Effects of Preform Deformation Behavior on the Properties

of the Poly(ethylene terephthalate) Bottles

B. Demirel,1 F. Daver2
1
Department of Materials Science and Engineering, Faculty of Engineering, Erciyes University,
Melikgazi 38039 Kayseri, Turkey
2
School of Aerospace, Mechanical and Manufacturing Engineering, RMIT University,
Bundoora, Victoria 3083, Australia

Received 11 December 2010; accepted 1 February 2012

DOI 10.1002/app.36925
Published online in Wiley Online Library (wileyonlinelibrary.com).

ABSTRACT: Plastics bottles made out of poly(ethylene tion to the activation of the blow pressure. The simulation
terephthalate) (PET) are usually produced by injection results obtained through the both simulation packages were
stretch blow molding. Optimization of the process parame- compared with experimental results. It was found that bot-
ters is necessary to achieve bottles with adequate top load tles of highest quality were produced if the sequencing of
and burst strength. However, doing so experimentally is axial stretching and radial inflation results in simultaneous
time-consuming and costly. To overcome this difficulty, biaxial deformation of the preform. Truly biaxial orientation
simulation packages based on finite element analysis meth- of PET molecules improved both top-load and burst
ods have been developed. In this study, process optimiza- resistances of the bottles. The structural simulation studies
tion of a 350-mL PET fruit juice bottle was carried out by performed by the ANYSYS simulation package validated
means of BlowView and ANSYS simulation packages. most of our experimental findings. V C 2012 Wiley Periodicals,

BlowView was used for the ISBM process simulation and Inc. J Appl Polym Sci 000: 000–000, 2012
ANSYS for structural analysis of the bottles. The bottles
were produced under different process conditions where Key words: simulation; stress; polyethylene terephthalate;
the timing of the stretch rod movement was varied in rela- injection molding; mechanical properties

INTRODUCTION Haddad et al. investigated the ISBM process for a

600-mL PET bottle utilizing B-SIM simulation soft-
Poly(ethylene terephthalate) (PET) is the material of
ware; for a given preform shape and temperature,
choice for fruit juice packaging due to its excellent
they obtained wall thickness and von Mises stress
clarity, good mechanical and barrier properties, and
distribution along the arc of the bottle. The simu-
ease of processing. PET bottles used for the fruit
lated bottle file was exported to the ANSYS finite
juice packaging are generally made by injection
element analysis software for structural analysis to
stretch blow molding (ISBM). An injection molded
verify the mechanical strength of the bottles. The
preform is deformed radially by the internal (blow)
ANSYS results were used to identify key areas of
pressure and axially by the stretch rod. The air pres-
structural weakness in relation to the thickness dis-
sure loading consists of two consecutive stages: pre-
tribution from B-SIM simulation.3 Yang et al. studied
blow and final blow. The preblow forms most parts
the processing of a 330-mL PET bottle using a finite
of the bottle with a low pressure, whereas the final
element model incorporating heat transfer between
blow exerts a higher pressure to form the intricate
the stretch rod, the preform, and the mold with a
details of the bottle.1 Both process parameters and
view to optimizing process conditions. Comparison
the preform design are known to affect bottle qual-
of the predicted wall thicknesses with the measured
ities such as top load and burst strength.2 The pro-
showed good agreement along most parts of the bot-
cess parameters comprise preform temperature, the
tle.4 In one of the earlier studies, an integrated simu-
timing of the stretching and blowing stages, stretch
lation of preform reheating, stretching and inflation
rod speed, and preblow and final-blow pressures.
was conducted in a two-stage ISBM process for the
There are numerous studies related to the effect
manufacture of 250-mL PET water bottle.5 Experi-
of process parameters on the bottle properties.3–7
mentally measured preform temperature profiles
were used in the simulations. The predicted bottle
wall thickness distribution was found to be within
Correspondence to: B. Demirel ([email protected]).
the limits of experimental scatter. McEvoy et al.
Journal of Applied Polymer Science, Vol. 000, 000–000 (2012) simulated the ISBM process for 1/3 and 2 L bottles
V
C 2012 Wiley Periodicals, Inc. based on production line setting.6 To evaluate the
2 DEMIREL AND DAVER

tudinal stresses within it, and the temperature distri-

bution along it. In our previous publications, we have
shown that the manufacturing of good quality PET
bottles depends on both bottle design and processing
parameters.8,9 In this study, we aimed to use an inte-
grative simulation method to investigate the effect
of stretch rod timing relative to blow pressure activa-
tion. To validate the prediction of the simulation, bot-
tle performance was compared with experimental
results. In the first stage of the work, BlowView
8.4 software was used to simulate the PET fruit juice
bottle production at four different processing condi-
tions; one being the standard conditions set by the
manufacturers of the ISBM machine. Then the ANSYS
software was used for structural analysis of the simu-
lated bottles. In the second stage of the work, bottles
were produced in the laboratory ISBM machine and
the performance of the bottles was assessed in terms
of top-load and burst resistances. Finally, the simula-
tion results obtained through the BlowView and the
ANSYS simulation packages were compared with the
results for the physical bottles.
Figure 1 The CAD of the bottle mold.
SIMULATION STUDY
flexibility of the complete finite element simulations, Material
variations of stretching times and speed were con-
sidered for a given preform and bottle. Thickness The PET resin used was a food grade (9921W) from
profiles of the bottles were compared favorably with Eastman Chemical Company (Tennessee) with an
the experimental results. In another study, the mecha- intrinsic viscosity of 0.80 dL/g and viscohyperelastic
nism of preform deformation for a 680-mL bottle was material model was used in simulation studies. The
observed by means of a transparent mold; the pre- model has been developed by Pham et al. to repre-
form growth was analyzed during stretching and sent the behavior of PET during the stretch-blow
blowing stage.7 The wall-thickness distributions of molding process.10
the bottles obtained under different preblow delay
time were compared. It was found that the delay time Bottle mold and preform design
of preblow is a key factor in the preform growth; the The bottle mold used in this study is a 350-mL PET
types of preform growths were found to depend on fruit juice bottle (Fig. 1). The preform design, which
the geometry and size of preform, the hoop and longi- is generated by BlowView version 8.4; and a typical

Figure 2 Preform profile (a) thickness and (b) temperature. [Color figure can be viewed in the online issue, which is
available at wileyonlinelibrary.com.]

Journal of Applied Polymer Science DOI 10.1002/app

PROPERTIES OF THE PET BOTTLES 3

Figure 3 Preform deformation steps—Model 1.

temperature profile of preform, which is obtained by 4. Standard Model: Preblow is applied shortly after
a thermal imaging camera (FLIR ThermaCAM PM the stretch rod touches the bottle base (Fig. 6).
595), are shown in Figure 2(a,b), respectively.
The process parameters used in the simulation
Process simulation and the experimental study are given in Table I. Pre-
form weight and preform temperature profile were
In this study, a total of three different processing kept the same for all the models used. The preblow
models were generated; each model has a different pressure and the final-blow pressure were set at 0.4
set of stretch rod movement and pressure profile as and 2 MPa, respectively, throughout the study. The
a function of time. Processing conditions suggested BlowView (version 8.4) simulation software was
by the manufacturers of the machine (AOKI, Japan) used to obtain thickness and stress distributions on
were also included under ‘‘standard model.’’ the bottles; whereas the ANSYS software was used
for the analysis of bottle top-load and burst pressure
1. Model 1: The stretch rod is engaged at the resistance.
same time with the preblow pressure (Fig. 3).
2. Model 2: The stretch rod moves only half way The BlowView and ANSYS simulation packages
down the bottle mold prior to the application
of the preblow (Fig. 4). The BlowView Plastic Blow molding software, which
3. Model 3: Preblow pressure is applied without has been developed at the Industrial Materials Insti-
any stretch rod movement generating free-blow tute-Canadian National Research Centre (IMI-CNRC),
conditions (Fig. 5). simulates and optimizes blow molding processes.11

Figure 4 Preform deformation steps—Model 2.

Journal of Applied Polymer Science DOI 10.1002/app

4 DEMIREL AND DAVER

Figure 5 Preform deformation steps—Model 3.

The simulation package helps to predict how the the unique concerns of pure structural simulations.
blow molded parts will preform before committing to It offers both linear and nonlinear structural capabil-
expensive tool manufacture. The software solves the ities. The software is capable of reading the output
nonisothermal solid mechanics constitutive equations files generated by BlowView simulation software for
specifically for the individual phases of the blow structural analysis purposes.12
molding process. It incorporates thermomechanical
material models: viscoelastic models for polyolefins
and viscohyperelastic material models for PET. The Simulation of top-load and burst pressure
resistances with ANSYS software
viscohyperelastic material model is based on the
assumption of an additive decomposition of the stress ISBM bottles are required to demonstrate high top-
into hyperelastic and viscous contributions. The load and burst pressure resistances. To compare the
model takes into account strain-rate and temperature effect of changes in the timing of the stretch rod
effects, effectively represents strain hardening proper- movement in relation to the blow pressure activation
ties of PET. Coefficients used to characterize the on the bottle properties, the processing of the bottles
hyperviscoelastic material properties are obtained by were simulated under different processing regimes
biaxial stretching tests performed on the Brucker by means of the BlowView software. The resultant
machine, available at the IMI-CNRC. simulation files were then transferred into the
The ANSYS simulation softwares have been devel- ANSYS software to analyze the burst pressure and
oped by ANSYS, a leading simulation software com- top-load resistance of the bottles. For the structural
pany. The ‘‘ANSYS Structural’’ module addresses analysis of the bottles, instead of using a constant

Figure 6 Preform deformation steps—standard model.

Journal of Applied Polymer Science DOI 10.1002/app

PROPERTIES OF THE PET BOTTLES 5

TABLE I TABLE III

Process Parameters used in the Simulation and Weight Distribution of the Bottle Sections
Experimental Study
Weight distribution (g)
Standard
Top Body Base
Model 1 Model 2 Model 3 model
Model 1 9.7 6 0.05 5.0 6 0.0 4.9 6 0.05
Preblow pressure 0.10 0.12 0.12 0.12
Model 2 9.5 6 0.05 5.0 6 0.0 5.1 6 0.05
start (s)
Model 3 9.4 6 0.05 5.0 6 0.0 5.2 6 0.05
Final blow pressure 0.18 0.14 0.12 0.12
Standard model 9.2 6 0.22 5.1 6 0.05 5.25 6 0.16
start (s)
Blow pressure 3.18 3.14 3.12 3.12
finish (s)
Stretch rod 0.10 0.10 3.00 0.30 EXPERIMENTAL STUDY
start (s)
Stretch rod 0.17 0.14 3.07 0.37 Injection stretch blow molding process parameters
finish (s)
AOKI SB III-100H-15 single-stage ISBM machine
was used for the bottle production in this experi-
mental study. The process parameters are provided
value to define the material properties, microstruc- in Table II.
ture-dependant mechanical performance models
were used to calculate typical mechanical properties
including Young’s Modulus, yield stress as shown Top-load strength
in our recent publication.13 Top-load strength assesses the overall durability of
In the top-load analysis, the constraints were the bottles necessary for filling and stacking the bot-
applied at the base section of the bottle. Loads vary- tles during manufacturing, storage and distribution.
ing from 200 to 400 N were then applied to the top The top-load strength tests were conducted using
section of the bottle. The maximum equivalent the INSTRON 4466 instrument equipped with a top-
stresses and the maximum deformation on the bottle load test platform. At least five bottles are tested to
surface were recorded. In the burst resistance analy- achieve an average value.
sis, the top section of the bottle was clamped, pres-
sures varying from 1.0 to 1.5 MPa were applied
Burst strength
inside the bottle. Maximum stress and deformation
values were then recorded. Burst strength, the pressure at which the bottle
bursts, provides an assessment of the overall stabil-
ity of the bottle under carbonation pressure of the
TABLE II content. It is particularly important in bottles
ISBM (AOKI SB III-100H-15) Process Parameters intended for carbonated beverages; to ensure bottles
Process parameters do not blow up at the filling stage and filled
Screw bottles do not expand excessively during storage
Diameter (mm) 38 and/or during bottle warming for pasteurization
Screw speed (rpm) 100 purposes. A Topwave Burst Tester (BR3000), with
Nozzle diameter (mm) 3 ‘‘Ramp Fill’’ capability, was used. At least five bot-
Hot runner block ( C)
tles were tested to achieve an average value.
Sprue 275
Block 275
Nozzle 295 Material distribution
Barrel temperature ( C)
Front 275 To measure the material distribution in the bottle,
Middle 275 the bottles were cut into three sections: ‘‘base,’’
Rear 270
Nozzle 275
‘‘body,’’ and ‘‘top’’ by a specially designed hot-wire
Injection pressure cutter (Fig. 1). The bottle sections were separately
Primary (MPa) 13.73 weighted on a precision scale and recorded for
Secondary (MPa) 5.88 assessment.
Injection speed (m/s) 200
Stretch blow molding
Mold temperature ( C) 15 Temperature profile on the preform
Preblow pressure (MPa) 0.4
Final blow pressure (MPa) 2 The thermal imaging camera (FLIR Systems
Machine oil temperature ( C) 50 ThermaCAM PM595)
Stretch rod speed (m/s) 1.0
The thermal imaging camera was used to record the
Stretch rod diameter (mm) 6
temperature profile of the preforms just before the

Journal of Applied Polymer Science DOI 10.1002/app

6 DEMIREL AND DAVER

TABLE IV TABLE VI
Experimental Burst Pressure Resistance and Volume Maximum Stress Values under the Burst Resistance
Expansion Values Simulations via ANSYS
Burst pressure Volume Internal Maximum stress (N/mm2)
(bars) expansion (%) pressure
(MPa) Model 1 Model 2 Model 3 Standard
Model 1 12.5 6 0.2 91.1 6 13
Model 2 5.2 6 2 4 6 0.9 0.8 134.07 287.36 135.01 124.55
Model 3 12.6 6 0.68 82.5 6 25.9 0.9 150.83 323.28 151.89 140.13
Standard model 13.2 6 0.27 99.8 6 8.0 1 167.59 359.2 168.77 155.71
1.1 184.35 395.12 185.65 171.29
1.2 201.11 431.04 202.52 186.86
1.3 217.86 466.96 219.4 202.43
preform is stretched and blown into the bottle mold.
The preform temperature profile was then input to
the BlowView simulation studies. The highest burst resistance value was achieved in
the standard model and the lowest one in Model 2
RESULTS AND DISCUSSION (Table IV). However, the burst resistance values for
the Model 1, the Model 3, and the standard model
Following the bottle production according to the are found to be very close to each other.
three different models and the standard model, the The top-load resistance of the bottles shows the
bottles were weighed to assess the material distribu- same trend as in burst resistance (Table V). The
tion under different processing conditions. Table III maximum top-load resistance value was reached for
shows the weight of top, body, and base of the bot- the bottles produced by standard model, closely
tles for all models. followed by the bottles produced according to the
In Model 1, stretch rod is activated at the same Model 3 and the Model 1. The high biaxial orienta-
time with the preblow pressure and the stretch rod tion of PET molecules would be achieved under
reaches all the way to the bottle base. During simultaneous axial and radial deformation (standard
stretching, the preform end-cap moves away from model) and under free blow conditions (Model 3).
the neck and stretches further, hence the preform That is most likely to be the reason for high mechan-
end-cap thins out, resulting in lower base weight. In ical performance of the bottles produced according
Model 2, stretch rod moves only half-way down the to standard model and Model 3.
bottle mold, much less axial stretching occurs before The bottles were produced by the BlowView 8.4
the activation of the blow. Hence, the base is heavier simulation software according to the three models
compared to the Model 1. In Model 3, the stretch proposed and the standard model currently in use.
rod is not used; free blow condition becomes preva- These simulated bottles were then transferred into
lent; this improves material distribution across the ANSYS software to carry out top-load and burst
bottle sections due to truly biaxial deformation of resistance tests. The burst resistance and top-load
the preform. In the standard model, simultaneous values obtained by ANSYS are given in Tables VI
axial and radial deformation of the preform is and VII, respectively.
observed as shown in Figure 6. The biaxial orienta- According to the burst resistance results obtained
tion of the preform results in a more even material through ANSYS simulation software, bottles pro-
distribution across all sections. When the PET mate- duced according to the Model 1 and the standard
rial is subjected to sequential orientation, stress lev- model were found to be stronger than the Model 2
els are higher compared to biaxial orientation and Model 3 bottles (Table VI). In this context, the
mode.14 Hence, under biaxial deformation mode, the burst pressure values obtained by means of ANSYS
preform deformation speed would be higher, simulation are in harmony with the actual burst
improving the overall strength of the bottles.

TABLE V TABLE VII

Experimental Top-Load Resistance and Compressive Maximum Stress Values under the Top-Load
Strength Values Simulations via ANSYS

Max compressive Extension Compressive Maximum stress (MPa)

load (N) (mm) strength (MPa) Load (N) Model 1 Model 2 Model 3 Standard
Model 1 321 6 30 5.26 6 0.54 5.43 6 0.51 200 4.98 15.71 16.7 5.52
Model 2 205 6 27 2.54 6 0.32 3.47 6 0.45 250 6.24 19.67 20.89 6.91
Model 3 374 6 11 6.03 6 0.31 6.3 6 0.19 300 7.49 23.63 25.08 8.3
Standard 394.54 6 31 6.2 6 0.09 6.67 6 0.52 350 8.73 27.53 29.27 9.67
model 400 9.98 31.49 33.46 11.06

Journal of Applied Polymer Science DOI 10.1002/app

PROPERTIES OF THE PET BOTTLES 7

pressure values obtained by the experimental study and radial inflation introduced simultaneous biaxial
(Table IV). deformation of the preform as demonstrated by the
Top-load resistance simulations carried out via simulation studies. The truly biaxial orientation of
ANSYS are shown in Table VII. While standard PET molecules improved the overall strength of the
model and Model 1 resulted in strong bottles, Model bottles. The structural simulation studies performed
2 and Model 3 bottles were found to be inferior in by ANYSYS simulation package validated most of
terms of their top-load resistance. The simulation our experimental findings.
results from ANSYS are not in harmony with the
experimental results for all models (Table V). In References
particular, Model 3 bottles showed high top-load
1. Rosato, D. V.; Rosato, D. V., Eds. Blow Moulding Handbook:
resistance experimentally. However, the simulation Technology, Performance, Markets, Economics—The Complete
via ANSYS predicts a very low top-load resistance. Blow Molding Operation; Hanser: Munich, 1989.
We were not able to validate our experimental find- 2. Brooks, D. W.; Giles, G. A., Eds. PET Packaging Technology;
ing for the top-load strength of the bottles produced Wiley-Blackwell Publishing, 2002.
under Model 3. 3. Haddad, H.; Masood, S.; Erbulut, D. U. Aust J Mech Eng
2009, 7, 69.
4. Yang, Z. J.; Harkin-Jones, E.; Menary, G. H.; Armstrong, C. G.
CONCLUSIONS Polym Eng Sci 2004, 44, 7.
5. Martin, L.; Stracovsky, D.; Laroche, D.; Bardetti, A.; Ben-
In this study, we used an integrative simulation Yedder, R.; Diraddo, R. ANTEC Tech Pap 1999, 982.
6. McEvoy, J. P.; Armstrong, C. G.; Crawford, R. J. Adv Polym
method to investigate how the timing of the stretch
Tech 1998, 17, 4, 339.
rod in relation to blow pressure activation affects the 7. Huang, H. X.; Yin, Z. S.; Liu, J. H. J Appl Polym Sci 2007, 103,
bottle properties and also aimed to validate our pre- 564.
dictions via experimental work. A total of four dif- 8. Demirel, B.; Daver, F. J Appl Polym Sci 2009, 114, 1126.
ferent processing models were considered. Preform 9. Demirel, B.; Daver, F. J Appl Polym Sci 2009, 114, 3811.
10. Pham, X. T.; Thibault, F.; Lim, L. T. Polym Eng Sci 2004, 44,
weight and preform temperature profile were kept
1464.
constant for all processing models. Each model had 11. Blowview Software. Available at: http://www.pacesimulations.
a different stretch rod and blow pressure activation com. Accessed on February 11, 2011.
timing, resulting in different preform deformation 12. ANSYS Software. Available at: http://www.ansys.com/.
behavior. Accessed on 11 Feb 2011.
We found that the 350-mL PET fruit juice bottle 13. Daver, F.; Demirel, B.; Sutanto, J.; Pang, C. W. J Appl Polym
Sci, DOI: 10.1002/app.36780.
produced by the standard model has achieved the 14. Martin, P. J.; Tan, C. W.; Tshai, K. Y.; McCool, R.; Menary, G.;
best qualities in terms of top-load and burst resistan- Armstrong, C. G.; Harkin-Jones, E. M. A. Plast Rubber
ces. In this model, the sequencing of axial stretching Compos 2005, 34, 276.

Journal of Applied Polymer Science DOI 10.1002/app