by the higher penetration in the WB specimens.
1818 A.M.S. Malafaia et al. / Procedia Engineering 2 (2010) 1815–1821

4 A. M. S Malafaia et al. / Procedia Engineering 00 (2010) 000–000

6,00
RA
maximum load (kN)
5,00
4,00 WA

3,00 WB
2,00
1,00
0,00
1,0E+03 1,0E+04 1,0E+05 1,0E+06 1,0E+07
N (cycles until the failure)

Fig. 1. Failure modes observed in the fatigue tests.

No comparison between the properties of welded joints and riveted joints has been found in the literature.
However, some studies covering tensile and fatigue tests can be used to evaluate the performance of the welded
joints of the present study regarding the quality of the fatigue results. Lin et al [10, 11] also studied two sets of
FSSW welded joints fabricated with different tools, named concave and flat tools. The flat tool leads to 1.94kN in
the tensile shear tests. In fatigue tests, applying nearly 1kN (~50% of maximum tensile load) the specimen reached
104 cycles and using approximately 0.8kN (~40% of maximum tensile load) the tested specimen attained 105 cycles.
The samples fabricated with the concave tool supported a higher load in the tensile tests (2.59 kN); in the fatigue
tests with 1.2kN applied as the maximum load (~46% of maximum tensile load) a 104 cycle life was obtained, and
tests with 0.8kN (~31% of maximum tensile load) provided nearly 105 cycles. In the present study the maximum
loads supported in the tensile tests were 5.45 kN for WA and 4.95 for WB. In fatigue tests, the WA specimens
supported 2.2 kN (~40% of maximum tensile load) to 104 cycles and 1.64 (~30% of maximum tensile load) to 2x105
cycles, and the WB specimens supported 2.73 kN (~55% of maximum tensile load) to 104 cycles, and 1.64 kN
(~33% of maximum tensile load) to 4x105 cycles. Considering fatigue results and the tensile shear test results as
reference, this work obtained near the same results of Lin et al [10, 11]. The WA specimens reached the shortest
lives, i.e., lives similar to those observed by Lin et al [10,11] for specimens fabricated using a flat tool. On the other
hand, the WB specimens presented the best results in a range close to Lin et al [10,11] results while using the
concave tool. Tran et al [12] results were also in the same level of loads and lives for two materials, AA5754-O and
AA6111-T4, presenting around 50% of the maximum load in the tensile shear tests for 104 cycles, and 30% for 105
cycles. Thus, these comparisons show the low resistance in fatigue of the welds produced by the FSSW process
when compared to tensile shear results. For instance, the riveted joints supported 80% of the maximum tensile load
for a life of 104 cycles, and 70% for almost 105 cycles.
Regarding failure modes the welded and riveted joints presented two different fractures depending on the applied
load. Table 3 presents the failure mode operating in each specimen. Riveted joints failed due to rivet shearing or
fretting crack nucleation followed by crack propagation in the plates. Figure 2 shows the two different failure
modes: rivet shearing, caused by the higher loads (Figure 2a) and crack propagation, after nucleation by fretting,
caused by the intermediate and lower loads (Figure 2b). The analysis performed by liquid penetrant testing was on
the specimen presenting rivet shearing in order to verify the presence of cracks in the plates. The results can be also
seen Figure 2. For the specimen tested with a higher load (5.8 kN), no cracks were found (Figure 2c), showing a
pure shearing failure. On the other hand, the specimens tested under 5.15 kN maximum load showed small cracks
(Figure 2d), demonstrating crack initiation caused by fretting.
 A.M.S. Malafaia et al. / Procedia Engineering 2 (2010) 1815–1821 1819

A. M. S Malafaia et al. / Procedia Engineering 00 (2010) 000–000 5

Table 3. Failure modes observed in the fatigue tests.

Specimens Failure mode

RA 01, 02, 05 and 12 rivet shearing
crack nucleation by fretting and crack
RA 03, 04, 06, 07, 08, 09, 10, 11 and 13
propagation in the plates
WA 02, 04, 05, 07, 08, 10, 12, 13, 14 and 16;
Shearing fracture
WB 01, 02, 05, 06, 12, 14 and 15
WA 01 and 15; WB 03, 04 and 07 crack propagation in the plates
WB 08 and 10 No failure (run-out)

(a) (b)

(c) (d)

Fig. 2. Failure modes observed in the rivet joints during the fatigue tests: (a) rivet shearing; (b) crack propagation perpendicularly
to load direction after crack nucleation by fretting; (c) and (d) liquid penetrant test inspection.

The welded joints produced by FSSW also presented two different failure modes (Table 2). Figure 3 depicts both
modes; the shearing fracture was the principal mode, observed in almost all specimens, and in some cases the crack
propagation occurred perpendicularly to the load in the base material. Shearing fracture mode was originated due to
brittleness of the bonded region and can be seen in Figures 3a and b. As cited above, some studies have verified a
shearing fracture mode and associated it with either a small resistance in the tensile shear tests or a low cycle in the
fatigue tests. However, according to Chang et al [14] and Lin et al [10, 11] studies, the thickness of plates was
around 1mm and the aluminium alloys AA 5xxx and AA 6xxx series were studied. As in the present study the plates
were 1.6 mm thickness and the alloy was AA2024-T3, more resistant than 5xxx and 6xxx series, deformations in the
plates were restricted, thus concentrating the load on the weld spot, and consequently favoring shearing fracture
mechanisms.
1820 A.M.S. Malafaia et al. / Procedia Engineering 2 (2010) 1815–1821

6 A. M. S Malafaia et al. / Procedia Engineering 00 (2010) 000–000

The failure mode responsible for crack propagation mechanism perpendicularly to the load direction was
observed only for low applied loads. Some details are presented in Figures 3d and e. This failure usually displays a
transverse crack in the width directions and was also reported [10-12] and always associated with high-cycle loading
conditions.

(a) (b)

(c) (d)

Fig. 3. Failure modes observed in the FSSW welded joints during fatigue tests: (a) and (b) rivet shearing; (c) and (d) crack
perpendicular to the load direction.

4. Conclusions

The failure modes of riveted joints observed in the fatigue tests were rivet shearing and crack nucleation by
fretting followed by crack propagation in the plates perpendicularly to the load direction. Shearing mechanism was
related with high loads and also occurred in tensile specimens. Nucleation by fretting occurred during low to
moderate load tests. Among the specimens that failed by rivet shearing, the one tested with higher load (5.8 kN) did
not present cracks in an inspection by liquid penetrant test. On the other hand, the specimen tested with 5.15 kN
(maximum load) presented some cracks originated by fretting.
Comparing to riveted joints, the FSSW joints exhibited shorter lives in fatigue. The main reasons were the weld
heterogeneity (compared with the homogeneous rivet) and the stress concentration at the end of the welds between
the plates. However, the FSSW fatigue results were in accordance with the literature, with fatigue lives between
104 and 105 cycles for loads between 50 and 30% of the maximum tensile shear resistance load respectively.
Regarding failure modes of the welded joints tested under fatigue, two types were observed, i.e. shearing fracture
and crack propagation in the plates perpendicularly to the load direction. The first mode was more frequent and was
the same presented in the tensile shearing tests, occurring for higher loads. The second mode, with transverse cracks,
was observed only in some high-cycles tests, with low load applied.
 A.M.S. Malafaia et al. / Procedia Engineering 2 (2010) 1815–1821 1821

A. M. S Malafaia et al. / Procedia Engineering 00 (2010) 000–000 7

Acknowledgements

The authors would like to acknowledge FAPESP (no 07/02093-5) and CNPq for the financial support and
Prof. Dr.R. T. Coelho (OPF-EESC-USP).

References

[1] Hancock, R. Friction welding of Aluminum Cuts Energy Cost by 99%. Welding Journal 2004;83:40.
[2] Awang, M.; Mucino, V. H.; Feng, Z.; David S. A. Thermo-Mechanical Modeling of Friction Stir Spot Welding (FSSW) Process: Use of
an Explicit Adaptive Meshing Scheme, 2005 SAE International, 2005-01-1251
[3] Lathabai, S.; Painter, M.J.; Cantin, G.M.D.; Tyagi, V.K. Friction spot joining of an extruded Al–Mg–Si alloy. Scripta Materialia
2006;55(10):899-902.
[4] Gerlich, A.; Su, P.; North, T. H. Tool penetration during friction stir spot welding of Al and Mg alloys. Journal Of Materials Science
2005;40:6473–81.
[5] Smith, C. B., Hinrichs, J. F. And Ruehl, P. C. Friction Stir And Friction Stir Spot Welding - Lean, Mean, And Green. Sheet Metal Welding
Conference XI, Paper 2-5, Sterling Heights, MI, May 11-14, 2004.
[6] Gerlich, A.; Su, P.; North, T.H.; Bendzsak, G.J. Friction Stir Spot Welding Of Aluminum and Magnesium Alloys. Materials Forum
2005;29:290-4.
[7] Wang, D.-A.; Lee, S.-C. Microstructures and Failure Mechanisms of Friction Stir Spot Welds of Aluminum 6061-T6 Sheets. Journal of
Materials Processing Technology 2007;186(1-3): 291-7.
[8] Mitlin, D.; Radmilovic, V.; Pan, T.; Chen, J.; Feng, Z.; Santella, M.L. Structure–properties relations in spot friction welded (also known
as friction stir spot welded) 6111 aluminum. Materials Science and Engineering 2006;A441:79–96.
[9] Feng, Z.M.; Santella, L.; David, S.A.; Steel, R.J.; Packer, S.M.; Pan, T. el al. Friction Stir Spot Welding of Advanced High-Strength
Steels – A Feasibility Study. 2005 SAE International, 2005-01-1248.
[10] Lin, P.-C.; Pan, J.; Pan, T. Failure modes and fatigue life estimations of spot friction welds in lapshear specimens of aluminum 6111-T4
sheets. Part 1: Welds made by a concave tool. International Journal of Fatigue 2008;30(1):74-89.
[11] Lin, P.-C.; Pan, J.; Pan, T. Failure modes and fatigue life estimations of spot friction welds in lapshear specimens of aluminum 6111-T4
sheets. Part 2: Welds made by a flat tool. International Journal of Fatigue 2008;30(1):90-105.
[12] Tran, V.-X.; Pan, J.; Pan, T. Fatigue behavior of aluminum 5754-O and 6111-T4 spot friction welds in lap-shear specimens.
International Journal of Fatigue 2008;30(12):2175-90.
[13] Chang, W.S.; Cho, H.J.; Kim, H.J.; Chun, C.K. Evaluation of Friction Spot Joining Weldability of Al Alloys for Automotive. Materials
Science Forum 2007;539-543:411-6.
[14] Malafaia, A.M.S. Avaliação das propriedades mecânicas de tração e fadiga, com monitoramento de trincas, de juntas de Al AA2024-T3
soldadas a ponto por fricção-mistura. Master’s thesis, University of São Paulo, 2009. Available at:
http://www.teses.usp.br/teses/disponiveis/88/88131/tde-19032009-155633/publico/Disser_Artur_Malafaia.pdf