International Journal of Plasticity 21 (2005) 1815–1833

www.elsevier.com/locate/ijplas

Analysis of tube hydroforming in a square

cross-sectional die
Yeong-Maw Hwang *, Wen-Chih Chen
Department of Mechanical and Electro-mechanical Engineering, National Sun Yat-Sen University,
Kaohsiung 804, Taiwan

Received in final revised form 6 September 2003

Available online 19 December 2004

Abstract

A mathematical model considering sliding friction between the tube and die is proposed to
explore the plastic deformation behavior of the tubes during a hydraulic expansion process in
a square cross-sectional die. This model is used to predict the forming pressure needed to hyd-
roform a circular tube into a square cross-section and the thickness distribution of the prod-
uct. FE simulations on tube expansion have also been carried out to compare the analytical
results with those by the proposed mathematical model. The effects of the friction coefficient
between the die and tube, upon the forming pressures needed and thickness distributions after
the expansion process are discussed. Experiments of tube expansion in a square die using a
self-designed apparatus are also conducted. The analytical results of forming pressure and
thickness distributions of the formed parts are compared with the experimental measurements
to verify the validity of this newly proposed model.
Ó 2004 Elsevier Ltd. All rights reserved.

Keywords: Tube hydroforming; Expansion process; Finite element simulation; Sliding friction

*
Corresponding author. Fax: +886 7 5254299.
E-mail address: [email protected] (Y.-M. Hwang).

0749-6419/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijplas.2004.09.004
1816 Y.-M. Hwang, W.-C. Chen / International Journal of Plasticity 21 (2005) 1815–1833

Notation

a half-width of the die

i the ith step or the number of all contact elements
j the jth element
K strength coefficient
L length of elements
m number of deforming elements
n strain hardening exponent
P forming pressure
R outer radius of the tube
S total elongation of all stretching elements
t thickness of the tube
V volume of the tube
r true stress
r

effective stress
e true strain

e effective strain
l friction coefficient between the tube and die

Superscripts
0
before stretching
00
after stretching

Subscripts
0 before expansion
1 the first element or the first step of the expansion process
f free expansion part
i the ith step or the ith element
j the jth element
l dummy variable
t thickness direction
y yielding point
h circumferential or tangential direction

1. Introduction

Nowadays, hydroforming processes have been widely applied to manufacturing

parts in various fields, such as automobile, aircraft and aerospace, and ship building
industry, due to the increasing demands for lightweight parts (Dohmann and Hartl,
1996).
 Y.-M. Hwang, W.-C. Chen / International Journal of Plasticity 21 (2005) 1815–1833 1817

Hydroforming is one kind of forming methods, with which the friction at the con-
tact interface between the workpiece and die can be reduced and better formability of
the product can be achieved, compared with bulk forming processes (Stupkiewicz
and Mroz, 2003). Tube hydroforming processes have more other advantages such
as part consolidation, weight reduction, lower tooling cost, fewer secondary opera-
tions, etc., over the traditional metal forming processes (Smith et al., 2003; Stough-
ton and Yoon, 2004; Chen et al., 2003; Li et al., 2003; Gotoh and Yamashita, 2003;
Yoon et al., 2004; Ho et al., 2004).
Concerning the studies on tube and pipe hydroforming processes, Sokolowski
et al. (2000) have carried out a series of simulations and experiments on the tube
formability test. Dohmann and Hartl (1996, 1997) have also undertaken a series
of investigations on tube hydroforming processes, such as manufacturing axisym-
metrical parts and T-shape parts by expansion and feeding. Asnafi (1999, 2000) pro-
posed a mathematical model to predict the forming pressure and the associated
feeding distance needed to hydroform a circular tube into a T-shape product without
wrinkling and bursting. Using the total strain theorem, Fuchizawa and Narazaki
(1989a,b) derived a mathematical model to analyze the relationships between the
pressure and bulged height as well as the strain distributions of the formed product
in the hydraulic bulging process. Woo (1973) proposed a numerical solution for anal-
ysis of the bulging process of a thin-walled tube under internal pressure and axial
force. The anisotropy of the tubular material is also assessed from the values of
the width/thickness strain ratio determined in the tension test (Woo and Lua,
1978). Banerjee (1975) proposed an analytical model to predict the pressure at
instability of short thin cylinders subjected to internal pressure in a bulging test.
Kridli et al. (2003) used the finite element code ABAQUS to examine the thick-
ness variation and the corner filling in tube hydroforming. They discussed the effects
of the strain-hardening exponent, initial tube wall thickness, and die corner radii on
corner filling and thickness distribution of the hydroformed tube. However, the effect
of the friction coefficient at the tube–die interface, which will significantly affect the
thickness distribution of the formed parts, was not mentioned. Rama et al. (2003)
proposed a two-dimensional approach for simulation of hydroforming expansion
of tubular cross-sections without axial feed. However, the sticking friction mode
was adopted, i.e., the thickness distribution will not be affected by the friction coef-
ficient at the tube–die interface. Koc and Altan (2002) used the two-dimensional
FEM to analyze the plastic deformation of tubes inside a simple sectional die during
tube hydroforming. Koc et al. (2001) conducted experiments of tube bulge hydro-
forming and proposed a simple model to determine the flow stress of tubular mate-
rials. Manabe and Amino (2002) discussed the effects of stress ratio, friction
coefficient, n-value, and anisotropic parameter r-values upon the wall thickness dis-
tribution and the fracture location of the formed tube during hydroforming of a
square product expanded from a round tube using combined axial load and internal
pressure. Their analytical results from the FE-analysis are in good agreement with
the experimental results using an A6063 tube. To get a constant ratio of incremental
axial and hoop strains, Johnson et al. (2004) proposed a numerical algorithm with a
FE model to control the pressure and end-feed increments during expansion of
1818 Y.-M. Hwang, W.-C. Chen / International Journal of Plasticity 21 (2005) 1815–1833

A6061-T4 tubes within a conical die shape and under free hydroforming conditions.
The analytical wall thickness distribution and cone radius are compared with exper-
imental results.
One of the present authors has also published a series of works on hydroforming-
related topics. For example, a simple mathematical model was proposed to estimate
the required forming pressure during tube expansion in a rectangular die (Hwang
and Chen, 2003). The forming pressure and the thickness of the formed product dur-
ing tube expansion in a square die (Hwang et al., 2000) and during T-shape protru-
sion with axial feeding (Hwang and Lin, 2003) were also discussed using the finite
element method. Considering the forming tube as an ellipsoidal surface, a mathemat-
ical model was proposed to examine the plastic deformation behavior of a thin-
walled tube during tube hydraulic bulging processes in an open die (Hwang and
Lin, 2002). The plastic flow pattern of a circular tube that is crushed into a rectan-
gular cross-section was discussed using the finite element method (Hwang and Altan,
2003). It is known that the internal pressures and the clamping forces using crushing
processes are much smaller than those by using hydraulic expansion processes, as
well as the thickness distribution of the formed product using crushing processes
is much uniform than that by the hydraulic expansion process. So far, however,
among published papers, there are currently few works that have considered the slid-
ing friction mode in hydraulically expanding a circular tube into a square cross-
section by analytical methods.
Automotive components such as rear-axles, subframes, front-axles, etc., are usu-
ally produced by the sequence of pre-bending and expansion (hydroforming) pro-
cesses from a long starting tube. During the expansion process, since axial feeding
provides limited material flow due to friction and geometry constraint, only pure
tube expansion occurs at the middle section, which means the deformation can be
regarded as a plane strain state. In the expansion process, the friction between the
tube and die affects the forming pressures or the thickness distribution of the formed
product greatly.
The present authors have developed a model considering the sticking friction
mode to predict the forming pressure and thickness distribution of formed parts dur-
ing expansion in a rectangular die (Hwang and Chen, 2003). In this paper, a math-
ematical model considering the sliding friction mode, in which the contact elements
stretch during the process, is developed. The forming pressure and the thickness dis-
tribution of the formed product are examined. The effects of the friction coefficient
upon the forming pressure and the thickness distribution of the product are
discussed.

2. Mathematical model

2.1. Analytical model of tube expansion

A mathematical model considering sliding friction mode is developed to predict

the forming pressures needed to expand a circular tube into a square cross-section,
 Y.-M. Hwang, W.-C. Chen / International Journal of Plasticity 21 (2005) 1815–1833 1819

and the thickness distribution of the product. During the formulation, the following
assumptions are employed:

1. the tube is a rigid-plastic body;

2. the von-Mises yielding criterion is adopted;
3. the plastic deformation of the tube is under a plane strain state;
4. the free expansion part of the tube is a circular arc and has a uniform thickness
distribution.

The configurations of the die and tube before and after the expansion process in a
square die are shown in Fig. 1, where t0 and R0 are the initial thickness and outer
radius of the tube, respectively. The schematic of the mathematical model for tube
expansion in a square die is shown in Fig. 2. Because of the symmetry of the square
die, only one quarter of the cross-section is shown. The expansion process is assumed
to be made up of many small procedure steps and the pressure increment for each

2a Center
2a Die
R
Ro
P
to
Corner
2a

2a

(a) (b)

Fig. 1. Configurations of the tube and die during the expansion process in a square die: (a) before
expansion; (b) after expansion.

Lj S
Li+1
)
+1
t1
t1

tf(i
t fi

tf1
tj

L1 1
R i+
Ri

R1

Ro Ro
(a) Step 1 (b) Step i+1

Fig. 2. Schematic of the analytical model during tube expansion in a square die.
1820 Y.-M. Hwang, W.-C. Chen / International Journal of Plasticity 21 (2005) 1815–1833

step is DP. Fig. 2(a) and (b) shows the configurations of the tube and die at the first
and ith steps, respectively. The free expansion part at the corner of the die is assumed
to be a circular arc, which radius and thickness at step i (shown by the dashed lines in
Fig. 2(b)) are denoted by Ri and tfi, respectively. The formulation of this model is
described below.
(i) Stress and strain states at the free expansion part.
The true strain in the thickness direction of the tube is et = ln(t0/t). From the
assumption of plane strain, the equivalent strain can be obtained as

e ¼ p2ffiffi3et : ð1Þ
Assuming the flow stress of the tubular material follows the power law of its equiv-
alent strain, the flow stress of the tube, r

, can be expressed as

¼ K
en ,
r ð2Þ
where K and n are the strength coefficient and strain-hardening exponent, respec-
tively, of the material. From the von-Mises yielding criterion and the assumption
of plane strain in the axial direction, we can get
pffiffi

¼ 23ðrh þ P Þ,
r ð3Þ
where, rh is the stress in the circumferential or tangential directions. When the thick-
ness-diameter ratio smaller than 10, the thin wall theory can be applied for analysis
of a cylindrical tank (Beer and Johnston, 1992), in which the walls offer little resis-
tance to bending and a uniform stress distribution in the wall can be assumed. The
critical internal pressure, P, needed to make the tube deform plastically, thus, can be
obtained from the force equilibrium as
t
P ¼ rh : ð4Þ
ðR
tÞ
Substituting Eq. (3) into the above equation yields
2 t
P ¼ pffiffiffi r

: ð5Þ
3 R
(ii) Geometrical relationship at the first step of the expansion process.
The internal pressure for the first step of the expansion process is set to be
P1 = DP. Let the length of the first contact element be L1 as shown in Fig. 2(a),
and its thickness be equal to that of the free expansion part, i.e., t1 = tf1 then, the out-
er radius of the free expansion part can be obtained as R1 = R0
L1 From volume
constancy
ph i ph i
V 0 ¼ R20
ðR0
t0 Þ2 ¼ 2L1 t1 þ R21
ðR1
tf 1 Þ2 , ð6Þ
4 4
we can get
s
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 ffi
4 4
tf 1 ¼ R 1 þ L1
R1 þ L1
t0 ð2R0
t0 Þ: ð7Þ
p p
 Y.-M. Hwang, W.-C. Chen / International Journal of Plasticity 21 (2005) 1815–1833 1821

From Eq. (5), the pressure needed to make the tube deform plastically at this situa-
tion can be obtained as
2 tf 1
P 001 ¼ pffiffiffi r

1 , ð8Þ
3 R1
where (00 ) indicates the state at the new geometrical situation. r
1 can be obtained
from Eqs. (1) and (2) as long as the current thickness, t, is known. If the pressure
given by Eq. (8), obtained under the current geometrical relationship, is equal to
the pre-set pressure, i.e., P 001 ¼ P 1 then, the pre-set L1 is determined as its element
length for the first step and the process continues to the second step. Otherwise,
L1 has to be readjusted until P 001 ¼ P 1 is satisfied.
(iii) Stress states for step i + 1 before stretching.
Assume the expansion process has proceeded to step i, and the internal pressure
for step i + 1 is set as Pi + 1 = Pi + DP, where DP is a small pressure increment. As
mentioned earlier, the contact elements will probably continue to deform plastically
during the expansion process, which will result in the movement or sliding of the
contact elements on the corner side. Accordingly, the geometrical relationships of
the free expansion part will change. The stress states and the geometrical relation-
ships before deformation or stretching are described below.
At this moment, the tangential stress to the die, imposed on the new contact ele-
ment i + 1 from the free expansion part, can be obtained from Eq. (4) as
Ri
tfi
r0hðiþ1Þ ¼ P iþ1 , ð9Þ
tfi
where ( 0 ) indicates the state before stretching. It is possible for the contact elements
to deform plastically due to the pulling force from the free expansion part. The stress
state in each element has to be checked. If its equivalent stress is equal to or larger
than its yielding stress, that element is regarded as a deforming or stretching element,
and all the elements on the corner side will move or slide. The tangential tension
stress acting on element j can be obtained approximately by subtracting the frictional
resistance from the tension stress r0hðiþ1Þ as below
t0iþ1 1 X i
r0hj ¼ r0hðiþ1Þ
lL0 P iþ1 ð1 6 j 6 iÞ: ð10Þ
t0j t0j l¼j l

The last term of the above equation is the total amount of the frictional resistance by
all the elements on the corner side of element j. From Eq. (3), the equivalent stress
for element j is
pffiffiffi
0 3 0

j ¼
r ðrhj þ P iþ1 Þ ð1 6 j 6 iÞ: ð11Þ
2
From Eq. (2), the yielding stress for element j is
!!n
0 2 t0

yj ¼ K pffiffiffi ln 0
r ð1 6 j 6 iÞ: ð12Þ
3 tj
1822 Y.-M. Hwang, W.-C. Chen / International Journal of Plasticity 21 (2005) 1815–1833

0j P r
If r
0yj , element j is regarded undergoing plastic deformation, and its length will
stretch and its thickness becomes thinner.
(iv) Geometrical relationships for step i + 1 after stretching.
Let the total elongation of all the contact elements be S, which is distributed over
the deforming elements only. Assume the individual elongation of each deforming
element is in proportion to its length before stretching, Lj. Then, the lengths of
the deforming elements after deformation or stretching become
signðjÞL0j
L00j ¼ L0j þ S ð1 6 j 6 iÞ, ð13Þ
P
i
signðlÞL0l
l¼1
00
where ( ) indicates the state after stretching, and sign(j) is set to be unity if element j
undergoes deformation (i.e., r
0j P r

0yj ), otherwise sign(j) = 0. The last term of the
above equation means the individual elongation of the deforming elements. The
thickness of element j after stretching can be obtained from volume constancy as
t0j L0j
t00j ¼ ð1 6 j 6 iÞ: ð14Þ
L00j
Let the length of the new contact element for step i + 1 be Li + 1. The volume per unit
width of the free expansion part can be obtained as (see Fig. 2(b)):
ph i
V f ðiþ1Þ ¼ R2iþ1
ðRiþ1
tf ðiþ1Þ Þ2 , ð15Þ
4
Riþ1 ¼ Ri
Liþ1
S: ð16Þ
Just as the assumption for the first step, the thickness of element i + 1 is equal to that
of the free expansion part, i.e., ti + 1 = tf(i + 1). Thus, from volume constancy
Vf(i + 1) + 2Li + 1tf(i + 1) = Vfi, the thickness of the free expansion part at step i + 1
can be obtained as
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 ffi
4 4
tf ðiþ1Þ ¼ Riþ1 þ Liþ1
Riþ1 þ Liþ1
tfi ð2Ri
tfi Þ: ð17Þ
p p
From Eq. (5), we can get the internal pressure needed at this situation as
2 tf ðiþ1Þ
P 00iþ1 ¼ pffiffiffi r

iþ1 : ð18Þ
3 Riþ1
If P 00iþ1 ¼ P iþ1 , the pre-set Li + 1 is determined. Otherwise, Li + 1 has to be readjusted
until P 00iþ1 ¼ P iþ1 .
(v) Stress states for step i + 1 after stretching.
The tangential or circumferential stress acting on element i + 1 can be obtained as
Riþ1
tf ðiþ1Þ
r00hðiþ1Þ ¼ P 00iþ1 : ð19Þ
tf ðiþ1Þ

The tangential tension stress acting on element j can be obtained as

 Y.-M. Hwang, W.-C. Chen / International Journal of Plasticity 21 (2005) 1815–1833 1823

X
iþ1
L00l
r00hj ¼ r00hðiþ1Þ
00
lP 00iþ1 ð1 6 j 6 iÞ: ð20Þ
l¼j
t l

The form and variables in the above equation are exactly the same as those in Eq.
(10), only with the values after stretching. The equivalent stress and yielding stress
in element j after stretching are given, respectively, as:
pffiffiffi
00 3 00

j ¼
r ðrhj þ P 00iþ1 Þ ð1 6 j 6 iÞ, ð21Þ
2
!!n
00 2 t0

yj ¼ K pffiffiffi ln 00
r ð1 6 j 6 iÞ: ð22Þ
3 tj

00j P r
If r
00yj is checked again. At this stage, if all the contact elements do not deform
any more, i.e., m = 0, then, the pre-set total elongation of all the contact elements, S,
is determined, and the process goes to the next step. Otherwise, S has to be re-
adjusted until m = 0. Because the thickness and length of the contact elements vary
with the expansion process, they have to be recorded before going to the next step.
The flow chart for the calculation of L, S and other variables at steps 1 and i + 1 is
shown in Fig. 3.

2.2. Finite element model

A static-implicit FE code ‘‘DEFORM’’ is adopted to analyze the plastic flow pat-

tern of a circular tube expanded into a square cross-section. The finite element code
is based on the flow formulation approach using an updated Lagrange procedure.
The basic equation for the finite element formulation from the variational approach
is
Z Z Z
dp ¼ r
_

de dV þ C ev d_ev dV

_ si dvi dS ¼ 0, ð23Þ
v v SF

where
qffiffi qffiffi
3 0 0 1=2 1=2

¼
r ðr r Þ ,
2 ij ij

e_ ¼ 2
e_ e_ ,
3 ij ij
e_ v ¼ e_ ii :

where C is a very large positive number, which is called the penalty constant and is
interpreted as the bulk modulus. d
e_ and d_ev are the variations in strain rate and vol-
umetric strain rate, respectively, derived from an arbitrary variation dvi. si is the sur-
face traction stress. By the finite-element discretization procedure, Eq. (23) can be
converted to non-linear algebraic equations. For the detailed descriptions of the
finite element theory and the modeling formulations, please refer to Kobayashi
et al. (1989).
During the simulations, the die is assumed to be rigid, the tube is rigid-plastic, and
the interface between the tube and die has a constant friction coefficient. Four-node
isoparametric elements are used. The tube is divided into about 1200 elements and
there are 7 layers of elements in the thickness direction. The configurations of the
1824 Y.-M. Hwang, W.-C. Chen / International Journal of Plasticity 21 (2005) 1815–1833

Start

Input forming
conditions, t0 , R0, ∆ P

i =1

Set L1 , P1 = ∆ P

Calculate R1 , t f 1, P1" Change L1

No
P1" = P1
Yes
i =i +1

Pi +1 = Pi + ∆P

Set S

Set Li +1

Calculate
'' , σ "
Change S Pi +1 θ (i +1) , σ θ"j Change Li +1
σ "j , σ "yj , m

No
Pi"+1 = Pi +1

Yes
No
m=0
Yes
No
Pi +1 = P final

Yes

Print t , L , P , R , etc.

Stop

Fig. 3. Flow chart used in the numerical calculations during tube expansion in a square die.
 Y.-M. Hwang, W.-C. Chen / International Journal of Plasticity 21 (2005) 1815–1833 1825

Fig. 4. Configurations of the meshes in the tube during the expansion process: (a) before expansion; (b)
after expansion.

meshes for the tube before and after expansion are shown in Fig. 4(a) and (b), respec-
tively. The iteration methods adopted for solving the nonlinear equations are New-
ton–Raphson and the direct iteration methods. The direct iteration method is used to
generate a good initial guess for Newton–Raphson method, whereas the Newton–
Raphson method is used for fast final convergence. The convergence criteria for
the iteration are the velocity error norm jjDvjj/jjvjj 6 0.0001, and the force error
norm jjDFjj/jjFjj 6 0.0002, where jjvjj is defined as (vTv)1/2. When the prescribed
internal pressure reaches a pre-set value, the simulation is stopped, and the corner
radius and the thickness distribution of the tube are extracted.

3. Analytical results and discussion

The forming conditions used in the analytical model and FE-simulations of tube
expansion in a square die are R0 = 30 mm, t0 = 2 mm, and a = 30 mm. The outside
diameter of the tube is the same as the die width. The flow stress of the tube material,
AISI 1008 used in the analytical model and FE-simulations is r
¼ 657:2
e0:24 MPa.
Fig. 5 shows the effects of the friction coefficient upon the internal pressure during
the expansion process in a square die. For the derivation of the analytical equations
with the sticking friction mode for predicting the internal pressure, please refer to
Hwang and Chen (2003). The pressures predicted using the sticking friction mode
are slightly larger than those from this model and FE simulations. From this figure,
it is known that the effect of friction coefficient upon the forming pressure is not sig-
nificant. The reason is probably as follows: The relative slip between the tube and die
is allowable during the expansion process. When the friction coefficient increases, the
1826 Y.-M. Hwang, W.-C. Chen / International Journal of Plasticity 21 (2005) 1815–1833

250
AISI 1008,a=30mm,R0=30mm,t0=2mm
Sticking(thin-wall)
200
Sticking(thick-wall)
Internal pressure P(MPa)

Sliding(µ=0.01)
150 Sliding(µ=0.05)
Sliding(µ=0.1)
FEM(µ=0.0)
100
FEM(µ=0.05)
FEM(µ=0.1)

50 to
a

Ro

0
0 5 10 15 20 25 30
Corner radius R(mm)
Fig. 5. Effects of friction coefficient on forming pressure during the expansion process in a square die.

relative slip between them becomes less significant, thus, the thickness at the free
expansion part becomes thinner. Whereas, the strain-hardening effect resulting from
more thinning leads to a larger flow stress at the free expansion part of the tube.
From Eq. (5), it is known that a larger r
and a smaller t will compensate for one an-
other. Accordingly, the relationship between P and R is not significantly affected by
the friction coefficient. Failure of a deforming material is largely dependent on the
plastic strain the material underwent. A larger friction coefficient will result in a thin-
ner thickness or a larger strain at the free expansion part, thus, it is more likely to
lead to failure of the tube for a larger friction coefficient at the die–tube interface.
Fig. 6 shows the effects of friction coefficient upon the thickness distribution after
expansion. The curvilinear length q is measured from the center of the die as shown
in the figure. Obviously, as the friction coefficient increases, the thickness becomes
more non-uniformly distributed. In the sticking and sliding models, the free expan-
sion part is assumed to have a uniformly distributed thickness. Thus, the thickness at
the free expansion part appears to be a constant value. From the FE-simulations, it
is noteworthy that the minimum thickness occurs at the very intersection of the
expansion part and the contact part, rather than at the corner part. The reason is
probably that the tube at the intersection (at about q = 25 mm) is subjected simulta-
neously to a horizontal tension stress and a vertical compression stress. The thick-
ness distribution from this model with l = 0.5 is quite close to that by the sticking
model, which means the contact situation for l P 0.5 can be regarded as the sticking
friction mode. On the other hand, the predicted thickness of the formed tube for
l = 0 from the FE simulations is not completely uniformly distributed; that is be-
cause different stress states exist in the contact part and the free expansion part.
 Y.-M. Hwang, W.-C. Chen / International Journal of Plasticity 21 (2005) 1815–1833 1827

1.1
AISI 1008,a=30mm,Ro=30mm,to=2mm
1

0.9
Relative thickness t/to

0.8

0.7

0.6 to
a
0.5 Ro

0.4
0 5 10 15 20 25 30 35
Length from center q(mm)

Sticking friction q=0 q=25mm

Sliding(µ=0.01)
Sliding(µ=0.05)
Sliding(µ=0.1)
Sliding(µ=0.5)
FEM(µ=0.0)
FEM(µ=0.05)
FEM(µ=0.1)
Fig. 6. Effects of friction coefficient on thickness distribution after tube expansion in a square die.

4. Experiments on tube expansion

The experimental apparatus used for tube expansion in a square die was designed
by our laboratory (Tsai, 2002). This testing machine is composed of three parts: a
tube expansion set, a hydraulic power system, and a pressure intensifier; and its max-
imum allowable working pressure inside the tube can reach 100 MPa. The tubes pro-
vided by the vendor are AA6063-T4. In order to increase their formability, the tubes
were further annealed at 410 °C for two hours before expansion experiments. The
initial outer diameter and thickness of the AA6063 tube are 76 and 4.83 mm, respec-
tively. The flow stress of the tubular material is obtained by a tensile test at an aver-
age strain rate of 0.01 s
1. The test piece is cut directly from the tubes for expansion
experiments. From the stress–strain curve, shown in Fig. 7, the flow stress can be
expressed approximately by a power law of its equivalent strain. The K, n values
obtained are 225.8 MPa and 0.402, respectively.
1828 Y.-M. Hwang, W.-C. Chen / International Journal of Plasticity 21 (2005) 1815–1833

160

120
Stress (MPa)

80

40
Curve fitting with σ = 228.5 ε 0.402
True stress-strain
Engineering stress-strain

0
0 0.2 0.4 0.6
Strain
Fig. 7. Flow stress of the tubular material used in the models.

The formed parts with different internal pressures are shown in Fig. 8, where their
length and width are 150 and 76 mm, respectively. The location, at which the thick-
ness distribution is measured, is at the middle part between the upper and lower ends
of the product, where the strain state can be regarded as a plane strain.
In order to obtain the friction coefficient l between the tube and die, the so-called
ring compression test is conducted. The aspect ratio of the test piece is 6:3:2. The
analytical reductions in the internal diameter of the ring by the FE simulations
and the measured values are shown in Fig. 9. The lubricant used for lubrication at
the tube–die interface is a kind of circulation oil R68, which is used in the hydraulic

Fig. 8. Products after hydraulic expansion with various internal pressures.

 Y.-M. Hwang, W.-C. Chen / International Journal of Plasticity 21 (2005) 1815–1833 1829

60 µ=0.5
OD=42mm, ID=21mm, H=14mm
50 0.25
Experiments
0.2
40 Lubrication
Decrease in internal diameter (%)

0.15
30 No lubrication

20
0.1
10
0.075
0

-10
0.05
-20
OD
-30 ID
0.025
-40
H

0.01
-50

-60
0 10 20 30 40 50 60 70
Reduction in height (%)

Fig. 9. Calibration curves and experimental data for determination of friction coefficient.

power system. From this figure, it is known that the analytical curve with l = 0.25
fits best with the experimental values with the same lubricant. The reason for obtain-
ing a high friction coefficient in this paper is that the circulation oil R68 is one kind
of liquid-state lubricant. The lubrication film of liquid-state lubricants is easily de-
stroyed at high-pressure interfaces, and the lubrication effect decreases. Solid-state
lubricants that can withstand high pressures are recommended.
Fig. 10 displays the comparisons of the internal pressures between the sticking
model (Hwang and Chen, 2003), sliding model, finite element simulation and exper-
imental results. The forming conditions are t0 = 4.83 mm, R0 = 38 mm, a = 38 mm
and l = 0.25 as shown in the figure. From this figure, it is known that the forming
pressure increases with the decrease of the corner radius.The forming pressure pre-
dicted by the thin-wall theorem of sticking model is the largest among the analytical
results. At the early stage, the FE-simulations are closer to the experimental values,
whereas, at the late stage, the analytical values by this model are closer to the exper-
imental results.
Fig. 11 shows the comparisons of thickness distributions after expansion between
analytical and experimental results. The thickness variation increases with increasing
friction coefficient. In the sticking model, the elements after contact are assumed to
1830 Y.-M. Hwang, W.-C. Chen / International Journal of Plasticity 21 (2005) 1815–1833

140
K=225.8MPa,n=0.402,a=38mm,
to=4.83mm,Ro=38mm,µ=0.25
120
Experiment
100 Sticking(thin-wall)
Internal pressure P(MPa)

Sticking(thick-wall)
80
Sliding
FEM

60

40
R
to
a

20
Ro
0
0 10 20 30 40
Corner radius R(mm)
Fig. 10. Comparisons of the internal pressures between mathematical models and experimental results.

be without movement, thus, the thickness at the center of the die is equal to the ini-
tial thickness of the tube. Clearly, the FE-simulations fit the experimental values
best, and it is known that the thinnest points occurred at around the intersection
of the free expansion part and contact part (at about q = 28 and 44 mm). On the
other hand, the thickness distribution by this model is closer to the experimental val-
ues than that by the sticking model. The predicted thinnest locations by this model
and FEM are all within or around the free expansion region. The main differences of
the analytical results between this model and FEM are the obtained thinnest thick-
ness. The stress state in the free expansion region is assumed to be uniform in this
analytical model, thus, the thickness throughout the free expansion region is equal
to that at the conjunction point of the contact and free expansion regions. However,
the real stress states at the middle part and the two ends of the expansion region are
different from each other, thus, there is thickness variation in the free expansion re-
gion as obtained by the FEM and experiments. The reason for the occurring of the
minimum thickness at the intersection of the free expansion and the contact regions
instead of at the middle part of the free expansion region is probably because the
tube at the conjunction point is subjected simultaneously to a horizontal tensile
stress and a vertical compressive stress (for q = 28 mm). The discrepancy of the min-
imum thicknesses between this model and the experimental value is about 5% of
the initial tube thickness, which is considered to be acceptable in engineering
 Y.-M. Hwang, W.-C. Chen / International Journal of Plasticity 21 (2005) 1815–1833 1831

1
K=225.8MPa, n=0.402, a=38mm,
0.95 to=4.83mm, Ro=38mm, µ=0.25

0.9

0.85
Relative thickness t/to

0.8

0.75

0.7

0.65 a

0.6 to
a

0.55 Ro
0.5
0 10 20 30 40 50 60 70
Length fromcenter q(mm)
q=0 q=30
Experiment
q=42
Sticking friction
Sliding
FEM
q=72

Fig. 11. Comparisons of the thickness distribution of the products between mathematical models and
experimental results.

applications. This numerical method, however, failed to account for the thickening
effect at the corner observed in the FEM and experimental results shown in Fig. 11.

5. Conclusions

The plastic deformation of a circular tube hydraulically expanded into a square

cross-section was explored using analytical models and the finite element method.
The forming pressures and thickness distributions of the formed products were dis-
cussed. The forming pressures predicted by this model indicate good agreement with
the experimental values and the thickness distributions using this model are much
closer to the experimental results than those by the sticking model. This model
can be extended to simulate the plastic deformation of the tube during the expansion
in a rectangular die, if the formulation is slightly modified and some contact condi-
tions are added to the flow chart of the numerical calculations. In the future, the
1832 Y.-M. Hwang, W.-C. Chen / International Journal of Plasticity 21 (2005) 1815–1833

effects of various forming conditions, such as the aspect ratio of the die, etc., upon
the forming results will be systematically discussed.

Acknowledgements

The authors extend their thanks to the National Science Council of the Republic
of China under Grant No. NSC 91-2212-E-110-016.

References

Asnafi, N., 1999. Analytical modeling of tube hydroforming. Thin-Walled Structures 34, 295–330.
Asnafi, N., Skogsgardh, A., 2000. Theoretical and experimental analysis of stroke-controlled tube
hydroforming. Material Science and Engineering A279, 95–110.
Banerjee, J.K., 1975. Limiting deformations in bulge forming of thin cylinders of fixed length.
International Journal of Mechanical Sciences 17, 659–662.
Beer, F.P., Johnston, E.R., 1992. Mechanics of materials, second ed. McGraw-Hill Inc., New York.
Chen, Z.T., Worswick, M.J., Cinotti, N., Pilkey, A.K., Lloyd, D., 2003. A linked FEM-damage
percolation model of aluminum alloy sheet forming. International Journal of Plasticity 19, 2099–2120.
Dohmann, F., Hartl, Ch., 1996. Hydroforming – a method to manufacture lightweight parts. Journal of
Materials Processing Technology 60, 669–676.
Dohmann, F., Hartl, Ch., 1997. Tube hydroforming: research and practical application. Journal of
Materials Processing Technology 71, 174–186.
Fuchizawa, S., Narazaki, M., 1989a. Analysis of bulge deformation of thin tube with cylindrical die.
Journal of Japanese Society for Technology of Plasticity 30, 116–122.
Fuchizawa, S., Narazaki, M., 1989b. Bulge deformation of thin copper tube with cylindrical die. Journal
of Japanese Society for Technology of Plasticity 30, 520–525.
Gotoh, M., Yamashita, M., 2003. An aspect of plasticity with compressibility. International Journal of
Plasticity 19, 383–401.
Ho, K.C., Lin, J., Dean, T.A., 2004. Modeling of spring back in creep forming thick aluminum sheets.
International Journal of Plasticity 20, 733–751.
Hwang, Y.M., Altan, T., 2003. Finite element analysis of tube hydroforming processes in a rectangular
die. Finite Elements in Analysis and Design 39, 1071–1082.
Hwang, Y.M., Chen, W.C., 2003. Analysis and finite element simulation of tube expansion in a
rectangular cross-sectional die. Proceedings of the Institution of Mechanical Engineers, Part B: Journal
of Engineering Manufacture 217, 127–135.
Hwang, Y.M., Lin, Y.K., 2002. Analysis and finite element simulation of the tube bulge hydroforming
process. Journal of Materials Processing Technology 125, 821–825.
Hwang, Y.M., Lin, Y.K., 2003. FE-simulations of T-shape tube hydroforming. Key Engineering Materials
233, 317–322.
Hwang, Y.M., Shr, S.G., Altan, T., 2000. Analysis of tube hydroforming in a square cross-sectional die.
International Journal of Engineering Simulation 1, 9–16.
Johnson, K.I., Nguyen, B.N., Davies, R.W., Grant, G.J., Khaleel, M.A., 2004. A numerical process
control method for circular-tube hydroforming prediction. International Journal of Plasticity 20, 1111–
1137.
Kobayashi, S., Oh, S.I., Altan, T., 1989. Metal forming and the finite element method. Oxford, New York.
Koc, M., Altan, T., 2002. Application of two dimensional (2D) FEA for the tube hydroforming process.
International Journal of Machine Tools & Manufacture 42, 1285–1295.
Koc, M., Aue-u-lan, Y., Altan, T., 2001. On the characteristics of tubular materials for hydroforming-
experimentation and analysis. International Journal of Machine Tools & Manufacture 41, 761–772.
 Y.-M. Hwang, W.-C. Chen / International Journal of Plasticity 21 (2005) 1815–1833 1833

Kridli, G.T., Bao, L., Mallick, P.K., Tian, Y., 2003. Investigation of thickness variation and corner filling
in tube hydroforming. Journal of Materials Processing Technology 133, 287–296.
Li, S., Hoferlin, E., Bael, A.V., Houtte, P.V., Teodosiu, C., 2003. Finite element modeling of plastic
anisotropy induced by texture and strain-path change. International Journal of Plasticity 19, 647–674.
Manabe, K., Amino, M., 2002. Effects of process parameters and material properties on deformation
process in tube hydroforming. Journal of Materials Processing Technology 123, 285–291.
Rama, S.C., Ma, K., Smith, L.M., Zhang, J.M., 2003. A two-dimensional approach for simulation of
hydroforming expansion of tubular cross-sections without axial feed. Journal of Materials Processing
Technology 141, 420–430.
Smith, L.M., Averill, R.C., Lucas, J.P., Stoughton, T.B., Matin, P.H., 2003. Influence of transverse
normal stress on sheet metal formability. International Journal of Plasticity 19, 1567–1583.
Sokolowski, T., Gerke, K., Ahmetoglu, M., Altan, T., 2000. Evaluation of tube formability and material
characteristics: hydraulic bulge testing of tubes. Journal of Materials Processing Technology 98, 34–40.
Stoughton, T.B., Yoon, J.W., 2004. A pressure-sensitive yield criterion under a non-associated flow rule
for sheet metal forming. International Journal of Plasticity 20, 705–731.
Stupkiewicz, S., Mroz, Z., 2003. Phenomenological model of real contact area evolution with account for
bulk plastic deformation in metal forming. International Journal of Plasticity 19, 323–344.
Tsai, G.W., 2002. Experiments and simulations of tube hydraulic forming. Master Thesis, National Sun
Yat-Sen University, Kaohsiung, Taiwan (in Chinese).
Woo, D.M., 1973. Tube-bulging under internal pressure and axial force. Journal of Engineering Materials
Technology 95, 219–223.
Woo, D.M., Lua, A.C., 1978. Plastic deformation of anisotropic tubes in hydraulic bulging. Journal of
Engineering Materials Technology 100, 421–425.
Yoon, J.W., Barlat, F., Dick, R.E., Chung, K., Kang, T.J., 2004. Plane stress yield function for aluminum
alloy sheets-part II: FE formulation and its implementation. International Journal of Plasticity 20,
495–522.