Measurement System Analysis (MSA) Using MINITAB

To reduce the total variation Measurement Variation should be identified.

Total
variation

Process Measurement
Variation Variation

Variation within Variation within Measurement Operator

process Sample variability variability

Short term Long term Location Spread

Part-to-part
Bias Stability Linearity Repeatability Reproducibility
Variation
Total Variability Product
Variability

Measurement
Variability

=
+

2
total 2
product 2
measurement system
 Measurement System Analysis

Off Line
Resolution ( Precision )
Bias ( Accuracy )
Linearity

On line
Repeatability
Reproducibility
Stability
 Resolution ( Precision )
Bias ( Accuracy )
 Gage R&R Evaluating

Type of measurement system variation : Width or Spread

Repeatability : is the variation in
measurements obtained with one
measurement instrument when used
several times by an appraiser while
measuring the identical characteristic on
the same part / condition. Repeatability
Reproducibility : is the variation in the
average of the measurements made by
different appraisers using the same
measurement instrument when measure
the identical characteristic on the same
part / condition.

Reproducibility
Ref. : MSA from Ford / GM 1990
 Type of Data

Discrete (attribute) Data : Qualitative data that can be

counted and recorded to analysis.. Example, a shaft
diameter measured by Go-no go gage.
X = Good or No Good,
X = Pass or Fail

Variable / Continuous data : Quantitative data

measured by measurement system.. Example, height or
weight of student.
X = 150, 160, 170 cm,
X = 60, 65, 70 kg
 Case Study
Scenario :
TherewasanissuefeedbackfromHDDmanufacturingplantin
Singapore. Drives were facingHSAhittopcover.TheFAreported that
allunitsgotcartridgeheightproblem.
PossibleRootCauses:Correlation,CartridgeinstallationorBoth?
10cartridgeswererandomlyselectedfromassemblylinesatSG
plant.Cartridgeheightsweremeasuredby3differenttools(A,Band
C)eachmeasured3timesthensentbacktoThailandforinvestigation

Thailand Action Items :

1. MSA needed Gauge R&R both TH and SG plants
2. Test Hypothesis Paired T-test
3. Correlation Linear Regression Analysis
 BasicInformationabouttheData
OpenMINITABFileMEASUREMENTSYSTEMANALYSISEXAMPLE CARTRIDGEHEIGHT.MPJ
 BasicInformationabouttheData

Descriptive Statistics: Thailand Reading

Variable N N* Mean SE Mean StDev Minimum Q1 Median

Thailand Reading 90 0 15.097 0.128 1.210 13.762 14.118 14.681

Variable Q3 Maximum
Thailand Reading 15.574 17.423

Descriptive Statistics: Singapore Reading

Variable N N* Mean SE Mean StDev Minimum Q1 Median

Singapore Reading 90 0 15.289 0.125 1.185 13.930 14.314 14.899

Variable Q3 Maximum
Singapore Reading 15.756 17.633
 NormalityTestofData
Mostofthestatisticaltools,exceptfornonparametrictest,requirethatthedatadistribute
normally.Sobeforewecanusethesetools,thenormalityofdatamustbevalidated.

10partswererandomlyselectedandcartridgeheightdatawerealreadyavailableforNormality
Test.
 Thesetwodatapointsare
likelyoutliners(therareand
extreme).Theycanbe
discardedandreplacedwith
newsamples.

Ideally,alldatapointsshouldlieonthestraightline,iftheyweredistributednormally.In
practice,duetorandomnatureofthedata,theywillnot.

WethentestthehypothesisofnormalityofdatabylookingatthePvalue.Thenull
hypothesisisdatadistributenormally.Typically,IfthePvalueislessthan5%wecanreject
thenullhypothesis.

Inthiscasethenullhypothesiscannotberejectedandwecansaythatthedataprobably
distributenormally.

IfthePvalueislessthan5%,wecandiscardafewsamples,recollectandtestagainuntilpass.
 Action 1: Gauge R&R
OpenMINITABFileMEASUREMENTSYSTEMANALYSISEXAMPLE CARTRIDGEHEIGHT.MPJ
 Gauge R&R Thailand
Two-Way ANOVA Table Without Interaction

Source DF SS MS F P IfthePvalueislessthan5%,
S/N 9 129.556 14.3951 1567.93 0.000
Operator 2 0.009 0.0043 0.47 0.626 theparameter(source)
Repeatability 78 0.716 0.0092 contributiontototal
Total 89 130.281
variation(sumsquaredSS)is
consideredsignificant.
Gage R&R

%Contribution Parttopartvariationis
Source VarComp (of VarComp)
Total Gage R&R 0.00918 0.57 significant.
Repeatability 0.00918 0.57
Reproducibility 0.00000 0.00
Operator 0.00000 0.00 Repeatabilityoftoolsisgood,
Part-To-Part 1.59844 99.43 only0.57%
Total Variation 1.60762 100.00
Reproducibility(differentamong
tools)istoosmalltomeasure.
Study Var %Study Var
Source StdDev (SD) (6 * SD) (%SV)
Total Gage R&R 0.09582 0.57490 7.56
Repeatability 0.09582 0.57490 7.56
Reproducibility 0.00000 0.00000 0.00
Operator 0.00000 0.00000 0.00
Part-To-Part 1.26429 7.58577 99.71
Total Variation 1.26792 7.60752 100.00
Theresidualerrormustdistributenormally:e=NID(0,2)
 Gauge R&R Singapore

Two-Way ANOVA Table Without Interaction

Source DF SS MS F P
S/N 9 124.177 13.7975 1422.24 0.000
Operator 2 0.004 0.0018 0.18 0.834
Repeatability 78 0.757 0.0097
Total 89 124.938

Gage R&R

%Contribution
Source VarComp (of VarComp)
Total Gage R&R 0.00970 0.63
Repeatability 0.00970 0.63 Repeatabilityoftoolsisgood,
Reproducibility 0.00000 0.00 only0.63%
Operator 0.00000 0.00
Part-To-Part 1.53198 99.37 Reproducibility(differentamong
Total Variation 1.54168 100.00 tools)istoosmalltomeasure.

Study Var %Study Var

Source StdDev (SD) (6 * SD) (%SV)
Total Gage R&R 0.09849 0.59097 7.93
Repeatability 0.09849 0.59097 7.93
Reproducibility 0.00000 0.00000 0.00
Operator 0.00000 0.00000 0.00
Part-To-Part 1.23773 7.42638 99.68
Total Variation 1.24164 7.44986 100.00
Theresidualerrormustdistributenormally:e=NID(0,2)
 Gage R&R Combined
(to see the effects of measurement locations)
Two-Way ANOVA Table Without Interaction

Source DF SS MS F P
Part 9 253.681 28.1867 3097.44 0.000
Tester 1 1.662 1.6623 182.67 0.000
Repeatability 169 1.538 0.0091
Total 179 256.881

Gage R&R

%Contribution Nowwecanseethatplant
Source VarComp (of VarComp)
Total Gage R&R 0.02747 1.72
measurementsdiffersignificantly,
Repeatability 0.00910 0.57 statisticallyspeaking.
Reproducibility 0.01837 1.15
Tester 0.01837 1.15
Part-To-Part 1.56542 98.28
ThereadingdifferencesbetweenTH&SGplants
Total Variation 1.59289 100.00 (1.15%)isstatisticallysignificant(Pvalue=0.000
above)whichneedtobeinvestigatedfurther.
Study Var %Study Var
Source StdDev (SD) (6 * SD) (%SV)
Total Gage R&R 0.16574 0.99443 13.13
Repeatability 0.09539 0.57236 7.56
Reproducibility 0.13553 0.81320 10.74
Tester 0.13553 0.81320 10.74
Part-To-Part 1.25117 7.50702 99.13
Total Variation 1.26210 7.57259 100.00
Theresidualerrormustdistributenormally:e=NID(0,2)
 Action 2: Test of Hypothesis

Test Hypothesis is required to check both Mean and Variation.

Remark : We are using the exactly same parts for both

Thailand and Singapore.

This allows us to use the paired T-test technique for hypothesis

testing. This technique is much less sensitive to part-to-part
variation than the T-test and Z-test.

Note that the paired T-test requires data be distributed

normally and collected (or measured) in pair.
 HypothesisTestofEqualVariances

Referringtothebasicinfoofdata,wecanseethatitislikelythatthevariancesofbothdata
setsarethesame.Ifthisistrue,wecanusepooledvarianceforsubsequentanalysis.

BasicInformationabouttheData
Descriptive Statistics: Thailand Reading

Variable N N* Mean SE Mean StDev Minimum Q1 Median

Thailand Reading 90 0 15.097 0.128 1.210 13.762 14.118 14.681

Variable Q3 Maximum
Thailand Reading 15.574 17.423

Descriptive Statistics: Singapore Reading

Variable N N* Mean SE Mean StDev Minimum Q1 Median

Singapore Reading 90 0 15.289 0.125 1.185 13.930 14.314 14.899

Variable Q3 Maximum
Singapore Reading 15.756 17.633
Test and CI for Two Variances: Thailand Reading, Singapore Reading

Method

Null hypothesis Sigma(Thailand Reading) / Sigma(Singapore Reading) = 1

Alternative hypothesis Sigma(Thailand Reading) / Sigma(Singapore Reading) not = 1
Significance level Alpha = 0.05

Statistics

Variable N StDev Variance

Thailand Reading 90 1.210 1.464
Singapore Reading 90 1.185 1.404

Ratio of standard deviations = 1.021

Ratio of variances = 1.043

95% Confidence Intervals

CI for
Distribution CI for StDev Variance
of Data Ratio Ratio
Normal (0.828, 1.259) (0.686, 1.584)
Continuous (0.787, 1.348) (0.620, 1.817)
Thissaidvariancescouldbeequal.

Tests

Test
Method DF1 DF2 Statistic P-Value ThissaidthenullhypothesisHo
F Test (normal) 89 89 1.04 0.844 (equalvariance)cannotbe
Levene's Test (any continuous) 1 178 0.05 0.827 rejectedat5%risk.Theriskof
beingwrongis84.4%torejectit.
HypothesisTestofDifferenceinMeansusingPairedTtest
Paired T-Test and CI: Thailand Reading, Singapore Reading

Paired T for Thailand Reading - Singapore Reading

Thissaidwecanrejectthe
N Mean StDev SE Mean
Thailand Reading 90 15.097 1.210 0.128 nullhypothesisHo (equal
Singapore Reading 90 15.289 1.185 0.125 reading)at5%risk.
Difference 90 -0.1922 0.1267 0.0134

95% CI for mean difference: (-0.2187, -0.1657)

T-Test of mean difference = 0 (vs not = 0): T-Value = -14.39 P-Value = 0.000

Thissaidthemeanreadings
ofcartridgeheightsbetween
TH&SGplantsislikelytobe
different,95%confident.

Conclusionmaybesomethinglikethis:

Itis95%confidentthatTHmeasures
cartridgeheightslowerthanSGby
0.1922onaverage.Thishelpsexplain
whysomeHSAhittopcoveratSGplant.
ThedifferencemustdistributenormallyforpairedTtest.
Whathappensifweusewrongstatisticaltoolstoanalyzethedata?

IfsomeoneusesTtesttoanalyzethedata,theresultswilllooklikethis.
 Whathappensifweusewrongstatisticaltoolstoanalyzethedata?
Two-Sample T-Test and CI: Thailand Reading, Singapore Reading

Two-sample T for Thailand Reading vs Singapore Reading

N Mean StDev SE Mean

Thailand Reading 90 15.10 1.21 0.13
Singapore Reading 90 15.29 1.18 0.12

Difference = mu (Thailand Reading) - mu (Singapore Reading)

Estimate for difference: -0.192
95% CI for difference: (-0.544, 0.160)
T-Test of difference = 0 (vs not =): T-Value = -1.08 P-Value = 0.283 DF = 178
Both use Pooled StDev = 1.1974

Asmentionedbeforethatthettestandztestarequitesensitivetoparttopartvariation,
which,inthiscase,isverylarge( =1.2)comparedtothedifference(0.192)thatwecan
observe,itisnotsurprisedthenthattheresultsfromthis2samplettestcannotconfirm
thedifferencebetweencartridgeheightreadingsbetweenTHandSGplants.
 Action 3: Finding SG-TH Reading Correlation
Conclusionfrompreviousanalysiswas:

Itis95%confidentthatTHmeasurescartridgeheightslowerthanSGby0.1922on
average.ThishelpsexplainwhysomeHSAhittopcoveratSGplant.

Thequestionimmediatelyfollowedis:howthereadingsarecorrelated?Ifcorrelations
exist,wecanthenfindnewcriteriaforcartridgeheightspectoscreenpartsforSGplants.

Wethenbeginthenextstepwithsimplelinearregression.
Regression Analysis: Thailand Reading versus Singapore Reading

The regression equation is

Thailand Reading = - 0.432 + 1.02 Singapore Reading (y=a+b*x)

Predictor Coef SE Coef T P

Constant -0.4317 0.1730 -2.50 0.014
Singapore Reading 1.01566 0.01128 90.03 0.000
Bothaandbaresignificant.

S = 0.126091 R-Sq = 98.9% R-Sq(adj) = 98.9% Regressionmodelcanexplain98.9%ofthe

totalspread(SS)indata.
Analysis of Variance

Source DF SS MS F P
Regression 1 128.88 128.88 8106.28 0.000 Regressionissignificantcomparedtoerror.
Residual Error 88 1.40 0.02
Total 89 130.28

Modelchecking:
IserrorNID(0,2)?
Ans:Yes
RegressionPlot
 The regression equation is
Thailand Reading = - 0.432 + 1.02 Singapore Reading

Ifmaximumcartridgeheight(beforeHSAhittingtopcover)atSGplantis17.5,whatshould
bethecartridgeheightspecforTHplantscreeningbeforesendingpartstoSG?

OverallCorrectiveActions:
TH: Cutinnewspecforcartridgeheightscreening.
UsetheregressionmodelabovetopredictreadingsatSG.
SetuptesterSPC,ProductSPCforyieldandperformancemonitoring.

SG: Tightcontrolofcartridgeinstallationprocess.
Continuousmonitoringandfeedback.