Download as pdf or txt
Download as pdf or txt
You are on page 1of 122

INTEGRATION OF PROCESS PLANNING AND

SCHEDULING FUNCTIONS
by
SRINIDHI MALLUR, B.E.
A THESIS
IN
INDUSTRIAL ENGINEERING
Submitted to the Graduate Faculty
of Texas Tech University in
Partial Fulfillment of
the Requirements for
the Degree of
MASTER OF SCIENCE
IN
INDUSTRIAL ENGINEERING

Approved

Accepted

May, 1992
ACKNOWLEDGEMENTS

I would like to express my grateful appreciation to several people for their

assistance in this research. Dr Hong-Chao Zhang, the chairperson of the committee,

provided me with excellent professional guidance.

My sincere thanks also goes to Dr. Jose Macedo, and Dr. Surya Liman, for their

support and suggestion during the course of the research work. Their comments and

criticism were very valuable for the research.

Finally, I would like to thank my parents, sister, and friends for their support,

and encouragement.
TABLE OF CONTENTS

ACKNOWLEDGMENTS 11

LIST OF TABLES Vl

LIST OF FIGURES Vll

CHAPTER

I. INTRODUCTION 1

II. PROCESS PLANNING AND SCHEDULING 5

2.1 Planning 5

2.1.1 Planning Defmed 5

2.1.2 Manufacturing Planning 6

2.2 Process Planning 7

2.2.1 Process Planning Defined 7.

2.2.2 Steps in Process Planning 11

2.2.3 Criteria Used in Process Planning 14

2.3 Computer-aided Process Planning 15

2.3.1 Databases in CAPP 18

2.4 Feature Based Process Planning 21

2.4.1 Feature System 22

2.4.1.1 Defmition of Feature 22

2.4.1.2 Types of Feature 22

lll
2.5 Models to Integrate Process Planning 28

2.6 Scheduling 34

III. LITERATURE REVIEW 42

3.1 Problem Definition 42

3.2 Literature Survey 46

IV. INTEGRATED PROCESS PLANNING MODEL 56

4.1 Objective 56

4.2 Generic Solution 57

4.3 Integrated Process Planning Model 59

4.3.1 Process Planning Module 64

4.3.2 Scheduling Module 73

4.3.3 Decision Making Module 74

4.4 Methodology 74

4.4.1 Rules for creating Setups 78

4.4.2 Heuristic Algorithm to Select Feasible Plans 80

4.4.2.1 Algorithm to Select Feasible Setups 83

4.4.2.2 Optimization Phase 85

4.4.2.3 Matrix Representation of the Problem 87

4.4.2.4 Heuristic to Assign the Setup to the Machine 88

v. SOLUTION METHODOLOGY 91

5.1 Generating Feature Relation Graph 91

5.2 Solution 92
lV
5.2.1 Initial Feasible Set 102

5.2.2 Optimization Algorithm 103

5.2.3 Matrix Solution 104

VI. CONCLUSION 106

BIBLIOGRAPHY 111

v
LIST OF TABLES

5.1 Resource Utilization 99

5.2 Relative Hamming Distance 100

5.3 Final Matrix 105

Vl
LIST OF FIGURES

2.1 Process Planning Overview 9

2.2 Generation of Process Plans 10

2.3 Expert Process Planner 20

2.4 Examples of Features 23

2.5 Form Feature Hierarchy 26

2.6 Feature Representation using Attributed Adjacency Graph 27

2.7 Non-Linear Process Planning 29

2.8 Closed Loop Process Planning 31

2.9 Distributed Process Planning 33

2.10 External Context of MRP II 36

2.11 External Context of OPT 38

2.12 Scheduling Database Files 40

4.1 Information Flow within a Factory 58

4.2 Functions Which Cell Controller Monitors 60

4.3 Command Execution Architecture 62

4.4 Integrated Planning Model 63

4.5 Integration between Process Planning and Scheduling 65

4.6 Scheme for Integration 66

Vll
4.7 Integration of Process Planning and Scheduling Functions 67

4.8 Feature Relation Graph 70

4.9 Alternative Process Plans versus Machines Matrix 75

4.10 Time Window 76

4.11 Flow Diagram of the Logic Used for Integration 90

5.1 Example Part 93

5.2 Possible Setups 94

Vlll
CHAPTER I

INTRODUCTION

Manufacturing function in today's environment is depending more and more on

the computerized technology. Many of the new developments like computer integrated

manufacturing, computer-aided process planning, computer-aided manufacturing, and

computer-aided desigr1 have totally depended on the development of the computer

technology. It has been seen that each one of the above stated fields have developed on

their own and have stayed very individualistic, using their own protocols and

technologies. Integrating these technologies has become a prime concern. Data from one

system cannot be easily ported to another system. Transfer algorithms have to be

developed so that the data can be used by some other system. All this is happening at

a much higher level of operation. But on the shop floor itself, technologies have

developed which are incompatible.

Any study into integration of functions has to address the above mentioned

scenariO. Functional integration can be achieved by some minor adjustments in the

functioning of the system itself. In this thesis, I have addressed the issue of "Integration

of Process Planning and Scheduling Functions." This topic receives a lot of attention due

to its strategic importance in the shop floor function.

Process Planning is a "function within the manufacturing environment which deals

with selecting the manufacturing processes and parameters to be used to transform a part

from its initial form to final shape according to design specifications" (Chryssolouris et

al., 1984, 1985, 19th CIRP). This definition clearly indicates that the process planning

1
function deals with the technological selection of the resources so that a raw material can

be converted into a marketable good. Most of the decisions in process planning are made

in the area of machine tool selection, machining process selection, fixture and tool

selection. Process planning is normally performed prior to the scheduling function. The

decision regarding the type of machine and processing is taken and sent and decision is

made know to scheduling function.

Scheduling function on the other hand is a final allocator. Scheduling mainly

considers the timing information while allocating resources to the tasks. Scheduling may

be done either on a dynamic or static consideration. Most of today' s scheduling systems

are based on static scheduling. Static scheduling assumes that the conditions on the shop

floor do not change in the time period for which the schedules are being created. This

assumption leads to making of unfeasible schedules. As the shop floor conditions change

at every instant, the above assumption fails to reflect the shop floor conditions

accurately. To make shop floor schedules which reflect the real time conditions, dynamic

scheduling is performed. In dynamic scheduling, no task is assigned to the resource

without being sure of its availability. All schedules are measured based on its

contribution to the performance criteria. Hence, scheduling unlike process planning is

aimed at improving the time performance of the system.

Process planning function being a technological decision maker about the process,

machine tools, operation sequence, etc., imposes tight constraint on the scheduling

function. Scheduling function is the fmal decision maker, which decides on the timing

issue of assignments. As scheduling has the direct interface with the shop floor, it bears

2
the brunt of wrong decisions high up in the hierarchial order. Being totally of two

contradictory school of thoughts, there is a lot of inconsistency in decision making

between the two functions, affecting the performance on the shop floor. Another

important decision flaw on part of the scheduling system is the assumption that the shop

floor is static (Khoshnevis. 1989), while this is not true. Due to the changing

conditions, the resource availability changes. This change affects the previous

assignments, and hence rendering most of the plans unfeasible. Most of the shop floor

supervisors make temporary changes in the schedule, trying to optimize the present

conditions on the shop floor. But in a global perspective, the whole performance of the

shop floor collapses. This local optimization leads to increased work in process,

increased makespan, etc. Furthermore, in the area of process selection, it has been

found that the process planners tend to select the most desirable process always, not

considering the availability of the process on the shop floor at the instant.

In summary, one can say that the problem of integration of scheduling and

process planning is mainly that of feedback. The two functions perform in total lack of

information feedback and hence make inconsistent decisions. I have proposed a possible

conceptual solution to the above mentioned problem in this thesis. The proposed model

is called Integrated Process Planning Model (IPPM). In this approach, process planning

and scheduling functions have been treated as two functions having same decision-making

priority and thereby considering multiple criteria for decision making. A feature-based

approach has been taken to deal with the process planning function. This approach

follows some obvious advantages as reported by Chang et al. As a optimization based

3
solution to this problem is virtually not possible due to its complexity. a heuristic

solution is suggested. The key area in this model is the matrix-based representation. The

underlying principle in this approach is that if the scheduling and process planning

information can be represented in the same domain it is easy to understand the problem

and obtain a solution which considers the two objectives simultaneously. A fuzzy set

operation based heuristic has been proposed to solve the problem of selection of feasible

process plans from all possible plans. These selections are based on the hamming

distance calculations between the various process plans.

This thesis has been structured into 6 chapters. Chapter II deals with the

definitions, functions, and types of process planning and scheduling. The problem has

been addressed in detail. Topics like feature based process planning, computer aided

process planning and other topics have been addressed. Chapter III starts with the

problem definition. Literature survey has been reported regarding the topic in concern.

Furthermore, literature about feature based CAPP and fuzzy set operations have been

reported. Chapter IV has been used to explain the proposed model and some of the

mathematical computations used in the model are shown. Chapter V has been used to

solve a numerical example, encompassing all the algorithms proposed in the Chapter IV.

Finally, the report concludes with Chapter VI, where the conclusion of the research is

reported.

4
CHAPTER IT

PROCESS PLANNING AND SCHEDULING

2.1 Planning

2.1.1 Planning Defmed

It has long been accepted that planning in a manufacturing environment 1s

essential in achieving the ultimate goal of "Completely Integrated Factory." Planning is

a key link between design and manufacturing. It also prevents the factory from becoming

chaotic, by providing decisive controls. Lot of work has been done in the area of

developing planning models, which can be used in industries of today. But most often

the results have been dismal. The models have failed to exactly represent the real time

conditions of the industry. This failure has been attributed to the complex and dynamic

nature of the planning domain.

Planning has often been defmed as the activity of devising means to achieve

desired goals under given constraints and with limited resources (Ham, 1988). Therefore,

one can clearly see that there are three ingredients of planning: (a) goals, (b )constraints,

and (c) resources. Feedback is an essential factor in planning. Execution of the plans

depends on how well a feedback cart be received. Based on the information from the

feedback, plans are modified and executed. This type of planning is dubbed as a "Closed

Loop Planning."

5
2.1.2 Manufacturing Planning

Manufacturing planning is a bridge between the objectives of design engineers and

manufacturing engineers. Designers prepare the blueprint of the pan to be manufactured

in terms of specifications, functionality, and reliability. Manufacturing engineers then

have to convert the intention of design engineers into a real product. Here, the

manufacturing engineers are constrained by the shop floor conditions and certain design

restrictions. Hence, it can be seen that the main aim of manufacturing planning is to

coordinate the two, design and manufacturing objectives, to provide a production facility

which can finally produce the desired product. There are technical and non-technical

aspects of manufacturing planning. As this thesis deals with the technical aspect, the

technical facet of manufacturing planning is explained.

Production Planning- Production planning deals with planning for the whole

production facility, and is done at a high level. All decisions regarding the types of

product to be manufactured, time of manufacture, personnel that will be in charge of the

production, and the facility that will be used for the production are made. This type of

planning is also referred to as scheduling, though most of the professionals believe that

scheduling forms a single function in production planning.

Process Planning- Process planning is a function which selects, describes, and

sequences the processes required to manufacture the product. Most of the process

planning functions regards the selection of the process as a primary goal and other

objectives like cost as a secondary goal. Process planning is often regarded as a low level

planning as it deals directly with the shop floor.

6
Process planning includes:

a.selection of machine tools,

b.selection of tools,

c.selection of machining operations and their sequence,

d.selection of jigs and fixtures,

e.calculation of cutting conditions, and

f.creating NC programs.

Operation Planning- Operation planning is perhaps the lowest type of planning on

the shop floor. Often, it follows the Process planning phase. In operation planning, actual

calculation of the parameters, design of jigs and fixtures, etc., is done so that the part

can be directly manufactured on the specific machine.

It can be seen that the above said functions form a hierarchical structure with

respect to their scope of execution. While production planning is done at a high level,

process and operation planning is done on the shop floor. Further, it is also evident that

the three planning functions should be studied separately. It is important that the three

functions be as integrated as possible, so that the main objective of "Integrated Factory"

can be achieved.

2.2 Process Planning


2.2.1 Process Planning Defmed

Process planning has been defined by the Society of Manufacturing Engineers as

---
"The systematic determination of the methods by which a product is to be manufactured
------~-··

economically and competitively." Output of process planning function is process plan.

7
Process plans are a set of manufacturing order which the facility must follow to convert

the raw material into a finished product. Figure 2.1 gives the overview of the process

planning method. Process plan can be represented as a graph where a node represents a

state of transition which the pan must undergo and the path itself represents the transition.

Every transformation is triggered off by a decision, which forms an instruction in the

process plan. Transition from one node to another shows a gain in information, i.e.,

more features are added to the product (material removed) to form the completed product.

The transition from one node to another requires a lot of information like, what process

to use, what machine and tool, and when it should be manufactured. Further,

information regarding the quality and quantity also forms a very important and governing

information. Hence, it can be seen that process planning provides information required

for transition from one state to another. A process plan establishes a set of operations to

be performed according to some set of allowable sequences. These sequences govern the

smooth transition of the pan. Therefore process plan can be described as "a set of

instructions describing the processes which allow a pan to make the transition from initial

state to the final state" (Chryssolouris et al., 19th CIRP). Figure 2.2 briefly shows how

the process plans are generated.

Process plans are made of work elements. Work elements correspond to the

operations, steps, and substeps. A set of work elements represent instructions.

Instructions are sent to the work unit level, where the instructions control the elementary

activity of producing the part. An operation consists of a number of steps, which is

nothing but an instruction describing how a part can make a transition. Most often, a

8
Select Raw Mlilerial

Crate Operation Sequence


I
Standard
Seq ~ Variant
Operation
Genentive ~ Decbion Locic l

/)c-.. '
o-·~~~

Feeds
Crate
Tool Plan Tool Layout Cut Sequence & detailed
operation
Speeds Instructions

Interactive Manufactur-
Graphics ability
Database

Figure 2.1

Process Planning Overview

9
Technological and Company Specific
Economic Criteria
Criteria

Figure 2.2

Generation of Process Plans

10
cut is regarded as a most elementary action. A tool operation is a collection of these cuts,

such that the same tool is used without interruption. Further, a fixing operation, which

gives rise to setups, is a grouping of tool operations that use the same fixing without any

interruption. A machine operation is a grouping of fixing operations which uses the same

machine without interruption. Hence, it can be seen that the instruction describes the

various groupings and orders the sequence within each group to achieve the transition.

In a traditional process plan, we are interested in machine grouping first, then in fixing

grouping, and finally in tool grouping. It is in this sense that machines are hierarchically

higher than tools.

2.2.2 Steps in Process Planning

The task of process planning is performed in series of steps (Alting et al., 1989).

The first step being the interpretation of the design data. This step usually involves

understanding the blueprint, or porting the design from the CAD package. In this stage,

the requirements of products , such as batch size, geometric configuration, raw material

property, dimension tolerances, surface roughness, heat treatment and hardness, and

some other special requirements, are studied and interpreted. The next step is the

selection of the machining process. Most of the time this selection depends on the

company's strategy. The strategy is to select a process or processes from turning,

milling, drilling, etc. After the selection of process comes a very important selection,

being selection of machine tools. The selection of machine tools can be either in the

form of selecting machining centers or individual machines itself. Selection of machine

tools normally is guided by many criterion like, availability, process capability, range

11
of machining operations, and production rate. Once the machine tool has been selected,

jigs and fixtures have to be selected based on the characteristics of the job. After the

selection of the fixtures. operation sequencing is done. This step consists of detailing

operations for a defined group of parts. Each operation is described in a comprehensive

manner. elucidating each detail of operation. The operation depends on the shape of the

part and the dimension. Datum surfaces or reference surfaces are selected so that the

position of the clamping devices can be planned. The position of the clamping device can

affect the machining process as it can restrict the tool motion. Further. the reference

surface also affects the machining accuracy of the referred surface, hence has to be

selected carefully. Required tools and the cutting parameters have to be assessed. Some

of the parameters which need attention are depth of cut, cutting speed and feed. This

will lead to the calculation of machining time, and non-machining time. Setup time,

material transfer time, and tool change time is calculated. The overall cost calculation

depends on the time . Finally, process sheets, operation sheets or route sheets, and part

programs are created and sent to the shop floor supervisor for execution. The steps

followed in process planning can be summarized in the point form as under:

1. interpretation of product design data,

2. selection of machining process,

3. selction of machine tools,

4. determination of fixtures and datum surfaces,

5. sequencing the operations,

6. selection of inspection devices,

12
7. determination of production tolerances,

8. determination of the proper cutting conditions,

9. calculation of the overall times, and

10. generating of process sheets including NC data.

Hence, it can be said that the function of process planning is to define what has

to be produced, how it should be manufactured and by what means. With the planning

of the operation sequences, the product is broken down into a series of machining

operations. The most suitable technology is selected and the operation is planned. Based

on the selected technology of production, the manufacturing equipment is selected.

Operation time planning follows the process planning closely. Based on the machine and

operation, operation times can be calculated. Operation time normally consists of:

I. The time for each operation of a process sequence,

2. The start and fmish times for process sequences,

3. The time intervals between successive operations, and

4. The total time to perform the tasks.

Requirement planning is then performed to plan the labor and equipment

requirement. Material planning is performed to select the raw material, its shape and

quantity is considered.

Decision points or variables in process planning is purely based on the process

planners experience and tends to be company specific. Most of the time variables can

involve all of the above mentioned planning points.

13
There are some important variables which a process planner has to g1ve

considerable thought. Some of them are:

1. Processing operations: Selection of the necessary machining operations and

sequences to generate the desired shape, size, finish, and tolerance of the part.

2. Machine tools: The selection of proper machine tool based on capability and

availability.

3. Workpiece material: Selection of the raw material, based on the design

features, and manufacturability studies.

4. Cutting tools: Selection of appropriate cutting tool based on operation, work

material, cutting conditions and other factors.

5. Cutting fluid: Selection of cutting fluid based on cutting conditions.

6. Machining variables: Selection of feed rate, cutting speed, depth of cut and

other machining parameters.

The selection of the above variables affects the material removal rate (l\1RR). The

MRR influences the economics of manufacturing greatly.

2.2.3 Criteria Used in Process Planning

Selection of the process plans are based on some selection criteria. While there

may be large alternative process plans possible to manufacture a particular part, most of

them may be unfeasible.

14
Hence, a higher order pruning is required to kill the process plans which are not

applicable on the shop floor. For this some criterion are developed. Some of the

important criterion can be (Bhaskaran, 1990):

1. minimize number of setups,

2. minimize processing steps, and

3. improve machining accuracy.

On closer observation. it can be seen that each one of the above stated criteria

actually contributes greatly to the machining economics. Each one of the criteria aims

at reducing the cost by making the manufacturing time of the job as short as possible.

Some of the economic consideration may be:

1. minimize production cost,

2. maximize production rate, and

3. maximize total profit.

2.3 Computer-aided Process Planning

Computer-aided process planning or CAPP is the application of computers in

process planning. With rapid development of computer aided techniques, research in

CAPP has progressed rapidly. CAPP systems are being extensively used as a part of

Computer Integrated Manufacturing environment. CAPP systems are based on the

prowess of the expert systems where logic can be easily written in form of objects. This

ease of programming has enhanced the development and use of CAPP systems.

15
Most widely felt advantages of computer in the area of data storage. retrieval and

data manipulation has been very well exploited in todays CAPP systems. There are

basically three approaches to CAPP. they are: (1) variant, (2) generative, (3) semi-

generative. Each approach has been explained as under.

Variant approach- Variant approach to process planning is similar to the traditional

manual approach. where the process plans for a new pan is created by merely recalling,

identifying, and retrieving an existing plan for a similar pan. Small modifications are

made on the retrieved plans to suit the new pan. Group technology may be used to store

the master plans. As the plans are only modified, it is difficult to cater to all the

geometrical combinations, thereby giving a sub-optimal plan. In variant approach, the

computer plays a role of a storage location and does not contribute to the generation of

the process plans itself. The advantage of such a system is that it is easy to maintain and

runs at lower costs.

Generative approach- As the name suggests, the process plans in the generative

approach are generated using the decision logic, tables, and some algorithms. The rules

of manufacturing and equipment capabilities are stored in a computer. The ideal approach

would be to generate a process plan using these rules without the intervention of the

process planner. The input to the generative CAPP system will be either in the form of

a user interface or direct porting of CAD data in a suitable format. (into the CAPP

system). Generative CAPP system uses feature algorithm and AI-based reasoning to

develop plans. The most important requirement in the generative system is the ability to

manipulate data to bring out the desired result. The data manipulation solely depends on

16
the complexity of the logic carried in the system. The logic itself should be auto driven

based on the self understanding of the situations rather than be driven by set of rules.

This is where the expert based systems fall short. To auto trigger the logic system a

anificial intelligence technique is required. Some applications have been found using

neural networks where over the period of time a system learns how to operate and hence

comes out with a solution. If there has to be a good data manipulation, the data itself

should be stored in such a way that the logic can fmd some correspondence between the

data and the situation. Based on the type of type data storage, the generative approach

has five alternative possibilities, they are:

a. Decision tables and decision trees. Developed in late 1950, this was an

effective means of capturing logic. The tables are composed of conditions, data, and

actions. Decision tables provide a graphical method of representing logic.

b. Decision tree approach. This has the same logic as the decision table and can

be converted from the decision tables. The tree approach provides a clear structural view

of the logical concept.

c. Decision tree/key word search. The information for the process planning is first

obtained from the classification of the part. This information or called as keys is used

to determine the path selection on the process decision tree.

d. Classification table/key word search.

17
e. The axiomatic approach. A process sequence model for a given part family is

developed, and then extended to incorporate logic for specific types of operations. This

extended logic is based on geometry, dimension, material. and special process

variations which are allowed for the part family.

The semi-generative approach was used as an alternative to the generative

approach. It was found that developing a truly generative approach was difficult and

hence pursue a pseudo-generative, wherein user interaction can be reduced by using such

features as standard operation sequence, decision table, and mathematical formulae. In

the semi-generative approach, a pre-process plan is developed and corrected before it is

implemented on the shop floor. All the required logic is available within the system. At

the first sight, the system works like a generative CAPP, but the final process plan has

to be examined and corrected for the errors before it can be implemented.

2.3.1 Databases in CAPP

As a result of the computer-based approaches, what were previously charts,

tables, blue books, etc., came to be known as the database. All of the required

information in the computers are stored in ft.les, which are a part of the database. In

CAPP, this database is called as a manufacturing database. Vast amount of information

can be stored in these databases. In fact, most of the success achieved in the area of

integration has been due to the possibility of databases.

18
All information is stored and can be retrieved from these locations. Database

managers which are computer codes written to search, update and or retrieve the

information has helped in enhancing the system performance to a great extent. Two

major developments have resulted in this phenomenon, one being the development of

time sharing capability and other being the development of real time operation on the

computers.

As mentioned before, databases are made up of datafiles. These datafi.les store

attributes regarding the object of interest. If the object of interest is a machine tool, then

information regarding its horsepower, dimension, capability, etc., will be stored. All

the files in a manufacturing database are related. The degree of relationship will vary

based on the operation being performed. One or many files may be opened to perform

a particular logical computation. Some typical manufacturing dataflles may be: part

design dataflle, tooling datafile, fixture datafile, machine tool datafile, raw material

dataflle, inventory datafile, machinability datafile, etc. Of all of the above dataflles,

machinability datafile is of special interest to us. Originally this datafile was structured

based on the machining data handbook. Recently the format changed, rather than storing

the data and simply retrieving them, the new system actually performs the engineering

analysis necessary to determine the recommended cutting parameters. There has been a

lot of discussion on machining parameter optimization. Some of the developed algorithms

find their way into this database. Figure 2.3 shows some of the databases used in process

planning.

19
Stock ixture
atabase atabase

Standard

Process

Plan

Fjxtu- Documenta-
nng tion

~gese
owl­

Figure 2.3

Expert Process Planner

20
2.4 Feature Based Process Planning

Currently, there has been a lot of research in feature-based process planning. The

motivation for such a approach as expressed by Tsang et al. is that the:

1. Expertise naturally exists in terms of features.

Observation of the human process planners has revealed that they naturally think

in terms of features when designing a process plan. Hence, it is easier to automate

process planning function when pans are considered as a set of features.

2. Blueprints description entails loss of information.

The blueprints lose a lot of information, with respect to the process planner. The

choices made by the designer are carried in the blueprint, but the rationale behind the

choices is omitted from being represented. As it is important for the process planner to

understand the operation of the part to prepare a good process plan, it is important that

the blueprints represent the functional significance of the pan. Being deprived of this

kind of information from the blueprint, the process planner makes a wild guess.

3. The feature approach is a federating concept.

By defining all the pans in terms of feature, the description will be very close to

what the process planner can understand. This will also result in taking the

manufacturability issue while developing the design.

21
2.4.1 Feature System

2.4.1.1 Defmition of Feature

Process planners define "feature" to be an area of interest. Another popular

definition considers feature to be "regions of pan having some manufacturing significance

in the context of machining." It can be seen that defmition of the feature can vary based

on its usage. Figure 2.4 gives some example of features. Hence. if a solid part is

regarded as solid object surrounded by surfaces, internal and external, then the features

can be regarded as local geometric entities that add to form the solid part. The addition

of the geometries is basically the approach taken by constructive solid geometry. In this

approach of defming parts, a boolean operation of features is performed to give the

finished part.

2.4.1.2 Types of Feature

The following are two types of features: (1) manufacturing and (2) process

planning

1. Manufacturing features are the features defined with respect to the

manufacturing activity. This type of feature is a simple tool trace, or a multiple tool

trace.

2. Process Planning features - With respect to process planning, feature can be

defmed as a "geometrical form and set of specification for which a process planning

process exists and this process is almost independent of the processes of the other features

of the pan." (Tsang et al., 1989).

22
Slot:

Hole:

Cylinder:

Figure 2.4

Examples of Features

23
It can be seen that process planning feature is a generalization of manufacturing

feature. Apart from manufacturing feature. process planning feature also includes

conjugate feature, super feature, and rough feature. Some simple features are as

follows: Planar features, including face, being regular, cluttered, boss, or irregular

face; Cylindrical features. including bore, being plain hole, tapping hole, counter bore

hole, trued bottom hole, conical bore, or blind hole; Cubic features, including

Lumiere, and Slot. being notch, or slot-in-bore; Profile features, including Contour.

being circular contour, or special contour; and other features like counterbore, chamfer,

or groove. Process planning features are a generalization of manufacturing fearures.

They include: (a) conjugate feature, (b) super feature, (c) rough feature.

(a) Conjugate feature, as the name indicates refers to a corresponding feature.

In machining we often come across female and male features. Female fearures are

depressions from which the material has to be removed. Male features are protrusions

corresponding to the female features. Female features are very helpful in preparing the

process plans as they give information about how much material has to be removed. On

the other hand male features are termed as conjugate features as they only give

information about how much material has to be left on the part.

(b) Super features are a group of features, grouped because of some common

reference which they share. This type of description allows the process planners to

naturally group individual features in more conceptual and functional chunks. For

example a group of holes sharing common center, two neighboring pockets, both being

mirror images.

24
(c) Rough features help in determining the fixing datum. Not all features are

smooth finished, and this feature forms the rough feature, which in tum forms the

reference surface for the first operation.

According to Gindy ( 1989), form feature is defmed as a "Geometric entity that

represents shape pattern that has some significance." He is of the view that form

geometry is made of an entry boundary, exit boundary, and a depth boundary. External

access directions are the directions from which the feature can be machined. Figure 2.5

shows the form feature classification as per Gindy's method of classification. Basically,

this figure suggests that there are main features, and the subfeatures derive characters

from main features. All sub-features are explained with respect to the main features and

cannot exist on their own. For example, chamfer is a sub-feature of a diameter and

cannot be found by itself. Feature representation is a very important topic that needs to

be addressed before we conclude this section on feature based process planning. All

features are represented either in the form of rules or a graph. Joshi and Chang in their

paper use the attributed adjacency graph (AAG). Figure 2.6 shows an example of their

approach to representing the features. All faces which are concave are shown with 0 and

otherwise with 1. An algorithm has been developed to understand the graph and

recognize the features. Joshi et al. talk about the feature relationship also. Feature

interactions occur in two ways. One being that the features have common edges between

them, and the other type of interaction being that the features share a common face.

25
Form Features

Not Through Real Imag-


Boss Hole Through Slot Notch Step Surface inary
Pocket Surface
Slot

Figure 2.5

Form Feature Hierarchy

26
Figure 2.6

Feature Representation Using AAG

27
2.5 Models to Integrate Process Planning
and Scheduling

There are three basic approaches to integrate process planning and scheduling.

The approach differs in terms of the type of information used in making process plans.

Based on weather there is a feedback from production control or production planning

department the integration approaches can be categorized as belonging to on of the three

categories. The three approaches include: ( 1) non-linear process planning (NLPP), (2)

closed-loop process planning (CLPP), and (3) distributed process planning (DPP).

1. Non-Linear Process Plarming (NLPP) got its name because of its inherent

property of being non-linear, i.e., the created plans do not form a linear structure,

instead are branched at every node. Figure 2.7 shows the structure of a NLPP. In this

method the Process Plarmer prepares all the possible process plans for a particular part.

This results in numerous optional plans. Most often, owing to such enormous number

of plans, they may be ranked based on the feasibility. NLPP allows plans to be created
--·-·----
before the part actually enters the manufacturing system. The underlying principle

followed is that, all problems which can be solved ahead of time or which can be

foreseen ahead of time should be solved, to reduce the decision making requirements on

the shop floor. When the decision is required, all that one needs to do is select the

alternative which best suites the situation, rather than make plans, which can be time

consuming.

The biggest disadvantage is felt in the situations where there are many parts in the

system. Plans made by NLPP tends to be enormous, creating storage problems. Further,

efficient pruning system is required to kill infeasible plans.

28
i
{K
s
Process Production
Planning f--.. Planning

Figure 2.7

Non-Linear Process Planning /


29
The advantage of this type of planning can be found in areas where there is no
...._____________
feedback system. As the dynamic condition of the shop floor cannot be reponed, it is

wise to be prepared for all contingencies. Referring to the figure, it can be seen that

plans are prepared and stored. Plans are stored with a priority number. The production

planning function selects these plans in order of the priority, checks for its applicability,

and finally executes it. If the plan is not suitable for the condition, then it will be

rejected in favor of some other more suitable plan, though with a smaller priority

number. Selection of the plan based on suitability can slow down the process to a great

extent.

NLPP is based on the ability to foresee the situation. There can be numerous

situation on__ t?e S~C>_P poor that a planner has to foresee. Creating a plan for each

situation can be extremely time consuming. Funher, it is virtually impossible to foresee

each and every situation on the shop floor. If an unforeseen situation were to occur, then

there would be no plan available for execution and bring the whole manufacturing unit

to a standstill.

2. Closed-Loop Process Planning commonly referred to as CLPP, forms a loop

type of architecture. Figure 2.8 shows a structure of CLPP. The information travels in
......____ ___________ -

a loop till it is verified. Once the information is verified, plans are made based on this

information. Typically, in manufacturing environment, process planning, production

planning, and production control interact to create and execute a plan.

30
Production
Planning
Production
Control

Figure 2.8

Closed-Loop Process Planning

31
In the previous architecture, due to the lack of availability of feedback facilities.

plans are created and checked for the suitability on the shop floor. In this approach.

plans are created only upon a feedback.

Feedback refers to the information about availability of resources in terms of

machine tools, jigs, fixtures, and tools. Process plans are created based on the available

resources. Hence, each plan is a feasible plan with respect to the production planning

function. As compared to the previous strategy, CLPP considers the dynamic condition

of the shop floor before creating the process plan.

Another advantage of this plan is the ability to create process plans as and when

they are needed. CLPP is performed in a sequential request-answer form. All the

decisions taken in this type of planning are constrained by the shop floor conditions.

3. Distributed Process Planning (DPP) is its inherent integrated approach to

creation of process plan. Figure 2.9 shows a structure of DPP. This approach is

commonly referred to as "Generative architecture." DPP allows simultaneous creation of

process and production plan. DPP divides the planning problem into two phases, namely:

pre-planning and fmal planning. In DPP, the job is studied in detail, all of the resources

required to machine the part are identified. After the identification, the requirements

are matched with the available resource. Flexibility of the system can be increased if all

the alternative required resources can be represented. The most important step in DPP

is to first understand what capabilities the machine tools have. For every machine tool,

operations are listed in decreasing order of desirability. Meanwhile, pre-process planning

is performed, wherein designs are understood, blank size is calculated, and required

32
Equipment Job

u
•c
.,·s
... c Machine Pre-Process
C..! Operation Planning
c. Capabilities (Operations)

r--.. t
Shop Available Operation Job Orders/
Status ~ <;>perat- f+ Matching ~ Opera-

··-
-=
c c
·- c
u

"'-.!
c.
Mainte-
nance
Work
lOllS tion
Sort-
ing
r--
Jobs
Manager
Interv-
ention
II
Calendar
II
~
"----"'
Free Seize Advance
Machine job and opera-
machine tion
(Operat- operat- pointer
ion) ion

Future Future
operation
calendar I-- -- job
opera-
tion
calen-
dar

Figure 2.9

Distributed Process Planning

33
operations are identified. The next step is to match the machine capability and the part

requirements. This is called as operation matching. A list of available machines and

their operation is matched with the desired operation on the pan. If a match is found,

the machine tool and the job are seized. Operation time can be computed for this match,

hence. giving the finish time of the machine and the job. This allows the scheduler to

predict when the job and machine will be ready to be assigned again.

DPP is a rational method for machine tool and process selection. Real time

information can be considered while making a process plan. Available resources are

constantly updated thereby resulting in a feasible solution.

2.6 Scheduling

As an order enters into the factory in form of a customer order or market demand,

the tasks of (a) Setting time and quantities of raw material release to the shop floor~

(b) Determining quantity to be moved through each operation for each type of

work in process over time~

(c) Comparing what actually happens on the shop floor with what was anticipated~

and (d) Forecasting the output of the finished goods remain to be done. The above

said jobs are mathematically daunting. Many of these problems are continuously solved

on the shop floor practically. Often scheduling is called as an optimization based

resource allocation. The main function of the scheduling is to assign specific task to

specific factory resource with specific stan and end times indicated.

34
A resource in a scheduling system is normally required for a time in order to carry

out an activity. Resource can be machine tool, labor. software, or any other material.

Scheduling is all about assigning these resources to specific tasks.

Scheduling rules are used to select tasks, or resources. The rules can be either

static, i.e., they can be applied at the beginning of the scheduling period and the result

in fixed schedule for the period, or they can be dynamic, i.e., changing over time.

Scheduling, most often follows capacity planning. In capacity planning, an idea of the

amount of available working hours on each resource is obtained which forms the

constraints on scheduling.

Scheduling, in general is concerned with makespan, work in process inventory,

and many other performance criterion. Basically there are three approaches to solving the

scheduling problem. They are: (1) manufacturing resource planning; (2) optimized

production technology; (3) just-in-time.

( 1) Manufacturing Resource Planning or l\1RP II was originally known as Material

Requirement Planning. Figure 2.10 shows the context of l\1RP II. It aimed at due date

performance. l\1RP II, unlike the classical scheduling approach of minimization of

makespan and flow time, approaches the problem by backward scheduling. Underlying

the assumption of such an approach is that all resources are used to convert raw material

to fmished product, and if the material can be scheduled to reach the resource at the right

time, resources are automatically scheduled. MRP II does not attempt to produce time

line for every resource in the factory, rather aims at specifying future events. Master

Production Schedule CMPS) forms the most important output of this scheduling process.

35
resources+bill or
Engineering material+routings+
material planning+
capacity planning
parameter
master production schedule
work calender ,...---------,
WIP
~---
stJitus Production
Shop
Floor / \ Control

I MRPU
)
.\ J
~ispatch production ....___ _ _ _ ___.
rorecast
list

Figure 2.10

External Context of :MRP II

36
(2) Optimized production teclmology (OPT) is used to schedule production off

line. The main consideration is such a scheduling approach is the resource utilization and

its dependency. Figure 2.11 shows the context of OPT. The distinguishing feature

between OPT and MRP II is the distinction which OPT makes between process batches

and the transfer batches. As the schedules in OPT are produced off line, it is incapable

of responding to the disruptions on the shop floor. Allowances are made while creating

schedules to overcome the lack of control and cater to any changes in the assumed course

of action. OPT is a fmite forward scheduler. Finite refers to its ability to respect given

capacity constraints and forward means that it deduces activity completion times for a

given start time and not the other way.

(3) Just-in-time, or JIT, was first employed by Toyota Motor Co., Japan. In this

system the shop floor is organized into flow lines built from cells of different machines

laid out in loop, staffed by operators trained to run all of them with autonomous

scheduling. Setup times are reduced in steps. Equipment is dedicated by product

families. Setup tasks are classified as internal if they require the machine to be stopped

and external if they can be performed without stopping the machine. JIT has a very good

application in terms of real time response. TIT responds to changes in the shop floor as

and when they occur.

In general, scheduling 1s concerned with information about the resources.

Information about labor absenteeism, machine tool availability, loading and unloading

times of machine tools and material handling, etc., are of special interest to the

scheduling function.

37
resources+bill of
Engineering material+routings
+material planning
+apacity planning
parameter

Produc-
Shop tion
Floor Control

dispatch list

Figure 2.11

External Context of OPT

38
As mentioned in the previous section, databases play a very important role in

storing these information. Figure 2.12 shows one such structure of the scheduling

database.

The mam arm of scheduling, as mentioned, IS to compute machine/job

assigrunents. These assignments are generally based on the desired performance criteria
......____ ---- -

and some company specific rules. Hence, all scheduling is basically done using some

son of rules. Scheduling as such may be categorized into two major types. One being

static scheduling and the other being dynamic scheduling. Static scheduling involves

assigning resources to tasks when the condition remains unchanged in the period for

which scheduling is being done. This approach assumes that all resources will be
~----~------

available when desired. Most of today's scheduling application is concentrated in this

-----
area.

Extensive computation procedures can be developed to get the best optimal

solutions. But due to the computational constraint, heuristics are developed and applied.

On the other hand, dynamic scheduling does not assume availability of resources. The

underlying principle of this type of scheduling is that the conditions change frequently,

i.e., the conditions are sufficiently dynamic and no assumptions can be made regarding

the availability of any resources. Scheduling for such conditions are done based on some

priority rules. R\lles are developed and all assigrunents follow these rules. The
'-,

assigrunent rule changes as the conditions change. These rules are aimed at optimizing

some performance criteria which has been set.

39
SCHEDULING
DATABASE

CODE DESCRIPTION

MCID MACHINE IDENTIFICATIO

JBLD JOB CURRENTLY LOADED

FT FINISHING TIME

QU JOBS ON QUEUE

MACHINE CALENDAR FILE

CODE DESCRIPTION

MCID MACHINE IDENTIFICATION

PtrriL PERCENTAGE trriLIZATION DESIRED


MXOVLD MAXIMUM OVERLOAD (PERCENT AGE OF
PtrriL)

CAPACITY DAT AFILE

Figure 2.12

Scheduling Database Files

40
Rules can be something like FIFO: First in-first out or Slack time/operation- assign

tasks those that have the smallest slack time etc. These rules help in prioritizing the task

and resource and hence contribute to the performance criteria.

41
CHAPTER III

LITERATURE SURVEY

3 .1 Problem Definition

As mentioned in the previous chapter, process planning and scheduling functions

have to be completed before a part can be manufactured. While process planning is

concerned with selection of operations/process and other parameters of operation,

scheduling is concerned with loading jobs on machines in an order so that the available

resources can be effectively used.

It is evident that scheduling function follows the process planning function. It is

also evident that the two functions are performed separately. Performing the two

functions separately leads to a lot of problems within a manufacturing system.

(Khoshnevis et al., 1989, Sundaram et al. 1988, Chryssolouris, 1984,1985, Tonshoff

et al.). Some of the problems follow.

Single criteria optimization is considered while developing a solution space.

Actual manufacturing conditions cannot be modeled by a single criteria. Simultaneous

consideration of various objectives is the best method to develop the solution. Single

criteria consideration allows the modeler to consider only one objective while

compromising some other important objectives. The solution so developed will result in

a solution which is optimal with respect to the considered objective, but with respect to

the system as a whole, the solution will be sub-optimal.

42
Process Plans are made with process planning criteria as the objective, they are

improved machining accuracy, improved tool life. desirability of a process for a feature.

etc., while scheduling is performed with some performance criteria as objective, they

being minimize makespan. flow time, etc. As each function aims at developing the

solution optimizing their set criteria, it is obvious that the solution space for the system

with their combined criterion will be sub-optimal.

Different criteria and their relative importance m decision making change

dynamically with the change in the shop floor condition. It is often found that the

process plans have to be modified before executing it on the shop floor. Process plans

are made with some assumed balanced condition. The plans become infeasible when

these conditions change due to disruptions. The main problem with process planning

function is that it assumes an unlimited availability of resource on the shop floor. The

resources are not always available, as they may be busy wit the current tasks,

breakdowns, scheduled maintenance, labor absenteeism, etc. Further, scheduling

criteria also changes with respect to the change in the shop floor. If a particular machine

is overloaded in a particular time window, it would be wise to schedule the jobs on

another compatible machine. While in another time window, the criteria could change

to minimization of processing time and hence assignment to those machines which allow

least processing time. It has been shown that, owing to the single fixed criteria approach

of obtaining the solution, 20% to 30% of the total load of particular period had to be

redirected. Also, it has been found that only a small amount of the orders actually

complies with the prepared schedule, thereby indicating that 20% to 30% of all the

-+3
process plans are not valid and have to be modified to be used on the shop floor. The

main reason for such high turnover in schedules and process plans are the rigid schedules

giving an inflexible due date. There is no flexibility provided in terms of alternative

routings. Hence, the main aim should be to improve the information to scheduling by

providing alternative plans.

It can be seen that no single criteria can be used to model the system, but the

criteria changes dynamically with changes on the shop floor. Process Planning and

scheduling have conflicting objectives. Process planning emphasizes on technological

aspect of a task, while scheduling emphasizes more on the timing aspect of the task.

Both scheduling and process planning functions are information generators. Both the

functions seek to assign the resources to the task by trying to attain their objective.

Scheduling lays emphasis on timing aspect. It tries to assign a task to the resource based

on some time optimization. Process planning function tries to assign the job purely based

on technological consideration. Resources are selected based on the desirability to

produce the feature, rather than its dynamic availability on the shop floor.

In summary, the following are true.

1. Scheduling follows process planning, it is restricted by the fixed process plans

which cannot be altered

2. Process plans assume unlimited resources at the shop floor. The resources

being: materials, machines, tools, fixtures, and operators. The resources are not

always available owing to disruptions, or simply because they are currently being used.

3. Process plans assume an idle factory

44
4. Process plans at the most recommend alternate processes

5. Desirable machines are repeatedly selected by various process planners

The consequences of the above said assumptions are the following:

Process plans are not completely followed on the shop floor and intuitive process

plans are generated on the spot by the supervisors to reduce the load on congested

machines, giving rise to sub optimal operating conditions.

Integration of the two functions is essential due to the above said consequences.

The integration will result in simultaneous consideration of objectives, resulting in

reduced production cost. Another important advantage of the integration will be the

possibility to have balanced loading on the machines. Further, by integrating the two

functions, the constraint on the scheduling function can be relaxed, thereby giving more

flexibility to the function. The added flexibility will allow scheduling to respond to the

changing conditions of the shop floor. The quick response will result in avoiding

overloading, and selection of an alternate machine when a previously selected machine

breakdown. An integrated approach will result in scheduling decisions made in the

planning office rather than on the shop floor. Flow rate of the jobs on the shop floor can

be kept uniform.

45
Summarizing, one can say that integration of the two functions will result in an

improved practice of planning and scheduling through re<Juc.tim:ljn scheduling conflicts,

elimination of scheduling and process planning as stand alone functions, reduction of

human intervention, reduction of production cycle time, improved utilization of

resources, quick response to change in shop floor events, and step closer to a completely

integrated factory.

3.2 Literature Survey

Many papers were reviewed pertaining to the topic of integration of process

planning and scheduling. It was found that not much work has been done in this area and

much is yet to be done. All the literature surveyed seemed to address the need to

integrate the two functions, though not giving much idea as to how they have to be done.

Chryssolouris and Chan in their paper "An integrated approach to process planning

and scheduling." address the issue of assigning production resources as a multiple criteria

decision making problem and suggests a decision-making technique suitable for the

manufacturing environment.

For the sake of applying the decision method, decision environment is assumed

to be made up of a hierarchial structure consisting of four distinct levels: ( 1) factory,

(2) job shop, (3) work center, and (4) work unit.

The decision model explained in the paper has been called as the MADEMA

approach. This approach is general in character and can be applied to any level. The

MADEMA (MAnufacturing DEcision MAking) considers a set of alternatives for the

execution of a particular production task; alternatives in this context are different

46
machine tools, operators, and handling devices needed to perform the task. The choice

of one of the alternative resource is made by an evaluation of relevant criteria/attributes.

This is done by creating a decision matrix, where the rows ALi (i=l.. .. n)

represent alternatives, the columns ATj represent the attribute and the entries a;j are the

values of the attributes for the corresponding alternatives. The proposed MADEMA in

the paper entails five steps:

1. Determine Alternatives (ALi)

Alternatives are prerequisites for any decision making process. Without

alternatives there is no flexibility in decision making. The resources should be

technologically feasible and available. Process planning has been used to determine

alternatives.

2. Determine Attributes (ATj)

Attributes are the criteria according to which the alternatives are to be selected.

Criteria can include time, quality, and many other variables.

3. Determine consequence with respect to the attributes for each alternative (~j)

Consequence are the values that the attributes take on at the time when decisions

are made. They are needed for evaluation and, eventually, to select the alternatives.

To determine the consequence, accurate values of the attributes have to be derived.

Time, cost, etc. can be derived from the out come of process planning.

4. Apply decision rules for choosing the best alternative

Decision rule is the way to select the best available alternative. The rule is most

cases are mathematical, which selects the alternatives based on highest utility value.

47
5. Select the best alternative

Following the decision rule, best alternative can then be selected.

A solved example of this approach is available in another of Chryssolouris et al.

paper titled "Decision-making on the factory floor: An integrated approach to process

planning and scheduling." They use AL 1 and A~ as two available alternatives and AT 1

and AT2 as two independent attributes. A numerical value for the utility of each

alternative is calculated. Further, preferences of the decision maker regarding the two

attributes are know in form weights on the attributes. It is reported that both the number

and the complexity of decisions could be enormous. It is reported that interactive

approach to determine the relative preference is inappropriate due to the time and cost

entailed. Even artificial intelligence is inefficient in doing so.

The main contribution of the MADEMA approach to the objective of real time

integration is the matrix based representation. MADEMA approach addresses the issue

of integration, but fails to consider the flexibility of process planning itself. Alternatives

suggested by MADEMA approach is merely alternative machine tools to manufacture a

feature. There is no consideration to determine if there are possible alternatives to the

features. The integration represented in this approach is not a true process planning and

scheduling integration. The suggested matrix shows the alternative machines, but does

not show alternate process plans. The process plan is assumed to be fiXed, the flexibility

only in the terms of machines, operators, etc.

48
As a multi-attribute integration is sought, the attributes are weighted based on the

decision maker's preference. Hence, the selected "best" alternatives reflect the decision

maker's preference and not necessarily the shop floor.

Khoshnevis and Chen, in their paper titled "Integration of process planning and

scheduling function," talk about the development of an automated process planning

environment in which the two problem domains of process planning and scheduling are

treated as a unified whole. In this new approach, the process plans for a given part may

be different as each arriving order may fmd the shop floor in an entirely different state.

A heuristic algorithm has been presented in this paper which integrates process planning

and scheduling to give good performance.

a. A dynamic list of available machine is maintained

b. A list stores the incoming parts with some priority number

c. From the list of features of the part, a feature is selected from the high priority

arts, that can be readily processes on the most desirable machine

d. Next plan time is advance to the first feature completion time, or the new part

arrival time

The above heuristic has been tried with two other types of planning, one being

manual and other being fixed process plan with optimization algorithm for scheduling.

Reported results indicate that the total processing time for the heuristic was the least,

while completion, flow, and tardiness were the highest. There was no saving in average

idle time or improvement in utilization. But it was reported that compared to other

methods, this method was easy to perform to obtain same results as other methods.

49
The new approach developed by Khoshnevis et al. considers the dynamic state

of the shop floor. By considering the features which for which the most desirable

machine is available, processing time and machining accuracy can be improved.

This method again fails to take full advantage of feature based process plans.

Features are considered separately. No feature representation scheme is proposed in this

model, where feature relationships can be determined and feature combinations can be

performed. The algorithm puts the parts in queue if it fails to find a most desirable

machine to manufacture the feature in question. The effect of feature based selection of

parts might lead to queuing of high priority parts resulting in tardiness. Further,

advancing the time to first feature completion time, leads to excessive planning. As

planning is done at every event, flexibility in terms of available machines is reduced.

This plan will result in making plans at every stage rather than improving the flexibility

of the plarming itself. Process planning criteria like reduction in setups, and processing

steps can be achieved by considering the possibility of combining features.

Tonshoff, Beckendorff, and Anders in their paper titled "FLEXPLAN- A concept

for intelligent process planning and scheduling" propose a model called FLEXible

PLANning to integrate process planning and scheduling. Referring to the need of just-in-

time process plans to respond to the changing shop floor condition, they propose a model

which keeps the time difference between planning and control. This difference is

proposed as they feel that TIT process plans is not feasible in the near future. In

FLEXPLAN, process plans are generated only once per workpiece, as compare to the

previous method of generating different process plans for every workpiece by Khoshnevis

50
et al. Any time consuming technological decision will be taken before manufacturing

starts. Process plans generated will comprise important alternatives of operation sequence

and resources which are considered to support rescheduling. Further, those operation

will be identified which can be performed parallel in time or those which are independent.

In the proposed FLEXPLAN, authors use a non-linear method of creating process

plans. Process plans are represented using petri nets. Pruning of plans are done at a high

level, where all alternatives will not have to be pruned. A feature-based part description

is used to port the part from the CAD to CAPP. On the first run, FLEXPLAN generates

a global non-linear operation sequence for the production of the workpiece. After the

global planning, a detailed operation sequence planner is called which looks up the

global operation sequence and starts the detailed planning per operation. By mapping

suitable work elements to the part, based on the required feature, process plans are

generated. If all the work element scan be applied without any conflict, a NLPP can be

prepared in a very straight forward manner. After the application of all the workelements,

time calculation is performed. Backward search, also commonly referred to as inverse

manufacturing operation, is performed to identify the required work elements to reach

for the final stage to the initial stage.

Most important step in FLEXPLAN is the creation of non-linear process plans.

These plans are created with the assumption that all major technological problems can be

anticipated in advance. A high level pruning reduces the requirement of pruning each

alternative plan. Pruning logic has been incorporated in the knowledge acquisition phase

of the FLEXPLAN. It will be very difficult to identify and defme the alternatives which

51
must not be pruned. Furthermore, excessive time will be spent in first of all identifying

the various possibilities that can occur on the shop floor. Creating alternative plans for

all the above possibilities can become computationally enormous.

Scheduling plays a very important role in improving the shop floor efficiency.

FLEXPLAN only considers the timing aspect of the scheduling, assuming that the criteria

will remain same throughout the operation period. When in reality, the scheduling

criteria also change with changing condition of the shop floor. An assumption that by

relaxing process planning constriants on the scheduling will improve the efficiency of the

scheduling is not entirely true. Simultaneous optimization of the scheduling and process

planning criteria is necessary. FLEXPLAN fails to consider the process plans in form of

setups, as reducing these should be the main criteria of process planning. By simply

providing alternative features and alternative machine tools to manufacture these features,

the problem of flexibility cannot be solved.

Sundaram and Fu in their paper titled "Process planning and scheduling - A

method of integration for productivity improvement" propose a schedule generation based

approach similar to that found in scheduling literature.

In this approach, Sundaram et al. propose that alternative machines should be

made available to perform each individual operation. The outcome of the process

planning function is used to develop the schedule. A key machine procedure is explained

to develop the schedules. An example problem has been worked out to enumerate the

methodology.

52
In the above methodology, scheduling is done after the process plans have been

created. It is clear that process planning and scheduling are not treated as a whole. No

special attention is given to the dynamic nature of the shop floor. This paper basically

tries to solve a scheduling problem without giving much attention to the process planning

stage. The planning is done well in advance, and total load is computed on each

machine. By using the shift principle, attempt has been made to make the loading a little

more balanced. But the biggest problem left to be answered is: what if one of the

assignments was to become infeasible, either due to machine breakdown or labor

absenteeism.

In summary, this paper really does not indicate a very good method to relax the

constraint on the scheduling, except that for a given operation alternative machines are

considered.

It is commonly felt that the criteria used to optimize the planning functions are

imprecise. A fuzzy concept was introduced by Zadeh, which gave rise to fuzzy set

theory. A process plan selection problem in manufacturing can be considered as a routing

problem, especially when an operation can be performed in many ways. It can be seen

that alternatives for both, operation and resources required to perform the operation can

be enormous. The selected machines have different process capability, processing cost,

flexibility to process variety of parts, different material characteristics, different scrap

generation, etc. The actual measure of these values can be non-commensurable and

imprecise. This impreciseness can be modeled usmg fuzzy concepts. Furthermore,

scheduling function also has immense imprecise specifications in terms of time constraints

53
and expression of rules providing advice about the sequencing of tasks. Hence, by using

fuzzy concepts, both scheduling and process planning criterion can be modelled and

solved simultaneously.

If E is set and A is a subset, then A c E. If an element x belongs to A, i.e., x

e A then its membership to A is given by }IA(x) = 1, if x e A the }IA(x) = 0.


If B is another subset of E, B c E and x e B. then }IA.'IB(x) = 1.

Further, J.lAnB(x) = J.lA(x). )J8 (x) called as boolean operation. J.lAvs(x) =1 if x e A u B.

All the above rules are true for a regular set. Let us consider a set A as a fuzzy subset;

{xI J.lA(x)}, 't/ x e E. If J.lA(x) takes its value in a set M, called the membership set M,

one may say that x takes its values in M through the function }IA(x). Then x ...... M. There

are lot of fuzzy set operations available from the literature. Only the operations

pertaining to this report has been shown here. For further reference, the reader is

encouraged to refer to some fuzzy set theory books. Principle of Intersection and Union

has been explained in this section.

Intersection: If A and B are two fuzzy subsets, then A n B is the largest fuzzy subset

contained at the same time in A and B, 1.e.

't/ x e E, J.lAnB(x) =MIN {J.lA(x),J.ls(x)}

Example: E = { X 1 , x2 , X3 }

A= {xtf-2, xi.8, xJ.6}

B = {xtf-5, xi.9, xJ.4}

A n B = { xtf-2, xi.8, xJ.4}

5-1-
Union:

Hamming distance: Relative hamming distance is given by;

55
CHAPTER IV

INTEGRATED PROCESS PLANNING MODEL (IPPM)

4.1 Objective

The main objectives of the Integrated Process Planning model are as following:

1. In view of the drawbacks of the current method of preparing process plans, as

stated in Chapter I, the main objective of this model is to truly integrate process planning

and scheduling functions.

2. To generate process plans which reflect the shop floor conditions and responds

to the dynamic nature of the manufacturing system.

In contradiction to the previous practice, IPPM produces plans "Just-In-Time."

This approach allows the plans to be prepared in the last minute when it is really needed.

Latest conditions of the shop floor can be included by preparing the plans just before

execution. Such consideration will lead to a dynamic process plan generation.

Furthermore, the plans generated by this method will result in different process plans for

the same part, every time it enters the manufacturing system. All plans are created based

on the resource availability,· instead of completely based on resource desirability.

3. To consider the objectives of both, process planning and scheduling function,

simultaneously and obtain best plans.

The underlying principle of the previous method of preparing process plans was

that, process planning function precedes the scheduling function hierarchically.

56
Figure 4.1 shows the factory information flow. This hierarchial structure allows

the process plans to be prepared based on the priorities of the process planning function.

thus imposing tight constraints on the scheduling function. If the two functions are

merged as a whole and the objectives are considered as a single set, plans that satisfy

both objectives can be selected.

4. To improve productivity of the manufacturing system by improving machine

utilization and reducing machining cost and time.

By specifying alternative machines to manufacture a particular feature, or by

specifying alternative features that can be manufactured, machine utilization can be

improved increasing the productivity and reducing cost of manufacture.

5. To improve integration between highly automated functions, thereby moving

closer to a completely integrated manufacturing system.

6. To improve the information storage, so that a common storage location can be

identified such that all manufacturing functions will be able to access this information.

4.2 Generic Solutions

Some generic solutions to solve the problem of integration are enumerated below.

The IPPM uses the proposed generic solutions to develop specific solutions in order to

solve the problem of integration.

1. Relax constraint on the scheduling function

- by specifying alternative machines to process a feature

- by specifying alternative features which can be processed at the time of planning

57
Job Order

Process
Planning

Scheduling l
Oprn
Plan-
Dispatching J
nning t - - - - - - - - - <

Moniloring

Production

Figure 4.1

Information Flow within a Factory

58
2. Select machine tools based on availability instead of purely based on desirability

3. Represent objectives of both scheduling and process planning on one

representation scheme so that they may be optimized simultaneously

4. Select process plans for various products based on "least difference," to reduce

scheduling problems

5. Features should be represented in such a way that all the relationships it has

with other features can be clearly seen

6. Store all the information from shop floor, process planning and scheduling

function in a central location so that information can be easily accessed by either

functions

7. Use only real time information from the shop floor to make plans

4.3 Integrated Process Planning Model (IPPM)

IPPM as such is structured along the line of Closed Loop Process Planning

(CLPP). The key feature of IPPM is its ability to prepare plans based on feedback from

the shop floor. At the functional level, a more one to one correspondence is sought to

match the task and resource. The distributed planning model fits well at the functional

level.

The underlying assumption of the model is that, a shop floor monitoring system

is available which monitors the shop floor and updates all the information. A shop floor

status file is used to store the updated information. All resources are monitored including

machine tools, labor, tools, jigs/fixtures, and material handling equipment. Figure 4.2

59
Inspection
at erial
nrorrnation
anagement

ixturing Feeds &


Cell
nrorrnation speeds

Process

Plans

Figure 4.2

Functions which Cell Controller Monitors

60
shows some of the functions which the cell controller controls. Figure 4.3 shows the

command execution architecture, by which the cell controller collects the information

from the shop floor and stores it into the shop floor status file. The integration itself is

based on two levels. Level 1 integration occurs at a factory level. Integration at this

level is aimed at integration of information. An integrated factory management system

is proposed, wherein the information generated by the shop floor, process planning

function, scheduling function, and other manufacturing functions are stored at a central

location. This architecture is shown in the Figure 4.4. The architecture is based on a

MRP II system. A factory database is proposed which stores all the information from

design, material management, personnel, marketing, and .MRP II system. Hence,

factory database is a central location accessed by top management level, where all the

factory information is available. This type of storage allows each manufacturing function

to share information. The material management group can order materials based on the

production planning information. 11RP II database is connected to this factory database.

Shop floor status, production planning information, and process planning information is

stored in this database. It can be seen that the level 1 integration aims at providing all

the information to all the functions when needed. 11RP II database most often stores all

the information directly related to the shop floor. Some of the information being; master

production schedule, bill of material, accounting, inventory control, material

requirement planning, work center information, routing, shop floor schedule. MRP II

database is a ideal integrator as it stores shop floor information. Shop floor monitoring

system updates its information into shop floor status file. This information is later

61
Factory Management Data

roduction

Control

"1:1
c
0:1
E
E
0
u
c
.::
:;
...:.<
<J

'-l

Shop floor

Cell Controller
Status File

Machine
on trollers

Figure 4.3

Command Execution Architecture

62
Factory Management System

Material
Manage
ment

Shop Floor
Status
File

Figure 4.4

Integrated Plaruling Model

63
transferred to MRP ll database later. Information about the part completion, machine

availability, tool availability. and schedule progress is noted and updated in the shop

floor status file. All this information is gathered by the cell controller which is most of

the time a dedicated computer having enough logic to control the shop floor.

Level 2 integration is the "functional integration." This integration mainly deals

with the integration of process planning and scheduling functions. As the main problem

is the integrity of decision making, process planning function has been treated as the

main information generator. The key integrator between the two functions is the

information itself. By using the shop floor and other related information, a common

decision making platform can be developed. Figure 4.5 shows the architecture to obtain

the integration between scheduling, process planning, and ~p II, using the proposed

IPPM model. This architecture has three modules: (1) Process Planning Module (PPM),

(2) Scheduling Module(SM), and (3) Decision Making Module (DMM). Figure 4.6

shows the scheme of the modules in the proposed plan. The information flow is shown.

Figure 4.7 shows the various modules and their functions.

4.3.1 Process Planning Module

Whole integration procedure depends upon this module. Process planning module

(PPM) is a expert system based module which has knowledge to make process plans.

PPM prepares the process plans in a slightly different manner, than the way currently

done. PPM, has many functions they being; feature recognition, feature representation,

feature selection, setup generation, and fmally creating process plans.

64
MRPII
Database

Bill Of
Material

Accounting

Database

Production
Control Routing

Shop Floor
Schedule
..
s
"CC
Final plans
Final plans Process
Scheduling Planning

Decision
Making
Module rocess planning criteria
Scheduling criteria L.,___ _ _ ___. Alternate Process plans
Available machines

Figure 4.5

Integration between Process Planning and Scheduling

65
To Cell Controller

-
,..
____.....,"""'
jManufactur en
::
~gResource en
&;
:l -· ..
...,Q.o .,Q'= ..-~-
::r
~
!Database Q
"0
........
"i:
0 ~
... ' ::!1 :;- ~ ;
::r -=
"i:
Q. .&:. 8., <ii"
-· =
.&:1 ..."' en
3" = :i"
Q
- Q.
....,Q =Q
IX
c;
c "' llO

"'

CAD Interface
Process Available machines Intelligent
Feature relationship Planning
Module
- scheduler

j
j >
<
!!:.
c; en
. ..
....... s:
n
"'c "'c
,!l
011 a'
"' ·-
Q.
c
s:
ii"
"'
::r

=? =·- c .. Q.

~
Vl s:~
-;
c
-;
c Q
c.,
=-
~-
... ~u ii: ii:
:c n (")

-~
Q.
.
2:
Q.
..___ Decision Making
Module
"' ;o
.,

:::!.

Matrix generation

Figure 4.6

Scheme for Integration

66
Process Sheet Manufacturing Cell Controller
1\_ ~~ource - -;...
CAD Information "'tabase /
1 Information from stored schedules
Feature Relation graph
"':I
3
Feature Recog- Feature
I ..
c.n

fi:
Intelligent Scheduler
- 8 ,-..,...-----____,,...l...-....,
1-'-'n:.:it:.:io::.:;n:..__ _ _-IRepresentatio ~ o._ Compute c
.c Performance
~ ug Criteria ~~
Select feature
~ Knowled ~ from
ge base __.. high priority job
~ ~ ~------~ ~
~~
..
i=""

·c:~ ~~~.
~

t lr Obtain Setups
based on
M/c capability 1
.....
Available Machines
.-------------1
0

·-g
~
..
~

·""5
;:'2
l;.c
~&:"1::1
Inference
Em~ine I ~
8
fi:,
Q. ~~~-------Jo.,
E~E
~~=
g_ee
8
All possible Process Plans _g Generate Dynamic jJ{.!)
c.n Rules
Process Planning Module

Available Machi~es

Decision Making Module

• Feasible Setups
Process Planning Criteria
-- • Setup vs Machine matrix

• Final Assignment
Final Plans

-
Final Plans

Figure 4.7

Integration of Process Planning and Scheduling Functions

67
As mentioned in Chapter I of this thesis, process planning is done in nine steps.

Though the sequence of the steps remain the same, the decisions about the resources

utilization is made as late as possible.

1. Interpretation of the design data. Interpretation of the design data forms the

first step of the IPPM model. The interpretation is mainly feature based. It is proposed

that the poning of the design data from the CAD to the PPM should be done based on

feature recognition. Literature review showed that immense work has been done in this

area. Either a expen-based system or a graph based system like attribute adjacency

graph(AAG) can be used to understand the feature. Using some of the above feature

recognition methods, features can be identified, and furthermore, their relationship with

other features can also be understood. It has been assumed that such a feature recognition

system is available and can be used for the model. Also, the process planning

knowledge stored in the process planning database will be helpful in recognizing the

feature interaction. The process planning database has information about feature

restrictions and feature interactions based on the knowledge of the process planner.

After recognizing the features in the pan, feature relationships in reference to the

manufacturing can be obtained by performing reverse manufacturing plarming. In reverse

manufacturing planning, features are generated as the planner moves from the fmished

pan to the raw material. As the planner moves from one stage to another, he removes

a immediate feature. This allows the planner to trace the feature relationships as being

either dependent or not dependent. As the planner moves from the finished stage to the

raw material stage, all the sub-features are recognized. For example, thread, chamfer,

68
etc, are sub-features while the diameter to which they are related is the main feature.

Reverse manufacturing planning helps in developing the Feature Relation Graph. which

is discussed in the following paragraph.

It is not just enough to recognize the feature interactions but it is very important

to represent the interactions accurately. IPPM depends on this representation to develop

alternative process plans quickly. In IPPM, the representation scheme is called as a

Feature Relation Graph (FRG). Figure 4.8 shows the Feature relation graph. In FRG.

all the relations the feature has with other features can be represented without ambiguity.

This representation actually helps in increasing the flexibility of the parts. Feature

relationships can be mutually exclusive or mutually dependent. In the current practice

of preparing process plans there is no means to represent the feature relationships.

thereby creating ambiguous relations. If the features are not related, i.e., they are

mutually exclusive, the features are termed as parallel features. If the features form

mutually dependent pairs, then they are called as collateral features. Collateral features

are found at different levels in the FRG, while parallel features can be found on the same

level. Sometimes in a process planning stage, alternative features can be specified,

which when manufactured form the same feature as the one originally required. Parallel

features can be manufactured at the same time without any restrictions, while collateral

features have to follow the technological sequence. While creating the process plans, all

parallel features lend flexibility to the system. Parallel features provide the alternatives

due to the fact that they can be processed independent of each other. As the features are

manufactured, they are deleted from the FRG, showing the features which needs to be

69
Finished Part

Figure 4.8

Feature Relation Graph

70
immediately processed. Interpretation of the design data is a very important function.

All feature representation is done in this step.

2. Selection of the machining process. Selection of machining process solely

depends on the technological consideration. In this step machining process like grinding,

milling, turning, drilling, etc., is selected.

3. Determination of fixtures and datum surfaces. Unlike the current method of

preparing process plan, IPPM performs the step of determining fixtures and datum

surfaces before selecting the machine tool. A dynamic list of machine tools and feedback

from the scheduling function informs the process knowledge base of PPM about the

available machines. Based on the capability of the machines, setups are created, using

the features from FRG. To create the setups, datum surfaces and fixtures have to be

decided. At this instant, all possible setups for a machine and for the high priority parts

are created.

4. Calculation of the overall time. Unlike the traditional method of preparing

process plans, time estimation is done at a very early stage. Cutting time can be

calculated based on some shop floor data estimation for the features. Predetermined time

standards is extensively used to estimate the time. This time calculation is used in

selection of machine tool which forms the next step of IPPM. Selection of datum and

fixtures provides information about alternative setups and overall time calculation gives

time information about the setups.

5. Selection of machine tool. Selection of the machine tool is the key function in

the integration of process planning and scheduling functions. In most of the current

71
practices, for every feature, the most desirable machine is selected based on the

machining proces. In IPP~l. the selection of the machine tool is done in a little different

m3flfter. The selection is first based on availability. As IPPM-aims at producing dynamic

proeess plans, availabilit)'-uf-m-aehine-too!s--play-a very important role in its selection.

In the section 4.4.2.4. the exact method of selection of machine tools is explained. The

selection of the machine tool must satisfy the technological requirement of process

planning function and also must satisfy the timing criteria of the scheduling function. All

machine tool selection in IPPM depends on the hamming distance calculation as explained

in the later section and the time to process a particular setup for a part.

6. Sequencing the operations. The fmal assignment of the machines to the setup

is done in the decision making module. This information is transferred to the PPM.

After receiving this information, detailed operation sequencing is done. In this step,

detailed planning of the cuts, feed, depth of cut, etc., is made.

7. Selection of the inspection device. Quality control procedure defines the type

of inspection device. Though not directly involved with the process planning function,

this step determines the fmal quality.

8. Determination of production tolerance. Production tolerances are calculated

based on the machine and process capability.

9. Determination of proper cutting conditions. In this step optimum cutting

conditions are calculated. Taylor's tool life equation, machining cost models, etc.,

guides the selection of the cutting condition.

72
10. Generating process sheets including NC data. After the decision about the

correct cutting conditions, NC programs, and other operation planning is performed,

and recorded in route sheets. This sheet is the execution order to the shop floor

operators. All operators follow the instruction on the route sheet.

PPM generates information about the different setups, and time to complete the

features in the setups. In tum, information about the available machines is supplied by

the scheduling function.

4.3.2 Scheduling Module

The scheduling module (SM) is also another expert system based module. It has

all the logic which any scheduler would have. Start and finish times of selected jobs are

calculated here. All system performance measurements are made in this module. Priority

rules for scheduling are changed based on the calculation of system performance, thereby

generating dynamic scheduling rules. This allows the planner to model the shop floor in

a more realistic terms. Performance of the shop floor can be enhanced by computing

rules individually for every machine and for every time window.

The decision-making module makes the fmal decision and informs the scheduler

and process planner. PPM performs the steps as mentioned earlier. The scheduling

module, based on the assignment, computes the start and fmish time for each

job/machine pair. This times are converted to material handling information based on the

setup time and other preparatory time.

73
4.3.3 Decision-Making Module

The decision making module (DMM) is the hean of the architecture proposed for

the integration of the two functions. This module consists of an algorithm to make

decisions. As IPPM proposes to use some fuzzy set operation to determine the setups,

and least processing time algorithm to assign the setup to the machine, all the required

logic for the computation is stored in this module. The decision-making module uses a

matrix based representation of the problem. Figure 4.9 shows the matrix and the

information they store. Matrix used in the DM is of the order m * n. In this matrix m

= n. M rows represent the scheduling information. being, the machines available. N

columns represent the selected setups for individual parts. This method of representation

allows both, process planning and scheduling information to be represented together.

The consequence, shown in the matrix is the time required by a setup on the machine.

4.4 Methodology

In this section, the methodology of integration will be explained. Functional

integration is achieved in the following steps.

Step 1: Decide on the value of time_window. The time_window is the period for

which plans are created. Figure 4.10 shows the concept of time_window. The span of

the time_window depends on the span for which the scheduler can predict the conditions

of the shop floor with confidence.

74
Information From Process Planning Function
Setup Selected for each job based on hamming distance
and its contribution to the Process Planning objectives

t
Alternate Process Plans

~ SUP 1 SUP2 - - - - - ~UPrr

Ml til - - - - - - t lm
Information
From - -
Scheduling
Function ~
c - -
:E
CJ

::;"' - -
41
:c
..!!
·;
... - -
<
- -
- -
Mn t nl - . . . . . tmn

Figure 4.9

Alternative Process Plan versus Machines Matrix

75
Run_time

Time_ window : Time for which scheduling is being done

Run-time : Time taken to run the code

Figure 4.10

Time Window

76
Time_ window is shifted as the planning is performed. IPPM performs all the

planning for only the time_ window in consideration. The main aim of IPPM is to make

plans to optimize the process planning and scheduling criteria for the time_ window being

considered.

Step 2: Understand the design data completely. Feature recognition algorithms

help in recognizing the features. As mentioned earlier either graph or expert system

based technique can be used.

Step 3: Develop Feature Relation Graph. All features in the pan have to be

represented in the graph. Features are interpreted in terms of its manufacturing

characteristics. In the FRG, all features are represented as a node. Every feature is

either connected to another feature or to the fmished product by means of a line with an

arrow head. The direction of the arrow head implies that "this feature when completed

will be followed by the feature at the end of the arrow head." Stan node of FRG is a raw

material stage and finish node is finished product. Features which can be manufactured

independent of each other are shown on the same level, while features which follow the

manufacture of another feature lies at a higher level. All sub-features, like chamfer,

threading, knurling, etc., form sequential features and have to represented at a higher

level than that of its parent feature. Features closer to the raw material are lower level

features and those closer to the finished product are higher level features. This

categorization is purely based on dependence of the features closer to the finished product

on the features closer to the raw material.

77
Step 4: This step involves creating alternative setups. FRG is extensively used in

creating the setups. First of all from the scheduler information about the available

machines is obtained. Setups are created based on the capability of the available

machines. Certain process planning objectives have to be satisfied while creating the

setups.

Some important objectives to be realized while creating setups are:

1. Minimize number of setups: This can be achieved by increasing the number

of features manufactured in a setup.

2. Minimize number of processing steps: Sometimes, by assigning jobs to less

desirable machine, number of processing steps is increased. This can lead to increased

cost, and scheduling problems.

3. Improve machining accuracy: Machining accuracy is immensely improved by

machining features with tolerances in the same setup. This can lead to setup restrictions,

reducing the flexibility.

4.4.1 Rules for creating setups

Some rules that need to be followed while creating the setups using the FRG are:

1. Select the first available machine

2. Start with the features closest to the raw material

3. Add as many features as possible in a setup based on the capability of the

machine, maintaining the restrictions as shown in the FRG

78
-+. Parallel features can be added in the same setup or can be used in a different
setup

5. While creating setups, try to add as many higher level features as possible.

6. Categorize the setups as Collateral setup; machining features which are

collateral in a setup. Sequential setup; machining features are sequential in the setup;

or Restriction setup; setups machining features which have feature restrictions. All these

setups are prioritized based on its contribution to a particular objective. Prioritization also

considers some intangible benefits. A sequential setup contributes extensively to the

reduction of setups. Furthermore, a sequential setup processes dependent features,

thereby reducing the restrictions from the FRG.

7. Obtain all the possible setups for every machine for as many parts as available

machines. It is very important to note that, only one setup can be assigned to one

machine at any instant. As the planning is being done for a particular time_window,

number of setups that can be scheduled must be equal to the number of available

machines.

Step 5: This step is performed by the decision making module. In this step the

problem is represented on a matrix as mentioned earlier. M * N ordered matrix is used


to represent the problem. Rows represent the available machines, which is obtained from

the scheduler. The available machines are those machines which will be unloaded in the

time_window for which the planning is being performed. Columns represent the feasible

setups. One setup for each part is selected so that the selected setups have the least

difference, reducing the scheduling problem.

79
When similar process plans are selected to manufacture products, similar machine

tools. jigs. fixtures. material handling equipment, etc.. is used. This simplifies the

planning procedure.

To find the feasible process plans, an algorithm usmg fuzzy set operation 1s

suggested. As mentioned earlier, there is a lot of impreciseness in the information

regarding the contribution of a particular setup to the objectives. This impreciseness lends

itself very well to fuzzy modelling. All setups are expressed in terms of their degree of

membership to a particular set. The set may be a collection of setups and their

contribution to the objective of reduction of setups.

4.4.2 Heuristic Algorithm to select feasible plans

This algorithm is applied once for every time_window

~ = Available machines; i = l, .... m

Ji = Set of high priority jobs that need to be scheduled; i = 1,.... n

Fij = Set of features in job J; j = 1,... n, i = 1,... k

Si = Setups for job J; i = 1, ... n

ASi = Setups for all jobs; i = 1,.... n

Oi = Set of features in the ith setup

A setup can be defmed as a set of 3 tuples

sij = {~. Ti, fi, t}

Where

~ = Machines required to machine this setup


Ti =Tools required to machine this setup

80
fi = Fixtures required to machine this setup

t = Type of the setup; COLlateral, SEQuential, REStricted.

The main aim of this step is to select setups which have the least difference in terms of

resource utilization. As mentioned earlier a setup is a combination of machines, tools,

and fixtures. Furthermore, the type of the setup decides on the priority of its selection.

For, V ASi, iEN.

The following column matrix can be defined;

Here, l, .. a = all fixtures used in ASi

b, .. c = all tools used in ASi

d, .. e = all available machines used in ASi

f = type of the setup.

X 1i = 1, if a tool t is used in ASi

0, if a tool t is not used in ASi.

xti = 1, if a fixture f is used in ASi

0, if a fixture f is not used in ASi

.xm = [0,1] the value lies between 0 and 1.

This value indicates the desirability of using machine m to process setup ASi. The

value depends upon the features being manufactured in the setup. Every feature will have

a degree of membership to a machine.

81
Based on the combination of the desirability of the features in this setup. a

resultant desirability to machine this setup on the available machine can be calculated.

Let:

mj be a fuzzy subset of all the machines available on the shop floor

fi be a set of features that can be machined on the shop floor

Then

is the degree of membership of ith feature in jth machine.

Hence, the resultant membership of a setup to the machine can be given by;

Three process plarming objectives are considered in IPPM. They being;

Z 1 = Minimize number of setups

~ = Minimize number of processing steps

Z3 = Improve machine accuracy

All setups are assigned a membership number for each one of the objective based

on the process planners judgment. Hence ~; i = 1,2,3 form three fuzzy subset

representing the three objectives. Membership of the setups to these subsets is given by;

82
If G is a fuzzy subset containing the effective contribution of the setups to the

three objectives, then the membership function is given by:

Hamming distance is the difference in resource utilization of the two setups. ASi

and ASj are the setups of two nonidentical parts. then, hamming distance (1"\j) is;

4.4.2.1 Algorithm to select the feasible setups

(a) Nk is the set of setups for job k. Hence, there are N" set of setups. That is

set of s 1, ..... sm forms set Nk such that;

(b) The main of IPPM is to construct a set S = { si}; such that

As the algorithm moves, at an intermediate stage, for a setup k,

To select these setups following algorithm is proposed;

Stepl:

At the start zero setups have been selected, hence;

sco> = ct>; k = 1.

Here k is a counter to check if all the high priority have been considered

83
Step 2:

In this step, initial entry into the setS is made. Unlike other methods of selecting

setups, it is proposed to select the setup solely based on its contribution to the objective,

not considering the time factor at all.

Hence, the contribution Ci is given by;

where,

-
h =1 - h
IS IS

Let ~be the setup with maximum contribution, that is

Step 3:

Add the above setup to the feasible set s<k) = s<k-l) + { ~}

If

k = n, then exit. s(n) = s is the feasible set.

Else

k = k++ and go to Step 2.

In this phase one setup is selected such that;

84
from a set of setups Nk for a job Jk. This method leaves all the previous selected setupsas

it is and proceeds to select one set from next job. In the above algorithm, setup entering

the initial feasible set, is the one with highest value of either ( 1) membership to fuzzy

subset G or (2) similarity between the kth setup being considered and s setups in the

feasible set s(kl.

As the hamming distance defmes the measure of dissimilarity,

-
hij

is the measure of similarity.

The next phase of the algorithm is to optimize the initial feasible set by

exchanging the setups within the set Nk. Every entry within S = { si}, is checked and

replaced with another setup which has a higher contribution than the one in the set.

4.4.2.2 Optimization Phase

Step 1: Make counter = 0; k = 1

Counter shows if any change in the entry of set S has been made. k is check if

all high priority parts have been checked.

Step 2: Exchanging within set Nk.

Compute contribution of all the setup in the set Nk with respect to the entries in

85
If the maximum contribution occurs in a

S;t!S'tliENk

then replace si with sk. The new set will be;

s = s- {sk} + {si} and make c = c++


Step 3: If k = n

and if c = 0 then

stop

else

repeat from step 1.

else if

k < n then

k = k++ and repeat from step 1.

The optimization algorithm results in a set which has setups with a minimum

difference and which contributes maximum to the objective. In the next chapter, an

example problem has been solved to explain the algorithm.

86
4.4.2.3 Matrix Representation of the Problem

After the set S has been computed. the next phase is to represent the information

in a matrix form so that the fmal assignment can be made. For this, the information

form the scheduler is also required, regarding the available machines. The problem itself

can be written as follows.

Objective:

n
MINE t IJ.. xxIJ.. < oo
j•l

Subjected to:

xij = 1; if ith machine is assigned to the jth setup

= 0; if not (1)

n
L xv= 1; "V j
i•l

(2)

m
E xij = 1; "V i.
j•l

(3)

87
The objective of this problem is to minimize the total assignment time such that

the individual assignment will be minimized. By making the left hand lesser than infinity

no assignment is made to the box which has no fmite processing time. An entry without

a finite processing time indicates that the setup cannot be processed on the specific

machine corresponding to the row. Constraint 1 puts a 1 in assigned boxes so that the

processing time may be added up. Constraints 2 and 3 ensure that all the setups and

machines are assigned, and only one assignment can be made in either a row or column.

This problem can be solved by classical solution methods like linear programming, etc.,

but due to difficulty is achieving solutions quickly, a heuristic is proposed to solve this

problem.

4.4.2.4 Heuristic to assign the setup to the machine

A simple heuristic to assign the setup to the machine based on least processing

time is as follows:

Step 1. From the matrix select the least tii"

Step 2. Assign ith machine to the jth setup,

Step 3. Remove the assigned setup and machine from the setup.

Step 4. If all the setups and machines are assigned;

Stop

Else proceed to step 5

Step 5. If there is a entry with Infinity which has to be assigned

go to Step 6

Else

88
go to step 1

Step 6. From the corresponding row or column which has the processing time equal to

infinity, select the least processing time. Cancel all the previous assignments, that is

enter in all the machines and setups into the matrix, and go to Step 2.

This heuristic assigns the setup to the machine based on the least processing time

form the matrix. This procedure of assignment has been found extremely good in most

occasions. In summary the steps involved in creating the matrix will be shown briefly

as in Figure 4.11.

As stated earlier, we will try to solve a example problem using all the above

algorithm and see how it performs.

89
CAD Data Feature Feature Recognition
Recognised Algorithm

Feature
Relation
Graph
' Feature
Relationship
~
~
"'...
CCI
CG
......
c
c
s:"'
"2 "'~
_I Knowledge/
i0 l5...
c c.
Alternate Tolerance ::..::
Available Machine analvsis to
Tools & Capabthty get set-up
Set-Ups

Fuzzy Ooeration Set-Up Process Planning


Vs Machine
Hamming Distance Matrix lJbjectives

-' CG
0 ptimize Scheduling Initial Shortest Processing c
Assignment
:.;: ...
:;
Objectives Time Rule
::;"' ~
0
c ::;

'
0
·;;;
Perform -~
Shirt Opera- Q
Balanced Load, tion

Final
'
Assignment
_.
lnlormataon'
to P.P &
Scheduler
c... :;......
Calculation
of start &
finish time
'
t;xecuhon or
the plan by
Scheduler
~ ~
....
.s &.
"'
(/)

Figure 4.11

Flow Diagram of the Logic Used for Integration

90
CHAPTER V

SOLUTION METHODOLOGY

This chapter mainly contains the solution to the proposed algoritluns in the

previous chapter. All proposed algorithms have been applied to obtain numeric solutions.

All the data used in this chapters are assumed for the solution purpose. Finally, solving

the matrix, an assign111ent of setup and machines are shown.

5.1 Generating Feature Relation Graph

A simple part has been considered to explain the methodology of generating the

Feature Relation Graph. FRG generation involves a lot of data transfer from the CAD

package to the IPPM modules. As explained in the previous chapter, an efficient

algorithm is required to understand the features from the information received form the

CAD package. Attributed adjacency graphs, expert rules, and many other feature

recognition algorithm can be used. Furthermore, algorithms are required to understand

the feature relationships after the features have been recognized. Generation of FRG

depends solely on this ability. As this topic is beyond the scope of this research, this

areas has been left to the future researchers, while the possibility of such an approach

has been exploited in this research.

Many orders are received on the shop floor. Every order has a due date. This due

date can be used to compute the priority of the orders. The priority rule normally

followed will be Slack time/ operation.

Slack Time Per Operation= due_date- time_now I number_of_features_to_be_machined

91
The priority of the parts keeps changing as the features are completed. This

priority has to be computed for every time window. A dynamic list has to be maintained

which updates every job that is unloaded from the shop floor. Another dynamic list is

maintained to update the machines available in the time_window. Using the capabilities

of the machine, setups are created according to the rules mentioned in the previous

chapter. Figure 5.1 shows one such simple part which needs to be scheduled on the shop

floor. Figure 5.2 shows how all the possible setups can be created.

5.2 Solution

For the purpose of this chapter, we will assume that there are three parts which

need to be scheduled. The method of obtaining the setups will be the same as that used

in the example part.

Let us assume that parts P 1 , P2 , P 3 are three parts which have the highest priority,

and needs to be scheduled. And let

be the possible setup for the corresponding parts.

Each set of the setups for the parts can be ordered in the set:

92
Diameter 2 Chamfer 2
Chamfer 1 Diameter 1 Keyway Diameter 3
/ / I
1

f12 / R8 '
• --
M30
30 25
so r !\
~~~
/ . '/ / / 1/

Jl \. .J
I ~

1
~ e 2 \ /Th
Hole fGroov 1

--
e~d

--
40 Gr
2x4S
~s ---- ~--- 30
2f4- 2x4S
---- r.--
45 45
175

Figure 5.1

Example Part

93
Final Part

/\ \ /'
THRD ~

v
CHMFI GROVI KEYWY CHMF2 GROV2 ~

HOLE DIAMI
I
DIAM2 v
DIAM3
~

\ Raw Material

Feature Relationship Graph

SUPl= DIAMI+CHMFl+GROVl+DIAM2+HOL SUPl:HOLE+CHAMF


SUP2=DIAM3+CHMF2+GROV2+DIAM2+THRD SUP2=HOLE
SUP3=KEYWY SUP3:DIAMl+DIAM2+DIAM3+GROVl+
GROV2+CHAMF1+CHAMF2+
THRD
SUP4=KEYWY

Figure 5.2

Possible Setups

94
Let the set of features that needs to be machined be given by;

f 1 , f2 , f3 , f-l, f5 , f6 , f7 , f8 .

Every setup mentioned above is a combination of features. The features are

obtained from the FRG and added as per the rules mentioned in the previous chapter.

The feature combination of the setups is as follows;

s 1 = f 1 + f 2 + f3

Sz = f:! + f4

S3 = f 1 + f3 + f.l

s4 = fs + f7

Ss = f6 + f7

S6 = f4 + f5 + f7

s7 = f6 + f 7 + f8

S8 = f7 + f8 .

Every feature has a degree of desirability for being machined on a particular

machine. This desirability can be represented in the form of a membership to the

machine subset. The desirability depends on a lot of factors. The factors can range from

setup requirement to the accuracy that can be achieved by machining a feature on the

machine. Most of the times this membership is dependent on the process planner's

experience on the shop floor. These memberships basically show the desirability of

processing a particular feature on the available machine. As the setups are combination

of the features, a desirability function can be derived, when a particular setup is used

on the machine. This calculation is shown later.

95
The feature membership function to the machine is given below:

mt = lf/.8. f~/.2, fi.7. fj.6, [ 5/.3, fJ.9, f,/1, fJ.ll

m2= lftf-3, f2/.l' f3/.5, fJ.6, f5/.9, fJ.7. f,/.4, fJ.2l

m3 = lftf-9, f2/.8, f3/.4, fj.5, f5/.3, fJ.2, f,/.8, fJ.9l

The membership function of the setup can be calculated by fmding the least

membership of the combination of features in the setup.

J.L (AS2) = MIN (jJ.l, JJ.6) = .1


ml

J.L"'J (AS2 ) = MIN (jJ.8, JJ.5) = .5

96
IJ.m 1 (AS~;) = MIN ifJ.9, f.,/1) = .9

1-1mz(AS5) = MIN ifJ.7, f.,/.4) = .4

1-lm 1(ASJ = MIN ifJ.6, /J.3, / 7/1) = .3

IJ."'l(ASJ = MIN ifJ.S, /J.3, .f-,/.8) = .3

IJ.m (AS1) = MIN ifJ.9, /,f.1, /pj.l) = .1


1

IJ."'l(AS7) = MIN ifJ.2, / 7/.8, /p).9) = .2

97
~~ (AS8) = MIN (/1/.4, /sf.2) = .2

The above equations give the membership of the setup to the machine. This

measure of this membership allows one to compare the setups and select the best. Most

of the time it can be seen that the setups may have a small processing time but contribute

very little to the objectives itself. The desirability of the setups gives a measure of the

performance of the machine while processing the setups. Hence, the algorithm will tend

to select setups which have a high desirability factor. Let Table 5.1 show the resource

utilization of the setups for the various parts.

98
Table 5.1: Resource Utilization Table

m3 .4 .5 .4 0 0 .3 .2 .8

m2 .1 .1 .3 .4 .4 0 0 .2

mt .2 .2 0 .3 .9 .3 .1 .1

f2 1 1 1 1 1 1 1 0

fl 0 1 1 0 0 0 0 1

t3 0 1 1 0 1 1 1 0

I
~ 1 0 1 1 0 0 1 1

tl 1 0 0 1 1 1 1 0
I

I st s2 s3 s4 Ss s6 s7 Sg

The next step is to calculate relative hamming distance . Using the formula given

in the previous chapter, hamming distance can be computed and stored in a matrix

format . Relative hamming distance will assure that the calculations will lie between 0

and 1. This allows comparison with other measures on the same scale.

Hamming distance calculations can be performed in the following manner:

h14 = 1/8 (0 + 0 + 0 + 0 + 0 + .1 + .3 + .4) = .8/8 = .1


ht2 = 1/8 (0 + 1 + 1 + 0 + 0 + .7 + .3 + .4) = 3.4/8 = .42.
Further hamming distance calculation can be done in similar manner.

99
Table 5.2 shows the relative hamming distances between the setups.

Table 5.2 : Relative hamming distance ~1

1 2 3 4 5 6 7 8

I 1 inf inf inf .10 .42 .28 .17 .45

2 inf inf .55 .75 .31 .42 .5

3 inf .48 .56 .45 .31 .31 ,I

4 inf inf .33 .22 .52


!

I
5 inf .16 .30 .85
II

6 inf inf inf II


II
II
7 inf inf I!

8 inf

Every setup is given a weight factor based on its type . The type depends on

whether, it is a RES, SEQ, or COL. Due to its contribution to the relaxing constraints

from the FRG, they are prioritized and used to compute the contribution. The weight

given to RES type setups is 1, SEQ = .5, and COL= .3 .

In the example, following weights are given;

Wi = {s1 = 1, s2 = .3, s3 = .5, s4 = .3, s5 = .3, s6 = .5, s7 = .5, s8 = .5} .

100
For the purpose of the solution we will consider three objectives for process

planning. The objectives have been explained in the previous chapter. By creating a

fuzzy subset for each one of the objectives, a measure of contribution of each setup to

the objective can be calculated. Again, this measure is given by a membership function.

A membership function can be derived either by fitting a curve equation or by

considering the process planner's experience. It is suggested that for the purpose of a

more realistic contribution, process planners should be suggested for the obvious reasons.

By getting an individual membership of setups for each one of the objectives, a

combined or resultant contribution can be obtained.

Membership of setups to the three objective subsets:

Z 1={ Stf-6, S2/.5 S3/.7, SJ.5, S5/.2, SJ.9, S7/.4,SJ1

~={ Stf-3, S2/.4 S 3/.8, Sj.8, S5/ 1, SJ.8, Si.9, Sg/.1}

Z 3 ={ Stf-2, S2/.4 Si.8, SJ 1, S5/.8, SJ.5, Si.8,SJ5 }

Jla(St) = Wi * Min { Stf.6,Stf.3, Stf.2}

= 1 * .2 = .2.

All other resultant memberships can be calculated based on the above equation.

By multiplying the weight factor with the total contribution, the setups are prioritized

based on their types, and will enter the solution in that order.

.2 . 12 .35 .15 .06 .25 .20 .05 .

101
Once the contribution of the setups is calculated, we can proceed with the

algorithm to select the initial feasible entries. As shown in the previous chapter, this

algorithm is done in two phases. The first phase makes the initial selections, and the

second one optimizes the selection in the set.

5.2.1 Initial Feasible Set

An algorithm to seleci initial feasible setups into a set is as follows:

Iteration 1:

Step 1: S(Oi = <1>; k = 1


Step 2: As 1"\s = oo, compare all J.lG(si)

J.lG(s 1) = .2; J.lG(s 2 ) = .12; J.lG(s 3) = .35

Step 3: s<ll = { s3 } is selected; k = 2.

Iteration 2:

Step 2: Compare Contribution & hamming distance

C4 =MIN{ .3, .513 } = .3

C5 = MIN { .06, .44 } = .06

MAX ci = c4
Step 3:

Iteration 3:

Step 2: C6 =:MINt .25, .55, .67 } = .25

102
c7 = WN t .2o. .69 .. 78 } = .2o

Cg = l\.1IN { .05, .69 .. 48 } = .48

Step 3:

Hence the initial feasible set is given by:

5.2.2 Optimization Algorithm

After the initial feasible set has been obtained, the optimization procedure gives

the final set.

Iteration 1:

Step 1: Counter= 0 ; K = 1.

Step 2: For every setup compare the contribution and hamming distance

cl = Iv1IN{ .2, .9, .72 } = .2

C 2 = Iv1IN{ .12, .45, .69 } = .12

C3 = Iv1IN t .35, .52, .55 } = .35

C 3 remains the maximum contribution hence s3 remains in the set.

Step 3: Set remains unchanged.

Iteration 2:

Step 2: K = 2

c4 = Iv1IN{ .15, .52, .67 } = .15

C5 = l\.1IN { .06, .44, .84 } = .06.

103
C4 is the maximum contribution and hence s4 remains in the set

Step 3: Set remains unchanged.

Iteration 3:

Step 2:

C6 =MIN{ .25, .55 .. 67 } = .25

c:! =MIN{ .20 •. 69, .78 } = .2o

C3 = MIN { .05, .69, .48 } = .05.

Step 3:

Set remains unchanged.

Therefore the final optimized set is:

S = ( S3 , S4 , S5 }

5.2.3 Matrix Solution

The last phase of the IPPM is the assignment of setup to the machine. Once the

optimized set of setups are found, available machines and setups can be represented on

a matrix and solved according to the algorithm proposed in the previous chapter.

104
Table 5.3 gives the processing time for the selected setups and available machines.

Table 5.3: Final Matrix

s, s~ s6

ml inf 5.6 -+.1

m., 4.1 9.1 inf

m3 5.3 5.4 3.2

Hence the fmal assignment is:

with L.tij = 12.9 time units.

This assignment is sent to the process planning and scheduling functions where

the route sheet and time calculations will be performed.

105
CHAPTER VI

CONCLUSION

The problem of integration between process planning and scheduling is a topic

which still needs extensive attention. It is difficult to solve process planning or scheduling

problems by some classical operation research techniques. The main problem that needs

to be solved to integrate the two functions is the problem of information integrity, and

objective integrity. Both, process planning and scheduling functions are resource

allocators. The problem of allocation is the prioritization of objectives. If the two

objectives can be combined sufficiently to make the two functions as a whole, plans with

high degree of feasibility can be made.

Main contributions from this research:

In this research work an attempt has been made to venture into areas like dynamic

process planning and scheduling. Unlike classical methods, here plans are made as

flexible as possible, thereby reducing constraints on scheduling. By using a matrix based

representation , the problem can be clearly represented simultaneously in most of the

current planning system, the problem of assignment is dealt at different levels and hence

show a lack of integrity in dealing with it. Also, the problem of integration is not just

assignment of setups to the machine, but selecting the right setups.

106
Selecting feasible setups from a set of possible setups is the key task in

integration. As the criteria for selection is many, a multi-criteria selection technique has

been used. Furthermore, a fuzzy based modelling of objectives allows the planner to

consider all the impreciseness.

Features form a membership function to the machine subset. The algorithm

considers the combined membership and features to the setup and hence the membership

of setup to the machine. This gives a better measure to select the setup into the feasible

set. Results from the algoritlun can be interpreted as a assignment which considers the

process planning objective and scheduling objective simultaneously.

Contributions of the proposed IPPM model are as following:

1. Setups requiring similar resources are selected. This reduces the scheduling

problem and reduces the dissimilarity on the shop floor. Similar setups are selected by

calculating the hamming distances. Process plans which require similar resources will

have a relatively smaller hamming distance from the ones which use dissimilar resources.

2. Prioritization of the setups allows to consider the setups based on their

contribution to the reduction of the constraints from the FRG. The prioritization 1s

considered in the algoritlun by giving a weight factor to the setups.

3. Using fuzzy membership to represent the impreciseness of information helps

in considering multiple criterion.

4. All setups are selected based on their contribution to the objectives and not

based wholly on timing criteria. This follows the principle of selecting the feasible setups

and then optimizing on them without compromising on performance.

107
5. Matrix solution algorithm has been proved to be near optimal in most of the

literature.

6. Planning can be made very flexible and almost realistically Just_in_time.

7. Matrix based problem representation allows the simultaneous consideration of

both process planning and scheduling functions.

8. Due to the simultaneous considertion of the problem all resource allocations

are made with proper consideration of both process planning and and scheduling

objectives.

9. As IPPM uses the matrix to ass1gn machines to setups, only available

resources are considered giving due care to the dynamic conditions of the shopfloor.

10. Feature Relation Graph used in the IPPM model is a very suitable method to

represent feature relationships of a feature. FRG can be used to create setups and also

monitor the progress of a part. As every time a feature is completed it is taken off from

the FRG, it forms an effective control method.

This research was carried out to give a conceptual solution to the problem of

integration of process planning and scheduling functions. The main idea was to address

the problem and see if there are possibilities of real time integration. The proposed IPPM

model can be further developed and coded so that it can be applied to some simulated

condition.

In summary, after reviewing the literature and working on this model, it was felt

that the main problem lies in considering the two functions (process planning and

scheduling) as a whole. The measure used by the two functions are separate, and hence

108
methods have to be devised so that the two objectives can be treated in the same domain.

The proposed model can be ideally used in a job shop type of environment. In this

environment, job orders enter the system in a random order, and plans have to be

modified based on the incoming orders. As job shop environment is a small quantity type

of operation, every pan which enters the physical production system can be treated as

a new part in itself. This allows the flexibility to change process plans for every job and

hence make the shop floor as flexible as possible. Hence, it can be seen that for the

same job, every time it enters the production system, it may follow a completely

different routing.

Some suggestions for future work are:

1. Building the form feature recognition system: A general form feature algorithm

should be developed based on the existing algorithms. Some features are difficult to

represent graphically and some are difficult to be explained in terms of rules. If a correct

blend can be achieved in feature representation and recognition schemes, data from CAD

can be easily and automatically ported into CAPP.

2. Building proper algorithm to generate Feature Relation Graph: The concept of

FRG proposed in this paper is at a very elementary stage of development. Formal rules

can be created based on the process planners and feature recognition system to generate

the feature hierarchy. This graph generation can be made automatic, with some

interactions with the process planner.

109
Creating a FRG has to simply follow the reverse boolean operation on the raw

material. Once the features are recognized from the CAD, these features can be either

added or subtracted from the raw material in a logical pattern to give the graph.

3. Expanding Process plannin~ knowledge base: A vast amount of knowledge has

to be stored in the knowledge base if this model has to be implemented. Knowledge

about feature/machine membership, setup generation rules, etc.

4. Building the model itself: The main of this research was to first of all identify

the areas which can be used for integration. Algorithm development and proving the

algorithm formed the main area of work. Implementation of the model is very important

to find out if the logic used is correct or otherwise. Hence, the most important work

needed to be done is that of creating the model and preparing the required computer

codes.

110
BIBLIOGRAPHY

Alting, L., and Zhang, H. C., 1989, Computer Aided Process Planning : The
State of the Art Survey, International Journal of Production Research,
Vol.27, No.4, pp. 553-585.

Bhaskaran. K., 1990, Process Plan Selection, International Journal of Production


Research, Vol.28, No.8, pp. 1527-1539.

Chang, T. C., Anderson, D. C., and Mitchell 0. R.. 1988, QTC- An Integrated
Design/Manufacturing/Inspection System for Prismatic Parts, Proceedings
of the 1988 ASME Computers in Engineering Conference, August 1-4.

Chang, T. C., and Wysk, R. A., 1981, An Integrated CAD/Automated Process


Planning System, AilE Transactions, Vol.13, No.3, pp. 223-233.

Chang, T. C., and Wysk, R. A., 1984, Integrating CAD and CAM through
Automated Process Planning, International Journal of Production Research,
Vol.22, No.5, pp. 877-894.

Chang, T. C., and Wysk, R. A., 1983, CAD/Generative Process Planning with
TIPPS, Journal of Manufacturing Systems, Vol.2, No.1, p. 223-233.

Chryssolouris, G., Chan, S., and Cobb,W., 1984, Decision Making on the
Factory Floor: An Integrated Approach to Process Planning and
Scheduling, Robotics and Computer-Integrated Manufacturing, Vol.l,
No.3/4, pp. 315-319.

Chryssolouris, G., and Chan, S., 1985, An Integrated Approach to Process


Planning and Scheduling, Annals of the CIRP, Vol.34/l/1985, pp. 413-
417.

Chryssolouris, G, and Gruenig, I., Process Planning Interface for Intelligent


Manufacturing Systems, 19th CIRP, pp. 181-188.

Chu, C. H., and Hayya, J. C., 1991, A Fuzzy Clustering Approach to


Manufacturing Cell Formation, International Journal of Production
Research, Vo1.29, No.7, pp. 1475-1487.

Dubois, D., and Prade, H., 1978, Operation on Fuzzy Numbers, International
Journal of System Science, Vo1.9, p. 613.

111
Dubois, D., Farreny, H., and Prade, H., 1986, Towards the Use of Fuzzy Rule-
Based Systems in the Monitoring of Manufacturing Systems, Software for
discrete Manufacturing, Elsevier Science Publishers B. V. (North-Holland).

Dyck, F. V., Tseng, M. M., and Lascoe, 0. D., 1976, A study of Computerized
Process Planning with Continuous and Discrete Data Base, AilE
Transactions, Vol.8, No.3, pp. 320-327.

Gindy, N. N. Z., 1989, A Hierarchical Structure for Form Features,


International Journal of Production Research, Vol.27, No.12. pp. 2089-
2103.

Graves, G. R., Yelamanchili, B., and Parks, C. M., An Interface Architecture


for CAD/CAPP Integration Using Knowledge Based Systems and Feature
Recognition Algorithms, International Journal of Computer Integrated
Manufacturing, Vol.1, No.2, pp. 89-100.

Ham, 1., Lu, S.C. Y., 1988, Computer-Aided Process Planning: The Present and
the Future, CIRP Annals, Vol. 37/2/1988.

Hayes, C., and Wright, P., 1988, Automating Process Planning: Using Feature
Interactions to Guide Search, Journal of Manufacturing Systems, Vol.8,
No.1, pp. 1-14.

Joneja, A., and Chang, T. C., 1991, Search Anatomy in Feature-Based


Automated Process Planning, Journal of Design and Manufacturing,
Vol.l, pp. 7-15.

Joshi, S., and Chang, T. C., 1988, Graph-based heuristics for Recognition of
Machined Features from a 3D Solid Model, Vol.20, No.2, pp 49-66.

Kanumury, M., and Chang, T.C., 1991, Process Planning in an Automated


Manufacturing Envirorunent, Journal of Manufacturing System, Vol.10,
No.1, pp. 67-78.

Karinthi, R. R., and Nau, D. S., 1989, Geometric Reasoning as a Guide to


Process Planning, Computers In Engineering, pp. 609-616.

Khoshnevis, B., and Chen, Q., 1989, Integration of Process Plarming and
Scheduling Functions, liE Integrated Systems Conference and Society for
Integrated Manufacturing Conference Proceedings, pp. 415-420.

112
Korhonen, P., Wallenius. J., and Duckstein. L., Multi Objective Linear
Programming over a Fuzzy Feasible Set, Applications of Fuzzy Set
Methodologies in Industrial Engineering. pp. 225-235.

Liang. G. S., and Wang, M. J. J., 1991, A Fuzzy Multi-criteria Decision Making
Method for Facility Site Selection, International Journal of Production
Research, Vo1.29, No.11, pp. 2313-2330.

Managaki, M., Kawagoe, M., and Naniwada, M., 1984, A Model and its
Implementation in Practical CAD/CAM Database, Computes in Industry,
No.4, pp. 319-327.

Prade, H., 1979, Using Fuzzy Set Theory in a Scheduling Problem: A Case
Study, Fuzzy Sets and Systems, Vol.2, pp. 153-165.

Rahman, M., and Narayanan, V., 1987, An Expert System for Process Planning,
Robotics and Computer Integrated Manufacturing, Vol.3, No.3, pp. 365-
372.

Singh, N., and Aneja, Y.P., 1990, Multi-objective Modelling and Analysis of
Process Planning in a Manufacturing System, International Journal of
System Science, Vol.24, pp. 621.

Singh, N., and Mohanty, B. K., 1991, A Fuzzy Approach to Multi-objective


Routing Problem with Applications to Process Planning in Manufacturing
Systems, International Journal of Production Research, Vol.29, No.6,
pp. 1161-1170.

Steudel, H. J., 1984, Computer Aided Process Planning: Past, Present, and the
Future, International Journal of Production Research, February, pp.253-
266.

Subramanyam, S., and Lu, R., 1987, A Characterization of the Process Planning
Task from an Artificial Intelligence Perspective, 19th CIRP Seminar.

Sundaram, R. M., and Cheng, T.J., 1986, Micro-computer Based Process


Planning using Geometric Programming, International Journal of
Production Research, January, pp.119-127.

Sundaram, R. M., and Fu, 1988, Process Planning and Scheduling- A Method
of Integration for productivity Improvement, Computers In Industrial
Engineering, Vol.l5, No.l-4, pp.296-301.

113
Tonshoff, H.K.. Beckendorff, U.. and Andres, N., 1989, FLEXPLAN- A
concept for Intelligent Process Planning and Scheduling, lnstiture for
Production Engineering and Machine Tools, University of Hannover.
Denmark.

\\'ang, H. P.. and Wysk, R. A., 1986, Applications of Microcomputers in


Automated process Planning, Journal of Manufacturing System, Vol.5,
No.2. pp.l 03-111.

114

You might also like