- Email: [email protected]
Mechanical properties and fatigue behavior of electromagnetic riveted lap joints influenced by shear loading
Journal of Manufacturing Processes 26 (2017) 226–239
Contents lists available at ScienceDirect
Journal of Manufacturing Processes journal homepage: www.elsevier.com/locate/manpro
Technical Paper
Mechanical properties and fatigue behavior of electromagnetic riveted lap joints influenced by shear loading Guangyao Li a,b , Hao Jiang a , Xu Zhang c , Junjia Cui a,b,∗ a b c
State Key Laboratory of Advanced Design and Manufacturing for Vehicle Body, Hunan University, Changsha, 410082, China Joint Center for Intelligent New Energy Vehicle, Shanghai, 200092, China Hunan University of Science and Technology, College of Mechanical and Electrical Engineering, Xiangtan, 411100, China
a r t i c l e
i n f o
Article history: Received 13 November 2016 Received in revised form 17 February 2017 Accepted 18 February 2017 Keywords: Electromagnetic riveting Fatigue behavior Mechanical properties Interference Microstructure
a b s t r a c t Electromagnetic riveting (EMR) has gained increasing attention as a relatively new mechanical joining technique in automobile industry. In this paper, the mechanical properties and fatigue behavior of electromagnetic riveted lap joints are discussed systematically. The rivet deformation, microstructure and hardness distribution of the formed rivets were investigated, which were also compared with regular pressure riveting (RPR). The results of shear strength showed that there was almost no difference between EMR and RPR, and the fatigue performance of EMR was about 1–3 times higher than that of RPR at any cyclic stress level. Quasi-static fracture analysis showed that shear fracture occurred in rivet shaft and the rupture appearance of two processes was similar. For fatigue failure, there were two fatigue failure modes for both processes: rivet shaft fracture under a higher cyclic stress and manufactured head fracture under a lower cyclic stress. Under the higher cyclic stress level, there was no big difference between two processes in the fatigue appearance. However, the fatigue cracks propagation zone of EMR sample fracture was significantly wider than that of RPR under a lower cyclic stress level, indicating a higher fatigue life of EMR samples. © 2017 Published by Elsevier Ltd on behalf of The Society of Manufacturing Engineers.
1. Introduction In automobile and aerospace manufacturing fields, the durability and security of machining parts are extremely important. Joining process as an inevitable assembling procedure is usually used in the sensitive area, and most strength failures occur in lap joints especially in transport aircraft [1]. Consequently, automotive body and aircraft fuselage consist of numerous lap joints which are attained by various joining techniques such as welding [2], bolting [3], and riveting [4]. However, welding defect caused by stress concentration and heat affected greatly reduce the joint reliability. By contrast, riveting process has some characteristics of reliable and stable joint quality, simple process, high efficiency and good seal [5]. Based on the advantages of riveting process, exhaustive studies have been conducted on the riveting lap joints. Hartman et al. [6] found that the rivet lap joints with a relatively higher rivet driven head diameter had better fatigue strength and Skorupa et al. [7]
∗ Corresponding author at: State Key Laboratory of Advanced Design and Manufacturing for Vehicle Body, Hunan University, Changsha, 410082, China. E-mail address: [email protected] (J. Cui).
investigated the effect of the driven head dimensions of the rivets on the strength of riveted joints comprehensively. The hole expansion (namely interference) was defined as the gap between expanded rivet shaft and the original hole. Müller [8] firstly measured the riveting interference and found the maximum value occurred near the driven head. Skorupa et al. [9] proved that a suitable interference had better mechanical properties and fatigue performance. Newman et al. [10] explored microstructure in the rivet cross section of the riveted sample with countersunk rivets, and found that the cracks initiated and grew from the faying surface. Literatures mentioned above mainly focused on the rivet process parameters studies such as rivet driven head dimensions, interference and cracks initiation. However, there are still some defects for the conventional riveting methods, such as easy slanting, cracking of the driven head, lower impact force and powerlessness for the bigger diameter rivets. Alternative riveting approach for structures is electromagnetic riveting (EMR), the main theory of which is based on the high-speed electromagnetic forming (EMF) (strain rate of 104 s−1 [11]) and regular riveting technology. As a new riveting method, EMR has been proved to have many advantages such as higher efficiency, the high-speed loading and the larger impact force. The advantage of EMR can further improve the joining quality of riveting lap-joint structures. For
http://dx.doi.org/10.1016/j.jmapro.2017.02.022 1526-6125/© 2017 Published by Elsevier Ltd on behalf of The Society of Manufacturing Engineers.
G. Li et al. / Journal of Manufacturing Processes 26 (2017) 226–239
example, Cao et al. [12] studied the interference in fiber composite riveting structures. The results showed the EMR technique had a better-distributed interference than hydraulic squeeze riveting and could partly prevent the destruction of the composite laminates. Experimental results obtained by Feng et al. [13] showed that both shear strength and pull-out strength of the EMR structures were significantly higher than that of the pneumatic riveting. Zhang et al. [14] investigated that the effect of driven head dimensions on microstructure and mechanical properties, and also found that 10 mm-2A10 electromagnetic riveted structures could not only improve both shear strength and pull-out strength, but also reduced the total weight compared with 6 mm-30CrMnSi bolted structures. Adiabatic shearing deformation is the significant characteristics of EMR, which could affect the microstructure properties in rivet driven heading and the strength of the rivet joints. Choo et al. [15] found that the high strain rate induced the precipitation hardening in adiabatic shear bands (ASBs) and led to the failure of aluminum rivets. Deng et al. [16] demonstrated that the deformation mechanism of EMR was the adiabatic shearing deformation and results from experiments and simulations showed that the maximum temperature was in ASBs of titanium alloy rivets. Zhang et al. [17] discovered that the driven head was divided into several zones with different mechanical properties by ASBs. Aforementioned work have mainly addressed the technical index (e.g. the interference) and the deformation mechanism in driven head, as well as its influence on the mechanical properties (e.g. shear and pull-out strength) of the EMR structures. The mechanical properties could only be used to evaluate the one-time or short-term bearing capacity of the structures, whereas the longterm performance evaluation in the industrial application usually adopts fatigue strength. However, few studies on the fatigue properties of EMR structures have been done so far. In addition, the relationship between the rivet deformation and microstructure after EMR has not been fully discussed. The aim of this paper is to investigate the rivet deformation mechanism, joining strength of carbon steel lap joints and the relationship between the two during the EMR. In addition, regular pressure riveting (RPR) with a quasi-static speed (2 mm/min) was employed to use as the comparing process. Firstly, the rivet deformation were measured after riveting, including rivet driven head dimensions and the hole expansion values of riveted sheets. And then the microstructure observation was conducted to under-
227
Fig. 1. The tensile stress-strain curve of Q235 carbon steel.
stand the microstructure evolution and hardness distribution of the deformed rivet. After that, experiments were performed to test the strength of the riveted lap joints, including shear and fatigue strength. Subsequently, the failure fractures appearance of shear and fatigue test were observed. Finally, the experimental results were analyzed and discussed. In general, this study is expected to provide further understanding of EMR and RPR lap joints for engineering application.
2. Materials and methods 2.1. Sample preparation Q235 hot rolling carbon construction steel rivets (similar to SS41 carbon steel for ISO standard) were used in this study. Oldersma [18] found that the galvanic corrosion between dissimilar materials would usually occur. Therefore, the same materials were selected as riveted sheets. Material properties of rivets and sheets are presented in Fig. 1, which are obtained by quasi-static tensile tests with a 2 mm/min velocity. The main chemical compositions are presented in Table 1.
Fig. 2. Geometry and dimensions of the riveted specimens.
228
G. Li et al. / Journal of Manufacturing Processes 26 (2017) 226–239
Table 1 Chemical compositions of the as-received rivets (wt.%).
2.2. Experimental methods
Material
C
Mn
Si
S
P
Fe
Q235
≤0.18%
0.35–0.80
≤0.30
≤0.040
≤0.040
Balance
Fig. 3. Schematic of formed rivet and driven head dimensions.
The riveted lap joint specimens for shear and fatigue tests consisted of two layers sheets with the prefabricated hole and a rivet. In order to avoid the bending of specimens under shear and fatigue loading, all the specimens were assembled a metallic gasket which had the same thickness with the sheets, as shown in Fig. 2. The thickness of riveted sheets was 4 mm and the diameter of rivet was 6 mm. The hole diameter of sheets was designed to 6.1 mm, and the length of the rivet shafts stretching out from the upper sheet was 8.0 mm. All the structure parameters were kept same for EMR and RPR experiments. According to the reference [19], D/D0 (D0 is the diameter of rivet, D is the diameter of the driven head) as a major structure parameter has a significant effect on the fatigue performance as shown in Fig. 3. To reduce the influence of the dimension of driven head on fatigue properties, the value of D/D0 was controlled around 1.5. Five repeated experiments were conducted to obtain a specified driven head dimension for EMR and RPR samples, respectively.
Fig. 4(a) shows the schematic of electromagnetic riveting, it can be seen that the electromagnetic riveting equipment mainly includes two parts: the setup of EMF (providing the energy for the riveting) and riveting mould. According to Weddeling et al. [20], the setup of EMF is mainly composed of a RLC-circuit and the storage energy of capacitors can be calculated by 1/2CU2 (C is the capacitance and U is the preset voltage). The preset voltage (U) is adjusted by a transformer, and the capacitors (C) are charged through direct current exported by rectifier circuit. When the charging of capacitors is completed, the storage energy would be discharged by closing the switch. Thus, high-amplitude alternating pulse currents can pass through the coil. The alternating currents would induce a stronger electromagnetic field in the surround, which would further lead to generating eddy currents in the metal (driver plate) within the magnetic field. Then the eddy currents would further induce electromagnetic field. Because the direction of the electromagnetic field induced by the coil is opposite to that by the driver plate, the powerful repulsive Lorenz force is generated between the two magnetic fields and pushed the punch to strike the rivets. The schematic of riveting process is shown in Fig. 4(b). It can be seen that the manufactured head is restricted in a die, and the driven head is formed under the punch impact to compress the two sheets. The driven head is deformed to drum-shaped due to contact friction between punch and rivets. EMR experiments were conducted with a HH54 handheld low voltage electromagnetic riveting guns (produced by Electroimpact ® company ) [21], as shown in Fig. 5. It could be seen that two electromagnetic riveting guns worked simultaneously and played a role in impacting and restriction respectively. The maximum discharge voltage of the equipment is 900 V. Higher voltage causes the faster impact velocity, so the diameter of the rivet driven head increased and the height of that decreased with the increasing of voltages. The optimal discharge parameters need to be explored by tests firstly. In this work, the discharge voltage of 800 V was proved suitable for deformations of the driven head. RPR experiments were conducted with 5985 tester with a 2 mm/min compression velocity. It can be seen that the punch and pedestal are used to rivet as shown in Fig. 6. In order to guarantee the consistency of driven head dimension, the compression range was set to 5.6 mm for EMR and RPR samples.
Fig. 4. The schematic of electromagnetic riveting: (a) electromagnetic setup and (b) the riveting process.
G. Li et al. / Journal of Manufacturing Processes 26 (2017) 226–239
229
Fig. 8. Comparing results of the driven head dimension (D/D0 ).
Fmin and Fmax are the minimum and maximum fatigue load in the sinusoidal wave cyclic loading, respectively. The maximum fatigue loads Fmax are determined by certain percentage of the maximum shear load (MSL). In this paper, Fmax were adopted by 70% MSL, 80% MSL, 90% MSL and 100% MSL, respectively. The average MSL of RPR and EMR was obtained by quasi-static test with a velocity of 2 mm/min using an Instron 8801 dynamic test machine. All fatigue tests were also implemented using the universal Instron 8801 test machine. The riveted specimens were firstly split along the axis of the rivet by Wire cut Electrical Discharge Machining (WEDM). After that, the specimens were mechanically polished and finally etched by a solution of 3ml-HNO3 and 100ml-C2 H5 OH. The microstructural observations were carried out with a Leica optical microscope, and hardness values were obtained with a HVS-1000 Digital Micro Vickers Hardness Tester. Rupture appearances after shear tests and fatigue tests were characterized with VEGA3 Scanning Electron Microscope of TESCAN after cleaning up by ultrasonic cleaning machine.
Fig. 5. HH54 Handheld low voltage electromagnetic riveting gun.
3. Results and discussion 3.1. Rivet deformation measurements
Fig. 6. The RPR experiments using Instron 5985 tester.
2.3. The strength tests and microstructure observations The fatigue tests were carried out using a sinusoidal wave shape at the frequency of 25 Hz/s, which is near inherent frequency of the car frame with a stress ratio (R) of 0.1, which is obtained by Eq. (1). R = Fmin /Fmax
In order to obtain a high quality riveting lap joint, the dimensions of the driven head (diameter and height) should be controlled accurately. In this study, five repeated experimental samples were measured to obtain the driven head dimensions (diameter and height) for EMR and RPR respectively. It can be seen that the shape of driven heads were similar between EMR and RPR as shown in Fig. 7. Fig. 8 shows the comparing results of driven head deformations (D/D0 ) between EMR and RPR. Table 2 shows the specific measured results of driven head dimensions. It could be seen that the
(1)
Fig. 7. Schematic diagram of the rivet driven head.
230
G. Li et al. / Journal of Manufacturing Processes 26 (2017) 226–239
Table 2 The rivet driven head dimensions (diameter and height) of EMR and RPR. Sample
1 2 3 4 5
EMR
RPR
Diameter of driven head (mm)
Height of driven head (mm)
Diameter of driven head (mm)
Height of driven head (mm)
9.21 8.99 9.02 9.11 9.20
2.84 3.21 3.22 2.95 2.82
8.99 9.02 9.14 9.01 9.22
3.17 3.18 2.99 3.27 2.77
Fig. 9. Comparison results of interference between EMR and RPR.
driven head dimension of all the samples had little difference and the values of D/D0 were all around 1.5. The comparing results also demonstrated that repeatability of the measured results for EMR and RPR samples were excellent. This could further contribute to compare the performance of EMR and RPR under the same driven head dimensions. During the riveting process, hole expansion of riveted sheets (interference fit) occurs due to the effect of mutual extrusion between rivet shaft and hole wall. Previous literature [7] had demonstrated that larger deformation of driven head has a tighter interference fit, and a suitable interference has also been proved to have better mechanical properties and fatigue performance. The interference (I) is defined as: I = [(di − d)/d] × 100%
(2)
where di and d are the expanded rivet diameter after riveting and original rivet hole diameter, respectively. In this study, all specimens were fabricated with a 6.1 mm original hole as shown in Fig. 2. To obtained the diameter of the deformed rivets, sheets were cut along the axis, and diameters of three typical locations were measured (position 1 was adjacent to the rivet driven head, position 2 was at the middle of the rivet shaft and position 3 was adjacent to the rivet manufactured head) as shown in Fig. 9. The results of the interference for EMR and RPR are shown in Table 3. It could be seen in Fig. 9 that three positions of interference for EMR had slighter fluctuation, indicating that the EMR had better interference fit behavior. From Table 3, it could be seen that the average value of interference I for EMR (1.59%) was higher than that of RPR (1.26%). The reason was that the stress wave propagation, reflection and stack existed in the EMR process (high-speed) as found by Meyers [22]. In this case, two EMR guns transmitted the stress waves in the both ends simultaneously, and then the stress waves encountered in the middle of the rivet shaft (posi-
Fig. 10. Microstructure observation in a section of the whole rivets and typical areas: (a) the EMR, (b) the RPR.
tion 2), which led to highest the interference fit in this position. Moreover, the interference for RPR decreased gradually from driven head to manufactured head, which was consistent with the results published by Müller [8].
G. Li et al. / Journal of Manufacturing Processes 26 (2017) 226–239
231
Table 3 The diameters of rivet shaft and interferences of different positions for EMR and RPR. Measured positions
1 2 3 Average
EMR
RPR
Diameter of rivet shaft (mm)
Interference (%)
Diameter of rivet shaft (mm)
Interference (%)
6.20 6.24 6.15 6.20
1.64 2.30 0.82 1.59%
6.22 6.19 6.12 6.18
1.97 1.48 0.33 1.26%
3.2. Microstructure and hardness distribution The fatigue and mechanical properties of metal materials significantly depend on microstructure. In order to further understand the performances of riveted structures obtained by EMR and RPR, microstructure observation of the rivets after riveting was carried out. Fig. 10(a) and (b) present the section metallographical structure of the deformed rivet after EMR and RPR, respectively. It could be seen that both of the rivets consisted of three parts (driven head, rivet shaft and manufacture head). Zone 1–3 were the typical areas in driven head, rivet shaft and manufactured head, respectively. These zones with high magnification were also exhibited below macrographs. It could be observed that two severe deformed areas existed in driven head (zone 1) and manufactured head (zone 3). For the rivet shaft (zone 2), it could be found that the original rivet microstructures were equiaxed grains with the size of 5 m. Grains were obviously elongated and distorted in the center of driven heads. In addition, although the deformations were severe in these areas, no microcracks were found in the driven head for two riveting processes. Moreover, the strengthening degree accorded with the deformation degree of grains, causing that the distributing law of hardness values was consistent to the microstructure deformations. Specifically, it could be demonstrated that the shear bands existed whether EMR or RPR in the driven head, and the difference between them was that the grain deformation of EMR was highly concentrated in the central of driven head, whereas the grain deformation of RPR was relatively more homogeneous. The reason for the phenomenon was that more severe material plastic flow occurred during EMR process (high-speed), while RPR process (quasi-static) could relieve deformation concentration to some extent. The microstructure of rivet shaft for EMR and RPR were roughly the same and the grains were almost not deformed due to the drum shape effect [14]. It illustrated that the two riveting methods had little effect on grain deformation of rivet shaft. At the manufactured head, it was found that the shape of manufactured head was almost not deformed. The grain deformation degree of two riveting methods also had little difference. It could be proved that the grain deformation in the manufactured head was caused during manufacturing process of rivets. Figs. 11 and 12 present the hardness distribution of manufactured head, driven head and rivet shaft, respectively. The hardness value of the undeformed rivets was around 180 HV. It showed that the hardness in the deformation zone was exceeded the original hardness because of the work hardening. Consequently, the hardness distribution law was consistent with the deformation degree. Fig. 11 shows that the hardness distribution of route 1 and route 2, it was indicated that the center of the manufactured head had the highest hardness which could reach about 248 HV and the hardness value in cross section (route 1) of the head were higher than other direction (route 2), indicating that the work hardening degree of route 1 was more serious than route 2. Fig. 12 compares the hardness of the driven head between EMR and RPR. The results showed that the distribution trend was similar. However, Fig. 12(a) and (b) depict that the hardness of the route 3 and route 4 (the upper half of driven head in position 3 and
Fig. 11. Vickers hardness distribution at manufactured head with the route 1 and 2.
4) after EMR were higher than that of RPR. The highest hardness of EMR and RPR in the route 3 and route 4 occurred in the center (position 3) and could reach about 257 HV and 245 HV, respectively. The high strain rate hardening effect of EMR could lead to a relatively higher hardness value, and the microscopic grain deformation of upper half and center was more concentrated and more serious. During the upsetting process of driven heads, the plastic flow was inhomogeneous due to friction effect [23]. The metal in the intermediate area was in the three-dimensional compressive stress state, causing more severe deformation occurred in these areas. The strengthening effect was remarkable in the intermediate areas. The large deformation in the intermediate areas also drove the plastic flow of the metals close to them. Consequently, the metals closer to the intermediate areas had relatively higher hardness values. Fig. 12(c) and (d) showed that the hardness difference in rivet shaft with the route 5 and 6 was also slight between EMR and RPR because the microscopic grain deformation of rivet shaft was not obvious, demonstrating that riveting methods had little effect on the hardness distribution of the rivet shaft. In conclusion, two riveting methods both had a reinforcement effect on the driven head, whereas the reinforcement effect of EMR was more obvious than RPR. 3.3. Shear and fatigue strength 3.3.1. Shear properties Repeated shear tests were conducted for EMR and RPR specimens respectively, and typical load-displacement curves and fracture feature are presented in Fig. 13. The repeated experimental results of MSL are shown in Table 4. The two curves can be divided into three stages: stage 1 (displacement from 0 to 0.25 mm) of elastic deformation, stage 2 (displacement from 0.25 to 3.0 mm) of plastic deformation, stage 3 (displacement from 3 to 3.5 mm) of failure. The variation trend was
232
G. Li et al. / Journal of Manufacturing Processes 26 (2017) 226–239
Fig. 12. Comparison results for the Vickers hardness of EMR and RPR at the rivet driven head and shaft: (a) route 3, (b) route 4, (c) route 5, (d) route 6.
deformation was relatively long, indicating that the samples had outstanding ability to bear the plastic deformation. It was also observed that the trend of two curves was the same and the MSL of EMR was about 10.3 kN which was slightly higher than that of RPR (10.1 kN), which demonstrated that the riveting methods had little influence on shear strength. Two curves had a sudden decline around the displacement of 3.0 mm, which revealed that the final failure occurred in this position. In addition, according to Eq. (3), the maximum shear strength was mainly depended on material properties and cross-sectional area. Fmax =
Fig. 13. The shear stress-strain curves and fracture samples of EMR and RPR.
Table 4 Maximum shear loads (MSL) for EMR tests and RPR tests.
Sample 1 Sample 2 Sample 3 Average
MSL for EMR tests (kN)
MSL for RPR tests (kN)
10.3 10.4 10.3 10.3
10.1 9.9 10.2 10.1
coincidence with the tensile curve of the rivet material shown in Fig. 2. Moreover, it could be found that the second stage of plastic
d2 4
(3)
where t represents shearing strength which is about 0.6 times of tensile strength, and d is the diameter of a rivet shaft. In this paper, the cross-section area at the interface of two sheets was the only difference which could be seen from Fig. 9. The diameter of rivet shaft in fracture position (position 2) using EMR was also slightly higher than that of RPR, and it proved that variation law of experimental results were in accordance with theoretical analysis. 3.3.2. Fatigue properties In order to compare the fatigue test results between EMR and RPR, the same maximum cycle stresses (average value of maximum shear strength with two different riveting methods) were taken into account. Fig. 14 shows the S-N curves of carbon steel riveted lap joints for EMR and RPR. The repeated experiment results of fatigue life are presented in Table 5. The data with the same cyclic stress and riveting method has little difference, indicating that the results can be used for comparison analysis. It could be observed from Fig. 14 that two riveting methods had the same trend of SN curves. The fatigue life of EMR was higher than that of RPR at
G. Li et al. / Journal of Manufacturing Processes 26 (2017) 226–239
Fig. 14. The S-N curves of carbon steel riveted lap joints for EMR and RPR.
any cyclic stress level. Specifically for high stress level, the fatigue life of EMR samples was about 3 times than that of RPR. The relation between the stress level and the gap between EMR and RPR in fatigue life was a negative correlation. There were some reasons
233
Fig. 15. Comparison results of displacement amplitude between EMR and RPR with the maximum cyclic stress.
that the riveting methods could affect the fatigue properties of carbon steel riveted lap joints. The previous studies [9,24] had proved
Fig. 16. SEM micrographs of the fracture appearance for shear test: (a) EMR, (b) RPR.
234
G. Li et al. / Journal of Manufacturing Processes 26 (2017) 226–239
Table 5 Repeated experiment results of fatigue life of carbon steel riveted lap joints for EMR and RPR. Maximum cyclic stress (MPa)
Fatigue life of EMR
Fatigue life of RPR
360.9
140455 102151 120654
57709 72820 50446
324.8
1010990 886521 806542
383608 596321 650596
288.7
1490768 1659241 1356910
1001686 1198930 1342318
253.0
2462250 2925625 2305464
2062037 1756208 1852365
that an appropriate interference fit could markedly improve the fatigue life. The reason was that a suitable interference fit could realize a close fit between rivet and hole. A higher interference fit would lead to stress concentration and even cracks around the contact area, and a lower interference fit would result in less contact area. In this study, the above results showed that the average interference fit of EMR was also larger than that of RPR. Therefore, a close fit between rivet shaft and sheet hole was formed during EMR process, which could effectively increase the contact area and decrease the average stress value per unit area. The above results also showed that the interference fit of EMR was more uniform, causing stress distribution also more uniform on the rivet shaft for EMR. In addition, the riveted joints would gradually form a gap under cyclic dynamic loading. However, a relative higher interference need not only more time to form the gap between rivet shaft and sheet hole, but could also reduce the impact and wear damage. Fig. 15 shows the displacement amplitude of EMR and RPR samples during fatigue cyclic loading process. It could be seen that the displacement amplitude of two curves increased with the increasing of fatigue life. The quick increase of the clearance could induce the impact damage and fatigue failure. The higher interference of EMR delayed the formation of clearance between rivet and hole wall during the cyclic loading process. Therefore, the displacement amplitude of EMR increased more slowly, indicating that EMR process could significantly improve the fatigue life of the riveted lap joints
Fig. 17. The schematic of the shear fracture process.
3.4. Fracture analysis 3.4.1. Shear fracture analysis Fig. 16 shows the fracture appearance of shear testing specimens for both EMR and RPR. It could be seen obviously that the shear fracture surface could be divided into two areas (zone 1 and zone 2). For zone 1, it was relatively smooth and flat, and almost no dimples occurred, which was a typical brittle shear zone characterized by shear slip. For zone 2, it was fairly rough and many tiny tensile plastic dimples appeared along the direction of shear force, which indicated that the original rivet material had excellent plasticity and presented the typical plastic ruptured mode. Therefore, the shear fractures for two rivet methods were both classified into plastic and brittle mixed ruptured mode. According to the fracture appearance and testing observation, the shear fracture process of the rivet lap joints could be illustrated as depicted in Fig. 17. The rivet firstly experienced the shear slip, resulting in forming the brittle shear (zone 1). The sudden rupture (zone 2) finally occurred when the rivet could not withstand the shear load. The area of zone 1 was nearly half of the rivet crosssection. The area for RPR was a bit larger than that of EMR, which
Fig. 18. The schematic of shear fatigue failure process.
was consistent with the contrastive result of the fracture displacement values (Fig. 13). It can be concluded that the appearance of shear fracture between EMR and RPR was similar, demonstrating that the rivet methods had no great effect on shear failure location and fracture appearance. In addition, it also caused little difference between EMR and RPR on the shear load. 3.4.2. Fatigue fracture analysis Fig. 18 presents the fatigue fracture process of the rivet lap joints. It could be seen that there were two failure modes for shear fatigue tests: rivet shaft fracture under a higher cyclic stress and manufactured head fracture under a lower cyclic stress. For the higher cyclic stress (≥320 MPa), the fracture process of tested specimens was similar to the shear fracture. The rivet shaft was unable to bear loop load and fractured after a short cyclic loading period [25]. For the lower cyclic stress (<320 MPa), the load was too small to destroy the rivet in a short period. A small clearance gradually formed between
G. Li et al. / Journal of Manufacturing Processes 26 (2017) 226–239
Fig. 19. SEM micrographs of fatigue fracture with global and typical zone views for failure mode 1: (a) EMR, (b) RPR.
235
236
G. Li et al. / Journal of Manufacturing Processes 26 (2017) 226–239
the rivet shaft and sheets, causing that the rivet became tilted. The rivet shaft still sustained the cyclic load in the vertical direction. However, the manufactured head and driven head of the rivet were also bearing the horizontal loads from sheets. The fatigue failure
eventually occurred at the interface between manufactured head and rivet shaft. From the above results and analysis, it could be found that the middle position of rivet shaft bear the maximum cyclic load under the higher cyclic stress. In addition, the manufactured head was
Fig. 20. (Continued)
G. Li et al. / Journal of Manufacturing Processes 26 (2017) 226–239
237
Fig. 20. SEM micrographs of fatigue fracture with global and typical zone views for failure mode 2: (a) EMR, (b) RPR.
Table 6 Comprehensive comparison between the EMR and RPR. Aspects
EMR
RPR
Comparing results
Joining quality
Shear load of 10.3 kN; Higher fatigue performance 1 millisecond At ambient temperature Electromagnetic forming system Discharging coil
Shear load of 10.1 kN; Lower fatigue performance 2 minutes Heating for large rivets Hydraulic system Hydraulic oil
Increased by 2% Increased by about 1–3 times Significantly reduced Green manufacturing Costs flat Convenient maintenance
Joining time per riveting Riveting condition Capital cost Maintenance
238
G. Li et al. / Journal of Manufacturing Processes 26 (2017) 226–239
hardly strengthened due to the constraint effect of restrict die during the riveting process. The manufactured head had the lower strength compared to rivet shaft and driven head, causing it to become a weak area. Meanwhile, it could also be found that the cyclic stress of 320 MPa is the critical point of two fatigue failure mode. The critical stress approximately equaled the yield strength (as shown in Fig. 2). Therefore, if the cyclic stress was equal or greater than 320 MPa, the rivet shaft would undergo plastic deformation in a shorter loading period. If the cyclic stress is under 320 MPa, the manufactured head with the lower strength would more quickly reached the bearing capacity than other parts. Fig. 19 presents the SEM fracture appearance of fatigue fracture mode 1. It could be seen that the fracture appearance was similar to that of the shear failure, and was also divided into two areas (zone 1 and zone 2). The surface of zone 1 was relatively flat, demonstrating that the zone was brittle rupture. It could be also seen that zone 1 was an obvious slip band. The slip direction was the same to load direction. The zone 2 had many tiny dimples and the rougher surface could be considered as the plastic fracture. All the dimples were extended along the same direction due to the cyclic shear stress. It could be found that the bottom of zone 2 for RPR was smoother than that of EMR, illustrating that the rivet still had a larger strength store for EMR than RPR. Consequently, the fatigue life of EMR lap joints was remarkably higher than that of RPR. Fig. 20 shows the SEM fracture appearance of fatigue fracture mode 2. It was observed that the fatigue fracture was a typical fatigue failure mode, and the rupture appearance was composed of three zones: fatigue crack initiation zone (zone 1), fatigue crack propagation zone (zone 2) and final fracture zone (zone 3). Zone 4 and 5 show the higher magnification of the fatigue crack propagation and final fracture zone respectively. In general, zone 3 for the two riveting processes had both tiny dimples and smooth fracture surface, indicating that this zone was the final fracture zone and the fracture mode was the mixed mode of the plastic and brittle rupture. The initiation fatigue cracks generally existed in opposite area of final fracture zone. The crack propagation direction was obviously toward to the fracture (zone 4). It could be also seen that fatigue trench lines were formed during crack propagation process, indicating that the propagation zone was induced by different crack initiations. So the striatum microstructures were not metal defects, demonstrating that the two riveting methods would not produce the initial defects. It could be seen from the fatigue crack propagation zone (zone 2) that the clear strip trend facilitated to deduce the extending direction of cracks. The dimples with various deep and size indicated that the fatigue fracture resulted from alternate overloading (zone 5). By comparing the fracture appearance of the two process joints, it could be found that the final fracture zone of RPR was larger than that of EMR. The fatigue propagation zone of RPR was smaller than that of EMR, which demonstrated that the tested RPR lap joint experienced a shorter cycle time in cracks propagation process [26]. Meanwhile, this is another reason for the increase in fatigue performance of EMR lap joints. Table 6 shows the comprehensive comparison between the EMR and RPR. The EMR structures have better fatigue performance and slightly higher shear properties than that of RPR. During the riveting process, the time-consuming of RPR is significantly longer than that of EMR, and the capital cost of the two process is almost equal. The damaged discharging coil can be replaced quickly and conveniently, and the production process is not affected. For large diameter rivets with higher strength, the EMR process is completed at ambient temperature relative to the RPR process. In summary, EMR process can meet the requirement of green manufacturing. This work investigated EMR and RPR through the comparing experimental method. However, the EMR process involves the electromagnetic-mechanical-thermal coupling effect, causing some deformation characteristics observed difficultly by the exper-
iments. In future works, numerical simulations can be performed to improve the deep understanding the difference between EMR and RPR. 4. Conclusions In order to study the microstructure and fatigue behavior of electromagnetic riveting and regular pressure riveting structures, process experiments and microstructure analysis were carried out. The conclusions obtained can be drawn as follows: (1) The hardness value of the most area in driven head for EMR was higher than that of RPR, indicating that the reinforcement of EMR in driven head was more effective than RPR. (2) The EMR structure exhibits much better fatigue strength than that of RPR due to larger and more uniform interference, and the maximum shear loads of EMR joint is slightly higher than that of RPR. (3) The fatigue failure mode was effected by rivet material yield stress and maximum cyclic stress. The fatigue fractures for the higher cyclic stress and the lower cyclic stress occurred on the rivet shaft and the manufactured head, respectively. The cyclic stress of 320 MPa is the critical point of two fatigue failure mode. Acknowledgements This project is supported by National Natural Science Foundation of China (No. 51405149) and the State Key Program of National Natural Science Foundation of China (No. 61232014). References [1] Skorupa A, Skorupa M. Riveted lap joints in aircraft fuselage: design, analysis and properties. Dordrecht, Heidelberg, New York, London: Springer; 2012. [2] Casalino G, Mortello M, Campanelli SL. Ytterbium fiber laser welding of Ti6Al4V alloy. J Manuf Process 2015;20:250–6. [3] Lee HC, Lee Y, Lee SY, Choi S, Lee DL, Im YT. Tool life prediction for the bolt forming process based on high-cycle fatigue and wear. J Mater Process Technol 2008;201:348–53. [4] Sun X, Khaleel MA. Performance optimization of self-piercing rivets through analytical rivet strength estimation. J Manuf Process 2005;7:83–93. [5] Chen NJ, Luo HY, Wan M, Chenot J. Experimental and numerical studies on failure modes of riveted joints under tensile load. J Mater Process Technol 2014;214:2049–58. [6] Hartman A. The influence of manufacturing procedures on the fatigue life of 2024-T3 Alclad riveted single lap joints. Report NLR TR 68072 U. Amsterdam: NLR; 1968. [7] Skorupa M, Skorupa A, Machniewicz T, Korbel A. Effect of production variables on the fatigue behaviour of riveted lap joints. Int J Fatigue 2010;32:996–1003. [8] Müller RPG. An experimental and analytical investigation on the fatigue behaviour of fuselage riveted lap joints. The significance of the rivet squeeze force, and a comparison of 2024-T3 and glare 3. Ph.D. Thesis. The Netherlands: Delft University of Technology; 1995. [9] Skorupa M, Machniewicz T, Skorupa A, Schijve J, Korbel A. Fatigue life prediction model for riveted lap joints. Eng Fail Anal 2015;53:111–23. [10] Newman Jr JC, Ramakrishnan R. Fatigue and crack-growth analyses of riveted lap-joints in a retired aircraft. Int J Fatigue 2016;82:342–9. [11] Psyk V, Risch D, Kinsey BL, Tekkaya AE, Kleiner M. Electromagnetic forming—a review. J Mater Process Technol 2011;211:787–829. [12] Cao ZQ, Cardew-Hall M. Interference-fit riveting technique in fiber composite laminates. Aerosp Sci Technol 2006;10:327–30. [13] Feng DG, Cao ZQ. Quality comparing analysis of electromagnetic riveting and pneumatic riveting. Forg Stamp Technol 2012;6:62–5. [14] Zhang X, Yu HP, Su H, Li CF. Experimental evaluation on mechanical properties of a riveted structure with electromagnetic riveting. Int J Adv Manuf Technol 2015;83:2071–82. [15] Choo V, Reinhal PG, Ghassaei S. Effect of high rate deformation induced precipitation hardening on the failure of aluminum rivets. J Mater Sci 1989;24:59–60. [16] Deng JH, Tang C, Fu MW, Zhan YR. Effect of discharge voltage on the deformation of Ti grade 1 rivet in electromagnetic riveting. Mater Sci Eng A 2014;591:26–32. [17] Zhang X, Yu HP, Li CF. Microstructure and mechanical properties of 2A10 aluminum alloy bar subjected to dynamic heading. J Mater Process Technol 2016;227:259–67. [18] Oldersma A. Fatigue of riveted joints. A literature survey and statistical analysis of existing test data. Report NLR CR 92401 L. Amsterdam: NLR; 1992. [19] Schijve J, Campoli G, Monaco A. Fatigue of structures and secondary bending in structural elements. Int J Fatigue 2009;31:1111–23.
G. Li et al. / Journal of Manufacturing Processes 26 (2017) 226–239 [20] Weddeling C, Walter V, Haupt P, Tekkaya AE, Schulze V, Weidenmann KA. Joining zone design for electromagnetically crimped connections. J Mater Process Technol 2015;225:240–61. [21] Brent WH. HH54 rugged and reliable handheld EMR. SAE aerofast conference 2009. [22] Meyers MA. Dynamic behavior of materials. Wiley-Interscience Press; 1994. [23] Zhang X, Yu HP, Li J, Li CF. Microstructure investigation and mechanical property analysis in electromagnetic riveting. Int J Adv Manuf Technol 2015;78:613–23.
239
[24] Li J, Zhang KF, Li Y, Liu P, Xia JJ. Influence of interference-fit size on bearing fatigue response of single-lap carbon fiber reinforced polymer/Ti alloy bolted joints. Tribol Int 2016;93:151–62. [25] Plaine AH, Suhuddin UFH, Alcântara NG, Santos JF. Fatigue behavior of friction spot welds in lap shear specimens of AA5754 and Ti6Al4V alloys. Int J Fatigue 2016;91:149–57. [26] Sharma C, Dwivedi DK, Kumar P. Fatigue behavior of friction stir weld joints of Al-Zn-Mg alloy AA7039 developed using base metal in different temper condition. Mater Des 2014;64:334–44.
Contents lists available at ScienceDirect
Journal of Manufacturing Processes journal homepage: www.elsevier.com/locate/manpro
Technical Paper
Mechanical properties and fatigue behavior of electromagnetic riveted lap joints influenced by shear loading Guangyao Li a,b , Hao Jiang a , Xu Zhang c , Junjia Cui a,b,∗ a b c
State Key Laboratory of Advanced Design and Manufacturing for Vehicle Body, Hunan University, Changsha, 410082, China Joint Center for Intelligent New Energy Vehicle, Shanghai, 200092, China Hunan University of Science and Technology, College of Mechanical and Electrical Engineering, Xiangtan, 411100, China
a r t i c l e
i n f o
Article history: Received 13 November 2016 Received in revised form 17 February 2017 Accepted 18 February 2017 Keywords: Electromagnetic riveting Fatigue behavior Mechanical properties Interference Microstructure
a b s t r a c t Electromagnetic riveting (EMR) has gained increasing attention as a relatively new mechanical joining technique in automobile industry. In this paper, the mechanical properties and fatigue behavior of electromagnetic riveted lap joints are discussed systematically. The rivet deformation, microstructure and hardness distribution of the formed rivets were investigated, which were also compared with regular pressure riveting (RPR). The results of shear strength showed that there was almost no difference between EMR and RPR, and the fatigue performance of EMR was about 1–3 times higher than that of RPR at any cyclic stress level. Quasi-static fracture analysis showed that shear fracture occurred in rivet shaft and the rupture appearance of two processes was similar. For fatigue failure, there were two fatigue failure modes for both processes: rivet shaft fracture under a higher cyclic stress and manufactured head fracture under a lower cyclic stress. Under the higher cyclic stress level, there was no big difference between two processes in the fatigue appearance. However, the fatigue cracks propagation zone of EMR sample fracture was significantly wider than that of RPR under a lower cyclic stress level, indicating a higher fatigue life of EMR samples. © 2017 Published by Elsevier Ltd on behalf of The Society of Manufacturing Engineers.
1. Introduction In automobile and aerospace manufacturing fields, the durability and security of machining parts are extremely important. Joining process as an inevitable assembling procedure is usually used in the sensitive area, and most strength failures occur in lap joints especially in transport aircraft [1]. Consequently, automotive body and aircraft fuselage consist of numerous lap joints which are attained by various joining techniques such as welding [2], bolting [3], and riveting [4]. However, welding defect caused by stress concentration and heat affected greatly reduce the joint reliability. By contrast, riveting process has some characteristics of reliable and stable joint quality, simple process, high efficiency and good seal [5]. Based on the advantages of riveting process, exhaustive studies have been conducted on the riveting lap joints. Hartman et al. [6] found that the rivet lap joints with a relatively higher rivet driven head diameter had better fatigue strength and Skorupa et al. [7]
∗ Corresponding author at: State Key Laboratory of Advanced Design and Manufacturing for Vehicle Body, Hunan University, Changsha, 410082, China. E-mail address: [email protected] (J. Cui).
investigated the effect of the driven head dimensions of the rivets on the strength of riveted joints comprehensively. The hole expansion (namely interference) was defined as the gap between expanded rivet shaft and the original hole. Müller [8] firstly measured the riveting interference and found the maximum value occurred near the driven head. Skorupa et al. [9] proved that a suitable interference had better mechanical properties and fatigue performance. Newman et al. [10] explored microstructure in the rivet cross section of the riveted sample with countersunk rivets, and found that the cracks initiated and grew from the faying surface. Literatures mentioned above mainly focused on the rivet process parameters studies such as rivet driven head dimensions, interference and cracks initiation. However, there are still some defects for the conventional riveting methods, such as easy slanting, cracking of the driven head, lower impact force and powerlessness for the bigger diameter rivets. Alternative riveting approach for structures is electromagnetic riveting (EMR), the main theory of which is based on the high-speed electromagnetic forming (EMF) (strain rate of 104 s−1 [11]) and regular riveting technology. As a new riveting method, EMR has been proved to have many advantages such as higher efficiency, the high-speed loading and the larger impact force. The advantage of EMR can further improve the joining quality of riveting lap-joint structures. For
http://dx.doi.org/10.1016/j.jmapro.2017.02.022 1526-6125/© 2017 Published by Elsevier Ltd on behalf of The Society of Manufacturing Engineers.
G. Li et al. / Journal of Manufacturing Processes 26 (2017) 226–239
example, Cao et al. [12] studied the interference in fiber composite riveting structures. The results showed the EMR technique had a better-distributed interference than hydraulic squeeze riveting and could partly prevent the destruction of the composite laminates. Experimental results obtained by Feng et al. [13] showed that both shear strength and pull-out strength of the EMR structures were significantly higher than that of the pneumatic riveting. Zhang et al. [14] investigated that the effect of driven head dimensions on microstructure and mechanical properties, and also found that 10 mm-2A10 electromagnetic riveted structures could not only improve both shear strength and pull-out strength, but also reduced the total weight compared with 6 mm-30CrMnSi bolted structures. Adiabatic shearing deformation is the significant characteristics of EMR, which could affect the microstructure properties in rivet driven heading and the strength of the rivet joints. Choo et al. [15] found that the high strain rate induced the precipitation hardening in adiabatic shear bands (ASBs) and led to the failure of aluminum rivets. Deng et al. [16] demonstrated that the deformation mechanism of EMR was the adiabatic shearing deformation and results from experiments and simulations showed that the maximum temperature was in ASBs of titanium alloy rivets. Zhang et al. [17] discovered that the driven head was divided into several zones with different mechanical properties by ASBs. Aforementioned work have mainly addressed the technical index (e.g. the interference) and the deformation mechanism in driven head, as well as its influence on the mechanical properties (e.g. shear and pull-out strength) of the EMR structures. The mechanical properties could only be used to evaluate the one-time or short-term bearing capacity of the structures, whereas the longterm performance evaluation in the industrial application usually adopts fatigue strength. However, few studies on the fatigue properties of EMR structures have been done so far. In addition, the relationship between the rivet deformation and microstructure after EMR has not been fully discussed. The aim of this paper is to investigate the rivet deformation mechanism, joining strength of carbon steel lap joints and the relationship between the two during the EMR. In addition, regular pressure riveting (RPR) with a quasi-static speed (2 mm/min) was employed to use as the comparing process. Firstly, the rivet deformation were measured after riveting, including rivet driven head dimensions and the hole expansion values of riveted sheets. And then the microstructure observation was conducted to under-
227
Fig. 1. The tensile stress-strain curve of Q235 carbon steel.
stand the microstructure evolution and hardness distribution of the deformed rivet. After that, experiments were performed to test the strength of the riveted lap joints, including shear and fatigue strength. Subsequently, the failure fractures appearance of shear and fatigue test were observed. Finally, the experimental results were analyzed and discussed. In general, this study is expected to provide further understanding of EMR and RPR lap joints for engineering application.
2. Materials and methods 2.1. Sample preparation Q235 hot rolling carbon construction steel rivets (similar to SS41 carbon steel for ISO standard) were used in this study. Oldersma [18] found that the galvanic corrosion between dissimilar materials would usually occur. Therefore, the same materials were selected as riveted sheets. Material properties of rivets and sheets are presented in Fig. 1, which are obtained by quasi-static tensile tests with a 2 mm/min velocity. The main chemical compositions are presented in Table 1.
Fig. 2. Geometry and dimensions of the riveted specimens.
228
G. Li et al. / Journal of Manufacturing Processes 26 (2017) 226–239
Table 1 Chemical compositions of the as-received rivets (wt.%).
2.2. Experimental methods
Material
C
Mn
Si
S
P
Fe
Q235
≤0.18%
0.35–0.80
≤0.30
≤0.040
≤0.040
Balance
Fig. 3. Schematic of formed rivet and driven head dimensions.
The riveted lap joint specimens for shear and fatigue tests consisted of two layers sheets with the prefabricated hole and a rivet. In order to avoid the bending of specimens under shear and fatigue loading, all the specimens were assembled a metallic gasket which had the same thickness with the sheets, as shown in Fig. 2. The thickness of riveted sheets was 4 mm and the diameter of rivet was 6 mm. The hole diameter of sheets was designed to 6.1 mm, and the length of the rivet shafts stretching out from the upper sheet was 8.0 mm. All the structure parameters were kept same for EMR and RPR experiments. According to the reference [19], D/D0 (D0 is the diameter of rivet, D is the diameter of the driven head) as a major structure parameter has a significant effect on the fatigue performance as shown in Fig. 3. To reduce the influence of the dimension of driven head on fatigue properties, the value of D/D0 was controlled around 1.5. Five repeated experiments were conducted to obtain a specified driven head dimension for EMR and RPR samples, respectively.
Fig. 4(a) shows the schematic of electromagnetic riveting, it can be seen that the electromagnetic riveting equipment mainly includes two parts: the setup of EMF (providing the energy for the riveting) and riveting mould. According to Weddeling et al. [20], the setup of EMF is mainly composed of a RLC-circuit and the storage energy of capacitors can be calculated by 1/2CU2 (C is the capacitance and U is the preset voltage). The preset voltage (U) is adjusted by a transformer, and the capacitors (C) are charged through direct current exported by rectifier circuit. When the charging of capacitors is completed, the storage energy would be discharged by closing the switch. Thus, high-amplitude alternating pulse currents can pass through the coil. The alternating currents would induce a stronger electromagnetic field in the surround, which would further lead to generating eddy currents in the metal (driver plate) within the magnetic field. Then the eddy currents would further induce electromagnetic field. Because the direction of the electromagnetic field induced by the coil is opposite to that by the driver plate, the powerful repulsive Lorenz force is generated between the two magnetic fields and pushed the punch to strike the rivets. The schematic of riveting process is shown in Fig. 4(b). It can be seen that the manufactured head is restricted in a die, and the driven head is formed under the punch impact to compress the two sheets. The driven head is deformed to drum-shaped due to contact friction between punch and rivets. EMR experiments were conducted with a HH54 handheld low voltage electromagnetic riveting guns (produced by Electroimpact ® company ) [21], as shown in Fig. 5. It could be seen that two electromagnetic riveting guns worked simultaneously and played a role in impacting and restriction respectively. The maximum discharge voltage of the equipment is 900 V. Higher voltage causes the faster impact velocity, so the diameter of the rivet driven head increased and the height of that decreased with the increasing of voltages. The optimal discharge parameters need to be explored by tests firstly. In this work, the discharge voltage of 800 V was proved suitable for deformations of the driven head. RPR experiments were conducted with 5985 tester with a 2 mm/min compression velocity. It can be seen that the punch and pedestal are used to rivet as shown in Fig. 6. In order to guarantee the consistency of driven head dimension, the compression range was set to 5.6 mm for EMR and RPR samples.
Fig. 4. The schematic of electromagnetic riveting: (a) electromagnetic setup and (b) the riveting process.
G. Li et al. / Journal of Manufacturing Processes 26 (2017) 226–239
229
Fig. 8. Comparing results of the driven head dimension (D/D0 ).
Fmin and Fmax are the minimum and maximum fatigue load in the sinusoidal wave cyclic loading, respectively. The maximum fatigue loads Fmax are determined by certain percentage of the maximum shear load (MSL). In this paper, Fmax were adopted by 70% MSL, 80% MSL, 90% MSL and 100% MSL, respectively. The average MSL of RPR and EMR was obtained by quasi-static test with a velocity of 2 mm/min using an Instron 8801 dynamic test machine. All fatigue tests were also implemented using the universal Instron 8801 test machine. The riveted specimens were firstly split along the axis of the rivet by Wire cut Electrical Discharge Machining (WEDM). After that, the specimens were mechanically polished and finally etched by a solution of 3ml-HNO3 and 100ml-C2 H5 OH. The microstructural observations were carried out with a Leica optical microscope, and hardness values were obtained with a HVS-1000 Digital Micro Vickers Hardness Tester. Rupture appearances after shear tests and fatigue tests were characterized with VEGA3 Scanning Electron Microscope of TESCAN after cleaning up by ultrasonic cleaning machine.
Fig. 5. HH54 Handheld low voltage electromagnetic riveting gun.
3. Results and discussion 3.1. Rivet deformation measurements
Fig. 6. The RPR experiments using Instron 5985 tester.
2.3. The strength tests and microstructure observations The fatigue tests were carried out using a sinusoidal wave shape at the frequency of 25 Hz/s, which is near inherent frequency of the car frame with a stress ratio (R) of 0.1, which is obtained by Eq. (1). R = Fmin /Fmax
In order to obtain a high quality riveting lap joint, the dimensions of the driven head (diameter and height) should be controlled accurately. In this study, five repeated experimental samples were measured to obtain the driven head dimensions (diameter and height) for EMR and RPR respectively. It can be seen that the shape of driven heads were similar between EMR and RPR as shown in Fig. 7. Fig. 8 shows the comparing results of driven head deformations (D/D0 ) between EMR and RPR. Table 2 shows the specific measured results of driven head dimensions. It could be seen that the
(1)
Fig. 7. Schematic diagram of the rivet driven head.
230
G. Li et al. / Journal of Manufacturing Processes 26 (2017) 226–239
Table 2 The rivet driven head dimensions (diameter and height) of EMR and RPR. Sample
1 2 3 4 5
EMR
RPR
Diameter of driven head (mm)
Height of driven head (mm)
Diameter of driven head (mm)
Height of driven head (mm)
9.21 8.99 9.02 9.11 9.20
2.84 3.21 3.22 2.95 2.82
8.99 9.02 9.14 9.01 9.22
3.17 3.18 2.99 3.27 2.77
Fig. 9. Comparison results of interference between EMR and RPR.
driven head dimension of all the samples had little difference and the values of D/D0 were all around 1.5. The comparing results also demonstrated that repeatability of the measured results for EMR and RPR samples were excellent. This could further contribute to compare the performance of EMR and RPR under the same driven head dimensions. During the riveting process, hole expansion of riveted sheets (interference fit) occurs due to the effect of mutual extrusion between rivet shaft and hole wall. Previous literature [7] had demonstrated that larger deformation of driven head has a tighter interference fit, and a suitable interference has also been proved to have better mechanical properties and fatigue performance. The interference (I) is defined as: I = [(di − d)/d] × 100%
(2)
where di and d are the expanded rivet diameter after riveting and original rivet hole diameter, respectively. In this study, all specimens were fabricated with a 6.1 mm original hole as shown in Fig. 2. To obtained the diameter of the deformed rivets, sheets were cut along the axis, and diameters of three typical locations were measured (position 1 was adjacent to the rivet driven head, position 2 was at the middle of the rivet shaft and position 3 was adjacent to the rivet manufactured head) as shown in Fig. 9. The results of the interference for EMR and RPR are shown in Table 3. It could be seen in Fig. 9 that three positions of interference for EMR had slighter fluctuation, indicating that the EMR had better interference fit behavior. From Table 3, it could be seen that the average value of interference I for EMR (1.59%) was higher than that of RPR (1.26%). The reason was that the stress wave propagation, reflection and stack existed in the EMR process (high-speed) as found by Meyers [22]. In this case, two EMR guns transmitted the stress waves in the both ends simultaneously, and then the stress waves encountered in the middle of the rivet shaft (posi-
Fig. 10. Microstructure observation in a section of the whole rivets and typical areas: (a) the EMR, (b) the RPR.
tion 2), which led to highest the interference fit in this position. Moreover, the interference for RPR decreased gradually from driven head to manufactured head, which was consistent with the results published by Müller [8].
G. Li et al. / Journal of Manufacturing Processes 26 (2017) 226–239
231
Table 3 The diameters of rivet shaft and interferences of different positions for EMR and RPR. Measured positions
1 2 3 Average
EMR
RPR
Diameter of rivet shaft (mm)
Interference (%)
Diameter of rivet shaft (mm)
Interference (%)
6.20 6.24 6.15 6.20
1.64 2.30 0.82 1.59%
6.22 6.19 6.12 6.18
1.97 1.48 0.33 1.26%
3.2. Microstructure and hardness distribution The fatigue and mechanical properties of metal materials significantly depend on microstructure. In order to further understand the performances of riveted structures obtained by EMR and RPR, microstructure observation of the rivets after riveting was carried out. Fig. 10(a) and (b) present the section metallographical structure of the deformed rivet after EMR and RPR, respectively. It could be seen that both of the rivets consisted of three parts (driven head, rivet shaft and manufacture head). Zone 1–3 were the typical areas in driven head, rivet shaft and manufactured head, respectively. These zones with high magnification were also exhibited below macrographs. It could be observed that two severe deformed areas existed in driven head (zone 1) and manufactured head (zone 3). For the rivet shaft (zone 2), it could be found that the original rivet microstructures were equiaxed grains with the size of 5 m. Grains were obviously elongated and distorted in the center of driven heads. In addition, although the deformations were severe in these areas, no microcracks were found in the driven head for two riveting processes. Moreover, the strengthening degree accorded with the deformation degree of grains, causing that the distributing law of hardness values was consistent to the microstructure deformations. Specifically, it could be demonstrated that the shear bands existed whether EMR or RPR in the driven head, and the difference between them was that the grain deformation of EMR was highly concentrated in the central of driven head, whereas the grain deformation of RPR was relatively more homogeneous. The reason for the phenomenon was that more severe material plastic flow occurred during EMR process (high-speed), while RPR process (quasi-static) could relieve deformation concentration to some extent. The microstructure of rivet shaft for EMR and RPR were roughly the same and the grains were almost not deformed due to the drum shape effect [14]. It illustrated that the two riveting methods had little effect on grain deformation of rivet shaft. At the manufactured head, it was found that the shape of manufactured head was almost not deformed. The grain deformation degree of two riveting methods also had little difference. It could be proved that the grain deformation in the manufactured head was caused during manufacturing process of rivets. Figs. 11 and 12 present the hardness distribution of manufactured head, driven head and rivet shaft, respectively. The hardness value of the undeformed rivets was around 180 HV. It showed that the hardness in the deformation zone was exceeded the original hardness because of the work hardening. Consequently, the hardness distribution law was consistent with the deformation degree. Fig. 11 shows that the hardness distribution of route 1 and route 2, it was indicated that the center of the manufactured head had the highest hardness which could reach about 248 HV and the hardness value in cross section (route 1) of the head were higher than other direction (route 2), indicating that the work hardening degree of route 1 was more serious than route 2. Fig. 12 compares the hardness of the driven head between EMR and RPR. The results showed that the distribution trend was similar. However, Fig. 12(a) and (b) depict that the hardness of the route 3 and route 4 (the upper half of driven head in position 3 and
Fig. 11. Vickers hardness distribution at manufactured head with the route 1 and 2.
4) after EMR were higher than that of RPR. The highest hardness of EMR and RPR in the route 3 and route 4 occurred in the center (position 3) and could reach about 257 HV and 245 HV, respectively. The high strain rate hardening effect of EMR could lead to a relatively higher hardness value, and the microscopic grain deformation of upper half and center was more concentrated and more serious. During the upsetting process of driven heads, the plastic flow was inhomogeneous due to friction effect [23]. The metal in the intermediate area was in the three-dimensional compressive stress state, causing more severe deformation occurred in these areas. The strengthening effect was remarkable in the intermediate areas. The large deformation in the intermediate areas also drove the plastic flow of the metals close to them. Consequently, the metals closer to the intermediate areas had relatively higher hardness values. Fig. 12(c) and (d) showed that the hardness difference in rivet shaft with the route 5 and 6 was also slight between EMR and RPR because the microscopic grain deformation of rivet shaft was not obvious, demonstrating that riveting methods had little effect on the hardness distribution of the rivet shaft. In conclusion, two riveting methods both had a reinforcement effect on the driven head, whereas the reinforcement effect of EMR was more obvious than RPR. 3.3. Shear and fatigue strength 3.3.1. Shear properties Repeated shear tests were conducted for EMR and RPR specimens respectively, and typical load-displacement curves and fracture feature are presented in Fig. 13. The repeated experimental results of MSL are shown in Table 4. The two curves can be divided into three stages: stage 1 (displacement from 0 to 0.25 mm) of elastic deformation, stage 2 (displacement from 0.25 to 3.0 mm) of plastic deformation, stage 3 (displacement from 3 to 3.5 mm) of failure. The variation trend was
232
G. Li et al. / Journal of Manufacturing Processes 26 (2017) 226–239
Fig. 12. Comparison results for the Vickers hardness of EMR and RPR at the rivet driven head and shaft: (a) route 3, (b) route 4, (c) route 5, (d) route 6.
deformation was relatively long, indicating that the samples had outstanding ability to bear the plastic deformation. It was also observed that the trend of two curves was the same and the MSL of EMR was about 10.3 kN which was slightly higher than that of RPR (10.1 kN), which demonstrated that the riveting methods had little influence on shear strength. Two curves had a sudden decline around the displacement of 3.0 mm, which revealed that the final failure occurred in this position. In addition, according to Eq. (3), the maximum shear strength was mainly depended on material properties and cross-sectional area. Fmax =
Fig. 13. The shear stress-strain curves and fracture samples of EMR and RPR.
Table 4 Maximum shear loads (MSL) for EMR tests and RPR tests.
Sample 1 Sample 2 Sample 3 Average
MSL for EMR tests (kN)
MSL for RPR tests (kN)
10.3 10.4 10.3 10.3
10.1 9.9 10.2 10.1
coincidence with the tensile curve of the rivet material shown in Fig. 2. Moreover, it could be found that the second stage of plastic
d2 4
(3)
where t represents shearing strength which is about 0.6 times of tensile strength, and d is the diameter of a rivet shaft. In this paper, the cross-section area at the interface of two sheets was the only difference which could be seen from Fig. 9. The diameter of rivet shaft in fracture position (position 2) using EMR was also slightly higher than that of RPR, and it proved that variation law of experimental results were in accordance with theoretical analysis. 3.3.2. Fatigue properties In order to compare the fatigue test results between EMR and RPR, the same maximum cycle stresses (average value of maximum shear strength with two different riveting methods) were taken into account. Fig. 14 shows the S-N curves of carbon steel riveted lap joints for EMR and RPR. The repeated experiment results of fatigue life are presented in Table 5. The data with the same cyclic stress and riveting method has little difference, indicating that the results can be used for comparison analysis. It could be observed from Fig. 14 that two riveting methods had the same trend of SN curves. The fatigue life of EMR was higher than that of RPR at
G. Li et al. / Journal of Manufacturing Processes 26 (2017) 226–239
Fig. 14. The S-N curves of carbon steel riveted lap joints for EMR and RPR.
any cyclic stress level. Specifically for high stress level, the fatigue life of EMR samples was about 3 times than that of RPR. The relation between the stress level and the gap between EMR and RPR in fatigue life was a negative correlation. There were some reasons
233
Fig. 15. Comparison results of displacement amplitude between EMR and RPR with the maximum cyclic stress.
that the riveting methods could affect the fatigue properties of carbon steel riveted lap joints. The previous studies [9,24] had proved
Fig. 16. SEM micrographs of the fracture appearance for shear test: (a) EMR, (b) RPR.
234
G. Li et al. / Journal of Manufacturing Processes 26 (2017) 226–239
Table 5 Repeated experiment results of fatigue life of carbon steel riveted lap joints for EMR and RPR. Maximum cyclic stress (MPa)
Fatigue life of EMR
Fatigue life of RPR
360.9
140455 102151 120654
57709 72820 50446
324.8
1010990 886521 806542
383608 596321 650596
288.7
1490768 1659241 1356910
1001686 1198930 1342318
253.0
2462250 2925625 2305464
2062037 1756208 1852365
that an appropriate interference fit could markedly improve the fatigue life. The reason was that a suitable interference fit could realize a close fit between rivet and hole. A higher interference fit would lead to stress concentration and even cracks around the contact area, and a lower interference fit would result in less contact area. In this study, the above results showed that the average interference fit of EMR was also larger than that of RPR. Therefore, a close fit between rivet shaft and sheet hole was formed during EMR process, which could effectively increase the contact area and decrease the average stress value per unit area. The above results also showed that the interference fit of EMR was more uniform, causing stress distribution also more uniform on the rivet shaft for EMR. In addition, the riveted joints would gradually form a gap under cyclic dynamic loading. However, a relative higher interference need not only more time to form the gap between rivet shaft and sheet hole, but could also reduce the impact and wear damage. Fig. 15 shows the displacement amplitude of EMR and RPR samples during fatigue cyclic loading process. It could be seen that the displacement amplitude of two curves increased with the increasing of fatigue life. The quick increase of the clearance could induce the impact damage and fatigue failure. The higher interference of EMR delayed the formation of clearance between rivet and hole wall during the cyclic loading process. Therefore, the displacement amplitude of EMR increased more slowly, indicating that EMR process could significantly improve the fatigue life of the riveted lap joints
Fig. 17. The schematic of the shear fracture process.
3.4. Fracture analysis 3.4.1. Shear fracture analysis Fig. 16 shows the fracture appearance of shear testing specimens for both EMR and RPR. It could be seen obviously that the shear fracture surface could be divided into two areas (zone 1 and zone 2). For zone 1, it was relatively smooth and flat, and almost no dimples occurred, which was a typical brittle shear zone characterized by shear slip. For zone 2, it was fairly rough and many tiny tensile plastic dimples appeared along the direction of shear force, which indicated that the original rivet material had excellent plasticity and presented the typical plastic ruptured mode. Therefore, the shear fractures for two rivet methods were both classified into plastic and brittle mixed ruptured mode. According to the fracture appearance and testing observation, the shear fracture process of the rivet lap joints could be illustrated as depicted in Fig. 17. The rivet firstly experienced the shear slip, resulting in forming the brittle shear (zone 1). The sudden rupture (zone 2) finally occurred when the rivet could not withstand the shear load. The area of zone 1 was nearly half of the rivet crosssection. The area for RPR was a bit larger than that of EMR, which
Fig. 18. The schematic of shear fatigue failure process.
was consistent with the contrastive result of the fracture displacement values (Fig. 13). It can be concluded that the appearance of shear fracture between EMR and RPR was similar, demonstrating that the rivet methods had no great effect on shear failure location and fracture appearance. In addition, it also caused little difference between EMR and RPR on the shear load. 3.4.2. Fatigue fracture analysis Fig. 18 presents the fatigue fracture process of the rivet lap joints. It could be seen that there were two failure modes for shear fatigue tests: rivet shaft fracture under a higher cyclic stress and manufactured head fracture under a lower cyclic stress. For the higher cyclic stress (≥320 MPa), the fracture process of tested specimens was similar to the shear fracture. The rivet shaft was unable to bear loop load and fractured after a short cyclic loading period [25]. For the lower cyclic stress (<320 MPa), the load was too small to destroy the rivet in a short period. A small clearance gradually formed between
G. Li et al. / Journal of Manufacturing Processes 26 (2017) 226–239
Fig. 19. SEM micrographs of fatigue fracture with global and typical zone views for failure mode 1: (a) EMR, (b) RPR.
235
236
G. Li et al. / Journal of Manufacturing Processes 26 (2017) 226–239
the rivet shaft and sheets, causing that the rivet became tilted. The rivet shaft still sustained the cyclic load in the vertical direction. However, the manufactured head and driven head of the rivet were also bearing the horizontal loads from sheets. The fatigue failure
eventually occurred at the interface between manufactured head and rivet shaft. From the above results and analysis, it could be found that the middle position of rivet shaft bear the maximum cyclic load under the higher cyclic stress. In addition, the manufactured head was
Fig. 20. (Continued)
G. Li et al. / Journal of Manufacturing Processes 26 (2017) 226–239
237
Fig. 20. SEM micrographs of fatigue fracture with global and typical zone views for failure mode 2: (a) EMR, (b) RPR.
Table 6 Comprehensive comparison between the EMR and RPR. Aspects
EMR
RPR
Comparing results
Joining quality
Shear load of 10.3 kN; Higher fatigue performance 1 millisecond At ambient temperature Electromagnetic forming system Discharging coil
Shear load of 10.1 kN; Lower fatigue performance 2 minutes Heating for large rivets Hydraulic system Hydraulic oil
Increased by 2% Increased by about 1–3 times Significantly reduced Green manufacturing Costs flat Convenient maintenance
Joining time per riveting Riveting condition Capital cost Maintenance
238
G. Li et al. / Journal of Manufacturing Processes 26 (2017) 226–239
hardly strengthened due to the constraint effect of restrict die during the riveting process. The manufactured head had the lower strength compared to rivet shaft and driven head, causing it to become a weak area. Meanwhile, it could also be found that the cyclic stress of 320 MPa is the critical point of two fatigue failure mode. The critical stress approximately equaled the yield strength (as shown in Fig. 2). Therefore, if the cyclic stress was equal or greater than 320 MPa, the rivet shaft would undergo plastic deformation in a shorter loading period. If the cyclic stress is under 320 MPa, the manufactured head with the lower strength would more quickly reached the bearing capacity than other parts. Fig. 19 presents the SEM fracture appearance of fatigue fracture mode 1. It could be seen that the fracture appearance was similar to that of the shear failure, and was also divided into two areas (zone 1 and zone 2). The surface of zone 1 was relatively flat, demonstrating that the zone was brittle rupture. It could be also seen that zone 1 was an obvious slip band. The slip direction was the same to load direction. The zone 2 had many tiny dimples and the rougher surface could be considered as the plastic fracture. All the dimples were extended along the same direction due to the cyclic shear stress. It could be found that the bottom of zone 2 for RPR was smoother than that of EMR, illustrating that the rivet still had a larger strength store for EMR than RPR. Consequently, the fatigue life of EMR lap joints was remarkably higher than that of RPR. Fig. 20 shows the SEM fracture appearance of fatigue fracture mode 2. It was observed that the fatigue fracture was a typical fatigue failure mode, and the rupture appearance was composed of three zones: fatigue crack initiation zone (zone 1), fatigue crack propagation zone (zone 2) and final fracture zone (zone 3). Zone 4 and 5 show the higher magnification of the fatigue crack propagation and final fracture zone respectively. In general, zone 3 for the two riveting processes had both tiny dimples and smooth fracture surface, indicating that this zone was the final fracture zone and the fracture mode was the mixed mode of the plastic and brittle rupture. The initiation fatigue cracks generally existed in opposite area of final fracture zone. The crack propagation direction was obviously toward to the fracture (zone 4). It could be also seen that fatigue trench lines were formed during crack propagation process, indicating that the propagation zone was induced by different crack initiations. So the striatum microstructures were not metal defects, demonstrating that the two riveting methods would not produce the initial defects. It could be seen from the fatigue crack propagation zone (zone 2) that the clear strip trend facilitated to deduce the extending direction of cracks. The dimples with various deep and size indicated that the fatigue fracture resulted from alternate overloading (zone 5). By comparing the fracture appearance of the two process joints, it could be found that the final fracture zone of RPR was larger than that of EMR. The fatigue propagation zone of RPR was smaller than that of EMR, which demonstrated that the tested RPR lap joint experienced a shorter cycle time in cracks propagation process [26]. Meanwhile, this is another reason for the increase in fatigue performance of EMR lap joints. Table 6 shows the comprehensive comparison between the EMR and RPR. The EMR structures have better fatigue performance and slightly higher shear properties than that of RPR. During the riveting process, the time-consuming of RPR is significantly longer than that of EMR, and the capital cost of the two process is almost equal. The damaged discharging coil can be replaced quickly and conveniently, and the production process is not affected. For large diameter rivets with higher strength, the EMR process is completed at ambient temperature relative to the RPR process. In summary, EMR process can meet the requirement of green manufacturing. This work investigated EMR and RPR through the comparing experimental method. However, the EMR process involves the electromagnetic-mechanical-thermal coupling effect, causing some deformation characteristics observed difficultly by the exper-
iments. In future works, numerical simulations can be performed to improve the deep understanding the difference between EMR and RPR. 4. Conclusions In order to study the microstructure and fatigue behavior of electromagnetic riveting and regular pressure riveting structures, process experiments and microstructure analysis were carried out. The conclusions obtained can be drawn as follows: (1) The hardness value of the most area in driven head for EMR was higher than that of RPR, indicating that the reinforcement of EMR in driven head was more effective than RPR. (2) The EMR structure exhibits much better fatigue strength than that of RPR due to larger and more uniform interference, and the maximum shear loads of EMR joint is slightly higher than that of RPR. (3) The fatigue failure mode was effected by rivet material yield stress and maximum cyclic stress. The fatigue fractures for the higher cyclic stress and the lower cyclic stress occurred on the rivet shaft and the manufactured head, respectively. The cyclic stress of 320 MPa is the critical point of two fatigue failure mode. Acknowledgements This project is supported by National Natural Science Foundation of China (No. 51405149) and the State Key Program of National Natural Science Foundation of China (No. 61232014). References [1] Skorupa A, Skorupa M. Riveted lap joints in aircraft fuselage: design, analysis and properties. Dordrecht, Heidelberg, New York, London: Springer; 2012. [2] Casalino G, Mortello M, Campanelli SL. Ytterbium fiber laser welding of Ti6Al4V alloy. J Manuf Process 2015;20:250–6. [3] Lee HC, Lee Y, Lee SY, Choi S, Lee DL, Im YT. Tool life prediction for the bolt forming process based on high-cycle fatigue and wear. J Mater Process Technol 2008;201:348–53. [4] Sun X, Khaleel MA. Performance optimization of self-piercing rivets through analytical rivet strength estimation. J Manuf Process 2005;7:83–93. [5] Chen NJ, Luo HY, Wan M, Chenot J. Experimental and numerical studies on failure modes of riveted joints under tensile load. J Mater Process Technol 2014;214:2049–58. [6] Hartman A. The influence of manufacturing procedures on the fatigue life of 2024-T3 Alclad riveted single lap joints. Report NLR TR 68072 U. Amsterdam: NLR; 1968. [7] Skorupa M, Skorupa A, Machniewicz T, Korbel A. Effect of production variables on the fatigue behaviour of riveted lap joints. Int J Fatigue 2010;32:996–1003. [8] Müller RPG. An experimental and analytical investigation on the fatigue behaviour of fuselage riveted lap joints. The significance of the rivet squeeze force, and a comparison of 2024-T3 and glare 3. Ph.D. Thesis. The Netherlands: Delft University of Technology; 1995. [9] Skorupa M, Machniewicz T, Skorupa A, Schijve J, Korbel A. Fatigue life prediction model for riveted lap joints. Eng Fail Anal 2015;53:111–23. [10] Newman Jr JC, Ramakrishnan R. Fatigue and crack-growth analyses of riveted lap-joints in a retired aircraft. Int J Fatigue 2016;82:342–9. [11] Psyk V, Risch D, Kinsey BL, Tekkaya AE, Kleiner M. Electromagnetic forming—a review. J Mater Process Technol 2011;211:787–829. [12] Cao ZQ, Cardew-Hall M. Interference-fit riveting technique in fiber composite laminates. Aerosp Sci Technol 2006;10:327–30. [13] Feng DG, Cao ZQ. Quality comparing analysis of electromagnetic riveting and pneumatic riveting. Forg Stamp Technol 2012;6:62–5. [14] Zhang X, Yu HP, Su H, Li CF. Experimental evaluation on mechanical properties of a riveted structure with electromagnetic riveting. Int J Adv Manuf Technol 2015;83:2071–82. [15] Choo V, Reinhal PG, Ghassaei S. Effect of high rate deformation induced precipitation hardening on the failure of aluminum rivets. J Mater Sci 1989;24:59–60. [16] Deng JH, Tang C, Fu MW, Zhan YR. Effect of discharge voltage on the deformation of Ti grade 1 rivet in electromagnetic riveting. Mater Sci Eng A 2014;591:26–32. [17] Zhang X, Yu HP, Li CF. Microstructure and mechanical properties of 2A10 aluminum alloy bar subjected to dynamic heading. J Mater Process Technol 2016;227:259–67. [18] Oldersma A. Fatigue of riveted joints. A literature survey and statistical analysis of existing test data. Report NLR CR 92401 L. Amsterdam: NLR; 1992. [19] Schijve J, Campoli G, Monaco A. Fatigue of structures and secondary bending in structural elements. Int J Fatigue 2009;31:1111–23.
G. Li et al. / Journal of Manufacturing Processes 26 (2017) 226–239 [20] Weddeling C, Walter V, Haupt P, Tekkaya AE, Schulze V, Weidenmann KA. Joining zone design for electromagnetically crimped connections. J Mater Process Technol 2015;225:240–61. [21] Brent WH. HH54 rugged and reliable handheld EMR. SAE aerofast conference 2009. [22] Meyers MA. Dynamic behavior of materials. Wiley-Interscience Press; 1994. [23] Zhang X, Yu HP, Li J, Li CF. Microstructure investigation and mechanical property analysis in electromagnetic riveting. Int J Adv Manuf Technol 2015;78:613–23.
239
[24] Li J, Zhang KF, Li Y, Liu P, Xia JJ. Influence of interference-fit size on bearing fatigue response of single-lap carbon fiber reinforced polymer/Ti alloy bolted joints. Tribol Int 2016;93:151–62. [25] Plaine AH, Suhuddin UFH, Alcântara NG, Santos JF. Fatigue behavior of friction spot welds in lap shear specimens of AA5754 and Ti6Al4V alloys. Int J Fatigue 2016;91:149–57. [26] Sharma C, Dwivedi DK, Kumar P. Fatigue behavior of friction stir weld joints of Al-Zn-Mg alloy AA7039 developed using base metal in different temper condition. Mater Des 2014;64:334–44.