30079_31.pdf

CHAPTER 31

CLASSIFICATION SYSTEMS

Dell K. Allen

Manufacturing Engineering Department

Retired from Brigham Young University

Provo, Utah

31.1 PART FAMILY

CLASSIFICATION

AND CODING

31.3.3 Process Taxonomy

970

31.3.4 Process Code

973

31.3.5 Process Capabilities

973

951

31.1.1 Introduction

951

31.1.2 Application

952

31.4 FABRICATION EQUIPMENT

CLASSIFICATION

31.1.3 Classification Theory

954

974

31.4.1 Introduction

974

31.1.4 Part Family Code

955

31.1.5 Tailoring the System

962

31.4.2 Standard and Special

Equipment 976

31.4.3 Equipment Classification 976

31.4.4 Equipment Code

31.2 ENGINEERING MATERIALS

TAXONOMY 962

31.2.1 Introduction 962

31.2.2 Material Classification 962

31.2.3 Material Code 964

31.2.4 Material Properties 965

31.2.5 Material Availability 966

31.2.6 Material Processability 966

977

31.4.5 Equipment Specification

Sheets

978

31.5 FABRICATION TOOL

CLASSIFICATION AND

CODING

981

31.5.1 Introduction

981

31.3 FABRICATION PROCESS

TAXONOMY

31.5.2 Standard and Special

Tooling 982

31.5.3 Tooling Taxonomy 982

31.5.4 Tool Coding 982

31.5.5 Tool Specification Sheets 984

967

31.3.1 Introduction

967

31.3.2 Process Divisions

969

31.1 PART FAMILY CLASSIFICATION AND CODING

31.1.1 Introduction

History

Classification and coding practices are as old as the human race. They were used by Adam, as

recorded in the Bible, to classify and name plants and animals, by Aristotle to identify basic elements

of the earth, and in more modern times to classify concepts, books, and documents. But the classi-

fication and coding of manufactured pieceparts is relatively new. Early pioneers associated with

workpiece classification are Mitrafanov of the USSR, Gombinski and Brisch, both of the United

Kingdom, and Opitz of Germany. In addition, there are many who have espoused the principles

developed by these men, adapted them and enlarged upon them, and created comprehensive workpiece

classification systems. It is reported that over 100 such classification systems have been created

specifically for machined parts, others for castings or forgings, and still others for sheet metal parts,

and so on. In the United States there have been several workpiece classification systems commercially

developed and used, and a large number of proprietary systems created for specific companies.

Why are there so many different part-classification systems? In attempting to answer this question,

it should be pointed out that different workpiece classification systems were initially developed for

Mechanical Engineers' Handbook, 2nd ed., Edited by Myer Kutz.

different purposes. For example, Mitrafanov apparently developed his system to aid in formulating

group production cells and in facilitating the design of standard tooling packages; Opitz developed

his system for ascertaining the workpiece shape/size distribution to aid in designing suitable pro-

duction equipment. The Brisch system was developed to assist in design retrieval. More recent sys-

tems are production-oriented.

Thus, the intended application perceived by those who have developed workpiece classification

systems has been a major factor in their proliferation. Another significant factor has been personal

preferences in identification of attributes and relationships. Few system developers totally agree as

to what should or should not be the basis of classification. For example: Is it better to classify a

workpiece by function as "standard" or "special" or by geometry as "rotational" or "non-

rotational"? Either of these choices makes a significant impact on how a classification system will

be developed.

Most classification systems are hierarchal, proceeding from the general to the specific. The hi-

erarchal classification has been referred to by the Brisch developers as a monocode system. In an

attempt to derive a workpiece code that addressed the question of how to include several related, but

non-hierarchal, workpiece features, the feature code or polycode concept was developed. Some clas-

sification systems now include both polycode and monocode concepts.

A few classification systems are quite simple and yield a short code of five or six digits. Other

part-classification systems are very comprehensive and yield codes of up to 32 digits. Some part

codes are numeric and some are alphanumeric. The combination of such factors as application,

identified attributes and relationships, hierarchal versus feature coding, comprehensiveness, and code

format and length have resulted in a proliferation of classification systems.

31.1.2 Application

Identification of intended applications for a workpiece classification system are critical to the selec-

tion, development, or tailoring of a system.

It is not likely that any given system can readily satisfy both known present applications and

unknown future applications. Nevertheless, a classification system can be developed in such a way

as to minimize problems of adaptation. To do this, present and anticipated applications must be

identified. It should be pointed out that development of a classification system for a narrow, specific

application is relatively straightforward. Creation of a classification system for multiple applications,

on the other hand, can become very complex and costly.

Figure 31.1 is a matrix illustrating this principle. As the applications increase, the number of

required attributes also generally increases. Consequently, system complexity also increases, but often

at a geometric or exponential rate, owing to the increased number of combinations possible. There-

fore, it is important to establish reasonable application requirements first while avoiding unnecessary

requirements and, at the same time, to make provision for adaptation to future needs.

In general, a classification system can be used to aid (1) design, (2) process planning, (3) materials

control, and (4) management planning. A brief description of selected applications follows.

Design Retrieval

Before new workpieces are introduced into the production system, it is important to retrieve similar

designs to see if a suitable one already exists or if an existing design may be slightly altered to

accommodate new requirements. Potential savings from avoiding redundant designs range in the

thousands of dollars.

Design retrieval also provides an excellent starting point for standardization and modularization.

It has been stated that "only 10-20% of the geometry of most workpieces relates to the product

function." The other 80-90% of the geometric features are often a matter of individual designer taste

or preference. It is usually in this area that standardization could greatly reduce production costs,

improve product reliability, increase ease of maintenance, and provide a host of other benefits.

One potential benefit of classification is in meeting the product liability challenge. If standard

analytic tools are developed for each part family, and if product performance records are kept for

those families, then the chances of negligent or inaccurate design are greatly reduced.

The most significant production savings in manufacturing enterprise begin with the design func-

tion. The function must be carefully integrated with the other functions of the company, including

materials requisition, production, marketing, and quality assurance. Otherwise, suboptimization will

likely occur, with its attendant frequent redesign, rework, scrap, excess inventory, employee frustra-

tion, low productivity, and high costs.

Generative Process Planning

One of the most challenging and yet potentially beneficial applications of workpiece classification is

that of process planning. The workpiece class code can provide the information required for logical,

consistent process selection and operation planning.

The various segments of the part family code may be used as keywords on a comprehensive

process-classification taxonomy. Candidate processes are those that satisfy the conditions of the given

I WORKPIECE ATTRIBUTES/CHARACTERISTICS/VALUES

//$//////$////t/*/ /

/ ///§/9/*/*/*/£/ ///// /

/•/*/£/*/////*/$/£/*/*/ /

APPUCAT.OMS /^ff^f^

Generative design

Design retrieval

Generative

process planning

Equipment

selection

Tool design

Time/cost

estimating

Assembly

planning

Quality planning

Production

scheduling

Parametric part

programming

Fig. 31.1 Attribute selection matrix.

basic shape and the special features and the size and the precision and the material type and the

form and the quality/time requirements.

After outputting the suitable processes, economic or other considerations may govern final process

selection. When the suitable process has been selected, the codes for form features, heat treatments,

coatings, surface finish, and part tolerance govern computerized selection of fabrication and inspection

operations. The result is a generated process plan.

Production Estimating

Estimating of production time and cost is usually an involved and laborious task. Often the results

are questionable because of unknown conditions, unwarranted assumptions, or shop deviations from

the operation plan. The part family code can provide an index to actual production times and costs

for each part family. A simple regression analysis can then be used to provide an accurate predictor

of costs for new parts falling in a given part family. Feedback of these data to the design group could

provide valuable information for evaluating alternative designs prior to their release to production.

Parametric and Generative Design

Once the product mix of a particular manufacturing enterprise has been established, high-cost, low-

profit items can be singled out. During this sorting and characterization process, it is also possible

to establish tabular or parametric designs for each basic family. Inputting of dimensional values and

other data to a computer graphics system can result in the automatic production of a drawing for a

given part. Taking this concept back one more step, it is conceivable that merely inputting a product

name, specifications, functional requirements, and some dimensional data would result in the gen-

eration of a finished design drawing. Workpiece classification offers many exciting opportunities for

productivity improvement in the design arena.

Parametric Part Programming

A logical extension of parametric design is that of parametric part programming. Although parametric

part programming or family of parts programming has been employed for some time in advanced

numerical control (NC) work, it has not been tied effectively to the design database. It is believed

that workpiece classification and coding can greatly assist with this integration. Parametric part pro-

gramming provides substantial productivity increases by permitting the use of common program

modules and reduction of tryout time.

Tool Design Standardization

The potential savings in tooling costs are astronomical when part families are created and when form

features are standardized. The basis for this work is the ability to adequately characterize component

pieceparts through workpiece classification and coding.

31.1.3 Classification Theory

This section outlines the basic premises and conventions underlying the development of a Part Family

Classification and Coding System.

Basic Premises

The first premise underlying the development of such a system is that a workpiece may be best

characterized by its most apparent and permanent attribute, which is its basic shape. The second

premise is that each basic shape may have many special features (e.g., holes, slots, threads, coatings)

superimposed upon it while retaining membership in its original part family. The third premise is

that a workpiece may be completely characterized by (1) basic shape, (2) special features, (3) size,

(4) precision, and (5) material type, form, and condition. The fourth premise is that code segments

can be linked to provide a humanly recognizable code, and that these code segments can provide

pointers to more detailed information. A fifth premise is that a short code is to be adequate for human

monitoring, and linking to other classification trees but that a bitstring (O's, 1's) that is computer-

recognizable best provides the comprehensive and detailed information required for retrieval and

planning purposes. Each bit in the bitstring represents the presence or absence of a given feature and

provides a very compact, computer-processable representation of a workpiece without an excessively

long code. The sixth premise is that mutually exclusive workpiece characteristics can provide unique

basic shape families for the classification, and that common elements (e.g., special features, size,

precision, and materials) should be included only once but accessed by all families.

E-Tree Concept

Hierarchal classification trees with mutually exclusive data (E-trees) provide the foundation for es-

tablishing the basic part shape (Fig. 31.2). Although a binary-type hierarchal tree is preferred because

it is easy to use, it is not uncommon to find three or more branches.

It should be pointed out, however, that because the user must select only one branch, more than

two branches require a greater degree of discrimination. With two branches, the user may say, "Is it

this or that?'" With five branches, the user must consider, "Is it this or this or this or this or

this?" The reading time and error rate likely increase with the number of branches at each node. The

E-tree is very useful for dividing a large collection of items into mainly exclusive families or sets.

Round

Solid Shapes |SSund' w/Devia''°ns

[Round, Bent C'Line

Rotational Dome

I O/T Solid pklli—

Basic Shape Ll2™i_

Columnar, Straight

Sheet Forms

iNon-Rotational Box.|jke So|j(js

Named Shapes

Fig. 31.2 E-tree concept applied to basic shape classification.

N-Tree Concept

The N-tree concept is based on a hierarchal tree with nonmutually exclusive paths (i.e., all paths may

be selected concurrently). This type of tree (Fig. 31.3) is particularly useful for representing the

common attributes mentioned earlier (e.g., form features, heat treatments, surface finish, size, preci-

sion, and material type, form, and condition).

In the example shown in Fig. 31.3, the keyword is Part Number (P/N) 101. The attributes selected

are shown by means of an asterisk (*). In this example the workpiece is characterized as having a

"bevel," a "notch," and a "tab."

Bitstring Representation

During the traversal of either an E-tree or an N-tree, a series of 1's and O's are generated, depending

on the presence or absence of particular characteristics or attributes. The keyword (part number) and

its associated bitstring might look something like this:

P/N-101 = 100101 • • • 010

The significance of the bitstring is twofold. First, one 16-bit computer word can contain as many

as 16 different workpiece attributes. This represents a significant reduction in computer storage space

compared with conventional representation. Second, the bitstring is in the proper format for rapid

computer processing and information retrieval. The conventional approach is to use lists and pointers.

This requires relatively large amounts of computation and a fast computer is necessary to achieve a

reasonable response time.

Keywords

A keyword is an alphanumeric label with its associated bitstring. The label may be descriptive of a

concept (e.g., stress, speed, feed, chip-thickness ratio), or it may be descriptive of an entity (e.g.,

cutting tool, vertical mill, 4340 steel, P/N-101). In conjunction with the Part Family Classification

and Coding System, a number of standard keywords are provided. To conserve space and facilitate

data entry, some of these keywords consist of one- to three-character alphanumeric codes. For ex-

ample, the keyword code for a workpiece that is rotational and concentric, with two outside diameters

and one bore diameter, is "Bll." The keyword code for a family of low-alloy, low-carbon steels is

Al. These codes are easy to use and greatly facilitate concise communication. They may be used as

output keys or input keys to provide the very powerful capability of linking to other types of hierarchal

information trees, such as those used for process selection, equipment selection, or automated time

standard setting.

31.1.4 Part Family Code

Purpose

Part classification and coding is considered by many to be a prerequisite to the introduction of group

technology, computer-aided process planning, design retrieval, and many other manufacturing activ-

* Bevel

Chamfer

* Corner/Edge c\\\^

Features

* Notch

Radius

O/T Above

Hole/Recess

Teeth/Thread/Knurl

Form Features

Bend

Boss

Keyword I P/N-1011 ^ . 4.

Fin

II * Projection

F'ange

pTab

Joggle/Louver

Fig. 31.3 N-tree concept applied to form features.

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: