Teaching Measurement Systems Analysis to Engineers Using R
Thomas Roth
[Link]@[Link]
Linearity and Stability
MSA Type II Gage Repeatability & Reproducibility
Measurement Systems Analysis (MSA) is based on
the reference manual of the Automotive Industry
Action Group (AIAG) which was first published in
1990. Based on this reference manual a variety of
company guidelines and other texts such as the
Band 5 of the Verein Deutscher Automobilbauer1
(VDA) originated, which in its 2nd edition, is
based on the Norm ISO 22514-7 Measurement
Process Capability and is to be published at the
end of 2011 [7]. In addition, the verification of
measuring equipment to ensure valid results is a
formal requirement of any ISO 9001 certified
corporation [4]. With the ISO 9001 topping one
million certifications in 2009 [3], Measurement
Systems Analysis should be an important aspect
within the statistical education of engineers.
Moreover, Measurement System Analysis is a
standardized procedure within a methodology.
For assessing the linearity of a gage, parts with
measurements covering the operating range of
the gage are taken. Each reference part is
repeatedly
measured
under
repeatability
conditions. A graphical and numerical analysis of
the measurements is done [1].
The focus of this study is to identify the source and
magnitude of the variation for a given gage in a
production environment. The quality characteristic of
parts is measured under reproducibility conditions
and the parts are sampled from the production
process. Using a Gage R&R study the repeatability
and reproducibility of the gage is determined.
Repeatability refers to the inherent precision of a
measurement system. Reproducibility is the part of
the overall variance that models the effect of
different e.g. operators performing measurements
on the same unit and a possible interaction between
different operators and parts measured within this
Gage R&R. The relation between these component is
given by [1]:
10
Methods for analyzing this design are given by
gageRR and plot.
gdo = gageRR(design); plot(gdo)
Components of Variation
-----------LINEARITY:
13.16667
The focus of this study is to determine the bias
and repeatability of the gage.
The quality
characteristic of a reference part is measured
repeatedly under repeatability conditions and the
repeatability and bias is determined. The
capability of the gage is then expressed in the
gage capability indices and (analogous to
the process capability indices and ) which
relates the reference interval of the measurements
to a portion of the tolerance interval of the
considered quality characteristic [8]. The portion
depends on company guidelines and usually takes
on values around 0.2.
example(cg)
10.4
RunChart
xtar 0.1 T
x
s
target
Cg
Cgk
= 9.97
= 0.1
= 10
= 1.41
= 1.3
10.2
x99.865%
Histogram of x - target
[Link]
H0 : Bias = 0
t-value: -3.493
p-value: 0.001
Density
10.0
target
-0.3
-0.2
-0.1
0.0
0.1
0.2
0.3
9.8
-0.4
xtar 0.1 T
10
8
x0.135%
11
12
Tolerance View
20
40
22
24
26
28
reproducibility
PartToPart
60
Index
80
100
120
20
40
60
Index
80
100
120
25
x
23
21
C
1
2
3
3
3
1
2
3
1
3
2
1
2
1
3
2
1
2
1
A
B
C
3
2
1
UCL = 2.25
R 1.27
LCL = 0.32
0.0
25
1
2
1
2
2.0
27
Operator
2
1
3
3.0
R Chart
C
Operator
Interaction Operator:Part
23
27
28
26
24
22
2
E
Part
C
Operator
Gage R&R
VarComp
Suppose 10 randomly chosen units were measured
by 3 randomly chosen operators. Each operator
measured each unit two times in a randomly chosen
order and the units were presented in a way they
could not be distinguished by the operators.
UCL = 23.81
x 22.93
LCL = 22.06
Operator
2Operator + 2PartsxOperator
x Chart
20
mean of Measurement
In here Parts models the variation between different
units of the same process. 2Parts is thus an estimate
of the inherent process variability. Repeatability is
modeled by 2Error and reproducibility by:
E
Part
Measurement by Operator
21
2Total = 2Parts + 2Operator + 2PartsxOperator + 2Error
9.6
bias:
difference
between
the
average
measurement and the reference value (accuracy)
repeatability: is given by the standard deviation of
subsequent measurements under repeatability
conditions (inherent precision)
reproducibility: is given by the average of
measurements for a change in one factor
repeatability conditions: fixed and defined
conditions i.e. identical operator, measurement
device, part and quality characteristic
reproducibility conditions: repeatability conditions
with one factor subject to change e.g. operator
reference interval: the range in which 95.5% or
99.73% of the measurements are to be expected
tolerance interval: upper lower specification limit
repeatability
component
MSA Type I Bias & Repeatability
Gages can have three types of impairments which
are a lack of linearity, a bias and variation. The
latter can be split into system immanent variation
of the measurement itself and variation introduced
by other factors such as operators using these
measurement systems:
20
totalRR
Concepts & Terminology
Measurement by Part
VarCompContrib
StudyVarContrib
ptRatio
Reference Values
Measurement
response(design) = c(23,22,22,22,22,25,23,)
1.5 2.0
0.0
-0.5
Bias
0.5
Single Bias
Mean Bias
Regression
95 % [Link]
The measurements are assigned to this design using
the response method.
0.5 1.0
1.0
example(gageLin)
design = gageRRDesign(O=3, P=10, M=2)
0.0
library(qualityTools)
The corresponding Gage R&R design can be created
using the gageRRDesign method of the
qualityTools package.
1.0
Analyze
Introduction
Measurement
USL
LSL
Mea
s
u
r
rove C
p
Im
De
f
e
in
ol
r
t
on
Ratio
totalRR
repeatability
reproducibility
Operator
Operator:Part
Part to Part
totalVar
1.664
1.209
0.455
0.455
0.000
1.781
3.446
StudyVarContrib P/T
0.695
0.592
0.364
0.364
0.000
0.719
1.000
1.55
1.32
0.81
0.81
0.00
1.60
2.23
Summary
MSA is covered in an obligatory introductory
statistics course for engineers at the TU Berlin
using the qualityTools package. Understanding
the methodology is enhanced by performing a
Gage R&R during class. In this activity the
thickness of each of six dice is measured three
times by three students in a randomized fashion
[5]. This classroom experiment proved to be
valuable for understanding the concept of
repeatability and reproducibility. The observed
variation is assigned easily to particular variance
components using the provided methods.
References
[1] A.I.A.G - Chrysler Corp., Ford Motor Co., General Motors Corp.: Measurement
Systems Analysis, Reference Manual. 4th ed., Michigan (USA)
[2] ISO (2010) ISO/DIS 22514-7: Statistical methods in process management -Capability and performance -- Part 7: Capability of measurement processes.
[3] The ISO-Survey of certifications 2009, 18th ed.
[4] ISO (2008). Quality management systems Requirements (ISO 9001:2008)
[5] Mathews, P.(2004): Design of Experiments with MINITAB. ASQ Quality Press.
ISBN 0-87389-637-8
[6] Roth, T. (2010). qualityTools: Statistical Methods for Quality Science.
[Link]
[7] Dietrich, E 2011, 'Wo liegen die Unterschiede', QZ - Qualitt und Zuverlssigkeit, no.
6, pp. 3034
[8] Dietrich, E & Schulze, A 2007, Eignungsnachweis von Prfprozessen.
Prfmittelfhigkeit und Messunsicherheit im aktuellen Normenumfeld, Hanser, Mnchen
[u.a.].
1Association
of German Car Manufacturers