MSA / GR&R

MSA / GR&R

MSA / GR&R

Validate
Validate Measurement
Measurement System
System

MSA / GR&R

Step 3: Validate Measurement System On Y

Measurement System
Data is from where ?
How do you get your data ?
What do you do with it ?
Sample checks taken to see that the data is true ?
If so, what were the results pass / fail ? (needs to pass
before you move to step 4 !)
NOTE different tool to be used for a GR&R if attribute or
continuous data
3

MSA / GR&R

Why
Why Measure?
Measure?
We Need to Know...
Dimensions
Outer Diameter, Inner Diameter, Length

Speeds
Cycle time, Feed Rate, Turning Speed

Qualities
Color, Finish, Porosity, Scratches

Quantities
# completed, # Good, # Bad

How much potential error do we introduce into

our process because of the way we measure???
4

MSA / GR&R

Warm-Up
Warm-Up Exercise
Exercise

You will be shown a paragraph of text

Individually you will count the letter f or F
You only have 40 seconds to count them

Not too difficult?

Would the class be a good gage?
5

MSA / GR&R

Warm-Up
Warm-Up Exercise
Exercise
The Necessity of Training Farm Hands in the Fatherly Handling of
Farm Live Stock is Foremost in the Eyes of Farm Owners. Since
the Forefathers of the Farm Owners Trained the Farm Hands for
First Class Farms in the Fatherly Handling of Farm Live Stock, the
Farm Owners Feel they should carry on with the Family Tradition
of Training Farm Hands of First Class Farmers in the Fatherly
Handling of Farm Live Stock Because they Believe it is the Basis
of Good Fundamental Farm Management

How Many Fs are in this Paragraph?

6

MSA / GR&R

Percent
Percent Effective
Effective
What would be an acceptable error rate
for this exercise?
90%, 95% , 99% or more?
% Effective = Number of people correct
Total people in the class
What if we had other paragraphs to count or
other letters to find?
Is this class a good measurement system?
7

MSA / GR&R

Warm-Up
Warm-Up Exercise
Exercise
The Necessity of Training Farm Hands in the Fatherly Handling of
Farm Live Stock is Foremost in the Eyes of Farm Owners. Since
the Forefathers of the Farm Owners Trained the Farm Hands for
First Class Farms in the Fatherly Handling of Farm Live Stock, the
Farm Owners Feel they should carry on with the Family Tradition
of Training Farm Hands of First Class Farmers in the Fatherly
Handling of Farm Live Stock Because they Believe it is the Basis
of Good Fundamental Farm Management

Did we just improve the measurement system?

8

MSA / GR&R

How
How to
to Assess
Assess Your
Your Measurement
Measurement System?
System?
We will look at a tool called GR&R
Attribute data
Variable data (optional module)
We will also look at what you can do when those
techniques do not apply
Electronic data, for example

MSA / GR&R

Electronic
Electronic Data
Data
What if your measurement data comes from an
electronic source, like SAP or some other data
management software?

Is there any human or gauging involved?

Discuss where your project data is coming from

10

MSA / GR&R

Example:
Example: Inventory
Inventory Reduction
Reduction Case
Case Study
Study
The retailer was collecting electronic data regarding inventory
weeks on-hand and stock turns through a retail management
software.
In order to determine the ability of this measurement system,
an audit of several stores was performed to hand-count
inventory levels and review stock turns.
There was a 100% match between the electronic and store
data.

An audit may be a suitable assessment of the

measurement system for data from an electronic source
11

MSA / GR&R
When
WhenGR&R
GR&Ris
isnot
not applicable
applicable how
how to
tovalidate
validate the
the
measurement
measurement system?
system?
Check for all records being complete
Check for missing values, gaps, zero records etc.

Create a Time Series Plot / Graph

Look for data that does not fit the norm
E.g. Airfreight cycle time (2 days) vs. Sea freight (30 days)

Review Subsets
Do like processes give similar results?
E.g. Do all containers on the same vessel have the same transit
time?

Data types
Is the data type consistent?
E.g. Has text been entered when numeric is expected etc.
12

MSA / GR&R
Example:
Example: Identifying
Identifying discrepancies
discrepancies in
indata
data

13

MSA / GR&R

Attribute
Attribute GR&R
GR&R

14

MSA / GR&R

Why
Why Study
Study Attribute
Attribute
Measurement
Measurement System?
System?

To determine how well quality inspectors are conforming

to known boundary samples
To determine how well quality inspectors are conforming
to themselves
To determine if the existing boundary samples provide
sufficient criteria for quality inspectors

Recall: Attribute data is discrete data, i.e. counts

15

MSA / GR&R

Principle
Principle of
of GR&R
GR&R
Parts representative of process

Operators making decisions

1. Does each operator make the same

assessment every time he/she assesses the
same part?
2. Do different operators apply different
standards?
3. Even when all operators agree every time,
do they make the right decision?

16

MSA / GR&R

Principle
Principle of
of GR&R
GR&R
Does each operator make the same assessment
every time he/she assesses the same part?

REPEATABILITY
Do different operators apply different standards?

REPRODUCABILITY
Even when all operators agree every time, do they
make the right decision

EFFECTIVENESS
17

MSA / GR&R

Acceptance
Acceptance Criteria
Criteria
% Appraiser Repeatability - indicates whether the
operator agrees with himself on both trials
% Score vs. Attribute - Indicates whether the operator
agrees on both trials with a known standard
Screen % Effective - Indicates whether operators agree
among themselves
Screen % Effective vs. Attribute - Indicates whether
operators agree with themselves and with the known standard

The target for all scores is 100%, but 80% or higher is

acceptable
18

MSA / GR&R

Attribute
Attribute GR&R
GR&R Guidelines
Guidelines

At least two, but preferably three or more operators

At least 10 units that represent 80% of the observed
process variation
Each unit is measured at least 2 times by each operator
The equipment must have adequate discriminating power
(minimum 10 X resolution)

19

MSA / GR&R

Attribute
Attribute GR&R
GR&R Procedure
Procedure
1.

Select 10 parts from the process that include both good

and bad samples

2.

Two appraisers independently measure each part twice

(in random order)

3.

Use the MSC_GRR.XLS spreadsheet to record the data

and calculate the Effective score

4.

An acceptable Effective score is 80% or higher

Is GR&R suitable for all transactional projects?

20

MSA / GR&R

Example
Example
Known Population
Sample #
Attribute
1
Pass
2
Pass
3
Pass
4
Pass
5
Pass
6
Fail
7
Fail
8
Fail
9
Fail
10
Pass

Operator #1
Try #1
Try #2
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Fail
Pass
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail

% APPRAISER SCORE(1) ->

% SCORE VS. ATTRIBUTE(2) ->

90.00%
80.00%

Operator #2
Try #1
Try #2
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail

100.00%
90.00%

Operator #3
Try #1
Try #2

Y/N
Agree
Y
Y
Y
Y
N
Y
Y
Y
Y
Y

Y/N
Agree
Y
Y
Y
Y
N
Y
Y
Y
Y
N

0%
0.00%

SCREEN % EFFECTIVE SCORE(3) -> 90.00%

SCREEN % EFFECTIVE SCORE vs. ATTRIBUTE (4) ->

80.00%

What do the numbers mean?

21

MSA / GR&R

Example
Example
Known Population
Sample #
Attribute
1
Pass
2
Pass
3
Pass
4
Pass
5
Pass
6
Fail
7
Fail
8
Fail
9
Fail
10
Pass

Operator #1
Try #1
Try #2
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Fail
Pass
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail

% APPRAISER SCORE(1) ->

(2)

% SCORE VS. ATTRIBUTE ->

90.00%
80.00%

Operator #2
Try #1
Try #2
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail

100.00%
90.00%

Operator #3
Try #1
Try #2

Y/N
Agree
Y
Y
Y
Y
N
Y
Y
Y
Y
Y

0%
0.00%

SCREEN % EFFECTIVE SCORE(3) -> 90.00%

SCREEN % EFFECTIVE SCORE vs. ATTRIBUTE (4) ->

Repeatability

Reproducibility

Y/N
Agree
Y
Y
Y
Y
N
Y
Y
Y
Y
N

Total Study
(looking for this
to be > 80.00%!!)

80.00%

22

MSA / GR&R

%
%Appraiser
Appraiser Score
Score
Known
Known Population
Population
Sample
#
Attribute
Sample #
Attribute
11
Pass
Pass
22
Pass
Pass
33
Pass
Pass
44
Pass
Pass
55
Pass
Pass
66
Fail
Fail
77
Fail
Fail
88
Fail
Fail
99
Fail
Fail
10
Pass
10
Pass

Operator
Operator #1
#1
Try
#1
Try
Try #1
Try #2
#2
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Fail
Pass
Fail
Pass
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail

(1)
(1)
%
% APPRAISER
APPRAISER SCORE
SCORE ->
->
(2)
(2)
%
% SCORE
SCORE VS.
VS. ATTRIBUTE
ATTRIBUTE ->
->

90.00%
90.00%
80.00%
80.00%

Operator
Operator #2
#2
Try
#1
Try
Try #1
Try #2
#2
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail

Operator
Operator #3
#3
Try
#1
Try
Try #1
Try #2
#2

Does
Does Not
Not Match
Match

100.00%
100.00%
90.00%
90.00%

Y/N
Y/N
Agree
Agree
YY
YY
YY
YY
NN
YY
YY
YY
YY
YY

Y/N
Y/N
Agree
Agree
YY
YY
YY
YY
NN
YY
YY
YY
YY
NN

0%
0%
0.00%
0.00%

(3)
SCREEN
SCREEN %
% EFFECTIVE
EFFECTIVE SCORE
SCORE(3) ->
-> 90.00%
90.00%
(4)
SCREEN
%
EFFECTIVE
SCORE
vs.
ATTRIBUTE
SCREEN % EFFECTIVE SCORE vs. ATTRIBUTE (4) ->
->

80.00%
80.00%

Appraiser Score calculated by the percent

match of these two columns

So:

9 Matches / 10 Pairs = 90%

23

MSA / GR&R

Screen
Screen %
% Effective
Effective
Known Population
Sample #
Attribute
1
Pass
2
Pass
3
Pass
4
Pass
5
Pass
6
Fail
7
Fail
8
Fail
9
Fail
10
Pass

Operator #1
Try #1
Try #2
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Fail
Pass
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail

All Do Not Match

% APPRAISER SCORE(1) ->

% SCORE VS. ATTRIBUTE(2) ->

90.00%
80.00%

Operator #2
Try #1
Try #2
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail

Operator #3
Try #1
Try #2

Indicated Here

100.00%
90.00%

Y/N
Agree
Y
Y
Y
Y
N
Y
Y
Y
Y
Y

Y/N
Agree
Y
Y
Y
Y
N
Y
Y
Y
Y
N

0%
0.00%

SCREEN % EFFECTIVE SCORE(3) -> 90.00%

SCREEN % EFFECTIVE SCORE vs. ATTRIBUTE (4) ->

80.00%

Screen % Effective Score calculated by the percent

match of the Appraisers to each other.
(Reproducibility)

So:

9 Matches / 10 Pairs = 90%

24

MSA / GR&R

Screen
Screen %
% Effective
Effective vs.
vs. Attribute
Attribute
Known Population
Sample #
Attribute
1
Pass
2
Pass
3
Pass
4
Pass
5
Pass
6
Fail
7
Fail
8
Fail
9
Fail
10
Pass

Operator #1
Try #1
Try #2
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Fail
Pass
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail

All Do Not Match

% APPRAISER SCORE(1) ->

% SCORE VS. ATTRIBUTE(2) ->

90.00%
80.00%

Operator #2
Try #1
Try #2
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Pass
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail
Fail

Operator #3
Try #1
Try #2

Indicated Here

100.00%
90.00%

Y/N
Agree
Y
Y
Y
Y
N
Y
Y
Y
Y
Y

Y/N
Agree
Y
Y
Y
Y
N
Y
Y
Y
Y
N

0%
0.00%

SCREEN % EFFECTIVE SCORE(3) -> 90.00%

SCREEN % EFFECTIVE SCORE vs. ATTRIBUTE (4) ->

80.00%

Screen % Effective Score vs. Attribute is calculated by

the percent match of the Screen Effective Score to
the Known Population (Total Study)

So:

8 Matches / 10 Pairs = 80%

25

MSA / GR&R

Attribute
Attribute GR&R
GR&R Extension
Extension

26

MSA / GR&R

Trouble-shooting
Trouble-shooting aa Poor
Poor Gage
Gage
Dominant Source of
Variation

Repeatability
1. Replace, repair, or adjust
equipment
2. If equipment is
performing to spec, either
live with the error or use
Signal averaging

Reproducibility
1. Re-train
2. Examine quality and skill
of standard processes
between operators

27

MSA / GR&R

Variable
Variable GR&R
GR&R
Variable GR&R uses the same principle but involves
more complex maths
We can understand more of where any problem is
coming from
Involves physical measuring devices and less human
assessment
This is an optional training module see your MBB

28

MSA / GR&R
Observed Process Variation

Measurement
Variation

Actual Process
Variation

Variation
due to
Gauge

Variation
due to Operators

Gauge R&R

Reproducibility

Repeatability
Accuracy

Calibration

Stability
Linearity
29

MSA / GR&R

GR&R
GR&R Summary
Summary

Is there a chance that measurement error is a high % of product /

service variation?

Run GR&R study

Attribute - If it is less then 80%, improve the measurement before

proceeding

Variable (short method) - Improve if % contribution >10% of

tolerance

GR&R is not always suited to Transactional processes - USE

WHAT MAKES SENSE

GR&R is used to determine the accuracy of our measurement

system and validate our data

Excessive variation in our measurement system must be

eliminated before proceeding with your process capability
analysis in step 4, Analyze phase
30