Industrial Statistics: Practical Methods and Guidance for Improved Performance

280 Pages·2010·3.64 MB·English

by  Anand M. Joglekar(auth.)

INDUSTRIAL STATISTICS
INDUSTRIAL STATISTICS
Practical Methods and Guidance for
Improved Performance
ANAND M. JOGLEKAR
Joglekar Associates
Plymouth, Minnesota
Copyright(cid:1)2010byJohnWiley&Sons,Inc.Allrightsreserved.
PublishedbyJohnWiley&Sons,Inc.,Hoboken,NewJersey.
PublishedsimultaneouslyinCanada.
Nopartofthispublicationmaybereproduced,storedinaretrievalsystem,ortransmittedinany
formorbyanymeans,electronic,mechanical,photocopying,recording,scanning,orotherwise,
exceptaspermittedunderSection107or108ofthe1976UnitedStatesCopyrightAct,withouteither
thepriorwrittenpermissionofthePublisher,orauthorizationthroughpaymentoftheappropriate
per-copyfeetotheCopyrightClearanceCenter,Inc.,222RosewoodDrive,Danvers,MA01923,
978-750-8400,fax[phone],oronthewebatwww.copyright.com.RequeststothePublisher
forpermissionshouldbeaddressedtothePermissionsDepartment,JohnWiley&Sons,Inc.,
111RiverStreet,Hoboken,NJ07030,201-748-6011,fax[phone],oronlineat
http://www.wiley.com/go/permissions.
LimitofLiability/DisclaimerofWarranty:Whilethepublisherandauthorhaveusedtheirbesteffortsin
preparingthisbook,theymakenorepresentationsorwarrantieswithrespecttotheaccuracyor
completenessofthecontentsofthisbookandspecificallydisclaimanyimpliedwarrantiesof
merchantabilityorfitnessforaparticularpurpose.Nowarrantymaybecreatedorextendedbysales
representativesorwrittensalesmaterials.Theadviceandstrategiescontainedhereinmaynotbe
suitableforyoursituation.Youshouldconsultwithaprofessionalwhereappropriate.Neitherthe
publishernorauthorshallbeliableforanylossofprofitoranyothercommercialdamages,including
butnotlimitedtospecial,incidental,consequential,orotherdamages.
Forgeneralinformationonourotherproductsandservices,ortechnicalsupport,pleasecontactour
CustomerCareDepartmentwithintheUnitedStatesat877-762-2974,outsidetheUnitedStates
at317-572-3993orfax317-572-4002.
Wileyalsopublishesitsbooksinavarietyofelectronicformats.Somecontentthatappearsinprint
maynotbeavailableinelectronicbooks.FormoreinformationaboutWileyproducts,visitour
websiteatwww.wiley.com
LibraryofCongressCataloging-in-PublicationData:
Joglekar,AnandM.
Industrialstatistics:practicalmethodsandguidanceforimproved
performance/AnandM.Joglekar.
p.cm.
Includesbibliographyreferencesandindex.
ISBN978-0-470-49716-6(cloth)
1. Processcontrol–Satisticalmethods.2. Qualitycontrol–Statistical
methods.3.Experimentaldesign.I. Title.
TS156.8.J622010
658.5072’7–dc22
2009034001
PrintedintheUnitedStatesofAmerica
10 9 8 7 6 5 4 3 2 1
To thememoryof myparents
and to Chhayaand Arvind
Thefollowingage-oldadvicedealswithrobustdesignandcontinuousimprovement
atthepersonallevel.
Youhavecontroloveryouractions,butnotontheirfruits.
Youshouldneverengageinactionforthesakeofreward,
norshouldyoulongforinaction.
Performactionsinthisworldabandoningattachments
andalikeinsuccessorfailure,
foryogaisperfectevennessofmind.
–BhagavadGita2.47–48
MahatmaGandhiencapsulatesthecentralmessageofGitainonephrase:nishkama
karma,selflessaction,workfreefromselfishdesires.Desireisthefueloflife;without
desirenothingcanbeachieved.Kama,inthiscontext,isselfishdesire,thecompulsive
cravingforpersonalsatisfactionatanycost.Nishkamaisselflessdesire.Karmameans
action.Gitacounsels—workhardintheworldwithoutanyselfishattachmentandwith
evennessofmind.
Mahatma Gandhi explains—By detachment I mean that you must not worry
whether the desired result follows from your action or not, so long as your motive
ispure,yourmeanscorrect.Itmeansthatthingswillcomerightintheendifyoutake
careofthemeans.Butrenunciationoffruitinnowaymeansindifferencetoresults.In
regardtoeveryactiononemustknowtheresultthatisexpectedtofollow,themeans
theretoandthecapacityforit.Hewho,beingsoequipped,iswithoutselfishdesirefor
theresultandisyetwhollyengrossedintheduefulfillmentofthetaskbeforehim,is
saidtohaverenouncedthefruitsofhisaction.Onlyapersonwhoisutterlydetached
andutterlydedicatedisfreetoenjoylife.Renounceandenjoy!
–AdaptedfromBhagavadGitabyEknathEaswaran
CONTENTS
PREFACE xi
1. BASICSTATISTICS: HOW TO REDUCE FINANCIALRISK? 1
1.1. Capital MarketReturns / 2
1.2. Sample Statistics / 5
1.3. PopulationParameters / 9
1.4. ConfidenceIntervals and Sample Sizes / 13
1.5. Correlation / 16
1.6. Portfolio Optimization / 18
1.7. Questions toAsk / 24
2. WHY NOT TODO THE USUAL t-TEST AND WHAT TO
REPLACE ITWITH? 27
2.1. What is at-Test and what is Wrong with It? / 29
2.2. ConfidenceInterval is Better Than at-Test / 32
2.3. HowMuch Data toCollect? / 35
2.4. Reducing Sample Size / 39
2.5. Paired Comparison / 41
2.6. Comparing TwoStandard Deviations / 44
vii
viii CONTENTS
2.7. Recommended Design and Analysis Procedure / 46
2.8. Questions toAsk / 46
3. DESIGN OFEXPERIMENTS: ISIT NOT GOING TO COST
TOO MUCHANDTAKE TOO LONG? 48
3.1. Why Design Experiments? / 49
3.2. Factorial Designs / 53
3.3. SuccessFactors / 59
3.4. Fractional Factorial Designs / 63
3.5. Plackett–Burman Designs / 66
3.6. Applications / 67
3.7. Optimization Designs / 71
3.8. Questions toAsk / 75
4. WHAT IS THE KEY TODESIGNING ROBUST PRODUCTS
AND PROCESSES? 77
4.1. The Keyto Robustness / 78
4.2. Robust Design Method / 83
4.3. Signal-to-Noise Ratios / 87
4.4. Achieving Additivity / 89
4.5. AlternateAnalysis Procedure / 92
4.6. Implications for R&D / 98
4.7. Questions toAsk / 100
5. SETTING SPECIFICATIONS: ARBITRARY ORIS THERE
A METHOD TO IT? 101
5.1. Understanding Specifications / 103
5.2. EmpiricalApproach / 106
5.3. Functional Approach / 107
5.4. MinimumLife Cycle Cost Approach / 114
5.5. Questions toAsk / 119
6. HOW TO DESIGN PRACTICAL ACCEPTANCESAMPLING
PLANSAND PROCESS VALIDATION STUDIES? 121
6.1. Single-Sample Attribute Plans / 123
6.2. Selecting AQL and RQL / 129
CONTENTS ix
6.3. Other Acceptance Sampling Plans / 133
6.4. DesigningValidationStudies / 135
6.5. Questions toAsk / 136
7. MANAGING AND IMPROVING PROCESSES: HOW TO
USE AN AT-A-GLANCE-DISPLAY? 138
7.1. Statistical Logic ofControlLimits / 139
7.2. Selecting Subgroup Size / 145
7.3. Selecting SamplingInterval / 147
7.4. Out-of-Control Rules / 150
7.5. Process Capability andPerformance Indices / 151
7.6. At-A-Glance-Display / 155
7.7. Questions toAsk / 157
8. HOW TOFINDCAUSESOF VARIATION BY JUST
LOOKING SYSTEMATICALLY? 159
8.1. Manufacturing Application / 161
8.2. Variance Components Analysis / 164
8.3. Planning for Quality Improvement / 166
8.4. Structured Studies / 168
8.5. Questions toAsk / 172
9. IS MY MEASUREMENT SYSTEM ACCEPTABLE AND
HOW TODESIGN,VALIDATE, AND IMPROVE IT? 174
9.1. Acceptance Criteria / 175
9.2. DesigningCost-EffectiveSampling Schemes / 178
9.3. DesigningaRobust Measurement System / 180
9.4. Measurement System Validation / 183
9.5. Repeatability and Reproducibility (R&R) Study / 185
9.6. Questions toAsk / 188
10. HOW TO USE THEORY EFFECTIVELY? 190
10.1. Empirical Models / 192
10.2. Mechanistic Models / 197
10.3. Mechanistic Model for Coat WeightCV / 201
10.4. Questions toAsk / 205
x CONTENTS
11. QUESTIONS AND ANSWERS 207
11.1. Questions / 208
11.2. Answers / 232
APPENDIX: TABLES 251
REFERENCES 259
INDEX 261
PREFACE
This book is based upon over 25 years of teaching and consulting experience
implementing statistical methods in a large number of companies in industries as
diverse as automotive, biotechnology, computer, chemical, defense, food, medical
device, packaging, pharmaceutical, and semiconductor among many others. The
consultingassignmentshaveresultedinmanysuccessstories—largecostreductions,
rapidproductdevelopment,regaininglostmarkets,dramaticreductionsinvariability,
andtroubleshootingmanufacturing.Overtenthousandparticipantshaveattendedmy
seminars onstatistical methods.All these interactions—thetechnical problems the
participants brought forward, the prior statistical knowledge they had, and the
questionstheyasked—haveshapedthewritingofthisbook.
Muchofthetechnicalworkinindustryreliesuponthecouplingofknownscientific
andengineeringknowledgewithnewknowledgegainedthroughactiveexperiments
and passive observations. Accelerating this data-based learning process to develop
high-quality, low-cost products and bringing products to market rapidly are key
objectivesinindustry.Thefactthatstatisticalmethodsareanecessaryingredientin
accomplishingtheseobjectivesisastruetoday,ifnotmoreso,asitwas25yearsago.
Theuseofstatisticalmethodsbyalltechnicalindividualsinindustry,whichnumberin
themillions,continuestobeanimportantneed.
Four major changes have occurred during the past several years that have
influencedthewritingofthisbook:
1. With the advent of personal computers and statistical software, the need to
understand statistical computations, in the detail necessary for hand calcula-
tions,hasreduceddramatically.Today,thejobofnumbercrunchingcanandis
delegatedtoasoftwarepackage.Thestatisticalcomputationsdescribedingreat
xi