GAUGE CAPABILITY (R & R) STUDIES
In any problem involving measurement, some of the observed variability will
be due to variability in the product itself and some will be due to measurement
error. The measurement error usually follows the normal distribution.
Expressed mathematically:
σ 2 observed = σ 2 parts + σ 2 gauge
Where σ2gauge results from measurement error.
We can extend the above to include the components of gauge variability:
σ 2 gauge = σ 2 repeatability + σ 2 reproducibility
σrepeatability is a measure of the equipment variation and σ reproducibility is a measure
of the appraiser variation. 5.15σrepeatability will encompass 99% of the
measurement error due to the equipment (inherent precision of the gauge).
5.15σreproducibility will encompass 99% of the measurement error due to the
appraisers (effect of different operators). 5.15σgauge is the gauge capability.
One Appraiser
Consider that 20 parts are each measured twice by one appraiser:
Part Trials Mean Range
1 2
1 21 20 20.5 1
2 24 23 23.5 1
3 20 21 20.5 1
4 27 27 27.0 0
5 19 18 18.5 1
6 23 21 22.0 2
7 22 21 21.5 1
8 19 17 18.0 2
9 24 23 23.5 1
10 25 23 24.0 2
11 21 20 20.5 1
12 18 19 18.5 1
13 23 25 24.0 2
14 24 24 24.0 0
15 29 30 29.5 1
16 26 26 26.0 0
17 20 20 20.0 0
18 19 21 20.0 2
19 25 26 25.5 1
20 19 19 19.0 0
1
Part Trials Mean Range
Σ = 446 Σ = 20
X = 446 / 20 = 22 .30
R = 20 / 20 =1.00
The standard deviation for repeatability can be estimated from:
σ rep ea tabityil = R / d 2 *
d2* is used (rather than d2) in gauge repeatability and reproducibility studies.
Values for d2* are given below:
Number Subgroup Size
of 2 3 4
Subgroup d2*
s
1 1.41 1.91 2.24
2 1.28 1.81 2.15
3 1.23 1.77 2.12
4 1.21 1.75 2.11
5 1.19 1.74 2.10
6 1.18 1.73 2.09
7 1.17 1.73 2.09
8 1.17 1.72 2.08
9 1.16 1.72 2.08
10 1.16 1.72 2.08
11 1.16 1.71 2.08
12 1.15 1.71 2.07
13 1.15 1.71 2.07
14 1.15 1.71 2.07
15 1.15 1.71 2.07
15 + 1.128 1.693 2.059
In this case there are 20 subgroups (parts) of subgroup size 2 (trials):
∴σrepeatabil ity =1.00 / 1.128 = 0.887
∴5.15 σrepeatabil ity = 5.15 (0.887 ) = 4.57
It is sometimes useful to compare the gauge capability with the product
tolerance using the %Tolerance:
%Tolerance = 100 (5.15 σgauge ) /( USL − LSL )
If, for the part in question, the USL = 60 and the LSL = 5:
2
%Tolerance = 100(4.57)/(60 – 5) = 8.3%
Values of %Tolerance of 10% or less imply adequate gauge capability.
Many Appraisers
If it is determined that for appraisers 2 and 3:
Appraiser 2, X =22 .28 and R =1.25
Appraiser 3, X =22 .60 and R =1.20
Then:
R =(1 / 3)( 1.00 +1.25 +1.20 ) =1.15
R X = X max −X min = 22 .60 −22 .28 = 0.32
∴σrepeatabil ity = R / d 2 * =1.15 / 1.128 =1.02
The standard deviation for reproducibility can be estimated from:
σ reproducib ility = (R X /d 2 *) 2 − σ 2 repeatabil ity /nr
In this case there is 1 subgroup (appraisers) of subgroup size 3.
n is the number of parts and r is the number of trials.
∴σ reproducib ility = (0.32/1.91 ) 2 −1.02 2
/20(2) = 0.045
∴5.15 σreproducib ility = 5.15 (0.045 ) = 0.233
Since:
σ 2 gauge = σ 2 repeatability + σ 2 reproducibility
σ 2 gauge = 1.022 + 0.0452 = 1.04, giving σ gauge = 1.02
∴5.15 σgauge = 5.15 (1.02 ) = 5.25
∴%Tolerance = 100 (5.15 σgauge ) /( USL − LSL ) = 100 (5.25 ) /( 60 − 5) = 9.6%
© R.L.Roome, 2004
APPENDIX
3
Some textbooks present the formulae in a slightly different form:
EV = K1 R
Where:
EV = Equipment Variation
EV = 5.15 σrepeatabil ity
K1 = 5.15 / d 2 (d2 is d2* for 15+ subgroups)
and:
AV = Appraiser Variation
AV =5.15 σ reproducib ility
AV = (K 2 R X ) 2 − EV 2 /nr
K 2 = 5.1 5/ d 2 * (d2* for 1 subgroup)
Typical values of K1 and K2 are:
Number of trials 2 3 4
K1 4.46 3.05 2.5
Number of Operators 2 3 4
K2 3.65 2.70 2.30
GR & R = AV 2 + EV 2
Where:
GR&R = Gauge Repeatability and Reproducibility
