Professional Documents
Culture Documents
Approval of Measurement Facility Capability / Specific Inspection Facility
Approval of Measurement Facility Capability / Specific Inspection Facility
R /B
Approval of measurement facility capability / ________________________________________
Standard
Specific inspection facility ________________________________________
Status Enforceable
TRANSLATION ADVISORY NOTICE: This document has been translated from French. In the event of any
dispute, only the French version is referred to as the reference text and is binding on the parties
Object Define the capability approval procedure for any specific inspection facility,
whatever its characteristics, used for dimensional measurement.
It applies to all specific inspection facilities defined above, for example inspection
fixtures, specific automatic inspection machines, inspection stations built into
machine tools, etc.
It is possible to apply this procedure for checking of a specific characteristic of an
all-purpose measurement facility, according to predetermined instructions, after
conventional geometric acceptance testing of the facility.
By analogy, application of this standard can be extended to inspection facilities for other
measurable variables.
History of versions
Class E41
Key words capabilite, aptitude, statistique, cmc, mesure, controle dimensionnel, capability, ability,
measurement, statistics, dimensional check
Language English
Contents
Page
1 General ...................................................................................................................................... 3
2 Definitions ................................................................................................................................. 4
1 General
A specific inspection facility, whether manual or automatic, is a facility only able to check the
characteristics for which it has been designed (e.g.: multidimension check).
The purpose of capability approval is to:
Provide a practical check that the equipment and the process are able to carry out the
operation for which they have been designed (functional aptitude),
Check that the degree of uncertainty in measurement is compatible with the tolerances for the
parts to be inspected (accuracy of measurement).
This type of check makes it possible to evaluate both uncertainties due to manufacture and those due to
design.
It is also used to check that the utilisation and calibration instructions are of operational standard.
Le The principal consists in measuring one or several parts first using metrological means and then using
the specific inspection facility, and then comparing the results.
2 Definitions
Calibration master (for specific inspection facility)
Measurement made on a part with dimensions generally close to those of parts to be checked and
with faults in form and temporal stability of dimensions compatible with the degree of precision
required, this being better than of the facility as s whole.
Repeatability
Dispersion of measurement results for a part, repeated in identical conditions and without
recalibration, over a short period:
Repeatability of calibration master measurements is equal to ± Ie where :
le = 2 se
and "se" is the experimental standard deviation over several calibration measurements
taken on the measurement facility.
Repeatability of measurements for a parts equal to ± Ir where :
Ir = 2 sr
and "sr" is the experimental standard deviation over several measurements of a part taken
on the measurement facility.
Tolerance bracket
lT = TB= upper and lower limits of tolerance on the characteristics checked on the facility.
IT
CMC =
2 Ig
3.1 Precautions
The finalised inspection facility is to be installed under the normal utilisation conditions specified by the
instructions for utilizations at stations.
Measurements are made on the inspection facility in strict accordance with the instruction in the
instruction manual for the inspection facility.
Calculate the experimental calibration standard deviation (se) on the inspection facility.
Repeatability of measurements on the inspection facility is equal to ± Ie where le = 2 se.
If le is greater than the value specified in paragraph 5, the inspection facility must be reviewed.
Note: If the nature of the clause checked does not require a calibration master, take se = 0.
If Ir is greater than the value specified in paragraph 5, the inspection facility must be reviewed before
carrying on with the acceptance procedure.
This check makes it possible to avoid complete acceptance if the inspection facility has a serious fault.
If no serious fault is encountered, go to the next operation.
Precautions:
The uncertainty for metrology measurement must be equal to or less than half of the theoretical degree of
uncertainty of measurement on the inspection facility
It is possible that geometrical faults in machining origins or deformation of a part may introduce e degree
of dispersion of measurement making it possible to fulfil the condition described above
4.1 Notation
p = number of parts
i = part index (from 1 to p)
m = number of measurement on a part or on calibration master on the inspection facility
j = measurement index (from 1 to m)
x = value obtained in metrology
y = value obtained on inspection facility
xi = conventional true value for part no i, measured in metrology
yij = value measured on inspection facility, on part no. i, during measurement no. j
J = average precision error for inspection facility
V = variance
s = experimental standard deviation = V
l = measurement uncertainty
Indices affecting these 3 variables:
a = amplification
l = linearity
r = repeatability
g = global
e = calibration master
metro = metrology
lT = tolerance bracket (TB) for the characteristic to be checked.
1 p
x=
p
∑ i = 1 xi = average of measurement on all parts in metrology
1 m
yi =
m
∑ j = 1 yij = average of m measurements on part No I on inspection facility
1 p
y=
p
∑ i = 1 yi = average of all measurement on all parts on inspection facility
Principle of calculation
The results of the (n) of the measurements (z) are considered as having the same distribution law.
Estimation of variance V (z) is given by the formula:
1 n
V(z) = ∑ (zi - z)2
n - 1 i =1
∑ dij2 - pm ( ∑ dij )
1 1 2
Vg =
pm - 1
1 1 m 2
Ve = ∑ mj = 1 y ej 2 - ∑ j = 1 y ej
m - 1 m
J= y - x
ig = J +2 Vg + Ve
6 Simplified acceptance
This procedure only applies in cases where it is impossible to apply the acceptance procedure
described in the paragraph above. (For example, if typical parts are missing on delivery on inspection
facility)
The procedure makes it possible to refuse a facility which fails to comply, but acceptance can only be
pronounced after the full acceptance procedure.
The method is identical to that for complete acceptance, but here only on part is measured (instead of
the minimum 5).
This part is measured in metrology, then 5 times on the inspection facility without modifying the
calibration, and lifting the part after each measurement.
The graphs and calculations described, in this annex are optional. They allow you to analyse the various
causes of uncertainty and to arrive at a "diagnosis" of them.
For example, it is possible to discern a predominant fault. If there is no predominant fault, results must be
interpreted with caution.
∑ = ∑ pi = 1
where n = m . p (Number of xi, yij points)
n ∑ xi . yi - ∑ xi . ∑ yi
a=
n ∑ x i2 - (∑ xi )
2
b = y - a.x
To determine whether this curve is significantly different from the y = x curve, it is possible to apply the
following statistical test:
1 2
The quantity: F = t + t a 2
2 y
Follows a Snédécor law with 2 and (p - 2) degrees of freedom (see table, paragraph 3 of annex) where:
p ( y - x )2
t y2 =
Vl
(a - 1)2 ∑ pi = 1 ( xi - x )2
t a2 =
Vl
(for Vl calculation see paragraph 2.2 of annex).
In Snédécor table (see paragraph 3 of annex), read the value for Flimit for γ1 = 2.
γ2 = p - 2 and p = 0.05 :
Si F ≤ Flimit the regression curve is not significantly different from y = x.
Si F > Flimit the regression curve is significantly different from y = x.
To determine whether this is due to the slope of the curve or to precision error, use the Student-Ficher
test: t a and t y according to a Student-Ficher law with (p - 2) degrees of freedom.
In the Student-Ficher table (see paragraph 3 of annex), read the value of t for γ = p - 2 and P = 0.05.
If t a > t is read in the table, the difference is due to the slope of the curve (amplification error).
If t y > t is read in the table, the difference arises from precision error.
2 Calculation of variances
1 1 2
V l = p - 2 ∑ i = 1 ei - p ∑ i = 1 ei
p
2 p
2
(a −1) . P 2.. − .. 1 ..( P )2
Va =
p ∑I+1 XI ∑I=1 xi
P
2.4 Remark
The relationship of one of these three variances to overall variance Vg gives an order of magnitude for the
percentage of variability explained by the variance chosen. However, this relationship is only an indication
as the sum of (Vr + Vl + Va) is not exactly equal to Vg, the third variance not being totally independent of
the first two. The aim of the calculation of these variances is simply to give an orientation for diagnosis.
3 Distribution tables
γ1= 2
γ2
P =005 DISTRIBUTION TABLE FOR F
1 199.5 (Snédécor Variable)
2 19.5
3 9.55 Value for F with probability P being exceeded
4 6.94
5 5.79 (F = S12 / S 22 )
6 5.14
7 4.74
8 4.46
9 4.26
10 4.10
11 3.98
12 3.88
13 3.80
14 3.74
15 3.68
16 3.63
17 3.59
18 3.55
19 3.52
20 3.49
P 0.05
γ
1 12.706
2 4.303 DISTRIBUTION TABLE FOR t
3 3.182 (Student law)
4 2.776
5 2.571 Values for t with probability P of being exceeded for absolute
6 2.447 valued'être dépassée en valeur absolue
7 2.365
8 2.306
9 2.262
10 2.228
11 2.201
12 2.179
13 2.160
14 2.145
15 2.131
16 2.120
17 2.110
18 2.101
19 2.093
20 2.086
Figure 1
x
INSPECTION INSPECTION
FACILITY FACILITY
y x
y
x
x
J x
x
x
x
x x
x
x x
METROLOGY METROLOGY
PRECISION ERROR
AMPLIFICATION ERROR
INSPECTION INSPECTION
+
FACILITY FACILITY + +
y y +
+
+ +
+ + +
+
+ +
+
+ +
+ +
+
+ +
+
+ +
+ + +
+ + +
+ + +
+ x x
+
METROLOGY + METROLOGY
Figure 2
INSPECTION
FACILITY
J
+
x
=
y
b
x+ E
=a L IN
Y N
IO
E SS
GR
RE
x
=
y
y
yij
REPEATABILITY
yi J
LINEARITY
AMPLIFICATION
PRECISION
x1 x1 x METROLOGY
= [yij –yi] + [yi – (axi + b)] + [(axi + b) – (xi + J)] + (xi + J) – xi]
REPEATABILITY LINEARITY AMPLIFICATION AVEARGE PRECISION
When it is not possible to obtain parts distributed over more than 0,6 TB (IT):
The parts must have an optimum distribution of their characteristics over a bracket 0.6 TB: a parts
sample aiming at this optimum objective is necessary in order to evaluate errors of accuracy, linearity
and amplification across the entire operating range of the inspection facility.
Ideally range should extend beyond the TB.
For practical reasons, this type of sample is not always possible at first reception of the facility; in this
case it is possible to refuse the facility if the CMC is outside of tolerances for a single part or for parts
extending avec less than 0.6 TB, but final acceptance can only be given for series production parts with
adequate dispersion.
Yi − Ls
where K =
IT
Thus to evaluate the risk of error of judgment R applying to a value from a facility, we calculate for each
K, the normal density of probability of risk, which can be determined from the table below.
This table, which characterizes the function of distribution of the reduced normal variable, determines the
probability P from the expression 4 x K x CMC, which allows us to evaluate the risk R from the statement
R = 1 - P.
Calculation of K
Ps = 1 - Rs = 0.95 (i.e. 95 %)
→ 4 x K x CMC = 1.64
→ K = 0.12
Yi − L s
K= = 0.12 → Yi − L s = 0.012 where IT = 0.1
IT
Thus the limits of the measurement range brought to a low limit and having a risk level greater than
5% are:
9.938 mm and 9.962 mm
And the limits for the measurement range with a risk greater than 5 % brought to the upper limit are:
10.038 mm and 10.062 mm
Uncertainty
Ie = 2 V e = 1.0954
Average
∑ yj = 49.000
y=
m
1 2
Repeatability variance Vr =
(m - 1)
∑ (y j - y) = 1.1111
Ir = 2 Vr
Uncertainty = 2.1082
m : number of measurement and
5 ≤ m ≤ 10
DECISION (1) Ir > IT/8 (2) Facility to be reviewed
IT/8 (3) IT = TB 6.2500 Ir ≤ IT/8 (2) Acceptable
∑ ∑ (dij - J)2
Difference overall variance: Vg = = 5.2600
pm - 1