1. 6 簡介6 SIGMA 起源及主要概念
6 SIGMA 組織及人員認證
基本統計應用
執行主要步驟 D-M-A-I-C
Case Study
QRA : 2001/8/27
2. The Origins of SixSigma
It originated in the 1980's as Motorola responded to the threat of Japanese competition which had far lower defective rates. The approach spread to AlliedSignal and to General Electric, whose Chief Executive Officer, Jack Welch, has been the most passionate advocate of SixSigma. Since introducing it worldwide in 1996, GE has made over $1 billion of cost savings.
3. 1979 年 ,當時 Motorola一位資深業務主管 Arthur Sundry 在高階主管會議上說:"Our quality levels really stink !"
Six Sigma Quality Program 此名稱乃為 Motorola工程師 Bill Smith所建議,為Robert Galvin所採納。何謂 6-Sigma ?Motorola :
5. Where Does “Six Sigma” Come From? Mikel J. Harry one of the Original Architects
Previously Headed Quality Function at ABB and Motorola
Now President/CEO of Six Sigma Academy in Phoenix, Arizona
Has Consulted for Texas Instruments, Allied Signal (and others)
Currently Retained by GE to Teach the Implementation,
Deployment and Application of Six Sigma Concepts & Tools Learning from Those Who Have had Success
With 6Will Accelerate its Implementation at GE
6. What Does “Sigma” Mean? Sigma is a Measure of the Consistency of a ProcessIt (is Also the 18th Letter in the Greek Alphabet!
7. Motorola公司認為數據是滿足顧客的關鍵,他 們常說:
1.如無法用數據表示我們所知的,那麼我們對它
所知不多
(If we cannot express what we know in numbers, we don't know much about it)
2.如果對它所知不多,就無法控制它 (If we don't know much about it, we cannot control it)
3.如果我們不能控制它,那只有靠運氣了
(If we cannot control it, we are at the mercy of chance)
8. Why “Six Sigma”?Proven Successful in “Quality-Demanding” Industries e.g.,
Motorola, Texas Instruments (many process steps in series)
Proven Method to Reduce Costs
Highly Quantitative Method – Science and Logic Instead of Gut Feel
Includes Manufacturing & Service (close to customer) and Provides Bridge to Design for Quality Concepts
Has Support and Commitment of Top ManagementIt Works!!!
10. 6 Overview Sigma3456SpellingMoneyTime1.5 Misspelled Words
per Page in a Book1 Misspelled Word
per 30 Pages in a Book1 Misspelled Word in
a set of Encyclopedias1 Misspelled Word in all
of the Books in a Small
Library$2.7 Million Indebtedness
per $1 Billion in Assets$570 Indebtedness
per $1 Billion in Assets$63,000 Indebtedness
per $1 Billion in Assets$2 Indebtedness
per $1 Billion in Assets3 1/2 Months
per Century2 1/2 Days
per Century30 Minutes
per Century6 Seconds
per Century6 is Several Orders of Magnitude Better Than 3!!!Sigma: A Measure of Quality
11. 6的品質水準0.002 DPPM , Cp=2中心值±1.5的偏移3.4 DPPM, Cpk=1.5或用 DPMO (Defects Per Million Opportunities)表示
12. 4.5
13. 偏移±1.5在動態的真實世界中,每一件事都在不停的變化著
溫度
溼度
工具的磨耗
原料的差異 ......
1.5乃由機率、估計、經驗而來參考資料:Evans, D.H. (1975) : Statistical Tolerancing : The State of the Art, Part III. Shifts and Drifts.
23. 6 品質方案的架構Total Customer Satisfaction縮短作業週期參與式管理Empowerment標竿研究........6 Quality Program, Cp, Cpk, DpuQUALITY METRICSIMPROVEMENT PROCESSIdentify product or services
Identify customer requirements
Identify your needs
Define the process
Mistake-proof the process and eliminate wasted effort
Ensure continuous improvementTOOLS FOR SIX SIGMADesign to standard parts/materials
Design to standard processes
Design for assembly
Part standardization
Supplier SPC
Process standardization
Statistical process control ......and more
25. So...What is Six Sigma? A Measurement System A Problem-Solving Approach A Disciplined Change Process“THE SIX SIGMA BREAKTHROUGH STRATEGY”MeasureAnalyzeImproveControl
26. *DefineMeasureAnalyzeImproveControlDMAICProject Selection
Team Formation
Identify CTQs
Y
Define Specs
Validate Measurement System
Define Defects
Baseline
Set Goals
Identify x’sFind and Confirm
vital few x’s
Pilot Soln
Y = f(x)Ensure Solution
is Sustainable
27. Key Customer
Opportunity to grow or threat of lossIs the customer telling us when the issue is?Six Sigma Project IdentifiedDefine-Scope itMeasure-How are we doingImprove-ImplementControl-Make sure it sticksProblem has quick fix?Fix it!Analyze-Root Cause Preliminary look at how we think we are doingSet up customer meetingDefine/Refine CTQ’s What’s important to them?How are we doing on these CTQ’sHow can we improve?Six Sigma Is A Growth ToolYESYESNONODMAICCustomer Improvement
Continuous Growth
28. DMAICDefine Measure Analyze Improve Control
Define the project goals and customer (internal and external) deliverables
Measure the process to determine current performance
Analyze and determine the root cause(s) of the defects
Improve the process by eliminating defects
Control future process performance
DMADVDefine Measure Analyze Design Verify
Define the project goals and customer (internal and external) deliverables
Measure and determine customer needs and specifications
Analyze the process options to meet the customer needs
Design (detailed) the process to meet the customer needs
Verify the design performance and ability to meet customer needs
30. StructureQuality Council Members: Labs & Functions
“Pipeline” & BB Project Priorities
Training & Certification
Measurements & Rewards
CommunicationsChampions Leadership: Overall Initiative
Project Funding
HR: Training & RewardsBlack Belts Lead 6 Project Teams
“Measure/Analyze”
“Improve/Control”
Out with Businesses
Here at CRDMaster Black Belts Teach 6
Mentor Black Belts
Monitor BB Projects
Work “Pipeline” Projects
A Resource PoolTeam Members Learn/Use 6 Tools
Work on BB Projects
Part of The Job
Out with Businesses 6 Projects with the GE Businesses
31. Champions or Leaders are those senior managers who will ensure that resources are available for training and projects, and who will be involved in project reviews. Champion
32. Role
. Mentor and train Blackbelts and others.
. Technical leader in GEMS& 6 sigma
. Mentor Greenbelts when referred by Blackbelts.
Responsibility
Develop & deploy 6 sigma tools & processes
Own 6 sigma technical development roadmap
Mentor Blakbelts/Greenbelts on Methodology
Ensure rigor of 6 sigma methodology
Assist in selection of Blackbelts
Drive project closure
Focus BBs,GBs on achieving business results
Commitment
100% dedicated
Certification
Mentor 20 successful projects
Master Blackbelt
33. BlackbeltRole
Lead teams implementing 6 sigma Methodology
Mentor Greenbeltsand refer to MBB as required.
Responsibility
Lead multiple projects to successful completion
Drive awareness of methodology and tools to team members and broader organization
Act as both technical and cultural change agent for quality
Participate in 6 sigma roadmap development
Mentor Greenbelts and attend GB reviews
Update GB project status color codes
Commitment
100% dedicated
Certification
2 successfully closed BBs projects
34. Greenbelt Role
Leads or participate in teams implementing the 6 sigma methodology on projects
Responsibility
Learn and apply methodology to projects in current job scope
Put 6 sigma on Goals & Objectives.
Proactively involve BBs, MBBs as required to advance project.
Work together with other GBs to achieve function’s 6 sigma goal.
Present project findings in the appropriate review sessions.
Commitment
Significant. Varies depending on quality of project definition and scope
Certification
2 successfully closed GBs projects
43. Z - Scale of MeasureZ =A Unit of Measure Equivalent
to the Number of Standard
Deviations that a Value is Away
from the Target Value-3.0-0.53.0Z - Values USLLSL2.50.53.0= Process MeanZ TTarget 0A 3 Process
44. Problem Solving ApproachCenter
ProcessReduce
SpreadXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXOff-TargetUnpredictableOn-Target 6Helps us Identify and Reduce VARIATION due to:
- Insufficient Process Capability
- Unstable Parts & Materials
- Inadequate Design Margin
45. DefineWhat do we want to know ?
-- What is our problem ?
-- What is our defect ?Practical Problem
Project CTQ’s
Solid Y, Spec Definition, and Defect Definition
Building a Justifiable Business Case
Approved Charter (R0)What did we learn ………Do we have a viable project ?
46. 特性要因圖
47. (本页无文本内容)
48. The % Variation Due to the Measurement Method
GAGE R & RGAGE:
The instrument used for making
Measurements that we want to validate.REPEATABILITY:
Does the same operator get the same results
When measuring the same part several times? REPRODUCIBILITY:
Do different operators get the same results
when measuring the same part several times?CAUTION: A
Calibration Sticker does not
imply that Gage R&R
is acceptable!
49. MeasureWhat do we want to know ?
Statistical Problem
Project Y, Spec Definition, and Defect Definition
Gage R&R on Y Data
Process Map
L1 to Show Z Values with Link to Financials
Rationalization for Focused Y
Descriptive Statistics for Focused YWhat did we learn ………Mean problem ? Variance ? Both ?
54. Changing Focus From Output to Process Y
Dependent
Output Effect
Symptom
Monitor X1. . . XN
Independent
Input-Process Cause
Problem
Control Identifying and Fixing Root Causes
Will Help us Obtain the Desired Outputf (X)Y =
55. Some Basic 6-Related ToolsScatter Diagram Over Slept Car Would
Not StartWeather Family
ProblemsOtherPareto Diagram Frequency
of
OccurrenceReasons for Being Late for WorkArrival
Time
at WorkTime Alarm Went Off
56. AnalyzeWhat do we want to know ?
Statistical Problem Understood
Link Process Map to Fishbone w/C, N, X Labeling
Ensure Data is Collected for All Xs
Drill Down to Vital Few Xs via ANOVA, GLM,
T-test, F-test
List of Vital Few X’s
Quantified Financial OpportunitiesWhat did we learn ………
What were the causal Xs ? How did we identify them ?
57. PM/CE/CNX/SOPProcess MappingWhy?* Process visual foundation for current situation and analysis
Aids in identifying bottlenecks, redundancies and waste
Look for Non Value Add
Look for Variables (CNX)How? Determine beginning and end of process to be mapped
Involve people knowledgeable about the process
Brainstorm steps & group in major process areas
Layout activities in sequence
Validate by physically walking through process
58. PM/CE/CNX/SOPCE: Cause & Effect
Sources of VariationOutput(s)
& Specs People Material MachineX
CC
NCNXX Measurement Method EnvironmentEvery variable on the diagram should be labeled as either:
C = Constant
N = Noise
X = Experimental variable or Factor
59. PM/CE/CNX/SOPSOP: Standard Operating ProceduresStandard Operating Procedure are rules that we define to ensure that we have
Consistent processes in everything we do.
Based on good judgment
Common sense
Engineering knowledge
Remember ISO 9000!
Make sure we have defined processes and that the rules are being obeyed by all.
60. 數據分析1.Sample size risk Is the risk of say there is a difference when
there really isn’t one(生產者冒險率)
Is the risk of not finding a difference when
there really is one (消費者冒險率)
/ is he magnitude or size of the difference been
tested.
This is sometimes called the test sensitivity.
61. A Basic Sample
Size TableApplies to
Continuous
Data Onlyδ/σ
62. For example :
The 1st adhesive has an average of 13.2 and a
standard deviation of 3.27
The practical significance is such that any alternative
Adhesive must have an average strength of 20 or more
to make the change worth while.
Q: For α=5% & β=10% what should be the sample size
in each level of the experiment?Ans: How big of a change is important?
δ/σ=(20-13.2)/3.27= 2.07
α=5% β=10%
So follow the table we find the sample size is 5
63. 2.T-test : Comparing Means
3.F-test : Comparing Variances
(Between 2 Groups)
4.ANOVA :(Analysis Of Variance)
(Between multi-groups)
64. ImproveStatistical Control
Poka-Yoke Plan (if Applicable)
Confirmed Vital X’s ……Are They Statistically Significant ?
Sample Size Calculation For Confirmation Run
DOE Plan (if Applicable)
Regression Equation Linked to statistical Problem
SOP Changes and/or Optimal Solution Identified
Plan to Implement SOP Changes and/or Optimal SolutionWhat contribution did each vital X have to the Y ?
How can we control the Xs ?What do we want to know ?
65. How do we determine a modelCorrelation & RegressionCorrelation tells if you have a relationship between two variables(Y and X, or two Xs)
Regression is used to identify the nature of the relationship, and be able to predict Y while better understanding Y and possibly improving the controllability of Y
66. CorrelationCorrelations (Pearson)
Correlation of pull and temp = -0.978
67. (本页无文本内容)
68. Regression Analysis (Minitab software)
The regression equation is
pull = 353 - 2.30 temp
Predictor Coef StDev T P
Constant 353.11 13.56 26.05 0.000
temp -2.29520 0.09001 -25.50 0.000
S = 0.3460 R-Sq = 95.6% R-Sq(adj) = 95.4%
Analysis of Variance
Source DF SS MS F P
Regression 1 77.832 77.832 650.28 0.000
Error 30 3.591 0.120
Total 31 81.423
Unusual Observations
Obs temp pull Fit StDev Fit Residual St Resid
12 150 9.3122 9.9752 0.1181 -0.6630 -2.04R
31 151 4.9778 5.8438 0.0864 -0.8660 -2.59R
69. DOE (Design of experiment)1.篩選主要因子
2.尋找最佳因子組合
3.結果再現性驗證找尋最佳之生產條件(製程參數 )
70. Design of Experiments6 Overview SCREENINGOPTIMIZATIONCHARACTERIZATION For Experiments
Involving a Large
Number of Factors
Useful in Isolating
the “Vital Few “ from
the “Trivial Many” For Experiments
Involving a Relatively
Small Number of Factors
Useful When Studying
Relatively Uncomplicated
Effects & Interactions For Experiments
Involving Only 2
or 3 Factors
Useful When Studying
Highly Complicated
Effects & RelationshipsDOE is More Effective Than Testing One Factor at a Time
71. ControlWhat do we want to know ?
Statistical Control
Poka-Yoke Plan (if Applicable)
Measurement of Final Capability Using Confirmation Run
Comparison of Before and After Distributions
How Do You Control or Poka-Yoke the Vital Xs ?
Is the Learning Transferrable Across the Business
……What’s the Plan ?
Who Owns the Project Documentation File ?
Final Financial Results Signed Off
What Spin Off Projects Are There ?How will this change the way we work ?
72. Mistake proofing
FMEA
Process Monitor(SPC, Control chart…)
Risk vs. Benefit
Control tools
73. Short Term CapabilityShort Term Capability Ratio(Cp)Cp =LSL-6USLExampleUSLLSL 3.0==-3.063.0-( - 3.0Cp =Cp =1LSLUSL2.5 0.53.0Process MeanTTargetA 3 ProcessThe Potential Performance of a Process, if it Were on Target
74. Long Term Capability (Cpk)CpCpk=Long Term Capability RatioExampleCp =1 (previous chart)Target = -0.5 =0Cpk1 - (-0.5-03 =Cpk =0.83-Off-Target Penalty Target - 3The Potential Performance of a Process, Corrected for an Off-Target MeanLSLUSL2.5 0.53.0Process MeanTTargetA 3 Process
75. LCLUCLRange ChartROut of Control ConditionLCLXUCLX Bar ChartSome Basic 6-Related ToolsLCL = Lower Control LimitUCL = Upper Control LimitX = MeanR= Average RangeMonitors Changes in Average or Variation Over Time
76.
Risk vs. Benefit
BenefitLowHighLowMedHighRisk122Perform
Confirmation
Runs33Just do it 2. Probably stop
3. “Hire a consultant unless you have high
confidence and experience”
77. 2
3
4
5
6308,537
66,807
6,210
233
3.4PPM SIGMA
LEVEL DEFECTS per
MILLION
OPPORTUNITYIRS Tax AdviceBest CompaniesAirline SafetyAverage CompanyGEAirline BaggageDoctor’s PrescriptionRestaurant BillsAverage Company in 3 to 4Range Some Sigma “Benchmarks”
78. 突 破 策 略ABCDEFG1選擇輸出特性2定義輸入積效指標3定義有效衡量系統4建立產品能力5定義達成績效目標6定義變異來源7篩選潛在因素8發掘變因關係9建立管制公差10建立有效衡量系統11決定流程能力12導入流程管制改 善 專 案突破要項流程認知定 義衡 量分析改善控制產品能
力分析流程能
力分析 專注於Y專注於X目標 = Y = f (X1,X2,….,Xn)
80. Problem Solving ApproachKey Components of “BREAKTHROUGH STRATEGY”MeasureAnalyzeImproveControl Identify CTQ &
CTP (Critical to
Process) Variables
Do Process
Mapping
Develop and
Validate Measurement
Systems Benchmark and
Baseline Processes
Calculate Yield
and Sigma
Target Opportunities
and Establish
Improvement Goals
Use of Pareto Chart
& Fishbone Diagrams Use Design of
Experiments
Isolate the
“Vital Few” from the
“Trivial Many”
Sources of Variation
Test for Improvement
in Centering
Use of Brainstorming
and Action Workouts
Set up Control
Mechanisms
Monitor Process
Variation
Maintain “In Control”
Processes
Use of Control
Charts and
Procedures A Mix of Concepts and Tools Will Also Integrate with NPI Process
81. Disciplined Change ProcessA New Set of QUALITY MEASURES Customer Satisfaction
Cost of Poor Quality
Supplier Quality
Internal Performance
Design for Manufacturability Will Apply to Manufacturing & Non-Manufacturing
Processes and be Tracked & Reported by Each Business
82. TargetUSLLSLTargetUSLLSLTargetUSLLSLCenter
ProcessReduce
SpreadOff-TargetUnpredictableOn-TargetDefectsProblem Solving Approach“Lower Specification Limit” “Upper Specification Limit”Less Variation Means Fewer Defects & Higher Process Yields
83. Baselining & Benchmarking an Existing Processp (x)DefectsBenchmarkBaseline Entitlement Benchmark.....A World-Class Performance Entitlement.....Achievable Performance Given
the Investments Already Made Baseline.....The Current Level of PerformanceBaselining = Current Process / Benchmarking = Ultimate Goal
84. Why 6-Sigma so successful ?
*It’s seeking “breakthrough” rather than “improvement”.
* Top management involvement (Champion)
* Focus on cross-functional critical business processes
* Full-time manpower dedication
* Quantitative data oriented -- $ (data stratification and
confirmation)
*A lot of training (especially statistical training)
* P-D-C-A
86. 6 Sigma is Process Knowledge6
sigmaScience
= Process knowledge
= 6 sigma quality
= Success
Art & Magic
=Confusion
=Fire Fighting
=Waste
We’ve All Seen the Results:
Teams &/or Individuals
*Flounder
*Panic
*Fail
*Cover-upMore process knowledge will improve our competitiveness
87. A Practical Example
(The “Cookbook”)6 Overview
88. 6.....and Baking BreadYEAST FLOURUsing a 12 Step ProcessThe “BETTER BREAD” Company
89. Step 1.....Selecting “Critical to Quality” (CTQs or Y)What is Important to the Customer?
Rise
Texture
Smell
Freshness
TasteY = Taste!!Measure
90. Step 2.....Defining Performance Standards for CTQs or YHow Could We Measure Taste (Y)?
Panel of Tasters
Rating System
of 1 to 10
Target: Average
Rating at 8
Desired: No
Individual Ratings
(“defects”) Below 7
Y = 1 2 3 4 5 6 7 8 9 10 TargetDefectsWorstBestBut.....Is this the Right System?Measure
91. Step 3.....Validating the Measurement System for YHow Could We Approach This?
Blindfolded Panel Rates
Several Loaf Samples
Put “Repeat” Pieces
from Same Loaf in
Different Samples
Consistent Ratings* on
Pieces from Same
Loaf = “Repeatability”
Consistent Ratings* on
Samples Across the
Panel = “Reproducibility”
“Repeatability” &“Reproducibility” Suggest Valid Measurement Approach Panel
Member Loaf 1 Loaf 2 Loaf 3 A 5 8 9
B 4 9 1
C 4 9 2
D 8 9 8
E 4 8 2
F 5 9 1
G 8 9 2* Within One Taste UnitMeasure
92. Step 4.....Establish Product Capability for Y (Taste)This is a 3 Process!7 Defects (ratings below 7)24 Ratings (from our panel)=.292292,000 Defects per
1,ooo,ooo LoavesOR7
6
5
4
3
2
11 2 3 4 5 6 7 8 9 10 # of
RatingsRating64321143Defects <7Target = 8AnalyzeHow Do We Approach This?
Bake Several Loaves
Under “Normal”
Conditions
Have Taster Panel
Again Do the Rating
Average Rating is 7.4
But Variation is
too Great for a 6 Process3 x 10 + 4 x 9 + 6 x 8 + 4 x 7 + 3 x 6 + 2 x 5 + 1 x 4 + 1 x 31 + 1 + 2 + 3 + 4 + 6 + 4 + 3
94. Step 6.....Identify Sources of Variation in Y (Taste)How do we Determine the Potential Sources of Variation (Xs)?
Have the Chefs Brainstorm
Some Likely Ones Might be:
- Amount of Salt Used
- Brand of Flour
- Baking Time
- Baking Temperature
- Brand of Yeast
YEAST FLOURMultiple Sources: Chefs, Suppliers, ControlsAnalyze
95. Step 7.....Screen Potential Causes of Variation (Xs)How do we Screen for Causes of Variation (Xs)?
Design an Experiment
Use Different Sources
of Potential Variation
Have Panel Rate
the Bread Used in
the Experiment
Results Lead to the
“Vital Few” CausesYEAST FLOURSourceConclusionNegligibleMajor CauseNegligibleMajor CauseNegligibleFocus on The “Vital Few”Improve
96. Step 8.....Discover Variable Relationships Between “Vital Few” (Xs) and YHow do we Find the Relationship Between the “Vital Few” (Xs) and Taste (Y)?
Conduct a More Detailed Experiment
Focus: Oven Temperature from 325
to 375 and 3 Brands of Flour
RUN# TEMP BRAND
1 325 A
2 325 B
3 325 C
4 350 A
5 350 B
6 350 C
7 375 A
8 375 B
9 375 C FLOUR FLOUR FLOURBrand ABrand BBrand CImproveResults: 350 & Brand A is Best Combination of Temperature & FlourNote: Time is a Factor
Only if Temperature
Changes Significantly
97. Step 9.....Establish Tolerances on “Vital Few” (Xs)How do we Ensure Oven Temperature is Controlled?
Data Suggests 350 ( 5 )
is best Temperature to
Reduce Taste Variation
Brand A Flour to be
Used Except in Case
of Emergency
“BETTER BREAD”
to Search for Better
Alternative Supplier
of Flour Just in Case FLOURBrand ABut.....Is Our Measurement System Correct?Improve
98. Step 10.....Validate the Measurement System for XsHow Could We Approach This?
Need to Verify the
Accuracy of Our
Temperature Gauges
Need for “Benchmark”
Instrumentation for
Comparison
Rent Some Other
“High End” Gauges
Compare the ResultsVerify that our Instruments are AccurateControl
99. Step 11.....Determine Ability to Control Vital Few XsHow Could We Approach This?
Check A Number
of Ovens
Monitor Temperatures
Over Time
Focus on the
Process Capability
Look for Degree of
VariationVariation OK But...Average is High (and the algorithm should be checked)30345 # of
OvensTemperature34635734734834935035135235335435535625201510 5Control
100. Step 12.....Implement Process Control System on XsWhat do we do Going Forward?
Check Ovens Daily
for Temperature Levels
Audit Usage Frequency
of Alternative Flour
Supplier (e.g., Brand C)
Periodically Reassemble
the Panel to Test Taste
Chart the ResultsAnd.....Plot the Data Over Time FLOUR“Brand C”354
353
352
351
350
349
348 1 3 5 7 9 11 13 15 17 19 21 23 25Control
101. How Do We Arrive at Sigma?Measuring & Eliminating Defects is the “Core” of Six SigmaMeasurement SystemIdentify the CTQsLook for Defects
in Products or
Services “Critical to Quality”
Characteristics or
the Customer
Requirements for a
Product or Service Count Defects
or failures to
meet CTQ
requirements in
all process steps Define Defect
Opportunities Any step in the
process where a
Defect could occur
in a CTQ Arrive at DPMO Use the SIGMA
TABLEConvert DPMO to
Sigma Defects Per Million
Opportunities2
3
4
5
6308,537
66,807
6,210
233
3.4PPM Defects per
Million of
Opportunity Sigma
Level