Journal of Materials Processing Technology 237 (2016) 31–47

Contents lists available at ScienceDirect

Journal of Materials Processing Technology

journal homepage: www.elsevier.com/locate/jmatprotec

Analysis of longitudinal weld seam defects and investigation of

solid-state bonding criteria in porthole die extrusion process of
aluminum alloy profiles
Junquan Yu, Guoqun Zhao ∗ , Liang Chen
Key Laboratory for Liquid-Solid Structural Evolution and Processing of Materials (Ministry of Education), Shandong University, Jinan, Shandong 250061, PR
China

a r t i c l e i n f o a b s t r a c t

Article history: In porthole die extrusion process of aluminum alloy profiles, the formation of longitudinal weld seams
Received 22 January 2016 (L-seams) is inevitable. Analysis of L-seam defects and investigation of solid-state bonding criteria are
Received in revised form 7 April 2016 important issues for practical production. In this study, a set of modular porthole extrusion dies with
Accepted 21 May 2016
different depths of welding chambers were designed and manufactured. The profiles extruded with dif-
Available online 24 May 2016
ferent depths of welding chambers were obtained by performing extrusion experiments.The welding
quality of extruded profiles was characterized by means of microstructure observation, tensile test and
Keywords:
fracture analysis. The true stress-strain curves of homogenized AA6063 aluminum alloy with various
Aluminum alloy profile
Extrusion
strain rates and deformation temperatures were obtained by means of isothermal hot compression test.
Weld seam defect Three-dimensional transient numerical simulation models of the porthole die extrusion processes were
Welding path established. Flow behaviors of aluminum alloy in porthole die extrusion process were investigated, and
Bonding criterion the formation processes of L-seams and their defects were revealed. The solid-state bonding processes
of metal particles were traced and their welding paths were determined. Finally, based on the plastic
deformation and diffusion mechanisms for closure behaviors of micro voids on contact interfaces, a new
dimensionless solid-state bonding criterion related to stress triaxiality, effective strain rate, temperature
and contact time was proposed. The bonding quality of L-seams of extruded profiles without any macro
defect was predicted successfully by using this new criterion. In addition, it was found that the forma-
tion of a macro hole in the profile extruded with a shallow welding chamber is attributed to metal flow
behavior, and does not relate to solid-state bonding process.
© 2016 Elsevier B.V. All rights reserved.

1. Introduction
Due to the advantages of recyclability, light-weight and excel-
lent corrosion resistance and mechanical properties, aluminum
alloy profiles play an important role in many fields, such as railway
transportation, shipbuilding, automobile industry, aerospace man-
ufacturing, electronic equipment and engineering structure. As a
typical plastic deformation process, hot extrusion process has been
widely used to produce various kinds of aluminum alloy profiles.
With the development of industry, the hollow section profiles with
large scale, thin-wall, multi-cavity and complex cross-section, as
shown in Fig. 1, are drawing more and more attention and the
demand for these profiles is increasing significantly. For the pur-
∗ Corresponding author.
E-mail addresses: yujunquansdu@163.com (J. Yu), zhaogq@sdu.edu.cn (G. Zhao),
chenliang@sdu.edu.cn (L. Chen).
pose of high productivity, this kind of aluminum alloy profiles are
usually extruded by using porthole die direct extrusion process.
In porthole die extrusion process of hollow section profiles,
when a preheated aluminum alloy metal is extruded into portholes,
it is firstly split into several separate streams by bridges, and then
the streams are rejoined together in welding chamber. Finally, the
rejoined metal breaks through the bearing of lower die and forms
a profile with longitudinal weld seams (L-seams) inside. Owing to
improper extrusion process parameter or die structure, unsound
bonding may occur in the forming process of a hollow section pro-
file. Kim et al. (2002) stated that the failure of a hollow section
profile mostly occurs along or around L-seams when the profile is
subjected to severe internal pressure in the practical use. Therefore,
L-seams are usually the weakest areas of hollow section profiles.
In practice, the welding quality of profiles with large scale, multi-
cavity and complex cross-section can be detected by means of burst
test. Fracture always occurs at the position where the L-seam has

http://dx.doi.org/10.1016/j.jmatprotec.2016.05.024
0924-0136/© 2016 Elsevier B.V. All rights reserved.
32 J. Yu et al. / Journal of Materials Processing Technology 237 (2016) 31–47
Fig. 1. Aluminum alloy profiles with large scale, thin-wall, multi-cavity and complex
cross-section.
poor welding quality, as shown in Fig. 2. Consequently, investi-
gating the formation mechanism of L-seams and their defects and
establishing criterion for welding quality prediction have impor-
tant theoretical significance and engineering application values for
extrusion die and extrusion process design to ensure the welding
quality of L-seams.
Defects of L-seams of profiles extruded by porthole dies can be
classified into two categories: macro and micro defects. The macro
defects refer to macro holes or gaps which are formed on unbond-
ing surfaces. The micro defects refer to micro voids, cracks, oxide
films, or other inclusions, which result in unsound bonding of metal
atoms on bonding interface. In practise, defects of L-seams are
mainly caused by improper extrusion die structure. Den Bakkera
et al. (2014) found that a very shallow welding chamber can cause
the appearance of a macro hole defect on profile’s cross-section.
Valberg et al. (1995) found that a relative shallow welding cham-
ber can lead to a transition of sticking friction to sliding friction at
the rear end of bridge. This transition results in the formation of
free metal surfaces which are pressed into each other by classical
pressure welding to form L-seams. As a result, some oxide particles
may exist in L-seams. Furthermore, Donati and Tomesani (2005)
found that a deep and large welding chamber contributes to the
improvement of welding quality and extrusion speed limit.
In order to evaluate the welding quality of L-seams of extruded
aluminum alloy profiles, researchers mainly carried out following
studies.
(1) Bonding criteria. Several bonding criteria were established
according to extrusion experimental results and theoretical
derivation. Akeret (1972) earliest proposed the maximum pres-
sure criterion, which is stated as that the maximum pressure
Fig. 2. Burst test for a hollow section aluminum alloy profile: (a) test equipment, (b) fracture location.
inside the welding chamber must exceed a critical limit to form
a sound weld seam. More recently, Plata and Piwnik (2000) pro-
posed the pressure-time criterion (Q criterion), which is stated
as that the integral of ratio of contact pressure to material flow
stress on time on welding surface must exceed a critical value to
obtain a sound weld seam. Finally, Donati and Tomesani (2004)
proposed the pressure-time-flow criterion (K criterion), which
is stated as that a sound weld seam can be formed once the
integral of ratio of contact pressure to material flow stress on
all possible welding paths on welding surface exceeds a critical
value.
(2) Numerical simulations. In the extrusion process with a port-
hole die, relevant physical variables such as pressure, effective
stress and metal flow velocity are important for welding quality
evaluation of L-seams, and these physical variables are usually
obtained by means of numerical simulation. Currently, the port-
hole die extrusion process is simulated mainly by using Deform
software based on Lagrange algorithm or HyperXtrude software
based on arbitrary Lagrange-Euler (ALE) algorithm. Based on
the bonding criteria and the numerical simulation softwares
mentioned above, Güley et al. (2013) studied the effect of die
design on the welding quality during solid state recycling of
AA6060 chips, and Chen et al. (2011) evaluated the welding
quality of a complex aluminum alloy multi-cavity wallboard of
high-speed train.
(3) Physical simulations. In fact, the formation of an L-seam in pot-
hole die extrusion process is belonged to solid-state bonding
process. Physical simulation means that the formation of an L-
seam in porthole die extrusion process is simulated by means
of some basic solid-state bonding tests, such as the uniaxial
hot compression test used by Edwards et al. (2009) and the flat
rolling experiment reported by Ceretti et al. (2009). By combin-
ing numerical simulation with uniaxial hot compression test,
Tang et al. (2014) evaluated the weld seam quality of a micro
channel tube according to the relationship of welding chamber
height, hydrostatic pressure and L-seam welding strength.
Although some research has been done to investigate the L-
seams’ defects and the bonding criteria in porthole die extrusion
process of aluminum alloy profiles, several critical problems still
need to be further classified.
Firstly, there are still some problems for the evaluation method
of welding quality of L-seams. Most of previous studies took the
average effect of the whole bonding zone of an L-seam on profile’s
cross-section as a standard to evaluate the welding quality of the
L-seam. Namely, only the overall effect of relevant physical quanti-
ties of all the particles in welding plane on the welding quality of an
L-seam is considered. However, the recent study of Yu et al. (2016)
showed that the welding quality in the whole bonding zone of an L-
seam on profile’s cross-section is usually variable. Since the failure
is easy to originate at the weakest bonding position, the welding
quality of each particle of an L-seam should be evaluated individ-
 J. Yu et al. / Journal of Materials Processing Technology 237 (2016) 31–47
Fig. 3. Experimental extrusion dies.
Fig. 4. Dimensions of bridges (Unit: mm).
ually. For this purpose, the true welding path of each particle of an
L-seam should be determined.
Secondly, the existing bonding criteria are still not perfect. For
the solid-state bonding process of aluminum alloy extrusion, the
existing bonding criteria mainly take pressure, effective stress,
velocity and time into consideration. Both the Q and K criteria con-
sider the effect of temperature on bonding quality just through the
effect of temperature on material flow stress. The experimental
results of Bariani et al. (2006) showed that the increase of the ratio
of hydrostatic pressure to effective stress allows complete weld-
ing to take place at a lower temperature. Ceretti et al. (2009) found
that the critical value of the Q criterion is a function of tempera-
ture. In addition, the experimental results of Valberg et al. (1995)
demonstrated that improper extrusion die structure can result in
an incomplete filling of metal in welding chambers and the for-
mation of gas pockets. In this case, the surfaces of split metal
streams are oxidized to some extent. Edwards et al. (2009) further
revealed that the surface stretching parameter (strain) is the most
influential parameter for the bonding process of specimens that
were covered with oxide layers. However, the K criterion based on
the experiments of Valberg et al. (1995) does not take strain into
consideration. Therefore, a more realistic and effective bonding cri-
terion needs to be established from the viewpoint of more extensive
influencing factors.
Thirdly, although some welding defects of L-seams were discov-
ered, there is still lack of a deep and detailed investigation on the
formation mechanism of the welding defects. Since the defects of
L-seams directly impact the performance of extruded profile, it is
necessary to study the formation mechanism of the welding defects
from both the macro and micro levels.
33
Based on the aforementioned reasons, in this study, firstly, a
set of modular porthole extrusion dies with different depths of
welding chambers were designed and manufactured. The profiles
extruded with different depths of welding chambers were obtained
by performing extrusion experiments. Welding quality of extruded
profiles was systematically characterized by means of microstruc-
ture observation, tensile test and fracture analysis. After that,
three-dimensional transient numerical simulation models of the
porthole die extrusion processes were established. Flow behaviors
of aluminum alloy in the porthole die extrusion process were inves-
tigated, and the formation processes of L-seams and their defects
were revealed. Solid-state bonding processes of metal particles
were traced and true welding paths were determined. Finally, based
on experimental results and solid-state bonding theory and by syn-
thesizing various factors affecting solid-state bonding process, a
new dimensionless solid-state bonding criterion was established.
This new criterion considered the actions of both plastic deforma-
tion and atomic diffusion from the viewpoint of closure behaviors
of micro voids on contact interfaces. According to this new bonding
criterion, the bonding quality of L-seams obtained by means of the
extrusion experiments was successfully predicted.
2. Experimental methods and procedure
2.1. Extrusion experiment
In this study, plate-shaped profiles were investigated, and the
extrusion dies used to extrude these profiles were designed in a
modular way, as shown in Fig. 3. This kind of extrusion die consists
of the upper die, lower die and bridge. The rear end of the bridges
34 J. Yu et al. / Journal of Materials Processing Technology 237 (2016) 31–47
Fig. 5. The position of different samples and the dimension of tensile specimen.
Fig. 6. Finite element model of the aluminum alloy profile extrusion process.
were arched and pointed. Fig. 4 gives the dimensions of the bridges,
where the h represents the depth of welding chambers. In order
to obtain the extruded profiles with different defects of L-seams,
welding chambers were designed as four different depths, namely,
h of the bridges is 5, 15, 25 and 35 mm, respectively.
The as-cast billets of AA6063 aluminum alloy with dimensions
of Ø152 mm × 450 mm were selected and the homogenization was
carried out at 530 ◦ C for 14 h and then cooled down to room tem-
perature in the air. Table 1 gives the chemical compositions of
AA6063 aluminum alloy. Extrusion experiments were performed
on a 16 MN extrusion press. The temperatures of container and
extrusion dies were 450 ◦ C and 500 ◦ C respectively, and the tem-
perature of the billets was 500 ◦ C. The ram speed was set to be
1.0 mms−1 . The extruded profiles were rapidly quenched with
spraying water. The extrusion loads were obtained by calculating
the sum of products of one main hydraulic pressure and two side
hydraulic pressures with the areas of cylinders respectively.
2.2. Characterization of microstructure and mechanical
properties of L-seams
After extrusion experiments, the plate-shaped profiles with
12 mm thickness, 80 mm width and about 8 m length were
obtained. The L-seam is located in the mid-plane of the profile in
width direction, as shown in Fig. 5. In order to ensure the stability
of the samples’ performance, all the samples were taken from the
middle parts of extruded profiles in length direction. The samples
Table 1
Chemical compositions of AA6063 aluminum alloy.
Element Si Fe Cu Mn Mg Cr Zn Ti Al
(wt%) 0.2–0.6 ≤0.35 ≤0.10 ≤0.10 0.45–0.90 ≤0.10 ≤0.10 ≤0.10 Bal.
for microstructure observation of L-seams were polished and then
etched in the solution of 1.0 ml HF, 1.5 ml HCl, 2.5 ml HNO3 and
95.0 ml H2 O. The light optical microscopy was used for microstruc-
ture observation. Due to the limited width of extruded profile, the
dimensions of the tensile specimens are not compliant with rel-
evant standard tensile samples. Nevertheless, the results of the
tensile tests are still reasonable for the comparison of the different
profiles obtained from extrusion experiments. The tensile speci-
mens are tested on a hydraulic testing machine with high pressure
clamps. All the tensile tests were performed at a constant speed
of 0.01 mms−1 , and repeated three times for each configuration to
ensure the accuracy and repeatability of the tests. After the tensile
test, the fracture surfaces of tensile specimens were observed by
means of scanning electron microscopy (SEM).
3. 3-D numerical simulation modeling
3.1. Elements meshing
The porthole die extrusion process was simulated by means of
Deform-3D software. Due to the symmetry of extrusion die and
 J. Yu et al. / Journal of Materials Processing Technology 237 (2016) 31–47
Fig. 7. True stress-strain curves of the homogenized AA6063 aluminum alloy under various strain rates and at the deformation temperatures of (a) 400 ◦ C, (b) 450 ◦ C, (c)
500 ◦ C, (d) 530 ◦ C.
Table 2
Physical properties of materials used for billet and extrusion tools.
Material Young’s modulus (Pa) Poisson’s ratio
AA6063 6.9E + 10 0.33
H13 tool steel 2.1E + 11 0.30
deformation process, a quarter of the numerical simulation model
was setup to reduce computational time, as shown in Fig. 6. The
components of the numerical simulation model include stem, con-
tainer, billet, bridge, upper die and lower die. All these components
were meshed into tetrahedral elements. Since the Lagrange algo-
rithm is used in Deform-3D software, the elements of deformation
metal are remeshed continuously in the simulation process. In
order to ensure the accuracy of simulation, the absolute mesh-
ing method was adopted. This meshing method means that the
minimum and maximum sizes of elements of deformation metal
are invariable in the extrusion process. However, the number of
elements increases accordingly with the increase of complexity of
geometrical shape of deformation metal. As a result, the geometri-
cal shape of deformation metal can be described accurately. Liu et al.
(2008) confirmed the advantage of the absolute meshing method
through simulating the extrusion process of a thin-walled square
magnesium tube. In addition, in order to predict defects of L-seams,
local mesh refinement was also used in welding area.
3.2. True stress-strain curves of homogenized AA6063 aluminum
alloy
In order to precisely describe the high-temperature plastic
deformation behavior and calculate the flow stress of homogenized
AA6063 aluminum alloy used in this study. The cylindrical speci-
mens of 15 mm in height and 10 mm in diameter were machined
from the homogenized AA6063 aluminum alloy billet, and the
isothermal hot compression tests were carried out by using Gleeble
35
Density (Kg/m3 ) Thermal conductivity [N/(s ◦ C)] Heat capacity [N/(mm2 ◦ C)]
2700 180.2 2.27
7760 28.4 5.6
3500 thermo-simulation machine. The deformation temperatures
were 400, 450, 500 and 530 ◦ C, and the strain rates were set to
be 0.01, 0.1, 1 and 10 s−1 , respectively. Fig. 7 gives the true stress-
strain curves of the homogenized AA6063 aluminum alloy under
various strain rates and at different deformation temperatures.
According to the obtained results of isothermal hot compression
tests, the values of true stresses and strains of a series of points on
the true stress-strain curves were imported into Deform-3D soft-
ware. When strain is larger than 0.8, the corresponding flow stress
is calculated by means of trend extrapolation.
3.3. Extrusion parameters and boundary conditions
All the extrusion process parameters in the numerical simula-
tion models were set in accordance with the extrusion experiments.
The extrusion ratio is 20.9. The physical properties of materials used
for billet and extrusion tools are listed in Table 2.
The boundary conditions in the numerical simulation model
mainly include heat transmission boundary conditions and fric-
tional conditions. According to the report of Fang et al. (2009),
the heat transfer coefficient between tooling and billet was set
as 11 N/s mm ◦ C, and the heat transfer coefficients between tool-
ing/billet and air were set as 0.02 N/s mm ◦ C. The friction behavior
between the die and the billet was described with the following
plastic shear friction model.
 = my (1)
36 J. Yu et al. / Journal of Materials Processing Technology 237 (2016) 31–47
Fig. 8. The macroscopic and microscopic observation results of L-seams on the cross-sections of profiles extruded with different depths of welding chambers.
where  is the shear frictional stress,  y the shear yield stress and
m is the friction factor. Based on both numerical simulation and
experimental work, Flitta and Sheppard (2003) found that the fric-
tion conditions should be variable in the extrusion process. In the
research work of Zhou et al. (2003), the friction factor of 0.3 was
set on the interface between the billet and the bearing of lower
die, and 0.9 between the billet and the container or the other areas
of the extrusion die. In this work, the friction factor on the inter-
face between the billet and the bearing of lower die was selected
as 0.3, while the friction factor on the interface between the bil-
let and the container or the other areas of the extrusion die was
respectively selected as 0.8, 0.9 and 1.0. The numerical simulations
of extrusion process for the above three pair of friction factors were
carried out by using Deform software. The results showed that sim-
ulated profile fronts and the maximum ram forces are the closest
to the experimental results when the friction factor m on the inter-
face between the billet and the container or the other areas of the
extrusion die was set as 0.9 and the friction factor on the interface
between the billet and the bearing of lower die was selected as 0.3.
4. Results and discussion
4.1. Microstructure and mechanical properties of L-seams
4.1.1. Microstructure
Fig. 8 shows the macroscopic and microscopic observation
results of L-seams on the cross-sections of the profiles extruded
by the experimental dies with the depth of welding chambers h = 5,
15, 25 and 35 mm, respectively. The macro-observation results of
L-seams show that there is no any macro welding defect on the
cross-sections of profiles extruded with h = 15, 25 and 35 mm, as
shown in Fig. 8(e), (i) and (m). However, an elliptic macro hole
exists in the central zone of cross-section of the profile extruded
with h = 5 mm, as shown in Fig. 8(a). This elliptic macro hole pen-
etrates through the whole profile along length direction and the
dimensions of this hole are around 5.0 mm in thickness direction
and 1.6 mm in width direction of the profile.
Further metallographic observation results show that the
microstructures of L-seams are very different along the thick-
ness direction of extruded profiles. For the profile extruded with
h = 5 mm, the microstructures of the zones B and C of the L-seam
are respectively presented in Fig. 8(c) and (d). The white boundaries
in the middle of profile’s cross-section represent the bonding inter-
face. For the profile extruded with h = 15 mm, the bonding interface
in the central zone A exhibits a dark boundary, as shown in Fig. 8(f).
This dark boundary gradually narrows down from the center to the
side of the cross-section along the thickness directions of the pro-
file and transforms to be a white boundary at the position of the
transition point T1 in the zone B, as shown in Fig. 8(g). The bond-
ing interface in the zone C exhibits a white boundary, as shown in
Fig. 8(h). Furthermore, by means of microstructure observation and
measurement, the lengths of the dark and white boundaries were
finally determined as 2.5 mm and 9.5 mm respectively, as shown
in Fig. 8(e). For the profile extruded with h = 25 mm, the L-seam’s
 J. Yu et al. / Journal of Materials Processing Technology 237 (2016) 31–47
Fig. 9. The engineering stress-strain curves obtained by tensile tests and the fracture angles of tensile specimens.
microstructure is also different along the thickness direction of the
profile. However, the dark boundary in the central zone A is rela-
tively narrow, and the length of this dark boundary is shortened to
1.5 mm and the white boundary is increased to 10.5 mm, as shown
in Fig. 8(j) and (i). The transition position is defined as T3 . For the
profile extruded with h = 35 mm, only white boundary exists on the
whole bonding interface of the L-seam, as shown in Fig. 8(n)–(p),
and thus the length of the white boundary is 12.0 mm. As a result,
it can be concluded that the depths of welding chambers have a
significant influence on the microstructure of L-seams.
4.1.2. Mechanical properties and fracture feature analyses
Fig. 9 gives the engineering stress-strain curves and the frac-
ture angles of tensile specimens. The T-1–T-4 represent the tensile
specimens taken from the profiles extruded with h = 5, 15, 25 and
35 mm, respectively. According to the tensile test results, the profile
extruded with h = 5 mm has the lowest tensile strength (264 MPa)
and elongation (0.054) due to the existence of the elliptic macro
hole. The fracture of the tensile specimen T-1 penetrates through
L-seam, and the fracture angle is about 40 ◦ . The yield and tensile
strengths of profiles extruded with h = 15, 25 and 35 mm are about
175 and 275 Mpa, but the entire engineering strains before fracture
(fracture elongations) and the fracture angles of the tensile speci-
mens are obviously different. The profile extruded with h = 15 mm
has a relatively large fracture elongation (0.398). However, the fail-
ure of the tensile specimen T-2 occurs almost along L-seam. The
fracture surface of T-2 is smooth, and the fracture angle is about 0 ◦ .
The profile extruded with h = 25 mm has a larger fracture elonga-
tion (0.455) than the profile extruded with h = 15 mm. The fracture
surface of the tensile specimen T-3 is unsmooth, and the fracture
angle is about 15 ◦ . The profile extruded with h = 35 mm has the
largest fracture elongation (0.500). The fracture surface of the ten-
sile specimen T-4 is also unsmooth, but the fracture angle is about
30 ◦ . For the tensile specimens T-2–T-4, the larger the fracture angle
is, the larger the fracture elongation is. In addition, it also can be
37
seen from Fig. 9 that the difference of the fracture elongations of
profiles extruded with h = 15, 25 and 35 mm is mainly attributed
to the difference of non-uniform plastic deformations occurring
in the process of necking, namely, ␧2 < ␧3 < ␧4 . In conclusion,
the depths of welding chambers have an obvious influence on the
mechanical properties of extruded profiles. With the increase of the
depth of welding chamber, the non-uniform plastic deformation
and the fracture angle of the tensile specimen increase.
In the ductile fracture process of tensile specimens, micro voids
easily originate at the positions where brittle substances exist such
as the second phase particles and impurities, or where the atomic
bonding is weak such as grain boundary, phase boundary and bond-
ing interface. With the increase of plastic deformation, these micro
voids gradually grow and propagate, and finally macro cracks are
formed and the specimen failure occurs. In this study, the fracture
locations, fracture angles and non-uniform plastic deformations
obtained from tensile tests directly reveal the atomic bonding
degrees of the bonding interfaces of L-seams. Overall, the bonding
interface of the L-seam of the profile extruded with h = 15 mm has
a weak atomic bonding, and the bonding interface of the L-seam of
the profile extruded with h = 35 mm has a strong atomic bonding.
While, the atomic bonding degree of the bonding interface of the
L-seam of the profile extruded with h = 25 mm is mediate.
Fig. 10 shows the facture morphologies of the tensile specimens
by means of SEM. Fig. 10(a) shows the morphologies of the unbond-
ing surface and the fracture surface of the specimen T-1. Some
equiaxial dimples with very small size and some elongated dimples
with relative big size are found in the fracture surface. Fig. 10(b)
shows the morphologies of the fracture surface of the T-2. There
are a large number of small and shallow equiaxial dimples in the
central region of the fracture surface and some elongated dimples
in the edge region adjacent to profile’s surfaces. Fig. 10(c) shows
the morphologies of the fracture surface of the T-3. There are also
many equiaxial dimples in the central region of the fracture surface
and some elongated dimples in the edge region adjacent to profile’s
38 J. Yu et al. / Journal of Materials Processing Technology 237 (2016) 31–47
Fig. 10. The fracture morphologies of the tensile specimens: (a) T-1, (b) T-2, (c) T-3, (d) T-4.
surfaces. However, the dimensions of the equiaxial dimples in the
central region have an obvious difference. The equiaxial dimples
in the zone 2, which is around the center of the fracture surface,
are small and shallow. While, the equiaxial dimples in the zone 1,
which is a bit far away from the center of the fracture surface, are
large and deep. The morphologies of the fracture surface of T-4 are
shown in Fig. 10(d). It is observed that there are many large and
deep quiaxial dimples in the center region of the fracture surface
and some elongated dimples in the edge region adjacent to pro-
file’s surfaces. As a result, it can be seen that the shape and size of
the dimples on the fracture surfaces of the tensile specimens are
different along the thickness direction of profiles.
The theoretical and experimental research of Li et al. (2011)
showed that the shape of the dimples on fractrue surface is deter-
mined mainly by the stress state. In the process of ductile fracture of
tensile specimens, the equiaxial dimple is formed due to the action
of normal stress, while the elongated dimple is formed owing to the
action of shearing stress, which were also demonstrated by Achouri
et al. (2013). In addtion, Guo et al. (2014) pointed out that the size
and depth of the dimples on fractrue surface are influenced by the
second phase particles, inclusions, grain boundaries, bonding inter-
faces, dislocation pileups and so on. Furthermore, Den Bakkera et al.
(2014) indicated that in the process of solid-state bonding, the size
and depth of the dimples on fractrue surfaces of weld seams reveal
the atomic bonding degrees of the bonding interfaces and corre-
spond to the ductility of extruded profiles. More specifically, large
and deep dimples represent strong atomic bonding and superior
ductility, while small and shallow dimples represent weak atomic
bonding and inferior ductility, which were also showed by Shen
et al. (2014).
Overall, the fracture surface of the tensile specimen T-1 has
equiaxial and elongated dimples, and the elongated dimples take
the dominant place. This is because that the shearing stress takes
the dominant place although the dimples were formed under the
combined action of the normal and shearing stress. The central
regions of the fracture surfaces of tensile specimens T-2–T-4 have
a large number of equiaxial dimples, but the edge regions of these
fracture surfaces have some elongated dimples. This is because that
micro voids firstly originate, grow, propagate and form micro cracks
in the central regions of tensile specimens under the action of ten-
sile stress, but with the propagation of the micro cracks, the edge
regions of tensile specimens are subjected to shearing stress. In
addition, since the failure of the tensile specimen T-2 occurs almost
along the L-seam of the profile extruded with h = 15 mm and the
fracture surface has lots of small and shallow dimples, it can be con-
cluded that the bonding quality of the central region of the L-seam
on the cross-section of profile extruded with h = 15 mm is poor.
While, the bonding quality of the central region of the L-seam on
the cross-section of profile extruded with h = 35 mm is good, since
the fracture of the tensile specimen T-4 penetrates through the L-
seam and the fracture surface has lots of large and deep dimples.
The fracture surface of tensile specimen T-3 has many small and
shallow dimples (in zone 2) similar to those in the central region of
L-seam of the profile extruded with h = 15 mm and many large and
deep dimples (in zone 1) similar to those in the central region of L-
seam of the profile extruded with h = 35 mm. Therefore, the central
region of the L-seam of the profile extruded with h = 25 mm has a
poor bonding quality, but the region adjacent to the central region
has a good bonding quality.
According to the aforementioned microstructure observation,
mechanical properties and fracture feature analyses, it can be con-
cluded that the dark bonding interfaces in Fig. 8 have poor bonding
quality and the white bonding interfaces have good bonding qual-
ity. The bonding interfaces of the L-seams of the profiles extruded
with h = 5, 15, 25 and 35 mm can be distinguished as the unbond-
ing surfaces, unsound and sound bonding zones. Fig. 11 gives the
dimension of each zone in detail. According to Cooper and Allwood
(2014a), the unsound bonding is formed because that the bonding
process does not establish close contact on the bonding interface
on the micrometer and millimeter scale, or that islands of con-
taminant (such as oxide) prevent full bonding of the interface at
the nano scale. As a result, the unsound bonding zones exhibit
dark boundaries in Fig. 8. In addition, although the sound bond-
ing zones have established close contact and have no contaminant,
their microstructure such as grain size may be different from the
other zones without L-seam. As a result, the sound bonding zones
exhibits white boundaries in Fig. 8.
4.2. Formation processes of L-seams and defects
4.2.1. Analysis of the accuracy of numerical simulation results
In order to reveal the formation processes of the L-seams and
their defects and obtain relevant physical quantities, 3-D transient
numerical simulations for the porthole die extrusion processes
 J. Yu et al. / Journal of Materials Processing Technology 237 (2016) 31–47
Fig. 11. The bonding quality of L-seams on the cross-sections of profiles extruded with different depths of the welding chambers: (a) h = 5 mm, (b) h = 15 mm, (c) h = 25 mm,
(d) h = 35 mm.
Fig. 12. Load-time curves obtained from numerical simulation and experiment with
h = 15 mm.
were conducted by using Deform-3D software. In order to verify
the reliability of the simulation results, the extrusion loads and
profile fronts obtained from extrusion experiments and numerical
simulations were compared accordingly.
Fig. 12 compares the load-time curves obtained from numeri-
cal simulation and experiment with the depth of welding chamber
h = 15 mm. Their maximum extrusion loads are 7.60 and 7.73 MN,
respectively. It is clear that the predicted maximum extrusion load
is in good agreement with the experimental one.
Fig. 13 shows the comparison of the profile fronts obtained
from experiments and simulations. Although the effect of the weld-
ing quality of the areas just behind the fronts on the profile front
geometries was not taken into consideration in numerical sim-
ulations, it can be seen that the sizes and shapes of extruded
profile fronts are in good agreement with the experimental results.
According to the aforementioned comparisons between the exper-
imental and simulation results, it can be concluded that the
numerical simulation models and results are reliable.
4.2.2. Formation processes of L-seams and the macro hole
Extrusion experimental results have shown that the L-seams of
the profiles extruded with h = 15, 25 and 35 mm have no macro
39
defects. However, there is an elliptic macro hole in the central
zone of cross-section of the profile extruded with h = 5 mm. Fig. 14
gives the formation process of the L-seam of the profile extruded
with h = 15 mm. In the porthole die extrusion process, when heated
metal flows into upper die cavity, it is firstly divided into two sepa-
rated metal streams by bridge. After that, the two separated metal
streams gradually flow toward the bottom of welding chamber.
When the metal streams have reached the bottom of welding cham-
ber, they flow sideways to fill welding chamber and a few of them
is pressed into the bearing of lower die to form profile fronts, as
shown in Fig. 14(a)–(d). This process can be called the dividing and
filling stage. With the extrusion proceeding, since the metal close
to the central axis of welding chamber has the fastest flow veloc-
ity, the two separated metal streams firstly contact to each other in
the central area of welding chamber and start to solid-state bond
to form welding seam under the conditions of high pressure and
temperature, as shown in Fig. 14(e)–(h). This process can be called
the contacting and partially welding stage. After metal fills welding
chamber completely and flows out of the bearing of the lower die to
form profile, the extrusion process enters into the steady welding
and forming stage (Fig. 14(i)–(l)), and a welding plane with stable
shape and size is formed in welding chamber and lower die bear-
ing, as shown in Fig. 14(l). From Fig. 14(k) and (l), it can be seen that
there is no any macro defect in the welding plane and in the L-seam
of extruded profile. In addition, the formation processes of the L-
seams of the profiles extruded with h = 25 and 35 mm are similar
to that extruded with h = 15 mm, and there are also no any macro
defect in welding planes and in L-seams of extruded profiles. There-
fore, it can be concluded that the welding quality of the L-seams of
the profiles extruded with h = 15, 25 and 35 mm depends on the
solid-state bonding behavior in welding planes.
Fig. 15 gives the formation process of the elliptic macro hole
in the central zone of cross-section of the profile extruded with
h = 5 mm. In the extrusion process with h = 5 mm, the welding pres-
sure is quite low due to the small height of welding chamber. As a
result, in the contacting and partially welding stage, the split metal
streams can not contact to each other in the central area of the
welding chamber and a hole is formed, as shown in Fig. 15(b) and
(c). With the extrusion proceeding, the hole gradually shrinks but
the flow resistance in the radial direction increases. When the flow
resistance in the radial direction is greater than that in lower die
bearing, the metal flows out of the bearing and forms profile. After
that, the shape and size of the hole are stable and the hole appears
in the extruded profile, as shown in Fig. 15(e) and (f). This hole
always exists in the steady welding and forming stage, and thus
it penetrates through the whole profile along length direction of
the extruded profile. It can be concluded that the macro hole in the
profile extruded with h = 5 mm is caused by metal flow behavior,
and does not relate to solid-state bonding process.
In porthole die extrusion process, the macro defect of L-seams
is easy to be obversed and can be directly predicted by means of
numerical simulation. However, the bonding quality of the L-seams
without any macro defect is difficult to be directly determined by
means of numerical simulation. Since the bonding quality of the
40 J. Yu et al. / Journal of Materials Processing Technology 237 (2016) 31–47

Fig. 13. Comparison of profile fronts obtained from the experiments and simulations with different depths of welding chambers: (a) h = 5 mm, (b) h = 15 mm, (c) h = 25 mm,
(d) h = 35 mm.

Fig. 14. Three stages of the formation process of an L-seam during porthole die extrusion: (a)–(d) dividing and filling stage, (e)–(h) contacting and partially welding stage,
(i)–(l) steady welding and forming stage.
 J. Yu et al. / Journal of Materials Processing Technology 237 (2016) 31–47
Fig. 15. Formation process of the macro hole of the profile extruded with h = 5 mm: (a)–(c) contacting and partially welding stage, (d)–(f) steady welding and forming stage.
L-seams without any macro defect is determined by the solid-state
bonding behavior in welding planes, investigating the bonding
mechanism and establishing bonding criterion become particularly
important. In the following sections, the extrusion processes of the
profiles extruded with h = 15, 25 and 35 mm are used to establish a
new bonding criterion.
4.3. Evaluation of solid-state bonding quality
4.3.1. Metal flow behavior and welding paths
In steady welding and forming stage, the separated metal
streams always encounter in the welding plane shown in Fig. 14(l)
and form an L-seam inside profile by means of solid-state bonding.
Therefore, for the profiles extruded with h = 15, 25 and 35 mm, the
flow behavior and the solid-state bonding process in the welding
planes directly determine the bonding quality of the profiles.
The extrusion experimental results in this study have shown
that the bonding quality of the L-seams of profiles extruded with
h = 15, 25 and 35 mm are variable along the thickness direction of
profiles. In order to investigate the solid-state bonding processes
of some special particles in each L-seam, five tracing points named
as P1 , P2 , P3 , P4 and P5 were set up in each L-seam at the position
of the die exit, as shown in Fig. 16(b). The distance between each
tracing point and the center of the profile’s cross-section is 0, 0.75,
1.25, 3.0 and 5.0 mm, respectively. After that, based on the particle
tracing algorithm of Deform-3D software, the position change of
each tracing point during extrusion is obtained. According to the
coordinate values of each tracing point, the metal particle flow lines
were determined, as shown as f1 –f5 in Fig. 16(c) and (d). In addition,
the initial welding positions of five tracing points, named as I1 , I2 , I3 ,
I4 and I5 , where metal just flows into the welding plane and starts
41
solid-state bonding, are also determined. In this way, in the welding
plane, the welding path of each tracing point is finally determined,
as shown in Fig. 16(f). By the same way, the metal particle flow
lines, initial welding positions and the welding paths in the steady
welding and forming stages for the extrusion processes with h = 25
and 35 mm can also be determined.
The velocity distributions in Fig. 16(e) and (f) show that there
is a dead metal zone in the welding chamber and welding plane.
According to the flow lines f1 –f5 shown in Fig. 16(c) and (d) and
the velocity distributions shown in Fig. 16(e) and (f), it can be seen
that, in the steady welding and forming stage, most of the metal in
the welding chamber bypasses the dead metal zone and flows into
the welding plane from initial welding positions with a relatively
high velocity. Then, the bypassed metal starts to solid-state bond
in the welding plane and gradually flows into the die bearing to
form the L-seam. Since the metal in the dead zone has a very low
velocity, it hardly flows into the bearing to form the L-seam. As a
result, the L-seam of profile is formed almost by the metal in the
main welding zone of the welding plane, as shown in Fig. 16(f).
According to the welding paths of tracing points, the changes of
physical qualities during the solid-state bonding can be obtained.
Fig. 17 gives the changes of the hydrostatic pressure, effective
stress, effective strain rate and temperature along the welding
paths of tracing points P1 –P5 in the solid-state bonding process for
the extrusion die with h = 15 mm. The solid-state bonding time for
tracing points P1 –P5 are 1.71, 4.73, 6.93, 13.99 and 14.19 s, respec-
tively. During the solid-state bonding, no matter which welding
path the metal flows along, with the increase of the bonding
time, the hydrostatic pressures overall show downward trends
and finally decrease to 0 Mpa (Fig. 17(a)), and both the effective
stresses and effective strain rates increase firstly and then decrease
42 J. Yu et al. / Journal of Materials Processing Technology 237 (2016) 31–47

Fig. 16. Related information in steady welding and forming stage for the extrusion process with h = 15 mm: (a) a quarter of geometric model, (b) five tracing points, (c) flow
lines in Y–Z plane, (d) flow lines in X–Y plane, (e) 3-D velocity distribution, (f) velocity distribution, initial welding positions and welding paths in welding plane.

Fig. 17. The changes of physical quantities along welding paths of tracing points P1 –P5 in the solid-state bonding process for the extrusion die with h = 15 mm: (a) hydrostatic
pressure, (b) effective stress, (c) effective strain rate, (d) temperature.

(Fig. 17(b) and (c)), and the temperatures increase slightly (Fig. 17
(d)). In addition, compared to the metal flowing along the welding
path of tracing points P5 , the metal flowing along the welding path
of tracing points P1 is subjected to a lower hydrostatic pressure,
larger effective stress and effective strain rate, and lower tempera-
ture. As a result, the bonding quality in the center of the L-seam
of profile extruded with h = 15 mm is worse, while the bonding
quality in the region far away from the center is good. In the solid-
state bonding processes for extrusion dies with h = 25 and 35 mm,
the change rules of the hydrostatic pressure, effective stress, effec-
tive strain rate and temperature along their corresponding welding
paths of tracing points P1 –P5 are similar to those for the extrusion
die with h = 15 mm. Consequently, the changes of physical quali-
ties obtained from the numerical simulation results qualitatively
explain the differences of the bonding quality of L-seams of profile
extruded with h = 15, 25 and 35 mm along the thickness direction.
 J. Yu et al. / Journal of Materials Processing Technology 237 (2016) 31–47
4.3.2. Extrusion bonding mechanism and a new bonding criterion
Inspired by the above research results, the authors of this paper
proposed a solid-state bonding mechanism of the L-seam in the
process of porthole die extrusion, as shown in Fig. 18. When sep-
arated metal streams just flow into the welding plane and contact
to each other, on a microscopic scale, the surfaces for bonding
are irregular and have many asperities. Therefore, there are many
micro voids on the contact interface, as shown in Fig. 18(a). With
the solid-state bonding proceeding along welding paths, the micro
voids gradually close to create a micro-level intimate contact under
the actions of plastic deformation and atomic diffusion, as shown in
Fig. 18(b). According to the report of Cooper and Allwood (2014a),
the inter-atomic and van der Waals forces are the major sources
of attraction between the atoms, and the inter-atomic force is
far greater than the van der Waals force. When intimate con-
tact is achieved, namely, the distances (d) of the atoms on the
contact interface are reduced to the 10 atomic spacings, the inter-
atomic attractive forces will form a joint, which means that a sound
solid-state bonding for L-seam of profile is achieved, as shown in
Fig. 18(c).
In order to predict solid-state bonding quality of L-seams in port-
hole die extrusion process, investigating void closure mechanism
and developing mathematical model of void closure behavior are of
great importance. Main void closure mechanisms include the plas-
tic deformation mechanism and the diffusion mechanism. In the
porthole die extrusion process, the plastic deformation mechanism
of the void closure can refer to that in other plastic deformation
processes such as forging and rolling. With the effect of the plas-
tic deformation, the voids gradually shrink and become flatten, as
shown in Fig. 18(1). Nakasaki et al. (2006) showed that the ratio of
mean normal stress to effective stress (stress triaxiality) and effec-
tive strain in the region surrounding the void are the dominant
factors for the void closure behavior, which can be described by
using following hydrostatic integration Gm ,
 ε̄  
m
Gm = dε̄
0
¯
where,  m is the mean normal stress (Mpa), ¯ is the effective stress
(Mpa), and ε̄ is the effective strain.
The diffusion mechanism of the void closure can refer to that
in the solid-state diffusion bonding process. Derby and Wallach
(1982) demonstrated that the diffusion mechanism of the void
closure mainly includes the surface source diffusion and the inter-
face source diffusion. The surface source diffusion means that the
atoms on or near the surface of void are transferred from the parts
with larger curvatures to those with smaller curvatures under the
driving of the difference of surface free energy by means of vol-
ume diffusion (path 2 in Fig. 18(2)) and surface diffusion (path
3 in Fig. 18(2)). While, the interface source diffusion means that
the atoms on or near the contact interface are transferred to the
void necks under the driving of the chemical potential gradient by
means of volume diffusion (path 4 in Fig. 18(3)) and surface diffu-
sion (path 5 in Fig. 18(3)). Furthermore, Ma et al. (2012) considered
that the surface source diffusion only results in the distribution
change of the atoms on the void surface, and not results in the
change of the void volume. Therefore, when the void closure occurs
under the action of the diffusion mechanism, the volume of void
is directly determined by the interface source diffusion. In order
to quantitatively predict the volume evolution of void, some geo-
metric models are proposed to describe the shape of void and
some mathematical models of void closure were established. For
example, Hill and Wallach (1989) assumed that the initial shape of
void is an ellipse, while Orhan et al. (1999) assumed that is a sine
wave, and the models for void closure in the diffusion bonding pro-
cess were derived respectively. However, since these mathematical
43
models are complex and the assumed shape of void directly deter-
mines the prediction precision, they are difficultly used to guide
practical process. Since the temperature, time and material have
obvious influences on the atomic diffusion, which were reported
by Cooper and Allwood (2014b), in order to describe the influence
of the atomic diffusion on the void closure behavior in porthole
die extrusion process, a diffusion integration GD was proposed as
follows,
 t  RT 
GD = k0 exp dt (3)
0
QD
where, t is the time (s), R is the universal gas constant
(8.314 J mol−1 K−1 ), T is the absolute temperature (K), QD is the dif-
fusion activation energy (J mol−1 ), and k0 is a coefficient related to
material and the surface conditions of metal for bonding.
It should be pointed out that, in the Eq. (3), the effect of temper-
ature on the void closure behavior is described as the exp(RT/QD )
instead of the Arrhenius type relationship exp(-QD /RT) used in dif-
fusion equations. The reason of this modification is as follows. In
porthole die extrusion process, usually, the ratio of mean normal
stress to effective stress, stress triaxiality  m /,
¯ is about 1.0–10.0.
On the one hand, if the product of  m /¯ and exp(-QD /RT) is used to
describe the effect of stress triaxiality and temperature on the void
closure behavior, since the orders of magnitude of the values of
 m /¯ and exp(-QD /RT) are considerably different, the effect of tem-
perature on the void closure behavior almost can not be reflected.
On the other hand, if the values of exp(-QD /RT) are modified as
the same order of magnitude of the values of  m /, ¯ a small tem-
perature difference will result in a large difference of the values
of exp(-QD /RT), which means that the exp(-QD /RT) gives an over
emphasis on the effect of temperature on the void closure behav-
ior. As a contrast, the values of exp(RT/QD ) have the same order of
magnitude of the values of  m /¯ and can properly reflect the effect
of temperature on the void closure behavior. Therefore, this study
used exp(RT/QD ) in the Eq. (3) to describe the effect of temperature
(2)
on the void closure behavior in porthole die extrusion process.
This study considers that in the whole process of void clo-
sure and sound bonding formation, both plastic deformation and
atomic diffusion are indispensable two steps. The dominant fac-
tors affecting void closure and sound bonding are temperature,
stress, strain, strain rate, time and material. Firstly, temperature
not only influences the flow stress of material but also affects
the atomic diffusion. Secondly, stress state influences the plas-
tic deformation behavior, while the stress gradient is a driving
force for atomic diffusion. Thirdly, strain and strain rate not only
influence the degree of contaminant broking and virgin material
exposing but also reveal the dislocation density surrounding voids.
The dislocation density influences the atomic diffusion. Strain can
be expressed as an integral of strain rate on time. Fourthly, time
influences the degree of atomic diffusion and plastic deformation.
Finally, the thermo-physical properties of material and the sur-
face conditions of metal for bonding also influence void closure
behavior and atomic diffusion. The surface conditions of metal for
bonding include the roughness, oxide and contaminant films. Dur-
ing porthole dies extrusion, sometimes a certain amount of oxide
layers or contaminant films exist on the surfaces of the split metal
streams due to the formation of gas pockets behind the die bridge.
As a result, in this study, a new dimensionless solid-state bonding
criterion named as the J criterion was proposed as follows,
 t  RT 
m ˙
J= k0 ε̄ exp dt (4)
¯ QD
0
where, ε̄˙ is the effective strain rate (s−1 ).
44 J. Yu et al. / Journal of Materials Processing Technology 237 (2016) 31–47

Fig. 18. Schematic diagram of solid-state bonding mechanism in porthole die extrusion process.

 t
4.3.3. Experimental validation and evaluation of the new bonding p
criterion Qad = dt (6)
0
¯
In order to validate the accuracy of the J criterion for predicting
solid-state bonding quality in porthole die extrusion process, the
where,  is the metal flow velocity (mm s−1 ), p is the contact pres-
values of J, which are integrals along the welding paths of tracing
sure (Mpa), i.e. the normal contact stress ( n ) on the welding plane.
points mentioned previously, were calculated. Since the L-seams
In order to comprehensively evaluate the action of p in Eqs. (5)
in this study are formed by the same material and the alloy ele-
and (6), this paper tried to use two definitions of p. One is the nor-
ments such as Mg and Si in AA6063 are very few, thus the void
mal contact stress ( n ) on the welding plane originally defined by
closure behavior with the action of the atomic diffusion is mainly
the K and Q criteria, another is the mean normal stress ( m ). The
due to the self-diffusion behavior of aluminum atoms. According to
corresponding Kad is respectively expressed as Kad  n and Kad  m ,
the report of Sohn (2004), the activation energy of self-diffusion of
and the corresponding Qad is respectively expressed as Qad  n and
aluminum atoms is 1.42 × 105 J mol−1 at the temperature range of
Qad  m .
450–650 ◦ C (QD = 1.42 × 105 J mol−1 ). Since the extrusion parame-
The above mentioned microstructure observation, mechanical
ters are the same for the extrusion experiments with h = 15, 25 and
properties and fracture feature analyses have shown that the L-
35 mm and there are no gas pockets behind the die bridges during
seams of profiles extruded with h = 15 and 25 mm respectively have
extrusion, thus the surface conditions of metal for bonding should
an unsound bonding zone with the length of 2.5 and 1.5 mm in their
be the same. Therefore, for simplicity, the value of k0 is assumed
central zones, and the L-seam of profile extruded with h = 35 mm
as 1.0. In addition, for comparison, the values of the Kad and the
has no unsound bonding zone. The distance from the centers to
Qad for each tracing point were also respectively calculated accord-
the bonding quality transition positions for L-seams of profiles
ing to the K criterion proposed by Donati and Tomesani (2004) and
extruded with h = 15 and 25 mm are 1.25 and 0.75 mm, respectively.
the Q criterion proposed by Plata and Piwnik (2000). The K and Q
Fig. 19(a) represents the bonding quality of five tracing points
criteria are expressed as follow,
on L-seams summarized according to the experimental results,
where 䊉 indicates sound bonding,  indicates unsound bonding,
 t and  indicates the transition point from unsound to sound bond-
p
Kad = dt (5) ing, i.e. sound bonding critical point. The dotted rectangle indicates
0
¯ unsound bonding area. For h = 15 mm, P1 and P2 are the unsound
 J. Yu et al. / Journal of Materials Processing Technology 237 (2016) 31–47 45

Fig. 19. The values of Kad , Qad and J calculated according to the welding paths of the tracing points P1 –P5 for the L-seams of profiles extruded with h = 15, 25 and 35 mm: (a)
bonding quality of the tracing points P1 –P5 on the L-seams, (b) Kad  n , (c) Qad  n , (d) Kad  m , (e) Qad  m , (f) J.

bonding points, P3 is the transition point (corresponding to T1 in the
above Section 4.1.1), and P4 and P5 are the sound bonding points.
For h = 25 mm, P1 is the unsound bonding point, P2 is the transition
point (corresponding to T3 in the above Section 4.1.1), and P3 –P5
are the sound bonding points. For h = 35 mm, P1 –P5 are all the sound
bonding points.
Therefore, according to the experimental results, the values of
the Kad , Qad and J for five tracing points should satisfy the following
conditions.
(1) The values of the Kad , Qad and J for P1 and P2 with h = 15 mm
and P1 with h = 25 mm should be respectively less than the cor-
responding values of the Kad , Qad and J for P1 with h = 35 mm.
46 J. Yu et al. / Journal of Materials Processing Technology 237 (2016) 31–47
(2) The values of the Kad , Qad and J for P1 and P2 with h = 15 mm
should be respectively less than the corresponding values of
the Kad , Qad and J for P3 with h = 15 mm.
(3) The values of the Kad , Qad and J for P1 with h = 25 mm should be
respectively less than the corresponding values of the Kad , Qad
and J for P2 with h = 25 mm.
(4) The values of the Kad , Qad and J for P3 with h = 15 mm should be
respectively equal to the corresponding values of the Kad , Qad
and J for P2 with h = 25 mm.
For the K and Q criteria, when p in Eqs. (5) and (6) is the nor-
mal contact stress ( n ), the values of the Kad  n , Qad  n calculated
according to the welding paths of the tracing points P1 –P5 for the
L-seams of profiles extruded with h = 15, 25 and 35 mm are respec-
tively shown in Fig. 19(b) and (c). It can be seen from Fig. 19(b)
that the value of the Kad  n for P1 with h = 25 mm is larger than that
for P2 with h = 25 mm. Obviously, the values of Kad  n in Fig. 19(b)
do not satisfy the condition (3). In addition, the value of Kad  n for
P3 with h = 15 mm is equal to 57.00 mm, and the value of Kad  n for
P2 with h = 25 mm is equal to 52.50 mm. The relative error of these
two values is 7.89%. It can be seen from Fig. 19(c) that the value
of Qad  n for P3 with h = 15 mm is equal to 16.18 s, and the value of
Qad  n for P2 with h = 25 mm is equal to 17.68 s. The relative error of
these two values is 8.48%.
For the K and Q criteria, when p in Eqs. (5) and (6) is the
mean normal stress ( m ), the values of the Kad  m , Qad  m calculated
according to the welding paths of the tracing points P1 -P5 for the
L-seams of profiles extruded with h = 15, 25 and 35 mm are respec-
tively shown in Fig. 19(d) and (e). It can be seen from Fig. 19(d) that
the value of the Kad  m for P1 with h = 25 mm is also larger than that
for P2 with h = 25 mm. Obviously, the values of Kad  m in Fig. 19(d)
also do not satisfy the condition (3). In addition, the value of Kad  m
for P3 with h = 15 mm is equal to 45.92 mm, and the value of Kad  m
for P2 with h = 25 mm is equal to 43.64 mm. The relative error of
these two values is 4.97%. It can be seen from Fig. 19(e) that the
value of Qad  m for P3 with h = 15 mm is equal to 14.20 s, and the
value of Qad  m for P2 with h = 25 mm is equal to 15.71 s. The relative
error of these two values is 9.61%.
Fig. 19(f) gives the values of the J criterion calculated according
to the welding paths of the tracing points P1 -P5 for the L-seams
of profiles extruded with h = 15, 25 and 35 mm. The value of J for
P3 with h = 15 mm is equal to 9.65, and the value of J for P2 with
h = 25 mm is equal to 9.63. The relative error of these two values is
0.21%. Obviously, in comparison with the K and Q criteria calculated
whether by using the normal contact stress ( n ) on the welding
plane or the mean normal stress ( m ), the J criterion proposed in
this study has the highest prediction accuracy.
It should be pointed out that the points above the critical lines in
Fig. 19(b)–(f) only indicate that the corresponding zones of bonding
interfaces have sound bonding, and the magnitudes of values of
these points do not represent the difference of bonding quality in
these zones. Therefore, the magnitudes of values of the points above
the critical lines can not be used to evaluate the accuracy of the
bonding criteria.
In this work, the welding process of a plate-shaped profile in
porthole die extrusion process was studied in detail. Generally
speaking, different shaped profiles may have different metal flow
behaviors, L-seam formation processes, welding defects, bonding
quality and microstructure. However, the formation mechanisms of
L-seams and welding defects, microstructure evolution in welding
area, and welding paths in main welding zone obtained in this work
will provide a useful guideline or reference for other shaped pro-
files. Especially, the J criterion proposed in this work finally comes
down to a function of mean normal stress, effective stress, effective
strain rate, temperature and time in solid-state bonding process.
Therefore, the J criterion can be used to evaluate the bonding quality
of different shaped profiles.
5. Conclusions
In this study, a set of modular porthole extrusion dies with
different depths of welding chambers were designed and manu-
factured. The profiles extruded with different depths of welding
chambers were obtained by performing extrusion experiments.The
welding quality of these extruded profiles was characterized by
means of microstructure observation, tensile test and fracture anal-
ysis. True stress-strain curves of homogenized 6063 aluminum
alloy with various strain rates and deformation temperatures were
obtained by means of isothermal hot compression test. Three-
dimensional transient numerical simulation models of porthole
die extrusion processes were established. Flow behaviors of alu-
minum alloy in porthole die extrusion process were investigated,
and the formation processes of L-seams and their defects were
revealed. Solid-state welding processes of metal particles were
traced and welding paths were determined. Finally, based on the
plastic deformation and diffusion mechanisms for closure behavior
of micro voids on contact interfaces, a new dimensionless solid-
state bonding criterion related to stress triaxiality, effective strain
rate, temperature and contact time was proposed. According to this
new bonding criterion, the bonding quality of L-seams obtained
by means of extrusion experiments was successfully predicted.
According this study, the following conclusions were drawn:
(1) The welding quality of an L-seam on the cross-section of the
profile extruded by the modular die with an arched and pointed
bridge is variable along the thickness direction of profile and
the welding defect tends to be formed in the central zone of
the profile’s cross-section. With the increase of the depth of
the welding chamber, the macro hole in the central zone of the
profile’s cross-section disappears and the length of the unsound
bonding zone decreases.
(2) The welding quality of L-seam of profile extruded by porthole
die is determined by two key factors: metal flow behavior and
solid-state bonding process. The metal flow behavior deter-
mines whether the macro welding defect is formed and the
solid-state bonding process determines the bonding quality of
the bonding interface. The formation of a macro hole in the pro-
file extruded with a shallow welding chamber is attributed to
metal flow behavior, and does not relate to solid-state bonding
process.
(3) There is a dead metal zone in the welding plane and the metal
in the dead metal zone almost does not participate the solid-
state bonding. The L-seam of profile without any macro welding
defect is formed almost by the metal in the main welding zone
of the welding plane through solid-state bonding.
Acknowledgements
The authors would like to acknowledge the financial support
from National Natural Science Foundation of China (51375270
and 51405268), Encouragement Foundation for Young Scholars of
Shandong Province (BS2014ZZ001).
References
Achouri, M., Germain, G., Dal Santo, P., Saidane, D., 2013. Experimental
characterization and numerical modeling of micromechanical damage under
different stress states. Mater. Des. 50, 207–222.
Akeret, R., 1972. Properties of pressure welds in extruded aluminium alloy
sections. J. Inst. Met. 10, 202.
Bariani, P.F., Bruschi, S., Ghiotti, A., 2006. Physical simulation of longitudinal
welding in porthole-die extrusion. CIRP Ann. Manuf. Technol. 55, 287–290.
 J. Yu et al. / Journal of Materials Processing Technology 237 (2016) 31–47
Ceretti, E., Fratini, L., Gagliardi, F., Giardini, C., 2009. A new approach to study
material bonding in extrusion porthole dies. CIRP Ann. Manuf. Technol. 58,
259–262.
Chen, H., Zhao, G.Q., Zhang, C.S., Guan, Y.J., Liu, H., Kou, F.J., 2011. Numerical
simulation of extrusion process and die structure optimization for a complex
aluminum multicavity wallboard of high-speed train. Mater. Manuf. Process.
26, 1530–1538.
Cooper, D.R., Allwood, J.M., 2014a. The influence of deformation conditions in
solid-state aluminium welding processes on the resulting weld strength. J.
Mater. Process. Technol. 214, 2576–2592.
Cooper, D.R., Allwood, J.M., 2014b. Influence of diffusion mechanisms in
aluminium solid-state welding processes. Procedia Eng. 81, 2147–2152.
Den Bakkera, A.J., Werkhovenb, R.J., Sillekensb, W.H., Katgermanc, L., 2014. The
origin of weld seam defects related to metal flow in the hot extrusion of
aluminium alloys EN AW-6060 and EN AW-6082. J. Mater. Process. Technol.
214, 2349–2358.
Derby, B., Wallach, E.R., 1982. Theoretical model for diffusion bonding. Metal Sci.
16, 49–56.
Donati, L., Tomesani, L., 2004. The prediction of seam welds quality in aluminum
extrusion. J. Mater. Process. Technol. 153, 366–373.
Donati, L., Tomesani, L., 2005. The effect of die design on the production and seam
weld quality of extruded aluminum profiles. J. Mater. Process. Technol. 164,
1025–1031.
Edwards, S.P., den Bakker, A.J., Zhou, J., Katgerman, L., 2009. Physical simulation of
longitudinal weld seam formation during extrusion to produce hollow
aluminum profiles. Mater. Manuf. Processes 24, 409–421.
Fang, G., Zhou, J., Duszczyk, J., 2009. Extrusion of 7075 aluminium alloy through
double-pocket dies to manufacture a complex profile. J. Mater. Process.
Technol. 209, 3050–3059.
Flitta, I., Sheppard, T., 2003. Nature of friction in extrusion process and its effect on
material flow. Mater. Sci. Technol. 19, 837–846.
Güley, V., Güzel, A., Jäger, A., Ben Khalifa, N., Tekkaya, A.E., Misiolek, W.Z., 2013.
Effect of die design on the welding quality during solid state recycling of
AA6060 chips by hot extrusion. Mater. Sci. Eng. A 574, 163–175.
Guo, J.F., Chen, H.C., Sun, C.N., Bi, G.J., Sun, Z., Wei, J., 2014. Friction stir welding of
dissimilar materials between AA6061 and AA7075 Al alloys effects of process
parameters. Mater. Des. 56, 185–192.
Hill, A., Wallach, E.R., 1989. Modelling solid-state diffusion bonding. Acta Metall.
37, 2425–2437.
47
Kim, K.J., Lee, C.H., Yang, D.Y., 2002. Investigation into the improvement of welding
strength in three-dimensional extrusion of tubes using porthole dies. J. Mater.
Process. Technol. 130, 426–431.
Li, H., Fu, M.W., Lu, J., Yang, H., 2011. Ductile fracture: experiments and
computations. Int. J. Plast. 27, 147–180.
Liu, G., Zhou, J., Duszczyk, J., 2008. FE analysis of metal flow and weld seam
formation in a porthole die during the extrusion of a magnesium alloy into a
square tube and the effect of ram speed on weld strength. J. Mater. Process.
Technol. 200, 185–198.
Ma, R.F., Li, M.Q., Li, H., Yu, W.X., 2012. Modeling of void closure in diffusion
bonding process based on dynamic conditions. Sci. China Technol. Sci. 55,
2420–2431.
Nakasaki, M., Takasu, I., Utsunomiya, H., 2006. Application of hydrostatic
integration parameter for free-forging and rolling. J. Mater. Process. Technol.
177, 521–524.
Orhan, N., Aksoy, M., Eroglu, M., 1999. A new model for diffusion bonding and its
application to duplex alloys. Mater. Sci. Eng. A 271, 458–468.
Plata, M., Piwnik, J., 2000. Theoretical and experimental analysis of seam weld
formation in hot extrusion of aluminum alloys, In: Seventh International
Aluminum Extrusion Technology Seminar. Omnipress, Chicago, IL, USA, pp.
205–211.
Shen, Z.K., Yang, X.Q., Yang, S., Zhang, Z.H., Yin, Y.H., 2014. Microstructure and
mechanical properties of friction spot welded 6061-T4 aluminum alloy. Mater.
Des. 54, 766–778.
Sohn, Y., 2004. 13—diffusion in metals. In: Totemeier, W.F.G.C. (Ed.), Smithells
Metals Reference Book. , eighth edition. Butterworth-Heinemann, Oxford, pp.
1–120.
Tang, D., Zhang, Q.Q., Li, D.Y., Peng, Y.H., 2014. A physical simulation of longitudinal
seam welding in micro channel tube extrusion. J. Mater. Process. Technol. 214,
2777–2783.
Valberg, H., Loeken, T., Hval, M., Nyhus, B., Thaulow, C., 1995. The extrusion of
hollow profiles with a gas pocket behind the bridge. Int. J. Mater. Prod.
Technol. 10, 222–267.
Yu, J.Q., Zhao, G.Q., Chen, L., 2016. Investigation of interface evolution,
microstructure and mechanical properties of solid-state bonding seams in hot
extrusion process of aluminum alloy profiles. J. Mater. Process. Technol. 230,
153–166.
Zhou, J., Li, L., Duszczyk, J., 2003. 3D FEM simulation of the whole cycle of
aluminium extrusion throughout the transient state and the steady state using
the updated Lagrangian approach. J. Mater. Process. Technol. 134, 383–397.