Corresponding author. Tel.: +34 96 590 3789; fax: +34 96 590 3826.
E-mail address: antonio.marcilla@ua.es (A. Marcilla).
2. After a short delay period, compressed nitrogen gas cores
out the molten polymer. The penetrating gas leaves behind
a polymer layer at the mold walls, attaining a molded part
with a polymer skin and an inner gas channel (Fig. 1b). Gas
penetration which takes place at this step is denominated
primary gas penetration. This step nishes when the mold
cavity has been entirely lled (Fig. 1c).
3. Gas is continuously injected to transmit the packing pres-
sure to the polymer. At this stage the polymer shrinkage is
counteracted by a growth of the gas core. The gas packing
pressure remains until all polymer material has solidied.
Gas penetration during post-lling step is termed secondary
gas penetration (Fig. 1d).
In full-shot GAIM the rst step of the short-shot process is
extended until the shot weight of molten polymer lls entirely
the mold cavity. The effects of the processing conditions on
the behaviour of gas penetration in full-shot GAIM have been
investigated by other researchers [3,4].
Back-to-screw process is similar to the full-shot process.
The difference is that plastic is pushed back into the injection
unit (in front of the screw). The last GAIM process, mold with
retractable cores, is based in the variation of the volume of the
mold cavity which is accomplished using a core. The initial vol-
ume of the mold is the minimum and when the gas is injected,
simultaneously, the core is pulled back to increase the cavity
volume.
0924-0136/$ see front matter 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.jmatprotec.2006.04.095
A. Marcilla et al. / Journal of Materials Processing Technology 178 (2006) 350357 351
Fig. 1. Description of the steps of the short-shot process (standard GAIM): (a)
polymer short-shot; (b) start of primary gas penetration; (c) end of primary gas
penetration; (d) secondary gas penetration.
GAIM offers a considerable number of advantages over con-
ventional injection molding. These advantages mainly originate
from the near constant value of the gas pressure throughout the
gas core as a result of the pressure drop in the gas core is insignif-
icant compared with the pressure drop in an equivalent molten
polymer [5]. The advantages over conventional injection mold-
ing are reduced part weight, injection pressure, clamp force,
shrinkage, warpage and residual stress, apart from better sur-
face nish [6]. Moreover GAIM enhanced design possibilities
and thick-sectioned components are possible without excessive
warpage, shrinkage and cooling time [7].
All these advantages are exploited only if the design and
processing parameters are perfectly understood. GAIM appli-
cation is more critical than conventional injection molding
due to the additional parameters which are introduced, such
as the numbers and location of gas injection points, the
amount of molten polymer injected (shot size), the gas delay
time, the injected gas pressure and the holding time for gas
injection.
Optimization of molds and products by trial-and-error meth-
ods, i.e., by modifying the mold and changing the process
parameters in a series of test moldings, is both expensive and
time consuming. With the advent of scientic-based computer-
aided engineering (CAE) simulation tools, engineers can eval-
uate alternative design routes and materials without physically
committing real material and machine time. They enable the
optimization of the injection molding process allowing the pre-
diction of the gas distribution in the molded part and the resulting
part properties [79]. In this way, the simulation of GAIM can
help elucidate the characteristics of the process and extend the
molding experience.
To perform a simulation, a CAE technique must be adopted
which involves the creation of a suitable model of the mold
cavity. This model should be optimized to apply on them the
numerical methods which are needed to carry out the sim-
ulation, i.e., nite difference method (FDM), nite element
method (FEM), boundary element method (BEM) and nite
volume method (FVM) [10]. These numerical methods require
discretization of the physical domain (geometric model) into
nite sub-domains (elements) obtaining a mesh model of the
mold cavity. Through the discretization, the continuous govern-
ing equations are converted into a set of discretize algebraic
equations to be solved by the use of the computer. There are
three main different types of mesh models based on the type
of the element used to construct them: one-dimensional tubu-
lar elements (strip mesh), two-dimensional triangular elements
(mid-plane mesh) and tetrahedral elements (three-dimensional
mesh). According to the model used, different analysis can be
performed: 2, 2.5 and 3-dimensional simulations, respectively
[11,12]. Obviously, the higher the complexity of the element
used the higher the accuracy of the simulation, albeit computa-
tion time is considerably increased.
The major benet of a two-dimensional analysis is speed
although their results are not equivalent to those produced with
more advanced analysis. In the case of GAIM simulation, strip
models lack the ability to predict an unsymmetrical polymer
wall distribution at any cross-section of the part.
Most of commercial injection molding simulation packages
are based on the so-called thin-lm or HeleShaw approxima-
tion developed by Hieber and Shen [13], in which the fact that
most injection parts are thin-walled is used to decouple owand
thermal effects in mid-plane and thickness directions; this has
led to what is known as the 2.5-dimensional approach [14]. As
a mid-plane model is generated by a surface, which is a two-
dimensional geometry entity, to attain a representation of the
three-dimensional geometry of the part, a numerical attribute
called thickness is assigned to all the at elements that compose
the mid-plane mesh. Many authors have investigated the relia-
bility of these commercial packages regarding to the simulation
of conventional injection molding [1520] and also concern-
ing GAIM for parts designed with gas channels [7,21,22]. This
approach has become the state-of-the-art for the simulation of
injection molding processes.
Some programs provide three-dimensional simulation of the
polymer GAIM process. However the analysis time is very
long, even using high-speed parallel processing CPUs. The time-
consuming issue is still more important if we are dealing with
the mold design stage [23,24]. Moreover, nowadays, these pro-
grams are not able to perform all the analysis that can be carried
out with a mid-plane model such as, cooling, warpage and stress
analysis. In accordance with this, in this work, it has been created
a mid-plane representationof the moldcavityinvolvinga consid-
erable geometric simplication due to the mid-plane mesh that
has to be created either from scratch or degenerated (collapsed)
from an existing three-dimensional CAD design model. Some
commercial simulation programs are able to deal automatically
with this task. Nevertheless, when for some areas of the part,
a unique mid-plane representation does not exist as occurs in
the part studied, the mid-plane creation must be accomplished
manually [25].
352 A. Marcilla et al. / Journal of Materials Processing Technology 178 (2006) 350357
Table 1
Properties of polypropylene ISPLEN PB 130 G1M
Properties (unit) Method Value
Physical
Melt ow rate (230
C) (kJ/m
2
) ISO 180 15
Izod unnotched impact strength (20
C) (kJ/m
2
) ISO 180 70
Thermal
Vicat softening temperature (
C) 235
Mold surface temperature (
C) 40
Injection speed (cm
3
/s) 90
Gas delay time (s) 0
Gas pressure (bar) 150
Gas holding time (s) 18
Cooling time (s) 15
Inlet coolant temperature (
C) 20
with a diameter of 14 mm for the two internal cross-beams (labelled 2 and 3).
The total dimensions of the part are 200 mm128 mm60 mm.
2.3. Injection molding machine and gas injection unit
The experiment was conducted using a 1300 kN Battenfeld TM1300/750
Unilog B4 injection molding machine equipped with a Airmould
-Modular
systemwhich is based in a pressure-controlled process and comprises four mod-
ules: gas supply (by nitrogen bottles), compressor, gas pressure control and gas
injection. The injection molding machine has a screw diameter of 50 mm which
provides a theoretical shot volume of 442 cm
3
(polystyrene) with a maximum
specic pressure of 1714 bar.
2.4. Process parameters
The xed process parameters are shown in Table 2. They were chosen based
on material and equipment suppliers recommended processing ranges. The
quantity of molten polymer injected just before gas injection started was opti-
mized as explained below. Using the optimum conditions 50 parts were molded
in order to achieve the stabilization of the characteristics in the molded parts
(weight was controlled). Finally, 20 parts more were produced and labelled to
be studied.
The polymer is injected into the mold cavity through a sprue gate and a single
gas needle was placed in the runner system in a special antechamber behind the
gate and the cavity (see Fig. 3). If the gas needle is placed directly into the
cavity the polymer ow will be divided by the gas needle and weld lines will be
generated provoking a decient propagation of the gas bubble [26].
Fig. 2. Part geometry (indication of the cross-sections used to compare the residual wall thickness).
A. Marcilla et al. / Journal of Materials Processing Technology 178 (2006) 350357 353
Fig. 3. Feed system of the mold with locations for polymer and gas injection.
2.5. Characterization of the molded parts by GAIM
To determine the accuracy of the simulation outcomes the gas bubble pen-
etration in the experimental parts has been studied. Furthermore, it has been
measured the residual wall thickness, which is dened as the thickness of the
polymer layer that is left behind at the mold walls after the gas front has passed.
According to Poslinski et al. [27] the residual wall thickness in GAIM is deter-
mined by two phenomena: the penetration of a gas into a viscous liquid, and the
growth of the solid layer.
To evaluate the two characteristics above mentioned the molded parts were
sectioned by parallel planes to its symmetry plane at 20 mmintervals throughout
the complete length of the part (200 mm).
2.6. Computer simulation
Molded parts by GAIM were simulated using the commercial software,
Moldow Plastics Insight (release 4.1). The rst stage of any simulation by the
nite element method (FEM) is called pre-process. This is the most time-
consuming step and consists of creating the geometric model of the mold cavity
to be analyzed. This can be performed by using either the simulation software
itself or one of the computer-aided design (CAD) computer programs. In this
work, an external CAD software, called GiD