8.. $3.

3.,ch

:14. 1}.

3W1», s

 

a...

a: .6 3,3.“

I. ..

A
.u .. .F.
. (5...:

11,. A7,!

 

Kuhnx u

This is to certify that the
dissertation entitled

A NEW METHODOLOGY OF PROCESS DESIGN FOR
RECONFIGURABLE MANUFACTURING SYSTEMS

presented by

YICHONG ZENG

has been accepted towards fulfillment
of the requirements for the

PhD. degree in MECHANICAL ENGINEERING

V .. _-. - - .

 

 

  

Major Professor's Signature

off/fl/(ng

Date

MSU is an Affinnative Action/Equal Opportunity Institution

 

 

 

LIBRARY
Michigan State
University

 

 

 

PLACE IN RETURN Box to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJClRC/DatoDuepes-sz

_. .‘_____.—.

 

 

A NEW METHODOLOGY OF PROCESS DESIGN FOR RECONFIGURABLE
MANUFACTURING SYSTEMS

By

Yichong Zeng

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Mechanical Engineering

2004

.& 4 4 .3 . n .. 11141 I .
.M . In ATP...‘II ‘I. .‘v’ii .( . ‘yllii‘t ‘03 .4 . It- {IE-III.‘II.'IV.III‘I;9 .‘uluttt-I' Y

ABSTRACT

A NEW METHODOLOGY OF PROCESS DESIGN FOR RECONFIGURABLE
MANUFACTURING SYSTEMS

By

Yichong Zeng

In response to the increasing complexity of the manufacturing environment and
the rapid change of market requirements, a new methodology for process design
is applied in Reconfigurable Manufacturing Systems (RMSs). The methodology is
used to model, design, and reconfigure the processes and to facilitate
collaborative manufacturing among different companies or different departments
within one company. The methodology can help to manage the enormous
quantity and variety of process data generated from the changes in functional
requirements, production facilities, and manufacturability. It can also improve

system’s flexibility to new production and adaptability to new process technology.

In this thesis, in order to apply the methodology in a real industrial practice, a
case study has been conducted to capture the processes and to set up a process
model for making a variety of pulleys at Focus: Hope in Detroit, Michigan. Based
on the process model, a hierarchy approach of constraint management is
represented for reconfiguring processes efficiently in reconfigurable
manufacturing systems. A unique aspect of this process model is that it uses a

grammar-based approach to capture both processes and constraints in hierarchy

It -‘hrnw..pu..uuhflfi.r . my .ILntI I- (. it‘ll- Iltaltlrtfit‘tt‘tr 1.

 

o. :9..|l...¢...t .

r u‘. :ko Jr... ..k.l\-E.-I..

structure, which facilitates a way to apply a multilevel multiple criteria decision-
making model (MMCDM) for process selection. Within the MMCDM model, the
grammar-based approach helps to decompose a complex selection problem into
several manageable chunks; therefore, an effective multiple criteria decision-
making method - Analytic Hierarchy Process (AHP) could be applied to the
process selection at any abstract level. The MMCDM model has been applied in
this dissertation on selecting processes, machines, and part families for
producing a part in a complex manufacturing environment. It can also be applied

on those problems with a great amount of data and many decisions to be made.

Copyright by
YICHONG ZENG
2004

To my wife, Sining Yu

ACKNOWLEDGMENTS

I would like to take this opportunity to express my appreciation to a number of

people, without whom my thesis could never have been completed.

My sincere gratitude goes to my advisor, Dr. Patrick Kwon. Dr. Kwon has
instructed me in many things since he has been my advisor at Michigan State
University. He provided me with valuable guidance, the influence of which on my
personal and technical development will be carried forward into my future
endeavors. I also want to thank my committee members: Dr. Brian T. Pentland,
Dr. Brian Feeny, and Dr. Dahsin Liu. I appreciate their reviewing my dissertation

and providing many valuable suggestions.
I would like to thank my parents, Xianglian Zeng and Guilian Wu. Without their
continuous support, love, and encouragement, I cannot reach this

accomplishment today.

I proudly share this accomplishment with my wife, Sining Yu. I would also like to

thank her for her everlasting support, patience, and love.

vi

TABLE OF CONTENTS

LIST OF TABLES .................................................................................. ix
LIST OF FIGURES ................................................................................ xi
LIST OF ABBREVIATIONS ..................................................................... xii
CHAPTER 1
INTRODUCTION ................................................................................... 1
1.1 Manufacturing Systems ................................................................. 2
1.1.1 Dedicated Manufacturing Systems .......................................... 2
1.1.2 Group Technology and Flexible Manufacturing Systems ............. 3
1.1.3 Reconfigurable Manufacturing Systems ................................... 4
1.2 Process Design for Reconfigurable Manufacturing Systems .................. 4
1.3 Introduction of Grammar Approach .................................................. 7
1.3.1 Shape Grammar ................................................................. 8
1.3.2 Graph Grammar ................................................................ 9
1.3.2.1 Design Features .................................................... 10
1.3.2.2 Process Grammar ................................................. 10
1.4 Analytic Hierarchy Process for Process Selection ............................ 14
1.5 Outline .................................................................................... 14
CHAPTER 2
CASE STUDY .................................................................................... 17
2.1 Background .............................................................................. 19
2.2 Modeling Processes ................................................................... 23
2.2.1 Genesis of Alternative Processes ........................................... 23
2.2.2 Representing the Overall Processes ........................................ 25
2.2.3 Constraints in Manufacturing ................................................. 34
2.3 Conclusion ............................................................................... 38
CHAPTER 3
CONSTRAINT MANAGEMENT ............................................................... 39
3.1 Introduction .............................................................................. 39
3.2 Process Model Hierarchy ............................................................ 41
3.3 Constraints for the Process Models .............................................. 43
3.3.1 Categorizations .................................................................. 43
3.3.2 Constraint List .................................................................... 45
3.3.3 Classification of Constraints .................................................. 49
3.4 Constraint Hierarchy ................................................................... 50
3.4.1 Production ......................................................................... 51
3.4.2 Machine ............................................................................ 52
3.4.3 Process ............................................................................. 57

vii

3.5 Process Reconfiguration with Constraint Hierarchy .......................... 58

3.5.1 Machine Assignment ......................................................... 58

3.5.2 Constraints Related to Process Hierarchy ............................... 59

3.6 Conclusion ............................................................................. 60

CHAPTER 4

IMPLEMENTATION OF PROCESS SELECTION ...................................... 62
4.1 Introduction ............................................................................ 62
4.1.1 Decision Making Method for Hierarchies ................................ 62

4.1.2 Consistency Analysis ......................................................... 66

4.1.3 Application of AHP on Manufacturing Problems ...................... 67

4.2 Multilevel Multiple Criteria Decision Making (MMCDM) Model .......... 71

4.3 A Case Study for Applying MMCDM ........................................... 73

4.4 Modification of Matrix for AHP Method ......................................... 82
4.4.1 Case Study without Matrix Modification ................................. 84

4.4.2 Case Study with Matrix Modification ...................................... 86

4.5 Selection of Part Family ............................................................ 90

4.6 Conclusion ............................................................................ 93

CHAPTER 5

CONCLUSION ................................................................................... 94
5.1 Contributions .......................................................................... 94

5.2 Future Work ........................................................................... 96
APPENDICES .................................................................................... 97
APPENDIX A. GLOSSARY .................................................................... 98
BIBIOGRAPHY ................................................................................. 102

viii

LIST OF TABLES

Table 1. Basic features of pulleys ............................................................ 31
Table 2. Machining process flow of pulley .................................................. 33
Table 3. Alternative processes due to assigned machine .............................. 33
Table 4. Typical process constraints for making a pulley .............................. 37
Table 5. Classification of Constraints ........................................................ 48
Table 6. Alternative processes due to the assigned machine(s).....................56
Table 7. Constraints associated to the tasks .............................................. 59
Table 8. Random index of the matrix ........................................................ 66
Table 9. Application of AHP on manufacturing problems .............................. 69
Table 10. Relative priorities with respect to objective constraints ................... 76
Table 11. Scale of relative importance ...................................................... 77

Table 12. Relative priorities of constraints with respect to alternative

processes ........................................................................... 78
Table 13. Global weight of machine with respect to the process ..................... 80
Table 14. Attributes of alternatives based on selected sub-altemative ............. 81
Table 15A. Relative priorities with respect to the goal ................................... 82
Table 158. Selection of alternative process for the goal ............................... 82

Table 16. Relative priorities with respect to drilling cost, boring cost,
tolerance, production rate, and milling cost without matrix
modification .......................................................................... 85

Table 17. Relative priorities with respect to drilling cost, boring cost,
Tolerance, production rate, and milling cost after matrix
modification ......................................................................... 86

Table 18. Relative priorities of constraints with respect to alternative

processes .............................................................................. 87
Table 19. Attributes of alternatives based on selected sub-alternative..............89
Table 20. Two part families of pulleys ....................................................... 91
Table 21. The difference of given part refer to the part families.......................92

Table 22. Relative priorities of part family with respect to weight, 0.0.,
I.D., and tolerance .................................................................. 92

LIST OF FIGURES

Figure
Figure
Figure
Figure
Figure
Figure
Figure
Figure

Figure

Figure 10.
Figure 11.
Figure 12.
Figure 13.

Figure 14.

Figure 15.
Figure 16.
Figure 17.

Figure 18.

1. A process for making a pulley ................................................... 12
2. Overall process of making pulleys ............................................. 26
3. The process flow in company B ................................................ 27
4. A detailed quotation process .................................................... 29
5. A process for finding suppliers .................................................. 29
6. A process for outsourcing ........................................................ 30
7. Process flow for machining ...................................................... 3O
8. Features of a pulley ................................................................ 32
9. Alternative processes due to tolerance requirement ...................... 32

Process Hierarchy in producing a pulley ..................................... 42
Constraint hierarchy for producing pulleys ................................... 50
Basic Features of a typical pulley .............................................. 53
Alternative Processes depending on the assigned machine............54

Change of constraints and alternative processes depending
on the assigned machine ........................................................ 56

Multilevel Multiple Criteria Decision Making (MMCDM) .................. 72
Case study of MMCDM ........................................................... 73
Case Study 2 ........................................................................ 84

The selection of part family for casting process ............................ 91

xi

LIST OF ABBREVIATIONS

AHP: Analytic Hierarchy Process

CAD: Computer Aided Design

CAM: Computer Aided Manufacturing

DMS: Dedicated Manufacturing Systems

FMS: Flexible Manufacturing Systems
GT: Group Technology

RMS: Reconfigurable Manufacturing Systems

xii

Chapter 1

INTRODUCTION

Today, manufacturing industries face many challenges, including global-scale
competition and rapid change of process technology. In order to be competitive,
a production system must be flexible enough to add new requirements and
functions and be open to adopt new process technologies if needed. These
requirements could be achieved by developing a new manufacturing system —
Reconfigurable Manufacturing System (RMS) with reconfigurable machine tools
[Koren et al. 1999]. However, the cost for hardware replacement would be
tremendous for many small manufacturing companies. Due to the limited budget,
many companies would like to reorganize their present production facilities and
process technologies, instead of replacing them with a completely new
production system. Therefore, the development of new methodologies for
process design and the related software can be more helpful than before to

improve the flexibility and manufacturability of a production system.

A methodology to model, design, and reconfigure into an optimal set of
processes for such production systems should enable:
0 Frequent introduction of new products and rapid adjustment of
manufacturability to market demands;
. Rapid integration of new functions and process technologies into existing

systems, and

0 Easy adaptation to variable quantities of products for niche marketing.
The methodology to model, design, and reconfigure processes must be able to
help the system to reconfigure quickly for a new production, be able to produce a
wide variety of products in any scale of production and be able to integrate a new

technology.

1.1 Manufacturing Systems
Most manufacturing industries use dedicated or flexible manufacturing systems
to produce their products. The type of manufacturing system used for an industry
depends on the product requirements including quantity, cost, and variety of
products. Each manufacturing system differs in production throughput,
equipment and production cost, flexibility of system configuration, and
adaptability to new functions and process technologies. Thus, the processes

being used for a particular product depend on the manufacturing systems.

1.1.1 Dedicated Manufacturing Systems
A dedicated manufacturing system (DMS) is a system designed for producing a
specific part at a high volume, which uses typically transfer line technology with
fixed tooling and rigid automation. When the product demand is high, the cost per
part is relatively low. The economic objective of a DMS is to cost-effectively
produce a specific part type at high volumes and required quality. Due to fixed
tooling and fixture during the production, the process for DMS is usually designed

for the production of a specific part only. It is difficult to reconfigure the processes

required by the changes of functionality, manufacturing process, production

facility, and manufacturability. The system configuration is also difficult to change.

1.1.2 Group Technology and Flexible Manufacturing Systems
Group technology (GT) is an approach to manufacturing in which similar parts
are identified and grouped together in order to take advantage of their similarities
in design and production [Groover 2002]. Similarities among parts permit them to
be classified into part families. A part family is a collection of parts that are similar
in geometric shape and size or require similar processing steps in their

manufacture.

A flexible manufacturing system (FMS) is a highly automated GT machine cell,
consisting of a group of processing stations (usually CNC machine tools),
interconnected by an automated material handling and storage system and
controlled by an integrated computer system. Because of the single-tool
operation of the CNC machines, the FMS throughput is lower than that of DMS.
The combination of high equipment cost and low throughput makes the cost per
part relatively high. The objective of a FMS is to make possible the cost-effective
manufacture of many types of parts, which can reduce changeover time on the
same system at the required volume and quality. The process for FMS can be
designed for production of a part family, which uses similar tooling and fixtures
for several types of parts. Due to the flexible tooling and fixtures, it is easy to

reconfigure the processes for changes within a flexible manufacturing system.

1.1.3 Reconfigurable Manufacturing Systems
As defined by Koren et al. [1999], a reconfigurable manufacturing system (RMS)
is designed for rapid change in the structure of both hardware and software
components, in order to rapidly adjust to the new production method, production
capacity and functionality in response to changes in market or regulatory
requirements. The reconfigurable manufacturing system will allow adding,
removing, or modifying specific process capabilities, controls, software, or
machine stmcture to adjust production capacity in response to changing market
demands or technologies. The objective of an RMS is to provide the functionality
and capacity that is needed, when it is needed. Accordingly, the process design

for RMS will be quite different compared with DMS and FMS.

1.2 Process Design for Reconfigurable Manufacturing Systems
A process design methodology is very important to a manufacturing system. The
design of a manufacturing system on the hardware and software components
usually defines the system properties such as the manufacturability, production
scale, operation and maintenance cost, and flexibility of system configuration.
Different from the manufacturing system, the methodology to model, design, and
reconfigure processes provides the organization of machines, processes,
constraints, and other involved manufacturing resources within the manufacturing
system. The methodology will be able to represent the processes in a

manufacturing system and be able to apply the constraints to evaluate the

alternative processes and machines in order to choose the best ones. It can
reconfigure the processes quickly as needed to provide a new production and
produce variety of products within the same manufacturing system. An effective
methodology of process design can help to improve a manufacturing system
economically by enhancing production capability and adaptability to changing

market.

As a key element in integrating design and manufacturing, process planning is
defined in the Tool and Manufacturing Engineer’s Handbook [1974] as “the
systematic determination of the methods by which a product is to be
manufactured, economically and competitively.” As part of process planning,
process design could be defined as “the systematic exploration and
determination of the methods by which a process is to be designed, economically

and competitively.”

As a reconfigurable manufacturing system provides flexible functionality and
capacity in response to changes, the process for RMS should be designed based
on the characteristics of RMS. The process designed for RMS should have
following basic characteristics: (a) modularity, (b) convertibility, and (c)

customization.

(a) Modularity

Modularity of a process for RMS is achieved by its modular construction using
grammar-based approach. The modularity structure of the processes makes it
possible to add/remove processes to/from the current process diagram to meet
the changing manufacturing environment. For example, broaching keyway is a
process module for the production of pulleys, since it can be added or removed

for producing a pulley based on specific functions.

(b) Convertibility

Some of the process modules are capable of changing their internal components
or functionalities. The components could be process, machine, or other
resources (companies or operators) related to the process module. A process
module can maintain its own functionalities by including different alternative
processes as its internal components; it can also change the module
functionalities by choosing an alternate machine with additional process
capabilities to produce more features. For example, casting hole and drilling hole
are two alternative internal processes of a module for producing holes. A
machining center and a turning center are two alternate machines to run turning
process. The machining center can also run milling process. Therefore, if a
process module is carried out in a machining center instead of a turning center,
the module can produce the auxiliary holes besides the original production

functionalities.

(c) Customization

The process for RMS is designed for a given part family. The customization of
process has two aspects: customized resource and customized constraints.
Customized resource means that manufacturing resources (such as
organizations, machines, and human) are built around the part families that are
being manufactured with only the resources needed for those specific parts,
thereby reducing the cost. Customized constraint means that process engineers
or managers use the constraints based on individual experience to choose the
best alternative processes, machines, and providers; thereby optimizing the
processes. Customization enables industries to maximize the usage of

manufacturing resources.

1.3 Introduction of Grammar Approach
A “grammar”, as presented in Webster's dictionary, is defined as “the study of the
classes of words, their inflections, and their functions and relations in the
sentence.” The “generative grammar” is then defined as “a description in the
form of an ordered set of rules for producing the grammatical sentences of a

Ianguagef

As presented in much of literature, grammar-based approaches are already
applied for engineering design in different fields. Most of these grammar-based
approaches are generative approaches. They can apply specific rules on a set of

basic terms (such as shapes, processes, etc.) for producing a sophisticated

product or generating a complex process for producing the product. Grammars
model the engineering design as a linguistic formalism, allowing design to be
viewed as a domain independent activity and positing interesting research
hypotheses. Based on the literature reviews, three different grammar-based
approaches: shape grammar [Stiny 1980, 1992; Fitzhom 1990; Aganival and
Cagan 1998, 1999, 2000], graph grammar [Fitzhom 1986; Pinilla et al. 1989; Fu
et al. 1993; Hoover and Rinderle 1989], and process grammar [Baldwin and

Chung 1995a; Chung et al. 2002; Kwon et al. 2002] are presented.

1.3.1 Shape Grammar
A shape grammar [Stiny 1980] derives designs in the language that it specifies
by successive application of shape transformation rules to some evolving shape,
starting with an initial shape. It can be used to describe how complex shapes are
built from simple entities and how a complex shape can be decomposed into
simpler sub-shapes. In particular, given a finite set of shapes (8) and its labels
(L), a finite set of shape rules (R) of the form a —> [3 transform a labeled shape a
in (S, L)+ into a labeled shape [3 in (S, L)°, where (S, L)+ is the set of all labeled
shapes made up of shapes in the set S and symbols in the set L, and (S, L)° is
the set that contains, in addition to all of the labeled shapes in the set (S, L)", the
empty labeled shape <Sq>, <D>. Shapes themselves can be transformed with

Boolean operations.

Parametric shape grammars are an extension of shape grammars in which
shape rules are defined by filling in the open terms in a general schema. An
assignment, 9, gives specific values to all the variables in a and [3. It determines
a shape rule, 9(a) —, 9(3), which can be applied on a labeled shape in the usual
way to generate a new labeled shape. Algebras of shapes can also be
augmented by weights to obtain new algebras in which the shape union
operation has been redefined to reflect different possible weights on different

entities [Stiny 1992].

The shape grammar has been used successfully for spatial design in the field of
architecture. There are also some limited applications of shape grammar to
engineering design. Fitzhom [1990] presented a shape grammar specifying the
languages of constructive solid geometry and boundary representations (i.e.,
realizable solids). Agarwal and Cagan [1998, 1999, 2000] have focused on
developing shape grammars for a class of individual products, such as
coffeemakers and MEMS resonators and associating cost expressions with the
grammar rules [Aganival et al. 1999]. Aganival and Cagan [2000] also proposed
the use of shape grammars as expert systems for geometry-based engineering

design.

1.3.2 Graph Grammar
A graph grammar is a mathematical formalism for manipulating symbols

representing graph vertices and edges [Schmidt and Cagan 1997]. Grammar

rules assign functions and forms to vertices and connect vertices with edges. The
edges of a graph signify the relationship between two vertices. A completed
graph described with a set of vertices and edges expresses a design. There are
two kinds of graph grammar: one is for feature design through the representation
of features and their functions; the other is for process design through the

representation of processes and associated tasks constraints.

1.3.2.1 Design Features
Graph grammars have been applied to a variety of design applications. The
applications involve graph representations at different levels of detail, from solid
models and features, to components and their behaviors, to complete devices.
Fitzhom [1986] used a graph grammar to describe physical solids. Pinilla et al.
[1989] proposed a graph grammar to describe shape features such as notches
and perpendicular faces. Fu et al. [1993] also presented a graph grammar for the
representation and transformation of machinable features with primitive forms
and to propagate constraints. Hoover and Rinderle [1989] presented a technique
to transform a graph-based representation of specifications into an arrangement

of physical components for single-speed mechanical power transmissions.

The grammar approach shows a flexibility of the abstraction to express a wide
range of function and form relationships for the machine design problem.
However, the grammar rules required for sophisticated representations may be

too complicated and difficult to refine.

10

1.3.2.2 Process Grammar

In this dissertation, the core of the proposed grammar-based approach for
reconfigurable system is a process grammar, which provides the theoretical
foundation to represent, manipulate and execute the design and manufacturing
process [Baldwin and Chung 1995a]. This grammar-based approach has been
implemented in a system called Manufacturing Infrastructure and Design
Automation System (MIDAS). In the MIDAS system, D&M processes are
modeled as a collection tasks. Our process flow graph depicts tasks, data, and
the relationships among them, describing the sequence of tasks for a larger
activity. There are four basic elements used to represent the process flow:

. Logical task: It is a task that can be decomposed into a set of subtasks
(logical tasks, atomic tasks, or selectors). Logical tasks are represented
using oval nodes in the process flow graph.

. Atomic task: It is a task that cannot be decomposed into a set of
subtasks. Atomic tasks are represented by using two-concentric oval
nodes in the process flow graph.

. Specification or data: It is the input for executing a task or the output for
a finished task. An output specification produced by a task can be
consumed by subsequent tasks as an input specification. Specifications
are represented using rectangular nodes in the process flow graph.

. Selector: A selector selects a specification or a parameter for a task.

Selectors are represented using diamond nodes in the process flow graph.

11

These elements can be combined into a process flow graph using directed arcs
to indicate the specifications used and produced by each task. Specifications

with no incoming arcs are the first inputs to the process flow.

 

 

Four basic symbols
Logical task: single oval nodes 0
Atomic task: double oval nodes ©
Specification: rectangle nodes E
Selector: diamond nodes <>
[Part blueprint I I [Machine DBI I

 

 

 

 

I Part blueprint II IMachine DBI I@

 

E=x>pand@|
Making pulley casting I I Selected machine ®>I

<=|
Abstract
I Final pulley I w

I Final pulley I

 

 

 

 

 

 

 

 

 

Figure 1. A process for making a pulley

Using these graphical elements, a process flow graph can be created in a top-
down fashion. Figure 1 shows a high-level process flow of a process for making a
pulley. Process flow graphs can describe processes in varying degrees of detail.
The overall process can be seen because the details are hidden within the tasks.
A graph containing many logical nodes describes what should be done without
describing how it should be done (for example, specifying which tools to be
used). Conversely, a graph in which all task nodes are either atomic or selectors
fully describes a complete methodology for a design. In a complicated product,
the overall process can be visualized more simply after the detail tasks are

hidden from the overall process.

12

The process grammar provides the mechanism for transforming a high-level
process flow graph into progressively more detailed process flow graphs. As will
be discussed further, the grammar consists of a set of production rules. A
production is a substitution that permits the replacement of a logical task node
with a flow graph that represents a possible way of performing the task. Several
production rules with the same left side flow graph imply alternative production
rules for the logical task. Several production rules for a logical task imply there
are alternative production rules for that task. Casting and machining are, for
example, two alternative productions for the task manufacturing. This capability is
critical to maintaining the usability and effectiveness of the overall framework.
Therefore, the process grammar naturally captures the hierarchical design

methodology and allows systematic exploration of process space.

There are two ways for applying the process grammar: expansion and
abstraction. Figure 1 shows that a simple process flow graph in left side can be
expanded into a detailed process flow graph with a set of subtasks or a detailed
process flow graph on the right side can be abstracted into a simple process flow

graph after the detailed tasks are hidden.

Note that there may be many alternatives for each of these process steps:

different technologies, different vendors (companies), and so on. And at each

step, there is a “make or buy” decision, as well. Also, for the overall process to

13

critical information (such as design changes) and ideally, they need to be able to
provide feedback early in the process concerning the feasibility and schedule for

their part of the work.

1.4 Analytic Hierarchy Process for Process Selection
The Analytic Hierarchy Process (AHP) is a multi-criteria decision making method
developed by Saaty [1980]. It can decompose a complex problem into a
hierarchical order and use matrix algebra to solve the multi-criteria decision
making problem. The process selection in a real manufacturing environment is a
multi-criteria decision making problem. Many constraints with objective or
subjective values are used to evaluate lots of alternative processes. This
evaluation could be very difficult to make for a complex process. Considering the
characteristics of the grammar-based approach, it is possible to decompose a
complex process where an engineer or manager has to make many selections
among many alternatives into a smaller number of selections at any abstract
level. Therefore, the AHP method could be used to solve the process selection
problem through matrix algebra in a real industry environment. This is the most

important part in the dissertation and will be presented in detail in chapter 4.

1.5 Outline
Grammars are a means to generate a space of alternatives to a design problem.
The generate-and-optimize approach to the design problem involves three steps

[Schmidt and Cagan 1997]: the first step is the application of a method to

14

generate alternatives; the second step is the evaluation of feasible designs
among alternatives; and finally the third step is the application of an optimization
technique to select the best design. This dissertation presents a methodology of
process design as applied to manufacturing in the following steps: first, a
grammar-based approach is applied to generate alternative processes using the
case study on the production of pulleys; second, the important constraints for
producing pulleys are identified, which will be used to evaluate the alternative
processes; then an optimization technique, the analytic hierarchy process (AHP)
method, is used to select the suitable process based on the hierarchy

presentation of both processes and constraints.

In this dissertation, the proposed design methodology will be applied in an
industrial case: production of pulleys. At first, we use process grammar to
represent the processes for producing a pulley within a company; then we
capture the important constraints for evaluating the alternative processes so as
to facilitate the process design; finally, a hierarchy model for process selection is

set up for the case, and an implement method of process selection is presented.

In chapter 2, the case study of pulley’s production is presented. In this chapter,
the processes include external processes (such as casting) and internal
processes. A grammar-based approach is used to present the processes in a

distributed manufacturing environment. The alternative processes are generated

15

with process grammar, and the key constraints are captured to evaluate the

alternative processes.

In chapter 3, the structures of both processes and the constraints are discussed.
The common constraints used in a real manufacturing environment are listed and
organized into four categories. With the process hierarchy and constraint
hierarchy, the process reconfiguration is illustrated using these hierarchies with

examples.

In chapter 4, the implementation of process selection is described. Based on the
hierarchy structure of both processes and constraints, a hierarchy decision-
making model is presented and an effective selection method is applied on the
model. Some matrix calculations and the related modifications based on the

specific industry case are presented.

16

Chapter 2

CASE STUDY

An advanced methodology to represent design and manufacturing processes is
becoming essential due to the increasing complexity of the manufacturing
environment and the rapid change of the market requirements. Many companies
distributed around the globe can collaborate on various aspects of designing and
manufacturing a product such as quotation, design, manufacturing, assembly,
distribution, and service. Even within a given company, each department
specializes in particular aspects of the company’s business; and various
departments work together to carry out various tasks such as bidding, quoting,
designing, manufacturing, quality control, purchasing, etc. Within such a complex
environment, if new and unexpected conditions arise during the production, an
ideal process may not necessarily be the same process used previously with a
different set of parameters and must be reconfigured after evaluating a new set

of constraints. Alternative processes must be considered.

In a collaborative environment, tasks are interdependent: one task in one
organization affects other tasks in another organization. Thus, achieving
effective coordination among the participants throughout the life cycle of a
product is a challenging problem. We believe that part of the solution lies in a
common process representation that can be understood and shared by the

participants. With such a representation, each participant can understand his or

17

her task in relation to the overall process. Even within a company, developing a
new process, buying a new machine, or hiring a novice or experienced operator
can lead to more alternative processes. Furthermore, with many unforeseen
changes in design or suppliers, capturing the alternative process enables us to
reconfigure the process with a new set of constraints. This provides the incentive

of this research based on a real industrial case.

The case presented in this chapter focuses on a collaborative manufacturing
environment, which has the following characteristics:
0 Distributed participants: the involved participants, such as the design
company and the manufacturing company, may be around the world
. Various functions: within a company, there may be different departments
such as a department of engineering, quality, and purchasing etc.
o Frequent changes: e.g., design change leads to reconfiguring processes
o lnterdependent tasks: the tasks executed by different participants may
interact on each other
The study will provide a new paradigm where the grammar-based process model
is used to capture and reconfigure design and manufacturing processes with the
constraints in a collaborative environment. To apply the model in an industrial
practice, a case study has been conducted to capture the processes for making a

variety of pulleys at Focus: Hope in Detroit, Michigan.

18

2.1 Background
Due to rapid changes in the market, manufacturing enterprises need to respond
quickly while reducing costs through collaboration and use of alternative
processes and resources. In a collaborative environment, process
reconfiguration is necessary due to the changes in functional requirements,
manufacturing processes, production facilities, and manufacturability [Nau et al.
1994]. Reconfiguring a process requires the re-evaluation of the constraints that

are relevant to the process.

A variety of techniques to ‘engineer’ the processes are being developed to
address this problem. In mass customization, many firms use process
postponement, process re-sequencing, and process standardization [Feitzinger
and Lee 1997]. In a distributed manufacturing environment, when a new product
is introduced, group technology (GT) can support an efficient search for a similar
product family and retrieve all the information about the design feature and
manufacturing processes of the product [Candadai et al. 1996]. Agent-based
approaches are also popular. For example, at the enterprise level, agent-based
techniques have been developed to support decision-making in distributed
design and facility planning teams [Ratchev et al. 2000]. This approach makes
an iterative matching of design, process, and facility attributed by using multilevel
resource capability representation within the extended enterprise. At the
manufacturing floor level, a multiple criteria decision-making approach was

developed to obtain the best machining cell structures even with conflicting

19

objectives [Malakooti and Yang 2002]. It shows how to formulate and solve
multi-objective problem (such as maximize the machine utilization rate and
minimize the number of duplicated machines, etc.) by generating efficient

alternatives and applying the algorithm on the machine-part cell formation.

A major challenge in the modeling process is to capture, represent, and evaluate
a variety of tasks without explicit dependencies [Park and Cutkosky 1999]. They
also reviewed several existing methods. For example, digraph represents
processes with nodes and directed edges, but it is not effective to follow
precedence relationships and detect circuits among tasks nodes in a large
model. Project Evaluation and Review Technique (PERT) [Wiest and Levy 1977]
is a popular model with nodes representing task completion milestones and arcs
representing individual tasks. While the length of an arc is proportional to the
duration of the corresponding task, PERT can predict the expected completion
dates of a project. However, it neither provides the concept of process iteration
(rollback) nor shows the constraints explicitly. The Structured Analysis and
Design Technique (SADT) [Ross 1977] supports a structured graphical notation
for specifying the tasks, information inputs, and outputs in a large and complex
process model. However, it is difficult to maintain and, thus, prone to ambiguous
interpretations. Petri nets are effective to capture and analyze dynamic states of
a modeled system. However, Petri nets do not provide particular mechanisms for
abstracting process details and cannot generate representation to help

differentiate between process-related information and unnecessary details

20

[Murata 1989]. Other methods typically have limitations in providing multiple
levels of abstraction, precedence relationships, and constraints among tasks and

work products.

The grammar-based approach demonstrated here has the potential to resolve
these problems [Chung et al. 2002 and Kwon et al. 2002] for the following
reasons. First, as we shall demonstrate, the representation used in the
grammar-based process model can explicitly capture alternative processes. In
addition, the constraints are captured and evaluated to reconfigure the process.
Second, the grammatical structure should be relatively easy to maintain since
processes are described in a hierarchical structure. Finally, the grammatical
structure of the process model prevents the formation of circuits or causal loops

in the process representation.

This grammar-based approach has been implemented in a system called
MIDAS. In MIDAS, detailed task descriptions can be abstracted into a simple,
higher-level task. In reverse, a higher-level logical task can be decomposed into
a set of subtasks (other logical and atomic tasks). An atomic task cannot be
decomposed. To capture and represent a process, four basic symbols are used:
Logical Tasks are represented using oval nodes, Atomic Tasks are represented
using two-concentric oval nodes, Specifications or Data are represented using
rectangular nodes, and Selectors are represented using diamond nodes.

Databases are a special kind of Specification, which are represented using

21

rectangular nodes with two vertical lines on two sides (left and right) of the

rectangle.

The graphical decomposition of tasks into subtasks provides each participant
with a shared representation of the overall process [Chung et al. 2002]. Each
task uses a set of input specifications and produces another set of output
specifications, which are then used by subsequent tasks as input specifications.
Directed arcs indicate the specifications used and produced by each task to
represent the flow of these specifications. The four basic symbols together with
directed arcs are used to represent a process-flow graph. This notational system
facilitates multiple layers of abstraction and the explicit representation of

constraints.

The process grammar provides a mechanism to abstract 3 complex set of highly
dependent tasks into a simple, higher-level task (logical task) or to expand a
higher-level task into a set of lower-level tasks. When more than one alternative
process exists, one process can be selected by evaluating a given set of
constraints. Our process-flow graph depicts tasks, data, and the relationships
among them, describing the sequence of tasks. This representation is easy to
maintain, especially for families of related processes, such as those described
here. As will be seen in this research, the process-flow graph is effective in
capturing, representing, and sharing various processes being practiced at Focus:

HOPE, in Detroit, Michigan.

22

2.2 Modeling Processes
In this study, we concentrate on the internal and external processes in a real
manufacturing environment, at a company named Focus: HOPE in Detroit,
Michigan. The processes are specially used to manufacture a variety of pulleys.
We will represent alternative processes and the constraints used to reconfigure

the processes for a variety of pulleys.

2.2.1 Genesis of Alternative Processes
In this research, we used the ontology defined in Chung et al. [2002], in
combination with the process grammar, to generate alternative processes.
Processes are selected based on the attributes of four kinds of entities: Tasks,
Data, Resources, and Organization. Because some of these entities can
change, it is not possible to simply store and retrieve detailed process plans for
production within a collaborative environment. Process plans should be stored at
a different level of abstraction under a different management level and

specialized to fit the current circumstances.

Tasks - Many manufacturing processes can be used to produce the same part.
In producing a pulley, instead of starting with casting, a pulley can be
manufactured completely by machining out of a bar stock. This will be cost-
effective to make a small quantity of pulleys. Process planning must be made

based on dimensions and production volume. Other factors identified by Zhang

23

et al. [1999] include the geometry complexity, tolerance, the number of setups,
machining cost, alternative operations, and other specifications. A manufacturing

company needs to determine what can be done in-house or by outside suppliers.

Data - The design requirement contained in the blue print of a part (such as
material, geometry, and tolerance etc.) must be considered to choose different
processes. A material must be selected from a material database. Alternative
processes must be adopted depending on the selected material. For example,
polymeric materials cannot be produced with the same process used for
ceramics because completely different resources and expertise are required.
Also, based on the geometry and the tolerance, alternative processes can be

used to produce the preferred part.

Resources - A process planner selects and assigns a machine from the list of
available machines for a task. Each machine may have distinct process
capabilities and accuracy. Even for the same type of machines, the dimension,
tolerance, or position accuracy of the part can be different. A combination of
machines, instead of one, can be used to meet the design requirement, which

generates alternative processes.

Organization - Choosing another supplier can result in alternative processes as

each company brings in a different process capability or professional practices

with other organizations. Cost and lead-time are two main attributes when

24

choosing a supplier. Some other attributes such as quality status, location, and
credit history will be considered as well. Within a company, various departments
may have distinct roles in a project. Based on the function of different

departments, a task may be assigned to more than one department.

2.2.2 Representing the Overall Process
This study has been carried out through a series of the interviews with the people
involved in making pulleys at Focus:HOPE. Due to many different kinds of
pulleys manufactured in this plant, a sample of twenty pulleys was selected to
understand the processes involved in making the pulleys. Using the symbols and
arcs defined earlier, the common process reflecting all twenty pulleys after a

series of abstraction is represented in Figure 2.

Capturing the Overall Process
The processes shown in Figure 2 start with a set of drawings and other
requirements such as time constraints and production quantity requested by
Company A. Based on the drawings, Company B (Focus:HOPE) starts the
quotation processes that includes not only the cost to manufacture the pulley but
also all of other related costs including cast, tools, gauges, human resources,
etc. Based on a set of quotation prices from various companies’ bids for the
project, Company A decides on a supplier. Once Company B is chosen as the
supplier, Company B has to evaluate a set of companies for casting, gauges, and

tooling typically based on its past experience dealing with each of the companies.

25

A similar bidding process is necessary for Company B in selecting the suppliers
for gauges, casting and tooling. From the selected suppliers, the casting, gauges
and tooling will be ordered. Most of these are made in a small quantity initially to
test these parts in the production. At the same time, the detailed process plan
such as a sequence of machining processes, assignment of resources and

operators, and other processes necessary to produce the pulley is worked out.

Pullebelue print) I CompanyA I

Quotation Process Venders Company B

Fixtures Resources & Materials

 

 

 

Selected Resources Suppliers

 

/ I [IX X \

L- - . L. \L
/ aking ”Min“ (king Gau) Making Tools
Received COSI 8‘ Cost 8. Received
/ Castin ime 2a ime 2b Gau-es
/ CompanyC .\ Compa}l),F / CompanyE
M

 

    
  

    

 

 

    
 

 

 

 

 

 

 

 

 

Cost & Time 1
Cost 8 Time
Machined Part Established Price I Lead Time

 

 

 

 

Figure 2. Overall process of making pulleys

Manufacturing Environment

Figure 2 illustrates the complete process to make a pulley at Focus:HOPE. The
input/output specifications for each process task also reflect the necessary data
flows among the tasks. Five companies (denoted as Companies A, B, C, D, and
E for convenience) are working together to produce a pulley. Company A, which

designed the pulley, sends a document with the requirements for the pulley to

26

Company B; Company B makes a macro-level process plan, including Quotation
Process for the pulley, finds suitable suppliers for casting, tooling, and gauging;
manufactures the pulley; and analyzes the time and cost. Companies C, D, E
are selected by company B to provide materials such as casting and consumable

resources such as cutting tools and gauges.

 

I Pulley (que printlj _

 

Dept. of purchase,
7 _' tooling, and quality
IQesources & Materials]
Operations Find Suppliers

 

    
 
 
  
    

  
       

  
 

 

 

 
 

 

 

  
 
  

  

 

 

 

 

Selécted I
Dept_ of Resources Suppliers
material,
finance, sales, R ‘ d R ,
°"9‘"°°""9- I “9;"; t 9.5‘I’mesI ICost a. Time 2I
manufacturing a eria S

 

 

 

 

 

 

 

 

 

 

 

 

 

/
Dept. of finance,
Dept. Of _ and sales
manufacturing
@t & Time Anal©
:/ \
IMachined PartI IEstablished PriceI I Lead Time I

 

 

 

 

 

 

 

Figure 3. The process flow in company B

Figure 3 shows the process to manufacture a pulley within Company B after
abstraction. In this scenario, many departments such as sales, finance,
materials, engineering, purchasing, and quality are engaged to execute the
quotation process and to determine the required fixtures, operations, and
consumable resources (such as gauges and cutting tools). In the quotation
process, macro-level process planning is carried out to estimate the total cost to
produce a pulley. The purchasing department finds one casting supplier from its

casting provider list; the tooling department orders the cutting tools from a tool

27

supplier; and the quality department orders the required gauges from gauge
providers. The manufacturing department starts the production of the pulley once
the required consumable resources, such as cutting tools, are available. Finally,
the finance and sales departments calculate the total cost and time to determine

the price and the lead-time of the pulley.

Al_t_e_r_rlative Processes
As shown in Figure 2, Company B performs tasks such as Quotation Process,
Find Suppliers, Cost & Time analysis, Machining and Outsourcing (ordering the
required casting, gauges, and tools). Each supplier executes their tasks of
making or purchasing casting, gauges, and cutting tools. All the processes will be

described in the following sections.

Quotation process - As shown in Figure 4, the logical task, Quotation Process,
can be decomposed into four main atomic tasks: Feature Analysis, Assign
Resources, Operation Analysis, and Capability Analysis. For example, Gauges,
Tools, and Casting from the suppliers have to pass a Capability Analysis. Here,
the failure in the Capability Analysis will be resolved by reselecting another
fixture, cutting tools, or gauges in order to meet the requirement. Iteration may be

required to find an acceptable match.

28

   

Pulle
Cami Anal
Verified features
OR

u—I

I Fixtures. tools
and nuaues DB

blue rint

   
 
 
 
 
 

 

Pulley (blue print)

 
    
 

   
  

 
 

z)

   
    

Quotation Process

    

Selected
causes

       
 
 

Resources &
Materials

   

Operations I

Figure 4. A detailed quotation process

 

 

 

 
  

    

Resources 8.
Materials

l-I

    
 

 

I-I

Tddrooupp
uppliers

I_I l-I

ind Casting ind Gauge
Suppliers Suppliers
-OR-
I-I
list1
Inl
Su . . lier

asting Supplie
Selection

Selected Casting
Suppller

   
  

   
  

   
      
   
 
 

 
   
 

-OR-
ICIost8t TimIeI ICIost8. Timfl
list2 llst3

III-I
Su- lier
V

auge Suppli-
Selection

Selected Gauges
Supplier

 

   

  

Find Suppliers

 

Isl-I
u - - lier

ooling Supplie
Selection

 

   

 

 

 

Selected Tools
Supplier

 

 

 

 

 

 

elected Resources
Suppliers

Purchase dept. Quality dept. Tooling dept.

Figure 5. A process for finding suppliers
Find Suppliers — Figure 5 shows the process in which the consumable
resources and material suppliers are selected from the list of suppliers. It
includes an atomic task called Find Suppliers and a selector called Supplier
Selection. A common logical task, Find Suppliers, can be used but specialized
based on the input specification as shown in Figure 5. For example, based on

the input specification, Gauges, the quality department performs the atomic task,

29

Find Gauge Suppliers, to select a suitable supplier to produce an output

specification, Selected Gauges Supplier.

 

Selected
Gauges Supplier

Selected

elected Resources
‘ Tools Supplier

Suppliers

 

Selected
asting Supplle

 

 

 

 

 
      
   
     

Resources
8. Materials

    
       

I:> -OR- -OR-
.

Outsourcing

     

     

Received Resources
8. Materials

Received Received Received

 

Casting Gauges Tools
Purchase Quality Tooling
department department department

Figure 6. A process for outsourcing

 

 

 

Resources & Received Resources
Materials 8. Materlals

 

Resources 8. Received Resources
Materials 8 Materials

analys

   

    
   
    

         

Machining Process

kostat Time II IMachined PartI IEosts Tlme1I I Machined Part I

 

 

 

Figure 7. Process flow for machining
Outsourcing process — Outsourcing is a process of purchasing required
resources or materials produced or resold by suppliers. In Figure 6, the input
specifications for the outsourcing process are Resources & Materials and

Selected Resources Suppliers. It is very similar to Find Suppliers in terms of

30

using a common logical task and specializing into a more specific Select

Suppliers.

Machining process - As shown in Figure 7, Process Analysis generates Cost &
Time related to Physical machining processes and the documents that include
CNC programs, process sheets, picture sheets, and audit sheets. Machining
includes the machining operations with assigned resources such as machines,
fixtures, and other consumable resources. Depending on geometry complexity,
tolerance, and machine, the machining process may have many alternatives.
Table 1 shows the basic structure of a pulley and its required manufacturing

processes. A pulley is shown in Figure 8 with some described features in

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table1.
Table 1. Basic features of pulleys
Structure Processes
Components Features
Wheel Front flange Facing
Flange Rear flange Facing
Smooth O.D. Turning
Wheel Wheel O'D' Stepped 00. Turning
Groove Grooving
Chamfer Turning
Hub face Front face FacIng
Rear face Facrng
Axial hole Casting, drilling, boring, reaming
Counter-bore Counter boring
Hub I'D' Inside cone Taper boring
Inside groove Snap ring grooving
Smooth O.D. Turning
HUD 00' Stepped O.D. Turning
Chamfer Boring
Auxiliary hole Drilling
Internal thread Tapping
Keyway Broaching
Others Part No. Stamping
Balance holes Balancing
Rib, Boss Casting

 

 

 

 

 

31

  
  
 
  
 
  
   

Hub on.
Axtelhole

Auxiliary Hole
Balance Hole

Wheel O.D.

I Groove

Wheel Flange

Figure 8. Features of a pulley

Machined wheel Machined wheel Machined wheel

<i-F inish boring)

Semi-Bored l.D.

       
   
       

   
    
     
   

:> -on-
Finish machining |.D. @m
Reaming ID.

Finish Finish Finish
machined hub machined hub machined hub

Alternative A1 Tolerance = 0.0025 - Tolerance 5 0.0025 inch.
0.005 inch.

 

 

Figure 9. Alternative processes due to tolerance requirement
After reviewing over twenty pulleys, the processes can be classified into two
distinct types. The alternative A2 requires a reference plane from the finish
machined wheel. The detailed processes show more alternative processes
caused by geometric complexity, such as the hole represented in Table 2. The
alternative A1 has two more alternative processes depending on the process
used to introduce the inner diameter of a pulley: casting or drilling. Figure 9

shows two alternative productions that lead to the attainable tolerance for a

32

specific machining feature: Inner Diameter (I.D.). For a general tolerance
(0.0025~0.005inch), finish boring can meet the requirement; however, a better

tolerance (50.0025 inch) can be achieved by combined Semi-Finish boring and

 

 

 

 

 

 

 

 

 

 

Reaming l.D.
Table 2. Machining process flow of pulley
Alternatives Tasks Detailed Alternative Processes
i. Casting hole; ii.
1. Core drilling hole; 2.Counter boring; Counter boring; iii.
1. Roughing hub 3. Semi-finish boring; 4.Boring chamfer; Rough boring; iv.
5.Facing hub face; Boring chamfer; v.
A1 Facing hub face.
. . 6.Facing flange; 7.Tuming flange O.D.;
2. Machining wheel 8. gr oovi n g;
3. Finishing hub 9. Finish boring; i. Reaming
. . . 10. Breach keyway; 11.balancing;
4. Finishing pulley 12. st a m pi n g.
. 1.turning flange O.D; 2.facing flange;
1. Roughing wheel 3. g r oovi n g:
2. Roughing hub 4.counter boring; 5.boring:
A2 3. Finishing wheel 6.facing flange;
. . . 7.facing hub face: 8.tuming hub O.D.;
4' Finishing Mb 9.taper boring cone.
. . . 10.Drilling holes; 11.Tapping thread;
5' Finishing pulley 12. Balancing; 13.Stamping.

 

 

 

 

 

 

Table 3. Alternative processes due to assigned machine

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Alternative 1 Alternative 2

1.Turning flaLge O.D. 1.Turning flange O.D.
2.Facing front flaLge 2.Facingfront flange

'5 3.Facing hub face 3.Facing hub face

.3 4.Grooving_groove 4.Facing rear flange

a) 5.Rough and semi-finish boring l.D. 5.Grooving_groove
6.Reaming front l.D. ,_ 6§ough boring front

. 2 3 7.Rough back boring 2

7.Facmg flange g rear I.D.

N 8.Facing rear hub face 8.Reamingl.D.

g- 9.Facing bottom face

‘65 10.ReaminLrear l.D.

‘0 9 Drilling auxiliary
11.Drilling auxiliary hole hole
12.Tapping thread 10.Tapping thread

 

 

 

 

 

 

Table 3 illustrates two alternative processes to be chosen depending on the

assigned lathe. The rest of the process that was not performed by the lathe was

33

not presented here. The size of the chuck limits a certain machining process.
Machine denoted M1 can hold the pulley in a bigger chuck diameter (516 inch),
which allows all of the features to be finished in one setup. The machine denoted
M2 has a smaller chuck diameter (:12 inch), which forces the machining process

to take two setups.

2.2.3 Constraints in Manufacturing

As discussed earlier, the constraints must be evaluated to choose alternative
processes. Some constraints are used to pre-evaluate the alternative processes
for a task. Other constraints are applied as post-evaluation functions to check
the output specification. The post-evaluation is usually required to verify fixture,
tool, gauge, and operation sequence before the selected processes are applied
on the full-scale production. The pre- and post-evaluation functions are

presented according to the entities discussed earlier.

Constraints for Task
The inner diameter is a dimension of the pulley that could be produced by casting
or machining (drilling and/or boring). For machining a larger inner diameter,
drilling and boring processes are required since the diameter of a drill is limited.
With casting process, additional mold material for the hole must be added. The
cost comparison between extra mold and the additional machining is important in

making a decision. As a general rule, the pre-evaluation used in this case is if

34

the diameter is less than 1 inch, choose machining and, otherwise, choose

casfing.

Pro-evaluation - In Table 2, the geometric complexity of the casting is evaluated
to choose a process flow for the machining process task: Alternative A1 is good
for a pulley with a simple geometry, while Alternative A2 is suitable for a pulley
with a more complex geometry. Figure 9 shows two alternative processes where
the pre-evaluation compares the required tolerance against the tolerance that
can be achieved with a given process alternative. Generally, a tolerance of
0.0025~0.005 inch can be produced by finish boring while a better tolerance (less
than or equal to 0.0025 inch) can be achieved by combining semi-finish boring

and reaming.

Post-Evaluation - A post-evaluation is used to check if the output specifications
meet the constraints. The Capability Analysis on Figure 4 runs the test
operations with the gauges, fixtures, and cutting tools. The tolerance is
evaluated on a part produced as the post-evaluation. When the tolerance cannot
be met, iteration may be required to choose another (1) fixture, (2) tool, (3) gauge,
and (4) new operation sequence to meet the design requirement. An
experienced engineer may require less iteration. If the post-evaluation of

Capability Analysis is satisfied, then they are ready for full-scale production.

35

Wraints for Organization
As shown in Figures 4 and 5, the task of Find Suppliers (casting, gauge, or
tooling) finds the suppliers based on their process capability and production scale.
Cost and lead-time are two main constraints to select appropriate companies.
However, location and credit history (including quality and on-time delivery) are
additional constraints sometimes considered by the company. Once the planned
processes have been determined, particular departments are assigned to
perform the tasks. The basic functions of a department are the main constraint in
choosing which department should be responsible for each task. Figures 5 and 6
show the purchasing department performing the tasks of Find Suppliers and
Outsourcing that related to casting while quality department performs the tasks

related to the gauges.

Constraints for Resow
As discussed earlier, machine specifications such as maximum chuck diameter
and maximum power of a machine can lead to the selection of an alternative
machining process. In addition, each cutting tool has a specific shape capability
and tolerance level, which leads to different operation sequences. Similarly, the
assignment of human resources such as the domain expertise of an engineer or
the experience level of an operator is an important constraint for choosing certain
processes. For the processes represented in Figure 4, a process engineer
should be assigned to the task of Operation Analysis, and a quality engineer can

be assigned to Gauge Supplier Selection.

36

Constraints for Making Pulleys

Table 4. Typical process constraints for making a pulley

 

Entities Constraints

1. Maximum drill diameter = 1 inch (evaluating drilling or casting ID.)

2. Extra cost (from extra core mold or additional machining) (evaluating
drilling or casting ID.)

3. Geometric complexity of pulley (evaluating alternative process type
A1 or A2 in Table 2)

4. Designed tolerance of ID. =0.0025 inch (evaluating “finish boring” or

Task “semi-finish boring and reaming")

5. Minimum tolerance of ID. =0.0025 inch (evaluating turning center or
drilling machine)

6. Capability test of tools (gauge, fixture, cutting tool) and operation
sequences to meet the design requirement (tolerance, shape)

7. Dimension (I.D., depth, thread, runout, etc.) to match available

smiles

8. Process capability of supplier = (casting, tooling, gauging}

9. Production scale of supplier

10. Bidding cost

11. Bidding lead-time

 

 

 

 

 

Organization 12. Location of supplier
13. Credit history (including quality and on-time delivery)
14. Department function ={purchasing, tooling, quality, engineering, sales,
finance, manufacturim}
Data 15. Material of pulley = {gray cast iron or ductile cast iron}
16. Material of cuflirgtool = {carbide or ceramic}
17. Maximum chuck diameter = 16 inch (turning center)
18. Min tolerance = 0.005 inch (turning center)
19. Maximum power = 44 kW (turning center)
20. Shape capability = (O.D., I.D., Groove, flat face, chamfer, thread}
(turning center)
21. Process capability = {tuming, drilling, facing, grooving, boring,
Resource reaming, taper boring) (turning center)

22. Domain expertise of engineer ={process engineer, quality engineer,
tooling engineer}

23. Experience level of operator

24. Specification of gauge (size, grade of thread, tolerance of the thread)

25. Specification of cutting tool (shape capability, tolerance level)

26. Others (max turning diameter, Max turning length, Max weight, axis
number, etc.)

 

 

 

 

The entities and constraints for the pulley’s production system are summarized in
Table 4. The constraints associated with each entity are important to reconfigure

processes. The changes in some or all of the four entities may lead to

37

reconfiguring the process. Therefore, related constraints will help identify the part
of the process to be modified instead of going through the whole process. Further
research to identify the relation among different entities and their constraints are

presented in the following chapter.

2.3 Conclusion
This case demonstrates that grammar-based processes models are a feasible
method of capturing and representing the processes even for the complex
sequence of tasks in an industrial setting. Further research will be required to
determine the range of situations where grammar-based models are superior to
various alternatives. However, this case study shows that process grammar is a
useful way to represent complex tasks and the constraints associated each task
are evaluated to choose among alternative processes. We have shown that the
interaction among Tasks, Data, Resources, and Organizations generates
alternative processes. The explicit graphical representation makes it easier to
see each participant’s task in relation to the overall process and show important
constraints. The ability to handle multiple levels of abstraction provides a natural
way to deal with changes in context. Thus, process grammars can facilitate
coordination among the departments within an organization as well as among the

organizations.

38

Chapter 3

CONSTRAINT MANAGEMENT

In a collaborative manufacturing environment, a challenging problem for process
management is to capture and model the key constraints for reconfiguring the
processes. This is an important issue especially in Reconfigurable Manufacturing
System (RMS) where the constraints can constantly evolve. The changes of the
constraints in RMS emanates from, for example, producing a variety of products,
introducing new machines, implementing design changes, etc. To demonstrate
this, a case study was carried out at Focus: HOPE, Detroit, Michigan in
producing a variety of pulleys, as presented in chapter 2. Numerous constraints
affecting the machining and other related processes for producing these pulleys
have been collected. This chapter presents a hierarchical approach to capture
and model the constraints that are important in reconfiguring processes in the

proposed grammatical approach.

3.1 Introduction
Process is one of the most essential concepts in the development of a
Reconfigurable Manufacturing System (RMS). However, the theoretical
development in the representation and management of processes has not been
well established to make a major impact on RMS. The important consequence of
the modeling process is that, as new and sometimes unexpected conditions arise

during the production, an ideal process may not necessarily be the previous

39

process with another set of parameters. In many situations, alternative
processes must be considered and the most suitable process must be selected.
Therefore, to implement RMSs effectively, we must be able to capture, represent,
and reconfigure the processes in the environment where the constraints and

requirements are constantly evolving.

The main goal of a RMS is placed on the dynamical problem of continuously
adapting the technological solution at a lower cost [Koren et al. 1999]. An ideal
process modeling method must be able to represent the processes and the
associated constraints; and, at the same time, it must capture the specific
changes in the processes and the constraints that bring about these changes. It
is essential that such changes must be introduced systematically and their

impact must be clearly understood [Sadiq and Orlowska 2000].

Being able to reconfigure the processes requires capturing the constraints that
influence the processes as well as recommencing and re-evaluating the
constraints. Chapter 2 demonstrated that a grammar-based approach can
accomplish process representation and reconfiguration very effectively. A
grammatical representation allows both processes and constraints to be
represented in a hierarchy structure. In order to demonstrate the proposed
approach, we have captured and represented the machining and other related
processes for twenty pulleys through an extensive industrial case study at Focus:

HOPE in Detroit, Michigan. Focus: Hope produces well over one hundred

4O

different pulleys and we have limited ourselves for convenience to only twenty
pulleys. We will show how to reconfigure the process for each pulley and to
select and evaluate the necessary constraints to reconfigure the process. This
allows us to understand the inner working of the constraints that impose on the

processes.

3.2 Process Model Hierarchy
As presented in chapter 2, the proposed grammar-based approach can represent
the processes in a hierarchy structure at any level of abstraction. In order to
reconfigure the processes, the required constraints must be captured and
organized. When a logical task is expanded into detailed tasks, the most suitable
process must be selected among the alternative processes by evaluating the
necessary constraints. In chapter 2, Figure 3 illustrates the high-level
(abstracted) process to manufacture a variety of pulleys at Focus: HOPE. The
input/output specifications for each process task also reflect the necessary data
flows among tasks. Each task is executed by each department working
independently or cooperatively with other departments. The task may include
more sub-(or detail) tasks [Zeng et al. 2003]. For example, the logical task called
‘Outsourcing’ is carried out to purchase the required consumable resources and
materials that are produced or resold by suppliers; it includes three atomic tasks: -

‘Making Casting’, ‘Making Gauges’, and ‘Making Tools”.

41

A process hierarchy helps a participant to understand the overall processes and
the detailed processes at an appropriate level of abstraction. With any change in
the requirement and constraints, the hierarchy structure can quickly identify the
processes that are affected by the change. Figure 10 shows a process hierarchy,
in which the processes required for producing a pulley are described in multiple
levels. Producing a pulley is the top-level goal, with quotation process, cost and
time analysis, outsourcing, and machining in the second level. The process tasks
in the third level are the sub-tasks of the second level tasks. The atomic tasks
such as cost and time analysis in the second level, or feature analysis, making
gauge, find gauge supplier, etc. in the third level cannot be expanded. The logic
tasks will have sub-tasks in a lower level. Because of the limit of space, Figure

10 shows a hierarchy up to four levels only.

Producing a pulley
Cost and
Time Anal sis Sunliers

      
   

   
 

 

 

 

Quotation

  
 

 

Roughing achinin Finishing Finishing
hub wheel hub

Figure 10. Process Hierarchy in producing a pulley

42

3.3 Constraints for Process Models

3. 3.1 Categorizations
When representing the processes for producing the pulleys manufactured at
Focus: HOPE, an enormous set of constraints are identified and captured.
Several categories of the machining and other related process constraints are

necessary and thus presented here briefly.

Genesis

The previous chapter shows that the alternatiVe processes are generated from
four entities: Task, Data, Resource, and Organization. Considering an alternative
process resolves the limitation and/or improves the capability of existing
processes. Consequently, the constraints that describe the limitation and
capability of each process must be captured in order to reconfigure the
processes. They can be categorized by four geneses: task, data, resource, and

organization.

Element

The constraints are as well categorized by four elements: feature, process,
machine, and production. These elements are identified in our case study, which
are typically understood as the key elements in manufacturing. The feature
constraints are those that arise directly from the design requirement such as
tolerance, geometry, and material. The process constraints arise from the

specifications of the process without regard to the individual machine. These kind

43

of constraints, such as the tolerance range of a specific process, are normally
presented in handbooks and textbook. The machine constraints arise from the
specifications of an individual machine such as horsepower and maximum
tolerance that can be achieved. Such constraints can be an independent set of
constraints for each machine. The production constraints arise from the
constraints related to a production project such as cost, lead-time, etc. In section

3.4, a constraint hierarchy will be presented based on this category.

Obligation

Jones et al. [1996] evaluated the obligation of constraints for assembly planning.
As performed in Jones et al. [1996], we have classified the constraints into four
categories: requirement, prohibition, optimization, and suggestions. The
constraints for machining and other processes have a similar obligation, which
are represented as the following:

. Requirement refers to a constraint that defines the design requirement
from the specific features in the blue print.

- Prohibition: refers to a constraint that rules out unsuitable alternative
processes for executing a task.

0 Optimization: refers to a constraint that selects the best process among
the suitable alternative processes to maximize or minimize a scalar
function such as cost or tolerance. (MIN and MAX).

0 Suggestion: refers to a constraint that helps planners to generate a

suitable process plan based on either handbook or their own experience.

44

3.3.2 Constraint List
The constraints listed below are identified from the case study. A brief definition
of each constraint is presented in this section. As discussed in chapter 2, these
constraints may be applied when selecting a supplier, choosing a process among
many alternatives and so on. The constraints are listed here based on the

obligation categories presented above.

Optimization
Max-Credit-History: Maximize by selecting a supplier with the best credit history
such as quality and on-time delivery.
Min-Cost: Minimize the overall cost for producing a pulley.
Min-Cost-Casting: Minimize the cost of casting by selecting a suitable supplier.
Min-Cost-Fixture: Minimize the cost of fixtures by either selecting a supplier or
making them in house.
Min-Cost-Gauge: Minimize the cost of gauge by selecting a suitable supplier.
Min-Cost-Hole: Minimize the cost of producing a hole. The cost may vary
depending on the processes used such as drilling hole or casting hole.
Min-Cost-Machining: Minimize the overall cost of machining for a part by
planning a suitable machining process.
Min-Cost-Tool: Minimize the cost of tooling.
Min-Distance: Minimize the distance from the location of supplier. It relates to

the constraint, Min-Cost.

45

Min-Setup: Minimize the number of setups for producing a part, which may vary
depending on a selected machine.
Min-Time: Minimize the required time to execute a process task.

Prohibition
Prh-Axis-Number: The maximum number of axis in a machine. As the number
of axis increases, the flexibility of a machine improves requiring a less number of
set-ups.
Prh-Chuck-Diameter: The range of a part diameter that may be chucked into a
machine. It prohibits executing certain processes in a particular machine.
Prh-Domain-Expertise: The domain expertise of an engineer. It prohibits some
human who can execute a task.
Prh-Experience-Level: The experience level of an operator in operating the
machine.
Prh-Function-Dept: The function that may be performed by a department in the
company, such as the departments of purchasing, tooling, quality, etc.
Prh-Hole-Diameter: The range of hole diameters that may be accomplished in a
machine.
Prh-Horsepower: The maximum horsepower of a selected machine.
Prh-Machine-Type: The type of machine used.
Prh-Material: The types of material processed in the machine.
Prh-Process-Machine: The list of processes that can be run in a machine.
Prh-Process-Supplier: The process capability of a supplier.

Prh-Scale-Supplier: The production capability of a supplier.

46

Prh-Shape-Type: The shape types of a work piece that may be accomplished in
a machine such as flat face, axial hole, chamfer, etc.
Prh-Spindle-Speed: The maximum speed of the spindle of a machine.
Prh-Time-Schedule: The time schedule of a machine available for a part.
Prh-Tolerance: The range of tolerances of features that may be accomplished in
a machine.
Prh-Turning-Diameter: The range of external diameters that may be
accomplished by turning process in a machine.
Prh-Turning-Length: The range of part lengths that may be accomplished by
turning process in a machine.
Prh-Weight: The maximum weight of a part that may be chucked in a machine.
Reguirement
Req-Hole-Diameter: Designed diameters of holes on a part.
Req-Material: Material Requirement for a part.
Req-Quantity: Quantity Requirement of a part.
Req-Shape-Type: Particular shape types of a part.
Req-Tolerance: Tolerance requirement for a part.
Suggestion
Sug-Drill-Diameter: Suggest the range of drill diameters for a typical drilling
process. It includes Maximum-DrilI-Diameter and Minimum-Drill-Diameter.
Sug-Geometry-Complexity: Suggest a certain complex geometry for a part

based on the geometry feature and the size of the part.

47

accomplished by any process.

Sug-Hole-Diameter: Suggest the range of hole diameters that can be

Table 5. Classification of Constraints

 

 

 

Constraint Name Obligation Element
Max-Credit-History Optimization Production
Min-Cost Optimization Production
Min-Cost-Casting Optimization Production
Min-Cost-Fixture Optimization Production
Min-Cost-Gauge Optimization Production
Min-Cost-Hole Optimization Production
Min-Cost-Machining Optimization Production
Min-Cost—Tooling Optimization Production
Min-Distance Optimization Production
Min-Setup Optimization Production
Min-Time Optimization Production
Prh-Axis-Number Prohibition Machine
Prh-Chuck-Diameter Prohibition Machine
Prh-Domain-Expertise Prohibition Production
Prh-Experience-Level Prohibition Production
Prh-Function-Dept Prohibition Production
Prh-Hole—Diameter Prohibition Machine
Prh-Horsepower Prohibition Machine
Prh-Material Prohibition Machine
Prh-Process-Machine Prohibition Machine
Prh-Process-Supplier Prohibition Production
Prh-Scale-Supplier Prohibition Production
Prh-Shape-Type Prohibition Machine
Prh-Spindle-Speed Prohibition Machine
Prh-Time-Schedule Prohibition Machine
Prh-Tolerance Prohibition Machine
Prh-Turning-Diameter Prohibition Machine
Prh-Turning-Length Prohibition Machine
Prh-Weight Prohibition Machine
Req-Hole-Diameter Requirement Feature
Req-Material Requirement Feature
Req-Quantity Requirement Production
Req-Shape-Type Requirement Feature
Req-Tolerance Requirement Feature
Sug-DrilI-Diameter Suggestion Process
Sug-Geometry-Complexity Suggestion Process
Sug-Hole-Diameter Suggestion Process
Sug-Prod—Rate Suggestion Process
Sug-Tolerance Suggestion Process
Sug-Material Suggestion Process
Sug-Shape-Type Suggestion Process

 

48

 

Sug-Material: Suggest a particular material type.

Sug-Prod-Rate: Suggest that a particular production rate.

Sug-Shape-Type: Suggest a particular shape that may be accomplished by a
process.

Sug-Tolerance: Suggest a particular tolerance range of a process.

3.3.3 Classification of Constraints
As discussed above, the constraints can be categorized based on obligation and
element. Table 5 shows the constraint list in alphabetical order identified in the
case study. With the name of the constraints listed in column one, column two
defines the obligation of each constraint. For example, the constraint Prh-Chuck-
Diameter (the chuck diameter range of a machine), as will be shown later,
prohibits a certain process, as an additional set-up may be needed to accomplish

a task with a smaller chuck.

Column three defines the element of each constraint. These categories help to
identify the constraints that are affected more efficiently in reconfiguring the
processes. For example, when considering a new casting supplier, the
constraints belongs to the element “production” must be evaluated so that the
constraint Min-Cost-Casting may be reevaluated to minimize the cost of casting,
and the constraints Prh-ScaIe-Supplier and Req-Quantity will be applied to

choose the best suitable supplier.

49

3.4 Constraint Hierarchy
Chapter 2 shows that the grammar-based approach demonstrates the benefit of
using a hierarchy structure for process reconfiguration. As the processes are
abstracted or expanded, the constraints must be transformed accordingly so that
they can be evaluated at an appropriated level of abstraction to make a choice

among feasible alternative processes.

A Complete Set
of Constraints

A etc. etc.
. Req- I Req- Req— . Turning achinin-

 

t

etc. etc. etc. ec.
‘ Sug- Min- Min- Min— Min- Min- Prh- Prh-
Drill- Sug— Sug- cost- cost- cost- cost- cost- urning chuck Prh- Prh-
. . Tol. Mat. . . .. . . Tol. rave
Dia. casting fixture tooling machlnlng gauge Dla. Dla.

Figure 11. Constraint hierarchy for producing pulleys
To show this, the constraints used in this study are presented and classified
based on the obligation and element constraints shown in Table 5. A constraint
hierarchy is used to evaluate the constraints effectively. Figure 11 shows the
constraint hierarchy for producing a pulley. The complete set of constraints can
be subdivided as process, feature, production, and machine. At a specific
abstracted level, any constraints at a lower level can be evaluated to reconfigure

the processes. In the following sections, we will provide the constraints identified

50

for each of the four elements and how the processes are affected by the

constraints.

3.4.1 Production
The cost is an important constraint for an OEM (Original Equipment Manufacturer)
to choose a supplier or to determine an appropriate process for a task. The
constraint of Min-Cost is a measurement of the overall cost for producing a
pulley, which includes the costs of machining, casting, tooling, gauge, fixture, etc.
The constraint of Min-Cost helps an OEM to choose a supplier based on its bids.
Focus: HOPE is one of the suppliers bidding for the project. Other possible
constraints such as the lead-time, quality, and the distance from a supplier are

not considered here for simplicity.

During the macro-level process planning stage, each supplier chooses a
particular process for machining, and selects suppliers for casting, tooling, fixture,
and gauge. Each department in the company must implement a set of constraints
that affects the tasks assigned. For example, the purchasing department uses
the constraint of Min-Cost-Casting to choose a casting supplier, tooling
department uses Min-Cost-Tooling to determine the tooling supplier, the quality
department uses Min-Cost-Gauge to select gauge supplier, and the engineering
department uses Min-Cost-Machining to choose a suitable process plan for
machining the pulley. Therefore, the cost for each process must be summed up

as a total cost.

51

A simple relationship exists among the above constraints. A total cost of a pulley
is Sum {Min-Cost-Casting, Min-Cost-Tooling, Min-Cost-Machining, Min-Cost-
Gauge, Min-Cost-Fixture, etc.}. Thus, if three companies A, B, and C are able
to produce the pulley; the constraint, Min-Cost, evaluates the total costs

submitted by the suppliers in selecting a supplier:

Sum(i) = Sum{Min-Cost-Casting(i), Min-Cost-Tooling(i), Min-Cost-
Machining(i), Min-Cost-Gauge(i), Min-Cost-Fixture(i), etc.}

for company i where i=A, B or C;

The supplier who bid the lowest can be selected by the following constraint and
the total cost for the pulley is

Min-Cost = Min{Sum(A), Sum(B), Sum(C)}.

3.4.2 Machine
The constraints (e.g.: the specifications of a particular machine) associated with
each type of machines are different. In addition, due to the process capability of
each type of machines, alternative processes must be generated according to
the type of the assigned machine. Even for the same type of machines, the
process capability depends on the status of the machine such as the typical
tolerance each machine can achieve. Depending upon the specification, the

processes must be reconfigured. For example, the turning center can drill the

52

center hole only, and another milling machine is required to finish the drilling of
auxiliary holes on the flange; however, for a machining center that performs both

turning and milling processes, the time for changing each setup could be saved.

Based on the assigned machine, the constraint Prh-Machine-Type can be
implemented for a high-level process. As shown in Figure 11, the constraints,
Prh-Chuck-Diameter and Prh-Tolerance, can be implemented to evaluate the
alternative processes for the sub-tasks in a low-level. This shows that a
constraint hierarchy can be used to evaluate the alternative processes in an

appropriate level related to the process hierarchy in the case study.

Wheel flange Wheel flange

   
  
  
 
 
  

Auxiliary hole

Figure 12. Basic Features of a typical pulley

53

 

Max chuck diameter< D2 Max chuck diameter >= 02

(a) Machine Ml: (b) Machine M2: (c) Machine M3:
Turning Center Tuming Center Machining Center

Figure 13. Alternative Processes depending on the assigned machine
The general features of a pulley are shown in Figure 12. Figure 13 shows the
alternative processes for the task, machining, generated for each of three
machines assigned to machine the wheel and hub of the pulley with the circular
holes. In Figure 13, machine M1 and M2 are turning centers, while M3 is a
machining center. For M1 and M2, additional milling center are required to finish
all of the features of circular holes. If the machining center is not available, the
processes related to a machining center can be disregarded. Other alternative
processes will be evaluated by the constraints coming with the turning center,

such as Prh-Chuck-Diameter and Prh-Tolerance.

Table 6 illustrates the alternative processes generated by each of the machines
assigned. Alternative 3 is for the machining center in which all features can be

done with only one setup. Alternative 1 and alternative 2 are the processes for

the turning center. The bigger chuck diameter enables it to finish all the features

with fewer setups.

A relationship between machine constraints and feature constraints exists. The
prohibition constraints of an assigned machine such as Prh-Tolerance, Prh-
Shape-Type and Prh-HoIe-Diameter must be related to the feature constraints
such as Req-Tolerance, Req-Shape-Type, and Req-Hole-Diameter as follows:
Req-Tolerance ; Prh-Tolerance

Req-Shape-Type ; Prh-Shape-Type

Req-Hole-Diameter _c_ Prh-Hole-Diameter

Above relationships represent that if a machine can accomplish a design feature
of a part, the design requirement (tolerance, shape-type, or hole-diameter) of the
feature will lie in (or belong to) the capability range (tolerance, shape-type, or
hole-diameter) of the machine. For example, Req-Tolerance represents the
tolerances of design features of a part, while Prh-Tolerance defines the
tolerance range of features that a machine can accomplish. If a designed hole
has a tolerance of 0.001 inch and a turning lathe can boring a hole with a smaller
tolerance up to 0.0005 inch; since the designed hole tolerance belongs to the
tolerance range of the turning lathe, the turning lathe could be used to produce

the hole.

55

Table 6. Alternative processes due to the assigned machine(s)

8.Reaming I.D. 8.Reaming l.D.
9.Drilling auxiliary hole

9. Reaming I.D.
11.Drilling auxiliary hole 9.Drilling auxiliary hole

12.Tapping thread 10.Tapping thread 10.Tapping thread

 

  

 

 

    

Machined wheel

(i-Flnish boring I.)
l:> -on-
Finish machining I.D. Finish bo®

Finish
machined huh

I Machined wheelI I Machined wheel I

 
 
  

    

 

    

Semi-Bored l.D.

Reaming I.D.

Flnish
machined hub

 
 
  

      
   

      
 

Finish
machined hub

    
        

Tolerance = 0.0025 ~ Tolerance=0.0005-
0.005 inch. 0.0025 Inch.

(a) Cincinnati turning machine

 

 

Machined wheel I I Machined wheel I I Machined wheel I

 

 

  
  

 
 

   

.OR.

Finish machining I.D. Finish boring I.D. Reaming I.D.
Finish Finish Finish
machined hub machined hub machined hub

Tolerance = 0.001 ~0.003 inch Tolerance = 0.001~0,003 inch
Diameter ofl.D.=0.1- 48 inch Diameter of I.D.=0.013~6 inch

Cost more, for low production Cost less, for any production

(b) Drilling machine [8]

 

 

i::>

     
   

     

           

Figure 14. Change of constraints and alternative processes depending
on the assigned machine

56

3.4.3 Process
Many of the constraints are related to the process capabilities such as the
geometry shape (Sug-Shape-Type) and dimensional tolerance (Sug-Tolerance).
When a specific process type is considered, its related constraints may be
applied. Drilling and casting are alternative processes for producing the internal
diameter for a pulley. The associated constraints for drilling or casting may be
applied to evaluate the alternative processes. The constraint Sug-DriII-Diameter
helps to identify the maximum drill diameter and to determine if the hole can be
accomplished directly by drilling. The constraint Sug-Tolerance and Sug-
Material can be applied to choose the specific casting process (e.g.: sand
casting or die casting) for producing the hole based on the tolerance requirement
(Req-Tolerance) of the hole and the material requirement (Reg-Material) of the
pulley.
A relationship exists between the process constraints and the feature constraints.
When planning the machining process for the hole of a pulley, some relationships
may be described as following:
Req-Hole-Diameter ; Sug-Hole-Diameter
Req-Tolerance g Sug-Tolerance
Req-Shape-Type ; Sug-Shape-Type
For drilling process, the following relationship, Req-Hole-Diameter c_:_ Sug-Drill-

Diameter, must be considered instead.

57

These relationships represent that if a process can be chosen to produce a
design feature of a part, the design requirement (tolerance, shape-type, or hole-
diameter) of the feature will lie in (or belong to) the capability range (tolerance,

shape-type, or hole-diameter) of the process.

3.5 Process Reconfiguration with Constraint Hierarchy
Constraint hierarchy provides a useful tool to help to reconfigure the process due
to the change of manufacturing systems such as buying a new machine,
developing a new process, and installing a new manufacturing system. Only the
constraints affected by the change need to be re—evaluated. The hierarchy
structure of constraint helps to identify the constraints that are re-evaluated and

to reconfigure the processes efficiently.

3.5.1 Machine Assignment
Figure 14 shows how the assignment of a new machine can affect the
constraints in choosing a process. At least two alternative processes for the task
of Finish Machining Internal Diameter (I.D.) exist. If a turning machine is
assigned as in Figure 14 (a), then the constraint Prh-Tolerance under the
machine element must be used. Each alternative process produces its tolerance
for the ID. However, if a drilling machine is assigned, both alternative processes
shown in Figure 14(b) can produce the same tolerance. Therefore, the constraint
Prh-Tolerance is not enough to decide between the alternatives since no

difference can be found. Another constraint in the machine element, Sug-Hole-

58

Diameter, may be used. If the given hole—diameter can be produced with two
alternative processes, another constraint in the production element such as Min-
Cost-Machining has to be evaluated.

Table 7. Constraints associated to the tasks

 

 

 

 

 

 

 

 

 

 

 

Tasks Constraints level
Level 2 Level 3 Level 4
Quotation Production Min-Cost, Min-
process Time
Find Production Min-Cost Min-Cost-Casting, Min-
suppliers Cost-Gauge, Min-Cost-
Tool
Machining Production, Min-Cost, Req- Min-Machining-Cost
feature tolerance
Machining Machine, Tuming-machine, Prh-Process-Machine,
process feature, Req-shape-type, Prh-Hole-Diameter, Sug-
process Req-Hole- Drill-Diameter, Sug-Hole—
Diameter Diameter

 

3.5.2 Constraints related to Process Hierarchy
Table 7 shows some tasks in the process hierarchy and the related constraints at
different levels in the constraint hierarchy. For reconfiguring the process, the
constraints at a specific level can be applied on the task at an appropriate level to
evaluate the alternative processes. When planning the task “Machining”, we may
apply the production constraints such as Min-Machining-Cost and the feature
constraints such as Req-Tolerance to choose alternative processes. However,
for executing the lower level task “Machining Process”, we need to apply the
machine constraints such as 'Prh-Process-Machine and Prh-HoIe-Diameter,
feature constraints such as Req-Shape-Type and Req-Hole-Diameter, and
process constraints such as Sug-Drill-Diameter and Sug-Hole-Diameter to

choose a suitable alternative process.

59

Table 7 illustrates the possible relationship between process hierarchy and the
constraint hierarchy shown in Figure 10 and Figure 11. The constraint at a
different level in the constraint hierarchy may be associated with the process task
at an appropriate level in the process hierarchy. The levels of constraints
present the priorities of the constraint to be applied on a process task. The lower
is the level (with a higher number), the higher priority has the constraint. The
lower-level constraints should be applied first on a process task before applying
higher-level constraints on the task. A process task at a higher abstract level
could use an upper-level constraint to evaluate the alternative processes. If a
constraint is affected by some change, only its associated constraints at upper
levels (with a lower number) need to be reevaluated for their associated process
tasks. Thus, the constraint hierarchy may help to identify the specific constraints
for the process reconfiguration as the change in the manufacturing system

affects the process.

3.6 Conclusion
In this paper, the hierarchy structures of both process models and the constraints
in a manufacturing environment are presented. The grammar-based approach
exhibits the following attractive features: (1) it captures and represents the
manufacturing process at any level of abstraction. This will help engineers to
understand the overall process and explore more detailed processes, if desired.

(2) It provides an effective method to reconfigure the processes with the

60

constraints affected by any change in manufacturing. (3) It constructs a
necessary foundation for an open-architecture system that manages the overall

process in a manufacturing environment.

61

Chapter 4

IMPLEMENTATION OF PROCESS SELECTION

In Chapter 3, both processes and constraints are represented in a hierarchy
structure and shown the effectiveness of the structure in a real manufacturing
environment. The hierarchies represent the relationships among the processes
and the constraints in the manufacturing environment. The proposed
grammatical approach can not only represent the processes in a hierarchical
structure but also capture and model the constraints in a hierarchical structure.
The constraint hierarchy may help us to reduce the number of alternative
processes and even to select an appropriate alternative process at any level of
abstraction. However, a major obstacle in implementing the hierarchies is how to
select a suitable process among many alternatives. A decision-making method
with the hierarchy structures intact (without getting into more detailed processes)
is required for the process selection problems in a complex manufacturing

environment.

4.1 Introduction

4.1.1 Decision Making Method for Hierarchies
The Analytic Hierarchy Process (AHP) developed by Saaty [1980] is one of the
multi-criteria decision making methods that decomposes a complex problem into
a hierarchical order. Three important steps for AHP are (1) structuring the

problem as a hierarchy consisting of a goal and subordinate elements

62

(decomposition), (2) making a pair-wise comparison between elements at each
level to determine the preferential order of the elements (evaluation), and (3)
propagating level-specific, local priorities to global priorities through matrix
algebra (synthesis). Subordinate levels of the hierarchy may include objectives,
scenarios, events, actions, outcomes, and alternatives. Alternatives to be
compared appear at the lowest level of the hierarchy. A decision at a higher level
will be made based on the global priorities of alternatives that are calculated
through matrix algebra beginning from a lower level. Pair-wise comparisons are
made between elements at a particular level with respect to elements in the level

above it.

When structuring a hierarchy for manufacturing related problems with the AHP
method, the subordinate levels of the hierarchy may include objectives,
constraints, and alternatives. These three constituents have been identified in
this study. The subordinate elements in the hierarchy could be processes,

constraints for evaluating alternatives, machines, organizations, and materials.

The AHP method presented by Saaty [1980] can be described as follows. Given
the elements (like constraints) of one level in a hierarchy and one element (like
process) of the next higher level, the elements of constraints are compared pair-
wise in their strength of influence on the process. Numbers reflecting the result
of the comparison are inserted into a matrix to find the eigenvector with the

largest eigenvalue. The eigenvector represents the priority ordering of each

63

constraint element on the process, and the eigenvalue is a measure of the

consistency of the judgment.

The pairwise comparison between elements helps to determine the priority
importance of the elements with respect to one element of the next higher level.
The relative importance can be denoted by a,-,-, the number indicating the strength

of element i when compared with element j. The matrix of these elements is

denoted A, or

A = (all) (4-1)
The weights (W1, W") are the measurements of the priority importance of
elements (1, n) with respect to one element of the next higher level. The

larger the weight, the more important is the element. Therefore, 8;} represents the

ratio of the weight of element ito element j. It is known as

a; = Wi/Wj L] = 1, ..., n (4.2)

For the decision-making problems with hierarchies, we want to determine the

weights of hierarchy elements at each level. With the matrix consisting of 8"], the

estimated weight vector w is found by solving the following eigenvector problem:

Aw = kmaxw (4.3)

64

where the matrix A=(a,-,-), and in,“ is the largest eigenvalue of A. In the consistent
case, there exists a”, = a,-,-- ajk for all i, j, and k; then n (the order of matrix) is the
only non-zero eigenvalue of A, and obviously n is the largest eigenvalue, lmax,
[Saaty 1980]. Thus, we can use the equation (4.3) to find the eigenvector w. In
the inconsistent case, the equation a”, = a,-,-- ajk may not exist; i.e., if the relative
importance of element 1 is greater than that of element 2 and the relative
importance of element 2 is greater than that of element 3, then the relation of
importance of element 1 need not be greater than that of element 3, a common
occurrence in human judgments. Thus, the largest eigenvalue, 11mm, of the matrix
A is not equal to its order n. The largest eigenvalue is the measure of the
consistency of the matrix A: the closer the largest (or principal) eigenvalue close
to n, the more consistent of the matrix A is. Therefore, if the largest eigenvalue
closes to n, we can still use the equation (4.3) to find the approximate
eigenvector w. Some concepts are used to evaluate the consistence of the result,

as presented later.

The eigenvector, w, represents the priority ordering, which can be normalized by
making all numbers in the eigenvector to be summed up to unity. Each number is
divided by the sum of all number in the eigenvector. In this chapter, the weights
are obtained by normalizing the eigenvector of the largest eigenvalue of the
evaluated matrix. Alternative weights represent the priorities of alternatives

with respect to one constraint of the next higher level; constraint weights

65

represent the priorities of constraints with respect to one objective of the next
higher level; and global weights represent the priorities of alternatives with

respect to the objective.

4.1.2 Consistency Analysis
There are several concepts used for consistency analysis with the AHP method
[Saaty 1980]: Consistence Index (C.l.), Random Index (R.l.), and Consistence
Ratio (C.R.). They are represented as following:

. Consistence Index: it represents the deviation from consistency, which
can be calculated by (linu- n)/(n-1), xmax is the largest eigenvalue of the
matrix, and n is the order of the matrix.

0 Random index: it is the average consistency index of a randomly
generated matrix from the scale 1 to 9. Its value varies depending on the
matrix order. Table 8 gives the order of the matrix and the average R.l.
[Saaty 1980].

Table 8. Random index of the matrix

 

Matrix order 1 2 3 4 5 6 7 8
Random Index 0 0 0.58 0.90 1.12 1.24 1.32 1.41

 

 

 

 

 

 

 

 

 

 

 

 

o Consistence Ratio: it is the ratio of CI. to average R.I. for the same order
matrix, which is calculated by GR. = C.I.lR.l.. For practical problems, the
decision matrix is considered to be consistent if has a consistency ratio of

0.10 or less, which means C.R.<0.1.

66

With the consistence ratio, we can evaluate the result (weight vector) calculated
from the matrix of elements. If the matrix is consistent, the weight vector is
acceptable, otherwise, the matrix needs to be revised; and the judgment of

relative importance between elements needs to be modified.

4.1.3 Application of AHP on Manufacturing Problems
The process selection in a real manufacturing environment is a multi-criteria
decision making problem. When evaluating each alternative process, many
constraints have objective or subjective values. These constraints are the criteria
to decide on an optimum process. As one of multi-criteria decision making
approaches, the AHP method uses a ratio scale to quantify relative weights for a
given set of criteria, which are represented in a hierarchy structure. Two features
differentiate AHP from other decision-making approaches [Akarte et al. 2001]: (1)
it provides a comprehensive structure to combine the intuitive objective and
subjective values during the decision making process, (2) it can judge the

consistency in the decision-making process.

The AHP method has been used on various manufacturing related problems.
Valerie and Hu [2002] used AHP for the problem of choosing manufacturing
system configurations with consideration of multiple performance criteria:
productivity, quality, convertibility, and scalability. Rangone [1996] stated that the
AHP framework could effectively compare the performance of manufacturing

departments. The overall performances are evaluated in three aspects: quality,

67

flexibility, and environmental compatibility. Luong [1998] combined the AHP and
database technology on a decision support system for the selection of computer
integrated manufacturing technologies. Tiwari and Banerjee [2001] presented an
AHP based decision support system to select the most suitable casting process
for a given product. Akarte et al. [2001] also discussed the selection of a casting
supplier by identifying the criteria for casting supplier assessment and
segregating them in four groups: product development capability, manufacturing
capability, quality capability, and cost and delivery. They implemented the AHP
method on a prototype web-based system. Chan [2003] further discussed an
interactive selection model for supplier selection process by systemizing the
earlier steps, such as the determination of buyer-supplier relationships and
formation of selection criteria before the implementation of the AHP. Yurdakul
and Cogun [2003] developed a multi-attribute selection procedure with AHP for
non-traditional machining processes. Abdi et al. [2003] presented a design
strategy for reconfigurable manufacturing systems. They applied the AHP model
on structuring the decision-making process for the selection of a manufacturing
system among feasible alternatives based on the RMS study. The summary of

the applications on manufacturing problems is described in Table 9.

68

Table 9. Application of AHP on manufacturing problems

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Groups Manufacturing Performances Others
problems Criteria Sub-criteria
Rangone Selection of 0 Quality Conformance-rate, inspection
(1996) manufacturing . Flexibility costs, rationalization degree,
d rt 1 . Environmental product flexibility, volume
epa men wmpafibiiiiy flexibility, technology
flexibility, solid waste, energy
consumption, factory green
image
Luong Selection of . Increased mfg lead lime Combined with
(1998) IM ' Increased productivtty database
- . Reduced inventory and work in progress
teChaomgclefD . Increased quality (reduced scraps and rework) “32182893; . d
(SUP as . . Increased flexibility ( ). °n$l 9"
r . . Increased integration in the oornpany the company
F MS, CAPP, characteristics
MRP) (production type
and main function
of a company)
Tiwari and Selection of - Material §uilabllity8iflexlbllity
Banerjee casting process ° elemefnw' °°ml"e"'ty ,
(2001) (such as sand . Dimen5lonal accuracy & surface finish
. . . Cost
casting, die
casting, shell
moldin
Akarte, Selection of - Product Melflmum ¢3$iin9 Size. Combined with a
Surendra, casting supplier development minimum section thickg‘e’ss, web-based
R . d (Web-base d) capability casting oomplexIty. so are a roach to
aVl. an . Mfg capability aid; pattern making, sand pp
Rangaraj . Quafity preparation, molding, core evaluate the
(2001) Capability making. melting and nonring: casting supplier.
. Cost and time heart treatment, machining,
dimensional tolerance,
surface roughness, testing
facilities; quality certification,
quality awards, total casting
cost, sample delivery time
Valerie Selection of - Productivity Expected production, Mean
and Hu manufacturing 0 Quality deviation, standard deviation,
2002 t . Convertibility number of flow-paths,
( ) sys _em _ . Scalability minimum increment,
configurations costfincrement
Chan Interactive (Criteria aI different levelS) __ . .. Before selection,
(2003) selection of Cost, quality, desrgn capability, technical capability, the buyer-supplier
. technological capability, performance history, . .
SUPPI'er management capability, degree of cooperation, relationships-need
financial performance, degree of closeness to be determined to
form selection
criteria
Yurdakul Selection of Ragtime. Salrlfacd? finlsth. Sglfafidflamatge. 02mg
_ - - ra II, per, oe tame er, ep lame erra c or
and non rt‘radittonal cylindrieal holes), depth/width ratio (for blind
09“" mac Inlng cavities), width of cut, material removal rate
(2003) processes
(such as AJM,
WJM, etc.)
Abdi and Selection of Level 11 planning horizons (long/medlum/short lenn): A design strategy
Labib manufacturing L2: decision makers (plant manager/shop floor for RMS is
manager/mfg designer):
(2003) systems (DMS, L3: objectives (responsiveness/product cost/product presented.
FMS, RMS) quality/inventory/operator skills);

 

 

L4: sub-criteria of L3.

 

 

69

 

As shown in Table 9, the criteria used for each manufacturing problem are
different. When selecting a suitable process among alternatives, different criteria
must be used for each manufacturing process due to the difference in the
problems of each process. Table 9 shows the main fields in which AHP could be
applied for following manufacturing problems:

0 Selection of organizations such as company or internal department

0 Selection of resources such as machines or suppliers

0 Selection of manufacturing systems such as FMS, RMS, DMS

- Selection of configurations for a specific system

0 Selection of processes and CIM technologies
There are two ways of matching machine alternatives for processes:

0 Associate each machining process with alternative machines in which the

process can be carried on.
0 Associate each machine with alternative processes that the machine can

carry on.

However, Rebstock and Kaula [2003] presented that there is no single hierarchy
for particular problems such as supplier selection and process selection. For
example, it is difficult to construct a hierarchy diagram to evaluate hundreds of
alternative processes at one level and make a selection with all the criteria

needed.

70

Considering the characteristics of the grammar-based approach, it is possible to
decompose a complex process where an engineer has to make many selections
among many alternatives into a smaller number of selections at any abstract
level [Zeng et al. 2003]. Thus, the AHP method combined with the grammar-
based approach becomes a convenient solution for the process selection

problem.

In this chapter, we will apply the AHP on the process selection problem for the
shop-floor environment. First, a hierarchy model is established with the process
constraints and machine constraints, which are used to evaluate alternative
processes and machines. The performances of each candidate process/machine
are then evaluated. Several case studies using the AHP method will be

described in this chapter.

4.2 Multilevel Multiple Criteria Decision Making (MMCDM) Model
The process selection could be a complex problem when selecting both a
process and machines. Various constraints are used to evaluate each alternative
processes as well as alternative machines. It is difficult to make a selection by
considering all alternatives at the same time. In many situations, the selected

process does not guarantee that the selected process is the most optimum one.

71

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Level 1: Objective i Goal i \
Level 2: Constraints I_C] C2 CL] I Procejss
selection
Level3: Alternative U1 I A2 I I A3 I J
pi'OCCSSCS
\
Lei/614: Sub- I Cll ][ CIZj chl I] C22] LC3I J
Constraints sub.
\1 I level
Level 5: Sub- seleaion
Al‘emat‘ves I All 1] A12 l[A1flLA2fi[A22][A3flI A32 1

 

 

J

Figure 15. Multilevel Multiple Criteria Decision Making (MMCDM)
Figure 15 shows a multi-level decision model. The main goal is to select a
suitable process among alternative processes after evaluating the constraints (or
criteria) at level 2. In addition, more selection processes will be required in order
to identify suitable machines to carry out the alternative processes, or to choose
sub-processes for each alternative process. The machines or sub-processes are
the sub-alternatives to be evaluated by the sub-constraints at level 4. Thus, in
Figure 15, the decision making model shows the selections at two abstract levels.
The first level is the process selection without the details of machines or sub-
processes. The second level is a selection, sub-alternative selection, to choose
suitable machines or sub-processes with more details involved. More selections
at lower levels may be required when other detail information is captured. This
model avoids making a selection among many alternatives. It decomposes a

complex selection problem as a hierarchy selection problem. Many selections

72

with many alternatives are decomposed into a smaller number of selections at

any abstract level with reasonable constraints evaluating reasonable alternatives.

With this multilevel multiple criteria decision making (MMCDM) model, we can
use the AHP method and the process/constraint hierarchy to represent a
complex selection process as several simple selection processes in the
hierarchical order by evaluating quantitative and/or qualitative constraints. The

following case studies will describe the detail evaluation.

4.3 A Case Study for Applying MMCDM

 

Level 1:

, _ I Process selection for central hole I
Objective

 

 

Level 2: I Cost I I Tolerance I I Max.diameter I Process

\ > / selection
><>4
rm. 3 Alt

 

 
  

Constraints

 
   

 

Alternative \ 1“ (
processes: )"$§
Level 4: . . . . .
Drilling Boring Min. Production
Sub- cost cost tolerance rate Machine
constraints \ A» 0 (’4 1 '
~ ‘9: <. 56 CCtlon
>«’\\
Level 5: TCI TC2 MC
Sub-

alternative
Figure 16. Case study of MMCDM
As shown in Figure 16, a five-level hierarchy shows how to select a process for
producing a hole with suitable machines. Level 1 represents the objective. Level
2 represents the process constraints for evaluating alternative processes at level

3. Level 4 contains the machine constraints for evaluating alternative machines

73

at level 5. In order to produce a central hole for a pulley, three alternative
processes (A1, A2, and A3) are represented as following:
0 A1 (casting, drilling, and boring): drilling and boring are used to finish
the inner diameter with a casting hole.
0 A2 (cutoff, drilling, and boring): drilling and boring are required to finish
the inner diameter with the blank from the cutoff of a bar stock.
. A3 (casting, boring): boring is used to finish the inner diameter with a
casting hole.
The constraints at level 2, such as cost, tolerance, and maximum diameter (of
the hole), are applied to evaluate the alternative processes. Before choosing the
suitable alternative process, a suitable machine should be selected with respect
to each alternative process. The constraints at level 4 (drilling cost, boring cost,
minimum tolerance, and production rate) are applied to evaluate the alternative
machines: two turning centers (TC1 and T02) and one machining center (MC).
The AHP method will be used to select a suitable process for producing the hole

with the most suitable machine.

Figure 16 shows a two-level selection model. The low-level selection, called
machine selection, chooses a suitable resource (machine) for each alternative
process of the next high level. The high-level selection, called the process
selection, chooses a suitable alternative process with the selected resource
(machine) to achieve the final objective. Therefore, the two-level selection

process always selects the best option at each level, which guarantees an

74

acceptable use of the selected process in the shop floor. The selected alternative

process at a higher level will use the most suitable machines at a lower level.

4.3.1 Solution Procedure
Step1:
When evaluating the alternatives with objective constraints (i.e. the constraints of
cost, tolerance, and production rate), we can obtain the alternative weights of the
alternatives by using following rules:
1. if the maximum value is desirable, the alternative weights can be
calculated by normalizing the values.
2. If the minimum value is desirable, the alternative weights can be
calculated by normalizing the reciprocal of values.
As shown in Tables 1OA-1OC, both the cost and the tolerance constraints prefer
the minimum value, so the alternative weights for them are calculated by
normalizing the reciprocal of the values. In Table 100, the production rate prefers
the maximum value, so the alternative weights are calculated by normalizing the

values of all machines directly.

75

Table 10A. Relative priorities with respect to
drilling cost (minim um prefer)

 

Table 10. Relative priorities with respect to objective constraints

Table 1013. Relative prio rities with
respect to boring cost (minimum prefer)

 

 

 

 

 

 

 

Drilling Reciproca Altern ative Boring Reciproca Altem ative
cost (5) l weights cost($) l weights
TC] 2 0.5 0.273 TC] 1 1 0.545
TC2 l 1 0-545 TC2 2 0.5 0.273
MC 3 0.33 0.182 MC 3 0,33 0.182

 

 

 

 

Table 10C. Relative prio rities with respect
to tolerance (minimum prefer)

 

 

 

 

Table 10D. Relative priorities with respect to
production rate (maximum prefer)

 

 

 

 

tolerance Reciprocal Altern ative Production rate Altern ative
weights (unit/hr) weights
TCI 0.001 1000 0.25 TCI l00 0.143
TC2 0.00] 1000 0.25 TC2 100 0.143
MC 0.0005 2000 0.5 MC 500 0.714

 

 

 

 

 

 

The alternative weight vectors from above tables are put into a matrix
(alternative matrix):

0.273 0.545 0.25 0.143
0.545 0.273 0.25 0.143
0.182 0.182 0.5 0.714

M1 :=

Tables 1OA-1OD represent the relative priorities of alternative machines with
respect to the different constraints. The alternative weight vectors for the
alternative machines with respect to all constraints are put into a matrix, called

the alternative matrix, for the machines.

Step 2:
The constraints used to evaluate alternative machines may be:
. Drilling cost, or boring cost: the cost of a process carried out by a machine.

0 Minimum tolerance: the best tolerance for a feature that can be produced by

a machine.

76

0 Production rate: the number of the part finished by a machine in a limit time.

The relative importance between constraints is obtained from the judgment of
experts. The values related to the judgment can be set based on Table 11. Table
11 presents the scale of relative importance, which was defined by Saaty [1980]
for pair-wise comparison. A value within a 1-9 scale is assigned to represent the
relative importance between two constraints. With the values obtained from pair-
wise comparison among the constraints, we can put them 'into a matrix, called
the comparison matrix, to find the eigenvector for the largest eigenvalue of the
matrix. The eigenvector will represent the weights of the constraints.

Table 11. Scale of relative importance

 

 

 

 

 

 

 

 

 

Intensity Definition Intensity Definition
Equal importance 7 Very strong importance
3 Moderate importance 9 Extreme importance
5 Strong importance 2,4,6,8 Intermediate values

 

Tables 12A-12C show three comparison matrixes with respect to alternative
processes A1, A2, and A3 respectively. For each alternative process, the relative
importance among constraints can be different, since the decision maker may
have a different opinion to the same constraints. For example, the constraint of
the production rate for alternative process A1 has a bigger importance than that
of alternative process A2; thus, in the comparison matrix, a bigger value (0.5) is
assigned to the production rate for A1 than that (0.33) for M. Therefore, the
three comparison matrixes represent the different relative importance between

the constraints with respect to the alternative processes.

77

The comparison matrix of the same constraints could be different for each
alternative process. Because the same constraints may have a different influence
on each alternative process, the relative importance and the weights of the
constraints for each alternative process are also different. For each alternative
process, the weights of the constraints are obtained by normalizing the
eigenvector of the largest eigenvalue of the comparison matrix. The weight
vectors for the constraints with respect to all alternative processes can be
grouped into a matrix, called the constraint matrix. In the constraint matrix, if
some constraint is not used to evaluate the machines, such as the constraint of
drilling cost in Table 12C, it indicates a zero value for the constraint weight of the
drilling cost with respect to the alternative process A3.
Table 12. Relative priorities of constraints with respect to alternative processes

Table 12A. Relative priorities of constraints with respect to Al (C.R.=O)

 

 

 

 

 

Drilling Boring Tolerance Production Eigenvector Constraint

cost cost rate (weights) weights
Drilling cost l 1 0.5 2 0.4 0.2222
Boring cost 1 l 0.5 2 0.4 0.2222
Tolerance 2 2 1 4 0.8 0.4444
Production rate 0.5 0.5 0.25 1 0.2 0.1111

 

 

 

 

 

 

 

Table 128. Relative priorities of constraints with respect to A2 (C.R.=0.156>0.1, inconsistent)

 

 

 

 

 

Drilling Boring Tolerance Production Eigenvector Constraint

cost cost rate (weights) weights
Drilling cost 1 l 0.25 3 0.351 0.2042
Boring cost 1 l 0.25 3 0.351 0.2042
Tolerance 4 4 l 2 0.85 0.4916
Production rate 0.33 0.33 0.5 1 0.177 0.1024

 

 

 

 

 

 

 

Table 12C. Relative priorities of constraints with respect to A3

 

 

 

 

Boring cost Tolerance Production Eigenvector Constraint
rate (weights) weights
Boring cost 1 1 2 0.667 0.4001
Tolerance l l 2 0.667 0.4001
Production rate 0.5 0.5 1 0.333 0.1998

 

 

 

 

 

 

78

The constraint weight vectors from above tables are put into a matrix (constraint
matrix):

0.2222 0.2042 0

M2 0.2222 0.2042 0.4001
" 0.4444 0.4916 0.4001

0.1111 0.1024 0.1998
The consistency of the comparison matrixes can be evaluated using the
consistency ratio (OR) represented in section 4.1.2. With the largest eigenvalue
of the matrix obtained through matrix calculation, the consistency analysis of the

matrixes in Tables 12A-12C are shown as the following:

Table 12A: the largest eigenvalue Ame, = 4, the consistency index CI. =
(ll-n)/(n-1)=(4-4)/(4-1)=0, and the random index R.I.=0.9 (matn'x order n=4); thus
the consistency ratio C.R.=C.I./R. I. = 0 <0.1; the comparison matrix is consistent.

Table 128: Am“: 4.422, CI. = 0.141, R.I.=0.9, C.R.= C.I./R.I. = .156 >0.1

Thus the matrix is inconsistent; a new judgment value among all constraints may

be required.

Table 126: lime, = 3, so C.I.=0, R.I.=0.58, C.R. =0. The comparison matrix

is consistent.

Step 3:

The matrixes obtained from steps 1-2 are multiplied as the following:

79

0.273 0.545 0.25 0.143 02222 02042 0

M1 := 0.545 0.273 0.25 0.143 M2 __ 0.2222 0.2042 0.4001
0.182 0.182 0.5 0.714 0.4444 0.4916 0.4001

0.1111 0.1024 0.1998

0.309 0.305 0.347
Ml-M2 = 0.309 0.305 0.238
0.382 0.393 0.416

The alternative matrix M1 is a 3 by 4 matrix that represents the alternative
weights of three alternative machines with respect to four machine constraints.
The constraint matrix M2 is a 4 by 4 matrix that represents the constraint weights
of four constraints with respect to four alternative processes. Table 13 shows the
result obtained from the matrix multiplication. The result represents the global
weights (or overall priority) of each machine with respect to alternative processes.
As shown in Table 13, the machine center is the best option for all alternative
processes A1, A2, and A3, since it has the biggest overall priority: 0.382, 0.393,
and 0.416.

Table 13. Global weight of machine with respect to the process

 

 

 

 

Overall priority Process A1 Process A2 Process A3
TC1 0.309 0.305 0.347
T02 0.309 0.305 0.238
MC 0.382 0.393 0.416

 

 

 

 

 

 

Step 4:

As the suitable machines are selected for all alternative processes, another
selection procedure similar to that in steps 1-3 is made for the process selection
at the next higher level. The data (i.e. actual cost and tolerance) related to the

selected machines will be transformed to the higher level for further comparison.

80

Assuming the casting cost is $4 per unit, Table 14 represents the relative

priorities of alternative processes with respect to the different process constraints.

The alternative weight vectors for alternative processes with respect to all

constraints are grouped into an alternative matrix, M3, for the processes.

Table 14. Attributes of alternatives based

on selected sub-altern ative

Table 14A. R elative priorities with
respect to cost (minimum prefer)

 

 

 

 

 

 

 

 

Cost (5) Tolerance Maximum Cost Reciprocal alternative
(inch) diameter (inch) weights
Al 10 0.0005 8 A1 10 0.1 0.2439
A2 6 0.0005 2 A2 6 0.167 0.4073
A3 0.001 50 A3 0.143 0.3488

 

 

 

 

 

 

 

Table 14C. Relative priorities with respect

to max. diameter (maximum prefer)

 

Table 148. Relative priorities with
respect to tolerance (minimum prefer)

 

 

 

 

 

 

 

 

Max. Alternative Tolerance Reciprocal Alternative
diameter weights weights
Al 8 0.1333 A1 0.0005 200 0.4
A2 2 0.0333 A2 0.0005 200 0.4
A3 50 0.8333 A3 0.001 100 0.2

 

 

 

 

 

 

 

The alternative weight vectors from above tables are put into a matrix

(alternative matrix):

Step 5:

0.2439 0.4 0.1333

M3 :=

0.4073 0.4 0.0333

0.3488 0.2 0.8333

The constraints used here for process selection are cost, tolerance, and

maximum diameter of the hole. As shown in Table 15A, the comparison matrix

represents the result of pair-wise comparison of the relative importance among

constraints. With the constraint weight vector calculated from the comparison

81

matrix, the constraint matrix M4, a 3 by 1 matrix, is obtained. It represents the
relative importance of the constraints with respect to the goal. When multiplying
M4 by the alternative matrix M3 from left side, the overall priority for each

alternative process can be obtained, as shown in Table 15B.

Table 15A. Rrelative priorities with respect to the goal

 

 

 

 

Cost Toleranc M ax. Eigen vector Constraint
e diameter (weights) weights
Cost 1 2 4 0.873 0.5717
Tolerance 0.5 1 2 0.436 0.2855
M ax. 0.25 0.5 1 0.218 0.1428
diameter

 

 

 

 

 

 

Table 158. Selection of alternative process for the goal

 

 

 

 

Alternative Global weight

A 1 Priority=0.2439"0.5717+0.4"0.2855+0.l 333*0.1428=0.273

A2 Priority=0.4073“0.57l7+0.4‘0.2855+0.0333*0.1428=0.352

A3 Priority=0.3488*0.57l7+0.2*0.2855+0.8333‘0.l428= 0.376
(selected)

 

 

Thus, the alternative A3 (Casting + Boring) will be selected with
the biggest priorities.

The constraint weight vector in Table 15A is put into a matrix (constraint
matrix):

0.5717
M4 := 0.2855
0.1428

Table 153 shows the result of matrix multiplication M3*M4.

4.4 Modification of Matrix for AHP Method

When applying the AHP method on the process selection problems in a real

manufacturing environment, some problems may exist:

82

a)

b)

How can a constraint matrix be constructed if some sub-objectives require

only part of the constraints instead of all constraints to evaluate the

alternatives?
How can an alternative matrix be constructed with respect to the

constraints if some constraints exclude some particular alternatives?

To solve the above problems, we use following methods to oonstnict the matrix

or revise the result of the matrix multiplication as needed.

1)

2)

3)

For the constraint matrix, if one of the sub-objectives does not use some
constraint to evaluate the alternatives, the constraint weight of the
constraint with respect to that sub-objective will be assigned zero value in
the constraint matrix.

For the alternative matrix, if one of the alternatives is excluded by some
constraint, the alternative weight of the alternative with respect to the
constraint will be assigned zero value in the alternative matrix.

For the result of matrix multiplication between the alternative matrix and
the constraint matrix, if there exists a zero value in the alternative matrix,
assuming aij =0, and at the same time, there exists an item bjk ¢ 0 in the
constraint matrix; the result of the matrix multiplication needs to be revised:
the item of the result matrix at the ith row and the kth column will be
changed to zero value. This change means that if an alternative is
excluded by a constraint that has a non-zero influence on one objective,

the alternative will not have any influence on the same objective.

83

The application of above rules will be demonstrated with the cases.

4.4.1 Case Study without Matrix Modification
Figure 17 shows a case where a new alternative process A4 and a new
constraint of milling cost are introduced comparing it to the case in section 4.3.
The process A4 uses casting and milling to produce the center hole of the pulley.
The alternative machines are one turning center, T01, and two machining
centers, M01 and MC2. In Figure 17, the constraints at level 4 are not used to
evaluate some alternatives at level 5. Turning center T01 is excluded by the
constraint milling cost, because it cannot carry on the milling process. Thus, a
different alternative matrix must be constructed in such situation. The relative

priorities in Tables 16A—16E are different from those in Tables 10A-10E.

 

Level 1; I Process selection for central hole I

Level 2:
Constraints

 

Level 3:
Alternative
processes

  
  

 

Level 4: - ' ' ' . Production Milling
Sub- rate cost
constraints

Level 5:
Sub-
alternative

t...—

 

1. A1: Casting+Drilling+Boring; A2: Cutoff+Drilling+Boring; A3:Casting+Boring;
A4:Casting+Milling.
2. Turning center: TCl; Machine Center: MCI and MC2

Figure 17. Case Study 2

84

Table 16. Relative priorities with respect to drilling cost, boring cost, tolerance,
production rate, and milling cost without matrix modification

Table 16A. Relative priorities with respect to

drilling cost (minimum prefer)

 

 

 

 

Drilling Recip roca Alternative
cost ($) 1 weights
TCI 2 0.5 0.273
MC] 1 1 0.545
MC2 3 0.33 0.182

 

 

 

 

Table 16C. Relative priorities with respect
to tolerance (minimum prefer)

 

 

 

 

 

 

 

 

Table 16B. Relative priorities with respect
to boring cost (minimum prefer)

 

 

 

 

 

 

 

 

Boring Reciprocal Alternative
cost ($) weights
TCl 1 1 0.545
MCI 2 0.5 0.273
MC2 3 0.33 0.182
Table 16D. Relative priorities with respect to

production rate (maximum prefer)

 

 

 

 

 

 

 

 

 

 

Tolerance Reciprocal Alternative Production rate Alternative
weights (“mt/hf) weights
TCI 0.001 1000 0.25 TCI 100 0143
MCI 0.001 1000 0.25 MCI 100 0143
MC2 0.0005 2000 0.5 MC2 500 07“
Milling cost ($) Reciprocal Alternative weights

Table 16E. Relative priorities MCI 5 0.2 0.667

with respect to milling cost MC2 10 0.1 0.333

 

 

 

 

For A1, there are four constraint required, assume equal weight for all constraints:
0.273 0.545 0.25 0.143

 

M12: 0.545 0.273 0.25 0.143 0.25 0.303
0.182 0.182 0.5 0.714 M21: 025 Ml-M2= 0.303
0.394

0.25

For A4, there are three constraints only. Assuming equal weight, then

0.33
0.333 0.167 0.667 0.385
= M2 := 0.33 M1«M2 =
0.667 0.833 0.333 033 0.605

Based on the result of the matrix multiplication, machining center 2 will be
selected for A1 and A4 since it has the biggest global weights 0.394 and 0.605

respectively.

85

4.4.2 Case Study with Matrix Modification

In section 4.4.1, the matrixes are not modified with the rules presented above.

The following steps will show how the matrixes are revised based on the real

case.

Step 1:

Tables 17A-17E show the relative priorities of alternative machines with respect

to different constraints.

Table 17. Relative priorities with respect to drilling cost, boring cost, tolerance,

Table 17A. Relative priorities with respect to

drilling cost (minimum prefer)

production rate, and milling cost after matrix modification

Table 17B. Relative priorities with respect to

boring cost (minimum prefer)

 

 

 

 

 

 

 

 

Drilling Reciprocal Alternative Boring Reciprocal Alternative
cost ($) weights cost ($) weights
TCI 2 0.5 0.273 TC] 1 1 0.545
TC2 1 I 0.545 TC2 2 0.5 0.273
MC 3 0.33 0.182 MC 3 0.33 0.182

 

 

 

 

Table 17C. Relative priorities with respect

 

 

 

 

Table 17D. Relative priorities with respect to

to tolerance (minimum prefer)

 

 

 

 

 

 

production rate (maximum prefer)

 

 

 

 

 

 

 

 

 

 

 

 

 

Tolerance Reciprocal Alternative Production Alternative
weights rate (unit/hr) weights
TCl 0.001 1000 0.25 TC] 100 0.143
TC2 0.001 1000 0.25 TC2 100 0.143
MC 0.0005 2000 0.5 MC 50 0.714
. . . . Milling cost ($) Reciprocal Alternative weights
Table 17E. Relatiye priorities TCI N / A 0 0
With respect to milling cost
MC] 5 0.2 0.667
MC2 10 0.1 0.333

 

 

 

 

Based on the alternative weight vectors shown above, the alternative matrix for

the machines is obtained as following:

M1 :=

0.273 0.545 0.25 0.143 0
0.545 0.273 0.25 0.143 0.667
0.182 0.182 0.5 0.714 0.333

86

Step 2:

Tables 18A-1BD show the constraint weights of the constraints with respect to

different alternative processes as following:

Table 18. Relative priorities of constraints with respect to alternative processes

Table 18A. Relative priorities of criteria with respect to Al

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Drilling Borin Tolerance Producti Milling Eigenvector Constraint
cost g cost on rate cost (weights) weights
Drilling cost 1 1 0.5 2 N/A 0.4 0.2222
Boring cost 1 1 0.5 2 N/A 0.4 0.2222
Tolerance 2 2 l 4 N/A 0.8 0.4444
Production 0.5 0.5 0.25 1 N/A 0.2 0.1111
rate
Milling cost N/A N/A N/A N/A N/A N/A 0
Table 183. Relative priorities of criteria with respect to A2
Drilling Boring Tolerance Production Milling Eigenvecto Constraint
cost cost rate cost r (weights) weights
Drilling cost 1 1 0.25 3 N/A 0.324 0.1900
Boring cost 1 l 0.25 3 N/A 0.324 0.1900
Tolerance 4 4 1 2 N/A 0.868 0.5091
Production rate 0.33 0.33 0.5 1 N/A 0.189 0.1109
Milling cost N/A N/A N/A N/A N/A N/A 0
Table 18C. Relative priorities of criteria with respect to A3
Drilling Boring Tolerance Production Milling Eigenvector Constraint
cost cost rate cost (weights) weights
Drilling cost N/A N/A N/A N/A N/A N/A 0
Boring cost N/A 1 l 2 N/A 0.667 0.4001
Tolerance N/A 1 l 2 N/A 0.667 0.4001
Production N/A 0.5 0.5 1 N/A 0.333 0.1998
rate
Milling cost N/A N/A N/A N/A N/A N/A 0
Table 18D. Relative priorities of criteria with respect to A4
Drilling Boring Tolerance Production Milling Eigenvector Constraint
cost cost rate cost (weights) weights
Drilling cost N/A N/A N/A N/A N/A N/A 0
Boring cost N/A N/A N/A N/A N/A N/A 0
Tolerance N/A N/A 1 l 2 0.667 0.4001
Production N/A N/A 0.5 0.5 1 0.333 0.1998
rate
Milling cost N/A N/A 1 l 2 0.667 0.4001

 

 

 

 

 

 

 

 

87

Based on the constraint weight vectors shown above, the constraint matrix for
the machines is obtained as the following:

i 0.2222 0.1900 0 0 \
0.2222 0.1900 0.4001 0

M2 := 0.4444 0.5091 0.4001 0.4001
0.1111 0.1109 0.1998 0.1998

\ 0 0 0 0.4001)

 

 

Thus, the result of the matrix multiplication between alternative matrix and
constraint matrix:

0.309 0.299 0.347 0.129
Ml-M2 = 0.309 0.299 0.238 0.395
0.382 0.403 0.416 0.476

Step 3:
The result of matrix multiplication needs to be revised since there exists zero
value in alternative matrix and non-zero item in constraint matrix

M1: a15= 0; and M2: D54 = 0.4001 15 0
According to the rule, the value of the item in the result of matrix multiplication C14
should be changed to zero; the turning center TC1 cannot be chosen for
alternative process A4, that is

0.309 0.299 0.347 0
MI-M2 := 0.309 0.299 0.238 0.395
0.382 0.403 0.416 0.476

Thus, the machining center 2 will be selected for the alternative processes A1,
A2, A3, and A4 since it has the biggest global weights: 0.382, 0.403, 0.416, and

0.476.

88

Step 4:

Table 19. Attributes ofalternatives based
on selected sub-alternative

Table 19A. Relative priorities with
respect to cost (minimum prefer)

 

 

 

 

 

 

 

 

 

 

Cost ($) Tolerance Maximum Cost Reciprocal Alternative
(inch) diameter (inch) weights
A1 10 0.0005 8 A1 10 0.1 0.208
A2 6 0.0005 2 A2 6 0.167 0.347
A3 0.001 50 A3 7 0.143 0.297
A4 14 0.0005 100 A4 14 0.071 0.148

 

 

 

 

 

 

 

 

Table 19C. Relative priorities with respect
to max. diameter (maximum prefer)

Table 198. Relative priorities with
respect to tolerance (minimum prefer)

 

 

 

 

 

 

 

 

 

 

 

 

Max. Alternative Tolerance Reciprocal Alternative weights
diameter weights

A1 8 0.05 A1 0.0005 200 0.286

A2 2 0.0125 A2 0.0005 200 0.286

A3 50 0.3125 A3 (”’0‘ 100 M42

A4 100 0.625 A4 0.0005 200 0.286

 

 

 

 

 

The alternative matrix for Tables 19A—1QC is

0.208 0.05 0.286
0.347 0.0125 0.286
0.297 0.3125 0.142
0.148 0.625 0.286

M3

Tables 19A-19C show the alternative weights of the alternative processes with
respect to the different constraints. The constraint matrix for the goal is same as

that for case study 1 in section 4.3. Then

0.5717 0174

'= 0.243
M4. 0.2855 M3M4 =

0.1428 0279

0.304

Thus, the alternative process A4 should be selected for producing the hole with

the biggest global weight 0.304.

89

4.5 Selection of Part Family
Present industry is facing a frequently changing market. The order from
customers could be various product types and quantities. One problem is that the
customer’s order may be too small to match the requirement of a minimum order
from the supplier. In this case, several parts with similar geometrical shapes or
sizes could be grouped into a part family to produce a profitable order for the

supplier. Thus, an effective method to identify the suitable part family is required.

Zeng et al. [2003] presented four sets of constraints (production, process,
machine, and feature) in a real manufacturing environment. The above case
studies show that the process, machine, and production constraints could be
applied on the MMCDM model with the AHP method. The feature constraints

could also be used to identify a suitable part family with the AHP method.

During the production of the pulleys, a problem emerges because the small order
of a particular pulley from the customer cannot meet the minimum order
requirement from the casting supplier. It is quite important for the machining
company to group different parts into a part family. Within the part family, all parts
can be finished with the same casting even though the sizes or geometrical
shape may vary a bit for each particular part. In the company, the part family for
pulleys may be defined based on their outside diameters since the maximum
chuck diameter of the machine limits the part family it can machine. In addition,

the inside diameter, weight, and tolerance of a pulley can help to distinguish a

90

part family for a particular pulley; and a suitable machine will be identified to

produce the part family.

 

Select part family for casting

 

 

(Weight II O.D. I I ID. I IToIerance

 

 

 

 

Group 1 I Group 2

Figure 18. The selection of part family for casting process

Table 20. Two part families of pulleys

 

 

 

Weight (lb.) O.D. (inch) l.D. (inch) Tolerance (inch)
Group 1 15 15 3 0.001
Group 2 13 14 2 0.005

 

 

 

 

 

 

Figure 18 shows a hierarchy for the selection of a part family. At the constraint
level, four constraints (including the outside diameter (O.D.), inside diameter
(I.D.), tolerance, and weight) are used to evaluate alternative part families. Group
1 and group 2 are two part families whose characteristics are shown in Table 20.
For each part family, only a particular type of casting is ordered from the casting
supplier. Once a new pulley is introduced, a suitable part family will be selected

for this pulley so that a casting for old pulley could also be used for this new part.

91

 

For the identification of the part family, we capture the differences between new

pulley and the part family with the constraints of weight, outside diameter, inside

diameter, and tolerance. Table 21 represents the difference of the new pulley

referred to the two part families. With these, AHP method can be used to select a

suitable part family for the given part with the largest overall priority. The solution

is similar to that in case study 1, as shown in Table 22A — 220.

Table 21. The difference of given part refer to the part families

 

 

 

 

 

 

 

 

 

Weight (1b.) O.D. (inch) 1.0. (inch) Tolerance (inch)
Part 12 14 4 0.0005
Difference to group 1 3 2 1 0.0005
Difference to group 2 1 1 2 0.0045

 

Table 22. Relative priorities of part family with respect to weight, O.D., I.D., and

tolerance

Table 22A. Relative priorities ofp art
family with respect to weight

Table 228. Relative priorities ofp art

family with respect to O .D.

 

 

 

Weight Reciprocal Alternative O.D. Reciprocal Alternative
weights weights
C51 3 0.33 0.25 01 2 0.5 0.333
02 l l 0.75 (32 1 1 0.667

 

 

 

 

 

 

 

Table 22C. Relativepriorities ofp art

family with respect to 1.0.

Table 22D. Relative priorities ofpart
family w ith respect to tolerance

 

 

 

i.D. Reciprocal Alternative Tolerance Reciprocal Alternative
weights weights
GI 1 1 0.667 01 0.0005 2000 0.9
02 2 0.5 0.333 C12 0.0045 10000/45 0.1

 

 

 

 

 

 

 

The alternative matrix for above tables is

l (0.25 0.333 0.667 0.9)

0.75 0.667 0.333 0.1
Assuming a distributed weight vector (0.3, 0.3, 0.3, 0.1) for the constraints

(weight, O.D., I.D., and tolerance), the result matrix is obtained as the following:

92

 

 

0.3

0.25 0.333 0.667 0.9 M2 ;= 0'3 0.465
M1 := 03, M1-M2 =
0.75 0.667 0.333 0.1 0.535
0.1

Therefore, the part is preferred to be included into group 2 with a bigger weight

0.535.

4.6 Conclusion
This chapter has demonstrated that AHP is an effective tool for evaluating and
selecting alternative processes and the related manufacturing resources (such as
machines). Combined with process grammar, AHP can be applied on a decision
making model — MMCDM, which is set up for the process selection problem in
the shop floor environment. It achieves several attractive features (1) the AHP
method can be applied on evaluating the alternatives (i.e. processes, machines,
and part families) at any level of abstraction. This will help decision makers to
understand the overall selection process and select the best alternatives at each
level. (2) With process grammar, a complex selection problem could be simplified
as hierarchy selection problems that can be solved with AHP method. (3) The
hierarchy structure of both processes and constraints represented in the process
model makes it easy to obtain a multilevel multiple criteria decision making model
for a complex process selection problem. (4) The proposed decision making
model could be applied on selecting manufacturing systems, processes,
machines, providers, part families, etc. when quantitative and/or qualitative

constraints are captured and put into the model.

93

 

Chapter 5

CONCLUSION

A methodology to model, design, and reconfigure a process has been described,
which will make a production system more flexible and adaptable to a new
production. This will reduce design and reconfiguring costs by:

. Representing the processes of a system with a grammar-based approach,
and making designer or manager understand the overall processes of the
system at any detail level of abstraction.

. Representing the system constraints in hierarchy stnicture and managing
them in process reconfiguration when change arises during the
production.

. Describing a method for process selection with the hierarchy structures of
both processes and constraints represented above.

0 Decomposing a complex decision-making problem as a smaller number of
simple selections at any abstract level by combining process grammar and

analytic hierarchy process method.

5.1 Contributions
This dissertation presents a grammar-based methodology of process design for
Reconfigurable Manufacturing Systems. Specifically, this research makes the

following contributions:

94

Conducted a case study for the application of process grammar on
modeling, designing, and reconfiguring processes in an industrial setting.
Described a grammar-based approach in capturing and managing the
processes in a distributed manufacturing environment.

Represented the important constraints used to evaluate the alternative
processes generated from the entities (task, data, organization, and
resource) related to collaborative manufacturing.

Described the hierarchy structures of processes and constraints that will
help to effectively reconfigure processes when change arises during the
production.

Defined the constraint categories and identified the relationships among
them when the constraints are used to evaluate the alternative processes
at a different abstract level.

Represented a multilevel multiple criteria decision-making (MMCDM)
model using the hierarchy structures of processes and constraints to make
the selected process optimum at any abstract level.

Applied the analytical hierarchy process (AHP) method on the selection of
process, machine, and part family and revised the matrix of AHP method
for matching the real industrial selection.

Constructed a foundation for an open-architecture system that manages
the overall processes effectively in a distributed manufacturing

environment.

95

5.2 Future Work
To extend this research, it is recommended that the following topics be
investigated in the future:
0 Mathematical descriptions for quantifying the constraints that are used to
evaluate the alternative processes.
a More well-defined relationships among constraints should be captured for
constraint transformation between different levels.
. Rules for automatically operating upon the constraints for evaluating the
alternative processes at different abstract level.
. Development of software and implementation of the developed

methodology in practice.

96

APPENDICES

97

APPENDIX A

GLOSSARY

Analytic Hierarchy Process A multi-criteria decision making method that
decomposes a complex problem into a hierarchical order. See Chapter 4.
Alternative Matrix A matrix in which each column vector represents the
alternative weights of the alternatives with respect to the elements (constraints)
of next higher level. See Chapter 4.

Alternative Weight The measurement of relative priorities of alternatives with
respect to the constraint that is used to evaluate the alternatives.

Atomic Task A task that cannot be decomposed into a set of subtasks. Atomic
tasks are represented using two-concentric oval nodes in the process flow graph.
Comparison Matrix A matrix in which each component value represents the
relative importance between the column constraint and the row constraint. See
Chapter 4.

Consistence Index (0.1.) The deviation from consistency of a matrix, which is
calculated by (Amax— n)/(n-1), 2mg, is the largest eigenvalue of the matrix, and n is
the order of the matrix. See Chapter 4.

Consistence Ratio (C.R.) The ratio of consistence index (C.l.) to average
random index (R.l.) for the same order matrix, which is calculated by OR. =
C.I./R.I.. See Chapter 4.

Constraint A condition that must be satisfied for an alternative (process,

resource, or organization) to be considered acceptable. See Chapter 2.

98

Constraint Hierarchy A hierarchy in which each constraints has an “is a type of“
relationship with its ancestors. See Chapter 3.

Constraint Matrix A matrix in which each column vector represents the
constraint weights of the constraints with respect to the elements (objectives) of
next higher level. See Chapter 4.

Constraint Weight The measurement of relative priorities of constraints with
respect to an objective that use all the constraints to evaluate alternatives.
Dedicated Manufacturing System A system designed for producing a specific
part at a high volume, which uses typically transfer line technology with fixed
tooling and rigid automation. See Chapter 1.

Flexible Manufacturing System A highly automated GT machine cell,
consisting of a group of processing stations (usually CNC machine tools),
interconnected by an automated material handling and storage system and
controlled by an integrated computer system. See Chapter 1.

Global Weight The measurement of relative priorities of alternatives with respect
to an objective that use constraints to evaluate the alternatives.

Graph Grammar A mathematical formalism for manipulating symbols
representing graph vertices and edges. See Chapter 1.

Group Technology An approach to manufacturing in which similar parts are
identified and grouped together in order to take advantage of their similarities in
design and production. See Chapter 1.

Hierarchy A series in which each element is graded or ranked. See Chapter 3.

99

Logical Task A task that can be decomposed into a set of subtasks (logical
tasks, atomic tasks, or selectors). Logical tasks are represented using oval nodes
in the process flow graph. See Chapter 1.

Methodology A body of practices, procedures, and rules used by those who
work in a discipline or engage in an inquiry.

Part Family A collection of parts that are similar either because of geometric
shape and size or because similar processing steps are required in their
manufacture. See Chapter 1.

Process Design The systematic exploration and determination of the methods
by which a process is to be designed, economically and competitively. See
Chapter 1.

Process Grammar A proposed formalism for representing, manipulating, and
executing sets of acceptable design processes. It provides the mechanism for
transforming a high-level process flow graph into progressively more detailed
process flow graphs. See Chapter 1.

Process Hierarchy A hierarchy in which each process has an “is a type of”
relationship with its ancestors. See Chapter 3.

Process Planning The systematic determination of the methods by which a
product is to be manufactured, economically and competitively. See Chapter 4.
Random Index (R.l.) The average consistency index of a randomly generated
matrix from the scale 1 to 9. See Chapter 1.

Reconfigurable Manufacturing System A reconfigurable manufacturing system

(RMS) is designed for rapid change in the structure of both hardware and

100

 

software components in order to rapidly adjust to new production method,
production capacity, and functionality in response to changes in market or
regulatory requirements. See Chapter 1.

Selector A selector selects a specification or a parameter for a task. Selectors
are represented using diamond nodes in the process flow graph. See Chapter 1.
Shape Grammar A method to derive designs in the language that it specifies by
successive application of shape transformation rules to some evolving shape,
starting with an initial shape. See Chapter 1.

Subtask A task component represented by one of the task nodes in a design
process grammar in which the component can be a logical task, an atomic task,
or a selector. See Chapter 1.

Task A step in a design process or an activity in a production. See Chapter 1.

101

 

BIBLIOGRAPHY

102

 

Bibliography

Abdi, MR. and Labib, AW. (2003). A Design Strategy for Reconfigurable
Manufacturing Systems (RMSs) Using Analytical Hierarchical Process (AHP): A
Case Study. lntematjonal Journal of Prodflon Research, 41 (10), 2273-2299.

Agarwal, M., Cagan, J. and Constantine, KG. (1999). Influencing generative
design through continuous evaluation: Associating costs with the coffeemaker
shape grammar. Agificial Intelligence for Engineering Design. Analvsis and
Manufacturing, 13, 253-275.

Agarwal, M. and Cagan, J. (2000). On the use of shape grammars as expert
systems for geometry-based engineering design. Artificial Inteflgence for
Engineering Design. Analvsis and Mantfiacturing, 14, 431-439.

Akarte, M.M, Surendra, N.V, Ravi, B. and Rangaraj, N. (2001). Web Based
Casting Supplier Evaluation Using Analytical Hierarchy Process. Journal of the
Operation Research Society, 52, 511-522.

Baldwin, R. and Chung, M.J. (1995a). Design Methodology Management: A
Formal Approach. IEEE Computer, 54-63.

Bralla, J.G. (1998). Design For Mangfacturabilitv Handbook. 2nd Edition,
McGraw-Hill, 4.45-4.58.

Candadai, A., Herrmann, J.W. and Minis, l. (1996). Applications of Group
Technology in Distributed Manufacturing. Journal of Intelligent Mangfacturing, 7,
271-291.

Chan, F.T.S. (2003). Interactive Selection Model For Supplier Selection Process:
An Analytical Hierarchy Process Approach. International Jogmal of Production
Research, 41 (15), 3549-3579.

Chung, M.J., Kwon, P. and Pentland, B. (2002). Making Process Visible: A
Grammatical Approach to Managing Design Processes. ASMI; Transactions.
flimal of Mechanical Design, 124, 364-374.

Chung, M.J., Kwon, P., Pentland, B. and Kim, S. (2002). A Process Management
System for Collaborative Manufacturing. IMECE, A SmeOSILfl on
Reconfigurable Mangfacturing Systems, New Orleans, LA.

Feitzinger, E. and Lee, H. (1997). Mass Customization at Hewlett-Packard: The
Power of Postponement. Harvard @siness Review, 75, 116-121.

103

 

Fitzhom, P. (1986). A linguistic formalism for engineering solid modeling. Graph-
grammars and their application to computer science, 3rd lntemational Workshop.
Springer-Verlag, New York.

Fu, 2., Pennington, A. and Saia, A. (1993). A graph grammar approach to feature
representation and transformation. lntematjonal Journal of Comter Integrateg
Manufacturing, 6(102), 137-151.

 

Groover, MP. (2002). Fundamentals of Mogem Mantnictunng, 2nd Edition, John
Wiley & Sons, INC.

Hoover, S. and Rindeiie, J. (1989). A synthesis strategy for mechanical devices.
Research in E_ngineering_l;)esigg, 1, 87-103.

Jones, RE. and Wilson, RH. (1996). A Survey of Constraints in Automated
Assembly Planning. Proc. 1996 IEEE Intl. Conf. On Robotics and Automation,
1525-1532.

Kalapakjian, S. and Schmid, SR. (2003). Manufacturing Processes for
Engineering Materials, Fourth Edition, Prentice Hall.

Koren, Y., Heisel, U., Jovane, F., Moriwaki, T., Pritschow, G., Ulsoy, G. and Van
Brussel, H. (1999). Reconfigurable Manufacturing Systems. Annafi of the CIRP.
48(2), 527-540.

Kwon, P., Chung, M.J. and Pentland, B. (2002). A Grammar Based Framework
for Integrating Design and Manufacturing. ASMI_E Transagions. Jogmal of
Manufacturing Science and Engineering, 124, 899-907.

Luong, L.H.S. (1998). A Decision Support System For The Selection Of
Computer-Integrated Manufacturing Technologies. Robotics anLComguter—
lntegratgd; Manufacturing, 14, 45-53.

Malakooti, B. and Yang, Z. (2002). Multiple Criteria Approach and Generation of
Efficient Alternatives for Machine-part Family Formation in Group Technology. E
TransacLions. 34, 837-846.

Millet, I. and Saaty, T. L. (1998). On the Relativity of Relative Measures —
Accommodating both Rank Preservation and Rank Reversals in the AHP.

Murata T. (1989). Petri nets: properties, analysis, and applications. Proc IEEE,
77(4), 541 -580.

Nau, D.S., Ball, M.O., Gupta, S.K., Minis, LE. and Zhang, G. (1994, December).

Design for Manufacture in Multi-enterprise Partnerships: Current Status and
Future Directions. ASMfiE¥Concurrent Product Design, 74, 117-125.

104

 

Park, H. and Cutkosky, MR. (1999). Framework for Modeling Dependencies in
Collaborative Engineering Processes. Research in Engineering Design, 11, 84-
102.

Pinilla, J., Finger, S. and Prinz, F. (1989, June 11-14). Shape feature description
using an augmented topology graph grammar. Preprints: NSF Engineering
Design Research Conference, Amherst, MA, 285-300.

Rangone, A. (1996). An Analytical Hierarchy Process Framework For comparing
The Overall Performance Of Manufacturing Departments. lntemational Journal of
Operations and Production Management, 16(8), 104-119.

Ratchev, S., Shiau, J. and Valtchanov, G. (2000). Distributed Product and Facility
Prototyping in Extended Manufacturing Enterprises. lntemational Journal
Proggction Reseggcfi, 38, 4495-4506.

Ross UT (1977). Structured Analysis (SA): A language for communicating ideas.
IEEE Trans Software fig, 3(1), 16-34.

Saaty, T.L., (1980), The AngMic HiergrchiLProcess. New York: McGraw-Hill.

Sadiq, W. and Orlowska, ME. (2000, October). On Business Process Model
Transformations. Proceeding of 19th lntematjonal Conference on a, Salt Lake

City.

Schmidt, LC. and Cagan, J. (1997). GGREADA: A Graph Grammar-Based
Machine Design Algorithm. Research in E_ngineeringgeSLQQ. 9, 195-213.

Schmoldt, D. L., Kangas, J. and Mendoza, GA. (2001). Basic Principles of
Decision Making in Natural Resources and the Environment. The Analyt_i_C_
Hierarchy Process in Nautical Resource and Environmentgl Decision Makigq, 1-
13. Netherlands: Kluwer Academic Publishers.

Society of Manufacturing Engineers (1974). Tool and Manufacturing Engineers
Handbook, 4th ed. Dearborn, Michigan.

Speredelozzi, V.M. and Hu, J. (2002). Selecting Manufacturing System
Configurations Based On Performance Using AHP. Transagtjons of NAMRI/SMI;
30, 637-644.

Tiwari, MK. and Banerjee, R. (2001). A Decision Support System For The
Selection Of A Casting Process Using Analytic Hierarchy Process. Production
Planning and Control, 12(7), 689-694.

Wabalickis, RN. (1987). Justification of FMS with the Analytic Hierarchy
Process. Mmal of Mangfacturing Systems, 7(3), 175-182.

105

Wiest JD. and Levy, F.K. (1977). A Management Gifie to PERT/CPM, Prentice-
Hall, Englewood Cliffs, NJ.

Yurdakul, M. and Cogun, C. (2003). Development of A Multi-Attribute Selection
Procedure for Non-Traditional Machining Processes. Journal Engineering
Manufacture, 217.

Zeng, Y., Kwon, P., Pentland, B.T. and Chahine, A. (2003). A Grammar-Based
Approach To Capturing And Managing Processes: An Industrial Case Study.
Transactions of NAMRI/SMI; 31, 555-562.

Zhang, Y., Feng, S., Wang, X., Tian, W. and Wu, R. (1999). Object Oriented

Manufacturing Resource Modeling for Adaptive Process Planning. Intemaflonal
Journal Production Research, 37, 4179-4195.

106