yan MINITAB

8/6/2019 yan MINITAB

1/50

MINITAB Applications in Six

Sigma

AgendaAgenda

13:0013:00--13:30 Registration13:30 Registration

13:3013:30--14:20 MSA14:20 MSA--MINITABMINITAB

14:2014:20--14:30 Break14:30 Break

14:3014:30--15:20 SPC15:20 SPC--MINITABMINITAB

15:2015:20

--15:30 Break15:30 Break

15:3015:30--16:20 Case study (Healthcare & Banking)16:20 Case study (Healthcare & Banking)

16:2016:20--16:30 Q&A16:30 Q&A

July 14, 2006 in TaipeiJuly 14, 2006 in Taipei


2/50

MSA-MINITAB

Application inApplication in

(Measurement Systems Analysis, MSA)(Measurement Systems Analysis, MSA)


3/50

3

Measuring System.

What is a Measuring System?What is a Measuring System?

The measuring device (Repeatability)The measuring device (Repeatability)

The person who is taking the measurementThe person who is taking the measurement

(Reproducibility)(Reproducibility)


4/50

4

Measurement Process

The ideal measurement system produceThe ideal measurement system produce truetruemeasurements every timemeasurements every time

Quality of the measurement system is characterized byQuality of the measurement system is characterized by

statistical propertiesstatistical properties

PropertiesProperties

Must be inMust be in Statistical ControlStatistical Control Variability must be smallVariability must be small compared to productcompared to product

specificationsspecifications

Variability must be smallVariability must be small comparedcomparedprocess variationprocess variation


5/50

5

Basic Model

2 2 2Total Product Measurement System = +

The Total Variation is equal to the real productvariation plus the variation due to the

measurement system

Mi i E i


6/50

6

Minita Exercise

Capability and Measurement

Error Assume we have a process that has anAssume we have a process that has an

actualactual sigma of 5 unitssigma of 5 units and aand a mean of 70mean of 70

unitsunits

Also assume we have a measurementAlso assume we have a measurement

system that has the same measurement errorsystem that has the same measurement error

as the processas the process -- sigma of 5 unitssigma of 5 units

LetLets use Minitab to simulate the effect ofs use Minitab to simulate the effect of

measurement error on process capabilitymeasurement error on process capability


7/50

7

Exercise - Continued

Follow these methodsFollow these methodsto create data to fit ourto create data to fit our

process model:process model:Creates a randomnormal distribution


8/50

8

Exercise - Continued

What value should we usefor the mean of the

measurement error?

Calculate observed

process behavior.


9/50

9

11010090807060504030

15

10

5

0

Observed

Frequency

LSL USL

ActualActualprocess variation -NoNomeasurement error

ObservedObserved processvariation -

WithWithmeasurement error

11010090807060504030

15

10

5

0

Process

Frequency

LSL USL

Results of Simulations


10/50

10

Effects of Measurement Error

AveragesAverages

VariabilityVariability

total product measurement 2 2 2

= +

total product measurement= +

Measurement SystemVariability - Determinedthrough R&R Study

AccuracyAccuracy

PrecisionPrecision


11/50

11

Sources of VariationProduct Variability

(Actual variability)

Product Variability

(Actual variability)

Measurement

Variability

Measurement

Variability

Total Variability

(Observedvariability)

Total Variability

(Observedvariability)


12/50

12

Two General Kinds of Data

ATTRIBUTE - Discrete, Counted Data

Ex: 1, 2, 3, 4 etcGood/Bad, Go/NoGo, Pass/Fail

Machine 1 , 2 , 3 ...

VARIABLES - Continuous, Measured Data

Ex: Weight = 10.2 Lbs

Thickness = 11.211 inches


13/50

13

10%10%

30%30% 30%30%

P/T

Ratio

P/T

Ratio %R&R%R&R

Red

Ye

llow

Green

15%15%

Main Question - Is my measurement system okay to use on my project?

GR&R Metrics


14/50

14

Part Operator Response

1 1 0.65

1 1 0.60

2 1 1.002 1 1.00

3 1 0.85

3 1 0.80 ...

MSA Exercise


15/50

15

Minitab - Gage R&R Studies


16/50

16

Minitab - Gage R&R Studies

GAGE Options

If There Is A

Tolerance For

This Part -

Put It Here

General Info

On Your Study


17/50

17

Misc:

Tolerance:

Reported by :

Date of study:Gage name:

0

1.1

1.0

0.9

0.8

0.70.6

0.5

0.4

0.3

321

Xbar Chart by Operator

Sample

Mean

Mean=0.8075UCL=0.8796

LCL=0.7354

0

0.15

0.10

0.05

0.00

321

R Chart by Operator

Sample

Range

R=0.03833

UCL=0.1252

LCL=0

10987654321

1.1

1.0

0.9

0.8

0.7

0.6

0.5

0.4

Part

OperatorOperator*Part Interaction

Average

1

2

3

321

1.1

1.0

0.9

0.8

0.70.6

0.5

0.4

Operator

By Operator

10987654321

1.1

1.0

0.9

0.8

0.70.6

0.5

0.4

Part

By Part

%Contribution

%Study Var

%Tolerance

Part-to-PartReprodRepeatGage R&R

200

100

0

Components of Variation

Percent

Gage R&R (ANOVA) for Response

Minitab - Output

Expected To

Be Out of Control

%R&R &

P/T Ratios

Gives Us The 1st

Look At

Discrimination

Graphical View

Of Results

Mi i b O


18/50

18

What decision do we make?

Minitab - Output

%P/TBest case: 10% Acceptable: 30%

% R&R

As a target, look for %R&R < 30%

Gage R&RStdDev Study Var %Study Var %Tolerance

Source (SD) (5.15*SD) (%SV) (SV/Toler)

Total Gage R&R 0.066615 0.34306 32.66 68.61Repeatability 0.035940 0.18509 17.62 37.02

Reproducibility 0.056088 0.28885 27.50 57.77

Operator 0.030200 0.15553 14.81 31.11

Operator*Part 0.047263 0.24340 23.17 48.68

Part-To-Part 0.192781 0.99282 94.52 198.56

Total Variation 0.203965 1.05042 100.00 210.08

Number of Distinct Categories = 4

%P/T% R&R


19/50

19

Reducing recruiting costs Due to high costs and significantDue to high costs and significant

time needed to screen resumestime needed to screen resumes

submitted electronically, the teamsubmitted electronically, the team

decides to evaluate the screeningdecides to evaluate the screeningprocess for resumes submitted viaprocess for resumes submitted via

the web.the web. Think about the response. ResumesThink about the response. Resumes

are evaluated by team membersare evaluated by team members--what type of data is this?what type of data is this?


20/50

20

Attribute Agreement Analysis

Binary responseBinary response-- Pass/Fail.Pass/Fail. A resume either proceeds to theA resume either proceeds to the

next level or not.next level or not. 4 people are responsible for4 people are responsible for

screening resumes. Each isscreening resumes. Each ispresented with 50 resumes,presented with 50 resumes,

randomly selected and alreadyrandomly selected and alreadyjudged to be either a pass or fail.judged to be either a pass or fail.


21/50

21

Attribute Agreement Analysis

Appraiser

Percent

WendyNinaDavidAlex

100

95

90

85

80

95.0% C I

Percent

Appraiser

Percent

WendyNinaDavidAlex

100

95

90

85

80

95.0% C I

Percent

Assessment Agreement

Within Appraisers Appraiser vs Standard


22/50

22

MSA - Summary

IntroduceIntroduceMeasurement Systems AnalysisMeasurement Systems Analysis

Define basic measurement termsDefine basic measurement terms

Outline procedure for performing a GageOutline procedure for performing a Gage

Study (Measurement Systems Analysis)Study (Measurement Systems Analysis)


23/50

SPC-MINITAB

Application inApplication in

(Statistical Process Control, SPC)(Statistical Process Control, SPC)


24/50

24

S = Statistical techniques used to examine process variation

C = Controlling the process through active management

P = Process, ANY Process

The Way We Manage Data - Today

SPC


25/50

25

Control Chart Methods

Where Did It Come From ?

19201920ss -- Western Electric / Dr. Walter ShewhartWestern Electric / Dr. Walter Shewhart

Used to identify Controlled & Uncontrolled VariationUsed to identify Controlled & Uncontrolled Variation

Controlled:Controlled: Common Cause or InherentCommon Cause or InherentVariationVariation

Uncontrolled: Special Cause or AssignableUncontrolled: Special Cause or AssignableVariationVariation

Uses Control Charts as main toolUses Control Charts as main tool


26/50

26

Types of Variation

Common vs. Special

COMMON CAUSECOMMON CAUSE Is present in every processIs present in every process

Is produced by the process itself (the way we doIs produced by the process itself (the way we do

business)business) Can be removed and/or lessened but requires aCan be removed and/or lessened but requires a

fundamental change in the processfundamental change in the process

A process isA process is Stable, PredictableStable, Predictable, and, andInIn--ControlControl

when only Common Cause Variation exists in thewhen only Common Cause Variation exists in the

processprocess

Types of Variation


27/50

27

SPECIAL CAUSESPECIAL CAUSE

UnpredictableUnpredictable

Typically large in comparison to Common CauseTypically large in comparison to Common Causevariationvariation

Caused by unique disturbances or a series of themCaused by unique disturbances or a series of them

Can be removed/lessened by basic process controlCan be removed/lessened by basic process control

and monitoringand monitoring

A process exhibiting Special Cause variation is saidA process exhibiting Special Cause variation is said

to beto be OutOut--ofof--ControlControl andandUnstableUnstable

Types of Variation

Common vs. Special

Mi i b C l Ch


28/50

28

Minitab - Control Charts

]

]


29/50

29

DATA PLOTTED OVER TIME

M

ONITORED

CHA

RACTERI

STIC

UCL

Center Line

LCL

UCL = Upper Control Limit / LCL = Lower Control Limit

Plotted Data

Key Component - Control

Charts


30/50

30

A set of standard tests have been created to help identifySPECIAL CAUSE events in our processes

We use the phrase Out of Control when a test (or rule)has been broken.

The tests we suggest:Pattern rule: If you see a pattern, the process is out of control

The tests we suggest:Pattern rule: If you see a pattern, the process is out of control

Process Control Tests

This means something unusual has happened -

Go check it out!!Go check it out!!

R l f St d d D i ti


31/50

31

1 Sigma

2 Sigma

3 Sigma

1 Sigma

2 Sigma

3 Sigma

60-75%

90-98%

99-99.9%

% of Data PointsUCL

LCL

TIMETIME

The ItemWe Are

Measuring

Rules of Standard Deviation

Where should the data lie?

Minitab Tests


32/50

32

Minitab Tests

Test #1

Test #2 .


33/50

33

In Control or Out of Control ? ___________________

If out of Control, which rule(s) is broken or condition(s) is present?_______________________________________________________

3020100

10

5

0

-5

Observation Number

IndividualValue

I Chart for C1

X=0.2800

3.0SL=5.416

-3.0SL=-4.856

UCL

LCL


34/50

34

In Control or Out of Control ? ___________________

If out of Control, which rule(s) is broken or condition(s) is present?_______________________________________________________


35/50

35

If a point falls beyond the upper or lower controlcontrol limit

does this mean we are making a defect for the customer?

Control Limits vs.

Specification Limits

UCL

LCL

TIMETIME

Control Limits vs.


36/50

36

Specification Limits

Process Control Limits are calculated based on dataProcess Control Limits are calculated based on data

from the process itselffrom the process itself They are based on +/They are based on +/-- 33 (99.73% of the process(99.73% of the process

variation is expected to fall between these limits)variation is expected to fall between these limits)

Product Specification LimitsProduct Specification LimitsARE NOTARE NOTfound on thefound on thecontrol chartcontrol chart

Understanding how the process matches up againstUnderstanding how the process matches up againstcustomer requirementscustomer requirementsISIS important to knowimportant to know

To determine how the process performs to CustomerTo determine how the process performs to CustomerExpectations, aExpectations, a Process Capability StudyProcess Capability Study is requiredis required

Different Variables Control Charts


37/50

37

I-MR Individuals - Moving Range

X-Bar-R Average -Range Chart

Different Variables Control Charts

0Subgroup 10 2021

222324252627

282930

IndividualValue

X=25.42

3.0SL=29.00

-3.0SL=21.83

0

1

2

3

4

5

MovingRange

1

R=1.348

3.0SL=4.404

-3.0SL=0.00E

I and MR Chart for C1

0Subgroup 10 20 30

65

70

75

80

SampleMean

1 1

X=72.81

3.0SL=80.71

-3.0SL=64.90

0

10

20

30

SampleRange

1

R=13.70

3.0SL=28.97

-3.0SL=0.00E

Xbar/R Chart for Output


38/50

Different Attribute Control Charts : Defects


39/50

39

C Number of defectsU Number of defects per unit

Different Attribute Control Charts : Defects

Exercise: What Type of Control Chart?


40/50

40

1. # of typos per sales contract (C,U)

2. Number of notebooks with defects in monthly production (P)

3. % of defective vehicles in monthly production (I-MR,P)

4. Per accounts receivable, amount of time it takes to close it (I-MR_

5. Number of transmissions with defects per 100 built (NP)

Exercise: What Type of Control Chart?

Requirements for Effective Use


41/50

41

q

of Control Charts

ManagementManagement

MUSTMUST

establish and support anestablish and support an

environment that promotes proper action andenvironment that promotes proper action andsupport to the information collected on the controlsupport to the information collected on the controlchartscharts

Control Charts are implementedControl Charts are implemented ONLYONLYon Keyon KeyProcesses on which improvement will bringProcesses on which improvement will bringbenefit to the organization and/or the customerbenefit to the organization and/or the customer

Data collected from the process is validatedData collected from the process is validatedthrough the use of athrough the use of a CAPABLECAPABLEmeasurementmeasurement

systemsystem


42/50

42

SPC - Summary

Linked Control Chart Methods to theLinked Control Chart Methods to theDMAIC roadmapDMAIC roadmap

Discussed different types of variationDiscussed different types of variation

Introduced various Control Chart typesIntroduced various Control Chart types

Discussed the interpretation of ControlDiscussed the interpretation of Control

ChartsCharts


43/50

Case study (Healthcare & Banking)


44/50

44

Healthcare industry

MD Anderson Cancer Center (TX, USA)MD Anderson Cancer Center (TX, USA)

19981998--20002000

CT process reCT process re--arrangedarranged

Service capacity increased 28%Service capacity increased 28%

H l h i d


45/50

45

Healthcare industry _

City hospital of Taipei

XX--ray transparency delivery time< 30ray transparency delivery time< 30 minsmins

KPIVsKPIVs

No dedicated delivery personnelNo dedicated delivery personnel

No specific processNo specific process

MSAMSA

Attribute dataAttribute data SPCSPC

Control chartsControl charts

H l h i d


46/50

46

Healthcare industry_

Local hospital in Taichung

TQIPTQIP

Unexpected return rate within 15 daysUnexpected return rate within 15 days--


47/50

47

Banking

1.1. Customer waiting time


48/50


49/50

49

Thank you

Q& AQ& A

Control charts For data in subgroups:For data in subgroups:


50/50

50

For data in subgroups:For data in subgroups:

subgroup means,subgroup means, XbarXbar XbarXbar

subgroup ranges, rsubgroup ranges, r R ChartR Chart subgroup standard deviations, ssubgroup standard deviations, s S ChartS Chart

XbarXbar and r on same screenand r on same screen XbarXbar and Rand R

XbarXbar and s on same screenand s on same screen XbarXbar and Sand S

I chart, MR chart, and R chart for subgroups on the same screenI chart, MR chart, and R chart for subgroups on the same screen II--MRMR--R/SR/S

(Between/Within)(Between/Within) For individual observations:For individual observations:

individual observationsindividual observations IndividualsIndividuals

moving rangesmoving ranges Moving RangeMoving Range

individual observations and moving ranges on same screenindividual observations and moving ranges on same screen II--MRMR

For subgroup combinations:For subgroup combinations: exponentially weighted moving averagesexponentially weighted moving averages EWMAEWMA

moving averagesmoving averages Moving AverageMoving Average

cumulative sumscumulative sums CUSUMCUSUM

individual observations or subgroup means according to their disindividual observations or subgroup means according to their distance from thetance from the

center linecenter line ZoneZone For short runs:For short runs:

standardized individual observations and moving ranges from shorstandardized individual observations and moving ranges from short run processest run processesZZ--MRMR

All Rights Reserved. 2000 Minitab, Inc.All Rights Reserved. 2000 Minitab, Inc.

Date post:	08-Apr-2018
Category:	Documents
Upload:	afxi3
View:	250 times
Download:	0 times

yan MINITAB

Documents