Gage R&R
Estimating measurement components
Gage capability and acceptability measures
Prof. Tom Kuczek, Purdue Univ.
Terms and Definitions
Repeatability refers to the measurement
variation obtained when one person
repeatedly measures the same item with the
same gage.
Reproducibility refers to the variation due to
different operators using the same gage
measuring the same item.
Estimated Common Cause
Our estimate of Common Cause Variation,
which is the variance of the actual product
measurement, is actually the sum of three
components:
The true product variation.
Variation due to different operators
(reproducibility).
Variance of measurement equipment error
(repeatability).
3
Measurement Component Analysis
Notation
Gage R&R: Designed Data Collection
In order to estimate these components of
variation, we do a standard Gage R&R study.
All such studies follow the following format:
We select a fixed number of parts.
We select a fixed number of operators.
Each operator measures each of the parts a
fixed number of times.
Layout of Typical Gage R&R Study
Analysis of Gage R&R Study Data
There are two typical Statistical tools for the
analysis of the data from Gage R&R studies:
The first, and most widely taught technique, is
the analysis of average Ranges.
Ranges are obtained from successive
replications to estimate error variance.
Ranges from averages between different
operators for the same part are used to
estimate operator variation.
8
Analysis of Variance for Gage R&R
The Analysis of Variance (ANOVA) can also be
used to analyze Gage R&R studies.
In ANOVA terminology, most Gage R&R
studies have an ANOVA type data structure.
A variance component analysis can easily be
done in most software packages.
The individual variance components provide
estimates of error, operator and true
product variance.
9
Typical example
10
Variance components of Model 1
Yijk Parti Operatorj repetition(ij ) k
11
ANOVA of previous example in JMP
12
Variance components from ANOVA
13
Variance components
Part has variance
p2
Operator has variance
o2
Repetition within Part Operator combination
has variance
e2
14
Variance component estimates
from Analysis of Variance
p2 530.889
o2 19.525
e2 12.446
15
Variance Component model 2
This includes an interaction term. In practice,
most Analyses use model 1.
Yijk Parti Operatorj Part * Operatorij repetition(ij ) k
16
Model 2 variance components
The variance components will include a term for
the interaction of operator and part.
m 2 p 2 o 2 op 2 e 2
For Model 2, Reproducibility will be
Reproducibility=
o 2 op 2
17
Running model 2 in JMP
18
Variance component estimates for Model 2
19
Model 1 vs. Model 2
An F-test (not shown) shows that the
Operator*Part term is not significant at alpha
=.10 (authors criteria), then we will use the
results of Model 1.
Operator*Part is rarely significant. As a result
many people leave it out in practice.
20
Conclusions from Gage R&R
Part has the greatest variability
Operator and repetition are negligible
Invest in the Process
The Measurement system here is not an issue
in the sense of Reproducibility and
Repeatability
Concentrate on the Process, put on Target
with Minimum Variation
21
Gage Capability
22
Gage Acceptability (usually specified by
contract as a numerical value)
23
Precision to Tolerance ratio
The Gage Precision to Tolerance ratio is generally
defined as
6 gauge / (USL LSL)
P/T =
Where for our example
gauge e 2 o 2 19.525 31.971 5.6543
P/T < .1 is generally considered acceptable by the
author.
24