Strength Modeling
Strength Modeling
Strength Modeling
Research Article
Keywords: Self-piercing riveting (SPR), aluminum alloy sheet, rivet aring, riveting strength model,
industrial internet of things (IIoT)
DOI: https://doi.org/10.21203/rs.3.rs-2052846/v1
License: This work is licensed under a Creative Commons Attribution 4.0 International License.
Read Full License
Strength modeling of Al-alloy sheet self-piercing riveting considering different failure
modes
a. School of Materials Science and Engineering, Shanghai Jiao Tong University, No. 800,
Forming, Shanghai Jiao Tong University, 800 Dongchuan Road, Minhang District,
c. State Key Laboratory of Metal Matrix Composites, Shanghai Jiao Tong University, 800
Abstract
Self-piercing riveting (SPR) has been widely utilized to connect metal components in
industry, and the mechanical properties of final product depend on the strength of SPR joint
which is experimentally measured through cross-tension and lap-shear tests. These tests are
destructive and the tested strength of specimen is not directly related to the actual strength of
the SPR product. In this study, the SPR process of aluminum alloy sheet was investigated and
the general empirical model of SPR strength was established by comprehensively considering
the factors including of the geometric dimensions of components and die, the material
properties of rivet and sheet, and the load - stroke curve of punch. The calculated strength values
of 4 group of SPR specimens were verified by the experimentally measured results. All
1 / 54
calculation errors are lower than 8%. An industrial internet of things (IIoT) was developed to
automatically realize the data transmission and strength calculation of the SPR process.
Keywords
Self-piercing riveting (SPR); aluminum alloy sheet; rivet flaring; riveting strength model;
1. Introduction
Self-piercing riveting (SPR) is one of the high-efficiency processes to connect sheet materials,
in which a rivet is pushed into the stacked sheets in a certain direction under the force of a
punch [1-3]. At the same time, the deformed metal is formed into a button shape with a mold
assembled on the other side of the sheet. In practice, the number of sheets prepared for SPR can
be more than double layers. As an instance, three layers of Al alloy sheets connected via SPR
process were studied by Han, et al. [4]. Moreover, SPR process is extremely suitable to connect
sheets that are made of similar or dissimilar materials [5]. Typically, SPR is successfully utilized
to connect low-strength metals such aluminum alloy and copper alloy. For example, Calabrese,
et al. [6] studied the SPR of AA6111 sheet and obtained the failure map for net-tension and
pull-out mechanisms. He, et al. [7] compared the joint performances of the SPR of copper alloy
H62 sheets and the SPR of Al-to-Cu sheets. Besides, high-strength materials can also be
connected via SPR process. Xie, et al. [8] investigated the cross-tension strength of the SPRed
cold formed steel sheets. Zhao, et al. [9] reported the SPR process of titanium alloy TA1 sheet.
Additionally, SPR process is applied to connect composite sheet. The SPR of the carbon fiber
reinforced polymer composite sheet was studied by Rao, et al. [10]. And the glass fiber
reinforced thermoplastic composite sheet was studied by Gay, et al. [11]. The sheet subjected
2 / 54
to SPR may also be covered with surface coating, such as the corrosion resistance coating and
electrical insulation coating reported by Han and Chrysanthou [12]. The candidates for rivet
material can be different according to the sheet material. the steel HSLA350 rivet was utilized
to connect aluminum alloy AA6111 sheets by Han, Chrysanthou and Young [4]. While
aluminum alloy rivets made of 6082-T6, 7108-T5, and 7278-T6 were utilized by Hoang, et al.
[13] to aluminum alloy 6060 sheets in three different tempers (temper W, temper T4, and temper
T6). The forming of SPR button depends on the die cavity which also has different designs.
Karathanasopoulos, et al. [14] compared the influences of different die tip design on the
The broad application of SPR is due to the its following advantages: the wide range of
material applicability even for the materials with poor weldability, high forming efficiency,
material saving (no need for pre-drilling, no remnant), energy saving (no need for preheating),
environmental friendly (no pollution emission), and lightweight manufacturing. SPR also offers
more advantages over other bonding processes, such as resistance spot welding (RSW). Sun
and Khaleel [15] found that SPR joints of metal sheets had better dynamic impact strength
fatigue strength than the joint of resistance spot welding. Also, Sun, et al. [16]found that the
fatigue strength of SPR joints was better than the joint of resistance spot welding. Currently,
some researches combine SPR process with other bonding methods to improve the interlock
strength of the joints. Ma, et al. [17] employed the friction self-piercing riveting (F-SPR) to
bond low ductility materials. Ma, et al. [18] also studied the bonding strength of F-SPRed
AA7075-T6 aluminum alloy joints. Yang, et al. [19] investigated the hybrid process of F-SPR
and adhesive bonding of Al7075-T6 alloy sheets where the adhesive was utilized for lubrication.
3 / 54
Yang, et al. [20] studied the Al-to-Mg sheet F-SPR and optimized the design of bottom die. Ma,
et al. [21] found that the rivet orientation significantly affect the quality of Al-to-Mg F-SPR
joint. Ying, et al. [22] studied the thermal self-piercing riveting (T-SPR) of AA7075-to-T6
sheets and the failure modes of the joints. Han, et al. [23] developed an innovative SPR to
connect the sheets made of glass fiber reinforced polymer (GFRP). Jiang, et al. [24] studied the
Electromagnetic self-piercing riveting (E-SPR) of CFRP-to-Al and steel-to-Al sheets and found
the structure of rivet had significant influences on the quality. All these researches have
extended the applications of SPR and improved the connecting quality of joints.
The major application fields of SPR include automotive industry (e.g. structural components),
electrical industry (e.g. busbar support), aircraft industry and so on. Due to the demand for
lightweighting of automobiles, new techniques have emerged in the field of SPR. Danyo [25]
highlighted that SPR is one of the key jointing methods in vehicle. The current tendencies of
the SPR applications in vehicle body include the connection of dissimilar materials of metal
and composite, and the jointing of high-strength metal and low-ductility metal. Wang, et al. [26]
remarked that the SPR of metal and composites sheets has become a major part of the new
jointing techniques. Karim, et al. [27] also pointed out that SPR is more suitable for bonding
dissimilar materials than other conventional jointing processes. However, there are also
challenges for the industrial applications of SPR recently including of the quality prediction on
the automatic production line, the high-efficiency optimization of the tool design for certain
SPR process, and the development of new SPR process. Many literatures focus on the quality
prediction of SPR joints, and parts of the issues include the metal deformation during SPR
process, the mechanical properties of SPR joint, the modeling of joint strength and so on.
4 / 54
Kyoung-Yun Kim, et al. [28] discussed the challenges facing SPR in the quality prediction, and
compared the main methods of quality prediction, including of experimental test, finite element
method (FEM), numerical equation model, and data-driven model. Notably, to establish a
robust quality prediction model for SPR is still a hard bone for engineers and researchers
because many unknown variables in the changing manufacturing scenes result in the individual
differences of SPR joints. Those unknown variables could be one or more of the following
aspects: the inhomogeneous properties, the internal defects as the nature of material, the uneven
deformation of rivet during operation, the pressure fluctuations of forming tools, the wear of
bottom die and so on. Actually, a single quality prediction method is insufficient to cover all
the concerns in practice. Nevertheless, the it is possible to establish a sound model to predict
SPR quality by controlling for variables, and the main steps could be summarized as follows:
(a) analyze the SPR processing, (b) selecting the key variables affecting joint quality, (c)
selecting the most suitable quality prediction model and establishing the model, (d) verifying
SPR process of metal sheets is a typical cold forming procedure, which is divided into three
deformation stages by Porcaro, et al. [29]: (a) under the pressure of punch, the rivet pierces
through the head-side sheet materials without significant plastic deformation; (b) the rivet
flaring increases rapidly and the bottom sheet metal starts to fill the die cavity. Generally, the
stacked sheets and bottom die are fixed during the processing. SPR process has been illustrated
in other studies. The aluminum-to-steel SPR with a steel rivet was divided into three stages
including of clamping, riveting and mold opening by Lou, et al. [30]. The A6060-to-A6060
sheet SPR with a high-strength steel rivet was divided into four stages by Porcaro, Hanssen,
5 / 54
Langseth and Aalberg [29] in terms of clamping, piercing, flaring and release of punch. The
SPR process was also divided into four stages named clamping, piercing, flaring and release of
punch by Su, et al. [31], as well as Zhang, et al. [32]. The SPR of steel sheet was divided into
six stages by Yan, et al. [33] including of die clamping, punch clamping, punch expert pressure,
rivet piercing, deformation, and forming. All the demonstrations on the procedure of SPR in
those literatures paid attentions to the tool motion and the material deformation, especially the
rivet piercing through the stacked sheets and flaring in the metal sheet under the compression
of punch. The rivet joint is a final representation of the component deformation, most of which
is recorded in the load-displacement curve of punch and the final shape of the rivet joint. It’s
necessary to discuss the factors affecting the component deformation, joint strength and joint
defects, which are expected to help improve the quality of SPR product.
The performances of SPR joint include strength index (e.g. cross-tension strength and lap-
shear strength), fatigue life, fretting wear, heat resistance and corrosive resistance. The cross-
tension strength and lap-shear strength of SPR joint are the maximum force value measured in
the cross-tension test and lap-shear test, respectively, which are widely utilized in industrial
practice. The fatigue behaviors of SPR joint are usually investigated by using the specimens
with lap-shear type, cross-tension type, U-shaped type, and coach-peel type. The cross-tension
specimen of cold form steel sheet was utilized by Xie, et al. [34] to study the tensile strength of
SPR. The finite element models of cross-tension, lap-shear, and coach-peel specimens were
built by Kang and Kim [35] to study the fatigue strength of SPRed Al5052 sheet. The specimens
with lap-shear type and U-shaped type were utilized by Wu, et al. [36] in order to investigate
the fatigue behaviors of SPRed AA6111-T4. The lap-shear and coach-peel specimens were
6 / 54
utilized by Presse, et al. [37] to estimate the fatigue life of SPRed Al-to-steel joint. Those
specimens can also be used to study the wear behaviors of SPR joint. The wear of lap-shear
specimen at the interface between sheet and rivet was studied by Chen, et al. [38]. Kotadia, et
al. [39] found that the corrosion condition has significant influences on the failure of the joint
of the coated Al-to-steel sheet SPR. All the above tests are destructive experiments which
cannot be directly applied to the actual SPR product. Hence, a theoretical calculation model is
useful to predict the quality of SPR joints in industrial practice. However, the robustness of the
strength model is first issue that should be considered when calculating the strength of SPR
joints.
In order to establish a theoretical model of the strength of metal SPR joint, it is necessary to
analyze which parameters have a significant effect on the strength of SPR joint. Parameters
related to the SPR strength can be divided into the following three categories: (a) the material
properties of sheet and rivet (e.g. density, Young’s module, yield strength and hardness), (b) the
geometric dimensions of components (e.g. the thickness of stacked sheet, the diameter and
length of rivet, the rivet flaring, the diameter and depth die cavity) [40], (c) the processing
parameters (e.g. the load - stroke curve of punch, riveting direction). Zhao, et al. [41] reported
that increasing the thickness of sheet in a certain range can increase the fatigue life of SPR joint
and decrease the fretting wear at the interface between stacked sheets. Haque and Durandet [42]
studied the steel-to-steel SPR and obtained the conclusion that increasing the diameter of rivet
can improve the strength of SPR joint under impact load testing. Moreover, Sun and Khaleel
[43] found that the flaring shape of rivet dominates the interlock of stacked sheets, so SPR
joints with larger rivet flaring have higher static strength. The load-stroke curve of punch can
7 / 54
record the details of materials deformation during SPR process, which was utilized to quantify
the rivet flaring by Sun and Khaleel [44]. The rivet flaring is one of the most important factors
due to the interlock strength increases with the value of rivet flaring increasing. Besides, the
coating of sheet is also an important factors affecting the fractional statement and the joint
The methods of SPR strength modeling generally include numerical simulation (e,g. FEM)
[46-48], theoretical analysis method, and data-driven modeling. Firstly, numerical simulation
is believed an effective way to analyze the time-evolution history of displacement, stress, and
strain field in the processing of cold metal deformation like SPR. From simulation result, it is
easily to obtain the deformation profile of rivet and stacked sheet during SPR process and to
directly measure the value of rivet flaring. The simulation results can also help to draw the load-
stroke curve of punch. Many studies use numerical simulation to build two-dimension (2D) or
three-dimension (3D) model of the SPR process. Using software LS-DYNA, Hoang, Porcaro,
Langseth and Hanssen [13] successfully applied a 2D axisymmetric model to the SPR of
aluminum alloy sheet with aluminum rivets. A 3D model of the Mg-to-Al SPR process was
established through ABAQUS by Moraes, et al. [49], and it was found that the strain hardening
caused by SPR process is one of the main contributions for the joint strength. Du, et al. [50]
studied the Al-to-steel sheet SPR process by using a 2D axisymmetric model established
through LS-DYNA based on radaptivity method. Modeling the SPR process of Al sheet with a
steel rivet was studied by Casalino, et al. [51], and the simulation results were validated by
the thermo-mechanical behaviors caused by friction during Al sheet SPR process. Other
8 / 54
researchers studied the structure strength of SPR joint by numerical simulation in order to
predict the mechanical behaviors and quality of SPR joints. Lukas Potgorschek [53] improved
curve of punch. Hönsch, et al. [54] proposed a 2D FE model and a 3D FE model to simulate
the SPR processing and the tension test of joint, respectively, and the simulation results of the
joint behaviors were consistent with experiment results. However, numerical simulation
method is not only time-consume in preparing the stringent input and complex settings of the
model, but also quite inaccurate because it neglects the material differences and dimensional
deviation of component to simplify the calculation model. Weighting the advantages and
Theoretical analysis is a candidate method for evaluating the strength of SPR joints by using
explicit empirical equations instead of complicated calculation based on finite element model.
For example, Sun and Khaleel [43] established two estimators to calculate the cross-tension
strength of SPR joints cracked in failure mode I (rivet tail pullout) and failure mode II (rivet
head pullout) respectively, and the failure mode of SPR joints can be predicted by comparing
the calculated values of cross-tension strength, where only 3 empirical coefficients and 3
material parameters were involved in each estimator. However, two important material
parameters in the estimator of failure mode I, diameter of clinched portion of rivet tail (Dc) and
effective material thickness on tail side (teff), were obtained by destructive examination. Haque,
et al. [55] developed a mathematical model to calculate rivet flaring (Δd) instead of using
destructive examination. Further, Haque and Durandet [42] associated the parameters Dc and teff
9 / 54
with the rivet flaring and directly calculated the cross-tension strength of SPR joint, where the
input variables involved empirical coefficient, punch displacement, and the design dimensions
of rivet, sheet and die cavity. They also developed a formula to calculate lap-shear strength
from the cross-tension strength of the steel-to-steel sheet SPR joint. Kim, et al. [56] developed
the analytical strength estimators for SPR joints for the lap-shear and the cross-tension modes,
where the numerical simulation was utilized to reveal the mechanical responses of the rivet and
sheets during the SPR process and the lap-shear and the cross-tension tests. The authors tried
to develop the strength estimators with using the calculated material properties rather than
empirical coefficients in order to simplify the calculation procedure. In their strength models,
the lap-shear strength is the product of the four important parameters are the tensile strength of
sheet metal, the effective contact thickness and diameter (geometric parameters), which are also
used in previous literatures, such as Sun and Khaleel [43] and Haque, Williams, Blacket and
Durandet [55]. However, a reference coefficient representing a ratio to uniform elongation was
also involved in their strength models. The prediction on the failure mode of rivet joint is
another important issue, which is related to the mechanical performance of the SPR joint.
Commonly, the failure mode is the results of the competency between the rivet head pullout
and rivet tail pullout. Therefore, the failure mode of a SPR joint can be curtained by comparing
the strength index of the rivet head pullout and rivet tail pullout, after the strength indexes been
calculated. Porcaro, et al. [57] studied the Eurocode 9 and reported the equation to calculate the
resistance force of the rivet head pullout and rivet tail pullout in cross-tension test, respectively,
where the resistance force of SPR joint is the product of the ultimate tensile strength of material
and geometric dimensions of rivet parts. Calabrese, Bonaccorsi, Proverbio, Di Bella and
10 / 54
Borsellino [6] improved the work of Porcaro, Hanssen, Langseth and Aalberg [57] and
These studies focused on the development of the cross-tension strength model of the SPR
joint based on rivet flaring, but the lap-shear strength models of failure mode I and II have not
been involved. Considering the various SPR processes and joint failure modes in actual
situations, it is necessary to improve the established strength models of SPR joint to extend
their applications, especially considering the different failure modes in the lap-shear test and
This work attempts to develop a general model to predict the strengths and failure modes of
the Al-to-Al sheet SPR joints with one steel rivet, where the strengths include lap-shear strength
and cross-tension strength, and the failure modes includes the rivet tail pullout and rivet head
pullout. It is also possible to calculate the strengths based on their relationships in mathematic
when one of the strengths is known. The destructive tests of cross-tension and lap-shear
specimens were conducted to verify the prediction results which contained the joint strength
and the joint failure mode. Furthermore, the established SPR strength models were embedded
in an industrial internet of things (IIoT) to in line monitor the connection quality of SPR product.
2. Experiment method
In present study, two layers of Al alloy sheets are stacked and connected with a steel rivet by
self-piercing riveting process. The schematic diagram of the SPR mold assemble is shown in
Fig. 1. The material of Al alloy sheets is A6063-T5, and the steel rivet material is 37Cr4 steel.
The chemical compositions of A6063-T5 and 37Cr4 are shown in Table 1 and Table 2,
11 / 54
respectively. The mechanical properties of A6063-T5 and 37Cr4 are shown in Table 3. The Al
alloy sheet under consideration contains two types: (a) type 1 is one-side coated Al sheet, and
(b) type 2 is double-side coated sheet. Compared with the Al alloy sheet and the steel rivet, the
insulation coating can be omitted in the study due to its small thickness and low mechanical
strength.
Fig. 1. The SPR schematic diagram of aluminum alloy sheet: (a) preparation, (b) forming, (c)
demolding.
Element Al Mg Si Cr Mn Ti Cu Zn Fe
Weight/% 97.5 0.45- 0.2- ≤0.1 ≤0.1 ≤0.1 ≤0.1 ≤0.1 ≤0.35
0.9 0.6
Element C Si Mn P S Cr
Weight/% 0.34 - 0.41 ≤0.4 0.6 - 0.9 ≤ 0.025 ≤0.035 0.9 - 1.2
12 / 54
The SPR device used for the Al sheet riveting is shown in Fig. 2. The bottom die is fixed on
the anvil to form the metal into a button, and the punch is placed along horizontal direction to
provide a push force for each riveting process. The rivets are beforehand prepared in a chain in
order to automatically supply rivet for the SPR. An infra-red inspection device has been
equipped to detect whether a steel rivet is already placed on the right position in each SPR
process. Besides, an automatic feeding system is equipped on the production line to keep
The SPR of Al alloy sheet is a procedure of cold metal deformation where a rivet is pressed
into the two or multiple layers of sheets under punch pushing. The load - stroke curve of the
punch is important data to determine the details of rivet deformation. For example, He, et al.
[58] made use of the punch stroke to calculate the rivet flaring. No doubt, The rivet flaring is
the primary factor that dominates the interlock of the stacked sheets and the strength of SPR
joint, as reported by Hoang, et al. [59]. To obtain the load-displacement curve of each SPR
process, an IIoT system was constructed based on hardware and software. The hardware the
13 / 54
IIoT makes use of a displacement sensor assembled on the side of the punch to measure the
displacement of the punch and a pressure sensor amounted on the hydraulic drive station of the
punch to measure the pressure change of the punch. Both the pressure sensor and displacement
sensor have been adjusted to realize the synchronicity of the data acquisition. An analog-to-
digital (A/D) card was utilized to convert the collected pressure-dislocation data from the
analog signal to the digital signal. Afterwards, the digital signals of the displacement and
pressure were transmitted to the software of the IIoT system aiming to sequentially complete
the data processing and the results uploading, including the calculation of SPR joint strength
and the uploading of calculation results to online database. The software of the IIoT is installed
Regarding of the automatic SPR production line (see Fig. 2), the time duration of each SPR
process in practice is limited in 3 seconds, and the time interval between two SPR processes is
limited in 8 seconds. The IIoT system has the capability to catch the pressure-dislocation data
of each SPR process in real time (3 seconds) and rapidly complete the relevant calculation
before the start of the next SPR process. The data sampling frequency of the displacement
sensor and the pressure sensor is set as 1000Hz. The raw data of the IIoT system is saved on
The monitored strength indexes of each SPR joint generally include the cross-tension
strength and lap-shear strength. In order to determine whether the predicted strength of each
SPR joint meets the quality requirements, a standard strength value is required as a reference.
The standard strength value of the cross-tension strength and lap-shear strength were directly
14 / 54
measured by conducting a series of deconstructive tests with SPR joint specimens, as shown in
Fig. 3.
Fig. 3. The SPR specimens of destructive tests: (a) cross-tension specimen of group 1, (b) lip-
shear specimen of group 1, (c) cross-tension specimen of group 3, and (d) lip-shear specimen
of group 4.
In this study, the SPRed specimens for destructive test were divided into four groups: (a)
group 1 is the sheet type 2 to sheet type 1 cross-tension specimen, (b) group 2 is the sheet type
2 to sheet type 1 lap-shear specimen, (c) group 3 is the sheet type 2 to sheet type 2 cross-tension
specimen, (d) group 4 is the sheet type 2 to sheet type 2 lap-shear specimen. The type 1 is one-
side coated sheet, and type 2 is double-side coated sheet. The cross-section views of the SPR
joints in the four groups were observed by using an optical microscope Olympus BX51M and
To analyze the metal deformation in SPR, numerical simulation was conducted by using a 2-
dimensional (2D) finite element (FE) model via software ABAQUS, as shown in Fig. 4. The
2D FE model involved the rigid tools (punch, holder, and die), and the deformable components
(rivet, top sheet and bottom sheet). The material properties of the rivet, the top sheet and the
15 / 54
bottom sheet were set as shown in Table 1. The dimensions of the tools and the deformable
components were the actual size value of relevant parts. The mass reference point of the punch
was applied a velocity of 0.1 m/s on, and the mass reference point of the holder was applied a
holding force of 100N. The degree of freedom of the bottom die was all fixed. The friction
factors between tools and components were set as follows: 0.2 at the interface between steel
rivet and Al alloy sheets, 0.22 at the interface between the Al alloy sheets, 0.2 at the interface
between the tools and the Al alloy sheets, 0.2 at the interface between the tools and the rivet.
The element type of rivet model is hybrid of rectangle and triangle, and the element type of
sheets is rectangle. The element type of the deformable body in the FE model is quadrilateral
element. The element of the top sheet is finer than that of the bottom sheet in order to reduce
the impaction of large distortion in the top sheet on the computation convergence. The coating
Fig. 4. The 2D axis symmetric FE model of SPR for the Al sheet and steel rivet.
16 / 54
3. Results and discussions
In investigating the procedure to fabricate a SPR joint of metal sheets in present industrial
production line, it is common sense to pay attentions to the metal deformation details which
reflect essential characteristics of the SPR process. In practice, the methods to sufficient the
simulation, both of which were coupled with the section view of SPR joint. The former method
provides a clear perspective of SPR joint under microscope view field, even though the
alternative method, numerical simulation performs more efficiently in providing the details of
continuous deformation during a SPR procedure, as shown in Fig. 5. The simulation results of
the 2D FE model clearly presents the metal deformation of rivet, top sheet and bottom sheet at
different punch stroke. With the steel rivet being driven into the op Al alloy sheet with a
thickness of 3mm, the predicted Mises effective stress in the rivet tail continuously is clamming
up to a high value above 800 MPa, which directly force the rivet tail to flare. The simultaneous
deformation occurs in the top alloy sheet that partial material of the top sheet is pierced away
and trapped in the hollow cavity of the rivet tail. When the punch stroke reaches 65%, the rivet
tail already pierces through the top sheet and drives the partial metal of the bottom sheet to fill
the cylinder mold cavity, as shown in Fig. 5(d). With the punch stroke increasing, the rivet
flaring increases and the bottom sheet tends to fill fully the entire mold cavity. Due to the
complexity of metal flowing in the mold cavity, the compress load of the punch against the
metal deformation increase rapidly. A cone structure in the bottom mold is designed to promote
17 / 54
the rivet flaring and to achieve a fully filled button. A final action of the punch is to stamp on
the top sheet to avoid sheet rebound and to reduce sheet gap, leading to the punch load to reach
its maximum value. Moreover, no significant deformation occurs in the rivet heat throughout
Fig. 5. FE simulation results of Mises stress at the different punch stroke: (a) 5%, (b) 25%, (c)
Furthermore, the predicted displacement distribution in Fig.6 suggests the metal flowing
during the SPR process. The maximum horizontal flowing of the sheet metal along X axis
direction is caused by the rivet piercing and the rivet flaring results in the tail deforming in the
opposite direction, which reveals the main principle of the SPR bounding and the generation of
18 / 54
the connection strength. In the vertical direction, the rivet travels a long distance through the
two metal sheet before it struggles to reach the final position. This travel can be divided into
two stages: (1) stage I of the rivet starts from it contacting the top sheet and ends with the rivet
tail going into the mold cavity; (2) stage II is the rivet flaring procedure. However, to define the
precise demarcation point of the two stages is an engineering problem for the various factors in
the real SPR process, such as the hardness of materials. Above all, the rivet flaring is one of the
Fig. 6. The predicted displacement field of SPR joint in (a) total magnitude, (b) X axis
More details of the metal deformation in SPR process can be found in the metallographic
picture of the joint cross-section, as shown in Fig. 7. As discussed in section 3.1, the rivet flaring
19 / 54
is one of the critical factor that dominated the connection strength of SPR joint. The measured
rivet flaring of the group 1 - 4 is 0.69±0.11 mm, 0.65±0.13 mm, 0.73±0.14 mm, and 0.62±
0.15 mm, respectively. The differences between those rivet flaring is caused by the essential
material properties and the key dimensional deviations of the components. The differences of
the material properties commonly includes the density, Young’s modules, Poisson rate, yield
strength and hardness, which have significant contribution to the rivet flaring deformation.
Especially, Haque and Durandet [42] reported that a small change of yield strength of sheet will
largely change the metal deformation in SPR process. The dimensional deviation of each SPR
process is basically the rivet diameter, sheet thickness, and mold cavity size (depth and
diameter). For example, the real rivet is not perfectly axisymmetric, which will cause a small
difference of the rivet flaring in the metallography. More details of the deformed rivet and Al
alloy sheet are shown in Fig. 8. No crack or folds were found at the interface between the rivet
Fig. 7. Optical images of SPR sample: (a) cross-tension specimen of group 1, (b) lip-shear
specimen of group 2, (c) cross-tension specimen of group 3, and (d) lip-shear specimen of
group 4.
20 / 54
Fig. 8. SEM images of SPR sample: (a) cross-section view of SPR joint, (b) rivet flaring, (c)
the center of the bottom button, and (d) the corner of the bottom button.
In spite of the known influence of those factors, the direct measurement of them in each
process is a particularly difficult problem and therefore the quantification with an alternative
method is quite attractive in industrial applications where time and cost are more meaningful.
The load-displacement curve provides a window for comprehensively characterizing the metal
deformation in the SPR process, which combines the various factors including of tool action
and the time evolution of metal deformation resistance. The current study collected the load-
displacement curves in the four groups of SPR specimen, as shown in Fig. 9. No significant
difference is found in those curves which can also be divided into two stages according to the
rivet flaring deformation: (a) stage I is the rivet piercing through the stacked sheets without
significant deformation, (b) stage II is the large deformation resistance from the bottom mold
forcing the rivet tail to flare. When the steel rivet is going to flare, a rapidly increasing of
21 / 54
deformation resistance give a birth to a step on the load-displacement curve, as shown in Fig.
10, where d 0 and d max is the distance of stage I and the total displacement of punch stroke in
(b) lip-shear specimen of group 2, (c) cross-tension specimen of group 3, and (d) lip-shear
specimen of group 4.
Fig. 10. The load-displacement curve of the SPR of two Al alloy sheets with a steel rivet.
22 / 54
3.3. SPR strength test results
Most of the current assessments on the SPR strength are cross-tension strength and lap-shear
strength, which are also considered in this study. With the four groups of the SPR specimens,
the destructive tests were implemented in the same testing conditions, where the failure models
and strength results of the specimens were compared and discussed in the following contents.
Regardless of the coating on the sheets, the failure modes of those specimens can be divided
into three categories (see Fig. 11): (a) failure mode I, the rivet tail pullout from the bottom sheet,
(b) failure mode II, the rivet head pullout from the top sheet, (c) failure mode III, rivet failure.
The failure of rivet (failure mode III), as reported by Haque and Durandet [42], is not found in
the current study perhaps, which is due to the yield strength of the steel rivet utilized here is
Fig. 11. The tested SPR specimens: (a) cross-tension specimen of group 1, (b) lip-shear
specimen of group 2, (c) cross-tension specimen of group 3, and (d) lip-shear specimen of
group 4.
The failure mode I is the main mode observed in the cross-tension tests (group 1 and group
3) while the failure mode II is the main mode observed in the lap-shear tests (group 2 and group
23 / 54
Fig. 12. The failure mode of the SPR specimens: (a,b) rivet head pullout (failure mode II) in
cross-tension tests, (c) rivet tail pullout (failure mode I) in lip-shear test, (d,e) rivet pullout
Fig. 13. The scheme of force distribution in the strength tests: (a) cross-tension test, (b) lap-
shear test.
In the cross-tension test, the top sheet transfers the tensile stress to the rivet heat (yellow area
in Fig. 13) along the vertical direction upwards and the bottom sheet transfer the tensile stress
to the flared rivet tail (red area in Fig. 13) along the vertical direction downwards. Although the
areas of the two parts in Fig. 13 (a) are similar, the riveted joint seems prefer to separate at the
24 / 54
interface between the rivet head and the top sheet. To quantify the failure behavior, the pull-
over strength of the rivet head can be calculated based on the BS 5950-5-1998 published by
𝑃𝑉 = 1.1𝑡1 𝑑ℎ 𝑝𝑦 (1)
where PV is the normal pull-over strength, t1 is the sheet thickness in contact with rivet head
(mm), dh is the diameter of rivet head (mm), py is the yield strength of the sheets. As an instance,
considering a specimen in group 1 (see Fig. 13a), the pull-over strength of the rivet head in the
SPR joint is PV = 1.1t1Dhpy = 1.1×2.5 mm×7.52 mm×120 MPa = 2481.6 N, and the pullout
strength of the rivet tail is PV = 1.1teffDtpy = 1.1×3.8 mm×7.12 mm×120 MPa = 3571.4 N. This
proofs that the rivet tail has higher strength than the rivet head in the cross-tension tests (group
1 and 3), which leads to the pullout of rivet head (failure mode I).
In the lap-shear test (see Fig. 13b), the pullout strength can be similarly calculated by Eq. (1).
As an instance, considering a specimen in group 2 (see Fig. 13b), the pull-over strength of the
rivet head is PV = 1.1t3Dhpy = 1.1×5.5 mm×7.52 mm×120 MPa = 5459.5 N, and the pullout
strength of the rivet tail is PV = 1.1t4Dtpy = 1.1×2.5 mm×7.12 mm×120 MPa = 2349.6 N. This
proofs that the rivet head has higher strength than the rivet tail in the lap-shear tests (group 1
and 3), which leads to the pullout of rivet tail (failure mode II).
In the above calculation, a constant of 120 MPa utilized as the yield stress is inconsistent
with the fact that the large deformation introduced work hardening can significantly improve
the hardness at location therefore the yield strength of the material surrounding the rivet tail
also increase in a linear proportion. As reported by Su, Lin, Lai and Pan [31], the hardness in
the large deformation of the SPRed 6111-T4 aluminum sheet rises to 160 HV while the hardness
25 / 54
of sheet matrix is only 85 HV. The microhardness of the SPRed AA5182-O joint reported by
Ma, et al. [61] can rise to 136 HV, which is much greater than the hardness (79 HV) of the base
metal. The hardness - yield stress linear relationship of Al alloy 7010 is reported by Tiryakioğlu,
where σY is the yield stress of alloy (MPa), HV is the Vickers hardness, and the scope is related
The strength of SPR joint is generally expressed by the maximum cross-tension force and
maximum lap-shear force, both of which can be measured by the experimental tested force-
displacement curve, as shown in Fig. 14. From the force-deformation curves in the cross-
tension tests of group 1 and 3, the average values of the maximum cross-tension force are
3474.6 ± 211.5 MPa and 3326.7 ± 202.7 MPa, respectively. In the case of lap-shear tests of
group 2 and 3, the average values of the maximum lap-shear force are 5690.4 ± 188.1 MPa and
6203.5 ± 336.0 MPa, respectively. A few abnormal test data are excluded from this statistic
because these abnormal data will cause the statistical results to be much greater or lower than
26 / 54
Fig. 14. Force-displacement curves: (a) cross-tension test of group 1, (b) lip-shear tests of
group 2, (c) cross-tension tests of group 3, and (d) lip-shear tests of group 4.
Comparing the tendency of all the force-deformation curves, it is clear that the metal
deformation process of the rivet tail pullout (failure mode I) in the lap-shear tests proceed more
smoothly than that of the rivet heat pull-over (failure mode II) in cross-tension tests. As the
desperation of the data displayed in Fig. 15, the average values of the lap-shear deformation in
the group 2 and 4 are 10.21 ± 0.55 mm and 9.33 ± 0.84 mm, respectively. While the average
values of the cross-tension deformation in group 1 and 3 are 49.77 ± 8.88 mm and 38.40 ±
13.75 mm, respectively. The linear fitting expressions of the cross-tension data are shown in
Eqs. (3) - (6) where FCT1, FCT2, σCT1, σCT2, d is the maximum force in group 1, the maximum
force in group 3, the maximum stress in group 1, the maximum stress in group 3 and the
deformation (unit: mm), respectively. The values of R2 of the four functions suggest the low
27 / 54
𝐹𝐶𝑇1 = 3079.82 + 7.92𝑑, 𝑅 2 = 0.1103 (3)
Fig. 15. The destructive test results: (a) force - deformation, and (b) stress - deformation.
numerical modeling has the disadvantages of low efficiency and poor operability. And the
cost-expensive, which is limited it application to the actual product. However, based on the
numerical analysis and experimental data, establishing an empirical calculation model of SPR
strength might overcome these limitations. As one of the most important mothed, the SPR
strength model based on the rivet flaring has been well developed and verified in a broad range
of sheet materials such as steel and aluminum alloy. Given the three component configuration
as illustrated in Fig. 16, the rivet flaring Δd is defined as the maximum deformation value of
28 / 54
the rivet tail in one side, where Dh, Dt, teff are the rivet head diameter, deformed rivet tail
Fig. 16. Key parameters related to the metal deformation in the SPR.
According to the metal deformation during SPR process discussed in sections, the rivet
piercing through a distance d0 without significant deformation, therefore the major rivet flaring
occurs in the stage II which starts from point O and completes at point C (see Fig. 16). The
position of point C is of significantly important due to it correlated to the following terms: (a)
the maximum diameter of the rivet flaring (Dt), (b) the maximum effective depth in the bottom
sheet (teff), (c) the maximum effective length of the deformed rivet (Lt), (d) the diameter of the
bottom die (Dd) which is the horizontal movement extreme of point C, and (e) the depth of
Constructing a right triangle △ODC, and extending the straight line OC to B which is located
the bottom of the die, another right triangle △OAB is built as illustrated in Fig. 16, where the
rivet flaring Δd equals to the length of CD. Due to the relationship of OD / OA =CD / AB, the
29 / 54
𝛥𝑑 = 𝑂𝐷 ∙ 𝐴𝐵/𝑂𝐴 (7)
Considering the point O as the end point of the rivet deformation stage I (see Fig. 10), the
vertical distance from point O to the top surface of rivet head is equal to d0. Then OD, AB and
𝑂𝐷 = 𝐿𝑡 − 𝑑0 = 𝐶1 ∙ 𝑑𝑚𝑎𝑥 − 𝑑0 , 0 ≤ 𝐶1 ≤ 1 (8)
𝐴𝐵 = 𝐶2 ∙ (0.5𝐷𝑑 − 𝑅𝑟 ), 0 ≤ 𝐶2 ≤ 1 (9)
𝑂𝐴 = 𝑡1 + 𝑡2 + ℎ − 𝑑0 (10)
where Lt is the length of deformed rivet; C1 is a empirical coenfficent related to the rivet length
L (0 ≤ C1 ≤ 1); dmax is the maxomum displacment of the punch which is measured from load-
strokee curve; d0 is the displacement of punch before rivet flaring; C2 is a empirical coenfficent
related to the rivet hardness (0 ≤ C2 ≤ 1); Dd is the diameter of the bottom die; Rr, t1, t2, h is rivet
radius, top sheet thickness, bottom sheet thickness, and the depth of bottom die, respectively.
The empirical expression of rivet flaring can be obtained by substituting Eqs. (8) - (10) into
An empirical estimator of failure mode II (rivet head pullout) is established Sun and Khaleel
where 𝐹ℎ𝑇 is rivet strength for rivet head pullout failure; ηh is empirical coefficient for the
material degradation of sheet due to rivet piercing; βh is empirical coefficient for head side sheet
30 / 54
bending induced thickness reduction, βh = 1 for t1 > 1.0mm, βh ≈ 0.7 for t1 ≤ 1.0mm; π ≈ 3.14;
Similarly, an empirical estimator of failure mode I (rivet tail pullout) is developed by Sun
and Khaleel [44] to calculate the cross-tension strength of SPR joint as follows:
where 𝐹𝑡𝑇 is rivet strength for rivet tail pullout failure; ηt is empirical coefficient for the material
degradation of tail-side sheet due to riveting, ηt = 1 for materials with elongation > 15%, ηt =
0.5 for extrusions or castings on tail end; βt is empirical coefficient for tail side sheet bending
induced thickness reduction, βt = 1 for t2 > 1.0 mm, βt ≈ 0.7 for t2 ≤ 1.0 mm; teff is effective
material thickness on the tail-side; Dt is diameter of rivet flaring; σt is yield strength of base
The diameter of rivet flaring Dt and effective material thickness on the tail-side teff are
ℎ−𝛥𝑑
𝑡𝑒𝑓𝑓 = 𝑡2 + (15)
2
ℎ−𝛥𝑑
𝐹𝑡𝑇 = 0.7𝜂𝑡 𝛽𝑡 ⋅ (𝑡2 − 2
)𝜋 ⋅ 2(𝑅𝑟 + 𝛥𝑑) ⋅ 𝜎𝑡 (16)
Condiserating the balance of failure mode I and II, the rivet strength FCT in cross-tension test
31 / 54
There are two ways to calculate the lap-shear strength of SPR joint: (a) direct calculation of
lap-shear strength, (b) indirect calculation from the cross-tension strength. As an instance,
Haque and Durandet [42] analyzed the strength formula of blind rivet joint in Eurocode 9
(prEN1999-1-4) and proposed two empirical models to directly calculate the SPRed cross-
tension strength of failure mode I and the lap-shear strength of failure mode II as follows:
3
𝐹𝐶𝑇 = 𝛼𝐶𝑇 ⋅ 𝜎𝑡 ∙ √𝐷𝑡 ∙ 𝑡𝑒𝑓𝑓 (13)
where FCT is cross-tension strength; αCT is empirical strength coefficient for cross-tension; FLS
is lap-shear strength; αLS is empirical strength coefficient for lap-shear; teff is effective material
thickness on the tail-side; Dt is diameter of rivet flaring; σt is yield strength of base material in
tail-side sheet. Then the relationship is estiblished between cross-tension strength and lap-shear
strength as follows:
𝛼 𝐷𝑡
𝐹𝐿𝑆 = (𝛼 𝐿𝑆 ⋅ √𝑡 ) ⋅ 𝐹𝐶𝑇 (19)
𝐶𝑇 𝑒𝑓𝑓
Xie, Yan, Yu, Mu and Song [34] compared the strength formulas in Eurocode 9 (prEN1999-
1-4), Eq. (13) proposed by Sun and Khaleel [44], Eq. (16) proposed by Haque and Durandet
standard AISI S100-2016 and Australia standard AS/NZS 4600-2005. They found that Eq. (13)
and Eq. (16) have higher accuracy while the prediction results of Eurocode 9 is more
conservative.
32 / 54
Considering the established models and the factors discussed above, general models for the
strength prediction of SPR joint based rivet flaring is proposed here in terms of test method and
Table 4 General strength models of SPR joint considering the different failure models.
Cross-tension
𝐹𝑡𝐶𝑇 = 𝑎𝑡 𝑡𝑒𝑓𝑓 𝐷𝑡 𝜎𝑡 (20)* 𝐹ℎ𝐶𝑇 = 𝑎ℎ 𝑡ℎ 𝐷ℎ 𝜎ℎ (21)
strength
Lap-shear
𝐹𝑡𝐿𝑆 = 𝑏𝑡 𝑡𝑒𝑓𝑓 𝐷𝑡 𝜎𝑡 (22)* 𝐹ℎ𝐿𝑆 = 𝑏ℎ 𝑡1 𝐷ℎ 𝜎ℎ (23)
strength
* where Dt, teff , and Δd are calculated by Eqs. (14), (15) and (11), respectively; empirical
coefficients at, ah, bt, and bh are obtained by experiment; th is the effective thickness of head-
side material which is calculated with the thickness of head-side sheet t1 and the highth of rivet
Condiserating the balance of failure mode I and II, the rivet strengths are expressed as Eqs.
In the current study, the relationships between cross-tension strength (failure mode I or
failure mode II) and lap-tension strength (failure mode I) are constructed with Eqs. (20), (21)
33 / 54
𝑏𝑡
𝑎𝑡
∙ 𝐹𝑡𝐶𝑇 , 𝑖𝑓 𝐹ℎ𝐶𝑇 ≥ 𝐹𝑡𝐶𝑇
𝐹𝑡𝐿𝑆 = { 𝑏𝑡 𝑡𝑒𝑓𝑓𝐷𝑡𝜎𝑡 (27)
𝑎ℎ 𝑡1 𝐷ℎ 𝜎ℎ
∙ 𝐹ℎ𝐶𝑇 , 𝑖𝑓 𝐹ℎ𝐶𝑇 < 𝐹𝑡𝐶𝑇
The Eq. (26) can be simplified as Eq. (27) when the SPR of similar sheets:
𝑏𝑡
𝑎𝑡
∙ 𝐹𝑡𝐶𝑇 , 𝑖𝑓 𝐹ℎ𝐶𝑇 ≥ 𝐹𝑡𝐶𝑇
𝐹𝑡𝐿𝑆 = { 𝑏 𝑡𝑒𝑓𝑓𝐷𝑡 (28)
𝑡
∙
𝑎 ℎ 𝑡 1 𝐷ℎ
∙ 𝐹ℎ𝐶𝑇 , 𝑖𝑓 𝐹ℎ𝐶𝑇 < 𝐹𝑡𝐶𝑇
The fist term in Eq. (28) is a constant depending on the coefficients in Eqs. (21) and (22).
The second term in Eq. (28) is a variable depending on the geometric dimensions of rivet and
sheets before and after deformation. The third terms in Eq. (28) is the obtained cross-tenison
strength (failure mode II). The above strength models will be verified and discussed in the next
section.
The above-mentioned models cannot be applied to quantify SPR joint strength before the
following inlet data is well prepared: (a) geometric dimensions including of Dh, Rr, t1, t2, h, and
Dd; (b) material properties σh and σt obtained by experimental test; (c) empirical coefficients C1,
C1, αh, and βt; (d) punch displacements d0 and dmax. For the same batch of raw material, the data
in group (a) and (b) can be assumed to be constant which depending on the suppliers’ stringent
quality control. The empirical coefficients in group (c) can be obtained by experimental method.
The punch displacements can be measured from the load-displacement curve of SPR process.
In addition, it also necessary to build an automated system with functions such as data
transmission, data processing, strength calculation, data storage, and abnormal data feedback
during each SPR process. The signal transmission in the proposed IIoT system is shown in the
Fig. 17.
34 / 54
Production lines
Riveter machine
Software on PC
Ok?
No
Database Controller
Fig. 17. Schematic diagram of the hardware and software in the proposed IIoT platform.
Example calculation is conducted by using the strength models (Eqs. 27 and 28) in section
3.4 and the 4 groups of specimens in destructive tests. Material parameters and empirical
coefficients for calculation models of SPR joint strength are given in Table 5 and 6, respectively.
The partial calculation results of the cross-tension strength and lap-shear strength are compared
t1 t2 t0 Rr L Dh Dd h σt σh
(mm) (mm) (mm) (mm) (mm) (mm) (mm) (mm) (MPa) (MPa)
3 3 0.4 2.65 9 7.75 10 2 120 120
35 / 54
Table 6 The empirical coefficients of the SPR joint strength model.
C1 C2 ηt βt ah bt
0.53 0.9 0.55 1 1.4 2
The absolute errors of the predictions on the cross-tension strength and the lap-shear strength
are shown in Fig. 18 (a) and (b), respectively. The average absolute errors of the 4 groups of
data are 5.35%, 5.39%, 4.46%, and 7.50%, respectively. Comparing the errors of the strength
prediction in Fig. 18 (a) and (b), established SPR strength models (Eqs. 27 and 28) perform
36 / 54
sound in predicting the joint strengths (FCT and FLS) of the single-sided or double-sided coated
sheets. The comparison of normalized predicted and measured strengths (FCT and FLS) is shown
in Fig.18 (c) and (d). The premise of the above calculations is that the riveting is perfect.
However, the abnormal situation that may actually occur is rivetless riveting, which can be
filtered out from the normal SPR by mornitoring whether its load-displacement curve passes
Fig. 18. Analysis on the prediction results: (a) predicted FCT and error (%), (b) predicted FLS
and error (%), (c) linear relationship between predicted FCT and measured FCT, and (d) linear
37 / 54
Fig. 19. Load-dislocation curves of rivetless riveting and normal riveting.
4. Conclusions
This study investigated the SPR process of two Al alloy sheets with one steel rivet in order
to establish the general model of the cross-tension strength and lap-shear strength considering
Numerical simulation of finite element method coupled with metallography was utilized
to reveal the cold metal deformation in SPR process. It is clear that the rivet travel through
stacked sheets can be divided into two stages: (a) stage I is the rivet piercing stacked sheets
and going into mold cavity; (b) stage II is the rivet flaring with the rivet severe deformation.
The rivet flaring causes the punch load rapidly increasing, therefore it is possible to
distinguish the starting and ending points of rivet flaring at the load-displacement curve of
punch.
The destructive tests of cross-tension and lap-shear specimens reveal the two failure modes
of SPR joint in this study: rivet tail pullout (failure mode I), and rivet head pullout (failure
mode II). The destructive test results provide credible evidences for the fact that the rivet
38 / 54
flaring is of significantly important to prevent the joint from debonding at the tail-side.
The rivet flaring can be calculated based on the geometric deformation of rivet and stacked
sheets, where the input parameters include of rivet dimensions (head diameter, radius, and
length), mold size (depth and diameter), sheet thickness and so on. This study provided a
new rivet flaring model in order to simplify the calculation procedure and improve the
calculation accuracy.
Base on the principle of rivet flaring, general strength models were proposed to calculate
the cross-tension strength and lap-shear strength from the rivet flaring, material thickness,
and material yield strength, regarding of the failure modes I and II. With the general
strength models of SPR joint, it easy to obtain the relationship between the cross-tension
strength and lap-shear strength, then calculating one of the strength value is possible if the
The general strength models were utilized in industrial internet of things in order to predict
the SPR quality in real time. The predicted results of SPR strength were compared with
the experimental results to verify the general strength models of SPR. In the case of the
SPR for one-side coated Al sheet and double-side coated sheet, the average absolute errors
are 5.35% (cross-tension test) and 5.39% (lap-shear test). In the case of the SPR for two
double-side coated sheets, the average absolute errors are 4.46% (cross-tension test) and
Nomenclature:
39 / 54
L mm rivet length
Dh mm diameter of rivet head
Rr mm rivet radius
H HV rivet hardness
Dd mm diameter of flat die
h mm depth of flat die
teff mm effective length of rivet in bottom sheet
th mm effective thickness of head-side material
Dt mm diameter of deformed rivet
d0 mm punch displacement before rivet flaring
dmax mm maximum punch displacement
C1 - coefficient depends on rivet length (0 ≤ C1 ≤ 1)
C2 - coefficient depends on rivet hardness (0 ≤ C2 ≤ 1)
ηt - empirical coefficient of material degradation
βt - empirical coefficient of sheet thickness reduction
σh MPa yield strength of heat-side sheet
σt MPa yield strength of tail-side sheet
Δd mm rivet flaring
αCT - empirical coefficient for cross-tension
αLS - empirical coefficient for lap-shear
at - empirical coefficient for cross-tension in failure mode I
ah - empirical coefficient for cross-tension in failure mode II
bt - empirical coefficient for lap-shear in failure mode I
bh - empirical coefficient for lap-shear in failure mode II
𝐹𝑡𝐶𝑇 N cross-tension strength in failure mode I
𝐹ℎ𝐶𝑇 N cross-tension strength in failure mode II
𝐹𝑡𝐿𝑆 N lap-shear strength in failure mode I
𝐹ℎ𝐿𝑆 N lap-shear strength in failure mode II
FCT N cross-tension strength
FLS N lap-shear strength
Author contribution: All authors contributed to the study conception and design. Material
preparation, data collection and analysis were performed by Yahui Liu, Zhiwang Zhu, Huipeng
Yu, and Jun Wang. The first draft of the manuscript was written by Yahui Liu and all authors
commented on previous versions of the manuscript. All authors read and approved the final
manuscript.
40 / 54
Data availability: Not applicable.
Declarations
Ethics approval: The present manuscript contains an original research; it has not been
published elsewhere, it has not been submitted simultaneously for publication elsewhere, and
Consent for publication: The present paper does not require any consent to publish since all
References
riveting process and the joint static lap shear strength. The International Journal of Advanced
https://doi.org/10.1007/s00170-017-0156-x
[3] Wu J, Chen C, Ouyang Y, Qin D, Li H (2021) Recent development of the novel riveting
https://doi.org/10.1007/s00170-021-07689-w
41 / 54
[4] Han L, Chrysanthou A, Young K W (2007) Mechanical behaviour of self-piercing riveted
multi-layer joints under different specimen configurations. Materials & Design 28: 2024-2033.
https://doi.org/10.1016/j.matdes.2006.06.015
alternate immersion test of self-piercing riveting aluminium joint. Materials & Design 46: 849-
856. https://doi.org/10.1016/j.matdes.2012.11.016
[7] He X, Zhao L, Deng C, Xing B, Gu F, Ball A (2015) Self-piercing riveting of similar and
dissimilar metal sheets of aluminum alloy and copper alloy. Materials & Design (1980-2015)
steel shear walls with self-piercing riveted connections. Thin-Walled Structures 131: 1-15.
https://doi.org/10.1016/j.tws.2018.06.028
[9] Zhao L, He X, Xing B, Zhang X, Cheng Q, Gu F, Ball A (2017) Fretting behavior of self-
piercing riveted joints in titanium sheet materials. Journal of Materials Processing Technology
[10] Rao H M, Kang J, Huff G, Avery K (2018) Impact of specimen configuration on fatigue
42 / 54
[11] Gay A, Lefebvre F, Bergamo S, Valiorgue F, Chalandon P, Michel P, Bertrand P (2016)
Fatigue performance of a self-piercing rivet joint between aluminum and glass fiber reinforced
https://doi.org/10.1016/j.ijfatigue.2015.10.004
riveted aluminium to high strength low alloy sheets with different surface coatings. Materials
connections using aluminium rivets. International Journal of Solids and Structures 47: 427-439.
https://doi.org/10.1016/j.ijsolstr.2009.10.009
investigation of the role of rivet and die design on the self-piercing riveting joint characteristics
https://doi.org/10.1016/j.jmapro.2021.07.049
[15] Sun X, Khaleel M A (2007) Dynamic strength evaluations for self-piercing rivets and
resistance spot welds joining similar and dissimilar metals. International Journal of Impact
[16] Sun X, Stephens E, Khaleel M (2007) Fatigue behaviors of self-piercing rivets joining
similar and dissimilar sheet metals. International Journal of Fatigue 29: 370-386.
https://doi.org/10.1016/j.ijfatigue.2006.02.054
43 / 54
[17] Ma Y, Niu S, Liu H, Li Y, Ma N (2021) Microstructural evolution in friction self-piercing
riveted aluminum alloy AA7075-T6 joints. Journal of Materials Science & Technology 82: 80-
95. https://doi.org/10.1016/j.jmst.2020.12.023
AA7075-T6 aluminum alloy joints. Journal of Materials Science & Technology 105: 109-121.
https://doi.org/10.1016/j.jmst.2021.07.026
[19] Yang B, Shan H, Liang Y, Ma Y, Niu S, Zhu X, Li Y (2022) Effect of adhesive application
https://doi.org/10.1016/j.jmatprotec.2021.117336
aluminum alloy to magnesium alloy using a flat die. Journal of Magnesium and Alloys.
https://doi.org/10.1016/j.jma.2020.12.016
Riveted and Friction Self-Piercing Riveted Aluminum Alloy and Magnesium Alloy Joints.
[22] Ying L, Gao T, Dai M, Hu P, Dai J (2021) Towards joinability of thermal self-piercing
riveting for AA7075-T6 aluminum alloy sheets under quasi-static loading conditions.
https://doi.org/10.1016/j.ijmecsci.2020.105978
44 / 54
[23] Han D, Yang K, Meschut G (2021) Mechanical joining of glass fibre reinforced polymer
[24] Jiang H, Gao S, Li G, Cui J (2019) Structural design of half hollow rivet for
electromagnetic self-piercing riveting process of dissimilar materials. Materials & Design 183:
108141. https://doi.org/10.1016/j.matdes.2019.108141
[25] Danyo M W, Self-piercing riveting (SPR) in the automotive industry: an overview, 2014,
pp. 171-180.
riveting method for joining of continuous carbon fiber reinforced composite and aluminum
https://doi.org/10.1016/j.compstruct.2020.113219
[27] Karim M A, Bae J-H, Kam D-H, Kim C, Choi W-H, Park Y-D (2020) Assessment of
https://doi.org/10.1016/j.surfcoat.2020.125726
[28] Kyoung-Yun Kim, Jaemun Sim, Noor-E Jannat, Fahim Ahmed, Ameri S (2019)
1150. https://doi.org/
10-20. https://doi.org/10.1016/j.jmatprotec.2005.05.048
45 / 54
[30] Lou M, Li Y, Wang Y, Wang B, Lai X (2014) Influence of resistance heating on self-
piercing riveted dissimilar joints of AA6061-T6 and galvanized DP590. Journal of Materials
[31] Su Z-M, Lin P-C, Lai W-J, Pan J (2015) Fatigue analyses of self-piercing rivets and
clinch joints in lap-shear specimens of aluminum sheets. International Journal of Fatigue 72:
53-65. https://doi.org/10.1016/j.ijfatigue.2014.09.022
[33] Yan W, Xie Z, Yu C, Song L, He H (2017) Experimental investigation and design method
for the shear strength of self-piercing rivet connections in thin-walled steel structures. Journal
connections in thin-walled steel structures. Journal of Constructional Steel Research 144: 211-
220. https://doi.org/10.1016/j.jcsr.2018.01.022
[35] Kang S-H, Kim H-K (2015) Fatigue strength evaluation of self-piercing riveted Al-5052
joints under different specimen configurations. International Journal of Fatigue 80: 58-68.
https://doi.org/10.1016/j.ijfatigue.2015.05.003
[36] Wu G, Li D, Lai W-J, Shi Y, Kang H, Peng Y, Su X (2021) Fatigue behaviors and
46 / 54
[37] Presse J, Künkler B, Michler T (2021) Stress-based approach for fatigue life calculation
1648(03)00274-6
metal Self-Piercing Riveting (SPR) joint and coating behaviour under corrosive environment.
[40] Liu Y, Li H, Zhao H, Liu X (2019) Effects of the die parameters on the self-piercing
riveting process. The International Journal of Advanced Manufacturing Technology 105: 3353-
3368. https://doi.org/10.1007/s00170-019-04567-4
behavior and fretting of self-piercing riveted joints in aluminum alloy 5052. Materials & Design
[42] Haque R, Durandet Y (2016) Strength prediction of self-pierce riveted joint in cross-
https://doi.org/10.1016/j.matdes.2016.07.029
https://doi.org/10.1016/s1526-6125(05)70085-2
47 / 54
[44] Sun X, Khaleel M A (2005) Strength estimation of self-piercing rivets using lower bound
limit load analysis. Science and Technology of Welding and Joining 10: 624-635.
https://doi.org/10.1179/174329305x57491
[45] Karim M A, Jeong T-E, Noh W, Park K-Y, Kam D-H, Kim C, Nam D-G, Jung H, Park
Y-D (2020) Joint quality of self-piercing riveting (SPR) and mechanical behavior under the
frictional effect of various rivet coatings. Journal of Manufacturing Processes 58: 466-477.
https://doi.org/10.1016/j.jmapro.2020.08.038
[46] Zhao H, Han L, Liu Y, Liu X (2021) Modelling and interaction analysis of the self-
pierce riveting process using regression analysis and FEA. The International Journal of
06519-9
[47] Wang D, Kong D, Xie C, Li S, Zong L (2022) Study on the effect of rivet die parameters
on joint quality of self-piercing riveting employed 3D modeling and MCDM method. The
https://doi.org/10.1007/s00170-022-08759-3
715-721. https://doi.org/10.1007/s00170-013-5072-0
https://doi.org/10.1016/j.engfracmech.2018.12.020
48 / 54
[50] Du Z, Duan L, Jing L, Cheng A, He Z (2021) Numerical simulation and parametric
multiphysics self piercing riveting process based on the finite element technique. Advances in
simulation of the SPR process using a thermo-mechanical finite element analysis. Journal of
https://doi.org/10.1016/j.jmatprotec.2016.05.001
[53] Lukas Potgorschek J D, Florian Hönsch, Christof Sommitsch, Stefan Kaufmann (2020)
pp. 413-418.
[54] Hönsch F, Domitner J, Sommitsch C, Götzinger B (2020) Modeling the Failure Behavior
of Self-Piercing Riveting Joints of 6xxx Aluminum Alloy. Journal of Materials Engineering and
[55] Haque R, Williams N S, Blacket S E, Durandet Y (2015) A simple but effective model
for characterizing SPR joints in steel sheet. Journal of Materials Processing Technology 223:
225-231. https://doi.org/10.1016/j.jmatprotec.2015.04.006
[56] Kim C, Min K M, Choi H, Kim H J, Lee M-G (2021) Development of analytical strength
estimator for self-piercing rivet joints through observation of finite element simulations.
49 / 54
International Journal of Mechanical Sciences 202-203: 106499.
https://doi.org/10.1016/j.ijmecsci.2021.106499
[58] He X, Pearson I, Young K (2008) Self-pierce riveting for sheet materials: State of the
https://doi.org/10.1016/j.jmatprotec.2007.10.071
https://doi.org/10.1016/j.ijsolstr.2012.05.039
Code of Practice for Design of Cold Formed Thin Gauge Sections. British Standard Institution,
London.
https://doi.org/10.1016/j.eng.2020.06.015
50 / 54
51 / 54
Caption of Tables and Figures
Fig. 1. The SPR schematic diagram of aluminum alloy sheet: (a) preparation, (b) forming, (c)
demolding.
Fig. 3. The SPR specimens of destructive tests: (a) cross-tension specimen of group 1, (b) lip-
shear specimen of group 1, (c) cross-tension specimen of group 3, and (d) lip-shear specimen
of group 4.
Fig. 4. The 2D axis symmetric FE model of SPR for the Al sheet and steel rivet.
Fig. 5. FE simulation results of Mises stress at the different punch stroke: (a) 5%, (b) 25%, (c)
Fig. 6. The predicted displacement field of SPR joint in (a) total magnitude, (b) X axis
ig. 7. Optical images of SPR sample: (a) cross-tension specimen of group 1, (b) lip-shear
specimen of group 2, (c) cross-tension specimen of group 3, and (d) lip-shear specimen of
group 4.
Fig. 8. SEM images of SPR sample: (a) cross-section view of SPR joint, (b) rivet flaring, (c)
the center of the bottom button, and (d) the corner of the bottom button.
(b) lip-shear specimen of group 2, (c) cross-tension specimen of group 3, and (d) lip-shear
specimen of group 4.
Fig. 10. The load-displacement curve of the SPR of two Al alloy sheets with a steel rivet.
52 / 54
Fig. 11. The tested SPR specimens: (a) cross-tension specimen of group 1, (b) lip-shear
specimen of group 2, (c) cross-tension specimen of group 3, and (d) lip-shear specimen of
group 4.
Fig. 12. The failure mode of the SPR specimens: (a,b) rivet head pullout (failure mode II) in
cross-tension tests, (c) rivet tail pullout (failure mode I) in lip-shear test, (d,e) rivet pullout
Fig. 13. The scheme of force distribution in the strength tests: (a) cross-tension test, (b) lap-
shear test.
Fig. 14. Force-displacement curves: (a) cross-tension test of group 1, (b) lip-shear tests of
group 2, (c) cross-tension tests of group 3, and (d) lip-shear tests of group 4.
Fig. 15. The destructive test results: (a) force - deformation, and (b) stress - deformation.
Fig. 16. Key parameters related to the metal deformation in the SPR.
Fig. 17. Schematic diagram of the hardware and software in the proposed IIoT platform.
Fig. 18. Analysis on the prediction results: (a) predicted FCT and error (%), (b) predicted FLS
and error (%), (c) linear relationship between predicted FCT and measured FCT, and (d) linear
Table 4 General strength models of SPR joint considering the different failure models.
53 / 54
Table 5 The material parameters of the SPR joint strength model.
54 / 54