Measurement System

Analysis

Quality Growth Services Pvt. Ltd.

Module Objective:
 What is Measurement System
 Understanding MSA
 Detailed discussion on Measurement Errors
– Variable
 Bias
 Linearity
 R&R
– Attribute studies

1-2 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 2

1
Quality Growth Services
 INTRODUCTION

Product
Quality

Ability of
Process Control

Ability to Measure

Quality of Measurement System

So, let us understand measurement process

1-3 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 3

Measurement Process
•Standard
•Work piece (part)
•Instrument Measurement Measurement
•Person result
•Procedure
•Environment

Decision (action)
Analysis

1-4 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 4

2
Quality Growth Services
 Properties of a Good Measurement System

• Adequate discrimination(resolution)
• Under statistical control
• Accuracy
• Precision

1-5 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 5

DISCRIMINATION

Ability of measuring the

smallest difference
1,2,3...,9,10..
got it!

required
resolution

6

1-6 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 6

3
Quality Growth Services
 DISCRIMINATION

Should be small relative to

• PROCESS VARIATION or
• SPECIFICATION LIMIT.

• Rule of 10’s i.e. 1/10th should be followed as a starting point. i.e. least
count/ resolution of equipment should be 1/10th of process Variation
(10 data categories).

1-7 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 7

DATA CATEGORIES:

• The number of groups in which the measurement data (results) can

be obtained by using the measurement system.
Example:
•Process variation : 3.93 - 4.06 mm
•Equipment : Vernier Caliper (0.02 mm)
•Group of readings : 3.94, 3.96, 3.98, 4.00, 4.02, 4.04, 4.06
•Data categories :7

What if measurement system has inadequate discrimination (resolution or data category)?

1-8 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 8

4
Quality Growth Services
 Properties of a Good Measurement System

• Adequate discrimination
(resolution)
• Under statistical control
• Accuracy
• Precision

1-9 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 9

What is the meaning of “Under statistical control”?

A process showing variation only due to

common causes are said to be under
STATISTICAL CONTROL

1 - 10 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 10

5
Quality Growth Services
 Statistical Control of Measurement System

- Common cause variations only

- No special cause variation

What if measurement system is not under statistical

control?

1 - 11 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 11

Statistical Control of Measurement System

- Generally, a variation found within Mean+3 Sigma (6
sigma spread) is considered as common cause variation.
- 6 sigma spread covers 99.73 % of the process

1 - 12 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 12

6
Quality Growth Services
 Properties of a Good Measurement System

• Adequate discrimination
(resolution)
• Under statistical control
• Accuracy
• Precision

1 - 13 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 13

Accuracy & Precision

Accurate &
Accurate
Precise

What?
Precise

1 - 14 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 14

7
Quality Growth Services
 IF Measurement System has accuracy error

ACTUAL VARIATION OBS. VARIATION

DUE TO MS ERROR

IT WILL CREATE LOCATION ERROR IN RESULT OF MEASUREMENT:

1 - 15 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 15

IF Measurement System has precision error

ACTUAL VARIATION OBS. VARIATION

DUE TO MS ERROR

IT WILL CREATE SPREAD ERROR IN RESULT OF MEASUREMENT:

1 - 16 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 16

8
Quality Growth Services
 So, we conclude…
total  process   MS Accuracy error

total
2
  2process   2MS Precision error

Observed Process = Actual Process + Measurement System Error

1 - 17 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 17

EFFECT OF MEASUREMENT SYSTEM ERROR ON DECISION

EFFECT ON PRODUCT CONTROL:

1a. CALLING A GOOD PART AS BAD PART (TYPE -I ERROR)

1b. CALLING A BAD PART AS GOOD PART (TYPE -II ERROR)

So, we need ‘Control’ on Measurement Process…

1 - 18 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 18

9
Quality Growth Services
 Measurement System Analysis

• Study of effect of Measurement system on

measurement result and
• Assessing their suitability
for Product or Process Control.

Errors in Measurement System

affect decision about Products & Processes
1 - 19 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 19

Measurement System Analysis

Measurement System Errors

Location Spread

Bias Repeatability

Linearity Reproducibility

Stability

1 - 20 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 20

10
Quality Growth Services
 Bias

–Difference between the observed

Reference
average of measurements and the Value
true value (reference value)
Bias
– on the same characteristics
– on the same part.

Observed
Average
Value

Bias

1 - 21 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 21

Linearity
– The difference of bias through the expected operating
(measurement) range of the equipment.
– This is change of bias with respect to size.

1 2 3
MEASURMENT POINTS
1 - 22 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 22

11
Quality Growth Services
 Stability (Drift)

The total variation in the measurements obtained with a

measurement system-
• on the same master or parts,
• when measuring a single characteristic,
• over an extended time period.

Stability is the change of bias over time

1 - 23 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 23

Repeatability (Within System Variation)

The variation in measurements obtained

• with one measurement instrument
• when used several times
• by one appraiser
• while measuring the identical characteristic
• on the same part.

repeatability= Rtrial/d2*=K1R, where K1=1/ d2* Repeatability

Note:Repeatability is commonly referred to as equipment variation(EV), although

this is misleading. In fact repeatability is within system (SWIPPE) variation

1 - 24 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 24

12
Quality Growth Services
 Reproducibility (Between system variation)
The variation in the average of the measurements
•made by different appraisers
•using the same measuring instrument
•when measuring the identical characteristic
•on the same part.

reproducibility= Rappraiser/d2*=K2R, where K2=1/ d2*

APPRAISER A C B
REPRODUCIBILITY

1 - 25 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 25

Gage Repeatability & Reproducibility(GRR)

An estimate of the combined variation of repeatability
and reproducibility.

2GRR = 2EV + 2AV

A C B
APPRAISER

1 - 26 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 26

13
Quality Growth Services
 Section-3

DETAILED EXPLANATION

1 - 27 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 27

R&R- STUDY

Step 1: SELECTION OF PARTS n>5 parts

Key Consideration
Must cover entire process variation

Step 2 :Mark Identification

Not visible to appraiser

Minimize the effect of within part variation

Example: Identify measurement location.

1 - 28 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 28

14
Quality Growth Services
 R&R- STUDY

Appraiser (k) : 2-3 routine appraisers

Measuring equipment : Calibrated routine equipment

Number of trials (r) : 2-3

Data collection:
•Using data collection sheet
•Under ROUTINE measurement condition
•in random order
•using Blind Measurement process

1 - 29 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 29

R&R- DATA COLLECTION: Xbar-R method

TRI
AL
OPER NO. PART AVERAGE

1 2 3 4 5 6 7 8 9 10

RAKESH 1 48.06 48.055 48.054 48.065 48.064 48.056 48.063 48.064 48.065 48.066 1

2 48.061 48.056 48.055 48.065 48.063 48.055 48.06 48.066 48.062 48.063 2

3 3

AVERAGE X bar a 4

RANGE R barc 5

Sharifi 1 48.06 48.057 48.053 48.065 48.052 48.055 48.063 48.064 48.065 48.063 6

2 48.06 48.056 48.056 48.065 48.053 48.054 48.06 48.065 48.065 48.063 7

3 8

AVERAGE X bar b 9

RANGE R barc 10

1 - 30 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 30

15
Quality Growth Services
 R&R- GRAPHICAL ANALYSIS (Manual)
1) For appraiser A, calculate Average (X bar) & Range(R) for each part and enter in
rows 4 & 5 respectively.
2) Do the same for appraisers B & C and enter results in rows 9, 10 and14, 15
respectively.
3) For appraiser A, calculate average (X bar a) of all the averages (row 4) and
average (R bara) of all the ranges (row 5) and enter in data sheet.
4) Calculate X bar b, R barb & X bar c, R barc for appraisers B & C also and enter the
results in data sheet.
5) Calculate average of all the observations (rows 4, 9 &14) of each part and enter
result in row 16.
6) Calculate Part range (Rp)= Difference of Max. and Min. of row 16 and enter in
data sheet (right most row 16).
7) Calculate X double bar =(X bar )/k and enter in data sheet (right most row 16).

1 - 31 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 31

Calculate
Calculate R&R- CALCULATION Appraiser
Average & Average &
Range Range

1 - 32 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 32

16
Quality Growth Services
Calculate Part R&R- CALCULATION
Average &
Range

1 - 33 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 33

R&R- Calculation (Manual)

– R double bar = (Rbar)/k.

– X bardiff = Max (Xbar) - Min (X bar)
– UCL Rbar = D4 R double bar (D4=3.27 for 2 trials & 2.58 for 3)
– LCL Rbar = D3 R double bar (D3= 0 for trials<7)
– UCL x bar = X double bar +A2 R double bar
(A2=1.88 for 2 trials & 1.02 for 3 trials).
– LCL x bar = X double bar -A2 R double bar

1 - 34 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 34

17
Quality Growth Services
 R&R- CALCULATION

Calculate
Calculate Calculate R
X bar diff
UCL Rbar double bar

Calculate
LCL x bar

35
1 - 35 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 35

R&R- GRAPHICAL ANALYSIS (Manual)

RANGE CHARTS

1 - 36 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 36

18
Quality Growth Services
 R&R- GRAPHICAL ANALYSIS: RANGE CHARTS

– Used to determine whether the process is in statistical control.

– The special causes need to be identified and removed
– Plot all the ranges for each part & each appraiser on range chart
– If all ranges are under control, all appraisers are doing the same job.
– If one appraiser is out of control, the method used differs from the
others.
– Repeat any readings that produced a range greater than the
calculated uclr using the same appraiser and part as originally used.
– Or, discard those values and re-average and re-compute R double
bar and the limiting value UCLR based upon the revised sample
size.
– Correct the special cause that produced the out of control condition.

1 - 37 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 37

R&R- GRAPHICAL ANALYSIS: AVERAGE CHARTS

1 - 38 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 38

19
Quality Growth Services
 R&R- ANALYSIS – NUMERICAL (Manual)
Calculate the following and record in report sheet
– Repeatability (EV) = Rdouble bar x K1
K1 = .8862 (2 trials), 0.5908(3 trials)

– Reproducibility (AV) =
X Bar Diff x K2
K2=0 .7071(2 app.) & 0.5231(3 app.)

– Repeatability & Reproducibility (GRR) = (EV)2 + (AV)2

– Part to part variation (PV) = RP x K3

– Total variation (TV) = (GRR)2 + (PV)2

1 - 39 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 39

R&R- CALCULATION

40
1 - 40 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 40

20
Quality Growth Services
 R&R- CALCULATION
– Calculate % variation and ndc as follows

%EV = 100 [EV/TV]

%AV = 100 [AV/TV]

%GRR = 100 [GRR/TV]

%PV = 100 [PV/TV]

No. of distinct categories (ndc)= 1.41(PV/GRR)

1 - 41 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 41

R&R- CALCULATION

42
1 - 42 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 42

21
Quality Growth Services
 TV Calculation- Different Approaches

Priority order:
1. PV approach
2. Surrogate process variation approach
3. Specification tolerance approach

1 - 43 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 43

TV Calculation- PV Approach

PV is calculated from Rp if parts

represent entire process variation

1 - 44 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 44

22
Quality Growth Services
 TV Calculation - Surrogate Process variation Approach

Process variation is historical process variation

from a stable and statistically controlled process

1 - 45 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 45

TV Calculation - Tolerance Approach

1 - 46 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 46

23
Quality Growth Services
 R&R- DECISIONS

GRR < 10% : Considered to be acceptable

10% < GRR < 30% : May be acceptable for some applications
 Should be approved by the Customer
GRR > 30% : Considered to be unacceptable

ndc >= 5 : Acceptable

Note: When looking at GRR and measurement variation it is important to
look at each application individually, to see what is required and how the
measurement is going to be used.

1 - 47 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 47

NDC Graphically:

GRR GRR GRR GRR

1 2 3 4 5 6 7 8 9 10 11

PART VARIATION

1 - 48 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 48

24
Quality Growth Services
 No. of Distinct Categories (NDC)

 AIAG suggests that when the number of categories is less

than 2, the measurement system is of no value for
controlling the process, since one part cannot be
distinguished from another. When the number of
categories is 2, the data can be divided into two groups,
say high and low. When the number of categories is 3, the
data can be divided into 3 groups, say low, middle and
high. A value of 5 or more denotes an acceptable
measurement system

1 - 49 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 49

R&R- DECISIONS

• When beginning to evaluate an organization’s

measurement system, it can be useful to set priorities on
which measurement systems to initially focus.
• When SPC is used for process control and indicates under
control and acceptable, the measurement system can be
considered acceptable for this use.
• If out of control situation is seen, MS should be evaluated
first.
• In general, when an apparent measurement issue exists,
assembly and set-up instruction for the gage to be reviewed
and verified.

1 - 50 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 50

25
Quality Growth Services
 MSA FOR NON -REPLICABLE Measurement Systems

Non Replicable Measurement System:

 Destructive Measurement System
 When the part being measured is destroyed by the
act of measuring

 Systems where the part changes on use/test

 When the part, itself, is not harmed by the
measurement process but the characteristic being
measured will change

1 - 51 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 51

MSA FOR NON -REPLICABLE MS

GRR STUDY
Sampling:
• Select 6 very similar parts (assumed to be identical) consecutively
or if possible split 1 sample in 6 parts
• Consider these 6 duplicate samples as sample 1
• Mark them 1-1, 1-2,……….1-6
• In similar manner select 10 samples (i.e 10 x 6= 60 parts as-
1-1, 1-2,…………….1-6
2-1, 2-2,……………..2-6
……………………………
10-1, 10-2, …………10-6

1 - 52 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 52

26
Quality Growth Services
 GRR FOR COMPLEX & NON REPEATABLE MS

Sampling: Key Considerations

a) All 1 to 6 parts (duplicate parts) should as much alike as

possible
b) However, there must be difference between one group of
samples and another group of samples.
c) That is duplicate parts within a group should be identical,
but between group of parts there should be difference.
d) The total combination should represent entire process
variation

1 - 53 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 53

GRR FOR COMPLEX & NON REPEATABLE MS

SAMPLING EXAMPLE:

1 - 54 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 54

27
Quality Growth Services
 GRR FOR COMPLEX & NON REPEATABLE MS

Data collection
Select the following:
• Routine appraiser (2-3),
• Routine measuring equipment
• Routine measurement venue
• Routine measurement environment

1 - 55 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 55

GRR FOR COMPLEX & NON REPEATABLE MS

Data collection (all preconditions apply):

• Use blind measurement process
 Parts in random order
 Appraiser in random order
 Part number should not be known to appraiser
 Appraiser should not be shown his previous observation
 Appraiser should not know other appraisers observations

1 - 56 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 56

28
Quality Growth Services
 GRR FOR COMPLEX & NON REPEATABLE MS
DATA COLLECTION EXAMPLE

1 - 57 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 57

MSA FOR COMPLEX & NON REPEATABLE MS

1 - 58 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 58

29
Quality Growth Services
 MSA FOR COMPLEX & NON REPEATABLE MS

1 - 59 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 59

MSA FOR COMPLEX & NON REPEATABLE MS

1 - 60 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 60

30
Quality Growth Services
 Recall: Bias

–Difference between the observed

Reference
average of measurements and the Value
true value (reference value)
Bias
– on the same characteristics
– on the same part.

Observed
Average
Value

Bias

1 - 61 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 61

DETERMINING BIAS
1. Obtain sample and determine reference value
2. Collect data
3. Determine Bias
4. Plot bias histogram
5. Compute Average bias
6. Compute Repeatability Standard deviation
7. Determine acceptability of repeatability
8. Determine t statistic for bias
9. Compute bias Confidence Interval and decision
making

1 - 62 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 62

31
Quality Growth Services
 Step 1: Obtain REFERENCE SAMPLE SELECTION

Key Consideration : Should be sufficiently

stable during study

Priority order
1. Sample piece else
2. Production part else
3. Similar other component else
4. Metrology standard

1 - 63 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 63

Step 1: DETERMINING REFERENCE VALUE

Minimize the effect of within part variation

Example: Identify measurement location.

Measure the part for n10 times

•In standard room/ tool room
•With a measurement equipment of better accuracy.
•Using standard measurement method

Reference Value = Average of measured value

1 - 64 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 64

32
Quality Growth Services
 Step- 2. DATA COLLECTION
Under routine measurement condition

For n > 10 times

Trials True Value  Obs Value
1 6.00 5.8
2 6.00 5.7
3 6.00 5.9
4 6.00 5.9
5 6.00 6.0
6 6.00 6.1
7 6.00 6.0
8 6.00 6.1
9 6.00 6.4
10 6.00 6.3
11 6.00 6.0
12 6.00 6.1
13 6.00 6.2
14 6.00 5.6
15 6.00 6.0

1 - 65 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 65

Step-3
 Determine Bias for each reading:
 biasi=xi- Reference Value

Trials True Value  Obs Value Bias

1 6.00 5.8 ‐0.2
2 6.00 5.7 ‐0.3
3 6.00 5.9 ‐0.1
4 6.00 5.9 ‐0.1
5 6.00 6.0 0.0
6 6.00 6.1 0.1
7 6.00 6.0 0.0
8 6.00 6.1 0.1
9 6.00 6.4 0.4
10 6.00 6.3 0.3
11 6.00 6.0 0.0
12 6.00 6.1 0.1
13 6.00 6.2 0.2
14 6.00 5.6 ‐0.4
15 6.00 6.0 0.0

1 - 66 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 66

33
Quality Growth Services
 Step 4. Graphical Analysis
Plot the bias as a histogram

Analyse if any special cause present.

If yes, identify & remove the cause and recollect data & re-analyse.
If not, proceed for numerical analysis.

1 - 67 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 67

Step 5. Compute Average BIAS

Trials True Value  Obs Value Bias
1 6.00 5.8 ‐0.2
2 6.00 5.7 ‐0.3
3 6.00 5.9 ‐0.1
4 6.00 5.9 ‐0.1
5 6.00 6.0 0.0
6 6.00 6.1 0.1
7 6.00 6.0 0.0
8 6.00 6.1 0.1
9 6.00 6.4 0.4
10 6.00 6.3 0.3
11 6.00 6.0 0.0
12 6.00 6.1 0.1
13 6.00 6.2 0.2
14 6.00 5.6 ‐0.4
15 6.00 6.0 0.0

Sum bias 0.1000
Avg bias 0.0067

1 - 68 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 68

34
Quality Growth Services
 Compute Repeatability Standard Deviation

Obs Value
Trials True Value  (x) x‐ Avg(x) Sq[x‐ Avg(x)]
1 6.00 5.8 ‐0.2067 0.0427
2 6.00 5.7 ‐0.3067 0.0940
To calculate repeatability 3 6.00 5.9 ‐0.1067 0.0114
(EV) you can also use R 4 6.00 5.9 ‐0.1067 0.0114
Bar/d2, but when sample 5 6.00 6.0 ‐0.0067 0.0000
6 6.00 6.1 0.0933 0.0087
size large (number of
7 6.00 6.0 ‐0.0067 0.0000
reading are large), Root
8 6.00 6.1 0.0933 0.0087
Mean Square method is 9 6.00 6.4 0.3933 0.1547
better. 10 6.00 6.3 0.2933 0.0860
11 6.00 6.0 ‐0.0067 0.0000
12 6.00 6.1 0.0933 0.0087
13 6.00 6.2 0.1933 0.0374
14 6.00 5.6 ‐0.4067 0.1654
15 6.00 6.0 ‐0.0067 0.0000
Sum (x) 90.1000 Sum= 0.6293
Avg (x) 6.0067 Sigma‐r 0.2120
1 - 69 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 69

7. Determine Acceptability of Repeatability

Where, TV= Process Standard Deviation

1 - 70 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 70

35
Quality Growth Services
 8. Determine Bias standard error

1 - 71 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 71

9. DETERMINE CONFIDENCE LIMIT

Alpha‐ T Value
– Lower limit (L) = bias – t b two‐trails α=0.05
Sample  DF
– Upper limit (U) = bias + t b Size
2 1 12.71
3 2 4.303
4 3 3.182
•t can be obtained from Table 5 4 2.776
•  (preferably 0.05) is a measure of confidence 6 5 2.571
7 6 2.447
8 7 2.365
9 8 2.306
10 9 2.262
– Lower limit (L) = .0067– 2.145* 0.0547=- 11
12
10
11
2.228
2.201
0.1106 13 12 2.179
14 13 2.16
– Upper limit (U) = 0067+ 2.145* 0.0547 15 14 2.145
=0.1240 16 15 2.131
17 16 2.12
18 17 2.11
19 18 2.101
20 T19 2.093

1 - 72 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 72

36
Quality Growth Services
 Decision Making

Bias is acceptable
at 100(1-)% confidence level
if
L<0<U

1 - 73 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 73

43.97
BIAS: Exercise 43.96
43.92
43.99
43.95
43.96

Ref. Value: 43.99 43.93

43.96
43.94
43.95
TV: 0.3389 43.93
43.98
43.95
43.94
43.91
43.97
43.96
43.94
43.92

1 - 74
44
© Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 74

37
Quality Growth Services
 Linearity
– The difference of bias through the expected operating
(measurement) range of the equipment.
– This is change of bias with respect to size.

1 2 3
MEASURMENT POINTS
1 - 75 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 75

Linearity

NO LINEARITY ERROR CONSTANT LINEARITY NON LINEAR

1
BIAS

-1

REFERENCE VALUE

1 - 76 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 76

38
Quality Growth Services
 LINEARITY STEPS
Determine Process Range

Select Reference Sample

Determine Ref. Value

Calculate Bias

Check Linear Relation

Draw best fit

Draw Confidence band

Determine Repeatability Error

Take Decision

1 - 77 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 77

DETERMINING LINEARITY

Determine operating range

due to process variation

Select g > 5 parts covering the operating range and

determine their reference value (x)

• Determine BIAS values for each

1 - 78 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 78

39
Quality Growth Services
 DETERMINING LINEARITY

1 - 79 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 79

Excel Results
– Statistics:
For the best fit line,
yi = axi + b Where,
xi = reference value
yi = bias average

xy – 1  x  y
gm
a = = slope
x2 – 1 ( x)2
gm

b = y – ax = intercept

1 - 80 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 80

40
Quality Growth Services
 Determine Confidence band for the Line: Complex
Calculation.
Use Excel
Work sheet
1/2
1 (x0 – x)2
Lower: b + ax0 – tgm-2,/2 +
gm s
(xi – x)2

1/2
1 (x – x)2
Upper: b + ax0 + tgm-2,/2 + 0 s
gm (xi – x)2

Where, x0 = Reference Value

s= y i – b yi – a xiyi
2

gm-2
81
1 - 81 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 81

Plot Lower & Upper band and the “bias = 0” line

82
1 - 82 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 82

41
Quality Growth Services
 Linearity Acceptability

when “bias = 0” lie line

entirely within the confidence bands of the fitted line.

83
1 - 83 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 83

Linearity Decision Criteria

line

Linearity acceptable if, “bias = 0” line lie

entirely within the confidence bands of the
fitted

1 - 84 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 84

42
Quality Growth Services
 Stability (Drift)

The total variation in the measurements obtained with a

measurement system-
• on the same master or parts,
• when measuring a single characteristic,
• over an extended time period.

Stability is the change of bias over time

1 - 85 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 85

DETERMINING STABILITY

Selection of reference standard: Refer bias

study.
Establish reference value :Refer bias study.
Data collection:
 Decide subgroup size
 Decide subgroup frequency
 Collect data for 20-25 subgroups

1 - 86 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 86

43
Quality Growth Services
 DETERMINING STABILITY

Analysis
Calculate control limits for X-R chart.
Plot data on chart
Analyse for any out of control situation.

Decision
Measurement system is stable & acceptable if
no out of control condition is observed other
wise not stable and needs improvement .

Remember these are same you draw for process control (SPC)

1 - 87 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 87

ATTRIBUTE MEASUREMENT SYSTEMS STUDY

Attribute measurement systems:

– Class of measurement systems where the measurement value is one of
a finite number of categories.
– The most common of these is a go/nogo gage
– There is a quantifiable risk when using any attribute measurement
systems
– The largest risk is at the category boundaries

LSL USL
LSL USL

MEASUREMENT SYSTEM GRAY AREAS

1 - 88 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 88

44
Quality Growth Services
 ATTRIBUTE MEASUREMENT SYSTEMS STUDY

Select sufficient samples

– Approximately 25% close to lower specification (conforming and
non-conforming

– Approximately 25% close to upper specification (conforming and

non-conforming

– Remaining both conforming and non conforming

– Note down the correct measurement attribute (true status).

– Decide the no. Of appraiser & no. Of trials.

– Record the measurement result in data sheet

1 - 89 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 89

ATTRIBUTE MEASUREMENT SYSTEMS STUDY

Data Collection:

1 - 90 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 90

45
Quality Growth Services
 ATTRIBUTE MEASUREMENT SYSTEMS
STUDY(Effectiveness Hypothesis based)
This method will enable you compare agreement within the appraiser, between the
appraiser and references as follows:
% Appraiser1 %Score vs Attribute2
Operator Operator Operator
Source #1 #2 #3 Operator #1 Operator #2 Operator #3

Total Inspected 20 20 20 20
# Matched 10 16 10 16
False Negative (operator biased toward
rejection) 0 0 0
False Positive (operator biased toward acceptance) 0 0 0
Mixed 10 4 0
95% UCL
Calculated Score 50.0% 80.0% 50.0% 80.0%
95% LCL
Screen % Effective Score3 Screen % Effective Score vs Attribute4
Total Inspected 20 20
# in Agreement 10 10
95% UCL
Calculated Score 50.0% 50.0%
95% LCL

91
1 - 91 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 91

ATTRIBUTE MEASUREMENT SYSTEMS STUDY

(PROBABILITY BASED)
Decision by Appraisers
–Correct decisions
– Calling good part as good : Good-Correct (GC)
– Calling bad part as bad : Bad-Correct (BC)
–Wrong decisions
– Calling good part as bad : False Alarm (FA)
– Calling bad part as good : Miss (M)
–For each appraiser count the data
– Total good-correct (GC)
– Total bad -correct (BC)
– Total correct (TC): total (GC+BC)
– Total false alarm (FA)
– Total miss (M)

1 - 92 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 92

46
Quality Growth Services
 ATTRIBUTE MEASUREMENT SYSTEMS STUDY

Decision Making

For each appraiser calculate :

– Effectiveness (E) = TC/Total decision
– Probability of false alarm (Pfa) = FA/ Total opportunity for FA
– Probability of miss (Pm) = M/ Total opportunity M
–Decision
Parameter acceptable marginal unacceptable
E >.90 .80 to .90 <.80
Pfa <.05 .05 to .10 >.10
Pm <.02 .02 to .05 >.05

1 - 93 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 93

Attribute Measurement Systems Study

Kappa Method
Compare agreement between the appraisers
using cross tabulation :

No. 30 in the first column first Row of above table has been calculated as under:
“when Operator A declared “1”i.e. Bad , Operator B also declared it as “1”i.e. Bad
Similarly, 4 in Second column first Row of above table has been calculated as under:
when Operator A declared “1”i.e. Bad , Operator B declared it as “2” i.e. Good

1 - 94 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 94

47
Quality Growth Services
 Attribute Measurement Systems Study

Kappa Method
Compare agreement between the appraisers
using cross tabulation :

No. 30 in the first column first Row of above table has been calculated as under:
“when Operator A declared “1”i.e. Bad , Operator B also declared it as “1”i.e. Bad
Similarly, 5 in Second column first Row of above table has been calculated as under:
when Operator A declared “1”i.e. Bad , Operator B declared it as “2” i.e. Good

1 - 95 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 95

Attribute Measurement Systems Study

ANALYSIS No. 1: (Hypothesis Based)
Compare agreement between the appraisers using cross
tabulation & kappa as follows
A*B Cross Tabulation
B
Total
1 2
1
Count O11 O21 R1
Expected Count E11 E21
A
2 Count O12 O22 R2
Expected Count E12 E22
Count C1 C2 Total
Total
Expected Count
• ORow,Column -- Observation  real data
• PRi -- Probability of being in Row i = Ri/Total
• PCi -- Probability of being in Column i = Ci/Total
• ERow,Column – Expected Value  Values expected if the two variables are independent

Expected count () = (row total x column total)/ grand total

QGS\MSA
1 - 96 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 96

48
Quality Growth Services
 Attribute Measurement Systems Study
po-pe
•Calculate kappa : Kappa =
1-pe
(52/60)- (30.5/60)
Kappa = = 0.727
1- 30.5/60

Where,
po = sum of observed proportion in diagonal cells
pe = sum of expected proportion in diagonal cells

1 - 97 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 97

Attribute Measurement Systems Study

Can also be used for comparing agreement
between the appraiser and references

1 - 98 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 98

49
Quality Growth Services
 Attribute Measurement Systems Study

Calculate appraiser-references kappa :

1 - 99 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 99

QUESTION – ANSWER

1 - 100 © Quality Growth Services Pvt Ltd. Unpublished proprietary work available only under license. All rights reserved. 100

50
Quality Growth Services