Binded 2

13
Journal of Materials Processing Technology 214 (2014) 1341–1351
Contents lists available at ScienceDirect
Journal of Materials Processing Technology

journal homepage: www.elsevier.com/locate/jmatprotec
Deformation analysis of double-sided tube hydroforming in

square-section die
Xiao-Lei Cui, Xiao-Song Wang ∗ , Shi-Jian Yuan
National Key Laboratory of Precision Hot Processing of Metals, Harbin Institute of Technology, Harbin 150001, Heilongjiang, PR China
a r t i c l e i n f o a b s t r a c t
Article history: In this paper, in order to explore the deformation behavior of double-sided tube hydroforming in square-
Received 3 September 2013 section die, effect of external pressure on the critical effective strain was theoretically analyzed firstly.
Received in revised form 15 January 2014 And then a special experimental setup was designed for double-sided tube hydroforming in which the
Accepted 1 February 2014
difficulty of simultaneous loading was overcome using two independent intensifiers and servo controlling
Available online 12 February 2014
while the sealing of external pressure was guaranteed by the O rings assembled in the interfaces of
mandrel/base plate and outer cylinder/base plate. Furthermore, 5A02-O aluminum alloy tubes with outer
Keywords:
diameter of 63 mm and thickness of 2 mm were investigated under different external pressures varying
Double-sided tube hydroforming
External pressure
from 0 to 80 MPa. At the same time, numerical simulation was conducted using the Abaqus/Explicit
5A02 aluminum alloy software. It is shown that increasing of external pressure has an effect on the fraction of grain boundaries,
Transition zone the number and size of the microvoids and the microhardness in the transition zone, and thus increases
Microstructure the critical effective strain in the transition zone. It can be concluded that the deformation ability of
the transition zone is improved by the external pressure in double-sided tube hydroforming of square-
section. This investigation shows that double-sided tube hydroforming is a potential forming method for
the fabrication of lightweight hollow structures using the tubes with low ductility.
© 2014 Elsevier B.V. All rights reserved.
1. Introduction As is known, material’s formability is not only related to its

inherent properties, but also has a strong dependence on the
With the urgent requirement of energy conservation and imposed stress state when it is undergoing plastic deformation
emission reduction in the fields of automobile and aerospace, (Lewandowski and Lowhaphandu, 1998). Motivated by such theo-
lightweight alloys, such as aluminum alloy and magnesium alloy, ries, double-sided tube hydroforming is proposed, where hydraulic
were proven to be the preferred alternative to traditional steel pressure is introduced on both sides of the tube and/or the axial
materials (Hirsch and Al-Samman, 2013). The structural members force is applied on the tube ends simultaneously, and then the tube
with different cross-sections have been widely used in aerospace is formed into the required shape under the pressure difference
and automobile industry due to their excellent bending perfor- and/or the axial force (Jain, 2003). Nucleation, growth and coa-
mance and high stability of assembly (Lang et al., 2004). However, lescence of micro voids in the tube materials can be suppressed
the onset of plastic instability is more likely to happen in the tran- effectively by applying external pressure on the outside surface,
sition zone because of the too large hoop stress in the process of which could change the stress state and increase the hydrostatic
corner filling for tubes with poor formability such as aluminum pressure in the deformation zone, thus the forming limit of the
alloy tubes and magnesium alloy tubes (Liu et al., 2006). Tube tube is improved.
hydroforming process is one of the most important forming tech- In 1966, Fuchs et al. (1966) found that the forming limit of
nology in which the formability of the tube can be improved tube can be improved remarkably by applying liquid pressure to
obviously by the combined action of axial force and internal pres- the inside and outside of the tube simultaneously. Since then both
sure. Nevertheless, the stress state applied on the tube can be domestic and international scholars started to study the effect of
simplified as plane stress state due to the negligible internal pres- through-thickness normal stress on the sheet metal formability.
sure compared with the in-plane stress (Dohmann and Hartl, 1996). Forming limit model considering through-thickness normal stress
were proposed on the basis of the classical theory of plastic instabil-
ity. Smith et al. (2003a) studied six different formability models for
∗ Corresponding author. Tel.: +86 451 86417917; fax: +86 451 86417917. suitability for finite element modeling of the double-sided high-
E-mail addresses: [email protected] (X.-L. Cui), [email protected] pressure tubular hydroforming process. Smith et al. (2003b) also
(X.-S. Wang), [email protected] (S.-J. Yuan).
http://dx.doi.org/10.1016/j.jmatprotec.2014.02.005
0924-0136/© 2014 Elsevier B.V. All rights reserved.
1342 X.-L. Cui et al. / Journal of Materials Processing Technology 214 (2014) 1341–1351
Fig. 1. Principle of double-sided tube hydroforming in a square-section die: (a) initial stage; (b) corner filling stage.
adopted Stoughton’s plane-stress model based on Hill’s 48 yield normal stress can be controlled artificially, has not appeared in the
criterion and proposed a new sheet metal formability model that published academic papers.
takes the through-thickness normal stress into account. Matin and In the present study, theoretical analysis of the effect of external
Smith (2005) improved their model proposed in 2003 by assuming pressure (through-thickness normal stress) on tube formability in
that the in-plane strain ratio as constant according to the prac- a square-section die was carried out based on the classical theory
tical observation. Jain and Wang (2005) have made a theoretical of plastic instability. Then a special experimental setup for double-
analysis of plastic instability for the tube under different load- sided tube hydroforming was designed and 5A02-O aluminum alloy
ing conditions, and then drawn a conclusion that the formability tubes were bulged in the experimental setup with a square-section
of the tube could be improved by imposing liquid pressure on die cavity under the combined action of internal pressure and
the tube outside no matter how much axial feeding participated external pressure. In addition, numerical simulation was conducted
in the process. In the study of effect of external pressure on the using the Abaqus/Explicit software. The thickness distribution and
plastic instability of the thin-wall tubes, Guven (2009) obtained effective strain distribution were validated detailedly with experi-
the circumferential limit strain and axial limit strain considering mental, theoretical and simulation study. Finally, the deformation
the effect of through-thickness normal stress using three differ- mechanism in the transition zone under different external pres-
ent necking criteria. Moreover, initial imperfection was assumed sure is clarified by the electron back scattering diffraction (EBSD),
to be inevitable for the actual sheet materials and the influence fracture morphology and micro hardness.
of through-thickness normal stress on formability was discussed
based on the M-K model. Allwood and Shouler (2009) proposed 2. Principle of double-sided tube hydroforming in a
a generalized forming limit diagram (GFLD) accounting for the square-section die
possibility that all six components of the stress and strain ten-
sors in sheet forming may be non-zero. Assempour et al. (2010) The principle of double-sided tube hydroforming in a square-
modified the FLC based on the M-K model by using Hill48 3D section die is shown in Fig. 1. Compared with the conventional
yield criterion. Nurcheshmeh and Green (2012) did the similar tube hydroforming of square section in which the tube is deformed
work and obtained both the FLC and FLSD with the effect of a toward to the die corner only by the function of internal pressure,
non-negligible through-thickness normal stress. The forming limit external pressure is introduced to the tube outside simultane-
models mentioned above considered the influence of through- ously in the double-sided tube hydroforming of square section
thickness normal stress by using three-dimensional yield criteria. and the tube is deformed under the hybrid effect of internal pres-
However, Banabic and Soare (2008) used a general method to sure and external pressure. The external pressure, as a supporting
describe the effect of the normal pressure while maintaining the role to increase the normal compressive stress, could change the
plane-stress characteristic of the model. In addition, Jain et al. stress state and increase the hydrostatic pressure for the tube, thus
(2004) established the merits of applying external counter pres- improve its deformation ability.
sure in tube hydroforming using the finite element simulation. The loading path for the double-sided tube hydroforming in a
Liu et al. (2012) simulated the effect of through-thickness normal square-section die is shown in Fig. 2. It should be ensured that the
stress on the sheet formability using the finite element software internal pressure is greater than the external pressure for a cer-
based on the ductile fracture criterion. Although Wang (2009) has tain value in the process of establishment for external pressure
investigated the role of through-thickness normal stress in viscous (T0–T1) owing to the shape feature of the tube. The specific rela-
pressure bulging of overlapping sheets and double-sided viscous tionship between internal pressure and external pressure is shown
pressure bulging, the normal stress induced by the overlapping as follows according to the mechanical equilibrium condition.
sheet or the viscous medium was uncontrollable and the shear t

stress also play an important role on the formability improvement pi = pe 1 + (1)
r
for the target sheet. As the above mentioned issues are concerned,
most investigations of influence of through-thickness normal stress where pi is the internal pressure, pe is the external pressure, t is
on sheet formability are focused on the theoretical analysis and the initial thickness of the tube and r is the inside radius of the
numerical simulation. The systematical experimental work about tube. In the stage of establishment for external pressure (T0–T1),
double-sided tube hydroforming, in which the through-thickness the internal pressure and external pressure are increased to their
target values respectively according to Eq. (1). In the corner filling
X.-L. Cui et al. / Journal of Materials Processing Technology 214 (2014) 1341–1351 1343
Fig. 3. Schematic diagram of stress status in straight wall area.
According to Fig. 3, the differential equilibrium equation for

straight wall area along the x-direction (circumferential direction)
can be written as follows.
( + d )t = Ndx + t (4)
where N is the positive pressure between the tube and the die.
After rearrangement and integral of Eq. (4), the circumference
Fig. 2. Schematic of loading path. stress in the straight wall area is expressed as below.
x C
= −N + (5)
t t
stage, the external pressure is maintained at the target value and
where C is the constant term.
the increase of internal pressure does not stop until the occurrence
When there is no external pressure applied on the tube outside,
of fracture on the tube. The tube is deformed under the function of
the circumference stress in the transition zone (x = 0) is = pi r/t.
pressure difference. When fracture occurs on the tube the internal
Substituting this boundary condition into Eq. (5), constant term C
and external pressure should be unloaded in the same time.
can be obtained and Eq. (5) can be written as below:
r − x
= pi (6)
3. Theoretical analysis t
As the tube hydroforming in a square-section die is concerned,
In general, materials locate in the transition zone firstly satisfy
plane strain state is usually adopted for the tube, and according to
the yield condition due to the higher circumferential stress in this
Mises yield criteria
zone than in the corner area and straight wall area when the tube
is bulged in a square-section die. As a result, the deformation will 2
− r = √ ¯ (7)
concentrate on the transition zone after the occurrence of necking 3
and then lead to the final fracture (Liu et al., 2006). Substituting Eq. (6) and the boundary condition r = −pi into
In the previous investigation, an analytical solution giving con- Eq. (7), the effective stress in straight wall area is expressed as
ditions at instability of a thin-walled tube subjected to internal following.
pressure and independent axial tension load was presented by √ √ r − x
Mellor (1962). Then the plastic instability criterion of tube under 3 3
¯ = ( − r ) = pi +1 (8)
the combined actions of internal pressure, external pressure and/or 2 2 t
axial force was studied by Jain and Wang (2005) based on the In tube hydroforming, the process will become unstable when
approach presented by Mellor (1962). The critical effective strain increase in internal pressure due to work hardening is balanced
in the corner area of the tube can be written as follows accord- by reduction in pressure due to geometry changes (Jain and Wang,
ing to the instability criterion mentioned above with only internal 2005). Assuming the friction coefficient = 0, and from Eq. (6)
pressure functioned:
t d d
t
dpi = ıε̄ + ıε̄ = 0 (9)
1 r dε̄ dε̄ r
ε= √ n (2)
3 Therefore
t 2 dY √ t 2
But with the integrated actions of internal pressure pi and exter- √ ıε̄ − 3 √ Y − pi ıε̄ = 0 (10)
r 3 dε r 3
nal pressure pe , critical effective strain in the corner area will
become to be the following expression: Rearranging Eq. (10), a new form of expression can be obtained
√ √
1
pe
1 dY 2 3Y − 3pi 3 1
ε= √ n 1+ (3) = = = (11)
Y dε 2Y + pi z
3
Hence, critical effective strain
It can be seen from Eqs. (2) and (3) that the formability in the
+ pi n
pi

corner area can be improved with the increasing external pressure. ε̄ = nz = n √ = √ 1+ (12)
3 3
However, when the tube is bulged in a square-section die, fracture
will occur in the transition zone, which may also be influenced by It can be found from Eq. (12) that in the single-sided tube hydro-
the deformation characteristic of straight wall area. forming in a square-section die the critical effective strain of the
The deformation characteristic of the straight wall area is very tube in the straight wall area depends on the work hardening expo-
different from the corner area. The schematic diagram for stress sta- nent and the internal pressure.
tus in the straight wall area is shown in Fig. 3. The friction coefficient However, as the double-sided tube hydroforming in a square-
between the tube outside and die is assumed to be . section die is concerned, the circumference stress in the transition
zone (x = 0) should be expressed as follows based on the mechanical

equilibrium condition.
pi r
r + t
= − pe (13)
t t
Recalling Eq. (5), the circumference stress is easily obtained in a
form of
x r r
= −(pi − pe ) + pi − pe − pe (14)
t t t
Combining Eqs. (7) and (14), the effective stress in straight wall
area is expressed as below.
√
3 r+t x
¯ = (pi − pe ) − (pi − pe ) (15)
2 t t
Using the same argument of ‘incipient bulging’ used by Mellor
(1962), assuming the friction coefficient = 0, and from Eq. (14)
t d d
t d
t
ıε̄ + ıε̄ + pe ıε̄ = 0 (16) Fig. 4. True stress–strain curve for 5A02-O aluminum alloy tube along the axial
r dε̄ dε̄ r dε̄ r direction.
Therefore
t 2 d¯ √ t 2 √ t tensile test was conducted on an Instron 5569 machine along the
√ ıε̄ − 3 √ ¯ − pi ıε̄ − 3 pe ıε̄ = 0 (17) axial direction of tube. The true stress–strain curve is shown in Fig. 4
r 3 dε r 3 r
and the material parameters are shown in Table 1.
Rearranging Eq. (17), a new form of expression can be obtained
√ √ 4.2. Experimental setup
1 d¯ 2 3¯ − 3(pi − pe ) 3( + pe ) 1
= = = (18)
¯ dε 2¯ + pi z The special experimental setup for double-sided tube hydro-
Hence, critical effective strain forming was designed and assembled, as is shown in Fig. 5. The
1 + pi n + pi n
pi − pe
experimental die is mainly composed of outer cylinder, mandrel,
ε̄ = nz = n √ = √ = √ 1+ (19) die cavity, clamping blocks and base plates. The cross-sectional

3 + pe
3 + pe 3 + pe
shape and length of the bulging area can be changed by replacing
When the friction coefficient = / 0, calculation process is very the die cavity and clamping blocks with others. The experimen-
complicated and no further analysis has been conducted in this tal die, which is different from the conventional hydroforming die
study. By comparing Eqs. (12) and (19) conclusion can be drawn that consist of up die and low die, was designed into closed cylinder
the external pressure has no effect on the critical effective strain of structure in order to solve the sealing problem for external pres-
the straight wall area under the assumption of zero friction in the sure. The sealing of internal liquid and external liquid was achieved
tube/die interface, because hydrostatic pressure has no influence by the O rings assembled in the spaces between mandrel and tube,
on the yield behavior of the tube in terms of Mises yield function. mandrel and base plate, outer cylinder and base plate. Moreover,
In conclusion, external pressure can improve the formability of the clamping force should be applied on the base plate through
the corner area for the tube, but has no effect on the critical effective press machine in order to avoid the separation of base plate and
strain of the straight wall area. The deformation ability of the tran- outer cylinder due to increasing external pressure.
sition zone, where is the most dangerous area for fracture, may be The experimental study of double-sided tube hydroforming in
improved by the external pressure theoretically. However, it should a square-section die was carried out on the 10 MN hydroforming
be noted that once the tube is deformed to contact the die closely machine in Harbin Institute of Technology. To carry out the double-
in straight wall area, there are two different extreme situations for sided hydroforming experiment, the hydroforming machine was
the tube/die interface. One is that external liquid could exist on the upgraded. Another intensifier (intensifier B) was introduced to sup-
tube/die interface completely in the straight area which is the same ply the external pressure while the intrinsic intensifier (intensifier
as in the corner area. The other situation is that the external liquid A) was used to supply the internal pressure. The two intensifiers
will be squeezed out from the tube/die interface, and in fact the were both controlled by servo valves. The control system was re-
external pressure has no possibility to exist on the tube/die inter- designed and a new double-pressure control mode was introduced.
face. The difference between these two situations is that the friction When experiments are performed the internal pressure was
condition on the tube/die interface. If the second condition is cor- provided by intensifier A through the inlet on the mandrel, while
rect in actual hydroforming process, the contact pressure between the external pressure was supplied by intensifier B through the inlet
the tube and die, N, is equal to the internal pressure pi . As a result, on the outer cylinder. In particular, the two independent intensi-
the friction force on the tube/die interface would increase with the fiers could be coupled together by the servo valves A and B, whose
increasing internal pressure due to the external pressure, and thus signals were supplied by the control system, to achieve accurate
cause more inhomogeneous deformation in the straight wall area,
which is disadvantageous for the corner filling of the tube. Table 1
Material parameters of 5A02-O aluminum alloy tube along the axial direction.
4. Experimental Material parameters Value
Yield stress, s (MPa) 78.3

4.1. Material Ultimate tensile strength, b (MPa) 182.9
Yield ratio 0.43
The as-received material used in this study is 5A02-O aluminum Total elongation, ı (%) 23.9
alloy tube with outer diameter of 63 mm and nominal thickness Strength coefficient, K (MPa) 347
Strain hardening exponent, n 0.245
of 2 mm, provided by Northeast Light Alloy Co., Ltd. The uniaxial
Fig. 5. Experimental setup of double-sided tube hydroforming process: (a) three-dimensional model; (b) experimental setup.
servo controlling. And then both of the internal pressure and exter- specimens for EBSD were cut from the transition zone, as shown
nal pressure could be increased or decreased strictly according to in Fig. 7, and prepared by electropolishing at room temperature
the loading path implemented into the control system in advance. within 8 s, in electrolytic solution containing 90% perchloric acid
The tube used in this study has a total length of 234 mm. The and 10% absolute ethyl alcohol. The EBSD data was collected with a
square-section die cavity has a side length of 63 mm, equal to the step size of 6.0 ␮m. Meanwhile, fracture morphology was observed
outer diameter of the tube, and a bulging length of 140 mm. Three by scanning electron microscope (SEM). In addition, microhardness
loading paths, in which the external pressure is 0 MPa, 40 MPa and in the transition zone was measured using the Vickers hardness
80 MPa respectively, were used in this study, as is shown in Table 2. tester.
4.3. Finite element model

5. Results and discussion
In order to investigate the effect of external pressure and fric-
tion on the tube/die interface on thickness distribution and strain Fig. 7 shows the cracked tubes deformed under different exter-
distribution of the tube, One quarter plane strain finite element nal pressures. The effect of external pressure and the related
model for double-sided tube hydroforming in a square die was microstructure evolution will be discussed in details as following.
established using the Abaqus/Explicit software, as is shown in Fig. 6.
Two symmetrical planes are constrained. The die was assigned to be
analytic rigid shell. The tube metal was modeled by the plane strain 5.1. Effect of external pressure on distribution of thickness and
solid element CPE4R (A 4-node bilinear plane strain quadrilateral, strain
reduced integration, hourglass control) and was divided into ten
layers in thickness direction. The internal pressure and external In order to estimate the thickness distribution more accurately,
pressure were introduced to the inside and outside of the tube at the thickness distribution of the initial tube (t0 ) along the circum-
the same time according to the loading paths presented in Table 2, ferential direction was measured firstly, and then the thinning ratio
to investigate the influence of external pressure on the deformation
behavior of tube. Moreover, different friction coefficients between
the tube and die ( = 0.2, 0.4 and 0.6) were used in the studying of
effect of friction on the strain distribution when only the internal
pressure functioned in the deformation process.
4.4. Microstructure and hardness measurement
The grain structure was analyzed by electron backscattering

diffraction (EBSD) technique on Quanta 200 FEG-SEM machine. The
Table 2
Three loading paths used in the experimental study.
Loading path Pressure/MPa
T0 T1 T2 T3
1 pi 0 0 25 0
pe 0 0 0 0
2 pi 0 42.7 65 0
pe 0 40 40 0
3 pi 0 85.4 105 0
Fig. 6. One quarter plane strain finite element model of double-sided tube hydro-
pe 0 80 80 0
forming in a square die.
Fig. 7. Cracked tubes under different external pressures: (a) 0 MPa; (b) 40 MPa; (c) 80 MPa.
can be obtained after the thickness distribution of the cracked tube

(t) was measured according to the following equation:
t − t0
Thinning ratio (%) = × 100 (20)
t0
The points for the thickness measurement are shown in Fig. 7,
in which point 1 and point 28 represent the midline of the straight
wall area while point 7 and point 22 locate at the transition zone.
The distance between adjacent points is about 2 mm. Then the aver-
age value of thickness of the three un-cracked corner is calculated
and the thickness distribution of the deformed tubes under differ-
ent external pressures is shown in Fig. 8(a). It can be found from
Fig. 8(a) clearly that higher external pressure can increase the thin-
ning ratio of the tube in the transition zone but has a little effect on
the thickness distribution in the straight wall area and corner area.
In general, the axial strain of the tube is close to zero when the
tube is bulged in a square-section die, so the strain state of the tube
can be regard as plane strain state in which the axial strain εz is
assumed to be zero (Yang and Ngaile, 2008). The thinning strain
εt could be obtained after the thinning ratio was measured. The
circumferential strain ε can be easily got based on the incompress-
ibility of the tube and the assumption of plane strain state.
ε = −εt (21)
Therefore, the effective strain could be expressed as follows

2 2
ε̄ = ε2 − ε εt + ε2t = √ |εt | (22)
3 3
The effective strain can be obtained directly from the thickness
measurement of the tube and the subsequent simple calculations.
The effective strain distributions of the cracked tubes under dif-
ferent external pressures are shown in Fig. 8(b). It is seen that the
distribution trends of the strain and thickness are almost same.
The maximum deformation occurs in the transition zone and has a
dependence on the external pressure.
From the aspect of theoretical analysis, the relationship between
the critical effective strain and external pressure could be obtained
according to Eqs. (3) and (19). However, the relationship between Fig. 8. Thickness and effective strain distribution of tubes under different external
pressures: (a) thickness; (b) effective strain.
the circumference stress and external pressure pe should be
obtained first based on the incompressibility. It can be expressed
as = 0 − pe , in which is the circumference stress when the
applied external pressure is pe and 0 is the circumference stress

when there is no external pressure imposed on the tube outside.
So Eqs. (3) and (19) can be expressed as follows:

1 pe
ε= √ n 1+ (23)
3 0 − pe

n pi − pe
ε̄ = √ 1+ (24)
3 0
It can be found that Eqs. (12) and (24) are the same actually,
in other words, the critical effective strain obtained from the equi-
librium equation of the straight wall has no dependence on the
external pressure. The values of circumference stress 0 and the
critical pressure difference pi − pe can be obtained based on the
numerical simulation results with the assistance of experimental
study, and they are about equal to 230 MPa and 20.4 MPa respec-
tively. As a result, the relationship of critical effective strain and
external pressure can be easily got according to Eqs. (23) and (24),
as is shown in Fig. 9. It can be found that increasing tendency of the Fig. 9. Comparison of the critical effective strain in the transition zone between
experimental results and theoretical prediction.
critical effective strain in the transition zone obtained from exper-
imental study is the same as obtained by the theoretical prediction
based on the instability criteria of the corner area. Moreover, the referring to the point of maximum load-bearing capacity, but neck-
data got from experimental study is a litter higher than the theoret- ing will play a dominant role in the subsequent deformation before
ical prediction. The reason of this phenomenon is that the critical fracture in which the thinning is unavoidable for many materials
effective strain obtained from classical plastic instability theory is in the actual deformation process.
Fig. 10. Effect of external pressure on the thickness of the tube: (a) external pressure = 0 MPa; (b) external pressure = 40 MPa; (c) external pressure = 80 MPa; (d) comparison
of thinning ratio between different external pressures.
Fig. 11. The effect of friction coefficient on the strain distribution: (a) = 0.2; (b) = 0.4; (c) = 0.6; (d) comparison of the PEEQ (equivalent plastic strain) distribution of
the middle layer of the tube under different friction conditions.
Fig. 10 shows the contour of through thickness normal strain localize at the transition zone earlier with higher friction force and
under different external pressure and the corresponding compar- lead to the earlier fracture for the tube.
ison of thinning ratio distribution when the pressure difference is It has been concluded from Fig. 8(b) that external pressure has
reach to 20.4 MPa. It can be found from Fig. 10 that the external a little effect on the thickness distribution in the straight wall area
pressure has no influence on the thickness distribution of the tube and corner area. In other words, the increasing of internal pressure
before the occurrence of necking instability no matter how much due to the existence of external pressure has a little function on the
the external pressure is. In fact, an appropriate ductile fracture cri- friction conditions on the tube/die interface in the experimental
terion should be implemented into the FEA software in order to study, which means that the external pressure could still exist on
predict the effect of external pressure on the thickness distribution the tube/die interface after the tube is deformed to contact with the
of the tube after occurrence of necking, which will be discussed in die. In a word, all the results presented above show that the external
the further investigation. pressure has a contribution to the improvement of the deformation
Furthermore, the contour of equivalent plastic strain under ability of the transition zone for the tube.
different friction conditions on the tube/die interface and the
corresponding comparison of equivalent plastic strain distribution 5.2. Effect of external pressure on microstructure
in the middle layer of the tube are shown in Fig. 11. It can be seen
from Fig. 11, increase of the friction coefficient on the tube/die Fig. 12 shows the microstructure of the as-received tube. From
interface could cause the more inhomogeneous deformation for Fig. 12(a), it is clearly seen that the grain size of the as-received
the tube. The higher friction force will make the deformation in the tube is close to equiaxed grain but not very uniform. Fig. 12(b)
straight wall area more difficult once the tube is in contact with gives the grain boundaries distribution of the as-received tube in
the die under internal pressure, and the deformation will mainly which the blue line represent 15–180◦ grain boundaries, the green
occur in the corner area and transition zone. On the contrary, line represent 5–15◦ grain boundaries and the red line represent
when the friction force on the tube/die interface is relatively low, 0–5◦ grain boundaries. The fraction of low angle grain boundaries
the deformation in the straight wall area is still possible in spite (LAGBs, <15◦ ) is about 62.8% while the fraction of high angle grain
of the occurrence of contact between tube and die, which could boundaries (HAGBs, >15◦ ) is about 37.2%.
make the tube obtain homogeneous deformation along the hoop It is generally known that low angle grain boundaries (LAGBs,
direction. In addition, the friction conditions have a little influence <15◦ ) is the result of the movement and rearrangement of dislo-
on the location and the equivalent plastic strain of the transition cations, and a high density of LAGBs suggest that the material has
zone under the same internal pressure. But the deformation may experienced more deformations (Biyikli et al., 2010). Fig. 13 shows
Fig. 12. Microstructure of as-received tube: (a) grain distribution; (b) grain boundaries distribution (Image Quality map).
the grain boundaries distribution (GBs) under different external 5.3. Effect of external pressure on fracture morphology
pressures and the comparison of grain boundaries number between
the as-received tube and the cracked tubes. It can be found that the Ductile fracture is a fracture mode that dominated by the plas-
number of LAGBs increased and the number of HAGBs decreased tic deformation. Ductile fracture in metals and alloys usually occurs
with the increase of external pressure in the transition zone. When via the nucleation, growth and coalescence of microscopy voids due
the external pressure is increased from 0 MPa to 80 MPa the frac- to the inclusions and the second phase, which is often referred to
tion of LAGBs increase from 80.8% to 86.6%. The results imply that as micro void coalescence (MVC). For the metals with good plas-
the amount of deformation in the transition zone is enhanced by ticity, such as low carbon steel, quenched and tempered steel and
increasing the external pressure. aluminum alloy, significant necking may occur before fracture, and
Fig. 13. GBs distribution under different external pressures: (a) 0 MPa, (b) 40 MPa, (c) 80 MPa, (d) comparison of GBs distribution under different external pressures.
Fig. 14. Effect of external pressure on fracture morphology: (a) 0 MPa; (b) 40 MPa; (c) 80 MPa.
the final fracture surface usually consists of microscopic dimples 5.4. Effect of external pressure on microhardness
and shear lip. For the polycrystalline material with higher ductility
such as pure aluminum and pure gold, the development of plastic Fig. 15 shows the dependence of Vickers hardness in the tran-
deformation is abundant and the percentage reduction of area can sition zone of the cracked tubes on the external pressure. It can
reach 100% which cause the occurrence of pure shear fracture (Liu be found from Fig. 15 that microhardness in the transition zone
et al., 2002). increases with the increase of external pressure from 0 to 80 MPa.
In order to reveal the effect of external pressure on the frac- When there is no external pressure applied on the tube outside,
ture behavior of 5A02 aluminum alloy tube when it is deformed the hardness of the tube in the transition zone is just 69.31 HV.
toward to the die corner, fracture surface under different external However, the hardness is increased to 71.76 HV when the exter-
pressures was surveyed using the scanning electron microscope nal pressure is 80 MPa, corresponding to an increment rate of
(SEM). Fig. 14 shows the fracture surface morphology of tubes 3.5%. In general, the deformation at room temperature for 5A02
under different external pressures, 0 MPa, 40 MPa and 80 MPa, aluminum alloys in annealed condition always experiences a sig-
respectively. Comparing the three pictures in Fig. 14, microscopic nificant strengthening with the increase of plastic strain, as shown
dimples almost exist everywhere with an average size of 30 ␮m in Fig. 4. In other word, higher strength for materials reveals that
when the external pressure is zero. When the external pres- they have experienced larger plastic deformation. Therefore, the
sure is increased to 40 MPa, the size of the microscopic dimples
is reduced to about 16 ␮m and the shear lips start to appear
on both sides in the thickness direction. Moreover, the micro-
scopic dimples only appear in the center of thickness direction of
the tube if external pressure of 80 MPa is imposed on the tube
outside, the quantity and size of the micro voids significantly
decrease.
In conclusion, the number and size of microscopic dimples,
as well as the proportion of microscopic dimples area, are all
reduced with the increase of external pressure due to the change
of the stress state. When the external pressure increase, the stress
state is transformed from approximate plane stress state to three-
dimensional stress state and the increased hydrostatic pressure
can strongly inhibit the nucleation, growth and coalescence of
micro voids. The greater the external pressure, the stronger of the
inhibitory action. Therefore, the fracture mode is changed from
MVC to pure shear fracture with the increase of external pressure
which could make the tube achieve higher plasticity. To sum up, the
fact that applying pressure on tube outside can inhibit the nucle-
ation, growth and coalescence of micro voids and thereby improve
the plastic deformation ability of the tube is confirmed in terms of
the microstructure. Fig. 15. Effect of external pressure on hardness in the transition zone.
increase of Vickers hardness caused by external pressure indicates Fundamental Research Funds for the Central Universities
that external pressure can improve the formability for the tube. (HIT.NSRIF.201134). The authors wish to express their grat-
itude to the funding. Special thanks to Mr. Peng Lin for his
6. Conclusions assistance in the microstructure analysis.
In the present work, effect of external pressure on the forma- References

bility of the tube in double-sided hydroforming in a square-section
die has been investigated from the aspects of theoretical analysis, Allwood, J.M., Shouler, D.R., 2009. Generalised forming limit diagrams showing
increased forming limits with non-planar stress states. International Journal of
experimental study and numerical simulation. Conclusions can be Plasticity 25 (7), 1207–1230.
drawn from this work as follows: Assempour, A., Nejadkhaki, H.K., Hashemi, R., 2010. Forming limit diagrams with the
existence of through-thickness normal stress. Computational Materials Science
48 (3), 504–508.
(1) Higher external pressure can increase the thinning ratio and Banabic, D., Soare, S., 2008. On the effect of the normal pressure upon the form-
the critical effective strain of the tube in the transition zone, ing limit strains. In: 7th International Conference and Workshop on Numerical
which is generally in agreement with the tendency predicted Simulation of 3D Sheet Metal Forming Processes, Interlaken, Switzerland, pp.
199–204.
by the classical plastic instability theory based on the mechan- Biyikli, E., Canadinc, D., Maier, H.J., Niendorf, T., Top, S., 2010. Three-dimensional
ical equilibrium condition in corner area. When the external modeling of the grain boundary misorientation angle distribution based on
pressure increases from 0 to 80 MPa the thinning ratio in the two-dimensional experimental texture measurements. Materials Science and
Engineering: A 527 (21–22), 5604–5612.
transition zone increases from about 12% to about 21%, corre-
Dohmann, F., Hartl, C., 1996. Hydroforming – a method to manufacture light-weight
sponding increase of critical effective strain is from 0.143 to parts. Journal of Materials Processing Technology 60 (1–4), 669–676.
0.252. Fuchs, F.J., Electric, W., Princeton, N.J., 1966, April. Hydrostatic pressure: its role in
(2) The number of LAGBs increases with the increasing external metal forming. Mechanical Engineering, 34–40.
Guven, U., 2009. Effect of the external pressure on the plastic instability of thin-
pressure in the transition zone. When the external pressure is walled tubes. Proceedings of the Institution of Mechanical Engineers, Part B:
increased from 0 to 80 MPa the fraction of LAGBs is increased Journal of Engineering Manufacture 223 (10), 1361–1365.
from 80.8% to 86.6%, which implies that the amount of defor- Hirsch, J., Al-Samman, T., 2013. Superior light metals by texture engineering:
optimized aluminum and magnesium alloys for automotive applications. Acta
mation in the transition zone is enhanced by increasing the Materialia 61 (3), 818–843.
external pressure. Jain, N., (Ph.D. dissertation) 2003. Modeling and Analysis of Dual Hydroforming
(3) With the increase of external pressure, the number and size of Process. Texas A&M University, Texas.
Jain, N., Wang, J., Alexander, R., 2004. Finite element analysis of dual hydroforming
microscopic dimples, as well as the proportion of its area, are processes. Journal of Materials Processing Technology 145 (1), 59–65.
all reduced due to the change of the stress state. Microscopic Jain, N., Wang, J., 2005. Plastic instability in dual-pressure tube-hydroforming pro-
dimples almost exist everywhere with an average size of 30 ␮m cess. Journal of Materials Processing Technology 47 (12), 1827–1837.
Lang, L.H., Wang, Z.R., Kang, D.C., Yuan, S.J., Zhang, S.H., Danckert, J., Nielsen, K.B.,
when the external pressure is zero, however, they only appear 2004. Hydroforming highlights: sheet hydroforming and tube hydroforming.
in the center of thickness direction of the tube if external pres- Journal of Materials Processing Technology 151 (1–3), 165–177.
sure of 80 MPa is imposed on the tube outside, the quantity and Lewandowski, J.J., Lowhaphandu, P., 1998. Effects of hydrostatic pressure on
mechanical behaviour and deformation processing of materials. International
size of the micro voids significantly decrease.
Materials Reviews 43 (4), 145–187.
(4) When there is no external pressure applied on the tube outside, Liu, G., Yuan, S., Teng, B., 2006. Analysis of thinning at the transition corner in tube
the hardness of the tube in the transition zone is just 69.31 hydroforming. Journal of Materials Processing Technology 177 (1–3), 688–691.
HV. However, the hardness is increased to 71.76 HV when the Liu, J., Wang, Z., Meng, Q., 2012. Numerical investigations on the influence of
superimposed double-sided pressure on the formability of biaxially stretched
external pressure is 80 MPa, corresponding to an increment rate AA6111-T4 sheet metal. Journal of Materials Engineering and Performance 21
of 3.5%, which means that the external pressure could increase (4), 429–436.
the deformation degree of the tube and make the tube achieving Liu, R., Liu, W., Liu, J., 2002. Mechanical Properties of Engineering Materials. Harbin
Institute of Technology Press, Harbin.
a greater work hardening. Matin, P.H., Smith, L.M., 2005. Practical limitations to the influence of through-
thickness normal stress on sheet metal formability. International Journal of
The conclusions summarized above reveal that external pres- Plasticity 21 (4), 671–690.
Mellor, P.B., 1962. Tensile instability in thin-walled tubes. Proceedings of the Institu-
sure applied on the tube outside surface could increase the tion of Mechanical Engineers, Part C: Journal of Mechanical Engineering Science
deformation ability of the transition zone in double-sided tube 4 (3), 251–256.
hydroforming in a square-section die. In further research, free Nurcheshmeh, M., Green, D., 2012. Influence of out-of-plane compression stress on
limit strains in sheet metals. International Journal of Material Forming 5 (3),
expansion of tube under integrated action of internal pressure, 213–226.
external pressure and/or axial force will be processed, and higher Smith, L.M., Ganeshmurthy, S., Alladi, K., 2003a. Double-sided high-pressure tubular
external pressure will be imposed on the tube outside, in order hydroforming. Journal of Materials Processing Technology 142 (3), 599–608.
Smith, L.M., Averill, R.C., Lucas, J.P., Stoughton, T.B., Matin, P.H., 2003b. Influence
to investigate the deformation behavior of the tube under three-
of transverse normal stress on sheet metal formability. International Journal of
dimensional stress state. Plasticity 19 (10), 1567–1583.
Wang, Z., (Ph.D. dissertation) 2009. Study on the Effects of Normal Pressure Loading
Mode on Sheet Metal Plastic Deformation Behavior. Harbin Institute of Technol-
Acknowledgments
ogy, Harbin.
Yang, C., Ngaile, G., 2008. Analytical model for planar tube hydroforming: prediction
This study was financially supported by the National Nat- of formed shape, corner fill, wall thinning, and forming pressure. International
ural Science Foundation of China (No.: 50975061) and the Journal of Mechanical Sciences 50 (8), 1263–1279.
14
Available online at www.sciencedirect.com
ScienceDirect
Procedia CIRP 23 (2014) 1 – 6
5th CATSConference
2014 - CIRP
onConference on Assembly Systems
Assembly Technologies and Technologies
and Systems
Electromagnetic joining of hybrid tubes for hydroforming

Verena Psyk*, Thomas Lieber, Petr Kurka, Welf-Guntram Drossel
Fraunhofer Institute for Machine Tools and Forming Technology, Reichenhainer Strasse 88, 09126 Chemnitz, Germany
* Corresponding author. Tel.: +46-371-5397-1731; fax: +49-371-5397-61731. E-mail address: [email protected]
Abstract
Joining of functionally adapted and weight-optimized semi-finished parts by electromagnetic compression is investigated taking into account
that the resulting component is further processed by hydroforming. For this purpose both ends of an aluminum tube are connected to steel
tubes. During the subsequent hydroforming of the hybrid tube the joints are exposed to extreme loading. Finite-Element-Simulation is applied
for characterizing and quantifying these loads. The electromagnetic joining process is analyzed, the joints are characterized by microstructural
analysis and the resulting properties of the joint are evaluated by model experiments.
© 2014 Published
© The Authors.by Published by Elsevier
Elsevier B.V. B.V.
This is an open access article under the CC BY-NC-ND license
Selection and peer-review under responsibility of the International Scientific Committee of 5th CATS 2014 in the person of the Conference
(http://creativecommons.org/licenses/by-nc-nd/3.0/).
Chair Prof.
Selection Dr.peer-review
and Matthias Putz [email protected].
under responsibility of the International Scientific Committee of 5th CATS 2014 in the person of the Conference
Chair Prof. Dr. Matthias Putz [email protected]
Keywords: Electromagnetic forming, joining, hydroforming
1. Introduction and motivation

1.1. Hydroforming as a lightweight construction technology
Against the background of climate change and global
Hydroforming is a technology with high potential
warming, social pressure as well as legal restraints force
considering forming of complex shapes from tube and sheet
industry not only to consider technological and economic
metal material as it is impressively shown by the application
aspects but also to attach great importance to the ecological
examples presented in [5-7].
tenability of new developments [1]. This concerns especially –
According to [8], the overall success of hydroforming
although not exclusively – the automotive industry. According
products significantly depends on the applied semi-finished
to WALTL, the reduction of CO2-emissions is one of the most
parts. In principle, tubes and hollow-profiles made of different
important challenges in automotive manufacturing and the
materials including typical lightweight materials as e.g.
consequent implementation of lightweight construction
aluminum alloys, high-strength steels, and under certain
concepts is a key strategy to cope with this demand [2].
conditions even magnesium alloys can be used, but maximum
Maximum weight reduction can only be achieved if all
benefit with regard to lightweight design can be achieved by
aspects of lightweight strategies including material as well as
applying tailored semi-finished parts. This means that
construction-related aspects are ideally exploited. As a result,
optimized distribution of the material and the material
the material distributions as well as material properties have to
properties in the finished component are achieved by applying
be ideally adapted to the according load profile and function
blanks or tubes with locally different thickness (tailored
of the component [3]. Frequently, this can only be achieved by
blanks), locally different grades (tailored heat treated blanks),
combining different materials – as it is e.g. done in the
or even locally different materials (hybrid blanks), so that in
multimaterial body of the Audi A3 [4] – and increasing the
contrast to the application of conventional material with
geometric complexity of the components. This leads to high
homogenous properties local oversizing is avoided [3].
demands on forming and joining technologies.
Tailored tubes for tube hydroforming can be round seam
welded or longitudinally seam welded [8].
2212-8271 © 2014 Published by 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/).
Selection and peer-review under responsibility of the International Scientific Committee of 5th CATS 2014 in the person of the Conference
Chair Prof. Dr. Matthias Putz [email protected]
doi:10.1016/j.procir.2014.10.063
2 Verena Psyk et al. / Procedia CIRP 23 (2014) 1 – 6
1.2. Electromagnetic forming and magnetic pulse welding of this special process combination. However, in contrast to
[17], where form-fit joints are preferred, the focus is set on
Welding of different materials is frequently related to magnetic pulse welding, here, because it is expected that the
severe problems or might even be completely impossible. In loads acting on the joint during the hydroforming process are
many of these cases, impact welding processes as magnetic extremely high and the according demands on the connection
pulse welding offer an interesting alternative. Magnetic pulse cannot be fulfilled by interference-fit or form-fit joints.
welding is a process variant of electromagnetic forming
frequently using the compression setup and resulting in a
2. Description of the manufacturing task
metallically bonded joint [9].
For electromagnetic compression a tubular workpiece is
For the analysis, an aluminum tube ((EN AW-6060-O;
positioned in a cylindrical inductor. Due to a sudden discharge
ø76 mm x 2.5 mm; length: 340 mm) shall be connected at
of a connected capacitor, damped sinusoidal current flows
both sides to steel tubes (St34 mod; ø76 mm x 2.5 mm; length:
through the inductor, inducing an according magnetic field
290 mm). The resulting hybrid tube part shall be mechanically
and a current in the tube– provided that the latter is made of
preformed (crash formed) to allow insertion into the
highly electrically conductive material. Due to the interactions
hydroforming die for the final forming with pressurized fluid.
of the currents and the magnetic field, Lorentz forces result
The welding zone has to withstand significant cross section
and if the resulting stresses in the material reach the yield
deformations during crash forming and for hydroforming leak
stress, workpiece compression is caused. During this
tightness of the connection has to be guaranteed. Considering
deformation the workpiece is accelerated within several
lightweight design aspects, the overlap of the joining partners
microseconds and can achieve velocities in the magnitude of
should be as short as possible. Here, a length of 10-30 mm is a
102 m/s and strain rates in the magnitude of 103 to 104 s-1 [10].
realistic value.
For welding tubes by electromagnetic compression, the
ends of the two joining partners are positioned coaxially
oriented, overlapping each other, and with a small gap in- 3. Quantification of the weld loading
between them in the cylindrical inductor. The above
mentioned procedure causes deformation of the outer tube, so In order to estimate the loads acting in the welding zone a
that it collides with the inner one. If the collision parameters simplified finite-element-analysis (FEA) of the crash forming
(i.e. especially the impacting angle and velocity) are suitable, and the hydroforming was carried out. Thereby, only a section
a jet effect occurs sweeping away oxides and impurities. Thus, of the work piece with an axial length of 90 mm including one
the workpiece surfaces are cleaned and accordingly reactive. joint in the middle of the regarded area was considered. The
Due to the close approximation and supported by the contact tubes were modeled in a simplified way as cylindrical
pressure, atomic bonding of the surfaces results, which is components with an overlap of 10 mm and also a simplified
often – but not necessarily – related to a wavy shape of the tool geometry was considered. The latter one was derived
contact zone. [11] The advantages of magnetic pulse welding from the geometry of the real part by extruding the
as high strength of the joint and none or significantly less approximately rectangular cross section in the joint area
formation of intermetallic phases compared to thermal joining featuring a width of 98 mm, a height of 42.5 mm and a corner
have frequently been described in literature [11-15]. radius of 15 mm.
1.3. Combined processes
In several cases, electromagnetic forming technologies

have successfully been combined with other, more
conventional forming technologies as extrusion, bending, or
deep drawing in order to increase forming limits [9]. In [16]
electromagnetic compression was already used for realizing
optimized semi-finished parts for hydroforming. The
technology was applied for locally adapting the circumference
of straight or bended tubes. Only by applying such pre-
contoured components in a subsequent hydroforming
operation the desired complex part geometry could be
produced in aluminum. In [17] combined electromagnetic
joining and subsequent hydroforming is suggested, but
information about the feasibility of the process combination is
Fig. 1. 3D-FEA model for simplified numerical analysis
missing here and has also not been published elsewhere.
To fill this gap, one aim of the joint research project
The 3D-FEA-model based on volume elements is shown in
„Skalierbare Module aus Antrieb und Achse für die
Fig. 1 as well as the considered process combination (crash
Elektromobilität – ESKAM“ is using joining by
forming and hydroforming). The aluminum tube (EN-AW
electromagnetic forming for producing hybrid tubes for
6060-O) is geometrically considered as former mentioned.
hydroforming and thus proving the feasibility and the potential
The steel tube is assumed as a St34 mod material with an outer
Verena Psyk et al. / Procedia CIRP 23 (2014) 1 – 6 3
diameter of 71 mm and a thickness of 2.5 mm. This implies a

former reduction of the steel tube diameter as necessary for
the magnetic pulse welding process.
Fig. 2. Shear force to be transferred by the joint
The tube ends are considered as clamped in axial

Fig. 3. Setup for magnetic pulse welding of hybrid tubes.
direction. In the joined area the tubes are assumed as contact
zones with sticking properties, so no slip and no contact
release may occur and a direct force transition is possible. In the experiments a capacitor charging voltage of 16 kV
After crash forming the hydroforming process starts with a corresponding to a capacitor charging energy of 42 kJ was
linear ramp up of pressure up to 800 bar. applied. Tests were carried out with and without a supporting
The resulting shear force within the tube interacting zone is mandrel. In principle for both cases metallically bonded joints
displayed in Fig. 2. During crash forming compressive forces could be realized. This was shown on the basis of specimens
are acting within the joined area. Within the hydroforming that were cut from the joining area. The influence of the
process those change to tensile forces due to the increasing supporting mandrel on the geometry and the microstructure
diameter of the tubes and respectively the material flow out of can be seen in Fig. 4.
the joined area. The maximum shear force of 42 kN is
achieved at a level of 500 bar. At this certain point the contact
zone is completely formed into the die, so that no further
increase of the shear force is possible.
4. Magnetic pulse welding of hybrid tubes
To enable magnetic pulse welding of the two tubes

featuring the same diameter, the steel tubes were subject to a
preceding mechanical forming aiming at reducing the
diameter at the ends. This preforming operation was
performed by Salzgitter Hydroforming. As a consequence the
steel tubes feature significant marks, which have been
smoothed by grinding. This preparation of the joining zone
simultaneously removes oxide layers, so that the surfaces
become more reactive.
The magnetic pulse welding experiments were carried out
at the Fraunhofer Institute for Machine Tools and Forming
Technology using a pulsed power generator PS103-25 Blue
Wave by PSTproducts with a capacitance of 330 μF and an
inductance of 85 nH (including the collector and inductor
connection). The according setup for the magnetic pulse
welding includes an inductor system that was especially Fig. 4. Contour and micrographs of joints a) without and b) with a mandrel
designed for this specific joining task (compare [18]). The
inductor system was built by Poynting GmbH. The principle Comparing the contour, joining without supporting mandrel
setup is shown in Fig.3. leads to more diameter reduction even deforming the inner
steel tube remarkably. The joining zone shows significant 5.1. Tensile tests
local curvature. Consequently, these joints can be expected to
feature stronger strain hardening effects and higher geometric As a first model experiment, tensile tests of the joint were
stiffness compared to the straighter ones achieved with a carried out. The according specimens were milled from the
mandrel. Both effects necessitate higher forces during hybrid tubes. The specimen geometry is shown in Fig. 5
hydroforming in order to reverse the bulging. As a together with the most important results of these tests – i.e.
consequence, more expensive equipment will be required for relevant strength values and photos of the torn specimens
hydroforming. indicating the failure mode.
However, what is even more important concerning the
feasibility of the combined magnetic pulse welding and
hydroforming is the consideration of the accumulated strain in
the hybrid tube over the complete production chain. The more
deformation is introduced in the mechanical preforming
operations and during magnetic pulse forming the less
formability is available for hydroforming (see also [16]).
Moreover, the diameter reduction introduced during
mechanical preforming and magnetic pulse welding has to be
reversed by hydroforming. Consequently, necessary
deformation during hydroforming increases if the compression
is more significant. This means that the compression should be
as slight as possible. Therefore, joining with a supporting
mandrel seems to be more appropriate considering the
influence of the different geometries on the parameters of the
subsequent hydroforming process as well as on the formability
of the hybrid tube.
Additionally, the microstructure of the steel tube clearly
shows an influence related to the application of a supporting
mandrel. In case of joining without mandrel the grain structure
close to the contact surface is significantly finer compared to
that farer away from the contact surface. In case of joining
with a supporting mandrel this effect was not observed. Due to
the extreme and local occurrence of the grain refinement, it is
not likely that higher deformations due to the more severe
compression and the curvature of tube surface in the joining
area are the reason for the phenomenon. However, the Fig. 5. Tensile tests of joints a) without and b) with a mandrel
microstructure reminds of a surface that has been hardened by
hammering. The most important conclusion from these tests is that,
The difference in the microstructures of the specimens with the exception of very few specimens, failure typically
joined with and without a mandrel suggests that also the occurred in the aluminum base material. This shows that the
impacting process develops differently. The unsupported steel endurable stress in the joint area is higher than the tensile
tube gives way to the moving aluminum tube by being stress of the aluminum tube, which is slightly below 100 MPa,
compressed itself. Thereby, the deceleration of the aluminum here. In case of the tubes welded with supporting mandrel,
tube is related to the deformation – i.e. at first the acceleration visual inspection of the joint area shows no damage or
and later the deceleration of the steel tube. This might include deterioration in this zone. For the specimens joined without
multiple small impacting processes similar to hammering the supporting mandrel, a slight detachment of the aluminum and
surface. In contrast, the steel tube that is supported by the the steel tube can be observed at the end of the joint. This
mandrel cannot be deformed significantly. In this case the phenomenon occurs already early in the process at a stress in
aluminum tube impacts with the steel tube and is stopped the range of 15-35 MPa and is related to a temporary drop of
abruptly. Consequently, the microstructure is similar to cases, the transferable tensile force and the according stress.
in which tubes are magnetic pulse welded to massive inner However, after overcoming this critical load, the joint strength
parts [19, 20]. is again higher than the tensile stress of the aluminum
material. Nevertheless, the joints produced with supporting
mandrel turn out to be better compared to the specimens
5. Estimation of the feasibility of the process chain on the joined without support.
basis of model experiments Apart from this general conclusion, the results of the
tensile tests can also be exploited for a first feasibility
If and how these differences influence the joint quality was estimation of the combined process. An important output
investigated by suitable model experiments. These tests also value from the numerical simulation, presented in section 3,
serve as a basis for a first estimation of the feasibility of was that during crash forming and subsequent hydroforming
hydroforming of magnetically pulse welded hybrid tubes. the joint has to withstand a maximum total force of
Verena Psyk et al. / Procedia CIRP 23 (2014) 1 – 6 5
approximately 42 kN. Assuming that this force is detachment of the steel and the aluminum tube might have
homogeneously distributed along the circumference of tube occurred (compare Fig. 6a, last forming step).
and joint, respectively, a circumferential section of 10 mm
length, corresponding to the width of the tensile specimens,
has to transfer a maximum force of 1.9 kN. This value
corresponds more or less to the yield stress of the aluminum
and the joined specimens, respectively. The maximum
transferable force is in the range of 2.5 kN. This suggests that
under the precondition that the hydroforming geometry can be
achieved with the regarded aluminum material, the joint can
be expected to withstand the load.
5.2. Cross section deformation
In order to evaluate the performance of the joint zone

during crash forming, a model experiment representing this
load case was developed. Similar to the numerical
simulations, a simplified tool geometry was considered.
However, the corner radii of the rectangular cavity were
disregarded here to further simplify the model experiment.
Since corner filling is achieved during pressurization in the
hydroforming process and not during crash forming this
simplification is reasonable.
The tool, consisting of a flat bottom, straight rectangular
walls with an adjustable distance in-between them, and an
open upper side, is mounted to a press. Wall distance was set
to 98 mm. During the press stroke, a flat punch dips into and
thus closes the cavity. The height of the resulting rectangle
depends on the press stroke. During the experiments it was
successively decreased in order to show the course of the
deformation and its influence on the joining area. The
minimum regarded height of 44 mm is in good agreement
with the real part’s cross section height in the joining zone.
Considering the initial outer diameter of the hybrid tube of
76 mm, this corresponds to a press stroke of 32 mm. In the
experiments, tube sections with a length of 195 mm were
tested. The joining zone was located in the middle of these Fig. 6. Cross section deformation tests of specimens a) joined without and
b) joined with a mandrel
specimens.
Fig. 6 shows the axial view into the hybrid tube at different
For a first estimation of the suitability of such deformed
forming stages as well as the joint after the last forming step.
hybrid tubes for being further processed by hydroforming, the
Analogues to the investigations shown above, hybrid tubes
watertightness of the specimens was tested. Therefore, one
joined with and without supporting mandrel are compared.
end of the hybrid tube was sealed before filling it with water
As it was already predicted by the simulation (compare
(no pressurization). The tube joined with supporting mandrel
Fig. 1 Crash forming) the tube buckles at the long sides of the
turned out to be watertight even after maximum deformation
rectangle during the cross section deformation. With
of its cross section, the tube joined without support showed
increasing press stroke and accordingly increasing
significant leakage in the region, where detachment of
deformation this effect becomes more pronounced. Moreover,
aluminum and steel tube was already assumed. This supports
it can be seen, that the effect is not absolutely symmetric.
the better evaluation of the tubes joined with mandrel already
Considering the aluminum tube, the buckling is more
made on the basis of the micrographic analysis and the tensile
significant at the lower side while buckling of the steel tube is
tests. It clearly proofs that for magnetic pulse welding of
more pronounced at the upper side. Higher wall-thickness and
hybrid tubes for hydroforming such a support is absolutely
accordingly higher stiffness of the joining zone leads to less
recommended. In the regarded case the hybrid tubes joined
buckling in this area compared to the individual tubes.
without support have to be assessed as unsuitable for further
However, it cannot be completely avoided here, either.
processing by hydroforming.
Comparing the hybrid tube joined with supporting mandrel
to that joined without any support, it is remarkable that in case
of joining without mandrel the aluminum tube buckles less,
while buckling of the steel tube and of the joining zone is
stronger. The axial view into the tube suggests that a local
6. Summary Verlag Wissenschaftlicher Scripten, 2011, ISBN: 978-3-642267-32-8, pp.

25-38.
[2] Waltl, H.: Produktionstechnische Herausforderungen an den
In order to provide optimized semi-finished parts for Karosseriebau durch aktuelle Trends. In: Manufacture of Car Bodies –
hydroforming, magnetic pulse welding of hybrid tubes Sustainable and Efficient; Verlag Wissenschaftlicher Scripten, 2011,
consisting of an aluminum middle section and steel end ISBN: 978-3-642267-32-8, pp. 13-23.
sections was suggested. To estimate the feasibility of the [3] Kleiner, M.; Chatti, S.; Klaus, A.: Metal forming techniques for
subsequent forming operations, a combined numerical and lightweight construction. Journal of Materials Processing Technology 177
(2006) 2-7.
experimental approach was applied. Numerical simulations [4] AUDI AG (editor): Workshop Audi A3. Product and Technology
were carried out, assuming an ideal joint. The forces to be Communications, May 2012, www.audi-mediaservices.com (shown on
transferred during further processing of the semi-finished October 11th, 2013).
parts hydroforming were quantified via the simulation. [5] Hartl, C.: Research and advances in fundamentals and industrial
Complementing magnetic pulse welding experiments were applications of hydroforming. Journal of Materials Processing
Technology 167 (2005) 383-392.
carried out. The weld properties and especially the influence [6] Lang, L. H.; Wang. Z. R.; Knag. D. C.; Yuan, S. J. Zhang, S. H.;
of applying an inner supporting mandrel were characterized Danckert, J.; Nielsen: Hydroforming highlights: sheet hydroforming and
by micrographic investigations, tensile tests and cross section tube hydroforming. Journal of Materials Processing Technology 151
deformation tests. It was shown that applying such a mandrel (2004) 165-177.
avoids unnecessarily strong deformations of the tubular [7] Vollertsen, F.: Accuracy in process chains using hydroforming. Journal of
Materials Processing Technology 103(2000) 424-433.
aluminum steel composite as well as local grain refinement in [8] Koç, M.; Altan, T.: An overall review of the tube hydroforming (THF)
the surface of the steel tube. technology. Journal of Materials Processing Technology 108 (2001) 384-
The tensile tests have shown that the magnetically pulse 393.
welded region is stronger compared to the aluminum, leading [9] Psyk, V.; Risch, D.; Kinsey, B. L.; Tekkaya, A. E.; Kleiner, M.:
to failure in the aluminum base material in nearly all cases. Electromagnetic Forming – A Review. Journal of Materials Processing
Technology, 211 (2011) 787–829.
Comparing the forces determined in the numerical simulation [10] Beerwald, C.: Grundlagen der Prozessauslegung und -gestaltung bei der
to those measured during tensile tests suggests that the joint elektromagnetischen Umformung. Dr.-Ing.-Dissertation, TU Dortmund,
will withstand the load if the hydroforming geometry can be 2005, ISBN 3-8322-4421-2.
achieved with the regarded aluminum material. However, [11] Shribman, V.; Tomer, Y.: Magnetic puse technology for improved tube
joining with supporting mandrel again shows a benefit. joining and forming. In: Tube & Pipe Technology, pp. 91-95, 2006.
[12] Kore, S. D.; Imbert, J.; Worswick, M. J.; Zhou, Y.: Electromagnetic
The cross section deformation tests and especially a impact welding of Mg to Al sheets. Science and Technology of Welding
watertightness test performed on the deformed specimens and Joining 14(6), 549-553, 2009.
finally proofed that for magnetic pulse welding of hybrid [13] Göbl, G.; Kaspar, J.; Herrmannsdörfer, T.; Brenner, B.; Beyer, E.:
tubes for hydroforming, applying a supporting mandrel is Insights into imtermetallic phases on pulse welded dissimilar metal joints.
absolutely recommended. The tubes joined without support In: Proceedings of the 4th International Conference on High Speed
Forming – ICHSF 2010, Columbus, pp. 127-136.
turned out to be untight and consequently not suitable for [14] Psyk, V.; Gershteyn, G.; Barlage, B.; Weddeling, C.; Albuja, B.;
further forming with a fluid medium, here. For the final Brosius, A.; Tekkaya, A. E.; Bach, F.-W.: Process Design for the
feasibility proof of the suggested process combination of Manufacturing of Magnetic Pulse Welded Joints. Key Engineering
magnetic pulse welding and subsequent hydroforming, Materials Vol. 473, S. 243-250, 2011.
technological tests will be carried out in close cooperation of [15] Aizawa, T.; Kashani, M.; Okagawa, K.: Application of magnetic pulse
welding for aluminium alloys and SPCC steel sheet joints. Welding in the
Fraunhofer Institute for Machine Tools and Forming world, Vol. 49, pp. 212-222, 2005.
Technology and Salzgitter Hydroforming GmbH. [16] Psyk, V.: Prozesskette Krümmen – Elektromagnetisch Komprimieren –
Innenhochdruckumformen für Rohre und profilförmige Bauteile. Dr.-Ing.
Acknowledgements Dissertation, TU-Dortmund, 2010, ISBN 978-3-8322-9026-9.
[17] Dudziak, K.-U.; Falke, J.: Gesamtbauteil aus wenigstens zwei
Einzelbauteilen und Verfahren zu dessen Herstellung. Patent application
The presented study has been performed by the authors DE 10 2008 048 216 A1, 2008.
within the joint research project „Skalierbare Module aus [18] Psyk, V.; Scheffler, C.; Drossel, W.-G.: Werkzeugauslegung für das
Antrieb und Achse für die Elektromobilität – ESKAM“. The elektromagnetische Fügen. Zeitschrift für wirtschaftlichen Fabrikbetrieb –
project has been funded by the Federal Ministry of Education ZWF 11/2013.
and Research and the consortium has the following partners: [19] Faes, K.; Baaten, T.; De Waele, W.; Debroux, N.: Joining of Copper to
Brass Using Magnetic Pulse Welding. In: Proceedings of the 4th
Hirschvogel Automotive Group, Groschopp AG, PTS- International Conference on High Speed Forming– ICHSF 2010,
Prüftechnik GmbH, REFU Elektronik GmbH, Wilhelm Vogel Columbus, pp. 84-96.
GmbH, Salzgitter Hydroforming GmbH & Co. KG, Wilhelm [20] Elsen, A.; Ludwig, M.; Schaefer, R.; Groche, P.: Fundamentals of
Funke GmbH & Co. KG, Fachhochschule Düsseldorf, EMPT-Welding. In: Proceedings of the 4th International Conference on
Fraunhofer Institute for Machine Tools and Forming High Speed Forming– ICHSF 2010, Columbus, pp. 117-126.
Technology, Aalen University, and University of Stuttgart.
References
[1] Meleghy, G.; Bayer, S.: Die Automobilzulieferindustrie – Perspektiven

innerhalb einer globalen Branche im Strukturwandel am Beispiel Tower
Automotive. In: Manufacture of Car Bodies – Sustainable and Efficient;
Materials and Design 91 (2016) 411–423
15
Materials and Design
journal homepage: www.elsevier.com/locate/matdes
Evaluation of the optimal working conditions for the warm sheet

HydroForming taking into account the yielding condition
Gianfranco Palumbo ⁎, Vito Piglionico, Antonio Piccininni, Pasquale Guglielmi, Luigi Tricarico
DMMM — Politecnico di Bari, viale Japigia 182, Bari, Italy
Article history: The present work aims at determining the optimal working conditions to manufacture an aluminium component
Received 12 May 2015 by warm sheet HydroForming. The following parameters were considered: temperature, blank holder force, oil
Received in revised form 22 October 2015 pressure; while, as output variables, die filling and maximum oil pressure. The optimization procedure was pre-
Accepted 28 November 2015
liminary based on numerical data. A finite element model was created and great attention was paid to the mate-
Available online 30 November 2015
rial modelling: an extensive characterization for evaluating the mechanical and deformative behaviour at
Keywords:
different temperatures was conducted; in addition, the effect of modelling different plastic behaviours was con-
Warm HydroForming sidered through the implementation of two yield criteria (the anisotropic Barlat '89 criterion and the isotropic
Al 6xxx series Von Mises one). Results from the optimization procedure indicated the room temperature as the optimal work-
Multi-objective optimization ing condition when implementing the Barlat '89 criterion, while the optimal temperature was higher (about
FEM 110 °C) when considering the isotropic formulation. It is relevant that, irrespective to the yield criterion, the
Yield criterion tuning of both the numerical models was possible, but HydroForming tests confirmed the validity only of the an-
isotropic criterion. In fact, results from the optimization procedure based on experimental data indicated the
room temperature as the optimal condition, highlighting the key role played by the yield criterion in the process
modelling.
© 2015 Elsevier Ltd. All rights reserved.
1. Introduction Starting from this scenario, it is quite clear that Al alloys can be con-
sidered the most attractive alternative to conventional carbon steel
In these last years, mass reduction and pollutant emissions limita- grades, especially for their excellent corrosion resistance. As widely re-
tion are considered the main tasks to be reached by car manufacturers. ported in literature, the main limiting drawback of Al alloys is the
The adoption of light alloys for vehicles lightweighting is one of the poor formability at room temperature, mainly due to the high percent-
most promising solutions to this problem. Hirsch demonstrated that re- age of alloying elements needed to ensure high mechanical properties
placing steel by aluminium (Al) leads to mass reductions ranging from [4]: if compared to traditional deep drawing steels, Al alloys exhibit
40 to 60% (reducing the mass of the car of 100 kg can save about 9 g (in the same working conditions) a formability reduction of almost
of CO2/km) with a subsequent improvement in fuel efficiency [1]. Car two thirds. In addition, low formability leads to the need of increasing
manufacturers have thus largely adopted Al alloys for producing auto- the number of forming steps that, in turn, makes not convenient the
motive components (bonnets, doors, front structures, wings, crash ele- massive adoption of such alloys. The abovementioned limitations can
ments and bumpers) since such a material allows a suitable be overcome adopting, for example, warm working conditions, able to
combination of improved surface quality and mechanical properties determine an improvement of the ductility due to the increased strain
(30% of weight reduction with no appreciable changing in car perfor- rate hardening [5], the reduction of the flow stress and the increase of
mance) [2]. In particular, 6xxx series Al alloys are preferred when high the material toughness [6]; the high strain rate hardening is determi-
strength and surface quality are needed, whereas 5xxx series Al alloys nant at elevated temperatures (about 300 °C) for improving the
are usually adopted when energy absorption and/or higher drawability Forming Limit Curve (FLC) of the material, while at lower temperatures
are requested [3]. (below 200 °C) the formability is also strongly affected by the strain
hardening [7]. Another solution for producing lightweight automotive
components using Al alloys is to adopt innovative stamping processes,
⁎ Corresponding author.
E-mail addresses: [email protected] (G. Palumbo),
like for example the sheet HydroForming (HF): in such a process the
[email protected] (V. Piglionico), [email protected] (A. Piccininni), punch is replaced by a fluid (oil) under pressure [8–11] and higher
[email protected] (P. Guglielmi), [email protected] (L. Tricarico). level of geometrical complexity and/or strain levels along with a more
http://dx.doi.org/10.1016/j.matdes.2015.11.113
0264-1275/© 2015 Elsevier Ltd. All rights reserved.
412 G. Palumbo et al. / Materials and Design 91 (2016) 411–423
Table 1 adopted, since it is assumed that friction influences the tube thinning
Chemical composition. during the process [15–16]. A good alternative approach is the adoption
Si % Fe % Cu % Mn % Mg % Cr % Zn % Ti % Al % of the inverse analysis, based on the minimization of the difference be-
0.67 0.42 0.23 0.10 0.92 0.18 0.04 0.07 Bal.
tween numerical and experimental data (an output parameter usually
affected by the friction): for example, Korkolis and Kyriakides compared
the experimental evolution of the axial strain of a formed tube with the
correspondent one predicted by numerical simulations in order to eval-
uniform strain distribution can be obtained. The simultaneous adoption uate the CoF value able to best fit experimental data [17].
of warm conditions and HF (Warm HydroForming, WHF) surely repre- As concerns the material modelling, it is necessary to perform an ac-
sents a logical consequence since it is able to combine the advantages curate mechanical and technological characterization of the Al alloy
coming from both the solutions. One of the drawbacks of the WHF pro- and, at the same time, to choose the most appropriate yield criterion
cess is the higher complexity in determining the working window, since (YC). The mechanical characterization can be carried out through ten-
it is affected by a larger number of variables than the process at room sile tests in order to determine flow stress curves (according to temper-
temperature, like: temperature, blank holder force (BHF) and oil pres- atures and strain rates) which usually gives a detailed description of the
sure. Despite the literature provides several examples of analytical material plastic behaviour, while the deformative behaviour can be de-
models useful to better understand such a process, the complex interac- fined evaluating FLCs; but the way (i.e. the load conditions) the material
tion among the parameters needs the adoption of a more suitable ap- enters the plastic field has to be properly chosen. The anisotropy of the
proach, for example based on a numerical model [12]. In fact, finite material can strongly affects the material plasticity, thus making the
element (FE) simulations give the possibility to investigate the process classical and more largely adopted isotropic Von Mises formulation
and the effect of the abovementioned parameters, thus avoiding the not suitable when modelling sheet forming processes. Hill, in 1948
trial and error approach in the process development stage; actually, [18], proposed a first simple anisotropic yield model in the form of a
such a technique is successfully applied to: (i) reduce the product de- quadratic function; since then, several formulations have been devel-
sign time, (ii) compare design options, and (iii) understand the way oped to account for the material anisotropy to obtain a better descrip-
the process parameters affect the manufacturing process [13]. The tion of the plastic behaviour and, at the same time, to answer to the
main drawback related to the FE approach is the need of tuning the increasing needs from the application field [19–20]. Alternatively,
model, in terms of both material implementation and tribological condi- some authors proposed new yield criteria, usually extensions of previ-
tions evaluation. In order to properly define the tribological conditions ous issued formulations, which revealed to be more suitable for the sin-
to model the interaction between blank and tools, even if the simple gle investigated case study [21]. On the other hand, great efforts were
Coulomb model is adopted, it is necessary to correctly tune the Coeffi- spent to enhance the process potentiality through several optimization
cient of Friction (CoF) according to the working conditions. Various so- procedures: generally, the most used approach is based on algorithms
lutions are reported in literature for the CoF evaluation: pin-on-disk test (in most of cases a gradient based type) which stop when the conver-
represents a well-assessed solution to evaluate the values to be imple- gence to a solution is reached [22–23]. Their limitation of leading to
mented within numerical models [14]; in other cases, the choice of local optima can be overcome by the adoption of more sophisticated so-
the friction test depends on the investigated forming process. In the lutions: genetic algorithms as well as simulated annealing algorithms
case of tube HF, for example, tube upsetting friction test is usually reveal to be more effective [24–26]. One of the most interesting
Fig. 1. Tensile test equipment: a) heating device; b) DIC acquisition feature.

G. Palumbo et al. / Materials and Design 91 (2016) 411–423 413
Fig. 2. Experimental set-up for the formability tests.
procedures is based on a statistical approach, involving the Response

Surface Methodology (RSM) for the Design of Experiment (DOE). It is
a powerful mean to reduce the number of numerical simulations, as
confirmed by the literature: for example such a methodology was
used for the optimization of the stamping process of an automotive
component or the control of the springback effects [27–30]. The present
work aims at determining, through a multi-objective optimization pro-
cedure, the working conditions for successfully manufacturing by WHF
an aluminium (AA6061-T6) benchmark component displaying varying
deformation ratios due to its geometry. The following process parame-
ters were considered: temperature, Blank Holder Force, oil pressure;
while, as output variables: the die filling and the maximum oil pressure.
The optimization procedure was at first carried out using numerical
data. A FE model was thus necessary to simulate the WHF process and
great attention was paid to the material modelling. A preliminary mate-
rial characterization aimed at evaluating the mechanical and deforma-
tive behaviour of the investigated material at different temperature
levels was carried out; in addition, the effect of modelling different plas-
tic behaviours was also investigated through the implementation of two
yield criteria (the anisotropic Barlat '89 criterion and the isotropic Von
Mises one). Friction conditions in the numerical models implementing
Fig. 3. Acquisition of a major strain map using the DIC system.
the two different yield criteria were properly tuned using the inverse
Fig. 4. The facility used for WHF tests: a) press machine (3D overview); b) main quotes of the case study.
Fig. 5. Electro-hydraulic press machine: a) main components; b) thermocouples positioning.
analysis methodology: in such a way both the FE models were able to and 220 °C) were investigated as well as three different orientations
reproduce the thickness profile experimentally measured on the with respect to the rolling direction (0°, 45° and 90°). Tensile tests
formed part. Finally, WHF tests were carried out and the optimization were performed on a 20 tons electromechanical INSTRON tensile test
procedure was repeated using data from experimental tests for valida- machine, equipped with the specifically designed furnace shown in
tion purposes. Fig. 1a, able to heat the specimen thanks to 9 radiant heaters positioned
around the specimen.
2. Material and methods The furnace allowed to monitor the whole test with an optical 3D
strain measurement system based on the Digital Image Correlation
2.1. The investigated aluminium alloy (DIC) technique. As reported in Fig. 1b, the strain field of the specimen
could be calculated using images acquired during the test; in order to
A single batch of AA6061, 0.5 mm thick, was purchased in the T6 correlate images, the surface of the specimens had to be properly pre-
condition. The chemical composition is reported in Table 1. pared painting a stochastic pattern.
It was used for the material characterization (the evaluation of the
mechanical and deformative behaviour) and for WHF experiments as 2.3. Evaluation of the deformative behaviour
well.
The deformative behaviour was determined by means of formability
2.2. Evaluation of the mechanical behaviour tests conducted at different temperature levels using a specific equip-
ment able to heat the blank (by the punch) and to acquire the strain
The mechanical behaviour was determined by means of tensile tests field during the whole test (the DIC system was embedded into the
in warm conditions. Dog bone specimens (gauge length: 100 mm; equipment). Tests were carried out using a Nakajima equipment assem-
width: 15 mm) were adopted. Three temperature levels (20 °C, 110 °C bled on the tensile test machine: in particular, the equipment is
Fig. 6. Digital Image Correlation system used for thickness reduction evaluation.
Fig. 7. Validation of the thickness measurements using the DIC system through the comparison with data measured by an ultrasonic thickness gauge.
Fig. 8. a) Numerical model; b) adopted profiles for BHF and pressure.
composed of a hemispherical punch (diameter 90 mm), a circular draw by the programmable logic control (PLC) of the press machine. The PLC
die and a blank holder ring; heating cartridges were used to heat up the uses temperature values coming from the thermocouples for activating/
punch. In addition, in order to acquire the strain field of the blank during deactivating the electric cartridges in the lower and the upper plate in
the test, the sensors of the DIC system were embedded in the equip- order minimize the difference between the set point and the actual tem-
ment. During each test the temperature in the centre of the blank perature in the blank holder and in the die. In fact, as shown in Fig. 5b,
(where rupture was expected to occur) was continuously monitored thermocouples are positioned in both the blank holder (TC2 and TC3)
by a pyrometer. Fig. 2 shows the formability equipment and its and the die (TC6): the heating phase is concluded when the tempera-
components. ture difference between the die cavity and the blank holder is smaller
A constant punch speed of 60 mm/min and an acquisition frequency than the threshold value (5 °C in the experiments of this work), thus as-
of 10 Hz were used for all tests. Both the punch and the blank were ap- suring a uniform temperature distribution. The blank to be formed was
propriately lubricated: this allowed to drastically reduce the friction and introduced into the tools (between the upper die and the Blank Holder)
forced the rupture in the central position of the specimen, as shown in once the system reached the target temperature and no lubricant was
the example reported in Fig. 3. used in the experiments.
In particular, the Analysis Methods AM3 (“Evaluation of the situation The forming fluid (oil) was introduced through the lower die at the
directly before the crack will happen”) detailed in the ISO 12004-1 same temperature of the tools; oil pressure was controlled through a
(2008) standard was implemented to calculate limit strains before the proportional electronic valve (maximum pressure: 350 bar) also man-
necking occurrence. aged by the PLC in order to reproduce any pressure profile defined by
the user. In the present work, investigated values of temperature and
2.4. Warm HydroForming facility and case study time exposure were respectively limited to 220 °C and 60 s, so that
the working condition did not alter the supply condition of the material.
WHF tests were carried out using the facility shown in Fig. 4a: it is a Thickness values in the longitudinal direction could be measured on the
2500 kN electro-hydraulic press machine (4 electric motors supported formed component adopting the methodology resumed in Fig. 6.
by additional 2 hydraulic cylinders for increasing the closing force) spe-
cifically designed and produced as a prototype.
In Fig. 4b the geometry of the investigated case study is shown: it is a
stepped die characterised by two depths cavities, respectively equal to Table 2
Adopted levels of the factors investigated by CCD plan.
12 mm and 20 mm.
The press machine is suitable for both WHF and SuperPlastic Point pmax/σy BHp0/σy BHpfin/pmax T
Forming tests, being equipped with heated tools for warm (up to Cube (upper value) 0.1 0.06 1.2 220
350 °C) and hot conditions (up to 1000 °C). Cube (lower value) 0.02 0.02 0.6 RT
According to Fig. 5a, heated tools for WHF (both the lower and the Axial (upper value) 0.1 0.06 1.2 220
upper one) contain a Nickel Alloy 600 plate embedding 6 electric car- Axial (lower value) 0.02 0.02 0.6 RT
Axial (centre) 0.06 0.04 0.9 110
tridges (9 kW in the bottom tool and 15 kW in the upper tool), managed
Table 3
Hollomon's model parameters and yield strength values according to temperature.
Temperature σY [MPa] K [MPa] n [−]
RT 280 435.2 0.081

110 °C 270 348.5 0.042
220 °C 220 243.7 0.02
integration points [31] were used for modelling the deformable blank
(initial thickness was set to 0.5 mm).
In addition, due to the symmetry of the case study, only half of the
blank was taken into account in order to reduce computational costs.
The symmetry of the blank was modelled constraining the translational
DoF in the x-direction and the rotational DoF in both y- and z-direction
of the nodes lying along the longitudinal axis [32-33]. The blank surface
was partitioned into two regions: the internal one (AOIL), on which the
oil pressure was applied, and the external one (ABHF), on which the BHF
was applied.
Fig. 9. Definition of the output parameter F%.
The blank holder force (BHF) and the oil pressure (P) were gradually
increased during the process according to the profiles plotted in Fig. 8b:
in such a way the initial value of both the blank holder force (BHF0) and
The undeformed blank, prepared as described in Section 2.1, was the pressure (P0) could be set thus avoiding inertial effects.
preliminary acquired by the DIC system; the same blank was again ac- The hardening behaviour of the investigated Al alloy was modelled
quired after the WHF test, thus obtaining the strain field of the formed using data from tensile tests; in particular, the Hollomon's model
part. (Eq. 1) was used to fit the flow stress curve in the rolling direction:
In order to check the accuracy and the effectiveness of the proposed
methodology for the thickness acquisition, data obtained through the σ ¼ Kεn : ð1Þ
DIC system were compared with the thickness values measured by
means of an ultrasonic thickness gauge. Fig. 7 shows data acquired In order to investigate the effect of adopting different yield criteria,
using both the methodologies. and according to literature [34–35], the Von Mises isotropic formulation
Since the fitting between data can be considered satisfactory, the was implemented and adopted as a reference for comparing the results
evaluation of the thickness distribution of the formed components when the anisotropy was taken into account. In particular, the Barlat '89
was carried out through the DIC system for all the analyses detailed in criterion [36], reported in Eq. 2, was used:
the present work.
Ф ¼ ajK1 þ K2 jm þ ajK1 −K2 jm þ cj2K2 jm ¼ 2σ m
Y ð2Þ
2.5. Numerical modelling where σY is the yield stress, K1 and K2 are the stress tensor invariants, a
and c are functions of the Lankford's parameters and the value of the ex-
The numerical model shown in Fig. 8a was created adopting the ponent m depends on the crystallographic structure (equal to 8 for FCC
commercial FE Explicit code LS-Dyna. The blank holder (not reported alloys). Such a criterion is not only considered suitable for modelling the
in figure) and the die were modelled as rigid surfaces while about Al behaviour but, which is fundamental for the robustness of the
10,000 Belytschko-Tsay shell elements with five through-thickness simulation, gave to the authors the possibility of determining the
Fig. 10. Overview of the modeFRONTIER workflow.

Fig. 11. a) Evolution of the longitudinal strain during a tensile test; b) normal anisotropy index according to longitudinal strain.
abovementioned constants through experimental tests. In fact, flow 2.6. Response Surface Methodology and multi-objective optimization
stress curves (along the three considered orientations) were directly
implemented in the model as well as the Lankford's parameter as func- In order to evaluate the effect of the most influencing process pa-
tion of the major strain. Rupture occurrence was numerically estimated rameters, i.e. the oil pressure, the initial and final value of the load ap-
directly implementing the FLCs [37], which were experimentally deter- plied by the blankholder (respectively, BHF0 e BHFfin) and the working
mined, too. temperature, a numerical campaign of FE simulations was scheduled ac-
Computational costs were further reduced also adopting the mass cording to a Central Composite Design (CCD) plan, being such an ap-
scaling technique, which allowed to dramatically reduce the simulation proach one of the most experimental design used in process
time. The complete tuning of the numerical model thus needed a sensi- optimization studies [38]. As factors of the CCD the maximum oil pres-
tivity analysis. In particular, results were analysed after changing both sure (pmax), the initial blank holder pressure (BHp0), the final (BHpfin)
the mass scaling and the element size and compared to experimental blank holder pressure and the temperature were considered. The max-
data. imum oil pressure was expressed as a percentage of the material yield
The minimum time increment was chosen in order to avoid any in- stress (σy), while the initial and the final (BHpfin) blank holder pressure
ertial effect (the kinetic energy resulted negligible with respect to both were normalized using σy and pmax, respectively. The initial (BHF0) and
the internal and the total energy). final (BHFfin) values of the BHF could consequently be calculated. The
Simulations using element dimensions ranging from 1 mm to 3 mm investigated temperature levels were the same used for the alloy
were also run in order to evaluate the effect of the mesh size on both the characterization.
thickness distribution and the failure prediction: even if a mesh size of The adopted CCD scheme, as listed in Table 2, was characterized by
2 mm already allowed to correctly predict the real thickness distribu- investigation points located at the centre and at each vertex of the
tion measured on the part, the dimension was set to 1 mm since such hyper-cube (four factors were considered); additional axial points
a value allowed to properly predict also the failure occurrence at the (called “star points”) were located at a distance α (equal to 1.0 in the
same pressure level used in experiments. present work) from the design centre.
Fig. 12. Forming limit curves according to temperature.

Fig. 13. Variation of the thickness reduction profile when changing the CoF value while keeping all process parameters (T = 220 °C, BHF = 76–131 kN, p = 10–84 bar) and the Yield con-
dition (Barlat '89) unchanged.
Results from FE simulations were analysed in terms of response var- taken into account evaluating the anisotropic behaviour of the material
iables able to properly describe the process output: pb%, which is de- in terms of Lankford's parameters (R0, R45 R90): strains along both the
fined as the ratio between the bursting pressure and the test pressure longitudinal and transversal direction (respectively major and minor
(pburst/pmax ∗ 100) and F% (Flatness), which is defined (see Fig. 9) as strain) were monitored in three different points up to the onset of neck-
the ratio between the flat part of the component measured along the ing (dashed line in Fig. 11a).
longitudinal direction and the deepest part of the die cavity (Lc/ R-values could be thus calculated as the ratio between the minor
Ld ∗ 100). strain and the thickness strain. The evolution of the normal anisotropy
Simulation results, collected in terms of the abovementioned output as a function of the major strain is reported in Fig. 11b.
variables, were used for creating Response Surfaces (RSs). In particular, Forming limit strains were determined for different specimen geom-
the commercial software modeFRONTIER was adopted: as described by etries according to the abovementioned ISO Standard, as plotted in Fig.
the schematic workflow in Fig. 10, for each output parameter an 12.
interpolant Radial Basis Function (RBF) was adopted for correlating Experimental data were thus fitted in order to define the Forming
the 4 input variables (I.V.) to the 2 Output Variables (O.V.). Limit Curves at each investigated temperature level (a first order regres-
Each RS was determined fitting data from simulations: the whole sion line was used for fitting data in the drawing region, while a polyno-
data set was divided into two groups, the Validation Set (about 10% of mial approximating function was used for the stretch side).
the starting data set) and the Training Set (containing the remaining
90%) [39,40]. The latter set was used to initially train the RS, while the 3.2. Surfaces interaction properties (Coefficient of Friction)
Validation Set allowed to evaluate the RS performance in terms of abso-
lute residuals. The choice for the most accurate RBF algorithm was based To accurately model the interaction between the tools and the blank,
on: (i) the RS fitting capability in terms of minimizing the value of the proper Coefficient of Friction (CoF) values were determined using the
Mean Leave One Out Error (MLOOE), usually optimal if less than 0.2; inverse analysis technique. It was based on the comparison between
(ii) the evaluation of the absolute residuals calculated using the Valida- the experimental thickness distribution acquired by means of the DIC
tion Set. system in the longitudinal direction and the correspondent numerical
Obtained RSs could be used in the optimization procedure for evalu- one. The value of the CoF was determined (for each level of the working
ating optimal working conditions able to maximize the Objective Func- temperature and each yield criterion) minimizing the difference be-
tions (O.F. in Fig. 10). In particular, a multi objective genetic algorithm tween experimental and numerical thickness distributions. For exam-
was used (MOGA-II), setting a number of successive generations equal ple, the graph in Fig. 13 clarifies how the CoF at the temperature of
to 100 [41]. 220 °C was determined for the numerical model implementing the
Barlat '89 yield criterion.
3. Material characterization It is worthy of notice that the FE model is strongly affected by such a
parameter. A complete overview of the obtained results is proposed in
3.1. Mechanical and deformative behaviour of the investigated alloy Fig. 14, in which numerically predicted thickness reductions along the
longitudinal path of the formed part have been compared with the cor-
An extensive mechanical characterization in terms of tensile tests respondent experimental ones. In the same figure CoF values have been
was carried out to investigate the mechanical behaviour of the Al reported for each temperature level and YC: it is important to highlight
alloy. The strength coefficient (K), the strain hardening exponent (n) that, through the inverse analysis, it was possible to determine proper
and the yield stress at the investigated temperature levels (when setting friction conditions able to tune any numerical model and thus to predict
the strain rate to 0.0013 s−1) have been reported in Table 3. the thickness distribution on the component. But, since the isotropic
Differences in the strength properties when considering the rolling yield formulation was not able to completely describe the material be-
(0°), the transverse (90°) and the intermediate (45°) direction were haviour when working conditions changed, it was necessary to adopt
Fig. 14. Numerical/experimental comparison of the thickness distributions: a) room temperature, BHF = 102–145 kN, p = 10–52 bar; b) T = 110 °C, BHF = 94–188 kN, p = 10–120 bar;
c) T = 220 °C, BHF = 76–131 kN, p = 10–84 bar.
different CoF values according to temperature. On the contrary, when level of pb% along with the highest values of F%. On the contrary, FE sim-
the YC taking into account the material anisotropy was used, a constant ulations assuming the isotropic yielding behaviour predict the exten-
CoF value could be set at any temperature level. sion of the region characterised by the highest values of pb% up to
about 130 °C and the shifting of the region characterised by the highest
4. Results and discussion values of F% towards the range 80°–130 °C. Such an effect of the YC on
the output F% is more clearly presented in Fig. 16, where the predicted
4.1. Numerical evaluation of the optimal working range extensions at different temperature levels of the flat region of the
deepest part have been highlighted on formed components.
According to the CCD scheme, a total number of 50 simulations were It is worthy of notice that, despite the accuracy of both the FE models
run using the FE models implementing the two different yield criteria. (the one implementing the Von Mises criterion and the one using the
Fig. 15 shows the contour plots concerning the RSs created using nu- Barlat '89 formulation) in predicting the thickness distribution thanks
merical data (in terms of F% and pb%) obtained from the FE model to the fine tuning of the CoF, the output variable F% remarkably changes
implementing the Barlat '89 YC (contours in Fig. 15a and Fig. 15b) and if the material anisotropy is neglected. According to the values reported
from the FE model implementing the isotropic YC (contours in Fig. 15c in Fig. 14, the estimated CoF at 110 °C is the lowest, thus clarifying why
and Fig. 15d). such an intermediate temperature level is more suitable in order to ob-
Comparing the contour plots in Fig. 15 it may be noted that the tain a well-formed component.
choice of the YC strongly affects final results: if results from the numer- Bubble charts in Fig. 17 are useful to understand how the values of
ical model implementing the anisotropic YC are considered, setting the the output variables predicted by the optimization algorithm for certain
working temperature to the lowest values allows to obtain the highest combinations of the input variables (virtual designs, represented by
Fig. 15. Contour plots of the RSs created using data from FE simulations: a) pb% (anisotropic YC), b) F% (anisotropic YC), c) pb% (isotropic YC), d) F% (isotropic YC).
triangles) change according to the working temperature; in the same the most accurate/robust results, a set of WHF experiments was de-
graphs also the designs used for evaluating the RSs (real data, represent- signed starting from the CCD plan adopted for numerical analyses. The
ed by circles) have been plotted. incremental space filler (ISF) algorithm was adopted to define new com-
Optimal virtual designs are the designs able to maximize the two binations of the factors (I.V.): since such an algorithm is able to generate
output variables; thus, they are located along the dashed lines in Fig. an uniform distribution of designs in the input space (it considers the
17, which represent the Pareto fronts. Such fronts suggest that it is pos- existing points in the database and adds sequentially new points
sible to manufacture a well-formed (the highest F%) and sound (pb% where the minimum distance from the existing points is maximized
equal to 100) component if the temperature is set to intermediate [39]), the original design space was uniformly filled. Results from WHF
values (ranging from 80 to 110 °C) when the alloy behaviour is tests were again analysed in terms of F% and pb%. In Fig. 18 tests conduct-
modelled as isotropic. On the contrary, if the anisotropic YC is imple- ed at different temperatures but setting the same value of the maximum
mented, the optimal working condition moves towards the room pressure (pmax/σy0 = 0.05) have been presented in terms of F%.
temperature. It is important to highlight that the numerical model implementing
the anisotropic YC was able to accurately predict the sheet capability of
4.2. Experiments for the validation of virtual designs filling the die cavity: in fact, results in terms of F% agree well with the
experimental ones. Moreover, if the working temperature was in-
In order to validate numerical results and to determine which YC is creased, not only the extension of the flat region decreased, but ruptures
able to properly describe the real material behaviour and to furnish also occurred. The accuracy of the numerical model implementing the
anisotropic YC was further confirmed by the comparison between nu-
merical and experimental thickness distributions along the longitudinal
direction: as Fig. 19 shows, a good overlapping could be achieved in all
the investigated working conditions.
The abovementioned experimental data were also used to create the
RSs for both the response variables in order to repeat the optimization
procedure: results from such an optimization based on RSs created
using experimental data have been presented in Fig. 20a, which shows
the Pareto Front composed by optimal virtual designs.
It may be noted that experiments indicated as optimal the working
conditions characterised by the lowest temperature values, confirming
the accuracy of the optimization results obtained by the FE model
implementing the anisotropic YC. Finally in Fig. 20b all Pareto fronts
(both numerical and the experimental ones) have been compared. It
may be noted that the Pareto front obtained using experimental data
(only the portion of the Pareto front characterized by the best F% values
was considered) agrees well with the one obtained using the FE model
implementing the Barlat '89 YC.
The matching between the Pareto fronts obtained using RSs based
Fig. 16. Extension of the flat region of the deepest part according to temperature using on both experimental and numerical data (using the Barlat '89 YC) con-
both YC. firms the effectiveness of the proposed optimization methodology. But
Fig. 17. Results from the optimization procedure: a) isotropic YC; b) anisotropic YC).
it is evident that, in order to make the FE model reliable and to obtain

robust results from the optimization procedure, it is not sufficient to
tune the model simply using thickness distributions. According to the
work by Gedikli [35], who investigated the capability of numerical
models implementing different yield criteria to predict real forming
process, the adoption of an anisotropic YC can determine slight differ-
ences in results with respect to the Von Mises isotropic formulation, es-
pecially if a proper tuning is carried out. But the thickness reduction of
the component is strictly related to the die cavity filling [42]: as a conse-
quence, according to obtained results, the anisotropy dramatically af-
fects the optimization procedure, thus highlighting the central role
played by the YC in the HF process modelling.
It is worthy of notice that the parameters for defining the adopted
Barlat '89 YC could be determined using only data from tests carried
out by the authors. Even if such a criterion is not the state of art for
sheet metal forming simulations of Al alloys, numerical results could
Fig. 18. Numerical/experimental comparison of the F%. be validated by WHF tests conducted in the present work, thus
Fig. 19. Thinning evolutions acquired using the DIC system compared with the correspondent numerical ones: a) BHF = 111–219 kN p = 10–208 bar; b) BHF = 63–157 kN p = 10–
120 bar; c) BHF = 55–115 kN p = 10–84 bar.
Fig. 20. a) Pareto fronts obtained from the procedure based on RSs created using experimental data, b) comparison between Pareto fronts obtained using RSs based on numerical and ex-
perimental data.
confirming the effectiveness of the adopted approach for taking into ac- References
count the material anisotropy.
[1] J. Hirsch, Recent development in aluminium for automotive applications, Trans.
Nonferrous Metals Soc. China 24 (2014) 1995–2002.
5. Conclusions [2] J. Hirsch, Aluminium in Innovative Light-Weight car design, Mater. Trans. 52 (2011)
818–824.
[3] W.S. Miller, L. Zhuang, J. Bottema, A.J. Wittebrood, P. De Smet, A. Haszler, A.
In the present work the possibility of adopting an optimization pro- Vieregge, Recent development in aluminium alloys for the automotive industry,
cedure based on Response Surfaces constructed using numerical data is Mater. Sci. Eng. A280 (2000) 37–49.
proposed and verified. But the need of adopting a suitable and properly [4] S. Novotny, M. Geiger, Process design for hydroforming of lightweight metal sheets
at elevated temperatures, J. Mater. Process. Technol. 38 (2003) 594–599.
tuned numerical model was demonstrated to play a key role in order to
[5] D. Li, A. Ghosh, Tensile deformation behavior of aluminum alloys at warm forming
make the optimization procedure robust. In fact, even if an extensive temperatures, Mater. Sci. Eng. A 352 (2003) 279–286.
characterization allowed to properly model the mechanical and defor- [6] N. Abedrabbo, F. Pourboghrat, F. Carsley, Forming of aluminum alloys at elevated
temperatures. Part 1. Material characterization, Int. J. Plast. 22 (2006) 314–341.
mative material behaviour at different temperatures, results from the
[7] T. Naka, G. Torikai, R. Hino, F. Yoshida, The effects of temperature and forming speed
optimization procedure based on numerical data drastically changed ac- on the forming limit diagram for type 5083 aluminum-magnesium alloy sheet, J.
cording the adopted yielding criterion: a well-formed component could Mater. Process. Technol. 113 (2001) 648–653.
be obtained at room temperature when the anisotropic behaviour was [8] S. Mahabunphachai, M. Koç, Investigations on the forming of aluminum 5052 and
6061 sheet alloys at warm temperatures, Mater. Des. 31 (2010) 2422–2434.
taken into account, while the adoption of the isotropic formulation [9] H.U. Lucke, Ch. Hartl, T. Abbey, Hydroforming, J. Mater. Process. Technol. 115 (2001)
shifted the optimal working window towards higher temperatures 87–91.
(about 110 °C). Such a result might be related to the possibility of eval- [10] K. Siegert, M. Haussermann, B. Lösch, R. Rieger, Recent developments in
hydroforming technology, J. Mater. Process. Technol. 98 (2000) 251–258.
uating a unique value of the Coefficient Of Friction for tuning the finite [11] P. Hein, F. Vollertsen, Hydroforming of sheet metal pairs, J. Mater. Process. Technol.
element model only when implementing the Barlat '89 criterion; on 87 (1999) 154–164.
the contrary, since the isotropic yield formulation revealed, as con- [12] A. Alaswad, K.Y. Benyounis, A.G. Olabi, Tube hydroforming process: a reference
guide, Mater. Des. 33 (2012) 328–339.
firmed by literature, to be not able to properly describe the material be- [13] M. Koç, Advances in tube hydroforming-an enabling technology for low-mass vehi-
haviour when changing the working conditions, it was necessary to set cle manufacturing-material, lubrication, loading, simulation issues, and alternatives,
different Coefficient Of Friction values according to temperature. Tsinghua Sci. Technol. 9 (2004) 527–545.
[14] L.H. Lang, Z.R. Wang, D.C. Kang, S.J. Yuan, S.H. Zhang, J. Danckert, K.B. Nielsen,
Experimental tests conducted to validate the results of the optimiza- Hydroforming highlights: sheet hydroforming and tube hydroforming, J. Mater.
tion procedure using numerical data confirmed the robustness of the fi- Process. Technol. 151 (2004) 165–177.
nite element model implementing the anisotropic yield criterion. Such [15] B.J. Kim, C.J. Van Tyne, M.Y. Lee, Y.H. Moon, Finite element analysis and experimental
confirmation of warm hydroforming process for aluminium alloy, J. Mater. Process.
tests demonstrated that the tuning procedure simply based on the
Technol. 187–188 (2007) 296–299.
best fitting of numerical and experimental data by properly changing [16] M. Planack, F. Vollersten, J. Woitschig, Analysis, finite element simulation and exper-
the Coefficient Of Friction is not sufficient to define the real material be- imental investigation of friction in tube hydroforming, J. Mater. Process. Technol.
haviour. In fact, the optimization procedure repeated using data from 170 (2005) 220–228.
[17] F. Vollersten, M. Plancak, On possibilities for the determination of the coefficient of
experimental tests indicated the room temperature as the optimal con- friction in hydroforming of tubes, J. Mater. Process. Technol. 125–126 (2002)
dition: the portion of the Pareto Front characterized by the best F% 412–420.
values revealed to agree quite well with the one obtained using the FE [18] Y.P. Korkolis, S. Kyriakides, Hydroforming of anisotropic aluminum tubes: Part II
analysis, Int. J. Mech. Sci. 53 (2011) 83–90.
model implementing the Barlat '89 criterion. [19] R. Hill, The mathematical theory of plasticity, Clarendon Press, New York: Oxford
As reported in previous studies focused on the optimization of the University Press, 1983.
HF process [27,43], the isotropic yield criterion is often assumed; in ad- [20] K. Mattiasson, M. Sigvant, An evaluation of some recent yield criteria for industrial
simulations of sheet forming processes, Int. J. Mech. Sci. 50 (2008) 774–787.
dition, a huge amount of numerical simulations is needed. On the con- [21] D. Banabic, F. Barlat, O. Cazacu, T. Kuwabara, Advances in anisotropy and formability,
trary, through the proposed methodology based on the optimization Int. J. Mater. Form. 3 (2010) 165–189.
procedure using (i) a suitable FE model and (ii) proper output variables [22] S. Comsa, G. Cosovici, P. Jurco, L. Paraianu, D. Banabic, Simulation of the
hydroforming process using a new orthotropic YC, J. Mater. Process. Technol. 157–
to create Response Surfaces from a small set of numerical simulations, 158 (2004) 67–74.
optimal working conditions for the Warm HydroForming process (also [23] K.J. Fann, P.Y. Hsiao, Optimization of loading conditions for tube hydroforming, J.
validated by experiments) can be determined. Mater. Process. Technol. 140 (2003) 520–524.
[24] J.B. Yang, B.H. Jeon, S.I. Oh, Design sensitivity analysis and optimization of the
hydroforming process, J. Mater. Process. Technol. 113 (2001) 666–672.
Acknowledgements [25] N. Abedrabbo, M. Worswick, R. Mayer, I. van Riemsdijkc, Optimization methods for
the tube hydroforming process applied to advanced high-strength steels with ex-
The research activities in the present work were carried out using perimental verification, J. Mater. Process. Technol. 209 (2009) 110–123.
[26] C.F. Castro, C.A.C. António, L.C. Sousa, Optimisation of shape and process parameters
the equipments funded by Region APULIA (project acronym: in metal forging using genetic algorithms, J. Mater. Process. Technol. 146 (2004)
TRASFORMA, grant number: 28). 356–364.
[27] M. Mirzaali, S.M.H. Seyedkashi, G.H. Liaghat, H. Moslemi Naeini, K. G Shojaee, Y.H. [36] F. Barlat, K. Lian, Plastic behaviour and stretchability of sheet metal. Part I: A yield
Moon, Application of simulated annealing method to pressure and force loading op- function for orthotropic sheets under plane stress condition, Int. J. Plast. 5 (1989)
timization in tube hydroforming process, Int. J. Mech. Sci. 55 (2012) 78–84. 51–66.
[28] T. Jansson, A. Andersson, L. Nilsson, Optimization of draw-in for an automotive sheet [37] G. Palumbo, A. Piccininni, Numerical-experimental investigations on the
metal part - an evaluation using surrogate models and response surfaces, J. Mater. manufacturing of an aluminium bipolar plate for the proton exchange membrane
Process. Technol. 159 (2005) 426–434. fuel cells by warm hydroforming, Int. J. Adv. Manuf. Technol. 69 (2013) 731–742.
[29] H. Naceur, Y.Q. Guo, S. Ben-Elechi, Response Surface Methodology for design of [38] D.C. Montgomery, G.C. Runger, Applied statistics and probability for engineer, 6th
sheet forming parameters to control springback effect, Comput. Struct. 84 (2006) ed. John Wiley & Sons, 2013.
1651–1663. [39] modeFRONTIER (v. 4.5.3) User’s Manual, 2014.
[30] A. Alaswad, A.G. Olabi, K.Y. Benyounis, Integration of finite element analysis and de- [40] M.D. Buhmann, D. Martin, Radial Basis Functions: Theory and Implementations,
sign of experiments to analyse the geometrical factors in bi-layered tube Cambridge University Press, UK, 2003.
hydroforming, Mater. Des. 32 (2011) 838–850. [41] K. Deb, Multi-Objective Genetic Algorithms: Problems Difficulties and Construction
[31] LS-DYNA Theory Manual, Livermore Software Technology Corporation, August of Test Problems, Evol. Comput. 7 (1999) 205–230.
2015. [42] R. Bihamta, Q.H. Bui, M. Guillot, A. D’Amours, A. Rahem, M. Fafard, Global optimiza-
[32] LS-DYNA Keyword User’s Manual Vol.l Vers. R7.0, Livermore Software Technology tion of the production of complex aluminium tubes by the hydroforming process,
Corporation, Feb. 2013. CIRP J. Manuf. Sci. Technol. 9 (2015) 1–11.
[33] LS-DYNA Keyword User’s Manual Vol. 2 Vers. R7.0, Livermore Software Technology [43] M.H. Parsa, P. Darbandi, Experimental and numerical analyses of sheet
Corporation, Feb. 2013. hydroforming process for production of an automobile body part, J. Mater. Process.
[34] M. Zampaloni, N. Abedrabbo, F. Pourboghrat, Experimental and numerical study of Technol. 198 (2008) 381–390.
stamp hydroforming of sheet metals, Int. J. Mech. Sci. 45 (2003) 1815–1848.
[35] H. Gedikli, Cora Ö. Necati, M. Koç, Comparative investigations on numerical model-
ling for warm hydroforming of AA5754-0 aluminum sheet alloy, Mater. Des. 32
(2011) 2650–2662.
16
International Journal of Machine Tools & Manufacture 50 (2010) 753–764
International Journal of Machine Tools & Manufacture

journal homepage: www.elsevier.com/locate/ijmactool
Evaluation of effect of flow stress characteristics of tubular material on

forming limit in tube hydroforming process
Woo-Jin Song a,1, Seong-Chan Heo b,2, Tae-Wan Ku b,2, Jeong Kim b,3, Beom-Soo Kang c,n
a
Industrial Liaison Innovation Center, Pusan National University, Gumjung-ku, Busan 609-735, Republic of Korea
b
Department of Aerospace Engineering, Pusan National University, Gumjung-ku, Busan 609-735, Republic of Korea
c
ERC/NSDM, Pusan National University, Gumjung-ku, Busan 609-735, Republic of Korea
a r t i c l e in fo abstract
Article history: The material properties for the analytical and numerical simulation in sheet metal processes, especially
Received 23 December 2008 in tube hydroforming process, are generally obtained from the uniaxial tensile test of raw sheet
Received in revised form material. However, the validation of the formability and reliability of the numerical simulation for the
21 May 2010
tube hydroforming process arises from the fact that the material characteristics of tubes are different
Accepted 21 May 2010
from those of the raw sheet materials. In order to determine the most suitable material property of the
Available online 1 June 2010
tubular material for the evaluation of forming limit on the THF process, the uniaxial tensile test for the
Keywords: specimens of the raw sheet metal and the roll-formed tube and the free bulge test for the roll-formed
Tube hydroforming tubular material are carried out in this paper. The forming limit curves are also derived using plastic
Forming limit curve
instability based on three kinds of necking criteria, which are Hill’s local necking criterion for sheet and
Plastic instability
Swift’s diffuse necking criteria for sheet and tube, to describe and explain the forming limits for the roll-
Flow stress
Uniaxial tensile test formed tubular material in the THF process. In order to acquire the informative data on the forming
Free bulge test limit curves in the THF process, the loading condition of the free bulge test is controlled. The proper
band from nearly necking initiation to nearly bursting initiation has been defined for the roll-formed
tubular material in the THF process. It can be concluded that the flow stress of the tubular material
should be determined from the actual free bulge test to find the practically valuable forming limit curve
for the THF process.
& 2010 Elsevier Ltd. All rights reserved.
1. Introduction useful for the tubular material even with the same grade and
composition. In other words, flow stress, strain hardening, yield
Since tube hydroforming (THF) process has been increasingly strength, etc., which are widely used as the material properties,
attracting attention in the automotive industry for the manufac- differ between the sheet and tubular materials [3–6].
turing of various components, to improve the productivity of During the period in which the new THF process is developed
hydroformed structures, there is a need to design the actual THF to replace the conventional process or to adapt it to the
process. Recently, the numerical simulation method, especially manufacturing of specific components, it is essential to perform
the finite element method (FEM), has been widely used to predict the analytical or numerical simulation for the developed process
and estimate the formability of the THF process [1,2]. However, before the detailed design and prototyping stages [7]. In this case,
validation of the formability and reliability of the numerical the suitable material properties should be applied in the
analysis for the THF processes arises from the fact that the simulation of the newly developed hydroforming process in order
material characteristics of tubes, which are formed using sheet to confirm the results from the analysis of the THF process.
material through rolling, welding and sizing operations, are However, the mechanical data from the tubular materials are
different from those of the raw sheet metal. Therefore, any hardly acquired in the THF processes. The material properties for
material properties taken from the sheet material may not be the analysis of sheet metal forming processes are generally
obtained from the uniaxial tensile test using the specimens of the
sheet material. However, the process conditions and the state
n
Corresponding author. Tel.: + 82 51 510 2310; fax: +82 51 513 3760. of stress in the THF processes are different from those in the
E-mail addresses: [email protected] (W.-J. Song), [email protected] conventional sheet metal forming processes. Thus, in order to
(S.-C. Heo), [email protected] (T.-W. Ku), [email protected] (J. Kim), analyze accurately the THF process, the analytical or numerical
[email protected], [email protected] (B.-S. Kang).
1
Tel.: + 82 51 510 2493; fax: + 82 51 514 3690.
simulation should adopt the material properties directly mea-
2
Tel.: + 82 51 510 1531; fax: + 82 51 513 3760. sured in a test method with a similar state of stress conditions to
3
Tel.: + 82 51 510 2477; fax: + 82 51 513 3760. the THF process. Fuchizawa and Narazaki [3] had conducted the
0890-6955/$ - see front matter & 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijmachtools.2010.05.010
754 W.-J. Song et al. / International Journal of Machine Tools & Manufacture 50 (2010) 753–764
free bulge test with nonconstant bulge width and then concluded from the actual free bulge test to find the practically valuable
that in the case of titanium tube, the material properties from forming limit curve for the THF process.
both tests, bulge and uniaxial tensile test, were different. Koc et al.
[4] proposed several methods to determine the tubular material
properties by experimental and analytical approaches. Sokolowski 2. Characterization of flow stress of tubular material in THF
et al. [5] described a practical test method for determining the process
flow stress of the tube material and compared among experi-
mental, analytical and numerical results, using the flow stress of 2.1. Flow stress based on uniaxial tensile test with sheet and
the tube material. However, few previous works compared the tubular material
material properties of the tubular material with the uniaxial
tensile and the free bulge tests. In order to evaluate the material property for the tubular
The reliable evaluation of formability in the THF processes can material in the THF process, two kinds of tests were executed. The
be considered as whether fracture or necking initiation in the material tests to obtain the mechanical property were conducted
formed product occurred or not. The necking initiations are for material in Table 1. Uniaxial tensile tests for the raw sheet
characterized by localized line thinning across the surface of tube metal and the roll-formed tubular material and free bulge test for
on the scale of tube thickness. Without a reliable prediction the roll-formed tubular material are performed.
method of forming severity, normal process and material
variations inevitably lead to necking and bursting for THF
2.1.1. Uniaxial tensile test for the raw sheet material
processing. The forming limit diagram, which had been initially
The specimens of the raw sheet material before the roll-
suggested by Gensamer and matured by Keeler and Backhofen [8]
forming process were prepared from three different directions,
and Goodwin [9], provides useful empirical information on the
i.e., the rolling direction (RD), 451 from the rolling direction and
forming severity in the absence of a neck or any kinds of defects
the transverse direction from the rolling direction (TD).
[10]. The dependence of the forming limit curve on strain path is
The results of the uniaxial tensile tests are summarized in
typical of all materials including steel alloys, copper and brass, as
Table 2. Stress–strain relationships of the raw sheet material are also
reported by Kleemola and Pelkkikangas [11]. The effects on strain
shown in Fig. 1. It is noted that the yield behavior of the sheet
path inhibit its utility regarding the analysis of processes that are
material was slightly different with respect to the test, but the work
inherently non-linear, such as the THF process [12]. Stoughton
hardening behavior of the sheet material is almost the same with
[13] and Stoughton and Zhu [14] rediscovered the effect and
respect to the test direction in sheet plane. It means that the process
proposed that it is necessary to use the forming limit stress
to manufacture the sheet material, i.e., the rolling process, does not
diagram in all forming operations in order to evaluate effectively
affect the material properties with planar direction in this material.
the forming severity. The forming limit curve based on strain has
However, the total elongation on 451 from the rolling direction is the
been already well understood. These models include the diffuse
greatest one among those from three test directions.
instability criterion described by Swift [15] that appears to define
qualitatively the characteristic shape of the strain-based forming
limit curve for positive minor strains. In addition, Hill [16] 2.1.2. Uniaxial tensile test for the roll-formed tubular material
described the relationship between the r value of the material and The specimens from the roll-formed tubular material are also
angle from the through-thickness neck direction to the uniaxial prepared at three different directions, i.e., the welding direction
tension axis. Zhu et al. [17] added the moment equilibrium in the (01), 901 from the welding direction and the opposite side (1801)
plane of the sheet to this analysis, leading to two solution classes, from the welding direction in Fig. 2. The uniaxial tensile test for
one applicable to the right-hand side of forming limit curve and the tubular specimens is very carefully performed in order to
the other the reduced solution obtained by Hill. Kulkarni et al. consider the curvature in the grip zone of the tubular specimens.
[18] introduced the geometric imperfection to predict the forming The test jigs, which have the same shape of the inner and outer
limit curve of aluminum alloy tubular materials numerically and curvatures of the tubular specimens, are adopted in the uniaxial
analytically and verified with the associated experiments. The tensile test of the tubular material.
necking pressures for some kind of aluminum alloy and steel alloy
materials were well predicted with the extended stress-based
forming limit curve (XSFLC), which is characterized by the mean Table 1
and equivalent stress, by Simha et al. [19] and Sorine et al. [20] Compositions of test material.
who investigated the prediction of necking phenomena in high
Component (WzT%)
strength steel tubes with multi-axial loading conditions.
In order to determine the most suitable material property of C 0.0927
the tubular material for the evaluation of forming limit on the THF Si 0.0076
process, the uniaxial tensile test for the specimens of the raw Mn 0.5195
P 0.0098
sheet metal and the roll-formed tube and the free bulge test for S 0.0093
roll-formed tubular material are carried out in this paper. The
forming limit curves are also derived using plastic instability
based on three kinds of necking criteria, which are Hill’s local
necking criterion for sheet and Swift’s diffuse necking criteria for Table 2
sheet and tube, to describe and explain the forming limits for the Uniaxial tensile test results of the raw sheet material.
roll-formed tubular material in the THF process. In order to
Test direction Ultimate Uniform Total
acquire the informative data for the forming limit curves in the strength elongation elongation
THF process, the loading condition of the free bulge test of the (MPa) (%) (%)
tubular blank is controlled. The proper band from nearly necking
initiation to nearly bursting initiation has been defined for the RD (Rolling direction) 391.76 27.38 32.80
451 385.96 25.18 32.84
roll-formed tubular material in the THF process. It is concluded TD (Transverse direction) 385.78 24.26 31.92
that the flow stress of the tubular material should be determined
W.-J. Song et al. / International Journal of Machine Tools & Manufacture 50 (2010) 753–764 755
700 stress and strain relationship based on the converted stress and
strain points from the free bulge test. Because of the lack of
600 informative data that can describe the fully plastic zone of the
material, it is assumed that the initial strain value on the effective
500 stress and strain relationship is the same as that from the uniaxial
True Stress [MPa]
tensile test of the roll-formed tubular material.

400 The material properties determined from the free bulge test of
the tubular blank are summarized in Table 4. One is obtained from
considering the initial strain effect, and the other is obtained from
300
the material flow stress model without the initial strain effect.
Due to the effect of the initial strain value, there is a slight
200
True S-S curve : 0°[RD]
difference between two kinds of flow stress models.
True S-S curve : 45°
100
True S-S curve : 90°[TD]
3. Analytical estimation of forming limit curve on THF process
0
0.0 0.1 0.2 0.3 0.4 0.5
The tubular blank in the free bulge test can be considered a
True Strain thin-walled, close-end tube with original thickness to and radius
700 ro under the internal hydraulic pressure p and axial load
Faxial_feeding, which are applied independently. The tubular material
600 can be assumed to be thin enough for the plane stress state to be
valid. From the constitutive equation the ratios of the plastic
Effective Stress [MPa]
500 strain increment are defined by

@f ðs,YÞ
400 deij ¼ dl ð1Þ
@sij
300 where s is effective stress, l is plastic multiplier and f is the

plastic potential identified as the scalar function that defines the
200 elastic limit surface. From the Von Mises yield criterion for
isotropic materials and under the plane stress condition
Sheet Tensile Test Data (s3 ¼ sr ffi0), the effective stress for the isotropic materials can
100
Curve-Fitted Data be written as follows:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
0 s ¼ s21 þ s22 s1 s2 ð2Þ
0.0 0.1 0.2 0.3 0.4 0.5
Effective Strain where s1 and s2 are principal circumferential and axial stress. The
plastic strain increments are given by considering the normality
Fig. 1. Stress–strain relationship in uniaxial tensile test of the specimen from the
condition, the incompressibility condition and the equivalent
raw sheet material: (a) true stress-strain curve and (b) curve-fitted stress-strain
curve. work definition, and can be expressed as follows:

de 1
de1 ¼ s1 s2
The results of the uniaxial tensile test for the roll-formed s 2
tubular material are summarized in Table 3. Stress–strain

relationships of the tubular material are shown in Fig. 3. It can de 1
de2 ¼ s2 s1 ð3Þ
be observed that the strength of the tubular material in the s 2
welding direction is higher than that of the others. It is well known
that the welding in the roll-forming process affects the high de3 ¼ ðde1 þ de2 Þ
strength characteristic of the roll-formed tubular material [21]. The
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
s ffi
effective stress and strain curve, which can represent the material 4 2
de ¼ 2
de1 þ de2 þde1 de2
properties of the tubular material, has been curve-fitted from the
3
averaged true stress, except that on the welding direction.
where de1, de2 and de3 are the mean plastic strain incremental
components along the principal circumferential, axial and radial
2.2. Flow stress based on the free bulge test with the tubular material
directions, respectively. The principal stresses in the tubular
material can be derived by
The free bulge tests of the tubular blank from yielding pressure
to bursting pressure are performed in several intervals, of which pr
s1 ¼
five stages are shown in Fig. 4. The internal pressure increment t
among the intervals is controlled at about 10 bar (1 MPa). At each pr 2 pFaxial_feeding Fsealing þ Ffriction
s2 ¼ ð4Þ
test interval, the actual internal pressure, bulge tip thickness and 2prt
maximum bulge height are measured. Using these measured data
in the free bulged tubes of the tubular material, the effective where r is current value of the tube’s radius and t is wall thickness
stress and strain relationship is determined using the analytical (t{r). Assuming that principal stresses maintain constant ratios
procedures as shown in Fig. 5. and directions, the ratio of strain increments also will not change:
The empty circle points in the upper graph in Fig. 4 indicate a ¼ s2 =s1 , b ¼ de2 =de1 . As plastic instability condition, three
the converted effective stress and strain values from the free different necking criteria are adopted in this study to derive the
bulge test, and the solid line indicates the curve-fitted effective analytical forming limit in the THF process.
Opposite Area
of Welding
180°
Welding Area
0°
Jig for the grip part
Intermediate
Area
90°
Fig. 2. Tubular specimens from the roll-formed tube and the curvature-jig for the grip zone of the tubular specimens.
Table 3
Uniaxial tensile test results of the roll-formed tubular material.
Test direction Ultimate strength (MPa) Uniform elongation (%) Total elongation (%)
01 (Welding area) 500.25 9.90 14.03

901 (Intermediate area) 400.24 16.50 26.31
1801 (Opposite area of welding area) 394.26 15.83 21.36
3.1. Plastic instability based on local necking criterion 3.2. Plastic instability based on diffuse necking criterion
Hill’s local necking criterion is introduced as the plastic 3.2.1. Diffuse necking criterion for tube
instability condition and then onset of plastic instability is As the second instability condition, it is assumed that plastic
assumed to occur when the following constraint is satisfied: instability occurs when dp ¼0 and dFaxial_feeding ¼ 0 simultaneously
ds1 ¼ s1(de1 +de2). Assuming that the stress ratio remains fixed [15]. This leads to the following simultaneous constraints:
the deformation of the neck ds2 ¼ s2(de1 + de2) is also satisfied ds1 ¼ s1(2de1 + de2) and ds2 ¼ s1de1 + s2de2. From Eq. (3) and the
[22]. From Eq. (3) and the instability constraints mentioned above constraints, the variation of the principal stress compo-
previously, the variation of the principal stress can be written as nents in terms of the principal hoop stress can be written as

de ða þ1Þ de ð3þ aÞ 2
ds1 ¼ s1 2 ds1 ¼ s1
s 2 s 2

de aða þ 1Þ de
ds2 ¼ s1 2 ð5Þ ds2 ¼ 1a þ a2 s21 ð8Þ
s 2 s
Based on Eqs. (1) and (5), the variation of the effective stress is From Eqs. (1) and (8), the variation of the effective stress can be
deduced as derived as
" # " #
ð1 þ aÞ að2a1Þ2 þ2ð2aÞð1 þ aÞ
ds ¼ sde ð6Þ ds ¼ sde ð9Þ
2ð1a þ a2 Þ1=2 4ð1a þ a2 Þ3=2
Thus the following instability criterion in terms of sub-tangent Thus the plastic instability condition in terms of sub-tangent Z
Z for the isotropic materials is obtained: for the THF process is obtained as
1 1 ds ð1 þ aÞ 1 1 ds að2a1Þ2 þ 2ð2aÞð1 þ aÞ
¼ r ð7Þ ¼ r ð10Þ
ZL s de 2ð1a þ a2 Þ1=2 ZT s de 4ð1a þ a2 Þ3=2
700 700
600 600

500 500
True Stress [MPa]
400 400
300 300
200 200
100 100 Bulge Test Data
True S-S curve : 180° Curve-Fitted Data
0 0
0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5
True Strain Effective Strain
700
600
12.98MPa 14.05MPa 15.06MPa 15.97MPa 16.89MPa
500
400
300
200
100 Tube Tensile Test Data

Curve-Fitted Data 20
0
0.0 0.1 0.2 0.3 0.4 0.5
16
Internal Pressure [MPa]
Effective Strain
Fig. 3. Stress–strain relationship in uniaxial tensile test of the specimen from the
roll-formed tubular material: (a) true stress-strain curve and (b) curve-fitted
12
stress-strain curve. Bursting
8
3.2.2. Diffuse necking criterion for sheet
The third assumption is that the condition for plastic
instability is satisfied when the load reaches a maximum value 4
along both principal directions [15]: dF1 ¼0 and dF2 ¼0, with
F1 ¼ pr2p Faxial_feeding Fsealing + Ffriction and F2 ¼prl, where l is tube
length. This condition leads to the following simultaneous 0
0 1 2 3 4
constraints: ds1 ¼ s1de1 and ds2 ¼ s2de2. From Eq. (3) and the
plastic instability constraints, the variation of the principal stress Process Time [sec]
components in terms of the principal circumferential stress can be Fig. 4. Stress–strain relationship in free bulge test of the roll-formed tubular
written as material.

de 1 de h ai
ds1 ¼ s1 s2 s1 ¼ 1 s21 3.3. Limit strains and limit stresses for forming limit curve
s 2 s 2

de 1 de 1
ds2 ¼ s2 s1 s2 ¼ a a s21 ð11Þ To obtain forming limit curves in terms of strains for the above
s 2 s 2
three cases, it is necessary to introduce the critical principal
strains ec1 , ec2 along the circumferential and axial direction,
Based on Eqs. (1) and (11), the variation of the effective stress is
respectively. With the work hardening law s ¼ Kðe0 þ eÞn ,
deduced as
Eqs. (7), (10) and (13) can be written as
" #
að2a1Þ2 þð2aÞ2 1 1 ds n W
ds ¼ s de ð12Þ ¼ ¼ ¼ i ði ¼ L,T,SÞ ð14Þ
4ð1a þ a2 Þ3=2 Zi s de ðe0 þ eÞ Xi
Thus, the plastic instability criterion based on the diffuse necking Assuming the proportional loading, the equivalent strain can be
criterion for sheet is obtained as described as
s ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

1 1 ds að2a1Þ2 þ ð2aÞ2 4 2
¼ r ð13Þ e¼ e1 þ e2 2 þ e1 e2 ¼ Ye1 ð15Þ

ZS s de 4ð1a þ a2 Þ3=2 3
l
2 w
2 1(z) h
Ffriction t
Fsealing
t0 1()
Faxial_feeding 1 r0
P
2 2(z) 2(z)
Free Body Diagram Stress - State 1()
Faxial_ feeding
w
Fsealing Fsealing
P
Fsealing Ffriction
Bulge Test Configuration

Flowchart
Apply the force equilibrium

Derive the effective stress & strain
equations
Determine the forces in F.B.D Curve fit using the effective values
Calculate the strength coefficient,

Derive the principal stress &
work hardening
strain values
exponent & initial strain
Fig. 5. Analytical determination of flow stress through the free bulge test.
Table 4 From Eq. (3), sc1 and sc2 , the circumferential and axial stress
Free bulge test results of the roll-formed tubular material. when the bursting occurred can be obtained as

Mathematical model s ¼ Kðe0 þ eÞn s ¼ K en 4
1
sc1 ¼ Kðe0 þ Yec1 Þn e ec1 þ ec2

3 2
Yield strength, se ¼ 0:002 ¼ sY , (MPa) 215.18 150.95
c 4 c n

c 1 c
Strength coefficient, K (MPa) 514.66 436.87 s2 ¼ Kðe0 þ Ye1 Þ e e2 þ e1 ð17Þ
3 2
Work hardening coefficient, n 0.362 0.171
Initial strain, e0 0.0904 –
4. Experimental determination of forming limit

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi on THF process
2
where Y ¼ ð4=3Þf1 þ b þ b g. According to Eqs. (14) and (15), the
limit strains based on the plastic instability yield as 4.1. Sequential free bulge test with internal pressure and
axial feeding
ðXi nWi e0 Þ
ec1 ¼ ði ¼ L,T,SÞ
YWi ð16Þ In order to find the forming limit curve for tubular material in
ec2 ¼ bec1
the THF process, a series of free bulge tests were executed. The
free bugle tests were conducted for material in Table 1 and Fig. 6 4.2. Evaluation between necking and bursting in the
shows the configuration of the free bulge test machine and die set. tubular material
The loading sequence of the free bulge tests is represented in
Fig. 7. Axial feed displacement with the same internal pressure For the majority of all cases, the bursting was distributed
level at each loading path, which means that different loading randomly in the perimeter of the roll-formed tubular blank. Also,
histories can be assigned to the tubular material in each free bulge bursting appeared in the heat-affected zone (HAZ) near the
test, was changed, as shown Fig. 7. As the result of excessive welding zone of the tubular blank. It means that these tubular
pressurizing with the specified loading paths during the free materials have good weldability and welding condition in the roll
bulge tests, the tubular blank that occurred in bursting is forming process. In the moment that the tubular blank with
illustrated in Fig. 8. excessive internal pressure is nearly burst, the actual phenomena
w
Initial Tube
Axial Feed
Internal Pressure
Axial Feed
Free Bulge Die Axial Feeding Punch

Free Bulge Test Chamber
Fig. 6. Configuration of the free bulge test machine and die set.
30
Loading Loading Loading Loading Loading Bursting Spot

Path 1 Path 2 Path 3 Path 4 Path 5
25 Loading Loading Loading
Path 6 Path 7 Path 8
20
15
10
0
0 2 4 6 8 10 12 14
Axial Feed Displacement [mm]
Fig. 7. Loading sequences in the free bulge test.

nearly necking initiation to nearly bursting initiation was

introduced and applied to a series of results of the free bulge
test for the seam-welded tubular material.
Fig. 9 explains the proposed definition of zone from necking to
bursting occurred in the seam-welded tubular material. In case (a)
in Fig. 9, the bursting occurred due to the material defects or low
surface roughness in the opposite area of the welding zone of the
tubular material. The major strain in this case has quite a higher
value than that of the welding zone in the tubular material.
On the other hand, in case (b) in Fig. 9, the bursting occurred due
to the low strength area (HAZ) near the welding zone of the
tubular material. The major strain in this case has quite a lower
value than that of the opposite area of the welding zone in the
tubular material. Based on this bursting phenomenon of the
seam-welded tubular material, all of the measured strains
in Fig. 9 can be grouped from the nearly necking initiation
area, intermediate area, to the nearly bursting initiation area in
Fig. 10.
Loading Loading Loading Loading

5. Comments of results
Path 1 Path 2 Path 3 Path 4
5.1. Comparison between the analytical and experimental

forming limits
As mentioned previously, three kinds of necking criteria were

used to derive the analytical forming limit curve in the free bulge
condition. The analytical forming limit curves with respect to each
necking criterion were determined with the material properties in
Table 5. There are four different types of material properties: (a)
that from the uniaxial tensile test with the raw sheet metal
considering the initial strain, (b) that from the uniaxial tensile test
with the roll-formed tubular material considering the initial
strain, (c) that from the free bulge test with the roll-formed
tubular material considering the initial strain and (d) that from
the free bulge test with the roll-formed tubular material ignoring
the initial strain.
Forming limit strain curves through the analytical approach
were shown and compared with a series of free bulge test results
in Fig. 11. Especially, for the region in lower minor strain value,
the forming limit curves based on the local necking criterion are
Loading Loading Loading Loading overestimated, and for the region in higher minor strain value, the
Path 5 Path 6 Path 7 Path 8 forming limit curves from the diffuse necking criterion for sheet
have the tendency of a high degree of prediction. And for the
Fig. 8. Results of the free bulge test with respect to the specified loading paths.
material property considering the initial strain value in the free
bulge test, the forming limit curves are overestimated in all cases.
The forming limit curve from the local necking criterion using the
flow stress through the free bulge test without the initial strain
of bursting in tubular blank should occur at the low quality region can predict adequately for the region in the negative minor
of the tube, i.e., material defects, low strength area (HAZ), low strain. For the region in the positive minor strain, the forming
surface roughness, etc. In general, the seam-welded tubular limit curve from the diffuse necking criterion for sheet using
material has the tendency of different tensile strengths in the flow stress through the free bulge test without the initial
circumferential directions of the tube. Especially the zone strain can be expressed adequately. From these results, in
containing the welding line of tubular material has the highest order to find the reasonable forming limit curve for the THF
tensile strength among all of the circumferential directions of the process, it is noted that all of the aspects in the analytical forming
tube. In the case that there is a tempering process after the limit curves and the results from the free bulge test should be
welding process, this phenomenon of the seam-welded tubular considered.
material could be weakened. Thus, during the free bulge test with The forming limit stress curves are derived using the forming
the internal pressure, the opposite area of the welded part of the limit curves based on strain, which are calculated from the
tubular material is more bulged than the welding area of the analytical approach with the necking criteria and the material
tubular material due to high tensile strength in the welding area, properties. And these forming limit stress curves were compared
as shown in Fig. 9. Therefore, the major strain in the opposite area with the converted stress values from the free bulge tests of the
is usually higher than that in the welding area of the tubular tubular blank in Fig. 12. For the material properties from the free
blank. In that case, some kind of definition about the major and bulge test with the initial strain value or without, the prediction
minor strain to deduce the forming limit curve is needed. of the forming limit is fairly in good agreement with the
Therefore, the proper definition to determine the band from converted test values. Especially, in terms of the necking
Bursting Spot
Nearly Bursting Area Relatively Overbulged

Area
Nearly Necking Area
Welding Area
Heat Affected Zone
(HAZ)
Bursting on the opposite area of HAZ
Bursting Spot
Welding Area
Heat Affected Zone
(HAZ)
Nearly Necking Area
Nearly Bursting Area

Relatively Overbulged
Area
Bursting on the area near HAZ
Fig. 9. Proposed definition of zone from necking to bursting in tubular material: (a) bursting on the opposite area of HAZ and (b) bursting on the area near HAZ.
initiation in the THF process, the forming limit stress curve for the
Nearly Bursting Area material property with the initial strain value from the free bulge
test of the tubular blank agrees well with the converted test
results. However, the forming limit stress curve based on the
diffuse necking criterion for tube estimates lower in the forming
Intermediate Area severity than the rest of the necking criteria. On the other hand,
for the material properties from the uniaxial tensile test with the
initial strain value, the forming limit of processes is quite highly
predicted by comparing with the converted test values. It is noted
Nearly Necking Area that the practical valuable forming limit stress curve in the THF
process should be found in combination with the reasonable
analytical results and the converted free bulge test results for the
0.40 specified material.
0.35
5.2. Evaluation of forming limit in THF process
0.30
Major Strain [ε1]
0.25 In this section, in order to obtain the practically valuable

forming limit stress curve in the THF process, the forming limit
Loading path 1 curves and forming limit stress curves determined in the previous
0.20
Loading path 2
section are rearranged with respect to the material characteristics
Loading path 3
0.15 about the test method and the test material condition. And those
Loading path 4
Loading path 5 curves were compared with the major, minor strain from the free
0.10 Loading path 6 bulge tests and the converted stress values. From these analyses,
Loading path 7 these partially selected forming limit curve and forming limit
0.05 Loading path 8 stress curve for the material property without the initial strain
0.00 value from the free bulge test of the tubular blank are redrawn in
-0.30 -0.25 -0.20 -0.15 -0.10 -0.05 0.00 0.05 0.10 0.15 Fig. 13. The lower solid thick curve in Fig. 13(a) indicates the
Minor Strain [ε2] selected forming limit curve in combination with the local
necking criterion for sheet and the diffuse necking criterion for
Fig. 10. Description of the proposed definition of zone from necking to bursting sheet. The upper limit that predicts the bursting on the bulged
using the measured strain data from the free bulge test. tube is determined as the offset curve containing all of the points
Table 5
Comparison of material properties with respect to test method and specimen type.
Specimen Raw sheet Roll-formed tube
Test Tensile test Tensile test Free bulge test Free bulge test
n n n
Mathematical model
s ¼ Kðe0 þ eÞ s ¼ Kðe0 þ eÞ s ¼ Kðe0 þ eÞ s ¼ K en
Yield strength, se ¼ 0:002 ¼ sY , (MPa) 261.86 349.84 215.18 150.95
Strength coefficient, K (MPa) 680.71 617.44 514.66 436.87
Work hardening coefficient, n 0.235 0.210 0.362 0.171
Initial strain, e0 0.0107 0.0904 0.0904 –
Bulge Test Data Bulge Test Data

Tube Bulge Test Mat'l Property [ε0 = 0.0] Tube Bulge Test Mat'l Property [ε0 = 0.0]
Tube Bulge Test Mat'l Property Tube Bulge Test Mat'l Property
Tube Tensile Test Mat'l Property Tube Tensile Test Mat'l Property
Sheet Tensile Test Mat'l Property Sheet Tensile Test Mat'l Property
0.50 0.50
0.45 0.45
0.40 0.40
0.35 0.35
Major Strain [ε1]

Major Strain [ε1]
0.30 0.30
0.25 0.25
0.20 0.20
0.15 0.15
0.10 0.10
0.05 0.05
0.00 0.00
-0.30 -0.25 -0.20 -0.15 -0.10 -0.05 0.00 0.05 0.10 0.15 0.20 0.25 0.30 -0.30 -0.25 -0.20 -0.15 -0.10 -0.05 0.00 0.05 0.10 0.15 0.20 0.25 0.30
Minor Strain [ε2] Minor Strain [ε2]
Bulge Test Data

Tube Bulge Test Mat'l Property [0 = 0.0]
Tube Bulge Test Mat'l Property
Tube Tensile Test Mat'l Property
Sheet Tensile Test Mat'l Property
0.50
0.45
0.40
0.35
Major Strain [ε1]
0.30
0.25
0.20
0.15
0.10
0.05
0.00
-0.30 -0.25 -0.20 -0.15 -0.10 -0.05 0.00 0.05 0.10 0.15 0.20 0.25 0.30
Minor Strain [ε2]
Fig. 11. Comparison of forming limit strain curves with respect to necking criterion: (a) local necking criterion for sheet; (b) diffuse necking criterion for tube and
(c) diffuse necking criterion for sheet.
from the free bulge tests. Forming limit stress curve to predict 6. Conclusions
reasonably the necking and bursting can be shown in Fig. 13(b),
which can be calculated using the forming limit curve data based In this study, the analysis of the determination of the forming
on the strain values in Fig. 13(a). limit curve using the analytical approach and the experimental
Bulge Test Data Bulge Test Data

Yield Locus [Tube Bulge Test, ε0 = 0.0] Yield Locus [Tube Bulge Test, ε0 = 0.0]
Yield Locus [Tube Bulge Test] Yield Locus [Tube Bulge Test]
Yield Locus [Tube Tensile Test] Yield Locus [Tube Tensile Test]
Yield Locus [Sheet Tensile Test] Yield Locus [Sheet Tensile Test]
Tube Bulge Test Mat'l Property [ε0 = 0.0] Tube Bulge Test Mat'l Property [ε0 = 0.0]
Tube Bulge Test Mat'l Property Tube Bulge Test Mat'l Property
Tube Tensile Test Mat'l Property Tube Tensile Test at'l Property
Sheet Tensile Test Mat'l Property Sheet Tensile Test Mat'l Property
800 800
700 700
600 600
Major Stress [σ1]
Major Stress [σ1]

500 500
400 400
300 300
200 200
100 100
0 0
-100 0 100 200 300 400 500 600 -100 0 100 200 300 400 500 600
Minor Stress [σ2] Minor Stress [σ2]
Bulge Test Data

Yield Locus [Tube Bulge Test, ε0 = 0.0]
Yield Locus [Tube Bulge Test]
Yield Locus [Tube Tensile Test]
Yield Locus [Sheet Tensile Test]
Tube Bulge Test Mat'l Property [ε0 = 0.0]
Tube Bulge Test Mat'l Property
Tube Tensile Test Mat'l Property
Sheet Tensile Test Mat'l Property
800
700
600
Major Stress [σ1]
500
400
300
200
100
0
-100 0 100 200 300 400 500 600
Minor Stress [σ2]
Fig. 12. Comparison of forming limit stress curves with respect to necking criterion: (a) local necking criterion for sheet; (b) diffuse necking criterion for tube and
(c) diffuse necking criterion for sheet.
results was performed. In terms of analyzing the necking criteria, the strain from the free bulge tests. In order to find the reasonable
forming limit curve for the material property without the initial forming limit curve for the THF process, it is noted that all of the
strain value from the free bulge test of the tubular blank is quite well aspects in the analytical forming limit curves and the results from
fitted with the major and minor strain from a series of free bulge the free bulge test should be considered. Also the practical valuable
tests. Especially in the region of the negative value of minor strain, forming limit stress diagram from the free bulge test has been
the forming limit curve based on the local necking criterion for sheet determined using the reasonable forming limit curve from the actual
is more reasonable than the others, but in the region of the positive test results. Consequently, it is shown that the forming limit curve
value of minor strain, the forming limit curve based on the diffuse through the free bulge test can play an important role in the
necking criterion for sheet is well-fitted with the major and minor evaluation of formability for the THF processes.
Bulge Test Data

government (MEST) (No. R0A-2008-000-2 0017-0), and the partial
Forming Limit Curve of Upper Margin support by grants-in-aid for the National Core Research Center
Forming Limit Curve of Lower Margin program from MOST/KOSEF (No. R15-2006-022-02002-0).
0.50
FLD0 = 32.0%
0.45
0.40 References
0.35
[1] J. Kim, S.W. Kim, W.J. Song, B.S. Kang, Analytical and numerical approach to
Major Strain [ε1]
0.30 prediction of forming limit in tube hydroforming, International Journal of

Mechanical Sciences 47 (7) (2005) 1023–1037.
0.25 [2] S.J. Kang, H.K. Kim, B.S. Kang, Tube size effect on hydroforming formability,
Journal of Materials Processing Technology 160 (1) (2005) 24–33.
0.20 [3] S. Fuchizawa, M. Narazaki, Bulge test for determining stress–strain char-
acteristics of thin tubes, Proceedings of the Fourth International Conference
0.15 on Technology of Plasticity (1993) 488–493.
[4] M. Koc, Y. Aue-u-lan, T. Altan, On the characteristics of tubular materials for
0.10 hydroforming: experimentation and analysis, International Journal of
FLD0 = 17.1%
Machine Tools and Manufacture 41 (2001) 761–772.
0.05 [5] T. Sokolowski, K. Gerke, M. Ahmetoglu, T. Altan, Evaluation of tube
formability and material characteristics: hydraulic bulge testing of tubes,
0.00 Journal of Materials Processing Technology 98 (2000) 34–40.
-0.30 -0.25 -0.20 -0.15 -0.10 -0.05 0.00 0.05 0.10 0.15 0.20 0.25 0.30 [6] D. Rees, Plastic flow in the elliptical bulge test, International Journal of
Minor Strain [ε2] Mechanical Sciences 37 (4) (1995) 373–389.
[7] F. Dohmann, C. Hartl, Hydroforming-applications of coherent FE-simulations
to the development of products and process, Journal of Materials Processing
Bulge Test Data
Technology 150 (2004) 18–24.
Yield Locus [Tube Bulge Test, ε0 = 0.0] [8] S.P. Keeler, W.A. Backhofen, Plastic instability and fracture in sheet stretched
over rigid punches, ASM Transactions Quarterly 56 (1963) 25–48.
Forming Limit Stress Curve of Upper Margin
[9] G.M. Goodwin, Application of strain analysis to sheet metal forming in the
Forming Limit Stress Curve of Lower Margin press shop, SAE Paper 680093 (1968).
[10] M.J. Painter, R. Pearce, Instability and fracture in sheet metal, Journal of
800 Physics D: Applied Physics 7 (1974) 992–1002.
[11] H.J. Kleemola, M.T. Pelkkikangas, Effect of predeformation and strain path on
the forming limits of steel, copper and brass, Sheet Metal Industries 63
700 (1977) 559–591.
[12] S.P. Keeler, Forming limits for hydroforming and other simulations, Interna-
tional EUROPAM Conference (2000).
600 [13] T.B. Stoughton, A general forming limit criterion for sheet metal forming,
International Journal of Mechanical Sciences 42 (2000) 1–42.
[14] T.B. Stoughton, X. Zhu, Review of theoretical models of the strain-based FLD
500 and their relevance to the stress-based FLD, International Journal of Plasticity
20 (2004) 1463–1486.
Major Stress [σ1]
[15] H.W. Swift, Plastic instability under plane stress, Journal of Mechanics and
Physics of Solids 1 (1952) 1–18.
400
[16] R. Hill, On discontinuous plastic states with special reference to localized
necking in thin sheets, Journal of Mechanics and Physics of Solids 1 (1952)
19–30.
300 [17] X. Zhu, K. Weinmann, A. Chandra, A unified bifurcation analysis of sheet
metal forming limits, Journal of Engineering Materials and Technology 3
(2001) 329–333.
200 [18] A. Kulkarni, P. Biswas, R. Narasimhan, A.A. Luo, R.K. Mishra, T.B. Stoughton,
A.K. Sachdev, An experimental and numerical study of necking initiation in
aluminum alloy tubes during hydroforming, International Journal of
100 Mechanical Sciences 46 (2004) 1727–1746.
[19] C.H.M. Simha, J. Gholipour, A. Bardelcik, M.J. Worswick, Prediction of necking
in tubular hydroforming using an extended stress-based forming limit curve,
0 Journal of Engineering Materials and Technology—Transaction of the ASME
-100 0 100 200 300 400 500 600 129 (2007) 36–47.
Minor Stress [σ2] [20] M. Sorine, C.H.M. Simha, I.V. Riemsdijk, M.J. Worswick, Prediction of
necking of high strength steel tubes during hydroforming—multi-
axial loading, International Journal of Mechanical Sciences 50 (2008)
Fig. 13. Rearranged forming limit strain and stress curves for THF processes: 1411–1422.
(a) forming limit strain curve and (b) forming limit stress curve. [21] J. Kim, Y.W. Kim, B.S. Kang, S.M. Hwang, Finite element analysis for bursting
failure prediction in bulge forming of a seamed tube, Finite Elements in
Analysis and Design 40 (2004) 953–966.
Acknowledgements [22] R. Hill, in: The Mathematical Theory of Plasticity, Oxford University Press,
New York, 1983.
This work was supported by the Korea Science and Engineering
Foundation (KOSEF) NRL Program grant funded by the Korea
17
Advances in Engineering Software 42 (2011) 815–820
Advances in Engineering Software

journal homepage: www.elsevier.com/locate/advengsoft
Experimental and finite element investigation of formability and failures

in bi-layered tube hydroforming
A.G. Olabi ⇑, A. Alaswad
School of Mechanical and Manufacturing Engineering, Dublin City University, Glasnevine, Dublin-9, Ireland
Article history: Optimization of the operating conditions is one of the most significant studies in the hydroforming pro-
Received 6 March 2009 cess, which affect the forming of successful components. In this work, a finite element model was estab-
Received in revised form 27 July 2010 lished for bi-layered tube hydroforming process using ANSYS LS-DYNA pre-processor and LS-DYNA solver
Accepted 13 May 2011
in order to predict the most efficient and acceptable operating condition for certain material properties
Available online 23 June 2011
and initial blank geometry.
Experiments were conducted to check the model validation and the resultant bulge height and thick-
Keywords:
ness reduction were seen in good agreement with the numerical results. The major failure modes of the
Multi-layered
Hydroforming
process were investigated. Different sets of loading paths (relations between axial feed and internal pres-
Optimization sure) were tested and compared to each others. Applying internal pressure in advance of the axial push-
Failure ing was found to give the best formability for the process.
Loading path Ó 2011 Elsevier Ltd. All rights reserved.
Finite element analysis
1. Introduction In the hydroforming process there are two modes of defects

which may appear. If the axial feed is too high in comparison with
Hydroforming is one of the unconventional metal forming pro- the internal pressure, then there is the chance of wrinkling and
cesses which is widely used in order to form complex shapes. Com- buckling. On the other hand if the internal pressure is too high
pared to other traditional manufacturing methods (such as compared to the end axial load there would be excess thinning
stamping), hydroforming offers better part quality (tighter toler- of the tube resulting in bursting of the tube. Thus bursting, wrin-
ances, increased rigidity) with lower manufacturing costs resulting kling and buckling are the main failure modes for any tube hydro-
from reduced number of forming and assembly operations. More- forming process. In order to apply the process successfully an
over, with the advent of new technological capabilities, particu- optimum loading condition should be established to avoid all the
larly more complex and improved control systems, hydroforming instabilities and failure modes which will enable proper deforma-
has become a flexible, reliable, and hence an attractive technology. tion of the product. Hydroforming process is one of the highly com-
Nowadays, hydroforming is widely used in making tubular parts of plicated processes to analyze; the analytical methods for plastic
different configurations used in automotive and aircraft industries, deformation in hydroforming have produced limited understand-
household appliances, and other applications. ing of the process. Numerical analysis and simulation have the po-
In the hydroforming process, the material is biaxially stretched tential to provide a much deeper understanding of the process and
by applying internal pressure through the hydraulic medium. In ad- hence allow for better design of process equipment.
vanced hydroforming processes, axial load is added with the inter- Tube hydroforming process was firstly established by Grey [1]
nal pressure to improve the formability of the process. During the in early 1939 to manufacture seamless metal T and X branches;
forming action, the tube blank is placed between the two die halves Rupture of the tube was avoided by controlling the pressure and
and the tube ends are held by two plungers. There are holes through the axial force. A number of research works have been reported
both plungers and a hydraulic fluid is pumped through these holes in the literature in recent years on the hydroforming process.
in order to build up the internal pressure inside the tube blank. The Dohman and Hartl [2] found that it is essential to have the knowl-
plungers are pushed inward to provide the compressive axial load. edge of the avoidance of failure cases as well as of the behaviour of
Both the internal pressure and the axial compressive loads are ap- the tube in the tool under the compressive forces that are exerted
plied simultaneously to increase the process formability. by the machine. Ahmed and Hashmi [3] developed theoretical esti-
mation of the internal pressure, axial load and clamping for hydro-
forming of certain materials and blank geometries. Hydroforming
⇑ Corresponding author. of tubes in X and T dies was investigated by Ray and McDonald
E-mail address: [email protected] (A.G. Olabi). [4]. Numerical model was found validated when numerical results
0965-9978/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.advengsoft.2011.05.022
816 A.G. Olabi, A. Alaswad / Advances in Engineering Software 42 (2011) 815–820
were compared with the experimental ones. Both modes of failure rigid plunger using ANSYS/LS-DYNA pre-processor. By taking
were numerically investigated. advantage of symmetry, a 1/8th section of the X-branch was mod-
Bi-layered tubing is suitable for chemical use in special environ- elled (Fig. 1).
ments (sea floor piping). Another important application is the The two layers were modelled using thin shell elements with
bimetallic CRA-lined pipe [5], which has a liner pipe made of cor- the Belytschko-Tsay element formulation with shell thickness
rosion-resistant-alloy (CRA) and an outer pipe made of low-cost change option activated [7]. A mesh convergence study was carried
steel, has been utilized in oil production, nuclear power plants out to minimize mesh refinement effect on the accuracy of the
and refining industry increasingly. Islam et al. [6] studied the mul- numerical results. It was found that 1140 and 1020 quadrilateral
ti-layered tube hydroforming process. Finite element simulations mapped meshed elements for the outer layer and inner layer,
using implicit code ANSYS were performed to explain stress distri- respectively, were able to give converged solutions within a rea-
bution in the hydroformed part. The multi-sheet hydroforming sonable amount of time and hence this mesh was used throughout
with the very thin middle layer was investigated by Lang et al. the study.
[7]. The process parameters effect on the process formability was The interfaces between the two layers, outer layer and die, the
discussed experimentally and numerically. Mac Donald [8] simu- both layers and the plunger were modelled with an advanced auto-
lated cross branch bulge forming of bimetallic tubes using finite matic surface-to-surface contact algorithm. A coefficient of friction
element method. A thin layer made of stainless steel was used to of 0.57 between the two layers and 0.15 in the rest of contacts was
protect a copper tube. Alaswad and Olabi [9] investigated the effect applied. The materials properties were obtained for the two layers
of geometrical factors on bi-layered tube hydroforming of X using a tensile test of a flattened part of the tubes used for the
branch. experiments. Bilinear kinematic hardening models were used for
In this paper, a finite element model is established for bi-lay- the both layers with the following parameters (see Table 1).
ered tube hydroforming using ANSYS LS-DYNA pre-processor and The material properties used for the rigid die and the plunger
LS-DYNA solver. Experiments have been conducted using the tube were of EN21 hardened tool steel. Although the die and the plunger
hydroforming machine of Dublin City University to check the were assumed to be rigid, realistic material properties were de-
numerical model validation. Based on the studied model, both fined because these values are used by the LS-DYNA code for calcu-
types of defects have been investigated and different sets of load- lation of the contact friction and stiffness [7]. The die was
ing paths were examined. A comparison of the used loading paths constrained for all degrees of freedom and the taper end plunger
was drawn. was constrained for all degrees of freedom except for Z-translation.
2. FE analysis 2.2. Experimental validation
2.1. FE modelling Experiments were conducted on the tube hydroforming ma-

chine shown in (Fig. 2) to form bi-layered X-branch components
In metal forming analysis using finite element method, an accu- from a combination of copper and brass tubes. The power source
rate modelling of the deformed body is important in order to
Table 1
achieve an accurate solution. A finite element model was created Mechanical properties of the two layers.
for bi-layered tube hydroforming using ANSYS LS-DYNA pre-pro-
cessor and LS-DYNA solver. Outer and inner layers were numeri- Mechanical properties Outer layer (annealed brass) Inner layer (copper)
cally hydroformed in X branch die with a die corner radius of Density (gm/cc) 8.80 8.98
3 mm using different settings of loading paths. Elastic modulus (GPa) 100 105
Poisson’s ratio 0.33 0.33
The finite element model was built in four parts: (a) outer tube Yield stress (MPa) 980 220
(L = 120 mm, OD = 24 mm, ID = 22 mm), (b) inner tube (L = Tangent modulus 0.59 0.21
120 mm, OD = 22 mm, ID = 20.3 mm), (c) rigid die and, (d) taper
Fig. 1. Simulation of bi-layered tube hydroforming.

A.G. Olabi, A. Alaswad / Advances in Engineering Software 42 (2011) 815–820 817
were found to agree well as the maximum deviation was within

±5% which leads to the numerical model validation (Table. 2).
3. Failures in hydroforming process
One of the significant purposes of the hydroforming process

simulation is to get marketable good products for a certain loading
condition. In order to get acceptable components, the load path
should be controlled. The main failures of the hydroforming pro-
cess which involves axial feed and internal pressure, are wrinkling,
buckling or bursting, thinning and thickening. Due to an excessive
axial feed in comparison to the internal pressure the tube could fail
by buckling or wrinkling. Alternatively, if the pressure is too high
with respect to axial feed there would be a chance of bursting
the tube.
By applying a high internal pressure compared to the axial feed,
the material is not able to flow uniformly over the length of the
tube, because the internal pressure is neutralized by the reaction
Fig. 2. Tube hydroforming machine of Dublin City University. force of the rigid die surface in the tube, except the die cavity area.
As a result, materials in this region experience more stretch than
any other area of the tube. Due to the continuous action of the
internal pressure the elongation increases continuously as long
as the pressure exists at the internal surfaces of the tube. Because
of the application of a relatively small axial feed the material sup-
ply is not sufficient towards the die cavity, where the branch top
Internal pressure [MPa]
becomes increasingly thinner. Big thickness reduction percentages

like the one studied in (Fig. 6) leads to the tube burst.
In hydroforming processes which involve end axial and internal
pressure, it is desirable that the end axial feed over the process
time should be appropriate to allow as much material as possible
to be fed into the forming zone. Appropriate material pushed into
the deformation zone helps in maintaining the wall thickness near
the highly strained areas within the desired limits. However, the
actual value of the axial feeding in the process is limited by the
occurrence of wrinkles or buckling. Typical wrinkling failure of
Axial feed [mm] the hydroformed component was shown in (Fig. 7) by applying a
Fig. 3. Experimental and simulation load paths. high axial feed compared to the internal pressure. In this simula-
tion, the material could not flow along the die cavity due to the
low internal pressure. This would result in the tube being wrinkled
or buckled in the certain region.
The comparison is drawn between the simulation results and
the practical experimental test, which is carried out in the hydro-
forming machine. This test is performed for a high axial displace-
ment (maximum displacement of the machine) and a lower
internal pressure compared to axial feed. (Fig 8a) shows the test re-
sult in the bulge forming machine. It is clear from the figure that
the multi layer tube failed through wrinkling, folding and squeez-
ing. To represent this failure by simulation, a further solution is
undertaken by applying the same boundary conditions as were ap-
plied in the practical test. (Fig 8b) depicts the simulation result. It
demonstrates that the failure patterns are apparently the same but
in real life they occur in different manner. In the practical case,
when the tube fails due to over loading, wrinkling and folding, as
soon as a leakage appears in the system, the internal pressure
Fig. 4. Experimental result. drops instantly. But in the case of simulation, the internal pressure
keeps acting on the surfaces until the last moment of the solution
even though the structure has already failed. Due to these difficul-
for the hydraulic system is a variable displacement piston pump ties the failure results differ from practical testing to simulation.
driven by a 7.5 kW electric motor. Although the branch is severely buckled, the total displacement
To verify the numerical results, an experimental test was car- is only 17 mm. The actual displacement of the failed specimen is
ried out with the same boundary conditions. Loading path used approximately 15 mm. Therefore it can be concluded that both
for the simulation was matched with the one recorder by the Lab- the practical and simulation results are in good agreement. It can
VIEW data acquisition system during the experiment (Fig. 3). be seen that a large amount of buckling occurs and both materials
Numerical results (Fig. 4) were compared with the experimental were under very high stress. The buckling occurred as there was
ones (Fig. 5) to check the numerical model validation. Bulge height not a high enough internal pressure to withstand the high axial
and bulge tip thickness were measured for both studies. Results displacement or axial force.
Fig. 5. Numerical result.
Thus the forming loading path (relation between internal pres-

Table 2
sure and axial feed) has to be adjusted properly in order to obtain a
Branch height and bulge top thickness comparison (experiment and simulation
results). successful formed component with the hydroforming process.
Experimental Numerical Error

4. Loading paths comparison
Bulge height (mm) 8.787 8.424 +4.13
Whole thickness at the branch top 1.600 1.622 1.37
(mm) Different loading paths as shown in (Fig. 9) were applied on
multi-layer tube hydroforming process. Loading path (A)
Fig. 6. Bursting failure due to big thickness reduction in the bulge top.
Fig. 7. Wrinkle forming in bi-layered tube hydroforming process.

A.G. Olabi, A. Alaswad / Advances in Engineering Software 42 (2011) 815–820 819
Fig. 8. (a) Failure in bulge forming machine for high axial feed and (b) Failure in FE simulation.
reduction. Formability of the process was increased when pressure

advanced loading paths (B and C) are used as high bulges can be
obtained within accepted range of wrinkle height. Pressure ad-

vanced loading paths avoid forming of big wrinkles from the early
stages by applying high internal pressure during initial forming;
however thickness reduction can be critical in this kind of loading
path and should be accurately observed. Axial feeding advanced
loading paths (D and E) were found to result in low bulges with
considerably big wrinkles. Big wrinkles can be formed from the ini-
tial stage because of the big feeding without enough internal pres-
sure to form the material in the die. This kind of wrinkles cannot be
flattened in the calibration stage especially when using multi-layer
Axial Feed [mm] tubes.
Fig. 9. Different loading paths: (A) Linear, (B and C) pressure advanced and (D and
E) axial feed advanced.
5. Conclusion
represents a linear loading path, (B and C) are pressure advanced From the paper, the following conclusion points were drawn:
paths in which the hydraulic pressure was raised to a certain mag-
nitude in advance of the axial pushing, while (D and E) are the feed 1. A finite element model was developed for bi-layered tube
advanced path in which axial feed was raised in advance of the hydroforming using ANSYS LS-DYNA pre-processor and LS-
internal pressure. All the studied loading paths have the same DYNA solver. Numerical model was found to be validated when
maximum internal pressure and axial feed. compared with the experimental results.
Based on the numerical model, bulge heights, wrinkles height, 2. Two different types of failure (bursting and wrinkling) were
and thickness reduction values were recorded and a comparison presented for bi-layered tube hydroforming process.
of the applied loading paths was presented. 3. Different sets of loading paths were tested. Best formability of
From (Fig. 10) it can be found that applying linear loading path the process was achieved when applying advanced internal
(A) results in small bulge height with small wrinkle and thickness pressure loading paths.
Change of parameter
Loading path
Fig. 10. Effect of different loading paths on the hydroformed part.

References [6] Islam MD, Olabi AG, Hashmi MSJ. Feasibility of multi-layered tubular
components forming by hydroforming and finite element simulation. J Mater
Process Technol 2006;174:394–8.
[1] Grey JE. Apparatus for making wrought metal ts. Patent no 2203,868; 1939. P.
[7] Lang L, Danckert J, Nielsen KB. Multi-layer sheet hydroforming: experimental
1–40.
and numerical investigation into the very thin layer in the middle. J Mater
[2] Dohmann F, Hartl Ch. Hydroforming – a method to manufacture light-weight
Process Technol 2005;170:524–35.
parts. J Mater Process Technol 1996;60(1–4):669–76.
[8] Mac Donald BJ. Non linear finite element simulation of complex bulge forming
[3] Ahmed M, Hashmi MSJ. Three-dimensional finite-element simulation of bulge
processes. PhD thesis. School of mechanical engineering. Ireland: Dublin City
forming. J Mater Process Technol 2001;119/1–3:387–92.
University; 2000.
[4] Ray P, Mac Donald BJ. Experimental study and finite element analysis of simple
[9] Alaswad A, Olabi AG. A numerical study of the effect of geometrical factors on
X-and T-branch tube hydroforming processes. Int J Mech Sci
bi-layered tube hydroforming. In: Proceeding of 7th European LS-DYNA
2005;47:1498–518.
conference.
[5] Wang X, Li P, Wang R. Study on hydro-forming technology of manufacturing
bimetallic CRA-lined pipe. Int J Mach Tools Manuf 2005;45(4–5):373–8.
18
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 1 9 8 ( 2 0 0 8 ) 381–390
journal homepage: www.elsevier.com/locate/jmatprotec
Experimental and numerical analyses of sheet

hydroforming process for production of an
automobile body part
M.H. Parsa a,∗ , P. Darbandi b

a School of Metallurgy and Materials Engineering, University College of Engineering, University of Tehran, P.O. Box 11155/4563,
Tehran, Iran
b Department of Metallurgy and Materials Engineering, Faculty of Engineering, University of Tehran, P.O. Box 11155/4563,
Tehran, Iran
Article history: An approach for substituting conventional manufacturing method of a three pieces shell
Received 9 April 2006 fender by one piece has been proposed. This approach is based on sheet hydroforming
Received in revised form process, which has many advantages over conventional stamping processes. For produc-
21 July 2007 tion of shell fender by sheet hydroforming, two possible manufacturing procedures have
Accepted 24 July 2007 been evaluated numerically, namely; pure stretching and draw-in. In addition, results of
both finite element simulations and experiments on small size specimens show superi-
ority of draw-in procedure. The advantage of draw-in over pure stretching procedure has
Keywords: been explained using thickness variation measurements. Furthermore, the draw-in proce-
Sheet hydroforming dure of shell fender has been optimized with respect to initial blank size, fluid pressure and
Auto body parts required tools using an explicit commercial forward finite element program. Finally, the
Stamping effects of sheet material and fluid pressure on the required deformation force in laboratory
Fluid pressure scale have also been evaluated.
FEM © 2007 Elsevier B.V. All rights reserved.
1. Introduction hydroforming advantages include, increasing drawing ratio,

better surface quality, less spring back, minimizing thick-
Sheet hydroforming process is one of the new technologies ness variations of the products, and reducing tooling cost
that seems to be capable of satisfying the industrial require- especially for nonsymmetrical components, etc., that lead
ments in the field of sheet metal forming. This technique, to possibility of manufacturing complex stampings with less
in which one of the tools is replaced by a soft tool was ini- difficulties with regards to rigid tools (Zhang and Danckert,
tially proposed about two decades ago, has many advantages 1998; Zhang, 1999; Kandil, 2003; Zampaloni et al., 2003; Lang
over the conventional stamping processes. Sheet hydroform- et al., 2004; Zhang et al., 2003). For clarifying the ability of
ing has the ability to solve inherent problems and limitations this process, sheet hydroforming has been subjects of many
associated with conventional processes. Typical tools for sheet research works (Zhang et al., 2004; Yossifon and Tirosh, 1989;
hydroforming consist of a punch, a blank holder, a pressure Thiruvarudchelvan and Lewis, 1999; Thiruvarudchelvan and
chamber, and a rubber diaphragm that is used for sealing of Travis, 2003; Rimkusa et al., 2000; Ahmetoglu et al., 2004; Hein
liquid in the pressurized chamber as shown in Fig. 1. Sheet and Vollertsen, 1999; Novotny and Hein, 2001), and it has
∗
Corresponding author. Tel.: +98 21 61114069; fax: +98 21 88006076.
E-mail addresses: [email protected], m h [email protected] (M.H. Parsa).
0924-0136/$ – see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.jmatprotec.2007.07.023
382 j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 1 9 8 ( 2 0 0 8 ) 381–390
Fig. 1 – Schematic of sheet hydroforming process.
been proposed as a good candidate for production of stamp-

ings in automotive industry (Lei et al., 2000; Kang et al., 2004;
Nakamura et al., 1994).
For design analysis, application of the conventional meth-
ods of evaluations for sheet hydroforming are very expensive
and time consuming, while the numerical and simulation
Fig. 2 – Flow diagram of the simulation steps.
approach such as finite element method can enhance existing
knowledge and reduce tool cost and lead-time. This is done by
predicting formability and providing virtual tryout before tool
construction in advance (Lei et al., 2000; Kang et al., 2004). of shell fender, capabilities of two methods for production of
Since different strategies can be selected for production of shell fender have been evaluated, numerically. Based on the
any stampings, in this research it has been attempted to sim- evaluations, the best possible procedure has been selected for
ulate the production of a complete hydroformed shell fender production of the shell fender. Then the sheet hydroforming
that substitutes two complex stampings, in order to select the process of integrated part has been optimized with respect
best possible procedure. The manufacturing procedure of the to the constraint that exerted by the die and fluid pressure
mentioned part has been analyzed from the blank design to using a commercial finite element program with capability of
the final product. Then, in order to evaluate the finite ele- large deformation simulation. Also the effects of sheet mate-
ment results, experimental examinations have been carried rials on the deformation behavior have been evaluated. Since
out on the down sized specimens. In the following sections, sheet hydroforming experiments had been carried out on the
the simulation and experimental procedures and results will 1:8 ratio scale of real size, hydroforming simulations were car-
be presented. ried out on both real and small sizes. Fig. 2 illustrates, the
simulation steps as flow diagram.
In the following sections brief discussions on the new inte-
2. Sheet hydroforming simulation grated part, simplified model, simulation steps and assumed
boundary conditions, will be discussed.
Based on the past experience, it is believed that sheet
hydroforming using a punch is suitable and practical for 2.1. Model description
manufacturing of nonsymmetrical complex shell fender that
consists of two welded stampings (Parsa and Shahabizaheh, Two parts shown in Fig. 3 constitute the conventional shell
submitted for publication). For defining the best manufactur- fender. Normally three steps are required to make this spec-
ing strategy for production of complete shell fender based on imen. These are (i) laser welding and pressing to make part
numerical simulations, the shape and size of initial blank has A, (ii) stamping of part B, and (iii) finally joining of parts A, B.
been determined first. Since both stretching and draw-in pro- But by noticing capabilities of the sheet hydroforming process,
duction procedures seem to be applicable for manufacturing production of complete shell fender in one step seems to be
j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 1 9 8 ( 2 0 0 8 ) 381–390 383
Fig. 3 – Conventional stamp welded fender shell parts: (A) tailor blank and (B) cup head.
Fig. 4 – The complete integrated model.

Fig. 5 – Initial blank shape.
possible. In Fig. 4 a simplified complete model is shown. This

is the shape that considered for later simulations and exper- sheets, the maximum biaxial strains are equal to 0.65 and 0.3,
imental works. The resulted model is very complicated and respectively.
difficult for production by conventional stamping processes.
2.3. Initial blank shape and size
2.2. Used materials
Defining the initial blank shape and size is a critical step
Three kinds of sheet metals, low carbon steel sheet (St14), in stamping production design especially for complicated
IF steel sheet and 2024 aluminum sheet with thickness of shapes. It is a well-known fact that, if the initial blank shape
0.8 mm have been used. The mechanical properties of used and size are not correctly designed, successful stamping oper-
sheet metals are presented in Table 1. In the case of low car- ation will not be attained even though well-designed tools
bon steel (St14), it has been found from forming limit diagram have been applied. Therefore, for designing the initial blank
of this material that the maximum biaxial strain is equal to shape and size, an inverse finite element program was used.
0.45 (Aretz, 2004). In the case of IF steel and 2024-O aluminum The data for the shape, size and thickness of the final product
Table 1 – Mechanical properties of used sheet materials

Material Yield strength (MPa) Ultimate strength (MPa) El. (%) Hardening model K (MPa) n
St14 180 323 43 = Kε n

625 0.27
IF 146.5 439.7 60 = Kεn 580 0.32
2024-O 96 240 20 = Kεn 380 0.28
Fig. 6 – Hydroforming tool models used for simulations: (a) punch; (b) pressure chamber with guide or semi-die device; (c)
schematic assembled model.
is processed with the aid of this program, which will subse- ent boundary conditions using a commercially available FEM
quently calculate the initial blank shape and size in one step code.
(Parsa and Pournia, 2004). The shape of the predicted blank for During all of simulations, tools (punch, pressure chamber
simulation of a full size draw-in procedure is shown in Fig. 5. In components and semi-die device) were considered to be rigid,
addition, 1/8 size blank was also used for simulations of draw- while sheet metals constitute deformable material. The fluid
in procedure in order to compare the simulations’ outcomes pressure was assumed to be uniformly distributed over the
with experimental results. surface of sheet metal whose opposite surface was located in
In the case of pure stretching, maximum circular blank front of the punch. Therefore, punch movements and liquid
diameter of 1200 mm for full scale and 150 mm for 1:8 scale pressure acted in opposite direction and could be varied, sepa-
were considered, since available sheet material for deforma- rately. The fluid pressure was varied from 0 to 30 Mpa. Contact
tion plays an important role. condition between the surface of sheet and tools was defined
as the surface to surface contact based on the nonlinear con-
2.4. Tools tact strategy (Harnau et al., 2005). Also it was assumed that the
Coulomb law of friction prevailed and the coefficient of fric-
Fig. 6 illustrates the shape of tools and their arrangements tion was considered be equal to 0.01. For simulation of shell
used for forming simulation of shell fender by sheet hydro- fender production, two cases, namely stretching and draw-in
forming. The mainly considered tools consisted of fluid were considered.
pressure chamber, semi-die device or guide and punch. In all In the case of draw-in procedure based on the considered
of simulations, the tools were considered to be rigid. tools configurations, contact could be established between
punch, blank holder and one side of sheet metal surface, while
2.5. Boundary conditions movement of other side sheet metal surface that is subjected
to fluid pressure in some limited area was assumed to be
The quality of final products manufactured using sheet hydro- controlled by semi-die device. Also it was assumed that the
forming method, largely depends on the process parameters periphery of the initial blanks had the possibility of moving
such as pressure, friction, size of initial blank and position of in three directions, rotating in two directions and deforming
the punch and semi-die device. To establish the proper pro- according to the shape of punch and semi-die device.
cess parameter combinations after preliminary initial blank In the case of stretching, contact could be established
calculations, several simulations were conducted under differ- between punch, blank holder and one side of the sheet metal
surface. It was considered that periphery of the initial blank extensometer, and some accessories to control the pressure.
was fixed and there was no possibility of movement or rota- The pressure container had been designed to withstand the
tion. pressurized fluid and gave enough space for the deformed
specimen. A tightly screwed cover plate was used to close
2.6. Optimization procedures the container at the top. The punch penetrated through the
concentric hole in the cover plate. To seal the container an
After determination of the best process for production of the o-ring was used under the cover plate. Also a 2 mm flat rub-
shell fender (draw-in process) and defining applicable range of ber diaphragm was used under the cover plate for sealing the
the fluid pressure variation, forward sheet hydroforming sim- pressurized container.
ulations of the shell fender were carried out. The main aims
of these series of simulations were attempted, (i) to obtain an 3.2. Experimental procedure
optimum blank shape and size, (ii) to define the best position
of the punch, (iii) to determine the most efficient combina- At the beginning of the sheet hydroforming test, after locating
tion of fluid pressure and punch stroke and (iv) to study the the initial blank in the desired position and orientation on the
necessity of using additional tool (guide or semi-die device) for upper surface of rubber diaphragm, the cover plate was tightly
minimizing the defects that might occur in some parts of the screwed on the container and then the punch was inserted in
work piece. There are two main criteria for selecting the opti- the position. The entire test rig was placed on a die set, which
mum conditions; (a) strain distribution should be uniform as was positioned in a single action press. To move the heavy
much as possible; (b) strain should not exceed the maximum upper plate of die set, two hydraulic jacks were used. Using
allowable biaxial strain of the sheet material defined by FLD a pressure feeder, the fluid pressure in the container could
diagrams. Several simulations were carried out and based on be increased while the punch would be moved downward
the mentioned two criteria, the optimum manufacturing con- separately to form a complex part. During the deformation,
ditions of shell fender were defined. The results of simulations internal pressure could be recorded by a pressure gauge, which
will be presented in Section 4. was assembled across the feeding fluid line. Also the draw-
ing force and punch stroke could be recorded using a load
cell and extensometer, respectively. Finally, after shaping of
3. Experimental tools and procedure shell fender and stopping the movement of hydraulic press,
the fluid pressure was released and then the upper plate of
3.1. Experimental tools die set lifted using hydraulic jacks.
For experimental tryout, the blank sizes of 1/8 full scale uti-
Initially it should be point out that the experimental tools had lized in simulations were used for pure stretching and draw
been designed for studying the effect of parameters on form- in procedures. In the case of pure stretching condition, the
ing behavior of small size shell fender in laboratory testing as periphery movements of 150 mm diameter circular blanks are
a means of sheet hydroforming performance evaluation for constraint by the beads. While in the case of draw-in condi-
production of full size shell fender. All the tests were carried tion, the initial blanks were located in the desired position and
out using 40-t hydraulic single action press. A photograph and orientation on upper surface of the rubber diaphragm with
a schematic of the used equipment for hydroforming test are possibility of free movement of periphery of blank.
shown in Fig. 7. The designed experimental test rig consisted In the preliminary test series that had been carried out for
of a pressure container, cover plate, punch, guide, load cell, revealing the capabilities of pure stretching and draw-in pro-
Fig. 7 – Experimental set-up.

Fig. 8 – The deformation of mesh for pure stretching of IF steel sheet to the depth of (a) 25 mm and (b) 30 mm compared with
(c) experimental results.
cedures, different sheet metals (low carbon steel sheet (St14), For evaluating the thickness and strain distributions in
IF steel sheet and 2024 aluminum sheet with thickness of specimen deformed by sheet hydroforming process in differ-
0.8 mm with mechanical properties shown in Table 1 were ent conditions, circular grid of 4 mm diameter was printed
used. In addition, various combinations of fluid pressure and on one face of sheet material. After deformation, the circu-
punch travel based on the simulations’ results were tested. lar grid transformed into ellipse and therefore strain could be
The fluid pressure varied from 0 to 20 MPa. The criterion for calculated based on the measurements of circular grid diam-
selection of the process parameters for further experiments eter change (ISO 12004, 1997). In all the experiments for every
was free tearing condition of the specimens. condition, three specimens were tested and the results were
In the next step after selection of optimum process con- reported.
ditions with regards to the preliminary experimental and
simulation results, the effect of using material flow controlling
device (semi-die or guide) with various designs were exam- 4. Results and discussion
ined. This device was also evaluated at various fluid pressures
and for different sheet materials during draw-in procedure of At first, it must be realized that which of the two processes,
sheet hydroforming process. In addition, the effect of using i.e. pure stretching or draw-in of sheet metal during hydro-
various steel sheets on the punch force in the case of apply- forming could lead to a sound shell fender. For answering
ing fluid pressure and without applying fluid pressure was this question, simulations of the mentioned cases were car-
investigated. ried out using commercial finite element software capable of
Fig. 9 – The pictures of (a) low carbon steel (St14) and (b) aluminum sheet deformed by sheet hydroforming set-up under the
condition of pure stretching.
Fig. 10 – Effective plastic strain distribution in sheet hydroforming of fender shell using low carbon (St14): (a) drawing and
(b) stretching.
simulating large deformation. In Fig. 8, as an example, the In the next step, sheet hydroforming of shell fender under
resulted deformation of mesh for pure stretching of IF steel the condition of draw-in was evaluated, both numerically
sheet to the depth of 25 and 30 mm was compared with the and experimentally. Results of simulation in this condition
experimental results. As illustrated, the deformation at the predicted possibility of manufacturing sound shell fender, if
nose of the punch was so concentrated that sound product suitable combination of fluid pressure and punch stroke were
could not be produced. Simulation and following experimen- used. If the selected fluid pressure at the beginning of defor-
tal evaluation for sheet hydroforming of aluminum and low mation was less than 7 MPa, too much wrinkling would appear,
carbon steel sheet (St14) in pure stretching condition led to while increasing of fluid pressure up to 15 MPa, helped in pre-
the same results. In Fig. 9 pictures of aluminum and low car- vention of wrinkling and further increase of fluid pressure had
bon steel sheet (St14) deformed using experimental set up no effect on the produced part. Based on the uniformity of
of sheet hydroforming are shown. As shown, deformation of strain distribution and lessening of biaxial strain from critical
these sheet materials under the condition of pure stretching value criteria, fluid pressure of 10 MPa was selected. For clari-
also lead to fracture. Therefore, sheet hydroforming under the fying the main difference between two procedures (stretching
condition of pure stretching for the production shell fender in and draw-in), the deformed meshes of low carbon steel (St14)
all rang of considered fluid pressure led to fracture and there- in two different conditions of sheet hydroforming have been
fore it was not considered practical for manufacturing of this compared in Fig. 10. Results of simulations show that even in
part. optimum fluid pressure of 10 MPa, producing sound part under
Fig. 11 – Path determined along the main length of the part for thickness measurements.
Fig. 12 – Comparison of thickness variation along the path

mentioned in Fig. 11 for stretching and drawing.
Fig. 14 – Sheet hydroforming drawing experiment without
using semi-die device.
the condition of pure stretching was impossible, while using

draw-in could lead to a sound product. Comparison of thick-
ness variations resulted from simulations of two methods, cult. In Fig. 13, the results of simulation for the production
along the main length of specimen shown in Fig. 11 reveals of shell fender with and without using semi-die device are
the main differences, which are shown in Fig. 12. Accordingly, shown. As illustrated, without using semi-die device, wrin-
under the condition of pure stretching, thickness was reduced kling appeared and it was not possible to produce a sound
locally at the normalize point of 0.4, while in the case of draw- product. In Fig. 14 shell fender produced without using flow
in such localized thinning would not appear. The position controlling device is shown. As presented, without using
of local thinning that had been predicted by simulation was semi-die device, wrinkling appeared that was consistent with
consistent with the position of the fractures in experimental simulation results. Therefore, for the production of such com-
results. plicated parts, using some device for controlling material flow
Productions of sound shell fenders using sheet hydroform- was necessary. Fig. 15 shows the shell fender produced using
ing under the condition of draw-in without some semi-die semi-die device during sheet hydroforming with draw-in con-
device for controlling of material flow seemed to be diffi- dition of low carbon steel sheet (St14). Using of IF steel sheet
Fig. 13 – Result of simulation for production of shell fender (a) with and (b) without using semi-die device.
Fig. 15 – Sheet hydroforming of low carbon steel sheet

(St14) using draw in condition and semi-die device.
Fig. 17 – Variation of thickness distribution in the final
product predicted by simulation and emerged from
experimental results using IF steel sheets with and without
under the same condition during sheet hydroforming for pro-
application of fluid pressure.
ducing the same part led to the successful operation. But in
the case of 2024-O aluminum sheet because of low formabil-
ity, sheet hydroforming operation did not lead to the sound
product. Therefore, experimental results confirmed that using
flow-controlling device for influencing material flow in the defined path in the final product predicted by simulation and
desired direction would prevent wrinkling during sheet hydro- emerged from experimental results using IF steel sheets with
forming of complicated products if sheet materials had the and without the application of fluid pressure are compared.
minimum required formability. There is a good correlation between the experimental and
By applying established desired condition for the produc- simulation results. Nevertheless, as shown in the figure there
tion of sound shell fender during sheet hydroforming process, are differences in the thickness variations for specimens pro-
it became possible to evaluate the variations of strain along duced with and without application of the fluid pressure. In
different directions related to the usage of different sheet the case of forming shell fender without using fluid pressure,
materials. Fig. 16 illustrates thickness variations predicted by strain concentration appeared that in most cases could lead
simulations and evaluated by experimental results using low to fracture. Therefore, it can be deduced that by applying fluid
steel carbon sheet (St14) along the path defined in Fig. 11. As it pressure, thickness variations become more uniform in the
can be seen, there is a good correlation between the simulation case of such a complex shape.
and experimental results in locating the thinning position that Use of fluid pressure during sheet hydroforming also influ-
was at the top of the punch. Also comparison of Figs. 12 and 16 enced the applied punch force. In Fig. 18 the variation of punch
shows that how draw-in condition would help to establish force with punch stroke for the cases of with and without
more uniform distribution of thickness than the pure stretch applying fluid pressure are compared. Slopes of curves espe-
condition. cially differ by applying fluid pressure at the beginning of the
As mentioned in the introduction, one feature of sheet process and this case is more apparent for IF steel sheet. This
hydroforming was the application of fluid pressure for behavior could be attributed to the effect of existing friction
influencing the deformation behavior and increasing the between tools and sheet metal. In addition, application of fluid
deformability of sheet. In Fig. 17, thickness distribution along pressure increased the necessary punch force especially at
the end of process. This increase in punch force was natu-
ral and appeared because of fluid pressure resistance. General
differences between the shapes of punch load–stroke curves
of St14 and IF steel sheets could be related to the different
mechanical properties of the mentioned sheet materials. As
illustrated, beside the effect of fluid pressure on the success
of forming process and uniformity of deformation, fluid pres-
sure had significant effect on the variations of punch load with
punch stroke.
5. Conclusion
Manufacturing of a complicated stamping shell fender has

been selected to evaluate the capability of sheet hydro-
Fig. 16 – The variation of thickness resulted from forming process for the production of complex parts whose
simulations and experiments on low steel carbon sheet manufacturing via conventional methods required several
(St14) in the path defined in Fig. 11. steps.
references
Ahmetoglu, M., Hua, J., Kulukuru, S., Altan, Taylan, 2004.

Hydroforming of sheet metal using a viscous pressure
medium. J. Mater. Process. Technol. 146, 97–107.
Aretz, H., 2004. Numerical restrictions of the modified maximum
force criterion for prediction of forming limits in sheet metal
forming. Model. Simul. Mater. Sci. Eng. 12, 677–692.
Harnau, M., Konyukhov, A., Schweizerhof, K., 2005. Algorithmic
aspects in large deformation contact analysis using
Solid-Shell elements. Comput. Struct. 83, 1804–1823.
Hein, P., Vollertsen, F., 1999. Hydroforming of sheet metal pairs. J.
Mater. Process. Technol. 87, 154–164.
ISO 12004, Metallic materials-Guidelines for the determination of
forming-limit diagrams, The international Organization for
Standardization, 1997(E).
Kandil, A., 2003. An experimental study of hydroforming deep
drawing. J. Mater. Process. Technol. 134, 70–80.
Kang, B.S., Son, B.M., Kim, J., 2004. A comparative study of
stamping and hydroforming processes for an automobile fuel
tank using FEM. Int. J. Mach. Tools Manuf. 44, 87–94.
Lang, L.H., Wang, Z.R., Kang, D.C., Yuan, S.J., Zhang, S.H.,
Danckert, J., Nielsen, K.B., 2004. Hydroforming highlights:
sheet hydroforming and tube hydroforming. J. Mater. Process.
Technol. 151, 165–177.
Lei, L.P., Kim, J., Kang, B.S., 2000. Analysis and design of
hydroforming process for automobile rear axle housing by
FEM. Int. J. Mach. Tools Manuf. 40, 1691–1708.
Nakamura, S., Sugiura, H., Onoe, H., Ikemoto, K., 1994.
Hydromechanical drawing of automotive parts. J. Mater.
Fig. 18 – Variation of punch force with punch stroke for the Process. Technol. 46, 491–503.
cases of application and without application of fluid Novotny, S., Hein, P., 2001. Hydroforming of sheet metal pairs
pressure. from aluminum alloys. J. Mater. Process. Technol. 115, 65–69.
Parsa, M.H., Pournia, P., 2004. Blank shape design based on
inverse finite element method using ideal forming theory and
First of all, the difference between pure stretching and a modified kinematics formulation. In: Proceedings of 7th
Biennial ASME Conference. Engineering Systems Design and
draw-in procedures have been examined using finite element
Analysis, Manchester, UK, July 19–22.
simulations and experiments. Simulations and experimen- M.H. Parsa, M. Shahabizaheh, Deformation of aluminum sheet
tal results showed that pure stretching procedure produced into complicated depressions using sheet hydroforming, Int. J.
defective product, while draw-in procedure could lead to the Adv. Manuf. Technol., submitted for publication.
production of a sound part in the case of a complex part. The Rimkusa, W., Bauer, H., Mihsein, M.J.A., 2000. Design of
reason for tearing of part during pure stretching procedure, load-curves for hydroforming applications. J. Mater. Process.
Technol. 108, 97–105.
could be attributed to the deformation concentrations at sharp
Thiruvarudchelvan, S., Lewis, W., 1999. A note on hydroforming
points. In the case of draw-in procedure, such concentrations
with constant fluid pressure. J. Mater. Process. Technol. 88,
did not appear and deformation distributed more uniformly 51–56.
than pure stretching case. Thiruvarudchelvan, S., Travis, F.W., 2003.
Additionally for production of shell fender using sheet Hydraulic-pressure-enhanced cup-drawing processes—an
hydroforming with draw-in procedure, because of the shape appraisal. J. Mater. Process. Technol. 140, 70–75.
complexity of shape and its influence on the material flow, Yossifon, S., Tirosh, J., 1989. On the permissible fluid-pressure
path in hydroforming deep drawing process—analysis of
some flow controlling or semi-die device was required for con-
failures and experiments. Trans. ASME 110, 146–152.
trolling material flow. In such a condition, sheet hydroforming Zampaloni, M., Abedrabbo, N., Pourboghrat, F., 2003.
could be used for the production of complex parts, through the Experimental and numerical study of stamp hydroforming of
optimization of blank geometry, fluid pressure, and shape and sheet metals. Int. J. Mech. Sci. 45, 1815–1848.
position of the semi-die device or guide. These results have Zhang, S.H., 1999. Developments in hydroforming. J. Mater.
been confirmed by the simulation outcomes and experimental Process. Technol. 91, 236–244.
works. Zhang, S.H., Danckert, J., 1998. Development of hydro-mechanical
deep drawing. J. Mater. Process. Technol. 83, 14–25.
Zhang, S.H., Zhou, L.X., Wang, Z.T., Xu, Yi, 2003. Technology of
Acknowledgment sheet hydroforming with a movable female die. Int. J. Mach.
Tools Manuf. 43, 781–785.
Zhang, S.h., Wang, Z.R., Xu, Y., Wang, Z.T., Zhou, L.X., 2004.
The authors wish to express their gratitude to IDRO of Min-
Recent development in sheet hydroforming technology. J.
istry of Mining and Industry in Iran for providing the financial
Mater. Process. Technol. 151, 237–241.
support of this research.
19
European Journal of Mechanics A/Solids 31 (2012) 67e76
European Journal of Mechanics A/Solids

journal homepage: www.elsevier.com/locate/ejmsol
Experimental and numerical investigation of localized thinning in hydroforming

of micro-tubes
W. Zhuang a, S. Wang b, J. Lin b, *, D. Balint b, Ch. Hartl c
a
State Key Laboratory of Automotive Dynamic Simulation, Jilin University, Changchun 130022, PR China
b
Department of Mechanical Engineering, Imperial College, London SW7 2AZ, UK
c
Faculty of Automotive Systems Engineering and Production Engineering, Köln D50679, Germany
Article history: An experimental program has been carried out for hydroforming of stainless steel micro-tubes. Under
Received 16 July 2010 careful control, it was found that failure takes place randomly, which is significantly different from
Accepted 7 June 2011 observations of failure in hydroforming of macro-tubes, where failure loads and locations are predictable.
Available online 23 July 2011
This occurs because wall thinning of micro-tubes in forming processes is non-uniform, i.e. localized
necking takes place randomly. To investigate the localized thinning mechanism, an integrated crystal
Keywords:
plasticity finite element (CPFE) modeling system has been developed. In this paper, a simplified plane
Hydroforming
strain CPFE model is presented and used to investigate the localized thinning and failure features in
Micro-tubes
Micro-mechanics modeling
hydroforming of micro-tubes. The crystal plasticity equations were implemented in the ABAQUS/Explicit
Crystal plasticity FE modeling FE code through a user-defined material subroutine, VUMAT. Single crystal, three-grain and polycrystal
FE models were generated to study the localized thinning/necking mechanism and the effect of differing
adjacent grain orientations, as well as the number of grains across the smallest specimen dimension, on
the necking features. It has been confirmed from the analyses that the localized thinning observed in
hydroforming of micro-tubes is significantly affected by the microstructure and grain orientations of the
material.
Ó 2011 Elsevier Masson SAS. All rights reserved.
1. Introduction and is capable of high productivity and accuracy in manufacturing

micro-tubes.
Hydroforming is a metal forming technology based on the Since an ordinary tubular component is generally much larger in
application of pressurized liquid to generate defined workpiece its smallest dimension (i.e. thickness) compared with the grain size
shapes from tubular materials or sheet metals. The tube hydro- of the metal from which it is made, conventional macro-mechanics
forming technique has been widely used in automotive, aircraft and (continuum) FE modeling is effective for simulating typical
sanitary industries in recent years (Ahmetoglu and Altan, 2000). hydroforming processes (Aue-u-lan et al., 2004; Kim et al., 2005;
Compared with conventional punching and drawing processes, Kang et al., 2005). However, in hydroforming of micro-tubes, the
tube hydroforming has the advantage of part consolidation, weight material grain size can be on the order of the smallest part
reduction, improved structural strength and stiffness, lower tooling dimension. Thus, any given region within a micro-part may contain
cost, fewer secondary operations, reduced dimensional variations, ten or fewer grains; when the number of grains across the smallest
and reduced scrap. Hydroforming enables forming hollow part dimension is sufficiently small, microstructural effects such as
complex-shaped components with integrated structures from strain non-uniformity and localization, size effects (Yefimov and
single initial workpieces (Hart, 2005). Modern electronic, tele- Van der Giessen, 2005) and twinning (Kowalczyk-Gajewska,
communication and medical technology products increasingly 2010) may play an important role in the deformation.
require tubular micro-components; the market for these products, To investigate the deformation features in hydroforming of
and the micro-tubes they depend upon, is growing rapidly. micro-tubes, a hydroforming test facility has been designed and
Hydroforming is an efficient and time saving production method, manufactured. Tests have been carried out for hydroforming of
micro-tubes made of stainless steel having an initial blank outer
diameter of 800 mm and a wall thickness of 40 mm. Fig. 1 shows some
* Corresponding author. examples of hydroformed stainless steel micro-tubes having a final
E-mail address: [email protected] (J. Lin). outer diameter of 1030 mm. Many tubes were formed using the same
0997-7538/$ e see front matter Ó 2011 Elsevier Masson SAS. All rights reserved.
doi:10.1016/j.euromechsol.2011.06.017
68 W. Zhuang et al. / European Journal of Mechanics A/Solids 31 (2012) 67e76
conventional continuum plasticity without the introduction of

imperfections (Larsson et al., 1982; Tvergaard, 1990; Dyson et al.,
1996); hence the need to account for the underlying crystal struc-
ture using crystal plasticity theory.
Due to the aforementioned differences in the deformation
characteristics of hydroforming at the micro- and macro-scales, the
workpiece of a micro-part cannot be regarded as a homogeneous
continuum for process-simulation purposes. Crystal-plasticity (CP)
theory, which assumes that homogenized crystallographic slip is the
plastic deformation mechanism for crystalline materials, has
attracted significant further attention in recent years because of its
ability to relate the plastic deformation of micro-parts to their
microstructures. CP models have been used successfully in the
modeling of important phenomena, such as micro-crack initiation,
crack propagation, fatigue, creep, etc. Crystal plasticity models based
on a single-crystal yield function have been proposed for hydro-
forming by several investigators (Zamiri et al., 2007; Zamiri and
Pourboghrat, 2009; Guan et al., 2006; Jansson et al., 2008; Korkolis
and Kyriakides, 2008; Nygards and Gudmundson, 2004; Houtte
et al., 2006). However, such models require sophisticated yield
functions and specially designed algorithms so that non-uniqueness
of stress can be avoided. Recently, Shi et al. (2010) studied the
localized necking of an aluminum tube under internal pressure.
Localized necking was found to be associated with a surface insta-
bility resulting from non-uniform deformation. EBSD data was used
to define the grain structure and orientations, with investigation
Fig. 1. Premature localized failure of formed micro-tubes. (a) Failure takes place
randomly. (b) Close-up view of the failure. into the effect of strain hardening and strain rate sensitivity on the
predicted necking behavior. Grain distribution effects were explored
by simulating a random grain structure with the same overall
texture as the EBSD measurement, and a structure where cube-
hydroforming parameters and it was found that rupture took place
oriented grains were replaced by grains having a random orienta-
at different locations as a result of localized necking. Fig. 2 shows
tion. This was sufficient to conclude that the spatial grain orientation
simulation results for a quarter section of a hydroformed micro-tube
distribution has a significant effect on necking in pressurized tubes
using a macro-mechanics modeling method (i.e. using conventional
(Shi et al., 2010). In the present work, a systematic study of the effect
continuum plasticity theory); it can be observed that the wall
of grain orientation distribution is the focus, with necking occurring
thickness of the tube is uniform and no localized necking is pre-
naturally due to strain localization in grains having orientations
dicted, unless large strain analyses together with imperfections (e.g.
locally favorable to crystallographic slip. An emphasis of the present
in the wall thickness) are used, which could give localized necking in
research is on generating virtual microstructures in order to
the tube wall when using a conventional continuum plasticity
systematically study the effect of relative grain orientations (Schmid
model, e.g. for a power-law isotropic hardening material (Larsson
factors) and size (relative to the tube thickness) on the features of the
et al., 1982; Tvergaard, 1990). Localized necking and the subse-
localized thinning, and how this relates to the local state of stress in
quent fracture and rupture of the tube cannot be captured using
hydroforming of micro-tubes.
The overall objective of the present study was the development
of an experimental method and an integrated, polycrystalline
crystal plasticity finite element (CPFE) modeling technique for
investigating the localized thinning features in the hydroforming of
micro-tubes. Rate-dependent crystal plasticity theory is used to
overcome the stress non-uniqueness. This theory has no yield
function, and the numerical algorithms are stable and much more
easily implemented in commercial FE software. Grains within the
tube workpiece, their distributions and orientations are generated
automatically using the previously developed VGRAIN software
system. A set of crystal viscoplasticity model equations is imple-
mented in the ABAQUS/Explicit FE code through a user-defined
material subroutine, VUMAT. Single crystals, crystals having three
grains contrived for studying orientation effects, and polycrystals
have been generated and localized thinning analyzed using the
simplified plane strain CPFE modeling technique.
2. Experimental program for hydroforming of micro-tubes
2.1. Test material, facility and procedures
Fig. 3a shows an example of a prototype micro-tube hydro-

Fig. 2. Virtually formed micro-tube using a macro-mechanics modeling method. forming test machine, which has been designed for investigating
W. Zhuang et al. / European Journal of Mechanics A/Solids 31 (2012) 67e76 69
the feasibility of the hydroforming concept for the production of linearly to a maximum value of 400 MPa to ensure the blank material is
micro-tubes. This machine enables micro-hydroforming of tubular fully in contact with the die surface to form the desired part.
components having diameters between 0.2 and 1 mm. It is equip-
ped with a spindle driven pressure intensifier which enables the 2.2. Examination of the features of tube failure
application of up to 4000 bar internal pressure, the closing force
being realized by a hydraulic drive, and the axial punches being Roughly one hundred micro-tubes having the same initial and
moved by linear actuators with spindle gears. final geometry were formed using exactly the same hydroforming
Tests for the hydroforming of micro-tubes have been carried out parameters. Despite carefully controlling the process, failure
using the machine. The length of the stainless steel tubular workpiece occurred in random locations for most of the formed micro-tubes;
is 16 mm, and the outer diameters of the initial and formed tubes are examples are shown in Fig. 1. Visual inspections showed that
800 mm and 1030 mm, respectively. The experimental process is shown fracture took place in different locations, and in a random manner,
in Fig. 3b. At the beginning of the process the tube blank (800 mm) is before the die cavity was filled completely by the expanding tube.
placed into a die cavity (1030 mm) that corresponds to the final shape of To investigate the random failure features, microstructure exami-
the component. The dies are closed and the closing force applied while nation of the material was carried out to look at the relationship
the tube is internally pressurized by a liquid to expand the component. between grain size, as well as grain orientation distribution, and
Additionally the tube ends are axially compressed by sealing punches. the failure mode of the material.
The final tube is formed under the simultaneously controlled action of In order to examine the microstructure at the failure location,
internal pressure and axial force. The forming pressure is applied a formed micro-tube was cross-sectioned at the location of the
Fig. 3. Experimental setup for the hydroforming of micro-tubes. (a) Micro-tube hydroforming machine. (b) Micro-tube hydroforming tooling and test set-up.
rupture using a micro-wire cutting technique. By mounting, pol- number of finite elements, Nel, within a single crystal grain
ishing and etching, the grains in a cross-section at the location of (a polycrystal is simply an assemblage of single crystals with
failure were revealed. Fig. 4 shows the sample, grains and grain displacement compatibility at the common boundaries). The a slip
boundaries at two locations in the cross section. First, it can be system is characterized by mai and saj . The number of slip systems
observed that the grain size is non-uniform in the vicinity of the and their orientations depends on the crystal lattice structure, e.g.
failure. Fig. 4c shows that the grain size is more uniform where no an FCC crystal has 4 independent primary slip planes, and each has
obvious localized necking has been observed; there is an average of 3 independent slip directions (i.e. a ¼ 1,2,.,12). At the beginning of
approximately 2 grains across the thickness at that location. deformation, sa ¼ 0 for all a, and the slip plane normals mai and slip
However, Fig. 4d shows some large grains close to the failure; in directions saj are the same for all of the elements within a single
some areas, only one grain across the thickness of the material can
crystal grain; the initial grain orientation is defined by VGRAIN
be observed. In addition, localized necking is evident at the internal
(introduced later) according to a probability distribution that
surface of the deformed tube, near the failure (Fig. 4d). Once
characterizes the polycrystal grain orientation distribution. When
localized necking occurs, failure takes place quickly. In the
plastic deformation occurs, mai and saj may have different values for
following sections, CPFE models have been created to study the
localized necking features observed in the micrographs. the elements within a grain (in the finite strain formulation
employed here; not the case when a small strain formulation is
3. Crystal viscoplasticity constitutive equations used). The use of mai and saj ensures that the same orientation is
assigned to all elements within a grain.
Crystal plasticity theories are used to represent the flow of Material strain hardening is specified based on the slip system
dislocations along slip systems in metallic crystals in terms of strain hardness, ga. The self, haa, and latent, hab, hardening moduli
resolved shear strains. In particular, crystalline slip is assumed to are defined by Asaro (1983) and Peirce et al. (1982), which are
obey Schmid’s law, i.e. the slipping rate g_ a in any particular slip directly related to accumulated shear strain g:
system a is related to the shear stress, sa, acting on that slip system.
The crystal plasticity theory used in this paper follows the pio- hab ¼ hðgÞhh0 sech2 jh0 g=ðgs g0 Þj a ¼ b (5)
neering work of Taylor (1938), Hill and Rice (1972) and Asaro qhðgÞ asb
(1983). The set of crystal viscoplasticity constitutive equations
used are summarized below: X
12
g¼ jga j (6)
sa ¼ mai sij saj (1) a¼1
where h0 is the initial hardening modulus, g0 is the initial shear

g_ a ¼ asgnð
_ sa Þðjsa =ga jÞn (2) strength, gs is the break through stress when plastic flow initiates
and q is the hardening factor. In Taylor’s isotropic hardening
X assumption, the self and latent hardening rates are assumed to be
g_ a ¼ hab g_ b ; b ¼ 1; 2.12 for an FCC crystal (3)
the same. Hence, the value of the hardening factor, q, is taken as one
b !
X a a for isotropic hardening. Cijkl is the fourth order stiffness tensor and
a a a
s_ ij ¼ Cijkl 3_ kl 3_ pkl ¼ Cijkl 3_ kl 0:5g_ sk ml þ sl mk (4) the indices i, j, k and l take values between 1 and 3. In the initial
a state (t ¼ 0), sij ¼ 0, ga ¼ 0, ga ¼ g0 and 3 kl ¼ 0. The material
X
Nel parameters for the equation set presented here are listed in Table 1
where mai ¼ 1= mai ;k is the average slip plane normal and for 316 L stainless steel, and Young’s modulus and Poisson’s ratio
Nel
k ¼1
are 193 GPa and 0.34, respectively (Harewood and McHugh, 2007).
X
Nel
a
sj ¼ 1= saj ;k is the average slip direction, with respect to the A high value of n is used here (Table 1) to reduce the viscoplastic
Nel effect of the material, as the hydroforming is carried out at low
k ¼1
temperature.
4. Computational procedure for CPFE analysis
4.1. The numerical procedure
An integrated micro-mechanics CPFE modeling system contains

three parts: microstructure generation and pre-processing, CPFE
analysis and post-processing (Cao et al., 2009a,b). In the pre-
processing stage, the so-called VGRAIN system introduced by Cao
et al. (2009b) has been used to generate virtual grain structures
according to parameters characterizing the microstructure. The
grain structure within a defined region is generated according to
the input values of average, maximum and minimum grain sizes.
Orientations of grains are assigned according to a probability
distribution either in a random form or with a chosen distribution
Table 1
The values of the material parameters for Eqs. (1)e(6).
n a_ (s1) h0 (MPa) gs (MPa) g0 (MPa)

Fig. 4. Failure of formed micro-tube: (a) Formed micro-tube with a crack, and (b), (c) 20.0 0.001 225 330 50
and (d) the microstructure of a micro-tube.
(e.g. gamma). The generated virtual grains with their orientation

information are input into ABAQUS/CAE for further pre-processing,
e.g. adding boundary and loading conditions.
The CP material model is implemented in ABAQUS for explicit
calculations via the user-defined material subroutine, VUMAT. The
explicit implementation used here was adapted from the implicit
algorithm developed by Huang (1991) and compared to that of
Harewood and McHugh (2007). In explicit finite element calcula-
tion procedures, the task can be split up easily and solved by
a number of processors. Hence, the VUMAT can be constructed with
a vectorized interface. This means that when a simulation is carried
out using multiple processors, the analysis data can be split up into
blocks and solved independently. Thus, vectorization can be
preserved in the writing of the subroutine in order that optimal
processor parallelization can be achieved.
The accelerations and velocities at a particular point in time are
assumed to be constant during a time increment and are used to
solve for the next point in time, i.e. a forward Euler integration is
carried out. To reduce dynamic effects, the ratio of the duration of
the load to the fundamental natural period of the model is taken to
be greater than five (Kutt et al., 1998); it has been found that by
keeping the ratio of kinetic energy to the total internal strain
energy <5%, dynamic effects in the model are negligible (Chung Fig. 5. Micro-mechanics (CPFE) model for tube hydroforming. Grains and grain
boundaries are shown in the figure; different colors indicate different orientations.
et al., 1998; Choi et al., 2002).
4.2. Plane strain CPFE model

5. Computational results
The geometry and the FE model of the cross-section of a micro-
tube are shown in Fig. 5. In the CPFE model, a quarter section of the 5.1. Grain orientation effects of single crystals
micro-tube was considered with symmetry boundary conditions
along X ¼ 0 and Y ¼ 0. The minimum, average and maximum grain To investigate the localized thinning features associated with the
sizes of the material were taken to be 25, 30 and 40 mm, respec- grain orientation and the applied load, a single crystal cross-section
tively, and 95% of the grains are within that range. Hence, there are model (single grain, uniform orientation) was employed. Fig. 6a
between 1 and 2 grains across the thickness of the tube section on shows a single FCC crystal with two sets of coordinate systems, the
average. The grains and their orientations are generated using the cubic crystal system and the global sample coordinate system,
VGRAIN system, and they are read into ABAQUS/CAE for further XeYeZ. An assumption is made that the cubic crystal system may
mesh generation, boundary conditions specification and loadings undergo only one rotation relative to the global system, around the
definition. The die is defined as a rigid part in ABAQUS/CAE. The Z-axis; this is depicted in Fig. 6a where one of the potential slip
maximum applied pressure is 400 MPa; this high pressure ensures systems, {111}<110>, is also indicated for reference. Hence, the
that the workpiece is deformed sufficiently to come into complete effect of arbitrary rotations is not considered; the rotation defined by
contact with the die. A friction coefficient of 0.1 is used when the the one angle a is sufficient to illustrate the effect of grain orientation
workpiece and the die are in contact during the forming process. on necking. Fig. 6b illustrates the grain texture (in the global coor-
For simplicity, a 2D plane strain CPFE analysis was carried out here, dinate system) relative to a position on the tube specified by the
but the crystal plasticity equations above are valid for 3D analyses. angle q. It is indicated in Fig. 6b that the direction of the hoop stress at
It is worth mentioning that large strain, polycrystal CPFE anal- position q, sq, the main driving force for forming the part, follows the
yses require considerable CPU time. A 2D simplification (e.g. plane tangent direction of the tube cross-section (thin tubes are consid-
strain) reduces the computation time significantly and it is still ered here), whereas the crystal orientation remains unchanged over
possible to capture the interesting features, such as localized the tube cross section. The {111}<110> slip system is indicated by
thinning, failure, etc., as observed in hydroforming of micro-tubes. the dotted lines, and the orientation of this slip system is specified
The developed CPFE process modeling technique can be readily relative to the [100] direction; the projection of the slip plane normal
used for 3D hydroforming simulations if the 3D grain structures can onto the XeY plane, n, and the slip direction, s, are indicated. It
be constructed effectively for the initial metal tube. should be noted that the angle between the hoop stress direction
In order to investigate the relationship between localized thin- and the projection of the slip plane normal onto the XeY plane is the
ning and microstructures of polycrystalline materials in hydro- same as that between the hoop stress direction and the slip plane
forming of micro-tubes, three FE models of a quarter cross-section normal itself, which is skewed in space relative to the XeY plane;
of a tube were studied. The three models have the same initial writing n ¼ N ðN,KÞ ^ K,^ where N is the normal to the slip plane and
geometry, but different crystal combinations: 1) A single crystal ^ is the unit vector in the Z direction, it is clear that the scalar product
K
used to investigate the relationship between FCC slip systems and between any vector in the XeY plane with n, hence the angle
the loading direction; 2) A three-grain crystal comprising a small between them, is equal to the scalar product between that vector
crystal section inserted symmetrically in the middle of the quarter and N. Therefore, calculations of Schmid factor are correct when
FE model, having a different orientation relative to the grains using the projection of the normal, as done here. Henceforth, n is
adjacent to it, for studying the relationship between orientation simply referred to as the normal to the slip plane. The Schmid factor
mismatch and necking; And, 3) polycrystal structures for investi- for crystal orientation a at location q on the tube is
gating the apparently random occurrence of local thinning for real cosðp=4 a þ qÞcosðp=4 þ a qÞ. Thus, the location where plastic
hydroforming cases. deformation occurs most readily on the {111}<110> system, the
Numerical investigations were carried out with a grain orienta-

tion of a ¼ 60 . A single crystal was used for the tube cross-section
and was analyzed using the loading and boundary conditions
described above. Fig. 7 shows the deformed tube cross-section
superimposed with contours of accumulated plastic shear strain. It
can be observed, as described above, that the thickness of the formed
tube varies due to the variation in the direction of the hoop stress. The
necking ratio, which is defined as the minimum thickness divided by
the maximum thickness, is 0.74 for the simulation shown in Fig. 7. The
thickness of the formed part for the single crystal structure gradually
varies from the minimum thickness section (29.3 mm) at the position
indicated by the crystal axis [100] to the maximum thickness section
(39 mm), where the slip direction {111}<110> (see Fig. 6a) and the
hoop stress are perpendicular (hence the resolved shear stress on the
{111}<110> system is zero), which is the position 45 clockwise from
the crystal axis [100]. The change in thickness for the formed tube is
mainly due to the variation of the angle between the slip system and
the local hoop stress direction. In addition, the hardening law of the
material would affect the behavior, but different hardening laws are
not explored in this study.
5.2. Effect of adjacent grain orientation on localized necking:

A three-grain model
A three-grain model was created as shown in Fig. 8. The grain

orientations are defined according to the convention shown in
Fig. 6, and the tube dimensions are the same as those defined in
Fig. 5. The two large grains, I and III, have the same size, and are
sectors of the quarter model spanning 42 . Direction (1) represents
Fig. 6. (a) FCC single crystal with one rotation relative to the micro-tube (global)
coordinate system, about the Z-axis; (b) Cross section model of hydroforming of
the crystal axis [100], which is at a ¼ 45 for both grains I and III.
a micro-tube with the slip direction represented by dotted lines; a gives the [100] The small grain, grain II, is a sector of the quarter model spanning
direction, and q a location on the tube in the global coordinate system. 6 . A series of CPFE analyses have been carried out while varying
the small grain orientation (direction (1); crystal axis [100]) from
a ¼ 0 to 45 , while the orientations of the two large grains remain
location of maximum resolved shear stress, occurs where q ¼ a,
fixed at a ¼ 45 .
which gives sa ¼ smax ¼ sq/2. The most difficult location to deform
With applied internal pressure, the tube blank is deformed to
plastically is where jq aj ¼ p/4, which gives sa ¼ smin ¼ 0; in other
match the die shape. Fig. 9a shows the computational results for the
words, the location most difficult to deform plastically is 45 from
hydroformed tubes with grain II orientations ([100] direction) of
the location that is easiest to deform plastically. It should be noted
a ¼ 0 , 20 and 30 relative to the global horizontal axis (see Fig. 8).
that sa ¼ 0 on the {111}<110> system at a certain location does not
When the crystal orientation for the small grain, grain II, is 0 , the
imply plastic deformation does not occur; other slip systems out of
plastic mismatch between it and the adjacent grains (I and III as
the plane may be activated instead, but plastic deformation will in all
shown in Fig. 8) is maximized and the ratio tII/tmin ¼ 1.56, where tII
cases be less than it is at a location where sa is greater on the {111}<
is the thickness at the middle section of grain II and tmin is the
110> system. For example, if a ¼ 45 , the greatest plastic deforma-
minimum thickness of the deformed tube section. The small grain
tion will occur at q ¼ 45 , and the least at q ¼ 0 and q ¼ 90 .
is difficult to deform in the a ¼ 0 orientation, because the Schmid
Since the crystal orientation is constant over the tube cross-
section, the direction of the hoop stress varies relative to the slip
direction and it can be envisaged that necking will occur at the
position where the resolved shear stress is maximum, which will be
at the location where the [100] and/or [010] directions are
perpendicular to the local direction of the hoop stress. The only case,
for the quarter model, in which both [100] and [010] are perpen-
dicular to the local hoop stress direction is a ¼ 0 ; for positive a, the
only location is that indicated by the [100] direction, and for nega-
tive a, the only location is that indicated by the [010] direction; in
the case of a polycrystal, there may be many locations where
necking occurs in the quarter model. The quarter model with
symmetry conditions assumes a somewhat unnatural symmetry of
the necking locations since for any a > 0 , the [010] direction is
effectively neglected and the slip in the other quadrants occurs by
symmetric reflections of that in the first quadrant about the global
coordinate axes. However, the necking locations are separated such
that they will not appreciably interact, and in any case, the quarter
model assumption has no influence on the qualitative behavior Fig. 7. Predicted localized necking for a single crystal structure with a grain orientation
demonstrated in the simulations, which is the purpose of this study. of 60 with respect to the sample (global) coordinate system.
Fig. 8. A three-grain structure; the orientation of the large grains I and III are fixed and
the orientation of grain II varies.
factor (hence the resolved shear stress) is zero in the center of the
grain (a ¼ 45 ), and remains relatively small even at its outer
extents since the grain size is small (6 span). The influence of this
hard central grain causes the thinnest section to be located away
from the interfaces between the small grain and the two large
grains, i.e. the constraint of the small grain being difficult to deform
plastically is felt by the neighboring grains until a sufficient
distance away from that interface is reached (it is a compromise
between the central grain, grain II, and the adjacent grains, as
grains I and III would prefer to localize at the interfaces with the
central grain, where their resolved shear stress is largest). However,
as the orientation of the small grain was changed to 20 and 30 ,
the ratio tII/tmin reduced to 1.34 and 1.13, respectively. It can be
observed from Fig. 9a that grain II deformed more as its orientation
varied from 0 to 30 , demonstrating a reduction in the localization Fig. 9. The dependence of localized necking behavior on the variation of the orien-
of plastic strain in the grains adjacent to grain II. Fig. 9b shows the tation of grain II, as depicted in Fig. 8. (a) Strain and necking variations as the orien-
variation of tII/tmin with grain II orientation; the ratio tII/tmin tation of grain II changes. (b) Thinning behavior as a function of relative grain
misorientation.
approaches 1 as the orientation of grain II varies from 0 to 45 .
5.3. Deformation effects for polycrystalline micro-tubes progresses very quickly and leads to localized failure of the
material.
Two cases were studied to investigate thinning for hydro- Fig. 10c shows the progression of thinning with deformation, t/
forming of micro-tubes with polycrystalline material microstruc- t0, where t0 is the initial thickness of the tube, for the three loca-
tures. In the two cases, the grain structures and grain orientations tions, A, B and C indicated in Fig. 10a and b, during the forming
of the tube blank were identical, but the degree of deformation was process. Location A represents the thinnest location and location C
different. This simulates the process of forming tubes with different the thickest. It can be observed that during the initial tube
final diameters from workpieces with the same initial diameter. expansion process, there is no obvious localized thinning taking
The initial dimensions and grain structure of the material blank are place. However, when the deformation progresses to a certain
shown in Fig. 5. For the first case (Fig. 10a), the radius of the die is extent (a deformation ratio above 1.05), more thinning occurs at
515 mm, thus, the ratio of final diameter to initial diameter, the location A. During the subsequent deformation, the localized
deformation ratio, is 1.3. For the case shown in Fig. 10b, the radius of thinning at A progresses very quickly and the thinning at location C
the die is 596 mm, hence the deformation ratio is 1.5. progresses slowly. This indicates that failure must take place at
Predicted localized thinning for the two cases is shown in location A. The other location of localized thinning, B, does not
Fig. 10a and b. It is clear that the wall thicknesses of the formed progress very much by the end of the deformation process as
tubes are non-uniform and the amount of localized thinning localization at A has taken over the deformation.
increases with deformation dramatically and non-proportionally. This localized thinning feature can be further demonstrated in
The ratios of the maximum and minimum values of wall thick- Fig. 11 for deforming the tube to the die with the radius of 596 mm
ness in Fig. 10a to those in Fig. 10b are 0.97 (31.9/32.9) and 0.64 (deformation ratio: 1.5). It should be noticed that the deformed
(13.7/21.4), respectively. It can be seen that the maximum value of tube is not perfectly circular before the material finally contacted
the wall thickness remains approximately the same as the defor- the die surface. Even if the maximum pressure is reached, some
mation progresses, but the minimum value of the wall thickness particular areas of the tube surface may have difficulty deforming
decreases sharply with further deformation. This indicates that to the die surface due to the slip systems of the grains. It can be seen
once necking (localized thinning) takes place at a location, it from Fig. 11a that once the necking takes place at specific locations,
Fig. 11. Thinning progress in forming a micro-tube to a deformation ratio of 1.5. (a)
Field plots for the deformation history, and, (b) normalized thinning profiles for
different deformation ratios, where R is the average radius of the tube at the given
deformation ratio.
orientations. The experimentally observed random localized thin-

ning and failure are also predicted by the CPFE analyses carried out
in this study. Furthermore, the CPFE analyses confirm that
conventional macro-mechanics (continuum) FE process modeling
techniques (see Fig. 2) cannot be used to predict localized failure in
hydroforming of micro-tubes unless imperfections are introduced
(Larsson et al., 1982; Tvergaard, 1990; Dyson et al., 1996).
5.4. Random thinning locations in polycrystal structures

Fig. 10. Comparison of thinning features for deformation ratios of (a) 1.3 and (b) 1.5,
and (c) necking development for the positions A, B and C.
Polycrystal structures and grain orientations are generated
using the VGRAIN system automatically as mentioned before and
shown in Fig. 5. To simulate the deformation and thinning behavior
it progresses quickly and most of the deformation in the forming of hydroformed micro-tubes taken from the same piece of original
process would occur at these locations. The thinnest location is at material, the grain structures are generated twice using the same
0.28p from the x-axis. This localized necking feature can be further microstructure control parameters defined above, with grain
illustrated in Fig. 11b, where the thinnest location is indicated by orientations for both micro-tube models assigned randomly based
the dotted line. on the orientation probability distribution within the VGRAIN
As discussed before, the necking location is related to the angle system, characterized by the maximum, minimum and average
between the slip direction and the local direction of the hoop stress, grain sizes. Grain structures and grain orientations may be different
but is also influenced by the orientation and size of neighboring between the two CPFE models, but they are within the ranges of the
grains. Thus, for a polycrystal case, the location and amount of material specification. The simulation results for the hydroformed
localized necking are difficult to control in practice if there are only micro-tubes are shown in Fig. 12. It can be seen that the minimum
one or two grains through the thickness of the tube, as shown by and maximum values of the wall thickness of the formed tubes,
the examples of localized failure in actual hydroforming of micro- shown in Fig. 12a (20.2 mm and 31.4 mm, respectively) and
tubes in Fig. 1; the location of failure is random and cannot be b (20.7 mm and 33.1 mm, respectively), are different for the two
predicted without a detailed knowledge of the grain sizes and cases studied. The wall thickness of the two hydroformed micro-
crystal axis is perpendicular to the local hoop stress direction, and

the thickest section is 45 from the location of the thinnest section
for a single crystal structure. Thus, the location of localized thinning
is easy to predict for a single crystal material but is difficult to
predict for polycrystal cases without a detailed knowledge of the
microstructure. However, the CPFE models developed in this study
can be used to guide tube workpiece design by identifying the
lower limit on the number of grains through the thickness such that
an acceptable uniformity of thickness is achieved during forming,
hence mitigating unpredictable failure, and a polycrystal grain
distribution design that facilitates thickness uniformity. It was also
demonstrated that adjacent grains with significant plastic
mismatch compromise in determining the neck location, i.e. the
grain that is easier to deform plastically may neck but not neces-
sarily where the resolved shear stress is largest; neighboring grains
that are difficult to deform plastically constrain those that are easy
to deform, and thereby influence the neck location. Lastly, the
simulations demonstrated that once necking takes place at the
location that is most susceptible to plastic deformation by crystal-
lographic slip, it progresses very quickly and leads to localized
thinning, then failure; other potential areas of localized necking are
subsequently irrelevant.
Acknowledgment
The authors thank the European Committee for their support on

the FP7 project “Multiscale Modelling for Multilayered Surface
Systems (M3-2S)”, Grant No: CP-FP 213600-2 M3-2S.
References
Ahmetoglu, M., Altan, T., 2000. Tube hydroforming, state of the art and future
trends. J. Mater. Process. Technol. 98, 25e33.
Asaro, R.J., 1983. Crystal plasticity. J. Appl. Mech. 50, 921e934.
Aue-u-lan, Y., Ngaile, G., Altan, T., 2004. Optimizing tube hydroforming using process
simulation and experimental verification. J. Mater. Process. Technol. 146, 137e143.
Fig. 12. Comparison of thinning features for microstructures generated twice by the Cao, J., Zhuang, W., Wang, S., Lin, J., 2009a. Development of a VGRAIN system for CPFE
VGRAIN system using the same control parameters. analysis in micro-forming applications. Int. J. Adv. Manuf. Tech. 47, 981e991.
Cao, J., Zhuang, W., Wang, S., Ho, K.C., Zhang, N., Lin, J., Dean, T.A., 2009b. An
integrated crystal plasticity FE system for microforming simulation.
J. Multiscale Model. 1, 107e124.
tubes having random grain orientations is non-uniform and diffi- Choi, H.-H., Hwang, S.-M., Kang, Y.H., Kim, J., Kang, B.S., 2002. Comparison of
cult to predict. This is a result of the spatial variation of grain size, implicit and explicit finite element methods for the hydroforming process of an
grain orientation and the constraints neighboring grains with automobile lower arm. Int. J. Adv. Manuf. Tech. 20, 407e413.
Chung, W.J., Cho, J.W., Belytschko, T., 1998. On the dynamic effects of explicit FEM in
different orientations have on one another; this cannot be captured
sheet metal forming analysis. Eng. Computations 15, 750e776.
using conventional macro-mechanics (continuum) FE techniques Dyson, B.F., Hayhurst, D.R., Lin, J., 1996. The ridged uniaxial testpiece: creep and
(see Fig. 2). It can also be observed that localized thinning occurs at fracture predictions using large-displacement finite element analysis. Proc. R.
different locations due to the randomness of the grain orientations Soc. Lond. A 452, 655e676.
Guan, Y., Pourboghrat, F., Barlat, F., 2006. Finite element modelling of tube hydro-
of the workpiece material, which is difficult to control in practice. forming of polycrystalline aluminium alloy extrusions. Int. J. Plast. 22,
2366e2393.
Harewood, F.J., McHugh, P.E., 2007. Comparison of the implicit and explicit finite
6. Conclusions element methods using crystal plasticity. Comput. Mater. Sci. 39, 481e494.
Hart, Ch, 2005. Research and advances in fundamentals and industrial applications
of hydroforming. J. Mater. Process. Technol. 167, 283e392.
Tests on hydroforming of micro-tubes showed that premature Hill, R., Rice, J.R., 1972. Constitutive analysis of elastic-plastic crystals at arbitrary
failure takes place at random locations for materials having strain. J. Mech. Phys. Solids 20, 401e413.
between 1 and 2 grains through the thickness of the initial tube Houtte, P.V., Kanjarla, A.K., Van Bael, A., Seefeldt, M., Delannay, L., 2006. Multiscale
modeling of the plastic anisotropy and deformation texture of polycrystalline
workpiece. The random failure and localized thinning features materials. Eur. J. Mech. A Solids 25, 635e648.
observed in the experiments cannot be predicted using macro- Huang, Y., 1991. A User-material Subroutine Incorporating Single Crystal Plasticity in
mechanics (continuum) FE techniques. Effects can be generally the ABAQUS Finite Element Program Harvard University Report, MECH 178.
Jansson, M., Nilsson, L., Simonsson, K., 2008. Tube hydroforming of aluminium
understood and anticipated for single crystal tubes with known
extrusions using a conical die and extensive feeding. J. Mater. Process. Technol.
crystal structures through a geometric analysis of the Schmid 198, 14e21.
factor, hence resolved shear stress, but this cannot be done for Kang, S.J., Kim, H.K., Kang, B.S., 2005. Tube size effect on hydroforming formability.
J. Mater. Process. Technol. 160, 24e33.
polycrystals. CPFE models have been developed, and these models
Kim, J., Kim, W.J., Kang, B.S., 2005. Analytical and numerical approach to prediction
can predict the important features of localized thinning due to of forming limit in tube hydroforming. Int. J. Mech. Sci. 47, 1023e1037.
grain structures and grain orientations. The research also shows Korkolis, Y.P., Kyriakides, S., 2008. Inflation and burst of aluminium tubes, Part II: an
that localized necking is related to the directions between the advanced yield function including deformation-induced anisotropy. Int. J. Plast.
24, 1625e1637.
crystal slip systems and the local hoop stress direction. The thinnest Kowalczyk-Gajewska, K., 2010. Modelling of texture evolution in metals accounting
location in the simulations is always in the section where the [100] for lattice reorientation due to twinning. Eur. J. Mech. A Solids 29, 28e41.
Kutt, L.M., Pifko, A.B., Nardiello, J.A., Papazian, J.M., 1998. Slow-dynamic finite Taylor, G.I., 1938. Plastic strains in metals. J. Inst. Met. 62, 307e324.
element simulation of manufacturing processes. Comput. Struct. 66, 1e17. Tvergaard, V., 1990. Bifurcation in elastic-plastic tubes under internal pressure. Eur.
Larsson, M., Needleman, A., Tvergaard, V., Storakers, B., 1982. Instability and failure of J. Mech. A Solids 9, 21e35.
internally pressurized ductile metal cylinders. J. Mech. Phys. Solids 30 (3), 121e154. Yefimov, S., Van der Giessen, E., 2005. Size effects in single crystal thin films:
Nygards, M., Gudmundson, P., 2004. Numerical investigation of the effect of non-local nonlocal crystal plasticity simulations. Eur. J. Mech. A Solids 24, 183e193.
plasticity on surface roughening in metals. Eur. J. Mech. A Solids 23, 753e762. Zamiri, A., Pourboghrat, F., 2009. A novel yield function for single crystals based on
Peirce, D., Asaro, R.J., Needleman, A., 1982. Material rate dependence and localized combined constraints optimization. Int. J. Plast. 26, 731e746.
deformation in crystalline solids. Acta Metall. 31, 1951e1976. Zamiri, A., Pourboghrat, F., Barlat, F., 2007. An effective computational algo-
Shi, Y., Wu, P.D., Lloyd, D.J., Embury, J.D., 2010. Crystal plasticity based analysis of rithm for rate-independent crystal plasticity based on a single crystal yield
localized necking in aluminum tube under internal pressure. Eur. J. Mech. A surface with an application to tube hydroforming. Int. J. Plast. 23,
Solids 29, 475e483. 1126e1147.
20
Available online at www.sciencedirect.com
International Journal of Mechanical Sciences 45 (2003) 1815 – 1848
Experimental and numerical study of stamp hydroforming

of sheet metals
M. Zampaloni, N. Abedrabbo, F. Pourboghrat∗
Department of Mechanical Engineering, Michigan State University, East Lansing, MI 48824-1226, USA
Received 17 December 2001; received in revised form 10 November 2003; accepted 18 December 2003
Abstract
The objectives of this research was to experimentally and numerically study the stamp hydroforming process
as a means for shaping aluminum alloy sheets. In stamp hydroforming, one or both surfaces of the sheet metal
are supported with a pressurized viscous 4uid to assist with the stamping of the part thereby eliminating the
need for a female die. The pressurized 4uid serves several purposes: (1) supports the sheet metal from the
start to the end of the forming process, thus yielding a better formed part, (2) delays the onset of material
failure and (3) reduces wrinkle formation. This paper focuses on the experimental and numerical results
of the stamp hydroforming process utilizing a 4uid pressure applied to one surface of the sheet metal. The
e7ects of applying a constant, varying and localized pressure to the surface of 3003-H14-aluminum sheet alloy
were evaluated. Experiments demonstrated draw depths improvements up to 31% before the material failed.
A failure prediction analysis by Hsu was also carried out to predict an optimal 4uid pressure path for the
varying 4uid pressure case. The commercial ;nite element analysis code Ls-Dyna3D was used to numerically
simulate the stamp hydroforming process. Both isotropic and anisotropic material models were used and their
predictions compared against the experimental results. The numerical simulations utilizing Barlat’s anisotropic
yield function accurately predicted the location of the material failure and the wrinkling characteristics of the
aluminum sheet.
? 2003 Elsevier Ltd. All rights reserved.
Keywords: Sheet hydroforming; Stamping; Fluid forming; Fluid pressure; Failure; Tearing; Wrinkling; Material
anisotropy
1. Introduction
There are a multitude of parts, both in size and complexity that can be made from sheet metals,
usually produced in large quantities on mechanical presses. There are signi;cant expenses associated
∗
Corresponding author. Tel.: +1-517-353-0819; fax: +1-517-353-1750.
E-mail address: [email protected] (F. Pourboghrat).
0020-7403/$ - see front matter ? 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijmecsci.2003.11.006
1816 M. Zampaloni et al. / International Journal of Mechanical Sciences 45 (2003) 1815 – 1848
Table 1
Advantages and disadvantages of the stamp hydroforming process
Advantages
Improved drawability Applied pressure delays fracture onset
Low wear rate of tooling Eliminates contact between forming tools
Reduced thinning in ;nal part Pressure aids in the uniform distribution of strains
Signi;cant economic savings Eliminates the need for a female die and heating/curing oven
Environmentally friendly Consolidates multiple stamping operations, eliminates some ;nishing operations and
reduces scrap/waste
Optimized processing Better ;nish and all inclusive heating/forming/curing
Complex shapes Forms complex parts with convex contours
Disadvantages
Cycle time Longer than traditional stamping
Temperature challenges DiIcult to maintain isothermal conditions throughout processing
Fluid pressure–punch stroke Optimal path may be part-speci;c
with the necessary tooling and the processes require a fair amount of ingenuity and skill to bring
the economic costs down to a suitable level. Numerous studies have been conducted on traditional
sheet stamping methods such as mechanical stretch forming and deep drawing, both methods that
require a male and female die for the proper forming of a ;nished part. McClintock [1] and Rice
and Tracey [2] conducted studies on sheet metals demonstrating a rapid decrease in fracture ductility
as a hydrostatic pressure, applied across the material, was increased. Clift et al. [3] and Hartley et
al. [4] demonstrated that for sheet metals, the use of a hydrostatic pressure prevented the initiation
and spreading of microcracks within the metallic material. Based on the success found using a
hydrostatic pressure to delay the onset of fracture, the idea of stamp hydroforming was investigated
both experimentally and numerically as a possible method for shaping aluminum and ferrous sheet
metals. The process of stamp hydroforming, unlike conventional stamping, involves supporting the
bottom of the sheet with a bed of viscous 4uid during the stamping process. This external support
provides a through-thickness compressive stress that delays the onset of tensile instabilities as well
as reduces the formation of wrinkles due to tensile frictional forces.
The advantages of the stamp hydroforming process are numerous and the process is receiving
signi;cant attention from both the automotive and aerospace industries. Advantages, illustrated in
Table 1, include improved formability of the blank due to the applied pressure by the 4uid, low wear
rate of dies and punch, a better distribution of plastic deformation when compared to conventional
stamping, signi;cant economic savings associated with the decreased tooling, and the potential for
reducing the amount of ;nishing work required [5].
A schematic of the hydroforming process, as shown in Fig. 1, represents a part that is being
formed by a hemispherical punch. At the start of the process the sheet is placed across the clamping
mechanism, as shown in Fig. 1(1). Fig. 1(2) shows the upper 4uid chamber lowered and the sheet
clamped securely between the two die halves, creating a seal for the upper 4uid chamber. The 4uid
is injected into the chamber and then pressurized. At this point in the process the punch begins to
travel into the sheet forcing it to take the hemispherical shape, ;nally deforming into a fully formed
M. Zampaloni et al. / International Journal of Mechanical Sciences 45 (2003) 1815 – 1848 1817
Fig. 1. A schematic of the stamp hydroforming process with a hemispherical punch. (1) Material is placed on the draw
bead with the 4uid chamber in the raised position. (2) Fluid chamber is lowered and ;lled. Material held in place by
clamping mechanism. (3) Punch begins to move upward. Fluid pressure in chamber is controlled to force the material to
conform to the shape of the punch. (4) Fluid is drained. The 4uid chamber is raised and the punch is lowered to remove
the part.
part after the punch penetrates deeper into the blank, Fig. 1(3). As the punch begins to deform the
material the volume in the upper 4uid chamber decreases thereby causing the pressure to increase
due to its incompressible nature. This 4uid pressure serves two main purposes. First, it delays the
onset of material fracture as described earlier [1–4] and secondly, it forces the material to conform
much more closely to the shape of the punch than the parts formed without the resisting 4uid. The
latter causes a larger volume of the sheet to deform uniformly at a lower punch stroke, thus delaying
excessive thinning and postponing the onset of material fracture. Once the punch has reached the
prescribed draw depth, the 4uid is drained and the upper chamber removed, Fig. 1(4).
2. Hydroforming challenges
The challenges that are present during the stamp hydroforming process can be classi;ed into
two broad categories: material and 4uid pressure. The material challenge refers to the choice and
behavior of the sheet metal. One of the major obstacles concerns the delicate balance between
the 4uid pressure and the ductility of the material chosen for the hydroforming process. The 4uid
pressure needs to be high enough to stretch and bend the work piece through its radius of curvature
to conform to the shape of the punch, yet the material needs to be ductile enough to form without
rupturing.
Fig. 2. A generic curve illustrating the optimum 4uid pressure–punch stroke path for the stamp hydroforming process.
The second challenge is the relationship between the 4uid pressure and the punch stroke during
the process. As shown by Yossifon and Tirosh [6–10] 4uid pressures within the upper 4uid chamber
that are too high will cause the material to bend to the radius of curvature of the punch much
faster than the ductility of the material may allow. This will lead to premature rupturing of the sheet
metal. On the other hand, if the 4uid pressure is too low the sheet may not stretch enough during
the process and wrinkle. Therefore, there is the need to establish an upper and lower limit on the
4uid pressure, as it relates to the punch stroke, to determine an optimum 4uid pressure–punch stroke
path to ensure limited rupturing and wrinkling failures of the ;nished part. A generic curve is shown
in Fig. 2 to help illustrate this idea.
In the stamp hydroforming of sheet metals the diIculty lies in ;nding this appropriate 4uid
pressure–punch stroke path while avoiding rupture and wrinkling instabilities. Lo et al. [11] and Hsu
and Hsieh [12] performed a series of experiments and analyses that established this 4uid pressure–
punch stroke path for the stamp hydroforming of metallic hemispherical cups.
In this paper, the stamp hydroforming experiments that were conducted with 3003-H14-aluminum
sheet alloy and a common ferrous sheet metal purchased o7 the shelf from a local supplier will be
described. The signi;cance of this work is in its thoroughness in studying stamp hydroforming both
experimentally and numerically, as described below:
(A) Experimental studies:
(1) Types of deformation:
(a) Pure stretch.
(b) Draw-in.
(c) Combination of draw-in and pure stretch.
(2) Types of pressure loading:
(a) Fluid pressure applied to one surface of the sheet metal:
(i) Constant 4uid pressure.
(ii) Varying 4uid pressure.
(b) Pressure applied at the punch–sheet interface, using a vinyl sheet.
(B) Numerical studies:

(1) Simulating sheet hydroforming process, with the explicit dynamic ;nite element analysis
(FEA) code Ls-Dyna 3D, using the above experimental conditions.
(2) Comparing predictions of the von Mises isotropic yield function against Barlat’s 1989
anisotropic yield function that was developed for aluminum sheets.
In the following sections the experimental work will be described, followed by the numerical anal-
ysis. In addition, some conclusions will be made and research directions for future papers will be
discussed.
3. Experimental work
3.1. Experimental apparatus
The experimental apparatus was built around an Interlaken 75 double action servo press, shown
in Fig. 3a, where the double action refers to the clamping mechanism moving independently of
the punch mechanism. The ability to independently control both the clamp and the punch a7ords
the opportunity for various modi;cations of the experimental procedure. The experimental process
started with a few modi;cations that were made to a limiting dome height (LDH) test setup, including
drilling ports for pressure measurement, removing the air during the ;lling process and for ;lling
and draining the 4uid chamber. Fig. 3b shows a picture of the in-house designed die that was used
to study the stamp hydroforming process.
Attached to the 4uid line is a regulator/controller that is used to accurately control the 4uid
pressure within the 4uid chamber as the sheet is deformed by the punch, as shown in Fig. 4. If the
pressure is too high, based on a user-de;ned pro;le, then the pressure in the system is reduced to
the appropriate level. If the pressure is too low then the regulator pulls additional pressurized 4uid
from a pressure vessel that is in-line with the rest of the system. A pressure intensi;er is used to
supply the necessary volume and pressure to the reservoir prior to the start of the hydroforming
process.
The experimental procedure used vegetable oil as a resisting 4uid. Due to its incompressible
nature, as the punch began to deform the sheet metal, the volume in the 4uid chamber decreased
causing the pressure to increase. In the case of the 4uid pressure being applied to the topside of the
sheet, the 4uid pressure served two main purposes. Firstly, it delayed the onset of material fracture
as described earlier [1–4]. Secondly, the 4uid pressure forced the material to conform much more
closely to the shape of the punch than the parts formed without the resisting 4uid. The latter also
caused a larger volume of the sheet to deform plastically at a lower punch stroke, thus delaying
shear localization and postponing the onset of fracture.
3.2. Hydroforming with hemispherical punch
The experimental setup was used to form 101:6 mm (4 in) diameter hemispherical cups using
3003-H14-aluminum alloy sheet and a common ferrous sheet material purchased o7 the shelf from
a local material supplier. The experiments that were conducted could be broken down into two
major categories: 4uid pressure applied from one side of the material, and a localized punch–sheet
contact region pressure applied using a stretchable vinyl sheet. It should be mentioned that in all
Fig. 3. (a) The modi;ed Interlaken servo press 75 used for the stamp hydroforming process. (b) the in-house designed
die set for the stamp hydroforming process with 4uid applied form one side of the draw blank [13].
the experiments the punch contacted the bottom of the sheet and the 4uid pressure was applied to
the top surface of the sheet (see Fig. 4).
Three di7erent types of experiments were performed using the 4uid pressure applied from one
side of the sheet metals:
(1) Evaluation of pure stretching.
(2) Allowing the material to draw-in.
(3) Combination of stretch and draw-in.
Each of the above three di7erent categories were further evaluated using the following pressure
loading:
(a) Constant 4uid pressure, Fig. 5a.

Fig. 4. The regulator and controller system used for sheet hydroforming from one side.
Constant Fluid Pressure Variable Fluid Pressure

Fluid Pressure (MPa )(Kn/mm^2)
3 16
14
2.5
12
Pressure (MPa)
2
10
1.5 8
6
1
4
0.5
2
0 0
0 5 10 15 20 25 30 35 40 0.000 5.000 10.000 15.000 20.000 25.000
(a) Punch Displacement (mm) (b) Punch Displacement (mm)
Fig. 5. Experimental 4uid pressure curves. (a) Constant, (b) varying 4uid pressure pro;les.
(b) Varying 4uid pressure, Fig. 5b.

(c) Localized hydrostatic pressure using a stretchable vinyl sheet.
Therefore, a total of nine di7erent combinations of experiments were conducted to evaluate the
sheet hydroforming process. Below, the results and signi;cant ;ndings of each test will be
discussed.
Fig. 6. Example of material sag in unsupported regions when a constant 4uid pressure is applied on one side of the draw
blank material (gap exaggerated to illustrate e7ect).
3.3. Constant 3uid pressure applied to one-side of the sheet
3.3.1. Pure stretch experiments

For the experiments conducted under pure stretch conditions a 1 mm (0:04 in) thick, square blank
(178 mm × 178 mm (7 in × 7 in)) was placed over a draw bead and clamped with a blank holding
force (BHF) of approximately 267 kN (60; 000 lbf ). To ensure no material draw-in, several tests
were conducted where the edge of the sheet on either side of the draw bead were marked with a
thin pencil before forming. After the forming, the location of these lines with respect to the draw
bead was checked for any possible movement. In all the tests it was veri;ed that the marked lines
did not move at all, con;rming the pure stretch condition of the sheet metal. After clamping the
sheet, the 4uid chamber was ;lled and then given an initial pressure up to 2758 kPa (400 psi).
As the pressure within the chamber increased, the sheet bulged towards the punch (away from
the 4uid chamber) prior to the punch beginning its movement into the sheet metal. This bulging,
schematically illustrated in Fig. 6, created a strain concentration around the rigid die corner, which
had a radius of curvature of 6:1 mm (0:24 in). With the constant 4uid pressures above 3448 kPa
(500 psi), the material sheared o7 at the sheet/die corner interface prior to the punch moving into
the 4uid chamber.
Maintaining the 4uid pressures below the critical 3448 kPa (500 psi) level led to increased draw
depths for the 3003-H14-aluminum alloy as illustrated in Fig. 7. Experiments were conducted at
Punch Force versus Displacement for 3003-H14 Aluminum Sheets under a constant fluid
pressure applied from one side of the material
40000
35000
30000
Punch Force (N)
25000
20000
2758 kPa
15000 2069 kPa
1370 kPa
10000
690 kPa
5000 0 kPa
0
0 5 10 15 20 25 30 35
Displacement (mm)
Fig. 7. Punch force versus displacement for 3003-H14-aluminum alloy sheet metal using a constant 4uid pressure applied
form one side of the draw blank, pure stretch experiments.
several pressure levels in order to quantify the upper bound of the 4uid pressure/punch stroke diagram
for the constant 4uid pressure, pure stretch experiments. Maintaining a constant 4uid pressure allowed
for an impressive increase in the forming depth of 12–31% over parts that were formed without the
resisting 4uid (i.e., conventional stamping). This improved formability could be attributed to several
factors, but is mostly caused by changes in the boundary conditions. One explanation could be that
when the sheet bulges in one direction (e.g., toward the punch) followed by a deformation in the
opposite direction, the in-plane and bending strains in the sheet will reverse, causing the sheet to
work harden. Depending on the amount of the work hardening, the resistance of the sheet to failure
will increase. Also, this reverse bending and stretching causes the entire sheet metal in the die cavity
to deform plastically and therefore strain localization over the punch surface will be delayed. Another
reason for the improved formability could be that when the initial bulging occurs it creates more
material in the die cavity to be deformed by the punch (see Fig. 6), in comparison with conventional
stamping where the length of the sheet in the die cavity is shorter (see Fig. 1(1)). As is discussed
in Section 5, this improved draw-depth was also observed in the numerical modeling.
3.3.2. Draw-in experiments

By reducing the blank holding force (BHF), the sheet was allowed to draw into the die cavity.
For these experiments a 4uid pressure of 483 kPa (70 psi) was applied to the sheet. The goal was to
study the wrinkling and tearing characteristics of the sheet formed with (hydroforming) and without
the 4uid pressure (conventional stamping) applied on one side of the material. As the sheet draws
into the die cavity, the compressive stresses developing in the sheet under the blank holder and
Fig. 8. Wrinkling in the aluminum sheet after draw-in. (a) No 4uid pressure, (b) 70 psi 4uid pressure applied to the
topside of the sheet.
in the unsupported region (between the punch–sheet interface and the die corner region) increases,
resulting in the eventual wrinkling of the sheet metal in these regions.
Figs. 8a and b show the wrinkles that form in a sheet formed without (conventional stamping) and
with a 4uid pressure of 483 kPa (70 psi). The wrinkles initiate in the 4ange area (under the blank
holder) and gradually extend toward the center of the cup, as the draw depth increases. It was noted
that although the loading was symmetrical, the wrinkles occurred along one side of the sheet metal.
It was later found that this was caused by the anisotropic properties of the 3003-H14-aluminum alloy
sheet, as shown in Table 4. This table shows that the values of the parameter (R=t =w ) are di7erent
along the rolling (R0 ), transverse (R90 ) and 45◦ direction (R45 ). The R values for an isotropic sheet
metal would be 1.0 everywhere.
Overall, the application of the 4uid pressure from one side of the sheet did not help with the
wrinkling problem. In fact, to some extent the application of the 4uid pressure exacerbated it by
creating larger compressive stresses, as the unsupported sheet was pushed over the surface of the
punch. The wrinkling problem became proportionally worse with the increase in the 4uid pressure
applied to one-side of the sheet.
3.3.3. Combination experiments

In the combination experiments, the sheet was ;rst drawn into the die cavity with a low 4uid
pressure of 483 kPa (70 psi), followed by locking the boundary and stretching the sheet until it
ruptured. During both the draw-in and combination experiments the presence of this initial pres-
surization caused the shear localization to occur earlier and led to the shearing of the sheet metal
as the punch moved into the 4uid chamber. Therefore, for these experiments rupture occurred at
shallower draw depths than those parts formed without the presence of a resisting hydrostatic 4uid.
The reason for the early rupture is that suIcient initial bulging of the sheet metal did not occur (as
in Section 3.3.1) to create additional material in the die cavity to help with the punch deformation.
Also, because of the extra tension that is created in the sheet by the 4uid pressure, excessive thinning
occurs in the punch–sheet contact region, resulting in an earlier failure in comparison with the case
of no 4uid pressure (conventional stamping) [14].
P unch Force versus D isplacem ent for V arying Fluid P ressure,

P ure S tretch E xperim ents
120000
110000
100000
90000
80000 Fe w/pressure
Punch Force (N)
70000 A l w/pressure
60000 Fe no pressure
A l no pressure
50000
40000
30000
20000
10000
0
0 5 10 15 20 25 30 35 40
D isplacem ent (m m )
Fig. 9. Punch force versus displacement for both the Al and Fe sheet materials using a varying 4uid pressure applied
from one side of the sheet blank, pure stretch experiments.
3.4. Varying 3uid pressure applied to one-side of the sheet
3.4.1. Pure stretch experiments

The goal of the varying 4uid pressure experiments was to try to delay the occurrence of the strain
localization by gradually increasing the pressure in the 4uid chamber (see Fig. 5b) as the punch
deformed the sheet, while maintaining an upper pressure bound of 2758 kPa (400 psi). The main
obstacle with these experiments was the control of the 4uid pressure. At times, the 4uid pressure
was found to spike at levels that were over twice the set boundary level of 2758 kPa (400 psi).
Though these spikes lasted for only milliseconds they were long enough to impart signi;cant stress
concentrations to the material. Several experiments were successfully run with both aluminum and
ferrous sheet metals, with overall results for the pure stretch case illustrated in Fig. 9. Parts that were
being formed using an applied varying hydrostatic 4uid pressure were rupturing at shallower punch
depths than those parts formed without any resisting 4uid pressure. These premature ruptures were
primarily due to excessive thinning of the sheet metal, caused by the extra tension created by the
applied pressure. The higher the 4uid pressure was, the earlier the sheet failed in these experiments.
3.4.2. Draw-in experiments

When the part was not rigidly clamped the material was allowed to draw into the 4uid chamber
as the part was formed. An interesting trend was found during this procedure. As the material was
drawn to a depth of approximately 12:7 mm (0:5 in) some wrinkling started to occur on the cup
surface. To try to eliminate this instability it was attempted to utilize the 4uid pressure to iron
out the wrinkles. A series of experiments were run and the results showed that the wrinkles were
not ironed out; in fact as the 4uid pressure was increased the wrinkling became more pronounced.
This was attributed to the increased compressive stresses in the hoop direction caused by the 4uid
pressure pushing the sheet over the punch surface, exacerbating the already wrinkled part.
3.4.3. Combination experiments

The third case evaluated with varying 4uid pressure was the investigation into the e7ects of a
combination of pure stretch and draw-in during the stamp hydroforming process. The material was
allowed to draw-in to a depth of 12:7 mm (0:5 in). As explained before, at this depth the material
that was drawn into the chamber began to exhibit wrinkling characteristics. During the next step of
the process the material was tightly clamped and 4uid introduced and pressurized. The goal was to
use the 4uid pressure to ;rst counteract the movement of the material into the chamber to delay the
onset of fracture while also ironing out the wrinkles that formed during the draw-in process.
So far the preliminary results have shown some promise but further minor modi;cations will need
to be made to the procedure to ensure a consistent wrinkling pattern during the draw-in phase. Again,
due to the 4uid pressure spikes and the unsupported regions the results were not as anticipated, parts
were either rupturing at shallower draw depths or the wrinkling that was occurring during the draw-in
phase were not removed but became more pronounced.
3.5. Localized hydrostatic pressure
In an attempt to explore di7erent design ideas, the process was conducted using a thin vinyl sheet
in place of the counteracting 4uid. The vinyl material was a stretchable material that was used to
simulate the e7ect of a localized hydrostatic pressure. As the punch moved into the sheet metal
the vinyl counteracted the motion and added a pressure at the location of the sheet metal that was
in contact with the punch. The goal of these experiments was to investigate the overall e7ect the
pressure has on the unsupported regions of the material during processing. Since the vinyl only
provided a counteracting force over the area of the material being shaped by the punch there was
no reverse bulging of the material within the unsupported regions of the sheet metal.
Fig. 10 illustrates the results achieved through the use of the localized pressure for the 3003-H14-
aluminum alloy sheets, while Fig. 11 represents the results for the ferrous sheets. In both cases the
use of the localized pressure increased the draw depth by 3–10%. This may indicate the advantages
of applying the 4uid pressure where only the punch and the sheet come into contact and will be
discussed further in Section 3.6.
By coating the punch with a thin layer of grease prior to forming, the surface area that contacts the
vinyl could be quanti;ed. Using this value an approximate equivalent pressure could be calculated
using the relationship between the forces applied by the punch and the surface area (P = F=A). For
the single vinyl sheet this was calculated as 148 kPa (21:5 psi), while the use of two vinyl sheets
was calculated as being equivalent to applying a 4uid pressure of 159 kPa (23 psi). Since the vinyl
sheet had no e7ect on the sheet metal that was not in contact with the punch then it was concluded
that this vinyl sheet was applying an equivalent 4uid pressure locally instead of globally. Therefore,
based on the increased draw depths achieved using a relatively low localized pressure, it could be
assumed that increases in the 4uid pressure, applied locally, should result in better formability of
the sheet metals.
Punch Force versus Displacement for 3003-H14 Aluminum using Localized

Hydrostatic Pressure, Pure Stretch Experiments
20000
19000
No Vinyl
Punch Force (N)
18000
1 Vinyl Sheet
2 Vinyl Sheets
17000
16000
15000
19 20 21 22 23 24 25 26 27
Displacem ent (m m )
Fig. 10. Force versus displacement for 3003-H14-aluminum sheets with a hydrostatic force applied by vinyl sheets, pure
stretch experiments.
Force versus Displacement for Ferrous Sheet Metal, Localized Hydrostaic

Pressure Experiments
49000
48000
47000
46000
Punch Force (N)
45000
44000
No Vinyl Sheet
43000
1 Vinyl Sheet
42000
2 Vinyl Sheets
41000
40000
32 33 34 35 36 37 38 39
Displacement (mm)
Fig. 11. Force versus displacement for ferrous sheet metal with a hydrostatic pressure applied by vinyl sheets.
Localized Hydrostatic Pressure Experiments using 3003-H14 Aluminum Alloy Metal

Blanks, Combination Experiments
20500
19500
18500
17500
16500
Punch Force (N)
15500
14500
13500
12500 Al No Vinyl
11500 Al 1 Vinyl
10500 Al 2 Vinyl
9500
8500
7500
12 14 16 18 20 22 24 26
Displacement (mm)
Fig. 12. Results for the combination experiments performed on the 3003-H14-aluminum alloy sheet, using vinyl sheet to
apply a localized pressure at the punch–sheet interface.
For the draw-in experiments the vinyl sheets showed no discernible trend. The vinyl requires a
tight clamp at the edges of the sheet so that as the part is being formed the vinyl counteracted this
movement. Since the part was allowed to draw-in, the vinyl material moved along with the part and
therefore applied no counteractive pressure as the part was formed.
The use of the localized pressure was also investigated during the combination experiments. Results
for the aluminum alloy could be found in Fig. 12, while the results for the ferrous sheet are shown in
Fig. 13. For both materials increased draw depths of approximately 1.5 –3% were achieved through
the use of the small-localized pressures.
3.6. Fluid pressure applied from both sides of the sheet material
A new experimental die, illustrated in Fig. 14, was designed that o7ers several advantages over
the existing die design. The new die ;lls the sealed bottom chamber with 4uid that is equalized
with the 4uid in the upper chamber. As the pressure in the upper 4uid chamber increases due to the
volume change, the displaced 4uid will be forced into the bottom chamber thereby equalizing the
pressure between the chambers.
This equal pressurization on both sides of the sheet metal allows for the support of the material
that is not in contact with the punch and prevents the sheet from an uncontrolled bulging in these
regions. On the other hand, the new die design would still be able to o7er the increased formability
caused by reverse loading the sheet (see Fig. 7), through creating a di7erential pressure between the
top and bottom chambers. In the case of equal pressurization, the material that is in contact with the
punch will experience a pressure that is representative of the localized hydrostatic pressure applied
Localized Hydrostatic Pressure Experiments using Fe Sheet Metal Blanks,

Combination Experiments
55000
50000
45000
40000
Punch Force (N)
35000
30000
Fe No Vinyl
25000
Fe 1 Vinyl
20000 Fe 2 Vinyl
15000
10000
12 14 16 18 20 22 24 26 28 30 32 34 36 38
Displacement (mm)
Fig. 13. Results for the combination experiments performed on the ferrous sheet, using a vinyl sheet to apply a localized
pressure at the punch–sheet interface.
Fig. 14. The dual 4uid pressure die design that allows the pressurized 4uid to be applied on both sides of the sheet
metal [13].
Fig. 15. Hydroforming process with a blank holding support [12].
Fig. 16. Hemispherical forming process [12].
by the vinyl sheet, resulting in deeper draw depth prior to material failure. This new die design
should also be able to use 4uid pressures above the limiting 3848 kPa pressure with the current one-
sided die design. Results from the experiments conducted with this new die, 4uid from both sides
of the sheet metal, will be presented in future papers.
The 4uid pressure-spiking problem discussed before will be alleviated through the implementation
of a constant 4uid pressure reservoir in-line with the current regulator/controller. This constant source
will allow for better tuning of the controllers and will help eliminate any spikes that could occur
outside the error of the physical devices.
4. Theoretical analysis
The limit theorem of plasticity has been used by Hsu and Hsieh [12] to study the hemispherical
punch hydroforming process. Following their analysis, the limit (failure) pressure for the hemispher-
ical punch has been determined for the experiments described in Section 3.4.
A geometrical interpretation of the hydroforming process is shown in Figs. 15 and 16. The work-
piece under consideration is divided into three zones according to the deformed shape: The re-
gion where the workpiece stays contact-free from the die (zone I), the unsupported area (zone II),
and where the workpiece has been compressed onto the surface of the punch (zone III). In their
analysis, Hsu and Hsieh [12] used the Tresca yield criterion and considered the normal anisotropy
of the material. Below, the set of equations derived by Hsu and Hsieh [12] for hemispherical sheet
hydroforming will be given. For the axisymmetric problem, the polar equilibrium equation in the
rim area is
d t
(tr ) + (r − ) + f(P1 ) = 0; (1)
dx r
where r and are radial and hoop stresses, respectively, r is the polar coordinate, t is the thickness,
P1 is the blank holding pressure, f(P1 ) = 2P1 is friction stress, and is the coeIcient of friction.
From the incremental theory of plasticity due to Hill [15], equivalent strain-rate is
1=2
1+R 2 2R 2
˙e = √ (˙r ) + ˙r ˙ + (˙ ) ; (2)
1 + 2R 1+R
where ˙r ; ˙ correspond to strain rate components and ˙e is the e7ective strain rate, and R is the
normal anisotropy parameter. Solving the equilibrium equation for zone I:
b b
f(P1 ) 0 (rI )n Rn=2
e
rI (r) = dr + dr; (3)
a t a r

G(r; h; )
rI = ln ; (4)
r
where
 2 1=2
 r 2 2 

 − 1+ cos 


 a a 


 


 2 


 + 2(1 − sin ) 1 − 

 a 
G(r; h; ) = a ; (5)

 


 + 2 1 + − cos 


 a a 2 







 h h 


 +2 − − 1 HV − −1  
a a a a
 

 h 

a+−h
 − − 160

−1 a a
= sin when ; (6)
a+ 
 h 
 − − 1¿0

a a
2(1 + R)
Re = ; (7)
1 + 2R
0 is a material constant (strength coeIcient), geometrical parameters ; a; b and h are de;ned in

Fig. 16, and the Heaviside unit step function is
 
  h 

h  1 when − − 1 ¿ 0  
a a
Hv − −1 = : (8)
a a 
 h 
 0 when − − 1 6 0  
a a
For zone II, with the continuity condition:
(a+) cos
0 (rII )n Rn=2
e
rII (r) = rI [r = (a + ) cos ] + dr; (9)
a r
F(r; h; )
rII = ln ; (10)
r
where
 1=2

 1 − sin + 1 + − − cos 


 a a 2 

 
√  
F(r; h; ) = 2a + (sin − cos ) (11)



a




 h h 

 + − − 1 HV − −1 
a a a a
and

−1 (a + ) cos − r
= sin : (12)

When combining the instability criterion of anisotropic material under biaxial plane strain condition
and the power–law hardening relationship, the critical 4uid pressure causing rupture in the workpiece
can be determined by equating the critical stress to Eq. (9), which can be expressed as
 
 2P1 [b − (a + ) cos ]
 


 0 t 


 


n+1   b n=2 G n 

1+R R e ln r
n n
√ = + dr (13)
1 + 2R 
 (a+) cos r 


 (a+) cos n=2 


 Re ln Fr
n 


 + dr 

a cos r
and
 n+1 1=2
n √1+R
 2n 1+2R 
= 1 +  − 1; (14)
a PQ
where PQ is the normalized blank holding pressure. To solve Eq. (13) for the experiments described
in Section 3.3, a general numerical method was used to calculate for the limit (failure) pressure as
follows: The Newton–Raphson method was used to solve for the value of the 4uid pressure as
f(PQ n )
PQ n+1 = PQ n − (15)
f (PQ n )
the derivative of the function was taken using

f(x + ) − f(x)
f (x) ≈ ; (16)

where is a small number (∼ 10−6 ). The Secant method was also used to ;nd the new value of
the pressure to avoid the calculation of the derivative and the results were similar. The value of the
integrals of F and G in Eq. (13) was calculated using the Gauss–Legendre quadrature with (n = 4)
weight points as follows:
1 n
f(x) d x ≈ wj f(xj ); (17)
−1 j=1
where
b
1

b−a a + b + (b − a)x
f(t) dt = f d x: (18)
a 2 −1 2
The following values for the constants were used in the calculations (taken from the experimental
setup) (Table 2):
Fig. 17 shows the predicted limit (failure) 4uid pressure from the numerical analysis and the
variable 4uid pressure applied in the actual experiment. Rupture in the sheet occurred when the
Table 2
Data input to numerical analysis
n R 0 y BHF t A b
0.0476 0.89 0.13 95:0 MPa 202:4 MPa 482:2 MPa 1 mm 50:8 mm 88:9 mm
(BHF = blank holding force).
Fluid Pressure Path
20.0
18.0
16.0
14.0
Fluid Pressure (Mpa)
12.0
10.0
8.0
6.0
4.0
2.0 Exprimental
Numerical Analysis
0.0
0.0 5.0 10.0 15.0 20.0 25.0 30.0
Punch Stroke (mm)
Fig. 17. Experimental 4uid pressure compared against numerical analysis from Ref. [12].
pressure crossed the limiting “failure” pressure curve. This compares well with the results shown
in Fig. 32, for varying 4uid pressure case. Although the analytical model developed in Ref. [12]
can be used as a guide for choosing the limit (failure) 4uid pressure, its application is limited
to hydroforming of hemispherical cups formed by varying 4uid pressure. Furthermore, its results
cannot be applied to experiments conducted in Section 3.3, where constant 4uid pressure is applied
throughout the forming process, which resulted in maximum deformation prior to rupture (e.g.,
compare Figs. 7 and 32).
Therefore, to predict the limiting (failure) 4uid pressure for hydroforming of complex parts, the
FEA method must be used. In this paper, FEA of hydroforming of a 3003-H14-aluminum alloy
sheet was carried out using Barlat Yield 89 anisotropic yield function. This numerical analysis will
be presented next.
5. Finite element analysis
An important goal of this research is to determine an optimum method of production for a new
product. To achieve a better understanding of the deformation of the sheet metal during the forming
process, numerical analysis was employed. The goal was to aid in the prediction of the ;nal part
geometry, compare results against experimental data and to reduce the amount of trial and error asso-
ciated with the experimental aspect of the work. For the stamp hydroforming process the numerical
study was performed using the explicit ;nite element code, Ls-Dyna3D [17].
Two types of ;nite element models were used for this study: quarter- and full-model. Two types
of element were used for the blank, shell elements with 7 integration points through the thickness
and solid elements consisting of ;ve layers with 1 integration point through the thickness. The initial
models were created using UnigraphicsJ and imported as IGES ;les. HypermeshJ was then used
to create the ;nite element mesh, assigning the boundary conditions, and build the Ls-Dyna input
deck for the analysis. The full-size ;nite element shell model used approximately 10,500 four-noded
shell elements, while the quarter-model, illustrated in Fig. 18, used 4600 shell elements.
The punch, die, and the blank holder were created using rigid material. The sheet blank was ini-
tially modeled using the isotropic material number 18 in Ls-Dyna (*MAT POWER LAW
PLASTICITY). This is the von Mises yield function de;ned as follows:
! = 12 Sij Sij − 13 Q 2 ; (19)
where Sij is the deviatoric stress, and Q is the hardening law.
This material uses the Hollomon power–law hardening rule de;ned as
Q = k Q n ; (20)
where “k” is the strength coeIcient, and “n” is the work hardening exponent.
To determine the hardening coeIcients for the AA3003-H14 material, a uniaxial tensile test was
carried out. The results are reported in Table 3.
Fig. 19 shows the experimental true stress–strain curve for the material and the curve ;t with the
power–law hardening model using the calculated values of “k” and “n”.
The fully integrated shell element (ELEFORM #16) with 7 integration points through the thickness
was used in the modeling of the blank. To prevent draw-in, the blank holder was given a constant
Fig. 18. A quarter-model of the hydroforming process using a hemispherical punch.
Table 3
Aluminum alloy, 3003-H14, properties
3003-H14 y (yield stress) E & (Poisson ratio)
150:6 MPa 69 GPa 0.33

k n Thickness (t)
202:2 MPa 0.0476 1 mm
force applied to its center of gravity (250 kN (56 kip)). Since the part has symmetric features, a
quarter-model was used for the same process, as shown in Fig. 18. In the quarter-model the edge
of the material on both forming axes (x; y) were constrained to enforce symmetry, while in the full
3D model no constraints were applied to the sheet.
Forming limit diagrams (FLD) displaying contour plots of the minor and major strains were used
to determine the locations, and the punch height, at which the sheet metal would fail due to tearing.
The Ls-Dyna code’s new post-processor has the ability to calculate the FLD plot and to project them
on the desired shell elements. This allows the location of the failure points to be easily determined.
To reduce the calculation time while maintaining accuracy, an adaptive meshing scheme was used.
In this scheme only those areas coming into contact are re;ned. The re;nement scheme uses an
angle change re;nement method with a maximum re;nement level of three. In addition, a one-pass
adaptivity with an approach criterion is used.
True Stress Vs. True Strain

y = 202.2ε 0.0476
200
180
160
140
True Stress (MPa)
120
100
80
60
40
20
0
0.00 0.01 0.02 0.03 0.04 0.05 0.06
True Strain (mm/mm)
Fig. 19. Punch force versus punch displacement, using k and n values from Table 3.
Punch Velocity
2.5
2
Velocity (mm/ms)
1.5
0.5
0
0 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 30
Time (ms)
Fig. 20. Punch velocity used for the ;nite element simulation.
The main goal of the numerical analysis was to verify the predictions of the hydroforming model
for the aluminum alloy. Once the accuracy of the numerical model is established, it could be used
as a design tool for other materials. Various examples were run including pure stretch (stamping),
draw-in, constant and varying 4uid pressure applied to one side of the sheet. Each of these cases
will be discussed next.
5.1. No 3uid pressure (pure stretch)
For the traditional stamping case the punch was given a trapezoidal velocity pro;le to ;t the curve
shown in Fig. 20 with a maximum punch speed of 2 mm=ms. The results of both the FLD-numerical
Fig. 21. Sheet metal forming without pressure (conventional stamping).
and experimental analysis for sheet stamping without pressure are shown in Fig. 21. The experimental
result shows that the material failure occurs at a stamping depth of 24:6 mm (0:969 in). For the
numerical result the FLD contour plot was used to determine the failure point. From the plot it was
determined that the initial failure would occur around the center of the sheet at a punch depth of
23:6 mm (0:93 in). This represents a 4% discrepancy between the experimental and the numerical
results, which for all practical purposes is acceptable.
Fig. 22 shows the punch force–displacement for the experimental and numerical analysis up to
the initial failure. It was noted that the experimental punch force was slightly higher than the
numerical predictions. The load cell used in the experimental investigations has a capacity of 311 kN
(70; 000 lbf ) with an error of 1%. If an error of less than 0.5% is assumed (1500 N (337 lbf )), and
this error is subtracted from the experimental data, then we get the results shown in Fig. 22 (Adjusted
force), which matches the numerical predictions very well.
5.2. Constant 3uid pressure applied to one-side of the sheet (pure stretch)
In the experimental procedure, the punch was ;rst raised until it touched the bottom of the sheet
and then stopped. Then, the 4uid pressure was increased to a desired value and kept constant at that
level before the punch was allowed to move up to deform the sheet. To simulate this in the numerical
model, the punch was controlled by the displacement pro;le shown in Fig. 23. The constant pressure
used in the simulation was ramped up over a time period of 7 ms. Similar to the experiments, at the
constant pressure level of 3448 kPa (500 psi), the numerical model predicted that the sheet would
fail, before the punch moves, along its corner radii due to strain localization (see Fig. 24).
In the experiments with a constant 4uid pressure of 2758 kPa (400 psi) it was found that the sheet
sometimes fails along the corner radii, as shown in Fig. 25, soon after the punch starts deforming
the sheet. With further investigation, it was found that due to an initial controller–regulator problem
in the experimental setup the 4uid pressure was not being kept at the desired constant level of
Experimental
Numerical Vs. Experimental Without Pressure K=202, n=0.0475
Adjusted Force
20000
18000
16000
14000
Punch Force (N)
12000
10000
8000
6000
4000
2000
0
15
41
69
95
20
43
70
01
0
.2
.4
.7
.9
.2
.4
.7
.0
.2
.5
.7
.0
0.
1.
2.
3.
5.
6.
7.
9.
10
11
12
13
15
16
17
19
20
21
22
24
Punch Displacement (mm)
Fig. 22. Predicted and measured punch force versus punch displacement curve.
Punch Displacement
50.0
45.0
40.0
35.0
Z-Displacement (mm)
30.0
25.0
20.0
15.0
10.0
5.0
0.0
0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00
Time (ms)
Fig. 23. Punch displacement pro;le for one-sided hydroforming simulation.

Fig. 24. FLD results. Failure is predicted at the edges for a 3448 kPa (500 psi) pressure.
Fig. 25. Pressure pro;le shown in Fig. 26 was applied to the topside of the sheet. Failure occurred at a draw depth of
0:32 in.
2758 kPa (400 psi), but instead was ramping up from 2758 kPa (400 psi) to 7000 kPa (1015 psi).
Fig. 26 shows the actual 4uid pressure curve, which was applied to the sheet, resulting in the
premature sheet failure, as shown in Fig. 25. To verify the robustness of our numerical modeling, it
Constant-variable Pressure Profile
8000.0
7000 kPa (1000psi)

7000.0
6000.0
Pressure (kPa)
5000.0
4000.0
2758 kPa (400 psi)

3000.0
2000.0
1000.0
0.0
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0
Time (ms)
Fig. 26. The actual pressure pro;le applied to the topside of a 3003-H14-aluminum sheet alloy.
was decided to apply the same pressure pro;le (Fig. 26) to the blank, modeled with solid elements,
and to calculate the punch height and location where the sheet fails, using plastic strain data. Fig. 27
shows the corresponding plastic strain contours developed on the upper surface (where the pressure
was applied) and the bottom surface of the hydroformed sheet metal. The model predicts that the
aluminum sheet would fail at its corner radii on the top (Fig. 27a). This compares very well with
the actual failure location observed in the experiments (Fig. 25). The numerically predicted punch
height at the failure point was 6:6 mm (0:26 in), which faired well against the experimental punch
height of 8:1 mm (0:32 in).
In pure stretch experiments, with constant 4uid pressure maintained throughout the whole process,
a signi;cant increase in the punch depth, before failure, was noticed (see Fig. 7). The numerical
model also predicts higher punch depths, before failure, when constant pressure is applied to the
topside of the sheet. Fig. 28 shows the numerical punch force–displacement plots, up to the failure
point, for several constant 4uid pressures. It is obvious that compared to conventional stamping
(no 4uid pressure), the hydroformed aluminum sheet reaches signi;cantly deeper punch depths,
before it fails. This increase in the punch depth improves with increasing the 4uid pressure, up to
a maximum value of 2758 kPa (400 psi). Fig. 29 shows the failure punch depth–pressure plot for
both the experiments and the numerical simulations. Fig. 30 shows the experimental and numerical
punch force–displacement curves for the case of constant 4uid pressure of 2758 kPa (400 psi), up to
the failure point. Besides the obvious improvement in the failure punch depth, the pressurized 4uid
also forces the sheet to better conform to the shape of the punch, resulting in a better-formed part
and a possible reduction in the springback, as shown in Fig. 31 (Fig. 32).
Fig. 27. Plastic strain contours, predicted with a solid model, for a 3003-H14-aluminum sheet alloy after the pressure
pro;le in Fig. 26 was applied to its topside. (a) Topside of the sheet, (b) bottom side of the sheet. Results clearly match
the failure location observed in the experiment, Fig. 25.
Numerical Punch Force vs. Displacement for 3003-H14 Aluminum Sheets Under a
Constant Fluid Pressure Applied From One Side of The Material
45000
2758 Kpa (400psi)
40000
35000
2069 Kpa (300psi)
30000
Punch Force (N)
1370 Kpa (200psi)

25000
690 Kpa (100psi)

20000
0 Kpa (0psi)
15000
10000
5000
0
0 5 10 15 20 25 30 35
Punch Displacement (mm)
Fig. 28. Numerical punch force–displacement results for 3003-H14-aluminum alloy sheet at constant 4uid pressures applied
to the topside of the sheet.
Failure Depth Vs. Pressure
35
30
25
Poly. (Numerical)
Failure Depth (mm)
Poly. (Experimental)
20
15
10
0
20 520 1020 1520 2020 2520 3020
Pressure (Kpa)
Fig. 29. Failure depth versus pressure results for 3003-H14-aluminum alloy sheet. Comparison between experiment and
numerical results.
5.3. Varying 3uid pressure
In the case of varying 4uid pressure, the pressure was applied to the top surface of the sheet
incrementally up to 14:4 MPa (2080 psi). In the numerical simulation, the punch was given a velocity
pro;le (Fig. 20), since in the experiments the punch was not stopped during the process of applying
the 4uid pressure. The 4uid pressure pro;le applied in the numerical analysis was taken from the
experimental data (see Fig. 5b). The timing of the pressure pro;le was changed in order for it to be
applied in the numerical analysis. Fig. 30 shows the predicted punch force, before failure, matching
the experimental values very well (3% error).
5.4. Draw-in results
Fig. 33 shows the resulting part when hydroforming experiments with draw-in were performed. It
is clear from Fig. 33, that the material is anisotropic and wrinkling occurs along the rolling direction
of the sheet. Initially in the numerical analysis, the von Mises isotropic material model (material
number 18 in Ls-Dyna: ∗MAT POWER LAW PLASTICITY) was used. Fig. 34 shows the resulting
deformed shape of the sheet using this material model. From this ;gure the isotropic nature of this
material model is clear, since the wrinkles occur along both axes of the material.
Punch Force Vs. Punch Displacement
45000
40000
35000
30000
Punch Force (N)
25000
20000
15000
10000
5000 Experimental
Numerical
0
0 5 10 15 20 25 30 35 40
Displacement (mm)
Fig. 30. Punch force versus punch displacement for the 400 psi constant 4uid pressure.
Fig. 31. Part shape comparison between: (a) conventional stamping (no 4uid pressure), and (b) a hydroformed sheet with
a constant 4uid pressure of 2758 kPa (400 psi) applied to its topside. Besides improved formability, the hydroformed part
conforms to the shape of the punch more closely too, possibly resulting in a reduced springback.
Barlat and Lian [16] developed an orthotropic material model for anisotropic sheet metals with
plane stress conditions. This material model uses the Lankford parameters (R0 ; R45 and R90 ) for the
de;nition of anisotropy as follows:
! = a|K1 + K2 |m + a|K1 − K2 |m + c|2K2 |m − 2Q m ; (21)
where
x + hy
K1 = ;
2

x + hy 2
K2 = + p2 +2xy : (22)
2
The anisotropic material constants (a; c and h) are obtained through the R0 ; R45 and R90 as
follows (17):

R0 R90
a=2−2 ;
(1 + R0 )(1 + R90 )
c = 2 − a;

R0 (1 + R90 )
h= : (23)
R90 (1 + R0 )
The anisotropy parameters (R0 ; R45 and R90 ) are determined from uniaxial tensile tests performed
in 0◦ , 45◦ and 90◦ measured from the rolling direction of the sheet. An iterative method is used to
;nd the “p” value. This particular material model is used in Ls-Dyna to simulate the anisotropic
behavior of the material (Material #36 in Ls-Dyna: ∗MAT 3-PARAMETER BARLAT). The material
properties for the AA3003-H14 used in this simulation are shown in Table 4 below.
Experimental, Variable
Punch Force vs. Displacement
Numerical,Variable
120000
100000
80000
Punch Force (N)
60000
40000
20000
0
0 5 10 15 20
Displacement (mm)
Fig. 32. Punch force versus punch displacement for varying 4uid pressure.
Fig. 33. Experimental results for the draw-in case with 4uid pressure of 70 psi applied to the topside of the sheet.
Fig. 34. Numerical draw-in simulation using an isotropic material model in LS-Dyna.
Table 4
Material properties for 3003-H14 using Barlat model
3003-H14 y (yield stress) E & (Poisson ratio)
150:6 MPa 69 GPa 0.33

k n Thickness (t)
202:2 MPa 0.0476 1 mm

R0 R45 R90
0.500 0.550 0.650
Fig. 35. Numerical draw-in simulation using an anisotropic material model (Barlat YLD-89) in LS-Dyna. A better com-
parison with the experiments (Fig. 33) could be seen.
Fig. 35 shows the resulting deformed shape of the sheet using this anisotropic material model. It
is clear that the numerical model correctly predicts the wrinkling behavior of the aluminum sheet
when utilizing Barlat’s material model.
6. Conclusions
A total of nine di7erent combinations of experiments were conducted to evaluate the sheet hy-
droforming process. These experiments included studying the e7ect of boundary condition (i.e.,
pure-stretch, draw-in and their combinations) and pressure loading (i.e., constant and varying 4uid
pressure and a localized pressure applied using a vinyl sheet) on the deformation of 3003-H14-
aluminum sheet alloy. Overall, the experimental results show that the sheet hydroforming process
is a viable processing method that deserves some additional attention based on the signi;cant ad-
vantages that it provides in potential cost savings (no female die needed), improved formability
and part shape accuracy. Further experimentation is needed, however, in order to validate the die
design that accounts for the dual 4uid pressures applied to the upper and lower surfaces of the sheet.
In addition, the investigation into the wrinkling and rupturing characteristics needs to be continued
and a characteristic punch stroke/4uid pressure path should be determined to minimize the occur-
rence of these instabilities. The numerical analysis of the hydroforming process with the commercial
;nite element code Ls-Dyna 3D seems to capture the failure and wrinkling characteristics of the
3003-H14-aluminum sheet alloys very well. The accuracy of the numerical predictions, however,
is very sensitive to the material properties of the sheet metal. Accurate material characterization
methods are required to ensure that the model can accurately capture the deformation behavior of
the material.
Acknowledgements
The authors wish to thank the Manufacturing Research Consortium at Michigan State University
and its member companies GM and Ford for providing the funding support for this research work.
References
[1] McClintock FA. A criterion for ductile fracture by the growth of holes. Journal of Applied Mechanics 1968;35:
363–71.
[2] Rice JR, Tracey DM. On the ductile enlargement of voids on triaxial stress ;elds. Journal of Mechanical Physics
and Solids 1969;17:201–17.
[3] Clift SE, Hartley P, Sturgess CEN, Rowe GW. Fracture prediction in plastic deformation process. International
Journal of Mechanical Science 1990;32(1):1–17.
[4] Hartley P, Pillinger I, Sturgess C. Numerical modeling of material deformation processes research development and
applications. Berlin: Springer; 1992.
[5] Youssef Y, Denault J. Thermoformed glass ;ber reinforced polypropylene: microstructure, mechanical properties and
residual stresses. Polymer Composites 1998;19(3):301–9.
[6] Tirosh J, Yossifon S, Eshel R, Betzer A. Hydroforming process of uniform wall thickness products. ASME Journal
of Engineering for Industry 1977;99:685–91.
[7] Yossifon S, Tirosh J, Kochavi E. On suppression of plastic buckling in hydroforming processes. International Journal
of Mechanical Science 1984;26:389–402.
[8] Yossifon S, Tirosh J. Rupture instability in hydroforming deep-drawing process. International Journal of Mechanical
Science 1985;27:559–70.
[9] Yossifon S, Tirosh J. Buckling prevention by lateral 4uid pressure in deep drawing. International Journal of
Mechanical Science 1985;27:177–85.
[10] Yossifon S, Tirosh J. On the permissible 4uid-pressure path in hydroforming deep drawing processes—analysis of
failures and experiments. Journal of Engineering for Industry 1988;110:146–52.
[11] Lo SW, Hsu TC, Wilson WRD. An analysis of the hemispherical-punch hydroforming process. Journal of Materials
Processing Technology 1993;37:225–39.
[12] Hsu TC, Hsieh SJ. Theoretical and experimental analysis of failure for the hemisphere punch hydroforming processes.
Journal of Manufacturing Science and Engineering 1996;118:434–8.
[13] Pourboghrat F, Zampaloni M, Benarad A. Hydroforming of composite materials. October 2003. United States Patent
No. 6631630.
[14] Pourboghrat F, Chu E. Springback in plane strain stretch/draw sheet forming. International Journal of Mechanical
Sciences 1995;36(3):327–41.
[15] Hill R. The mathematical theory of plasticity. Oxford: Clarendon Press; 1950.
[16] Barlat F, Lain J. Plastic behavior and stretchability of sheet metals. Part I: a yield function for orthotropic sheets
under plane stress conditions. International Journal of Plasticity 1989;5:51–66.
[17] Hallquist JO. Ls-Dyna user’s manual. Livermore Software Technology Corporation, 1999.
21
Experimental investigation of friction coefficient in tube hydroforming
Hyae Kyung YI 1, Hong Sup YIM 1, Gun Yeop LEE 1, Sung Mun LEE 1, Gi Suk CHUNG 2, Young-Hoon MOON 1
1. Engineering Research Center for Net Shape and Die Manufacturing, School of Mechanical Engineering,
Pusan National University, Busan 609-735, Korea;
2. Technical Research laboratories, POSCO, Jeonnam 545-090, Korea
Received 21 April 2010; accepted 10 September 2010
Abstract: The friction coefficient between tube and die in guide zone of tube hydroforming was obtained. In hydroforming, the tube
is expanded by an internal pressure against the tool wall. By pushing the tube through tool, a friction force at the contact surface
between the tube and the tool occurs. In guiding zone, the friction coefficients between tube and die can be estimated from the
measured axial feeding forces. In expansion zone, the friction coefficients between tube and die can be evaluated from the measured
geometries of expanded tubes and FE analysis.
Key words: tube hydroforming; friction coefficient; lubricant; axial force; internal pressure
1 Introduction
The hydroforming is a process to expand metal

tubes through high fluid pressure from inside in a closed
forming die[1−3]. The hydroforming process is widely
applied to the automotive industry, such as body, chassis
and frame, and rapidly spread to other industrial fields
[4−7]. In hydroforming process, the tube is expanded by
an internal pressure against the tool wall. By pushing the
tube through the tool, a friction force at the contact Fig.1 Friction zones in tube hydroforming process
surface between the tube and the tool occurs.
In a typical tube hydroforming process, there are tube hydroforming. And the friction coefficients are
two different zones shown in Fig.1, which are the evaluated for different materials, die surface coating,
so-called guiding zone and the expansion zone[8−9]. In viscosity of lubricants and internal pressures. The friction
the guiding zone, the tube and the die are in full contact coefficients between tube and die can be estimated from
during the process and the material is pushed into the the measured axial feeding forces in guiding zone and
deformation zone by axial feeding cylinders. In from the measured geometries of expanded tubes and FE
expansion zone, the deforming material at the analysis in expansion zone. The effects of the various
die-material interface can undergo surface expansion. internal pressures, viscosity of lubricants, tube materials,
The tube is expanded by an internal pressure against the tube size and die surface coating on the friction forces
tool wall. By pushing the tube through the tool, a friction and friction coefficients were discussed.
force at the contact surface between the tube and the tool
occurs. Therefore, the friction at the die-tube interface is 2 Experimental
affected by the formability in tube hydroforming. With
low friction, more homogeneous deformation can be 2.1 Process variable
achieved[10−11]. For this study, STKM11A and SS321, 304, 409
In this study, the friction test method is proposed to tubes were prepared in order to evaluate the effect of
obtain the friction coefficient between tube and die in tube material and size. The tubes with outer diameters of
Corresponding author: Young-Hoon MOON; Tel: +82-51-5102472; E-mail: [email protected]

Hyae Kyung YI, et al/Trans. Nonferrous Met. Soc. China 21(2011) s194−s198 s195
50.8 mm and 76.2 mm and thickness of 2.0 mm are used.
The lubricants used in the friction tests can be divided
into two groups: liquid-state and solid-state lubricants.
The liquid lubricants with different viscosity and Teflon
film were used, as well as no lubricant for comparison
purposes. Each tube was tested at different internal
pressures in the range of 10−50 MPa. The test was
repeated using combination of tube, lubricant, internal
pressure and die surface coating, respectively.
2.2 Friction test in guiding zone

There are two friction laws that are commonly used
in metal forming: Coulomb’s friction law and shear
friction law. While Coulomb’s law is more suitable for
sliding contact with little deformation, the shear friction
law is suitable for sliding objects subjected to bulk
plastic deformation. The deformation modes in guiding
zone of tube hydroforming are dominated by sliding,
hence, Coulomb’s law may be applicable. Assuming
Coulomb’s law, the interface friction coefficient can be Fig.2 Friction test in guiding zone: (a) Schematic drawing;
determined by Eq.(1). The friction force Ff and normal (b) Appearance of testing apparatus
load Fn are both measured by hydroforming system.
μ=Ff/Fn (1)
It assumed that the coefficient of friction is uniform
at the interface throughout the whole tube length and the
normal stress is equal to the internal pressure.
Fn=πpiDoL (2)
where pi is the internal pressure; Do is the out-diameter
of tube; and L is length of tube. From Eq.(2), the friction
coefficient can be expressed as
μ=Ff/(πpiDoL) (3)
Fig.3 Cross-section geometries of corner-fill test model
In friction test of guiding zone, a lubricated tube is
placed in a die with the same size as the out-diameter of FE simulations, using the commercial code
the tube, as shown in Fig.2. The die is closed by the ram LS-DYNA, were conducted for three models under
movement of a press. At the same time, the tube ends are various frictional conditions. And the FE analysis results
loaded and sealed by two punches, and the tube is filled were compared with the measured geometries of
with fluid. And tube is pushed through a die by axial hydroformed tube. And the friction coefficients at
punch. Finally, the internal pressure, axial force and axial various lubricants were obtained at similar geometries by
displacement are measured continuously throughout the comparing the experimental and FE analysis results.
test. The test tubes and die are cleaned with acetone
before the friction test, then the surface of tube is 3 Results and discussion
smeared uniformly with various lubricants. And the total
sliding distance is 30 mm. 3.1 Friction coefficient in guiding zone
The relationships between the friction coefficients
2.3 Friction test in expansion zone and the internal pressure at various lubrications are
Three corner-fill tests were carried out to evaluate shown in Fig.4. The results from the friction test of
the friction coefficient for the expansion zone. Fig.3 guiding zone show that the viscosity of lubricant and
shows the cross-section geometries of corner-fill test internal pressure increases with decreasing the friction of
model. The tube is constrained by the die and can expand coefficient of STKM11A tube. The friction coefficient
into the corners of the die with various lubricants. The without lubricant is higher than that with liquid lubricant.
corner-fill distance and thickness of expanded tube were And the friction coefficient with liquid lubricant is
measured. higher than that of Teflon film. The friction coefficient
s196 Hyae Kyung YI, et al/Trans. Nonferrous Met. Soc. China 21(2011) s194−s198
Fig.4 Friction coefficient distribution with different lubricants Fig.5 Friction coefficient distribution with different lubricants,
at internal pressures (STKM11A) diameters and internal pressures
decreases with increasing the internal pressure. The

friction coefficients without lubricant are about 0.084 at
10 MPa, 0.034 at 30 MPa, 0.020 at 50 MPa. The friction
coefficient decreases with increasing viscosity of
lubricant at same internal pressure. The friction
coefficient is 0.052 without lubricant, whereas those
using viscosity 180, 320, 400 of liquid lubricant and
Teflon are 0.031, 0.020, 0.018 and 0.002 at 20 MPa. And
the friction coefficient is more sensitive to internal
pressure with decreasing the viscosity of lubricant. The
friction coefficient with Teflon film is about 0.002 at
every condition and shows the lowest friction coefficient.
Fig.5 shows the effect of tube diameter on the
friction coefficient. The friction coefficients of 180
lubricant are 0.031 for d 50.8, 0.027 for d 76.2 at 20 MPa,
and 0.021 for d 50.8, 0.020 for d 76.2 at 30 MPa. The
friction coefficient decreases with decreasing the
diameter at same internal pressure and lubricant
conditions. The normal force Fn increases with
increasing the contact area, which is more effective than
the increase of the friction force with friction coefficient
increasing.
The relationships between the friction coefficient
and the strength of material at various internal pressures
and lubrications are shown in Fig.6(a). The effects of
internal pressure and lubricant viscosity are like the
preceding. The friction coefficients of 180 lubricant are Fig.6 Friction coefficient distribution with different lubricants,
0.031 for STKM11A, 0.012 for SS321 at 20 MPa, and materials and internal pressures: (a) STKM11A and SS321;
0.020 for STKM11A, 0.008 for SS321 at 30 MPa. (b) SS304 and SS409
Fig.6(b) shows the effect of the material, therefore,
the friction coefficient decreases with increasing the 0.031 at 20 MPa, irrespective of the die coating.
strength of material under condition of same lubricant Therefore, in lubrication condition, the die coating is
internal pressure. independent on friction force.
Fig.7 shows the effect of die condition at various
internal pressure conditions. The friction coefficients are 3.2 Friction coefficient in expansion zone
0.037 for coated die, 0.052 for uncoated die at 20 MPa. FE simulation results and measured geometry of
However, the friction coefficient has similar value as expanded tube are given in Tables 1, 2 and Fig.8. The
Hyae Kyung YI, et al/Trans. Nonferrous Met. Soc. China 21(2011) s194−s198 s197
Table 1 Measured geometry of expansion tubes with friction coefficient
Friction Triangular Square Hexagonal

coefficient ① ② ③ ④ ① ② ③ ④ ① ② ③ ④
0.001 1.86 1.75 30.50 9.29 1.85 1.78 28.21 12.46 1.89 1.78 27.18 15.00
0.130 1.92 1.81 28.64 9.22 1.90 1.80 27.58 12.06 1.92 1.81 26.30 14.45
0.150 1.93 1.84 28.60 9.23 1.95 1.82 27.57 12.10 1.92 1.81 26.31 14.50
0.168 1.94 1.82 28.53 9.21 1.98 1.80 27.50 12.04 1.95 1.84 26.01 14.24
Table 2 Measured geometry of expansion tubes with lubricants

Triangular Square Hexagonal
Lubricant
① ② ③ ④ ① ② ③ ④ ① ② ③ ④
Teflon 1.86 1.79 30.60 9.29 1.85 1.78 28.25 12.46 1.89 1.78 27.20 15.00
400 1.95 1.82 28.64 9.22 1.91 1.82 27.64 12.10 1.93 1.81 26.30 14.45
320 1.95 1.82 28.54 9.23 1.94 1.82 27.54 12.10 1.93 1.81 26.32 14.50
180 1.94 1.82 28.55 9.21 1.94 1.83 27.55 12.04 1.95 1.84 26.00 14.24
from 0.095 to 0.089 for lubricants 180−400. The friction

coefficient decreases with increasing the strength of
material. The maximum thinning of thickness
represented at contact area of die and tube. The contact
length is longer in the order of hexagonal, square,
triangular shape. The triangular shaped die has the
longest corner fill length. The friction coefficient with
Teflon film is obtained a similar value, about 0.001.
4 Conclusions
The friction coefficient decreases with increasing

the strength and diameter of materials, lubricants
viscosity and internal pressures. Also the friction
Fig.7 Effect of die coating and lubricant on COF coefficient without lubricant decreases with decreasing
die coating; however, the friction forces with lubricant
are not influenced by die coating. The friction coefficient
in guiding zone is lower than that in expansion zone. The
friction coefficient at expansion zone decreases with
increasing the lubricants viscosity and is not sensitive of
the internal pressure. The friction coefficients are similar
for different geometries.
Fig.8 Measured geometries of expansion friction test model: Acknowledgement

① Thickness at contact region; ② Thickness at expansion This work was partially supported by grants-in-aid
region; ③ Maximum expansion length; ④ Contact length; for the National Core Research Center Program from
⑤ Half of contact length MEST/KOSEF
friction coefficients were deducted by comparing FE References

analysis and measured geometries of expanded tubes. In
the case of STKM11A, the friction coefficients are [1] SOKOLOWSKI T, GEKE K, AHMETOGLU M, ALTAN T.
obtained from 0.168 to 0.130 for lubricants 180−400. Evaluation of tube formability and material characteristics: Hydraulic
SS304 has lower friction coefficient than STKM11A, bulge testing of tubes [J]. Journal of Materials Processing Technology,
s198 Hyae Kyung YI, et al/Trans. Nonferrous Met. Soc. China 21(2011) s194−s198
2000, 98: 34−40. Materials Processing Technology, 1999, 91: 236−245.
[2] HWANG Y M, ALTAN T. Process fusion: Tube hydroforming and [8] HWANG YRONG MAW, HUANG LI SHAN, ALTAN TAYLAN.
crushing in a square die [J]. Journal of Engineering Manufacture, Friction tests in tube hydroforming [J]. Proceedings of the Institution
2004, 218B: 169−174. of Mechanical Engineers, Part B: Journal of Engineering
[3] VOLLERTSEN F, PLANCAK. On possibilities for the determination Manufacture, 2005, 219: 587−594.
of the coefficient of friction in hydroforming of tubes [J]. Journal of [9] NGAILE G, JAEGER S, TAYLAN A. Lubrication in tube
Materials Processing Technology, 2002, 125: 412−420. hydroforming (THF) Part 1. Lubrication mechanisms and
[4] LANG L H, WANG Z R, KANG D C, YUAN S J, ZHANG S H, development of model tests to evaluate lubricants and die coatings in
DANCKERT J, NIELSEN K B. Hydroforming highlights: Sheet the transition and expansion zones [J]. Journal of Materials
hydroforming and tube hydroforming [J]. Journal of Materials Processing Technology, 2004, 146: 108−115.
Processing Technology, 2004, 151: 165−177. [10] MUAMMER KOC, ALTAN TAYLAN. Prediction of forming limits
[5] AHMETOGLU M, SUTTER K, LI X J, ALTAN T. Tube and parameters in the tube hydroforming process [J]. Int J Machine
hydroforming: Current research, applications and need for training Tools & Manufacture, 2002, 42: 123−138.
[J]. Journal of Materials Processing Technology, 2000, 98: 224−231. [11] LEI L P, KANG B S, KANG S J. Prediction of the forming limit in
[6] KWAN C T, LIN F C. Investigation of T-shape tube hydroforming hydroforming process using the finite element method and a ductile
with finite element method [J]. The International Journal of fracture criterion [J]. Journal of Materials Processing Technology,
Advanced Manufacturing Technology, 2003, 21: 420−425. 2001, 113: 673−679.
[7] ZHANG S H. Developments in hydroforming [J]. Journal of (Edited by LI Yan-hong)
22
Journal of Manufacturing Processes 15 (2013) 256–262
Contents lists available at SciVerse ScienceDirect
Journal of Manufacturing Processes

journal homepage: www.elsevier.com/locate/manpro
Technical paper
Failure analysis of hydroforming of sandwich panels

Jyhwen Wang a,b,∗ , Cheng-Kang Yang b
a
Department of Engineering Technology and Industrial Distribution, Texas A&M University, College Station, TX, USA
b
Department of Mechanical Engineering, Texas A&M University, College Station, TX, USA
Article history: Sandwich panels are commonly used in various applications to improve the stiffness-to-weight ratio of
Received 4 September 2012 structure components. While producing flat sandwich panel is relatively straightforward, manufacturing
Received in revised form 17 January 2013 of shaped sandwich components can be a challenging task. This paper presents the use of hydroforming
Accepted 29 January 2013
technique in forming bi-layered and sandwich materials with an open-cell foam core. Hydraulic bulge
Available online 17 March 2013
experiments were conducted to form bi-layered and sandwich blanks into a dome shape. Various failure
modes were observed from the experiments. Finite element simulations were conducted to understand
Keywords:
the different failure mechanisms that could occur during the deformation process. The investigation
Hydroforming
Sandwich panels can facilitate the selection of geometry and property of the constituents of the sandwich material for
Failure analysis successful hydroforming.
© 2013 The Society of Manufacturing Engineers. Published by Elsevier Ltd. All rights reserved.
1. Introduction with metal face sheets and perforated sheet metal core. The exper-
imental results showed that, due to bending, the dominant failure
Light-weight sandwich panels are commonly used in various mechanism is core shear. From numerical analysis, it was also found
industries to improve the stiffness and reduce the weight of struc- that the required core shear strength is proportional to the face
ture components. In aviation and aerospace industry, sandwich sheet strength. Jackson et al. [3] conducted experiments in incre-
panels are widely adopted in body and airfoil design of the air- mental forming of sandwich panels. With the loading from the
craft to obtain a better stiffness to weight ratio. In automotive forming tool, the soft core was non-uniformly deformed. From the
industrial, sandwich panels are designed as shock-absorbing and previous work, it can be observed that while forming sandwich pan-
impact-resistant materials in racing car. These sandwich panels els, the high contact force between the tool/die and the sandwich
generally consist of metal face sheets, such as steel or aluminum blank generally results in crushing or fracture of the lower strength
sheets, and a porous light-weight core, such as a honeycomb or core material.
metal/polymer foams. In order to prevent damage of the open cell foam core while
While producing flat sandwich panels is relatively simple, man- forming shaped sandwich panels, sheet hydroforming technique
ufacturing of shaped sandwich panel can be costly and time can be considered. In the automotive industry, sheet hydroforming
consuming. The face sheets are formed in a stamping press first. The process is used to produce components with complicated shapes
shaped core is then produced from expensive machining or mold- that are difficult to form using traditional stamping processes. In
ing process. Finally, an adhesive bonding operation is required to sheet hydroforming, the hydraulic fluid can uniformly exert pres-
assemble the face sheets and the core. sure on the workpiece surface. The technique could reduce tooling
An alternative to the traditional approach is to produce flat contact, avoid excessive localized deformation, reduce springback,
sandwich panels first and then forms the panels into the desired and improve the part strength. Sheet hydroforming has been widely
shapes. The approach, however, generally results in the collapse studied. Hill developed an analytical model to describe the defor-
of foam core. Elzey and Wadley [1] analyzed the open die forg- mation of the metal sheet in a bulge test [4]. Chakrabraty et al.
ing of structural sandwich with porous core. It was shown that improved Hill’s work by including hardening coefficient. The effects
the deformation process can lead to a non-uniform core density of work hardening on the thickness and the dome height are also
and thickness distribution (indication of core collapse). Mohr [2] investigated [5]. Shang et al. developed a model to predict the influ-
conducted deep drawing experiments using thin sandwich blanks ence of the die radius on sheet hydroforming [6]. They also found
that allowing draw-in of the flange can improve the formability
of the sheet metal [7]. Hein et al. presented the forming mech-
∗ Corresponding author at: Department of Engineering Technology and Industrial anisms at different areas in free hydroforming of a single blank
Distribution, Texas A&M University, College Station, TX, USA. Tel.: +1 979 845 4903. [8]. Controlling the blank holder force to improve the formability
E-mail address: [email protected] (J. Wang). of the sheet was reported by Shulkin et al. [9]. A comprehensive
1526-6125/$ – see front matter © 2013 The Society of Manufacturing Engineers. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jmapro.2013.01.008
J. Wang, C.-K. Yang / Journal of Manufacturing Processes 15 (2013) 256–262 257
Fig. 2. Hydroforming of a sandwich panel (a stretch forming process). (a) Initial

configuration and (b) hydroforming.
to the adhesive bond, the core material and the bottom sheet could
be formed into the upper die cavity.
2.1. Specimen preparation and experimental setup
Both bi-layered and sandwich panels were prepared for hydro-

forming experiments. Aluminum 5052 sheet with a thickness of
0.5 mm was used as face sheets. The diameter of the top face sheet
was 241.3 mm, and the bottom face sheet of the sandwich panels
had a diameter of 152.4 mm. The extra surface area of the upper face
sheet is used for clamping such that the lower chamber (containing
Fig. 1. Specimens for hydroforming experiments. (a) Bi-layered specimen and (b)
sandwich specimen. hydroforming fluid) can be sealed. The core material was prepared
from Duocel open-cell foam blocks with 40 pores per 25.4 mm.
The foam blank thickness was 6.35 mm. The primary material of
review of sheet hydroformig processes can be found in [10] by Duocel foam was 6101-T6 aluminum alloy, and the relative den-
Vollertsen. sity ratio was 8%. Since the bi-layered and sandwich panels are to
In the present work, the feasibility of using sheet hydroforming be subjected to stretch forming, polyurethane adhesive was used
techniques to form bi-layered and sandwich panels is demon- to provide good flexibility. The bonding strength between the face
strated. The process is similar to the bulge forming test [11] that is sheet and the foam depends on the surface preparation. To pro-
used to test the biaxial formability of sheet metal. A porous mate- duce blanks with different bonding strength, some of the face sheets
rial, such as a metallic or polymer foam, can be attached to a sheet were roughed by sand paper and were cleaned by using acetone.
metal to form a bi-layered panel. An additional bottom face sheet The specimens were placed between a clamping plate and
can be attached to the other side of the porous material to make a hydroforming die with an opening of 177.8 mm. During an exper-
sandwich panel. With different specimens and deformation con- iment, the hydraulic pressure was slowly increased and the bulge
ditions, various failure modes were observed. The experimental forming process was stopped at a preset pressure or when a sharp
results can be explained by the results obtained from numerical pressure drop was detected. The specimen was then inspected for
(finite element) simulations, and a better understanding of the fail- any failure.
ure mechanism is achieved.
2.2. Experimental results
2. Hydroforming experiments From the experiments, it was observed that limited deformation
was achieved in bulge forming of sandwich blanks. The bi-layered
To produce shaped panels, bi-layered blanks (Fig. 1a) and sand- panel, on the other hand, can be deformed into a deeper dome
wich blanks (Fig. 1b) with open-cell foam core are used in the shape. That is, while the top face sheet was stretched, the foam
hydroforming experiments. The premise is that during the form- blank was “pulled” and deformed with the face sheet. During hydro-
ing process, the hydraulic fluid is allowed to flood the pore of the forming, the fluid can flow into the porous open cell foam core.
open-cell foam without collapsing the foam core. Fig. 2 depicted the Thus, no foam damage (collapsing of foam core) was observed. The
process. Fig. 2a showed that the upper face sheet was first clamped. failures in forming the bi-layered and sandwich specimens can be
Then the hydraulic fluid can fill the sealed chamber without dam- classified into three categories.
aging the foam core. In Fig. 2b, with continuous increase of the The first type of failure is top face sheet fracture. It can be
hydraulic pressure, the upper face sheet can be deformed. And due detected from a sudden pressure drop. The fracture location can
258 J. Wang, C.-K. Yang / Journal of Manufacturing Processes 15 (2013) 256–262
Fig. 5. The delamination in three layer sandwich panel.
Fig. 3. Fracture occurred at the top face sheet. understanding of the failure mechanisms. Hydroforming of differ-
ent bi-layered and sandwich panels were analyzed using ABAQUS
be at the radius of the clamping die or at the area near the edge to investigate the effects of material properties and bonding
of the foam blank as shown in Fig. 3. As the top face sheet was strength on the success or failure of the process. Due to the sym-
stretched over a hard corner (the die radius or the edge of the foam), metry, an axisymmetric model was used to simulate the sandwich
the deformation was localized and the fracture occurred. No frac- panels (as shown in Fig. 6) and the bi-layered panels. The linear
ture near the pole of the dome was observed as the area was more hybrid element was utilized in the foam and face sheets, and the
uniformly deformed. cohesive elements COHAX4 was adopted in the adhesive layer. The
Delamination between face sheet and foam blank was observed hydraulic fluid can pass through the edge of the open cell foam
in many experiments, especially in forming sandwich panels. It was and flood the pores. The pressure loading of ramping up to 2 MPa
found that delamination can take place between the top face sheet was applied on the inner surface of the top face sheet for both the
and the foam (Fig. 4), or between the bottom face sheet and the bi-layered and sandwich panels. The friction coefficient for the con-
foam. The different failure modes could be due to the difference in tacts between sheet metal, foam, and the die/holder was assigned
bonding strength at the two interfaces. It should be noted that the to 0.1.
sandwich blank can delaminate into three separate pieces at the
end of the experiment. Finite element simulations were conducted
3.1. Material parameters
(and presented in the next section) to investigate the different fail-
ure mechanisms.
In the simulation, 0.8 mm thickness aluminum 5052 sheet was
The third failure mode was the circumferential buckling of the
selected as the material of the face sheet. The mechanical prop-
bottom face sheet. The failure can occur without complete delam-
erties of aluminum 5052 were considered isotropic material and
ination initially. As the sandwich panel was further deformed,
the elastic/perfect plastic material model was assumed to describe
separation of the foam blank from one of the face sheets can be
the constitutive behavior of the material. For aluminum 5052, the
observed. Fig. 5 shows that, along with the foam blank the buck-
Young’s modulus, yielding strength, and the Poisson’s ratio are set
led bottom face sheet was separated from the top face sheet.
at 70.3 GPa, 89.6 MPa, and 0.33, respectively.
It is believed that the buckling was due to the induced hoop
The open-cell foam, Duocel (from ERG Aerospace Corporation),
stress and the non-uniform adhesive strength at the interface
has 8% relative density. The matrix of cells and ligaments is com-
between the bottom face sheet and the foam blank. As this failure
pletely repeatable, regular, and uniform throughout the entirety
mode was not common and was resulting from defects in speci-
of the material. The properties of the foam are shown in Table 1
men preparation, it was not further studied in the finite element
[12].
analysis.
The mechanical properties of foam materials have been widely
studies [13–17]. In order to investigate the effects of foam prop-
3. Finite element analysis erties on the results of hydroforming process, the constitutive
model developed by Gibson and Ashby [13] was adopted. The
As various failure modes were observed in the experiments,
finite element simulations were conducted to obtain a better
Fig. 4. Delamination between the top face sheet and the foam blank. Fig. 6. Finite element model of a sandwich panel.
Table 1
Mechanical properties of Duocel foam (8% relative density) [12].
Compression strength 2.53 MPa

Tensile strength 1.24 MPa
Shear strength 1.31 MPa
Modulus of elasticity (compression) 103.08 MPa
Modulus of elasticity (tension) 101.84 MPa
Shear modulus 199.95 MPa
density-dependent Young’ modules E* and plastic collapse strength

∗ are given by
pl
E∗
∗
= C1 (1)
Es s
Fig. 7. The normal stress and displacement of the adhesive bond between the face
∗
pl ∗ 3/2 sheet and the foam.
= C2 (2)
ys s debonding of the adhesive. Hence, the energy toughness Gnc can be
where Es , S and ys are the Young’s modulus, density, and yield approximated as
strength of the raw material in foam, respectively, and * is the 1 1 2
density of the foam. C1 and C2 are geometry related constants. Gnc = ı = (5)
2 2 Ẽ/t
From the experimental data [13], the values of the constants are
recommended that C1 = 1 and C2 = 0.3. The Poisson ratio of open- where t is the thickness of the adhesive and Ẽ is the modulus defined
cell foam was 0.3. The open-cell foam was also assumed to have as Ẽ/t = /ı. For simplicity, the behavior of the adhesive in the first
an elastic-perfect plastic behavior during the analysis. In the finite and the second shear directions was assumed the same as that of the
element hydroforming simulations, the aluminum Al6101-T6 used normal direction in the present study. The base case properties of
as the raw material of foam has the properties of Es = 68.9 GPa and the adhesive are given in Table 2. To investigate different potential
ys = 193 Mpa. The relative density of the foam was set at differ- failure modes, a number of different adhesive strength values were
ent values from 8% to 24%. The htydrostatic pressure effect on the used in the simulations.
mechanical property of the open cell foam was not considered. In the finite element analysis, once the energy failure criterion,
Eq. (3), is satisfied in a cohesive element, the element is not able to
take any further load, and then the element is removed to depict
3.2. Cohesive zone model
the delamination.
One of the failure modes observed from the experiments is the

4. Result and discussion
delamination between the face sheet and the foam core. As cohesive
zone models have been successfully used to analyze the interface
The objective of the present finite element simulations is to
fracture of composite materials [18–21], a cohesive zone model was
investigate the causes of various failure modes observed in the
adopted to describe the delamination behavior in the finite element
hydroforming experiments. It was found that the simulation results
simulation. In this model, the adhesion in the normal and shear
are consistent with the experimental observations. The simulation
direction were assumed independent. To characterize the mixed
results can be used to guide the selection of the face sheet, the foam,
mode adhesive fracture, an energy-based failure criterion was used.
and the adhesive for hydroforming processes.
The interface fracture occurs when:
Gn Gs1 Gs2 4.1. Failure modes in hydroforming of bi-layered panels
+ + =1 (3)
GnC C
Gs1 C
Gs2
There were primarily three competing failure modes in hydro-
where Gn , Gs1 , and Gs2 are the energy release rate in the normal, first forming of bi-layered panels. Fig. 8 shows that the delamination
shear and second shear direction, and the GnC , Gs1C , and GC were the
s2 can be observed for panels with low bonding strength at the sheet
energy toughness under each mode. The energy release rate is given metal–foam interface. In this model, the panel has a low adhesive
by strength (1.5 MPa) and foam density (8%). Fig. 9 shows that, with
n
higher adhesive strength (6.0 MPa) and density (24%), the face sheet
Gn = (ın )dın fracture at the clamping die radius or at the edge of the foam block
0 can occur before delamination. With increasing strength or thick-
s1
ness of the face sheet, the fracture is delayed and the failure mode
Gs1 = (ıs1 )dıs1 (4)
can shift to delamination failure.
0 s2 The maximum attainable dome height of the bi-layered pan-
Gs2 = (ıs2 )dıs2 els was typically greater than that of the sandwich panels. This
0 is because that the localized deformation of the top face sheet is
where and are displacement-related normal and shear stress, not as severe, and there is no debonding of the bottom face sheet
and ın , ıs1 , and ıs2 are the normal, first shear and second shear
displacements [22]. Table 2
The energy toughness is equal to the area under the Material properties of the base case adhesive.
stress–displacement curve in the traction–separation law. As Thickness, t 0.3 mm

shown in Fig. 7, the normal stress is proportional to the dis- Modulus, Ẽ 150 MPa
placement ı before the normal stress reached to a maximum value. Strength of adhesive, 10 MPa
Energy toughness, Gnc 100 N/m
Beyond the maximum normal stress, the steep drop indicates the
Fig. 11. Bottom face sheet delamination in hydroforming of sandwich panels with
Fig. 8. Delamination in hydroforming of bi-layered panels. a lower strength in the bottom adhesive.
mostly near the edge of the foam block. As a result, the top face sheet
fracture can easily occur. If the top face sheet has a high strength
to delay fracture, the delamination of the bottom face sheet can
occur. For the model in Fig. 11, the foam density is 8%; and the
adhesive strengths are 6.0 MPa and 1.5 MPa for the top and bot-
tom interfaces, respectively. As can be observed in the figure, the
delamination, due to insufficient adhesion strength at the bottom
interface, was initiated near the perimeter/edge of the foam and
sheet metal blanks. The delamination can shift to the top interface
only when the adhesion strength at the bottom interface is much
stronger than the adhesion strength at the top.
Compared to forming bi-layered panels (Fig. 8), a lower dome
height was obtained when forming sandwich panels (Fig. 10) given
the same material properties for the face sheet, adhesive, and foam.
In the analysis, the foam is not evaluated for damage as no failure
criterion for the foam was specified.
Fig. 9. Face sheet fracture in hydroforming of bi-layered panels.
As shown earlier, the mechanical properties of the foam is deter-
mined by its relative density. For low relative density (i.e. 8%), the
delamination was initiated at the bottom interface. The upper inter-
to be considered. Since no deformation limit is specified for the face can then delaminate at a later stage. A high relative density
elastic-perfect plastic foam model, no foam failure was predicted. leads to a high Young’s modulus, which in turn makes the foam
The result is in agreement with the experiments that, although difficult to bend. As a result, shown in the simulation, the bottom
the ligaments of the foam were plastically deformed, no significant interface was not subject to significant stress, and the delamination
damage was noted. occurred at the top interface first.
4.2. Failure modes in hydroforming sandwich panels 4.3. Parametric study in hydroforming of bi-layered panels
From finite element simulation, it was found that the defor- The effect of the foam’s relative density on the attainable dome
mation behavior of sandwich panels was different from that of height in the bulge test of bi-layered panels was further investi-
bi-layered panels. The model in Fig. 10 has a foam density of 8% gated quantitatively. The maximum dome height at the onset of
and an adhesive strength of 1.5 MPa for both top and bottom inter- delamination was plotted against the adhesive strength for the
faces. The figure shows the bottom face sheet provides high bending foam blank with 8%, 16% and 24% relative density. Fig. 12 shows
rigidity and thus the deformation of the top face sheet took place the plots in three sizes of foam diameters, 101.6 mm, 127.0 mm, and
152.4 mm. The filled mark presented the final dome height without
delamination at 2 MPa maximum hydraulic pressure.
It can be seen that the gain in the dome height due to the
improved adhesion strength was initially significant (large slop) for
panels with low relative density foam. Also note that panels with
high relative density foam required a stronger adhesive strength
to carry the foam during the hydroforming process to achieve a
specified dome height (e.g. a horizontal line at 10 mm). It can also
be observed that for a given adhesive strength (e.g. a vertical line
at 3 MPa), a panel with a higher relative density foam (24%) had
a lower dome height compared to a panel with a lower relative
density foam (8%). To reach a certain dome height, the difference
of the required adhesive strength between each foam density was
decreased with increased foam diameter as observed from Fig. 12a
through c.
Fig. 10. Face sheet fracture near the edge of the foam in hydroforming of a sandwich For low relative density foam (8%), the effect of foam diame-
panel. ter is further illustrated in Fig. 13. Without curve fitting, the data
Fig. 13. The maximum attainable dome height vs. adhesive strength for blanks
different foam diameter for bi-layered panels.
5. Conclusion
Forming sandwich panels with conventional stamping die has

been proven difficult as the foam core can be easily damaged dur-
ing the deformation process. In this paper, experimental work and
numerical simulation are presented to demonstrate the feasibil-
ity of using sheet hydroforming technique to form bi-layered and
sandwich panels. With open-cell foam core, the hydraulic fluid can
flood the pore without collapsing the foam. The approach, how-
ever, can lead to other failure modes such as face sheet fracture and
delamination. Through finite element simulation of the conducted
bulge test, the followings can be concluded:
• The qualitative results of finite element simulation are in agree-

ment with experimental observation. The failure modes depend
on the properties of the face sheet, the foam, and the adhesive.
• Compared to forming a bi-layered panel, forming sandwich panel
is a more difficult task as the delamination of the bottom face
sheet can easily occur.
• A low relative density of the foam can reduce the stiffness of the
foam core and improve the maximum dome height in the bulge
test.
• Finite element simulation of hydroforming processes can facil-
itate the identification of failure modes, and can be used to
determine the appropriate constituents (face sheet, foam, and
adhesive) of the sandwich panels.
• Additional FEA simulations can be conducted to investigate the
failures of sandwich (three-layered) panels. Other material and
tooling parameters such as the face sheet thicknesses and prop-
erties and clamping die radius can be included in the future study.
Acknowledgements
Fig. 12. Maximum attainable dome height vs. adhesive strength for foam with 8 mm The work was supported by a grant from National Science Foun-
thickness and different diameters (D) for bi-layered panels: (a) D = 101.6 mm, (b) dation (CMMI-0825986). The authors would like to thank Professor
D = 127 mm, and (c) D = 152.4 mm.
Farhang Pourboghrat at Michigan State University for conducting
some of the experiments.
References
show that the dome height can reach a plateau despite an increase
in adhesive strength. This could be due to continued stress build [1] Elzey DM, Wadley HNG. Open die forging of structurally porous sandwich pan-
up without bulge deformation prior to delamination. For the panel els. Metallurgical and Materials Transactions A 1999;30A:2689–99.
with smaller foam blank, the plateau was reached at a lower dome [2] Mohr D. On the role of shear strength in sandwich sheet forming. International
Journal of Solids and Structures 2005;42:1491–512.
height compared to the panel with larger foam blank (16 mm for [3] Jackson KP, Allwood JM, Landert M. Incremental forming of sandwich panels.
D = 101.6 mm and 21 mm for D = 152.4 mm). Journal of Materials Processing Technology 2008;204:290–303.
[4] Hill R. A theory of the plastic bulging of a metal diaphragm by lateral pressure. [14] Andrews E, Sanders W, Gibson LJ. Compressive and tensile behavior of alu-
Philosophical Magazine 1950;7:1133–42. minum foams. Materials Science & Engineering A 1999;270:113–24.
[5] Chakrabraty J, Alexander JM. Hydrostatic bulging of circular diaphragms. The [15] Deshpande VS, Fleck NA. Isotropic constitutive models for metallic foams. Jour-
Journal of Strain Analysis for Engineering Design 1970;5:155–61. nal of the Mechanics and Physics of Solids 2000;48:1253–83.
[6] Shang HM, Shim VPW. A model study of the effect of the size of the die shoulder [16] Ashby MF, Evans A, Fleck NA, Gibson LJ, Hutchinson JW, Wadley HNG. Metal
in hydroforming. Journal of Mechanical Working Technology 1984;10:307–23. foams: a design guide. Boston: Butterworth-Heinemann; 2000.
[7] Shang HM, Qin S, Tay CJ. Hydroforming sheet metal with intermittent changes [17] Peroni L, Avalle M, Peroni M. The mechanical behaviour of aluminium foam
in the draw-in condition of the flange. Journal of Materials Processing Technol- structures in different loading conditions. International Journal of Impact Engi-
ogy 1997;63:72–6. neering 2008;35:644–58.
[8] Hein P, Vollertsen F. Hydroforming of sheet metal pairs. Journal of Materials [18] Sha JB, Yip TH. Deformation and fracture behavior of beams composed of alu-
Processing Technology 1999;87:154–64. minum foam core and ceramic Al2 O3 under monolithic bending. Metallurgical
[9] Shulkin LB, Posteraro RA, Ahmetoglu MA, Kinzel GL, Altan T. Blank holder force and Materials Transactions A 2005;36:771–83.
(BHF) control in viscous pressure forming (VPF) of sheet metal. Journal of Mate- [19] Li S, Thouless MD, Waas AM, Schreder JA, Zavattiei PD. Use of model-I cohesive-
rials Processing Technology 2000;98:7–16. zone model to describe the fracture of an adhesively-bonded polymer–matrix
[10] Vollertsen F. State of the art and perspectives of hydroforming of tubes and composite. Composites Science and Technology 2005;65:281–93.
sheets. Journal of Materials Science and Technology 2001;17:321–4. [20] Li S, Thouless MD, Waas AM, Schreder JA, Zavattiei PD. Use of a cohesive-zone
[11] ASTM E2712. Standard Test Method for Bulge-Forming Superplastic Metallic model to analyze the fracture of a fiber reinforced polymer–matrix composite.
Sheet. ASTM International, 100 Barr Harbor Drive, West Conshohocken, PA Composites Science and Technology 2005;65:537–49.
19428; 2009. www.astm.org [21] Li S, Thouless MD, Waas AM, Schreder JA, Zavattiei PD. Mixed-mode cohesive-
[12] ERG Aerospace Corporation, http://www.ergaerospace.com/Aluminum- zone models for fracture of an adhesively bonded polymer–matrix composite.
properties.htm Engineering Fracture Mechanics 2006;73:64–78.
[13] Gibson LJ, Ashby MF. Cellular Solids: Structure and Properties. Cambridge: [22] Dassault Systèmes. ABAQUS Analysis User’s Manual, version 6.9-EF, 25.9.6;
Cambridge University; 1997. 2009.

Binded 2

Uploaded by

Copyright:

Available Formats

Binded 2

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Binded 2

Uploaded by

Copyright:

Available Formats

13

Journal of Materials Processing Technology 214 (2014) 1341–1351

Contents lists available at ScienceDirect

Journal of Materials Processing Technology

Deformation analysis of double-sided tube hydroforming in

1. Introduction As is known, material’s formability is not only related to its

Fig. 3. Schematic diagram of stress status in straight wall area.

According to Fig. 3, the differential equilibrium equation for

zone (x = 0) should be expressed as follows based on the mechanical

Yield stress, s (MPa) 78.3

4.3. Finite element model

4.4. Microstructure and hardness measurement

The grain structure was analyzed by electron backscattering

Loading path Pressure/MPa

can be obtained after the thickness distribution of the cracked tube

Therefore, the effective strain could be expressed as follows

applied external pressure is pe and 0 is the circumference stress

In the present work, effect of external pressure on the forma- References

Available online at www.sciencedirect.com

Electromagnetic joining of hybrid tubes for hydroforming

1. Introduction and motivation

1.3. Combined processes

In several cases, electromagnetic forming technologies

diameter of 71 mm and a thickness of 2.5 mm. This implies a

Fig. 2. Shear force to be transferred by the joint

The tube ends are considered as clamped in axial

4. Magnetic pulse welding of hybrid tubes

To enable magnetic pulse welding of the two tubes

5.2. Cross section deformation

In order to evaluate the performance of the joint zone

6. Summary Verlag Wissenschaftlicher Scripten, 2011, ISBN: 978-3-642267-32-8, pp.

[1] Meleghy, G.; Bayer, S.: Die Automobilzulieferindustrie – Perspektiven

Contents lists available at ScienceDirect

Materials and Design

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

Evaluation of the optimal working conditions for the warm sheet

Fig. 1. Tensile test equipment: a) heating device; b) DIC acquisition feature.

Fig. 2. Experimental set-up for the formability tests.

procedures is based on a statistical approach, involving the Response

Fig. 5. Electro-hydraulic press machine: a) main components; b) thermocouples positioning.

Fig. 8. a) Numerical model; b) adopted proﬁles for BHF and pressure.

Temperature σY [MPa] K [MPa] n [−]

RT 280 435.2 0.081

Fig. 10. Overview of the modeFRONTIER workﬂow.

Fig. 12. Forming limit curves according to temperature.

it is evident that, in order to make the FE model reliable and to obtain

Contents lists available at ScienceDirect

International Journal of Machine Tools & Manufacture

Evaluation of effect of ﬂow stress characteristics of tubular material on

tensile test of the roll-formed tubular material.

500 strain increment are deﬁned by

300 where s is effective stress, l is plastic multiplier and f is the

Jig for the grip part

01 (Welding area) 500.25 9.90 14.03

Effective Stress [MPa]

100 Tube Tensile Test Data

¼ r ð13Þ e¼ e1 þ e2 2 þ e1 e2 ¼ Ye1 ð15Þ

Free Body Diagram Stress - State 1()

Bulge Test Configuration