Procedia CIRP 18 ( 2014 ) 3 – 8
Available online at www.sciencedirect.com
2212-8271 © 2014 Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/3.0/).
Peer-review under responsibility of the International Scientifi c Committee of the “International Conference on Manufacture
of Lightweight Components – ManuLight 2014”
doi: 10.1016/j.procir.2014.06.098
ScienceDirect
International Conference on Manufacture of Lightweight Components – ManuLight2014
Extrusion of aluminum tubes with axially graded wall thickness and
mechanical characterization
M. Negendank
1*
, S. Müller
1
, W. Reimers
2
1 Extrusion Research and Development Center, TU Berlin, Germany
2 Chair Metallic Materials, TU Berlin, Germany
* Corresponding author. Tel.: +49 (0) 30 314 725 16; fax: +49 (0) 30 314 725 03; E-mail address: [email protected]
Abstract
In this study the indirect extrusion of seamless aluminum tubes with variable wall thickness was investigated. Therefore, an axially
moveable stepped mandrel was applied. Investigations revealed that wall thickness transitions can either be graded over the tube
length or very sharp. The microstructures in thin-walled and thick-walled tube sections were investigated. The local variation of the
extrusion ratio and with that the tube wall thickness, product velocity and product temperature during the process lead to
significantly different local microstructures at T
B
=400°C. At T
B
=500°C the microstructure was homogeneously recrystallized with
similar grain size in all different tube sections. Furthermore, the mechanical tube properties were characterized by three point
bending tests.
© 2014 The Authors. Published by Elsevier B.V.
Selection and peer-review under responsibility of the International Scientific Committee of the “International Conference on
Manufacture of Lightweight Components – ManuLight2014” in the person of the Conference Chair Prof. A. Erman Tekkaya.
Keywords: Extrusion, variable wall thickness, tailored tubes, aluminum, microstructure, three point bending test
1. Introduction
Modern lightweight constructions more and more
demand for customized solutions in order to meet
customers’ requirements of increasing complexity [1].
Therefore, in the recent history approaches towards a
flexibilization of extrusion processes and its products
were investigated.
The manufacturing of bend profiles during hot metal
extrusion can avoid post extrusion bending. On the one
hand in [2] three-dimensional bend profiles were
extruded by regulating the material flow in the die. On
the other hand a guiding device behind the extrusion
press was applied to create in-line three-dimensional
bend extrusions [3].
Since the distribution of loads usually is not constant
over the profile’s length, the development of extruded
profiles with locally adapted cross sections would be
desirable. Jäger et al. [4] showed the feasibility of local
cross section reductions of hollow profiles by applying
in-line electromagnetic compression subsequent to
extrusion.
Further, it would be attractive for innovative
lightweight constructions to adapt the wall thickness of
extruded hollow profiles according to the actual load.
For this reason laboratory-scale extrusion trials using an
axially movable conic mandrel and lead as billet material
were used in order to show the feasibility of wall
thickness variations by changing the inner tube diameter
during extrusion [5]. Negendank et al. [6] used a stepped
mandrel and a small industrial size 8MN extrusion press
to show the feasibility of producing multiple wall
thickness variations along tube direction for AA6060.
Another possibility of changing the wall thickness of
extruded hollow profiles is to vary the outer diameter by
using movable die bearings [7]. In [8] both approaches
(moving mandrel as well as movable die segments) were
combined and hollow profiles with variable inner and
outer tube diameter were extruded.
Aim of the current study was to extrude AA6060
tubes with axially variable cross sections and to analyze
the product geometry as well as the local microstructure
© 2014 Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/3.0/).
Peer-review under responsibility of the International Scientifi c Committee of the “International Conference on Manufacture
of Lightweight Components – ManuLight 2014”
4 M. Negendank et al. / Procedia CIRP 18 ( 2014 ) 3 – 8
in different tube areas. Furthermore, the mechanical
properties of tube sections with different wall thickness
should be investigated by three point bending tests.
2. Experimental Procedure
2.1. Extrusion of seamless aluminum tubes with variable
wall thickness
For the extrusion experiments conducted in this study
the indirect extrusion process was used. The diameters
of the die and the container were D
die
=41mm and
D
C
=125mm. A thermocouple in the die bearing was
applied to measure the profile temperature. The die was
fixed on the hollow stem and therefore was heated
indirectly by the container’s heat prior to the trials. The
variation of the wall thickness (t) of seamless aluminum
tubes during the extrusion process was achieved by an
axially moveable stepped mandrel (Fig. 1). The mandrel
featured a tip diameter of 32mm with a shaft diameter of
37mm. Such as the die, the mandrel was also heated
indirectly by the container’s heat. Thus, the temperatures
of die and mandrel were 50-100°C lower than the set
container temperature and the billet temperature T
B
. In
order to change the tube cross section the mandrel was
first moved in extrusion direction. This procedure is
based on the well-known process of extrusion with
moving mandrel. Once the mandrel step arrives in the
die, the wall thickness will decrease. In order to
manufacture multiple wall thickness transitions along the
tube’s length, the mandrel needs to be drawn back. Since
the control system of the extrusion press did not allow
mandrel motion against extrusion direction (ED) during
the extrusion process, it was necessary to interrupt the
extrusion before drawing the mandrel back to its starting
position. Two variations of the draw back process were
investigated. For the first the mandrel was simply drawn
back using the billet piercing system of the press. The
second alternative included the release of the pressure
from the billet and subsequent repositioning of the
mandrel to its starting position. Afterwards the extrusion
was continued with extruding a thick-walled section
again. The principle of the process is displayed in Fig. 1.
For the experiments homogenized, predrilled billets
of AA6060 were used and heated to billet temperature
T
B
in an induction furnace. By using the described
procedure two different extrusion trials were carried out
and the wall thickness was changed multiple times
during each extrusion. The ram velocity v
ram
was set to
0.9mm/s. Both tubes were water quenched at the end of
the hollow stem about 0.9m behind the die in order to
‘freeze’ the as extruded state and compare it to the T6
heat-treated state in future investigations. Table 1
presents the significant process parameters for the
extrusion trials. Therein D
M
describes the diameter of
mandrel tip and mandrel shaft and R is the local
extrusion ratio.
The microstructures of tube sections with constant
wall thickness were analyzed after anodizing with
barker’s reagent. The mean grain size was then
determined by linear intercepts of grain boundaries.
Fig. 1. Principle of extrusion of tubes with variable wall thicknesses
using an axially moving stepped mandrel (ED)
Table 1. Parameters of extrusion trials
No. TB D
M t R vram
[°C] [mm] [mm]
[mm/s]
1 500 32 4.5 22:1 0.9
37 2.0 46:1
2 400 32 4.5 22:1 0.9
37 2.0 46:1
2.2. 3-point bending tests
The mechanical properties of the extruded tubes were
characterized by 3-point bending tests of tube sections.
Aim was to investigate the influence of the local wall
thickness (local degree of deformation) and the
microstructure on the bending strength of the extruded
tubes. These tests were performed on a 20t
tension/compression machine using a bending mandrel.
The diameter of the round bending mandrel tip was
20mm. Fig. 2 shows the experimental setup.
Thin-walled (t=2.0mm) and thick-walled (t=4.5mm)
sections of each hollow profile with length of 250mm
were tested. The distance between the supporting rolls
was set to 100mm and the testing speed to 20mm/min.
During the tests the force and the displacement of the
machine’s traverse was measured. With the maximal test
force (F
max
) the maximal bending moment M
b,max
was
calculated using equation 1 [9], where L is the distance
between the supporting rolls.
(eq. 1)
Mb,max Fmax L
4
5
M. Negendank et al. / Procedia CIRP 18 ( 2014 ) 3 – 8
For round tubes the area moment of inertia (I
y
) is given
in eq. 2 [10] by:
(eq. 2)
D
o
and D
i
describe the outer and inner tube diameter.
With eq. 1 and eq. 2 the bending stress V
b
can be
calculated in dependence of the distance z from the
neutral fiber with eq. 3 [9]:
(eq. 3)
V
b
will be maximal at the tube surface (edge stress),
where z is half of the outer tube diameter.
Fig. 2. Experimental setup for 3-point bending tests of tubes
3. Results and discussion
3.1. Extrusion of seamless aluminum tubes with variable
wall thickness
Fig. 3 presents an example of a force-ram-
displacement diagram gained during the experiments. In
contrast to conventional indirect extrusion there is no
steady state. Instead, multiple peaks of the overall force
(F
ov
) and the ram velocity (v
ram
) can be noticed. This can
be explained as follows. At the beginning of each wall
thickness variation cycle the mandrel is positioned with
its smaller cross section in the die orifice. Thus, when
the force increases and reaches the flow stress of the
material at the given conditions, the material starts to
flow and a thick-walled tube section will be extruded.
When the mandrel step reaches in front of the die, the
gap between die and mandrel surface reduces.
Subsequently, the wall thickness decreases and the
extrusion ratio increases. This increase leads to higher
necessary extrusion forces, thus F
ov
increases. But
contrary to the expectation of a more or less linear
increase, a sudden decrease in F
ov
accompanied with an
abrupt acceleration of v
ram
was observed before
increasing.
Fig. 3. Extrusion diagram showing multiple cycles of wall thickness
variations for TB=400°C
There are two possibilities that could explain this
phenomenon. As previously found by finite element
analysis (FEA) [6, 11], due to the inwardly directed
material flow and lack of resistance by the moving
mandrel, the outer tube surface loses contact to the die
bearing and also to the mandrel tip as well as to the
mandrel step when the wall thickness is varied. That
means a reduction of the friction area and thus lower
necessary F
ov
. The second possibility is the backpressure
acting on the area of the mandrel step. The material in
the deformation zone in front of the die creates this
pressure. As the distance between mandrel step and die
decreases the pressure in that area increases resulting in
higher extrusion force requirements. Once the mandrel
step has entered into the die and the backpressure is
overcome, the extrusion force decreases abruptly as can
be seen in Fig. 3. At the same time v
ram
increases
significantly leading to the peaks in the v
ram
curve.
Afterwards v
ram
remains about constant and F
ov
increases
due to the higher extrusion ratio, since the wall thickness
now is reduced from 4.5mm to 2.0mm. When the
mandrel shaft has entered about 15mm into the die the
extrusion process was interrupted in order to draw the
mandrel back to its initial position. The shear drops of
the F
ov
-curves in Fig. 3 represent these positions. The
different procedures of mandrel drawback can also be
distinguished in the force-ram displacement diagram.
When the pressure on the billet is relieved before the
mandrel is drawn back, the ram position changes about
some centimeters against extrusion direction. This was
not observed when the billet remained under pressure.
After the mandrel was drawn back, the extrusion was
Iy
S
64 Do
4Di
4
V
bz
Mb,max
Iy
z
6 M. Negendank et al. / Procedia CIRP 18 ( 2014 ) 3 – 8
continued with another cycle of wall thickness variation,
starting by extruding a thick-walled tube section. The
development of the temperature measured in the die
bearing qualitatively follows the trend of F
ov
.
3.2. Characterization of tube dimensions in transition
zones between two wall thicknesses
Subsequent to the extrusion experiments the tubes
were examined. A longitudinal section of transition areas
in Fig. 4 proofs that the tube wall thickness was
successfully changed during extrusion.
Fig. 4. Longitudinal section of tube with variable wall thickness
showing abrupt and graded transition, (ED )
Fig. 5. Development of outer tube diameter Do and wall thickness t in
graded wall thickness transition, ED
Caused by the process three different kinds of wall
thickness transitions were observed. The first type is
called continuous or graded transition since the wall
thickness is varied continuously over a relatively long
profile length (right side in Fig 4). This transition is
generated when the mandrel step enters into the die.
Since the gap between mandrel step and die reduces
gradually with mandrel displacement, the tube wall
thickness is also gradually decreased. The wall thickness
development over the transition length is displayed in
Fig. 5. It can be learned that the reduction of the wall
thickness from t=4.5mm to 2.0mm takes place over a
profile length of about 150mm. At the beginning the
decrease is relatively sharp but then becomes gradually.
On the other hand when the extrusion is interrupted
and the mandrel drawn back before resuming the
extrusion, the wall thickness changes abruptly over a
very short tube length. On the left side Fig. 4 shows that
a wall thickness transition from t=2.0mm to t=4.5mm
was achieved within a profile length of 1mm when the
billet pressure is released before drawing the mandrel
back to its starting position. When the pressure is not
released before the drawback the result can be abrupt
wall thickness transitions that are stretched in extrusion
direction. Fig. 6 shows that in this case the wall
thickness transition takes places over an extended tube
length of 30mm. This can be explained by the reducing
extrusion ratio when the mandrel shaft is pulled out of
the die. When R is reduced, the remaining billet pressure
can still be sufficiently high enough to force the material
to flow through the now increased orifice between die
and mandrel.
Fig. 6. Longitudinal section of stretched abrupt transition manufactured
without releasing pressure from billet for mandrel drawback, (ED )
Thus, with abrupt wall thickness transitions tubes can
be strengthened (‘tailored’) with increased cross sections
in very localized profile areas that are highly stressed.
Furthermore, it was revealed that in the wall thickness
transitions the tubes’ surfaces were dent. In order to give
a quantitative evaluation of these observations, the
development of the outer tube diameter was also
determined in Fig. 5 for the continuous transition and for
the abrupt transition (with relieved billet pressure) in
Fig. 7.
Fig. 7. Development of outer tube diameter Do and wall thickness t in
abrupt wall thickness transition with released billet pressure, ED
7
M. Negendank et al. / Procedia CIRP 18 ( 2014 ) 3 – 8
As previously found [11], the reduced outer tube
diameter in the graded transitions forms due to the lack
of mandrel resistance against the inwardly directed
material flow when the mandrel step is right in front of
the die as well as in the die bearing. On the other hand in
abrupt transitions manufactured without releasing the
billet pressure a surface dent also developed (Fig. 6). In
this case the material does not flow parallel to the
extrusion direction after mandrel draw back but in
direction of the mandrel surface. Thus, the previously
extruded thin-walled section in contact with the die
bearing is ripped off the bearing and a surface dent is
formed [11]. Abrupt transitions formed after releasing
the billet pressure for mandrel drawback have not yet
been investigated by FEA. But the reason for the locally
increased outer tube diameter (Fig. 7) should also be
attributed to the material flow. As the material starts
flowing inwardly in direction of the mandrel after the
extrusion is resumed subsequent to the mandrel’s draw
back, it could have been deflected by the mandrel
surface and directed towards the tube surface, resulting
in the increased outer diameter. This effect will be
further investigated in future work by FEA and detailed
analysis of the microstructure in these regions.
3.3. Analysis of tube microstructures
The analysis of the tube microstructures revealed two
significantly different types. Firstly, fully recrystallized
microstructure consisting of relatively big, equiaxed
grains was found at the higher billet temperature of
TB=500°C in all tube cross sections of t=2.0mm,
t=4.5mm, the graded transition as well as the abrupt
transition. Fig. 8 gives a characteristic example of this
kind of microstructure. Fully recrystallized grains were
also found at the lower TB=400°C, when the extrusion
ratio and thus the profile velocity as well as the local
profile temperature were high, thus improving the
conditions for recrystallization [12].
The mean grain sizes of the tube sections of constant
wall thicknesses (t=4.5mm and t=2.0mm) are given in
table 2. It can be learned that the microstructure and the
grain size in these specific sections are very similar for
TB=500°C. Thus, the differences in local degree of
deformation, product velocity as well as the product
temperature do not have an influence on the local
microstructures in these specific areas. A comparison of
the grain size of the thin walled areas between both tubes
revealed a lower grain size for TB=400°C. Due to the
lower TB the energy for grain growth during
recrystallization is lower resulting in smaller grain size.
On the other hand when the extrusion ratio and the
product velocity are low as they are during extrusion of
the thick-walled section at TB=400°C, the
microstructure is dominated by thin grains, which are
elongated in extrusion direction (Fig. 9). Obviously
these extrusion conditions did not generate sufficient
energy for the activation of recrystallization processes.
With higher magnification it was found that the grain
boundaries are serrated. These serrations develop during
the deformation and are an indication of dynamic
recovery processes [13]. Because of the morphology of
the elongated grains (very thin and long) a grain size
determination could not be given.
Table 2. Grain size [μm] in dependence of local tube wall thickness t
and TB
t [mm] TB=500°C TB=400°C
2.0 132 81
4.5 126 deformed
Fig. 8. Fully recrystallized microstructure with equiaxed grains
TB=500°C, t=2.0mm), ED
Fig. 9. Microstructure with grains elongated in ED (TB=400°C,
t=4.5mm), ED
3.4. Mechanical tube properties in 3-point bending tests
During the three point bending tests force-
displacement curves were gained. From these curves the
maximal forces were used in order to calculate the
maximal bending moments (eq. 1) and the edge stresses
(eq. 3) on the tube surface. The results are given in table
3.
8 M. Negendank et al. / Procedia CIRP 18 ( 2014 ) 3 – 8
When the calculated edge stresses are compared, it
can be learned that for t=2.0mm as well as for t=4.5mm
higher stresses were found for the tube extruded at the
higher billet temperature of TB=500°C. This observation
can be explained by the higher solubility of especially
Mg- and Si-atoms in the aluminum solid solution at the
higher TB. Since the tubes were water quenched behind
the hollow stem, precipitation could be reduced leading
to higher strength of the aluminum solid solution. This
behavior was previously observed for the results of
Vickers hardness measurements in [14].
Table 3. Results of three point bending tests
TB
[°C]
t
[mm]
Fmax
[kN]
Mb, max
[Nm]
Iy
[mm4]
Vb
[MPa]
500 2.0 4.8 120 44766 54
500 4.5 19.3 483 84429 116
400 2.0 3.5 88 45625 39
400 4.5 15.2 380 84082 92
4. Conclusions
1. The wall thickness of aluminum tubes (AA6060) was
successfully varied multiple times along tube
direction during extrusion by applying axially
moving a stepped mandrel.
2. Two significantly different wall thickness transition
types were produced. The transition can either be
manufactured gradually when the mandrel step is
continuously moved into the die. Or the transition
can be very sharp (abrupt) when the extrusion is
stopped for drawing the mandrel back to its starting
position. Thus, load adapted (tailored) aluminum
tubes with thick-walled sections located only in
highly stressed areas can be manufactured. Since less
stressed profile areas can be produced with lower
wall thicknesses, the overall profile weight can be
reduced.
3. The outer tube diameter in wall thickness transitions
differed from the set values. This is due to the
inwardly directed material flow and lack of
resistance by the mandrel during the transitions [11].
4. The variation of the local extrusion ratio and thus the
product velocity as well as the product temperature
during the experiments lead to significantly different
microstructures in thin-walled and thick-walled
sections of the tube extruded at TB=400°C. At
TB=500°C the microstructure consisted of
recrystallized grains with similar grain size in all
parts of the tube.
5. Results of three point bending tests revealed that
sections of a tube extruded at TB=500°C showed
higher strength than sections with same wall
thicknesses extruded at TB=400°C. This was
explained by solid solution hardening.
Acknowledgements
The German Research Foundation (DFG) under project
number MU2963/5 founds this project.
References
[1] Otto Fuchs KG Meinerzhagen, Tailored aluminium tubes –
extruded tubes with variable wall thickness, International
Aluminium Journal Vol. 89, 10, 2013, pp. 35-36
[2] K. B. Müller: Bending of extruded profiles during extrusion
process, International Journal of Machine Tools and Manufacture,
Vol. 46, 2006, pp. 1238-1242
[3] S. Chatti, M. Hermes, M. Kleiner, Three-Dimensional Bending of
Profiles by Stress Superposition Advanced Methods in Material
Forming, 2007, pp. 101-118
[4] A. Jäger, D. Risch, A.E. Tekkaya, Thermo-mechanical processing
of aluminum profiles by integrated electromagnetic compression
subsequent to hot extrusion, Journal of Materials Processing
Technology, Vol. 211, 2011, pp. 936-943
[5] T. Makiyama, M. Murata, Materials Science Forum, Vols. 396-
402, (2002), pp. 513-520
[6] M. Negendank, S. Müller, W. Reimers, Extrusion of tailored
alumimun tubes, Proceedings of the 8th World Congress
Aluminium Two Thousand, 14.-18.5.2013, Milan, Italy, ISBN
978-88-87971-0-2
[7] Selvaggio, A., Chatti, S., Ben Khalifa, N., Tekkaya, A. E.,
New developments in extrusion of profiles with variable
curvatures and cross-sections. In: Proceedings of the Tenth
International Aluminium Extrusion Technology Seminar ET´12,
Miami, Florida (USA), 2012, pp. 505-512
[8] K. Hara, N.Ohtake, K. Horita, Variable-Shape Extrusion of
Aluminum Square Pipes using DLC coated dies, New Diamond
and Frontier Carbon Technology, Vol. 14, 2004, pp. 331-34004
[9] B. Heine, Werkstoffprüfung Ermittlung von Werkstoffeigen-
schaften, 2003, Carl Hanser Verlag, München Wien, ISBN 3-446-
22284-7
[10] K.-H. Grote, J. Feldhusen, Dubbel Taschenbuch für den
Maschinenbau, 22. Auflage, 2007, Springer, Berlin, Heidelberg,
New York, ISBN 978-3-540-49714-1
[11] M. Negendank, S. Müller, W. Reimers, Finite element analysis of
material flow and characterization of tube dimensions of
indirectly extruded tailored aluminum tubes, Proceedings of
THERMEC 2013, Las Vegas, Nevada, USA, accepted manuscript
[12] E. D. Sweet, S. K. Caraher, N. V. Danilova, Xinquan Zhang,
Effects of Extrusion Parameters on Coarse Grain Surface Layer in
6xxx Series Extrusions, In: Proceedings of the Eighth
International Aluminum Extrusion Technology Seminar, Vol. I,
2004, pp. 115-126
[13] J.F. Humphreys, M. Hatherly, Recrystallization and Related
Annealing Phenomena, second edition, 2004, Elsevier
[14] M. Negendank, S. Müller, W. Reimers, Microstructural
development and mechanical properties of indirectly extruded
tailored aluminum tubes, International Congress on Light
Materials –Science and Technology (LightMAT2013), 2013,
Bremen, Germany,
http://www.dgm.de/tagungen/?tgnr=1171&cat=&edate=05.09.201
3&lg=en
