You are on page 1of 12

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/245387235

Parameter study of stretch–blow moulding process of polyethylene


terephthalate bottles using finite element simulation

Article in Proceedings of the Institution of Mechanical Engineers Part B Journal of Engineering Manufacture · August 2010
DOI: 10.1243/09544054JEM1853

CITATIONS READS

11 26,652

3 authors, including:

Saeed Bagherzadeh
Purdue University
21 PUBLICATIONS 317 CITATIONS

SEE PROFILE

All content following this page was uploaded by Saeed Bagherzadeh on 13 October 2015.

The user has requested enhancement of the downloaded file.


1217

Parameter study of stretch–blow moulding process


of polyethylene terephthalate bottles using finite
element simulation
S Bagherzadeh1*, F R Biglari1, and K Nikbin2
1
Mechanical Engineering Department, Amirkabir University of Technology, Tehran, Iran
2
Mechanical Engineering Department, Imperial College London, London, UK

The manuscript was received on 15 October 2009 and was accepted after revision for publication on 26 November 2009.

DOI: 10.1243/09544054JEM1853

Abstract: In this paper, the analysis of the stretch–blow moulding (SBM) process of poly-
ethylene terephthalate (PET) bottles is studied by the finite element method (FEM). In this
simulation, owing to the symmetry of bottle geometry, the parts are considered as an axisym-
metric model. A hyperelastic constitutive behaviour was calibrated using material data avail-
able in literature in variant high temperatures and strain rates and was used in the numerical
simulation. Hydrostatic pressure with convention heat transfer has been used instead of a
blowing process. Comparisons of numerical results with experimental observations demon-
strate that the model can predict an overall trend of thickness distribution, especially in regions
between 30 mm to 90 mm from the bottle bottom. However, some differences can be seen in
regions 10 mm and 125 mm. These results can be used for an overall prediction of bottle
properties such as final bottle thickness and a defect-free production because they are governed
by orientation and crystallinity, which are highly temperature and strain dependent. Also, with
definition of the critical area in bottle forming, the parametric studies are conducted on the
effect of friction condition, heat transfer coefficient, and initial pre-blowing air entrance time
delay on bottle thickness. Through the study, it becomes clear that the proposed model is
applicable for simulating the stretch-blow moulding process of PET bottles, and is capable of
offering helpful knowledge in the production of bottles and the design of an optimum preform.

Keywords: constitutive behaviour, critical area, effective parameters, finite element analysis,
PET, stretch-blow moulding

1 INTRODUCTION bottles with desired outer shapes. Finally, the bottles


are cooled and ejected from the mould (Fig. 1).
Injection stretch–blow moulding (ISBM) is one of the Finite element analysis (FEA) can be used to
major methods in blow-moulding processes for simulate the blowing process for optimization pur-
making hollow polyethylene terephthalate (PET) poses. The prediction of the deformation process of
bottles. In this process, the PET resin is first injected polymers has been the subject of widespread
into an axisymmetric U-shaped mould to make research over the past three decades. A lot of research
structurally amorphous preforms called ‘preform’. efforts are based on experimental techniques to
Then, the preforms are heated in an infrared oven realize accurate deformation mechanics of polymers
higher than the glass transition temperature. Next, it [1, 2]. Other investigations have tended to simulate or
is stretched simultaneously with a stretch rod and predict the mechanical behaviour of the polymer
blown up with high-pressure air inside, to create under different loading conditions. Schmidt et al. [3]
presented a viscoelastic model for PET materials
*Corresponding author: Mechanical Engineering Department, in FEA of the stretch-blow moulding process based
Amirkabir University of Technology, Hafez Avenue, Tehran, on an updated Lagrangian method. This simulation
Iran. showed a reasonable material movement and defor-
email: sd.bagherzadeh@yahoo.com mation process but very limited accuracy in thick-

JEM1853 Proc. IMechE Vol. 224 Part B: J. Engineering Manufacture


1218 S Bagherzadeh, F R Biglari, and K Nikbin

ness prediction. Ogden’s model of elasticity is widely stretch rod. Particular attention was paid to the
used in thermoforming to give relatively accurate thermal and contact modelling, material model, and
thickness predictions. However, it neglects the rate selection of proper element types.
effect of deformation [4]. Christensen [5] proposed a
viscoelastic model consisting of elastic and viscous
components. The strain rate tensor appeared in the 2 CONSTITUTIVE BEHAVIOUR OF PET
viscous part of the formulation to consider the strain
rate effect. However, this model could not represent According to Fig. 1, in the stretch–blow moulding
the behaviour of PET very well. Yang et al. [6] sug- process, the bottle wall experiences a biaxial tension
gested the modified Christensen model as a visco- in two axial and hoop directions. Then, the numerical
hyperelastic model. This study, performed on rubber, model of large biaxial deformation of the material at
stated that the elastic part was as Mooney-Rivilin and elevated temperature is taken into account for a high
the viscous part was expressed with invariants of the strain rate which is similar to whatever takes place in
deformation tensor. Pham [7] presented phenomen- the SBM process. Also, in the preform heating step
ological hyperelastic models such as Ogden and before stretching, the preform is heated from ambi-
Mooney-Rivilin that predict the strain hardening rea- ent temperature to the temperature of the forming
sonably well, but not the strain rate effect. Haberstroh process by infrared radiation from a series of radiator
and Detrois [8] reported some experimental work on ovens. This is the main reason for defining the
the orientation and crystallization of PET films sub- material behaviour of PET in the different tempera-
jected to uniaxial and biaxial drawing under industrial tures. The material behaviour under biaxial defor-
processing conditions. Denardin et al. [9] demon- mation can be obtained using the membrane-
strated the large plastic deformation of PET. The inflation rheometer that has been carried out by
stress–strain curves had a strong dependency to tem- Haberstroh and Detrois [8]. The results of these
perature and strain rate. The stress–strain curves measurements are stress–strain curves as shown in
Fig. 2 for PET at very high strain rates (10 s1) in the
obtained at different temperatures and strain rates can
three high temperature levels as occurs during the
be useful for engineering applications.
SBM process.
In the last two decades, numerical modelling
The stress–strain plot at elevated temperatures
(mainly finite element modelling (FEM)) has received
shows a significant increase after a certain stretch
increasing attention in the simulation of plastic
ratio. A small increase in stress value is seen at an
moulding. In the literature [10–13], the numerical
elongation of 2 per cent owing to the high amount of
simulation of the SBM process has been investigated
straining induced by crystallization. This is normally
by an FE method. Pham et al. [14] have conducted
called the strain hardening effect. A hyperelastic strain
the simulation of SBM process using BlowSIM soft-
energy function of third order was used as a con-
ware, which has some shortcomings in the definition
stitutive model for the deformation material behaviour
of true material behaviour.
[15]. The material is considered to be incompressible.
This study presents a fully coupled temperature– The specific strain energy W is expressed as a function
displacement finite element simulation of SBM of of the first invariant I1 of the strain tensor
PET bottles using ABAQUS. In this model, the con-
stitutive behaviour of PET is modelled as hyperelastic W ¼ C10 ðI1  3Þ þ C20 ðI1  3Þ2 þ C30 ðI1  3Þ3 ð1Þ
materials. Hydrostatic pressure and convection heat
transfer has been used for air flow modelling. The gap
conductance model have been used for modelling of
the preform thermal contact with both a mould and

Fig. 1 Stretch–blow moulding process steps Fig. 2 Stress–strain curve for PET at 10 s1 strain rate

Proc. IMechE Vol. 224 Part B: J. Engineering Manufacture JEM1853


Parameter study of stretch–blow moulding process 1219

where W is the strain energy potential, Cij are mat- Ci0 constants have been determined using the least-
erial parameters which may be a function of tem- squares method and literature [8]. In Fig. 2, A, B, C, and
perature, and I1 is first invariant that measures the D are polynominal constants in the s versus l curve
distortion in the material. Also I1 can be written as fits. The Ci0 constants can be obtained from three
values of s versus arbitrary values of l that were cal-
I1 ¼ l21 þ l22 þ l23 ð2Þ culated from equation s ¼ Al3 þ Bl2 þ Cl þ D. Then
where li are the principle stretches for equibiaxial these three pairs of s versus l were adopted into
deformation l1 ¼ l2 ¼ l, assuming that the material is equation (6) to obtain the Ci0 constants. In the next
incompressible. Therefore the first invariant can be step, the Ci0 constants have been entered into ABAQUS
achieved as material properties as the third-order strain energy
model. In ABAQUS code a strain energy potential (W)
l1 l2 l3 ¼ 1 ð3Þ was used rather than a Young’s modulus and Poisson’s
ratio, to relate stresses to strains in hyperelastic mat-
1 erials. In this paper, the material parameters have been
l3 ¼ ð4Þ obtained from published experimental results [8] of
l2
biaxial tension of PET under high strain rate at different
With substitution of equations (3) and (4) into equa- temperatures. A different test data curve fit was eval-
tion (2) uated using embedded hyperelasticity models in
ABAQUS. Comparison of imported test data showed a
I1 ¼ 2l2 þ l4 ð5Þ
close agreement with the third-order reduced poly-
Applying the theory of virtual work yields for the nominal model. Therefore in this simulation, the
stress obtained Ci0 constants based on the third reduced
polynominal model have been used to describe PET
1 @W @I1 1 behaviour during the SBM process.
s¼ ¼ ½C10 þ 2C20 ð2l2 þ l4  3Þ
2 @I1 @l 2
þ 3C30 ð2l2 þ l4  3Þ2 
·ð4l  4l5 Þ ð6Þ 3 NUMERICAL SIMULATION

where s is the stretching stress and C10, C20, and C30 Figure 3(a) shows the FE model that was used in this
are material constants in the hyperelastic model. The study. Because of model symmetry only half of the

Fig. 3 FE model: (a) meshed preform; (b) initial temperature distribution of the preform

JEM1853 Proc. IMechE Vol. 224 Part B: J. Engineering Manufacture


1220 S Bagherzadeh, F R Biglari, and K Nikbin

geometry is modelled, and axisymmetric elements relation shown in Table 3. When g > 50 mm the heat
were adopted. Numerical simulations have been transfer is deemed negligible. When g is limited to
performed using FE code ABAQUS 6.7 as a fully between 0.001 mm and 50 mm, it represents the
coupled temperature-displacement with ABAQUS/ convective heat transfer between the preform outer
Standard solver. Figure 3(b) shows the initial tem- surface and the ambient air. When g 6 0.001 mm, the
perature distribution in the preform. mould and the preform gradually contact each other
In this model, preform and stretch rod were with a varying conductivity from h0 ¼ 10 W/K m2 to
assumed as deformable parts and the mould was h1 ¼ 500 W/K m2. The same approach is used to
considered as a rigid part. Some of the geometrical model heat transfer between preform inner surface
data for modelled preform and mould have been lis- and the stretch rod, except that only heat transfer
ted in Table 1. after contact is considered, since convective heat
Both the stretch rod and mould materials are made transfer has been accounted for by equation (7).
of aluminium and assumed to behave as linear elastic
materials. The properties of the aluminium and PET
materials at room temperature are shown in Table 2.
In this simulation, the heat transfer between the 4 EXPERIMENTAL STUDY
high-pressure air and the inner surface of preform
that does not contact the stretch rod was modelled by Experimental works of the stretch–blow moulding
surface convection. The heat transfer as surface process were based on the production-line of man-
convection in ABAQUS (option *Sfilm) has been for- ufactured PET bottles. The stretch–blow moulding
mulated as follows [16] process was performed on the Sidel blowing
machine, which makes the 330 ml PET bottles. The
qa ¼ ha ðu  ua Þ ð7Þ preforms experience a heating step in the induction
heating oven before entrance to the mould cavity
where u is the preform temperature, ua the high-
with an approximate temperature gradient as shown
pressure air temperature (30  C), qa the heat flux on
in Fig. 3(b). The PET material with BG-781 industrial
preform inner surface, and ha is the film coefficient
grade was used for this work. Some of the adjusted
(ha ¼ 10 W/K m2 [13]). Also, the heat transfer between
and measured data in the Sibel machine for the SBM
the mould and the preform outer surface was mod-
process have been listed in Table 4.
elled by gap conductance theory in ABAQUS as below
Before processing, the preform was marked by
[16]
points at specific positions along the complete length
qm ¼ hm ðu  um Þ ð8Þ of the preform; these were used for verification of
material movement during the process. Thickness
where um is the mould temperature (10  C), hm is the measurements were conducted using a sensitive
gap conductance coefficient, and qm is the heat flux
per unit area from points on the mould surface to
points on preform outer surface. The gap con- Table 3 The clearance-gap conductance rela-
ductance coefficient (hm) depends on the gap (g) tion in thermal model [13]
between the two surfaces. This study uses a g–hm
Thermal conduction Gap between surfaces,
coefficient, hm (W/K m2) g (mm)
Table 1 Preform and mould geometries
500 0
Middle Inner Overall 10 0.001
Length radius radius thickness 0 50
Component (mm) (mm) (mm) (mm)

Preform 77 23 10 2.6
Mould 151 60 25 – Table 4 Preform type and measured process
conditions

Parameters Value
Table 2 Material properties Preform (grade, weight) PET BG-781, 20 gr
  3 
Average heating temperature ( C) 105
Mat. E: GPa C: J/kg C a: 1/ C r: kg/m K: W/m C v Pre-blowing starting (s) 0.15
Pre-blowing pressure (MPa) 0.6
Al 70 963 1.2 · 105 2740 159 0.3 Main blowing starting time (s) 0.436
PET – 1200 7 · 105 1050 0.24 – Main blowing pressure (MPa) 3.8
Ejection time (s) 1.826
(Note: E: elastic modulus, C: specific heat, a: coefficient of ex- Stretching displacement (mm) 69.6
pansion, r: density, K: conductivity, v: Poisson ratio)

Proc. IMechE Vol. 224 Part B: J. Engineering Manufacture JEM1853


Parameter study of stretch–blow moulding process 1221

adjustable fine-loading micrometer (Mitutoyo-Series the bottle bottom. An over-stretch phenomenon can
227-Japan). be seen in the experimental results of thickness dis-
tribution between 110 mm to 130 mm, which may be
attributable to the higher temperature of that region
of the mould and a faster thinning effect in that
5 RESULTS AND DISCUSSION
region. There are several reasons for the difference
5.1 Thickness distribution and material between these results, such as temperature-depen-
movement dent viscous behaviour of PET and measurement
error at wall thickness.
From simulation results, eight segments of the mod- Figure 6 shows the material movement at three
elled deformation process are shown in Fig. 4. As they defined positions in the preform. The solid lines
show, the deformation process generally consists of represent the FE model movement and the dashed
three steps. First, the preform is expanded by the
stretch rod (full stretching step). After that, low air
pressure is applied to prevent the preform from
necking while touching the stretch rod from the
bottom (pre-blowing step). After the stretch rod
reaches the final position, the full blow pressure is
applied and the bottle is deformed in a radial direc-
tion (main blowing step).
Figure 5 compares the predicted bottle wall thick-
ness distribution obtained from numerical modelling
with the experimental measurements. A general
agreement in trend of thickness distribution can be
seen in the regions between 30 mm to 90 mm from

Fig. 5 Predicted bottle thickness distribution

Fig. 4 Bottle formation process Fig. 6 Material movement at three specific points

JEM1853 Proc. IMechE Vol. 224 Part B: J. Engineering Manufacture


1222 S Bagherzadeh, F R Biglari, and K Nikbin

lines represent the measured points that are mea- meters such as the crystallization and mechanical
sured from experimental observations. It can be seen strength of the final bottles.
that the simulation models the deformation of
underneath the bottle area with lower accuracy. 5.3 The critical areas for tearing tendency
These differences in the numerical predictions com- in the SBM process
pared with experimental observations can be attrib-
uted to either highly viscous behaviour of PET or that The simulation is expected to predict the defects and
the infrared camera could not accurately measure the critical regions in the process by a careful examin-
initial preform temperature distribution, which ation of the thickness distribution and material
greatly affects the process. movement. A tearing phenomenon in the bottle walls
is a common defect that can be seen in those regions
with very high equivalent plastic strain. Figure 9
5.2 Temperature distribution and cooling
presents equivalent plastic strain distribution on the
history
The modelled temperature distribution has been
presented in the bottle bottom and neck region at the
end of the process (Figs 7(a) and (b)). It can be
observed that the two ends remain warm, with great
temperature differences between the inner and outer
surfaces, while at this time the other part of the bottle
is comparatively cold.
The history of temperature that represents the
cooling rate is plotted in Fig. 8. The temperature
records of four nodes that are located at inner/outer
surfaces of two locations are shown in Fig. 8. It can be
noticed that the temperature drops significantly from
one location to another. The cooling rates on the
outer surface are generally much higher than the
inner surface at the pre-blowing stage because of Fig. 8 Temperature history during process for neck and
the low thermal conductivity of the PET material. bottom locations (cooling history)
These differences can be related to various para-

Fig. 7 Temperature distribution contour for neck and bottom regions

Proc. IMechE Vol. 224 Part B: J. Engineering Manufacture JEM1853


Parameter study of stretch–blow moulding process 1223

Fig. 9 Final equivalent plastic strain distribution in outer/ Fig. 10 Effect of friction coefficient on bottle bottom
inner surfaces thickness

bottle inner and outer surfaces. As can be observed,


the strain distribution has a gradual trend in most
parts of the regions except at the bottle bottom
region. The main reason is that the stretch rod con-
tacts with preform for a long time during the process.
This can lead to different heat transfer conditions in
inner and outer surfaces of the preform, and various
strain distribution through the thickness. In Fig. 9,
this significant difference between plastic strains in
the inner and outer area of the bottle bottom leads to
shearing. It can also be a reason for an increased
tendency towards tearing defect in this region.

6 PARAMETER STUDY Fig. 11 Bottle bottom thickness distribution for various


heat transfer coefficients
6.1 The effect of process parameters on
bottom bottle thickness Similar to the friction coefficient, the conduction
The most common problems in the SBM process are heat transfer coefficient affects only the bottom
the thickening of the bottle bottom, tearing at the region, which is initially in contact with the stretch
contact regions with the stretching rod, and non- rod. A higher h1 leads to a thicker bottle bottom. This
uniform wall thickness distribution. These problems is due to a faster cooling effect of higher h1values and
are usually observed in the production line of PET prevention of stretching in the bottom region of the
bottles and are overcome by production parameters. bottle.
The optimal adjustable production parameters aim
to attain uniform bottle wall thickness. Therefore, 6.2 Study on the preform growth and air
three important parameters such as friction coeffi- entrance time
cient, convection heat transfer coefficient, and pre-
blowing air entrance time are studied in this paper. Figure 12 shows forces acting on a segment of pre-
Figure 10 shows a parameter study on the effect of form during stretch blowing. Consider a segment of
the sticking friction coefficient (m) on the bottle the preform with a length of l as illustrated in Fig. 12,
bottom thickness. Any increase of friction coefficient where the forces acting on the preform segment are
(m) between the rod and inner bottle surfaces could also shown. The stretch force, F, imposed by the
result in an increase of bottom thickness and stretch rod, and the air pressure, P, were assumed to
mechanical strength in this region owing to more be constant in the stretching and blowing stage. The
contact cooling with the rod surface. This might also longitudinal force caused by only stretching the pre-
increase the amount of material waste. form is equal to F. After imposing the air pressure, the
Also, Fig. 11 presents effects of the heat transfer preform was bulged at different inflation rates along
coefficient (h1) on the bottle bottom thickness. its length. While being stretched and blown

JEM1853 Proc. IMechE Vol. 224 Part B: J. Engineering Manufacture


1224 S Bagherzadeh, F R Biglari, and K Nikbin

Fig. 12 (a) Schematic preform obtained by only stretching; (b) forces acting on preform segment

simultaneously, the longitudinal


 force acting on the As can be seen, the two parameters are only
preform is equivalent to F þ PpR2m , where Rm is dependent on the thickness d and internal radius R of
the maximum internal radius along the preform. the preform. According to initial preform shape, the
The longitudinal and hoop stresses acting on the thickness is changed throughout the preform wall.
preform segment, denoted as s1 and sh, respec- So, the value of m was different along the preform
tively, can be calculated by [17] walls, which can result in variable longitudinal stress
8 9 while stretching the preform. The thickness for the
> F F F
>
< 2 ¼ ¼ ðstretchedÞ >
>
= region near the preform support ring was smaller
pd þ 2pRd pdðd þ 2RÞ pm
sL ¼ than that of other regions. Although the temperature
>
> F þ PpRm2
F þ PpRm2
>
: ¼ ðstretched-blownÞ > ; near the support ring was relatively lower (as shown
pdðd þ 2RÞ pm
previously in Fig. 3(b)), the strain of this region was
ð9Þ larger than that of other regions of the preform after
being stretched by the rod owing to larger long-
2RlP R itudinal stress. This leads to an obvious decrease of
sh ¼ ¼ P ¼ Pn ð10Þ
2ld d the thickness for this region. The internal radius,
however, was not decreased much. So the value of
where d and R are the instantaneous thickness and
the n for this region was much larger than that of
internal radius profiles along the preform during the
other regions. These observations show that the
stretching and blowing respectively. As shown in
initial thickness distribution of preform and time
equations (9) and (10), two parameters, m and v, are
delay in pre-blowing air entrance have considerable
defined for further analyses
effects on final bottle thickness.
m ¼ dðd þ 2RÞ ð11Þ For investigation of air entrance time effect,
three sets of FE simulation with different initial pre-
n ¼ R=d ð12Þ blowing delay times were performed. These delay
times (Td) were 0.15, 0.18, and 0.2 s. The observations

Proc. IMechE Vol. 224 Part B: J. Engineering Manufacture JEM1853


Parameter study of stretch–blow moulding process 1225

during deformation steps showed that at Td ¼ 0.15 s thinning of the bottle bottom as well as thicker bottle
the preform is separated from the stretching rod after upper wall are expected (Fig. 13(a).
the air entrance. This allows shorter contact between In the other two cases (Td ¼ 0.18 s, 0.2 s) less
the bottle bottom and stretching rod with hotter thinning in the bottle bottom is observed, as well
material inflation in this region. Therefore, increased as less thickening in the upper regions of the
bottle. The difference between these models is in
fact in the initial preform stretching time or pre-
blowing time delay. With a higher pre-blowing air
pressure delay time (Td) experiences the further
stretching by rod and will have a more uniform
final overall wall thickness. In addition, a longer
stretching time can result in a significant increase
in tensile stress in the stretched preform and
occasional tearing defect. Figure 14 shows an
equivalent von Mises stress distribution on the
stretched preform. Figure 15 shows bottle wall
thickness for three different pre-blowing delay
times (different stretching time). In this figure,
more pre-blowing delay times will result in less
thickness variation. This is highly desirable to
achieve a uniform bottle thickness but increases
the probability of tearing at the same time.
It is clear from Figs 14 and 15 that the last case with
higher pre-blowing delay time has less thickness
variation and a higher tendency towards tearing
defect because of higher stress distribution. There-
fore, it is very desirable to determine an optimum
pre-blowing stretching time delay.

7 CONCLUSION

An investigation of a fully coupled temperature–


displacement simulation of SBM of PET bottles has
been presented using ABAQUS code. Extensive
FEAs have been carried out to predict the defor-
mation and the distribution of strain, temperature,
and thickness. Comparisons of numerical results
with the experimental observations demonstrate a
close agreement in the trend of thickness distribu-
tion in lower-middle regions of the bottle. The
difference in material movement deformation can
be attributed to the viscous behaviour of PET. The
result of strain and stress distributions showed that
the most critical area is the region near the bottle
bottom for a higher tendency towards tearing
defects. A parameter study on the effective para-
meters of the process showed that the conductance
heat transfer coefficient, friction condition, and
time delay of pre-blowing time delay effect the
final bottle thickness and other properties. It was
found that higher values of sticking friction and
conductive heat transfer coefficients increase the
thickness of the bottle bottom. More pre-blowing
time delay results in less thickness variation. It is
Fig. 13 Three deformation steps of SBM process with: concluded that the last case with higher pre-blow-
(a) Td ¼ 0.15 s; (b) Td ¼ 0.18 s; (c) Td ¼ 0.20 s ing delay time (Td ¼ 0.2 s) has less thickness

JEM1853 Proc. IMechE Vol. 224 Part B: J. Engineering Manufacture


1226 S Bagherzadeh, F R Biglari, and K Nikbin

Fig. 14 Von Mises stress distribution for three different delay times of air blowing on stretched preform

REFERENCES

1 Garbarski, J. The application of an exponential-type


function for the modeling of viscoelasticity of solid
polymers. J. Polymer Engng Sci., 1992, 32, 107–114.
2 Krishnaswamy, P., Tuttle, M. E., Emery, A. F., and
Ahmad, J. Finite element modeling of the time-
dependent behavior of nonlinear ductile polymers.
Polymer Engng Sci., 1992, 32(16), 1086–1096.
3 Schmidt, F. M., Agassant, J. F., Bellet, M., and Desoutter,
L. Viscoelastic simulation of PET stretch/blow moulding
process. J. Non-Newtonian Fluid Mechanic, 1996, 64,
19–42.
4 Kouba, K., Baetos, O., and Valchopoulos, L. Computer
simulation of thermoforming in complex shapes.
Polymer Engng Sci., 2004, 32, 699–704.
Fig. 15 Final bottle wall thickness comparison at three 5 Christensen, R. M. Theory of viscoelasticity, an intro-
various pre-blowing air pressure entrance times duction, 1982 (Academic Press, Boston).
6 Yang, L. M., Shim, V. P. W., and Lim, C. T. A visco-
variation and a higher tendency towards tearing hyperelastic approach to modelling the constitutive
defect because of higher stress distribution. There- behaviour of rubber. Int. J. Impact Engng, 2000, 24, 545–
fore, an optimum blowing process is expected to be 560.
7 Pham, X. T. Modelling and characterization of PET.
achieved with a pre-blowing delay time between
International Report, Industrial Material Institute/NRC,
Td ¼ 0.18 and 0.2 s. 2002.
8 Haberstroh, E. and Detrois, C. Capabilities and
limitations of injection blow moulding simulation, 1999
€ r Kunststoff verarbeitung (IKV), RWTH
(Institut fu
 Authors 2010 Aachen).
9 Denardin, E. L. G., Tokumoto, S., and Samios, D.
Stress–strain behavior of poly (ethylene terephthalate)

Proc. IMechE Vol. 224 Part B: J. Engineering Manufacture JEM1853


Parameter study of stretch–blow moulding process 1227

(PET) during large plastic deformation by plane strain APPENDIX


compression: The relation between stress–strain curve
and thermal history, temperature and strain rate. Rheol. Notation
Acta, 2005, 45, 142–150.
10 McEvoy, J. P., Armstrong, C. G., and Crawford, R. J. C specific heat factor
Simulation of the stretch blow moulding process of Cij constants of material
PET bottles. Adv. Polymer Technol., 1998, 17, 339– E elastic modulus
352. F preform stretch force
11 Menary, G. H., Armstrong, C. G., Crawford, R. J., and
ha film coefficient (preform-air)
McEvoy, J. P. Modeling of poly(ethylene terephthalate)
hm gap conductance coefficient
in injection stretch–blow moulding. Plastic Rubber
I1 first invariant of strain tensor
Composite, 2000, 29, 360–370.
12 McEvoy, J. P., Armstrong, C. G., and Crawford, R. J.
K conductivity
Simulation of the stretch blow molding process of PET l preform segment length
bottles. Adv. Polymer Technol., 1998, 17, 339–352. m sticking friction coefficient
13 Schmidt, F. M., Agassant, J. F., and Bellet, M. Experi- P air pressure on preform
mental study and numerical simulation of the injection qa heat flux on preform inner surface
stretch/blow molding process. Polymer Engng Sci., qm Heat flux per unit area
1998, 38, 1399–1412. Rm max. internal radius along the preform
14 Pham, X. T., Thibault, F., and Lim, L. T. Modeling W specific strain energy
and simulation of stretch blow molding of polyethylene
terephthalate. Polymer Engng Sci., 2004, 44, 1460– a coefficient of expansion
1472. d instantaneous thickness
15 Yeoh, O. H. Some forms of the strain energy function u preform temperature
for rubber. Rubber Chem. Tech., 1993, 66, 745–771. ua high-pressure air temperature
16 ABAQUS Theory Manual, user help documentation, um mould temperature
Version 6.7.
l principle stretch
17 Huang, H. X., Yin, Z. S., and Liu, J. H. Visualization
study and analysis on preform growth in polyethylene
r density
terephthalate stretch blow molding. J. Appl. Polymer sh hoop stress acting on the preform
Sci., 2007, 103, 564–573. sl longitudinal stress acting on the preform

JEM1853 Proc. IMechE Vol. 224 Part B: J. Engineering Manufacture

View publication stats

You might also like