21st Century Manufacturing Episode 4 doc

Functional requirements FR might well be the performanceissues described for computers amount of RAM and so forth and automobiles ters COP might then be the specific design parameters fo

Step IIIanufacturinspecifics

3.1a

3.1b

Evaluation ofrequirements

RelativeAccording ~c:~~~~~ numerical Rank

to r~~~7.:er class ranking of ordering(1-10) c00~g)or

FigureJ.t The mapping process in QFD and Compton's(1997) example for a

textbook (from Engineering Management by Compton,© 1997.Reprinted bypermission of Prentice-Hall, Inc., Upper Saddle River, NJ)

Figure 3,1b is Compton's hypothetical example for a textbook This ranks thetopics from "easy to understand" as number 1, down to "good resale value" asnumber 8 In another matrix, these customer requirements such as "easy to under-

examples," "organization of material," "number of diagrams," and so on Through a

product characteristics are finally identified and launched into the process planning

department would select layouts, artistic style, and even the weight and dimensions

of the book to reflect the desires of the customers

One can also imagine QFD applied to automobile design and fabrication pose a consumer group typical of Volvo owners-values crash resistance above all

left side might be "avoid crushing a door from a side impact," and the "how" alongthe top might be "an attainable stress level." In the "manufacturing specifics" thenecessary sheet metal thickness and grade of steel might then emerge along the top.Sheet thickness and grade will also influence the final cost of the product in the man-ufacturing table.

There is a drawback to the QFD method for the design of mass-produced ucts such as standard automobiles OFD assumes (a) the existence of distinct marketsectors and (b) that everyone inside each sector has the same ordered set of prefer-ences QFDfor a group breaks down if the preferences of the group members are notthe same (see Hazelrigg, 1996) For example, the group of consumers who value crash

prod-if that group of people does not like the style, price, or color selections, the resistance characteristic may not be high enough to boost sales Having said this, OFD

crash-there is a limited range of product characteristics (Hauser and Clausing, 1988).3.4.2Axiomatic Design

An alternative to OFD has been developed by Sub (1990) and colleagues It alsobegins with customer needs Three steps follow: identifying functional requirements(FR), design parameters (DP), and process variables (PV) In this regard, the map-ping shown in Figure 3.2 below is philosophically the same as Figure 3.1a for OFD.The idea is to map the qualitative desires of the customer to more concreteengineering terms Functional requirements (FR) might well be the performanceissues described for computers (amount of RAM and so forth) and automobiles

ters COP) might then be the specific design parameters for a computer chip or a carengine needed to attain the FR Process variables (PV) would then be the semicon-ductor or production-line technologies needed to create the design of the chip or theautomobile engine

To make these mappings converge as the design team moves across Figure 3.2,Suh and colleagues propose two axioms:

1 Maintain the independence of the functional requirements (FRs).

2 Minimize the information content of the design

The first axiom aims to create a design where the functional elements are decoupled.For example, Suh analyzes a variety of designs for an injection molding machine (see

~'P"

User

needs

Fixure3.Z The framework for axiomatic

design (from The Principles of Design by

Nam P Suh, © 1990byOxford UniversityPress.Jnc.)

Suh, 1990, 72-78) In some designs, three functional requirements (FRs)-themelting rate of the plastic, the flow rate, and the pressure rise in the extruder-arcall affected by just one design parameter (DP) the rotational speed of the screw (the

coupled design is criticized because it will be impossible to regulate any of the threefunctional requirements (FRs) independently As a result, alternative designs areexplored where individual functional requirements (FRs) can be controlled inde-pendently of a screw mechanism

The mathematical way to create such independence is to set up a "designmatrix," A, where (FR)=A (DP).The matrix elements define the nature of the rela-tionship between each of the functional requirements (FR;) and each design param-eter(DP j). The individual elements of the matrix are given by

'J SDPj

A "decoupled'' product design will be achieved if A is a diagonal matrix In otherwords, A should he a r.quare matrix and its nonzero elements should appear only on

pling Various designs can thus be analyzed The aim is to arrive at a situation where

any specific function of the device is related to one and only one design parameter.

Speaking colloquially, if one were to "tweak" that design parameter, it should onlyinfluence that one functional requirement and not cascade into other functions ofthe device

Intuitively, the second axiom is quickly grasped It advises the designer tocreate simple subcomponents and devices In fact this is also one of the key ideas in

They advocate simplifying the shape of individual components in an assembly andsimplifying the fit of one component with another, for example, using press fits wherepossible rather than screws This is further described in Chapter 8 with reference tothe redesign of IBM's ProPrinter

3.5 THE ANALYTICAL PHASE OF DESIGN

Soon after the conceptual design and the high-level analysis are in place, detaildesigns should be pinned down Although these can be done with simple "desktopCAD/drawing systems," most of today's CAD programs encourage their users to addsoftware modules to their basic CAD package These include, but are not limited to,(a) constraint-based design and parametric modeling, (b) design for manufacture/assembly/environment (DFMlAIE) scoring, and (c) finite element analysis (FEA)

3.5.1Constraint-Based Design and Parametric Modeling

Rather than build up a model with specific dimensions, it is often useful to createconstraints between certain features or lines Consider the simple cover plate to a

outer dimensions of the plate are usually 112 X 68 millimeters (4.5 X 2.72 inches).

one can find light switch covers where the outer shape is a popular cartoon character.However, in the United States, there is a central rectangular slot for the actual light

The dimensions of the central slot and the positions of the screw holes must always

cover can thus be wildly imaginative on the outside contours-but it must be strictlyconstrained on these inner, more engineering-like dimensions

Another example from Shah and Mantyla (1995) is shown in Figure 3.3:

• Line 3 is parallel to line 5

• Line 2 is a circular are, tangent to lines 1 and 3

• Line 4 is oriented at0:.to line 3

• Line 1 is horizontal and lengthb.

• Line 5 is perpendicular to 1 and lengtha.

A more sophisticated development of such constraints is parametric modeling Thisprocedure allows fast scaling of an object into unique variations from one original.Consider a simple aluminum can for soft drinks (figure 3.4) Rather thanspecify that the dimensions are 120 millimeters (4.75 inches) high and 62 millimeters(2.5 inches) in diameter the ratio4.7512.5 ==1.9 might be used The height and diam-eter are then described in proportion to each other without specific dimensionsbeing used Then the model can be scaled up or down without having to numericallyrespecify each of the new dimensions

Figure 3.3 Example parametric

definitions (from Parameter and Feature-Based CAD/CAM, Jami Shahand M Mantyla, © 1995 Reprinted bypermission of John Wiley&Sons Inc.)

4ParaUel(line 3 line 5)

• Line B parallel to line 5

• Line2 acirculararc,tangenttolines! and 3

• Line4 orientedatatoline3

• Line 1 horizontal and length b

Figure 3.4 Parametric scaling of a soft drink can

3.5.2Design for Assembly, Manufacturing, and the

Environment {DFA/M/EI

the designer or design team Ideally, designs can be modified early in the design

techniques to analyze a particular design for its manufacturability include the designfor manufacturing and assembly (DMFA) software tools by Boothroyd andDewhurst (1999) Theirs is a commercialized product that designers can access by

CD or the Web.A suite of tools is available that contain, for example, DFM softwarefor machining, DFM software for sheet metalworking, DFE software to assess envi-ronmental impacts, and their best-known DFA module for evaluating assembly.The DFA for assembly module involves two key ideas:

• The quality of individual subcomponents must be high Also, their numbermust be reduced as much as feasible

• Assembly operations must be as simple as possible For example, factory outs should be orderly, the shape of individual components should be simple,design features should simplify the assembly of one component with another,and assembly operations should not fight gravity

lay-Their DFM for machining module "Machining for Windows" assists a designer withthe following issues: developing operation and process plans, obtaining cost esti-

ning for production

3.5.3Analysis and Decision-Based Design

Finite element analysis (PEA) allows the optimization of material use and

perfor-detail design process In many cases this perfor-detailed engineering analysis provides

tech-Despite the apparent rigor of FEA calculations, they should be interpretedwith caution In particular, the cummun "safety Iactors" that have been developedover many decades for design work should still be applied This is because there isuncertainty in the engineering materials that are used today, and also in theboundary conditions that are used for the finite element modeL Siddall's (1970)

based engineering design (Hazelrigg, 1996) In general the designer does not havethe luxury of measured, or a priori known, values of stress due to loading; instead the

today's steelmaking and other production methods, the designer does not know cise values for the yield strength and similar properties; again there is a probabilityfunction The diagram shows the great dangers that can arise when the "tails" of thetwo probability functions overlap in the "failure zone."

pre-A key question arises: What does a designer do to address the "failure zone"?The answer critically depends on the preferences, or values, of the designer For rou-tine consumer products the goal of a design is to make money, and more is better(Hazelrigg, 1996)

With "making money" as the main objective, the designer might choose rial properties in such a way that "an occasional failure is acceptable"; for example,

mate-consumer might be temporarily disappointed by a product failure But if a fast,

cour-utility, u, will be acceptable In other words, a consumer product will not be signed and overcostly" from choosing material properties that are "oversafe'' so that

"overde-no failures will occur

By sharp contrast, for the aerospace and aircraft industry, a designer shouldreevaluate the goal of making money and balance it with "stability and reliability."

In the extreme case of a nuclear weapon, mentioned at the beginning of Chapter 2,society demands "reliability" rather than "cost effectiveness" as the main designgoal In Figure 3.5, this means that the two curves should move apart on thexaxisleaving no possibility for a failure zone

Figure3.5 A frequency distributionfor predicting the strength ofa design Itshows the inevitable spread in materialproperties on the right and loadingstresses on the left (courtesy of JamesSiddall,1970)

Stress due

to loading

Yield stress

of material

With these thoughts in mind the designer can now begin to create a specificdesign with specific dimensions.

3.6 THE DETAILED PHASE OF DESIGN

In the "old days," the creative designers would pass their drawings down to the drafting

appropriate tolerances With the advent of CAD stations, it was obvious that much of this

and-drop positioning from the mouse The detailed CAD methods that are available fallinto several categories, including wire frame methods and solid-geometry techniques

by step to show the construction techniques However, reading these tutorials, or buyingthe bookTeach Yourself AutoCAD, is no substitute for going to a CAD workstation, log-ging in, and persevering through the tedious little commands that eventually create a niceisometric viewand the three principal orthogonal viewsof a fairly standard object Detaileddesign using CAD is like driving a car: no amount of tutorials in high school drivers' edu-cation classes can prepare you for the actual experience of cornering, even at 10 mph

For detail design (as opposed to conceptual design) the utility of traditional

CAD programs as a communication tool is undeniable Drawings remain consistent,

component designs to check for clearance, flushness, perpendicularity, and other related qualities Although this process is often tedious, it is a great improvementover manual methods CAD has also been useful in articulating precise details so thatdesigners can better understand their implications during each design iteration.Most contemporary CAD tools have both wire frame and solid modeling capa-bilities Solid modeling provides a sense of form usually lacking in wire frames Solid

fit-provides insights into the manufacture (particularly the machining) of parts andassemblies Solid modeling by both constructive solid geometry (CSG) and destruc-tive solid geometry (OSG) is described later in this chapter

3.7 THREETUTORIALS: AN OVERVIEW

The object pictured in Figure 3.6 is an example of a prototype of a joystick for the user of

a virtual reality (VR) environment The joystick has a somewhat heavy mass as its base

virtual environment appropriately Information is transferred to the computer via a par~allelport

With the exception of the handle, the device is fabricated from componentswith rather simple geometries The joystick will now be designed with three differentCAD methods:

• Wire frame using the AutoCAD package (Section 3.8)

• Constructive solid geometry (CSG) using the AutoCAD package (Section 3.10)

F1pre 3.6 Virtual reality joystick (from a student group led by Ryan Inouye).

The three tutorials show the characteristics of each CAD technique In addition the

some aspects of parametric based design.'

3.8 FIRST TUTORIAL: WIREFRAME CONSTRUCTION

Simple wire frame CAD systems use basic mathematical and computer graphic nologies wire frame programs begin by allowing users to choose points from a local orrelative reference frame (the choice of the local frame is often arbitrary) These pointsare then mapped onto a global reference frame Finally,lines are drawn between points.The final image is therefore a connection of lines that may be hard to viewclearly For example, (a) lines that should be hidden, perhaps representing the back

tech-front face will be possible.A quick glance ahead to the figures depicting wire frameconstructions shows these difficulties with visualization

IThe use ofAutoCAD,SolidWorks,and SDRC is not intended to be an endorsement of these ucts or a deliberate exclusion of the other CAD products listed in the URLs at the end of the chapter Fur-

prod-some parametric design.A variety of CAD systems was chosen for the chapter to deliberately show prod-some

Althoughwire frame models lack the advantages ofsolidrepresentations, theyare nevertheless useful The inner workings of wire frame programs are also simple

to understand and thus adapt for special user-generated purposes Also, the puter's calculations mirror those commonly found in linear algebra, dynamics, androbotics classes

com-The fact that most wire frame programs require less powerful computers hasuntil very recently been the most compelling reason for their use However, the

associated figures) and the availability of cheaper, more powerful machines nowbring solid modeling tools to the average desktop machine

For illustrative purposes, consider the base of the joystick.Byselecting anumber of desired lengths for the base, a series of lines can be developed to outline

only half of the object and thenmirror it across a central plane 'This generalapproach will be used here The advantage is that the overall time taken is less, and

right half, plan view of the base.The halfway point of the line segment that representsthe back of the base will be the mirroring plane origin,8.Therefore, point a is at coor-dinatex=0, y=0,Z=0, or more simply (0,0,0).The line beginning at point8wasdrawn with thelinecommand The line travels from the middle of the back bottomedge of the base to the end of that edge The line then travels toward the front of the

of the base to r, and finally moves across the front bottom edge

Figures 3.7 and 3.8 provide the results of some of these initial suggestions onhow big the base of the object should be Figure 3.7 is the top view of the externaledges of half of the base Figure 3.8 is an isometric view of the same lines.It providesinsights into the three-dimensional nature of the lines

The viewpoint in Figure 3.8 was accomplished with thevpoint command inAutoCAD The setting for the viewpoint is (1,-1,1),or positive in thexdirection, neg-

fFront of joystick baseMirroring

Figure 3JI Isometric view of one-half of the base (viewing from the back of thejoystick where the wires can be seen in the earlier photograph).

ative in theydirection, and positive in theZdirection, with each axis held to 1:1 portions (i.e., isometric) While making drawings, it is a good idea to experiment withthe overall view by changing the viewpoint proportions to see the effects of fore-shortening

pro-The line starting at point b was drawn in a similar manner according to thedimensions for the top surface of the object All points on the b line have a constant,

z dimension above the points on the a line The lines originating at c and d follow thesamex-yprofile Each of these lines has its own, constantzcoordinate.Points a, b, e, and d were specified with Cartesian coordinates The choice ofcoordinates was dictated by the desired object size and ease of calculation Thepoints

at c and d were first determined with Cartesian coordinates However, thelinesthatoriginate at e and d, going along the angled portion and the front edge, have beendetermined using polar coordinates Points e and f were determined by a fixed dis-tance and an angle from points c and II.Temporary construction lines establish all thepoints for the shape df and ceo but are later removed They are there temporarily tovalidate the correct shape of the slanting surface along ref The outline of the baseexterior begins to become clear in Figure 3.9 Themirror command was used toduplicate the objects across they-zplane atx=O.Figure 3.9 contains some over-

hangs and extraneous line segments The overhangs can be trimmed with the trim

first clicking on thc excess segments jl, hk, I'm, and en as the objects to trim Theerase

command then leaves segments ee, df, gb, Ij, ej, and bf

The outline of the object is revealed by constructing lines between points 0 and

hob andj,j andq,p andf,fand e,and e and r, all with thelinecommand For example,lines along the z direction (between e andf,jand h, ) at the intersection of planescomplete the general outline of the object Figure 3.10 presents the results thus far.With the general outline of the object described, the generation of inner details

Left side to be mirrored

Right side of base

Bock

Figure 3.9 Outline of the base exterior before the trimming process.

F1gure3.10 Outline of the base exteriorafter the trimming process

A good place to start is the shallow rounded pocket in the bottom face, becausemany consumer electronics products, including telephones, have a cover plate on the

into The cover plate will close over the larger internal cavity that houses the printedcircuit boards and wires The plate generally follows the contour of the bottom sur-face of the object For the joystick, the base plate itself is a flat trapezoid withroundedcorners

A shallow rounded pocket will thus be drawn in the next few images to modate the plate This task begins in Figure 3.11 First, the straight lines of a trape-

accom-will then be connected, next rounded, and then copied and repeated for the depth ofthe cavity

In the first construction step, the lineiis drawn parallel to the base, at a small

Ilemporaryoonstructionlines(ceanddf)

Figure 3.11 Initial formation of the cavity in the base exterior

of the object on each side Lineiibegins by placing a tiny line of lengthwcular to the other edge The full extent of linefiis then constructed from the end point

perpendi-of this short construction line w It goes past the front perpendi-of the object, parallel to the

angled bottom edge The line is completed by extending it with the extend command

such that it intersects linei,The extend command finds the intersection cally Line iii is constructed in a similar manner from the short construction line w2from the front face Once one of the lines marked "il" is constructed on the right side

automati-of the object it is mirrored to the left, across the y axis, to form the image.

The next step is to trim lines I,ii,andiiiso that only the trapezoid remains.The

cor-ners of the trapezoid are then rounded with the fillet command The fillet command in

AutoCAD can be quite confusing to a new user because the default fillet radius is alwaysset to zero The procedure is to enter the fillet command,hit the"r"key (for radius) whenthe dialog line appears in the command area, and then enter the desired fillet radius

conunand has been used on each comer of the trapezoid, selecting both intersecting lines

created for the curved region of each fillet The trapezoid has thus been transformedfrom four individual lines to eight segments: four lines plus four fillets

At this point it is convenient to join all eight segments comprising the zoidal pocket into a single line This is done with the pedit (polyedit) command This

trape-procedure begins by typing pedit and hitting enter When prompted, a segment of the

designer would like to make it one To make it one, the designer can simply entery(es) A set of polyline options is then presented, and the best choice is "[" for join.Each segment to be included in the polyline that caps the top of the handle is thenselected, remembering there are eight segments in this case Pressing "enter" joins allthe segments into one entity Joined polylines can be separated into their original

<Line of lengthw

Figure3.12 Cornpletion of the cavity inthe hMe extenor

To create the depth of the cavity on the bottom, the copy command is used The existing polyline is copied and repeated a short distance along the positive z axis.This

cavity depth The preceding procedures are used to generate Figure 3.12

The outline of the base cavity could have been drawn with the rectang and/or polygon commands These commands automatically make polylines; however, they

are not preferred because individual segments might be edited out from the originalpolylines when the final smoothing command is applied The only way to avoid this

it is a simpler process to start by drawing line segments and then joining them intopolylines

There are additional cavities within the object that house the accelerometersand circuitry, as shown in Figure 3.13

The process begins by drawing a series of lines in the shape of a large rectangle,

a much smaller rectangle, and a trapezoid These are shown in Figure 3.13 going in a

"northeast" direction First constructions are best done on the lower level with laterextrusions in+z.

The corners of the large rectangle and trapezoid are then filleted (the radiusneeds to be reset to a smaller value) The common edge of the large rectangle and

common edge between the trapezoid and the smaller rectangle is also removed,leaving the open space between b andII,and the segments are joined as a polyline.The outline of the rectangles is then copied to a certain distance in the z direction,thus creating a pocket Next, lines are drawn at the vertical intersecting edges-a,b,

Coand d-to show depth The trapezoid is also copied, but to a greater distance in the

z direction Progress thus far is illustrated in Figure 3.13, which shows the main blockfrom earlier drawings, the shallow recess for the base plate, the medium depth pocket

on the left, and the deeper trapezoidal pocket to the right Wires can be run betweenthe electronic components in each pocket through the small slotabedin between therectangular and trapezoidal pockets

A few more details remain The gripping handle is attached to the base with

interior cavities

in the bottom of the handle Therefore, so as not to interfere with the electronicsinside the inner cavities, it is necessary to provide large countersinks for the boltheads

The counter-bore for the large bolt head is represented by a large circle, andthe hole for the screw thread part by a smaller concentric one Thus, originally, two

one for the through-hole that will take the screw thread part

Circles are created with thecircle command The user is prompted for thecenter and radius of the circle The circles are then copied to the appropriate posi-

from the distance ab in the earlier figures.Thus in Figure 3.14,the upper smaller cles are coincident with the top surface, meaning they are through-holes for thescrew threads

cir-Figure 3.14 also illustrates that circleyintersects one of the vertical surfaces ofthe rectangular pocket Circleywas modified by trimming the entire right-side half.Such manual operations are a major disadvantage of wire frame modeling.Ambiguous drawings can result if they are not attended to, which may result in costlymistakes during manufacture

Figure 3.14 is now completed to the extent required for this illustration Screwholes for the base plate and a hole for the parallel port cable have been omitted here,but are eventually required

To provide greater clarity, internal lines can be changed to hidden lines Withhidden lines the object meets all the criteria required of typical multiview drawings.Each of the six possible orthogonal viewpoints can be obtained with thevpoim

command, but some hidden lines may have to be manually adjusted as the viewpoint

what difficult to read The cylinders (i.e., through-holes) removed hom the base are

cult to see depth and surfaces without concentrating

Development of the wire frame model has made no reference to bility or even physical feasibility The drawing is simply a set of lines and curves

manufacture-Flgure 3.14 Complete 3-D drawing oftheVR joystick base.

point to the wire frame model to render the surfaces.This rendering makes the object

representation However, since we are still working only with a wire frame, this sion of rendering does not make the computer understand the surfaces Just to emphasize this last point: all the understanding so far is in the eye and

ver-the brain of ver-the beholder-ver-the human CAD designer Actually, the human designercould have drawn an "Escher-like art image" that would be impossible to manufac-

ture, and the wire frame model would have happily accepted it!

3.9 SOLID MODELING OVERVIEW

3.9.'Introduction

In contrast to the previous wire frame methods, solid modeling creates objects that

devoted to the formal aspects of these topics For example, the details are

compre-hensively described in Geometric and Solid Modeling by Hoffmann (1989) and in Computer Graphics: Principles and Practices by Foley,van Dam, Feiner, and Hughes

(1992).The notes below on Boolean set operations, b-rep, and CSG are arranged in

form is gratefully acknowledged

In the wire frame tutorial (Section 3.8), the person sitting at the CAD system's user interface mostly clicks on points and connects them with lines to create the joy-

the CAD system's user interface constructs the joystick by adding, subtracting, or intersecting individual bodies (Section 3.10) or destructs the joystick by starting with

a large block and subtracting smaller bodies (Section 3.11) Such construction or

disassembling "Lego blocks" of different shapes and sizes.The operations are done