Measurement System Analysis

Measurement System Analysis
Welcome to the seminar

On
Measurement System Analysis -
4th Edition
By
Vinay Kumbhar
Vinay S K 1
INTRODUCTION OF FACULTY
Name : Vinay S. K.
More than 10 years Experience in Automotive Industry,
Home appliances and Quality Management systems.
Areas of Work : ISO 9001, TS 16949, ISO 17025 (NABL), and
Core tools
(Manufacturing Industry, Calibration Lab, Consultancy
Service
Email – vinayskumbhar@gmail.com
Vinay S K 2
1
Please keep your

mobiles Switched off /
on Silent Mode…
Vinay S K 3
OBJECTIVE
To provide
 Basic concept of measurement system
 Detail application methodology of all statistical

studies for variable and attribute data
 Methods to meet ISO/TS-16949 requirements
Vinay S K 4
2
Today’s Agenda
 Brief Changes and overview of 4th edition.
 Objectives
 Introduction
 ISO/TS 16949 Requirements
 Measurement System Analysis
 Selection of Instrument ( Discrimination )
 Statistical Properties of Measurement System
 Bias , Linearity, Stability & Gauge Repeatability & Reproducibility
 Graphical Analysis
 Attribute data Study & Conclusion
Vinay S K 5
Brief Changes in 4th edition

 Calibration and relationship to external systems ;e.g., ISO 17025.
 Analysis of Results
 The use of GRR guidelines as threshold criteria alone is not an

acceptable practice for determining the acceptability of a
Measurement system.
 Individual graphical analysis recommended.
 Application of Measurement system needs to be considered to have

lower GRR.
 Bias – calculation methods changed.
 Chapter no 4 and 5 –clubbed and made one chapter no 4

Vinay S K 6
3
Over View of Changes in 4th edition

 Calibration System
A Calibration System consists of both the documented and

technical processes that assure measurement traceability is
disseminated to the “shop floor”.
– Critical for assuring accuracy, precision and reliability of

measurements
– Calibration Events provide the links or traceability to higher

level standards
Vinay S K 7
Vinay S K 8
4
Vinay S K 9
GRR Analysis And Target Pp

Pp (or Ppk) target value
– use when sufficient samples to represent the process are not available and
an existing process with similar process variation is not available or the
new process is expected to have less variability than an existing process
• Use the following TV in the GRR analysis:
Vinay S K 10
5
APQP And MSA

PLAN AND DEFINE
In the Plan and Define Program stage the team is to identify preliminary special
product and process characteristics.
The organization’s product quality planning team should build on this listing and
reach consensus through the evaluation of the technical information.
The technical information should come from:
– the drawings
– industry standards such as SAE
– Engineering specifications
This is to be documented on the appropriate Control Plan
Vinay S K 11
 Gages/testing equipment requirements should also

be identified at this time.
 The organization’s product quality planning team

should add these requirements to the Timing Chart.
 Progress should be monitored to assure that required

timing is Met
Vinay S K 12
6
After gauge Test Equipment identified

The team must assess that proposed gauge /test system design:
– Understand which traceable standard are appropriate for the required testing
– The required frequency that the standards are to be used to assess the accuracy
of the measurement system
– Cost of the testing
– An estimate of the proposed system accuracy, precision, repeatability,

reproducibility, etc based on similar parts and similar measurement systems.
– The time required for the testing
– The frequency of the testing
This is a joint effort between design, quality and the supplier
Vinay S K 13
 The characteristics to be measured should be agreed to as early in the

program as possible. This should help prevent late gauging changes.
 The gauging methods should be established up front.
 This should also help prevent late gauging changes.
 The gauging must be used on the prototype build. An early gage study can be
performed. While the full range of process variation will probable not be
represented, as preliminary study can be performed.
 The gage can be verified at the prototype stage to ensure that the required
features are being correctly measured. This will help eliminate late changes to
the gauging method.
 Before using a gage to measure the product, the gage
must have acceptable bias and linearity, and repeatability
Vinay S K 14
7
Introduction
A decision to adjust a manufacturing process not is commonly based on
measurement data. Measurement data are also used in analytical studies to know
the relation between two or more variables, which in turn will lead to a better
understanding of process.
Benefit of using measurement data will depend on the Quality of data. Much of the
variations in a set of measurements is due to the interaction between the
measurement system and its environment .If the interaction generates too much of
the variations then the quality of the data may be so low that the data are not
useful.
If the Quality of the data is not acceptable, then it must be improved. This is usually
accomplished by improving the measurement system, rather than by improving the
data themselves.
Vinay S K 15
Effects of Measurement Systems
Effect on Product Decisions :
After measuring the part one of the action that can be taken to determine
the status of the part.
wrong decisions will be some times made
Eg,, some times good parts will be called as bad parts – Producers risk –
Eg
false alarm
Some times bad parts will be called as good part – Customer’s risk -
Miss rate.
Vinay S K 16
8
Introduction…
Variation :
It is the difference between two things or an entity .
Measurement System Variation :
Variation present in the measurement system affects the process results
Total variation :
It is the total variation observed in the process outputs
A) There can be variation due to Process
B) There can be variation due to Measurement system
Vinay S K 17
Introduction
Variation due
to Interaction
Total
Variation Appraiser
Variation
Equipment
Part to MSV
Within Part Variation
Part
Variation Variation
Always Measurement System variation must be
smaller than manufacturing process variation
Vinay S K 18
9
Selection of Instruments
Based on its Discrimination / Resolution
Discrimination : Capability of an instrument to indicate even

small changes of the measured characteristic.
Discrimination Decision Rules :
 Resolution of 1/10th of tolerance or process spread
 Study gage discrimination during APQP and test before PPAP
 For Continuous Improvement , 1/10th of tolerance may be

inadequate , MSA recommends 1/10th of six sigma (Total
manufacturing standard deviation).
Vinay S K 19
Statistical Properties of Measurement Systems
Accuracy : Absolute correctness of measurements compared with a

known standard. It is quantified by evaluating bias.
Precision: Ability to the measurement system to reproduce or

duplicate readings. It is quantified by evaluating repeatability as
well as reproducibility.
Refer Illustration 1 :
Measurement System Variation can be characterized by
1. Location : Stability, Bias, & Linearity
2. Width or Spread : Repeatability & Reproducibility
Vinay S K 20
10
BIAS – WHAT IT IS ?
Ref.
Value BIAS= X bar - Ref.Value
X bar
Vinay S K 21
BIAS - STEPS
1. Capture reference value
• Use masters calibrated- traceable to
standards
• Measure by better instrument than the
instrument being measured
• By Layout inspection
2. Measure the part/ master by Min. 10 times
Vinay S K 22
11
BIAS - STEPS
3. Calculate BIAS for individual reading
4. Calculate ,  bias, t limits
APPENDIX C t distribution table
Vinay S K 23
ACCEPTANCE CRITERIA
Zero must lie between t Limits of

Bias distribution or
Bias should be within acceptable gage
error as defined in Calibration Proc.
Vinay S K 24
12
Bias :
 Difference between observed average of measurements
and reference value
 Reference value also known as accepted reference value

or master value, is a value that serves as an agreed upon
reference for the measured values
 A reference value can be determined by averaging

several measurements with a higher level of measuring
equipment
Vinay S K 25
BIAS DATA SHEET

Part No.:- Date:-
Part Name:- Characteristics:
Instrument Used for Ref:-
Instrument under study:
Trial Reference value Observed Value
Reference value = Observed Value =
Bias =
Vinay S K 26
13
 BIAS CASE STUDY
 The inside diameter of a coupling is measured fifteen

times repeatedly by the same operator. The data is
given below.
 The reference value is 6.00.

 Bias = Observed Average – Reference Value
 OD Values observed:
5.8, 5.7, 5.9, 5.9, 6.0,
6.1, 6.0, 6.1, 6.4, 6.3,
6.0, 6.1, 6.2, 5.6, 6.0
Vinay S K 28
BIAS CASE STUDY
Ref. Value = 6.00 mm

Number of samples, n = 15
Average = 6.0067 mm
Bias = 6.0067 – 6.0000 = 0.0067 mm
r= Rbar / d2* = (6.4-5.6)/3.55333 = 0.22514
d2* from Appendix C for 15 trials
bias =  r / Sq. root of n
0.22514 / sq. root of 15
0.22514 / 3.873 = 0.05813
t-Static= bias / bias
0.0067 / 0.05813 = 0.1217
Upper & Lower Limits of [ bias +&– [bias x t(v,1-L/2]
t-static@ 95% Confidence = v, the degree of freedom, = n-1 = 15-1 =14
for confidence level of 95% & L= 0.05, factor for C/L =(1-L/2)
= (1-0.05/2) = 0.975
from t-static table for v=14 & 95%confidence level value t
value for bias is 2.51
[0.0067+&-{ 0.05813 * 2.51}
EV %= 100(EV/TV)= EV=  r , TV = process variation or tol/6
t-static Upper limit = +0.1526

t-static Lower limit = -0.1395
CONCLUSION: As observed t-static is between upper & lower limits our Bias Study is acceptable.
However observed t-static is not exactly in the middle or very near to mid value some action has to
be initiated before next study is due.
Vinay S K 29
14
 CASE STUDY ON BIAS
Vinay S K 30
LINEARITY
Difference in Bias over the Operating range

of the Gage
+ve Bias
- ve Bias
X bar X bar 31
Vinay S K
15
Linearity - Steps
1. Collect 5 parts / masters over the

operating range of the gage through process
variation
2. Capture reference value for the 5 parts
3. Measure at least 12 times in a random

manner to cover all 5 parts
Vinay S K 32
Linearity – Steps
4. Calculate Average, BIAS and enter the data

in a Spread sheet
5. Draw Best fit line for the average, Lower

limit and Higher limit of Bias.
Vinay S K 33
16
Linearity – Acceptance criteria

ZERO LINE MUST BE CONTAINED WITHIN THE
TWO LIMIT LINES OR WITHIN ACC.CRITERIA OF
CAL ERROR
+ VE Upper t limits
0
Lower t limits
- VE
Is this MS acceptable ?
Vinay S K 34
LINEARITY STUDY – DATA SHEET
Date:
Instrument under study:-
Part  1 2 3 4 5
Ref val 
1
2
T 3
R 4
I
A 5
L 6
S
7
8
9
10
11
12
Observed Avg
Bias
Vinay S K 35
17
LINEARITY STUDY – Example
Vinay S K 36
Stability study – What?

Variation of Bias w.r.t time when measured over a
Same Master
T1
T2
T3
Vinay S K 37
18
Stability :
Stability (or drift) is 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.
Procedure for Stability study:

 Decide the part and characteristics for the study.
 Identify the location of measurement on the part.
 Make this part as a master part and ensure that the part is preserved till the study is
completed.
 Decide the subgroup size (3 to 5), no. Of subgroup (20) & frequency (every day) of data
collection.
 As per the frequency, measure the master part 3 to 5 times and record the details in
the Xbar-R chart.
 After completion of data collection, calculate each subgroup avg and range.
Vinay S K 38
Procedure for Stability study ( Contd..) :
 Calculate the upper control limit & lower control limit for range ( R ).
 Calculate the upper control limit & lower control limit for X double bar.
 Calculate the standard deviation (Sigma)
 Calculate the Sigma X bar.
 Calculate the 't‘ value.
 After calculating the 't' value compared that value with 't' distribution
table value.
 If 't' calculated value is less than or equal to 't' table value

@95% confidence level. It means variation is accepted.
Vinay S K 39
19
 Stability -example
Vinay S K 40
Repeatability (equipment variation)

Repeatability is 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.
Reproducibility (appraiser variation)

Reproducibility is the variation in the average of the
measurements made by different appraisers using the same
measuring instruments when measuring the identical
characteristic on the same part.
Gage Repeatability and Reproducibility study

Should be carried out for all measurement systems referred in
control plan.
Vinay S K 41
20
Procedure for Gage R&R study:
 Identify the measurement system for which R&R study to be conducted.
 Decide the part and the characteristics to be measured.
 Select the instrument and ensure the instrument is calibrated and is within
the acceptance criteria.
 Collect 5 to 10 parts randomly from a lot of production. Ensure the parts

collected are representative of the total process variation.
 Identify the location on the part where the measurement has to be done
repeatedly.
 Number each part.
 Choose 2 or 3 operators and have each operator measure all parts 2 or 3

times. Enter these results in the data sheet.
 Ensure that the readings are not shown to appraisers.
 Calculate AV, EV, PV, R&R & TV as per the data sheet and calculation sheet.
Vinay S K 42
Procedure for Gage R & R study - ( Contd….)

 Interpret the results based on the acceptance criteria given below :
R&R % below 10% - M.S. Is acceptable.

R&R % between 10 & 30% - M.S. Can be acceptable under
Concession, based on the cost of
repair, importance on product
quality,etc..
R&R% greater than 30% - Not acceptable. M.S. Needs
Improvement.
NDC =/> 5
Notes:
If EV is greater than AV – Act on instrument & method
If AV is greater than EV – Act on appraisers method of
checking, training, make gage design
more sturdy, etc..
Refer Annexure 4 for the Data Sheet & Analysis Sheet
Vinay S K 43
21
Graphical Analysis :
It is used to analyze the results of a measurement system. It provides
insight in to the variation model and inter- relationship.
-Average Chart
Since the group of parts used in the study represents the process
variation. Approximately one half or more of the averages should fall
outside the control limits.
- If data shows this pattern then the measurement system should be

adequate to detect part to part variation and the measurement system
can provide useful information to analyzing and controlling the process.
- If less than half falls outside the control limits then measurement
system lacks adequate effective resolution or the sample does not
represent the expected process variation.
Vinay S K 44
Graphical Analysis
-Range Chart
ranges of the multiple readings by each appraiser on each part

is plotted on range control chart.
if all ranges are within control all appraisers are doing the same
job.
if one appraiser is out of control limit , the method used by

appraiser differs from other.
If all Appraisers have some out of control ranges, Measurement

system is inadequate as appraiser methods –techniques needs
improvement.
Vinay S K 45
22
Attribute data Measurement
An Attribute Gage :
 Compares each part to a specific set of limits and accepts the
part if the limits are satisfied.( It detects the Nonconforming
parts)
 Is designed to accept / reject a set of master parts
 Cannot indicate how good or bad a part is, only whether the
part is accepted or rejected ( Pass/Fail )
Vinay S K 46
Attribute Gage Study :

Miss: Part is bad. Decision given is good.
False Alarm: Part is good. Decision given is bad.
Procedure for long method:
• Decide the no. Parts to be selected, no. Of appraisers & no. Of trials
for conducting the study.
• Identify all parts by a unique number and inspect the part to decide
the master value. This has to be approved by a expert.
• Have each appraiser measure the parts randomly for the no. Of trials
decided.
• Record the results as “good / bad” or “go/no go” or “accept / reject”,
etc..
Vinay S K 47
23
Procedure ( Contd…)
• After completion of collecting the data calculate the probability of
miss and probability of false alarm.
• Based on the probability of miss and false alarm, obtain the bias (miss)
and bias (false alarm) from the table.
• Calculate the effectiveness and bias factor as per the calculation
sheet.
• Interpret the effectiveness results as defined below.
Eff is > 0.9 - M.S is effective.
Eff is betwn 0.7 – 0.9 - M.S is acceptable under Concession
based on the importance, cost of
repair, etc..
Eff is < 0.7 - M.S is not effective.
Vinay S K 48
Data collection sheet
SL. Master
Appraiser and trails
NO status
I II III IV V
1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
Vinay S K 49
24
Data evaluation sheet

Sl.No. Parameters Appraiser
I II III IV V
1 Total no.of samples (N)
2 No. of Trials (n)
3 No of good samples (NG)
4 No of bad samples (NB)
5 No of Miss (NM)
6 No of False Alarm (NFA)
7 No of Good Decision (NGD)
8 Probability of Miss P(M) = NM / NB X n
9 Probability of False Alarm P (FA) = N
FA / NG X n
10 Effectiveness (E) = N GD / N X n
11 * Conclusion
* Refer table
Vinay S K 50

Long Method
Evaluation criteria for inspection capability studies involving attribute data.
Parameter Acceptable Marginal Unacceptable
E 0.9 or More 0.8 to 0.9 Less than 0.8
P (FA) 0.05 or Less 0.05 to 0.1 More than 0.1
P (Miss) 0.02 or Less 0.02 to 0.05 More than 0.05
B 0.80 to 1.20 0.50 to 0.80 or Less than 0.5 or
1.2 to 1.5 More than 1.5
Special cases in computing bias

P (FA) P (Miss) B Decision or action
0 More than 0 0 Unacceptable
More than 0 0 No value Use E,P(FA),and P(Miss) directly
0 0 No value This is same as E=1 since ,P(FA)= P(Miss)
acceptable
More than 0.5 0.5 or Less More than 1.5 Unacceptable
0.5 or Less More than 0.5 Less than 0.5 Unacceptable
More than 0.5 More than 0.5 No value Bias unimportant, study is
unacceptable.Based on P(Miss) and
P(FA) being More than 0.5
Vinay S K 51
25
 Constants for evaluating attribute data

P(FA) or P(MISS) B(FA) or B(MISS) P(FA) or P(MISS) B(FA) or B(MISS)
0.01 0.0264 0.17 0.2541
0.02 0.0488 0.18 0.2613
0.03 0.0681 0.19 0.2707
0.04 0.0863 0.20 0.2803
0.05 0.1040 0.21 0.2874
0.06 0.1200 0.22 0.2966
0.07 0.1334 0.23 0.3034
0.08 0.1497 0.24 0.3101
0.09 0.1526 0.25 0.3187
0.10 0.1758 0.26 0.3251
0.11 0.1872 0.27 0.3312
0.12 0.1989 0.28 0.3372
0.13 0.2107 0.29 0.3429
0.14 0.2227 0.30 0.3485
0.15 0.2323 0.31 0.3538
0.16 0.2444 0.32 0.3572
Vinay S K 52
0.33 0.3621
0.34 0.3668
0.35 0.3712
 Constants for evaluating attribute data
0.36 0.3739
0.37 0.3778
0.38 0.3814
0.39 0.3836
0.40 0.3867
0.41 0.3885
0.42 0.3910
0.43 0.3925
0.44 0.3945
0.45 0.3961
0.46 0.3970
0.47 0.3977
0.48 0.3984
0.49 0.3989
0.50 0.3989
Vinay S K 53
26
 List of Measurement Systems

SL. Part Param Instrum Manufact
No. No eter ent urer Variable Attribute
Max/
Min
Bias Line Stabili R&R
arity ty
Vinay S K 54
Conclusion :
MSA study has become important since it can be applied

under the following areas
 Criterion to accept new instrument
 Comparison of one measuring device against another
 Basis for evaluating a gage suspected of being deficient
 Comparison of equipment before and after repair
 Required element for calculating process variation
 Develop Gage Performance Curve which indicates the
probability of gage accepting a part of some true value
Limitation of standard are it is Difficult to use in
destructive testing , Some Processes ,Product
Characteristics & Tests have no defined industry or
National standards
Vinay S K 55
27
Any
QUESTIONS ???
Vinay S K 56
Thank You !
Vinay S K 57
28

Measurement System Analysis - 4th Edt - Handouts 2

Caricato da

Informazioni sul documento

Descrizione originale:

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Measurement System Analysis - 4th Edt - Handouts 2

Caricato da

Copyright:

Formati disponibili

Welcome to the seminar