Copyright Warning & Restrictions

The copyright law of the United States (Title 17, United

States Code) governs the making of photocopies or other
reproductions of copyrighted material.

Under certain conditions specified in the law, libraries and

archives are authorized to furnish a photocopy or other
reproduction. One of these specified conditions is that the
photocopy or reproduction is not to be “used for any
purpose other than private study, scholarship, or research.”
If a, user makes a request for, or later uses, a photocopy or
reproduction for purposes in excess of “fair use” that user
may be liable for copyright infringement,

This institution reserves the right to refuse to accept a

copying order if, in its judgment, fulfillment of the order
would involve violation of copyright law.

Please Note: The author retains the copyright while the

New Jersey Institute of Technology reserves the right to
distribute this thesis or dissertation

Printing note: If you do not wish to print this page, then select
“Pages from: first page # to: last page #” on the print dialog screen
The Van Houten library has removed some of the
personal information and all signatures from the
approval page and biographical sketches of theses
and dissertations in order to protect the identity of
NJIT graduates and faculty.
 ABSTRACT

Title: COMPUTER-AIDED PROCESS PLANNING

(CAPP) - SORICH
Author: SOHAIL PIRZADA
Degree: Master of Science
Date: December, 1991
Advisor: Steve Kotefski
Department: Manufacturing Engineering

The last two decades have witnessed the increasing role of

the computer in the process planning function. This has been
further enhanced by the advent of the knowledge based expert
system, and its impact on Computer Aided Process Planning
(CAPP). CAPP has emerged as a strategic link between design
and manufacture. This thesis discusses the various methods
used in the process planning function. Group Technology (GT)
plays a pivotal role in establishing CAPP. The utilization
of Artificial Intelligence techniques in CAPP is listed.
This study also presents an Interactive Software adaptive to
product/part configuration variables.
 COMPUTER-AIDED PROCESS PLANNING
(CAPP)
SORICH

by

SOHAIL PIRZADA

Thesis Submitted To The Faculty Of Graduate Division Of The

New Jersey Institute of Technology
In Partial Fulfillment Of The Requirement For The Degree Of
Master Of Science
Manufacturing Engineering

December, 1991
 APPROVAL SHEET

Title of The Thesis: COMPUTER-AIDED PROCESS PLANNING

(CAPP) - SORICH

Name of The Candidate: SOHAIL PIRZADA

Thesis and Abstract Approved Date Approved

Ze Kotdes' ki /f/
Professor
Department of Manufacturing
Engineering Technology

Dr. Raj Sodhi

Professor
Department of Manufacturing
Engineering

Nouri Levy
Professor
Department of Mechanical
Engineering
 VITA SHEET

Name: SOHAIL PIRZADA

Degree & Date: Master of Science, December, 1991

Education:

Institute Degree Major Date

NED University of B.E (B.S) Mechanical 1989

Engineering &
Technology, Karachi,
Sindh - Pakistan

New Jersey Institute M. S Manufacturing 1991

of Technology,
New Jersey
 To

my

MOTHER

Nasim Akhtar Pirzada

 ACKNOWLEDGEMENT

The author wishes to express his sincere gratitude to

Professor Steve Kotefski of Manufacturing Engineering

Technology Department at New Jersey Institute Of Technology

for his assistance, guidance and encouragement as Graduate

Advisor.
 CONTENTS

CHAPTER 1
Computer Aided Process Planning (CAPP) 2

CHAPTER 2

Process Planning Methods 5

2.1. Engineering Design 13
2.2. Dimensioning 16
2.3. Tolerances 18
2.4. Geometric Tolerances 24

2.5. Maximum Material Condition (MMC) 28

2.6. Regardless of Feature Size (RFS) 29
2.7. Least Material Condition (LMC) 29
2.8. Datum 30
2.9. Basic 30
2.10. Machine Tooling 31
2.11. Jigs 32
2.12. Fixtures 34
2.13. Dies 35

CHAPTER 3

Economical Production Criteria 38

3.1. Automation 39

3.1.1. Types of automation 40

CHAPTER 4
Manual Process Planning 46

CHAPTER 5
Computer-Aided Process Planning 52

5.1. Benefits of CAPP 56

5.2. Retrieval CAPP systems 57

5.2.1. The Preparatory Stage 60
5.2.2. A Production Stage 62
5.2.3. Database Structure 63
5.2.4. Search Procedure 64
5.2.5. Plan Editing and Parameter
Selection 65
5.3. Generative CAPP systems 67
5.3.1. Forward and Backward Planning 72
5.3.2. The Input Format 75
a. Code 77
b. Description Language 77
c. CAD Models 81

5.3.3. Decision Logic 82

5.3.3.1. Decision Trees 86
5.3.3.2 Decision Tables 95
5.4. BORICH CAPP System Designed for Thesis 96

CHAPTER 6
Group Technology in CAPP 99
6.1. Part-Family Formation 103
6.2. Visual Methods 108

6.3. Production Flow Analysis (PFA) 109

6.3.1. Rank Order Cluster Algorithm 111

6.4. Coding and Classification 114

6.4.1. Hierarchical 117

6.4.2. Chain (matrix) 118

6.4.3. Hybrid 118

6.5. Group-Technology Coding System 121

6.5.1. The Opitz System 125

6.5.2. The Vuoso-Praha System 130

6.5.3. The KK-3 System 132

6.5.4. The CODE System 135

6.5.5. DCLASS Systems 138

6.5.6. The MICLASS System 140
6.6. Benefits Of Group Technology 145

6.7. Machine Selection 146

CHAPTER 7

CAPP Using Artificial Intelligence 151

7.1. AI and Process Planning 152

CHAPTER 8

Process Planning Systems - A Survey 157

8.1. APPAS 157

8.2. AUTOPLAN 157

8.3. CAM-I CAPP 158

8.4. ACAPS 158
8.5. ICAPP 158

8.6. AUTOCAPP 158

8.7. MIPLAN 159

8.8. GARI 159

8.9. GENPLAN 159

8.10. SHAPES 160

8.11. TIPPS 160

8.12. TOJICAPP 160

8.13. MICRO-CAPP 161

8.14. GEOMETRIC PROGRAMMING 161

8.15. CAPP and Dynamic Process Selection

Techniques 161

CHAPTER 9

Description of SORICH CAPP System 164

9.1. Design of System 164

9.2. Use Of The SORICH System 168

CHAPTER 10
CAPP - The Future 171

CHAPTER 11

Conclusion 175

BIBLIOGRAPHY 177


LIST OF FIGURES

FIGURE 1. Standard Material Shapes 10

FIGURE 2. Process Plan 12

FIGURE 3. Idea Sketch Prepared By Leonardo

da Vinci (1452-1519) 15

FIGURE 4. Tolerancing: Bilateral And Unilateral.

Dashed Lines Show The Tolerance
Limits 19
FIGURE 5. Surface Control Symbols 21
FIGURE 6. Lay Symbols 23
FIGURE 7. Illustration Of Some Addition Part
Conditions 24
FIGURE 8. Using Notes & Using Symbols 26
FIGURE 9. Plate Drilling Jig 33
FIGURE 10. Typical Process-Planning System 53
FIGURE 11. Process-Planning Bridges Design And
Manufacturing 55
FIGURE 12. General Procedure For Using One Of The
Retrieval Computer-Aided Process
Planning Systems 59
FIGURE 13. Production Stage 62
FIGURE 14. Process-Parameter File 66
FIGURE 15. Component To Be Planned: Final Design 73
FIGURE 16. AUTAP Data Input
(Eversheim et al., 1980) 78
FIGURE 17. CIMS/DEC Component Modeling 79
FIGURE 18. GIRI Component Modeling 80
FIGURE 19. Decision Table And Decision-Tree
Representation 84
FIGURE 20. Structured Flow Chart Corresponds To
A Decision-Tree 87
FIGURE 21. Implementation Of A Decision-Tree In
A Program 91
FIGURE 22. Input To DCTREE 93
FIGURE 23. SORICH Decision-Tree 94
FIGURE 24. Design Family 100
FIGURE 25. Composite Components 102
FIGURE 26. Two Parts Of Identified Shape And
Size But Different Manufacturing
Requirements 104
FIGURE 27. Thirteen Parts With Similar
Manufacturing Process Requirements
But Different Design Attributes 104
FIGURE 28. Process-Type Layout 106
FIGURE 29. Group Technology Layout 106
FIGURE 30. PFA Matrix 110
FIGURE 31. Rank-Order Cluster Algorithm 113
FIGURE 32. Hierarchical Structure 116
FIGURE 33. Hybrid Structure 119
FIGURE 34. Opitz Coding And Classification
System 120


FIGURE 35. Rotational Component (H. Opitz, A

Classification System To Describe
Workpieces) 121
FIGURE 36. Example Of A KK-3 Coding System 134
FIGURE 37. A Portion Of The CODE System Of MDSI 137
FIGURE 38. DCLASS Structure 138
FIGURE 39. Workpart 142
FIGURE 39 Computerized MICLASS System

Determination Of Code Number For

Workpart 143
FIGURE 40. Comparison (Typical) Of A Turned-Part
Dimension As A Function Of Machine
Capacity 147
FIGURE 41. Comparison Of Maximum Speeds And Feeds

(Machine Capacity) With Maximum Used 148

 LIST OF TABLES

TABLE 1. OPERATIONS AND MACHINES FOR THE

MACHINING OF SURFACEC 8

TABLE 2. RECOMMENDED HEIGHT VALUE 22

TABLE 3. GEOMETRIC TOLERANCING SYMBOLS 25

TABLE 4. PARTIAL COMPONENT PROCESS SUMMARY 109

TABLE 5. CHAIN STRUCTURE 118

TABLE 6. FORM CODE (DIGITS 1-5) FOR ROTATIONAL

PARTS IN THE OPITZ SYSTEM. PART CLASSES

0, 1, AND 2 126

TABLE 7. THE VUOSO-PRAHA CODE 131

TABLE 8. STRUCTURE OF THE KK-3 CODING SYSTEM
(ROTATIONAL COMPONENTS) 132
TABLE 9. FUNCTIONAL DELINEATION OF THE KK-3
CODING SYSTEM 133
TABLE 10. MACROS OF SORICH CAPP SYSTEM 166
CHAPTER 1
 2

. Computer Aided Process Planning (CAPP)

A common task practiced in most manufacturing industries is

process planning. Process planning is the function within a

manufacturing facility that establishes which processes and

parameters are to be used (as well as those machines capable

of performing these processes) to convert a piece part from
its initial form to a final form predetermined (usually by a
design engineer) in an engineering drawing. Alternatively,
process planning could be defined as the act of preparing
detailed work instructions to produce (machine or assemble)
a part. Required inputs to the process planning are:

• Geometric Features
• Dimensions
• Tolerances
• Materials
• Finishes.

These inputs are analyzed and evaluated in order to select

an appropriate sequence of processing operations based upon
specific, available machinery and workstations.

A process plan needs to generate operational details or

outputs such as:
 3

• Operation sequence
• Feeds
• Depth of cut
• Required tooling
• Machine tools
• Standard times and costs
• Workpiece holding devices(jigs)
• Material Removal Rate

In general, a process plan is prepared using available

design data and manufacturing knowledge. The process
planning function has been described as "a subsystem that is

responsible for the conversion of design data to work

instructions".

A major thrust has been to promote the use of Computer-Aided

Process Planning (CAPP) as a necessary, logical interface

between Computer-Aided Design (CAD), and Computer-Aided

Manufacturing (CAM).

Current CAD systems tend to concentrate on automated design,

solids modeling, drafting, and tool path generation for

numerically controlled machine tools.

CAM systems tend to concentrate on automated manufacturing

and robot usage. Integration of these two procedures could
be brought about ideally through the use of CAPP.
 4

Softwares are available to perform these tasks, but at quite

an expense. To demonstrate the theory of CAPP, the software
package, Quattro-Pro was customized to produce a route sheet
and bill of materials. The products of the Intrex
Corporation (a custom furniture company) were used as a
model for this thesis. Most CAPP systems are component

oriented. Keeping the Intrex Corporation in mind (which is a

product oriented company) we designed our CAPP system to
function on a product oriented basis. However, the system
can be adapted to serve as a component based CAPP system.
 5

CHAPTER 2
 6

2. Process Planning Methods

There are essentially three possible approaches to process

planning:

• Manual
• Computer-assisted variant
• Computerized Generative

It is stated by H.J. Steudel that it may be more appropriate

to use a particular approach under a given set of
circumstances.

Closely related to process planning are the functions of

determining appropriate cutting conditions for the machining
operations and setting the time standards for the
operations. All three functions:

• Planning the process

• Determining the cutting conditions and
• Setting the time standards

These tasks have traditionally been carried out as tasks

with a very high manual content. They are also typically
routine tasks in which similar or even identical decisions
 7

are repeated over and over. Today, these kinds of decisions

are being made with the aid of computers.

The initial material can take a number of forms, the most

common of which are bar stock, plate, castings, forgings, or

maybe just a slab (of any geometry). The slab of material is

normally a burnout cut to some to some rough dimensions.
This metal slab can consist of almost any geometry. The
selection of raw stock requires some study to ensure its

specified engineering properties and economical viability.

With these raw materials as a base, the process planner must

prepare a list of processes to convert this normally
predetermined material into a predetermined final shape. The
processes used by a process planner in a discrete-part
metal-manufacturing industry are listed in Table-1. Some of

the operations given in Figure 1 are often considered

subsets of some major category (e.g., facing can be

considered a subset of turning and reaming can be considered

a subset of drilling).

The process plan in Figure 2 (Intrex Process plan) is

sometimes called an operation sheet, route sheet, or

operation planning summary. A detailed plan usually contains

the route, processes, process parameters , and machine and
tool selections. In a more general sense, a process is
 8

TABLE 1. OPERATIONS AND MACHINES FOR THE MACHINING OF

SURFACES
 9

TABLE 1 (Continued).
 10

FIGURE 1. Standard Material Shapes.

 11

called an operation (including manual operations). The route

is the operation sequence. The process plan provides the
instructions for the production of the part. These
instructions dictate the cost, quality, and rate of
production ; therefore, process planning is of utmost
importance to the production system.

In a conventional production system, a process plan is

created by a process planner, who examines a new part
(engineering drawing) and then determines the appropriate
procedure to produce it. The previous experience of the
process planner is critical to the success of the plan.
Planning, as practiced today, is as much an art as it is a
formal procedure.

As mentioned previously, there are numerous factors that

effect process planning. The design (shape), dimensions,
tolerance, surface finish, size, material type, quantity,
and the manufacturing system itself all contribute to the
selection of operations and the operation sequence.
In addition to operation sequencing and operation selection,
the selection of tooling, jigs and fixtures (sometimes the
design of jigs and fixtures) is also a major part of the
process planning function. The tooling portion includes
selection of both the tool itself and the machine on which
the tool is used. There is a limited set of commercially
available tools, with different shapes, diameters, lengths,
 12

Intrex Corporation
Manufacturing Methods(Process Plan)

Family: The Taper Series

Description: 54" Conference Table
Part: Apron
Department: Table
Note: All dimensions are final dimensions after the operation.
# Operation Machine/ Setup Description
Description Equipment
1 Select 4 strips for Square Check if strips are twisted
4 aprons standing on edges, if twisted more
than 1/32" over 3" in width, select
other strip(s).
2 Trim 4 strips to Table Saw Use 10" dia., 60 teeth,TCG(triple
Width: - 3" chip grind) Carbide tipped circular
saw blade. Note: Trim both edges of
strip.
3 Cut 4 strips to Table Saw 42" dia. insert, 42" stopper, jig
Length: - 26 13/16" # j-aa196. Lay strip on insert.
Use 10" dia., 60 teeth, TCG Carbide
tipped circular saw blade. Slide
strip(s) on track of insert to check
if strip is twisted, if twisted
select other strip(s). Note: Trim
both ends.

FIGURE 2. Process Plan.

numbers of teeth, and alternative tool materials.. The most

commonly used tool materials are high speed steel (HSS) and
tungsten carbide (WC). In recent years, ceramic tools have
 13

also become reasonably popular. WC can machine metal at a

higher cutting speed than HSS, but costs more and is
difficult or impossible to regrind. WC is also brittle and
is not recommended for interrupted cuts. All these
conditions have to be taken into account to select an
appropriate tool.

Jigs and fixtures are devices to guide a tool or hold a

workpiece for better machining. Very few standard devices
are available. Even with standard devices, application is
normally left to the machine operator. Process planning,
however, should consider the impact of tooling and jigs and
fixtures on the quality of the product during the selection
of the production operations.

Some of the prerequisites for the process planners are

given.

2.1. Engineering Design

Engineering design is the partial realization of a

designer's concept. The designer normally cannot directly
transform a concept into a physical item. Instead, the
designer conveys the idea to others through an alternative
 14

medium, such as an engineering drawing, and then the

manufacturing engineer or machinist produces the design.
When a farmer needed a tool prior to the industrialized age,
he normally went to a blacksmith and told the blacksmith the
shape and size of the tool required. Because most tools were
simple and did not require significant accuracy , the
blacksmith would get a pretty good picture of a hoe or a
plow through the verbal description. If the blacksmith still
did not understand, the farmer could sketch the tool on the
dirt floor of the blacksmith shop.

As product design requirements became more complex, a

picture became necessary to relate the information to
others. Multiview orthographic drawing have long been
adopted by engineers as the standards tool to represent a
design. Design information can be passed from the designer
to others who are well trained in reading such drawings. The
object a designer draws on paper can be interpreted and
reconstructed in a viewer's mind. The capability of
transforming an object from one medium to another (e.g.,
from a two-dimensional three view drawing to a three-
dimensional picture) and an understanding of the rules of
drawing are prerequisite to pass design information from the
designer to others without error.

There are several methods available to represent an

engineering drawing. The conventional method is drafting on
 15

FIGURE 3. Idea Sketch Prepared By Leonardo da Vinci

(1452-1519).
paper with pen or pencil. Manual drafting is tedious and
requires a tremendous amount of patience and time. Recently,
 16

computer-aided drafting systems have been implemented to

improve drafting efficiency. The major objective of these
systems is to assist the draftsman with tedious drawing and
redrawing. Partially completed or completed drawings from a
graphics tablet or screen can be stored in a computer and
retrieved when needed.

Engineering drawing is a universal language used to

represent a designer's idea to others. It is the most
accepted medium of communication in all phases of industrial
and engineering work. In ancient times, before multiview
drawing standards/concepts were adopted, perspective
drawings were normally employed. The great master of art
during the Renaissance, Leonardo da Vinci, designed several
machines and mechanical components (which still amazes the
contemporary designers) using perspective sketches, Figure
3. Today, pictorial drawings are still used to supplement
other design representation.

2.2. Dimensioning

A design is expected to convey a complete description of

every detail of a part. However, dimensioning is as
important as the geometric information. It saves money


17

directly by providing for maximum producibility of the part

through maximum production. It provides uniformity and
convenience in drawing delineation and interpretation,
thereby reducing controversy and guesswork.

According to the American National Standards Institute

(ANSI) standards, the following are the basic rules that
should be observed in dimensioning any drawing:

2.2.1. Show enough dimensions so that the intended sizes

and shapes can be determined without calculating or assuming
any distances.

2.2.2. State each dimension clearly, so that it can be

interpreted in only one way.

2.2.3. Show the dimensions between points, lines, or

surfaces that have a necessary and specific relation to each
other or that control the location of other components or
mating parts.

2.2.4. Select and arrange dimensions to avoid

accumulations of tolerances that may permit various

interpretations and cause unsatisfactory mating of parts and

failure in use.
 18

2.2.5. Show each dimension only once.

2.2.6. Where possible, dimension each feature in the view

where it appears in profile, and where its true shape
appears.

2.2.7. Wherever possible, specify dimensions to make use

of readily available materials, parts, tools, and gages.
Savings are often possible when drawings specify:

a. Commonly used materials in stock sizes

b. Parts generally recognized as commercially standard

c. Sizes that can be produced with standard tools and

inspected with standard gages

d. Tolerances from accepted published standards.

2.3. Tolerances

Since it is impossible to produce the exact dimension

specified, a tolerance is used to show the acceptable

 19

variation in a dimension. The higher the quality a product

has, the smaller the tolerance value specified. Tighter

tolerances are translated into more careful production

procedures and more rigorous inspection. It ensures that
design dimensional and tolerance requirements, as they
relate to actual function, are specifically stated and thus
carried out. It ensures interchangeability of mating parts

FIGURE 4. Tolerancing: Bilateral And Unilateral. Dashed

Lines Show The Tolerance Limits.

at assembly. It is increasingly becoming the "spoken word"

throughout industry, the military, and, internationally, on
engineering drawing documentation. Every engineer or
technician involved in originating or reading a drawing
 20

should have a working knowledge of this new state of the

art.

There are two types of tolerances:

2.3.1. Bilateral tolerance

2.3.2. Unilateral tolerance

Figure 4 shows bilateral and unilateral tolerance.

Unilateral tolerances, such as 1.00 +0.00
+0. -0.05/ specify
dimensional variation from the basic size (i.e., decrease)
in one direction in relation to the basic size; for example,

1.00 +0 00 -0.05 = 0.95 to 1.00

The basic location where most dimension lines originate is

the reference location, cumulative errors can be eliminated.
Most mechanical parts contain both working surfaces and
nonworking surfaces. Working surfaces are those for items
such as bearings, pistons, and gear teeth, for which optimum
performance may require control of the surface
characteristics. Nonworking surfaces, such as the exterior
walls of an engine block, crankcase, or differential
housings, seldom require surface control. For surfaces that
require surface control, control surface symbols can be
used. Figure 5 shows how these symbols are used.
 21

FIGURE 5. Surface Control Symbols.

In the symbol, several surface characteristics are

specified. The roughness height is the roughness value as

normally related to the surface finish. It is the average
amount of irregularity above or below an assumed center
line. It is expressed in microinches (micro inch = 0.000001
in.) or, in the metric system, in micrometer (micro meter =
0.000001 m). Recommended roughness heights are given in

Table 2. Lay is another property of machined surface. It is

 22

the direction of the predominant surface pattern, produced

by tool marks or grain of the surface, ordinarily dependent
upon the production methods used. Lay symbols are listed in

Figure 6 .

TABLE 2. RECOMMENDED HEIGHT VALUE

Roughness value
(Ain.) Type of surface Purpose

1000 Extremely rough Used for clearance surfaces only where

good appearance is not required
500 Rough Used where vibration, fatigue, and stress
concentration are not critical and close
tolerances are not required
250 Medium Most popular for general use where
stress requirements and appearance
are essential
125 Average smooth Suitable for mating surfaces of parts held
together by bolts and rivets with no
motion between them
63 Better-than-average For close fits or stressed parts except
finish rotating shafts, axles, and parts sub-
ject to extreme vibration
32 Fine finish Used where stress concentration is high
and for such applications as bearings
16 Very fine finish Used where smoothness is of primary
importance. such as high-speed shaft
bearings, heavily loaded bearings, and
extreme tension members
8 Extremely fine finish Use for such parts as surfaces of cylin-
produced by cy- der
lindrical grinding,
honing, lapping,
or butting
4 Superfine finish Used on areas where packings and rings
produced by hon- must slide across the surface where
ing, lapping. buff- lubrication is not dependable
ing, or polishing
 23

FIGURE 6. Lay Symbols.

 24

2.4. Geometric Tolerances

Conventional methods of dimensioning only provide

information concerning size and surface condition. A

component can be produced without a guarantee of
interchangeability. For example, in Figure 7, both
components (b) and (c) satisfy the dimension specified in
(a), that is, the diameter of components (b) and (c) is
0.501 in. over the entire length of the component.
Obviously, both (b) and (c) are not desirable. However, as
specified, both (b) and (c) meet specifications.

FIGURE 7. Illustration Of Some Addition Part Conditions.

 25

Geometric Tolerances specifies the tolerance of geometric

characteristics. Basic geometric characteristics as defined
by the ANSI Y14.5M 1982 standard that are used as the
building blocks for geometric dimensioning and Tolerances
are given in Table 3.

TABLE 3. GEOMETRIC TOLERANCING SYMBOLS

 26

Using Symbols

FIGURE 8.
 27

The general use of symbols instead of notes on a drawing

provides a number of advantages. Figure 8 below incorporate
the geometric characteristic symbols with datum and feature
control symbols. Some of the advantages of symbols over
notes are:

2.4.1. The symbols have uniform meaning. A note can be

stated inconsistently, with a possibility of
misunderstanding.

2.4.2. Symbols are compact, quickly drawn, and can be

placed on the drawing where the control applies; symbols
adapt readily to computer applications.

Notes require much more time and space, tend to be scattered

on the drawing, often appear as footnotes which separate the
note from the feature to which it applies.

2.4.3. Symbols are the international language and

surmount individual language barriers.

Notes may require translation if the drawing is used in

another country.

2.4.4. Symbols can be applied with drafting templates or

computer techniques and retain better legibility in various
forms of copy reproduction.
 28

2.4.5. Geometric Tolerances symbols follow the

established precedent of other well-known symbol systems,
e.g., electrical and electronic, welding, surface texture,

etc.

To specify the geometric tolerances, reference

features--either planes, lines, or surfaces--can be
established:

2.5. Maximum Material Condition (MMC)

One of the fundamental and most important of geometric

dimensioning and Tolerances is Maximum Material Condition.
The condition in which a feature of size contains the
maximum amount of material within the stated limits of size:
for example, minimum hole diameter and maximum shaft
diameter.
 29

2.6. Regardless of Feature Size (RFS)

The term used to indicate that a geometric tolerance or

datum reference applies at any increment of size of the
feature within its size tolerance. Unlike maximum material
condition, the "regardless of feature size" principle
permits no additional positional, form or orientation
tolerance, no matter to which size the related features are
produced. It is really the independent form of dimensioning
and Tolerances which has always been used prior to the
introduction of the MMC principle.

2.7. Least Material Condition (LMC)

The term used is the condition in which a feature of size

contains the least amount of material within the stated
limits of size, for example, maximum hole diameter, minimum
shaft diameter. LMC is the condition opposite to MMC. For
example, a shaft is at least material condition when it is
at its high limit of size.
 30

2.8. Datum

A theoretically exact point, axis, or plane derived from the

true geometric counterpart of a specified datum feature. A
datum is the origin from which the location or geometric
characteristics of features of a part are established. Datum
surfaces and datum features are actual part surfaces or
features used to establish datums. They include all the
surface or feature inaccuracies.

2.9. Basic

A numerical value used to describe the theoretically exact

size, profile, orientation, or location of a feature or
datum target. It is used as the basis from which permissible
variations are established by tolerances on other dimensions
or in notes.

Use of a Basic dimension, which is a theoretical and exact

value, requires also tolerance stating the permissible
variation from this exact value (most often relative to a
position angularity or profile requirement). A basic
dimension states only half the requirement. To complete it,
 31

a tolerance must be associated with the features involved in

the Basic dimension.

2.10. Machine Tooling

The Machine Tooling portion includes selection of both the

tool itself and the machine on which the tool is used.
Machine tooling refers to the selection and design of
cutting tools, jigs, fixtures, and dies required to perform
a specific operation. This field, which is generally called
tool engineering, has much to do with the operating
efficiency of machines. A decision as to whether an air
chuck or a mechanical chuck should be used; whether a
cemented carbide tool or one made from high-speed steel will
perform better; what speeds and feeds produce the most
satisfactory results; whether water, oil, or air will serve
best as a coolant - these are typical problems encountered
by the process planner.
 32

2.11. Jigs

Jigs are fixtures used in production drilling, tapping,

boring, and reaming operations. Their function is:

2.11.1. To reduce the cost of operation.

2.11.2. To increase production

2.11.3. To assure high accuracy

2.11.4. To provide interchangeability

A jig must hold the part and guide the cutting tool. Such a
device is a good example of the transfer of skill from a
mechanic to an accessory part of a machine; thus the
operation may be performed accurately by an unskilled
operator. This is well illustrated in the drilling of four
holes in a plate by using a plate or channel jig similar to
the one shown in Figure 9. The jig is made with hardened
steel bushings which locate the positions of the four holes
accurately. Any number of plates may be clamped to this jig,
and each part will be identical with the other.

Jigs perform the same function but differ widely in

appearance according to the shape and design of the part to
 33

FIGURE 9. Plate Drilling Jig.

be worked on. Classification is based on their general

appearance and construction.

A jig should be designed to provide quick and easy loading

and unloading. Likewise, clamping devices must be positive,

and the design should be such that there is no question

about the proper location of the part in the jig. Clearance

is usually provided under drill bushings to allow chips to
escape without having to go through them. This is important
if much metal is to be removed. In addition, provision
should be made for rapid cleaning of chips from the jig.

Most jigs utilize standard parts such as drill bushings,

thumb screws, jig bodies, and numerous other parts. Jigs are
 34

not limited to drilling operations but are also used on

tapping, counterboring, and reaming operations.

2.12. Fixtures

A fixture is a work-holding and work-supporting device that

is securely held or fixed to a machine. Fixture, unlike
jigs, do not guide the tool. Their primary function is:

2.12.1. To reduce the cost of operation

2.12.2. To increase production

2.12.3. To enable complex-shaped, and often heavy, parts

to be machined by being held rigidly to a machine.

Fixtures are used most on lathes, turret lathes, milling

machines, boring equipment, shapers, and planers.

Fixtures to hold a particular part are made from a grey cast

iron or from steel plate by welding or bolting. It is fixed-
that is bolted, clamped, or "set" with a low-melting alloy-
to the machine. A fixture has the locating pins or machined
blocks against which the workpiece is tightly held by
 35

clamping or bolting. In order to assure interchangeability,

the locating devices are made from hardened steel. Many
fixtures are massive because they, like a machine frame, may
have to withstand large dynamic forces. Since all fixtures
are between the workpiece and the machine, their rigidity
and the rigidity of their attachment to the machine are
paramount. Duplex fixtures have been built to allow for the
loading or unloading of one side of fixture while the
machining operation is taking place on a part clamped to the
other side.

2.13. Dies

Dies for both the hot and cold-chamber machines are similar
in construction because there is little difference in the
method of holding and operating them. They are made in two
sections to provide a means of removing the castings and are
usually equipped with heavy dowel pins to keep the halves in
proper alignment. Metal enters the stationary side when the
die is locked in closed position. As the die opens, the
ejector through the die force the casting from the cavity
and fixed cores. The dies are provided with a separate
mechanism for moving the ejector plate or movable cores. The
life of these molds depends on the metal cast and may range
 36

from 10,000 fillings, if brass castings are made, to several

million if zinc is used.

It is always desirable to provide vents and small overflow

wells on one side of a die to facilitate the escape of air
and to catch surplus metal that has passed through the die
cavity. In spite of this provision, there is a certain
amount of flash metal that must be trimmed off in the
finishing operation.

For large or complex castings, a single-cavity mold is used.

If the quantity of castings to be produced is large and they
are relatively small in size, a multiple-cavity die can be
used. A combination die is one that has two or more
cavities, each of which is different. They are frequently
made up of insert blocks that can be removed so that die
blocks can be substituted. Most dies are provided with
channels for water cooling to keep the die at the correct
temperature for rapid production.
 37

CHAPTER 3
 38

3. Economical Production Criteria

The increasing need for finished goods in large quantities

has, in the past, led engineers to search for and develop
new methods for production. Many individual developments in
the various branches of manufacturing technology have been
made and have allowed the increased production of improved
finished goods at lower cost. The cost of a product is based
on expenditures for raw materials, machines, labor, sales,
warehousing, and overhead. Machine and labor costs are
inexorably related and make up, along with raw materials
expenditures, the bulk of production costs. When a material
is chosen, the process, including the machine, may be
automatically specified, or if a machine is available, the
raw material that can be used may be limited. One could say
that the purpose of economical production is to produce a
product at a profit. This infers that the cost must be
acceptable and competitive, also, a demand for the product
must exist, or else it must be created.

Since the first use of machine tools, there has been a

gradual, but steady, trend toward making machines more
efficient by combining time and labor. To meet these needs,
machine tools have become complex both in design and in
control. Automatic features have been built into many
machines, and some are completely automatic. This technical
 39

development has made it possible for industry to attain a

high production rate with the accompanying low labor cost
that is an essential development for any society wishing to
enjoy high living standards.

3.1. Automation

Automation was first coined by the automotive industry to

describe methods used for automatically controlling various
machines and the conveying of workpieces between these
machines. Automation is a technology concerned with the
application of mechanical, electronic, and computer-based
systems to operate and control production. Coordination is
not necessarily automation. Automation implies self-
correction sch as the governor controlling the speed of an
engine or the thermostat controlling the temperature in a
house. On a machine doing a turning operation the feedback
control would feed the tool in slightly if the diameter
became too large. In other words, automation gives to the
machine some power of choice or ability to correct itself if
certain prescribed limits are exceeded. Automation is very
tricky. One can't be successful just by automating the
facility. This technology includes:
 40

• Automatic machine tools to process parts

• Automatic assembly machines

• Industrial robots

• Automatic material handling and storage systems

• Automatic inspection systems for quality control

• Feedback control and computer process control

• Computer systems for planning, scheduling

• Computer systems for data collection and decision

making to support manufacturing activities.

3.1.1. Types of automation

Automation production systems in manufacturing can best be

classified into three basic types:

3.1.1.1. Fixed Automation

3.1.1.2. Programmable automation

 41

3.1.1.3. Flexible automation

3.1.1.1. Fixed Automation is a system in which the

sequence of processing (or assembly) operations is fixed by
the equipment configuration. The typical features of fixed
automation are:

• High initial investment for custom-engineered equipment

• High production rates

• Relatively inflexible in accommodating product changes

The economic justification for fixed automation is found

with very high demand rates and volumes. Examples of fixed
automation include mechanized assembly lines (starting
around 1913-- the product moved along mechanized conveyors,
but the workstations along the line were manually operated)
and machining transfer lines (beginning around 1924).

3.1.1.2. Programmable automation, the production

equipment is designed with the capability to change the
operations to accommodate different product configurations.
Some of the typical features that characterize programmable

automation include:
 42

• High investment in general-purpose equipment

• Low production rates relative to fixed automation

• Flexibility to deal with changes in product

configuration

• Most suitable for batch production

Automated production systems that are programmable are used

in low and medium-volume production. The parts or products
are typically made in batches. To produce each new batch of
a different product, the system must be reprogrammed with
the set of machine instructions that correspond to the new
product. Examples of programmable automation include
numerically controlled machine tools (first prototype
demonstrated in 1952) and industrial robots (initial
applications around 1961), although the technology has its
roots in the Jacquard loom (1801).

3.1.1.3. Flexible automation is an extension of

programmable automation. The concept of flexible automation
has developed only over the last 15 or 20 years, and the
principles are still evolving. A flexible automated system
is one that is capable of producing a variety of products
(or parts) with virtually no time lost for changeovers from
one product to the next. There is no production time lost
 43

while reprogramming the system and altering the physical

setup (tooling, fixtures, machine settings). Consequently,
the system can produce various combinations and schedules of
products, instead of requiring that they be made in separate
batches. The features of flexible automation can be
summarized as follows:

• High investment for a custom-engineered system

• Continuous production of variable mixtures of products

• Medium production rates

• Flexibility to deal with product design variations

The essential features that distinguish flexible automation

from programmable automation are:

a. The capacity to change part programs with no lost

production time.

b. The capability to change over the physical setup, again

with no lost production time.

The above features allow the automated production system to

continue production without the downtime between batches
that is characteristic of programmable automation. Changing
 44

the part programs is generally accomplished by preparing the

programs off-line on a computer system and electronically

transmitting the programs to the automated production

system. Therefore, the time required to do the programming
for the next job does not interrupt production on the
current job. Advances in computer systems technology are
largely responsible for this programming capability in

flexible automation. Changing the physical setup between

parts is accomplished by making the changeover off-line and
then moving it into place simultaneously as the next part

comes into position for processing.

Although it is the intention of the thesis to emphasize the

use of computers in manufacturing, but first, how to prepare

a process plan manually is shown.

 45

CHAPTER 4
 46

4. Manual Process Planning

There are variations in the level of detail found in route

sheets among different companies and industries. In the one
extreme, process planning is accomplished by releasing the
part print to the production shop with the instructions
"make to drawing." Most firms provide a more detail list of
steps describing each operation and identifying each work
center. The traditional approach to process planning has
been to manually examine an engineering drawing and develop
a detailed process plan based upon knowledge of process and
machine capabilities, tooling, materials, and shop
practices. The process planner, given a specific material,
must endeavor to build a list of processes for converting
the given raw material into its desired final geometry. In
any case, it is traditionally the task of the manufacturing
engineers or industrial engineers in an organization to
write these process plans for the new part designs to be
produced by the shop. Manual process planning leans very
heavily upon the judgement and experience of the planner. It
is the manufacturing engineer's responsibility to determine
an optimal routing for each new part design and requires
well trained, experienced personnel who are well versed in
shop floor practices. However, individual engineers each
have their own opinions about what constitutes the best
 47

routing and would always exhibit the personal preferences

and prejudice of the individual.

For a model shop, where all the machinists are highly

skilled in several machines and most parts produced are one
of a kind, the process plan is usually nothing but a list of
workstation routes. The remaining detail is left to the
machinists. In the case where a part is produced in an
entirely automated transfer line, the process plan contains
a breakdown of every detail. Machines, tools, fixtures, and
dies are designed to execute the required operations.
However, these are extreme cases; in most machine shop,
small-batch production is the norm.

In order to prepare a process plan, a process planner has to

have the following knowledge:

• Ability to interpret an engineering drawing

• Familiarity with manufacturing processes and practice

• Familiarity with tooling and fixtures

• Know what resources are available in the shop

• Know how to use reference books, such as machinability

data handbooks
 48

• Ability to do computations on machining time and cost

• Familiarity with the raw materials

• Know the relative costs of processes, toolings, and raw

materials.

Objective/Criterion in selectivity of a Process Plan, the

following are some steps that have to be taken:

a. Economic Criteria

a.l. Minimum processing time

a.2. Minimum material cost

b. Study the overall shape of the part. Use this

information to classify the part and determine the type
of workstation needed.

c. Thoroughly study the drawing. Try to identify all

manufacturing features and notes.

d. Determine the best raw material shape to use if

raw stock is not given.


49

e. Identify datum surfaces. Use information on datum

surfaces to determine the setups.

f. Select machine for each setup.

g. Determine the rough sequence of operations

necessary to create all the features for each setup.

h. Sequence the operations determined in the previous

step. Check whether there is any interference or
dependency between operations. Use this information to

modify the sequence or operations.

i. Select tools for each operation. Try to use the

same tool for several operations if possible. Keep in

mind the trade-off on tool-change time and estimated

machining time.

Select or design fixtures for each setup.

•

k. Select or design jigs for each setup.

1. Evaluate the plan generated thus far and make necessary

modifications.

m. Select cutting parameters for each operation.

 50

n. Prepare the final process-plan document.

One has to keep in mind that during each of these steps,

decisions are made based on an evaluation of many factors.
For example, tool selection is based on the feature to be
created and other related features as well as the machine to
be used. Operations selection depends on the features to be
created and the capability of the machine selected. Machine
selection is normally determined by the operations required.

The following are some of the disadvantages of manual

process planning:

a. The required experience for a manual planner requires

years of hard work to accumulate.

b. The experience thus developed would not apply to new

processes.

c. Experience gathered would represent approximate data

only.

The disadvantages of this method has left to investigate the

automation of the process planning function.

 51

CHAPTER 5
 52

5. Computer-Aided Process Planning

Because of the problems encountered with manual process

planning, attempts have been made in recent years to capture
the logic, judgement and experience required for this
important function and incorporate them into computer

programs. The use of computer resources to aid the process

planner in a systematic determination of proper methods to
be used in the production processes is called CAPP. These
systems manage the storage, retrieval, distribution, and
maintenance of the process plan library. They have been
extended to include the logic used by process planners in
choosing alternate process methods. Early attempts to

automate process planning consisted primarily of building

computer-assisted systems for report generation, storage,

and retrieval of plans. A database system with a standard

form editor is what many early systems encompassed.

Formatting of plans was performed automatically by the
system. Process planners simply filled in the details. The
storage and retrieval of plans are based on part number,

part name, or project ID. When used effectively, these

systems can save up to 40% of a process planner's time.

Perhaps the best known automated process planning system is

the CAM-I automated process-planning system, CAPP (CAM-I
stands for Computer-Aided Manufacturing--International, a
 53

FIGURE 10. Typical Process-Planning System.

 54

nonprofit industrial research organization). In CAPP,

previously prepared process plans are stored in a database.
When a new component is planned, a process plan for a
similar component is retrieved and subsequently modified by
a process planner to satisfy special requirements. The
technique involved is called Group-Technology (GT) based
variant planning. The typical organization of a variant
process-planning system is shown in Figure 10. During the
last decade or so, there has been much interest in
automating the task of process planning of CAPP systems. The
shop-trained people who are familiar with the details of
machining and other processes are gradually retiring, and
these people will be unavailable in the future to do process
planning. An alternative way of accomplishing this function
is needed, and CAPP systems are providing this alternative.
Recent developments in Computer-Aided Process Planning have
been focused on eliminating the process planner from the
entire planning function. Computer-Aided Process Planning
can reduce some of the decision making required during a
process planning. It has the following advantages:

a. It can reduce the skill required of a planner.

b. It can reduce the process-planning time.

c. It can reduce both process-planning and manufacturing

costs.
 55

d. It can create more consistent plans.

e. It can produce more accurate plans.

f. It can increase productivity.

Process planning is the critical bridge between design and

manufacturing, Figure 11 illustrates this situation. Design
information can be translated into manufacturing language
only through process planning. Today, both automated design,
Computer-Aided Design (CAD) and manufacturing, Computer-
Aided Manufacturing (CAM) have been implemented.
Integrating, or bridging, these functions require automated
process planning.

FIGURE 11. Process-Planning Bridges Design And

Manufacturing.
 56

5.1. Benefits of CAPP

Among the benefits derived from computer-aided process

planning, some of them are the following:

a. Process Rationalization and Standardization: Automated

process planning leads to more logical and consistent
plans than when process planning is done completely
manually. Standard plans tend to result in lower
manufacturing costs and higher product quality.

b. Increased Productivity of Process Planners: The

systematic approach and the availability of standard
process plans in the data files permit more work to be
accomplished by the process planners. One system was
reported to increase productivity by 600%.

c. Reduced Lead Time for Process Planning: Process

planners working with the CAPP system can provide route
sheets in a shorter lead time compared to manual
preparation.

d. Improved Legibility: Computer-prepared route sheets are

neater and easier to read than manually prepared route
sheets.
 57

f. Incorporation of Other Application Programs: The CAPP

program can be interfaced with other application
programs, such as cost estimating, work standards, and
others.

Modern CAPP systems are focused around two technologies.

These technologies are called:

5.2. Retrieval CAPP systems

5.3. Generative CAPP systems

5.2. Retrieval CAPP systems

The Retrieval-type computer-aided process planning systems

also called Variant CAPP systems requires a catalog of
standard process plans to be stored in a computer systems.
The part being planned is classified and coded using the
principles of Group Technology (GT). GT approach is required

for parts classification and coding system to organize the

computer files and to permit efficient retrieval of the
appropriate process plan. The variant approach uses library
retrieval procedures to find standard plans for similar
components. A process plan that can be used by a family of
 58

components is called a standard plan. A standard plan is

stored permanently in the database with a family code number
as its key. The standard plans are created manually by
process planners. The standard process plans are based on
current part routings in use in the factory, or on an ideal
plan that is prepared for each family. There is no
limitation to the detail that a standard plan can contain.
However, it must contain at least a sequence of fabrication
steps or operations. For some new parts, when a standard
plan is retrieved, a certain degree of modification is
usually necessary in order to use the plan on a new
component. The development of the data base of these process
plans requires substantial effort.

The retrieval method and the logic in variant systems are

predicated on the grouping of parts into families. Common
manufacturing methods can then be identified for each
family. Such common manufacturing methods are represented by
standard plans.

The mechanism of standard-plan retrieval is based on part

families. A family is represented by a family matrix that
includes all possible members. In use, a retrieval CAPP
system operates as illustrated in Figure 12. The user begins
by deriving the GT code number for the component for which
the process plan is to be determined. With this code number,
a search is made of the part family file to determine if a
 59

standard route sheet exists for the given part code. If the
file contains a process plan for the part, it is retrieved
and displayed for the user. The standard process plan is
examined to determine whether any modifications are
necessary. It might be that the new part has the same code

FIGURE 12. General Procedure For Using One Of The Retrieval

Computer-Aided Process Planning Systems.
 60

number, there are minor differences in the processes to make

the part. The user edits the standard plan accordingly. It
is this capacity to alter an existing process plan that
gives the retrieval system its other name: Variant CAPP
system.

If the file does not contain a standard process plan for the
given code number, the user may search the computer file for
a similar or related code number for which a standard route
sheet does exist. Either by editing an existing process
plan, or by starting from scratch, the user writes the route
for the new part. This route sheet becomes the standard
process plan for the new part code number. In general,
variant process-planning systems have two operational
stages:

5.2.1. The Preparatory Stage

5.2.2. A Production Stage

5.2.1. The Preparatory Stage

The Preparatory work is required when a company first starts

implementing a variant system. During the preparatory stage,
existing components are coded, classified, and subsequently
grouped into families. The first step is to choose an
 61

appropriate coding system. The coding system must cover the

entire spectrum of parts produced in the shop. It must be
unambiguous and easy to understand. The special features
exist on the parts must be clearly identified by the coding
system. An existing coding system may be adopted and then

modifications can be made for the specific shop. The coding

of existing components can be a tedious task. Before it can

be done, a thorough study of the inventory of drawings and
process plans has to be completed so that an orderly coding
task can be conducted. The personnel involved in coding must
be trained. They must have the precise understanding of the

coding system. They must generate the identical code for the

same component when they work independently. Inconsistent

coding of components results in redundant and erroneous data

in the database.

After coding is completed, part families can be formed. A

family matrix is then constructed for each part family. Due

to the large number of components involved, a computer

should be used to help construct family matrices. The next

step is to prepare standard process plans for part families.
The preparatory stage is a labor-intensive process. It

requires a tremendous amount of effort. Whatever is prepared

for shop A can be used only for shop A. The system structure
and software can be used by other shops, but the database

must be prepared uniquely by and for each shop.

 62

5.2.2. The Production Stage

The production stage occurs when the system is ready for

use. New components now can be planned. An incoming
component is first coded.

FIGURE 13. Production Stage.

The code is then input to a part-family search routine to

find the family to which the component belongs. The family
number is then used to retrieve a standard plan. The human
 63

planner may modify the standard plan to satisfy the

component design. For a frequently produced part, it might

be desirable to perform the search by direct code matching.

In this case, a process plan (not a standard plan) for an
existing part is retrieved. Figure 13 shows the flow of the
production stage. Some other functions, such as parameter
selection and standard time calculations, can also be added
to make the system more complete.

5.2.3. Database Structure

Because of the large amount of information, database systems

play an important role in variant process planning. A .

database is no more than a group of cross-referenced data

files. The database contains all the necessary information
for an application and can be accessed to by several
different programs for specific applications. There are
three approaches to construct a database:

a. Hierarchical

b. Network

c. Relational
 64

Although the concept and structure of these approaches are

very different, they can serve the same purpose.

For commercial programming, there are several available

database management systems, such as CODASYL, ORACLE, RBASE,
UNIFY, dBASE-V, Quatro Pro, and Lotus 1-2-3. These systems

are high-level languages for database construction and

manipulation. Of course, a database can always be written
using procedural languages such as COBOL, FORTRAN, and C. No
matter what approach and language are used, the basic
structure of the database is the same.

5.2.4. Search Procedure

Once the preparatory stage has been completed, the variant

planning system is ready for production. The basic idea of
the variant system is to retrieve process plans for similar
components. The search for a process plan is based on the
search of a part family to which the component belongs. When
the part family is found, the associated standard plan can
be easily retrieved.
 65

5.2.5. Plan Editing and Parameter Selection

Before a process plan can be released to the shop, some

modification of the standard plan is necessary, and process
parameters must be added to the plan. There are two types of
plan editing: one is the editing of the standard plan itself
in the database, and the other is the editing of the plan
for the component. Editing a standard plan implies that a
permanent change in the stored plan be made. This editing
must be handled very carefully because the effectiveness of
a standard plan affects the process plans generated for the

entire family of components. Aside from the technical

considerations of file maintenance, the structure of the

database must be flexible enough for expansion and additions

and deletions of data records.

Editing a process plan for a component requires the same

expertise as editing a standard plan. However, it is a

temporary change and , therefore, does not affect any other

component in the family. During the editing process, the

standard plan has to be modified to suit the specific needs

of the given component. Some operations or entire records

have to be removed and others must be changed. Additional
operations may also be required to satisfy the design. A

text editor is usually used at this stage.

 66

FIGURE 14. Process-Parameter File.

A complete process plan includes not only operations, but

also process parameters.

Figure 14 shows the structure for the parameter file. Data

in the file are linked so that we can go through the tree to

find the feed and speed for an operation.

 67

The parameter file can be integrated into variant process

plan to select process parameters automatically. Information
such as depth of cut and cutter diameter can be retrieved
directly from the process plan for each operation.

5.3. Generative CAPP systems

Generative CAPP systems is a process planning system that

synthesizes process information in order to process plans
for parts automatically. It uses an automatic, computerized
system consisting of decision logic, formulae, technology
algorithms, and geometry-based data to make decisions
required for a process plan. The computer would employ a set
of algorithms to progress through the various technical and
logical decisions toward a final plan for manufacturing.
Inputs to the system would include a comprehensive
description of the workpart. This may involve the use of
some form of part code number to summarize the workpart
data, but it does not involve the retrieval of existing
standard plans. Instead, the generative CAPP system
synthesizes the design of the optimum process sequence,
based on an analysis of part geometry, material, and other
factors which would influence manufacturing decisions. The
problem of designing a generative CAPP system is considered
 68

part of the field of expert systems, a branch of artificial

intelligence. An expert system is a computer program that
draws on the organized expertise, it is capable of solving
complex problems that normally require one or more human
experts who has years of education and experience. Process
planning fits within the scope of that definition.

There are several ingredients required in fully generative

process planning system.

a. The first ingredient is the technical knowledge of

manufacturing and the logic that is used by successful
process planners must be captured and coded into a
computer program. In an expert system as it would be
applied to process planning, the knowledge and logic of
the human process planners is incorporated into a so-
called "knowledge base". The generative CAPP system
would then use that knowledge base to solve process
planning problems (i.e., create route sheet).

b. The second ingredient in generative process planning is

a computer-compatible description of the part to be
produced. This description contains all of the
pertinent data and information needed to plan the
process sequence. Two possible ways of providing this
description are:
 69

b.l. The geometric model of the part that is developed on a

CAD system during product design.

b.2. A GT code number of the part that defines the part

features in significant detail.

c. The third ingredient in a generative CAPP system is the

capability to apply the process knowledge and planning
logic contained in the knowledge base to a given part
description. In other words, the CAPP system uses its

knowledge base to solve a specific problem--planning

the process for a new part. The problem-solving

procedure is referred to as the "inference engine" in
the terminology of expert system. By using its
knowledge base and inference engine, the CAPP system

synthesizes a new process plan from scratch for each

new part it is presented.

Other planning functions, such as machine selection, tool

selection, and process optimization, can also be automated

using generative planning techniques.

The generative planning techniques/approach has the

following advantages:

a. It can generate consistent process plans rapidly.

 70

b. New components can be planned as easily as existent

components.

c. It can potentially be interfaced with an automated

manufacturing facility to provide detailed and up-to-
date control information.

Decisions on process selection, process sequencing, etc.,

are all made by the system. However, transforming component
data and decision rules into a computer-readable format is
still a major obstacle to be overcome before generative
planning systems become operational. Successful
implementation of this approach requires the following key
developments:

a. Process-planning knowledge must be identified and

captured.

b. The part to be produced must be clearly and precisely

defined in a computer-readable format (e.g., three-
dimensional model and GT code).

c. The captured process-planning knowledge and the part

description data must be incorporated into a unified
manufacturing database.
 71

Today the term "Generative Process Planning' is often

loosely used. Systems with built-in decision logic are often
called generative process-planning systems. The decision
logic consists of the unusual ability to check some
conditional requirements of the component and select a
process. Some systems have decision logic to select several
"canned" process-plan fragments and combine them into a
single process plan. However, no matter what kind of
decision logic is used and how extensively it is used, the
system is usually categorized as a generative system.

Generative process plan design requires long term

investments of people, machinery, and time. Generative
process planning techniques can use forward planning or
backward planning. Forward planning involves modifying the
workpiece until it attains the features required by the
finished product. Backward planning involves starting with
the finished product, and filling it to the shape of the
unmachined workpiece. Each machining process is considered
as a filling process.

There are several generative process-planning systems used

in the industry (e.g., AUTOPLAN OF METCUT [Vogel, 1979;
Vogel and Adlard, 1981], and CPPP of United Technology [Dunn
and Mann, 1978]).
 72

Ideally, a generative process-planning system is a turnkey

system with all the decision logic coded in the software.

The system possesses all the necessary information for

process planning; therefore, no preparatory stage is

required. This is not always the case, however. In order to
generate a more universal process-planning system, variables
such as process limitation, process capabilities, and

process costs must be defined prior to the production stage.

Systems such as CPPP (Kotler, 1980a, 1980b) require user-

supplied decision logic (process models) for each component

family. A wide range of methods have been and can be used
for generative process planning.

5.3.1. Forward and Backward Planning

In variant process planning, process plans are retrieved

from a database. A direction for the planning procedure does
not exist because plans are simply linked to a code.
However, in generative process planning, when process plans
are generated, the system must define an initial state in
order to reach the final state (goal). The path taken
(initial-to-final or final-to-initial) represents the
sequence of processes. For example, the initial state is the
raw material (workpiece) and the final state is the
 73

FIGURE 15. Component To Be Planned: Final Design.

component design. If the planner works on modifying the raw

workpiece until it takes on the final design qualities, the

type of component is shown in Figure 15. The raw workpiece

is a 6.0-in. X 3.0 in. X 2.5 in. block. Starting from the

top surface S1, using forward planning. A milling process is

 74

used to create the final dimension for Si. After Si has been
planned, S2 can be planned. For S2, a chamfering process is
first selected, and then the hole is drilled and tapped. The
progression begins from the raw material and proceeds to the
finished requirements.

Backward planning uses a reverse procedure. Assuming that we

have a finished component, the goal is to fill out to the
unmachined workpiece shape. Each machining process is
considered a filling process. A drilling process can fill a
hole; a reaming process can fill a thin wall (cylinder); and
so on. When applied to the example component, the bottom
most surface is planned first. A tapping process is selected
(this reduces the threaded surface to a smaller-diameter
hole with a rough surface finish). Drilling is then selected
to fill the hole, and so forth, until we finally obtain the
block.

Forward and backward planning may seem similar, however,

they affect the programming of the system significantly.
Planning each process can be characterized by a precondition
of the surface to be machined and a post-condition of the
machining (its results). For forward planning, we must know
the successor surface before we select a process, because
the post-condition of the first process becomes the
precondition for the second process. For example, when we

selected drilling for the threaded hole, we knew that the

 75

thread was going to be cut. Therefore, we rough drilled the

hole using a smaller drill. Otherwise, we might have chosen
a larger drill and no thread could be produced. Backward
planning eliminates this conditioning problem since it
begins with the final surface forms and processes are
selected to satisfy the initial requirements. The transient
surface (intermediate surface) produced by a filling process
is the worst precondition a machining process can accept
(i.e., depth of cut left for finish milling. etc.). Any
filling process that can satisfy the transient surface can
be selected as the successor process.

In forward planning, the objective surface must always be

maintained even though several operations must be taken to
guarantee the result. On the other hand, backward planning
starts with the final requirements (which helps to select
the predecessor process) and searches for the initial
condition or something less accurate (which is easy to
satisfy).

5.3.2. The Input Format

The input format of a process-planning system affects the

ease with which a system can be used and the capability of
 76

the system. A system using a very long special description

language as its input is more difficult to use. The
translation from the original design (either an engineering
drawing or a CAD model) to a specific input format may be
tedious and difficult to automate. In this case, it is
probably easier and faster to plan a component manually than
to prepare the input. However, such input can provide more
complete information about a component, and more planning
functions can be accomplished using the input. This does not
imply that a system using a long and special descriptive
language always provides more planning functions.

Many different input formats have been used in process-

planning systems. Although no/few systems use the same input
format, we can categorize the input for these systems into
the following classes:

a. Code

b. Description Language

c. CAD Models
 77

a. Codes

As discussed in the variant approach, GT codes can be used

as input for variant process-planning systems. Some
generative systems such as APPAS (Wysk, 1977) and GENPLAN
(Tulkoff, 1981) also use part codes as input. Codes used in
generative systems are more detailed and sometimes mix code
digits with explicitly defined parameter values. Since a
code is concise, it is easy to manipulate. When process
capabilities are represented by a code, a simple search
through the process capability to match the component code
will return the desired process. In order to determine the
process sequence, a code for the entire component is
appropriate because it provides global information.

Description Language

Specially designed part-description language can provide

detailed information for process-planning systems. A
language can be designed to provide all of the information
required for the necessary functions of a process-planning
system. The format can be designed such that functions can
easily accomplish their task from the information provided.
 78

AUTAP system (Eversheim et al., 1980) uses a language

similar to a solid modeling language. A component is

described by the union of some primitives and modifiers.

Figure 16 shows the description of a rotational component.

FIGURE 16. AUTAP Data Input (Eversheim et al., 1980).

 79

CYCLE (cylinder), CHAL (chamfer left), CHAR (chamfer right),

UNCUL (undercut), and RADIR (radius right-curved chamfer)
are primitives and modifiers. A process planner can model
the component using the language. Materials, processes,
machine selection, and time estimates can be selected by the
system using the input model. Although reasonably complex
components can be modeled, this language lacks a complete
set of Boolean operators, and modeling a complex component
may be difficult. The process sequence is also affected
directly by the sequence with which a component is modeled.
Although the system models a component from left to right,
it does not reduce the number of possible models for a
component to a single description.

FIGURE 17. CIMS/DEC Component Modeling.

 80

Another system, CIMS/PRO, developed by Iwata et al. (1980),

uses an input language called CIMS/DEC (kakino et al.,
1977). In the CIMS/DEC systems, component shape is modeled

by sweeping (translation or rotation) to generate surfaces,

Figure 17. In CIMS/PRO, a pattern-recognition module

automatically identifies machined surfaces, such as planes,
cylinders, threaded shafts, holes, and grooves. Twenty six
different types of machined surfaces can be characterized.
Tool approach can also be determined (every machined surface
has a set of approach directions predefined). This input
language can model both rotational (by rotation sweep) and
boxlike (by translation sweep) components. It is most
powerful for modeling rotational components without axial-
shape elements. However, if the component is very complex,
it is difficult to model.
GARI (Descotte and Latombe,
1981) is an Artificial
Intelligence (AI) problem
solver. It is one of the
earliest attempts to use
AI in process planning.
A component can be described
by some system words such as
diameter and surface finish,
Figure 18.
 81

Rules can then be applied to determine the processes and

machines needed to produce a part. The knowledge base (where

process and machine capabilities are stored) uses the same

set of system words; therefore, decisions can be reached by

searching the knowledge in order to satisfy the input
description. The input description must, however, be
prepared by a human operator. For a complex component, the
translation of the original design to this input language
can be tedious and difficult.

There are many other systems such as CPPP [Kotler, 1980a,

1980b] and AUTOTECH [Tempelhof, 1980] that use their own

special-shape code, dimensions, and technological data.
Although description languages can provide complete
information for process-planning functions, the main problem
(the difficulty to generate the original design
automatically) is still unresolved. The next class of input
format is aimed at eliminating this problem.

c. CAD Models

Since a design can be modeled effectively in a CAD system,

using a CAD model as input to a process-planning system can
eliminate the human effort of translating a design into a
 82

code or other descriptive form. The increased use of CAD in

industry further points to the benefits of using CAD models
for process-planning data input.

A CAD model contains all the detailed information about a

design. It can provide information for all the planning
functions. However, an algorithm to identify a general
machined surface from a CAD model does not currently exist.
Additional code is needed to specify the machined surface
shape from the raw material shape.

CADCAM (Chang and Wysk, 1981) uses a CAD model for its
input. Several other systems (AUTOPLAN [Vogel and Adlard,
1981] and GENPLANO also use a CAD database interactively for
tool and fixture selection. There is tremendous potential
for using CAD data for process-planning input. However,
substantial work has to be done.

5.3.3. Decision Logic

In a generative process-planning system, the system decision

logic is the core of the software and directs the flow of
program control. The decision logic determines how a process
or processes are selected. The major function of the
decision logic is to match the process capabilities with the
 83

design specification. Process capabilities can be described

by "IF....THEN...." expressions. Such expressions can be
translated into logical statement in a computer program.
Perhaps the most efficient way to translate these
expressions is to code process-capability expressions
directly into a computer language. Information in handbooks
or process boundary tables can be easily translated using a
high-level computer language.

Several methods can be used to describe the decision

structure of process planning. The knowledge-representation
methods are related directly to the decision logic in these
systems. The static data are the representation and the
dynamic use of the data becomes the decision logic. The
following are the list of the decision logic as applied to
process-planning systems. The list is by no means complete.
However, this classification forms a handy framework for the
discussion.

5.3.3.1. Decision Trees

5.3.3.2 Decision Tables

Decision trees and decision tables are ways of describing or

specifying the various actions (decisions) associated with
combinations of (input) conditions. Both decision trees and
decision tables have been employed to help systematize
 84

decision making. With the advent of the digital computer,

increased use of these tools have occurred. Both can be
described by the following example. The activity deals with
a weekend decision that can be described as follows:

"If it is raining, I will go to the arcade and play video

games."

"If it is not raining and hot, I will go to the beach."

"If it is not raining and cool, I will go on a picnic with a

friend."

FIGURE 19. Decision Table And Decision Tree

-

Representation.
 85

This plan can also be represented by either a decision table

or a decision tree, Figure 19.

In a decision table, T implies true to a specified

condition, F implies false, and a blank entry implies that
we do not care. Only when all conditions in a decision table
column are met is the marked action taken.

In a decision tree, the conditions are specified on the

branches of the tree. When all of the branches leading to an
action hold true, the action at the terminal point is taken.
This system can be used represent the decision for the
weekend decision plan given, as well as knowledge base for
process planning.

Both decision trees and decision tables are tools to assist

in decision making. Since decision rules must cover all
possible situations, they must be well thought out before
such a tool can be used for process planning. For a given
set of decision rules, one can list the conditions and
actions, as shown in the example, and then translate the
actions into decision tree or decision table form. Whatever
can be represented by a decision tree can also be
represented by decision table. The primary difference is the
ease and elegance of presentation and programming when a
computer is used.
 86

5.3.3.1. Decision Trees

A decision tree is a natural way to represent process

information. Conditions (IF) are set on the branches of the
tree and predetermined actions can be found at the junction
of each branch.

A decision tree can be implemented as either;

a. computer code or
b. presented as data.

When a decision tree is implemented in computer code, the

tree can be directly translated into a program flow chart.
The root is the start node, Figure 19, and each branch is a
decision node. Each branch has a decision statement (a true
condition and a false condition). At each junction, an
action block is included for the true condition. For a false
condition, another branch might be taken or the process
might be directed to the end of the logic block. When the
false condition includes another branch, these two branches
are said to branch from an OR node. When the false condition
goes directly to the end of an action block (which is rooted
from the same decision statement), the current branch and
the following branch are part of the same AND node. A
 87

decision statement can be a predicate or a mathematical

expression.

FIGURE 20. Structured Flow Chart Corresponds To A Decision-

Tree.
 88

Figure 20 shows a sample decision tree and its flow-chart

representation. It can be written in a programming language
(pseudo-code) as follows:

;root
if El then do Ni enddo
else if E7 then do A5 enddo

else endif endif stop

;node Ni

procedure N1

if E2 then do N2 enddo
else if E3 then do A4 enddo

else endif endif return

;node N2

procedure N2

if E4 then do Al enddo

else if E5 then do A2 enddo

else if E6 then do A3 enddo

else endif endif endif return.

 89

This Language (or interpreter) format allows for the easy

construction of decision trees that are frequently used in
generative process-planning systems. Similar language
formats using FORTRAN, Pascal, C, and so on, have been
developed for general-purpose algorithms. APPAS (Wysk, 1977)
is a typical example of decision-tree logic used for process
planning. Although the approach is easy to implement,
system-expansion work can be difficult, especially for a
programmer other than the original author of the system.

When implementing a decision tree in data form, another

program (system program) is required to interpret the data
and achieve the decision-tree flow. This approach is more
difficult to develop (for the system program). However, once
the system program has been developed, the implementation
and system maintenance are significantly lessened. Again,
however, it can be extremely difficult to add a function
that was not originally included in the system program.
There are many methods that can be used to design such a
system program. A simple example is presented to
demonstrated a basic structure that one can use.

We call this example system DCTREE, DCTREE uses a query

procedure to obtain design information and then print the
final conclusions. In DCTREE, there are three major
components:
 90

a. The decision-tree data

b. A compiler

c. A system run-time module

Part (a) is supplied by a user who translates a decision

tree from graph form into DCTREE input-language format. The
DCTREE compiler compiles the input and saves it in a
computer-usuable format. Finally, the system run-time module
uses the compiled decision tree to generate questions, make
decisions, and print out conclusions.

We first look at the input language. These are two parts of

each input. Expression definition and Tree-structure
definition, Figure 21. In the expression definition, each
expression is preceded by an expression identifier (ID).
Each ID must be unique. An expression with an ID initial of
Q or A (query or action) is simply stored in a buffer. Other
expressions are compiled as condition expressions. A
condition expression (such as &1 < or = 0.002) uses postfix
notation and stack operations. A variable (&1) causes the
run-time module to input a real number and store it on a
stack. Therefore, these expressions can be compiled as a
simple code instead of using a simple constant or a variable
datum.
 91

FIGURE 21. Implementation Of A Decision-Tree In

A Program.
 92

the tree structure is represented by expression IDs and

pointers. For instance, an arrow (--->) represents "point
to." The syntax is given below:

Eno = root branch (source)

E ni = expression number (destination action)
A i = execution action
| = either E's or A's, but not both

During compilation, each is assigned an address in the tree

structure file. E n i's in parentheses are substituted by
pointers. Ai's are marked with negative values to indicate
their actions. Figure 22 shows how a decision tree can be
represented by DCTREE.
The system run-time module performs the I/O and decision
making.
 93

FIGURE 22. Input To DCTREE.

A decision tree used in the design of SORICH CAPP system is

illustrated in the Figure 23.
 94

FIGURE 23. SORICH Decision-Tree

(Product Oriented).
 95

5.3.3.2. Decision Tables

Decision tables have long been used to present complex

engineering data. Decision tables can also be easily
implemented on a computer. Using decision tables for process
planning, however, normally requires a special preprocessor
program or computer language to implement the table and
control the operation of the table. Such software is

generally called a decision-table language. A decision-table

language consists of:

a. A base language
b. A decision table
c. An outer language

A base language is the foundation of a decision-table

language. For example, FORTAB of the RAND Corporation and
S/360 DLT of the IBM Corporation use FORTRAN as their base

language (McDaniel, 1970). DETAB/65 (Silberg, 1971) of

SIGPLAN of the ACM (Association of Computing Machinery) uses

COBOL as its base language. A base language is extended to

include statements that can describe a decision table in a

more easily implemented manner. A preprocessor is
occasionally written to translate the decision-table
language program into its base-language program.
 96

The decision table is the most essential part of a decision-

table-language program. It is represented in its original
table format. Decision-table techniques can be used to
simplify and/or parse a complex table.

The third element (outer language) is used to control the

decision table. Decision table does not contain any
input/output or control statements; therefore, it is not a
complete program. An outer language can eliminate this void.

5.4. SORICH CAPP System Designed for Thesis

The CAPP system produced using the Quattro Pro software is a

cross between Retrieval Process Planning and Generative
Process Planning. It uses the concept from generative
process planning of asking questions of the user that help
describe the product to be manufactured. Upon answering
these questions a code number is generated. This code number
is then used to retrieve the appropriate files. This is
where the variant type concepts are used. If the system does
not find a file that matches exactly the code number which
was generated, it allows the user to retrieve the closest
code number file and make the necessary changes. The exact
 97

way in which the software performs these tasks will be

described in the section on software use.
 98

CHAPTER 6
 99

6. Group Technology in CAPP

Since the beginning of human culture, people have tried to

apply reason to their actions. One important way to apply
reason is to relate similar things. Biologists classify
items into genus and species. We relate to such things as
mammals, marsupials, batrachians, amphibians, fish,
mollusks, crustaceans, birds, reptiles, worms, insects, and
so on. A chicken is a bird with degenerated wings. Tigers,
jaguars, and domestic cats are all members of a single
family.

The same concept applied to natural phenomena can also be

applied to fabrication and information phenomena. When a
vast amount of information has to be kept and ordered, a
taxonomy is normally employed. Librarians use taxonomies to
classify books in libraries. Similarly, in manufacturing,
thousands of items are produced yearly. When one looks at
the parts that construct the product, the number is
exceptionally large. Each part has a different shape, size,
and function. However, when one looks closely, one may again
find similarities among components in Figure 24, a dowel and
a small shaft may be very similar in appearance but
different in function. Spur gears of different sizes need
the same manufacturing processes and vary only in size.
 100

Therefore, it appears that manufactured components can be

classified into families similarly to biological families or

FIGURE 24. Design Family.

 101

library taxonomies. Parts classified into families similarly

to biological families or library taxonomies. Parts
classified and grouped into families produce a much more
tractable database for management.

Group Technology (GT) is a manufacturing philosophy in which

similar parts are identified and grouped together to take
advantage of their similarities in manufacturing and design.
Although this simple concept has been in existence for a
long time, it was not until 1958 that S. P. Mitrofanov, a
Russian engineer, formalized the concept in his book "The
Scientific Principles of Group Technology". Group Technology
(GT) has been defined (Solaja, 1973) as follows:

"Group Technology is the realization that many problems are

similar, and that by grouping similar problems, a single
solution can be found to a set of problems thus saving and
effort."

Although the definition is quite broad, one usually relates

group technology only to production applications. In
production systems, group technology can be applied in
different areas. For component design, it is clear that many
components have a similar shape (Figure 24). Similar
components, therefore, can be grouped into design families
and a new design can be created by simply modifying an
existing component design from the same family. By using
 102

this concept, composite components can be identified.

Composite components are parts that embody all the features
of a design family or design subfamily. An example is shown
in Figure 25. Components in the family can be identified
from features of the composite components.

FIGURE 25. Composite Components.

 103

For manufacturing purposes, GT represents a greater

importance than simply a design philosophy. Components that
are not similar in shape may still require similar
manufacturing processes. For example, in Figure 10.3, most
components have different shapes and functions, but all
require internal boring, face milling, hole drilling, and so
on. Therefore, it can be concluded that the components in
the figure are similar. The set of similar components can be
called a production family. From this, process-planning work
can be facilitated. Since similar processes are required for
all family members, a machine cell can be built to
manufacture the family. This makes production planning and
control much easier, since only similar components are
considered for each cell. Such a cell oriented layout is
called a group-technology layout or cellular layout.

6.1. Part-Family Formation

One of the major benefits derived from GT applications is

part family formation for efficient work flow. Efficient
work flow can result from grouping machines logically so
that material handling and setup can be minimized. Parts can
frequently be grouped so that the same tooling and fixtures
can be used. When this occurs, a major reduction in setup
 104

FIGURE 26. Two Parts Of Identified Shape And Size But

Different Manufacturing Requirements.

FIGURE 27. Thirteen Parts With Similar Manufacturing Process

Requirements But Different Design Attributes.
 105

results. Machines can also be grouped so that the amount of

handling between machining operations also can be minimized.
A part family is a collection of parts which are similar
either because of geometric shape and size or because
similar processing steps are required in their
manufacturing. The parts within a family are different, but
their similarities are close enough to merit their
identification as members of the part family. Figure 26 and
27 show two part families. The two parts shown in Figure 26
are similar from a design viewpoint but quite different in
terms of manufacturing. The parts shown in Figure 27 might
constitute a part family in manufacturing, but their
geometry characteristic do not permit them to be grouped as
a design part family.

The part family concept is central to design retrieval

systems and most current computer-aided process planning
schemes. Another important manufacturing advantage derived
from grouping workparts into families can be explained with
reference to Figures 28 and 29. Figure 28 shows a process-
type (functional) layout for batch production in a machine
shop. It is one of the most common machine layouts used in
industry As can be seen in the figure, machines are laid out
with respect to type. Mills, lathes, drills, and grinders
are clustered so that they reside in a single department.
This layout is efficient in the allocation of tooling and
fixturing within departments. It also allows a supervisor to
 106

FIGURE 28. Process-Type Layout.

FIGURE 29. Group Technology Layout.

 107

accumulate a significant amount of information concerning a

limited set of operations within the department.
Unfortunately, functional layout also requires that the
product flows throughout the entire system in a somewhat
random manner (as can be seen in the figure). Material
handling can be significantly reduced if parts requiring
similar operations are clustered into a "part family", and
the machines required to produce the part family are
organized into a cell. This process is known as cellular
layout and is illustrated in Figure 29. As can be seen from
the figure, the flow of product through the system is much
more direct and material handling can be reduced
significantly.

The basis of cellular, or GT layout is part-family

formation. Family formation is based on parts production or
more specifically, their manufacturing features. Components
requiring similar processes are grouped into the same
family.

There is no rigid rule that can be applied to form families.

Users must set their own definitions of what a family is or
should be. A general rule in forming the family is that all
the parts must be related. For production-flow analysis, all
parts in a family must require similar routings. A user may
want to put only those parts having exactly the same routing
sequence into a family.


108

The biggest single obstacle in changing over to group

technology from a traditional production shop is the problem
of grouping parts into families. Before grouping can start,
information concerning the design and processing of all the
existing components must be collected from existing part and
processing files. There are three general methods for
solving this problem. All three methods are time consuming
and involve the analysis of much data by properly trained
personnel. The three methods are:

6.2. Visual Methods

6.3. Production Flow Analysis (PFA)

6.4. Parts Classification and Coding System

6.2. Visual Methods

The visual inspection method is the least sophisticated and

least expensive method. It involves the classification of
parts into families by looking at either the physical parts
or photographs and arranging them into similar groupings. It
is obvious that when we have many components, visual
inspection will be difficult to use. This method is
generally considered to be the least accurate of the three.
 109

TABLE 4. PARTIAL COMPONENT PROCESS SUMMARY

Component Code Processes OP Code Sequence

A-112 1110 SAW01, LATHE02, GRIND05, INSPO6

A-115 6514 MILL02, DRL01, INSPO3
A-120 2110 SAW01, LATHE02, GRIND05, INSPO6
A-123 2010 SAWOI, LATHE01, INSPO6
A-131 2110 SAWOI, LATHE02, INSPO6
A-212 7605 MILL05, INSPO3
A-230 6604 MILL05, INSPO3
A-432 2120 SAWOI, LATHE02, INSPO6
A-451 2130 SAWOI, LATHE02, INSPO6
A-510 7654 MILL05, DRL01, GRIND06, INSPO6
A-511
A-S 11
A-512
A-550
A-556
B-105
B-107
B-108
8-109
B-115
B-116
B-117
B-118
B-119
B-120

6.3. Production Flow Analysis (PFA)

For example, if we have a total of 24 components in our

minishop. After coding them, we can obtain a summary in
Table 4. We can easily group parts into families with the


1 10

help of computerized methods such as production-flow

analysis.

N N 0
r:
MM
N N
- 7 ¢ « a7 ¢
7 7 %;

. _
SAW01 / / / /
LATHE01 /
LATHE02 / / /
DRLO1

MILL02 /
MILL05 / /
GRINDO5

GRIND06 / x /.
INSPO3 / / /
INSPO6 / / / / _ / / /

FIGURE 30. PFA Matrix.

Production Flow Analysis (PFA) was introduced by J. L.

Burbidge to solve the family-formation problem for
manufacturing cell design (Burbidge, 1971, 1975). Many
 111

researchers have subsequently developed algorithms to solve

the problem. In PFA, a large matrix (incidence matrix) is
constructed. Each row represents an OP code, and each column
in the matrix represents a component, Figure 30. We can
define the matrix as Mij, where i designates the OP codes,
and j designates components (Mij = 1 if component j has OP
code i; otherwise, Mij = 0). The objective of PFA is to
bring together those components that need the same or a
similar set of OP codes in clusters.

6.3.1, Rank Order Cluster Algorithm

J. R. King (1979) presented a rank-order cluster algorithm

that is quite simple. We use his method to show how
component families can be determined in our shop. King's
algorithm can be stated as follows:

Step 1. Calculate total weight of each column as:

wi = -- 2 i M..3

Step 2. If wi is in ascending order, go to step 3, or

rearrange in ascending order.
 112

Spep 3. Calculate total weight of each row as:

wi = — 2j Mii

Step 4. If in ascending order stop, else rearrange and

return to step 1.

The rank-order clustering algorithm sorts the matrix into a

diagonal block structure. The diagonal blocks are not always
mutually exclusive. Final judgement has to be made by the
user.

One of the major drawbacks of applying this algorithm is the

need of storing the binary word. The word length is max(n,
m), where n is the number of machines, and m is the number
of components. For a moderate problem with 50 machines and
2000 components, it is impossible to calculate the weights
before sorting. For the machine weights, 50 words of 2000
bits are needed. A word of 2000 bits requires 250 bytes. To
overcome this problem, direct comparison of elements, either
row or column, can be used. A digit-by-digit comparison is
performed, beginning from the most significant digit. Each
row or column of the matrix is treated as a binary number;
no weight is ever calculated. Unfortunately, this procedure
has a computational complexity of a cubic order, namely,
 113

FIGURE 31. Rank-Order Cluster Algorithm.

 114

O[ij(i+j)] (King and Nakornchai, 1982), where i and j are

the number of rows and columns, respectively.

Figure 31 shows the procedure of rearranging the PFA matrix

in Figure 30. Note that the last two rows in Table 3 were
not used in Figure 31 in order to simplify the example.

After we obtain the final matrix, we can determine

(arbitrarily) that components Al23, Al20, A131, A432, A451,
and A112 form a family that needs SAW01, LATHE01, LATHE02,
and GRIND05. A115, A212, A230, and A510 form the second
family.

6.4. Coding and Classification

Parts classification and coding systems are divided into one

of the following four general categories:

a. CAD--Systems based on part design attributes

b. CAM--Systems based on part production attributes

C. CAB--Systems based on part business attributes

d. Combinations

CAD/CAM--Systems based on both design and production

attributes

CIM—Systems based on business, design and production

attributes

Coding is a process of establishing symbols to be used for

meaningful communication. Classification is a separate
process in which items are separated into groups based on
the existence or absence of characteristic attributes.
Coding can be used for classification purposes, and
classification requirements must be considered during the
construction of coding scheme. Therefore, coding and
classification are closely related.

Before a coding scheme can be constructed, a survey of all

component features must be completed and then code values
can be assigned to the features. The selection of relevant
features depends on the application of the coding scheme.
For example, tolerance is not important for design
retrieval, therefore, it is not feature in a design-oriented
coding system. However, in a manufacturing oriented coding
system, tolerance is indeed an important feature. Because
the code structure affects its length, the accessibility and
the expendability of a code (and the related database ) is
 116

FIGURE 32. Hierarchical Structure.

 117

of importance. There are three different types of code

structure in GT coding system:

6.4.1. Hierarchical

6.4.2. Chain (matrix)

6.4.3. Hybrid.

6.4.1. Hierarchical

A Hierarchical structure is also called a monocode. In a

monocode, each code number is qualified by the preceding

characters. For example, in Figure 32, fourth digit

indicates threaded or not threaded for a 322X family. One

advantage of a hierarchical structure is that it can

represent a large amount of information with very few code

positions. A drawback is the potential complexity of the

coding system. Hierarchical codes are difficult to develop

because of all the branches in the hierarchy that must be

defined.

A hierarchical type structure is being used in our custom

designed CAPP system. In this way we can represent more code
numbers with Quattro-Pro's limited eight digit file name
allowance.
 118

6.4.2. Chain (matrix)

A chain structure is called a polycode. Every digit in the

code position represents a distinct bit of information,
regardless of the previous digit.

TABLE 5. CHAIN STRUCTURE

Digit position I 2 3 4
Class of feature External shape Internal shape Holes
Possible value
1 Shape 1 Shape 1 Axial
2 Shape 2 Shape 2 Cross
3 Shape 3 Shape 3 Axial and
Cross
4

In Table 5, a chain-structured coding scheme is presented. A

2 in the third position always means a cross hole no matter
what numbers are given to positions 1 and 2. Chain codes are
compact and are much easier to construct and use. The major
drawback is that they cannot be as detailed as hierarchical
structures with the same number of coding digits.

6.4.3. Hybrid.

The third type of structure, the hybrid structure, is a

mixture of the hierarchical and chain structures Figure 33.

Most existing coding system use a hybrid structure to obtain

 119

FIGURE 33. Hybrid Structure.

the advantages of both structures. A good example is the

widely used Optiz (1970) code Figure 34. There are more than

100 GT coding systems used in industry today. The structure

selected is based primarily on the application.

The physical coding of a component can be shown best by

example. In Figure 35, a rotational component is coded using

the Optiz system. By going through each code position, the

resulting code becomes 01112. This code represents this

component and all others with similar shape and diameter. A

definition of the specific Optiz code attributes is given

later in Figure 36.

 120

FIGURE 34. Opitz Coding And Classification System.

 121

FIGURE 35. Rotational Component (H. Opitz, A Classification

System To Describe Workpieces).

6.5. Group-Technology Coding System

Both drafting and geometric modeling are detailed

representations of an engineering design. They provide
detailed information concerning the component to be made and
are essential in conveying the design for manufacturing.
However, as with many decision-making processes, too much
 122

information may make the decision more difficult. For

example, when reading a magazine or journal, one seldom
begins by reading each sentence on successive pages.
Instead, you peruse the table of contents first in order to
locate interesting technical information. By doing so,
candidate articles can be located more quickly. However, the
title may not convey all of the necessary information.
Therefore, the reader would typically scan the abstract. The
abstract is summary that represents the article without
great detail. Reading the abstract of an article normally
provides you with the insight to continue reading. Parts
lists corresponding to a table of contents have been around
for a long time, and although it is impractical to write
abstracts for CAD model, A similar concept can be applied
using coding.

Group Technology (GT) is an appropriate tool for this

purpose. Coding, a GT technique, can be used to model a
component without all the detail. When constructing a coding
system for a component's representation, there are several
factors to be considered. They include:

a. The population of components (i.e., rotational,

prismatic, deep drawn, sheet metal, etc.)

b. The detail the code should represent

 123

C. The code structure: chain, hierarchical, hybrid

d. The digital representation(i.e., binary, octal,

decimal, alphanumeric, hexadecimal, etc.)

The population of component contributes to the variety of

shapes. For example, the population in the United States

include virtually all races that exist on earth. In a sense,

it is necessary to distinguish race, hair color, eye color,
and so on. However, in a nation such as China or Japan, it
is not worthwhile to record skin color, hair color, and so
on, because these items are virtually invariant. In
component coding, it is also true that only those features
that vary have to be included in the code . When designing
or using a coding scheme, two properties must hold true: the
code must be:

a. Unambiguous and

b. Complete.

The code has to be concise. If a 100-digit code and a 10-

digit code can both represent components in population space
completely and unambiguously the 10-digit code is normally
necessary. For example, the basic Opitz code shown in Figure
32 uses five digits to describe the shape. Five digits can
represent 10 5 combinations. With this set, it is not
 124

possible to show a large amount of detail of a component.

Some codes are significantly longer, for example, the KK-3
of Japan (Japan Society, 1980), which has 21 digits and
contains multiple digits for single features, and MICLASS of
TNO (Houtzeel and Schilperoot, 1970), which has a 12-digit
code. For some computer-aided process planning systems
• (e.g., APPAS (Wysk, 19773), a detailed surface code is used
instead of a code for the entire component. The decision on
how much detail the code should represent depends solely on
the application. The selection of a code structure again
depends on the application.

The last consideration in coding-system construction is the

code digits. Several positional alternatives can be selected
(from binary to alphanumeric). However, this selection
yields different precisions for the different schemes. For
example, an N-digit code with different coding features
yields the following combinations of code:

Binary 2 (011)
Octal 8 (0,1,....,7)
Decimal 10 (0,1, ,9)
Hexadecimal 16 (0,1,...., 9,A,....,F)

Alphanumeric (26 + 10) (0,1,...., 9,A,.....2)

Although alphanumeric systems are the most compact (the same

amount of information can be represented by fewer digits),
 125

the difficulty of handling both numerical and alphabetical

characters makes alphanumerics less attractive.

6.5.1. The Opitz System

The Opitz coding system (Opitz, 1970) was developed by H.

Opitz of the Aachen Tech University in the West Germany. The
code uses a mixed code structure. However, except for the
first digit, it resembles a chain structure more closely.
Table 6. The Opitz code is a 9-digits code, it is probably
the best known coding system. First 5-digits describe the
geometry (form/shape) and rest 4-digits are supplementary
code. The geometric code can represent parts of the
following variety:

• Rotational

• Flat

• Long

• Cubic

A dimension ratio is further used in classifying the

geometry:
TABLE 6. FORM CODE (DIGITS 1-5) FOR ROTATIONAL PARTS IN THE
OPITZ SYSTEM. PART CLASSES 0, 1, AND 2
TABLE 6 (Continued).
TABLE 6 (Continued).
 129

• Length/Diameter ratio is used to classify rotational

components

• Length/Width and Length/Height ratios are used to

classify non-rotational components.

The Opitz geometric code uses 5-digits, representing:

a. Component Class

b. Basic Shape

C. Rotational-Surface Machining

d. Plane-Surface Machining

e. Auxiliary Holes, Gear Teeth, and Forming

Primary, secondary, and auxiliary shapes can be represented

using the five geometric digits.

A supplemental code containing four digits is usually

appended to the Opitz code. The first digit represents the
major dimension (either diameter or edge length). The
approximate component size can then be determined by using
the dimension ratio specified in the geometry. The dimension
range is specified from 0.8 to 80 in. Dimensions of less
 130

than 0.8 in. and greater than 80 in. are represented by a 0

or 9 code, respectively. The material type, raw material
shape, and accuracy are represented by digits 2,3, and 4.
The Opitz code is concise and easy to use. It has been
adopted by many companies as their coding subsystem. Several
CAM-1 CAPP systems currently use an Opitz-based coding
system.

6.5.2. The Vuoso-Praha System

The Vuoso-Praha code is a 4-digit coding system that

characterizes a part by Kind, Class, Group, and Material.
The code is illustrated in Table 7.

This type of coding is typically used for rough part

classification so as to identify the type of department that
would produce the part. For example, the part shown in
Figure 33 would be classified as

3 Rotational workpiece, with through hole

3 D = 51.2 mm; L/D = 0.6

0 Smooth

7 Material not specified

 131

TABLE 7 . THE VUOSO-PRAHA CODE

 132

6.5.3. The KK-3 System

The KK-3 coding system is a general-purpose classification

and coding system for machining parts. KK-3 was developed by.

the Japan Society for the Promotion of Machine Industry

(JSPMI) (Japan Society, 1980). Parts to be classified are

primarily metal-cutting and grinding components. KK-3 was

first presented in 1976, and uses a 21-digit decimal system.

The code structure for rotational components is shown in

Tables 8 and 9.

TABLE 8. STRUCTURE OF THE KK-3 CODING SYSTEM (ROTATIONAL

COMPONENTS)

Digit Items (Rotational components)

1 Parts General classification

name
2 Detail classification

3 General classification
Materials
4 Detail classification

5 Chief Length
dimensions
6 Diameter

7 Primary shapes and ratio of major dimensions

8 External surface and outer primary shape

9 Concentric screw threaded parts

10 External Functional cut-off parts

- 12
surface
11 a) Extraordinary shaped parts

12 `6. Forming
ii
13 -13 Cylindrical surface
c
14 "I Internal primary shape
-c
c Internal
15 0 Internal curved surface
t
. surface
16 . Internal flat surface and cylindrical surface
- o
17 .. surface
cn
18 Regularly located holes
Nonconcentric
19 holes Special holes

20 Noncutting process

21 Accuracy
 40
'-■,........„11„,........., IIZ
4 5 6 7 8 9
"----..----
0 1 2 3
I I

Spur,
Internal Bevel Hypo id Worm Screw Sprocket Special Round 0
0 Gears helical Other(s) Gears
gear (s) earls) gear(s) gear(s) gear(s) wheel gear vessel 14-3
gear(s)
0
Spindle,
e' Shafts, Counter Lead Screwed Round Eccentric Splined Cross Round Round Shafts,
1 t arbor,
f. spindles shaft screw (s) shaft rods) shaft (s) shaft shaft column casting spindles
C mai n
drive t4
a. - 1
E
2 8 Main
Pulley(s) Clutch Brake(s) Impellerls) Piston(s)
Round
Other(s) Flange Chuck (s)
Labyrinth Main tri
drive tables seal(s) drive
T. H
c
- .o_ 4
z
ro Guiding Sleeves, Bearing Dial Index Guiding
3 t' Bearing(s) Rollerts) Cylinder Other(s) Cams) Others
cc parts bushing metal plate(s) plates} parts
1 -3
1-1
O
Fixing Socket, Fastening Round Fixing
4 Cohorts) Pin(s) Others) Handles Spool (s) Screws) Others
part spacer screws links part
O
rJ
5 1-3

tri
T.
6 o
c
..7.
- ,
0
• C
0 0
. z
O
1-1
9
G)
ti)
to
'-3
 134

Because KK-3 is much greater in length than Opitz and CODE,

more information can be represented.

Code digit Item Component condition Code

1 1 Control valve 0
Name
2 (others) 9
3 Copper bar 7
1 Material
4
5 Dimension length 80 mm 2
6 Dimension diameter 60 mm 2
7 Primary shape and ratio
of chief dimension VD 1.3 2
8 External surface With functional tapered
surface 3
9 Concentric screw None 0
10 Functional cutoff None 0
11 Extraordinary shaped None 0
12 Forming None 0
13 Cylindrical surface ≥ 3 None 0
14 Internal primary Piercing hole with dia.
variation, NO cutoff 2
15 Internal curved surface None 0
16 Internal flat surface None 0
17 End surface Flat 0
18 Regularly located hole Holes located on
circumferential line 3
19 Special hole None 0
20 Noncutting process None 0
21 Accuracy Grinding process on
external surface 4

FIGURE 36. Example Of A KK-3 Coding System.

 135

The KK-3 code includes two digits for the component-name

(functional-name) classification. The first digit classifies
the general function, such as gears, shafts, drive, and
moving parts, and fixing parts. The second digit describes
more detailed functions. For example, included in a single
family, there are spur gears, bevel gears, worm gears, and
so on. With two digits, KK-3 can classify 100 functional
names for rotational and nonrotational components. However,
at times, this can be as confusing as it is complete. KK-3
also classifies materials using two-code digits. The first
digit classifies material type and the second digit
classifies shape of the raw material. Dimensions and
dimension ratios are also classified. Some redundancy can be
found, that is, length, diameter, and length/diameter ratios
are classified in KK-3 using 13 digits of code (much more
detail than either the Opitz or CODE systems). An example of
coding a component using KK-3 is illustrated in Figure 36.

6.5.4. The CODE System

The Code system is a parts classification and coding system

developed and marketed by Manufacturing Data Systems, Inc.
(MDSI), of Ann Arbor, Michigan. Its most universal
application is in design engineering for retrieval of part
 136

design data, but it also has applications in manufacturing

process planning, purchasing, tool design, inventory

control.

The CODE system is a Hexadecimal (8-digits) coding scheme.

For each digit there are 16 possible values (0,1,....,
9,A,....,F) which are used to describe the part's design and
manufacturing characteristics. The initial digit position
indicates the basic geometry of the part and is called the

Major Division of the CODE system. This digit would be used

to specify whether the shape was a cylinder, flat piece,

block, or other. The interpretation of the remaining seven

digits depends on the value of the first digit, but these
remaining digits form a chain-type structure. Hence the CODE

system possesses a hybrid structure.

The second and third digits provide additional information

concerning the basic geometry and principal manufacturing
process for the part. Digits 4,5, and 6 specify secondary

manufacturing processes such as threads, grooves, slots, and

so forth. Digits 7 and 8 are used to indicate the overall

size of the part (e.g., diameter and length of the turned

part) by classifying it into one of 16 size ranges for each
of two dimensions. Figure 37 shows a portion of the

definitions for digits 2 through 8, given that the part has

already been classified as a cylindrical geometry (Major
Division 1 for concentric parts other than profiled).
 137

FIGURE 37. A Portion Of The CODE System Of MDSI.

 138

6.5.5. DCLASS Systems

FIGURE 38. DCLASS Structure.

 139

The DCLASS system was developed by Del Allen at Brigham

Young University (Allen and Smith, 1980). DCLASS was

intended to be decision-making and classification system

(thus the name DCLASS). It is a tree-structured system, as

it can be seen in Figure 38, it can generate codes for

components, materials, processes, machines, and tools. For

components, an 8-digit code is used:

Digits 1-3 Basic Shape

Digit 4 Form Feature
Digit 5 Size
Digit 6 Precision
Digits 7 & 8 Material

In DCLASS, each branch represents a condition, and a code

can be found at the terminal (junction) of each branch.

Multiple passes of the decision tree allow a complete code

to be found. Figure 38 illustrates the process by showing a

portion of the decision tree for components. One pass on

this tree generates the first three digits of the code. Code
construction is established using certain roots. (N nodes

and E nodes). At N nodes, all branches can be true. At E

nodes, only one branch can be true.

 140

6.5.6. The MICLASS System

MICLASS stands for Metal Institute Classification System and

was developed by TNO, the Netherlands Organization for
Applied Scientific Research. It was started in Europe about
five years before introduced in United States in 1974.
Today, it is marketed in the United States by the

Organization for Industrial Research in Waltham,

Massachusetts. The MICLASS system was developed to help

automate and standardize a number of design, production, and

management functions. These include:

• Standardization of engineering drawings

• Retrieval of drawings according to classification
number
• Standardization of process routing
• Automated process planning
• Selection of parts for processing on particular groups

of machine tools
• Machine tool investment analysis

The MICLASS classification number can range from 12 to 30

digits. The first 12 digits are a universal code that can be
applied to any part. Up to 18 additional digits can be used
to code data that are specified to the particular company or
industry. For example, lot size, piece time, cost data, and
 141

operation sequence might be included in the 18 supplementary

digits.

The workpart attributes coded in the first 12 digits of the

MICLASS number are as follows:

Digit 1 Main shape

Digits 2 & 3 Shape elements
Digit 4 Position of shape element
Digits 5 & 6 Main dimensions
Digit 7 Dimension ratio
Digit 8 Auxiliary dimension
Digits 9 & 10 Tolerance code
Digits 11 & 12 Material codes

One of the unique features of the MICLASS system is that

parts can be coded using a computer interactively. To

classify a given part design, the user responds to a series
of questions asked by the computer. The number of questions
depends on the complexity of the part. For a simple part, as
few as seven questions are needed to classify the part. For
an average part, the number of questions ranges between 10
and 20. On the basis of the responses to its questions, the

computer assigns a code number to the part. Because the

system developer, TNO, is an international organization, the

program was written to converse in any of four languages:

English --- French - German - Dutch
 142

TORE 39 . Workpart .
FIGURE
 143

FIGURE 39 (Continued). Computerized MICLASS System.

Determination Of Code Number For Workpart.

 144

Also it can operate in either inches or metric, or both.

MICLASS is also one of the earliest interactive coding
systems, Figure 39 illustrates an interactive coding
session. MICLASS has been adapted by many U.S. industries.
Several application programs based on MICLASS are currently
available, such as the MULTIPLAN and MULTICAPP variant
process-planning systems.

6.6. Benefits Of Group Technology

Broadly defined, Group Technology (GT) is the application of

knowledge about groups. More specially, GT implies the
notion of recognizing and exploring and exploiting
similarities in three distinct way:

a. By performing like activities together

b. By standardizing similar tasks

c. By efficiently storing and retrieving information about

recurring problems.
 145

It should be regarded not as a tool or an application but as

a management philosophy. In particular, GT is not a
classification and coding system, nor is it equivalent to
cellular manufacturing; the first is a tool to help
implement GT, the second is an application of GT.

Although group technology is expected to be an important

principle in future production plants, it has not yet
achieved the widespread application which might be expected.
While there is currently a high level of interesting GT,
fairly little is known about its uses. There are several
reasons for this. First, there is a problem of rearranging
the machines in the plant into GT cells. Many companies have
been inhibited from adopting GT because of the expense and
disruption associated with the transition to GT machines
cells. Second, there is the problem of identifying part
families among the many components produced in the plant.
Usually associated with this problem is the expense of parts
classification and coding. Not only this procedure is
expensive, but it also requires a considerable investment in

time and personnel resource. Managers often feel that these

limited resources can be better be allocated to other
projects than group technology with its uncertain future

benefits. Finally, it is common for companies to encounter a

general resistance among its operating personnel when
changeover to a new system is contemplated.
 146

When these problems are solved and group technology is

applied, the company will typically realize benefits in the
following areas:

* Product design

* Tooling and setups

Materials handling

Production and inventory control

Employee satisfaction

Process planing procedures

6.7. Machine Selection

A common characteristic of U.S. industry is underutilization

of expensive processing equipment. The underutilization can

take two forms:

a. Much of the machine time is idle and totally

unproductive;
 147

b. Many of the parts assigned to a specific machine are

far below the capacity of the machine.

FIGURE 40. Comparison (Typical) Of A Turned-Part Dimension

As A Function Of Machine Capacity.
 148

FIGURE 41. Comparison Of Maximum Speeds And Feeds (Machine

Capacity) With Maximum Used.

 149

By grouping closely matched parts into a part family,

machines can be more fully utilized from both a scheduling
as well as a capacity standpoint.

By using a part coding system, similar parts with similar

feature dimension specifications can be assigned to the same
part family , and machines corresponding to the minimum
product specification can be selected rather than

overspecifying (and overpaying) the processing. Figure 40

illustrates this phenomenon for lathe parts. In the figure,
it can be seen that only few percent of the parts being

machined require the full lathe swing or length.

Furthermore, the speed and feed capacity of the lathe can
also be over-specified. See Figure 41. Since additional
features on machine tools require additional cost, be
careful not to over-specify a machine and not to use the
features.
 150

CHAPTER 7
 151

7. CAPP Using Artificial Intelligence

Artificial intelligence (AI), is a new and rapidly

developing field of computer science. It deals with
computers performing human like functions such as reasoning
and interpretation. AI may also take the form of simulating
the human senses of vision, hearing, touch, and even smell.

These senses usually serve as a source of information for

the human brain to use in making observations, drawing
conclusions, problem solving, natural language processing,
and formulating decisions. Computer systems are being
developed which operate with the same basic approach to
problem solving, rather than the fixed, sequential,
programmed logic of conventional computers.

Recent advances in the field of AI demonstrate that

practical AI based systems are possible. AI has achieved

considerable success in the development of expert systems

since the mid-1960s. Edward Feigenbaum of Stanford

University, one of the leading researchers in expert

systems, has defined an expert system as:

" an intelligent computer program that uses knowledge and

inference procedures to solve problems that are difficult

enough to require significant human expertise for their

solution. Knowledge necessary to perform at such a level,

 152

plus the inference procedures used, can be thought of as a

model of the expertise of the best practitioners of the
field".

This area of AI has concentrated on the construction of

high-performance programs in specialized professional
domains--a pursuit that has encouraged an emphasis on the
knowledge that underlies human expertise and has
simultaneously decreased the apparent significance of
domain-independent problem-solving theory. A new set of
principles, tools, and techniques has emerged that forms the
basis of knowledge engineering, which is a new term adopted
by the researchers in this field in order to describe this
new discipline.

7.1. Al and Process Planning

Today, with rapidly diminishing number of experienced

process planners in industry, there is an urgent need to
automate the process-planning function. Many computer-aided
process planning (CAPP) systems have been developed as a
result of this urgency. The complexity, in addition to the
variety of tasks in preparing a process plan, requires a
significant amount of time from an experienced process
planner in almost all existing CAPP systems.
 153

The successful application of artificial intelligence in

many science and engineering areas reveals that AI is
applicable to the process planning and benefits have been
reported.

A productive CAPP system must contain a tremendous amount of

knowledge--facts about the machine shop and rules about
arranging machine operations. Furthermore, the system should
be flexible because facts and rules in the database require
constant updating. This is especially true in today's
manufacturing environment. In a traditional CAPP system,
manufacturing knowledge is coded line by line in the
program's statements. Any modification to the facts and
rules require rewriting the original program. In other
words, a traditional CAPP program cannot learn new knowledge
unless it is explicitly rewritten. The inflexibility of
traditional methodoligies has retarded the implementation of
CAPP systems, which are the most important factors in
CAD/CAM linkages.

An expert system stores knowledge in a special manner so

that it is possible to allow pieces of knowledge to be
added, delete, and modify from the system without having to
modify other segments of the system. Such a system would
allow initial knowledge to be encoded into the system, and
once the system was operating, it could be augmented by
additional information. A list of operations could be
 154

created for the surface to be machined. The prerequisites

and restrictions of each process must be considered before
it is performed. A software control structure must search
through the knowledge base until a satisfactory process is
identified. The system must interact with the user until it
has determined the process to produce every surface.

Wolfe and Kung describe the development of a generative

process plan using an expert system methodology. The system
aims to extract form features of a part from a CAD System.
These features, along with details about tolerance and
surface finish, are placed in a data base which is used by
expert system to generate a process plan. Perhaps the
earliest plan generator to be developed using AI techniques
was GARI. GARI showed that as the number of production rules
increased, the possibility of conflicts between them also
increased. This would require appropriate conflict breaking
methods to be enunciated. In addition, it is possible that a

rule may also have an exception which may lead to delays due
to tedious conflict resolution.

Hummel developed an expert system to generate process plans

to machine a part in one fixturing setup at a machine tool.
The use of localized process planning capabilities for each
control module in a distributed computing atmosphere is
recommended. This would require a hierarchical network of
independent modules whose planning spectrum would reflect
 155

the span of the control at that level in the hierarchy. Each

module would have a local knowledge base and a set of
requirements that would be machined with the capabilities of
an individual machining tool. It is imperative that each
module in the hierarchy be able to access the knowledge
bases of other modules in the system. The machine tool
planner developed could work as a segment of a larger
system, responsible for generating machining instructions
required to process a part in a single fixturing setup at a
machine tool, or as a stand alone system, dedicated to a
single machine.
 156

CHAPTER 8
 157

8. Process Planning Systems -A Survey

There exist today both variant and generative process plans.

Process planning systems have been historically been
developed by a single organization or in an academic
atmosphere. Following is the list of some of the CAPP
systems that have been described in the open literature. The
list is not intended to be exhaustive, nor are the
attributes intended to be critical judgement.

8.1. APPAS

Generative system, exclusively for parts produced on

machining centers. Uses a special code to provide technical
data and information describing each surface. Complex
decision tree structure used.

8.2. AUTOPLAN

Generative system, provides additional details such as

jig and fixture selection, and material and stock
determination. Does not claim to be CAPP system, instead a
system to aid process planner. Uses graphical capability.
 158

8.3. CAM-I CAPP

A database management system with appropriate retrieval

logic. Interactive with user specified GT coding technique.

8.4. ACAPS

A semi-generative system that allows interactive coding

of surfaces. Considers available capacity and suggests a
possible machining sequence. Can handle holes, slots, plane
and turned surfaces.

V.V. ICAPP

Developed for milling, drilling, and boring, the

sequence of operations is decided by the part's features,
material, dimensions and tolerances. Calculates speeds,
operation and production times.

8.6. AUTOCAPP
An interactive system for process planning of turned
parts, output includes speeds, feeds, number of cuts,
operations, handling and machining time. Planner has to
 159

indicate surfaces to be machined, accuracy and surface

finish.

8.7. MIPLAN

A special purpose, interactive word processor that

allows a planner to construct and edit process plans from
standard plans. New plans can be created, stored plans
retrieved and edited, or an incomplete plan completed.

8.8. GARI

An expert system, it consists of a knowledge base and a

general purpose problem solver. Knowledge stored as
production rules. Successive iterations used to generate
plan.

8.9. GENPLAN

A Generative system, capable of instantly creating work

instructions for the manufacturing of aircraft parts and
subassemblies. Analyzes a part geometry, material, and
specifies tools and processes needed.
 160

8.10. SHAPES

Part profiles are entered in 'SHAPE' elements. The data

on the material and component size are used to determine the
machine tool, and collets, chucks, and centers to hold
workpiece.

8.11. TIPPS

Generative system, that addresses model decomposition

and shape identification. The facility's capability and
required processes are matched to devise a process plan. A

combination of AI and decision tree approaches are used.

Inputs are in the form of CAD boundary representation.

8.12. TOJICAPP

System for rotational parts. Combines generative and

variant planning. Can be used to generate a new plan
interactively. Outputs include cutting parameters, time
estimates, production times, and cost estimates.
 161

8.13. MICRO-CAPP

Interactive variant system, system commands are all two

letter neumonics. Coding and part classification is through
GT.

8.14. GEOMETRIC PROGRAMMING

Techniques used to select optimal machining parameters

for a wide variety of operations. Inputs required-speed,
machine tool, feed, and HP. Can optimize cost or time. Does
not determine operation sequence or the machines to be used.

8.15. CAPP and Dynamic Process Selection Techniques

The primary objective of most process planning systems has

been to develop a sequence of operations based upon
technical and economical criteria. This does not consider
the dynamic conditions of the machine shop. A versatile,

flexible manufacturing system would require a CAPP that was

dynamic in nature. The system must be able to generate

alternative, feasible plans. Dynamic process selection would
involve the capacity to decide which process would be used
to manufacture a part, taking into consideration current
 162

facility status. This assumes the alternative processes

exist to machine a part. This means that a part that could
be processed on a particular class of machines could also be
processed on a different class of machines using
significantly different manufacturing procedures.
Nof and Barash state that production rate and machine
utilization are two criteria to be monitored in a facility.
It was found that when dynamic allocation of either the
original plan or the alternative were undertaken, the
production rate and the machine utilization improved.
However, product quality may be dependent upon the
manufacturing process used. In a particular facility, since
most parts are not manufactured just once, but repetitively,
it is a sound decision to prepare alternate process plans
for each part.
 163

CHAPTER 9
 164

9. Description of SORICH CAPP System

The SORICH CAPP System is not a specialized software package

as the other CAPP systems available on the market. It is a
customized macro based program using the spreadsheet
software package Quattro Pro. It is a concept that is being
demonstrated based on the ideas of CAPP. This type of CAPP
system can be designed on any of the spreadsheet softwares
available on the market, such as but not limited to; Quattro
Pro, Lotus 1-2-3, V-P Planner and Excel. The system must be
designed around a particular company's manufacturing ideas
and philosophies.

9.1. Design of System

The SORICH CAPP system was designed for use with the
spreadsheet package Quattro Pro, but as indicated above it
can be adapted to any spreadsheet package. The system is
designed and written in the same way as any program would be
done, using the MACROS, Table 10, and by following logical
programming practices and the limitations and coding set out
by the software being used. SORICH was designed to be as
user friendly as possible. The need was for the user to be
 165

able to enter the desired information with limited room for

error. Several different attempts were made at reaching this
goal. There were no real step by step instructions to follow

and trial and error and continued searching through the

Quattro Pro manual was needed to complete this task.

Upon running the program, the first step to occur is the

removal of any existing information from a previous run,
this is performed automatically by the macro. The macro then
accesses custom menus that prompt the user to make
selections. The custom menus are designed by the programer,
when accessed they resemble a typical pull down menu similar
to those in most commercial softwares. The user can make his
selection from the menus and these selections then determine
the code number of the part to be produced. The macro will
then allow the user to import, view, alter, save and print
the Route Sheets and Bill of Materials for the particular

code number selected.

 166

TABLE 10. MACROS OF SORICH CAPP SYSTEM

MACRO {;THE FOLLOWING 4 COMMANDS}

CAPP- {;DELETE THE OLD INFORMATION}
{;TO ALLOW FOR NEW INFO}
{BLANK CHOICES}
{BLANK ROUTE }
{BLANK EOM}
BLANK DRAWING)
{}
{;THE FOLLOWING DISPLAYS A}
;MESSAGE UNTIL ENTER IS HIT)
;
:MESSAGE MSG1,25,10,0)
THE FOLLOWING 7 COMMANDS)
UF CUSTOM MENUS YOUTr:}
;CH'DICES DETERMINE THE CODE}
:=11,ER FOR THE NEW FART}
TYPT MENU:
IF CHOICE1< : . iBRANCH CAPP)
Y.EN . :CALL SHAPE ti
!

MENU}
FINIE?] MENU%
77 CHC':C 7 4=1)iY 7 NUCATL WOOD MENU)
IF CHOICE4=2;:MEN'jCALL COLOR MENU)
CHOICE4=3):MENUCALL METAL MENU1
ME]•::::CALL TOT MENU}
l'.F.E. !;UCALL CASTER MENU:-
,THE FC , L - C , WING 3
ONVEY-.7 USERS RESPONSES)
:"INTC- ' A VALUE;
/EVCO:E'CODE1 -
/EVCODE'C -
/EVCOI , E - COII , E3 -
%
THE FOLLOWING 3 COMMANDS)
HBRING OF CUSTOM MENUS)
MEN'jCALL IMPORT MENU)
0:ESSAGE MSG2,15,10,0)
M 7 NUCALL VIEW MENU;
;MENUCALL SAVE MENU)
;ME::UCALL F -1-',INT MENU}
-QUIT)
 167

TABLE 10. (Continued)

MSGI DURING EXECUTION OF CUSTOM

MENU'S YOU MUST HIGHLIGHT
YOUR CHOICE BY USE OF THE
ARROW KEYS AND HIT ENTER.
** HIT ENTER TO CONTINUE **
MSG2 UPON CHOOSING TO VIEW YOU WILL HAVE
FREEDOM TO MOVE THROUGHOUT THE
SPREADSHEET WITH THE ARROW KEYS AND
AND MAKE ANY NECESSARY CHANGES.
TYPE IN CHANGE AND HIT ARROW KEY
HIT ENTER WHEN ALL CHANGES ARE MADE.
" HIT ENTER TO CONTINUE **
INITATTON FOR THIS SELECTION IS
CN - .ER PROCESS, MAKE ANOTHER SELECTION.
*' HIT ENTER TO CONTINUE **
DC:WN THE CODE NUBER THAT
7S DIS:-ThAY N 7 X 7 . USr THIS CODE
N= ,7 T-, T TM -i'C)RT FILES . IF AN
E:XACT MATCH 1::(_- )ES NOT EXIST CHOOSE
TH: 1777 E WTTE THE CLOSEST NUMERIC
FILES EN2, :NG IN —1:" ARE B ,'
ROUTE
SHE:. AND F 71.;', T:NroN1 7, IN ,.
• HIT ENTER TO CONTINUE **
CO 7 Y DDW3'C THE CODE NUMEER THAT
- S DIS':-LAYED NEXT. USE THIS CODE
NjY 7- 7 R TO SAVr THE N 7 W FILE CREATED
EN -1 WITH "R' FOR ROUTE SHEET
OR END WITH "B" FOR BUM.
** HIT ENTER TO CONTINUE *,
T-- Nr , THE FYTOWING CODE NUMBER
W_TH ' R ' FfiR ROUTE SiirrT.
*" HIT ENTER TO CONTINUE **
YSGE END THE FOLLOW:NG CODE NUMBER
WITH "B - FOR EOM.
* HIT ENTER TO CONTINUE **
Y.S3; EN: , THE FOLLOWING CODE NUMBER
W7TH ";" FOR DRAWING.
*' H:T ENTER TO CONTINUE **
 168

9.2. Use Of The SORICH System

The following is a procedure for the operation of the SORICH

CAPP system. SORICH is a Quattro Pro program and must be
executed in a Quattro Pro environment. 5 1/4 disk that
accompanies with the THESIS, has the SORICH file.

Step 1. Start Quattro Pro and load SORICH by entering the

following commands:

a. / File Retrieve

b. Now type SORICH and hit enter

Step 2. Begin Macro execution by entering the following

commands:

a. / Tools Macro Execute

b. Now type SORICH and hit enter<

Step 3. The customized pulldown menus will be displayed in

the upper left hand corner of the screen.

Step 4. Use the arrow keys to highlight your choice.

 169

Step 5. A brief description is given in the lower left

hand corner of the screen while the choice is
highlighted.

Step 6. Use the enter key to make the proper selection.

Step 7. The first seven pulldown menus are used to make

selection which produces the code number.

Step 9. The following pulldown menus will allow the user

to import, view, alter, save and print the route

sheet and BOM
 170

CHAPTER 10
 171

10. CAPP - The Future

In 1984 it was forecasted that by 1990 over 50% of all

process plans produced would be developed by CAPP systems.

Although this seems that this ambitious estimate may not

come true, the importance of CAPP systems cannot be over

emphasized. The decreased in the availability of trained,

experienced personnel to serve as process planners

accentuates the importance of CAPP. The continuous increase

in labor cost, the never ending struggle to reduce

production cost, and increased productivity will force

manufacturing organizations to adopt CAPP systems.

Improvement in software, its transportable nature, and the

advent of powerful desktop computers will be an incentive

for small manufacturing organizations to adopt CAPP.

Increased efforts will be made to integrate CAD and CAM

through CAPP, and reduce the time for the job to proceed
from the drawing board to the production. This would cause
designers and process planners to work together, designing

parts with ease of manufacture as a primary consideration.

It would cause the development of a unified data base that

could be accessed by the process planner and the design

engineer. Most CAPP systems require the user to input the

specific GT code. This could be eliminated if the system
could automatically create and mark surfaces. Dynamic
 172

process selection systems that suggest process plans based

upon current facility load should be developed.

The recent spurt in the development and implementation of

knowledge based expert systems will certainly have its
effect on CAPP. The AI approach with its elegant data
representation and retrieval techniques, coupled with sound

decision logic programmed into system software should have

widespread applications in CAPP. Applications could include

processes such as metal forming, welding, casting, lesser

used processes such as broaching, and traditional CAPP

applications such as turning and drilling. A major hurdle

would be the gathering of reliable data in order to create

good, comprehensive knowledge bases.

CAPP systems have to expand their horizons to serve a

greater variety of manufacturing industries. Currently, CAPP
systems concentrate on machining processes, especially in

processes involving turning and hole generation. Recent CAPP

system applications have included operations such as cold
forging, and deep drawing. Increased use of CAPP systems

would make installation of CAPP systems more viable

financially, and help to increase productivity levels in all

manufacturing industries.

The SORICH CAPP System designed for the thesis is a product

oriented software, it is a concept that is being

 173

demonstrated based on the ideas of CAPP. This type of CAPP

system can be designed on any of the spreadsheet softwares
available on the market. The system must be designed around
a particular company's manufacturing ideas and philosophies.
 174

CHAPTER 11
 175

11 . Conclusion

The desire to improve productivity and decrease costs have

developed the need to reduce the time required by a part to
go from preliminary design to actual manufacture. CAPP will
reduce the time spent from product development to the
commencement of production in a product's life cycle.
Enhancements in available computing power coupled with
reductions in computer hardware costs have made it more
viable to use CAPP. Recent trends indicate that knowledge
based expert systems have greatly altered CAPP's scope. It
is forecasted that expert systems would be used to aid
process planning involving most manufacturing processes. A
hurdle to this would be the development of comprehensive
data bases for the applications under considerations.

The importance of CAPP, as necessary CAD/CAM interface, is

emerging. Researchers are striving to produce a generative
process planning system using a CAD input interface,
requiring with very little human intervention. In the
future, research will employ interdisciplinary approach
involving experienced process planners, design engineers,
and computer scientists as well as industrial engineers.
 176

BIBLIOGRAPHY
 177

1. Alan, D. K., 'Generative Process Planning Using The

DCLASS Information System', Monograph # 4, Computer Aided
Manufacturing Laboratory, Brigham Young University, Provo,
Utah, 1979.

2. Alan, D. K., & Smith, P. R., 'Computer-Aided Process

Planning', Computer Aided Manufacturing Laboratory, Brigham
University, Provo, Utah, 1980.

3. Amstead, B. H., Ostwald, Phillip F., & Begeman, Myron

L., 'Manufacturing Processes', John Wiley & Sons, New York,
1979.

4. Berra, P. B., 'Investigation of Automated Planning And

Optimization of Metal Working Processes', Doctoral

Dissertation, Purdue University, West Lafayette, Indiana,

1968.

5. Caldeira, E., 'Integration of CAPP And Computerized

Standard Data', Proceedings - Fall IIE Conference, 1985, PP,

106-109.

6. Chang, Tien-Chien, & Wysk, Richard A., Wang, Hsu Pin,

'Computer-Aided Manufacturing', Prentice Hall, New Jersey,
1991.
 178

7. Chang, Tien-Chien, 'Interfacing CAD and CAM - A Study

In Hole Design °, Master Thesis, Virginia Teach, Blacksburg,

Virginia, 1980.

8. Chang, Tien-Chien, Wysk, Richard A., & Davis, R. P.,

'Interfacing CAD and CAM - A Study in Hole Design',
Computing and Industrial Engineering, Vol. 6, # 2, 1982, PP.

91-102.

9. Chang, Tien-Chien, Wysk, Richard A., 'An Integrated

CAD/Automated Process Planning System', AIIE Transactions,

Vol. 13, # 3, 1981, PP. 223-233.

10. Chang, Tien-Chien, 'TIPPS-A Totally Integrated Process

Planning System', Doctoral Dissertation, Virginia Teach,

Blacksburg, Viginia, 1982.

11. Chang, Tien-Chien, 'Computer Aided Process Planning

Today and in the Future'. Proceedings - Fall IIE Conference,

Nov. 1983, PP. 349-355.

12. Chang, Tien-Chien, & Wysk, Richard A., 'Integrating CAD

and CAM through Automated Process Planning', International
Journal of Production Research, Vol. 22, # 5, 1984, PP. 877-
894.
 179

13. Chang, Tien-Chien, & Wysk, Richard A., 'An introduction

to automated Process Planning Systems', Prentice Hall, NJ,

1985.

14. Descotte, Y., & Latombe, J. C., 'GARI: A Problem Solver

that Plans How To Machine Mechanical Parts', Proceedings -

IJCAI 7, Vancover, Canada, PP. 766-772, Aug. 1981.

15. El-Gomayel, J., & Abou-Zeid, M. R., "Piece Part Coding

And The Optimization of Process Planning', Proceedings -
North American Metal Working Research Conference, May 1975.

16. El-Midany, T. T., & Davis, B. J., 'AUTOCAP - A dialog

System For Planning The Sequence Of Operations For Turning
Components', International Journal Of Machine Tool Research,

Vol. 21, # 3/4, 1981, PP. 175-191.

17. Emerson, C., & HAM, I., 'An Automated Coding And
Process Planning System Using A DEC PDP-10', Computing And
Industrial Engineering, Vol. 6, # 2, 1982, PP. 159-168.

18. Eshel, G., Barash, M., & Chang, Tien-Chien, 'A Rule

Based System For Automatic Generation Of Process Outlines

For Deep Processes', Proceedings - ASME Winter Annual

Meeting, Miami Beach, Florida, Nov. 1985, PP. 1-18.

 180

19. Eskiciouglu, H. & Davies, B. J., 'An Interactive

Process Planning System For Prismatic Parts', International
Journal Of Machine Tool Research, Vol. 21, # 3/4, 1981, PP.
193-206.

20. Ferreira, P., Puls, F., Chang, Tien-Chien, & Liu, C.

R., 'A CAD/CAM Systems For Complex Surfaces', Automated
Process Planning Systems, ASME Winter Annual Meeting, Miami
Beach, Florida, Nov. 1985, PP. 189-199.

21. Foster, Lowell W., 'GEO-METRICS II', Addison-Wesley

Publishing Company, Inc., California, 1986.

22. Foston, Arthur L., Smith, Carolena L., & Au Tony,

"Fundamentals of Computer Integrated Manufacturing',

Prentice Hall, New Jersey, 1991.

23. Groover, Mikell P. 'Automation, Production Systems and

Computer-Integrated Manufacturing', Prentice Hall, New

Jersey, 1987.

24. Groover, Mikell P., & Zimmers, Emory W. Jr., 'CAD/CAM,

Computer Aided Design and Manufacturing', Prentice Hall, New
Jersey, 1984.

25. Halevi, G., 'The Role of Computers in Manufacturing

Processes', John Wiley & Sons, New York. 1980.

 181

26. Ham, I., 'Group Technology', Handbook of Industrial

Engineering, Edited by G. Salvendy, John Wiley & Sons, New
York. 1982.

27. Hannam, R. G., & Plummer, J. C. S., 'Capturing

Production Engineering Practice within a CAD/CAM System',
International Journal of Production Research, Vol. 22, #. 2,
1984, PP. 267-280.

28. Hegland, D. E., 'Out in Front with CAD/CAM at Lockheed

- Georgia', Production Engineering, Vol. 28, # 11, 1981, PP.
45-48.

29. Hutzel, A., 'Integrating CAD/CAM through Group

Technology', Proceedings Of the Annual Meeting and Technical
Conference of The Numerical Control Society, Dallas, Texas,
May 1981, PP. 430-444.

30. Hsu-Pin Wang & Wysk, Richard A., 'Micro Computer Based
Process Planning Systems', Proceedings Fall IIE
Conference, December 1985, PP. 145-154.

31. Hummel, K. E., ' An Expert Machine Tool Planner',

Computers in Engineering, Vol. 2, 1985, PP. 367-373.
 182

32. ITT-Shop Practice Manual, P 9938, ITT Defence

Communications Division, New Jersey, 1968.

33. Link, C. H., 'CAPP - CAM - I Automated Process Planning

System', Proceedings - Annual Meeting & Technical Conference
of The Numerical Control Society, Cincinnati, Ohio, 1976,
PP. 401-408.

33. Meenakshi Sundaram, R., & Ta Jen Cheng, 'Micro Computer

Based Process Planning Using Geometric Programming',
International Journal of Production Research, Vol. 24, # 1,

1986, PP. 119-127.

34. Nau, D. S., & Chang, Tien-Chien, 'A Knowledge-Based

Approach To Generative Process Planning', Automated Process

Planning Systems, Proceedings ASME Winter Annual Meeting,

Miami Beach, Florida, November 1985, PP. 65-72.

35. Nof, S. Y., & Barash, M. M., 'Dynamic Process Selection

Procedures and Their Effect On Machine Configuration',
International Journal of Machine Tool Research, Vol. 20, PP.

137-146.

36. Opitz, H., 'A Classification Scheme To Describe Work

Pieces', Pergamon Press, Elmsfort, New York, 1970.

 183

37. Philips, R. H., 'A Computerized {Process Planning

System Based Upon Component Classification', Doctoral
Dissertation, Purdue University, West Lafayette, Indiana,
1978.

38. Powers, John H. Jr, 'Computer-Automated Manufacturing',

Mcgraw-Hill Book Company, New York, 1987.

39. Rebholz, M., 'Computer Aided Planning In Cold Forging',

CIRP Annual, Vol. 29, # 1, 1980, PP. 173-177.

40. Riggs, James L., 'Production Systems: Planning Analysis

and Control', John Wiley & Sons, New York, 1987.

41. Schaffer, G., 'GT Via Automated Process Planning',

American Machinist, May 1980, PP. 119-122.

42. Schaffer, G., 'Implementing CIM', American Machinist,

Special Report, # 736, August 1989, PP. 151-174.

43. Smith, R. N., 'Computer Aided Fully Generative Process

Planning', Manufacturing Engineering, May 1981, PP. 98-99.

44. Steudel, H. J., 'Computer Aided Process Planning, Past,

Present, Future', International Journal of Production
Research, Vol. 22, # 1984, PP. 253-266.
 184

45. Tirupatikumara, S., Joshi, S., Moodie, C. L., Kashyap,

R. L., & Chang, Tien-Chien, 'Expert Systems In Industrial

Engineering', International Journal of Production Research,

Vol. 24.

46. Tulkoff, J., 'Lockheed's GENPLAN', Proceedings of The

Annual Meeting and Technical Conference of The Numerical

Control Society, Dallas, Texas, May 1981, PP. 41; 7-421.

47. Vogel, S. A., 'Use Of GT In Process Planning',

Computers in Manufacturing Systems, ASME Planning, Control,
and Robots, Presented at The Winter Meeting of The ASME,

Chicago, II. 1980.

48. Vogel, S. A., & Adlard, E. J., 'The Autoplan Process

Planning System', Proceedings of The Annual Meeting and

Technical Conference of The Numerical Control Society, 1981,

PP. 442-429.

49. Weill, R., Spur, G., & Evershan, W., 'Survey of

Computer Aided Process Planning Systems', CIRP Annals, Vol.
31, # 2, 1982, PP. 539-551.

50. Wolfe, P. M., 'Automatic Process Planning Using

Artificial Intelligence', Proceedings-Annual IIE Conference,
May 1984, PP. 387-395.
 185

51. Wysk, Richard A., Barash, N. M., & Moodie, C. M., 'Unit
Machining Operations: An Automated Process Planning and

Selection Program', Journal of Engineering For Industry,

Vol. 102, November 1980, PP. 297-302.

52. Wysk, Richard A., 'An Automated Process Planning and

Selection Program: APAS', Doctoral Dissertation, Purdue
University, West Lafayette, Indiana 1977.

53. Zhang, S., Gao, W. D., 'TOJICAP-A System of Computer

Aided Process Planning System', CIRP Annals, Vol. 33, # 1

1984, PP. 299-301.