RSC Advances

View Article Online

PAPER View Journal | View Issue

Bubble morphological evolution and surface defect

Published on 18 August 2015. Downloaded by University of Wollongong on 25/04/2016 09:21:01.

formation mechanism in the microcellular foam

Cite this: RSC Adv., 2015, 5, 70032
injection molding process
Lei Zhang, Guoqun Zhao,* Guiwei Dong, Shuai Li and Guilong Wang

Received 25th April 2015
Accepted 30th July 2015
DOI: 10.1039/c5ra07512b
www.rsc.org/advances
The filling stage of the Microcellular Foam Injection Molding (MFIM) process is a three phase flow process of
polymer melt, super critical fluid (SCF) and air. It not only includes the nucleation and growth of spherical
bubbles, but also the morphological evolution of the bubbles such as deformation, bursting, and vanishing.
There are usually silver marks, spiral lines, pits and other defects on the product surface. In order to
effectively control the surface quality, it is significant to reveal the morphological evolution law of
bubbles and the formation mechanism of surface defects in the filling stage of MFIM. This paper
established an incompressible, non-isothermal, and unsteady three-dimensional mathematical model of
multiphase flow. A new setting method of the boundary conditions with the exhaust function on the
mold cavity walls was proposed. The problem of temperature solution divergence on the interface
between the two phases with a high viscosity ratio was solved through the coupling algorithm of energy
equation and PIMPLE loop. The tracking accuracy of micron grade bubbles interface in macroscopic
scale flow field was improved though the Adaptive Meshing Refining (AMR) technique. Based on the
abovementioned mathematical model, the influence law of the temperature field and velocity field on
the bubble morphological evolution in the thickness cross-section of the injection flow field was
obtained. The deformation, bursting and vanishing process of bubbles with different initial sizes and
locations in the shear and fountain flow field was predicted. Combined with a short shot experiment, the
formation mechanism of pits, silver marks and collapses on the product surface manufactured by MFIM
was also revealed.

which the cooling contraction of a polymer melt provides a

1 Introduction
The Microcellular Foam Injection Molding (MFIM) process is a
polymer molding process, producing plastic parts with micron
level bubbles inside, in which a thermoplastic polymer is used
as the matrix and Super Critical Fluid (SCF) as the physical
foaming agent.1–3 This technology conforms to the development
trend of light-weight and environment protection. It has supe-
riority in material saving, energy conservation, dimensional
accuracy of product, short molding cycle and so on. Consider-
able attention has been paid to the MFIM process in automo-
bile, electric appliance, communication equipment, biological
tissue engineering4,5 and other elds.
Compared to the batch foaming process, the bubble evolu-
tion process of MFIM is more complicated. From previous
research,6–8 it was found that there are two foaming stages in
MFIM. One stage occurs in the lling process, in which the
pressure drop between the runner and cavity causes polymer
melt foaming. The other stage occurs in the cooling process, in
Key Laboratory for Liquid-Solid Structural Evolution and Processing of Materials
(Ministry of Education), Shandong University, Jinan, Shandong 250061, PR China.
E-mail: zhaogq@sdu.edu.cn; Fax: +86-531-88392811; Tel: +86-531-88393238
pressure drop for the nucleation and growth of a foaming agent.
The rst stage belongs to a dynamic foaming process, whereas
the second stage belongs to a static foaming process. These two
foaming stages have different foaming degrees and bubble
morphologies for different process parameters, product struc-
tures and polymer materials. As a result, versatile pore struc-
tures will appear inside the MFIM products and complex
morphology on the surface. In this way, the process control of
MFIM becomes complex and difficult. This two-stage foaming
process oen causes serious silver marks and swirl marks on
the surface of the MFIM products. Because these defects have
not been solved effectively until now, the popularization and
development of MFIM technology are restricted.9
To eliminate the abovementioned defects on the surface of
the MFIM product, it is necessary to ascertain the morpholog-
ical evolution law of micron grade bubbles and reveal the
forming mechanism of various defects. At present, many
hypotheses about the forming mechanism of swirl marks and
silver marks have been proposed. Cha and Yoon10 considered
that swirl marks were caused by the gas trapped between the
mold surface and polymer melt. Pierick et al.11 suggested that
the silver marks were caused by the bubbles at the melt surface

70032 | RSC Adv., 2015, 5, 70032–70050 This journal is © The Royal Society of Chemistry 2015

Paper
dragging against a mold wall. According to Andreas et al.,12 due
to a sudden pressure drop as shooting polymer melt into the
mold cavity, the polymer melt at the ow front becomes over-
saturated with dissolved gas, and consequently bubbles are
nucleated at the ow front. The nucleated bubbles at ow front
are moved to the interface between the polymer melt and cavity
wall and show pronounced silver and swirl marks on the surface
of the nal part. Chen et al.13,14 also believed that the appear-
ance of silver marks can be attributed to the foaming bubbles
Published on 18 August 2015. Downloaded by University of Wollongong on 25/04/2016 09:21:01.
that are brought to the surface via fountain ow. From the
abovementioned research, it is concluded that the formation of
swirl marks and silver marks on the product surface is closely
related to the lling ow eld of MFIM. Nevertheless, the
abovementioned hypotheses have not theoretically revealed the
formation mechanism of surface defects under the action of a
polymer melt ow eld. Unfortunately, the existing bubble
nucleation theory15–17 and growth theory18–23 also cannot
account for the evolution law of non-spherical bubbles.
Currently, visualization experiments are mainly used to
study the bubble morphological evolution law in a polymer melt
ow uid. Villamizar and Han24,25 observed that the deforma-
tion phenomenon of bubbles was affected by the surrounding
uid ow by a visual mold with windows on both sides. Mah-
moodi et al.26 found that the bubble migration velocity can
transcend the movement velocity of the polymer front, and
observed the cracking phenomena of bubbles. Wong and Park27
built a set of experiment simulation devices to observe the
nucleation, growth and morphological evolution of bubbles in a
shear ow. Ahmadzai et al.28 studied the inuence of mold
structure and process parameters on bubble rupture in the
injection lling stage. With the help of the abovementioned
mold visualization technology, the evolution process of bubbles
in MFIM can be observed directly. This has played an important
role in understanding the formation of the internal structure
and surface morphology of MFIM parts. However, mold visu-
alization technology has some obvious deciencies. As we all
know, the size of the bubble in a MFIM part reaches the micron
level, and the distance of bubble migration belongs to the
centimeter level. The scale difference makes it difficult for mold
visualization technology to simultaneously complete the clear
capture of the bubble shape and track the entire process of
bubble displacement. Furthermore, due to the constraints of
transparency of a polymer material and the shape of the visual
mold and the size of the observing window, it is difficult to
acquire the entire morphological evolution process in a MFIM
using mold visualization technology.
Compared to the abovementioned mold visualization
method, the numerical simulation provides an effective method
to study the morphological evolution of bubbles in MFIM,
which is not restricted by the polymer material or mold shape
and can simultaneously capture the bubble shape and track the
entire process of bubble displacement. To date, several
interface-tracking methods (e.g., level set method,29,34 diffusion-
interface model,31 volume of uid method,35,36 and lattice
Boltzmann method37) and discrete methods (nite element
method,29 nite volume method,30,32 nite difference method,32
and boundary integral method30,33) can be used to describe the
This journal is © The Royal Society of Chemistry 2015
View Article Online
RSC Advances
morphological evolution of the interface between two types of
uid. With these numerical simulation methods, a wide variety
of mathematical models have been established to simulate the
deformation of a bubble29,32,36,37 or droplet30–35 in simple shear
ow. Most of the simulation results29–31,33,35–37 are in agreement
with the experimental observations. However, the actual melt
ow eld in the injection molding process is not a simple shear
ow eld, but a complex ow eld coupled by the nonlinear
shear ow, fountain ow and complex 3D ow eld geometry.
Furthermore, the aforementioned studies did not consider the
inuence of an inhomogeneous temperature eld and shear
thinning phenomenon on the uid viscosity. To date, the
mathematical modeling and numerical simulation of the
bubble morphological evolution in a MFIM ow eld have not
been carried out.
To perform a numerical simulation of the bubble morpho-
logical evolution in the lling process of MFIM, this study
established an incompressible, unsteady, non-isothermal 3D
mathematical model of multiphase ow in the MFIM process,
wherein the interfacial tension model of SCF/polymer proposed
by Park et al.38 and SCF-concentration-updated Cross–Carreau
model proposed by Lee and Park39 were used to calculate
interface tension and melt viscosity. The Navier–Stokes equa-
tion with the inertia item was used to solve the velocity eld and
pressure eld, and the VOF with a compression item was
adopted to track the interface between two arbitrary phases. To
solve the abovementioned model accurately, a new set of
boundary conditions of gas ltration on the mold cavity wall
was proposed. To solve the divergence problem of temperature
eld on the interface between two phases with a large viscosity
ratio, the energy equation was introduced into the PIMPLE
algorithm, which is usually used to solve the velocity and
pressure eld, and the coupling algorithm of velocity, pressure
and temperature was nally established. The accurate tracking
of a micron level bubble interface in the large scale ow eld
was realized by adopting the Adaptive Mesh Rening (AMR)
technology40 based on eld variables distribution. Based on the
abovementioned established model, this study obtained the
inuence law of temperature eld and velocity eld on the
bubble morphological evolution in the thickness cross-section
of the melt ow eld of the MFIM process. The deformation,
bursting and vanishing of bubbles with different initial sizes or
locations in the shear and fountain ow eld were predicted.
The related short shot experiments of MFIM were performed.
The bubble morphologies and surface defects were observed by
SEM. Based on the numerical and experimental results, the
formation mechanism of pits, silver marks and collapses on the
product surface molded by MFIM was nally revealed.
2 Mathematical model and
calculation method
2.1 Governing equations
The mathematical model in this study was established to
research the bubble deformation stage aer the nucleation and
growth stage in the MFIM process. Considering the practical
RSC Adv., 2015, 5, 70032–70050 | 70033

RSC Advances
uid state and calculation amount, we assume that (1) polymer
melt, air and SCF are all incompressible materials, (2) there is
no chemical reaction and phase transition in the system, (3) the
ow type is laminar, and (4) the force of gravity is ignored.
According to the abovementioned assumption, the conservation
of mass, momentum and energy are given as follows:
1. Mass equation is expressed as
V$U ¼ 0
Published on 18 August 2015. Downloaded by University of Wollongong on 25/04/2016 09:21:01.
where U is the velocity eld and V$ is the divergence operator.
2. Momentum equation is expressed as
vðrUÞ
þ V$ðrUUÞ ¼ V$T þ rfb
vt
where r is the density, T is the stress tensor, fb is the interfacial
tension and t is time. According to the generalized Newton's
laws of internal friction, T can be expressed as


2 front-tracking method43 and the moving mesh method,40 this
T ¼  P þ mV$U I þ 2m3
3 method is still an appropriate interface tracking method for
1  complex 3D fountain ow eld of a mixed system with a large
3¼ VU þ ðVUÞT
2 viscosity ratio.44
where P is the pressure eld, m and 3 are the kinematic viscosity
and the strain rate tensor of the mixed system, respectively, and
I is the unit tensor. By substituting eqn (1) and (3) into (2), the
following momentum equation was obtained.
vðrUÞ  
þ V$ðrUUÞ ¼ Vp þ rfb þ V$m VU þ ðVUÞT
vt
3. Energy equation is expressed as
 
v rCp T  
þ V$ rUCp T ¼ V$ðkVTÞ þ mg_ 2
vt
where Cp and k are the isobaric heat capacity and the heat
conductivity of the mixed system, respectively, g_ is the modulus
of the strain rate tensor 3 and its expression is as follows:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
g_ ¼ j3ðUÞj ¼ 23ðUÞ : 3ðUÞ (7)
Physical property F in the interface between the arbitrary two
phases can be solved by the arithmetic mean of the physical
properties Fi of each phase and is expressed as
X
n
 
F¼ ai Fi ; F ¼ Cp ; k; r; m
1 k ¼ V$
where i means the ith uid, ai is the phase fraction of uid i in
the FVM mesh element of the multiphase interface, and n
means the number of phases in the mesh element.
2.2 Interface tracking method
To study the evolution of the interface in the mixed system
composed of polymer melt, SCF and air, this study indirectly
determined the transient location of the multi-phase interface
by the polymer melt phase fraction, apolymer, SCF phase fraction,
70034 | RSC Adv., 2015, 5, 70032–70050
View Article Online
Paper
aSCF, and air phase fraction, aair, in the mesh, wherein the
phase fraction of each phase was solved using the VOF method
with a compression item proposed by Weller41 as follows:
vai
þ V$ðUai Þ þ V$½Ur ai ð1  ai Þ ¼ 0 (9)
vt
where V$[Urai(1  ai)] is the compression item, and the phase
fraction ai of the uid i in a mesh element can be regarded as
(1) aði; x; y; z; tÞ ¼
8
< 1; for the point ðx; y; z; tÞ inside fluid i
0; for the point ðx; y; z; tÞ outside fluid i
:
0\a\1; for the point ðx; y; z; tÞ inside the interface area
(2)
This method is a tracking method dening the interface with
a limited thickness.42 Although an articial error is inevitably
introduced into the fusion and failure process of the interface,
in comparison to zero interface thickness methods, such as the
(3)
dealing with the bubble morphological evolution in the
(4)
2.3 Interfacial tension
In the mold cavity, there are three pairs of interfaces (polymer/
(5) SCF, polymer/air and SCF/air) between the different uids.
Among these interfaces, the interfacial tensions of polymer/air
and SCF/air are much smaller than that of polymer/SCF
because of the curvature difference. Furthermore, they will
not inuence the bubble morphology directly. Therefore, the
(6) interfacial tensions of polymer/air and SCF/air are ignored in
this paper. To accurately describe the bubble morphology in the
shear and fountain ow eld, the interfacial tension between
the SCF and polymer melt is translated into volume force form
using the following Continuum Surface Force (CSF) model
proposed by Brackbill et al.45
fb ¼ skVa (10)
where s is the surface tension coefficient between the polymer
melt and SCF and k is the curvature of the free interface and is
dened as
(8)

Va
(11)
jVaj
The abovementioned surface tension coefficient s is a
function of temperature and pressure. This can be solved by a
second order linear regression equation form38 expressed as
follow:
s ¼ D0  DTT  DPP + DTPTP (12)
where the equation coefficients D0, DT, DP, DTP are associated
with the material properties of polymer/SCF mixed system and
This journal is © The Royal Society of Chemistry 2015
 View Article Online

Paper RSC Advances

can be obtained using the pendant drop method in SCF with a
high temperature and pressure.
2.4 Viscosity model
In the mathematic model established in this study, the viscosity
of SCF and air is respectively assumed to be constant. The
viscosity of the polymer melt is calculated through the viscosity
model considering the SCF plasticizing effect on a polymer
melt46,47. At present, the plasticizing mechanism is explained
Published on 18 August 2015. Downloaded by University of Wollongong on 25/04/2016 09:21:01.
Fig. 1 shows a schematic of the boundary conditions for the
2D and 3D model of the MIFM lling process. Among these
boundary conditions, Gwall belongs to the no-slip type, Gprole
and Gvent belong to the hybrid type for realizing the exhaust
function of mold cavity. Table 1 lists the complete boundary
conditions in this study.
For the boundary condition of the cavity prole Gprole,
when apolymer < c1 X aSCF < c2 X aair > c3, the condition of
Gprole–air{T ¼ Tprole, VU ¼ 0, P ¼ P0} is used. When apolymer $ c1

W aSCF $ c2 W aair # c3, the condition of Gprole–polymer{T ¼

through the free volume theory of polymer.48 Lee and Park39 Tprole, U ¼ 0, VP ¼ 0} is used. For the boundary condition of
proposed the following Cross–Carreau viscosity model of a intensifying venting Gvent, when apolymer < C1 X aSCF < C2 X
homogeneous system of polystyrene/supercritical CO2 aair > C3, the condition of Gvent–air{T ¼ Tvent, VU ¼ 0, P ¼ P0} is
(PS/SCCO2) by introducing the concentration of SCCO2 and used, when apolymer $ C1 W aSCF $ C2 W aair # C3, the condition
pressure into the Arrhenius equation of the zero shear viscosity. of Gvent–polymer{T ¼ Tvent, U ¼ 0, VP ¼ 0} is used.
h0 c1, c2, c3, C1, C2, C3, in the abovementioned conditions are
h¼ (13)

a
ð1mÞ the judgment parameters. Considering the same ow behaviors
h0 g_ a
1þ of SCF and polymer melt on Gprole and Gvent, c1, c2, C1, C2 are
s
suggested to be dened in the range of 0.001–0.01 to effectively

 control the polymer overow to not more than 1% on the
b
h0 ¼ Aexp þ bP þ jC (14) boundary mesh. c3, C3 should be dened to ensure the free-
T  Tr
owing exhaust and avoid numerical instability. In this study,
where h0 and h denote the zero shear viscosity and dynamic 0.001, 0.001, 0.9, 0.01, 0.01 and 0.8 were assigned to c1, c2, c3,
viscosity, respectively. C is the mass concentration fraction of C1, C2, C3, respectively.
SCF. A, b, b, j, Tr, s, a, and m are the model parameters related Considering the coupling relationship of velocity and pres-
to the material, which can be obtained using online measure- sure in the momentum equation, this study dened the
ments in a foaming extruder with a Gas Counter Pressure (GCP) boundary condition of velocity and pressure of each physical
control equipment.39,49 boundary as different types such as the Dirichlet or Neumann
The kinematic viscosity, m, in the momentum equation can boundary condition.
be calculated by
h
m¼ (15)
r

2.5 Boundary condition

The solution of the abovementioned mathematical model
requires a set of appropriate boundary conditions. Hetu et al.50
proposed a type of hybrid exhaust boundary conditions based Fig. 1 Schematic of the boundary conditions of the 2D and 3D model
on the two phase ow model, in which the phase fraction of of the MIFM filling stage. (a) 2D model, (b) 3D model.
polymer was denoted by a with the value ranging from 0 to 1
and the critical phase fraction of polymer was denoted by ac.
When a $ ac, the hybrid boundary condition is expressed as a Table 1 Application method of the complete boundary conditions for
no-slip boundary condition. When a < ac, it is expressed as the the multiphase flow model of the MIFM filling process
exhaust boundary condition. The value of ac is the key to control
Boundary 2D model 3D model Boundary condition
the exhaust. However, it is difficult to ensure the cavity lling of
polymer melt and emptying of air with just one parameter ac. Ginlet Line CD Face efgh {T ¼ Tinlet, U ¼ Uinlet, VP ¼ 0}
Normally, the volume fraction of lost polymer in the boundary Gwall Line AC Face adeh {T ¼ Twall, U ¼ 0, VP ¼ 0}
mesh may be up to 20–50%. Line BD Face bcfg
Gprole Face abhg Gprole–air{T ¼ Tprole,
To solve the abovementioned problems, this paper proposed
VU ¼ 0, P ¼ P0}
a set of new hybrid exhaust boundary conditions of the multi- Face cdef Gprole–polymer{T ¼ Tprole, U ¼ 0,
phase model and its transformation criterion. Considering the VP ¼ 0}
practical structure characteristics of an injection mold, two Gvent Line AB Face abcd Gvent–air{T ¼ Tvent,
types of exhaust boundary conditions with different trans- VU ¼ 0, P ¼ P0}
Gvent–polymer{T ¼ Tvent, U ¼ 0,
formation criteria are put forward. One is the boundary condi-
VP ¼ 0}
tions of the cavity prole, and the other is the boundary Gempty Face ABCD Empty
conditions in the lling end of the cavity.

This journal is © The Royal Society of Chemistry 2015 RSC Adv., 2015, 5, 70032–70050 | 70035
 View Article Online

RSC Advances Paper

2.6 Calculation methods and procedures developed based on OpenFOAM,51 which is an open source
database of computational uid mechanics. The whole calcu-
In this study, coupled problem of the velocity eld and pressure
lation ow chart of this program is shown in Fig. 4, where
eld in the momentum equation was solved using the PIMPLE
Fig. 4(b) is the PIMPLE loop of the whole ow chart. The
algorithm.51,52 Unfortunately, this research involves the solution
calculation method and procedures of this program are stated
of the energy conservation equation on the interface between
as follows.
two phases with a large viscosity ratio. If the traditional PIMPLE
(1) Generate the initial mesh of the calculation model. Input
algorithm is used as the solution method, a divergence of
temperature eld with unreasonable uid ow is easily caused, the physical parameters, velocity, pressure, phase fraction,
as shown in Fig. 2(a). To solve this problem, this study intro- temperature of each material. Set the boundary condition
Published on 18 August 2015. Downloaded by University of Wollongong on 25/04/2016 09:21:01.

parameters and time step control parameters. Start the rst

duces the energy equation into the PIMPLE algorithm and
time step loop.
increases the coupling of the temperature eld, velocity eld
(2) Calculate the time step according the Courant number,
and pressure eld. Consequently, a more reasonable tempera-
then calculate the polymer viscosity and mixed material
ture eld distribution is acquired using the abovementioned
parameters of multiphase ow. Solve the phase fraction equa-
method, as shown in Fig. 2(b).
tion, and enter into the PIMPLE iteration loop.
Furthermore, to obtain more accurate bubble morphological
characteristics, the mesh size in the interface area cannot be (3) The PIMPLE iteration loop includes the inside and
over 1% of the bubble diameter. If one wants to calculate the outside loops. Aer solving the velocity and temperature elds
in the outside of the SIMPLE iteration loop, enter into the inside
macroscopic eld quantities in MFIM and precisely capture the
loop of the PISO iteration to solve the pressure eld. Then,
bubble interface at the micron level simultaneously, a mesh
calculate the ux on the surface of the mesh and correct the
with a uniform and small size will lead to huge computational
velocity eld. Judge the velocity and pressure elds are
overhead due to the large scale difference between the bubble
converged or not. If converged, break the PISO loop and return
and mold cavity. To solve this problem, the AMR technique53,54
to SIMPLE loop. If not, continue the PISO loop until PISO iter-
was used in this research. In each time step of calculation
ation is converged.
process, the mesh surrounding the interface was automatically
(4) Determine whether the converged results of PISO loop
rened according to the distribution state of phase. Fig. 3 shows
satisfy the convergence criterion of SIMPLE algorithm. If satis-
the mesh division result through the AMR, and the mesh on the
fying, break the PIMPLE loop. If not, continue the SIMPLE loop
ow front and polymer surface was deleted for clarity. As shown
until the SIMPLE iteration is converged. According to the
in Fig. 3, the mesh on the polymer ow front interface and
converged results, calculate the shear stress tensor, strain rate
bubble interface was rened, and the interfaces between the
tensor and viscous dissipation eld, and update the mesh by
different phases are clear and smooth.
AMR.
In this study a program based on the FVM called the Non-
(5) Check whether the mold cavity is completely lled or not.
isothermal Multiphase Flow Simulation System (N-MPFS) was
If not, the program will enter into the next time step and repeat

Fig. 2 Comparison of temperature results on the interface between the polymer and air acquired by the traditional algorithm and the algorithm
of this study. (a) Result of the traditional PIMPLE algorithm, (b) result of the algorithm of this study.

Fig. 3 Mesh division result using AMR technology.

70036 | RSC Adv., 2015, 5, 70032–70050 This journal is © The Royal Society of Chemistry 2015
 View Article Online

Paper RSC Advances

Published on 18 August 2015. Downloaded by University of Wollongong on 25/04/2016 09:21:01.

Fig. 4 Calculation flow chart of the program N-MPFS. (a) Whole flow chart, (b) flow chart of the PIMPLE loop.

the abovementioned procedures (2)–(4) until the polymer 2.596  105, respectively,38 and the parameters (A, b, b, j, Tr, s,
completely lls the mold cavity. a, m) in the polymer viscosity model (eqn (13)–(15)) are assigned
0.8298, 1272 K, 6.032  1013 Pa1, 66.51, 341.1 K, 23 160 Pa,
1.877, and 0.265, respectively.39 The physical parameters of the
3 Model validation uid for the model validation are presented in Table 2.
Compared to the MFIM technology, the deformation law of the Fig. 6 compares the experimental observation and simulation
bubble in Microcellular Foam Extrusion (MFE) ow eld has results of the bubble deformation process in the converging die.
been revealed more explicitly. Yoo and Han55,56 examined the a, b, c, and d on the horizontal axis in the gure represent the
MFE process, established a set of visualization system of the different positions on the neutral plane of the ow eld, where a
MFE process, and observed the deformation of a bubble in a
converging extrusion mold, as shown in Fig. 5. The observed
bubble had an initial diameter of 0.1 mm and located on the
center line of the ow eld with a distance of 1.3 mm from the
outlet. The temperature of the polymer melt and mold was 200

C. Although they used a chemical blowing agent (sodium
bicarbonate was used to produce carbon dioxide), the process
they researched was also the bubble deformation stage aer the
bubble growth, which is similar with proposed mathematical
model. In this way, Han's results are appropriate to validate the
present mathematical model.
In the numerical simulation process, the initial led vari-
ables and boundary conditions are kept the same as those in
literature,55 and the other model parameters that were not
reported in the literature are determined by referencing other
literature. The parameters D0, DT, DP, DTP in the interfacial Fig. 5 Schematic of the converging die in the visualization system
tension model (eqn (12)) are assigned 38.7032, 0.0559, 0.01, and built by Han et al.55

This journal is © The Royal Society of Chemistry 2015 RSC Adv., 2015, 5, 70032–70050 | 70037

RSC Advances
Table 2 Physical parameters of the fluids used for model validation
Density Specic isobar heat capacity
Materials (kg m3) (J kg1 K1)
PS 954.17 2100
CO2 11.29 1011
represents the farthest distance from the mold outlet and d
Published on 18 August 2015. Downloaded by University of Wollongong on 25/04/2016 09:21:01.
represents the location nearest the outlet. Vertical coordinates
show the bubble shape parameter D,57 which can be calculated
by the equation shown in Fig. 6. The D value of a round bubble
equals 0, and the D value of a tenuous shape of bubble is close to
1. The solid black bubbles in the gure are those observed from
Yoo's experiment, and the hollow bubbles in the gure are those
obtained from the simulation of this study.
Aer comparing two groups of bubble shapes shown in
Fig. 6, both the deformation law and the shapes of the two
groups of bubbles are identical in the entire extrusion process.
The bubble shape is gradually changed from an egg shape to a
fusiform shape and then to a slender shape with the progress of
the extrusion process. The D values of bubbles on the four
positions obtained from the visual observation are 0.41, 0.66,
0.8, and 0.92, and the D values of bubbles obtained from the
simulation are 0.44, 0.70, 0.85, and 0.94. Obviously, the two
groups of bubble have a similar D value of the bubble on each
position. The greatest error of the D value between two groups of
bubbles is less than 6.8%. The mathematical model established
in this study to simulate bubble deformation process in a
viscous uid is feasible.
However, the D value obtained from the model established in
this study is slightly lower than the result of visual observation
(average 5.1%). This means that the bubble shape of visual
observation is slightly more slender than that of the simulated
result. In the practical MFE process, the polymer elasticity in the
converging MFE mold will cause the uid pressure to increase
near the outlet, and this increasing pressure will squeeze the
bubble to be increasingly slender. Therefore, the reason for the
Fig. 6 Comparison of the experimental observation results and
simulation results.
70038 | RSC Adv., 2015, 5, 70032–70050
View Article Online
Paper
Heat conductance Kinematic viscosity
(W m1 K1) (m2 s1)
0.15 Eqn (13)–(15)
0.03157 2  106
abovementioned small error may be caused by ignoring the
elasticity of the polymer melt in the constitutive model used in
the mathematical model.
In contrast to the MFE process, the injection molded parts
are designed using the uniform thickness rules. There is rarely
converging ow eld in the injection mold cavity, and the rising
pressure caused by the polymer elasticity will be inconspicuous.
In this way, the mathematic model established in this study
ignores the fact that inuence of polymer elasticity on the
bubble shape is feasible for simulating the morphological
evolution of a bubble in the lling process of MFIM.
4 MFIM experiment study
To further validate the feasibility of the mathematical model
established in this study, clarify the morphological evolution
law of bubbles in the MFIM process and reveal the formation
mechanism of defects on the surface of the MFIM part, a ex-
ural sample (ASTM D790) was chosen to do the MFIM short shot
experiment with a 40% injection content, as shown in Fig. 7(b).
The research team of authors developed a SCF generator and its
measuring conveying system. They were mounted on a MA3200
conventional injection molding machine manufactured by
Ningbo Haitian plastic machinery group co., Ltd. The shut-off
nozzle was provided by Herzog, Switzerland. A specially
designed screw with a diameter of 60 mm and a length to
diameter ratio of 22 : 1 was installed in the machine. A sample
MFIM mold was designed and manufactured. Supercritical N2
(SCN2) with a purity of 99.99% was used as the physical foaming
agent. High Impact Polystyrene (HIPS, PH-888 G) provided by
Zhenjiang Chimei Chemical Industry Co. Ltd was selected as
the matrix polymer. The polymer was dried for more than 10
hours at 80  C before use. The process parameters for the MFIM
short shot experiment are shown in Table 3.
Fig. 7 shows the SEM images on the surface of exural
sample with 40% short shot, where Fig. 7(a) shows a SEM image
(200) of the sample surface, Fig. 7(b) shows the amplied
image of the front-end of exural sample and Fig. 7(c) shows the
SEM image (30) of the front-end of exural sample. This can
clearly distinguish the silver marks (groove), pit defects on the
surface and front-end of the exural sample. In Fig. 7(a), there
are a sea of zonal grooves and some pits marked by the dotted
line on the surface of the exural sample. In the Fig. 7(c), there
are many fusiform pits surrounded by the bubble skin on the
front-end of polymer. The reasons and mechanism of the
formation of these defects morphology on the sample surface
will be revealed in the following section according to the
numerical simulation and short shot experiment results. The
This journal is © The Royal Society of Chemistry 2015
 View Article Online

Paper RSC Advances

Published on 18 August 2015. Downloaded by University of Wollongong on 25/04/2016 09:21:01.

Fig. 7 SEM images of the surface of flexural sample with 40% short shot. (a) SEM image (200) of the sample surface, (b) amplified image of the
front-end of flexural sample, (c) SEM image (30) of the front-end of flexural sample.

Table 3 Process parameters of the MFIM short shot experiment of the

flexural sample

Process parameters Value

Injection delay time (s) 10

Injection rate (mm s1) 10
Injection time (s) 2.5
Melt temperature ( C) 240
Mold temperature ( C) 50
Supercritical uid level (wt%) 0.3
Cooling time (s) 20 Fig. 8 Schematic of the initial state of a bubble for simulating
Cooling water temperature ( C) 10 morphological evolution in MFIM, where plane A is the central plane of
the injection flow field in the thickness direction (neutral plane), plane
B is the central plane of injection flow field in the width direction.

geometric and process parameters of the numerical simulation

are the same as those of the experiment, but due to the Table 4 Numerical simulation scheme of the morphological evolu-
restriction of the experimental conditions, it is difficult to tion of a bubble with different initial diameters and positions
guarantee the consistency of material physical parameters of
Scheme no. h/mm L/mm d/mm
these two methods such as the viscosity and interfacial tension
coefficient. 1 1.6 1 0.15
2 1.6 1 0.2
3 1.6 1 0.25
4 1.6 1.5 0.2
5 1.6 0.5 0.2
5 Results and discussion 6 2.4 1 0.15
7 2.4 1 0.2
The morphological evolution of a bubble in the exural sample 8 2.4 1 0.25
(ASTM D790) lling stage of the MFIM process was simulated 9 2.0 1 0.2
using the abovementioned mathematical model. The melt 10 2.8 1 0.2
lling state of 0.2 s in MFIM was dened as the initial state for
the numerical simulation of bubble morphological evolution.
Fig. 8 shows a schematic of the initial simulation state. The X 5.1 Morphological evolution of a bubble on the neutral
direction is the polymer lling direction, the Y direction is the plane of injection ow eld
cavity width direction, and Z direction is the cavity thickness
direction. All the results of the later numerical simulation will Fig. 9 shows the numerical simulation result of the bubble
evolution process of Scheme 2 (h ¼ 1.6 mm, L ¼ 1 mm, d ¼
use the same coordinate system, as shown in Fig. 8. The initial
0.2 mm) in Table 4. All the images in Fig. 9 are the sectional view
bubble was spherical with d as the diameter, and the initial
(Y ¼ 6.35 mm) of the 3D result. The upper half of Fig. 9(a)–(i)
coordinates of the center of spherical bubble are dened as (L,
and (o) shows the temperature distribution of the polymer melt
6.35, h). Different values of h, L and d are assigned to study the
and the bottom half shows the velocity distribution of the
morphological evolution of a bubble with different initial
polymer melt on the X axis direction. Fig. 9(j)–(n) show the
diameters and positions in MFIM process, as listed in Table 4.
magnication of velocity distribution of the polymer melt on
Herein, the initial physical variables and boundary condi-
tions refer to the process parameter in Table 3, and the coeffi- the X direction during a lling time of 1.05 s to 1.3 s.
Fig. 9(a)–(g) shows the deformation and burst process of the
cients of the viscosity model and the interfacial tension model
bubble, and Fig. 9(h), (i), and (o) show the formation process of
are consistent with those in Section 3 of this study. Table 5 lists
the polymer surface defects. The corresponding relationship
the uid physical property parameters required here.

This journal is © The Royal Society of Chemistry 2015 RSC Adv., 2015, 5, 70032–70050 | 70039
 View Article Online

RSC Advances Paper

Table 5 Physical property parameters in simulating bubble morphological evolution in the MFIM process

Density Specic isobar heat capacity Heat conductance Kinematic viscosity

Materials (kg m3) (J kg1 K1) (W m1 K1) (m2 s1)

HIPS 927.68 2100 0.18 Eqn (13)–(15)

N2 13.14 1068 0.040 2  106
Air 1.078 1008 0.027 1.8  105
Published on 18 August 2015. Downloaded by University of Wollongong on 25/04/2016 09:21:01.

between the gure icon, lling time and the maximum melt
ow velocity is listed in Table 6.
Fig. 9(a) to (c) show that the bubble shape in the injection
ow eld changes from circular to elliptical and then crescent.
This is because that the trailing melt ows faster than the front
melt on the X direction, and there is a velocity gradient on the X
direction of the injection ow eld. The trailing melt will
squeeze the bubble and concurrently push it approaching
toward the ow front. This phenomenon was also conrmed by
the visualization experiment research results of Mahmoodi
et al.26
Fig. 9(d)–(g) presents the bubble burst process at the ow
forefront. The bubble becomes extremely attened under the
action of a fountain ow, and the bubble interface can be
divided into the front wall and back wall, as shown in Fig. 9(c).
When the lling time was 0.55 s, the front wall of the bubble
almost fused with the ow front interface, and the bubble is
about to burst. Aer bubble bursting, the back wall of the
bubble becomes the new ow front of the lling polymer, as
shown in Fig. 9(e). The bubble skin is formed aer the bubble
bursts, as shown by the solid circle mark in Fig. 9. The bubble
skin is conveyed to the mold surface by fountain ow and
continually fuses into the polymer melt with shrinking size, as

Fig. 9 Numerical simulation result of the bubble evolution process of Scheme 2 in Table 4.

70040 | RSC Adv., 2015, 5, 70032–70050 This journal is © The Royal Society of Chemistry 2015
 View Article Online

Paper RSC Advances

Table 6 Corresponding relationships between the figure icon, filling
time and the maximum melt flow velocity
Figure icon Time (s)
a 0
b 0.15
c 0.35
d 0.55
e 0.6
f 0.8
Published on 18 August 2015. Downloaded by University of Wollongong on 25/04/2016 09:21:01.
1.6 mm), and the initial X directional location and diameter of
the bubble is changed to study the different morphological
evolution of the bubble. Fig. 11 shows the bubble morpholog-
Max. Ux (m s1)
ical evolution of numerical simulation Schemes 1–3 (h ¼
0.0161 0.8 mm, L ¼ 1 mm), where the bubble diameters are dened as
0.0172 d1 ¼ 0.15 mm, d2 ¼ 0.2 mm, and d3 ¼ 0.25 mm. The three rows
0.0192 of images correspond to the simulation Schemes 1–3. As shown
0.0206
0.0209
in Fig. 11, the shapes of the bubbles with three different
0.022 diameters change from circle to vertical oval and then crescent.

g 0.9 0.0225 The bubble in the rst row of the image with the smallest size
h 1.05 0.0231 (d1 ¼ 0.15 mm) produces a bubble skin with a relatively small
i 1.15 0.0235 size aer burst. The bubble skin atrophies on the ow front
j 1.05 0.0231
k 1.1 0.0231
interface and vanishes before touching the mold cavity wall, as
l 1.15 0.0231 shown in 0.8 s image of the rst row. The second row of images
m 1.2 0.0231 in Fig. 11 gives the evolution process of the bubble with an
n 1.25 0.0231 initial diameter of 0.2 mm. The residual bubble skin aer
o 1.7 0.0252 bursting has a larger size, and it vanishes aer touching the
mold cavity wall under the action of a fountain ow, as shown in
0.95 s image of the second row. As the initial diameter of bubble
increases continuously, the residual bubble skin aer bursting
shown in Fig. 9(e)–(g). At last, it disappears. The partially
gradually becomes larger, as shown in the third row of Fig. 11.
enlarged Fig. 9(j)–(n) show the disappearing process of the
The bubble skin vanishes aer touching the mold cavity wall
bubble skin.
and sliding a distance along it, as shown in the 1.15 s image of
In addition, from Fig. 9, it can be also be found that a
the third row.
collapse phenomenon behind the bubble skin on the sample
Fig. 11 shows that the bubble skin will vanish on the ow
surface appears when the bubble skin disappears, as shown by
front interface when the initial position of a size bubble is
the dotted elliptical mark in Fig. 9. This collapse phenomenon
located on the neutral plane of the ow eld. It will not be
was also found from the SEM image of the exural sample
conveyed to the mold surface by the fountain ow. As a result, it
surface molded by the short shot experiment, which is marked
will not cause surface defects on the plastic part. Therefore, not
by dotted circle in Fig. 10. This indicates that the collapse
all the burst bubbles will lead to a defect on the surface of the
phenomenon really exists. The formation reason and condition
plastic part. However, when the initial position of a large bubble
about the collapse phenomenon will be further analyzed.
is located on the neutral plane of ow eld, the bubble skin will
be pushed to the mold cavity surface by the fountain ow and
5.2 Morphological evolution of a bubble with different cause defects on the surface of the plastic part. The larger the
initial positions and sizes on the neutral plane of injection diameter of the bubble is, more serious the defects on the
ow eld surface of plastic part will be.
In this section, it is assumed that the initial location of a bubble To further reveal the bubble burst and disappearance
is located on the thickness center surface of a sample (h ¼ process of the bubble skin, this study also examined the bubble
burst process on the ow front interface, and the simulation
result is shown in Fig. 12. Fig. 12(a) shows the ow front and the
viewing angle of a bubble. Fig. 12(b) gives a locally amplied
area of a pit in the SEM image in Fig. 7(c), and it really reects
the morphology of the pit on the ow front interface. Fig. 12(c)
shows the simulation result of the bubble bursting process on
the ow front interface obtained from the simulation Scheme 1.
As shown in the 0.6 s image of Fig. 12(c), the bubble shape
changed to an oval shape before bursting. As the polymer
continuously lls in the mold cavity, the bubble begins to burst,
and the crevasse on the ow front interface exhibits an oval
shape. From the images of 0.604 s to 0.64 s shown in Fig. 12(c),
in this short period of time, the crevasse expanded rapidly. As a
result, the size of the crevasse gradually approaches to that of
the burst bubble, and the size of bubble skin continuously
shrinks and nally evolves into a narrow annulus. When the
lling time reaches 0.7 s, the bubble skin on the ends of the
Fig. 10 SEM image (150) of flexural sample surface obtained from minor axis of the oval disappears completely, and the bubble
the short shot experiment. skin on the ends of the long axis still has a little rest. Compared

This journal is © The Royal Society of Chemistry 2015 RSC Adv., 2015, 5, 70032–70050 | 70041
 View Article Online

RSC Advances Paper

Published on 18 August 2015. Downloaded by University of Wollongong on 25/04/2016 09:21:01.

Fig. 11 Morphological evolution of a bubble in the front part of the melt obtained from simulation Schemes 1–3.

Fig. 12 Bubble bursting process on the flow front. (a) Flow front and viewing angle of a bubble, (b) SEM image of the pit, (c) simulation result of
the bubble bursting process on the flow front interface.

to the SEM image of the crevasse morphology shown in the bubble skin disappears completely. Compared to the results
Fig. 12(b), the simulation results are in good agreement with of the moment before, the crevasse at this moment has been
those of the experiments. When the lling time reaches 0.8 s, obviously stretched by the fountain ow on the long axis of the

70042 | RSC Adv., 2015, 5, 70032–70050 This journal is © The Royal Society of Chemistry 2015

Paper
bubble. The bubble aer bursting forms a spindle pit
morphology on the ow front interface, as shown in the B–B
sectional view in Fig. 12(c).
To study the morphological evolution law of bubble with
different initial positions in the X direction, simulation
Schemes 5, 2 and 4 (h ¼ 1.6 mm, d ¼ 0.2 mm) were performed.
Fig. 13 gives the numerical simulation results of these schemes.
The three rows of images in Fig. 13 correspond to the results of
simulation Schemes 5, 2 and 4, where L1 ¼ 0.5 mm, L2 ¼ 1 mm
Published on 18 August 2015. Downloaded by University of Wollongong on 25/04/2016 09:21:01.
and L3 ¼ 1.5 mm; Fig. 13(a)–(c) show the partially enlarged
drawing of the residual bubble skin with three different
positions.
Fig. 13 shows that the deformation, bursting and dis-
appearing process of a bubble with different initial positions are
quite different. When the initial position of the bubble is near
the inlet of the ow eld (value of L is the smallest), the simu-
lation results are shown in the rst row of image in Fig. 13. The
bubble shape becomes an approximate ellipse with long axis
along the X direction in the early deformation stage, as shown in
the image of 0.05 s in the rst row. It changes to an approximate
ellipse with a long axis along the Z direction, as shown in the
image of 0.3 s in the rst row. Subsequently, it evolves to a
crescent shape under the squeezing action of the fountain ow
eld, as shown in the image of 0.8 s in the rst row. Finally, the
bubble bursts and its skin is gradually pushed onto the mold
cavity wall surface. When the initial position of the bubble is
Fig. 13 Morphological evolution of a bubble in the front part of a polymer melt obtained from simulation Schemes 5, 2 and 4.
This journal is © The Royal Society of Chemistry 2015
View Article Online
RSC Advances
relatively far away from the inlet of the ow eld (value of L is
larger), the morphology of the bubble before bursting is
different from that of smaller L, as shown in the second and
third row of the image in Fig. 13. The bubble did not evolve to
the approximate ellipse with long axis along the X direction in
the early deformation stage, but directly evolved to the
approximate ellipse with a long axis along the Z direction.
Furthermore, different initial positions of the bubble corre-
spond to the different bubble bursting times. The shorter the
distance from the initial position of the bubble to the ow front
interface is, the earlier the bubble bursting time is. At the same
time, the bubbles with different initial positions have different
skin thicknesses aer bursting. As the distance from the initial
position of bubble to ow front interface increases, the thick-
ness of the bubble skin aer bursting decreases, as shown as t1
in Fig. 13(a), t2 in Fig. 13(b) and t3 in Fig. 13(c). This is because
that the initial bubble with a longer distance to the ow front
interface has been subject to a long time of action of fountain
ow before bursting, its shape becomes extremely at aer
being fully squeezed and the bubble skin aer bursting is
relatively thinner.
Fig. 14 compares the numerical simulation results of the
polymer surface collapse depth for the different simulation
Schemes 5, 2 and 4 at the lling time of 1.7 s. The curves in the
gure represent the interface shapes of polymer surface for
different cases. The D1–D3 represent the maximum depth of the
RSC Adv., 2015, 5, 70032–70050 | 70043

RSC Advances
polymer melt surface collapse of the simulation Schemes 5, 2
and 4, respectively. The relationship of three collapse depths is
D3 > D2 > D1. It can be noted from Fig. 13 and 14 that as the
distance from the initial bubble position to the ow front
interface decreases, the thickness of the bubble skin aer its
bursting and the depth of the polymer surface collapse behind
the bubble skin increases. This is because in the melt lling
process, the formed bubble skin turns into the obstacle of the
rear melt, preventing the polymer surface behind the bubble
Published on 18 August 2015. Downloaded by University of Wollongong on 25/04/2016 09:21:01.
skin from maintaining the plane state, and nally makes the
polymer surface separate from the mold wall and form a
collapse defect. The thinner the bubble skin is, the smaller the
impediment imposed on the rear melt is, and the shallower the
surface collapse of plastic part is, and vice versa.
5.3 Morphological evolution of the bubble deviating from
the neutral plane of injection ow eld
Fig. 15 shows the morphological evolution of a bubble with the
initial position deviating from the neutral plane of the melt ow
eld in the lling process of MFIM, which corresponds to
simulation Scheme 7 (h ¼ 2.4 mm, L ¼ 1 mm, and d ¼ 0.2 mm).
The upper part of Fig. 15(a)–(j) and (p)–(t) shows the tempera-
ture distribution of the polymer melt, and the lower part is the
velocity (X directional) distribution diagram of the polymer
melt. Fig. 15(p)–(t) show a partially enlarged drawing of the
back-end bubble skin, and Fig. 15(k)–(o) show the partially
enlarged drawing of temperature distribution of the front-end
bubble skin. Fig. 15(a)–(f) shows the deformation and
bursting process of the bubble. Fig. 15(g)–(i) shows the forma-
tion process of surface defect of polymer. Table 7 lists the cor-
responding relation between the gure icon, lling time and the
maximum value of melt ow velocity.
As shown in Fig. 15(a)–(g), under the action of fountain ow,
the bubble approaches the ow front interface, and at the same
time, deviates from the neutral plane. It is then pushed to the
mold cavity wall and nally bursts on the mold surface.
Fig. 15(h)–(j) gives the formation of a surface pit and the
evolution process of the front-end and back-end residual bubble
skin aer bubble bursting. Fig. 15(h) shows that when the
bubble is bursting, the bubble skins between the front-end and
back-end of the pit have an obvious morphological difference,
and with the progress of polymer melt lling the mold cavity,
Fig. 14 Comparison of the polymer surface collapse depths of
numerical simulation Schemes 5, 2 and 4 at the filling time of 1.7 s.
70044 | RSC Adv., 2015, 5, 70032–70050
View Article Online
Paper
the evolution process of these two parts of the bubble skin is
obviously different.
The morphology of a bubble depends on the ow eld
environment in which it lies. As shown in Fig. 15(a)–(g), in the
initial stage of deformation, the bubble is located in the high
temperature area, wherein the ow ability of the polymer melt
is favorable. The bubble is pushed easily to the mold cavity
surface under the action of a fountain ow. The back-end of
the bubble gets close to the mold surface rstly, as shown in
Fig. 15(e). At this moment, the melt around the back-end of
bubble is cooled down gradually and its ow ability decreases
gradually. As a result, the movement of the whole bubble,
particularly the back half part of the bubble, begins to slow
down along the polymer lling direction. However, the melt
near the front-end of the bubble has a relatively strong ow
ability and continues to ow along the lling direction
because its temperature is still relatively high. As a result, the
bubble is elongated into a slender shape, as shown in
Fig. 15(e)–(g).
Fig. 15(g) shows the moment of the bubble bursting aer it
touches the mold cavity wall. Interestingly, the bubble
bursting position is not the bubble's back-end approaching
the mold cavity wall in the earliest, but the front-end. The
reason is that the bubble's back-end is difficult to touch the
mold wall subject to the obstruction of polymer melt
condensing layer, whereas the uid ability of the melt near
the bubble's front-end is better; the front-end can continu-
ously approach and nally touch the mold wall to burst. The
two parts of the bubble skin formed aer bursting exhibit
completely different evolution processes.
Fig. 15(k)–(o) display the evolution process of the front-end
bubble skin. Because the temperature of melt surrounding it
is high, the residual bubble skin is gradually swallowed and
eventually disappears. Fig. 15(p)–(t) show the evolution process
of the back-end bubble skin. The temperature of the melt
surrounding it is low. The back-end bubble skin is relatively far
away from the fountain ow eld and barely inuenced by
fountain ow. Its location is relatively unchanged. However, its
shape changes with decreasing length, thinning thickness and
decreasing gap with the internal melt.
As shown in Fig. 15(j), when the polymer lling time reaches
1.7 s, the collapse phenomenon is found behind the front-end
bubble skin on the surface of the plastic part, but the surface
near the back-end bubble skin remains at. This is because the
formation of collapse defects behind the bubble skin requires
an appropriate temperature range of the polymer melt. On one
hand, if the temperature is too high, the polymer melt with
strong uid ability can quickly engulf the bubble skin and the
engulfed bubble skin has a little inuence on the rear melt
surface. As a result, rear melt surface collapse will not occur.
The simulation result of the simulation Scheme 1 shown in
Fig. 11 shows a similar circumstance. On the other hand, if the
temperature is too low, the polymer melt surface will conden-
sate quickly and its shape is hard to be changed and the surface
collapse of the polymer melt behind the bubble skin will not
appear, as shown in 15(p)–(t).
This journal is © The Royal Society of Chemistry 2015

Paper
Published on 18 August 2015. Downloaded by University of Wollongong on 25/04/2016 09:21:01.
Fig. 15 Morphological evolution of a bubble obtained from the simulation Scheme 7.
5.4 Morphological evolution of a bubble with different
initial diameters and positions deviating from the neutral
plane of the injection ow eld
This section mainly studies the morphological evolution of a
bubble with different initial sizes and positions on the Z axis
direction, i.e., different initial positions of bubble deviating
from the neutral surface of the injection ow eld. Fig. 16 gives
the simulation results of simulation Schemes 6–8 (h ¼ 2.4 mm,
L ¼ 1 mm), wherein the bubble initial position is located on the
section of thickness with the different initial diameters, d1 ¼
0.15 mm, d2 ¼ 0.2 mm, and d3 ¼ 0.25 mm. The X axis denotes
the coordinate along the ow eld direction, wherein the inlet is
the origin of the coordinate. The Z axis denotes the distance to
This journal is © The Royal Society of Chemistry 2015
View Article Online
RSC Advances
the neutral surface of the ow eld. Fig. 16(a)–(c) give the
simulation results corresponding to simulation Schemes 6–8,
respectively, wherein Fig. 16(d) shows the enlarged drawing of
areas circled by the dotted rectangular box in Fig. 16(a)–(c).
Fig. 16 shows that the length of surface pits formed by the
bubbles with different initial size is different, shown as 8.1 mm,
11.1 mm and 12.2 mm in Fig. 16(a)–(c). With increasing initial
diameter of the bubble, the length of surface pits increases.
What is the reason for this length difference? Obviously, the
volume or size difference of the bubbles is not enough to cause
such a distinct length difference. From Fig. 16(a)–(c), it can be
seen that the evolution of these three back-end bubble skin is
similar. Even aer the long evolution time, the morphology of
three back-end bubble skins is also similar and their positions
RSC Adv., 2015, 5, 70032–70050 | 70045
 View Article Online

RSC Advances Paper

Table 7 Corresponding relations between the figure icon, filling time

and the maximum value of the melt flow velocity

Figure icon Time (s) Max. Ux (m s1)

a 0 0.0154
b 0.05 0.0158
c 0.1 0.0165
d 0.2 0.0177
e 0.35 0.0192
f 0.45 0.020
Published on 18 August 2015. Downloaded by University of Wollongong on 25/04/2016 09:21:01.

g 0.7 0.0215
h 0.75 0.0218
i 0.85 0.0223
j 1.7 0.0254
k 0.75 —
l 0.85 —
m 1 —
n 1.2 —
o 1.7 —
p 0.75 0.0218
q 0.85 0.0223
r 1 0.023
s 1.2 0.0237
t 1.7 0.0254

is basically xed, as shown at a-2, b-2 and c-2 in Fig. 16(d).

Nevertheless, the evolution of these three front-end bubble
skins is quite different. The authors considered that this
difference is caused by the different bursting locations of the
bubbles.
As shown in a-1, b-1 and c-1 of Fig. 16(d), for the bubble with
the smallest initial diameter, most part of the front-end bubble
skin touched the mold cavity wall when it bursts. However, for
the bubble with the largest initial diameter, the front-end
bubble skin does not touch the mold cavity wall when it
bursts. The front-end skin of the bubble with a smallest size
touches the mold cavity wall in a large area and is obviously
subject to the cooling action from the mold. It is hard to
disappear because it has the poorest uid ability among the
three bubble skins. In contrast, for the bubble with a large
diameter, because the front-end bubble skin is located in the
high temperature area of polymer melt with better ow ability, it
almost disappears at the last moment of simulation, as shown
in Fig. 16(b) and (c). For the bubble with a small diameter, the Fig. 16 Morphological evolution of the bubble obtained by simulation
front-end bubble skin still remains at the last moment of the Scheme 6–8 (h ¼ 2.4 mm, L ¼ 1 mm). (a) Simulation result of Scheme
6, (b) simulation result of Scheme 7, (c) simulation result of Scheme 8,
simulation due to the skin in a low temperature area, as shown
(d) partially enlarged drawing of the simulation results.
in Fig. 16(a). Therefore, this ow ability difference of different
front-end bubble skins causes a large difference in the
displacement distance of front-end bubble skins in the polymer
defect on the surface of the MFIM parts. In contrast, bubble
lling process.
bursting on the mold surface produces a shorter pit or an
In the formation process of the pits, the back-end bubble
unobvious silver mark defect.
skins is basically xed but the front-end bubble skins continu-
From a-3 shown in Fig. 16(d), it can be also seen that the
ously migrate forwards, and nally the pits are elongated
collapse phenomenon appears behind the front-end bubble
signicantly. According to this viewpoint, it can be concluded
skin on the polymer surface, but it is not found near the back-
that the bursting location of the bubbles is the main factor
end bubble skin. This is because the bubble skin works as a
inuencing the length of the pit. The bubble bursting on the
barrier to the subsequent polymer melt in the ow process, and
triple point of polymer melt, mold and air easily leads to a
the melt surface with high ow ability is difficult to keep the
longer pit, and actually, this type of long pit is the silver mark
plane state. From a-3 in Fig. 16(d), it can be also found that the

70046 | RSC Adv., 2015, 5, 70032–70050 This journal is © The Royal Society of Chemistry 2015

Paper
rear area of front-end bubble skin is the bottom of the pit, which
has not touched the mold surface. This area has good ow
ability and can easily form a collapse phenomenon. However,
the polymer melt near the back-end bubble skin has been
cooled down and its ow ability becomes poor. Thus, it is hard
to form the collapse phenomenon.
To further reveal the morphological evolution of the pit
caused by the bubble bursting on the mold cavity surface,
Fig. 17 gives the corresponding numerical simulation results.
Published on 18 August 2015. Downloaded by University of Wollongong on 25/04/2016 09:21:01.
Fig. 17(a) presents the ow front and the viewing angle of a
bubble. Fig. 17(b) provides the partially enlarged drawing of
Fig. 7(a) with the camera site located behind the ow front to
display the real morphology of the pit on the plastic part
surface. Fig. 17(c) shows the numerical simulation result of the
bubble bursting process and the morphological evolution of pit
on the mold cavity surface obtained by the mathematical model
established in this study.
Fig. 17 shows that the bubble has not been formed into a
slender shape before bursting, as shown in A–A section in
Fig. 17(c). Aer bubble bursting, the residual bubble skin
quickly touches the mold wall surface. As stated above, this
early contact of the skin with the mold wall surface leads to a
decrease in the ow ability of the bubble skin, and a relatively
short pit is caused, as shown in Fig. 17(b). The morphology of
the bubble when it bursts determines the subsequent evolution
of the pit. From the B–B section of Fig. 17(c), it can be seen that
there is a distance between the pit bottom and mold cavity wall,
and the pit depth changes from shallow to deep along the
polymer lling direction. In the following-up evolution process,
the pit bottom extends to beneath the front polymer, as shown
in the D–D section of Fig. 17(c).
According to the simulation results, the reason for this type
of pit morphology lies in the ow ability difference between the
Fig. 17 Morphological evolution of the pit caused by bubble bursting on the mold cavity surface. (a) Flow front and the viewing angle of the
bubble, (c) SEM image of the pit, (d) numerical simulation result of the bubble bursting process and morphological evolution of the pit.
This journal is © The Royal Society of Chemistry 2015
View Article Online
RSC Advances
bottom and the crevasse of the pit. Compared to the pit crevasse
cooled by mold cavity wall, the pit bottom is not subject to the
cooling inuence of the mold cavity wall and thereby has better
ow ability. This difference results in a relative displacement
between the pit crevasse and bottom. At last, the pit morphology
with a deep front and shallow rear is nally formed, as shown in
Fig. 17(b). Through the abovementioned comparison and
analysis, it can be seen that the simulation results about the
position and morphology of the pit are in good agreement with
those of the short shot experiment.
Fig. 18 shows the morphological evolution of the initial
bubble deviating from the neutral plane of the injection ow
eld with three distances (h1 ¼ 0.4 mm, h2 ¼ 0.8 mm, h3 ¼ 0.12
mm), which correspond to the simulation Scheme 9, 7 and 10,
respectively, listed in Table 4. From Fig. 18, the rst bursting
bubble is neither the bubble whose initial position is closest to
the polymer surface, as shown in Fig. 18(c), nor the bubble
whose initial position is farthest from the polymer surface, as
shown in Fig. 18(a), but the bubble whose initial position is
located in the middle, as shown in Fig. 18(b). This is because the
migration of bubble interface is hampered by the polymer
condensation layer and not subjected to the squeezing action of
fountain ow, as shown in Fig. 18(c). As a result, the bubble is
difficult to reach the mold cavity surface and burst. However, if
the temperature of the polymer melt surrounding the bubble is
high, this type of elongated bubble is still possible to burst. This
paper has not yet studied the effect of the mold cavity surface
temperature on the bubble deformation, bursting behavior and
bubble skin evolution process. These effects are still worthy of
study in the future.
As shown in Fig. 18(c), the slender bubble is stretched
further by the polymer melt in the lling direction and the back
end of bubble is nally separated into two small bubbles. c-1
RSC Adv., 2015, 5, 70032–70050 | 70047

RSC Advances
Published on 18 August 2015. Downloaded by University of Wollongong on 25/04/2016 09:21:01.
Fig. 18 Morphological evolution of a bubble obtained from simulation
Schemes 9, 7 and 10. (a) Result of Scheme 9, (b) result of Scheme 7, (c)
result of Scheme 10, (d) partially enlarged drawing of the simulation
results.
70048 | RSC Adv., 2015, 5, 70032–70050
View Article Online
Paper
Fig. 19 Temperature field, velocity field and hodograph of the flow
field at filling time of 1.5 s for simulation Scheme 10.
and c-2 in Fig. 18(d) give the enlarged drawing of these two
bubbles. The front-end of bubble begins to extend to the ow
eld center under the action of an injection ow eld, as shown
in the elliptical wireframe of Fig. 18(c). Fig. 19 gives the
temperature eld, velocity eld and hodograph of the ow eld
at the lling time of 1.5 s for simulation Scheme 10. It can be
surmised that there is a velocity component pointing to the
neutral surface of the velocity eld around the bubble front-end
interface. This velocity component leads to the front-end of
bubble extending to the center of the ow eld.
6 Conclusion
This study established an incompressible, unsteady, and non-
isothermal 3D mathematical model of multiphase ow in the
MFIM process. A new set of boundary conditions and its
applying method were proposed. The energy equation was
coupled into the PIMPLE loop algorithm and the temperature
divergence problem on the interface between two phases with
large viscosity ratio was effectively overcome. The tracking
accuracy of micron grade bubbles interface in macroscopic
scale ow eld was improved signicantly using the physical
eld variable-based AMR technology. With the established
mathematical model and MFIM short shot experiment of the
exural sample, the bubble morphological evolution in the
lling process of MFIM was obtained. The following conclu-
sions were drawn:
1. The established model successfully predicted the surface
collapse phenomenon near the pit formed by bubble bursting,
and the existence of the collapse was veried in the SEM image
of plastic part surface. This collapse could only appear in a
certain range of polymer melt temperature, and the barrier
action of the bubble skin to polymer melt ow is the main
reason of causing this collapse.
2. The formation mechanism of the pit with a deep front and
shallow rear on the plastic part surface was revealed. This type
of pit is formed only when a bubble bursts on the mold cavity
wall. This is the ow ability difference of the polymer melt in the
bottom and the crevasse of the pit to cause this type of pit with a
deep front and shallow rear.
3. The morphological evolution of the bubble before
bursting can be divided into two cases according to the bubble
This journal is © The Royal Society of Chemistry 2015

Paper
initial position: the bubble on the neutral surface of the ow
eld and the bubble deviating from the neutral surface of the
ow eld. The former experiences the morphological evolution
of a circle, oval, and crescent and continuously approaches the
ow front. The latter experiences the morphological evolution
of a circle, teardrop, and slender and continuously approaches
the mold cavity wall. Nevertheless, in the melt lling process, it
is not always true that the burble bursts easily when the
distance between the mold cavity wall and the bubble decreases.
Published on 18 August 2015. Downloaded by University of Wollongong on 25/04/2016 09:21:01.
Once the polymer condensation layer is formed, it will hamper
the bubbles approaching the mold cavity wall. Without the
transport of a fountain ow and a relatively high temperature of
polymer melt, it is difficult for the bubbles to reach the mold
cavity wall to burst.
4. The morphological evolution of a bubble aer bursting
can be divided into three cases according to the bubble bursting
position: (1) bubble bursting on the interface between the melt
and air, i.e., ow front, (2) bubble bursting on the interface
between melt and mold, i.e., mold wall, (3) bubble bursting on
the triple point of polymer melt, mold and air. In the rst case,
the residual skins formed aer bursting of the bubbles with a
small initial size or those whose nucleation position is close to
the ow front can disappear before touching the mold cavity
wall without any deleterious effect on the product surface
quality. In the second case, the burst bubble is affected by the
cooling effect of the mold cavity wall, and it is difficult to stretch
to form a thin silver defect but a pit with a deep front and
shallow rear. In the third case, the burst bubbles most likely
result in the silver mark defects. This is because there is a ow
ability difference between front-end bubble skin and back-end
bubble skin aer bubble bursting, and the pit is elongated
and a silver mark morphology is formed.
5. A comparison of the numerical simulation results with
those of short shot experiment indicated that the model
established in this study can accurately capture the phase
interface of multiphase in the 3D ow eld and simulate the
overall process of bubble morphological evolution in the lling
process of MFIM. Using this model, we can obtain the bubble
bursting and disappearing process, which is difficult to catch
with the visual equipment. Furthermore, the boundary condi-
tions and their applying method provide a reliable guarantee for
an accurate description of the bubble morphological evolution
in the lling process of the MFIM.
Acknowledgements
This research study was supported by the Program for Chang
Jiang Scholars and Innovative Research Team in the University
of Ministry of Education of China (No. IRT0931) and the
Research Award Fund for Shandong Province Excellent Inno-
vation Team (No. 2012-136).
References
1 J. E. Martini, Master's thesis, University of Massachusetts,
1981.
This journal is © The Royal Society of Chemistry 2015
View Article Online
RSC Advances
2 J. E. Martini, F. A. Waldman and N. P. Suh, Annu. Tech. Conf.
Soc. Plast. Eng., 1982, 28, 674–676.
3 J. Y. Xu, Microcellular Injection Molding, ed. R. F. Grossman
and D. Nwabunma, John Wiley & Sons, 2011, vol. 9, ch. 1,
pp. 1–11.
4 A. Kramschuster and L. S. Turng, J. Biomed. Mater. Res., Part
B, 2010, 92, 366–376.
5 H. Y. Mi, X. J. Max, R. Salick, L. S. Turng and X. F. Peng,
J. Mater. Res., 2014, 29(8), 911–922.
6 G. L. Wang, G. Q. Zhao, J. C. Wang and L. Zhang, Polym. Eng.
Sci., 2014, 55, 807–835.
7 G. W. Dong, G. Q. Zhao, Y. J. Guan, G. L. Wang and
X. X. Wang, J. Appl. Polym. Sci., 2014, DOI: 10.1002/
app.40365.
8 S. Li, G. Q. Zhao, G. L. Wang, Y. J. Guan and X. X. Wang,
J. Cell. Plast., 2014, 50(5), 415–435.
9 J. J. Lee, L. S. Turng, E. Dougherty and P. Gorton, Polymer,
2011, 52, 1436–1446.
10 S. W. Cha and J. D. Yoon, Polym.-Plast. Technol. Eng., 2005,
44(5), 795–803.
11 D. E. Pierick, J. R. Anderson, S. W. Cha, L. Q. Chen,
J. F. Stevenson and D. E. Laing, US Pat., 6 884 823, 2005.
12 A. N. J. Spoerrer, D. S. Bangarusampath and V. Altstaedt,
Cell. Polym., 2008, 27(2), 101–121.
13 S. C. Chen, W. H. Liao and R. D. Chien, Int. Commun. Heat
Mass Transfer, 2012, 39(8), 1125–1131.
14 S. C. Chen, P. S. Hsu and S. S. Hwang, J. Appl. Polym. Sci.,
2013, 127(6), 4769–4776.
15 C. D. Han and H. Yang, J. Appl. Polym. Sci., 1984, 29(12),
4465–4470.
16 J. Colton and N. Suh, Polym. Eng. Sci., 1987, 27(7), 485–492.
17 N. S. Ramesh, G. A. Campbell and D. H. Rasmussen, Polym.
Eng. Sci., 1994, 34(22), 1685–1697.
18 C. D. Han and H. J. Yoo, Polym. Eng. Sci., 1981, 21(9), 518–
533.
19 M. Amon and C. D. Denson, Polym. Eng. Sci., 1984, 24(13),
1026–1034.
20 A. Arefmanesh, S. G. Advani and E. E. Michaelides, Polym.
Eng. Sci., 1990, 30(20), 1330–1337.
21 N. S. Ramesh, D. H. Rasmussen and G. A. Campbell, Polym.
Eng. Sci., 1991, 31(23): 1657–1664.
22 A. Osorio and L. S. Turng, Polym. Eng. Sci., 2004, 44(12),
2274–2287.
23 S. N. Leung, C. B. Park, D. L. Xu, H. B. Li and R. G. Fenton,
Ind. Eng. Chem. Res., 2006, 45(23), 7823–7831.
24 C. A. Villamizar and C. D. Han, Polym. Eng. Sci., 1978, 18(9),
699–710.
25 C. D. Han and C. A. Villamizar, Polym. Eng. Sci., 1978, 18(9),
687–698.
26 M. Mahmoodi, A. H. Behravesh, S. A. M. Rezavand and
A. Pashaei, J. Appl. Polym. Sci., 2010, 116(6), 3346–3355.
27 A. Wong and C. B. Park, Polym. Test., 2012, 31(3), 417–424.
28 A. Ahmadzai, A. H. Behravesh, M. T. Sarabi and P. Shahi,
J. Cell. Plast., 2014, 50(3), 279–300.
29 S. B. Pillapakkam and P. Singh, J. Comput. Phys., 2001,
174(2), 552–578.
RSC Adv., 2015, 5, 70032–70050 | 70049

RSC Advances
30 I. B. Bazhlekov, P. D. Anderson and E. H. M. Han, J. Colloid
Interface Sci., 2006, 298(1), 369–394.
31 P. Yue, J. J. Feng, C. Liu and J. Shen, J. Fluid Mech., 2005, 540,
427–437.
32 J. Lee and C. Pozrikidis, Comput. Fluids, 2006, 35(1), 43–60.
33 P. J. A. Janssen and P. D. Anderson, Phys. Fluids, 2007, 19,
043602.
34 A. M. Ardekani, S. Dabiri and R. H. Rangel, Phys. Fluids,
2009, 21(9), 093302.
Published on 18 August 2015. Downloaded by University of Wollongong on 25/04/2016 09:21:01.
35 K. Verhulst, R. Cardinaels, P. Moldenaers, S. Ahami and
Y. Renardy, J. Non-Newtonian Fluid Mech., 2009, 156(1), 44–
57.
36 S. S. Rabha and V. V. Buwa, Chem. Eng. Sci., 2010, 65(1), 527–
537.
37 Y. K. Wei, Y. Qian and H. Xu, J. Comput. Multiphase Flows,
2012, 4(1), 111–118.
38 H. Park, R. B. Thompson, N. Lanson, C. Tzoganakis,
C. B. Park and P. Chen, J. Phys. Chem. B, 2007, 111(15),
3859–3868.
39 M. Lee, C. B. Park and C. Tzoganakis, Polym. Eng. Sci., 1999,
39(1), 99–109.
40 W. Huang and R. D. Russell, Adaptive moving mesh methods,
ed. S. S. Antman, J. E. Marsden and L. Sirovich, Springer,
New York, 2011.
41 H. G. Weller, G. Tabor, H. Jasak and C. Fureby, Comput.
Phys., 1998, 12(6), 620–631.
42 C. W. Hirt and B. D. Nichols, J. Comput. Phys., 1981, 39(1),
201–225.
70050 | RSC Adv., 2015, 5, 70032–70050
View Article Online
Paper
43 S. O. Unverdi and G. Tryggvason, J. Comput. Phys., 1992,
100(1), 25–37.
44 W. Martin, Microuid. Nanouid., 2012, 12(6), 841–886.
45 J. U. Brackbill, D. B. Kothe and C. Zemach, J. Comput. Phys.,
1992, 100(2), 335–354.
46 H. E. Park and J. M. Dealy, Macromolecules, 2006, 39(16),
5438–5452.
47 D. Raps, T. Köppl, A. R. de Anda and V. Altstädt, Polymer,
2014, 55(6), 1537–1545.
48 H. Fujita and A. Kishimoto, J. Polym. Sci., 1958, 28(118), 547–
567.
49 J. R. Royer, J. M. D. Simone and S. A. Khan, J. Polym. Sci., Part
B: Polym. Phys., 2001, 39(23), 3055–3066.
50 J. F. Hetu, D. M. Gao, A. Garcia-Rejon and G. Salloum, Polym.
Eng. Sci., 1998, 38(2), 223–236.
51 OpenFOAM, http://www.openfoam.com, accessed October
2014.
52 J. A. Horacio, M. D. Santiago, M. J. Gimeneza and
M. N. Norberto, Mech. Comput., 2013, 32, 995–1011.
53 P. Yue, C. Zhou, J. J. Feng, C. F. Ollivier-Gooch and H. H. Hu,
J. Comput. Phys., 2006, 219(1), 47–67.
54 I. Ginzburg and G. Wittum, J. Comput. Phys., 2001, 166(2),
302–335.
55 H. J. Yoo and C. D. Han, Polym. Eng. Sci., 1981, 21(2), 69–75.
56 H. J. Yoo and C. D. Han, J. Rheol., 1981, 25(1), 115–137.
57 G. I. Taylor, Proc. R. Soc. London, Ser. A, 1934, 501–523.
This journal is © The Royal Society of Chemistry 2015