characterization and simulation of material distribution - qucosa

Characterization and Simulation of Material Distribution

and Fiber Orientation in Sandwich Injection Molded Parts

Von der Fakultät für Maschinenbau der

Technischen Universität Chemnitz

genehmigte

Dissertation

zur Erlangung des akademischen Grades

Doktor-Ingenieur (Dr.-Ing.)

vorgelegt

von M. Eng. Somjate Patcharaphun

geboren am 25.12.1972 in Bangkok, Thailand

Gutachter: Prof. Dr.-Ing. G. Mennig

Prof. Dr.-Ing. J. Wortberg

Prof. Dr.-Ing. Habil. B. Wielage

Tag der Einreichung: 21.05.2006

Tag der Verteidigung: 29.10.2006

URL: http://archiv.tu-chemnitz.de/pub/2006/0184

ISBN: 3-939382-04-3 (978-3-939382-04-1)

Bibliographic Description

Author: Patcharaphun, Somjate

Topic: Characterization and Simulation of Material Distribution and Fiber Orientation in

Sandwich Injection Molded Parts

A Dissertation submitted to the Faculty of Mechanical Engineering, Institute of Mechanical and

Plastics Engineering, Chemnitz University of Technology, 2006.

133 Pages, 69 Figures, 14 Tables, 144 References

Abstract

In this work, the material distribution, structure of fiber orientation and fiber attrition in

sandwich and push-pull injection molded short fiber composites are investigated, regarding the

effect of fiber content and processing parameters, given its direct relevance to mechanical

properties. The prediction of the tensile strength of conventional, sandwich and push-pull

injection molded short fiber composites are derived by an analytical method of modified rule of

mixtures as a function of the area fraction between skin and core layers. The effects of fiber

length and fiber orientation on the tensile strength are studied in detail. Modeling of the

specialized injection molding processes have been developed and performed with the simulation

program in order to predict the material distribution and the fiber orientation state. The second-

order orientation tensor ( ) approach is used to describe and calculate the local fiber

orientation state. The accuracy of the model prediction is verified by comparing with

corresponding experimental measurements to gain a further basic understanding of the melt flow

induced fiber orientation during sandwich and push-pull injection molding processes.

11a

Key words: Sandwich injection molding, Push-Pull injection molding, Fiber orientation

distribution, Fiber length distribution, Material distribution, Mechanical properties, Numerical

simulation.

Acknowledgements

This work is based on research conducted between March 2003 and April 2006 at the Institute of

Mechanical and Plastics Engineering at the University of Chemnitz. A number of people have

contributed to the completion of this thesis, and deserve to be thanked.

First and foremost, I would like to express my sincere gratitude to my supervisor, Prof. Dr.-Ing.

Günter Mennig for his invaluable guidance and encouragement throughout my studies. His

support during my pursuit of the doctor degree will always be appreciated.

Special thanks to Dr.-Ing. Hannes Michael for his kindness and encouragement. I enjoyed the

many hours of lively exchanges (technical or not) that we had and look forward to future

collaborations.

I would like to express my thanks to Dipl.-Ing. Helmut Püschner, and other colleagues at the

laboratory, who provided invaluable support concerning the experimental part. Thanks also go to

M.Tech. Kaushik Banik, M.Sc. Bin Zhang, Loic Bouteruche and Anne Hewitson, for interesting

discussions, many valuable experimental data, and great friendships.

I wish to extend my thanks to TARGOR GmbH, BUNA GmbH, BASF AG, and BAYER GmbH,

Germany for the cost free supply of materials. Financial support from the Faculty of Engineering,

Kasetsart University, Thailand is gratefully acknowledged.

Most importantly, my deepest thanks go to my parents, my sisters, and my wife for their

dedication and inspiration that enabled me to reach this milestone in my life. Thank you for

keeping me afloat when I was down and thank you, by always being there when I needed you,

for reminding me of the goodness of life.

Chemnitz, 2006 Somjate Patcharaphun

Contents

Acknowledgements

Bibliographic Description

Nomenclature V 1. Introduction 1

1.1 Conventional injection molding process 1

1.2 Two-component injection molding process 3

1.2.1 Co-injection (Sandwich) molding 3

1.2.2 Gas- and Water-assisted injection molding 5

1.2.3 Overmolding 7

1.3 Specialized injection molding techniques for enhancing

properties of thermoplastics and composites 7

1.3.1 Multiple live-feed injection molding 8

1.3.2 Push-pull processing 9

1.3.3 Sequential injection molding 10

1.4 Simulation of the injection molding and specialized processes 11

1.4.1 Simulation of the conventional injection molding process 11

1.4.2 Simulation of some specialized injection molding processes 13

1.5 Research objectives 15

1.6 Outline of the thesis 16

Contents II

2. Molding of Short Fiber Reinforced Composites 17

2.1 Rheology of short fiber composites 18

2.2 Microstructure of injection molded short fiber composites 20

2.2.1 Fiber orientation 20

2.2.2 Fiber attrition during molding 21

2.3 Mechanical properties 23

2.4 Predictive methods of tensile strength for short fiber composites 25

2.4.1 Modified rule of mixtures (MROM) 26

2.4.2 Area fraction method 28

3. Modeling of the Injection Molding Process 34

3.1 Governing equations 34

3.2 Predicting fiber orientation 37

3.2.1 Characterizing orientation 38

3.2.2 Flow-induced fiber orientation 40

3.2.3 Numerical simulation of fiber orientation for injection molding 42

4. Experimental and Simulation Procedures 43

4.1 Materials and processing conditions 43

4.1.1 Sandwich injection molding 43

4.1.2 Push-pull injection molding 46

4.2 Microstructure analyses 48

4.3.1 Skin/core material distribution 48

4.3.2 Fiber orientation analysis 49

4.3.3 Fiber length analysis (Fiber attrition) 51

4.3 Mechanical testing 52

4.4 Process simulation 53

4.4.1 Pre-processing 53

Contents

III

4.4.2 Simulation approach 55

4.4.2.1 Simulation of skin/core material distribution

in sandwich injection molding 55

4.4.2.2 Simulation of 3-D fiber orientation distribution

in sandwich and push-pull injection moldings 56

5. Experimental Results and Discussion 61

5.1 Comparison between conventional and sandwich injection moldings 61

5.1.1 Fiber orientation distribution 61

5.1.2 Fiber length distribution (Fiber attrition) 66

5.1.3 Mechanical properties 68

5.2 Comparison between conventional and push-pull injection moldings 70

5.2.1 Geometry of weldlines 70

5.2.2 Fiber orientation in weldline areas 71

5.2.3 Effects of holding pressure difference and fiber concentration

on penetration length of weldline 76

5.2.4 Fiber length distribution in weldline areas 79

5.2.5 Weldline strength 82

5.3 Prediction of tensile strength for short fiber reinforced composites 84

6. Comparison between Simulation and Experiment 89

6.1 Sandwich injection molding 89

6.1.1 Effect of skin/core volume fraction on the skin/core

material distribution 89

6.1.2 Effect of processing parameters on the skin/core material

distribution 92

6.1.2.1 Effect of skin and core melt temperatures 92

6.1.2.2 Effect of skin and core injection flow rates 94

6.1.2.3 Effect of mold temperature 98

Contents IV

6.1.3 Effect of glass fiber content on the skin/core material distribution 99

6.1.4 Case study 101

6.2 Simulation of fiber orientation in sandwich injection molding 104

6.3 Simulation of fiber orientation in push-pull injection molding 111

7. Conclusions 115

8. References 119

9. Curriculum Vitae 137

Nomenclature

Symbol Meaning Unit

----------------------------------------------------------------------------------------------------------------

CUσ Ultimate strength of the composite MPa

fσ Ultimate strength of the fiber MPa

fV Volume fraction of the fiber -

mV Volume fraction of the matrix -

mσ Stress developed in the matrix MPa

0f Fiber orientation efficiency factor -

lf Fiber length efficiency factor -

na Proportion of fibers making an angle nϕ with

respect to the applied load or flow direction -

l Fiber length mμ

cl Critical fiber length mμ

d Diameter of fiber mμ

τ Interfacial shear strength between fiber and matrix MPa

mτ Shear strength of the matrix MPa

CF Total load sustained by the composite N

LF Load carried by longitudinal fibers N

TF Load carried by transverse (or random) fibers N

C

L

AA Area fraction between the skin region and the

cross-sectional area of specimen -

Nomenclature

VI

Symbol Meaning Unit

----------------------------------------------------------------------------------------------------------------

C

T

AA Area fraction between the core region and the

cross-sectional area of specimen -

ULσ Ultimate tensile strength of the skin material MPa

UTσ Ultimate tensile strength of the core material MPa

skinf0 Fiber orientation efficiency factors for the skin layer -

coref0 Fiber orientation efficiency factors for the core layer -

skinA Cross-sectional area of skin material 2mm

coreA Cross-sectional area of core material 2mm

ρ Density 3/ mkg

P Pressure Pa

pC Specific heat at constant volume 11.. −− KkgJ

T Temperature °C

v Specific volume kgm /3

•

S Rate of heat generation due to chemical reaction 3/ mW

u Velocity vector -

g Body force vector -

q Heat flux vector -

∇ Gradient operator -

tDD Substantial derivative -

τ Extra stress tensor -

η Non-Newtonian viscosity sPa.

γ& Strain rate tensor -

k Heat conduction coefficient -

I Identity matrix -

α Compressibility coefficient -

Nomenclature

VII

Symbol Meaning Unit

----------------------------------------------------------------------------------------------------------------

β Thermal expansion coefficient -

gh Melt-mold heat transfer coefficient -

( )φθψ , Orientation distribution function -

ija Second order orientation tensor -

ijkla Fourth order orientation tensor -

λ Shape factor of particle -

er Aspect ratio of the ellipsoid -

IC Fiber interaction coefficient -

bδ Thickness fraction of the core material -

0Lxi Measured distance ratio between length of

measurement and total length of specimen -

iϕ Angle between the individual fiber and the local

flow direction °

iNϕ Number of fibers with a certain angle to the local

flow direction -

lΔ% Percent difference between the number average

fiber length inside the granules and the overall glass

fiber length inside the molded part %

Gl Average fiber length inside the granules mμ

jl Local fiber length inside the individual layers of

sectioned part mμ

1. Introduction

1.1 Conventional injection molding process

Injection molding process is one of the most widely used operations in the polymer

processing industry. It is characterized by high production rate, high automation, and accurate

dimensional precision. Products ranging from as small as plastic gears to as large as

automobile bumpers can be injection molded. Injection molding process is accomplished in

an injection molding machine (Figure 1.1) which basically consists of two essential

components; the injection unit and the clamping unit. The function of the former is to melt

the polymer and inject it into the mold cavity, whereas the clamping unit holds the mold,

opens and closes it automatically, and ejects the finished products.

Figure 1.1 Schematic drawing of a typical injection molding machine. [1]

Introduction 2

The most common type of injection molding machine is the in-line reciprocating screw type.

The screw both rotates and undergoes axial reciprocating motion. When the screw rotates, it

acts like a screw extruder, melting and pumping the polymer. When it moves axially, it acts

like an injection plunger, pushing the polymer melt into the mold cavity. The screw is

generally driven by a hydraulic motor and its axial motion is activated and controlled by

hydraulic system. The raw material is supplied to the injection molding machine through the

feed hopper, which is located on top of the injection unit. The screw takes in the material and

conveys it to the screw tip. On its way, the plastic passes through heated barrel zones, while

the rotation of the screw results in a continuous rearrangement of the plastic material in the

flights of the screw. Shear and heating from the barrel wall cause a largely homogeneous

heating of the material. The conveying action of the screw builds up the pressure in front of

the tip. This pressure pushes back the screw. As soon as there is enough supply of melt in

front of the screw, the screw moves forward to inject the molten material into the mold cavity.

The injection molding process can be subdivided into four stages: (a) injection, (b) packing,

(c) cooling, and (d) ejection. The cycle begins when the mold closes, followed by the

injection of the polymer into the mold cavity. Once the cavity is completely filled, a holding

pressure is maintained to compensate for material shrinkage. As soon as the gate is

completely frozen, no more material can be injected and, the packing pressure is released and

the screw turns, feeding the next shot to the front of the screw. When the part is sufficiently

cool, the mold opens and the part can be taken out for further cooling to the ambient

temperature.

Introduction

3

1.2 Two-component injection molding processes

During the past two decades, numerous attempts have been made to develop injection

molding process to produce products with special design features and properties. Two

component injection molding being an alternative process derived from conventional

injection molding has created a new era for additional applications, more design freedom, and

special structural features. These efforts have resulted in a number of processes, including:

• Co-injection (Sandwich) molding

• Gas- and Water-assisted injection molding (GAIM and WAIM)

• Overmolding

• Further two-component injection methods e.g. Insert molding and Rotating mold

techniques are beyond the scope of this section

1.2.1 Co-injection (Sandwich) molding

Sandwich injection molding is an extension of the standard injection molding technology

which allows for two components to be sequentially injected into the mold in order to

fabricate products with a layered structure. This processing technology was first invented by

Garner and Oxley of ICI [2]. Figure 1.2 shows a schematic principle of the sandwich

injection molding process. The formation of the skin and the core structure can be explained

by the molding process. A given percentage of the skin material is first injected into the

cavity to form the skin layer. As the fastest material in the center of the flow reaches the flow

front, it splits to the outer wall of the mold and freezes forming a frozen layer or skin layer.

This is called “Fountain flow” as schematically illustrated in Figure 1.3. Prior to the skin

material’s reaching the end of the cavity, the second material is injected to form the core.

This core material develops a second flow front pushing the skin material ahead of it until the

cavity is nearly filled and finally a much smaller amount of the skin material is injected to

seal the gate. The last injection of skin material is important to clean all core material out of

the gate area and ensure that no core material will be injected into the next part during the

initial skin material injection.

Introduction 4

A

B B

B

A

A

B

(1) (2)

(3) (4)

A

Figure 1.2 The sandwich injection molding process works by first injecting the skin material

(1, 2) then switching to the core material (3). A small amount of skin material can seal the

gate to purge the core material away from the sprue (4).

Mold Wall

Fountain Flow

Skin MaterialCore Material

Solidified Skin Layer


Wal

l Thi

ckne

ss

Melt Front

Flow Direction

Figure 1.3 Schematic of polymer melt flow profile across the thickness during sandwich

injection molding process.

Introduction

5

The resulting skin/core geometry of sandwich molded parts provides a number of advantages

because the different material properties can be incorporated into the same part, as

demonstrated in Figure 1.4. It is often desirable for the skin material to have a superior

appearance, while the strength and rigidity of the part is strongly dependent upon the core

material [3-5]. Typically, the core material will be less costly than the skin material, which

can yield potential cost savings. This is often achieved by using recycled material as the core.

Sandwich injection also exploits to use a foam core. In this case, large parts with hard and

glossy surfaces can be molded without the need for high clamping forces since shrinkage is

compensated by the expansion of the core material. In applications, where thin-walled

products are more suitable, low weight, high stiffness products with reinforcing ribs can be

molded economically [6-7].

Skin material

Core material Skin material

Core material

Figure 1.4 Sandwich injection moldings. [5]

1.2.2 Gas and Water-assisted injection molding

Gas-assisted injection molding (GAIM) is an important variant of the traditional technology

for injection molding of thermoplastics. In the simplest terms, gas-assisted molding process

begins like any conventional injection molding process with the injection of polymer melt

into a cavity. Only a partial volume of melt is injected and a short shot is purposely produced

(see Figure 1.5). At the end of the polymer injection stage, compressed gas, usually nitrogen

(due to its relative inertness and availability) is injected through the central core of the melt

similar to sandwich injection molding. The gas drives the molten polymer further into the

mold, until it is filled completely. The penetrating gas, acting now as the core material, leaves

Introduction 6

a polymer layer at the mold wall, yielding a product with a polymer skin and a hollow core.

The gas can either be injected through a needle in the nozzle, or directly into the mold

through separate gas injection needles. After the mold has been entirely filled, gas is used to

transmit the packing pressure to the polymer that is being cooled. Any shrinkage of the

polymer material near the gas channel is compensated for by an enlargement of the gas core.

Once all polymer material has solidified, the gas pressure is released. The product is then

further cooled until it has retained sufficient rigidity to be ejected from the mold. The most

important characteristic of GAIM is the fact that the pressure drop in the gas core is

negligibly small compared to the pressure drop in an equivalent molten polymer.

Consequently, the pressure can be considered constant throughout the gas core, which

accounts for most of the advantages of GAIM, such as reduction of raw material, weight of

product, cycle time, clamping force, sink marks and residual stresses, and enhancement of

design possibilities [1, 8-10].

P = 0P = 0

(1) (2)(2)

(3) (4)

Vented cavityVented cavity

Figure 1.5 Schematic showing the various stages of the gas assisted injection molding process:

(1) Melt injection; (2) Gas injection; (3) Packing phase; (4) Part ejection. [8]

Water-assisted injection molding (WAIM) appears at the beginning of the 70s but its real

development started at the Institute für Kunststoffverarbeitung (IKV), a plastic processing

development center in Germany, in 1998. This process is similar to GAIM except that it uses

water instead of nitrogen. The aim of developing WAIM is to reduce cooling cycle times in

the production of hollow or partly hollow parts [11].

Introduction

7

1.2.3 Overmolding

The overmolding process is a versatile and increasingly popular injection molding process

that provides increased design flexibility for making multi-color or multi-functional products

at reduced cost. This technique permits one-step joining of two or more polymers (e.g.

rigid/flexible or rigid/rigid) into parts, which do not require any further finishing operations

[12-13]. Besides the economical advantages, the process offers the possibility of obtaining a

broad range of mechanical properties of the end products [14-15]. The typical applications of

the overmolding process are the combination of multi-colored areas within one part and the

soft-touch applications (e.g. handles). For instance, two-component plastic parts can be

produced by a two-component injection molding machine, which introduces sequentially

different polymers into a special mold through separate runner systems. After molding the

preform of the first component, the cavity part for the second component is activated by

removing a metal core (core back mold) or opening the mold and transporting the preform

into a second cavity (e.g. rotating mold base). The second polymer is then delivered by the

second injection unit into the newly formed cavity through its independent runner system and

the final part is ejected after packing and cooling phases. This method is sometimes referred

to as in-mold assembly, since the resulting part effectively acts as an assembly of two

materials rather than as a layered structure.

1.3 Specialized injection molding techniques for enhancing properties of

thermoplastics and composites

Defects such as weldlines, sink marks, and warpage are caused by melt fronts collision,

unbalanced flow, uneven cooling and non-uniform internal stress. Varying the processing

parameters can result in the modification of the molded part outlook, physical and mechanical

properties [1, 8, 16]. The modifications, however, are often slight and not quantified, and they

also rely upon the expertise of the operator who uses his experience and art to determine the

processing parameters. During the last decade, several techniques have been developed using

different approaches in order to improve the molding properties, e.g. weldline strength, by

controlling the melt flow pattern of the polymer as it is being shaped [17-19]. This concept

has been applied to a wide range of thermoplastic matrix composites especially with glass-

Introduction 8

fiber reinforced thermoplastics [20-25]. There have been many such improvements, but three

in particular stand out. The first, multiple live-feed injection molding, auxiliary equipment is

incorporated into the standard molding machine. For two others, push-pull processing and

sequential injection molding, require special molding machine and modified tooling for

optimal success.

1.3.1 Multiple live-feed injection molding

The multiple live-feed injection molding process, is also known as Shear Controlled

Orientation Injection Molding (SCORIM), developed at Brunel University, and licensed by

British Technology Group [26]. This process achieves significant improvement and control

over part properties by using a special injection head that splits the melt flow in the mold into

two streams (see Figure 1.6). Once the mold is filled or during the packing stage, the multi-

live feed system’s hydraulic pistons begin moving forward and backward in an alternating

fashion. As one live feed piston pushes downward, it forces melt through the runner and

cavity up into the second live feed cylinder. The process then reverses, and the melt flows in

the opposite direction. The principle advantages of the process are; enhanced and controlled

orientation of fiber or flake fillers, significant reduction of weldline effects and controlled

modification of the microstructure of injection molded unfilled plastics, especially in liquid

crystal polymers (LCPs) [20-21, 27].

Wel

dlin

eW

eldl

ine

Conventional injection unit

Hydraulic cylinders and pistons

Runner system

Multiple live-feedprocessing head

Figure 1.6 Schematic of the multiple live-feed injection molding process. [8]

Introduction

9

1.3.2 Push-pull processing

The push-pull injection molding process is a melt oscillation technique, which is very similar

to SCROIM. It was originally unveiled by Klöckner Ferromatik Desma at the K’89 show. As

shown in Figure 1.7, the push-pull injection molding system includes two injection units and

a two gate mold. The cavity is firstly filled simultaneously by the melt from both the units via

the two separate gates. After the two melt fronts meet, the weldline is formed and the filling

phase is subsequently switched to the holding phase. The material solidifies starting at the

cavity wall but there is still molten core and then the first push-pull stroke begins. The control

software program allows the definition of several holding pressures for one stroke from either

the first or the second injection unit. From one of the injection units, polymer melt is pressed

into the cavity resulting in the molten core being pushed through the gate back into the other

injection unit and thus the geometry of weldline is deformed to a tongue shape. As the

material flows back and forth through the mold, molecular orientation is continuously created

and subsequently frozen in as the material solidifies from the outer layers toward the hot core.

By keeping the molten polymer in laminar motion during solidification, the molded parts

acquire an oriented structure throughout the volume. If the mold is complex and the melt has

to flow around obstacles, the motion will create better mixing in the area behind the obstacles

and reduce the weakening effect of the weldline by dispersing them throughout the part and

eliminates void, cracks, and micro-porosities in large cross-section molding. The number of

strokes can be selected by taking into account the part’s thickness. When all the strokes are

completed, cooling phase follows. As the thickness of frozen layer increases with the number

of strokes within the holding time, the total cycle time is not notably increased as compared

to conventional injection molding [23].

Secondary injection unitPrimary injection unit

Primary runner Overflow runner

Multigated mold

Figure 1.7 Schematic principle of push-pull injection molding process. [8]

Introduction 10

1.3.3 Sequential injection molding

Sequential injection molding is an increasingly used manufacturing processing technique

presenting the advantages over traditional injection molding. The process is generally used in

large parts, which are difficult to pack from one central area. Sequential valve gating is used

to control the filling of parts and each valve gate is independently opened and closed at a pre-

determined event (time, screw position, cavity pressure, etc.) providing complete control of

cavity fill. This technique can minimize the pressure loss in the system and also can be used

to control the location of weldline, as illustrated in Figure 1.8, in order to ensure that the

weld is positioned away from the critical area, and thus improving the product’s performance

[24, 25].

Resultant Part

a) Classical Weldline

b) Middle Disturbance

c) Side Disturbance

Figure 1.8 Schematic illustration of sequential injection molding (experimental mold,

developed at TU Chemnitz): Filling study by sequentially opening and closing the valve gates.

[24, 25]

Introduction

11

1.4 Simulation of the injection molding and specialized processes

1.4.1 Simulation of the conventional injection molding process

There are several milestones in the history of Computer-Aided Engineering (CAE). The

analysis of mold filling in injection molding started with the work of Spencer and Gilmore

[28] in the early 1950’s. They employed an empirical equation for capillary flow and coupled

it with a quasi steady-state approximation to calculate the filling time. Since then, different

methods have been proposed to describe the molding cycle with varying degrees of

complexity. One-dimensional rectangular flow was proposed by Ballman et al. [29] and

Staub [30]. Harry and Parrot [31] considered a one-dimensional quasi-steady state flow

analysis coupled with an energy balance equation. Williams and Lord [32] made a significant

contribution by considering all the components of a one-dimensional non-isothermal flow. A

similar model was presented by Thienel and Menges [33] using a different solution technique.

In order to study a more representative one-dimensional flow, a number of analyses were

carried out on the radial filling of a center gated disc mold. Kamal and Kenig [34-35]

proposed an integrated mathematical treatment of the filling, packing, and cooling stages of

the injection molding cycle. Similar simulations were carried out by Berger and Gogos [36],

and Wu et al. [37]. However, it was not until the 1970’s when the development and

application of computer simulations to injection molding intensified. In particular, Stevenson

and co-workers [38] analyzed one-dimensional flow in a center-gated disc. Lord and

Williams [39] studied the one-dimensional filling behavior in rectangular cavity geometry.

Nunn and Fenner [40] modeled one-dimensional tubular flow of polymer melts, which was

later extended by Hieber et al. [41] to simulate the polymer flow in a non-circular tube under

non-isothermal condition.

The general characteristics of injection molding are that the part thickness is much smaller

than the overall part dimension and the polymer melts are highly viscous due to their long

molecular chain structure. As a result, the ratio of inertia force to the viscous forces (as

characterized by the dimensionless Reynolds number) is in the order of 10-3 -10-4. This makes

the Hele-Shaw flow formulation [42], which is based on the creeping-flow lubrication model,

an appropriate candidate for analyzing the flow in typical injection molded parts. In addition

to neglecting the fluid inertia, the Hele-Shaw flow formulation also omits calculation of the

Introduction 12

velocity component and thermal convection in the gapwise direction. Compared with heat

conduction in the gapwise direction, heat conduction in the planar directions is also neglected.

Other commonly adopted simplifications include neglecting the transverse flow at the melt

front region (the fountain flow behavior), viscous convection (drag force) and heat

conduction on the lateral wall surfaces, and mapping of gapwise solutions at the flow

junctions and where the wall thickness changes. Accordingly, the usage of computational

resources including computational storage and CPU time can be reduced considerably

compared with the case of a full three-dimensional simulation. In this approach, three-

dimensional geometry is represented with one-dimensional tubular elements and two-

dimensional triangular thin-shell elements for which the wall thickness is implicitly specified

as an attribute; i.e. a “mid-plane” mesh has to be created either from collapsed or from an

existing three-dimensional CAD design model. Those one- and two-dimensional elements are

numerically divided into several “layers” (typically 8-20) in the gapwise direction for details

of the variables under consideration. While the governing equations of mold filling and

packing are being solved by the finite-element method (FEM), finite-difference method

(FDM) is applied in the gapwise direction and the temporal domain. By doing so, the

transient behavior and variation of the variables in the gapwise direction can be captured.

Since the gapwise velocity component is not calculated and the mesh model only represents

the “shape” of the part geometry, the Hele-Shaw flow formulation is sometimes called 2.5-

dimensional (2.5-D) simulation. Although the governing equations and the geometry are

simplified, the Hele-Shaw flow model became the standard numerical framework for various

commercial software packages and research codes [43-45] and has been extended or

incorporated by other researchers [46-49] e.g. simulation of polymer melt flow during the

filling and packing phase, fiber orientation, shrinkage and warpage. However, the Hele-Shaw

flow formulation has its limitations owing to the inherent creeping-flow and thin-wall

assumptions. For example, the shell element employed in the Hele-Shaw model needs the

construction of the mid-plane, which is time-consuming [50]. Furthermore, it cannot

accurately model the three-dimensional flow behaviors, particularly important when molding

with fiber reinforced systems [51], within thick and complex geometries or at the melt fronts

(fountain flows), regions where the part thickness changes abruptly or separate melt fronts

meet (weldlines), and regions around special part features such as bosses, corners, and/or ribs

as compared to those obtained by the three-dimensional (3-D) simulation model [50, 52-54].

Introduction

13

The interest in 3-D simulation of injection molding has increased tremendously in the past

few years. Several commercial and research-oriented 3-D CAE simulation programs for

injection molding have been developed [52-53]. In particular, Hetu et al. [52] developed a 3-

D finite-element program for predicting the velocity and pressure fields governed by

generalized Stokes equations. In addition to the temperature field, they also solved the

position of flow fronts using the pseudo-concentration method. Zachert and Michaeli [53]

analyzed polymer flow at the region of a sudden thickness change during injection molding

using both a Hele-Shaw flow formulation and a 3-D approach. Chang and Yang [54]

developed the numerical simulation for 3-D mold filling based on an implicit finite-volume

method (FVM). Their work was later commercialized and extended to cover various stages in

injection molding and special molding processes. Pichelin and Coupez [55] analyzed the 3-D

mold filling of an incompressible fluid and the shape of the fountain flow front using an

implicit discontinuous Taylor-Galerkin scheme. Han et al. [56] predicted the fluid flow

advancements and pressure variation in the microchip encapsulation process using a 3-D

FEM based on a generalized Hele-Shaw formulation. By treating the polymer density as a

function of pressure and temperature, Haagh et al. [57] incorporated the compressibility of

the polymer melt in a 3-D mold filling process. Rajupalem et al. [58] and Talwar et al. [59]

used an equal-order velocity-pressure formulation to solve the Navier-Stokes equations in

their 3-D simulation of mold filling/packing phases.

1.4.2 Simulation of some specialized injection molding processes

Over recent years, as for the conventional injection molding, the numerical simulations of co-

injection, gas-assisted injection, SCORIM, and push-pull processing are mostly based on the

thin wall, Hele-Shaw approximation. Simulation of the sequential sandwich injection

molding process was first carried out by Turng and Wang [60] in order to predict the skin and

core melt front progression and the distribution of the two layers by calculating the residence

time of the particles that enter the mold cavity. Schlatter et al. [61-62] used a special transport

equation to characterize the displacement of the interface between skin and core for the

sequential injection of polymers. Visualization and simulation of the sandwich mold filling

process have been presented by Lee et al. [63-64]. They developed a simulation approach

based on the Hele-Shaw approximation and kinematics of interface to calculate the two-phase

flow and the interface evolution during filling in simultaneous sandwich molding. Jaroschek

[65] studied the distribution of the core material during the filling process of sandwich

Introduction 14

injection molding using multiple gates, a variation of cascade control, with specific valves

and standard hot runner manifolds. The experimental results were also compared with those

obtained by the Moldflow simulation package using a layered 2.5-D flow approach and FEM

grid representing the housing geometry (loudspeaker box). His findings suggest that it is

possible to use a double hot runner system that allows even large components to be produced

by sandwich molding using multiple gating in case where the simulation program provides

sufficient accuracy. Chen et al. [66-67] utilized an algorithm based on the control volume

finite-element method combined with a particle-tracing scheme using a dual-filling-parameter

technique to predict the advancements of both melt front and gas front during the gas-assisted

injection process. Gao et al. [68] also used this volume tracking technique with the Galerkin

Finite Element model to simulate the filling stage of the gas-assisted injection molding

process, particularly the gas penetration phenomenon involving the gas-polymer interaction.

Wang et al. [69] compared the experimental and simulation results of gas penetration in terms

of different shot sizes, delay time and gas pressure. They suggested that an improper

modeling can cause artificial fast cooling in the gas channels, which will hinder the gas

penetration in the numerical simulation. Pittman et al. [70] simulated the cooling and

solidification of polymer melt during SCORIM by using the one-dimensional transient model.

A non-Newtonian, temperature-dependent viscosity is used, together with temperature-

dependent thermal properties and latent heat of solidification. Recent work [71] investigates

the weldline strength and the fiber orientation in the weldline region of push-pull processed

parts, with respect to the number of push-pull strokes and the holding pressure differences

between both the injection units. The experimental results are also compared with those

obtained by the simulation package using a 2.5-D model (based on Hele-Shaw

approximation). A good agreement has been obtained between the predictions and the

measurements, thus showing the usefulness of the commercial software in helping the design

engineer to identify the location of weldline and fiber orientation state within the weldline

areas.

Initial work on the 3-D simulation of the gas-assisted injection molding process was done by

Khayat et al. [72] which used a boundary-element method (BEM). Their contribution reduces

to the analysis of isothermal, incompressible, Newtonian fluid flow in simple 3-D geometries.

Haagh et al. [73] presents a 3-D model based on the finite-element method and a pseudo-

concentration technique for tracking the flow interfaces. Ilinca et al. [74] used a pressure

stabilized Petrov-Galerkin method to solve the Navier-Strokes equations in their 3-D

Introduction

15

numerical model for gas-assisted injection molding. An additional pressure stabilization term

was included compared with the standard Galerkin method. The position of the polymer/air

interfaces was also tracked using the pseudo-concentration method. Ilinca et al. [75] also used

this numerical model to simulate 3-D numerical model to simulate 3-D co-injection molding.

The polymer/air and skin/core polymer interfaces were tracked by solving two additional

transport equations.

1.5 Research objectives

The main objective of this work is to investigate the capability of the sandwich injection

molding technique for enhancing the orientation of fibers within the molded parts. The

influences of glass fiber concentration and processing parameters on the material distribution,

fiber orientation and fiber attrition are examined. Additionally, one of the specialized

injection molding techniques – push-pull processing – is employed in order to improve the

fiber orientation within the weldline area. The effect of processing parameters including the

number of push-pull strokes and the holding pressure differences between both the injection

units have been studied. The degradation of the fiber length caused by the alternating shear

field is also investigated.

The prediction of the tensile strength of conventional, sandwich and push-pull injection

molded short fiber reinforced composites are derived by an analytical method of modified

rule of mixtures (MROM) as a function of the area fraction between skin and core layers. The

effects of fiber length and fiber orientation on the tensile strength of short fiber reinforced

composites are also studied in detail. This model provides the necessary information to

determine what fiber length distribution and what fiber orientation distribution are required to

achieve a desired composite strength.

Material distribution and fiber orientation structures of sandwich and push-pull processed

parts are predicted by the 2.5 and 3-D numerical analyses. The predictions solve the full

balance equations of mass, momentum, and energy for a generalized Newtonian fluid. The

second-order orientation tensor ( ) approach is used to describe and calculate the local

fiber orientation state. The accuracy of model predictions is extensively evaluated by

11a

Introduction 16

comparing with corresponding experimental measurements to gain a further basic

understanding of the relationship between the processing conditions, the fiber orientation

distribution and the properties of the final injection molded part.

1.6 Outline of the thesis

In Chapter 2 the fundamentals of rheology, general behavior and predictive methods for short

fiber reinforced composites are introduced. The influence of parameters on mechanical and

physical properties is described. The mathematical formulations used for the flow model and

the basics of fiber orientation prediction are summarized in Chapter 3. Chapter 4 details the

experimental procedures including materials, processing conditions, morphology observation

and measurement of mechanical properties. The process simulations of sandwich and push-

pull injection molding processes are also presented. In Chapter 5, the morphology

developments with different processing types are discussed. It provides the evidences of

mechanical properties, fiber orientation, and fiber length distribution within sandwich and

push-pull injection molded parts compared with those obtained using conventional injection

molding. In Chapter 6, predictions of the skin/core material distribution and fiber orientation

are compared with the experimental results. A number of simulations of sandwich and push-

pull injection molding in a dumbbell part are carried out in order to investigate the influence

of various parameters. Finally, the conclusions are presented in Chapter 7.

2. Molding of Short Fiber Reinforced

Composites

A short fiber composite consists of a polymer matrix reinforced by fibers of much smaller

length as compared with the overall dimensions of the fabricated structure. These reinforced

polymers have been developed to fill the mechanical property gap between the continuous

fiber laminates used as primary structures by the aircraft and aerospace industry and the neat

polymers used in low-load-bearing applications. Although short fiber composites do not

achieve the characteristic mechanical values which can be obtained with continuous fiber

laminates [76]. However, they can be processed with the same techniques used for unfilled

thermoplastics, e.g. injection molding of short fiber composites for the high volume

production purposes and the ability to be molded into complex shapes. Furthermore, their

intrinsic recyclability is rapidly being recognized as a strong driving force for their further

application [77]. The molding processes of short fiber composites requires the compounded

material to be heated and then forced to flow under the application of a high pressure in order

to conform to the shape of the mold cavity. If injection molding is the chosen fabrication

route, then the flow processes involved in mold filling can be very complex and result in a

marked orientation of the fibers in the final part. The microstructure and morphology that are

produced depend on how well the fibers are dispersed, the range of fiber lengths and

diameters, how the fibers interact with the mold walls and each other during flow, the heat

transfer in the mold, and the geometry of the mold. This in turn will have a major effect on

the anisotropy of the mechanical and physical properties of the injection molded component

[76].

Molding of Short Fiber Reinforced Composites

18

This chapter will introduce a short review of rheological properties and microstructure of

short fiber reinforced composites including the fiber attrition during processing operations.

Topics also include the mechanical properties and the predictive methods for short fiber

reinforced composites.

2.1 Rheology of short fiber composites

The rheological properties of short fiber composites may differ in detail from those of normal

unfilled polymers, but they are not considerably different. In practice the methodology for

measuring the basic flow properties of fiber filled melts follows closely that used for

characterizing the flow properties of unfilled materials. Indeed, the shear viscosity can be

measured using most of the normal techniques, i.e. cone and plate, capillary, dynamic

mechanical, etc. Detailed experimental techniques used for characterizing rheological

properties can be found elsewhere [78].

Much of the early work on the rheology of filled polymers was concerned with dilute

suspensions. The flow properties of such melts will differ significantly from those of filled

thermoplastics of commercial grades, where individual particles are in close proximity to

each other. Figure 2.1 shows a plot of log shear viscosity versus log shear rate for three

grades of polycarbonate, containing 0, 20 and 35 by weight of short glass fibers (%wt),

respectively. At low shear rates the presence of the fibers causes an appreciable increase in

viscosity, but it is noteworthy that the viscosity values for the three materials converge to a

very similar value at the higher shear rates. Essentially the same pattern of behavior is shown

by many other fiber reinforced grades of polymers [79]. The similarity in viscosity values at

high shear rates, for filled and unfilled polymers, is an important factor in explaining the

successful exploitation of these materials, since very little additional power will be required

to process the filled materials [80].

The basic rheological data for fiber filled polymers are derived from tests performed using

very simple geometries. In practice, however, the flow geometries occurring in technological

processing equipment will be very complex. The fiber orientation distribution (FOD) in a

molded part can be qualitatively interpreted through an appreciation of the fiber orientation

resulting from certain basic categories of flow:


19

• Shear flow – as will occur in a straight tube or duct

• Convergent flow – in the simplest case this will occur when a fluid passes from a

wide to a narrow cross-section.

• Divergent flow (also referred to as extensional flow) – in the simplest case this will

occur when a fluid passes from a narrow to a wide cross-section.

Convergent flow leads to an alignment of fibers parallel to the flow direction. Divergent flow,

on the other hand, tends to orient the fibers orthogonally to the flow direction. Frequently,

this is observed as fiber filled melts pass from a narrow gate into the mold cavity.

1 10 100 1000 10000 10000010

100

1000

10000

Temp. = 306.7 oC

Visc

osity

(Pa.

s)

Shear Rate (1/s)

PC (Makrolon 2800) PC filled with 20 wt% short-glass-fiber (Makrolon 8020) PC filled with 35 wt% short-glass-fiber (Makrolon 8030)

Figure 2.1 Apparent viscosity versus shear rate for three polycarbonates containing different

amounts of short glass fibers. (Moldflow database)


20

2.2 Microstructure of injection molded short fiber composites

2.2.1 Fiber orientation

As the fiber suspension flows into the mold, one must know the fiber orientation structure not

only to determine its rheological behavior but also to estimate its mechanical performance.

Flow and deformation of the suspension change the orientation of the fibers flowing in it.

These orientations are subsequently frozen in as the material solidifies and become a key

feature of the microstructure of the finished composite. If the fibers are randomly oriented,

the mechanical and physical properties will be isotropic. If the fibers are aligned in one

direction, the composite will be stiffer and stronger in that direction compared with any other

direction. The distribution of orientations in a molded part could be quite diverse. One region

may have random fiber orientation, while others may have preferred alignment in certain

directions.

In the injection molded short fiber composites, a characteristic layer structure is observed,

with the fibers oriented in quite different manners according to their location through the

thickness, as schematically demonstrated in Figure 2.2. These general features are apparent in

studies of fiber orientation distribution found in the literature [80-89]. In the skin region, the

fiber orientation is predominately parallel to the flow direction. This is due to, as the melt fills

the mold, there is fountain flow which initially orients the fibers perpendicular to the main

flow direction. Fountain flow causes the melt to be deposited on the mold wall with the

alignment direction parallel to the mold fill direction. Here it solidifies rapidly and this

alignment is retained in the solid article. Further behind the melt front, shear flow dominates

and produces fairly uniform levels of fiber alignment. The fiber orientation and the thickness

of this region are influenced by non-isothermal effects and by injection speed [80-81]. In

contrast, the core of the molding contains fibers mainly aligned perpendicular to the flow

direction due to a slower cooling rate and lower shearing.


21

Core layer

Skin layer

Skin layer

Z

X Y (Flow direction)

Figure 2.2 Schematic diagram of molding indicating the fiber orientation in the skin and core

layers.

During the injection molding process of short fiber composites, the distribution of fiber

orientation is governed by a variety of factors. These include the type and/or shape [82, 90],

concentration of fibers [83-84], the gate designs and/or flow geometry [85-86]. The

processing conditions such as injection flow rate, injection pressure, and melt temperature

can also significantly alter the proportions of the oriented regions [87-88]. It has also been

shown that the mechanical and physical properties of injection molded short fiber composites

depend critically on the fiber orientation distribution in the final product [82-89].

2.2.2 Fiber attrition during molding

One of the major concerns in producing fiber reinforced parts by injection molding is fiber

breakage during processing. Many experiments have established the fact that the fibers get

damaged during processing and fiber length may be reduced by an order of magnitude [91-

93]. This reduction in length can potentially reduce the reinforcing efficiency of the fibers,

thus substantially reducing the mechanical properties of the composite [94-95]. The reduction


22

in fiber length during compounding and molding can be explained by the three common

mechanisms including; fiber-flow, fiber-fiber and fiber-wall interactions.

• Fiber-flow interactions: most flows in injection molding process are a combination of

extensional and shear deformations. In a purely elongational flow, fibers tend to align along

the stretching direction and hence are under tension, fibers are rather unlikely to break under

this mode. However, in shear flows, fibers rotate across the streamlines and may bend to their

critical radius of curvature and buckle under viscous forces transmitted by the polymer melt.

• Fiber-fiber interactions in concentrated suspensions can cause fiber overlaps, which

will induce bending stresses in the fibers, resulting in breakage. The effect of fiber volume

fraction on breakage was studied by von Turkovic and Erwin [91], who found that glass

fibers in polystyrene had the identical length distribution at the exit of an extruder for volume

fractions ranging from 1 to 20%. Their results indicated that the fiber length was reduced by

an order of magnitude after processing and that the average final length was insensitive to

initial fiber length distribution.

• Fiber-wall interactions: in injection molding, the effect of mold and screw geometries

plays a crucial role in fiber length reduction. Wider mold channels have been shown to

reduce breakage [84-86]. Also, the gate region of the mold may be a key factor in fiber

degradation. Bailey and Rzepka [96] studied long fiber materials under various material and

processing conditions in a plunger molding machine, a conventional injection molding

machine, and an extruder. Mold configurations with a small gate and a generous gate were

studied. Fiber loadings from 30-60% were used. Their study showed that a larger gate

produced final fiber lengths with a mean of 0.99 mm, whereas a smaller gate had a mean

fiber length of only 0.49 mm. They also found that there was substantially more damage in

the skin region than in the core. This could be attributed to a high shear rate near the mold

surface coupled with fiber interactions with the mold wall.


23

2.3 Mechanical properties

The general stress-strain curves observed in glass fiber reinforced thermoplastics are

illustrated in Figure 2.3. It is well known that the addition of glass fibers results in an

enhancement of the stiffness and strength of the composite [82-85, 89, 94-95]. It also can be

seen that both the stiffness and strength increase with fiber concentration. However, one very

important consequence of utilizing large volume fractions of fibers is that although stiffness

is increased significantly, there is not necessarily a pro rata change in strength [85, 89, 95].

Furthermore, the work to fracture decreases rapidly as the concentration of fibers is increased

[82-83, 85], i.e. the composite will only tolerate small impact energies.

As with all materials that are uniaxially oriented, the mechanical properties of a highly

aligned composite depend critically on the angle between the applied stress and fiber

orientation direction. Figure 2.3 also shows some experimental tensile strength data obtained

on moderately well-aligned composites of glass fiber in polypropylene. It can be seen that a

large composite strength is only obtained when the stress direction is close to the fiber

orientation axis. This sensitivity to angle becomes much more acute as the ratio of fiber

strength to matrix strength increases. The anisotropy in the mechanical properties has

important implications for the behavior of partially aligned short fiber composites. A cross-

section cut from a simple tensile test specimen, molded in short glass fiber reinforced

polycarbonate, reveals a complex fiber orientation distribution (see Figure 2.4). It is clear that

when a stress is applied to this bar many of the fibers will be at quite large angles with

respect to the stress direction. Overall, then the stiffness of the bar will be significantly less

than for fully aligned fibers.


24

Te

nsile

stre

ss,

(M

Pa)

Strain, (%)

0

10

20

30

40

50

60

70

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5

SFRPP30 parallel to flow directionSFRPP30 transverse to flow directionSFRPP40 parallel to flow directionSFRPP40 transverse to flow directionLFRPP30 parallel to flow directionLFRPP30 transverse to flow directionLFRPP40 parallel to flow directionLFRPP40 transverse to flow direction

ε

σ

Flowdirection

Figure 2.3 Typical stress-strain curves for glass fiber reinforced polypropylene at various

fiber volume fractions, fiber lengths, and testing locations. [89] (The description of

abbreviations is shown in section 4.1)

Another parameter of key importance in influencing the mechanical properties of composites

is the fiber length. As can be seen from Figure 2.3, when the stress is applied parallel to the

fiber axis in a uniaxially aligned fiber reinforced composite, the greatest stiffness and strength

occur when the fibers are very long compared with their diameter. With short fiber

composites, however, the fibers are usually of the order of half a millimeters in length so as

to enable the composites to be processed easily using, e.g. injection molding. In this case the

stiffness and strength of the composite will be lower [86, 89, 95]. Hence the strength of any

given injection molded composite part will depend critically on both the fiber length

distribution and fiber orientation distribution.


25

Fiber orientationparallel to theflow direction

Fiber orientationparallel to theflow direction

Random-in-planealignment of fibers

FlowFlow directiondirection

Skin

laye

rC

ore

laye

rSk

in la

yer

Figure 2.4 Optical photomicrograph showing the fiber orientation pattern across the thickness

of tensile test bar.

2.4 Predictive methods of tensile strength for short fiber composites

Tensile strength is one important property of engineering materials. One of the basic

motivations for the use of composite materials as engineering materials is the high tensile

strength that can be achieved by incorporating high strength fibers into a matrix since the

fibers carry most of the load. Over the last decade, several theoretical models have been

proposed in order to predict the modulus and strength of short fiber composites. One is the

laminate analogy, which combines the micro-mechanics of joining different phases with the

macro-mechanics of lamination theory. The success of the laminate approximation is strongly

dependent upon the assumption of physical volume averaging combined with an ability to

estimate the properties of the individual plies, each of which contains uniaxially oriented

fibers. This approach has been used successfully to predict strength, modulus, stress-strain

behavior [97-98], and flexural stiffness [99]. The other major approach is the modified rule of


26

mixtures (MROM), which has been mostly used to predict the modulus and strength of short

fiber composites by taking into consideration the effects of fiber length and orientation

distribution [100-103]. In general, all of the proposed methods have shown good agreement

with experimental results. Although the laminate and MROM methods are usually used to

estimate the strength for short fiber composites, the procedures to estimate the strength of

sandwich and push-pull injection molded parts have not been established.

In this section, therefore, the model used for predicting the ultimate tensile strength (UTS) of

sandwich and push-pull injection molded part will be introduced. This predictive method is

also based on a MROM as a function of the area fraction between skin and core layers (so

called area fraction method). The advantage of this method over the traditional method is that

the weldline strength of push-pull processed part and the UTS of sandwich injection molded

part, containing different fiber concentration between skin and core material, being able to

estimate. In order to take into account the influence of fiber length as well as fiber orientation,

the fiber orientation efficiency factor ( ) and fiber length efficiency factor ( ) also can be

accommodated.

0f lf

2.4.1 Modified rule of mixtures (MROM)

The modified rule of mixtures is often used to predict the tensile strength of short fiber

composites. The formula of MROM is given by

mmfflCU VVff σσσ += 0 (2.1)

where CUσ and fσ are the ultimate strength of the composite and fiber, respectively; and

denote the volume fraction of the fiber and matrix;

fV

mV mσ is the stress developed in the

matrix; and are the fiber orientation and fiber length efficiency factors, which depend

on various parameters such as fiber volume fraction and processing conditions, and are only

fitted empirically [95]. By using the Voigt average [79] and dividing the reinforcement into

groups of uniaxially aligned fibers, is determined by

0f lf

0f

∑=n

nnaf ϕ40 cos (2.2)


27

where is the proportion of fibers making an angle na nϕ with respect to the applied load or

flow direction. The efficiency of fiber reinforcement for several situations is presented in

Table 2.1, this efficiency is taken to be unity for an oriented fiber composite in the alignment

direction, and zero perpendicular to it.

Table 2.1 Reinforcement efficiency of fiber reinforced composites for several fiber

orientations and at various directions of stress application. [79]

Reinforcement Efficiency, (f 0 )

Parallel to fibers 1Perpendicular to fibers 0

Fiber randomly and uniformly Any direction in the plane tributed within a specific plane of the fibers

ber randomly and uniformlytributed within three-

imensions in space

3/8

Any direction 1/5

Stress DirectionFiber Orientation

ibers parallel

dis

Fidis d

All F

If the fiber length ( is uniform, the fiber length efficiency factor can be obtained from )l

cl l

lf2

= for l < (2.3) cl

ll

f cl 2

1−= for l ≥ (2.4) cl

where the critical minimum fiber length. This critical length is given by cl

τ

σ2

dl f

c = (2.5)

where is the fiber diameter and d τ the interfacial shear strength between fiber and matrix.

In the case of a strong interfacial bond, τ is limited by the shear strength of the matrix ( )mτ .

Assuming isotropy of the matrix this results in

3mσ

τ = (2.6)


28

If the fiber length is not uniform, the model can be given by

mmll i

cff

ll c

iffCU V

llVf

llV

fcii

σσσ

σ +⎥⎦

⎤⎢⎣

⎡⎟⎟⎠

⎞⎜⎜⎝

⎛−+⎥

⎦

⎤⎢⎣

⎡= ∑∑

⟩⟨ 21

2 00 (2.7)

The first and second terms in this expression represent the contributions of the fiber length

being shorter and longer than , respectively. cl

As always one should be fully aware of all assumptions that lie behind any model, which in

this case are:

• Stress transfer across the interface increases linearly from the tips of the fiber inwards

to some maximum value

• No fiber-matrix deboding occurs

• The fiber orientation factor is independent of strain and is the same for all fiber

lengths

• The composite matrix properties are the same as the resin properties

• The fiber strength is known (which may also be different from a textbook value or

even a measurement on the fibers used to produce the test samples)

• τ is independent of loading angle

• Fiber diameter is monodisperse

2.4.2 Area fraction method

The deviation of this model to predict the tensile strength of short fiber composite can begin

by considering the total load sustained by the composite ( )CF is equal to the loads carried by

longitudinal fibers and transverse (or random) fibers, which was proposed by Akay and

Barkley [85], defined as

TLC FFF += (2.8)


29

From the definition of stress ( AF σ= ) and the expression for , and in terms of

their respective stresses, the ultimate tensile strength of the composite (

CF LF TF

CUσ ) can be rewritten

as:

TUTLULCCU AAA σσσ += (2.9)

or ⎟⎟⎠

⎞⎜⎜⎝

⎛+⎟⎟

⎠

⎞⎜⎜⎝

⎛=

C

TUT

C

LULCU A

AAA

σσσ (2.10)

where C

L

AA

is the area fraction between the skin region and the cross-sectional area of

specimen; and C

T

AA

is the area fraction between the core region and the cross-sectional area of

specimen. The UTS of the skin region, ULσ , where the fibers near the part surface are

generally aligned in the flow direction or tensile axis, is given by:

mmll i

cff

ll c

iffUL V

llVf

llV

fci

skin

i

skinσσ

σσ +⎥

⎦

⎤⎢⎣

⎡⎟⎟⎠

⎞⎜⎜⎝

⎛−+⎥

⎦

⎤⎢⎣

⎡= ∑∑

⟩⟨ 21

2 00 (2.11)

The ultimate tensile strength of the core region, UTσ , where the fiber orientation is

predominately transverse or random to the flow direction, this equation can be written as:

mmll i

cff

ll c

iffUT V

llVf

llV

fci

core

i

coreσσ

σσ +⎥

⎦

⎤⎢⎣

⎡⎟⎟⎠

⎞⎜⎜⎝

⎛−+⎥

⎦

⎤⎢⎣

⎡= ∑∑

⟩⟨ 21

2 00 (2.12)

Therefore, the UTS of short fiber reinforced composites ( UCσ ) with respect to the effects of

fiber length and fiber orientation can be evaluated with following equation:


30

⎟⎟⎠

⎞⎜⎜⎝

⎛

⎥⎥⎦

⎤

⎢⎢⎣

⎡+⎟⎟

⎠

⎞⎜⎜⎝

⎛⎟⎟⎠

⎞⎜⎜⎝

⎛−+⎟⎟

⎠

⎞⎜⎜⎝

⎛

+⎟⎟⎠

⎞⎜⎜⎝

⎛

⎥⎥⎦

⎤

⎢⎢⎣

⎡+⎟⎟

⎠

⎞⎜⎜⎝

⎛⎟⎟⎠

⎞⎜⎜⎝

⎛−+⎟⎟

⎠

⎞⎜⎜⎝

⎛=

∑∑

∑∑

⟩⟨

⟩⟨

C

Tmm

ll i

cff

ll c

iff

C

Lmm

ll i

cff

ll c

iffCU

AAV

llVf

llV

f

AAV

llVf

llV

f

ci

core

ci

core

ci

skin

i

skin

σσσ

σσσ

σ

21

2

21

2

00

00

(2.13)

where and are the fiber orientation efficiency factors for the skin and core layers,

respectively. The schematic diagram of cross-sectional area for conventional injection

molded composites is depicted in Figure 2.5a.

skinf 0 core

f0

Equation (2.13) can be expressed in terms of the UTS for the sandwich injection moldings as

below:

⎟⎟⎠

⎞⎜⎜⎝

⎛

⎥⎥⎦

⎤

⎢⎢⎣

⎡+⎟⎟

⎠

⎞⎜⎜⎝

⎛⎟⎟⎠

⎞⎜⎜⎝

⎛−+⎟⎟

⎠

⎞⎜⎜⎝

⎛

+⎟⎟⎠

⎞⎜⎜⎝

⎛

⎥⎥⎦

⎤

⎢⎢⎣

⎡+⎟⎟

⎠

⎞⎜⎜⎝

⎛⎟⎟⎠

⎞⎜⎜⎝

⎛−+⎟⎟

⎠

⎞⎜⎜⎝

⎛=

∑∑

∑∑

⟩⟨

⟩⟨

C

coremm

ll i

cff

ll c

iff

C

skinmm

ll i

cff

ll c

iffCU

AAV

llVf

llV

f

AAV

llVf

llV

f

ci

core

ci

core

ci

skin

i

skin

σσσ

σσσ

σ

21

2

21

2

00

00

(2.14)

where and skinA coreA are the cross-sectional area of skin and core materials (see Figure 2.5b).

Equation (2.13) also can be employed for sandwich injection molded composites, containing

different fiber concentration between skin and core material (see Figure 2.5c). For example,

when the skin and core materials filled with 40 and 20 wt% of fiber, the expression can be

written as follows:

⎟⎟⎠

⎞⎜⎜⎝

⎛

⎥⎥⎦

⎤

⎢⎢⎣

⎡+⎟⎟

⎠

⎞⎜⎜⎝

⎛⎟⎟⎠

⎞⎜⎜⎝

⎛−+⎟⎟

⎠

⎞⎜⎜⎝

⎛

+⎟⎟⎠

⎞⎜⎜⎝

⎛

⎥⎥⎦

⎤

⎢⎢⎣

⎡+⎟⎟

⎠

⎞⎜⎜⎝

⎛⎟⎟⎠

⎞⎜⎜⎝

⎛−+⎟⎟

⎠

⎞⎜⎜⎝

⎛

+⎟⎟⎠

⎞⎜⎜⎝

⎛

⎥⎥⎦

⎤

⎢⎢⎣

⎡+⎟⎟

⎠

⎞⎜⎜⎝

⎛⎟⎟⎠

⎞⎜⎜⎝

⎛−+⎟⎟

⎠

⎞⎜⎜⎝

⎛=

∑∑

∑∑

∑∑

⟩⟨

⟩⟨

⟩⟨

C

Tmm

ll i

cff

ll c

iff

C

coremm

ll i

cff

ll c

iff

C

skinmm

ll i

cff

ll c

iffCU

AAV

llVf

llV

f

AAV

llVf

llV

f

AAV

llVf

llV

f

ci

coresub

ci

coresub

ci

core

ci

core

ci

skin

i

skin

σσσ

σσσ

σσσ

σ

21

2

21

2

21

2

20.

20

.

20

20

40

40

00

00

00

(2.15)


31

AL

AT

AC = AL + AT

Skin material

Core material

ASkin

ACore

AC = ASkin + ACore

Skin material

Core material

ASkin

ACore

AC = ASkin + ACore + AT

AT

(a)

(b)

(c)

Figure 2.5 Schematic illustration of cross-sectional area indicating the skin and core regions:

(a) Conventional injection molded short fiber composite, (b) Sandwich injection molded part,

and (c) Sandwich injection molded short fiber composite.

In addition, the prediction of weldline strength, WLσ , of short fiber reinforced composites for

conventional and push-pull injection moldings also can be calculated by rewriting the

Equation (2.13) as :

⎟⎟⎠

⎞⎜⎜⎝

⎛

⎥⎥⎦

⎤

⎢⎢⎣

⎡+⎟⎟

⎠

⎞⎜⎜⎝

⎛⎟⎟⎠

⎞⎜⎜⎝

⎛−+⎟⎟

⎠

⎞⎜⎜⎝

⎛

+⎟⎟⎠

⎞⎜⎜⎝

⎛

⎥⎥⎦

⎤

⎢⎢⎣

⎡+⎟⎟

⎠

⎞⎜⎜⎝

⎛⎟⎟⎠

⎞⎜⎜⎝

⎛−+⎟⎟

⎠

⎞⎜⎜⎝

⎛=

∑∑

∑∑

⟩⟨

⟩⟨

C

Coremm

ll i

cff

ll c

iff

C

Skinmm

ll i

cff

ll c

iffWL

AAV

llVf

llV

f

AAV

llVf

llV

f

ci

core

ci

core

ci

skin

i

skin

σσσ

σσσ

σ

21

2

21

2

00

00

(2.16)


32

where and are the cross-sectional areas where the fibers are aligned randomly and

perpendicular to the flow direction, respectively (see Figure 2.6). In the case of push-pull

injection molded composites, is the region with the predominant fiber orientation in the

flow direction.

skinA coreA

coreA

ACore

ASkin

ASkinAC ACore= +

ACoreACore

ASkinASkin

ASkinAC ACore= +ASkinASkinACAC ACoreACore= +

ASkin

ACore

ASkinAC ACore= +

ASkinASkin

ACoreACore

ASkinAC ACore= +ASkinASkinACAC ACoreACore= +

Conventional injection molded part

Push-pull injection molded part

Figure 2.6 Schematic illustration of cross-sectional area at the weldline position for

conventional and push-pull injection molded short fiber composites.

3. Modeling of the Injection Molding

Process

3.1 Governing equations

The polymer melt flow is governed by the three fundamental laws of physics, i.e. the

principles of conservation of mass, momentum, and energy, which are expressed as

Continuity: ( ) 0=⋅∇+∂∂ u

tρρ (3.1)

Momentum: gPtDuD ρτρ +⋅∇+−∇= (3.2)

Energy: ( )•

+∇+⋅∇⎥⎦

⎤⎢⎣

⎡∂∂

−⋅−∇= SuuTPTq

tDTDC

vp :τρ (3.3)

where ρ is the density, P is the pressure, is the specific heat at constant volume, pC T is

the temperature, v is the specific volume, is the rate of heat generation due to chemical

reaction, is the velocity vector,

•

S

u g is the body force vector, q is the heat flux vector, ∇ is

the gradient operator,tD

D is the substantial derivative, and τ is the extra stress tensor. To

equate the numbers of unknowns and equations, it requires additional relationships among the

variables, such as rheological constitutive equation, an equation of state for polymer, a

thermal constitutive equation and/or equations for cure and kinetics [45]. For example, the

Modeling of the Injection Molding Process 34

rheological constitutive equation of polymer melts is typically modeled by the generalized

Newtonian viscosity model:

γητ &2= (3.4)

with ( )[ ]Tuu ∇+∇=21γ& (3.5)

where η is the non-Newtonian viscosity, γ& is the strain rate tensor, and u is the velocity

vector.

Due to the typically large number of elements and nodes associated with 3-D simulation, it is

necessary to neglect less significant terms in the governing equations. Otherwise, the

computational time and memory requirement would become too excessive to justify for the

3-D simulation. For example, if the polymer melts are assumed to be incompressible during

filling, the conservation equations of mass and momentum in equations (3.1) and (3.2)

become the Navier-Stokes equations [57]:

0=⋅∇ u (3.6)

( ) gPtDuD ργηρ +⋅∇+−∇= &2 (3.7)

To take into account the effect of inertia on the polymer melt flows (e.g. branching flows or

jetting), the inertia term in the left-hand side of equation (3.3) may not be ignored [59].

Whether the body force (e.g. gravity) term, gρ , in the momentum equation can be neglected

depends on the geometry and material of the system [57]. If the inertia is negligible, the

Navier-Stokes equations can be further reduced to the Stokes equations:

( ) gP ργη +⋅∇+−∇= &20 (3.8)

Through the dimensional analysis (see [73]), the energy equation can be reduced to a simpler

form as:

Modeling of the Injection Molding Process

35

( ) γγηρ &&:2+∇⋅∇=⎟⎟⎠

⎞⎜⎜⎝

⎛∇⋅+

∂∂ TkTu

tTCp (3.9)

where is the heat conduction coefficient. k

After the entire cavity is completely filled, the pressure throughout the cavity increases

rapidly as additional polymer melt is packed into the cavity to compensate for the volumetric

shrinkage. During the packing stage, the melt compressibility can no longer be neglected.

Haagh et al. [73] used equation (3.1) as the governing equation to account for polymer

shrinkage in both filling and post-filling stages. Hetu et al. [52] employed the following

governing equations for the compressible flow during the packing stage. The compressibility

effects are included by incorporating the equation of state in the continuity equation.

⎟⎟⎠

⎞⎜⎜⎝

⎛∇⋅+

∂∂

−⎟⎟⎠

⎞⎜⎜⎝

⎛∇⋅+

∂∂

=⋅∇− TutTPu

tPu βα (3.10)

⎟⎠⎞

⎜⎝⎛ ⋅∇−⋅∇+−∇= uIP ηγη

3220 & (3.11)

( ) γγηβρ && :2+∇⋅∇+⎟⎟⎠

⎞⎜⎜⎝

⎛∇⋅+

∂∂

=⎟⎟⎠

⎞⎜⎜⎝

⎛∇⋅+

∂∂ TkPu

tPTTu

tTCp (3.12)

where I is the identity matrix, α is the compressibility coefficient, and β is the thermal

expansion coefficient. The density and the coefficients α and β can be simply calculated

from the equation of state [45].

After the gate freezes off, no more polymer melt enters the cavity and the cooling stage

begins. In this stage, the convection and dissipation terms in the energy equation can be

neglected since the velocity of a polymer melt in the cooling stage is almost zero [52].


3.2 Predicting fiber orientation

As pointed out earlier, knowing an average fiber orientation direction for the complete part is

not sufficient because the local spatial orientation also plays an important role in determining

properties such as strength and toughness. Hence, considerable research has been done with

the intent of learning how to predict fiber orientation in injection molded parts. Mathematical

models have been developed and incorporated into simulations of injection molding with the

emphasis on predicting the influence of mold geometry, processing conditions, and material

properties on the final orientation pattern. Most recent works on the fiber orientation analysis

have their origin from Jeffery [104], who derived the equation of orientation change of an

ellipsoidal particle immersed in the homogeneous flow field based on hydrodynamics. This

equation is available only in a dilute suspension regime where the interaction between fibers

is negligible. In a non-dilute suspension regime, statistical consideration of the fiber

orientation is required to model the ensemble of interacting fiber particles. In this case, the

orientation distribution function is the only model that can describe the distribution of fiber

orientation completely.

For an efficient numerical simulation of the orientation state of fibers, Advani and Tucker

[105] reviewed the orientation tensor, which was originally introduced by Hand [106]. The

meaning of the orientation tensor is similar to the Fourier series of the orientation distribution

function. In this approach, only a few components are required to represent the state of

orientation at each spatial point. This advantage has made the orientation tensor, especially

the second order tensor, to be widely used in calculation of the fiber orientation [47, 107-108]

and material property prediction [109-111] in short fiber composites. The orientation tensor is

independent of the coordinate system, making it advantageous for numerical simulation and

evaluations of orientation. The only weakness of the orientation tensor approach is that a

closure approximation is required to close the governing equation which is the main reason

for causing errors between calculated and measured values. Hybrid closure approximation

proposed by Advani and Tucker [105, 112] has been considered in many applications.

Recently, several approaches have been introduced to propose more accurate closure

approximations [113-115].


37

The following section will review the present state-of-the-art modeling of fiber orientation in

molding short fiber composites. Topics covered include characterizing orientation, fiber

orientation mechanics for a collection of fibers, closure approximation and incorporation of

these ideas in manufacturing process model for injection molding.

3.2.1 Characterizing orientation

A complete description of the distribution of fiber orientation begins by considering a single

fiber whose orientation can be represented by a unit vector p as shown in Figure 3.1.

Components of the vector p are described by angles θ and φ in spherical coordinates as

follows:

φθ cossin1 =p φθ sinsin2 =p θcos3 =p (3.13)

Describing the orientation of individual fibers is ineffective since the composites contain

numerous short fibers. Thus, the concept of probabilistic distribution was introduced to fully

describe the distribution of fiber orientation in three dimensions. The probability distribution

function for orientation, also known as the orientation distribution function, ( )φθψ , , is

defined as the probability of fiber lying between angles θ and θθ d+ , φ and φd . The

probability distribution function must satisfy two physical conditions. First, one end of the

fiber is indistinguishable from the other end, so ψ must be periodic:

( ) ( )πφθπψφθψ +−= ,, (3.14)

Second, every fiber must have some direction, so the integral over all possible directions or

the orientation space must be equal to unity:

( ) 1sin,2

0 0

=∫ ∫ φθθφθψππ

dd (3.15)

This is known as the normalization requirement. If the orientation statistics change with

position, ψ is a function of x , , y z in addition to θ and φ .


p

2

1

3

θ

φ

p

2

1

3

θ

φ

Figure 3.1 Characterization of the fiber orientation in a coordinate system

The distribution function can be approximated by measuring the orientations of a large

number of fibers selected from a region where the distribution function is a complete and

unambiguous description of the fiber orientation state. However, this distribution function

depends not only upon the angles θ and φ but also upon spatial position. The resulting

approach would be highly computationally intensive with large storage requirement, making

it somewhat inappropriate for numerical simulation. For a more efficient method of

numerically simulating the orientation state of fiber, Advani and Tucker [105] used

orientation tensors. Such tensors are defined as the dyadic products of the unit vector p

averaged over all possible directions, with ψ as the weighting function. The definitions of

second ( ) and fourth ( ) order orientation tensors can be defined as: ija ijkla

( )dppppa jiij ψ∫= (3.16)

and ( )dppppppa lkjiijkl ψ∫= (3.17)

There are a number of physical interpretations of these tensors. They can be thought of as a

generalization of orientation parameters, as moments of the distribution function, or as a

series expansion of the distribution function. Advani and Tucker [105] provide a complete

review. They have shown that these tensors are free from a priori assumptions about the


39

shape of the distribution function and can be readily transformed from one coordinate frame

of reference to another complying with the rules of tensor transformation, that is

jiij aa = (3.18)

The normalization condition such as Equation (3.15) implies that the trace of is unity: ija

1=iia (3.19)

The tensor description substitutes less number of scalar quantities for the distribution function

to describe orientation. For example, for planar orientation, only two of the four components

are independent. For the 3-D case, only five of the nine components are independent.

Viscosity and elastic stiffness are normally fourth-order tensors, and one would normally

require only a fourth-order orientation tensor to predict the effect of orientation on these

fourth-order properties [105]. By approximating the fourth-order tensor in term of second-

order tensors, it is possible to predict the effect of orientation on fourth-order properties using

just the second-order orientation tensor [112].

3.2.2 Flow-induced fiber orientation

Jeffery modeled accurately the motion of a single fiber immersed in a large body of

incompressible Newtonian fluid [104]. However, fibers in concentrated suspensions behave

somewhat differently. A number of researchers have observed fiber orientation in

concentrated suspensions [80-89, 96]. All of them reported fiber orientation behavior that is

qualitatively similar to single-fiber motion. In elongational flows, fibers align along the

direction of stretching, aligning with the streamlines in converging flows and normal to the

streamlines in diverging flows. Shear flows tend to orient fibers in the flow direction.

However, these researchers do not observe perfect alignment of fibers, which the theory

predicts. Fibers flowing in concentrated suspension are so close to each other that they not

only interact hydrodynamically but may physically collide with each other, causing more

erratic motion that violates Jeffery’s assumptions.

The dynamic behavior of a concentrated suspension is quite complex, as the rheology is

closely coupled with the fiber orientation structure. So far, a phenomenological model


proposed by Folgar and Tucker has proven useful [116]. They model the interaction between

the fibers by introducing an additional term in the equation of motion for single-fiber motion.

This term is similar to a diffusive term, and the effective diffusivity is made proportional to

the strain rate, as interactions take place only when the suspension is deforming. A

dimensionless “interaction coefficient” term, typically of the order of 10IC -2, served to

match their experimental results [116-117]. The equation of motion can then be combined

with the equation that conserves fibers in the orientation space to produce the equation of

change for fiber orientation in terms of distribution function and orientation tensors [105,

116-117]. For example, the equation of change for the second-order tensor is given by:

( ) ( ) ijijIijklklkjikkjikkjikkjikij aCaaaaatD

aD322

21

21

−+−++−−= δγγγγλωω &&&& ( ) (3.20)

where ijδ is the unit tensor, j

i

i

jij x

vxv

∂∂

−∂

∂=ω and

j

i

i

jij x

vxv

∂∂

+∂

∂=γ& are the vorticity and the

rate of deformation tensors, respectively. A shape factor of particle is defined as ( )( )1

12

2

+

−=

e

e

rr

λ

where is the aspect ratio of the ellipsoid, is the experimental interaction coefficient

depending on the particle geometry and concentration, and

er IC

2ijij γγ

γ&&

& = is the effective shear

rate.

3.2.3 Numerical simulation of fiber orientation for injection molding

Over the last decade, several numerical techniques have been developed to determine the

fiber orientation in injection molded composites by using the generalized Hele-Shaw (GHS)

formulation [47, 108, 113, 118-122]. However, because of the simplifying assumptions used

in GHS flow model [42], the so-called 2.5-D model reaches its limits in the simulation of the

thick-walled moldings, complex geometrical configurations (such as bosses, corners, and

ribs), and at the melt front (fountain flow) regions. Therefore, the fully 3-D simulation model

should be able to generate complementary and more detailed information related to the flow

characteristics in injection molded parts than the one obtained when using a 2.5-D model.

This will be particularly important while molding with fiber reinforced systems.

4. Experimental and Simulation

Procedures

4.1 Materials and processing conditions

4.1.1 Sandwich injection molding

To investigate the influence of processing parameters and glass fiber content on the material

distribution in sandwich injection molding, the polystyrene (PS 165H), unfilled polycarbonate

(Makrolon 2800), and polycarbonate filled with 20 and 35 wt% short glass fiber (Makrolon

8020 and Makrolon 8345) were used. The materials were supplied in granular form by BASF

Chemical Co., Ltd. and BAYER GmbH, Germany, respectively. The sandwich injection

molded part (dumbbell shape) was carried out on an ARBURG ALLROUNDER two-

component injection molding machine (Model: 320S 500-350). The core materials were

colored prior to injection to facilitate the identification of the interface between skin and core

materials. All the specimens were molded after the machine had attained a steady state with

respect to the preset melt and mold temperatures. The selection of the variation range for each

injection parameter was based on the filling technique recommended by the material supplier

and ARBURG’s operating instructions [123]. The effects of the skin/core ratio, molding

parameters, and glass fiber content on the core thickness were varied on three levels, while

other variables were maintained at a constant level throughout this study. Table 4.1 and 4.2

summarize the materials and the various parameter settings. These four materials are

respectively designated as PS, PC, SFRPC20, and SFRPC35. The sandwich combinations

considered are PC/PC, PC/SFRPC20, PC/SFRPC35, SFRPC20/PC, and SFRPC35/PC. In this

nomenclature, the first constitutes the skin while the second designates the core.

Experimental and Simulation Procedures 42

Table 4.1 Material and variation of parameter settings used for investigating the effect of

processing conditions on the skin/core material distribution in sandwich injection molding.

Material Grade Supplier

Polystyrene PS 165 H BASF

1st-Plasticator (Skin) 2nd-Plasticator (Core)

N In

M

* T

ozzle temperature ( °C) 210 / 230 / 260 210 / 230 / 260jection flow rate (ccm / s) 9.25 / 18.5 / 37.0 6.5 / 13.5 / 27.0 / 54.0

* Injection volume (%by volume) 35 / 40 / 45 65 / 60 / 55

old temperature was set at 40 °CTotal cooling time = 40 sec.

otal volume of part = 37.0 ccm

Processing ConditionsSandwich Molding

Table 4.2 Materials and processing parameters used for investigation of the effect of mold

temperature and glass fiber content on the skin/core material distribution in sandwich injection

molding.

Skin Material Core Material (Skin/Core)

PC PC PC/PCPC PC+SGF 20 wt% PC/SFRPC20PC PC+SGF 35 wt% PC/SFRPC35

PC+SGF 20 wt% PC SFRPC20/PCPC+SGF 35 wt% PC SFRPC35/PC

1st-Plasticator (Skin) 2nd-Plasticator (Core)

NInj

ozzle temperature ( oC) 300 300ection flow rate (ccm/s) 14.95 16.65njection volume (% by volume) 50 50

old temperatures were set at 40 °C / 80 °C /120 °CTotal cooling time = 40 sec.

otal volume of part = 37.0 ccm

Sandwich Molding Sample Code


* I

M

* T

Experimental and Simulation Procedures

43

Polypropylene compounded with short glass fiber is one of the most commercially important

materials and exhibits the most significant development and growth. Therefore, to investigate

the effect of processing technique on the fiber orientation, fiber attrition and mechanical

properties, the unfilled polypropylene (PP-H 1100 L, supplied by TARGOR) and

polypropylene filled with 20 and 40 wt% short glass fiber (PP32G10-0 and PP34G10-9,

supplied by BUNA) were used. The test specimens were also molded on the same machine

which can be employed both for conventional injection molding and sandwich molding.

Injection speed and skin/core volume ratio were chosen as follows: the speed of the first

injection unit (skin material) was kept higher in order to achieve a good surface finish and to

prevent premature solidification of the melt, whereas lower speed was used for the second

injection unit (core material). The latter was done in order to assess the uniform core

extension along the flow direction without the breakthrough of the core material at the far end

of the bar [124-125]. Several settings were tried and those leading to an overall satisfying

quality with regard to visual properties were finally chosen. The mold temperature was 55 °C

and the five heating zones (from nozzle to feed zone) were set to 250 °C, 240 °C, 230 °C, 220

°C, and 210 °C, respectively. The materials and processing parameters used for single and

sandwich molding specimens, containing different short glass fiber contents between skin and

core materials, are given in Table 4.3 and 4.4.

Table 4.3 Materials used for investigation of the effect of processing technique on the fiber

orientation, fiber attrition, phase separation, and mechanical properties.

No. Sample Code

1 PP2 SFRPP203 SFRPP40

Sample CodeSkin Material Core Material (Skin/Core)

4 PP+SGF 20 wt% PP SFRPP20/PP5 PP PP+SGF 20 wt% PP/SFRPP206 PP+SGF 20 wt% PP+SGF 20 wt% SFRPP20/SFRPP207 PP+SGF 40 wt% PP SFRPP40/PP8 PP PP+SGF 40 wt% PP/SFRPP409 PP+SGF 40 wt% PP+SGF 20 wt% SFRPP40/SFRPP2010 PP+SGF 40 wt% PP+SGF 40 wt% SFRPP40/SFRPP40

Single Molding

Sandwich Molding

PP PP+SGF 20 wt% PP+SGF 40 wt%


Table 4.4 Processing conditions used for investigation of the effect of processing technique on

the fiber orientation, fiber attrition, phase separation, and mechanical properties of single and

sandwich injection moldings.

1st-Plasticator 2nd-Plasticator

Injection pressure (bar) 1000 1000 1000Holding pressure (bar) 800 - 800

Holding time (sec) 25 - 25Back pressure (bar) 20 20 20Cooling time (sec) 40 - 40

Injection flow rate (ccm/s) 18.5 18.5 8.8Screw speed (m/min) 12 12 12

Injection volume (ccm), (%) 37 (100%) 14.8 (40%) 22.2 (60%)

Sandwich MoldingProcessing Conditions Single Molding

4.1.2 Push-pull injection molding

Compared to the conventional injection molding process, the push-pull technique is different

in the way that the mold is fitted with at least two gates. The cavity is firstly filled

simultaneously by the melt from both the units via the two separate gates (see Figure 4.1).

After the two melt fronts meet, the weldline is formed and the filling phase is subsequently

switched to the holding phase. The material solidifies starting at the cavity wall but there is

still molten core and then the first push-pull stroke begins. The control software program

allows the definition of several holding pressures for one stroke from either the first or the

second injection unit. From one of the injection units, polymer melt is pressed into the cavity

resulting in the molten core being pushed through the gate back into the other injection unit

and thus the geometry of weldline is deformed to a tongue shape. One of these movements is

termed as push-pull 1 stroke. The number of strokes can be selected by taking into account

the part’s thickness. When all the strokes are completed, cooling phase follows. As the

thickness of frozen layer increases with the number of strokes within the holding time,

therefore the total cycle time is not notably increased as compared to conventional injection

molding [23].


45

MoldMold unitunit

First First plasticatorplasticator

Seco

nd

Seco

nd p

last

icat

orpl

astic

ator

Weldline

(b)(b) (c)(c)(a)(a)

Figure 4.1 Schematic principle of push-pull injection molding process (a) First step, (b)

Second step, and (c) Third step.

The materials used in this study were unfilled polycarbonate (Makrolon 2800) and

polycarbonate filled with 20 and 35 wt % short-glass-fiber (Makrolon 8020 and Makrolon

8345). The dumbbell-shaped specimens were injection molded on ARBURG two-component

injection molding machine. Besides the push-pull processing, conventional injection molding

was also carried out for reference purpose. The material injected from both the units was same

with an addition of a small amount of pigment in one of them to be used as a tracer material.

All the specimens were molded only after the machine had attained a steady state with respect

to the preset melt and mold temperatures. The processing parameters used for push-pull

molding are given in Tables 4.5 and 4.6, respectively.


Table 4.5 Processing conditions used in this study.

* T

Processing Conditions 1st-Plasticator 2nd-Plasticator

Nozzle temperature (oC) 300 300Injection speed (ccm/s) 30 30Injection pressure (bar) 1000 1000

* Injection volume (% by volume) 62.5 37.5

Total cooling time = 45 sec. Mold temperature = 100 oCotal volume of part = 24.0 ccm

Table 4.6 Parameter settings for push-pull 1, 2, and 3 strokes

Difference of Holding Difference of Holding Difference of Holding Holding Pressure Time Holding Pressure Time Holding Pressure Time

ΔP (bar) (sec) ΔP (bar) (sec) ΔP (bar) (sec)

PC / PC

120

120

Push-Pull 3 strokes

(# 1st / # 2nd)

Sample Code Push-Pull 1 stroke Push-Pull 2 strokes

70 4

4

5

54

70 5 4

120

220

220 4

120 4

5

4

120 4

5220

70 5

4

SFRPC20 / SFRPC20

SFRPC35 / SFRPC35

10

10

10

120

120

120

120

4.2 Microstructure analyses

4.2.1 Skin/core material distribution

For investigating the skin/core material distribution, tensile specimens were cut along the

flow direction through the middle at five different locations, as shown in Figure 4.2. The

sections were then mounted on a stage, after polishing with the help of a metallurgical


47

technique. The thickness fraction of the core material ( bδ ) was assessed by optical

microscopy (OLYMPUS model PMG3) and computer aided image analysis (a4i Analysis

version 5.1 and Image-Pro Plus). The measurements were taken at every 6 mm from the gate,

which corresponded to the measured distance ratio ( 0Lxi ) between length of measurement

and total length of specimen (170 mm).

170 mm

Flowdirection

6 mm

4 mm

bδ

Longitudinal area

Skin material

Core material

Figure 4.2 Location of sections for material distribution analysis [126].

4.2.2 Fiber orientation analysis

Polarized light microscopy and computer aided image analysis were also utilized in order to

investigate the fiber orientation distribution. For observing fiber orientation, tensile

specimens were cut parallel to the flow direction into various layers parallel at the middle of

specimen (or weldline position for push-pull molded part) as shown schematically in Figure

4.3. The sections were then polished using a metallurgical technique and mounted on a stage.

In the present case, 500 fibers per sample were measured, establishing histograms and

calculating the fiber orientation variation across only half the thickness of the sections

assuming the symmetry of flow. In order to determine planar fiber orientation in the skin and


core layers, the second order orientation tensor , introduced by Advani and Tucker [105],

was calculated using the following equation:

11a

∑=

=i

i

N

niN

aϕ

ϕϕ 1

211 cos1 (4.1)

where iϕ is the angle between the individual fibers and the local flow direction and is

the number of fibers with a certain angle

iNϕ

iϕ to the local flow direction. For perfect alignment

along the flow axis, the orientation average ( ) would be equal to 1, whereas when fibers are

randomly oriented to the flow direction it would be 0.5.

11a

X Y (Flow direction)

Z

Weldline Position

20 mm

Figure 4.3 Location of areas for fiber orientation analysis [23, 127].


49

4.2.3 Fiber length analysis (Fiber attrition)

For the investigation of fiber lengths within the skin and core layer, the tensile specimens

were cut into seven sections, as shown in Figure 4.4. For the separation of skin and core

materials microtome technique was employed. Short-glass-fibers were isolated from the

composite materials by using an incineration method, according to DIN EN 60. Magnified

fiber images were then digitized semi-automatically with the help of Image-Pro Plus software

running on a personal computer. The fiber length distribution (FLD) was determined by the

average fiber length which was calculated from a minimum of 500 length measurements on

fibers recovered from the incineration of the specimen sections. The percent difference

between the average fiber length inside the granules and the overall glass fiber length inside

the molded part ( ) was used to describe the results. For this purpose the following

equation was employed:

lΔ%

100% ×⎥⎦

⎤⎢⎣

⎡ −=Δ

G

Gj

lll

l (4.2)

with being the average fiber length inside the granules and the local fiber length inside

the individual layers (skin and core layers) of the sections of the parts.

Gl jl

1

2

4

3

6

7

5

25 mm25 mm

25 mm

25 mm

25 mm

25 mm

20 mm

170 mm

Figure 4.4 Location of sections for fiber length distribution analysis.


4.3 Mechanical testing

Geometry of tensile specimens is shown in Figure 4.5a, according to the recommendation of

DIN EN ISO 527-1/1A/5. Impact bars were obtained from the tensile bars by removing the

clamping parts; these rectangular specimens are of thickness 4 mm, width 10 mm and length 80

mm. A V-notch of 45o ± 1o and a root radius of 0.25 ± 0.05 mm were made by sawing with a

razor blade. The dimension of the Charpy V-notch impact specimen is illustrated in Figure

4.5b, referring to the standard of DIN EN ISO 179/1 e A.

Test specimen

Sprue

Gate

Runner

(a) (b)

Figure 4.5 (a) Dimensions of the injection molded specimen and (b) Geometry of Charpy V-

notch impact.

The molded tensile specimens were tested on Zwick 1464 mechanical tester at a crosshead

speed of 5 mm/min for a sample gage length of 50 mm. For each molding condition, five

dumbbell-shaped specimens were tested and the average values of the maximum tensile stress

were used for analysis. Charpy impact tests were conducted on a CEAST impact tester model

6545 using the specimens with a V-notch. The tests were carried out with impact energy of 1

Joule and a sample span length of 80 mm. The average values of notched Charpy impact

strength (kJ/m2) were obtained again from a group of five specimens.


51

4.4. Process simulation

4.4.1 Pre-processing

A simple simulation project can be subdivided into three sections. In the first stage, a model

is created, the boundary conditions are assigned and the calculation is set up. In the

subsequent processing step, the calculation algorithm is applied to the meshed model and

resulting files are automatically saved. This means that the results are not visible immediately.

When displaying results (post-processing), first one has to decide about which information

contained in the extensive result data should be presented. Then, the corresponding data can

be read from the result files and can be displayed graphically. Results from the first

calculation are interpreted. Then the decisions to modification of the part geometry or the

process parameters are made and a new calculation is performed. Additionally, the results

from an analysis can be the basis for the following simulation. For example the results from a

filling analysis are used as a basis for the packing analysis. A cooling analysis can be

attached to a packing analysis. Figure 4.6 shows the typical sequence performed when

simulating the injection molding process.

Open a new project

Create mold model Open / Createa part model

Mesh mold model Create and edit the mesh

If required createrunner system

Assign boundaryconditions

Set up analysis

PrePre--processingprocessing

Create report

Change partgeometry

Change boundaryconditions

Run analysis Create and display results

Are the resultsreasonable ?

Perform furtherAnalysis if

required

If yes

If no

Open a new project

Create mold model Open / Createa part model

Mesh mold model Create and edit the mesh

If required createrunner system

Assign boundaryconditions

Set up analysis

PrePre--processingprocessing

Create report

Change partgeometry

Change boundaryconditions

Run analysis Create and display results

Are the resultsreasonable ?

Perform furtherAnalysis if

required

If yes

If no

Figure 4.6 Typical sequence performed in injection molding simulation.


4.4.2 Simulation approach

4.4.2.1 Simulation of skin/core material distribution in sandwich injection molding

To investigate the effects of processing parameters and glass fiber content on the skin/core

material distribution, the commercial software package, Moldflow, has been utilized in order

to predict the flow behavior and thickness fraction of the core material in the sandwich

injection molding process. The midplane model was first created by a CAD program

(Solidwork) and was meshed by creating triangular elements on the surfaces (3517 nodes,

6296 elements) as demonstrated in Figure 4.7. In the present implementation for the

sandwich injection molding process, each polymer obeys the governing equations for

generalized Hele-Shaw flow of inelastic, non-Newtonian fluids under non-isothermal

conditions [44].

Injection Point

8.0006.7505.5004.2503.000

Mesh thickness (mm)

Figure 4.7 2.5-D mesh model used for the simulation of skin/core material distribution in

sandwich injection molding.


53

4.4.2.2 Simulation of 3-D fiber orientation distribution in sandwich and push-pull injection

moldings

At present, a commercial software package for 3-D simulation of two-component injection

molding including the fiber orientation analysis is not yet available. However, from the

process setting of this program (where the user can input the processing parameters i.e. an

injection location, selection of material, definition of mold and melt temperature) it is

possible to control the melt flow profile during the filling phase similar to the melt flow

development during sequential sandwich and push-pull injection molding processes. By

using only one material and controlling the ram speed profile during injection phase for

sandwich injection molding process [128] or utilizing the controlled valve gate with the hot

runner system for push-pull processing [129]. In this study, the 3-D models of sandwich and

push-pull processing have been developed and performed with the aid of Moldflow

simulation program in order to predict the 3-D fiber orientation distribution in the sandwich

and push-pull injection molded parts. First, the 3-D model was created by Solidwork program

and was then meshed by creating tetrahedral elements within the CAD model (see Figures 4.8

and 4.9). This mesh model was then simulated by using the 3-D finite element analysis,

which is based on fluid mechanics and heat transfer calculations. The material properties used

for computer simulation of skin/core material distribution in sandwich injection molding and

for predicting of 3-D fiber orientation distribution in sandwich and push-pull injection

moldings are summarized in Tables 4.7

Table 4.7 Material properties (Moldflow database).

Thermal Conductivity Specific Heat Glass Transition

(W / m °C) (J / kg °C) Temperature (°C)

PP + 40% Short Glass Fiber * Celstran PP-GF40 0.140 at 275 °C 2243 at 275 °C 138

PS PS 165H 0.155 at 230 °C 1975 at 230 °C 92

PC Makrolon 2800 0.173 at 300 °C 1700 at 300 °C 139

PC + 20% Short Glass Fiber Makrolon 8020 0.185 at 300 °C 1530 at 300 °C 150

PC + 35% Short Glass Fiber Makrolon 8345 0.209 at 300 °C 1400 at 300 °C 150

* The material properties of PP filled with 40% short glass fiber from TARGOR (PP34G10-9) are not available in Moldflow database

Materials Trade Name


Injection Point

Figure 4.8 3-D mesh model used for simulation of fiber orientation during sandwich injection

molding (59,798 nodes, 337,358 tetrahedral elements).

Figure 4.9 3-D mesh model used for simulation of fiber orientation during push-pull

processing (22,352 nodes, 119,355 tetrahedral elements).

Overflow Cavity # 1

Overflow Cavity # 2

Main Cavity

Injection Point

Valve Gate # 1

Valve Gate # 2

Valve Gate # 3

Valve Gate # 4

Hot Runner System


55

• Sandwich injection molding

The sandwich injection molding simulation was performed by injecting 40 % vol. of material

(Celstran 40 wt% GF) into the cavity with the injection flow rate of 18.5 ccm/s. After

approximately 1 second, the rest of material (60 %vol.) was then injected into the cavity with

a slower injection flow rate of 8.88 ccm/s until the end of the injection shot. The comparison

of simulation results between single and sandwich injection molding processes is shown in

Figure 4.10.

Single injection molding Sandwich injection molding

40 %vol. of material

Figure 4.10 Three-dimensional simulation results of the melt flow front during the filling

stage for single and sandwich injection moldings.


• Push-pull processing simulation

To simulate the conventional injection molded part with weldline, valve gates # 2 and # 3

were initially opened whereas valve gates # 1 and # 4 were closed. The simulation started

when the polymer melt was injected from the injection point and passed valve gates # 2 and #

3 into the main cavity and a weldline was formed by the merging of two melt fronts at the

middle of the part, as demonstrated in Figure 4.11a. After the main cavity had been

completely filled, the simulation was switched to the packing phase and the push-pull

processing could be simulated by alternatively opening and closing the valves as described

previously [71,129]. Figure 4.11b to 4.11d demonstrate the simulation results of the push-pull

process, for the number of push-pull 1, 2, and 3 strokes, respectively. In the case of the push-

pull 1 stroke, valve gate # 3 was closed after the main cavity had been filled and a weldline

had been produced and at the same time valve gate # 4 was opened. During this period,

depending on the setting time of the valve gate, the molten polymer flowed continuously

from the main cavity through valve gate # 4 into overflow cavity # 2. For the push-pull 2

strokes, after the first stroke was completed, valve gates # 2 and # 4 were closed and valve

gates # 1 and # 3 were activated. Then the polymer melt flowed in the reverse direction from

the main cavity through valve gate # 1 into overflow cavity # 1. In the case of the push-pull 3

strokes, the third stroke of the push-pull processing simulation started after the second stroke

was completed by closing valve gates # 1 and # 3 and opening valve gates # 2 and # 4. The

polymer melt was again pushed back through the corresponding cavity system.


57

Weldline Position

Valve Gate # 1

Valve Gate # 2Valve Gate # 3

Valve Gate # 4

Overflow Cavity # 2 Overflow Cavity # 1

Valve Gate # 1


Valve Gate # 4


Push-Pull 1 stroke

Valve Gate # 1


Valve Gate # 4


1st stroke 2nd stroke

Valve Gate # 1


Valve Gate # 4


2nd stroke1st stroke

3rd stroke

(a) (b)

(c) (d)

Fill time (sec)

11.268.4435.6292.8140.000

Figure 4.11 Evolution of the melt flow front during the simulation: (a) Conventional injection

molding with weldline, (b) Push-pull injection molding 1 stroke, (c) Push-pull injection

molding 2 strokes, and (d) Push-pull injection molding 3 strokes.

Experimental and Simulation Procedures 58 Experimental and Simulation Procedures 58

5. Experimental Results and Discussion

5.1 Comparison between conventional and sandwich injection moldings

5.1.1 Fiber orientation distribution

As Figure 5.1a shows, it was found that there are a number of distinct regions within the

moldings with different fiber alignments. This has also been identified by several other

studies [17, 20, 47, 80-89,120]. The layer at the mold wall, referred to as the surface layer,

tends to have fibers randomly oriented or slightly flow aligned. This randomly oriented

region is caused by fountain flow near the melt front [113]. In particular, fibers from the core

region near the melt front move outward to the wall passing through the fountain flow region.

In the skin region, the fiber orientation is predominately parallel to the flow direction. This is

due to elongational forces arising during fountain flow at the front and to shear flow after the

front has passed. In contrast, a random in plane alignment of fibers is observed in the core

layer due to a slower cooling rate and lower shearing. Moreover, the micrographs clearly

reveal that voids are mostly located within the core layer. This observation has also been

reported in previous works [85, 96,130]. The presence of voids is mainly attributed to

shrinkage during the cooling stage of injection molding where the molten core undergoes

shrinkage away from the solidified skin layer [130]. Figure 5.2 shows the measured values

for the fiber orientation tensor ( 11a ) vs. the relative thickness (zi / h). From the measured

results obtained in this study, it is interesting to note that the core of the molding appears to

contain a random alignment of fibers ( 11a ≈ 0.5) rather than highly aligned transverse to the

flow direction ( 11a ≈ 0). The explanation for this would be related with the effect of geometry

such as a clamping part of tensile specimen (at the entrance region) where the converging

flow field is established. This can lead to a higher velocity gradient of the flowing melt along

the flow path, and thus resulting in an increase of the fiber orientation within the core layer. It

Experimental Results and Discussion 60

also can be observed that the 11a in the core region of SFRPP20 is higher than in the core of

SFRPP40. These results are in agreement with the work carried out by previous researchers

[83-85] in that an increase of the thickness of the core layer of injection molded short fiber

reinforced thermoplastics appears to be more pronounced as the glass fiber content increases.

Figure 5.1 Optical micrographs in the Z-Y plane of (a) single and (b) sandwich injection

moldings.

VoidsVoids

FlowFlow DirectionDirection

SFRPP20SFRPP20 SFRPP40SFRPP40

Skin Skin LayerLayer

CoreCore LayerLayer

SurfaceSurface LayerLayer

Y

Z(a)

SFRPP20/SFRPP20SFRPP20/SFRPP20 SFRPP40/SFRPP40SFRPP40/SFRPP40

FlowFlow DirectionDirectionY

Z

(b)

Experimental Results and Discussion

61

Figure 5.2 Variation of the 11a component of orientation tensor through the half thickness of

single and sandwich injection molded specimens.

Figure 5.1b shows the fiber orientation inside the skin and core layers of sandwich molded

specimens with different glass-fiber contents. For SFRPP20/SFRPP20 and

SFRPP40/SFRPP40, it can also be seen in Figure 5.2 that the values for 11a within the core

layer are higher than those obtained for single injection moldings. This can be due to two

possible reasons. Firstly, it can be the result of the melt flow front of the first injected

material (which will become the skin material) develops a parabolic velocity profile, near the

mold wall the fibers are generally aligned in the flow direction due to the high velocity

gradient. Prior to the skin material’s reaching the end of the cavity, the second material is

injected to form the core. This material develops a second flow front, pushing the skin

material ahead of it. As shown in Figure 5.3, the velocity at the center of the core material is

higher than that at the skin flow front, because the material injected first solidifies as it comes

into contact with the cold wall of the mold. The solidified skin material can act as a second

mold wall inside the mold cavity, narrowing of the flow channel. Thus the higher the velocity

gradient of core material, the higher is the fiber orientation in the core layer. Secondly, the

slower the injection speed of the second material during the filling stage, the higher the

thickness of the solidified skin layer restricting the cross-sectional area available for the

0.0 0.2 0.4 0.6 0.8 1.00.4

0.5

0.6

0.7

0.8

0.9

1.0

Surface Midplane

Orie

ntat

ion

Tens

or C

ompo

nent

(a11

)

Orie

ntat

ion

Tens

or C

ompo

nent

(a11

)

Relative Thickness (zi / h)

Experimental Results

SFRPP20 SFRPP40 SFRPP20/SFRPP20 SFRPP40/SFRPP40

0.4

0.5

0.6

0.7

0.8

0.9

1.0


flowing melt. This leads to a higher velocity gradient that tends to increase the fiber

orientation of the adjacent melt layer. These results are in agreement with the work carried

out by previous researchers [47, 81, 119] in that the fibers at the mid-plane become more

flow aligned and the thickness of the core region decreases with the thickness of the cold

boundary layer increasing.

Figure 5.3 Schematic of polymer melt flow profile and resulting fiber layer structure in

sandwich injection molded short fiber composites. [127]



PrimaryPrimary Skin Skin LayerLayer

CoreCore LayerLayer

SecondarySecondary Skin Skin LayerLayerTotal Skin Total Skin LayerLayer

PrimaryPrimary Skin Skin LayerLayer

CoreCore LayerLayer


Skin Material

Mold WallMold Wall

Core Material

Velocity Profile of Skin Material

Velocity Profile of Core Material



63

The photomicrographs of longitudinal area of sandwich specimens are illustrated in Figures

5.4a through 5.4e. These pictures clearly indicate that the fibers are highly oriented parallel to

the local flow direction within the skin region, as presented in Figures 5.4a to 5.4c.

Furthermore, in the core region, the higher degree of fiber orientation and lesser voids

(Figures 5.4c-e) of sandwich molded parts are thought to be caused by the shape of the

velocity profile and the thickness of the frozen layer near the wall, as stated earlier.

Figure 5.4 Optical micrographs of longitudinal area (Z-Y plane) of sandwich injection

molded parts: (a) SFRPP20/PP, (b) SFRPP40/PP, (c) SFRPP40/SFRPP20, (d) PP/SFRPP20,

and (e) PP/SFRPP40.


SFRPP20SFRPP20

SFRPP40SFRPP40

PPPPPPPP

PPPP PPPP


(a) (b) (c)

(d) (e)

Y (Flow direction)

Z


5.1.2 Fiber length distribution (Fiber attrition)

Figure 5.5 shows the percent differences between the mean fiber length of the granules and

the overall glass fiber length inside the molded part ( lΔ% ). In all the cases, the fiber length

is much lower in the injection moldings than in the granules, which is due to the fact that

fiber length is always reduced to a limiting value depending on melt viscosity, the intensity of

the shear field and the residence time [91-92]. Furthermore, it is obvious that the mean fiber

length for each subdivision of tensile specimens decreases with the increase of the glass fiber

concentration. This has been observed by many authors [82, 87, 92, 94-95, 99, 102] who

mainly attribute a higher fiber concentration to a higher degree of fiber-fiber interaction and

increased fiber-wall contacts. Moreover, it can be seen that the fiber attrition inside the skin

layer of the injection moldings is higher than that in the core layers. Our experiments with

simple molded geometry showed that the reduction of fiber length for SFRPP20 is

approximately 10-15% in the core and about 20-25% in the skin layer. As pointed out by

previous researchers [83, 131], the fiber length is obviously higher within the core layer than

in the skin layer. This is due to the following mold filling characteristics. The core

component is filled with relatively low shear force compared to the skin component, where

the melt begins to solidify as soon as it comes into contact with the cold mold wall. Therefore,

less deformation is applied to the fibers at the center which results in a higher average fiber

length in the core region. For the higher fiber concentration, a higher degree of fiber

degradation inside the skin and core layers, which accounts for approximately 30% in the

core layer and 40-45% in the skin layer, can be observed. The occurrence of more

pronounced fiber length degradation for the higher fiber concentration is believed to arise

from an increased fiber-fiber interaction in the more viscous melt.

With respect to fiber attrition in the longitudinal direction of the bar, it can be noted that the

effects of different processing types and glass-fiber concentrations do not lead to significant

changes of fiber length. In all the cases, only insignificant differences between the

subdivisions were observable. Probably the effect of the simple mold geometry, used for this

investigation, on fiber length destruction is smaller than that of complicated mold geometry

[132]. Comparing the effect of sandwich and single injection molding processes on the fiber

length inside the skin region, only minor differences were observed. As mentioned earlier,

this is due to a higher shear rate near the mold surface and fiber interactions with the mold

wall. The effect of different processing types on fiber length, however, is more pronounced in


65

the core region. The fiber length distribution within the core layer of the sandwich molding

(SFRPP20/SFRPP20) is slightly lower than the values obtained for the single injection

molding (SFRPP20). For a higher fiber loading, the fiber length inside the core region of

SFRPP40 and SFRPP40/SFRPP40 becomes higher. This can not only be explained by the

narrower flow channel and the higher shear rate occurring during the sandwich molding

process, but also by the higher fiber loading itself, which results in more frequent fiber-fiber

interactions and, thus, higher fiber destruction in the core region of sandwich moldings.

Figure 5.5 Fiber attrition in the skin and core layers at various positions of tensile specimens:

(a) Single and sandwich molded parts containing 20 wt% short glass fibers and (b) Single and

sandwich molded parts containing 40 wt% short glass fibers.

5.1.3 Mechanical properties

Figure 5.6 illustrates the tensile and impact properties of sandwich molding specimens

containing different glass fiber concentrations within the skin and core materials in

comparison to those of single injection molding specimens. It is generally known that the

addition of glass fibers results in an enhancement of the tensile and impact properties [76, 82-

1 2 3 4 5 6 7-50

-40

-30

-20

-10

0-50

-40

-30

-20

-10

0

%Δ l

Position

SFRPP40 (Core layer) SFRPP40 (Skin layer) SFRPP40/SFRPP40 (Core layer) SFRPP40/SFRPP40 (Skin layer)

%Δ l SFRPP20 (Core layer) SFRPP20 (Skin layer) SFRPP20/SFRPP20 (Core layer) SFRPP20/SFRPP20 (Skinlayer) (a)

(b)


89, 95-96, 102-103]. The mechanical properties of PP co-injected with glass-fiber reinforced

PP are generally at an intermediate level between those of PP and glass-fiber reinforced PP

alone [133]. It is interesting to note that, for the sandwich injection moldings

(SFRPP20/SFRPP20 and SFRPP40/SFRPP40), the maximum tensile stress and impact

strength are higher than for the single injection moldings (SFRPP20 and SFRPP40). This

improvement of the mechanical properties is considered to be due to a higher degree of fiber

orientation within the core layer (the influence of voids is neglected since the area fraction

between the total area of voids and the cross-sectional area of the specimen is approximately

less than 0.25%). However, comparing the mechanical properties of sandwich molding and

single injection molding, it can be observed that the maximum tensile stress and the impact

strength of sandwich specimens are not as high as one might have expected. This is probably

due to the higher fiber attrition that occurs during sandwich molding process and this fiber

shortening can reduce the fiber reinforcing efficiency [134-136].

Figure 5.6 Maximum tensile stress and impact strength of conventional and sandwich

injection molded parts containing different glass fiber contents.

PP

SFRP

P20/

PP

SFRP

P40/

PP

PP/S

FRPP

20

PP/S

FRPP

40

SFRP

P20

SFRP

P20/

SFRP

P20

SFRP

P40/

SFRP

P20

SFRP

P40

SFRP

P40/

SFRP

P40

0

20

40

60

80

100

Max

imum

Ten

sile

Str

ess

(MPa

) Maximum Tensile Stress Impact Strength

0

2

4

6

8

10

12

14

Impa

ct S

tren

gth

(kJ/

m2 )


67

5.2 Comparison between conventional and push-pull injection moldings

5.2.1 Geometry of weldlines

Figure 5.7 shows the weldline geometries of conventional and push-pull processed specimens

for SFRPC35 / SFRPC35, which are molded at several holding pressure differences. In the

case of conventional injection molding process, a relatively straight weldline is produced

when the holding pressures on both sides are kept same. The geometry of weldline is

deformed to the tongue shape with difference in holding pressure. It has been found that the

position of the weldline at the mold surface does not change even when the pressure

difference is increased. The tongue-shaped weldlines can also be observed for SFRPC20 /

SFRPC20 and PC / PC in the same way. The geometry of weldline for push-pull 2 and 3

strokes processed specimens can be easily observed in the longitudinal area (Z-Y plane), (as

presented later in Figures 5.10 and 5.11).

Figure 5.7 Weldline geometry of SFRPPC35/SFRPC35.

ΔΔP = 0 barP = 0 bar






WeldlineWeldlinePositionPosition

ConventionalConventional

PushPush--PullPull 1 1 strokestroke



PushPush--PullPull 2 2 strokesstrokes


Holding Holding PressurePressureDifferenceDifference ((ΔΔP)P)


5.2.2 Fiber Orientation in Weldline Areas

A typical result of fiber orientation in the weldline area is shown in Figure 5.8. As expected,

it can be seen that fiber orientation in the weldline region consists mainly of fibers which are

parallel to the weldline surface, i.e., perpendicular to the flow direction. This is associated

with the fountain flow phenomena at the melt flow front, as investigated by previous works

[137-142]. As one moves away from the weldline region, the orientation pattern is similar to

that observed in the non-welded specimens.

Figure 5.8 Optical micrographs showing the fiber orientation distribution in the weldline

region of the PC filled with 35 wt% of short-glass-fiber at the midplane location (X-Y plane).

Weldline region

Fountain flow at

the melt flow front


69

The micrographs of the weldline geometries and the fiber orientation pattern across the

thickness of the push-pull processed specimens are illustrated in Figures 5.9 - 5.11 and the

measured values of orientation tensor (a11) within weldline area are shown in Figure 5.12. It

has been found that in the surface layer of weldline position, the fibers still show

perpendicular alignment to the flow direction. This is due to the fact that the melt begins to

solidify as soon as it comes into contact with the cold mold wall. For the push-pull 1 stroke

and 2 strokes processed specimens, it can be clearly seen that the fibers within the weldline

region are highly aligned parallel to the local flow direction and the degree of fiber

orientation in the core layer of push-pull 1 stroke processed specimen increases with

increasing holding pressure difference (see Figure 5.12). However, in the case of the push-

pull 3-strokes processed sample, it should be noted that the third stroke of push-pull does not

produce any major changes in the fiber orientation within the weldline region, particularly in

the core layer, where the fibers are preferentially oriented perpendicular to the flow direction,

as presented in Figures 5.11a. This can be attributed to the heat transfer characteristic of

molten polymer, as reported by Moldflow database, where the thermal conductivity of the

unfilled PC (Makrolon 2800) was 0.173 W/m.K and 0.209 W/m.K for PC filled with 35 wt%

short-glass-fibers (see Table 4.7). In the case of PC filled with 35 wt% short-glass-fibers the

viscosity increases more rapidly and reaches its no-flow temperature sooner than for the

unfilled PC, and thus it could not be manipulated as much as the unfilled PC with the lower

thermal conductivity (see Figure 5.11b). Therefore, the molten core is expected to become

thinner during the push-pull process as – stroke by stroke – solidified layers deposit on the

mold wall, so that the resistance against the screw motion increases.


Figure 5.9 Optical micrographs showing the fiber orientation pattern across the thickness (Z-

Y plane) of the push-pull 1 stroke processed specimen.

1 2

3

44

12

34

5

5


71

Figure 5.10 Optical micrographs showing the fiber orientation pattern across the thickness

(Z-Y plane) of the push-pull 2 strokes processed specimen.

12

1st 2nd

1

2


Figure 5.11 (a) Optical micrographs showing the fiber orientation pattern across the thickness

(Z-Y plane) of the push-pull 3 strokes processed specimen and (b) Weldline geometry of

push-pull 3 strokes processed specimen for unfilled PC.

1st

3rd

Fiber alignment perpendicular to the flow direction

1st

3rd

(a)

(b)


73

Figure 5.12 Variation of the 11a component of orientation tensor through the half thickness of

conventional and push-pull processed specimens in the weldline area.

5.2.3 Effects of holding pressure difference and fiber concentration on penetration

length of weldline

The optical micrographs of penetration length of weldline for the unfilled PC and PC filled

with different glass-fiber contents (20 and 35 wt %) using various holding pressure

differences are shown in Figures 5.13 to 5.15. It can be seen that the higher the difference in

holding pressure, the longer is the penetration length of weldlines, as expected. However, it

should be noted that the relationship between the pressure difference and the penetration

length of weldline is not linear, (see Figure 5.16) suggesting that this increase in the

penetration length of weldline is not solely due to an increase in holding pressure differences,

but may involve some other parameters including the compressibility of molten polymer

during the holding stage and also the effect of pressure on the viscosity of melt, in that the

greater the pressure the higher the melt viscosity [143]. In comparing the penetration length

0.0 0.2 0.4 0.6 0.8 1.00.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Conventional Push-pull 1 stroke (ΔP = 70 bar) Push-pull 1 stroke (ΔP = 120 bar) Push-pull 1 stroke (ΔP = 220 bar) Push-pull 2 strokes (ΔP = 120 bar) Push-pull 3 strokes (ΔP = 120 bar)

Experimental results

MidplaneSurface

Orie

ntat

ion

Tens

or C

ompo

nent

(a11

)

Orie

ntat

ion

Tens

or C

ompo

nent

(a11

)


0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0


of weldline for filled and unfilled PC, it can be observed that the higher the glass fiber

content added into the polymer melt, the shorter the penetration length of the weldline. For

the maximum holding pressure difference, it has been found that the maximum penetration

length of the unfilled PC melt is around 24-25 mm, while in the case of filled PC melt (20

and 35 wt%), it is around 16-18 mm away from the initial weldline position. The decrease in

penetration length of weldline in filled PC melt can be attributed to an increase in viscosity in

the flowing polymer melt [144]. This appears to contradict results of a previous work on the

filled and unfilled PC in which the shortest penetration length of weldline for unfilled PC was

reported [145].

Figure 5.13 Penetration length of weldline across the sample thickness of the molded part for

the unfilled PC at various holding pressure differences.

ΔP = 0 bar

ΔP = 70 bar

ΔP = 120 bar

ΔP = 220 bar

Penetration length of weldline


75


the PC filled with 20 wt% short glass fibers at various holding pressure differences.


the PC filled with 35 wt% short glass fibers at various holding pressure differences.

ΔP = 0 bar

ΔP = 70 bar

ΔP = 120 bar

ΔP = 220 bar


ΔP = 0 bar

ΔP = 70 bar

ΔP = 120 bar

ΔP = 220 bar



Figure 5.16 Relationship between the holding pressure difference and penetration length of

weldline for PC with various glass-fiber contents.

5.2.4 Fiber length distribution in weldline areas

The histogram representing the fiber length distribution for the granules of PC filled with 20

and 35 wt% short-glass-fibers (20G and 35G) is shown in Figure 5.17a. In general, it can be

seen that the higher the fiber loading, the shorter the fiber length, as has been reported by

many authors [82, 87, 92-95, 99]. A further fragmentation of fibers also occurred during

injection molding as presented in Figure 5.17b. Our experiments with simple geometry

showed that the reduction of fiber length is approximately 8-14 %. This is due to the high

shear rates during injection molding when the melt containing fibers has to pass through gates

which are primarily narrow channels of flow. Table 5.1 shows the percent differences

between the mean fiber length of the granules and the mean fiber length ( lΔ% ) within the

weldline area of conventional and push-pull processed parts. It has been found that the effects

of holding pressure difference and the number of push-pull strokes do not lead to significant

changes of fiber length. This may be due to the fact that fiber length is always reduced to a

0 50 100 150 200 2500

5

10

15

20

25Pe

netra

tion

Leng

th o

f Wel

dlin

e (m

m)

Difference of Holding Pressure, ΔP (bar)

PC / PC SFRPC20 / SFRPC20 SFRPC35 / SFRPC35


77

limiting value depending on melt viscosity, the intensity of the shear field and the residence

time [91-92] and thus the fibers can not undergo further fragmentation during push-pull

processing.

Figure 5.17 Histograms representing fiber length distribution (a) within PC granules

containing 20 and 35 wt% short glass fibers and (b) within injection molded article without

weldline.

0-25

25-5

050

-75

75-1

0012

5-15

015

0-17

517

5-20

020

0-22

522

5-25

025

0-27

527

5-30

030

0-32

532

5-35

035

0-37

537

5-40

040

0-42

542

5-45

045

0-47

547

5-50

050

0-52

552

5-55

055

0-57

557

5-60

062

5-65

065

0-67

567

5-70

070

0-72

5

0.00

0.05

0.10

0.15

0.20

0.25

0.30

Average aspect ratio (L/D) = 7.866

G35

Average aspect ratio (L/D) = 10.093μ = 172.209 μm

μ = 129.879 μm

Rel

ativ

e Fi

ber L

engt

h D

istr

ibut

ion

Fiber Length (μm)

G20

Fiber length (μm)

(a)

(b)

0-25

25-5

050

-75

75-1

0012

5-15

015

0-17

517

5-20

020

0-22

522

5-25

025

0-27

527

5-30

030

0-32

532

5-35

035

0-37

537

5-40

040

0-42

542

5-45

045

0-47

547

5-50

050

0-52

552

5-55

055

0-57

557

5-60

062

5-65

065

0-67

567

5-70

070

0-72

5

0.00

0.05

0.10

0.15

0.20

0.25

0.30

SFRPC35Average aspect ratio (L/D) = 7.088

μ = 114.586 μm

Average aspect ratio (L/D) = 9.691

μ = 151.416 μm

Rel

ativ

e Fi

ber L

engt

h D

istr

ibut

ion

Fiber length (μm)

SFRPC20

Fiber length (μm)


Table 5.1 Fiber length in the weldline area of conventional and push-pull processed parts.

Difference of Holding Average Fiber Average Aspect (# 1st / # 2nd) Pressure, ΔP (bar) Length (μm) Ratio (L/D)

70 157.028 10.183 8.815

120 150.395 9.703 12.667

220 152.689 9.821 11.335

70 116.802 6.913 10.069

120 112.120 7.159 13.673

220 116.223 6.966 10.514



SFRPC20 / SFRPC20

Push-Pull 2 strokes

Push-Pull 3 strokes

11.260SFRPC35 / SFRPC35 120 115.254 7.086

Sample Code

%Δ l

SFRPC20 / SFRPC20 120 148.754 9.744 13.620

10.412

SFRPC35 / SFRPC35 120 118.736 6.853 8.580

SFRPC20 / SFRPC20 120 154.278 10.143

SFRPC35 / SFRPC35

Sample Code

%Δ l

Sample Code Push-Pull 1 stroke

%Δ l


79

5.2.5 Weldline strength

The photographs of the samples broken during the tensile testing are illustrated in Figure 5.18.

In the case of SFRPC20/SFRPC20 and SFRPC35/SFRPC35, the tensile failure was brittle

without any neck formation and it occurred at the weldline position. The explanation for this

would be related to the existence of the perpendicular fiber alignment around the weldline

surface, which can be a source of stress concentration within the part surface and thus it

would be easy to break at this position when compared to another position away from the

weldline area. On the other hand, for the unfilled PC specimens, the failure was ductile with

necking initiated across the gauge length. This behavior is also in accordance with previous

observation [145].

Figure 5.18 Appearance of fractured specimens of SFRPC35 / SFRPC35.

The maximum tensile stresses of conventional and push-pull processed parts, for PC

containing different short-glass-fiber contents (0, 20, and 35 %wt) are shown in Figure 5.19.

For the unfilled PC, it is found that the presence of the weldline and the effect of push-pull

processing do not have any significant influence on the maximum tensile stress. While, in the

case of short glass fiber reinforced PC, the maximum tensile stress of the samples with







WeldlineWeldlinePositionPosition

ConventionalConventional






Holding Holding PressurePressureDifferenceDifference ((ΔΔP)P)


weldline is significantly lower than the values obtained for the samples without weldline.

Furthermore, the weldline strength of injection molded PC composites is found to decrease

with the content of reinforcing fibers in the composites. These behaviors are also in

accordance with previous investigations using different composite materials [71, 137-142,

145-146].

Figure 5.19 Maximum tensile stress of conventional and push-pull processed specimens

containing different glass fiber contents.

In comparing the weldline strength of injection molded parts produced by conventional and

push-pull techniques, it can be seen that the weldline strength of the push-pull 1 stroke

processed parts increases with increasing the holding pressure differences. In case of the

highest holding pressure difference (�P = 220 bar), the weldline strength is higher than that

of the conventional molding. In fact the differences are in the range of 40-45%. For the push-

pull 2 strokes processed part, it is also observed that the weldline strength is superior to the

one produced by the conventional injection molding (approximately 25-30 %). This increase

in the weldline strength is supposed to be caused by the higher degree of fiber orientation

PPP

1 st

roke

(Δ

P=70

bar

)

PPP

1 st

roke

(Δ

P=12

0 ba

r)

PPP

3 st

roke

s(Δ

P=12

0 ba

r)

With

out

Wel

dlin

e

PPP

1 st

roke

(ΔP=

220

bar)

PPP

2 st

roke

s (Δ

P=12

0 ba

r)

Wel

dlin

e

0

20

40

60

80

100

120

140

Max

imum

Ten

sile

Str

ess

(MPa

)


0

20

40

60

80

100

120

140

Max

imum

Ten

sile

Str

ess

(MPa

)

PPP

1 st

roke

(Δ

P=70

bar

)

PPP

1 st

roke

(Δ

P=12

0 ba

r)

PPP

3 st

roke

s(Δ

P=12

0 ba

r)

With

out

Wel

dlin

e

PPP

1 st

roke

(ΔP=

220

bar)

PPP

2 st

roke

s (Δ

P=12

0 ba

r)

Wel

dlin

e

0

20

40

60

80

100

120

140

Max

imum

Ten

sile

Str

ess

(MPa

)


0

20

40

60

80

100

120

140

Max

imum

Ten

sile

Str

ess

(MPa

)


81

along the flow direction within the weldline areas, as mentioned earlier. However, there is

only minor difference in weldline strength observed between push-pull 3 strokes and

conventional injection moldings (around 7-10%). This is due to the perpendicular fiber

alignment in weldline area particularly in the core layer of push-pull 3 strokes processed part,

as stated earlier.

5.3 Prediction of the tensile strength for short fiber reinforced composites

The purpose of developing a theoretical model is to explain and predict the experimental

results. Additionally, the theoretical model should also be able to be verified by the existing

experimental results. As described in the section 2.4.1, the critical fiber length ( cl ) and the

interfacial shear strength between fiber and matrix (τ ) can be calculated by Equations 2.5

and 2.6 for given composites system if the fiber length distribution, the fiber diameter ( d )

and the matrix strength ( mσ ) are given. The required parameters for predicting the strength of

short fiber reinforced composites are given in Tables 5.2 to 5.4. Since the fiber volume

fraction, the matrix strength and the area fraction between skin and core layer have been

given experimentally (the measured data of fiber orientation and length, glass fiber content of

granules and area fraction of skin and core layers are summarized in Appendices A to F),

then the predicted values of UTS for the conventional, sandwich and push-pull injection

molded composites can be estimated following Equations 2.13 to 2.16. The theoretically

calculated results, together with the experimentally determined UTS and weldline strength

are shown in Figures 5.20 to 5.22. It can be seen from Figure 5.20 that the model predictions

show a reasonable agreement with the experimental values, although there are still some

differences in both the cases. The calculated results indicate that the UTS increases with the

increase of fiber volume fraction and the UTS of PP sandwich injected with glass fiber

reinforced PP (PP/SFRPP and SFRPP/PP) are at an intermediate level between those of PP

and glass fiber reinforced PP alone. In addition the predicted results also show the UTS of

sandwich injection moldings are higher than that of single injection moldings. This is due to a

higher degree of fiber orientation within the core layer (or a higher area fraction of skin layer,

C

L

AA ) of sandwich injection molded composites, as mentioned in section 5.1.4, though the


fiber length within the core layer of the sandwich moldings are slightly lower than the values

obtained for the single injection moldings.

Figure 5.20 Comparison of experimental and theoretically calculated UTS results for

conventional and sandwich injection molded short glass fiber reinforced polypropylene.

SFR

PP20

/PP

PP/S

FRPP

20

SFR

PP20

SFR

PP20

/SFR

PP20

SFR

PP40

/PP

PP/S

FRPP

40

SFR

PP40

/SFR

PP20

SFR

PP40

SFR

PP40

/SFR

PP40

0

20

40

60

80

100

120

M

axim

um T

ensi

le S

tres

s (M

Pa)

Max

imum

Ten

sile

Str

ess

(MPa

)

Experimental Results Theoretically Calculated Results

0

20

40

60

80

100

120


83

SFR

PP20

/PP

PP/S

FRPP

20SF

RPP

20SF

RPP

20/S

FRPP

20SF

RPP

40/P

PPP

/SFR

PP40

SFR

PP40

/SFR

PP20

SFR

PP40

/SFR

PP40

SFR

PP40

σm

(M

Pa)

28.5

28.5

28.5

28.5

28.5

28.5

28.5

28.5

28.5

Mea

sure

men

tτ

(MPa

)16

.45

16.4

516

.45

16.4

516

.45

16.4

516

.45

16.4

516

.45

Cal

cula

ted

[79]

σf (M

Pa)

3450

3450

3450

3450

3450

3450

3450

3450

3450

[76]

d (μ

m)

1212

1212

1212

1212

12M

easu

rem

ent

l c (μ

m)

1258

.36

1258

.36

1258

.36

1258

.36

1258

.36

1258

.36

1258

.36

1258

.36

1258

.36

Cal

cula

ted

[79]

Vf

0.2

0.2

0.2

0.2

0.4

0.4

0.4/

0.2

0.4

0.4

Mea

sure

men

t

Vm

0.8

0.8

0.8

0.8

0.6

0.6

0.6/

0.8

0.6

0.6

Mea

sure

men

t

AL/A

C

__

0.91

30.

965

__

_0.

938

0.83

4M

easu

rem

ent

AT

/AC

_

0.04

0.08

70.

035

_0.

063

0.05

50.

062

0.16

6M

easu

rem

ent

AS

kin/A

C0.

440.

42_

_0.

410.

420.

42_

_M

easu

rem

ent

AC

ore/A

C0.

560.

54_

_0.

590.

517

0.52

5_

_M

easu

rem

ent

Para

met

ers

Sam

ples

Sour

ce

Tabl

e 5.

2 P

aram

eter

s us

ed in

the

theo

retic

al c

alcu

latio

n of

UTS

for

sing

le a

nd s

andw

ich

inje

ctio

n m

olde

d sh

ort g

lass

fibe

r

rein

forc

ed P

P c

ompo

site

s.


PPP

2 st

roke

sPP

P 3

stro

kes

With

out

( ΔP

= 70

bar

)( Δ

P =

120

bar)

( ΔP

= 22

0 ba

r)( Δ

P =

120

bar)

( ΔP

= 12

0 ba

r)W

eldl

ine

σm

(M

Pa)

60.3

360

.33

60.3

360

.33

60.3

360

.33

60.3

3M

easu

rem

ent

τ (M

Pa)

34.8

334

.83

34.8

334

.83

34.8

334

.83

34.8

3C

alcu

late

d [7

9]σ

f (M

Pa)

3450

3450

3450

3450

3450

3450

3450

[76]

d (μ

m)

1515

1515

1515

15M

easu

rem

ent

l c (μ

m)

742.

8974

2.89

742.

8974

2.89

742.

8974

2.89

742.

89C

alcu

late

d [7

9]

Vf

0.2

0.2

0.2

0.2

0.2

0.2

0.2

Mea

sure

men

t

Vm

0.8

0.8

0.8

0.8

0.8

0.8

0.8

Mea

sure

men

t

AS

kin/A

C

0.44

0.76

0.61

0.48

0.60

0.56

0.92

Mea

sure

men

t

AC

ore/A

C

0.56

0.24

0.39

0.52

0.40

0.44

0.08

Mea

sure

men

t

PPP

2 st

roke

sPP

P 3

stro

kes

With

out

( ΔP

= 70

bar

)( Δ

P =

120

bar)

( ΔP

= 22

0 ba

r)( Δ

P =

120

bar)

( ΔP

= 12

0 ba

r)W

eldl

ine

σm

(M

Pa)

60.3

360

.33

60.3

360

.33

60.3

360

.33

60.3

3M

easu

rem

ent

τ (M

Pa)

34.8

334

.83

34.8

334

.83

34.8

334

.83

34.8

3C

alcu

late

d [7

9]σ

f (M

Pa)

3450

3450

3450

3450

3450

3450

3450

[76]

d (μ

m)

1515

1515

1515

15M

easu

rem

ent

l c (μ

m)

742.

8974

2.89

742.

8974

2.89

742.

8974

2.89

742.

89C

alcu

late

d [7

9]

Vf

0.35

0.35

0.35

0.35

0.35

0.35

0.35

Mea

sure

men

t

Vm

0.65

0.65

0.65

0.65

0.65

0.65

0.65

Mea

sure

men

t

AS

kin/A

C

0.41

0.81

0.62

0.52

0.59

0.50

0.89

Mea

sure

men

t

AC

ore/A

C

0.59

0.19

0.38

0.48

0.41

0.50

0.11

Mea

sure

men

t

Para

met

ers

Sour

ceW

eldl

ine

SFR

PC20

/SFR

PC20

PPP

1 st

roke

Para

met

ers

SFR

PC35

/SFR

PC35

Sour

ceW

eldl

ine

PPP

1 st

roke

Tabl

e 5.

3 P

aram

eter

s us

ed in

the

theo

retic

al c

alcu

latio

n of

wel

dlin

e st

reng

th fo

r con

vent

iona

l and

pus

h-pu

ll in

ject

ion

mol

ded

shor

t gla

ss fi

ber r

einf

orce

d P

C c

ompo

site

s.


85

Table 5.4 Mean glass fiber length of injection molded composites.

A comparison between the experimental and predicted results of weldline strength for

conventional and push-pull injection molded short glass fiber reinforced PC composites are

illustrated in Figures 5.21 and 5.22. The predicted results are in satisfactory agreement with

the experiments in that the weldline strength of injection molded composites decreases with

the increase of the fiber volume fraction and the UTS of weldline-containing parts is very

much lower than the values obtained for the parts without weldline. Furthermore, it is evident

that the weldline strength increases with the increase of holding pressure difference (�P) and

the increasing of number of strokes does not have any significant influence on the UTS.

However, it should be noted that the predicted UTS results are still higher than the

experimental ones. The reasons for this are twofold; firstly, the parameters used in the

calculation ( fσ and τ ) are given by various independent methods from literature [76, 79].

The conditions of the tests may be different, which would lead to an error in the calculation.

It is observed in Equations 2.2 to 2.5 that all the orientation measures are independent of the

fiber strength. Only the fiber length efficiency factor ( lf ) depends on the fiber strength,

SFRPP20 (Skin layer) 333.04

SFRPP20 (Core layer) 375.65

SFRPP40 (Skin layer) 184.38 Conventional (Weldline) 151.416

SFRPP40 (Core layer) 232.22 Push-pull 1 stroke (ΔP = 70 bar) 157.028

Push-pull 1 stroke (ΔP = 120 bar) 150.395

Push-pull 1 stroke (ΔP = 220 bar) 152.689

Push-pull 2 strokes (ΔP = 120 bar) 154.278

SFRPP20/SFRPP20 (Skin layer) 332.73 Push-pull 3 strokes (ΔP = 120 bar) 148.754

SFRPP20/SFRPP20 (Core layer) 370.93 Conventional (Without Weldline) 151.416

SFRPP40/SFRPP40 (Skin layer) 183.15

SFRPP40/SFRPP40 (Core layer) 230.95

SFRPP20/PP (Skin layer) 344.05 Conventional (Weldline) 114.586

PP/SFRPP20 (Core layer) 379.81 Push-pull 1 stroke (ΔP = 70 bar) 116.802

SFRPP40/PP (Skin layer) 190.68 Push-pull 1 stroke (ΔP = 120 bar) 112.12

PP/SFRPP40 (Core layer) 224.24 Push-pull 1 stroke (ΔP = 220 bar) 116.223

SFRPP40/SFRPP20 (Skin layer) 193.19 Push-pull 2 strokes (ΔP = 120 bar) 118.736

SFRPP40/SFRPP20 (Core layer) 373.68 Push-pull 3 strokes (ΔP = 120 bar) 115.254

Conventional (Without Weldline) 114.586

SFRPC35/SFRPC35

μ

μ

Push-pull processed parts

SFRPC20/SFRPC20

Single moldings μ

Sandwich moldings μ

(μm)

(μm)

(μm)

(μm)


which is the value known with the least degree of accuracy. Therefore, if fσ is not known

with sufficient accuracy, the absolute value of lf may be inaccurate. Secondly, the errors

may arise due to the consideration of the uniform fiber alignment, flaw-free molding, and

both the longitudinal and transverse layers experience the same strain, whereas these

assumptions are difficult to obtain in thermoplastic composites.

Figure 5.21 Comparison of experimental and theoretically calculated weldline strength for

conventional and push-pull injection moldings with 20 %wt short glass fiber reinforced

polycarbonate.

0

20

40

60

80

100

120

140

Max

imum

Ten

sile

Str

ess

(MPa

)

Max

imum

Ten

sile

Str

ess

(MPa

)

Experimental Results for SFRPC20/SFRPC/20

Theoretically Calculated Results for SFRPC20/SFRPC20

0

20

40

60

80

100

120

140

PPP

1 st

roke

(Δ

P=70

bar

)

PPP

1 st

roke

(Δ

P=12

0 ba

r)

PPP

3 st

roke

s(Δ

P=12

0 ba

r)

With

out

Wel

dlin

e

PPP

1 st

roke

(ΔP=

220

bar)

PPP

2 st

roke

s (Δ

P=12

0 ba

r)

Wel

dlin

e


87

Figure 5.22 Comparison of experimental and theoretically calculated weldline strength for

conventional and push-pull injection moldings with 35 %wt short glass fiber reinforced

polycarbonate.

0

20

40

60

80

100

120

140

Max

imum

Ten

sile

Str

ess

(MPa

)

Max

imum

Ten

sile

Str

ess

(MPa

)

Experimental Results for SFRPC35/SFRPC35

Theoretically Calculated Results for SFRPC35/SFRPC35

0

20

40

60

80

100

120

140

PPP

1 st

roke

(Δ

P=70

bar

)

PPP

1 st

roke

(Δ

P=12

0 ba

r)

PPP

3 st

roke

s(Δ

P=12

0 ba

r)

With

out

Wel

dlin

e

PPP

1 st

roke

(ΔP=

220

bar)

PPP

2 st

roke

s (Δ

P=12

0 ba

r)

Wel

dlin

e

6. Comparison between Simulation and

Experiment

6.1 Sandwich injection molding

6.1.1 Effect of skin/core volume fraction on the skin/core material distribution

In sandwich injection molding, one of the major tasks is to find out the proper ratio between

the skin and the core materials which is needed to obtain an optimum skin/core sandwich

structure in the molded part. In this study, the virgin polystyrene (PS) was employed in order

to investigate the effect of processing parameters on the skin/core material distribution. In the

case of a simple mold geometry (dumbbell shaped parts), three volume fractions of the core

material (in terms of cavity volume percentage, vol.%), ranging from 55 to 65 vol.% were

injected with the melt temperature of 230 °C. The mold temperature was set at 40 °C and the

skin and core injection flow rates were kept at 18.5 and 27.0 ccm/s, respectively.

Figure 6.1 shows the experimental and numerical results of the effect of varying core volume

fractions on the skin/core material distribution. It is found that the core volume fraction of 60

vol.% produces sandwich injection molded parts without any defect, whereas the lowest core

volume fraction (55 vol.%) shows a large amount of skin material at the far end of the cavity

due to less penetration of the core melt along the flow direction. On the contrary, increasing

the core volume fraction from 60 to 65 vol.%, results in a breakthrough phenomenon because

the amount of skin material is too low and the core melt can easily catch up with the flow

front of the skin material, which generates a defective part and leads to the molding being

discarded.

Comparison between Simulation and Experiment 90

Figure 6.1 Influence of core volume fraction on the material distribution at the end of filling

process: (a) Experiment and (b) Simulation.

For purpose of comparison, a heat transfer coefficient (�) of 1,200 W/m2-K [147] is used in

this calculation and the wall thickness of the model is divided into 20 layers. It can be seen

from Figure 6.1b that the resemblance between numerical simulation and experiment is

strikingly good. Moreover, the prediction by simulation program is also in accordance with

the measured data obtained from the Image-Pro Plus Analysis software, as shown in Figure

6.2, which represents the thickness fraction of the core material at various positions along the

bar. However, it should be noted that the values for the thickness fraction of the core material

are slightly different in the simulation and the experimental results. The measured values are

higher than the predicted values. Following are some possible reasons for these discrepancies.

Firstly, the thermal conductivity of polymer (k) used in the calculations is assumed to be

constant [44]. However, it was found that this property varied considerably with temperature

BreakthroughBreakthrough of of corecore material material at at thethe end of end of thethe barbar

55 vol.% of Core Material



(a) Experimental Results

(b) Simulation Results

55 55 vol.%vol.% of of CoreCore MaterialMaterial



Distribution of polymer A and B

Polymer B

Polymer A

Thickness fraction, polymer B

0.96260.72200.48130.24070.000

Comparison between Simulation and Experiment

91

[148]. Secondly, the heat transfer coefficient between the polymer and the mold metal is

more significant for thin wall moldings. In this study, the calculated values of thickness

fraction were still lower than the measured ones, although the suggested value of 1,200

W/m2-K was employed in the simulation [147]. According to a recent measurement of the

heat transfer coefficient [149], it was found that this value is approximately 550 W/m2-K.

Thirdly, the errors may arise due to the use of dimensional analysis to simplify the governing

equations, which assumes the cavity to be thin and flat in that the ratio of cavity thickness to

cavity length is much lower than unity ( 1⟨⟨= LHδ ) [44]. Therefore, errors can occur in the

region containing out-of-plane flow, such as thick sections of the part, which cannot be

accurately modeled by 2.5-D analysis. Finally, the errors may arise due to the assumption of a

steady state and incompressibility of the fluid in the calculations, whereas these are difficult

to achieve during the processing of the polymeric materials.

Figure 6.2 Effect of core volume fraction on the thickness fraction of core material at various

positions of specimen: (a) Experiment and (b) Simulation.

0.0 0.2 0.4 0.6 0.8 1.00.00.10.20.30.40.50.60.70.80.91.0

0.00.10.20.30.40.50.60.70.80.91.0

Relative Cavity Length (xi / L0)

Simulation Results 55 vol.% of Core Material 60 vol.% of Core Material 65 vol.% of Core Material

Thic

knes

s Fr

actio

n of

Cor

e M

ater

ial (δ

/ b)

Experimental Results 55 vol.% of Core Material 60 vol.% of Core Material 65 vol.% of Core Material (a)

(b)

Breakthrough of Core Material


6.1.2 Effect of processing parameters on the skin/core material distribution

6.1.2.1 Effect of skin and core melt temperatures

The effect of skin and core melt temperature on the thickness fraction of the core material is

shown in Figures 6.3 and 6.4. The injection flow rates of skin and core melt were set at 18.5

and 27.0 ccm/s, respectively, while the injection volumes of skin and core polymers were

kept at 40 and 60 vol.%, respectively. Experimental results are in good agreement with the

numerical simulation results. In Figures 6.3a and 6.4a, only the skin melt temperature is

varied and the core melt temperature is kept unchanged. It can be seen that an increase in the

skin melt temperature induces a decrease in skin thickness, which results in a thicker core

layer near the gate region and a larger amount of skin polymer in areas that are remote from

the gate. In turn, for the lower skin melt temperature, the thicker skin layer is formed near the

gate area. This can also lead to an increase in the penetration length of the core melt at the far

end of the cavity.

Figure 6.3 Influence of skin/core melt temperature on material distribution at the end of

filling process: (a) Effect of skin melt temperature and (b) Effect of core melt temperature.

Skin / Core = 210oC / 230oC






Skin / Skin / CoreCore = 210= 210ooC / 230C / 230ooC C







0.96260.72200.48130.24070.000

Experimental results Simulation results

(a)

(b)


93

Figure 6.4 Comparison of predicted and experimental results: (a) Effect of skin melt

temperature and (b) Effect of core melt temperature on core thickness fraction of molded part

at various positions of tensile specimens.

0.0 0.2 0.4 0.6 0.8 1.00.00.10.20.30.40.50.60.70.80.91.0

0.00.10.20.30.40.50.60.70.80.91.0

Thic

knes

s Fr

actio

n of

Cor

e M

ater

ial (δ

/ b)


Simulation Results Skin Temp./Core Temp. =230 / 210 oC Skin Temp./Core Temp. =230 / 230 oC Skin Temp./Core Temp. =230 / 260 oC

Experimental Results Skin Temp./Core Temp. = 230 / 210 oC Skin Temp./Core Temp. = 230 / 230 oC Skin Temp./Core Temp. = 230 / 260 oC

0.0 0.2 0.4 0.6 0.8 1.00.00.10.20.30.40.50.60.70.80.91.0

0.00.10.20.30.40.50.60.70.80.91.0


Simulation Results Skin Temp./Core Temp. = 210 / 230 oC Skin Temp./Core Temp. = 230 / 230 oC Skin Temp./Core Temp. = 260 / 230 oC

Thic

knes

s Fr

actio

n of

Cor

e M

ater

ial (δ

/ b)

Experimental Results Skin Temp./Core Temp. = 210 / 230 oC Skin Temp./Core Temp. = 230 / 230 oC Skin Temp./Core Temp. = 260 / 230 oC

(a)

(b)


The effect of the core melt temperature on material distribution is shown in Figures 6.3b and

6.4b. The results show that the effect of the core melt temperature on material distribution is

more pronounced than that of the skin melt temperature. The core layer near the gate region

is thicker when the core melt temperature is lower. In addition, when the core material is less

viscous, the core melt front advancement increases substantially. This can not only be

explained by the higher core melt temperature and lower core melt viscosity, but also by the

local rate of cooling at the mold center which is relatively slow when compared to that near

the mold wall, this can be explained by the low thermal conductivity of the polymer melt.

Both of the effects just mentioned can lead to the flow front of the core material stretching

easily which results in a reduction of the thickness of the core near the gate, while the core

thickness away from the gate is increased.

6.1.2.2 Effect of skin and core injection flow rates

Figures 6.5 and 6.6 show the effect of skin and core injection flow rate on the thickness

fraction of the core polymer at the end of the filling process. Both the skin and core melt

temperatures were set at 230 °C and the injection volume of skin and core polymers were 40

and 60 vol. %, respectively. It was found that the effect of the skin injection flow rate does

not lead to significant changes in the skin/core configuration of molded parts, as presented in

Figures 6.5a and 6.6a. In all the cases, only insignificant differences with regard to the

thickness fraction values for the core material can be noted between the measured positions.

Probably this can be attributed to the higher cooling rate at the mold wall, which has a greater

effect on the skin melt viscosity than the rise in temperature caused by shear heating. From

the simulation results, it can also be seen that the higher the skin injection flow rate, the lower

the thickness fraction of the skin layer near the gate. This is associated with the shear

thinning behavior of the polymer melt.


95

Figure 6.5 Influence of skin/core injection flow rate on material distribution at the end of

filling process: (a) Effect of skin injection flow rate and (b) Effect of core injection flow rate.

54.0 54.0 ccm/sccm/s



6.5 6.5 ccm/sccm/s

Breakthrough of core material Breakthrough of core material at the end of the barat the end of the bar

CoreCore injectioninjection flowflow raterate


Polymer B

Polymer A

Skin / Core = 9.25 / 27.0 ccm/s

Skin / Core = 18.5 / 27.0 ccm/s

Skin / Core = 37.0 / 27.0 ccm/s

Skin / Core = 18.5 / 54.0 ccm/s

Skin / Core = 18.5 / 27.0 ccm/s

Skin / Core = 18.5 / 13.5 ccm/s

Skin / Core = 18.5 / 6.5 ccm/s

Skin Skin / / CoreCore = 9.25 = 9.25 / 27.0 / 27.0 ccm/sccm/s



Skin / Core = 18.5 / 54.0 ccm/s

Skin / Core = 18.5 / 27.0 ccm/s

Skin / Core = 18.5 / 13.5 ccm/s

Skin / Core = 18.5 / 6.5 ccm/s


0.96260.72200.48130.24070.000

(a)

(b)

Experimental results Simulation results


Figure 6.6 Comparison of predicted and experimental results: (a) Effect of skin injection flow

rate and (b) Effect of core injection flow rate on core thickness fraction of molded part at

various positions of tensile specimens.

0.0 0.2 0.4 0.6 0.8 1.00.00.10.20.30.40.50.60.70.80.91.0

0.00.10.20.30.40.50.60.70.80.91.0

Thic

knes

s Fr

actio

n of

Cor

e M

ater

ial (δ

/ b)


Simulation Results Skin / Core Injection Flow Rate = 37.0 / 27.0 ccm/s Skin / Core Injection Flow Rate = 18.5 / 27.0 ccm/s Skin / Core Injection Flow Rate = 9.25 / 27.0 ccm/s

Experimental Results Skin / Core Injection Flow Rate = 37.0 / 27.0 ccm/s Skin / Core Injection Flow Rate = 18.5 / 27.0 ccm/s Skin / Core Injection Flow Rate = 9.25 / 27.0 ccm/s

0.0 0.2 0.4 0.6 0.8 1.00.00.10.20.30.40.50.60.70.80.91.0

0.00.10.20.30.40.50.60.70.80.91.0

Thic

knes

s Fr

actio

n of

Cor

e M

ater

ial (δ

/ b)


Simulation Results Skin / Core Injection Flow Rate = 18.5 / 54.0 ccm/s Skin / Core Injection Flow Rate = 18.5 / 27.0 ccm/s Skin / Core Injection Flow Rate = 18.5 / 13.5 ccm/s Skin / Core Injection Flow Rate = 18.5 / 6.5 ccm/s

Experimental Results Skin / Core Injection Flow Rate = 18.5 / 54.0 ccm/s Skin / Core Injection Flow Rate = 18.5 / 27.0 ccm/s Skin / Core Injection Flow Rate = 18.5 / 13.5 ccm/s Skin / Core Injection Flow Rate = 18.5 / 6.5 ccm/s

(a)

(b)


97

A variation of the core injection flow rate has a more significant effect on the material

distribution than a change in the skin injection flow rate, which is evident from Figures 6.5b

and 6.6b. A decrease in the core injection flow rate leads to a significant reduction of the core

thickness fraction near the gate region, while the penetration length of the core polymer

increases substantially and breakthrough occurs at the far end of the cavity. The reason for

this phenomenon can be related to the slower core injection flow rate, leading to the skin

polymer gaining more time to solidify against the mold wall, which results in a thicker

solidified skin layer close to the gate (see Figure 6.7). This can lead to an increase in the core

melt front advancement. In the case of breakthrough, the skin material does not reach the far

end of mold. Consequently, the core melt front catches up with the skin melt front and ends

up at the far end of the cavity. In contrast, the higher core injection flow rate results in a

higher thickness fraction of the core material near the gate leaving a large amount of skin

polymer at the end of the cavity. This can be explained by the fact that the faster the core

injection flow rate, the higher the shear rate near the mold wall; hence the shear thinning

behavior is more pronounced for the skin polymer.

Figure 6.7 Effect of core injection flow rate on thickness distribution of solidified skin

material.

Thicker solidified skin material GateGate

GateGate Thinner solidified skin material

Skin / Core injection flow rate = 18.5 / 6.5 ccm/s

Skin / Core injection flow rate = 18.5 / 54.0 ccm/s


6.1.2.3 Effect of mold temperature

The effect of the mold temperature variation on the skin/core material distribution is shown in

Figure 6.8. In this study the unfilled PC was used for both the skin and core materials, which

were injected with the same melt temperature of 300 °C. The injection volume of core

material was kept at 50 vol.% and the injection flow rate of skin and core melt were

maintained at constant levels of 14.95 and 16.65 ccm/s, respectively. Both of experimental

and predicted results indicate that increasing mold temperature induces an increase in the

core thickness. This is due to the fact that higher mold temperature allows slower cooling of

the polymer melt, resulting in a thinner frozen layer of the skin material, i.e. higher thickness

fraction of core material. On the other hand, at the lower mold temperature, the thicker skin

layer is formed near the gate region. This can also lead to an increase in the penetration

length of the core melt front at the end of the cavity.

Figure 6.8 Effect of mold temperature on the thickness fraction of core material at various

positions of specimen: (a) Experiment and (b) Simulation.

0.0 0.2 0.4 0.6 0.8 1.00.00.10.20.30.40.50.60.70.80.91.00.00.10.20.30.40.50.60.70.80.91.0

Thic

knes

s Fr

actio

n of

Cor

e M

ater

ial (δ

/ b)


Simulation Results Mold Temperature = 40oC Mold Temperature = 80oC Mold Temperature = 120oC

Experiental Results Mold Temperature = 40oC Mold Temperature = 80oC Mold Temperature = 120oC (a)

(b)


99

6.1.3 Effect of glass fiber content on the skin/core material distribution

Figure 6.9 shows the effect of the glass fiber content within the skin material on the thickness

fraction of core material. In this investigation, the unfilled PC, PC filled with 20 and 35 wt%

were injected with the skin flow rate of 14.95 ccm/s and 16.65 ccm/s for the core material.

The mold and melt temperature were set at 300 °C and 80 °C, respectively. It can be seen that

the higher the glass fiber content in the skin material, the thicker the frozen layer of skin

material is formed. This can also be attributed to the heat transfer characteristic of molten

polymer, as mentioned in section 5.2.2, where the thermal conductivity of the PC filled with

35 wt% short-glass-fibers was higher than that of unfilled PC. Thus, the viscosity of PC filled

with 35 wt% short glass fibers increases more rapidly and solidifies sooner than for the

unfilled PC, resulting in the thicker solidified skin layer.

Figure 6.9 Effect of glass fiber content within the skin material on core thickness fraction of

molded part at various positions of tensile specimens: (a) Experiment and (b) Simulation.

0.0 0.2 0.4 0.6 0.8 1.00.00.10.20.30.40.50.60.70.80.91.0

0.00.10.20.30.40.50.60.70.80.91.0

Thic

knes

s Fr

actio

n of

Cor

e M

ater

ial (δ

/ b)


Simulation results PC/PC SFRPC20/PC SFRPC35/PC

Experimental Results PC/PC SFRPC20/PC SFRPC35/PC (a)

(b)


The effect of the glass fiber content within the core material on the thickness fraction of core

material can be observed further in Figure 6.10. The results indicate that the thickness

fraction of core material increases with increasing the glass fiber content. This is associated

with the higher the glass fiber content added into the polymer melt, the higher the viscosity of

the flowing polymer melt [78, 144]. The viscosity versus shear rate curve for three materials

(PC, SFRPC20, and SFRPC35) is shown in Figure 2.1.

Figure 6.10 Effect of glass fiber content within the core material on core thickness fraction of

molded part at various positions of tensile specimens: (a) Experiment and (b) Simulation.

0.0 0.2 0.4 0.6 0.8 1.00.00.10.20.30.40.50.60.70.80.91.0

0.00.10.20.30.40.50.60.70.80.91.0

Thic

knes

s Fr

actio

n of

Cor

e M

ater

ial (δ

/ b)


Simulation Results PC/PC PC/SFRPC20 PC/SFRPC35

Experimental Results PC/PC PC/SFRPC20 PC/SFRPC35 (a)

(b)


101

6.1.4 Case study

This experimental study deals with the sandwich injection of a housing part as illustrated in

Figure 6.11. In this case, the sandwich molding technology allows to combine a conductive

filled skin material with a cheaper unfilled or recycled core material based on the same

material for good skin-core adhesion. The industrial goal is to fill a cheaper or recycled

polymer in the core as much as possible and without breakthrough phenomenon. In this study,

the sequential injection was performed in the following steps, as described in section 1.2.1, a

given percentage of the skin material (PS, transparent) is first injected into the cavity,

followed by the injection of the core material (PS, blue color) and finally a much smaller

amount of the skin material is injected to seal the gate. The material and processing

parameters used for the computer prediction are given in Table 6.1 and the amounts of

injected material (in term of cavity volume percentage) are varied according to Table 6.2.

Figure 6.11 Housing and 2.5-D meshed model used in this study.

5.0004.1503.3002.4501.600

Mesh thickness (mm)


Table 6.1 Material and processing conditions.

Table 6.2 Amount of injected material.

A comparison between the experimental and simulation results for the housing part is

illustrated in Figures 6.12 to 6.13. In Figure 6.12b the simulation result of the core material

distribution is given in terms of core thickness fraction. The blue area designates the area

which is only occupied by the skin material, while the yellow-green areas indicate the areas

occupied by both the skin and core materials. The simulation results of skin and core material

distribution for different injection volume percentage are given in Figure 6.13. In the case of

Run # 1 (A-B-A = 50-45-5 % by volume), the blue areas are the breakthrough areas as

predicted by the numerical simulation where the core component has broken through the skin

component and can be seen at the part surface, which would lead to discarding of molded part

(see Figure 6.13a). On the other hand, by slightly changing the content of skin and core

materials (Run # 2, A-B-A = 55-40-5 % by volume), it can be seen that the entire surface is

occupied by skin material (see Figure 6.13b). Finally, at the end of filling, one observes a

very good agreement between prediction and experiment.

1st-Plasticator 2nd-Plasticator 1st-Plasticator (Skin Material) (Core Material) (Skin Material)

1 50% 45% 5%2 55% 40.5% 4.5%

* Total volume of part = 30 ccm

Run #

Injection Volume (%by volume)

Material Grade SupplierPolystyrene PS 165 H BASF

1st-Plasticator 2nd-Plasticator (Skin Material) (Core Material)

Melt temperature ( oC ) 230 230Injection flow rate (ccm/s) 18.5 27Injection pressure (bar) 1000 1000Holding pressure (bar) _ 800Holding time (sec.) _ 25

Mold temperature = 40 oCTotal cooling time = 40 sec.



103

Figure 6.12 Sandwich injection of a housing part: (a) Experimental and (b) Simulation results.

Figure 6.13 (a) Material distribution of skin and core polymers at the end of filling: Injection

volume of A-B-A = 50- 45-5 % (Run # 1).

The breakthrough of core material


Polymer B

Polymer A


0.96260.72200.48130.24070.000

(a) (b)

Skin material

Skin material


Figure 6.13 (b) Material distribution of skin and core polymers at the end of filling: Injection

volume of A-B-A = 55- 40-5 % (Run # 2).

6.2 Simulation of fiber orientation in sandwich injection molding

Comparison between 2.5-D and 3-D simulation of fiber orientation and measurement

In comparing the 2.5-D simulation results for the fiber orientation tensor ( 11a ) across the half

thickness for single and sandwich injection moldings, as depicted in Figure 6.14, there is no

significant discrepancy in both cases. Furthermore, as can be seen from Figure 6.15 the

predicted values of 11a obtained in 2.5-D model are not in accordance with the measurements.

The simulation results show the higher value of 11a within the core region compared to the

skin region, whilst the measured results show the lower value of 11a within the core layer.

This may be probably caused by some factors. Firstly, it can be associated with the effect of

geometry such a clamping part of tensile specimen where the converging flow field is

established. This can lead to a higher velocity gradient of the flowing melt along the flow

path, and thus resulting in an increase of the fiber orientation within the core layer, as

mentioned in section 5.1.1. Secondly, the errors may arise due to the use of dimensional

analysis to simplify the governing equations, which omits the calculation of velocity

No breakthrough


Polymer B

Polymer A


105

component and thermal convection in the gap wise direction [44-45]. Finally, the Hele-Shaw

flow formulation also neglects the transverse flow at the melt front region (the fountain flow

behavior) which has a significant effect on the evolution of fiber orientation during injection

mold filling [150].

Figure 6.14 2.5-D simulation results for the orientation tensor ( 11a ) component at different

layers over the half thickness of single and sandwich injection molded parts.

Single Molding Sandwich Molding

Flow directionFlow directionFiber orientation tensor

Relative thickness = 0 (Surface)

Relative thickness = 1.000 (Midplane)

Relative thickness = 0.120





Relative thickness = 0 (Surface)

Relative thickness = 1.000 (Midplane)






Converging flow


Figure 6.15 Comparison of the 2.5-D model prediction and measurement of fiber orientation

tensor ( 11a ) across the half thickness at the middle position of single and sandwich injection

moldings (tensile specimen) for PP filled with 20 wt% of short glass fiber.

The 2.5-D model of the rectangular bar (without clamping part) was used in order to neglect

the effect of converging channel. It can be seen from Figure 6.16 that the fiber orientation

predictions of the rectangular bar give a better qualitative agreement with the experimental

measurements for the orientation profiles as compared to that of the tensile geometry, in

which the values of 11a become lower in the core region. However, there is still no significant

difference in the predicted value of 11a between single and sandwich moldings and the

predicted values of 11a within the skin layer obtained from 2.5-D model are still

underestimated.

0.0 0.2 0.4 0.6 0.8 1.00.4

0.5

0.6

0.7

0.8

0.9

1.0

Midplane


SFRPP20 SFRPP20/SFRPP20

Orie

ntat

ion

Tens

or C

ompo

nent

(a11

)

Orie

ntat

ion

Tens

or C

ompo

nent

(a11

)


2.5-D Simulation Results


0.4

0.5

0.6

0.7

0.8

0.9

1.0

Surface


107

Figure 6.16 Comparison of the 2.5-D model prediction and measurement of fiber orientation

tensor ( 11a ) across the half thickness at the middle position of single and sandwich injection

moldings (rectangular specimen) for PP filled with 20 wt% of short glass fiber.

Figure 6.17 represents the 3-D simulation result of fiber orientation tensor ( 11a ) at the end of

filling for single injection molding process (injection flow rate of 18.5 ccm/s). It can be seen

that the predicted values of 11a agree reasonably well with the measurements. The simulation

results show the higher degree of fiber orientation in the skin layer as compared to the core

layer, where the fibers are oriented randomly to the flow direction ( 11a ≈ 0.5). This would be

associated with the effect of geometry i.e. a clamping part of tensile specimen, as stated

earlier. It is also indicated in Figure 6.18a that the calculated values of 11a within the core

region of SFRPP20 are higher than that of SFRPP40. This is due to an increase in the

viscosity of the polymer melt, which has more fiber content [78, 144]. The higher the

viscosity of polymer melts the narrower the shear zone during the filling phase, and thus

resulting in a lower degree of fiber orientation in the core region. Furthermore, from the 3-D

simulation results as illustrated in Figures 6.18b to 6.20, it is interesting to note that the

0.0 0.2 0.4 0.6 0.8 1.00.4

0.5

0.6

0.7

0.8

0.9

1.0



MidplaneSurface

Orie

ntat

ion

Tens

or C

ompo

nent

(a11

)

Orie

ntat

ion

Tens

or C

ompo

nent

(a11

)


2.5-D Simulation Results


0.4

0.5

0.6

0.7

0.8

0.9

1.0


sandwich injection moldings show higher values of 11a within the core region as compared to

the single injection moldings. The change in the 11a values is thought to be caused by the

thicker solidified skin layer which develops during the flow of sandwich injection molded

short fiber composites, as mentioned in section 5.1.1. Therefore, the velocity profile of the

following core melt can be sharper and can cause the fibers in the core layer to be more

aligned in the flow direction.

Figure 6.17 3-D simulation result of fiber orientation tensor ( 11a ) for single injection molded

part.

Fiber orientation tensor (a11)

1.000

0.750

0.500

0.250

0.000


109

Figure 6.18 Comparison of the 3-D model prediction and measurement of 11a across the half

thickness at the middle position: (a) single injection molded specimens for PP filled with 20

wt% and 40 wt% of short glass fiber (b) single and sandwich injection molded specimens for

PP filled with 40 wt% of short glass fiber.

0.0 0.2 0.4 0.6 0.8 1.00.4

0.5

0.6

0.7

0.8

0.9

1.0

Orie

ntat

ion

Tens

or C

ompo

nent

(a11

)

3-D Simulation Results

Single molding Sandwich molding



MidplaneSurface

Orie

ntat

ion

Tens

or C

ompo

nent

(a11

)


0.4

0.5

0.6

0.7

0.8

0.9

1.0

0.0 0.2 0.4 0.6 0.8 1.00.4

0.5

0.6

0.7

0.8

0.9

1.0

Single Injection Moldings

MidplaneSurface


SFRPP20 SFRPP40

Orie

ntat

ion

Tens

or C

ompo

nent

(a11

)

Orie

ntat

ion

Tens

or C

ompo

nent

(a11

)


3-D Simulation Results

SFRPP20 SFRPP40

0.4

0.5

0.6

0.7

0.8

0.9

1.0

(a)

(b)


Figure 6.19 Comparison of the orientation tensor ( 11a ) component across the thickness for

single and sandwich injection moldings.

Figure 6.20 Comparison of the orientation tensor ( 11a ) component at the mid-plane layer (z =

2.0 mm) for single and sandwich injection moldings.

Flow direction

Flow direction

Single Injection Molding

Sandwich Injection Molding

Sandwich Injection Molding Single Injection Molding


111

6.3 Simulation of fiber orientation in push-pull injection molding

Effect of holding pressure difference and the number of push-pull strokes on 3-D fiber

orientation in weldline areas

Figures 6.21 and 6.22 represent the 3-D simulation results of fiber orientation tensor ( 11a ) for

conventional and push-pull injection molding processes. In the case of conventional injection

molding with weldline, the predicted results indicate that the fibers near the part surface are

randomly oriented. Near the midplane of the part (core layer), the fibers are mainly aligned

perpendicular to the flow direction, which is caused by the fountain flow effect at the melt

front [137-142]. For the push-pull injection moldings, the simulation results show that not

only the degree of fiber orientation increases with increasing holding pressure differences,

but also an increase in the number of strokes does not produce any major changes in fiber

orientation within the weldline area compared to push-pull 1 stroke. These findings are also

in good agreement with previous experimental work concerning the fiber orientation

distribution within the weldline areas of push-pull processed parts. In addition, it can be seen

from Figure 6.22 that the predicted values of orientation tensor components ( 11a ) agree

reasonably well with corresponding experimental measurements. The predicted values of 11a

across the part thickness show the same tendency as the measured ones, although there is still

a slight discrepancy in both cases. One possible factor that may cause the differences between

calculation and measurement is the fiber-fiber interaction coefficient (CI). According to

previous findings [119-120, 122] concerning the comparison between the numerical and

experimental fiber orientation in injection molded part, it has been found that the fiber

orientation predictions are quite sensitive to this coefficient especially near the surfaces of the

part. They also suggested that CI = 0.01 gives a good agreement for the skin layer, whereas CI

= 0.001 is a better choice for the core layer.


Figure 6.21 3-D Simulation results of fiber orientation tensor ( 11a ): (a) at midplane layer and

(b) across the part thickness of conventional and push-pull injection moldings, for PC with 35

wt% short glass fibers.

(b)

Push-Pull 1 stroke

Push-Pull 2 strokes

Push-Pull 3 strokesWeldlineWeldline




11stst

11stst 22ndnd

22ndnd11stst

33rdrd

Z

X

Y Z

X

(a)

WeldlineWeldline PositionPositionFlowFlow DirectionDirection FlowFlow DirectionDirection




1st

1st

1st

2nd

2nd

3rd


113

0.0 0.2 0.4 0.6 0.8 1.00.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Orie

ntat

ion

Tens

or C

ompo

nent

(a11

)

MidplaneSurface

Orie

ntat

ion

Tens

or C

ompo

nent

(a11

)


Weldline Push-Pull 1 stroke (ΔP = 70 bar) Push-Pull 1 stroke (ΔP = 120 bar) Push-Pull 1 stroke (ΔP = 220 bar) Push-Pull 2 strokes (ΔP = 120 bar) Push-Pull 3 strokes (ΔP = 120 bar)

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Figure 6.22 Predicted and measured values of 11a orientation tensor component across the

half thickness at the weldline position of conventional and push-pull injection moldings, for

PC with 35 wt% short glass fibers.

Simulation Results


0.0 0.2 0.4 0.6 0.8 1.00.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Orie

ntat

ion

Tens

or C

ompo

nent

(a11

)

MidplaneSurface

Orie

ntat

ion

Tens

or C

ompo

nent

(a11

)


Weldline Push-Pull 1 stroke (ΔP = 70 bar) Push-Pull 1 stroke (ΔP = 120 bar) Push-Pull 1 stroke (ΔP = 220 bar) Push-Pull 2 strokes (ΔP = 120 bar) Push-Pull 3 strokes (ΔP = 120 bar)

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

7. Conclusions

The present study was concentrated on investigating the capabilities of the sandwich and

push-pull injection molding processes to enhance the orientation of fibers and its effect on the

attrition of fiber length since these are critical to the mechanical performance of short fiber

composites. The accuracy of the model prediction was verified by comparing with the

corresponding experimental measurements. Results and conclusions obtained during this

study can be summarized as:

• A sandwich injection molding technique was employed to improve the mechanical

properties of short glass fiber reinforced thermoplastics (SFRTs) with respect to the fiber

orientation and fiber attrition within the skin and core layers. The results confirmed the

expected rise in the maximum tensile stress and impact strength as the concentration of the

short glass fibers was increased. The mechanical properties of sandwich injection moldings

were observed to be higher than those of single injection moldings, which can be attributed to

the higher fiber orientation within the core layer. The results obtained by analyzing the fiber

attrition inside the skin and core regions in the longitudinal direction of tensile specimens

showed that the degree of fiber degradation inside the skin layers was higher than in the core

layers. There were only minor differences in the skin region fiber length observed between

sandwich and single injection molding processes, this effect was more pronounced in the core

region and for the higher fiber concentration.

• A theoretical model derived by an analytical method of modified rule of mixtures

(MROM) as a function of the area fraction between skin and core layers has been introduced

to predict the ultimate tensile strength (UTS) of conventional, sandwich and push-pull

injection molded composites. The effects of fiber length, fiber orientation and fiber content

on the tensile strength of SFRTs were studied in detail. The present model was verified to

Conclusions 116

existing experimental results, and for all cases, the predictions showed satisfying agreement.

The differences between calculated and experimental data may result from some parameters

and assumptions made in the derivation of the equations, which would lead to errors in the

calculations.

• A push-pull injection molding technique was employed to enhance the weldline

strength of short glass fiber reinforced polycarbonate with respect to the fiber orientation and

the fiber length distribution in the weldline areas. The effects of processing parameters

including the number of push-pull strokes, holding pressure differences between both of the

injection units and the effect of glass fiber concentration have been studied. It was found that

the weldline strength of the push-pull 1 stroke processed parts increases with increasing

penetration length of weldline. An increase of the number of strokes did not produce any

major changes in the weldline strength compared to that of push-pull 1 stroke processed part.

The fiber attrition within the weldline area was not significantly affected by the holding

pressure difference and the number of push-pull strokes. An expected nonlinear relationship

between the holding pressure difference and the penetration length of weldline was also

observed, which can be associated with the rheological behavior and heat transfer

characteristics of polymer melt in the cavity.

• The effects of processing parameters and glass fiber concentration on the skin/core

material distribution in sandwich injection molded parts were investigated and extensively

verified against the predicted results performed by the commercial simulation package

(Moldflow). Both the simulated and the experimental results indicated that in order to obtain

an optimum encapsulated skin/core structure in the sandwich injection molded part, it is

necessary to select proper core volume fraction and processing parameters. The results

suggest that the most important processing parameter for controlling the breakthrough

phenomenon was the core injection flow rate, while the skin injection flow rate did not have

any significant influence on the thickness fraction of the core material. The thickness fraction

of core material increases with either an increase in mold temperature and skin melt

temperature or a decrease in core melt temperature. An increase in solidified thickness of skin

material or an increase in thickness fraction of core material, which has higher glass fiber

content can be associated with the heat transfer characteristic and the viscosity of molten

polymer. If a higher proportion of core material is needed for industrial purposes, either the

skin melt temperature or the core injection flow rate has to be increased, or the core melt

Conclusions

117

temperature has to be decreased. A good agreement between simulation and experimental

results indicate that the simulation program can be used as a valuable tool for the prediction

of melt flow behavior during sandwich injection process.

• Structure of fiber orientation in sandwich and push-pull injection molded short fiber

composites have been predicted by the 2.5-D and 3-D numerical analyses. The predictions

solve the full balance equations of mass, momentum, and energy for a generalized Newtonian

fluid. The second-order orientation tensor approach was used to describe and calculate the

local fiber orientation state. Changing the numerical values of orientation tensor ( 11a ) clearly

shows the difference in the capabilities of simulation model. The predicted results obtained

from the 2.5-D was found to be less accurate than that of 3-D model. This is due to the use of

dimensional analysis to simplify the governing equations, which omits the calculation of

velocity component and thermal convection in the gap wise direction. In addition, the Hele-

Shaw flow formulation also neglects the transverse flow at the melt front region (the fountain

flow behavior) which has a significant effect on the evolution of fiber orientation during

injection mold filling. For the 3-D model, the predicted and measured results of 11a were

found to be in a good agreement. However, slight discrepancies were observed at the center

and close to the mold wall which may result from the fiber-fiber interaction coefficient used

in the calculation.

Conclusions 118

8. References

[1] Donald V. Rosato and Dominick V. Rosato, Injection Molding Handbook: The

Complete Molding Operation Technology, Performance, Economics, Second Edition,

Chapman & Hall, New York, 1995.

[2] P.J. Garner and D.F. Oxley, British Patent, 1968, 156(1): 1156-1217.

[3] C. Donovan, K.S. Rabe, W.K. Mammel and H.A. Lord, Recycling plastic by two-shot

molding, Polymer Engineering and Science, 1975, 15(11): 774-780.

[4] W. Fillmann, Verarbeitung treibmittelhaltiger Thermoplaste nach dem

Mehrkomponentenspritzgießverfahren, Kunststoffe, 1977, 67(4): 206-216.

[5] J. MaRoskey, A new low cost co-injection system that reduced product costs, 30th

Annual Conference & Design Recognition of the SPI Structural Plastics Division

(Structural Plastic 2002), 14-16 April 2002, Michigan.

[6] E. Escales, Das ICI-Sandwich-Spritzgießverfahren, Kunststoffe, 1970, 60(11): 847-

852.

[7] H. Eckardt and S. Davies, An introduction to multi-component (sandwich) injection

molding, Plastic & Rubber International, 1979, 4(2): 72-74.

[8] R.A. Malloy, Plastic Part Design for Injection Molding: An Introduction, Hanser

Publishers, New York, 1994.

References

120

[9] K.C. Rush, Gas-assisted injection molding: The new age of plastic molding

technology, Society of Plastics Engineers Annual Technical Conference, Technical

Papers, 1989: 1014-1018.

[10] J. Avery, Gas-Assisted Injection Molding: Principle and Applications, Hanser


[11] W. Michaeli, A. Brunswick and C. Kujat, Reducing cooling time with water-assisted

injection molding, Kunststoffe Plast Europe, 2000, 90(8): 25-28.

[12] E. Costalas and H. Krauss, Optimized bonding of composites in over-injection,

Kunststoffe Plast Europe, 1995, 85(11): 1887-1891.

[13] J. Rothe, Special injection molding methods, Kunststoffe Plast Europe, 1997, 87(11):

1564-1581.

[14] K. Kuhmann and G.W. Ehrenstein, The influence of flow condition and injection

parameters on the bond strength of compatible material combinations, Society of

Plastics Engineers Annual Technical Conference, Technical Papers, 1997: 451-428.

[15] D.B. Tachalamov and A.M. Cunha, Structure development and mechanical properties

of overmolded parts, Society of Plastics Engineers Annual Technical Conference,

Technical Papers, 2003: 725-729.

[16] A.W. Birley, B. Haworth and J. Batchelor, Physics of Plastics: Processing, Properties

and Materials Engineering, Hanser Publishers, New York, 1992.

[17] P.S. Alan and M.J. Bevis, Multiple live-feed injection molding, Plastic Rubber and

Composites Processing and Applications, 1987, 7(1): 3-10.

[18] H. Becker, G. Fischer and U. Müller, Gegentakt-Spritzgießen technischer Formteile,

Kunststoffe, 1993, 83(3): 165-169.

[19] F. Johannaber and W. Michaeli, Handbuch Spritzgießen, Hanser Verlag, 2001.

References 121

[20] P.J. Hine, R.A. Duckett, I.M. Ward, P.S. Alan and M.J. Bevis, A comparison of short

glass fiber reinforced polypropylene plates made by conventional injection molding

and using shear controlled injection molding, Polymer Composites, 1996, 17(3): 400-

407.

[21] R.L. Reis, A.M. Cunha and M.J. Bevis, Shear controlled orientation injection molding

of polymeric composites with enhanced properties, Society of Plastics Engineers

Annual Technical Conference, Technical Papers, 1998: 487-491.

[22] H.-C. Ludwig, G. Fischer and H. Becker, A quantitative comparison of morphology

and fiber orientation in push-pull processed and conventional injection molded parts,

Composites Science and Technology, 1995, 53(2): 235-239.

[23] S. Patcharaphun and G. Mennig, Investigation on weldline strength of short-glass-

fiber reinforced polycarbonate manufactured through push-pull-processing technique,

Journal of Reinforced Plastics and Composites, 2006, 25(4): 421-435.

[24] I. Kühnert, V. Stoll and G. Mennig, Sequential injection molding: The influence on

weldlines, The 21st annual meeting of the polymer processing society (PPS), Leipzig,

Germany, 19-23 June 2005, ISBN 3-86010-784-4.

[25] G. Mennig and I. Kühnert, Untersuchungen und Verbesserung der Bindenahtfestigkeit

von technischen Bauteilen aus kurzfaserverstärkten Thermoplasten,

Forschungsbericht des Forschungskuratoriums Maschinebau e.V., FKM-Heft 283,

AiF-Abschlussbericht 2004.

[26] P.S. Allan and M.J. Bevis, British Patent No. 2170-140-B, 1987.

[27] L. Wang, P.S. Alan and M.J. Bevis, Enhancement of internal weldline strength in

thermotropic liquid crystal polymer moldings, Plastic Rubber and Composites

Processing and Applications, 1995, 23(3): 139-150.

[28] R.S. Spencer and G.D. Gilmore, Some flow phenomena in the injection molding of

polystyrene, Journal of Colloid Science, 1951, 6(2): 118-132.

References

122

[29] R.L. Ballman, T. Schusman and H.L. Toor, Injection molding: Flow of a molten

polymer into a cold cavity, Industrial and Engineering Chemistry, 1959, 51(7): 847-

850.

[30] R.B. Staub, Effect of basic polymer properties on injection molding behavior, Society

of Plastics Engineers Annual Technical Conference, Technical Papers, 1961: 345-348.

[31] D.H. Harry and R.G. Parrot, Numerical simulation of injection molding filling,

Polymer Engineering and Science, 1970, 10(4): 209-214.

[32] G. Williams and H.A. Lord, Mold-filling studies for the injection molding of

thermoplastic materials. Part 1: The flow of plastic materials in hot- and cold-walled

circular channels, Polymer Engineering and Science, 1975, 15(8): 553-568.

[33] P. Thienel and G. Menges, Mathematical and experimental determination of

temperature, velocity, and pressure fields in flat molds during the filling process in

injection molding of thermoplastics, Polymer Engineering and Science, 1978, 18(4):

314-320.

[34] M.R. Kamal and S. Kenig, The injection molding of thermoplastics part 1:

Theoretical model, Polymer Engineering and Science, 1972, 12(4): 294-301.

[35] M.R. Kamal and S. Kenig, The injection molding of thermoplastics part 2:

Experimental test of model, Polymer Engineering and Science, 1972, 12(4): 302-308.

[36] J.L. Berger and C.G. Gogos, A numerical simulation of the cavity filling process with

PVC in injection molding, Polymer Engineering and Science, 1973, 13(2): 102-112.

[37] P.C. Wu, C.F. Huang and C.G. Gogos, Simulation of the mold-filling process,


[38] J.F. Stevenson, A. Galskoy, K.K. Wang, I. Chen and D.H. Reber, Injection molding in

disk-shaped cavities, Polymer Engineering and Science, 1977, 17(9): 706-710.

References 123

[39] H.A. Lord and G. Williams, Mold-filling studies for the injection molding of

thermoplastic materials. Part 2: The transient flow of plastic materials in the cavities

of injection-molding dies, Polymer Engineering and Science, 1975, 15(8): 569-582.

[40] R.E. Nunn and R.T. Fenner, Flow and heat transfer in the nozzle of an injection

molding machine, Polymer Engineering and Science, 1977, 17(11): 811-818.

[41] C.A. Hieber, R.K. Upadhyay and A.I. Isayev, Non-isothermal polymer flow in non-

circular runners, Plastic Engineering, 1983, 39(3): 48.

[42] C.A. Hieber and S.F. Shen, A finite-element/finite-difference simulation of the

injection-molding filling process, Journal of non-Newtonian Fluid Mechanics, 1980,

7(1): 1-32.

[43] E.C. Bernhardt, Computer Aided Engineering for Injection Molding, Hanser


[44] P. Kennedy, Flow Analysis of Injection Molds, Hanser Publishers, New York, 1995.

[45] C.L. Tucker, Computer Modeling for Polymer Processing, Hanser Publishers, New

York, 1995.

[46] H.H. Chiang, C.A. Hieber and K.K. Wang, A unfilled simulation of the filling and

post-filling stages in injection molding Part 1: Formulation, Polymer Engineering and

Science, 1991, 31(2): 116-124.

[47] M. Gupta and K. K. Wang, Fiber orientation and mechanical properties of short-fiber-

reinforced injection-molded composites: Simulated and experimental results, Polymer

Composites, 1993, 14(5): 367-382.

[48] H.H. Chiang, K. Himasekhar, N. Santhanam and K.K. Wang, Integrated simulation of

fluid-flow and heat-transfer in injection-molding for the prediction of shrinkage and

warpage, Journal of Engineering Materials and Technology: Transaction of the

ASME, 1993, 115(1): 37-47.

References

124

[49] J. Wortberg and G. Burmann, Computer design of hot-runner systems, Kunststoffe

1992, 82(2): 91-94.

[50] S. Roth, B. Küster and H. Sura, Practical comparison of simulation techniques: 2.5-D

or 3-D?, Plast Europe, 2004, 94(7): 65-67.

[51] B.E. VerWeyst, C.L. Tucker III, P.H. Foss and J.F. O’Gara, Fiber orientation in 3-D

injection molded features: Prediction and experiment, International Polymer

Processing, 1999, 14(4): 409-420.

[52] J.F. Hetu, D.M. Gao, A. Garcia-Rejon and G. Salloum, 3-D finite element method for

the simulation of the filling stage in injection molding, Polymer Engineering and

Science, 1998, 38(2): 223-236.

[53] J. Zachert and W. Michaeli, Simulation and analysis of three-dimensional polymer

flow in injection molding, Journal of Reinforced Plastics and Composites, 1998,

17(10): 955-962.

[54] R.Y. Chang and W.H. Yang, Numerical simulation of mold filling in injection

molding using a three-dimensional finite volume approach, International Journal for

Numerical Methods in Fluids, 2001, 37: 125-148.

[55] E. Pichelin and T. Coupez, Finite element solution of the 3-D mold filling problem for

viscous incompressible fluid, Computer Methods in Applied Mechanics and

Engineering, 1998, 163(1-4): 359-371.

[56] R. Han, L.H. Shi and M. Gupta, Three-dimensional simulation of microchip

encapsulation process, Polymer Engineering and Science, 2000, 40(3): 776-785.

[57] G.A.A.V. Haagh and F.N. Van de Vosse, Simulation of three-dimensional polymer

mold filling processes using a pseudo-concentration method, International Journal for

Numerical Methods in Fluids, 1998, 28(9): 1355-1369.

References 125

[58] V. Rajupalem, K. Talwar and C. Friedl, Three-dimensional simulation of the injection

molding process, Society of Plastics Engineers Annual Technical Conference,

Technical Papers, 1997: 670-673.

[59] K. Talwar, F. Costa, V. Rajupalem, L. Antanovski and C. Friedl, Three-dimensional

simulation of plastic injection molding, Society of Plastics Engineers Annual

Technical Conference, Technical Papers, 1998: 562-566.

[60] L.S. Turng and V.W. Wang, Simulation of co-injection and gas-assisted injection

molding, Society of Plastics Engineers Annual Technical Conference, Technical

Papers, 1991: 297-299.

[61] G. Schlatter, A. Davidoff, J.F. Agassant and M. Vincent, Numerical simulation of the

sandwich injection molding process, Society of Plastics Engineers Annual Technical

Conference, Technical Papers, 1995: 456-460.

[62] G. Schlatter, A. Davidoff, J.F. Agassant and M. Vincent, An unsteady multi-fluid

flow model: Application to sandwich injection molding process, Polymer Engineering

and Science, 1999, 39(1): 78-88.

[63] D.J. Lee, A.I. Isayev and J.L. White, Simultaneous sandwich injection molding:

Simulation and experiment, Society of Plastics Engineers Annual Technical


[64] C.T. Li, D.J. Lee and A.I. Isayev, Interface and encapsulation in simultaneous co-

injection molding of disk: Two-dimensional simulation and experiment, Society of

Plastics Engineers Annual Technical Conference, Technical Papers, 2002: 465-469.

[65] C. Jaroschek, Passgenaue Verteilung des Kernmaterials, Kunststoffe 2004, 94(5): 68-

71.

[66] S.C. Chen, K.F. Hsu and K.S. Hsu, Analysis and experimental study of gas

penetration in gas-assisted injection-molded spiral tube, Journal of Applied Polymer

Science, 1995, 58(4): 793-799.

References

126

[67] S.C. Chen, K.F. Hsu and K.S. Hsu, Polymer melt flow and gas penetration in gas-

assisted injection molding of a thin part with gas channel design, International Journal

of Heat and Mass Transfer, 1996, 39(14): 3003-3019.

[68] D.M. Gao, K.T. Nguyen, R.A. Garcia and G. Salloum, Numerical modeling of the

mold filling stage in gas-assisted injection molding, International Polymer Processing,

1997, 12(3): 267-277.

[69] T.J. Wang, H.H. Chiang, X. Lu and F. Lihwa, Computer simulation and experimental

verification of gas-assisted injection molding, Journal of Injection Molding

Technology, 1998, 2(3): 120-127.

[70] J.F.T. Pittman, P. Aguirre and D. Ding, Development of a computer simulation of

SCORIM: In-mold shearing with cooling and solidification, International Polymer

Processing, 1996, 11(3): 248-256.

[71] R. Brunotte, I. Kühnert and G. Mennig, Einfluss von pulsierendem Nachdruck auf die

Festigkeit von Bindenähten, Tagungsband, TECHNOMER’03, 18, Fachtagung über

Verarbeitung und Anwendung von Polymeren, 13-15 November 2003, Chemnitz,

ISBN 3-00-012510-8.

[72] R.E. Khayat, A. Derdouri and L.P. Hebert, A 3-dimentional boundary-element

approach to gas-assisted injection molding, Journal of non-Newtonian Fluid

Mechanics, 1995, 57(2-3): 253-270.

[73] G.A.A.V. Haagh, H. Zuidema, F.N. Van de Vosse, G.W.M. Peters and H.E.H. Meijer,

Towards a 3-D finite element model for the gas-assisted injection process,

International Polymer Processing, 1997, 12(3): 207-215.

[74] F. Ilinca and J.F. Hetu, Three-dimensional finite element solution of gas-assisted

injection molding, International Journal for Numerical Methods in Engineering, 2002,

53(8): 2003-2017.

References 127

[75] F. Ilinca and J.F. Hetu, Three-dimensional simulation of multi-material injection

molding: Application to gas-assisted and co-injection molding, Polymer Engineering

and Science, 2003, 43(7): 1415-1427.

[76] S.M. Lee, Handbook of Composite Reinforcements, VCH Publishers, New York,

1992.

[77] D.W. Clegg, Recycling of polymer composites, Handbook of Polymer-Fiber

Composites, Polymer Science and Technology Series, Longman Scientific &

Technical, New York, 1994.

[78] F.N. Cogswell, Polymer Melt Rheology: A guide for industrial practice, George

Godwin, London, 1981.

[79] I.S. Miles and S. Rostami, Multicomponent Polymer Systems, Polymer Science and

Technology Sires, Longman Scientific & Technical, New York, 1992.

[80] P.F. Bright, R.J. Crowson and M.J. Folkes, Study of effect of injection speed on fiber

orientation in simple moldings of short glass fiber filled polypropylene, Journal of

Material Science, 1978, 13(11): 2497-2506.

[81] P. Gerard, J. Raine and J. Pabiot, Characterization of fiber and molecular orientations

and their interaction in composite injection molding, Journal of Reinforced Plastics

and Composites, 1998, 17(10): 922-934.

[82] S.Y. Fu, B. Lauke, E. Mäder, C.Y. Yue and X. Hu, Fracture resistance of short-glass-

fiber-reinforced and short-carbon-fiber-reinforced polypropylene under Charpy

impact load and its dependence on processing, Journal of Materials Processing

Technology, 1999, 89-90: 501-507.

[83] J. Karger-Kocsis and K. Friedrich, Fatigue crack propagation in short and long fibre-

reinforced injection-molded PA 6.6 composites, Composites, 1988, 19(2): 105-114.

References

128

[84] M. Akay and D. Barkley, Processing-structure-property interaction in injection

molded glass-fiber-reinforced polypropylene, Composite Structure, 1985, 3(3-4): 269-

293.

[85] M. Akay and D. Barkley, Fiber orientation and mechanical behavior in reinforced

thermoplastic injection moldings, Journal of Materials Science, 1991, 26(10): 2731-

2742.

[86] M. Pechulis and D. Vautour, The effect of thickness on the tensile and impact

properties of reinforced thermoplastics, Society of Plastics Engineers Annual

Technical Conference, Technical Papers, 1997: 1860-1864.

[87] P. Singh and M.R. Kamal, The effect of processing variables on microstructure of

injection molded short fiber reinforced polypropylene composites, Polymer

Composites, 1989, 10(5): 344-351.

[88] S.E. Barbosa and J.M. Kenny, Analysis of the relationship between processing

conditions fiber orientation-final properties in short fiber reinforced polypropylene,

Journal of Reinforced Plastic and Composites, 1999, 18(5): 413-420.

[89] P. Hess, Mechanischer Eigenschaftsvergleich von langglasfaserverstärkten

Thermoplasten mit kurzglasfaserverstärkten Kunststoffen, Diplomarbeit, Technische

Universität Chemnitz, 2004.

[90] S. Hashemi, K.J. Din and P. Low, Fracture behavior of glass bead-filled

poly(oxymethylene) injection moldings, Polymer Engineering and Science, 1996,

36(13): 1807-1820.

[91] R. von Turkovic and L. Erwin, Fiber fracture in reinforced thermoplastic processing,


[92] V.B. Gupta, R.K. Mittal, P.K. Sharma, G. Mennig and J. Wolters, Some studies on

glass-fiber-reinforced polypropylene, Part 1: Reduction in fiber length during

processing, Polymer Composites, 1989, 10(1): 8-15.

References 129

[93] B. Franzen, C. Klason, J. Kubat and T. Kitano, Fiber degradation during processing of

short fiber reinforced thermoplastics, Composites, 1989, 20(1): 65-76.

[94] V.B. Gupta, R.K. Mittal, P.K. Sharma, G. Mennig and J. Wolters, Some studies on

glass-fiber-reinforced polypropylene, Part 2: Mechanical properties and their

dependence on fiber length, interfacial adhesion, and fiber dispersion, Polymer

Composites, 1989, 10(1): 16-27.

[95] J. Denault, T. Vu-Khanh and B. Foster, Tensile properties of injection molded long

fiber thermoplastic composites, Polymer Composites, 1989, 10(5): 313-321.

[96] R. Bailey and B. Rzepka, Fiber orientation mechanisms for injection molding of long

fiber composites, International Polymer Processing, 1991, 6(1): 35-41.

[97] J.C. Halpin and J.L. Kardos, Strength of discontinuous reinforced composites: Fiber

reinforced composites, Polymer Engineering and Science, 1978, 18(6): 496-504.

[98] J.L. Kardos and J.C. Halpin, Short predicting the strength and toughness of fiber

composites, Macromolecular Symposia, 1999, 147: 139-153.

[99] M. Xia, H. Hamada and Z. Maekawa, Flexural stiffness of injection molded glass

fiber reinforced thermoplastics, International Polymer Processing, 1995, 10(1): 74-81.

[100] H. Fukuda and T.W. Chou, A probabilitistic theory of the strength of short fiber

composites with variable fiber length and orientation, Journal of Materials Science,

1982, 17(4): 1003-1011.

[101] P.A. Templeton, Strength predictions of injection molding compounds, Journal of

Reinforced Plastics and Composites, 1990, 9(3): 210-225.

[102] S.Y. Fu and B. Lauke, Effects of fiber length and fiber orientation distributions on the

tensile strength of short fiber reinforced polymers, Composites Science and

Technology, 1996, 56(10): 1179-1190.

References

130

[103] J.L. Thomason, Micromechanical parameters from macromechanical measurements

on glass reinforced polypropylene, Composites Science and Technology, 2002, 62(10-

11): 1455-1468.

[104] G.B. Jeffery, The Motion of Ellipsoidal Particles Immersed in a Viscous Fluid,

Proceedings of the Royal Society of London A102, 1922: 161-179.

[105] S.G. Advani and C.L. Tucker III, The use of tensors to describe and predict fiber

orientation in short fiber composites, Journal of Rheology, 1987, 31(8): 751-784.

[106] G.L. Hand, A theory of anisotropic fluids, Journal of Fluid Mechanics, 1962, 13(1):

33-46.

[107] S.G. Advani and C.L. Tucker III, A numerical simulation of short fiber orientation in

compression molding, Polymer Composites, 1990, 11(3): 164-173.

[108] S.T. Chung and T.H. Kwon, Numerical simulation of fiber orientation in injection

molding of short fiber reinforced thermoplastics, Polymer Engineering and Science,

1995, 35(7): 604-618.

[109] P.D. Patel and D.C. Bogue, The effect of molecular orientation on the mechanical

properties of fiber filled amorphous polymers, Polymer Engineering and Science,

1981, 21(8): 449-456.

[110] C.W. Camacho, C.L. Tucker III, S. Yalvac and R.L. McGee, Stiffness and thermal

expansion predictions for hybrid short fiber composites, Polymer Composites, 1990,

11(4): 229-239.

[111] N.M. Neves, G. Isdell, A.S. Pouzada and P.C. Powell, On the effect of the fiber

orientation on the flexural stiffness of injection molded short fiber reinforced

polycarbonate plates, Polymer Composites, 1998, 19(5): 640-651.

[112] S.G. Advani and C.L. Tucker III, Closure approximations for 3-dimensional structure

tensors, Journal of Rheology, 1990, 34(3): 367-386.

References 131

[113] R.S. Bay and C.L. Tucker III, Fiber orientation in simple injection molding, Part 1:

Theory and numerical methods, and Part 2: Experimental results, Polymer

Composites, 1992, 13(4): 317-341.

[114] J.S. Cintra and C.L. Tucker III, Orthotropic closure approximations for flow-induced

fiber orientation, Journal of Rheology, 1995, 39(6): 1095-1122.

[115] K.H. Han and Y.T. Im, Modified hybrid closure approximation for predicting of flow-

induced fiber orientation, Journal of Rheology, 1999, 43(3): 569-589.

[116] F. Folgar and C.L. Tucker III, Orientation behavior of fibers in concentrated

suspensions, Journal of Reinforced Plastics and Composites, 1984, 3(2): 98-119.

[117] W.C. Jackson, S.G. Advani and C.L. Tucker III, Predicting the orientation of short

fibers in thin compression moldings, Journal of Composite Materials, 1986, 20(6):

539-557.

[118] M.R. Kamal and A.T. Mutel, The prediction of flow and orientation behavior of short

fiber reinforced melts in simple flow systems, Polymer Composites, 1989, 10(5): 337-

343.

[119] P.H. Foss, J.P. Harris, J.F. O’Gara, L.P. Inzinna, E.W. Liang, C.M. Dunbar, C.L.

Tucker III and K.F. Heitzmann, Prediction of fiber orientation and mechanical

properties using C-Mold and Abaqus, Society of Plastics Engineers Annual Technical


[120] A. Larsen, Injection molding of short fiber reinforced thermoplastics in a center-gated

mold, Polymer Composites, 2000, 21(1): 51-64.

[121] N.M. Neves, A.J. Pontes and A.S. Pouzada, Experimental validation of morphology

simulation in glass fiber reinforced polycarbonate disc, Journal of Reinforced Plastics

and Composites, 2001, 20(6): 452-465.

References

132

[122] A.J. Pontes, N.M. Neves and A.S. Pouzada, The role of the interaction coefficient in

the prediction of the fiber orientation in planar injection moldings, Polymer

Composites, 2003, 24(3): 358-366.

[123] ARBURG GmbH + Co, A Brief Guide into Injection Molding, 1997.

[124] G. Schlatter, J.-F Agassant, A. Davidoff and M. Vincent, An unsteady multi-fluid

flow model: Application to sandwich injection molding process, Polymer Engineering

and Science, 1999, 39(1): 78-88.

[125] R. Seldezn, Co-injection molding: Effect of processing on material distribution and

mechanical properties of a sandwich molded plate, Polymer Engineering and Science,

2000, 40(5): 1165-1176.

[126] S. Patcharaphun and G. Mennig, Simulation and experimental investigations of

material distribution in the sandwich injection molding process, Polymer-Plastics

Technology and Engineering (in press).

[127] S. Patcharaphun and G. Mennig, Properties enhancement of short-glass-fiber

reinforced thermoplastics via sandwich injection molding technique, Polymer

Composites, 2005, 26(6): 823-831.

[128] S. Patcharaphun and G. Mennig, Three-Dimensional Simulation and Experimental

Investigations of Fiber Orientation in Sandwich Injection Molded Parts,

TECHNOMER’05, 19. Fachtagung über Verarbeitung und Anwendung von

Polymeren, November 2005, Chemnitz, Germany.

[129] S. Patcharaphun, B. Zhang and G. Mennig, Simulation of Three-Dimensional Fiber

Orientation in Weldline Areas during Push-Pull-Processing, Journal of Reinforced

Plastics and Composites (submitted).

[130] M. Akay and D. O’Regan, Generation of voids in fiber reinforced thermoplastic

injection moldings, Plastics, Rubber and Composites Processing and Applications,

1995, 24(2): 97-102.

References 133

[131] D. O’Regan and M. Akay, The distribution of fiber lengths in injection molded

polyamide composite components, Journal of Materials Processing Technology, 1996,

56(1-4): 282-291.

[132] M. Sanou, B. Chung and C. Cohen, Glass fiber-filled thermoplastics, Part 2: Cavity

filling and fiber orientation in injection molding, Polymer Engineering and Science,

1985, 25(16): 1008-1016.

[133] D.A. Messaoud, B. Sanschagrin and A. Derdouri, Co-injection molding: Effect of

processing on material distribution and mechanical properties of short-glass-fiber-

reinforced polypropylene test bar, Society of Plastics Engineers Annual Technical


[134] P.A. Eriksson, A.C. Albertsson, P. Boydell, G. Prautzsch and J.-A.E. Manson,

Prediction of mechanical properties of recycled fiberglass reinforced polyamide 66,

Polymer Composites, 1996, 17(6): 830-839.

[135] S.R. Tremblay, P.G. Lafleur and A. Ait-Kadi, Effect of injection parameters on fiber

attrition and mechanical properties of polystyrene molded parts, Society of Plastics

Engineers Annual Technical Conference, Technical Papers, 1999: 432-436.

[136] U. Yilmazer and M. Cansever, Effect of processing conditions on the fiber length

distribution and mechanical properties of glass fiber reinforced nylon-6, Polymer

Composites, 2002, 23(1): 61-71.

[137] A. Savadori, A. Pelliconi and D. Romanini, Weldline resistance in polypropylene

composites, Plastics and Rubber Processing and Applications, 1983, 3(3) : 215-221.

[138] B. Fisa and M. Rahmani, Weldline strength in injection molded glass fiber-reinforced

polypropylene, Polymer Engineering and Science, 1991, 31(18): 1330-1336.

[139] B. Fisa, J. Dufour and T. Vu-Khanh, Weldline integrity of reinforced plastics: Effect

of filler shape, Polymer Composites, 1987, 8(6): 408-418.

References

134

[140] A. Vaxman, M. Narkis, A. Siegman and S. Kenig, Weld-line characteristics in short

fiber reinforced thermoplastics, Polymer Composites, 1991, 12(3): 161-168.

[141] V. M. Nadkarni and S. R. Ayodhya, The influence of knit-lines on the tensile

properties of fiber-glass reinforced thermoplastics, Polymer Engineering and Science,

1993, 33(6): 358-367.

[142] F. Meyer and I. Kühnert, Untersuchungen zur Bindenahtfestigkeit von technischen

Bauteilen aus kurzfaserverstärkten Thermoplasten, Forschungsbericht des

Forschungskuratoriums Maschinebau e.V., FKM-Heft 265, AiF-Abschlussbericht

2001.

[143] N. Sombatsompop, K. Liolios, M.H. Mohd Jamel and A.K. Wood, Techniques for

pressure-density-temperature measurements in polymer melts, Polymers & Polymer

Composites, 1997, 5(4): 259-264.

[144] J. Crown, M.J. Folkes and P.F. Bright, Rheology of short-glass-fiber reinforced

thermoplastics and it application to injection molding 1. Fiber motion and viscosity

measurement, Polymer Engineering and Science, 1980, 20(14): 925-933.

[145] K. Tomari, H. Takashima and H. Hamada, Improvement of weldline strength of fiber

reinforced polycarbonate injection molded articles using simultaneous composite

injection molding, Advances in Polymer Technology, 1995, 14(1): 25-34.

[146] A. Meddad and B. Fisa, Weldline strength in glass fiber reinforced polyamide 66,


[147] K. Ainoya and O. Amano, Accuracy of filling analysis program, Society of Plastics

Engineers Annual Technical Conference, Technical Papers, 2001: 726-730.

[148] N. Sombatsompop and A. K. Wood, Measurement of thermal conductivity of

polymers using an improved Lee’s disc apparatus, Polymer Testing, 1997, 16(3): 203-

223.

References 135

[149] G. Mennig, K. Lunkwitz and D. Lehmann, Chemische Oberflächenmodifizierung

beim Spritzgießen und dessen Wechselwirkung mit dem Verarbeitungsverhalten,

DFG-Abschlussbericht, Technische Universität Chemnitz, 2005.

[150] K.-H. Han and Y.-T. Im, Numerical simulation of three-dimensional fiber orientation

in injection molding including fountain flow effect, Polymer Composites, 2002, 23(2):

222-238.

Curriculum Vitae

Personal data Name: Somjate Patcharaphun

Date of birth: 25 December 1972

Birthplace: Bangkok, Thailand

Nationality: Thai

Marital status: Married

Education 1979 – 1985 Primary school, Santa Cruz Suksa School, Bangkok, Thailand

1986 – 1990 Secondary school, Taweetapisek School, Bangkok, Thailand

1991 – 1994 Bachelor of Engineering (B.Eng.), Faculty of Engineering, King

Mongkut’s Institute of Technology Thonburi (KMITT), Bangkok,

Thailand

1998 – 2000 Master of Engineering (M.Eng.), School of Energy and Materials, King

Mongkut's University of Technology Thonburi (KMUTT), Bangkok,

Thailand

Professional experience 1994 – 1996 Project engineer, Isuzu Motor (Thailand) Co., Ltd.

1996 – 1997 Technical consultant and instructor, Italthai Development Co., Ltd.

1997 – 1998 Project engineer, Telecom Asia Co., Ltd.

since 2000 Lecturer, Department of Materials Engineering Faculty of Engineering,

Kasetsart University, Thailand.

characterization and simulation of material distribution - qucosa

Documents