Fg. 22.
Comparison
of
the
contact
times.
0.5
0.48
0.46
0.44
0.42
Y
0.4
0.38
0.36
0.34
0.32
0.3
0.28
h
Experimerital
Study
and
Numerical
Simulation
of
the
Injection Stretch/Blow
Molding
Process
Inner surface Outer surface
1
2
3
45
6
7
8
9
No
Capt.
T
1'
the dissipation energy related to high deformations,
which increases the temperature locally. The compari-
son between the computed thickness distribution and
the experimental data at the end of the process is
shown
in
Flg.
19.
The agreement
is
better with
a
non-
isothermal model than with an isothermal one.
The temperature di:jtribution at the end of the pro-
cess
is
represented
in
Fig.
20
at
different locations
(bottom of the bottle, details of the mold, neck of the
bottle). The contact between the stretch rod and the
bottom of the preform
as
well
as
the contact between
the bottle mold and the outer surface of the preform
induces high temperature gradients throughout the
thickness (maximum
50°C),
which just@ the volumic
approach
in
order to obtain
an
accurate description of
the temperature distribution. More enlightening
is
to
plot the temperature distribution along three material
lines, which are located, respectively,
at
the neck of
the preform
(yA
=
290
mm), at the center of the pre-
form
(y,
=
250
mm),
and at the bottom of the preform
(x,
=
6.9
mm) (see
F'ig.
214.
For each material line,
the temperature distribution
is
represented vs. the
nondimensional cumlinear coordinate
s
in
the thick-
ness direction
(Rg.
21
b).
We
note that the temperature
increases from
the
upper material line
(A)
to
the lower
one (C). This
is
due
1.0
the expansion of the preform,
which occurs from the neck to the bottom of the bot-
tle.
So,
the contact between the preform and the mold
occurs later for line
C
than for line
A.
More interesting
is
that
the temperature distribution along
line
C
at
the end of the process is partly higher
than
the
initial
temperature (see dashed line), which
is
related to the
energy dissipation during inflation. This phenomenon
has been observed experimentally.
Contact times versus location of contact sensors for
the isothermal model are plotted in
Flg.
22
and com-
pared with experiments. Contact times computed
using Oldroyd
B
model are closer
than
the Newtonian
one. Consequently, it appears that the kinetic of con-
tact
(and consequently the kinetic of blowing) depends
on the rheological behavior.
6)
CONCLUSION
Experiments have been conducted on
a
well-instru-
mented stretch blow molding machine: stretching
forces
as
well
as
contact times between the polymer
and the mold have been recorded. The bottle thick-
ness distribution has been measured for various pro-
cessing
parameters.
A
coupled model for the thermodynamic of the
air
and for the thermomechanical idation of the parison
has been proposed.
A
finite element model and a
vis-
coelastic differential constitutive equation have been
used. Viscous dissipation,
as
well
as
the temperature
gradient between the mold and the molten polymer,
has been considered. The comparison between experi-
mental and numerical rod stretching forces
is
fair.
The discrepancy
is
more important when considering
the thickness distribution and the contact kinetic.
Further developments
will
necessitate
a
better
un-
derstanding of the rheology of
PET
during the stretch
blow molding operation (amorphous, i.e., liquid at the
beginning
of the process, and semicrystalline, i.e.,
solid
at
the end of the process).
ACKNOWLEDGMENT
This research was supported by Side1 Company and
the Rench 'Ministere de
la
recherche"
(MHT
no
9OA
136).
POLYMER
ENGINEERIA'G AND SCIENCE,
SEPTEMBER
1998,
Vol.
38,
No.
9
1411