6 ssgb amity bsi msa p

73
Module-6 Copyright © 2012 BSI. All rights reserved. Module-6 Measurements Systems Analysis

Upload: ankursingh

Post on 30-Sep-2015

12 views

Category:

Documents


0 download

TRANSCRIPT

  • Module-6

    Copyright 2012 BSI. All rights reserved.

    Module-6Measurements Systems Analysis

  • Measurements Systems Analysis - Agenda

    1. Is our data accurate?

    Repeatability & Reproducibility

    Accuracy & Precision

    DMAIC

    Measurements System Variation

    Bias, Linearity, Stability, Repeatability, Reproducibility, Calibration, Gauge R&R

    2. Variable Gauge R&R

    Copyright 2012 BSI. All rights reserved.

    2

    Parts, Operators, Variation

    Is the gauge good?

    Workshop

    3. Attribute Gauge R&R

    Workshop

    4. Appendix

    Analysis of Variance (ANOVA)

  • Gauge R & R is a means of assessing the repeatability and reproducibility of our measurement systems.

    Gauge R & R studies are carried out in order to discover how much of the process variation is due to the measurement device and measurement methods.

    Is our Data Accurate?

    Copyright 2012 BSI. All rights reserved.

    3

    Dimension

    ?

  • Define ImproveMeasure Control Control Critical xs

    Monitor ys1 5 10 15 20

    10.2

    10.0

    9.8

    9.6

    Upper Control Limit

    Lower Control Limit

    Analyse Characterise xs

    Optimise xs

    y=f(x1,x2,..)

    y

    x

    . . .

    . . .

    . .

    . . .

    . . .

    Identify Potential xs

    Analyse xs

    Run 1 2 3 4 5 6 7

    1 1 1 1 1 1 1 1

    Effect

    C1 C2

    C4

    C3

    C6C5

    Select Project Define Project

    Objective Form the Team

    Map the Process

    Define Measures (ys)

    Evaluate Measurement System

    Determine Process

    DMAIC Improvement Process

    Copyright 2012 BSI. All rights reserved.

    4

    Validate Control Plan

    Close Project

    y

    Phase Review

    Set Tolerances for xs Verify Improvement

    15 20 25 30 35

    LSL USL

    Phase Review

    Select Critical xs

    Phase Review

    1 1 1 1 1 1 1 12 1 1 1 2 2 2 23 1 2 2 1 1 2 24 1 2 2 2 2 1 15 2 1 2 1 2 1 26 2 1 2 2 1 2 17 2 2 1 1 2 2 18 2 2 1 2 1 1 2

    x

    xx

    xx

    xx

    xx

    x

    x

    Identify Customer Requirements

    Identify Priorities Update Project File

    Phase Review

    Determine Process Stability

    Determine Process Capability

    Set Targets for Measures

    15 20 25 30 35

    LSL USL

    Phase Review

  • Measurement is accurate but not precise

    Measurement is precise but not accurate

    Measurement Accuracy & Precision

    Copyright 2012 BSI. All rights reserved.

    5

    Measurement is accurate and precise

    precise accurate

  • Measurement

    Accuracy

    Bias

    Linearity

    Measurement System Variation

    Copyright 2012 BSI. All rights reserved.

    6

    MeasurementSystemVariation

    Reproducibility

    Repeatability

    Stability

    Precision

  • Bias

    Bias

    Copyright 2012 BSI. All rights reserved.

    7

    ObservedAverage

    TrueValue

    Bias is the difference between the observed average of the measurements and the true value.

  • Me

    a

    s

    u

    r

    e

    d

    V

    a

    l

    u

    e Non-Linearity

    Gauge is measuring lower than true value at high end

    Linearity

    Copyright 2012 BSI. All rights reserved.

    8

    Linearity is the difference in bias values over the expected operating range of the measurement gauge.

    Reference Value

    M

    e

    a

    s

    u

    r

    e

    d

    V

    a

    l

    u

    e

  • Stability

    Stability

    Copyright 2012 BSI. All rights reserved.

    9

    Time2

    Time1

    Stability is the variation (differences) in the average over extended periods of time usingthe same gauge and appraiser to repeatedly measure the same part

  • Repeatability

    Repeatability

    Copyright 2012 BSI. All rights reserved.

    10

    Repeatability is the variation between successive measurements of the same part, samecharacteristic, by the same person using the same gauge.

  • Reproducibility

    Reproducibility

    Copyright 2012 BSI. All rights reserved.

    11

    Operator2

    Operator1

    Reproducibility is the difference in the average of the measurements made by differentpeople using the same instrument when measuring the identical characteristic on thesame pieces.

  • Measurement System Variation

    Accuracy

    Stability

    Bias

    Linearity Calibration

    Copyright 2012 BSI. All rights reserved.

    12

    Reproducibility

    Repeatability

    Stability

    Precision Gauge R&R

  • Calibration

    The Bias of a gauge can be assessed by repeat measurements of a known reference unit

    This can be extended across the operating range of the gauge in a Gauge Linearity Study

    The Stability of the gauge can be assessed by control charting a reference unit

    Should not routinely recalibrate, instead if reference unit tests outside the control limits, then

    Copyright 2012 BSI. All rights reserved.

    13

    Should not routinely recalibrate, instead if reference unit tests outside the control limits, then

    re-calibrate

    If measurement device requires frequent recalibration, attempt to improve stability

  • Gauge R & R is a means of assessing the repeatability and reproducibility of our measurement systems.

    Gauge R & R studies are carried out in order to discover how much of the process variation is due to the measurement device and measurement methods.

    Gauge R & R

    Copyright 2012 BSI. All rights reserved.

    14

    Dimension

    ?

  • Variable Gauge R&R

    Copyright 2012 BSI. All rights reserved.

    Variable Gauge R&R

  • Variable Gauge R&R

    Requirements:

    A minimum of two operators (recommend 3 or 4)

    At least 10 parts which should be chosen to represent the full range of manufacturing variation

    (it may be acceptable to use fewer parts in some special cases)

    Copyright 2012 BSI. All rights reserved.

    16

    Part 1

    Part 4

    Part 2

    Part 3

    Part 10

    Part 5

    Each part should be measured two or three times in a random order

    Operators should not be aware of the previous result when measuring the same part

  • There are two methods available:

    1. Analysis of Variance (ANOVA)

    2. X-Bar and R

    Variable Gauge R&R

    Copyright 2012 BSI. All rights reserved.

    17

    The ANOVA method is:

    the recommended approach

    takes into account any interactive effect between operator and part

  • Reproducibility

    Part-to-PartVariation

    Operator

    Operator

    Variable Gauge R&R

    OverallVariation

    Copyright 2012 BSI. All rights reserved.

    18

    MeasurementSystemVariation

    Repeatability

    Operatorby part

    Interaction

    We want the Part-to-Part component to be large!

  • Part Operator 1 Operator 2 Operator 31 0.65 0.55 0.501 0.60 0.55 0.552 1.00 1.05 1.052 1.00 0.95 1.003 0.85 0.80 0.803 0.80 0.75 0.804 0.85 0.80 0.804 0.95 0.75 0.805 0.55 0.40 0.45

    Variable Gauge R&R - Example

    Copyright 2012 BSI. All rights reserved.

    19

    5 0.55 0.40 0.455 0.45 0.40 0.506 1.00 1.00 1.006 1.00 1.05 1.057 0.95 0.95 0.957 0.95 0.90 0.958 0.85 0.75 0.808 0.80 0.70 0.809 1.00 1.00 1.059 1.00 0.95 1.0510 0.60 0.55 0.8510 0.70 0.50 0.80

  • Open Worksheet: Gauge R&R

    The Part numbers being measured

    Operators performing measurements

    Each operator measures each part twice

    Individual measurements

    Variable Gauge R&R - Minitab

    Copyright 2012 BSI. All rights reserved.

    20

    measurements Individual measurements

    In Minitab the data is entered in single columns

  • Variable Gauge R&R - Minitab

    Stat>Quality Tools>Gage Study>Gage R& R (Crossed)Enter Part, Operator, MeasurementCheck ANOVA Method

    Copyright 2012 BSI. All rights reserved.

    21

  • Variable Gauge R&R - Minitab

    Two-Way ANOVA Table With Interaction

    Source DF SS MS F PPart 9 2.05871 0.228745 39.7178 0.000Operator 2 0.04800 0.024000 4.1672 0.033Part * Operator 18 0.10367 0.005759 4.4588 0.000Repeatability 30 0.03875 0.001292Total 59 2.24913

    Gage R&R

    Copyright 2012 BSI. All rights reserved.

    22

    Gage R&R

    %ContributionSource VarComp (of VarComp)Total Gage R&R 0.0044375 10.67Repeatability 0.0012917 3.10Reproducibility 0.0031458 7.56

    Operator 0.0009120 2.19Operator*Part 0.0022338 5.37

    Part-To-Part 0.0371644 89.33Total Variation 0.0416019 100.00

    p

  • Variance Component Estimates

    OverallVariation0.0416019 Reproducibility

    0.0031458

    Part-to-PartVariation0.0371644 Operator

    0.0009120

    Operator

    Copyright 2012 BSI. All rights reserved.

    23

    Repeatability0.0012917

    MeasurementSystemVariation0.0044375

    Operatorby part

    Interaction0.0022338

    Variances are additive!

  • Variable Gage R&R Standard Deviations

    Study Var %Study Var

    Source StdDev (SD) (6 * SD) (%SV)Total Gage R&R 0.066615 0.39969 32.66

    Repeatability 0.035940 0.21564 17.62

    This is the gauge standard deviation, R&R = 0.066615Remember that standard deviations are not additive!

    Copyright 2012 BSI. All rights reserved.

    24

    Repeatability 0.035940 0.21564 17.62

    Reproducibility 0.056088 0.33653 27.50

    Operator 0.030200 0.18120 14.81

    Operator*Part 0.047263 0.28358 23.17

    Part-To-Part 0.192781 1.15668 94.52

    Total Variation 0.203965 1.22379 100.00

  • We would like the total measurement system variation (Gauge R&R) to be as small as possible. Calculate the percentage of the process tolerance taken up by the measurement system variation, represented by 6 x the gauge standard deviation. This is known as %Precision/Tolerance or %P/T.

    Interpreting the Results

    Copyright 2012 BSI. All rights reserved.

    25

    %P/T.

    The Process Tolerance is equivalent to the difference between the upper and lower specification limits (USL LSL).

    Tolerance Process6%100%P/T R&R=

  • Is the Gauge Good?

    % P/T(6R&R/Process Tolerance)

    Acceptability

    0 - 10% Very Good (Six Sigma Gauge)

    10 - 30% May be Acceptable

    Copyright 2012 BSI. All rights reserved.

    26

    The interpretation will also depend on the current level of process variation

    >30% Probably Not Acceptable

    Note that these guidelines are as recommended in Measurement Systems Analysis Third Edition published in March 2002 as part ofQS-9000 and developed in conjunction with AIAG.

  • Is the Gauge Good?

    % R&R

    If the %P/T is greater than 10%, then a secondary calculation can be used to decide whether the gauge can be used during the DMAIC activity.

    Comparing R&R to the current process variation indicates whether the measurement device is currently causing a problem. This is known as %R&R.

    We need an independent estimate of the process (total) variation (the value from the Gauge

    Copyright 2012 BSI. All rights reserved.

    27

    We need an independent estimate of the process (total) variation (the value from the Gauge R&R is based on only a few samples)

    We would like the measurement standard deviation to be less than the total standard deviation

    50% 100%R&%R

    tal)Process(to

    R&R

  • 1. Comparing the gauge variation to the process tolerance:

    This is greater than 10% so the gauge will not be good enough for six sigma. As the process improves the gauge will become a problem. To improve this gauge we should start by addressing the reproducibility.

    28.5%100%1.40.06666

    100%Tolerance

    %P/T 6 R&R ===

    Interpreting the Results

    Copyright 2012 BSI. All rights reserved.

    28

    addressing the reproducibility.

    2. Comparing the gauge variation to the process variation:

    This is less than 50% so the gauge is not the limiting factor at the moment. We can use this gauge for process improvement.

    37%100%0.18

    0.0666 100%R&%R

    tal)Process(to

    R&R===

  • Open Worksheet: GaugeR&RStat>Quality Tools>Gage Study>Gage R& R (Crossed)Enter Part, Operators, MeasurementCheck ANOVA MethodSelect Options: Study Variation: 6

    Process Tolerance: 1.4Historical standard deviation: 0.18

    Two-Way ANOVA Table With Interaction

    Variable Gauge R&R - Minitab

    Copyright 2012 BSI. All rights reserved.

    29

    Two-Way ANOVA Table With Interaction

    Source DF SS MS F PPart 9 2.05871 0.228745 39.7178 0.000Operator 2 0.04800 0.024000 4.1672 0.033Part * Operator 18 0.10367 0.005759 4.4588 0.000Repeatability 30 0.03875 0.001292Total 59 2.24913

  • Components of Variation

    Components of VariationGage R&R

    %ContributionSource VarComp (of VarComp)Total Gage R&R 0.0044375 10.67

    Repeatability 0.0012917 3.10Reproducibility 0.0031458 7.56

    Operator 0.0009120 2.19Operator*Part 0.0022338 5.37

    Part-To-Part 0.0371644 89.33Total Variation 0.0416019 100.00

    Copyright 2012 BSI. All rights reserved.

    30

    Study Var %Study Var %Tolerance %ProcessSource StdDev (SD) (6 * SD) (%SV) (SV/Toler) (SV/Proc)Total Gage R&R 0.066615 0.39969 32.66 28.55 37.01

    Repeatability 0.035940 0.21564 17.62 15.40 19.97Reproducibility 0.056088 0.33653 27.50 24.04 31.16

    Operator 0.030200 0.18120 14.81 12.94 16.78Operator*Part 0.047263 0.28358 23.17 20.26 26.26

    Part-To-Part 0.192781 1.15668 94.52 82.62 107.10Total Variation 0.203965 1.22379 100.00 87.41 113.31

    Number of Distinct Categories = 4

  • Components of Variation

    120

    100

    % Contribution

    % Study Var

    % Process

    % Tolerance

    Gage name:

    Date of study :

    Reported by :

    Tolerance:

    Misc:

    Components of Variation

    Gage R&R (ANOVA) for Measurement

    Copyright 2012 BSI. All rights reserved.

    31

    P

    e

    r

    c

    e

    n

    t

    Part-to-PartReprodRepeatGage R&R

    100

    80

    60

    40

    20

    0

    % Tolerance

  • Part to Part Measurements

    1.1

    1.0

    Gage name:

    Date of study :

    Reported by :

    Tolerance:

    Misc:

    Measurement by Part

    Gage R&R (ANOVA) for Measurement

    Copyright 2012 BSI. All rights reserved.

    32

    Part

    10987654321

    0.9

    0.8

    0.7

    0.6

    0.5

    0.4

  • Operator by Part Interaction

    1.1

    1.0

    Operator

    1

    2

    3

    Gage name:

    Date of study :

    Reported by :

    Tolerance:

    Misc:

    Operator * Part Interaction

    Gage R&R (ANOVA) for Measurement

    Copyright 2012 BSI. All rights reserved.

    33

    Part

    A

    v

    e

    r

    a

    g

    e

    10 9 8 7 6 5 4 3 2 1

    0.9

    0.8

    0.7

    0.6

    0.5

    0.4

    3

  • Measurements by Operator

    1.1

    1.0

    Gage name:

    Date of study :

    Reported by :

    Tolerance:

    Misc:

    Measurement by Operator

    Gage R&R (ANOVA) for Measurement

    Copyright 2012 BSI. All rights reserved.

    34

    Operator

    321

    0.9

    0.8

    0.7

    0.6

    0.5

    0.4

  • Xbar and R Chart by Operator

    M

    e

    a

    n

    1.0

    0.8

    __X=0.8075UCL=0.8796

    LCL=0.7354

    1 2 3

    Gage name:

    Date of study :

    Reported by :

    Tolerance:

    Misc:

    Xbar Chart by Operator

    Gage R&R (ANOVA) for Measurement

    Copyright 2012 BSI. All rights reserved.

    35

    S

    a

    m

    p

    l

    e

    R

    a

    n

    g

    e

    0.12

    0.08

    0.04

    0.00

    _R=0.0383

    UCL=0.1252

    LCL=0

    1 2 3

    S

    a

    m

    p

    l

    e

    M

    0.8

    0.6

    0.4

    X=0.8075LCL=0.7354

    R Chart by Operator

  • Rounding Errors

    Rounding is another component of measurement variation which needs to be minimised

    It can be shown that to avoid rounding error getting in the way of achieving six sigma quality, it

    is necessary to have a minimum of 14 discrete values between the upper and lower specification

    For one-side specifications, there need to be at least 7 discrete values between the process

    Copyright 2012 BSI. All rights reserved.

    36

    average and the specification limit

    UG 37

  • Rounding Errors - Interpolating

    If possible interpolate between graduation marks

    For example, thermometers are frequently marked to the nearest degree but can be read to the nearest 0.2 degrees, even if the last digit is not entirely accurate

    Interpolating frequently reduces and never increases the measurement variation

    Copyright 2012 BSI. All rights reserved.

    37

  • Improving the Measurement System

    Gauge incapable:

    Repeatability (Gauge)

    Take multiple measurements and use average (short term fix)

    Mistake proofing (e.g. provision of tooling to hold part during measurement)

    May need maintenance

    Reproducibility (Operators)

    Copyright 2012 BSI. All rights reserved.

    38

    Use 1 operator (short term fix during improvement only)

    Have several operators measure the part and take the average (short term fix)

    Ensure consistency (training, SOPs, WIS, )

    Mistake proofing (e.g. provision of tooling to hold part during measurement)

    Calibrations on the gauge dial may not be clear

    Reproducibility Operator x Part Interaction

    Identify cause of interaction and then as Operator

  • Destructive Gauge R&R

    Destructive gauge testing means that it is impossible to carry out repeat tests!

    To complete an assessment of a destructive gauge it is therefore necessary to assume homogeneity within batches.

    If there is much more difference in parts between batches than within batches, then a standard variable Gauge R & R may be sufficient.

    Copyright 2012 BSI. All rights reserved.

    39

  • Workshop Variable Gauge R&R

    Using the provided measuring device and products carry out a Gauge R&R

    Use three operators and measure each part twice

    Ensure that the order of measuring is randomised

    Analyse the data using Minitab

    Copyright 2012 BSI. All rights reserved.

    40

    Analyse the data using Minitab

    What could you do, if anything to improve the Measurement System?

    Prepare a short report detailing your findings

  • Attribute Gauge R&R

    Copyright 2012 BSI. All rights reserved.

    Attribute Gauge R&R

  • Attribute Gauge R&R

    A Gauge R&R study can also be carried out on attribute data

    Using attribute data, we would have a problem with the measurement system if:

    Operators disagree with each others evaluation of a piece

    The same operator gains different results from a repeat evaluation of the same piece

    Copyright 2012 BSI. All rights reserved.

    42

  • Attribute Measurement System

    An attribute measurement system compares each part to a standard and either accepts or rejects

    the part.

    The screen effectiveness is the ability of the attribute measurement system to properly

    discriminate good from bad.

    Copyright 2012 BSI. All rights reserved.

    43

    Screen effectiveness of 100% is desirable.

  • Conducting Attribute Gauge R&R

    1. Select a minimum of 30 parts from the process. These parts should represent the full

    spectrum of process variation (good parts, defective parts, borderline parts).

    2. An expert inspector performs an evaluation of each part, classifying it as Good or Not

    Good.

    Copyright 2012 BSI. All rights reserved.

    44

    3. Independently and in a random order, each of 2 or 3 operators should assess the parts as

    Good or Not Good.

    4. Calculate effectiveness scores.

  • Attribute Gauge R&R

    Column containing parts being assessed

    Text column containing expert

    assessment (can use words or numbers but

    must be consistent)

    Column containing parts being assessed

    Text column containing expert

    assessment (can use words or numbers but

    must be consistent)

    Open Worksheet: Attribute Gage R&RMinitab Data Layout:

    Copyright 2012 BSI. All rights reserved.

    45

    Text column containing operator

    performing measurements

    Text column containing results of measurements (can

    use words or numbers but must be consistent)

    Text column containing operator

    performing measurements

    Text column containing results of measurements (can

    use words or numbers but must be consistent)

  • Attribute Gauge R&R

    Stat>Quality Tools>Attribute Agreement AnalysisEnter Results in Attribute Column, Part in Samples, Appraiser in Appraisers and Expert in Known standard/attributeClick on Results button and select Percentages

    Copyright 2012 BSI. All rights reserved.

    46

  • Attribute Gauge R&R - Results

    Attribute Agreement AnalysisWithin AppraiserAssessment AgreementAppraiser # Inspected # Matched Percent (%) 95.0% CI A 30 28 93.3 ( 77.9, 99.2)B 30 30 100.0 ( 90.5, 100.0)C 30 30 100.0 ( 90.5, 100.0)# Matched: Appraiser agrees with him/herself across trials.

    Appraiser A was not consistent on two out of

    thirty parts inspected

    Copyright 2012 BSI. All rights reserved.

    47

    Each Appraiser vs StandardAssessment Agreement

    Appraiser # Inspected # Matched Percent (%) 95.0% CI A 30 28 93.3 ( 77.9, 99.2)B 30 29 96.7 ( 82.8, 99.9)C 30 29 96.7 ( 82.8, 99.9)# Matched: Appraiser's assessment across trials agrees with standard.

    Appraiser A disagreed with expert on two parts,

    Appraiser B and C disagreed with expert on

    one part

  • Attribute Gauge R&R - Results

    Assessment Disagreement # Not Good/ # Good/

    Appraiser Good Percent (%) Not Good Percent (%) # Mixed Percent (%) A 0 0.0 0 0.0 2 6.7 B 1 6.7 0 0.0 0 0.0 C 1 6.7 0 0.0 0 0.0 # Not Good/Good: Assessments across trials = Not Good / standard = Good.# Good/Not Good: Assessments across trials = Good / standard = Not Good.# Mixed: Assessments across trials are not identical.

    Between AppraisersAssessment Agreement Appraiser A,B and C agreed

    Appraiser B assessed one part as Not Good when the

    standard (expert) assessed it as Good

    Assessment Disagreement # Not Good/ # Good/

    Appraiser Good Percent (%) Not Good Percent (%) # Mixed Percent (%) A 0 0.0 0 0.0 2 6.7 B 1 6.7 0 0.0 0 0.0 C 1 6.7 0 0.0 0 0.0 # Not Good/Good: Assessments across trials = Not Good / standard = Good.# Good/Not Good: Assessments across trials = Good / standard = Not Good.# Mixed: Assessments across trials are not identical.

    Between AppraisersAssessment Agreement Appraiser A,B and C agreed

    Appraiser B assessed one part as Not Good when the

    standard (expert) assessed it as Good

    Copyright 2012 BSI. All rights reserved.

    48

    Assessment Agreement# Inspected # Matched Percent (%) 95.0% CI

    30 26 86.7 ( 69.3, 96.2)# Matched: All appraisers' assessments agree with each other.

    All Appraisers vs StandardAssessment Agreement# Inspected # Matched Percent (%) 95.0% CI

    30 26 86.7 ( 69.3, 96.2)# Matched: All appraisers' assessments agree with standard.

    Appraiser A,B and C agreed on 26 out of 30 parts

    inspected

    Appraiser A,B and C all agreed with the standard on 26 out of 30 parts inspected

    Assessment Agreement# Inspected # Matched Percent (%) 95.0% CI

    30 26 86.7 ( 69.3, 96.2)# Matched: All appraisers' assessments agree with each other.

    All Appraisers vs StandardAssessment Agreement# Inspected # Matched Percent (%) 95.0% CI

    30 26 86.7 ( 69.3, 96.2)# Matched: All appraisers' assessments agree with standard.

    Appraiser A,B and C agreed on 26 out of 30 parts

    inspected

    Appraiser A,B and C all agreed with the standard on 26 out of 30 parts inspected

  • Attribute Gauge R&R - Results

    100

    95

    95.0% C I

    Percent100

    95

    95.0% C I

    Percent

    Date of study:

    Reported by:

    Name of product:

    Misc:

    Assessment Agreement

    Within Appraisers Appraiser vs Standard

    Copyright 2012 BSI. All rights reserved.

    49

    Appraiser

    P

    e

    r

    c

    e

    n

    t

    CBA

    90

    85

    80

    Appraiser

    P

    e

    r

    c

    e

    n

    t

    CBA

    90

    85

    80

  • Attribute Gauge R&R - Results

    The target effectiveness is always 100%

    Possible Corrective Actions include:

    Operator Training

    Clarification of Standards

    Copyright 2012 BSI. All rights reserved.

    50

    Simplification of Standards

    Conversion to Variable Data

  • Workshop Attribute Gauge R&R

    From your team select two expert inspectors.

    The experts should select 20 sweets, roughly half good (pass) and half bad (fail).

    Some sweets should be borderline.

    Carry out a Gauge R&R

    Use two operators and measure each part twice (if more time available use three operators)

    Copyright 2012 BSI. All rights reserved.

    51

    Use two operators and measure each part twice (if more time available use three operators)

    Ensure that the order of measuring is randomised

    Analyse the data

    What could you do, if anything, to improve the Measurement System?

    Prepare a short report detailing your findings.

  • Measurement Systems Analysis - Summary

    Measurement errors can account for a large proportion of the variation in our measures (ys)

    We must evaluate our measurement systems before assessing process stability or process

    capability

    Errors in measurement systems can come from a variety of sources

    Copyright 2012 BSI. All rights reserved.

    52

    Errors in measurement systems can come from a variety of sources

    Action should be taken to improve the capability of our measurement systems if they are found

    to be inadequate

  • Define ImproveMeasure Control Control Critical xs

    Monitor ys1 5 10 15 20

    10.2

    10.0

    9.8

    9.6

    Upper Control Limit

    Lower Control Limit

    Analyse Characterise xs

    Optimise xs

    y=f(x1,x2,..)

    y

    x

    . . .

    . . .

    . .

    . . .

    . . .

    Identify Potential xs

    Analyse xs

    Run 1 2 3 4 5 6 7

    1 1 1 1 1 1 1 1

    Effect

    C1 C2

    C4

    C3

    C6C5

    Select Project Define Project

    Objective Form the Team

    Map the Process

    Define Measures (ys)

    Evaluate Measurement System

    Determine Process

    DMAIC Improvement Process

    Copyright 2012 BSI. All rights reserved.

    53

    Validate Control Plan

    Close Project

    y

    Phase Review

    Set Tolerances for xs Verify Improvement

    15 20 25 30 35

    LSL USL

    Phase Review

    Select Critical xs

    Phase Review

    1 1 1 1 1 1 1 12 1 1 1 2 2 2 23 1 2 2 1 1 2 24 1 2 2 2 2 1 15 2 1 2 1 2 1 26 2 1 2 2 1 2 17 2 2 1 1 2 2 18 2 2 1 2 1 1 2

    x

    xx

    xx

    xx

    xx

    x

    x

    Identify Customer Requirements

    Identify Priorities Update Project File

    Phase Review

    Determine Process Stability

    Determine Process Capability

    Set Targets for Measures

    15 20 25 30 35

    LSL USL

    Phase Review

  • Appendix - ANOVA

    Copyright 2012 BSI. All rights reserved.

    Appendix - ANOVA

  • ANOVA Table - Construction

    Construction of an Analysis of Variance (ANOVA) table requires the following:

    1. Identification of the Sources (Components) of Variation

    2. Calculation of the Sum of Squares due to each Source of Variation

    3. Assignment of the appropriate Degrees of Freedom

    Copyright 2012 BSI. All rights reserved.

    55

    3. Assignment of the appropriate Degrees of Freedom

    4. Calculation of the Mean Squares

    5. Calculation of the F-Ratio

  • Analysis of Variance (ANOVA) allows the decomposition of the variability in the Gauge R&R study.

    The components of variation in the Gauge R&R study are:

    2Part = Variation due to the different parts

    1. Components of Variation

    Copyright 2012 BSI. All rights reserved.

    56

    Part = Variation due to the different parts2Operator = Variation due to different operators2Operator x Part = Variation due to the interaction between operator

    and part2Repeatability = Variation due to gauge repeatability2Total =

    2Part +

    2Operator +

    2Operator x Part +

    2Repeatability

  • The total sum of squares is calculated as follows:

    Strictly speaking the sum of squares column is the sum of squares around the mean, known as the corrected sum of squares. We always use the corrected sum of squares when estimating variation.

    ( ) ( )n

    yyyySSTotal

    222

    ==

    2. Calculation of the Sum of Squares

    Copyright 2012 BSI. All rights reserved.

    57

    use the corrected sum of squares when estimating variation.

    ( ) ( ) 2491.21234.393725.416045.483725.41

    3725.4180.0...........00.100.160.065.0

    45.4880.0.............00.100.160.065.0

    222

    222222

    ====

    =+++++=

    =+++++=

    n

    yySS

    y

    y

    Total

  • ( ) ( ) ( ) ( ) ( )

    ( ) ( ) ( ) ( ) ( )6045.48

    600.4.........80.405.640.3

    .........

    22222

    2210

    23

    22

    21

    ++++=

    ++++=

    Part

    pPart

    SS

    n

    yn

    PPPPSS

    The sum of squares due to parts is calculated as follows:

    Calculation of the Sum of Squares

    Copyright 2012 BSI. All rights reserved.

    58

    0587.21234.391821.41 ==PartSS

    Where:P1, P2, P3..P10 are the Sums for each Parti.e the Sum of the 6 measurements made on each part.np is the number of individual measurements of each part.

  • The sum of squares due to operators is calculated as follows:

    (((( )))) (((( )))) (((( )))) (((( ))))(((( )))) (((( )))) (((( )))) (((( ))))

    6045.48

    2055.1635.1555.16 2222

    223

    22

    21

    ++++++++====

    ++++++++====

    Operator

    o

    Operator

    SS

    n

    yn

    OOOSS

    Calculation of the Sum of Squares

    Copyright 2012 BSI. All rights reserved.

    59

    Where:O1, O2, O3 are the Sums for each Operatori.e the sum of the 20 measurements made by each operator.no is the number of measurements made by each operator.

    0480.01234.391714.39 ========OperatorSS

  • The sum of squares due to the interaction between operators and parts is calculated as follows:

    ( ) ( ) ( ) ( )( ) ( ) ( ) ( ) 0587.20480.0

    6045.48

    265.1........00.225.1

    ..........

    2222

    22103

    221

    211

    ++=

    ++=

    PartOperator

    PartOperatorPO

    PartOperator

    SS

    SSSSn

    yn

    POPOPOSS

    Calculation of the Sum of Squares

    Copyright 2012 BSI. All rights reserved.

    60

    Where:O1P1, O1P2,.O3P10 are the Sums for each Operator & Part Combinationi.e the sum of the 2 measurements made by each operator on each part.nOxP is the number of measurements made by each operator on each part.

    1037.00587.20480.01234.393338.41

    0587.20480.0602

    ==

    PartOperator

    PartOperator

    SS

    SS

  • The sum of squares due to repeatability is obtained by subtraction:

    ityRepeatabil = SSSSSSSSSS PartOperatorOperatorPartTotal

    Calculation of the Sum of Squares

    Copyright 2012 BSI. All rights reserved.

    61

    0387.01037.00480.00587.22491.2ityRepeatabil ==SS

  • Source of Variation

    Between PartsBetween OperatorsOperator x Part

    Sum of Squares

    2.05870.04800.1037

    Calculation of the Sum of Squares

    Copyright 2012 BSI. All rights reserved.

    62

    Operator x PartRepeatability

    Total

    0.10370.0387

    2.2491

  • Degrees of Freedom is a statistical concept relating to the number of paired comparisons required to distinguish between items.

    For example, we need to find the tallest person out of 3 people. 2 comparisons would be required:

    3. Degrees of Freedom

    Copyright 2012 BSI. All rights reserved.

    63

    people. 2 comparisons would be required:

    Person 1 v Person 2Tallest v Person 3

    We would then know who the tallest person is.

  • The following rules apply to Degrees of Freedom:

    DF for a Factor (Main Effect) = (Number of Levels) 1

    DF for interactions = Product of the DF of the Factors involved

    Rules for Degrees of Freedom

    Copyright 2012 BSI. All rights reserved.

    64

    DF for interactions = Product of the DF of the Factors involved

    DF for Repeatability = (Product of Factor Levels) x (Repeats 1)

    Total DF = (Number of Individual Results) - 1

  • Source of Variation

    Between PartsBetween Operators

    Sum of Squares

    2.05870.0480

    Degreesof

    Freedom

    92

    Degrees of Freedom

    Copyright 2012 BSI. All rights reserved.

    65

    Between OperatorsOperator x PartRepeatability

    Total

    0.04800.10370.0387

    2.2491

    21830

    59

  • The Mean Square is calculated as follows:

    Mean Square = (Sum of Squares) / (Degrees of Freedom)

    Source of Variation Sum of Squares DF Mean Square

    4. Calculation of the Mean Squares

    Copyright 2012 BSI. All rights reserved.

    66

    Between PartsBetween OperatorsOperator x PartRepeatability

    Total

    2.05870.04800.10370.0387

    2.2491

    921830

    59

    0.22870.02400.00580.0013

  • Source of Variation

    Between PartsBetween OperatorsOperator x PartRepeatability

    Total

    Sum of Squares

    2.05870.04800.10370.0387

    2.2491

    DF

    92

    1830

    59

    Mean Square

    0.22870.02400.00580.0013

    F-Ratio

    39.434.144.46

    5. Calculation of the F-Ratio

    Copyright 2012 BSI. All rights reserved.

    67

    Total 2.2491 59

    The F-Ratio is used to test the significance of each source of variation. F-Ratio for Parts = (MSParts) / (MSOperator x Part)F-Ratio for Operators = (MSOperators) / (MSOperator x Part)F-Ratio for Operator x Part = (MSOperators x Parts) / (MSRepeatability)

  • Mean Square

    The Mean Square column is expected to contain the following components of variation. This expected mean square is only applicable to this current study, where we have 3 operators, 10 parts and 2 repeat measurements. For other studies, the number of the components will change. (Fortunately, Minitab can do this for us!)

    Expected Mean SquareSource

    Estimating Components of Variation

    Copyright 2012 BSI. All rights reserved.

    68

    Mean Square

    0.2287

    0.0240

    0.0058

    0.0013

    Expected Mean Square

    2ityRepeatabil

    2ityRepeatabil

    2PartOperator

    2ityRepeatabil

    2PartOperator

    2Operator

    2Repeatability

    2PartOperator

    2Part

    2

    220

    26

    +

    ++

    ++

    Source

    Parts

    Operators

    Operator x Part

    Repeatability

  • Mean Square

    0.2287

    0.0240

    0.0058

    0.0013

    Expected Mean Square

    2

    2ityRepeatabil

    2PartOperator

    2ityRepeatabil

    2PartOperator

    2Operator

    2ityRepeatabil

    2PartOperator

    2Part

    2

    220

    26

    +

    ++

    ++

    Source

    Parts

    Operators

    Operator x Part

    Repeatability

    Estimating Components of Variation

    Copyright 2012 BSI. All rights reserved.

    69

    0.0013 2 ityRepeatabilRepeatability

    00225.0

    0045.00013.00058.02

    0058.02

    0013.0

    2PartOperator

    2PartOperator

    2ityRepeatabil

    2PartOperator

    2ityRepeatabil

    =

    ==

    =+

    =

  • Estimating Components of Variation

    Mean Square

    0.2287

    0.0240

    0.0058

    Expected Mean Square

    2ityRepeatabil

    2PartOperator

    2ityRepeatabil

    2PartOperator

    2Operator

    2ityRepeatabil

    2PartOperator

    2Part

    2

    220

    26

    +

    ++

    ++

    Source

    Parts

    Operators

    Operator x Part

    Copyright 2012 BSI. All rights reserved.

    70

    00091.020

    0013.0)00225.0(2)0240.0(

    20240.020

    0240.0220

    2Operator

    2ityRepeatabil

    2PartOperator

    2Operator

    2ityRepeatabil

    2PartOperator

    2Operator

    =

    =

    =

    =++

    0.00132

    ityRepeatabilRepeatability

  • Mean Square

    0.2287

    0.0240

    0.0058

    Expected Mean Square

    2ityRepeatabil

    2PartOperator

    2ityRepeatabil

    2PartOperator

    2Operator

    2ityRepeatabil

    2PartOperator

    2Part

    2

    220

    26

    +

    ++

    ++

    Source

    Parts

    Operators

    Operator x Part

    Estimating Components of Variation

    Copyright 2012 BSI. All rights reserved.

    71

    0.00132

    ityRepeatabilRepeatability

    03715.06

    0013.0)00225.0(22287.0

    22287.06

    2287.026

    2Part

    2ityRepeatabil

    2PartOperator

    2Part

    2ityRepeatabil

    2PartOperator

    2Part

    =

    =

    =

    =++

  • 00130.0

    00225.0

    00091.003715.0

    2ityRepeatabil

    2PartOperator

    2Operator

    2Part

    =

    =

    =

    =

    Estimating Components of Variation

    Copyright 2012 BSI. All rights reserved.

    72

    04161.000130.000225.000091.003715.0

    00130.0

    2Total

    2ityRepeatabil

    2PartOperator

    2Operator

    2Part

    2Total

    ityRepeatabil

    =+++=

    +++=

    =

    We have established estimates of each of the components of variation!

  • OverallVariation0.04161 Reproducibility

    0.00316

    Part-to-PartVariation0.03715 Operator

    0.00091

    Operator

    Variance Component Estimates

    Copyright 2012 BSI. All rights reserved.

    73

    MeasurementSystemVariation0.00446

    Repeatability0.00130

    Operatorby part

    Interaction0.00225

    Variances are additive!