Pré-visualização3 páginas / 392
Pesquisar no resumo do documento
SPC Caleulatians for Control Limits Average df Measurements — Average Of Averages — Range — Average of Ranges USL—Upper Specification Limit LSL—Lower Specificatioa Limit Notation: UCL —Upper Control Limit LCL —Lower Control Limit CL —Center Line n Sample Size PCR — Process Cápability Ratio 3 — Process Standard Deviation a gp Variables Data (7 and R Control Charts) x Conto; Chart n Ao Ds D, UCL + AR x 2 L880 — G000 3.267 3 1023 - G000 2574 4 ops 0000 2282 I 5 057 000 24 2326 ROontrol Char 6 Oag3 0000 200 253 UCi=R Dy 7 o49 0076 1924 2704 Ds 3 COS G136 1864 2847 o 0357 - Qltd 1816 2970 o 10 0308 023. 197 308 Capability Study Cp = (USL - LSL)/(66); where é = Eid, Attribute Data (p, np, é, and 4 Control Charts) nontompotnes dêFectives Control Chart Formulas Defectês np (number of c(coumt of 4 (count of p ffacson)) nonconforming) nonconformânces) nonconformaneesfniey > np é E Pe np+3 pd) CRENÇA El Fa- D pad e-mE Ha vares, use H a must be . nrzusthe Fn varies, usem - dr individual n; a constant a constant or indívidnal n; a Guide to Univariate Process Monitoring and Control TE E penal Other Wiley books by Dougias €. Montgomery Website: vryrwwiley.com/coilege/montgomery Engineering Statistics, Third Edison By Montgomery Ranger and Hubele Introduction to engineerirg statistics, with topical coverage aporaprinte for à one semester course. À modest mathematical jevel, and an epplied approach, Applied Statistics and Probability for Engineers, Third Edition By Montgomery and Runger Introducúon to engineering statistics, with topical coverage appropriate for either à one- or iwo-semester course. An applied approach Io soiving real-world engineering problems. Probability and Statistics in Engineering, Fourth Edition By Hines, Montgomery, Goldsmam, and Borror Wensite: www. wiley.com/college/bines For a first tvo-semester caurse in applied probability and stadistios for undergraduato smdents, or a one-semester refresher for graduate students, covering probabilicy from the start. Introduction to Statistical Quality Control, Fifth Edition By Douglas C. Montgomery For a fest course in statistical quality contro? Stasistical “echniques are emphasized throughont, with à strong engineeriag and management oriemation. Desiga and Analysis of Experiments, Fifth Edition By Douglas C. Monsgomery “Am introduction to the design and analysis of experiments, with the modest prerequisite of a first course in statistical methods. Introduction to Linear Regression Analysis, Third Edition By Montgomery, Peck, and Viring A comprehensive and thoroughly up-to-date look at regression analysis, still the most widely used technique in statistics today. * Experiments, Second Edition By Myers and Monigomery Website: wyrwswiley.comicallege/myers The expleration and optimization of response surfaces, for graduate courses in experi- mental desiga, and for applied siatsticians, engincers, end chemical and physical solentisis. Generalized Linear Models: With Applications in Engineering and the Sciences By Myers, itontgomeny and Vining Website: wryae.wiley.comvcollegeimyers An istroductory text or reference on Generalized Lincar Models (GLMS). The range of thcoreticel topics anc applications appeals boti to srdents and practicing professionals. Respunse Surface Methodology: Process and Product Optimization Using Designed | | | introduction to Statistical Quality Control . Fifth Edition Douglas C, Montgomery Arizona State University WILEY John Wiley & Sons, Inc. cin Ele deny tr capo. Jg Pos e as. Sata Das ig Dio Say Sos Trama rd Mp Sa gy bina Snóee book seas ty 10/12imes Roman ty Argosy Poblisbing Sendeos, and peates end bound by “This ooo was iypesesio E Dunnedey (Wind) The cover was pentes by Pheenix Color Compeat inelado áustained yield bar- sas memfctced dy ml ao fas managémem BOTITA DO O genes go Vi harvetog peniples nsue hs dh mom vesting é is imbectando. y nor exceed (ie amount 0E new aro. “shock ponta om aci pager, E . Copyrigit 692005 by John “Wiley & Sons, Toc. AJ riglus reserved. e Bar Cora ÃO, Wiuefis der rs Pio asian aa men of E copy fes to ths Copyright Clenranc 22 a as O 2 Publisher for permission shout a A IS, ACO, fax (978) 750-4470. Requests ta tho Pu or do eis E Beni E MREDO WILEY. COM, To order books or for customer a cal 1-80D-CALLWILEY(205-3945) Dougls € o real Qua Cone, Bio aum.asóars D.47-66122:8 (Wiley Intermanionel Edition) Prime o the Unived States of Amesios. 10937654321 About the Author Dongias C. Montgomery is the Arizona Stare University Foundation Professor of Engineering and Professor of Statistics, He received his B.S., M.S, and PhD. degrees from Virginia Polytechnic Instirute, all in engineering. From 1969 to 1984 he was a fac- ulty member of the School of Industrial &: Systems Engineering at the Georgia Instituir of Technology; from 1984 to 1988 he was aí the University of Washington, where he held the John M, Fluke Distinguished Chair of Manuficmring Engineering, was Professor of Mechanical Engireering, and was Director of the Program in Indusuial Engineering. Dr. Montgomery has research and teaching interests in engineering statistics inciud- ing statistical quality-cortxol techniques, design of experiments, regression analysis and empirical model building, and the application of operations research methodology to prob- lems in menufacturing systems, He has authored and coautiored move “ham 160 technical papers in these fields and is an author Of twelve osher books. Dr. Montgomery is a Fellow ot the American Society for Quality, a Fellew of the Ame-ican Statistical Association, 2 Fellow of the Royal Statistical Society, a Fehcw of the Institute of Industrial Engineers, and an elected member of the Internationa! Statisúcal Instimute. He is à Shewhart Megalist of tie American Sociery for Quality, and he has also received the Brumbaugh Award, the William G, Hunter Award, and the Shewell Award from the ASQ, He is à rec Ellis R. Ou Award. He is a focmer editor of the Journal of Qualiry Technology, is one cf the current chief editors of Quality and Retiabiliry Engineering International, and serves om the editorial boards of several joumals. INTRODUCTION AUDIENCE “This book is about the use of modem statistical methods for quelisy control and improve- mer. 1 provides comprehensive coverage of the subject from basic principles to state-of- the-art concepis and applications. The objective is to give the reader a sound “understanding of the principles aud the basis for applying them in à variety of situations. Although statistical Iechniques are emphasized thronghout, the boclk bas 2 strong engi- neering and management ortentarion. Extensive knowledge of statistics is not a prerequi- site for using this book. Readers whose backgrouné includes à basic course in statistical methods will find much cf the material in this hook easily accessible, “The book is an omgrowth ofover 30 years of teaching, research, and consulting ia the apeli- cation o? statistica! methods for industrial problems, E is designed as a textbook far srudents enrolled ir colloges and universities, who are smdying engines ing, statistics, management, and reiated: fieids and are taking a first course in statis! quality control. The basic quality-control course is often taught at the junior or senior level. AlÍ of the standard topics for this course are covered in detai?. Some more advanced material is also available in the book, and this could be used with advanced nudergraduztes who have had somo previous exposure to the basics or in a course aimed at geaduate students. T have also used the text materials extensively in programs jor professional practitioners, including quality and reli- ability enginecrs, manufncturing ard development engincers, prodect designers, managers, precurement specialists, marketing personnel, technicians end Inboratory analysts, inspeo- tors, and operators. Many professionals have also used the material for self-smudy. CHAPTER ORGANIZATION AND TOPICAL COVERAGE Chapter 1 is an introduction to the philosophy and basic concepis of quality improvement. Tt notes that quelity has become a major business strategy and that organizations that suc- cessfally improve quality can increase their productivity, enhance their market penetra- ton, and achieve greater profitabiliy and a strong competitivo advantage. Some of the menegeriel and implementation aspects of quality improvement are included. PREFACE Following the introductory chapter, thc book is divided into five párts. Part Eis à description of statistical methods nseftl in quality improvement. Topios covered include sampling and descriptive statistios, the basic notions of protability and probabiliry disti- bmejons, point and interval estimation of parameters, end statistical hypothesis tesing These topies are nsually covesed in a basic course in statistical methods; however, their presentation in this text is from the qualy-engincering viewpoint, My experience has been that even readers with É strong statistical background will find the approsch to this matcxial nseful and somevihat differerr from à standard statistios textbook, Part Jf contains four chapters covering the basic methods of statistical process control (SPC) and methods for process capabiliry analysis, Even thongh several SPC problem solviag tools are discussed (including Pareto charts, cause-and-cífect diagrams, for example), tte primary foces in this section is on lhe Shewhart control chart. The Sheswiart control chart is certainly ag: new, but its usc in modem-day business and industcy às of tremendous valve. There are four chaptecs in Part IL that present some moxe advanced SPC methods. included are the cumalative sm and exponentially weighted moving average control cliaris (Chapter 8), several important univariate control charts such as procedures for shor: procuctíon runs, antocorrelated data, and multple sueam processes (Chepter 9), multi variate process monitoring and control (Chaprer 10), and ieechack adjustment techniques (Chapter 11). Some of tis meterial is at à higher level than Par HE but much ofitÃs ecces- sible by advanced undergraduates or first-year pradvate students. This material forros the basis af à second course in statstical quality control and improvement for this audience. Pert TV contains rwo chapters that show how statistically designed experiments cam be used for process design, development, and improvement. Chapter 12 presents the funcia- mental concepts of designed experimente and introduces the reeder to factorial anú fiac- cional factorial designs, with particular emphasis on the two-level system of designs. “These designs are used extensively in industry for factor screening and process character- ization. Although the treatment of the subject is not extensive and is no substinte for a formal conse ia experimental desiga, it will enable tha reader to appreciate more sophis- Ecared examples of experimental desigr. Chapter 13 introduces response aurface methods and designs, ilustrates evolutionary operation (EVOP) for process monitoring, and shows how sutistically designed experiments can be used for process robusmess sudies. Chapiers 12 and 13 emphesizo the important interrelationship berweer statistical process control and experimental design for process improvement. Two chepress deal with acceptance sampling in Part V. The focus is om lotby-lot acceptance sacnpling, although there is some discussion cf continuous sampling and MIL STD 1235C ins Chapter 14. Other sampling topies presented include various aspects oé the desiga of acceptance-sampling plans, à discussion of MIL STD 105E, MIL ST) 414 (and their civilian countesparts, ANSIASQC Z1.4 and ANSIASQC 719), and other techniques suoh as chain sampling and sip-lot sampling. -Shroughont tho book, guidelines are given for selecdag the proper type of statistical techutique to use in a wide variery of situaticas, Addiionally extensive references to jour. nai aráoles and other sechnical literature should assist the reader in applying he methods described. SUPPORTING TEXT MATERIALS Computer Software “The computer plays an important roie in a modem quality-contro! conrse. This edition of the book uses Minitab às the primary illustrative software package, Instruciors may order this book wi a student version of Minitab included, | strongly Tecominend that she conrse have 2 meanirgfol computing component. To request this book with Minitab included, contact your loczl Wiley represensative. You can find your local Wiley representative by going 10 www wiley.com end Cicking on the tab for “Who's My Ren?” . Supplementa! Text Material Thave yicitien à set of supplemental matecial to angment mam of lhe chapters in the book. The Supplementsl material contains topics that 1 could not easily fit into that chapter with- out serionsiy disrupting the flow, The topics arc shown ia the Table of Contents for the took and inthe individual chapter qutlines, Some pf this material consists of proofs or der- ivations, new copies OF à (sometimes) more advanced nature, supporting details concerm- ing remares or concepts presented in the text, and answers to frequently asked questions The smpplemental material provides an interestng set of accompanying ceadings for ary- one curicus about the field. It is available on the World Wide Web pege chat supporis the book, located at www wiley-com/college/montgomery. Student Resouree Manual mm The text contains answers to most Of the odii-numbered exermises. A Student Resource Mara] às available from John Wiley 4 Sons that presents comprehensive annotateg solu- tions to these same odd-mumbered problems. This is an excellent study aid that many text users will find extremely helpful. The Student Resouroe Manual mey he orderee in a ser with the text or purchasad separately. Contect your local Wiley representative to vequesr the set for your nookstare or purchase the Student Resource Manual from the Wiley website, Instruetor's Materials The instractor's secrion ox the texthook website contains the following: 2. Solutions to the text problems 2. The supplemental text material described above à. A serof Microsoft? PowesPoint slides for the basic SPC course 4, Data sets from the book, in electronic form The instructors section is for instruetor use only and is password-protected. Visit the Instructor Companion Site porton Of the website, located at www wiley.comicollege” montgomery. to register for à password. 2 World Wide Web Page “he Web page for the book is accessible tarongh the Wiley home page. Tt contains the supplemental text material and the cara sers in electronic form, It vailt also be used co post items of interest to text users. The website address às wwawswiley com/college/montgomery. Click on the cover of the text you are using. KNOWLEDGMENTS May people have generousty coetributed their time and Jnowledge of statistics and quality improvement to this took. 1 wouid like to thank Dr. Bill Woodall, Dr Doug Hamtáns, Dr. Joe Sullivan, De George Rmger, Dr, Bert Kears, Dr. Bob Hogg, Mr. Bric Zsegel, Dr. Joe Pignáricllo, Di. John Ramberg, Dx. Emic Seniga, Dt. Ensique Del Castillo, Dx. Sarah Streett, and De. Tim Alloway for their thosough and insigitful comments on this and. previous editions. They genermusy shared may of their ideas and teaching expert ences ivith me, leading to substantial improvements in the boot. Over the years since the firsc edition was published, 1 have received assistance. and ideas from a great meny other people. À complete list of colicagues with whom Lhaveinter- acted would be impossible to enumerate. However, some of the major contributors and their professional affliations are ns follows: Ds, Mary R, Anderson-Rovdand, Dr. Dvagne Rolls, and Dr. Norma E Eubele, Auzona State University; Mr. Seymour M. Selig, formeriy of the Office of Neval Research; Dr. Lymwood A. Johnson, Dr. Ressell G, Hedkes, De, David E. Eyffe, and Dx E. M. Wadswonh, It, Georgia Instinto of Technology; De Siarad Preblu and De Robert Rodriguez, SAS Institute; Dx. Richard L. Sto:ch and Dr. Cheistina M, Mastrangelo, University aí Washington; Dr. Cynthia A. Lowry, formenty of “Texas Christian Universiry; Dr. Smiley Cheng, Dr. John Brewster, Dr: Buan Macpherson, and Dr Fied Spiring, the University of Manitoba; Dr. Joseph D. Moder, University of Miami; Dr. Frank B. Alt, Universicy of Maryland; Dr. Kenneth E. Case, Oklahoma State University; Dr. Daniel R. MeCarvile, Dr. Lisa Custer, De. Pat Spagon, and Mr. Robert Stuare, al formerly of Motorola; Dr. Richard Post, Intel Corporation; Dr. Dale Sevier, Hiybritech; Mr. Jotm À. Butora, Me, Leon V. Mason, Mr Lloyd X. Collins, Mr. Dana D, Lesher, Mt. Roy E. Dent, Mr. Mark Fazey, Ms, Kathy Schuster, Mc. Dan Fritãe, De 1. 8. Gardiner, Mx. Ariel Rosentrates, Mt, Lolly Marwah, Me. Ed Sobleicher Mr. Amiin Weiner, asd Ms. Elaine Baechtle, IBM; Mx. Thomas C, Bingham, Mr. K. Dick Vaugha, Mr. Rabert EeDoui, Mi John Black, Mr Jack Wires, Dt Julian Anderson, Mr. Richard Altire, and Me Case Nielsen, The Being Company: Ms. Karen Madison, Me. Don Walton, andMr Mile Goza, Alcoa; Mr. Harry Peterson-Nedry, Ridgccrest Vineyards ard The Chehalem Group: De Russell À. Boyles, formeily of Precision Castparts Corporarion; De Sadre Khalesei and Mr, Franz Wagnet, Signeries Corporation; Mr, Larry Newton and Mr. C. T Howie, gia Pacific Corporation; Me, Robert V. Bexley, Monsanto Chemicals; Dr. Craig Fox Dr Thomas L. Sadosly, Mr. James P. Walker, and Me Jon Belvins, The Coca-Cola Company, Mr. BI Wagner and Mr AI Paciseno, Liton Industries; Mr, Fobm M Fiuke, e, Jota Flu Manufacturing Company; Dr. Paul Tobias, formerly of TBM and Semitech De William DuMouchel and Ms, Janes Olson, BBN Sofiware Producis Corporation, ? would also Uke o acknowledge che imany contributions of ay late partmer in Staxistical PRERACE ix Productiviry Consultants, Mr. Summer S, Averett, AL, ox these indivi ã , Avetett, AL, of these individuals and many otix Have contributed to my knoWledgo of the quality improvement field. ves The eititórial and production staff at Wiley, particulariy Ms. Cheriry Robey and Me, Wayne Andêrson, with velom Twokad for many years, and Ms. Jenny Welter, they ave find much patience with the over the years and have contributed grealy towards the suo- es cf his book. De. Cheryl L. Tennings mede many valuablo contributions by her care- a checkáng of the manuseript and procf materiale, and in prepering the solutions, E also e o po Chair of the Department of Industrial Engineering at Asizona State ersity, for bis support and for providing a tertific environment pd g ment in which to teach and o I tre pe vaçipus professional societies and publishers who have given permission reptoduce th=ir materials in my text. Permission credit is E egos fia edit is acknowledged at appropriate Tam also indebisd to the many organizations that h ave sponsored my research and m graduate students for a mumber of yezrs, inchuding he mrearber companies of the National Science FoundationIndustry/Universiry Cooperntivo Rescarch Center im Quelity and Reliahilivy Engineering at Arizona Stste Universioy, the Office of Naval Research, th National Science Foundation, the Aluminum Company of America, and the BM Corporation. Finally, L would like to thank the nsany users of the previous editions of this baok, including students, practicing professionals, end my academic colleagues. Many of the changes and (hopefully) improvements ia this edition of th de eng a e book are the direct result Douglas €. Montgomery Tempe, Arizona E EE ema ão Contents CRAPTI Quality Improvement ih he Modern Business Environment 1 Chapter Overview and Leaming Objectives 1 11 The Meaning 9f Qualiiy end Quality improvement 2 ;-L1 Dimensions of Quality 2 11.2 Quality Engineering Terminology 6 1.2 A Brief History of Quality Control and Improvement 8 1-3 Btaristical Methods for Quality Control and improvement 11 1-4 Management Aspests of Quality Enprovement 15 1.4.1 Quality Pnilosophy and Management Suategios 16 14,2 The Link Betweon Quality end Prodncuivity 27 143 Quality Costs 28 144 Legal Aspecis of Quality 33 1.45 Implementing Quality Improvement 35 PARTI Stsristical Metbods Useful in Quality Conaol ad Improvement 39 CHAPTER 2 Modeling Process Quaity 41 Chapter Overview and Leaming Objecives 41 21 Doseribirg Variatioo 42 211 The Stem-and-LcafPlor 42 212 The Hisogam dá 213 Nemerical Summary of Data 47 214 The Box Plot 50 21,5 Probabiliy Disuributions 51 23 Important Discreto Distribuiions 55 .3 The Hypesgeomesric Dismibution 55 22.2 The Binomisl Disuibution 57 22.3 The Poisson Distribution 58 2.2.4 The Pascal and Related Disuibutioos 59 Important Continnous Disitudons 6i 23.1 The Normal Distibudon 61 2:3,2 The Lognormal Distributon 66 3 The Exponential Distribution 69 2.3.4 The Gamma Diseribution 70 2.3.5 The Weibull Distribution 72 2-5 Probability Plots 74 2-4.) Normal Probability Plois 74 2.42. Other Probability Plots 76 2.5 Some Useful Approximations 77 23,3 The Binomial Approximation to the Hypergcomento 77 23,2 The Poisson Approximaton to the Binomial 78 2:53 The Nocmal Approximation to the Binomial 78 25: Commente on Approximations 79 2 CHAPTER 3 Enferences about Process Qualy 86 Chapter Overview and Leaming Objectives 86 3] Siatistios and Sampling Distributions 87 3-1.1 Sampling from à Normal Dismibution 88 3-1. Sampling from à Bemoulb Dismibution 92 3-13 Sampling frem a Poisson Distribution 92 3-2 Point Estimation of Process Parameters 93 3.3 Statistical Inference for a Single Sample 96 3-2.1 Inference an the Mem Of à Population, arianos Known 97 3.3.2 The Use of P-Values for Hypothesis Testing 100 xii comtaNTS , 4-33 Inference va the Mean Df a Normal Distribudon, Varianca Unkmown 101 3:34 Inference on the Variance of a Normal Distibudos 105 3-3. Inference 08 a Population Proportion 407 3.3.6 The Probabiity of Type Il Error and Sample Size Decisicns 109 3-4 Statistical Inference for Two Samples 112 3-4,1 Inference for a Difference in Means, Variances Known 113 3:42 Inference for a Difitrence in Means of Two Normal Distiibutians, Variances Unknown 116 343 Inference on the Variances of Two Normal Distribudons 125 3-4.4 Inference on Two Population Praportons 126 3-5 NWhas If There Are More Than Two Populerions? “The Analysis of Variance 128 3.5.1 An Example 128 3.5.2 The Analysis of Variance 130 35,3 Checking Assurmptions: Residual Analysis 137 PARTH Basis Methods of Staústical Process Control and Capabitity Analysis 145 CHAPTER 4 Methods and Phvilasophy of Sististical Process Conrol 147 Chapter Overviesy anti Learming Objective 147 41 Tomodection 148 42 Chance and Assignable Causes of Quality ” variation 148 43. Sististical Basis of the Control Char 150 4.3.1 Basie Principles 150 4:32 Choice of Control Limits 158 43.3 Sample Size and Sempling Prequenzy 160 434 Racional Subgroups 162 4.35 Analysis of Paerms on Control Chars 164 436 Discussion of Sensitizing Rules for Como! Chars 166 43,7 Phase 1 and Phase If af Control Chart Application 168 44 The Rest aí the “Magnificent Seven” 169 4.5 Implemening SPC 175 &5 An Applicadon of SPC +76 47 Nonmanufacavring Applications of Statisical Process Control 283 CHAPTER 5 Conto] Charts for Variables 194 Chapeer Overview and Leaming Objentives 194 54 Inreducdon 195 5-2 Control Chacis for Z and R 196 52,1 Statistical Basis of the Chars 196 3.22 Development and Use of F aná R Chars 206 5-2,3 Charts Based on Standard Vatdes 212 5-2.4 Incerprezaton of X and A Chans 214 5.2.5 The Effees of Nonrormality on X and RChas 216 5.2.5 The Opetating-Characteristo Function 217 5:9.7 The Average Run Length for bexChar 220 5-3 Control Cherts for 7 ands 222 5.34 Construction and Operation of Z and sChets 222 5.3.2 The E and s Control Charts with Veriable Sampte Size 227 5:33 The” Control Char 231 5.4 The Shevibart Control Chars for Individual Measurements 231 5-5 “Stramary of Procedures for £, R, and s Chars 242 5.6 Applications of Variables Control Charts 243 CHAPTER 6 Control Charts for Anribures 265 Chepter Overview and Leaming Objectives 265 6-1 Inmoducion 266 6-2 The Control Chert for Practoa Nonconforming 266 6-2. Development aad Operation of the Control Char 268 6-2 Varisble Sample Size 280 6-2.3 Nonmenufaciwring Applications 284 6-2.4 The Operating Characteristic Function and Average Run Length Calculations 286 6-3 Controy Char's for Nonconformitios (Defects) 288 Ex 63.1 Procedures with Constant Sample 89 63.2 Procedures with Variable Sample Size 298 6-3.3 Demerit Systems 300 634 The Operating-Characieristic Function 302 633 Dealing with Low Defeer Levels 304 63.6 Nonmanufacturing Applications 306 fd Choice Between Armibutes and Variables Control Chars 306 6-5 Guidelines for Implementing Control Chars 311 CHAPRER 7 Process and Measurement System Capability Analysis 326 Chapter Overview ané Leaming Objectivos 326 7-1 Introduction 327 7-2 Precess Capebility Analysis Using a Ilistogram or a Probabilisy Plot 329 72.1 Using the Histogram 329 7.2.2 Probability Plotting 331 73 Process Capability Ratics 333 73,1 Use and Interprotaticn of Cp 333 73.2 Process Capabilivy Ratio fot an Off-Center Process 328 733 Normality and the Process Capabiliry Ratio - 339 7:34 More about Process Centering 341. 7:25 Confidence Iitervals and Tests on Process Capabiiiy Ratias 343 3-4 Process Capability Analysis Using a Control Chan 349 Process Capabiliy Analyeis Using Designed Bxperiments 351 76 Gauge and Mensurement System Capabilicy Studies 352 7-6.1 Basic Concepts OF Galge Capabiltty 352 7.62 The Analysis of Variance Meihod 358 7-6.3 Coniidence Intervels in Gauge R & R Sudies 362 +64 False Defectives and Passed Defectives 364 77 Sertng Specificadon Limits on Diserere Components 367 TF Linear Combinadons 367 72 Nomlineer Combinations 371 coNTENTS iii 7-8 Estiimacing the Natural Tolerance Limits of a Process 374 7-8.1 Tolerance Limits Based on the Normal Distribution 374 7-8.2 Nonpárametric Tolerance Limits 375 PART HI Other Stavistical Pencess-Manitoring and Control Techniques 383 CHAPTER 8 Cumulative Sum and Exponentially Weigheed Moving Average Control Charts 385 Chapter Overview and Leaning Objectivos 385 8-1 The Comelative Sum Contrel Chart 386 81.1 Basic Principles; The Cusum Control Chart [or Manitoring the Process Mean 386 8-1.2 The Tabular or Algoritâmic Cusum for Monitoring the Process Mean 390 8-13 Recommendations for Cusum Design 395 &-1.4 The Standardized Cusum 397 Improving Cusum Responsiveness for Lorge Shifis 398 8-1.6 “The Fast Initial Response or Headstart Pestere 398 One-Sided Cusums ad0 8:16 A Cusum for Monitoring Process Vasiabilicy 401 8:19 Rational Subgroupe 402 8110 Cusums for Other Sample Statistics 402 8-1.11 The VMask Procedure 403 82 The Exponentislly Weighted Moving Average Control Cher 405 8:21 The Exponentiaily Weightea Moving Average Contra] Char: for Monitoring the Process Mean 406 82.2 Design of an EWMA Control Chart 411 823 Robusmess cf the EMA to Nannormality 412 82.4 Rational Subgrovps 413 2.5 Extensions of the EWMA 413 83 The Moving Average Control Chart 447 7 connams ' “BAPTER 9 tbeç Univariare Statistical Process Monitoring and control Techniques 423 haprer Overview and Leaming Objectives 423 Statisdical Process Contro] for Short Production Runs 424 9-1 x end R Charis for Show Production Runs 425 9-1.2 Amibutes Control Charts for Short Productos Runs 427 92-13 Other Metheds 428 Mudified and Acceprance Control Chars 429 Modified Conerol Limits for the E Chan 429 3:22 Acceprance Control Charts 433 Control Chars for Muhiple-Seream Processes 434 9-3.1 Mullple-Btream Processes 434 8:32 Group Control Chars 434 9.3.3 Cther Apprraches 437 SPC With Antocarrelated Process Data 438 9-4.1 Sources and Effects of Autocorreletion in Process Deta 438 5-4.2 Madel-Based Approaches 443 9.4.3 A Model-Pree Approach 452 -5 Adaptive Sampling Procecures 485 -6 Bxonotuio Design of Control Charts 457 9.5.1 Desigring à Controt Char 457 9.6.2 Process Characteristics 458 9.63 Cost Parameters 458 9.64 Early Work and Semiecoaomie Designs 460 9-6.5 An Economic Model of the * Control Char 461 9-6.6 Olher Work 466 "7 Cuscore Charts 467 +8 The Changepoint Model for Process Monitoring 470 +9 Overview of Other Procedures 472 59.1 Toc Wear 472 9.9.2 Control Charts Based on Quher Sample staústics 473 9.53 Fi Contol Problems 474 994 Preconmol 474 9.9.5 Tolerance Interval Contro! Charts A76 9.9.6 Monitoring Processes with Censored Dam 476 2.9.7 Nonparanenio Consro) Charts 477 + CHAFTER 10 Multivariate Prncess Monitoring 2ad Con] 486 Chapier Overview anê Learnlag Objectives 486 10-1 The Multiveriate Qualiy-Coatrol Problema 487 202 Descripticn of Multivariare Data 489 10-2.1 The Multivariate Normal Distribution 489 10-2.2 The Sample Mean Vector and Covarience Metix 490 1€-3 The Hotelling T? Control Char 49] 20-31 Subgrouped Data 491 10-32 Individual Observations 500 10-4 The Multvariaie EWIMA Control Cher 504 10-5 Regressioa Agjusment 507 10-6 Contro: Charts for Monitoring Variabiliry 51% 10.7 Latent Stmicimre Methods 513 20-7.1 Principal Components 514 10-72 Partat Least Squares 519 20-8 Profile Monitoring 520 CHAPTER 11 Engineering Process Contro) and SPC 526 Chapter Overview and Leaming Objectves 527 -1 Process Monitoring and Process Regulation 527 -2 Process Control by Sesdback Adjustment 526 11-2.1 A Simple Adiustment Sckeme: Integra! Conol 528 44-22 The Agjusment Char 535 11.2.3 Variaons o: the Adjusument Chert 536 1)-24 Onher Types of Peedback Comtrollecs 540 11-3 Combining SPC and EPC 541 PART IV Process Design and Ieeprovement with Designed Experiments 547 CHAPTER 12 Eactorial and Erctionat Pacrorial Experiments for Process Design and Improvement 549 Chapeer Overview and Esaming Objectivos 349 12-14 What is Experimenta! Design? 550 12-2 Examples of Designed Experiments In Process improvement 551 12:3 Guidelines for Designing Experiments 12.4 Faciorial Experiments 557 1241 An Example 560 1242 Statistical Aseiysis S60 124,3 Residual Analysis 566 225 The 2º Faciorial Design 567 125,4 The 22 Design 567 125.2 The 2º Design fork 23 Factors 873 12-53 A Single Replicate of the 2 Desigo 585 12-54 Addírion of Center Points to the 2 Design 580 12-55 Blocking and Confounding in the 2Desim 393 12-6 Fractional Replicaton of tho 2*Desizn 595 12-6.1 The One-Half Fraction of che Design 595 126.2 Smaler Fractons: The 2º Practional Factorial Design 602 CHAPTER 13 Process Optmizados with Designed Experiments 611 13-1 Response Surftce Methods and Designs 612 13-11 The Method of Steepest Ascent 614 13-12 Analysis of a Second-Oçder Response. Surface 616 132 Process Robusmess Studies 621 13-21 Background 621 132.2 The Response Surface “ Process Robusimess Studies 13.3 Rvolntionery Operation 631 PART V Acoeprance Sampling 643 CHAPTER 4 Lot-by-Lot Acceprance Sarmpling for Attributes 645 Chapter Overview and Leaming Objectivos 645 14-] The Acceptance-Samplins Problem 646 14-1.1 Advantages and Disadvantages of Sampling 647 14-12 Types of Sampling Plans 648 t4-1.3 LotFormadon 649 14-14 Rendom Sampling 649 14-1.5 Guidelines for Using Acceptance Sampling 650 14.2 Single-Sarmpling Plans for átuibutes 652 14-2.1 Definition of a Single-Sampling Plan 652 14-22 The OC Curve 652 conrewrs xy 14-23 Designing a Singlo-Sampling Flan sith a Specified OC Curve 657 14-2.4 Rectifyivg Inspection 658 14-3 Double, Multiple, and Sequential Sempliag 662 143.1 Double-Sampling Plans 662 14-32 Muliple-Sampling Plans 668 143.3 Sequential-Sampling Pians 669 244 Military Standard HOST (ANSVASQC ZL.4, ISO 2859) 672 14-41 Deseription of ie Standard 672 14.2 Procedure 674 14-4,3 Discussion 679 24-5 The Dodge-Romig Sampliog Flans 681 14-5.1 AOQL Plans 682 145.2 LTPD Pins 685 14-5,3 Estimation of Process Average 685 CEAPTER 15 Other Acreptance-Sampling Techniques 688 154 Acceptance Sampling by Varidbles 689 15-11 Advantages and Disadvantages of Variables Sampling 689 15-12 Types of Samoliog Plans Aveilable 690 15-13 Cantion in the Use f Variables Sampling 691 15-2 Designing a Variables Saxapling Plan with a Specified OC Curve 692 15.3 MIL STD 424 (ANSUASQC 719) 694 25:31 General Descripion of the Standard 694 2532 Use ofthe Tables 695 1533 Discussion OF MIL STD 414 end ANSUASQCZIS 698 15-4 Other Varigbles Barmpling Proceduces 699 154,1 Sesmpiing by Variables to Give Assurance Regarding the Lot or Procass Mar 699 15-42 Sequentiat Sampling by Variables 700 15-5 Chain Sampling 700 15-6 Contímious Sampling 702 156,1 sp 703 15.62 Other Continuous-Sampliag Plans 705 15-7 Slip-Lcr Sampling Piaos 706 Appendix 713 1 Summary of Common Probability Disuributions Often Useá in Statisticat Qualy Control 745 EL. Cumuiative Standará Norma Distribution 716 UU. Fercentage Points of the x? Distribution 718 IV. Percentage Points of the : Distribution 719 Y. Percentago Points of the E Distribution 720 o. O mm xy conTENTS t ' VI. Factors for Constructing VasiaDles Control “Chars 735 . VIT, Factors for Two-Sited Normal Tolerance. Limits 726 VII, Factors for One-Sided Normal Tolerance Limits 727 Quality Improvement in the Modem Business Environment Bibliography 729 Answes 10 Selecied Exercises 713 Index 743 CHAPTER OUTLINE 1.1 THEMBANHNG OF QUALITY AND QUALITY 1.4 MANAGEMENT ASPECTS OF QUALITY IMPROVEMENT IMPROVEMENT 11.4 Dimensions of Qualicy 14.4 Quality Philosoghy 1-1.2 Quality Engineering Terminology and Managemene Strenegies 1.2 A BRIEF HISTORY OF QUALITY CONTROL 142 The Link Berween Qualicy and Productiviey AND IMPROVEMENT . 143 (Qualicy Coses STATISTICAL METHODS POR QUALITY 1.44 Legal Aspeecs of Qualiry CONTROL AND IMPROVEMENT 14,5 Implemencing Qualicy Improvement CHAPTER OVERVIEW AND LEARNING OBJECTIVES This book is about the use of statistical methods and other problem-solving techmiques to improve the quelity of the products used by our society. These products consist of mannfactared goods such as automobiles, computers, and clothing, às weil as services such as the generation and distribution of electrical energy, public transportation, banting, retailíng, and health care. Quality improvement methods can be applied to any ares within a company or organization, including manufactarng, process development, engineering design, finance and accounting, marketing, distribution and logistics, and field servics of produeis. This text presents the technical tools that are needed to achieve quality improvemeny in these organizations. 1a this chapter we give the basic definitions of quality, quality improvement, and other i qualiy engineering termino.ogy. We also discuss the historical development of quality i improvement methodology and overview the statistical tools essential for modem profes- Í sional practice. A brief discussion of some maragement and business aspects for imple- menting quality improvement is also given, Ater careful study of this chapter you should de able to do the following: Define and éiscuss qualãty and quality improvement Discuss the different cimensions of qualiry Discuss the evolution of modem quality improvement methods Discuss the role that variability and statistical methods play in controlling and improving quality 5 Desenibe the quality management philosophies of W. Edwards Deming, Joseph M. Turao, and Armand V. Feigenbaum Bum CHABTER 1 QUALITY IMPROVEMENT IN THE MODERN BUSINESS ENVIRONMENT 6. Discuss total quality management, the Malcolm Baldrige National Quality Award, Six-Sigma, and quality systems anc standards, 7. Explain the links between quality and produetviry and between quality and cost 8. Discuss product liabikity Discuss the three functions: quality planning, quality assurance, and quality control and improvement 4-1 THE MEANING OF QUALITY AND QUALITY IMPROVEMENT x We may define quality in many ways. Most people have a conceprual understanding of queliry as relating to one or more desireble characteristics that a product or service should possess, Although this conceptual understanding is cerainty a Useful atming point, we veilt give a more precise and usefil definition. . Cuality has become one of the most important consumer decision factors in the seles- Hon ámang compening products and services. The phenomenor is widespread, regardless of whether the consumer iz at individual, am industrial organization, a retail store, or a mil. itary defense program. Consequently, understanding and improving quality key factors are leading to business success, growih, and enhanced competiiveness. There is à qubstantial remim on investment from improved quality end from successfuliy eraploying quality as an integral past of overall business strategy Tn tis section we provide operation! defii- tons É quality and quality improvernenc, We begin vdth à brief discussion of the differ- ent dimensions of quality and some besic terminology. 1.1.1 Dimensions of Quality “Lhe quality of a produci cam be evaluated in several ways. eis often very importancio dif. feentinte these different dimensions of quality. Garvin (1987) provides am excellent cis- cussion of eight components or dimensions Of quatiiy. We sumamarize his key poinis conceming these dimensions of qualie as follows: 1. Performance (will the product do the intended job?) Potential customers usually evaluate a producr to determáne if it will perform cer. taia specific functions end determine how ywell it performs them. For example, you conlá evaluate spreadsheet software packages for à PC ta derermine which data manipuletion operations they perform. You may discover thai one outper- forms another with respect to the execution speed. Reliability (how often does the producer fail?) . . Complex products, such as many appliances, automobiles, or eixplanes, wil usu- aliy require some repair over iheir service life. For example, you should expert that am automobile will require occasional repair, but if the car requizes frequent cepair, que say that it is unreliabls. There are many industries in which the cus» tomer's view of quality is greaily impacted by the relZability dimension of qualiry. 3. Durability (how long does he product last?) “This is the effective service life of “he product. Customers obviously want prod- ucts that perform sarisfactorily over a long period of time. The automobile and » te THEMBANING OF QUALITY AND.QUALITS IMPROVEMENT 3 major appliance industries are examples of businesses where this dimension of quality is very important to most customers. 4: Serviceability (how easy is it to repair the product?) There are many industries in which the customer's view of quality is direcily influenced by how quickly and economicaliy a repair or routine maintenance activity can be accomplished. Examples include the appliance and antomobile industries and many types of service industries (ow long did it tale a credit card compamy to correct af esror in your bi?) 8. Aestheties (what does the product look like?) “This is the visual appeal of the product, often talking into aceoust factors such as style, color, shage, packaging aematives, tackile characteristics, and other sen- sory feacures. For exmnple, soft-driak beverage manufacamers have relied on the visual appeal of their packaging to differentiate their product from other com- petitors, 6. Features (what does the producr do?) Usually, customers associare high quality with products thar have ádded features; tharis, those thar have features beyond the basie performance of the competition. For example, you might consider a sprendsheet sofisware package to be of shpe- rior quality df ix had builtin statistical analysis feamres while its competitors did not. ? Porceived Quality (what is the reputation of the company ar its product?) in many cases, customers rely on the post repuratior of the company concerning quality of its products. This reputation is directly influenced by failures of the produet thar are highly visible to the puhlie or that require product recalls, and by how the customer is trented when & quality-related problem with the product às reported. Perceived quality, customer loyalty, and repeated business are closely interconnected. For example, if you make regular business trips using à particular aíriine, and the flight almost always amives on time and the airline company does not lose or damage your luggage, you will probably prefer to fiy om hat carrier instead of its competitors. 8 Conformanee to Standards (is the product made exactly as the designer intended?) We usually think of a higa-quality product as one that exactly meets the require- ments placed on il. For example, how well does the hood Gt on a new car? Is it periectly flush with the fender height, and is the gap exactly the same oa all sides? Manufactuzed parts that do not exactiy meet the designer's requirements can cause significant quality problems when they are used as the components of a more complex assembly. An automobile consists of several thousand paris. If each one is just slightly 100 big or wo small, many Of the components will not fit together properly, and the vebicie (or its major subsysteme) may nor perform as the designer intended. We see from the foregoing discussion that quelity is indeed a muitifacered entry. Conseguentiy, a simplé answer to questions such às "What is qualiry?” or “What is quality improvement?" is not easy. The traditional definition of quality às based on the viewpoimr that products and services must meet the requirements of those who use them. 4 CAAPTER 1 QUALITY IMPROVEMENT THE MODERIM NUSÍHESS ENVIRONMENT Definition Quality means fitness for use, +rhete are tra general aspecis of Siness for use: quality of design and quality of con- formance. All goods and services are produced in various grades or levels o? quality. These vasiarions iu grades or levels cf quality are intentional, and, consequently, the aprogriate technical term s quali of design. For example, all automobiles have as their basic objective providing safe transportation for the consunf” However, automobiles dif- fer with respect to size, oppointments, epesrance, and performance, These differences are fo rege of intentional design differences berween the Eypes o? automobites. These design Vsesences include the types of aterialo used in constramticn, specifications or: the com- ponents, relinbiity obtained ihrough engineering development of engines and drive trains, and other accessories or equipment. Tte quality of conformanco is how well the produc: conforms 10 the specifications sequited by te design, Qualy oÉ conformanse is influenced by à number Of factors, including the choice of manufectucing procestes, the training ané smpervision of the work» force, the types Of process comrols, tests, and inspection activities that are employed, the extent to which these procedures are followed, and tie motivaúon of he «woriforce to aehieve quality. Tafortunatety, dhis definition has become associeted mare «with the conformence aspect of quelity than seita design. This is in part ue to the lack of formal education most designers and engincers receive in quality engineering methodology. This also leads to much Jess focus on lhe customer and more oÉ& uconformance-to-specifications” approach ro quality regardiess of whether the product, even sihen produced to standards, was act- atly *Bforauee" by lhe customer. Áleo, there is sal] a widespread belief that quality às a probjem dia cam be deal veith solely in manufactncing, or that the only way quality cam de improved is by "gold-plating” tho produet. We prejer a modern definition o? quality: Desipition i Quality is inversely propordonsi to variabiliy. Note that this definion implies thar if variability in the important charactesisdes of à product decreases, the quality vf the product increases. sam example of the operational effvotiveness cf this definition, a few years ago, ne of the automobile compênies in the United States performed a ecmpecative study of a er Ao Lg Saca o Ne vaic Tordo dna dc vara jo aersally good As são poa fer Bob Hop as pone out," real He Chinese od bt | doa't wa to as it very mig 1 THE MEANING OF QUALITY AND QUALITY IMPROVEMENT S qe United Siates Tai Teger UsT States Figure 1-3 Warcanty costs for mensemissions. Figure 1-2 Disvitudis leal alma n igure 4-2 Distibutions of ori axa OE critical dismensigus for a transmissi i “ smsmision dem vma manufactered in a domestic plant and by à Japanese supplios A ali o any lia on repair cons indica é des vas 2 ing dE fem production, with the Japarese-prodi i navio co asso o 1, A pacote Gay to desova cao E his in cost and performance, the compa e É À ny trensmissions from each plan, disessembled tbem, Asa sore al quai Hansi f , nd measured several critical quality Figure 1-2.is i SE 2 general recanto eres o ty. Note dt de di aracteristies for lhe transimissi i à dusen o + isão missions manufactured in the Uni Sis us tp Sho 15% f he aid of de apenas ping ha very Temnoãe cone vt ven be produced, Ta fact, the plant was produciag at a quality level Sit vs it go, Eno o he generally asp vi quali ii de comp Bloveur, ie Fepaneso lan produced tranamissione for which the same ctícal câmeis eo vm 2 bout 25% fe speifcanon band. As ecl, ico ico derabty | : e critical quality charactecistá ge-buil trems- isso in eoeepaisor so those built he aa Ga Ee Japan tao E are tw > ms er ae tvi quere Be; Wy lêo pemas do lj? Hom ley da a ao e “why” question is obvious frora examination of Fig. 1-1. Red ed vaca ha rent txinsiated Jo lover cost, Furtenner, he Iapaniatsoilstrane risons sed gens soc sol ca sore quis, nd ue general pec ty dh ester as suport oe bt domesaly. Fer eai mê enty clama me le and the reduction of wasted time, eff y ro : ic fort, and money. Thus, quali y proportional to variability. Furthermore, it can be comimunicated Ve de in a language that everyone (particularly money. eryone (particularly managers and executivos) understands—nemely, How di je lid the Japanese do this? The answer hes in the systematic and efigotive use of the methods desedbed in this hook, Tt als s fe n of quality ' sedbed in this hook. à ok. Tt also leads to the following definidos K Definition Quality improvement is the reduction oÉ variability in processes and products. 6 CHAPTER: QUALITY IMPROVEMENT IN THE MODERR DUSINESS ENVIRONMENT Excessive variability in process performance often restlts im waste, For example, com- sider the wasied money, time, and eifost that is associated will the repeirs represented is Eig. 1-1. Therefore, an altemare and frequently very useful definition is char quality improvement is ho reduction of waste. This definition is particularly estective in servico industries, where there may norbe às many things that cam be directly measured like the transmission critical dimensions in Fig. 1-2). In service industries, 4 quality problem may be an error or a mistake, the correction of wiich requises eifort and expense, By improving the service process, this wasted effort ané expense cam be ded, q now present some quality engineering terminology that is used throughout the dock. X 11.2 Quality Engineering Terminology Every produer possesses à numiber Of element that joindy describe var the user or com- cumes tainkes of 28 quality. These parameters are often called quality characteristics. Sometimes these are called critical-to-qnality (CTQ) characteristics. Quatity charactec. istios may be Of sevoral types: 1. Physical: length, weight, voltage, viscosity 2. Sensory: mete, appearance, color 3. Time Orientation: reliaDiliry, durability, serviceahility Ncte that the different types of quality characteristics can relate directly or indirecily to the dimensions of quality discussed in the previous section. o o Quility engineering is lhe set of operational, managerial, and engineering actvicios that à company nses to ensure that the quality characteristics of a product are at tbe nom inal or tequired levels, The techniques discussed in the book form much of the basio meibodology used )y engincers and other technical professionals to achieve these goals. Most organizations find it difficult (and expensive) to provide the customer with producis that have quality charaeterisios that are always identical frara unit to uni:, or dee at levels that match customer expectations. A majar reasoa for this is variability. There is a certain amount of vasiabiliry in every product, consequently, no two producis are ever identical. Por exastple, the thiclness of the blades on a jet furbine engine impeller is not «identical even on the same impeller. Blade thíclmess vil also diffez between impellers. TÉ this variation in blade thickness is small, then K may have no impact on tie customer. However, if the variation is large, then the customer may perceive the unit to be undesir- able and unacceptable, Sources of ihis variabiliry include differences in materials, difer- ences in he performance and operation of the mantfncturing equipment, and difrencos is the way the operators pesforra their tasks. This line Of thinking Ted to the previous def- átion oé qualiry improvement. e a can osly be described in statistica] terms, stadistical methods play central role in quality improveraent efforts. In the application of statistical methods to quality enginsering, it is fairls typical to classify data on queliy chavanteristes as either attributes or variables date. Variables data are usually continuous measurements, such as length, voltage, or viscosity. Auributes duia, on the other harid, aro usually discrete dae, often taking the form of counts. We will Gesorbe siatistical-bascd quality enginecring tools jor desling with both types of data. 154 TREMEANING OF QUALITY ANDLQUALTY DPROVBMENT 7 dty characteristics are often evalunted relative to specifications. For a manufao- fured product, the specificarions are the desired measurements for the Quaiity characteris- tios of the components and subassemblies that make up the aroduct, às well as the desired values for the quality characteristics in the final prodict For example, the diameter of à shaft used in an ausomobile ranemission canno: be t00 large or ir will not Bt into “he mat- ing bearing, nor can it be tea small, resulting in 2 loose À, causing vibration, wear, and early failure of the assembly. In the service industries, specifications are typically in terms Of the maximum amount of time to process am order or to provide a particular service. A value of a measurement that corresponds to the desired value for that quality char- acteristic às called the nominal or target value for thar characteristic. These tasget values are usually bounded by a tange af values that, most typically, we believe wall be suffi- clentiy close to the target so 25 to not impact the function or performance 0€ the product ifthe quality characteristic is in that range. The Incgest aliowabie value for e quality char- acteristic is called the upper specification limit (USL), and the smailest allowable value for a quality characteristic is called the lower specification limit (LSL). Some quality characteristics have specification limits on only one side of the target. For example, the compressive strength of à component used in an automobile bumper Hke)y has à target value aná e lower specification limit, bué nor ar ugper specification Ymit. Specification are usvally che resulr of the engineering design process for the product. Traditionaliy, design engineers have arrived at é product design configuration through the use of engineering science principles, wltich aften results in the designer specifying the targer values for the critical design partmeters. Then prototype construction and testing follow. This testing is often done in à very unstructured manner, without the use OÉ statis- tically based experimental design procedures, and without much interaccion with or Imowledge of the manvfacmring processes that must produce the component paris and final product, However, through this general procedure, the specification limits are usually determined by the design engineer. Then the final product is released to manufactucing. We refer ta this ps the over-the-wall approach to design. Problems in produet quality usually are greater when the avercthe-wall anproach to desiga is used. Tm this approach, specifications are often set without regard to the inberest variability out exists in materials, processes, and other parts of the system, wEich results in components or products thar are nonconforming: that is, that fail to meet one or mors of its specifications. À specific type of failure is called a nonconformity. A nonconform- ing product is not necessariiy uh for use; for example, a derergent may have à concen- tration of active ingreCients that is below the lower speciloadom Umit, bt it may still perform acceptably if the customer uses à greater amount of the produet. A nonconform- ing produe: is considered defective if it has one ar more defects, which are nonconformi- tes thar are serious enough to significantly affect the safe or effective use of the product. Obvionsly, failure on the port of a company to improve its manufactring processes can also cause nonconformities and defects. “The over-the-wall design process has been the subject of much attention in the last 20 years. CADICAM systems have done much to antomate the design process and to more elfzetively translate specifications into manufacturing acévitics and processes. Design for manuacomrabiliry and assembly has emerged as am important part af overcoming the inherent problems with the over-the-wall approach to design, and most engineers receive some background on those areas loday as part of their formal education. The recent emphasis on concurrent engincering hos stcessed o feain approsch to design, with spe- cialists in manufucturing, quality engineering, and other disciplines working together with CUAPISR 1 QUALITY IMPROVEMENT EN'THE MODERN BUSINESS ENVIRONMENT the produce desiguer ar the eaiest stages of the product design process. Eurbermore, the dono Use of lhe quality improvement methodology im this boo. at all leves of tre process used in produer desiga, development, and manufheturing, plays a crucial cols in quality improvement. : 12 A BRIEF BISTORY CF QUALITY CONTROL AND IMPROVEMENT Couaticy alves has beer am integoal part of virtually al products and services. However, our ewareness of its importance end tho insxoduetion af formal me:hods for quality con- ttol and improvement have beca an evolitionary development. Table t-1 presents a time- fine OF some of the important milestones in this evolutionary process We will bricfly discnss some of the events on this timeline, Broderole W. Taylor introduced some prieíples of seiertião management 8 mass production industries began to develop prior to 1900, Taylor pioneered dividing work into Task so that the product could be manufactured and assembled more easily. Elis work led to substantial improvements in productiviry. Alea, because of standardized production and assemblv methods, the qualicy of mansfactured goods tas positiveiy impacted as well. However, long with the standardization of work methods came the conter of work undando a standard tme to accomplish the wark, or à specified number OE jmits that “aust be proquczd per period. Erantk Gilbreth and others extended ins concept to the study cf motion and work design, Much of chis had a positive impact em productíviry, but it oftem de emphasized the quality aspect of worie Furthermore, if caniod to extreaa, “work stan- durds Have the risk of halting innovation and continuous improvement, «ich we recog- nize today as being a vital aspect of all work activities. Sratiatiea methods and their applicadion in quality improvement have ha a long his- roxy: da 1994, Walter À, Shewhart oÉ the Bell Telephone Laboratories developed lhe sta- sical control-chast concept, which is often considere ths formal beginning of statistical cquelity control. Toward the end of the 19205, Harolá F. Dodge and Harry G, Romig, both dE Bell Telephone Laboratories, developed ataisticady based aorennnma sampling as am Altemative to 100% inspection. By the middo of the 19305, statistical quality-control Aethods were in váde use at Westem Electric, the mamufactacing arm of the Bell System. However, the value of statistical quali:y control was not «widely recognized by industry wWocld War TI saw e greatly expanded use and acceptance of statistical quality-conmol. . concepts in manufactusing industries. Wartime experience made it Apparert thar statistical leclmiques were necessary to control and improve produei qualizy. The American Society for Quaiity Conttol was formed in 1946. This organization promen the use OF quality improvement techniques for al types of produeis and services, It offers a numbe: of con- feineta, technical publications, ant tnining programs in quality assuesan The 1950s and 1060 sato the emergence of reliability engineering, the introduction of several impo- dt cextnoohs on statistical quality control, and the vicwpoini that quality is à way of man aging the organization. ho 1050, designed experiments for produet and process improvemen! medo first introduced ig the United States, The infiei applicatious were in fe chemical industry. These memods were svidely exploited in the chemical industry, and :hey arc often cited as one Of the primary rensous that the U.S. chemical industey is one of the most compeútive the world and tas lost ne business o foreign companies, The spread of these methods i 1 13.4 BRIBF HISTORY OF QUALITY CONTROL AND IMPROVEMENT Table 1.1 A Timeline of Qualicy Merhods tro tono as - 1900-1930 1907 1907-1908 1908 1915-1819 1918 19208 1922-1923 1924 1928 1934 172 1m2-1933 1933 1938 1940 1940-1943 1942 1942-1946 1944 1946 1846-1949 1948 1950 19505 Quait js largely determined by the eáfore cf an individual crafiman Bi bio nodes mas Iecangeso pasto simply assembly |. Taylor iniroduces “Scientific Mana; E” princir vi A a E e gement” principles to divide w cast pocomplihed uno er exrent) acceptance sampling. Ta adéfti techniques, à number of other statistical (o: i x ade ue, ols are useful in ansfyzing quality problems and improving pe performance of prodnetion processes. The role of same of these tools is pu im E tê, which presexts a production process as à system with à set of inputs oúitpi. The inputs Xy gs... «2, Are controllable factors, such as termperainres, pres- aures, feed rates, and other process vari i process variables, The inputs Zy Xp - -« 2, are uncontrollable Contraiobla inputs t TMigasirernam e. + Elgin | Moslging : + . contro : 1 Bros ————+ [= val coorte Output Product neon pars Eigure 13. Produecon process inpoo end our EST UE O me TO er Ve . 13. STAFISTICAL MEIHDDE PO QUALITY CONTROL AND IMPROVEMENT 13 as USINESS ENVIROISY 2 CHASTERI QUALITY MPROVEMENT IN THE MODER! ad (or difficult o contro) inputs, en as environmental factors or prego fa ri s |. e A ã +turing process transfomms a e as se tos “The output variable y às à measure | E / A / Exished product that has several quality o + e quali. . SH proces cui art is one of me arimary techniques of statistical process contei or Í “s . dot chart is shown in ig 1-4. This hantplots the avernges ot mel e = : . RE que eis samples token from ae process versus fime (or he : : y E a as à center line (CL) nd upper and lower control limite (UCL. , e A a | jet). The chart ha A im AN crio E 1.6), he center line represents were tis process charaeterisão should | : la aro deterstined fal] if there are no nnusual sources of variability present Me comeal Costa A 5, e it is iderations thot we will discuss im Ds o. From some simple statistical consi dias ja Chapem 4 5 aut ie Ji! to the output variable(s) in à sy Classically, control chaxis are appl do as sea i ey cam be úsefiliy applied to the in Fig, jed, However, in Some cases epic o beto k v g technique; control chart is à very vseful process monil to] e DS coa of vaia are presen, sample average vá pcensiãe de cota To ei 5 id be made and co: a some invesgarion of the process should be made and ico 1 o Unusual sources of variability taken. Systematic use ué a control ch to reduce variability. a aa emo experiment is extremely helpful in discovering the der es a datos of à . A designed expo à ly oharactenistic of interest in the process. À des r a o stemadioly vanng the conirollable input factors in the ay duct parameters. j ect these factors have on the output product E stcaly je spesimems doe invalinlo in reducing the variability in the quality cierageris ln th levels oé ho controle vaiabis that opímize prertmê mu fsemance. Olten significant brecktarougis in process performance and product q) iram using designcá experiments aaa . o o type of designed experiment às the factarial design, m vaia fa i ao rether in such a way that all possible combineiions o factor a me a Eleito 13 shows two possible facrorial designs for the process in Fig. reis ad É =2andp=3 cortrollable factors. 1n Fig 1-5a the Tactors have two] a Pa na posible est conbinacion nfs fctoril oxptrimem Mom CE TT Quarto 1a Fig. 1:55 there are fee factors each at wo lee, via em experiment od eg rest combinadons arranged arthe comes ofa cube. Tae isinaos sho er ess 0É the cube represent the process performance at each combination of ue + et Sarmpte average “Time tor sample numbarl Figure 14 A cypioal control chart fel Tua actors, =, ar x (19 Tha dacioe, x, ag, nd sã Figare 1.5. Factors designs for he process in Fig. 1-3. factors xy, Xy, and 2, 1 is clear that some combinaúôns of factor jevels produce berter vesults than others, For example, increasing x, from low to high increases the average level of the process output ana could shift it off the target value (7). Firthermore, process vari- ability seems to be substantialiy reduced when we operate the process along the back edge of the cube, where 2, and à; are at their high levels. Designed experiments are a major off-line quality-contral tool, because they are often used during developmen activities and the enriy steges of marnfacturing, rather than as à routine on-line or in-proceas procedure. They play a crucial role in reducing vasiabiliry. Once we have identified a lise of importam varisbles that affect tis process output, it is usually necessary; to model the relationship berween the influential input variables and the output quality characteristios. Statistical techniques useful in constructing such mod- ls includ> regression aralysis and éme series analysis. Deteiled discussions of designed exgeriments, tegression analysis, and time series modeling are in Montgomery (2001), Montgomery, Peck, and Vining (2001), and Box, Jenkins, and Reinsei (1994). When the important variables have been identified and the natere of the relationship beswsea the important variables and the process output has been quentifisa, then an on-line statistical prooess-control technique for monitoring and surveillance of the process can be employed with considerable efeciiveness. Techniques such as control charts can he nseé to monitor the process outpnt and detecl when changes in the inputs are required to bring the process back to am in-control state. The models that relate the influentia! inputs to process untpuis help determine the namre and magnitude of the adjustments required. in many processes, once the d$namic nature of the relationships between the inputs and the cutpuis are understood, it may be possible to rontinely adjust the process so that fare values of the product charanteristies will be approximately 09 target. This rontine adjust- ment is often called engineering control, automatic control, or ferdback control. We «will brigf£y discuss these Iypes of process contro] schemes in Chapter 11 and illustrate how Statistical Process Control (or SPC) methods can be successfully integrated into a manu- Tacturing system iv which engineering control is in use. The third arca of quality control and improvement that we discuss is acceptance sam- pfing, This is closely connecied with inspection and testing of producr, which is one of she earliest aspects of quality control, dating back to long before statistical methodology was developed for quality improvement. Inspection can occur at many points in à process. Acceptance sampling, defined as the inspection and classification of a sample Of uníts selected at random from a larger batch or lot and the ultimate decision about disposition uá CRABTER 3 QUALITY IMPROVEMENTIN THE MODERN BUSINESS ENVIRONMENT of the Jor, usually occurs at avo points: incoming raw materigis or eorrponents, =x fisial production. . “Seral different vasasions of acceptance sampling are shown in Fig, Ló la Fig. 1-5a, the inspection operation is performed immediately following production, before the prod- ct s siipped to the customer. This is usually called ontgoing inspection. Figure 1-6b “ihustcates incomíng inspection; Shar, à simadon io wiich lots of hatohes of product are sampled as they are received irom the suppller. Various lor-dispositioring decisions are Alustrated in Fig. 1-6. Sampled lots may either be accepted or rejected, Items im arejecied jot ave typicaliy either acrapped or recycled, or they may be reworked or replaced with gpod usb. This latter case is ofien called rectifying inspection, 'Madem quality assusanos systems usually place less ermpltasis om acenando sam pling and estemnpt to malke stadsdical process control amd designed experiments the focus rlheirefforts. Acceptance sampling tends to reinforce lhe “conformande to specification” view of quality and does not have any feedback into cither the production process ur enginsesing design or development that wontá necessanly lená to quality improvement. “iguze 1.7 shows the typical evolution in the use ql these mclaniques in most organi- ratono, At the lowest level of many, management may be completely avandã of quality issues, and lhece is ihely to be no effective organized quality improvement effort. Frequently there wil be some modest epplicatons of acoeptance-sampling and inspection methods, usually for incoming parts and metexíals. The firs: activity se maturity increases jeto intensify the use of sampling inspection, The use of sampling wall increase until we sealize that qualiey cannot be irapectod or tested into the product dat that point the organization usually begins to focus on process improvement statisdeal process control and experimental desiga potendally have malte impacts on Tanifactusing, produei design aetivitss, and process development, The systemano intro- quetion of these methods usually mars the ste cf aubstantal qualith cos, and produe- tvi improvements in he organization. Ae the highest levels of matuci.y, companies use Aueigned experiments and satistical process cool meibiodo intensively and make rela tively modest use of acceptance sawplir “The primary objective of quality engincering eiforo às the systematio reduction of cariability iu che Key quality characteistics of the product. Figure 18 shows how his happens over time. In the early stages, Wheu acceptance sampling is the major technique jn use, process “fallout or units that do not conform to he specifications, constitute à rig percentage É the process antpet. The intoduetion o statistical process contol will FREE qe tm) Duigoing 'nspectior sip 4h feceivingncaming inspection Aecemt te D'spositian ol lts Figuce 6 Varindons af accopranee sampliog. 14 MANAGEMENT ASPECTS OF QUALITY IMPROVEMENT 15 1c0 Acceptnce specistoa Sama | dimit / Ê Prozess | Process mem E conte ! e Ê Ê Loner Desian specification o e a experiente E Aesspiance — Stasial Design ot oo” preaiol am Figure 1.7 Phase dicgram of the f Pi ruda vo 1.4, Pato digam ot toc Plgure 8. Applicdon OF quaiy-sgiesán cuatyranginecing metros a tc ori ve eins an he stabilize the process and reduce the variability. However, is not satisfactory just to meet eguixements-—furiher reduction of variability usually leads to better product performance and enhance competitive position, as vas vividly demonstrated in the automebile trams. mission example discussed earlier. Statistically designed experiments can be employed in conjunction with statistical process control to j cess variability in nearly al e e minimize pros hs Pp; ability ) 1.4 MANAGEMENT ASPECTS OF QUALITY IMPROVEMENT Statistical ieciques, including SPC ane designed experiment, along with other problem solving tools are the technical basis for quality control and improvement lower, to be used most effectively, Ciese techniques must ne implemented within ané be part of a man- agement syster that is focused om qualty improvement The management system of am organization must be o-ganized to property direct the overall quality improvement philos- cply and ensure ts deploymentin al aspects of the business. The eifective. anagement cf quality involves suecessful execution of three activities: quality planni ue esses ená quaitty control ard improvement. sum juality planning is a strategio activity, and dt às just as vital izatior soc Bias ss te produ develop lr he fail plan, e me oting plan, and plans for the utilization of human resources. Witaoat a siracegio quality plan, im enorsaons amouax é time, money, and effort will be wasted by he organization desing th fauey designs, mamufeconing defecis, Geld failures, and customer com- pluimi. Quality plescing involves ideníing onsomers, both exteaa and those the operato interaal to the business, and identifying their needs (this is someúmes called lis- tening to the volee of the customer). Then produess or services that meet or exceed eus fomer expectations must be developed. The eigi dimensions of quality discussed in Ses É. ne am ira a of ais. Tue copio res then determine ow these produeis and services sil be cealized, Plaaning dor quality ireprovement on a specific, systemado basis is also a vital part of this process. Quality asgurance is the set o activities that ensures the quality leves of products and services are properly maintained and that supplier and customer quality issues are properly resolved. Documentation of the quelity system is am irporant component, Quality system documentation involves four components: policy, jpocedares, «york jhsricions and specification, and records. Policy generally deats with sihar is to De done 16 CHAPIERI QUALITY IMPROVEMENT IM THE MODERN BUSINESS ENVIRONMENT and way, vibile procedures focus on the methods and personne) shat will implement pol- iey. Work instructions and speeificaãons are usually product, department, tool-. or iachine-oriented, Records are a way of documenting Hhe polícies, procedures, and work instructions that have been followea. Records are also used to track specific units or datohes oÉ prodircr, so that it cen be determined exactly how they were produced. Records are often vital in providing data for dealing with customer complaints, corrective netions, and, if necessary, produer recalis. Development, maintenance, and control of docwmenta- Eor are importent quality Assurance functions. One example of document controLis ensur- ing that specifications and work instructions developed for operating personnel refleo: the latest design and engineering changes. Quality control and iriprovement involve the set of ecuvities used to ensure tharthe producis and services meer requirsinerts and are improved om a continuous basis, Since variability is often a major source of poor Quailty, statistical cectiniques, including SPC and designed experiments, are the major tools of quality control and improvement. Quality improvement is often done on a project-by-projset hasis and involves teams led by per- sonnel with specialized knowledge of scatistical methods and experience in applying them * Projects should be selected so thar they have significant business impact and are linked «with she overall business goals for quality identified during the planning process. The tech- niques in this book are integral to successful quality control and improvement. The next section provides a brief overview of some of the key elements of quality management, We discuss some of the important quatity philosophies; qualioy systems and standards; the link berween quality and productivity and quality and cost; economic and legal implications of quality; and some aspects of implementation. The thves aspects of quality planning, quality assurance, and quality contro) and improvement wii be voven inro the discussion. 1.41 Quality Philosophy and Management Strategies Many people have contributed to the statistical methodology of quality improvement. However, ia terms of implementation and management philosophy, tee individuals emerge as she leaders: W. E. Deming, 3. M. Jurar, and A. V. Feigenbaum. We now reviesy the approaches and plilosophy of those leaders in quality management. W. Edwards Deming W Edwards Deming was educated in engineering ard physics ar the University of vyoming and Yale University. He worked fo: Western Blenzic and was inflnenced greatly by Walter A. Shewharr, the developer cé the control char. After leaving Western Plecrrio, Deming held goveramenk jobs with the U.S. Departuent of Agricultero and the Bureau of the Census. Dwing World War IL, Deming worked for he War Deparement and the Census Bureau. Following the war, he became à consultant to Japanese industries and convinced eejr top smanagement of the power of statistical methods and the importance of quality às a competitive weapon. Tais comunftiteat to and use OÉ statistical methods has been a key element im the expansion of Japan's industry and economy. The Japanese Union of Bcientists and Engineers orsated the Deming Prize for quality improvemene in bis honor. Until his death in 1994, Deming was an active consultant and spenlcer; he was am inspira tonal force for quality improvement in this courtry and around the world. He femly beliored that tas responsibílicy for quality rests with management, that ís, most of the opportunities for Quality improvement require management action, and very few opportu- 14 MANAGEMBNT ASFECES DE QUALITY IMPROVEMENT 17 nities lie. as the workforce or aperator level, Demin; sh crid pis Hs de vor g was à harsh critic of many American de Deming philosophy is an important framework for implementing quality and productivity improvement. This philosophy is summarized in his 14 pojnts for manage- ment, We now give a brief statement and discussion of Deming's 14 points: 2. Cicate a constancy of purpose focused on the improvemen of products and ser- vices. Constantly try to improve product design and performance, Investment in research, development, and innovarioa will have long-term payback to the E k pon à new philosophy tha recognizes we are in a different economi era. ejeet poor workmanship, defective products, or bad service, Xt costs as much to produce a defectivo unit as it does to produce a good ore (and sometimes mare). The cost os desling with scrap, rework, and ather losses created by defec- tives is an enormous drain on Company resources. » Do not vely au máss inspecrion to “conirol” quality. AI iuspecrion cam dos sort out defectives, and at this point it is ico Iate because we have already paid io produce these defscives, Inspeetion typically occurs too late in the process, iris ekpensive, and tis oftes ineffective, Quality results from prevention of defee- tives ihrough process improvement, not inspection. Do not award business ta supplisrs on the basis oÉ price alone, but also consider guelity. Price is à meaningful measure of a supplier's produot only if it is con- sidered in relation to a measure of quality. In other words, the total cust of the item must be considered, nox just the purchase price. When quality is consid- ered, th lowest bidder frequently is not the low-cost supplier. Preference should be given to suppliars viho use modem methods of quality improvement in their business and who can demonstrate process control and capability. Focus on continucus improvement. Constantly try to improve the production E ane sem Involve the workforce in these activities and make use of statistical methods, particularly Ei ii H q a by the statistically based problem-soiving tools dis- Practice modem training methods and invest in on-the-job training for all employees. Everyone shovld be trained in the technical aspects of their jub, and ia modern quality: and produceivity-improvement methods as well. The training should encourage all employees to practice “hese methods every day. Improve leadership, and practice modem supervision methods. Supervision should nor consist merely of passive surveiliance of workers but should be focused on helping the employees improve the system in which they work, The number ore goal of supervision shoclá be to improve the work system and the produer. Drive out fear. Many woriters are afraid to ask questions, report problems, or point out conditions that 2x6 barriers to quality and esfective production. In many organizations the economic loss associared váth fa: às large; only man- agement cam eliminate fear, Break dowa the bartiers between functional areas of the business. Teamwork among diffocent organizational mnits is essential for effective quality and pro- Guctivity improvement to take place. E 18 CHAPTERA QUALITY IMPROVEMENT IN TRE MODERN BUSINES SS E ENVIRCIMENT 40. Elimioato targer, slogans, ani mumerioal goels fo: (se workforce A target stéh às "aero defects” is useless withouta plan for the achievement of this objective. Ta fact, these slogaos and “programs” axe usually conntesproduetive. Work to improve the system and provide information on that. 11. Eliminate numerical quoras and work standards. These standards have bistori Cai seen ser without regard to qualiy. Woxk srandards are often syrantoms cf damagement's inebility 10 understand the work process and 1o provide am ele. cive management system focused on improving his process. 1. Remove the baniers thar discourage employees from doing their jobs. Management must lstea to emplojee saggestions, comments, and complaínis “The person sho js doing the job laws the most about it aná usually bas valor able ideas about hove to malke the process work more effectively, The workforce js al important participam im che business, and not just a opponemt ia colieo- tive bargaining. 13. Instinto 20 ongoing program of education for all employees. Education in sir ple, porverful statistical techniques should be mandatory for al employees. Use de ho basic SPC problem-solving tocis, particulary te coutol char, should Vecome vwidespread in the business. As these chars become widespread and às employees understand thedr uses, they vn be more líely to Toner the ceuses of poor quality and to idemify process impeovemento. Education is a way of malgng everyone parmers in the quality improvement process. 14, Crente à simicinre in top management that vel] vigononsly edvocate the first 33 poínes. As we read Deinhag's 14 points we noúice that there às a smong emphasis on change. Also, fe role of management in guiding his change process is of dominating imporáras However, that should be changed, and how should this change process de started? Fer example, if we want to improve the yield of a sermicendueto- manyfacturing process, what shoulé we do? Iris in this area that statistical methods come into pay most frequenty. To improve the semicondutor process, we «must determine which controliable factors in the arocess influence the number of defective units produced. To answer this question, we. aist Colleer data on the process and see how the system reaois to change in the process variables. Statistical methods, snch as designed expernmenis and control charts, cas com * tribute to these activities. Joseph M. Juran . e bora ja 1904. He is one of the founding fntters of the quality conmol and a men feta, He word for Walter A, Sheet at AT&:T Bel Laborttocios end tas Been at the ieading edge of quality improvement ever since. He was invited to speak to Japagnso industry leaders es hey 3egan their industrial tcansformanon in lhe easly 19505. a as autãos (uith Ecant M. Gryve) of the Quality Conto? Handbook, s sandard nas for quality methods and improvement sínce is intal pablicadom in 1957. “Th Juraa quality management piilosophy fecuses on three components: piaraiho, control, and improvement, These exe known 16 lhe Juan Trilogy, Às we have nojet pr Siousty, planning involves identiying ememel customers and determining tora needs, “Then produ or services Nat cesponá to ihese customer nerd axe destgned andior devel- oped, and the processos for producing these product cr services axe them developed. The 14 MANAGEMENT ASPECTS OF QUALITY IMPRCVEMBN: planning process shoutd also involve planning for qualiry improvement ou a regular (oyp- jcally annwal) basis, Contiol is employed by the operating forces of the business to ensure feat he product or service meets the requirements. SPC is one of the primary tools of con- sro). Tmprovement aims to achisve performance and quality levels that are higher than cur- rent levels. Juran emphasizes thar improvement must be on à project-by-projecr basis “These projects ace ispically idencifed at the plenning stage of the trilogy. Improvement Gam either be contimucus (or incremental) or by brealethrongh. Typicalty, à breaketrongh improvement is the -esult cf studying the process and idendíying a set of changes that result in a large, relatively rapid improvement in performance. Designed experiments are an importam: lool that can be used to achieve brealethrongh. Atmand V. Feigenheurm. Fejgenbaum fist intoduced the concept of company-wide quality control in his historio bogis Total Quality Control (the first edition was published in 1951), This book influenced mach of the essly philosophy of quality management in Japan in the early 19505. Ta fact, many Japanese companies used the name "oral quality control” to cescribe thei: efforis. He proposed a ihrse-step approach to improving quality: quality leaderehip, quality tech- nojogy, aná organizatioia commivment E quality technology, Feigenbaum means st tistical methods and other technical and engineering methods, such as the ones discussed im tis book Feigerbenra is concerned with organizational structure and a systems approach to improving quality, He proposed a 19-step improvement process, of wkich uso of statist- cal methods was step 17. He initially suggested "hat much of the technical capabitity De concentrated in a specialized department. This is ju contras to the more modern view tha knowieégo and use of statistical tools need to be widespread. However, the organizational aspects of Fegenhaum's work are important, as quality improvement does not usunly spring forth as a “grass roots” activity; it requires a lot of management comumitment so make jt work. The brief descriptcas of the plilosophies of Deming, Juran, and Feigenbeum have bighlighted both the common aspects and differences of their viewpoints, Ta this aathor's opinion, there are more similarides than differences among thera, and the similariltes are var is important. AM three of these pioneers sress the importance cf quality as en essem- tiai competitive weapon, the important role that management must play in implementing quality improvement, and the importance of statstica: meihods aud techniques in He “quality transformacion'" of an organization. Total Quality Management “Total quality manapement (or TOM) is à strategy for implementing and managing quality improvement zetivities on an organization-ywide basis. TOM began in he early 19805, with the philosophies of Doming and Jurso às the focal point, It evolved into a broader speo- trem Of concepts and ideas, involving perticipative organizatioas and work culture, cus- tomer facas, suppler quality improvement, integratioa of the quality system with business goals, and many other activities o focus il elements of ibe crganizatioa avoná ie qualisy improvement gos!. Tupically, organizations that have implemented a TQM approach to quality improvement have quality councils or high-level teams that deal with strategio quality initiatives, workforce-level teams that focus on routine production or busiaess ativíties, and eross-fonciional teams that adáress specific quality improvement issues. 2 CHAPTER 1 QUALTY IMPROVEMENT UN THE MODERN BUSINESS ENVIRONMENT TCM hes only had moderate success for à variety ol'reasons, but frequently beceutse there is insufficiemt effort devoted to widespread utilizaton Of the technical tools of variability reduction. Many organizations saw the missioa of TQM as one of training. Consequentiy, many TOM sfforis engaged in widespread rraining of the woridorce in the philosophy Of quality improvement avd à few basic methods, This training was usually placed in the hands of human resolrces deparments, and much of it was ineffective. The trainers oftey had no real idea about what methods shonld be iaught, and success was asu- ally measured by the percentage of the workforce that had been “trained,” not by whether any measurabie impaer on business results had been achieved. Some general reasons for the Jack of conspicvous sucoess of TQM include (1) lack of topdowa, high-level manage- met commitment and involvement; (2) inadequate use of statistical methods and insuf- cient recognition of varinbility reduction as a prime objective; (3) general as ogposed to specific business-results-oriemed objectives; and (4) 100 much emphasis on widespread sraining as opposed to focused technical educarion. Another reason for the erratic sucoess of TQM is that many managers and executivas have regarded ir às just mother “program” to improve quality. During the 19505 and 4960, programs such às zero defects and value engineering abounded, but they dad Ht- He real impact on quality and productivity improvement. During the heyday of TQM in the 19808, another pogular program was the quality is free initiative, in which manage- ment worked on identifying the cost of quality (or the cost of nonquality, as the “quality às free” devotess so cleveriy pur 1). Indeed, idensification of quality costs can be very bse- ful (we discuss quality costs in Section 1-4.3), but the “quality is free” practítloners ofien hnd no idea about what to do to acmally improve meny types of complex industrial processes. Ta fact, the leaders of this initistive had no knowledge abourstatstical meshod- ology and completely failed to understand “ts role in quality improvement. When TQM is wcapped around am ineffective program such as this, disastez às often che result, Quality Systorme ond Standards The Imernational Standards Organization (founded in 1946 in Geneva, Switzerland), Ienowen as ISO, has developed a series of standards for quality systems. The fist standards cuere issued in 1987. The current version of the standard às Imowa as the IS 9000 serios, — Xe is a gencrio standard, broadly soplicabie to ey type of organization, and ir às often used to demonstrate a supplier"s ability to control “ts processes. The three standards of ISO 9000 are: = ISO 9000:2000 Quality Management Systerm-—Tundarnentels and Vocabulary ISO 900::2000 Quality Management System-—Requirements ISO 9004:2000 Quatity Management Systero —Guíelines for Performance Improvement 1SO 9000 is also an American Natiorai Standards Institute and an ASQ standard The 180 9001:2000 standard has eight clauses: (1) Scope, (2) Normative References, (2) Definitions, (4) Quality Management Systems, (5) Management Responsibilhy, () Resoureo Management (7) Product (or Service) Realization, and (8) Measarement, Analysis, and Improvement. Clanses 4 through 8 are the most important, ané Meir key components and requirements ave shows in Table 1-2. To become certified under the ISO stendard, à company Iaust select à registrar and prepare for a certifica- tion audit by this registrar. There às no single independent authority that licenses, reg- ulates, monitors, or qualiÃes registrers. As we will discuss later, this às à serious problem vith-the ISO system. Prepering for the certification audit involves many ectivitios, Tá MANAGEMENT ASPECTS OP QUALITY IMPROVEMENT 21 “Table 1-2 ISO 9003:2000 Requirements so “1 53 54 35 so el 62 &s 64 70 a 72 73 7 75 76 ao 81 22 23 sa 8 Quality Management System Generel Requirements The organiesticn shall establish, document, implemen:, and main sh, docur q ain a queliry management system continually improve its effectiveness 's gecordanço with the requizemeats of the ia ando Documentation Requirements . Quality management system documentaiion with inciud j K é À mem inciude a quality policy an qualicy objectives; a queli “mau, docente prnedre, dormenta to enure ici lescng, peso am cama processes; and records required hy the imternatioael standard. Mansgement System Management Conmitmont à. Communication of meetir.g customer, sramros ú customer, sraratory, and regularory eequisements b. Establisbing à quality policy e e. Estublishing quality objectives & Conducting management reviews e. Ensuring that resources are evailabie Top menngemet shall ensure that customer requirements are deterrmin: à ai embencing customer satisfection. emas gre de 6 anã re met vit te sim ot Maua goment shall estoblish a quality policy. Management shall ensuze that quality objectives shell be estabji ished. Management 4 - núng occurs for the qualley mansgement system. geme tal ensuro dat plo Management shall ensuze (hat sesponsibilítios and suthorites are deined and comununicated. Management shell review the qualioy management system at regular intervals. Resoures Management The arganizsticn shall determine and provide needed resources. Wotkeis will be provided necessary education, training, skills, and experience. The organizetion shall dererrméne, provide, and mainrain «he infrastruct e eve it te produes requirements. º eu pega to peso conto ganized joe organization shall determine and smanage the work environment needed to achicwo confemity to Te at de d hr achie Product or Service Realization “The organiuatico al plan and develop processes needed for produot or service realizatica, The organfeetico ehal determino sequirerments ss specificd y customers “The organizáticn shall plan and control the design and development for its products or services. The organizstica shall cusuro that unchased material or product a Terei 2 product conforms ty specified purchase The owganizetica ehal! plan and carry out productioo and service under controlled conditions. Tre organiastico dba) detrrne le monitoring and measurêmenn ob underistan an dh sonioring easvring devioes nveded (o provide evidence of conformi ã i dad essi pe ernisy Of products Or services to determined Measurement, Anelysis, ond Improvement The organization shall plan aod implement the monitoring, measurement, analysis, and improvement process for continual improvement end conformity to cequirements. “The arganizaticn shell monitor information relating ta customer percepticns. The organization shall ensure that produst that does not conf 4 iene o ts com molied to prevent its unintended uso ar dolivory. nto squdemens is Hifi and com The organization shall detessalne, cole, and arvelyze dita to demonstrate the suiability and effective. ntss OÉ the quality mauagenent System, including ». Customer sagisfaction b. Conformence data e. Trend dara à, Supplier dam The organizatico shall continually tmmprove the effectiveness of the quality management system. “Asdapted from the 180 9001:2000 Standaré, Intersarional Standards Ocganizarion, Geneva, Switzerland, 2003, 22 CHAPIER 4 QUALESY DMPROVEMENT IN THE MODSRI BUSINESS ENVIRONMENT including (usually) ar inital or phase audit thar cheéks he present quality menope- ment system dgainst the standard. This is usually folowed by establishing teams to ensure that all components of the Key clause are developed and impiemented, reining of personnel, develoçing applizasle documentation, and developing and instalÉng al nes components of the quality system that may be required. Then the certification audit akes place, Tf the company às certified, then pedodi surveillanee audits by the regis tras continue, sually on ar amava (or perhaps six-moath) schedule, Maay organizations have required their suppliers so decome certified under ISO 5000, or one of the standards that are more indusery-specific. Examoles of thuso industey- specific quality system standards are AS 9100 for the aerospace industry: ISO/TS Lóado anã Q8 9000 for the automotive industey; and TL. 9000 for the telecommnnications indus- try, Many components of these standards ave very similar to those of ISO 050. Much of the facus of ISO 2000 (ané of the industry-specific standards) is cm form: documentation of the quality system; that is, on quality Assurance activities. Organizations usually must makes extensive eiforts to bring their documesration into line with the reguirements É the standards; this às the Achilles! heel of ISO SOCO and other standards. There is far 100 much effort devoted to documentation, paperwork, and book Keeping and not nearly encugh to actually reducing variability end improving processes and products, Ferheomore, many of the third-pary registress, auditors, and consuitants that wrork in (his area ars not sufficiontly educared or experienced enough ia thé technical roots requised fer quality improvement or how these tools shoutd be deployed. They are all too often unaware of whar constitates modem engineerisg and stenistcal practice, end ustally aro familiar with only the most eJementary techniques. Therefore, they conceatrate targsty on the documentation, record-ecping, and popervork aspects of certification. There is also evidence that ISO certification or vertificadon under one of the other indusery-specifio standards does line to prevent poor qualicy products from being desigreg, minufaciared, and delivered to se enstomer, For example, in 1999-2000, ihere «iete numero!s incigents of rollover socidents involving Ford Explorer vehicles equinped with Bridpestone/Firestone tires and there were nearly 300 deaths in the United States alone atributed to these aceidents, which led to a recall by Bridgestone/Fixostone of, approximately 6.5 milion tires. Apparent, many of the tres involved in these incidente are memtiacred at the Bridgestone/Firestone plant in Decir, IMinois. a an article on e story ja Time magazine (September 18, 2000), there was a photograph (p. 38) of the siga at the entrance of the Decaur plant which stated thet the plant was "QS 9000 “Certified? and “ISO 14001 Certified” (ISO 14001 is an environmental standard). Although the assignable causes underiying these incidents have nor beem fully discovered, there are clear indicators that despite quality systems certificaion, Bridgestone/Firestone experi- enced significant quality problems. Tr has bear estimated thar ISO cenification activities are (approximately) a 40 biliton dollar annual business, worldwide. Much of this money flows to the registrars, auditors, acid consultants, This amour does not include th interual costs incurred by orgenizatioas to achieve registration, such as the thousands of hours of enginecring and management aSfort, travel, interna! maining, end internal auditing, Tt is not Clear whether any significant fracxion of this expenditure has made its way to the bottom line of the registesed organi »ations, Purthermore, there is no assurance thar certification has a0y real impaet on qualiry (as in the Bridgestone/Firestone tire incidents.) Many quality engineering suthorítics fest chat ISO certificetion is Iargely a waste of effort. Often, organizations would be far better cof to “Just say no to ISO" and spend à smail frnction of that 40 bilion dollars on their quality systems and anodher levger fraction om amoaningful variability resuctiom efforis, 14 MANAGEMENT ASPECTS OP QUALITY IMPROVEMENT 23 develop their own intemal (or perhaps indesty- iry-based) quality standards, sig enforce them, and pocket the difference. > gal sds, gone The Malcolm Baldrige National (Quality Award “The Melcoim Baldrige National Quality Award (MBNQA) was created dy the U.S. Congress in 1987. It is gives annually to recognize U.S. corporations for performance excellence, Awards ave given to organizations in five categories: mannfacturing, service, small businoss, health care, an «dueation. Three awards may be given ench year in ench category. Meny organizadions compete for the awards, and many companies use the per- formence excellence criteria for self-assessment. The award is administered by NIST (the National Bureau of Standards and Technology). “The performance excellonce criteria and their inserrelasionships arc shown in Pig. 1-9. “The point vales for these citeia in the MENA are shown in Taile 1-3, The criterio are directed vowards results, where results are à composite of cus-omer saisfnetion and retem- ton, market haso and new market development, product/servine qualisy, productivity and operational effectiveness, human resources development, snpplier performance, and public/corporate citizenship. The criteria are nonprescriptive, That is, the focus is on results, not the use of specific procedures or tools. “The MBNÇA process is shown in Sig, 1-10. Aa appficant sendo the completed applh- cntion to NIST. This application is theu subjectad 10 à firstogund ceviss by a team cf Balárige examiners. The board of Baldrige examiners consists of highly qualifed volun- teers from a variery of fields. Judges evaluate the scoring on the applicatio to determine if the applicant will Continue to consensus, During the consensus phase, a group of examin- &xs who scored the original application determines a consensus socre for each of the items, Once consensus is reached and a consensus report wrilten, judges the make a site-visit determination. À site visit tynically às a one-week visit by a team Of four to six examiners «lo produce a site-visit report, The site-visit reports are used by the judges as the basis of determining the fina! MBNQA winners, gafieattonaimiadt = Relihoneihe arg Enenêe. é “2 “25, E 'Straqagi: umano semi] | seg: demo E ausinei E a Enc pan Figore 19 The sinicie ofihe MBNQA perforreance excetionce crtecia (Soure; Fouadation for the Malcolm Balcrige Natonai Qualy Sard, 2002 Critoia dor Performance Excellence.) 2 CHAPTER 1 QUALITY IMPROVEMENT IN THE MODERN BUSINESS ENVIRONMENT Table 1.3 Posformenes Excellence Categories and Point Valuda 1 Lendership 2.1 Lendership System .....ico oiee 22 Company Responsibility and Citizenshio ... 2 Strategic Plenning 21 Suretegy Development Process 2.2 Company Strategy . 3 Customer nad Market Focus . à1 Customer and Market Knowledge 3.2 Customer Satisínciion and Relationship Enhancement . 4 Information and Analysis 41 Mensurement and Analysis of Performance .... 42 Tnformation Management .. 5 Human Resouree Focus 5.3 Work Systems .. ne no 52 Employes Educaton, Trínivg, and Development 53 Employee WellBeing end Sntisthesion 6 Process Managoment 61 Management of Produer and Servios Processos 62 Management of Business Peocessos . 63 Management of Suppor: Processos 7 Business Results 71 Customer Results coco 7.2 Financial and Maes Results . 73 Hume Resoures Results . 74 Organizational Resuls ......oo. Total Polués 120 1 so E as0 1000 Fexappiication a: oe Eqede DE Blog Feedback Begor to Agplicant get, Fesoback Report to Seport to Apolicant Agaficant Yes, Pengback Rapert Eigure 110 MBNQA process. (Souces Fouedetor for dhe Malcolm Bajdiige Notonal Quality Anac, 2002 Crheria for Pe “oruguto Excellence, 14 MANAGEMENT ASPECTS OF QUALITY REROVEMENT 25 As shown in Pig, 1-9, feedback reports ars provided to the applicant at up to thres stages of the MENA prucess. Many organizadons have found these reperts very helpful and use them as the basis of planning for overal improvement of the organization and for driving improvement in business results, SieSigma Products with maoy compocents typicaliy have many opportunities for failure or defecas to oceur, Motorala developed the sbe-sigma program in the Jate 1980s as a response to the demand for their products. The focus Of sie-sigma is reducing variebility in key product quality characteristios to the level ar which failure or defects are extremely unhikely. Figure 1-11g shows à zormal provability distribution as a mode! for à quality char- acteristic with the specification limits at three standard deviations on either side of the mean. Now it tums out thar in this situation the probability of producing 2 product within these specifications is 0.9973, which corresponds to 2700 parts per million (ppm) defective, This is referred to as three-sigma quality performance, and it actu- ally sounds presty good. However, suppose we have a praduot that consists of an assem- dJy of 100 components or parts and all 100 of these parts must be nondefective for the So da = Spec, Limit Percant insida Squos ppm Enfoctia 41 Sigma sam 3x7300 =2 Sigma D5As assag aS Sigma soa 2700 às Sigma 999937 & =5 Sigma 98 ogo94a 057 à6 Sigma 99.999999a 600 te) Nocmal distritutlon fezes at the larger (M) elo, 150 USL T Mo 40 430 vio 457 v6g snee. Unit Percentinstéo epocs ppm Detective 43 Sigma 3025 oro a2 Sigma óaas éogco *3 Sigma 9232 Ei) dá Sigma 95,2980 “2 *5 Sigma E) 233 *6 Sigma 98saogéol 34 tó) Soma! ltributlan with te enem abifie By 1.50" form ae target Figltre 1.11 The Movarola six-aigma concept 26 CHAPISRI QUALICY IMPROVEMENT IN THE MODERN BUSINESS ENVIRONMENT product to function satisfnctorily. The probability that any specific unit of product is nondefeotive is 0.9973 x 0.9973x..-x 0.997 09973708 == 0.7631 That is, about 23.7% of the products produced under three-sigma qualicy «will be defectve. This is not am neceptablo situation, because many prnducts used by today's soci- ty axe made up Of many components Even a relesively simple service aetiv'ty, such as a visit by a family of four to a fast-food restmmrart, cam involve the assembly Of several dozen components. An automobile has about 200,000 components and am airpiane hes several milhoa! “The Motorola six-sigma consept is to reduce the varfabilicy in the process so that the specification Tmits are six standard deviations from the mena. Then, as shown in Fig, i-l Io, chore will only be about 2 parts per billion defective. Under six-sigma quality, the proba- boy hat any specific unit OF the hypothetical product above às nondefective is 0,9999998, or 0.2 ppm, a much Serter simuation. When the sbe-sigma concept iwas inilially developed, an asstinption was made that - sen the process reached be six-sigma quality level, the process mean was sil subjece o disturbances tha could cause it to shift by as much às 1.5 standard deviations off tas ger, This situation is Shosem in Fig. 1-1 1b. Under this soenario, a six-sígrna process would proânce about 3.4 ppm defective, There às an appasent incoasistency in this. As vie will discuss in Chapter 7 on process capatilisy, we can only rnale prediotioas about process performance when the process is stable; that is, winen the meso (and standard dovinon, too) is canstaut. É the mean is Aiiting arouad, and ends Up as much 85 1,9 standard devistions of target, a prediction of 3,4 ppm defeetive may not be very reliable, because the meam mig shift by more Man the calenve” 1.5 standard deviations. Process performance isn't predictable unless the process beltaviar is stable. However, no process or systera is ever truly stable, and even im the best of cituatioos, disturbances ocour. The concept of a six-sigma process is one way to mudel this behavior. Like al] models, it's probably not exacely right, bue ichas provem to be a nseful way to ihinhe about process performance. Motorola established six-sigma as both am chjestivo for the corporation and as a focal point for process and product qualioy improvement elforis. Tn recert yeseo, six-sigma has spread beyond Motora and hes come so encormpass much more T has become a program Tor imoroving corporate business performance by both improving quality and peylis attention to reducing cost. Companies involved ir. a si-sigma eifor. utilize specially Cined individuals, called Black Pelts (BB6) and Master Blacx Belts (MBBs), to lead feno focused ou projects thas Nave both quality and business (economic) impaets for the organdeaion, The BBo and MBBs havo specialized trôning and education op sratistical, Trethods and the quality and process improvement toois ia chis textbook that equips them o function as team leaders, facilitators, and problem solvecs. The paper by Hloesl (2001) desenlbes tsc components of à typical BB educadon program. BBs and MBBs udlize a specifie five-step problem-soiving aparogob: Define, Mensure, Analize, fmprove, and Control (MATO), The MAI framework utilizes control chart, designed experimenta, process copability analysis, measurement systezns capabílity studies, and mamy other basic statistical tools. Sbe-sigma has beem much more successful than is predecensors, notatly TM The project-by-project approach and the focus on obteinhag; improvemene in battom-line busi- as resulta has been instrimental ia obtaining managenmewt comraitmens (o sit-sigma. 14 MANAGEMENT ASPECTS OP QUALITY IMPROVEMENT 27 Another major component in obtaining success is driving the propez deployment of stz- tistical methods into the right places in the organization. The DMAIC problem-solving tramework is an important part of this, Just-in-Time, Lean Manufacturing, Poka-Yolte, and Others There have been many initintives devotnd to improving the production system. Some cf these include the Just-in-Time aporosck emphasing in-process invenlory reduction, rapid ser-up, and a pull-ype production system; Poka-Yoke or mistake-proofing of processes; the “Tnyota production system and other Tapanese manufactatimg techniques (veith once popular management books by those names); reengincering; thscry O canstraints; agile menufieturing; lean mamufieruriag; and so om. Mosr of these “programs” devote far too lítte artencion 10 variability reduction Tes virmally impossible to reduce the in-process inventory or operate à pull-type, agile, ar lem production system when a large ard unpre- dictsblo fracion of the process output is defective. Such efforis will not achieve their full potential without à major focus on statistical methods for process improvemeat and vari- ability reduction to aceompamy tbem, 1:42 The Link Between Quality and Productivity Producing higl-enaltwy producis in the modera industrial enviromnent is not easy. À siz- nificant aspeer of the problem is the rapid evolation of tecimology. The last 20 years have seen au explosion of technology ia such diverse fields es electronios, metallurgy, Cerari- es, composite materials, diotechmology, and the chemical and pharmaceutical sciences, wikich has resulted in Many new producis and services. For example, in the electronios fietd the development of the integrated circuit has revolutionized the design and manufac- ture of computers and many electronic office products, Basic integrated circuit technology hos been supplanted dy large-scale integration (LSD) ad very Iasge-soale integration (VLSD technology, with corresponding developments in scinicondnotor design and man- ufoemring. When technological asvences cecur rapídiy and when the new technologies are used quickdy to exploit competitive advantages, the problems of designing and mano- tacturing products af superior quality are greatly complioated. Often, too litle attention is paid to achieving sl] dimensions Cf an optimal process: economy, efficiency, procuctiviy, and queliy, Effective qualioy improvement am be instramental ia increasing productivity and reducing zost. To illusirate, consider the men- ufnctur of 4 mechanical component used in a copier machine. The parts are mamsinctared in a machining process at à rate CÊ approximately 100 parts per day. For various reasons, the process is operating at à first-pass yield of about 75%. (That is, ebout 75% of he process ontput conforms to specificadons, and about 25% Of the output is nonconform- ing) About 60% of the Fallout (the 25% nonconforming) can be remorked into an accent- able procuct, and the rest mus: be serapped, The direct manufactacing cost through táis stage of producion per part is approximately 820. Pares that can be reworked incur an ateitona processing charge of 84, Therefore, the manufnetaring cos! per good par pro- juced is 520(100)+ $4(15) 30 Nore that the total yield from this process, after reworlâng, is 90 good pas Costigood par: = s2289 per day.