_tMEAbND3Gu

Svcic~j' (Jf,Dlu:'.IIC,~ l"ngmeer.1Edited by

Hans-Peter Heim

H. PotentePlastics Design Library

Copyright © 2001, Plastics Design Library. All rights reserved.

ISBN 1-884207-91-X

Library of Congress Control Number: 2001091835

Published in the United States of America, Norwich, NY by Plastics Design Library a division of

William Andrew Inc.

Information in this document is subject to change without notice and does not represent a commitment

on the part of Plastics Design Library. No part of this document may be reproduced or transmitted in

any form or by any means, electronic or mechanical, including photocopying, recording, or any infor-

mation retrieval and storage system, for any purpose without the written permission of Plastics Design

Library.

Comments, criticism and suggestions are invited and should be forwarded to Plastics Design Library.

Plastics Design Library and its logo are trademarks of William Andrew Inc.

Please Note: Great care is taken in the compilation and production of this volume, but it should be

made clear that no warranties, express or implied, are given in connection with the accuracy or com-

pleteness of this publication, and no responsibility can be taken for any claims that may arise. In any in-

dividual case of application, the respective user must check the correctness by consulting other

relevant sources of information.

The use of general descriptive names, registered names, trademarks, etc. in this publication does not

imply, even in the absence of a specific statement, that such names are exempt from the relevant protec-

tive laws and regulations and therefore free for general use.

Manufactured in the United States of America.

Plastics Design Library, 13 Eaton Avenue, Norwich, NY 13815 Tel: 607/337-5080 Fax: 607/337-5090

Preface

Over the next few decades, the development of our society will be characterized by two par-ticular factors:1. the increasing importance of the medical sector, coupled with growing requirements on medical engineering products.2. a continuing key role for mobility within society, with growing emphasis on finding a solution to the associated ecological problems.

Plastics will play a crucial role in both cases. With properties ranges that can be widelyadjusted and ease of processing, plastics can be used to produce highly-integrated, custom-ized product solutions for medical-engineering applications, as well as products for theautomotive sector and telecommunications. The plastics sector is far from having exhaustedthe innovation potential that exists. What is required are material, process engineering, andmechanical engineering-based approaches to innovation which will make it possible torespond to ever more demanding applications or the substitution of other materials by plas-tics.

The development trends currently emerging in plastics engineering are essentially thecombination of different processes (e.g., injection molding and surface refinement), thecombination of different materials (e.g., plastic/metal composites), the integration of a widerange of functions within a single component, the improvement of surface and/or opticalproperties, and also reduced material consumption and the recyclability of the materialsemployed.

Much of the innovation potential lies in the injection molding process. Special injec-tion molding processes such as the thin-wall technique, micro-molding, hybrid processesand multi-component processes, are playing an increasingly large role in the processing ofplastics. At the same time, the requirements placed on product quality, production qualityand also part precision are rising, while costs are expected to be reduced wherever possible.This combination of objectives can only be achieved by an appropriate increase in produc-tivity from process-engineering innovations and a reduction in the number of process steps.

From the microeconomic angle, the challenge facing a plastic-processing companytoday is one of retaining its market position through readiness and ability to innovate. Thisapplies particularly to the opening up of new application potential and the application tech-nology that this requires, as well as to the crucial aspect of employee skills. Specialistknow-how is essential in order to survive on the market – in other words, injection molding

vi Preface

companies are called upon to acquire know-how on new materials and material combina-tions as well as on process variants and special injection molding processes.

It is at this point that the book commences. It constitutes a collection of experimentaland theoretical studies in the field of special injection molding processes. The papers thathave been brought together in this book were presented at International SPE (Society ofPlastic Engineers) Conferences over the period 1996 to 2000. The presenters of the papersare scientists and representatives from eminent institutes and companies who have made akey contribution towards the continuing development of injection molding technologythrough their work. The manuscripts printed in the book represent an extract of their currentwork and research results. This book is not therefore intended as a textbook but rather asspecialist reading, progressing far beyond the basic principles of application technology ininjection molding.

A broad range of specialized subjects is covered, including process engineering, mate-rial science, mechanical engineering, and mold engineering. The topics treated are dividedinto the following subject areas, which follow on from an initial overview of the specialinjection molding processes that are currently in use:

Gas Assisted Injection Molding (GAIM)• influence of process parameters• molded part design and mold layout• the special processes of powder injection and reaction injection molding

Thin Wall Molding• influences on molded part quality• avoidance of errors during application• influence on cycle time and costs

Molding of Micro Parts and Micro Structures• reproduction of high-precision surface structures• optical applications• simulation techniques and mechanical engineering

Improving Material Properties• modified plastics for automotive applications• improved molded part properties through vibrational molding

Molding of Composites• compression molding• vacuum assisted liquid molding• resin transfer molding• inmold decoration

Preface vii

Mold Making and Plasticisation• automation technology• mold engineering for special injection molding processes• twin-screw injection molding• screw optimization

This book presents the state of the art in all the areas presented. The authors of the indi-vidual contributions give their views on the uses and limitations of new injection moldingtechnologies. They thus offer the know-how that is required to exploit the innovation poten-tial of plastics engineering.

Helmut Potente and Hans-Peter Heim Paderborn, June 2001

Table of Contents

Preface vHelmut Potente and Hans-Peter Heim

One-shot Manufacturing: What is Possible with New Molding Technologies 1James F. Stevenson

Chapter 1: Gas Assisted Injection Molding 15Gas Assist Injection Molding. The North American Legacy 15

Jack Avery Flow Directions in the Gas Assisted Injection Molding Technology 27

Young Soo Soh and Chan Hong ChungGas-assisted Injection Molding: Influence of Processing Conditions and Material Properties 35

Kurt W Koelling and Ronald C KaminskiCover Part as an Application Example for Gas-assisted Injection Molded Parts 43

Michael Hansen Molded Part Design for the Gas Injection Technique 51

H. Potente and H.-P. HeimDesign Optimization of Gas Channels for an Air Cleaner Assembly Using CAE Simulations 57

D.M. Gao, A. Garcia-Rejon, G. Salloum and D. BaylisThe Occurrence of Fiber Exposure in Gas Assist Injection Molded Nylon Composites 65

Shih-Jung Liu and Jer-Haur ChangSaving Costs and Time by Means of Gas-assisted Powder Injection Molding 73

Christian Hopmann, Walter MichaeliGas-assisted Reaction Injection Molding (GRIM): Application of the Gas Injection Technology to the Manufacturing of Hollow Polyurethane Parts 79

I. Kleba, E. Haberstroh

Chapter 2: Thin Wall Molding 89Thin Wall Processing of Engineering Resins: Issues and Answers 89

Larry Cosma

ii Table of Contents

Effects of Processing Conditions and Material Models on the Injection Pressure and Flow Length in Thinwall Parts 99

A. J. Poslinski10 Common Pitfalls in Thin-Wall Plastic Part Design 107

Timothy A. PalmerFlow Instabilities in Thin-wall Injection Molding of Thermoplastic Polyurethane 113

Christian D. Smialek, Christopher L. SimpsonPressure Loss in Thin Wall Moldings 121

John W. Bozzelli, Jim Cardinal, and Bill FierensIntegrating Thin Wall Molder’s Needs into Polymer Manufacturing 127

W. G. Todd, H. K. Williams, D. L. WiseThinning Injection Molded Computer Walls 133

Lee Hornberger and Ken Lown

Chapter 3: Molding Micro Parts and Micro Structures 143Transcription of Small Surface Structures in Injection Molding – an Experimental Study 143

Uffe R. Arlø, Erik M. KjærInjection Molding of Sub- m Grating Optical Elements 149

R. Wimberger-FriedlProcess Analysis and Injection Molding of Microstructures 157

Alrun Spennemann and Walter MichaeliSimualtion of the Micro Injection Molding Process 163

Oliver Kemmann, Lutz Weber, Cécile Jeggy, Olivier Magotte, and François Dupret

Chapter 4: Manufacturing of Composites 171Melt Compression Molding (MCM) a One-shot Process for In-mold Lamination and Compression Molding by Melt Strip Deposition 171

Georg H. KuhlmannIn-mold Lamination Back Compression Molding 187

Thomas HuberAnalysis and Characterization of Flow Channels during Manufacturing of Composites by Resin Transfer Molding 193

R. V. Mohan, K. K. Tamma, S. Bickerton, S. G. Advani and D. R. ShiresOptimization of Channel Design in VARTM Processing 209

Roopesh Mathur, Suresh G. Advani and Bruce K. Fink

µ

Table of Contents iii

Injection Compression Molding. A Low Pressure Process for Manufacturing Textile-Covered Mouldings 215

Carsten Brockmann, Walter Michaeli Kurz-Hastings Inmold Decoration 223

Roy Bomberger

Chapter 5: Improving Material Properties 229High Impact Strength Reinforced Polyester Engineering Resins for Automotive Applications 229

Mengshi Lu, Kevin Manning, Suzanne Nelsen, and Steve Leyrer Control of Internal Stresses in Injection Molded Parts Through the Use of Vibrational Molding, “RHEOMOLDINGSM”, Technology 237

Akihisa Kikuchi, Marc Galop, Harold L. Brown, and Alexander Bubel Experimental Determination of Optimized Vibration-assisted Injection Molding Processing Parameters for Atactic Polystyrene 245

Alan M. Tom, Akihisa Kikuchi, and John P. CoulterVibrated Gas Assist Molding: Its Benefits in Injection Molding 253

J.P. Ibar

Chapter 6: Mold Making and Plasticisation 259Advances in Stack Molding Technology 259

Vincent Travaglini and Henry RozemaAdvanced Valve Gate Technology for Use in Specialty Injection Molding 267

John Blundy, David Reitan, and Jack SteeleIn-mold Labeling for High Speed, Thin Wall Injection Molding 273

Gary FongAdvances in Fusible Core Technique 281

E. Schmachtenberg and O. SchröderProcessing Glass-filled Polyethylene on a Twin-screw Injection Molding Extruder 287

David Bigio, Rajath Mudalamane, Yue Huang and Saeid Zerafati Injection Molding by Direct Compounding 295

Bernd KlotzImprovement of the Molded Part Quality: Optimization of the Plastification Unit 301

S. Boelinger and W. Michaeli

iv Table of Contents

Non-return Valve with Distributive and Dispersive Mixing Capability 307Chris Rauwendaal

Index 313

One-shot Manufacturing: What is Possible with NewMolding Technologies

James F. StevensonGenCorp Technology Center, Akron, OH 44305, USA

INTRODUCTION

New molding technologies1 together with a revolution in thinking about how to design andmanufacture products2-3 have merged to open exciting new possibilities in polymer partmanufacturing. The new technologies offer versatile, cost effective forms of materials andmore unified, efficient production methods. Even greater benefits, especially from consoli-dation, exist for parts previously made of metal. These processes move forward and, insome cases, realize the goal of forming a complex product in a single manufacturing step, orone-shot manufacturing.

Conventional manufacturing processes generally employ homogeneous materials andsimple primary shaping processes to form components which were then assembled by vari-ous joining methods. These subassemblies then go through various secondary operationsand ultimately are combined to form the finished product. This sequential process is laborintensive, time consuming, and costly; it requires large inventories and long changeovertimes and is prone to produce scrap.

This paper presents and analyzes common features of ten of the new material-processtechnologies. These technologies, along with chapters on predicting orientation andwarpage by T.A. Osswald and on lean molding by Colin Austin are presented in greaterdepth in a recent book Innovation in Polymer Processing: Molding.1

All of these process and material innovations are based on antecedent technology andwell-known physical principles. The key to realizing these innovations was conceptual.

NEW TECHNOLOGIES

The innovative molding technologies are summarized in Tables A1-A10 which also includea listing of advantages/disadvantages, applications, and materials. These tables are self con-tained; readers are referred to them as independent sources of information and as descriptivematerial for the processes cited in Tables 1-3.

2 Special Molding Techniques

PRODUCT ELEMENTS

The framework given in Tables 1-3 facilitates classification and comparison of the newmolding technologies. It serves both as a guide to applications and as a means of locatingopportunities and gaps in the new technologies. Technologies at similar locations in thetables can be considered as potential alternatives for each other.

In the tables materials are divided into polymers, consisting of rubber and plastics, andnonpolymers, primarily solids or gases. Solids, typically fibers, serve as reinforcements,whereas gases reduce density or increase stiffness for a given cross-sectional area by distrib-uting material to increase the moment of inertia. The term macroscopic refers to dimensionsthat are on the order of the thin part dimension, e.g., a 2-mm diameter void in a 3-mm rod.Microscopic means dimensions two orders of magnitude or more smaller (e.g., a 300-layerlaminate) than the part thickness.

Achieving one-shot manufacturing requires an optimal combination of MATERIALS,PROCESS, and GEOMETRY.

MATERIALS

In terms of MATERIALS, the innovative molding technologies are classified accord-ing to COMPOSITION, one or more POLYMERS with or without NONPOLYMERS, andSCALE, MICROSCOPIC or MACROSCOPIC, as shown in Table 1.

* Processes designated LPM (Low Pressure Molding) are described in Table A4; those designated MMP (Multimaterial Multi-process) are given in Table A10.

Table 1. Materials: composition and scale

Scale

Composition

Polymers combined with

Other polymers Gas Solids

Macroscopic Blow Molding Coinjection Molding (MMP)* Multimaterial Molding (MMP)In-Mold Coating (MMP) Dual Molding (LPM)*

Gas-Assisted Molding Liquid-Gas Molding (LPM) Blow Molding

Laminate Molding (LPM)

Microscopic Lamellar Molding Microcellular Plastics Controlled Density (LPM)

Sheet Composites Reactive Liquid Molding

One-shot Manufacturing 3

Two of the innovative processes shown in Table 1, lamellar molding and microcellularplastics, serve primarily as the means of generating a unique material on the microscopicscale and secondarily for shaping the material. Other processes listed in the table combinepolymers and nonpolymers on a microscopic or macroscopic scale to from products wherespecific properties of the multiple materials meet local functional needs. This localized opti-mization ultimately enhances overall product performance.

PROCESSING

PROCESSING is the bridge between the unshaped raw MATERIALS and the GEOMETRY(MACROSCOPIC STRUCTURE and SIZE & SHAPE) of the product as shown in Table 2.

Many processes under MACROSCOPIC STRUCTURE in Table 2 allow the combina-tion of a polymer with other polymer(s) or nonpolymers on a macroscopic scale to give

Table 2. Processing: materials and geometry

Materials

Geometry

Macroscopic structure Size & shape

Laminate Segmented Large Hollow

Polymers In-Mold Coat-ingCoinjectionBlow MoldingDual Molding (LPM)

Multimaterial Molding (MMP),Blow Molding

Sheet CompositesIn-Mold Coating (MMP)Injection-Compres-sion (MMP)

Dual Molding (LPM)Blow Molding

Poly-mers and non-polymers

Solid Laminate Molding (LPM)Reactive Liquid Molding

Laminate Molding (LPM)Sheet CompositesReactive Liquid Molding

Gas Blow MoldingGas-Assisted MoldingLiquid Gas Molding (LPM)Fusible Core

Blow MoldingControlled Density Molding (LPM)

Blow MoldingGas-Assisted Mold-ingLiquid Gas Mold-ing (LPM)Fusible Core


LAMINATES which have uniform properties over the surface of the part but variable proper-ties through the thickness.

SEGMENTED PARTS which exhibit a variation of properties along the surface but areuniform over a given cross-section.

Combinations of laminated and segmented parts are possible, for example gas-assistedparts with hollow sections only in certain regions of the part.

In terms of part SIZE & SHAPE, innovative technologies are generally needed whenthe part is LARGE, especially when a cosmetic surface is required, or when the part has aCOMPLEX, often hollow shape, particularly when it is load bearing.

GEOMETRY

Part GEOMETRY can be considered in terms of FUNCTION where the part is locatedon a scale ranging from ENCLOSURE (containers or panels, often with cosmetic surfaces) toLOAD BEARING, and COMPLEXITY which allows for geometric complexity ranging fromSYMMETRIC (planar or axisymmetric) to fully THREE DIMENSIONAL.

Table 3. Geometry: function and complexity


Table 3 suggests that molded parts and the associated innovative processes to makethem generally range from more or less symmetric enclosures in the upper left to threedimensional load bearing parts in the lower right.

REFERENCES1 Stevenson, J.F. (Ed.) Innovation in Polymer Processing: Molding, Hanser, Munich, Hanser-Gardner, Cincinnati (1996).2 Womack, J.P., Jones, D.T., and Roos, D., The Machine that Changed the World, Macmillan, New York (1990).3 Gooch, J., George, M., and Montgomery, D., America Can Compete, Institute of Business Technology, Dallas (1987).


APPENDIX

Table A1 and A2. Gas-assisted injection molding: process [1.1] and simulation [2.1]

PROCESS: Nitrogen gas under high pressure is injected through the nozzle or mold wall into plastic partially filling a mold. The gas flows preferentially through local thick sections with hot interiors and pushes the plas-tic ahead to fill the mold.SIMULATION: Commercial software now available to predict gas flow paths, polymer thickness, clamp force, and contraction during cooling for various geometries and process variables including gas pressure, injection time, and prefilled polymer volume.Simulations and experiment generally show• increasing gas pressure decreases fill time, gas penetration distance, and (by conservation of mass) poly-

mer wall thickness,• melt temperature has a variable effect on gas penetration length,• increasing delay time before the start of gas injection increases wall thickness and gas penetration

length,• increasing gas injection time increases gas penetration distance,• decreasing the prefilled polymer volume fraction increases the penetration length until a critical level

when gas blows through.• increasing gas pressure level and time decreases shrinkage.Simulations are generally able to predict undesirable air traps and gas penetration into thin sections. Simula-tion of a freezer bottom part converted to gas-assisted molding showed a 70% reduction in packing pressure, feasibility of using a less expensive material, and reduced warpage due to lower, more uniform pressure and higher part stiffness. Eight design guidelines are given based on both experiment and computer simulation [2.1].

Advantages/Disadvantages Applications Materials

PROCESS: Part weight and cooling time can be reduced up to 50%. Sink marks are eliminated. Warpage is reduced. Clamp force and injection pressure are lower. Part stiffness is increased because of the higher moment of inertia. Licensing is necessary. SIMULATION: Simulation helps identify optimal process conditions including runner layout and size, and location and timing of gas introduction. Software is available but new.

Handles, Panels with Ribs, Appli-ance/machine Housings (TV ben-zels) Automotive Parts (Clutch Pedals, Mirror Housings)

ABS, PA, PE, PP, PS, PPO, PC, PBTP, PC/PBTP, SAN, TPE, TPU


Table A3. Fusible core injection molding [3.1]

Complex hollow parts are formed by injection molding plastic around a fusible alloy core which is subse-quently removed by melting. The fusible (or lost) core typically is cast form a bismuth-tin alloy with a eutec-tic melting point of 138oC. The molten metal fills a split steel mold from the bottom and then cools for 2 min to produce a heavy core with a mirrorlike surface. The still-hot core is positioned by a robot in a steel mold and plastic is injected.Flow channels are designed to balance forces around the core during filling to prevent core movement. For thermoplastics the injection temperature, e.g. 290oC for polyamide, can be well above the melting point of the core since the relatively high thermal diffusivity of the metal maintains a low interface temperature. After demolding, cores are melted out in a large bath or by induction heating or by injecting heat transfer fluid inside hollow cores.


Plastic parts made by fusible core technology have a weight and cost advantage over metal parts. Fusible core molding eliminates the need for mechanically complex molds or join-ing separately molded parts. Interior surfaces of fusible core parts are smooth which increases gas flow. Disadvantages are loss or oxidation of expensive core metal and need for robots to handle heavy cores.

Air intake manifolds, tennis racquets, pump parts

PAPoly(etherarylke-tone)

Table A4. Low pressure molding [4.1]

Low pressure molding, as developed by Siebolt Hettinga, enables a number of other molding technologies.In Low Pressure Molding (LPM) the mold cavity is filled at low speed through large gates with a controlled pressure profile in the shape of a broad inverted U. LPM has no packing stage and no cushion. The melt tem-perature profile is controlled by adjusting screw speed and flow resistance during plastication. LPM works better with low viscosity semicrystalline materials and is not suitable for thin-wall parts. Slow injection, lower melt and higher mold temperatures reduce residual stress to allow demolding at a higher temperature to maintain or reduce cycle times. For larger parts, low clamp force can be achieved using multiple valve gates with programmed opening [4.2]. Lower clamp force allows use of self clamping molds and multistation injectors. Laminate Molding involves molding plastic at low pressure directly behind textile, film, or metal. In Liquid Gas Injection Molding, a volatile liquid is injected at low pressure into the melt and then vaporizes to form hollow channels in the part. The liquid condenses and is absorbed in the part. Dual Molding, similar to Bayer’s Multishell Molding, forms an integrated hollow part by overmolding at low pressure an assembly formed from separately molded parts. In Controlled Density Molding the mold is partially opened once a skin has formed to give a low density interior.



Substantial capital costs savings result from the use of presses with a lower camp force or self clamping molds. Laminate Molding saves on assembly and adhesive costs in fabric/plastic lam-inates.

Low Pressure Molding: Interior Vehicle Panels, Bumper Fascia. Laminate Molding: Fabric/Plastic Seats, Vehicle Trim Panels. Liq-uid-Gas Assist Molding: Large Chairs, Chair Bases. Dual and Shell Molding: Manifolds, Pump Bodies, Valves, and Fittings. Low Density Molding: Fittings, Elec-tronic Enclosures, Table Tops.

Thermoplas-tics, especially polyolefins, thermosets

Table A5. Advanced blow molding [5.1]

The advanced blow molding technologies described below have greatly extended the versatility and facili-tated product design. Deep-Draw Double-Wall Molding employs a mold with four hinged slides and an advancing core which close in a programmed manner around a partially inflated parison to shape a deep draw part. Press Blow Molding is used to form panels between shallow male and female mold halves which press together certain sections and inflate other sections to form hollow stiffening ribs.Three Dimensional (3D) Blow Molding forms serpentine three-dimensional parts without excessive scrap by manipulating the parison and positioning it in a convoluted mold cavity. Positioning the parison can be accomplished by (1) translating in two directions the mold which is titled at an angle, (2) movement of the parison by robotic arms in a mold with multiple sections which close sequentially, and (3) guiding the parison through the mold by sucking air along the length of the mold. Multimaterial Blow Molding employs multiple materials sequentially along the part length, in layers over the part thickness, or on opposite sides of the parison. New material developments include molding of 0.3-in fiber reinforced materials and foam layers [5.2]. Computer simulation of blow molding has been developed by A.C. Technology, Ithaca, NY in cooperation with G.E.


Deep Draw Technology increases draw (depth-to-length) ratio from 0,3 to 0.7 and allows forming of parts with undercuts, ribs, and noncircular crosssections. Multimaterial applications allow soft surfaces on structural parts, flexible conduits with rigid con-nectors, or parts with opposite sides of different properties. 3D Blow Molding consolidates complex parts and enhances func-tion.

Insulated containers with foam, Planters, Conduits, Air Ducts, Bumpers, Equipment Panels, Instru-ment Panels, Portable Toi-lets, Golf Cases, Arm Rests, Gasoline Filler Tubes, Gas Tanks.

PE, PS, PP, POA, ABS, PPE elastomers

Table A4. Low pressure molding [4.1]


Table A6. Thermoplastic sheet composites [6.1]

Production of thermoplastic sheet composites involves two steps: (1) FORMATION of fiber reinforced sheets (prepreg) by polymer impregnation and sheet consolidation and (2) SHAPING of the sheets. Large volume competitive sheet FORMATION processes are continuous Melt Impregnation (e.g. Azdel sheet by extruding polypropylene onto a continuous “swirled’ fiber glass mat), and Slurry Deposition in which long fibers and polymer powder with dispersing agents are deposited on a moving screen similar to paper making. Other processes are Powder Impregnation (powder and fiber consolidated by pultrusion, double belt press, or compression molding), Reactive Pultrusion, and Commingling (intertwining different fibers) and Coweaving polymer fibers, and reinforcing fibers. SHAPING techniques for consolidated prepreg include Melt-Phase Stamping (prepreg covering the mold cavity is heated with infrared and shaped in a fast closing press), Fast Compression Molding (thick charge flows during mold closing), and Solid State Stamping (semicrystalline plastics below their melting point are stamped into parts with simple geometries in 15 sec. Other shaping methods include Pultrusion of prepreg tapes, Diaphragm Molding (preform between plastically deformable diaphragms shaped by hard tooling), Rubber Pad Molding, Hydroforming (rubber bladder inflated hydrauli-cally, Vacuum Forming in an autoclave, and Flexible Resin Transfer Molding (sheets of resin and fiber between elastomeric diaphragms are consolidated, then shaped).


Extrusion melt impregnation allows high fiber contents and longer fibers which give improved mechanical properties. Slurry deposition employs shorter fibers which allow greater flow and more complex parts. Cycle times are short.

Automotive Body Panels, Other Components, Air-craft Components

PP, PE, PA, PBT, PET, PVC, PC, PEEK, PSU, PPS


Table A7. Reactive liquid composite molding [7.1]

Reactive Liquid Composite Molding (RLCM) proceeds in two steps: (1) PREFORM FORMATION by orga-nizing loose fibers into a shaped preform, and (2) IMPREGNATION of the fibers with a low viscosity react-ing liquid. The reacting material may be thermally activated by heat transfer in the mold or mixing activated by impingement of two reactive streams. Simulations of flow and reaction, a recent innovation in RLCM, allow determination of vent and weld line locations, fill times, and control of ‘racetracking’ in terms of gate locations, mat permeability, and processing conditions. Commercial success requires (1) fast reaction and (2) efficient preform formation. Cycle time for thermally active systems can be decreased by using higher mold temperatures and heating the preform. Innovative processes for PREFORMING include: Thermoformable Mat heated by IR to melt the binder and pressed into shape by one or two moving platens while supported by a hold/slip edge clamp to reduce wrinkling.Automated Directed Fiber Performers employ multiple delivery systems to create a surface veil, a chopped roving layer, and continuous roving with loops, all of which are fused by hot air. The SCRIMP process channels resin flow between layers of fibers or along internal networks. Water Slurry Deposition positions fibers by water flow through a contoured screen and sets them with hot air. Innovations to reduce costs by combining process steps include: Direct Part Forming combines sheet formation and shaping, e.g. heating porous sheet and then consolidating and shaping in a compression mold. The Hot Air Preformer produces performs by either directed fiber or thermoplastic mat forming. The Cut-N-Shoot process combines preforming and molding steps consecutively in the same tool. Bladder inflation inside a mold shapes the preform and forms the mold wall during filling.


Low pressure and temperature processing by RLCM allow the use of inexpensive light-weight tools, especially for prototyping. RLCM allows customizing reinforcement to give desired local properties and part consolidation via complex 3D geometries.

Marine and Pool-side Products, Sanitary ware, Caskets, Automo-tive Panels, Vehi-cle Suspension Links

Isocyanate based resins (mixing activated) Unsaturated polyester and sty-rene (thermal activated)


Table A8. MIcrocellular plastics: formation and shaping [8.1]

Extremely small closed cells from 0.1 to 10 microns in diameter can be formed in most plastics by dissolving gas in the plastic, typically supercritical CO2, and then rapidly reducing the pressure and increasing tempera-ture in a controlled manner to cause homogeneous and likely heterogeneous nucleation and growth of gas bubbles. The bubbles are to be smaller than naturally occurring flaws in the polymer so mechanical properties are not compromised. Particles in PS(HI) can be sites for heterogeneous nucleation [9.2]. Short diffusion paths, elevated temperatures, and gases in the supercritical state are necessary to achieve the high diffusion rate and high gas concentration needed for commercial use. Extrusion with gas injection is an efficient process to saturate polymer with gas. Manufacturing issues are determining sequence of pressure, temperature, and shaping geometries to nucleate and form cells without disruption and to shape product with-out distortion. Microcellular technology is covered by several patents and is offered for licensing by Axiom-atics, Woburn, MA


The extremely small bubbles give weight reductions of 10-90%, no reduction in specific mechanical proper-ties, appearance of a solid opaque surface, and foaming of thin sections. Fatigue resistance is observed to increase. Environmental advantages are use of atmo-spheric gases and lower material use

Siding, Pipes, Aircraft Parts, Athletic Equipment, Machine Housings, Automotive Com-ponents, Food Containers, Artificial Paper, Thermal Insulation, Fibers for Apparel and Carpets

ABS, PE, PET, PMMAPS, PS(HI), PP, PUPVC, SMC, Fluoropolymers, Poly(methylpen-tene)

Table A9. Lamellar injection molding [9.1]

In Lamellar Injection Molding (LIM), two (or three) materials are extruded separately and combined with a 3 (or 5)-layer feedblock with multipliers to form a melt stream with hundreds of layers. This stream is injection molded to from parts with an irregular lamination pattern. The third material may be an adhesive. The layer structure, as assessed by oxygen permeability, shows (1) undesirable high permeability when too few layers allow easy passage around barrier layers, (2) low permeability (300-fold reduction) at 60-600 layers equal to theoretical minimum for lamellar structure, and (3) increased permeability as extremely thin laminates break up to from discontinuous domains (blends). LIM technology is offered for licensing by the Dow Chemical Company



Only machine modifications needed are addition of feedblock and multipliers. LIM does not require multiple channels or sequenced valving used in coinjection molding and can easily be applied to complex parts or multicavity molds. Parts can be molded with high barrier properties to gases and hydrocarbons at lower costs than monolayer materials. Scrap can be recycled by incorporation into the major component or by conventional methods since LIM materials are compatible. Optical clarity (reduced haze) is improved compared to blends because the ordered LIM morphology reduces light scattering. LIM struc-ture, with sheetlike continuous component selected for specific properties (controlled thermal expansion, increased load bear-ing, and temperature resistance), offers distinct property enhancements compared to blends.

Structural Parts (dimensional stabil-ity, temperature/chemical resistance) Housewares/Durables (clarity, temperature and solvent resis-tance) Containers for Food and Chemicals (gas. hydrocarbon barriers)Automotive Reser-voirs (fluid/heat resistance).

PC/PET, PC/PBT, PO/ad/EVOH, PET/PEN, PO/ad/PA, PS/PA6, PC/TPU, TP/T-LCP, filled/unfilled, brit-tle/ductile, vir-gin/recycle

Table A10. Multimaterial multiprocess (MMP) technology [10.1]

The use of multiple materials and processes is the overarching technology in achieving one-hot manufactur-ing for large and/or complex parts. Material thermal expansion differences can be dealt with by flexible joints, (adhesives), process sequence to minimize distortion, sliding at interfaces (incompatible materials), and design for minimum distortion. Common multimaterial multiprocess technologies include Injection Compression Molding in which resin is injected into a partially open mold which closes, requiring less clamp force and producing less residual stress in the part. Multimaterial Molding in which a material is shaped, the mold is altered, and a second (or subsequent) material is shaped. Shaping processes are combinations of injection and compression molding and stamping. In-Mold Coating in which a thin thermoset coating is injected onto an injection or compression molded part in a closed mold. ‘Mono-Sandwich’ Coinjection Injec-tion Molding in which a small extruder, operated intermittently, pumps a skin layer into the front of the main injection unit for subsequent coinjection. The Alpha 1 machine at GE Plastics, with two injection units, a long stroke vertical press, and shuttle table, allows combinations of compression molding, (gas-assisted) injection molding, and stamping. Other MMP technologies are described in tables on Low Pressure Molding, Advanced Blow Molding, and Lamellar Molding.

Table A9. Lamellar injection molding [9.1]


TABLE ABBREVIATIONSad adhesiveABS Acrylonitrile-butadiene-styrene copolymerEVOH Poly(ethylene-co-vinyl alcohol)PA PolyamidePBT Poly(butylene terephthalate)PC PolycarbonatePE PolyethylenePEN Poly(ethylene 2,6-nathalenedicarboxylate)PET Poly(ethylene terephthalate)PMMA Poly(methylmethacrylate)PO PolyolefinPPS Poly(phenylene sulfide)PP PolypropylenePS PolystyrenePSU PolysulfonePPO PolyphenyleneoxidePPE Poly(phenylene ether)SAN Poly(styrene-co-acrylonitrile)SMC Sheet Molding CompoundT-LCP Thermotropic Liquid Crystal PolymerTPU Thermoplastic PolyurethaneTPE Thermoplastic Elastomer

TABLE REFERENCES1.1 Eckardt, H., “Gas-Assisted Injection Molding,” Ref. [1].2.1 Turng, L.S. “Computer-Aided Engineering for the Gas-Assisted Injection Molding Process,” Ref [1]3.1 Hauck, C. , Schneiders, A., “Injection Molding with Fusible Core Technology,” Ref. [1].4.1 Hettinga, S., “Controlled Low Pressure Injection Molding,” Ref[1].4.2 Turng, L.S., Chiang, H., Stevenson, J.F., Plast. Eng., p.33, Oct. 1995.


The advantages of more than one material and/or process include design flexibility, tailored perfor-mance, effective material use, lower labor costs, improved quality through automation, reduced sec-ondary operations, less auxiliary equipment, and more recycle use. Multiple materials allow advan-tageous combinations such as multiple colors (auto-motive lens), flexible/rigid (conduits with connectors), and consolidated/strong (plastic/metal composite) and cost /barrier or strength (laminate structure).

Telephone booth molded from 132 lbs of structural foam on a 2500-ton press with three injection units Air vent with molded mov-able louvers made from incompatible materials. Automotive bumper with injection molded fascia over a stamped beam. Multi-color automotive taillights.

Combinations of ther-moplastic, thermoset, and reinforcements subject to constraints of product perfor-mance, limitations on distortion, and inter-face requirements (adherent or incom-patible)

Table A10. Multimaterial multiprocess (MMP) technology [10.1]


5.1 Sugiura, S., “Developments in Advanced Blow Molding,” Ref. [1].5.2 Myers, J., Mod. Plast., p.64, June 1995. 6.1 Bigg, D.M., “Manufacturing and Formation of Thermoplastic Sheet Composites,” Ref. [1]. 7.1 Castro, J.M., “Reactive Liquid Composite Molding,” Ref. [1]. 8.1 Suh, N.P., “Microcellular Plastic,” Ref. [1]. 8.2 Campbell, G.A. and Rasussen, D.H., U.S. Patents 5,369135, 5,358,675 (1994). 9.1 Barger, M.A., Schrenk, W.J., “Lamellar Injection Molding Process for Multiphase Polymer Systems,” Ref. [1].10.1 Avery, J.A., “Multimaterial Multiprocess Technology,” Ref. [1].

Chapter 1: Gas Assisted Injection MoldingGas Assist Injection MoldingThe North American Legacy

Jack AveryGE Plastics, USA

Let's look at some of the specifics. What impact will the expiration of the original Frederichpatent have? As you would expect, there are a variety of opinions. After having discussedthis with a variety of people, it is my opinion that it will have little actual impact on thedevelopment of gas-assist injection molding. Why? Most applications use in-runner orin-article options. These variations provide the most flexibility to introduce gas into the partand consequently the most flexibility to optimize both the design of the component and theutilization of the process.

Design will dictate the best solution for gas injection. But there will always be someapplications where through the nozzle technology is the best choice. Typically these arehandles, symmetrical components, and similar type applications. Typical applications ofgas-assist injection molding include:• Business machine chassis/ housings• Material handling pallets• Furniture: chairs, tables• Handles• Automobile bumpers• Automobile trim• Television housings• Golf club shafts

Not only has progress been made in licensing and utilization of gas-assist injectionmolding technology, advancements in computer simulation, design and variations in thetechnology continue.

GAS ASSIST INJECTION MOLDING TECHNOLOGIES

Several variations of gas-assisted injection exist. Most are patented and require licensing topractice.


Commercial Technologies License RequiredCinpres YesGAIN YesJohnson Controls NoAirmould (Battenfeld) NoHelga (Hettinga Industries) Yes

Another trend is that machine manufacturers are entering into licensing agreementswith Cinpres or GAIN or both. This enables them to integrate the control system forgas-assist injection into their machines as an option. Companies having such agreementsinclude:• Cincinnati Milacron Cinpres• Engle Machinery - Worldwide GAIN• HPM Industries GAIN• Husky Cinpres

Gas assist-injection molding is a global process. One of the drivers is that globalOEM's recognize its value and are beginning to apply the technology on a broad base.Examples are:• Samsung• Mitsubishi Consumer Electronics• Ford• Xerox

Another significant factor in the globalization of gas-assist injection molding is that thetechnology suppliers have a global presence. Battenfeld, Cinpres and GAIN all have rela-tionships that enable them to serve the worldwide market, either directly or on a primarylicenser basis. In addition, Cinpres is opening an office in Singapore to serve Asia (Table 1).

Table 1.

N. America Europe Japan

Airmould Battenfeld Battenfeld Tsukishima

Cinpres Cinpres Ltd. Cinpres Ltd. Mitsubishi Gas Chemical

GAIN GAIN GAIN Asahi

Helga Hettinga Hettinga Toray

Johnson Controls JCI JCI JCI

Gas Assist Injection Molding 17

LICENSING GAS-ASSIST INJECTION TECHNOLOGIES

A significant difference between the use of gas-assist injection technology in Europe andthe rest of the world is the need for licensing. Two companies, Cinpres and GAIN hold avariety of patents on this technology. To use any of these patented technologies anywhere inthe world, and to ship a product manufactured using gas-assist technology anywhere in theworld, a license is required.

Three licensing options are available:• Patent only license• Development license• Full technology license

A patent only license provides protection to use technology patented by the licenser.An example of a patent only license is Engle who requires that a GAIN Technologies patentonly license be taken prior to or in conjunction with purchase of Engle equipment.

Machine manufacturers with these agreements include:Cinpres GAIN

Engle X XHusky X -Klockner Ferromatik Desma X -Mannesman Demag X -

A development license can be taken for the development phase of an application.Under this arrangement, the technology is in your facility and access to support from thetechnology supplier is available. In some cases, this license is used to evaluate the technol-ogy for a specific application.

The development license must be converted to a full technology or full manufacturinglicense prior to going into production. Each technology supplier has variations on the typeof license and fees. A summary of the license fees and "hardware" cost associated withusing gas assist technology in production is in Table 2. It is necessary to review each indi-vidual case with the technology licenser to determine actual costs and conditions.


(1) Licensing fees and details vary depending upon each application and supplier. It is essential to obtain from each supplier relative to a specific application(2) Per mold license is also available on a lifetime basis: $12.5 - 75,000(3) "HELGA" Package includes required equipment and rights to practice(4) Available only as an option on new or as a retrofit on existing Johnson Controls' machines(5) Patents pending(6) Airmold process per Airmold specification, no license required(7) Includes pressure generator which can be used with additional machines

APPLICATION DEVELOPMENT

What is different in the application development process? The most important step is todetermine if the application is appropriate for the gas-assist injection process.

Table 2. Gas-assist injection molding licensing information (1)

Manufacturing license fee

"Gas injection" equipmentAdditional costs

"royalties"

Airmold (Battenfeld)(6)

None Single Machine, Single Injection Point, Base Price $110 000, Expandable(7)

None

Cinpres(5) $60,000 Single $35,000, Multiple $58 - 95,000

Based on: Material Usage or Tooling Fee or Flat Fee for Parts

Epcon - Single $55,000 Multiple $77,500 None

GAIN Per mold $1.5 -15,000/yr per facility $25 - 250,000/yr(2)

Single $25 - 50,000 Multiple $35 - 85,000

None

HELGA (Hettinga)(3)

None HELGA Package $70 - 75,000 None

Johnson Controls Multinozzle/Sequential Gas Assist(4)

None Integrated into machine controls $30 -50,000

None

Nitrojection $25,000 $45 - 85,000 None


How is this accomplished? The first step is to complete a thorough assessment of theperformance requirements. Then the material, process and design of the component can bedetermined. Factors which make gas assist injection the process of choice include:• High stiffness to weight ratio is required • Part design allows for hollow rib geometry • Tight tolerances are required • A hole in the part surface can be tolerated, or the hole can be sealed • Improved surface is desirable vs. structural foam molded parts

Once the performance requirements have been determined and gas assist injection hasbeen selected as the appropriate process, part and tool design proceed. It is important to takeinto consideration details which are different from standard injection molding. They are:

PART DESIGN CONSIDERATIONS

• Sizing of gas channels• Gas channel layout• Location of gas injection point(s)

TOOL DESIGN CONSIDERATIONS

• Gate size for the through-the-nozzle or in runner gas injection• Gas nozzle location - for in-runner and in-article gas injection, material must cover the

gas nozzle prior to gas introduction• Location of gas nozzles in the tool to prevent interference with cooling lines, slides,

ejector systems, etc.Where can you obtain assistance? Four primary sources of assistance exist:

1) Gas-assist injection technology supplier (Cinpres, GAIN, Battenfeld, etc.)2) Material supplies may provide assistance if the application is a fit for their materials3) For OEM's the third source may be a molder who has experience with gas assist technology4) Consultants who specialize in gas-assist technology (such as Caropreso Associates) or firms that can provide modeling assistance, i.e. Plastics and Computer

As a general comment, licensees find that even though assistance is available, the finaltest is to learn through the experience of putting parts into production. A useful recommen-dation of many molders is to do a prototype tool for the first few applications. This will pro-vide you with some leeway for changes prior to the production tool. Also, in many cases,the prototype tool can be used for process validation, preproduction or initial parts while theproduction tool is being completed. Another option is to prototype a section of the part, i.e.


one-quarter. This could provide information critical to design and construction of the moldbut would not provide parts for evaluation. Each application needs must be considered.

ADDED COST USING GAS-ASSIST TECHNOLOGY

One factor often overlooked in the development of a program using gas-assist injectionmolding technology is the added cost involved. In addition to the licensing fees and royal-ties, equipment costs plus the nitrogen used in the process must be taken into account. Also,the cost of the mold may be higher than for standard injection molding since, except forthrough the nozzle technology, gasinjection nozzles must be integrated into the mold.

These added costs must be recovered. Some factors which may contribute to recover-ing costs are:• Parts consolidation resulting in fewer molds, less machine utilization and reduced or

elimination of assembly• Use of lower tonnage machines• Improved part quality• Reduced cycle time• Less scrap• Lower weight (lighter and less material)

Delphi Interior and Lighting Systems had significant challenges to overcome to meetGM requirements for door systems for new vehicles. Some of the requirements were:• Reduced systems cost• Systems mass reduction• Reduced assembly time• Improved system quality

Gas-assisted injection molding was selected as the process technology to meet thesedemanding requirements due to the benefits it offered:• High strength-to-weight ratio• Low molded-in stress provides excellent dimensional stability• Parts consolidation opportunities• Weight reduction• Design flexibility• Reduced part cost

When Delphi initiated the program in the early 1990's, gas-assist injection molding hadnot demonstrated the capability to deliver all of these benefits in a production environment.What did Delphi do to reduce the risk of employing a new technology in this high visibilityprogram? They brought in a gas-assist technology supplier and a material supplier who wasactively developing gas-assist design and process technology for their materials.


A four year collaboration produced the following results:• Parts consolidation: 61 parts to 1 part• Assembly time reduction from 330 sec to about 60 sec• Lighter weight (up to 1.5 Kg/vehicle)• Reduced tooling requirements and investment costs• Improved material handling• Better assembly ergonomics• Improved quality (fewer squeaks and rattles)• Reduced operational noise• Improved corrosion resistance• Best-in-class serviceability• Up to 10% piece price savings

This sounds like a successful program. The most important lesson is that the key partic-ipants were involved from the inception and worked through the development programtogether.

TECHNOLOGY DEVELOPMENTS

Technology developments continue. Design guidelines have been developed and published.The lead has been taken by material suppliers who have developed this information for usewith their customers in developing applications that employ gas-assist injection moldingtechnology. In addition, the Structural Plastics Division of the Society of the Plastics Indus-try has a compilation of papers relating to gas-assist injection molding technology whichhave been presented at their annual conferences.

Process technology development continues. Mitsubishi Gas Chemical has developed avariation of the Cinpres process - the "full shot" process. Instead of short shooting the moldand packing it out with gas, in the "full shot" process, the mold is filled with polymer andthe gas is used only for packing.

Nozzle design is one of the critical areas of gas-assist injection molding. For this to bea commercially viable process, injection of the gas through the nozzle must be as troublefree as injecting the material. This is an area of ongoing development by a wide variety ofsuppliers.

Cinpres has introduced a new "directional" nozzle design that ensures the flow of gasinto the mold in the same direction as resin flow. A nozzle with a 90 degree tip whereby thegas exits at right angles, ensures the flow of nitrogen into the mold in the same direction asresin flow. This configuration is claimed to prevent blemishes on the surface of the part thatappear opposite the nozzle.


Xaloy Inc. has introduced an upgraded nozzle that is claimed to increase durability, andreduce cost. This redesigned nozzle uses a hydraulically or pneumatically actuated needle toshut off melt flow when the hold pressure is released at the end of the injection cycle. Thisprevents gas in the sprue and runner system from flowing back into the nozzle.

Nitrogen source and recovery play a critical role in the cost and quality of componentsproduced using gas-assist injection technology. Studies underway indicate that nitrogenpurity is critical, especially when using engineering thermoplastic resins.

Impurities (i.e., oxygen) can result in oxidation and burning in the mold. Severalon-site nitrogen separation systems are available for prices ranging from $10,000 - $70,000depending on volume required.

Gas recovery is a topic of intense discussion, as it offers cost reduction opportunities.Recovery rates of 70-90% are attainable with some resins. However, volatiles can be pickedup by the gas flowing through the molten materials resulting in contamination or fouling ofthe recovery system and/or clogging of gas injection needles. Much more work needs to bedone in this area.

Design, analysis and material optimization are also critical elements. The full potentialof gas-assist technology will not be realized without continuing developments in theseareas. Dow, DuPont, GE Plastics all are working to optimize design for their materials. A.C.Technology, Plastics and Computers Inc. and Moldflow provide specialized software fordesign optimization and process simulation. Work continues as new releases of softwareincorporating more sophisticated approaches to the gas-assist technology continue to becommercialized.

In addition, gas-assist injection technology development continues. Cinpres and GAINcontinue their developmental work, and variations of the basic technology continue to beintroduced.

Another approach was introduced in late 1995 by EPCON Gas Systems Inc. Thisapproach is based on three key elements:• Shot control• Simplified gas pins• Pressure controlEPCON has applied for patents on their approach to shot control. It is based upon "strategi-cally" located thermocouples at points in the mold where the material flow front needs tostop before gas pressure completes the mold filling. The thermocouple simultaneously sig-nals the gas injection unit to begin the gas flow and the molding machine to stop its hydrau-lic pumps which end the injection forward sequence.

Their approach to gas pins is similar to the concept used for mold vents - the holes aresmall enough so that the material skins cover but allows the gas to pass through. According


to EPCON, "Microscopic holes have been incorporated to allow enough volume of gas topass through."

To date, I am not aware of any EPCON systems in commercial production. It remainsto be seen how well this technology performs.

Their approach to gas pins is similar to the concept used for mold vents - the holes aresmall enough so that the material skins cover but allows the gas to pass through. Accordingto EPCON, "Microscopic holes have been incorporated to allow enough volume of gas topass through."

EQUIPMENT

Variations on gas-assist injection molding continue to evolve. The latest is Battenfeld'sMonomodule. This unit is for use when gas is to be injected at a single point. Then only onepressure regulator is required. The handheld unit provides the capability to be programmeddirectly. The data is stored in the Monomodule.

SUMMARY

We all know that gas-assist injection molding technology languished for several years inNorth America due to litigation threats. How was this "fear" overcome?

As with any industry, there are leaders and there are followers. The leaders were con-vinced that the upside for productivity and design flexibility were worth the risk. Theystepped forward and developed applications using gas-assist injection molding technology.The initial applications took longer to develop than standard injection molding or withstructural foam. But once they have developed the technology, they have an advantage. This"confidence factor" is a competitive advantage that can be exploited and translated into abottom-line advantage over the followers.

My opinion as to why gas-injection molding technology is developing so rapidly inNorth America, has been the collaboration of the members of the supply chain involved indeveloping an application. In Europe, the machine suppliers have been the primary driversof gas-assist technology. Even though they had an earlier start, I feel that North America ismoving faster and has applied the technology to more challenging applications.

How have we been able to do this? The OEM's and material suppliers recognized thepotential this technology offered and joined forces with technology suppliers to apply it toapplications that offered a high return. The GM-Delphi Super Plug application is an excel-lent example of this.

For this process technology to become mainstream, progress must be made in severalareas.


1) The "knowledge gap" must be overcome. It must be able to be done right the firsttime. The development cycle for components molded using gas-assist technology must beno longer than for a standard injection or structural foam molded component.

2) Recognition that gas-assist injection is not a "fix-it" for existing problems. As withany process, to optimize value, the product must be optimally designed for a material andprocess.

3) Additional costs of using gas-assist injection must be understood and minimized.The value of components molded using this technology must be recovered.

4) Equipment consistency and reliability issues (gas injection control and nozzle) mustbe eliminated.

5) "Fear of litigation" must be eliminated!Commercialization of gas-assist injection technology can take several paths. Three of

which might be:1) Technology becomes widely understood and low cost to implement. Most convert-

ers become "experts" and utilize.2) A group of converters specialize in gas-assist injection molding. Their expertise

enables them to develop components quickly and cost effectively, positioning them as thesuppliers of choice for gas-assisted injection molded components.

3) Captive converters become the largest users of gas-assist injection. They have thelarge volume applications which enable them to amortize the added cost.

What is needed for this technology to prosper? Higher levels of collaboration. TheOEM must involve the processor, material supplier and tool maker at the program inception.This will ensure that all required input is provided early and that costly changes and/or com-promises are eliminated or minimized. This will also develop an increased sense of owner-ship by the members of the supply chain.

Another area of need is increased education of the design community. For proper appli-cation of gas-assist injection molding, the designer must understand the process, itsstrengths and weaknesses in order to minimize design flaws which result in process, perfor-mance and economic issues.

Will the rapid growth continue? Not unless the technology becomes widely understoodand is demonstrated to provide cost effective solutions. Designers must become confidentnot only in designing components, but also that mold makers and converters have the capa-bility to implement the technology.

SOURCESBattenfeld of America, Inc., 31 James P., Murphy Industrial Highway, W. Warwick, R.I. 02893, Phone (401) 823-0700, Fax (401) 823-5641


Cinpres Limited, Waterworks Plaza, Building, 3135 South State Street, Suite 108, Ann Arbor, MI 48108, Phone (313) 663-7700, Fax (313) 663-7615Epcon Gas Systems, Rochester Hills, MI, Phone (810) 651-9661, Fax (812) 650-8293Gain Technologies, Inc. 6400 Sterling Drive, North, Sterling Heights, MI 48312, Phone (810) 826-8900, Fax (8I0) 826-8906Hettinga Technologies, Inc., 2123 N.W., 11th Street, Des Moines, IA 50325, Phone (515) 270-6900, Fax (515) 270-1333Johnson Controls, Inc., 10501 Highway, M-52, Manchester, MI 48158, Phone (313) 428-8371, Fax (313) 428-0143Kontor- JPI Technologies Inc., 35 Gibson, Lake Dr., Box 220, Palgrove, ON, LON 1PO, Canada Phone (905) 880-2600, Fax (905) 880-2599AC Technology/C-Mold, 11492 Bluegrass, Parkway, Suite 100, Louisville, Ky 40299, Phone (502) 266-6727, Fax (502) 266-6654Moldflow Pty, Limited, 4341 S. Westnedge, Suite 2208, Kalamazoo, MI 49008, Phone (616) 345-4812, Fax (616) 345-4816Plastics & Computers, 14001 Dallas, Parkway, Suite 1200, Dallas, TX 75240, Phone (214) 934-6705, Fax (214) 934-6755

Flow Directions in the Gas AssistedInjection Molding Technology

Young Soo Soh and Chan Hong ChungDepartment of Chemical Engineering, Kyungbook, 712-714 Korea

INTRODUCTION

Gas assisted injection molding process produces parts with many advantages includingClass A finish with no sink marks, reduced cycle time, lower injection pressure, lowerclamping tonnage, reduced part warpage, greater design freedom, and more.

In the gas assisted process, an inert gas is injected into the center of the flow of plastic.A combination of the high surface tension of the plastic and the lower viscosity of the hottermolten material in the center of thicker sections, such as ribs and bosses, confines the gas toform hollow area in the thicker sections of the parts.

Most gas and assisted injection molded parts may be categorized into two types: Theparts consist merely of single thick section through which the gas penetrates, and the partsconsists of a nominal thin wall with gas channels traversing the parts. The latter are moredifficult to design and process because the gas may not just flow through the channel butpenetrate into the nominal thin walled. These parts are expected to be designed such that thegas cores out all the channels without penetrating into the thin walls.

To design molds such that the gas cores out all the thick sections and not the thin walls,one needs to predict and understand the preferred direction of gas in the process. In thispaper, we use a method to relate the preferred gas direction with the process variables. Themethod requires a knowledge on the relations between resistance for the gas flow and pro-cesses variables such as resin flow length, cross section area of cavity, melt temperature,and existence of short shot. A simulation package was used to confirm the method.

Commercial packages simulate the flow of gas. At a mold design stage, the commer-cial package plays a very important role to prevent blow through or fingering phenomena bysimulating the gas flow. At the pilot production or first mold trial stage, the package alsoplays an important role to make perfect parts. If, however, the packages are not available atthe molding shop or instant solution to prevent blow-through or fingering is necessary at themold trial stage, the equations in this paper are very useful to treat the troubles. When atrouble shooting engineer modify the virgin mold without a flow analysis package, initially


he needs qualitative information described in the theory, not quantitative. After qualitativequestions are answered, the quantitative solution comes from trial and error. For example, ifthe theory tells that the channel size be enlarged, the shop mold maintenance technician willenlarge the channel diameter a little and try the molding cycle, which then will be followedby another small channel enlargement, if necessary, until satisfactory parts come out of themold.

THEORY

Although the process is unsteady state, steady state flow equations may be used to explainthe gas flow directions. The equation for the steady state flow of a Newtonian fluid betweeninfinite parallel flat plates is given by1

[1]

whereL length of plate in direction of flow a distance between plates

p pressure drop across the distance V average velocity

neglecting end effects.The equation for the steady state flow of pseudo plastic liquids between infinite paral-

lel flat plates is given by

[2]

where m, n power law indices Q flow rate

The steady state flow of a Newtonian liquid through conduit with diameter D is givenby1

[3]

The steady state flow of pseudo plastic liquids through conduit with radius R is givenby

[4]

When the direction of gas path is discussed for the gas assisted injection molding, theterm "the direction of least resistance" is commonly used. When more than two paths are

∆p 12µVL

a2

-----------------=

∆

∆p2L-------

Q 3n 1+( )πn

------------------------- n m

a3n 1+( )

------------------- =

∆p 32µVL

D2

-----------------=

∆p2L-------

Q 3n 1+( )πn

------------------------- n m

R3n 1+( )

-------------------- =

Flow Directions 29

competing for the direction of gas, the gas prefers the direction of less resistance. Veryoften, this resistance is explained by temperature, length, or distance between plates. Thedegree of resistance is proportional to the pressure drop requirement in equation [1] through[4].

Using these equations, the preferred gas directions can be predicted. Equations [1] and[3] are easier than equations [2] [4] to use, and equations [1] and [3] are as correct as equa-tions [2] and [4] to answer qualitative questions such as "which one is the least resistancepath?"

Theoretically, there exists pressure drop both along the gas phase and along the poly-mer melt phase. However viscosity of gas is less than 0.1% of apparent viscosity of polymerresin, and the pressure drop along the gas phase can be considered negligible. Thus the pres-sure of the gas may be considered the same for all regions of gas not only in the stage of sta-tionary packing stage, which comes from fluid statics theory, and but also in the phase offirst dynamic stage. Hence, only the pressure drop of the resin is necessary for the discus-sion of the degree of the resistance.

The resistance increases with the increase of viscosity, path length between gas frontand melt front, velocity of the melt, and decreases with the increase of path cross sectionarea. The path cross section area here excludes the frozen layer of mold cavity

RESULTS AND DISCUSSIONS

Consider that a polymer melt was injected in the middle of a 8 mm diameter pipe to a totalmelt length of 60 mm, followed by gas injection at the point 0.5 mm left of the melt injec-tion point, and we now try to pick the preferred direction of gas. With the equation 1, onecan calculate that the resistance to the left hand side for the gas to move with velocity, V, isproportional to

where ( p)L is pressure drop requirement to the left hand side direction and LL is distancebetween gas injection point and left melt front. The resistance to the right hand side for thegas to move with velocity, V, is proportional to

One only needs to compare LL with LR as all the remaining variables are the same. LLis less than LR and ( p)L is less than ( p)R. Thus the resistance of the flow to the left handside direction is smaller, and the left hand side direction is the preferred direction for gasflow.

∆p( )L

32µVLL

D2

--------------------=

∆

∆p( )R

32µVLR

D2

--------------------=

∆ ∆


In Figure 1, a Mold Flow simulationresults are shown, which is consistent with themethod given here.

Consider another case where a pipe with 8mm diameter is connected to a pipe with 4 mmdiameter. The melt length from the center is thesame for both directions. The gas is injected atthe center and we now try to pick the preferreddirection of gas. With the equation 1, one can

calculate that the resistance to the left hand side is proportional to

and the resistance to the right hand side is proportional to

Thus

and

( p)L is less than ( p)R. and the resistance ofthe flow to the right hand side direction issmaller, and the right hand side direction is pre-ferred direction of gas flow. In Figure 2, MoldFlow simulation is shown for the case, which isconsistent with the method given here.

Consider the third case where a pipe with 8mm diameter is connected to a pipe with 4 mmdiameter, where gas is injected at the pointwhere two pipes are connected. The melt lengthfrom the gas injection point is different at each

side. We now try to pick the preferred direction of gas. With the equation 1, one can calcu-late that the resistance to the left hand side is proportional to


∆p( )L

32µVLL

DL2

--------------------=

∆p( )R

32µVLR

DR2

--------------------=

∆p( )L 32µV 30 16⁄( )=

∆p( )R 32µV 30 64⁄( )=

∆ ∆

∆p( )L

32µVLL

DL2

--------------------=

Figure 1. Gas injection - Case I.

Figure 2. Gas injection - Case II.

POLYMER SHUT OFF

GAS INJECTION

polymer polymer

POLYMER SHUT OFF

polymerpolymer

GAS INJECTION

30mm 30mm

D = 8mm

D = 8mm D = 8mm29.5mm 30.5mm

D = 4mm

Flow Directions 31

Thus

and

( p)L is smaller than ( p)R and the resistanceof the flow to the right hand side direction isgreater, and the left hand side direction is pre-ferred direction of gas flow. In Figure 3, MoldFlow simulation results is shown for the case,which is consistent with the method given here.

Consider case 4, where a pipe with diame-ter 7 mm is connected to a cavity of 7 mm thick-ness formed by two parallel plates, where gas isinjected at the point where two cavities are con-

nected. The melt length from the gas injection point is the same at each side. With the equa-tion 1, one can calculate that the resistance to the left hand side is proportional to


Thus

and

( p)L is greater than ( p)R and the resistance ofthe flow to the left hand side direction is greater,and the right hand side direction is preferreddirection of gas flow. In Figure 4, Mold Flowsimulation is shown for the case, which is con-sistent with the method given here.

∆p( )R

32µVLR

DR2

--------------------=

∆p( )L 32µV 10 16⁄( )=

∆p( )R 32µV 100 64⁄( )=

∆ ∆

∆p( )L

32µVLL

DL2

--------------------=

∆p( )R

12µVLR

a2

--------------------=

∆p( )L 32µV 20 49⁄( )=

∆p( )R 12µV 20 49⁄( )=

∆ ∆

Figure 3. Gas injection - Case III.

Figure 4. Gas injection - Case IV.

POLYMER SHUT OFF

D = 8mm

POLYMER SHUT OFF

GAS INJECTION

polymer

GAS INJECTION polymer

polymer

20mm 20mm

10mm 100mm

D = 4mm


Consider case 5, where a 7 mm diameter pipe is connected to a cavity of 7 mm thick-ness formed by two parallel plates, where gas is injected at the point where two cavities areconnected. The melt length from the gas injection point is 40 mm at the plates and the samefor the pipe. With the equation 1, one can calculate that the resistance to the left hand side isproportional to


Thus

and

( p)L is greater than ( p)R and the resistance ofthe flow to the left hand side direction is greater,and the right hand side direction is preferreddirection of gas flow. In Figure 5, Mold Flowsimulation is shown for the case, which is con-sistent with the method given here.

Consider case 6, where a 7 mm diameterpipe is connected to a cavity of 7 mm thicknessformed by two parallel plates, where gas isinjected at the point where two cavities are con-nected. The melt length from the gas injectionpoint is longer at the plates with 60 mm. With

the equation 1, one can calculate that the resistance to the left hand side is proportional to


Thus

∆p( )L

32µVLL

DL2

--------------------=

∆p( )R

12µVLR

a2

--------------------=

∆p( )L 32µV 20 49⁄( )=

∆p( )R 12µV 40 49⁄( )=

∆ ∆

∆p( )L

32µVLL

DL2

--------------------=

∆p( )R

12µVLR

a2

--------------------=

∆p( )L 32µV 20 49⁄( )=

Figure 5. Gas injection - Case V.

POLYMER SHUT OFF

GAS INJECTION

D = 7mm

polymer

20mm 40mm

Flow Directions 33

and

( p)L is smaller than ( p)R and the resistance of the flow to the right hand side direction isgreater, and the left hand side direction is the preferred direction of gas flow. In figure 6,Mold Flow simulation is shown for the case, which is consistent with the method givenhere.

REFERENCES1 W. L. McCabe, J. C. Smith, and P. Harriot, Unit Operations of Chemical Engineering, 4 th Ed. McGraw-Hill, 1986.

∆p( )R 32µV 60 49⁄( )=

∆ ∆

Gas-assisted Injection Molding: Influence ofProcessing Conditions and Material Properties

Kurt W KoellingDept. of Chemical Engineering, The Ohio State University, Columbus, Ohio 43210, USA

Ronald C KaminskiThe Geon Company, One Geon Center, Avon Lake, Ohio 44012, USA

INTRODUCTION

In general, gas-assisted injection molding can be described by a simple three-step process.1

A short shot of molten polymer initially fills 75-98% of the mold cavity under the ram speedcontrol of the injection molding machine. After a short delay period, compressed nitrogengas cores out the molten polymer, filling the remainder of the mold. The third step, or thegas packing stage, occurs as a result of the volumetric shrinkage of the polymer melt. As theplastic solidifies, the gas expands into the volume created by shrinkage, locally packing outthe part.

In 1935, Fairbrother conducted the first experiments investigating this flow phenomenausing a viscous Newtonian solution. He found that m, the fractional coverage or fraction ofliquid deposited on the walls of the tube after bubble penetration, is a function of the capil-lary number, Ca, for capillary numbers up to 0.009.2 The fractional coverage, m, is definedfor tube-shaped geometries as:

m = Ap/ At = 1 - (rb/R) 2 [1]where Ap is the polymer cross-sectional area, At is the tube cross-sectional area, rb is theradius of the gas bubble, and R is the radius of the tube. The capillary number is defined asthe product of the bubble velocity, Ub, and the viscosity of the fluid, , divided by the fluidsurface tension, , or:

Ca = Ub / [2]Taylor investigated this problem further in 1961 and ran experiments that extended to

capillary numbers of two.3 Cox found that the fractional coverage reached an asymptoticvalue of m = 0.60 for capillary numbers greater than ten for viscous Newtonian fluids.4

With interest renewed in this problem because of gas injection molding, Poslinski andStokes conducted similar isothermal experiments using silicone liquid pastes that behavedas Bingham fluids.5,6 Capillary numbers of up to 800 were obtained and showed good cor-

ηΓ

η Γ


relation with previous work at low capillary numbers. However, the fractional coverage ulti-mately approached a value of m = 0.564 at high capillary numbers.5 An isothermalcomputer model was developed to simulate the gas-liquid dynamics to validate the experi-mental results. It was discovered that the hydrodynamic layer of molten polymer depositedby the passage of the gas bubble was much larger than the associated frozen layer developedat the mold-melt interface.

Polymer solutions, including a viscous, Newtonian fluid, an elastic Boger fluid withNewtonian shear viscosities, and a shear-thinning polymer solution have been used in iso-thermal experiments to characterize the role of melt rheology in the gas-assist process.6

Fractional coverage data from the shear-thinning fluid compared well with the Newtoniandata at very low capillary numbers, but dropped abruptly at a capillary number of two andapproached a limiting value much lower than m = 0.60. The Boger fluid also showed goodagreement at low capillary numbers, but began to climb at a capillary number of approxi-mately six, ultimately reaching a value of m = 0.75.

Simple correlations for the actual gas-assisted injection molding process have alsobeen developed.7-11 It was shown that the wall thickness of the molded part has a depen-dence on the residual time of the gas bubble.7,8 The residual time is defined for each pointon the flow path of the gas bubble as the difference in time between the passing polymerfront and the moving bubble tip. This research indicates that increasing residual times resultin higher fractional coverage. Others, however, have performed research to demonstrate acapillary number dependence by accounting for the solid wall thickness build-up as a func-tion of residual time.9-11

EXPERIMENTAL BACKGROUND

A spiral tube mold cavity of 0.0127 m (0.5 in)diameter with a flowlength of 0.585 m (23 in)was utilized for the molding experiments. SixKistler melt pressure transducers, model 6159Aand 6157 type, were mounted flush in the fullspiral mold cavity. The placement of the firstfour transducers allowed measurements of theadvancing gas bubble in the region filled duringthe polymer injection step of the process. Thefifth and sixth transducers were situated in aregion filled during the gas filling step of thecycle.

Figure 1 - Experimental apparatus for gas-assisted injec-tion molding trials.

CCD video camera

Controlled volume gas injection unit

75 ton injection molding machine

High resolution monitor,S-VHS video

recorder, and 486 DX2 PC with dataacquisition board

Influence of Processing Conditions 37

The trials were conducted at the laboratory facilities of the Ohio State University Engi-neering Research Center for Net Shape Manufacturing (ERC/NSM). The equipment used tomold the test parts is shown in Figure 1. An all-electric 75 ton ACT-B Cincinnati Milacroninjection molding machine was utilized in conjuction with a single cylinder Cinpres gasinjection unit equipped with a Cinpres II gas nozzle for a constant volume GIM process.

The melt pressure transducer measurements of the gas bubble advancement were veri-fied in preliminary trials by replacing the moving side core plate with a two-inch thick, cir-cular, borosilicate glass plate, fixtured into a protective steel frame. This produced a cross-section that was one-half of the full tube. A mirror positioned at a 45 degree angle wasplaced directly behind the glass window. Mounted above the mirror was a Cohu CCD videocamera, model 4915-2001. The video image captured by the camera was recorded by a highresolution, S-VHS videotape recorder at a rate of 60 frames/second. While the video imagesfrom the GIM process were recorded, the signals from five melt pressure transducers,amplified by Kistler model 5012 charge amplifiers, were fed to a data acquisition system. AKeithley MetraByte DAS - 1600 series data acquisition board, installed in an IBM-compati-ble 486 DX2 computer, sampled the amplified signals at a rate of 200 samples/second perchannel. Comparisons between the videotaped images and the pressure profiles, were thenperformed to verify the measurement technique.

During the molding experiments, an electronic Mettler AE-100 balance was used toweigh each set of full spiral parts, ensuring as much repeatability as possible during the tri-als by monitoring the shot weight. A Sheffield model RS30 co-ordinate measuring machine(CMM) evaluated the gas bubble area of each cross-section from the molded spirals. Threepoints were measured on the part exterior, along with one point near the gas bubble center.Twenty-five points around the circumference of the bubble surface were then measured.These points were connected to the initial probe position to calculate the bubble area bydividing it into twenty-five small triangles.

Three transparent injection molding gradecompounds were utilized in this study: a generalpurpose polystyrene (Dow Styron 685 D), a rigidpoly(vinyl chloride) (GEON 87781), and a highviscosity polycarbonate (GE Lexan 101). Eachcompound provided the opportunity to examinethe effects of polymer rheology on the resultingfractional coverage or wall thickness of themolded test spirals. Figure 2 displays the shearrate dependent viscosity of each material at its'base melt temperature, as predicted by the Cross-

Figure 2. Cross-exponential viscosity characterizations of experimental materials.

η

(Pa-

s)

10000

1000

100

10

10.01 0.1 1 10 100 1000 10000 100000

γ (sec )-1

PS

PVC

PC


exponential model. Model parameters were taken from the material selection database ofAC Technology's injection molding simulation software.

RESULTS AND DISCUSSION

The role of the melt rheology and the gas bubble velocity during the deposition of thehydrodynamic polymer layer is shown by the response of the wall thickness to changes inthe gas compression rate. Figures 3 and 4 show the fractional coverage as a function of dis-tance down the spiral for three different piston speeds of the gas cylinder. In Figure 3, alarge change in wall thickness is shown for the polystyrene depending on the piston speed ofthe gas cylinder. The slowest gas compression rate created a more uniform wall thicknessdistribution, before dropping off in the region filled during gas bubble penetration. Con-versely, the fastest gas piston speed caused a non-uniform, decreasing wall thickness distri-bution.

In contrast to the polystyrene, the response of the polycarbonate to the changes in pis-ton speeds was negligible until the end of the gas bubble advancement. As Figure 4 shows,there was no change in wall thickness until a flowlength of 0.375 m was reached. After thatpoint, slow piston speeds cause slightly thicker walls and fast piston speeds cause the wallsto thin, as with the polystyrene.

Subsequent trials were performed with pressure transducers mounted in the spiralmold. Since the times at which the polymer and gas fronts reached each transducer could bedetermined from the melt pressure profiles, residual times as a function of flowlength wereobtained. The time delay between the end of polymer injection and the beginning of gasinjection was the same for all three gas piston speeds for polycarbonate and polystyrene.The residual times, however, change as the gas compression rate changes. The lowest gaspiston speed produced the highest, most uniform distribution of residual times along the spi-ral flow path. In contrast, the highest setting lowered the residual times, with a resulting dis-tribution that is less uniform than either of the other two settings.

Figure 3. Fractional coverage vs. flowlength for three gas piston speeds using polystyrene (Tmelt = 487 deg F, Gas delay = 2.4 sec).

Figure 4. Fractional coverage vs. flowlength for three gas piston speeds using polycarbonate (Tmelt = 624 deg F, Gas delay = 2.4 sec).


residual times discussed above. Recalling the large changes in fractional coverage and theshear-thinning behavior of the polystyrene, the velocities for the three settings in Figure 5are not surprising. The interaction between the polymer melt rheology and the advancinggas front can be seen, as the velocities for the high setting diverge from those for the lowand base settings. As the gas bubble accelerates due to the decreasing resistance associatedwith polymer deposition on the mold walls far in front of the bubble tip, the melt shear-thins. This further decreases the resistance to gas bubble acceleration, causing faster bubbleacceleration, higher shear rates and continually decreasing viscosities and resistances toflow.

Since the polycarbonate has a large, upper Newtonian viscosity before a transition to ashear thinning regime and the polystyrene displays shear-thinning behavior over the entirerange of interest, it is reasonable to expect that the polycarbonate would show little responseto changes in gas piston speed and bubble velocities until they were large enough to causethe melt to shear-thin. As expected, the polycarbonate bubble velocities in Figure 6 showedvery different rates of gas front acceleration from those of the polystyrene. Comparing wellwith the bubble velocities, the fractional coverage did not begin to change as a function ofthe gas piston speed until the bubble velocities began to diverge and rapidly accelerate.

The issue of the relative importance of the gas bubble velocity in comparison to thethermal influences in the mold cavity may be addressed through plots of Fourier number vs.fractional coverage. The Fourier number, Fo, is defined as:

Fo = tr / Rt2

where is the thermal diffusivity, tr is the residual time, and Rt is the tube radius. Figures 7and 8 show the dependence of the wall thickness on the Fourier number. These plots aredivided into two regions where the polymer deposition on the mold walls is either as a result

α

α

Figure 5. Bubble velocities vs. flowlength for three gas piston speeds using polystyrene (Tmelt = 492 deg F, Gas delay = 2.65 sec).

Figure 6. Bubble velocities vs. flowlength for three gas piston speeds using polycarbonate (Tmelt = 637 deg F, Gas delay = 2.05 sec).

Flowlength (m)

Figures 5 and 6 demonstrate gas bubble velocities which were calculated from the

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50

100 100

10 10

1 1

0.1 0.1

0.01 0.01

Flowlength (m)

U (

m/s

ec)

U (

m/s

ec)

1.8 in/sec 1.8 in/sec3.1 in/sec 3.1 in/sec5 in/sec 6.2 in/sec

bb


of a polymer or a gas bubbl e dr iving force. In the region filled under polymer control wheresmall bubbl e velocities are common, thermal i nf luences cluster the fract ional coverageresults in a genera lly increasing tr end. In the region filled under gas control where the gasvelocities are highly varied, it is apparent t hat the wall thi ckness shows litt le or no depen-dence on the res idual time.

CONCLUSIONS

Gas-assisted injection molding experi ment s were per formed using a spiral tube mold andthree common injection molding gr ade compounds : polystyrene, polyvinyl chloride, andpolycarbona te. By measuri ng the wall thickness along the flow path of the gas bubble, theresidua l time, gas bubbl e velocity, and m aterial pr opert ies were found to be int errelated andresponsible for changes in fractional wall coating thickness of as much as 20%. The shapeof the shear rate dependent viscosity curve for ea ch ma ter ial was found to provide an indica-tion of the ability to cha nge the wall th ickness coverage by direct manipulation of the gasbubble velocity through proce ss param eters su ch as gas piston speed and pr e-charge gaspressure . Polymer melts which begin to shear -th in at low shear rates are more sensitive tochanges in gas pressure and gas piston speed, while those polymers that have significantupper Newtonian r egions ar e relatively insensitive to these changes.

ACKNOWLEDGMENT

This work was suppor ted by the Engineer ing Research Center for Net Shape Manufacturingat the Ohio State University and The Geon Company . The aut hor s would like to thank Dr.Clive Copsey and Mr . Scott Weir for their help in providing technical advice and a ssistanceand Mr . Alfred Geiger for hi s effor ts in collecting and summarizi ng resul ts from the PVCexper iments.

Figure 7. Dependence of wall thickness on Fou r ier nu m-ber u sing polycarb onate.

Figure 8. Dependence of wall thickness on Fouri er nu m-ber us ing polystyrene.

Fra

ctio

na

l co

vera

ge

Fra

ctio

na

l co

vera

ge

F illed under Gas Contr ol F illed under Gas Contr olF illed under Polymer Contr ol F illed under Polymer

Contr ol

0.8

0.75

0.7

0.65

0.6

0.55

0.5

0.45

0.4

0.8

0.7

0.6

0.5

0.4

0.001 0.01 0.1 1 0.00001 0.0001 0.001 0.01 0.1


REFERENCES1 Zheng, T., Glozer, G. and Altan, T, ERC for Net Shape Manufacturing at OSU Report, 1992.2 Fairbrother, F. and Stubbs, A.E , J. Chem. Soc., 1, 1935, pp. 527-529.3 Taylor, G.I., J. Fluid Mech., 10, 1961, pp. 161-165.4 Cox, B.G., J. Fluid Mech., 14, 1962, pp. 81-96.5 Poslinki, A.J. and Stokes, V.K , ANTEC Proceedings, 39, 1993, pp. 68-73.6 Huzyak, P., and Koelling, K., 1994 International Conference on Gas-assisted Injection Molding Technology, Ohio State University, November 1994.7 Findeisen, H., Diploma thesis at the IKV Aachen, 1991, Betreuer: A. Lanvers.8 Findeisen, H., 1994 International Conference on Gas-assisted Injection Molding Technology, Ohio State University, November 1994.9 Brockmann, C., Zheng, T., and Altan, T., ERC for Net Shape Manufacturing at OSU Report, 1994.10 Zheng, T., Koskey, J., and Altan, T, 1994 International Conference on Gas-assisted Injection Molding Technology, Ohio State University, November 1994.11 Zheng, T., and Altan, T., ERC for Net Shape Manufacturing at OSU Report, 1994.

Cover Part as an Application Examplefor Gas-assisted Injection Molded Parts

Michael HansenMack Molding Company, 608 Warm Brook Road, Arlington, VT 05250, USA

INTRODUCTION

The gas-assisted inj ection molding pro cess is in use now for sev-eral year s offer ing new technical and creative possibi lities forinj ection mol ding. Af ter a bri ef survey of the pr inciple sequenceof the process and basic pr ocess physics this paper comments onan applicat ion example for a cover part and pr ovides solutions forthe pr oblems found d uri ng the pr ocess of fixing existing issues onthis tool.

In Figur e 1, the pr inciple sequen ce of the pro cess is shown.In this process, the pressur ized nitrogen gas is injected into themelt to penetrate the part via a network of thicker cross-sectionedgas-channels. The pr ocess consists of everything f rom a partial toa volumetric filling of a cavity with polymer melt, as in compactinjection molding. This phase is followed by the injection of com-pressed gas, usual ly ni trogen, because of its availability, cost andiner tness. There is a vari ety of gas-assisted injection molding pro-cesses. In most of the processes, the gas is inj ected into the hotcore of the mel t through the nozzle and the sprue/runner system,or directly into the cavity via one or more gas needles. Due to analmost constant pressure in all areas with gas penetr at ion, a goodand even pr essur e distri but ion and t ransmission is guaranteed

across the molded parts. After the end of the gassing phase, the pressure is released either bygas recycling or blowing the gas into the atmosphere. As soon as ambient pressur e isreached, the molded par t can be ejected.

There are three main basic categori es of appl icat ions and some combi nat ions of these.The categories are as in the following. The first group includes tube- or rod-shaped part ssuch as for example clothes hangers and grab handles. The second category consists of large

Figur e 1. Sequence of the gas-assisted injection molding pro-cess.

the meltInjection of

Gas injectionandsubsequentfollow-uppressure

Venting andpart release


~

Figure 3. Influential factors for the production of gas-assisted injection molded parts.

-Design and layo11t of the partbased on e1."perlence

..CoRstJ11ction of a prototype tool

.-Te6iing or tht pl'Ototype tool

...-Detection of weaknesses

.-Rem011'lll ofweaknesies tltrongltmodiflcat1on I adaptation

(gfomcil'Y , illjcction point locatiOlls,2})5 ~ectlon locatIon)

.Layout or a nlM~ produdiOll tool bum cover-shaped structural parts with a network ofon the fmdings fium the prototype gas channels often combined with the rib struc-tool n,-outs .

ture of those parts such as e.g. business machmeFigure 2. Part and tool design for gas-assisted injection housings, automotive panels and outdoor furni-molded parts. ture. The third group encircles complex parts

consisting of both thin and thick sections where the process is often used mostly for part

integration by consolidating several assembled parts into a single design. Typical examplesare armrests or mirrors as designed for automotive applications.

The characteristic of the gas-assisted injection molding process leads to both variousadvantages and inherent problems associated with the design and processing. The key factorin being successful in producing with one of the gas-assisted injection molding processes is

to use the advantages to your benefits and to avoid the difficulties by optimizing part design,

material selection for the application as well as to choose the optimum technique.

STRATEGIES FOR PART AND TOOL DESIGN

In Figure 2, the particular steps for the optimization of tool and part design are listed. Thebuilding of a mass production tool is based on the knowledge and experience gained fromexperiments. In many cases, the use of programs for mold filling simulation is a good helpto optimize the final design.

INFLUENTIAL FACTORS

There are four different basic groups of influential factors involved in the production of gas-assisted injection molded parts as shown in Figure 3. These groups are material properties,processing parameters, part design and the gas-assisted injection molding technique. Sub-stantial is a part design adapted to the gas-assisted injection molding process. It is veryimportant, that designers take the specific process conditions and requirements associated

Cover Part as an Application Example 45

with these techniques into account in advance while developing new part designs. The nec-essary adaptation work and the costs associated herewith can be reduced to a minimum.

Furthermore the knowledge about interdependencies/interactions between part design,runner/gate-type and location, course and size of the gas-channels in connection with thefilling pattern after melt injection and processing parameters are a very important factor forachieving an optimized and stable production process.

The above-described procedure can be used for all application examples and offerssolutions for all the process-related phenomena/problems in each category of parts. Thecourse to optimize every single part differs a little bit from application to application. Butthe key principle is always to analyze the influential factors and adjust and improve those asmuch as necessary. With this better understanding of the process more and more of theadvantages can be used in order to utilize the high and still growing potential of the gas-assisted injection molding process in daily production and for new future developments.

In the following the experiences in fixing an existing production tool is discussed. Theanalysis starts in this case with a try-out to determine the problems associated with tool/design. After the weaknesses are determined the necessary adaptations are will be imple-mented and tested in a subsequent try-out. In order to achieve the best possible result for theproduction all the above-mentioned influential factors were investigated while testing andanalyzing the tool.

GAS–ASSISTED INJECTION MOLDED COVER PART

Because of the physics of the gas-assisted injection molding process cover parts require anetwork of gas-channels to provide the guidance for the gas entering into the hot meltthrough nozzle or gas pin. The more than one-dimensional flow path of cover-shaped struc-tural parts ask for an excellent balancing of the injection point location, the gassing locationand the melt filling pattern before gas injection. These are only some of the boundary condi-tions in connection with a cover application, which need to be taken into consideration. Thesize of the part reviewed in the following is 985 mm x 560 mm with a 3 mm wall thickness.

The material used in this application is a glass-fiber filled PPE/PS resin. The materialwas chosen for this application because of structural requirements, heat and chemical resis-tance. The cavity surface of the finished part is a class A textured surface. The problem withthe glass-fibers is to create an acceptable surface finish for the subsequent texture paintingprocess without showing visible sink, hesitation or flow marks. The material solidifiesquickly once the flow front stagnates. In areas with weld lines problems can occur to pro-duce an acceptable surface without glass-fibers showing up at the surface. If those areasshow up another secondary filling/sanding operation would be necessary.


PROBLEM DESCRIPTION

In case of this cover application the main issuesare as described in the following. The melt fill-ing pattern before gas-injection has to match thegas-channel network along the structural ribs.The task is to determine the appropriate degreeof pre-fill with polymer before gas-injection,which makes possible to produce parts withoutthe above-mentioned surface imperfections. Theconfiguration of the injection sprue, gas pin andthe network of gas- channels are shown in Fig-ure 4. This con-figuration of gas and melt injec-tion creates certain boundary conditions for thepre-fill with melt, which are illustrated in thefollowing chapter. Figure 5 shows the injectionsprue and the gas pin location as well as one ofthe filling stages before gas injection.

Very important for the production of the partis to avoid sink marks on the cavity as occurringin areas around screw bosses and thick ribs run-ning across the length of the part in center and onthe side opposite to the screw boss locations.

FINDINGS FROM PRACTICALMOLDING EXPERIMENTS

In the try-outs with the existing tool the next stepwas to find weaknesses/necessary adaptations forcorrecting the problems occurring in the massproduction. Prerequisite for running a successful

production is to establish a stable molding process.One important prerequisite for the production of gas-assisted injection molded parts

are the filling pattern before gas injection. This needs to be adapted to the planned or exist-ing course of the gas-channel network. The part design is dependent upon size and shape, aswell as the course and length of the gas-channels. The information for the final layout canbe obtained from short shot studies as well as from mold filling simulation programs.

Figure 4. Core side of cover part with gas-channel net-work.

Figure 5. Filling pattern showing the location of sprue and gas pin.


Another important factor is to check the process for repeatability in production. Tocompensate for slight fluctuations in the injection stage with polymer it is very important toensure the same conditions for the subsequent gas injection via gas pin. Changes in the shortshot sequence can create a narrow processing window for the gas-assisted injection moldingprocess and under certain circumstances it results in a borderline running process.

Due to the lower flow resistance there is a significant foreflow visible in areas wheregas-channels are running along ribs or across surfaces. This flow behavior determines thedegree of pre-fill with polymer before gas injection. As a result of the gas pin position andthe gas-channel connecting the gassing location the whole surface area around the gas pinlocation needs to be completely filled before gas injection. Otherwise a gas blow-throughwould occur.

The height (51 mm) and a 3 mm wall thickness of the main high rib running almostacross the whole center portion of the part results in a sink mark potential on the later tex-ture painted surface. There is also a thick rib running almost all the way along the part onthe side of the 6 hinges as shown in Figure 4. This rib is a bigger problem from a processingprospective because there are more than one gas–channel intersecting with this rib. A sinkmark far away from the gassing location can only be avoided when the gas hollows out thearea where the rib intersects with the main surface. To achieve this there are at least 3 differ-ent gas-bubbles necessary created via the different gas-channels attached to the rib.

The most critical factor in producing these parts is to figure out a way to keep the gasbubble penetrating through the hot melt always in the same areas of the part. This is the keyto a successful and stable serial production. Slight fluctuations in gas-bubble extension andpath can lead to significant changes in the molded part. The ability to keep the gas-bubblepenetration in certain limits is the boundary line between running a successful application orhaving a constant fight with a production problem.

In connection with the processing there is a very important issue with gas-assistedinjection molded parts, which is often even more critical than the process itself. It is thedesign of the molded part. In the case of this part there are 4 areas with screw bosses and arib pattern as shown in Figure 6 a) that create a mounting plain. These create one of the pro-duction problems. The rib pattern around the screw boss can result in a significant visiblesink mark on the surface. All those areas are located far away from the melt and gas injec-tion and represent areas which are filled and packed late in the process of producing thepart. The gas-channels need as shown in Figure 6 b) to be at least close enough to pack theareas around the boss areas.

In Figure 7 a) the course of the gas penetrating the melt is shown. The gas flows fromthe gas pin in direction to the high center rib and the gas flow is diverted into 6 differentdirections following the gas-channel network of the part. The gas flow is split up is more


often diverted flowing towards the last filledareas of the part as shown in Figure 7 b). Themore flow paths are provided for the gas, themore problem areas can occur on the moldedpart.

The most critical areas from showing sink marks are the thick ribs running along thelength of the part. Due to the wall thickness and height the gas pressure needs to preventthose areas from sinking in. The rib on the side opposite the boss areas is very critical forpacking. As shown in Figure 8 a) and b) the gas-bubble splits up reaching the thick rib andflows in two opposite directions. Important here is to achieve a stable gas-channel patternwith a gas-bubble entering the root of the rib from both sides (see Figure 8 b) packing outthe melt to avoid a sink mark. This can be achieved by extending the gassing time and pres-sure to allow the gas to compensate the shrinkage in those thick-walled areas.

Figure 6. Screw boss areas with rib pattern for creating a mounting plane.

Figure 7. Gas distribution from the gas injection point following the gas-channel network.


CONCLUDING REMARKS

The application example shows both the diffi-culties of distributing the gas along several dif-ferent paths as well as the ability to pack areasfar away from the gas injection location. Tomake sure that sink mark can be avoided it isnecessary to provide a consistent compact injec-tion molding phase. It also demonstrates thatextending the pin length in those screw bossareas were the key to void sink marks along withan adapted gassing time and pressure. As long asthe gas-channel can penetrate close or into thoseareas, visible sink marks can be avoided asshown in Figure 7 b). The nitrogen gas has theability to pack ribs far away from the gas injec-tion without leaving a visible sink mark on thesurface. Prerequisite to achieve this goal is toprovide the consistency of metering and con-stant conditions in the compact filling phase.Then the gas always follows the same flowpaths and the gas-channel distribution along ribsand mass accumulations is nearly constant.

Based on those results very little adaptations to the gas-channel network were necessary.Basically only one gas-channel needed to be tapered down at the very end to avoid slightsink marks. In other areas a slight chamfer helped to guide the gas in certain locations. Thehigh rib running across the length of the part didn’t show any visible marks on the cavitysurface after optimizing the gas processing parameters. This part is a very good examplethat in a lot of applications the sometimes so-called minor side issues are creating the big-gest production problems and those can be identified by analyzing all the influential factors,which contribute to the production of a gas-assisted injection molded part.

REFERENCES1 Hansen, M. “Anwendungsbeispiele fuer Gas-innendruckformteile,” (Topic: “Application Examples for the Gas-Assisted Injection Molding Process”), seminar “Processing Technology in Injection Molding”, at the SKZ in Wuerzburg,

Germany, June 19972 Hansen, M. “Verfahrenstechnische Grundlagen zur Auslegung von Gasinnendruck-formteilen”, (Topic: “Processing Basics for the Design of Gas-Assisted Injection Molded Parts”), Ph.D. thesis, Shaker Publishers, Aachen, 1996

Figure 8. Gas-channel course close to boss areas and along the rib parallel to main cross rib.


3 Potente, H. “Anwendung des GID-Verfahrens am Hansen, M. Beispiel eines Haltegriffes”, (Topic: Burgdorf, D. Application of the Gas-Assisted Injection Molding Technology exemplary for an Oven Handle”), Plastverarbeiter 46

(1995) 3, p. 40-514 Hansen, M. Application Examples for Gas-Assisted Injection Molded Parts Structural Plastics ‘99, Boston, p. 95-1085 Hansen, M. Application Examples for Gas-Assisted Injection Molded Parts Journal of Injection Molding Technology, Vol. 3, No. 3, p. 141-153.

H. Potente and H.-P. Heim

lnstitut fur Kunststofftechnik, Universitiit-GH Paderborn, Germany

INTRODUCTION

Over the past few years, a large number of variants of the injection moulding process havebeen introduced offering properties optimally tailored to the applications and targets inquestion. The gas injection technique (GIT) is a multi-component process, similar to thesandwich technique, which involves different raw materials being successively injected intothe mould, leading to a layered moulding structure.

By injecting inert gas into the molten plastic, hollow spaces are created in areas of themoulded part with higher wall thicknesses. The use of this process can be motivated eitheron purely process engineering grounds or for design reasons. In many cases, design require-ments are placed on parts which only permit cost-efficient production by the gas injectiontechnique and, at the same time, the process offers a number of technical advantages. Vari-ous drawbacks have to be set against these advantages -especially the considerably moredifficult design of the part and the more complex process control.7

The different variants of the gas injection technique can be distinguished on the basisof the nature of the gas introduction or the type of melt injection. The gas can be introducedeither via the machine nozzle or via a mould injector.8 Both processes have their specificadvantages and drawbacks for the particular application involved.7

To form the hollow spaces, melt is pushed out of the liquid centre of the moulded partand replaced by gas, and hence it is necessary for a similar-sized volume to be available orto be created in the moulded part in order to receive this displaced material. The mould cav-ity is only filled partially to begin with, by the short-shot process, and the residual fillingwith gas conducted in such a way that melt is conveyed into the as yet unfilled volume bythe gas guidance geometry (thick-walled area of the moulded part).5

In the ancillary cavity process, by contrast, the moulded part undergoes complete volu-metric filling and, when the gas is injected in, a connected volume is opened up by themachine control system. The melt is conveyed into this area by the gas guidance geometryas the gas is being introduced.7

Both cases call for highly elaborate balancing of the displaced volume and the volumeto be filled.


The gas injection technique has found a large number of areas of application over thepast few years in particular. The moulded part geometries for which the gas injection tech-nique is used similarly vary over a broad range. It has become established practice to distin-guish between.rod-shaped, thick-walled moulded parts.thin-walled moulded parts with gas guidance ribs and.thin-walled moulded parts with thick points in certain parts.

The actual shape of a moulded part in these categories results from the requirementsplaced on the part design and function, making it necessary for the process employed to beadapted to the properties desired in the moulded part being produced, with consideration todesign guidelines. The cause-and-effect correlations for the gas injection technique are cor-

respondingly complex.

PROBLEM

When it comes to the design optimization of a gas injection moulding, different publicationscontain information on the design of the thick-walled areas of the moulded part in respect ofcross-section and wall thickness ratios as well as in terms of the radii employed. This pro-vides design guidelines, which have already been incorporated in a systematic design proce-dure in the main.7 These guidelines relate to the gas guidance geometries, for which highlydetailed design recommendations are given. They constitute the basis for what is set outbelow and ought to be borne in mind whenever a GIT moulding is being de-signed.

POSITION OF THE MEL T AND GAS INJECTION POINTS

In conjunction with the structural layout of the moulded part, it must be stated that the posi-tion of the melt and gas injection points is of particular importance, in addition to the designguidelines.2,6 The position of the melt gate must be optimized with allowance for the meltflow and coordinated with the position of the gas injection point.l Since the moulding com-pound can only be displaced to-wards the end of the flow path, i.e. towards unfilled areas ofthe moulding, the fosition of the melt gate and the gas injection point determine the courseof the gas bubble. Apart from this, the associated high pressure requirement causes moltenplastic to be displaced from the thick-walled gas guidance geometry to thin-walled areas ofthe moulding, which can lead to instabilities in the process sequence.

In order to completely fill the thin-walled areas, a sufficiently high pressure must beprovided with the gas. The problem here is the stagnation of the melt after its injection,while the gas pressure is being built up, which is inherent to the process.

Figure 1 shows the pressure development at different measurement points in the moul-ded part. A pressure gradient is clearly evident along the flow path. This is only eliminated

Molded Part Design 53

Figure 2. Schematic diagram of an unfavourable combi-nation of gas guidance rib position and residually-filledarea.

Figure I. Pressure development over time in the gasinjection process, from Michaeli, W., Lanvers, A.: Gasin-jektionsverfahren, p. 248.

as the gas bubble advances, giving an isobaric state prevails along the length of the gaschannel.4

Areas of the moulded part that are a long way from the gas injection point first have arelatively low pressure acting on them at the time the gas is introduced. If this is not suffi-cient to displace the melt, the flow front will freeze, producing a gramophone record effecton the surface. the moulded part will only be fully shaped with a high gas pressure. It canbe concluded that this effect will be all the stronger the higher the pressure requirement inthe residually-filled area is.

The gas follows the course of least resistance. The higher the pressure requirement forresidual filling, therefore, the higher the tendency for fingers to form in areas adjacent to thehollow space guidance rib. In extreme cases, the course followed by the gas can deviatecompletely from the gas guidance rib if there are other areas of the moulding that are easierto displace.

The schematic diagram in Figure 2 shows an example of such a case. The position ofthe so-called residually-filled area, i.e. the part of the mould cavity that is not filled withmelt prior to the injection of the gas, is thus of decisive importance.

Prefilling thus constitutes a crucial aspect of moulded part design. In critical cases, justa slight variation in the amount of compound initially injected in can lead to the problemsthat have been indicated.

Two requirements can thus be placed on the layout of GIT mouldings:.the residually-filled area must be located at the end of the desired gas channel.thin-walled areas must be completely filled with melt before the gas is introduced.

Penetration of the gas in

the flat area of the moulding

and blow-through

residually-filled area

gas guidance rib

gas injection


PRESENTATION OF THE PROBLEM TAKING A SAMPLE MOULDED PART

The moulded part shown in Figure 3 was devel-,\rea. ..fthe c".jty not IiIlcd oped for different studies of the gas injection

prior to the .tart of g~rod..o:tjOn ~echnique. Thi~ is a test mold whic.h de~ibera~ely

-/ ~ " mcorporates different process engmeenng diffi-

culties, such as the thick-walled annular region,and also the gas guidance geometry divided intotwo flow directions. In order to produce a low-warpage and low-shrinkage moulded part, thethick-walled areas of the moulding are to be hol-lowed-out by the gas injection technique.

First of all, it is necessary to establish thegas guidance geometry and the required positionof the residually-tilled area(s), so that the meltand gas injection points can be determined. After

different preliminary considerations, it was decided that gas injection should be performedin the annular area and that the gas bubble should propagate around the circular ring and inthe two rod-shaped, thick-walled areas.

It must be borne in mind here that it is, of course, not possible to achieve a ring fullysurrounded by a gas bubble and, hence, apart from the two residually-tilled areas in the frontdomes, an unfilled area must also remain in the ring after partial filling. The target residu-ally-tilled areas are shown in Figure 3.

The requirements for the establishment of one or more appropriate melt injectionpoints are thus:.partial filling with the unfilled areas that are marked in Figure 3.proportional distribution of the volumes of the un-filled areas in accordance with the

volume of melt to be displaced (with different degrees of pre-filling) and.complete filling of the thin-walled area during partial filling.

(;as n..'.'dl"position

Figure 3. Presentation of the residually-filled areas

ESTABLISHMENT OF OPTIMUM MEL T INJECTION POINTSWITH A SAMPLE MOULDED PART

To determine the melt injection points, use was made of finite element calculations to simu-late filling behavior with different gate positions. Different calculations were performedwith melt injection in the central thin-walled area, inter alia. By contrast to injection in theedge area, this position also constitutes an advantage in mould engineering terms. Figure 4shows the calculation results for three different gate positions.

55Molded Part Design

Figure 4. Filling simulation for three different gate positions (marked by the arrow).

mc11 injoction po;n",

00impro,"" artick ~comctty

2Figure 5. Presentation of the improved article geometryand melt injection points.

It is clear that these injection points will notgive satisfactory results. The flow pattern shows Figure 6. Presentation of different degrees of filling forthat the desired division of the residually-filled the optimised article geometry.

areas is not possible on account of the flowdirection dictated by the perforations. The left-hand picture shows an excessively high fill-

ing level in the thick-walled domes. When the injection point is moved towards the centre of

the moulded part, disproportionately high filling of the ring area results.This effect can be attributed to the melt guidance through the bars in the thin-walled

area of the moulding. Since a different gate position cannot be expected to improve the flow

pattern, structural changes were made to the moulded part. Figure 5 shows the changed arti-

cle geometry.As the Figure 5 shows, additional bars were incorporated in the region of the perfora-

tions in order to guide the melt. This produces a clearly improved flow front course which

better meets the requirements on the flow behavior defined at the outset.As Figure 6 shows, virtually symmetric mould filling results in the thin-walled region.

The depiction of different degrees of filling shows that the requirements in respect of theresidually-filled areas can also be fulfilled. An estimate of the unfilled residual volumes for


different degrees of filling shows that the three unfilled areas are more or less proportionalto the displaced volume. This ensures that the prefilling can be aligned to the different gasbubble cross-sections to be expected.

CONCLUSIONS

The example outlined illustrates the basic approach to the establishment of appropriate meltinjection points for the gas injection technique. It becomes clear that even slight changes tothe article geometry and the position of the gate can lead to changes in the course of theflow front, which have an extremely negative impact on application of the gas injectiontechnique. The flow pattern is thus one of the decisive parameters for the process reliabilitythat can be attained with the GIT and for the quality of the moulded parts produced.

Filling simulations employing finite element calculations represent an appropriatemethod for obtaining information on the application of the GIT right at the developmentphase. Through appropriate balancing of the filling behavior and the unfilled areas of thecavity after partial filling, it can be ensured that the conditions for application of the GIT,i.e. the three main requirements on appropriate melt injection points, can be fulfilled.

The results presented deal exclusively with the filling behavior of the moulded parts.The extent to which other results of an FE simulation, such as pressure and temperature dis-tribution, can be used and interpreted specifically for the GIT is to be checked in further

investigations.

REFERENCES

23

4

5

6

Eckardt, Helmut: Gas-Assisted Injection Molding, in: Stevenson, J. F.: Innovation in Polymer Processing: Molding,Carl Hanser Verlag, Munich Vienna New York, 1996.Jaroschek, Christoph: Elegant? Preiswert? Oder sogar beides?, Kunststoffe 87 (1997) 9, pp. 1172-1176.K1otz, B.: Voraussetzungen im Bereich der Formteilgestaltung fUr die Anwendung des Gasinnendruckver-fahrens,Transferzentrum Aachen Kunststofftechnik, SpritzgieBtechnisches Kolloquium 1990, pp. 36-57.Michaeli, W.; Lanvers, A.: SpritzgieBen transparent gernacht -Neue Entwick1ungen bei der ProzeBsimulation, Teil1:CAE- Techniken fUr das Zweikomponen-tenspritzgieBen und das Gasinjektionsverfahren, Plaste und Kautschuk39 (1992) 7, pp. 241-248.Moritzer, Elmar: Ph1lnomenorientierte ProzeB-und Formteiloptimierung von thermoplastischen Gasinjektions-( GIT)-SpritzgieBartikeln, Dissertation an der Universitlit-GH Paderborn, Shaker Verlag, Aachen 1997.Renger, M.: Das Gasinnendruckverfahren -eine SpritzgieBvariante mit besonderen Mt\glichkeiten, SUddeutschesKunststoff-Zentrum WUrzburg, Fachtagung 18.-19. September 1990, pp. 101-136.Rennefeld, Christoph: Konstruktive Optimierung von Thermoplastformtei1en und SpritzgieBwerkzeugen fUr dieGasinnendrucktechnik, Dissertation an der Universitlit GH Paderborn, Shaker Ver1ag, Aachen, 1996.Shah, Suresh: Gas Injection Molding: Current Practices, ANTEC '91, pp. 1494-1506.

7

8

1

Design Optimization of Gas Channels for an AirCleaner Assembly Using CAE Simulations

D.M. Gao, A. Garcia-Rejon, G. Salloum and D. BaylisIndustrial Materials Institute - National Research Council Canada, 75, blvd. de Mortagne,

Boucherville, Quebec J4B 6Y4, Canada

INTRODUCTION

The typical gas-assisted injection moulding process (short shot process) can be subdividedinto the following steps: a) polymer filling to a predetermined percentage cavity filling b)gas injection; and c) packing stage. During polymer filling the cavity is partly filled (up to80%). Shortly after the end of the polymer injection, the gas is injected to hollow-out thegas channels until the cavity is completely filled. The relative melt/gas flowrate and theswitch-over time between polymer and gas injections will determine the amount to be hol-lowed. Due to the gas penetration inside the polymer melt during the gas injection phase,the amount of material needed, the level of injection pressure as well as the clamp tonnagerequired, the resulting shrinkage/warpage and sink marks on the part can be greatly reduced.

Due to its versatility in the production of mouldings of much greater complexity whichcombine thick and thin walls, hollow sections and elongated shapes, gas assisted injectionmoulding has resulted in the production of single parts replacing multipart assemblies andtherefore a substantial reduction in manufacturing costs. Designers now have much greaterfreedom to incorporate thick and thin sections in the same moulding that will result in ribsand flow leaders; higher stiffness to weight ratios; reduced cycle times; and higher dimen-sional stability.

The air cleaner assembly, analyzed in this project, is a good example of an open-chan-nel part. These parts have a thin wall with the gas channels traversing the part similar toconventional ribs. These parts are more difficult to design and mould because the gas maypenetrate into the thin walled sections of the part (fingering).

The optimal layout of the gas channels - relative location of gas channels and gas injec-tion points to polymer gates - within a cavity should create a polymer filling pattern inwhich the lowest pressure will be located near the end of the channels. In order to avoid fin-gering the channels should be oriented in the direction of the melt flow. Channel size andgeometry have to be chosen in such a way to minimize race track effects while maintaining


the structural advantages offered by gas assisted injection moulding. In the case of multiplegas channels it is also very important to avoid an unbalanced gas penetration in the differentchannels.

The use of computer aided flow analysis in the case of gas assisted injection mouldingcan be of great help to the part designer, mould maker and processor in the determination ofparameters such as: i) gas channel design; ii) % of polymer to be injected and its optimalinjection point; iii) gas injection locations; iv) preset volumes and pressures for gas injec-tion; and v) filling patterns and operating conditions for optimal polymer wall thickness;best quality surface and minimal shrinkage and warpage.

The primary objectives of this project are to determine, through computer aided flowanalysis, the gas channel design (size and location) as well as the optimal moldability dia-gram (polymer injection speed, flow rate, temperature range, gas pressure, etc.) for an aircleaner assembly moulded using a 30% glass fibre reinforced polypropylene.

In order to provide a useful information to the part designer, mould maker and moulder,several gas channel designs as well as different moulding conditions are considered. A num-ber of important factors such as polymer flow pattern, gas penetration, tooling constraints,etc. are taken into consideration in the optimization of the moulding conditions.

NUMERICAL MODEL

During the gas penetration into the polymer melt, three distinct flow regions during the gasinjection stage can be identified: 1) gas penetration region, 2) polymer melt region, and 3)empty or unfilled region.

Region 1 is initially filled by the polymer. The gas penetrates into the polymer melt andcreates a gas core. During the gas penetration, the gas is transmitting the pressure requiredto advance the melt front. The polymer skin layer between the gas and the mould walls isstagnant. It is assumed that the skin layer consists of a solidified layer and an adhered layer.The solidified layer is formed by the polymer freezing upon contact with the cold mould.The adherence between the polymer flow and the solidified layer creates the adhered layer.Regions 2 and 3 are identical to those encountered in conventional injection mould fillingexcept that two moving boundaries for the polymer melt region are present.

In this study, the polymer melt is considered as a Generalized Newtonian Fluid, i.e. theviscosity is a function of shear rate and temperature. The flow is assumed to be quasi-steadystate and the inertia terms are neglected due to the low Reynolds numbers encountered inmolten polymer flow. Since most parts produced by gas-assisted injection moulding have ashell like geometry, i.e. the part thickness is much smaller than other part dimensions, thelubrication approximation (Hele-Shaw flow)1 can be used for modelling the global flowbehavior in the mould cavity. A dimensional analysis of the energy equation shows that the

Design Optimization 59

heat conduction in the flow direction can be neglected since the thickness of the cavity ismuch smaller than the other two dimensions. The convection in the gapwise direction is alsoneglected.

The pressure equation is solved using the Galerkin finite element method. A three nodetriangular element was chosen to approximate the pressure. The energy equation is dis-cretized using the finite difference method. The time dependent derivative of the tempera-ture is approximated by backward finite difference. A control volume approach has beenemployed to track both, the flow front advancement as well as the gas-polymer interface. Athickness fraction of polymer skin (Fs) is associated to each control volume in order to rep-resent the three distinct regions present during the filling phase. Fs is defined as the ratio ofthe thickness of the polymer skin to the total thickness of the part. Fs = 1 represents an ele-ment completely filled with polymer, while Fs = 0 represents an empty element. 0< Fs <1represents an element where the gas has penetrated through and therefore the polymer layerhas become a skin layer. Details concerning the simulation model development and itsnumerical implementation are given elsewhere.2-4

SIMULATION RESULTS

GEOMETRY

Figure 1 shows the geometry of the air cleanerassembly. Two gas channel network designs arepresented here in order to illustrate the designoptimization procedure. In both cases square gaschannels having a 10x10 mm cross section wereused.

Even though the gas channel size is gener-ally small compared to the overall part geome-try, as the flow velocity is mucher high insidethe gas channels than in the thinner sections, alocal fine mesh is needed to provide an adequaterepresentation of the gas penetration which is theprimary concern in gas channel design.

MATERIAL PROPERTIES

The material used for the moulding trials is a30% glass reinforced polypropylene (Thermofil-

P6-30FG-0153). The Carreau-WLF model was used to describe the melt viscosity as func-tion of the shear rate and temperature.

Figure 1. Geometry of air cleaner assembly.


The coefficients for the Carreau-WLF model are given in Table 1.

RESULTS FOR DESIGN 1

The first gas channel design comprises three primary independent gas channels in whichtwo of them have transversal branches touching the part edges (See Figure 2). This designallows maximum gas penetration. The gas penetration in all gas channels generates a uni-form pressure over the entire part, therefore reducing the cooling time and the parts’swarpage. A centre polymer gate is considered to provide a balanced filling prior to the gaspenetration. Three gas injection points were used to core the three primary gas channels.During the actual moulding this design will require a three gas cylinder machine. The loca-tion of the polymer gate and gas injection points are shown in Figure 2. Figure 3 shows thepolymer flow front advancement before the gas injection. The grey scale represents thefilled domain as a function of time. For this gas channel network, the polymer filled regionis stretched in the same direction as the orientation of the primary gas channels. This is dueto the fact that the polymer follows the path of least resistance, e.g. inside the gas channels.

Table 1. Carreau model coefficients (P6-30FG-0153)

Zero shear rate viscosity, η0, Pa.s 1.1213 E+04

Reciprocal shear rate, λ, 1/s 6.4200 E-01

Exponent, n 6.8750 E-01

Reference temperature, Tref, oC 2.0035 E+02

Standard temperature, Ts , oC 4.9050 E+01

No flow temperature, oC 173

Melt density, g/cm3 0.84

Thermal conductivity, kcal/mhC 0.211

Specific heat, J/gC 1.49

ηη0At

1 λAtγ·+( )

n----------------------------=

Atlog8.86 Tref Ts–( )

101.6 Tref Ts–+( )---------------------------------------------

8.86 T Ts–( )101.6 T Ts–+( )

---------------------------------------–=


After the polymer injection, gas was simulta-neously injected from the three gas injectionpoints. The final gas penetration is represented inFigure 4. The last filled region was located nearthe exhaust pipe (top left of the Figure). Sincethe gas channel 3 (at the right of Figure 4) islocated across the part, the gas penetrated theentire gas channel and then leaked out. Due tothe strongly stretched polymer prefill pattern (seeFigure 3), the gas penetration inside the centergas channel was blocked by the stagnant polymermeeting the part edges. Therefore the centre gaschannel has no practical use in this design.

RESULTS FOR DESIGN 2

Figure 5 shows the second design studied in thisproject. The centre gas channel has been

removed and three transverse gas channels have been added to connect the two previouslyindependent gas channels and to increase the part rigidity. A centre polymer gate is used andthe gas will be injected through two gas injection points located at the low end of the twoprimary gas channels (see Figure 5). The final gas penetration is shown in Figure 6. Accord-

Figure 2. Polymer and gas injection locations for Design 1.

Figure 3. Polymer prefill filling pattern for Design 1.

Figure 4. Final gas penetration for Design 1.

Polymer Gate

Gas injection points

Gas channels

Gas channel 3

Gas channel 1

Last filled area

Gas channel 2

Gas finguering


ing to the resulting filling pattern (not presented here), the gas advanced at almost the samespeed in both gas channels, therefore ensuring an even penetration. With this modifieddesign the two primary gas channels are entirely hollowed. The transverse branches are alsohollowed by the gas except for the centre portions which are being further investigated. Asexpected, the pressure distribution is quite uniform due to the uniform gas penetration (seeFigure 7). Figure 8 shows the temperature field. It can be seen that the region around thecentre polymer gate has cooled down faster. The polymer skin contained inside the gaschannels, being hotter, will cool down during the gas packing phase.

Figure 5. Polymer and gas injection locations for Design 2. Figure 6. Final gas penetration for Design 2.

Figure 7. Pressure distribution for Design 2 (unit: psi). Figure 8. Temperature distribution for Design 2 (unit: oC)


CONCLUSIONS

In this work the gas channel design optimization of an air cleaner assembly was conductedusing a numerical model for gas assisted injection moulding developed at NRC's IndustrialMaterial Institute. Two gas channel designs are presented as well as the simulation resultson gas penetration, pressure distribution and temperature field. The CAE simulation tech-niques have shown their effectiveness when used for the optimization of this industrial part.

ACKNOWLEDGEMENTS

The authors would like to thank Siemens Electric Ltd for allowing to present the results ofthis project.

REFERENCES1 C.A. Hieber and S.F. Shen, J. Non-Newtonian Fluid Mechanics, 7, (1980). 2 D.M. Gao, K.T. Nguyen, A. Garcia-Rejon and G. Salloum, Proceedings SPE ANTEC’96, 638-643, Indianapolis (1996). 3 D.M. Gao, K.T. Nguyen and G. Salloum, NUMIFORM'95, Eds. Shan-Fu Shen & Paul Dawson, 1125-1130, Ithaca, NY (1995). 4 K.T. Nguyen and D.M. Gao, ASME Annual Meeting, MD-Vol. 49/HTD-Vol. 283, 89-103, Chicago (1994).

The Occurrence of Fiber Exposure in Gas AssistInjection Molded Nylon Composites

Shih-Jung Liu and Jer-Haur ChangPolymer Rheology and Processing Lab, Mechanical Engineering, Chang Gung University,

Tao-Yuan 333, Taiwan, R.O.C.

INTRODUCTION

Gas assist injection molding1,2 has received extensive attention in recent years, due to itsflexibility in the design and manufacture of plastic parts. In this process, the mold cavity ispartially filled with the polymer melt followed by the injection of inert gas into the core ofthe polymer melt. Gas assist injection molding can produce parts incorporating both thickand thin sections with less shrinkage and warpage and better surface finish. It requires lowerclamping force than the conventional injection molding process.3-5

Today, short glass-fiber reinforced thermoplastic composites have the fastest rate ofdevelopment and represent 30% of the composite market. Gas assist injection molding ofglass-fiber reinforced composites has the capability of producing parts having thick and thinsections with a good structured rigidity. However there are still some unsolved problemsthat confound the overall success of this technology. Parts roughness occurring on the sur-face of molded composites caused by inappropriate mold design and processing condition isone of them.

To the best knowledge of the authors, no research paper has ever discussed the above-mentioned problem. The purpose of this report was to study the surface roughness phenom-enon occurring in the gas assist injection molded composite parts. The materials used wereshort glass-fiber filled Nylon-6 composites. Experiments were carried out on an 80-toninjection molding machine equipped with a high pressure nitrogen-gas injection unit. Two“float-shape” axisymmetric cavities were used. Various processing variables were studied interms of their influence on parts surface quality. The final goal of this research is to explainthe mechanisms of rough surface formation, so that steps can be taken to improve the sur-face quality of molded composites.


EXPERIMENTAL

The resins used in this study were commercially available grade 15% and 35% E-glass-fiber filled Nylon-6 composites (Ginar Chem., Taiwan).6 The fiber in the composites has adiameter of 10 m and an approximate aspect ratio of 10, as measured by the resin sup-plier.6 Table 1 lists the characteristics of the composite materials. Gas assist injection mold-ing experiments were conducted on an 80-ton injection molding machine (Taichung Mach.VS-80, Taiwan). A high-pressure nitrogen gas injection unit (Gas Injection model PPC-1000, U.K.) was attached to the machine.6

Two “float-shape”cavities were used inthis study. Figure 1shows the dimensions ofthese cavities. One ofthem has a diameter d(=9 mm) at the ends and

a diameter 2d at the center portion, i.e. d-2d-d, while theother has the same geometry except a size variation of d-4d-d. The mold has two movable plugs in the runners. By turn-ing the plug, the way the flow enters each cavity can be con-

Table 1. The processing variables as well as the values used in the experi-ments.

Processing parameters

a b c d e f

Melt temperature,

oC

Mold temperature,

oC

Melt filling speed,%

Short-shot size,

mm

Gas pressure,

bar

Gas injection delay time, s

1 280 65 55 42 50 5.0

2 285 80 65 43 60 6.0

3 290 95 75 44 70 7.0

4 295 110 85 45 80 8.0

5 300 125 95 46 90 9.0

µ

Figure 1. Layout and dimensions of mold cavity.

Figure 2. Schematically the distribution of parts roughness and the positioning of the measuring probe.

The Occurrence of Fiber Exposure 67

trolled. Only one cavity was used for the experiments each time. The temperature of themold was regulated by an oil-circulating mold temperature control unit.

After molding, the surface roughness of moldedcomposites was measured. A roughness meter (Hommel-werke model T2000, Germany) with a measurable rangeof ±200 m was used. Eight lines equally spaced acrossthe surface length of a cylinder (see Figure 2) wereselected for evaluation. The length of each line was 8mm, which was the longest measurable distance of theroughness meter at a time. Both the maximum individualpeak-to-peak height Rmax and the arithmetic averageroughness Ra were used for parts evaluation. Rmax isdefined as the maximum peak-to-valley dimensionobtained from the five sampling length le within the eval-uation length lm (ISO 4287/1).

The arithmetic average value Ra of filtered roughness profile is defined by the follow-ing equation (ISO 4287/1):

[1]

which was determined from deviation about the center line within the evaluation length lm.The measured eight roughness Rmax and Ra were integrated to obtain the final surfaceroughness values.

Various processing variables were studied in terms of their influence on parts surfacequality: melt temperature, mold temperature, melt filling speed, short-shot size, gas pres-sure, and gas injection delay time. Table 1 lists these processing variables as well as the val-ues used in the experiments.

RESULTS

Gas assist injection molding experiments were conducted with an 80-ton injection moldingmachine equipped with a high-pressure nitrogen gas injection unit.6 All molded compositesexhibited severe surface roughness, beginning from the divergent portion to beyond the cen-ter areas of the parts. The d-2d-d parts showed a spiral roughness distribution on the surface.The d-4d-d parts exhibited a wider and more severe distribution of roughness on the surface.Figure 2 shows schematically the rough surface distribution of the parts. Since the moldedd-4d-d part exhibited a greater severity of the surface roughness phenomenon, it was usedfor the subsequent investigations.

µ

Ra1lm----- y xd

0

lm

∫=


Various processing variables were studied in terms of their influence on surface rough-ness of the molded parts. Table 1 lists these processing variables as well as the values usedin the experiments. To mold the parts, one arbitrary processing condition was chosen as areference �Table 1). After molding, a roughness meter was used to measure the surface qual-ity of the parts. A total of eight lines (8mm long for each line) equally spaced across the sur-face length of a cylinder (Figure 2) were selected for evaluation. By changing one of theparameters in each test, we were able to understand the effect of every factor on the surfaceroughness of gas assist injection molded composites.

CONTENT OF GLASS FIBER

The effect of the content of glass fiber on the parts surface roughness was studied. VirginNylon, 15% and 35% glass fiber reinforced Nylon composites were used to mold the parts.It was found that the 35% filled composites exhibited the most severe rough surfaces, whilethe virgin Nylon exhibited none of the surface roughness phenomenon. By setting one arbi-trary set of processing conditions in Table 1 (the shaded one) as an example, the measuredroughness values (Ra) were 1.6, 3.8 and 8.5 respectively for the molded virgin Nylon, 15%and 35% glass fiber filled Nylon parts. Here the 35% filled Nylon material exhibited themost serious surface roughness; it was therefore selected for the subsequent experimentalinvestigations

MELT TEMPERATURE AND MOLD TEMPERATURE

The effect of melt temperature on parts surface roughness was investigated. Thirty five per-cent glass fiber reinforced Nylon was used for the experiments. Five temperatures of thepolymer melt were set in the gas assist injection molding experiments: 280, 285, 290, 295and 300oC. The measured results in Figure 3 shows that part surface roughness decreased

Figure 3. Effect of melt temperature on the surface rough-ness of gas assist injection molded composites.

Figure 4. Effect of mold temperature on the surface rough-ness of gas assist injection molded composites.


with melt temperature. The effect of mold temperature was also studied. Five different moldtemperatures were selected, from 65 to 125oC (Table 1). The experimental result in Figure 4shows that surface roughness of gas assist injection molded Nylon composites decreasedwith mold temperature.

SHORT-SHOT SIZE

The gas assist injection molded composites in the experiments were subjected to differentshort-shot size melt fillings (screw displacement ranges from 42 to 46 mm). The measuredresult does not show an obvious effect of short-shot size on the surface roughness of gasassist injection molded composites.

MELT FILLING SPEED

The effect of melt filling speed on the surface quality of molded composites was also stud-ied. Different filling speeds, from 55 to 95% of the maximum available speed of the injec-tion molding machine, were selected. The experimental result shows that the surfaceroughness of the parts generally decreased with increased melt filling speed.

GAS INJECTION PRESSURE AND GAS INJECTION DELAY TIME

Parts were molded with different gasinjection pressures. Five different gaspressures (50 to 90 bars) were selected tomold the composites. The result suggeststhat one can decrease the surface rough-ness by increasing the gas injection pres-sure. The effect of gas injection delaytime on the surface roughness was alsostudied. The gas delay times selected forthe experiments were between 5 to 9 sec-onds. Parts were not moldable for the gasdelay time shorter than 5 seconds in thisstudy. The measured surface roughness ofthe molded parts in The result suggeststhat one can improve the surface qualityof the parts by shortening the gas injectiondelay time.

7DEOH � lists the effect of various pro-cessing parameters on the surface roughness of gas assist injection molded composite parts,based on the above experimental results.

Table 2. Effect of processing parameterson the surface roughness of gas assistinjection molded composite parts

Processing parameterSurface

roughness

Content of glass fiber (�) (�)

Melt temperature (�) (�)

Mold temperature (�) (�)

Melt filling speed (�) (�)

Gas pressure (�) (�)

Gas injection delay time (�) (�)


DISCUSSION

A series of short-shot experiments of the polymer filling have been completed. As a poly-mer melt is injected from a small size gate or runner into a large size cavity, the phenome-non of “jetting” or irregular flow can be observed.6 The melt emerging from the gate formsa jet that rapidly advances until it is stopped by the mold wall. Regular flow forward fillingcommences subsequently. It has been experimentally observed6 that jetting can occur when-ever the dimension of the fluid stream is smaller than the smallest dimension in the planeperpendicular to the flow. It is thus related both to the gate size and to the degree of extru-date swelling of the melt, rather than to the level of that axial momentum. Filled polymers,which swell less than unfilled melts, exhibit jetting at lower filling rates.6 In this study, bothd-2d-d and d-4d-d parts exhibited jetting flows during filling. Parts molded by the d-4d-dcavity exhibited more severe jetting and irregular flow than those molded by the d-2d-d cav-ity. This might be due to the fact that the d-4d-d cavity has geometry of larger size-variationfrom the end to the center.

In this study, jetting and irregular flows6 occur in the molded composites. The polymermelt emerging from one end of the mold forms a jet that rapidly advances until it is stoppedby the mold wall. As soon as the jetted melt contacts the mold wall it tends to cool. Thepolymer shrinks much more than the glass fiber6 and leaves the fiber exposed to the partssurface. Parts roughness may thus form at the jetting and irregular flow filled surfaces. Reg-ular flow forward filling commences subsequently after the jetting flow filling.6 As the meltfilling stops, the high-pressure gas is injected and pushes the composite against the moldwall.1 This minimizes the shrinkage of the materials and the consequent parts roughnessexcept at the jet filled areas, which have been cooled down and left the fibers exposed to theparts surface. In the experiments, the d-4d-d parts were found to have more serious jet fill-ing than the d-2d-d parts. This might explain why the d-4d-d parts exhibited more severesurface roughness.

For the factors selected in the experiments, a higher melt temperature was found toimprove the surface quality of the molded composites (Figure 3). Increasing the melt tem-perature keeps the material hot for a longer time for the gas pressure packing. This will min-imize the shrinkage of the polymer3 as well as the surface roughness.

As soon as the melt begins to enter the cavity it starts to cool. In order to minimize thesurface roughness, the temperature of the melt must remain high enough for a period suffi-cient for the gas to pack the composite.6 This is aided by having a.high mold temperature sothat the melt will not cool too rapidly (Figure 4).

In gas assist injection molding, the gas pressure acts as a holding pressure just as in thatof conventional injection molding.2 One can decrease the surface roughness by increasingthe gas pressure. It is mainly due to the fact that increasing the gas pressure decreases the


shrinkage of the Nylon matrix1 as well as the possible exposure of the glass fibers. This willmake the final part surface smoother.

Increasing the gas injection delay time increases the cooling time of the polymer melt.It becomes more difficult for the composites to be packed by the subsequent Nitrogen gas.Filling the mold cavity as rapidly as possible should minimize the surface roughness andobtain parts with the best surface quality.

CONCLUSIONS

This report has examined the effect of different processing factors on surface roughness ofgas assist injection molded composites. The following conclusions can be drawn based onthe current study.• The occurring surface roughness mainly resulted from the exposure of glass fiber on the

surface of gas assist injection molded Nylon composites.• When designing a part, one should avoid the occurrence of jetting or irregular flows

(e.g., by adopting a smaller size-variation geometry design) during filling to eliminatethe surface roughness phenomenon.

• One can also improve the surface quality of a gas assist injection molded compositesby: increasing the melt and mold temperature, increasing the melt filling speed, increas-ing the gas pressures, or shortening gas injection delay time.In this study, the mechanism of surface roughness formation has been explained and

steps can thus be taken to ensure that the roughness can be minimized. This provides signif-icant advantages in improving product quality of gas assist injection molded composites.

ACKNOWLEDGEMENT

The authors would like to express their gratitude to the National Science Council of Taiwan,R.O.C. for their funding support under the grant NSC89-2216-E-182-005.

REFERENCES1 S. Shah, SPE-ANTEC Tech. Paper, 37, 1494(1991).2 L. S. Turng, SPE-ANTEC Tech. Paper, 39, 74(1993).3 S.C. Chen, K.F. Hsu and K.S. Hsu, Num. Heat Trans., 28, 121(1995).4 S.H. Parng and S.Y. Yang, Inter. Polym. Proc., 13, 318(1998).5 S.C. Chen, N.T. Cheng and S.Y Hu, J. Appl. Polym. Sci., 67, 1553(1998).6 S.J. Liu and J.H. Chang, Polym. Comp., (in press) (1999).

Saving Costs and Time by Means of Gas-assistedPowder Injection Molding

Christian Hopmann, Walter MichaeliInstitute of Plastics Processing (IKV), Aachen, Germany

INTRODUCTION

Powder injection molding is a near net-shapemanufacturing technique for complex metal orceramic components. The first step in the manu-facturing process (Figure 1) is the careful choiceof the ingredients of the compound that in generalconsists of a polymeric binder system and themetal or ceramic powder. Binder system is a mix-ture of all additives and auxiliaries necessary toallow the feedstock being processed by an injec-tion molding machine. It may contain plasticiz-ers, lubricants, mold release agents, wettingagents and other. The binder system has to fulfilthree tasks. First, it has to provide a sufficient

flowability to enable the processing by injection molding. Afterwards the binder has to pro-vide mechanical stability to the green part to allow for handling. Finally the binder has to beremovable from the part without any residue. The feedstock typically has a volume loadingof 60%.

Following the compounding the second step is the injection molding of the feedstock.The machine equipment and processing is very similar to conventional injection molding ofthermoplastics. Due to the high powder content the protection against wear of the barrel,nozzle and mold has to be improved. Regarding the process there are some peculiarities pri-marily concerning the course of cavity pressure over time. This is because of the rheologicalbehavior of the feedstock as discussed below.

Injection molding was conducted on a machine with a clamping force of 600 kN and athree section screw with a diameter of 25 mm and a L/D-ratio of 20. The injection moldingstage has a considerable influence on the dimensional stability of the moldings. The aim is

Figure 1. Powder injection molding process.

Sintering

Compounding Injection Moulding

Debinding


to achieve a homogeneous density distribution, since any density differences will immedi-ately result in inhomogeneous shrinkage and warpage.

Finally the binder is removed from the molding and the part is sintered to its final den-sity. There are several possibilities to extract the binder from the molding i.e. thermal treat-ment and catalytic debinding. However binder removal and sintering may take several hoursup to some days.

CHARACTERIZATIO N OF THE FEEDSTOCK

In order to process ceramic feedstock it isvery important to consider its specific rheo-logical and thermal properties.1,2 As to thehigh filler content of about 60% there is a significant reduction in the flowability of thefeedstock which should be met by adding additives of low viscosity to the binder.

Figure 2 shows a comparison of the viscosity of a ceramic feedstock and a polypropy-lene. The binder bases on a wax/PP-system. It can be shown that the feedstock's viscosity isslightly higher than that of the polymer but it is within the processing range of commonthermoplastics.

Besides the flowability of the feedstock, thermal properties are influenced by the highpowder loading (Table 1). Regarding the thermal material properties density, thermal con-ductivity and heat capacity or rather the thermal diffusivity derived from these three proper-ties, it is obvious that it is seven times higher for ceramic feedstock than for polypropylene.

Because thermal diffusivity characterizes the velocity of temperature adjustmentwithin the injection molding tool it is proved with regards to processing that the feedstocksolidifies very fast. The effective holding pressure time is substantially reduced whichdemands for a higher filling and holding pressure in order to produce moldings of high qual-ity.

Figure 2. Flowability of CIM-feedstock.

Table 1. Thermal properties of CIM-feedstock in comparison to purepolypropylene

PP CIM

density ρ, g/cm3 0.9 2

heat capacity cp, kJ/kg 2 1

thermal conductivity λ, W/mK 0.2 1.4

thermal diffusivity a, mm2/s 0.1 0.7

Vis

coci

ty [

Pas

]

10

10

10

1010101010 4

4

3

3

2

2

1

1

Shear rate [1/s]

PPCeramic-feedstock

υm,PP =235 C

υm,CIM =135 C

Saving Cost and Time 75

Furthermore, the outer layers in the flow channel are considerably thicker than thoseoccurring in the processing of pure thermoplastics and a rapid sealing of the gate can beexpected. Due to a complete filling of the cavity the injection speed has to be on a level thatis sufficiently high. On the other hand an excessive injection speed results in jetting of mate-rial and separation effects of powder from binder having negative effect on part's quality.Altogether the relevant parameters of the powder injection molding process have to remainin a fairly narrow window.3

WHY GAS-ASSISTED INJECTION MOLDING?

Using gas-assisted injection molding, which leads to inner hollow sections, the consumptionof raw material can be cut while the stiffness of the part remains on a high level. It is possi-ble as well to increase part's volume while keeping the consumption of material constant.For these reasons gas-assisted injection molding results in a more cost effective productionof plastics. Altogether the use of gas-assisted injection molding in powder injection moldingprovides the same advantages as in the processing of thermoplastics.4

Moreover the transfer of gas-assisted injection molding to powder injection moldinghas some additional advantages:

1. Due to the increased surface and the lower material consumption besides the savingof cycle time a considerable saving for debinding and sintering can be expected. This leadsto an improvement of cost effectiveness because these periods are clearly higher than thecycle time of the injection molding process.

2. During injection molding of compact parts molecular orientations and internalstresses are induced by flow effects for the entire holding pressure phase in order to com-pensate volume shrinkage. Preferably they can be found near the gate. These orientationsand internal stresses may relax during debinding and result in irreversible defect of themolding. Using gas-assisted injection molding, orientations caused by holding pressurephase are eliminated due to the constant pressure level over flow length. In this phase massflow is reduced to very low level so that resulting moldings have a better dimensional stabil-ity and are low-stress and low-warpage in comparison to compact parts. This has not only afavorable effect on the quality of the moldings but also on the error rate.

3. Due to the gas pressure on constant level moldings with higher differences in wallthickness are producible. Internal stresses, which may cause the damage of the part duringthe following process steps, are avoided to a great extent.

In addition to these advantages the use of gas-assisted injection molding in powderinjection molding offers a new range of application as well as an increased degree of inte-gration. Nevertheless, up to now the use of gas-assisted injection molding in powder injec-tion molding has not been a matter of research. There are only a few references that certify


the feasibility of this technique.5,6 For this rea-son IKV is about to gain systematic knowledgeabout the use of gas-assisted injection moldingin powder injection molding.

OBJECTIVE S

Within the scope of research at IKV the combination of gas-assisted injection molding and powder injection molding willbe realized. This will include the analysis of all process stepsfrom compounding to the ceramic part. This enables thedetection of all relevant process parameters and their conse-quences on the quality of the part.

It is essential to focus the results on the properties of theceramic product and not only to consider the quality of thegreen part. The relevant quality features are the error rate andthe resulting wall thickness as well as their distribution overthe flow path of melt.

When using gas-assisted injection molding in powderinjection molding the very poor elastic properties of the melt

are a main problem (Figure 3). This behavior favours jetting of material which results in apoor reproducibility of cavity filling particularly when producing moldings with high cross-sections. This point has to be met by solutions relating to the mold and relating to the pro-cess engineering.

Beyond the elastic properties elongational viscosity and shear viscosity play an impor-tant role with respect to the processability of the feedstock. Considering the high shear vis-cosity the acceleration of the gas bubble caused by a decreasing pressure difference betweenthe flow front of the feedstock and the gas bubble (Figure 4) leads to a crack initiation andpropagation. On the other hand the elongational viscosity hinders the frontal flow at theflow front and brings about the risk of defects on the flow front.

INV ESTIGATIONS

First, some fundamental investigations about the feasibility of "gas-assisted powder injec-tion molding (GAPIM)" have been conducted. A consisting injection molding tool,equipped for gas-assisted injection molding but not adapted to ceramic feedstock, has beenused in order to gain some basic information about the behavior or ceramic feedstock duringthis process. This information has been very important for a successful design of an opti-mized mold.

Figure 3. Defects caused by jetting of material.

Figure 4. Acceleration of the gas bub-ble.

flow front of the gas flow front of the melt

solidifiedouter layer

Saving Cost and Time 77

Furthermore a new injection molding toolhas been designed and manufactured with spe-cial respect to gas-assisted powder injectionmolding. The cavity shows a molding that isvery similar to a screw and has different diame-ters (Figure 5). It enables getting informationabout.possible size of molding and critical areaslike changing diameters. In order to analyze dif-ferent variations of gas-assisted injection mold-ing in powder injection molding the use of anoverspill cavity is possible.

Following the fixing of a suitable operating parameter set systematical analyses aboutthe connection between process parameters and molding's properties had to be found. Therelevant process parameters are varied close to the operating parameter set and the effect onpart’s properties is investigated. The diameter of the gas bubble respectively the resultingwall thickness of the molding, the length of the gas bubble and the distribution of the gasbubble over the flow length are considered as the most relevant quality characteristics.Investigations about possible defects like cracks or voids complement the research work.

CONCLUSIONS

The feasibility of combination of gas-assisted injectionmolding and powder injection molding has been foundout within a first step. Using an existing injection moldparts could be produced (Figure 6). However, due to thefact that this mold has not been adapted to the proper-ties of ceramic feedstock demolding was unsuccessful

again and again and a lot of moldings have beendestroyed.

Nevertheless it has been proved that it is possible to combine gas-assisted injectionmolding and powder injection molding using the selected feedstock.

Regarding the first results it is predictable that resulting wall thickness will be very thinin comparison to gas-assisted injection molding of unfilled thermoplastics. Due to the highelongational viscosity the feedstock used preferably tends to a solid flow profile over flowchannel height. This leads to a displacement of the layer near the wall by the gas-bubble.

Figure 5. Design of the GAPIM-Tool.

Figure 6. Sample (detail).

gate

gas nozzle

overspill cavity(optional)

gas bubble

melt


REFERENCES1 Kurzbeck, S., Kaschta, J., Münstedt, H.: Rheological behavior of a filled wax system, Rheologica Acta 35 (1996) 5,

S. 446-4572 Münker, M.: Untersuchung zur Spritzgießverarbeitung keramikpulvergefüllter Formmassen und rechnerische Simulation des Formteilbildungsprozesses, unveröffentlichte Diplomarbeit am IKV, RWTH Aachen, 19953 Hopmann, C., Knothe, J.: Verarbeitung von ultrahoch-gefüllten organischen Silizium/Siliziumnitrid-Dispersionen nach dem Pulverspritzgießverfahren, Abschlußbericht zum BMBF-Verbundprojekt 03 N 3000 C, IKV Aachen, 19974 Lanvers, P.: Analyse und Simulation des Kunststoff-Formteilbildungsprozesses bei der Gasinjektionstechnik (GIT),

Dissertation an der RWTH Aachen, 19925 N.N.: Powder injection offers new choice for molders, Modern Plastics (1997) 12, S. 146 N.N.: Spritzguß mit Hohlräumen, DKG 75 (1998) 6, S. 77 Pohl, T.: Einsatz der Gasinjektionstechnik beim Pulverspritzgiessen, unpublished thesis, IKV, RWTH Aachen, released soon, supervisor: C. Hopmann.

Gas-assisted Reaction Injection Molding (GRIM):Application of the Gas Injection Technology

to the Manufacturing of Hollow Polyurethane Parts

I. Kleba, E. HaberstrohInstitut für Kunststoffverarbeitung (IKV), Pontstraße 49, D-52062 Aachen, Germany

INTRODUCTION

Gas-assisted Injection Molding (GIM) is one of the most promising special injection mold-ing technologies for processing thermoplastics and continues to gain market shares.Between 1997 and 1998 a growth in use of 10% by North American molders has beenstated.1 Once limited to thick-walled moldings, nowadays this processing technology isapplied to manufacture a broad range of plastic products including complex and thin-walledparts. As its main advantages raw material and weight reduction, shorter cycle times, mini-mization of warp phenomena as well as compensation of sink marks can be cited. Moreoverrecent investigations have shown that the gas injection technology is capable to realize partswith functional hollow spaces, e.g., media pipes. This could be demonstrated for injectionmolding of thermoplastics as well as hot-curing liquid silicone rubber (LSR).2,3

These advantages already indicate the great potential of application of a gas-assistedreaction injection molding (GRIM) technology for reactive polyurethane (PU) systems.Polyurethanes are known for their broad range of material properties which can be variedfrom soft and elastic to hard and stiff. Hence the combination of these material advantageswith the processing potential of the gas injection technology could lead to an improvementof existing PU processes and products. As an example, the GRIM technology could be usedto improve the surface quality due to the possibility to apply an internal gas holding pres-sure. In addition it could lead to weight, raw material and hence cost reduction similar to thebenefits of GIM. Moreover the GRIM technology could also enable interesting new PUapplications. For example, it could be used to manufacture gas pockets in soles of shoes in aone-step process to realize special elastic properties.

However, due to the significant differences in the material behavior during the moldingprocess between thermoplastics and reactive polyurethane systems the transference of theexperiences gathered in gas-assisted injection molding of thermoplastics has to be regarded


critically. Against this background numerous investigations concerning the material behav-ior during the gas-assisted molding process have been performed.

DESCRIPTION OF THE PROCESS

The processing steps of the GRIM concept areshown in Figure 1. In accordance with the GIMprocess the first step is a short shot with the reac-tive PU mixture, characterized by the degree offilling FG. After a certain period of time, the so-called (gas) delay time, td, which enables a pre-reaction of the PU mixture the gas is injectedinto the core area of the PU mixture. During thisprocess the gas bubble propagates inside thecenter of the mixture while due to the fountainflow at the flow front the PU skin material is dis-placed from the inside to the boundary area of

the cavity. When the cavity is filled volumetric a gas holding pressure is applied until anappropriate degree of curing is reached. After this holding pressure time, th, the gas pressureis released and the part can be demolded.

RHEOKINETIC MATERIAL BEHAVIOR

One of the most important influencing parameters of a stable gas bubble propagation is theviscosity. For thermoplastics the change of viscosity during the molding process is mainlyinfluenced by cooling effects of a hot melt in a cooled mold. This solidification behaviorleads to a lower viscosity and flow resistance in the middle area of the cavity so that a cen-tered gas bubble propagation is preferred as desired. In the opposite the molding of reactivePU systems is influenced by an exothermic chemical coupled with physical processes. Thusthe rheological behavior of PU systems is depending on temperature but also on the degreeof conversion. This so-called rheokinetic material behavior leads to rather complex viscos-ity profiles inside the mold depending on time, mold temperature and the reaction kineticsof the PU system.

Moreover PU systems are generally characterized by a significant lower viscosity levelcompared to thermoplastic melts. Since it is obvious that the realization of a stable gas bub-ble propagation is easier in higher viscous liquids the GRIM process can be expected to bemore difficult to control. This underlines the significance of the pre-reaction phase. Oneconsequence of the lower viscosity is that a short shot and gas injection against the gravityis preferred as shown in Figure 1.

Figure 1. Schematic description of the Gas-assisted Reaction Injection Molding (GRIM) process.

reactive PU mixturegas injectionshort shot with

volumetric fillinggas holding

pressure curingpressure release

demolding

mixing headgas injector

A BMH MH MH MH

Gas-assisted Reaction Injection Molding 81

These pre-considerations indicate the importance of analyzing the rheokinetic materialbehavior especially during the pre-reaction phase to realize a stable gas bubble propagation.Since the rheological properties inside the mold are inaccessible for measurements a mathe-matical model has to be developed. This implies a mathematical description of the tempera-ture field which is influenced by heat conduction and heat generation effects due to theexothermic reaction. It can be modeled by the equation4

[1]

where , cp, and ρ are the thermal conductivity, heat capacity, and density, respectively. Itcan be shown that assuming a constant density and heat capacity the heat source for thePU specific isocyanate/polyol reaction close to stoichiometry can be approximated by5

[2]

with r degree of conversion,k0 absolute rate constant of the reaction, EA activation energy,R gas constant,n order of reaction,T absolute temperature,

adb maximal temperature difference max - 0 under adiabatic conditions.Taking into account equation 2 and assuming the thermal conductivity to be constant,

equation 1 can be written as

+ [3]

The parameters k0, EA, n and adb have been obtained by evaluating the exothermictemperature rise of the PU system under adiabatic conditions.6 The thermal material charac-teristics , cp, and ρ have been determined by means of a pvT- and DSC-analysis of thecured material.

Equation [3] as well as the determined material characteristics have been implementedinto the Finite Element program ABAQUS. A comparison of the calculated temperaturedevelopment inside a circular mold (D = 20 mm) with appropriate temperature measure-ments has shown a good agreement. Moreover several FTIR measurements under isother-mal conditions at different temperatures have been performed to determine the degree ofconversion versus time. Also in this case a good agreement with the calculated results couldbe observed.

To model the change of viscosity η as a function of temperature ϑ and degree of con-version r the following empirical equation7 has been used

ρ cp⋅ ∂ϑ∂t------- ∇λ∇ϑ Φ'''

·+=

λΦ'''·

Φ'''·

∆ϑadbρcpk0

EA–

RT--------- 1 r–( )n

exp=

∆ϑ ϑ ϑ

∂ϑ∂t------- ∆ϑadbk0

EA–

RT--------- 1 r–( )n

exp=λ

ρcp--------∇2ϑ

∆ϑ

λ


[4]

where is the temperature dependent initial viscosity of the mixture. The functionality has been determined by measuring the viscosity of the initial reactants at various tem-

peratures and by applying a mixing rule. In addition numerous measurements concerningthe time depending isothermal viscosity development of the PU mixture at different temper-atures have been performed. Analyzing these data it could be shown that the coefficient B ofequation [4] is a rather complex function of degree of conversion and temperature.

The FEA code for the temperature field has been extended by equation [4] as well asby the resulting dependencies of the parameters and B. To verify this algorithm the

results of calculation for the viscosity develop-ment under isothermal conditions of the reactingPU system at different temperatures have beencompared with the appropriate measurements. Inall cases a good approximation of the measure-ments could be registered.

To sum up it can be said that the developedmodel and the determined material characteris-tics enable a good approximation of the viscosityfield during the pre-reaction phase. Figure 2shows some calculations of viscosity profilesover the radius of a tubular molding at distinctreaction times for two different molding cross

ity data will be exemplarily used to interpret the observations of the molding experiments.

MOLDING INVESTIGATIONS

PROCEDURE, SET-UP AND QUALITY CRITERIA

The delay time as one of the most important influencing parameters of the GRIM processhas already been mentioned. Furthermore the mold temperature, W, has a significant influ-ence on the reaction kinetics, the temperature development and hence on the viscosity rise.Both of these process parameters have been varied in a broad range and their influence onthe part quality has been analyzed. Furthermore the influence of the gas injection pressure,pGas, or the velocity of gas bubble propagation, respectively, has been investigated.

The molding experiments have been performed using a modular mold which enables tovary the geometry of the part from D = 5 to 30 mm (circular cross section) and from B = 40

η ϑ r,( ) η0 ϑ( ) Br( )exp=

η0

η0 ϑ( )

η0

ϑ

times and different molding diameters (m = 60°C).ϑ

radius r (mm) radius r (mm)

D = 8 mm D = 20 mm

sections (D = 8 and 20 mm) and a mold temperature of 60°C. In the following these viscos-

-4 -2 0 2 4 -10 -5 0 5 10

Figure 2. Calculated viscosity profiles at distinct reaction


to 60 mm (rectangular cross section) using different mold inserts. The length of the partswas kept constant with L = 500 mm.

The PU system used was a slow reacting model formulation. It shows a marked ther-moplastic phase during its curing reaction, is characterized by a low degree of cross-linkingand belongs to the group of segmented flexible polyurethanes. Dosing and mixing of the ini-tial reactants were performed manually and the injection of the mixture into the cavity usinga pressure pot or a plunger injection unit. As an appropriate degree of filling a value ofFG=65% has been identified in pre-investigations and has been assessed for any furtherinvestigation.

As the main quality criterion the thickness of the PU skin of the final hollow article,designated as residual wall thickness r, at different cross sections over the entire length (orheight) of the gas bubble has been determined. Furthermore the specific residual wall thick-ness, rspec, which is defined as the ratio of residual wall thickness and radius of the partR=D/2, as well as the means r’ and r’spec have been used to characterize the part quality.


The investigations concerning the variation ofthe gas injection pressure (1 - 10*105 Pa) haveshown that in the entire pressure range a stablegas bubble propagation is possible. However, ata pressure level of 1 - 3*105 Pa a comparablylarge deviation from the mean residual wallthickness could be observed while for pGas >3*105 Pa the parts show a homogenous residualwall thickness over the entire length of the gasbubble with no significant influence of the pres-sure on the mean residual wall thickness.Against this background for all further investiga-tion a gas pressure of 5*105 Pa has been applied.

Figure 3 (upper diagrams) shows the (spe-cific) residual wall thickness as a function of molding diameter D for different delay timesas well as the appropriate maximum and minimum values ( W = 60°C). It can be seen thatat a long delay time of td = 360 s and for larger part diameters (D ≥ 15 mm) the resultingdeviations within the residual wall thickness are small. In contrast to this for D ≤ 15 mm agas injection even at higher pressure levels was impossible. This can be attributed to a fasterincrease of viscosity over the entire cross section (see Figure 2) due to an increasing influ-ence of the mold temperature on the rate of curing with decreasing molding diameters.

ϑ

Figure 3. Influence of delay time on the (specific) resid-ual wall thickness for different molding diameters (m = 60°C).

ϑ


Moreover it could be observed that for D = 8 mm a gas delay time of 300 s already leads toa very slow gas bubble propagation and to a significant decrease of the residual wall thick-ness from the beginning to the upper end of the gas bubble. In consideration of the calcu-lated viscosity data shown in Figure 2 it can be said that in this case the maximum viscositylevel (800 Pa s near the mold wall and 1400 Pa s in the middle of the part) is reached.

At a medium delay time of 270 s a good gas bubble structure with a low deviation fromthe mean residual wall thickness can be observed for all investigated cross sections. A simi-lar observation has been made for short delay times of 180 s and molding diameters ≤ 15mm (Figure 3, lower diagrams). Furthermore Figure 3 demonstrates that for short delaytimes and molding diameters of D ≤ 10 mm a marked good part quality could be realized.

For molding diameters larger than 15 mm and short delay times an significant increaseof deviation of the residual wall thickness with increasing molding diameters as well as sec-ondary gas bubbles in the bottom area of the part were observed. Moreover it can be saidthat in this case the residual wall thickness decreases from the bottom of the main gas bub-ble to the top end. These phenomena can be attributed to an increasing influence of gravity.It can be expected that during and after the gas injection process the relatively low viscosePU mixture flows back to the bottom of the part and in front of the gas injection gate. As aconsequence for voluminous moldings a longer delay time and hence higher viscosity levelis necessary to reach a homogenous gas bubble structure as can be observed for td = 270 s -360 s (Figure 3). As an example, for D = 20 the minimal delay time was found to be 270 s.This corresponds with a viscosity between 15 Pa s in the middle of the part and 170 Pa s atthe mold walls (Figure 2).

The phenomenon of back flow of the skin material has also been observed for themolding experiments with the larger rectangular cross section (B = 40 mm) as illustrated in

Figure 4. On the one hand at a short delay timeof 225 s this back flow leads to the generation ofthe mentioned secondary gas bubbles during thegas injection process. On the other hand in thefirst stage of the holding pressure period theback flow due to the still low viscosity levelleads to an accumulation of skin material at thebottom of the part. However, an increase ofdelay time of 45 s is already sufficient to preventthe back flow effect. Moreover, Figure 4 illus-trates that with an increasing delay time the

Figure 4. Influence of delay time on the residual wall thickness (B = 40 mm, m = 60°C).ϑ


mean residual wall thickness as well as the deviation decrease. This could also beenobserved for the other molding diameters.

To sum up it can be said that for long delay times the processing window is limited byan upper viscosity level and the resulting flow resistance. For the investigated PU systemthe mold temperature is dominating the reaction kinetics and the change of viscosity (seeFigure 2). Thus for smaller cross sections the upper limit of the processing window concern-ing the delay time is reached at shorter delay times. In contrast to this the lower limit iseffected by an increasing influence of gravity with a decreasing viscosity level.

Figure 5 shows the mean residual wallthickness as a function of molding diameter fordifferent mold temperatures (20°C and 60°C)and a medium delay time of 270 s. As mentionedabove for m = 60°C the gas bubble shows agood quality with a low deviation of the residualwall thickness for each of the investigated crosssections. In contrast to this at a marked low moldtemperature of 20 °C just the moldings with adiameter of D ≤ 12 mm are showing a compara-bly good gas bubble quality. For larger moldingdiameters a more significant deviation of resid-ual wall thickness can be observed, especiallyfor the largest cross section of D = 30 mm. Inthis case an uneven gas bubble structure in thebottom area of the part can be seen. This phe-nomenon can also be attributed to the abovedescribed back flow effect due to a slower reac-tion and lower viscosity level under these moldtemperature conditions.

For the larger rectangular cross sections afurther interesting effect could be observed. Atlow mold temperatures the moldings show alocal decrease of residual wall thickness in theheight of the mixture level of the short shot (Fig-ure 6, top). This decrease of residual wall thick-ness decreases with increasing mold

temperatures and is absent for mold temperatures higher than 60°C. An explanation for thiseffect is that due to the thermal conditions (low mold and air temperature on top of the mix-

ϑ

Figure 5. Residual wall thickness vs. molding diameter at different mold temperatures (td = 270 s).

Figure 6. Influence of mold temperature on the residual wall thickness for different molding geometry (td = 270 s).


ture level, exothermic heat generation in the middle area of the molding) the material closeto the mixture level shows a slower viscosity rise due to a slower reaction compared to thematerial in the middle area of the residual part. At the beginning of the gas injection processthis low viscose mixture is displaced to the mold walls. During the gas bubble propagationthis material is continuously cleared away due to the shear flow resulting in the observedgas bubble extension when it passes this area. In the opposite at higher mold temperatures amore homogenous mixture can be expected since the mold temperature was found to be inthe range of the temperature level in the middle area of the residual part. Hence in this casethe extension phenomenon is absent.

A further result of the process investigations concerning the influence of mold temper-ature is that below a molding diameter of 15 mm the mean residual wall thickness seems tobe independent from the mold temperature while for D ≥ 15 mm an increase of the residualwall thickness with the mold temperature could be observed (Figure 6, bottom). This behav-ior can be attributed to the faster reaction and viscosity growth close to the mold wall whilefor smaller cross sections the mold temperature is influencing the development of viscosityover the entire cross section resulting in more homogenous properties of the PU mixture.

CONCLUSIONS

It could be proved that for various cross sections the realization of a gas injection processfor reactive PU systems is possible. Depending on the processing parameters, in particularthe gas delay time and the mold temperature, a very good part quality can be obtained.Moreover it has been shown that the developed mathematical description of the temperatureand viscosity field is capable to explain the observed processing behavior.

However, at the moment these issues are restricted to simple tubular molding geome-tries and to slow reacting PU systems with a marked thermoplastic phase during the solidifi-cation reaction. In further investigations it has to be proved in what respect the GRIMprocess can be realized for more complex, praxis relevant parts as well as with faster react-ing and with non-segmented, highly crosslinking PU systems. An additional interesting taskfor future investigations is the use of foaming PU systems.

ACKNOWLEDGMENT

The investigations set out in this report received financial support from the Bayer AG,Leverkusen (Germany), to whom we extend our special thanks.

REFERENCES1 William, N., Gas-assist continues to make advances, Modern Plastics Int., (1999) 5, p. 34-36.2 Michaeli, W. and Brunswick, A., Herstellung medienführender Leitungen mit GIT”, Kunststoffe 88 (1998) 1, p. 34-39.


3 Michaeli, W., Brunswick, A. and Henze E., LSR-Bauteile mit funktionellen Hohlräumen, Kunststoffe 88 (1998) 9, p. 1404.

4 N.N., VDI-Wärmeatlas, VDI-Verlag, Düsseldorf, 1984.5 Fleischer, D., Sandwichstrukturen aus randschicht-verstärktem Polyurethan-Hartschaum, PHD thesis at the RWTH Aachen, 19976 Macosko, C. W., RIM – Fundamentals of Reaction Injection Molding, Hanser Publisher, Munich, Vienna, New York, 19897 Malkin, A.Y., Michaeli, W. et al., Rheologie reaktiver Systeme: Ermittlung von Berechnungsgrundlagen für

die Herstellung verstärkter Kunststoffe, final report of the research project (Volkswagen-Stiftung) No. I/71477, Aachen, 1998.

Chapter 2: Thin Wall MoldingThin Wall Processing of Engineering Resins:

Issues and Answers

Larry CosmaPrincipal Processing Engineer, GE Plastics

FILL TIME AND FLOW LENGTH: CRUCIAL CONCERNS

The most important factor in successful thin wall molding, around which all other process-ing issues revolve, is cavity fill time. As a wall section decreases, it becomes more difficultto get the material to flow the distances required for success in thin wall molding.

The flow length of a resin is the maximum distance a material can flow before the meltfront stops moving. One way to understand this is as a ratio that compares the length a mate-rial must travel into the mold with the thickness of the part's wall section. If the total lengthof flow for a part is 250 mm (10 in) and the wall section is 2.5 mm (0.100 in), the Length toThickness Ratio (L:T) is 100:1. At conventional wall thicknesses such as 3.0 mm (0.118 in),most molders can easily achieve this ratio, even when using PC and PC/ABS resins.

As wall sections decrease to 2.0 mm and under, the "skins" formed also will bereduced, but proportionately not nearly as much as the overall wall. The ratio of frozen-skinlayer to molten-core layer increases. When this happens, there is relatively less material inthe molten core or cross-section of the part to finish flowing and pack out the part. There-fore, the once easy-to-hit 100:1 ratio becomes more difficult to reach during processing.And when wall sections drop to 1.0 mm (0.040 in), it becomes difficult to achieve an L:Tratio of even 70:1.

For greater success at thinner walls in a given application, it's best to use conservativeL:T values for material flow. The most common techniques for reducing the L:T ratioinclude use of multiple gates or placing a gate near the center of the part. Designing with alow L:T ratio also will help reduce molded-in stress levels, increase the ability to pack outsinks at the end flow, and help ensure a more even shrinkage rate throughout the part, result-ing in less warpage.

INCREASED INJECTION PRESSURES

Molding a part with thinner walls requires shorter fill time because walls freeze and closeoff the channel of molten material faster. Since the amount of material in the wall section isless, heat dissipates more quickly, causing the material to "freeze off." Therefore, for the


injected material to properly fill the mold, lesstime is available than for parts with thickerwalls.

Injection pressures of 16,000 to 20,000 psiare typical for applications such as notebookcomputer housings, that is, parts with 1.2 mm to2.0 mm walls (0.050 in to 0.080 in). For wallsbelow 1.2 mm (0.050 in), pressures up to 35,000psi may be required. Typically, to reduce filltime, injection pressure must be increased. Forexample, consider a part with a 3 mm (0.118 in)wall section and a maximum flow length of 225mm (8.9 in). The part's L:T ratio is 75:1.Depending on the material used and the melt

temperatures employed, this part could be filled in a leisurely three seconds and packedafterward. In contrast, a similar part with a 1.0 mm (0.040 in) wall section would be limitedto a maximum flow length of 75 mm (2.95 in) to maintain the L:T ratio of 75:1. Even withthe much shorter flow length, this becomes difficult to do: the fill time in this situationwould have to take place within 0.5 seconds. As wall sections drop below 0:5 mm (0.020in), the fill time may drop below 0.1 seconds (Figure 1).

In short, all other recommendations, techniques, and guidelines stem from the need forcomplete and accurate cavity fill in very little time.

EQUIPMENT PROBLEMS AT THIN WALL CONDITIONS

When fill times drop below 1.0 second, several issues arise regarding molding equipment.These include:• Moving the screw forward at sufficient speed• Having sufficient pressure to move the screw forward at the required speed• Controlling the hydraulics to stop the screw's forward momentum when needed• Controlling the material pressure in the cavity to prevent overpacking• Platen flexure

Another machine-related problem too often overlooked is material residence at highertemperatures. As a general recommendation for its own materials, GE Plastics suggestsusing 40 percent to 70 percent of the barrel shot capacity on every cycle. GE developed thisguideline by calculating the total time the material spends in the barrel while it is being pro-cessed.

Figure 1. Injection pressure and fill time vs. wall thick-ness.

Pressure

Time

WALL THICKNESS

1mm 2mm 3mm

Thin Wall Processing 91

Since polymer bonds break down with exposure to too much heat for too long a time, itis important that the material not remain at elevated temperatures for too long. Because thinwall applications require less material to form parts with the same relative projected area,most machines now have barrel shot sizes that are too large for the thin walled parts beingmolded in them.

Downsizing barrels for thin wall applications may be required under some circum-stances. As an example, one leading thin wall molder has 180 t (200 T) molding machinesfitted with a shot capacity as small as 142 g (5 oz).

These equipment issues can be resolved, but possibly not on existing equipment. Cus-tomized machinery for thin wall molding is now being offered by several leading machinerysuppliers.

In general, standard microprocessor-controlled machines with closed-loop functionsare suitable for thin wall applications such as notebook computer housings, which are in the1.2 mm to 2.0 mm (0.050 in and 0.080 in) range. These microprocessor-controlled machinesusually give fast and accurate response to operator commands. Presses to mold parts withwalls below 1.2 mm (0.050 in) are more specialized than those typically found in cus-tom-molding facilities.

TOOLING PROBLEMS AT THIN WALL CONDITIONS

The fast injection speeds and higher pressures of thin wall molding also create toolingissues. These problems include venting, tool erosion, core-cavity alignment, plate flexure,and part release. A closer look at each of these problems, and their remedies, will be instruc-tive.

VENTING

Venting is an issue in thin wall molding. Because of the high injection speeds. there is lesstime for gases to escape from the cavity. A pressure buildup of hot gases in the cavity causesdifficulty in mold filling. This buildup of heated gas can be so great as to cause actual burn-ing or carbonization of the plastic melt. In extreme cases, the buildup of heat, pressure, andvolatiles can even etch the steel of the mold.

The problem is solved by adding vents from the mold cavity, especially where flowfronts converge and trap gases. Vents need not be any deeper or wider than in conventionaltooling. There simply needs to be more of them. It also may be useful to vent core pins, ribs,bosses, and ejector pins. Another solution is vacuum venting. Instead of venting into theatmosphere, vents enter a tightly sealed system with a vacuum pump attached to it. Eachtime the mold closes, atmospheric air is vacuumed out of the cavity before injection starts.


TOOL EROSION

Material flowing into a mold can gradually wear away the tool steel. Most erosion generallytakes place near the gates that fill the part. Erosion takes place in the molding of conven-tional parts, and is exacerbated in thinner walls because of the higher fill rates. Filled gradescan accelerate erosion. Erosion can be minimized through the use of hardened tool steelssuch as A-2, D-2, and M-2.

CORE-CAVITY ALIGNMENT

In thin wall processing, the high pressures tend to push the core and cavity in differentdirections. This can result in parts with changes in wall thickness at different locations. Cru-cial tolerances, as well as the ability to fill the cavity, can be compromised.

Core and cavity can be aligned through the use of mold interlocks. Typically, two orfour interlocks are used in the parting line surfaces of the mold.

PLATE FLEXURE

Mold plates must be thick enough and bolted together tightly enough so they don't move.But at high pressures, movement can occur unless this is accounted for in tool design. Usingextra heavy steel plates, support pillars, dowel pins, and bolts adds the necessary stability.

PART RELEASE

The tighter the cavity is packed, the more theplastic tends to adhere to features in the mold. Ifthe mold is overpacked, the material can gripsurface features. Also, some materials adhere tothe tool steel more than others. Using more andlarger ejector pins and sleeves is necessary (Fig-ure 2).

Commercially available mold coatingsadded to the surface of the tool steel will alsohelp in releasing parts.

AESTHETICS

Portable telecommunications devices such as notebook computers and cellular phones havebecome popular consumer items. They must be attractive, particularly at the point of pur-chase.

Gate vestiges are often undesirable. and direct gating often will leave gate blush andother flow marks. Valve gating is an excellent way to improve surface appearance. Careful

Figure 2. Part ejection.

Thinwall EjectorLayout

Conventional EjectorLayout


site selection of the valve gate may allow for direct gating to the appearance side of a part inan area that subsequently will be covered by a decal or other secondary operation. Newvalve gates, such as those supplied by Kona Corp., have shown exceptional part aestheticswhen placed opposite an appearance surface in 1 mm (0.040 in) walls.

Thinner wall sections often will reduce sink marks seen near ribs and bosses, but"ghosting" from wall transitions may be more evident. Painting can hide flaws, but it willincrease the cost of the part and has environmental concerns. Often, adding texture is a bet-ter and far less expensive way to disguise surface flaws or make them less noticeable. Theheavier the texture, the more it will hide cosmetic blemishes.

Another decorative option drawing much attention has been in-mold decoration.In-mold decoration has been used for years in the automotive industry, especially in back-litinstrument panel fascias. Now, the process is being explored to decorate portable comput-ers. Through in-mold decoration, notebook and subnotebook computer housings can be cus-tomized with a virtually unlimited numbers of colors, prints, designs, and logos. In thisprocess, thin, clear, pre-printed films are inserted into the mold before injection. The filmadheres to the plastic, forming a single part with a superior even dramatic, appearance.

Films comes in a variety of materials, such as polycarbonate, vinyl, and polyester.They range in thickness from 0.25 mm to 0.75 mm (0.010 in to 0.030 in). Depending on thegeometry of the surface to be covered, the film may be a flat sheet which forms to the minorcontours of the cavity or is preformed to the contours of the part. Film can be held in placeduring the molding process in several ways, including vacuuming, registration pins, staticelectricity, and core geometry. Costs are figured on a per square-foot basis.

As an example, a polycarbonate film 0.25 mm (0.010 in) printed and cut to fit the coverof a notebook computer may cost about 50 cents. Total costs would include film loading andadjustments in cycle time.

Another issue in thin wall aesthetics can be knit lines. While multiple gating may helpimprove L:T ratios, they create more knit lines. Again, texture will help hide knit lines.Also, multiple "live feed" processes such as the SCORIMTM process can reduce or com-pletely eliminate knit lines. In this licensed process, material is delivered by two runner sys-tems to opposite sides of the part. Two pistons, agitating in alternating movement, keep themolten core "live" longer and minimize if not eliminate unattractive knit lines.

Achieving a good surface appearance when using glass- or mineral-filled materials canbe a further challenge to high aesthetics. Fast injection rates, hot mold temperatures, andheavy textures can help hide the fillers in these materials. Some resins, such as LEXAN SPpolycarbonate resins with 10 or 20 percent glass filler, flow very well with little or no glassshowing on the surface.


MATERIALS

Choosing the right materials, of course, is crucial for any application. The correct material,one that will perform according to all of the design and processing criteria, can mean thedifference between success or failure in a part. The proper material can enhance productdesign and improve manufacturing by creating a broader processing window. On the otherhand, the incorrect resin can render an otherwise good design unworkable and difficult toprocess.

Material selection is a process of elimination. Key application requirements are definedand ranked, and these requirements are translated into physical properties. The propertiesare then compared to the resin families offered by materials suppliers. In many cases, theactual molding of parts will help finalize material selection.

Every application is different and each presents unique design and processing chal-lenges. But from surveys of product designers, a list of general materials requirements, inorder of importance, can be presented:

FLOW LENGTH

The most critical property of a thin wall part is the flow length. The material must be able tofill the mold. Often, resin suppliers compare relative flow lengths of resins by using spiralflow test numbers. However, designing parts and molds using spiral flow data can be mis-leading and dangerous to the success of an application. Spiral flow data represents the max-imum flow of a material through a channel. Seldom will an application exhibit pure channelflow. So for design purposes, it's best to be conservative.

IMPACT STRENGTH

A part bets its durability and toughness from the material. It is important to understand theimpact requirements of the assembled part, especially portable devices. The resin's physicalperformance will generally fall within specifications offered by the supplier, assuming itwas properly processed within recommended guidelines. Unfortunately, this can be a bigassumption when it comes to thin wall molding.

AESTHETICS

The part must look good. Since the majority of designers prefer the cost advantages ofunpainted parts, additional processing challenges are created.


STIFFNESS

Stiffness of the application becomes a critical, issue as the part gets smaller and the wallthickness is reduced. Stiffness is most often achieved with part design and assembly tech-niques rather than the flexural modulus of the resin.

HEAT RESISTANCE

The kinds of portable applications typical of thin wall molding (e.g., notebook computers)require better thermal performance than stationary applications (desktops) because of thedifferent end-use requirements and smaller package size.

FLAME RETARDANCE

Because FR materials create added headaches for molders, they should be used only whennecessary. From a design standpoint, there are few differences between FR and non-FRproducts. Typically, cellular phones do not require FR materials, while notebook computersrequire UL 94* V-1.

MECHANICAL INTEGRITY

As it relates to the final assembly of the part, mechanical integrity is often overlooked. Butassembly methods and material selected can greatly impact the weight of the part, as well asits total cost. Materials used for thin wall applications require, processing freedom and supe-rior performance to withstand rigorous molding environments and sometimes abusive end-use conditions.

COMMON MOLDING ERRORS

The most common error made in processing thin wall applications relates to temperature. Tomake mold filling easier, processors will often raise material temperatures above recom-mendations. Turning up barrel temperature reduces the viscosity of the material, enhancingflow, and since thin wall parts are more difficult to fill, the temptation is always to increaseheat.

However, higher temperatures can result in loss of physical properties, and the firstproperty to suffer is impact. Degraded material will generally produce brittle parts; ductilityvalues will not be as great as those reported by the materials supplier. The best guidelinehere is to stay within the material supplier's recommendations for drying time and tempera-tures, barrel temperatures, and residence time in the barrel and hot runner system.

Another common mistake is thinking a part can be filled with more pressure, whenwhat it really needs is a shorter fill time. Fill time is a dynamic process. Added pressure


alone won't guarantee optimum processing. In fact, it can cause problems such as molded-instress, anisotropic shrinkage, and warpage.

FASTER CYCLE TIMES

Getting the heat out of a part in a short amount oftime has a notable advantage: cycle times can bereduced. The 3.0 mm part (0.118 in) mentionedearlier may have a cycle time of 40 seconds orgreater. But when the walls are reduced to 1.0 mm(0.040 in), the cycle can be expected to be below20 seconds (Figure 3).

Some molders are achieving cycle timesunder seven seconds. These cycles are so fast thatrobots must be used to remove parts, since gravitycannot clear them from the mold halves fastenough. Cycles this fast are easy to achieve withhigh performance molding machinery equippedwith accumulators on the clamp and injectionunits, plus advanced hydraulics and valves. Care-ful attention must be paid to the cold runner sys-tem or sprue for fastest cycle times.

The faster times of thin wall molding also influence the heat history of the materialwithin the barrel. In addition, the guideline of a 40 percent to 70 percent shot size vs. capac-ity may not always apply. Since the cycle time may be one third of that used for conven-tional molding, a lower percentage of barrel capacity may be acceptable in someapplications. Molders should verify this by testing properties of the molded parts they areproducing, regardless of the machinery sizes and cycle times.

CONCLUSIONS

As consumer demand grows for smaller, lighter computers, cellular phones, and other tele-communications and data storage products, the need for advancements in thin wall technol-ogy will grow proportionately.

Material suppliers are responding by delivering new resins that will meet the imposingphysical demands of the OEM designer. They are producing resins that flow well, offerimpact strength, and provide good aesthetics, stiffness, and heat properties. For their part,designers are beginning to better understand the differences in thin wall design, particularlywith respect to the issues of impact, stiffness, and manufacturability.

Figure 3. Typical cycle time range.


However, the greatest gain in successful processing of thin wall applications rests withthe tool designer and the molder. Increasing flow length can best be influenced by usingvery fast injection speeds coupled with the special tooling techniques and machinery dis-cussed here.

Thin wall technology can't be learned overnight. It requires an investment of time aswell as money, but it's an investment that potentially offers an enormous return.

Effects of Processing Conditions and MaterialModels on the Injection Pressure and

Flow Length in Thinwall Parts

A. J. PoslinskiGE Corporate Research and Development, Schenectady, New York, 12309, USA

INTRODUCTION

Tradeoffs between machine capabilities, production rates, and structural performance usu-ally result in plastic parts that are about 2 to 3.5 mm (80 to 140 mil) thick. However, withthe recent trends toward portable communication and computer miniaturization, the conven-tional design envelope no longer applies to products such as personal pagers, cellularphones, and notebooks. Plastic shells on the order of 1 to 2 mm (40 to 80 mil) have beensuccessfully integrated with electronic components that provide the necessary stiffness.1

Because the tool modifications corresponding to this step change in wall thickness result infaster cooling and greater deformation of the molten plastic during processing, one goal ofthis work is to compare the skin and core structure of conventional and thinwall parts. Thehighly oriented skin and the extent of the randomly oriented core strongly influence thepotential for anisotropic shrinkage and nonuniform warpage.

The main difficulty facing thinwall designers and processors is achieving longer flowlengths without introducing knitlines caused by multiple gates. Inevitably, elevated melttemperatures, faster injection speeds, and higher injection pressures must be used to reachthe current 100:1 and 150:1 flow length to wall thickness ratios for cellular phones andnotebooks.1 The wrong combination of thinwall process settings may lead to material deg-radation and an increased number of shows that the rheological and impact performance ofthese polycarbonate materials does not significantly change within the temperature rangerecommended by resin manufacturers. However, thinwall parts molded at higher tempera-tures are more likely to fail in a brittle manner when exposed to sudden impact conditions.Clearly, thinwall molding is best accomplished by maintaining the melt temperature withinthe recommended limits. The present work expands this concept further by performing aparametric analysis to identify the primary variables affecting the injection pressure and themaximum flow length of thinwall resins and to suggest the optimum combination of processsettings for thinwall molding.


MATERIAL PROPERTIES

An accurate prediction of mold filling requires a viscosity model that is valid over the entirerange of shear rates and temperatures encountered during processing. An equation that pro-vides a good fit of the viscosity data is the generalized Cross model:3

[1]

where denotes the zero-shear viscosity, and , n, and * are material constants. Thetemperature and pressure dependence of the viscosity is indirectly captured through witha simple exponential or the Williams-Landel-Ferry (WLF) expression��

[2]

[3]

where Ei and Wi are material constants, P is the pressure, T is the temperature, and Tg is theambient glass transition temperature. Because data on the pressure dependence of the vis-cosity is not readily available, an average value for the coefficient E3 can be estimated overthe temperature range of interest.4 Likewise, the temperature data since it represents thepressure dependence of the glass transition temperature.

The thermodynamic properties that influence the heat transfer taking place duringmold filling include the specific heat, the thermal conductivity and the density. Constantproperties are used if the molten plastic is assumed incompressible. When this approxima-tion is relaxed, the specific heat and the thermal conductivity remain unchanged, but thetemperature and pressure dependence of the density is described with the double-domainTait equation of state:5

[4]

The specific volume V is equivalent to the reciprocal of the density , Vo is the spe-cific volume under ambient pressure, and Bi are material constants. The Tait equation isapplied separately in the molten and solid states, yielding two sets of coefficients above andbelow the glass transition temperature of Tg - W3P.

The numerical calculations presented in this report are based on the viscosity data of animpact-modified polycarbonate resin (LEXAN® UL6339R grade). Table 1 lists the materialconstants for the Cross exponential and WLF temperature models. In both cases, the powerlaw index n is set to 0.2, and the shear stress constant *, which is related to the onset ofshear-thinning behavior, is fixed at 5.17 x 105 Pa. A constant density of 1000 kg/m3 is used

η ηo 1ηoγ·

τ∗---------

α

+

n 1–( ) n⁄

=

ηo α τηo

ηo E1 E2 T⁄( ) E3P( )expexp=

ηo W1

W T Tg– W3P–( )–

51.6 T Tg–( )+----------------------------------------------exp=

V 1 ρ⁄ Vo 1 0.0894 1 P B⁄+( )ln–[ ]= =

Vo B1 B2 T Tg–( )+=

B B3 B4 T Tg–( )–[ ]exp=

ρ

τ

101Effect of Processing Conditions

to enforce incompressibility; otherwise, the spatial variation of the density is predicted withthe Tait equation of state and the corresponding material constants listed in Table 2. A con-stant specific heat of 2056 I/kg K, a constant thermal conductivity of 0.25 W /m K, and anambient glass transition temperature of 144°C complete the required set of material proper-ties.

Table 1. Material constants in SI units for the Cross model

Type 1 2 3

Exponential l.l9xlO-7 l.26xlO4 l.7xlO-8

WLF 5.05xlO12 30.9 l.9xlO-7

Table 2. Material constants in SI units for the Tait equation

3 4Type 1 2

2.99xlO8 1.71xl0-3solid 8.54xlO-4 l.59xlO-7

melt 8.54xlO-4 5.62xlO-7 l.83xlO8 3.99xlO-3

Including the pressure dependence of the zero-shear viscosity in the material modelrequires further consideration. Normally, the pressure coefficient E3 or W3 is set to zero whenthe exponential function in Equation [2] or the WLF function in Equation [3] is fit to theexperimental data. The other two constants El and E2 or W 1 and W 2 are then associated withthe ambient pressure, which is set to zero as a reference point. Independently changing E3 orW 3 results in the viscosity increasing above the measured values for pressures larger than zero.

Because viscosity measurements are usually pressures, the reference point for the viscosity iso-bars should be greater than zero. This simply means that the viscosity level, which is controlledby El or W 1 needs to be adjusted when nonzero values of E3 or W 3 are imposed. One way to

account for the pressures generated during the viscosity tests is to f1rst calculate the averagemean pressure in the capillary die and then modify the viscosity coefficients accordingly. In thecase of the polycarbonate resin considered in this study, the average mean pressure based on

capillary viscosity data6 at a temperature of 305°C (580°F) is approximately 17 MPa (2.5 ksi) forshear rates ranging from 1000 to 2000 s-I. Using this value as the reference point, the exponen-tial and WLF material constants, El and Wl in Table 1, are reduced to 8.88 x 10-8 Pa s and3.135 x 1012 Pa s, respectively. The corrected values shift the viscosity isobars downward, so


WKDW WKH\ FURVV RYHU DW WKH UHIHUHQFH SUHVVXUH DQG WKH YLVFRVLW\ OHYHO PHDVXUHG ZLWK WKH FDSLOODU\

UKHRPHWHU�

NUMERICAL CONSIDERATIONS

Injection molding simulations are performed with a pro-cess model that simulates the flow of molten plastic intoany arbitrary molding geometry.7 The part geometry isshown in Figure 1. Specifically, the main componentsinclude a machine nozzle, a sprue, and a quarter disk.The quarter disk represents the radial flow in actualmold filling; its size and thickness matches the typicalaspect ratio of thinwall parts and the mold geometryused in a related experimental study.8 The additionalpressure drop in the sprue is modeled with a cold runner,and the additional shear heating at higher flow rates inthe nozzle is modeled with a hot runner. The hot mate-rial forming the shot in front of the nozzle and the

dynamics of the injection unit are replaced with constant melt temperature and variablescrew velocity conditions at the nozzle entrance.

SKIN AND CORE STRUCTURE

Figure 2 shows a surface and contour plot of the cross-sectional temperature distributionpredicted for an 82 cm3/s (5 in3/s) flow rate and a wall thickness of 1 mm (40 mils) at theend of filling. The radial and axial coordinates correspond to the flow and thickness direc-tions, respectively. In particular, the left and right edges represent the gate entrance and theend of flow; whereas, the upper and lower edges represent the mold wall and the midplaneof the quarter disk. The surface topology shows that the temperatures at the gate entranceand near the midplane are almost identical due to convection of heat in the flow direction,and a steep temperature gradient develops near the mold wall as a caused by high shear ratesraises the temperature by as much as 30°C (80°F) near the upper left corner of the domain.The contours reveal that a band of higher temperatures extends from this region along thediagonal toward the lower right corner. The temperature spike increases at higher flow rates,disappears completely when the molten plastic is injected at approximately 16 cm3/s (1 in3/s), and is not affected by wall thickness.

The temperature distributions can be used to identify the solidified material in contactwith the mold wall. The solid thickness is practically nonexistent near the gate entrance; it isalso reduced near the melt front as a result the fountain flow approximations. Because the

Figure 1. Schematic drawing of the part geom-etry.

1/2 in2.5 in

1/4 in2 in

5/16 in 6 in

0.04 in

Effect of Processing Conditions 103

temperature distribution shown in Figure 2 is notstrongly affected by the wall thickness, thenumerical calculations confirm that the solidlayer is similar for both conventional and thin-wall parts, constituting about 2% of the totalthickness. At the lower flow rates of 16 to 82cm3/s (1 to 5 in3/s), the solid layer is on the orderof 10%.

Another component that makes up the skinregion is the shear zone next to the solid layer.The size of the shear zone is similar for bothconventional and thinwall parts; however, con-siderably higher shear rates are observed in the

latter case. The shear zone makes up approximately 30% of the total thickness, so that thetotal skin region is on the order of 40%.

INJECTION PRESSURE PREDICTIONS

If mold filling is not constrained by the limitations of the injection unit, the pressure at thesprue entrance increases in an approximately linear manner as the melt front advances fur-ther into the part. The final value required to completely fill the entire mold cavity is definedas the injection pressure. The results indicate that the lower viscosities at higher melt tem-peratures and faster injection speeds reduce the pressure requirements. Whereas the mini-mum pressures for conventional molding are obtained around 164 cm3/s (10 in3/s), theinjection pressures for thinwall molding continue to decrease at flow rates even as high as492 cm3/s (30 in3/s). Evidently, the additional heat dissipation at the higher shear rates inthinwall parts compensates for the higher pressures required to achieve faster injectionspeeds.

The effect of various material models on the injection pressure at 82 cm3/s (5 in3/s) and305°C (580°F) demonstrates that slightly higher predictions are obtained when the spatialvariation of the density is taken into account because some of the pressure is used to com-press the material. However, including the effect of compression work raises the melt tem-perature slightly, so that the injection pressures are reduced back to the levels predicted withthe incompressible material model. Significantly higher pressures are predicted when thepressure dependence of the viscosity is included; although, the correction for the pressuresgenerated during capillary measurements results in somewhat lower values. Furthermore,the difference between the exponential and WLF temperature models becomes more appar-

Figure 2. Cross-sectional temperature distribution for an 82 cm3/s (5 in3/s) flow rate and a wall thickness of 1 mm (40 mils) at the end of filling.

Radial Distance (In) Axial D

istan

ce (m

il)

Tem

pera

ture

(F

) 500

400

300

200

100

0

2

046

8

16

20210270

350

310

360


ent; the WLF function predicts a steeper viscosity rise with pressure, and thus higher injec-tion pressures.

FLOW LENGTH PREDICTIONS

If mold filling is constrained by the limitations of the injection unit, the flowrate decreasesafter the injection pressure reaches the maximum allowable value. When the molten plasticstops flowing and does not reach the end of the mold cavity, the flow length is defined as thefarthest distance from the gate attained by a short shot. The numerical calculations indicatethat longer flow lengths are obtained at higher melt temperatures. The flow length is notstrongly affected by the flow rate, except at the higher pressure limits, when the longer flowpath provides more time for the melt to heat up at the higher flow rates. An examination ofthe cross-sectional temperatures reveals that the short shot is not caused by freeze off;rather, the molding machine does not provide enough force to push the material further. Theflow length trends obtained with various material models are similar to the injection pres-sure trends. The pressure dependence of the viscosity has the greatest effect, resulting inshorter flow lengths.

SUMMARY

A parametric analysis has been performed to investigate the process mechanics of thinwallmolding.: Numerical calculations confirm that the skin and coy structure in thinwall parts issimilar to the structure i conventional parts; however, considerably higher shear rates areobserved in the shear zone of the skin region The injection pressure is reduced with highertemperatures and faster injection speeds, and the resin flow length is increased with highermelt temperatures and larger machine capabilities. The spatial variations of the density andthe additional compression heat resulting from higher thinwall pressures do not significantlyaffect these two variables. However, experimental evidence is needed to validate the higherinjection pressures and shorter flow lengths caused by the viscosity pressure dependence.The present analysis suggests that thinwall molding is best accomplished within the temper-ature range recommended by resin manufacturers and on molding machines that providefaster injection speeds and higher pressure limits.

ACKNOWLEDGEMENTS

This work was supported by GE Plastics. The authors also wish to thank Jack Berkery forperforming the numerical simulations. Also, helpful modeling suggestions by Toni Gennariare duly appreciated.

Effect of Processing Conditions 105

REFERENCES1 Thinwall Technical Guide, 2nd ed., Wireless Network Telecommunications Group, GE Plastics, Pittsfield,

Massachusetts (1995).2 A. J. Poslinski, L. O'Connell, P. R. Oehler, SPE Annual Technical Papers 42, submitted (1996).3 C. A. Hieber and H. H. Chiang, Rheologica Acta 28, 321-332 (1989).4 C. A. Hieber, in Injection and Compression Molding Fundamentals, A. 1. Isayev, ed., 1-136, Marcel Dekker, Inc., New York (1987).5 P. Zoller, in Polymer Handbook, J. Brandrup and E. H. Immergut, eds., John Wiley and Sons, New York (1989).6 Engineering Design Database and Design Guide, Commercial Technology Division, GE Plastics, Pittsfield,

Massachusetts (1989).7 C -MOLD CAE Software, AC Technology, Ithaca, New York (1995).8 A. J. Poslinski and G. Tremblay, SPE Annual Technical Papers 42, submitted (1996).

10 Common Pitfalls in Thin-Wall Plastic Part Design

Timothy A. PalmerBayer Corporation, 100 Bayer Road, Pittsburgh, PA 15205, USA

DEFINITION OF THIN-WALL

For the purposes of this paper, a thin-wall part is defined as one injection molded in an engi-neering thermoplastic resin (e.g. PC, PC/ABS, PA6), having projected area greater than 8square inches and nominal wall thickness less than 0.060" (1.5 mm). Today, many thin-wallapplications push beyond this defined limit and use nominal wall thicknesses less than0.040" (1.0 mm).

PITFALL #1:DESIGNING WITH TOO MUCH VARIATION FROM THE NOMINAL WALL

THICKNESS

After the molten resin is injected into the mold cavity, different areas of the plastic partexperience different levels of volumetric shrinkage proportional to wall thickness. In con-ventional moldings packing pressure is applied to force more molten material into thethicker areas, minimizing the effects of differential shrinkage.

Unlike conventional parts, molten resin in thin-wall parts solidifies only a few secondsafter the end of fill, giving packing pressure little time to act. The thinnest walls solidifybefore significant volumetric shrinkage can occur. Thicker areas take longer to freeze, expe-riencing very high volumetric shrinkage. In the worst case, material around the gate cansolidify before any area of the part can be adequately packed-out.

The notion that molten plastic follows the path of least resistance is especially true inthin-wall housings. Often, advancing flow will simply not fill the thinnest areas of a part,creating either non-fill or gas entrapment.

Because of these difficulties, thin-wall parts should be designed with uniform wallthickness as much as possible. This allows molded parts with low differential volumetricshrinkage, improved dimensional quality and reduced chance of cosmetic problems causedby non-fill or gas entrapment. However, the decision to use nominal wall design must bemade early in the design cycle due to the restrictions it may impose. Often, additional wallthickness must be added to the inside of a housing opposite areas such as label recesses to


maintain the nominal wall thickness. Note that as with conventional parts, sharp edges inthe flow path and at internal corners should be avoided.

PITFALL #2:USING IMPROPER RIB TO WALL THICKNESS RATIO

The thick section formed by the intersection of a rib and the nominal wall tends to experi-ence greater volumetric shrinkage than the rest of the part, causing sink opposite the rib. Inconventional housings, rib base thickness is based on a percentage of the attached nominalwall, varying from 50 to 66% depending on the degree of cosmetic perfection desired. Thisdesign practice acts to reduce the thick section and make it easier to pack-out, largely elimi-nating visible sink.

When standard rib design rules are applied to thin-wall parts, the resulting rib designsare usually too thin to fill properly, especially after draft is added. If the ribs can be filled,freeze-off usually occurs well before the rest of the part, with shrinkage much different thanin the attaching nominal wall. To allow the ribs to fill properly, a 1:1 rib to wall thicknessratio can be used in walls less than about 0.050" thick. Any resulting sink marks tend to bemuch less noticeable than with conventional parts, especially if the opposing surface is tex-tured. In a thin-wall part, there is much less material at the rib/wall intersection to shrinkand cause sink than in conventional molded parts.

PITFALL #3:CONSIDERING ONLY EASY-FLOW RESINS FOR THIN-WALL

APPLICATIONS

Thermoplastic resins are often available in a range of molecular weights. Grades with lowermolecular weight typically have lower melt viscosity and flow farther under the same pres-sure than their higher molecular weight counterparts. Unfortunately, easier flow usuallycomes at the expense of physical properties such as yield strength and impact strength. Inaddition, a material's resistance to UV light and chemical attack are reduced with decreasingmolecular weight.

Because thin-wall applications can be difficult to fill, the expected flow properties oflow molecular weight resins seem desirable. Figure 1 shows the difference in predicted fill-ing pressure between high and low molecular weight grades of polycarbonate for a samplehousing. Mold-filling analysis results for the 0.040" (1.0 mm) nominal wall show thatregardless of molecular weight, high-performance injection molding equipment is probablyrequired. In this case, using a lower molecular weight resin may sacrifice material propertieswithout significantly reducing production costs.

10 Common Pitfalls 109

PITFALL #4:RELYING ON FIBER-REINFORCED RESINS TO PROVIDE RIGIDITY

The structural rigidity of a thin-wall housing is greatly reduced versus its thick-wall coun-terpart due to the reduction in section modulus. From the standard engineering beam bend-ing formula (w/both ends simply supported), the maximum deflection is inverselyproportional to the thickness cubed, so under identical loads, a beam 0.040" thick hasdeflection 8 times a wall 0.080" thick. A potential solution for thin-wall housings is to use afiber-filled resin, which typically increases the material modulus by about 50% (10% glassfiber-filled). However, maximum deflection is only inversely proportional to the materialmodulus, so the unfilled beam only deflects 1.5 times more than the fiber-filled one.Because the wall thickness effect dominates over the effect of fiber reinforcement, the rigid-ity of thin-wall housings cannot be expected to compare to thick-wall, conventional hous-ings. Rigidity of thin-wall applications will still depend on assembly with the product'sother internal components, regardless of the resin used.

Impact properties are also important for thin-wall housings given their widespread usein hand-held products prone to being dropped. Fortunately, thinner walls may performslightly better in a drop impact because more flexible walls have better energy absorption.However, the addition of fillers can sharply reduce these properties. For example, thenotched izod impact strength of 0.125" thick polycarbonate is reduced from 17 ft lb/in to 2 ftlb/in when 10% glass is added. These examples suggest that the liabilities of fiber-filledmaterials may outweigh their benefits in most thin-wall parts.

Figure 1.

Resin Melt Flow Index (g/10min)

35 20 12

Injection Rate - 15 cu. in./sec.F i l l i n g P r e s s u r e v s . W a l l T h i c k n e s s

Housing Nominal Wall Thickness (in.)

Inje

cti

on

Pre

ss

ure

(p

si)

35000

30000

25000

20000

15000

10000

5000

0

Centrally Gated Housing, Single Hot Drop

0.750 "

9.012.0"

"


PITFALL #5:IMPROPERLY LOCATING GATES

Thin-wall applications push thermoplastic resins and standard injection molding equipmentto their respective limits, but properly locating gates is often overlooked as a way to widenthe available processing window. Unfortunately, gate locations are often chosen after partdesigns are finalized, leaving only a few locations where gate vestige is allowed. A betterapproach is to pick gate locations early in the design cycle to optimize filling, and then posi-tion label areas or other styling to conceal any remnant of vestige.

In conventional as well as thin-wall parts, filling pressure is minimized when all of thelast areas to fill do so simultaneously. This phenomenon is known as balanced filling andpromotes uniform solidification and packing of the part. When wall thickness is uniform ina thin-wall part, gate locations should be chosen so that the longest flow paths from all gatesare equal in length.

However, if a thin-wall part has non-uniform wall thickness, truly balanced filling isdifficult to achieve. In fact, some degree of filling imbalance may actually improve themoldability of a non-uniform wall part. Mold-filling analysis is required to optimize suchcases. When analyzing a thin-wall part, the mold-filling analyst should always consider thepart and the delivery system (e.g., three-plate runner, hot manifold), because pressure con-sumed in these components can have a much greater effect on flow balance in thin-wallparts than in conventional designs.

PITFALL #6:USING SLOW INJECTION RATES

While high injection pressures are required to fill thin-walled parts, delivering the moltenresin at a sufficient injection rate is also an important parameter. To prevent early freeze-off,the molding machine must inject material at a rate high enough to produce shear heating atthe flow-front. Once the flow-front temperature begins to drop, the pressure required toadvance it can quickly exceed press capabilities, resulting in non-fill.

Today's closed-loop, electronic controls allow nearly any injection rate to be set at thepress, but close examination of the actual ram velocity vs. position trace may show that thedesired injection rate can only be achieved over a small portion of the injection cycle, if atall. In this case, a "high-performance" injection molding press designed specifically for highinjection rates will be required. Such machines have the ability to deliver high pressure atvery high injection rates through the use of accumulators or other methods.

10 Common Pitfalls 111

PITFALL #7:USING MORE GATES THAN NECESSARY

In many thin-wall applications, numerous gatesare used when fewer would be suitable becausethe material is not expected to flow more than afew inches beyond the gate. However, as men-tioned in #6, significant flow in thin walls is pos-sible when flow front velocity is high enough.The rapid freeze-off expected in thin walls typi-cally occurs because the flow front velocity istoo low to generate shear heat ing.

While the ability to maintain highflow-front velocity is largely dependent on thecapabilities of the injection molding press, the

number of gates used also plays an important role. Assuming radial flow from a pin-pointstyle gate, the flow front velocity is inversely proportional to the distance flowed. If asquare housing is fed through a centrally located gate (Figure 2), flow front velocity at theend of fill is Q/2 Rt, where Q is injection rate, R is the radial distance flowed and t is partthickness. When multiple gates are used to fill the part, flow distance is reduced, but theinput flow rate must be divided among the gates. In this example, the four gate system hashalf the flow front velocity of the single gate system at the end of fill. The par t with a single,center gate has higher flow front velocity at the end of fill, no major knitlines and avoids gasentrapment at the center of the part.

PITFALL #8:UNDERSIZING GATES

Because higher injection rates are used in thin-wall molding, larger gates are required toprevent cosmetic damage caused by excessive gate shear. Externally heated hot drops orvalve-gated drops allow large gate diameters with clean degating. The following formulacan be used to estimate the required pin-point or hot-tip gate orifice diameter.

Here, the diameter D is a function of Q, the volumetric flow rate from the nozzle, n, thenumber of gates and the shear rate limit. For engineering thermoplastics the shear ratelimit is usually 20,000-40,000 1/s, depending on the shear-sensitivity of the resin. Use alimit of 20,000 1/s for shear-sensitive resins. Note that this formula assumes equal flow

π

D 32Qnπγ----------3=

γ

Figure 2.

InputFlowRate, Q R/2

Q/4

Q

t

R

vv

RRV=Q/2πRt V=Q/4πRt


passes through each gate. It can also be used tosize tunnel gates, which should have at least a 20°included angle and be at a 45° angle to the partingline.

If a three-plate runner is used, large gatesmay cause damage the thin nominal wall duringdegating. This can be avoided if a reinforcingdome is used opposite the gate as shown in Figure3. Keep in mind that pressure imbalance betweenmultiple drops in cold, three-plate runners may bemore than with hot runner systems.

PITFALL #9:UNDERESTIMATING CLAMP TONNAGE REQUIREMENTS

In thin-wall molding, it is not uncommon for the process window to be limited by the moldblowing open due to high cavity pressures. With conventional parts, clamp tonnage esti-mates of 3 tons per square inch are often adequate. Thin-wall applications must typicallyallow for more than 5 tons of clamp per square inch of the mold cavity projected area. If thepart to be filled is large, the mold and backup plates should be about twice as thick as con-ventional parts to prevent flexing during high-pressure injection.

PITFALL #10:INADEQUATE VENTING IN THE TOOL

The fast injection rates used in thin-wall molding require larger parting line vents, primarilyto prevent flow hesitation as air is pushed from the cavity at the end of fill. However, thehigher injection pressures and better flowing resins used increase the risk of parting lineflash. A mold designed with a generous number of thinner vents may be the best compro-mise. Proper venting in the areas where air is chased at the end of fill is especially critical.Air trapped ahead of a quickly converging flow front can significantly increase filling pres-sure requirements.

Figure 3. Suggested pin-point gate detail for thin-wall parts requiring large gates.

Gate DiameterReinforcing Dome~2.0 mm

~1.0 mmNominal Wall Thickness

Flow Instabilities in Thin-wall Injection Molding ofThermoplastic Polyurethane

Christian D. Smialek, Christopher L. SimpsonPlastics Engineering Dept., University of Massachusetts, Lowell, USA

INTRODUCTION

Thin-wall molding is conventionally defined as molding parts that have a thickness of 1 mmor less and a surface area of at least 50 cm2.5 Thin-wall molding has been around fordecades, but due to the narrow processing window it did not catch on. The earliest applica-tions for thin-wall molded parts were in the container and packaging industries. Recent re-interest in thin-wall injection molding is due to economic and environmental concerns whenit was realized that products could be made lighter, more compact, and less expensive, aswell as made quicker due to the reduced time for cooling the part during processing.7

In the past processing equipment could not generate the high pressures that are neededin thin-wall injection molding. As processing equipment became more robust and as therewere revolutionary advancements in resin and in process control equipment, thin-wallmolding was realized as a viable alternative to conventional molding. Advancements in pro-cess control systems have enabled the processor to have more precise control over the entiremolding process. This enabled the production of parts with tighter and higher tolerances.

While process control systems allowed one to have better control over our processingwindow, the introduction of high flow resins in the 80’s and the single site catalyst of the90’s allowed for expansion of the previously found narrow processing window.1 Resinswith narrow molecular weight distribution exhibit better properties and more stability athigh rates of flow. This makes them ideal for thin-walled applications were high injectionpressures and flow-rates are the norm.

Market demands of the computer and telecommunications industries has fueled the lat-est interest in thin-wall molding.7 Each new product line demands smaller and lighter prod-ucts and as a result the thin-walled parts have been attractive by providing housings at areduced weight, size, and cost.

Thin-wall elastomeric injection molding, specifically TPU, has not been evaluated,extensively, to date. Reasons for this center on the (still existing) processing difficultiesassociated with the resiliency and high viscosity of the material. However, it should be real-


ized that gaskets, seals, and other products can now be made thinner and less expensivelywith thin-wall injection molding.

EXPERIMENTAL

The virgin resin chosen for this experiment was Desmopan – 453 (Miles; lot no: 3922238-28-1). Before processing the resin was dried for 12 hours at 88°C in a desiccant drier(Novatec NPH), since polyurethane is a hygroscopic material.

The cavity was 140 mm in length, 51 mm in width and had a uniform thickness of 1mm. The sprue was conical and had an average diameter of 7.5 mm and a length of 70 mm.The runner was full round with a diameter of 10 mm and length of 15 mm.

The resin was molded on a 150-ton injection molding machine (Reed, 5-ounce, TGIIseries). Fill pressure and melt temperature were changed for each trial that was tested. Filltime, pack/hold time, pack/hold pressure, mold temperature, cooling time, and back pres-sure remained constant throughout the experiment. The melt temperatures and fill pressuresfor the experiment are listed in Table 1, while the constant parameters are listed in Table 2.

The molding trials began at the lowest temperature and highest pressure. While mold-ing at a set temperature, the pressures were varied until all the pressures for that temperaturehad been tested. When the pressure was changed after one trial had been completed (at con-stant temperature), the machine was kept in cycle and the first three parts were discarded toensure proper conditions. After all of the trials had been completed at any one temperature,the machine was taken out of cycle and material was purged so that the actual melt tempera-ture could be measured with a portable thermocouple device (Atkins digital thermocouple;

Table 1. Experimental processingvariables

Melt temperatureFill pressure

(melt pressure)

238°C 59 MPa

232°C 61 MPa

224°C 61, 85, 109 MPa

215°C 85, 109, 133 MPa

207°C 133 MPa

Table 2. Experimental constants

Parameter Experimental value

Fill time 2.0 s

Pack/hold time 6.0 s

Cooling time 8.0 s

Mold temperature 43°C

Pack/hold pressure 55 MPa (melt pressure)

Back pressure 0.7 MPa (hydraulic pres.)

Flow Instabilities 115

Type J, model 396). After this, the temperature profile was changed, the molding cycle wasresumed and samples were produced until the temperatures had reached the new set-pointsand stabilized. Ten samples were kept for each trial.

This same part, sprue and runner included, was also modeled in full, (no simplifica-tions) on MOLDFLOW, an injection molding analysis software package. A three-dimen-sional multi-laminate filling analysis was performed at all of the processing temperaturesused in the actual experiment. The fill time used for this computer-aided analysis was 0.35seconds, the filling time output by the RPC (process analysis) feature on the injection mold-ing machine. That is, the actual time that it took to fill the mold, not the time that was usedfor fill time. Desmopan 386 was used for the software analysis since the 453 resin was notavailable on the database and this resin was the most similar TPU with respect to thermal,rheological, and physical properties.

In addition to the processing and software analysis equations predicting shear rate, ,shear stress, , and pressure drop, P, were solved. All three analyses were compared.

Tables pertinent to this report are displayed within the report while figures appear atthe end of the text.

RESULTS

PROCESSING

During processing it was determined that there was aminimum temperature and pressure required for fullyfilling the cavity. The experiment began at low tempera-tures, and until the temperature reached 224°C it was notpossible to fill the cavity, even when the pressure was setat 133 MPa. The temperature increased as each trial wascompleted, and it was found that at 238°C, when thematerial showed significant degradation, the minimumpressure needed to fill the cavity was 59 MPa.

It was also determined that the acceptable process-ing range of the melt temperature (was) from 224-232°Cwhile the pressure to fill ranged from 109-61 MPa in thatrange. Figure 1 displays the processing window for thin-wall molding, with respect to melt temperature and fillpressure superimposed on the processing window sug-gested by the resin supplier for conventional molding.

Additionally, it was noted that when the injectionpressure was 109 MPa or greater, the surface texture of

γτ ∆

Figure 1. Processing window (melt tempera-ture vs. fill pressure) for conventional and thin-wall injection molding of Desmopan – 453.

240

230

220

210

2000 50 100 150{MPa}

USERUSERUSERUSERUSERUSERUSERUSERUSERUSER


the part was not smooth like the surface of the mold, but rather marked with surface depres-sions, such as those found on the surface of a golf ball.

SOFTWARE ANALYSIS

The computer model indicated that the model would fill at all temperatures other than207°C, which was the lowest temperature used in the experiment. The software was capableof providing values of the required pressure for filling, Pmax, maximum shear rate, max,maximum shear stress, max, percent skin, % skin, and percent throughput, % thrpt, for eachanalysis. Table 3 displays these data for each trial.

The maximum allowable values for and for the material as well as a generic filling

pressure maximum were also provided by thesoftware package. Table 4 exhibits these values.

THEORETICAL EQUATIONS

Since traditional equations for determiningshear rate, , pressure drop, P, and shearstress, , are geometry specific, the mold wassplit into distinct regions, each of unique geome-try. The first region consisted of the sprue andthe runner. These two sections were modeled

together, as a cylinder of 7.5 mm diameter and 85 mm length. The other section was theactual cavity, which was modeled as a rectangular slit. For the shear rate calculation, thegate (also a slit) dimensions of 10 mm width and 1 mm depth were used. The followingequations were used to estimate the shear rate though each section.

Table 3. Computer aided filling results

Tmelt, oC Pmax, MPa max, MPa max, s

-1 % Skin % Thrpt

238 43 0.24 40,500 22.0 100

232 51 0.27 42,150 22.5 100

224 65 0.31 43,600 24.3 100

315 87 0.55 52,400 26.5 100

207 100 - - 38.3 runner only

γτ

τ γ

γτ

γ ∆τ

Table 4. Computer based maxima

Parameter Maximum value

max 40,000 s-1

max 0.300 MPa

Filling pressure 100 MPa

γ

τ


Cylinder

Slit

where a is the apparent shear rate (s-1), Q is the volumetric flow rate (mm3/s), d is the cyl-inder diameter (mm), w is the slit width (mm), and h is the slit height (mm).4 Since elas-tomers, as are most thermoplastics, are characterized as exhibiting pseudoplastic flow, thefollowing correction was made to the apparent shear rate:

where c is the corrected shear rate (s-1), a is the apparent shear rate (s-1), and n is thepower law index. From a plot of viscosity, , vs. shear rate, , the slope of the graph isequal to n-1. Said plot was provided by the supplier.8 This TPU has a power law index of0.35.

Once the corrected shear rates were found and the viscosity through each section wasdetermined from the same plot of vs. , the pressure drop through each section could becalculated. The following equations were used to perform this operation:

Cylinder

Slit

where P is the pressure drop (MPa), is the viscosity (Pa-s), L is the length of the section(mm), r is the radius of the cylinder (mm),4 and is the geometry factor (for a slit = 1.5).6

In the same fashion, when P was found it was possible to determine the shear stress,, developed in the cavity during filling. The shear stress developed at minimum pressure to

fill and maximum pressure available to fill, as well as the filling pressure at which is equalto the critical were determined using the following equation;

where is the shear stress (MPa), dP/dx is the pressure gradient in the cavity (MPa/mm),and h is the height of the cavity (mm).3 Table 5 demonstrates the values that were calculatedduring this analysis.

TPU FLOW TYPE

A very interesting observation noted during this experiment was the flow-profile of theTPU. A dual-plug profile was observed in most short-shot trials, rather than the traditionalsingle, centered plug profile. In parts that were not filled more than 20 mm, there was not

γa 32Q πd3⁄=

γa6Q

wh2

---------=

γ

γc γa3n 1+( )

4n--------------------=

γ γη γ

η γ

∆P 8Qη L

πr4

--------=

∆P 12QηLφ

wh3

---------=

∆ ηφ

∆τ

ττ

τ dPdx-------=

h2---

τ


enough time or space for the profile to developand the flow that was observed was due toentrance effects at the gate. That is, the materialwas fanning out. In all other shots that were notfull, this dual profile could be observed.

DISCUSSION

FILLING PRESSURE, TEMPERATURE,SHEAR RATE

The computer-aided filling analysis suggestedthat the cavity could be filled with 43 MPa at thehighest temperature, 238°C, while the general-ized equations predicted that 54 MPa would berequired. In actual processing 59 MPa was theminimum pressure needed at 238°C. However,at this temperature the resin was severelydegraded, so in actuality, 61 MPa was the mini-mum pressure needed in order to produce a goodpart.

The software also indicated that the cavitycould be filled when the melt temperature was

215°C. During processing it was found that the cavity could not be filled until the tempera-ture of the melt was 224°C.

Although the software understated the minimum pressure and temperature required forfilling, the analysis of the shear rate and shear stress were, most likely, more accurate, espe-cially when compared to the predictive equations. This is due to the fact that the softwarewas able to take into account the skin layer that forms and entrance effects at the gate.

SHEAR STRESS

During flow a shear stress distribution develops in the cavity, with the shear stress at thefreezing skin layer near the wall at a maximum value, retreating to a minimum value in thehot flowing center.5 The critical shear stress represents the value above which primarybonds in the polymer backbone can be broken during flow. If the shear force of flow is toohigh it will overcome the frictional force between the mold wall and the skin layer. This willact to tear some of the skin layer away from already frozen polymer (slip-stick phenom-ena).5 This leads to serious cosmetic defects in the molding. This effect is magnified in thin-wall molding.

Table 5. Theoretical results

Parameter Result

c (sprue) 12,410 s-1

c (gate) 17,810 s-1

(sprue) 100 Pa-s

(gate,cavity) 50 Pa-s

P (sprue) 2.3 MPa

P (cavity) 51.7 MPa

P (total) 54.0 MPa

(at min. pressure) 0.19 MPa

(at max. pressure) 0 .49 MPa

P (at = critical) 84.0 MPa

γ

γ

η

η

∆

∆

∆

τ

τ

τ τ


Upon subsequent analysis of the molded parts it was foundthat some of the samples molded at pressures greater than 109MPa demonstrated this slip-stick phenomena. These parts werecharacterized by depressions in the top and bottom surfaces,along the whole length of the part. This pressure of 109 MPa wasgreater than the 84 MPa that was predicted to induce the onset ofslippage.

To investigate this effect, the core side of the mold was sur-faced with a fine aluminum-oxide powder (vapor-honed)which acted to increase the roughness of the surface, andcoefficient of sliding friction between the polymer andthe cavity surface. Samples were then molded and ana-lyzed. It was found that the rougher surface led to a morestable flow, and a more cosmetically appealing surfaceon the molding. Thus, from a practical standpoint, it isbetter to create a mold with a rougher surface to preventslippage during filling.

TPU FLOW TYPE

Typically, fountain flow profiles are observed in thermo-plastic injection molding. This flow-profile is character-ized by parabolic velocity profiles at the melt front withrespect to the width, with the maximum melt velocity atthe center and zero velocity at either wall, as demon-strated in Figure 2.2 However, as illustrated in Figure 3,the TPU studied herein showed what appeared to be adual-plug profile with the maximum velocity of the meltat each wall retreating to a reduced velocity in the center.This observation appears to be totally unique and wasunreported in previous literature.

The implications of such unique rheology mayexplain the appearance of gas traps in the part, rather than gas burns at the corners as wouldnormally be expected for a mold without venting. In addition this flow suggests the pres-ence of a weld line running down the center of the part even though there was only one gatefor the part and there were no obstructions during flow. Both the air trap and weld line areshown in Figure 4.

Figure 2. Typical flow in a rectan-gular slit cavity.

Figure 3. Typical melt front profile experi-enced during thin-wall molding of polyure-thane elastomer.

Figure 4. Weld line and air trap location in completely filled samples.

Flow History

Gate Flow Direction

Gate Weld Line

Air Trap

FOUNTAIN FLOW

Frozen Layer

Flow Front


This behavior could represent a new challenge in mold design and processing forTPU’s. The formation of this flow-type and possible theories for its formation will continueto be researched and discussed.

CONCLUSIONS

Although there are uncertainties in with thin-wall molding of TPU’s, such as the flow pro-file, it was found that it is possible to mold TPU’s for thin-wall applications. It was alsofound that while it is possible to predict behavior during filling, the results of the predictionsshould be used only as a starting point for processing.

FUTURE WORK

Future experiments will be designed to study thin-wall injection molding of other TPU’s aswell as thermoplastic polyolefin elastomers (TPO’s).

These experiments will center on investigating the unique flow profile found in thisexperiment. Additionally, physical properties, such as tensile properties, will be determinedand related to the processing conditions at which the samples were produced.

ACKNOWLEDGEMENTS

The authors would like to extend sincere thanks to our advisor Dr. Nick R. Schott ofUMASS-Lowell, Dr. S. J. Grossman (also from Lowell), Stephen Guberski, and to Lee Plas-tics for providing the mold, machine, and material for this experiment. We would also liketo thank the following people at Lee Plastics for their assistance: Leo Montagna Jr., DanWagner, and Steve Leele.

REFERENCES1 Belcher, Don and Whetten, Alan (1994), Processing Effects for High Speed Thin Wall Injection Molding of Polyethylene Improved Processing (IP) Resins, ANTEC’94, p.5932 Kramer, Nanda M. G., How Plastics Flow Into and Within a Cavity, KONA – The News, October 1993.3 McKelvey, James M., Polymer Processing, Wiley, New York, 1962.4 Oehler, Peter R. (1996), Estimation of Machine Requirements and Process Optimization of Thinwall Injection Molding, ANTEC’96, p.5725 Schott, Nick R., Thin Wall Injection Molding: Strategies for Processing and Applications for Consumer Electronics, Lowell, 19956 Stevens, M. J., and Covas, J. A., Extruder Principles and Operation, Chapman & Hall, New York, 1995.7 Thinwall® Technical Guidebook for Electronics Applications, GE Plastics, October 1995.8 Urethane Elastomer Engineering Handbook, Miles Polymer Division, October 1992.

Pressure Loss in Thin Wall Moldings

John W. BozzelliInjection Molding SolutionsJim Cardinal, Bill Fierens

General Polymers Division, Ashland Chemical Company

INTRODUCTION

As the injection molding industry continues to mature, the development trends begin toemerge. One of the many trends today is this drive to thinner-wall parts. Designers are driv-ing more functions into parts to save assembly costs. Thinner walls are used to save weightand plastic which also reduces part costs. These trends do save costs push a molders' pro-cessing window narrower and narrower. What can the molder do to cope with this trendwith respect to machines and processing strategy?

Within processing there are four key plastic variables that define a part: • Plastic temperature • Plastic pressure • Plastic flow rate • Plastics cooling rate and time

This list is not in order of any priority and of these this paper will address only the keyvariable, plastic pressure as it relates to part complexity and thin walling. This is not to saythat plastic pressure is the most important. We are simply singling out plastic pressure withrespect to a part becoming more complex and thinner walled. The point being that both ofthese trends require more melt pressure to fill and pack the mold as well as higher clamptonnage to keep the mold closed.

This paper compares the pressure requirements to fill a rectangular cavity and pack itto normal molding pressures as the nominal wall decreases from 2.54 - 1.27 mm (0.100 -0.050 inch). The study merely reports experimental pressure readings for three different res-ins. Two polystyrenes, one crystal general purpose and one high impact, along with apolypropylene. While the general trend toward higher required pressure is well acceptedthere is a scarcity of actual data published. This data is needed to correctly specify machinesand understand the magnitude of clamp tonnage required to do thin-wall molding. With theproper machine and the correct processing strategies for thin-wall molding the molder mayincrease his processing capabilities and maximize profits.


EXPERIMENTAL

Tests were done under controlled conditions. Molding was done on a 100 metric ton Man-nesman Demag digital hydraulic molding machine with a 15.6 intensification ratio. Themold was a wedge action variable depth mold built by DME. Part dimensions were 38.1mm (1.50 inches) wide by 152.4 mm (6.00 inches) long with the nominal wall thicknessstated in the data tables. Melt temperature setpoints were at 232oC (450oF) which yieldedand actual melt temperature of 235oC (455oF). Mold temperature was set at 38oC (100oF)

Filling the part was done under velocity control. There was always abundant pressureon the pump side of the flow control valve to assure velocity control. Hydraulic pressure inthe injection cylinder varied shot to shot compensating for viscosity variations during fill.Fill was terminated at a stroke position that provided a part 99% full. Hydraulic pressurewas monitored at a 100 hertz through a strain gauge hydraulic pressure transducer. Data wasplotted and stored via computer. Hydraulic pressure at transfer was converted to plasticpressure in the nozzle via the machine's intensification ratio and reported as the pressure tofill the part 99% full.

Cavity pressures were measured at two locations along the parts center line near thegate and near the last area to fill. The measurements were made via strain gauge transducersbehind 6.04 mm (0.125 inch) ejector pins. The distance between the ejector pins was 140mm (5.50 inches). This is the distance used to calculate the pressure loss reported in the dataTables 1-3.


Tables 1 through 3 present the data in both SI and English units. Data is presented for filltimes, time to fill the parts 99%, from 1.44 seconds to 0.42 seconds, at nominal wall thick-ness' of 2.54 mm (0.100 inch) and 1.27 mm (0.050 inch). This nominal wall change repre-sents two length to thickness ratios: 60: 1 for the 2.54 mm thickness and 120:1 for the 1.27mm thickness. The highest of which barely qualifies as thin wall molding as sometimesdefined. Thin wall molding length to thickness ratios can go to 300:1 ratios.

Table 1. Pressure loss vs. nominal wall thickness and flow rate data for highimpact polystyrene ( STYRON 484); (SI units)

Nominal wall, mm

Fill time, s

Inject pressure to fill

part, MPa

Hold melt pressure,

MPa

Post gate pressure,

MPa

Last fill pressure,

MPa

Pressure loss across part, MPa

Pressure loss per cm of flow, MPa

2.54 1.44 60.9 48.6 31.8 21.4 10.4 5.0

Pressure Loss 123

2.54 0.82 65.6 35.0 31.0 22.0 9.0 4.4

2.54 0.42 76.3 31.0 31.8 22.9 8.9 4.3

1.27 1.44 95.0 60.3 49.4 17.3 32.1 15.5

1.27 0.83 89.8 49.4 41.0 15.1 25.9 12.5

1.27 0.43 93.7 37.2 40.7 19.3 21.5 10.4

Table 1A. Pressure loss vs. nominal wall thickness and flow rate data for highimpact polystyrene ( STYRON 484); (English units)

Nominal wall, in.

Fill time, s


part, psi

Hold melt pressure,

psi

Post gate pressure, psi

Last fill pressure,

psi

Pressure loss across part, psi

Pressure loss per inch of

flow, psi

0.100 1.44 8,827 7,045 4,618 3,111 1,507 287

0.100 0.82 9,513 5,075 4,501 3,195 1,306 249

0.100 0.42 11,070 4,500 4,610 3,318 1,292 246

0.050 1.44 13,784 8,744 7,161 2,511 4,650 886

0.050 0.83 13,027 7,162 5,941 2,190 3,751 714

0.050 0.43 13,588 5,392 5,905 2,793 3,112 593

Table 2. Pressure loss vs. nominal wall thickness and flow rate data forpolypropylene ( SB-823; 20 MRF); (SI units)

Nominal wall, mm

Fill time, s


part, MPa

Hold melt pressure,

MPa

Post gate pressure,

MPa

Last fill pressure,

MPa



2.54 0.82 42.1 30.3 29.0 20.6 8.4 4.1

2.54 0.44 52.8 33.5 31.9 22.7 9.2 4.5

Table 1. Pressure loss vs. nominal wall thickness and flow rate data for highimpact polystyrene ( STYRON 484); (SI units)

Nominal wall, mm

Fill time, s


part, MPa

Hold melt pressure,

MPa

Post gate pressure,

MPa

Last fill pressure,

MPa




1.27 0.90 60.2 55.5 55.3 30.9 24.4 11.8

1.27 0.40 69.8 46.5 46.8 24.4 22.3 10.8

Table 2A. Pressure loss vs. nominal wall thickness and flow rate data forpolypropylene ( SB-823; 20 MRF); (English units)

Nominal wall, in.

Fill time, s


part, psi

Hold melt pressure,

psi


Last fill pressure,

psi



flow, psi

0.100 0.82 6,100 4,400 4,200 2.985 1,215 231

0.100 0.44 7,658 4,853 4,628 3,290 1,338 255

0.050 0.90 8,726 8,053 8,025 4,485 3,540 674

0.050 0.40 10,120 6,740 6,781 3,543 3,238 617

Table 3. Pressure loss vs. nominal wall thickness and flow rate data for crys-tal general purpose polystyrene ( STYRON 685D, 2MFR); (SI units)

Nominal wall, mm

Fill time, s


part, MPa

Hold melt pressure,

MPa

Post gate pressure,

MPa

Last fill pressure,

MPa



2.54 0.42 73.9 45.2 33.6 22.4 11.2 5.4

1.27 0.43 99.6 51.7 42.1 17.0 25.1 12.1

Table 2. Pressure loss vs. nominal wall thickness and flow rate data forpolypropylene ( SB-823; 20 MRF); (SI units)

Nominal wall, mm

Fill time, s


part, MPa

Hold melt pressure,

MPa

Post gate pressure,

MPa

Last fill pressure,

MPa



Pressure Loss 125

Interpretation of the data provides some insights to requirements for thin-wall molding.First, the data indicates the thinner the nominal wall the greater the injection melt pressuresrequired to fill and pack the cavity. This is expected, however the magnitude of the pressureincrease is noteworthy. In each case the pressure loss more than doubled, in some cases ittripled, that of the thicker wall data. Higher thin-wall ratios would drive this pressure losseven higher. This brings to question the typical rules of thumb for calculating requiredclamp tonnage. The data demonstrates the need for significantly higher clamp tonnage. Nor-mal requirements typically stated for these relatively easy-flow commodity resins will beelevated to tonnage quoted for engineering grade resins. Processors need to have availableexact pressure loss data for a given distance relative a specified wall thickness so they canproperly specify machine clamp requirements.

The large pressure loss is the cause for the higher clamp pressure requirements. Tominimize pressure loss through the part, various injection rates were tried. While data is notcomplete for each resin a trend is clear: fast injection rates produce less pressure lossthrough the part. In going from the slow 1.44 second fill time to the faster 0.43 second filltime, pressure loss decreased dramatically. This provides better part uniformity which mayprovide better performance, less dimensional variances and less warp. However, fasterinjection rates do require greater melt pressures to drive the plastic into the cavity. This is acritical point in specifying thin wall molding machines. Melt pressures of 140 MPa (20,000psi) are common today but may not be sufficient to tomorrow's downsized thin wall part.

This data showing that faster fills allow for lower pressure losses provide a strategy forfilling thin-wall molds: faster fill rates produce parts with less pressure gradient from gate tolast area to fill. The molder can improve his process capability by using fast fill speeds.

SUMMARY

Data has been presented that shows the effect of reducing nominal wall thickness on thepressure requirements of molding three commodity resins. Going from a 60:1 to 120:1 flow-length to thickness ratios more than doubled the pressure loss in the cavity. Data also shows

Table 3A. Pressure loss vs. nominal wall thickness and flow rate data forcrystal general polystyrene ( STYRON 685D, 2 MFR); (English units)

Nominal wall, in.

Fill time, s


part, psi

Hold melt pressure,

psi


Last fill pressure,

psi



flow, psi

0.100 0.42 10,713 6,552 4,872 3,248 1,624 309

0.050 0.43 14,442 7,502 6,113 2,472 3,641 694


that faster injection rates provide significantly lower pressure loss within the cavity whilerequiring higher injection melt pressures. This data supports the position that for thin-wallmolding higher clamp pressures and higher melt pressure will be required of moldingmachines to control thin-wall processing. It is important that more of this type of data begenerated for both commodity and engineering grade resins so that molders and processorscan correctly specify molding equipment and processing strategies.

ACKNOWLEDGMENTS

The authors would like to thank Ashland Chemical and Dow Plastics, particularly GaryRademacher for their support of this work.

Integrating Thin Wall Molder’s Needs into PolymerManufacturing

W. G. Todd, H. K. Williams, D. L. WiseEquistar Chemicals, LP

INTRODUCTION

One of the most frustrating problems for resin manufacturers is how to relate injection-molding parameters back to manufacturing synthesis conditions and laboratory quality con-trol (QC) measurements. This article describes a unique use of existing QC-measured resinproperties to predict relative molding cycle times for high-flow polyethylene (PE) resins.The introduction of the “Isometric Spiral Flow Chart” (Figure 4) provides the basis for thisnew approach. A nomogram for optimizing injection molding melt temperatures when tran-sitioning from lot-to-lot is also presented.

DISCUSSION

Molders of rigid food packaging containers and promotional drink cups generally havewell-defined processing needs and related methods to measure process consistency andmolded part performance. Likewise, the polymer manufacturer has well-defined manufac-turing and analytical methods for characterizing resin properties and physical properties.How well a resin supplier is able to translate polymer manufacturing measurements back tothe molder’s process and the end-use applications often determines the degree of success forboth the resin supplier and the molder. Table 1 attempts to define these inter-relationshipsbetween the molder’s processing requirements and the polymer producer’s process mea-

surements. The information in Table 1 shows the injec-

tion molder can easily measure some of hisneeds and the manufacturer can relate thoseneeds through TS (Technical Service) laboratorymeasurement. In other areas, the customer doesnot have a well-defined measurement of hisneeds. The problem is further compoundedbecause even if the customer’s measurement canFigure 1. Typical cycle times versus spiral flow number.


be correlated to a measured TS lab measurement, how does the TS measurement relate to aplant QC measurement? For example, molding cycle time correlates well to TS laboratoryspiral flow measurements as illustrated in Figure 1. The Equistar spiral flow number (SFN)is the number of centimeters of flow produced when molten resin at 227oC is injected into along, spiral-channel insert (half-round 0.635 x 0.157 x 127 cm) at a constant pressure of 6.9MPa. Estimated shear rate is approximately 10,000 reciprocal seconds. The equivalentASTM Method for SFN is D3123.

* Melt Index (MI2)** Melt Flow Ratio (MI20/MI2)

Now that we have established a relationship between cycle time and SFN, how dopolyethylene resin properties influence SFN? Typically a resin manufacturer changes poly-merization catalyst systems, modifies reactor configuration or adjusts reactor-operatingparameters, such as temperature, ethylene and hydrogen concentrations, to vary molecularweight (MW) and molecular weight distribution (MWD). Melt index, MI2, is measured inthe QC lab and is used as an indication of resin molecular weight. It is defined as the num-ber of grams of polymer extruded in ten minutes as measured by ASTM Method D1238.The higher the melt index, the lower the molecular weight and melt viscosity which meansthe resin processes more easily. Melt flow ratio (MFR or MI20/MI2) is a calculated QC labnumber, which is used as an indication of MWD. It is calculated by dividing a melt indexmeasured at a high shear rate (MI20) by a melt index measured at a low shear rate (MI2). Alow MFR indicates a narrow MWD; conversely a larger number indicates a broad MWD

Table 1. Molding requirements versus resin physical properties

Molder Polymer producer

End-use requirements How measuredTech. service lab.

measurementPlant QC-

measurement

Production rates Cycle time Spiral flow MI2* and MFR**

Stacking strength Top load Flex modulus Density, MI2 and MFR

Toughness Drop impact Izod impact - ambient Density, MI2 and MFR

Cold temp. impact Drop impact Izod impact – freezer Density, MI2 and MFR

Dimensional control Shrinkage, lid fit, nesting ASTM shrinkage Density, MI2 and MFR

Warpage Visual, printability Part deformation Density, MI2 and MFR

Integrating Thin Wall Molder’s Needs 129

polymer. In general, a broader MWD resin flows easier than a narrow MWD resin at a givenmelt index. This article defines high-flow polyethylene resins as those resins with melt indi-ces above 20 and MFR’s between 20 and 40. Figure 2 plots commercially available high-flow resins as functions of melt index and MFR. Individual resins are identified by theirMI 2.

Regression of Equistar laboratory spiral flow data for the resins shown in Figure 2resulted in the correlation shown in Figure 3, which is defined by the following equation:

SFN = 10.44 + 1.016 x Sqrt (MI20) [1]Remembering that

MI20 = MI2 x MFR [2]

Isometric spiral flow values were superimposed on Figure 2, as shown in Figure 4.With this easy-to-read chart, one can rapidly determine how one resin performs versusanother one with regard to cycle time. The effect of melt index and MFR on spiral flow is

Figure 2. Commercially available high flow resins. Figure 3. Spiral flow number versus MI20 correlation.

Figure 4. Isometric spiral flow chart commercially available high flow resins.

Melt Flow Ratio MI20/MI2

Mel

t In

dex

, g/1

0 m

in

140

120

100

80

60

40

20

0

18 20 22 24 26 28 30 32 34 36 38

SFN, cm

70

65

60

55

5045403530


very graphic. Molders can easily determine which resins satisfy their required cycle times.An equally valuable tool would be a chart that predicts how a given resin fills a particularmold. To accomplish this, the isometric SFN values in Figure 4 were converted into moldaspect ratios as shown in Figure 5. The aspect ratio of a mold is calculated by dividing thelength of melt flow by the average wall thickness of the part. Using Figure 5, the customerand resin manufacturer can easily determine the range of MI2s and MFRs that will fill agiven mold.

Recently, Equistar established spiral flow specifications for all high-flow HDPE resinsand began reporting the spiral flow number for each lot on the shipping Certificate of Anal-ysis (COA). The customer can compare the spiral flow of an incoming lot of resin with the

Figure 5. Isometric aspect ratio commercially available high flow resins.

Figure 6. Melt viscosity as function of spiral flow and melt temperature.

Melt Temperature, C

Sp

iral

Flo

w n

um

ber

, cm

210 230 250 270 290 310 330

65

60

55

50

45

40

35

30

25

20

Lines of Constant Melt Viscocity

Melt Flow Ratio, MI20/MI2

18 20 22 24 26 28 30 32 34 36 38

160

140

120

100

80

60

40

20

0

Mel

t in

dex

, g/1

0 m

in

AspectRatio

440

400

360

320280240200

Integrating Thin Wall Molder’s Needs 131

spiral flow of the lot on-hand and readily estimate how the new lot will process relative tothe lot currently in production. For example, if the lot currently being run has a SFN of 50cm and the new lot has a SFN of 55 cm, the new lot should process at an approximately 10%faster rate.

To further aid the customer in adjusting his or her processing conditions, Figure 6, Iso-metric Melt Viscosity Chart as a function of SFN and melt temperature, was developed fromlaboratory, capillary Rheometer, melt viscosity data. For a given SFN and melt temperature,a point can be located on a constant melt viscosity line. By tracing along this viscosity lineto the new incoming lot’s SFN, the required melt temperature to compensate for the differ-ence in SFN's between the two lots can be read. Adjusting the melt temperature to compen-sate for the difference in SFNs minimizes changes in cycle time.

To simplify this compensation process, the special nomogram shown in Figure 7 wasdeveloped. The left-hand vertical line represents the SFN of a given lot of resin and the rightvertical line is the extruder melt temperature used to process the resin. The center verticalline is a Polymer Melt Viscosity Index, which is a relative scale from 0 percent to 100 per-cent of the melt viscosities used to develop the nomogram. To use the nomogram, the cus-tomer draws a straight line between the SFN of the lot currently being run and the extrudermelt temperature. This locates a fixed point on the center Melt Viscosity Index. The line isthen rotated about this fixed Melt Viscosity Index point to the new incoming resin lot’sSFN. The recommended new extruder melt temperature is read from the right side PolymerMelt Temperature Line.

For example, if a molder was transitioning from a resin with a SFN of 50 cm at a melttemperature of 230oC to a resin with a SFN of 45 cm, the melt temperature required to

Figure 7. Spiral flow temperature adjustment nomogram.

Sp

iral

Flo

w N

um

ber

, cm

Mel

t T

emp

erat

ure

, C

32.5

35.0

37.5

40.0

42.5

45.0

47.5

50.0

52.5

55.057.560.062.565.067.570.072.5

100

80

60

0

180

190

200

210

220

230

240

250

260

270

280290300310320330340

A

D C

B

E

20

40


maintain the same cyclic time would determined by: 1) drawing a straight line (solid line)between Point A, SFN of 50 cm, and Point B, 230oC melt temperature; 2) locating Point C,50%, on the Viscosity Index line; 3) extend a straight line (dashed line) from Point D, newSFN of 45 cm, through Point C to Point E, which gives a new melt temperature of 251oC.

SUMMARY

The use of SFN’s to describe the flow properties of a given resin in thin wall injection moldsuniquely combines the effect of both the resin’s MI2 and MWD. In addition, superimposingisometric SFN’s and Aspect Ratios onto resin product maps graphically depicts how oneresin will perform in a given mold versus another resin. Furthermore, the thin wall moldercan minimize transition losses between resin lots and between resin grades by using the Iso-metric Melt Viscosity Chart or the associated Nomogram to adjust injection melt tempera-tures.

ACKNOWLEDGEMENTS

The authors would like to acknowledge the following Equistar Chemicals associates whodeveloped the laboratory data used to generate the correlations presented in this article: JeanMerrick-Mack, Scott Nolan, Charlie Smith, Kirby Perry, Jim Hale and Mark Gregurek.

Thinning Injection Molded Computer Walls

Lee Hornberger and Ken LownSanta Clara University, Santa Clara, CA, USA

BACKGROUND

Several new plastic materials and processes have been developed in the last few years whichfacilitate the production of high quality "thin" walled plastic parts. These new technologieshave enabled the production of cosmetic injection molded parts with wall thicknesses lessthan 1 mm.1,2 This form of molding has been readily adopted by the cellular phone and por-table computer industry because thin walled parts provide valuable product weight and costsavings. But, designers of larger parts such desktop or workstation computer housings havehad little interest in thinning the walls of their products and have continued to design theirparts with wall thickness in the range of 3 mm. However, the current market trend of reduc-ing computer cost and the environmental impact of large quantities of plastic materialmakes it worth exploring the potential material savings of thinner walled products.

REQUIREMENTS FOR COMPUTER HOUSINGS

Typical computer housings such as those madeby Sun Microsystems are made from two to fourparts each in the order of 400 mm long by 400mm wide. A typical computer housing paneldesigned by Sun is illustrated in Figure 1.

Housing panels such as this one must bestructurally sound, cosmetically attractive, resis-tant to weathering and meet flammabilityrequirements for office equipment. They mustalso be easy to mold consistently for a price wellbelow $10 with volumes in the order of 100,000parts. Consequently, most computer housingsare made from engineering resins such as flameretardant ABS, PC/ABS blends or Polycarbonate

which meet these specifications at reasonable cost.

Figure 1. Sun’s housing panel.


DESIGNING THINNER WALLED HOUSINGS

Of the many requirements for computer housings three typically drive the design: structuralproperties, cosmetics and cost. Thinning the nominal wall effects all three of these areas.

The first and minimum structural requirement for a housing is stiffness. The stiffnessof the computer is a measure of its ability to withstand the loads it is subjected to without anoticeable change in shape. The loads on a housing are varied. They may be due to theweight of a monitor or spring loads from EMI shields or even customer abuse so the hous-ing must have some minimum stiffness in all directions.

The stiffness of the housing is a combination of its geometric stiffness, measured by itsmoment of inertia, I, and its material stiffness, measured by the modulus of elasticity, E, ofits plastic. The overall stiffness of a box can be quantified by its "EI" product. The higherthis product the stiffer the housing. Thinning the housing wall dramatically decreases itsmoment of inertia which is proportional to the cube of wall thickness. The designer cancompensate, somewhat, for a loss in geometric stiffness by changing to a stiffer material.

Another important structural property of a housing is its load bearing capacity. This isof particular importance for detailed design features such as snaps and bosses. The load afeature can bear is proportional to its material yield strength and cross-sectional area. Con-sequently, thinning feature walls reduces the load they can tolerate and must be compen-sated for by changing to stronger materials.

Impact strength is another important structural property in computer housings as theyare often subjected to intentional or unintentional physical blows from their human masters.Designers enhance the ruggedness of the boxes by selecting materials with high impactstrengths. Most office computers and telephones are made from materials with notchedimpact strengths of 160 J/m (3 ft-lbf) or greater and this seems a reasonable range for hous-ing materials. Decreasing the wall thickness of a housing does not directly change its impactstrength. However, impact strength may be lost when stiffer, stronger materials are substi-tuted in these applications.

The cosmetics or appearance of a housing may also be affected by thinning its nominalwall if it increases the occurrence of sinkmarks or weldlines. In conventional 3 mm design,the wall thickness of internal ribs and bosses are specified as 60% of the nominal wall inorder to minimize visual sink. Using this same approach in the design of thinner walledparts yields much weaker ribs and bosses which may not support their loads. Designers maythen be forced to increase wall thickness and trade sinkmarks for strength. However, it isreported that 1 mm thick panels made from glass filled materials can have rib widths 100%of the nominal wall without sink.3

Another cosmetic issue impacted by the use of a thinner walls in housings is weldlines.Weldlines occur when melt fronts meet within the part. Normally, designers minimize weld-

Thinning Injection Molded Computer Walls 135

lines by limiting the number and location of gates needed to fill a part. Housing tool design-ers typically use one or two direct gates to mold these parts in three millimeters. However,more gates are needed to fill thinner walled parts as the thinner walls constrict the flow ofmaterial and accelerate the cooling of the material. The final, and most important design element which is affected by thinning the nominalwall is the manufacturing cost of the part. This cost is made of three components: materialcost, processing cost, and tooling cost. Thinning the wall reduces the reduces material costby decreasing the volume of material required to fill the part. It decreases process cost byreducing cooling and injection time. Tooling cost, however, are reported to increase whenmolds are designed for thinner walled parts1,3-5 as more steel is required to resist the higherpressures needed to push these materials into the mold. In addition, if glass filled materialsare used for these designs, harder, more expensive steels must be used due to resist the abra-sive nature of these materials.3

TRADE-OFF STUDY METHODOLOGY

Manufacturers of computer housings such as Sun Microsystems, have been reluctant to usethinner walls in their products because the benefits relative to the apparent risk have beenunclear. As part of Sun’s study of thin wall design, the authors designed a Sun housing panelwith 1, 2 and 3 mm walls and evaluated these panels relative to their resultant structure, cos-metics and cost. The panel analyzed was that sketched in Figure 1.

The structural properties of the three panels were compared by analyzing their relativestiffness, flexural strength and impact strength. In order to do this, the geometric stiffness ofeach of the three panels was calculated from its wall thickness and the width of its largestcross-section. For comparison purposes, this value was calculated with and without the sixribs that straddled the part. The modulus, E, used to evaluate the EI product varied for eachas different materials were recommended by General Electric and industry standards foreach wall thickness.2,3 For the 3 mm wall a flame retardant KJB ABS was chosen as it iscommonly used in these applications. For the 2 mm design, a PC/ABS high flow 2950Cycoloy3 was recommended as it provided acceptable flammability resistance at this thick-ness, and high flow at the high pressures and temperatures used in this type of molding. Italso has increased stiffness, impact and yield strength over ABS. For the 1 mm panel a highmodulus Lexan SP76043 was recommended as by GE, Shieldmate and Apple for its struc-tural value and cosmetics in this wall thickness.

The relative load bearing capacity of the three panels was compared by evaluating theirperformance in a standard cantilever snap feature under load. The snap feature was modeledby a 25.4 mm long by 12.7 mm wide cantilever beam clamped at one end with a concen-


trated load applied at the free end. The maximum load, P, the beam can withstand withoutyielding was calculated from the standard cantilever beam formula:

P= Sy wt2 / (6l) where:

Sy tensile yield strength of the plastic t wall thickness of the panel l length of cantilever (25.4 mm) w width of cantilever (12.7 mm)

The impact strength of the panels was assumed to be equal to their notched Izod impactstrengths which are shown in Table 1. These values were listed on the data sheets for thesematerials.

To evaluate the relative cosmetics of the panels, the authors estimated the number ofgates needed to fill each part based on recommended L/T ratios for thin wall molding byGeneral Electric.3 To minimize sinksmarks, the parts were designed so that the rib to wallthickness ratio in the 1 mm parts were 1:1 and in the 2 and 3 mm parts were 0.6:1 (60%) asrecommended.3

Manufacturing cost were calculated from material, processing and tooling cost throughthe following relationships:

Part Cost = Mp+ Pp + Tpwhere:

Mp Material cost per part Pp Processing cost per part

Table 1. Properties of plastic materials

Property Lexan SP 7604 Cycoloy 2950 Cycolac KJB ABS

Modulus, E, GPa 6.08 2.58 2.27

Yield strength, Sy, MPa 74 64 38

Impact strength, notched, J/m

53 458 213

Density, 10-6 kg/mm3 1.36 1.18 1.22

Market price, $/kg 11.86 8.87 6.47


Tp Tooling cost per part Material cost, Mp, were calculated from the following equation and the material propertiesin Table 1:

Mp=VpxρmxCm where:

Vp the total part volume ρm the weight density of the material (kg/mm3) Cm cost of the material per kg in custom colors.

The processing cost per part was calculated from the molding cycle time and assumedmolding machine rates. Secondary cost were not included in this evaluation. Cycle time wasestimated using the methods of Poli and Dixon6 which calculates cycle time from consider-ations of wall thickness, part complexity and required surface finish.

Injection machine cost were assumed to be $75/hour for a standard 3 mm wall part,$100/hour for a 2 mm wall and $120/hour for a 1 mm wall. Molding thinner walls requireshigher injection pressures and more expensive higher tonnage injection molding machines.In addition, 1 mm parts need specialized machines with smaller barrels and programmedinjection to aid flow and minimize degradation of these materials.3 Rates were chosen fromthe authors experience in California area.

The tooling cost per part was derived from total tool cost distributed over 100,000parts. Tooling cost was estimated with Poli and Dixon’s methodology.7 In this technique thecomplexity of the part due to features such as ribs, holes and bosses as well as its size andtexture requirements are quantified relative to a baseline part (a simple plastic washer). Theresultant complexity factor becomes a multiplier of the known cost of the washer tooling.

The baseline washer tool was specified by Dixon and Poli’s as requiring 200 hours8 ofmachining and approximately $1000 of material. To estimate the washer tooling cost forthis study, a machining cost of $75/hour was used which resulted in a total tool cost of$16,000.

A 30% surcharge was added to the estimated cost of the 1 mm tool as this was thereported burden for increasing the strength of the mold for high pressure molding, increas-ing its hardness for the high materials and for the increased number of ejector pins needed toremove the fragile part.3-5

RESULTS

The calculated structural properties, cosmetics and cost for the three designs are dis-played in Tables 2, 3 and 4. From Table 2 it is obvious that the loss of geometric rigidity, I,with decreasing wall thickness is difficult to compensate for even with a glass filled poly-carbonate which has a modulus more than 2.5 times that of the ABS. Even with the higher


modulus material, the net stiffness, EI, for the 1 mm walled part was 10 times less than thatof the original 3 mm walled part in both the unribbed and ribbed configurations.

The loss in stiffness, however, in going to a 2 mm from a 3 mm wall was fairly minor.Impact strength actually improved when Cycoloy replaced ABS in the 2 mm version of thepart. The reverse of this result occurred when the glass-filled polycarbonate was used in the1 mm part. Here, a large loss in impact strength was the price paid for the gain in stiffness.

The load bearing capacity of the model snap feature (reported in Table 2) was dimin-ished by 25% in the 2 mm part and 75% in the 1 mm part relative to the 3 mm version. This

Table 2. Structural properties of housing panels

Wall thickness, mm 1 2 3

Stiffness, I, mm4 (6 ribs) 110 460 1362

Stiffness, I, mm4 (no ribs) 39 312 1053

Stiffness, EI, GPa-mm4 (ribbed) 669 1187 3097

Snap load capacity, Newtons 7.3 21.3 28.38

Table 3. Cosmetics of housing panels


Flow length, mm 90-150 180-300 30

Minimum gates 3 2 1

Minimum weldlines 2 1 0

Table 4. Manufacturing cost of housing panels


Material cost, $ 3.55 4.60 4.83

Processing cost, $ 1.60 1.75 1.61

Tooling cost, $ 1.26 0.97 0.97

Total part cost, $ 6.41 7.32 7.41


loss in strength and impact coupled with an increase in stiffness of the snap severely limitsits allowable deflection and utility in a 1 mm part. The decrease in load capacity in the 2 mmnominal wall design is not as severe and is somewhat offset by the lower modulus andhigher impact strength of the PC/ABS.

The cosmetics of the part also suffer as its wall is thinned to 1 mm. As evidenced inTable 3, the flow length predictions for the 1 mm thick walls are nearly half of that for the 2and 3 mm walls and this increases the number of gates required to fill this part. The threegates and two weldlines estimated to fill the 1 mm walled part listed in Table 3 are probablya low estimate as this calculation did not fully account for the poor flow of the glass filledmaterial

It is apparent from the data in Table 4 that the most beneficial asset of molding a 1 mmpart is the decrease in part cost. The estimated total manufacturing cost for the 1 mm thickversion of the part is a dollar less than that of its 3 mm counterpart. This 14% savings comesfrom the decrease in material needed to fill the part.

Additional cost savings resulted from a decrease in cycle time for the thinner partsCycle time for the 1 mm part was estimated, in this study, to be 48 seconds for the 1 mm, 63seconds for the 2 mm and 77 seconds for the 3 mm. Cycle time savings, however, were off-set by the increased molding machine cost and tooling for the thinner walled parts.

No unique modifications had to be made to the 3 mm tool to produce 2 mm partsassuming high quality steel and conservative tooling design was used for 3 mm walls. Con-sequently, the cost for 2 mm tool was identical to that of 3 mm. Although there were savingson material volume and cycle time, the net cost of a 2 mm made from the PC/ABS blendwas only 1.5% less than the original 3 mm part made from FR ABS. If FR ABS is accept-able for the 2 mm part than the manufacturing cost can be reduced by 20% to a total of$5.41.

REALITY CHECK

A number of companies like Sun are in the process of evaluating the potential of thin wallmolding and their experiences appear to support and supplement the quantitative data of thisanalysis.

Shieldmate has had extensive experience molding small thin walled parts and foundthat their cycle times were significantly lowered. They were able to mold small 2 mm partsin 30 seconds. This is nearly half of the value estimated in this study but their parts were lessthan 150 mm in overall length and width. Shieldmate has also molded 1 mm parts 200 mm x50 mm with a single gate but have found that they needed high quality tools and moreexpensive customized molding machines to produce quality parts.1


Ron Lenox of Triquest11 also reported cycle times in the order of 35-45 seconds for 1.5mm wall thickness parts which were roughly 200x280 mm He was able to use PC/ABS inthis application but noted a significant increase in tooling cost.

Apple successfully molded 1.5 and 2 mm walls in their laptop products without paint-ing but they found that they needed glass fillers in their material to maintain the structuralrigidity of the housing. It also increased their tooling cost. This result paralleled the findingsof this report.9

Eastman Kodak has experimented with thinning the walls on large plastic parts (greaterthan 300x 600 mm).10 They saw a decrease in cycle time between 30 and 40% when theydecreased their wall thickness from 3 to 2.5 mm. However, they noted significant manufac-turing yield problems due to warp and burning when they made parts below 3 mm thick.Kodak predicts that these losses would be diminished with molding machines with pro-grammed injection and plastic raw material with tightly controlled specifications.

ADDITIONAL TRADE-OFFS

There are other issues that are equally important in the design decision relative to thin wallmolding that are more difficult to quantify than the three design areas discussed in thispaper. Warpage of parts is one of these. As reported in the Kodak test, decreasing the wallthickness may increase warpage of large parts by altering its cooling. Glass fillers in thinwall materials may also change the way in which the part warps.

Manufacturing yield during molding was another factor difficult to quantify in thisstudy. Whenever, a new technology is implemented in manufacturing there are always star-tup problems and these lower production yields. As reported by Kodak, this is also the casewith thin walled parts.

The degradation of the mechanical properties of the material in thin wall molding is athird issue of concern. In order to lower the viscosity of the plastic during molding andincrease its flow through thin walls the molders generally increases the melt temperature ofthe material and raise injection velocity. Pushing these process variables to their limits canlead to degradation of the material and the loss of its stiffness and impact strength.3

SUMMARY & CONCLUSIONS

Clearly, there is a large potential for cost savings by thinning the walls of a plastic part.According to the costing model used in this study a savings between 14 and 20% is possibleon large computer housing parts produced in the range of 100,000 parts. Currently designersand molders are shying away from this approach because these savings do not appear to off-set the loss of stiffness, impact strength, and load bearing capacity shown in this paper.


However, the movement toward thinning the walls in larger housings appears to beinevitable as the benefits in cost are so attractive. The steps in this direction may be verygradual. Designers will slim the walls by tenths of millimeters rather than by whole milli-meters. This gradual approach which enables continuous improvements and cost savingsworked well in thinning smaller parts such as cellular phones and cameras10 and may workwell in the computer industry.

Important for the implementation of thin walls in computer housings is the develop-ment of grades of materials that can overcome the limitations discussed in this paper. Mate-rials are needed with high stiffness, impact, strength, and flow properties which can beconsistently produced. As was evidenced in the trade-off analysis of this study, it will be thedevelopment of these materials rather than tooling or molding machines which will drivethis technology.

REFERENCES1 Matecki, S. Shieldmate, Personal conversation.2 Fassett, J., "Thin Wall Molding: Differences in Processing over Standard Injection Molding," Proceedings of

ANTEC ‘96, Indianapolis, May 1996, p 430. 3 "Thinwall: Technical Guide for Electronics Applications," GE Plastics Technical Bulletin. 4 Marques, R, SPM, Personal conversation. 5 Rowe, D., Apple Computer, Personal conversation. 6 Dixon, John R. and Poli, Corrado, Engineering Design and Design for Manufacturing, Field Stone Publishers , Conway, Mass., 1995. p15-1 and 11- 1NBC-575, October 1995. P.7-1 7 Ibid, p.11-1. 8 Ibid, p.15-2. 9 Sabi Khan, Apple Computer, Personal conversation. 10 Bill Moncha, Eastman Kodak, Personal conversation. 11 Ron Lenox, Triquest, Personal conversation.

Chapter 3: Molding Micro Parts and Micro StructuresTranscription of Small Surface Structures in

Injection Molding – an Experimental Study

Uffe R. Arlø, Erik M. KjærDanish Polymer Centre, Technical University of Denmark

THEORY

PHYSICAL MODEL (HYPOTHESIS)

During filling of the cavity the plastic melt is transported from the center towards the coldmold wall by fountain flow where the melt solidifies. At the time the melt reaches the moldwall only a low pressure is present (the pressure in the melt front is atmospheric). Later inthe process the frozen layer at the surface grows and pressure is built up.

Ideal conditions for a good surface transcription would be• Plastic material with a high temperature (a thin frozen layer)• Large forces to press the plastic material around the surface structures, i.e. high pres-

sure.However, these two criteria are not met at the same time. As described above the plas-

tic frozen layer grows with time and so does the pressure. Therefore surface transcriptioncan be regarded as a compromise between temperature and pressure. It is hypothesized thatsurface transcription is formed by three mechanisms:• An immediate formation where the molten plastic flows in the surface valleys and over

the peaks as it flows over the surface. This mechanism takes place during the fillingstage. This mechanism will be denoted flow formation.

• A formation that takes place after the molten plastic has passed over the surface duringthe subsequent pressure build-up. This formation could be regarded as a thermoform-ing-like deformation of the frozen layer around the surface structures. The mechanismis predominant in the post-filling stage. This mechanism will be denoted press forma-tion.

• A modification of the surface structure due to thermal effects such as shrinkage andstress relaxation as the heat from the plastic parts core is transported through the sur-face. This mechanism will be denoted thermal modification.The first two mechanisms are the basis for this study.


PROCESS PARAMETERS

In practice the hypothesized physical phenomena are controlled by process parameters onthe injection molding machine and equipment. A parameter space for the process could con-sist of• Melt temperature (controlled by barrel heater bands)• Mould temperature (controlled by coolant flow and/or temperature)• Injection rate• Switch over point• Holding pressure• Holding pressure time• Cooling time• Other parameter controlling plastification and mold opening.

In agreement with the aim of studying the mechanisms described above, temperatureand (holding) pressure are selected. The temperature of interest is of course the temperatureof the plastic, where the melt temperature as set on the machine and the mold temperaturecorresponds to the initial value and boundary condition of the heat transfer problem. Inother words both melt temperature and mold temperature are means of controlling thedevelopment of temperature in the plastic material during the process. For this study themelt temperature is selected. Further the injection rate is included influencing both the tem-perature of the plastic (due to friction and time for cooling) and the pressure gradient.

EXPERIMENTAL PROCEDURE

PROCESS EQUIPMENT AND MATERIAL

The experiment was carried out with a 2 mm thick rulertype part with an edge gate and a cold runner in a twoplate mold (see Figure 1). Opposite the gate end, the cav-ity had a band with a rough spark eroded surface. Both C-MOLD simulation and the physical parts show that themelt front almost forms a straight line as is passes therough surface band hence assuring uniform orientationover the band. The part was made in polystyrene BASF143E and produced in an Engel 135 injection moldingmachine.

Figure 1. Part and runner (fill pattern C-MOLD).

Transcription of Small Surface Structures 145

DESIGN OF EXPERIMENT

I order to investigate the two mechanisms the design of the experiment was divided into twosets, one to study the flow formation and one to study the press formation. In the first setinjection rate was varied over three levels and the melt temperature over two levels whileholding pressure was kept at a minimum (as close to zero as the machine allowed). In thesecond set the holding pressure was varied over three levels and the melt temperature overtwo levels while injection rate was set at maximum. The experiments were carried out withfull factorial variation. Process parameters are varied over a range corresponding to process-ing intervals recommended by the material supplier.

SURFACE CHARACTERIZATION

The surface roughness was measured on a Talysurf Surtronic 2D mechanical stylus instru-ment. As roughness parameter the Ra value was selected due to its wide spread application.The application of stylus characterization for comparing mold surfaces with plastic part sur-faces suffers an inherent error in that valleys on the mold surface transcribes into peaks onthe plastic part surface. Due to its physical dimension the stylus is unable to register deepvalleys correctly while is capable of registering high peaks. This problem can be overcomeby making a silicone replica of the mold surface and using this for comparison with theplastic part. However, since the aim of this study is to compare the plastic parts producedunder different process conditions with each other such a replication has not been carriedout.

The rough mold surface has an Ra value of 12.6 m.As a supplement to the stylus measurement scanning electron microscope images has

been processed for the mold wall and selected parts.

RESULTS

The experiment reveals that surface tran-scription consists of both the hypothe-sized flow formation and press formation(see Figures 2 and 3). From the first set ofexperiments where holding pressure is setto a minimum it can be seen that the tran-scription is improved progressively withhigher injection rate. This can be contrib-uted to the fact that high injection ratesreduces time for cooling of the melt and/or the fact that high injection rates can

µ

Figure 2. Effects during filling.


induce higher pressure gradients. In thesecond set of experiments where pressformation is investigated it is evidentthat higher pressure up to a certain pointimproves transcription. It is surprisingthat the transcription quality falls at thehighest level of holding pressure. Thisphenomenon is observed for both tem-peratures and cannot be explained bymeasurement uncertainty of processvariance.

In both sets of experiments theincreased temperature improves transcriptionquality. Higher temperatures are observed toshift the roughness upwards. The shift is morepronounced for the experiments with minimumholding pressure than for the experiments withholding pressure suggesting that the flow for-mation mechanism is more sensitive to (the ini-tial) melt temperature than the press formationis.

Recently a study1 at The Technical Univer-sity of Denmark concerning transcription ofspecific micro structures on the surfacesshowed results supporting some of these find-ings.

By comparing SEM images of the plasticparts produced with “max. setting” (high tem-perature, high injection rate, high holding pres-sure) and “min. settings” (low temperature, lowinjection rate, low holding pressure) it is obvi-ous that the “max. setting” part has a greaterroughness than the “min. setting” sample (seeFigures 4, 5 and 6).

Figure 3. Effects during post-filling.

Figure 4. Rough spark eroded mold surface, SEM image.

Figure 5. Plastic part surface “max. settings”, SEM image.

Transcription of Small Surface Structures 147

CONCLUSIONS

Surface transcription in injection molding repre-sents complex mechanisms and depends onpressure, temperature and the viscoelastic prop-erties of the plastic. Two mechanisms have beenidentified: Flow formation and press formation.Generally increased temperature results in bettersurface transcription, while increased pressureleads to better surface transcription up to a cer-tain (high) level of pressure. Surprisingly thetranscription pressure dependence shows a localmaximum and the transcription quality drops forthe highest level of pressure.

ACKNOWLEDGEMENTS

The authors wish to thank Lotte Due Teilade, Ph.D. for a critical review of this paper.

REFERENCES

1 Sørensen, Johansen: Micro injection molding, 1999, Technical University of Denmark.

Figure 6. Plastic part surface “min. settings”, SEM image.

Injection Molding of Sub- µmGrating Optical Elements

R. Wimberger-FriedlPhilips Research Laboratories, Eindhoven, The Netherlands

INTRODUCTION

Injection molding of optical components has become a high volume business mainly due tothe enormous success of optical recording. The requirements with respect to the optical per-formance are very stringent for all parts in the optical light path, viz., the recording substrateand a number of components in the optical pick up. This requires a very accurate shape rep-lication and low optical anisotropy as induced by the molding related stresses.

Microstructure replication is important for both the media as well as components in theplayer. The information on the substrate is present in the form of small pit structures or asgrooves in the case of recordable media. The pits are of sub-micron dimension but rathershallow. Still the replication is challenging for high recording density substrates like theDVD because of the large area and the thin substrate which makes filling difficult. In theoptical pick up of the player also diffractive optical elements are used for instance for split-ting of the main laser beam into three beams for track following but also for the deflectionof the returning beam onto the detector area. Such diffractive gratings can be produced inhigh volume by injection molding already. In certain players an optical element with l/4retardation is necessary for optimized beam splitting. Such elements are usually made frompolished quartz laminates which are intrinsically expensive to produce. It is known thatgrating structures with a pitch considerably smaller than the wavelength of light do behaveas retardation elements due to a so-called form birefringence effect which makes the effec-tive refractive index in the structure depend on the direction of polarization.1 By injectionmolding such structures one could replace the expensive quartz element by a cost effectiveplastic component.

In the following the design, fabrication of structured mold surfaces and the results ofthe replication by straight injection molding will be presented.


GRATING DESIGNThe grating geometry was designed with the aid of a numerical package called Cyclopdeveloped at our laboratory by P. Urbach.2 This is an exact treatment of the problem bysolving the Maxwell equations in a finite element representation of the physical gratinggeometry. The simplest geometry is a rectangular grating with pitch p, height h and dutycycle dc (polymer to air ratio). In Figure 1 the effect of the pitch is demonstrated for theretardation and the intensity of the emerging 0th order beams with polarization parallel andperpendicular to the grating direction Te and Tm, respectively. As can be seen for the indi-cated wavelength of 785 nm and refractive index of Polycarbonate, above a pitch of 0.5 mthe intensities drop quickly due to upcoming higher order diffraction. Below the retardationincreases strongly with pitch. Therefore a pitch just below 0.5 seems the best choice. In Fig-ure 2 the effect of the grating depth is illustrated. As expected, the retardation increases lin-early with depth. The required depth for a retardation of /4 is 1.56 m. One can also seethat the transmission oscillates with depth. This means that such grating structures haveanti-reflective properties due to interference. Higher values of transmission could beachieved with non-rectangular shapes at the expense of efficiency in retardation. The thirddesign parameter is the duty cycle. Simulation shows an asymmetry in the behavior with amaximum retardation at dc = 0.6. The resulting design will have an aspect ratio of > 5! Thisis clearly a challenge for the insert production as well as the replication.

One possibility to facilitate manufacture is to split the function in two /8 gratingstructures for the top and bottom surface, respectively. In that case the depth is reduced toabout one half of the value shown above.

µ

λ µ

λ

Figure 1. Calculated effect of the pitch variation on intensity and retardation of transmitted 0th

-order beams.

Figure 2. Calculated effect of structure depth on retardation and intensity of transmitted 0th-order beams.

pitch (microns)

retardation Int Te IntTmretardation Int Te IntTm

depth (microns)0.00 0.10 0.20 0.30 0.40 0.50 0.600.00 0.50 1.00 1.50 2.00

785 nm; n=1.573dc= 0.6; h= 1560nm

reta

rdat

ion

[deg

]

1 tra

nsm

1 tra

nsm

reta

rdat

ion

[deg

]

785nm; n=1.573dc=0.6; p= 480nm

100

95

90

85

80

80

95

90

85

80 85

90

95

100100120

100

60

40

20

0

Injection Molding of Optical Elements 151

EXPERIMENTAL

OPTICAL INSERT MANUFACTURE

Several techniques were used to obtain the structures. The basic step in obtaining suchstructures is E-beam lithography. The E-beam was used either for making a lithographicmask for the wafer stepper or an in-situ mask for consecutive etching steps or to write thedesired structure directly in a thick photo-resist layer. The resist itself is not consideredstrong enough to be used in a mold. Therefore the structures first were replicated in a differ-ent material. One obvious way of doing this is by electroplating Ni. In the case the photore-sist is only used as etching mask the structures were obtained by etching a 80 nm thick Crlayer. The Cr structure then served as an etching mask for the reactive-ion etching (RIE) inSiO2.3 The fused quartz structures can then be used directly as mold insert or replicated firstin electroplated Ni as well. The resist was Success ST 3, a DUV resist of BASF. 0.8 mmthick resist structures were also obtained from DUV exposure with a wafer stepper ASMLPAS 5500/90 from a mask written with E-beam.

MOLDING

The structures as obtained on typically 2 mm thick sub-strates were attached to steel rods. The shape of the cav-ity which determines the product shape is sketched inFigure 3. The diameter of the area containing the gratingstructure was 5 mm in the largest case, the thickness typ-ically 1 mm. The molding experiments were carried outon a 35 tons Engel machine with 20 mm screw at PhilipsPMF in Eindhoven. The polymer was Makrolon CD2000 (polycarbonate of Bayer AG, FRG). Its thermo-rheological properties are described elsewhere.4 Themolding parameters, viz. the injection rate and tempera-ture, mold temperature and packing pressure, were varied

in a limited range only. In contrast to other research groups5 we did not use thermal cyclingof the mold but employed a constant mold temperature in a conventional molding process.

RESULTS

INSERTS

Three different inserts were used for molding, i.e., RI-etched quartz with a depth of 1.6 and0.8 m and electroplated Ni with a depth of 0.8 m. The etching of quartz in a depth of 1.6

m turned out to be very difficult. The trenches either were not open completely or theµ µ

µ

Figure 3. Schematic mold construction.


walls tumbled. The processing window is not broad enough forsafe industrial production. The structures of 0.8 m were lesscritical although one always needs to etch an array of structureswith steps in etching time due bad predictability of the etchdepth. The structures written by E-beam into thick resist alwaysshowed a negative slope and were therefore not used for replica-tion. Instead DUV lithography was used. Ni replica’s wereobtained from the resist structures for use in the mold. Figure 4shows SEM micrographs of fractured cross-sections of the insertstructures obtained by different techniques, as indicated.

MOLDING

For the replication of microstructures by a thermoplastic materialthe contact temperature is of major importance. The polymermelt has to flow into the channels of sub-micron height. The heattransfer to the walls occurs within a microsecond. The filling

would have to occur considerably faster in orderto prevent freeze-off. Only when the contacttemperature is above the no-flow temperatureenough time is left for complete filling. The con-tact temperature with a given polymer melt isonly determined by the mold temperature andthe thermal properties of the mold material. Theeffect of mold temperature is demonstratedclearly in Figure 5, where the depth of the repli-cated grating is plotted as a function of the moldtemperature for the case of a Ni insert with 0.8

m. Only at a mold temperature of 152oC,which is above the dilatometric Tg, the struc-tures are completely filled. Despite the highmold temperature it is still possible to conduct a

regular molding cycle but the products are not according to specs with respect to flatnessand birefringence of the PC bulk. The molded structures are very good replicas of the moldinsert as can be seen in Figure 6. Even the standing waves which are typical for DUV lithog-raphy are still visible on the walls of the PC grating. The duty cycle, however, is lower, i.e.the walls are thinner than according to the mold structure. This is attributed to shrinkage.

µ

µ

Figure 4. Insert structures as obtained by RI etching (top) and DUV lithography (bottom).

Figure 5. Measured depth of the replicated structures in PC vs. mold temperature as obtained from Ni insert.

1000

800

600

400

200

0144 146 148 150 152 154

T mold [degC]

dep

th [

nm

]


With a moldinsert of fused quartzthe contact tempera-ture will be muchhigher at the same

mold temperature. Therefore the filling will bemuch easier. In fact no filling problems wereencountered with mold temperature of 140oC.SEM micrographs of cross-sections of struc-tures molded in this way are shown in Figure 7.As can be seen the structures are completelyfilled but there is a shape deviation. The struc-tures are thicker at the top than at the bottom. Atthe moment this is not quite clear whether this isa filling phenomenon or a consequence of someelongation during the release of the structures.The secondary structure at the fracture surfaceis due to the fracture which indeed is very diffi-cult to achieve with Polycarbonate. Sofar onlystructures of 0.8 m have been shown. But sur-prisingly enough we were also able to moldstructures with double the depth. A major prob-lem turned out to be the observation of the rep-lication fidelity as such structures cannot beinvestigated easily. One technique which worked well is milling by a focussed ion beam.There the structure is first coated by a W layer and then ablated at a certain area. In this way

µ

Figure 7. Fracture surface and surface SEM micrographs of PC structure as repli-cated from SiO2 insert.

Figure 6. SEM micrographs of fractured edge of PC structures.

Figure 8. SEM micrographs of defect (top) and FIB machined detail at 45 degrees of 2 m deep grating.µ


it becomes possible to view the structures from aside. A typical result is shown in Figure 8.The contrast in the SEM is not very good but one can clearly see that the structures areapproximately 2 m deep! To our best knowledge this is a record for replication with injec-tion molding. As can be seen also in Figure 8 is that the walls start tumbling which willaffect the local duty cycle and consequently the optical function. The inserts are very vul-nerable and in practice we always found local defects like missing walls. Also the etching isvery critical as mentioned above.

The mold temperature is also important for the total shrinkage of the polymer after vit-rification. The shrinkage of the bulk leads to a relative displacement of the replicated struc-tures with respect to the mold structures and will destroy the fragile walls. Therefore themold temperature has to be chosen close to the vitrification temperature anyway.

CONCLUSIONS

• Grating structures with l/4 retardation have been designed for use in CD recorders with785 nm wavelength.

• Such rectangular structures have a pitch below 0.5 m and depths in excess of 1.5 m.• Inserts with these dimensions were indeed manufactured by RI-etching in SiO2. The

etching process window, however, is very narrow and defects are always observed inthe structures.

• By splitting the function over the two surfaces of the retardation element the depth ofthe structure is reduced to approx. 0.8 m. Such structures can be produced in both Niand SiO2.

• The replication from SiO2 inserts by straight injection molding yields a stable and feasi-ble process for mass fabrication.

• Structures with a record depth of 2 m and a pitch of 0.5 m were replicated by injec-tion molding with polycarbonate.

ACKNOWLEDGEMENT

The author likes to acknowledge the contribution of many colleagues at Philips which wasessential for success of the project. W. Ophey, J. van Haren and R. Merkx for their assis-tance with the optical simulations, H. van Helleputte and E. van der Heuvel for the lithogra-phy and etching of the structures on the mold inserts, J.Godfried of PMF for help with themolding experiments, W. Coumans for the electroplating and J. de Bruin for the fabricationof the inserts.

µ

µ µ

µ

µ µ


REFERENCES1 Born, M., and Wolf, E., Principles of Optics, Pergamon, New York, 1980.2 Urbach, P.H., and Merkx, R.T.M., Finite Element Simulation of Electromagnetic Plane Wave at Gratings for Arbitrary Angles of Incidence, in Mathematical and Numerical Aspects of Wave Propagation Phenomena, Eds. Cohen,

G., Halpern, L., and Joly, P., SIAM, 1991.3 Chapman, B., Glow Discharge Processes, J. Wiley, New York, 1980.4 Wimberger-Friedl, R., and de Bruin, J., Rheol. Acta, 30 (1991), 329 and ibid. 419.5 e.g. Ehrig, F. Klein, H., Rogalla, A., Ziegmann, C., Micro technology: New Dimensions in Plastics Processing, in 19 Kunststofftechnisches Kolloquium des IKV, 1998, 1.

Process Analysis and Injection Moldingof Microstructures

Alrun Spennemann and Walter MichaeliInstitute for Plastics Processing (IKV), Aachen, Germany

INTRODUCTION

The injection molding of microstructures represents a key technology for the economic pro-duction of medium and large series of microstructured moldings and the assembly ofmicro-systems. During the last years fundamental research on the injection molding ofmicrostructures has been done at the Institute for Plastics Processing (IKV) at Aachen Uni-versity of Technology (RWTH Aachen). For these investigations a suitable injection mold-ing machine and an appropriate mold technology were provided. A modular stem mold wasdesigned for various different mold inserts. This mold contains a vario-thermal temperingsystem.Thus the mold is first heated close to melt temperature before injection with an elec-tric heating and then cooled down to ejection temperature by a fluid tempering unit.1

In a first step demonstrator cavities were used to analyze the suitability of selected lowviscous materials to fill a microstructured mould easily. These demonstrators, e.g. honeycomb structures, were moulded in LiGA-cavities. The LiGA-technique is a special technol-ogy often used in micro technology that allows the precise production of microstructureswith very high aspect ratio and high quality surfaces. For aspect ratios < 5, traditional pro-cesses, e.g. micro-cutting, micro spark-erosion and laser erosion are supposed to be a lessexpensive alternative.1

INJECTION MOLDING OF MICROSTRUCTURED PARTS

This paper will compare the three processing technologies mentioned concentrating on thetechnological limits (smallest size of structures, cavity materials, freedom of design,...) andthe quality of molded parts. Systematic trials were carried out to find a process window ofinjection molding parameters. After 500 shots the quality of the cavities was controlled.

Ten different cavities were manufactured by microcutting, micro spark-erosion andlaser-erosion. The geometry of these cavities varies in width and depth of the ditches andthe width of the bridges. Moreover, different ejection slopes are realised in some cavities.The mould inserts are 8 mm in diameter with a structured area of 6 mm in diameter.


The design of the cavities follows IKV investigations and hints from publications1-3 aswell as advice from the manufacturer (Fraunhofer Institut fur Produktionstechnologie (IPT),Aachen; Fraunhofer Institut fuer Lasertechnik (ILT), Aachen; Laboratorium fuer Werkzeug-maschinen and Betriebslehre (WZL), Aachen; Ronda AG, Schweiz).

For systematic injection molding trials, standard parameters were defined for two ther-moplastic materials: polyacetal (POM) and polycarbonate (PC). The parameters for the pro-cessing of POM (Hostaform C52021, Ticona GmbH, Frankfurt) are: mass temperature of220°C, temperature of the oil tempering unit 90°C (i.e. a cavity temperature of about 190°Cbecause of the additional electric heating) and injection pressure 250 bar. PC (Makrolon2205, Bayer AG, Leverkusen) was molded at 310°C mass temperature, 500 bar injectionpressure and the oil tempering unit was set at 120°C. Figure 1 gives an example of a moldedpart (POM) from a spark-eroded cavity. The cycle times vary from 40 to 60 s.

The injection molding parameters were varied systematically to analyze their influenceon the process. Parts molded at a higher temperature of the oil tempering unit (105°C) areoverfilled (Figure 2, left), but when the oil temperature was set as low as 60°C, the surfaceis not molded well (Figure 2, right). After 500 shots, the small bridges (20 m) in the cavityµ

Figure 1. REM shot of a molded part (standard molding parameters, POM, spark-eroded cavity). Figure 2. Variation of cavity temperature (POM, spark-

eroded cavity).

500 µm

variation of geometry in the cavity:width ofbridges

(µm)(µm) (µm)

width ofditches ditches

depth of

left: 10 to 40 300 200

right: 20 300 100 to 400

1 mm

1 mm1 mm

Figure 3. REM shots pf molded parts.

cavity: micro-cutmaterial: POM

500 µmcavity: laser-erodedmaterial: PC

Process Analysis 159

showed damages, because the mold material was thermally influenced by the spark-erosionprocess. Therefore, bridges in spark-eroded cavities should at least have a width of 30 m.

Against that, the micro-cut cavity is rather tough. A disadvantage of micro-cutting isthe limitation to non-ferrous metal for the processing with diamond tools. The molded parts(POM) are a detailed copy of the cavity SPAN 3 (Figure 3, left) and were ejected withoutany problems.

The laser erosion process provides variability as well in design of the structures as inthe choice of the cavity material. Unfortunately, the erosion process was not optimized, sothat the surface of the structures is rather rough. The molded part (Figure 3, right) repro-duces these surface faults exactly. As the cavities do not show any damage after the injec-tion molding trials, this process technology should be improved and is then very suitable forthe production of micro cavities.

DEVELOPMENT OF A NEW MACHINE TECHNOLOGY

In micro injection molding another task besides the injection molding of small parts (> 1 g)with microstructured details as described above has to be considered: the direct productionof micro parts, i.e. parts with a part weight down to a milligram (mg). Until now there are nosuitable injection molding machines available for the production of single micro parts, soinjection molders produce big, but precise sprues to achieve the necessary shot weight.4-6

Figure 4 gives an example: The two parts shown areraytracing elements in the headlights of a Maerklin MiniClub railway engine. Made from polymethylmethacrylate(PMMA) both of them together have a part weight of 0.0335g, but the shot weight including the gate is 0.5549 g, so thatthe weight of a single part is only about 3% of the total shotweight. The regrind of the sprue cannot be used for the samearticle, as the quality of the recycling material is not goodenough for raytracing. So 94% of the material are wasted.Considering costs of up to $60/kg of special material e.g. formedical applications, this waste can be an important cost fac-tor.

Figure 4 also illustrates the specific problems that come along with such small shotweights: The size of the granules used in standard injection molding limits the size of theplastification screws to 14 mm diameter minimum, i.e. that when the screw moves just 1mm, about 0.185 g plastic material are injected. And even just one granule of PMMAweighs 0.024 g, which is more than one of the parts shown in Figure 4.

µ

Figure 4. Molded parts (0.0335 g), gate (0.5214 g) and PMMA granule (0.024 g).


Until now several machine builders offer modified standard machines with very smallscrews for the injection molding of small parts from 5 g to 0.5 g shot weight. One machinebuilder presents a suitable injection molding machine with integrated quality control andhandling for micro parts (part weight down to mg, but - including the gate - higher shotweights about 0.01 g) on the Plastics and Rubber Fair K '98, Duesseldorf.6 Since severalyears the IKV is involved in the development of special plastification units for small shotweights. A combined screw plastification and plunger injection system was developed byIKV and Ferromatik Milacron Maschinenbau GmbH, Malterdingen, Germany and is soldwith Ferromatik machines since 1994. This plastification unit has a preferred shot weight of0.1 to 1 g.1

To open new dimensions in the size of minimum shot weight (< 0.01 g), IKV is nowdeveloping a micro injection molding machine that meets the molder's demands. Thesedemands were defined by reference molders and by the experiences made during the pro-cess analyses with the different cavities as described above. The injection pressure variesbetween 150 and 600 bar. The cavity has to be evacuated to 0.5 bar to avoid burn marks andsoiling. Mass temperatures go up to 400°C for some engineering plastics. The mold temper-ing (fluid tempering) varies between 60 and 180°C and an additional local heating of thecavity up to mass temperature has to be realised.

So the components of the new machine concept have to consider the followingdemands: the plastic material must not melt in the material feeding, but in a small meteringzone to avoid material degradation. The dosing has to be controlled properly without soak-ing in air. A homogeneous tempering of the plastification unit with a good thermal separa-tion of nozzle and mold is important. For the injection of mg-shot weights the dosing has tobe very exact and the material has to be injected fast and without leakage. All components

should be dismantled and cleaned easily.In the following the important elements and functions of the

new machine and the injection molding process are explained.7

Figure 5 shows nozzle [1] and mold (two plates [2] and [3]) indetail. The conical nozzle is tempered separately and well insu-lated [4] against the other machine components. The spree plate[3] is very narrow to keep the spree volume [5] as small as possi-ble. The plate [2] on the movable platen side transmits theclamping force onto spree plate and nozzle.

The process starts with the injection of the molten mass [6]into the cavity [7] using an injection plunger (see Figure 6). Theplunger is driven by an electric motor. Nozzle and mold platesare heated up to the temperature of the molten mass. After injec-Figure 5. Nozzle and mold.

Process Analysis 161

tion, the holding pressure avoids shrinkage whilethe cavity is cooled and the plastic freezes. At theend of the holding phase, the nozzle is cooleddown rapidly by the injection of liquid gas (CO2),so that the mass in the nozzle freezes at once andshrinks (see Figure 7a). At this moment new mol-ten mass can be metered. Then the gate is ejectedand teared off the molten mass in the cavity. Fig-

ure 7b explains the ejection of the molded part that is cooled down by a fluid tempering. Toeject the part, the mold plates [2 and 3] have to be opened. During ejection the mold plate[3] is on the nozzle to heat up again and to hold back leaking melt. Then the mold closes andis heated up to the temperature of the molten mass. So the next production cycle can start.

As the concept of the plastification and the injection unit is new and completely differ-ent from those available on the market, IKV has applied for a patent.7

OUTLOOK

In the next steps of the project, the machine components will be designed and built accord-ing to the described concept. A suitable mold technology has to be adapted following theIKV experiences in micro injection molding. An important aim is the reduction of the sizeof the mold to realize an effective and homogeneous tempering. The behavior of the newmachine/mold system has to be tested. Investigations about the produced part quality willfollow.

Figure 6. Injection molding machine - start of process cycle.

Figure 7. Dosing, ejection of the gate (a) and of the molded part (b).


ACKNOWLEDGEMENTS

The investigations set out in this report received financial support by the Deutsche Fors-chungsgemeinschaft (DFG), to whom we extend our thanks.

REFERENCES1 A. Rogalla: Analyse des Spritzgiessens mikrostrukturierter Bauteile aus Thermoplasten, Ph. D. Thesis, RWTH Aachen, 1998.2 W. Michaeli, A. Rogalla, A. Spennemann, C. Ziegmann: Mikrostrukturierte Formteile aus Kunststoffgestalten, F & M Feinwerktechnik, Mikrotechnik, Mikroelektronik, 106 (1998) 9, p. 642-645.3 M. Weck, S. Fischer: Ultraprazisionstechnik fur die Werkzeugbearbeitung, Froc. IKV-Seminar Innovative

Produktionstechnologien fur das Spritzgiel3en von Klein- and Mikrostrukturbauteilen aus Kunststoff, Aachen, 1997.4 W. Gotz: Mikroteile in der halben Zykluszeit herstellen, Industrieanzeiger 18 (1998), p. 40-41.5 M. Kleinebrahm: Der Weg zum Mikrospritzgiessen, Proc. Micro Engineering, Stuttgart, 1998.6 C. Kukla, H. Loibl, H. Detter, W. Hannenheim: Mikrospritzgiessen - Ziele einer Projektpartnerschaft, Kunststoffe 88 (1998) 9, p. 1331-1336.7 W. Michaeli, A. Spennemann, B. Lindner, E. Koning, J. Zabold: Verfahren zum Spritzgiessen von Mikroformteilen aus thermoplastischen Kunststoffen mit einer geringen Angussmasse, applied for patent, 1998.

Simulation ofthe Micro Injection Molding Process

Oliver Kemmann and Lutz Weber Institut für Mikrotechnik Mainz GmbH, Germany

Cécile Jeggy, Olivier Magotte, and François DupretUniversité Catholique de Louvain, Belgium

INTRODUCTION

Market analysis for microsystems1 show, that 40billion US$ will be spent on micro devicesmainly in the automotive and communicationindustries until the year 2002. Key products areacceleration and pressure sensors and increas-ingly components for the computer industry(read/write-heads for hard disks, flat displaymonitors etc.). Even if polymer parts have nottaken over the market of these silicon basedproducts, yet, they are already performing excel-lently in the fields of medical technology, bio-technology or passive plastic components foroptical networks. Examples include micromotors and gears (Figure 1), optical switches(Figure 2), glucose and blood pressure sensors aswell as components for minimal invasive sur-gery. A multibillion dollar market for micro-structured parts with typical part-structuredimensions from several micrometers up to 100µm can be expected. And injection molding isstill the most common process for cost effectivemass production even in the field of microstruc-tured parts.

Figure 1. Micro motor, with gear box.

Figure 2. Optical switch.


Tools like software packages for the injection molding process are very common inindustry, since the early eighties,2-4 because development time and cost must be decreasedpermanently. Filling, post-filling even shrinkage and warpage of plastic parts can thereforebe calculated. All producers of plastic parts are using simulation tools to decrease the devel-opment costs, select proper machines and choose the right material from material data basesall in the sense of a cost effective mass production of plastic parts. Fast running 2½D codesprovide excellent results with flat and thin, so-called standard injection molding parts. How-ever, to conquer the very new market of micro injection molding it is extremely important tolearn from the huge amount of experience in the field of standard injection molding and toprovide the same tools. Therefore initial simulations with commercial software were run atthe Institute for Microtechnology Mainz (IMM) to check their suitability for the micro-injection molding process. The characteristic shapes and extremely small dimensions ofmicro parts as well as unique flow front shapes with different materials quickly reveal that a2½D solution is not sufficient anymore to describe all the effects. Therefore, to describe thefilling of micro structures a 3D transient code is under development at the UniversitéCatholique de Louvain (UCL).5

DIFFERENCES BETWEEN STANDARD AND MICRO INJECTION MOLDING

In general micro injection molding is the production of plastic parts with structure dimen-sions in the micron or sub-micron range. Micro structured mold inserts are produced withthe help of ultra precision processes. These inserts are attached to standard molds as knownfrom conventional injection molding. Especially the LIGA technology6 allows the produc-tion of metal mold inserts with structures in the micron or even sub micron range, e.g. byattaching many micro test structures to one base plate.

To achieve proper filling of micro structured parts significant modifications to the stan-dard process must be made. Due to the extremely low surface roughness of the mold cavitywalls, demolding without a standard draft of 3° becomes feasible. However, during demold-ing any lateral offset has to be avoided. Otherwise it can be observed, that structures areripped or sheared off the ground plate. To support the filling of small cavities, especiallywith a high aspect ratio (height against smallest lateral dimension) the so-called variother-mal heating is used. In contrast to standard cooling, which keeps the mold temperature at acertain temperature below transition temperature, the surface of the mold insert is heated upwith the help of an inductive heating7 almost to the melt temperature in order to gain alower melt viscosity during filling. Compressed air causes problems, too. So the air in themold must be evacuated by a vacuum pump. This is necessary to provide complete part fill-

Simulation of the Micro Injection Molding 165

ing as well as to prevent the “Diesel effect” where polymer is burned by the compressed, hotair at the bottom of the insert.

The state of the art in micromolding allows the development of microstructured plasticparts with hardly any restriction in design. Suitable machines and molds are available. Onthe other hand, there is a lack of basic understanding of the flow behavior of plastics inmicrostructures. Hence, suited test structures as well as simulation tools must be developed.

INITIAL 2½D SIMULATIONS AND MOLDING TRIALS

Almost three years ago, IMM started with initial simulations of the microstructure filling.Using the C-MOLD software,8 the filling of test structures (Figure 3)9 was simulated (Fig-ure 4). These test structures were filled via a ground plate, while the two walls building across were 50 µm and 10 µm thick and 100 µm high. The material was POM. As observedfrom the performed short shots, the thinner wall was simultaneously filled from the groundplate and the thicker wall, thus was causing a weld line through the smaller wall structure.Another effect is, that the upper right corner is filled last. The calculation results show theseeffects sufficiently but, taking a closer look at the edges of the structure walls, one can seemelt front effects, which are in the direction of flow, and cannot be predicted by usual 2½Dsoftware tools.

To understand this, it is necessary to take a look at the simplifications made whendescribing the filling behavior with conventional codes. The usually flat and thin characterof conventional injection molding parts allows to make these simplifications, which makethe calculations fast and easy. The dimensional character of the micro structures is howevernot thin and flat anymore, and therefore does not fulfill the requirements needed to allow thesimplifications used in standard simulation packages. Also visco-elastic- or surface-tension-

Figure 3. Test structure molding, POM. Figure 4. Simulation result, C-MOLD filling.


effects are not taken into account since they are not important for standard parts.10 Even themeshing of micro structures with 2D mesh generators is difficult and mostly impossible.Modeling e.g. blind hole parts as they are typical for LIGA components reveal the problemof double generating volumes, since the real micro structure is much smaller than the stabi-lizing ground plate. So the volume around the connecting node is generated twice. A similareffect is known from standard parts as the “step effect” where a sudden change of wallthickness cannot be modeled since a mid-plane 2D mesh cannot describe asymmetrical wallthickness distributions. Beside these modeling and numerical simplifications which lead toincorrect prediction of the filling, the material (especially the rheological data used) is takenfrom data bases for macroscopic applications and scaled down to the sub-millimeter range.To provide more suitable data, especially for viscosity, and to be able to verify the simula-tion results, it is thus necessary to investigate the filling of microstructures directly, by usingdifferent materials.

Therefore, a proper simulation of the micro injection molding process requires a 3Dcode and mesh generator as well as accurate data for micro range applications. In the fol-lowing sections, a first approach to a simulation software suitable even for LIGA parts,which are right now the smallest molded parts, is developed while the injection moldingtests are performed to provide the data needed to verify the results from the simulations.

3D SIMULATION APPROACH

In the 3D model, two simulation scales, the ground plate and the micro-part scales, must beconsidered in order to predict numerically the injection molding of LIGA-produced micro-parts. Two meshes must thus be generated in order to simulate the filling process for suchmicro-part. First, the filling of the ground plate is predicted using a mesh, whose finite ele-ment scale is quite larger than the micro-part size. Hence, the number of nodal unknowns isnot prohibitively high (it should be noted that conventional injection molding softwarecould generally be used to perform this first simulation). Secondly, the filling of the micro-structure is performed using a 3D mesh. Both the micro-structure and a reduced part of theground plate are covered by this second mesh, in order to allow imposing boundary condi-tions obtained from the first simulation. However, the second simulation certainly cannot beperformed using conventional injection molding software, since the second mesh clearlyexhibits three-dimensional features.

The objective of performing filling simulations is to predict the motion of the flowfront(s) (which are true surfaces of arbitrary shape), together with the evolution of theunknown fields. For that purpose, the 3D time-dependant software uses two basic modules(viz. the so-called flow and geometrical solvers), which are processed using a decoupledalgorithm.


The flow solver is devoted to calculate the velocity, pressure and temperature fieldsusing a generalized Newtonian model with a temperature and shear rate dependent viscos-ity. As this approach turns out to be not completely satisfactory for simulating the filling oftruly 3D micro-parts, visco-elastic effects will be considered at a later stage. While classicalfinite elements are used for space discretization, Eulerian integration schemes are used fortime integration.

The geometrical solver is devoted to move the front(s), track front-front and front-wallmeetings if any, and to create a finite element mesh covering the flow domain occupied bythe fluid at every time step of the simulation. Moreover, an “extrapolation mesh” coveringthe region located between the fronts at successive times tn and tn+1 is generated in order toallow extrapolation of the fields calculated by the flow solver onto the new flow domain(this is required because Eulerian time integration is performed). A similar method wasused11 in order to perform 2 ½ D molding simulations.

The re-meshing algorithm is based on the Delaunay triangulation principles and imple-mented using the node-insertion scheme developed by George.12 Exact geometry algo-rithms are used in order to avoid the dramatic effect of round-off errors occurring incomputational geometry procedures.

NEW MICRO INJECTION MOLDING TRIALS

In order to provide sufficient material and flow data for the simulation of the micro injectionmolding process it is extremely necessary to investigate the flow behavior of various ther-moplastic engineering polymers. Therefore, typically used flow spirals are replaced bynewly designed structures based on the experience gained in former investigations at IMM.New inserts are derived from these guidelines to investigate the limits of part filling inmicro injection molding. The trials are carried out with different engineering polymers withknown excellent flow characteristics.

Test structures with lateral dimensions between2.5 µm and 20 µm (Figure 5) are mounted onto aground plate and filled via a regular runner system. Afilm gate connects the ground plate with the runner sys-tem. With systematic injection molding tests the flowbehavior of the polymer through the molded structureshas been investigated. An injection molding machine(ARBURG Allrounder 370C 800 – 100) and the fol-lowing materials have been used: polyoxymethylene(POM) as a standard material for micro injection mold-ing and an unfilled polyphenylenesulfide (PPS)

Figure 5. LIGA test mold inserts.


because of its extremely low viscosity atmelting temperature. The final parts havebeen investigated with the help of a SEM byobserving the degree as well as the quality ofthe filling.

The POM parts (Figure 6, left) showvery good filling behavior. Even the thinchannels of the part are filled very well. Tak-ing a closer look at the smallest ring (Figure6, right) shows, that the demolding process isbending the thin structure. Compared to thePOM parts, the PPS structures are less wellreplicated. Only the two outer circles are suf-ficiently filled (Figure 7, left). Taking acloser look at the melt front of the largest ring(Figure 7, right) reveals a unique shape with alot of small weld lines in it. Showing such a

significant difference between the different parts, forces the need for a simulation softwarehelping to foresee such behavior for future applications.

OUTLOOK

With the increasing use of micro injection molding in micro fabrication, the need for a soft-ware tool providing important information during the design stage of the part developmentwill increase, too. Therefore the ground has to be prepared to develop a specific simulationsoftware for micro injection molding. Moreover, the results gained from injection moldingtests with many different materials should be collected in a database, and software program-mers will take all the differences between conventional and micro injection molding intoaccount by using this data base. Within the rush into 3D software tools for the standardinjection molding process, the special role of micro injection molding must not be forgotten.

ACKNOWLEDGMENTS

The authors wish to thank the European Commission and the consortium of the BRITE-Euram Project BRPR-CT97-0430.

REFERENCES1 R. Wechsung, J. C. Eloy, Market Analysis for Microsystems – an interim report from NEXUS Task Force, Proc.

EUROSENSORS XI, Warschau (1997).

Figure 6. POM molding results.

Figure 7. PPS molding results.


2 H.H. Chiang, C.A. Hieber and K.K. Wang, Polym. Eng. Sci., 31 (1991) 116;125.3 C.A. Hieber and S.F. Chen, J. Non-Newtonian Fluid Mech., 7 (1980) 1.4 K.K. Wang and V.W. Wang, in : A.I. Isayev (ed.), Injection and Compression Molding Fundamentals, Marcel Dekker, New York, 1987.5 C. Jeggy, O. Magotte and F. Dupret, Numerical simulation of the micro-injection moulding process , in: Proc. 2nd ESAFORM Conference on Material Forming, Guimarães, Portugal (April 1999), 117;120.6 W. Ehrfeld, D. Münchmeyer, Nucl. Inst. and Meth. in Phys. Research, A303, p. 523-531, (1991)7 C. Schaumburg, W. Ehrfeld, W. Schinköthe, Th. Walther. L. Weber, Microsystems Technology 98, Proceedings, Berlin, p. 679, (1998)8 C-Mold, User Manual, Advanced CAE Technology, Inc. Ithaca, (1997)9 M. Hörr, Diploma Thesis, FH Darmstadt and the Institute for Microtechnology Mainz GmbH, (1997)10 J. Zachert, Analysis and Simulation of Three-Dimensional Polymer-Flow in Injection Moulding, Aachen (1998)11 F. Dupret and al., Modelling and simulation of injection molding, in Advances in the Flow and Rheology of

Non-Newtonian Fluids, D.A. Siginer, D. De Kee and R.P. Chhabra (ed.), Rheology Series, Elsevier, 1998.12 P.-L. George and H. Borouchaki, Triangulation de Delaunay et maillage, Hermes edition, (1997).

Chapter 4: Manufacuring of CompositesMelt Compression Molding (MCM) a One-shot

Process for In-mold Lamination and CompressionMolding by Melt Strip Deposition

Georg H. KuhlmannDieffenbacher UTW

PREFACE

Since about 10 years processes for simultaneous moulding of carriers and decorative lami-nation (IML - in-mold lamination) are steadily replacing conventional methods. This devel-opment was primarily initiated by the automotive industry with the objective to be preparedfor future trends such as:• growing demands for better and more comfortably appointed interiors of passenger

cars and - to a lesser extent - of vans, busses, and trucks achievable e.g. by an increasedapplication of textile coverstock and leather substitutes both preferably with a soft touch

• the necessity of cost reduction i.e. by fewer manufacturing steps and less manuallabour including finishing

• more safety e.g. by application of materials with higher impact and without splinters orsharp rupture lines after accidents as well as the use of foam paddings

• ecological concerns to be overcome by lamination without adhesives i.e. solvent mat-ters and yet with better adhesion of the laminate, furthermore by composites suitable forrecycling or uncritical incineration of waste or used parts.

• preservation of fossil energy by reduced vehicle weights also easing the strain on traf-fic surfaces as well as by substitution of processes heavy on energy like GMT (Azdel)preperation and forming

• a fair chance for agriculturally orientated economies replacing industrial fiber by regen-erative fibers.The technologies described by the term "low pressure injection moulding" can substan-

tially contribute to achieve these objectives. Meanwhile other industries i.e. not connectedwith the automotive industry e.g. furniture and packaging material manufactures are suc-cessfully applying the processes - a trend gaining forceful momentum by excellent resultsobtained by compression moulding of melt strips of long glass fiber reinforced thermoplas-tics (LFT) into technical, non-laminated parts.


LOW PRESSURE INJECTION MOULDING TECHNICS

Low pressure injection moulding technics have alot in common to justify the definition. They arenot a fundamentally new technology but theclever combination of known technical methodsfurther developed and improved for the purpose.

At present three technics are steadily gain-ing importance - a development with growingmomentum since wider use reveals the out-standing capabilities of the processes.

These innovative processes are:• backinjection including the injection/com-

pression• melt flow compression moulding and• backcompression by melt strip deposition for two applications i.e. in-mold lamination

(IML) and compression forming of fiber reinforced thermoplastics (LFT).Low pressure injection moulding technics have a lot in common to justify the definition:• Predominantly hydraulic clamping units - vertical or horizontal - are applied, modified

from clamping unit for conventional injection moulding.• Plastication occurs by means of a single screw extruder.• The melt is injected into a mould - closed or open - by a conventional injection unit

adapted for high plastication and injection rates.• All low pressure injection moulding processes are capable of in-mould lamination

(IML) of decorative coverstock.• Part forming is performed at low internal mould pressure originally not exceeding

approx. 100 bar (i.e. 1450 psi) also established as the borderline for economical in-mould lamination (IML) i.e. about the maximum sustained by coverstock materials.With the advent of LFT (long fiber reinforced thermoplastics) compression mouldinginternal mould pressures up to 200 bar (i.e. 2900 psi) are applied - an acceptabledemarcation line between low and high pressure injection moulding. Generally speak-ing the internal mould pressure for the low pressure technologies amounts to 15 to 60%of high pressure applications.

• Most development efforts are dedicated to the reduction and limitation of internalmould pressure during the forming cycle. These are areas influenced by the machinesand the pertinent software e.g. - melt injection profiles - pressure build-up and compres-sion speed profiles - clamping force decompression profiles - reduction of flow length

Figure 1. Low pressure injection molding process.

Melt Compression Molding 173

by sequential gate valve actuation (cascade valve control) or variable melt strip deposi-tion.Of course mould design is a decisive factor for the moulding success e.g. by dimen-

sioning and location of the sprue gates, dimensioning of shear edges, flow aids, cooling andejector technics, etc.

This paper is primarily concentrating on technics based on the melt strip deposition.As there are many different terms for the various low pressure injection moulding technics,it appears to be useful to briefly identify the other methods.

BACKINJECTION

From the many denominations "backinjection"seems to be the most descriptive and probablythe most popular.

The process is performed on conventional,mainly horizontal injection moulding machinesor - in increasing numbers on special machineswith relative to the clamping force - large mouldmounting areas and purpose built injection unitswith high injection rate and low injection pres-sure.

The coverstock is inserted and located in anopen mould - a shear edge mould permittingdraw-in of the coverstock during the closingcycle to avoid wrinkles and damage by stretchingof the fabric and yet flash-tight during the injec-

tion. In order to prevent weakening joint lines (also a potential source of wrinkles), meltpenetration, destruction of any foam backs and/or special textile effects like piles, plush fin-ish or leather grain embossing on foils injection occurs through carefully arranged gateswith pneumatic needle shut-off nozzles which are actuated for injection in a specificsequence described as cascade control.

Moulds for backinjection are quite sophisticated. Apart from a complicated hot runnersystem incorporating the shut-off nozzles with their pertinent drives also all other mouldelements like ejector, core pulls and slides have to be accommodated in the injection sidemould half. Ejectors etc. are not acceptable on the decorative side.

A variant is the injection/compression cycle during which - sometimes after a pre-forming stroke for the coverstock - the carrier material is injected in a partially open mould.

Figure 2. Principle of backinjection cycle. 1 - insert cov-erstock, 2 - clamping and injection, 3 - cooling, 4 - demolding.

1 2

4 3


By closing the gap the part is formed and laminated. The mould corresponds to a backinjec-tion mould. The method has similarities with melt flow compression moulding.

By backinjection remarkable results can be achieved provided the limitations of thetechnology which are especially valid for larger parts are recognized e.g.• restricted influence on coverstock preservation - fabric or foil - without barrier back fin-

ish• back finish is required for a save process• significant effect on foam layers• no genuine soft touch• sensitive process because of risk of wrinkles or damage to the coverstock• rather complicated mould system which may be heavy on maintenance.

MELT FLOW COMPRESSION MOULDING

Melt flow moulding for short (ignoring all theother names) is performed on vertical clampingunits.

The coverstock - perhaps preformed fordeep parts - is inserted into an open mould. Thenthe mould is partially closed. The carrier stock isinjected from below through a hot runner systemand several generously dimensioned gates withpneumatically actuated needle shut-off nozzles.The melt available as cakes around the gates iscompression formed into the part by closing theremaining mould gap.

Shear edge moulds with hot runner systemssimilar to those for backinjection are applied.Mould cost especially for large parts is probablythe only drawback for a wider acceptance of theprocess. Another may be a rather diffuse patentsituation in some countries.

A melt flow compression moulding plant issimilar to backcompression equipment. However normally fitted with one injection unitonly which is permanently attached to the lateral inlet of the hot runner block i.e. motionaxes as with backcompression equipment are not required. Backcompression machines areavailable equipped for melt flow compression moulding.

Figure 3. Principle of flow compression cycle. 1 - insert coverstock, 2 - partial closing and injection, 3 - compres-sion forming and cooling, 4 - demolding.


BACKCOMPRESSION - MCM

The term "backcompression" is quite wellaccepted for a process based on compressionmoulding of a melt strip deposited in an openmould.

Backcompression describes the process dur-ing which a coverstock cutting is placed on a meltstrip for simultaneous compression moulding andlamination (IML) of interior automotive parts.

In recent months the application also basedon melt strip deposition of mostly fiber rein-forced thermoplastic stock (LFT) with subse-quent compression moulding e.g. ofnonlaminated structural car parts attracts growingattention.

MCM, short for melt compression mould-ing, appears to be a more comprehensive namewith the distinction IML for the previous back-compression process and LIFT for all long fiberreinforced applications.

MCM-IML

A typical MCM-IML cycle is performed as follows• The cycle starts with an open mould in a vertical press.• A horizontal injector equipped with a deposition head moves into the mould depositing

a melt strip in the lower mould half during the retraction movement (x-axis).• A flat or preformed decorative cutting is placed on the melt strip.• The press closes moulding the part by compression.• At the end of the cooling cycle the press opens for part demolding.

Thermoplastics are used as carrier materials - predominantly PP unfilled or talc filledand to a lesser extent ABS, ABS/PC alloys, and PA.

There is a vast variety of coverstock materials e.g.• woven and non-woven fabrics with various finish like pile and plush, and many colors

including sensitive dark blue. Barrier layers are the exception even for fabrics as light as<200 g/m2 (i.e. <0.7 oz/ft2)

Figure 4. Principle of MCM-IML cycle. 1 - melt strip deposition, 2 - insert coverstock, 3 - compression form-ing and cooling, 4 - demolding.


• foils (TPO, PVC) with an embossed surface are increasingly used with excellent preser-vation of the grain.The coverstock is normally placed on top of the melt strip in order to minimize the

exposure time to the heat of the melt. With today's fast cycles coverstock insertion prior tothe melt strip deposition is possible however excluded for materials with foamback or foilswithout back protection (barrier layer).

Lower stock temperature, intensified mould cooling, and thinner walls have lead to sig-nificantly shorter cycle intervals. Lower temperature and decreased internal mould pres-sures more recently supported by ram decompression and a length controlled retractioncycle are protecting the coverstock most efficiently.

With cycle intervals around 60 s MCM-IML is revealing its versatile capabilities andeconomical feasibility. Originally mainly intended for large area parts like complete doortrims there is a rapidly increasing interest in applications for smaller parts e.g. B and C pil-lars, map holders, IP flaps, etc.

Since the introduction of MCM-IML for industrial production less than 6 years agothe process has seen significant progress in application engineering as shown on the chartbelow.area of change previously todayback protection of- same as for depending on naturecoverstock backinjection of material none or littlemoulding of assembly simple shapes only all shapes requiredfixtures including elements of- 120 mm (4.7 in)of

depth and with undercutsstock temperature (PP) at 230-240°C 180-190°Cdeposition die orifice (450-460°F) (360-375°F)internal mould pressure min. 60 bar 50 bar

(870 psi) (725 psi)melt strip deposition speed < 100 mm/s > 300 mm/s

(< 4 in/s) (> 12 in/s)mould temperature upper/ 40°/40-70°C 10-20°/20°Clower half (100°/100-160°F) (50-70°/70°F)wall thickness achievable 2.5 - 3.5 mm 1.8 - 2.0 mm

(0.10-0.14 in) (0.07-0.08 in)typical cycle time with 120 (105) s 67 (55) smanual (robotic) insertionand demolding


Multi-cavity moulds are common already e.g.• 2 left + 2 right center console parts for BMW series 5 with pile carpet lamination i.e. 4

parts per cycle of 57 s or• 2 left + 2 right door trim inserts for BMW series 3 i.e. 4 parts per cycle of 60 s both with

robotic coverstock placing and parts demolding.There are many valid reasons for the ready acceptance of the MCM-IML process by

the industry like• numerous degrees of freedom of parts design• moulding of high/deep assembly fixtures including undercuts• elimination of 180° back wraps previously necessary for optical reasons• 90° back wraps also with recess• sensitive treatment of delicate coverstock e.g. fabrics with pile and plush finish,

embossed foils, also material with foam back• genuine uniform soft touch also on large area parts• none or little protective backfinish of decorative coverstock• lamination in one heat (and cycle) i.e. IML resulting in• absence of adhesives i.e. of aggressive solvent matters• secure intimate durable bonding of lamination• low content of manual labour• low internal stresses because of the compression cycle• most rejects, waste, and used parts are recyclable or suitable for incineration i.e. no

waste for classified disposal• lower cost of investment as with conventional process• uniform industrially reproducible high quality

In fact there are few compelling reasons advising against a wide use of theMCM-IML process. Such reasons are• parts with areas too small for controlled deposition of the melt strip• significant undercuts at the fringes of the laminated area and• unavailability of space - even after tilting of the core - to securely deposit the melt strip.

It is appropriate to say that a comparison of cost i.e. for investment and manufacturingof parts produced by conventional high pressure moulding with subsequent press laminationto parts produced by the integrated MCM-IML method will result in cost savings up to 40%.On average a cost reduction of 20 to 25% is a realistic assumption.


MCM-LFT

The request from within the car component industry to develop the MCM-LFT process ismotivated by the intention to replace glass mat reinforced thermoplastics (GMT or Azdel)and to an extent SMC wherever possible for reasons like• cost reduction (material and processing)• simplification of process• reduction of thermal stress on the PP polymer• lower cost of investment• floor space requirements• more flexibility in regard to material analysis and characteristics etc.

At present there are two methods which have matured to continuous production appli-cation. The processing of long glass fiber reinforced granules (LFG) which come as littlerods and are readily available in Europe in lengths from > 10 mm (0.4 in) to 25 mm (1 in)from sources like (in alphabetical order)• Appryl (Elf-Atochem) of France, brand name "Pryltex"• Borealis of Finland, brand name "Nepol"• DSM of the Netherlands, brand name "Stamylan" and• Ticona (Hoechst) of Germany, brand name "Compel"

The pertinent equipment will also process recycled GMT (Azdel) material or blendswith LFG.

Parts already produced are bumper carriers, undercovers, and battery holders. Instru-ment panel carriers are under development.

In-line compounding and subsequent forming of LFT starting from roving.The process runs through the following stages:

• Roving filaments are unwound from bobbins under rupture control and guided throughtubes to a preheating station.

• The rovings are pulled through a pultrusion head fed by a single screw extruder with PPmelt for roving impregnation.

• The impregnated roving is taken in by screws of a corotating twinscrew extruder at apoint where the main matrix is already molten.

• Fiber length is determined by screw rotation.• The LFT melt is extruded into preforms to be placed in a compression mould or dis-

charged into an injection unit.Recycled material (GMT-Azdel, LFT) may be added by means of an additional single

screw extruder at the end of the compounding phase.Fiber lengths are varying however come with a length concentration around 10 mm

(0.4 in) to 80 mm (3.1 in).


The process is successfully applied for the production of VW Passat frontends since alonger period of time.

A typical MCM-LFT cycle is identical to the MCM-IML cycle exclusive of the cov-erstock placing phase. The stock temperature at the orifice of the deposition die will bearound 240°C to 260°C (- 460° to 500°F). However the cooling time aided by the better heatconductivity of glass fiber and the absence of coverstock will be about the same as with theMCM-IML technique.

Based on a wall thickness of 2 to 3 mm (- 0.08 to 0.12 in) a total cycle time of 60 sec bymanual demolding and 45 sec by robotic demolding can be achieved. Early demolding withsubsequent postcooling will shorten the cycle and help to control internal stresses.

The internal mould pressure required may vary from 80 bar (- 116 psi) to approx. 150bar (- 2180 psi) depending on the structure of the part, fiber lengths, and fiber content andhence the flowing properties of the stock.

For the MCM-LFT process most development and application efforts are directed tothe preservation of the original fiber lengths. Glass fiber appears to be mainly affected bypressure resulting in immediate contact of the fibers. Detrimental pressure may result frompressure development in the final flights and at the tip of plastication screws i.e. backpres-sure and/or by narrowing of the melt passage for extrusion or injection.

Adequate conditions during extrusion are provided for LFG (pellet) processing by pre-paring the melt on a single screw extruder with the following characteristics:• low compression, low friction screw of 30 D length, screw diameter min 90 mm (3.5

in) preferably 130 mm (5.1 in) with the low rotation speed of max 30 rpm• deep cut, thermolator controlled intake zone with tangential intake pocket• external heating by heater collars on the barrel supported by a screw core temperature

control• axially movable screw for back pressure control and melt ejection.

The above extruder is also capable of processing recycled material from GMT-Azdelor LFG.

Due to the most careful plastication conditions output capacities are in the range of 160kg/h (350 lb/h) for screw diameter 90 mm (3.5 in) to 240 kg/h (530 lb/h) for a 130 mm (5.1in) screw. Extruders with larger screws are available.

Twin screw extruders with corotating screw are designed for high mixing capacities atlow stock pressure i.e. they are high efficiency compounding extruders. They are not suit-able to process LFG or recycled materials.


MCM EQUIPMENT

In principle the main units of a MCM-IML plant are identical to those for MCM-LFT,adapted only for specific application engineering requirements.

CLAMPING UNIT

MCM clamping units are designed as hydraulic vertical 4 column presses. As cycles startwith an open mould for subsequent compression moulding special emphasis is laid on amechanically sturdy design including platens with low deflection, generously dimensionedcolumns with extent and precise guidings of the upper moving platen.

The guidings are about 2.5 times the length compared to conventional clamping unitsplus additional stiffening elements. During MCM-IML and most MCM-LFT applicationsnone or moderate tilting momenta occur which will be absorbed by the mechanical designfeatures.

In rare cases i.e. involving parts with a complicated structure, melt with poor flowingproperties, etc. requiring higher internal mould pressures along with critical tilting momentaa frame press with closed loop platen parallelism control may be necessary.

MCM clamping units are available with two hydraulic systems (reference: 8000 kNi.e. 880 US-t):

Figure 5. Schematic drawing of MCM-plant. 1. clamping unit, 2. injection unit, 3. melt strip deposition die, 4. hydraulic sys-tems, 5. electrical and electronic control units.


Hydraulic system based on variable displacement pumps with the following datarelevant to the process

closing speed mm/s 500 in/s 19.7 opening speed mm/s 330 in/s 13.0 compression speed mm/s 25 in/s 1.0 pressure build-up s < 2 s < 2

This specification is adequate for most MCM-IML applications. Whereas closing andopening speeds have marginal influence on the performance only, pressure build-up andhence compression speed may have decisive bearing on the moulding quality because of theinstantaneous flow in a cold mould and under high friction. Therefore an accumulatorassisted hydraulic system is advisable for certain MCM-IML applications and is essentialfor most MCM-LFT applications. The pertinent critical data for a Accumulator hydraulicsystem are

closing speed mm/s 800 in/sec 31.5 opening speed mm/s 800 in/sec 31.5 compression speed mm/s =50 in/sec = 2 pressure build-up s 0.9 sec 0.9

Clamping units with clamping forces from 4000 kN (440 US-t) to 15 000 kN (1650US-t) are available. At present units of 8000 kN (880 US-t) and 10 000 kN (1100 US-t)appear to be preferred. Platens are still mostly dimensioned to customer's needs. Howeverplatens with a mould mounting area (x by y) of 1400x1900 mm (i.e. 55x75 in) are much indemand as being suitable to accommodate two cavity moulds for standard left/right doortrims. The clamping units do not require pits.

INJECTION UNIT

MCM injection units follow the design principles of conventional injection unit modifiedfor increased plastication capacity and high injection rates at moderate injection pressures.

For MCM-IML plants the following frame work of data are an average specificationscrew diameter mm 70 - 110 in 2.8 - 4.3screw length /�' 20 - 24 /�' 20 - 24screw rotation speed min-1 140 - 170 rpm 140 - 170plastication capacity cm3/s 54 - 110 in3/s 3.3 - 6.7injection pressure bar = 700 psi =10 150injection rate cm3/s 700 - 1600 in3/s 42.7 - 97.6injection volume cm3 1100- 4100 in3 67.1 - 250.2


For MCM-LFT the corresponding data arescrew diameter mm 90 - 160 in 3.5 - 6.3screw length /�' 28 - 30 /�' 28 - 30screw rotation speed min-1 < 30 rpm < 30plastication capacity cm3/s 40 - 70 in3/sec 2.4 - 4.3injection pressure bar = 400 psi =5800injection rate cm3/s 720 - 2100 in3/sec 43.9 - 128.2injection volume cm3 1700 - 8000 m3 103.7 - 488.2

Screw core temperature control and pellet preheating devices will increase the plasti-cation capacity.

For the left and right part feature during one cycle MCM-IML injection unit are usu-ally equipped with 2 injection units.

For the melt deposition movements the injection units are mounted on motion axes i.e. x-axis (horizontal motion parallel with screw axis) driven by servo valve controlled hydraulic cylinder

speed max. mm/s 1500 in/s 59.1y-axis (horizontal motion perpendicular to screw axis) driven by ball screw spindle with servo AC drive

speed max. mm/s 500 in/s 19.7z-axis (vertical motion perpendicular to screw axis) driven by ball screw spindle with servo AC drive

speed max. mm/s 200 in/s 7.9In most cases the z-axis is executed for mold height adjustment only driven by synchronizedlifting spindles with AC motor

speed max. mm/s 10 in/s 0.4Strokes of axes are defined by the platen size or mould mounting area respectively if 2

injection units are provided x- and y-axes move independently.User demands for larger injection volume and LFT capabilities are calling for larger

i.e. heavier injection unit. This comes along with the request for faster movements of theaxes. This has lead to new developments separating the plastication function from injec-tion.

The plastication unit is a stationary single screw extruder purpose built for the specifictask i.e. thermoplastic or LFT processing. Size and weight is no restricting factor. The screwis still axially movable for fast melt transfer into the injectors.

Melt strip deposition is performed by one or two separate plunger injectors which aremounted on individual 3-axes coordinate table. As the weight of the injector is only about15% of an average size integrated plastication/injection unit fast dynamic movements by


servo controlled rack and pinion drives becomereality e.g.x-axis 1800 mm/s i.e. 71.0 in/sy-axis 1000 mm/s i.e. 39.4 in/sz-axis 500 mm/s i.e. 19.7 in/sspeeds which will help to shorten cycle inter-valls. The drive elements applied permit highaccuracy of positioning of the deposition die inthe mould (± 0.2 mm i.e. ± 0.008 in).

Plunger injection units yield high shot pre-cision and consistency. Melt transfer occursunder first-in-first-out conditions. During melttransfer from extruder to injection chamber theplunger is retracted in a controlled movement

for accurate dosing of the shot size.As there is no friction from a full screw as with screw injection a rather low injection

pressure is applicable i.e. < 250 bar i.e. < 3625 psi - conditions especially favorable for LFTprocessing.

Figure 6. MCM-plant with stationary extruder and 2 injection units (side view-rear view). 1 - stationary extruder, 2 - plunger injector, 3 - axes coordinate table, 4 - melt strip deposition die, 5 - hydraulic station, 6 - control cabinet.

Figure 7. MCM-plant with stationary extruder and 2 injection units (top view). 1 - stationary extruder, 2 - plunger injector, 3 - axes coordinate table, 4 - melt strip deposition die, 5 - hydraulic station, 6 - control cabinet.


MELT STRIP DEPOSITION DIE

Melt strip deposition dies are basically flat extrusion dies. They consist of two mould halves(sandwich design) with a melt feeding channel followed by a coat hanger type computeroptimized distribution channel. The standard version has a fixed lip gap and workingwidth. The actual strip dimensioning is achieved by interchangeable orifice attachmentsmounted on the die body by a quick mounting device.

In order to obtain a short distance between the orifice and the deposition surface ofthe mould the feeding channel is arranged in the injector side mould half i.e. deflected by90° to permit a perpendicular position of the die in relation to the mould. Lips and die bodyare heated.

The melt strip cutter consists of a cutting blade horizontally sliding over the orifice bya pneumatic drive. To prevent heat build-up the blade stays in the retracted position betweenthe cutting cycle. A hydraulic rotary shut-off valve in the adaptor between injector and dieretains the melt under pressure before injection thus preventing melt leakage from the die.

The importance of reduction of flow lengths was discussed earlier. Deposition of meltstrips close to the fringes of the cavity is a proven method. Deposition dies with a continu-ous width control will contribute to this target.

A patented device attached to a deposition die as described above is equipped with anumber of narrow cutting blades instead of one only. Each blade is actuated by an individualpneumatic drive. The cutting and retracting pattern is freely programmable interfacing withinjection rate to account for the changing cross-section of the melt strip. The device alsoallows to spare islands.

For melt accumulation in certain part areas a most reactive program to increase theinjection rate i.e. to accelerate screw or piston by a speed profile is available. This featurecan be further supported by a speed profile of the x-axis.

This unit has been successfully used to produce door trims without insert i.e. some kindof a frame.

Fig. 8 shows a MCM plant now being used in continuous production. The plantincludes a coverstock loading and demolding robot. The plant represents the present state ofthe art as described above.

OUTLOOK

The future will see developments in the following areas:1. refinement of the MCM-IML backcompression process to the efficiency totally

comparable to the basic conventional high pressure injection moulding process in regard toquality, productivity, including automation

2. replacement of the GMT-Azdel process by MCM-LFT for most application


3. introduction of natural fiber to partly substitute glass fiber and for improvement ofparts not yet reinforced and

4. replacement of some SMC applications by using the melt strip deposition methodswith new BMC type materials.

MCM will be an important process of the future.

Figure 8. MCM equipment layout.

In-mold Lamination Back Compression Molding

Thomas HuberSwiss Impulse, Inc., the partnership of Georg Kaufmann AG, Switzerland and

Delta Tooling, Auburn Hills, MI, USA

INTRODUCTION

During the process the coverstock is mechanically fused to the substrate through the influ-ence of heat. Peeling the coverstock off the substrate is no longer possible unless a weaklayer (e.g. foam) can be split.

The application has the following advantages:• no gluing or welding required• no special plant ventilation necessary• no solvent recovery necessary• no surface preparation needed• no tooling cost for blank substrate• no tooling cost for fixtures, vacuum forms, etc.• no storage of blank substrates• no fogging from interior car parts• exceeds all standard peel and pull tests• lower overall price per piece (up to 30% savings possible)• if coverstock and substrate are from the same family of plastic (e.g. PP) the part can be

recycledThe application has the following disadvantages:

• small parts cannot be properly filled by the melt strip• difficult tooling for significant undercuts at the edge of the parts• vertical press and special plastifier unit needed

The application has one big challenge:• coverstock

BACKGROUND

Conventional compression molding has been practiced for years. Best known application issheet compression and GMT (Glass Mat Technology). In the GMT process heated squares


of material or sheets are formed into shape with a mold installed in a fast closing, powerfulvertical press.

Back compression molding is also related to back injection molding, also known aslow pressure or reversed molding. However, pressure (up to 80/100 bars at the main nozzle)and high temperatures (220oC) are still needed to fill the mold. The coverstock is placedinto the mold before closing. Especially at the nozzles the coverstock is fully exposed dur-ing the filling of the mold and has to withstand local elevated pressure and heat. Good moldmakers can alleviate some of these challenges, but for larger parts there is no alternative to aspecialized hot runner system and controlled valve gating.1

Production and demand for back compression molded parts are rapidly growing inEurope. We estimate that approximately 40,000 to 75,000 parts are formed in this fashionevery day.

In the US, Ford Motor Company is producing almost all the door panels in a similarprocess which, however, requires a specially designed three layer vinyl coverstock.

At least a dozen US companies are experimenting with IML or are already in produc-tion.

PROCESS PARAMETERS

In back compression molding the clamping pressure can be significantly reduced, whichallows for presses with large platen and reasonably sized hydraulic units.

Clamping force for back compression molding is 5,000 to 7,000 kN for each m2 of theprojected cavity surface.2

There are no good measurements available, but it is estimated that the actual local pres-sure interacting with the compressed coverstock is only a fraction of the numbers above andcertainly far less than the local peak pressure seen at the drops during the filling and com-pacting of the mold with a hot runner.

The preferred clamping speed should be onthe slow side to achieve an even “rollout” of themelt strip, also avoiding abrupt forces on thecoverstock.

The molten plastic (melt strip) is placedinto the open mold (at zero pressure), onto thecore side through a wide nozzle, or it can beperked up through distribution channels (hotrunner) into the open or semi-closed mold. Thepolypropylene melt strip has a temperature of180-190oC. That is a 15% to 20% reduction in

Figure 1.

CAVITY

DECOR

NOZZLE

COOLING

HYDRAULIC CYLINDER

EJECTOR PLATE

In-mold Lamination 189

energy, when compared to injection molding. As a result, we have less internal stress, fastercooling cycles, and considerable energy savings (Figure 1).

The coverstock is either placed manually or by a robot on top of the molten plastic orhung in a frame attached to the moving cavity side. There is ample time (approximately 20seconds) to place the coverstock and close the press after the extrusion head is retracted.

PROCESS EQUIPMENT

Several machine manufacturers are offering equipment for the process, including a doublenozzle for filling simultaneously two cavity molds.

Most often used is a 8,000 kN vertical press. However, larger units are already on orderfrom various parts’ producers.

The injection unit (plastifier) can be moved along 2 or 3 axis. Special wide nozzle dieswith a cut-off device are needed. Such dies have heated and adjustable lips with an openingrange of 3 to 10 mm.

APPLICATIONS

Testing of the parts by the European automobile and furniture manufacturers producedexcellent results:• Lower internal stress in the substrate, resulting in less warping. BMW and Mercedes

are now installing most of these parts in the luggage compartments of their station wag-ons. Less warping also means that these parts do not need an additional trim to cover upproblems, which might develop later on.

• Quality and price lead to the decision for Mercedes to use the process for all the differ-ent doors in the new A-Class van.

• Robust appearance, easier installation, and price convinced a large US truck manu-facturer to retool for IML parts.

• Many flat parts (door inserts, console inserts, side wall panels) are excellent applica-tions for IML with back compression molding.

• Economical coverstock (felt) for trunk parts e.g. access doors and floor sections can beused in connection with a high content of recycled material.

ADVANTAGES

The following advantages can be listed for the back compression molding process:• less thermal exposure for the coverstock• no peak pressure, when compared with the pressure generated at the nozzles of injection

molding


• uniform wall thickness not a must any longer• especially suited for carpet and foam backed cover-stock (vinyl, fabric)• excellent results with larger parts (rear door claddings measuring 2 by 5 feet across)• tooling cost comparable to a conventional injection mold for the blank substrate• no hot runner system, no valve gating, no heating elements• adjustable filling (fill location and amount) for best results• less tool maintenance• in the substrates a high content of recycled materials can be used• with proper tooling a high degree of detailing in back of substrate possible • and last but not least: 20% to 30% reduction in piece price

One of the European automobile executives is quoted: “Presently we see no othertechnology that can achieve the same quality for a lower price!”

MOLD DESIGN

It is impossible to list or discuss all the design features of in-mold laminating tools. Thetooling is less complicated and less crowded, since the entire hot runner system does notexist. Complex detailing is often requested by the designers, especially in the trunk sectionof the car. Walls, clips, hooks, “dog houses”, etc. are designed into the back of the substrate.Such IML parts can double as holders of first aid kits, carriers for lifting jacks, or othertools.

The line of draw is very important, since only the movement and force of the press isfilling the last corner of the mold. These forces have to be properly vectored, otherwise,bonding between coverstock and the substrate is put in jeopardy. With IML the pressure andwear and tear on the diving edges3 are greater than in back injection molding. Forming thepart and closing the mold occur during the same time. Special guide plates are necessary toequalize the forces. The ejector plate is hydraulically operated and has to be fully retractedprior to deposit the melt strip onto the mold.

Cooling is very important to minimize internal stress and discoloration of the cover-stock. During the cooling cycle the coverstock acts as an insulation, and an asymmetric heatprofile will occur in the wallstock. The distribution and layout of the cooling channels isvery important (Figure 2).

Conventional design experiences and computer software are not applicable. There is nomold flow software available for this technology. Instinct and experience with IML moldingis important. Every part is unique in its demand. Collected design criteria for back injectionmolding can only serve as a guideline. Mold design for back compression molding hasbecome an art again. Common sense and experience with the application is important.

In-mold Lamination 191

CAVITY SIDE

Figure

Core Side

PART DESIGNFigure 2.

In-mold lamination is only going to be success-ful if the designers and stylists of the car compa-nies and tier one suppliers can be convinced totake full advantage of the process. In Europemany designers have been visiting GK for spe-cial design courses and discussions for the bestsolutions and compromises. It is important thatthese discussions are taking place as early aspossible during the design phase. The crucialpoints remain the same:

Raw Edges and 180° Folds (Figure 3)Figure 4. IML is producing a strong bond between

coverstock and substrate all the way to the edge.There is no danger that the coverstock will peel off at the edge!

IML parts (without hand-wrap) are possible for even the most discriminating taste, forappearance and rattle free performance.

Wall Thickness, Ratios to Ribs (Figure 4)In general the same rules apply as for normal injection molding. However, a ratio of 1

to 1 without any sink marks is not uncommon.Walls up to a depths of 120 mm are possible without filling problems.It becomes possible to design parts with different wall thickness. The drawings shows a

back rest for an office chair.4 Ribs are needed for structural stiffness.Different Coverstock Materials (Figure 5)The practical range of different coverstock applications is restricted by the diving

edges of the mold. It is therefore important to know what coverstock material will be used.


However, it is conceivable that differentmold halves are used for different production.Such halves could also be designed for partswhich will be flocked or covered with leather.Toady's CAD/CAM systems allow extremeclose tolerances, which make interchangeablehalves possible.

COVERSTOCK

Figure 5. The selection of the proper coverstock is impor-

tant. The specifications from the car manufactur-ers usually do not reflect in-mold lamination material.

As a mold maker we are not an expert in textile and vinyl. We recommend, however,that every molder is lobbying with the coverstock suppliers for close cooperation regardingIML:.all materials have to have a stretch factor, which should be omni directional.if the material is not dense enough or sensitive to heat, a protective layer is needed (e.g.

fabric or fleece) which is laminated to the back of the coverstock.colors have to withstand fading while exposed to the heat during the molding process."stiff' coverstock combined with curved parts most likely have to be preformed.the back, or back layer, of the coverstock has to be compatible with plastic used for

forming the substrate in order to achieve proper bonding.thin film (e.g. on exterior car parts) has to be extremely flexible, not showing any

stretch marks when elongated

CONCLUSIONS

IML back compression molding already has a proven track record in Europe and has estab-lished itself as an economical solution to produce a variety of interior car parts. As morevertical presses and plastifier units are placed into production the technology will also findits way into other applications, like the furniture industry.

REFERENCES

Plast Europe 4/94, Vol 84, G. Bagusche, Unitemp SA, Switzerland.Example: A one cavity door panel projection is 0.5 m2. Necessary clamping force (or press size) is max. 3,500 kN,or in other words, you need a press size of approx. 400 US tons.Diving edge is the specially designed parting line to accommodate the coverstock without leaving any gaps for flashing.Plast Europe 7/97, Volume 87; M. Zwetz, Waldshut, Germany.

OFFICE CHAIRBACKREST

Analysis and Characterization of Flow Channelsduring Manufacturing of Composites

by Resin Transfer Molding

R. V. Mohan, K. K. Tamma, S. Bickerton, S. G. Advani and D. R. ShiresProcesses and Properties Branch, Materials Division, U. S. Army Research Laboratory,

Aberdeen Proving Grounds, MD 21005, USA

INTRODUCTION

The success of Liquid Composite Molding (LCM) processes such as Resin Transfer Mold-ing (RTM) depends on the complete impregnation of the fibrous reinforcing preform. Varia-tion in volume fraction around bends and corners creates gaps around the fiber preformwithin the mold cavity. The deformation of the fiber mat in the preforming stage,1-3 causedby bending, shearing and stretching, creates variations in permeabilities and reductions inthickness in the sheared preform regions. These physical variations of the preform duringpreform lay-up in the mold create regions of higher porosity, hence increasing the local per-meability. Such regions offer the injected resin paths of least resistance, significantly alter-ing the shape of the resin flow front, and the injection and mold pressures. Thisphenomenon is called race tracking4,5 and often leads to dry spot formation and otherdefects.

While race tracking has traditionally been viewed as an undesired effect in RTM,related processes such as Vacuum Assisted RTM (VARTM) have employed race trackingeffects to improve and facilitate mold filling. Race tracking may be purposely createdwithin a mold through the use of “flow channels” or “runners” to improve the flow distribu-tion, and direct the resin to areas that may be difficult to fill successfully. Pre-determinedflow channels may also reduce the time required to fill the mold under constant pressureinjection, and reduce the injection and mold pressures in constant flow rate injection. Thebeneficial effects of flow channels have not been explored fully in the context of RTM, eventhough the potential exists. Process simulation studies demonstrating the effect of flowchannels on the flow front progression and pressure histories in RTM structural componentsrelevant to army applications are presented by Mohan et al.6 A clear understanding of theeffects of mold filling due to race tracking caused by channels is needed to fully explore thebenefits obtainable through the use of flow channels.


Manufacturing process simulations under the paradigm of Virtual Manufacturing havebeen successfully employed for RTM processes. Various manufacturing process modelingsimulation tools analyzing the progression of the resin inside a mold cavity are currentlyavailable and have been applied successfully to various net shape parts manufactured byRTM.7-10 Race tracking effects due to channels can be accounted for in mold filling simula-tions by applying constant values of higher permeability within regions of a mold represent-ing the flow channels.11-13 The process simulation predictions depend on the modelsemployed to predict the permeabilities applied to the flow channels. Experimental studiesinvolving the flow visualization in transient molds as well as investigation of flow charac-teristics using Laser Doppler Anemometry (LDA) techniques to experimentally measure thevelocity profiles of the fully developed race tracking flow have been conductedearlier11,14,15 to study the race tracking phenomenon in liquid composites molding. Thesestudies correlate and justify the use of equivalent permeability models with the experimentalobservations.11 Experimental investigations studying flow behavior due to the presence ofgaps along the edges of a planar rectangular mold cavity have been conducted and corre-lated with simulations. These studies involved the application of equivalent permeabilitiesto the air gaps present within the mold, justifying their use in process simulations.12

The present work experimentally investigates mold filling in the presence of flowchannels within a simple mold configuration. Investigations involve different sized squarechannels on the face of a simple rectangular plate mold cavity. Experimental flow front andpressure measurements (injection and mold pressures) are employed to analyze and charac-terize the flow through channels and determine the optimal equivalent permeability modelfor the flow channels. Experimental observations based on transient flow fronts and tran-sient pressures are compared with simulations based on various equivalent permeabilitymodels which are employed in simulations. The equivalent permeability models provide agood effective process design tool to understand the flow and pressure behavior when flowchannels are involved and provide a simulation tool to ensure successful impregnation offiber preforms. The experimental data and numerical simulations presented, demonstrateand validate the effect of flow channels in reducing the injection and mold pressures andredistributing the flow.

EXPERIMENTAL STUDY: EFFECT OF FLOW CHANNELS

Experiments to study and understand the effects of flow channels on flow front progressionand pressure histories were completed using a simple two piece mold, the male and femalepieces constructed from 0.0254 meters (1 inch) clear acrylic plate. The mold cavity is a pla-nar square 0.3556 by 0.3556 meters (14 by 14 inches), having a thickness of approximately0.003175 meters (1/8 of an inch). There is a single injection port at the center of the plate,

Analysis and Characterization of Flow Channels 195

while a hole of 0.0127 meters (1/2 inch) diame-ter is cut in the center of each preform. The cut

in the preform eliminates the need for the fluid to flow through the thickness of the preformat the injection site. Pressure data was collected using pressure transducers screwed into thefemale portion. The pressure transducers were located at the injection port, 2 inches and 3inches away from the injection ports along the central symmetric lines. To study the effectof channels, grooves were cut on the face of the male mold plate. The injection port is at thecenter of the male piece. The flow channels have square cross-sections and two differentsized channels were employed. The cross-sectional dimensions of these channels are0.003175 x 0.003175 meters (1/8 x 1/8 inches) and 0.001984 x 0.001984 meters (5/64 x 5/6inches). The geometry of the part with and without channels is shown in Figure 1. The malepiece with the channels cut is shown in Figure 2 (a). With the injection at the center of theplate, these channels create a preferential path for the resin flow.

Figure l. Flat plate mold geometry.

Figure 2. Schematic of male mold piece and experimen-tal setup.

ALL DIMENSIONS IN INCHES

ALL DIMENSIONS IN INCHES

(a) Male mold piece with grooves(a) Flat plate part - no channels

(b) Flat plate part - with channels

(b) Experimental setup


Experiments were completed with the constant flowrate injection system representedschematically in Figure 2 (b). The transparent mold permitted the fluid flow front to be visi-ble throughout the experiment and recorded to video tape. A video camera was placeddirectly above the mold capturing a plan view of the flow front progression. The experimen-tal system consists of a control mechanism which continually monitors the pressures dropsand adjusts the pressure applied to the pressure pot to give a constant flowrate injection.16

The experimental system permits continuous measurement of fluid flowrate into the moldand pressure just prior to the injection port. The pressure transducers attached to the femalemold piece measured pressures directly within the mold. Due to the corrosive nature, andthe cleaning challenges provided by the actual resin systems, a substitute fluid was used.Since we are concerned only with the mold filling stage of RTM, curing need not be consid-ered. The fluid chosen for the flow visualization studies was a mixture of corn syrup, waterand clothing dye. The corn syrup mixture is Newtonian and the viscosity is easily character-ized.

Several experiments were conducted to study the effect of channels on flow front pro-gression and mold pressure histories. Two different styles of preform were studied, continu-ous strand random fiberglass mat, and a continuous strand stitched fiberglass mat, with towsrunning in the 0, +45, and -45 degree directions. Two volume fractions were studied foreach material, 19% and 26% for the random mat, and 44% and 51% for the stitched matrespectively. Each fiber preform configuration involved three sets of experiments, one withno channels in the male mold and the two experiments with the two different sized channels.The viscosity and flowrate were maintained to be nearly equal within each set of experi-ments. The resulting cavity thicknesses for each experiment were measured, as this is animportant parameter provided to the numerical simulations presented later. Due to varyingamounts of flexure of the acrylic mold pieces, the nominal cavity thickness of 0.003175meters (1/8 of an inch) could not be assumed.

Digitized video frames from several experiments are presented in Figures 3 and 4. Fig-ure 3 presents frames captured at 4, 10, and 21 seconds into the 19% volume fraction ran-dom mat experiments. Frames are shown from each of the three separate experiments.Figure 4 presents frames captured at 15, 37, and 49 seconds into the 44% volume fractionstitched mat experiments. The effect of the channels on the flow distribution is clearly seenfrom these video images. These figures also clearly demonstrate the increased deformationof the flow front as the channel sizes are increased.

Figure 5 compares the three injection pressures and pressures at the 2 inch transducerlocation obtained from the random mat experiments presented in Figure 3. Figure 6 com-pares the three injection pressures and pressures at the 3 inch transducer location obtainedfrom the stitched mat experiments presented in Figure 4. The pressure histories shown here


clearly illustrate the reduction in injection and mold pressure due to the presence of thechannel. This reduction could be used effectively in making large composite structures withlow pressure RTM.

FLOW MODELING

The flow modeling simulations for RTM are based on modeling the resin flow through afiber preform as a pressure driven flow through a porous medium characterized by Darcy'slaw.17 Darcy's law relates the velocity field to the pressure gradient, through the fluid vis-cosity and the fiber preform permeability. Permeability is a measure of the resistance a fluidexperiences when flowing through a porous media and is an important material characteris-tic involved in flow simulations. The mold filling simulations involve solving both the pres-sures and tracking of the flow fronts which can be multiple as well as converging anddiverging inside a mold cavity, modeled by a Eulerian numerical mesh geometry. Mold fill-ing simulations based on finite element-control volume approaches4,18,19,20 are based onsolving the transient mold filling problem as a quasi-steady state problem along with anexplicit time advancement scheme for advancing the flow front and tracking the filledregions.

Figure 3. Video frames of flow fronts - Random mat; Top row (No channels); Middle row (5/64 inch channels); Bottom row (1/8 inch channels).

Figure 4. Video frames of flow fronts - Stitched mat; Top row (No channels); Middle row (5/64 inch channels); Bottom row (1/8 inch channels).

4 Seconds 10 Seconds 21 Seconds 15 Seconds 17 Seconds 49 Seconds


The flow modeling simulations utilized in this paper for tracking the flow and pres-sures inside a closed mold cavity are based on a pure finite element based methodology forRTM simulations developed by Mohan et al.7 These computational developments for theRTM flows are based on a transient mass balance equation for the resin mass while employ-ing Darcy's law for the velocity fields in conjunction with an implicit filling technique totrack the flow fronts and filled regions. The transient mass balance equation for the resinmass inside the mold cavity involves the unknown variables pressure and fill factor, ,which denotes the fill nature of the mold cavity region. Finite element discretizations areintroduced for the pressure and the fill factor field to yield a system of discretized equations.

Ψ

Figure 5. Effect of channels on pressures : ra ndom mat experimental compar isons. Figure 6. Effect of channels on pressures stitched mat

experimental compar isons.

9000080000700006000050000

40000

40000

40000

30000

30000

3000020000

20000

20000

10000

10000

10000

0

00

00

0 00

10000

35000

35000

25000

25000

15000

15000

5000

5000

12000

8000

6000

4000

2000

5

5 1010

10 10

15

15

20

20

20

20

3030

303025

25 35

35

40

40

50

50

60

60

70

70Time, s

Time, s Time, s

Time, s

(b) Pressures at 2 inch transducer location (b) Pressures at 3 inch transducer location

No channels1/8 inch channels

No channels1/8 inch channels

5/64 inch channels 5/64 inch channels

Pre

ssu

re, P

a

Pre

ssu

re, P

aP

ress

ure

, Pa

Pre

ssu

re, P

a

(a) Injection pressures (a) Injection pressures

No Channels No Channels1/8 inch channels 1/8 inch channels5/64 inch channels 5/64 inch channels


The resulting discretized system of equations for the pressure and fill factor are solved in aniterative manner till mass convergence is obtained. The theoretical and computationaldetails and the numerical and physical advantages of this methodology are discussed else-where.7,8,21 One significant advantage of the flow modeling strategy employed here is thatthe computed location of the flow front at any instant of time does not depend on the timestep size employed to reach that stage. With effective models to characterize the permeabil-ities of flow channels, the flow modeling simulations can be effectively used to study thefiber impregnation and mold pressures when the flow channels are involved to determineand investigate optimal process parameters and mold designs.6

FLOW CHANNELS AND RTM SIMULATIONS

The flow channels are open regions, not containing any porous material. The existing RTMprocess modeling simulation tools can be effectively employed by characterizing the flow inthe channels in terms of average flow capacity in the channel cross-sectional area. Thechannels are thus treated as a porous medium whose material characteristics are quantifiedby a variable called equivalent permeability. Equivalent permeability approaches have beenverified both theoretically and experimentally11,14,15 and have been employed effectively inRTM flow simulations to model the constitutive flow behavior due to race tracking in sim-ple one-dimensional mold geometries.12,11

In this paper, flow in channels is modeled based on equivalent permeability models.Permeabilities of flow channels in our numerical models have been quantified with equiva-lent channel permeabilities. The equivalent permeability in terms of Darcy's law is definedas the average velocity divided by the driving force. The steady state flow in channels canbe described by a fully developed pressure driven flow as described by the Poisson equa-tion, with zero-velocity at the mold surface, and slip velocity at the interface.11,22 For thetype of square channels involved in this study, two different permeability models are avail-able and the equivalent permeabilities of these models are presented in Table 1. Other mod-els, including the parallel plate model, tend to be over predictive.

Model 2, presented in Table 1, includes effects on flow resistance in the channel due tothe presence of the permeable wall of the preform.11 However, the resulting expression forequivalent permeability is complex, and requires the specification of the permeability of thepreform, and alpha, an empirical slip constant. Though preform permeability is typicallywell known, alpha is very difficult to measure, and has no real physical significance.14,15 Asignificant simplification can be made if it is assumed that we have zero velocity at the per-meable wall of the preform. The resulting formulation based on a fully developed duct flowthrough a rectangular cross-sectional channel is presented as model 1 in Table 1.23 Whenemploying this expression, both the preform permeability and alpha are not required, and

Special Molding Techniques200

Table 1. Equivalent permeability for rectangular channel sections

the formulation is simpler to use. Studies indicated that there was some error in the expres-sion presented in reference 11 for model I. All of the simulated results presented in this paper

are based on equivalent permeabilities calculated from model I. Simulations were carriedusing both, however it was found that there was never more than an 8% diffference in per-meability for various alpha values. This resulted in a small adjustment to the predicted pres-sure histories and flow front progressions. Also, as we had no measurements of alpha, littlejustification could be found in applying the permeable wall model.

ANAL VSIS AND DISCUSSION

In this section, the flow fronts and pressure histories from the numerical process modelingsimulations are compared and discussed with the experimental results, to understand thebehavior of the equivalent permeability models in simulations. The process modeling simu-lations are based on 2.5-D thin mold geometries and flow models. Simulations presentedhere were carried out to compare the flow fronts and pressure histories with the experimen-tal observations and measurements.

Since the experiments were considered to be completed before the fluid reaches theend of the preform, we are interested only in the predicted flow front progression and pres-sure histories before the flow fronts reach the edge of our simulation model. The fiber pre-forms employed in the experiments were square sections of dimensions 0.3302 x 0.3302meters (13 x 13 inches), which corresponds to a half width of 0.1651 meters (6.5 inches).Hence for comparisons with experiments, we are interested in simulated flow fronts up untilthe point that they have travelled a maximum of 0.1651 meters (6.5 inches) from the injec-


tion point. Taking advantage of symmetry, simulations were completed using a mesh havinga quarter circular plate of radius 0.2032 meters (8.0 inches), composed of 4 noded quadrilat-eral elements with appropriate thicknesses representing the cavities measured from theexperiments. Two-dimensional quadrilateral elements are employed to model channels.

Figure 7 shows the flow front contourswhen no channels are involved for the randommat and stitched experiments detailed earlier.All the simulated flow front contours presentedare staggered to show details on a finer scaleduring the time of interest for which the experi-mental results are available. Simulations arebased on the physical parameters listed in Table2, which are measured during the experiments.The fiber volume fraction for the random mat is0.197 and that of the stitched mat is 0.444. The

channel permeabilities for the 0.003175 meters (1/8 inch) channels is computed to be3.5427E-07 m2, and for the 0.001984 meters (5/64 inch) channels is 1.3838E-07 m2 basedon model 1 discussed earlier.

Table 2. Physical measured parameters for simulations

Random mat experiment

ExperimentPermeability

Flow rate, m3/s

Viscosity, Pas Thickness, mK11, m

2 K22, m2

No channels 4.56E-09 4.56E-09 9.552E-06 0.228 0.0038862

1/8 channels 4.56E-09 4.56E-09 9.643E-06 0.226 0.0038862

5/64 channels 4.56E-09 4.56E-09 9.562E-06 0.228 0.0038862

Stitched mat experiment

No channels 2.16E-09 5.657E-10 1.953E-06 0.092 0.0032512

1/8 channels 2.16E-09 5.657E-10 1.947E-06 0.0925 0.0032512

5/64 channels 2.16E-09 5.657E-10 1.947E-06 0.0925 0.0032512

Figure 7. Simulated flow fronts - No channels; (Refined contours represent flow fronts at successive 2.0 second intervals for random mat and 4.3 second intervals for stitched mats).

Stitched MatRandom Mat


From Figure 7, it is clear that the simulated flow fronts with no channels match the

Figure 8. Simulated flow fronts - 1/8 inch channels ; Random mat; (Refined contours represent flow fronts at successive 2.0 second intervals).

Figure 9. Simulated flow fronts - 1/8 inch channels ; Stitched mat; (Refined contours represent flow fronts at successive 4.3 second intervals).

Figure 10. Random mat: Comparison of pressure histo-ries; 1/8 inch channels.


Isometric View Plan View Isometric View Plan View

Exp. - Inj. Exp. - Inj.Sim. - Inj. Sim. - Inj.


Plan V;ew Isometric View Plan View

Figure 9. Simulated flow fronts -118 inch channels ;Stitched mat; (Refined contours represent flow fronts atsuccessive 4.3 second intervals).

Figure 8. Simulated flow fronts -118 inch channels ;Random mat; (Refined contours represent flow fronts atsuccessive 2.0 second intervals).

40000

35000

"' 300000..25000

QJ~ 2000014~ 15000

0. 10000

5000

0

60000Exp. -InjSim. -Inj

EXp. .Inj. .Sim. .Inj. - 50000

~Q, 40000

4)~ 30000III01GIIoi 20000Q,

10000

:;

015 20 25 30 35

Time, 8

0 5 10 15 20 25 30 35'l'i1\1e, S

0 5 10

(a) Injection pressures (a) Injection pressures

35 10 15 20 25 30 35

Time, s0 5 10 15 20 25

Time, s30 0 s

(b) Pressures at 2, 3 inch transducer locations(b) Pressures at 2, 3 inch transducer locations



From Figure 7, it is clear that the simulated flow fronts with no channels match the

Isometric View


experimental results well. The simulated andexperimental pressure histories (not presentedhere) were in agreement. These results providethe authors with confidence in the experimentalprogram and mold filling simulation process.

The simulated flow fronts of the randommat experiment having 0.003175 m (1/8 inch)square section channels are shown in Figure 8,and those for the stitched mat experiments areshown in Figure 9. From these figures and theexperimental observations presented earlier, it isclear that the simulated flow fronts provide avery good qualitative match. The simulated flowfronts based on the smaller 0.001984 m squaresection channels showed similar behavior andare not presented here. Pressure histories, whichis a more rigorous test, based on simulations arecompared with experimental pressure historiesfor both channel sizes in Figures 10 and 11 forthe case of random mat, and in Figures 12 and 13for the stitched mat experiments.

Comparisons of the experimental andnumerically predicted flow fronts indicate thatthe fluid flows faster through the channels in theexperiments, compared to simulations. This maybe attributed to how the injection conditions are

modeled in simulations. Simulations are based on distributing the injected flow through thenodes comprising the inner radius and channels; Hence it is seen that there is some flow intothe fiber preforms even before the channels are filled. In reality, it is more likely that themajority of fluid will flow into the channels, with a smaller amount flowing directly into thepreform. Injection conditions must be modeled very elaborately to capture this behavior.

The 2.5-D thin cavity flow simulations cannot capture accurately the 3-D flow effectsseen in the experiments. The nature of the simulation geometry used drives the flow into thepreform as soon as the channels are filled, while during the experiments, the fluid must flowthrough the thickness of the preform after leaving the channels and spreading into the fiberpreform. This effect is more noticeable in preforms with higher volume fractions, as resis-tance to flow through the thickness increases.

Figure 12. Stitched mat: Comparison of pressure histo-ries; 1/8 inch channels.


By further improving the modeling of flow at the injection ports, the flow fronts basedon the simulations may be further improved, thus providing better comparisons with experi-mental observations. A full 3-D flow simulation would take into account the flow throughthe thickness, but will be computationally expensive and would require transverse perme-ability data which is difficult to measure.24,25 However, the thin 2.5-D flow simulations,used in conjunction with equivalent permeability models, provide very good qualitativecomparisons that improve as the experiments proceed, and the flow fronts move away fromthe injection point and flow channels. The equivalent permeability models based on ductflow are based on steady state models while mold filling flow is highly transient. This is truemore in the initial stages, which are not effectively simulated.

A comparison of the experimental and simulated pressure histories provide some inter-esting observations. For the case of the random fiber mat with lower volume fraction, theexperimental pressures at the 2 and 3 inch pressure locations start early compared to thesimulations. This start can be attributed to the early differences in flow front progressionseen between the experimental and simulated flow fronts, and the lower volume fraction ofthe random mat providing little resistance to flow through the thickness of the preform.These properties allow the fluid to reach the pressure transducers quickly. In the stitchedmat experiments presented here, the experimental pressures at the 2 and 3 inch locations lagbehind the simulated pressures. This is interesting to note, as fluid fills the channels veryquickly during these experiments. This can be attributed to the finite time it takes for thefluid to penetrate through the thickness of the preform, initiating a delayed time response ofthe transducers located on the face of the female mold below the fiber preform in the exper-iments. This time lag could be effectively addressed if full 3-D flow simulations were used.

Figure 13. Stitched mat: Comparison of pressure histories; 5/64 inch channels.


The pressure histories also show significant differences in the slope and characteristicsbetween experiments and simulations which can be attributed to the combination of effectsdiscussed earlier. Experimental uncertainties in the measurement of cavity thickness, vis-cosity, flowrate and other general errors during experimentation, can also attribute to differ-ences in pressure curve characteristics seen in the experiments and the simulations.

Furthermore, despite more sophisticated analysis required to accurately capture thetransient and 3-D flow effects with channels, the 2.5-D process simulation models, used inconjunction with equivalent permeability models, provide effective process design and anal-ysis tools for understanding the effects of channels on mold filling. These simulations canbe used as an effective tool to predict problems than may occur during mold filling. Theseallow the designer to make modifications before the mold designs have been finalized in avirtual environment, ultimately leading to process maturation.

CONCLUDING REMARKS

The positive effects of controlled race tracking during composite manufacturing by RTMhave been demonstrated through experiments involving a simple flat plate mold geometry.While race tracking is normally considered to be undesirable, flow channels can be effec-tively employed in RTM mold designs to improve the flow distribution and reduce moldpressures. As channels serve to reduce pressure within a mold cavity, injection flowratesmay be increased while keeping pressures within permissible levels thereby improving theprocess cycle times. Experimental flow visualization results presented have clearly demon-strated the dramatic effect of channels, even in the simple mold cavity geometry employedhere. The experimental flow front progressions and pressure histories have been comparedwith process simulation results based on a pure finite element based methodology. The flowin the channels is modeled to be Darcian, quantified by equivalent permeabilities, modeledon steady state fully developed flow through a rectangular cross-sectional duct geometry.Comparisons indicate that process simulation models based on 2.5-D thin geometries andflow models, in conjunction with equivalent permeability models, provide effective processdesign and analysis tools for understanding the influence of channels during mold filling.However, a more elaborate analysis of the injection conditions for simulations, equivalentpermeability characterization models to account for the transient flow nature, and three-dimensional models to take into account physically observed 3-D flow effects are neededfor a more accurate analysis and understanding of the channel effects on the local flow andpressure histories near a fiber preform. Further investigations involving constant pressureexperiments and analyses to include 3-D flow effects through full three-dimensional effectsare planned.


ACKNOWLEDGMENTS

This work is supported by ARO grant number DAAH04-96-1-0172 through University ofMinnesota, and by ARL grant number DAAL01-95-K-0086 through University of Dela-ware. Special thanks to Mr. Nam Ngo of University of Minnesota for the earlier work onflow modeling studies. Thanks are also due to Dr. Andrew Mark, Mr. Walter Roy and Dr.Shawn Walsh of U. S. Army Research Laboratory for their support and encouragement.

REFERENCES� L. Fong and S. G. Advani. The role of drapability of fiber preforms in resin transfer molding.

In 9th Americal Society for Composites, page 1246, 1994.2 S. Bickerton, S. G. Advani, L. Fong, and, K. Fickie. Effect of draping of fiber preform on process

parameters during manufacturing with resin transfer molding.. In 11 th Annual ESD Advanced Composites Conference and Exposition, Dearborn, MI, November 1995.3 P. Simacek and S. G. Advani. Permeability model for a woven fabric. Polymer Composites (in press).4 M. V. Bruschke and S. G. Advani. A finite element/control volume approach to mold filling in anisotropic porous media. Polymer Composites, 11(6):398-405, December 1990.5 K. Han, C. H. Wu, and L. J. Lee. Characterization and simulation of resin transfer molding - race tracking and dry spot formation. In 9th Annual ASM-ESD Advanced Composites Conference and Exposition,

pages 286-300, Dearborn, MI, 1993. 6 R. V. Mohan, D. R. Shires, K. K. Tamma, and N. D. Ngo. Flow channels/fiber impregnation studies for the process modeling/analysis of complex engineering structures manufactured by resin transfer molding. In ASME International Mechanical Engineering Congress and Exposition, Atlanta, GA, November 1996.7 R. V. Mohan, N. D. Ngo, K. K. Tamma, and K. D. Fickie. On a pure finite element based methodology for resin transfer mold filling simulations. In R. W. Lewis and P. Durbeta, editors, Numerical Methods for Thermal Problems, volume IX, pages 1287-1310, Atlanta, GA, July 1995. Pineridge Press.8 R. V. Mohan, N. D. Ngo, K. K. Tamma, and K. D. Fickie. A pure finite element based methodology for resin transfer mold filling simulations. Technical Report ARL-TR-975, U. S. Army Research Laboratory, March 1996.9 M. V. Bruschke and S. G. Advani. A numerical approach to model non-isothermal, viscous flow with free surfaces through fibrous media. International Journal of Numerical Methods in Fluids, 19:579-603, 1994.10 B. Liu, S. Bickerton, and S. G. Advani. Modeling and simulation of resin transfer molding : Gate control, venting and dry spot prediction. Composites - Part A, 27A:135-141, 1996.11 J. Ni, Y. Zhao, L. J. Lee, and S. Nakamura. Analysis of race tracking phenomenon in liquid composite molding. In 11th Annual ESD Advanced Composites Conference and Exposition, Dearborn, MI, November 1995.12 S. Bickerton and S. G. Advani. Characterization of corner and edge permeabilities during mold filling in resin transfer molding. In ASME AMD-MD Summer Annual Meeting, Los Angeles, CA, June 1995.13 S. Bickerton and S. G. Advani. Experimental investigation and flow visualization of the resin transfer molding process in a non-planar geometry. Composite Science and Technology (in press).14 S. Gupte and S. G. Advani. Non-darcy flow near the permeable boundary of a porous medium:

An experimental investigation using Ida. Experiments in Fluids (in press).15 S. Gupte and S. G. Advani. Flow near the permeable boundary of aligned fiber preforms.

Polymer Composites (in press).16 J. Mogavero. Compression characterization and resin infiltration of multi layered preforms in resin transfer molding. Master's thesis, University of Delaware, October 1996.


17 H. Darcy. Les Fontaines Publiques de la Ville de Dijon. Delmont, Paris, 1856.18 C. A. Fracchia, J. Castro, and C. L. Tucker. A finite element/control volume simulation of resin transfer mold filling. In American Society for Composites - 4 th Technical Conference, Lancaster, PA, 1995.19 W. B. Young, K. Han, L. Fong, and L. J. Lee. Flow simulation in molds with preplaced fiber mats. Polymer Composites, 12(6):391403, December 1991.20 F. Trouchu, R. Gauvin, and D. M. Gao. Numerical analysis of the resin transfer molding process by the finite element method. Advances in Polymer Technology, 12(4):329-342, 1993.21 R. V. Mohan et al. Process modeling and implicit tracking of moving fronts for threedimensional thick composites manufacturing. In AIAA-96-0725, 34th Aerospace Sciences Meeting, Reno, NV, January 1996.22 G. S. Beavers and D. D. Joseph. Boundary conditions at a natually permeable wall. Journal of Fluid Mechanics, 30(1):197-207, 1967.23 S. Bickertorn et al. Important mold filling issues in liquid composite molding processes

Modeling and experiments. In To appear in the Proceedings of the Annual Meeting of the Society of Plastics Engineers : ANTEC 97, Toronto, Ontario, May 1997.

24 R. S. Parnas, J. G. Howard, T. L. Luce, and S: G. Advani. Permeability characterization. part 1: A proposed standard reference fabric for permeability. Polymer Composites, 16(6):429-445, December 1995.25 T. L. Luce, S. G. Advani, J. G. Howard, and R. S. Parnas. Permeability characterization. part 2: Flow behavior in multiple-layer preforms. Polymer Composites, 16(6):446-458, December 1995.

Optimization of Channel Designin VARTM Processing

Roopesh Mathur and Suresh G. AdvaniCenter for Composite Materials, and Department of Mechanical Engineering,

University of Delaware, Newark, DE 19716, USABruce K. Fink

Army Research Laboratory, Aberdeen Proving Ground, MD 21005, USA

INTRODUCTION

The Vacuum Assisted Liquid Molding VARTM(Figure 1) process1,2 has been used for the man-ufacture of large composite parts. In this pro-cess, a preform is placed in an open mold and aplastic vacuum bag placed on top of the mold. Avacuum is created in the mold using a vacuumpump. A resin source is connected to the mold.As vacuum is drawn through the mold, resininfuses into the preform.

VARTM is finding increasing application inthe manufacture of large parts with complexgeometry such as panels of all-composite buses,

railroad cars and armored vehicle components.3 These parts are manufactured after the pro-cess has gone through a costly and time consuming development cycle. This developmentcycle is empirical and experimental and requires considerable amount of effort and exper-tise. However, it is unclear whether the manufacturing process is efficient and cost-effec-tive. The actual processing conditions may differ from that of the development cycle on apart-to-part basis. Hence there is a need for the scientific study of the manufacturing processand the development of objective process efficiency criteria that will facilitate cost-effectivemanufacturing. The increasing complexity of components and new processes such as Co-injection Resin Transfer Molding4 will lead to an increasing need for optimization of RTMbased processes. The filling and cure of the part can be simulated by several packages

Figure 1. Vacuum assisted resin transfer molding.

Fiber preform

Vacuum pump

MoldResin injection

Fiber preformunder vacuum

Resin injectionResin impregnates

fibers and cures Cured part


developed for this purpose, such as Liquid Injection Molding Software or LIMS,5 which hasbeen developed at the University of Delaware.

The process of resin infusion of large parts with complex geometry and varying pre-form properties have been examined. Often, such parts have channels cut in them in order toensure that resin reaches all sections of the part. The configurations of the channels deter-mine the mold filling time and minimize dry spot formation. However since the volumeoccupied by the channels contains pure resin, the weight of the finished part is higher. Thecosts of tooling and setting-up a complex system of channels should also be taken intoaccount.

A channel configuration having wide channels reaching all parts of the mold would beable to fill the part in less time. But the costs in weight and tooling have to be accounted for,to reflect manufacturing needs. A Process Performance Index is proposed here, which pro-vides an objective measure of manufacturing efficiency and enables the comparison of dif-ferent channel configurations. The performance index has been tested out on a test part,which has complex shape, large size and varying material properties in the preform and theresults are discussed.

A PROCESS PERFORMANCE INDEX

A Process Performance Index can be defined as a quantification of manufacturing efficiencyand cost effectiveness. It should contain important parameters and criteria that effect themanufacturing process. In this study, three different criteria have been considered: the timeto fill the part, the area of the surface area of the part occupied by the channels and thelength of the channels. The time to fill is the most important of the three factors and itreflects the fact that the configuration of the channels with respect to the fibers in the pre-form determines the time it takes to fill. The area of the channels and their length is propor-tional to the dead weight of resin in the channels, which should be kept to a minimum. Thelength of the channels reflects the cost of machining the channels into the part as well as thepressure required to deliver the resin to the channel. These have different magnitudes andneed to be scaled to the same magnitude for the purpose of comparison. Hence a simple Pro-cess Performance Index can be formulated, which includes all the three criteria as well astheir scaling factors:

[1]

The scaling factor for the time to fill is the time taken for the filling if the entire parthad one value of permeability and the resin was injected along the width. Thus the flow isone-dimensional along the length and the time to fill can be calculated by an analyticalexpression:

Jtfi lltfi l l ∗-----------

Achannels

Achannels∗

--------------------------l channels

lchannels∗

------------------------+ +=

Optimization of Channel Design 211

[2]

The characteristic length is the perimeter of the part and the characteristic area of thechannels is taken as the product of the highest dimension of the preform and the averagethickness of the preform.

In practice, some of the manufacturing criteria merit more importance than the others.The time to fill a given part may be an overriding concern while the cost of tooling andweight are unimportant. For applications where the dead weight of resin is important, suchas aerospace applications, the second criteria should be given more weight. The toolingcosts may be significant for manufacture of a complex part which requires a complicatedchannel configuration with a number of resin inlets. This is reflected in the following Pro-cess Performance Index containing weights for the different criteria.

[3]

APPLICATION

The part evaluated here is a two-dimensionalderivation from a section of an all-compositebus that is under development by NorthropGrumman. It has the dimensions as shown (Fig-ure 2) and has thick sections made of a differentmaterial which corresponds to the wheel wellson the actual part. Thus we have a large complexpart with different permeability properties in dif-ferent sections. The part is symmetrical aboutthe centerline thus enabling the detailed analysisof the flow of resin in one half of the part. Forthis part, the characteristic time to fill is 6 hours,the characteristic length is 14.6 m and the char-

acteristic area is 0.16 m2.Six possible channel configurations have been evaluated here using the Process Perfor-

mance Index defined in the previous sections (Figure 3). The permeability of the material inthe thick section was simulated by considering it as 25 plies of random-mat material com-pressed down to a thickness of 7.6 cm. The permeability of the material in the rest of thepart was determined experimentally. It is a stitched fabric having two-dimensional anisot-ropy with the ratio Kxx/Kyy=0.5. The effective permeability of the channels was determined

tfi l l ∗φµl

2

2k∆P--------------=

J λ1

tf il ltf i ll ∗----------- λ2

Achannels

Achannels∗

-------------------------- λ3

lchannels

lchannels∗

------------------------+ +=

Figure 2. Section of Northrop Grumman ATTB.

Whe

el B

ase

(thi

ck s

ectio

n)

1.22m 0.17m

6.1m


using the relation for flow in a rectangular channel. The values of the permeabilities, vol-ume fractions and thickness of each section are given in Table 1.

The six channel configurations were meshed using PATRAN and the simulations per-formed using LIMS. The filling of the part with one of the channel configurations are given

Table 1. Properties of the preform material

Material Permeability K 11, m2 Permeability K22, m

2 Thickness, cm

Volume fraction

Woven material (1) 10-9 2x10-9 2.54 0.5

Random material (2) 10-10 10-10 7.62 0.3

Table 2. Times to fill, length of channel and area occupied for different chan-nel configurations

Configuration number

Fill time, h t/t*Length of

channels, ml/l*

Area of channels, m2 A/A*

1 5.00 0.84 6.1 0.42 0.16 1.0

2 3.93 0.66 12.2 0.83 0.16 1.0

3 0.41 0.07 6.1 0.42 0.16 1.0

4 5.41 0.91 9.1 0.63 0.12 0.75

5 5.78 0.968 9.62 0.66 0.16 1.0

Figure 3. Channel configurations for optimization study.

Optimization of Channel Design 213

as a TECPLOT contour plot (Figure 4). The filling times were deter-mined and the value of the Process Performance Index evaluated withdifferent sets of parameters. The times to fill are given in Table 2. Thetime to fill, the area occupied on the part face by the channels and thelength of the channels are scaled down using the scaling factorsdescribed earlier. The PPI is calculated with the following set ofweights:All equally important: Length unimportant: Areas unimportant:

The values of the PPIs for the 3 different sets of parameters aregiven in Table 3.


From the results, it can be seen that the channel configuration number 3 gives the best valuefor the performance index in all three cases. This is due to the higher number of injectionlocations and the location of the channels such that the resin has a smaller distance to travelthan in the other cases. Configuration 2 has a lower time to fill than configuration 1, but hasa higher value of PPI.

Thus if the dead weight of the resin and the tooling costs are taken into account, thenthe second configuration is probably more inefficient. If an penalty is assigned for thehigher number of injection ports, then the configuration 3 would have a comparable PPI.Channel configurations 4 and 5 have comparable or higher values of the PPI than the otherconfigurations. Thus the branching configurations employed don’t seem to have any effect

Table 3. Process performance index for 3 different sets of parameters

Configuration number

1 2.25 1.88 1.36

2 2.16 1.74 1.59

3 1.49 1.11 0.59

4 2.28 1.72 1.61

5 2.63 2.03 1.73

λ1 λ2 λ3 1= = =

λ1 λ3 1 λ2, 0.1= = =

λ1 λ2 1 λ3, 0.1= = =

λ1 λ2 λ3 1= = = λ1 λ3 1 λ2, 0.1= = = λ1 λ2 1 λ3, 0.1= = =

Figure 4. Channel con-figuration 3: TECPLOT result of flow simula-tion using LIMS (time to fill: 0.41 hrs).


on the efficiency of the manufacturing process. This is due to the length that the resin has totraverse to reach all parts of the mold with one injection port only and hence the mold fillingtimes are higher.

CONCLUSIONS

A Process Performance Index to measure the efficiency of the VARTM manufacturing pro-cess with channels was postulated. Different configurations of channels were evaluated forthe filling a large part with complex geometry and varying material properties using the PPI.The PPI was found to provide a good measure of the cost-effectiveness of manufacturingand enables a quantitative evaluation of the variables vital for efficient manufacturing.

ACKNOWLEDGMENTS

This work was prepared through the participation in the Composite Materials Research Col-laborative Program sponsored by the Army Research Laboratory under Cooperative Agree-ment DAAL01-96-2-0048. The authors would like to acknowledge the help of Mr. Rod Donof the Army Research Laboratory and Mr. Simon Bickerton of Mechanical Engineering,University of Delaware with this work.

REFERENCES1 Gallez, X.E. and S.G. Advani, “Numerical Simulations for Impregnation of Fiber Preforms in Composites

Manufacturing, Fourth Intntl. Conference on Flow Processes in Composite Materials, University of Wales, 19962 Seemann II, “ Plastic Transfer Molding Techniques for the Production of Fiber Reinforced Plastic Structures”, United States Patent 4,902,215, Feb 20, 1990.3 Pike, T., McArthur, M. and Schade, D. , “Vacuum Assisted Resin Transfer Molding of a Layered Structural Laminate for Application on Ground Combat Vehicles”, Proc. of 2th Intntl. SAMPE Tech. Conference, 19964 Gillio, E.F. et. al., “ Manufacturing of Composites with the Con-Injection Process”, 38th Structures, Structural Dynamics and Materials Conference, AIAA, 19975 M. V. Bruschke and S. G. Advani, "A Numerical Approach to Model Non-isothermal, Viscous flow with Free Surfaces through Fibrous Media, International Journal of Numerical Methods in Fluids, 19, pp.575-603 (1994).

Injection Compression Molding.A Low Pressure Process for Manufacturing

Textile- Covered Moldin gs

Carsten Brockmann, Walter MichaeliInstitut für Kunststoffverarbeitung, Pontstraße 49, Aachen D-52062, Germany

INTRODUCTION

Especially in the automotive, but also in the furniture industry, there is a growing demandfor decorated mouldings.1,2 Decorated mouldings can be produced on a standard injectionmoulding machine, using conventional injection moulding (IM) or injection compressionmoulding (ICM). Both processes have some advantages compared to the decoration ofmouldings by glueing, because both are one step processes with a good reproducibility andthe avoidance of adhesives.

When producing decorated mouldings directly in an injection mould, the decorationmaterial is stressed by high temperatures and cavity pressures. This very often leads tostrong decoration material damage.1,3-5 One major type of damage is the collapsing of thefoam layer of the decoration materials, which are typically used in the car interior. Thisfoam layer shall provide a "soft touch"-effect to give the car interior a comfortable appear-ance. If the foam layer collapses this effect is lost.

ICM is a special process of IM, which is able to reduce the cavity pressure.3,6,7 Thisprocess starts with the closing of the mould until the compression gap is reached. At thispoint of time the mould is already sealed, because it is equipped with shear edges. In thenext step the melt volume, which is needed to fill the moulding, is injected into the cavityand the shut-off nozzle is closed afterwards. In the last step the mould is closed completelyby the compression movement. Due to this movement the melt is spread throughout the cav-ity until it is filled.

Finally the melt is compressed, because the injected melt volume is usually higher thanthe cavity volume. This is necessary to compensate the volume shrinkage. In conventionalIM this is done by the packing phase, which is only exceptionally used in ICM.

The major goal of the experiments introduced in this paper is to compare conventionalIM and ICM in regard of cavity pressure reduction and the quality of decorated mouldings.


Futhermore the major process parameters of the injection compression moulding process,which influence the quality of the decorated mouldings, shall be found out and optimized.

EXPERIMENTAL SETUP

The experiments are carried out on a 150 t Mannesmann Demag Ergotech IM machine,which is equipped with an ICM control. A special test mould (Figure 1), equipped withshear edges and facilities to preform and fix the decoration material is used. Four pressuretransducers are mounted into the mould for cavity pressure measurements. These measure-ments are stored and visualized with a PC based data acquisition system.

All experiments are run with the polypropylene Vestolen P2000, which is an easilyflowing PP-type with a MVI of 55 (Tm = 230°C, load 2,16 kg). Four different textile decora-tion materials are used, which are provided by Viktor Achter GmbH & Co KG, Viersen. Allof them consist of three layers including a PU foam layer.


The process conditions of ICM and conventional IM differ strongly in some points.Whereas the injection speed is the most important influence on the cavity pressure in IM, inICM the clamping force and the injected melt volume have strong effects. This is due to thefinal compression phase. The higher the clamping force, the higher is the cavity pressure, ascan be seen in Figure 2. The first experiments were run without decoration material. To beable to compare both processes, parts with the same weights have been produced.8

The influence of the melt volume and clamping force becomes clear when the processis explained in a p,v,T-diagram, as shown in Figure 3.8-10 From 1 to 2 the isothermal injec-

Figure 1. Molding and mold concept.

Injection Compression Molding 217

tion phase can be found, which is connected with the first pressure increase. Due to thecompression gap the cavity is not filled completely. Therefore the pressure drops in thedelay time between injection and compression phase (point 3). With the beginning of thecompression phase the melt starts to flow again, which leads to another pressure increaseuntil point 4 is reached. Here the cavity is filled completely and the melt is now compressedby the clamping force until point 5 is reached. Finally the pressure decreases again (5 to 6)because of the cooling of the melt and the coupled volume shrinkage. Point 7 represents theopening of the mould. If an IM machine with optional pressure-controlled clamp force isused, the course in the p,v,T-diagram is drawn from 4' to 5'. In this case the pressure staysconstant once a given clamp force is reached.

For in-mould surface-decoration the volume shrinkage is usually not as important as itis for conventional IM. The reason for this is the decoration material, which covers opticalproblems such as sinkmarks. Therefore only small or no packing pressures are applied in IMprocess. In analogy to this it is possible to reduce the cavity pressure for ICM by a smallerinjected melt volume. The course in the p,v,T-diagram would then look like as drawn by the

Figure 2. Influencing parameters for IM and ICM. Figure 3. ICM in a p,v,T diagram.

pres

sure

(ba

r)pr

essu

re (

bar)

spez

ific

volu

me

IM:injection speed

__ 60 mm/s

. . . 20 mm/s

____ 100mm/s

ICM:clamp forcechosen (measured)

____ 90kN (246kN)__ 51kN (175kN). . . 25kN (120kN)

time (s)

time (s)temperature


dashed line from 5* to 6* . Figure 4 shows a comparison of the pressure courses, in the casethat decoration material is used.

Another process parameter for the ICM process, which has strong influence on the cav-ity pressure is the compression speed. With higher speed the pressure to fill the cavityincreases as well as the maximum pressure, caused by the melt compression (Figure 5). Thefirst effect is due to lower shear rates and the second due to a longer cooling time and whichleads to a reduced melt volume, when the compression phase starts.

The height of the compression gap influences the pressure course only during the injec-tion phase. The differences are, compared to the pressure maximum, relative small.

In the first experiments the IM parameter injection speed was the same for both pro-cesses, conventional IM and ICM. With this setup it was easier to isolate the influences ofIMC process, because the injection moulding parameters stay constant. But this experimen-tal setup did not take the stress time into account, which was found to be very important.The stress time is defined as the time, when the decoration material is loaded with pressureand high temperatures.

Figure 4. Comparison of pressure courses for IM and ICM with decoration material.

Figure 5. Influence of compression speed on the cavity pressure.

time (s)

time (s)time (s)

time (s)

mou

ld p

ositi

on (

mm

)m

ould

pos

ition

(m

m)

pres

sure

(ba

r)pr

essu

re (

bar)

cavity pressure

cavity pressure

cavi

ty p

ress

ure

(bar

)ca

vity

pre

ssur

e (b

ar)

injection compressionmoulding

injection mouldingmould position

mould position

compression speed:. . . 5% (0.9 mm/s)__ 15% (2.7 mm/s)____ 25% (4.5 mm/s)

compression speed:. . . 5% (0.9 mm/s)__ 15% (2.7 mm/s)____ 25% (4.5 mm/s)


Figure 6 shows the residual foam thicknessdisplayed over the flowpath. The residual foamthickness is measured one week after moulding.This Figure shows important information for theproducer. He is interested in the absolute thick-ness of the foam layer but also in the distributionover the part. Strong differences in thicknesslead to visual marks on the parts surface. In Fig-ure 6 results of experiments with constant injec-tion speed are shown. This means, that all ICMexperiments have longer stress times than theIM experiments, because after the injectionphase, which takes the same time for both pro-cesses, there is the compression phase. This

results in residual foam thicknesses for ICM, which are as small as for IM or even smaller,although the cavity pressures are lower a can be seen in Figure 4.

In Figure 7 results of later experiments are shown.11 Here the ICM experiments havethe same stress times as the ones done by IM. For these experiments at first a reference partis produced by IM. The injection time for this part is then used as the reference processingtime for all following ICM experiments. This means, that the injection time plus the com-pression time can only take as long as this reference time. Therefore higher injection speedsare used for the ICM experiments and the compression phase takes place in the remainingtime. The results of ICM are in this case much more uniform. In case of a compression gapof 3 mm the residual foam thickness is between 60 and 80% (compression gap = 3 mm) ofthe original thickness of the decoration material. For decreasing compression gaps the partsquality is getting worse. There are bigger differences in the thickness distribution and the

Figure 6. Comparison of residual foam thickness for IM and ICM (different stress times).

Figure 7. Comparison of residual foam thickness for IM and ICM (equal stress times).

Figure 8. Residual foam thickness for simultaneous com-pression.

resi

dual

foam

thic

knes

s (m

m)

resi

dual

foam

thic

knes

s (%

)

ScrewPosition

40 mm

45 mm

50 mm

flow path length (mm)


0 20 40 60 80 100 120

0 10 20 30 40 50 60 70 80 90 100 110

Compression Gap = 2 mm


resi

dual

foam

thic

knes

s (%

)

CompressionGap

0 mm2 mm3 mm

0 20 40 60 80 100 120

3.5

3.0

2.5

2.0

1.5

1.0

0.5

100

100

90

90

80

80

70

70

60

60

50

50

40

40

30

30

20

20

10

10

0

0

IM textile 1IM textile 2ICM textile 1ICM textile 2


minimum thickness is decreasing. The curve with 0 mm compression gap represents con-ventional IM and has strong differences over the flow path.

Further improvement of the parts quality can be reached by using the option "simulta-neous compression". In this case the compression phase starts already during the injectionphase. The starting point of compression can be defined by the screw position. The stresstime can further be decreased with this option, which leads to a more uniform and higherresidual foam thickness. In addition the melt flow becomes more uniform without the typi-cal stopping between the injection and compression phase. Figure 8 shows the influence of astarting point variation on the residual foam thickness. The compression gap (2 mm) and theinjection speed (150 mm/s) stay constant. For all different starting positions the foam thick-ness is more uniform and the absolute values are higher than in the experiment shown inFigure 7.

CONCLUSIONS

In-mould surface-decoration is an economic way of producing textile decorated mouldings.But still there are quality problems, like decoration material damage, especially if the size ofthe moulding exceeds certain bounds. Responsible for this damage are high cavity pressureand the melt temperature.

Especially the cavity pressure can be reduced by the use of ICM. Furthermore theexperiments show, that the stress time is an important parameter for final parts quality. WithICM it is, compared to conventional IM, possible to reduce cavity pressures and at the sametime the stress time. Further reduction of the stress time is possible by using simultaneouscompression. With this option a further improvement of the quality of the decorated mould-ing is possible.

An optimization of the ICM process is possible especially by changing the parameterscompression gap and compression speed.

ACKNOWLEDGEMENTS

We appreciate the help of Mannesmann Demag Kunststofftechnik, Viktor Achter GmbH &Co KG and Vestolen GmbH. The investigations set out in this report received financial sup-port from the Ministry of Economics (BMWi) and from the AiF e.V., to whom we extendour thanks.

REFERENCES1 Annen, D., Analysis of the part developement process for the production of decorated mouldings by in-mould

surface-decoration, unpublished diploma thesis at the IKV, Aachen, 1993, supervisor: S. Galuschka. 2 Galuschka, S., In-mould surface-decoration - manufacturing of textile-covered injection mouldings, dissertation at the RWTH, Aachen, 1994.


3 Michaeli, W., Galuschka, S., In-mould surface-decoration: Analysis of the boundary conditions. Teil 1: Plastverarbeiter 44 (1993) 3, S. 102-106, Teil 2: Plastverarbeiter 44 (1993) 4, S. 62-68. 4 Münker, M., Developement and set-up of two tests for the characterisation of decoration materials, internal report at the IKV, Aachen, 1996 supervisor: C. Brockmann, A. Oelgarth 5 Steinbichler, G., Gießauf, J., In-mould surface-decoration with textiles and films, Kunststoffberater 41 (1996) 3,

p. 16-21. 6 Jaeger, A., e.a., Machine technique and processing for in-mould surface-decoration, Kunststoffe 81 (1991) 10,

p. 869-874. 7 Hansen, J., Analysis of injection compression moulding in respect to the opportunity to decrease pressures during mould filling, unpublished thesis at the IKV, Aachen, 1994, supervisor: S. Galuschka. 8 Wodke, T., Optimizatioin and comparison of the injection moulding and the injection compresion moulding process in respect to the part in in-mould surface-decoration, unpublished diploma thesis at the IKV, Aachen, 1996,

supervisor: C. Brockmann. 9 Yang, S.Y., Lien, L. Experimental Study on the Injection Compression Molding of Parts with Precision Contours,

International Polymer Processing XI (1996) 2, p. 188-190. 10 Knappe, W., Lampl, A., About the cycle course in injection compression moulding of thermoplasts, Kunststoffe 74 (1984) 2, S. 79-83. 11 Kuckertz, M., Comparison of injection moulding and injection compression moulding in respect to the suitability for

in-mould surface-decoration, unpublished thesis at the IKV, Aachen, 1996, supervisor: C. Brockmann.

Kurz-Hastings Inmold Decoration

Roy BombergerKurz-Hastings, Inc.

The Kurz World Group is the "State of the Art" in Decoration Technology. Technical andCost improvements as well as automation are the factors most important to any moderninjection molding plant. Kurz-Hastings IMD process makes is possible to produce finisheddecorated parts in one operation. This process has tremendous advantages to any other dec-orating methods for suitable applications.

In many cases the appearance of the piece part has to be improved. This can be done bydecorating their surface and thus achieving a simulated appearance of woodgrains, marbles,metallizations or various combinations. Some products may require special artwork or spe-cial coating on the top surface.

There are many ways to rationalize the additional operations to decorate parts such aspick and place from a molding machine to hot stamp machine or heat transfer machine. Thisrequires secondary operations and additional space.

The idea of decorating inmold is not new in the USA. Parts have been produced for twodecades by placing pre-cut foil inserts in the mold. This method was done with random orcontinuous patterns. It has its limitations due to position accuracy. The IMD combines themold and hot stamping technology into a single process developed in foil manufacturingand design of the parts.

The IMD process offers a number of important advantages when compared with thetraditional method such as printing, painting, metallizing (chrome plating) and conventionalhot stamping.1. The process is environment friendly.2. The decorated area forms an integral bond to the substrate offering better surface proper-ties.3. The IMD process will not add any additional time to the molding cycle. Because mostpeople use robotics to remove parts, the film will be advanced at the same time as the partsare removed.4. Multi-color decoration is achievable in a single cycle.5. This easy change of foil during molding permits trouble free production of different sur-face decorations. Also, it permits you to change foil to give a different look to the productyou are running.


THE IMD PROCESS

The IMD process uses special modified hot stamp foils. These foils are guided through themold in its open position. During injection, the pressure and heat of the molten materialtransfers the printed layer from the polyester carrier and bonds it to the plastic substrate. Afoil feeding and positioning device is used along with special modified foil with registrationmarks for positioning of film. The foil feed device is attached to the moveable platen of theinjection molding machine. The molding press must have two stage injection. Auxiliaryequipment such as mold temperature controller and hopper dryers, ensure optimal moldingconditions.

THE SEQUENCE OF THE IMD PROCESS

The roll of IMD foil is held by the upper part of the device. The foil is fed from top to bot-tom of the mold by this device and used film is rewound by the bottom roller. The printedlayers of this film are transferred by the hot plastics being injected against it. No die cuttingor trimming of the foil is required in this process. As the press opens, a robot removes thefinished part from the mold and the foil advances to the next image.

THE FOIL FEED DEVICE

The device and its control cabinet are supplied in a unit. The mounted unit will be affixed tothe molding machine by a base plate (20 mm thick). The Kurz IMD device can be used withany injection molding machine of modern design.

Three types of IMD devices can be used depending on the positioning tolerances andfeeding requirements.

IMD-JB TYPE

Due to a very precise feeding system this offers positioning tolerances of mm in eachof two X/Y directions. The required positioning registration mark printed on the foil is dif-ferent from the MA device (see below). This IB type provides the best possible positioningaccuracy of the decoration.

IMD-MA TYPE

This device is also equipped with sensors for separate positioning in the X/Y direction. Themaintained positioning accuracy of this device is plus minus 0.1 mm.

Both devices IB & MA feed the foil with an adjustable speed until the sensor govern-ing the vertical positioning reads the start of the printed positioning mark. The final posi-

0.05±

Kurz-Hastings Inmold Decoration 225

tioning, with a slower speed, follows until the end of the mark is read. The machine closesafter the foil is in position (controlled by the device) and the mold cycle starts.

Each of the devices offer three control possibilities:a. Controlled by infra-red sensor. This must be used when individual images are processed.b. Control by timing. In this case, the foil-feed amount is a result of the combination ofspeed and time entered into the control cabinet of the device.c. Manual control. Used basically for setting up.

IMD-T

The unit is a time device only which can be used for continuous pattern. NOTE: This unitcan not be converted to a sensor unit.

All of the IMD devices are available in three sizes corresponding with the width of foil:250 mm, 350 mm and 400 mm. Devices tailored for special applications are available onrequest.

IMD FOIL

The IMD foils are specially modified hot stamping foils using a polyester carrier film 23 to75 micron thick. The choice of the suitable carrier film thickness depends on the decorationdepth and geometric shape, the size of the part, the position and the types of gating and typeof plastic material. We have developed several special foils for the process. They are:a. Metallized high gloss foil (Gold, Silver and Metallic colors)b. Single color (Unicolor) foilc. Woodgrain, Marble, Granite of various designsd. Brushed Silvers, Golds & Mattee. Multicolor pictures - 7 to 9 colors printed in registerf. Continuous patternsg. Selective or partial metallization

The modern printing machines at Kurz permit multiple tones and shades and partialmetallization, also full metallization. We can also have in the image matte and gloss surfaceeffects.

The adhesion layer of the IMD foil is specifically formulated for use with differentmolding material thus ensuring good bond between the plastic and the decoration layer offoil in every case. Foil suitable for the following plastics are available:

1. PS 7. PC/ABS2. SAN 8. PBTP3. ABS 9. PPE


4. PP 10. ABSAN5. PPO 11. PMMA (acrylic)6. PA 6 PA6.6 12. PC * (under consideration of special parameters)

TOOLING FOR INMOLD

When designing new tools for IMD process, the following points should be observed:1. The foil must have an unobstructed passage between the guide pillars and pass close tothe parting surface of the mold.2. The gating must ensure that the foil presses against the cavity surface during injection.This is usually done with edge or submarine (tunnel) gates. Also the hot tip runner systemworks well.3. Regular types as well as 3 plate molds are in use for IMD. Slides and cams, etc. are possi-ble provided they are designed in a way preventing the damage or pinching of the foil dur-ing their operation.4. Since the inmold feeding device is attached to the moveable platen of the moldingmachine, the decorated surface of the part must be formed by a cavity or cavities located inthe moving half of the mold. This construction requires the ejectors to operate from thefixed half (stationary side) of mold. Mechanical or hydraulic activation of ejector system isused.5. All cavities in multiple cavity mold must be covered by the width of foil. Also foil can berun partially if located in front of the injection point.6. The part can be filled by one or more gates. Particular care must be taken not to createexcessive weld lines where two or more streams of plastic meet. The foil will not cover nordisguise such weld lines and appearance faults will result.7. Adequate venting (0.0007 to 0.004 depth) must be provided to permit the air trappedbetween the carrier film and the surface of the cavity to escape. The position and width ofthese vents depends on the gate and gating position and also on the shape of the moldedpart. Moldmakers should realize that they do not need as much cavity pressure as normalbecause of the foil in the mold.8. All cores forming openings in the decorated surface must be located in the fixed andmoveable half of the foil (core split).9. Cavities can use textured finish or high polish surface depending on the required appear-ance.10. The decoration of 3 dimension surfaces must not result in excessive elongation of theIMD foil. Local overstretching of the foil can be prevented by providing for sufficientlylarge radius, chamfer and soft transition when designing the surface to be decorated. Theparting line of the decorated part must also be taken into account.

Kurz-Hastings Inmold Decoration 227

11. Temperature controllers should be available to maintain mold temperature. This willrange from 85 degrees to 180°F Sometimes a separate controller may be needed on bothhalves of the mold.12. Both photo cell sensors can be accommodated within the mold frame in suitable cut outsin the mold frame or located outside the mold.

POINTS TO CONSIDER WHEN DESIGNING IMD PARTS & MOLDS

1. The Depth of Decoration: The corner radii of square or rectangular parts should at leastmatch the decoration depth (if decoration depth is 2 mm, the minimum radius should also be2 mm). A minimum 10 degrees draft in the molding direction will help separate a carrierfilm at mold opening. All inmold radii should be polished. This allows the foil to movewhile the cavity fills.The decoration depth of minimum 0.020 should be maintained even in cases of totally flatdecorated surfaces. This tensioning of the foil over the edge of the part puts tension on thecarrier and this prevents the foil from wrinkling.The gating should be located as centrally as possible for parts of irregular shape. Roundparts are basically gated in the center to help in maintaining the diameter tolerance. In casesrequiring multiple gating, the gates should be located as close to each other as practical tominimize the effect of flow and weld lines in the part.2. Parts with textured surfaces can be achieved by using foil with texture and/or texturingthe mold surface. Textured IMD foil can be matte, brushed or have area of mixed appear-ance (this will require a separate operation during manufacturing) such as high gloss andmatte together. The depth of pattern in the mold surface must not exceed 0.006 with edgesand corners having radii of at least 0.030 inch.3. Plastic material that will be suitable for IMD. Resin with a high flow and easy melt arepreferred. (It is best to fill the mold relatively fast. This helps maintain a uniform tempera-ture on the foil surface allowing the adhesive to bond to the plastic part).4. Clamping pressure of molding machine should be as low as possible. This permits a littlemovement of the foil in case of decorating a large area.5. Elongation of foil: Test results have shown that the stretchability of nonmetallized foilsmust not exceed 20%. This has been achieved under optimal molding conditions. However,each part design should be evaluated separately taking into account its material, shape, sizeand the specific decoration being applied.6. Gates should be as large as possible to facilitate even and fast filling of the cavities.7. Wall thickness of the parts should be at least 0.040. Thinner wall section may not insureproper adherence of the foil to the plastic.


8. Positioning of the part (parts) in mold; long or narrow parts should be positioned horizon-tally in the mold. This orientation will assist in shortening mold cycle.

THE ECONOMICS OF THE IMD PROCESS

The IMD process combines several traditional production and decorating steps in one oper-ation. Therefore, when comparing the cost of IMD with the costs of traditionally decoratedparts, examples should be used where multiple operations and multiple colors would other-wise be required. This IMD process is favorable in cost reduction because of much lowerreject rates and the elimination of storage and handling between operation. The estimatedsavings in real terms vary between 10% and 50% depending on the individual part.

The IMD process eliminates environmental problems as no paints or solvents areinvolved so there are no concerns of VOC's.

Chapter 5: Improving Material PropertiesHigh Impact Strength Reinforced Polyester

Engineering Resins for Automotive Applications

Mengshi Lu, Kevin ManningHoechst Research and Technology

Suzanne Nelsen, Steve LeyrerTicona, Summit, NJ

INTRODUCTION

For engineering plastics to be used in large parts applications such as automobiles, the mate-rial must meet the performance requirements such as stiffness and toughness. In addition,the material should have good processability to be injection moldable.

For short-fiber reinforced engineering plastic materials, as a rule of thumb, stiffermaterials tend to be more brittle. A review of mechanical properties of a variety grades ofpolyester engineering resins manufactured by Ticona supports this generalization. Figure 1illustrates the relationship between impact strength and flexural modulus for several polyes-ter materials. Unfilled Vandar thermoplastic alloys such as Vandar 4602 and Vandar 6000are super-tough at room temperature but have low flexural modulus compared to neat PBT.A similar trend exists between break elongation and flex modulus as shown in Figure 2.

Figure 1. The correlation between impact strength and flex modulus for some commercially available polyester resins.

Figure 2. The correlation between break elongation and flex modulus for some commercially available polyester resins.


Vandar 4361, a 30% glass filled and impact modified PBT resin, has a good combination ofstiffness and toughness. One objective of this work was to generate a PET based resin whichhas high impact strength and maximum break elongation while maintains the highest possi-ble flexural modulus.

Semi-crystalline polymers such as polyesters are pseudo-ductile materials, i.e., theyexhibit high crack initiation energy but low crack propagation energy under impact. Whenthere is a sharp notch or crack present, the parts made from this kind of material may fail ina catastrophic manner without absorbing much energy during impact. Toughening semi-crystalline polymers including polyester is a common practice in the plastics industry. Thegeneral approach is to create a second dispersed rubbery phase which helps initiate matrixplastic deformation leading to dissipating a large amount of energy before fracture. To beeffective the rubber particle size has to be within a certain range. The optimal rubber parti-cle size range varies from polymer to polymer. In general, for aromatic polyesters like PETand PBT or polyamides like nylon 6 and nylon 66, the optimal rubber particle size range isbetween 0.1 to 1 micron.1-3 To achieve such fine rubber dispersion, some kind of reactivecompatibilization is usually needed since most rubbers are neither miscible nor compatiblewith polyesters or polyamides.4,5 Selection or design of the proper impact modifier andproper compatibilizer is the key to developing a material which has the desired properties.

In this report, we discuss the development of Impet Hi 430, a toughened, reinforcedPET based engineering resin. This material has both high toughness and good processabilityto allow molding of large parts. The effect of rubber level and glass level on the mechanicalproperties will be reviewed. Structure/property relationship will.also be discussed.

EXPERIMENTAL

The polyethylene terephthalate used, Impet® 100, has an IV of 0.68 and melt viscosity of

1560 poise at 280oC and 1000 1/sec. A 14 m fiberglass is used as reinforcement. Threeelastomers were evaluated as impact modifiers. The first two, designated as A and B, havefunctional groups that can react with the PET carboxyl end groups. The third elastomer, C,does not contain any reactive groups.

All the materials were compounded using either a 30 mm or a 40 mm WernerPfleiderer ZSK twin screw co-rotating extruder. All components except glass were pre-blended using a tumbler mixer before compounding. The glass was fed downstream. Theextrudates were cooled in a water trough before being pelletized. Normally, prior to moldingthe pellets were dried at 135oC for 4 hours or at 93oC overnight using a dehumidifyingoven. Various standard injection molding machines like a 30 ton BOY machine were used tomold test bars in accordance with either ASTM or ISO standards. All samples were condi-tioned at 23oC and 50% relative humidity for 48 hours before testing.

µ

High Impact Strength 231

The apparent melt viscosity and melt stability were measured by Kayeness Rheometerat 280oC and 1000 1/sec. All samples were dried in a vacuum oven at 150oC for one and ahalf hours prior to testing. The rubber morphology of selected compounds were examinedon fracture surface using a Jeol JSM-T200 Scanning Electron Microscopy. To enhancephase contrast the rubber phase was selectively etched by soaking the samples in xylene forhalf a hour.


One existing material, Impet 320, was first reviewed. The impact strength and modulus ofthis resin are shown in Figure 1. The impact strength was deemed inadequate for structuralapplications where greater toughness is desired. This material contains 15% glass and lowlevel nonfunctionalized impact modifier.

It seemed necessary to use a higher level of functionalized impact modifier to achievebetter toughness. Elastomers A and B both contain reactive groups which can chemicallyinteract with PET. This interaction may help generate proper rubber morphology necessaryfor toughening. For comparison, the nonfunctionalized elastomer C was also used. All themodifications were based on 15% glass reinforced formulation. Figure 3 illustrates theresponse of melt viscosity of the resins as a function of the level of these three elastomers.The rheological information gives a good indication of the interaction between the elas-tomers and PET matrix. Elastomers A and B both seem to have strong interaction with PETwhile elastomer C seems to have little interaction with PET. Figure 4 shows the notchedIzod impact strength of these resins versus the elastomer level for each elastomer. For elas-tomers which have a strong interaction with PET, i.e., A and B, increasing the loading level

Figure 3. The melt viscosity of 15% glass reinforced res-ins as a function of rubber content of elastomers A, B and C.

Figure 4. The notched Izod impact strength of 15% glass reinforced resins as a function of rubber content of elas-tomers A, B and C.

% Elastomer % Elastomer

8000

7000

6000

5000

4000

3000

2000

1000

0 000

100

200

300

400

15% glass filled15% glass filled

Elastomer A

Elastomer AElastomer B Elastomer B

Elastomer CElastomer C

280 C

5 510 10 1515 20 2025 25


leads to higher impact strength. On the otherhand, increasing the loading level of elastomer Cfrom 5 to 20%, seems to have no effect onimpact strength. Elastomer A imparts the besttoughening improvement.

The morphology of the resins that contains20% of elastomers A, B, and C are shown in Fig-ures 5a to 5c. When elastomers A and B wereused, the average rubber particle size was below1 m. While particles of elastomer C are irregu-lar in shape with an average particle size wellabove 1 m. The difference in morphology mayexplain why elastomers A and B are betterimpact modifiers than elastomer C. Using elas-tomer A renders a better impact strength, it also

increases the melt viscosity of the resin significantly. A resin with good impact strength anda relatively lower melt viscosity is more desirable from a processing point of view.

The chemical structure of elastomer A has a similar backbone to that of elastomer C. Itis suspected that the two elastomers are at least compatible (if not miscible) due to this sim-ilarity in structure. One idea was to explore the possibility of using a combination of elas-tomers A and C as impact modifiers. The hypothesis was that elastomer A could serve asboth an impact modifier and as a compatibilizer between the PET matrix and elastomer C.Diluting elastomer A with C could lead to lower melt viscosity. While keeping the total rub-

µ

µ

Figure 5. The SEM graphs of 15% glass filled resin con-taining 20% rubber.

Figure 6. The melt viscosity of 15% glass reinforced res-ins as a function of the concentration of elastomer A in the total rubber content.

10000

8000

6000

4000

2000

00 10 20 30 40 50 60 70 80 90 100

% Elastomer A In Total Rubber

15% glass filled

Melt

Visc

ocity

at 1

000

1/se

c (p

oise

)

280 C


ber content constant at 20%, various combination of elastomers A and C were used asimpact modifiers for the 15% glass reinforced resin. Figure 6 shows the melt viscosity ver-sus the concentration of A in the total rubber. In general, as more elastomer A was used, theresin had higher melt viscosity. Figure 7 shows the tensile strength and the flexural modulusof these resins as a function of the concentration of A. When the rubber is composed of100% of the nonreactive elastomer C, the compound has low tensile strength and low flex-ural modulus. The addition of even small amount of the functional elastomer A leads toimmediate increase in both properties. Further increase in elastomer A concentration, how-ever, imparts no additional improvement. It appears that there is a minimum amount of A(perhaps ~ 20% in the total rubber content) that is needed to provide compatibilization forthe nonfunctionalized elastomer C with the PET matrix. Adding more than this minimumamount of elastomer A provides no further improvement in tensile properties.

The impact properties of these compounds is plotted as a function of elastomer A con-centration in the total rubber content, see Figure 8. Similar to the tensile properties, adding asmall amount of elastomer A to elastomer C leads to a significant increase in both notchedand unnotched Izod impact strength. Adding more than 20% of elastomer A in the total rub-ber content seems to impart little further improvement in impact strength.

There is a clear advantage of using a combination of elastomers A and C, preferably20% of A in the total rubber content. First, it has notched Izod impact strength of 190 J/m,only slightly lower than using 100% of elastomer A.

Second, it has almost exact tensile properties as using 100% of elastomer A. Third, themelt viscosity of this compound is much lower than using 100% of elastomer A. The rubbermorphology of this compound is shown in Figure 9. It is interesting to point out that the

Figure 7. The tensile strength and flex modulus of 15% glass reinforced resins as a function of the concentration of elastomer A in the total rubber content.

Figure 8. The notched and unnotched Izod impact strength of 15% glass reinforced resins as a function of the concentration of elastomer A in the total rubber con-tent.


average rubber particle size is less than 1 m.The particles appears to be uniform in size, con-firming the hypothesis that elastomers A and Care compatible. One would expect a bi-modalparticle size distribution if they were not com-patible.

This resin is reinforced and yet it exhibits ayield point even at sub-ambient temperatures asshown in Figure 10. The existence of a yieldpoint indicates ductile fracture behavior which isnot common for reinforced resins.

Figure 11 shows the effect of glass contenton the flexural modulus and impact strength of resins which contain a combination of 4/16A/C elastomers. While the impact strength seem to stay more or less the same, the flexuralmodulus increases almost linearly as the glass level is increased.

CONCLUSIONS

We have successfully developed a new PET based engineering resin product line, i.e., ImpetHi. The first product, Impet Hi 430, has a good combination of stiffness and toughness. Thisresin also has excellent processibility. This resin was successfully molded into very largeparts. This product can be UV stabilized for exterior applications, and the natural productcan be colored using available concentrates. This material finds use in large molded partsthat experience a large temperature range in use. Other potential applications include recre-ation vehicles, automotive assemblies, furniture and appliances.

µ

Figure 9. The SEM graph of 15% glass filled resin con-taining 4% A and 16% C.

Figure 10. The stress-strain curves of Impet Hi 430 at various temperatures.

Figure 11. The impact strength and flex modulus of 4/16 A/C toughened resins as a function of glass level.


ACKNOWLEDGMENT

The authors would like to thank Mr. A. Towne, Dr. Rong Chen, Physical Testing Lab ofHR&T and Polymer Processing Team of Ticona for technical assistance.

REFERENCES1 Epstein, B. N. U.S. Patent, 4,172,859, 1979. 2 Wu, S. Polymer, 1985, 26, 1855.3 Flexman, E. A., in “Toughened Plastics I: Science and Engineering”, Ed. Riew C. K. and Kinloch, A. J., Advances in Chemistry Series 233, American Chemical Society, Washington, DC 1993.4 Lu, M., Keskkula, H. and Paul, D. R., Polymer, 1993, 34, 1874.5 Lu, M., Keskkula, H. and Paul, D. R., J. Appl. Polym. Sci., 1995, 58, 1175.

Control of Internal Stresses in Injection Molded PartsThrough the Use of Vibrational Molding,

“RHEOMOLDING SM”, Technology

Akihisa Kikuchi, Marc Galop, Harold L. Brown, Alexander BubelTherMold Partners L.P., Stamford, Connecticut, USA

INTRODUCTION

In the injection molding processes, there are many parameters of importance. These vari-ables can be divided into two major types: 1) controls of a molding machine; and 2) poly-mer rheological behavior. Vibrational molding technology was applied to an injectionmolding process in an effort to control rheological behavior during processing and toimprove the properties of materials. Use of the vibration technique results in more orderedstructures from part to part and reduces the effect of certain independent variables duringprocessing.1

A PROBLEM IN INJECTION MOLDING PRACTICE

In molded products, the internal stresses, which are sometimes referred to as residualstresses or molded-in stresses, are the result of the morphologies of a molded part. In simpleterms, a part has high internal stresses if molecules are more oriented; a part has low internalstresses if molecules are more relaxed. The morphologies which determine the properties ofmolded parts are the result of the rheological behavior of polymer during processing. Therheological behavior is controlled by the processing temperature, applied pressure to thepolymer melt, and shear stress resulting from polymer flow. In general, these factors are

controlled poorly during injection molding prac-tice.

During processing the molecules tend to be

sheared, as during injection molding. Figure 1schematically depicts the cavity melt flow look-ing at a part’s thickness during the injectionmolding processes.2 Orientation consists of a

Figure 1. Cavity melt flow looking at a part’s thickness.

Gate

Surface Highly Oriented

more oriented than relaxed, particularly when

Flow Front

Velocity P

rofile

ExtensionallyOriented SkinFrom StretchingCore Orientation

From Bulk Shear

Sub-Surface OrientationFrom High Shear Near Wall


controlled system of stretching plastic molecules to improve their strength, stiffness, opti-cal, electrical, and other properties. For example, molecular orientation results in increasedstiffness, strength, and toughness as well as liquid and gas permeation, crazing, microc-racks, and others in the direction or plane of the orientation.3 However, orientation resultingfrom the melt flow during processing can be deliberate or undesirable depending on thedirection of applied load or stresses when used as a finished product. When the orientationis accidental and unfavorable, which is very common in injection molding practice, it is typ-ically referred to as so-called residual stresses or molded-in stresses. The term residualstresses identifies the system of stresses that are in effect locked into a part, even withoutexternal forces acting on it.

Orientation or residual stresses play an important role in toughness enhancementbecause toughness is primary based on the mechanics of craze formation and shear band(craze and flaws) formation. The shear bands determine the fracture mode and toughness ofa polymer when subjected to impact loads. The amount of energy dissipated depends onwhether the material surrounding the flaws deforms plastically. For toughness enhancementthe residual stresses play an important role in the suppression of craze formation, by avoid-ing the stress state that promotes brittle fracture. For instance, at room temperature an ori-ented PS is a brittle, glassy, amorphous polymer, whereas a uniaxial oriented PS is highlyanisotropic. High tensile strength, elongation, and resistance to environmental stress crazingand cracking are achieved in the direction of orientation. However, an oriented PS is weakerand more susceptible to stress crazing in its transverse direction than is an unoriented PS.Biaxially oriented PS is strong and tough in all directions.4

Many factors influence orientation or residual stresses, such as the design of the part,the design of the mold, and the processing conditions. There are three typical causes ofresidual stresses in the injection molding process: 1) Nonuniform shear action - as can beseen in Figure 1, molecules are highly oriented near the mold walls as compared to the bulkand this shear tendency due to the melt flow behavior results in nonuniform orientationthrough a part thickness and part width with respect to the flow direction; 2) Nonuniformheating and cooling - the introduction of residual stresses can be the result of nonhomoge-neous plastic deformation occurring during thermal and mechanical actions, arising fromchanges in either volume or shape. Thermal treatments like quenching and annealing intro-duce changes in physical and mechanical properties. For example, with sheet plastic thestresses created by quenching are the result of uneven cooling, when the surfaces cool fasterthan the core. This produces nonuniform volume changes and properties throughout thethickness. The compressive stresses on the surfaces of the quenched plastic produce tensilestresses in the core, which maintain the equilibrium of the forces;4 3) Nonuniform partdimension - varying wall thickness from thick and thin sections in a part induces residual

Control of Internal Stresses 239

stresses. This is due to the different rates of shrinkage (causing warpage), and possibly voidformation in the thick portion. Since the parts in a mold solidify from their outer surfacestoward the center, sinks will tend to form on the surface of a thick portion.5

VIBRATIONAL MOLDING CONCEPT

With the problems in mind, vibrational molding technology has been applied to the injectionmolding process. The patents for the specific types of vibrational molding technology areowned by TherMold Partners L.P., Stamford, Connecticut.6,7,8 The concept of vibrationalmolding is based on the oscillation of a molten polymer to control rheological behavior ofpolymers within the mold cavity during processing. Oscillating polymer melt within thecavity controls the effects obtained from shear stresses, such as molecular orientation. Thisconcept was verified by using a compression molding technique.

When the vibrational molding technology is applied to the injection molding, the vibra-tion force can be induced during two stages, the injection stage and holding stage. When thevibration is induced during the injection stage, the melt flow behavior is altered as com-pared to the flow in the injection molding process. When the vibration is induced during thepacking stage, it generates a different type of pressure gradient/distribution as compared tothe gradient/distribution in the injection molding process. This pressure gradient generatedby vibrational molding results in "In-Mold Flow" and controls the internal friction.9 "In-Mold Flow" modifies the flow advancement. There is induced a distinct flow pattern ascompared to the flow pattern resulting from the conventional molding. Also, the amounts ofshear effects on the melt can be increased or decreased because vibration frequencies andamplitudes control the internal friction which results from shear stresses and hydrostaticpressure.8 This vibration allows for the simulation of various cooling rates that can behigher than the values obtainable by conventional conductive cooling. In addition, highpressure processing is achieved by controlling hydrostatic pressure. Therefore, the effects ofhigh cooling rates or the effects of lower temperature processing as well as high pressureprocessing can be obtained by inducing vibrations while maintaining normal molding con-ditions. The cooling rates through the thermokinetic transition temperatures, such as meltingtemperatures, Tm, or glass transition temperatures, Tg, can be high, and these transition tem-peratures can be raised. Additionally, more uniform temperature gradients are obtainablesince the vibration energy is transmitted to the entire part. This internal friction controlsmolecular orientation/relaxation which in turn result in controlling internal stresses as wellas molded part properties.


VIBRATIONAL MOLDINGAPPLICATION METHODS

There are five types of vibrational molding appli-cation methodologies, which are referred to asRheojector (Vibration Generating System - VGS)I, II, II-A, III, IV and IV-A, however the VGS IVand IV-A systems, which are much moreadvanced and more flexible than other vibrationgenerating systems, are not disclosed since theyare pending patent approval.

Figure 2 depicts the schematic of the VGS Isystem with the rotary valve in its initial position.

The VGS I system was installed between the mold die and the barrel, and this system wasused for the proof of concept. This system is very similar to the two stage injection moldingmachine. The injection screw injects the melt into the chamber when the rotary valve is inthe initial position, then the rotary valve rotates 90 degrees and the plunger injects the meltinto the mold cavity. During the injection and/or holding stage, the plunger vibrates tomanipulate the melt. Figure 3 depicts the schematic of the VGS II and II-A. The VGS II sys-tem embeds piston(s) within the mold cavity or the mold, and these pistons manipulate themelt. To install the VGS II system, the mold die needs to be modified. The VGS II-A systemuses the same concept as the VGS II system, however the mold modification is very mini-mal. Instead of placing the pistons within the mold, the VGS II-A system places the pistonsat the parting line of the mold. Figure 4 depicts the schematic of the VGS III system. Theconcept of the VGS III system is very similar to the VGS I system, however the rotary valveused for the VGS I was removed in order to have two stage processing. The sensors, thecontrols of VGS I and VGS II, and the controls of the injection machine were interfaced

Figure 2. Schematics of VGS I. Figure 3. Schematics of VGS II & II-A.

Figure 4. Schematics of VGS III.


through the data acquisition board to a personal computer. The entire process which was thecombination of vibrational molding technique and the injection molding was monitored andcontrolled by the external computer.


The objective of this study was to enhance the part properties by applying the vibrationalmolding technique to the injection molding process.

The materials used for this particular study were Polystyrene (PS), Polycarbonate (PC)and Polyethylene Terephthalate (PET). The VGS IV system was used for this study toimplement the vibrational molding (VMT) technology to injection molding processes. Themolding conditions were the same for both VMT and injection molded specimens. TSCinternal stress measurements were performed for both VMT and reference samples underidentical conditions. At first, the reference samples were produced under the conditionswhich meets the resin manufacturers’ performance specification. The VMT samples wereproduced just by simply applying the vibration while using the processing conditionsobtained during the reference sample production.

The internal stresses were measured byThermally Stimulated Current (TSC) Spectrom-eter provided by TherMold Partners L.P., Ther-mal Analysis Instrumentation Division. Figure 5depicts the picture of TSC experiment station.TSC measures molecular mobility by observingthe displacement current generated by dipolemotion. Typically molecules are oriented in cer-tain directions during processing and trappedwithin a part after solidification. When a part isheated up to a certain temperature (i.e. Tg orTm), molecules relax and the displacement cur-rents measured in a proportion to the amount ofstress originally in the specimen. The results is a

spectrum of current versus temperature. The total amount of internal stresses in a part can bederived from the area under the curve of current-temperature curve obtained in the TSCdata. In layman’s term, a part has low internal stresses if the area under the TSC curve issmall, and a part has high internal stresses if the area under the TSC curve is large.11

Figure 5. Picture of thermally stimulated current (TSC) spectrometer.



The results shown in Figure 6 was obtained fromPS specimens. Figure 6 depicts the comparison

of the stress-strain curve between the vibrational molding (VMT) specimen and the injec-tion molded specimen. As seen in Figure 6, the VMT specimen clearly shows the improve-ment in toughness (area under the stress-strain curve) as well as tensile strength. Forinstance, the elongation at break was boosted up from 6.17 to 8.19% by VMT, and theimprovement was 32.7%. The tensile strength was boosted up from 45.36 MPa (6578 psi) to55.47 MPa (8043 psi), and the improvement was 22.3%.

The results shown in Figure 7, 8 and 9 were obtained from PC specimens. Figure 7depicts the comparison of the stress-strain curve between the vibrational molding (VMT)specimen and the injection molded specimen. Figure 8 shows the TSC’s internal stress mea-surement data for the same specimens. Figure 10 shows the birefringence pattern for injec-tion molded PC lens and VMT PC lens, respectively. As seen in Figure 7, the VMTspecimen clearly shows the improvement in toughness (area under the stress-strain curve) aswell as tensile strength. For instance, the elongation at break was boosted up from 75.19%to 104.00% by VMT, and the improvement was 38.3%. The tensile strength was boosted up

Figure 6. Stress-strain curve for PS. Figure 7. Stress-strain curve for PC.

Figure 8. TSC measurement data for PC injection mold-ing VMT.

Figure 9. Birefringence pattern of PC lens.


from 55.75 MPa (8084 psi) to 69.39 MPa (10061psi), and the improvement was 24.5%. Figure 8clearly shows the difference in the area under thecurrent-temperature curve between the VMT sample and the injection molded sample. Bycomparing the birefringence pattern of injection molded lens and VMT lens in Figure 9, sig-nificant differences in the formation of birefringence are seen. The injection molded lenshas more discontinuity in birefringence, which reveals nonuniform morphology. However,the VMT lens part has more uniformity, which reveals relatively more uniform morphology.

The results shown in Figure 10 and 11 were obtained from PET specimens. Figure 10depicts the comparison of the stress-strain curve between the vibrational molding (VMT)specimen and the injection molded specimen. Figure 11 shows the comparison of the TSC’sinternal stress measurement data for the VMT (4 curves as indicated as Vibrational Mold-ing) and injection molded specimens (4 curves as indicated as Regular Molding). As seen inFigure 10, the VMT specimen shows slight improvement in toughness (area under thestress-strain curve). The elongation at break was improved from 216.3 to 226.9% by VMT,and the improvement was 4.9%. However, it must be noted that the primary objective ofproducing the PET samples was to reduce the processing (melt) temperature while main-taining the part clarity. The test results shown in Figure 11 clearly shows the difference inthe amount of internal stresses (area under the current-temperature curve) between the VMTsample and the injection molded sample. The VMT curves (indicated as Vibrational Mold-ing) were more like flat lines as compared with the injection molding (Regular Molding)curves. The reduction of the internal stresses due to VMT was as much as 80%. In addition,the VMT curves show more repeatability than the injection molding. This results indicatesthat VMT had less deviation in the amount of internal stresses in the part than the injectionmolding.

By considering the test data and the earlier theoretical explanation, reducing the inter-nal stresses in a part (or making a more relaxed part) improves the toughness. However,

Figure 10. Stress-strain curve for PET.

Figure 11. TSC measurement data for PET.


according to the earlier explanation the more oriented part, which has more internal stresses,should have higher strength than a more relaxed part. This was not seen in our experimentaldata. This may be because vibrational molding produces parts with more ordered structureswhile maintaining the relaxed structures instead of having the molecules under tension and/or compression within a part.

CONCLUSION

Based on the test results, using vibrational molding can provide the following advantages:1) vibrational molding can reduce the internal stresses without having secondary operationsuch as annealing. 2) vibrational molding can reduce the internal stresses to improve part’stoughness without sacrificing other tensile properties such as tensile strength. 3) vibrationalmolding can produce a part with improved performance without sacrificing a cycle time.

REFERENCE1. Akihisa Kikuchi and Robert F. Callahan, "Quality Improvements Resulting From RHEOMOLDING Technology During Injection Molding Processes", SPE ANTEC’96 proceedings, pp. 764-768.2. Donald V. Rosato et al, 1991, Designing with Plastics and Composites; A Handbook. New York: Van Nostrand

Reinhold, pp. 617.3. Donald V. Rosato et al, 1991, Designing with Plastics and Composites; A Handbook. New York: Van Nostrand

Reinhold, pp. 118.4. Donald V. Rosato et al, 1991, Designing with Plastics and Composites; A Handbook. New York: Van Nostrand

Reinhold, pp. 791-794.5. Donald V. Rosato et al, 1991, Designing with Plastics and Composites; A Handbook. New York: Van Nostrand

Reinhold, pp. 791-794.6. Doug Smock, Nov. 1993, ''Good vibes boost part properties in quantum leap; 'Rheomolding' will greatly expand

application for plastics. Developer is seeking license agreement now.", Plastics World, Vol. 51, pp. 14-15.7. Sara Ferris, Dec. 1993, "New technology vibrates melt, affects plastic microstructure.", Plastics Machinery &

Equipment, Vol. 22, No. 12, pp.37.8. John De Gaspari, Mar. 1994, "Melt Flow Oscillation Improves Part Properties", Plastics Technology, Vol. 40, No. 3, pp.21-23.9. Akihisa Kikuchi and Robert F. Callahan, "Quality Improvements Resulting From RHEOMOLDING Technology During Injection Molding Processes", SPE ANTEC’96 proceedings, pp. 764-768.10. J. P. Ibar, 1981, "Rheomolding: A new Process to mold polymeric materials", Polym. Plast. Technol. Eng.11. Marc Galop, “Characterization of Internal Stresses Using The Thermally Stimulated Current Technique”,

prepared for publication.

Experimental Determination of OptimizedVibration-assisted Injection Molding

Processing Parameters for Atactic Polystyrene

Alan M. Tom, Akihisa Kikuchi, John P. CoulterLehigh University, USA

INTRODUCTION

As manufacturing becomes increasingly competitive, improved processing technology iscontinuously needed. One such new technology, vibration assisted injection molding,appears promising. Several mechanisms through which to implement the technology havebeen studied during the past decade, but the complete understanding and effective deploy-ment of the concept remains to be realized. Vibrational Assisted Injection Molding (VAIM)was initially implemented and introduced experimentally to the injection molding process inthe early 1980’s. To date, the concept has shown promising results in its ability to improveaesthetic, mechanical, thermal, and optical properties of molded parts. Although conven-tional injection molding techniques had made polymer manufacturing popular, associatedfinal products are often not defect free. Typical final product defects include sinks and voidswithin the material, warpage from residual stresses created upon solidification, and weld-lines that affect appearance and structural integrity. One of VAIM’s potentially biggest con-tributions to the manufacturing industry related to injection molded plastic parts will be inits ability to reduce material property variations, and hence enhance quality control throughthe reduction of rejects.

In it’s simplest form, VAIM applies mechanically induced vibrational forces to poly-mer flow melts during the injection molding process. Once applied to the plastic, the vibra-tional forces can be controlled to enhance and change the micro as well as the macrostructures of the polymer material either at specific locations within the part or throughoutthe molded part. Improved properties of plastic molded parts include mechanical properties(tensile and impact strength), thermal properties (glass transition temperature and melt tem-perature), optical properties (transparency and birefringence), and aesthetic properties(weldlines, warpage, and color uniformity). Current modern day systems that have beendeveloped include SCORIM (Shear-Controlled Orientation Injection Molding) by ScortecInc.,1-4 Push-Pull process,3,5-7 Injection Spin Process,8,9 Moving Boundary technique,10,11


and Rheomolding by TherMold Partners.2,12-18 These modern VAIM systems developedwithin the past decade have been experimentally implemented, but a complete understand-ing on a scientific level of both the mechanical and material property improvements haveyet to be realized. For instance, it is known theoretically that process conditions such astemperature, pressure and cooling rate affect the degree of crystallinity, the size, type anddistribution between crystallites and the amorphous phases. These crystal structures in turnaffect the physical and mechanical properties of the final plastic parts such as polymer glasstransition temperature and melt temperature. However, experimental results using VAIMfall short of scientifically explaining what effects specific processing variables such as meltoscillation type, frequency, initiation time, and duration have on material properties, specif-ically glass transition and melt temperatures. Therefore, a controlled scientific experimentwas necessary to evaluate and determine optimal molding conditions and the effects of allprocessing conditions on final product properties.


EXTERNAL HARDWARE

As shown in Figure 1, a Gateway 2000 GP6series personal computer was used as an exter-nal medium to run LabVIEW software. Lab-VIEW software enabled the user to varyvibrational molding parameters such as fre-quency, amplitude, vibration start and stop timeby interacting with a 15 ton BOY injectionmolding machine via a National Instruments A/D converter and DAQ relay board. Two poweramplifier relays were used to ramp up control

output digital signals emitted by the DAQ board in an effort to assure sufficient electricalpower was being generated to operate the hydraulic compression and decompression valvesof the injection molding machine. Mold temperatures were controlled through the use of aReglomat RT20 thermolater that supplied heated water flow in a closed loop network ofhoses to an ASTM standard tensile test mold. Tensile test specimens produced in the shapeof dogbones were in compliance with ASTM D638-91, standard test methods for tensiletesting of plastics.

VAIM SOFTWARE DEVELOPMENT

In an effort to supply oscillatory pressure vibrations to polymer melt, computer softwarewas written with the aid of LabVIEW to physically simulate the manual operation of push-

Figure 1. Experimental hardware and software setup.

Control Box withDAQ and Amplifier

BOY 15 SInjection Molding Machine

Reglomat RT20Thermolater

Gateway 2000 GP6Personal Computer

Vibration-assisted Injection Molding 247

ing and releasing both injection and decompres-sion buttons on the injection molding machine inan alternating fashion. Choosing to control theinjection molding machine in this manner, ver-sus writing software that utilized a square wavefunction generator, allowed for the advantage ofdoing experimental vibrational tests with con-stant stroke amplitude while still being able tovary other machine processing variables such asvibration frequency. The software activation of

screw injection and decompression motions is shown in Figure 2.

CONTROL SPECIMEN TESTING

Polystyrene single gated control test specimens were molded using conventional injectionmolding procedures and were used as a baseline reference for comparative measures againstthose molded with mechanical vibration. In an effort to obtain optimal machine moldingparameters for baseline tensile test specimens, Taguchi’s method was implemented on sixmachine parameters at three levels of operation, thereby producing 729 separate experi-ments. Five test specimens were produced for each control experiment for an overall total of3,645 polystyrene dogbones.

TESTING OF VAIM SPECIMENS

In a similar fashion, vibrational molded samples were produced and experimentally testedfor optimal molding conditions within a particular processing window. Four moldingparameters were chosen to investigate the preliminary optimal VAIM molding conditions.These were Delay Time to Begin Vibration, Oscillation Frequency, Vibration Duration, andDuty Cycle. Each parameter was tested at three levels of operation producing a total of 81experiments. Each experiment produced 5 VAIM samples for an overall total of 405 testspecimens.

The Delay Time to Begin Vibration variable is the amount of time that the machineremained in idle after polymer had been injected into the mold cavity before mechanicalvibrations were activated. This variable was tested at 0, 0.5, and 1.0 seconds. A Delay Timeof 0 seconds represents an immediate activation of the vibration control mechanism follow-ing complete polymer injection.

Previous studies have indicated that large increases in polymer viscosity occur at lowoscillating frequencies. Therefore, it was determined that 1, 2, and 3 Hz frequencies wouldbe tested. The VAIM Vibration Duration parameter was tested at 5, 10, and 15 seconds and

Figure 2. Injection and decompression process.

Compression

Decompression


represents the length of time that vibration oscil-lations were applied to polymer melts.

The Duty Cycle, Tinj/Tdec, represents theratio of machine injection time to decompres-sion time for one complete cycle of vibration.Figures 3(a) and 3(b) depict examples of dutycycles for a 2:1 (i.e. 2 to 1) ratio of injectiontime to decompression time for 1 Hz and 2 Hzfrequencies, respectively. For a 1Hz frequency,

the period of one complete cycle will be 1 second, and therefore during a 2:1 duty cycle, themachine injection and decompression positions will be sustained for 0.66 s and 0.33 srespectively. The fact that the injection time was always longer relative to the decompres-sion time was necessitated by the observation that any duty cycle ratio less than one pro-duced parts that were not completely filled and exhibited sink marks. This was a directresult of less than minimal packing pressure needed to counteract the effects of the observ-able defects. As shown in Figure 3(b), for a 2 Hz frequency, the period of one vibrationcycle was reduced to 0.5 s and the machine injection and decompression positions shouldthen be held for 0.33 s and 0.17 s respectively.

TENSILE TEST SPECIMEN AND INSTRUMENTATION

A single gated ASTM tensile test specimen in the shape of a ‘dogbone’ was used to deter-mine the optimized injection molding parameters for control and VAIM experiments. Testspecimens were produced and tested under the guidelines of ASTM D638-91 on an MTIPhoenix-486 tensile test machine with a load cell capacity of 10,000 lbs and a cross headspeed of 0.1 in/min. Dimensions of each specimen, for the purpose of determining crosssectional area, were measured and recorded using a digital caliper.

PRELIMINARY OPTIMAL MOLDING PARAMETERS

Preliminary results for the optimal molding conditions on VAIM single gated, polystyrene,tensile test specimens proved to be conclusive in setting a general trend that could be usedas guidelines for the further study necessary to completely optimize VAIM molding condi-tions. Utilizing Taguchi’s method, a main effects analysis was done on the ultimate tensilestrength (UTS) results of 405 VAIM specimens and plotted against 4 individual moldingparameters shown in Figure 4(a), (b), (c), and (d). Note that each data point plotted is repre-sentative of an average UTS determined from 135 samples.

With molding machine parameters set at the same settings as the optimized controlsample settings, the results of optimal Delay Time to Begin Vibration is observed to be 0.5

Figure 3. (a) 2:1 Duty cycle for a 1 Hz frequency. (b) 2:1 Duty cycle for a 2 Hz frequency.

Time (sec)Time (sec)

(a) (b)

Inje

ctio

n S

igna

l

Inje

ctio

n S

igna

l

2:1 Duty Cycle for 1Hz Frequency

00 0.66 1.01.0 0.33 0.50 0.88

-1-1

00

+1+1

2:1 Duty Cycle for 2Hz Frequency


s. As shown in Figure 4(a), the longer it took to initiate vibratory motion, the greater the riskfor polymer solidification within the mold cavity which in turn reduced the effects of VAIM.The VAIM Vibrational Frequency parameter in Figure 4(b), showed increasing results withincreasing frequency and therefore a 3 Hz frequency was chosen as the optimal value. Thisis partly due to the effects of shear thinning which reduces polymer viscosity and increasesthe flow characteristics of the polymer material being molded. The optimal molding condi-tion for Vibration Duration was determined to be 5 s, as depicted by the results of Figure4(c). There is a high probability in existence that any machine vibratory motion exertedafter this time period (i.e. 10 s or 15 s) would not be felt locally by the solidified polymerwithin the mold. In fact, VAIM test specimens created with 10 s and 15 s duration’s exhib-ited sink marks that were a direct result of insufficient packing pressure during solidifica-tion. This suggests that at some critical time during the polymer injection process,oscillatory vibrational motion should cease in order to allow for a normal, necessary pack-ing procedure. Any polymer packing within the mold done by oscillatory pressure vibra-tions after this critical time was futile and actually was observed to degrade polymerstrength. A VAIM Duty Cycle of 2:1 was chosen as the optimal injection and decompres-

Figure 4. Preliminary experimental results showing VAIM parameters and their effects on final product ultimate tensile strength (UTS).

Vibration Duration (sec)

Delay Time to Begin Vibration (sec) Frequency Hz

Main Effects AnalysisMain Effects Analysis

Main Effects AnalysisMain Effects AnalysisUTS vs. Delay Time to Begin Vibration UTS vs. Frequency

UTS vs. Vibration Duration

Ulti

mat

e T

ensi

le S

tren

gth

Ulti

mat

e T

ensi

le S

tren

gth

Ulti

mat

e T

ensi

le S

tren

gth

Ulti

mat

e T

ensi

le T

est (

psi)

7100

71007100

7100

7200

7000

7000

7000

7000

6900

6900

6900

6900

6800

6800

6700

0 0.5 11

2

2 3

4 65 10 15

(psi

)(p

si)

(psi

)


sion time ratio that produced parts with the highest ultimate tensile strength, as shown inFigure 4(d).

REFINED VAIM OPTIMAL MOLDING PARAMETERS

During a secondary phase of the research,refined VAIM optimized molding conditionswere determined for a processing window thatincluded vibrational molding machine frequen-cies of up to 8 Hz. From an observed generaltrend previously obtained during the preliminarystudy, it was determined that the four levels ofparameter testing could now be narrowed downto 2 by eliminating variations in Duty Cycle andDelay Time to Begin Vibration to 2 to 1 and 0.5seconds, respectively. The results of the remain-ing two parameters, frequency and vibration

duration, versus UTS are plotted in Figure 5.In summary, the experimental results obtained in the determination of refined optimal

VAIM processing conditions on single gated polystyrene test specimens are shown in Table1. It is interesting to note that VAIM polystyrene test specimens produced an average tensilestrength at 8 Hz of 8,293 psi which is an increase in tensile strength of 1,463 psi when com-pared to control samples previously tested that exhibited an average maximum tensilestrength of 6,830 psi. This corresponded to a 17.6% increase in ultimate tensile strength.Also, reference control specimens exhibited a 273 psi variation in tensile strength whileVAIM specimens molded at optimal conditions exhibited only 106 psi strength variations.This is a reduction in tensile strength variation of over 60% and definitely shows that VAIM

Figure 5. Refined experimental results showing optimal VAIM frequency of 8 Hz and maximum vibration duration of 6 sec-onds correlating to tensile strengths of 8157 and 7867psi, respectively.

Table 1. Refined optimal VAIMmolding conditions for polystyrene

Delay time to begin vibration 0.5 s

Vibration frequency 8 Hz

Vibration duration 6.0 s

Duty cycle (Tinj/Tdec) 2:1

Frequency (Hz) Vibration Duration (sec.)

Main Effects:Part 2 Main Effects:Part 2UTS vs. Frequency UTS vs. Vibration Duration

Ulti

mat

e T

ensi

le S

tren

gth

Ulti

mat

e T

ensi

le S

tren

gth

(psi

)

(psi

)

3 4 55 6 6 77 88

820081008000

7900

7900 7800

7800

77007700

7600

760075007400

7850

7750

7650


deserves some merit as to improving final part quality. Also worthy of discussing is the factthat out of 81 VAIM experiments completed, only 3 produced ultimate tensile strengths thatwere actually lower than the average ultimate tensile strength result obtained from controlspecimens. The common factor shared by all three cases was that they were all molded witha Delay Time to Begin Vibration of 1 second. Again, this shows a minimal affect on poly-mer rheology that was translated to a less than minimal affect on the polymer final morphol-ogy.

CONCLUSION AND FUTURE WORK

The application of mechanical pressure vibrations to conventional injection molding pro-cesses has shown definite advantages and improvements to final product mechanical prop-erties as well as a 60% decrease in product quality variations. Besides showing a maximumincreased result in ultimate tensile strength of 21% for a commercial grade polystyrenematerial, VAIM has partially confirmed the theoretical effects of low vibrational frequenciesand their impact on material morphology. Although an amorphous polymer, such as Poly-styrene used in this experiment, exhibited improvements in tensile strength, the resultsshould not be considered equally true for crystalline and reinforced polymers since theyhave not been experimentally tested. Also, their molecular structure and flow property char-acteristics that characterize these types of materials vary substantially from amorphouspolymers.

Initial attempts at determining the validity of VAIM is in its infancy and the ground-work has been set for future studies. Future work with this novel injection molding processwill require determining VAIM effects on other polymer materials and expanding the rangeof testing conditions. For example, it was confirmed from experimental results that theoreti-cal low frequencies contributed greatly to final product morphology, but there is a concernas to the limitations of what exactly “low frequencies” mean. There is an experimental trendin VAIM data showing increased strength for increasing frequencies, but surely there is acritical frequency at which the advantages of expending more energy to exert higher vibra-tory oscillations will not out weigh the cost or effort of producing a stronger part. This gen-eral line of thinking can also be applied to other variables such as Vibration Duration, andTinj/Tdec, which must have the outer limits of their processing window set. Also, futurestudies should include the effects of amplitude dependency if any. Future studies willinclude the aid of pressure transducers equipped within the mold, and continued extensiveresearch on polymer behavior to determine a scientific knowledge of the local characteristicvariations and their relationship to the applied external global parameters.


REFERENCES1 Grossman, E.M. Scorim - Principles, Capabilities and Applications in ANTEC '95. 1995: Society of Plastics Engineers. p. 461-476.2 Ibar, J.P., Control of Polymer Properties by Melt Vibration Technology: A Review. Polymer Engineering and Science, 1998. 38(1): p. 1-20.3 Kalay, G. and M.J. Bevis, The Effect of Shear Controlled Orientation in Injection Moulding on the Mechanical

Properties of An Aliphatic Polyketone. Journal of Polymer Science Part B - Polymer Physics, 1997. 35(3): p. 415-430.4 Malloy, R., G. Gardner, and E. Grossman. Improving Weld Line Strengths Using A Multi-Live Feed Injection Molding Process. in ANTEC '93. 1993: Society of Plastics Engineers. p. 521-529.5 Theberge, J., IM alternatives produces performance advantages, in Plastics Engineering. 1991. p. 27-31.6 Ludwig, H.-C., G. Fischer, and H. Becker, Quantitative Comparison of Morphology and Fibre Orientation in Push-Pull Processed and Conventional Injection Moulded Parts. Composites Science and Technology, 1995. 53(2): p. 235-239.7 Michaeli, W. and S. Galuschka. Procedure for Increasing the Weldline Strength of Injection Molded Parts. in

ANTEC '93. 1993: Society of Plastics Engineers. p. 534-542.8 Cao, B., et al. Injection-Spin Process. in ANTEC '94. 1994: Society of Plastics Engineers. p. 2603-2606.9 Cao, B., et al., Injection-Spin Process, in Plastics Engineering. 1994, Society of Plastics Engineers. p. 47-49.10 Gardner, G. and R. Malloy. A Moving Boundary Technique To Strengthen WeldLine In Injection Molding. in

ANTEC '94. 1994: Society of Plastics Engineers. p. 626-630.11 Gardner, G.P. and R.A. Malloy. Use of The Moving Boundary Molding Technique to Straighten Weld Lines. in

ANTEC '96. 1996: Society of Plastics Engineers. p. 685-691.12 Ibar, J.P. Control of Performance of Polymers and their Blends through Melt Vibration Technology. Critical Review.

in ANTEC '96. 1996: Society of Plastics Engineers. p. 769-773.13 Ibar, J.P., Control of performance of polymers and their blends through melt vibration technology. Critical review. Progress in Rubber & Plastics Technology, 1997. 13(1): p. 17-25.14 Kikuchi, A. and R.F. Callahan. Enhancement of Molded Product Properties Through The Use of Rheomolding

Technology on Injection Molding Processes. in RETEC (PD3). 1996: Society of Plastics Engineers. p. 165-181.15 Kikuchi, A. and R.F. Callahan. Quality Improvements Resulting From Rheomolding Technology During Injection

Molding Processes. in ANTEC '96. 1996: Society of PLastics Engineers. p. 764-768.16 Kikuchi, A., et al. Molded Product Properties Enhancement through the Use of Rheomolding Technology on Injection Molding Processes. in ANTEC '97. 1997: Society of PLastics Engineers. p. 436-440.17 Kikuchi, A., et al. Control of Internal Stresses In Injection Molded Parts Through The Use of Vibrational Molding, "Rheomolding", Technology. in ANTEC '98. 1998: Society of Plastics Engineers. p. 2233-2237.18 Kikuchi, A., J.P. Coulter, and P. Santiago. Vibration-Assisted Injection Molding Technology For Improved

Manufacturing. in Competing in a Global Manufacturing Environment. 1999. Behtlehem, Pennsylvania: Lehigh University Center for Manufacturing Systems Engineering. p. 79-87.

Vibrated Gas Assist Molding:Its Benefits in Injection Molding

J.P. IbarEKNET Research, P.O. Box 385, New Canaan, CT 06840, USA

INTRODUCTION

It is well known to those skilled in molding polymeric materials that such defects as weldlines, sink marks, and warpage of the final part are caused by melt fronts collision, unbal-anced flow, uneven cooling, non-uniform internal stress and non-homogenous nucleationand growth of crystals as the part solidifies. Varying the processing parameters (e.g., tem-perature, pressure, flow rates, filling and packing time in the case of injection molding etc.)can result in the modification of the molded part outlook and final product's physical prop-erties, but the modifications are often slight and not quantified, and they also rely, to a largeextent, upon the expertise of the molding operator who uses his experience and art to deter-mine the molding processing parameters, the so-called "processing window". Unfortunately,more often than less, the conventional wisdom for a good processing window involvesincreasing the clamp tonnage, sometimes as high as 25-50,000 psi of pressure, which sub-stantially contributes to the price of the injection molding equipment.

The methods of gas assist molding have demonstrated their great usefulness in injec-tion molding to hollow parts out and induce an excellent surface finish. Most of the benefitsfrom applying vibrational energy to the molding process are well-documented.1 The idea ofusing pressurized gases to transmit vibrational energy to plastic as it enters and fills themold - and thereby improve the physical and mechanical properties of the molded item - isthe subject of a recent patent.2 The technology, called Vibrogaim (Vibration Gas InjectionMolding) is the next generation vibration molding technique providing melt manipulationcapabilities during molding. Vibrogaim's innovation comes from the use of pressurized gasas the tool for delivering the vibrations from generating devices into the plastic melt. Thegas is the means of controlling and maintaining vibration in the system. The present paperexplores the various benefits of vibrating gas during injection molding of plastics under gasvibration. Gas can be inserted in the mold prior to melt injection, and vibrated from the sub-sonic to the arsenic range in order to modify the filling process. Gas can also be inserted


prior to injection and, acting like a pressurized vibrated gas spring, help induce orientationbenefits during filling completion.

Vibrated air pressure, localized in specifically designed air-runners distributed aroundthe runners and inside the mold, helps fill and pack the mold, core out hollow parts and bal-ance flow in multicavity molds.

Compared to earlier processes that apply the vibrational energy to the entire mold inorder to affect the melt1,2 Vibrogaim, by employing pressurized gases to induce the reso-nance, opens up new ways to influence molding results.

One unique advantage of this technology is the ability to alter the morphology of theplastic and upgrade one or more mechanical properties, such as tensile strength, modulusand impact. Other benefits include improved surface finish and the potential to controlnucleation and crystal growth rates, reduce internal stresses and warpage, and cut down onsink marks and voids. Based on developmental work with Vibro-molding technology, up toa 25% gain in crystallinity in polypropylene can be achieved with a resulting improvementin low temperature impact and clarity. Also other advantages in molding process itself, suchas smoother melt flow and new latitude in controlling mold flow mechanisms.

The basic idea in Vibrogaim is its use of pressurized gas to manipulate the melt as itenters and fills the mold in order to upgrade or otherwise alter the properties of the part. Thepressurized gas fills the mold assembly, vibrations generated by electrical or mechanicaltransducers. The gas may be introduced directly into the mold from the back of the cavityand/ or through the injection nozzle. It also can be introduced, via channels or shaped cham-bers, at specific locations in the mold where the vibration effects are wanted, near the run-ners and gates.

A chemically inert gas, such as nitrogen, is used wherever there will be contact withthe melt (cavity, runner system, nozzle etc.). Air is used in the channels and chambers andother passages that are external to the flow path.

The Vibrogaim process can apply many types of vibratory energy to the molding pro-cess, depending on the effect wanted on the part or the process. The energy modes rangefrom subsonic, low frequency vibration (1 to 30 Hertz), easily obtained with help of pneu-matic piston actuators, going all the way up to the ultra-sonic range15,000 to 20,000 Hertz.In many cases, several waves of vibration are used simultaneously or in a sequence at vari-ous locations. The advantage of using air/gas is that it is quite straightforward to combinedifferent modes of various frequencies, various amplitudes. The different modes combinetheir effect on the plastic melt, increasing the elasticity of the melt for the shear oscillationof low frequency, and modifying the mechanism of flow through the runners and in the cav-ity, through the higher frequency modes. Such flexibility in melt manipulation is not avail-able from other techniques which do not use gas as transfer material.

Vibrated Gas Assist Molding 255

VIBRATED GAS MODIFIES THE FILLING PROCESS

In a second application of the use of vibrated gas (air in this instance), vibrated air inside anetwork of air runners located in the mold, near the gate and around plastic runners, is acti-vated to produce a reduction of the friction coefficient at the wall surfaces, allowing either abetter flow, a different type of flow or greater throughput. Alternatively, as already men-tioned, gas inserted in the cavity prior to filling can be put into oscillation and/or resonanceto influence the mechanism of filling, away from the common fountain flow at the tip. In thecase of knit lines, a short shot just ahead of the knit line is suspended by the gas pressure andhigh frequency vibrations propagated towards the flow front to achieve superior weldingresults. The gas is then released to allow the completion of the filling process.

VIBRATED GAS ACTS AS SPRING TO MODIFY MELT

One of the most novel applications of the technology involves pressurizing the mold cavitybefore the shot. The compressed gas (which may vary from 50 to 3,000 psi) acts like aspring, keeping the melt "floating" for a few seconds before it is allowed to completely fillthe cavity. During that time, the melt is subjected to oscillation by a series of low-frequency(10 to 30 Hz) pressure pulses which induce a large increase of its elasticity. The vibratoryenergy pumped into the melt changes the mechanism of deformation during subsequent fill-ing of the cavity, in particular it favors flow-imposed alignment of the polymer molecules,so that a higher percentage of the molecular bonds are forced into an orientation deforma-tion mechanism.

Orientation of the molecular bonds is the basis for existing strengthening by orientationprocesses and has been shown to directly affect the mechanical and physical properties ofthe molded plastic. Tensile strength, modulus and other mechanical properties can beimproved, depending on the specific process parameters. The process of providing.orienta-tion benefits during injection molding requires the right combination of process variables,or the properties' improvements are non-existent. The gas pressure and the vibration charac-teristics must be in the right range to oscillate and reorient the melt. Also, the plastic’s rheo-logical temperature cannot be too warm (no more than 45oC above its (frequencydependent) glass transition temperature, so it won't have time to relax and loose the orienta-tion before it can fill and pack the mold.

Gas pressure can be used in other ways inside the mold besides molecular orientation:helping to fill and pack the mold, core out hollow parts, and help balance flow in multi-cav-ity molds. These applications require precise control of temperature and pressure as well asinstantaneous timing.


In addition to the low frequency vibrations that produce the molecular reorientationdescribed above, the Vibrogaim process also uses vibration frequencies up in the ultrasonicrange (15 kHz to 20 kHz or higher). These frequencies have other types of effects on theproperties of the molded part or process. For example, ultrasonics can be used to induce alocal fusion of the weld lines or adjust the balance between the rate of crystallization andcrystal growth in semi-crystalline plastics. Smaller crystals can boost impact resistance,light transmission and chemical resistance, while larger crystals can improve mechanicalstrength.

In addition, high-frequency vibrations accelerate the stress relaxation of semicrystal-line and amorphous plastics. The faster the return to the equilibrium state, the less thechances of molded-in stress. Another benefit is reduced friction between the melt and therunner walls and other areas in the flow path. Minimizing friction translates into fastermolding and better part surfaces as well as lower shear stress and, in the case of opticalproducts, birefringence.

CONTROLS

Most of the benefits from applying vibrational energy to the molding process are well-docu-mented.1 Vibrogaim's innovation comes from the use of pressurized gas as the tool fordelivering the vibrations from generating devices into the plastic melt. The gas is the meansof controlling and maintaining resonance in the system.

Resonance is the condition in which the components of a system - in this case, the plas-tic, mold cavity, mold passageways, etc. - vibrate at the natural frequency of the system. Thetrick is to find the natural frequency, or to tune the system in order to extract the energyfrom the vibrations and produce a physical change in the targeted item. This is done step bystep with use of an acoustic chamber connected to the mold cavity during filling. The reso-nance frequency is determined as a function of time as the cavity is filled with hot melt.

The Vibrogaim technology uses a sophisticated computer control system designed tomonitor and maintain resonant conditions in the mold assembly or other machine compo-nents. The closed loop, high speed system integrates inputs from the pressure, temperatureand vibration sensors and adjusts these parameters instantly to track with the events of themolding process.

Stored programs in the computer, specific for each job, dictate the sequence of controlactions, including when the gas is injected into the mold or nozzle or the air into the moldchannels, at what pressure, when venting occurs, when vibration is initiated and at what fre-quency and power level. The principal hardware items for the Vibrogaim technology con-sists of the vibration sources, gas mixing units, gas and air pressurizing/injection units,valves and other control devices. The process hardware and the computer system are sup-

Vibrated Gas Assist Molding 257

plied as a portable package, capable of being used on any molding press. Specific applica-tions in the automotive industry will be presented at the meeting.

CONCLUSIONS

The present paper explores the processing of injection molded plastics under gas vibration.Vibrated gas can be used for several purposes. 1. Gas can be inserted and vibrated in themold prior to melt injection to modify the filling process mechanism, fuse knit lines, healsink lines and other defects due to flow imperfections. 2. Compressed Vibrated Gas can actlike a pressurized vibrated gas spring, which helps induce orientation benefits in the shortshot during filling completion. 3. Vibrated air pressure, localized in specifically designedair-runners distributed around the runners and inside the mold, helps fill and pack the mold,core out hollow parts and balance flow in multi-cavity molds. 4. Vibrated Gas can also beused to tag parts for recognition during recycling or later inspection.

Vibrogaim technology uses pressurized and vibrated (or pulsed) gas to manipulate themelt as it enters and fills the mold in order to modify the flow pattern (especially when knitlines form) and/or alter the properties of the part. The pressurized gas fills the mold assem-bly, vibrations are.generated by pneumatic, electrical or mechanical transducers. The gasmay be introduced directly into the mold from the back of the cavity and/ or through theinjection nozzle. It also can be introduced, via channels or shaped chambers, at specificlocations in the mold where the vibration effects are wanted, near the runners and gates. Thepaper reviews hardware and controls requirements to apply this novel technique to injectionmolding.

REFERENCES1 Plastics Blow Molding Handbook, Norman Lee Editor, Van Nostrand Reinhold, New York (1990). 2 P. J. Zuber, Relationship of Materials and Design to Gas-Assist Injection Molding Application Development in

Molding 95, ECM Fith International Conference and Exhibit, March 27-29, 1995, New Orleans. Also see: Gas-Assist Injection Molding, "Design and Processing guide for GE Resins", a GE Plastics publication. H. Eckardt, Annual

Conference, SPI Structural Division, 18, 57. S. Shah and D. Hlavaty, ANTEC 91 Reprints, 1479 (1991). S. Shah, "Gas Injection Molding:Current Practices", ANTEC 91 Reprints, 1494 (1991). K. Beattie, "Developments in Cinpres Gas Injection Nozzles", Molding 95, Fith International Conference and Exhibits, March 27-29 1995. P.L. Medina,

L.S. Turng, V.W. Wang, "Understanding and Evaluating Gas-Assisted Injection Molding Applications via ComputerSimulation", paper presented in the Structural Plastics 19th Annual Conference, Society of the Plastics Industry, Atlanta,

Georgia, April 1991.3 J.P. Ibar, Control of Polymer Properties by Melt Vibration Technology. A Review. Polym. Eng. Sci., 38(1), 1 (1998).

The References provides a list of all Patents.4 J.P. Ibar, Method for Exerting Stress Tensor to Molding Material, US Patent 5,543,092.5 J.P. Ibar, U.S. Patent 5,605,707 and PCT Application, Method and Apparatus for Controlling Gas Assisted Injection Molding to Produce Hollow and Non-Hollow Plastic Parts and Modify Their Physical Characteristics (1995).6 B. Miller, Closed-Loop Controller Aids Gas-Assist Control, Plastics World, July 1995, p.13.

Chpater 6: Mold Making and PlasticizationAdvances in Stack Molding Technology

Vincent Travaglini and Henry RozemaTradesco Mold Limited, Rexdale, Ontario, Canada

INTRODUCTION

In any industry, increasing competition coupled with the need for higher levels of productiv-ity motivates suppliers to provide products with higher quality, in a shorter time and at alower price. In an industry where processes are already very efficient, the most effectiveway to accomplish this is through technical innovation.

With respect to injection mold making, one of the major technological advancementswas the development of the conventional two face stack mold. The ability to mold on twofaces provides twice the output from the same molding machine. Prior to this parts weretypically molded on one face (single face) in either single or multi cavity configurations.Molding on two faces has been widely accepted over the last two decades.

Over the last few years, there have been further technological advances that improveboth the productivity and the flexibility of molding operations. These advances include: 1. Four face (level) stack molds. 2. Quick Product Change, QPC, systems. 3. Two cavity stack molds for large parts.

Four level stack molds essentially quadruple the output from single face molds. QPCmolds allow molders to switch from one product to another in less than one hour in both sin-gle face and two face stack mold configurations. Two cavity stack molds allow large parts tobe molded in a back to back configuration, thus doubling the machine capacity.

The key to each mold configuration is based on the development of the valveless melttransfer system. For each of the stack mold configurations listed, the following paper willdetail the engineering principles and technical application of this new hot runner technol-ogy.

CONVENTIONAL TECHNOLOGY

The simplest mold design, from a runner point of view, is the single cavity, single face mold.The machine nozzle injects plastic directly into the cavity. The single face mold can also be


extended to a multi cavity layout. In this case,the machine nozzle injects the melt into a runnersystem that feeds each individual cavity.

With the introduction of the conventionaltwo face stack mold, parts could now be moldedon two faces. This mold design has been widelyaccepted in the thin wall injection moldingindustry, typically with hot runner designs. Themelt is injected into an extended sprue (spruebar) which is attached to a manifold system. Themanifold feeds each cavity on both molding lev-els (see Figure 1 & Figure 2)

From a rheological standpoint, since the runner has increased in length, both residencetime and pressure losses in the runner system have increased. Attention must be paid toensure the hot runner is balanced and all cavities are fed at the same pressure and tempera-ture.

The largest advantage to this design is the doubling of machine capacity. By placingthe cavities back to back, the stack mold takes advantage of injection force cancellationsuch that the same machine as a single face mold can be used. A 10-15% increase in ton-nage is required to offset the force that the machine nozzle applies to the sprue bar.

Figure 1. Runner layout.

Figure 2. 8 cavity (2x4) conventional stack mold.

SPRUE BAR

VMTSVMTSVMTS

VMTS

SPRUE BAR

FOUR FACE STACK MOLD TWO CAVITY STACK MOLD

SINGLE FACE MOLD CONVENTIONAL STACK MOLD QUICK PRODUCT CHANGE

STACK MOLD

MELT PAIR

Advances in Stack Molding Technology 261

The sprue bar in the conventional stack mold design is typically located on the center-line of the mold. This results in two major drawbacks. First, the cavities cannot be placed inthe center of the mold, they must be placed around it. Secondly, parts can be damaged uponejection since they may come in contact with the hot sprue bar on the first molding level.

TECHNOLOGICAL ADVANCEMENT

The engineering challenge arose with the idea to extendthe stack mold design from two to four levels. A Valve-less Melt Transfer System [VMTS] was designed in orderto pass the melt across the mold parting line. This VMTSis engineered to provide a tight seal each time the mold isclosed and the melt is injected at pressures in the range of20,000 psi. It is a self compensating design that allows forvariations in nozzle thermal expansion. When the mold isopened and the VMTS is disengaged, drool is avoideddue to the self decompression of the central hot runnersystem. There are no heavy wear items as associated withvalve shut off systems (see Figure 3).

APPLICATIONS

FOUR FACE STACK MOLDS

The four face stack mold is essentially two stackmolds placed back to back. Due to increased moldshut height and plasticizing requirements, it is wellsuited for the high volume production of shallowparts such as thin wall lids, for the packagingindustry (see Figure 4).

The first application of the VMTS was in thefour face stack mold design. The melt is injectedinto a heated sprue bar along the mold centerline.It is then transferred to the non operator side of themold where it splits in two directions. In each

direction the melt passes through the VMTS as it crosses the parting line and enters into itsrespective hot runner block. From there it is routed through a fully balanced hot runner man-ifold where it feeds individual nozzles for injection into the cavities (see Figure 1). Theentire hot runner system is thermally and rheologically balanced such that the first cavity on

Figure 3. Hot runner system for a two cavity stack mold.

Figure 4. 96 cavity (4x24) four level stack mold.


the first level and the last cavity on the fourth level are filled at the same time, temperatureand pressure. With the extended runner length, shear rates, pressure drops and shear heattemperature rise are considered when sizing all runner lines.

The primary advantage to the four level stack mold is increased plant efficiency. With-out modifying the injection molding machine, output is essentially quadrupled over a singleface mold. Savings are realized by decreased manufacturing costs which can be passed on tothe customer. Cycle times are extended by fractions of a second since more time is requiredto open and close the mold over all four faces.

As in the conventional stack mold, the four level stack mold also takes advantage of theforce cancellation by positioning the cavities in a back to back arrangement. By placing anadditional two levels in the mold, clamp tonnage is not increased over the conventionalstack mold.

Other technical features include synchronized opening provided by either a harmoniclinkage system or series of splines. The linkage system not only ensures equal openings onall four levels but also activates stripper plates to eject the parts. If the spline centeringdevice is used, standard part ejection can be performed by cam actuated linkages or throughthe use of internal hydraulic or pneumatic cylinders. Due to additional mold weight, thethree center sections of the mold are supported on the machine tie bars using adjustablemold supports.

Since its inception in 1991, the four face stack mold has been successfully designedand manufactured in various cavitations. After the first 8 cavity (4x2) mold there have beennumerous additional four level stack molds produced; the latest being a 96 cavity (4x24)mold running at a six second cycle. With each new design, thermally and rheologically bal-anced hot runner systems are engineered to meet the increasing demands of molding envi-ronments.

QUICK PRODUCT CHANGE, QPC, SYSTEMS

QPC refers to tooling designed for the quick changeover of core and cavity module setswithout having to change over the entire mold. The main technological objective in thedevelopment of this mold design was to reduce the time required to change from one prod-uct to another using a large multi cavity stack mold. The intention is to produce groups ofrelatively low volume products in an efficient stack molding environment. Technical fea-tures include: • QPC mold operates in the same molding machine as a conventional stack mold • QPC mold operates at the same cycle time as a conventional mold • the hot runner system is thermally and rheologically balanced


• shot to shot change over time is less than one hour for both mechanically and air ejectedparts in the stack mold configuration

• the mold design allows upgrading from a single face mold to a stack mold • machine shut height remains the same from one product to the next

The QPC system consists of twobasic components: a unique hot runnercarrier frame and interchangeable coreand cavity sets which are housed in platemodules. All necessary water, air andelectrical connections are designed toremain permanently within the carrierframe. Thus when a product change ismade, there is no time lost in re-installingthese services (see Figure 5).

The sprue bar in the conventionalstack mold design presents an obstacle inapplying a quick mold change technique.

It is virtually impossible to access and remove plates or modules from the stationary sidewithout removing the entire mold. With the application of the VMTS, this problem is elimi-nated. The melt is injected at the mold centerline and is immediately transferred around thecore and cavity modules via a hot runner manifold. The plastic then crosses the mold partingline with the VMTS and passes the melt to a fully balanced, central distribution manifold.This manifold feeds the individual nozzles for injection into the cavities (see Figure 1).

The QPC design s major advantage is the flexibility that the quick mold changeoversoffers molders. This includes the ability to minimize inventory levels and costs and main-tain $Just in Time# delivery schedules. Molders with high product changeovers will realizelower manufacturing costs per part due to the time savings in mold changeovers. The modu-lar system allows molders to start projects initially on a small scale. Expansion from a singleface to a stack mold can be done using the original core and cavity module sets. As a resultmolders can compete with high efficiency tooling without the high initial capital invest-ment.

Latest advances include modular QPC stack molds for the production of cutlery items.These molds apply hot and cold runner technology along with cam followed ejector systemson both molding levels. In addition, a design of a four level stack mold incorporating theQPC system has been completed.

Figure 5. 24 cavity (2x12) Quick Product Change, QPC, system.


TWO CAVITY STACK MOLDS FOR LARGE PARTS

Typically large parts are molded in single cavitytools due to tonnage and platen size limitations.Given conventional technology, in order toincrease the cavitation of the mold in either thesingle face or stack mold configuration, a largertonnage machine is required. The technical chal-lenge was to increase the cavitation of the moldwithout having to change the press. Similar tothe QPC hot runner design, the VMTS wasapplied to the two cavity stack mold. By placingthe cavities in a back to back configuration, theresult is a two cavity stack mold that runs in the

same molding machine as does the single cavity (see Figure 6). As with the QPC design, the melt is injected at the mold centerline into a hot runner

manifold housed within the stationary core backing plate and transferred around the coreand cavity set on the stationary side. It then crosses the VMTS into a simple hot runner man-ifold that feeds both parts (see Figure 1). This concept has also been expanded to feedingback to back valve gates.

This technological advancement provides molders with a considerable advantage overtheir single cavity competitors. Output is doubled without having to invest into a largermachine to increase cavitation. As with the four level mold, plant efficiency increases andmanufacturing costs decrease dramatically.

This technology has been initially applied to the low pressure molding of heavier wallitems with low L/t ratios. Recently, it has been applied to molding thin wall containers in aback to back configuration with a wall thickness of 0.7 mm (0.028") and a resulting L/t ratioof 300.

CONCLUSIONS

The development of the valveless melt transfer system has opened numerous doors in stackmold design. The ability to transfer high pressure melt around cavities and across partinglines has provided for the #next generation# stack molds such as: 1. Four face (level) stack molds. 2. Quick Product Change, QPC, systems. 3. Two cavity stack molds for large parts.

Figure 6. 2 cavity (2x1) stack mold.


The four level and two cavity stack molds provide molders with significantly lowermanufacturing costs and increased productivity from existing molding machines. QuickProduct Change molds offer flexibility to molding operations, making shorter runs moreeconomical.

Advanced Valve Gate Technology for Usein Specialty Injection Molding

John Blundy, David Reitan, Jack SteeleIncoe Corporation

INTRODUCTION

The use of valve gate hot runner systems has been generally accepted in our industry as amore precise method of controlling the gate vestige, allowing the user positive open/closecapability. The use of larger orifices would allow for faster fill and part stress reduction.However, due to the complexity, these systems do present areas of concern. Considerationfor operating the mechanism either hydraulically or pneumatically must be properlydesigned to provide reliability. Contamination generated by material passing around theshut-off pin can cause flow lines on the finished part with some resins. Absolute control ofthe open/close position, which is usually based on time, can present difficulties. Advancesin system design, computer technology, materials and manufacturing processes haveallowed great improvements in valve gate systems. These advanced systems will bereviewed in the following four sections of this report.

CO-INJECTION

Co-injection, the molding of two similar or dis-similar resins in the same molded part creating aseparate skin of one material and a separate coreof a second material, requires the use of valvegates. Advanced valve gate systems are key tosuccessful single cavity, multi-cavity or sequen-tial molding for co-injection applications. Thespecial valve gate cylinder provides for three

unique positions for resin flow and controls the volume of skin and core material (Figure 1).Position one starts the injection of the skin (surface) layer into the cavity. This is followedby the second position which starts the flow of the core material while continuing the flowof skin material. After the desired amount of core is achieved, the pin moves to the second-ary position allowing the skin material to fully encapsulate the core. Finally, pack pressure

Figure 1.

POSITION 1 POSITION 2 POSITION 3


is applied prior to shutting off both skin and corematerials. This unique three position actuator,along with the shut-off pin and mixing pin, pro-vide the necessary mechanical flow channelsand precise shut-off sequence for successful co-injection processing. This advanced system isthe result of a joint technology between INCOECorporation and Bemis Corporation. INCOE’spatented valve gate technology along withBemis’ processing “know how”, provide thestimulus for improved co-injection technology.This part (Figure 2) illustrates a part perfor-mance advantage in that the overall rigidity ofthe arm rest was improved while allowing thedesired cosmetic requirements to be maintained.Various features and benefits can be realized by

the co-injection process including improved engineering properties, lower product cost andcycle time reduction.

GAS-ASSIST VALVE GATE

When applying gas-assist technology for applications where the gasis introduced in the mold, in either the part or runner system, valvegate technology is, in most cases, required. This process requires avalve gate to allow for proper pack-out and to keep the plastic fromflowing back into the manifold. This example (Figure 3) is typical ofany gas-assist system and can be duplicated for multiple cavities orfor sequential fill applications. The systems are designed such thatall valve gates are closed prior to the gas injection cycle.

CLEAR-FLO TM VALVE GATE

The development of the Clear-FloTM valve gate system (Figure 4) has advanced the tech-nology by removing the valve gate shut-off pin from the material until the exact point ofclose. This advanced design concept opens the door to the use of shear/heat sensitive mate-rials in valve gate applications. The flow channel is increased and flow separation “which isinherent in conventional valve gates” is eliminated. This larger flow channel also minimizesflow induced material stress. The design allows the material to maintain a consistent veloc-

Figure 2.

Figure 3.

Advanced Valve Gate Technology 269

ity from the manifold entrance to the gate area.These advantages corrected a processing problem

shown in this lawn tractor hood application (Figure 5). The quality issue was a shadow orstreak created by shear sensitive colored resin. The shut-off pin separates the material flowcreating a difference in velocity which would cause the color concentrate to be somewhatdarker, thus, creating an imperfection on the part. The larger flow channel provided a bettercondition for the engineering resin. The rate of reject was reduced from 18% to near zero.Color changes are dramatically improved as the absence of the shut-off pin eliminates theflow separation condition as well as creating an area that is not directly removed by the newresin.

This advanced design also incorpo-rates several other advantages. First, astraight shut-off pin is used (Figure 6).This concept provides positive sealbetween the shut-off pin and pin chan-nel, which virtually eliminates plasticleaks, as well as providing for animproved gate appearance. Additionally,the actuator provides for improved reli-ability by reducing the components from19 to 9 and uses either pneumatic orhydraulic power (Figure 7).

Figure 4.

Figure 5. Figure 6.

Figure 7.


SEQUENTIAL GATE CONTROL SYSTEM

Further, improving the technology available in the area of valve gate pinposition control has been achieved using a newly developed Sequential GateControl System. It is the result of developing and using proprietary softwarewith the latest in Computer technology. Based on Windows Software, thissystem can be used on any valve gate system and can be interfaced with allinjection molding machines. This development is especially beneficial forsequential molding, co-injection or the Clear-FloTM systems (Figure 8) sincepin position directly influences molded part quality. The controller providesprecise control based on linear movement of the injection screw. The systemis designed to allow for two open and close sequences per valve gate cycle.The sequential valve gate system can be activated by time and/or linear posi-tion using either the inch or metric scale. The system includes many uniquefeatures such as system ready protection, modem hook-up for trouble shoot-

Figure 8.

Figure 9.

Advanced Valve Gate Technology 271

ing or upgrade, mold file storage for quick set-up security code and 24 hour printable runhistory. The system can control up to 40 separate valve gates for positive open and closeactuations (Figure 9).

SUMMARY

Valve gate systems continue to be a very important part of many of today’s molding applica-tions. Valve gates are being used for co-injection both single cavity and multi-cavity molds.Valve gates are, in most cases, mandatory for gas-assisted molding. Valve gate systems arealso used for sequential molding, family molds and over molding. The refinements in valvegate technology will continue to advance the capabilities of the plastic molding industry.

In-mold Labeling for High Speed,Thin Wall Injection Molding

Gary FongTradesco Mold Ltd.

INTRODUCTION

This paper introduces a simple and cost effective IML system developed for high speed,multi-cavity injection molds. After this brief introduction, some of the basic as well as spe-cific design issues will be discussed. In no way can this article cover the many facets of IMLsystem design. To conclude, a brief overview of future design considerations will be pre-sented.

The concept of in-mold labeling has existed for over 14 years.1 IML have been appliedto all forms of plastic packaging: thermoforming, blow molding, and injection molding.IML technology can be found in European, Asian, and other foreign molding shops. IMLsystems are readily available for high volume packaging applications. However, IML sys-tems for high speed, multi-cavity injection molding applications are rather expensive. Thetechnology ranges from simple mechanical swing arms, to complex robotic arms with com-puter control. The type of IML system to incorporate into a process largely depends on theproduction level, as well as the cost of post-printing. In general, an IML system willincrease the cost of a mold by 10%-40%.2

The main drive for this project was to develop an IML system that was designed inconjunction with the mold. Currently, it is the trend for the injection machine supplier, or arobotics supplier to implement the IML system. This leads to a larger IML developmentteam, involving multiple companies, and a certain amount of confusion. It is believed thatthe concurrent design of the mold and IML system will result in a synergistic molding sys-tem.

This IML system was developed for a four-cavity lid mold. However, future adapta-tions of this system could see its use in multi-level, multi-cavity molds or even single cavity,single face tools.

TECHNICAL ISSUES

There are many issues involved with an IML system, some of which are


• label properties• spatial limitations• mechanically or electrically driven• robustness of the IML system• material selection of IML components• path of transfer arm into the mold space• label positions at start and end of cycle• continuous label feed mechanism

LABELS

The labels used for this project were made from oriented polypropylene (OPP), pre-cut tosize. The surface was treated with an antistatic coating to facilitate label separation by theIML system. The labels were sized at 91.80 mm [3.614”] by 111.5 mm [4.389”], with fourcorner fillets of 25.40 mm [1.000”]. The thickness were 0.080 mm [0.003”].

No static charging devices were used to assist label handling in this system. Rather, avacuum was used wherever the label was required to ‘adhere’ to a surface, namely the pick-up plate and the cavity surface.

BASIC DESIGN CONSIDERATIONS

Spatial limitations are a major concern for an IML system. The IML system requires suffi-cient room to move its transfer arm from the label holder into the mold space. The path ofentry into the mold space can be accomplished either from the top or the side of the mold. Ingeneral, a mechanism at the bottom of the mold will interfere with falling parts during thepart ejection stage.

The working space for the IML system also dictates its design. In general, the largerthe work space, the simpler the design requirements. The available space is dictated by thetie-bar space, mold size, as well as molding environment. A large tie-bar space, with a smallmold and a very spacious molding environment will provide the most workspace for theIML system. In reality, this situation does not occur very often.

This IML design was chosen to enter through the side of the mold space. Room wasmade available during the design of the mold to accommodate an IML system. The molderwas also willing to further accommodate the IML system, by increasing the size of thesafety gates on the machine. The complete removal of the gate was avoided, as only a fewinches of extra space was needed.

The key criteria when choosing an IML system are speed, reliability, simplicity, androbustness.

In-mold Labeling 275

The core system for this IML system is based on an electro-mechanical design. Thevarious pneumatic pistons and vacuum generators are controlled with a PLC. The motion ofthe transfer arms are mechanically controlled by a spline, driven by the moving plates of theinjection machine.

The speed of this IML system is controlled by the rate at which the mold opens andcloses. By tying the system directly to the mold, it does not increase the cycle time. Thespeed of the IML system is very important for high-speed injection molds. A delay, due toan IML system’s slower speed, can strongly affect the molder’s bottom-line.

The mechanical control over the motions of the transfer arm provides the molder withpeace of mind, as the sliding arm should never be caught between the closing plates. Aswell, the repeatability of the transfer arm’s motion is maintained due to its mechanical con-nection to the mold.

The ‘Mechanical Part Removal System’ was chosen as the core technology behind theprototype IML design. The ‘Mechanical Part Removal System’ was originally designed toremove plastic lids from a multi-level, multi-cavity mold. This system was mounted directlyonto the side of the mold. The simplistic design of this mechanism allows for easy mainte-nance whenever required.

Perhaps one of the greater concerns for an IML system is robustness. The systemshould, ideally, last forever. The ‘Mechanical Parts Removal System’ has successfully oper-ated in the field through 3 years of continual service. This in itself is a good measure of thebasic system’s robustness.

SPECIFIC DESIGN CONSIDERATIONS

Simply speaking, most IML systems have both a transfer arm and a label holder. Both com-ponents in this IML prototype are bolted to the side of the mold with holder arms. The moldis first dropped into the machine, then the IML assembly is attached to the side of the mold.Assembly of the various IML components must be manageable by a lone technician. Thusthe system should be lightweight and compact.

LABEL TRANSFER ASSEMBLY

The following is a quick overview of the label transfer assembly components; the assemblyincludes the transfer arm, pick-up plates, linkage, bearing housing, and spline. Their func-tion, and the material choices made for the components are discussed. Refer to Figure 1 forpart identification.

The system’s transfer arm is made from aluminum to reduce the inertial loads beingapplied onto the links. Though some loading is applied to segments of the arm, it wasn’t feltto be significant enough to warrant concern. A steel transfer arm would have created prob-


lems with regards to the sliding wear and linkage strengths. The linear motion of the transferarm is controlled through a set of linear guide rails. The linear guides limit the slidingmotion to a lateral direction.

The pick-up plate is attached to the transfer arm with four pins and a piston. The pistonis pneumatically controlled to drive the plate forward and back while the pins maintainalignment. The labels are affixed to the plate through the use of vacuum holes. The pick-upplate is made from aluminum, for its light weight and its low strength as compared to themold steel. If the label plate, for whatever reason strikes the molding surface, then the platewould deform rather than the tool. This protects the tool.

The linkage system transmits the force from the bearing housing to the transfer arm. Itis made from steel to maintain its strength and durability. The wear at the joints is reducedthrough the use of special washers and self-lubricating bushings. As these componentspotentially cycle between 5-8 seconds, wear issues were of some concern. The weight of thelinkage arms was also an issue, and as such, the linkage arms were designed for maximumstrength, while maintaining a reasonable weight.

Figure 1. Picture of the in-mold labeling transfer arm assembly.

bearing housing

spline

tie bar

linkage

linear guide rails

Direction of Motion

transfer arm

magazine

Mold

pick-up plate


The bearing housing, in turn, drives the linkage system. Thebearing housing is fixed onto the moving side of the mold. Thebearing housing is made from both steel and aluminum parts.Dynamic components in the bearing housing follow the profile ofthe spline to provide the motion needed to drive the transfer arm.The bearings used in the housing were chosen to maximize, ratherthan reduce, contact area. A small bearing contact would gougeinto the surface of the spline in long-term applications.

The spline is the source of power for this mechanical system.The spline is designed to provide the desired angle of twist to the

bearing housing. Figure 3. shows the spline used in theprototype. CNC tool paths created from the 3D geometrywere used on a steel rod to obtain the desired profile.After the piece was completed, it was.checked with theCMM machine to ensure that the proper rotational anglewas imbued upon the spline.

One end of the spline is fixed onto the stationary sideof the mold. The other end is held by the bearing housing,which is fixed to the moving side of the mold. When themold opens and closes, the linear force from the injectionmachine is transformed into a rotational force by thespline’s profile. This rotational motion is then transmittedby the bearing housing, through the linkage system andtransformed back into a linear force by the linear guiderails on the transfer arm. Thus any error in the spline pro-file would affect the transfer arm’s linear movement. This

potential problem is addressed later in this article.

LABEL HOLDER

The label holder for this IML system was designed to accommodate lids of various sizes.Though further development is required, this label holder design provides some insight intothe various requirements for a label holder. The label holder will also be referred to as themagazine. Figure 4 shows an assembly of the magazine.

Several factors were considered for the design of the magazine: ‘quick change’ capa-bility, and the label position with respect to the pick-up plates. Since the labels themselvesare relatively small, the use of a magazine system is not too unreasonable.

Figure 2. Picture of the in-mold labeling system on the four cav-ity mold.

Figure 3. Spline used to drive the IML sys-tem.


The magazine was designed to holdapproximately 300 labels. The labels can bechanged during production by simply replacingone magazine with another. The magazine slidealong grooves in the holder arms, which arefixed to the sides of the mold plates. This ‘quickchange’ characteristic was required for this IMLsystem, as the molder had multiple label designsfor the same part. The label holder was madefrom aluminum to minimize the handlingweight.

The labels are held within four thin plates.These four plates can be adjusted to change thelabel’s position in the magazine, which alsochanges the label’s position on the cavity sur-face.

As mentioned earlier, an error in the splineprofile will cause the transfer arm to deviatefrom its desired position. This deviation, though,should be fairly constant as the mold is generally

opened and closed at a consistent distance. This deviation can be corrected by adjusting thefour plates in the magazine. For example, if the transfer arm is deviated by X mm, then thefour plates can be adjusted by –X mm to correct the problem. The labels would still bealigned to the center of the cavity surface. This methodology can be applied to fix othererrors in the mechanical system as well.

There is an advantage and disadvantage for this method of alignment. On one hand,full flexibility is provided to correct alignment issues. On the other hand, it is very time con-suming to adjust each of the four plates for each magazine. Future developments will deter-mine a median solution that allows positioning control, while minimizing adjustment time.

FUTURE DESIGN CONSIDERATIONS

Some of the goals in the next phase of this project would be to improve the label holderunits, reduce the weight of the system, and further optimize the system’s design.

Since space was available along the sides of the mold, a concept of a label magazinewas simple to adopt. However, if the labels were much larger, the magazines would becomea problem. The use of a lighter material would be needed for the magazine.

Figure 4. Fully adjustable label holder (magazine).


A different approach for large label applications could involve the use of intermediatelabel handling systems. The intermediate handling systems would transfer the labels fromoutside the injection machine space to a fixed position for the transfer arm. By moving thelabels outside the injection machine, problems with label size, storage, and orientationwould be reduced significantly. Unfortunately, the design of a reliable and robust intermedi-ate handling system could prove difficult.

Development of an ‘infinite’ label supply system will also be considered. The mainobjective being to facilitate the refilling of labels, without stopping the injection machine.The prototype magazine design is considered a continuous label feed mechanism, thoughonly for a relatively short period of time.

Attempts will also be made to reduce the size of the various components in this system.This decrease in size will reduce the loads on the system, as well as reduce the cost frommaterial and machining requirements. Strength and robustness of the system will not becompromised. Scalability is a large issue for this IML system, as adaptation to a large orsmall mold is desired. Further investigation into making the transfer arm components scal-able will be included.

There are many ways to design an IML system. There are even more ways to improveon one. However, with the introduction of this specific system, it is envisioned that in-moldlabeling will soon be cost effective for all types of molds, operating at various productionlevels.

ACKNOWLEDGMENTS

I would like to thank Vincent Travaglini, Don Hersey, Henry Rozema, and Joe Klanfar, atTradesco Mold Ltd., for their support throughout this project. I would also like to thankMichael Sava, an advisor from the National Research Council Canada, for his support.

REFERENCES1 Mike Fairley, Overview of the in-mould label market- size, growth, and types; presented at In Mould Labelling

Innovations and Opportunities, March 25-26, 1996.2 Achim Franken, Injection-moulding equipment and robotics; presented at In Mould Labelling Innovations and

Opportunities, March 25-26, 1996.

Advances in Fusible Core Technique

E. Schmachtenberg, O. SchröderUniversity of Essen, Institute for Plastics in Mechanical Engineering (IKM),

Altendorfer Str. 3, 45127 Essen, Germany

INTRODUCTION

The basic production sequence of fusible coretechnology is shown in Figure 1. A metal core isproduced from a low-melting metal alloy, usu-ally a eutectic tin/bismuth alloy. It is theninserted into an injection mold where it is encap-sulated by the polymer. Afterwards, the core isremoved in a heating bath where the plastic partcomes into contact with a heated liquid medium,which is removed in the subsequent cleaningprocess.1

In the product development, the feasibilitystudies must take into account each process step and the interactions between the materialsused (polymer, core material, mold materials and heating medium). For example, the corematerial must not melt or soften during injection molding. The polymer must be compatibleto the heating medium, and the molten metal must not corrode the material of the core-cast-ing mold.

SUITABILITY OF RAW MATERIALS

Melting out of the core usually takes place in a bath of modified polyglycolether as heat-transfer medium. Melt removal can be assisted by induction coils to make this process stepfaster. Nevertheless, the same heat-transfer medium is used. Thus, using conventional pro-duction plants and methods the product designer has to make sure that the polymer is com-patible to the modified polyglycolether at a temperature of 160°C.

Table 1 shows a survey of the materials investigated so far at the University of Essen/Germany, concerning the compatibility to the heat-transfer medium. The materials classi-fied as suitable are currently undergoing further feasibility studies concerning the process

Figure 1. Basic production sequence of fusible core tech-nology.

encapsulatingraw materials core removal

cleaning ofpartscasting of

cores

part


step of injection molding. These studieshave take into account the aspects describedin the following.

PREVENTING CORE MELTINGAND CORE SOFTENING

During injection molding there are twomajor problems which may occur: shiftingof the core and melting of the core.

The problem of shifting of the core infusible core technology can be dealt with ina similar way to that used in conventionalinjection molding. It should be taken intoaccount that the core material has a compar-atively low stiffness and strength which are

highly temperature dependent. Figure 2 shows the corresponding stress/strain curves. It isevident that core shifting in fusible core technology is a question of temperature. The higherthe maximum temperature of the core the more high mechanical loading of the core has tobe avoided. A appropriate choice of the gate position and additional core supports may benecessary.

Core melting occurs as a result of a local and/or global temperature increase of thecore. The local temperature increase is particularly prevalent in the gate region of the cavity.This Problem can be estimated by the contact temperature:

[1]ϑcontact

bpϑp bCϑC+

bp bC+--------------------------------=

Table 1. Compatibility of raw materialswith heating medium.

Raw materialGlass fibre

contentAssessment

PA6 and PA66 30 to 50% Suitable

PPA 33% Suitable

PBT none/30% Suitable

PET 30% Suitable

PPE 20 to 30% Suitable

PPE + S/B 20% Suitable

PVDF none Suitable

PFA none Suitable

POM none Restricted suitable

PP 30% Restricted suitable

POM cop. none unsuitable

PES none unsuitable

PSU none/30% unsuitable

Figure 2. Stress-strain-curves of tin/bismuth.

strain (%)st

ress

(N

/mm

)2

Advances in Fusible Core Technique 283

with

[2]

The index "p" stands for "polymer" where as the index "c" stands for "core". The con-tact temperature contact depends on the temperatures of the polymer melt p and the temper-ature of the core c which is approximately the temperature of the cavity. Moreover, thecontact temperature is influenced by the properties of the materials, thermal conductivity, ,thermal diffusivity, a, density, ρ, and the specific heat capacity cp.

However, equitation [1] takes not into account the global increase of temperature of thecore during injection molding. This effect can be theoretically detected by calculating athermal balance across the core. A characteristic parameter, the melting index (MI), can bedefined for the risk of melting:

[3]

It is a function of the characteristic parameters of the materials used, and the processparameters. The specific enthalpy difference hp of the polymer from melt to solid state isdependent on both, material properties and process parameters. The temperature of theinjection molding cavity can be regarded as the starting temperature, 0, the melting pointof the core material as m,c.

As the melting index increases, the corethickness must increase in relation to the wallthickness of the plastic part to prevent impermis-sible heating of the core. Figure 3 shows the crit-ical and non-critical geometrical proportionscalculated from the thermal balance for a part(cylindrical core) produced by fusible core tech-nique. Since this is a simplified but practicableapproximation, the diagram shows a relativelywide scattering range, for which the theory doesnot give a clear prediction.1

Practical tests at the University of Essen, in which the temperature profiles in cores fordifferent geometrical proportions and process parameters were recorded show, however,that the "mixed range" can be regarded as largely non-critical as regards core melting.

COSTS

Fusible core technology is generally in competition with manufacturing processes in whichparts injection molded without undercuts are subsequently joined by welding, snap connec-

bi λiρicp i,=

ϑ ϑϑ

λ

MIρp

ρc-----=

∆hp

cp c, ϑm c, ϑ0–( )---------------------------------------

∆

ϑϑ

Figure 3. Number of non-plastic elements for different distribution rules.

non-critical range

mixed range

critical range

Melting Indax (MI)

d co

re d

iam

eter

D o

uter

dia

met

er


tion, screwing or overmolding. If parts can still meetspecifications, production costs for parts produced bymulticomponent technology are often expected to belower.

Parts produced by fusible core technology are char-acterized by high technical requirements, which cannot beachieved in the same way by multicomponent technology.Prevalent reasons for manufacturing plastic parts by fus-ible core technology are high demands on the accuracy ofthe inner geometries and the lifetime of the part (hightemperatures and media). Moreover, fusible core technol-ogy provides the possibility to integrate functions [2].

Typical parts which such requirements are shown inFigure 4 and Figure 5. The intake manifolds shown in Fig-ure 4 are parts that can be produced in large quantities onhighly automated plants. These production facilitiesrequire investments of the order of US $3 to 5 million.Figure 5 shows pump housings that are manufactured insmall to medium production runs. They require smallerand more flexible production plants, which usually have alower degree of automation. These plants are not designedfor a specific part, but can be used for different parts.Investment strongly depends on the degree of automationand is only about US $0.5 to 1 million, making them eco-

nomic.

DEVELOPMENT POTENTIAL

In the plant conception, it must be taken into account that the cycle time for core productionis often significantly longer than the cycle time for injection molding. This is a factor thatsignificantly impairs the economy of the process chain. Usually twice as many productionunits must therefore be purchased for core casting than for injection molding. Current R+Dwork is concentrated on reducing the cycle time for core casting. In particular, researchesare investigating the use of alternative mold materials and material coatings for core castingmolds. Basically, the use of materials with high thermal conductivity reduces cycle times.

However, many such materials are corroded by the molten tin/bismuth alloy, or them-selves change the properties of the molten metal. Furthermore, materials with high thermalconductivity lead to low contact temperatures in the core casting process, as Figure 7 shows.

Figure 4. Intake manifolds (Photos: MANN+HUMMEL and SIEMENS AT).

Figure 5. Hot water applications.

Advances in Fusible Core Technique 285

The contact temperature on first contact of the molten metalwith the mold has, in turn, a significant effect on the surfacequality.

Therefore, at the University of Essen an analyticalinstrument has been developed for investigation the compati-bility of mold materials and material coatings that can beused (Figure 6). The first investigations show that materialssuch as copper, brass and aluminium are attacked after evencomparatively short contact with the molten metal. With theuse of aluminium, there is the added problem that the molten

tin/bismuth alloy undergoes a significant viscos-ity increase and eventually becomes unusable.Anodized aluminium on the other hand has amuch higher resistance. This indicates that itmay be a suitable alternative to tool steel. Theanodized layer also leads to higher contact tem-peratures. Practical tests with molds made fromanodized aluminium show that better surfacequality of the cores can be reached compared tosteel molds. Ongoing research work at the IKMis investigating this mold material in detail.

Different mold coatings are also beinginvestigated. If a surface coating is found with a sufficiently high lifetime and significantlylower thermal diffusivity a further cycle time reduce with a simultaneous improvement insurface quality can be achieved. Figure 7 shows that, for example, a PTFE coating alwaysleads to contact temperatures above the melting point of tin/bismuth. As tests show, thisleads to a higher freedom of shape of the core, since flow turbulences during filling do notLQHYLWDEO\ lead to surface defects. Moreover, a thin coating does not affect cycle time. How-ever, pure PTFE coatings only have a poor lifetime.3

A wealth of experience is available in the processing of PA66 by fusible core technol-ogy. For almost all other materials, there is a question mark over their suitability for thisspecial process. Extensive investigations are therefore underway into the suitability of alter-native polymer materials.

CONCLUSION

Although fusible core technology is often regarded sceptically as regards production costs,technically demanding parts can be manufactured economically. The exploitation of the

Figure 6. Testing of mold materials.

Figure 7. Contact temperatures of mold materials.

rotating shaftof tested material

tin/bismuth

heated cylindre

PTFEaluminum, anodizedsteelaluminumco

ntac

t tem

pera

ture

( c

)o

melting point of tin/bismuth140160

120100

80

604020

20 40 60 80 1000

10 30 50 70 90


potential available with mold technology and polymeric materials will in future make fus-ible core technology suitable for further products and product groups.

REFERENCES1 Polifke, M.: Ph.D. Thesis, University of Essen, Germany, 19992 Schmachtenberg, E., Polifke, M.: From an Intake Manifold to Pump Casing, Kunststoffe (plast europe) 86 (1996) 3,

p 16-173 Schmachtenberg, E., Schröder, O.: Fusible Core Technology, Kunststoffe (plast europe) 89 (1999) 9, p 36-38.

Processing Glass-filled Polyethyleneon a Twin-screw Injection Molding Extruder

David Bigio, Rajath Mudalamane, Yue HuangUniversity of Maryland

Saeid ZerafatiElf Atochem

BACKGROUND

Glass fibers are a common filler added to polymers in order to improve mechanical proper-ties of the raw polymer, especially the impact strength and stiffness. The traditional route toproducing fiber reinforced parts involves blending the fibers into raw polymer in a Twin-Screw Extruder followed by pelletization. The pellets are then molded using an InjectionMolding Machine to form the final parts. All these steps cause fiber attrition. The focus ofthese experiments is to determine final glass fiber lengths under different operating condi-tions and compare these lengths to results reported by other researchers who have used aconventional route. One such paper is by Averous et al.10 They blended 4.5 mm glass fiberstrands into polypropylene using a twin screw extruder and then injection molded tensiletesting bars. They then conducted fiber length and distribution studies on the parts. Theyreport final glass fiber lengths of 0.47 mm (number average) and 0.7 mm (weight average).

In their recent paper, Fu et al.1 present a review of literature on the effect of processingon fiber length retention. The important conclusions they make are: 1) most of the fiberdamage occurs in the injection molding machine, rather than in the compounder, 2) lowscrew speeds and relatively high barrel temperatures minimized fiber breakage, 3) lowerback pressures and generous gate and runner dimensions are recommended to preserve thefiber aspect ratio during molding, 4) back pressure has a more dramatic effect upon the fiberlength than does the injection speed. This is understandable because the pellets containingglass fibers have to go through the melting section in the molding machine, which is aregion of high shear and high frictional forces.

THE TIME

The Twin-screw Injection Molding Extruder (TIME) is a novel injection molding machinethat is capable of both blending and extrusion in one step. Because it is a one step process,


the fibers never go through the entire extrusionprocess as well as the pelletization which limitsthe fiber size and nor the melting section in theTIME, but are blended into the molten polymerbefore injection. This machine is explained indetail in the following sections. The screw partof this machine is based on a Non-Intermeshing,counter-rotating Twin-Screw Extruder (NITSE).Figure 1 shows a schematic diagram of thismachine. This machine differs from a NITSE in

that one of the screws is capable of axial movement and has a non-return valve on the end.This enables the screw to inject and mold parts.

BACKGROUND OF PROCESSING ON THE NITSE

Nichols and Stellar3 compounded 1/8" glass fibers into nylon 6 on a NITSE. They chosethat device because it can be operated at low shear rates, even at high screw speeds. Theyalso expected higher glass fiber aspect ratios due to the higher free volume of the tangentialscrews. They found that tensile strength, flexural modulus, impact strength were weaklyaffected by the operating conditions but varied linear with increasing fiber content whichranged from 10% to 40 wt.%. Distributive mixing refers to the process of blending materialswhen there is no resistance to the process (e.g., paints) whereas dispersive mixing is neededwhen there is resistance in the nature of interfacial tension (immiscible polymers) or solidsagglomerates (e.g., carbon black). The mixing of glass fibers needs to overcome any possi-ble agglomeration, without breaking the glass, and then distribute throughout the polymerevenly. The NITSE has been shown to have superior distributive mixing characteristics tothe SSE4 and Co-rotating TSE5 especially when the flights of the screws are set at a staggerof 50%.6,7 The screw to.screw fluid transfer, with the resultant reorientation of the flow,which is important for efficient mixing, has been shown to be due to a local pressure gradi-ent across the nip region.8 Dispersive mixing is accomplished through cylinders or reverseflight elements.8,9 The advantage of this design is that a high shear stress can be appliedevenly to the flow. This condition is optimal for redistributing the glass fibers throughoutthe material without high stresses that could break glass fibers.

DESCRIPTION OF THE TIME

A single screw, 50-ton (clamp force), all electric IMM built by Cincinnati Milacron wasused as the base machine. The barrel and screw were removed from the machine and a newtransmission was fabricated and installed in their place. A long drive shaft connects the

Figure 1. Schematic diagram of the TIME.

Material flowdirection

Feed 2 Meltseals

Feed 1

Compoundingcylinders

(a)

(b)

Processing Glass-filled Polyethylene 289

injection screw or main screw to the injectiondrive unit of the molding machine. This allowsthe injection screw to move axially to injectmaterial. An auxiliary shaft can also be drivenby the transmission, thus converting it to a twinscrew machine. The auxiliary screw drive wasslaved to the main screw drive so that it turns theauxiliary screw at the same speed as the mainscrew whenever the extruder/plasticizer axis isactivated. Thus, the TIME operates essentiallyas a single screw machine during injection andpack, when the screws don’t turn. During theretraction step, both screws turn to melt andblend material. The auxiliary shaft cannot move

axially but can rotate to perform blending duties in a NITSE mode. The most important fac-tors that influenced the design of this transmission were 1) the close spacing between thetwo screws, 2) one screw needs to move axially with respect to the other.

A barrel is constructed by assembling barrel sections, supplied by the NFM-WeldingEngineers and the transmission, supplied by Spirex Corp. These sections are NITSE stan-dard 0.8”(20.32 mm), which are mounted on to the face of the transmission and securedwith tie rods. The screw sections are made by NFM-Welding Engineers and their outerdiameter is 0.8”(20.32 mm). The last barrel section is single screw and the auxiliary screwstops before this section. The main screw continues into the last barrel section. The pool ofmolten material for injection is formed in this section. A non-return valve on the end of themain screw prevents material from flowing backwards into the twin-screw region (see Fig-ure (2)). It also builds up pressure that pushes the injection screw back to obtain the requiredshot size.

EXPERIMENTS

MATERIALS

The raw fibers have a mean length of 5mm. The polyethylene used in the tests was an injec-tion molding grade Alathon® made by DuPont.

PART PRODUCTION

Polymer was fed using a loss-in-weight feeder. This resin goes through a melting sectionwith tapering screws and compounding cylinders (restrictive elements) for pressure genera-tion. Glass fibers are then added to the molten polymer through a downstream feed port.

Figure 2. Drawing of the last two barrel sections showing the transition from twin to single screw (not to scale).


Since glass-fibers require relatively gently mixing, no special mixing, elements are used.The gently mixing screw-to-screw transfer flow accomplishes the required dispersion ofglass fibers. The barrel temperatures are set at 320ºF for all zones. Dumbbells shaped tensiletesting specimens are molded under different blending conditions. A range of glass fibercontents from 10% to 40% was tested and twenty parts were made at each fiber content. Arange of screw speeds (during screw retraction) from 100 to 250 rpm was also tested at eachfiber composition. The changing screw speed at the same feed rates results in a differentamount of screw fill.

SCREW DESIGN

Figure (3) shows the screw design used for this process. Since there are no special kneadingelements that are used in a NITSE, a simple plot of screw root diameter v/s length providesample information about the screw design. Compounding cylinders are used as restrictiveelements and are labeled.

The melting section consists of a gradually tapering screw that pressurizes material topush it over the cylinders. The melting section is followed by the second feed region wheredeep screws (11.48 mm R.D.) are used to accommodate the addition of low-bulk density

Figure 3. Photograph showing fully assembled screws mounted on to the transmission.


glass-fiber (~0.45 g/cc). As it was mentioned before, the mixing flow caused by screw-to-screw transfer is used to blend the fibers into the polymer. The final section has shallowconveying elements (17.37 mm R.D.) which generate enough pressure to push materialthrough the non-return valve. It also serves to seal the material in the melt pool and preventit from flowing up through the second feed port.

Figure 4. The screw design used for these tests.

Figure 5. Image of a fiber bundle prior to processing. This bundle is 5 mm long.

Figure 6. Image of the fibers after blending into polyeth-ylene and separation by pyrolysis.

Main ScrewCompounding

cylinders

PressurePorts

Feed Port Feed PortAuxiliary Screw

Barrel IV Barrel Barrel II Barrel I

X-Axis: Screw Length (inches)Y-Axis: Root diameter (inches)

18 16 14 12 10 8 6 4 2 00

1

2

3

4


FIBER LENGTH DETERMINATION

The polypropylene was burned off in an oven in two hours at 900ºF. The glass fibers leftbehind were spread on microscope slides and studied under an optical microscope. Imagesof the fibers were captured using a CCD camera mounted on the microscope and processedin an image analysis program to determine fiber lengths. A population of about 300 fiberswas included in the average for each measurement.

Examination of the samples shows that the chosen screw design is successful in incor-porating and distributing the glass fibers into the matrix. This was accomplished in a verymild screw design which imparts little stress. The mixing was due to the transfer from oneto another in the apex region. Results from preliminary studies of glass fiber lengths are pre-sented at this time. Preliminary tests have shown average final fiber size of 0.814 mm. Thisis a 30% improvement over the lengths observed by other researchers. The lengths wereshorter than expected. This is attributed to the fact that a standard non-return valve was usedin the machine. This has a restricted flow path and can cause fiber attrition. For future tests,an improvised screw tip, similar to a smear-tip will be used.

CONCLUSIONS

This paper has demonstrated that it is capable of processing and injecting glass fibers in thesingle step process on the TIME. Glass fiber length improvement of 30% compared toreported lengths has been realized in initial tests. Future tests will incorporate a new designof the non-return valve to remove the limiting flow path.

ACKNOWLEDGEMENTS

The authors express their gratitude to the Spirex Co. for their continued support and speedyhandling of the fabrication process and repairs. Thanks are also due to the Adell PlasticsInc., for generously supplying the materials for testing. The invaluable assistance of MarcelMeissel, Detlef Knoeller and Mauricio Justiniano is also appreciated.

BIBLIOGRAPHY1 Fu S. Y., Lauke, B., 0äder E., Hu. X., Yue, C.Y., “Fracture resistance of short-glass-fiber-reinforced and short-carbon-

fiber-reinforced polypropylene under Charpy impact load an its dependence on processing”, J. Mat. Processing Tech., 89-90 (1999), p501-507.2 Mudalamane, R., Bigio, D.I., Tomayko, D.C., Zerafati, S., ‘Development of a Twin-screw Injection Molding Extruder’, ANTEC ’99.3 Nichols, R. and Stellar M., Plastics Compounding, Vol. 9, No. 4,(1986)4 Howland, C. and L. Erwin, SPE ANTEC, 113-115, 1982.5 Bigio, D. I., K. Cassidy, M. DeLappa, and Baim, W., ‘Starve-Fed Flow in Co-Rotating Twin Screw Extruders’,

International Polymer Processing Journal, Vol. VII , 2, p. 111-116 June, 1992.6 Conners, M., BS Thesis, MIT, 1985


7 Bigio, D.I. and Baim W., ‘A Study of the Mixing Abilities of the Counter-rotating, Non-intermeshing Twin Screw Extruder Using a Newtonian Fluid’, Adv. in Poly. Tech., Vol. 11, No. 1, p. 135-141, Feb, 19928 Bigio, D., M. Ramanthan, and P. Herman, Adv. in Poly. Tech., Vol. 12, No 4, 353-360 (1993)9 Hagberg, C., M. Shah, and D. Bigio, SPE-ANTEC, (1995)10 Avérous, L., Quantin, J-C. and Crespy, A., ‘Determination of the microtexture of reinforced thermoplastics by image analysis’, Composites Sci. and Tech., 58(1998), p. 377-387.

Injection Molding by Direct Compounding

Bernd KlotzKrauss-Maffei Kunststofftechnik, Germany

INTRODUCTION

Direct compounding has long since been established in sheet-, profile- and pipe-extrusion,where the high cost-advantages of single-stage article production are appreciated.5 Com-pared to extrusion applications, direct compounding of injection molded articles is compar-atively unknown territory.

The IMC-process (Injection Molding Compounding) enables filled or reinforced plas-tics to be direct-compounded immediately before injection molding. This offers two advan-tages to the molder: his material costs are reduced and he gains in flexibility.

SAVINGS POTENTIAL

The economic success of injection molding to the greater part relies on this process produc-ing fully finished articles during each cycle, employing gas-injection technology, multi-component injection molding, the lost core technique1 or the back-injection method,amongst others.

All these processes have in common, that they become effective, once the melt hasbeen compounded. Cost-savings result especially with the production of fully finished and/or assembled articles. Because expenditure for downstream finishing stages is either dis-pensed with or reduced considerably.

Contrary to downstream finishing, rationalization of material costs has hardly beenattempted, if pigmentation at the machine is ignored. But that example in particular shows,that molders would be able to save the greater part of the added costs for colored granulate,and would in addition become decidedly more flexible, where store keeping, color meteringas well as shade of color are concerned.

These arguments apply to a much greater degree to the compounding of filler and rein-forcement materials. The savings potential becomes obvious, if one considers, that thematerial costs represent roughly 80% of the production costs for consumer articles. Evenwith technical components, they still amount to about 50%.


Vast amounts of chalk- or talcum-filled polypropylene for instance are molded into themost varied products. Whereas an injection molding screw disperses pigments added asmasterbatch very well during granulate plasticizing, a twin-screw compounding extruder isrequired for producing blends, due to the larger amounts of fillers added in powder-form.3,4

These extruders, that have proved themselves superbly with the continuous production ofblends, are poorly suited however for discontinuous operation, characteristic for the injec-tion molding process.

KEY-FUNCTION OF THE DOUBLE-SHOT POT INJECTION SYSTEM

With the IMC-process, a double-shot pot is the link between the continuous compoundingprocess and the discontinuous injection molding side. This allows the high melt quality ofcompounding to be directly employed for injection molding. As subsequently explained indetail, this novel combination of compounding extruder, double shot pot and conventionalclamping unit, allows• filled plastics materials to be compounded directly on the modified injection molding

machine,• the compounding extruder to be run continuously and yet to• achieve for the injection molding machine as a whole the machine performance custom-

ary for injection molding, even if operation is interrupted. It is an additional advantage of this machine design, that due to the separation of the

twin-screw’s melt compounding from the injection molding sequence, the two units can beoptimized independently of each other.

CONTROL CONCEPT

Parameters are set through the machine control system. All the usual compounding-specificsetting parameters for the twin-screw extruder are to be found in the control system’sparameter list, so that the compounding expert is unrestricted as well, when optimizing.

Settings of the dicontinuous molding process - injection, holding pressure and cooling,up to the demolding stage – as well as for the mold movement on the injection moldingmachine, are unrestricted, compared to the standard machine, with regard to the kind as wellas the variety of the setting possibilities. Parameters of the “double-shot pot” injection unitare set on the two-component machine’s control system.

The parameter governing the metering stroke of the double-shot pot’s specific fillingweight, in conjunction with the actual cycle-time, is converted directly into an output capac-ity m° (kg/h), that has to be achieved. Conventional injection molding machine-setting phi-losophies thus remain intact for the machine setter.

Injection Molding by Direct Compounding 297

THE “IMC”-RANGE OF MACHINES

Such machine-modules as gravimetric metering units, twin-screw extruders, double-shotpots can in almost every case be adapted to virtually every clamping unit by simple designmodifications.

“IMC”INJECTION- AND PLASTICIZING UNIT

The “IMC” injection- and plasticizing unit is a modular combination of twin-screw extruderand double-shot pot, that have been matched to each other with regard to achievable shotweights. There is a choice of 4 units, that are listed in the Table 1 below.

AREAS OF APPLICATION

The advantages of the IMC-machine range come into their own during long-term produc-tion runs, typical for garden-furniture- or carbody-components, where materials or moldsare seldomly changed.

Thus, for instance, a manufacturer of leisure furniture processes several thousand tonsof chalk-filled PP annually into garden chairs and tables. That was the reason, why the firstprototype application was designed for the production of armrests for garden chairs, at ashot weight of about 1000 grams.

FUNCTIONS SEQUENCE

Based on the example of the prototype application, the subsequent text describes the flow-path of the basic material components, starting in the gravimetric metering unit and finish-ing as the completed article.

The conventional single-screw plasticizing unit has been replaced with a close-mesh-ing, co-rotating twin-screw compounding extruder (40Ax36D). It produces the compoundfor the armrests from polypropylene, pigmented masterbatch and chalk, delivering the mix

Table 1

SP3500/40 IMC SP5700/50 IMC SP14700/60 IMC SP19000/60 IMC

max. swept volume, ccm

1953 3611 7632 11133

maximum injection pressure, bar

1742 1580 1931 1707


as a homogenous melt. The extruder is equipped with degassing, in order to dispense withhaving to pre-dry the PP-granulates, as well as the chalk.

A gravimetric mixing- and metering unit transfers the pigmented masterbatch-enrichedPP-granulate to the extruder’s feed-throat opening. So that the feed-throat is not subjected tohard wear, and in order to achieve good distribution without any agglomerate forming, thechalk is added only, when the granulate has already melted. A feed position some way downthe side of the extruder barrel serves that purpose. Chalk is fed into the melt flow by a sec-ond gravimetric metering unit, and a twin-screw conveying unit (model ZSFE 34). Due tothe dispersing effect of the co-rotating twin-screws, the chalk is broken down very finelyand uniformly distributed, so that a homogeneously melted compound results.

The compounded melt is metered continuously into one of the double-shot pot’s cylin-ders by the extruder, through a transfer-line and a two-way valve. The double-shot pot unithas been designed as plunger system, for efficient scavenging of the cylinders. Heaterbandskeep the melt in the transfer pipes, the valve and the injection cylinders at processing tem-perature. During metering to the first-in-first-out principle, the plunger moves back insidethe injection cylinder. Once the metering time has elapsed, the two-way valve changes overto the second cylinder without interruption, while the extruder keeps running, therebychanging the cylinder functions.

Immediately afterwards, the plunger of the freshly charged cylinder starts a productioncycle by injecting the melt. During that phase the plunger/cylinder unit operates in the sameway as the injection unit of a plunger-type injection molding machine; this is followed bythe process stages of holding pressure, cooling, article demolding and nozzle contact for thenext injection cycle. So that the compounding extruder continues reliably in continuousoperation, the metering time is set, so that it is only fractionally longer than the injectionmolding cycle time. Hydraulic drives have been installed for generating the metering strokeor the injection pressure, aided by the plungers.

The injection molding compounder operates in a very material-protective manner, dueto• the single-step working operation, which dispenses with the additional thermal stress

encountered with separate compounding,• the twin-screw extruder having a very tight residence-time profile, in contrast to the tri-

ple-zone injection molding screw,• no locally high shear-forces occurring either in the extruder or in the buffer store, so that

there is no risk of material overheating and• the melt-conducting channels and buffer stores having been designed according to rheo-

logical aspects.

Injection Molding by Direct Compounding 299

PRODUCING WITH IMC

When starting-up the plant from cold, the procedure is similar to that of running-up an injec-tion molding machine. First of all, the extruder, the double-shot pot and the connectingpipes are brought up to temperature. Then gravimetric metering opens the supply of thegranulate components, and plasticizing starts.

The initially produced melt is discharged through a special valve, installed downstreamof the extruder. As soon as stable operating conditions have been established in the extruder,the start-up valve changes over to its home-position, the melt charges one of the injectioncylinders, and the earlier mentioned production sequence starts. The pre-heated injectionmolding compounder can be started-up almost as quickly, as an injection molding machineof identical size.

Under normal running conditions, i. e. during trouble-free production of molded arti-cles, the two cylinders operate in synchronization with the cycle time, either as melt bufferor as injection unit, as described. If operation had to be interrupted for a brief period, one ofthe cylinders will additionally have been charged with the material volume of the secondcylinder’s storage capacity, once metering is completed.

It is possible to lower the extruder’s output rate to a material-specific minimum byclosed-loop control during such an interruption of operation. To that end, the control systemthrottles down the material supply and the screw RPM, so that the filling rate of the extruder- and thus of the melt quality – remains constant.

Only when an interruption takes longer than 30 minutes will the plant have to be shutdown. When processing PP, it is not necessary to lower the operating temperature. Once thefault has been rectified, the plant is re-started within a few minutes, as described above.

By and large, the injection molding compounder therefore shows a behavior in all theimportant operating conditions - like start-up, normal operation, malfunction -that comesclose to that of the injection molding machine’s performance during production runs.

PROFITABILITY, FLEXIBILITY

As mentioned at the beginning, profitability of direct compounding arises from the achiev-able savings in material costs. That potential must therefore be determined specifically foreach application. In the case of chalk-filled PP, a compound containing chalk at 30 percentby weight, available in the trade, presently costs about 1.80 DM/kg in white, and coloredblue about 3.40 DM/kg. Based on prices of 1.35 DM/kg for unfilled PP and 0.40 DM/kg forchalk, the costs for material produced on the injection molding compounder amount to

0.7 kg PP × 1.35 DM/kg = 0.95 DM0.3 kg chalk × 0.23 DM/kg = 0.07 DM1 kg of compound 1.02 DM


That represents a cost difference across the board.from 0.78 DM/kg for white materialto 2.38 DM/kg for colored compounds, (without even taking the costs for the color-master-batch into consideration). These amounts are available for amortization of the machine’sadditional equipment as well as for possibly covering any other additional outlay, such asquality assurance. For typical products, an amortization period of from less than 1 and 3years results, which is clearly reduced with rising annual material costs.

In addition to the cost advantages, the molder gains flexibility with the purchasing andstockpiling of materials, as well as the determining of amounts of filler additives and rein-forcing media. Indirect cost advantages result, when the cycle time is reduced, due to ahigher content of filler (the cooling time gets shorter, because the lower proportion of PPmeans, that there is less to be dissipated from the polypropylene melt), so that productivityincreases.

OUTLOOK

During the production of medium-sized to larger moldings from modified plastics materials,the injection molding compounder opens up savings opportunities, which have hithertobeen impossible to exploit. This for instance applies to filled materials, or those containingflame-retardant additives, as employed by the electrical appliances industry. Even largemoldings consisting of different compounds (e. g. glass-fibre reinforced plastics) such asbumpers, fascias, internal door coverings, rear-trunk claddings, can be produced by thismethod.

Ultimately the injection molding compounder will be able to open up yet another areaof application: the gentle, material-protective compounding of difficult to process, highquality plastics of narrow processing lee-ways. During the melting process of such materi-als, single screw plasticizing units can reach their limits - amongst other things, due to theirbroad residence time profile. The injection molding compounder on the other hand is capa-ble of homogeneously melting these plastics materials, custom-made as reactor-blend, forinstance. Transferred for injection to the double-shot pot at high plasticizing rates and with-out causing thermo-mechanical damage.

LITERATURE1 Rothe, J.: Sonderverfahren des Spritzgießens. Kunststoffe 87 (1997) 11, 1564-1582.2 Bürkle, E.; Rehm, G.; Eyerer, P.: Hinterspritzen und Hinterpressen – Lagebericht zur Niederdrucktechnik. Kunststoffe 86 (1996) 3, 298-307.3 Allen, P. S.; Bevis, M. J.; Hornby, P. R.: New direct compounding/injection unit for molding composites. Modern

Plastics International, April 1987, 38-39.4 DE 40 21 922 A 1. Putsch, P.: Kombiniertes Compoundier-Spritzgußverfahren und Vorrichtung zur Durchführung dieses Verfahrens. Offenlegungsschrift vom 16. 1. 92.5 Horst Kurrer: Gefüllte Polyolefine direkt extrudieren, Kunststoffe 83 (1993) 17-21.

Improvement of the Molded Part Quality:Optimization of the Plastification Unit

S. Boelinger, W. MichaeliInstitut für Kunststoffverarbeitung (IKV), Pontstr. 49, D-52062 Aachen, Germany

INTRODUCTION

Because of its varied area of responsibility the plastification unit is very important for theinjection molding process. The process steps carried out by the plastification unit are thefeeding of the material, the conveying of the material, the melting and the homogenizationof the melt. As consequence the following demands are made on the plastification unit:1-6

• the supply of the necessary mass flow with a high melting capacity,• a good energetic operation behavior,• a good feeding behavior and a constant convey of the material over the whole screw dis-

placement,• the achievement of a good thermal, mechanical and material melt homogeneity,• a good residence time of the material to reduce the thermal degradation,• a good reproducibility of the melting process and• a wide area of use.

Traditionally three-zone plastification screws, also called standard screws, were usedin the injection molding process. The injection molding materials and the machine technol-ogy are more and more adapted to the different process groups. Because of this fact standardscrews are limited frequently in their efficiency so that not all demands on the plastificationunit can be fulfilled.

One important industrial process group are fast running processes like the productionof packaging. The use of multi-cavity and multi-floor mould bases requires a high plasticis-ing output from the plastification unit. Furthermore new developments in the mould tech-nology lead to a quicker movement of the mould base during the injection cycle. Theoptimization of the control technology of the injection molding machines allow the produc-tion with a higher injection speed so that cycle time can be reduced, too. But the reductionof the cycle time can be limited by the melting capacity of the plastification unit, if the dos-ing is the dominant operation step of the injection molding process and if a high melt vol-


ume is needed. In this case the necessary amount and quality of the melt can often be notachieved using the established screw geometries.

In contrast to the fast running processes other product groups, like optical parts, requirean extremely good thermal homogeneity of the melt. Good quality parts can only be pro-duced using extremely slow injection speeds and a controlled temperature in the barrel.Injection molding materials used for the application of optical parts, like PC, PMMA, COCor transparent PA, react very sensitive against heat. Due to the long cycle times which arenormal during the production of optical parts the use of standard screws is limited often.

It is a problem for the design of plastification screws, that the target values “melt qual-ity” and “the minimization of the cycle time” behave in different ways. As design criteriathe application of the screw has to be taken into account. It must be the aim to reach an opti-mized area of both demands, cycle time and melt quality. Furthermore it should be possibleto use the screw design for a varied choice of injection molding materials.

BASIC INVESTIGATIONS OF DIFFERENT SCREW CONCEPTS

In cooperation with an injection molding machine manufacturer several screw geometrieswere tested with different injection molding materials.1 During the investigations the influ-ence of several dosing parameters on the melting capacity, the energy consumption of thescrew and the melt homogeneity were analyzed. Furthermore it was the aim to quantify themodification of the screw geometry which is normally based on internal company experi-ence.

The used screw geometries were a three zone screw (standard screw), a multi-threadedscrew and three barrier screws. Differences of the three barrier screw concepts (I-III) werethe thread gradient of the main and the barrier flight and the length of the screw zones. Asbasic screw geometries a diameter (D) of 60 mm and a length of 20 D were chosen. To ana-lyze the influence of shear and mixing elements the basic screws were lengthened with addi-tional elements to a new screw length of 25 D. For the investigations a defined volume of 10liters was injection molded with the materials PP and PE. Besides the parameters back pres-sure, rotational speed and melt temperature the residence time of the material in the barrelwas varied. Therefore the dosing time and the cycle time were changed in combination. Inthis first investigations the influence of the parameters was analyzed separately. But theinteractions of the parameters were taken into account qualitatively at the interpretation ofthe results.

Some examples of the results are explained below. The main emphasis is put on theanalysis of the plasticising output (melting capacity) of the plastification unit in correlationwith the varied parameters. Figure 1 shows the correlation between the plasticising output

Improvement of the Molded Part Quality 303

and the dosing position (scaled as multiple of thescrew diameter D) for the screws with the length25D and the material PP.

For the use of the barrier screw concepts itwas found that a degressive profile of the barreltemperatures is good for processing. Because ofthis fact only the results for a degressive barreltemperature profile from the feeding zone to thenozzle (250°C-230°C) are shown. For the use ofthe barrier screw concepts the plasticising outputwas reduced with increasing dosing volumes.

Furthermore differences in the plasticising output can be noticed for the different barrierscrew geometries. The use of the barrier concept I leads to a small drop of the plasticisingoutput whereas the barrier screw concept II reduces the plasticising output about 15%. Incomparison to the barrier screw concepts the investigated standard screw and the multi-threaded screw only show slight changes in the plasticising output.

In a second step the influence of the residence time and as consequence the influenceof the heat conduction was investigated. Figure 3 shows the plasticising output as functionof the residences time for PP and the 25D screws. Here could be noticed that for short resi-dence times the plasticising outputs are close together. But with growing residence timesthere are differences in the achieved plasticising outputs. For the standard and the multi-threaded screw the plasticising output is not much influenced by the residence time. In con-trast to this fact the plasticising output increases with longer residence times using the bar-rier screw concepts. Again there is a difference between the different types of barrierscrews. The increase in the plasticising output comes to 20% for the barrier screw concept I.

Figure 1. Plasticizing output as function of the dosing position (D = 60 mm; L = 25 D; material PP).

Figure 2. Plasticizing output as function of the residence time (D = 60 mm; L = 25 D; material PP).

Figure 3. Plasticising output as function of the residence time (D = 60 mm; L = 25 D; material PE) barrier screw III.


I

jII

""""""~..,. ., ~ ..,-, -

100 -..,I-'.--P'oI'«2~~~.l4~:..:CI-.

W

I 80

70

i :

0

30 20

,

10

-w O

0 20 0 00 80 100 120 10160 160 200 220

R.aide- 1imO 1&1 ~~(m'SI -

Figure 5. Medium energy consumption as function of the

circumferential speed (in principle).Figure 4. Plastizising oputput as function of the residence

time (D = 60 mm; L = 20 D; material PE).

The barrier screw concept II even comes to an increase in the plasticising output of 40%. Asconclusion with the use of barrier screws an improvement of the plasticising output about56% (concept II) referring to the plasticising output of the standard screw can be achieved.

The characteristic curves of the plasticising output in dependency on the residence timefor PE and the 25 D screws are shown in Figure 3. In principle the same correlations can benoticed as they are shown above for the material PP. In comparison to the standard screw theplasticising output could be increased about 78% using the barrier concept II. All in all theachieved plasticising outputs are higher for the injection molding material PE than for PP.Reasons for this fact are the different material properties, like the material density and the

melting enthalpy.The results for the plastification screws without shear and mixing elements (20D

screws) are shown in Figure 4. Here again the plasticising output is given as function of theresidence time for the use of the material PP. Again the courses are similar to the courses of

the 25D screws as they are shown on Figure 3.Another important aspect for the evaluation of screw concepts is the medium energy

consumption during the dosing. Figure 5 shows the medium power consumption as functionof the circumferential speed. The graph only shows the courses for the different screw con-cepts in principle. With increasing circumferential speed a higher energy consumption isrequired. Though all courses which are shown on Figure 5 are close to each other there is atendency of a lower energy consumption for the use of barrier screws.

FURTHER INVESTIGATIONS BASED ON A MODULAR DESIGN OF THEPLASTIFICATION SCREW

The basic investigations have revealed that changes in the screw geometry have a greatinfluence on the plasticising output. But due to the high costs of the screw manufacturing

Improvement of the Molded Part Quality 305

the number of screws used for the investigations is limited. One solution for this problem isthe design of a modular screw. Components of this modular screw are the load-bearing

shaft, also called mandrel, and the screwelements which are on the mandrel (Fig-ure 6). Important is the protection of thescrew elements against twisting. There-fore feathers, multi-wedge toothing or aprofile splinted shaft can be used. Addi-tionally the screw elements are fixedwith the screw tip. Due to the heat in thebarrel the screw elements and the man-drel might show differences in their

elongation. This effect is compensated using springs.7

This screw concept has been used in the area of extrusion, especially for compounding,for a long time. It is necessary for compounding to adapt the screw geometry to the extru-sion material. In this area threading elements with different geometries, shear and mixingelements are combined useful. Furthermore it is possible to vary the material of the screwelements, if abrasive materials are.used for extrusion. For example nitride steel can bereplaced by full-hardened tool steel.7

For the injection molding process the adaptation of the screw geometry to the processconditions is important, too. Especially for laboratory tests advantages can be gained usingmodular screws. This type of screws allows to change the screw elements purposefully tovary characteristics of the screw geometry, e.g. thread gradients or depths. Further advan-tages are the possibilities to change the length of the screw zones and to use hardened orcoated screw elements, if this is useful for the application.9-11 Injection molding materialstaken to produce high quality optical parts for example show the tendency to stick on thescrew materials normally used for standard applications. This stick effect could be reducedusing surface coated screw materials.

For the adaptation of the modular screw from the extrusion to the injection moldingprocess it is important to work out a new screw design. In contrast to the extrusion processthe plasticising is not a continuous process step and uneven load is put on the screw. Onetask of the injection screw is the dosing. During this process step the screw moves back-wards with additional rotation. Afterwards the screw is responsible for the injection processand it functions as a piston. The melt is injected into the mould base by an axial progressivemovement of the screw. For the final design of the modular screw the trigger of the chang-ing torque and the buckling behavior of the screw must be taken into account. Furthermore

Figure 6. Load-bearing shaft (mandrel) with fixed screw ele-ments.8

screw element

shaft nut load-bearing shaft (mandrel) feather


the screw elements are not allowed to be sheared off the mandrel during the injection mold-ing process.

CONCLUSIONS AND NEW GOALS

Because of the possibility to change the single screw elements purposely, the modular screwconcept is a good basis for the investigation of the plastification behavior of injection mold-ing machines in laboratory tests. For the experiments with the modular screw the statisticalmethod for experiment design will be used so that interactions of the parameters can betaken into account. As objective the plasticising output and the melt quality will be investi-gated in dependency on the process parameters and the screw geometry. In addition to thematerial homogeneity of the melt the axial and radial temperature distribution of the meltwill be evaluated.

ACKNOWLEDGEMENT

The presented basic investigations of different screw concepts were carried out in coopera-tion with the company Mannesmann Demag Ergotech GmbH, Schwaig, Germany.

REFERENCES1 Mehler, C., Untersuchung verschiedener Schnek-kengeometrien von Spritzgießmaschinen; not published diploma thesis at the IKV, Aachen, Germany (1998), Supervisor: S. Bölinger, S. Pahlke (Demag Ergotech GmbH, Schwaig).2 Mehler, C., Schimmel, D., Pahlke, S., Dreizonenschnecke oder Barriereschnecke für Polyolefine?, ERGOpress 2:1 10-13 (1999) 1.3 Bürkle, E., Qualitätssteigerung beim Spritzgießen als Aufgabe des Plastifiziersytems, Kunststoffe 78:4 289-295 (1989) 4.4 Bürkle, E., Bauer, M., Würtele, M., Spritzgießschnecken - Kompromisse definieren ihr Einsatzspektrum, Kunststoffe 87:10 1272-1286 (1997).5 Potente, H., Entwicklung bei der Auslegung von Plastifizierschnecken, Plastverarbeiter 43:11 118-127 (1992).6 Wortberg, J., Mahlke, M., Effen, N., Barriereschnecken steigern Homogenität der Schmelze, Kunststoffe 84:9 1131-1138 (1994).7 Uhland, E., Kunststoff-Extrusionstechnik I Maschinenbauliche Grundlagen – Doppel-schneckenextruder,

gleichläufig, Carl Hanser Verlag, München, Wien, 527-535 (1989).8 Meier, U., Kunststoff-Extrusionstechnik I Maschinenbauliche Grundlagen - Andere Extruder, Carl Hanser Verlag, München, Wien, 536-539 (1989).9 Peters, H., Mit flammgespritzten Schnecken gegen abrasiven Verschleiß, Kunststoff-Berater 12 891-892 (1973).10 Lülsdorf P., Verschleißschutz bei Spritzgieß-maschinen, Kunststoffe 86:6 776-782 (1996).11 N. N., Panzerschichten erhöhen die Lebensdauer von Schnecke und Zylinder, Kunststoff-Berater 12 887-889 (1973).

Non-return Valve with Distributive andDispersive Mixing Capability

Chris RauwendaalRauwendaal Extrusion Engineering, Inc., Los Altos Hills, California 94022, USA

INTRODUCTION

The screw of the plasticating unit of an injection molding machine (IMM) typically consistsof a single stage, single flighted conveying screw with a non-return valve at the end. Mixingsections are usually not incorporated into the screw design. One reason for this is the factthat most plasticating units a relatively short; the typical length-to-diameter ratio is 20:1 inIMMs. This does not leave much space to incorporate a mixing element. Another reasonmay be the mistaken believe that mixing is not very important in the injection molding pro-cess.

A convenient method to improve the mixing capability of the plasticating unit of anIMM is to design the non-return valve (NRV) such that it has mixing capability. Such adual-purpose NRV allows an increase in mixing capability without affecting the melting andconveying capability of the plasticating unit. This paper will describe a NRV mixer basedon the CRD mixing technology developed for single screw extruders.

BACKGROUND

In polymer mixing we usually distinguish between distributive and dispersive mixing. Dis-tributive mixing aims to improve the spatial distribution of the components without cohe-sive resistance playing a role; it is also called simple or extensive mixing. In dispersivemixing cohesive resistances have to be overcome to achieve finer levels of dispersion; dis-persive mixing is also called intensive mixing. The cohesive component can consist ofagglomerates where a certain minimum stress level is necessary to rupture the agglomerate.It can also be droplets where minimum stresses are required to overcome the interfacialstresses and deform the droplet to cause break-up.

Dispersive mixing is usually more difficult to achieve than distributive mixing. Singlescrew extruders are generally considered to be poor dispersive mixers while twin screwcompounding extruders have much better dispersive mixing capability. However, when weanalyze the mixing process in co-rotating twin screw extruders,1 it is clear that the main


mixing action does not occur in the intermesh-ing region but in the region between the pushingflight flank and the barrel. This is particularlytrue when the flight helix angle is large as it is inkneading disks. This being the case, there is noreason that the same mechanism cannot be usedin single screw extruders.

The reason that twin screw extruders makegood dispersive mixers is that the space between

the pushing flight flank and the barrel is wedge shaped and thus creates elongational flow asthe material is force through the flight clearance. Shear flow is not very efficient in achiev-ing dispersive mixing because particles in the fluid are not only sheared they are alsorotated, see Figure 1 top. In elongational flow particles undergo a stretching type of defor-mation without any rotation; rheologists call this “irrotational flow,” see Figure 1 bottom.

In the past it was considered to be difficult to generate elongational flow in mixingdevices. It turns out, however, that this is not the case. The new dispersive (CRD) mixersdeveloped by Chris Rauwendaal2-4 create elongational flow two ways, see Figure 2. Byusing a slanted pushing flight flank so that the material is stretched as it is forced throughthe flight clearance and by using tapered slots in the flights. The tapered slot accelerates thefluid as it flows through the slots and thus creates elongational deformation. The dashedarrows in Figure 2 indicate the screw velocity, the solid arrows indicate the melt velocityrelative to the screw.

The tapered slots in the flights serve to increase distributive mixing as well as disper-sive mixing. If the material is not randomized in its passage through the mixer, only theouter shells of the fluid will be dispersed leaving the inner shells undispersed.5 Therefore, itis critical to incorporate both distributive and dispersive mixing ability within the mixer.Figure 3 shows the CRD5 mixer with curved flight flanks and tapered slots in the flights.

Figure 2. Two methods of creating elongational flow in the CRD mixer.

Figure 1. Dispersion in shear and elongational flow.

barrel

Curved flight tank Tapered flight slot

Non-return Valve 309

The slots divide the flights into wiping.and mix-ing flight segments. Each wiping segment is fol-lowed by three mixing segments.

The initial design of the CRD mixer2 wasdeveloped using the concept of the passage dis-tribution function,6 while the final geometry wasdeveloped using a detailed three-dimensionalflow analysis. The 3D flow analysis was per-formed using BEM flow,7 a boundary elementflow analysis package originally developed atthe University of Wisconsin, Madison by Profes-sor Osswald and co-workers. The BEM analysisallows a complete description of flow, so that the

stresses, the number of passes over the mixingflights, the number of passes through the tapered slots, residence time, etc. can be quantifiedfor a large number of particles. The CRD mixer may well be the first complex mixingdevice developed solely based on engineering calculations and computer modeling.

APPLICATION TO INJECTION MOLDING

Mixing is not only important in extrusion, it is equally important in injection molding. CRDmixing elements can be added to an injection screw. Most injection screws have a non-return valve (NRV) at the end of the screw to prevent the molten plastic flowing back intothe screw during injection. It is possible to incorporate mixing capability into the NRV tocombine two functions within one device. Since the NRV is short, there is little room avail-able to incorporate mixing capability into the NRV. The slide ring NRV is the most suitablewith respect to adding mixing capability.

The action of the ring check valve isillustrated in Figure 4. When the screwrotates, plastic is conveyed forward,melted, and mixed. The plastic meltaccumulates at the discharge end of thescrew and pushes the screw back againsta controlled pressure. As the screwmoves backward, the check ring is

dragged to the most forward position against a stop at the end of the screw, see Figure 4a.The stop is usually a star shaped shoulder with four to six points. When the check ring restsagainst the stop plastic melt can flow through the valve.

Figure 3. Three dimensional rendering of a CRD5 mixer.

Figure 4. The ring valve in open (a) and closed (b) position.


When the screw moves forward,the check ring is dragged to the mostrearward position against the check ringseat forming a seal, see Figure 4b. Inthis position the valve is closed and theplastic melt is thus prevented from leak-ing back into the screw channel duringinjection. Because of the relative move-ment between the check ring and thestop, stops will wear over time andeventually have to be replaced.

Mixing capability can be designedinto the slide ring valve by locating mix-ing pins on the inside of the slide ring asshown in Figure 5. The pins are elon-gated in the axial direction to achieve anacceleration of the fluid as it is passingbetween two pins. The resulting elonga-

tional flow results in effective dispersive mixing. The same elongated pins are also locatedon the outside of the stop, resulting in a second exposure to elongational flow with furtherdispersive mixing action. The large number of pins located on the slide ring and stop inducea large number of splitting and reorientation events, resulting in efficient distributive mixingaction.

The CRD NRV provides a convenient and cost efficient method to improve the mixingcapability of injection molding screws. Good mixing action can be incorporated simply byexchanging the conventional NRV with a CRD NRV. Figure 6 left shows a solid model ofthe CRD valve with the slide in the forward (open) position. Figure 6 right shows the slidering with the internal mixing pins. During screw rotation the slide ring will rotate with thescrew but at a lower rotational speed. This results in a mixing action somewhat similar tothe Twente Mixing Ring.10

EXPERIENCE FROM THE FIELD

As of November 1999 over forty extruders are running with CRD mixers. The current appli-cations are color concentrates, foamed plastic, post-consumer-reclaim with calcium carbon-ate, medical applications, heat shrinkable tubing, carbon black dispersion in polyolefins,and blown film extrusion of low melt index metallocenes. All current applications are onsingle screw extruders in sizes ranging from 19 to 200 mm (0.75 to 8.0 inch). Three screws

Figure 5. CRD non-return valve for injection molding screws (open position).

Figure 6. Solid model of the CRD non-return valve (left) with detail of just the slide ring (right).

Nose Slide ring Shoulder

Threaded stud

Slide ring detail

Non-return Valve 311

are being manufactured for injection molding and one set of screws for twin screw extru-sion.

In the manufacture of color concentrates it was possible to produce even the most diffi-cult material, phthalate blue, with good quality using a single screw extruder. The foamedplastic application polystyrene was foamed with a physical blowing agent carbon dioxide, ademanding application requiring very good dispersive and distributive mixing. The foamedproduct was found to have uniform cell size and excellent surface quality, while the extru-sion process was very stable with respect to melt pressure and temperature. In the blownfilm application major problems were poor film quality and the inability to feed more than3% fluff with the virgin plastic without generating gels in the film. The CRD mixing screwwas able to improve film quality and handle up to 12% fluff without generating gels.

The good results with respect to gels indicate that elongational mixing devices caneffectively disperse gels as opposed to shear mixing devices. Luciani and Utracki foundsimilar results in experiment using their extensional flow mixer.9 The experience from thefield was obtained with a fifth generation mixing device, the CRD5, see Figure 3. Thismixer has four parallel flights with tapered slots in the flights. Each wiping flight segment isfollowed by three mixing flight segments. Tapered slots separate the flight segments. Thewiping segments are offset such that the mixer wipes the entire barrel surface. Completewiping of the barrel surface is important to achieve efficient heat transfer between the plas-tic melt and the extruder barrel.

CONCLUSIONS

Mixing devices that split and reorient the fluid while generating strong elongational flowcan achieve both efficient distributive and dispersive mixing. CRD mixers have proven theireffectiveness in extrusion;2-4 however, their use is not limited to single screw extruders. Thesame mixing devices can be used in injection molding. Perhaps the most convenient methodto enhance the mixing capability in injection molding is to design mixing elements into thenon-return valve.

This paper describes a slide ring valve/mixer based on the patented mixing technol-ogy.8 The slide ring has multiple internal grooves that are tapered while the nose-piece hasexternal, tapered grooves. The tapered grooves are formed by elongational pins. The pinssplit and reorient the fluid thus causing efficient distributive mixing. The tapered groovesaccelerate the fluid and cause elongational flow with efficient dispersive mixing.

The high mixing efficiency of the CRD mixer is due to the generation of strong elonga-tional flow and the fact that all fluid elements make multiple passes through the high stressregions. Elongational flow not only achieves more effective dispersion; it also creates less


viscous dissipation than shear flow. As a result, the power consumption and temperature risein the CRD are less than in mixing devices that rely on shear flow.

The CRD valve/mixer is easy to manufacture and can be mounted on existing injectionscrews. Since the non-return valve on many injection screws is removable, it is very easy toreplace a conventional valve with a CRD valve/mixer. As a result, the change-over can bedone in a short period of time. Actual results of the NRV mixer obtained from injectionmolding trials will be presented at the ANTEC meeting as well as animations showing themovement of the various components of the mixer during the injection molding cycle.

REFERENCES1 Polymer Mixing, A Self-Study Guide, C. Rauwendaal, Carl Hanser Verlag, Munich (1998).2 “A New Dispersive Mixer for Single Screw Extruders,” 56 th SPE ANTEC, Atlanta, GA, 277- 283, Chris Rauwendaal, Tim Osswald, Paul Gramann, and Bruce Davis (1998).3 “Experimental Study of a New Dispersive Mixer,” Chris Rauwendaal, Tim Osswald, Paul Gramann, Bruce Davis,

Maria del Pilar Noriega, and O. A. Estrada, 57 th SPE ANTEC, New York, 167-176 (1999).4 “Design of Dispersive Mixing Sections,” Chris Rauwendaal, Tim Osswald, Paul Gramann, and Bruce Davis,

International Polymer Processing, Volume 13, pp. 28-34 (1999).5 “Kinematics and Deformation Characteristics as a Mixing Measure in the Screw Extrusion Process,” T.H. Kwon,

J.W. Joo, and S.J. Kim, Polym. Eng. Sci., V. 34, N. 3, 174-189 (1994).6 “The Distribution of Number of Passes Over the Flights in Single Screw Melt Extruders,” Z. Tadmor and

I. Manas-Zloczower, Advances in Polymer Technology, V. 3, No. 3, 213-221 (1983).7 BEMflow, Boundary Element Fluid and Heat Transfer Simulation Program, ©1996 The Madison Group: PPRC.8 C. Rauwendaal, U.S. Patent 5,932,159.9 “The Extensional Flow Mixer”, A. Luciani and L. A. Utracki, Intern. Polymer Processing, 9, 4, 299-309 (1996).10 A.J. Ingen-Housz and S.A. Norden, Intern. Polym. Process., 10, 120 (1995).

AABS 135acrylic 194aesthetic properties 245agglomerates 288amorphous 256anisotropic shrinkage 99antistatic 274appearance 134applications 15armored vehicle 209attrition 287automotive 93, 163, 171, 175, 189, 215

Bbackcompression 175barrel 91, 137beam splitting 149binder 74 removal 74Bingham fluids 35birefringence 243, 245blood pressure sensors 163Boger fluid 36bosses 134brittle fracture 238buses 209

Ccarbon black 288

catalyst system 128cavity pressure 215cellular phone 92, 99, 133ceramic feedstock 74Cinpres 17clamp tonnage 121clamping force 65, 73clamping speed 188clarity 254communication 163complexity 4composition 2compounding 295compression molding 175, 187computer 99, 133containers 113, 127conversion 81cooling 164 rate 246core 99cosmetic 133cost 20, 63, 135craze 238Cross model 100crystallinity 246, 254cycle time 20, 63, 79, 284, 302

DDarcy's law 199decoration 93, 215

Index

314 Index

defects 193degradation 99design guidelines 58 optimisation 58Desma 17Diesel effect 165dimensional analysis 64 stability 63distributive mixing 288, 308drink cups 127drying time 95DSC 81

Eejector pin 122elasticity 134elongated shapes 63elongational flow 308EMI shield 134energy dissipated 238Engle 17ergonomics 21erosion 92etching 151Eulerian time integration 167

Ffiber length 287fill time 89, 90film 93flow front 76 length 89, 94

marks 45, 92 pattern 61food packaging 127forward momentum 90fountain flow 143FTIR 81function 4furniture 189, 215fusible core technology 281

Ggain 17gas assist 15, 27, 36, 43, 63, 65, 79 bubble 58 penetration 38 propagation 81 channel 59 injection 45, 57 pins 22gaskets 114gate blush 92geometry 3, 57glass fiber 45, 287Glass Mat Technology 187glass transition temperature 245globalization 16guidance rib 59

Hheat conduction 65 dissipation 90Hele-Shaw flow 64hesitation 45

Index 315

Hettinga 18high flow resins 113hollow sections 63honey comb structures 157Husky 17

Iimmiscible polymers 288impact strength 94, 287inductive heating 164injection molding compounding 295injection speed 91inmold decoration 223 labeling 273 lamination 171innovation 1instrument panel 93interfacial tension 288interlock 92internal stress 243

KKlockner Ferromatik 17knit line 93

Llamination 175Laser Doppler Anemometry 194laser erosion 157licensing 15lithography 151

Mmachine 137

capacity 260Mannesman Demag 17mathematical model 81medical 159melt 231, 237 filling 45metering units 297micro spark-erosion 157micro-cutting 157microstructures 157micro-systems 157miniaturization 99mixing capability 307mold cavity 35 injector 57molecular weight distribution 113morphology 237

NNewtonian fluids 28, 35, 64nitrogen 254notebook 92, 99nozzle 57

Oone-shot manufacturing 1optical anisotropy 149optical components 149orientation 255overlooked 90overpacking 90

Ppackaging 113, 301

316 Index

PBT 230permeability 193PET 230photo-resist 151piston speed 38plant efficiency 262plasticating unit 301, 307platen 90Poisson equation 199polarization 149poly(vinyl chloride) 37polyamide 285, 302polycarbonate 37, 93, 99, 158, 302polyesters 230polyethylene 127polymer flow 237, 245polymethylmethacrylate 159, 302polyoxymethylene 158, 167polyphenylensulfide 167polypropylene 121, 188, 254, 274, 287, 296polystyrene 37, 121, 238, 247polyurethane 79porosity 193powder injection molding 73pressure 90, 121, 143 control 22 loss 260processing parameters 253 window 94, 113, 253pseudo-ductile 230psuedoplastic flow 117PTFE 285

Qquality 20Quick Product Change 263

Rrace tracking 193railroad cars 209ram speed 35reaction injection molding 79recordable media 149repeatability 47residence time 95, 260residual stress 237, 245resin transfer molding 193resonance 256reversed molding 188Reynolds number 64rheokinetic behavior 81rheological behavior 237 properties 81rheology 37, 73ribs 134roughness 65

Sscale 2screw speed 90screws 301seals 114SEM 153, 232semicrystalline 256shape 58shear band 238

Index 317

thinning 36shot control 22shrinkage 74 rate 89silicone 79simulation 46single site catalyst 113sink marks 79, 93sintering 74skin 89, 99slip-stick phenomenon 118sprue 260stack mold 259stiffness 95, 134, 230, 287stretching deformation 308

TTait equation 100temperature 144tensile strength 255texture painting 45thermal conductivity 283thin-wall 121tin/bismuth alloy 284tool erosion 91tooling 91toughness 230, 238TPU 113transparency 245twin screw extruder 287, 308

VVacuum Assisted Liquid Molding 209Valveless Melt Transfer System 261variothermal heating 164

venting 91vertical press 189vibration 254Vibration Gas Injection Molding 253vibrational energy 253 force 245 molding 237viscosity 27, 37, 81, 113, 164, 231 elongational 76 shear 76volume 57

Wwall thickness 38, 135warpage 1, 63, 74, 89, 99weight 20 reduction 79weldline 134

_tMEAbND3Gu

Documents

ohio state university

special molding techniques

advanced blow molding

pseudo plastic liquids

standard injection molding

assisted injection molding

assist injection molding

low capillary numbers