Western Michigan University

ScholarWorks at WMU

Master's Theses Graduate College

8-1995

Development of a Mold Trial Procedure for Establishing a Robust

Process
Marc R. Vander Kooi

Follow this and additional works at: https://scholarworks.wmich.edu/masters_theses

Part of the Industrial Engineering Commons

Recommended Citation
Vander Kooi, Marc R., "Development of a Mold Trial Procedure for Establishing a Robust Process" (1995).
Master's Theses. 4859.
https://scholarworks.wmich.edu/masters_theses/4859

This Masters Thesis-Open Access is brought to you for

free and open access by the Graduate College at
ScholarWorks at WMU. It has been accepted for inclusion
in Master's Theses by an authorized administrator of
ScholarWorks at WMU. For more information, please
contact [email protected].
DEVELOPMENT OF A MOLD TRIAL PROCEDURE FOR
ESTABLISHING A ROBUST PROCESS

by

Marc R. Vander Kooi

A Thesis
Submitted to the
Faculty of The Graduate College
in partial fulfillment of the
requirements for the
Degree of Master of Science
Department of Industrial and Manufacturing Engineering

Western Michigan University

Kalamazoo, Michigan
August 1995
 ACKNOWLEDGMENTS

I would like to express my sincere appreciation to a few of the individuals

who contributed in the completion of this thesis. First, to the members of my thesis

committee, Dr. Paul Engelmann, Michael Monfore, and Dr. Sam Ramrattan, I extend

my sincere appreciation for all of your guidance and support throughout my academic

career. Second, to Kris Dykstra, for the many hours spent editing.

Third, I would like to express my sincere appreciation to all of my loved ones.

Melissa DeGroff, who continuously provides me with support and encouragement in

everything I do. My parents, Rich and Linda Vander Kooi, for their never ending

confidence and support. I would also like to express my sincere appreciation to my

brothers, sister, and grandparents for their continued support.

And last, but not least, I would like to thank the PCIM Consortium which

includes R&TI of West Michigan, ADAC Plastics, Batts Inc., Cascade Engineering,

DuPont, IBM, Prince Corporation, and Wright Plastic Products. Without the effort of

these companies, this thesis would not have been possible. I extend a special thanks

to all of the people from the consortium member companies that I had the

opportunity to work with. You have made this research an invaluable experience.

Marc Vander Kooi

11
 DEVELOPMENT OF A MOLD TRIAL PROCEDURE FOR
ESTABLISHING A ROBUST PROCESS

Marc R. Vander Kooi, M.S.

Western Michigan University, 1995

To reduce variation in the injection molding process it is essential to

understand the effect of process, mold, and material variables on critical part

characteristics. Gaining a thorough understanding of these variables is often difficult

due to the limited time manufacturers typically have available for mold trials. To

address this issue, the Premier Class Injection Molding consortium (PCIM) has

performed numerous experiments relating to shot-to-shot repeatability. Using the

PCIM data set and experimentation procedures as a base, a procedure was developed

to collect data during a mold trial. Incorporating small experiments and statistical

analysis, the new procedure allows meaningful data to be efficiently and effectively

collected. These data are used to define a stable molding process which contributes

to the development of a more robust process. The procedure also provides consistent

and thorough documentation of the process. In addition, this study reviews the

economics of applying the new mold trial procedure.

 TABLE OF CONTENTS

ACKNOWLEDGEMENTS................................................................................... 11

LIST OF TABLES...............................................................'.................................. vm

LIST OF FIGURES................................................................................................ IX

CHAPTER

I. INTRODUCTION................................................................................ 1

Industry Growth............................................................................. I

Product and Process Variation....................................................... 2

Deficiencies in Mold Trial Methodology...................................... 5

PCIM Consortium.......................................................................... 8

Prior PCIM Research..................................................................... 8

Problem Statement........................................................................ 10

Research Objectives...................................................................... 11

Assumptions.................................................................................. 11

Limitations.................................................................................... 13

II. REVIEW OF RELATED LITERATURE.......................................... 14

Review of Process Variables........................................................ 14

Pressure............................................................................. 15

Temperature...................................................................... 16

Time.................................................................................. 17

Speed.................................................................................. 17

lll
 Table of Contents-Continued

CHAPTER

Current Process Set-up Experiments............................................ 18

Gate Seal........................................................................... 18

On-Machine Rheology Curve........................................... 19

Mold Temperature Mapping............................................. 19

Cavity-to-Cavity Evaluation............................................. 20

Optimizing Plastication.................................................... 21

Current Mold Trial Procedures.................................................... 22

Molding Area Diagram Techniques................................ 22

RJG Rigorous Mold Trial Procedure............................... 23

Hunkar MAD Approach................................................... 25

Shot-to-Shot Repeatability Data.................................................. 28

Shot-to-Shot Key Variables............................................. 28

Transferability of Key Variables..................................... 29

Experimental Procedures................................................ 30

Part Attribute Measurement Correlation........................ 32

Proposed Control Strategies....................................................... 33

Cavity Pressure...............--:-:!............................................. 33

Hydraulic Injection Pressure Integral............................. 35

Nozzle Melt Temperature Integral................................. 37

IV
 Table of Content-Continued

CHAPTER

On-line Rheology............................................................. 38

III. METHODOLOGY........................................................................... 41

Step 1 ........................................................................................... 41

Step 2........................................................................................... 42

Step 3 ........................................................................................... 42

Step 4........................................................................................... 42

Step 5 ........................................................................................... 43

Step 6........................................................................................... 43

Step 7........................................................................................... 43

Step 8........................................................................................... 44

Step 9........................................................................................... 44

IV. RESULTS........................................................................................ 46

Current Mold Trial Techniques................................................... 46

Preparation Guidelines................................................................ 54

Development of Short Experiments............................................ 54

Establishing Fill Speed.................................................... 55

Establishing Pack/Hold Time.......................................... 56

Establishing Back Pressure and Screw Speed................. 56

Mold Surface Temperature Mapping.............................. 57

V
 Table of Contents-Continued

CHAPTER

Part Surface Temperature Mapping ................................ 62

Quenching Procedure.................................................................. 65
. . .
D etermmmg Quench T 1me............................................. 66

Quenched Dimensions Versus Cooled Dimensions....... 69

Development of a Standard Designed Experiment (DOE)......... 75

Instruction Booklet. ..................................................................... 76

Computer Software Specification............................................... 76

Evaluation of Procedures............................................................. 76

Mold Trial Case Studies.............................................................. 77

Case Study # 1 Process Results........................................ 80

Case Study #2 Process Results........................................ 85

Case Study #3 Process Results........................................ 88

Case Study Mold Trial Time Resources......................... 92

Case Study Process Documentation................................ 93

Case Study Conclusions.................................................. 93

V. CONCLUSIONS AND RECOMMENDATIONS........................... 98

Conclusions................................................................................ 98

Recommendations.................................................................... 102

VI
 Table of Contents-Continued

APPENDICES

A. Interview Questions........................................................................ 104

B. PCIM Mold Trial Questionnaire..................................................... 106

C. Mold Trial Preparation Guidelines................................................. 110

D. Experiment Instructions................................................................. 114

E. Spreadsheet Program Examples..................................................... 144

F. Case Study #1 Mold and Process Documentation

Current Procedure........................................................................... 153

G. Case Study #1 Mold and Process Documentation

New Procedure .............................................................................. 158

H. Case Study #2 Mold and Process Documentation

Current Procedure......................................................................... 171

I. Case Study #2 Mold and Process Documentation

New Procedure............................................................................... 179

J. Case Study #3 Mold and Process Documentation

Current Procedure.......................................................................... 191

K. Case Study #3 Mold and Process Documentation

New Procedure.............................................................................. 193

L. Human Subjects Protocol Approval.............................................. 203

BIBLIOGRAPHY................................................................ :........................... 205

Vll
 LIST OF TABLES

1. Part, Material, and Machine Size Matrix for the Quench

Correlation Experiments................................................................................. 69

2. Analysis of Variance p-values Comparing Quenched Versus

Normally Cooled Dimensions......................................................................... 72

3. Levene Test of Homogeneity of Variance p-values Comparing

Quenched Versus Normally Cooled Dimensions ........................................... 74

4. Case Study #1 Target Dimensions, Means, and Variances............................. 81

5. Case Study #2 Target Dimensions, Means, and Variances............................ 85

6. Case Study #3 Target Dimensions, Means, and Variances............................ 89

7. Case Study Summary of Target Dimensions, Means, and Variances............. 94

8. Hours Required to Perform Mold Trials in the Case Studies......................... 95

9. Case Study Summary of Target Dimensions, Means, and Variances............ 99

10. Total Mold Trial Cost Comparison............................................................. 101

Vlll
 LIST OF FIGURES

1. Increase in Injection Molded Products........................................................... 1

2. Taguchi's Loss Function...............·····························:····································· 3

3. Viscosity Curve............................................................................................. 20

4. Two-dimensional Molding Area Diagram.................................................... 23

5. Three-dimensional Molding Area Diagram.................................................. 23

6. Hold Pressure Experiment Data.................................................................... 27

7. Mean Hydraulic Injection Pressure and Coefficient of Variation................ 36

8. Mean Screw Injection Velocity and Coefficient of Variation...................... 37

9. Mean Nozzle Melt Temperature and Coefficient of Variation.................... 38

10. Modified Extended Nozzle.......................................................................... 39

11. Titles of People Surveyed............................................................................ 48

12. Years in Molding Industry........................................................................... 48

13. Average Number of Mold Trials................................................................. 49

14. Average Hours Spent Per Mold Trial.......................................................... 49

15. Number of People Involved .....·................................................................... 50

16. People Present at Mold Trial....................................................................... 50

17. Person Determining Part Acceptability....................................................... 51

18. Criteria for Acceptable Parts....................................................................... 51

IX
 List of Figures-Continued

19. Number of Parts Saved................................................................................. 52

20. Standard Documentation During the Mold Trials........................................ 53

21. Mold Surface Temperature Locations.......................................................... 59

22. Overall Means and Variances of Each Measurement Method..................... 60

23. Location Specific Data................................................................................. 61

24. Overall Means and Variances of Each Measurement Method..................... 62

25. Location Specific Data................................................................................. 62

26. Correlation Plot at 65°F Water..................................................................... 64

27. Correlation Plot at 170 °F Water................................................................... 65

28. Quench Time Versus Shrink Arrest Time.................................................... 68

29. ABS Length.................................................................................................. 72

30. PC Length..................................................................................................... 72

31. Nylon 6/6 Length.......................................................................................... 73

32. PP Length ..................................................................................................... 73

33. PC Tensile Bar Thickness............................................................................ 74

34. PA 6/6 Tensile Bar Thickness.... .'................................................................. 74

35. Case Study #1 Hole Diameter Distribution Using the Current
Mold Trial Procedure................................................................................... 81

36. Case Study #1 Hole Diameter Distribution Using the New

Mold Trial Procedure................................................................................... 82

X
 List of Figures-Continued

37. Case Study #2 Width Distribution Using the Current

Mold Trial Procedure................................................................................... 86

38. Case Study #2 Width Distribution Using the New

Mold Trial Procedure................................................................................... 86

39. Case Study #3 Hole Diameter Distribution Using the Current

Mold Trial Procedure................................................................................... 89

40. Case Study #3 Hole Diameter Distribution Using the New

Mold Trial Procedure................................................................................... 90

XI
 CHAPTER I

INTRODUCTION

Industry Growth

Over the past 10 years, the U.S. plastics industry has experienced enormous

growth. In 1984 the industry produced approximately 40 billion pounds of material

("Barometer," 1986). The current production of material has reached approximately

Increase in Demand for

Injection Molded Products
j

i-
12 -------------
!:!.,

1
0
� 8 ...

a5 6 ---
Q)
j
� 4 .I. ..
&l
ill 2
j-
� o- 1991-92 1992-93
g
Years

Figure 1. Increase in Injection Molded

Products.

69 billion pounds per year, this represents an increase of nearly 75 percent ("Key

markets," 1993).
 This growth is the result of a steady increase in the demand for injection

molded products. As shown in Figure 1, demand rose approximately 11 percent

between 1992 and 1993 (Smock, 1993). This was 2.5 percent higher than demand

between 1991 and 1992, which increased 9.5 percent (Smock, 1993).

The increased demand for injection molded products, growing global

competition, and the continuous rise in resin prices, has forced molders to increase

process efficiency and reduce scrap to maintain or improve profit margins. Reduction

in waste is also a goal from the land fill perspective. Plastics account for about 14-20

percent of the waste stream by volume ("Landfill," 1989).

Product and Process Variation

Within the injection molding industry, there is a growing need among molders

to improve consistency and performance of their products. For example, excessive

part size and appearance variation can cause scrap directly at the press and during

secondary operations such as painting, welding, and assembly (Groleau, 1989). This

product variation can be very costly to a company. Following interviews with several

injection molders, it was found that scrap rates generally average between 2% and

5%, a significant cost to the process. Combined with loss of machine and personnel

time, there is potential for substantial reduction in profit.

A modern perspective on the cost of production variation is illustrated by

Taguchi's loss function (Ross, 1988). Taguchi's loss function addresses total process
 3
costs, including the loss to society. The loss to society are costs incurred in the

production process and those encountered by the customer while using the product

(Ross, 1988). As a product attribute deviates from its optimum, the cost to society is

increased as seen in Figure 2. The farther away the attribute deviates from the

optimum value, the more loss is incurred, often at an exponential or geometric rate.

Measurement

Figure 2. Taguchi's Loss Function.

Loss occurs even when the measurement is within the standard specifications. This is

in marked contrast to the idea of an acceptable range defined by specification limits.

Taguchi's loss function quantifies the real value of reducing variation in the product

(Ross, 1988).

As shown by Taguchi's loss function, there are costs associated with any

degree of product variation. Similarly, there are costs involved in reducing,

monitoring, and rectifying product variation. These costs are known as quality costs.
 4
The three main quality costs are often defined as prevention , appraisal, and failure.

Prevention costs are associated with the investment a company makes to prevent

variation in the process. Prevention costs incorporate the review and update of

specifications and procedures, preventative maintenance, and personnel training.

Appraisal costs include product inspection by the quality department; and evaluations

of processes, new materials and methods (Chauvel & Andre, 1985). Failure costs are

typically broken into two categories, internal and external failure (Hartley, 1990).

Internal failures are associated with extra consumption of materials and labor,

manufacturing losses, variance in materials, and rework (Chauvel et al., 1985).

External failure costs are incurred when defective product reaches the customer.

External failure costs include product liability, warranties, and loss of business. An

interesting relationship exists between the different quality costs and total quality

cost. When the cost of prevention versus the total quality cost is compared, a strong

correlation is observed. As prevention efforts increase, the total quality cost

decreases (Chauvel & Andre, 1985). In contrast, when appraisal or failure costs are

incurred, the total quality cost increases. This fact supports the argument that

prevention efforts, prior to production, have a direct and positive influence on a

company's profit margin. To improve quality, defects must be eliminated through the

use of prevention measures, rather than incurring failure costs such as rework

(Boteler, 1993). This suggests that structured mold trial procedures, which represent

prevention practice, can produce a substantial "return on investment" for reducing

quality costs. If manufacturing can identify the problems associated with the process
 5
and resolve them early on, substantial downstream quality costs can be avoided.

To best identify process problems prior to production, the process variation

must be studied. Variation can be divided into components, often defined as common

and special cause. Common cause variation is due to random noise or uncontrollable

effects in the process. Special cause variation is the result of assignable (non­

random) effects in the process. Special cause variation can often be controlled once

it is adequately defined. Since the injection molding process is complex, it is usually

advantageous to study variation through the use of statistical design of experiments

(DOE). DOE is a statistically based research method used to systematically study the

effect of process variables on product and process characteristics (Rauwendaal,

1993 ). As a result of identifying and understanding the causes of process variation, a

more robust process, or suitable process control strategy can be established.

Deficiencies in Mold Trial Methodology

Injection molding is often considered an art rather than a disciplined science

(Groleau, 1991). This philosophy of the molding process needs to become a thing of

the past. Without a complete understanding of the effect process parameters have on

the part dimensions and properties, shot-to-shot consistency can not be efficiently

achieved. Lack of shot-to-shot consistency directly and negatively impacts

profitability.

To achieve a new set of quality standards, the process must be capable of

producing parts of low variability, every shot and every production run. The problem
 6
facing the injection molding industry is the inability to efficiently establish a process

with low variability among parts. As quality becomes a mind set based on

communication rather than improvisation, patience must become a key ingredient for

improved quality (Chauvel et al., 1985). Part of the patience philosophy requires

sufficient allocation of time for thorough mold trials. Thorough mold trials are not

common practice among many molders. The usual approach is to force in a mold

trial during production using the first available person, ask them to quickly make one

"decent" part, and call it "good" (Groleau, 1991). Strictly subjective analysis methods

are common during a mold trial, and can be a poor and costly practice. Subjective

analysis provides limited information regarding the shot-to-shot consistency of the

process. It also tends to produce disinformation, reducing the efficiency of the mold

trial. Subjective analysis does not provide data to evaluate concerns such as cavity­

to-cavity differences, key process variables or process consistency.

A statistically capable process is one able to produce parts to specification.

The process must be able to absorb normal variation in the machines, materials,

environments, and procedures, while continuing to produce parts to specification

without process adjustment. A recent estimate found that less than 5 percent of

injection molding processes are statistically capable (Groleau, 1991). Some have·

blamed the low percentage on unpredictability of the process. However, process

unpredictability is a misconception. Most processes are predictable if the proper

information is gathered to understand the cause and effect relationships (Groleau,

1991). Although most processes are reasonably predictable, many molders have not
 7
yet established a procedure to systematically define and understand the cause and

effect relationships. Therefore, lack of rigorous mold trials is the major contributor

to the lack of process capability (Groleau, 1991).

The root of the problem is the lack of knowledge required to perform rigorous

mold trials and the perceived complexity of the process variables. Typical set-up

personnel either have not been trained or are afraid to apply statistically designed

experiments to define the process. Often this is the result of management not

understanding the potential of applying statistically designed experiments.

Therefore, time is generally not allocated for such practices. As a result, little

attention is given to mold trials based on statistical methodology and the discipline

required to perform them. Mold trials based on statistically designed experiments

permit the process to be rigorously tested prior to production (Groleau, 1991).

Collection, analysis and proper interpretation of statistical data provide a means to

view the process objectively. Statistical methods also help to evaluate and correct the

design and manufacture of a mold, to achieve shot-to-shot consistency.

Proper process instrumentation must be applied to gather objective data

during the mold trial. Instrumentation may include mold cavity transducers, melt

temperature probes, screw position transformers, and hydraulic pressure transducers.

Using instrumentation during a mold trial quantifies the process and assists in

understanding the characteristics of the mold.

 8
PCIM Consortium

The Plastics Computer Integrated Manufacturing (PCIM) consortium was.

formed in March of 1991 (currently known as the Premier Class Injection Molding

consortium). The function of the PCIM consortium is to research, develop, and

deploy technology in the injection molding industry. The PCIM consortium was

formed through the Research and Technology Institute of West Michigan. The PCIM

consortium is currently comprised of several injection molders, a material supplier

and a group of local colleges and universities. The consortium members include

ADAC Plastics, The Batts Group Inc., Cascade Engineering, DuPont, IBM

Corporation, Prince Corporation, Wright Plastic Products, Ferris State University,

Grand Rapids Community College, and Western Michigan University. The research

performed through the PCIM consortium is driven by the needs of the industry. The

consortium works together to provide the tools and information necessary to improve

productivity and thus, profitability of the member companies. The PCIM consortium

provides an avenue for the member companies and educational institutions to pool

their resources and capital together to fund a variety of research projects determined

by the consortium (Research and Technology Institute of West Michigan, 1991).

Prior PCIM Research

Since the inception of the consortium, a number of research projects have

been performed. Prior research topics included feed throat and oil temperature
 9
effects, communications network integration, data acquisition system

implementation, and practical on-line rheology for injection molding (Speirs, 1993,

December; Fisher, Hom, & Schulte, 1991; Creighton, Mcinerney, & Soto, 1992;

Speirs, Thayer, Wilson, & Massei, 1993, November).

In April of 1992, a project proposal was submitted to the consortium to

research the area of shot-to-shot repeatability. The purpose of the study was to

develop a strategy to control the magnitude and variation in weight and dimensions

of injection molded parts (Engelmann, Monfore, Campbell, Marshall, & Sommers,

1993). The project was accepted and research began in the fall of 1992. This

research project was the first step of the shot-to-shot repeatability project. Following

a thorough literature search, ten most frequently cited independent process variables

(according to industry experts), were chosen for evaluation in laboratory experiments

(Campbell, Marshall, & Sommers, 1992). The materials used in the study were

selected based on a 1991 survey of the consortium member companies and agreed

upon by the consortium ("Technology Survey," 1991). The materials selected were

three of the top five resins used among the consortium member companies. The first

material selected for the study was polycarbonate. Following the two polycarbonate

experiments, research was conducted in the same manner on polypropylene and

nylon 6/6. In total, two designed experiments were run on each material. The first

experiments were screening experiments and the second were higher resolution

experiments used to better understand the results of the initial screening experiments

(Engelmann, Monfore, Dykstra, Huff, Schoon, & Vander Kooi, 1994). All of the
 10
primary experimentation was performed at the Applied Technology Center (ATC) at

Grand Rapids Community College. Concurrently, a transferability experiment using

polycarbonate was performed at Western Michigan University. This experiment

evaluated the machine-to-machine transferability of the polycarbonate lab data

gathered at the ATC.

The initial transferability study proved to be very promising and prompted

further research into the transferability of the lab data (Engelmann et al., 1994). The

transferability studies were begun during the summer of 1994, and consisted of two

experiments using polypropylene and two using nylon 6/6. These experiments were

performed at the facilities of the consortium members using their molds, machines,

and materials. The purpose of this research was to determine how well the shot-to­

shot repeatability lab data transferred to production machines and molds using

different grades of materials. Achieving an understanding of the transferability of the

lab data was an important step in applying the data to the production environment.

Problem Statement

Few molders apply a rigorous mold trial procedure to document and identify

the relevant data of the injection molding process ( Groleau, 1991 ). This deficiency is

commonly the result of a lack of knowledge in applying statistical tools, such as

designed experiments. Many molders run a few parts, visually inspect the parts to

make sure they look good and begin production. There is often little effort made to

understand and document the characteristics of the mold and process. As a result, the
 11
mold and process are the source of continuous problems for technicians and

engineers in the production environment. Understanding and documenting the

characteristics of the mold and process are the key to establishing effective process

control. Without this effort, the mold may run an inefficient cycle and/or produce

parts that do not meet customer requirements.

Research Objectives

This research focused on the development of procedures which allow

efficient, effective, and statistically valid mold set-up. The objectives are as follows:

1. Use of a rigorous mold trial procedure for establishing process parameters

should result in development of a more robust process than achievable by traditional

methods. For this study, a more robust process is defined as providing a lower

process variance and a critical dimension closer to the target dimension.

2. The rigorous mold trial procedure should provide consistent and thorough

documentation of the process.

3. The economic impact of applying the mold trial procedure will be defined.

Assumptions

1. The consortium members will allow the observation and review of their

current mold trial procedures and techniques.

2. The consortium members will provide opportunities to test the mold trial

procedures at their facilities.

 12
3. The consortium members will provide sufficient opportunity to perform

comparisons between new mold trial procedures and current mold trial techniques.

4. The consortium members will provide materials, molds, and machinery for

evaluation of the mold trial procedures.

5. Sufficient time will be available during the mold trials to thoroughly test

new mold trial procedures.

6. Software can be purchased for data analysis that is acceptable to all

consortium members.

7. The data set from the transferability experiments is valid in the production

environment.

8. For molds used in the study, the mold cooling system, ejection system,

resin distribution system, and cavity geometry are designed according to standard

mold design principles.

9. The injection molding equipment has been maintained according to an

adequate routine maintenance schedule.

10. The injection molding process can be defined and modeled.

11. A computer can be present at the machine during the mold trial to record

data.

12. At least 60 shots can be saved from the evaluation of the mold trial for

review later. This makes it possible to evaluate parts from a DOE after normal

shrinkage has ceased.

13. The mold trial procedure will be applicable to both crystalline and
 13
amorphous resins.

Limitations

1. The number of verification experiments were limited by the commitment

of the consortium members and their time constraints..

2. Collection of cavity pressure data will be limited by company to company

mold differences. Adding transducer slots to each mold was strongly suggested, but

will be at the discretion of each company.

3. PCIM transferability data has only been gathered on polypropylene and

nylon 6/6. Since these are both crystalline resins, the amorphous data set is currently

incomplete.
 CHAPTER II

REVIEW OF RELATED LITERATURE

The following chapter reviews topics which are relevant to the development

of a new mold trial procedure. Initially, the process variables used in the mold trial

procedure are reviewed. This review is essential to identify the significance of the

process variables chosen for the mold trial procedure. Following the process variable

review is an examination of existing process set-up experiments and mold trial

procedures. This segment is an overview of the methods and procedures currently

taught and implemented in the industry. In addition, the findings and experimental

procedures from the earlier PCIM shot-to-shot repeatability studies are reviewed.

This section discusses the findings and procedures from which the mold trial

procedure is based upon. Finally, four process control strategies are examined. The

process control strategies are examined to provide information concerning techniques

which may be applied to signature a process after it has been defined by the mold

trial procedure.

Review of Process Variables

The injection molding process is a complex system containing as many as one

hundred independent process variables. These independent process variables can be

14
 15
broken down into four main categories. The categories are pressure, temperature,

time and speed. A review of earlier literature has shown the focus to be upon a

relatively small group of process variables (Campbell et al., 1992). The following

section contains a brief description of the variables and how they effect the process.

Pressure

Injection pressure, sometimes referred to as boost or first stage pressure, is the

pressure applied to move the screw forward from the start of injection until the

transfer to pack and hold pressure control. According to Groleau and Bozzelli, by

applying decoupled molding methods, the injection pressure can vary to counteract

viscosity variations (Bozzelli, Groleau, Kirkland, & Colby, 1993).

Pack pressure is the pressure applied to the screw for a specific amount of

time following the transfer point. Pack pressure provides enough pressure to

completely fill the mold cavity and compensate for material shrinkage. Hold

pressure is the pressure applied, following the pack pressure, to keep the material in

the cavity until the gate solidifies. Research by Harry has found pack and hold

pressure to be a reliable predictor of product attributes (Harry, 1991). The research

performed by Western Michigan University has shown pack and hold pressure to be

the most significant variables for controlling product attributes when molding

polypropylene and nylon 6/6 (Dykstra, Engelmann, Huff, Monfore, Schoon, &

Vander Kooi, 1994). Experimentation by Ricketson and Wang at Cornell University

concluded that pack/hold pressure had the greatest effect on part thickness for
 16
polystyrene and polypropylene (Ricketson & Wang, 1987).

Back pressure is commonly referred to as the resistance on the back of the

screw during the screw recovery. Back pressure is used to improve the consistency of

the melt during the recovery stage. It is also one of the key variables effecting the

melt uniformity and temperature (Dym, 1986). Hunkar cites back pressure as one of

the most significant variables to establish first, when using the Hunkar molding area

diagram (MAD) approach (Hunkar, 1994).

Temperature

Melt or plastic temperature is the actual temperature of the polymer. It is not

the temperature of the barrel zones. In-line melt temperature can be measured very

accurately using an infrared melt probe (Miller, 1992). The simplest alternative is to

measure the melt temperature using a hand held probe. The importance of melt

temperature on hydraulic and cavity pressure has been shown at McGill University (

Kamal, Patterson, & Gomes, 1986). In addition, Groleau found that the melt

temperature significantly effects nozzle pressure and fill time (Groleau, 1989). The

effects of melt temperature on part weight, which can correlate well with product

dimensions, have also been reported to be significant by researchers at Ohio State

University (Srinivasan, Srinivasan, & Maul, 1992).

Mold or cavity temperature refers to the surface temperature of the mold. The

actual surface temperature of the mold is usually not directly controlled. Instead, the

temperature of the fluid flowing through the mold cooling circuits is controlled using
 17
a mold temperature controller. For most materials, the part quality is largely

dependent on the consistency and uniformity of the mold surface temperature (Dym,

1987). To achieve consistent and efficient cooling, Groleau and Bozzelli have shown

the significance of balancing and establishing turbulent flow through all of the

cooling circuits (Bozzelli et al., 1993). Studies by Liu and Manzione, at AT&T Bell

laboratories, have shown mold temperature to significantly influence dimensional

precision in amorphous resins (Liu & Manzione, 1993).

Pack time is the duration the pack pressure is applied. Pack time should be

applied long enough to allow the cavity to finish filling and push extra material into

the cavity to compensate for shrinkage. Hold time is the duration the hold pressure is

applied. Hold time should be applied until the material at the gate has solidified,

with a few exceptions. This prevents material from discharging out of the cavity,

thus creating an undersized part. Research by Harry has reported the significance of

providing the proper pack and hold time to achieve part consistency (Harry, 1990).

Screw speed is the revolutions per minute the screw is turning during screw

recovery. Screw speed has a significant effect on the plastication process. As screw

speed increases, screw recovery time decreases. However, an increase in screw speed

may require a longer screw channel length to provide thorough melting of the plastic
 18
material. Therefore, screw speed has a significant effect on the melt consistency and

temperature (Nunn, 1986).

Injection velocity or fill rate is the rate at which the polymer is injected into

the cavity. According to Groleau, in most cases the polymer should be injected into

the mold as fast as possible to take advantage of the smaller incremental changes in

viscosity due to the non-Newtonian characteristics of the polymer. Filling the mold

rapidly allows the polymer to flow easily and reduces the amount of viscosity change

caused by slight fill time variations (Groleau, 1991 ).

Current Process Set-up Experiments

To assist in optimizing the injection molding process, several simple

experiments have been developed by experts in the industry. The application of gate

seal experiments and on-machine rheology curves experiments have been stressed by

John Bozzelli and Rod Groleau. The use of mold and part temperature mapping

techniques and cavity-to-cavity comparison have also been emphasized in the plastics

industry. In addition, Dow has developed an experiment to define the parameter

settings for optimum plastication. The following section will briefly summarize the

application, as well as, the methodology associated with each procedure.

Gate Seal

Gate seal experiments are used to determine the time needed for the plastic at

the gate to become frozen or sealed. There are a couple of methods that can be used
 19
for this determination dependent upon whether or not cavity pressure instrumentation

is available. The first method uses a cavity pressure transducer near the gate of the

part to monitor the pressure in the gate area. Without cavity pressure transducers, the

parts must be weighed during the experiment. The parts are weighed and recorded as

the hold time is increased by one second intervals. This procedure is continued until

an increase in the hold time does not increase the part weight. This is the point at

which the gate is considered sealed (Groleau, 1993).

On-Machine Rheolo� Curve

When defining the proper fill speed, on-machine rheology curves are often

generated. The on-machine rheology curve provides data to determine the fill speed

required to take advantage of the non-Newtonian behavior of the polymer (Groleau,

1991). The on-machine rheology curve experiment requires the mold to be filled at a

wide range of velocities. Using the fill time and the injection pressure recorded for

each velocity setting, and the intensification ratio for the machine, the relative

viscosity can be calculated. By plotting relative viscosity data versus the reciprocal

of the fill time, a viscosity curve is created. The fill time should be somewhere in the

flat area to the right of the bend, as shown in Figure 3 (Groleau, 1993).

Mold Temperature Mapping

According to Groleau, mold temperature mapping is an important part of a

rigorous mold trial (Groleau, 1991). Before probing the mold surface, the mold must
 20
be run long enough to stabilize the mold temperature. Prior to probing the mold

surface, a strategic mold map should be created. This will help to efficiently and

l/Fill Time

Figure 3. Viscosity Curve.

accurately measure the mold surface temperatures. The map should measure all of

the major surfaces, concentrating on the comers and other hard to cool areas

(Groleau, 1991).

Cavity-to-Cavity Evaluation

In order to study cavity-to-cavity differences two things are essential, a stable

process and a mold steel dimensional layout. Comparison of the average part

dimensions and mold steel dimensions are critical to determine if the variation is the

result of mold steel dimensions, process variation or other unknown factors

(Sammons, 1994).

Twenty five shots should be taken over at least an eight hour time period to
 21
properly evaluate the cavity-to-cavity relationship (Sammons, 1994 ). By plotting the

mold steel dimensions versus the average part dimensions, the cavity-to-cavity

differences can be assessed. If the differences are a result of the steel dimensions, a

mean steel dimension can be targeted to change the mold steel. The same target

dimension should be used for all of the cavities. Following changes to the mold, a

second evaluation should be performed to confirm the results (Sammons, 1994 ).

Optimizing Plastication

Dow Plastics has developed a procedure to optimize the plastication variables

based on laboratory and industrial experience. The independent process variables

used are the barrel temperature profile, screw speed, and back pressure. The

dependent variables used in the procedure are screw motor pressure, screw output,

and melt temperature. The procedure was developed to assist experienced molders in

determining the upper and lower limits for the process variables. This procedure is

not required for every mold trial, but is recommended when a new material or

machine is being used (Martin, 1994 ).

This procedure leads the individual through a 2-4 hour experiment in which

the independent process variables are manipulated and the dependent variables are

recorded. Plotting the independent variables versus the dependent variables will

yield the optimum settings. To optimize plastication, the independent variables

should be set to allow the screw motor pressure to operate at the midpoint of its

maximum range. In addition, the independent variable settings should maximize the
 22
screw output and provide the desired melt temperature (Martin, 1994).

Current Mold Trial Procedures

Many molders define the process settings simply by trial and error. In

addition, process set-up personnel often establish the parameters in various sequences

and according to their personal experiences. Personal experience is vital to process

development but, the application of a structured mold trial procedure could

significantly improve the results. To aid in the process development several

procedures have been defined including, molding area diagram techniques, RJG's

rigorous mold trial procedure, and Hunkar's MAD approach.

Molding Area Diagram Techniques

The molding area diagram (MAD) is a simplified approach used to achieve

quality parts. The simplest form of the molding diagram is a two dimensional model.

This diagram is created by plotting two variables versus one another, for example,

hold pressure versus mold temperature. The outline or borders of the diagram

represent the area in which a quality part can be produced, see Figure 4 (Rosato,

1986).

A refinement of the two-dimensional diagram is the three-dimensional

molding diagram. This diagram is created by plotting three variables versus each

other (for example, melt temperature versus hold pressure versus mold temperature).

This method produces a molding volume diagram (MVD) similar to the one shown in
 23
Figure 5. The MVD provides a comprehendible visual aid which can be used as a

guide in the production environment (Rosato, 1986).

l Short Shot Arca

---t)li► Variable I

Figure 4. Two-dimensional Molding

Area Diagram.

Figure 5. Three-dimensional Molding Area

Diagram.

RJG Rigorous Mold Trial Procedure

The mold trial procedure outlined by Rod Groleau stresses monitoring the
 24
process from the point of view of the polymer (Groleau, 1991). The use of the

variables pertaining to the polymer allows the mold to be evaluated independent of

the machine. Prior to a mold trial, a series of minimum conditions should be

established. The runners, gates and cavities of the mold should be balanced to

provide uniform filling. The cooling system in the mold should provide uniform

cooling. The mold should be of rigid construction to reduce flexing at the parting

line as well as incorporate the correct shrink factors. With these conditions

established, the methodology to define the process settings can be used effectively

(Groleau, 1991).

The following steps have been suggested by Groleau to establish a capable

process (Groleau, 1991). The steps are as follows:

1. A mold surface temperature map should be created to identify

nonuniformly cooled areas.

2. The mold and melt temperatures should be set to the manufacturer's

recommended mid-range.

3. The filling and packing stages should be decoupled to allow the maximum

fill speed to be achieved.

4. The mold should be filled as fast as possible to take advantage of the non­

Newtonian behavior of the material.

5. The part should be packed out to provide good cavity definition. Good

cavity definition is defined as a part which appears visually correct and is free of

sink. The pack pressure must also not blow open the mold causing flash or part
 25
deformation.

6. The gate seal time should be determined using either cavity pressure or part

weight techniques.

7. The plastic conditions should be recorded, including the melt temperature,

fill time, peak hydraulic pressure, peak cavity pressure and the actual mold

temperature. Some of the critical machine conditions should also be recorded,

including the overall cycle time, recovery time, back pressure, mold open and close

times, screw velocity profile and cushion size.

8. The cooling circuit flow rates and temperatures should be recorded along

with machine nozzle specifications.

9. The material viscosity should be checked, using on-machine techniques, to

verify the material is the same in subsequent mold trials.

Hunkar MAD Approach

The molding area diagram (MAD) approach, as defined by Hunkar, has been

suggested as a means to determine the optimum process settings and control limits

(Hunkar, 1994). According to Hunkar, the MAD approach can produce results with

close to the same precision as a statistically designed experiment, with minimal

expense (Hunkar, 1994). However, this method can be less efficient in terms of

number of experimental runs.

Based upon earlier work, two prioritized lists of critical process variables has

been comprised. The first of these lists is used for packing related experiments. A
 26
packing related experiment uses part dimensions as the dependent measure. The

variables used for these experiments, in prioritized order, are hold pressure, back

pressure, boost pressure, injection speed, hold time, plastication time, and mold

temperature.

The second list of process variables is used for filling related experiments.

These experiments use distortion data or stress related data as the dependent

measure. In these experiments, the following variable order is used for optimization:

back pressure, injection speed, mold temperature, hold pressure, boost pressure,

plastication time, and hold time.

Beginning with the first variable on the list, an experiment is conducted for

each of the variables to determine the range in which an acceptable part can be run

(often referred to as the process window). This type of experimentation is often

referred to as a one-factor-at-a-time experiment. Prior to performing the experiment,

a critical part dimension is identified as the dependent measure. The tolerance on the

critical part dimension serves as the range used to determine the process window.

For example, to determine the hold pressure process window, the hold pressure is set

to a pressure which is capable of filling the part to approximately 99% full. The hold

pressure is then increased in small steps with part measurements taken at each

incremental step. This process is continued until the measurement of the critical part

dimension exceeds the maximum allowable size. Upon review of the data, the

process window for the hold pressure can be defined as the pressure range capable of

producing a part that meets the critical dimension (see Figure 6).
 27
Following the determination of the process window for each variable, the

variable is set to the mid-point of the established process window for the remainder

Shot Number ----'►

Figure 6. Hold Pressure Experiment Data.

of the experiments. The same experimental technique is performed on each of the

remaining variables on the list. According to Hunkar, the oil temperature and barrel

temperature should be monitored. However these variables normally show no

significant variation and are not optimized in the experiments. Once the optimization

of the process variables is complete, the mechanical aspects of the process are set.

These include the mold open time, mold close time, and overall cycle time.

The critical aspect of this approach is the required ability to measure the parts

at press side. This saves measurement time and space required to store parts while

they cool prior to measurement. Hunkar has determined a strong correlation exists

between parts measured hot and parts measured 24 hours after molding (Hunkar,

1994 ). In order to accurately measure hot parts, the shrink process must be
 28
temporarily slowed down. This is normally achieved using a quenching process. In

an interview with Russ Henderly, an applications engineer with Hunk:ar Labs, the

following details were revealed with regard to the process used (Henderly, 1994 ).

The parts are quenched in a 62° Fahrenheit water bath. Quench time is dependent on

the material being measured and is determined through.experimentation prior to

measurement. Materials tested by Hunk:ar include polypropylene, polyethylene,

acrylic, and polycarbonate.

Shot-to-Shot Repeatability Data

Shot-to-Shot K�y Variables

Seven designed experiments were performed to determine the relationships

and significance of process variables on part size, weight and tensile strength for

selected grades of polycarbonate (PC), polypropylene (PP), and nylon (PA) 6/6. The

experiments examined the effect the process variables had on the mean and variance

of the product attributes measured. The variables used in the experiments were:

barrel temperature, coolant temperature, back pressure, screw speed, ram speed, hold

time, pack pressure, and hold pressure.

Of the eight variables studied, a small group of variables influenced the

process for each resin. Often the set of variables that had the most effect on the mean

part size, weight and strength were different from those parameters that significantly

effected the variation of the attribute. The factors found to significantly effect the
 29
mean part size, weight and strength for polycarbonate were the ram speed and

pack/hold pressure. These factors were followed in importance by barrel temperature

and coolant temperature (Engelmann et al., 1994). For polypropylene parts, the

significant variables effecting the part attributes (size, weight, and strength) were

hold pressure and barrel temperature. These were followed by pack pressure and to a

lesser extent screw speed and coolant temperature (Engelmann et al., 1994). The

most important variables influencing the mean size, weight, and strength of nylon 6/6

parts were pack/hold pressure, barrel temperature, and coolant temperature

(Engelmann et al., 1994).

Transferability of Key Variables

Following the laboratory experiments, industrial validation experiments were

performed using facilities, machines, and molds of the consortium members. The

industrial experiments were performed to determine the transferability of the

laboratory process data. A total of four industrial experiments were performed, two

with polypropylene and two with nylon 6/6.

In general, only a few of the variables effecting mean part attributes were

shown to transfer and be significant between the PP experiments. These variables

included hold pressure, pack pressure, coolant temperature and barrel temperature.

The variance effects for PP did not show any significant areas of transferability

(Dykstra et al., 1994).

Some variables transferred between the experiments for PA 6/6. These

 30
variables include pack/hold pressure, barrel temperature, and coolant temperature.

The variance effects for PA 6/6 also did not show significant areas of transferability

between the experiments (Dykstra et al., 1994).

It was apparent that for both the PP and PA 6/6 data the independent variables

effecting variation in the dependent variables lacked significant transferability

between the experiments. This could lead one to speculate that the tool geometry and

construction may play a significant role in the process variability (Dykstra et al.,

1994).

For the industrial experiments, machines ranging from 85 to 700 tons were

used. Due to the range of machines used it was concluded that the transferability of

the significant variables seems to be unaffected by the machine size (Dykstra et al.,

1994).

The transferability of the data among the industrial experiments seemed to

have a higher correlation than between the laboratory and industrial data. As a result,

the good correlation is speculated to be due to the part geometry . The parts molded

in the lab were tensile bars which had no restricting geometric features, such as right

angles. The parts molded in the industrial setting contained more restrictive

geometric features thus, creating complexity (Dykstra et al., 1994).

Experimental Procedures

To successfully collect, measure, and analyze data during an experiment, a set

of standard procedures should be observed. Standard procedures made it possible to

 31
perform all of the experimentation uniformly and reduce the experimental error. The

following section briefly describes a portion of the procedures and guidelines utilized

in the PCIM Shot-to-Shot Repeatability study.

When establishing the experimental high and low process settings, a rigorous

evaluation of the parameters is required. The parameters that are not altered during

the experiment should be set at the recommended mid-points according to the

material specifications. During experimentation, there should always be at least an

.125 inch cushion size under any of the experimental conditions. The injection stage

' should be run decoupled. Extruder delay and mold open timers should be used to

keep the cycle time and time the material spends in front of the screw prior to

injection, uniform for all of the experimental runs. The water flow in the tool should

be checked to insure turbulent flow through all of the circuits (Engelmann et al.,

1994).

Before collecting data, the machine should be run in cycle for about two hours

to allow the process to stabilize. Shot weights should be recorded in a spreadsheet

and graphed simultaneously to assist in monitoring trends in the process stability.

Following process changes, shots should be saved for each run of the experiment only

after the shot weight graph displays no upward or downward slope (Engelmann et al.,

1994).

The following guidelines can reduce error in part measurement and data

analysis. In order to view the normal distribution of the process, at least six shots

should be gathered from each run for part measurement. Due to the fact hand held
 32
calipers are very susceptible to operator induced error, gages and fixtures should be

used whenever possible. Gages and fixtures can reduce the measurement error. Data

entry errors during measurement can be reduced by having one person take the

measurement while a second person records the measurement into a computer

spreadsheet. Scatterplots should also be used to inspect the data for entry errors prior

to analysis (Engelmann et al., 1994).

Part Attribute Measurement Correlation

A new understanding of part attribute correlation has been obtained through

research performed by the PCIM consortium on shot-to-shot repeatability. While

molding standard tensile bars in laboratory experiments, a high correlation was

generally apparent between the part attributes measured (shot weight, part weight,

length, width and thickness). The high correlation of the part attributes support the

idea of using one of the part attribute measurements as a reasonable predictor of the

other measurements (Engelmann et al., 1994).

However, when experimentation was performed using real production molds

in an industrial environment, the results were quite often different. Many times a

correlation was only apparent between a couple of the part attributes (Dykstra et al.,

1994). The lack of a reasonable correlation between the part attributes clearly

indicated that the use of a single part attribute measurement as the quality predictor

may be an inaccurate assumption. Before assuming a single part attribute

measurement can be used as the quality predictor, part attribute data must be
 33
gathered and evaluated. Without evaluation of the part attribute data, assumptions

should not be made regarding the true predictive ability of a single part attribute

measurement.

Proposed Control Strategies

Due to the complex nature of the injection molding process, several control

strategies have been studied. These strategies include: cavity pressure, hydraulic

injection pressure integral, nozzle melt temperature integral, and on-line rheology.

The use of such control strategies can provide a method to measure and control the

process by monitoring the characteristics of the polymer in real time. Efforts have

been made to develop these strategies into a closed-loop system but, a fully reliable

system has not yet been developed. Currently, the best application of these strategies

is to establish a process signature. The signature provides a measure of the process

that is not machine dependent. The ability to signature the process may improve the

transferability from machine-to-machine and aid in solving molding problems.

Cavity Pressure

Cavity pressure is currently the most comprehensive control strategy for

monitoring the injection molding process. The use of cavity pressure makes it

possible to detect all of the changes in the process that reflect in the quality of the

molded part (Groleau, 1984). Changes in shear rate, screw speed, injection speed,

barrel and mold temperatures and resin shifts, which all effect the melt viscosity, are
 34
all visible in the cavity pressure data (Miller, 1989). Several techniques are possible

when monitoring cavity pressure. The first and most often implemented technique is

to monitor the process using a single cavity pressure transducer placed near the end

of fill of the part. Part quality prediction can be determined by establishing the end­

of-fill cavity pressure window during the initial mold trial (Groleau, 1984). In more

sophisticated control systems, the end of fill cavity pressure is used to transfer from

boost pressure to hold pressure. The use of cavity pressure to transfer from boost to

hold is typically not necessary in simple applications however, precision molding

requires high repeatability, which is capable with cavity pressure transfer. When the

end of fill pressure is used for transfer, the placement of the transducer is very

important. The transducer should be placed far enough from the end of fill to

compensate for the time lag between the signal and the response (Miller, 1989).

Implementation of a second cavity transducer is necessary when molding tight

tolerance parts or large parts with long flow lengths. This approach requires the use

of two transducers, one located near the end of fill and the other near the gate. By

using two transducers in these locations, the pressure profile across the cavity can be

monitored to determine part quality consistency. Using transducers in these locations

can provide automatic quality control and troubleshooting information regarding the

part's dimensions and weight (Groleau, 1989).

For control of hot runner molds, it is recommended to monitor near the end of

fill of each drop area. It has also been suggested the pressure profile in one drop area

be monitored by placing a transducer near the hot drop and one near the end of fill
 35
(Groleau, 1984). To control transfer using cavity pressure in a multi-cavity mold, a

transducer should be placed in the runner system half the flow length from the gate

and away from the sprue and other intersections (Miller, 1989). When monitoring

multi-cavity molds, it is ideal to have transducers in all cavities. If cavity pressure

transducers are impractical in every cavity, it is recommended that one cavity be

monitored and a correlation be established between that cavity pressure and the other

cavities.

Hydraulic Injection Pressure Integral

The second control strategy is currently being researched and developed by

Speight, Coates and Hull at the Mechanical & Manufacturing

Engineering/Interdisciplinary Research Centre (IRC) in Polymer Science and

Technology, University of Bradford, Bradford, UK. This approach is based on using

a portion of the hydraulic injection pressure curve to monitor viscosity changes. The

results of their research demonstrates a high correlation between the nozzle melt

pressure and the hydraulic injection pressure during the primary injection stage

(Speight, Coates, & Hull, 1994). Due to the high correlation between the two

pressures, measuring the hydraulic injection pressure is equally capable of detecting

viscosity changes when compared to measuring the nozzle pressure. Through the

application of accurate high speed machine and process monitoring equipment, a

precise hydraulic injection pressure curve can be generated. Analysis of the

hydraulic injection pressure curve over 5 cycles showed the coefficient of variation
 36
for the nozzle and hydraulic pressure to be the lowest and most constant near the end

of the primary injection stage. This area is between 1.26 and 1.76 seconds, as shown

in Figure 7. This area of the curve provides the most precise area in which to

calculate the hydraulic injection pressure curve integral. The research also shows the

Time (seconds)

----0- Mean hydraulic Injection pressurt __.,_ Coefficient ofvartatlon

Note. From "Modem Sensor Technology and Benefits for Injection

Molding Process Control" by R.G. Speight, E.P. Yazbak, J.B. Hull and
P.D. Coates, 1994.

Figure 7. Mean Hydraulic Injection Pressure and Coefficient of Variation.

effect of screw velocity control on the coefficient of variation during the primary

injection phase (Speight et al., 1994). As the screw reaches the set velocity, the

velocity becomes stable while the coefficient of variation reaches it's lowest point for

the hydraulic injection pressure, as shown in Figure 8. Therefore, the precision of the

hydraulic injection pressure integral measurement is highly dependent on the relative

performance of the machine being used (Speight et al., 1994). The use of the
 37

. 00 Meaa based oa 5 cycles �00

35
22.50
20.00 �

II
g 2�00��-1- ---+----+---+rl-"---+---+----tll:-1-;. 17.50 j
-l-4-l>-----+-+-�----1------------,-� 1�00 �

i
!_
2<l00
12.50
15.00 H¾-l�r++..q,cA-++-�....c�H---t---t---tt+-t---:i 10.00 'cj

i
I
SI 'ii
7.50
► 10.00 .m:.-Jl-�..,JJ....----11o:i:4-�--+--+----t---+- ---tH-1HtH
5.00
5.00 -ff-,,-0-......+----+-=---llitt-<1,jt--t-t-t-i.----t---t- -----tH-:�
2.50

0.00 t'll-----+---+----+---+------+-----===1=------f 0.00

0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00

Time (1ecoad1)

--o-- Meao screw lojeclioa velocity - Coefficient orvarialloo

�- From "Modem Sensor Technology and Benefits for Injection

Molding Process Control" by R.G. Speight, E.P. Yazbak, J.B. Hull and
P.D. Coates, 1994.

Figure 8. Mean Screw Injection Velocity and Coefficient of Variation.

hydraulic injection pressure integral yields a value that can be used to monitor the viscosity

variations of the polymer (Speight et al., 1994).

Nozzle Melt Temperature Intewl

Controlling and monitoring the process through nozzle melt temperature

measurement is another strategy being researched and developed by the University of

Bradford. Infrared melt temperature devices are required to accurately measure the

nozzle melt temperature. The accuracy and reaction time of a standard thermocouple

is not sufficient to collect real time data (Speight et al., 1994). To gather relevant

temperature data, the melt temperature instrumentation requires optimization. This

optimization permits the instrument to measure the polymer melt deformation

 38
heating effects. Upon review ofthe coefficient ofvariation for the nozzle melt

temperature during the injection stage, shown in Figure 9, the variation curve

Mean based on 5 cycle,

7.00 -t--<f--"""'"""T- --,---- ,----.--- �--�-�--� 5.00

4.50
6.00 -f-t----+---+---t----+--+----±:;;,m�"'t-'lr-----J
5.00 +--++ll':\--1---A--+---l----lHc----c,kjz'----+--+---'-ta------d. 3.50
4.00 t
- t;;
lt :E
3.00 �
.. ,: 4.00

1SI jf! 3.00 2.50

2.00 .,
!
!
�I! JX. 2.00
1
. 50
1.00 �
i
1.00 � --rl-'+---+--4'------,f----+-- -- ·l----+ -....ca--Hf---,--1
0.SO
0,00 b'---+---+------1f----+---�----+---+------+ 0.00
0.00 0.25 0.50 0,75 1.00 1.25 1.50 1.75 2.00

Time (seconds)

--0-- Mean nozzle melt temperature ---- Coefficient of variation

Note. From "Modem Sensor Technology and Benefits for Injection

Molding Process Control" by R.G. Speight, E.P. Yazbak, J.B. Hull and
P.D. Coates, 1994.

Figure 9. Mean Nozzle Melt Temperature and Coefficient ofVariation.

resembled the hydraulic and nozzle melt pressure curves. Similar to the hydraulic

injection pressure integral, the nozzle melt temperature integral is calculated in the

area ofleast variation (towards the end ofthe primary injection stage). When

compared to using hydraulic injection pressure or nozzle pressure, nozzle melt

temperature is more sensitive to viscosity changes in the polymer and thus is a better

measure ofmaterial viscosity (Speight et al., 1994).

On-line Rheology

The last control strategy in this discussion is on-line rheology. On-line

 39
rheology is based on measuring the pressure drop between two points in the molding

system. Experimentation at the University of Massachusetts at Lowell, has been

performed using extended nozzles retrofitted with pressure transducers, shown in

Figure 10. The transducers are used to measure the pressure drop in the nozzle. By

measuring the pressure drop, the shear stress can be calculated. The apparent shear

rate can also be calculated by knowing the volumetric flow rate and nozzle radius.

Since the shear stress and shear rate are known, the viscosity can be calculated by

dividing the shear stress by the shear rate (Dym, 1987).

1tPR4 where µ = viscosity

= P = pressure
µ -----
8QL R = radius of a cylinder
Q = amount of material flow/ sec
L = length of cylinder

Figure 10. Modified Extended Nozzle.

Using an extended nozzle provides an accurate and dependable measure of

the polymer viscosity during the fill stage of the molding process (Malloy, Chen, &

Orroth, 1988). Although an extended nozzle has been proven to be accurate and

dependable, it can be inconvenient due to the length and special machining required
 40
(Speirs, Thayer, Wilson, & Massei, 1993). As a result, an alternative method has

been researched at Ferris State University with the support of the PCIM consortium.

This alternative approach involves measuring the pressure drop between the

nozzle and the sprue puller pin in the mold. This method utilizes a pressure

transducer applied to a standard nozzle and a pressure transducer located behind the

sprue puller pin in the mold (Speirs et al., 1993). The research focused on reducing

the need to build special equipment and instead applying standard instrumentation.

The conclusion was the melt viscosity can be measured using this "on-line" rheology

method, but required further development to reduce the experimental error.

Reducing the experimental error would significantly improve the reproducibility of

the data (Speirs et al., 1993). It was also found that the viscosity could not be

calculated for any specific instance in time due to the variation in the injection

velocity (Speirs et al., 1993).

 CHAPTER III

METHODOLOGY

The project methodology was developed around a progression of nine steps.

The following discussion is broken out by steps.

Step 1

The first step in this project was a review of current mold trial techniques.

This was done to establish a baseline understanding, allowing comparison to new

procedures developed by the project. Individuals managing the mold trial process

were interviewed from each PCIM consortium company to identify techniques

currently used by injection molding set-up personnel,. Following these interviews,

actual mold trials were observed and documented. These observations provided

information on how process parameters are set, the order in which procedures are

performed, and time required to perform mold trials. As a final measure, a

questionnaire was distributed to set-up personnel from each of the consortium

member companies. The questionnaire was used to gather additional data pertaining

to mold trial activities.

Utilizing the interview, observation, and questionnaire data, a report was

prepared and distributed to the members of the consortium comparing the interview

41
 42
information with the mold trial observation information. In addition, the report

summarized questionnaire responses and discussed common practices and

differences among the companies. These data serve as the baseline description of

current mold trial practices.

Step2

Basic mold trial guidelines, to be followed prior to mold trials, were

developed to ensure proper experimental techniques. These standards were

summarized in the form of a reference booklet. This booklet includes information,

such as calibration guidelines, turbulent coolant flow qualifications, and mold setting

criteria.

Step 3

Methods were developed to establish optimum back pressure, screw speed,

fill speed, and pack/hold time (gate seal time). Methods were also developed for

mold surface and part surface temperature mapping. These methods were based on

techniques currently applied in industry, experimental data, and standard scientific

methodology.

Step4

A method to perform part measurement at the molding machine was

developed to improve the efficiency of the mold trial. Part quenching experiments
 43
were conducted to determine quench times, temperatures, and the correlation

between quenched and unquenched dimensions. Using experimental data, a

procedure was created for quenching and measuring parts at the molding machine.

Step 5

Earlier PCIM studies identified the three most critical injection molding

variables as melt temperature, coolant temperature, and pack/hold pressure. These

three variables were not optimized in step 3 and needed to be addressed. To

determine the set point of these variables, a standard design of experiments (DOE)

was utilized to optimize these variables. A procedure was written to guide data

collection and data analysis from the DOE. To improve the efficiency of the data

collection and analysis, an analysis spreadsheet was developed. The spreadsheet

allows data to be logged, analyzed, and graphed simultaneously during the

experiment.

Step 6

An instruction book detailing the procedures was developed. The book

includes topics such as, instructions to perform short experiments, data interpretation

guidelines, data collection procedures, and part measurement instructions.

Step 7

Suitable computer software was defined for data collection and statistical data
 44
analysis. Software selection was based on the requirements defined by the mold trial

procedure.

Step 8

The integrity of the mold trial procedure was tested by a series of field trials.

Set-up personnel from consortium member companies evaluated the mold trial

procedure. The mold trials were observed for clarity, efficiency, accuracy, and

thorough documentation of the process. Following the mold trials, the procedures

were edited according to the comments and observations made by the set-up

personnel.

Step9

Several case studies were performed to evaluate the performance of the new

mold trial procedure. In each case study, a mold was selected and the process was

established by set-up personnel using current mold trial methods and the new mold

trial procedure. After establishing the process using current methods, 50 shots were

collected to determine the process mean and variance. Following this period, the

process was set back to the original settings. The process was then established a

second time, using the new mold trial procedure. 50 shots were again collected to

determine the process mean and variance. The process means and variances were

compared to access the magnitude of improvement achieved using the new mold trial

procedure. The process documentation from each experiment was gathered and
 45
compared for thoroughness. Using the means and variance data, and the times

required to perform the mold trials, the economics of applying the new mold trial

procedure were accessed.

 CHAPTERIV

RESULTS

Current Mold Trial Techniques

To effectively study the implementation of a new mold trial procedure,

current industry mold trial procedures needed to be defined and understood. Three

different techniques were used to gather information regarding current mold trial

procedures. These included interview sessions, mold trial observations, and a

questionnaire.

The interview sessions were conducted with individuals who facilitate mold

trial processes in each consortium company. During the interview session, a standard

set of prepared questions were asked (Appendix A). These questions were designed

to gather specific information regarding mold trial methods uses by each company.

The second information gathering tool involved observing actual mold trials. The

purpose of this observation was to document the actual procedure used and compare

it with the methods described during the interview sessions. Information was

recorded regarding mold trial times, process documentation, and the procedure used.

The third technique utilized to gather information was a general mold trial procedure

questionnaire (Appendix B). The questionnaire gathered additional information

46
 47
regarding process documentation, personnel involved, and time spent performing

mold trials within the consortium. The questionnaire was distributed throughout the

consortium to individuals involved in the mold trial process. The selection of these

individuals was left to the discretion of each company. This was done to ensure

appropriate personnel responded to the questionnaire.

The information gathered from interview sessions and observations were

compared. This was done to determine whether the information from interview

sessions and observations correlated. The responses from the questionnaire were

tallied and reviewed using bar graphs. Graphs were used to discern the distribution

of the responses to each question. Averages and ranges were subsequently

determined for each numerical response. Questions with non-numerical responses

were summarized using pareto graphing methods.

Procedures outlined during the interview sessions were generally followed

throughout the observed mold trials. However, this may have been due to the vague

nature of the procedures outlined by the companies.

In general, the process variable settings (melt temperature, mold temperature,

pack/hold pressure, screw speed, etc.) were determined by part appearance and

personal experience. Only once during eight observation sessions were part

dimensions used to judge part acceptability . One of the companies periodically

utilized short experiments to determine fill speed and gate seal. These short

experiments appeared quite useful for determining the set points of some process

variables. In addition to personal experience, this method allowed the set points to
 48
be based on actual process data.

The questionnaire was distributed to gather a sample set of information

pertaining to mold trial procedures among the consortium companies. It should be

noted that responses on the questionnaire were to be based on mold trials used to

define variable settings for production and resolve mold function issues. These data

do not include mold trials used for mock builds or customer requests. Process

engineers and mold/process technicians appeared to make up the majority of the

personnel responding to the questionnaire (see Figure 11). Production supervisors

and other engineering disciplines represented a minority of participants. Participants

ranged in molding experience from l to l O+ years, with most in either the 4 to 6 year

range, or the 10+ years group (see Figure 12).

Titles of People Surveyed Years in Molding Industry

10+
Procea,'MoldTechnicilln
8

e!
4

Less than 1
0 1 23 456 0 1 2 3 4 5 6 7
Number al R-
Number of Responses

Figure 11. Titles of People Surveyed. Figure 12. Years in Molding Industry.

Highlights of the data from the questionnaires are as follows:

 49
1. The average number of mold trials prior to releasing a mold to production

was 7.5 with responses ranging from 3 to 14 (see Figure 13).

� -------
Average Number of Mold Trials

15+
14 ------­
VJ 13
�
1- 10 ------­
- 9
� 8
Cl) 7
.c 6
E s
� 4

1+-----+---+------+------t
0 1 2 3
Number of Responses

Figure 13. Average Number of Mold Trials.

2. The number of hours spent per mold trial averaged 5.2 with responses

ranging from 2 to 10+ hours (see Figure 14).

Average Hours Spent Per Mold Trial

10+

-
-
8

I!!
4

Less than 1
0 1 2 3 4 5
Number of Responses

Figure 14. Average Hours Spent Per Mold Trial.

 50
3. The most common response for the number of people involved during a

standard mold trial was 2 (see Figure 15).

Number of People Involved

6+

g- 5
Q)

1
Q)

4 1

3 2

z 2 10

1 4

0 2 4 6 8 10 12
Number of Responses

Figure 15. Number of People Involved.

4. A majority of the mold trials are performed by either the process engineer

or the process/mold technician. Quite often the project engineer is also present

during the mold trial (see Figure 16).

People Present at the Mold Trial

Process/Mold Technician
Project Engineer
Process Engineer ��-
Mold Setter
Production Supervisor
Operator
New Product Technician
Tooling Engineer
Quality Engineer
Manufacturing Engineer
Tooling Room Rep.
Mold Process Development Engineer +-1-+--+--+-+-+-'

0 3 6 9 12 1518
Number of Responses

Figure 16. People Present at the Mold Trial.

 51
5. During the mold trial, the project engineer or process engineer are usually

responsible for determining acceptability of the molded parts (see Figure 17).

Person Determining Part Acceptability

Project Engineer 11
Process Engineer 0
Process/Mold T ectinician
Quality Engineer
Manufacturing Engineer
Tooling Engineer
Quality Inspector
Launch Engineer
Mold Development Engineer

0 2 4 6 8 101214
Number of Responses

Figure 17. Person Determining Part Acceptability.

6. According to the questionnaire, during the mold trial, part acceptance is

most often based on both part appearance and part dimensions (see Figure 18).

However, this fact conflicts with the actual mold trial observation data. Part

dimensions were only used once during the observed mold trials.

Criteria for Acceptable Parts

Part Appearance 1!
I
Part Dimensions 16

Part Weight �
Previous Run Data I
Part Temperature I 1

Function and Fit I 1

0 5 10 15 20
Number of Responses

Figure 18. Criteria for Acceptable Parts.

 52
7. Data regarding the average number of parts saved during a mold trial,

appeared to be divided into two distinct groups (see Figure 19). The first group saved

2 to 4 parts and the second group saved 20 to 500. These data suggest a very

different mold trial strategy.

Number of Parts Saved

1
0 1g14
n
r
Q)

g
0 1 2 3 4 5 6
Number of Responses

Figure 19. Number of Parts Saved.

8. Process documentation varied among companies and personnel performing

the mold trials. Process settings sheets were produced by all respondents. A majority

also created a mold water diagram and gate seal plot. Aside from these few

similarities, there seemed to be large variations in process documentation (see Figure

20).

The data suggest that methods and documentation are not uniform among the

different personnel performing mold trials. In addition, these data suggest an average

of 39 total hours are spent performing mold trials, prior to releasing a mold to

production. According to the data, this time varied from 6 to 140 hours, depending
 53

Standard Documentation During

the Mold Trial
Process Settings Sheet 19
Gate Seal Plot 11
Mold Water Diagram ·11.
Cavity-to-Cavity Evaluations 8
Mold Temperature Map 7
Or>-Machine Rheology Curve 7
Part Temperature Map 6
Mold Water Flow Rates 6
Process Window Diagrams 5
Cavity Pressure Traces 4
Melt Temperature 3
Fill Weight -2
Improvement Notes -2
Mold Temp. Not a Map -2
Part Surface Temp. Not a Map -2
Cause for Sample Delays 1
-�
Peak Plastic PSI -1

0 5 10 15 20
Number of Responses

Figure 20. Standard Documentation During the Mold Trial.

on the mold. These data were subsequently used to define current procedures and

were utilized in comparisons with new procedures developed by this research.

Through this portion of the study it was determin�d that procedures outlined

by the companies were generally followed. However, probably because procedures

are not rigidly defined, there are areas of the mold trial process which are not uniform

among personnel in the same company. This lack of uniformity implies that well

defined and consistent procedures are not always utilized. This also presented an

opportunity to evaluate whether implementing a new rigorous mold trial procedure

 54
could improve the current mold trial process. This evaluation will aid in determining

whether a more rigorous mold trial procedure is practical. Furthermore, a

determination of the cost-to-benefit ratio could clarify the viability of any procedural

changes.

Preparation Guidelines

To gather consistent data during a mold trial and during production, it is

essential for the molding machine and mold to be properly and consistently prepared.

Lack of proper and consistent preparation may induce additional process variation.

Adhering to a set of standard preparation guidelines prior to mold trials and

production runs, improves consistency and comparability of data between mold trials.

To assist in preparation of the mold and machine prior to mold trials and

production, a preparation guidelines were created. These guidelines were developed

using previous PCIM research standards and data, current mold setup publications,

and personal experience. The guidelines include machine calibration specifications

developed by the PCIM consortium. In addition, the guidelines provide tips for

improving mold and machine setup consistency. A copy of the guidelines may be

found in Appendix C.

Development of Short Experiments

Procedures used to establish process settings often rely on subjective analysis

techniques. However, objective analysis techniques provide measurable data from

 55
which to make decisions. Objective data gathering can be done through a series of

short designed experiments. Variables addressed in these short designed experiments

included fill speed, pack/hold time, back pressure, and screw speed (Appendix D). In

addition procedures were developed for mold and part surface measurement

(Appendix D).

Establishing Fill Speed

On-machine rheology curves are an objective method to determine the fill

speed required to take advantage of the non-Newtonian behavior of the polymer. The

fundamentals used to determine optimum fill speed were drawn from on-machine

rheology techniques developed by Rod Groleau and John Bozzelli of RJG

Technologies (Groleau, 1993). Other similar rheology methods were also reviewed

and referenced (Advanced, 1990; Tobin, 1992; Whelan, 1994). The fill speed

experimental procedure for this study was primarily based on methods outlined by

RJG Technologies and other sources.

The procedure requires the mold to be filled at a wide range of velocities.

Using the fill times, injection pressures, and intensification ratio of the molding

machine, relative viscosity can be calculated. By plotting the relative viscosity versus

the reciprocal fill time, a viscosity curve is created.

Along with the procedures for performing the experiment, interpretation

guidelines were written to aid in correctly determining the optimum fill speed. In

addition, a spreadsheet program was created to log, graph, and print the experimental
 56
data for permanent documentation (Appendix E).

Establishing Pack/Hold Time

Gate seal experiments are an objective method often used to determine the

time needed for the plastic at the gate to become frozen.or sealed. The experimental

method for determining pack/hold time, also known as gate seal time, was drawn

from standard techniques outlined in numerous publications (Groleau, 1993;

Advanced, 1990; Tobin, 1992; Whelan, 1994). A review ofthese methods found

them to generally be equivalent. Therefore, the procedure for determining pack/hold

time was based upon these published standard techniques.

The procedure requires part weights to be recorded as the pack/hold time is

increased by one second intervals. This process is continued until an increase in the

pack/hold time does not increase the part weight. This is the point at which the gate

is considered sealed.

In addition to the procedure, a spreadsheet program was created to log, graph,

and print data from the experiment for permanent documentation (Appendix E).

Establishing Back Pressure and Screw Speed

Back pressure and screw speed were initially reviewed through laboratory

exploration ofa method discussed by Martin, 1994. Experiments were performed

using several materials under different molding conditions. It was determined this

experimental method did not provide reliable data. Consequently, this method was
 57
not incorporated into the mold trial procedures. Following discussions with area

injection molding process development experts, an objective procedure was

developed to establish the back pressure and screw speed. While, the procedure is

not as thorough as those used to determine fill speed and pack/hold time, the

procedure does provide objective data.

The procedure for determining back pressure measures the inherent back

pressure induced by the material and melt temperature. The back pressure is set

slightly higher than the inherent back pressure and the melt temperature is measured

to insure the desired temperature is maintained.

The procedure for determining screw speed measures the recovery time

versus the cooling time and melt temperature. The screw speed is set so the screw is

capable of recovering in slightly less time than the cooling time required. The melt

temperature is then measured to insure the desired temperature is maintained.

Mold Surface Temperature Mapping

Defining and documenting mold surface temperature is an important

component of a new mold trial. Evaluation of mold surface temperature distribution

provides a method to warn when cooling conditions have changed. These changes

may be the result of inaccurate water hookup, restricted waterlines, or other water

system problems. Measurement of the mold surface temperature allows cooling

efficiency and temperature distribution to be evaluated. These variables may then be

optimized prior to production.

 58
A well defined procedure to measure and document the mold surface

temperature must be followed to insure measurement precision. The short amount of

time often provided for mold trials, requires the method be efficient. A lack of

published information regarding mold temperature mapping procedures, required the

design and execution of experiments to define an appropriate procedure. The

purpose of this study was to determine the method most capable of quickly and

precisely measuring the mold surface temperature over a wide range of mold

temperatures.

Sample data was gathered using an 85 ton VanDom, with a standard ASTM

test bar mold housed in a master unit die (MUD) unit. An AEC Truetemp series

mold temperature controller was used to regulate mold cooling water temperature.

Material molded in the experiment was a polypropylene copolymer. Mold surface

temperature was measured using an Omega Handheld Microprocessor Digital

Thermometer with a J type thermocouple surface probe.

A mold surface map was initially prepared which identified precise locations

on the mold surface where temperature measurements were to be taken. The

experiment was performed at two water temperature treatment levels, 65°F and

170 °F, to simulate a large temperature range. Three methods of measurement were

compared at each temperature level. Within each method, four predetermined

locations were measured on each half of the mold, providing a total of eight test spots

(see Figure 21). In each of the test spots, three replicate measurements were taken in

random order between test spots to control measurement error.

 59

Top Top

0 4 ©

®
©

Moving Half Stationary Half

Figure 21. Mold Surface Temperature Locations.

Method 1: (a) interrupt the process after each cycle, (b) measure one test spot

on the mold surface.

Method 2: (a) interrupt the process after every two cycles, (b) measure one

test spot on the mold surface.

Method 3: (a) interrupt the process after each cycle, (b) measure two test spots

on the mold surface.

The data were screened for entry errors using scatterplot graphs. Following

data screening, the data were analyzed using standard analysis of variance techniques.

The analysis was conducted separately on each of the two cooling water temperatures

used in the experiment. Each analysis of variance tested for average differences

between measurement methods. A 95% confidence level was used to establish

significant differences. A Levene test of homogeneity of the variances was also

executed at a 95% confidence interval (Statsoft, 1995). This analysis evaluated the

difference between the variances of each measurement method. In addition, data

 60
from each location on the mold surface map were graphed using box and whisker

plots. These graphs were used to supplement a determination of the optimal

measurement method.

The results from the analysis of variance and Levene test at the 65°F water

temperature produced p-values of .516 and .577, respectively. This indicated there

were no significant differences (p>.05) between the means or variances of the

measurement methods tested (see Figure 22). However, a review of the location

specific data suggested four of the eight locations (locations 1, 2, 3, and 6), within

method 3 produced a much larger standard error than other methods. In the

remaining locations, the standard error appeared comparable (see Figure 23). This

indicated method 3, two measurements after each cycle, was not a preferred method

of mold surface temperature measurement at low temperatures.

Average Mold Surface Temperature

65 F Water Temperature
ss�---------------�

1--- - - - - - - -
83

J � =: 1_._
81 ··•

79 .. --=;===---------- ---

�a__,1 L · � I�

69 I �Id.Dev.
67 '-----�---�---�----'
D zStd.Err.
2 3 a Mean
Method

Figure 22. Overall Means and Variances of Each Measurement Method.

 61

Mold Temperatures by Location

·· ···
65 F Water Temperature

F · •
• ;•
• · ·i-·
· · 1 t · · · · l·· · · i · · ]f·
· t::::::;::::=.:-.\! · · ;;;;:
· · i · · · · · l ·::J1
90 : : : : : : : :

� :::�:--! ::±::@: :J L::· :·�:1.:. :-t:�::J

:t:
� : LF:. :;;:::·1· ·::::;:::·1..i .::::;:::·1i :::·]j LF:. :·�:·1·! · ;;;;;;::· ·: · · ;;;;;;;· •1i · ·: :· ]1 F:�· · :,.....· ·i · ·:�· · 1 · ·:�· ·1! · ·: ·1]
1 2 3 1 2 3 1 2 3
LOCATION: LOCATION: LOCATION:
1 2 3

: : :
j �
66
( ; � ; : � 3 : � �

�:f-L:··..,.: :!i:::.··�···!· ·:+: c:·:··]1 f···�·

Ii LOCATION: LOCATION: LOCATION:

�·:::]1
� • 5

2
L:.::::::jJ :::+
: : \::: :t 2 3
I "5td.DeY.
D ,t$td. Err.
LOCATION: LOCATION:
7 8 D Melln

Method

Figure 23. Location Specific Data.

°
The results from the analysis of variance and Levene test at the l 70 F water

temperature produced p-values of .208 and .058, respectively. These values indicate

there were no significant differences (p>.05) between means or variances of the

measurement methods tested (see Figure 24). However, the variances were

significantly different at a 90% confidence level, with method 2 showing the lowest
°
variance. A review of 170 F location specific data suggested high variance is

dependent on the test spot more than the test method (see Figure 25).

The data from this study indicated that for both high and low mold

temperatures, there were no statistically significant differences (p>.05) among means

and variances of the measurement methods studied. However, when using low water

temperatures, method 3 was more variable at some locations, suggesting it is a less

precise method under some conditions. In addition, at a 90% confidence level,

method 1 variances were larger at higher temperatures. Although differences

measurement after every two cycles, as the preferred method of mold surface
 62
Average Mold Surface Temperature
170 F Water Temperature
168r----�---- �---�---- -----,

167

166

165
l l I
� 164 ·

I . . .... -· : ·r·:.
1!·1 �·�· ·· · · ·· · · ··
j 163
I
162
�
161

160 ················!···
· ·········+······
159 ··· ···· ·· ···· ······· ··--·· ·· ·····..-·. I :J:Std.Dev.
· ·· · · ··
158�---�----�---�----� D ¼Std.Err.
2 3 a Mean
Method

Figure 24. Overall Means and Variances of Each Measurement Method.

Mold Temperature by Location

170 F Water Temperature

1�;�* �::::::: J::d,;:: ::r:: : 1 r•±:':3=:::T::::: :l �:: r:r: �: : 1

� ... ... �...... ..
·· · · · · ··· · ·· · · ·· ·
···••;••·· ···· '·········· ;· ·· ·· · ·· · · ••:•• ······••:••··· ····;··· ·······
1 2 3 1 2 3 1 2 3

i l�t::$�•*:d k:�t:t::• I t::•*•t:=r:::i

LOCATION: LOCATION: LOCATION:
1 2 3

� 1 2 3 1 2 3 1 2 3
LOCA TION: LOCA;ION: LOCA tON:

�:: $;:�i%i:: :I · :@: :l

{E_ 4

�i;� t . . ! . . . ; . . . ; ... . �:t· ::;;;;:.!

.)..... ; . . . . t· · · · ·
167 : ; ; . : :

I ±Std. Dev.
1 2 3 1 2 3 D ±Std. Err.
LOCATION: LOCATION:
7 8 a Mean
Method

Figure 25. Location Specific Data.

between methods were relatively small, these data suggest method 2, one

temperature measurement. The data from this study was used to write a mold surface

mapping procedure.

Part Surface Temperature Mapping

In situations where mold surface temperature changes very rapidly between

 63
molding cycles, it is often difficult to obtain an accurate surface temperature

measurements. This is particularly true when using super chilled water, ora large

mold. In both these situations, by the time a measurement is taken from the surface,

the surface may have reached equilibrium with the mold cooling water. Therefore, a

procedure to measure part surface temperature was reviewed as a substitute for mold

surface temperature measurement. An experiment was subsequently performed to

evaluate the correlation between mold surface temperature and part surface

temperature.

Sample data in the experiment were gathered using an 85 ton VanDom

injection molding machine with a standard ASTM test bar mold housed in a MUD

unit. Temperature of the water circulated to the mold was regulated with an AEC

Truetemp series mold temperature controller. Material molded in the experiment

was a polypropylene copolymer. The surface of the mold and parts were measured

using an Omega Handheld Microprocessor Digital Thermometer with a J type

thermocouple surface probe.

The mold surface temperature data used for the correlation experiment were

gathered by creating a mold surface temperature map following method 2, as outlined

in the previous section of this document (Mold Surface Temperature Mapping). As a

first step, a part surface map was prepared. This map targeted the same test spots

measured during mold surface mapping experimentation. The experiment was

performed at two water temperature treatment levels, 65 °F and 170 °F, to simulate a

large temperature range. Within each temperature treatment, four test spots were
 64
measured on both sides of the part. The location of these part test spots corresponded

to locations tested in the mold surface temperature experiment. Three measurements

were taken in random order at each of the eight test spots. The three measurements

were averaged, providing a single mean for each test spot.

Data from the experiment were examined for data entry errors using

scatterplot graphing techniques. Following the screening process, plots of mold

surface versus part surface temperatures were created for the two water cooling

temperature treatments. These data were also evaluated using simple linear

(Pearson) correlation statistics.

The linear correlation for the data using 65°F water was .58 (see Figure 26).

Correlation Plot at 65 F Water

123�-----------------�

121
0
o
g: 119 f······················i······················•·················-·····'················· ·····•·····•··················'·········•·•·-········i•···•······•· ···:;:,--.''°'�
...................1

I
:...----
� 0 -------
117 �· · ...............; .............. .. . .; ...................... +· · ·····•···· .;·-·•·········::;;;,-;.c.:··········· ·+·····················;........... . · ·i
,
!i'!l �
115 f· · ················+················· ···i··················•c;;;.-c.: .............. +.......................;...................... ;.......................•.....................1

i
111
------- --------:--
" 113 f···· ··············:::,,-''°"··············7············'···········i-······················t·······················i········ ·············-r···•·•·········· ·····t··················· ·1
�
o
0

109'---�-�--�-�--�-�-�-�
70 72 74 76 78 80 82 84 86
Mold Surface Temperature (F}

Figure 26. Correlation Plot at 65°F Water.

The linear correlation for the data using l 70°F water was .43 (see Figure 27). Both of

these correlations were statistically insignificant (p<.05) and suggest a limited

correlation between mold and part surface measurements. In addition, a review of

the correlation graph data suggest the maximum measurement error in these two
 65
Correlation Plot at 17 0 F Water
180 .-----------,---------,------------�

178 •··················..················•·····································•····························•·········•·····································'·················· ····················•

0
� 176 �·····································•·························•······•·····•························•·············'-·····································•·······································1

i
0

174 t·· ··································c··-····················•·•··•·······t···············-····················t·····································;· ··················=..-c.

-

! 172 ---------
.� ,.,-
__,-
0

·--------
� o �
t 170 f···· ································•···················-················i;;-""'·····················-···•················-···················i············•················· ·····•·�
�
168
� 0
166�--�---�---�---�--�
160 161 162 163 164 165
Mold Surface Temperature (F)

Figure 27. Correlation Plot at 170°F Water.

treatments was approximately 8 ° F.

These data suggest use of part surface temperature mapping as a substitute for

mold surface temperature mapping is not suggested for detecting small differences in

mold surface temperature uniformity (<8°F). However, these data do suggest that if

one were mapping a hard to reach mold or a mold running chilled water, part surface

temperatures may be useful in identifying large mold surface temperature uniformity

differences (>8 °F). Using the knowledge gained from this experiment, a procedure

was written for part surface temperature mapping.

Quenching Procedure

The capacity to measure parts immediately after they are molded is of great

value in gaining a real time understanding of the process characteristics. Currently,

dimension data is not generally collected and correlated to the process at the molding

machine. This is generally related to concerns with part shrinkage. If measurements

 66
are taken on hot parts, the parts are shrinking while the measurement is being taken.

Shrinkage of the part during measurement can induce large measurement error and

reduce the reliability of the data. Error of this type produces a significant barrier to

performance and analysis of a designed experiment at the molding machine.

To circumvent this problem, a technique for temporarily arresting shrinkage

of the part is required. Arresting or controlling shrinkage during measurement

increases both accuracy and precision. One means of arresting shrinkage is water

quenching the parts before measurement. A study was constructed to investigate this

technique. The objectives of this study were as follows: (a) to determine quench

conditions adequate to arrest the shrinkage long enough for part measurement, (b) to

determine the correlation between dimensions of quenched parts and normally cooled

parts, and (c) to determine if quenching effects variation in part measurements.

Determining Quench Time

A quench time study was developed to determine the amount of time a part

needed to be quenched in order to arrest the shrinkage long enough to dry the part

and take three measurements. Arrest time was established by gauging the time

required to measure the length, width, and thickness of a tensile bar. Five seconds

were added to the time to account for error in the process. For this study, the amount

of time required to take three accurate measurements was determined to be 15

seconds. Materials used in the study include polypropylene copolymer (PP), ABS,

polycarbonate (PC), and nylon 6/6 (PA 6/6).

 67
Tap water, at 68 F in a five gallon bucket, was used for quenching. Water
°

temperature was monitored throughout the experiment using an Omega Handheld

Microprocessor Digital Thermometer with a J type thermocouple. Measured parts

were standard ASTM tensile bars produced using an 85 ton VanDom molding

machine. Part lengths were measured using a fixture and a MT100M Heidenhain

digital plunger gage measuring to .00001 inches.

Parts were produced at two different melt temperature treatments for each of

the four test materials. For each melt temperature and material type, parts were

produced and a tensile bar was removed from the runner system. The tensile bar was

then quenched for five seconds and placed into the measurement fixture. The length

of the tensile bar was measured with the plunger gage. The dimension was

monitored and the time required for the dimension to shrink .001 inch was recorded.

Six replicate sample parts were taken for each quench time tested (5, 15, 20, and 30

second intervals). This process was repeated for all materials, at both high and low

melt temperature treatments. In addition, quench water temperature was recorded for

each sample.

Within each material group, there was no significant interaction effect of melt

temperature and quench time on shrink arrest time (see Figure 28). However, quench

time effects on shrink arrest time were unique for each material group. PP exhibited

a relatively linear relationship between quench time and arrest time. Adequate arrest

time occurred at a 20 second quench time. ABS exhibited a more pronounced and

less linear relationship between shrink arrest time and quench time. The desired
 68
arrest time occurred at a 20 second quench time. However, due to the nonlinearity, at

a 30 second quench time, the resulting arrest time was more than 60 seconds (see

Figure 28). PA 6/6 and PC exhibited similar shrink arrest time responses to quench

Quench Time Versus Shrink Arrest Time

70�-------------------�

60

50 .....

Cl)
� 40
Cl)
E
i= 30
iii

-"' 20 -o- PP (390F)

C
·•O·· PP (430F)
w 10 ··+·· PC (490F)
-..- PC (530F)
0 .......................... . -•- NYLON (535F)
-·■·· NYLON (565F)
-- ABS (440F)
-10 �---�-------�---�---�
5 15 20 30 ···•··· ABS (480F)

Quench Time (Sec)

Figure 28. Quench Time Versus Shrink Arrest Time.

time. Shrink arrest time effects were generally higher for these materials as

compared to PP. PA 6/6 at 535°F and PC data both show the desired arrest time

could be achieved at a 15 second quench time (see Figure 28). Data for PA 6/6 at

565°F suggested the desired arrest time could be achieved at approximately 17

seconds of quench time (see Figure 28). 200 samples were quenched during a 3 hour

period in this study. Throughout the experiment the quench water temperature

variation was limited to the range of 68°F to 72° F.

 69
Data from this study indicated, for a .125 inch thick part, a 20 second quench

time would arrest the shrinkage in all of the materials studied, long enough for

measurements to be taken. Also, the change of 4 °F in quench water temperature did

not significantly effect the results of this study. This was determined by replicating

the PP experiment at the completion of the study. These data suggested quenching

water temperature does not require sophisticated control to be effective.

Quenched Dimensions Versus Cooled Dimensions

The first objective of this study is to determine the correlation between

quenched and unquenched part dimensions. The second objective is to determine

whether part quenching effects variation. At the conclusion of this study, it was

determined to what extent quenched part measurements could be used during a mold

trial to predict final dimensions.

In this study, a series of different parts were molded using ABS, PC,

PA 6/6, and PP. Parts molded were tensile bars, an automotive bracket, and a clear

lens (see Table 1).

Table l

Part, Material, and Machine Size Matrix for the Quench Correlation Experiment

Part Material(s) Machine Size

Tensile Bars ABS, PC, PA 6/6, PP 85 Ton
Automotive Bracket ABS, PC, PA 6/6 220 Ton
Clear Lens PC 85 Ton
 70
Parts were measured using Mitutoyo Digimatic digital calipers, a Mitutoyo

Handheld plunger gage, and a MTl00M Heidenhain digital plunger gage and fixture.

Parts were molded in machines ranging from 85 to 220 tons. Parts were quenched in

a five gallon plastic bucket using 65 °F to 70°F tap water.

In the experiment, the tensile bars and automotive brackets, both

approximately .125 inches thick, were quenched for 20 seconds. The lenses,

approximately .28 inches thick, were quenched for 40 seconds. The 40 second

quench time was determined by quenching parts for 20, 30 and 40 seconds and

comparing the quenched measurements with the normally cooled dimensions. These

data suggest that 40 seconds of quench time produced dimensions comparable to

normally cooled dimensions. This also suggests an approximately linear relationship

between part thickness and quench time.

Eight parts were quenched and measured after being molded for each material

and part combination. The part molded immediately after each quenched part, was

saved and allowed to air cool normally for 48 hours before measurement. Different

dimensional measurements were taken for each part. Length, width and thickness

dimensions were measured on each of the tensile bars. Length and width dimensions

were measured on the automotive brackets. Width dimensions were measured on the

lens'.

Data gathered from each material and part combination were evaluated using

standard analysis of variance techniques. This analysis tested the average differences

between the means of the quenched dimensions and the normally cooled dimensions.
 71
A 95% confidence level was used to establish the statistical significance of the

differences. A Levene test of homogeneity of the variances was also executed at a

95% confidence level. This analysis tested for differences in the variances of the

quenched and normally cooled dimensions. Graphs of the data were used to

supplement a determination of the correlation.

Analysis of variance results for tensile bars indicated a statistically significant

difference in the length for ABS, PC, and PP (p<.05), but not for nylon 6/6 (p>.05).

The width dimension showed a significant difference in the ABS tensile bars only.

The thickness dimension demonstrated no significant differences between the means

for any material (see Table 2). Analysis of the automotive bracket data displayed

statistically significant differences in the length and width dimensions for both PC

and nylon 6/6 parts. The ABS automotive bracket demonstrated a significant

difference in the length dimension only (see Table 2). The lens data, molded in PC

only, indicated a statistically significant difference between the means of the width

dimensions (see Table 2).

Although, means in some data sets were statistically significant, many of the

differences averaged less than .002 inches. These differences are likely of little

practical importance. The mean quenched ABS dimensions were approximately .000

to .001 inch larger than normally cooled dimensions (an example is shown in Figure

29). The quenched PC dimensions were approximately .000 to .002 inches smaller

than the normally cooled dimensions (an example is shown in Figure 30). The

quenched PA 6/6 dimensions were approximately .000 to .003 inches larger than the
 72
normally cooled dimensions (an example is shown in Figure 31). QuenchedPP

Table 2

Analysis of Variance p-valuesComparing Quenched Versus

NormallyCooled Dimensions

Part and Material p-value Length p-value Width p-value Thickness

Tensile Bar -ABS .000* .000* ----
Tensile Bar -PC .023* 1.00 .151
Tensile Bar -PA 616 .293 .246 .065
Tensile Bar -PP .000* .074 .174
Auto Bracket -ABS .009* .090 NIA
Auto Bracket -PC .000* .010* NIA
Auto Bracket -PA 616 .000* .000* NIA
Lens -PC NIA .040* NIA

* Significant at a 95% confidence level.

ABS Automotive Bracket Length PC Automotive Bracket Length

p=.009• p= .ooo•
5.11115 � ----- - �
- 5.9157 r--------------,
- ------------- - ,,.- -
5.9665

t
5.9845

:::::::::��:::::::::::::::::
:/::
� 5.1188
-----------------------/·------------

I 5.984

5.9835 ----------------------�---------
5.9855

5.985

5.9845
-
5.983 -'--- +-----+----" 5.1164 -'----+-------�
Normally Cooled Normally Cooled

Figure 29_ ABS Length. Figure 30. PC Length_

length dimensions were .01 l inches larger than the normally cooled dimensions (see

Figure 32)_ The quenching process appeared to induce approximately 90% of the
 73
Nylon 6/6 Automotive Bracket Length PP Tensile Bar Length
p=.�
5.833 �---''----- ---� p=.ooo•
a003�----- -- -�
5.932 6.001
5.9119
i
I
:[ 5.931
5.997
5.93 5.
� 995
5.993
5.1129
5.991
5.928 �-----+---- 5.989 +----+----- -
Nonrally
--- Cooled Nomially+--
Cooled ---'
�

Figure 31. Nylon 6/6 Length. Figure 32. PP Length.

polypropylene's total shrink. However, the difference between quenched and

normally cooled PP width and thickness dimensions was less than .001 inch. These

data suggest quenched part dimensions are a reasonable predictor ofnormally cooled

dimensions.

The Levene test ofhomogeneity ofthe variances showed the difference in

variance between quenched and normally cooled parts was statistically significant in

only 3 of19 cases, PC and PA 6/6 tensile bar thickness, and PP tensile bar length (see

Table 3). Two ofthe three significant cases, thickness ofPC and PA 6/6 tensile bars,

were oflittle practical concern because in both cases, only 2 ofthe 10 sample

measurements varied by .001 inch (see Figures 33 and 34). These data indicate that

in approximately 95% ofall cases, variance is not significantly effected using

quenched parts.

This study suggests that quenching parts for measurement at the molding

machine is effective. For ABS, PC, and PA 6/6 the quenching process produces

dimensions that are close (.000-.003 inches) to the normally cooled dimensions.
 74
Table 3

Levene Test of Homogeneity of Variance p-values Comparing Quenched Versus

Normally Cooled Dimensions

Part and Material p-value Length p-value Width p-value Thickness

Tensile Bar - ABS 1.00 .334 ----
Tensile Bar - PC .354 ---- .001*
Tensile Bar - PA 616 .356 .293 .000*
Tensile Bar - PP .000* .081 1.00
Auto Bracket - ABS .334 .230 NIA
Auto Bracket - PC .230 .149 NIA
Auto Bracket - PA 616 .512 .219 NIA
Lens - PC .149 NIA NIA

* Significant at a 95% confidence level.

PC Tensile Bar Thickness PA 6/6 Tensile Bar Thickness

1
f.:: :7\
--·····-····n······-········•-····
0.1235 --------- 0.1245 ---------

i
0.124

� 0.122 -• .i ■ \. ■ ■ ■ ■
I
j■
I
0.1235

0.123
.;
......... •·
.......... ....... .................. .

0.1215 '--+---+---+---+--__;-----;__ 0.1225 '--+---+-�-+--+--+-->--+---+--+-'

1 2 3 4 5 8 7 II 8 10 1 2 3 4 5 8 7 II 8 10
a.,.- a.,.

!... Narnwlv Coaled ... �

Figure 33. PC Tensile Bar Figure 34. PA 616 Tensile Bar
Thickness. Thickness.

Quenched PP dimensions, smaller than one inch, were within .00 I inch of normally

cooled dimensions. Quenched PP part dimensions, larger than one inch, were not as
 75
close to the dimensions of normally cooled PP parts. However, since quenching

appeared to induce 90% of the air cooled PP shrinkage, final PP dimensions could be

predicted. The most important finding of this study was that part to part variance did

not appear to be practically effected by the quenching process.

This suggested the quenching process could be an effective tool to study

process variance during mold trials at the molding machine during mold trials. The

data from this study was subsequently used to develop a quench procedure for

collecting data during a designed experiment (Appendix D).

Development of a Standard Designed Experiment (DOE)

The short experiments previously discussed in this chapter, do not address the

three most critical variables identified in the PCIM shot-to-shot repeatability studies.

These variables include barrel temperature, mold temperature, and pack/hold

pressure. To objectively determine the settings of these variables, a statistically

designed experiment was developed as part of the mold trial procedure. The design

selected was a 23 full factorial, with a replicated center point (Box, Hunter, Hunter,

1978). Data gathered in the DOE was used to develop predictive models and test the

models for significant non-linearity. In addition, minimum significance values were

calculated to aid in determining which process variables and/or interactions

significantly effected each dependent variable measured. These data allowed

determination of variable settings which should provide desired dimensions and

reduced process variation. To aid in data collection and analysis, a commercial

 76
spreadsheet program, Quattro Pro Version 5.0, was utilized (Borland, 1993). The

spreadsheet program allowed input of measurement data during the experiment.

Once the required data was collected and entered into the spreadsheet, the

spreadsheet program analyzed the data and produced descriptive graphs to aid in

setting process variables (Appendix E). The data and graphs could also be printed

and added to permanent process documentation. Following development of the

standard DOE, instructions for performing the DOE and data entry were written.

Instruction Booklet

Following completion of all of the pieces of the procedure, an instruction

booklet was assembled (Appendix D). This consisted of bringing together all of the

instructions and guidelines written for the procedure. These instructions and

guidelines were arranged in a three ring binder for easy reference.

Computer Software Specification

Software selection was based primarily upon the statistical analysis

requirements. It was determined that standard linear regression analysis was

necessary. Quattro Pro Version 5.0 was selected as the spreadsheet package because

it offered the required capabilities and had worked well in earlier PCIM research.

Evaluation of Procedures

Once a rough draft of the procedure for determining fill speed, pack/hold
 77
time, back pressure, and screw speed was developed, it was reviewed at two of the

consortium companies. Portions of the procedure were reviewed by a process

engineer at each company. The review was conducted during actual mold trials. The

suggested experiments were performed by the process engineer strictly following the

documentation. During each experiment, comments and concerns of the process

engineer were recorded. Following the review, comments and concerns were used to

revise the procedure.

Mold Trial Case Studies

To evaluate performance of the new mold trial procedure developed, three

case study comparisons were conducted. These case study comparisons were used to

compare current mold trial methods with the new mold trial procedure. Comparisons

were made between part dimension means and variances for each procedure, and

with dimension tolerances for each part. In addition, comparisons were made

between the time resources required to perform the mold trials and documentation

gathered during the mold trial.

In each of the three case studies, substantially different machines, molds and

materials were used. The first case study was run on a 250 ton Toshiba injection

molding machine at Wright Plastic Products (Sheridian, Michigan). An eight cavity

clip mold was used. The clips were molded of Nylon 6 (Wellman 22LHI6XEN).

The second case study was run using an 85 ton Cincinnati Milicron injection molding

machine at Prince Corporation (Holland, Michigan). A single cavity trim plate mold
 78
was used. The trim plates were molded of polycarbonate (GE EM3110). The third

case study was run using a 390 ton Toshiba injection molding machine at Cascade

Engineering (Grand Rapids, Michigan). A single cavity isolator mold was used. The

isolators were molded of an impact modified polypropylene copolymer (Aristech

TI4120C).

To perform the new mold trial, a computer, spreadsheet software, and several

measurement instruments were required. The computer was a 486DX33 laptop

computer with 8 megabytes of RAM. The spreadsheet program, Quattro Pro Version

5.0, was used to calculate and graph viscosity curves, during the fill speed

experiment, and gate seal curves, during the pack/hold time experiment. The

spreadsheet program was also used to collect and analyze data during each designed

experiment (DOE). Mold surface temperatures were measured using an Omega

Handheld Microprocessor Digital Thermometer with a K type thermocouple surface

probe. Melt temperature was measured using an Omega Handheld Microprocessor

Digital Thermometer with a J type thermocouple melt probe. Part and/or shot

weights were measured using an Acculab electronic digital scale graduated to .01

grams. Part dimensions were measured using gage pins, Mitutoyo Digimatic digital

calipers, and measurement fixtures with Mitutoyo plunger gages.

In each case study, the process was first developed by a set-up person from

the host company. The set-up person developed and documented the process

according to the company's current mold trial procedure. When the set-up person

determined an acceptable process had been developed, the amount of time used to
 79
develop the process was recorded. In addition, process documentation from the mold

trial was collected. The machine was allowed to run at the determined process

settings for 30 minutes to stabilize the process based on prior PCIM research

(Bodfish, Dennison, Howard, & Engelmann, 1992). Following stabilization, 50

consecutive shots were collected to measure process mean and variance. This data

defined the mean and variance of the current procedure. Process settings were then

altered to return the process to an unstable state. The author then redeveloped the

process following the new mold trial procedure as outlined in Appendix D. Using

data gathered in the experiments, predictive models were developed. This provided

the means to identify and predict process settings which would deliver desired part

dimensions and/or minimum variance. After the process settings were established, a

mold surface temperature map was generated. In addition, a critical process data

sheet was filled out (Appendix D). At the completion of the new mold trial

procedure, the amount of time required to perform the mold trial was recorded.

Process documentation generated during the mold trial was also gathered. As a

limited test of the predictive model, a set of processing conditions were tested which

would be expected to center the process on the desired dimension. The machine was

run for 30 minutes at these desired process settings. After stabilization, 50

consecutive shots were collected to measure process mean and variance. These

results defined the mean (optimized) and variance (non-optimized) of the new

procedure.

Measurement data were tested using a Student's t-test executed at a 95%

 80
confidence level. The first t-tests was performed to determine whether the

difference in dimension means between procedures and the target dimension were

statistically significant. Additional t-tests were performed to determine whether the

difference in dimension means between procedure treatments were statistically

significant. A Levene test of homogeneity of the variances was also executed at a

95% confidence level. This analysis tested for differences in the variances between

procedure treatments. The DOE data were also reviewed to determine significance

of each effect in the predictive models, and to identify strategies for optimizing the

process. The amount of time required to perform each mold trial procedure was also

evaluated to determine the difference in time resources required. In addition,

documentation from the mold trials was compared, to supplement a determination of

the most thorough documentation.

Case Study #1 Process Results

In the first case study, performed at Wright Plastic Products, the target or

nominal dimension for the hole diameter of the part was .422 +/- .0025 inches (see

Table 4). Mean dimensions from both current and new procedures showed highly

significant differences from the target dimension (p<.001). The difference in mean

dimensions between procedure treatments were also highly significant (p<.001). The

mean dimension from all eight cavities using the current procedure was .417 inches

and the mean dimension using the new procedure was .419 inches (see Figures 35

and 36). Although the new procedure did not exactly center the process on the target,
 81
Table 4

Case Study #1 Target Dimensions, Means, and Variances

Target Dimension .422 +/- .005

Current Procedure Mean .417
New Procedure Mean .419
Current Procedure Variance .0000041
New Procedure Variance .0000049

Case Study #1 (Current Procedure)

Specifications: Lower Specif,cation Limit (LSL): .419500
Nominal Specification Limit: .422000
Upper Specihcation Limn ( USL): .424500
-3.s LSL NOMINAL +3.s USL
140

120

100

80

u: 60

40

20

0.41 0.412 0.414 0.416 0.418 0.42 0.422 0.424 0.426 0.428
Hole Diameter (Inches)

Figure 35. Case Study #1 Hole Diameter Distribution Using the

Current Mold Trial Procedure.

the new procedure was able to produce a mean dimension statistically closer to the

target dimension than the current procedure. The deviation of the new procedure

from the desired target was determined to be a result of problems in the assumptions

regarding the relationship between quenched and normally cooled parts.

 82

Case Study #1 (New Procedure)

Specifications: lower Specification limit (LSL): .419500
Nominal SpeciflCalion limit: .422000
Upper Specification limit (USL): .424500
-3.s LSL NOMINAL USL +3.s
130
120
110
100
90
>- 80
70

50
40
30
20
10
0
0.41 0.412 0.414 0.416 0.418 0.42 0.422 0.424 0.426 0.428
Hole Diameter (Inches)

Figure 36. Case Study #1 Hole Diameter Distribution Using the

New Mold Trial Procedure.

Improved centering of the process is related to the predictive power of the

DOE results. In the new mold trial procedure, predictive models were developed

using quenched dimensions. The data show the largest possible dimension, in the

processing window explored, was .423 inches (Appendix G). Referring to the earlier

quenching studies, it was determined quenched nylon part dimensions are

approximately .001 inches larger than normally cooled dimensions. Therefore, the

process was set to achieve a .423 quenched dimension, which it was assumed, would

result in a process which produced .422 inch average hole diameters.

Upon review of the mean hole diameter produced using the new procedure, it

was found the mean hole diameter was actually .419 inches. This meant the

quenched dimension was approximately .004 inches larger than normally cooled

dimensions. Accordingly, the scale from the predictive models for the hole diameter
 83
would need to be reduced by .004 inches, to accurately estimate normally cooled

dimensions.

A test of procedure variances indicated a significant difference between

procedure treatments (p=.006). The variance of the hole diameter using the current

procedure was .0000041, and .0000049 using the new procedure (see Table 4).

While these data show a small improvement in the variance using the current

procedure versus the new procedure, it must be stressed that during the new

procedure, centering the process, not reduction of the variance, was prioritized. Data

gathered using the new procedure did provide a means to reduce variation, which will

be addressed below. The Cavity-to-Cavity Differences graph, from the DOE data,

suggested a large portion of process variance observed was due to cavity-to-cavity

differences (Appendix G).

Further review of the DOE data revealed additional useful information

(results shown Appendix G). The Varl significance value, under the Dependent

Variable 1 Mean Effects graph, suggested only variable 1, barrel temperature, had a

statistically significant effect on the mean hole diameter (Appendix G). The

Dependent Variable l Variance Effects graph also shows only variable 1 significantly

effects hole diameter variance. Therefore, control of the hole diameter mean and

variation should focus primarily on barrel temperature. These data suggested a

reduction in barrel temperature would increase the hole diameter. In contrast, the

Dependent Variable l Variance Effects graph suggested variable 1, barrel

temperature, had a negative effect on hole diameter variation. Conversely, increasing

 84
barrel temperature reduces hole variation. This effect can also be observed by

reviewing the actual dimensions for each run of the DOE (Appendix G). Runs 6

through 10, high barrel temperature treatments, show a 50% reduction in variation

over runs set at lower barrel temperatures.

After adjusting the predictive models by .004 inches, these data suggest the

target dimension cannot be achieved in the process window explored by the DOE.

This strongly suggests a change in the cavity dimension to achieve the target

dimension. Due to the fact increasing the barrel temperature would reduce hole

diameter variation, the cavity dimension should be modified to take advantage of this

response. Because variable 2, mold temperature, has virtually no effect on the hole

diameter, this variable can be set at any point in the range used in the DOE. It could

for example be set at conditions which would reduce process costs. Variable 3 is

very nearly significant and should be considered along with variable 1 when

determining the modification required to the cavity. A review of the Interaction xlx3

Dependent Variable 1 Mean graph can be used to develop a useful combination of

settings (Appendix G). This graph suggested if variable 3 is set at the high level

(40%), and variable 1 is raised to the high level (530°F), the quenched hole diameter

would be .419 inches, which should produce a .415 inch normally cooled hole

diameter (Appendix G). Therefore, to achieve the target dimension of .422 inches

and reduce variation, the diameter of the core pins should be increased by .007

inches. If possible, the print dimension could also be changed to .415 inches. The

new procedure allowed process adjustment decisions to be made without numerous

 85
additional mold trials. At most, two additional mold trials may be required. First, to

verify the predictions before modifying the cavity. Second, to verify the cavity

modifications.

Case Study #2 Process Results

In the second case study, performed at Prince Corporation, the target or

nominal dimension for the part width was 3.165 +/- .010 inches (see Table 5). Mean

dimensions from both current and new procedures showed highly significant

differences from the target dimension (p<.001). The difference in mean dimension

Table 5

Case Study #2 Target Dimensions, Means, and Variances

Target Dimension 3.165 +/- .010

Current Procedure Mean 3.167
New Procedure Mean 3.166
Current Procedure Variance .00000025
New Procedure Variance .00000006

between procedure treatments was also highly significant (p<.001). The mean

dimension using the current procedure was 3.167 inches and the mean dimension

using the new procedure was 3.166 inches (see Figures 37 and 38). These data show

the new mold trial procedure produced a mean dimension statistically closer to the

target dimension than the current procedure, during this initial mold trial.
 86

Case Study lf2 (Current Procedure)

SpeciflCaliona: Lo,wr SpeciflCalion Limit (LSL): 3.15500
Nominal SpeciflCalion Limit: 3.16500
Upper Specification Limit (USL): 3.17500
LSL NOMINAL -3.s +3.s USL

40 .. ................ ................ ...............

35 .. ........................... .........................

r: : :::: ::: : : : : :: : :: :: : : :
30 ········•··········· ·

or 15 .. ....................
10 .........................

5 .. .. .......... ............ .. ····•···•·· ·······... ·.......

---"""--""'""----------.J....J
������������;§�����!i�i!i��!f!f���E���s����
0 '-'------------.......

MMMMMMMMMMM MMMMMMMMMMMMMMMMMMM MMMMMMMMMMM

Width (lnchN)

Figure 37. Case Study #2 Width Distribution Using the Current

Mold Trial Procedure.

Case Study #2 (New Procedure)

Specifications: L<MW Specification Limit (LSL): 3.15500
Nominal Specification Limit: 3.16500

55.---------------------------,
Upper Specification Limit (USL): 3.17500
LSL NOMINAL -3.s +3.s USL

50 �·+· .......... ......... . I

45 ........................... : :
; ;
40 �+................ ... ...................................... ..................... Hf½!l......,.............................................................................J ... J

.· 1
......... ...... ............... ,.,,�.. ...,:....................................................... ........................ ..j
35

··· ·h:
>- 30 :
� 25
:
,
... ............................................................................... ,...
� 20 .. .. ......... .......
15 ................................ ............................. '!'�
10 ............. ........ .. . ....................... .. ... f·: ·····•····..···..·..········•· ...········· ..··•··········· .......................--i, .. J
:
........._
� L.. .. ..._ ...._
....._
..._ ...._•._
...._ ...._
...._
...._
...._ ...._
..•._._ .. ._
.._ .•.._
...•_
..._ . ...............1;......._
... .._ ... ..........___________..__,
..

����������l��§������i�i!f��!i�l��EEB�s�r���
M MMM MM MM MM M MM MM MMMMM M MM MMMM MMM MM MMMM MMM M M
Width (ll"ICMS)

Figure 38. Case Study #2 Width Distribution Using the New Mold
Trial Procedure.

The predictive models generated in the new mold trial procedure allowed the

process to be set closer to the target dimension during this initial mold trial. Based
 87
on the earlier quenching studies, the process was set to produce a quenched

dimension of 3.166 inches, which it was assumed, would produce a normally cooled

dimension of 3.165 inches.

Upon further review of the mean part width produced using the new

procedure, it was determined the quenched part widths were equal to normally cooled

part widths. Therefore, no scale adjustments were required on the predictive models.

A test of variances also indicated a highly significant difference between

procedure treatments (p<.001). The variance of the width dimension using the

current procedure was .00000025 and .00000006 using the new procedure (see Table

5). These data suggest a 75% reduction in variance from this preliminary setting

using the new mold trial procedure.

A review of the Dependent Variable 1 Mean Effects graph suggest only

variables 2 and 3, mold temperature and pack/hold pressure, significantly effect the

part width mean (Appendix I). In addition, the Dependent Variable 1 Variance

Effects graph shows variable 1, barrel temperature, is the only variable significantly

effecting the width variance (Appendix I).

To center the process mean, mold temperature and pack/hold pressure,

variables 2 and 3, are the key variables. A review of the Interaction x2x3 Dependent

Variable 1 Mean graph showed that by setting variable 3 at the low setting (500 PSI),

and variable 2 at approximately 80% of the way between the low and midpoint

setting (120°F), a 3 .165 width dimension should be achieved (Appendix I). An

alternate method to achieve a 3.165 inch part width, taken from the same graph,
 88
°
would be to set variable 2 at the low setting (100 F), and variable 3 at the midpoint of

the DOE range (600PSI). The most appropriate strategy from these two could be

decided based on their impact on processing cost. A review of the Dependent

Variable 1 Variance Effect graph shows variable 1 having a negative effect on part

width variation (Appendix I). Therefore, variable 1 should be set at the high setting

used in the DOE, to reduce variation. These predictive models allow the process

adjustment decisions to be determined without numerous additional mold trials. The

settings would only need to be quickly verified in a subsequent mold trial.

Case Study #3 Process Results

In case study #3, performed at Cascade Engineering, the target or nominal

dimension for the part hole diameter of the part was 3.822 +/- .040 inches (see Table

6). Mean dimensions for both current and new procedures showed highly significant

differences from the target dimensions (p<.001). The difference in mean dimension

between procedure treatments was also significant (p=.015). The mean dimension

using the current procedure was 3.785 inches and the mean dimension using the new

procedure was 3.790 inches (see Figures 39 and 40). The new procedure was able to

produce a mean dimension statistically closer to the target dimension as compared to

the current procedure.

The improvement in process centering using the new procedure is related to

the predictive ability of the DOE results. Using the predictive models, a quenched
 89
Table 6

Case Study #3 Target Dimension, Means, and Variances

Target Dimension 3.822 +/- .040

Current Procedure Mean 3.785
New Procedure Mean 3.790
Current Procedure Variance .000107
New Procedure Variance .000081

Case Study #3 (Current Procedure)

Specilicalions: Lower Specification Limit (LSL): 3.78200
Nominal Specification Limit: 3.82200
Upper Specification Limit (USL): 3.86200

-3.s LSL +3.s NOMINAL USL

22
20
18
16

,.. 14
12
i° 10
it
8
6
4
2

3.76 3.77 3.78 3.79 3.8 3.81 3.82 3.83 3.84 3.85 3.86 3.87
Hole Diameter (lnchN)

Figure 39. Case Study #3 Hole Diameter Distribution Using the

Current Mold Trial Procedure.

dimension of 3.830 inches was determined to be the necessary processing target

during the mold trial (Appendix K). This value was selected with the assumption

normally cooled dimensions would be an average of .008 inches smaller than

quenched dimensions. This assumption was based on the quenching study performed
 90
Case Study #3 (New Procedure)
Specifications: Lower Specification Limit (LSL): 3.78200
Nominal Specification Limit:3.82200
Upper Specification Limit (USL): 3.86200
-3.s LSL +3.s NOMINAL USL
26�---------------------�
24
22 •····················;.................
20•·······•···························
18•·····•·····•························
lj' 16
ai 14 �····················'.··· ······················
g" 12 �···················,·· ·······················
it 10 •······················ ...................../
8 •····················S··
6 � ...................•..................,
4 �·····•·············'··········
2

3.76 3.77 3.78 3.79 3.8 3.81 3.82 3.83 3.84 3.85 3.86 3.87
Hole Diameter (Inches)

Figure 40. Case Study #3 Hole Diameter Distribution Using the

New Mold Trial Procedure.

earlier in this project.

After reviewing the mean hole diameter produced using the new procedure, it

was determined this assumption was not valid. The normally cooled part was

actually .040 inches smaller than the quenched dimensions. In this case study, use of

quenched dimensions to predict final dimensions was not as useful as it was for PC

and PA. However, the actual bias (.040 inches) was identified between quenched and

normally cooled dimensions for this part and material. The predictive models could

still be used to predict the part dimension, by adjusting the scales .040 inches.

A test of variances show an insignificant difference between procedure

treatments (p=.081 ). The variance of the hole diameter using the current procedure

was .000107, and .000081 using the new procedure (see Table 6).

Further review of the DOE data revealed additional information which could
 91
be used for process improvement (Appendix K). The Dependent Variable 1 Mean

Effects graph suggested only variable 1, barrel temperature, had a statistically

significant effect on the mean hole diameter (Appendix K). The Dependent Variable

1 Variance Effects graph suggested variable 3, pack/hold pressure, had twice the

effect on variance as variables 1 and 2, barrel temperature and mold temperature

(Appendix K). After adjusting the scale on the predictive models by the bias

correction factor of .040 inches, these data suggested the target dimension of 3.822

inches could not be achieved in the process window explored by the DOE (Appendix

K). These data strongly suggested a modification was required to the cavity to

achieve the target dimension. Due to the fact variables 1,2, and 3 all have a large

negative effect on variance, these variables should all be set at the higher level used

in the DOE. In addition, setting variable 1 at the high level reduced variation and

also increased the hole diameter (440°F). To determine the dimension achievable

from setting all of the variables at the high setting, a review of both Interaction xlx2

and Interaction xlx3 Dependent Variable 1 Means graphs is necessary (Appendix K).

Following adjustments of the scales by the bias factor of .040 inches, these graphs

suggested that if all the variables are set high, a 3.785 inch hole diameter is

achievable. Therefore, to achieve the target dimension and reduce variation, the

process variables should remain at the high settings and the diameter of the cavity

should be increased by .037 inches. If possible, the print dimension could also be

modified to 3. 785 inches.

 92
Case Study Mold Trial Time Resources

Mold trials on single cavity molds required an average of 1.5 hours to perform

using current procedures and 5 hours using the new procedures. The mold trial using

an eight cavity mold required 1.75 hours to perform using current procedures and 6

hours using the new procedure. The actual time taken to perform the mold trial is

dependent on the number of cavities in the mold. Molds with multiple cavities

require more samples to be taken during the DOE and therefore require more time to

perform. The new mold trial procedure takes approximately 3 times the amount of

time required to perform the initial mold trial using current procedures. However,

the current procedure yields less than optimum process settings. Using current

procedures, numerous additional mold trials will inevitably be performed to continue

searching for process settings capable of centering the process and reducing

variation. The large number of mold trials required using current procedures is

clearly evident in the data from the review of current mold trial techniques. Using

current procedures, these data suggested an average of 7 mold trials are performed to

develop a capable process. Conversely, the new procedure provides significantly

more information than current procedures during the initial mold trial and provides a

clear strategy for optimizing the process mean and variance. Therefore, using the

new procedure could result in as few as two mold trials to develop a centered process

with low variation. The first mold trial would be used to define the process, followed

by one additional mold trial to verify the predictive models.

 93
Case Study Process Documentation

Objective evaluation of the process documentation was a difficult task. The

documentation required during a mold trial is often dependent on the company's

definition of thorough documentation. In all case studies, basic machine variable

setting sheets were used. In two of the case studies, current mold trial procedures

produced mold water diagrams, mold surface temperature documentation, and gate

seal graphs. In one case study, current mold trial procedures produced a viscosity

curve as part of routine documentation (Appendix F, H, and J). Documentation from

the new mold trial procedure includes all of the aforementioned items. In addition,

graphs were produced to aid in determination of barrel temperature, mold

temperature, and pack/hold pressure settings capable of centering the process and

reducing variation (Appendix G, I, and K). With multi-cavity molds, the new mold

trial procedure also produced graphs quantifying cavity-to-cavity differences

(Appendix G). The additional documentation produced using the new procedure

provided an excellent process improvement and monitoring tool.

Case Study Conclusions

In all three case studies, the process was initially centered more accurately

using the new mold trial procedure (see Table 7). However, the real power of the

new procedure lay in the subsequent ability to optimize the process mean and

variance. In one of the three case studies, process variation was reduced 50% using
 94
the new mold trial procedure. In the other two case studies, the variances were either

insignificantly different or slightly different between treatments (see Table 7).

However, in all of the case studies, the new procedure provided data to aid in

developing a strategy for further centering the process and reduction of variation.

The current procedure provided no such information or capability.

Table 7

Case Study Summary of Target Dimensions, Means, and Variances

Case Study #1 Case Study #2 Case Study #3

Target Dimension .422 +/- .005 3.165 +/- .010 3.822 +/- .040
Current Procedure Mean .417 3.167 3.785
New Procedure Mean .419 3.166 3.790
Current Procedure Variance .0000041 .00000025 .000107
New Procedure Variance .0000049 .00000006 .000081

During mold trials using current procedures, process settings were determined

by subjective review of the parts. This method did not provide the set-up person with

any indication whether or not part dimensions were near the target dimension.

Proper centering appeared to be largely a matter of serendipity. Current procedures

also had no capacity to measure part variation, or show the effect of process variables

on variation. Therefore, additional mold trials are inevitably required to continue

searching for a properly centered process with low variation. There is no expectation

that a true optimum can be reached using current procedures.

 95
Better centering of the mean dimension results from using objective data to

make process setting decisions. The new procedure provided objective data

generated from the DOE. These data provided an efficient and accurate means of

centering the process on the target dimension and reducing variation. The objective

data gathered in the new mold trial procedure provided immediate feedback on the

effect of process variables on means and variances. Due to feedback acquired during

the mold trial, significantly more information was gathered in a single mold trial.

The additional information provided a method for determining a strategy to center the

• process and reduce variance without numerous additional mold trials. In most cases,

only one additional mold trial would be required to verify the process improvement

strategy developed using the predictive models.

In these case studies, the new mold trial procedure required more time

resources than current procedures (see Table 8). However, according to the

questionnaire used in the earlier study of current mold trial procedures among

consortium members, average mold trials require 5.2 hours to complete. This

Table 8

Hours Required to Perform Mold Trials in the Case Studies

Case Study #1 Case Study #2 Case Study #3

Current Procedure 1.75 Hours 1.5 Hours 1.42 Hours
New Procedure 6 Hours 5 Hours 5 Hours

average corresponds well with the 5.3 hour average required to perform the new
 96
mold trial procedure in these case studies. Although current procedures required 1/3

the time to perform, in these case studies, additional mold trials would definitely be

required to continue searching for process settings capable of centering the process

and reducing variation. According to data from the study of current mold trial

techniques, an average of 7 mold trials are required to-establish a capable process.

On the contrary, using the new mold trial procedure should result in a single mold

trial to define the process, followed by one additional mold trial to verify the

predictive models. Or, if the predictive models suggest a cavity steel modification,

four mold trials may be required. An example of this scenario is shown below.

1st Mold Trial: used to define the process (a cavity modification is required

according to the predictive models).

2nd Mold Trial: verify predictive models and potential cavity modification

prior to cutting steel.

3rd Mold Trial: after modifying the cavity, reevaluate the accuracy of the

predictive models.

4th Mold Trial: verify the new process settings determined by the predictive

models.

Documentation produced using the new mold trial procedure was found to be

more thorough than documentation produced using current mold trial procedures.

The thoroughness of documentation results from the large data set produced using the

DOE. Thorough documentation produced in the new mold trial procedure provided

identification of key process variables and their effects on part attributes. This type
 97
of information provided an excellent tool for developing a process improvement

strategy. In addition, thorough documentation of material viscosity, gate seal time,

and mold surface temperatures, provided a means of detecting process shifts after the

mold is released to production. Thorough documentation provides a useful tool for

process troubleshooting and monitoring. It is important to remember that in order to

determine when and why a process shifts, one must identify and document the

process center, normal process variation, and the significant variables effecting the

process.
 CHAPTER V

CONCLUSIONS AND RECOMMENDATIONS

Conclusions

The part quenching process, developed in this study, proved useful in

estimating final part dimensions at the molding machine during mold trials.

Quenched part dimensions worked well in predicting normally cooled part

dimensions using ABS, PC, and PA. However, quenched part dimensions did not

work as well for predicting normally cooled part dimensions using PP. Though the

estimate of the initial PP bias was poor at the time of the trial, normally cooled parts

could be saved during the mold trial and measured at a later time. These

measurements could then be correlated with quenched dimensions. The difference

between quenched and normally cooled part dimensions can then be determined and

the predictive models can be adjusted accordingly.

The data from this study suggest apptication of the new mold trial procedure

can significantly improve the mold trial process. The new mold trial procedure

utilized actual process and part dimension data to objectively determine process

settings. The new mold trial procedure implemented experimental techniques to

systematically evaluate the effect of process variables on part dimensions. In all

98
 99
three case studies, the process was initially centered more accurately using the new

mold trial procedure (see Table 9). Although reduction of variance was not a priority

in the initial use of the predictive model, some favorable response in process

variation was observed. In one of the three case studies, initial process variance was

reduced 50% using the new mold trial procedure. In the other two case studies, the

variances were slightly different or insignificantly different between procedure

treatments (see Table 9). Even though in two cases the variances were not

substantially reduced, the new procedure did provide mean and variance effects

Table 9

Case Study Summary of Target Dimensions, Means, and Variances

Case Study #1 Case Study #2 Case Study #3

Target Dimension .422 +/- .005 3.165 +/- .010 3.822 +/- .040
Current Procedure Mean .417 3.167 3.785
New Procedure Mean .419 3.166 3.790
Current Procedure Variance .0000041 .00000025 .000107
New Procedure Variance .0000049 .00000006 .000081

graphs, predictive models, and cavity-to-cavity difference evaluation, from which a

process improvement strategy could be developed (Appendix E). This scientific

approach to data collection allows mold trial time to be used more efficiently. This

information allows the process settings to be determined without performing

numerous additional mold trials. Without the predictive models, additional mold of
 100
trials would need to be perfonned to continue searching for process settings capable

centering and reducing variation in the part dimension.

The case studies showed the new mold trial procedure required approximately

5.3 hours to perfonn, compared to 1.5 hours required for current procedures.

However, following the initial mold trial, the new procedure had more accurately

centered the process and had defined the effect of each process variable studied on

the dependent part attributes. The data gathered in a single 5 to 6 hour mold trial

using the new procedure, has the potential to eliminate many of the current mold

trials required to continue randomly searching for a capable process. The large

number of mold trials required using current procedures is clearly evident in the data

from the review of current mold trial techniques. Using current procedures, these

data suggested an average of 7 mold trials are perfonned to develop a capable

process. Conversely, using the new procedure should result in a single mold trial to

define the process, followed by one additional mold trial to verify the predictive

models. Or, if the predictive models suggest a cavity steel modification, four mold

trials may be required. An example of this scenario is shown below.

1st Mold Trial: used to define the process (a cavity modification is required

according to the predictive models).

2nd Mold Trial: verify predictive models and potential cavity modification

prior to cutting steel.

3rd Mold Trial: after modifying the cavity, reevaluate the accuracy of the

predictive models.
 101
4th Mold Trial: verify the new process settings determined by the predictive

model.

Economically, additional mold trials are quite expensive and therefore

undesirable. Mold trials require machine time, mold preparation time, material

consumption, and personnel time, to name a few. Based upon this data and

observations over the last 6 months of this research, it is reasonable to project the

number of mold trials, directed at developing a capable process, could be reduced by

as much as 50%. Economically, the new mold trial procedure can be very cost

effective (see Table 10). In should be noted, the entire new procedure only needs to

be performed once. After using the new procedure to define the process, the mold

trials should be shorter and cost less then mold trials using the current procedures.

Table 10

Total Mold Trial Cost Comparison

Costs Current Procedure New Procedure

Machine Time $75 (1.5 hours at $50/hr) $265 (5.3 hours at $50/hr)
Personnel $38 (1.5 hours at $25/hr) $133 (5.3 hours at $25/hr)
Material $45 (30 lb at $1.50/lb) $120 (80 lb at $1.50/lb)
Total Cost for Initial Mold Trial $157 $518
Cost Per Mold Set 7@$75 4@$75
Cost for Additional Trials 6@$157 3@$157
Total Process Development Cost $1624 $1289

Therefore, if 7 mold trials were required using the current procedure, the total cost to
 102
develop a capable process would be $1624. If only 4 mold trials were required using

the new procedure, the total cost to develop a capable process would be $1289.

Therefore, the new mold trial procedure demonstrates the potential for a reduction in

mold trial cost. In addition, using the new mold trial procedure to thoroughly define

the process, provides the potential to achieve a significant improvement in process

capability.

Recommendations

1. The new mold trial procedure is capable of gathering large amounts of

information in a relatively short time period. However, all portions of the procedure

may not be required for every mold. Simple molds, without critical dimensions, may

warrant using the procedure without the DOE. The determination of which portions

to utilize in which situations is the responsibility of each individual company.

2. The entire mold trial procedure does not need to be performed during a

single mold trial. For instance, the procedure, without the DOE, may be performed

during the initial mold trials. Once the mold functional problems have been

corrected, the DOE may then be performed during a subsequent mold trial. However,

the DOE must never be performed prior to optimization of fill speed, pack/hold time,

back pressure, and screw speed. These variables must be optimized prior to

performing the DOE to reduce the opportunity for these variables to significantly

effect the process.

3. Continued research in the area of part quenching is recommended.

 103
Optimization of the quenching process could improve the predictability of normally

cooled dimensions. Also, quench studies should be performed using addition

materials.

4. Further investigation of part surface temperature mapping is required to

determine the best applications for it to be used effectively.

5. Data should be collected to evaluate long term effects of applying the new

mold trial procedure versus current procedures.

 Appendix A

Interview Questions

104
 105
Interview Questions:

1) What do you consider to be a simple, average, and complex mold? Please use
example parts or describe the characteristics.

2) On the average, for simple, average and complex molds, how many times is a
mold trial performed prior to releasing the mold for production?

3) What is the average time spent, for simple average and complex molds, per
mold trial?

4) How many people are involved in the actual process of performing a mold
trial?

5) Who does the mold trials (job function)?

6) Does the same person always perform the mold trial on an individual mold?
If not, how many different people perform the mold trials on a mold?

7) What sort of standard documentation is filed regarding the process set-up?

8) Describe for me, in as much detail as possible, the procedures followed by the
set-up personnel when performing a new mold trial.

9) How do you determine which parts are saved?

10) How many parts are saved for measurement and evaluation following the
mold trial?

11) What is the criteria for an acceptable part?

12) Who determines if the parts are acceptable?

 Appendix B

PCIM Mold Trial Questionnaire

106
 107

Premier Class Injection Molding (PCIM)

Mold Trial Questionnaire
This questionnaire is designed to collect information regarding the mold trial process
used by your company. While filling out this form keep in mind the questions are
focused on mold trials performed to test the function of the mold and to define the
process parameter settings. This does not include mold trials used to produce parts
for mock builds or customer requests.

Company: ____________ Date:--------

1) Which of the following are part of your job function? (Check all that apply)

□ Engineering management 0 Production supervision

0 Machine and process set-up □ Project or program engineering
□ Mold engineering □ Quality inspection
□ Mold setting □ Research and development
□ Process development □ Other:--------
□ Process monitoring □ Other:--------

2) What is your job title?

□ Manufacturing Engineer □ Project Engineer

□ Mold Setter □ Quality Engineer
□ Process Engineer □ Quality Inspector
□ Process/Mold Technician □ Tooling Engineer
□ Production Supervisor D Other:-------

3) How many years have you worked in the injection molding industry?

□ Less than 1 □ 1 □ 2 □ 3 □ 4 □ 5 □ 6 □ 7 □ 8 □ 9 0 1 O+

4) On the average, how many mold trials are performed to establish the process
settings and test the function of the mold prior to releasing it to production?

□ 1 □ 2 □ 3 □ 4 □ 5 □ 6 □ 7 □ 8 □ 9 □ 10 □ 11 □ 12 □ 13 □ 14 □ 15+
 108
5) On the average, how many hours are spent per mold trial?

D Less than 1 D 1 D 2 D 3 D 4 D 5 D 6 D 7 D 8 D 9 D 1 O+

6) How many people are present and involved in performing an average mold trial?

D 1 D 2 D 3 D 4 D 5 D 6+

7) On the average, which individuals are present while actually performing a mold
trial? (Check all that apply)

D Manufacturing Engineer D Project Engineer

D Mold Setter D Quality Engineer
D Process Engineer D Quality Inspector
D Process/Mold Technician D Tooling Engineer
D Production Supervisor D Other: -------

8) For a given mold, does the same individual perform the initial mold trial and
subsequent mold trials?

D Always D Usually D Sometimes D Rarely D Never

9) Who makes the final determination whether parts are acceptable during the mold
trial? (If more than one check all that apply)

D Manufacturing Engineer D Project Engineer

D Mold Setter D Quality Engineer
D Process Engineer D Quality Inspector
D Process/Mold Technician D Tooling Engineer
D Production Supervisor D Other:

10) As standard practice, what criteria is used to define an acceptable part during
the mold trial process? (Check all that apply)

D Part Appearance
D Part Dimensions
D Part Weight
D Other:-------------------
□ Other:-------------------
 109
11) On the average, how many shots are saved during the mold trial to be used for
measurement and evaluation following the mold trial? (Fill in the blank)

12) As a standard practice, what documentation of the process is gathered during

the mold trial? (Check all that apply)

□ Cavity pressure traces □ Mold water diagram

□ Cavity-to-cavity evaluations □ On-machine rheology curve
□ Gate seal plot □ Part temperature map
□ Mold water flow rates □ Process settings sheet
□ Mold temperature map □ Process window diagrams
□ Other:-------------------
□ Other:-------------------
□ Other:-------------------
 Appendix C

Mold Trial Preparation Guidelines

110
 111

Mold Trial Guidelines

Environment Monitoring:
The humidity and temperature of the environment during the mold trial can greatly
affect the process. Therefore, it is recommended that the humidity and temperature
be monitored and recorded during the mold trial. Thes_e data can be used to
understand the effects environmental conditions have on the process.

Mold Preparation:
Consistency in preparation of the mold is a critical step in the mold trial and
production process. The manner in which the mold is hung in the press and plumbed
must remain unchanged between the mold trials and production. The following
guidelines should be closely followed while developing the process and similarly in
the production environment.

• Orientation of the mold in the machine should be the same every time. This
means the top of the mold should be identified and marked with the word
"TOP". This will help to insure the top is ALWAYS the top when it is hung in
the molding machine. Changing the way the mold is hung in the machine can
change the way the parts eject and occasionally, the way the parts fill.

• The nozzle type, radius and orifice size should always be checked and recorded
to be sure the same tip is used each time the mold is run.

• Consistent water hookup is important to produce uniform cooling across the

mold.

- To initially determine the best water hookup, obtain a copy of the mold
drawings showing the water circuits. Using the mold drawing, one can
determine which circuits in the mold are shorter than others and they can be
connected. This will allow the circuits in the mold to all be closer to the same
length. Similarly, this means that the hoses going to the mold from the water
supply should all be approximately the same length. This is done to equalize the
distance the water travels in all circuits.

- Implement the fewest cooling circuits that will provide turbulent flow and less
than a 5-degree F change in temperature between the inlet and outlet.
Remember, this must be measured to be sure the proper conditions have been
 112
met.

- Sufficient water flow is required to achieve efficient cooling in the water

circuits. Water flow must be high enough to achieve turbulent flow. Without
turbulent flow, the water does a poor job of extracting the heat from the mold.
The amount of water necessary to promote turbulent flow is dependent on the
diameter of the cooling line and the temperature of the cooling fluid being used.
It is important to check and record the flow through each circuit during the mold
trial process. Good initial documentation makes it possible to verify that the
flow rates are the same during each run. The following chart provides the
gallons per minute (GPM) necessary to achieve turbulent flow using water as the
cooling fluid.

Flow Rates to Achieve Turbulent Flow (GPM)

p·1pe s·1ze 1/D 140 ° F 100 ° F 50 ° F 32 ° F
1/8 5/16 .25 .50 .75 1.25
1/4 7/16 .40 .70 1.25 1.50
3/8 9/16 .50 .75 1.50 2.00
1/2 11/16 .60 1.00 2.00 2.50
Table Courtesy ofRJG and Associates

Note: Twice the flow rate is required when using 50/50 ethylene glycol/water mix.

- Be sure to always measure the water flow and pressure through each circuit in
the mold. An efficient cooling circuit should measure less than a 5 degree F rise
in coolant temperature and a 5 PSI drop in coolant pressure across each cooling
circuit.- The mold should be marked showing the inlet and outlet for each
internal water circuit. This is often marked for example "#1 IN" and "#1 OUT",
"#2 IN'" and "#2 OUT".

- When plumbing the mold, be mindful of water drips that result from removing
the water lines. To aid in resolving this problem, whenever possible place the
circuit connection hoses on the top of the mold. Also, have the input and output
lines from the source coming in the bottom or sides of the mold. This will help
prevent water from dripping into the cavities while the mold is being removed
from the machine.

- After plumbing the mold for the first time, create a detailed diagram showing
the manner in which the mold is plumbed. The diagram should show the
 113
location of all water lines on the mold. The diagram should also show the hose
length used in each location. This will provide a guide to insure the water is
hooked up the same way each time.

- The electrical connections on the mold should be located on the top whenever
possible. Locating them on the top of the mold wil1 help prevent electrical
shorts resulting from water leaks.

Material Preparation:
• If colorants are added to a natural material be sure the proper amount of colorant
is applied. Also make sure the natural material and colorant are thoroughly
mixed to avoid color shifts.

• If the material requires drying, be sure to follow the recommended drying

specifications provided by the resin supplier. To insure the material is properly
dried before use, measure the moisture level in the material using a moisture
analyzer. The moisture level of the material should also be tested periodically
during mold trials and production to determine the consistency of the moisture
content.

• Drying is also recommended when using a vented barrel system to insure all of
the moisture is removed before processing. If the material is not pre-dried and is
processed in a vented barrel, there is no way to determine what percentage of
moisture the vent is removing.
 AppendixD

Experiment Instructions

114
 115

Suggested Mold Trial Procedure Order

1) Gather material process data sheets

2) Assure proper water flow in the mold

3) Assure material is sufficiently dried (if required)

4) Develop a "generic" process at the required cycle time

5) Perform back pressure experiment

6) Perform fill speed experiment

7) Perform pack/hold time experiment

8) Perform screw speed experiment

9) Establish DOE settings

10) Perform DOE

11) Set process variables based on the experimental data

12) Determine part fill weight

13) Create a mold temperature map

14) Fill out the Critical Data sheet and machine set-up sheet
 116
Before Beginning Optimization Experiments
• Before performing any of the optimization experiments, develop a "generic"
process capable of producing full parts at the desired cycle time. The following
is a checklist to help setup a "generic".process.

□ Barrel temperature set at mid-point of material spec sheet

□ Mold temperature set at mid-point of material spec sheet

□ Back pressure set at mid:..point of material spec sheet

□ Screw speed set at mid-point of material spec sheet

□ Filling the mold at a "reasonable rate"

□ Enough pack/hold pressure and time to produce full parts

□ Mold running equal to or below quoted cycle time

□ Mold running a continuous and constant cycle

□ Maintaining a cushion

□ Enough cooling time to eject the part without deformation

• For the best results, perform the experiments in the order they appear. Or, for
instance, if the back pressure experiment is not needed, start with the fill speed
experiment then perform the pack/hold time experiment and finally the screw
speed experiment. If the fill speed is significantly changed folloWing the
pack/hold time experiment, it is advised that the pack/hold time experiment be
performed again using the new fill speed.
 117

Establishing Back Pressure:

1) Set the barrel temperature profile to the midpoint recommended by the material
supplier with the appropriate profile.

2) Set the back pressure at zero or as close to this as possible.

3) Set the screw RPM's to half of the machine's maximum available screw speed.

4) Rotate the screw and monitor the inherent back pressure or resistance of the
material on a pressure gage or the controller screen.

5) This back pressure value is a good indicator of the natural resistance of the
material. Therefore, to achieve consistent mixing the back pressure should be
set slightly higher than this value. For example, if the inherent back pressure
reads 88 psi, set the back pressure to around 100 psi.
 118
Establishing Fill Speed:
1) Set the machine injection pressure to the maximum allowable pressure (usually
around 2000 psi).

2) Set the pack and hold pressures to 0 psi (on machines not capable of 0 psi, set to
1 psi)

3) Set the maximum injection time to 10 seconds.

4) Determine the intensification ratio of the machine being used. The

intensification ratio is the ratio of the cross sectional area of the injection
hydraulic cylinder versus the cross sectional area of the check ring or screw.

5) Set the shot size and cutoff position to fill the mold approximately 95% full
during the injection stage while maintaining a small cushion.

6) Cycle the press with normal mold and melt temperatures for approximately 10
shots.

7) Slowly increase the fill speed while monitoring the parts to be sure the mold is
only filling to about 95% capacity. As the increased fill speed produces larger
parts, adjust the cutoff position be maintain a 95% full part.

8) When maximum fill speed is reached, adjust the cutoff position to achieve a part
that is approximately 95% filled. The maximum fill speed may be governed by
either machine ability or mold limitations such as severe bums or flash.

9) Record the fill speed setting, fill time, and peak injection pressure.

10) Reduce the fill speed about 20% and let the mold cycle 3 times.

11) Again record the fill speed setting, fill time and peak injection pressure.

12) Repeat steps 9 and 10 until at least 10 data points have been collected.

13) To product a well defined curve, 10 to 12 data points should be collected.

14) Record the melt and mold temperatures currently being run.

15) Calculate the relative viscosity and shear rate for each data point using the
worksheet on the next page or use the viscosity curve spreadsheet.
 119
16) Plot relative viscosity versus shear rate to produce the viscosity curve for the
mold.
 120

Viscosity Curve Data Sheet

Plastic Pressure Relative Viscosity

Fill Fill Injection (Inj. Pressure (Plastic
Shot Speed Time (t) Pressure x Intensification) lit Pressure x t)
1
2
3
4

5
6

7
8
9

10
11
12

Shear Rate (1/t)

 121
Interrupting the Viscosity Curve

The following are key things to remember when reviewing the viscosity curve to
determine the optimum fill speed:

• The optimum fill speed should be somewhere in the flat area to the right of the
bend in the curve (see Figure 1). The fill speed set in the flat area of the curve
will reduce fill time variation effects on material viscosity. Ideally the part
should be filled as fast as possible under a controllable speed. However, due to
gate blush and other molding faults the fill speed may need to be reduced to a
speed on the left end of the flat to eliminate such molding faults. Remember, this
experiment determines the fill speed to be used (in/sec) not the fill time. Using
the fill speed (in/sec) determined, adjust the shot size and cut-off position to
produce a· 95% full part.

Figure 1.

• When filling the part at high speeds, be careful to watch for the point when a
reduction in fill speed produces the same fill time and injection pressure as the
previous setting. If this condition is noticed, it means the machine cannot fill the
part at the last speed setting. At this speed setting the machine is not filling at a
controlled rate. Be aware that if this occurs and the injection pressure reading is
not at the maximum of the machine, it is often the lack of ability of the machine
to build full pressure in a short length of time. See the highlighted data in Table 1
for an example of this problem. In this case, the fastest controllable speed is
somewhere between 4.0 in/sec and 3.5 in/sec. Be sure the fastest fill speed is at a
controlled rate.
 122
Table 1

Fill Speed Viscosity Curve Data Example

Cycle Fill Speed (in/sec) Fill Time (sec) Injection Pressure

1 5.0 .21 1894
2 4.5 .22 1897
3 4.0 .21 1890
4 3.5 .37 1862
5 3.0 .59 1789
 123
Establishing Pack/Hold Time:
1) Run the mold with the proper fill speed under normal melt temperature and mold
temperature conditions.

2) Set the pack/hold pressure high enough to produce full visually acceptable parts.

3) Set with the pack/hold time at 3 seconds and add the time removed from the
pack/hold time to the cooling time so the cycle remains constant. Run 2 shots,
weigh the parts from the shots, and record the average weight. For example, if
pack/hold time was 10 seconds, set it to 3 seconds and add 7 seconds to the
cooling time.

4) Increase the pack/hold time to 4 seconds, collect 2 more shots, weigh the parts
from the shots, and record the average weight. As the pack/hold time is
increased, reduce the cooling time to keep the cycle time constant.

5) Continue to increase the pack/hold time in 1 second intervals and collect part
weight data. When the part weight is unchanged by an increase in pack/hold time
the gate has sealed. Add approximately 2 second to this time to establish the gate
seal time or the pack/hold time. Adding 2 seconds is to insure gate seal under
slightly different conditions.

6) Record the melt and mold temperatures used.

 124
Establishing Screw Speed:
1) Set the barrel temperature profile to the midpoint recommended by the material
supplier with the appropriate profile. Also, set the back pressure to the value
determined by the back pressure experiment.

2) Set the screw RPM's to half of the machine's maximum available screw speed.

3) With the machine in cycle, adjust the screw speed to a speed that can recover in
about two seconds less than the required cooling time.

4) Once the screw speed has been established, back off the barrel, purge, and check
the melt temperature. This is to be sure the screw speed is not over shearing the
material and creating a significant amount of additional heat in the melt.
 125
Determining Fill Weight:
To help with the transfer of a mold from one machine to another, the fill weight
should be determined. The fill weight is the weight of the part(s) at about 95% full.
Fill weight provides a target for determining the transfer position on a different
machine therefore, allowing the part to be filled to the same point before applying
pack/hold pressure. The following procedure is for determining the fill weight.

1) After the process has been established, set the pack/hold pressure to O psi, or to 1
psi, cycle the mold. If the process is setup correctly, this should produce parts
that are about 95% full.

2) Weigh all of the sample parts together, minus the runner, to determine the
average fill weight.

3) Record the fill weight on the Temperature Map/Water Diagram/Critical Data

sheet.
 126
Mold Surface Temperature Mapping:
1) Create a sketch of the surface of each side of the mold showing the cavities,
cavity numbers, and runner system.

2) Determine multiple locations on and around the cavity and core to measure
surface temperatures. Be sure to select enough locations to thoroughly map the
mold surface.

3) After the mold has run at least 50 shots during the mold trial, interrupt the cycle
and probe one of the locations from the mold temperature map. After the
measurement has been taken, place the machine back in cycle and record the
temperature. It is important that the measurement be taken quickly so the change
in mold temperature is minimized.

4) Let the mold cycle 2 shots and repeat step 3 measuring a different location on the
mold surface.

5) Continue step 4 until all locations on the mold surface map have been measured.

6) Randomly measure each location a total of three times using the same procedure
and determine the average for each location. Using the average of three
measurements reduces the effect of measurement error.

7) Record the average mold temperature for each location on the mold surface
temperature map.

8) Review the mold surface temperature map to be sure the cooling is uniform. If
the temperature across the mold is non-uniform, reevaluate the way the cooling
circuits are hooked up to determine if something can be changed to improve the
cooling. If this is unsuccessful, use the supply valves to adjust the flow of each
circuit to equalize the surface temperature. If supply valves are used to even out
the surface temperature, the design of the cooling channels in the mold should be
reevaluated and permanently adjusted to prevent set-up mistakes during
production.

Note: When using a temperature probe always take the reading after it has
stabilized.

The following page shows an example of a mold surface temperature map.

 127
Example of a mold surface temperature map:

Mold Surface Temperature Map

160F 157F
171F 170F

173 F
170F

167F
167
�

169F 171F
161F 170F

Cavity Half Core Half

 128
Mold Surface Temperature Map/Water Diagram/Critical
Data
Mold Number: ----- Date:------

Material:------- Lot#: ------ Nozzle Type: ____

Nozzle Orifice:----- Nozzle Radius: Melt Temp: _____

Fill Weight: ______ Part Weight:--- Fill Speed(in/sec):__

Gate Seal Time:-----

Cavity Half (Moving Half) Core Half (Stationary Half)

Water Temp: Water Temp:

NOTES:
 129

Part Surface Temperature Mapping:

Part surface temperature mapping proves useful when the mold surface temperature
can not be effectively measured. Ineffective measurement is often the result of the
cooling system significantly changing the surface before it can be probed between
cycles. Often this happens when running chilled water or large molds, which are
difficult to quickly reach into and measure. Due to the thermal conductivity of
plastics, the heat stays in the part longer than in the mold steel therefore, making it
easier to measure. If a mold surface temperature map· has already been created, a part
surface temperature map is normally not necessary.

I) Create a sketch of the part.

2) Determine location(s) to measure temperature on the part surface of each cavity.

These locations should correspond with critical locations on the mold surface. Be
sure to select enough locations to thoroughly map the mold temperature
distribution.

3) After the mold has run at least 50 shots, probe one of the locations from the part
temperature map. To improve measurement consistency, when the part falls from
the mold count to five before probing the part surface. Doing this for each part
measured keeps the time prior to measuring consistent.

4) Select the next shot and repeat step 3 measuring a different location on the part
surface. It is important to hold the probe firmly to the part surface to accurately
measure the part temperature. In addition, the way the part and probe are held
while measuring must not vary from measurement to measurement. Variations in
part and probe handling can cause large measurement error.

5) Continue step 4 until each of the surface map locations have been randomly
measured three times. Using the average of three measurements reduces the
effect of measurement error.

6) Record the average part temperature for each location on the part surface
temperature map.

Note: When using a temperature probe always take readings after it has stabilized.

The documentation is the same as that used for mold surface temperature
measurement, except the part surface temperatures are recorded on the map rather
than the actual mold surface temperatures.
 130
Part Surface Temperature Map/Water Diagram/Critical
Data
Mold Number:----- Date:-----

Material:------- Lot#: ----- Nozzle Type:____

Nozzle Orifice: ----- Nozzle Radius:-- Melt Temp:_____

Fill Weight ------ Part Weight___ Fill Speed(in/sec):__

Gate Seal Time: -----

Cavity Half (Moving Half) Core Half (Stationary Half)

Water Temp: Water Temp:

NOTES:
 131
Establishing Process Settings for the DOE
The goal of this process is to establish high, middle, and low variable settings for the
DOE that will produce parts that are visually acceptable at both extremes
(underpacked, overpacked).

1) Determine high and low barrel temperatures, mold cooling temperatures and
pack/hold pressures to attempt to use for the experiment. These initial high and
low settings should be determined through previous trials of the mold, personal
experience, and material suppliers process data sheets.

2) Set the process to produce parts at the target cycle time using the predetermined
fill speed, pack/hold time, back pressure and screw speed.

3) Set the barrel temperature at the high barrel temperature setting determined in
step 1.

4) Set the mold cooling temperature and pack/hold pressure to the high settings.
These process settings often represent the conditions under which the mold will
be easiest to fill. Cycle the mold at these settings to be sure the part is not flashed
and is able to run a continuous cycle. If a flashed part is produced, decrease one
or more of the variables slightly until a non-flashed part is produced. When
making slight adjustments, adjust each of the three variables slightly as opposed
to a large adjustment in one variable.

5) Keeping the barrel temperature at the high setting, cycle the mold at the other
experimental combinations of the high and low mold cooling temperatures and
pack/hold pressures. This step is to be sure the mold will produce full parts and
run a continuous cycle.

6) Reduce the barrel temperature to the low setting.

7) Set the mold cooling temperature and the pack/hold pressure to the low settings.
At these settings, be sure a full part is produced (may have slight sink) and the
process is capable of running a continuous cycle. If the parts are not full or the
cycle can not run continuously, increase one or more of the variables slightly until
acceptable parts are produced.

8) Keeping the barrel temperatures at the low setting, cycle the mold at the other
experimental combinations of high and low mold cooling temperatures and
pack/hold pressures. Again, this step is to be sure the mold will produce full parts
and run a continuous cycle.
 132
** Be sure that at all of the parameter combinations, at least an 1/e inch cushion
is maintained.

9) Record the high, midpoint, and low variable settings in the experimental design
matrix provided on the next page. Negative ones(-1) are the low setting, positive
ones (1) are the high settings and zeros (0) are the midpoint of the high and low
settings.
 133
DOE Matrix
Run Variable 1 Variable 2 Variable 3
(Barrel Temperature) (Mold Temperature) (Pack/Hold Pressure)

1 -1 +1 +1
2 -1 -1 -1
3 -1 -1 +1
4 -1 +1 -1
5 0 0 0
6 +1 -1 -1
7 +1 +1 +1
8 +1 +1 -1
9 +1 -1 +1
1 O(Replicate 0 0 0
of run 5)

Notes:
 134
Quench Water
The water used for part quenching should be standard tap water. Water should be
placed in a container large enough to submerge the entire part. A 5 gallon plastic
bucket is sufficient for most parts. The temperature ofthe water used should be
between 60 °F to 75°F. The water temperature should be periodically checked
throughout the experiment to be sure the temperature has not changed more than 5°F.
Ifthe temperature changes by more than 5°F, add cooler water to lower the
temperature back to the original temperature.

Starting the DOE Analysis Program

1) To start the DOE analysis program, enter Quattro Pro(version 5.0 or higher).

2) Open the file called ANALYSIS.WBl, from the A: drive or C: drive, ifloaded.

3) With the mouse, click the start button on the title page to execute the program.

4) Enter the high and low variable settings in the table, then click the large button.

5) Enter the date and tool number in the space provided on the DATA page.

Performing the DOE

Follow the steps and guidelines below to perform the DOE.

1) Set the three experimental variables to the settings indicated in the design matrix.

2) Allow the mold to cycle until the temperatures appear relatively stable (at least 2
hours ifthe machine used was started cold).

3) As each shot is molded, clip any excess vestige off ofthe sprue and weight the
shot.

4) Enter the shot weight into the spreadsheet DATA page in the Shot Wt column
corresponding with the proper shot number.

5) As the shot weight data is entered, it will be graphed on the right side ofthe
screen.
 135
6) Continue steps 3 and 4 until
the data on the graph shows no
slope trends. The weights may
vary shot-to-shot, but the
overall trend must show no
slope (see Figure 2).

7) Once the graph data shows no GOOD(No Slope) BAD(Slope)

slope, begin collecting
samples.
Figure 2. Graphing Examples
Sampling Guidelines:
• Gather a minimum of 1 sample from each cavity.

• To review cavity-to-cavity differences gather a minimum of 2 samples per cavity.

• Gather a minimum of 6 samples total.

Examples: To review cavity differences using a 4 cavity mold, collect 2

measurements from each cavity making a total of 8 samples.
Or collect 1 measurement from each cavity and two additional
measurements selected randomly from the 4 cavities, for a
total of 6 measurements.

Using a 2 cavity mold, collect 3 measurements from each

cavity making a total of 6 samples.

Using a 1 cavity mold, collect 6 measurements from the same

cavity making a total of 6 samples.

8) Clip any excess vestige off of the sprue and weigh the shot.

9) Enter the shot weight into the spreadsheet DATA page in the Shot Wt column
(see Figure 3).

10) Randomly select one of the parts and clip it from the runner system. Be sure you
keep track of the cavities measured to be sure all the cavities are sampled at least
twice.
 136

Shot Run Keep Shot Wt Dep Var 1 Dep Var 2 Cavity

Date: 6/21/95
Tool#: TL1234
1 1 1 21.82 0.417 1
2 1 1 21.88 0.415 1
3 1 1 21.88 0.416 1
4 1 1 21.89 0.417 1
5 1 1 21.90 0.416 1
6 1 1 21.88 0.416 1
7 21.89
8 21.58
9 21.60
10 21.57

Figure 3. Sample of DOE Software Data Entry Page

11) If part weight is measured, weigh the part and enter the value in the Dep Var 2
column of the spreadsheet (see Figure 3).

12) Place the part in the quench water fully submerged for the recommended quench
time (see Table 2).

Table 2

Part Thickness Quench Times

Part Thickness (in) Quench Time (sec)

.0625 10
.125 20
.1875 30
.250 40

13) Remove the part from the water and quickly wipe the excess water from the part.

14) Quickly take the desired measurement.

 137
15) Enter the value into the spreadsheet in the Dep Var 3 column, or if part weight
is not recorded, use the Dep Var 2 column (see Figure 3).

16) Enter a 1 in the Keep column corresponding to the part sampled (see Figure 3).

17) Enter the run number (1-10) in the Run column corresponding to the part
sampled (see Figure 3).

18) Enter the cavity number in the Cavity column corresponding to the part sampled
(see Figure 3).

Note: The quenching, measurement, and data entry will often take more time than
one cycle, but the mold must run continuously during the measurement.

19) Clear the extra parts and wait for the next molded part.

20) Repeat steps 8 through 19 until all of the samples required for the run have been
collected.

21) Once all the samples have been collected, leaving the mold cycling, make the
process setting changes according to the design matrix.

22) When the settings have been changed and double checked, begin with step 3
again measuring and entering shot weight into the spreadsheet program. Do not
put a 1 in the Keep column or a number in the Run column corresponding to
these values (see Figure 3).

Note: Mold temperature changes usually require about 5 to 10 minutes to stabilize.

Barrel temperature changes usually require about 20 to 30 minutes to stabilize.
Pack/Hold pressure changes usually require about 2 to 3 minutes to stabilize.
However, use the shot weight graphs to determine when the process has stabilized.

Continue steps 3 through 22 until all of the runs have been completed in the design
matrix.
 138

Graph Data Interpretation

To aid in interpreting the data gathered in the DOE, graphs are generated in the
DOE spreadsheet program. The graphs provide information on mean effects and
variance effects for shot weight and two additional dependent variables (determined
by the experimenter). In addition graphs are generated to aid in determining the
process variable setpoints to achieve desired dimensions or weight characteristics

Mean and Variance Effects Graphs

Mean and variance effects

graphs show the effect each process
variable and interactions between Shot Weight Mean Effects
the process variables have on the
part. These graphs are used in
determining which variables have
the most significant effect on both
part characteristic means and part
-.., 0.4

ffi 0.2
characteristic variations. In
0
addition, the graphs show the
direction of the effect (positive or -0.2 1=+====+====;::::=:==.::====i=====i::::l
x1 x2 x1x2 x3 x1x3 x2x3
negative). For example, Figure 4 Variable
shows the mean effects for shot
weight. Variable 1 appears to have
the largest effect on the shot Figure 4. Shot Weight Mean Effects
weight. Variable I has
approximately twice the effect than
do variables 2 and 3. The Dependent Variable 1
interaction effects show minimal Mean Effects
effect on the shot weight. 0.6
On each mean and variance 0.4
effects graph appears two � 0.2
significance lines across the graph. Ill
0
If the effect bar of a process -0.2
variable extends past one of the -0.4
significance lines, the variable is -0.6
x1 x2 x1x2 x3 x1x3 x2x3
statistically significant. For Variable
example, the significance lines for
the data in Figure 4 suggests only
variable 1 would be considered Figure 5. Dependent Variable 1 Mean·
statistically significant. Effects
The graph in Figure 4 reveals
 139
that variable 1 is the most significant for controlling shot weight. Also, variable 1 is
shown having a positive effect on the shot weight. Therefore, an increase to the
setting of variable 1 would increase the shot weight. If the effect were negative, an
increase in variable 1 would decrease the shot weight.
In contrast, the graph in Figure 5, shows all of the variables having
approximately the same effect on the dependent variable 1. Dependent variable 1
could be part length, width, hole diameter, or any other measurable part dimension.
This graph suggests that none of the variables are more important than the others or
significant for controlling part length.
The graph of variance effects
for shot weight, shown in Figure 6,
indicates that process variable 2 Shot Weight Variance Effects
has the most effect on variance of
the shot weight. Also, variable 2
has a negative effect. Therefore, to
help reduce variation in shot 0
weight, the setting of variable 2 ,:;
� -0.5
should be increased. If the effect w
of variable 2 were positive, -1
lowering the variable 2 setting
-1.5 �----<---<------+>-----+�
would reduce variation in the shot x1 x2 x1x2 x3 x1x3 x2x3
weight. Variable

Mean and variance effects

graphs for dependent variables 1 Figure 6. Shot Weight Variance Effects
and 2 can be interpreted in the
same way. This information is produced to help determine the critical variable(s)
and their effect on part characteristics.

Data Curvature Evaluation

To aid in determining the degree of curvature in the linear model, a curvature

evaluation table is generated in the DOE spreadsheet. The table provides a minimum
value and a curvature value for each dependent variable measured. If the curvature
value is less than the minimum value, the curvature of the data is not statistically
significant and is linear. If the curvature value is larger than the minimum value, the
curvature of the data is statistically significant. If the curvature of the data is
significant, the linear regression model will only be accurate at the end points of the
linear model. Therefore, any points along the graph lines will be inaccurate.

Process Variable and Interaction Graphs

The process variable and interactions graphs are used to determine the setpoints
 140
of the process variables to achieve the desired part characteristic. These graphs are to
be used in conjunction with the mean effects graphs.
A review of effects graphs -------------------.
will establish which of the
Interaction 1 &3
process variables or interactions
are significant. If an interaction ... Dependent Variable 1 Mean
� 1.57 ,------ -,--------::,4
is significant, focus on the
corresponding interaction i 1.56
� 1.55
graph. For example, if the � 1.54
interaction between process g_ 1.53
variable 1 and 3 is significant � 1.5 2 �--+-----1-------+-------j
.30
for dependent variable 1(part -1 -0.5 0 0.5
Variable 1
width) and process variable 2 is
relatively insignificant, focus \•Variable 3(-) *Variable 3(+) I
would be placed on the
interaction graph (see Figure 7). Figure 7. Interaction 1&3 for Dependent
If the target width dimension Variable 1 Mean
was 1.53 inches, set the process
to achieve a width of 1.54
inches on the graph.
Remember, graph data Dependent Variable 2
generated during the mold trial Means
1.32 .---------------,
is produced using quenched
� 1.3
dimensions. This means the 2i
i1.2a

. .•. . .
scale on the left side of the �1.26
graphs must be adjusted i1.24 ...
.

accordingly. For example, if §.1.22

the data for the graph in Figure � 1.2
7 was generated for a nylon 1.18 �-------+---+------4---+--�
-1 -0.5 0 0.5
\•Variable 1 •Variable 2•Variable 3I
part, the target dimension on the
graph should be approximately
1.54 to achieve a normally
cooled dimension of 1.53 Figure 8. Dependent Variable 2 Means
inches (see the Quench Study in
chapter 4). Therefore, process variable 1 would be set at the low setting used in the
experiment and process variable 3 would be set 30% of the way between the
midpoint and high setting used in the experiment. For example, if the midpoint was
400 PSI and the high was 450 PSI, set the process variable at 415 PSI. In addition,
process variable 2 would be set at the midpoint of the range used in the experiment.
If none of the interactions and only one of the variables are significant, focus on
the corresponding mean graph. For example using the graph in Figure 8, if process
variable 2 is the significant variable and the target dimension is 1.25 inches, process
variable 2 should be set 50% of the way between the midpoint and high setting used
 141
in the experiment. Due the insignificant effect on the dimension, process variables 1
and 3 can be set anywhere in the range used in the experiment as long as it does not
increase process variation.
The further explain data interpretation, the following example steps through
actual experimental data. This experimental data was generated molding ABS parts.
The width of the part, Dependent Variable 1, was declared the critical dimension.
The target width of the part is .405 inches. From the quench studies performed in
earlier PCIM research, we know that ABS quenched dimensions are approximately
.001-.002 inches larger than normally cooled dimensions. Therefore the target
quenched dimension would be approximately .407 inches.
The first data to review is the Dependent Variable 1 Mean Effects graph shown
in Figure 9, and the Varl Significant value below the graph. This graph shows only
process variable 2 and 3 are statistically significant. Variable 1 and the interactions

Dependent Variable 1 Dependent Variable 1

Mean Effects Variance Effects
0.025 3E-07 r------------,
0.02 2E-07+---------
0.015
1E-07
1j
ffi
0.01
0.005
i 0
-1E-07
0
-0.005 -2E-07
-0.01 -3E-07 +----+---+---+--+--+----I
x1 x2 x1x2 x3 x1x3 x2x3 x1 x2 x1x2 x3 x1x3 x2x3
Variable Variables

Figure 9. Dependent Variable 1 Figure 10. Dependent Variable 1

Mean Effects Variance Effects

Varl Significant .00291414

have an insignificant effect on the Dependent Variable 1 means in this experiment.

This allows focus to be placed on variables 2 and 3 to shift the mean width value.
The Dependent Variable 1 Variance Effects graph, Figure 10, shows variable 1
to have the largest effect on width variation. Therefore, variable 1 should be set to
reduce width variation.
Before interpreting the graphs to determine the process settings, the Curvature
Evaluation table must be reviewed (see Table 3). The curvature evaluation for the
width dimension data, VAR 1, shows the curvature value is less than the minimum
value. This means the curvature in the model is statistically insignificant. Because
the curvature is insignificant, the graphs can be used to predict the part width.
 142
Table 3

Curvature Evaluation Table

SWT VARl VAR2

Curvature .04 .00 ERR
Minimum .12 .01 ERR

Variable 1 does not significantly effect the mean width and therefore can be set
anywhere in the range used in the DOE. This can also be seen by the horizontal
nature of the variable 1 line in Figure 11. Figure 11 shows the effect each variable
has on dependent variable 1, assuming the other two variables are set at their mid­
points. It is known that variable 1 has a negative effect on the width variance.
Therefore, increasing variable 1 to the high setting used in the experiment should

Dependent Variable 1 Interaction x2x3

Means ... Dependent Variable 1 Mean

I"
0.44 �-- ------�

c--;r--=·- -+------=
...
0.-46
£ 0.43 +---------------< 0
i --·-::_·
lmapay. otVlrillll<3Mid-1)0inl
.,
j 0.42 +--,-==-----�=---------i > .44

c=·-·=· =- = - =· -=---±::::
> 0.42 -=-- - ;;;.;;··- :::.:-···· ....
�0.41
0
----
-g 0.4 t------:J1r"°-------===------1
======-J.
�-------+---
.4
!.0 38
£!l
io.39 +--,_...c...----------i
o 0.38
-1 --0.5 0 0.5
---< . -1 �.5 0 0.5

Level
Variable 2

I• Variable 1 _,_ Variable 2 • Variable 3j

Figure 11. Dependent Variable 1 Figure 12. Interaction of Variables 2

Means and 3 for Dependent
Variable 1 Means

reduce width variation.

Because variables 2 and 3 are most important for effecting the mean width
dimension, refer to Interaction xlx2 Dependent Variable 1 Mean graph (see Figure
12). The target quenched dimension of .406 inches can be achieved by setting
variable 3 at the mid-point and variable 2 at 50% of the way between the mid-point
and the low setting used in the experiment.
Or, if the target quenched dimension was .425 inches, variable 3 would be set at
the high setting and variable 2 would be set at 60% of the way between the mid-point
 143
and the high setting (see Figure 12).
 AppendixE

Spreadsheet ProgramExamples

144
 145
Spreadsheet Program Example

Date: 6/12195 Melt Temp: 500

Tool Number: Mold Temp: 120

Viscosity Curve
lntesification Ratio = 101

Shot Fill Speed Fill Time (tl . lni. Pressure Relative Viscosity 1/t
1 6 0.76 1997 15177 1.32
2 5 0.83 1998 16583 1.20
3 3.5 0.93 1983 18442 1.08
4 3 1.02 1843 18799 0.98
5 2.5 1.2 1674 20088 0.83
6 2 1.45 1516 21982 0.69
7 1.5 1.87 1350 25245 0.53
8 1 2.73 1199 32733 0.37
9 0.5 5.33 1090 58097 0.19
10 0.3 8.78 1107 97195 0.11

Viscosity Curve
100000
i .\

:z,;.
·u; 80000 �
\
.. . . ..
.

>
(/) 60000 l ..
;
I
i

---�
.:? 40000 T ·
ro I ..._________
20000-:- . . . ·········· .. . . .

O'
0 0.2 0.4 0.6 0.8 1.2 1.4
1/t Reciprocal Seconds
 146
Spreadsheet Program Example

Date: 6/12/95 Melt Temp: 500

Tool Number: Mold Temp: 120

Gate Seal Curve

Pack/H 0Id Ti1me a
PrtWe1g,
" ht
4 38.67
5 38.72
6 38.79
7 38.84
8 38.89
9 38.92
10 38.93
11 38.93
12
13
14
15
16
17
18
19
20

Gate Seal Curve

38.95 -, ---------------,
T ......---
38.9 f
t

t;
-
"5>38.85
"Q)
'
S: 38.8 .i.
r
-e
� 38.75 '.
•
38.7 .!..........
! .
/
38.65 _____,____________
0 5 10 15 20
Pack/Hold Time
 147
Spreadsheet Program Example

Shots 1-100
Shot Run Keeo Shot Wt Dep Var 1 Deo Var 2 Cavitv
Date: 6/21/95
22.1 , ----------,

22 �.
,--
ri . . . . . . . . . . . . . . . . . . . . . . . . 1
Tool#: TL1234
1 21.82 0.417
21.9L ... lL ,\ !
2 21.88 0.415 218 Li ! l . . . . · · · · · IL.':\- I
I . .. . .. .... . ...... J J
L
21.88 211 � 1
3
4 21.89 0.417 21eLLc::::::J.. �., -· : � .ii
�--
"'

,,
5 21.90 0.416 2.:S 1 12 23 34 45 56 66 TT 88 99 1
! __ ·•
6 21.88 0.416
7 21.89 0.414
8 21.58 I
9 21.60 Shots 101-200
I
10 21.57 22�-------�i
11 21.59 I I r'l i
12 21.61 1 21.95 f r!
+ "
I

13 21.62 I I
I 2ut/
14 21.62 1
r. . . . . .
1

15 21.62 .L(:.''' ..... .... ..... ........................1

16 21.62
21.85

I , !
17 21.63 21 8
· 101 110 119 128 137 1-46 155 16"4 173 182
18 21.62
19 21.63
20 21.63
21 2 21.64 0.402 Shots 201-300
22 2 21.62 0.400
23 21.64 0.400 I I
1()(),---------�
2
24 21.65 0.400
80 �-.
2
25 2 21.64 0.400 60 f. · · ..... .. ........ ....... .. ·
26 2 21.65 0.399 .a L......
27 21.65
28 22.06
20 t .. ........................
I
I
29 22.05 0'

30 3 22.05 0.428
201 211 221 231 241 251 261 271 281 291

31 3 22.06 0.428
32 3 22.05 0.428
I
33 3 22.07 0.429 Shots 301-400
34 3 22.06 0.427
I
35 3 22.08 0.428
100 �: --------�: I

36 21.65
......!

80;!
80 ,....... .................. .

37 21.64
38 21.65 I

39 21.64
40 -
I

40 21.65 20'
41 21.62 0I i

42 21.61
J01 311 321 331 341 351 361 371 381 391

43 21.62
44 21.62
45 21.58
46 21.60
47 21.59
48 21.58
49 21.59
 148
Spreadsheet Program Example

so 21.59
51 4 21.59 0.395
52 4 21.56 0.394
53 4 21.58 0.395
54 4 21.59 0.395
55 4 21.57 0.396
56 4 21.57 0.396
57 21.58
58 21.59
59 21.59
60 21.59
61 21.59
61 21.6
62 21.58
63 21.59
64 5 21.59 0.401
65 5 21.6 0.401
66 5 21.6 0.4
67 5 21.58 0.401
68 5 21.59 0.4
69 5 21.60 0.400
70 21.65
71 21.66
72 21.64
73 21.63
74 21.67
75 21.65
76 21.66
77 21.67
78 21.64
79 21.66
80 21.65
81 6 21.63 0.404
82 6 21.68 0.404
83 6 21.67 0.405
84 6 21.66 0.403
85 6 21.67 0.405
86 6 21.67 0.405
87 21.94
88 21.91
89 21.89
90 21.88
91 21.86
92 21.85
93 21.83
94 21.83
95 21.83
96 21.81
97 7 21.81 0.413
98 7 21.79 0.414
99 7 21.79 0.413
100 7 21.80 0.412
101 7 21.79 0.414
102 7 21.80 0.413
 149
Spreadsheet Program Example

103 21.80
104 21.55
105 21.55
106 21.55
107 8 1 21.55 0.395
108 8 .• 1 21.54 0.396
109 8 1 21.53 0.396
110 8 1 21.54 0.395
111 8 1 21.55 0.396
112 8 1 21.54 0.397
113 21.84
114 21.86
115 21.88
116 21.91
117 21.92
118 21.91
119 21.94
120 21.95
121 21.96
122 21.96
123 21.96
124 21.96
125 21.96
126 21.96
127 9 21.98 0.425
128 9 21.98 0.425
129 9 21.98 0.425
130 9 21.99 0.424
131 9 21.99 0.426
132 9 21.99 0.425
133 21.85
134 21.85
1 :35 21.83
136 21.86
137 21.86
138 21.87
139 21.86
140 10 21.87 0.416
141 10 21.87 0.417
142 10 21.87 0.416
143 10 21.88 0.417
144 10 21.86 0.417
145 10 21.88 0.415
146
147
148
149
150
151
152
153
154
155
156
 150
Spreadsheet Program Example

Shot Weight Mean Effects Shot Weight Variance Effects

0.0003 ,--- -------�
0.4 .,....--- - -- - --- --
0.0002 I
0.3
1 0.0001 L.
_ 0.2 r····
� 0.1 J·
... .!
I ii o,L
w
:c
W O I -0.0001 !°

-0.1 �...... -0.0002 °[

1
-0.0003 - ---,------- --
x1 x2 x1X2 x3 x1x3 x2x3 x1 x2 x1x2 x3 x1x3 x2x3
Variable Variables

-·--- !
SWT Significant: 0.048932918
Dependent Variable 1 Dependent Variable 1
o.025 ·-_M__
e_ � � ect�s �- --� Varia nce Eff e cts
JE-O? .
an Eff

0.02 I 2E-07 <·························· ·············

0.015 i... ...................! 1E-07
0.01 1 ..
, I
u . . ..... ..............!
ii 0
� 0.005 I �
-1E--07 i·
-0.005 -2E-07 I··
-0.01 I -3E--07 �- - -------�
x1 x2 x1x2 x3 x1x3 x2x3 x1 x2 x1x2 x3 x1x3 x2x3
1

Variable Variables

Var1 Si nificant: 0.00291414

Dependent Variable 2 Dependent Variable 2
100 - ---'-M"-'C..C"-C..C"-''-"c.=.cc'--- --­
'
ea n Effects
I 100 .,... __ an c_
v_a _ ri_
_ e_E� ff_ e_ct_ s___
�

80 ;

t; 60 ; · I I
I
I !
80 :·
I

60 ,.
�
40 w 40 ·

20 20
I
_
! __________-' I 0 �· --- - ---- ---
x1 x2 x1x2 x3 x1x3 x2x3 x1 x2 x1x2 x3 x1x3 x2x3
0

Variable i Variabfes

Var2 Significant: ERR

 151
Spreadsheet Program Example

1�1i=:�7�i
Shot Weight Dependent Variable 1
� 044
. I ·
Means I
:,1l0.43 l ······ ·· ······· ········· ···· · � I
·
·!!042l ec:::· / I
� o 41 l . • =:--:-------...c::::: • .. I
��:: ✓ ····· · · · · · · · · · · · · · · · · · · · · · · · · · •· 1 lo::i:L:=1
0
o:3a !.__________�
-1 -0.5 0 0.5 -1 -0.5 0 0.5
Level Level
i• Variable 1 • Variable 2 • Variable 3 I , ... Variable 1 • Variable 2 • Variable 3 I

7 Dependent Variable 2
Means
100
I"'
so 1
I . ....... I
Curvature Evaluation
I� I
i! soi
SWT Var1 Var2
I
Curvature 0.04 0.00 ERR
40 Minimum 0.12 0.01 ERR
I� C: I;
·· ·· · · ·· · ·
2:�1-··
I
·· · · · ··· · · ··
_· _ _ ·_ I
I! -+-
--1
----- - _
-0_.5
__.....o__,__ o_.5
_
_ � �

I . __
i• _
Vanable 1 • Jaria�e 2 • Variable 3
_�
I
I

Interaction x1x2 Interaction x2x3

22.4 .,'

r�=-=2 ,
J
22 ,
Shot Weigh� Mean ,
I I I
Shot Weigh� Mean
•
:!: 21.9 1· ......... ........ ...........
22
I i
���:L . ·
�
.. . . . . . .. !

-�21.8 f-··· ··········· ··········1·-·· · ···· ·······1

· · I

r�- i j ..
I
� 21.7 1......... · · ·· ...........
· · · ··· ·· . 1
ili21.6 ........ . i. . j
.;. rn 21.4 !... . - - - · ·· · ·· · - ·· ·····:· ··
---� I
21.5 '--------'-----� 21.2
-1 --0.5 0
Vanable 1
0.5 1 i -1 -0.5 0
Vanable 2
0.5 1 I
! • Variable 2(-) * Variable 2(•)1 1 • Variable 3(-) .,. Variable 3(•)1

Interaction x1x3
22.2 �hot Weight1 Mean

i
- 22 ! ......... .... ..... ..................,1..... .

i! 21.a i ..... . ...... lI

� 21.61····· ···· ·········i
·!
2 1. 4 �======!:=====� I

-0.5 0 0.5
Variable 1
-1

: • Variable 3(-) • Variable 3(•,!

 152
Spreadsheet Program Example

Interaction x1x2 I Interaction x2x3

Dependent Variable 1 Mean
I Dependent Variable 1 Mean

· ··r
:[!��H
'; 0. 42 -======;======w 0.45

l 0 415
t .. ·i I -;;
rr .......... . ...... ... . ...... _ji ..... .. .... ... ... .
·
:: 0.41
�0. 405
I;
}o3�: l.. . .. 1 � . I � 0.38 I , I
-1 --0.5 0 0.5 I -1 --0.5 o o.5 , I
Variable 1 Variable 2
I

1
-variable 2(-) .,.. Variable 2(+)I
I i• Variable3(-) .,.. Variable3(•)!

Interaction x1x3
Dependent Variable 1 Mean
�-----------<
: 0.44
� 0.4:i
� 0.42
t ... .... .. ....
t ... . ... . . . ·+ j·· · ·
•
04 · ··
�c: () � 1••··· .. . .... .. . J
X, 0.39 ....
1
l. ............ ............... · �
..............
� 038 I i
-1 --0.5 0 0.5 1 I
Variable 1 I
!
!•Variable3(-) •Variable3(•)! !
Interaction x1x2 Interaction x2x3
Dependent Variable 2 Mean Dependent Variable 2 Mean
"'100 ·�-----,-- ----- �100,------------�

� soir ... . ....... . ..···· ·····•··r•• i.

� ....... .... .. .. ... .. 1
· · ···· · ····· ·· · · Ii
..1.1 •..
! l··
·� 00 · -g 80 i--··
······ ·····1
0
�40
I 6
�40,
[ 20
I I I 20 -
x_ 1
� o�----�-----�I I � a�·--------
I ;

-1 --0.5 0 0.5 1 ! _, -0.5 0 0. 5 1 I

Variable 1 Variable 2

i • Variable 2(-) • Variable 2(+) I i • Variable3(-) • Variable3(+)I

Interaction x1x3
Dependent Variable 2 Mean
� 100.------�----�
� soi• .j ·!
� so 1- l- ·I
' ]l 40 i . i .. ·1
[ 20 1 · j
� 0·�----�-----�
-1 --0.5 0 0.5
Variable 1

I• Variable3(-) •Vanable3(+)/
 Appendix F

Case Study #1 Mold and Process Documentation

Current Procedure

153
MOLD NUMBEH 'l'HOUBLESHOO'rI NG
OA'l'E 6-72-fS--
MACHINE NUMBEH :53
NOZZLE 'l'EMPERA'l'UHE t:;L/0
ZONE 1 TEMPERATURE 5.t-6 n
ZONE 2 'l'EMPERA'l'URE 6Z-0 �
ZONE 3 TEMPERATURE
(I)

6(0
INJEC'rION HIGH PHESSUHE
-�
PACK PRESSUHE ;!,&7, ......
=l:i::
HOLD PRESSURE ,5P££!J f!-&�
CLAMP IIIGII PRESSURE 11'2 n
CLAMP LOW PRESSURE J;rJ�
/.c;� s.
(I)

------
BACK PRESSURE
""O
INJECTION SPEED 1 �0/4
INJECTION SPEED 2 (")
(I)
0..

--
INJECTION SPEED 3
INJECTION SPEED 4
INJECTION SPEED 5 0
(")

MOLD 'l'OUCH I '7.l

MOLD LOCK ()
SIIO'l' SIZE 1/5 g·
INJEC'rION IIIGH 'l'IME lo
-
INJECTION PACK TIME
INJEC'l'ION HOLD TIME -
COOLING 'l'IME 17 --·
TRANSFER POSITION /5" CHECKED
DECOMPRESSION ? CHECKED
SCHEW SPEED �
CHECKED ......
7, 91
V,
CUSHION I\PPHOVEO
 155
Case Study # I - Current Procedure Documentation

SET-OP SHEET
CUSTOMER:., PART NANE: C,L/ P
PART NUMBER: MOLD J:
MOLD FITS PR.F.:'SSES: 33 PREFERRED PRESS: 33
DO NOT RUN IN PRESS:

MOLD E!OT RUNNER

LENGTH: VOLTAGE: -
'WIDTS:
1�
; OF ZONES: -
HEIGET: TIP TYPE: NY(,_IJtJ RADIUS: 3/f ORIFICE:
WEIGHT: NOZZLE LENGTS: 2- (Ne+(

ErYDRAOLIC CORE PINS: LIMIT SWITCHES:..­

SEQ1JENCE: E!OW TO E!OOK OP:

SPECL:U. EQUIPMENT:

PERC:::NTAGE OF OPER.l\.TOR OTILZZATION:

,½
MATERIAL SPECIFICATIONS
MATE..�BL: /,J c. U t1,4t,I
a.ESIN �= Z..Z.t-H.J: "X�tV
IS DRYING REQUIRED?.
DRYING TEMPERATURE:
COLORANT: 70/ wl{r-r& DRYING TIME:
COLOR MIX: 3; (00 TEMPERATURE AT E!OPPER:
PERC�AGE �GR.IND: Z5 % MAXIMUM MOISTURE PERCENTAGE:

MATERL'!\.L LBS/1000 ?CS: t,/,z_ TOTAL SHOT W'!': 9{ 3

PART WEIGET: t/,' {:, � ·
f

CYCLZ TIME: 32.. 75 SEC PARTS/BOOR: PCS

SOX: I
PAC.KAGING
P!ECES/30X: /'JCJO
MOLD BEATER T"'...MPERATURE:
MOLD 'F...MPERATURE (MOVABLE) :
MOLD TEMPERATURE (STATIONARY) :
/SrJ
/Go
/5 0 F
=
?

SOXES/S'iGD: SLIDE/CORE TEMPERATURE: ?

?IECES/S'iC!D:
.KNOCKOUTS HOT RUNNER TEMPERATURES
LENGTS: IN ZONE l: F ZONE 5: F
DIAMETER: IN ZONE 2: F ZONE 6: F
':'IED IN?:
EJECTOR SPEED:
ZONE
ZONE
3:
4:
F
"" ZONE 7:
ZONE 8:
F
?
 SAMPLE PfWCESS Pll0CEOUl1E

MOLD MAP
CUSTOMER OAlE 0-2/-9.S- �
("')
SI)
Jdl"r j �/If Vl
(1)
. PAlll # ENG

. MOLD# SAMPLED OY

. MACII# 33
.MAlERIAU zz LHf 6 XG,AJ tJfltrE
.....
=It:

LOT# ("')
MFG. RECOMMENDED MATERIAL lEMP. PROCESS RANGE.
20�530·;::­
MFG. RECOMMEIIDED MATERIAL INJ. RA TE. MAcf/;'1 tit11 a
. MFG. RECOMMENDED MOLD TEMP.

. IS TIIE TOOL BALAIICEO?

I 75 ..,_ ZZ:5 °,C--
/JJ..O 1rJ J�,ol.r o« ""'c
('")
(1)

. FILL TIME [LO

�
. CAVITY I TIIAT FILLS LAST.
_:lI... 0
. SI IOT WEIGltr 90".4. FU� 0 IOO'll. 95 'l RUNNEII /
0
('")

NOTE TOOL IEMf'. OU MOLD MAP.

�
DI.ti I --..
MATERIAL MELT 1 EMP 30/30 RULE fi'-:z_z__ oF
.WAlER FLOW CORES V/51/AL CAVITY VtsvtlL .Ju o·
--------
,�
. PROCESS SI IEET COMPLE'I EU

LBS. OF MA l ERIAL USED

2_5
SAMPLES LOCA I ION Al-ID QTY.

. MACIIINE READY FOR Pl10DUC IION.

. MOLD REAOY fOH PHOllUCl ION. YES ✓- NO

'/£2 NOTE WI IY ON BACK.

. HElURN SAMPLE REQUEST StlEET WITI I SAMPLE SMOT. lb3'(

......
Pag• I
 157
Case Study # I - Current Procedure Documentation

-
--- ---- --· ·------------- ---·-----·-· -·----- ----- ---- ------ .
---------------- ----- - ---- -- - ·- ------- -·- -·-- ·
----·-- ·--·-·•--- ·----- -·--- --- -··---------·••---•-- ·----------•··----
- - -·- ---- •-·-------- ---·---·----·- - . . ··· -•-----
' ' -

-C-- --

----- -··· ·-- ·-- ---- <:r--· -- .. -

.. r--,- --- -

·-
 Appendix G

Case Study # 1 Mold and Process Documentation

New Procedure

158
 159

Case Study #1 - New Procedure Documentation

Mold Surface Temperature Map/Water Diagram/Critical Data

Mold Number· Date: (p/ 7 Z. f 95
�\\,,c...r\
Material: '2.. 2.. L H:;: loxEtJ Lot#:_____ Nozzle Type: t-1,,lo:n
/

l/ ,, 3; ,,
Nozzle Orifice:.__,_l_._f>,____ Nozzle Radius:� Melt Temp: 515. F
Fill Weight: t l. 7 J Part Weight: All"� 11. 7� Fill Speed(in/sec):____2_Q_J0
Q,v. ti: 8 -= q, <.,8
Gate Seal Time: IO Sec j

3,, \

I
i
�
7
, IA2.

r H\
-(
I----- .,,,i:ii"
I L\2 lt.\2 j
"
... L _J _J
..,
ui

c�
I I
! 0

7
�
8.
_I
l� L
&.
0

L../1
0

Ou.\ I
.t:::
L{
Cv.t2
L _J
'

Iv'\'::> 01.;..,-\--":i

\L\0 r
Cavity Half (Moving Half) Core Half (Stationary Half)
° °
Water Temp: l 40 F Water Temp:

NOTES:
 160
Case Study #1 - New Procedure Documentati9n

Date: 6/22/95 Melt Temp: 520

Tool Number: Mold Temp: 160

Viscosity Curve
lntesification Ratio = 101

Shot Fill Soeed Fill nme ltl lni. Pressure Relative Viscositv 1/t
1 70 0.77 1300 10010 1.30
2 60 0.9 1250 11250 1.11
3 50 1.11 1150 12765 0.90
4 40 1.45 1000 14500 0.69
5 30 2.2 875 19250 0.45
6 20 4.52 700 31640 0.22
7 15 9.52 750 71400 0.11
8 0 ERR
9 0 ERR
10 0 ERR

Viscosity Curve
80000 -,---- ------- ------,

·;;; 60000
8
� 40000
.?!
i 20000
a::
0 "--------------___,

0 0.2 0.4 0.6 0.8 1.2 1.4

1/t Reciprocal Seconds
 161
Case Study #1 - New Procedure Documentation

Date: 6/22195 520

160
Melt Temp:
Tool Number: Mold Temp:

Gate Seal Curve

Part We1a1
. ht
4
Pack/H 0Id T",me

5
6 9.55
7 9.6
8 9.59
9 9.56
9.68
9.67
10

12 9.67
11

13 9.66
14 9.66
15
16

18
17

19
20

Gate Seal Curve

9.68 .--------,------�

J:•: · ··· · · · ··• • ·r:.• ·•.;.··•·•·t•·• ·•·•·• • • ·• ·• • • • • • • • · •··I

9.66
....... .......... .. . .............. .......... / � .....................................
1

-e 9.6 + ···
.
11_ 9.58 1
/ \/
9.56
... ....... 1 I • ... ... .. . . ..
9.54 +------+----+-- -
�---+- ----,
0 5 10 15 20
Pack/Hold Time
 162
Case Study #1 - New Procedure Documentation

Shot Weight Mean Effects Shot Weight Variance Effects

0.01 ·r1------------,

I
I
2 ,, --------------,
0.005
1.5 •· ···· ········! i

i 1;
· · · · . . .. ·! j 0
� 0.5 l

■-----"----- ---ij
. .. . ..... ························
1
-0.005
0 �-- --
l
-0.5 _____________, -0.01
1
·•� ----,,------------<
X1 x2 x1x2 x3 x1x3
x1 x2 x1x2 x3 x1x3 x2x3 x2x3
Variable Variables

SWT Significant: 0.355448938

Dependent Variable 1
0.001 Mea n E=ff=e=ct=s___ 1E-06 ·---'-""-""'-'='-"'-"="'----
_
0.0005;
o i __.._______,--11-._,--i
0 I !
•lE-06 !
� -0.0005
I
j I
t: I
� -0.001 ..
I
w ·2E-06 i. . . ................ ......... ......... .......................... .1

-0.001s L l I
-JE-06 r··
-0.002 I· .. ,I
!
-0.0025 .�· ----------' --4E--06 -� · ----::---,,------
x1 x2 x1x2 x3 x1x3 x2x3 ' x1 x2 x1X2 x3 x1x3 x2x3
Variable Variables

____V_a_r1_S�ig.nificant: 0.001114818
Dependent Variable 2 Dependent Variable 2
=e=ct=s'-----�
100 __ _M=ea=n�E-ff 100 .---V
�a �·a=
=n n= �
ce �e
Eff =c='--------,
ts
,
I I
60

u 60 ' j
oof
60 f·
-1
w 40 W 40 I ,.

!
20 20 I
o----------- 0 i�x1----=---=------
x1 x2 x1x2 x3 x1x3 x2x3 x2 x1X2 x3 x1x3 x2x3
Vartable Variables

Var2 Significant: ERR

 163
Case Study #1 - New Procedure Documentation

------ ·----- ·--------- ---· -

Shot Weight Dependent Variable 1
Means Means

PU���I
i�::� 1.� . . . . . . . . . . . . ... . . . · I
·
98.5 ·- ---------�

t •� • • ·.!
04
� . 21j
i
•�"��
0.42 1

95 ··· ·····
i5)-0. 18 i.I
0.4 19 ac .
···· �·
.. . . � i
4 .
9�17
·
1 1 ------- -----'·
1
O 0. 417 '-
-1 -0.5 0 0.5 -1 -0.5 0 0.5
Level Level

1
.,.. Variable 1 -variable 2 • Variable 31 i * Variable 1 - Vanable 2 • Variable 31

Dependent Variable 2 Curvature Evaluation

Means
100 ·------- ---- --,
I ! SWT Var1 Var2
ERR
i
N
:1l 80
l Curvature -0.01 0.00
� so 1··· ... .......... ...... ... ....... . .... ......... .... .. .... I I Minimum 1.04 0.00 ERR

1 40

. . . . ... . . . . . . . . . . . . . . . . . . . ... I I
f
!
!
I I Process Variable High and Low Settings
20
O·�--�--....,.........,· I
-1 -a.5 o o.5 Var 1 Var 2 Var 3
i• Vanable 1 • Vana�e 2 • Variable 31
1 High
Low
530
510
175
140
40
20
!

Interaction x1 x2
i Interaction x2x3
Shot Weight Mean
i
I
Shot Weight Mean
----
----=---::-+------l
-----
I

t···
99
_91s+ ········· ·········· I

I . . ..·.:. . ·1l
- - -
o, I
l:i
I
l:' 98 t,

I j
!: I �==�.::::. .::::. . .=.... L,i,,.L----�
...... !
�::� \.. I , ..
� gs ��:- .....
:::c

::; ..,:•=·=· ====±=====�·' 94�-----�----�

-1 --0.5 0 0.5 1 I -1 -0.5 0 05
.
Vanable1
I Variable 2

; • Variable 2(·) -<r Variable I

2(+) i l•Variable3(-) ·•Variable3(•)1

Interaction x1 x3
98 .5 ..

� 9�
·�
_-_-

i� t·,. .
_-_-
_-_-
_- _-
_-
Shot Weight Mean
- �
· ·• .....- +.,
========· =�.
.. . ,
I
- 96.5 I
! II
� 96 .... •·:· ....... ...... · · ·•-•1
"'95.5 . ......... ..,
...
95 '"'======!::====:=:!I
-1 -0.5 0 0.5
Variable 1

i• Variable 3(-) * Variable 3(+) I

 164
Case Study # 1 - New Procedure Documentation

I
Interaction x1x2 Interaction x2x3 I

n���
Dependent Variable 1 Mean Dependent Variable 1 Mean
I
'; 0.423
I I ,. 0.421 •.
·c 0.4205
�
� 0.42 i·
+.===========
. . .... . . ····· ··· ·t· · · ······· ··· .···············
r
· · · · · ·· 1 I
I
··········11
5l o.4195 .L · ······················
-,,
x. 0.419
� 0.4185.
I
r 1
... ··..:. ! I
.
I

-1 -0.5 0 0.5 -1 -0.5 0 0.5

Variable 1 Variable 2

i•Variable2(·) ·.-Variable2(+)j i•Variable 3(·) •Vanable3(•)1

Interaction x1x3
Dependent Variable 1 Mean \
; 0 423 -=-----�------ I

jg:�t1
� 042
"Eo.419 \
�-
.. =- . . . .
�
.: . . . : ! I
1. .......�, .................1
. ... ·� ·� i
I

{g· : ��t ....... . . . .. 1··· '� I

� o.416 • I
.1 -a.5 o o.5 1
Variable 1 I
\
1 • Variable 3(-) • Variable3(•)I I

Interaction x1x2 Interaction x2x3

Dependent Variable 2 Mean Dependent Variable 2 Mean

r· ··· ·
�100�--------- ----,
·· ··· · · ·1 �100··--------------,

.
··· ········ 1
·� 80 � 80 ' ........ ..
. ·· · •· I
·
:l5 '

·r. . . .... .. · ·
1 T•··
r

�
:: 60 I . .. .. .. � 60 .... ......

l.. :
1

·1
- Ij-..
1 i 40
i 20 i · .... l..

0 · '-- · -----�1-----�·
!
0 I
Q. I
I

-1 -0.5 0 0.5 1 I
� Q '------'--------'
-1 -0.5 0 0.5
Vanable 1 Variable2
\

I I
I

i• Variable2(·) • Variable2(•) , • Vanable3(-) ..-Variable 3(•)I

I

Interaction x1x3
Dependent Variable 2 Mean
N 100 -- ------------,
.!!! 80
�
+·· 1....
............ ........
� 60 I ..... \
i" 40 t-I -
8 20 · · ··
. I I
� 0 '------�'-----�

-1 -0.5 0 0.5
Variable 1

j • Variable3(·) • Vanable3(+) j
 165
Case Study # 1 - New Procedure Documentation

Cavity-to-Cavity Differences
Dependent Variable 1

� 0.42

;i 0.41

::i;
04
· 1 2 3 4 5 6 7 8 9 10 11 12
Cavity

Cavity-to-Cavity Differences
Dependent Variable 2
100 ·,------------i

�
0
80
C:
"' 60
�c: 40
.,
� 20

O 1 2 3 4 5 6 7 8 9 10 11 12
Cavity
i• Mean • VarianceI
 166
Case Study # 1 - New Procedure Documentation

Shots 1-100
ShotJ Run I KeepI Shot Wt IDep Var 1 IDep Var 21 Cavity 98,----------

g:: �� : (:JLl-l :;>tr:;11w_: -

�'f�Pmt.Dillil'F'-11 Date: 6/22195
Tool#:
1 95_54
2 96_10 99 _5 4----- ------ ---------- -----,-----"-1---------,--------- ------- -
3 t I I
4
95_75
95_98 gs:_i ---f-
' I
t
5 1 1 97_50 0.419 _5 1
1 11 21 31 41 51 81 70 80 90 100
6 1 1 97_30 0.422 8 \
7 1 1 97_48 0.421 7
8 1 1 97.48 0.422 1

f � _ -• · · · · · • - • · • - • · · • - • -
9 1 1 97_19 0.417 6 Shots 101-200
10 1 1 97_10 0.421 2
11 1 1 97_33 0.425 4
12 1 1 97.45 0.425 3
13 96_04
14 96_10 i72 - ---------------- ------- ------------------

-\ift- ::: : •
• --- - - - --- - - - - -
15 96_08
16 96_12
;: i
17 2 1 96_13 0.417 6 -8
98 101 111 121 131 141 151 181 171 181 191
18 2 1 96_07 0.419 2
19 2 1 96_12 0.424 4
20 2 1 96_13 0.420 8
21 2 1 95_89 0.419 7 Shots 201-300
22 2 1 96_17 0.425 3 100,------------,
23 2 1 96_23 0.420 5 80 ------------- -------------- ----- ----------- ---------------------------
24 2 1 96_13 0.422 1
25 97_37 80 ------------ ------ --- --------------------- ---- ------------ --------
26 97.40 40 ------- - -- - ---- -- -------- ---- -- - ----- ----- --------------
27 97_39
20 ------ ------- - -- - --- ----------- --
28 3 1 97.40 0.421 2
29 3 1 97_52 0.419 5 0
201 211 221 231 241 251 261 271 281 291
30 3 1 97_55 0.421 8
31 3 1 97_57 0.426 4
32 3 1 97_54 0.421 6
33 3 1 97_59 0.420 7 Shots 301-400

1:r - - - - - - - ---- - --- - - - ---- -

34 3 1 97_58 0.423 3
35 3 1 97_59 0.422 1
36 97_53
37 97.47

�/
38 97_58
39 97_51
40 97_53
41 4 1 95_85 0.423 1
301 311 321 331 341 351 381 371 381 391
42 4 1 95_50 0.422 3
43
44
4
4
1
1
95_57
95_54
0.420
0.419
8
2
'------------I
45 4 1 95_58 0.416 6
46 4 1 95_55 0.424 4
47 4 1 95_70 0.419 7
48 4 1 95_53 0.420 5
49 96_65
 167
Case Study #1 - New Procedure Documentation

50 96.83
51 96.74
52 96.70
53 5 96.64 0.422 4
54 5 96.70 0.417 6
55 5 96.71 0.419 2
56 5 96.64 0.419 8
57 5 96.54 0.422 3
58 5 96.67 0.419 1
59 5 96.60 0.418 .7
60 5 96.66 0.418 5
61 95.86
61 95.93
62 95.92
63 95.87
64 95.88
65 6 96.13 0.419 7
66 6 96.20 0.417 6
67 6 96.17 0.420 3
68 6 96.23 0.419 1
69 6 96.20 0.419 2
70 6 96.19 0.417 5
71 6 96.15 0.418 8
72 6 96.04 0.421 4
73 97.12
74 97.07
75 97.13
76 97.10
77 97.08
78 7 97.30 0.421 3
79 7 97.51 0.422 4
80 7 97.23 0.420 2
81 7 97.13 0.419 8
82 7 97.15 0.419 6
83 7 97.34 0.417 5
84 7 97.40 0.420 1
85 7 97.27 0.418 7
86 95.55
87 95.53
88 95.54
89 8 96.03 0.419 8
90 8 95.57 0.418 1
91 8 95.60 0.418 2
92 8 95.60 0.417 7
93 8 95.77 0.421 3
94 8 96.03 0.416 5
95 8 95.64 0.417 6
96 8 95.70 0.421 4
97 97.46
98 97.45
99 97.39
100 97.24
101 97.36
102 9 97.35 0.418 7
 168
Case Study #1 - New Procedure Documentation

103 9 97.42 0.419 1

104 9 97.84 0.417 5
105 9 97.51 0.421 4
106 9 97.47 0.422 3
107 9 97.32 0.419 8
108 9 97.67 0.420 2
109 9 97.64 0.417 6
110 97.10
111 97.09
112 97.05
113 96.72
114 96.78
115 96.65
116 10 96.74 0.421 2
117 10 96.68 0.421 1
118 10 96.87 0.417 6
119 10 96.68 0.419 7
120 10 96.67 0.419 5
121 10 96.65 0.423 4
122 10 96.91 0.420 8
123 10 96.69 0.421 3
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
1AA
 169
Case Study #I - New Procedure Documentation
SET-OP SHEET
CUSTOMER: PART�: {;'LIP
PART NUMBER: . -. MOLD t:
MOLD FITS PRE'SS�S'":-33 PREFERRE.LJ .t>�SS:3 3
DO NOT RUN IN PRESS:

MOLD HOT RUNNER

LENGTH: VOLTAGE: -
WIDTH: t OF ZONES: -
HEIGHT:
WEIGHT: z /AICI/
TIP TYPE:,A/ff-.1.J,N
NOZZLE LENGTH:
RADIUS: 34 ORIFICE: f-s-

HYDRAULIC CORE PINS: LIMIT SWITCHES:­

SEQUENCE: HOW TO HOOK OP:

SPECIAL EQUIPMENT:
SPECIAL INSTRUCTIONS:

PERCENTAGE OF OPERATOR UTILIZATION: Y:r

MATERIAL SPECIFICATIONS
MATERIAL= we t-l.A1A..A, IS DRYING REQUIRED?
l(&-S
RESIN t: ZZi-llY:: � Y&il.1
COLORANT: l,V/ WNrr&­
DRYING TEMPERATURE:
DRYING TIME:
/ ocJ F
HRS
3
COLOR MIX: 3 .; /tJ(} o F
PERCENTAGE REGRIND: Z,S- % �=�S� :�:��NTAGE:/oO %

_MATERIAL LBS/1000 PCS: TOTAL SHOT WT: RUNNER WEIGHT:

PART WEIGHT:

CYCLE TIME: 33., '7( SEC PARTS/HOUR: PCS

PACKAGING MOLD HEATER TEMPERATURE: /'/0 F

BOX: I MOLD TEMPERATURE (MOVABLE) : /'/0 F
PIECES/BOX: /;;oo MOLD TEMPERATURE (STATIONARY) : /-i<:J F
BOXES/SKID: SLIDE/CORE TEMPERATURE: F
PIECES/SKID:
KNOCKOUTS HOT RUNNER TEMPERATURES
LENGTH: IN ZONE l: F ZONE 5: F
DIAMETER: IN ZONE 2: F ZONE 6: F
TIED IN?: ZONE 3: F ZONE 7: F
EJECTOR SPEED: ZONE 4: F ZONE 8: F
MOLD NUMBER 'l'ROUBLESIIOOTING
DA'l'E 1,; - zz -Cfc.
MACIIINE NUMBER
-:s-z.
NOZZLB 'l'EMPERA'l'UHE c;·;o
ZONE 1 'l'EMPERA'l'URE '5/u n
�
-5Lo
VJ
ZONE 2 'l'EMPEHA'l'UIIE (1)
ZONE 3 TEMPERATURE
000 E"

t1o
INJEC'l'ION IIIGII PHESSURE
�
PACK PRESSUHE
IIOLD PRESSUllB
�/, ......
=It:
S.P(/::.,dl 7-t?Z
CLAMP IIIGII Pll8SSUR8 c�
(1)
CLAMP LOW PHESSUllE �
�
DACK PRESSUHE
LQZ. '"Cl
INJECTION SPEED 1 .�
-
a
(")

--
(1)
INJECTION SPEED 2 0..
INJECTION SPEED 3
INJECTION SPEED 4
INJECTION SPEED 5 - 0
(")
MOLD 'l'OUCII /.,zj,

g_
MOLD LOCK
{)
SIIO'l' SIZE 4?
INJEC'l'ION IIIGII 'l'IME I:.? 0
::s

-LZ
INJECTION PACK TIME
INJEC'l'ION IIOLD 'l'IME
/6
a
COOLING 'l'IME
TRANSFER POSITION CIIBCKIW
DECOMPllESSION ? CIIECKED
SCllEW SPEED CHECKED
5-Q/4
CUSIIION
Z'/7 APPllOVED
......
-..J
0
 Appendix H

Case Study #2 Mold and Process Documentation

Current Procedure

171
 172

Case Study #2 - Current Procedure Documentation

FIRST SHOT CHECKLIST

Tool#: Project: c- So.Ji

S.B.U. Tool Eng.: _ ___,�-rr _ __
I
Material: L'M')\1 C

Prototype Mold Proces; Eng: l(o

Production Mold Tech Cell T ech:

YES NO NIA MOLD BASE

� 1.) Ejector system returns per mold design?
C7 2.) Clamp slots per mold design?
3.) Are water lines numbered and named in or out?
4.) Adequate relief around water nipple?
✓ 5.) TL painted on all sides?
.._,/ 6.) Mold safety straps (2) with keeper holes?

---
._,/ 7.) Locating ring as specified on tool design?

----
:::::::---- 8.) Eyebolt holes for balanced hanging and handling per mold design?
9.) Information stamped in base per tooling standards?

----- --- -::;:::----,-

10.)
11.)
12.)
Radius on sprue bushing per mold design?
Knock out pattern/relief for knockouts per mold design?
Is water flow restricted?
13.) Does mold have tippers per mold design?
14.) Is ejection guided per mold design?
15.) Are runner blocks or rotating sprue to mold design?

CAVITY INFORMATION
1. ) Are vents around cavities per mold design?
2.) Is parting line relieved around cavity?
- 3.) Do large standing steel tool have inter1ocks per mold design?
4.) Are ejector pins flush per tooling standards?
--- 5.) Do lifters seat correctly (flush)?
6.) Is sprue puller pin to tool design?
7.) Are runners per mold design?
8.) Are cold slug wells to tool design?
9.) Do subgates release from tooVpart cleanly?
._/ 10.) Is mold polished to tooling standards on visible surface?
11.) Is TL# and Cavity# stamped in cavity per part design?
/ -------- 12.) If left/right family mold, are RH parts scribed to part design?
.?
13.) Are ribs polished to tooling standards?
------- 14.) Do slides and horn pins actuate and return?
--
----
15.)

16.)
Are slides with up and down motion have keepers/springs to
keep in place?
Is accelerated ejection working correctly?
.....-- 17.) Is cover release working correctly?

AevisP.d: 10/?7l<M
 173
Case Study #2 - Current Procedure Documentation.-
�1110 11\c..
t� PoWCJ:a.� �
�" ��� ec·

27/JUN/95 08:04:15 PAGE 001

�,ETUF' SHEET

l"IACHINE #

CLAMF' EJECTOfi
CLOSE VELOCIT Y lB.00 INIS START EJECT SETPOINT 6 ' ,..,._..-' IN
CLOSE SLOUDOUN 7,50 IN
FORUARD LIMIT 1.80 HI
l"IOLD TOUCH 2.01 IN EJECT FORUARD DUELL 0,00 SEC
PULSES l COUNT
BF:EAKAUAY SPEED 5,00 IN/S
OF'EN FAST 6.50 IN RETRACT LIMIT 0,50 IN
OPEN VELOCITY 20,00 IN/S EJECT RETRACT OVEl�F: I DE 12.00 IN
OF' El'l SLOUDOUN 9.25 IN
OF'EN LIMIT 9.75 IN

TEl"IF'E fi' 1nu RES

NOZZLE �� 5 0 DEG F ZONE 555 DEG F
ZDNE ,.,,_ 545 DEG F
ZONE 3 �;45 DEG F
AL�1RM E;Al'ID 20 %

Til"IERS
CYCLE ALAF:M LIMIT 60.0 SEC COOLING 8.00 SEC
HIJECT HIGH 15,00 SEC EXTRUDER DEI.AY 0.50 EC !,
PACK 7,50 SEC OF'EN DUELL 0,00 SEC
HDLD 0,00 SEC

INJECTION CONTROL DATA

SHOT SIZE 5.00 rn INJ HI F'RS LMT L000 F'SI F'ACK 1 750 F'SI
VELOCITY 1 3. 72 Il'USEC HYO :<FER F'RS 10 F'SI F' AC f( 2 750 F'SI
SIJ [TCHF'DINT () %
VELOCITY 2 0, 00 IN/SEC HOLD 1 0 F'S I
SWITCHF'IJINT 2 0 % HOLD 2 0 F'Sl
VELDCITY "' .I 0 .00 IN/S[:C
Si.I ITCHF'O HIT 3 0 % ADAF"TIVE SHOT CONH:OL OFF
VELIJCIT Y 4 0,00 IN/S[C EF:ROF: CORRECTION 10 .,,.
S l,J ITCHf' 0 I tH 4 0 % CUSHION 0.20 I l'-J
VELDCIT Y 5 0,00 IN/SEC
TRANSFER POS 0.76 HI

EXTRUDER CONTROL DATA

EXTRUDER SPEED INIT 30 % BACK PRESSURE INIT 50 F'SI
EXTRUDER SPEED FINAL 30 % BACK PRESSURE FINAL 50 F'Sl
DECOMPRESS DISTANCE 0.05 IN
 174
Case Study #2 - Current Procedure Documentation

27/JUN/95 08:04=31 PAGE 001

SETUP SHEET

FEATURE SELECT
HYO EJT ON
EJT DURIO OFF
SPRU BRK OFF
TRY AGAIN OFF
HI SPD EXT MTR ON
PART DETECT OFF
CORE TIMERS OFF
MODE OF TRANSFER POSITION

MOU HIG COFiE!,

SET CORE CLAMP POSITION 0.00 IN SET COf/E TIME
SELECTION: AO 0.00 SEC
PULL CORE CLAMP POSITION 0.00 IN PULL CORE TIME
SELECTION: BO 0.00 SEC
 .·�.... ,(,,w:;;,c.
Tonttqge.i'.'.;i!1�-·
!;·'· �� t•·c·.-.
..-., -.-. ·, .ID s;..,, ·-.;
· ''ill[P.l 't/JU1t'• _-
� .���·,�.'
- ' .
-·
;<,r· .. �:w·j&�
• -_ ,..,u. ·ttfr��

I
.
I I I I I
.......... ..... ...., "'1 ,. • ' '_ _..¥.,J,.... ' Uc ·-· .
'Do all checks at start-up and color changes -,.,.'(
n
I
Samele Historv Loa
. I Fill Peak Low Low Finish t-1msn <.;ycl Melt Part Temp Steel Temp

-
� l �
# I Dato I Tech. ID I Material I Color I Fill wt. 1 FIii WI. 21 Time Pis Press PSI Time Pa,1 wt. 1 Part WI. 2 Hoe t T•"ll Cav. 1 Cav. 2I Cove, Ejector V,
I Ii,/,� I ,P Il ...">11° I�, ... IJi..,c. I .e� l11.99c- lsi.ro I -,.c-O IJ9, v I I _,, I ,n l,�...�I - I e,'( IB� 0

Issues: Recommended Action: Action Taken:

z
�
�
N
n
s:::
e:,.,,. "1,..sl
Notes: ,, ""Ps �\' Int.
Date a
I I I I I I I I I I I I I I I ""O
Issues: Recommended Action: Action Taken:
(")
0
0.

�
t,
0
(")

Notes: lnr. §0
I I I I I I T T I I I I I I I
Date
a
Issues: Recommended Action: Action Taken:
�-
0
::s

Notes: Int.
Date
'Issues should state reasons for any 11arlatlbn from original process and tool conditions encountered
.....
-..J
VI
 176
Case Study #2 - Current Procedure Documentation

RHEOLOGY
TL NUMBER MACHINE 85 DATE 6/27/95

melt temp 557 mold temp 85 material type em3110 color (cone?) black
intensification ratio = 11.00 material lot#

Fill Peak Plastic 1/ Relative

Shot Time H d. Pres Pressure . Fill Time Viscosit
1 7.35 805 8855 0.14 65084
2 4.2 800 8800 0.24 36960
3 3 850 9350 0.33 28050
4 2.3 903 9933 0.43 22846
5 1.9 968 10648 0.53 20231
6 1.62 1034 11374 0.62 18426
7 1.43 1148 12628 0.70 18058
8 1.27 1225 13475 0.79 17113
9 1.18 1293 14223 0.85 16783
10 1.05 1358 14938 0.95 15685
11 0.98 1421 15631 1.02 15318
12 0.9 1471 16181 1.11 14563
13 0.85 1508 16588 1.18 14100
14 0.8 1553 17083 1.25 13666
15 0.75 1622 17842 1.33 13382
16 0.73 1693 18623 1.37 13595
17 0.7 1756 19316 1.43 13521
18 0 #DIV/0! 0
19 0 #DIV/0! 0
20 0 #DIV/0! 0

SHEAR RHEOLOGY
70Cffi

6CXXX)

5CXXXl

4CXXX)

·;;:
30Cffi
!
20Cffi

lOCffi

0
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60
flow rate
 177
Case Study #2 - Current Procedure Documentation

GATE SEAL
TL NUMBER ----- MACHINE
co Id tool
---------
85 DATE 6/27/95

melt temp 557 mold temp _-""85-"--_ material type em3110 color (conc?) _b_la_ck__ _
intensification ratio = 11 material lot# _______ _
_
Pack Part
Time we· ht
1 37.66
Gate Seal
2 38.46
3 39.02
4 39.32
5 39.61
6 39.61
7 39.62
8 39.61
9 39.61
10 39.62
11 39.62
12 39.61
13
14
15
16
17
18
19
20 38.5

.i
:r;
LOW PSI 750

38

Pack Time
 178
Case Study #2 - Current Procedure Documentation

___
DATE l.,/ J 7/<IJ
TECHNICIAN
_....__
WATER LINE DIAGRAM
Tool Nuni:>er • . - -
Please note -Temp in, Temp Out. and GPM for each Loop.

....

COVER 2.. C:, !""-

OPERATOR �
CAVITY SIDE
� �f'"'-

EJECTOR

OPERATOR
SIDE I CAVITY
 Appendix I

Case Study #2 Mold and Process Documentation

New Procedure

179
 180
Case Study #2 - New Procedure Documentation

Mold Surface Temperature Map/Water Diagram/Critical Data

Mold Number: -· - - Date: 0/27/95
I

Material: Lqf<(I 3110 Pc Lot#: L AY8oY Nozzle Type: GP

Nozzle Orifice:__,_Y _ �-- Nozzle Radius: 3/y ''
I�uQ" Melt Temp: 5 -i 3 "f='
Fill Weight: 3 5. 08 t Part Weight: 3 9, "'7 Fill Speed(insec): 2 .1

Gate Seal Time: ·7 -:,e c..

,,. i.
·>·-"\
........._
\\100
)
)

..,.,
\1u.
, 10)
0
I e8.
--
f!
8.
'--- \___
\

-
\
5)
0 V\ \

1
(� c) v,lc
o�,rt 1

Cavity Half (Moving Half) Core Half (Stationary Half)

Water Temp: I r--r. f
O 0
Water Temp: IOO F l..,A..J

NOTES:
 181
Case Study #2 - New Procedure Documentation

Date: 6/27/95 Melt Temp: 520

Tool Number: Mold Temp: 120

Viscosity Curve
lntesification Ratio = 11 I

Shot Fill Speed Fill Time ltl lni. Pressure Relative Viscositv 1/t
1 6 0.87 1997 19111 1.15
2 5 0.88 1998 19341 1.14
3 3.5 0.93 1983 20286 1 08
4 3 1.02 1843 20678 0.98
5 2.5 1.2 1674 22097 0.83
6 2 1.45 1516 24180 0.69
7 1.5 1.87 1350 27770 0.53
8 1 2.73 1199 36006 0.37
9 0.5 5.33 1090 63907 0.19
10 0.3 8.78 1107 106914 0.11

Viscosity Curve
.
120000 �------------�
?:-100000 +
·;;; ··\
8 80000 �·I ......\.
"' I

> 60000 +I
m 40000 T
.�

-a;
O:: 20000 t·
0 �1-�--�----------,--
0 0.2 0.4 0.6 0.8 1.2
1 /t Reciprocal Seconds
 182
Case Study #2 - New Procedure Documentation

Date: 6/27/95 Melt Temp: 520

Tool Number: Mold Temp: 120

Gate Seal Curve

Pack/Hold Time Part Weiaht
4 38.67
5 38.72
6 38.79
7 38.81
8 38.81
9 38.8
10
11
12
13
14
15
16
17
18
19
20

Gate Seal Curve

- 38.78
. 38.
3����
76
...
j·····
. .... .... / ..
..... ..___
.... ....../. .... , .......•.....

i
i l
� 38.74
� 38.72
tt t·
•······
I
a. 38.7
38.68
38.66
t
l

J...l ___ ......_____�---+---- . . · 1·

0 5 10 15 20
Pack/Hold Time
 183
Case Study #2 - New Procedure Documentation

Shot Weight Mean Effects Shot Weight Variance Effects

o .5 0.01 -------------,
I 1
0.4 t····· ·· o.ooa I
_ 0.3
o
t·
I

u 0.006 i
I
� 0.2 r· �0.0041
0.002 i

O·
x1 x2 x1x2 x3 x1x3 x2x3 x1 x1x2 x3 xlx3 x2x3
Variable Variables I
- .I
SWT Significant: 0. 164190004
Dependent Variable 1 Dependent Variable 1
Mean Effects Variance Effects
0.004 .-------- -- 2E--07 �--------�
1
0.003 \... ··········••·••··· ·••· ············· ·····················1
I I

u 0.002 Ii

� � -IE--07 · i
w 0.001 I 1
•2E--07 ,·· ·············· !
o -JE--07 ,. · ······································· ·· ··· .. J
--0.001 ....____________, -4E--07 I
xi x2 x1x2 x3 x1x3 x2x3 xi x2 x1x2 x3 x1x3 x2x3
Variable Variables

Var1 Si nificant: 0.001141602

Dependent Variable 2 Dependent Variable 2
Mean Effects Variance Effects
0.025 I
I
0.005 ·1-----------,!
a�-- � 0.00
I I 4
r'-
i

l
0.003
fo1s i �
1: I
w 0.01 · w 0.002 ·
1 ·j 1
a=! � 0.001 i
o _I
xi x2 x1x2 x3 x1x3 x2x3
aL xi x2 x1x2 x3
� - Li
x1x3 x2x3
Variable Variables

Var2 Significant: 0.025555604

 184
Case Study #2 - New Procedure Documentation

Interaction x1x2 Interaction x2x3

Dependent Variable 1 Mean Dependent Variable 1 Mean
';
� 3.172 �··:::::.:.:..:::::::::::::::=""'-''t""--�
3.174 �-----�----�

3.17 c::..:: ····································t··

U:�: .:.�-·· _· _· ·_ · =· · =· ·,,,.,· ·,-c:·""· • ..::•• ••..::+;=••=· ••·=· · =· ·=· · =· ·=·· =· ·=· · =· ·:::- ·�
� ··········· ··· ··· ·········· ·- - ........... ........... ..
3.168 ·····
1'.

c j · · t ..

� 3.162 �-----�- ---�

-1 --0.5 0 0.5
Variable 1

I• Variable2(-) -Variable2(+)i i• Variable3(-) •Variable3(+)i

.•
r
Interaction x1x3
Dependent Variable 1 Mean

:1. ••··•••••••·• •••··•••••••·••••·•�

;; .
Q. 3, :1;66,!�[
� 3.165
-1
I!!:=======-:.__.______
--0.5 0
..
.
•
... ....•........�

0.5 1 I
Variable 1 1
!
/•Variable 3(-) • Variable3(+)i I
Interaction x1x2 Interaction x2x3

-� : ,
Dependent Variable 2 Mean Dependent Variable 2 Mean

1�1.��
';:2.78 I ';: 278 ,-------,------­

;i,����:
:;; 2 7 ' ··· ! A

I -=--
�268 '. , � 2.68 :. -·•••· .. .,
� 266 _. I
� 2.66 a-
-1 -0.5 0 0.5 -1 -0.5 0 0.5
Variable 1 Variable 2

: • Variable 2(-) • Variable2(+)i 1 • Vanable 3(-) • Variable 3(•ti

Interaction x1x3
Dependent Variable 2 Mean

' I l§ �-?�
. i;: 2.78 .-------------

'
' �268 1.......
� 2.66 --------�------
_, --0.5 0 0.5
Variab6e 1

f•Vanable3(-) •Variable3(•)i
 185
Case Study #2 - New Procedure Documentation

Shot Weight Dependent Variable 1

Means

i:'::t�l
Means

_!::t· ·· · · · · · · · · · · · · · · · · · · · · . · /· ·;
� 3.172 -------------..-�

f!!:�L.:
c;
.c 39
.. ..:..._,,,..... ··��/.:
rI' · ··· · · ··· _,,,.---
t-==• 1/
�: :
· · ··· ····· ··· · ···· ··· ····· ······ I 1

!::: I ·:� :�···

(/)
� I -✓
· .
O 3.164 .1�---,-:---,,----,----�
1'

o 05 ·· ... .. j -1 -05 0 0.5

Level Level
i..,.
.
Variable 1 • Variable 2 • Variable 3 j i..,. Variable 1 • Vanable 2 • Variable j3

I epende���ariable 2
"'�
Cuivature Evaluation

E!��I!
SWT Var1 Var2
Cuivature 0.14 0.00 0.01
Minimum 0.45 0.00 0.07

Process Variable High and Low Settings

I 2.68 . I Var 1 Var 2 Var 3

I
.1 -a.5 o o.5
j I High 560 150 700
I j-.. Variable 1 • Jana� 2 • Variable 3
I Low 520 100 500

Interaction x1x2 Interaction x2x3

Shot Weight Mean

r . . . . �.�1
Shot Weight Mean
39.5 ,----
---r------

J!::: I ----
� 39.2 :.
:=--""·- __.;._______
<IJ 39.1
39 -----�----�
-1 -0.5 0 05 1 ! -1 -0.5 0 0.5
Vanat>le 1 . Variable 2

;• Vanable 2(-) * Variable 2(+)j I• Variable 3(-) ... variable J(+)j

Interaction x1x3
39.8
_ 39.6
c-====+=======t
Shot Weight Mean

i
I
I
39
f 394 I I

! j�t· I
_g 38.8 :. _c...---'-
::.. --·· ·
(/) 3 8.6 ,i,.c�· __.,.,.,.,,--:;-- ·
384 ______,________,
-1 -0.5 0 0.5
. Variable 1

i • Variable 3(-l ..-variable 3(•)1

 186
Case Study #2 - New Procedure Documentation

Shots 1-100

I: .x1
Shot Run Keep I Shot Wt Dep Var 1 Dep Var 2 Cavitv
6127/95
39:�.... .
. �

=� �L:/ J'v-"'
· · · "A
395,..... �. /
1 39.34
2 39.36
3 39.35 39 L f · tIt!':
l
.
4 1 1 39.36 0.436 2.698
5 1 1 39.36 0.436 2.698 3Ul·
38.6 ··� i 11 21 31 '1 51 81 70 110 90 1�
6 1 1 39.34 0.437 2.697
7 1 1 39.36 0.437 2.696
8 1 1 39.36 0.434 2.695
9 1 1 39.39 0.435 2.696 Shots 101-200
10 1 1 39.34 0.436 2.695 39.15 �-------�
11 38.89
12 38.89 39.1

13 38.88
i. . ...
39.05

14 38.89 39
15 2 1 38.89 0.429 2.693
16 2 1 38.89 0.430 2.694 38.95

17 2 1 38.87 0.431 2.692 38 9

· 101 111121 131 141151 181 171 181 191
18 2 1 38.88 0.431 2.693
19 2 1 38.88 0.432 2.694
20 2 1 38.86 0.430 2.693
21 2 1 38.89 0.430 2.694 Shots 201-300
22 39.38
23 39.36 :�

t:r . ··. .· .· .· .·... . . . .·.·.·. ·. · · · .

1 ............................... •
24 39.38
25 3 1 39.39 0.432 2.696 110 � ..... ...............
. ..................... ······ ..............

26 3 1 39.47 0.431 2.695 .0 i

i
27 3 1 39.67 0.432 2.696
28 3 1 39.37 0.432 2.697
29 3 1 39.38 0.432 2.695 201 211 221 231 241 291 2fi1 271 281 291
I
i
30 3 1 39.56 0.433 2.695
31 38.93
32 38.91 I
33 38.91 Shots 301-400 I

34 38.89
35
36 4 1
38.86
38.80 0.433 2.695
i
37
38
4
4
1
1
38.81
38.78
0.433
0.432
2.693
2.693 I
39 4 1 38.77 0.433 2.695
40 4 1 38.77 0.433 2.693
41 4 1 38.78 0.432 2.695
42 39.18
43 39.11
44 39.10
45 38.93
46 38.99
47 39 01
48 39.00
49 5 1 38.99 0.433 2.696
 187
Case Study #2 - New Procedure Documentation

50 5 38.99 0.433 2.696

51 5 39.02 0.432 2.695
52 5 39.01 0.433 2.694
53 5 39.05 0.432 2.696
54 5 39.00 0.433 2.695
55 39.10
56 39.03
57 39.00
58 39.01
59 39.05
60 6 39.14 0.431 2.694
61 6 39.18 0.430 2.693
61 6 39.20 0.431 2.694
62 6 39.17 0.431 2.694
63 6 39.07 0.431 2.694
64 6 39.27 0.431 2.694
65 39.97
66 39.67
67 39.65
68 39.98
69 39.77
70 39.74
71 7 39.61 0.437 2.696
72 7 39.89 0.437 2.964
73 7 39.89 0.437 2.965
74 7 39.61 0.437 2.696
75 7 39.99 0.437 2.697
76 7 39.87 0.438 2.697
77 38.99
78 38.82
79 8 38.82 0.433 2.694
80 8 38.95 0.434 2.694
81 8 38.90 0.434 2.694
82 8 38.92 0.434 2.694
83 8 38.90 0.435 2.695
84 8 38.89 0.435 2.693
85 38.99
86 38.99
87 38.95
88 38.94
89 39.00
90 9 38.97 0.433 2.694
91 9 39.08 0.433 2.694
92 9 39.08 0.432 2.694
93 9 39.13 0.432 2.694
94 9 39.10 0.433 2.695
95 9 38.97 0.432 2.695
96 39.22
97 39.18
98 39.20
99 10 39.24 0.435 2.697
100 10 39.19 0.434 2.696
101 10 39.15 0.434 2.695
102 10 38.92 0.433 2.696
 188
Case Study #2 - New Procedure Documentation

103 10 38.93 0.432 2.694

104 10 38.99 0.433 2.696
105
106
107
108
109
110
111
112
113
114
115
 189
Case Study #2 - New Procedure Documentation

27/JUN/95 13:36:09 PAGE 001

SETUP SHEET
MACHIN E # l
CL�1MP EJECTOR
CLOSE VELDC!TY 18.00 IN/S STAF-T EJECT SETPDINT 6.75 IN
CLOSE SLOWDOWN 7.50 IN
FORWARD LIMIT 1 . 80 IN
MOLD TOUCH 2.01 IN EJECT FORWARD DWELL 0.00 SEC
PULSES 1 COUNT
BREAKAWAY SPEED 5.00 IN/S
OF'EN FAST 6.50 IN RETRACT LIMIT 0.50 IN
OPEN VELIJCITY 20.00 IN/S EJECT RETRACT OVERRIDE 12.00 HI
OPEN SLOWDOWt� 9.25 IN
OPEN LIMIT 9.75 IN
TEMPERATURES
NOZZLE 51+0 DEG F ZONE 1 530 DEG F
ZONE 2 520 DEG F
ZONE 3 �;io DEG F
ALARM BAND 20 %

TIMEFi S
CYCLE ALARI"! LIMIT 60.0 SEC COOLING 7.50 SEC
INJECT HIGH 15.00 SEC EXTRUDER DELAY 0.50 SEC
F'ACK 8.00 SEC OF'EN DWELL 0.00 SEC
HOLD 0.00 SEC
INJECTION CONTROL DATf4
SHDT SIZE 3.00 IN !NJ HI PRS LMT 2000 F'S! F'ACK 1 700 F'S!
VELOCITY 1 2.70 IN/SEC HYD XFER PRS 10 PSI F'ACK 2 700 PSI
SW ITCHF'DI NT 1 0 %
IJ ELO CITY 2 0.00 IN/SEC HOLD 1 0 F'SI
SWITCHF'OINT 2 0 % HOLD 2 0 F" SI
VELOCITY 3 0.00 IN/SEC
�iWITCHF'OINT 3 0 % ADAPTIVE SHOT CONTROL OFf'
1
JELOCITY 4 0.00 IN/SEC ERRIJR CORRECT I (Jl'j 10 %
SW ITCHPOINT 4 0 % CUSHION 0.20 HI
VELOC !TY 5 0.00 IN/SEC
HANSFER F'OS 0.76 IN
EXTRUDER CONTROL DATA
EXTRUDER SPEED INIT 35 % BACK PRESSURE INIT 50 PSI
EXTRUDER SPEED FINAL 35 % BACK PRESSURE FINAL 50 PSI
DECOMPRESS DISTANCE 0.05 HI
 190
Case Study #2 - New Procedure Documentation

27/JUN/95 13•36•26 PAGE 001

SETUP :,HEET

FEATURE SELECT
HYD EJT O�I
E.JT OIJRID OFF
SPRU 8RK OFF
TRY AGAIN OFF
HI SPD EXT MTR ON
PART DETECT OFF
CORE TIMERS OFF
MODE OF TRANSFrn POSITION

MOUING COl�ES
SET CORE CLAMP POSITION 0.00 IN SET CORE TIME
SEl.ECTION: AO 0.00 SEC
PULL CORE CLAMP POSITION 0.00 IN PULL CORE TIME
SEL ECTION: 80 0.00 SEC
 Appendix J

Case Study #3 Mold and Process Documentation

Current Procedure

191
 192
Case Study #3 - Current Procedure Documentation

TOSHIBA SET-UP SHEET FOR EAST PLANT

CE#: PART# NAME: - �,.bj+rr:.
PRESS#: .J.:-- 1- MATERIAL: V? Clo � P, 11 11 ,,t_ DATE: LP ct BY: -

- - - - - - - - CLAMPTROL - - - - - - - - - - TIMER- -
EJECT LS 20 ··. LS 2A PCL LS2 PCH TR1

I } Z 3Q 030 �o (), 1 l-0 &;, 3

VE VC1 VC2 TR3
LJ I t,o
Z w.5 /�

ss
TR4

L[O
LS32 LS31 V02 V01

22 Bto 2' c)
--
><
THICKNESS LS3 LS38 LS30 LS3A TR 42

NEXT
THICKNESS '288 2 Sj zoo Z0
-
LS100 LS30 TR 41

- X
,.__--------------INJECTROL ---------------

Jo Z8 5e,,C C_:j 0. 1e ·· $o/

PH 3 PH 2 PH 1 P1 CYCLE TIME BP

TRH 2
3Q
TRH 1
l--/ {) q� F,· /J �. Z?. c
RPMS ACTUAL

2 z 3 G
VH Vl4 Vl3 Vl2 VI1 SRN

-- )_2,_ qo t.-; J
-
LS 4 LS 4C LS 48 LS 4A LS 5 LS 10

) LP I !..{ 25 2/'YJM
-�---------HEATR OL-------------- I.OLD TEMPERA TUAE
HN H1 H2 H3 H4 HS

u()
ACTUAL ACTUAL ACTUAL ACTUAL ACTUAL ACTUAL CO VER

L../ t-.J {) L--j L-/ C ½ !:;£ '-/ 3 £

SETTING SETTING SETTING SETTING SETTING SETTING MO VING

&a
 Appendix K

Case Study #3 Mold and Process Documentation

New Procedure

193
 194
Case Study #3 - New Procedure Documentation

Mold Surface Temperature Map/Water Di�gram/Critical Data

Mold Number:_____ Date: Lt,(29 (95
P?
Material:A<"·,�tech TJL!t20C. Lot#: l�G>-0 ½<.;nil. Nozzle Type: GP
1;. /"
Nozzle Orifice:__.._--''1___ Nozzle Radius:�" Melt Temp: Lf 28 ° f="
5 \. 9 S :l 5 5. J8 Fill Speed(insec): 7 0 Jc,
Fill Weight: Part Weight:
J
Gate Seal Time: /5 + "-e(,, (eo .. lc\ -r-.olr-- �-z� ;"' r-ea-::,,,-Ao..lok cyc_\c +;�)

-:r.,., 3
7°t 7LJ;
Q.,_-\- :i
.,
ii
tl+
3"''1
Bo 77
Ov.tli

V
I"' I Ou.\ I I"2 O,..,:t'?.

Cavity Half (Moving Half) Core Half (Stationary Half)

Water Temp: _Q
__o or
r-
roar
Water Temp: \.Cl_ r
{
NOTES:
 195
Case Study #3 - New Procedure Documentation

Date: 6/29/95 Melt Temp: 420

Tool Number: Mold Temp: 70

Viscosity Curve
lntesification Ratio = 101

Shot Fill Soeed Fill Time ltl lni. Pressure Relative Viscositv 1/t
1 80 0.24 600 1440 4.17
2 60 0.33 475 1568 3 03
3 40 0.5 380 1900 2.00
4 30 0.71 300 2130 1.41
5 20 1.24 250 3100 0.81
6 10 3.12 220 6864 0.32
7 7 5.33 200 10660 0.19
8 0 ERR
9 0 ERR
10 0 ERR

Viscosity Curve

.... . ·.,1i
.�::: i,t \ ... I
i :��� tt �\.
� 8000

� 200 --�-----.�=::.:..;...:::.::.�::.:::.;; · · ..
�

0 2 3 4 5
1 /t Reciprocal Seconds
 196
Case Study #3 - New Procedure Documentation

Date: 6/29/95 Melt Temp: 420

Tool Number: Mold Temp: 70

Gate Seal Curve

Pack/Hold Time Part Weiaht
4 54.13
5 54.205
6 54.405
7 54.58
8 54.65
9 54.825
10 54.935
11 54.985
12 55.05
13 55.01
14 55.225
15 55.255
16
17
18
19
20

Gate Seal Curve

::
:: 1·· ...... .. Ir::-. . . ... . j
1: 55 . .. . . . .-"..,,__ ..... I
,..- I

54 8
.Q>
Q) Jl/ ..........., .. .,.......... l

�
t:: 54.6
. ..· .... ....... ... ... ...... ......__.-:"
. 7··..
··· ··
•
I
a. 54.4 -;-
ffl /

54.2 L. ,., ,-. ········I

54 -l---------+----+----
1

0 5 10 15 20
Pack/Hold Time
 197
Case Study #3 - New Procedure Documentation

Shot Weight Mean Effects Shot Weight Variance Effects

0.2
rl
0.15 t
=.-----------,!
.................................. I
0.006 �. ----------
0.005 !_ ···························. ······
· ·················•····

0.1 • ···· ·· ····· ···· · · I 0.004 !

ij
I j 0.05; 0.003 i
� 0.002 l.. ...... ........I
0 ;­ I
0.001 I..
f··
•••••••••••••.••.j

-0.05 · · · ························1
O·
-0.1 i i -0.001 � - --� --1 -- x 3--x 1-x3_____;
x2x3
X1 x2 X1 x2 x3 x1x3 x2x 3 x1 x2 x x2
Variable Variables

SWT Significant: 0.074348474

Dependent Variable 1 Dependent Variable 1
Mean Effects Variance Effects
0.01 r--------------, SE-06 ·-. --------­
'
0.008 I ........... ... i
a
I
l
0.006 f·
. I
j 0.004
..... a.002
/ ••·•·••·••·1
····· 1
i -SE-06 : ············ ····. I
0
r w I
-1E-0 5 '·········-- ....... .......
I
-0.002 I I
-0 004 �--------___,J -1.SE-05 ._____________,;
1
x ,a 1
x >c2 x3 1
x x3 x2x3
x1 ,a x1>c2 3x x1x3 x2x3
Vartabte Variables

Var1 Si nificant: 0.003132376

Dependent Variable 2 Dependent Variable 2
Mean Effects Variance Effects
100 ,------------ 100 -----------
!
80 ! ·· I
:
80 I······ ....... · .. ... ...
I
�
�
60 ! i 60 ·
i: ' I
w 40 • w 40 ·

20 20 •·
a------------' o-----------
1x 2x x1x2 3x 1
x x3 x2x3 1
x 2x 1
x x2 3x 1
x x3 2
x x3
Vartabte Variables

Var2 Significant: ERR

 198
Case Study #3 - New Procedure Documentation

Shot Weight Dependent Variable 1

Means ·Means
55.5 ·-�---------� - 3.83 -----------

l�il�i
� 3.825 L. .. ... . ... .�..
� I .... / I

55'- ____ !
1 3:::1·· �1
� 3.811. /
i
...
-1 -0.5 0 0.5 -1 -0.5 0 0.5
Level Level
j....Variable 1 • Variable 2 •Variable 31 i• Variable 1 • Vanable 2 •Vanable 3 I
I
I Dependent Variable 2 Curvature Evaluation
Means
j 100
SWT Var1 Var2
IN i
I" 80 ·1··
I Curvature -0.07 0.00 ERR
!i
I'" 60 j. ... . j' Minimum 0.19 0.01 ERR
.�
I
I :ii 40 I
l� i ... ....... .. I Process Variable High and Low Settings
,o
l � 201···
0 '-' __________..---,
-1 -0.5 0 0.5 Var 1 Var 2 Var 3
High 440 80 40
i.,.. Variable 1 • Vanabie 2 • Vanable 31
Low 400 60 15

Interaction x1x2 Interaction x2x3

Shot Weight Mean Shot Weight Mean
55 · 55.4 -----------
.s , I ;; 55.35 ::::---
1: 55.5 ; . .�
i554!·
I �-··········· l I
� -! ------�- j
·?

�::���-
r· . . �
... . :
:-----...
55.3

55 .
5 2
j :� : I ...... . .. ......... ... �l
55 �5 .
�

--�
'

(/) 5�k--?:'.:: I i I
-0.5 0 0.5 -1 -0.5 0 0.5 1 '
Variable 1 Variable 2
-1
I !
!•Variable2(·) ¼Variable2(•>1 • Variable 3(-) �-variable 3(•)1

Interaction x1x3
Shot Weight Mean
55.5, :::::::!
'.
1J:::r· � i
�ss.21• · ··1
� · ·
Cl) 55. 1 F'""'
i
T
55 '-' -----�-----
-1 -0.5 0 0.5
Variable 1

:• Variable 3(-) • Variable 3(+) I

 199
Case Study #3 - New Procedure Documentation

Interaction x1x2 Interaction x2x3 I

Dependent Variable 1 Mean Dependent Variable 1 Mean
I
3.83 I ; 3 824
i .....---'!'
:g 3.822 �...................
l5 i ---------
•
' . I I

�:
3
�
:;;s�········ ··•··•••••·••• 1
l ::.....•· �
· t:�:�� �
T� i
I I I

}�:;:! .....
···············
� 3.81
i------
"" 1

� 3.805 I
-1 -0.5 0 0.5 -1 -0.5 o o.5 1 I
Variable1 Variable2
!

I- Variable 2(-) • Variable 2(•) I i•Variable3(-) •Vanable3(•>!

I
Interaction x1x3 I
Dependent Variable 1 Mean
';; 3.83 ··-------------:
I
I

\'.§��!
� 3.805 ·'------�----�
.1 -0.5 o o.5 1
I
I
Variable1
I
I• Variable 3(-) • Variable 3(•)I I
Interaction x1x2 I Interaction x2x3
Dependent Variable 2 Mean
I
';: 100 ,---------,------,
I Dependent Variable 2 Mean

I r· ····· •·
';:100,•------,-----
I \

....
--,
80 .
·1 ·
�
80

Ii
!

60 . 1 I1 > 60
I
:'.
� I 1 ·•·r···· ...
40
� 20 I . : !C 20 II 4
0
. i .
� o�----�-----� I II �
X. i
0------------
-1 -0. 5 0 0.5 1 ! -1 -0. 5 0 0. 5 1 ,
Variable1 Variable 2

1 • Variable2(·) • Variable 2(•) I 1 • Variable3(-) • Variable 3(•) I

Interaction x1x3
Dependent Variable 2 Mean
';: 100,�-----�------,
j.
I :� I
� 80 i ... .. ... ... .... ..

!· · ... .. . !. . . I
� 20 · ·· ······ J-
-0

� 0'------�------
-1 -0.5 0 0.5
Variable1

!• Vanable3(-) •Variable 3(•)j

 200
Case Study #3 - New Procedure Documentation

Shots 1-100
Shot I Run IKeeo I Shot Wt lDeo Var 1 IDeo Var 2 I Cavitv
b���-
55.7 ----------,
6/29/95
��:� •
T
1 55.17
2 55.51
3 55.46
4 54.97
5 55.06 S4.9 1 11 21 31 .., 51 81 70 90 90 100
6 1 1 55.07 3.808

·:r
7 1 1 55.18 3.808
8 1 1 55.27 3.811
9 1 1 55.08 3.811 Shots 101-200

· · · · · · · · · · · · · · · · · · · · ··· ·· I
10 1 1 55.13 3.815
11 1 1 55.25 3.809
12 55.11

:[
13 55.17
14 55.15
15 55.27
16 55.35
17 55.24 0 -
101 111 121 131 141 151 181 171 181 191
18 55.20
19 2 1 55.20 3.817
20 2 1 55.15 3.829
21 2 1 55.15 3.817 Shots 201-300
22 2 1 55.32 3.812 100----------,
23 2 1 55.23 3.813
24 2 1 55.21 3.815
25 55.12 . .................

26 55.17 40
27 55.20
20 . .
28 3 1 55.14 3.813
29 3 1 55.21 3.814 0
201 211 221 231 2•1 251 261 271 291 291
30 3 1 55.30 3.820

!
31 3 1 55.22 3.814
32 3 1 55.13 3.812
33 3 1 55.11 3.821 Shots 301-400

':f · l
34 55.20

:t - i
35 55.15
36 55.18
37 4 1 55.15 3.819
38 4 1 55.18 3.816
39 4 1 55.16 3.805
40 4 1 55.23 3.813
41 4 1 55.13 3.815 301 311 321 331 341 351 361 371 381 391
42 4 1 55.05 3.815
43 55.26
44 55.43
45 55.39
46 5 1 55.39 3.818
47 5 1 55.38 3.813
48 5 1 55.34 3.820
49 55.14
 201
Case Study #3 - New Procedure Documentation

50 5 55.35 3.819
51 5 55.39 3.821
52 5 55.28 3.821
53 55.45
54 55.44
55 55.64
56 6 55.55 3.818
57 6 55.42 3.825
58 6 55.48 3.827
59 6 55.41 3.829
60 6 55.41 3.822
61 6 55.41 3.823
61 55.13
62 55.18
63 55.23
64 55.23
65 55.22
66 7 55.17 3.819
67 7 55.33 3.824
68 7 55.21 3.824
69 7 55.20 3.824
70 7 55.50 3.819
71 7 55.31 3.826
72 55.42
73 55.27
74 55.32
75 8 55.29 3.821
76 8 55.30 3.816
77 8 55.50 3.825
78 8 55.26 3.819
79 8 55.20 3.827
80 8 55.31 3.822
81 55.31
82 55.36
83 55.36
84 9 55.36 3.824
85 9 55.36 3.818
86 9 55.44 3.821
87 9 55.65 3.823
88 9 55.44 3.820
89 9 55.39 3.827
90 55.32
91 55.31
92 10 55.30 3.820
93 10 55.43 3.815
94 10 55.26 3.821
95 10 55.34 3.816
96 10 55.42 3.816
97 10 55.19 3.818
98
99
100
101
102
 202·
Case Study #3 - New Procedure Documentation

TOSHIBA SET-UP SHEET FOR EAST PLANT

CE#:
PREss #: Zot.J 1

- - - - - - - - CLAMPTROL - - - - - - - - - - TIMER- -
EJECT LS 20 LS 2A PCL LS2 PCH TR1

\ \ Z3o 3o So -9 ·10 \l."?:)

VE VC1 VC2 TR3

�2 L?5 15 !-\; 0
LS32 LS31 VO2 VO1 TR4

·22 o<oeo �a 55 7-.0

><
11-ilCKNESS LS3 LS3B LS30 LS3A TR 42

0
NEXT 288 25 0 -Zoo 7-D
11-ilCKNESS
TA 41

><
LS 100 LS 30

0 a 0
-- -------------���TRQ---------------
PH 3 PH 2 PH 1 P1 CYCLE TIME BP

lB 28 2.5 9q Z8 Sec.... 1-f

TAH 2 TRH 1 APMS ACTUAL

2.o \':).O Co5

VH Vl 4 Vl3 Vl2 VI 1 SAN

\1 () 0 C 70,o uo
LS 4 LS 4C LS 48 LS 4A LS 5 LS 10

11.� 0 0 0 ·7- S.0 7

-----------HEATAOL-------------- l.'OLD TEMPERATURE
HN H1 H2 H3 H4 HS
ACTUAL ACTUAL ACTUAL ACTUAL ACTUAL ACTUAL COVER

�2G L./2 Ci) �23 �1 9 too

SETTING SETTING SETTING SETTING SETTING SETTING MOVING

t.425 '-!2-S Y2o Lf2_o (vC

 Appendix L

Human Subjects Protocol Approval

203
 204

Human Subjects lnstrtutional Review Board Kalamazoo, Michigan 49008-3899

616 387-8293

WESTERN MICHIGAN UNIVERSllY

.':-,;,_�lK
'�

Date: Jan 31, 1995 , l\·

To: Vander Kooi, Mark \
G
, ' '"
\}.
From: Richard Wright, Interim Chait\y'$i'

Re: HSIRB Project Number 95-01-12

This letter will serve as confirmation that your research project entitled "Development of a mold
trial procedure for establishing a robust process" has been approved under the expedited
category of review by the Human Subjects Institutional Review Board. The conditions and
duration of this approval are specified in the Policies of Western Michigan University. You may
now begin to implement the research as described in the application.

Please note that you must seek specific approval for any ch anges in this design. You must also
seek reapproval if the project extends beyond the termination date. In addition if there are any
unanticipated adverse or unanticipated events associated with the conduct of this research, you
should immediately suspend the project and contact the Chair of the HSIRB for consultation.

The Board wishes you success in the pursuit of your research goals.

Approval Termination: Jan 31, 1996

xc: Engelman , IENG

 BIBLIOGRAPHY

Advanced Process Engineering. (1990). Injection Molding Reference Guide.

Advanced Process Engineering: Seminole, Florida.

Barometer. (1986). Modem Plastics, 63(1), 11.

Besterfield, D.H. (1986). Quality Control (2nd ed.). Englewood Cliffs, New Jersey:
Prentice-Hall.

Bodfish, AD., Dennison, A.H., Howard, L.M, & Engelmann, P.V. (1992, December).
Noise Affecting Shot-to-Shot Repeatability in the Injection Molding Process.
(Report no. 91/019-10-05). (Research and Technology Institute of West
Michigan, Grand Rapids).

Borland. (1993). Quattro Pro for Windows Version 5.0 [Computer program]. Scotts
Valley, CA: Borland International, Inc.

Boteler, J.L. (1993). Using prevention techniques. Quality Progress, 26(7), 105-107.

Box, G.E.P., Hunter, W.G., & Hunter, J.S. (1978). Statistics for Experimenters. New
York: John Wiley & Sons.

Bozzelli, J., Groleau, R., Kirkland, C., & Colby, P. Sr. (1993). How to achieve
repeatable production for quality parts, Details of the plastics world productivity
challenge, What did we do and why? The results!. Technical Papers, Vol 39, pp.
574-578, Brookfield, CT: Society of Plastics Engineers.

Campbell, J.P., Marshall, S.N., & Sommers, J.P. (1992, November). Injection
molding control strategies. (Report no. 007B). (Research and Technology
Institute of West Michigan, Grand Rapids).

Chauvel, AM., & Andre, Y.A. (1985). Quality cost: better prevent than cure. Quality
Progress, 18(9), 29-32.

Creighton, J.P., Mcinemey, J.A., & Soto, E.G. (1992). Further development of PCIM
- Data acQuisition system. Unpublished manuscript, Western Michigan
University, Department of Engineering Technology, Kalamazoo, MI.

205
 206
Dykstra, K.K., Engelmann, P.V., Huff, RA., Monfore, M.D., Schoon, M.T., &
Vander Kooi, M.V. (1994, December). Transferability of injection molding
process data. (Report no. 007E).(Research and Technology Institute of West
Michigan, Grand Rapids).

Dym, J.B. (1986). The injection molding machine operation, Injection Molding
Handbook (pp. 84-113). New York: Van Nostrand Reinhold.

Dym, J.B. (1987). Injection Molds and Molding: A Practical Manual (2nd ed.). New
York: Van Nostrand Reinhold.

Engelmann, P.V., Monfore, M.D., Campbell, J.P, Marshall, S.N., & Sommers, J.P.
(1993, April). Shot-to-shot repeatability for injection molders. (Report no. 007C).
(Research and Technology Institute of West Michigan, Grand Rapids).

Engelmann, P.V., Monfore, M.D., Dykstra, K.K., Huff, RA., Schoon, M.T., &
Vander Kooi, M.V. (1994, June). Injection molding shot-to-shot repeatability.
(Report no. 007D). (Research and Technology Institute of West Michigan, Grand
Rapids).

Fischer, J.W., Hom, M.E., & Schulte, E.W. (1991). Communications network.
Unpublished manuscript, Western Michigan University, Department of
Engineering Technology, Kalamazoo, MI.

Groleau, RJ. (1984). Intelligent molds. Modem Plastics, fil.(11), 73-74.

Groleau, RJ. (1989). Injection molded part size control. Technical Papers, Vol. 35,
pp. 1693-1697, Brookfield, CT: Society of Plastics Engineers.

Groleau, RJ. (1991). Rigorous mold tryouts-the key to SPC. Technical Papers, Vol.
37, pp. 346-351, Brookfield, CT: Society of Plastics Engineers.

Groleau, RJ. (1993, March). RJG Seminars: Injection Molding and Process Control.
(Available from RJG Associates, 2640 Aero Park Drive, Traverse City, Michigan,
49684).

Harry, D.H. (1990). Injection machine set-up stability. Technical Papers, Vol. 36, pp.
340-342, Brookfield, CT: Society of Plastics Engineers.

Harry, D.H. (1991). Injection molding machine control algorithms. Technical Papers,
Vol. 37, pp. 383-386, Brookfield, CT: Society of Plastics Engineers.
 207
Hartley, JR (1992). Concurrent engineering. Cambridge, Massachusetts:Productivity
Press.

Hunkar, D.B. (1994). MAD approach to the determination of optimum machine

setups in injection molding of thermoplastics. Technical Papers, Vol. 40, pp. 721-
728, Brookfield, CT: Society of Plastics Engineers.

Key markets post solid gains in 1993. (1993). Modem Plastics, 71(1), 73-83.

Landfill-plastics volume figures keep piling up. (1989). Modem Plastics, 66(13), 12-
14.

Liu, C., & Manzione, L.T. (1993). Process studies in precision injection molding.
Technical Papers, Vol. 39, pp. 1085-1087, Brookfield, CT: Society of Plastics
Engineers.

Malloy, R, Chen, S., & Orroth, S. (1988). Melt viscosity measurements using an
instrumented injection molding nozzle. Technical Papers, Vol. 34, pp. 279-284,
Brookfield, CT: Society of Plastics Engineers.

Martin, M. (1994). What's your plasticating 'window'? Plastics Technology. 40(13),

36-39.

Miller, B. (1989). Cavity pressure control keeps molding quality on target. Plastics
World, 47(11), 62-66.

Miller, B. (1992). New sensors tune up equipment's vital signs. Plastics World, 50(9),
32-34.

Nunn, RE. (1986). The reciprocating screw process, Injection Molding Handbook
(pp. 56-83). New York: Van Nostrand Reinhold.

Rauwendaal, C. (1993). SPC, statistical process control in extrusion. New York:

Hanser.

Research and Technology Institute of West Michigan. (1991, March). Plastics

Computer Integrated Manufacturing. A multi-client research and development
consortium proposal. Grand Rapids, MI: Author.

Ricketson, RC., & Wang, K.K. (1987). Injection molding process control based on
empirical models. Technical Papers, Vol. 33, pp. 231-234, Brookfield, CT:
Society of Plastics Engineers.
 208
Rosato, D.V. (1986). Process control technology, Injection Molding Handbook (pp.
277-314). New York: Van Nostrand Reinhold.

Ross, P.J. (1988). Taguchi techniQues for Quality engineering. New York: McGraw­
Hill.

Sammons, C. (1994). Steel and dimensional analysis: How to make mold

modifications. Injection Molding. 2(10), 47-52.

Smock, D. (1993). Rates rise slightly for injection molding time. Plastics World,
.ll(12), 20-23.

Speight, R.G., Coates, P.D., & Hull, J.B. (1994). The use of process measurements
for real time injection moulding process control. Technical Papers, Vol. 40, pp.
696-700, Brookfield, CT: Society of Plastics Engineers.

Speirs, R.G. (1993, December). Conclusions on feed throat and hydraulic oil
temperature effec ts experimentation. (Report no. 056B). (Research and
Technology Institute of West Michigan, Grand Rapids).

Speirs, R.G., Thayer, K.A., Wilson, B., & Massei, S. (1993, November). Practical on
line rheology for injection molding. (Report no. 008B). (Research and
Technology Institute of West Michigan, Grand Rapids).

Statsoft. (1994). Statistica Release 4.5 [Computer program]. Tulsa, OK: Statsoft, Inc.

Technology Survey Preliminary Results/2. (1991, December). Unpublished survey

data.

Tobin, W.J. (1992). Qualifications, Startu,ps, and Tryouts oflnjection Molds. WJT
Associates: Louisville, Colorado.

Whelan, T., & Goff, J. (1994). The Dynisco Injection Molders Handbook, 1st
Edition. Dynisco Instruments: Sharon, Massachusetts.